diff options
| -rw-r--r-- | Documentation/kernel-parameters.txt | 5 | ||||
| -rw-r--r-- | Documentation/trace/events.txt | 9 | ||||
| -rw-r--r-- | Documentation/trace/ring-buffer-design.txt | 955 | ||||
| -rw-r--r-- | arch/x86/kernel/ftrace.c | 4 | ||||
| -rw-r--r-- | include/linux/ftrace_event.h | 4 | ||||
| -rw-r--r-- | include/linux/ring_buffer.h | 1 | ||||
| -rw-r--r-- | include/trace/ftrace.h | 18 | ||||
| -rw-r--r-- | kernel/kprobes.c | 30 | ||||
| -rw-r--r-- | kernel/trace/ftrace.c | 92 | ||||
| -rw-r--r-- | kernel/trace/kmemtrace.c | 145 | ||||
| -rw-r--r-- | kernel/trace/ring_buffer.c | 940 | ||||
| -rw-r--r-- | kernel/trace/trace.c | 208 | ||||
| -rw-r--r-- | kernel/trace/trace.h | 37 | ||||
| -rw-r--r-- | kernel/trace/trace_events.c | 69 | ||||
| -rw-r--r-- | kernel/trace/trace_events_filter.c | 169 | ||||
| -rw-r--r-- | kernel/trace/trace_functions.c | 4 | ||||
| -rw-r--r-- | kernel/trace/trace_functions_graph.c | 164 | ||||
| -rw-r--r-- | kernel/trace/trace_sched_switch.c | 57 | ||||
| -rw-r--r-- | kernel/trace/trace_selftest.c | 1 | ||||
| -rw-r--r-- | kernel/trace/trace_stack.c | 9 | ||||
| -rw-r--r-- | kernel/trace/trace_stat.c | 7 | ||||
| -rw-r--r-- | kernel/trace/trace_stat.h | 2 | ||||
| -rw-r--r-- | kernel/trace/trace_workqueue.c | 32 | ||||
| -rwxr-xr-x | scripts/recordmcount.pl | 1 |
24 files changed, 2302 insertions, 661 deletions
diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index dd1a6d4bb74..81cdb7d5e38 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt @@ -2476,6 +2476,11 @@ and is between 256 and 4096 characters. It is defined in the file trace_buf_size=nn[KMG] [FTRACE] will set tracing buffer size. + trace_event=[event-list] + [FTRACE] Set and start specified trace events in order + to facilitate early boot debugging. + See also Documentation/trace/events.txt + trix= [HW,OSS] MediaTrix AudioTrix Pro Format: <io>,<irq>,<dma>,<dma2>,<sb_io>,<sb_irq>,<sb_dma>,<mpu_io>,<mpu_irq> diff --git a/Documentation/trace/events.txt b/Documentation/trace/events.txt index f157d7594ea..2bcc8d4dea2 100644 --- a/Documentation/trace/events.txt +++ b/Documentation/trace/events.txt @@ -83,6 +83,15 @@ When reading one of these enable files, there are four results: X - there is a mixture of events enabled and disabled ? - this file does not affect any event +2.3 Boot option +--------------- + +In order to facilitate early boot debugging, use boot option: + + trace_event=[event-list] + +The format of this boot option is the same as described in section 2.1. + 3. Defining an event-enabled tracepoint ======================================= diff --git a/Documentation/trace/ring-buffer-design.txt b/Documentation/trace/ring-buffer-design.txt new file mode 100644 index 00000000000..5b1d23d604c --- /dev/null +++ b/Documentation/trace/ring-buffer-design.txt @@ -0,0 +1,955 @@ + Lockless Ring Buffer Design + =========================== + +Copyright 2009 Red Hat Inc. + Author: Steven Rostedt <srostedt@redhat.com> + License: The GNU Free Documentation License, Version 1.2 + (dual licensed under the GPL v2) +Reviewers: Mathieu Desnoyers, Huang Ying, Hidetoshi Seto, + and Frederic Weisbecker. + + +Written for: 2.6.31 + +Terminology used in this Document +--------------------------------- + +tail - where new writes happen in the ring buffer. + +head - where new reads happen in the ring buffer. + +producer - the task that writes into the ring buffer (same as writer) + +writer - same as producer + +consumer - the task that reads from the buffer (same as reader) + +reader - same as consumer. + +reader_page - A page outside the ring buffer used solely (for the most part) + by the reader. + +head_page - a pointer to the page that the reader will use next + +tail_page - a pointer to the page that will be written to next + +commit_page - a pointer to the page with the last finished non nested write. + +cmpxchg - hardware assisted atomic transaction that performs the following: + + A = B iff previous A == C + + R = cmpxchg(A, C, B) is saying that we replace A with B if and only if + current A is equal to C, and we put the old (current) A into R + + R gets the previous A regardless if A is updated with B or not. + + To see if the update was successful a compare of R == C may be used. + +The Generic Ring Buffer +----------------------- + +The ring buffer can be used in either an overwrite mode or in +producer/consumer mode. + +Producer/consumer mode is where the producer were to fill up the +buffer before the consumer could free up anything, the producer +will stop writing to the buffer. This will lose most recent events. + +Overwrite mode is where the produce were to fill up the buffer +before the consumer could free up anything, the producer will +overwrite the older data. This will lose the oldest events. + +No two writers can write at the same time (on the same per cpu buffer), +but a writer may interrupt another writer, but it must finish writing +before the previous writer may continue. This is very important to the +algorithm. The writers act like a "stack". The way interrupts works +enforces this behavior. + + + writer1 start + <preempted> writer2 start + <preempted> writer3 start + writer3 finishes + writer2 finishes + writer1 finishes + +This is very much like a writer being preempted by an interrupt and +the interrupt doing a write as well. + +Readers can happen at any time. But no two readers may run at the +same time, nor can a reader preempt/interrupt another reader. A reader +can not preempt/interrupt a writer, but it may read/consume from the +buffer at the same time as a writer is writing, but the reader must be +on another processor to do so. A reader may read on its own processor +and can be preempted by a writer. + +A writer can preempt a reader, but a reader can not preempt a writer. +But a reader can read the buffer at the same time (on another processor) +as a writer. + +The ring buffer is made up of a list of pages held together by a link list. + +At initialization a reader page is allocated for the reader that is not +part of the ring buffer. + +The head_page, tail_page and commit_page are all initialized to point +to the same page. + +The reader page is initialized to have its next pointer pointing to +the head page, and its previous pointer pointing to a page before +the head page. + +The reader has its own page to use. At start up time, this page is +allocated but is not attached to the list. When the reader wants +to read from the buffer, if its page is empty (like it is on start up) +it will swap its page with the head_page. The old reader page will +become part of the ring buffer and the head_page will be removed. +The page after the inserted page (old reader_page) will become the +new head page. + +Once the new page is given to the reader, the reader could do what +it wants with it, as long as a writer has left that page. + +A sample of how the reader page is swapped: Note this does not +show the head page in the buffer, it is for demonstrating a swap +only. + + +------+ + |reader| RING BUFFER + |page | + +------+ + +---+ +---+ +---+ + | |-->| |-->| | + | |<--| |<--| | + +---+ +---+ +---+ + ^ | ^ | + | +-------------+ | + +-----------------+ + + + +------+ + |reader| RING BUFFER + |page |-------------------+ + +------+ v + | +---+ +---+ +---+ + | | |-->| |-->| | + | | |<--| |<--| |<-+ + | +---+ +---+ +---+ | + | ^ | ^ | | + | | +-------------+ | | + | +-----------------+ | + +------------------------------------+ + + +------+ + |reader| RING BUFFER + |page |-------------------+ + +------+ <---------------+ v + | ^ +---+ +---+ +---+ + | | | |-->| |-->| | + | | | | | |<--| |<-+ + | | +---+ +---+ +---+ | + | | | ^ | | + | | +-------------+ | | + | +-----------------------------+ | + +------------------------------------+ + + +------+ + |buffer| RING BUFFER + |page |-------------------+ + +------+ <---------------+ v + | ^ +---+ +---+ +---+ + | | | | | |-->| | + | | New | | | |<--| |<-+ + | | Reader +---+ +---+ +---+ | + | | page ----^ | | + | | | | + | +-----------------------------+ | + +------------------------------------+ + + + +It is possible that the page swapped is the commit page and the tail page, +if what is in the ring buffer is less than what is held in a buffer page. + + + reader page commit page tail page + | | | + v | | + +---+ | | + | |<----------+ | + | |<------------------------+ + | |------+ + +---+ | + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +This case is still valid for this algorithm. +When the writer leaves the page, it simply goes into the ring buffer +since the reader page still points to the next location in the ring +buffer. + + +The main pointers: + + reader page - The page used solely by the reader and is not part + of the ring buffer (may be swapped in) + + head page - the next page in the ring buffer that will be swapped + with the reader page. + + tail page - the page where the next write will take place. + + commit page - the page that last finished a