diff options
Diffstat (limited to 'tools')
81 files changed, 13738 insertions, 3161 deletions
diff --git a/tools/perf/Documentation/Makefile b/tools/perf/Documentation/Makefile index 5457192e1b4..bdd3b7ecad0 100644 --- a/tools/perf/Documentation/Makefile +++ b/tools/perf/Documentation/Makefile @@ -35,7 +35,7 @@ man7dir=$(mandir)/man7 # DESTDIR= ASCIIDOC=asciidoc -ASCIIDOC_EXTRA = +ASCIIDOC_EXTRA = --unsafe MANPAGE_XSL = manpage-normal.xsl XMLTO_EXTRA = INSTALL?=install diff --git a/tools/perf/Documentation/examples.txt b/tools/perf/Documentation/examples.txt new file mode 100644 index 00000000000..8eb6c489fb1 --- /dev/null +++ b/tools/perf/Documentation/examples.txt @@ -0,0 +1,225 @@ + + ------------------------------ + ****** perf by examples ****** + ------------------------------ + +[ From an e-mail by Ingo Molnar, http://lkml.org/lkml/2009/8/4/346 ] + + +First, discovery/enumeration of available counters can be done via +'perf list': + +titan:~> perf list + [...] + kmem:kmalloc [Tracepoint event] + kmem:kmem_cache_alloc [Tracepoint event] + kmem:kmalloc_node [Tracepoint event] + kmem:kmem_cache_alloc_node [Tracepoint event] + kmem:kfree [Tracepoint event] + kmem:kmem_cache_free [Tracepoint event] + kmem:mm_page_free_direct [Tracepoint event] + kmem:mm_pagevec_free [Tracepoint event] + kmem:mm_page_alloc [Tracepoint event] + kmem:mm_page_alloc_zone_locked [Tracepoint event] + kmem:mm_page_pcpu_drain [Tracepoint event] + kmem:mm_page_alloc_extfrag [Tracepoint event] + +Then any (or all) of the above event sources can be activated and +measured. For example the page alloc/free properties of a 'hackbench +run' are: + + titan:~> perf stat -e kmem:mm_page_pcpu_drain -e kmem:mm_page_alloc + -e kmem:mm_pagevec_free -e kmem:mm_page_free_direct ./hackbench 10 + Time: 0.575 + + Performance counter stats for './hackbench 10': + + 13857 kmem:mm_page_pcpu_drain + 27576 kmem:mm_page_alloc + 6025 kmem:mm_pagevec_free + 20934 kmem:mm_page_free_direct + + 0.613972165 seconds time elapsed + +You can observe the statistical properties as well, by using the +'repeat the workload N times' feature of perf stat: + + titan:~> perf stat --repeat 5 -e kmem:mm_page_pcpu_drain -e + kmem:mm_page_alloc -e kmem:mm_pagevec_free -e + kmem:mm_page_free_direct ./hackbench 10 + Time: 0.627 + Time: 0.644 + Time: 0.564 + Time: 0.559 + Time: 0.626 + + Performance counter stats for './hackbench 10' (5 runs): + + 12920 kmem:mm_page_pcpu_drain ( +- 3.359% ) + 25035 kmem:mm_page_alloc ( +- 3.783% ) + 6104 kmem:mm_pagevec_free ( +- 0.934% ) + 18376 kmem:mm_page_free_direct ( +- 4.941% ) + + 0.643954516 seconds time elapsed ( +- 2.363% ) + +Furthermore, these tracepoints can be used to sample the workload as +well. For example the page allocations done by a 'git gc' can be +captured the following way: + + titan:~/git> perf record -f -e kmem:mm_page_alloc -c 1 ./git gc + Counting objects: 1148, done. + Delta compression using up to 2 threads. + Compressing objects: 100% (450/450), done. + Writing objects: 100% (1148/1148), done. + Total 1148 (delta 690), reused 1148 (delta 690) + [ perf record: Captured and wrote 0.267 MB perf.data (~11679 samples) ] + +To check which functions generated page allocations: + + titan:~/git> perf report + # Samples: 10646 + # + # Overhead Command Shared Object + # ........ ............... .......................... + # + 23.57% git-repack /lib64/libc-2.5.so + 21.81% git /lib64/libc-2.5.so + 14.59% git ./git + 11.79% git-repack ./git + 7.12% git /lib64/ld-2.5.so + 3.16% git-repack /lib64/libpthread-2.5.so + 2.09% git-repack /bin/bash + 1.97% rm /lib64/libc-2.5.so + 1.39% mv /lib64/ld-2.5.so + 1.37% mv /lib64/libc-2.5.so + 1.12% git-repack /lib64/ld-2.5.so + 0.95% rm /lib64/ld-2.5.so + 0.90% git-update-serv /lib64/libc-2.5.so + 0.73% git-update-serv /lib64/ld-2.5.so + 0.68% perf /lib64/libpthread-2.5.so + 0.64% git-repack /usr/lib64/libz.so.1.2.3 + +Or to see it on a more finegrained level: + +titan:~/git> perf report --sort comm,dso,symbol +# Samples: 10646 +# +# Overhead Command Shared Object Symbol +# ........ ............... .......................... ...... +# + 9.35% git-repack ./git [.] insert_obj_hash + 9.12% git ./git [.] insert_obj_hash + 7.31% git /lib64/libc-2.5.so [.] memcpy + 6.34% git-repack /lib64/libc-2.5.so [.] _int_malloc + 6.24% git-repack /lib64/libc-2.5.so [.] memcpy + 5.82% git-repack /lib64/libc-2.5.so [.] __GI___fork + 5.47% git /lib64/libc-2.5.so [.] _int_malloc + 2.99% git /lib64/libc-2.5.so [.] memset + +Furthermore, call-graph sampling can be done too, of page +allocations - to see precisely what kind of page allocations there +are: + + titan:~/git> perf record -f -g -e kmem:mm_page_alloc -c 1 ./git gc + Counting objects: 1148, done. + Delta compression using up to 2 threads. + Compressing objects: 100% (450/450), done. + Writing objects: 100% (1148/1148), done. + Total 1148 (delta 690), reused 1148 (delta 690) + [ perf record: Captured and wrote 0.963 MB perf.data (~42069 samples) ] + + titan:~/git> perf report -g + # Samples: 10686 + # + # Overhead Command Shared Object + # ........ ............... .......................... + # + 23.25% git-repack /lib64/libc-2.5.so + | + |--50.00%-- _int_free + | + |--37.50%-- __GI___fork + | make_child + | + |--12.50%-- ptmalloc_unlock_all2 + | make_child + | + --6.25%-- __GI_strcpy + 21.61% git /lib64/libc-2.5.so + | + |--30.00%-- __GI_read + | | + | --83.33%-- git_config_from_file + | git_config + | | + [...] + +Or you can observe the whole system's page allocations for 10 +seconds: + +titan:~/git> perf stat -a -e kmem:mm_page_pcpu_drain -e +kmem:mm_page_alloc -e kmem:mm_pagevec_free -e +kmem:mm_page_free_direct sleep 10 + + Performance counter stats for 'sleep 10': + + 171585 kmem:mm_page_pcpu_drain + 322114 kmem:mm_page_alloc + 73623 kmem:mm_pagevec_free + 254115 kmem:mm_page_free_direct + + 10.000591410 seconds time elapsed + +Or observe how fluctuating the page allocations are, via statistical +analysis done over ten 1-second intervals: + + titan:~/git> perf stat --repeat 10 -a -e kmem:mm_page_pcpu_drain -e + kmem:mm_page_alloc -e kmem:mm_pagevec_free -e + kmem:mm_page_free_direct sleep 1 + + Performance counter stats for 'sleep 1' (10 runs): + + 17254 kmem:mm_page_pcpu_drain ( +- 3.709% ) + 34394 kmem:mm_page_alloc ( +- 4.617% ) + 7509 kmem:mm_pagevec_free ( +- 4.820% ) + 25653 kmem:mm_page_free_direct ( +- 3.672% ) + + 1.058135029 seconds time elapsed ( +- 3.089% ) + +Or you can annotate the recorded 'git gc' run on a per symbol basis +and check which instructions/source-code generated page allocations: + + titan:~/git> perf annotate __GI___fork + ------------------------------------------------ + Percent | Source code & Disassembly of libc-2.5.so + ------------------------------------------------ + : + : + : Disassembly of section .plt: + : Disassembly of section .text: + : + : 00000031a2e95560 <__fork>: + [...] + 0.00 : 31a2e95602: b8 38 00 00 00 mov $0x38,%eax + 0.00 : 31a2e95607: 0f 05 syscall + 83.42 : 31a2e95609: 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax + 0.00 : 31a2e9560f: 0f 87 4d 01 00 00 ja 31a2e95762 <__fork+0x202> + 0.00 : 31a2e95615: 85 c0 test %eax,%eax + +( this shows that 83.42% of __GI___fork's page allocations come from + the 0x38 system call it performs. ) + +etc. etc. - a lot more is possible. I could list a dozen of +other different usecases straight away - neither of which is +possible via /proc/vmstat. + +/proc/vmstat is not in the same league really, in terms of +expressive power of system analysis and performance +analysis. + +All that the above results needed were those new tracepoints +in include/tracing/events/kmem.h. + + Ingo + + diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt index 1dbc1eeb4c0..0ff23de9e45 100644 --- a/tools/perf/Documentation/perf-record.txt +++ b/tools/perf/Documentation/perf-record.txt @@ -29,13 +29,71 @@ OPTIONS Select the PMU event. Selection can be a symbolic event name (use 'perf list' to list all events) or a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a - hexadecimal event descriptor. + hexadecimal event descriptor. -a:: - system-wide collection + System-wide collection. -l:: - scale counter values + Scale counter values. + +-p:: +--pid=:: + Record events on existing pid. + +-r:: +--realtime=:: + Collect data with this RT SCHED_FIFO priority. +-A:: +--append:: + Append to the output file to do incremental profiling. + +-f:: +--force:: + Overwrite existing data file. + +-c:: +--count=:: + Event period to sample. + +-o:: +--output=:: + Output file name. + +-i:: +--inherit:: + Child tasks inherit counters. +-F:: +--freq=:: + Profile at this frequency. + +-m:: +--mmap-pages=:: + Number of mmap data pages. + +-g:: +--call-graph:: + Do call-graph (stack chain/backtrace) recording. + +-v:: +--verbose:: + Be more verbose (show counter open errors, etc). + +-s:: +--stat:: + Per thread counts. + +-d:: +--data:: + Sample addresses. + +-n:: +--no-samples:: + Don't sample. + +-R:: +--raw-samples:: +Collect raw sample records from all opened counters (typically for tracepoint counters). SEE ALSO -------- diff --git a/tools/perf/Documentation/perf-report.txt b/tools/perf/Documentation/perf-report.txt index 8aa3f8c8870..59f0b846cd7 100644 --- a/tools/perf/Documentation/perf-report.txt +++ b/tools/perf/Documentation/perf-report.txt @@ -24,6 +24,12 @@ OPTIONS --dsos=:: Only consider symbols in these dsos. CSV that understands file://filename entries. +-n +--show-nr-samples + Show the number of samples for each symbol +-T +--threads + Show per-thread event counters -C:: --comms=:: Only consider symbols in these comms. CSV that understands @@ -33,6 +39,28 @@ OPTIONS Only consider these symbols. CSV that understands file://filename entries. +-w:: +--field-width=:: + Force each column width to the provided list, for large terminal + readability. + +-t:: +--field-separator=:: + + Use a special separator character and don't pad with spaces, replacing + all occurances of this separator in symbol names (and other output) + with a '.' character, that thus it's the only non valid separator. + +-g [type,min]:: +--call-graph:: + Display callchains using type and min percent threshold. + type can be either: + - flat: single column, linear exposure of callchains. + - graph: use a graph tree, displaying absolute overhead rates. + - fractal: like graph, but displays relative rates. Each branch of + the tree is considered as a new profiled object. + + Default: fractal,0.5. + SEE ALSO -------- linkperf:perf-stat[1] diff --git a/tools/perf/Documentation/perf-sched.txt b/tools/perf/Documentation/perf-sched.txt new file mode 100644 index 00000000000..1ce79198997 --- /dev/null +++ b/tools/perf/Documentation/perf-sched.txt @@ -0,0 +1,41 @@ +perf-sched(1) +============== + +NAME +---- +perf-sched - Tool to trace/measure scheduler properties (latencies) + +SYNOPSIS +-------- +[verse] +'perf sched' {record|latency|replay|trace} + +DESCRIPTION +----------- +There's four variants of perf sched: + + 'perf sched record <command>' to record the scheduling events + of an arbitrary workload. + + 'perf sched latency' to report the per task scheduling latencies + and other scheduling properties of the workload. + + 'perf sched trace' to see a detailed trace of the workload that + was recorded. + + 'perf sched replay' to simulate the workload that was recorded + via perf sched record. (this is done by starting up mockup threads + that mimic the workload based on the events in the trace. These + threads can then replay the timings (CPU runtime and sleep patterns) + of the workload as it occured when it was recorded - and can repeat + it a number of times, measuring its performance.) + +OPTIONS +------- +-D:: +--dump-raw-trace=:: + Display verbose dump of the sched data. + +SEE ALSO +-------- +linkperf:perf-record[1] diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt index 0d74346d21a..484080dd5b6 100644 --- a/tools/perf/Documentation/perf-stat.txt +++ b/tools/perf/Documentation/perf-stat.txt @@ -40,7 +40,7 @@ OPTIONS -a:: system-wide collection --S:: +-c:: scale counter values EXAMPLES diff --git a/tools/perf/Documentation/perf-timechart.txt b/tools/perf/Documentation/perf-timechart.txt new file mode 100644 index 00000000000..1c2ed3090cc --- /dev/null +++ b/tools/perf/Documentation/perf-timechart.txt @@ -0,0 +1,38 @@ +perf-timechart(1) +================= + +NAME +---- +perf-timechart - Tool to visualize total system behavior during a workload + +SYNOPSIS +-------- +[verse] +'perf timechart' {record} + +DESCRIPTION +----------- +There are two variants of perf timechart: + + 'perf timechart record <command>' to record the system level events + of an arbitrary workload. + + 'perf timechart' to turn a trace into a Scalable Vector Graphics file, + that can be viewed with popular SVG viewers such as 'Inkscape'. + +OPTIONS +------- +-o:: +--output=:: + Select the output file (default: output.svg) +-i:: +--input=:: + Select the input file (default: perf.data) +-w:: +--width=:: + Select the width of the SVG file (default: 1000) + + +SEE ALSO +-------- +linkperf:perf-record[1] diff --git a/tools/perf/Documentation/perf-top.txt b/tools/perf/Documentation/perf-top.txt index 539d0128972..4a7d558dc30 100644 --- a/tools/perf/Documentation/perf-top.txt +++ b/tools/perf/Documentation/perf-top.txt @@ -3,36 +3,122 @@ perf-top(1) NAME ---- -perf-top - Run a command and profile it +perf-top - System profiling tool. SYNOPSIS -------- [verse] -'perf top' [-e <EVENT> | --event=EVENT] [-l] [-a] <command> +'perf top' [-e <EVENT> | --event=EVENT] [<options>] DESCRIPTION ----------- -This command runs a command and gathers a performance counter profile -from it. +This command generates and displays a performance counter profile in realtime. OPTIONS ------- -<command>...:: - Any command you can specify in a shell. +-a:: +--all-cpus:: + System-wide collection. (default) + +-c <count>:: +--count=<count>:: + Event period to sample. + +-C <cpu>:: +--CPU=<cpu>:: + CPU to profile. + +-d <seconds>:: +--delay=<seconds>:: + Number of seconds to delay between refreshes. --e:: ---event=:: +-e <event>:: +--event=<event>:: Select the PMU event. Selection can be a symbolic event name (use 'perf list' to list all events) or a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a - hexadecimal event descriptor. + hexadecimal event descriptor. --a:: - system-wide collection +-E <entries>:: +--entries=<entries>:: + Display this many functions. + +-f <count>:: +--count-filter=<count>:: + Only display functions with more events than this. + +-F <freq>:: +--freq=<freq>:: + Profile at this frequency. + +-i:: +--inherit:: + Child tasks inherit counters, only makes sens with -p option. + +-k <path>:: +--vmlinux=<path>:: + Path to vmlinux. Required for annotation functionality. + +-m <pages>:: +--mmap-pages=<pages>:: + Number of mmapped data pages. + +-p <pid>:: +--pid=<pid>:: + Profile events on existing pid. + +-r <priority>:: +--realtime=<priority>:: + Collect data with this RT SCHED_FIFO priority. + +-s <symbol>:: +--sym-annotate=<symbol>:: + Annotate this symbol. Requires -k option. + +-v:: +--verbose:: + Be more verbose (show counter open errors, etc). + +-z:: +--zero:: + Zero history across display updates. + +INTERACTIVE PROMPTING KEYS +-------------------------- + +[d]:: + Display refresh delay. + +[e]:: + Number of entries to display. + +[E]:: + Event to display when multiple counters are active. + +[f]:: + Profile display filter (>= hit count). + +[F]:: + Annotation display filter (>= % of total). + +[s]:: + Annotate symbol. + +[S]:: + Stop annotation, return to full profile display. + +[w]:: + Toggle between weighted sum and individual count[E]r profile. + +[z]:: + Toggle event count zeroing across display updates. + +[qQ]:: + Quit. + +Pressing any unmapped key displays a menu, and prompts for input. --l:: - scale counter values SEE ALSO -------- diff --git a/tools/perf/Documentation/perf-trace.txt b/tools/perf/Documentation/perf-trace.txt new file mode 100644 index 00000000000..41ed75398ca --- /dev/null +++ b/tools/perf/Documentation/perf-trace.txt @@ -0,0 +1,25 @@ +perf-trace(1) +============== + +NAME +---- +perf-trace - Read perf.data (created by perf record) and display trace output + +SYNOPSIS +-------- +[verse] +'perf trace' [-i <file> | --input=file] symbol_name + +DESCRIPTION +----------- +This command reads the input file and displays the trace recorded. + +OPTIONS +------- +-D:: +--dump-raw-trace=:: + Display verbose dump of the trace data. + +SEE ALSO +-------- +linkperf:perf-record[1] diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 9c6d0ae3708..b5f1953b614 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -158,13 +158,43 @@ uname_P := $(shell sh -c 'uname -p 2>/dev/null || echo not') uname_V := $(shell sh -c 'uname -v 2>/dev/null || echo not') # If we're on a 64-bit kernel, use -m64 -ifneq ($(patsubst %64,%,$(uname_M)),$(uname_M)) - M64 := -m64 +ifndef NO_64BIT + ifneq ($(patsubst %64,%,$(uname_M)),$(uname_M)) + M64 := -m64 + endif endif # CFLAGS and LDFLAGS are for the users to override from the command line. -CFLAGS = $(M64) -ggdb3 -Wall -Wstrict-prototypes -Wmissing-declarations -Wmissing-prototypes -std=gnu99 -Wdeclaration-after-statement -Werror -O6 +# +# Include saner warnings here, which can catch bugs: +# + +EXTRA_WARNINGS := -Wcast-align +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wformat +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wformat-security +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wformat-y2k +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wshadow +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Winit-self +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wpacked +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wredundant-decls +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wstack-protector +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wstrict-aliasing=3 +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wswitch-default +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wswitch-enum +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wno-system-headers +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wundef +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wvolatile-register-var +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wwrite-strings +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wbad-function-cast +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wmissing-declarations +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wmissing-prototypes +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wnested-externs +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wold-style-definition +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wstrict-prototypes +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wdeclaration-after-statement + +CFLAGS = $(M64) -ggdb3 -Wall -Wextra -std=gnu99 -Werror -O6 -fstack-protector-all -D_FORTIFY_SOURCE=2 $(EXTRA_WARNINGS) LDFLAGS = -lpthread -lrt -lelf -lm ALL_CFLAGS = $(CFLAGS) ALL_LDFLAGS = $(LDFLAGS) @@ -223,7 +253,7 @@ SPARSE_FLAGS = -D__BIG_ENDIAN__ -D__powerpc__ # Those must not be GNU-specific; they are shared with perl/ which may # be built by a different compiler. (Note that this is an artifact now # but it still might be nice to keep that distinction.) -BASIC_CFLAGS = +BASIC_CFLAGS = -Iutil/include BASIC_LDFLAGS = # Guard against environment variables @@ -288,11 +318,12 @@ export PERL_PATH LIB_FILE=libperf.a -LIB_H += ../../include/linux/perf_counter.h +LIB_H += ../../include/linux/perf_event.h +LIB_H += ../../include/linux/rbtree.h +LIB_H += ../../include/linux/list.h +LIB_H += util/include/linux/list.h LIB_H += perf.h LIB_H += util/types.h -LIB_H += util/list.h -LIB_H += util/rbtree.h LIB_H += util/levenshtein.h LIB_H += util/parse-options.h LIB_H += util/parse-events.h @@ -305,7 +336,9 @@ LIB_H += util/strlist.h LIB_H += util/run-command.h LIB_H += util/sigchain.h LIB_H += util/symbol.h +LIB_H += util/module.h LIB_H += util/color.h +LIB_H += util/values.h LIB_OBJS += util/abspath.o LIB_OBJS += util/alias.o @@ -328,21 +361,32 @@ LIB_OBJS += util/usage.o LIB_OBJS += util/wrapper.o LIB_OBJS += util/sigchain.o LIB_OBJS += util/symbol.o +LIB_OBJS += util/module.o LIB_OBJS += util/color.o LIB_OBJS += util/pager.o LIB_OBJS += util/header.o LIB_OBJS += util/callchain.o +LIB_OBJS += util/values.o +LIB_OBJS += util/debug.o +LIB_OBJS += util/map.o +LIB_OBJS += util/thread.o +LIB_OBJS += util/trace-event-parse.o +LIB_OBJS += util/trace-event-read.o +LIB_OBJS += util/trace-event-info.o +LIB_OBJS += util/svghelper.o BUILTIN_OBJS += builtin-annotate.o BUILTIN_OBJS += builtin-help.o +BUILTIN_OBJS += builtin-sched.o BUILTIN_OBJS += builtin-list.o BUILTIN_OBJS += builtin-record.o BUILTIN_OBJS += builtin-report.o BUILTIN_OBJS += builtin-stat.o +BUILTIN_OBJS += builtin-timechart.o BUILTIN_OBJS += builtin-top.o +BUILTIN_OBJS += builtin-trace.o PERFLIBS = $(LIB_FILE) -EXTLIBS = # # Platform specific tweaks @@ -371,6 +415,39 @@ ifeq ($(uname_S),Darwin) PTHREAD_LIBS = endif +ifneq ($(shell sh -c "(echo '\#include <libelf.h>'; echo 'int main(void) { Elf * elf = elf_begin(0, ELF_C_READ_MMAP, 0); return (long)elf; }') | $(CC) -x c - $(ALL_CFLAGS) -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -o /dev/null $(ALL_LDFLAGS) > /dev/null 2>&1 && echo y"), y) + msg := $(error No libelf.h/libelf found, please install libelf-dev/elfutils-libelf-devel); +endif + +ifdef NO_DEMANGLE + BASIC_CFLAGS += -DNO_DEMANGLE +else + has_bfd := $(shell sh -c "(echo '\#include <bfd.h>'; echo 'int main(void) { bfd_demangle(0, 0, 0); return 0; }') | $(CC) -x c - $(ALL_CFLAGS) -o /dev/null $(ALL_LDFLAGS) -lbfd > /dev/null 2>&1 && echo y") + + ifeq ($(has_bfd),y) + EXTLIBS += -lbfd + else + has_bfd_iberty := $(shell sh -c "(echo '\#include <bfd.h>'; echo 'int main(void) { bfd_demangle(0, 0, 0); return 0; }') | $(CC) -x c - $(ALL_CFLAGS) -o /dev/null $(ALL_LDFLAGS) -lbfd -liberty > /dev/null 2>&1 && echo y") + ifeq ($(has_bfd_iberty),y) + EXTLIBS += -lbfd -liberty + else + has_bfd_iberty_z := $(shell sh -c "(echo '\#include <bfd.h>'; echo 'int main(void) { bfd_demangle(0, 0, 0); return 0; }') | $(CC) -x c - $(ALL_CFLAGS) -o /dev/null $(ALL_LDFLAGS) -lbfd -liberty -lz > /dev/null 2>&1 && echo y") + ifeq ($(has_bfd_iberty_z),y) + EXTLIBS += -lbfd -liberty -lz + else + has_cplus_demangle := $(shell sh -c "(echo 'extern char *cplus_demangle(const char *, int);'; echo 'int main(void) { cplus_demangle(0, 0); return 0; }') | $(CC) -x c - $(ALL_CFLAGS) -o /dev/null $(ALL_LDFLAGS) -liberty > /dev/null 2>&1 && echo y") + ifeq ($(has_cplus_demangle),y) + EXTLIBS += -liberty + BASIC_CFLAGS += -DHAVE_CPLUS_DEMANGLE + else + msg := $(warning No bfd.h/libbfd found, install binutils-dev[el] to gain symbol demangling) + BASIC_CFLAGS += -DNO_DEMANGLE + endif + endif + endif + endif +endif + ifndef CC_LD_DYNPATH ifdef NO_R_TO_GCC_LINKER # Some gcc does not accept and pass -R to the linker to specify @@ -381,12 +458,6 @@ ifndef CC_LD_DYNPATH endif endif -ifdef ZLIB_PATH - BASIC_CFLAGS += -I$(ZLIB_PATH)/include - EXTLIBS += -L$(ZLIB_PATH)/$(lib) $(CC_LD_DYNPATH)$(ZLIB_PATH)/$(lib) -endif -EXTLIBS += -lz - ifdef NEEDS_SOCKET EXTLIBS += -lsocket endif @@ -642,6 +713,12 @@ builtin-help.o: builtin-help.c common-cmds.h PERF-CFLAGS '-DPERF_MAN_PATH="$(mandir_SQ)"' \ '-DPERF_INFO_PATH="$(infodir_SQ)"' $< +builtin-timechart.o: builtin-timechart.c common-cmds.h PERF-CFLAGS + $(QUIET_CC)$(CC) -o $*.o -c $(ALL_CFLAGS) \ + '-DPERF_HTML_PATH="$(htmldir_SQ)"' \ + '-DPERF_MAN_PATH="$(mandir_SQ)"' \ + '-DPERF_INFO_PATH="$(infodir_SQ)"' $< + $(BUILT_INS): perf$X $(QUIET_BUILT_IN)$(RM) $@ && \ ln perf$X $@ 2>/dev/null || \ @@ -697,6 +774,9 @@ builtin-init-db.o: builtin-init-db.c PERF-CFLAGS util/config.o: util/config.c PERF-CFLAGS $(QUIET_CC)$(CC) -o $*.o -c $(ALL_CFLAGS) -DETC_PERFCONFIG='"$(ETC_PERFCONFIG_SQ)"' $< +util/rbtree.o: ../../lib/rbtree.c PERF-CFLAGS + $(QUIET_CC)$(CC) -o util/rbtree.o -c $(ALL_CFLAGS) -DETC_PERFCONFIG='"$(ETC_PERFCONFIG_SQ)"' $< + perf-%$X: %.o $(PERFLIBS) $(QUIET_LINK)$(CC) $(ALL_CFLAGS) -o $@ $(ALL_LDFLAGS) $(filter %.o,$^) $(LIBS) diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index 722c0f54e54..1ec74161581 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -10,85 +10,37 @@ #include "util/util.h" #include "util/color.h" -#include "util/list.h" +#include <linux/list.h> #include "util/cache.h" -#include "util/rbtree.h" +#include <linux/rbtree.h> #include "util/symbol.h" #include "util/string.h" #include "perf.h" +#include "util/debug.h" #include "util/parse-options.h" #include "util/parse-events.h" - -#define SHOW_KERNEL 1 -#define SHOW_USER 2 -#define SHOW_HV 4 - -#define MIN_GREEN 0.5 -#define MIN_RED 5.0 - +#include "util/thread.h" static char const *input_name = "perf.data"; -static char *vmlinux = "vmlinux"; static char default_sort_order[] = "comm,symbol"; static char *sort_order = default_sort_order; +static int force; static int input; static int show_mask = SHOW_KERNEL | SHOW_USER | SHOW_HV; -static int dump_trace = 0; -#define dprintf(x...) do { if (dump_trace) printf(x); } while (0) - -static int verbose; +static int full_paths; static int print_line; static unsigned long page_size; static unsigned long mmap_window = 32; -struct ip_event { - struct perf_event_header header; - u64 ip; - u32 pid, tid; -}; - -struct mmap_event { - struct perf_event_header header; - u32 pid, tid; - u64 start; - u64 len; - u64 pgoff; - char filename[PATH_MAX]; -}; - -struct comm_event { - struct perf_event_header header; - u32 pid, tid; - char comm[16]; -}; - -struct fork_event { - struct perf_event_header header; - u32 pid, ppid; -}; - -struct period_event { - struct perf_event_header header; - u64 time; - u64 id; - u64 sample_period; -}; - -typedef union event_union { - struct perf_event_header header; - struct ip_event ip; - struct mmap_event mmap; - struct comm_event comm; - struct fork_event fork; - struct period_event period; -} event_t; +static struct rb_root threads; +static struct thread *last_match; struct sym_ext { @@ -97,323 +49,6 @@ struct sym_ext { char *path; }; -static LIST_HEAD(dsos); -static struct dso *kernel_dso; -static struct dso *vdso; - - -static void dsos__add(struct dso *dso) -{ - list_add_tail(&dso->node, &dsos); -} - -static struct dso *dsos__find(const char *name) -{ - struct dso *pos; - - list_for_each_entry(pos, &dsos, node) - if (strcmp(pos->name, name) == 0) - return pos; - return NULL; -} - -static struct dso *dsos__findnew(const char *name) -{ - struct dso *dso = dsos__find(name); - int nr; - - if (dso) - return dso; - - dso = dso__new(name, 0); - if (!dso) - goto out_delete_dso; - - nr = dso__load(dso, NULL, verbose); - if (nr < 0) { - if (verbose) - fprintf(stderr, "Failed to open: %s\n", name); - goto out_delete_dso; - } - if (!nr && verbose) { - fprintf(stderr, - "No symbols found in: %s, maybe install a debug package?\n", - name); - } - - dsos__add(dso); - - return dso; - -out_delete_dso: - dso__delete(dso); - return NULL; -} - -static void dsos__fprintf(FILE *fp) -{ - struct dso *pos; - - list_for_each_entry(pos, &dsos, node) - dso__fprintf(pos, fp); -} - -static struct symbol *vdso__find_symbol(struct dso *dso, u64 ip) -{ - return dso__find_symbol(kernel_dso, ip); -} - -static int load_kernel(void) -{ - int err; - - kernel_dso = dso__new("[kernel]", 0); - if (!kernel_dso) - return -1; - - err = dso__load_kernel(kernel_dso, vmlinux, NULL, verbose); - if (err) { - dso__delete(kernel_dso); - kernel_dso = NULL; - } else - dsos__add(kernel_dso); - - vdso = dso__new("[vdso]", 0); - if (!vdso) - return -1; - - vdso->find_symbol = vdso__find_symbol; - - dsos__add(vdso); - - return err; -} - -struct map { - struct list_head node; - u64 start; - u64 end; - u64 pgoff; - u64 (*map_ip)(struct map *, u64); - struct dso *dso; -}; - -static u64 map__map_ip(struct map *map, u64 ip) -{ - return ip - map->start + map->pgoff; -} - -static u64 vdso__map_ip(struct map *map, u64 ip) -{ - return ip; -} - -static struct map *map__new(struct mmap_event *event) -{ - struct map *self = malloc(sizeof(*self)); - - if (self != NULL) { - const char *filename = event->filename; - - self->start = event->start; - self->end = event->start + event->len; - self->pgoff = event->pgoff; - - self->dso = dsos__findnew(filename); - if (self->dso == NULL) - goto out_delete; - - if (self->dso == vdso) - self->map_ip = vdso__map_ip; - else - self->map_ip = map__map_ip; - } - return self; -out_delete: - free(self); - return NULL; -} - -static struct map *map__clone(struct map *self) -{ - struct map *map = malloc(sizeof(*self)); - - if (!map) - return NULL; - - memcpy(map, self, sizeof(*self)); - - return map; -} - -static int map__overlap(struct map *l, struct map *r) -{ - if (l->start > r->start) { - struct map *t = l; - l = r; - r = t; - } - - if (l->end > r->start) - return 1; - - return 0; -} - -static size_t map__fprintf(struct map *self, FILE *fp) -{ - return fprintf(fp, " %Lx-%Lx %Lx %s\n", - self->start, self->end, self->pgoff, self->dso->name); -} - - -struct thread { - struct rb_node rb_node; - struct list_head maps; - pid_t pid; - char *comm; -}; - -static struct thread *thread__new(pid_t pid) -{ - struct thread *self = malloc(sizeof(*self)); - - if (self != NULL) { - self->pid = pid; - self->comm = malloc(32); - if (self->comm) - snprintf(self->comm, 32, ":%d", self->pid); - INIT_LIST_HEAD(&self->maps); - } - - return self; -} - -static int thread__set_comm(struct thread *self, const char *comm) -{ - if (self->comm) - free(self->comm); - self->comm = strdup(comm); - return self->comm ? 0 : -ENOMEM; -} - -static size_t thread__fprintf(struct thread *self, FILE *fp) -{ - struct map *pos; - size_t ret = fprintf(fp, "Thread %d %s\n", self->pid, self->comm); - - list_for_each_entry(pos, &self->maps, node) - ret += map__fprintf(pos, fp); - - return ret; -} - - -static struct rb_root threads; -static struct thread *last_match; - -static struct thread *threads__findnew(pid_t pid) -{ - struct rb_node **p = &threads.rb_node; - struct rb_node *parent = NULL; - struct thread *th; - - /* - * Font-end cache - PID lookups come in blocks, - * so most of the time we dont have to look up - * the full rbtree: - */ - if (last_match && last_match->pid == pid) - return last_match; - - while (*p != NULL) { - parent = *p; - th = rb_entry(parent, struct thread, rb_node); - - if (th->pid == pid) { - last_match = th; - return th; - } - - if (pid < th->pid) - p = &(*p)->rb_left; - else - p = &(*p)->rb_right; - } - - th = thread__new(pid); - if (th != NULL) { - rb_link_node(&th->rb_node, parent, p); - rb_insert_color(&th->rb_node, &threads); - last_match = th; - } - - return th; -} - -static void thread__insert_map(struct thread *self, struct map *map) -{ - struct map *pos, *tmp; - - list_for_each_entry_safe(pos, tmp, &self->maps, node) { - if (map__overlap(pos, map)) { - list_del_init(&pos->node); - /* XXX leaks dsos */ - free(pos); - } - } - - list_add_tail(&map->node, &self->maps); -} - -static int thread__fork(struct thread *self, struct thread *parent) -{ - struct map *map; - - if (self->comm) - free(self->comm); - self->comm = strdup(parent->comm); - if (!self->comm) - return -ENOMEM; - - list_for_each_entry(map, &parent->maps, node) { - struct map *new = map__clone(map); - if (!new) - return -ENOMEM; - thread__insert_map(self, new); - } - - return 0; -} - -static struct map *thread__find_map(struct thread *self, u64 ip) -{ - struct map *pos; - - if (self == NULL) - return NULL; - - list_for_each_entry(pos, &self->maps, node) - if (ip >= pos->start && ip <= pos->end) - return pos; - - return NULL; -} - -static size_t threads__fprintf(FILE *fp) -{ - size_t ret = 0; - struct rb_node *nd; - - for (nd = rb_first(&threads); nd; nd = rb_next(nd)) { - struct thread *pos = rb_entry(nd, struct thread, rb_node); - - ret += thread__fprintf(pos, fp); - } - - return ret; -} - /* * histogram, sorted on item, collects counts */ @@ -440,7 +75,7 @@ struct hist_entry { struct sort_entry { struct list_head list; - char *header; + const char *header; int64_t (*cmp)(struct hist_entry *, struct hist_entry *); int64_t (*collapse)(struct hist_entry *, struct hist_entry *); @@ -584,7 +219,7 @@ static struct sort_entry sort_sym = { static int sort__need_collapse = 0; struct sort_dimension { - char *name; + const char *name; struct sort_entry *entry; int taken; }; @@ -600,7 +235,7 @@ static LIST_HEAD(hist_entry__sort_list); static int sort_dimension__add(char *tok) { - int i; + unsigned int i; for (i = 0; i < ARRAY_SIZE(sort_dimensions); i++) { struct sort_dimension *sd = &sort_dimensions[i]; @@ -837,17 +472,6 @@ static void output__resort(void) } } -static void register_idle_thread(void) -{ - struct thread *thread = threads__findnew(0); - - if (thread == NULL || - thread__set_comm(thread, "[idle]")) { - fprintf(stderr, "problem inserting idle task.\n"); - exit(-1); - } -} - static unsigned long total = 0, total_mmap = 0, total_comm = 0, @@ -860,18 +484,20 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) char level; int show = 0; struct dso *dso = NULL; - struct thread *thread = threads__findnew(event->ip.pid); + struct thread *thread; u64 ip = event->ip.ip; struct map *map = NULL; - dprintf("%p [%p]: PERF_EVENT (IP, %d): %d: %p\n", + thread = threads__findnew(event->ip.pid, &threads, &last_match); + + dump_printf("%p [%p]: PERF_EVENT (IP, %d): %d: %p\n", (void *)(offset + head), (void *)(long)(event->header.size), event->header.misc, event->ip.pid, (void *)(long)ip); - dprintf(" ... thread: %s:%d\n", thread->comm, thread->pid); + dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid); if (thread == NULL) { fprintf(stderr, "problem processing %d event, skipping it.\n", @@ -879,15 +505,15 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) return -1; } - if (event->header.misc & PERF_EVENT_MISC_KERNEL) { + if (event->header.misc & PERF_RECORD_MISC_KERNEL) { show = SHOW_KERNEL; level = 'k'; dso = kernel_dso; - dprintf(" ...... dso: %s\n", dso->name); + dump_printf(" ...... dso: %s\n", dso->name); - } else if (event->header.misc & PERF_EVENT_MISC_USER) { + } else if (event->header.misc & PERF_RECORD_MISC_USER) { show = SHOW_USER; level = '.'; @@ -906,12 +532,12 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) if ((long long)ip < 0) dso = kernel_dso; } - dprintf(" ...... dso: %s\n", dso ? dso->name : "<not found>"); + dump_printf(" ...... dso: %s\n", dso ? dso->name : "<not found>"); } else { show = SHOW_HV; level = 'H'; - dprintf(" ...... dso: [hypervisor]\n"); + dump_printf(" ...... dso: [hypervisor]\n"); } if (show & show_mask) { @@ -934,10 +560,12 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) static int process_mmap_event(event_t *event, unsigned long offset, unsigned long head) { - struct thread *thread = threads__findnew(event->mmap.pid); - struct map *map = map__new(&event->mmap); + struct thread *thread; + struct map *map = map__new(&event->mmap, NULL, 0); + + thread = threads__findnew(event->mmap.pid, &threads, &last_match); - dprintf("%p [%p]: PERF_EVENT_MMAP %d: [%p(%p) @ %p]: %s\n", + dump_printf("%p [%p]: PERF_RECORD_MMAP %d: [%p(%p) @ %p]: %s\n", (void *)(offset + head), (void *)(long)(event->header.size), event->mmap.pid, @@ -947,7 +575,7 @@ process_mmap_event(event_t *event, unsigned long offset, unsigned long head) event->mmap.filename); if (thread == NULL || map == NULL) { - dprintf("problem processing PERF_EVENT_MMAP, skipping event.\n"); + dump_printf("problem processing PERF_RECORD_MMAP, skipping event.\n"); return 0; } @@ -960,16 +588,17 @@ process_mmap_event(event_t *event, unsigned long offset, unsigned long head) static int process_comm_event(event_t *event, unsigned long offset, unsigned long head) { - struct thread *thread = threads__findnew(event->comm.pid); + struct thread *thread; - dprintf("%p [%p]: PERF_EVENT_COMM: %s:%d\n", + thread = threads__findnew(event->comm.pid, &threads, &last_match); + dump_printf("%p [%p]: PERF_RECORD_COMM: %s:%d\n", (void *)(offset + head), (void *)(long)(event->header.size), event->comm.comm, event->comm.pid); if (thread == NULL || thread__set_comm(thread, event->comm.comm)) { - dprintf("problem processing PERF_EVENT_COMM, skipping event.\n"); + dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n"); return -1; } total_comm++; @@ -980,16 +609,25 @@ process_comm_event(event_t *event, unsigned long offset, unsigned long head) static int process_fork_event(event_t *event, unsigned long offset, unsigned long head) { - struct thread *thread = threads__findnew(event->fork.pid); - struct thread *parent = threads__findnew(event->fork.ppid); + struct thread *thread; + struct thread *parent; - dprintf("%p [%p]: PERF_EVENT_FORK: %d:%d\n", + thread = threads__findnew(event->fork.pid, &threads, &last_match); + parent = threads__findnew(event->fork.ppid, &threads, &last_match); + dump_printf("%p [%p]: PERF_RECORD_FORK: %d:%d\n", (void *)(offset + head), (void *)(long)(event->header.size), event->fork.pid, event->fork.ppid); + /* + * A thread clone will have the same PID for both + * parent and child. + */ + if (thread == parent) + return 0; + if (!thread || !parent || thread__fork(thread, parent)) { - dprintf("problem processing PERF_EVENT_FORK, skipping event.\n"); + dump_printf("problem processing PERF_RECORD_FORK, skipping event.\n"); return -1; } total_fork++; @@ -998,42 +636,26 @@ process_fork_event(event_t *event, unsigned long offset, unsigned long head) } static int -process_period_event(event_t *event, unsigned long offset, unsigned long head) -{ - dprintf("%p [%p]: PERF_EVENT_PERIOD: time:%Ld, id:%Ld: period:%Ld\n", - (void *)(offset + head), - (void *)(long)(event->header.size), - event->period.time, - event->period.id, - event->period.sample_period); - - return 0; -} - -static int process_event(event_t *event, unsigned long offset, unsigned long head) { switch (event->header.type) { - case PERF_EVENT_SAMPLE: + case PERF_RECORD_SAMPLE: return process_sample_event(event, offset, head); - case PERF_EVENT_MMAP: + case PERF_RECORD_MMAP: return process_mmap_event(event, offset, head); - case PERF_EVENT_COMM: + case PERF_RECORD_COMM: return process_comm_event(event, offset, head); - case PERF_EVENT_FORK: + case PERF_RECORD_FORK: return process_fork_event(event, offset, head); - - case PERF_EVENT_PERIOD: - return process_period_event(event, offset, head); /* * We dont process them right now but they are fine: */ - case PERF_EVENT_THROTTLE: - case PERF_EVENT_UNTHROTTLE: + case PERF_RECORD_THROTTLE: + case PERF_RECORD_UNTHROTTLE: return 0; default: @@ -1043,24 +665,6 @@ process_event(event_t *event, unsigned long offset, unsigned long head) return 0; } -static char *get_color(double percent) -{ - char *color = PERF_COLOR_NORMAL; - - /* - * We color high-overhead entries in red, mid-overhead - * entries in green - and keep the low overhead places - * normal: - */ - if (percent >= MIN_RED) - color = PERF_COLOR_RED; - else { - if (percent > MIN_GREEN) - color = PERF_COLOR_GREEN; - } - return color; -} - static int parse_line(FILE *file, struct symbol *sym, u64 start, u64 len) { @@ -1069,7 +673,7 @@ parse_line(FILE *file, struct symbol *sym, u64 start, u64 len) static const char *prev_color; unsigned int offset; size_t line_len; - u64 line_ip; + s64 line_ip; int ret; char *c; @@ -1109,7 +713,7 @@ parse_line(FILE *file, struct symbol *sym, u64 start, u64 len) const char *path = NULL; unsigned int hits = 0; double percent = 0.0; - char *color; + const char *color; struct sym_ext *sym_ext = sym->priv; offset = line_ip - start; @@ -1122,7 +726,7 @@ parse_line(FILE *file, struct symbol *sym, u64 start, u64 len) } else if (sym->hist_sum) percent = 100.0 * hits / sym->hist_sum; - color = get_color(percent); + color = get_percent_color(percent); /* * Also color the filename and line if needed, with @@ -1191,7 +795,7 @@ static void free_source_line(struct symbol *sym, int len) /* Get the filename:line for the colored entries */ static void -get_source_line(struct symbol *sym, u64 start, int len, char *filename) +get_source_line(struct symbol *sym, u64 start, int len, const char *filename) { int i; char cmd[PATH_MAX * 2]; @@ -1237,7 +841,7 @@ get_source_line(struct symbol *sym, u64 start, int len, char *filename) } } -static void print_summary(char *filename) +static void print_summary(const char *filename) { struct sym_ext *sym_ext; struct rb_node *node; @@ -1253,12 +857,12 @@ static void print_summary(char *filename) node = rb_first(&root_sym_ext); while (node) { double percent; - char *color; + const char *color; char *path; sym_ext = rb_entry(node, struct sym_ext, node); percent = sym_ext->percent; - color = get_color(percent); + color = get_percent_color(percent); path = sym_ext->path; color_fprintf(stdout, color, " %7.2f %s", percent, path); @@ -1268,19 +872,25 @@ static void print_summary(char *filename) static void annotate_sym(struct dso *dso, struct symbol *sym) { - char *filename = dso->name; + const char *filename = dso->name, *d_filename; u64 start, end, len; char command[PATH_MAX*2]; FILE *file; if (!filename) return; - if (dso == kernel_dso) - filename = vmlinux; + if (sym->module) + filename = sym->module->path; + else if (dso == kernel_dso) + filename = vmlinux_name; start = sym->obj_start; if (!start) start = sym->start; + if (full_paths) + d_filename = filename; + else + d_filename = basename(filename); end = start + sym->end - sym->start + 1; len = sym->end - sym->start; @@ -1291,13 +901,14 @@ static void annotate_sym(struct dso *dso, struct symbol *sym) } printf("\n\n------------------------------------------------\n"); - printf(" Percent | Source code & Disassembly of %s\n", filename); + printf(" Percent | Source code & Disassembly of %s\n", d_filename); printf("------------------------------------------------\n"); if (verbose >= 2) printf("annotating [%p] %30s : [%p] %30s\n", dso, dso->name, sym, sym->name); - sprintf(command, "objdump --start-address=0x%016Lx --stop-address=0x%016Lx -dS %s", (u64)start, (u64)end, filename); + sprintf(command, "objdump --start-address=0x%016Lx --stop-address=0x%016Lx -dS %s|grep -v %s", + (u64)start, (u64)end, filename, filename); if (verbose >= 3) printf("doing: %s\n", command); @@ -1343,12 +954,12 @@ static int __cmd_annotate(void) int ret, rc = EXIT_FAILURE; unsigned long offset = 0; unsigned long head = 0; - struct stat stat; + struct stat input_stat; event_t *event; uint32_t size; char *buf; - register_idle_thread(); + register_idle_thread(&threads, &last_match); input = open(input_name, O_RDONLY); if (input < 0) { @@ -1356,13 +967,18 @@ static int __cmd_annotate(void) exit(-1); } - ret = fstat(input, &stat); + ret = fstat(input, &input_stat); if (ret < 0) { perror("failed to stat file"); exit(-1); } - if (!stat.st_size) { + if (!force && input_stat.st_uid && (input_stat.st_uid != geteuid())) { + fprintf(stderr, "file: %s not owned by current user or root\n", input_name); + exit(-1); + } + + if (!input_stat.st_size) { fprintf(stderr, "zero-sized file, nothing to do!\n"); exit(0); } @@ -1389,10 +1005,10 @@ more: if (head + event->header.size >= page_size * mmap_window) { unsigned long shift = page_size * (head / page_size); - int ret; + int munmap_ret; - ret = munmap(buf, page_size * mmap_window); - assert(ret == 0); + munmap_ret = munmap(buf, page_size * mmap_window); + assert(munmap_ret == 0); offset += shift; head -= shift; @@ -1401,14 +1017,14 @@ more: size = event->header.size; - dprintf("%p [%p]: event: %d\n", + dump_printf("%p [%p]: event: %d\n", (void *)(offset + head), (void *)(long)event->header.size, event->header.type); if (!size || process_event(event, offset, head) < 0) { - dprintf("%p [%p]: skipping unknown header type: %d\n", + dump_printf("%p [%p]: skipping unknown header type: %d\n", (void *)(offset + head), (void *)(long)(event->header.size), event->header.type); @@ -1428,23 +1044,23 @@ more: head += size; - if (offset + head < stat.st_size) + if (offset + head < (unsigned long)input_stat.st_size) goto more; rc = EXIT_SUCCESS; close(input); - dprintf(" IP events: %10ld\n", total); - dprintf(" mmap events: %10ld\n", total_mmap); - dprintf(" comm events: %10ld\n", total_comm); - dprintf(" fork events: %10ld\n", total_fork); - dprintf(" unknown events: %10ld\n", total_unknown); + dump_printf(" IP events: %10ld\n", total); + dump_printf(" mmap events: %10ld\n", total_mmap); + dump_printf(" comm events: %10ld\n", total_comm); + dump_printf(" fork events: %10ld\n", total_fork); + dump_printf(" unknown events: %10ld\n", total_unknown); if (dump_trace) return 0; if (verbose >= 3) - threads__fprintf(stdout); + threads__fprintf(stdout, &threads); if (verbose >= 2) dsos__fprintf(stdout); @@ -1467,13 +1083,18 @@ static const struct option options[] = { "input file name"), OPT_STRING('s', "symbol", &sym_hist_filter, "symbol", "symbol to annotate"), + OPT_BOOLEAN('f', "force", &force, "don't complain, do it"), OPT_BOOLEAN('v', "verbose", &verbose, "be more verbose (show symbol address, etc)"), OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, "dump raw trace in ASCII"), - OPT_STRING('k', "vmlinux", &vmlinux, "file", "vmlinux pathname"), + OPT_STRING('k', "vmlinux", &vmlinux_name, "file", "vmlinux pathname"), + OPT_BOOLEAN('m', "modules", &modules, + "load module symbols - WARNING: use only with -k and LIVE kernel"), OPT_BOOLEAN('l', "print-line", &print_line, "print matching source lines (may be slow)"), + OPT_BOOLEAN('P', "full-paths", &full_paths, + "Don't shorten the displayed pathnames"), OPT_END() }; @@ -1492,7 +1113,7 @@ static void setup_sorting(void) free(str); } -int cmd_annotate(int argc, const char **argv, const char *prefix) +int cmd_annotate(int argc, const char **argv, const char *prefix __used) { symbol__init(); diff --git a/tools/perf/builtin-help.c b/tools/perf/builtin-help.c index 0f32dc3f3c4..4fb8734a796 100644 --- a/tools/perf/builtin-help.c +++ b/tools/perf/builtin-help.c @@ -3,6 +3,7 @@ * * Builtin help command */ +#include "perf.h" #include "util/cache.h" #include "builtin.h" #include "util/exec_cmd.h" @@ -277,7 +278,7 @@ static struct cmdnames main_cmds, other_cmds; void list_common_cmds_help(void) { - int i, longest = 0; + unsigned int i, longest = 0; for (i = 0; i < ARRAY_SIZE(common_cmds); i++) { if (longest < strlen(common_cmds[i].name)) @@ -415,9 +416,10 @@ static void show_html_page(const char *perf_cmd) open_html(page_path.buf); } -int cmd_help(int argc, const char **argv, const char *prefix) +int cmd_help(int argc, const char **argv, const char *prefix __used) { const char *alias; + load_command_list("perf-", &main_cmds, &other_cmds); perf_config(perf_help_config, NULL); @@ -454,6 +456,7 @@ int cmd_help(int argc, const char **argv, const char *prefix) break; case HELP_FORMAT_WEB: show_html_page(argv[0]); + default: break; } diff --git a/tools/perf/builtin-list.c b/tools/perf/builtin-list.c index fe60e37c96e..d88c6961274 100644 --- a/tools/perf/builtin-list.c +++ b/tools/perf/builtin-list.c @@ -10,11 +10,12 @@ #include "perf.h" -#include "util/parse-options.h" #include "util/parse-events.h" +#include "util/cache.h" -int cmd_list(int argc, const char **argv, const char *prefix) +int cmd_list(int argc __used, const char **argv __used, const char *prefix __used) { + setup_pager(); print_events(); return 0; } diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index d18546f37d7..a5a050af8e7 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -15,6 +15,9 @@ #include "util/string.h" #include "util/header.h" +#include "util/event.h" +#include "util/debug.h" +#include "util/trace-event.h" #include <unistd.h> #include <sched.h> @@ -34,15 +37,19 @@ static int output; static const char *output_name = "perf.data"; static int group = 0; static unsigned int realtime_prio = 0; +static int raw_samples = 0; static int system_wide = 0; +static int profile_cpu = -1; static pid_t target_pid = -1; static int inherit = 1; static int force = 0; static int append_file = 0; static int call_graph = 0; -static int verbose = 0; static int inherit_stat = 0; static int no_samples = 0; +static int sample_address = 0; +static int multiplex = 0; +static int multiplex_fd = -1; static long samples; static struct timeval last_read; @@ -59,24 +66,6 @@ static int file_new = 1; struct perf_header *header; -struct mmap_event { - struct perf_event_header header; - u32 pid; - u32 tid; - u64 start; - u64 len; - u64 pgoff; - char filename[PATH_MAX]; -}; - -struct comm_event { - struct perf_event_header header; - u32 pid; - u32 tid; - char comm[16]; -}; - - struct mmap_data { int counter; void *base; @@ -88,7 +77,7 @@ static struct mmap_data mmap_array[MAX_NR_CPUS][MAX_COUNTERS]; static unsigned long mmap_read_head(struct mmap_data *md) { - struct perf_counter_mmap_page *pc = md->base; + struct perf_event_mmap_page *pc = md->base; long head; head = pc->data_head; @@ -99,7 +88,7 @@ static unsigned long mmap_read_head(struct mmap_data *md) static void mmap_write_tail(struct mmap_data *md, unsigned long tail) { - struct perf_counter_mmap_page *pc = md->base; + struct perf_event_mmap_page *pc = md->base; /* * ensure all reads are done before we write the tail out. @@ -202,47 +191,49 @@ static void sig_atexit(void) kill(getpid(), signr); } -static void pid_synthesize_comm_event(pid_t pid, int full) +static pid_t pid_synthesize_comm_event(pid_t pid, int full) { struct comm_event comm_ev; char filename[PATH_MAX]; char bf[BUFSIZ]; - int fd; - size_t size; - char *field, *sep; + FILE *fp; + size_t size = 0; DIR *tasks; struct dirent dirent, *next; + pid_t tgid = 0; - snprintf(filename, sizeof(filename), "/proc/%d/stat", pid); + snprintf(filename, sizeof(filename), "/proc/%d/status", pid); - fd = open(filename, O_RDONLY); - if (fd < 0) { + fp = fopen(filename, "r"); + if (fp == NULL) { /* * We raced with a task exiting - just return: */ if (verbose) fprintf(stderr, "couldn't open %s\n", filename); - return; + return 0; } - if (read(fd, bf, sizeof(bf)) < 0) { - fprintf(stderr, "couldn't read %s\n", filename); - exit(EXIT_FAILURE); - } - close(fd); - /* 9027 (cat) R 6747 9027 6747 34816 9027 ... */ memset(&comm_ev, 0, sizeof(comm_ev)); - field = strchr(bf, '('); - if (field == NULL) - goto out_failure; - sep = strchr(++field, ')'); - if (sep == NULL) - goto out_failure; - size = sep - field; - memcpy(comm_ev.comm, field, size++); - - comm_ev.pid = pid; - comm_ev.header.type = PERF_EVENT_COMM; + while (!comm_ev.comm[0] || !comm_ev.pid) { + if (fgets(bf, sizeof(bf), fp) == NULL) + goto out_failure; + + if (memcmp(bf, "Name:", 5) == 0) { + char *name = bf + 5; + while (*name && isspace(*name)) + ++name; + size = strlen(name) - 1; + memcpy(comm_ev.comm, name, size++); + } else if (memcmp(bf, "Tgid:", 5) == 0) { + char *tgids = bf + 5; + while (*tgids && isspace(*tgids)) + ++tgids; + tgid = comm_ev.pid = atoi(tgids); + } + } + + comm_ev.header.type = PERF_RECORD_COMM; size = ALIGN(size, sizeof(u64)); comm_ev.header.size = sizeof(comm_ev) - (sizeof(comm_ev.comm) - size); @@ -250,7 +241,7 @@ static void pid_synthesize_comm_event(pid_t pid, int full) comm_ev.tid = pid; write_output(&comm_ev, comm_ev.header.size); - return; + goto out_fclose; } snprintf(filename, sizeof(filename), "/proc/%d/task", pid); @@ -267,7 +258,10 @@ static void pid_synthesize_comm_event(pid_t pid, int full) write_output(&comm_ev, comm_ev.header.size); } closedir(tasks); - return; + +out_fclose: + fclose(fp); + return tgid; out_failure: fprintf(stderr, "couldn't get COMM and pgid, malformed %s\n", @@ -275,7 +269,7 @@ out_failure: exit(EXIT_FAILURE); } -static void pid_synthesize_mmap_samples(pid_t pid) +static void pid_synthesize_mmap_samples(pid_t pid, pid_t tgid) { char filename[PATH_MAX]; FILE *fp; @@ -294,7 +288,7 @@ static void pid_synthesize_mmap_samples(pid_t pid) while (1) { char bf[BUFSIZ], *pbf = bf; struct mmap_event mmap_ev = { - .header.type = PERF_EVENT_MMAP, + .header = { .type = PERF_RECORD_MMAP }, }; int n; size_t size; @@ -313,6 +307,10 @@ static void pid_synthesize_mmap_samples(pid_t pid) if (*pbf == 'x') { /* vm_exec */ char *execname = strchr(bf, '/'); + /* Catch VDSO */ + if (execname == NULL) + execname = strstr(bf, "[vdso]"); + if (execname == NULL) continue; @@ -323,7 +321,7 @@ static void pid_synthesize_mmap_samples(pid_t pid) mmap_ev.len -= mmap_ev.start; mmap_ev.header.size = (sizeof(mmap_ev) - (sizeof(mmap_ev.filename) - size)); - mmap_ev.pid = pid; + mmap_ev.pid = tgid; mmap_ev.tid = pid; write_output(&mmap_ev, mmap_ev.header.size); @@ -342,14 +340,14 @@ static void synthesize_all(void) while (!readdir_r(proc, &dirent, &next) && next) { char *end; - pid_t pid; + pid_t pid, tgid; pid = strtol(dirent.d_name, &end, 10); if (*end) /* only interested in proper numerical dirents */ continue; - pid_synthesize_comm_event(pid, 1); - pid_synthesize_mmap_samples(pid); + tgid = pid_synthesize_comm_event(pid, 1); + pid_synthesize_mmap_samples(pid, tgid); } closedir(proc); @@ -357,7 +355,7 @@ static void synthesize_all(void) static int group_fd; -static struct perf_header_attr *get_header_attr(struct perf_counter_attr *a, int nr) +static struct perf_header_attr *get_header_attr(struct perf_event_attr *a, int nr) { struct perf_header_attr *h_attr; @@ -373,7 +371,7 @@ static struct perf_header_attr *get_header_attr(struct perf_counter_attr *a, int static void create_counter(int counter, int cpu, pid_t pid) { - struct perf_counter_attr *attr = attrs + counter; + struct perf_event_attr *attr = attrs + counter; struct perf_header_attr *h_attr; int track = !counter; /* only the first counter needs these */ struct { @@ -387,7 +385,7 @@ static void create_counter(int counter, int cpu, pid_t pid) PERF_FORMAT_TOTAL_TIME_RUNNING | PERF_FORMAT_ID; - attr->sample_type = PERF_SAMPLE_IP | PERF_SAMPLE_TID; + attr->sample_type |= PERF_SAMPLE_IP | PERF_SAMPLE_TID; if (freq) { attr->sample_type |= PERF_SAMPLE_PERIOD; @@ -401,22 +399,33 @@ static void create_counter(int counter, int cpu, pid_t pid) if (inherit_stat) attr->inherit_stat = 1; + if (sample_address) + attr->sample_type |= PERF_SAMPLE_ADDR; + if (call_graph) attr->sample_type |= PERF_SAMPLE_CALLCHAIN; + if (raw_samples) { + attr->sample_type |= PERF_SAMPLE_TIME; + attr->sample_type |= PERF_SAMPLE_RAW; + attr->sample_type |= PERF_SAMPLE_CPU; + } + attr->mmap = track; attr->comm = track; attr->inherit = (cpu < 0) && inherit; attr->disabled = 1; try_again: - fd[nr_cpu][counter] = sys_perf_counter_open(attr, pid, cpu, group_fd, 0); + fd[nr_cpu][counter] = sys_perf_event_open(attr, pid, cpu, group_fd, 0); if (fd[nr_cpu][counter] < 0) { int err = errno; if (err == EPERM) die("Permission error - are you root?\n"); + else if (err == ENODEV && profile_cpu != -1) + die("No such device - did you specify an out-of-range profile CPU?\n"); /* * If it's cycles then fall back to hrtimer @@ -435,7 +444,7 @@ try_again: printf("\n"); error("perfcounter syscall returned with %d (%s)\n", fd[nr_cpu][counter], strerror(err)); - die("No CONFIG_PERF_COUNTERS=y kernel support configured?\n"); + die("No CONFIG_PERF_EVENTS=y kernel support configured?\n"); exit(-1); } @@ -463,22 +472,31 @@ try_again: */ if (group && group_fd == -1) group_fd = fd[nr_cpu][counter]; + if (multiplex && multiplex_fd == -1) + multiplex_fd = fd[nr_cpu][counter]; - event_array[nr_poll].fd = fd[nr_cpu][counter]; - event_array[nr_poll].events = POLLIN; - nr_poll++; - - mmap_array[nr_cpu][counter].counter = counter; - mmap_array[nr_cpu][counter].prev = 0; - mmap_array[nr_cpu][counter].mask = mmap_pages*page_size - 1; - mmap_array[nr_cpu][counter].base = mmap(NULL, (mmap_pages+1)*page_size, - PROT_READ|PROT_WRITE, MAP_SHARED, fd[nr_cpu][counter], 0); - if (mmap_array[nr_cpu][counter].base == MAP_FAILED) { - error("failed to mmap with %d (%s)\n", errno, strerror(errno)); - exit(-1); + if (multiplex && fd[nr_cpu][counter] != multiplex_fd) { + int ret; + + ret = ioctl(fd[nr_cpu][counter], PERF_EVENT_IOC_SET_OUTPUT, multiplex_fd); + assert(ret != -1); + } else { + event_array[nr_poll].fd = fd[nr_cpu][counter]; + event_array[nr_poll].events = POLLIN; + nr_poll++; + + mmap_array[nr_cpu][counter].counter = counter; + mmap_array[nr_cpu][counter].prev = 0; + mmap_array[nr_cpu][counter].mask = mmap_pages*page_size - 1; + mmap_array[nr_cpu][counter].base = mmap(NULL, (mmap_pages+1)*page_size, + PROT_READ|PROT_WRITE, MAP_SHARED, fd[nr_cpu][counter], 0); + if (mmap_array[nr_cpu][counter].base == MAP_FAILED) { + error("failed to mmap with %d (%s)\n", errno, strerror(errno)); + exit(-1); + } } - ioctl(fd[nr_cpu][counter], PERF_COUNTER_IOC_ENABLE); + ioctl(fd[nr_cpu][counter], PERF_EVENT_IOC_ENABLE); } static void open_counters(int cpu, pid_t pid) @@ -506,6 +524,7 @@ static int __cmd_record(int argc, const char **argv) pid_t pid = 0; int flags; int ret; + unsigned long waking = 0; page_size = sysconf(_SC_PAGE_SIZE); nr_cpus = sysconf(_SC_NPROCESSORS_ONLN); @@ -516,10 +535,14 @@ static int __cmd_record(int argc, const char **argv) signal(SIGCHLD, sig_handler); signal(SIGINT, sig_handler); - if (!stat(output_name, &st) && !force && !append_file) { - fprintf(stderr, "Error, output file %s exists, use -A to append or -f to overwrite.\n", - output_name); - exit(-1); + if (!stat(output_name, &st) && st.st_size) { + if (!force && !append_file) { + fprintf(stderr, "Error, output file %s exists, use -A to append or -f to overwrite.\n", + output_name); + exit(-1); + } + } else { + append_file = 0; } flags = O_CREAT|O_RDWR; @@ -539,6 +562,17 @@ static int __cmd_record(int argc, const char **argv) else header = perf_header__new(); + + if (raw_samples) { + read_tracing_data(attrs, nr_counters); + } else { + for (i = 0; i < nr_counters; i++) { + if (attrs[i].sample_type & PERF_SAMPLE_RAW) { + read_tracing_data(attrs, nr_counters); + break; + } + } + } atexit(atexit_header); if (!system_wide) { @@ -546,16 +580,22 @@ static int __cmd_record(int argc, const char **argv) if (pid == -1) pid = getpid(); - open_counters(-1, pid); - } else for (i = 0; i < nr_cpus; i++) - open_counters(i, target_pid); + open_counters(profile_cpu, pid); + } else { + if (profile_cpu != -1) { + open_counters(profile_cpu, target_pid); + } else { + for (i = 0; i < nr_cpus; i++) + open_counters(i, target_pid); + } + } if (file_new) perf_header__write(header, output); if (!system_wide) { - pid_synthesize_comm_event(pid, 0); - pid_synthesize_mmap_samples(pid); + pid_t tgid = pid_synthesize_comm_event(pid, 0); + pid_synthesize_mmap_samples(pid, tgid); } else synthesize_all(); @@ -586,17 +626,29 @@ static int __cmd_record(int argc, const char **argv) int hits = samples; for (i = 0; i < nr_cpu; i++) { - for (counter = 0; counter < nr_counters; counter++) - mmap_read(&mmap_array[i][counter]); + for (counter = 0; counter < nr_counters; counter++) { + if (mmap_array[i][counter].base) + mmap_read(&mmap_array[i][counter]); + } } if (hits == samples) { if (done) break; - ret = poll(event_array, nr_poll, 100); + ret = poll(event_array, nr_poll, -1); + waking++; + } + + if (done) { + for (i = 0; i < nr_cpu; i++) { + for (counter = 0; counter < nr_counters; counter++) + ioctl(fd[i][counter], PERF_EVENT_IOC_DISABLE); + } } } + fprintf(stderr, "[ perf record: Woken up %ld times to write data ]\n", waking); + /* * Approximate RIP event size: 24 bytes. */ @@ -623,10 +675,14 @@ static const struct option options[] = { "record events on existing pid"), OPT_INTEGER('r', "realtime", &realtime_prio, "collect data with this RT SCHED_FIFO priority"), + OPT_BOOLEAN('R', "raw-samples", &raw_samples, + "collect raw sample records from all opened counters"), OPT_BOOLEAN('a', "all-cpus", &system_wide, "system-wide collection from all CPUs"), OPT_BOOLEAN('A', "append", &append_file, "append to the output file to do incremental profiling"), + OPT_INTEGER('C', "profile_cpu", &profile_cpu, + "CPU to profile on"), OPT_BOOLEAN('f', "force", &force, "overwrite existing data file"), OPT_LONG('c', "count", &default_interval, @@ -645,16 +701,21 @@ static const struct option options[] = { "be more verbose (show counter open errors, etc)"), OPT_BOOLEAN('s', "stat", &inherit_stat, "per thread counts"), + OPT_BOOLEAN('d', "data", &sample_address, + "Sample addresses"), OPT_BOOLEAN('n', "no-samples", &no_samples, "don't sample"), + OPT_BOOLEAN('M', "multiplex", &multiplex, + "multiplex counter output in a single channel"), OPT_END() }; -int cmd_record(int argc, const char **argv, const char *prefix) +int cmd_record(int argc, const char **argv, const char *prefix __used) { int counter; - argc = parse_options(argc, argv, options, record_usage, 0); + argc = parse_options(argc, argv, options, record_usage, + PARSE_OPT_STOP_AT_NON_OPTION); if (!argc && target_pid == -1 && !system_wide) usage_with_options(record_usage, options); diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index 135b7837e6b..19669c20088 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -10,43 +10,45 @@ #include "util/util.h" #include "util/color.h" -#include "util/list.h" +#include <linux/list.h> #include "util/cache.h" -#include "util/rbtree.h" +#include <linux/rbtree.h> #include "util/symbol.h" #include "util/string.h" #include "util/callchain.h" #include "util/strlist.h" +#include "util/values.h" #include "perf.h" +#include "util/debug.h" #include "util/header.h" #include "util/parse-options.h" #include "util/parse-events.h" -#define SHOW_KERNEL 1 -#define SHOW_USER 2 -#define SHOW_HV 4 +#include "util/thread.h" static char const *input_name = "perf.data"; -static char *vmlinux = NULL; -static char default_sort_order[] = "comm,dso"; +static char default_sort_order[] = "comm,dso,symbol"; static char *sort_order = default_sort_order; -static char *dso_list_str, *comm_list_str, *sym_list_str; +static char *dso_list_str, *comm_list_str, *sym_list_str, + *col_width_list_str; static struct strlist *dso_list, *comm_list, *sym_list; +static char *field_sep; +static int force; static int input; static int show_mask = SHOW_KERNEL | SHOW_USER | SHOW_HV; -static int dump_trace = 0; -#define dprintf(x...) do { if (dump_trace) printf(x); } while (0) -#define cdprintf(x...) do { if (dump_trace) color_fprintf(stdout, color, x); } while (0) +static int full_paths; +static int show_nr_samples; -static int verbose; -#define eprintf(x...) do { if (verbose) fprintf(stderr, x); } while (0) +static int show_threads; +static struct perf_read_values show_threads_values; -static int full_paths; +static char default_pretty_printing_style[] = "normal"; +static char *pretty_printing_style = default_pretty_printing_style; static unsigned long page_size; static unsigned long mmap_window = 32; @@ -56,435 +58,59 @@ static char *parent_pattern = default_parent_pattern; static regex_t parent_regex; static int exclude_other = 1; -static int callchain; - -static u64 sample_type; - -struct ip_event { - struct perf_event_header header; - u64 ip; - u32 pid, tid; - unsigned char __more_data[]; -}; - -struct mmap_event { - struct perf_event_header header; - u32 pid, tid; - u64 start; - u64 len; - u64 pgoff; - char filename[PATH_MAX]; -}; - -struct comm_event { - struct perf_event_header header; - u32 pid, tid; - char comm[16]; -}; - -struct fork_event { - struct perf_event_header header; - u32 pid, ppid; -}; - -struct period_event { - struct perf_event_header header; - u64 time; - u64 id; - u64 sample_period; -}; - -struct lost_event { - struct perf_event_header header; - u64 id; - u64 lost; -}; - -struct read_event { - struct perf_event_header header; - u32 pid,tid; - u64 value; - u64 format[3]; -}; - -typedef union event_union { - struct perf_event_header header; - struct ip_event ip; - struct mmap_event mmap; - struct comm_event comm; - struct fork_event fork; - struct period_event period; - struct lost_event lost; - struct read_event read; -} event_t; - -static LIST_HEAD(dsos); -static struct dso *kernel_dso; -static struct dso *vdso; - -static void dsos__add(struct dso *dso) -{ - list_add_tail(&dso->node, &dsos); -} - -static struct dso *dsos__find(const char *name) -{ - struct dso *pos; - - list_for_each_entry(pos, &dsos, node) - if (strcmp(pos->name, name) == 0) - return pos; - return NULL; -} - -static struct dso *dsos__findnew(const char *name) -{ - struct dso *dso = dsos__find(name); - int nr; - - if (dso) - return dso; - - dso = dso__new(name, 0); - if (!dso) - goto out_delete_dso; - - nr = dso__load(dso, NULL, verbose); - if (nr < 0) { - eprintf("Failed to open: %s\n", name); - goto out_delete_dso; - } - if (!nr) - eprintf("No symbols found in: %s, maybe install a debug package?