write. + +The commit page only is updated by the outer most writer in the +writer stack. A writer that preempts another writer will not move the +commit page. + +When data is written into the ring buffer, a position is reserved +in the ring buffer and passed back to the writer. When the writer +is finished writing data into that position, it commits the write. + +Another write (or a read) may take place at anytime during this +transaction. If another write happens it must finish before continuing +with the previous write. + + + Write reserve: + + Buffer page + +---------+ + |written | + +---------+ <--- given back to writer (current commit) + |reserved | + +---------+ <--- tail pointer + | empty | + +---------+ + + Write commit: + + Buffer page + +---------+ + |written | + +---------+ + |written | + +---------+ <--- next positon for write (current commit) + | empty | + +---------+ + + + If a write happens after the first reserve: + + Buffer page + +---------+ + |written | + +---------+ <-- current commit + |reserved | + +---------+ <--- given back to second writer + |reserved | + +---------+ <--- tail pointer + + After second writer commits: + + + Buffer page + +---------+ + |written | + +---------+ <--(last full commit) + |reserved | + +---------+ + |pending | + |commit | + +---------+ <--- tail pointer + + When the first writer commits: + + Buffer page + +---------+ + |written | + +---------+ + |written | + +---------+ + |written | + +---------+ <--(last full commit and tail pointer) + + +The commit pointer points to the last write location that was +committed without preempting another write. When a write that +preempted another write is committed, it only becomes a pending commit +and will not be a full commit till all writes have been committed. + +The commit page points to the page that has the last full commit. +The tail page points to the page with the last write (before +committing). + +The tail page is always equal to or after the commit page. It may +be several pages ahead. If the tail page catches up to the commit +page then no more writes may take place (regardless of the mode +of the ring buffer: overwrite and produce/consumer). + +The order of pages are: + + head page + commit page + tail page + +Possible scenario: + tail page + head page commit page | + | | | + v v v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +There is a special case that the head page is after either the commit page +and possibly the tail page. That is when the commit (and tail) page has been +swapped with the reader page. This is because the head page is always +part of the ring buffer, but the reader page is not. When ever there +has been less than a full page that has been committed inside the ring buffer, +and a reader swaps out a page, it will be swapping out the commit page. + + + reader page commit page tail page + | | | + v | | + +---+ | | + | |<----------+ | + | |<------------------------+ + | |------+ + +---+ | + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + ^ + | + head page + + +In this case, the head page will not move when the tail and commit +move back into the ring buffer. + +The reader can not swap a page into the ring buffer if the commit page +is still on that page. If the read meets the last commit (real commit +not pending or reserved), then there is nothing more to read. +The buffer is considered empty until another full commit finishes. + +When the tail meets the head page, if the buffer is in overwrite mode, +the head page will be pushed ahead one. If the buffer is in producer/consumer +mode, the write will fail. + +Overwrite mode: + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + ^ + | + head page + + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + ^ + | + head page + + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + ^ + | + head page + +Note, the reader page will still point to the previous head page. +But when a swap takes place, it will use the most recent head page. + + +Making the Ring Buffer Lockless: +-------------------------------- + +The main idea behind the lockless algorithm is to combine the moving +of the head_page pointer with the swapping of pages with the reader. +State flags are placed inside the pointer to the page. To do this, +each page must be aligned in memory by 4 bytes. This will allow the 2 +least significant bits of the address to be used as flags. Since +they will always be zero for the address. To get the address, +simply mask out the flags. + + MASK = ~3 + + address & MASK + +Two flags will be kept by these two bits: + + HEADER - the page being pointed to is a head page + + UPDATE - the page being pointed to is being updated by a writer + and was or is about to be a head page. + + + reader page + | + v + +---+ + | |------+ + +---+ | + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-H->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + + +The above pointer "-H->" would have the HEADER flag set. That is +the next page is the next page to be swapped out by the reader. +This pointer means the next page is the head page. + +When the tail page meets the head pointer, it will use cmpxchg to +change the pointer to the UPDATE state: + + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-H->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +"-U->" represents a pointer in the UPDATE state. + +Any access to the reader will need to take some sort of lock to serialize +the readers. But the writers will never take a lock to write to the +ring buffer. This means we only need to worry about a single reader, +and writes only preempt in "stack" formation. + +When the reader tries to swap the page with the ring buffer, it +will also use cmpxchg. If the flag bit in the pointer to the +head page does not have the HEADER flag set, the compare will fail +and the reader will need to look for the new head page and try again. +Note, the flag UPDATE and HEADER are never set at the same time. + +The reader swaps the reader page as follows: + + +------+ + |reader| RING BUFFER + |page | + +------+ + +---+ +---+ +---+ + | |--->| |--->| | + | |<---| |<---| | + +---+ +---+ +---+ + ^ | ^ | + | +---------------+ | + +-----H-------------+ + +The reader sets the reader page next pointer as HEADER to the page after +the head page. + + + +------+ + |reader| RING BUFFER + |page |-------H-----------+ + +------+ v + | +---+ +---+ +---+ + | | |--->| |--->| | + | | |<---| |<---| |<-+ + | +---+ +---+ +---+ | + | ^ | ^ | | + | | +---------------+ | | + | +-----H-------------+ | + +--------------------------------------+ + +It does a cmpxchg with the pointer to the previous head page to make it +point to the reader page. Note that the new pointer does not have the HEADER +flag set. This action atomically moves the head page forward. + + +------+ + |reader| RING BUFFER + |page |-------H-----------+ + +------+ v + | ^ +---+ +---+ +---+ + | | | |-->| |-->| | + | | | |<--| |<--| |<-+ + | | +---+ +---+ +---+ | + | | | ^ | | + | | +-------------+ | | + | +-----------------------------+ | + +------------------------------------+ + +After the new head page is set, the previous pointer of the head page is +updated to the reader page. + + +------+ + |reader| RING BUFFER + |page |-------H-----------+ + +------+ <---------------+ v + | ^ +---+ +---+ +---+ + | | | |-->| |-->| | + | | | | | |<--| |<-+ + | | +---+ +---+ +---+ | + | | | ^ | | + | | +-------------+ | | + | +-----------------------------+ | + +------------------------------------+ + + +------+ + |buffer| RING BUFFER + |page |-------H-----------+ <--- New head page + +------+ <---------------+ v + | ^ +---+ +---+ +---+ + | | | | | |-->| | + | | New | | | |<--| |<-+ + | | Reader +---+ +---+ +---+ | + | | page ----^ | | + | | | | + | +-----------------------------+ | + +------------------------------------+ + +Another important point. The page that the reader page points back to +by its previous pointer (the one that now points to the new head page) +never points back to the reader page. That is because the reader page is +not part of the ring buffer. Traversing the ring buffer via the next pointers +will always stay in the ring buffer. Traversing the ring buffer via the +prev pointers may not. + +Note, the way to determine a reader page is simply by examining the previous +pointer of the page. If the next pointer of the previous page does not +point back to the original page, then the original page is a reader page: + + + +--------+ + | reader | next +----+ + | page |-------->| |<====== (buffer page) + +--------+ +----+ + | | ^ + | v | next + prev | +----+ + +------------->| | + +----+ + +The way the head page moves forward: + +When the tail page meets the head page and the buffer is in overwrite mode +and more writes take place, the head page must be moved forward before the +writer may move the tail page. The way this is done is that the writer +performs a cmpxchg to convert the pointer to the head page from the HEADER +flag to have the UPDATE flag set. Once this is done, the reader will +not be able to swap the head page from the buffer, nor will it be able to +move the head page, until the writer is finished with the move. + +This eliminates any races that the reader can have on the writer. The reader +must spin, and this is why the reader can not preempt the writer. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-H->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +The following page will be made into the new head page. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-H->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +After the new head page has been set, we can set the old head page +pointer back to NORMAL. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |-H->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +After the head page has been moved, the tail page may now move forward. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |-H->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + + +The above are the trivial updates. Now for the more complex scenarios. + + +As stated before, if enough writes preempt the first write, the +tail page may make it all the way around the buffer and meet the commit +page. At this time, we must start dropping writes (usually with some kind +of warning to the user). But what happens if the commit was still on the +reader page? The commit page is not part of the ring buffer. The tail page +must account for this. + + + reader page commit page + | | + v | + +---+ | + | |<----------+ + | | + | |------+ + +---+ | + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-H->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + ^ + | + tail page + +If the tail page were to simply push the head page forward, the commit when +leaving the reader page would not be pointing to the correct page. + +The solution to this is to test if the commit page is on the reader page +before pushing the head page. If it is, then it can be assumed that the +tail page wrapped the buffer, and we must drop new writes. + +This is not a race condition, because the commit page can only be moved +by the outter most writer (the writer that was preempted). +This means that the commit will not move while a writer is moving the +tail page. The reader can not swap the reader page if it is also being +used as the commit page. The reader can simply check that the commit +is off the reader page. Once the commit page leaves the reader page +it will never go back on it unless a reader does another swap with the +buffer page that is also the commit page. + + +Nested writes +------------- + +In the pushing forward of the tail page we must first push forward +the head page if the head page is the next page. If the head page +is not the next page, the tail page is simply updated with a cmpxchg. + +Only writers move the tail page. This must be done atomically to protect +against nested writers. + + temp_page = tail_page + next_page = temp_page->next + cmpxchg(tail_page, temp_page, next_page) + +The above will update the tail page if it is still pointing to the expected +page. If this fails, a nested write pushed it forward, the the current write +does not need to push it. + + + temp page + | + v + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +Nested write comes in and moves the tail page forward: + + tail page (moved by nested writer) + temp page | + | | + v v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +The above would fail the cmpxchg, but since the tail page has already +been moved forward, the writer will just try again to reserve storage +on the new tail page. + +But the moving of the head page is a bit more complex. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-H->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +The write converts the head page pointer to UPDATE. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +But if a nested writer preempts here. It will see that the next +page is a head page, but it is also nested. It will detect that +it is nested and will save that information. The detection is the +fact that it sees the UPDATE flag instead of a HEADER or NORMAL +pointer. + +The nested writer will set the new head page pointer. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-H->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +But it will not reset the update back to normal. Only the writer +that converted a pointer from HEAD to UPDATE will convert it back +to NORMAL. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-H->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +After the nested writer finishes, the outer most writer will convert +the UPDATE pointer to NORMAL. + + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |-H->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + + +It can be even more complex if several nested writes came in and moved +the tail page ahead several pages: + + +(first writer) + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-H->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +The write converts the head page pointer to UPDATE. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |--->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +Next writer comes in, and sees the update and sets up the new +head page. + +(second writer) + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-H->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +The nested writer moves the tail page forward. But does not set the old +update page to NORMAL because it is not the outer most writer. + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-H->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +Another writer preempts and sees the page after the tail page is a head page. +It changes it from HEAD to UPDATE. + +(third writer) + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-U->| |---> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +The writer will move the head page forward: + + +(third writer) + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-U->| |-H-> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +But now that the third writer did change the HEAD flag to UPDATE it +will convert it to normal: + + +(third writer) + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |--->| |-H-> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + + +Then it will move the tail page, and return back to the second writer. + + +(second writer) + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |--->| |-H-> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + + +The second writer will fail to move the tail page because it was already +moved, so it will try again and add its data to the new tail page. +It will return to the first writer. + + +(first writer) + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |--->| |-H-> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +The first writer can not know atomically test if the tail page moved +while it updates the HEAD page. It will then update the head page to +what it thinks is the new head page. + + +(first writer) + + tail page + | + v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-H->| |-H-> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +Since the cmpxchg returns the old value of the pointer the first writer +will see it succeeded in updating the pointer from NORMAL to HEAD. +But as we can see, this is not good enough. It must also check to see +if the tail page is either where it use to be or on the next page: + + +(first writer) + + A B tail page + | | | + v v v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |-H->| |-H-> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +If tail page != A and tail page does not equal B, then it must reset the +pointer back to NORMAL. The fact that it only needs to worry about +nested writers, it only needs to check this after setting the HEAD page. + + +(first writer) + + A B tail page + | | | + v v v + +---+ +---+ +---+ +---+ +<---| |--->| |-U->| |--->| |-H-> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + +Now the writer can update the head page. This is also why the head page must +remain in UPDATE and only reset by the outer most writer. This prevents +the reader from seeing the incorrect head page. + + +(first writer) + + A B tail page + | | | + v v v + +---+ +---+ +---+ +---+ +<---| |--->| |--->| |--->| |-H-> +--->| |<---| |<---| |<---| |<--- + +---+ +---+ +---+ +---+ + diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index d94e1ea3b9f..8e9663413b7 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -417,10 +417,6 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr, unsigned long return_hooker = (unsigned long) &return_to_handler; - /* Nmi's are currently unsupported */ - if (unlikely(in_nmi())) - return; - if (unlikely(atomic_read(¤t->tracing_graph_pause))) return; diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h index a81170de7f6..ac8c6f8cf24 100644 --- a/include/linux/ftrace_event.h +++ b/include/linux/ftrace_event.h @@ -103,6 +103,8 @@ void trace_current_buffer_discard_commit(struct ring_buffer_event *event); void tracing_record_cmdline(struct task_struct *tsk); +struct event_filter; + struct ftrace_event_call { struct list_head list; char *name; @@ -118,7 +120,7 @@ struct ftrace_event_call { int (*define_fields)(void); struct list_head fields; int filter_active; - void *filter; + struct event_filter *filter; void *mod; atomic_t profile_count; diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h index 29f8599e6be..7fca71693ae 100644 --- a/include/linux/ring_buffer.h +++ b/include/linux/ring_buffer.h @@ -170,7 +170,6 @@ unsigned long ring_buffer_overruns(struct ring_buffer *buffer); unsigned long ring_buffer_entries_cpu(struct ring_buffer *buffer, int cpu); unsigned long ring_buffer_overrun_cpu(struct ring_buffer *buffer, int cpu); unsigned long ring_buffer_commit_overrun_cpu(struct ring_buffer *buffer, int cpu); -unsigned long ring_buffer_nmi_dropped_cpu(struct ring_buffer *buffer, int cpu); u64 ring_buffer_time_stamp(struct ring_buffer *buffer, int cpu); void ring_buffer_normalize_time_stamp(struct ring_buffer *buffer, diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h index f64fbaae781..25d3b02a06f 100644 --- a/include/trace/ftrace.h +++ b/include/trace/ftrace.h @@ -25,7 +25,7 @@ #define __array(type, item, len) type item[len]; #undef __dynamic_array -#define __dynamic_array(type, item, len) unsigned short __data_loc_##item; +#define __dynamic_array(type, item, len) u32 __data_loc_##item; #undef __string #define __string(item, src) __dynamic_array(char, item, -1) @@ -51,13 +51,14 @@ * Include the following: * * struct ftrace_data_offsets_<call> { - * int <item1>; - * int <item2>; + * u32 <item1>; + * u32 <item2>; * [...] * }; * - * The __dynamic_array() macro will create each int <item>, this is + * The __dynamic_array() macro will create each u32 <item>, this is * to keep the offset of each array from the beginning of the event. + * The size of an array is also encoded, in the higher 16 bits of <item>. */ #undef __field @@ -67,7 +68,7 @@ #define __array(type, item, len) #undef __dynamic_array -#define __dynamic_array(type, item, len) int item; +#define __dynamic_array(type, item, len) u32 item; #undef __string #define __string(item, src) __dynamic_array(char, item, -1) @@ -120,7 +121,7 @@ #undef __dynamic_array #define __dynamic_array(type, item, len) \ - ret = trace_seq_printf(s, "\tfield:__data_loc " #item ";\t" \ + ret = trace_seq_printf(s, "\tfield:__data_loc " #type "[] " #item ";\t"\ "offset:%u;\tsize:%u;\n", \ (unsigned int)offsetof(typeof(field), \ __data_loc_##item), \ @@ -210,7 +211,7 @@ ftrace_format_##call(struct trace_seq *s) \ #undef __get_dynamic_array #define __get_dynamic_array(field) \ - ((void *)__entry + __entry->__data_loc_##field) + ((void *)__entry + (__entry->__data_loc_##field & 0xffff)) #undef __get_str #define __get_str(field) (char *)__get_dynamic_array(field) @@ -282,7 +283,7 @@ ftrace_raw_output_##call(struct trace_iterator *iter, int flags) \ #undef __dynamic_array #define __dynamic_array(type, item, len) \ - ret = trace_define_field(event_call, "__data_loc" "[" #type "]", #item,\ + ret = trace_define_field(event_call, "__data_loc " #type "[]", #item, \ offsetof(typeof(field), __data_loc_##item), \ sizeof(field.__data_loc_##item), 0); @@ -328,6 +329,7 @@ ftrace_define_fields_##call(void) \ #define __dynamic_array(type, item, len) \ __data_offsets->item = __data_size + \ offsetof(typeof(*entry), __data); \ + __data_offsets->item |= (len * sizeof(type)) << 16; \ __data_size += (len) * sizeof(type); #undef __string diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 0540948e29a..ef177d653b2 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -103,7 +103,7 @@ static struct kprobe_blackpoint kprobe_blacklist[] = { #define INSNS_PER_PAGE (PAGE_SIZE/(MAX_INSN_SIZE * sizeof(kprobe_opcode_t))) struct kprobe_insn_page { - struct hlist_node hlist; + struct list_head list; kprobe_opcode_t *insns; /* Page of instruction slots */ char slot_used[INSNS_PER_PAGE]; int nused; @@ -117,7 +117,7 @@ enum kprobe_slot_state { }; static DEFINE_MUTEX(kprobe_insn_mutex); /* Protects kprobe_insn_pages */ -static struct hlist_head kprobe_insn_pages; +static LIST_HEAD(kprobe_insn_pages); static int kprobe_garbage_slots; static int collect_garbage_slots(void); @@ -152,10 +152,9 @@ loop_end: static kprobe_opcode_t __kprobes *__get_insn_slot(void) { struct kprobe_insn_page *kip; - struct hlist_node *pos; retry: - hlist_for_each_entry(kip, pos, &kprobe_insn_pages, hlist) { + list_for_each_entry(kip, &kprobe_insn_pages, list) { if (kip->nused < INSNS_PER_PAGE) { int i; for (i = 0; i < INSNS_PER_PAGE; i++) { @@ -189,8 +188,8 @@ static kprobe_opcode_t __kprobes *__get_insn_slot(void) kfree(kip); return NULL; } - INIT_HLIST_NODE(&kip->hlist); - hlist_add_head(&kip->hlist, &kprobe_insn_pages); + INIT_LIST_HEAD(&kip->list); + list_add(&kip->list, &kprobe_insn_pages); memset(kip->slot_used, SLOT_CLEAN, INSNS_PER_PAGE); kip->slot_used[0] = SLOT_USED; kip->nused = 1; @@ -219,12 +218,8 @@ static int __kprobes collect_one_slot(struct kprobe_insn_page *kip, int idx) * so as not to have to set it up again the * next time somebody inserts a probe. */ - hlist_del(&kip->hlist); - if (hlist_empty(&kprobe_insn_pages)) { - INIT_HLIST_NODE(&kip->hlist); - hlist_add_head(&kip->hlist, - &kprobe_insn_pages); - } else { + if (!list_is_singular(&kprobe_insn_pages)) { + list_del(&kip->list); module_free(NULL, kip->insns); kfree(kip); } @@ -235,14 +230,13 @@ static int __kprobes collect_one_slot(struct kprobe_insn_page *kip, int idx) static int __kprobes collect_garbage_slots(void) { - struct kprobe_insn_page *kip; - struct hlist_node *pos, *next; + struct kprobe_insn_page *kip, *next; /* Ensure no-one is preepmted on the garbages */ if (check_safety()) return -EAGAIN; - hlist_for_each_entry_safe(kip, pos, next, &kprobe_insn_pages, hlist) { + list_for_each_entry_safe(kip, next, &kprobe_insn_pages, list) { int i; if (kip->ngarbage == 0) continue; @@ -260,19 +254,17 @@ static int __kprobes collect_garbage_slots(void) void __kprobes free_insn_slot(kprobe_opcode_t * slot, int dirty) { struct kprobe_insn_page *kip; - struct hlist_node *pos; mutex_lock(&kprobe_insn_mutex); - hlist_for_each_entry(kip, pos, &kprobe_insn_pages, hlist) { + list_for_each_entry(kip, &kprobe_insn_pages, list) { if (kip->insns <= slot && slot < kip->insns + (INSNS_PER_PAGE * MAX_INSN_SIZE)) { int i = (slot - kip->insns) / MAX_INSN_SIZE; if (dirty) { kip->slot_used[i] = SLOT_DIRTY; kip->ngarbage++; - } else { + } else collect_one_slot(kip, i); - } break; } } diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 1e1d23c2630..094863416b2 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1016,71 +1016,35 @@ static int __ftrace_replace_code(struct dyn_ftrace *rec, int enable) { unsigned long ftrace_addr; - unsigned long ip, fl; + unsigned long flag = 0UL; ftrace_addr = (unsigned long)FTRACE_ADDR; - ip = rec->ip; - /* - * If this record is not to be traced and - * it is not enabled then do nothing. + * If this record is not to be traced or we want to disable it, + * then disable it. * - * If this record is not to be traced and - * it is enabled then disable it. + * If we want to enable it and filtering is off, then enable it. * + * If we want to enable it and filtering is on, enable it only if + * it's filtered */ - if (rec->flags & FTRACE_FL_NOTRACE) { - if (rec->flags & FTRACE_FL_ENABLED) - rec->flags &= ~FTRACE_FL_ENABLED; - else - return 0; - - } else if (ftrace_filtered && enable) { - /* - * Filtering is on: - */ - - fl = rec->flags & (FTRACE_FL_FILTER | FTRACE_FL_ENABLED); - - /* Record is filtered and enabled, do nothing */ - if (fl == (FTRACE_FL_FILTER | FTRACE_FL_ENABLED)) - return 0; - - /* Record is not filtered or enabled, do nothing */ - if (!fl) - return 0; - - /* Record is not filtered but enabled, disable it */ - if (fl == FTRACE_FL_ENABLED) - rec->flags &= ~FTRACE_FL_ENABLED; - else - /* Otherwise record is filtered but not enabled, enable it */ - rec->flags |= FTRACE_FL_ENABLED; - } else { - /* Disable or not filtered */ - - if (enable) { - /* if record is enabled, do nothing */ - if (rec->flags & FTRACE_FL_ENABLED) - return 0; - - rec->flags |= FTRACE_FL_ENABLED; - - } else { + if (enable && !(rec->flags & FTRACE_FL_NOTRACE)) { + if (!ftrace_filtered || (rec->flags & FTRACE_FL_FILTER)) + flag = FTRACE_FL_ENABLED; + } - /* if record is not enabled, do nothing */ - if (!(rec->flags & FTRACE_FL_ENABLED)) - return 0; + /* If the state of this record hasn't changed, then do nothing */ + if ((rec->flags & FTRACE_FL_ENABLED) == flag) + return 0; - rec->flags &= ~FTRACE_FL_ENABLED; - } + if (flag) { + rec->flags |= FTRACE_FL_ENABLED; + return ftrace_make_call(rec, ftrace_addr); } - if (rec->flags & FTRACE_FL_ENABLED) - return ftrace_make_call(rec, ftrace_addr); - else - return ftrace_make_nop(NULL, rec, ftrace_addr); + rec->flags &= ~FTRACE_FL_ENABLED; + return ftrace_make_nop(NULL, rec, ftrace_addr); } static void ftrace_replace_code(int enable) @@ -1375,7 +1339,6 @@ struct ftrace_iterator { unsigned flags; unsigned char buffer[FTRACE_BUFF_MAX+1]; unsigned buffer_idx; - unsigned filtered; }; static void * @@ -1438,18 +1401,13 @@ static int t_hash_show(struct seq_file *m, void *v) { struct ftrace_func_probe *rec; struct hlist_node *hnd = v; - char str[KSYM_SYMBOL_LEN]; rec = hlist_entry(hnd, struct ftrace_func_probe, node); if (rec->ops->print) return rec->ops->print(m, rec->ip, rec->ops, rec->data); - kallsyms_lookup(rec->ip, NULL, NULL, NULL, str); - seq_printf(m, "%s:", str); - - kallsyms_lookup((unsigned long)rec->ops->func, NULL, NULL, NULL, str); - seq_printf(m, "%s", str); + seq_printf(m, "%pf:%pf", (void *)rec->ip, (void *)rec->ops->func); if (rec->data) seq_printf(m, ":%p", rec->data); @@ -1547,7 +1505,6 @@ static int t_show(struct seq_file *m, void *v) { struct ftrace_iterator *iter = m->private; struct dyn_ftrace *rec = v; - char str[KSYM_SYMBOL_LEN]; if (iter->flags & FTRACE_ITER_HASH) return t_hash_show(m, v); @@ -1560,9 +1517,7 @@ static int t_show(struct seq_file *m, void *v) if (!rec) return 0; - kallsyms_lookup(rec->ip, NULL, NULL, NULL, str); - - seq_printf(m, "%s\n", str); + seq_printf(m, "%pf\n", (void *)rec->ip); return 0; } @@ -2312,7 +2267,6 @@ ftrace_regex_write(struct file *file, const char __user *ubuf, } if (isspace(ch)) { - iter->filtered++; iter->buffer[iter->buffer_idx] = 0; ret = ftrace_process_regex(iter->buffer, iter->buffer_idx, enable); @@ -2443,7 +2397,6 @@ ftrace_regex_release(struct inode *inode, struct file *file, int enable) iter = file->private_data; if (iter->buffer_idx) { - iter->filtered++; iter->buffer[iter->buffer_idx] = 0; ftrace_match_records(iter->buffer, iter->buffer_idx, enable); } @@ -2543,7 +2496,6 @@ static void g_stop(struct seq_file *m, void *p) static int g_show(struct seq_file *m, void *v) { unsigned long *ptr = v; - char str[KSYM_SYMBOL_LEN]; if (!ptr) return 0; @@ -2553,9 +2505,7 @@ static int g_show(struct seq_file *m, void *v) return 0; } - kallsyms_lookup(*ptr, NULL, NULL, NULL, str); - - seq_printf(m, "%s\n", str); + seq_printf(m, "%pf\n", v); return 0; } diff --git a/kernel/trace/kmemtrace.c b/kernel/trace/kmemtrace.c index 1edaa9516e8..dda53ccf749 100644 --- a/kernel/trace/kmemtrace.c +++ b/kernel/trace/kmemtrace.c @@ -239,12 +239,52 @@ struct kmemtrace_user_event_alloc { }; static enum print_line_t -kmemtrace_print_alloc_user(struct trace_iterator *iter, - struct kmemtrace_alloc_entry *entry) +kmemtrace_print_alloc(struct trace_iterator *iter, int flags) { - struct kmemtrace_user_event_alloc *ev_alloc; struct trace_seq *s = &iter->seq; + struct kmemtrace_alloc_entry *entry; + int ret; + + trace_assign_type(entry, iter->ent); + + ret = trace_seq_printf(s, "type_id %d call_site %pF ptr %lu " + "bytes_req %lu bytes_alloc %lu gfp_flags %lu node %d\n", + entry->type_id, (void *)entry->call_site, (unsigned long)entry->ptr, + (unsigned long)entry->bytes_req, (unsigned long)entry->bytes_alloc, + (unsigned long)entry->gfp_flags, entry->node); + + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + return TRACE_TYPE_HANDLED; +} + +static enum print_line_t +kmemtrace_print_free(struct trace_iterator *iter, int flags) +{ + struct trace_seq *s = &iter->seq; + struct kmemtrace_free_entry *entry; + int ret; + + trace_assign_type(entry, iter->ent); + + ret = trace_seq_printf(s, "type_id %d call_site %pF ptr %lu\n", + entry->type_id, (void *)entry->call_site, + (unsigned long)entry->ptr); + + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + return TRACE_TYPE_HANDLED; +} + +static enum print_line_t +kmemtrace_print_alloc_user(struct trace_iterator *iter, int flags) +{ + struct trace_seq *s = &iter->seq; + struct kmemtrace_alloc_entry *entry; struct kmemtrace_user_event *ev; + struct kmemtrace_user_event_alloc *ev_alloc; + + trace_assign_type(entry, iter->ent); ev = trace_seq_reserve(s, sizeof(*ev)); if (!ev) @@ -271,12 +311,14 @@ kmemtrace_print_alloc_user(struct trace_iterator *iter, } static enum print_line_t -kmemtrace_print_free_user(struct trace_iterator *iter, - struct kmemtrace_free_entry *entry) +kmemtrace_print_free_user(struct trace_iterator *iter, int flags) { struct trace_seq *s = &iter->seq; + struct kmemtrace_free_entry *entry; struct kmemtrace_user_event *ev; + trace_assign_type(entry, iter->ent); + ev = trace_seq_reserve(s, sizeof(*ev)); if (!ev) return TRACE_TYPE_PARTIAL_LINE; @@ -294,12 +336,14 @@ kmemtrace_print_free_user(struct trace_iterator *iter, /* The two other following provide a more minimalistic output */ static enum print_line_t -kmemtrace_print_alloc_compress(struct trace_iterator *iter, - struct kmemtrace_alloc_entry *entry) +kmemtrace_print_alloc_compress(struct trace_iterator *iter) { + struct kmemtrace_alloc_entry *entry; struct trace_seq *s = &iter->seq; int ret; + trace_assign_type(entry, iter->ent); + /* Alloc entry */ ret = trace_seq_printf(s, " + "); if (!ret) @@ -345,29 +389,24 @@ kmemtrace_print_alloc_compress(struct trace_iterator *iter, if (!ret) return TRACE_TYPE_PARTIAL_LINE; - /* Node */ - ret = trace_seq_printf(s, "%4d ", entry->node); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - - /* Call site */ - ret = seq_print_ip_sym(s, entry->call_site, 0); + /* Node and call site*/ + ret = trace_seq_printf(s, "%4d %pf\n", entry->node, + (void *)entry->call_site); if (!ret) return TRACE_TYPE_PARTIAL_LINE; - if (!trace_seq_printf(s, "\n")) - return TRACE_TYPE_PARTIAL_LINE; - return TRACE_TYPE_HANDLED; } static enum print_line_t -kmemtrace_print_free_compress(struct trace_iterator *iter, - struct kmemtrace_free_entry *entry) +kmemtrace_print_free_compress(struct trace_iterator *iter) { + struct kmemtrace_free_entry *entry; struct trace_seq *s = &iter->seq; int ret; + trace_assign_type(entry, iter->ent); + /* Free entry */ ret = trace_seq_printf(s, " - "); if (!ret) @@ -401,19 +440,11 @@ kmemtrace_print_free_compress(struct trace_iterator *iter, if (!ret) return TRACE_TYPE_PARTIAL_LINE; - /* Skip node */ - ret = trace_seq_printf(s, " "); + /* Skip node and print call site*/ + ret = trace_seq_printf(s, " %pf\n", (void *)entry->call_site); if (!ret) return TRACE_TYPE_PARTIAL_LINE; - /* Call site */ - ret = seq_print_ip_sym(s, entry->call_site, 0); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - - if (!trace_seq_printf(s, "\n")) - return TRACE_TYPE_PARTIAL_LINE; - return TRACE_TYPE_HANDLED; } @@ -421,32 +452,31 @@ static enum print_line_t kmemtrace_print_line(struct trace_iterator *iter) { struct trace_entry *entry = iter->ent; - switch (entry->type) { - case TRACE_KMEM_ALLOC: { - struct kmemtrace_alloc_entry *field; - - trace_assign_type(field, entry); - if (kmem_tracer_flags.val & TRACE_KMEM_OPT_MINIMAL) - return kmemtrace_print_alloc_compress(iter, field); - else - return kmemtrace_print_alloc_user(iter, field); - } - - case TRACE_KMEM_FREE: { - struct kmemtrace_free_entry *field; - - trace_assign_type(field, entry); - if (kmem_tracer_flags.val & TRACE_KMEM_OPT_MINIMAL) - return kmemtrace_print_free_compress(iter, field); - else - return kmemtrace_print_free_user(iter, field); - } + if (!(kmem_tracer_flags.val & TRACE_KMEM_OPT_MINIMAL)) + return TRACE_TYPE_UNHANDLED; + switch (entry->type) { + case TRACE_KMEM_ALLOC: + return kmemtrace_print_alloc_compress(iter); + case TRACE_KMEM_FREE: + return kmemtrace_print_free_compress(iter); default: return TRACE_TYPE_UNHANDLED; } } +static struct trace_event kmem_trace_alloc = { + .type = TRACE_KMEM_ALLOC, + .trace = kmemtrace_print_alloc, + .binary = kmemtrace_print_alloc_user, +}; + +static struct trace_event kmem_trace_free = { + .type = TRACE_KMEM_FREE, + .trace = kmemtrace_print_free, + .binary = kmemtrace_print_free_user, +}; + static struct tracer kmem_tracer __read_mostly = { .name = "kmemtrace", .init = kmem_trace_init, @@ -463,6 +493,21 @@ void kmemtrace_init(void) static int __init init_kmem_tracer(void) { - return register_tracer(&kmem_tracer); + if (!register_ftrace_event(&kmem_trace_alloc)) { + pr_warning("Warning: could not register kmem events\n"); + return 1; + } + + if (!register_ftrace_event(&kmem_trace_free)) { + pr_warning("Warning: could not register kmem events\n"); + return 1; + } + + if (!register_tracer(&kmem_tracer)) { + pr_warning("Warning: could not register the kmem tracer\n"); + return 1; + } + + return 0; } device_initcall(init_kmem_tracer); diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index a330513d96c..da2c59d8f48 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -322,6 +322,14 @@ struct buffer_data_page { unsigned char data[]; /* data of buffer page */ }; +/* + * Note, the buffer_page list must be first. The buffer pages + * are allocated in cache lines, which means that each buffer + * page will be at the beginning of a cache line, and thus + * the least significant bits will be zero. We use this to + * add flags in the list struct pointers, to make the ring buffer + * lockless. + */ struct buffer_page { struct list_head list; /* list of buffer pages */ local_t write; /* index for next write */ @@ -330,6 +338,21 @@ struct buffer_page { struct buffer_data_page *page; /* Actual data page */ }; +/* + * The buffer page counters, write and entries, must be reset + * atomically when crossing page boundaries. To synchronize this + * update, two counters are inserted into the number. One is + * the actual counter for the write position or count on the page. + * + * The other is a counter of updaters. Before an update happens + * the update partition of the counter is incremented. This will + * allow the updater to update the counter atomically. + * + * The counter is 20 bits, and the state data is 12. + */ +#define RB_WRITE_MASK 0xfffff +#define RB_WRITE_INTCNT (1 << 20) + static void rb_init_page(struct buffer_data_page *bpage) { local_set(&bpage->commit, 0); @@ -403,21 +426,20 @@ int ring_buffer_print_page_header(struct trace_seq *s) struct ring_buffer_per_cpu { int cpu; struct ring_buffer *buffer; - spinlock_t reader_lock; /* serialize readers */ + spinlock_t reader_lock; /* serialize readers */ raw_spinlock_t lock; struct lock_class_key lock_key; - struct list_head pages; + struct list_head *pages; struct buffer_page *head_page; /* read from head */ struct buffer_page *tail_page; /* write to tail */ struct buffer_page *commit_page; /* committed pages */ struct buffer_page *reader_page; - unsigned long nmi_dropped; - unsigned long commit_overrun; - unsigned long overrun; - unsigned long read; + local_t commit_overrun; + local_t overrun; local_t entries; local_t committing; local_t commits; + unsigned long read; u64 write_stamp; u64 read_stamp; atomic_t record_disabled; @@ -489,6 +511,390 @@ void ring_buffer_normalize_time_stamp(struct ring_buffer *buffer, } EXPORT_SYMBOL_GPL(ring_buffer_normalize_time_stamp); +/* + * Making the ring buffer lockless makes things tricky. + * Although writes only happen on the CPU that they are on, + * and they only need to worry about interrupts. Reads can + * happen on any CPU. + * + * The reader page is always off the ring buffer, but when the + * reader finishes with a page, it needs to swap its page with + * a new one from the buffer. The reader needs to take from + * the head (writes go to the tail). But if a writer is in overwrite + * mode and wraps, it must push the head page forward. + * + * Here lies the problem. + * + * The reader must be careful to replace only the head page, and + * not another one. As described at the top of the file in the + * ASCII art, the reader sets its old page to point to the next + * page after head. It then sets the page after head to point to + * the old reader page. But if the writer moves the head page + * during this operation, the reader could end up with the tail. + * + * We use cmpxchg to help prevent this race. We also do something + * special with the page before head. We set the LSB to 1. + * + * When the writer must push the page forward, it will clear the + * bit that points to the head page, move the head, and then set + * the bit that points to the new head page. + * + * We also don't want an interrupt coming in and moving the head + * page on another writer. Thus we use the second LSB to catch + * that too. Thus: + * + * head->list->prev->next bit 1 bit 0 + * ------- ------- + * Normal page 0 0 + * Points to head page 0 1 + * New head page 1 0 + * + * Note we can not trust the prev pointer of the head page, because: + * + * +----+ +-----+ +-----+ + * | |------>| T |---X--->| N | + * | |<------| | | | + * +----+ +-----+ +-----+ + * ^ ^ | + * | +-----+ | | + * +----------| R |----------+ | + * | |<-----------+ + * +-----+ + * + * Key: ---X--> HEAD flag set in pointer + * T Tail page + * R Reader page + * N Next page + * + * (see __rb_reserve_next() to see where this happens) + * + * What the above shows is that the reader just swapped out + * the reader page with a page in the buffer, but before it + * could make the new header point back to the new page added + * it was preempted by a writer. The writer moved forward onto + * the new page added by the reader and is about to move forward + * again. + * + * You can see, it is legitimate for the previous pointer of + * the head (or any page) not to point back to itself. But only + * temporarially. + */ + +#define RB_PAGE_NORMAL 0UL +#define RB_PAGE_HEAD 1UL +#define RB_PAGE_UPDATE 2UL + + +#define RB_FLAG_MASK 3UL + +/* PAGE_MOVED is not part of the mask */ +#define RB_PAGE_MOVED 4UL + +/* + * rb_list_head - remove any bit + */ +static struct list_head *rb_list_head(struct list_head *list) +{ + unsigned long val = (unsigned long)list; + + return (struct list_head *)(val & ~RB_FLAG_MASK); +} + +/* + * rb_is_head_page - test if the give page is the head page + * + * Because the reader may move the head_page pointer, we can + * not trust what the head page is (it may be pointing to + * the reader page). But if the next page is a header page, + * its flags will be non zero. + */ +static int inline +rb_is_head_page(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page *page, struct list_head *list) +{ + unsigned long val; + + val = (unsigned long)list->next; + + if ((val & ~RB_FLAG_MASK) != (unsigned long)&page->list) + return RB_PAGE_MOVED; + + return val & RB_FLAG_MASK; +} + +/* + * rb_is_reader_page + * + * The unique thing about the reader page, is that, if the + * writer is ever on it, the previous pointer never points + * back to the reader page. + */ +static int rb_is_reader_page(struct buffer_page *page) +{ + struct list_head *list = page->list.prev; + + return rb_list_head(list->next) != &page->list; +} + +/* + * rb_set_list_to_head - set a list_head to be pointing to head. + */ +static void rb_set_list_to_head(struct ring_buffer_per_cpu *cpu_buffer, + struct list_head *list) +{ + unsigned long *ptr; + + ptr = (unsigned long *)&list->next; + *ptr |= RB_PAGE_HEAD; + *ptr &= ~RB_PAGE_UPDATE; +} + +/* + * rb_head_page_activate - sets up head page + */ +static void rb_head_page_activate(struct ring_buffer_per_cpu *cpu_buffer) +{ + struct buffer_page *head; + + head = cpu_buffer->head_page; + if (!head) + return; + + /* + * Set the previous list pointer to have the HEAD flag. + */ + rb_set_list_to_head(cpu_buffer, head->list.prev); +} + +static void rb_list_head_clear(struct list_head *list) +{ + unsigned long *ptr = (unsigned long *)&list->next; + + *ptr &= ~RB_FLAG_MASK; +} + +/* + * rb_head_page_dactivate - clears head page ptr (for free list) + */ +static void +rb_head_page_deactivate(struct ring_buffer_per_cpu *cpu_buffer) +{ + struct list_head *hd; + + /* Go through the whole list and clear any pointers found. */ + rb_list_head_clear(cpu_buffer->pages); + + list_for_each(hd, cpu_buffer->pages) + rb_list_head_clear(hd); +} + +static int rb_head_page_set(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page *head, + struct buffer_page *prev, + int old_flag, int new_flag) +{ + struct list_head *list; + unsigned long val = (unsigned long)&head->list; + unsigned long ret; + + list = &prev->list; + + val &= ~RB_FLAG_MASK; + + ret = (unsigned long)cmpxchg(&list->next, + val | old_flag, val | new_flag); + + /* check if the reader took the page */ + if ((ret & ~RB_FLAG_MASK) != val) + return RB_PAGE_MOVED; + + return ret & RB_FLAG_MASK; +} + +static int rb_head_page_set_update(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page *head, + struct buffer_page *prev, + int old_flag) +{ + return rb_head_page_set(cpu_buffer, head, prev, + old_flag, RB_PAGE_UPDATE); +} + +static int rb_head_page_set_head(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page *head, + struct buffer_page *prev, + int old_flag) +{ + return rb_head_page_set(cpu_buffer, head, prev, + old_flag, RB_PAGE_HEAD); +} + +static int rb_head_page_set_normal(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page *head, + struct buffer_page *prev, + int old_flag) +{ + return rb_head_page_set(cpu_buffer, head, prev, + old_flag, RB_PAGE_NORMAL); +} + +static inline void rb_inc_page(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page **bpage) +{ + struct list_head *p = rb_list_head((*bpage)->list.next); + + *bpage = list_entry(p, struct buffer_page, list); +} + +static struct buffer_page * +rb_set_head_page(struct ring_buffer_per_cpu *cpu_buffer) +{ + struct buffer_page *head; + struct buffer_page *page; + struct list_head *list; + int i; + + if (RB_WARN_ON(cpu_buffer, !cpu_buffer->head_page)) + return NULL; + + /* sanity check */ + list = cpu_buffer->pages; + if (RB_WARN_ON(cpu_buffer, rb_list_head(list->prev->next) != list)) + return NULL; + + page = head = cpu_buffer->head_page; + /* + * It is possible that the writer moves the header behind + * where we started, and we miss in one loop. + * A second loop should grab the header, but we'll do + * three loops just because I'm paranoid. + */ + for (i = 0; i < 3; i++) { + do { + if (rb_is_head_page(cpu_buffer, page, page->list.prev)) { + cpu_buffer->head_page = page; + return page; + } + rb_inc_page(cpu_buffer, &page); + } while (page != head); + } + + RB_WARN_ON(cpu_buffer, 1); + + return NULL; +} + +static int rb_head_page_replace(struct buffer_page *old, + struct buffer_page *new) +{ + unsigned long *ptr = (unsigned long *)&old->list.prev->next; + unsigned long val; + unsigned long ret; + + val = *ptr & ~RB_FLAG_MASK; + val |= RB_PAGE_HEAD; + + ret = cmpxchg(ptr, val, &new->list); + + return ret == val; +} + +/* + * rb_tail_page_update - move the tail page forward + * + * Returns 1 if moved tail page, 0 if someone else did. + */ +static int rb_tail_page_update(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page *tail_page, + struct buffer_page *next_page) +{ + struct buffer_page *old_tail; + unsigned long old_entries; + unsigned long old_write; + int ret = 0; + + /* + * The tail page now needs to be moved forward. + * + * We need to reset the tail page, but without messing + * with possible erasing of data brought in by interrupts + * that have moved the tail page and are currently on it. + * + * We add a counter to the write field to denote this. + */ + old_write = local_add_return(RB_WRITE_INTCNT, &next_page->write); + old_entries = local_add_return(RB_WRITE_INTCNT, &next_page->entries); + + /* + * Just make sure we have seen our old_write and synchronize + * with any interrupts that come in. + */ + barrier(); + + /* + * If the tail page is still the same as what we think + * it is, then it is up to us to update the tail + * pointer. + */ + if (tail_page == cpu_buffer->tail_page) { + /* Zero the write counter */ + unsigned long val = old_write & ~RB_WRITE_MASK; + unsigned long eval = old_entries & ~RB_WRITE_MASK; + + /* + * This will only succeed if an interrupt did + * not come in and change it. In which case, we + * do not want to modify it. + * + * We add (void) to let the compiler know that we do not care + * about the return value of these functions. We use the + * cmpxchg to only update if an interrupt did not already + * do it for us. If the cmpxchg fails, we don't care. + */ + (void)local_cmpxchg(&next_page->write, old_write, val); + (void)local_cmpxchg(&next_page->entries, old_entries, eval); + + /* + * No need to worry about races with clearing out the commit. + * it only can increment when a commit takes place. But that + * only happens in the outer most nested commit. + */ + local_set(&next_page->page->commit, 0); + + old_tail = cmpxchg(&cpu_buffer->tail_page, + tail_page, next_page); + + if (old_tail == tail_page) + ret = 1; + } + + return ret; +} + +static int rb_check_bpage(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page *bpage) +{ + unsigned long val = (unsigned long)bpage; + + if (RB_WARN_ON(cpu_buffer, val & RB_FLAG_MASK)) + return 1; + + return 0; +} + +/** + * rb_check_list - make sure a pointer to a list has the last bits zero + */ +static int rb_check_list(struct ring_buffer_per_cpu *cpu_buffer, + struct list_head *list) +{ + if (RB_WARN_ON(cpu_buffer, rb_list_head(list->prev) != list->prev)) + return 1; + if (RB_WARN_ON(cpu_buffer, rb_list_head(list->next) != list->next)) + return 1; + return 0; +} + /** * check_pages - integrity check of buffer pages * @cpu_buffer: CPU buffer with pages to test @@ -498,14 +904,19 @@ EXPORT_SYMBOL_GPL(ring_buffer_normalize_time_stamp); */ static int rb_check_pages(struct ring_buffer_per_cpu *cpu_buffer) { - struct list_head *head = &cpu_buffer->pages; + struct list_head *head = cpu_buffer->pages; struct buffer_page *bpage, *tmp; + rb_head_page_deactivate(cpu_buffer); + if (RB_WARN_ON(cpu_buffer, head->next->prev != head)) return -1; if (RB_WARN_ON(cpu_buffer, head->prev->next != head)) return -1; + if (rb_check_list(cpu_buffer, head)) + return -1; + list_for_each_entry_safe(bpage, tmp, head, list) { if (RB_WARN_ON(cpu_buffer, bpage->list.next->prev != &bpage->list)) @@ -513,25 +924,33 @@ static int rb_check_pages(struct ring_buffer_per_cpu *cpu_buffer) if (RB_WARN_ON(cpu_buffer, bpage->list.prev->next != &bpage->list)) return -1; + if (rb_check_list(cpu_buffer, &bpage->list)) + return -1; } + rb_head_page_activate(cpu_buffer); + return 0; } static int rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer, unsigned nr_pages) { - struct list_head *head = &cpu_buffer->pages; struct buffer_page *bpage, *tmp; unsigned long addr; LIST_HEAD(pages); unsigned i; + WARN_ON(!nr_pages); + for (i = 0; i < nr_pages; i++) { bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()), GFP_KERNEL, cpu_to_node(cpu_buffer->cpu)); if (!bpage) goto free_pages; + + rb_check_bpage(cpu_buffer, bpage); + list_add(&bpage->list, &pages); addr = __get_free_page(GFP_KERNEL); @@ -541,7 +960,13 @@ static int rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer, rb_init_page(bpage->page); } - list_splice(&pages, head); + /* + * The ring buffer page list is a circular list that does not + * start and end with a list head. All page list items point to + * other pages. + */ + cpu_buffer->pages = pages.next; + list_del(&pages); rb_check_pages(cpu_buffer); @@ -573,13 +998,14 @@ rb_allocate_cpu_buffer(struct ring_buffer *buffer, int cpu) spin_lock_init(&cpu_buffer->reader_lock); lockdep_set_class(&cpu_buffer->reader_lock, buffer->reader_lock_key); cpu_buffer->lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; - INIT_LIST_HEAD(&cpu_buffer->pages); bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()), GFP_KERNEL, cpu_to_node(cpu)); if (!bpage) goto fail_free_buffer; + rb_check_bpage(cpu_buffer, bpage); + cpu_buffer->reader_page = bpage; addr = __get_free_page(GFP_KERNEL); if (!addr) @@ -594,9 +1020,11 @@ rb_allocate_cpu_buffer(struct ring_buffer *buffer, int cpu) goto fail_free_reader; cpu_buffer->head_page - = list_entry(cpu_buffer->pages.next, struct buffer_page, list); + = list_entry(cpu_buffer->pages, struct buffer_page, list); cpu_buffer->tail_page = cpu_buffer->commit_page = cpu_buffer->head_page; + rb_head_page_activate(cpu_buffer); + return cpu_buffer; fail_free_reader: @@ -609,15 +1037,22 @@ rb_allocate_cpu_buffer(struct ring_buffer *buffer, int cpu) static void rb_free_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer) { - struct list_head *head = &cpu_buffer->pages; + struct list_head *head = cpu_buffer->pages; struct buffer_page *bpage, *tmp; free_buffer_page(cpu_buffer->reader_page); - list_for_each_entry_safe(bpage, tmp, head, list) { - list_del_init(&bpage->list); + rb_head_page_deactivate(cpu_buffer); + + if (head) { + list_for_each_entry_safe(bpage, tmp, head, list) { + list_del_init(&bpage->list); + free_buffer_page(bpage); + } + bpage = list_entry(head, struct buffer_page, list); free_buffer_page(bpage); } + kfree(cpu_buffer); } @@ -760,15 +1195,17 @@ rb_remove_pages(struct ring_buffer_per_cpu *cpu_buffer, unsigned nr_pages) atomic_inc(&cpu_buffer->record_disabled); synchronize_sched(); + rb_head_page_deactivate(cpu_buffer); + for (i = 0; i < nr_pages; i++) { - if (RB_WARN_ON(cpu_buffer, list_empty(&cpu_buffer->pages))) + if (RB_WARN_ON(cpu_buffer, list_empty(cpu_buffer->pages))) return; - p = cpu_buffer->pages.next; + p = cpu_buffer->pages->next; bpage = list_entry(p, struct buffer_page, list); list_del_init(&bpage->list); free_buffer_page(bpage); } - if (RB_WARN_ON(cpu_buffer, list_empty(&cpu_buffer->pages))) + if (RB_WARN_ON(cpu_buffer, list_empty(cpu_buffer->pages))) return; rb_reset_cpu(cpu_buffer); @@ -790,15 +1227,19 @@ rb_insert_pages(struct ring_buffer_per_cpu *cpu_buffer, atomic_inc(&cpu_buffer->record_disabled); synchronize_sched(); + spin_lock_irq(&cpu_buffer->reader_lock); + rb_head_page_deactivate(cpu_buffer); + for (i = 0; i < nr_pages; i++) { if (RB_WARN_ON(cpu_buffer, list_empty(pages))) return; p = pages->next; bpage = list_entry(p, struct buffer_page, list); list_del_init(&bpage->list); - list_add_tail(&bpage->list, &cpu_buffer->pages); + list_add_tail(&bpage->list, cpu_buffer->pages); } rb_reset_cpu(cpu_buffer); + spin_unlock_irq(&cpu_buffer->reader_lock); rb_check_pages(cpu_buffer); @@ -949,21 +1390,14 @@ rb_reader_event(struct ring_buffer_per_cpu *cpu_buffer) } static inline struct ring_buffer_event * -rb_head_event(struct ring_buffer_per_cpu *cpu_buffer) -{ - return __rb_page_index(cpu_buffer->head_page, - cpu_buffer->head_page->read); -} - -static inline struct ring_buffer_event * rb_iter_head_event(struct ring_buffer_iter *iter) { return __rb_page_index(iter->head_page, iter->head); } -static inline unsigned rb_page_write(struct buffer_page *bpage) +static inline unsigned long rb_page_write(struct buffer_page *bpage) { - return local_read(&bpage->write); + return local_read(&bpage->write) & RB_WRITE_MASK; } static inline unsigned rb_page_commit(struct buffer_page *bpage) @@ -971,6 +1405,11 @@ static inline unsigned rb_page_commit(struct buffer_page *bpage) return local_read(&bpage->page->commit); } +static inline unsigned long rb_page_entries(struct buffer_page *bpage) +{ + return local_read(&bpage->entries) & RB_WRITE_MASK; +} + /* Size is determined by what has been commited */ static inline unsigned rb_page_size(struct buffer_page *bpage) { @@ -983,22 +1422,6 @@ rb_commit_index(struct ring_buffer_per_cpu *cpu_buffer) return rb_page_commit(cpu_buffer->commit_page); } -static inline unsigned rb_head_size(struct ring_buffer_per_cpu *cpu_buffer) -{ - return rb_page_commit(cpu_buffer->head_page); -} - -static inline void rb_inc_page(struct ring_buffer_per_cpu *cpu_buffer, - struct buffer_page **bpage) -{ - struct list_head *p = (*bpage)->list.next; - - if (p == &cpu_buffer->pages) - p = p->next; - - *bpage = list_entry(p, struct buffer_page, list); -} - static inline unsigned rb_event_index(struct ring_buffer_event *event) { @@ -1024,6 +1447,8 @@ rb_event_is_commit(struct ring_buffer_per_cpu *cpu_buffer, static void rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer) { + unsigned long max_count; + /* * We only race with interrupts and NMIs on this CPU. * If we own the commit event, then we can commit @@ -1033,9 +1458,16 @@ rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer) * assign the commit to the tail. */ again: + max_count = cpu_buffer->buffer->pages * 100; + while (cpu_buffer->commit_page != cpu_buffer->tail_page) { - cpu_buffer->commit_page->page->commit = - cpu_buffer->commit_page->write; + if (RB_WARN_ON(cpu_buffer, !