\n", name); - - dsos__add(dso); - - return dso; - -out_delete_dso: - dso__delete(dso); - return NULL; -} - -static void dsos__fprintf(FILE *fp) -{ - struct dso *pos; - - list_for_each_entry(pos, &dsos, node) - dso__fprintf(pos, fp); -} - -static struct symbol *vdso__find_symbol(struct dso *dso, u64 ip) -{ - return dso__find_symbol(kernel_dso, ip); -} - -static int load_kernel(void) -{ - int err; - - kernel_dso = dso__new("[kernel]", 0); - if (!kernel_dso) - return -1; - - err = dso__load_kernel(kernel_dso, vmlinux, NULL, verbose); - if (err) { - dso__delete(kernel_dso); - kernel_dso = NULL; - } else - dsos__add(kernel_dso); - - vdso = dso__new("[vdso]", 0); - if (!vdso) - return -1; - vdso->find_symbol = vdso__find_symbol; +static char callchain_default_opt[] = "fractal,0.5"; - dsos__add(vdso); - - return err; -} - -static char __cwd[PATH_MAX]; -static char *cwd = __cwd; -static int cwdlen; - -static int strcommon(const char *pathname) -{ - int n = 0; - - while (pathname[n] == cwd[n] && n < cwdlen) - ++n; - - return n; -} - -struct map { - struct list_head node; - u64 start; - u64 end; - u64 pgoff; - u64 (*map_ip)(struct map *, u64); - struct dso *dso; -}; - -static u64 map__map_ip(struct map *map, u64 ip) -{ - return ip - map->start + map->pgoff; -} - -static u64 vdso__map_ip(struct map *map, u64 ip) -{ - return ip; -} - -static inline int is_anon_memory(const char *filename) -{ - return strcmp(filename, "//anon") == 0; -} - -static struct map *map__new(struct mmap_event *event) -{ - struct map *self = malloc(sizeof(*self)); - - if (self != NULL) { - const char *filename = event->filename; - char newfilename[PATH_MAX]; - int anon; - - if (cwd) { - int n = strcommon(filename); - - if (n == cwdlen) { - snprintf(newfilename, sizeof(newfilename), - ".%s", filename + n); - filename = newfilename; - } - } - - anon = is_anon_memory(filename); - - if (anon) { - snprintf(newfilename, sizeof(newfilename), "/tmp/perf-%d.map", event->pid); - filename = newfilename; - } - - self->start = event->start; - self->end = event->start + event->len; - self->pgoff = event->pgoff; - - self->dso = dsos__findnew(filename); - if (self->dso == NULL) - goto out_delete; - - if (self->dso == vdso || anon) - self->map_ip = vdso__map_ip; - else - self->map_ip = map__map_ip; - } - return self; -out_delete: - free(self); - return NULL; -} - -static struct map *map__clone(struct map *self) -{ - struct map *map = malloc(sizeof(*self)); - - if (!map) - return NULL; - - memcpy(map, self, sizeof(*self)); - - return map; -} - -static int map__overlap(struct map *l, struct map *r) -{ - if (l->start > r->start) { - struct map *t = l; - l = r; - r = t; - } - - if (l->end > r->start) - return 1; +static int callchain; - return 0; -} +static char __cwd[PATH_MAX]; +static char *cwd = __cwd; +static int cwdlen; -static size_t map__fprintf(struct map *self, FILE *fp) -{ - return fprintf(fp, " %Lx-%Lx %Lx %s\n", - self->start, self->end, self->pgoff, self->dso->name); -} +static struct rb_root threads; +static struct thread *last_match; +static struct perf_header *header; -struct thread { - struct rb_node rb_node; - struct list_head maps; - pid_t pid; - char *comm; +static +struct callchain_param callchain_param = { + .mode = CHAIN_GRAPH_REL, + .min_percent = 0.5 }; -static struct thread *thread__new(pid_t pid) -{ - struct thread *self = malloc(sizeof(*self)); - - if (self != NULL) { - self->pid = pid; - self->comm = malloc(32); - if (self->comm) - snprintf(self->comm, 32, ":%d", self->pid); - INIT_LIST_HEAD(&self->maps); - } - - return self; -} - -static int thread__set_comm(struct thread *self, const char *comm) -{ - if (self->comm) - free(self->comm); - self->comm = strdup(comm); - return self->comm ? 0 : -ENOMEM; -} - -static size_t thread__fprintf(struct thread *self, FILE *fp) -{ - struct map *pos; - size_t ret = fprintf(fp, "Thread %d %s\n", self->pid, self->comm); - - list_for_each_entry(pos, &self->maps, node) - ret += map__fprintf(pos, fp); - - return ret; -} - - -static struct rb_root threads; -static struct thread *last_match; - -static struct thread *threads__findnew(pid_t pid) -{ - struct rb_node **p = &threads.rb_node; - struct rb_node *parent = NULL; - struct thread *th; - - /* - * Font-end cache - PID lookups come in blocks, - * so most of the time we dont have to look up - * the full rbtree: - */ - if (last_match && last_match->pid == pid) - return last_match; - - while (*p != NULL) { - parent = *p; - th = rb_entry(parent, struct thread, rb_node); - - if (th->pid == pid) { - last_match = th; - return th; - } - - if (pid < th->pid) - p = &(*p)->rb_left; - else - p = &(*p)->rb_right; - } - - th = thread__new(pid); - if (th != NULL) { - rb_link_node(&th->rb_node, parent, p); - rb_insert_color(&th->rb_node, &threads); - last_match = th; - } - - return th; -} - -static void thread__insert_map(struct thread *self, struct map *map) -{ - struct map *pos, *tmp; - - list_for_each_entry_safe(pos, tmp, &self->maps, node) { - if (map__overlap(pos, map)) { - if (verbose >= 2) { - printf("overlapping maps:\n"); - map__fprintf(map, stdout); - map__fprintf(pos, stdout); - } - - if (map->start <= pos->start && map->end > pos->start) - pos->start = map->end; - - if (map->end >= pos->end && map->start < pos->end) - pos->end = map->start; - - if (verbose >= 2) { - printf("after collision:\n"); - map__fprintf(pos, stdout); - } +static u64 sample_type; - if (pos->start >= pos->end) { - list_del_init(&pos->node); - free(pos); +static int repsep_fprintf(FILE *fp, const char *fmt, ...) +{ + int n; + va_list ap; + + va_start(ap, fmt); + if (!field_sep) + n = vfprintf(fp, fmt, ap); + else { + char *bf = NULL; + n = vasprintf(&bf, fmt, ap); + if (n > 0) { + char *sep = bf; + + while (1) { + sep = strchr(sep, *field_sep); + if (sep == NULL) + break; + *sep = '.'; } } + fputs(bf, fp); + free(bf); } - - list_add_tail(&map->node, &self->maps); -} - -static int thread__fork(struct thread *self, struct thread *parent) -{ - struct map *map; - - if (self->comm) - free(self->comm); - self->comm = strdup(parent->comm); - if (!self->comm) - return -ENOMEM; - - list_for_each_entry(map, &parent->maps, node) { - struct map *new = map__clone(map); - if (!new) - return -ENOMEM; - thread__insert_map(self, new); - } - - return 0; -} - -static struct map *thread__find_map(struct thread *self, u64 ip) -{ - struct map *pos; - - if (self == NULL) - return NULL; - - list_for_each_entry(pos, &self->maps, node) - if (ip >= pos->start && ip <= pos->end) - return pos; - - return NULL; + va_end(ap); + return n; } -static size_t threads__fprintf(FILE *fp) -{ - size_t ret = 0; - struct rb_node *nd; - - for (nd = rb_first(&threads); nd; nd = rb_next(nd)) { - struct thread *pos = rb_entry(nd, struct thread, rb_node); - - ret += thread__fprintf(pos, fp); - } - - return ret; -} +static unsigned int dsos__col_width, + comms__col_width, + threads__col_width; /* * histogram, sorted on item, collects counts @@ -515,11 +141,13 @@ struct hist_entry { struct sort_entry { struct list_head list; - char *header; + const char *header; int64_t (*cmp)(struct hist_entry *, struct hist_entry *); int64_t (*collapse)(struct hist_entry *, struct hist_entry *); - size_t (*print)(FILE *fp, struct hist_entry *); + size_t (*print)(FILE *fp, struct hist_entry *, unsigned int width); + unsigned int *width; + bool elide; }; static int64_t cmp_null(void *l, void *r) @@ -541,15 +169,17 @@ sort__thread_cmp(struct hist_entry *left, struct hist_entry *right) } static size_t -sort__thread_print(FILE *fp, struct hist_entry *self) +sort__thread_print(FILE *fp, struct hist_entry *self, unsigned int width) { - return fprintf(fp, "%16s:%5d", self->thread->comm ?: "", self->thread->pid); + return repsep_fprintf(fp, "%*s:%5d", width - 6, + self->thread->comm ?: "", self->thread->pid); } static struct sort_entry sort_thread = { - .header = " Command: Pid", + .header = "Command: Pid", .cmp = sort__thread_cmp, .print = sort__thread_print, + .width = &threads__col_width, }; /* --sort comm */ @@ -573,16 +203,17 @@ sort__comm_collapse(struct hist_entry *left, struct hist_entry *right) } static size_t -sort__comm_print(FILE *fp, struct hist_entry *self) +sort__comm_print(FILE *fp, struct hist_entry *self, unsigned int width) { - return fprintf(fp, "%16s", self->thread->comm); + return repsep_fprintf(fp, "%*s", width, self->thread->comm); } static struct sort_entry sort_comm = { - .header = " Command", + .header = "Command", .cmp = sort__comm_cmp, .collapse = sort__comm_collapse, .print = sort__comm_print, + .width = &comms__col_width, }; /* --sort dso */ @@ -600,18 +231,19 @@ sort__dso_cmp(struct hist_entry *left, struct hist_entry *right) } static size_t -sort__dso_print(FILE *fp, struct hist_entry *self) +sort__dso_print(FILE *fp, struct hist_entry *self, unsigned int width) { if (self->dso) - return fprintf(fp, "%-25s", self->dso->name); + return repsep_fprintf(fp, "%-*s", width, self->dso->name); - return fprintf(fp, "%016llx ", (u64)self->ip); + return repsep_fprintf(fp, "%*llx", width, (u64)self->ip); } static struct sort_entry sort_dso = { - .header = "Shared Object ", + .header = "Shared Object", .cmp = sort__dso_cmp, .print = sort__dso_print, + .width = &dsos__col_width, }; /* --sort symbol */ @@ -631,18 +263,23 @@ sort__sym_cmp(struct hist_entry *left, struct hist_entry *right) } static size_t -sort__sym_print(FILE *fp, struct hist_entry *self) +sort__sym_print(FILE *fp, struct hist_entry *self, unsigned int width __used) { size_t ret = 0; if (verbose) - ret += fprintf(fp, "%#018llx ", (u64)self->ip); + ret += repsep_fprintf(fp, "%#018llx %c ", (u64)self->ip, + dso__symtab_origin(self->dso)); + ret += repsep_fprintf(fp, "[%c] ", self->level); if (self->sym) { - ret += fprintf(fp, "[%c] %s", - self->dso == kernel_dso ? 'k' : '.', self->sym->name); + ret += repsep_fprintf(fp, "%s", self->sym->name); + + if (self->sym->module) + ret += repsep_fprintf(fp, "\t[%s]", + self->sym->module->name); } else { - ret += fprintf(fp, "%#016llx", (u64)self->ip); + ret += repsep_fprintf(fp, "%#016llx", (u64)self->ip); } return ret; @@ -669,26 +306,26 @@ sort__parent_cmp(struct hist_entry *left, struct hist_entry *right) } static size_t -sort__parent_print(FILE *fp, struct hist_entry *self) +sort__parent_print(FILE *fp, struct hist_entry *self, unsigned int width) { - size_t ret = 0; - - ret += fprintf(fp, "%-20s", self->parent ? self->parent->name : "[other]"); - - return ret; + return repsep_fprintf(fp, "%-*s", width, + self->parent ? self->parent->name : "[other]"); } +static unsigned int parent_symbol__col_width; + static struct sort_entry sort_parent = { - .header = "Parent symbol ", + .header = "Parent symbol", .cmp = sort__parent_cmp, .print = sort__parent_print, + .width = &parent_symbol__col_width, }; static int sort__need_collapse = 0; static int sort__has_parent = 0; struct sort_dimension { - char *name; + const char *name; struct sort_entry *entry; int taken; }; @@ -703,9 +340,9 @@ static struct sort_dimension sort_dimensions[] = { static LIST_HEAD(hist_entry__sort_list); -static int sort_dimension__add(char *tok) +static int sort_dimension__add(const char *tok) { - int i; + unsigned int i; for (i = 0; i < ARRAY_SIZE(sort_dimensions); i++) { struct sort_dimension *sd = &sort_dimensions[i]; @@ -775,8 +412,146 @@ hist_entry__collapse(struct hist_entry *left, struct hist_entry *right) return cmp; } +static size_t ipchain__fprintf_graph_line(FILE *fp, int depth, int depth_mask) +{ + int i; + size_t ret = 0; + + ret += fprintf(fp, "%s", " "); + + for (i = 0; i < depth; i++) + if (depth_mask & (1 << i)) + ret += fprintf(fp, "| "); + else + ret += fprintf(fp, " "); + + ret += fprintf(fp, "\n"); + + return ret; +} +static size_t +ipchain__fprintf_graph(FILE *fp, struct callchain_list *chain, int depth, + int depth_mask, int count, u64 total_samples, + int hits) +{ + int i; + size_t ret = 0; + + ret += fprintf(fp, "%s", " "); + for (i = 0; i < depth; i++) { + if (depth_mask & (1 << i)) + ret += fprintf(fp, "|"); + else + ret += fprintf(fp, " "); + if (!count && i == depth - 1) { + double percent; + + percent = hits * 100.0 / total_samples; + ret += percent_color_fprintf(fp, "--%2.2f%%-- ", percent); + } else + ret += fprintf(fp, "%s", " "); + } + if (chain->sym) + ret += fprintf(fp, "%s\n", chain->sym->name); + else + ret += fprintf(fp, "%p\n", (void *)(long)chain->ip); + + return ret; +} + +static struct symbol *rem_sq_bracket; +static struct callchain_list rem_hits; + +static void init_rem_hits(void) +{ + rem_sq_bracket = malloc(sizeof(*rem_sq_bracket) + 6); + if (!rem_sq_bracket) { + fprintf(stderr, "Not enough memory to display remaining hits\n"); + return; + } + + strcpy(rem_sq_bracket->name, "[...]"); + rem_hits.sym = rem_sq_bracket; +} + static size_t -callchain__fprintf(FILE *fp, struct callchain_node *self, u64 total_samples) +callchain__fprintf_graph(FILE *fp, struct callchain_node *self, + u64 total_samples, int depth, int depth_mask) +{ + struct rb_node *node, *next; + struct callchain_node *child; + struct callchain_list *chain; + int new_depth_mask = depth_mask; + u64 new_total; + u64 remaining; + size_t ret = 0; + int i; + + if (callchain_param.mode == CHAIN_GRAPH_REL) + new_total = self->children_hit; + else + new_total = total_samples; + + remaining = new_total; + + node = rb_first(&self->rb_root); + while (node) { + u64 cumul; + + child = rb_entry(node, struct callchain_node, rb_node); + cumul = cumul_hits(child); + remaining -= cumul; + + /* + * The depth mask manages the output of pipes that show + * the depth. We don't want to keep the pipes of the current + * level for the last child of this depth. + * Except if we have remaining filtered hits. They will + * supersede the last child + */ + next = rb_next(node); + if (!next && (callchain_param.mode != CHAIN_GRAPH_REL || !remaining)) + new_depth_mask &= ~(1 << (depth - 1)); + + /* + * But we keep the older depth mask for the line seperator + * to keep the level link until we reach the last child + */ + ret += ipchain__fprintf_graph_line(fp, depth, depth_mask); + i = 0; + list_for_each_entry(chain, &child->val, list) { + if (chain->ip >= PERF_CONTEXT_MAX) + continue; + ret += ipchain__fprintf_graph(fp, chain, depth, + new_depth_mask, i++, + new_total, + cumul); + } + ret += callchain__fprintf_graph(fp, child, new_total, + depth + 1, + new_depth_mask | (1 << depth)); + node = next; + } + + if (callchain_param.mode == CHAIN_GRAPH_REL && + remaining && remaining != new_total) { + + if (!rem_sq_bracket) + return ret; + + new_depth_mask &= ~(1 << (depth - 1)); + + ret += ipchain__fprintf_graph(fp, &rem_hits, depth, + new_depth_mask, 0, new_total, + remaining); + } + + return ret; +} + +static size_t +callchain__fprintf_flat(FILE *fp, struct callchain_node *self, + u64 total_samples) { struct callchain_list *chain; size_t ret = 0; @@ -784,11 +559,18 @@ callchain__fprintf(FILE *fp, struct callchain_node *self, u64 total_samples) if (!self) return 0; - ret += callchain__fprintf(fp, self->parent, total_samples); + ret += callchain__fprintf_flat(fp, self->parent, total_samples); - list_for_each_entry(chain, &self->val, list) - ret += fprintf(fp, " %p\n", (void *)chain->ip); + list_for_each_entry(chain, &self->val, list) { + if (chain->ip >= PERF_CONTEXT_MAX) + continue; + if (chain->sym) + ret += fprintf(fp, " %s\n", chain->sym->name); + else + ret += fprintf(fp, " %p\n", + (void *)(long)chain->ip); + } return ret; } @@ -807,8 +589,20 @@ hist_entry_callchain__fprintf(FILE *fp, struct hist_entry *self, chain = rb_entry(rb_node, struct callchain_node, rb_node); percent = chain->hit * 100.0 / total_samples; - ret += fprintf(fp, " %6.2f%%\n", percent); - ret += callchain__fprintf(fp, chain, total_samples); + switch (callchain_param.mode) { + case CHAIN_FLAT: + ret += percent_color_fprintf(fp, " %6.2f%%\n", + percent); + ret += callchain__fprintf_flat(fp, chain, total_samples); + break; + case CHAIN_GRAPH_ABS: /* Falldown */ + case CHAIN_GRAPH_REL: + ret += callchain__fprintf_graph(fp, chain, + total_samples, 1, 1); + case CHAIN_NONE: + default: + break; + } ret += fprintf(fp, "\n"); rb_node = rb_next(rb_node); } @@ -826,33 +620,26 @@ hist_entry__fprintf(FILE *fp, struct hist_entry *self, u64 total_samples) if (exclude_other && !self->parent) return 0; - if (total_samples) { - double percent = self->count * 100.0 / total_samples; - char *color = PERF_COLOR_NORMAL; - - /* - * We color high-overhead entries in red, mid-overhead - * entries in green - and keep the low overhead places - * normal: - */ - if (percent >= 5.0) { - color = PERF_COLOR_RED; - } else { - if (percent >= 0.5) - color = PERF_COLOR_GREEN; - } + if (total_samples) + ret = percent_color_fprintf(fp, + field_sep ? "%.2f" : " %6.2f%%", + (self->count * 100.0) / total_samples); + else + ret = fprintf(fp, field_sep ? "%lld" : "%12lld ", self->count); - ret = color_fprintf(fp, color, " %6.2f%%", - (self->count * 100.0) / total_samples); - } else - ret = fprintf(fp, "%12Ld ", self->count); + if (show_nr_samples) { + if (field_sep) + fprintf(fp, "%c%lld", *field_sep, self->count); + else + fprintf(fp, "%11lld", self->count); + } list_for_each_entry(se, &hist_entry__sort_list, list) { - if (exclude_other && (se == &sort_parent)) + if (se->elide) continue; - fprintf(fp, " "); - ret += se->print(fp, self); + fprintf(fp, "%s", field_sep ?: " "); + ret += se->print(fp, self, se->width ? *se->width : 0); } ret += fprintf(fp, "\n"); @@ -867,6 +654,46 @@ hist_entry__fprintf(FILE *fp, struct hist_entry *self, u64 total_samples) * */ +static void dso__calc_col_width(struct dso *self) +{ + if (!col_width_list_str && !field_sep && + (!dso_list || strlist__has_entry(dso_list, self->name))) { + unsigned int slen = strlen(self->name); + if (slen > dsos__col_width) + dsos__col_width = slen; + } + + self->slen_calculated = 1; +} + +static void thread__comm_adjust(struct thread *self) +{ + char *comm = self->comm; + + if (!col_width_list_str && !field_sep && + (!comm_list || strlist__has_entry(comm_list, comm))) { + unsigned int slen = strlen(comm); + + if (slen > comms__col_width) { + comms__col_width = slen; + threads__col_width = slen + 6; + } + } +} + +static int thread__set_comm_adjust(struct thread *self, const char *comm) +{ + int ret = thread__set_comm(self, comm); + + if (ret) + return ret; + + thread__comm_adjust(self); + + return 0; +} + + static struct symbol * resolve_symbol(struct thread *thread, struct map **mapp, struct dso **dsop, u64 *ipp) @@ -886,6 +713,14 @@ resolve_symbol(struct thread *thread, struct map **mapp, map = thread__find_map(thread, ip); if (map != NULL) { + /* + * We have to do this here as we may have a dso + * with no symbol hit that has a name longer than + * the ones with symbols sampled. + */ + if (!sort_dso.elide && !map->dso->slen_calculated) + dso__calc_col_width(map->dso); + if (mapp) *mapp = map; got_map: @@ -902,8 +737,8 @@ got_map: if ((long long)ip < 0) dso = kernel_dso; } - dprintf(" ...... dso: %s\n", dso ? dso->name : "<not found>"); - dprintf(" ...... map: %Lx -> %Lx\n", *ipp, ip); + dump_printf(" ...... dso: %s\n", dso ? dso->name : "<not found>"); + dump_printf(" ...... map: %Lx -> %Lx\n", *ipp, ip); *ipp = ip; if (dsop) @@ -923,6 +758,58 @@ static int call__match(struct symbol *sym) return 0; } +static struct symbol ** +resolve_callchain(struct thread *thread, struct map *map __used, + struct ip_callchain *chain, struct hist_entry *entry) +{ + u64 context = PERF_CONTEXT_MAX; + struct symbol **syms = NULL; + unsigned int i; + + if (callchain) { + syms = calloc(chain->nr, sizeof(*syms)); + if (!syms) { + fprintf(stderr, "Can't allocate memory for symbols\n"); + exit(-1); + } + } + + for (i = 0; i < chain->nr; i++) { + u64 ip = chain->ips[i]; + struct dso *dso = NULL; + struct symbol *sym; + + if (ip >= PERF_CONTEXT_MAX) { + context = ip; + continue; + } + + switch (context) { + case PERF_CONTEXT_HV: + dso = hypervisor_dso; + break; + case PERF_CONTEXT_KERNEL: + dso = kernel_dso; + break; + default: + break; + } + + sym = resolve_symbol(thread, NULL, &dso, &ip); + + if (sym) { + if (sort__has_parent && call__match(sym) && + !entry->parent) + entry->parent = sym; + if (!callchain) + break; + syms[i] = sym; + } + } + + return syms; +} + /* * collect histogram counts */ @@ -935,6 +822,7 @@ hist_entry__add(struct thread *thread, struct map *map, struct dso *dso, struct rb_node **p = &hist.rb_node; struct rb_node *parent = NULL; struct hist_entry *he; + struct symbol **syms = NULL; struct hist_entry entry = { .thread = thread, .map = map, @@ -948,36 +836,8 @@ hist_entry__add(struct thread *thread, struct map *map, struct dso *dso, }; int cmp; - if (sort__has_parent && chain) { - u64 context = PERF_CONTEXT_MAX; - int i; - - for (i = 0; i < chain->nr; i++) { - u64 ip = chain->ips[i]; - struct dso *dso = NULL; - struct symbol *sym; - - if (ip >= PERF_CONTEXT_MAX) { - context = ip; - continue; - } - - switch (context) { - case PERF_CONTEXT_KERNEL: - dso = kernel_dso; - break; - default: - break; - } - - sym = resolve_symbol(thread, NULL, &dso, &ip); - - if (sym && call__match(sym)) { - entry.parent = sym; - break; - } - } - } + if ((sort__has_parent || callchain) && chain) + syms = resolve_callchain(thread, map, chain, &entry); while (*p != NULL) { parent = *p; @@ -987,8 +847,10 @@ hist_entry__add(struct thread *thread, struct map *map, struct dso *dso, if (!cmp) { he->count += count; - if (callchain) - append_chain(&he->callchain, chain); + if (callchain) { + append_chain(&he->callchain, chain, syms); + free(syms); + } return 0; } @@ -1004,7 +866,8 @@ hist_entry__add(struct thread *thread, struct map *map, struct dso *dso, *he = entry; if (callchain) { callchain_init(&he->callchain); - append_chain(&he->callchain, chain); + append_chain(&he->callchain, chain, syms); + free(syms); } rb_link_node(&he->rb_node, parent, p); rb_insert_color(&he->rb_node, &hist); @@ -1076,14 +939,15 @@ static void collapse__resort(void) static struct rb_root output_hists; -static void output__insert_entry(struct hist_entry *he) +static void output__insert_entry(struct hist_entry *he, u64 min_callchain_hits) { struct rb_node **p = &output_hists.rb_node; struct rb_node *parent = NULL; struct hist_entry *iter; if (callchain) - sort_chain_to_rbtree(&he->sorted_chain, &he->callchain); + callchain_param.sort(&he->sorted_chain, &he->callchain, + min_callchain_hits, &callchain_param); while (*p != NULL) { parent = *p; @@ -1099,11 +963,14 @@ static void output__insert_entry(struct hist_entry *he) rb_insert_color(&he->rb_node, &output_hists); } -static void output__resort(void) +static void output__resort(u64 total_samples) { struct rb_node *next; struct hist_entry *n; struct rb_root *tree = &hist; + u64 min_callchain_hits; + + min_callchain_hits = total_samples * (callchain_param.min_percent / 100); if (sort__need_collapse) tree = &collapse_hists; @@ -1115,7 +982,7 @@ static void output__resort(void) next = rb_next(&n->rb_node); rb_erase(&n->rb_node, tree); - output__insert_entry(n); + output__insert_entry(n, min_callchain_hits); } } @@ -1125,35 +992,72 @@ static size_t output__fprintf(FILE *fp, u64 total_samples) struct sort_entry *se; struct rb_node *nd; size_t ret = 0; + unsigned int width; + char *col_width = col_width_list_str; + int raw_printing_style; - fprintf(fp, "\n"); - fprintf(fp, "#\n"); - fprintf(fp, "# (%Ld samples)\n", (u64)total_samples); + raw_printing_style = !strcmp(pretty_printing_style, "raw"); + + init_rem_hits(); + + fprintf(fp, "# Samples: %Ld\n", (u64)total_samples); fprintf(fp, "#\n"); fprintf(fp, "# Overhead"); + if (show_nr_samples) { + if (field_sep) + fprintf(fp, "%cSamples", *field_sep); + else + fputs(" Samples ", fp); + } list_for_each_entry(se, &hist_entry__sort_list, list) { - if (exclude_other && (se == &sort_parent)) + if (se->elide) + continue; + if (field_sep) { + fprintf(fp, "%c%s", *field_sep, se->header); continue; - fprintf(fp, " %s", se->header); + } + width = strlen(se->header); + if (se->width) { + if (col_width_list_str) { + if (col_width) { + *se->width = atoi(col_width); + col_width = strchr(col_width, ','); + if (col_width) + ++col_width; + } + } + width = *se->width = max(*se->width, width); + } + fprintf(fp, " %*s", width, se->header); } fprintf(fp, "\n"); + if (field_sep) + goto print_entries; + fprintf(fp, "# ........"); + if (show_nr_samples) + fprintf(fp, " .........."); list_for_each_entry(se, &hist_entry__sort_list, list) { - int i; + unsigned int i; - if (exclude_other && (se == &sort_parent)) + if (se->elide) continue; fprintf(fp, " "); - for (i = 0; i < strlen(se->header); i++) + if (se->width) + width = *se->width; + else + width = strlen(se->header); + for (i = 0; i < width; i++) fprintf(fp, "."); } fprintf(fp, "\n"); fprintf(fp, "#\n"); +print_entries: for (nd = rb_first(&output_hists); nd; nd = rb_next(nd)) { pos = rb_entry(nd, struct hist_entry, rb_node); ret += hist_entry__fprintf(fp, pos, total_samples); @@ -1162,23 +1066,18 @@ static size_t output__fprintf(FILE *fp, u64 total_samples) if (sort_order == default_sort_order && parent_pattern == default_parent_pattern) { fprintf(fp, "#\n"); - fprintf(fp, "# (For more details, try: perf report --sort comm,dso,symbol)\n"); + fprintf(fp, "# (For a higher level overview, try: perf report --sort comm,dso)\n"); fprintf(fp, "#\n"); } fprintf(fp, "\n"); - return ret; -} + free(rem_sq_bracket); -static void register_idle_thread(void) -{ - struct thread *thread = threads__findnew(0); + if (show_threads) + perf_read_values_display(fp, &show_threads_values, + raw_printing_style); - if (thread == NULL || - thread__set_comm(thread, "[idle]")) { - fprintf(stderr, "problem inserting idle task.\n"); - exit(-1); - } + return ret; } static unsigned long total = 0, @@ -1207,32 +1106,35 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) char level; int show = 0; struct dso *dso = NULL; - struct thread *thread = threads__findnew(event->ip.pid); + struct thread *thread; u64 ip = event->ip.ip; u64 period = 1; struct map *map = NULL; void *more_data = event->ip.__more_data; struct ip_callchain *chain = NULL; + int cpumode; + + thread = threads__findnew(event->ip.pid, &threads, &last_match); if (sample_type & PERF_SAMPLE_PERIOD) { period = *(u64 *)more_data; more_data += sizeof(u64); } - dprintf("%p [%p]: PERF_EVENT_SAMPLE (IP, %d): %d: %p period: %Ld\n", + dump_printf("%p [%p]: PERF_RECORD_SAMPLE (IP, %d): %d/%d: %p period: %Ld\n", (void *)(offset + head), (void *)(long)(event->header.size), event->header.misc, - event->ip.pid, + event->ip.pid, event->ip.tid, (void *)(long)ip, (long long)period); if (sample_type & PERF_SAMPLE_CALLCHAIN) { - int i; + unsigned int i; chain = (void *)more_data; - dprintf("... chain: nr:%Lu\n", chain->nr); + dump_printf("... chain: nr:%Lu\n", chain->nr); if (validate_chain(chain, event) < 0) { eprintf("call-chain problem with event, skipping it.\n"); @@ -1241,11 +1143,11 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) if (dump_trace) { for (i = 0; i < chain->nr; i++) - dprintf("..... %2d: %016Lx\n", i, chain->ips[i]); + dump_printf("..... %2d: %016Lx\n", i, chain->ips[i]); } } - dprintf(" ... thread: %s:%d\n", thread->comm, thread->pid); + dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid); if (thread == NULL) { eprintf("problem processing %d event, skipping it.\n", @@ -1256,15 +1158,17 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) if (comm_list && !strlist__has_entry(comm_list, thread->comm)) return 0; - if (event->header.misc & PERF_EVENT_MISC_KERNEL) { + cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + + if (cpumode == PERF_RECORD_MISC_KERNEL) { show = SHOW_KERNEL; level = 'k'; dso = kernel_dso; - dprintf(" ...... dso: %s\n", dso->name); + dump_printf(" ...... dso: %s\n", dso->name); - } else if (event->header.misc & PERF_EVENT_MISC_USER) { + } else if (cpumode == PERF_RECORD_MISC_USER) { show = SHOW_USER; level = '.'; @@ -1272,16 +1176,20 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) } else { show = SHOW_HV; level = 'H'; - dprintf(" ...... dso: [hypervisor]\n"); + + dso = hypervisor_dso; + + dump_printf(" ...... dso: [hypervisor]\n"); } if (show & show_mask) { struct symbol *sym = resolve_symbol(thread, &map, &dso, &ip); - if (dso_list && dso && dso->name && !strlist__has_entry(dso_list, dso->name)) + if (dso_list && (!dso || !dso->name || + !strlist__has_entry(dso_list, dso->name))) return 0; - if (sym_list && sym && !strlist__has_entry(sym_list, sym->name)) + if (sym_list && (!sym || !strlist__has_entry(sym_list, sym->name))) return 0; if (hist_entry__add(thread, map, dso, sym, ip, chain, level, period)) { @@ -1297,20 +1205,23 @@ process_sample_event(event_t *event, unsigned long offset, unsigned long head) static int process_mmap_event(event_t *event, unsigned long offset, unsigned long head) { - struct thread *thread = threads__findnew(event->mmap.pid); - struct map *map = map__new(&event->mmap); + struct thread *thread; + struct map *map = map__new(&event->mmap, cwd, cwdlen); + + thread = threads__findnew(event->mmap.pid, &threads, &last_match); - dprintf("%p [%p]: PERF_EVENT_MMAP %d: [%p(%p) @ %p]: %s\n", + dump_printf("%p [%p]: PERF_RECORD_MMAP %d/%d: [%p(%p) @ %p]: %s\n", (void *)(offset + head), (void *)(long)(event->header.size), event->mmap.pid, + event->mmap.tid, (void *)(long)event->mmap.start, (void *)(long)event->mmap.len, (void *)(long)event->mmap.pgoff, event->mmap.filename); if (thread == NULL || map == NULL) { - dprintf("problem processing PERF_EVENT_MMAP, skipping event.\n"); + dump_printf("problem processing PERF_RECORD_MMAP, skipping event.\n"); return 0; } @@ -1323,16 +1234,18 @@ process_mmap_event(event_t *event, unsigned long offset, unsigned long head) static int process_comm_event(event_t *event, unsigned long offset, unsigned long head) { - struct thread *thread = threads__findnew(event->comm.pid); + struct thread *thread; - dprintf("%p [%p]: PERF_EVENT_COMM: %s:%d\n", + thread = threads__findnew(event->comm.pid, &threads, &last_match); + + dump_printf("%p [%p]: PERF_RECORD_COMM: %s:%d\n", (void *)(offset + head), (void *)(long)(event->header.size), event->comm.comm, event->comm.pid); if (thread == NULL || - thread__set_comm(thread, event->comm.comm)) { - dprintf("problem processing PERF_EVENT_COMM, skipping event.\n"); + thread__set_comm_adjust(thread, event->comm.comm)) { + dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n"); return -1; } total_comm++; @@ -1341,18 +1254,33 @@ process_comm_event(event_t *event, unsigned long offset, unsigned long head) } static int -process_fork_event(event_t *event, unsigned long offset, unsigned long head) +process_task_event(event_t *event, unsigned long offset, unsigned long head) { - struct thread *thread = threads__findnew(event->fork.pid); - struct thread *parent = threads__findnew(event->fork.ppid); + struct thread *thread; + struct thread *parent; + + thread = threads__findnew(event->fork.pid, &threads, &last_match); + parent = threads__findnew(event->fork.ppid, &threads, &last_match); - dprintf("%p [%p]: PERF_EVENT_FORK: %d:%d\n", + dump_printf("%p [%p]: PERF_RECORD_%s: (%d:%d):(%d:%d)\n", (void *)(offset + head), (void *)(long)(event->header.size), - event->fork.pid, event->fork.ppid); + event->header.type == PERF_RECORD_FORK ? "FORK" : "EXIT", + event->fork.pid, event->fork.tid, + event->fork.ppid, event->fork.ptid); + + /* + * A thread clone will have the same PID for both + * parent and child. + */ + if (thread == parent) + return 0; + + if (event->header.type == PERF_RECORD_EXIT) + return 0; if (!thread || !parent || thread__fork(thread, parent)) { - dprintf("problem processing PERF_EVENT_FORK, skipping event.\n"); + dump_printf("problem processing PERF_RECORD_FORK, skipping event.\n"); return -1; } total_fork++; @@ -1361,22 +1289,9 @@ process_fork_event(event_t *event, unsigned long offset, unsigned long head) } static int -process_period_event(event_t *event, unsigned long offset, unsigned long head) -{ - dprintf("%p [%p]: PERF_EVENT_PERIOD: time:%Ld, id:%Ld: period:%Ld\n", - (void *)(offset + head), - (void *)(long)(event->header.size), - event->period.time, - event->period.id, - event->period.sample_period); - - return 0; -} - -static int process_lost_event(event_t *event, unsigned long offset, unsigned long head) { - dprintf("%p [%p]: PERF_EVENT_LOST: id:%Ld: lost:%Ld\n", + dump_printf("%p [%p]: PERF_RECORD_LOST: id:%Ld: lost:%Ld\n", (void *)(offset + head), (void *)(long)(event->header.size), event->lost.id, @@ -1387,50 +1302,30 @@ process_lost_event(event_t *event, unsigned long offset, unsigned long head) return 0; } -static void trace_event(event_t *event) +static int +process_read_event(event_t *event, unsigned long offset, unsigned long head) { - unsigned char *raw_event = (void *)event; - char *color = PERF_COLOR_BLUE; - int i, j; - - if (!dump_trace) - return; + struct perf_event_attr *attr; - dprintf("."); - cdprintf("\n. ... raw event: size %d bytes\n", event->header.size); + attr = perf_header__find_attr(event->read.id, header); - for (i = 0; i < event->header.size; i++) { - if ((i & 15) == 0) { - dprintf("."); - cdprintf(" %04x: ", i); - } - - cdprintf(" %02x", raw_event[i]); - - if (((i & 15) == 15) || i == event->header.size-1) { - cdprintf(" "); - for (j = 0; j < 15-(i & 15); j++) - cdprintf(" "); - for (j = 0; j < (i & 15); j++) { - if (isprint(raw_event[i-15+j])) - cdprintf("%c", raw_event[i-15+j]); - else - cdprintf("."); - } - cdprintf("\n"); - } + if (show_threads) { + const char *name = attr ? __event_name(attr->type, attr->config) + : "unknown"; + perf_read_values_add_value(&show_threads_values, + event->read.pid, event->read.tid, + event->read.id, + name, + event->read.value); } - dprintf(".\n"); -} -static int -process_read_event(event_t *event, unsigned long offset, unsigned long head) -{ - dprintf("%p [%p]: PERF_EVENT_READ: %d %d %Lu\n", + dump_printf("%p [%p]: PERF_RECORD_READ: %d %d %s %Lu\n", (void *)(offset + head), (void *)(long)(event->header.size), event->read.pid, event->read.tid, + attr ? __event_name(attr->type, attr->config) + : "FAIL", event->read.value); return 0; @@ -1442,33 +1337,31 @@ process_event(event_t *event, unsigned long offset, unsigned long head) trace_event(event); switch (event->header.type) { - case PERF_EVENT_SAMPLE: + case PERF_RECORD_SAMPLE: return process_sample_event(event, offset, head); - case PERF_EVENT_MMAP: + case PERF_RECORD_MMAP: return process_mmap_event(event, offset, head); - case PERF_EVENT_COMM: + case PERF_RECORD_COMM: return process_comm_event(event, offset, head); - case PERF_EVENT_FORK: - return process_fork_event(event, offset, head); - - case PERF_EVENT_PERIOD: - return process_period_event(event, offset, head); + case PERF_RECORD_FORK: + case PERF_RECORD_EXIT: + return process_task_event(event, offset, head); - case PERF_EVENT_LOST: + case PERF_RECORD_LOST: return process_lost_event(event, offset, head); - case PERF_EVENT_READ: + case PERF_RECORD_READ: return process_read_event(event, offset, head); /* * We dont process them right now but they are fine: */ - case PERF_EVENT_THROTTLE: - case PERF_EVENT_UNTHROTTLE: + case PERF_RECORD_THROTTLE: + case PERF_RECORD_UNTHROTTLE: return 0; default: @@ -1478,36 +1371,22 @@ process_event(event_t *event, unsigned long offset, unsigned long head) return 0; } -static struct perf_header *header; - -static u64 perf_header__sample_type(void) -{ - u64 sample_type = 0; - int i; - - for (i = 0; i < header->attrs; i++) { - struct perf_header_attr *attr = header->attr[i]; - - if (!sample_type) - sample_type = attr->attr.sample_type; - else if (sample_type != attr->attr.sample_type) - die("non matching sample_type"); - } - - return sample_type; -} - static int __cmd_report(void) { int ret, rc = EXIT_FAILURE; unsigned long offset = 0; unsigned long head, shift; - struct stat stat; + struct stat input_stat; + struct thread *idle; event_t *event; uint32_t size; char *buf; - register_idle_thread(); + idle = register_idle_thread(&threads, &last_match); + thread__comm_adjust(idle); + + if (show_threads) + perf_read_values_init(&show_threads_values); input = open(input_name, O_RDONLY); if (input < 0) { @@ -1518,13 +1397,18 @@ static int __cmd_report(void) exit(-1); } - ret = fstat(input, &stat); + ret = fstat(input, &input_stat); if (ret < 0) { perror("failed to stat file"); exit(-1); } - if (!stat.st_size) { + if (!force && input_stat.st_uid && (input_stat.st_uid != geteuid())) { + fprintf(stderr, "file: %s not owned by current user or root\n", input_name); + exit(-1); + } + + if (!input_stat.st_size) { fprintf(stderr, "zero-sized file, nothing to do!\n"); exit(0); } @@ -1532,11 +1416,28 @@ static int __cmd_report(void) header = perf_header__read(input); head = header->data_offset; - sample_type = perf_header__sample_type(); + sample_type = perf_header__sample_type(header); - if (sort__has_parent && !(sample_type & PERF_SAMPLE_CALLCHAIN)) { - fprintf(stderr, "selected --sort parent, but no callchain data\n"); - exit(-1); + if (!(sample_type & PERF_SAMPLE_CALLCHAIN)) { + if (sort__has_parent) { + fprintf(stderr, "selected --sort parent, but no" + " callchain data. Did you call" + " perf record without -g?\n"); + exit(-1); + } + if (callchain) { + fprintf(stderr, "selected -g but no callchain data." + " Did you call perf record without" + " -g?\n"); + exit(-1); + } + } else if (callchain_param.mode != CHAIN_NONE && !callchain) { + callchain = 1; + if (register_callchain_param(&callchain_param) < 0) { + fprintf(stderr, "Can't register callchain" + " params\n"); + exit(-1); + } } if (load_kernel() < 0) { @@ -1575,12 +1476,12 @@ more: size = 8; if (head + event->header.size >= page_size * mmap_window) { - int ret; + int munmap_ret; shift = page_size * (head / page_size); - ret = munmap(buf, page_size * mmap_window); - assert(ret == 0); + munmap_ret = munmap(buf, page_size * mmap_window); + assert(munmap_ret == 0); offset += shift; head -= shift; @@ -1589,14 +1490,14 @@ more: size = event->header.size; - dprintf("\n%p [%p]: event: %d\n", + dump_printf("\n%p [%p]: event: %d\n", (void *)(offset + head), (void *)(long)event->header.size, event->header.type); if (!size || process_event(event, offset, head) < 0) { - dprintf("%p [%p]: skipping unknown header type: %d\n", + dump_printf("%p [%p]: skipping unknown header type: %d\n", (void *)(offset + head), (void *)(long)(event->header.size), event->header.type); @@ -1619,36 +1520,92 @@ more: if (offset + head >= header->data_offset + header->data_size) goto done; - if (offset + head < stat.st_size) + if (offset + head < (unsigned long)input_stat.st_size) goto more; done: rc = EXIT_SUCCESS; close(input); - dprintf(" IP events: %10ld\n", total); - dprintf(" mmap events: %10ld\n", total_mmap); - dprintf(" comm events: %10ld\n", total_comm); - dprintf(" fork events: %10ld\n", total_fork); - dprintf(" lost events: %10ld\n", total_lost); - dprintf(" unknown events: %10ld\n", total_unknown); + dump_printf(" IP events: %10ld\n", total); + dump_printf(" mmap events: %10ld\n", total_mmap); + dump_printf(" comm events: %10ld\n", total_comm); + dump_printf(" fork events: %10ld\n", total_fork); + dump_printf(" lost events: %10ld\n", total_lost); + dump_printf(" unknown events: %10ld\n", total_unknown); if (dump_trace) return 0; if (verbose >= 3) - threads__fprintf(stdout); + threads__fprintf(stdout, &threads); if (verbose >= 2) dsos__fprintf(stdout); collapse__resort(); - output__resort(); + output__resort(total); output__fprintf(stdout, total); + if (show_threads) + perf_read_values_destroy(&show_threads_values); + return rc; } +static int +parse_callchain_opt(const struct option *opt __used, const char *arg, + int unset __used) +{ + char *tok; + char *endptr; + + callchain = 1; + + if (!arg) + return 0; + + tok = strtok((char *)arg, ","); + if (!tok) + return -1; + + /* get the output mode */ + if (!strncmp(tok, "graph", strlen(arg))) + callchain_param.mode = CHAIN_GRAPH_ABS; + + else if (!strncmp(tok, "flat", strlen(arg))) + callchain_param.mode = CHAIN_FLAT; + + else if (!strncmp(tok, "fractal", strlen(arg))) + callchain_param.mode = CHAIN_GRAPH_REL; + + else if (!strncmp(tok, "none", strlen(arg))) { + callchain_param.mode = CHAIN_NONE; + callchain = 0; + + return 0; + } + + else + return -1; + + /* get the min percentage */ + tok = strtok(NULL, ","); + if (!tok) + goto setup; + + callchain_param.min_percent = strtod(tok, &endptr); + if (tok == endptr) + return -1; + +setup: + if (register_callchain_param(&callchain_param) < 0) { + fprintf(stderr, "Can't register callchain params\n"); + return -1; + } + return 0; +} + static const char * const report_usage[] = { "perf report [<options>] <command>", NULL @@ -1661,7 +1618,16 @@ static const struct option options[] = { "be more verbose (show symbol address, etc)"), OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, "dump raw trace in ASCII"), - OPT_STRING('k', "vmlinux", &vmlinux, "file", "vmlinux pathname"), + OPT_STRING('k', "vmlinux", &vmlinux_name, "file", "vmlinux pathname"), + OPT_BOOLEAN('f', "force", &force, "don't complain, do it"), + OPT_BOOLEAN('m', "modules", &modules, + "load module symbols - WARNING: use only with -k and LIVE kernel"), + OPT_BOOLEAN('n', "show-nr-samples", &show_nr_samples, + "Show a column with the number of samples"), + OPT_BOOLEAN('T', "threads", &show_threads, + "Show per-thread event counters"), + OPT_STRING(0, "pretty", &pretty_printing_style, "key", + "pretty printing style key: normal raw"), OPT_STRING('s', "sort", &sort_order, "key[,key2...]", "sort by key(s): pid, comm, dso, symbol, parent"), OPT_BOOLEAN('P', "full-paths", &full_paths, @@ -1670,13 +1636,21 @@ static const struct option options[] = { "regex filter to identify parent, see: '--sort parent'"), OPT_BOOLEAN('x', "exclude-other", &exclude_other, "Only display entries with parent-match"), - OPT_BOOLEAN('c', "callchain", &callchain, "Display callchains"), + OPT_CALLBACK_DEFAULT('g', "call-graph", NULL, "output_type,min_percent", + "Display callchains using output_type and min percent threshold. " + "Default: fractal,0.5", &parse_callchain_opt, callchain_default_opt), OPT_STRING('d', "dsos", &dso_list_str, "dso[,dso...]", "only consider symbols in these dsos"), OPT_STRING('C', "comms", &comm_list_str, "comm[,comm...]", "only consider symbols in these comms"), OPT_STRING('S', "symbols", &sym_list_str, "symbol[,symbol...]", "only consider these symbols"), + OPT_STRING('w', "column-widths", &col_width_list_str, + "width[,width...]", + "don't try to adjust column width, use these fixed values"), + OPT_STRING('t', "field-separator", &field_sep, "separator", + "separator for columns, no spaces will be added between " + "columns '.' is reserved."), OPT_END() }; @@ -1696,7 +1670,8 @@ static void setup_sorting(void) } static void setup_list(struct strlist **list, const char *list_str, - const char *list_name) + struct sort_entry *se, const char *list_name, + FILE *fp) { if (list_str) { *list = strlist__new(true, list_str); @@ -1705,10 +1680,15 @@ static void setup_list(struct strlist **list, const char *list_str, list_name); exit(129); } + if (strlist__nr_entries(*list) == 1) { + fprintf(fp, "# %s: %s\n", list_name, + strlist__entry(*list, 0)->s); + se->elide = true; + } } } -int cmd_report(int argc, const char **argv, const char *prefix) +int cmd_report(int argc, const char **argv, const char *prefix __used) { symbol__init(); @@ -1718,9 +1698,10 @@ int cmd_report(int argc, const char **argv, const char *prefix) setup_sorting(); - if (parent_pattern != default_parent_pattern) + if (parent_pattern != default_parent_pattern) { sort_dimension__add("parent"); - else + sort_parent.elide = 1; + } else exclude_other = 0; /* @@ -1729,11 +1710,17 @@ int cmd_report(int argc, const char **argv, const char *prefix) if (argc) usage_with_options(report_usage, options); - setup_list(&dso_list, dso_list_str, "dso"); - setup_list(&comm_list, comm_list_str, "comm"); - setup_list(&sym_list, sym_list_str, "symbol"); - setup_pager(); + setup_list(&dso_list, dso_list_str, &sort_dso, "dso", stdout); + setup_list(&comm_list, comm_list_str, &sort_comm, "comm", stdout); + setup_list(&sym_list, sym_list_str, &sort_sym, "symbol", stdout); + + if (field_sep && *field_sep == '.') { + fputs("'.' is the only non valid --field-separator argument\n", + stderr); + exit(129); + } + return __cmd_report(); } diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c new file mode 100644 index 00000000000..ea9c15c0cdf --- /dev/null +++ b/tools/perf/builtin-sched.c @@ -0,0 +1,2004 @@ +#include "builtin.h" +#include "perf.h" + +#include "util/util.h" +#include "util/cache.h" +#include "util/symbol.h" +#include "util/thread.h" +#include "util/header.h" + +#include "util/parse-options.h" +#include "util/trace-event.h" + +#include "util/debug.h" + +#include <sys/types.h> +#include <sys/prctl.h> + +#include <semaphore.h> +#include <pthread.h> +#include <math.h> + +static char const *input_name = "perf.data"; +static int input; +static unsigned long page_size; +static unsigned long mmap_window = 32; + +static unsigned long total_comm = 0; + +static struct rb_root threads; +static struct thread *last_match; + +static struct perf_header *header; +static u64 sample_type; + +static char default_sort_order[] = "avg, max, switch, runtime"; +static char *sort_order = default_sort_order; + +#define PR_SET_NAME 15 /* Set process name */ +#define MAX_CPUS 4096 + +#define BUG_ON(x) assert(!(x)) + +static u64 run_measurement_overhead; +static u64 sleep_measurement_overhead; + +#define COMM_LEN 20 +#define SYM_LEN 129 + +#define MAX_PID 65536 + +static unsigned long nr_tasks; + +struct sched_atom; + +struct task_desc { + unsigned long nr; + unsigned long pid; + char comm[COMM_LEN]; + + unsigned long nr_events; + unsigned long curr_event; + struct sched_atom **atoms; + + pthread_t thread; + sem_t sleep_sem; + + sem_t ready_for_work; + sem_t work_done_sem; + + u64 cpu_usage; +}; + +enum sched_event_type { + SCHED_EVENT_RUN, + SCHED_EVENT_SLEEP, + SCHED_EVENT_WAKEUP, +}; + +struct sched_atom { + enum sched_event_type type; + u64 timestamp; + u64 duration; + unsigned long nr; + int specific_wait; + sem_t *wait_sem; + struct task_desc *wakee; +}; + +static struct task_desc *pid_to_task[MAX_PID]; + +static struct task_desc **tasks; + +static pthread_mutex_t start_work_mutex = PTHREAD_MUTEX_INITIALIZER; +static u64 start_time; + +static pthread_mutex_t work_done_wait_mutex = PTHREAD_MUTEX_INITIALIZER; + +static unsigned long nr_run_events; +static unsigned long nr_sleep_events; +static unsigned long nr_wakeup_events; + +static unsigned long nr_sleep_corrections; +static unsigned long nr_run_events_optimized; + +static unsigned long targetless_wakeups; +static unsigned long multitarget_wakeups; + +static u64 cpu_usage; +static u64 runavg_cpu_usage; +static u64 parent_cpu_usage; +static u64 runavg_parent_cpu_usage; + +static unsigned long nr_runs; +static u64 sum_runtime; +static u64 sum_fluct; +static u64 run_avg; + +static unsigned long replay_repeat = 10; +static unsigned long nr_timestamps; +static unsigned long nr_unordered_timestamps; +static unsigned long nr_state_machine_bugs; +static unsigned long nr_context_switch_bugs; +static unsigned long nr_events; +static unsigned long nr_lost_chunks; +static unsigned long nr_lost_events; + +#define TASK_STATE_TO_CHAR_STR "RSDTtZX" + +enum thread_state { + THREAD_SLEEPING = 0, + THREAD_WAIT_CPU, + THREAD_SCHED_IN, + THREAD_IGNORE +}; + +struct work_atom { + struct list_head list; + enum thread_state state; + u64 sched_out_time; + u64 wake_up_time; + u64 sched_in_time; + u64 runtime; +}; + +struct work_atoms { + struct list_head work_list; + struct thread *thread; + struct rb_node node; + u64 max_lat; + u64 total_lat; + u64 nb_atoms; + u64 total_runtime; +}; + +typedef int (*sort_fn_t)(struct work_atoms *, struct work_atoms *); + +static struct rb_root atom_root, sorted_atom_root; + +static u64 all_runtime; +static u64 all_count; + + +static u64 get_nsecs(void) +{ + struct timespec ts; + + clock_gettime(CLOCK_MONOTONIC, &ts); + + return ts.tv_sec * 1000000000ULL + ts.tv_nsec; +} + +static void burn_nsecs(u64 nsecs) +{ + u64 T0 = get_nsecs(), T1; + + do { + T1 = get_nsecs(); + } while (T1 + run_measurement_overhead < T0 + nsecs); +} + +static void sleep_nsecs(u64 nsecs) +{ + struct timespec ts; + + ts.tv_nsec = nsecs % 999999999; + ts.tv_sec = nsecs / 999999999; + + nanosleep(&ts, NULL); +} + +static void calibrate_run_measurement_overhead(void) +{ + u64 T0, T1, delta, min_delta = 1000000000ULL; + int i; + + for (i = 0; i < 10; i++) { + T0 = get_nsecs(); + burn_nsecs(0); + T1 = get_nsecs(); + delta = T1-T0; + min_delta = min(min_delta, delta); + } + run_measurement_overhead = min_delta; + + printf("run measurement overhead: %Ld nsecs\n", min_delta); +} + +static void calibrate_sleep_measurement_overhead(void) +{ + u64 T0, T1, delta, min_delta = 1000000000ULL; + int i; + + for (i = 0; i < 10; i++) { + T0 = get_nsecs(); + sleep_nsecs(10000); + T1 = get_nsecs(); + delta = T1-T0; + min_delta = min(min_delta, delta); + } + min_delta -= 10000; + sleep_measurement_overhead = min_delta; + + printf("sleep measurement overhead: %Ld nsecs\n", min_delta); +} + +static struct sched_atom * +get_new_event(struct task_desc *task, u64 timestamp) +{ + struct sched_atom *event = calloc(1, sizeof(*event)); + unsigned long idx = task->nr_events; + size_t size; + + event->timestamp = timestamp; + event->nr = idx; + + task->nr_events++; + size = sizeof(struct sched_atom *) * task->nr_events; + task->atoms = realloc(task->atoms, size); + BUG_ON(!task->atoms); + + task->atoms[idx] = event; + + return event; +} + +static struct sched_atom *last_event(struct task_desc *task) +{ + if (!task->nr_events) + return NULL; + + return task->atoms[task->nr_events - 1]; +} + +static void +add_sched_event_run(struct task_desc *task, u64 timestamp, u64 duration) +{ + struct sched_atom *event, *curr_event = last_event(task); + + /* + * optimize an existing RUN event by merging this one + * to it: + */ + if (curr_event && curr_event->type == SCHED_EVENT_RUN) { + nr_run_events_optimized++; + curr_event->duration += duration; + return; + } + + event = get_new_event(task, timestamp); + + event->type = SCHED_EVENT_RUN; + event->duration = duration; + + nr_run_events++; +} + +static void +add_sched_event_wakeup(struct task_desc *task, u64 timestamp, + struct task_desc *wakee) +{ + struct sched_atom *event, *wakee_event; + + event = get_new_event(task, timestamp); + event->type = SCHED_EVENT_WAKEUP; + event->wakee = wakee; + + wakee_event = last_event(wakee); + if (!wakee_event || wakee_event->type != SCHED_EVENT_SLEEP) { + targetless_wakeups++; + return; + } + if (wakee_event->wait_sem) { + multitarget_wakeups++; + return; + } + + wakee_event->wait_sem = calloc(1, sizeof(*wakee_event->wait_sem)); + sem_init(wakee_event->wait_sem, 0, 0); + wakee_event->specific_wait = 1; + event->wait_sem = wakee_event->wait_sem; + + nr_wakeup_events++; +} + +static void +add_sched_event_sleep(struct task_desc *task, u64 timestamp, + u64 task_state __used) +{ + struct sched_atom *event = get_new_event(task, timestamp); + + event->type = SCHED_EVENT_SLEEP; + + nr_sleep_events++; +} + +static struct task_desc *register_pid(unsigned long pid, const char *comm) +{ + struct task_desc *task; + + BUG_ON(pid >= MAX_PID); + + task = pid_to_task[pid]; + + if (task) + return task; + + task = calloc(1, sizeof(*task)); + task->pid = pid; + task->nr = nr_tasks; + strcpy(task->comm, comm); + /* + * every task starts in sleeping state - this gets ignored + * if there's no wakeup pointing to this sleep state: + */ + add_sched_event_sleep(task, 0, 0); + + pid_to_task[pid] = task; + nr_tasks++; + tasks = realloc(tasks, nr_tasks*sizeof(struct task_task *)); + BUG_ON(!tasks); + tasks[task->nr] = task; + + if (verbose) + printf("registered task #%ld, PID %ld (%s)\n", nr_tasks, pid, comm); + + return task; +} + + +static void print_task_traces(void) +{ + struct task_desc *task; + unsigned long i; + + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + printf("task %6ld (%20s:%10ld), nr_events: %ld\n", + task->nr, task->comm, task->pid, task->nr_events); + } +} + +static void add_cross_task_wakeups(void) +{ + struct task_desc *task1, *task2; + unsigned long i, j; + + for (i = 0; i < nr_tasks; i++) { + task1 = tasks[i]; + j = i + 1; + if (j == nr_tasks) + j = 0; + task2 = tasks[j]; + add_sched_event_wakeup(task1, 0, task2); + } +} + +static void +process_sched_event(struct task_desc *this_task __used, struct sched_atom *atom) +{ + int ret = 0; + u64 now; + long long delta; + + now = get_nsecs(); + delta = start_time + atom->timestamp - now; + + switch (atom->type) { + case SCHED_EVENT_RUN: + burn_nsecs(atom->duration); + break; + case SCHED_EVENT_SLEEP: + if (atom->wait_sem) + ret = sem_wait(atom->wait_sem); + BUG_ON(ret); + break; + case SCHED_EVENT_WAKEUP: + if (atom->wait_sem) + ret = sem_post(atom->wait_sem); + BUG_ON(ret); + break; + default: + BUG_ON(1); + } +} + +static u64 get_cpu_usage_nsec_parent(void) +{ + struct rusage ru; + u64 sum; + int err; + + err = getrusage(RUSAGE_SELF, &ru); + BUG_ON(err); + + sum = ru.ru_utime.tv_sec*1e9 + ru.ru_utime.tv_usec*1e3; + sum += ru.ru_stime.tv_sec*1e9 + ru.ru_stime.tv_usec*1e3; + + return sum; +} + +static u64 get_cpu_usage_nsec_self(void) +{ + char filename [] = "/proc/1234567890/sched"; + unsigned long msecs, nsecs; + char *line = NULL; + u64 total = 0; + size_t len = 0; + ssize_t chars; + FILE *file; + int ret; + + sprintf(filename, "/proc/%d/sched", getpid()); + file = fopen(filename, "r"); + BUG_ON(!file); + + while ((chars = getline(&line, &len, file)) != -1) { + ret = sscanf(line, "se.sum_exec_runtime : %ld.%06ld\n", + &msecs, &nsecs); + if (ret == 2) { + total = msecs*1e6 + nsecs; + break; + } + } + if (line) + free(line); + fclose(file); + + return total; +} + +static void *thread_func(void *ctx) +{ + struct task_desc *this_task = ctx; + u64 cpu_usage_0, cpu_usage_1; + unsigned long i, ret; + char comm2[22]; + + sprintf(comm2, ":%s", this_task->comm); + prctl(PR_SET_NAME, comm2); + +again: + ret = sem_post(&this_task->ready_for_work); + BUG_ON(ret); + ret = pthread_mutex_lock(&start_work_mutex); + BUG_ON(ret); + ret = pthread_mutex_unlock(&start_work_mutex); + BUG_ON(ret); + + cpu_usage_0 = get_cpu_usage_nsec_self(); + + for (i = 0; i < this_task->nr_events; i++) { + this_task->curr_event = i; + process_sched_event(this_task, this_task->atoms[i]); + } + + cpu_usage_1 = get_cpu_usage_nsec_self(); + this_task->cpu_usage = cpu_usage_1 - cpu_usage_0; + + ret = sem_post(&this_task->work_done_sem); + BUG_ON(ret); + + ret = pthread_mutex_lock(&work_done_wait_mutex); + BUG_ON(ret); + ret = pthread_mutex_unlock(&work_done_wait_mutex); + BUG_ON(ret); + + goto again; +} + +static void create_tasks(void) +{ + struct task_desc *task; + pthread_attr_t attr; + unsigned long i; + int err; + + err = pthread_attr_init(&attr); + BUG_ON(err); + err = pthread_attr_setstacksize(&attr, (size_t)(16*1024)); + BUG_ON(err); + err = pthread_mutex_lock(&start_work_mutex); + BUG_ON(err); + err = pthread_mutex_lock(&work_done_wait_mutex); + BUG_ON(err); + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + sem_init(&task->sleep_sem, 0, 0); + sem_init(&task->ready_for_work, 0, 0); + sem_init(&task->work_done_sem, 0, 0); + task->curr_event = 0; + err = pthread_create(&task->thread, &attr, thread_func, task); + BUG_ON(err); + } +} + +static void wait_for_tasks(void) +{ + u64 cpu_usage_0, cpu_usage_1; + struct task_desc *task; + unsigned long i, ret; + + start_time = get_nsecs(); + cpu_usage = 0; + pthread_mutex_unlock(&work_done_wait_mutex); + + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + ret = sem_wait(&task->ready_for_work); + BUG_ON(ret); + sem_init(&task->ready_for_work, 0, 0); + } + ret = pthread_mutex_lock(&work_done_wait_mutex); + BUG_ON(ret); + + cpu_usage_0 = get_cpu_usage_nsec_parent(); + + pthread_mutex_unlock(&start_work_mutex); + + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + ret = sem_wait(&task->work_done_sem); + BUG_ON(ret); + sem_init(&task->work_done_sem, 0, 0); + cpu_usage += task->cpu_usage; + task->cpu_usage = 0; + } + + cpu_usage_1 = get_cpu_usage_nsec_parent(); + if (!runavg_cpu_usage) + runavg_cpu_usage = cpu_usage; + runavg_cpu_usage = (runavg_cpu_usage*9 + cpu_usage)/10; + + parent_cpu_usage = cpu_usage_1 - cpu_usage_0; + if (!runavg_parent_cpu_usage) + runavg_parent_cpu_usage = parent_cpu_usage; + runavg_parent_cpu_usage = (runavg_parent_cpu_usage*9 + + parent_cpu_usage)/10; + + ret = pthread_mutex_lock(&start_work_mutex); + BUG_ON(ret); + + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + sem_init(&task->sleep_sem, 0, 0); + task->curr_event = 0; + } +} + +static void run_one_test(void) +{ + u64 T0, T1, delta, avg_delta, fluct, std_dev; + + T0 = get_nsecs(); + wait_for_tasks(); + T1 = get_nsecs(); + + delta = T1 - T0; + sum_runtime += delta; + nr_runs++; + + avg_delta = sum_runtime / nr_runs; + if (delta < avg_delta) + fluct = avg_delta - delta; + else + fluct = delta - avg_delta; + sum_fluct += fluct; + std_dev = sum_fluct / nr_runs / sqrt(nr_runs); + if (!run_avg) + run_avg = delta; + run_avg = (run_avg*9 + delta)/10; + + printf("#%-3ld: %0.3f, ", + nr_runs, (double)delta/1000000.0); + + printf("ravg: %0.2f, ", + (double)run_avg/1e6); + + printf("cpu: %0.2f / %0.2f", + (double)cpu_usage/1e6, (double)runavg_cpu_usage/1e6); + +#if 0 + /* + * rusage statistics done by the parent, these are less + * accurate than the sum_exec_runtime based statistics: + */ + printf(" [%0.2f / %0.2f]", + (double)parent_cpu_usage/1e6, + (double)runavg_parent_cpu_usage/1e6); +#endif + + printf("\n"); + + if (nr_sleep_corrections) + printf(" (%ld sleep corrections)\n", nr_sleep_corrections); + nr_sleep_corrections = 0; +} + +static void test_calibrations(void) +{ + u64 T0, T1; + + T0 = get_nsecs(); + burn_nsecs(1e6); + T1 = get_nsecs(); + + printf("the run test took %Ld nsecs\n", T1-T0); + + T0 = get_nsecs(); + sleep_nsecs(1e6); + T1 = get_nsecs(); + + printf("the sleep test took %Ld nsecs\n", T1-T0); +} + +static int +process_comm_event(event_t *event, unsigned long offset, unsigned long head) +{ + struct thread *thread; + + thread = threads__findnew(event->comm.pid, &threads, &last_match); + + dump_printf("%p [%p]: perf_event_comm: %s:%d\n", + (void *)(offset + head), + (void *)(long)(event->header.size), + event->comm.comm, event->comm.pid); + + if (thread == NULL || + thread__set_comm(thread, event->comm.comm)) { + dump_printf("problem processing perf_event_comm, skipping event.\n"); + return -1; + } + total_comm++; + + return 0; +} + + +struct raw_event_sample { + u32 size; + char data[0]; +}; + +#define FILL_FIELD(ptr, field, event, data) \ + ptr.field = (typeof(ptr.field)) raw_field_value(event, #field, data) + +#define FILL_ARRAY(ptr, array, event, data) \ +do { \ + void *__array = raw_field_ptr(event, #array, data); \ + memcpy(ptr.array, __array, sizeof(ptr.array)); \ +} while(0) + +#define FILL_COMMON_FIELDS(ptr, event, data) \ +do { \ + FILL_FIELD(ptr, common_type, event, data); \ + FILL_FIELD(ptr, common_flags, event, data); \ + FILL_FIELD(ptr, common_preempt_count, event, data); \ + FILL_FIELD(ptr, common_pid, event, data); \ + FILL_FIELD(ptr, common_tgid, event, data); \ +} while (0) + + + +struct trace_switch_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char prev_comm[16]; + u32 prev_pid; + u32 prev_prio; + u64 prev_state; + char next_comm[16]; + u32 next_pid; + u32 next_prio; +}; + +struct trace_runtime_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char comm[16]; + u32 pid; + u64 runtime; + u64 vruntime; +}; + +struct trace_wakeup_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char comm[16]; + u32 pid; + + u32 prio; + u32 success; + u32 cpu; +}; + +struct trace_fork_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char parent_comm[16]; + u32 parent_pid; + char child_comm[16]; + u32 child_pid; +}; + +struct trace_sched_handler { + void (*switch_event)(struct trace_switch_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*runtime_event)(struct trace_runtime_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*wakeup_event)(struct trace_wakeup_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*fork_event)(struct trace_fork_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); +}; + + +static void +replay_wakeup_event(struct trace_wakeup_event *wakeup_event, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct task_desc *waker, *wakee; + + if (verbose) { + printf("sched_wakeup event %p\n", event); + + printf(" ... pid %d woke up %s/%d\n", + wakeup_event->common_pid, + wakeup_event->comm, + wakeup_event->pid); + } + + waker = register_pid(wakeup_event->common_pid, "<unknown>"); + wakee = register_pid(wakeup_event->pid, wakeup_event->comm); + + add_sched_event_wakeup(waker, timestamp, wakee); +} + +static u64 cpu_last_switched[MAX_CPUS]; + +static void +replay_switch_event(struct trace_switch_event *switch_event, + struct event *event, + int cpu, + u64 timestamp, + struct thread *thread __used) +{ + struct task_desc *prev, *next; + u64 timestamp0; + s64 delta; + + if (verbose) + printf("sched_switch event %p\n", event); + + if (cpu >= MAX_CPUS || cpu < 0) + return; + + timestamp0 = cpu_last_switched[cpu]; + if (timestamp0) + delta = timestamp - timestamp0; + else + delta = 0; + + if (delta < 0) + die("hm, delta: %Ld < 0 ?\n", delta); + + if (verbose) { + printf(" ... switch from %s/%d to %s/%d [ran %Ld nsecs]\n", + switch_event->prev_comm, switch_event->prev_pid, + switch_event->next_comm, switch_event->next_pid, + delta); + } + + prev = register_pid(switch_event->prev_pid, switch_event->prev_comm); + next = register_pid(switch_event->next_pid, switch_event->next_comm); + + cpu_last_switched[cpu] = timestamp; + + add_sched_event_run(prev, timestamp, delta); + add_sched_event_sleep(prev, timestamp, switch_event->prev_state); +} + + +static void +replay_fork_event(struct trace_fork_event *fork_event, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + if (verbose) { + printf("sched_fork event %p\n", event); + printf("... parent: %s/%d\n", fork_event->parent_comm, fork_event->parent_pid); + printf("... child: %s/%d\n", fork_event->child_comm, fork_event->child_pid); + } + register_pid(fork_event->parent_pid, fork_event->parent_comm); + register_pid(fork_event->child_pid, fork_event->child_comm); +} + +static struct trace_sched_handler replay_ops = { + .wakeup_event = replay_wakeup_event, + .switch_event = replay_switch_event, + .fork_event = replay_fork_event, +}; + +struct sort_dimension { + const char *name; + sort_fn_t cmp; + struct list_head list; +}; + +static LIST_HEAD(cmp_pid); + +static int +thread_lat_cmp(struct list_head *list, struct work_atoms *l, struct work_atoms *r) +{ + struct sort_dimension *sort; + int ret = 0; + + BUG_ON(list_empty(list)); + + list_for_each_entry(sort, list, list) { + ret = sort->cmp(l, r); + if (ret) + return ret; + } + + return ret; +} + +static struct work_atoms * +thread_atoms_search(struct rb_root *root, struct thread *thread, + struct list_head *sort_list) +{ + struct rb_node *node = root->rb_node; + struct work_atoms key = { .thread = thread }; + + while (node) { + struct work_atoms *atoms; + int cmp; + + atoms = container_of(node, struct work_atoms, node); + + cmp = thread_lat_cmp(sort_list, &key, atoms); + if (cmp > 0) + node = node->rb_left; + else if (cmp < 0) + node = node->rb_right; + else { + BUG_ON(thread != atoms->thread); + return atoms; + } + } + return NULL; +} + +static void +__thread_latency_insert(struct rb_root *root, struct work_atoms *data, + struct list_head *sort_list) +{ + struct rb_node **new = &(root->rb_node), *parent = NULL; + + while (*new) { + struct work_atoms *this; + int cmp; + + this = container_of(*new, struct work_atoms, node); + parent = *new; + + cmp = thread_lat_cmp(sort_list, data, this); + + if (cmp > 0) + new = &((*new)->rb_left); + else + new = &((*new)->rb_right); + } + + rb_link_node(&data->node, parent, new); + rb_insert_color(&data->node, root); +} + +static void thread_atoms_insert(struct thread *thread) +{ + struct work_atoms *atoms; + + atoms = calloc(sizeof(*atoms), 1); + if (!atoms) + die("No memory"); + + atoms->thread = thread; + INIT_LIST_HEAD(&atoms->work_list); + __thread_latency_insert(&atom_root, atoms, &cmp_pid); +} + +static void +latency_fork_event(struct trace_fork_event *fork_event __used, + struct event *event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + /* should insert the newcomer */ +} + +__used +static char sched_out_state(struct trace_switch_event *switch_event) +{ + const char *str = TASK_STATE_TO_CHAR_STR; + + return str[switch_event->prev_state]; +} + +static void +add_sched_out_event(struct work_atoms *atoms, + char run_state, + u64 timestamp) +{ + struct work_atom *atom; + + atom = calloc(sizeof(*atom), 1); + if (!atom) + die("Non memory"); + + atom->sched_out_time = timestamp; + + if (run_state == 'R') { + atom->state = THREAD_WAIT_CPU; + atom->wake_up_time = atom->sched_out_time; + } + + list_add_tail(&atom->list, &atoms->work_list); +} + +static void +add_runtime_event(struct work_atoms *atoms, u64 delta, u64 timestamp __used) +{ + struct work_atom *atom; + + BUG_ON(list_empty(&atoms->work_list)); + + atom = list_entry(atoms->work_list.prev, struct work_atom, list); + + atom->runtime += delta; + atoms->total_runtime += delta; +} + +static void +add_sched_in_event(struct work_atoms *atoms, u64 timestamp) +{ + struct work_atom *atom; + u64 delta; + + if (list_empty(&atoms->work_list)) + return; + + atom = list_entry(atoms->work_list.prev, struct work_atom, list); + + if (atom->state != THREAD_WAIT_CPU) + return; + + if (timestamp < atom->wake_up_time) { + atom->state = THREAD_IGNORE; + return; + } + + atom->state = THREAD_SCHED_IN; + atom->sched_in_time = timestamp; + + delta = atom->sched_in_time - atom->wake_up_time; + atoms->total_lat += delta; + if (delta > atoms->max_lat) + atoms->max_lat = delta; + atoms->nb_atoms++; +} + +static void +latency_switch_event(struct trace_switch_event *switch_event, + struct event *event __used, + int cpu, + u64 timestamp, + struct thread *thread __used) +{ + struct work_atoms *out_events, *in_events; + struct thread *sched_out, *sched_in; + u64 timestamp0; + s64 delta; + + BUG_ON(cpu >= MAX_CPUS || cpu < 0); + + timestamp0 = cpu_last_switched[cpu]; + cpu_last_switched[cpu] = timestamp; + if (timestamp0) + delta = timestamp - timestamp0; + else + delta = 0; + + if (delta < 0) + die("hm, delta: %Ld < 0 ?\n", delta); + + + sched_out = threads__findnew(switch_event->prev_pid, &threads, &last_match); + sched_in = threads__findnew(switch_event->next_pid, &threads, &last_match); + + out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid); + if (!out_events) { + thread_atoms_insert(sched_out); + out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid); + if (!out_events) + die("out-event: Internal tree error"); + } + add_sched_out_event(out_events, sched_out_state(switch_event), timestamp); + + in_events = thread_atoms_search(&atom_root, sched_in, &cmp_pid); + if (!in_events) { + thread_atoms_insert(sched_in); + in_events = thread_atoms_search(&atom_root, sched_in, &cmp_pid); + if (!in_events) + die("in-event: Internal tree error"); + /* + * Take came in we have not heard about yet, + * add in an initial atom in runnable state: + */ + add_sched_out_event(in_events, 'R', timestamp); + } + add_sched_in_event(in_events, timestamp); +} + +static void +latency_runtime_event(struct trace_runtime_event *runtime_event, + struct event *event __used, + int cpu, + u64 timestamp, + struct thread *this_thread __used) +{ + struct work_atoms *atoms; + struct thread *thread; + + BUG_ON(cpu >= MAX_CPUS || cpu < 0); + + thread = threads__findnew(runtime_event->pid, &threads, &last_match); + atoms = thread_atoms_search(&atom_root, thread, &cmp_pid); + if (!atoms) { + thread_atoms_insert(thread); + atoms = thread_atoms_search(&atom_root, thread, &cmp_pid); + if (!atoms) + die("in-event: Internal tree error"); + add_sched_out_event(atoms, 'R', timestamp); + } + + add_runtime_event(atoms, runtime_event->runtime, timestamp); +} + +static void +latency_wakeup_event(struct trace_wakeup_event *wakeup_event, + struct event *__event __used, + int cpu __used, + u64 timestamp, + struct thread *thread __used) +{ + struct work_atoms *atoms; + struct work_atom *atom; + struct thread *wakee; + + /* Note for later, it may be interesting to observe the failing cases */ + if (!wakeup_event->success) + return; + + wakee = threads__findnew(wakeup_event->pid, &threads, &last_match); + atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid); + if (!atoms) { + thread_atoms_insert(wakee); + atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid); + if (!atoms) + die("wakeup-event: Internal tree error"); + add_sched_out_event(atoms, 'S', timestamp); + } + + BUG_ON(list_empty(&atoms->work_list)); + + atom = list_entry(atoms->work_list.prev, struct work_atom, list); + + if (atom->state != THREAD_SLEEPING) + nr_state_machine_bugs++; + + nr_timestamps++; + if (atom->sched_out_time > timestamp) { + nr_unordered_timestamps++; + return; + } + + atom->state = THREAD_WAIT_CPU; + atom->wake_up_time = timestamp; +} + +static struct trace_sched_handler lat_ops = { + .wakeup_event = latency_wakeup_event, + .switch_event = latency_switch_event, + .runtime_event = latency_runtime_event, + .fork_event = latency_fork_event, +}; + +static void output_lat_thread(struct work_atoms *work_list) +{ + int i; + int ret; + u64 avg; + + if (!work_list->nb_atoms) + return; + /* + * Ignore idle threads: + */ + if (!strcmp(work_list->thread->comm, "swapper")) + return; + + all_runtime += work_list->total_runtime; + all_count += work_list->nb_atoms; + + ret = printf(" %s:%d ", work_list->thread->comm, work_list->thread->pid); + + for (i = 0; i < 24 - ret; i++) + printf(" "); + + avg = work_list->total_lat / work_list->nb_atoms; + + printf("|%11.3f ms |%9llu | avg:%9.3f ms | max:%9.3f ms |\n", + (double)work_list->total_runtime / 1e6, + work_list->nb_atoms, (double)avg / 1e6, + (double)work_list->max_lat / 1e6); +} + +static int pid_cmp(struct work_atoms *l, struct work_atoms *r) +{ + if (l->thread->pid < r->thread->pid) + return -1; + if (l->thread->pid > r->thread->pid) + return 1; + + return 0; +} + +static struct sort_dimension pid_sort_dimension = { + .name = "pid", + .cmp = pid_cmp, +}; + +static int avg_cmp(struct work_atoms *l, struct work_atoms *r) +{ + u64 avgl, avgr; + + if (!l->nb_atoms) + return -1; + + if (!r->nb_atoms) + return 1; + + avgl = l->total_lat / l->nb_atoms; + avgr = r->total_lat / r->nb_atoms; + + if (avgl < avgr) + return -1; + if (avgl > avgr) + return 1; + + return 0; +} + +static struct sort_dimension avg_sort_dimension = { + .name = "avg", + .cmp = avg_cmp, +}; + +static int max_cmp(struct work_atoms *l, struct work_atoms *r) +{ + if (l->max_lat < r->max_lat) + return -1; + if (l->max_lat > r->max_lat) + return 1; + + return 0; +} + +static struct sort_dimension max_sort_dimension = { + .name = "max", + .cmp = max_cmp, +}; + +static int switch_cmp(struct work_atoms *l, struct work_atoms *r) +{ + if (l->nb_atoms < r->nb_atoms) + return -1; + if (l->nb_atoms > r->nb_atoms) + return 1; + + return 0; +} + +static struct sort_dimension switch_sort_dimension = { + .name = "switch", + .cmp = switch_cmp, +}; + +static int runtime_cmp(struct work_atoms *l, struct work_atoms *r) +{ + if (l->total_runtime < r->total_runtime) + return -1; + if (l->total_runtime > r->total_runtime) + return 1; + + return 0; +} + +static struct sort_dimension runtime_sort_dimension = { + .name = "runtime", + .cmp = runtime_cmp, +}; + +static struct sort_dimension *available_sorts[] = { + &pid_sort_dimension, + &avg_sort_dimension, + &max_sort_dimension, + &switch_sort_dimension, + &runtime_sort_dimension, +}; + +#define NB_AVAILABLE_SORTS (int)(sizeof(available_sorts) / sizeof(struct sort_dimension *)) + +static LIST_HEAD(sort_list); + +static int sort_dimension__add(char *tok, struct list_head *list) +{ + int i; + + for (i = 0; i < NB_AVAILABLE_SORTS; i++) { + if (!strcmp(available_sorts[i]->name, tok)) { + list_add_tail(&available_sorts[i]->list, list); + + return 0; + } + } + + return -1; +} + +static void setup_sorting(void); + +static void sort_lat(void) +{ + struct rb_node *node; + + for (;;) { + struct work_atoms *data; + node = rb_first(&atom_root); + if (!node) + break; + + rb_erase(node, &atom_root); + data = rb_entry(node, struct work_atoms, node); + __thread_latency_insert(&sorted_atom_root, data, &sort_list); + } +} + +static struct trace_sched_handler *trace_handler; + +static void +process_sched_wakeup_event(struct raw_event_sample *raw, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_wakeup_event wakeup_event; + + FILL_COMMON_FIELDS(wakeup_event, event, raw->data); + + FILL_ARRAY(wakeup_event, comm, event, raw->data); + FILL_FIELD(wakeup_event, pid, event, raw->data); + FILL_FIELD(wakeup_event, prio, event, raw->data); + FILL_FIELD(wakeup_event, success, event, raw->data); + FILL_FIELD(wakeup_event, cpu, event, raw->data); + + if (trace_handler->wakeup_event) + trace_handler->wakeup_event(&wakeup_event, event, cpu, timestamp, thread); +} + +/* + * Track the current task - that way we can know whether there's any + * weird events, such as a task being switched away that is not current. + */ +static int max_cpu; + +static u32 curr_pid[MAX_CPUS] = { [0 ... MAX_CPUS-1] = -1 }; + +static struct thread *curr_thread[MAX_CPUS]; + +static char next_shortname1 = 'A'; +static char next_shortname2 = '0'; + +static void +map_switch_event(struct trace_switch_event *switch_event, + struct event *event __used, + int this_cpu, + u64 timestamp, + struct thread *thread __used) +{ + struct thread *sched_out, *sched_in; + int new_shortname; + u64 timestamp0; + s64 delta; + int cpu; + + BUG_ON(this_cpu >= MAX_CPUS || this_cpu < 0); + + if (this_cpu > max_cpu) + max_cpu = this_cpu; + + timestamp0 = cpu_last_switched[this_cpu]; + cpu_last_switched[this_cpu] = timestamp; + if (timestamp0) + delta = timestamp - timestamp0; + else + delta = 0; + + if (delta < 0) + die("hm, delta: %Ld < 0 ?\n", delta); + + + sched_out = threads__findnew(switch_event->prev_pid, &threads, &last_match); + sched_in = threads__findnew(switch_event->next_pid, &threads, &last_match); + + curr_thread[this_cpu] = sched_in; + + printf(" "); + + new_shortname = 0; + if (!sched_in->shortname[0]) { + sched_in->shortname[0] = next_shortname1; + sched_in->shortname[1] = next_shortname2; + + if (next_shortname1 < 'Z') { + next_shortname1++; + } else { + next_shortname1='A'; + if (next_shortname2 < '9') { + next_shortname2++; + } else { + next_shortname2='0'; + } + } + new_shortname = 1; + } + + for (cpu = 0; cpu <= max_cpu; cpu++) { + if (cpu != this_cpu) + printf(" "); + else + printf("*"); + + if (curr_thread[cpu]) { + if (curr_thread[cpu]->pid) + printf("%2s ", curr_thread[cpu]->shortname); + else + printf(". "); + } else + printf(" "); + } + + printf(" %12.6f secs ", (double)timestamp/1e9); + if (new_shortname) { + printf("%s => %s:%d\n", + sched_in->shortname, sched_in->comm, sched_in->pid); + } else { + printf("\n"); + } +} + + +static void +process_sched_switch_event(struct raw_event_sample *raw, + struct event *event, + int this_cpu, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_switch_event switch_event; + + FILL_COMMON_FIELDS(switch_event, event, raw->data); + + FILL_ARRAY(switch_event, prev_comm, event, raw->data); + FILL_FIELD(switch_event, prev_pid, event, raw->data); + FILL_FIELD(switch_event, prev_prio, event, raw->data); + FILL_FIELD(switch_event, prev_state, event, raw->data); + FILL_ARRAY(switch_event, next_comm, event, raw->data); + FILL_FIELD(switch_event, next_pid, event, raw->data); + FILL_FIELD(switch_event, next_prio, event, raw->data); + + if (curr_pid[this_cpu] != (u32)-1) { + /* + * Are we trying to switch away a PID that is + * not current? + */ + if (curr_pid[this_cpu] != switch_event.prev_pid) + nr_context_switch_bugs++; + } + if (trace_handler->switch_event) + trace_handler->switch_event(&switch_event, event, this_cpu, timestamp, thread); + + curr_pid[this_cpu] = switch_event.next_pid; +} + +static void +process_sched_runtime_event(struct raw_event_sample *raw, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_runtime_event runtime_event; + + FILL_ARRAY(runtime_event, comm, event, raw->data); + FILL_FIELD(runtime_event, pid, event, raw->data); + FILL_FIELD(runtime_event, runtime, event, raw->data); + FILL_FIELD(runtime_event, vruntime, event, raw->data); + + if (trace_handler->runtime_event) + trace_handler->runtime_event(&runtime_event, event, cpu, timestamp, thread); +} + +static void +process_sched_fork_event(struct raw_event_sample *raw, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_fork_event fork_event; + + FILL_COMMON_FIELDS(fork_event, event, raw->data); + + FILL_ARRAY(fork_event, parent_comm, event, raw->data); + FILL_FIELD(fork_event, parent_pid, event, raw->data); + FILL_ARRAY(fork_event, child_comm, event, raw->data); + FILL_FIELD(fork_event, child_pid, event, raw->data); + + if (trace_handler->fork_event) + trace_handler->fork_event(&fork_event, event, cpu, timestamp, thread); +} + +static void +process_sched_exit_event(struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + if (verbose) + printf("sched_exit event %p\n", event); +} + +static void +process_raw_event(event_t *raw_event __used, void *more_data, + int cpu, u64 timestamp, struct thread *thread) +{ + struct raw_event_sample *raw = more_data; + struct event *event; + int type; + + type = trace_parse_common_type(raw->data); + event = trace_find_event(type); + + if (!strcmp(event->name, "sched_switch")) + process_sched_switch_event(raw, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_stat_runtime")) + process_sched_runtime_event(raw, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_wakeup")) + process_sched_wakeup_event(raw, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_wakeup_new")) + process_sched_wakeup_event(raw, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_process_fork")) + process_sched_fork_event(raw, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_process_exit")) + process_sched_exit_event(event, cpu, timestamp, thread); +} + +static int +process_sample_event(event_t *event, unsigned long offset, unsigned long head) +{ + char level; + int show = 0; + struct dso *dso = NULL; + struct thread *thread; + u64 ip = event->ip.ip; + u64 timestamp = -1; + u32 cpu = -1; + u64 period = 1; + void *more_data = event->ip.__more_data; + int cpumode; + + thread = threads__findnew(event->ip.pid, &threads, &last_match); + + if (sample_type & PERF_SAMPLE_TIME) { + timestamp = *(u64 *)more_data; + more_data += sizeof(u64); + } + + if (sample_type & PERF_SAMPLE_CPU) { + cpu = *(u32 *)more_data; + more_data += sizeof(u32); + more_data += sizeof(u32); /* reserved */ + } + + if (sample_type & PERF_SAMPLE_PERIOD) { + period = *(u64 *)more_data; + more_data += sizeof(u64); + } + + dump_printf("%p [%p]: PERF_RECORD_SAMPLE (IP, %d): %d/%d: %p period: %Ld\n", + (void *)(offset + head), + (void *)(long)(event->header.size), + event->header.misc, + event->ip.pid, event->ip.tid, + (void *)(long)ip, + (long long)period); + + dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid); + + if (thread == NULL) { + eprintf("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + + if (cpumode == PERF_RECORD_MISC_KERNEL) { + show = SHOW_KERNEL; + level = 'k'; + + dso = kernel_dso; + + dump_printf(" ...... dso: %s\n", dso->name); + + } else if (cpumode == PERF_RECORD_MISC_USER) { + + show = SHOW_USER; + level = '.'; + + } else { + show = SHOW_HV; + level = 'H'; + + dso = hypervisor_dso; + + dump_printf(" ...... dso: [hypervisor]\n"); + } + + if (sample_type & PERF_SAMPLE_RAW) + process_raw_event(event, more_data, cpu, timestamp, thread); + + return 0; +} + +static int +process_event(event_t *event, unsigned long offset, unsigned long head) +{ + trace_event(event); + + nr_events++; + switch (event->header.type) { + case PERF_RECORD_MMAP: + return 0; + case PERF_RECORD_LOST: + nr_lost_chunks++; + nr_lost_events += event->lost.lost; + return 0; + + case PERF_RECORD_COMM: + return process_comm_event(event, offset, head); + + case PERF_RECORD_EXIT ... PERF_RECORD_READ: + return 0; + + case PERF_RECORD_SAMPLE: + return process_sample_event(event, offset, head); + + case PERF_RECORD_MAX: + default: + return -1; + } + + return 0; +} + +static int read_events(void) +{ + int ret, rc = EXIT_FAILURE; + unsigned long offset = 0; + unsigned long head = 0; + struct stat perf_stat; + event_t *event; + uint32_t size; + char *buf; + + trace_report(); + register_idle_thread(&threads, &last_match); + + input = open(input_name, O_RDONLY); + if (input < 0) { + perror("failed to open file"); + exit(-1); + } + + ret = fstat(input, &perf_stat); + if (ret < 0) { + perror("failed to stat file"); + exit(-1); + } + + if (!perf_stat.st_size) { + fprintf(stderr, "zero-sized file, nothing to do!\n"); + exit(0); + } + header = perf_header__read(input); + head = header->data_offset; + sample_type = perf_header__sample_type(header); + + if (!(sample_type & PERF_SAMPLE_RAW)) + die("No trace sample to read. Did you call perf record " + "without -R?"); + + if (load_kernel() < 0) { + perror("failed to load kernel symbols"); + return EXIT_FAILURE; + } + +remap: + buf = (char *)mmap(NULL, page_size * mmap_window, PROT_READ, + MAP_SHARED, input, offset); + if (buf == MAP_FAILED) { + perror("failed to mmap file"); + exit(-1); + } + +more: + event = (event_t *)(buf + head); + + size = event->header.size; + if (!size) + size = 8; + + if (head + event->header.size >= page_size * mmap_window) { + unsigned long shift = page_size * (head / page_size); + int res; + + res = munmap(buf, page_size * mmap_window); + assert(res == 0); + + offset += shift; + head -= shift; + goto remap; + } + + size = event->header.size; + + + if (!size || process_event(event, offset, head) < 0) { + + /* + * assume we lost track of the stream, check alignment, and + * increment a single u64 in the hope to catch on again 'soon'. + */ + + if (unlikely(head & 7)) + head &= ~7ULL; + + size = 8; + } + + head += size; + + if (offset + head < (unsigned long)perf_stat.st_size) + goto more; + + rc = EXIT_SUCCESS; + close(input); + + return rc; +} + +static void print_bad_events(void) +{ + if (nr_unordered_timestamps && nr_timestamps) { + printf(" INFO: %.3f%% unordered timestamps (%ld out of %ld)\n", + (double)nr_unordered_timestamps/(double)nr_timestamps*100.0, + nr_unordered_timestamps, nr_timestamps); + } + if (nr_lost_events && nr_events) { + printf(" INFO: %.3f%% lost events (%ld out of %ld, in %ld chunks)\n", + (double)nr_lost_events/(double)nr_events*100.0, + nr_lost_events, nr_events, nr_lost_chunks); + } + if (nr_state_machine_bugs && nr_timestamps) { + printf(" INFO: %.3f%% state machine bugs (%ld out of %ld)", + (double)nr_state_machine_bugs/(double)nr_timestamps*100.0, + nr_state_machine_bugs, nr_timestamps); + if (nr_lost_events) + printf(" (due to lost events?)"); + printf("\n"); + } + if (nr_context_switch_bugs && nr_timestamps) { + printf(" INFO: %.3f%% context switch bugs (%ld out of %ld)", + (double)nr_context_switch_bugs/(double)nr_timestamps*100.0, + nr_context_switch_bugs, nr_timestamps); + if (nr_lost_events) + printf(" (due to lost events?)"); + printf("\n"); + } +} + +static void __cmd_lat(void) +{ + struct rb_node *next; + + setup_pager(); + read_events(); + sort_lat(); + + printf("\n -----------------------------------------------------------------------------------------\n"); + printf(" Task | Runtime ms | Switches | Average delay ms | Maximum delay ms |\n"); + printf(" -----------------------------------------------------------------------------------------\n"); + + next = rb_first(&sorted_atom_root); + + while (next) { + struct work_atoms *work_list; + + work_list = rb_entry(next, struct work_atoms, node); + output_lat_thread(work_list); + next = rb_next(next); + } + + printf(" -----------------------------------------------------------------------------------------\n"); + printf(" TOTAL: |%11.3f ms |%9Ld |\n", + (double)all_runtime/1e6, all_count); + + printf(" ---------------------------------------------------\n"); + + print_bad_events(); + printf("\n"); + +} + +static struct trace_sched_handler map_ops = { + .wakeup_event = NULL, + .switch_event = map_switch_event, + .runtime_event = NULL, + .fork_event = NULL, +}; + +static void __cmd_map(void) +{ + max_cpu = sysconf(_SC_NPROCESSORS_CONF); + + setup_pager(); + read_events(); + print_bad_events(); +} + +static void __cmd_replay(void) +{ + unsigned long i; + + calibrate_run_measurement_overhead(); + calibrate_sleep_measurement_overhead(); + + test_calibrations(); + + read_events(); + + printf("nr_run_events: %ld\n", nr_run_events); + printf("nr_sleep_events: %ld\n", nr_sleep_events); + printf("nr_wakeup_events: %ld\n", nr_wakeup_events); + + if (targetless_wakeups) + printf("target-less wakeups: %ld\n", targetless_wakeups); + if (multitarget_wakeups) + printf("multi-target wakeups: %ld\n", multitarget_wakeups); + if (nr_run_events_optimized) + printf("run atoms optimized: %ld\n", + nr_run_events_optimized); + + print_task_traces(); + add_cross_task_wakeups(); + + create_tasks(); + printf("------------------------------------------------------------\n"); + for (i = 0; i < replay_repeat; i++) + run_one_test(); +} + + +static const char * const sched_usage[] = { + "perf sched [<options>] {record|latency|map|replay|trace}", + NULL +}; + +static const struct option sched_options[] = { + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_BOOLEAN('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_END() +}; + +static const char * const latency_usage[] = { + "perf sched latency [<options>]", + NULL +}; + +static const struct option latency_options[] = { + OPT_STRING('s', "sort", &sort_order, "key[,key2...]", + "sort by key(s): runtime, switch, avg, max"), + OPT_BOOLEAN('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_END() +}; + +static const char * const replay_usage[] = { + "perf sched replay [<options>]", + NULL +}; + +static const struct option replay_options[] = { + OPT_INTEGER('r', "repeat", &replay_repeat, + "repeat the workload replay N times (-1: infinite)"), + OPT_BOOLEAN('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_END() +}; + +static void setup_sorting(void) +{ + char *tmp, *tok, *str = strdup(sort_order); + + for (tok = strtok_r(str, ", ", &tmp); + tok; tok = strtok_r(NULL, ", ", &tmp)) { + if (sort_dimension__add(tok, &sort_list) < 0) { + error("Unknown --sort key: `%s'", tok); + usage_with_options(latency_usage, latency_options); + } + } + + free(str); + + sort_dimension__add((char *)"pid", &cmp_pid); +} + +static const char *record_args[] = { + "record", + "-a", + "-R", + "-M", + "-f", + "-m", "1024", + "-c", "1", + "-e", "sched:sched_switch:r", + "-e", "sched:sched_stat_wait:r", + "-e", "sched:sched_stat_sleep:r", + "-e", "sched:sched_stat_iowait:r", + "-e", "sched:sched_stat_runtime:r", + "-e", "sched:sched_process_exit:r", + "-e", "sched:sched_process_fork:r", + "-e", "sched:sched_wakeup:r", + "-e", "sched:sched_migrate_task:r", +}; + +static int __cmd_record(int argc, const char **argv) +{ + unsigned int rec_argc, i, j; + const char **rec_argv; + + rec_argc = ARRAY_SIZE(record_args) + argc - 1; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + + for (i = 0; i < ARRAY_SIZE(record_args); i++) + rec_argv[i] = strdup(record_args[i]); + + for (j = 1; j < (unsigned int)argc; j++, i++) + rec_argv[i] = argv[j]; + + BUG_ON(i != rec_argc); + + return cmd_record(i, rec_argv, NULL); +} + +int cmd_sched(int argc, const char **argv, const char *prefix __used) +{ + symbol__init(); + page_size = getpagesize(); + + argc = parse_options(argc, argv, sched_options, sched_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + if (!argc) + usage_with_options(sched_usage, sched_options); + + if (!strncmp(argv[0], "rec", 3)) { + return __cmd_record(argc, argv); + } else if (!strncmp(argv[0], "lat", 3)) { + trace_handler = &lat_ops; + if (argc > 1) { + argc = parse_options(argc, argv, latency_options, latency_usage, 0); + if (argc) + usage_with_options(latency_usage, latency_options); + } + setup_sorting(); + __cmd_lat(); + } else if (!strcmp(argv[0], "map")) { + trace_handler = &map_ops; + setup_sorting(); + __cmd_map(); + } else if (!strncmp(argv[0], "rep", 3)) { + trace_handler = &replay_ops; + if (argc) { + argc = parse_options(argc, argv, replay_options, replay_usage, 0); + if (argc) + usage_with_options(replay_usage, replay_options); + } + __cmd_replay(); + } else if (!strcmp(argv[0], "trace")) { + /* + * Aliased to 'perf trace' for now: + */ + return cmd_trace(argc, argv, prefix); + } else { + usage_with_options(sched_usage, sched_options); + } + + return 0; +} diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index 2e03524a1de..16af2d82e85 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -42,11 +42,13 @@ #include "util/util.h" #include "util/parse-options.h" #include "util/parse-events.h" +#include "util/event.h" +#include "util/debug.h" #include <sys/prctl.h> #include <math.h> -static struct perf_counter_attr default_attrs[] = { +static struct perf_event_attr default_attrs[] = { { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_TASK_CLOCK }, { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CONTEXT_SWITCHES}, @@ -60,11 +62,8 @@ static struct perf_counter_attr default_attrs[] = { }; -#define MAX_RUN 100 - static int system_wide = 0; -static int verbose = 0; -static int nr_cpus = 0; +static unsigned int nr_cpus = 0; static int run_idx = 0; static int run_count = 1; @@ -75,42 +74,77 @@ static int null_run = 0; static int fd[MAX_NR_CPUS][MAX_COUNTERS]; -static u64 runtime_nsecs[MAX_RUN]; -static u64 walltime_nsecs[MAX_RUN]; -static u64 runtime_cycles[MAX_RUN]; +static int event_scaled[MAX_COUNTERS]; -static u64 event_res[MAX_RUN][MAX_COUNTERS][3]; -static u64 event_scaled[MAX_RUN][MAX_COUNTERS]; +struct stats +{ + double n, mean, M2; +}; -static u64 event_res_avg[MAX_COUNTERS][3]; -static u64 event_res_noise[MAX_COUNTERS][3]; +static void update_stats(struct stats *stats, u64 val) +{ + double delta; -static u64 event_scaled_avg[MAX_COUNTERS]; + stats->n++; + delta = val - stats->mean; + stats->mean += delta / stats->n; + stats->M2 += delta*(val - stats->mean); +} -static u64 runtime_nsecs_avg; -static u64 runtime_nsecs_noise; +static double avg_stats(struct stats *stats) +{ + return stats->mean; +} -static u64 walltime_nsecs_avg; -static u64 walltime_nsecs_noise; +/* + * http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance + * + * (\Sum n_i^2) - ((\Sum n_i)^2)/n + * s^2 = ------------------------------- + * n - 1 + * + * http://en.wikipedia.org/wiki/Stddev + * + * The std dev of the mean is related to the std dev by: + * + * s + * s_mean = ------- + * sqrt(n) + * + */ +static double stddev_stats(struct stats *stats) +{ + double variance = stats->M2 / (stats->n - 1); + double variance_mean = variance / stats->n; -static u64 runtime_cycles_avg; -static u64 runtime_cycles_noise; + return sqrt(variance_mean); +} + +struct stats event_res_stats[MAX_COUNTERS][3]; +struct stats runtime_nsecs_stats; +struct stats walltime_nsecs_stats; +struct stats runtime_cycles_stats; + +#define MATCH_EVENT(t, c, counter) \ + (attrs[counter].type == PERF_TYPE_##t && \ + attrs[counter].config == PERF_COUNT_##c) #define ERR_PERF_OPEN \ -"Error: counter %d, sys_perf_counter_open() syscall returned with %d (%s)\n" +"Error: counter %d, sys_perf_event_open() syscall returned with %d (%s)\n" static void create_perf_stat_counter(int counter, int pid) { - struct perf_counter_attr *attr = attrs + counter; + struct perf_event_attr *attr = attrs + counter; if (scale) attr->read_format = PERF_FORMAT_TOTAL_TIME_ENABLED | PERF_FORMAT_TOTAL_TIME_RUNNING; if (system_wide) { - int cpu; + unsigned int cpu; + for (cpu = 0; cpu < nr_cpus; cpu++) { - fd[cpu][counter] = sys_perf_counter_open(attr, -1, cpu, -1, 0); + fd[cpu][counter] = sys_perf_event_open(attr, -1, cpu, -1, 0); if (fd[cpu][counter] < 0 && verbose) fprintf(stderr, ERR_PERF_OPEN, counter, fd[cpu][counter], strerror(errno)); @@ -120,7 +154,7 @@ static void create_perf_stat_counter(int counter, int pid) attr->disabled = 1; attr->enable_on_exec = 1; - fd[0][counter] = sys_perf_counter_open(attr, pid, -1, -1, 0); + fd[0][counter] = sys_perf_event_open(attr, pid, -1, -1, 0); if (fd[0][counter] < 0 && verbose) fprintf(stderr, ERR_PERF_OPEN, counter, fd[0][counter], strerror(errno)); @@ -132,13 +166,8 @@ static void create_perf_stat_counter(int counter, int pid) */ static inline int nsec_counter(int counter) { - if (attrs[counter].type != PERF_TYPE_SOFTWARE) - return 0; - - if (attrs[counter].config == PERF_COUNT_SW_CPU_CLOCK) - return 1; - - if (attrs[counter].config == PERF_COUNT_SW_TASK_CLOCK) + if (MATCH_EVENT(SOFTWARE, SW_CPU_CLOCK, counter) || + MATCH_EVENT(SOFTWARE, SW_TASK_CLOCK, counter)) return 1; return 0; @@ -149,12 +178,11 @@ static inline int nsec_counter(int counter) */ static void read_counter(int counter) { - u64 *count, single_count[3]; - ssize_t res; - int cpu, nv; + u64 count[3], single_count[3]; + unsigned int cpu; + size_t res, nv; int scaled; - - count = event_res[run_idx][counter]; + int i; count[0] = count[1] = count[2] = 0; @@ -165,6 +193,7 @@ static void read_counter(int counter) res = read(fd[cpu][counter], single_count, nv * sizeof(u64)); assert(res == nv * sizeof(u64)); + close(fd[cpu][counter]); fd[cpu][counter] = -1; @@ -178,29 +207,36 @@ static void read_counter(int counter) scaled = 0; if (scale) { if (count[2] == 0) { - event_scaled[run_idx][counter] = -1; + event_scaled[counter] = -1; count[0] = 0; return; } if (count[2] < count[1]) { - event_scaled[run_idx][counter] = 1; + event_scaled[counter] = 1; count[0] = (unsigned long long) ((double)count[0] * count[1] / count[2] + 0.5); } } + + for (i = 0; i < 3; i++) + update_stats(&event_res_stats[counter][i], count[i]); + + if (verbose) { + fprintf(stderr, "%s: %Ld %Ld %Ld\n", event_name(counter), + count[0], count[1], count[2]); + } + /* * Save the full runtime - to allow normalization during printout: */ - if (attrs[counter].type == PERF_TYPE_SOFTWARE && - attrs[counter].config == PERF_COUNT_SW_TASK_CLOCK) - runtime_nsecs[run_idx] = count[0]; - if (attrs[counter].type == PERF_TYPE_HARDWARE && - attrs[counter].config == PERF_COUNT_HW_CPU_CYCLES) - runtime_cycles[run_idx] = count[0]; + if (MATCH_EVENT(SOFTWARE, SW_TASK_CLOCK, counter)) + update_stats(&runtime_nsecs_stats, count[0]); + if (MATCH_EVENT(HARDWARE, HW_CPU_CYCLES, counter)) + update_stats(&runtime_cycles_stats, count[0]); } -static int run_perf_stat(int argc, const char **argv) +static int run_perf_stat(int argc __used, const char **argv) { unsigned long long t0, t1; int status = 0; @@ -240,7 +276,8 @@ static int run_perf_stat(int argc, const char **argv) /* * Wait until the parent tells us to go. */ - read(go_pipe[0], &buf, 1); + if (read(go_pipe[0], &buf, 1) == -1) + perror("unable to read pipe"); execvp(argv[0], (char **)argv); @@ -253,7 +290,8 @@ static int run_perf_stat(int argc, const char **argv) */ close(child_ready_pipe[1]); close(go_pipe[0]); - read(child_ready_pipe[0], &buf, 1); + if (read(child_ready_pipe[0], &buf, 1) == -1) + perror("unable to read pipe"); close(child_ready_pipe[0]); for (counter = 0; counter < nr_counters; counter++) @@ -269,7 +307,7 @@ static int run_perf_stat(int argc, const char **argv) t1 = rdclock(); - walltime_nsecs[run_idx] = t1 - t0; + update_stats(&walltime_nsecs_stats, t1 - t0); for (counter = 0; counter < nr_counters; counter++) read_counter(counter); @@ -277,46 +315,38 @@ static int run_perf_stat(int argc, const char **argv) return WEXITSTATUS(status); } -static void print_noise(u64 *count, u64 *noise) +static void print_noise(int counter, double avg) { - if (run_count > 1) - fprintf(stderr, " ( +- %7.3f%% )", - (double)noise[0]/(count[0]+1)*100.0); + if (run_count == 1) + return; + + fprintf(stderr, " ( +- %7.3f%% )", + 100 * stddev_stats(&event_res_stats[counter][0]) / avg); } -static void nsec_printout(int counter, u64 *count, u64 *noise) +static void nsec_printout(int counter, double avg) { - double msecs = (double)count[0] / 1000000; + double msecs = avg / 1e6; fprintf(stderr, " %14.6f %-24s", msecs, event_name(counter)); - if (attrs[counter].type == PERF_TYPE_SOFTWARE && - attrs[counter].config == PERF_COUNT_SW_TASK_CLOCK) { - - if (walltime_nsecs_avg) - fprintf(stderr, " # %10.3f CPUs ", - (double)count[0] / (double)walltime_nsecs_avg); + if (MATCH_EVENT(SOFTWARE, SW_TASK_CLOCK, counter)) { + fprintf(stderr, " # %10.3f CPUs ", + avg / avg_stats(&walltime_nsecs_stats)); } - print_noise(count, noise); } -static void abs_printout(int counter, u64 *count, u64 *noise) +static void abs_printout(int counter, double avg) { - fprintf(stderr, " %14Ld %-24s", count[0], event_name(counter)); - - if (runtime_cycles_avg && - attrs[counter].type == PERF_TYPE_HARDWARE && - attrs[counter].config == PERF_COUNT_HW_INSTRUCTIONS) { + fprintf(stderr, " %14.0f %-24s", avg, event_name(counter)); + if (MATCH_EVENT(HARDWARE, HW_INSTRUCTIONS, counter)) { fprintf(stderr, " # %10.3f IPC ", - (double)count[0] / (double)runtime_cycles_avg); + avg / avg_stats(&runtime_cycles_stats)); } else { - if (runtime_nsecs_avg) { - fprintf(stderr, " # %10.3f M/sec", - (double)count[0]/runtime_nsecs_avg*1000.0); - } + fprintf(stderr, " # %10.3f M/sec", + 1000.0 * avg / avg_stats(&runtime_nsecs_stats)); } - print_noise(count, noise); } /* @@ -324,12 +354,8 @@ static void abs_printout(int counter, u64 *count, u64 *noise) */ static void print_counter(int counter) { - u64 *count, *noise; - int scaled; - - count = event_res_avg[counter]; - noise = event_res_noise[counter]; - scaled = event_scaled_avg[counter]; + double avg = avg_stats(&event_res_stats[counter][0]); + int scaled = event_scaled[counter]; if (scaled == -1) { fprintf(stderr, " %14s %-24s\n", @@ -338,110 +364,29 @@ static void print_counter(int counter) } if (nsec_counter(counter)) - nsec_printout(counter, count, noise); + nsec_printout(counter, avg); else - abs_printout(counter, count, noise); - - if (scaled) - fprintf(stderr, " (scaled from %.2f%%)", - (double) count[2] / count[1] * 100); + abs_printout(counter, avg); - fprintf(stderr, "\n"); -} - -/* - * normalize_noise noise values down to stddev: - */ -static void normalize_noise(u64 *val) -{ - double res; + print_noise(counter, avg); - res = (double)*val / (run_count * sqrt((double)run_count)); + if (scaled) { + double avg_enabled, avg_running; - *val = (u64)res; -} + avg_enabled = avg_stats(&event_res_stats[counter][1]); + avg_running = avg_stats(&event_res_stats[counter][2]); -static void update_avg(const char *name, int idx, u64 *avg, u64 *val) -{ - *avg += *val; - - if (verbose > 1) - fprintf(stderr, "debug: %20s[%d]: %Ld\n", name, idx, *val); -} -/* - * Calculate the averages and noises: - */ -static void calc_avg(void) -{ - int i, j; - - if (verbose > 1) - fprintf(stderr, "\n"); - - for (i = 0; i < run_count; i++) { - update_avg("runtime", 0, &runtime_nsecs_avg, runtime_nsecs + i); - update_avg("walltime", 0, &walltime_nsecs_avg, walltime_nsecs + i); - update_avg("runtime_cycles", 0, &runtime_cycles_avg, runtime_cycles + i); - - for (j = 0; j < nr_counters; j++) { - update_avg("counter/0", j, - event_res_avg[j]+0, event_res[i][j]+0); - update_avg("counter/1", j, - event_res_avg[j]+1, event_res[i][j]+1); - update_avg("counter/2", j, - event_res_avg[j]+2, event_res[i][j]+2); - if (event_scaled[i][j] != -1) - update_avg("scaled", j, - event_scaled_avg + j, event_scaled[i]+j); - else - event_scaled_avg[j] = -1; - } - } - runtime_nsecs_avg /= run_count; - walltime_nsecs_avg /= run_count; - runtime_cycles_avg /= run_count; - - for (j = 0; j < nr_counters; j++) { - event_res_avg[j][0] /= run_count; - event_res_avg[j][1] /= run_count; - event_res_avg[j][2] /= run_count; - } - - for (i = 0; i < run_count; i++) { - runtime_nsecs_noise += - abs((s64)(runtime_nsecs[i] - runtime_nsecs_avg)); - walltime_nsecs_noise += - abs((s64)(walltime_nsecs[i] - walltime_nsecs_avg)); - runtime_cycles_noise += - abs((s64)(runtime_cycles[i] - runtime_cycles_avg)); - - for (j = 0; j < nr_counters; j++) { - event_res_noise[j][0] += - abs((s64)(event_res[i][j][0] - event_res_avg[j][0])); - event_res_noise[j][1] += - abs((s64)(event_res[i][j][1] - event_res_avg[j][1])); - event_res_noise[j][2] += - abs((s64)(event_res[i][j][2] - event_res_avg[j][2])); - } + fprintf(stderr, " (scaled from %.2f%%)", + 100 * avg_running / avg_enabled); } - normalize_noise(&runtime_nsecs_noise); - normalize_noise(&walltime_nsecs_noise); - normalize_noise(&runtime_cycles_noise); - - for (j = 0; j < nr_counters; j++) { - normalize_noise(&event_res_noise[j][0]); - normalize_noise(&event_res_noise[j][1]); - normalize_noise(&event_res_noise[j][2]); - } + fprintf(stderr, "\n"); } static void print_stat(int argc, const char **argv) { int i, counter; - calc_avg(); - fflush(stdout); fprintf(stderr, "\n"); @@ -460,10 +405,11 @@ static void print_stat(int argc, const char **argv) fprintf(stderr, "\n"); fprintf(stderr, " %14.9f seconds time elapsed", - (double)walltime_nsecs_avg/1e9); + avg_stats(&walltime_nsecs_stats)/1e9); if (run_count > 1) { fprintf(stderr, " ( +- %7.3f%% )", - 100.0*(double)walltime_nsecs_noise/(double)walltime_nsecs_avg); + 100*stddev_stats(&walltime_nsecs_stats) / + avg_stats(&walltime_nsecs_stats)); } fprintf(stderr, "\n\n"); } @@ -499,7 +445,7 @@ static const struct option options[] = { "stat events on existing pid"), OPT_BOOLEAN('a', "all-cpus", &system_wide, "system-wide collection from all CPUs"), - OPT_BOOLEAN('S', "scale", &scale, + OPT_BOOLEAN('c', "scale", &scale, "scale/normalize counters"), OPT_BOOLEAN('v', "verbose", &verbose, "be more verbose (show counter open errors, etc)"), @@ -510,14 +456,15 @@ static const struct option options[] = { OPT_END() }; -int cmd_stat(int argc, const char **argv, const char *prefix) +int cmd_stat(int argc, const char **argv, const char *prefix __used) { int status; - argc = parse_options(argc, argv, options, stat_usage, 0); + argc = parse_options(argc, argv, options, stat_usage, + PARSE_OPT_STOP_AT_NON_OPTION); if (!argc) usage_with_options(stat_usage, options); - if (run_count <= 0 || run_count > MAX_RUN) + if (run_count <= 0) usage_with_options(stat_usage, options); /* Set attrs and nr_counters if no event is selected and !null_run */ @@ -528,7 +475,7 @@ int cmd_stat(int argc, const char **argv, const char *prefix) nr_cpus = sysconf(_SC_NPROCESSORS_ONLN); assert(nr_cpus <= MAX_NR_CPUS); - assert(nr_cpus >= 0); + assert((int)nr_cpus >= 0); /* * We dont want to block the signals - that would cause diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c new file mode 100644 index 00000000000..4405681b313 --- /dev/null +++ b/tools/perf/builtin-timechart.c @@ -0,0 +1,1158 @@ +/* + * builtin-timechart.c - make an svg timechart of system activity + * + * (C) Copyright 2009 Intel Corporation + * + * Authors: + * Arjan van de Ven <arjan@linux.intel.com> + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; version 2 + * of the License. + */ + +#include "builtin.h" + +#include "util/util.h" + +#include "util/color.h" +#include <linux/list.h> +#include "util/cache.h" +#include <linux/rbtree.h> +#include "util/symbol.h" +#include "util/string.h" +#include "util/callchain.h" +#include "util/strlist.h" + +#include "perf.h" +#include "util/header.h" +#include "util/parse-options.h" +#include "util/parse-events.h" +#include "util/svghelper.h" + +static char const *input_name = "perf.data"; +static char const *output_name = "output.svg"; + + +static unsigned long page_size; +static unsigned long mmap_window = 32; +static u64 sample_type; + +static unsigned int numcpus; +static u64 min_freq; /* Lowest CPU frequency seen */ +static u64 max_freq; /* Highest CPU frequency seen */ +static u64 turbo_frequency; + +static u64 first_time, last_time; + + +static struct perf_header *header; + +struct per_pid; +struct per_pidcomm; + +struct cpu_sample; +struct power_event; +struct wake_event; + +struct sample_wrapper; + +/* + * Datastructure layout: + * We keep an list of "pid"s, matching the kernels notion of a task struct. + * Each "pid" entry, has a list of "comm"s. + * this is because we want to track different programs different, while + * exec will reuse the original pid (by design). + * Each comm has a list of samples that will be used to draw + * final graph. + */ + +struct per_pid { + struct per_pid *next; + + int pid; + int ppid; + + u64 start_time; + u64 end_time; + u64 total_time; + int display; + + struct per_pidcomm *all; + struct per_pidcomm *current; + + int painted; +}; + + +struct per_pidcomm { + struct per_pidcomm *next; + + u64 start_time; + u64 end_time; + u64 total_time; + + int Y; + int display; + + long state; + u64 state_since; + + char *comm; + + struct cpu_sample *samples; +}; + +struct sample_wrapper { + struct sample_wrapper *next; + + u64 timestamp; + unsigned char data[0]; +}; + +#define TYPE_NONE 0 +#define TYPE_RUNNING 1 +#define TYPE_WAITING 2 +#define TYPE_BLOCKED 3 + +struct cpu_sample { + struct cpu_sample *next; + + u64 start_time; + u64 end_time; + int type; + int cpu; +}; + +static struct per_pid *all_data; + +#define CSTATE 1 +#define PSTATE 2 + +struct power_event { + struct power_event *next; + int type; + int state; + u64 start_time; + u64 end_time; + int cpu; +}; + +struct wake_event { + struct wake_event *next; + int waker; + int wakee; + u64 time; +}; + +static struct power_event *power_events; +static struct wake_event *wake_events; + +struct sample_wrapper *all_samples; + +static struct per_pid *find_create_pid(int pid) +{ + struct per_pid *cursor = all_data; + + while (cursor) { + if (cursor->pid == pid) + return cursor; + cursor = cursor->next; + } + cursor = malloc(sizeof(struct per_pid)); + assert(cursor != NULL); + memset(cursor, 0, sizeof(struct per_pid)); + cursor->pid = pid; + cursor->next = all_data; + all_data = cursor; + return cursor; +} + +static void pid_set_comm(int pid, char *comm) +{ + struct per_pid *p; + struct per_pidcomm *c; + p = find_create_pid(pid); + c = p->all; + while (c) { + if (c->comm && strcmp(c->comm, comm) == 0) { + p->current = c; + return; + } + if (!c->comm) { + c->comm = strdup(comm); + p->current = c; + return; + } + c = c->next; + } + c = malloc(sizeof(struct per_pidcomm)); + assert(c != NULL); + memset(c, 0, sizeof(struct per_pidcomm)); + c->comm = strdup(comm); + p->current = c; + c->next = p->all; + p->all = c; +} + +static void pid_fork(int pid, int ppid, u64 timestamp) +{ + struct per_pid *p, *pp; + p = find_create_pid(pid); + pp = find_create_pid(ppid); + p->ppid = ppid; + if (pp->current && pp->current->comm && !p->current) + pid_set_comm(pid, pp->current->comm); + + p->start_time = timestamp; + if (p->current) { + p->current->start_time = timestamp; + p->current->state_since = timestamp; + } +} + +static void pid_exit(int pid, u64 timestamp) +{ + struct per_pid *p; + p = find_create_pid(pid); + p->end_time = timestamp; + if (p->current) + p->current->end_time = timestamp; +} + +static void +pid_put_sample(int pid, int type, unsigned int cpu, u64 start, u64 end) +{ + struct per_pid *p; + struct per_pidcomm *c; + struct cpu_sample *sample; + + p = find_create_pid(pid); + c = p->current; + if (!c) { + c = malloc(sizeof(struct per_pidcomm)); + assert(c != NULL); + memset(c, 0, sizeof(struct per_pidcomm)); + p->current = c; + c->next = p->all; + p->all = c; + } + + sample = malloc(sizeof(struct cpu_sample)); + assert(sample != NULL); + memset(sample, 0, sizeof(struct cpu_sample)); + sample->start_time = start; + sample->end_time = end; + sample->type = type; + sample->next = c->samples; + sample->cpu = cpu; + c->samples = sample; + + if (sample->type == TYPE_RUNNING && end > start && start > 0) { + c->total_time += (end-start); + p->total_time += (end-start); + } + + if (c->start_time == 0 || c->start_time > start) + c->start_time = start; + if (p->start_time == 0 || p->start_time > start) + p->start_time = start; + + if (cpu > numcpus) + numcpus = cpu; +} + +#define MAX_CPUS 4096 + +static u64 cpus_cstate_start_times[MAX_CPUS]; +static int cpus_cstate_state[MAX_CPUS]; +static u64 cpus_pstate_start_times[MAX_CPUS]; +static u64 cpus_pstate_state[MAX_CPUS]; + +static int +process_comm_event(event_t *event) +{ + pid_set_comm(event->comm.pid, event->comm.comm); + return 0; +} +static int +process_fork_event(event_t *event) +{ + pid_fork(event->fork.pid, event->fork.ppid, event->fork.time); + return 0; +} + +static int +process_exit_event(event_t *event) +{ + pid_exit(event->fork.pid, event->fork.time); + return 0; +} + +struct trace_entry { + u32 size; + unsigned short type; + unsigned char flags; + unsigned char preempt_count; + int pid; + int tgid; +}; + +struct power_entry { + struct trace_entry te; + s64 type; + s64 value; +}; + +#define TASK_COMM_LEN 16 +struct wakeup_entry { + struct trace_entry te; + char comm[TASK_COMM_LEN]; + int pid; + int prio; + int success; +}; + +/* + * trace_flag_type is an enumeration that holds different + * states when a trace occurs. These are: + * IRQS_OFF - interrupts were disabled + * IRQS_NOSUPPORT - arch does not support irqs_disabled_flags + * NEED_RESCED - reschedule is requested + * HARDIRQ - inside an interrupt handler + * SOFTIRQ - inside a softirq handler + */ +enum trace_flag_type { + TRACE_FLAG_IRQS_OFF = 0x01, + TRACE_FLAG_IRQS_NOSUPPORT = 0x02, + TRACE_FLAG_NEED_RESCHED = 0x04, + TRACE_FLAG_HARDIRQ = 0x08, + TRACE_FLAG_SOFTIRQ = 0x10, +}; + + + +struct sched_switch { + struct trace_entry te; + char prev_comm[TASK_COMM_LEN]; + int prev_pid; + int prev_prio; + long prev_state; /* Arjan weeps. */ + char next_comm[TASK_COMM_LEN]; + int next_pid; + int next_prio; +}; + +static void c_state_start(int cpu, u64 timestamp, int state) +{ + cpus_cstate_start_times[cpu] = timestamp; + cpus_cstate_state[cpu] = state; +} + +static void c_state_end(int cpu, u64 timestamp) +{ + struct power_event *pwr; + pwr = malloc(sizeof(struct power_event)); + if (!pwr) + return; + memset(pwr, 0, sizeof(struct power_event)); + + pwr->state = cpus_cstate_state[cpu]; + pwr->start_time = cpus_cstate_start_times[cpu]; + pwr->end_time = timestamp; + pwr->cpu = cpu; + pwr->type = CSTATE; + pwr->next = power_events; + + power_events = pwr; +} + +static void p_state_change(int cpu, u64 timestamp, u64 new_freq) +{ + struct power_event *pwr; + pwr = malloc(sizeof(struct power_event)); + + if (new_freq > 8000000) /* detect invalid data */ + return; + + if (!pwr) + return; + memset(pwr, 0, sizeof(struct power_event)); + + pwr->state = cpus_pstate_state[cpu]; + pwr->start_time = cpus_pstate_start_times[cpu]; + pwr->end_time = timestamp; + pwr->cpu = cpu; + pwr->type = PSTATE; + pwr->next = power_events; + + if (!pwr->start_time) + pwr->start_time = first_time; + + power_events = pwr; + + cpus_pstate_state[cpu] = new_freq; + cpus_pstate_start_times[cpu] = timestamp; + + if ((u64)new_freq > max_freq) + max_freq = new_freq; + + if (new_freq < min_freq || min_freq == 0) + min_freq = new_freq; + + if (new_freq == max_freq - 1000) + turbo_frequency = max_freq; +} + +static void +sched_wakeup(int cpu, u64 timestamp, int pid, struct trace_entry *te) +{ + struct wake_event *we; + struct per_pid *p; + struct wakeup_entry *wake = (void *)te; + + we = malloc(sizeof(struct wake_event)); + if (!we) + return; + + memset(we, 0, sizeof(struct wake_event)); + we->time = timestamp; + we->waker = pid; + + if ((te->flags & TRACE_FLAG_HARDIRQ) || (te->flags & TRACE_FLAG_SOFTIRQ)) + we->waker = -1; + + we->wakee = wake->pid; + we->next = wake_events; + wake_events = we; + p = find_create_pid(we->wakee); + + if (p && p->current && p->current->state == TYPE_NONE) { + p->current->state_since = timestamp; + p->current->state = TYPE_WAITING; + } + if (p && p->current && p->current->state == TYPE_BLOCKED) { + pid_put_sample(p->pid, p->current->state, cpu, p->current->state_since, timestamp); + p->current->state_since = timestamp; + p->current->state = TYPE_WAITING; + } +} + +static void sched_switch(int cpu, u64 timestamp, struct trace_entry *te) +{ + struct per_pid *p = NULL, *prev_p; + struct sched_switch *sw = (void *)te; + + + prev_p = find_create_pid(sw->prev_pid); + + p = find_create_pid(sw->next_pid); + + if (prev_p->current && prev_p->current->state != TYPE_NONE) + pid_put_sample(sw->prev_pid, TYPE_RUNNING, cpu, prev_p->current->state_since, timestamp); + if (p && p->current) { + if (p->current->state != TYPE_NONE) + pid_put_sample(sw->next_pid, p->current->state, cpu, p->current->state_since, timestamp); + + p->current->state_since = timestamp; + p->current->state = TYPE_RUNNING; + } + + if (prev_p->current) { + prev_p->current->state = TYPE_NONE; + prev_p->current->state_since = timestamp; + if (sw->prev_state & 2) + prev_p->current->state = TYPE_BLOCKED; + if (sw->prev_state == 0) + prev_p->current->state = TYPE_WAITING; + } +} + + +static int +process_sample_event(event_t *event) +{ + int cursor = 0; + u64 addr = 0; + u64 stamp = 0; + u32 cpu = 0; + u32 pid = 0; + struct trace_entry *te; + + if (sample_type & PERF_SAMPLE_IP) + cursor++; + + if (sample_type & PERF_SAMPLE_TID) { + pid = event->sample.array[cursor]>>32; + cursor++; + } + if (sample_type & PERF_SAMPLE_TIME) { + stamp = event->sample.array[cursor++]; + + if (!first_time || first_time > stamp) + first_time = stamp; + if (last_time < stamp) + last_time = stamp; + + } + if (sample_type & PERF_SAMPLE_ADDR) + addr = event->sample.array[cursor++]; + if (sample_type & PERF_SAMPLE_ID) + cursor++; + if (sample_type & PERF_SAMPLE_STREAM_ID) + cursor++; + if (sample_type & PERF_SAMPLE_CPU) + cpu = event->sample.array[cursor++] & 0xFFFFFFFF; + if (sample_type & PERF_SAMPLE_PERIOD) + cursor++; + + te = (void *)&event->sample.array[cursor]; + + if (sample_type & PERF_SAMPLE_RAW && te->size > 0) { + char *event_str; + struct power_entry *pe; + + pe = (void *)te; + + event_str = perf_header__find_event(te->type); + + if (!event_str) + return 0; + + if (strcmp(event_str, "power:power_start") == 0) + c_state_start(cpu, stamp, pe->value); + + if (strcmp(event_str, "power:power_end") == 0) + c_state_end(cpu, stamp); + + if (strcmp(event_str, "power:power_frequency") == 0) + p_state_change(cpu, stamp, pe->value); + + if (strcmp(event_str, "sched:sched_wakeup") == 0) + sched_wakeup(cpu, stamp, pid, te); + + if (strcmp(event_str, "sched:sched_switch") == 0) + sched_switch(cpu, stamp, te); + } + return 0; +} + +/* + * After the last sample we need to wrap up the current C/P state + * and close out each CPU for these. + */ +static void end_sample_processing(void) +{ + u64 cpu; + struct power_event *pwr; + + for (cpu = 0; cpu < numcpus; cpu++) { + pwr = malloc(sizeof(struct power_event)); + if (!pwr) + return; + memset(pwr, 0, sizeof(struct power_event)); + + /* C state */ +#if 0 + pwr->state = cpus_cstate_state[cpu]; + pwr->start_time = cpus_cstate_start_times[cpu]; + pwr->end_time = last_time; + pwr->cpu = cpu; + pwr->type = CSTATE; + pwr->next = power_events; + + power_events = pwr; +#endif + /* P state */ + + pwr = malloc(sizeof(struct power_event)); + if (!pwr) + return; + memset(pwr, 0, sizeof(struct power_event)); + + pwr->state = cpus_pstate_state[cpu]; + pwr->start_time = cpus_pstate_start_times[cpu]; + pwr->end_time = last_time; + pwr->cpu = cpu; + pwr->type = PSTATE; + pwr->next = power_events; + + if (!pwr->start_time) + pwr->start_time = first_time; + if (!pwr->state) + pwr->state = min_freq; + power_events = pwr; + } +} + +static u64 sample_time(event_t *event) +{ + int cursor; + + cursor = 0; + if (sample_type & PERF_SAMPLE_IP) + cursor++; + if (sample_type & PERF_SAMPLE_TID) + cursor++; + if (sample_type & PERF_SAMPLE_TIME) + return event->sample.array[cursor]; + return 0; +} + + +/* + * We first queue all events, sorted backwards by insertion. + * The order will get flipped later. + */ +static int +queue_sample_event(event_t *event) +{ + struct sample_wrapper *copy, *prev; + int size; + + size = event->sample.header.size + sizeof(struct sample_wrapper) + 8; + + copy = malloc(size); + if (!copy) + return 1; + + memset(copy, 0, size); + + copy->next = NULL; + copy->timestamp = sample_time(event); + + memcpy(©->data, event, event->sample.header.size); + + /* insert in the right place in the list */ + + if (!all_samples) { + /* first sample ever */ + all_samples = copy; + return 0; + } + + if (all_samples->timestamp < copy->timestamp) { + /* insert at the head of the list */ + copy->next = all_samples; + all_samples = copy; + return 0; + } + + prev = all_samples; + while (prev->next) { + if (prev->next->timestamp < copy->timestamp) { + copy->next = prev->next; + prev->next = copy; + return 0; + } + prev = prev->next; + } + /* insert at the end of the list */ + prev->next = copy; + + return 0; +} + +static void sort_queued_samples(void) +{ + struct sample_wrapper *cursor, *next; + + cursor = all_samples; + all_samples = NULL; + + while (cursor) { + next = cursor->next; + cursor->next = all_samples; + all_samples = cursor; + cursor = next; + } +} + +/* + * Sort the pid datastructure + */ +static void sort_pids(void) +{ + struct per_pid *new_list, *p, *cursor, *prev; + /* sort by ppid first, then by pid, lowest to highest */ + + new_list = NULL; + + while (all_data) { + p = all_data; + all_data = p->next; + p->next = NULL; + + if (new_list == NULL) { + new_list = p; + p->next = NULL; + continue; + } + prev = NULL; + cursor = new_list; + while (cursor) { + if (cursor->ppid > p->ppid || + (cursor->ppid == p->ppid && cursor->pid > p->pid)) { + /* must insert before */ + if (prev) { + p->next = prev->next; + prev->next = p; + cursor = NULL; + continue; + } else { + p->next = new_list; + new_list = p; + cursor = NULL; + continue; + } + } + + prev = cursor; + cursor = cursor->next; + if (!cursor) + prev->next = p; + } + } + all_data = new_list; +} + + +static void draw_c_p_states(void) +{ + struct power_event *pwr; + pwr = power_events; + + /* + * two pass drawing so that the P state bars are on top of the C state blocks + */ + while (pwr) { + if (pwr->type == CSTATE) + svg_cstate(pwr->cpu, pwr->start_time, pwr->end_time, pwr->state); + pwr = pwr->next; + } + + pwr = power_events; + while (pwr) { + if (pwr->type == PSTATE) { + if (!pwr->state) + pwr->state = min_freq; + svg_pstate(pwr->cpu, pwr->start_time, pwr->end_time, pwr->state); + } + pwr = pwr->next; + } +} + +static void draw_wakeups(void) +{ + struct wake_event *we; + struct per_pid *p; + struct per_pidcomm *c; + + we = wake_events; + while (we) { + int from = 0, to = 0; + char *task_from = NULL, *task_to = NULL; + + /* locate the column of the waker and wakee */ + p = all_data; + while (p) { + if (p->pid == we->waker || p->pid == we->wakee) { + c = p->all; + while (c) { + if (c->Y && c->start_time <= we->time && c->end_time >= we->time) { + if (p->pid == we->waker) { + from = c->Y; + task_from = c->comm; + } + if (p->pid == we->wakee) { + to = c->Y; + task_to = c->comm; + } + } + c = c->next; + } + } + p = p->next; + } + + if (we->waker == -1) + svg_interrupt(we->time, to); + else if (from && to && abs(from - to) == 1) + svg_wakeline(we->time, from, to); + else + svg_partial_wakeline(we->time, from, task_from, to, task_to); + we = we->next; + } +} + +static void draw_cpu_usage(void) +{ + struct per_pid *p; + struct per_pidcomm *c; + struct cpu_sample *sample; + p = all_data; + while (p) { + c = p->all; + while (c) { + sample = c->samples; + while (sample) { + if (sample->type == TYPE_RUNNING) + svg_process(sample->cpu, sample->start_time, sample->end_time, "sample", c->comm); + + sample = sample->next; + } + c = c->next; + } + p = p->next; + } +} + +static void draw_process_bars(void) +{ + struct per_pid *p; + struct per_pidcomm *c; + struct cpu_sample *sample; + int Y = 0; + + Y = 2 * numcpus + 2; + + p = all_data; + while (p) { + c = p->all; + while (c) { + if (!c->display) { + c->Y = 0; + c = c->next; + continue; + } + + svg_box(Y, c->start_time, c->end_time, "process"); + sample = c->samples; + while (sample) { + if (sample->type == TYPE_RUNNING) + svg_sample(Y, sample->cpu, sample->start_time, sample->end_time); + if (sample->type == TYPE_BLOCKED) + svg_box(Y, sample->start_time, sample->end_time, "blocked"); + if (sample->type == TYPE_WAITING) + svg_waiting(Y, sample->start_time, sample->end_time); + sample = sample->next; + } + + if (c->comm) { + char comm[256]; + if (c->total_time > 5000000000) /* 5 seconds */ + sprintf(comm, "%s:%i (%2.2fs)", c->comm, p->pid, c->total_time / 1000000000.0); + else + sprintf(comm, "%s:%i (%3.1fms)", c->comm, p->pid, c->total_time / 1000000.0); + + svg_text(Y, c->start_time, comm); + } + c->Y = Y; + Y++; + c = c->next; + } + p = p->next; + } +} + +static int determine_display_tasks(u64 threshold) +{ + struct per_pid *p; + struct per_pidcomm *c; + int count = 0; + + p = all_data; + while (p) { + p->display = 0; + if (p->start_time == 1) + p->start_time = first_time; + + /* no exit marker, task kept running to the end */ + if (p->end_time == 0) + p->end_time = last_time; + if (p->total_time >= threshold) + p->display = 1; + + c = p->all; + + while (c) { + c->display = 0; + + if (c->start_time == 1) + c->start_time = first_time; + + if (c->total_time >= threshold) { + c->display = 1; + count++; + } + + if (c->end_time == 0) + c->end_time = last_time; + + c = c->next; + } + p = p->next; + } + return count; +} + + + +#define TIME_THRESH 10000000 + +static void write_svg_file(const char *filename) +{ + u64 i; + int count; + + numcpus++; + + + count = determine_display_tasks(TIME_THRESH); + + /* We'd like to show at least 15 tasks; be less picky if we have fewer */ + if (count < 15) + count = determine_display_tasks(TIME_THRESH / 10); + + open_svg(filename, numcpus, count, first_time, last_time); + + svg_time_grid(); + svg_legenda(); + + for (i = 0; i < numcpus; i++) + svg_cpu_box(i, max_freq, turbo_frequency); + + draw_cpu_usage(); + draw_process_bars(); + draw_c_p_states(); + draw_wakeups(); + + svg_close(); +} + +static int +process_event(event_t *event) +{ + + switch (event->header.type) { + + case PERF_RECORD_COMM: + return process_comm_event(event); + case PERF_RECORD_FORK: + return process_fork_event(event); + case PERF_RECORD_EXIT: + return process_exit_event(event); + case PERF_RECORD_SAMPLE: + return queue_sample_event(event); + + /* + * We dont process them right now but they are fine: + */ + case PERF_RECORD_MMAP: + case PERF_RECORD_THROTTLE: + case PERF_RECORD_UNTHROTTLE: + return 0; + + default: + return -1; + } + + return 0; +} + +static void process_samples(void) +{ + struct sample_wrapper *cursor; + event_t *event; + + sort_queued_samples(); + + cursor = all_samples; + while (cursor) { + event = (void *)&cursor->data; + cursor = cursor->next; + process_sample_event(event); + } +} + + +static int __cmd_timechart(void) +{ + int ret, rc = EXIT_FAILURE; + unsigned long offset = 0; + unsigned long head, shift; + struct stat statbuf; + event_t *event; + uint32_t size; + char *buf; + int input; + + input = open(input_name, O_RDONLY); + if (input < 0) { + fprintf(stderr, " failed to open file: %s", input_name); + if (!strcmp(input_name, "perf.data")) + fprintf(stderr, " (try 'perf record' first)"); + fprintf(stderr, "\n"); + exit(-1); + } + + ret = fstat(input, &statbuf); + if (ret < 0) { + perror("failed to stat file"); + exit(-1); + } + + if (!statbuf.st_size) { + fprintf(stderr, "zero-sized file, nothing to do!\n"); + exit(0); + } + + header = perf_header__read(input); + head = header->data_offset; + + sample_type = perf_header__sample_type(header); + + shift = page_size * (head / page_size); + offset += shift; + head -= shift; + +remap: + buf = (char *)mmap(NULL, page_size * mmap_window, PROT_READ, + MAP_SHARED, input, offset); + if (buf == MAP_FAILED) { + perror("failed to mmap file"); + exit(-1); + } + +more: + event = (event_t *)(buf + head); + + size = event->header.size; + if (!size) + size = 8; + + if (head + event->header.size >= page_size * mmap_window) { + int ret2; + + shift = page_size * (head / page_size); + + ret2 = munmap(buf, page_size * mmap_window); + assert(ret2 == 0); + + offset += shift; + head -= shift; + goto remap; + } + + size = event->header.size; + + if (!size || process_event(event) < 0) { + + printf("%p [%p]: skipping unknown header type: %d\n", + (void *)(offset + head), + (void *)(long)(event->header.size), + event->header.type); + + /* + * assume we lost track of the stream, check alignment, and + * increment a single u64 in the hope to catch on again 'soon'. + */ + + if (unlikely(head & 7)) + head &= ~7ULL; + + size = 8; + } + + head += size; + + if (offset + head >= header->data_offset + header->data_size) + goto done; + + if (offset + head < (unsigned long)statbuf.st_size) + goto more; + +done: + rc = EXIT_SUCCESS; + close(input); + + + process_samples(); + + end_sample_processing(); + + sort_pids(); + + write_svg_file(output_name); + + printf("Written %2.1f seconds of trace to %s.\n", (last_time - first_time) / 1000000000.0, output_name); + + return rc; +} + +static const char * const timechart_usage[] = { + "perf timechart [<options>] {record}", + NULL +}; + +static const char *record_args[] = { + "record", + "-a", + "-R", + "-M", + "-f", + "-c", "1", + "-e", "power:power_start", + "-e", "power:power_end", + "-e", "power:power_frequency", + "-e", "sched:sched_wakeup", + "-e", "sched:sched_switch", +}; + +static int __cmd_record(int argc, const char **argv) +{ + unsigned int rec_argc, i, j; + const char **rec_argv; + + rec_argc = ARRAY_SIZE(record_args) + argc - 1; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + + for (i = 0; i < ARRAY_SIZE(record_args); i++) + rec_argv[i] = strdup(record_args[i]); + + for (j = 1; j < (unsigned int)argc; j++, i++) + rec_argv[i] = argv[j]; + + return cmd_record(i, rec_argv, NULL); +} + +static const struct option options[] = { + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_STRING('o', "output", &output_name, "file", + "output file name"), + OPT_INTEGER('w', "width", &svg_page_width, + "page width"), + OPT_END() +}; + + +int cmd_timechart(int argc, const char **argv, const char *prefix __used) +{ + symbol__init(); + + page_size = getpagesize(); + + argc = parse_options(argc, argv, options, timechart_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + + if (argc && !strncmp(argv[0], "rec", 3)) + return __cmd_record(argc, argv); + else if (argc) + usage_with_options(timechart_usage, options); + + setup_pager(); + + return __cmd_timechart(); +} diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index cf0d21f1ae1..1ca88896eee 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -23,14 +23,18 @@ #include "util/symbol.h" #include "util/color.h" #include "util/util.h" -#include "util/rbtree.h" +#include <linux/rbtree.h> #include "util/parse-options.h" #include "util/parse-events.h" +#include "util/debug.h" + #include <assert.h> #include <fcntl.h> #include <stdio.h> +#include <termios.h> +#include <unistd.h> #include <errno.h> #include <time.h> @@ -54,10 +58,11 @@ static int system_wide = 0; static int default_interval = 100000; -static u64 count_filter = 5; +static int count_filter = 5; static int print_entries = 15; static int target_pid = -1; +static int inherit = 0; static int profile_cpu = -1; static int nr_cpus = 0; static unsigned int realtime_prio = 0; @@ -65,17 +70,29 @@ static int group = 0; static unsigned int page_size; static unsigned int mmap_pages = 16; static int freq = 0; -static int verbose = 0; - -static char *sym_filter; -static unsigned long filter_start; -static unsigned long filter_end; static int delay_secs = 2; static int zero; static int dump_symtab; /* + * Source + */ + +struct source_line { + u64 eip; + unsigned long count[MAX_COUNTERS]; + char *line; + struct source_line *next; +}; + +static char *sym_filter = NULL; +struct sym_entry *sym_filter_entry = NULL; +static int sym_pcnt_filter = 5; +static int sym_counter = 0; +static int display_weighted = -1; + +/* * Symbols */ @@ -89,11 +106,238 @@ struct sym_entry { unsigned long snap_count; double weight; int skip; + struct source_line *source; + struct source_line *lines; + struct source_line **lines_tail; + pthread_mutex_t source_lock; }; -struct sym_entry *sym_filter_entry; +/* + * Source functions + */ + +static void parse_source(struct sym_entry *syme) +{ + struct symbol *sym; + struct module *module; + struct section *section = NULL; + FILE *file; + char command[PATH_MAX*2]; + const char *path = vmlinux_name; + u64 start, end, len; + + if (!syme) + return; + + if (syme->lines) { + pthread_mutex_lock(&syme->source_lock); + goto out_assign; + } + + sym = (struct symbol *)(syme + 1); + module = sym->module; + + if (module) + path = module->path; + if (!path) + return; + + start = sym->obj_start; + if (!start) + start = sym->start; + + if (module) { + section = module->sections->find_section(module->sections, ".text"); + if (section) + start -= section->vma; + } + + end = start + sym->end - sym->start + 1; + len = sym->end - sym->start; + + sprintf(command, "objdump --start-address=0x%016Lx --stop-address=0x%016Lx -dS %s", start, end, path); + + file = popen(command, "r"); + if (!file) + return; + + pthread_mutex_lock(&syme->source_lock); + syme->lines_tail = &syme->lines; + while (!feof(file)) { + struct source_line *src; + size_t dummy = 0; + char *c; + + src = malloc(sizeof(struct source_line)); + assert(src != NULL); + memset(src, 0, sizeof(struct source_line)); + + if (getline(&src->line, &dummy, file) < 0) + break; + if (!src->line) + break; + + c = strchr(src->line, '\n'); + if (c) + *c = 0; + + src->next = NULL; + *syme->lines_tail = src; + syme->lines_tail = &src->next; + + if (strlen(src->line)>8 && src->line[8] == ':') { + src->eip = strtoull(src->line, NULL, 16); + if (section) + src->eip += section->vma; + } + if (strlen(src->line)>8 && src->line[16] == ':') { + src->eip = strtoull(src->line, NULL, 16); + if (section) + src->eip += section->vma; + } + } + pclose(file); +out_assign: + sym_filter_entry = syme; + pthread_mutex_unlock(&syme->source_lock); +} + +static void __zero_source_counters(struct sym_entry *syme) +{ + int i; + struct source_line *line; + + line = syme->lines; + while (line) { + for (i = 0; i < nr_counters; i++) + line->count[i] = 0; + line = line->next; + } +} + +static void record_precise_ip(struct sym_entry *syme, int counter, u64 ip) +{ + struct source_line *line; + + if (syme != sym_filter_entry) + return; + + if (pthread_mutex_trylock(&syme->source_lock)) + return; -struct dso *kernel_dso; + if (!syme->source) + goto out_unlock; + + for (line = syme->lines; line; line = line->next) { + if (line->eip == ip) { + line->count[counter]++; + break; + } + if (line->eip > ip) + break; + } +out_unlock: + pthread_mutex_unlock(&syme->source_lock); +} + +static void lookup_sym_source(struct sym_entry *syme) +{ + struct symbol *symbol = (struct symbol *)(syme + 1); + struct source_line *line; + char pattern[PATH_MAX]; + char *idx; + + sprintf(pattern, "<%s>:", symbol->name); + + if (symbol->module) { + idx = strstr(pattern, "\t"); + if (idx) + *idx = 0; + } + + pthread_mutex_lock(&syme->source_lock); + for (line = syme->lines; line; line = line->next) { + if (strstr(line->line, pattern)) { + syme->source = line; + break; + } + } + pthread_mutex_unlock(&syme->source_lock); +} + +static void show_lines(struct source_line *queue, int count, int total) +{ + int i; + struct source_line *line; + + line = queue; + for (i = 0; i < count; i++) { + float pcnt = 100.0*(float)line->count[sym_counter]/(float)total; + + printf("%8li %4.1f%%\t%s\n", line->count[sym_counter], pcnt, line->line); + line = line->next; + } +} + +#define TRACE_COUNT 3 + +static void show_details(struct sym_entry *syme) +{ + struct symbol *symbol; + struct source_line *line; + struct source_line *line_queue = NULL; + int displayed = 0; + int line_queue_count = 0, total = 0, more = 0; + + if (!syme) + return; + + if (!syme->source) + lookup_sym_source(syme); + + if (!syme->source) + return; + + symbol = (struct symbol *)(syme + 1); + printf("Showing %s for %s\n", event_name(sym_counter), symbol->name); + printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter); + + pthread_mutex_lock(&syme->source_lock); + line = syme->source; + while (line) { + total += line->count[sym_counter]; + line = line->next; + } + + line = syme->source; + while (line) { + float pcnt = 0.0; + + if (!line_queue_count) + line_queue = line; + line_queue_count++; + + if (line->count[sym_counter]) + pcnt = 100.0 * line->count[sym_counter] / (float)total; + if (pcnt >= (float)sym_pcnt_filter) { + if (displayed <= print_entries) + show_lines(line_queue, line_queue_count, total); + else more++; + displayed += line_queue_count; + line_queue_count = 0; + line_queue = NULL; + } else if (line_queue_count > TRACE_COUNT) { + line_queue = line_queue->next; + line_queue_count--; + } + + line->count[sym_counter] = zero ? 0 : line->count[sym_counter] * 7 / 8; + line = line->next; + } + pthread_mutex_unlock(&syme->source_lock); + if (more) + printf("%d lines not displayed, maybe increase display entries [e]\n", more); +} /* * Symbols will be added here in record_ip and will get out @@ -110,6 +354,9 @@ static double sym_weight(const struct sym_entry *sym) double weight = sym->snap_count; int counter; + if (!display_weighted) + return weight; + for (counter = 1; counter < nr_counters-1; counter++) weight *= sym->count[counter]; @@ -157,7 +404,7 @@ static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se) static void print_sym_table(void) { int printed = 0, j; - int counter; + int counter, snap = !display_weighted ? sym_counter : 0; float samples_per_sec = samples/delay_secs; float ksamples_per_sec = (samples-userspace_samples)/delay_secs; float sum_ksamples = 0.0; @@ -173,7 +420,7 @@ static void print_sym_table(void) pthread_mutex_unlock(&active_symbols_lock); list_for_each_entry_safe_from(syme, n, &active_symbols, node) { - syme->snap_count = syme->count[0]; + syme->snap_count = syme->count[snap]; if (syme->snap_count != 0) { syme->weight = sym_weight(syme); rb_insert_active_sym(&tmp, syme); @@ -193,7 +440,7 @@ static void print_sym_table(void) samples_per_sec, 100.0 - (100.0*((samples_per_sec-ksamples_per_sec)/samples_per_sec))); - if (nr_counters == 1) { + if (nr_counters == 1 || !display_weighted) { printf("%Ld", (u64)attrs[0].sample_period); if (freq) printf("Hz "); @@ -201,7 +448,9 @@ static void print_sym_table(void) printf(" "); } - for (counter = 0; counter < nr_counters; counter++) { + if (!display_weighted) + printf("%s", event_name(sym_counter)); + else for (counter = 0; counter < nr_counters; counter++) { if (counter) printf("/"); @@ -226,71 +475,330 @@ static void print_sym_table(void) printf("------------------------------------------------------------------------------\n\n"); + if (sym_filter_entry) { + show_details(sym_filter_entry); + return; + } + if (nr_counters == 1) printf(" samples pcnt"); else - printf(" weight samples pcnt"); + printf(" weight samples pcnt"); - printf(" RIP kernel function\n" - " ______ _______ _____ ________________ _______________\n\n" - ); + if (verbose) + printf(" RIP "); + printf(" kernel function\n"); + printf(" %s _______ _____", + nr_counters == 1 ? " " : "______"); + if (verbose) + printf(" ________________"); + printf(" _______________\n\n"); for (nd = rb_first(&tmp); nd; nd = rb_next(nd)) { - struct sym_entry *syme = rb_entry(nd, struct sym_entry, rb_node); - struct symbol *sym = (struct symbol *)(syme + 1); - char *color = PERF_COLOR_NORMAL; + struct symbol *sym; double pcnt; - if (++printed > print_entries || syme->snap_count < count_filter) + syme = rb_entry(nd, struct sym_entry, rb_node); + sym = (struct symbol *)(syme + 1); + + if (++printed > print_entries || (int)syme->snap_count < count_filter) continue; pcnt = 100.0 - (100.0 * ((sum_ksamples - syme->snap_count) / sum_ksamples)); - /* - * We color high-overhead entries in red, mid-overhead - * entries in green - and keep the low overhead places - * normal: - */ - if (pcnt >= 5.0) { - color = PERF_COLOR_RED; - } else { - if (pcnt >= 0.5) - color = PERF_COLOR_GREEN; - } - - if (nr_counters == 1) + if (nr_counters == 1 || !display_weighted) printf("%20.2f - ", syme->weight); else printf("%9.1f %10ld - ", syme->weight, syme->snap_count); - color_fprintf(stdout, color, "%4.1f%%", pcnt); - printf(" - %016llx : %s\n", sym->start, sym->name); + percent_color_fprintf(stdout, "%4.1f%%", pcnt); + if (verbose) + printf(" - %016llx", sym->start); + printf(" : %s", sym->name); + if (sym->module) + printf("\t[%s]", sym->module->name); + printf("\n"); } } -static void *display_thread(void *arg) +static void prompt_integer(int *target, const char *msg) +{ + char *buf = malloc(0), *p; + size_t dummy = 0; + int tmp; + + fprintf(stdout, "\n%s: ", msg); + if (getline(&buf, &dummy, stdin) < 0) + return; + + p = strchr(buf, '\n'); + if (p) + *p = 0; + + p = buf; + while(*p) { + if (!isdigit(*p)) + goto out_free; + p++; + } + tmp = strtoul(buf, NULL, 10); + *target = tmp; +out_free: + free(buf); +} + +static void prompt_percent(int *target, const char *msg) +{ + int tmp = 0; + + prompt_integer(&tmp, msg); + if (tmp >= 0 && tmp <= 100) + *target = tmp; +} + +static void prompt_symbol(struct sym_entry **target, const char *msg) +{ + char *buf = malloc(0), *p; + struct sym_entry *syme = *target, *n, *found = NULL; + size_t dummy = 0; + + /* zero counters of active symbol */ + if (syme) { + pthread_mutex_lock(&syme->source_lock); + __zero_source_counters(syme); + *target = NULL; + pthread_mutex_unlock(&syme->source_lock); + } + + fprintf(stdout, "\n%s: ", msg); + if (getline(&buf, &dummy, stdin) < 0) + goto out_free; + + p = strchr(buf, '\n'); + if (p) + *p = 0; + + pthread_mutex_lock(&active_symbols_lock); + syme = list_entry(active_symbols.next, struct sym_entry, node); + pthread_mutex_unlock(&active_symbols_lock); + + list_for_each_entry_safe_from(syme, n, &active_symbols, node) { + struct symbol *sym = (struct symbol *)(syme + 1); + + if (!strcmp(buf, sym->name)) { + found = syme; + break; + } + } + + if (!found) { + fprintf(stderr, "Sorry, %s is not active.\n", sym_filter); + sleep(1); + return; + } else + parse_source(found); + +out_free: + free(buf); +} + +static void print_mapped_keys(void) +{ + char *name = NULL; + + if (sym_filter_entry) { + struct symbol *sym = (struct symbol *)(sym_filter_entry+1); + name = sym->name; + } + + fprintf(stdout, "\nMapped keys:\n"); + fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", delay_secs); + fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", print_entries); + + if (nr_counters > 1) + fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(sym_counter)); + + fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", count_filter); + + if (vmlinux_name) { + fprintf(stdout, "\t[F] annotate display filter (percent). \t(%d%%)\n", sym_pcnt_filter); + fprintf(stdout, "\t[s] annotate symbol. \t(%s)\n", name?: "NULL"); + fprintf(stdout, "\t[S] stop annotation.\n"); + } + + if (nr_counters > 1) + fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", display_weighted ? 1 : 0); + + fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", zero ? 1 : 0); + fprintf(stdout, "\t[qQ] quit.\n"); +} + +static int key_mapped(int c) +{ + switch (c) { + case 'd': + case 'e': + case 'f': + case 'z': + case 'q': + case 'Q': + return 1; + case 'E': + case 'w': + return nr_counters > 1 ? 1 : 0; + case 'F': + case 's': + case 'S': + return vmlinux_name ? 1 : 0; + default: + break; + } + + return 0; +} + +static void handle_keypress(int c) +{ + if (!key_mapped(c)) { + struct pollfd stdin_poll = { .fd = 0, .events = POLLIN }; + struct termios tc, save; + + print_mapped_keys(); + fprintf(stdout, "\nEnter selection, or unmapped key to continue: "); + fflush(stdout); + + tcgetattr(0, &save); + tc = save; + tc.c_lflag &= ~(ICANON | ECHO); + tc.c_cc[VMIN] = 0; + tc.c_cc[VTIME] = 0; + tcsetattr(0, TCSANOW, &tc); + + poll(&stdin_poll, 1, -1); + c = getc(stdin); + + tcsetattr(0, TCSAFLUSH, &save); + if (!key_mapped(c)) + return; + } + + switch (c) { + case 'd': + prompt_integer(&delay_secs, "Enter display delay"); + break; + case 'e': + prompt_integer(&print_entries, "Enter display entries (lines)"); + break; + case 'E': + if (nr_counters > 1) { + int i; + + fprintf(stderr, "\nAvailable events:"); + for (i = 0; i < nr_counters; i++) + fprintf(stderr, "\n\t%d %s", i, event_name(i)); + + prompt_integer(&sym_counter, "Enter details event counter"); + + if (sym_counter >= nr_counters) { + fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(0)); + sym_counter = 0; + sleep(1); + } + } else sym_counter = 0; + break; + case 'f': + prompt_integer(&count_filter, "Enter display event count filter"); + break; + case 'F': + prompt_percent(&sym_pcnt_filter, "Enter details display event filter (percent)"); + break; + case 'q': + case 'Q': + printf("exiting.\n"); + exit(0); + case 's': + prompt_symbol(&sym_filter_entry, "Enter details symbol"); + break; + case 'S': + if (!sym_filter_entry) + break; + else { + struct sym_entry *syme = sym_filter_entry; + + pthread_mutex_lock(&syme->source_lock); + sym_filter_entry = NULL; + __zero_source_counters(syme); + pthread_mutex_unlock(&syme->source_lock); + } + break; + case 'w': + display_weighted = ~display_weighted; + break; + case 'z': + zero = ~zero; + break; + default: + break; + } +} + +static void *display_thread(void *arg __used) { struct pollfd stdin_poll = { .fd = 0, .events = POLLIN }; - int delay_msecs = delay_secs * 1000; + struct termios tc, save; + int delay_msecs, c; - printf("PerfTop refresh period: %d seconds\n", delay_secs); + tcgetattr(0, &save); + tc = save; + tc.c_lflag &= ~(ICANON | ECHO); + tc.c_cc[VMIN] = 0; + tc.c_cc[VTIME] = 0; + +repeat: + delay_msecs = delay_secs * 1000; + tcsetattr(0, TCSANOW, &tc); + /* trash return*/ + getc(stdin); do { print_sym_table(); } while (!poll(&stdin_poll, 1, delay_msecs) == 1); - printf("key pressed - exiting.\n"); - exit(0); + c = getc(stdin); + tcsetattr(0, TCSAFLUSH, &save); + + handle_keypress(c); + goto repeat; return NULL; } +/* Tag samples to be skipped. */ +static const char *skip_symbols[] = { + "default_idle", + "cpu_idle", + "enter_idle", + "exit_idle", + "mwait_idle", + "mwait_idle_with_hints", + "ppc64_runlatch_off", + "pseries_dedicated_idle_sleep", + NULL +}; + static int symbol_filter(struct dso *self, struct symbol *sym) { - static int filter_match; struct sym_entry *syme; const char *name = sym->name; + int i; + + /* + * ppc64 uses function descriptors and appends a '.' to the + * start of every instruction address. Remove it. + */ + if (name[0] == '.') + name++; if (!strcmp(name, "_text") || !strcmp(name, "_etext") || @@ -302,37 +810,17 @@ static int symbol_filter(struct dso *self, struct symbol *sym) return 1; syme = dso__sym_priv(self, sym); - /* Tag samples to be skipped. */ - if (!strcmp("default_idle", name) || - !strcmp("cpu_idle", name) || - !strcmp("enter_idle", name) || - !strcmp("exit_idle", name) || - !strcmp("mwait_idle", name)) - syme->skip = 1; - - if (filter_match == 1) { - filter_end = sym->start; - filter_match = -1; - if (filter_end - filter_start > 10000) { - fprintf(stderr, - "hm, too large filter symbol <%s> - skipping.\n", - sym_filter); - fprintf(stderr, "symbol filter start: %016lx\n", - filter_start); - fprintf(stderr, " end: %016lx\n", - filter_end); - filter_end = filter_start = 0; - sym_filter = NULL; - sleep(1); + pthread_mutex_init(&syme->source_lock, NULL); + if (!sym_filter_entry && sym_filter && !strcmp(name, sym_filter)) + sym_filter_entry = syme; + + for (i = 0; skip_symbols[i]; i++) { + if (!strcmp(skip_symbols[i], name)) { + syme->skip = 1; + break; } } - if (filter_match == 0 && sym_filter && !strcmp(name, sym_filter)) { - filter_match = 1; - filter_start = sym->start; - } - - return 0; } @@ -340,12 +828,13 @@ static int parse_symbols(void) { struct rb_node *node; struct symbol *sym; + int use_modules = vmlinux_name ? 1 : 0; kernel_dso = dso__new("[kernel]", sizeof(struct sym_entry)); if (kernel_dso == NULL) return -1; - if (dso__load_kernel(kernel_dso, NULL, symbol_filter, 1) != 0) + if (dso__load_kernel(kernel_dso, vmlinux_name, symbol_filter, verbose, use_modules) <= 0) goto out_delete_dso; node = rb_first(&kernel_dso->syms); @@ -367,8 +856,6 @@ out_delete_dso: return -1; } -#define TRACE_COUNT 3 - /* * Binary search in the histogram table and record the hit: */ @@ -381,6 +868,7 @@ static void record_ip(u64 ip, int counter) if (!syme->skip) { syme->count[counter]++; + record_precise_ip(syme, counter, ip); pthread_mutex_lock(&active_symbols_lock); if (list_empty(&syme->node) || !syme->node.next) __list_insert_active_sym(syme); @@ -407,13 +895,13 @@ static void process_event(u64 ip, int counter, int user) struct mmap_data { int counter; void *base; - unsigned int mask; + int mask; unsigned int prev; }; static unsigned int mmap_read_head(struct mmap_data *md) { - struct perf_counter_mmap_page *pc = md->base; + struct perf_event_mmap_page *pc = md->base; int head; head = pc->data_head; @@ -461,26 +949,6 @@ static void mmap_read_counter(struct mmap_data *md) last_read = this_read; for (; old != head;) { - struct ip_event { - struct perf_event_header header; - u64 ip; - u32 pid, target_pid; - }; - struct mmap_event { - struct perf_event_header header; - u32 pid, target_pid; - u64 start; - u64 len; - u64 pgoff; - char filename[PATH_MAX]; - }; - - typedef union event_union { - struct perf_event_header header; - struct ip_event ip; - struct mmap_event mmap; - } event_t; - event_t *event = (event_t *)&data[old & md->mask]; event_t event_copy; @@ -509,9 +977,9 @@ static void mmap_read_counter(struct mmap_data *md) old += size; - if (event->header.type == PERF_EVENT_SAMPLE) { + if (event->header.type == PERF_RECORD_SAMPLE) { int user = - (event->header.misc & PERF_EVENT_MISC_CPUMODE_MASK) == PERF_EVENT_MISC_USER; + (event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK) == PERF_RECORD_MISC_USER; process_event(event->ip.ip, md->counter, user); } } @@ -537,8 +1005,8 @@ int group_fd; static void start_counter(int i, int counter) { - struct perf_counter_attr *attr; - unsigned int cpu; + struct perf_event_attr *attr; + int cpu; cpu = profile_cpu; if (target_pid == -1 && profile_cpu == -1) @@ -548,9 +1016,10 @@ static void start_counter(int i, int counter) attr->sample_type = PERF_SAMPLE_IP | PERF_SAMPLE_TID; attr->freq = freq; + attr->inherit = (cpu < 0) && inherit; try_again: - fd[i][counter] = sys_perf_counter_open(attr, target_pid, cpu, group_fd, 0); + fd[i][counter] = sys_perf_event_open(attr, target_pid, cpu, group_fd, 0); if (fd[i][counter] < 0) { int err = errno; @@ -575,7 +1044,7 @@ try_again: printf("\n"); error("perfcounter syscall returned with %d (%s)\n", fd[i][counter], strerror(err)); - die("No CONFIG_PERF_COUNTERS=y kernel support configured?\n"); + die("No CONFIG_PERF_EVENTS=y kernel support configured?\n"); exit(-1); } assert(fd[i][counter] >= 0); @@ -661,6 +1130,7 @@ static const struct option options[] = { "system-wide collection from all CPUs"), OPT_INTEGER('C', "CPU", &profile_cpu, "CPU to profile on"), + OPT_STRING('k', "vmlinux", &vmlinux_name, "file", "vmlinux pathname"), OPT_INTEGER('m', "mmap-pages", &mmap_pages, "number of mmap data pages"), OPT_INTEGER('r', "realtime", &realtime_prio, @@ -673,9 +1143,11 @@ static const struct option options[] = { "only display functions with more events than this"), OPT_BOOLEAN('g', "group", &group, "put the counters into a counter group"), - OPT_STRING('s', "sym-filter", &sym_filter, "pattern", - "only display symbols matchig this pattern"), - OPT_BOOLEAN('z', "zero", &group, + OPT_BOOLEAN('i', "inherit", &inherit, + "child tasks inherit counters"), + OPT_STRING('s', "sym-annotate", &sym_filter, "symbol name", + "symbol to annotate - requires -k option"), + OPT_BOOLEAN('z', "zero", &zero, "zero history across updates"), OPT_INTEGER('F', "freq", &freq, "profile at this frequency"), @@ -686,10 +1158,12 @@ static const struct option options[] = { OPT_END() }; -int cmd_top(int argc, const char **argv, const char *prefix) +int cmd_top(int argc, const char **argv, const char *prefix __used) { int counter; + symbol__init(); + page_size = sysconf(_SC_PAGE_SIZE); argc = parse_options(argc, argv, options, top_usage, 0); @@ -715,6 +1189,7 @@ int cmd_top(int argc, const char **argv, const char *prefix) delay_secs = 1; parse_symbols(); + parse_source(sym_filter_entry); /* * Fill in the ones not specifically initialized via -c: diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c new file mode 100644 index 00000000000..e9d256e2f47 --- /dev/null +++ b/tools/perf/builtin-trace.c @@ -0,0 +1,297 @@ +#include "builtin.h" + +#include "util/util.h" +#include "util/cache.h" +#include "util/symbol.h" +#include "util/thread.h" +#include "util/header.h" + +#include "util/parse-options.h" + +#include "perf.h" +#include "util/debug.h" + +#include "util/trace-event.h" + +static char const *input_name = "perf.data"; +static int input; +static unsigned long page_size; +static unsigned long mmap_window = 32; + +static unsigned long total = 0; +static unsigned long total_comm = 0; + +static struct rb_root threads; +static struct thread *last_match; + +static struct perf_header *header; +static u64 sample_type; + + +static int +process_comm_event(event_t *event, unsigned long offset, unsigned long head) +{ + struct thread *thread; + + thread = threads__findnew(event->comm.pid, &threads, &last_match); + + dump_printf("%p [%p]: PERF_RECORD_COMM: %s:%d\n", + (void *)(offset + head), + (void *)(long)(event->header.size), + event->comm.comm, event->comm.pid); + + if (thread == NULL || + thread__set_comm(thread, event->comm.comm)) { + dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n"); + return -1; + } + total_comm++; + + return 0; +} + +static int +process_sample_event(event_t *event, unsigned long offset, unsigned long head) +{ + char level; + int show = 0; + struct dso *dso = NULL; + struct thread *thread; + u64 ip = event->ip.ip; + u64 timestamp = -1; + u32 cpu = -1; + u64 period = 1; + void *more_data = event->ip.__more_data; + int cpumode; + + thread = threads__findnew(event->ip.pid, &threads, &last_match); + + if (sample_type & PERF_SAMPLE_TIME) { + timestamp = *(u64 *)more_data; + more_data += sizeof(u64); + } + + if (sample_type & PERF_SAMPLE_CPU) { + cpu = *(u32 *)more_data; + more_data += sizeof(u32); + more_data += sizeof(u32); /* reserved */ + } + + if (sample_type & PERF_SAMPLE_PERIOD) { + period = *(u64 *)more_data; + more_data += sizeof(u64); + } + + dump_printf("%p [%p]: PERF_RECORD_SAMPLE (IP, %d): %d/%d: %p period: %Ld\n", + (void *)(offset + head), + (void *)(long)(event->header.size), + event->header.misc, + event->ip.pid, event->ip.tid, + (void *)(long)ip, + (long long)period); + + dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid); + + if (thread == NULL) { + eprintf("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + + if (cpumode == PERF_RECORD_MISC_KERNEL) { + show = SHOW_KERNEL; + level = 'k'; + + dso = kernel_dso; + + dump_printf(" ...... dso: %s\n", dso->name); + + } else if (cpumode == PERF_RECORD_MISC_USER) { + + show = SHOW_USER; + level = '.'; + + } else { + show = SHOW_HV; + level = 'H'; + + dso = hypervisor_dso; + + dump_printf(" ...... dso: [hypervisor]\n"); + } + + if (sample_type & PERF_SAMPLE_RAW) { + struct { + u32 size; + char data[0]; + } *raw = more_data; + + /* + * FIXME: better resolve from pid from the struct trace_entry + * field, although it should be the same than this perf + * event pid + */ + print_event(cpu, raw->data, raw->size, timestamp, thread->comm); + } + total += period; + + return 0; +} + +static int +process_event(event_t *event, unsigned long offset, unsigned long head) +{ + trace_event(event); + + switch (event->header.type) { + case PERF_RECORD_MMAP ... PERF_RECORD_LOST: + return 0; + + case PERF_RECORD_COMM: + return process_comm_event(event, offset, head); + + case PERF_RECORD_EXIT ... PERF_RECORD_READ: + return 0; + + case PERF_RECORD_SAMPLE: + return process_sample_event(event, offset, head); + + case PERF_RECORD_MAX: + default: + return -1; + } + + return 0; +} + +static int __cmd_trace(void) +{ + int ret, rc = EXIT_FAILURE; + unsigned long offset = 0; + unsigned long head = 0; + struct stat perf_stat; + event_t *event; + uint32_t size; + char *buf; + + trace_report(); + register_idle_thread(&threads, &last_match); + + input = open(input_name, O_RDONLY); + if (input < 0) { + perror("failed to open file"); + exit(-1); + } + + ret = fstat(input, &perf_stat); + if (ret < 0) { + perror("failed to stat file"); + exit(-1); + } + + if (!perf_stat.st_size) { + fprintf(stderr, "zero-sized file, nothing to do!\n"); + exit(0); + } + header = perf_header__read(input); + head = header->data_offset; + sample_type = perf_header__sample_type(header); + + if (!(sample_type & PERF_SAMPLE_RAW)) + die("No trace sample to read. Did you call perf record " + "without -R?"); + + if (load_kernel() < 0) { + perror("failed to load kernel symbols"); + return EXIT_FAILURE; + } + +remap: + buf = (char *)mmap(NULL, page_size * mmap_window, PROT_READ, + MAP_SHARED, input, offset); + if (buf == MAP_FAILED) { + perror("failed to mmap file"); + exit(-1); + } + +more: + event = (event_t *)(buf + head); + + size = event->header.size; + if (!size) + size = 8; + + if (head + event->header.size >= page_size * mmap_window) { + unsigned long shift = page_size * (head / page_size); + int res; + + res = munmap(buf, page_size * mmap_window); + assert(res == 0); + + offset += shift; + head -= shift; + goto remap; + } + + size = event->header.size; + + + if (!size || process_event(event, offset, head) < 0) { + + /* + * assume we lost track of the stream, check alignment, and + * increment a single u64 in the hope to catch on again 'soon'. + */ + + if (unlikely(head & 7)) + head &= ~7ULL; + + size = 8; + } + + head += size; + + if (offset + head < (unsigned long)perf_stat.st_size) + goto more; + + rc = EXIT_SUCCESS; + close(input); + + return rc; +} + +static const char * const annotate_usage[] = { + "perf trace [<options>] <command>", + NULL +}; + +static const struct option options[] = { + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_BOOLEAN('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_END() +}; + +int cmd_trace(int argc, const char **argv, const char *prefix __used) +{ + symbol__init(); + page_size = getpagesize(); + + argc = parse_options(argc, argv, options, annotate_usage, 0); + if (argc) { + /* + * Special case: if there's an argument left then assume tha + * it's a symbol filter: + */ + if (argc > 1) + usage_with_options(annotate_usage, options); + } + + + setup_pager(); + + return __cmd_trace(); +} diff --git a/tools/perf/builtin.h b/tools/perf/builtin.h index 51d168230ee..e11d8d231c3 100644 --- a/tools/perf/builtin.h +++ b/tools/perf/builtin.h @@ -16,11 +16,14 @@ extern int check_pager_config(const char *cmd); extern int cmd_annotate(int argc, const char **argv, const char *prefix); extern int cmd_help(int argc, const char **argv, const char *prefix); +extern int cmd_sched(int argc, const char **argv, const char *prefix); +extern int cmd_list(int argc, const char **argv, const char *prefix); extern int cmd_record(int argc, const char **argv, const char *prefix); extern int cmd_report(int argc, const char **argv, const char *prefix); extern int cmd_stat(int argc, const char **argv, const char *prefix); +extern int cmd_timechart(int argc, const char **argv, const char *prefix); extern int cmd_top(int argc, const char **argv, const char *prefix); +extern int cmd_trace(int argc, const char **argv, const char *prefix); extern int cmd_version(int argc, const char **argv, const char *prefix); -extern int cmd_list(int argc, const char **argv, const char *prefix); #endif diff --git a/tools/perf/command-list.txt b/tools/perf/command-list.txt index eebce30afbc..00326e230d8 100644 --- a/tools/perf/command-list.txt +++ b/tools/perf/command-list.txt @@ -4,7 +4,10 @@ # perf-annotate mainporcelain common perf-list mainporcelain common +perf-sched mainporcelain common perf-record mainporcelain common perf-report mainporcelain common perf-stat mainporcelain common +perf-timechart mainporcelain common perf-top mainporcelain common +perf-trace mainporcelain common diff --git a/tools/perf/design.txt b/tools/perf/design.txt index f71e0d245cb..f1946d107b1 100644 --- a/tools/perf/design.txt +++ b/tools/perf/design.txt @@ -18,10 +18,10 @@ underlying hardware counters. Performance counters are accessed via special file descriptors. There's one file descriptor per virtual counter used. -The special file descriptor is opened via the perf_counter_open() +The special file descriptor is opened via the perf_event_open() system call: - int sys_perf_counter_open(struct perf_counter_hw_event *hw_event_uptr, + int sys_perf_event_open(struct perf_event_hw_event *hw_event_uptr, pid_t pid, int cpu, int group_fd, unsigned long flags); @@ -32,9 +32,9 @@ can be used to set the blocking mode, etc. Multiple counters can be kept open at a time, and the counters can be poll()ed. -When creating a new counter fd, 'perf_counter_hw_event' is: +When creating a new counter fd, 'perf_event_hw_event' is: -struct perf_counter_hw_event { +struct perf_event_hw_event { /* * The MSB of the config word signifies if the rest contains cpu * specific (raw) counter configuration data, if unset, the next @@ -93,7 +93,7 @@ specified by 'event_id': /* * Generalized performance counter event types, used by the hw_event.event_id - * parameter of the sys_perf_counter_open() syscall: + * parameter of the sys_perf_event_open() syscall: */ enum hw_event_ids { /* @@ -159,7 +159,7 @@ in size. * reads on the counter should return the indicated quantities, * in increasing order of bit value, after the counter value. */ -enum perf_counter_read_format { +enum perf_event_read_format { PERF_FORMAT_TOTAL_TIME_ENABLED = 1, PERF_FORMAT_TOTAL_TIME_RUNNING = 2, }; @@ -178,7 +178,7 @@ interrupt: * Bits that can be set in hw_event.record_type to request information * in the overflow packets. */ -enum perf_counter_record_format { +enum perf_event_record_format { PERF_RECORD_IP = 1U << 0, PERF_RECORD_TID = 1U << 1, PERF_RECORD_TIME = 1U << 2, @@ -228,7 +228,7 @@ these events are recorded in the ring-buffer (see below). The 'comm' bit allows tracking of process comm data on process creation. This too is recorded in the ring-buffer (see below). -The 'pid' parameter to the perf_counter_open() system call allows the +The 'pid' parameter to the perf_event_open() system call allows the counter to be specific to a task: pid == 0: if the pid parameter is zero, the counter is attached to the @@ -258,7 +258,7 @@ The 'flags' parameter is currently unused and must be zero. The 'group_fd' parameter allows counter "groups" to be set up. A counter group has one counter which is the group "leader". The leader -is created first, with group_fd = -1 in the perf_counter_open call +is created first, with group_fd = -1 in the perf_event_open call that creates it. The rest of the group members are created subsequently, with group_fd giving the fd of the group leader. (A single counter on its own is created with group_fd = -1 and is @@ -277,13 +277,13 @@ tracking are logged into a ring-buffer. This ring-buffer is created and accessed through mmap(). The mmap size should be 1+2^n pages, where the first page is a meta-data page -(struct perf_counter_mmap_page) that contains various bits of information such +(struct perf_event_mmap_page) that contains various bits of information such as where the ring-buffer head is. /* * Structure of the page that can be mapped via mmap */ -struct perf_counter_mmap_page { +struct perf_event_mmap_page { __u32 version; /* version number of this structure */ __u32 compat_version; /* lowest version this is compat with */ @@ -317,7 +317,7 @@ struct perf_counter_mmap_page { * Control data for the mmap() data buffer. * * User-space reading this value should issue an rmb(), on SMP capable - * platforms, after reading this value -- see perf_counter_wakeup(). + * platforms, after reading this value -- see perf_event_wakeup(). */ __u32 data_head; /* head in the data section */ }; @@ -327,9 +327,9 @@ NOTE: the hw-counter userspace bits are arch specific and are currently only The following 2^n pages are the ring-buffer which contains events of the form: -#define PERF_EVENT_MISC_KERNEL (1 << 0) -#define PERF_EVENT_MISC_USER (1 << 1) -#define PERF_EVENT_MISC_OVERFLOW (1 << 2) +#define PERF_RECORD_MISC_KERNEL (1 << 0) +#define PERF_RECORD_MISC_USER (1 << 1) +#define PERF_RECORD_MISC_OVERFLOW (1 << 2) struct perf_event_header { __u32 type; @@ -353,8 +353,8 @@ enum perf_event_type { * char filename[]; * }; */ - PERF_EVENT_MMAP = 1, - PERF_EVENT_MUNMAP = 2, + PERF_RECORD_MMAP = 1, + PERF_RECORD_MUNMAP = 2, /* * struct { @@ -364,10 +364,10 @@ enum perf_event_type { * char comm[]; * }; */ - PERF_EVENT_COMM = 3, + PERF_RECORD_COMM = 3, /* - * When header.misc & PERF_EVENT_MISC_OVERFLOW the event_type field + * When header.misc & PERF_RECORD_MISC_OVERFLOW the event_type field * will be PERF_RECORD_* * * struct { @@ -397,7 +397,7 @@ Notification of new events is possible through poll()/select()/epoll() and fcntl() managing signals. Normally a notification is generated for every page filled, however one can -additionally set perf_counter_hw_event.wakeup_events to generate one every +additionally set perf_event_hw_event.wakeup_events to generate one every so many counter overflow events. Future work will include a splice() interface to the ring-buffer. @@ -409,11 +409,11 @@ events but does continue to exist and maintain its count value. An individual counter or counter group can be enabled with - ioctl(fd, PERF_COUNTER_IOC_ENABLE); + ioctl(fd, PERF_EVENT_IOC_ENABLE); or disabled with - ioctl(fd, PERF_COUNTER_IOC_DISABLE); + ioctl(fd, PERF_EVENT_IOC_DISABLE); Enabling or disabling the leader of a group enables or disables the whole group; that is, while the group leader is disabled, none of the @@ -424,16 +424,16 @@ other counter. Additionally, non-inherited overflow counters can use - ioctl(fd, PERF_COUNTER_IOC_REFRESH, nr); + ioctl(fd, PERF_EVENT_IOC_REFRESH, nr); to enable a counter for 'nr' events, after which it gets disabled again. A process can enable or disable all the counter groups that are attached to it, using prctl: - prctl(PR_TASK_PERF_COUNTERS_ENABLE); + prctl(PR_TASK_PERF_EVENTS_ENABLE); - prctl(PR_TASK_PERF_COUNTERS_DISABLE); + prctl(PR_TASK_PERF_EVENTS_DISABLE); This applies to all counters on the current process, whether created by this process or by another, and doesn't affect any counters that @@ -447,11 +447,11 @@ Arch requirements If your architecture does not have hardware performance metrics, you can still use the generic software counters based on hrtimers for sampling. -So to start with, in order to add HAVE_PERF_COUNTERS to your Kconfig, you +So to start with, in order to add HAVE_PERF_EVENTS to your Kconfig, you will need at least this: - - asm/perf_counter.h - a basic stub will suffice at first + - asm/perf_event.h - a basic stub will suffice at first - support for atomic64 types (and associated helper functions) - - set_perf_counter_pending() implemented + - set_perf_event_pending() implemented If your architecture does have hardware capabilities, you can override the -weak stub hw_perf_counter_init() to register hardware counters. +weak stub hw_perf_event_init() to register hardware counters. diff --git a/tools/perf/perf.c b/tools/perf/perf.c index 4eb72593370..19fc7feb9d5 100644 --- a/tools/perf/perf.c +++ b/tools/perf/perf.c @@ -12,6 +12,8 @@ #include "util/cache.h" #include "util/quote.h" #include "util/run-command.h" +#include "util/parse-events.h" +#include "util/string.h" const char perf_usage_string[] = "perf [--version] [--help] COMMAND [ARGS]"; @@ -25,6 +27,8 @@ struct pager_config { int val; }; +static char debugfs_mntpt[MAXPATHLEN]; + static int pager_command_config(const char *var, const char *value, void *data) { struct pager_config *c = data; @@ -56,6 +60,15 @@ static void commit_pager_choice(void) { } } +static void set_debugfs_path(void) +{ + char *path; + + path = getenv(PERF_DEBUGFS_ENVIRONMENT); + snprintf(debugfs_path, MAXPATHLEN, "%s/%s", path ?: debugfs_mntpt, + "tracing/events"); +} + static int handle_options(const char*** argv, int* argc, int* envchanged) { int handled = 0; @@ -122,6 +135,22 @@ static int handle_options(const char*** argv, int* argc, int* envchanged) setenv(PERF_WORK_TREE_ENVIRONMENT, cmd + 12, 1); if (envchanged) *envchanged = 1; + } else if (!strcmp(cmd, "--debugfs-dir")) { + if (*argc < 2) { + fprintf(stderr, "No directory given for --debugfs-dir.\n"); + usage(perf_usage_string); + } + strncpy(debugfs_mntpt, (*argv)[1], MAXPATHLEN); + debugfs_mntpt[MAXPATHLEN - 1] = '\0'; + if (envchanged) + *envchanged = 1; + (*argv)++; + (*argc)--; + } else if (!prefixcmp(cmd, "--debugfs-dir=")) { + strncpy(debugfs_mntpt, cmd + 14, MAXPATHLEN); + debugfs_mntpt[MAXPATHLEN - 1] = '\0'; + if (envchanged) + *envchanged = 1; } else { fprintf(stderr, "Unknown option: %s\n", cmd); usage(perf_usage_string); @@ -228,9 +257,7 @@ static int run_builtin(struct cmd_struct *p, int argc, const char **argv) if (use_pager == -1 && p->option & USE_PAGER) use_pager = 1; commit_pager_choice(); - - if (p->option & NEED_WORK_TREE) - /* setup_work_tree() */; + set_debugfs_path(); status = p->fn(argc, argv, prefix); if (status) @@ -262,11 +289,14 @@ static void handle_internal_command(int argc, const char **argv) { "record", cmd_record, 0 }, { "report", cmd_report, 0 }, { "stat", cmd_stat, 0 }, + { "timechart", cmd_timechart, 0 }, { "top", cmd_top, 0 }, { "annotate", cmd_annotate, 0 }, { "version", cmd_version, 0 }, + { "trace", cmd_trace, 0 }, + { "sched", cmd_sched, 0 }, }; - int i; + unsigned int i; static const char ext[] = STRIP_EXTENSION; if (sizeof(ext) > 1) { @@ -349,6 +379,49 @@ static int run_argv(int *argcp, const char ***argv) return done_alias; } +/* mini /proc/mounts parser: searching for "^blah /mount/point debugfs" */ +static void get_debugfs_mntpt(void) +{ + FILE *file; + char fs_type[100]; + char debugfs[MAXPATHLEN]; + + /* + * try the standard location + */ + if (valid_debugfs_mount("/sys/kernel/debug/") == 0) { + strcpy(debugfs_mntpt, "/sys/kernel/debug/"); + return; + } + + /* + * try the sane location + */ + if (valid_debugfs_mount("/debug/") == 0) { + strcpy(debugfs_mntpt, "/debug/"); + return; + } + + /* + * give up and parse /proc/mounts + */ + file = fopen("/proc/mounts", "r"); + if (file == NULL) + return; + + while (fscanf(file, "%*s %" + STR(MAXPATHLEN) + "s %99s %*s %*d %*d\n", + debugfs, fs_type) == 2) { + if (strcmp(fs_type, "debugfs") == 0) + break; + } + fclose(file); + if (strcmp(fs_type, "debugfs") == 0) { + strncpy(debugfs_mntpt, debugfs, MAXPATHLEN); + debugfs_mntpt[MAXPATHLEN - 1] = '\0'; + } +} int main(int argc, const char **argv) { @@ -357,7 +430,8 @@ int main(int argc, const char **argv) cmd = perf_extract_argv0_path(argv[0]); if (!cmd) cmd = "perf-help"; - + /* get debugfs mount point from /proc/mounts */ + get_debugfs_mntpt(); /* * "perf-xxxx" is the same as "perf xxxx", but we obviously: * @@ -380,6 +454,7 @@ int main(int argc, const char **argv) argc--; handle_options(&argv, &argc, NULL); commit_pager_choice(); + set_debugfs_path(); if (argc > 0) { if (!prefixcmp(argv[0], "--")) argv[0] += 2; diff --git a/tools/perf/perf.h b/tools/perf/perf.h index d3042a6ba03..8cc4623afd6 100644 --- a/tools/perf/perf.h +++ b/tools/perf/perf.h @@ -1,7 +1,13 @@ #ifndef _PERF_PERF_H #define _PERF_PERF_H -#if defined(__x86_64__) || defined(__i386__) +#if defined(__i386__) +#include "../../arch/x86/include/asm/unistd.h" +#define rmb() asm volatile("lock; addl $0,0(%%esp)" ::: "memory") +#define cpu_relax() asm volatile("rep; nop" ::: "memory"); +#endif + +#if defined(__x86_64__) #include "../../arch/x86/include/asm/unistd.h" #define rmb() asm volatile("lfence" ::: "memory") #define cpu_relax() asm volatile("rep; nop" ::: "memory"); @@ -35,20 +41,26 @@ #define cpu_relax() asm volatile("" ::: "memory"); #endif +#ifdef __sparc__ +#include "../../arch/sparc/include/asm/unistd.h" +#define rmb() asm volatile("":::"memory") +#define cpu_relax() asm volatile("":::"memory") +#endif + #include <time.h> #include <unistd.h> #include <sys/types.h> #include <sys/syscall.h> -#include "../../include/linux/perf_counter.h" +#include "../../include/linux/perf_event.h" #include "util/types.h" /* - * prctl(PR_TASK_PERF_COUNTERS_DISABLE) will (cheaply) disable all + * prctl(PR_TASK_PERF_EVENTS_DISABLE) will (cheaply) disable all * counters in the current task. */ -#define PR_TASK_PERF_COUNTERS_DISABLE 31 -#define PR_TASK_PERF_COUNTERS_ENABLE 32 +#define PR_TASK_PERF_EVENTS_DISABLE 31 +#define PR_TASK_PERF_EVENTS_ENABLE 32 #ifndef NSEC_PER_SEC # define NSEC_PER_SEC 1000000000ULL @@ -68,6 +80,8 @@ static inline unsigned long long rdclock(void) #define __user #define asmlinkage +#define __used __attribute__((__unused__)) + #define unlikely(x) __builtin_expect(!!(x), 0) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ @@ -76,12 +90,12 @@ static inline unsigned long long rdclock(void) _min1 < _min2 ? _min1 : _min2; }) static inline int -sys_perf_counter_open(struct perf_counter_attr *attr, +sys_perf_event_open(struct perf_event_attr *attr, pid_t pid, int cpu, int group_fd, unsigned long flags) { attr->size = sizeof(*attr); - return syscall(__NR_perf_counter_open, attr, pid, cpu, + return syscall(__NR_perf_event_open, attr, pid, cpu, group_fd, flags); } diff --git a/tools/perf/util/abspath.c b/tools/perf/util/abspath.c index 61d33b81fc9..a791dd46726 100644 --- a/tools/perf/util/abspath.c +++ b/tools/perf/util/abspath.c @@ -50,7 +50,8 @@ const char *make_absolute_path(const char *path) die ("Could not get current working directory"); if (last_elem) { - int len = strlen(buf); + len = strlen(buf); + if (len + strlen(last_elem) + 2 > PATH_MAX) die ("Too long path name: '%s/%s'", buf, last_elem); diff --git a/tools/perf/util/alias.c b/tools/perf/util/alias.c index 9b3dd2b428d..b8144e80bb1 100644 --- a/tools/perf/util/alias.c +++ b/tools/perf/util/alias.c @@ -3,7 +3,7 @@ static const char *alias_key; static char *alias_val; -static int alias_lookup_cb(const char *k, const char *v, void *cb) +static int alias_lookup_cb(const char *k, const char *v, void *cb __used) { if (!prefixcmp(k, "alias.") && !strcmp(k+6, alias_key)) { if (!v) diff --git a/tools/perf/util/cache.h b/tools/perf/util/cache.h index 393d6146d13..6f8ea9d210b 100644 --- a/tools/perf/util/cache.h +++ b/tools/perf/util/cache.h @@ -3,6 +3,7 @@ #include "util.h" #include "strbuf.h" +#include "../perf.h" #define PERF_DIR_ENVIRONMENT "PERF_DIR" #define PERF_WORK_TREE_ENVIRONMENT "PERF_WORK_TREE" @@ -17,6 +18,7 @@ #define PERFATTRIBUTES_FILE ".perfattributes" #define INFOATTRIBUTES_FILE "info/attributes" #define ATTRIBUTE_MACRO_PREFIX "[attr]" +#define PERF_DEBUGFS_ENVIRONMENT "PERF_DEBUGFS_DIR" typedef int (*config_fn_t)(const char *, const char *, void *); extern int perf_default_config(const char *, const char *, void *); @@ -50,7 +52,6 @@ extern const char *perf_mailmap_file; extern void maybe_flush_or_die(FILE *, const char *); extern int copy_fd(int ifd, int ofd); extern int copy_file(const char *dst, const char *src, int mode); -extern ssize_t read_in_full(int fd, void *buf, size_t count); extern ssize_t write_in_full(int fd, const void *buf, size_t count); extern void write_or_die(int fd, const void *buf, size_t count); extern int write_or_whine(int fd, const void *buf, size_t count, const char *msg); diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index ad3c2857896..3b8380f1b47 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -4,52 +4,159 @@ * Handle the callchains from the stream in an ad-hoc radix tree and then * sort them in an rbtree. * + * Using a radix for code path provides a fast retrieval and factorizes + * memory use. Also that lets us use the paths in a hierarchical graph view. + * */ #include <stdlib.h> #include <stdio.h> #include <stdbool.h> #include <errno.h> +#include <math.h> #include "callchain.h" +#define chain_for_each_child(child, parent) \ + list_for_each_entry(child, &parent->children, brothers) -static void rb_insert_callchain(struct rb_root *root, struct callchain_node *chain) +static void +rb_insert_callchain(struct rb_root *root, struct callchain_node *chain, + enum chain_mode mode) { struct rb_node **p = &root->rb_node; struct rb_node *parent = NULL; struct callchain_node *rnode; + u64 chain_cumul = cumul_hits(chain); while (*p) { + u64 rnode_cumul; + parent = *p; rnode = rb_entry(parent, struct callchain_node, rb_node); - - if (rnode->hit < chain->hit) - p = &(*p)->rb_left; - else - p = &(*p)->rb_right; + rnode_cumul = cumul_hits(rnode); + + switch (mode) { + case CHAIN_FLAT: + if (rnode->hit < chain->hit) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + break; + case CHAIN_GRAPH_ABS: /* Falldown */ + case CHAIN_GRAPH_REL: + if (rnode_cumul < chain_cumul) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + break; + case CHAIN_NONE: + default: + break; + } } rb_link_node(&chain->rb_node, parent, p); rb_insert_color(&chain->rb_node, root); } +static void +__sort_chain_flat(struct rb_root *rb_root, struct callchain_node *node, + u64 min_hit) +{ + struct callchain_node *child; + + chain_for_each_child(child, node) + __sort_chain_flat(rb_root, child, min_hit); + + if (node->hit && node->hit >= min_hit) + rb_insert_callchain(rb_root, node, CHAIN_FLAT); +} + /* * Once we get every callchains from the stream, we can now * sort them by hit */ -void sort_chain_to_rbtree(struct rb_root *rb_root, struct callchain_node *node) +static void +sort_chain_flat(struct rb_root *rb_root, struct callchain_node *node, + u64 min_hit, struct callchain_param *param __used) +{ + __sort_chain_flat(rb_root, node, min_hit); +} + +static void __sort_chain_graph_abs(struct callchain_node *node, + u64 min_hit) +{ + struct callchain_node *child; + + node->rb_root = RB_ROOT; + + chain_for_each_child(child, node) { + __sort_chain_graph_abs(child, min_hit); + if (cumul_hits(child) >= min_hit) + rb_insert_callchain(&node->rb_root, child, + CHAIN_GRAPH_ABS); + } +} + +static void +sort_chain_graph_abs(struct rb_root *rb_root, struct callchain_node *chain_root, + u64 min_hit, struct callchain_param *param __used) +{ + __sort_chain_graph_abs(chain_root, min_hit); + rb_root->rb_node = chain_root->rb_root.rb_node; +} + +static void __sort_chain_graph_rel(struct callchain_node *node, + double min_percent) { struct callchain_node *child; + u64 min_hit; + + node->rb_root = RB_ROOT; + min_hit = ceil(node->children_hit * min_percent); + + chain_for_each_child(child, node) { + __sort_chain_graph_rel(child, min_percent); + if (cumul_hits(child) >= min_hit) + rb_insert_callchain(&node->rb_root, child, + CHAIN_GRAPH_REL); + } +} - list_for_each_entry(child, &node->children, brothers) - sort_chain_to_rbtree(rb_root, child); +static void +sort_chain_graph_rel(struct rb_root *rb_root, struct callchain_node *chain_root, + u64 min_hit __used, struct callchain_param *param) +{ + __sort_chain_graph_rel(chain_root, param->min_percent / 100.0); + rb_root->rb_node = chain_root->rb_root.rb_node; +} - if (node->hit) - rb_insert_callchain(rb_root, node); +int register_callchain_param(struct callchain_param *param) +{ + switch (param->mode) { + case CHAIN_GRAPH_ABS: + param->sort = sort_chain_graph_abs; + break; + case CHAIN_GRAPH_REL: + param->sort = sort_chain_graph_rel; + break; + case CHAIN_FLAT: + param->sort = sort_chain_flat; + break; + case CHAIN_NONE: + default: + return -1; + } + return 0; } -static struct callchain_node *create_child(struct callchain_node *parent) +/* + * Create a child for a parent. If inherit_children, then the new child + * will become the new parent of it's parent children + */ +static struct callchain_node * +create_child(struct callchain_node *parent, bool inherit_children) { struct callchain_node *new; @@ -61,91 +168,150 @@ static struct callchain_node *create_child(struct callchain_node *parent) new->parent = parent; INIT_LIST_HEAD(&new->children); INIT_LIST_HEAD(&new->val); + + if (inherit_children) { + struct callchain_node *next; + + list_splice(&parent->children, &new->children); + INIT_LIST_HEAD(&parent->children); + + chain_for_each_child(next, new) + next->parent = new; + } list_add_tail(&new->brothers, &parent->children); return new; } +/* + * Fill the node with callchain values + */ static void -fill_node(struct callchain_node *node, struct ip_callchain *chain, int start) +fill_node(struct callchain_node *node, struct ip_callchain *chain, + int start, struct symbol **syms) { - int i; + unsigned int i; for (i = start; i < chain->nr; i++) { struct callchain_list *call; - call = malloc(sizeof(*chain)); + call = malloc(sizeof(*call)); if (!call) { perror("not enough memory for the code path tree"); return; } call->ip = chain->ips[i]; + call->sym = syms[i]; list_add_tail(&call->list, &node->val); } - node->val_nr = i - start; + node->val_nr = chain->nr - start; + if (!node->val_nr) + printf("Warning: empty node in callchain tree\n"); } -static void add_child(struct callchain_node *parent, struct ip_callchain *chain) +static void +add_child(struct callchain_node *parent, struct ip_callchain *chain, + int start, struct symbol **syms) { struct callchain_node *new; - new = create_child(parent); - fill_node(new, chain, parent->val_nr); + new = create_child(parent, false); + fill_node(new, chain, start, syms); + new->children_hit = 0; new->hit = 1; } +/* + * Split the parent in two parts (a new child is created) and + * give a part of its callchain to the created child. + * Then create another child to host the given callchain of new branch + */ static void split_add_child(struct callchain_node *parent, struct ip_callchain *chain, - struct callchain_list *to_split, int idx) + struct callchain_list *to_split, int idx_parents, int idx_local, + struct symbol **syms) { struct callchain_node *new; + struct list_head *old_tail; + unsigned int idx_total = idx_parents + idx_local; /* split */ - new = create_child(parent); - list_move_tail(&to_split->list, &new->val); - new->hit = parent->hit; - parent->hit = 0; - parent->val_nr = idx; + new = create_child(parent, true); - /* create the new one */ - add_child(parent, chain); + /* split the callchain and move a part to the new child */ + old_tail = parent->val.prev; + list_del_range(&to_split->list, old_tail); + new->val.next = &to_split->list; + new->val.prev = old_tail; + to_split->list.prev = &new->val; + old_tail->next = &new->val; + + /* split the hits */ + new->hit = parent->hit; + new->children_hit = parent->children_hit; + parent->children_hit = cumul_hits(new); + new->val_nr = parent->val_nr - idx_local; + parent->val_nr = idx_local; + + /* create a new child for the new branch if any */ + if (idx_total < chain->nr) { + parent->hit = 0; + add_child(parent, chain, idx_total, syms); + parent->children_hit++; + } else { + parent->hit = 1; + } } static int __append_chain(struct callchain_node *root, struct ip_callchain *chain, - int start); + unsigned int start, struct symbol **syms); -static int -__append_chain_children(struct callchain_node *root, struct ip_callchain *chain) +static void +__append_chain_children(struct callchain_node *root, struct ip_callchain *chain, + struct symbol **syms, unsigned int start) { struct callchain_node *rnode; /* lookup in childrens */ - list_for_each_entry(rnode, &root->children, brothers) { - int ret = __append_chain(rnode, chain, root->val_nr); + chain_for_each_child(rnode, root) { + unsigned int ret = __append_chain(rnode, chain, start, syms); + if (!ret) - return 0; + goto inc_children_hit; } - return -1; + /* nothing in children, add to the current node */ + add_child(root, chain, start, syms); + +inc_children_hit: + root->children_hit++; } static int __append_chain(struct callchain_node *root, struct ip_callchain *chain, - int start) + unsigned int start, struct symbol **syms) { struct callchain_list *cnode; - int i = start; + unsigned int i = start; bool found = false; - /* lookup in the current node */ + /* + * Lookup in the current node + * If we have a symbol, then compare the start to match + * anywhere inside a function. + */ list_for_each_entry(cnode, &root->val, list) { - if (cnode->ip != chain->ips[i++]) + if (i == chain->nr) + break; + if (cnode->sym && syms[i]) { + if (cnode->sym->start != syms[i]->start) + break; + } else if (cnode->ip != chain->ips[i]) break; if (!found) found = true; - if (i == chain->nr) - break; + i++; } /* matches not, relay on the parent */ @@ -153,22 +319,27 @@ __append_chain(struct callchain_node *root, struct ip_callchain *chain, return -1; /* we match only a part of the node. Split it and add the new chain */ - if (i < root->val_nr) { - split_add_child(root, chain, cnode, i); + if (i - start < root->val_nr) { + split_add_child(root, chain, cnode, start, i - start, syms); return 0; } /* we match 100% of the path, increment the hit */ - if (i == root->val_nr) { + if (i - start == root->val_nr && i == chain->nr) { root->hit++; return 0; } - return __append_chain_children(root, chain); + /* We match the node and still have a part remaining */ + __append_chain_children(root, chain, syms, i); + + return 0; } -void append_chain(struct callchain_node *root, struct ip_callchain *chain) +void append_chain(struct callchain_node *root, struct ip_callchain *chain, + struct symbol **syms) { - if (__append_chain_children(root, chain) == -1) - add_child(root, chain); + if (!chain->nr) + return; + __append_chain_children(root, chain, syms, 0); } diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index fa1cd2f71fd..43cf3ea9e08 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -2,22 +2,44 @@ #define __PERF_CALLCHAIN_H #include "../perf.h" -#include "list.h" -#include "rbtree.h" +#include <linux/list.h> +#include <linux/rbtree.h> +#include "util.h" +#include "symbol.h" +enum chain_mode { + CHAIN_NONE, + CHAIN_FLAT, + CHAIN_GRAPH_ABS, + CHAIN_GRAPH_REL +}; struct callchain_node { struct callchain_node *parent; struct list_head brothers; - struct list_head children; - struct list_head val; - struct rb_node rb_node; - int val_nr; - int hit; + struct list_head children; + struct list_head val; + struct rb_node rb_node; /* to sort nodes in an rbtree */ + struct rb_root rb_root; /* sorted tree of children */ + unsigned int val_nr; + u64 hit; + u64 children_hit; +}; + +struct callchain_param; + +typedef void (*sort_chain_func_t)(struct rb_root *, struct callchain_node *, + u64, struct callchain_param *); + +struct callchain_param { + enum chain_mode mode; + double min_percent; + sort_chain_func_t sort; }; struct callchain_list { - unsigned long ip; + u64 ip; + struct symbol *sym; struct list_head list; }; @@ -28,6 +50,12 @@ static inline void callchain_init(struct callchain_node *node) INIT_LIST_HEAD(&node->val); } -void append_chain(struct callchain_node *root, struct ip_callchain *chain); -void sort_chain_to_rbtree(struct rb_root *rb_root, struct callchain_node *node); +static inline u64 cumul_hits(struct callchain_node *node) +{ + return node->hit + node->children_hit; +} + +int register_callchain_param(struct callchain_param *param); +void append_chain(struct callchain_node *root, struct ip_callchain *chain, + struct symbol **syms); #endif diff --git a/tools/perf/util/color.c b/tools/perf/util/color.c index 9a8c20ccc53..e88bca55a59 100644 --- a/tools/perf/util/color.c +++ b/tools/perf/util/color.c @@ -11,7 +11,8 @@ static int parse_color(const char *name, int len) }; char *end; int i; - for (i = 0; i < ARRAY_SIZE(color_names); i++) { + + for (i = 0; i < (int)ARRAY_SIZE(color_names); i++) { const char *str = color_names[i]; if (!strncasecmp(name, str, len) && !str[len]) return i - 1; @@ -28,7 +29,8 @@ static int parse_attr(const char *name, int len) static const char * const attr_names[] = { "bold", "dim", "ul", "blink", "reverse" }; - int i; + unsigned int i; + for (i = 0; i < ARRAY_SIZE(attr_names); i++) { const char *str = attr_names[i]; if (!strncasecmp(name, str, len) && !str[len]) @@ -164,7 +166,7 @@ int perf_color_default_config(const char *var, const char *value, void *cb) return perf_default_config(var, value, cb); } -static int color_vfprintf(FILE *fp, const char *color, const char *fmt, +static int __color_vfprintf(FILE *fp, const char *color, const char *fmt, va_list args, const char *trail) { int r = 0; @@ -189,6 +191,10 @@ static int color_vfprintf(FILE *fp, const char *color, const char *fmt, return r; } +int color_vfprintf(FILE *fp, const char *color, const char *fmt, va_list args) +{ + return __color_vfprintf(fp, color, fmt, args, NULL); +} int color_fprintf(FILE *fp, const char *color, const char *fmt, ...) @@ -197,7 +203,7 @@ int color_fprintf(FILE *fp, const char *color, const char *fmt, ...) int r; va_start(args, fmt); - r = color_vfprintf(fp, color, fmt, args, NULL); + r = color_vfprintf(fp, color, fmt, args); va_end(args); return r; } @@ -207,7 +213,7 @@ int color_fprintf_ln(FILE *fp, const char *color, const char *fmt, ...) va_list args; int r; va_start(args, fmt); - r = color_vfprintf(fp, color, fmt, args, "\n"); + r = __color_vfprintf(fp, color, fmt, args, "\n"); va_end(args); return r; } @@ -222,10 +228,12 @@ int color_fwrite_lines(FILE *fp, const char *color, { if (!*color) return fwrite(buf, count, 1, fp) != 1; + while (count) { char *p = memchr(buf, '\n', count); + if (p != buf && (fputs(color, fp) < 0 || - fwrite(buf, p ? p - buf : count, 1, fp) != 1 || + fwrite(buf, p ? (size_t)(p - buf) : count, 1, fp) != 1 || fputs(PERF_COLOR_RESET, fp) < 0)) return -1; if (!p) @@ -238,4 +246,31 @@ int color_fwrite_lines(FILE *fp, const char *color, return 0; } +const char *get_percent_color(double percent) +{ + const char *color = PERF_COLOR_NORMAL; + /* + * We color high-overhead entries in red, mid-overhead + * entries in green - and keep the low overhead places + * normal: + */ + if (percent >= MIN_RED) + color = PERF_COLOR_RED; + else { + if (percent > MIN_GREEN) + color = PERF_COLOR_GREEN; + } + return color; +} + +int percent_color_fprintf(FILE *fp, const char *fmt, double percent) +{ + int r; + const char *color; + + color = get_percent_color(percent); + r = color_fprintf(fp, color, fmt, percent); + + return r; +} diff --git a/tools/perf/util/color.h b/tools/perf/util/color.h index 5abfd379582..58d597564b9 100644 --- a/tools/perf/util/color.h +++ b/tools/perf/util/color.h @@ -15,6 +15,9 @@ #define PERF_COLOR_CYAN "\033[36m" #define PERF_COLOR_BG_RED "\033[41m" +#define MIN_GREEN 0.5 +#define MIN_RED 5.0 + /* * This variable stores the value of color.ui */ @@ -29,8 +32,11 @@ int perf_color_default_config(const char *var, const char *value, void *cb); int perf_config_colorbool(const char *var, const char *value, int stdout_is_tty); void color_parse(const char *value, const char *var, char *dst); void color_parse_mem(const char *value, int len, const char *var, char *dst); +int color_vfprintf(FILE *fp, const char *color, const char *fmt, va_list args); int color_fprintf(FILE *fp, const char *color, const char *fmt, ...); int color_fprintf_ln(FILE *fp, const char *color, const char *fmt, ...); int color_fwrite_lines(FILE *fp, const char *color, size_t count, const char *buf); +int percent_color_fprintf(FILE *fp, const char *fmt, double percent); +const char *get_percent_color(double percent); #endif /* COLOR_H */ diff --git a/tools/perf/util/config.c b/tools/perf/util/config.c index 3dd13faa6a2..8784649109c 100644 --- a/tools/perf/util/config.c +++ b/tools/perf/util/config.c @@ -47,10 +47,12 @@ static int get_next_char(void) static char *parse_value(void) { static char value[1024]; - int quote = 0, comment = 0, len = 0, space = 0; + int quote = 0, comment = 0, space = 0; + size_t len = 0; for (;;) { int c = get_next_char(); + if (len >= sizeof(value) - 1) return NULL; if (c == '\n') { @@ -158,17 +160,18 @@ static int get_extended_base_var(char *name, int baselen, int c) name[baselen++] = '.'; for (;;) { - int c = get_next_char(); - if (c == '\n') + int ch = get_next_char(); + + if (ch == '\n') return -1; - if (c == '"') + if (ch == '"') break; - if (c == '\\') { - c = get_next_char(); - if (c == '\n') + if (ch == '\\') { + ch = get_next_char(); + if (ch == '\n') return -1; } - name[baselen++] = c; + name[baselen++] = ch; if (baselen > MAXNAME / 2) return -1; } @@ -353,13 +356,13 @@ int perf_config_string(const char **dest, const char *var, const char *value) return 0; } -static int perf_default_core_config(const char *var, const char *value) +static int perf_default_core_config(const char *var __used, const char *value __used) { /* Add other config variables here and to Documentation/config.txt. */ return 0; } -int perf_default_config(const char *var, const char *value, void *dummy) +int perf_default_config(const char *var, const char *value, void *dummy __used) { if (!prefixcmp(var, "core.")) return perf_default_core_config(var, value); @@ -471,10 +474,10 @@ static int matches(const char* key, const char* value) !regexec(store.value_regex, value, 0, NULL, 0))); } -static int store_aux(const char* key, const char* value, void *cb) +static int store_aux(const char* key, const char* value, void *cb __used) { + int section_len; const char *ep; - size_t section_len; switch (store.state) { case KEY_SEEN: @@ -528,6 +531,8 @@ static int store_aux(const char* key, const char* value, void *cb) store.offset[store.seen] = ftell(config_file); } } + default: + break; } return 0; } @@ -551,7 +556,7 @@ static int store_write_section(int fd, const char* key) strbuf_addf(&sb, "[%.*s]\n", store.baselen, key); } - success = write_in_full(fd, sb.buf, sb.len) == sb.len; + success = (write_in_full(fd, sb.buf, sb.len) == (ssize_t)sb.len); strbuf_release(&sb); return success; @@ -599,7 +604,7 @@ static int store_write_pair(int fd, const char* key, const char* value) } strbuf_addf(&sb, "%s\n", quote); - success = write_in_full(fd, sb.buf, sb.len) == sb.len; + success = (write_in_full(fd, sb.buf, sb.len) == (ssize_t)sb.len); strbuf_release(&sb); return success; @@ -617,6 +622,7 @@ contline: switch (contents[offset]) { case '=': equal_offset = offset; break; case ']': bracket_offset = offset; break; + default: break; } if (offset > 0 && contents[offset-1] == '\\') { offset_ = offset; @@ -740,9 +746,9 @@ int perf_config_set_multivar(const char* key, const char* value, goto write_err_out; } else { struct stat st; - char* contents; - size_t contents_sz, copy_begin, copy_end; - int i, new_line = 0; + char *contents; + ssize_t contents_sz, copy_begin, copy_end; + int new_line = 0; if (value_regex == NULL) store.value_regex = NULL; diff --git a/tools/perf/util/debug.c b/tools/perf/util/debug.c new file mode 100644 index 00000000000..e8ca98fe0bd --- /dev/null +++ b/tools/perf/util/debug.c @@ -0,0 +1,95 @@ +/* For general debugging purposes */ + +#include "../perf.h" + +#include <string.h> +#include <stdarg.h> +#include <stdio.h> + +#include "color.h" +#include "event.h" +#include "debug.h" + +int verbose = 0; +int dump_trace = 0; + +int eprintf(const char *fmt, ...) +{ + va_list args; + int ret = 0; + + if (verbose) { + va_start(args, fmt); + ret = vfprintf(stderr, fmt, args); + va_end(args); + } + + return ret; +} + +int dump_printf(const char *fmt, ...) +{ + va_list args; + int ret = 0; + + if (dump_trace) { + va_start(args, fmt); + ret = vprintf(fmt, args); + va_end(args); + } + + return ret; +} + +static int dump_printf_color(const char *fmt, const char *color, ...) +{ + va_list args; + int ret = 0; + + if (dump_trace) { + va_start(args, color); + ret = color_vfprintf(stdout, color, fmt, args); + va_end(args); + } + + return ret; +} + + +void trace_event(event_t *event) +{ + unsigned char *raw_event = (void *)event; + const char *color = PERF_COLOR_BLUE; + int i, j; + + if (!dump_trace) + return; + + dump_printf("."); + dump_printf_color("\n. ... raw event: size %d bytes\n", color, + event->header.size); + + for (i = 0; i < event->header.size; i++) { + if ((i & 15) == 0) { + dump_printf("."); + dump_printf_color(" %04x: ", color, i); + } + + dump_printf_color(" %02x", color, raw_event[i]); + + if (((i & 15) == 15) || i == event->header.size-1) { + dump_printf_color(" ", color); + for (j = 0; j < 15-(i & 15); j++) + dump_printf_color(" ", color); + for (j = 0; j < (i & 15); j++) { + if (isprint(raw_event[i-15+j])) + dump_printf_color("%c", color, + raw_event[i-15+j]); + else + dump_printf_color(".", color); + } + dump_printf_color("\n", color); + } + } + dump_printf(".\n"); +} diff --git a/tools/perf/util/debug.h b/tools/perf/util/debug.h new file mode 100644 index 00000000000..437eea58ce4 --- /dev/null +++ b/tools/perf/util/debug.h @@ -0,0 +1,8 @@ +/* For debugging general purposes */ + +extern int verbose; +extern int dump_trace; + +int eprintf(const char *fmt, ...) __attribute__((format(printf, 1, 2))); +int dump_printf(const char *fmt, ...) __attribute__((format(printf, 1, 2))); +void trace_event(event_t *event); diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h new file mode 100644 index 00000000000..2c9c26d6ded --- /dev/null +++ b/tools/perf/util/event.h @@ -0,0 +1,104 @@ +#ifndef __PERF_RECORD_H +#define __PERF_RECORD_H +#include "../perf.h" +#include "util.h" +#include <linux/list.h> + +enum { + SHOW_KERNEL = 1, + SHOW_USER = 2, + SHOW_HV = 4, +}; + +/* + * PERF_SAMPLE_IP | PERF_SAMPLE_TID | * + */ +struct ip_event { + struct perf_event_header header; + u64 ip; + u32 pid, tid; + unsigned char __more_data[]; +}; + +struct mmap_event { + struct perf_event_header header; + u32 pid, tid; + u64 start; + u64 len; + u64 pgoff; + char filename[PATH_MAX]; +}; + +struct comm_event { + struct perf_event_header header; + u32 pid, tid; + char comm[16]; +}; + +struct fork_event { + struct perf_event_header header; + u32 pid, ppid; + u32 tid, ptid; + u64 time; +}; + +struct lost_event { + struct perf_event_header header; + u64 id; + u64 lost; +}; + +/* + * PERF_FORMAT_ENABLED | PERF_FORMAT_RUNNING | PERF_FORMAT_ID + */ +struct read_event { + struct perf_event_header header; + u32 pid, tid; + u64 value; + u64 time_enabled; + u64 time_running; + u64 id; +}; + +struct sample_event{ + struct perf_event_header header; + u64 array[]; +}; + + +typedef union event_union { + struct perf_event_header header; + struct ip_event ip; + struct mmap_event mmap; + struct comm_event comm; + struct fork_event fork; + struct lost_event lost; + struct read_event read; + struct sample_event sample; +} event_t; + +struct map { + struct list_head node; + u64 start; + u64 end; + u64 pgoff; + u64 (*map_ip)(struct map *, u64); + struct dso *dso; +}; + +static inline u64 map__map_ip(struct map *map, u64 ip) +{ + return ip - map->start + map->pgoff; +} + +static inline u64 vdso__map_ip(struct map *map __used, u64 ip) +{ + return ip; +} + +struct map *map__new(struct mmap_event *event, char *cwd, int cwdlen); +struct map *map__clone(struct map *self); +int map__overlap(struct map *l, struct map *r); +size_t map__fprintf(struct map *self, FILE *fp); + +#endif diff --git a/tools/perf/util/exec_cmd.c b/tools/perf/util/exec_cmd.c index d3929226315..2745605dba1 100644 --- a/tools/perf/util/exec_cmd.c +++ b/tools/perf/util/exec_cmd.c @@ -1,9 +1,11 @@ #include "cache.h" #include "exec_cmd.h" #include "quote.h" + +#include <string.h> + #define MAX_ARGS 32 -extern char **environ; static const char *argv_exec_path; static const char *argv0_path; @@ -51,7 +53,7 @@ const char *perf_extract_argv0_path(const char *argv0) slash--; if (slash >= argv0) { - argv0_path = strndup(argv0, slash - argv0); + argv0_path = xstrndup(argv0, slash - argv0); return slash + 1; } diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 450384b3bbe..e306857b2c2 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -7,10 +7,9 @@ #include "header.h" /* - * + * Create new perf.data header attribute: */ - -struct perf_header_attr *perf_header_attr__new(struct perf_counter_attr *attr) +struct perf_header_attr *perf_header_attr__new(struct perf_event_attr *attr) { struct perf_header_attr *self = malloc(sizeof(*self)); @@ -43,9 +42,8 @@ void perf_header_attr__add_id(struct perf_header_attr *self, u64 id) } /* - * + * Create new perf.data header: */ - struct perf_header *perf_header__new(void) { struct perf_header *self = malloc(sizeof(*self)); @@ -86,6 +84,46 @@ void perf_header__add_attr(struct perf_header *self, self->attr[pos] = attr; } +#define MAX_EVENT_NAME 64 + +struct perf_trace_event_type { + u64 event_id; + char name[MAX_EVENT_NAME]; +}; + +static int event_count; +static struct perf_trace_event_type *events; + +void perf_header__push_event(u64 id, const char *name) +{ + if (strlen(name) > MAX_EVENT_NAME) + printf("Event %s will be truncated\n", name); + + if (!events) { + events = malloc(sizeof(struct perf_trace_event_type)); + if (!events) + die("nomem"); + } else { + events = realloc(events, (event_count + 1) * sizeof(struct perf_trace_event_type)); + if (!events) + die("nomem"); + } + memset(&events[event_count], 0, sizeof(struct perf_trace_event_type)); + events[event_count].event_id = id; + strncpy(events[event_count].name, name, MAX_EVENT_NAME - 1); + event_count++; +} + +char *perf_header__find_event(u64 id) +{ + int i; + for (i = 0 ; i < event_count; i++) { + if (events[i].event_id == id) + return events[i].name; + } + return NULL; +} + static const char *__perf_magic = "PERFFILE"; #define PERF_MAGIC (*(u64 *)__perf_magic) @@ -96,7 +134,7 @@ struct perf_file_section { }; struct perf_file_attr { - struct perf_counter_attr attr; + struct perf_event_attr attr; struct perf_file_section ids; }; @@ -106,6 +144,7 @@ struct perf_file_header { u64 attr_size; struct perf_file_section attrs; struct perf_file_section data; + struct perf_file_section event_types; }; static void do_write(int fd, void *buf, size_t size) @@ -154,6 +193,11 @@ void perf_header__write(struct perf_header *self, int fd) do_write(fd, &f_attr, sizeof(f_attr)); } + self->event_offset = lseek(fd, 0, SEEK_CUR); + self->event_size = event_count * sizeof(struct perf_trace_event_type); + if (events) + do_write(fd, events, self->event_size); + self->data_offset = lseek(fd, 0, SEEK_CUR); @@ -169,6 +213,10 @@ void perf_header__write(struct perf_header *self, int fd) .offset = self->data_offset, .size = self->data_size, }, + .event_types = { + .offset = self->event_offset, + .size = self->event_size, + }, }; lseek(fd, 0, SEEK_SET); @@ -185,6 +233,8 @@ static void do_read(int fd, void *buf, size_t size) if (ret < 0) die("failed to read"); + if (ret == 0) + die("failed to read: missing data"); size -= ret; buf += ret; @@ -213,9 +263,10 @@ struct perf_header *perf_header__read(int fd) for (i = 0; i < nr_attrs; i++) { struct perf_header_attr *attr; - off_t tmp = lseek(fd, 0, SEEK_CUR); + off_t tmp; do_read(fd, &f_attr, sizeof(f_attr)); + tmp = lseek(fd, 0, SEEK_CUR); attr = perf_header_attr__new(&f_attr.attr); @@ -231,12 +282,58 @@ struct perf_header *perf_header__read(int fd) lseek(fd, tmp, SEEK_SET); } + if (f_header.event_types.size) { + lseek(fd, f_header.event_types.offset, SEEK_SET); + events = malloc(f_header.event_types.size); + if (!events) + die("nomem"); + do_read(fd, events, f_header.event_types.size); + event_count = f_header.event_types.size / sizeof(struct perf_trace_event_type); + } + self->event_offset = f_header.event_types.offset; + self->event_size = f_header.event_types.size; + self->data_offset = f_header.data.offset; self->data_size = f_header.data.size; - lseek(fd, self->data_offset + self->data_size, SEEK_SET); + lseek(fd, self->data_offset, SEEK_SET); self->frozen = 1; return self; } + +u64 perf_header__sample_type(struct perf_header *header) +{ + u64 type = 0; + int i; + + for (i = 0; i < header->attrs; i++) { + struct perf_header_attr *attr = header->attr[i]; + + if (!type) + type = attr->attr.sample_type; + else if (type != attr->attr.sample_type) + die("non matching sample_type"); + } + + return type; +} + +struct perf_event_attr * +perf_header__find_attr(u64 id, struct perf_header *header) +{ + int i; + + for (i = 0; i < header->attrs; i++) { + struct perf_header_attr *attr = header->attr[i]; + int j; + + for (j = 0; j < attr->ids; j++) { + if (attr->id[j] == id) + return &attr->attr; + } + } + + return NULL; +} diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index b5ef53ad4c7..a0761bc7863 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -1,12 +1,12 @@ #ifndef _PERF_HEADER_H #define _PERF_HEADER_H -#include "../../../include/linux/perf_counter.h" +#include "../../../include/linux/perf_event.h" #include <sys/types.h> #include "types.h" struct perf_header_attr { - struct perf_counter_attr attr; + struct perf_event_attr attr; int ids, size; u64 *id; off_t id_offset; @@ -16,9 +16,11 @@ struct perf_header { int frozen; int attrs, size; struct perf_header_attr **attr; - off_t attr_offset; + s64 attr_offset; u64 data_offset; u64 data_size; + u64 event_offset; + u64 event_size; }; struct perf_header *perf_header__read(int fd); @@ -27,10 +29,18 @@ void perf_header__write(struct perf_header *self, int fd); void perf_header__add_attr(struct perf_header *self, struct perf_header_attr *attr); +void perf_header__push_event(u64 id, const char *name); +char *perf_header__find_event(u64 id); + + struct perf_header_attr * -perf_header_attr__new(struct perf_counter_attr *attr); +perf_header_attr__new(struct perf_event_attr *attr); void perf_header_attr__add_id(struct perf_header_attr *self, u64 id); +u64 perf_header__sample_type(struct perf_header *header); +struct perf_event_attr * +perf_header__find_attr(u64 id, struct perf_header *header); + struct perf_header *perf_header__new(void); diff --git a/tools/perf/util/help.c b/tools/perf/util/help.c index 17a00e0df2c..fbb00978b2e 100644 --- a/tools/perf/util/help.c +++ b/tools/perf/util/help.c @@ -26,7 +26,7 @@ static int term_columns(void) return 80; } -void add_cmdname(struct cmdnames *cmds, const char *name, int len) +void add_cmdname(struct cmdnames *cmds, const char *name, size_t len) { struct cmdname *ent = malloc(sizeof(*ent) + len + 1); @@ -40,7 +40,8 @@ void add_cmdname(struct cmdnames *cmds, const char *name, int len) static void clean_cmdnames(struct cmdnames *cmds) { - int i; + unsigned int i; + for (i = 0; i < cmds->cnt; ++i) free(cmds->names[i]); free(cmds->names); @@ -57,7 +58,7 @@ static int cmdname_compare(const void *a_, const void *b_) static void uniq(struct cmdnames *cmds) { - int i, j; + unsigned int i, j; if (!cmds->cnt) return; @@ -71,7 +72,7 @@ static void uniq(struct cmdnames *cmds) void exclude_cmds(struct cmdnames *cmds, struct cmdnames *excludes) { - int ci, cj, ei; + size_t ci, cj, ei; int cmp; ci = cj = ei = 0; @@ -106,8 +107,9 @@ static void pretty_print_string_list(struct cmdnames *cmds, int longest) printf(" "); for (j = 0; j < cols; j++) { - int n = j * rows + i; - int size = space; + unsigned int n = j * rows + i; + unsigned int size = space; + if (n >= cmds->cnt) break; if (j == cols-1 || n + rows >= cmds->cnt) @@ -208,7 +210,7 @@ void load_command_list(const char *prefix, void list_commands(const char *title, struct cmdnames *main_cmds, struct cmdnames *other_cmds) { - int i, longest = 0; + unsigned int i, longest = 0; for (i = 0; i < main_cmds->cnt; i++) if (longest < main_cmds->names[i]->len) @@ -239,7 +241,8 @@ void list_commands(const char *title, struct cmdnames *main_cmds, int is_in_cmdlist(struct cmdnames *c, const char *s) { - int i; + unsigned int i; + for (i = 0; i < c->cnt; i++) if (!strcmp(s, c->names[i]->name)) return 1; @@ -271,7 +274,8 @@ static int levenshtein_compare(const void *p1, const void *p2) static void add_cmd_list(struct cmdnames *cmds, struct cmdnames *old) { - int i; + unsigned int i; + ALLOC_GROW(cmds->names, cmds->cnt + old->cnt, cmds->alloc); for (i = 0; i < old->cnt; i++) @@ -283,7 +287,7 @@ static void add_cmd_list(struct cmdnames *cmds, struct cmdnames *old) const char *help_unknown_cmd(const char *cmd) { - int i, n = 0, best_similarity = 0; + unsigned int i, n = 0, best_similarity = 0; struct cmdnames main_cmds, other_cmds; memset(&main_cmds, 0, sizeof(main_cmds)); @@ -345,7 +349,7 @@ const char *help_unknown_cmd(const char *cmd) exit(1); } -int cmd_version(int argc, const char **argv, const char *prefix) +int cmd_version(int argc __used, const char **argv __used, const char *prefix __used) { printf("perf version %s\n", perf_version_string); return 0; diff --git a/tools/perf/util/help.h b/tools/perf/util/help.h index 56bc15406ff..7128783637b 100644 --- a/tools/perf/util/help.h +++ b/tools/perf/util/help.h @@ -2,8 +2,8 @@ #define HELP_H struct cmdnames { - int alloc; - int cnt; + size_t alloc; + size_t cnt; struct cmdname { size_t len; /* also used for similarity index in help.c */ char name[FLEX_ARRAY]; @@ -19,7 +19,7 @@ static inline void mput_char(char c, unsigned int num) void load_command_list(const char *prefix, struct cmdnames *main_cmds, struct cmdnames *other_cmds); -void add_cmdname(struct cmdnames *cmds, const char *name, int len); +void add_cmdname(struct cmdnames *cmds, const char *name, size_t len); /* Here we require that excludes is a sorted list. */ void exclude_cmds(struct cmdnames *cmds, struct cmdnames *excludes); int is_in_cmdlist(struct cmdnames *c, const char *s); diff --git a/tools/perf/util/include/asm/system.h b/tools/perf/util/include/asm/system.h new file mode 100644 index 00000000000..710cecca972 --- /dev/null +++ b/tools/perf/util/include/asm/system.h @@ -0,0 +1 @@ +/* Empty */ diff --git a/tools/perf/util/include/linux/kernel.h b/tools/perf/util/include/linux/kernel.h new file mode 100644 index 00000000000..a6b87390cb5 --- /dev/null +++ b/tools/perf/util/include/linux/kernel.h @@ -0,0 +1,29 @@ +#ifndef PERF_LINUX_KERNEL_H_ +#define PERF_LINUX_KERNEL_H_ + +#ifndef offsetof +#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER) +#endif + +#ifndef container_of +/** + * container_of - cast a member of a structure out to the containing structure + * @ptr: the pointer to the member. + * @type: the type of the container struct this is embedded in. + * @member: the name of the member within the struct. + * + */ +#define container_of(ptr, type, member) ({ \ + const typeof(((type *)0)->member) * __mptr = (ptr); \ + (type *)((char *)__mptr - offsetof(type, member)); }) +#endif + +#ifndef max +#define max(x, y) ({ \ + typeof(x) _max1 = (x); \ + typeof(y) _max2 = (y); \ + (void) (&_max1 == &_max2); \ + _max1 > _max2 ? _max1 : _max2; }) +#endif + +#endif diff --git a/tools/perf/util/include/linux/list.h b/tools/perf/util/include/linux/list.h new file mode 100644 index 00000000000..dbe4b814382 --- /dev/null +++ b/tools/perf/util/include/linux/list.h @@ -0,0 +1,18 @@ +#include "../../../../include/linux/list.h" + +#ifndef PERF_LIST_H +#define PERF_LIST_H +/** + * list_del_range - deletes range of entries from list. + * @begin: first element in the range to delete from the list. + * @end: last element in the range to delete from the list. + * Note: list_empty on the range of entries does not return true after this, + * the entries is in an undefined state. + */ +static inline void list_del_range(struct list_head *begin, + struct list_head *end) +{ + begin->prev->next = end->next; + end->next->prev = begin->prev; +} +#endif diff --git a/tools/perf/util/include/linux/module.h b/tools/perf/util/include/linux/module.h new file mode 100644 index 00000000000..b43e2dc21e0 --- /dev/null +++ b/tools/perf/util/include/linux/module.h @@ -0,0 +1,6 @@ +#ifndef PERF_LINUX_MODULE_H +#define PERF_LINUX_MODULE_H + +#define EXPORT_SYMBOL(name) + +#endif diff --git a/tools/perf/util/include/linux/poison.h b/tools/perf/util/include/linux/poison.h new file mode 100644 index 00000000000..fef6dbc9ce1 --- /dev/null +++ b/tools/perf/util/include/linux/poison.h @@ -0,0 +1 @@ +#include "../../../../include/linux/poison.h" diff --git a/tools/perf/util/include/linux/prefetch.h b/tools/perf/util/include/linux/prefetch.h new file mode 100644 index 00000000000..7841e485d8c --- /dev/null +++ b/tools/perf/util/include/linux/prefetch.h @@ -0,0 +1,6 @@ +#ifndef PERF_LINUX_PREFETCH_H +#define PERF_LINUX_PREFETCH_H + +static inline void prefetch(void *a __attribute__((unused))) { } + +#endif diff --git a/tools/perf/util/include/linux/rbtree.h b/tools/perf/util/include/linux/rbtree.h new file mode 100644 index 00000000000..7a243a14303 --- /dev/null +++ b/tools/perf/util/include/linux/rbtree.h @@ -0,0 +1 @@ +#include "../../../../include/linux/rbtree.h" diff --git a/tools/perf/util/list.h b/tools/perf/util/list.h deleted file mode 100644 index e2548e8072c..00000000000 --- a/tools/perf/util/list.h +++ /dev/null @@ -1,603 +0,0 @@ -#ifndef _LINUX_LIST_H -#define _LINUX_LIST_H -/* - Copyright (C) Cast of dozens, comes from the Linux kernel - - This program is free software; you can redistribute it and/or modify it - under the terms of version 2 of the GNU General Public License as - published by the Free Software Foundation. -*/ - -#include <stddef.h> - -/* - * These are non-NULL pointers that will result in page faults - * under normal circumstances, used to verify that nobody uses - * non-initialized list entries. - */ -#define LIST_POISON1 ((void *)0x00100100) -#define LIST_POISON2 ((void *)0x00200200) - -/** - * container_of - cast a member of a structure out to the containing structure - * @ptr: the pointer to the member. - * @type: the type of the container struct this is embedded in. - * @member: the name of the member within the struct. - * - */ -#define container_of(ptr, type, member) ({ \ - const typeof( ((type *)0)->member ) *__mptr = (ptr); \ - (type *)( (char *)__mptr - offsetof(type,member) );}) - -/* - * Simple doubly linked list implementation. - * - * Some of the internal functions ("__xxx") are useful when - * manipulating whole lists rather than single entries, as - * sometimes we already know the next/prev entries and we can - * generate better code by using them directly rather than - * using the generic single-entry routines. - */ - -struct list_head { - struct list_head *next, *prev; -}; - -#define LIST_HEAD_INIT(name) { &(name), &(name) } - -#define LIST_HEAD(name) \ - struct list_head name = LIST_HEAD_INIT(name) - -static inline void INIT_LIST_HEAD(struct list_head *list) -{ - list->next = list; - list->prev = list; -} - -/* - * Insert a new entry between two known consecutive entries. - * - * This is only for internal list manipulation where we know - * the prev/next entries already! - */ -static inline void __list_add(struct list_head *new, - struct list_head *prev, - struct list_head *next) -{ - next->prev = new; - new->next = next; - new->prev = prev; - prev->next = new; -} - -/** - * list_add - add a new entry - * @new: new entry to be added - * @head: list head to add it after - * - * Insert a new entry after the specified head. - * This is good for implementing stacks. - */ -static inline void list_add(struct list_head *new, struct list_head *head) -{ - __list_add(new, head, head->next); -} - -/** - * list_add_tail - add a new entry - * @new: new entry to be added - * @head: list head to add it before - * - * Insert a new entry before the specified head. - * This is useful for implementing queues. - */ -static inline void list_add_tail(struct list_head *new, struct list_head *head) -{ - __list_add(new, head->prev, head); -} - -/* - * Delete a list entry by making the prev/next entries - * point to each other. - * - * This is only for internal list manipulation where we know - * the prev/next entries already! - */ -static inline void __list_del(struct list_head * prev, struct list_head * next) -{ - next->prev = prev; - prev->next = next; -} - -/** - * list_del - deletes entry from list. - * @entry: the element to delete from the list. - * Note: list_empty on entry does not return true after this, the entry is - * in an undefined state. - */ -static inline void list_del(struct list_head *entry) -{ - __list_del(entry->prev, entry->next); - entry->next = LIST_POISON1; - entry->prev = LIST_POISON2; -} - -/** - * list_del_range - deletes range of entries from list. - * @beging: first element in the range to delete from the list. - * @beging: first element in the range to delete from the list. - * Note: list_empty on the range of entries does not return true after this, - * the entries is in an undefined state. - */ -static inline void list_del_range(struct list_head *begin, - struct list_head *end) -{ - begin->prev->next = end->next; - end->next->prev = begin->prev; -} - -/** - * list_replace - replace old entry by new one - * @old : the element to be replaced - * @new : the new element to insert - * Note: if 'old' was empty, it will be overwritten. - */ -static inline void list_replace(struct list_head *old, - struct list_head *new) -{ - new->next = old->next; - new->next->prev = new; - new->prev = old->prev; - new->prev->next = new; -} - -static inline void list_replace_init(struct list_head *old, - struct list_head *new) -{ - list_replace(old, new); - INIT_LIST_HEAD(old); -} - -/** - * list_del_init - deletes entry from list and reinitialize it. - * @entry: the element to delete from the list. - */ -static inline void list_del_init(struct list_head *entry) -{ - __list_del(entry->prev, entry->next); - INIT_LIST_HEAD(entry); -} - -/** - * list_move - delete from one list and add as another's head - * @list: the entry to move - * @head: the head that will precede our entry - */ -static inline void list_move(struct list_head *list, struct list_head *head) -{ - __list_del(list->prev, list->next); - list_add(list, head); -} - -/** - * list_move_tail - delete from one list and add as another's tail - * @list: the entry to move - * @head: the head that will follow our entry - */ -static inline void list_move_tail(struct list_head *list, - struct list_head *head) -{ - __list_del(list->prev, list->next); - list_add_tail(list, head); -} - -/** - * list_is_last - tests whether @list is the last entry in list @head - * @list: the entry to test - * @head: the head of the list - */ -static inline int list_is_last(const struct list_head *list, - const struct list_head *head) -{ - return list->next == head; -} - -/** - * list_empty - tests whether a list is empty - * @head: the list to test. - */ -static inline int list_empty(const struct list_head *head) -{ - return head->next == head; -} - -/** - * list_empty_careful - tests whether a list is empty and not being modified - * @head: the list to test - * - * Description: - * tests whether a list is empty _and_ checks that no other CPU might be - * in the process of modifying either member (next or prev) - * - * NOTE: using list_empty_careful() without synchronization - * can only be safe if the only activity that can happen - * to the list entry is list_del_init(). Eg. it cannot be used - * if another CPU could re-list_add() it. - */ -static inline int list_empty_careful(const struct list_head *head) -{ - struct list_head *next = head->next; - return (next == head) && (next == head->prev); -} - -static inline void __list_splice(struct list_head *list, - struct list_head *head) -{ - struct list_head *first = list->next; - struct list_head *last = list->prev; - struct list_head *at = head->next; - - first->prev = head; - head->next = first; - - last->next = at; - at->prev = last; -} - -/** - * list_splice - join two lists - * @list: the new list to add. - * @head: the place to add it in the first list. - */ -static inline void list_splice(struct list_head *list, struct list_head *head) -{ - if (!list_empty(list)) - __list_splice(list, head); -} - -/** - * list_splice_init - join two lists and reinitialise the emptied list. - * @list: the new list to add. - * @head: the place to add it in the first list. - * - * The list at @list is reinitialised - */ -static inline void list_splice_init(struct list_head *list, - struct list_head *head) -{ - if (!list_empty(list)) { - __list_splice(list, head); - INIT_LIST_HEAD(list); - } -} - -/** - * list_entry - get the struct for this entry - * @ptr: the &struct list_head pointer. - * @type: the type of the struct this is embedded in. - * @member: the name of the list_struct within the struct. - */ -#define list_entry(ptr, type, member) \ - container_of(ptr, type, member) - -/** - * list_first_entry - get the first element from a list - * @ptr: the list head to take the element from. - * @type: the type of the struct this is embedded in. - * @member: the name of the list_struct within the struct. - * - * Note, that list is expected to be not empty. - */ -#define list_first_entry(ptr, type, member) \ - list_entry((ptr)->next, type, member) - -/** - * list_for_each - iterate over a list - * @pos: the &struct list_head to use as a loop cursor. - * @head: the head for your list. - */ -#define list_for_each(pos, head) \ - for (pos = (head)->next; pos != (head); \ - pos = pos->next) - -/** - * __list_for_each - iterate over a list - * @pos: the &struct list_head to use as a loop cursor. - * @head: the head for your list. - * - * This variant differs from list_for_each() in that it's the - * simplest possible list iteration code, no prefetching is done. - * Use this for code that knows the list to be very short (empty - * or 1 entry) most of the time. - */ -#define __list_for_each(pos, head) \ - for (pos = (head)->next; pos != (head); pos = pos->next) - -/** - * list_for_each_prev - iterate over a list backwards - * @pos: the &struct list_head to use as a loop cursor. - * @head: the head for your list. - */ -#define list_for_each_prev(pos, head) \ - for (pos = (head)->prev; pos != (head); \ - pos = pos->prev) - -/** - * list_for_each_safe - iterate over a list safe against removal of list entry - * @pos: the &struct list_head to use as a loop cursor. - * @n: another &struct list_head to use as temporary storage - * @head: the head for your list. - */ -#define list_for_each_safe(pos, n, head) \ - for (pos = (head)->next, n = pos->next; pos != (head); \ - pos = n, n = pos->next) - -/** - * list_for_each_entry - iterate over list of given type - * @pos: the type * to use as a loop cursor. - * @head: the head for your list. - * @member: the name of the list_struct within the struct. - */ -#define list_for_each_entry(pos, head, member) \ - for (pos = list_entry((head)->next, typeof(*pos), member); \ - &pos->member != (head); \ - pos = list_entry(pos->member.next, typeof(*pos), member)) - -/** - * list_for_each_entry_reverse - iterate backwards over list of given type. - * @pos: the type * to use as a loop cursor. - * @head: the head for your list. - * @member: the name of the list_struct within the struct. - */ -#define list_for_each_entry_reverse(pos, head, member) \ - for (pos = list_entry((head)->prev, typeof(*pos), member); \ - &pos->member != (head); \ - pos = list_entry(pos->member.prev, typeof(*pos), member)) - -/** - * list_prepare_entry - prepare a pos entry for use in list_for_each_entry_continue - * @pos: the type * to use as a start point - * @head: the head of the list - * @member: the name of the list_struct within the struct. - * - * Prepares a pos entry for use as a start point in list_for_each_entry_continue. - */ -#define list_prepare_entry(pos, head, member) \ - ((pos) ? : list_entry(head, typeof(*pos), member)) - -/** - * list_for_each_entry_continue - continue iteration over list of given type - * @pos: the type * to use as a loop cursor. - * @head: the head for your list. - * @member: the name of the list_struct within the struct. - * - * Continue to iterate over list of given type, continuing after - * the current position. - */ -#define list_for_each_entry_continue(pos, head, member) \ - for (pos = list_entry(pos->member.next, typeof(*pos), member); \ - &pos->member != (head); \ - pos = list_entry(pos->member.next, typeof(*pos), member)) - -/** - * list_for_each_entry_from - iterate over list of given type from the current point - * @pos: the type * to use as a loop cursor. - * @head: the head for your list. - * @member: the name of the list_struct within the struct. - * - * Iterate over list of given type, continuing from current position. - */ -#define list_for_each_entry_from(pos, head, member) \ - for (; &pos->member != (head); \ - pos = list_entry(pos->member.next, typeof(*pos), member)) - -/** - * list_for_each_entry_safe - iterate over list of given type safe against removal of list entry - * @pos: the type * to use as a loop cursor. - * @n: another type * to use as temporary storage - * @head: the head for your list. - * @member: the name of the list_struct within the struct. - */ -#define list_for_each_entry_safe(pos, n, head, member) \ - for (pos = list_entry((head)->next, typeof(*pos), member), \ - n = list_entry(pos->member.next, typeof(*pos), member); \ - &pos->member != (head); \ - pos = n, n = list_entry(n->member.next, typeof(*n), member)) - -/** - * list_for_each_entry_safe_continue - * @pos: the type * to use as a loop cursor. - * @n: another type * to use as temporary storage - * @head: the head for your list. - * @member: the name of the list_struct within the struct. - * - * Iterate over list of given type, continuing after current point, - * safe against removal of list entry. - */ -#define list_for_each_entry_safe_continue(pos, n, head, member) \ - for (pos = list_entry(pos->member.next, typeof(*pos), member), \ - n = list_entry(pos->member.next, typeof(*pos), member); \ - &pos->member != (head); \ - pos = n, n = list_entry(n->member.next, typeof(*n), member)) - -/** - * list_for_each_entry_safe_from - * @pos: the type * to use as a loop cursor. - * @n: another type * to use as temporary storage - * @head: the head for your list. - * @member: the name of the list_struct within the struct. - * - * Iterate over list of given type from current point, safe against - * removal of list entry. - */ -#define list_for_each_entry_safe_from(pos, n, head, member) \ - for (n = list_entry(pos->member.next, typeof(*pos), member); \ - &pos->member != (head); \ - pos = n, n = list_entry(n->member.next, typeof(*n), member)) - -/** - * list_for_each_entry_safe_reverse - * @pos: the type * to use as a loop cursor. - * @n: another type * to use as temporary storage - * @head: the head for your list. - * @member: the name of the list_struct within the struct. - * - * Iterate backwards over list of given type, safe against removal - * of list entry. - */ -#define list_for_each_entry_safe_reverse(pos, n, head, member) \ - for (pos = list_entry((head)->prev, typeof(*pos), member), \ - n = list_entry(pos->member.prev, typeof(*pos), member); \ - &pos->member != (head); \ - pos = n, n = list_entry(n->member.prev, typeof(*n), member)) - -/* - * Double linked lists with a single pointer list head. - * Mostly useful for hash tables where the two pointer list head is - * too wasteful. - * You lose the ability to access the tail in O(1). - */ - -struct hlist_head { - struct hlist_node *first; -}; - -struct hlist_node { - struct hlist_node *next, **pprev; -}; - -#define HLIST_HEAD_INIT { .first = NULL } -#define HLIST_HEAD(name) struct hlist_head name = { .first = NULL } -#define INIT_HLIST_HEAD(ptr) ((ptr)->first = NULL) -static inline void INIT_HLIST_NODE(struct hlist_node *h) -{ - h->next = NULL; - h->pprev = NULL; -} - -static inline int hlist_unhashed(const struct hlist_node *h) -{ - return !h->pprev; -} - -static inline int hlist_empty(const struct hlist_head *h) -{ - return !h->first; -} - -static inline void __hlist_del(struct hlist_node *n) -{ - struct hlist_node *next = n->next; - struct hlist_node **pprev = n->pprev; - *pprev = next; - if (next) - next->pprev = pprev; -} - -static inline void hlist_del(struct hlist_node *n) -{ - __hlist_del(n); - n->next = LIST_POISON1; - n->pprev = LIST_POISON2; -} - -static inline void hlist_del_init(struct hlist_node *n) -{ - if (!hlist_unhashed(n)) { - __hlist_del(n); - INIT_HLIST_NODE(n); - } -} - -static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h) -{ - struct hlist_node *first = h->first; - n->next = first; - if (first) - first->pprev = &n->next; - h->first = n; - n->pprev = &h->first; -} - -/* next must be != NULL */ -static inline void hlist_add_before(struct hlist_node *n, - struct hlist_node *next) -{ - n->pprev = next->pprev; - n->next = next; - next->pprev = &n->next; - *(n->pprev) = n; -} - -static inline void hlist_add_after(struct hlist_node *n, - struct hlist_node *next) -{ - next->next = n->next; - n->next = next; - next->pprev = &n->next; - - if(next->next) - next->next->pprev = &next->next; -} - -#define hlist_entry(ptr, type, member) container_of(ptr,type,member) - -#define hlist_for_each(pos, head) \ - for (pos = (head)->first; pos; \ - pos = pos->next) - -#define hlist_for_each_safe(pos, n, head) \ - for (pos = (head)->first; pos && ({ n = pos->next; 1; }); \ - pos = n) - -/** - * hlist_for_each_entry - iterate over list of given type - * @tpos: the type * to use as a loop cursor. - * @pos: the &struct hlist_node to use as a loop cursor. - * @head: the head for your list. - * @member: the name of the hlist_node within the struct. - */ -#define hlist_for_each_entry(tpos, pos, head, member) \ - for (pos = (head)->first; \ - pos && \ - ({ tpos = hlist_entry(pos, typeof(*tpos), member); 1;}); \ - pos = pos->next) - -/** - * hlist_for_each_entry_continue - iterate over a hlist continuing after current point - * @tpos: the type * to use as a loop cursor. - * @pos: the &struct hlist_node to use as a loop cursor. - * @member: the name of the hlist_node within the struct. - */ -#define hlist_for_each_entry_continue(tpos, pos, member) \ - for (pos = (pos)->next; \ - pos && \ - ({ tpos = hlist_entry(pos, typeof(*tpos), member); 1;}); \ - pos = pos->next) - -/** - * hlist_for_each_entry_from - iterate over a hlist continuing from current point - * @tpos: the type * to use as a loop cursor. - * @pos: the &struct hlist_node to use as a loop cursor. - * @member: the name of the hlist_node within the struct. - */ -#define hlist_for_each_entry_from(tpos, pos, member) \ - for (; pos && \ - ({ tpos = hlist_entry(pos, typeof(*tpos), member); 1;}); \ - pos = pos->next) - -/** - * hlist_for_each_entry_safe - iterate over list of given type safe against removal of list entry - * @tpos: the type * to use as a loop cursor. - * @pos: the &struct hlist_node to use as a loop cursor. - * @n: another &struct hlist_node to use as temporary storage - * @head: the head for your list. - * @member: the name of the hlist_node within the struct. - */ -#define hlist_for_each_entry_safe(tpos, pos, n, head, member) \ - for (pos = (head)->first; \ - pos && ({ n = pos->next; 1; }) && \ - ({ tpos = hlist_entry(pos, typeof(*tpos), member); 1;}); \ - pos = n) - -#endif diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c new file mode 100644 index 00000000000..804e0238273 --- /dev/null +++ b/tools/perf/util/map.c @@ -0,0 +1,97 @@ +#include "event.h" +#include "symbol.h" +#include <stdlib.h> +#include <string.h> +#include <stdio.h> + +static inline int is_anon_memory(const char *filename) +{ + return strcmp(filename, "//anon") == 0; +} + +static int strcommon(const char *pathname, char *cwd, int cwdlen) +{ + int n = 0; + + while (n < cwdlen && pathname[n] == cwd[n]) + ++n; + + return n; +} + + struct map *map__new(struct mmap_event *event, char *cwd, int cwdlen) +{ + struct map *self = malloc(sizeof(*self)); + + if (self != NULL) { + const char *filename = event->filename; + char newfilename[PATH_MAX]; + int anon; + + if (cwd) { + int n = strcommon(filename, cwd, cwdlen); + + if (n == cwdlen) { + snprintf(newfilename, sizeof(newfilename), + ".%s", filename + n); + filename = newfilename; + } + } + + anon = is_anon_memory(filename); + + if (anon) { + snprintf(newfilename, sizeof(newfilename), "/tmp/perf-%d.map", event->pid); + filename = newfilename; + } + + self->start = event->start; + self->end = event->start + event->len; + self->pgoff = event->pgoff; + + self->dso = dsos__findnew(filename); + if (self->dso == NULL) + goto out_delete; + + if (self->dso == vdso || anon) + self->map_ip = vdso__map_ip; + else + self->map_ip = map__map_ip; + } + return self; +out_delete: + free(self); + return NULL; +} + +struct map *map__clone(struct map *self) +{ + struct map *map = malloc(sizeof(*self)); + + if (!map) + return NULL; + + memcpy(map, self, sizeof(*self)); + + return map; +} + +int map__overlap(struct map *l, struct map *r) +{ + if (l->start > r->start) { + struct map *t = l; + l = r; + r = t; + } + + if (l->end > r->start) + return 1; + + return 0; +} + +size_t map__fprintf(struct map *self, FILE *fp) +{ + return fprintf(fp, " %Lx-%Lx %Lx %s\n", + self->start, self->end, self->pgoff, self->dso->name); +} diff --git a/tools/perf/util/module.c b/tools/perf/util/module.c new file mode 100644 index 00000000000..3d567fe59c7 --- /dev/null +++ b/tools/perf/util/module.c @@ -0,0 +1,509 @@ +#include "util.h" +#include "../perf.h" +#include "string.h" +#include "module.h" + +#include <libelf.h> +#include <gelf.h> +#include <elf.h> +#include <dirent.h> +#include <sys/utsname.h> + +static unsigned int crc32(const char *p, unsigned int len) +{ + int i; + unsigned int crc = 0; + + while (len--) { + crc ^= *p++; + for (i = 0; i < 8; i++) + crc = (crc >> 1) ^ ((crc & 1) ? 0xedb88320 : 0); + } + return crc; +} + +/* module section methods */ + +struct sec_dso *sec_dso__new_dso(const char *name) +{ + struct sec_dso *self = malloc(sizeof(*self) + strlen(name) + 1); + + if (self != NULL) { + strcpy(self->name, name); + self->secs = RB_ROOT; + self->find_section = sec_dso__find_section; + } + + return self; +} + +static void sec_dso__delete_section(struct section *self) +{ + free(((void *)self)); +} + +void sec_dso__delete_sections(struct sec_dso *self) +{ + struct section *pos; + struct rb_node *next = rb_first(&self->secs); + + while (next) { + pos = rb_entry(next, struct section, rb_node); + next = rb_next(&pos->rb_node); + rb_erase(&pos->rb_node, &self->secs); + sec_dso__delete_section(pos); + } +} + +void sec_dso__delete_self(struct sec_dso *self) +{ + sec_dso__delete_sections(self); + free(self); +} + +static void sec_dso__insert_section(struct sec_dso *self, struct section *sec) +{ + struct rb_node **p = &self->secs.rb_node; + struct rb_node *parent = NULL; + const u64 hash = sec->hash; + struct section *s; + + while (*p != NULL) { + parent = *p; + s = rb_entry(parent, struct section, rb_node); + if (hash < s->hash) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + rb_link_node(&sec->rb_node, parent, p); + rb_insert_color(&sec->rb_node, &self->secs); +} + +struct section *sec_dso__find_section(struct sec_dso *self, const char *name) +{ + struct rb_node *n; + u64 hash; + int len; + + if (self == NULL) + return NULL; + + len = strlen(name); + hash = crc32(name, len); + + n = self->secs.rb_node; + + while (n) { + struct section *s = rb_entry(n, struct section, rb_node); + + if (hash < s->hash) + n = n->rb_left; + else if (hash > s->hash) + n = n->rb_right; + else { + if (!strcmp(name, s->name)) + return s; + else + n = rb_next(&s->rb_node); + } + } + + return NULL; +} + +static size_t sec_dso__fprintf_section(struct section *self, FILE *fp) +{ + return fprintf(fp, "name:%s vma:%llx path:%s\n", + self->name, self->vma, self->path); +} + +size_t sec_dso__fprintf(struct sec_dso *self, FILE *fp) +{ + size_t ret = fprintf(fp, "dso: %s\n", self->name); + + struct rb_node *nd; + for (nd = rb_first(&self->secs); nd; nd = rb_next(nd)) { + struct section *pos = rb_entry(nd, struct section, rb_node); + ret += sec_dso__fprintf_section(pos, fp); + } + + return ret; +} + +static struct section *section__new(const char *name, const char *path) +{ + struct section *self = calloc(1, sizeof(*self)); + + if (!self) + goto out_failure; + + self->name = calloc(1, strlen(name) + 1); + if (!self->name) + goto out_failure; + + self->path = calloc(1, strlen(path) + 1); + if (!self->path) + goto out_failure; + + strcpy(self->name, name); + strcpy(self->path, path); + self->hash = crc32(self->name, strlen(name)); + + return self; + +out_failure: + if (self) { + if (self->name) + free(self->name); + if (self->path) + free(self->path); + free(self); + } + + return NULL; +} + +/* module methods */ + +struct mod_dso *mod_dso__new_dso(const char *name) +{ + struct mod_dso *self = malloc(sizeof(*self) + strlen(name) + 1); + + if (self != NULL) { + strcpy(self->name, name); + self->mods = RB_ROOT; + self->find_module = mod_dso__find_module; + } + + return self; +} + +static void mod_dso__delete_module(struct module *self) +{ + free(((void *)self)); +} + +void mod_dso__delete_modules(struct mod_dso *self) +{ + struct module *pos; + struct rb_node *next = rb_first(&self->mods); + + while (next) { + pos = rb_entry(next, struct module, rb_node); + next = rb_next(&pos->rb_node); + rb_erase(&pos->rb_node, &self->mods); + mod_dso__delete_module(pos); + } +} + +void mod_dso__delete_self(struct mod_dso *self) +{ + mod_dso__delete_modules(self); + free(self); +} + +static void mod_dso__insert_module(struct mod_dso *self, struct module *mod) +{ + struct rb_node **p = &self->mods.rb_node; + struct rb_node *parent = NULL; + const u64 hash = mod->hash; + struct module *m; + + while (*p != NULL) { + parent = *p; + m = rb_entry(parent, struct module, rb_node); + if (hash < m->hash) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + rb_link_node(&mod->rb_node, parent, p); + rb_insert_color(&mod->rb_node, &self->mods); +} + +struct module *mod_dso__find_module(struct mod_dso *self, const char *name) +{ + struct rb_node *n; + u64 hash; + int len; + + if (self == NULL) + return NULL; + + len = strlen(name); + hash = crc32(name, len); + + n = self->mods.rb_node; + + while (n) { + struct module *m = rb_entry(n, struct module, rb_node); + + if (hash < m->hash) + n = n->rb_left; + else if (hash > m->hash) + n = n->rb_right; + else { + if (!strcmp(name, m->name)) + return m; + else + n = rb_next(&m->rb_node); + } + } + + return NULL; +} + +static size_t mod_dso__fprintf_module(struct module *self, FILE *fp) +{ + return fprintf(fp, "name:%s path:%s\n", self->name, self->path); +} + +size_t mod_dso__fprintf(struct mod_dso *self, FILE *fp) +{ + struct rb_node *nd; + size_t ret; + + ret = fprintf(fp, "dso: %s\n", self->name); + + for (nd = rb_first(&self->mods); nd; nd = rb_next(nd)) { + struct module *pos = rb_entry(nd, struct module, rb_node); + + ret += mod_dso__fprintf_module(pos, fp); + } + + return ret; +} + +static struct module *module__new(const char *name, const char *path) +{ + struct module *self = calloc(1, sizeof(*self)); + + if (!self) + goto out_failure; + + self->name = calloc(1, strlen(name) + 1); + if (!self->name) + goto out_failure; + + self->path = calloc(1, strlen(path) + 1); + if (!self->path) + goto out_failure; + + strcpy(self->name, name); + strcpy(self->path, path); + self->hash = crc32(self->name, strlen(name)); + + return self; + +out_failure: + if (self) { + if (self->name) + free(self->name); + if (self->path) + free(self->path); + free(self); + } + + return NULL; +} + +static int mod_dso__load_sections(struct module *mod) +{ + int count = 0, path_len; + struct dirent *entry; + char *line = NULL; + char *dir_path; + DIR *dir; + size_t n; + + path_len = strlen("/sys/module/"); + path_len += strlen(mod->name); + path_len += strlen("/sections/"); + + dir_path = calloc(1, path_len + 1); + if (dir_path == NULL) + goto out_failure; + + strcat(dir_path, "/sys/module/"); + strcat(dir_path, mod->name); + strcat(dir_path, "/sections/"); + + dir = opendir(dir_path); + if (dir == NULL) + goto out_free; + + while ((entry = readdir(dir))) { + struct section *section; + char *path, *vma; + int line_len; + FILE *file; + + if (!strcmp(".", entry->d_name) || !strcmp("..", entry->d_name)) + continue; + + path = calloc(1, path_len + strlen(entry->d_name) + 1); + if (path == NULL) + break; + strcat(path, dir_path); + strcat(path, entry->d_name); + + file = fopen(path, "r"); + if (file == NULL) { + free(path); + break; + } + + line_len = getline(&line, &n, file); + if (line_len < 0) { + free(path); + fclose(file); + break; + } + + if (!line) { + free(path); + fclose(file); + break; + } + + line[--line_len] = '\0'; /* \n */ + + vma = strstr(line, "0x"); + if (!vma) { + free(path); + fclose(file); + break; + } + vma += 2; + + section = section__new(entry->d_name, path); + if (!section) { + fprintf(stderr, "load_sections: allocation error\n"); + free(path); + fclose(file); + break; + } + + hex2u64(vma, §ion->vma); + sec_dso__insert_section(mod->sections, section); + + free(path); + fclose(file); + count++; + } + + closedir(dir); + free(line); + free(dir_path); + + return count; + +out_free: + free(dir_path); + +out_failure: + return count; +} + +static int mod_dso__load_module_paths(struct mod_dso *self) +{ + struct utsname uts; + int count = 0, len; + char *line = NULL; + FILE *file; + char *path; + size_t n; + + if (uname(&uts) < 0) + goto out_failure; + + len = strlen("/lib/modules/"); + len += strlen(uts.release); + len += strlen("/modules.dep"); + + path = calloc(1, len); + if (path == NULL) + goto out_failure; + + strcat(path, "/lib/modules/"); + strcat(path, uts.release); + strcat(path, "/modules.dep"); + + file = fopen(path, "r"); + free(path); + if (file == NULL) + goto out_failure; + + while (!feof(file)) { + char *name, *tmp; + struct module *module; + int line_len; + + line_len = getline(&line, &n, file); + if (line_len < 0) + break; + + if (!line) + goto out_failure; + + line[--line_len] = '\0'; /* \n */ + + path = strtok(line, ":"); + if (!path) + goto out_failure; + + name = strdup(path); + name = strtok(name, "/"); + + tmp = name; + + while (tmp) { + tmp = strtok(NULL, "/"); + if (tmp) + name = tmp; + } + name = strsep(&name, "."); + + /* Quirk: replace '-' with '_' in sound modules */ + for (len = strlen(name); len; len--) { + if (*(name+len) == '-') + *(name+len) = '_'; + } + + module = module__new(name, path); + if (!module) { + fprintf(stderr, "load_module_paths: allocation error\n"); + goto out_failure; + } + mod_dso__insert_module(self, module); + + module->sections = sec_dso__new_dso("sections"); + if (!module->sections) { + fprintf(stderr, "load_module_paths: allocation error\n"); + goto out_failure; + } + + module->active = mod_dso__load_sections(module); + + if (module->active > 0) + count++; + } + + free(line); + fclose(file); + + return count; + +out_failure: + return -1; +} + +int mod_dso__load_modules(struct mod_dso *dso) +{ + int err; + + err = mod_dso__load_module_paths(dso); + + return err; +} diff --git a/tools/perf/util/module.h b/tools/perf/util/module.h new file mode 100644 index 00000000000..8a592ef641c --- /dev/null +++ b/tools/perf/util/module.h @@ -0,0 +1,53 @@ +#ifndef _PERF_MODULE_ +#define _PERF_MODULE_ 1 + +#include <linux/types.h> +#include "../types.h" +#include <linux/list.h> +#include <linux/rbtree.h> + +struct section { + struct rb_node rb_node; + u64 hash; + u64 vma; + char *name; + char *path; +}; + +struct sec_dso { + struct list_head node; + struct rb_root secs; + struct section *(*find_section)(struct sec_dso *, const char *name); + char name[0]; +}; + +struct module { + struct rb_node rb_node; + u64 hash; + char *name; + char *path; + struct sec_dso *sections; + int active; +}; + +struct mod_dso { + struct list_head node; + struct rb_root mods; + struct module *(*find_module)(struct mod_dso *, const char *name); + char name[0]; +}; + +struct sec_dso *sec_dso__new_dso(const char *name); +void sec_dso__delete_sections(struct sec_dso *self); +void sec_dso__delete_self(struct sec_dso *self); +size_t sec_dso__fprintf(struct sec_dso *self, FILE *fp); +struct section *sec_dso__find_section(struct sec_dso *self, const char *name); + +struct mod_dso *mod_dso__new_dso(const char *name); +void mod_dso__delete_modules(struct mod_dso *self); +void mod_dso__delete_self(struct mod_dso *self); +size_t mod_dso__fprintf(struct mod_dso *self, FILE *fp); +struct module *mod_dso__find_module(struct mod_dso *self, const char *name); +int mod_dso__load_modules(struct mod_dso *dso); + +#endif /* _PERF_MODULE_ */ diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 4d042f104cd..13ab4b842d4 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -1,24 +1,32 @@ -#include "../perf.h" #include "util.h" +#include "../perf.h" #include "parse-options.h" #include "parse-events.h" #include "exec_cmd.h" #include "string.h" - -extern char *strcasestr(const char *haystack, const char *needle); +#include "cache.h" +#include "header.h" int nr_counters; -struct perf_counter_attr attrs[MAX_COUNTERS]; +struct perf_event_attr attrs[MAX_COUNTERS]; struct event_symbol { - u8 type; - u64 config; - char *symbol; - char *alias; + u8 type; + u64 config; + const char *symbol; + const char *alias; +}; + +enum event_result { + EVT_FAILED, + EVT_HANDLED, + EVT_HANDLED_ALL }; +char debugfs_path[MAXPATHLEN]; + #define CHW(x) .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_##x #define CSW(x) .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_##x @@ -40,15 +48,15 @@ static struct event_symbol event_symbols[] = { { CSW(CPU_MIGRATIONS), "cpu-migrations", "migrations" }, }; -#define __PERF_COUNTER_FIELD(config, name) \ - ((config & PERF_COUNTER_##name##_MASK) >> PERF_COUNTER_##name##_SHIFT) +#define __PERF_EVENT_FIELD(config, name) \ + ((config & PERF_EVENT_##name##_MASK) >> PERF_EVENT_##name##_SHIFT) -#define PERF_COUNTER_RAW(config) __PERF_COUNTER_FIELD(config, RAW) -#define PERF_COUNTER_CONFIG(config) __PERF_COUNTER_FIELD(config, CONFIG) -#define PERF_COUNTER_TYPE(config) __PERF_COUNTER_FIELD(config, TYPE) -#define PERF_COUNTER_ID(config) __PERF_COUNTER_FIELD(config, EVENT) +#define PERF_EVENT_RAW(config) __PERF_EVENT_FIELD(config, RAW) +#define PERF_EVENT_CONFIG(config) __PERF_EVENT_FIELD(config, CONFIG) +#define PERF_EVENT_TYPE(config) __PERF_EVENT_FIELD(config, TYPE) +#define PERF_EVENT_ID(config) __PERF_EVENT_FIELD(config, EVENT) -static char *hw_event_names[] = { +static const char *hw_event_names[] = { "cycles", "instructions", "cache-references", @@ -58,7 +66,7 @@ static char *hw_event_names[] = { "bus-cycles", }; -static char *sw_event_names[] = { +static const char *sw_event_names[] = { "cpu-clock-msecs", "task-clock-msecs", "page-faults", @@ -70,22 +78,22 @@ static char *sw_event_names[] = { #define MAX_ALIASES 8 -static char *hw_cache[][MAX_ALIASES] = { - { "L1-d$", "l1-d", "l1d", "L1-data", }, - { "L1-i$", "l1-i", "l1i", "L1-instruction", }, +static const char *hw_cache[][MAX_ALIASES] = { + { "L1-dcache", "l1-d", "l1d", "L1-data", }, + { "L1-icache", "l1-i", "l1i", "L1-instruction", }, { "LLC", "L2" }, { "dTLB", "d-tlb", "Data-TLB", }, { "iTLB", "i-tlb", "Instruction-TLB", }, { "branch", "branches", "bpu", "btb", "bpc", }, }; -static char *hw_cache_op[][MAX_ALIASES] = { +static const char *hw_cache_op[][MAX_ALIASES] = { { "load", "loads", "read", }, { "store", "stores", "write", }, { "prefetch", "prefetches", "speculative-read", "speculative-load", }, }; -static char *hw_cache_result[][MAX_ALIASES] = { +static const char *hw_cache_result[][MAX_ALIASES] = { { "refs", "Reference", "ops", "access", }, { "misses", "miss", }, }; @@ -110,6 +118,136 @@ static unsigned long hw_cache_stat[C(MAX)] = { [C(BPU)] = (CACHE_READ), }; +#define for_each_subsystem(sys_dir, sys_dirent, sys_next) \ + while (!readdir_r(sys_dir, &sys_dirent, &sys_next) && sys_next) \ + if (sys_dirent.d_type == DT_DIR && \ + (strcmp(sys_dirent.d_name, ".")) && \ + (strcmp(sys_dirent.d_name, ".."))) + +static int tp_event_has_id(struct dirent *sys_dir, struct dirent *evt_dir) +{ + char evt_path[MAXPATHLEN]; + int fd; + + snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", debugfs_path, + sys_dir->d_name, evt_dir->d_name); + fd = open(evt_path, O_RDONLY); + if (fd < 0) + return -EINVAL; + close(fd); + + return 0; +} + +#define for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next) \ + while (!readdir_r(evt_dir, &evt_dirent, &evt_next) && evt_next) \ + if (evt_dirent.d_type == DT_DIR && \ + (strcmp(evt_dirent.d_name, ".")) && \ + (strcmp(evt_dirent.d_name, "..")) && \ + (!tp_event_has_id(&sys_dirent, &evt_dirent))) + +#define MAX_EVENT_LENGTH 512 + +int valid_debugfs_mount(const char *debugfs) +{ + struct statfs st_fs; + + if (statfs(debugfs, &st_fs) < 0) + return -ENOENT; + else if (st_fs.f_type != (long) DEBUGFS_MAGIC) + return -ENOENT; + return 0; +} + +struct tracepoint_path *tracepoint_id_to_path(u64 config) +{ + struct tracepoint_path *path = NULL; + DIR *sys_dir, *evt_dir; + struct dirent *sys_next, *evt_next, sys_dirent, evt_dirent; + char id_buf[4]; + int sys_dir_fd, fd; + u64 id; + char evt_path[MAXPATHLEN]; + + if (valid_debugfs_mount(debugfs_path)) + return NULL; + + sys_dir = opendir(debugfs_path); + if (!sys_dir) + goto cleanup; + sys_dir_fd = dirfd(sys_dir); + + for_each_subsystem(sys_dir, sys_dirent, sys_next) { + int dfd = openat(sys_dir_fd, sys_dirent.d_name, + O_RDONLY|O_DIRECTORY), evt_dir_fd; + if (dfd == -1) + continue; + evt_dir = fdopendir(dfd); + if (!evt_dir) { + close(dfd); + continue; + } + evt_dir_fd = dirfd(evt_dir); + for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next) { + snprintf(evt_path, MAXPATHLEN, "%s/id", + evt_dirent.d_name); + fd = openat(evt_dir_fd, evt_path, O_RDONLY); + if (fd < 0) + continue; + if (read(fd, id_buf, sizeof(id_buf)) < 0) { + close(fd); + continue; + } + close(fd); + id = atoll(id_buf); + if (id == config) { + closedir(evt_dir); + closedir(sys_dir); + path = calloc(1, sizeof(path)); + path->system = malloc(MAX_EVENT_LENGTH); + if (!path->system) { + free(path); + return NULL; + } + path->name = malloc(MAX_EVENT_LENGTH); + if (!path->name) { + free(path->system); + free(path); + return NULL; + } + strncpy(path->system, sys_dirent.d_name, + MAX_EVENT_LENGTH); + strncpy(path->name, evt_dirent.d_name, + MAX_EVENT_LENGTH); + return path; + } + } + closedir(evt_dir); + } + +cleanup: + closedir(sys_dir); + return NULL; +} + +#define TP_PATH_LEN (MAX_EVENT_LENGTH * 2 + 1) +static const char *tracepoint_id_to_name(u64 config) +{ + static char buf[TP_PATH_LEN]; + struct tracepoint_path *path; + + path = tracepoint_id_to_path(config); + if (path) { + snprintf(buf, TP_PATH_LEN, "%s:%s", path->system, path->name); + free(path->name); + free(path->system); + free(path); + } else + snprintf(buf, TP_PATH_LEN, "%s:%s", "unknown", "unknown"); + + return buf; +} + static int is_cache_op_valid(u8 cache_type, u8 cache_op) { if (hw_cache_stat[cache_type] & COP(cache_op)) @@ -134,13 +272,19 @@ static char *event_cache_name(u8 cache_type, u8 cache_op, u8 cache_result) return name; } -char *event_name(int counter) +const char *event_name(int counter) { u64 config = attrs[counter].config; int type = attrs[counter].type; + + return __event_name(type, config); +} + +const char *__event_name(int type, u64 config) +{ static char buf[32]; - if (attrs[counter].type == PERF_TYPE_RAW) { + if (type == PERF_TYPE_RAW) { sprintf(buf, "raw 0x%llx", config); return buf; } @@ -177,6 +321,9 @@ char *event_name(int counter) return sw_event_names[config]; return "unknown-software"; + case PERF_TYPE_TRACEPOINT: + return tracepoint_id_to_name(config); + default: break; } @@ -184,47 +331,74 @@ char *event_name(int counter) return "unknown"; } -static int parse_aliases(const char *str, char *names[][MAX_ALIASES], int size) +static int parse_aliases(const char **str, const char *names[][MAX_ALIASES], int size) { int i, j; + int n, longest = -1; for (i = 0; i < size; i++) { - for (j = 0; j < MAX_ALIASES; j++) { - if (!names[i][j]) - break; - if (strcasestr(str, names[i][j])) - return i; + for (j = 0; j < MAX_ALIASES && names[i][j]; j++) { + n = strlen(names[i][j]); + if (n > longest && !strncasecmp(*str, names[i][j], n)) + longest = n; + } + if (longest > 0) { + *str += longest; + return i; } } return -1; } -static int -parse_generic_hw_symbols(const char *str, struct perf_counter_attr *attr) +static enum event_result +parse_generic_hw_event(const char **str, struct perf_event_attr *attr) { - int cache_type = -1, cache_op = 0, cache_result = 0; + const char *s = *str; + int cache_type = -1, cache_op = -1, cache_result = -1; - cache_type = parse_aliases(str, hw_cache, PERF_COUNT_HW_CACHE_MAX); + cache_type = parse_aliases(&s, hw_cache, PERF_COUNT_HW_CACHE_MAX); /* * No fallback - if we cannot get a clear cache type * then bail out: */ if (cache_type == -1) - return -EINVAL; + return EVT_FAILED; + + while ((cache_op == -1 || cache_result == -1) && *s == '-') { + ++s; + + if (cache_op == -1) { + cache_op = parse_aliases(&s, hw_cache_op, + PERF_COUNT_HW_CACHE_OP_MAX); + if (cache_op >= 0) { + if (!is_cache_op_valid(cache_type, cache_op)) + return 0; + continue; + } + } + + if (cache_result == -1) { + cache_result = parse_aliases(&s, hw_cache_result, + PERF_COUNT_HW_CACHE_RESULT_MAX); + if (cache_result >= 0) + continue; + } + + /* + * Can't parse this as a cache op or result, so back up + * to the '-'. + */ + --s; + break; + } - cache_op = parse_aliases(str, hw_cache_op, PERF_COUNT_HW_CACHE_OP_MAX); /* * Fall back to reads: */ if (cache_op == -1) cache_op = PERF_COUNT_HW_CACHE_OP_READ; - if (!is_cache_op_valid(cache_type, cache_op)) - return -EINVAL; - - cache_result = parse_aliases(str, hw_cache_result, - PERF_COUNT_HW_CACHE_RESULT_MAX); /* * Fall back to accesses: */ @@ -234,93 +408,338 @@ parse_generic_hw_symbols(const char *str, struct perf_counter_attr *attr) attr->config = cache_type | (cache_op << 8) | (cache_result << 16); attr->type = PERF_TYPE_HW_CACHE; - return 0; + *str = s; + return EVT_HANDLED; } -static int check_events(const char *str, unsigned int i) +static enum event_result +parse_single_tracepoint_event(char *sys_name, + const char *evt_name, + unsigned int evt_length, + char *flags, + struct perf_event_attr *attr, + const char **strp) { - if (!strncmp(str, event_symbols[i].symbol, - strlen(event_symbols[i].symbol))) - return 1; + char evt_path[MAXPATHLEN]; + char id_buf[4]; + u64 id; + int fd; + + if (flags) { + if (!strncmp(flags, "record", strlen(flags))) { + attr->sample_type |= PERF_SAMPLE_RAW; + attr->sample_type |= PERF_SAMPLE_TIME; + attr->sample_type |= PERF_SAMPLE_CPU; + } + } - if (strlen(event_symbols[i].alias)) - if (!strncmp(str, event_symbols[i].alias, - strlen(event_symbols[i].alias))) - return 1; - return 0; + snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", debugfs_path, + sys_name, evt_name); + + fd = open(evt_path, O_RDONLY); + if (fd < 0) + return EVT_FAILED; + + if (read(fd, id_buf, sizeof(id_buf)) < 0) { + close(fd); + return EVT_FAILED; + } + + close(fd); + id = atoll(id_buf); + attr->config = id; + attr->type = PERF_TYPE_TRACEPOINT; + *strp = evt_name + evt_length; + + return EVT_HANDLED; } -/* - * Each event can have multiple symbolic names. - * Symbolic names are (almost) exactly matched. - */ -static int parse_event_symbols(const char *str, struct perf_counter_attr *attr) +/* sys + ':' + event + ':' + flags*/ +#define MAX_EVOPT_LEN (MAX_EVENT_LENGTH * 2 + 2 + 128) +static enum event_result +parse_subsystem_tracepoint_event(char *sys_name, char *flags) { - u64 config, id; - int type; - unsigned int i; - const char *sep, *pstr; + char evt_path[MAXPATHLEN]; + struct dirent *evt_ent; + DIR *evt_dir; - if (str[0] == 'r' && hex2u64(str + 1, &config) > 0) { - attr->type = PERF_TYPE_RAW; - attr->config = config; + snprintf(evt_path, MAXPATHLEN, "%s/%s", debugfs_path, sys_name); + evt_dir = opendir(evt_path); - return 0; + if (!evt_dir) { + perror("Can't open event dir"); + return EVT_FAILED; } - pstr = str; - sep = strchr(pstr, ':'); - if (sep) { - type = atoi(pstr); - pstr = sep + 1; - id = atoi(pstr); - sep = strchr(pstr, ':'); - if (sep) { - pstr = sep + 1; - if (strchr(pstr, 'k')) - attr->exclude_user = 1; - if (strchr(pstr, 'u')) - attr->exclude_kernel = 1; + while ((evt_ent = readdir(evt_dir))) { + char event_opt[MAX_EVOPT_LEN + 1]; + int len; + unsigned int rem = MAX_EVOPT_LEN; + + if (!strcmp(evt_ent->d_name, ".") + || !strcmp(evt_ent->d_name, "..") + || !strcmp(evt_ent->d_name, "enable") + || !strcmp(evt_ent->d_name, "filter")) + continue; + + len = snprintf(event_opt, MAX_EVOPT_LEN, "%s:%s", sys_name, + evt_ent->d_name); + if (len < 0) + return EVT_FAILED; + + rem -= len; + if (flags) { + if (rem < strlen(flags) + 1) + return EVT_FAILED; + + strcat(event_opt, ":"); + strcat(event_opt, flags); } - attr->type = type; - attr->config = id; + if (parse_events(NULL, event_opt, 0)) + return EVT_FAILED; + } + + return EVT_HANDLED_ALL; +} + + +static enum event_result parse_tracepoint_event(const char **strp, + struct perf_event_attr *attr) +{ + const char *evt_name; + char *flags; + char sys_name[MAX_EVENT_LENGTH]; + unsigned int sys_length, evt_length; + + if (valid_debugfs_mount(debugfs_path)) return 0; + + evt_name = strchr(*strp, ':'); + if (!evt_name) + return EVT_FAILED; + + sys_length = evt_name - *strp; + if (sys_length >= MAX_EVENT_LENGTH) + return 0; + + strncpy(sys_name, *strp, sys_length); + sys_name[sys_length] = '\0'; + evt_name = evt_name + 1; + + flags = strchr(evt_name, ':'); + if (flags) { + /* split it out: */ + evt_name = strndup(evt_name, flags - evt_name); + flags++; } + evt_length = strlen(evt_name); + if (evt_length >= MAX_EVENT_LENGTH) + return EVT_FAILED; + + if (!strcmp(evt_name, "*")) { + *strp = evt_name + evt_length; + return parse_subsystem_tracepoint_event(sys_name, flags); + } else + return parse_single_tracepoint_event(sys_name, evt_name, + evt_length, flags, + attr, strp); +} + +static int check_events(const char *str, unsigned int i) +{ + int n; + + n = strlen(event_symbols[i].symbol); + if (!strncmp(str, event_symbols[i].symbol, n)) + return n; + + n = strlen(event_symbols[i].alias); + if (n) + if (!strncmp(str, event_symbols[i].alias, n)) + return n; + return 0; +} + +static enum event_result +parse_symbolic_event(const char **strp, struct perf_event_attr *attr) +{ + const char *str = *strp; + unsigned int i; + int n; + for (i = 0; i < ARRAY_SIZE(event_symbols); i++) { - if (check_events(str, i)) { + n = check_events(str, i); + if (n > 0) { attr->type = event_symbols[i].type; attr->config = event_symbols[i].config; + *strp = str + n; + return EVT_HANDLED; + } + } + return EVT_FAILED; +} - return 0; +static enum event_result +parse_raw_event(const char **strp, struct perf_event_attr *attr) +{ + const char *str = *strp; + u64 config; + int n; + + if (*str != 'r') + return EVT_FAILED; + n = hex2u64(str + 1, &config); + if (n > 0) { + *strp = str + n + 1; + attr->type = PERF_TYPE_RAW; + attr->config = config; + return EVT_HANDLED; + } + return EVT_FAILED; +} + +static enum event_result +parse_numeric_event(const char **strp, struct perf_event_attr *attr) +{ + const char *str = *strp; + char *endp; + unsigned long type; + u64 config; + + type = strtoul(str, &endp, 0); + if (endp > str && type < PERF_TYPE_MAX && *endp == ':') { + str = endp + 1; + config = strtoul(str, &endp, 0); + if (endp > str) { + attr->type = type; + attr->config = config; + *strp = endp; + return EVT_HANDLED; } } + return EVT_FAILED; +} + +static enum event_result +parse_event_modifier(const char **strp, struct perf_event_attr *attr) +{ + const char *str = *strp; + int eu = 1, ek = 1, eh = 1; + + if (*str++ != ':') + return 0; + while (*str) { + if (*str == 'u') + eu = 0; + else if (*str == 'k') + ek = 0; + else if (*str == 'h') + eh = 0; + else + break; + ++str; + } + if (str >= *strp + 2) { + *strp = str; + attr->exclude_user = eu; + attr->exclude_kernel = ek; + attr->exclude_hv = eh; + return 1; + } + return 0; +} + +/* + * Each event can have multiple symbolic names. + * Symbolic names are (almost) exactly matched. + */ +static enum event_result +parse_event_symbols(const char **str, struct perf_event_attr *attr) +{ + enum event_result ret; + + ret = parse_tracepoint_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + ret = parse_raw_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; - return parse_generic_hw_symbols(str, attr); + ret = parse_numeric_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + ret = parse_symbolic_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + ret = parse_generic_hw_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + return EVT_FAILED; + +modifier: + parse_event_modifier(str, attr); + + return ret; } -int parse_events(const struct option *opt, const char *str, int unset) +static void store_event_type(const char *orgname) +{ + char filename[PATH_MAX], *c; + FILE *file; + int id; + + sprintf(filename, "/sys/kernel/debug/tracing/events/%s/id", orgname); + c = strchr(filename, ':'); + if (c) + *c = '/'; + + file = fopen(filename, "r"); + if (!file) + return; + if (fscanf(file, "%i", &id) < 1) + die("cannot store event ID"); + fclose(file); + perf_header__push_event(id, orgname); +} + + +int parse_events(const struct option *opt __used, const char *str, int unset __used) { - struct perf_counter_attr attr; - int ret; + struct perf_event_attr attr; + enum event_result ret; - memset(&attr, 0, sizeof(attr)); -again: - if (nr_counters == MAX_COUNTERS) - return -1; + if (strchr(str, ':')) + store_event_type(str); - ret = parse_event_symbols(str, &attr); - if (ret < 0) - return ret; + for (;;) { + if (nr_counters == MAX_COUNTERS) + return -1; - attrs[nr_counters] = attr; - nr_counters++; + memset(&attr, 0, sizeof(attr)); + ret = parse_event_symbols(&str, &attr); + if (ret == EVT_FAILED) + return -1; - str = strstr(str, ","); - if (str) { - str++; - goto again; + if (!(*str == 0 || *str == ',' || isspace(*str))) + return -1; + + if (ret != EVT_HANDLED_ALL) { + attrs[nr_counters] = attr; + nr_counters++; + } + + if (*str == 0) + break; + if (*str == ',') + ++str; + while (isspace(*str)) + ++str; } return 0; @@ -335,12 +754,55 @@ static const char * const event_type_descriptors[] = { }; /* + * Print the events from <debugfs_mount_point>/tracing/events + */ + +static void print_tracepoint_events(void) +{ + DIR *sys_dir, *evt_dir; + struct dirent *sys_next, *evt_next, sys_dirent, evt_dirent; + int sys_dir_fd; + char evt_path[MAXPATHLEN]; + + if (valid_debugfs_mount(debugfs_path)) + return; + + sys_dir = opendir(debugfs_path); + if (!sys_dir) + goto cleanup; + sys_dir_fd = dirfd(sys_dir); + + for_each_subsystem(sys_dir, sys_dirent, sys_next) { + int dfd = openat(sys_dir_fd, sys_dirent.d_name, + O_RDONLY|O_DIRECTORY), evt_dir_fd; + if (dfd == -1) + continue; + evt_dir = fdopendir(dfd); + if (!evt_dir) { + close(dfd); + continue; + } + evt_dir_fd = dirfd(evt_dir); + for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next) { + snprintf(evt_path, MAXPATHLEN, "%s:%s", + sys_dirent.d_name, evt_dirent.d_name); + fprintf(stderr, " %-42s [%s]\n", evt_path, + event_type_descriptors[PERF_TYPE_TRACEPOINT+1]); + } + closedir(evt_dir); + } + +cleanup: + closedir(sys_dir); +} + +/* * Print the help text for the event symbols: */ void print_events(void) { struct event_symbol *syms = event_symbols; - unsigned int i, type, prev_type = -1; + unsigned int i, type, op, prev_type = -1; char name[40]; fprintf(stderr, "\n"); @@ -348,7 +810,7 @@ void print_events(void) for (i = 0; i < ARRAY_SIZE(event_symbols); i++, syms++) { type = syms->type + 1; - if (type > ARRAY_SIZE(event_type_descriptors)) + if (type >= ARRAY_SIZE(event_type_descriptors)) type = 0; if (type != prev_type) @@ -358,16 +820,33 @@ void print_events(void) sprintf(name, "%s OR %s", syms->symbol, syms->alias); else strcpy(name, syms->symbol); - fprintf(stderr, " %-40s [%s]\n", name, + fprintf(stderr, " %-42s [%s]\n", name, event_type_descriptors[type]); prev_type = type; } fprintf(stderr, "\n"); - fprintf(stderr, " %-40s [raw hardware event descriptor]\n", + for (type = 0; type < PERF_COUNT_HW_CACHE_MAX; type++) { + for (op = 0; op < PERF_COUNT_HW_CACHE_OP_MAX; op++) { + /* skip invalid cache type */ + if (!is_cache_op_valid(type, op)) + continue; + + for (i = 0; i < PERF_COUNT_HW_CACHE_RESULT_MAX; i++) { + fprintf(stderr, " %-42s [%s]\n", + event_cache_name(type, op, i), + event_type_descriptors[4]); + } + } + } + + fprintf(stderr, "\n"); + fprintf(stderr, " %-42s [raw hardware event descriptor]\n", "rNNN"); fprintf(stderr, "\n"); + print_tracepoint_events(); + exit(129); } diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h index e3d552908e6..30c60811284 100644 --- a/tools/perf/util/parse-events.h +++ b/tools/perf/util/parse-events.h @@ -1,13 +1,25 @@ - +#ifndef _PARSE_EVENTS_H +#define _PARSE_EVENTS_H /* * Parse symbolic events/counts passed in as options: */ +struct option; + +struct tracepoint_path { + char *system; + char *name; + struct tracepoint_path *next; +}; + +extern struct tracepoint_path *tracepoint_id_to_path(u64 config); + extern int nr_counters; -extern struct perf_counter_attr attrs[MAX_COUNTERS]; +extern struct perf_event_attr attrs[MAX_COUNTERS]; -extern char *event_name(int ctr); +extern const char *event_name(int ctr); +extern const char *__event_name(int type, u64 config); extern int parse_events(const struct option *opt, const char *str, int unset); @@ -15,3 +27,8 @@ extern int parse_events(const struct option *opt, const char *str, int unset); extern void print_events(void); +extern char debugfs_path[]; +extern int valid_debugfs_mount(const char *debugfs); + + +#endif /* _PARSE_EVENTS_H */ diff --git a/tools/perf/util/parse-options.c b/tools/perf/util/parse-options.c index b3affb1658d..6d8af48c925 100644 --- a/tools/perf/util/parse-options.c +++ b/tools/perf/util/parse-options.c @@ -20,7 +20,8 @@ static int get_arg(struct parse_opt_ctx_t *p, const struct option *opt, if (p->opt) { *arg = p->opt; p->opt = NULL; - } else if (p->argc == 1 && (opt->flags & PARSE_OPT_LASTARG_DEFAULT)) { + } else if ((opt->flags & PARSE_OPT_LASTARG_DEFAULT) && (p->argc == 1 || + **(p->argv + 1) == '-')) { *arg = (const char *)opt->defval; } else if (p->argc > 1) { p->argc--; @@ -52,6 +53,12 @@ static int get_value(struct parse_opt_ctx_t *p, case OPTION_SET_INT: case OPTION_SET_PTR: return opterror(opt, "takes no value", flags); + case OPTION_END: + case OPTION_ARGUMENT: + case OPTION_GROUP: + case OPTION_STRING: + case OPTION_INTEGER: + case OPTION_LONG: default: break; } @@ -129,6 +136,9 @@ static int get_value(struct parse_opt_ctx_t *p, return opterror(opt, "expects a numerical value", flags); return 0; + case OPTION_END: + case OPTION_ARGUMENT: + case OPTION_GROUP: default: die("should not happen, someone must be hit on the forehead"); } @@ -295,6 +305,8 @@ int parse_options_step(struct parse_opt_ctx_t *ctx, return parse_options_usage(usagestr, options); case -2: goto unknown; + default: + break; } if (ctx->opt) check_typos(arg + 1, options); @@ -313,6 +325,8 @@ int parse_options_step(struct parse_opt_ctx_t *ctx, ctx->argv[0] = strdup(ctx->opt - 1); *(char *)ctx->argv[0] = '-'; goto unknown; + default: + break; } } continue; @@ -335,6 +349,8 @@ int parse_options_step(struct parse_opt_ctx_t *ctx, return parse_options_usage(usagestr, options); case -2: goto unknown; + default: + break; } continue; unknown: @@ -455,6 +471,13 @@ int usage_with_options_internal(const char * const *usagestr, } break; default: /* OPTION_{BIT,BOOLEAN,SET_INT,SET_PTR} */ + case OPTION_END: + case OPTION_GROUP: + case OPTION_BIT: + case OPTION_BOOLEAN: + case OPTION_SET_INT: + case OPTION_SET_PTR: + case OPTION_LONG: break; } @@ -485,7 +508,7 @@ int parse_options_usage(const char * const *usagestr, } -int parse_opt_verbosity_cb(const struct option *opt, const char *arg, +int parse_opt_verbosity_cb(const struct option *opt, const char *arg __used, int unset) { int *target = opt->value; diff --git a/tools/perf/util/parse-options.h b/tools/perf/util/parse-options.h index a1039a6ce0e..2ee248ff27e 100644 --- a/tools/perf/util/parse-options.h +++ b/tools/perf/util/parse-options.h @@ -90,21 +90,24 @@ struct option { intptr_t defval; }; -#define OPT_END() { OPTION_END } -#define OPT_ARGUMENT(l, h) { OPTION_ARGUMENT, 0, (l), NULL, NULL, (h) } -#define OPT_GROUP(h) { OPTION_GROUP, 0, NULL, NULL, NULL, (h) } -#define OPT_BIT(s, l, v, h, b) { OPTION_BIT, (s), (l), (v), NULL, (h), 0, NULL, (b) } -#define OPT_BOOLEAN(s, l, v, h) { OPTION_BOOLEAN, (s), (l), (v), NULL, (h) } -#define OPT_SET_INT(s, l, v, h, i) { OPTION_SET_INT, (s), (l), (v), NULL, (h), 0, NULL, (i) } -#define OPT_SET_PTR(s, l, v, h, p) { OPTION_SET_PTR, (s), (l), (v), NULL, (h), 0, NULL, (p) } -#define OPT_INTEGER(s, l, v, h) { OPTION_INTEGER, (s), (l), (v), NULL, (h) } -#define OPT_LONG(s, l, v, h) { OPTION_LONG, (s), (l), (v), NULL, (h) } -#define OPT_STRING(s, l, v, a, h) { OPTION_STRING, (s), (l), (v), (a), (h) } +#define OPT_END() { .type = OPTION_END } +#define OPT_ARGUMENT(l, h) { .type = OPTION_ARGUMENT, .long_name = (l), .help = (h) } +#define OPT_GROUP(h) { .type = OPTION_GROUP, .help = (h) } +#define OPT_BIT(s, l, v, h, b) { .type = OPTION_BIT, .short_name = (s), .long_name = (l), .value = (v), .help = (h), .defval = (b) } +#define OPT_BOOLEAN(s, l, v, h) { .type = OPTION_BOOLEAN, .short_name = (s), .long_name = (l), .value = (v), .help = (h) } +#define OPT_SET_INT(s, l, v, h, i) { .type = OPTION_SET_INT, .short_name = (s), .long_name = (l), .value = (v), .help = (h), .defval = (i) } +#define OPT_SET_PTR(s, l, v, h, p) { .type = OPTION_SET_PTR, .short_name = (s), .long_name = (l), .value = (v), .help = (h), .defval = (p) } +#define OPT_INTEGER(s, l, v, h) { .type = OPTION_INTEGER, .short_name = (s), .long_name = (l), .value = (v), .help = (h) } +#define OPT_LONG(s, l, v, h) { .type = OPTION_LONG, .short_name = (s), .long_name = (l), .value = (v), .help = (h) } +#define OPT_STRING(s, l, v, a, h) { .type = OPTION_STRING, .short_name = (s), .long_name = (l), .value = (v), (a), .help = (h) } #define OPT_DATE(s, l, v, h) \ - { OPTION_CALLBACK, (s), (l), (v), "time",(h), 0, \ - parse_opt_approxidate_cb } + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l), .value = (v), .argh = "time", .help = (h), .callback = parse_opt_approxidate_cb } #define OPT_CALLBACK(s, l, v, a, h, f) \ - { OPTION_CALLBACK, (s), (l), (v), (a), (h), 0, (f) } + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l), .value = (v), (a), .help = (h), .callback = (f) } +#define OPT_CALLBACK_NOOPT(s, l, v, a, h, f) \ + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l), .value = (v), (a), .help = (h), .callback = (f), .flags = PARSE_OPT_NOARG } +#define OPT_CALLBACK_DEFAULT(s, l, v, a, h, f, d) \ + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l), .value = (v), (a), .help = (h), .callback = (f), .defval = (intptr_t)d, .flags = PARSE_OPT_LASTARG_DEFAULT } /* parse_options() will filter out the processed options and leave the * non-option argments in argv[]. diff --git a/tools/perf/util/path.c b/tools/perf/util/path.c index a501a40dd2c..fd1f2faaade 100644 --- a/tools/perf/util/path.c +++ b/tools/perf/util/path.c @@ -17,7 +17,7 @@ static char bad_path[] = "/bad-path/"; * Two hacks: */ -static char *get_perf_dir(void) +static const char *get_perf_dir(void) { return "."; } @@ -38,8 +38,9 @@ size_t strlcpy(char *dest, const char *src, size_t size) static char *get_pathname(void) { static char pathname_array[4][PATH_MAX]; - static int index; - return pathname_array[3 & ++index]; + static int idx; + + return pathname_array[3 & ++idx]; } static char *cleanup_path(char *path) @@ -161,20 +162,24 @@ int perf_mkstemp(char *path, size_t len, const char *template) } -const char *make_relative_path(const char *abs, const char *base) +const char *make_relative_path(const char *abs_path, const char *base) { static char buf[PATH_MAX + 1]; int baselen; + if (!base) - return abs; + return abs_path; + baselen = strlen(base); - if (prefixcmp(abs, base)) - return abs; - if (abs[baselen] == '/') + if (prefixcmp(abs_path, base)) + return abs_path; + if (abs_path[baselen] == '/') baselen++; else if (base[baselen - 1] != '/') - return abs; - strcpy(buf, abs + baselen); + return abs_path; + + strcpy(buf, abs_path + baselen); + return buf; } diff --git a/tools/perf/util/quote.c b/tools/perf/util/quote.c index f18c5212bc9..2726fe40eb5 100644 --- a/tools/perf/util/quote.c +++ b/tools/perf/util/quote.c @@ -162,12 +162,16 @@ static inline int sq_must_quote(char c) return sq_lookup[(unsigned char)c] + quote_path_fully > 0; } -/* returns the longest prefix not needing a quote up to maxlen if positive. - This stops at the first \0 because it's marked as a character needing an - escape */ -static size_t next_quote_pos(const char *s, ssize_t maxlen) +/* + * Returns the longest prefix not needing a quote up to maxlen if + * positive. + * This stops at the first \0 because it's marked as a character + * needing an escape. + */ +static ssize_t next_quote_pos(const char *s, ssize_t maxlen) { - size_t len; + ssize_t len; + if (maxlen < 0) { for (len = 0; !sq_must_quote(s[len]); len++); } else { @@ -192,22 +196,22 @@ static size_t next_quote_pos(const char *s, ssize_t maxlen) static size_t quote_c_style_counted(const char *name, ssize_t maxlen, struct strbuf *sb, FILE *fp, int no_dq) { -#undef EMIT -#define EMIT(c) \ - do { \ - if (sb) strbuf_addch(sb, (c)); \ - if (fp) fputc((c), fp); \ - count++; \ +#define EMIT(c) \ + do { \ + if (sb) strbuf_addch(sb, (c)); \ + if (fp) fputc((c), fp); \ + count++; \ } while (0) -#define EMITBUF(s, l) \ - do { \ - int __ret; \ - if (sb) strbuf_add(sb, (s), (l)); \ - if (fp) __ret = fwrite((s), (l), 1, fp); \ - count += (l); \ + +#define EMITBUF(s, l) \ + do { \ + int __ret; \ + if (sb) strbuf_add(sb, (s), (l)); \ + if (fp) __ret = fwrite((s), (l), 1, fp); \ + count += (l); \ } while (0) - size_t len, count = 0; + ssize_t len, count = 0; const char *p = name; for (;;) { @@ -273,8 +277,8 @@ void write_name_quoted(const char *name, FILE *fp, int terminator) fputc(terminator, fp); } -extern void write_name_quotedpfx(const char *pfx, size_t pfxlen, - const char *name, FILE *fp, int terminator) +void write_name_quotedpfx(const char *pfx, ssize_t pfxlen, + const char *name, FILE *fp, int terminator) { int needquote = 0; @@ -306,7 +310,7 @@ char *quote_path_relative(const char *in, int len, len = strlen(in); /* "../" prefix itself does not need quoting, but "in" might. */ - needquote = next_quote_pos(in, len) < len; + needquote = (next_quote_pos(in, len) < len); strbuf_setlen(out, 0); strbuf_grow(out, len); @@ -314,7 +318,7 @@ char *quote_path_relative(const char *in, int len, strbuf_addch(out, '"'); if (prefix) { int off = 0; - while (prefix[off] && off < len && prefix[off] == in[off]) + while (off < len && prefix[off] && prefix[off] == in[off]) if (prefix[off] == '/') { prefix += off + 1; in += off + 1; diff --git a/tools/perf/util/quote.h b/tools/perf/util/quote.h index 5dfad89816d..a5454a1d1c1 100644 --- a/tools/perf/util/quote.h +++ b/tools/perf/util/quote.h @@ -53,7 +53,7 @@ extern size_t quote_c_style(const char *name, struct strbuf *, FILE *, int no_dq extern void quote_two_c_style(struct strbuf *, const char *, const char *, int); extern void write_name_quoted(const char *name, FILE *, int terminator); -extern void write_name_quotedpfx(const char *pfx, size_t pfxlen, +extern void write_name_quotedpfx(const char *pfx, ssize_t pfxlen, const char *name, FILE *, int terminator); /* quote path as relative to the given prefix */ diff --git a/tools/perf/util/rbtree.c b/tools/perf/util/rbtree.c deleted file mode 100644 index b15ba9c7cb3..00000000000 --- a/tools/perf/util/rbtree.c +++ /dev/null @@ -1,383 +0,0 @@ -/* - Red Black Trees - (C) 1999 Andrea Arcangeli <andrea@suse.de> - (C) 2002 David Woodhouse <dwmw2@infradead.org> - - This program is free software; you can redistribute it and/or modify - it under the terms of the GNU General Public License as published by - the Free Software Foundation; either version 2 of the License, or - (at your option) any later version. - - This program is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - GNU General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program; if not, write to the Free Software - Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA - - linux/lib/rbtree.c -*/ - -#include "rbtree.h" - -static void __rb_rotate_left(struct rb_node *node, struct rb_root *root) -{ - struct rb_node *right = node->rb_right; - struct rb_node *parent = rb_parent(node); - - if ((node->rb_right = right->rb_left)) - rb_set_parent(right->rb_left, node); - right->rb_left = node; - - rb_set_parent(right, parent); - - if (parent) - { - if (node == parent->rb_left) - parent->rb_left = right; - else - parent->rb_right = right; - } - else - root->rb_node = right; - rb_set_parent(node, right); -} - -static void __rb_rotate_right(struct rb_node *node, struct rb_root *root) -{ - struct rb_node *left = node->rb_left; - struct rb_node *parent = rb_parent(node); - - if ((node->rb_left = left->rb_right)) - rb_set_parent(left->rb_right, node); - left->rb_right = node; - - rb_set_parent(left, parent); - - if (parent) - { - if (node == parent->rb_right) - parent->rb_right = left; - else - parent->rb_left = left; - } - else - root->rb_node = left; - rb_set_parent(node, left); -} - -void rb_insert_color(struct rb_node *node, struct rb_root *root) -{ - struct rb_node *parent, *gparent; - - while ((parent = rb_parent(node)) && rb_is_red(parent)) - { - gparent = rb_parent(parent); - - if (parent == gparent->rb_left) - { - { - register struct rb_node *uncle = gparent->rb_right; - if (uncle && rb_is_red(uncle)) - { - rb_set_black(uncle); - rb_set_black(parent); - rb_set_red(gparent); - node = gparent; - continue; - } - } - - if (parent->rb_right == node) - { - register struct rb_node *tmp; - __rb_rotate_left(parent, root); - tmp = parent; - parent = node; - node = tmp; - } - - rb_set_black(parent); - rb_set_red(gparent); - __rb_rotate_right(gparent, root); - } else { - { - register struct rb_node *uncle = gparent->rb_left; - if (uncle && rb_is_red(uncle)) - { - rb_set_black(uncle); - rb_set_black(parent); - rb_set_red(gparent); - node = gparent; - continue; - } - } - - if (parent->rb_left == node) - { - register struct rb_node *tmp; - __rb_rotate_right(parent, root); - tmp = parent; - parent = node; - node = tmp; - } - - rb_set_black(parent); - rb_set_red(gparent); - __rb_rotate_left(gparent, root); - } - } - - rb_set_black(root->rb_node); -} - -static void __rb_erase_color(struct rb_node *node, struct rb_node *parent, - struct rb_root *root) -{ - struct rb_node *other; - - while ((!node || rb_is_black(node)) && node != root->rb_node) - { - if (parent->rb_left == node) - { - other = parent->rb_right; - if (rb_is_red(other)) - { - rb_set_black(other); - rb_set_red(parent); - __rb_rotate_left(parent, root); - other = parent->rb_right; - } - if ((!other->rb_left || rb_is_black(other->rb_left)) && - (!other->rb_right || rb_is_black(other->rb_right))) - { - rb_set_red(other); - node = parent; - parent = rb_parent(node); - } - else - { - if (!other->rb_right || rb_is_black(other->rb_right)) - { - rb_set_black(other->rb_left); - rb_set_red(other); - __rb_rotate_right(other, root); - other = parent->rb_right; - } - rb_set_color(other, rb_color(parent)); - rb_set_black(parent); - rb_set_black(other->rb_right); - __rb_rotate_left(parent, root); - node = root->rb_node; - break; - } - } - else - { - other = parent->rb_left; - if (rb_is_red(other)) - { - rb_set_black(other); - rb_set_red(parent); - __rb_rotate_right(parent, root); - other = parent->rb_left; - } - if ((!other->rb_left || rb_is_black(other->rb_left)) && - (!other->rb_right || rb_is_black(other->rb_right))) - { - rb_set_red(other); - node = parent; - parent = rb_parent(node); - } - else - { - if (!other->rb_left || rb_is_black(other->rb_left)) - { - rb_set_black(other->rb_right); - rb_set_red(other); - __rb_rotate_left(other, root); - other = parent->rb_left; - } - rb_set_color(other, rb_color(parent)); - rb_set_black(parent); - rb_set_black(other->rb_left); - __rb_rotate_right(parent, root); - node = root->rb_node; - break; - } - } - } - if (node) - rb_set_black(node); -} - -void rb_erase(struct rb_node *node, struct rb_root *root) -{ - struct rb_node *child, *parent; - int color; - - if (!node->rb_left) - child = node->rb_right; - else if (!node->rb_right) - child = node->rb_left; - else - { - struct rb_node *old = node, *left; - - node = node->rb_right; - while ((left = node->rb_left) != NULL) - node = left; - child = node->rb_right; - parent = rb_parent(node); - color = rb_color(node); - - if (child) - rb_set_parent(child, parent); - if (parent == old) { - parent->rb_right = child; - parent = node; - } else - parent->rb_left = child; - - node->rb_parent_color = old->rb_parent_color; - node->rb_right = old->rb_right; - node->rb_left = old->rb_left; - - if (rb_parent(old)) - { - if (rb_parent(old)->rb_left == old) - rb_parent(old)->rb_left = node; - else - rb_parent(old)->rb_right = node; - } else - root->rb_node = node; - - rb_set_parent(old->rb_left, node); - if (old->rb_right) - rb_set_parent(old->rb_right, node); - goto color; - } - - parent = rb_parent(node); - color = rb_color(node); - - if (child) - rb_set_parent(child, parent); - if (parent) - { - if (parent->rb_left == node) - parent->rb_left = child; - else - parent->rb_right = child; - } - else - root->rb_node = child; - - color: - if (color == RB_BLACK) - __rb_erase_color(child, parent, root); -} - -/* - * This function returns the first node (in sort order) of the tree. - */ -struct rb_node *rb_first(const struct rb_root *root) -{ - struct rb_node *n; - - n = root->rb_node; - if (!n) - return NULL; - while (n->rb_left) - n = n->rb_left; - return n; -} - -struct rb_node *rb_last(const struct rb_root *root) -{ - struct rb_node *n; - - n = root->rb_node; - if (!n) - return NULL; - while (n->rb_right) - n = n->rb_right; - return n; -} - -struct rb_node *rb_next(const struct rb_node *node) -{ - struct rb_node *parent; - - if (rb_parent(node) == node) - return NULL; - - /* If we have a right-hand child, go down and then left as far - as we can. */ - if (node->rb_right) { - node = node->rb_right; - while (node->rb_left) - node=node->rb_left; - return (struct rb_node *)node; - } - - /* No right-hand children. Everything down and left is - smaller than us, so any 'next' node must be in the general - direction of our parent. Go up the tree; any time the - ancestor is a right-hand child of its parent, keep going - up. First time it's a left-hand child of its parent, said - parent is our 'next' node. */ - while ((parent = rb_parent(node)) && node == parent->rb_right) - node = parent; - - return parent; -} - -struct rb_node *rb_prev(const struct rb_node *node) -{ - struct rb_node *parent; - - if (rb_parent(node) == node) - return NULL; - - /* If we have a left-hand child, go down and then right as far - as we can. */ - if (node->rb_left) { - node = node->rb_left; - while (node->rb_right) - node=node->rb_right; - return (struct rb_node *)node; - } - - /* No left-hand children. Go up till we find an ancestor which - is a right-hand child of its parent */ - while ((parent = rb_parent(node)) && node == parent->rb_left) - node = parent; - - return parent; -} - -void rb_replace_node(struct rb_node *victim, struct rb_node *new, - struct rb_root *root) -{ - struct rb_node *parent = rb_parent(victim); - - /* Set the surrounding nodes to point to the replacement */ - if (parent) { - if (victim == parent->rb_left) - parent->rb_left = new; - else - parent->rb_right = new; - } else { - root->rb_node = new; - } - if (victim->rb_left) - rb_set_parent(victim->rb_left, new); - if (victim->rb_right) - rb_set_parent(victim->rb_right, new); - - /* Copy the pointers/colour from the victim to the replacement */ - *new = *victim; -} diff --git a/tools/perf/util/rbtree.h b/tools/perf/util/rbtree.h deleted file mode 100644 index 6bdc488a47f..00000000000 --- a/tools/perf/util/rbtree.h +++ /dev/null @@ -1,171 +0,0 @@ -/* - Red Black Trees - (C) 1999 Andrea Arcangeli <andrea@suse.de> - - This program is free software; you can redistribute it and/or modify - it under the terms of the GNU General Public License as published by - the Free Software Foundation; either version 2 of the License, or - (at your option) any later version. - - This program is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - GNU General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program; if not, write to the Free Software - Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA - - linux/include/linux/rbtree.h - - To use rbtrees you'll have to implement your own insert and search cores. - This will avoid us to use callbacks and to drop drammatically performances. - I know it's not the cleaner way, but in C (not in C++) to get - performances and genericity... - - Some example of insert and search follows here. The search is a plain - normal search over an ordered tree. The insert instead must be implemented - int two steps: as first thing the code must insert the element in - order as a red leaf in the tree, then the support library function - rb_insert_color() must be called. Such function will do the - not trivial work to rebalance the rbtree if necessary. - ------------------------------------------------------------------------ -static inline struct page * rb_search_page_cache(struct inode * inode, - unsigned long offset) -{ - struct rb_node * n = inode->i_rb_page_cache.rb_node; - struct page * page; - - while (n) - { - page = rb_entry(n, struct page, rb_page_cache); - - if (offset < page->offset) - n = n->rb_left; - else if (offset > page->offset) - n = n->rb_right; - else - return page; - } - return NULL; -} - -static inline struct page * __rb_insert_page_cache(struct inode * inode, - unsigned long offset, - struct rb_node * node) -{ - struct rb_node ** p = &inode->i_rb_page_cache.rb_node; - struct rb_node * parent = NULL; - struct page * page; - - while (*p) - { - parent = *p; - page = rb_entry(parent, struct page, rb_page_cache); - - if (offset < page->offset) - p = &(*p)->rb_left; - else if (offset > page->offset) - p = &(*p)->rb_right; - else - return page; - } - - rb_link_node(node, parent, p); - - return NULL; -} - -static inline struct page * rb_insert_page_cache(struct inode * inode, - unsigned long offset, - struct rb_node * node) -{ - struct page * ret; - if ((ret = __rb_insert_page_cache(inode, offset, node))) - goto out; - rb_insert_color(node, &inode->i_rb_page_cache); - out: - return ret; -} ------------------------------------------------------------------------ -*/ - -#ifndef _LINUX_RBTREE_H -#define _LINUX_RBTREE_H - -#include <stddef.h> - -/** - * container_of - cast a member of a structure out to the containing structure - * @ptr: the pointer to the member. - * @type: the type of the container struct this is embedded in. - * @member: the name of the member within the struct. - * - */ -#define container_of(ptr, type, member) ({ \ - const typeof( ((type *)0)->member ) *__mptr = (ptr); \ - (type *)( (char *)__mptr - offsetof(type,member) );}) - -struct rb_node -{ - unsigned long rb_parent_color; -#define RB_RED 0 -#define RB_BLACK 1 - struct rb_node *rb_right; - struct rb_node *rb_left; -} __attribute__((aligned(sizeof(long)))); - /* The alignment might seem pointless, but allegedly CRIS needs it */ - -struct rb_root -{ - struct rb_node *rb_node; -}; - - -#define rb_parent(r) ((struct rb_node *)((r)->rb_parent_color & ~3)) -#define rb_color(r) ((r)->rb_parent_color & 1) -#define rb_is_red(r) (!rb_color(r)) -#define rb_is_black(r) rb_color(r) -#define rb_set_red(r) do { (r)->rb_parent_color &= ~1; } while (0) -#define rb_set_black(r) do { (r)->rb_parent_color |= 1; } while (0) - -static inline void rb_set_parent(struct rb_node *rb, struct rb_node *p) -{ - rb->rb_parent_color = (rb->rb_parent_color & 3) | (unsigned long)p; -} -static inline void rb_set_color(struct rb_node *rb, int color) -{ - rb->rb_parent_color = (rb->rb_parent_color & ~1) | color; -} - -#define RB_ROOT (struct rb_root) { NULL, } -#define rb_entry(ptr, type, member) container_of(ptr, type, member) - -#define RB_EMPTY_ROOT(root) ((root)->rb_node == NULL) -#define RB_EMPTY_NODE(node) (rb_parent(node) == node) -#define RB_CLEAR_NODE(node) (rb_set_parent(node, node)) - -extern void rb_insert_color(struct rb_node *, struct rb_root *); -extern void rb_erase(struct rb_node *, struct rb_root *); - -/* Find logical next and previous nodes in a tree */ -extern struct rb_node *rb_next(const struct rb_node *); -extern struct rb_node *rb_prev(const struct rb_node *); -extern struct rb_node *rb_first(const struct rb_root *); -extern struct rb_node *rb_last(const struct rb_root *); - -/* Fast replacement of a single node without remove/rebalance/add/rebalance */ -extern void rb_replace_node(struct rb_node *victim, struct rb_node *new, - struct rb_root *root); - -static inline void rb_link_node(struct rb_node * node, struct rb_node * parent, - struct rb_node ** rb_link) -{ - node->rb_parent_color = (unsigned long )parent; - node->rb_left = node->rb_right = NULL; - - *rb_link = node; -} - -#endif /* _LINUX_RBTREE_H */ diff --git a/tools/perf/util/run-command.c b/tools/perf/util/run-command.c index a3935343091..2b615acf94d 100644 --- a/tools/perf/util/run-command.c +++ b/tools/perf/util/run-command.c @@ -262,7 +262,7 @@ int run_hook(const char *index_file, const char *name, ...) { struct child_process hook; const char **argv = NULL, *env[2]; - char index[PATH_MAX]; + char idx[PATH_MAX]; va_list args; int ret; size_t i = 0, alloc = 0; @@ -284,8 +284,8 @@ int run_hook(const char *index_file, const char *name, ...) hook.no_stdin = 1; hook.stdout_to_stderr = 1; if (index_file) { - snprintf(index, sizeof(index), "PERF_INDEX_FILE=%s", index_file); - env[0] = index; + snprintf(idx, sizeof(idx), "PERF_INDEX_FILE=%s", index_file); + env[0] = idx; env[1] = NULL; hook.env = env; } diff --git a/tools/perf/util/strbuf.c b/tools/perf/util/strbuf.c index 464e7ca898c..5249d5a1b0c 100644 --- a/tools/perf/util/strbuf.c +++ b/tools/perf/util/strbuf.c @@ -16,7 +16,7 @@ int prefixcmp(const char *str, const char *prefix) */ char strbuf_slopbuf[1]; -void strbuf_init(struct strbuf *sb, size_t hint) +void strbuf_init(struct strbuf *sb, ssize_t hint) { sb->alloc = sb->len = 0; sb->buf = strbuf_slopbuf; @@ -92,7 +92,8 @@ void strbuf_ltrim(struct strbuf *sb) void strbuf_tolower(struct strbuf *sb) { - int i; + unsigned int i; + for (i = 0; i < sb->len; i++) sb->buf[i] = tolower(sb->buf[i]); } @@ -264,7 +265,7 @@ size_t strbuf_fread(struct strbuf *sb, size_t size, FILE *f) return res; } -ssize_t strbuf_read(struct strbuf *sb, int fd, size_t hint) +ssize_t strbuf_read(struct strbuf *sb, int fd, ssize_t hint) { size_t oldlen = sb->len; size_t oldalloc = sb->alloc; @@ -293,7 +294,7 @@ ssize_t strbuf_read(struct strbuf *sb, int fd, size_t hint) #define STRBUF_MAXLINK (2*PATH_MAX) -int strbuf_readlink(struct strbuf *sb, const char *path, size_t hint) +int strbuf_readlink(struct strbuf *sb, const char *path, ssize_t hint) { size_t oldalloc = sb->alloc; @@ -301,7 +302,7 @@ int strbuf_readlink(struct strbuf *sb, const char *path, size_t hint) hint = 32; while (hint < STRBUF_MAXLINK) { - int len; + ssize_t len; strbuf_grow(sb, hint); len = readlink(path, sb->buf, hint); @@ -343,7 +344,7 @@ int strbuf_getline(struct strbuf *sb, FILE *fp, int term) return 0; } -int strbuf_read_file(struct strbuf *sb, const char *path, size_t hint) +int strbuf_read_file(struct strbuf *sb, const char *path, ssize_t hint) { int fd, len; diff --git a/tools/perf/util/strbuf.h b/tools/perf/util/strbuf.h index 9ee908a3ec5..d2aa86c014c 100644 --- a/tools/perf/util/strbuf.h +++ b/tools/perf/util/strbuf.h @@ -50,7 +50,7 @@ struct strbuf { #define STRBUF_INIT { 0, 0, strbuf_slopbuf } /*----- strbuf life cycle -----*/ -extern void strbuf_init(struct strbuf *, size_t); +extern void strbuf_init(struct strbuf *buf, ssize_t hint); extern void strbuf_release(struct strbuf *); extern char *strbuf_detach(struct strbuf *, size_t *); extern void strbuf_attach(struct strbuf *, void *, size_t, size_t); @@ -61,7 +61,7 @@ static inline void strbuf_swap(struct strbuf *a, struct strbuf *b) { } /*----- strbuf size related -----*/ -static inline size_t strbuf_avail(const struct strbuf *sb) { +static inline ssize_t strbuf_avail(const struct strbuf *sb) { return sb->alloc ? sb->alloc - sb->len - 1 : 0; } @@ -122,9 +122,9 @@ extern void strbuf_addf(struct strbuf *sb, const char *fmt, ...); extern size_t strbuf_fread(struct strbuf *, size_t, FILE *); /* XXX: if read fails, any partial read is undone */ -extern ssize_t strbuf_read(struct strbuf *, int fd, size_t hint); -extern int strbuf_read_file(struct strbuf *sb, const char *path, size_t hint); -extern int strbuf_readlink(struct strbuf *sb, const char *path, size_t hint); +extern ssize_t strbuf_read(struct strbuf *, int fd, ssize_t hint); +extern int strbuf_read_file(struct strbuf *sb, const char *path, ssize_t hint); +extern int strbuf_readlink(struct strbuf *sb, const char *path, ssize_t hint); extern int strbuf_getline(struct strbuf *, FILE *, int); diff --git a/tools/perf/util/string.h b/tools/perf/util/string.h index 3dca2f654cd..bf39dfadfd2 100644 --- a/tools/perf/util/string.h +++ b/tools/perf/util/string.h @@ -5,4 +5,7 @@ int hex2u64(const char *ptr, u64 *val); +#define _STR(x) #x +#define STR(x) _STR(x) + #endif diff --git a/tools/perf/util/strlist.c b/tools/perf/util/strlist.c index 025a78edfff..7ad38171dc2 100644 --- a/tools/perf/util/strlist.c +++ b/tools/perf/util/strlist.c @@ -64,6 +64,7 @@ int strlist__add(struct strlist *self, const char *new_entry) rb_link_node(&sn->rb_node, parent, p); rb_insert_color(&sn->rb_node, &self->entries); + ++self->nr_entries; return 0; } @@ -155,8 +156,9 @@ struct strlist *strlist__new(bool dupstr, const char *slist) struct strlist *self = malloc(sizeof(*self)); if (self != NULL) { - self->entries = RB_ROOT; - self->dupstr = dupstr; + self->entries = RB_ROOT; + self->dupstr = dupstr; + self->nr_entries = 0; if (slist && strlist__parse_list(self, slist) != 0) goto out_error; } @@ -182,3 +184,17 @@ void strlist__delete(struct strlist *self) free(self); } } + +struct str_node *strlist__entry(const struct strlist *self, unsigned int idx) +{ + struct rb_node *nd; + + for (nd = rb_first(&self->entries); nd; nd = rb_next(nd)) { + struct str_node *pos = rb_entry(nd, struct str_node, rb_node); + + if (!idx--) + return pos; + } + + return NULL; +} diff --git a/tools/perf/util/strlist.h b/tools/perf/util/strlist.h index 2fb117fb4b6..921818e44a5 100644 --- a/tools/perf/util/strlist.h +++ b/tools/perf/util/strlist.h @@ -1,7 +1,7 @@ #ifndef STRLIST_H_ #define STRLIST_H_ -#include "rbtree.h" +#include <linux/rbtree.h> #include <stdbool.h> struct str_node { @@ -11,7 +11,8 @@ struct str_node { struct strlist { struct rb_root entries; - bool dupstr; + unsigned int nr_entries; + bool dupstr; }; struct strlist *strlist__new(bool dupstr, const char *slist); @@ -21,11 +22,17 @@ void strlist__remove(struct strlist *self, struct str_node *sn); int strlist__load(struct strlist *self, const char *filename); int strlist__add(struct strlist *self, const char *str); +struct str_node *strlist__entry(const struct strlist *self, unsigned int idx); bool strlist__has_entry(struct strlist *self, const char *entry); static inline bool strlist__empty(const struct strlist *self) { - return rb_first(&self->entries) == NULL; + return self->nr_entries == 0; +} + +static inline unsigned int strlist__nr_entries(const struct strlist *self) +{ + return self->nr_entries; } int strlist__parse_list(struct strlist *self, const char *s); diff --git a/tools/perf/util/svghelper.c b/tools/perf/util/svghelper.c new file mode 100644 index 00000000000..a778fd0f4ae --- /dev/null +++ b/tools/perf/util/svghelper.c @@ -0,0 +1,488 @@ +/* + * svghelper.c - helper functions for outputting svg + * + * (C) Copyright 2009 Intel Corporation + * + * Authors: + * Arjan van de Ven <arjan@linux.intel.com> + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; version 2 + * of the License. + */ + +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <string.h> + +#include "svghelper.h" + +static u64 first_time, last_time; +static u64 turbo_frequency, max_freq; + + +#define SLOT_MULT 30.0 +#define SLOT_HEIGHT 25.0 + +int svg_page_width = 1000; + +#define MIN_TEXT_SIZE 0.001 + +static u64 total_height; +static FILE *svgfile; + +static double cpu2slot(int cpu) +{ + return 2 * cpu + 1; +} + +static double cpu2y(int cpu) +{ + return cpu2slot(cpu) * SLOT_MULT; +} + +static double time2pixels(u64 time) +{ + double X; + + X = 1.0 * svg_page_width * (time - first_time) / (last_time - first_time); + return X; +} + +/* + * Round text sizes so that the svg viewer only needs a discrete + * number of renderings of the font + */ +static double round_text_size(double size) +{ + int loop = 100; + double target = 10.0; + + if (size >= 10.0) + return size; + while (loop--) { + if (size >= target) + return target; + target = target / 2.0; + } + return size; +} + +void open_svg(const char *filename, int cpus, int rows, u64 start, u64 end) +{ + int new_width; + + svgfile = fopen(filename, "w"); + if (!svgfile) { + fprintf(stderr, "Cannot open %s for output\n", filename); + return; + } + first_time = start; + first_time = first_time / 100000000 * 100000000; + last_time = end; + + /* + * if the recording is short, we default to a width of 1000, but + * for longer recordings we want at least 200 units of width per second + */ + new_width = (last_time - first_time) / 5000000; + + if (new_width > svg_page_width) + svg_page_width = new_width; + + total_height = (1 + rows + cpu2slot(cpus)) * SLOT_MULT; + fprintf(svgfile, "<?xml version=\"1.0\" standalone=\"no\"?> \n"); + fprintf(svgfile, "<svg width=\"%i\" height=\"%llu\" version=\"1.1\" xmlns=\"http://www.w3.org/2000/svg\">\n", svg_page_width, total_height); + + fprintf(svgfile, "<defs>\n <style type=\"text/css\">\n <![CDATA[\n"); + + fprintf(svgfile, " rect { stroke-width: 1; }\n"); + fprintf(svgfile, " rect.process { fill:rgb(180,180,180); fill-opacity:0.9; stroke-width:1; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.process2 { fill:rgb(180,180,180); fill-opacity:0.9; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.sample { fill:rgb( 0, 0,255); fill-opacity:0.8; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.blocked { fill:rgb(255, 0, 0); fill-opacity:0.5; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.waiting { fill:rgb(214,214, 0); fill-opacity:0.3; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.WAITING { fill:rgb(255,214, 48); fill-opacity:0.6; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.cpu { fill:rgb(192,192,192); fill-opacity:0.2; stroke-width:0.5; stroke:rgb(128,128,128); } \n"); + fprintf(svgfile, " rect.pstate { fill:rgb(128,128,128); fill-opacity:0.8; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c1 { fill:rgb(255,214,214); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c2 { fill:rgb(255,172,172); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c3 { fill:rgb(255,130,130); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c4 { fill:rgb(255, 88, 88); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c5 { fill:rgb(255, 44, 44); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c6 { fill:rgb(255, 0, 0); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " line.pstate { stroke:rgb(255,255, 0); stroke-opacity:0.8; stroke-width:2; } \n"); + + fprintf(svgfile, " ]]>\n </style>\n</defs>\n"); +} + +void svg_box(int Yslot, u64 start, u64 end, const char *type) +{ + if (!svgfile) + return; + + fprintf(svgfile, "<rect x=\"%4.8f\" width=\"%4.8f\" y=\"%4.1f\" height=\"%4.1f\" class=\"%s\"/>\n", + time2pixels(start), time2pixels(end)-time2pixels(start), Yslot * SLOT_MULT, SLOT_HEIGHT, type); +} + +void svg_sample(int Yslot, int cpu, u64 start, u64 end) +{ + double text_size; + if (!svgfile) + return; + + fprintf(svgfile, "<rect x=\"%4.8f\" width=\"%4.8f\" y=\"%4.1f\" height=\"%4.1f\" class=\"sample\"/>\n", + time2pixels(start), time2pixels(end)-time2pixels(start), Yslot * SLOT_MULT, SLOT_HEIGHT); + + text_size = (time2pixels(end)-time2pixels(start)); + if (cpu > 9) + text_size = text_size/2; + if (text_size > 1.25) + text_size = 1.25; + text_size = round_text_size(text_size); + + if (text_size > MIN_TEXT_SIZE) + fprintf(svgfile, "<text x=\"%1.8f\" y=\"%1.8f\" font-size=\"%1.8fpt\">%i</text>\n", + time2pixels(start), Yslot * SLOT_MULT + SLOT_HEIGHT - 1, text_size, cpu + 1); + +} + +static char *time_to_string(u64 duration) +{ + static char text[80]; + + text[0] = 0; + + if (duration < 1000) /* less than 1 usec */ + return text; + + if (duration < 1000 * 1000) { /* less than 1 msec */ + sprintf(text, "%4.1f us", duration / 1000.0); + return text; + } + sprintf(text, "%4.1f ms", duration / 1000.0 / 1000); + + return text; +} + +void svg_waiting(int Yslot, u64 start, u64 end) +{ + char *text; + const char *style; + double font_size; + + if (!svgfile) + return; + + style = "waiting"; + + if (end-start > 10 * 1000000) /* 10 msec */ + style = "WAITING"; + + text = time_to_string(end-start); + + font_size = 1.0 * (time2pixels(end)-time2pixels(start)); + + if (font_size > 3) + font_size = 3; + + font_size = round_text_size(font_size); + + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\">\n", time2pixels(start), Yslot * SLOT_MULT); + fprintf(svgfile, "<rect x=\"0\" width=\"%4.8f\" y=\"0\" height=\"%4.1f\" class=\"%s\"/>\n", + time2pixels(end)-time2pixels(start), SLOT_HEIGHT, style); + if (font_size > MIN_TEXT_SIZE) + fprintf(svgfile, "<text transform=\"rotate(90)\" font-size=\"%1.8fpt\"> %s</text>\n", + font_size, text); + fprintf(svgfile, "</g>\n"); +} + +static char *cpu_model(void) +{ + static char cpu_m[255]; + char buf[256]; + FILE *file; + + cpu_m[0] = 0; + /* CPU type */ + file = fopen("/proc/cpuinfo", "r"); + if (file) { + while (fgets(buf, 255, file)) { + if (strstr(buf, "model name")) { + strncpy(cpu_m, &buf[13], 255); + break; + } + } + fclose(file); + } + return cpu_m; +} + +void svg_cpu_box(int cpu, u64 __max_freq, u64 __turbo_freq) +{ + char cpu_string[80]; + if (!svgfile) + return; + + max_freq = __max_freq; + turbo_frequency = __turbo_freq; + + fprintf(svgfile, "<rect x=\"%4.8f\" width=\"%4.8f\" y=\"%4.1f\" height=\"%4.1f\" class=\"cpu\"/>\n", + time2pixels(first_time), + time2pixels(last_time)-time2pixels(first_time), + cpu2y(cpu), SLOT_MULT+SLOT_HEIGHT); + + sprintf(cpu_string, "CPU %i", (int)cpu+1); + fprintf(svgfile, "<text x=\"%4.8f\" y=\"%4.8f\">%s</text>\n", + 10+time2pixels(first_time), cpu2y(cpu) + SLOT_HEIGHT/2, cpu_string); + + fprintf(svgfile, "<text transform=\"translate(%4.8f,%4.8f)\" font-size=\"1.25pt\">%s</text>\n", + 10+time2pixels(first_time), cpu2y(cpu) + SLOT_MULT + SLOT_HEIGHT - 4, cpu_model()); +} + +void svg_process(int cpu, u64 start, u64 end, const char *type, const char *name) +{ + double width; + + if (!svgfile) + return; + + + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\">\n", time2pixels(start), cpu2y(cpu)); + fprintf(svgfile, "<rect x=\"0\" width=\"%4.8f\" y=\"0\" height=\"%4.1f\" class=\"%s\"/>\n", + time2pixels(end)-time2pixels(start), SLOT_MULT+SLOT_HEIGHT, type); + width = time2pixels(end)-time2pixels(start); + if (width > 6) + width = 6; + + width = round_text_size(width); + + if (width > MIN_TEXT_SIZE) + fprintf(svgfile, "<text transform=\"rotate(90)\" font-size=\"%3.8fpt\">%s</text>\n", + width, name); + + fprintf(svgfile, "</g>\n"); +} + +void svg_cstate(int cpu, u64 start, u64 end, int type) +{ + double width; + char style[128]; + + if (!svgfile) + return; + + + if (type > 6) + type = 6; + sprintf(style, "c%i", type); + + fprintf(svgfile, "<rect class=\"%s\" x=\"%4.8f\" width=\"%4.8f\" y=\"%4.1f\" height=\"%4.1f\"/>\n", + style, + time2pixels(start), time2pixels(end)-time2pixels(start), + cpu2y(cpu), SLOT_MULT+SLOT_HEIGHT); + + width = (time2pixels(end)-time2pixels(start))/2.0; + if (width > 6) + width = 6; + + width = round_text_size(width); + + if (width > MIN_TEXT_SIZE) + fprintf(svgfile, "<text x=\"%4.8f\" y=\"%4.8f\" font-size=\"%3.8fpt\">C%i</text>\n", + time2pixels(start), cpu2y(cpu)+width, width, type); +} + +static char *HzToHuman(unsigned long hz) +{ + static char buffer[1024]; + unsigned long long Hz; + + memset(buffer, 0, 1024); + + Hz = hz; + + /* default: just put the Number in */ + sprintf(buffer, "%9lli", Hz); + + if (Hz > 1000) + sprintf(buffer, " %6lli Mhz", (Hz+500)/1000); + + if (Hz > 1500000) + sprintf(buffer, " %6.2f Ghz", (Hz+5000.0)/1000000); + + if (Hz == turbo_frequency) + sprintf(buffer, "Turbo"); + + return buffer; +} + +void svg_pstate(int cpu, u64 start, u64 end, u64 freq) +{ + double height = 0; + + if (!svgfile) + return; + + if (max_freq) + height = freq * 1.0 / max_freq * (SLOT_HEIGHT + SLOT_MULT); + height = 1 + cpu2y(cpu) + SLOT_MULT + SLOT_HEIGHT - height; + fprintf(svgfile, "<line x1=\"%4.8f\" x2=\"%4.8f\" y1=\"%4.1f\" y2=\"%4.1f\" class=\"pstate\"/>\n", + time2pixels(start), time2pixels(end), height, height); + fprintf(svgfile, "<text x=\"%4.8f\" y=\"%4.8f\" font-size=\"0.25pt\">%s</text>\n", + time2pixels(start), height+0.9, HzToHuman(freq)); + +} + + +void svg_partial_wakeline(u64 start, int row1, char *desc1, int row2, char *desc2) +{ + double height; + + if (!svgfile) + return; + + + if (row1 < row2) { + if (row1) { + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row1 * SLOT_MULT + SLOT_HEIGHT, time2pixels(start), row1 * SLOT_MULT + SLOT_HEIGHT + SLOT_MULT/32); + if (desc2) + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\"><text transform=\"rotate(90)\" font-size=\"0.02pt\">%s ></text></g>\n", + time2pixels(start), row1 * SLOT_MULT + SLOT_HEIGHT + SLOT_HEIGHT/48, desc2); + } + if (row2) { + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row2 * SLOT_MULT - SLOT_MULT/32, time2pixels(start), row2 * SLOT_MULT); + if (desc1) + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\"><text transform=\"rotate(90)\" font-size=\"0.02pt\">%s ></text></g>\n", + time2pixels(start), row2 * SLOT_MULT - SLOT_MULT/32, desc1); + } + } else { + if (row2) { + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row2 * SLOT_MULT + SLOT_HEIGHT, time2pixels(start), row2 * SLOT_MULT + SLOT_HEIGHT + SLOT_MULT/32); + if (desc1) + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\"><text transform=\"rotate(90)\" font-size=\"0.02pt\">%s <</text></g>\n", + time2pixels(start), row2 * SLOT_MULT + SLOT_HEIGHT + SLOT_MULT/48, desc1); + } + if (row1) { + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row1 * SLOT_MULT - SLOT_MULT/32, time2pixels(start), row1 * SLOT_MULT); + if (desc2) + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\"><text transform=\"rotate(90)\" font-size=\"0.02pt\">%s <</text></g>\n", + time2pixels(start), row1 * SLOT_MULT - SLOT_HEIGHT/32, desc2); + } + } + height = row1 * SLOT_MULT; + if (row2 > row1) + height += SLOT_HEIGHT; + if (row1) + fprintf(svgfile, "<circle cx=\"%4.8f\" cy=\"%4.2f\" r = \"0.01\" style=\"fill:rgb(32,255,32)\"/>\n", + time2pixels(start), height); +} + +void svg_wakeline(u64 start, int row1, int row2) +{ + double height; + + if (!svgfile) + return; + + + if (row1 < row2) + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row1 * SLOT_MULT + SLOT_HEIGHT, time2pixels(start), row2 * SLOT_MULT); + else + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row2 * SLOT_MULT + SLOT_HEIGHT, time2pixels(start), row1 * SLOT_MULT); + + height = row1 * SLOT_MULT; + if (row2 > row1) + height += SLOT_HEIGHT; + fprintf(svgfile, "<circle cx=\"%4.8f\" cy=\"%4.2f\" r = \"0.01\" style=\"fill:rgb(32,255,32)\"/>\n", + time2pixels(start), height); +} + +void svg_interrupt(u64 start, int row) +{ + if (!svgfile) + return; + + fprintf(svgfile, "<circle cx=\"%4.8f\" cy=\"%4.2f\" r = \"0.01\" style=\"fill:rgb(255,128,128)\"/>\n", + time2pixels(start), row * SLOT_MULT); + fprintf(svgfile, "<circle cx=\"%4.8f\" cy=\"%4.2f\" r = \"0.01\" style=\"fill:rgb(255,128,128)\"/>\n", + time2pixels(start), row * SLOT_MULT + SLOT_HEIGHT); +} + +void svg_text(int Yslot, u64 start, const char *text) +{ + if (!svgfile) + return; + + fprintf(svgfile, "<text x=\"%4.8f\" y=\"%4.8f\">%s</text>\n", + time2pixels(start), Yslot * SLOT_MULT+SLOT_HEIGHT/2, text); +} + +static void svg_legenda_box(int X, const char *text, const char *style) +{ + double boxsize; + boxsize = SLOT_HEIGHT / 2; + + fprintf(svgfile, "<rect x=\"%i\" width=\"%4.8f\" y=\"0\" height=\"%4.1f\" class=\"%s\"/>\n", + X, boxsize, boxsize, style); + fprintf(svgfile, "<text transform=\"translate(%4.8f, %4.8f)\" font-size=\"%4.8fpt\">%s</text>\n", + X + boxsize + 5, boxsize, 0.8 * boxsize, text); +} + +void svg_legenda(void) +{ + if (!svgfile) + return; + + svg_legenda_box(0, "Running", "sample"); + svg_legenda_box(100, "Idle","rect.c1"); + svg_legenda_box(200, "Deeper Idle", "rect.c3"); + svg_legenda_box(350, "Deepest Idle", "rect.c6"); + svg_legenda_box(550, "Sleeping", "process2"); + svg_legenda_box(650, "Waiting for cpu", "waiting"); + svg_legenda_box(800, "Blocked on IO", "blocked"); +} + +void svg_time_grid(void) +{ + u64 i; + + if (!svgfile) + return; + + i = first_time; + while (i < last_time) { + int color = 220; + double thickness = 0.075; + if ((i % 100000000) == 0) { + thickness = 0.5; + color = 192; + } + if ((i % 1000000000) == 0) { + thickness = 2.0; + color = 128; + } + + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%llu\" style=\"stroke:rgb(%i,%i,%i);stroke-width:%1.3f\"/>\n", + time2pixels(i), SLOT_MULT/2, time2pixels(i), total_height, color, color, color, thickness); + + i += 10000000; + } +} + +void svg_close(void) +{ + if (svgfile) { + fprintf(svgfile, "</svg>\n"); + fclose(svgfile); + svgfile = NULL; + } +} diff --git a/tools/perf/util/svghelper.h b/tools/perf/util/svghelper.h new file mode 100644 index 00000000000..cd93195aedb --- /dev/null +++ b/tools/perf/util/svghelper.h @@ -0,0 +1,28 @@ +#ifndef _INCLUDE_GUARD_SVG_HELPER_ +#define _INCLUDE_GUARD_SVG_HELPER_ + +#include "types.h" + +extern void open_svg(const char *filename, int cpus, int rows, u64 start, u64 end); +extern void svg_box(int Yslot, u64 start, u64 end, const char *type); +extern void svg_sample(int Yslot, int cpu, u64 start, u64 end); +extern void svg_waiting(int Yslot, u64 start, u64 end); +extern void svg_cpu_box(int cpu, u64 max_frequency, u64 turbo_frequency); + + +extern void svg_process(int cpu, u64 start, u64 end, const char *type, const char *name); +extern void svg_cstate(int cpu, u64 start, u64 end, int type); +extern void svg_pstate(int cpu, u64 start, u64 end, u64 freq); + + +extern void svg_time_grid(void); +extern void svg_legenda(void); +extern void svg_wakeline(u64 start, int row1, int row2); +extern void svg_partial_wakeline(u64 start, int row1, char *desc1, int row2, char *desc2); +extern void svg_interrupt(u64 start, int row); +extern void svg_text(int Yslot, u64 start, const char *text); +extern void svg_close(void); + +extern int svg_page_width; + +#endif diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 78c2efde01b..fd3d9c8e90f 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -3,15 +3,27 @@ #include "string.h" #include "symbol.h" +#include "debug.h" + #include <libelf.h> #include <gelf.h> #include <elf.h> const char *sym_hist_filter; +enum dso_origin { + DSO__ORIG_KERNEL = 0, + DSO__ORIG_JAVA_JIT, + DSO__ORIG_FEDORA, + DSO__ORIG_UBUNTU, + DSO__ORIG_BUILDID, + DSO__ORIG_DSO, + DSO__ORIG_NOT_FOUND, +}; + static struct symbol *symbol__new(u64 start, u64 len, const char *name, unsigned int priv_size, - u64 obj_start, int verbose) + u64 obj_start, int v) { size_t namelen = strlen(name) + 1; struct symbol *self = calloc(1, priv_size + sizeof(*self) + namelen); @@ -19,7 +31,7 @@ static struct symbol *symbol__new(u64 start, u64 len, if (!self) return NULL; - if (verbose >= 2) + if (v >= 2) printf("new symbol: %016Lx [%08lx]: %s, hist: %p, obj_start: %p\n", (u64)start, (unsigned long)len, name, self->hist, (void *)(unsigned long)obj_start); @@ -35,7 +47,7 @@ static struct symbol *symbol__new(u64 start, u64 len, self = ((void *)self) + priv_size; } self->start = start; - self->end = start + len - 1; + self->end = len ? start + len - 1 : start; memcpy(self->name, name, namelen); return self; @@ -48,8 +60,12 @@ static void symbol__delete(struct symbol *self, unsigned int priv_size) static size_t symbol__fprintf(struct symbol *self, FILE *fp) { - return fprintf(fp, " %llx-%llx %s\n", + if (!self->module) + return fprintf(fp, " %llx-%llx %s\n", self->start, self->end, self->name); + else + return fprintf(fp, " %llx-%llx %s \t[%s]\n", + self->start, self->end, self->name, self->module->name); } struct dso *dso__new(const char *name, unsigned int sym_priv_size) @@ -61,6 +77,8 @@ struct dso *dso__new(const char *name, unsigned int sym_priv_size) self->syms = RB_ROOT; self->sym_priv_size = sym_priv_size; self->find_symbol = dso__find_symbol; + self->slen_calculated = 0; + self->origin = DSO__ORIG_NOT_FOUND; } return self; @@ -140,12 +158,13 @@ size_t dso__fprintf(struct dso *self, FILE *fp) return ret; } -static int dso__load_kallsyms(struct dso *self, symbol_filter_t filter, int verbose) +static int dso__load_kallsyms(struct dso *self, symbol_filter_t filter, int v) { struct rb_node *nd, *prevnd; char *line = NULL; size_t n; FILE *file = fopen("/proc/kallsyms", "r"); + int count = 0; if (file == NULL) goto out_failure; @@ -181,15 +200,17 @@ static int dso__load_kallsyms(struct dso *self, symbol_filter_t filter, int verb * Well fix up the end later, when we have all sorted. */ sym = symbol__new(start, 0xdead, line + len + 2, - self->sym_priv_size, 0, verbose); + self->sym_priv_size, 0, v); if (sym == NULL) goto out_delete_line; if (filter && filter(self, sym)) symbol__delete(sym, self->sym_priv_size); - else + else { dso__insert_symbol(self, sym); + count++; + } } /* @@ -212,7 +233,7 @@ static int dso__load_kallsyms(struct dso *self, symbol_filter_t filter, int verb free(line); fclose(file); - return 0; + return count; out_delete_line: free(line); @@ -220,7 +241,7 @@ out_failure: return -1; } -static int dso__load_perf_map(struct dso *self, symbol_filter_t filter, int verbose) +static int dso__load_perf_map(struct dso *self, symbol_filter_t filter, int v) { char *line = NULL; size_t n; @@ -258,7 +279,7 @@ static int dso__load_perf_map(struct dso *self, symbol_filter_t filter, int verb continue; sym = symbol__new(start, size, line + len, - self->sym_priv_size, start, verbose); + self->sym_priv_size, start, v); if (sym == NULL) goto out_delete_line; @@ -286,13 +307,13 @@ out_failure: * elf_symtab__for_each_symbol - iterate thru all the symbols * * @self: struct elf_symtab instance to iterate - * @index: uint32_t index + * @idx: uint32_t idx * @sym: GElf_Sym iterator */ -#define elf_symtab__for_each_symbol(syms, nr_syms, index, sym) \ - for (index = 0, gelf_getsym(syms, index, &sym);\ - index < nr_syms; \ - index++, gelf_getsym(syms, index, &sym)) +#define elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) \ + for (idx = 0, gelf_getsym(syms, idx, &sym);\ + idx < nr_syms; \ + idx++, gelf_getsym(syms, idx, &sym)) static inline uint8_t elf_sym__type(const GElf_Sym *sym) { @@ -307,6 +328,26 @@ static inline int elf_sym__is_function(const GElf_Sym *sym) sym->st_size != 0; } +static inline int elf_sym__is_label(const GElf_Sym *sym) +{ + return elf_sym__type(sym) == STT_NOTYPE && + sym->st_name != 0 && + sym->st_shndx != SHN_UNDEF && + sym->st_shndx != SHN_ABS; +} + +static inline const char *elf_sec__name(const GElf_Shdr *shdr, + const Elf_Data *secstrs) +{ + return secstrs->d_buf + shdr->sh_name; +} + +static inline int elf_sec__is_text(const GElf_Shdr *shdr, + const Elf_Data *secstrs) +{ + return strstr(elf_sec__name(shdr, secstrs), "text") != NULL; +} + static inline const char *elf_sym__name(const GElf_Sym *sym, const Elf_Data *symstrs) { @@ -315,7 +356,7 @@ static inline const char *elf_sym__name(const GElf_Sym *sym, static Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep, GElf_Shdr *shp, const char *name, - size_t *index) + size_t *idx) { Elf_Scn *sec = NULL; size_t cnt = 1; @@ -326,8 +367,8 @@ static Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep, gelf_getshdr(sec, shp); str = elf_strptr(elf, ep->e_shstrndx, shp->sh_name); if (!strcmp(name, str)) { - if (index) - *index = cnt; + if (idx) + *idx = cnt; break; } ++cnt; @@ -346,56 +387,81 @@ static Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep, idx < nr_entries; \ ++idx, pos = gelf_getrela(reldata, idx, &pos_mem)) -static int dso__synthesize_plt_symbols(struct dso *self, Elf *elf, - GElf_Ehdr *ehdr, Elf_Scn *scn_dynsym, - GElf_Shdr *shdr_dynsym, - size_t dynsym_idx, int verbose) +/* + * We need to check if we have a .dynsym, so that we can handle the + * .plt, synthesizing its symbols, that aren't on the symtabs (be it + * .dynsym or .symtab). + * And always look at the original dso, not at debuginfo packages, that + * have the PLT data stripped out (shdr_rel_plt.sh_type == SHT_NOBITS). + */ +static int dso__synthesize_plt_symbols(struct dso *self, int v) { uint32_t nr_rel_entries, idx; GElf_Sym sym; u64 plt_offset; GElf_Shdr shdr_plt; struct symbol *f; - GElf_Shdr shdr_rel_plt; + GElf_Shdr shdr_rel_plt, shdr_dynsym; Elf_Data *reldata, *syms, *symstrs; - Elf_Scn *scn_plt_rel, *scn_symstrs; + Elf_Scn *scn_plt_rel, *scn_symstrs, *scn_dynsym; + size_t dynsym_idx; + GElf_Ehdr ehdr; char sympltname[1024]; - int nr = 0, symidx; + Elf *elf; + int nr = 0, symidx, fd, err = 0; + + fd = open(self->name, O_RDONLY); + if (fd < 0) + goto out; + + elf = elf_begin(fd, ELF_C_READ_MMAP, NULL); + if (elf == NULL) + goto out_close; + + if (gelf_getehdr(elf, &ehdr) == NULL) + goto out_elf_end; + + scn_dynsym = elf_section_by_name(elf, &ehdr, &shdr_dynsym, + ".dynsym", &dynsym_idx); + if (scn_dynsym == NULL) + goto out_elf_end; - scn_plt_rel = elf_section_by_name(elf, ehdr, &shdr_rel_plt, + scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt, ".rela.plt", NULL); if (scn_plt_rel == NULL) { - scn_plt_rel = elf_section_by_name(elf, ehdr, &shdr_rel_plt, + scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt, ".rel.plt", NULL); if (scn_plt_rel == NULL) - return 0; + goto out_elf_end; } + err = -1; + if (shdr_rel_plt.sh_link != dynsym_idx) - return 0; + goto out_elf_end; - if (elf_section_by_name(elf, ehdr, &shdr_plt, ".plt", NULL) == NULL) - return 0; + if (elf_section_by_name(elf, &ehdr, &shdr_plt, ".plt", NULL) == NULL) + goto out_elf_end; /* - * Fetch the relocation section to find the indexes to the GOT + * Fetch the relocation section to find the idxes to the GOT * and the symbols in the .dynsym they refer to. */ reldata = elf_getdata(scn_plt_rel, NULL); if (reldata == NULL) - return -1; + goto out_elf_end; syms = elf_getdata(scn_dynsym, NULL); if (syms == NULL) - return -1; + goto out_elf_end; - scn_symstrs = elf_getscn(elf, shdr_dynsym->sh_link); + scn_symstrs = elf_getscn(elf, shdr_dynsym.sh_link); if (scn_symstrs == NULL) - return -1; + goto out_elf_end; symstrs = elf_getdata(scn_symstrs, NULL); if (symstrs == NULL) - return -1; + goto out_elf_end; nr_rel_entries = shdr_rel_plt.sh_size / shdr_rel_plt.sh_entsize; plt_offset = shdr_plt.sh_offset; @@ -412,9 +478,9 @@ static int dso__synthesize_plt_symbols(struct dso *self, Elf *elf, "%s@plt", elf_sym__name(&sym, symstrs)); f = symbol__new(plt_offset, shdr_plt.sh_entsize, - sympltname, self->sym_priv_size, 0, verbose); + sympltname, self->sym_priv_size, 0, v); if (!f) - return -1; + goto out_elf_end; dso__insert_symbol(self, f); ++nr; @@ -430,79 +496,63 @@ static int dso__synthesize_plt_symbols(struct dso *self, Elf *elf, "%s@plt", elf_sym__name(&sym, symstrs)); f = symbol__new(plt_offset, shdr_plt.sh_entsize, - sympltname, self->sym_priv_size, 0, verbose); + sympltname, self->sym_priv_size, 0, v); if (!f) - return -1; + goto out_elf_end; dso__insert_symbol(self, f); ++nr; } - } else { - /* - * TODO: There are still one more shdr_rel_plt.sh_type - * I have to investigate, but probably should be ignored. - */ } - return nr; + err = 0; +out_elf_end: + elf_end(elf); +out_close: + close(fd); + + if (err == 0) + return nr; +out: + fprintf(stderr, "%s: problems reading %s PLT info.\n", + __func__, self->name); + return 0; } static int dso__load_sym(struct dso *self, int fd, const char *name, - symbol_filter_t filter, int verbose) + symbol_filter_t filter, int v, struct module *mod) { - Elf_Data *symstrs; + Elf_Data *symstrs, *secstrs; uint32_t nr_syms; int err = -1; - uint32_t index; + uint32_t idx; GElf_Ehdr ehdr; GElf_Shdr shdr; Elf_Data *syms; GElf_Sym sym; - Elf_Scn *sec, *sec_dynsym; + Elf_Scn *sec, *sec_strndx; Elf *elf; - size_t dynsym_idx; - int nr = 0; + int nr = 0, kernel = !strcmp("[kernel]", self->name); elf = elf_begin(fd, ELF_C_READ_MMAP, NULL); if (elf == NULL) { - if (verbose) + if (v) fprintf(stderr, "%s: cannot read %s ELF file.\n", __func__, name); goto out_close; } if (gelf_getehdr(elf, &ehdr) == NULL) { - if (verbose) + if (v) fprintf(stderr, "%s: cannot get elf header.\n", __func__); goto out_elf_end; } - /* - * We need to check if we have a .dynsym, so that we can handle the - * .plt, synthesizing its symbols, that aren't on the symtabs (be it - * .dynsym or .symtab) - */ - sec_dynsym = elf_section_by_name(elf, &ehdr, &shdr, - ".dynsym", &dynsym_idx); - if (sec_dynsym != NULL) { - nr = dso__synthesize_plt_symbols(self, elf, &ehdr, - sec_dynsym, &shdr, - dynsym_idx, verbose); - if (nr < 0) - goto out_elf_end; - } - - /* - * But if we have a full .symtab (that is a superset of .dynsym) we - * should add the symbols not in the .dynsyn - */ sec = elf_section_by_name(elf, &ehdr, &shdr, ".symtab", NULL); if (sec == NULL) { - if (sec_dynsym == NULL) + sec = elf_section_by_name(elf, &ehdr, &shdr, ".dynsym", NULL); + if (sec == NULL) goto out_elf_end; - - sec = sec_dynsym; - gelf_getshdr(sec, &shdr); } syms = elf_getdata(sec, NULL); @@ -517,17 +567,34 @@ static int dso__load_sym(struct dso *self, int fd, const char *name, if (symstrs == NULL) goto out_elf_end; + sec_strndx = elf_getscn(elf, ehdr.e_shstrndx); + if (sec_strndx == NULL) + goto out_elf_end; + + secstrs = elf_getdata(sec_strndx, NULL); + if (secstrs == NULL) + goto out_elf_end; + nr_syms = shdr.sh_size / shdr.sh_entsize; memset(&sym, 0, sizeof(sym)); - self->prelinked = elf_section_by_name(elf, &ehdr, &shdr, - ".gnu.prelink_undo", - NULL) != NULL; - elf_symtab__for_each_symbol(syms, nr_syms, index, sym) { + if (!kernel) { + self->adjust_symbols = (ehdr.e_type == ET_EXEC || + elf_section_by_name(elf, &ehdr, &shdr, + ".gnu.prelink_undo", + NULL) != NULL); + } else self->adjust_symbols = 0; + + elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) { struct symbol *f; + const char *elf_name; + char *demangled; u64 obj_start; + struct section *section = NULL; + int is_label = elf_sym__is_label(&sym); + const char *section_name; - if (!elf_sym__is_function(&sym)) + if (!is_label && !elf_sym__is_function(&sym)) continue; sec = elf_getscn(elf, sym.st_shndx); @@ -535,25 +602,51 @@ static int dso__load_sym(struct dso *self, int fd, const char *name, goto out_elf_end; gelf_getshdr(sec, &shdr); + + if (is_label && !elf_sec__is_text(&shdr, secstrs)) + continue; + + section_name = elf_sec__name(&shdr, secstrs); obj_start = sym.st_value; - if (self->prelinked) { - if (verbose >= 2) + if (self->adjust_symbols) { + if (v >= 2) printf("adjusting symbol: st_value: %Lx sh_addr: %Lx sh_offset: %Lx\n", (u64)sym.st_value, (u64)shdr.sh_addr, (u64)shdr.sh_offset); sym.st_value -= shdr.sh_addr - shdr.sh_offset; } - f = symbol__new(sym.st_value, sym.st_size, - elf_sym__name(&sym, symstrs), - self->sym_priv_size, obj_start, verbose); + if (mod) { + section = mod->sections->find_section(mod->sections, section_name); + if (section) + sym.st_value += section->vma; + else { + fprintf(stderr, "dso__load_sym() module %s lookup of %s failed\n", + mod->name, section_name); + goto out_elf_end; + } + } + /* + * We need to figure out if the object was created from C++ sources + * DWARF DW_compile_unit has this, but we don't always have access + * to it... + */ + elf_name = elf_sym__name(&sym, symstrs); + demangled = bfd_demangle(NULL, elf_name, DMGL_PARAMS | DMGL_ANSI); + if (demangled != NULL) + elf_name = demangled; + + f = symbol__new(sym.st_value, sym.st_size, elf_name, + self->sym_priv_size, obj_start, v); + free(demangled); if (!f) goto out_elf_end; if (filter && filter(self, f)) symbol__delete(f, self->sym_priv_size); else { + f->module = mod; dso__insert_symbol(self, f); nr++; } @@ -566,44 +659,135 @@ out_close: return err; } -int dso__load(struct dso *self, symbol_filter_t filter, int verbose) +#define BUILD_ID_SIZE 128 + +static char *dso__read_build_id(struct dso *self, int v) +{ + int i; + GElf_Ehdr ehdr; + GElf_Shdr shdr; + Elf_Data *build_id_data; + Elf_Scn *sec; + char *build_id = NULL, *bid; + unsigned char *raw; + Elf *elf; + int fd = open(self->name, O_RDONLY); + + if (fd < 0) + goto out; + + elf = elf_begin(fd, ELF_C_READ_MMAP, NULL); + if (elf == NULL) { + if (v) + fprintf(stderr, "%s: cannot read %s ELF file.\n", + __func__, self->name); + goto out_close; + } + + if (gelf_getehdr(elf, &ehdr) == NULL) { + if (v) + fprintf(stderr, "%s: cannot get elf header.\n", __func__); + goto out_elf_end; + } + + sec = elf_section_by_name(elf, &ehdr, &shdr, ".note.gnu.build-id", NULL); + if (sec == NULL) + goto out_elf_end; + + build_id_data = elf_getdata(sec, NULL); + if (build_id_data == NULL) + goto out_elf_end; + build_id = malloc(BUILD_ID_SIZE); + if (build_id == NULL) + goto out_elf_end; + raw = build_id_data->d_buf + 16; + bid = build_id; + + for (i = 0; i < 20; ++i) { + sprintf(bid, "%02x", *raw); + ++raw; + bid += 2; + } + if (v >= 2) + printf("%s(%s): %s\n", __func__, self->name, build_id); +out_elf_end: + elf_end(elf); +out_close: + close(fd); +out: + return build_id; +} + +char dso__symtab_origin(const struct dso *self) { - int size = strlen(self->name) + sizeof("/usr/lib/debug%s.debug"); - char *name = malloc(size); - int variant = 0; + static const char origin[] = { + [DSO__ORIG_KERNEL] = 'k', + [DSO__ORIG_JAVA_JIT] = 'j', + [DSO__ORIG_FEDORA] = 'f', + [DSO__ORIG_UBUNTU] = 'u', + [DSO__ORIG_BUILDID] = 'b', + [DSO__ORIG_DSO] = 'd', + }; + + if (self == NULL || self->origin == DSO__ORIG_NOT_FOUND) + return '!'; + return origin[self->origin]; +} + +int dso__load(struct dso *self, symbol_filter_t filter, int v) +{ + int size = PATH_MAX; + char *name = malloc(size), *build_id = NULL; int ret = -1; int fd; if (!name) return -1; - self->prelinked = 0; + self->adjust_symbols = 0; + + if (strncmp(self->name, "/tmp/perf-", 10) == 0) { + ret = dso__load_perf_map(self, filter, v); + self->origin = ret > 0 ? DSO__ORIG_JAVA_JIT : + DSO__ORIG_NOT_FOUND; + return ret; + } - if (strncmp(self->name, "/tmp/perf-", 10) == 0) - return dso__load_perf_map(self, filter, verbose); + self->origin = DSO__ORIG_FEDORA - 1; more: do { - switch (variant) { - case 0: /* Fedora */ + self->origin++; + switch (self->origin) { + case DSO__ORIG_FEDORA: snprintf(name, size, "/usr/lib/debug%s.debug", self->name); break; - case 1: /* Ubuntu */ + case DSO__ORIG_UBUNTU: snprintf(name, size, "/usr/lib/debug%s", self->name); break; - case 2: /* Sane people */ + case DSO__ORIG_BUILDID: + build_id = dso__read_build_id(self, v); + if (build_id != NULL) { + snprintf(name, size, + "/usr/lib/debug/.build-id/%.2s/%s.debug", + build_id, build_id + 2); + free(build_id); + break; + } + self->origin++; + /* Fall thru */ + case DSO__ORIG_DSO: snprintf(name, size, "%s", self->name); break; default: goto out; } - variant++; fd = open(name, O_RDONLY); } while (fd < 0); - ret = dso__load_sym(self, fd, name, filter, verbose); + ret = dso__load_sym(self, fd, name, filter, v, NULL); close(fd); /* @@ -612,39 +796,233 @@ more: if (!ret) goto more; + if (ret > 0) { + int nr_plt = dso__synthesize_plt_symbols(self, v); + if (nr_plt > 0) + ret += nr_plt; + } out: free(name); + if (ret < 0 && strstr(self->name, " (deleted)") != NULL) + return 0; return ret; } +static int dso__load_module(struct dso *self, struct mod_dso *mods, const char *name, + symbol_filter_t filter, int v) +{ + struct module *mod = mod_dso__find_module(mods, name); + int err = 0, fd; + + if (mod == NULL || !mod->active) + return err; + + fd = open(mod->path, O_RDONLY); + + if (fd < 0) + return err; + + err = dso__load_sym(self, fd, name, filter, v, mod); + close(fd); + + return err; +} + +int dso__load_modules(struct dso *self, symbol_filter_t filter, int v) +{ + struct mod_dso *mods = mod_dso__new_dso("modules"); + struct module *pos; + struct rb_node *next; + int err; + + err = mod_dso__load_modules(mods); + + if (err <= 0) + return err; + + /* + * Iterate over modules, and load active symbols. + */ + next = rb_first(&mods->mods); + while (next) { + pos = rb_entry(next, struct module, rb_node); + err = dso__load_module(self, mods, pos->name, filter, v); + + if (err < 0) + break; + + next = rb_next(&pos->rb_node); + } + + if (err < 0) { + mod_dso__delete_modules(mods); + mod_dso__delete_self(mods); + } + + return err; +} + +static inline void dso__fill_symbol_holes(struct dso *self) +{ + struct symbol *prev = NULL; + struct rb_node *nd; + + for (nd = rb_last(&self->syms); nd; nd = rb_prev(nd)) { + struct symbol *pos = rb_entry(nd, struct symbol, rb_node); + + if (prev) { + u64 hole = 0; + int alias = pos->start == prev->start; + + if (!alias) + hole = prev->start - pos->end - 1; + + if (hole || alias) { + if (alias) + pos->end = prev->end; + else if (hole) + pos->end = prev->start - 1; + } + } + prev = pos; + } +} + static int dso__load_vmlinux(struct dso *self, const char *vmlinux, - symbol_filter_t filter, int verbose) + symbol_filter_t filter, int v) { int err, fd = open(vmlinux, O_RDONLY); if (fd < 0) return -1; - err = dso__load_sym(self, fd, vmlinux, filter, verbose); + err = dso__load_sym(self, fd, vmlinux, filter, v, NULL); + + if (err > 0) + dso__fill_symbol_holes(self); + close(fd); return err; } int dso__load_kernel(struct dso *self, const char *vmlinux, - symbol_filter_t filter, int verbose) + symbol_filter_t filter, int v, int use_modules) { int err = -1; - if (vmlinux) - err = dso__load_vmlinux(self, vmlinux, filter, verbose); + if (vmlinux) { + err = dso__load_vmlinux(self, vmlinux, filter, v); + if (err > 0 && use_modules) + err = dso__load_modules(self, filter, v); + } + + if (err <= 0) + err = dso__load_kallsyms(self, filter, v); + + if (err > 0) + self->origin = DSO__ORIG_KERNEL; + + return err; +} + +LIST_HEAD(dsos); +struct dso *kernel_dso; +struct dso *vdso; +struct dso *hypervisor_dso; + +const char *vmlinux_name = "vmlinux"; +int modules; + +static void dsos__add(struct dso *dso) +{ + list_add_tail(&dso->node, &dsos); +} + +static struct dso *dsos__find(const char *name) +{ + struct dso *pos; + + list_for_each_entry(pos, &dsos, node) + if (strcmp(pos->name, name) == 0) + return pos; + return NULL; +} + +struct dso *dsos__findnew(const char *name) +{ + struct dso *dso = dsos__find(name); + int nr; + + if (dso) + return dso; + + dso = dso__new(name, 0); + if (!dso) + goto out_delete_dso; - if (err < 0) - err = dso__load_kallsyms(self, filter, verbose); + nr = dso__load(dso, NULL, verbose); + if (nr < 0) { + eprintf("Failed to open: %s\n", name); + goto out_delete_dso; + } + if (!nr) + eprintf("No symbols found in: %s, maybe install a debug package?\n", name); + + dsos__add(dso); + + return dso; + +out_delete_dso: + dso__delete(dso); + return NULL; +} + +void dsos__fprintf(FILE *fp) +{ + struct dso *pos; + + list_for_each_entry(pos, &dsos, node) + dso__fprintf(pos, fp); +} + +static struct symbol *vdso__find_symbol(struct dso *dso, u64 ip) +{ + return dso__find_symbol(dso, ip); +} + +int load_kernel(void) +{ + int err; + + kernel_dso = dso__new("[kernel]", 0); + if (!kernel_dso) + return -1; + + err = dso__load_kernel(kernel_dso, vmlinux_name, NULL, verbose, modules); + if (err <= 0) { + dso__delete(kernel_dso); + kernel_dso = NULL; + } else + dsos__add(kernel_dso); + + vdso = dso__new("[vdso]", 0); + if (!vdso) + return -1; + + vdso->find_symbol = vdso__find_symbol; + + dsos__add(vdso); + + hypervisor_dso = dso__new("[hypervisor]", 0); + if (!hypervisor_dso) + return -1; + dsos__add(hypervisor_dso); return err; } + void symbol__init(void) { elf_version(EV_CURRENT); diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h index 2c48ace8203..6e849071640 100644 --- a/tools/perf/util/symbol.h +++ b/tools/perf/util/symbol.h @@ -3,8 +3,34 @@ #include <linux/types.h> #include "types.h" -#include "list.h" -#include "rbtree.h" +#include <linux/list.h> +#include <linux/rbtree.h> +#include "module.h" +#include "event.h" + +#ifdef HAVE_CPLUS_DEMANGLE +extern char *cplus_demangle(const char *, int); + +static inline char *bfd_demangle(void __used *v, const char *c, int i) +{ + return cplus_demangle(c, i); +} +#else +#ifdef NO_DEMANGLE +static inline char *bfd_demangle(void __used *v, const char __used *c, + int __used i) +{ + return NULL; +} +#else +#include <bfd.h> +#endif +#endif + +#ifndef DMGL_PARAMS +#define DMGL_PARAMS (1 << 0) /* Include function args */ +#define DMGL_ANSI (1 << 1) /* Include const, volatile, etc */ +#endif struct symbol { struct rb_node rb_node; @@ -13,6 +39,7 @@ struct symbol { u64 obj_start; u64 hist_sum; u64 *hist; + struct module *module; void *priv; char name[0]; }; @@ -22,11 +49,13 @@ struct dso { struct rb_root syms; struct symbol *(*find_symbol)(struct dso *, u64 ip); unsigned int sym_priv_size; - unsigned char prelinked; + unsigned char adjust_symbols; + unsigned char slen_calculated; + unsigned char origin; char name[0]; }; -const char *sym_hist_filter; +extern const char *sym_hist_filter; typedef int (*symbol_filter_t)(struct dso *self, struct symbol *sym); @@ -41,10 +70,23 @@ static inline void *dso__sym_priv(struct dso *self, struct symbol *sym) struct symbol *dso__find_symbol(struct dso *self, u64 ip); int dso__load_kernel(struct dso *self, const char *vmlinux, - symbol_filter_t filter, int verbose); + symbol_filter_t filter, int verbose, int modules); +int dso__load_modules(struct dso *self, symbol_filter_t filter, int verbose); int dso__load(struct dso *self, symbol_filter_t filter, int verbose); +struct dso *dsos__findnew(const char *name); +void dsos__fprintf(FILE *fp); size_t dso__fprintf(struct dso *self, FILE *fp); +char dso__symtab_origin(const struct dso *self); + +int load_kernel(void); void symbol__init(void); + +extern struct list_head dsos; +extern struct dso *kernel_dso; +extern struct dso *vdso; +extern struct dso *hypervisor_dso; +extern const char *vmlinux_name; +extern int modules; #endif /* _PERF_SYMBOL_ */ diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c new file mode 100644 index 00000000000..45efb5db0d1 --- /dev/null +++ b/tools/perf/util/thread.c @@ -0,0 +1,175 @@ +#include "../perf.h" +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include "thread.h" +#include "util.h" +#include "debug.h" + +static struct thread *thread__new(pid_t pid) +{ + struct thread *self = calloc(1, sizeof(*self)); + + if (self != NULL) { + self->pid = pid; + self->comm = malloc(32); + if (self->comm) + snprintf(self->comm, 32, ":%d", self->pid); + INIT_LIST_HEAD(&self->maps); + } + + return self; +} + +int thread__set_comm(struct thread *self, const char *comm) +{ + if (self->comm) + free(self->comm); + self->comm = strdup(comm); + return self->comm ? 0 : -ENOMEM; +} + +static size_t thread__fprintf(struct thread *self, FILE *fp) +{ + struct map *pos; + size_t ret = fprintf(fp, "Thread %d %s\n", self->pid, self->comm); + + list_for_each_entry(pos, &self->maps, node) + ret += map__fprintf(pos, fp); + + return ret; +} + +struct thread * +threads__findnew(pid_t pid, struct rb_root *threads, struct thread **last_match) +{ + struct rb_node **p = &threads->rb_node; + struct rb_node *parent = NULL; + struct thread *th; + + /* + * Font-end cache - PID lookups come in blocks, + * so most of the time we dont have to look up + * the full rbtree: + */ + if (*last_match && (*last_match)->pid == pid) + return *last_match; + + while (*p != NULL) { + parent = *p; + th = rb_entry(parent, struct thread, rb_node); + + if (th->pid == pid) { + *last_match = th; + return th; + } + + if (pid < th->pid) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + th = thread__new(pid); + if (th != NULL) { + rb_link_node(&th->rb_node, parent, p); + rb_insert_color(&th->rb_node, threads); + *last_match = th; + } + + return th; +} + +struct thread * +register_idle_thread(struct rb_root *threads, struct thread **last_match) +{ + struct thread *thread = threads__findnew(0, threads, last_match); + + if (!thread || thread__set_comm(thread, "swapper")) { + fprintf(stderr, "problem inserting idle task.\n"); + exit(-1); + } + + return thread; +} + +void thread__insert_map(struct thread *self, struct map *map) +{ + struct map *pos, *tmp; + + list_for_each_entry_safe(pos, tmp, &self->maps, node) { + if (map__overlap(pos, map)) { + if (verbose >= 2) { + printf("overlapping maps:\n"); + map__fprintf(map, stdout); + map__fprintf(pos, stdout); + } + + if (map->start <= pos->start && map->end > pos->start) + pos->start = map->end; + + if (map->end >= pos->end && map->start < pos->end) + pos->end = map->start; + + if (verbose >= 2) { + printf("after collision:\n"); + map__fprintf(pos, stdout); + } + + if (pos->start >= pos->end) { + list_del_init(&pos->node); + free(pos); + } + } + } + + list_add_tail(&map->node, &self->maps); +} + +int thread__fork(struct thread *self, struct thread *parent) +{ + struct map *map; + + if (self->comm) + free(self->comm); + self->comm = strdup(parent->comm); + if (!self->comm) + return -ENOMEM; + + list_for_each_entry(map, &parent->maps, node) { + struct map *new = map__clone(map); + if (!new) + return -ENOMEM; + thread__insert_map(self, new); + } + + return 0; +} + +struct map *thread__find_map(struct thread *self, u64 ip) +{ + struct map *pos; + + if (self == NULL) + return NULL; + + list_for_each_entry(pos, &self->maps, node) + if (ip >= pos->start && ip <= pos->end) + return pos; + + return NULL; +} + +size_t threads__fprintf(FILE *fp, struct rb_root *threads) +{ + size_t ret = 0; + struct rb_node *nd; + + for (nd = rb_first(threads); nd; nd = rb_next(nd)) { + struct thread *pos = rb_entry(nd, struct thread, rb_node); + + ret += thread__fprintf(pos, fp); + } + + return ret; +} diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h new file mode 100644 index 00000000000..32aea3c1c2a --- /dev/null +++ b/tools/perf/util/thread.h @@ -0,0 +1,22 @@ +#include <linux/rbtree.h> +#include <linux/list.h> +#include <unistd.h> +#include "symbol.h" + +struct thread { + struct rb_node rb_node; + struct list_head maps; + pid_t pid; + char shortname[3]; + char *comm; +}; + +int thread__set_comm(struct thread *self, const char *comm); +struct thread * +threads__findnew(pid_t pid, struct rb_root *threads, struct thread **last_match); +struct thread * +register_idle_thread(struct rb_root *threads, struct thread **last_match); +void thread__insert_map(struct thread *self, struct map *map); +int thread__fork(struct thread *self, struct thread *parent); +struct map *thread__find_map(struct thread *self, u64 ip); +size_t threads__fprintf(FILE *fp, struct rb_root *threads); diff --git a/tools/perf/util/trace-event-info.c b/tools/perf/util/trace-event-info.c new file mode 100644 index 00000000000..af4b0573b37 --- /dev/null +++ b/tools/perf/util/trace-event-info.c @@ -0,0 +1,540 @@ +/* + * Copyright (C) 2008,2009, Steven Rostedt <srostedt@redhat.com> + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; version 2 of the License (not later!) + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + */ +#define _GNU_SOURCE +#include <dirent.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdarg.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <sys/wait.h> +#include <pthread.h> +#include <fcntl.h> +#include <unistd.h> +#include <ctype.h> +#include <errno.h> +#include <stdbool.h> + +#include "../perf.h" +#include "trace-event.h" + + +#define VERSION "0.5" + +#define _STR(x) #x +#define STR(x) _STR(x) +#define MAX_PATH 256 + +#define TRACE_CTRL "tracing_on" +#define TRACE "trace" +#define AVAILABLE "available_tracers" +#define CURRENT "current_tracer" +#define ITER_CTRL "trace_options" +#define MAX_LATENCY "tracing_max_latency" + +unsigned int page_size; + +static const char *output_file = "trace.info"; +static int output_fd; + +struct event_list { + struct event_list *next; + const char *event; +}; + +struct events { + struct events *sibling; + struct events *children; + struct events *next; + char *name; +}; + + + +static void die(const char *fmt, ...) +{ + va_list ap; + int ret = errno; + + if (errno) + perror("trace-cmd"); + else + ret = -1; + + va_start(ap, fmt); + fprintf(stderr, " "); + vfprintf(stderr, fmt, ap); + va_end(ap); + + fprintf(stderr, "\n"); + exit(ret); +} + +void *malloc_or_die(unsigned int size) +{ + void *data; + + data = malloc(size); + if (!data) + die("malloc"); + return data; +} + +static const char *find_debugfs(void) +{ + static char debugfs[MAX_PATH+1]; + static int debugfs_found; + char type[100]; + FILE *fp; + + if (debugfs_found) + return debugfs; + + if ((fp = fopen("/proc/mounts","r")) == NULL) + die("Can't open /proc/mounts for read"); + + while (fscanf(fp, "%*s %" + STR(MAX_PATH) + "s %99s %*s %*d %*d\n", + debugfs, type) == 2) { + if (strcmp(type, "debugfs") == 0) + break; + } + fclose(fp); + + if (strcmp(type, "debugfs") != 0) + die("debugfs not mounted, please mount"); + + debugfs_found = 1; + + return debugfs; +} + +/* + * Finds the path to the debugfs/tracing + * Allocates the string and stores it. + */ +static const char *find_tracing_dir(void) +{ + static char *tracing; + static int tracing_found; + const char *debugfs; + + if (tracing_found) + return tracing; + + debugfs = find_debugfs(); + + tracing = malloc_or_die(strlen(debugfs) + 9); + + sprintf(tracing, "%s/tracing", debugfs); + + tracing_found = 1; + return tracing; +} + +static char *get_tracing_file(const char *name) +{ + const char *tracing; + char *file; + + tracing = find_tracing_dir(); + if (!tracing) + return NULL; + + file = malloc_or_die(strlen(tracing) + strlen(name) + 2); + + sprintf(file, "%s/%s", tracing, name); + return file; +} + +static void put_tracing_file(char *file) +{ + free(file); +} + +static ssize_t write_or_die(const void *buf, size_t len) +{ + int ret; + + ret = write(output_fd, buf, len); + if (ret < 0) + die("writing to '%s'", output_file); + + return ret; +} + +int bigendian(void) +{ + unsigned char str[] = { 0x1, 0x2, 0x3, 0x4, 0x0, 0x0, 0x0, 0x0}; + unsigned int *ptr; + + ptr = (unsigned int *)(void *)str; + return *ptr == 0x01020304; +} + +static unsigned long long copy_file_fd(int fd) +{ + unsigned long long size = 0; + char buf[BUFSIZ]; + int r; + + do { + r = read(fd, buf, BUFSIZ); + if (r > 0) { + size += r; + write_or_die(buf, r); + } + } while (r > 0); + + return size; +} + +static unsigned long long copy_file(const char *file) +{ + unsigned long long size = 0; + int fd; + + fd = open(file, O_RDONLY); + if (fd < 0) + die("Can't read '%s'", file); + size = copy_file_fd(fd); + close(fd); + + return size; +} + +static unsigned long get_size_fd(int fd) +{ + unsigned long long size = 0; + char buf[BUFSIZ]; + int r; + + do { + r = read(fd, buf, BUFSIZ); + if (r > 0) + size += r; + } while (r > 0); + + lseek(fd, 0, SEEK_SET); + + return size; +} + +static unsigned long get_size(const char *file) +{ + unsigned long long size = 0; + int fd; + + fd = open(file, O_RDONLY); + if (fd < 0) + die("Can't read '%s'", file); + size = get_size_fd(fd); + close(fd); + + return size; +} + +static void read_header_files(void) +{ + unsigned long long size, check_size; + char *path; + int fd; + + path = get_tracing_file("events/header_page"); + fd = open(path, O_RDONLY); + if (fd < 0) + die("can't read '%s'", path); + + /* unfortunately, you can not stat debugfs files for size */ + size = get_size_fd(fd); + + write_or_die("header_page", 12); + write_or_die(&size, 8); + check_size = copy_file_fd(fd); + if (size != check_size) + die("wrong size for '%s' size=%lld read=%lld", + path, size, check_size); + put_tracing_file(path); + + path = get_tracing_file("events/header_event"); + fd = open(path, O_RDONLY); + if (fd < 0) + die("can't read '%s'", path); + + size = get_size_fd(fd); + + write_or_die("header_event", 13); + write_or_die(&size, 8); + check_size = copy_file_fd(fd); + if (size != check_size) + die("wrong size for '%s'", path); + put_tracing_file(path); +} + +static bool name_in_tp_list(char *sys, struct tracepoint_path *tps) +{ + while (tps) { + if (!strcmp(sys, tps->name)) + return true; + tps = tps->next; + } + + return false; +} + +static void copy_event_system(const char *sys, struct tracepoint_path *tps) +{ + unsigned long long size, check_size; + struct dirent *dent; + struct stat st; + char *format; + DIR *dir; + int count = 0; + int ret; + + dir = opendir(sys); + if (!dir) + die("can't read directory '%s'", sys); + + while ((dent = readdir(dir))) { + if (strcmp(dent->d_name, ".") == 0 || + strcmp(dent->d_name, "..") == 0 || + !name_in_tp_list(dent->d_name, tps)) + continue; + format = malloc_or_die(strlen(sys) + strlen(dent->d_name) + 10); + sprintf(format, "%s/%s/format", sys, dent->d_name); + ret = stat(format, &st); + free(format); + if (ret < 0) + continue; + count++; + } + + write_or_die(&count, 4); + + rewinddir(dir); + while ((dent = readdir(dir))) { + if (strcmp(dent->d_name, ".") == 0 || + strcmp(dent->d_name, "..") == 0 || + !name_in_tp_list(dent->d_name, tps)) + continue; + format = malloc_or_die(strlen(sys) + strlen(dent->d_name) + 10); + sprintf(format, "%s/%s/format", sys, dent->d_name); + ret = stat(format, &st); + + if (ret >= 0) { + /* unfortunately, you can not stat debugfs files for size */ + size = get_size(format); + write_or_die(&size, 8); + check_size = copy_file(format); + if (size != check_size) + die("error in size of file '%s'", format); + } + + free(format); + } +} + +static void read_ftrace_files(struct tracepoint_path *tps) +{ + char *path; + + path = get_tracing_file("events/ftrace"); + + copy_event_system(path, tps); + + put_tracing_file(path); +} + +static bool system_in_tp_list(char *sys, struct tracepoint_path *tps) +{ + while (tps) { + if (!strcmp(sys, tps->system)) + return true; + tps = tps->next; + } + + return false; +} + +static void read_event_files(struct tracepoint_path *tps) +{ + struct dirent *dent; + struct stat st; + char *path; + char *sys; + DIR *dir; + int count = 0; + int ret; + + path = get_tracing_file("events"); + + dir = opendir(path); + if (!dir) + die("can't read directory '%s'", path); + + while ((dent = readdir(dir))) { + if (strcmp(dent->d_name, ".") == 0 || + strcmp(dent->d_name, "..") == 0 || + strcmp(dent->d_name, "ftrace") == 0 || + !system_in_tp_list(dent->d_name, tps)) + continue; + sys = malloc_or_die(strlen(path) + strlen(dent->d_name) + 2); + sprintf(sys, "%s/%s", path, dent->d_name); + ret = stat(sys, &st); + free(sys); + if (ret < 0) + continue; + if (S_ISDIR(st.st_mode)) + count++; + } + + write_or_die(&count, 4); + + rewinddir(dir); + while ((dent = readdir(dir))) { + if (strcmp(dent->d_name, ".") == 0 || + strcmp(dent->d_name, "..") == 0 || + strcmp(dent->d_name, "ftrace") == 0 || + !system_in_tp_list(dent->d_name, tps)) + continue; + sys = malloc_or_die(strlen(path) + strlen(dent->d_name) + 2); + sprintf(sys, "%s/%s", path, dent->d_name); + ret = stat(sys, &st); + if (ret >= 0) { + if (S_ISDIR(st.st_mode)) { + write_or_die(dent->d_name, strlen(dent->d_name) + 1); + copy_event_system(sys, tps); + } + } + free(sys); + } + + put_tracing_file(path); +} + +static void read_proc_kallsyms(void) +{ + unsigned int size, check_size; + const char *path = "/proc/kallsyms"; + struct stat st; + int ret; + + ret = stat(path, &st); + if (ret < 0) { + /* not found */ + size = 0; + write_or_die(&size, 4); + return; + } + size = get_size(path); + write_or_die(&size, 4); + check_size = copy_file(path); + if (size != check_size) + die("error in size of file '%s'", path); + +} + +static void read_ftrace_printk(void) +{ + unsigned int size, check_size; + char *path; + struct stat st; + int ret; + + path = get_tracing_file("printk_formats"); + ret = stat(path, &st); + if (ret < 0) { + /* not found */ + size = 0; + write_or_die(&size, 4); + goto out; + } + size = get_size(path); + write_or_die(&size, 4); + check_size = copy_file(path); + if (size != check_size) + die("error in size of file '%s'", path); +out: + put_tracing_file(path); +} + +static struct tracepoint_path * +get_tracepoints_path(struct perf_event_attr *pattrs, int nb_events) +{ + struct tracepoint_path path, *ppath = &path; + int i; + + for (i = 0; i < nb_events; i++) { + if (pattrs[i].type != PERF_TYPE_TRACEPOINT) + continue; + ppath->next = tracepoint_id_to_path(pattrs[i].config); + if (!ppath->next) + die("%s\n", "No memory to alloc tracepoints list"); + ppath = ppath->next; + } + + return path.next; +} +void read_tracing_data(struct perf_event_attr *pattrs, int nb_events) +{ + char buf[BUFSIZ]; + struct tracepoint_path *tps; + + output_fd = open(output_file, O_WRONLY | O_CREAT | O_TRUNC | O_LARGEFILE, 0644); + if (output_fd < 0) + die("creating file '%s'", output_file); + + buf[0] = 23; + buf[1] = 8; + buf[2] = 68; + memcpy(buf + 3, "tracing", 7); + + write_or_die(buf, 10); + + write_or_die(VERSION, strlen(VERSION) + 1); + + /* save endian */ + if (bigendian()) + buf[0] = 1; + else + buf[0] = 0; + + write_or_die(buf, 1); + + /* save size of long */ + buf[0] = sizeof(long); + write_or_die(buf, 1); + + /* save page_size */ + page_size = getpagesize(); + write_or_die(&page_size, 4); + + tps = get_tracepoints_path(pattrs, nb_events); + + read_header_files(); + read_ftrace_files(tps); + read_event_files(tps); + read_proc_kallsyms(); + read_ftrace_printk(); +} diff --git a/tools/perf/util/trace-event-parse.c b/tools/perf/util/trace-event-parse.c new file mode 100644 index 00000000000..f6a8437141c --- /dev/null +++ b/tools/perf/util/trace-event-parse.c @@ -0,0 +1,2967 @@ +/* + * Copyright (C) 2009, Steven Rostedt <srostedt@redhat.com> + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; version 2 of the License (not later!) + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * The parts for function graph printing was taken and modified from the + * Linux Kernel that were written by Frederic Weisbecker. + */ +#define _GNU_SOURCE +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <ctype.h> +#include <errno.h> + +#undef _GNU_SOURCE +#include "../perf.h" +#include "util.h" +#include "trace-event.h" + +int header_page_ts_offset; +int header_page_ts_size; +int header_page_size_offset; +int header_page_size_size; +int header_page_data_offset; +int header_page_data_size; + +static char *input_buf; +static unsigned long long input_buf_ptr; +static unsigned long long input_buf_siz; + +static int cpus; +static int long_size; + +static void init_input_buf(char *buf, unsigned long long size) +{ + input_buf = buf; + input_buf_siz = size; + input_buf_ptr = 0; +} + +struct cmdline { + char *comm; + int pid; +}; + +static struct cmdline *cmdlines; +static int cmdline_count; + +static int cmdline_cmp(const void *a, const void *b) +{ + const struct cmdline *ca = a; + const struct cmdline *cb = b; + + if (ca->pid < cb->pid) + return -1; + if (ca->pid > cb->pid) + return 1; + + return 0; +} + +void parse_cmdlines(char *file, int size __unused) +{ + struct cmdline_list { + struct cmdline_list *next; + char *comm; + int pid; + } *list = NULL, *item; + char *line; + char *next = NULL; + int i; + + line = strtok_r(file, "\n", &next); + while (line) { + item = malloc_or_die(sizeof(*item)); + sscanf(line, "%d %as", &item->pid, + (float *)(void *)&item->comm); /* workaround gcc warning */ + item->next = list; + list = item; + line = strtok_r(NULL, "\n", &next); + cmdline_count++; + } + + cmdlines = malloc_or_die(sizeof(*cmdlines) * cmdline_count); + + i = 0; + while (list) { + cmdlines[i].pid = list->pid; + cmdlines[i].comm = list->comm; + i++; + item = list; + list = list->next; + free(item); + } + + qsort(cmdlines, cmdline_count, sizeof(*cmdlines), cmdline_cmp); +} + +static struct func_map { + unsigned long long addr; + char *func; + char *mod; +} *func_list; +static unsigned int func_count; + +static int func_cmp(const void *a, const void *b) +{ + const struct func_map *fa = a; + const struct func_map *fb = b; + + if (fa->addr < fb->addr) + return -1; + if (fa->addr > fb->addr) + return 1; + + return 0; +} + +void parse_proc_kallsyms(char *file, unsigned int size __unused) +{ + struct func_list { + struct func_list *next; + unsigned long long addr; + char *func; + char *mod; + } *list = NULL, *item; + char *line; + char *next = NULL; + char *addr_str; + char ch; + int ret; + int i; + + line = strtok_r(file, "\n", &next); + while (line) { + item = malloc_or_die(sizeof(*item)); + item->mod = NULL; + ret = sscanf(line, "%as %c %as\t[%as", + (float *)(void *)&addr_str, /* workaround gcc warning */ + &ch, + (float *)(void *)&item->func, + (float *)(void *)&item->mod); + item->addr = strtoull(addr_str, NULL, 16); + free(addr_str); + + /* truncate the extra ']' */ + if (item->mod) + item->mod[strlen(item->mod) - 1] = 0; + + + item->next = list; + list = item; + line = strtok_r(NULL, "\n", &next); + func_count++; + } + + func_list = malloc_or_die(sizeof(*func_list) * func_count + 1); + + i = 0; + while (list) { + func_list[i].func = list->func; + func_list[i].addr = list->addr; + func_list[i].mod = list->mod; + i++; + item = list; + list = list->next; + free(item); + } + + qsort(func_list, func_count, sizeof(*func_list), func_cmp); + + /* + * Add a special record at the end. + */ + func_list[func_count].func = NULL; + func_list[func_count].addr = 0; + func_list[func_count].mod = NULL; +} + +/* + * We are searching for a record in between, not an exact + * match. + */ +static int func_bcmp(const void *a, const void *b) +{ + const struct func_map *fa = a; + const struct func_map *fb = b; + + if ((fa->addr == fb->addr) || + + (fa->addr > fb->addr && + fa->addr < (fb+1)->addr)) + return 0; + + if (fa->addr < fb->addr) + return -1; + + return 1; +} + +static struct func_map *find_func(unsigned long long addr) +{ + struct func_map *func; + struct func_map key; + + key.addr = addr; + + func = bsearch(&key, func_list, func_count, sizeof(*func_list), + func_bcmp); + + return func; +} + +void print_funcs(void) +{ + int i; + + for (i = 0; i < (int)func_count; i++) { + printf("%016llx %s", + func_list[i].addr, + func_list[i].func); + if (func_list[i].mod) + printf(" [%s]\n", func_list[i].mod); + else + printf("\n"); + } +} + +static struct printk_map { + unsigned long long addr; + char *printk; +} *printk_list; +static unsigned int printk_count; + +static int printk_cmp(const void *a, const void *b) +{ + const struct func_map *fa = a; + const struct func_map *fb = b; + + if (fa->addr < fb->addr) + return -1; + if (fa->addr > fb->addr) + return 1; + + return 0; +} + +static struct printk_map *find_printk(unsigned long long addr) +{ + struct printk_map *printk; + struct printk_map key; + + key.addr = addr; + + printk = bsearch(&key, printk_list, printk_count, sizeof(*printk_list), + printk_cmp); + + return printk; +} + +void parse_ftrace_printk(char *file, unsigned int size __unused) +{ + struct printk_list { + struct printk_list *next; + unsigned long long addr; + char *printk; + } *list = NULL, *item; + char *line; + char *next = NULL; + char *addr_str; + int ret; + int i; + + line = strtok_r(file, "\n", &next); + while (line) { + item = malloc_or_die(sizeof(*item)); + ret = sscanf(line, "%as : %as", + (float *)(void *)&addr_str, /* workaround gcc warning */ + (float *)(void *)&item->printk); + item->addr = strtoull(addr_str, NULL, 16); + free(addr_str); + + item->next = list; + list = item; + line = strtok_r(NULL, "\n", &next); + printk_count++; + } + + printk_list = malloc_or_die(sizeof(*printk_list) * printk_count + 1); + + i = 0; + while (list) { + printk_list[i].printk = list->printk; + printk_list[i].addr = list->addr; + i++; + item = list; + list = list->next; + free(item); + } + + qsort(printk_list, printk_count, sizeof(*printk_list), printk_cmp); +} + +void print_printk(void) +{ + int i; + + for (i = 0; i < (int)printk_count; i++) { + printf("%016llx %s\n", + printk_list[i].addr, + printk_list[i].printk); + } +} + +static struct event *alloc_event(void) +{ + struct event *event; + + event = malloc_or_die(sizeof(*event)); + memset(event, 0, sizeof(*event)); + + return event; +} + +enum event_type { + EVENT_ERROR, + EVENT_NONE, + EVENT_SPACE, + EVENT_NEWLINE, + EVENT_OP, + EVENT_DELIM, + EVENT_ITEM, + EVENT_DQUOTE, + EVENT_SQUOTE, +}; + +static struct event *event_list; + +static void add_event(struct event *event) +{ + event->next = event_list; + event_list = event; +} + +static int event_item_type(enum event_type type) +{ + switch (type) { + case EVENT_ITEM ... EVENT_SQUOTE: + return 1; + case EVENT_ERROR ... EVENT_DELIM: + default: + return 0; + } +} + +static void free_arg(struct print_arg *arg) +{ + if (!arg) + return; + + switch (arg->type) { + case PRINT_ATOM: + if (arg->atom.atom) + free(arg->atom.atom); + break; + case PRINT_NULL: + case PRINT_FIELD ... PRINT_OP: + default: + /* todo */ + break; + } + + free(arg); +} + +static enum event_type get_type(int ch) +{ + if (ch == '\n') + return EVENT_NEWLINE; + if (isspace(ch)) + return EVENT_SPACE; + if (isalnum(ch) || ch == '_') + return EVENT_ITEM; + if (ch == '\'') + return EVENT_SQUOTE; + if (ch == '"') + return EVENT_DQUOTE; + if (!isprint(ch)) + return EVENT_NONE; + if (ch == '(' || ch == ')' || ch == ',') + return EVENT_DELIM; + + return EVENT_OP; +} + +static int __read_char(void) +{ + if (input_buf_ptr >= input_buf_siz) + return -1; + + return input_buf[input_buf_ptr++]; +} + +static int __peek_char(void) +{ + if (input_buf_ptr >= input_buf_siz) + return -1; + + return input_buf[input_buf_ptr]; +} + +static enum event_type __read_token(char **tok) +{ + char buf[BUFSIZ]; + int ch, last_ch, quote_ch, next_ch; + int i = 0; + int tok_size = 0; + enum event_type type; + + *tok = NULL; + + + ch = __read_char(); + if (ch < 0) + return EVENT_NONE; + + type = get_type(ch); + if (type == EVENT_NONE) + return type; + + buf[i++] = ch; + + switch (type) { + case EVENT_NEWLINE: + case EVENT_DELIM: + *tok = malloc_or_die(2); + (*tok)[0] = ch; + (*tok)[1] = 0; + return type; + + case EVENT_OP: + switch (ch) { + case '-': + next_ch = __peek_char(); + if (next_ch == '>') { + buf[i++] = __read_char(); + break; + } + /* fall through */ + case '+': + case '|': + case '&': + case '>': + case '<': + last_ch = ch; + ch = __peek_char(); + if (ch != last_ch) + goto test_equal; + buf[i++] = __read_char(); + switch (last_ch) { + case '>': + case '<': + goto test_equal; + default: + break; + } + break; + case '!': + case '=': + goto test_equal; + default: /* what should we do instead? */ + break; + } + buf[i] = 0; + *tok = strdup(buf); + return type; + + test_equal: + ch = __peek_char(); + if (ch == '=') + buf[i++] = __read_char(); + break; + + case EVENT_DQUOTE: + case EVENT_SQUOTE: + /* don't keep quotes */ + i--; + quote_ch = ch; + last_ch = 0; + do { + if (i == (BUFSIZ - 1)) { + buf[i] = 0; + if (*tok) { + *tok = realloc(*tok, tok_size + BUFSIZ); + if (!*tok) + return EVENT_NONE; + strcat(*tok, buf); + } else + *tok = strdup(buf); + + if (!*tok) + return EVENT_NONE; + tok_size += BUFSIZ; + i = 0; + } + last_ch = ch; + ch = __read_char(); + buf[i++] = ch; + } while (ch != quote_ch && last_ch != '\\'); + /* remove the last quote */ + i--; + goto out; + + case EVENT_ERROR ... EVENT_SPACE: + case EVENT_ITEM: + default: + break; + } + + while (get_type(__peek_char()) == type) { + if (i == (BUFSIZ - 1)) { + buf[i] = 0; + if (*tok) { + *tok = realloc(*tok, tok_size + BUFSIZ); + if (!*tok) + return EVENT_NONE; + strcat(*tok, buf); + } else + *tok = strdup(buf); + + if (!*tok) + return EVENT_NONE; + tok_size += BUFSIZ; + i = 0; + } + ch = __read_char(); + buf[i++] = ch; + } + + out: + buf[i] = 0; + if (*tok) { + *tok = realloc(*tok, tok_size + i); + if (!*tok) + return EVENT_NONE; + strcat(*tok, buf); + } else + *tok = strdup(buf); + if (!*tok) + return EVENT_NONE; + + return type; +} + +static void free_token(char *tok) +{ + if (tok) + free(tok); +} + +static enum event_type read_token(char **tok) +{ + enum event_type type; + + for (;;) { + type = __read_token(tok); + if (type != EVENT_SPACE) + return type; + + free_token(*tok); + } + + /* not reached */ + return EVENT_NONE; +} + +/* no newline */ +static enum event_type read_token_item(char **tok) +{ + enum event_type type; + + for (;;) { + type = __read_token(tok); + if (type != EVENT_SPACE && type != EVENT_NEWLINE) + return type; + + free_token(*tok); + } + + /* not reached */ + return EVENT_NONE; +} + +static int test_type(enum event_type type, enum event_type expect) +{ + if (type != expect) { + die("Error: expected type %d but read %d", + expect, type); + return -1; + } + return 0; +} + +static int test_type_token(enum event_type type, char *token, + enum event_type expect, char *expect_tok) +{ + if (type != expect) { + die("Error: expected type %d but read %d", + expect, type); + return -1; + } + + if (strcmp(token, expect_tok) != 0) { + die("Error: expected '%s' but read '%s'", + expect_tok, token); + return -1; + } + return 0; +} + +static int __read_expect_type(enum event_type expect, char **tok, int newline_ok) +{ + enum event_type type; + + if (newline_ok) + type = read_token(tok); + else + type = read_token_item(tok); + return test_type(type, expect); +} + +static int read_expect_type(enum event_type expect, char **tok) +{ + return __read_expect_type(expect, tok, 1); +} + +static int __read_expected(enum event_type expect, char *str, int newline_ok) +{ + enum event_type type; + char *token; + int ret; + + if (newline_ok) + type = read_token(&token); + else + type = read_token_item(&token); + + ret = test_type_token(type, token, expect, str); + + free_token(token); + + return 0; +} + +static int read_expected(enum event_type expect, char *str) +{ + return __read_expected(expect, str, 1); +} + +static int read_expected_item(enum event_type expect, char *str) +{ + return __read_expected(expect, str, 0); +} + +static char *event_read_name(void) +{ + char *token; + + if (read_expected(EVENT_ITEM, (char *)"name") < 0) + return NULL; + + if (read_expected(EVENT_OP, (char *)":") < 0) + return NULL; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + + return token; + + fail: + free_token(token); + return NULL; +} + +static int event_read_id(void) +{ + char *token; + int id; + + if (read_expected_item(EVENT_ITEM, (char *)"ID") < 0) + return -1; + + if (read_expected(EVENT_OP, (char *)":") < 0) + return -1; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + + id = strtoul(token, NULL, 0); + free_token(token); + return id; + + fail: + free_token(token); + return -1; +} + +static int event_read_fields(struct event *event, struct format_field **fields) +{ + struct format_field *field = NULL; + enum event_type type; + char *token; + char *last_token; + int count = 0; + + do { + type = read_token(&token); + if (type == EVENT_NEWLINE) { + free_token(token); + return count; + } + + count++; + + if (test_type_token(type, token, EVENT_ITEM, (char *)"field")) + goto fail; + free_token(token); + + type = read_token(&token); + /* + * The ftrace fields may still use the "special" name. + * Just ignore it. + */ + if (event->flags & EVENT_FL_ISFTRACE && + type == EVENT_ITEM && strcmp(token, "special") == 0) { + free_token(token); + type = read_token(&token); + } + + if (test_type_token(type, token, EVENT_OP, (char *)":") < 0) + return -1; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + + last_token = token; + + field = malloc_or_die(sizeof(*field)); + memset(field, 0, sizeof(*field)); + + /* read the rest of the type */ + for (;;) { + type = read_token(&token); + if (type == EVENT_ITEM || + (type == EVENT_OP && strcmp(token, "*") == 0) || + /* + * Some of the ftrace fields are broken and have + * an illegal "." in them. + */ + (event->flags & EVENT_FL_ISFTRACE && + type == EVENT_OP && strcmp(token, ".") == 0)) { + + if (strcmp(token, "*") == 0) + field->flags |= FIELD_IS_POINTER; + + if (field->type) { + field->type = realloc(field->type, + strlen(field->type) + + strlen(last_token) + 2); + strcat(field->type, " "); + strcat(field->type, last_token); + } else + field->type = last_token; + last_token = token; + continue; + } + + break; + } + + if (!field->type) { + die("no type found"); + goto fail; + } + field->name = last_token; + + if (test_type(type, EVENT_OP)) + goto fail; + + if (strcmp(token, "[") == 0) { + enum event_type last_type = type; + char *brackets = token; + int len; + + field->flags |= FIELD_IS_ARRAY; + + type = read_token(&token); + while (strcmp(token, "]") != 0) { + if (last_type == EVENT_ITEM && + type == EVENT_ITEM) + len = 2; + else + len = 1; + last_type = type; + + brackets = realloc(brackets, + strlen(brackets) + + strlen(token) + len); + if (len == 2) + strcat(brackets, " "); + strcat(brackets, token); + free_token(token); + type = read_token(&token); + if (type == EVENT_NONE) { + die("failed to find token"); + goto fail; + } + } + + free_token(token); + + brackets = realloc(brackets, strlen(brackets) + 2); + strcat(brackets, "]"); + + /* add brackets to type */ + + type = read_token(&token); + /* + * If the next token is not an OP, then it is of + * the format: type [] item; + */ + if (type == EVENT_ITEM) { + field->type = realloc(field->type, + strlen(field->type) + + strlen(field->name) + + strlen(brackets) + 2); + strcat(field->type, " "); + strcat(field->type, field->name); + free_token(field->name); + strcat(field->type, brackets); + field->name = token; + type = read_token(&token); + } else { + field->type = realloc(field->type, + strlen(field->type) + + strlen(brackets) + 1); + strcat(field->type, brackets); + } + free(brackets); + } + + if (test_type_token(type, token, EVENT_OP, (char *)";")) + goto fail; + free_token(token); + + if (read_expected(EVENT_ITEM, (char *)"offset") < 0) + goto fail_expect; + + if (read_expected(EVENT_OP, (char *)":") < 0) + goto fail_expect; + + if (read_expect_type(EVENT_ITEM, &token)) + goto fail; + field->offset = strtoul(token, NULL, 0); + free_token(token); + + if (read_expected(EVENT_OP, (char *)";") < 0) + goto fail_expect; + + if (read_expected(EVENT_ITEM, (char *)"size") < 0) + goto fail_expect; + + if (read_expected(EVENT_OP, (char *)":") < 0) + goto fail_expect; + + if (read_expect_type(EVENT_ITEM, &token)) + goto fail; + field->size = strtoul(token, NULL, 0); + free_token(token); + + if (read_expected(EVENT_OP, (char *)";") < 0) + goto fail_expect; + + if (read_expect_type(EVENT_NEWLINE, &token) < 0) + goto fail; + free_token(token); + + *fields = field; + fields = &field->next; + + } while (1); + + return 0; + +fail: + free_token(token); +fail_expect: + if (field) + free(field); + return -1; +} + +static int event_read_format(struct event *event) +{ + char *token; + int ret; + + if (read_expected_item(EVENT_ITEM, (char *)"format") < 0) + return -1; + + if (read_expected(EVENT_OP, (char *)":") < 0) + return -1; + + if (read_expect_type(EVENT_NEWLINE, &token)) + goto fail; + free_token(token); + + ret = event_read_fields(event, &event->format.common_fields); + if (ret < 0) + return ret; + event->format.nr_common = ret; + + ret = event_read_fields(event, &event->format.fields); + if (ret < 0) + return ret; + event->format.nr_fields = ret; + + return 0; + + fail: + free_token(token); + return -1; +} + +enum event_type +process_arg_token(struct event *event, struct print_arg *arg, + char **tok, enum event_type type); + +static enum event_type +process_arg(struct event *event, struct print_arg *arg, char **tok) +{ + enum event_type type; + char *token; + + type = read_token(&token); + *tok = token; + + return process_arg_token(event, arg, tok, type); +} + +static enum event_type +process_cond(struct event *event, struct print_arg *top, char **tok) +{ + struct print_arg *arg, *left, *right; + enum event_type type; + char *token = NULL; + + arg = malloc_or_die(sizeof(*arg)); + memset(arg, 0, sizeof(*arg)); + + left = malloc_or_die(sizeof(*left)); + + right = malloc_or_die(sizeof(*right)); + + arg->type = PRINT_OP; + arg->op.left = left; + arg->op.right = right; + + *tok = NULL; + type = process_arg(event, left, &token); + if (test_type_token(type, token, EVENT_OP, (char *)":")) + goto out_free; + + arg->op.op = token; + + type = process_arg(event, right, &token); + + top->op.right = arg; + + *tok = token; + return type; + +out_free: + free_token(*tok); + free(right); + free(left); + free_arg(arg); + return EVENT_ERROR; +} + +static int get_op_prio(char *op) +{ + if (!op[1]) { + switch (op[0]) { + case '*': + case '/': + case '%': + return 6; + case '+': + case '-': + return 7; + /* '>>' and '<<' are 8 */ + case '<': + case '>': + return 9; + /* '==' and '!=' are 10 */ + case '&': + return 11; + case '^': + return 12; + case '|': + return 13; + case '?': + return 16; + default: + die("unknown op '%c'", op[0]); + return -1; + } + } else { + if (strcmp(op, "++") == 0 || + strcmp(op, "--") == 0) { + return 3; + } else if (strcmp(op, ">>") == 0 || + strcmp(op, "<<") == 0) { + return 8; + } else if (strcmp(op, ">=") == 0 || + strcmp(op, "<=") == 0) { + return 9; + } else if (strcmp(op, "==") == 0 || + strcmp(op, "!=") == 0) { + return 10; + } else if (strcmp(op, "&&") == 0) { + return 14; + } else if (strcmp(op, "||") == 0) { + return 15; + } else { + die("unknown op '%s'", op); + return -1; + } + } +} + +static void set_op_prio(struct print_arg *arg) +{ + + /* single ops are the greatest */ + if (!arg->op.left || arg->op.left->type == PRINT_NULL) { + arg->op.prio = 0; + return; + } + + arg->op.prio = get_op_prio(arg->op.op); +} + +static enum event_type +process_op(struct event *event, struct print_arg *arg, char **tok) +{ + struct print_arg *left, *right = NULL; + enum event_type type; + char *token; + + /* the op is passed in via tok */ + token = *tok; + + if (arg->type == PRINT_OP && !arg->op.left) { + /* handle single op */ + if (token[1]) { + die("bad op token %s", token); + return EVENT_ERROR; + } + switch (token[0]) { + case '!': + case '+': + case '-': + break; + default: + die("bad op token %s", token); + return EVENT_ERROR; + } + + /* make an empty left */ + left = malloc_or_die(sizeof(*left)); + left->type = PRINT_NULL; + arg->op.left = left; + + right = malloc_or_die(sizeof(*right)); + arg->op.right = right; + + type = process_arg(event, right, tok); + + } else if (strcmp(token, "?") == 0) { + + left = malloc_or_die(sizeof(*left)); + /* copy the top arg to the left */ + *left = *arg; + + arg->type = PRINT_OP; + arg->op.op = token; + arg->op.left = left; + arg->op.prio = 0; + + type = process_cond(event, arg, tok); + + } else if (strcmp(token, ">>") == 0 || + strcmp(token, "<<") == 0 || + strcmp(token, "&") == 0 || + strcmp(token, "|") == 0 || + strcmp(token, "&&") == 0 || + strcmp(token, "||") == 0 || + strcmp(token, "-") == 0 || + strcmp(token, "+") == 0 || + strcmp(token, "*") == 0 || + strcmp(token, "^") == 0 || + strcmp(token, "/") == 0 || + strcmp(token, "==") == 0 || + strcmp(token, "!=") == 0) { + + left = malloc_or_die(sizeof(*left)); + + /* copy the top arg to the left */ + *left = *arg; + + arg->type = PRINT_OP; + arg->op.op = token; + arg->op.left = left; + + set_op_prio(arg); + + right = malloc_or_die(sizeof(*right)); + + type = process_arg(event, right, tok); + + arg->op.right = right; + + } else { + die("unknown op '%s'", token); + /* the arg is now the left side */ + return EVENT_NONE; + } + + + if (type == EVENT_OP) { + int prio; + + /* higher prios need to be closer to the root */ + prio = get_op_prio(*tok); + + if (prio > arg->op.prio) + return process_op(event, arg, tok); + + return process_op(event, right, tok); + } + + return type; +} + +static enum event_type +process_entry(struct event *event __unused, struct print_arg *arg, + char **tok) +{ + enum event_type type; + char *field; + char *token; + + if (read_expected(EVENT_OP, (char *)"->") < 0) + return EVENT_ERROR; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + field = token; + + arg->type = PRINT_FIELD; + arg->field.name = field; + + type = read_token(&token); + *tok = token; + + return type; + +fail: + free_token(token); + return EVENT_ERROR; +} + +static char *arg_eval (struct print_arg *arg); + +static long long arg_num_eval(struct print_arg *arg) +{ + long long left, right; + long long val = 0; + + switch (arg->type) { + case PRINT_ATOM: + val = strtoll(arg->atom.atom, NULL, 0); + break; + case PRINT_TYPE: + val = arg_num_eval(arg->typecast.item); + break; + case PRINT_OP: + switch (arg->op.op[0]) { + case '|': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + if (arg->op.op[1]) + val = left || right; + else + val = left | right; + break; + case '&': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + if (arg->op.op[1]) + val = left && right; + else + val = left & right; + break; + case '<': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + switch (arg->op.op[1]) { + case 0: + val = left < right; + break; + case '<': + val = left << right; + break; + case '=': + val = left <= right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + case '>': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + switch (arg->op.op[1]) { + case 0: + val = left > right; + break; + case '>': + val = left >> right; + break; + case '=': + val = left >= right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + case '=': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + + if (arg->op.op[1] != '=') + die("unknown op '%s'", arg->op.op); + + val = left == right; + break; + case '!': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + + switch (arg->op.op[1]) { + case '=': + val = left != right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + + case PRINT_NULL: + case PRINT_FIELD ... PRINT_SYMBOL: + case PRINT_STRING: + default: + die("invalid eval type %d", arg->type); + + } + return val; +} + +static char *arg_eval (struct print_arg *arg) +{ + long long val; + static char buf[20]; + + switch (arg->type) { + case PRINT_ATOM: + return arg->atom.atom; + case PRINT_TYPE: + return arg_eval(arg->typecast.item); + case PRINT_OP: + val = arg_num_eval(arg); + sprintf(buf, "%lld", val); + return buf; + + case PRINT_NULL: + case PRINT_FIELD ... PRINT_SYMBOL: + case PRINT_STRING: + default: + die("invalid eval type %d", arg->type); + break; + } + + return NULL; +} + +static enum event_type +process_fields(struct event *event, struct print_flag_sym **list, char **tok) +{ + enum event_type type; + struct print_arg *arg = NULL; + struct print_flag_sym *field; + char *token = NULL; + char *value; + + do { + free_token(token); + type = read_token_item(&token); + if (test_type_token(type, token, EVENT_OP, (char *)"{")) + break; + + arg = malloc_or_die(sizeof(*arg)); + + free_token(token); + type = process_arg(event, arg, &token); + if (test_type_token(type, token, EVENT_DELIM, (char *)",")) + goto out_free; + + field = malloc_or_die(sizeof(*field)); + memset(field, 0, sizeof(field)); + + value = arg_eval(arg); + field->value = strdup(value); + + free_token(token); + type = process_arg(event, arg, &token); + if (test_type_token(type, token, EVENT_OP, (char *)"}")) + goto out_free; + + value = arg_eval(arg); + field->str = strdup(value); + free_arg(arg); + arg = NULL; + + *list = field; + list = &field->next; + + free_token(token); + type = read_token_item(&token); + } while (type == EVENT_DELIM && strcmp(token, ",") == 0); + + *tok = token; + return type; + +out_free: + free_arg(arg); + free_token(token); + + return EVENT_ERROR; +} + +static enum event_type +process_flags(struct event *event, struct print_arg *arg, char **tok) +{ + struct print_arg *field; + enum event_type type; + char *token; + + memset(arg, 0, sizeof(*arg)); + arg->type = PRINT_FLAGS; + + if (read_expected_item(EVENT_DELIM, (char *)"(") < 0) + return EVENT_ERROR; + + field = malloc_or_die(sizeof(*field)); + + type = process_arg(event, field, &token); + if (test_type_token(type, token, EVENT_DELIM, (char *)",")) + goto out_free; + + arg->flags.field = field; + + type = read_token_item(&token); + if (event_item_type(type)) { + arg->flags.delim = token; + type = read_token_item(&token); + } + + if (test_type_token(type, token, EVENT_DELIM, (char *)",")) + goto out_free; + + type = process_fields(event, &arg->flags.flags, &token); + if (test_type_token(type, token, EVENT_DELIM, (char *)")")) + goto out_free; + + free_token(token); + type = read_token_item(tok); + return type; + +out_free: + free_token(token); + return EVENT_ERROR; +} + +static enum event_type +process_symbols(struct event *event, struct print_arg *arg, char **tok) +{ + struct print_arg *field; + enum event_type type; + char *token; + + memset(arg, 0, sizeof(*arg)); + arg->type = PRINT_SYMBOL; + + if (read_expected_item(EVENT_DELIM, (char *)"(") < 0) + return EVENT_ERROR; + + field = malloc_or_die(sizeof(*field)); + + type = process_arg(event, field, &token); + if (test_type_token(type, token, EVENT_DELIM, (char *)",")) + goto out_free; + + arg->symbol.field = field; + + type = process_fields(event, &arg->symbol.symbols, &token); + if (test_type_token(type, token, EVENT_DELIM, (char *)")")) + goto out_free; + + free_token(token); + type = read_token_item(tok); + return type; + +out_free: + free_token(token); + return EVENT_ERROR; +} + +static enum event_type +process_paren(struct event *event, struct print_arg *arg, char **tok) +{ + struct print_arg *item_arg; + enum event_type type; + int ptr_cast = 0; + char *token; + + type = process_arg(event, arg, &token); + + if (type == EVENT_ERROR) + return EVENT_ERROR; + + if (type == EVENT_OP) { + /* handle the ptr casts */ + if (!strcmp(token, "*")) { + /* + * FIXME: should we zapp whitespaces before ')' ? + * (may require a peek_token_item()) + */ + if (__peek_char() == ')') { + ptr_cast = 1; + free_token(token); + type = read_token_item(&token); + } + } + if (!ptr_cast) { + type = process_op(event, arg, &token); + + if (type == EVENT_ERROR) + return EVENT_ERROR; + } + } + + if (test_type_token(type, token, EVENT_DELIM, (char *)")")) { + free_token(token); + return EVENT_ERROR; + } + + free_token(token); + type = read_token_item(&token); + + /* + * If the next token is an item or another open paren, then + * this was a typecast. + */ + if (event_item_type(type) || + (type == EVENT_DELIM && strcmp(token, "(") == 0)) { + + /* make this a typecast and contine */ + + /* prevous must be an atom */ + if (arg->type != PRINT_ATOM) + die("previous needed to be PRINT_ATOM"); + + item_arg = malloc_or_die(sizeof(*item_arg)); + + arg->type = PRINT_TYPE; + if (ptr_cast) { + char *old = arg->atom.atom; + + arg->atom.atom = malloc_or_die(strlen(old + 3)); + sprintf(arg->atom.atom, "%s *", old); + free(old); + } + arg->typecast.type = arg->atom.atom; + arg->typecast.item = item_arg; + type = process_arg_token(event, item_arg, &token, type); + + } + + *tok = token; + return type; +} + + +static enum event_type +process_str(struct event *event __unused, struct print_arg *arg, char **tok) +{ + enum event_type type; + char *token; + + if (read_expected(EVENT_DELIM, (char *)"(") < 0) + return EVENT_ERROR; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + + arg->type = PRINT_STRING; + arg->string.string = token; + arg->string.offset = -1; + + if (read_expected(EVENT_DELIM, (char *)")") < 0) + return EVENT_ERROR; + + type = read_token(&token); + *tok = token; + + return type; +fail: + free_token(token); + return EVENT_ERROR; +} + +enum event_type +process_arg_token(struct event *event, struct print_arg *arg, + char **tok, enum event_type type) +{ + char *token; + char *atom; + + token = *tok; + + switch (type) { + case EVENT_ITEM: + if (strcmp(token, "REC") == 0) { + free_token(token); + type = process_entry(event, arg, &token); + } else if (strcmp(token, "__print_flags") == 0) { + free_token(token); + type = process_flags(event, arg, &token); + } else if (strcmp(token, "__print_symbolic") == 0) { + free_token(token); + type = process_symbols(event, arg, &token); + } else if (strcmp(token, "__get_str") == 0) { + free_token(token); + type = process_str(event, arg, &token); + } else { + atom = token; + /* test the next token */ + type = read_token_item(&token); + + /* atoms can be more than one token long */ + while (type == EVENT_ITEM) { + atom = realloc(atom, strlen(atom) + strlen(token) + 2); + strcat(atom, " "); + strcat(atom, token); + free_token(token); + type = read_token_item(&token); + } + + /* todo, test for function */ + + arg->type = PRINT_ATOM; + arg->atom.atom = atom; + } + break; + case EVENT_DQUOTE: + case EVENT_SQUOTE: + arg->type = PRINT_ATOM; + arg->atom.atom = token; + type = read_token_item(&token); + break; + case EVENT_DELIM: + if (strcmp(token, "(") == 0) { + free_token(token); + type = process_paren(event, arg, &token); + break; + } + case EVENT_OP: + /* handle single ops */ + arg->type = PRINT_OP; + arg->op.op = token; + arg->op.left = NULL; + type = process_op(event, arg, &token); + + break; + + case EVENT_ERROR ... EVENT_NEWLINE: + default: + die("unexpected type %d", type); + } + *tok = token; + + return type; +} + +static int event_read_print_args(struct event *event, struct print_arg **list) +{ + enum event_type type; + struct print_arg *arg; + char *token; + int args = 0; + + do { + arg = malloc_or_die(sizeof(*arg)); + memset(arg, 0, sizeof(*arg)); + + type = process_arg(event, arg, &token); + + if (type == EVENT_ERROR) { + free_arg(arg); + return -1; + } + + *list = arg; + args++; + + if (type == EVENT_OP) { + type = process_op(event, arg, &token); + list = &arg->next; + continue; + } + + if (type == EVENT_DELIM && strcmp(token, ",") == 0) { + free_token(token); + *list = arg; + list = &arg->next; + continue; + } + break; + } while (type != EVENT_NONE); + + if (type != EVENT_NONE) + free_token(token); + + return args; +} + +static int event_read_print(struct event *event) +{ + enum event_type type; + char *token; + int ret; + + if (read_expected_item(EVENT_ITEM, (char *)"print") < 0) + return -1; + + if (read_expected(EVENT_ITEM, (char *)"fmt") < 0) + return -1; + + if (read_expected(EVENT_OP, (char *)":") < 0) + return -1; + + if (read_expect_type(EVENT_DQUOTE, &token) < 0) + goto fail; + + event->print_fmt.format = token; + event->print_fmt.args = NULL; + + /* ok to have no arg */ + type = read_token_item(&token); + + if (type == EVENT_NONE) + return 0; + + if (test_type_token(type, token, EVENT_DELIM, (char *)",")) + goto fail; + + free_token(token); + + ret = event_read_print_args(event, &event->print_fmt.args); + if (ret < 0) + return -1; + + return 0; + + fail: + free_token(token); + return -1; +} + +static struct format_field * +find_common_field(struct event *event, const char *name) +{ + struct format_field *format; + + for (format = event->format.common_fields; + format; format = format->next) { + if (strcmp(format->name, name) == 0) + break; + } + + return format; +} + +static struct format_field * +find_field(struct event *event, const char *name) +{ + struct format_field *format; + + for (format = event->format.fields; + format; format = format->next) { + if (strcmp(format->name, name) == 0) + break; + } + + return format; +} + +static struct format_field * +find_any_field(struct event *event, const char *name) +{ + struct format_field *format; + + format = find_common_field(event, name); + if (format) + return format; + return find_field(event, name); +} + +static unsigned long long read_size(void *ptr, int size) +{ + switch (size) { + case 1: + return *(unsigned char *)ptr; + case 2: + return data2host2(ptr); + case 4: + return data2host4(ptr); + case 8: + return data2host8(ptr); + default: + /* BUG! */ + return 0; + } +} + +unsigned long long +raw_field_value(struct event *event, const char *name, void *data) +{ + struct format_field *field; + + field = find_any_field(event, name); + if (!field) + return 0ULL; + + return read_size(data + field->offset, field->size); +} + +void *raw_field_ptr(struct event *event, const char *name, void *data) +{ + struct format_field *field; + + field = find_any_field(event, name); + if (!field) + return NULL; + + return data + field->offset; +} + +static int get_common_info(const char *type, int *offset, int *size) +{ + struct event *event; + struct format_field *field; + + /* + * All events should have the same common elements. + * Pick any event to find where the type is; + */ + if (!event_list) + die("no event_list!"); + + event = event_list; + field = find_common_field(event, type); + if (!field) + die("field '%s' not found", type); + + *offset = field->offset; + *size = field->size; + + return 0; +} + +int trace_parse_common_type(void *data) +{ + static int type_offset; + static int type_size; + int ret; + + if (!type_size) { + ret = get_common_info("common_type", + &type_offset, + &type_size); + if (ret < 0) + return ret; + } + return read_size(data + type_offset, type_size); +} + +static int parse_common_pid(void *data) +{ + static int pid_offset; + static int pid_size; + int ret; + + if (!pid_size) { + ret = get_common_info("common_pid", + &pid_offset, + &pid_size); + if (ret < 0) + return ret; + } + + return read_size(data + pid_offset, pid_size); +} + +struct event *trace_find_event(int id) +{ + struct event *event; + + for (event = event_list; event; event = event->next) { + if (event->id == id) + break; + } + return event; +} + +static unsigned long long eval_num_arg(void *data, int size, + struct event *event, struct print_arg *arg) +{ + unsigned long long val = 0; + unsigned long long left, right; + + switch (arg->type) { + case PRINT_NULL: + /* ?? */ + return 0; + case PRINT_ATOM: + return strtoull(arg->atom.atom, NULL, 0); + case PRINT_FIELD: + if (!arg->field.field) { + arg->field.field = find_any_field(event, arg->field.name); + if (!arg->field.field) + die("field %s not found", arg->field.name); + } + /* must be a number */ + val = read_size(data + arg->field.field->offset, + arg->field.field->size); + break; + case PRINT_FLAGS: + case PRINT_SYMBOL: + break; + case PRINT_TYPE: + return eval_num_arg(data, size, event, arg->typecast.item); + case PRINT_STRING: + return 0; + break; + case PRINT_OP: + left = eval_num_arg(data, size, event, arg->op.left); + right = eval_num_arg(data, size, event, arg->op.right); + switch (arg->op.op[0]) { + case '|': + if (arg->op.op[1]) + val = left || right; + else + val = left | right; + break; + case '&': + if (arg->op.op[1]) + val = left && right; + else + val = left & right; + break; + case '<': + switch (arg->op.op[1]) { + case 0: + val = left < right; + break; + case '<': + val = left << right; + break; + case '=': + val = left <= right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + case '>': + switch (arg->op.op[1]) { + case 0: + val = left > right; + break; + case '>': + val = left >> right; + break; + case '=': + val = left >= right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + case '=': + if (arg->op.op[1] != '=') + die("unknown op '%s'", arg->op.op); + val = left == right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + default: /* not sure what to do there */ + return 0; + } + return val; +} + +struct flag { + const char *name; + unsigned long long value; +}; + +static const struct flag flags[] = { + { "HI_SOFTIRQ", 0 }, + { "TIMER_SOFTIRQ", 1 }, + { "NET_TX_SOFTIRQ", 2 }, + { "NET_RX_SOFTIRQ", 3 }, + { "BLOCK_SOFTIRQ", 4 }, + { "TASKLET_SOFTIRQ", 5 }, + { "SCHED_SOFTIRQ", 6 }, + { "HRTIMER_SOFTIRQ", 7 }, + { "RCU_SOFTIRQ", 8 }, + + { "HRTIMER_NORESTART", 0 }, + { "HRTIMER_RESTART", 1 }, +}; + +static unsigned long long eval_flag(const char *flag) +{ + int i; + + /* + * Some flags in the format files do not get converted. + * If the flag is not numeric, see if it is something that + * we already know about. + */ + if (isdigit(flag[0])) + return strtoull(flag, NULL, 0); + + for (i = 0; i < (int)(sizeof(flags)/sizeof(flags[0])); i++) + if (strcmp(flags[i].name, flag) == 0) + return flags[i].value; + + return 0; +} + +static void print_str_arg(void *data, int size, + struct event *event, struct print_arg *arg) +{ + struct print_flag_sym *flag; + unsigned long long val, fval; + char *str; + int print; + + switch (arg->type) { + case PRINT_NULL: + /* ?? */ + return; + case PRINT_ATOM: + printf("%s", arg->atom.atom); + return; + case PRINT_FIELD: + if (!arg->field.field) { + arg->field.field = find_any_field(event, arg->field.name); + if (!arg->field.field) + die("field %s not found", arg->field.name); + } + str = malloc_or_die(arg->field.field->size + 1); + memcpy(str, data + arg->field.field->offset, + arg->field.field->size); + str[arg->field.field->size] = 0; + printf("%s", str); + free(str); + break; + case PRINT_FLAGS: + val = eval_num_arg(data, size, event, arg->flags.field); + print = 0; + for (flag = arg->flags.flags; flag; flag = flag->next) { + fval = eval_flag(flag->value); + if (!val && !fval) { + printf("%s", flag->str); + break; + } + if (fval && (val & fval) == fval) { + if (print && arg->flags.delim) + printf("%s", arg->flags.delim); + printf("%s", flag->str); + print = 1; + val &= ~fval; + } + } + break; + case PRINT_SYMBOL: + val = eval_num_arg(data, size, event, arg->symbol.field); + for (flag = arg->symbol.symbols; flag; flag = flag->next) { + fval = eval_flag(flag->value); + if (val == fval) { + printf("%s", flag->str); + break; + } + } + break; + + case PRINT_TYPE: + break; + case PRINT_STRING: { + int str_offset; + + if (arg->string.offset == -1) { + struct format_field *f; + + f = find_any_field(event, arg->string.string); + arg->string.offset = f->offset; + } + str_offset = *(int *)(data + arg->string.offset); + str_offset &= 0xffff; + printf("%s", ((char *)data) + str_offset); + break; + } + case PRINT_OP: + /* + * The only op for string should be ? : + */ + if (arg->op.op[0] != '?') + return; + val = eval_num_arg(data, size, event, arg->op.left); + if (val) + print_str_arg(data, size, event, arg->op.right->op.left); + else + print_str_arg(data, size, event, arg->op.right->op.right); + break; + default: + /* well... */ + break; + } +} + +static struct print_arg *make_bprint_args(char *fmt, void *data, int size, struct event *event) +{ + static struct format_field *field, *ip_field; + struct print_arg *args, *arg, **next; + unsigned long long ip, val; + char *ptr; + void *bptr; + + if (!field) { + field = find_field(event, "buf"); + if (!field) + die("can't find buffer field for binary printk"); + ip_field = find_field(event, "ip"); + if (!ip_field) + die("can't find ip field for binary printk"); + } + + ip = read_size(data + ip_field->offset, ip_field->size); + + /* + * The first arg is the IP pointer. + */ + args = malloc_or_die(sizeof(*args)); + arg = args; + arg->next = NULL; + next = &arg->next; + + arg->type = PRINT_ATOM; + arg->atom.atom = malloc_or_die(32); + sprintf(arg->atom.atom, "%lld", ip); + + /* skip the first "%pf : " */ + for (ptr = fmt + 6, bptr = data + field->offset; + bptr < data + size && *ptr; ptr++) { + int ls = 0; + + if (*ptr == '%') { + process_again: + ptr++; + switch (*ptr) { + case '%': + break; + case 'l': + ls++; + goto process_again; + case 'L': + ls = 2; + goto process_again; + case '0' ... '9': + goto process_again; + case 'p': + ls = 1; + /* fall through */ + case 'd': + case 'u': + case 'x': + case 'i': + bptr = (void *)(((unsigned long)bptr + (long_size - 1)) & + ~(long_size - 1)); + switch (ls) { + case 0: + case 1: + ls = long_size; + break; + case 2: + ls = 8; + default: + break; + } + val = read_size(bptr, ls); + bptr += ls; + arg = malloc_or_die(sizeof(*arg)); + arg->next = NULL; + arg->type = PRINT_ATOM; + arg->atom.atom = malloc_or_die(32); + sprintf(arg->atom.atom, "%lld", val); + *next = arg; + next = &arg->next; + break; + case 's': + arg = malloc_or_die(sizeof(*arg)); + arg->next = NULL; + arg->type = PRINT_STRING; + arg->string.string = strdup(bptr); + bptr += strlen(bptr) + 1; + *next = arg; + next = &arg->next; + default: + break; + } + } + } + + return args; +} + +static void free_args(struct print_arg *args) +{ + struct print_arg *next; + + while (args) { + next = args->next; + + if (args->type == PRINT_ATOM) + free(args->atom.atom); + else + free(args->string.string); + free(args); + args = next; + } +} + +static char *get_bprint_format(void *data, int size __unused, struct event *event) +{ + unsigned long long addr; + static struct format_field *field; + struct printk_map *printk; + char *format; + char *p; + + if (!field) { + field = find_field(event, "fmt"); + if (!field) + die("can't find format field for binary printk"); + printf("field->offset = %d size=%d\n", field->offset, field->size); + } + + addr = read_size(data + field->offset, field->size); + + printk = find_printk(addr); + if (!printk) { + format = malloc_or_die(45); + sprintf(format, "%%pf : (NO FORMAT FOUND at %llx)\n", + addr); + return format; + } + + p = printk->printk; + /* Remove any quotes. */ + if (*p == '"') + p++; + format = malloc_or_die(strlen(p) + 10); + sprintf(format, "%s : %s", "%pf", p); + /* remove ending quotes and new line since we will add one too */ + p = format + strlen(format) - 1; + if (*p == '"') + *p = 0; + + p -= 2; + if (strcmp(p, "\\n") == 0) + *p = 0; + + return format; +} + +static void pretty_print(void *data, int size, struct event *event) +{ + struct print_fmt *print_fmt = &event->print_fmt; + struct print_arg *arg = print_fmt->args; + struct print_arg *args = NULL; + const char *ptr = print_fmt->format; + unsigned long long val; + struct func_map *func; + const char *saveptr; + char *bprint_fmt = NULL; + char format[32]; + int show_func; + int len; + int ls; + + if (event->flags & EVENT_FL_ISFUNC) + ptr = " %pF <-- %pF"; + + if (event->flags & EVENT_FL_ISBPRINT) { + bprint_fmt = get_bprint_format(data, size, event); + args = make_bprint_args(bprint_fmt, data, size, event); + arg = args; + ptr = bprint_fmt; + } + + for (; *ptr; ptr++) { + ls = 0; + if (*ptr == '%') { + saveptr = ptr; + show_func = 0; + cont_process: + ptr++; + switch (*ptr) { + case '%': + printf("%%"); + break; + case 'l': + ls++; + goto cont_process; + case 'L': + ls = 2; + goto cont_process; + case 'z': + case 'Z': + case '0' ... '9': + goto cont_process; + case 'p': + if (long_size == 4) + ls = 1; + else + ls = 2; + + if (*(ptr+1) == 'F' || + *(ptr+1) == 'f') { + ptr++; + show_func = *ptr; + } + + /* fall through */ + case 'd': + case 'i': + case 'x': + case 'X': + case 'u': + if (!arg) + die("no argument match"); + + len = ((unsigned long)ptr + 1) - + (unsigned long)saveptr; + + /* should never happen */ + if (len > 32) + die("bad format!"); + + memcpy(format, saveptr, len); + format[len] = 0; + + val = eval_num_arg(data, size, event, arg); + arg = arg->next; + + if (show_func) { + func = find_func(val); + if (func) { + printf("%s", func->func); + if (show_func == 'F') + printf("+0x%llx", + val - func->addr); + break; + } + } + switch (ls) { + case 0: + printf(format, (int)val); + break; + case 1: + printf(format, (long)val); + break; + case 2: + printf(format, (long long)val); + break; + default: + die("bad count (%d)", ls); + } + break; + case 's': + if (!arg) + die("no matching argument"); + + print_str_arg(data, size, event, arg); + arg = arg->next; + break; + default: + printf(">%c<", *ptr); + + } + } else + printf("%c", *ptr); + } + + if (args) { + free_args(args); + free(bprint_fmt); + } +} + +static inline int log10_cpu(int nb) +{ + if (nb / 100) + return 3; + if (nb / 10) + return 2; + return 1; +} + +/* taken from Linux, written by Frederic Weisbecker */ +static void print_graph_cpu(int cpu) +{ + int i; + int log10_this = log10_cpu(cpu); + int log10_all = log10_cpu(cpus); + + + /* + * Start with a space character - to make it stand out + * to the right a bit when trace output is pasted into + * email: + */ + printf(" "); + + /* + * Tricky - we space the CPU field according to the max + * number of online CPUs. On a 2-cpu system it would take + * a maximum of 1 digit - on a 128 cpu system it would + * take up to 3 digits: + */ + for (i = 0; i < log10_all - log10_this; i++) + printf(" "); + + printf("%d) ", cpu); +} + +#define TRACE_GRAPH_PROCINFO_LENGTH 14 +#define TRACE_GRAPH_INDENT 2 + +static void print_graph_proc(int pid, const char *comm) +{ + /* sign + log10(MAX_INT) + '\0' */ + char pid_str[11]; + int spaces = 0; + int len; + int i; + + sprintf(pid_str, "%d", pid); + + /* 1 stands for the "-" character */ + len = strlen(comm) + strlen(pid_str) + 1; + + if (len < TRACE_GRAPH_PROCINFO_LENGTH) + spaces = TRACE_GRAPH_PROCINFO_LENGTH - len; + + /* First spaces to align center */ + for (i = 0; i < spaces / 2; i++) + printf(" "); + + printf("%s-%s", comm, pid_str); + + /* Last spaces to align center */ + for (i = 0; i < spaces - (spaces / 2); i++) + printf(" "); +} + +static struct record * +get_return_for_leaf(int cpu, int cur_pid, unsigned long long cur_func, + struct record *next) +{ + struct format_field *field; + struct event *event; + unsigned long val; + int type; + int pid; + + type = trace_parse_common_type(next->data); + event = trace_find_event(type); + if (!event) + return NULL; + + if (!(event->flags & EVENT_FL_ISFUNCRET)) + return NULL; + + pid = parse_common_pid(next->data); + field = find_field(event, "func"); + if (!field) + die("function return does not have field func"); + + val = read_size(next->data + field->offset, field->size); + + if (cur_pid != pid || cur_func != val) + return NULL; + + /* this is a leaf, now advance the iterator */ + return trace_read_data(cpu); +} + +/* Signal a overhead of time execution to the output */ +static void print_graph_overhead(unsigned long long duration) +{ + /* Non nested entry or return */ + if (duration == ~0ULL) + return (void)printf(" "); + + /* Duration exceeded 100 msecs */ + if (duration > 100000ULL) + return (void)printf("! "); + + /* Duration exceeded 10 msecs */ + if (duration > 10000ULL) + return (void)printf("+ "); + + printf(" "); +} + +static void print_graph_duration(unsigned long long duration) +{ + unsigned long usecs = duration / 1000; + unsigned long nsecs_rem = duration % 1000; + /* log10(ULONG_MAX) + '\0' */ + char msecs_str[21]; + char nsecs_str[5]; + int len; + int i; + + sprintf(msecs_str, "%lu", usecs); + + /* Print msecs */ + len = printf("%lu", usecs); + + /* Print nsecs (we don't want to exceed 7 numbers) */ + if (len < 7) { + snprintf(nsecs_str, 8 - len, "%03lu", nsecs_rem); + len += printf(".%s", nsecs_str); + } + + printf(" us "); + + /* Print remaining spaces to fit the row's width */ + for (i = len; i < 7; i++) + printf(" "); + + printf("| "); +} + +static void +print_graph_entry_leaf(struct event *event, void *data, struct record *ret_rec) +{ + unsigned long long rettime, calltime; + unsigned long long duration, depth; + unsigned long long val; + struct format_field *field; + struct func_map *func; + struct event *ret_event; + int type; + int i; + + type = trace_parse_common_type(ret_rec->data); + ret_event = trace_find_event(type); + + field = find_field(ret_event, "rettime"); + if (!field) + die("can't find rettime in return graph"); + rettime = read_size(ret_rec->data + field->offset, field->size); + + field = find_field(ret_event, "calltime"); + if (!field) + die("can't find rettime in return graph"); + calltime = read_size(ret_rec->data + field->offset, field->size); + + duration = rettime - calltime; + + /* Overhead */ + print_graph_overhead(duration); + + /* Duration */ + print_graph_duration(duration); + + field = find_field(event, "depth"); + if (!field) + die("can't find depth in entry graph"); + depth = read_size(data + field->offset, field->size); + + /* Function */ + for (i = 0; i < (int)(depth * TRACE_GRAPH_INDENT); i++) + printf(" "); + + field = find_field(event, "func"); + if (!field) + die("can't find func in entry graph"); + val = read_size(data + field->offset, field->size); + func = find_func(val); + + if (func) + printf("%s();", func->func); + else + printf("%llx();", val); +} + +static void print_graph_nested(struct event *event, void *data) +{ + struct format_field *field; + unsigned long long depth; + unsigned long long val; + struct func_map *func; + int i; + + /* No overhead */ + print_graph_overhead(-1); + + /* No time */ + printf(" | "); + + field = find_field(event, "depth"); + if (!field) + die("can't find depth in entry graph"); + depth = read_size(data + field->offset, field->size); + + /* Function */ + for (i = 0; i < (int)(depth * TRACE_GRAPH_INDENT); i++) + printf(" "); + + field = find_field(event, "func"); + if (!field) + die("can't find func in entry graph"); + val = read_size(data + field->offset, field->size); + func = find_func(val); + + if (func) + printf("%s() {", func->func); + else + printf("%llx() {", val); +} + +static void +pretty_print_func_ent(void *data, int size, struct event *event, + int cpu, int pid, const char *comm, + unsigned long secs, unsigned long usecs) +{ + struct format_field *field; + struct record *rec; + void *copy_data; + unsigned long val; + + printf("%5lu.%06lu | ", secs, usecs); + + print_graph_cpu(cpu); + print_graph_proc(pid, comm); + + printf(" | "); + + field = find_field(event, "func"); + if (!field) + die("function entry does not have func field"); + + val = read_size(data + field->offset, field->size); + + /* + * peek_data may unmap the data pointer. Copy it first. + */ + copy_data = malloc_or_die(size); + memcpy(copy_data, data, size); + data = copy_data; + + rec = trace_peek_data(cpu); + if (rec) { + rec = get_return_for_leaf(cpu, pid, val, rec); + if (rec) { + print_graph_entry_leaf(event, data, rec); + goto out_free; + } + } + print_graph_nested(event, data); +out_free: + free(data); +} + +static void +pretty_print_func_ret(void *data, int size __unused, struct event *event, + int cpu, int pid, const char *comm, + unsigned long secs, unsigned long usecs) +{ + unsigned long long rettime, calltime; + unsigned long long duration, depth; + struct format_field *field; + int i; + + printf("%5lu.%06lu | ", secs, usecs); + + print_graph_cpu(cpu); + print_graph_proc(pid, comm); + + printf(" | "); + + field = find_field(event, "rettime"); + if (!field) + die("can't find rettime in return graph"); + rettime = read_size(data + field->offset, field->size); + + field = find_field(event, "calltime"); + if (!field) + die("can't find calltime in return graph"); + calltime = read_size(data + field->offset, field->size); + + duration = rettime - calltime; + + /* Overhead */ + print_graph_overhead(duration); + + /* Duration */ + print_graph_duration(duration); + + field = find_field(event, "depth"); + if (!field) + die("can't find depth in entry graph"); + depth = read_size(data + field->offset, field->size); + + /* Function */ + for (i = 0; i < (int)(depth * TRACE_GRAPH_INDENT); i++) + printf(" "); + + printf("}"); +} + +static void +pretty_print_func_graph(void *data, int size, struct event *event, + int cpu, int pid, const char *comm, + unsigned long secs, unsigned long usecs) +{ + if (event->flags & EVENT_FL_ISFUNCENT) + pretty_print_func_ent(data, size, event, + cpu, pid, comm, secs, usecs); + else if (event->flags & EVENT_FL_ISFUNCRET) + pretty_print_func_ret(data, size, event, + cpu, pid, comm, secs, usecs); + printf("\n"); +} + +void print_event(int cpu, void *data, int size, unsigned long long nsecs, + char *comm) +{ + struct event *event; + unsigned long secs; + unsigned long usecs; + int type; + int pid; + + secs = nsecs / NSECS_PER_SEC; + nsecs -= secs * NSECS_PER_SEC; + usecs = nsecs / NSECS_PER_USEC; + + type = trace_parse_common_type(data); + + event = trace_find_event(type); + if (!event) { + printf("ug! no event found for type %d\n", type); + return; + } + + pid = parse_common_pid(data); + + if (event->flags & (EVENT_FL_ISFUNCENT | EVENT_FL_ISFUNCRET)) + return pretty_print_func_graph(data, size, event, cpu, + pid, comm, secs, usecs); + + printf("%16s-%-5d [%03d] %5lu.%09Lu: %s: ", + comm, pid, cpu, + secs, nsecs, event->name); + + pretty_print(data, size, event); + printf("\n"); +} + +static void print_fields(struct print_flag_sym *field) +{ + printf("{ %s, %s }", field->value, field->str); + if (field->next) { + printf(", "); + print_fields(field->next); + } +} + +static void print_args(struct print_arg *args) +{ + int print_paren = 1; + + switch (args->type) { + case PRINT_NULL: + printf("null"); + break; + case PRINT_ATOM: + printf("%s", args->atom.atom); + break; + case PRINT_FIELD: + printf("REC->%s", args->field.name); + break; + case PRINT_FLAGS: + printf("__print_flags("); + print_args(args->flags.field); + printf(", %s, ", args->flags.delim); + print_fields(args->flags.flags); + printf(")"); + break; + case PRINT_SYMBOL: + printf("__print_symbolic("); + print_args(args->symbol.field); + printf(", "); + print_fields(args->symbol.symbols); + printf(")"); + break; + case PRINT_STRING: + printf("__get_str(%s)", args->string.string); + break; + case PRINT_TYPE: + printf("(%s)", args->typecast.type); + print_args(args->typecast.item); + break; + case PRINT_OP: + if (strcmp(args->op.op, ":") == 0) + print_paren = 0; + if (print_paren) + printf("("); + print_args(args->op.left); + printf(" %s ", args->op.op); + print_args(args->op.right); + if (print_paren) + printf(")"); + break; + default: + /* we should warn... */ + return; + } + if (args->next) { + printf("\n"); + print_args(args->next); + } +} + +static void parse_header_field(char *type, + int *offset, int *size) +{ + char *token; + + if (read_expected(EVENT_ITEM, (char *)"field") < 0) + return; + if (read_expected(EVENT_OP, (char *)":") < 0) + return; + /* type */ + if (read_expect_type(EVENT_ITEM, &token) < 0) + return; + free_token(token); + + if (read_expected(EVENT_ITEM, type) < 0) + return; + if (read_expected(EVENT_OP, (char *)";") < 0) + return; + if (read_expected(EVENT_ITEM, (char *)"offset") < 0) + return; + if (read_expected(EVENT_OP, (char *)":") < 0) + return; + if (read_expect_type(EVENT_ITEM, &token) < 0) + return; + *offset = atoi(token); + free_token(token); + if (read_expected(EVENT_OP, (char *)";") < 0) + return; + if (read_expected(EVENT_ITEM, (char *)"size") < 0) + return; + if (read_expected(EVENT_OP, (char *)":") < 0) + return; + if (read_expect_type(EVENT_ITEM, &token) < 0) + return; + *size = atoi(token); + free_token(token); + if (read_expected(EVENT_OP, (char *)";") < 0) + return; + if (read_expect_type(EVENT_NEWLINE, &token) < 0) + return; + free_token(token); +} + +int parse_header_page(char *buf, unsigned long size) +{ + init_input_buf(buf, size); + + parse_header_field((char *)"timestamp", &header_page_ts_offset, + &header_page_ts_size); + parse_header_field((char *)"commit", &header_page_size_offset, + &header_page_size_size); + parse_header_field((char *)"data", &header_page_data_offset, + &header_page_data_size); + + return 0; +} + +int parse_ftrace_file(char *buf, unsigned long size) +{ + struct format_field *field; + struct print_arg *arg, **list; + struct event *event; + int ret; + + init_input_buf(buf, size); + + event = alloc_event(); + if (!event) + return -ENOMEM; + + event->flags |= EVENT_FL_ISFTRACE; + + event->name = event_read_name(); + if (!event->name) + die("failed to read ftrace event name"); + + if (strcmp(event->name, "function") == 0) + event->flags |= EVENT_FL_ISFUNC; + + else if (strcmp(event->name, "funcgraph_entry") == 0) + event->flags |= EVENT_FL_ISFUNCENT; + + else if (strcmp(event->name, "funcgraph_exit") == 0) + event->flags |= EVENT_FL_ISFUNCRET; + + else if (strcmp(event->name, "bprint") == 0) + event->flags |= EVENT_FL_ISBPRINT; + + event->id = event_read_id(); + if (event->id < 0) + die("failed to read ftrace event id"); + + add_event(event); + + ret = event_read_format(event); + if (ret < 0) + die("failed to read ftrace event format"); + + ret = event_read_print(event); + if (ret < 0) + die("failed to read ftrace event print fmt"); + + /* + * The arguments for ftrace files are parsed by the fields. + * Set up the fields as their arguments. + */ + list = &event->print_fmt.args; + for (field = event->format.fields; field; field = field->next) { + arg = malloc_or_die(sizeof(*arg)); + memset(arg, 0, sizeof(*arg)); + *list = arg; + list = &arg->next; + arg->type = PRINT_FIELD; + arg->field.name = field->name; + arg->field.field = field; + } + return 0; +} + +int parse_event_file(char *buf, unsigned long size, char *system__unused __unused) +{ + struct event *event; + int ret; + + init_input_buf(buf, size); + + event = alloc_event(); + if (!event) + return -ENOMEM; + + event->name = event_read_name(); + if (!event->name) + die("failed to read event name"); + + event->id = event_read_id(); + if (event->id < 0) + die("failed to read event id"); + + ret = event_read_format(event); + if (ret < 0) + die("failed to read event format"); + + ret = event_read_print(event); + if (ret < 0) + die("failed to read event print fmt"); + +#define PRINT_ARGS 0 + if (PRINT_ARGS && event->print_fmt.args) + print_args(event->print_fmt.args); + + add_event(event); + return 0; +} + +void parse_set_info(int nr_cpus, int long_sz) +{ + cpus = nr_cpus; + long_size = long_sz; +} diff --git a/tools/perf/util/trace-event-read.c b/tools/perf/util/trace-event-read.c new file mode 100644 index 00000000000..1b5c847d2c2 --- /dev/null +++ b/tools/perf/util/trace-event-read.c @@ -0,0 +1,514 @@ +/* + * Copyright (C) 2009, Steven Rostedt <srostedt@redhat.com> + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; version 2 of the License (not later!) + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + */ +#define _LARGEFILE64_SOURCE + +#include <dirent.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <getopt.h> +#include <stdarg.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <sys/wait.h> +#include <sys/mman.h> +#include <pthread.h> +#include <fcntl.h> +#include <unistd.h> +#include <ctype.h> +#include <errno.h> + +#include "../perf.h" +#include "util.h" +#include "trace-event.h" + +static int input_fd; + +static int read_page; + +int file_bigendian; +int host_bigendian; +static int long_size; + +static unsigned long page_size; + +static int read_or_die(void *data, int size) +{ + int r; + + r = read(input_fd, data, size); + if (r != size) + die("reading input file (size expected=%d received=%d)", + size, r); + return r; +} + +static unsigned int read4(void) +{ + unsigned int data; + + read_or_die(&data, 4); + return __data2host4(data); +} + +static unsigned long long read8(void) +{ + unsigned long long data; + + read_or_die(&data, 8); + return __data2host8(data); +} + +static char *read_string(void) +{ + char buf[BUFSIZ]; + char *str = NULL; + int size = 0; + int i; + int r; + + for (;;) { + r = read(input_fd, buf, BUFSIZ); + if (r < 0) + die("reading input file"); + + if (!r) + die("no data"); + + for (i = 0; i < r; i++) { + if (!buf[i]) + break; + } + if (i < r) + break; + + if (str) { + size += BUFSIZ; + str = realloc(str, size); + if (!str) + die("malloc of size %d", size); + memcpy(str + (size - BUFSIZ), buf, BUFSIZ); + } else { + size = BUFSIZ; + str = malloc_or_die(size); + memcpy(str, buf, size); + } + } + + /* trailing \0: */ + i++; + + /* move the file descriptor to the end of the string */ + r = lseek(input_fd, -(r - i), SEEK_CUR); + if (r < 0) + die("lseek"); + + if (str) { + size += i; + str = realloc(str, size); + if (!str) + die("malloc of size %d", size); + memcpy(str + (size - i), buf, i); + } else { + size = i; + str = malloc_or_die(i); + memcpy(str, buf, i); + } + + return str; +} + +static void read_proc_kallsyms(void) +{ + unsigned int size; + char *buf; + + size = read4(); + if (!size) + return; + + buf = malloc_or_die(size); + read_or_die(buf, size); + + parse_proc_kallsyms(buf, size); + + free(buf); +} + +static void read_ftrace_printk(void) +{ + unsigned int size; + char *buf; + + size = read4(); + if (!size) + return; + + buf = malloc_or_die(size); + read_or_die(buf, size); + + parse_ftrace_printk(buf, size); + + free(buf); +} + +static void read_header_files(void) +{ + unsigned long long size; + char *header_page; + char *header_event; + char buf[BUFSIZ]; + + read_or_die(buf, 12); + + if (memcmp(buf, "header_page", 12) != 0) + die("did not read header page"); + + size = read8(); + header_page = malloc_or_die(size); + read_or_die(header_page, size); + parse_header_page(header_page, size); + free(header_page); + + /* + * The size field in the page is of type long, + * use that instead, since it represents the kernel. + */ + long_size = header_page_size_size; + + read_or_die(buf, 13); + if (memcmp(buf, "header_event", 13) != 0) + die("did not read header event"); + + size = read8(); + header_event = malloc_or_die(size); + read_or_die(header_event, size); + free(header_event); +} + +static void read_ftrace_file(unsigned long long size) +{ + char *buf; + + buf = malloc_or_die(size); + read_or_die(buf, size); + parse_ftrace_file(buf, size); + free(buf); +} + +static void read_event_file(char *sys, unsigned long long size) +{ + char *buf; + + buf = malloc_or_die(size); + read_or_die(buf, size); + parse_event_file(buf, size, sys); + free(buf); +} + +static void read_ftrace_files(void) +{ + unsigned long long size; + int count; + int i; + + count = read4(); + + for (i = 0; i < count; i++) { + size = read8(); + read_ftrace_file(size); + } +} + +static void read_event_files(void) +{ + unsigned long long size; + char *sys; + int systems; + int count; + int i,x; + + systems = read4(); + + for (i = 0; i < systems; i++) { + sys = read_string(); + + count = read4(); + for (x=0; x < count; x++) { + size = read8(); + read_event_file(sys, size); + } + } +} + +struct cpu_data { + unsigned long long offset; + unsigned long long size; + unsigned long long timestamp; + struct record *next; + char *page; + int cpu; + int index; + int page_size; +}; + +static struct cpu_data *cpu_data; + +static void update_cpu_data_index(int cpu) +{ + cpu_data[cpu].offset += page_size; + cpu_data[cpu].size -= page_size; + cpu_data[cpu].index = 0; +} + +static void get_next_page(int cpu) +{ + off64_t save_seek; + off64_t ret; + + if (!cpu_data[cpu].page) + return; + + if (read_page) { + if (cpu_data[cpu].size <= page_size) { + free(cpu_data[cpu].page); + cpu_data[cpu].page = NULL; + return; + } + + update_cpu_data_index(cpu); + + /* other parts of the code may expect the pointer to not move */ + save_seek = lseek64(input_fd, 0, SEEK_CUR); + + ret = lseek64(input_fd, cpu_data[cpu].offset, SEEK_SET); + if (ret < 0) + die("failed to lseek"); + ret = read(input_fd, cpu_data[cpu].page, page_size); + if (ret < 0) + die("failed to read page"); + + /* reset the file pointer back */ + lseek64(input_fd, save_seek, SEEK_SET); + + return; + } + + munmap(cpu_data[cpu].page, page_size); + cpu_data[cpu].page = NULL; + + if (cpu_data[cpu].size <= page_size) + return; + + update_cpu_data_index(cpu); + + cpu_data[cpu].page = mmap(NULL, page_size, PROT_READ, MAP_PRIVATE, + input_fd, cpu_data[cpu].offset); + if (cpu_data[cpu].page == MAP_FAILED) + die("failed to mmap cpu %d at offset 0x%llx", + cpu, cpu_data[cpu].offset); +} + +static unsigned int type_len4host(unsigned int type_len_ts) +{ + if (file_bigendian) + return (type_len_ts >> 27) & ((1 << 5) - 1); + else + return type_len_ts & ((1 << 5) - 1); +} + +static unsigned int ts4host(unsigned int type_len_ts) +{ + if (file_bigendian) + return type_len_ts & ((1 << 27) - 1); + else + return type_len_ts >> 5; +} + +static int calc_index(void *ptr, int cpu) +{ + return (unsigned long)ptr - (unsigned long)cpu_data[cpu].page; +} + +struct record *trace_peek_data(int cpu) +{ + struct record *data; + void *page = cpu_data[cpu].page; + int idx = cpu_data[cpu].index; + void *ptr = page + idx; + unsigned long long extend; + unsigned int type_len_ts; + unsigned int type_len; + unsigned int delta; + unsigned int length = 0; + + if (cpu_data[cpu].next) + return cpu_data[cpu].next; + + if (!page) + return NULL; + + if (!idx) { + /* FIXME: handle header page */ + if (header_page_ts_size != 8) + die("expected a long long type for timestamp"); + cpu_data[cpu].timestamp = data2host8(ptr); + ptr += 8; + switch (header_page_size_size) { + case 4: + cpu_data[cpu].page_size = data2host4(ptr); + ptr += 4; + break; + case 8: + cpu_data[cpu].page_size = data2host8(ptr); + ptr += 8; + break; + default: + die("bad long size"); + } + ptr = cpu_data[cpu].page + header_page_data_offset; + } + +read_again: + idx = calc_index(ptr, cpu); + + if (idx >= cpu_data[cpu].page_size) { + get_next_page(cpu); + return trace_peek_data(cpu); + } + + type_len_ts = data2host4(ptr); + ptr += 4; + + type_len = type_len4host(type_len_ts); + delta = ts4host(type_len_ts); + + switch (type_len) { + case RINGBUF_TYPE_PADDING: + if (!delta) + die("error, hit unexpected end of page"); + length = data2host4(ptr); + ptr += 4; + length *= 4; + ptr += length; + goto read_again; + + case RINGBUF_TYPE_TIME_EXTEND: + extend = data2host4(ptr); + ptr += 4; + extend <<= TS_SHIFT; + extend += delta; + cpu_data[cpu].timestamp += extend; + goto read_again; + + case RINGBUF_TYPE_TIME_STAMP: + ptr += 12; + break; + case 0: + length = data2host4(ptr); + ptr += 4; + die("here! length=%d", length); + break; + default: + length = type_len * 4; + break; + } + + cpu_data[cpu].timestamp += delta; + + data = malloc_or_die(sizeof(*data)); + memset(data, 0, sizeof(*data)); + + data->ts = cpu_data[cpu].timestamp; + data->size = length; + data->data = ptr; + ptr += length; + + cpu_data[cpu].index = calc_index(ptr, cpu); + cpu_data[cpu].next = data; + + return data; +} + +struct record *trace_read_data(int cpu) +{ + struct record *data; + + data = trace_peek_data(cpu); + cpu_data[cpu].next = NULL; + + return data; +} + +void trace_report(void) +{ + const char *input_file = "trace.info"; + char buf[BUFSIZ]; + char test[] = { 23, 8, 68 }; + char *version; + int show_version = 0; + int show_funcs = 0; + int show_printk = 0; + + input_fd = open(input_file, O_RDONLY); + if (input_fd < 0) + die("opening '%s'\n", input_file); + + read_or_die(buf, 3); + if (memcmp(buf, test, 3) != 0) + die("not an trace data file"); + + read_or_die(buf, 7); + if (memcmp(buf, "tracing", 7) != 0) + die("not a trace file (missing tracing)"); + + version = read_string(); + if (show_version) + printf("version = %s\n", version); + free(version); + + read_or_die(buf, 1); + file_bigendian = buf[0]; + host_bigendian = bigendian(); + + read_or_die(buf, 1); + long_size = buf[0]; + + page_size = read4(); + + read_header_files(); + + read_ftrace_files(); + read_event_files(); + read_proc_kallsyms(); + read_ftrace_printk(); + + if (show_funcs) { + print_funcs(); + return; + } + if (show_printk) { + print_printk(); + return; + } + + return; +} diff --git a/tools/perf/util/trace-event.h b/tools/perf/util/trace-event.h new file mode 100644 index 00000000000..693f815c942 --- /dev/null +++ b/tools/perf/util/trace-event.h @@ -0,0 +1,245 @@ +#ifndef _TRACE_EVENTS_H +#define _TRACE_EVENTS_H + +#include "parse-events.h" + +#define __unused __attribute__((unused)) + + +#ifndef PAGE_MASK +#define PAGE_MASK (page_size - 1) +#endif + +enum { + RINGBUF_TYPE_PADDING = 29, + RINGBUF_TYPE_TIME_EXTEND = 30, + RINGBUF_TYPE_TIME_STAMP = 31, +}; + +#ifndef TS_SHIFT +#define TS_SHIFT 27 +#endif + +#define NSECS_PER_SEC 1000000000ULL +#define NSECS_PER_USEC 1000ULL + +enum format_flags { + FIELD_IS_ARRAY = 1, + FIELD_IS_POINTER = 2, +}; + +struct format_field { + struct format_field *next; + char *type; + char *name; + int offset; + int size; + unsigned long flags; +}; + +struct format { + int nr_common; + int nr_fields; + struct format_field *common_fields; + struct format_field *fields; +}; + +struct print_arg_atom { + char *atom; +}; + +struct print_arg_string { + char *string; + int offset; +}; + +struct print_arg_field { + char *name; + struct format_field *field; +}; + +struct print_flag_sym { + struct print_flag_sym *next; + char *value; + char *str; +}; + +struct print_arg_typecast { + char *type; + struct print_arg *item; +}; + +struct print_arg_flags { + struct print_arg *field; + char *delim; + struct print_flag_sym *flags; +}; + +struct print_arg_symbol { + struct print_arg *field; + struct print_flag_sym *symbols; +}; + +struct print_arg; + +struct print_arg_op { + char *op; + int prio; + struct print_arg *left; + struct print_arg *right; +}; + +struct print_arg_func { + char *name; + struct print_arg *args; +}; + +enum print_arg_type { + PRINT_NULL, + PRINT_ATOM, + PRINT_FIELD, + PRINT_FLAGS, + PRINT_SYMBOL, + PRINT_TYPE, + PRINT_STRING, + PRINT_OP, +}; + +struct print_arg { + struct print_arg *next; + enum print_arg_type type; + union { + struct print_arg_atom atom; + struct print_arg_field field; + struct print_arg_typecast typecast; + struct print_arg_flags flags; + struct print_arg_symbol symbol; + struct print_arg_func func; + struct print_arg_string string; + struct print_arg_op op; + }; +}; + +struct print_fmt { + char *format; + struct print_arg *args; +}; + +struct event { + struct event *next; + char *name; + int id; + int flags; + struct format format; + struct print_fmt print_fmt; +}; + +enum { + EVENT_FL_ISFTRACE = 1, + EVENT_FL_ISPRINT = 2, + EVENT_FL_ISBPRINT = 4, + EVENT_FL_ISFUNC = 8, + EVENT_FL_ISFUNCENT = 16, + EVENT_FL_ISFUNCRET = 32, +}; + +struct record { + unsigned long long ts; + int size; + void *data; +}; + +struct record *trace_peek_data(int cpu); +struct record *trace_read_data(int cpu); + +void parse_set_info(int nr_cpus, int long_sz); + +void trace_report(void); + +void *malloc_or_die(unsigned int size); + +void parse_cmdlines(char *file, int size); +void parse_proc_kallsyms(char *file, unsigned int size); +void parse_ftrace_printk(char *file, unsigned int size); + +void print_funcs(void); +void print_printk(void); + +int parse_ftrace_file(char *buf, unsigned long size); +int parse_event_file(char *buf, unsigned long size, char *system); +void print_event(int cpu, void *data, int size, unsigned long long nsecs, + char *comm); + +extern int file_bigendian; +extern int host_bigendian; + +int bigendian(void); + +static inline unsigned short __data2host2(unsigned short data) +{ + unsigned short swap; + + if (host_bigendian == file_bigendian) + return data; + + swap = ((data & 0xffULL) << 8) | + ((data & (0xffULL << 8)) >> 8); + + return swap; +} + +static inline unsigned int __data2host4(unsigned int data) +{ + unsigned int swap; + + if (host_bigendian == file_bigendian) + return data; + + swap = ((data & 0xffULL) << 24) | + ((data & (0xffULL << 8)) << 8) | + ((data & (0xffULL << 16)) >> 8) | + ((data & (0xffULL << 24)) >> 24); + + return swap; +} + +static inline unsigned long long __data2host8(unsigned long long data) +{ + unsigned long long swap; + + if (host_bigendian == file_bigendian) + return data; + + swap = ((data & 0xffULL) << 56) | + ((data & (0xffULL << 8)) << 40) | + ((data & (0xffULL << 16)) << 24) | + ((data & (0xffULL << 24)) << 8) | + ((data & (0xffULL << 32)) >> 8) | + ((data & (0xffULL << 40)) >> 24) | + ((data & (0xffULL << 48)) >> 40) | + ((data & (0xffULL << 56)) >> 56); + + return swap; +} + +#define data2host2(ptr) __data2host2(*(unsigned short *)ptr) +#define data2host4(ptr) __data2host4(*(unsigned int *)ptr) +#define data2host8(ptr) __data2host8(*(unsigned long long *)ptr) + +extern int header_page_ts_offset; +extern int header_page_ts_size; +extern int header_page_size_offset; +extern int header_page_size_size; +extern int header_page_data_offset; +extern int header_page_data_size; + +int parse_header_page(char *buf, unsigned long size); +int trace_parse_common_type(void *data); +struct event *trace_find_event(int id); +unsigned long long +raw_field_value(struct event *event, const char *name, void *data); +void *raw_field_ptr(struct event *event, const char *name, void *data); + +void read_tracing_data(struct perf_event_attr *pattrs, int nb_events); + +#endif /* _TRACE_EVENTS_H */ diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h index b4be6071c10..9de2329dd44 100644 --- a/tools/perf/util/util.h +++ b/tools/perf/util/util.h @@ -39,10 +39,6 @@ /* Approximation of the length of the decimal representation of this type. */ #define decimal_length(x) ((int)(sizeof(x) * 2.56 + 0.5) + 1) -#if !defined(__APPLE__) && !defined(__FreeBSD__) && !defined(__USLC__) && !defined(_M_UNIX) -#define _XOPEN_SOURCE 600 /* glibc2 and AIX 5.3L need 500, OpenBSD needs 600 for S_ISLNK() */ -#define _XOPEN_SOURCE_EXTENDED 1 /* AIX 5.3L needs this */ -#endif #define _ALL_SOURCE 1 #define _GNU_SOURCE 1 #define _BSD_SOURCE 1 @@ -50,6 +46,7 @@ #include <unistd.h> #include <stdio.h> #include <sys/stat.h> +#include <sys/statfs.h> #include <fcntl.h> #include <stddef.h> #include <stdlib.h> @@ -80,6 +77,8 @@ #include <netdb.h> #include <pwd.h> #include <inttypes.h> +#include "../../../include/linux/magic.h" + #ifndef NO_ICONV #include <iconv.h> @@ -308,6 +307,7 @@ static inline int has_extension(const char *filename, const char *ext) #undef isspace #undef isdigit #undef isalpha +#undef isprint #undef isalnum #undef tolower #undef toupper diff --git a/tools/perf/util/values.c b/tools/perf/util/values.c new file mode 100644 index 00000000000..1c15e39f99e --- /dev/null +++ b/tools/perf/util/values.c @@ -0,0 +1,230 @@ +#include <stdlib.h> + +#include "util.h" +#include "values.h" + +void perf_read_values_init(struct perf_read_values *values) +{ + values->threads_max = 16; + values->pid = malloc(values->threads_max * sizeof(*values->pid)); + values->tid = malloc(values->threads_max * sizeof(*values->tid)); + values->value = malloc(values->threads_max * sizeof(*values->value)); + if (!values->pid || !values->tid || !values->value) + die("failed to allocate read_values threads arrays"); + values->threads = 0; + + values->counters_max = 16; + values->counterrawid = malloc(values->counters_max + * sizeof(*values->counterrawid)); + values->countername = malloc(values->counters_max + * sizeof(*values->countername)); + if (!values->counterrawid || !values->countername) + die("failed to allocate read_values counters arrays"); + values->counters = 0; +} + +void perf_read_values_destroy(struct perf_read_values *values) +{ + int i; + + if (!values->threads_max || !values->counters_max) + return; + + for (i = 0; i < values->threads; i++) + free(values->value[i]); + free(values->pid); + free(values->tid); + free(values->counterrawid); + for (i = 0; i < values->counters; i++) + free(values->countername[i]); + free(values->countername); +} + +static void perf_read_values__enlarge_threads(struct perf_read_values *values) +{ + values->threads_max *= 2; + values->pid = realloc(values->pid, + values->threads_max * sizeof(*values->pid)); + values->tid = realloc(values->tid, + values->threads_max * sizeof(*values->tid)); + values->value = realloc(values->value, + values->threads_max * sizeof(*values->value)); + if (!values->pid || !values->tid || !values->value) + die("failed to enlarge read_values threads arrays"); +} + +static int perf_read_values__findnew_thread(struct perf_read_values *values, + u32 pid, u32 tid) +{ + int i; + + for (i = 0; i < values->threads; i++) + if (values->pid[i] == pid && values->tid[i] == tid) + return i; + + if (values->threads == values->threads_max) + perf_read_values__enlarge_threads(values); + + i = values->threads++; + values->pid[i] = pid; + values->tid[i] = tid; + values->value[i] = malloc(values->counters_max * sizeof(**values->value)); + if (!values->value[i]) + die("failed to allocate read_values counters array"); + + return i; +} + +static void perf_read_values__enlarge_counters(struct perf_read_values *values) +{ + int i; + + values->counters_max *= 2; + values->counterrawid = realloc(values->counterrawid, + values->counters_max * sizeof(*values->counterrawid)); + values->countername = realloc(values->countername, + values->counters_max * sizeof(*values->countername)); + if (!values->counterrawid || !values->countername) + die("failed to enlarge read_values counters arrays"); + + for (i = 0; i < values->threads; i++) { + values->value[i] = realloc(values->value[i], + values->counters_max * sizeof(**values->value)); + if (!values->value[i]) + die("failed to enlarge read_values counters arrays"); + } +} + +static int perf_read_values__findnew_counter(struct perf_read_values *values, + u64 rawid, const char *name) +{ + int i; + + for (i = 0; i < values->counters; i++) + if (values->counterrawid[i] == rawid) + return i; + + if (values->counters == values->counters_max) + perf_read_values__enlarge_counters(values); + + i = values->counters++; + values->counterrawid[i] = rawid; + values->countername[i] = strdup(name); + + return i; +} + +void perf_read_values_add_value(struct perf_read_values *values, + u32 pid, u32 tid, + u64 rawid, const char *name, u64 value) +{ + int tindex, cindex; + + tindex = perf_read_values__findnew_thread(values, pid, tid); + cindex = perf_read_values__findnew_counter(values, rawid, name); + + values->value[tindex][cindex] = value; +} + +static void perf_read_values__display_pretty(FILE *fp, + struct perf_read_values *values) +{ + int i, j; + int pidwidth, tidwidth; + int *counterwidth; + + counterwidth = malloc(values->counters * sizeof(*counterwidth)); + if (!counterwidth) + die("failed to allocate counterwidth array"); + tidwidth = 3; + pidwidth = 3; + for (j = 0; j < values->counters; j++) + counterwidth[j] = strlen(values->countername[j]); + for (i = 0; i < values->threads; i++) { + int width; + + width = snprintf(NULL, 0, "%d", values->pid[i]); + if (width > pidwidth) + pidwidth = width; + width = snprintf(NULL, 0, "%d", values->tid[i]); + if (width > tidwidth) + tidwidth = width; + for (j = 0; j < values->counters; j++) { + width = snprintf(NULL, 0, "%Lu", values->value[i][j]); + if (width > counterwidth[j]) + counterwidth[j] = width; + } + } + + fprintf(fp, "# %*s %*s", pidwidth, "PID", tidwidth, "TID"); + for (j = 0; j < values->counters; j++) + fprintf(fp, " %*s", counterwidth[j], values->countername[j]); + fprintf(fp, "\n"); + + for (i = 0; i < values->threads; i++) { + fprintf(fp, " %*d %*d", pidwidth, values->pid[i], + tidwidth, values->tid[i]); + for (j = 0; j < values->counters; j++) + fprintf(fp, " %*Lu", + counterwidth[j], values->value[i][j]); + fprintf(fp, "\n"); + } +} + +static void perf_read_values__display_raw(FILE *fp, + struct perf_read_values *values) +{ + int width, pidwidth, tidwidth, namewidth, rawwidth, countwidth; + int i, j; + + tidwidth = 3; /* TID */ + pidwidth = 3; /* PID */ + namewidth = 4; /* "Name" */ + rawwidth = 3; /* "Raw" */ + countwidth = 5; /* "Count" */ + + for (i = 0; i < values->threads; i++) { + width = snprintf(NULL, 0, "%d", values->pid[i]); + if (width > pidwidth) + pidwidth = width; + width = snprintf(NULL, 0, "%d", values->tid[i]); + if (width > tidwidth) + tidwidth = width; + } + for (j = 0; j < values->counters; j++) { + width = strlen(values->countername[j]); + if (width > namewidth) + namewidth = width; + width = snprintf(NULL, 0, "%llx", values->counterrawid[j]); + if (width > rawwidth) + rawwidth = width; + } + for (i = 0; i < values->threads; i++) { + for (j = 0; j < values->counters; j++) { + width = snprintf(NULL, 0, "%Lu", values->value[i][j]); + if (width > countwidth) + countwidth = width; + } + } + + fprintf(fp, "# %*s %*s %*s %*s %*s\n", + pidwidth, "PID", tidwidth, "TID", + namewidth, "Name", rawwidth, "Raw", + countwidth, "Count"); + for (i = 0; i < values->threads; i++) + for (j = 0; j < values->counters; j++) + fprintf(fp, " %*d %*d %*s %*llx %*Lu\n", + pidwidth, values->pid[i], + tidwidth, values->tid[i], + namewidth, values->countername[j], + rawwidth, values->counterrawid[j], + countwidth, values->value[i][j]); +} + +void perf_read_values_display(FILE *fp, struct perf_read_values *values, int raw) +{ + if (raw) + perf_read_values__display_raw(fp, values); + else + perf_read_values__display_pretty(fp, values); +} diff --git a/tools/perf/util/values.h b/tools/perf/util/values.h new file mode 100644 index 00000000000..cadf8cf2a59 --- /dev/null +++ b/tools/perf/util/values.h @@ -0,0 +1,27 @@ +#ifndef _PERF_VALUES_H +#define _PERF_VALUES_H + +#include "types.h" + +struct perf_read_values { + int threads; + int threads_max; + u32 *pid, *tid; + int counters; + int counters_max; + u64 *counterrawid; + char **countername; + u64 **value; +}; + +void perf_read_values_init(struct perf_read_values *values); +void perf_read_values_destroy(struct perf_read_values *values); + +void perf_read_values_add_value(struct perf_read_values *values, + u32 pid, u32 tid, + u64 rawid, const char *name, u64 value); + +void perf_read_values_display(FILE *fp, struct perf_read_values *values, + int raw); + +#endif /* _PERF_VALUES_H */ diff --git a/tools/perf/util/wrapper.c b/tools/perf/util/wrapper.c index 6350d65f6d9..4574ac28396 100644 --- a/tools/perf/util/wrapper.c +++ b/tools/perf/util/wrapper.c @@ -7,7 +7,7 @@ * There's no pack memory to release - but stay close to the Git * version so wrap this away: */ -static inline void release_pack_memory(size_t size, int flag) +static inline void release_pack_memory(size_t size __used, int flag __used) { } @@ -59,7 +59,8 @@ void *xmemdupz(const void *data, size_t len) char *xstrndup(const char *str, size_t len) { char *p = memchr(str, '\0', len); - return xmemdupz(str, p ? p - str : len); + + return xmemdupz(str, p ? (size_t)(p - str) : len); } void *xrealloc(void *ptr, size_t size) |