(--max_count))) + return; + if (RB_WARN_ON(cpu_buffer, + rb_is_reader_page(cpu_buffer->tail_page))) + return; + local_set(&cpu_buffer->commit_page->page->commit, + rb_page_write(cpu_buffer->commit_page)); rb_inc_page(cpu_buffer, &cpu_buffer->commit_page); cpu_buffer->write_stamp = cpu_buffer->commit_page->page->time_stamp; @@ -1044,8 +1476,12 @@ rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer) } while (rb_commit_index(cpu_buffer) != rb_page_write(cpu_buffer->commit_page)) { - cpu_buffer->commit_page->page->commit = - cpu_buffer->commit_page->write; + + local_set(&cpu_buffer->commit_page->page->commit, + rb_page_write(cpu_buffer->commit_page)); + RB_WARN_ON(cpu_buffer, + local_read(&cpu_buffer->commit_page->page->commit) & + ~RB_WRITE_MASK); barrier(); } @@ -1078,7 +1514,7 @@ static void rb_inc_iter(struct ring_buffer_iter *iter) * to the head page instead of next. */ if (iter->head_page == cpu_buffer->reader_page) - iter->head_page = cpu_buffer->head_page; + iter->head_page = rb_set_head_page(cpu_buffer); else rb_inc_page(cpu_buffer, &iter->head_page); @@ -1122,6 +1558,163 @@ rb_update_event(struct ring_buffer_event *event, } } +/* + * rb_handle_head_page - writer hit the head page + * + * Returns: +1 to retry page + * 0 to continue + * -1 on error + */ +static int +rb_handle_head_page(struct ring_buffer_per_cpu *cpu_buffer, + struct buffer_page *tail_page, + struct buffer_page *next_page) +{ + struct buffer_page *new_head; + int entries; + int type; + int ret; + + entries = rb_page_entries(next_page); + + /* + * The hard part is here. We need to move the head + * forward, and protect against both readers on + * other CPUs and writers coming in via interrupts. + */ + type = rb_head_page_set_update(cpu_buffer, next_page, tail_page, + RB_PAGE_HEAD); + + /* + * type can be one of four: + * NORMAL - an interrupt already moved it for us + * HEAD - we are the first to get here. + * UPDATE - we are the interrupt interrupting + * a current move. + * MOVED - a reader on another CPU moved the next + * pointer to its reader page. Give up + * and try again. + */ + + switch (type) { + case RB_PAGE_HEAD: + /* + * We changed the head to UPDATE, thus + * it is our responsibility to update + * the counters. + */ + local_add(entries, &cpu_buffer->overrun); + + /* + * The entries will be zeroed out when we move the + * tail page. + */ + + /* still more to do */ + break; + + case RB_PAGE_UPDATE: + /* + * This is an interrupt that interrupt the + * previous update. Still more to do. + */ + break; + case RB_PAGE_NORMAL: + /* + * An interrupt came in before the update + * and processed this for us. + * Nothing left to do. + */ + return 1; + case RB_PAGE_MOVED: + /* + * The reader is on another CPU and just did + * a swap with our next_page. + * Try again. + */ + return 1; + default: + RB_WARN_ON(cpu_buffer, 1); /* WTF??? */ + return -1; + } + + /* + * Now that we are here, the old head pointer is + * set to UPDATE. This will keep the reader from + * swapping the head page with the reader page. + * The reader (on another CPU) will spin till + * we are finished. + * + * We just need to protect against interrupts + * doing the job. We will set the next pointer + * to HEAD. After that, we set the old pointer + * to NORMAL, but only if it was HEAD before. + * otherwise we are an interrupt, and only + * want the outer most commit to reset it. + */ + new_head = next_page; + rb_inc_page(cpu_buffer, &new_head); + + ret = rb_head_page_set_head(cpu_buffer, new_head, next_page, + RB_PAGE_NORMAL); + + /* + * Valid returns are: + * HEAD - an interrupt came in and already set it. + * NORMAL - One of two things: + * 1) We really set it. + * 2) A bunch of interrupts came in and moved + * the page forward again. + */ + switch (ret) { + case RB_PAGE_HEAD: + case RB_PAGE_NORMAL: + /* OK */ + break; + default: + RB_WARN_ON(cpu_buffer, 1); + return -1; + } + + /* + * It is possible that an interrupt came in, + * set the head up, then more interrupts came in + * and moved it again. When we get back here, + * the page would have been set to NORMAL but we + * just set it back to HEAD. + * + * How do you detect this? Well, if that happened + * the tail page would have moved. + */ + if (ret == RB_PAGE_NORMAL) { + /* + * If the tail had moved passed next, then we need + * to reset the pointer. + */ + if (cpu_buffer->tail_page != tail_page && + cpu_buffer->tail_page != next_page) + rb_head_page_set_normal(cpu_buffer, new_head, + next_page, + RB_PAGE_HEAD); + } + + /* + * If this was the outer most commit (the one that + * changed the original pointer from HEAD to UPDATE), + * then it is up to us to reset it to NORMAL. + */ + if (type == RB_PAGE_HEAD) { + ret = rb_head_page_set_normal(cpu_buffer, next_page, + tail_page, + RB_PAGE_UPDATE); + if (RB_WARN_ON(cpu_buffer, + ret != RB_PAGE_UPDATE)) + return -1; + } + + return 0; +} + static unsigned rb_calculate_event_length(unsigned length) { struct ring_buffer_event event; /* Used only for sizeof array */ @@ -1200,96 +1793,93 @@ rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer, struct buffer_page *commit_page, struct buffer_page *tail_page, u64 *ts) { - struct buffer_page *next_page, *head_page, *reader_page; struct ring_buffer *buffer = cpu_buffer->buffer; - bool lock_taken = false; - unsigned long flags; + struct buffer_page *next_page; + int ret; next_page = tail_page; - local_irq_save(flags); - /* - * Since the write to the buffer is still not - * fully lockless, we must be careful with NMIs. - * The locks in the writers are taken when a write - * crosses to a new page. The locks protect against - * races with the readers (this will soon be fixed - * with a lockless solution). - * - * Because we can not protect against NMIs, and we - * want to keep traces reentrant, we need to manage - * what happens when we are in an NMI. - * - * NMIs can happen after we take the lock. - * If we are in an NMI, only take the lock - * if it is not already taken. Otherwise - * simply fail. - */ - if (unlikely(in_nmi())) { - if (!__raw_spin_trylock(&cpu_buffer->lock)) { - cpu_buffer->nmi_dropped++; - goto out_reset; - } - } else - __raw_spin_lock(&cpu_buffer->lock); - - lock_taken = true; - rb_inc_page(cpu_buffer, &next_page); - head_page = cpu_buffer->head_page; - reader_page = cpu_buffer->reader_page; - - /* we grabbed the lock before incrementing */ - if (RB_WARN_ON(cpu_buffer, next_page == reader_page)) - goto out_reset; - /* * If for some reason, we had an interrupt storm that made * it all the way around the buffer, bail, and warn * about it. */ if (unlikely(next_page == commit_page)) { - cpu_buffer->commit_overrun++; + local_inc(&cpu_buffer->commit_overrun); goto out_reset; } - if (next_page == head_page) { - if (!(buffer->flags & RB_FL_OVERWRITE)) - goto out_reset; - - /* tail_page has not moved yet? */ - if (tail_page == cpu_buffer->tail_page) { - /* count overflows */ - cpu_buffer->overrun += - local_read(&head_page->entries); + /* + * This is where the fun begins! + * + * We are fighting against races between a reader that + * could be on another CPU trying to swap its reader + * page with the buffer head. + * + * We are also fighting against interrupts coming in and + * moving the head or tail on us as well. + * + * If the next page is the head page then we have filled + * the buffer, unless the commit page is still on the + * reader page. + */ + if (rb_is_head_page(cpu_buffer, next_page, &tail_page->list)) { - rb_inc_page(cpu_buffer, &head_page); - cpu_buffer->head_page = head_page; - cpu_buffer->head_page->read = 0; + /* + * If the commit is not on the reader page, then + * move the header page. + */ + if (!rb_is_reader_page(cpu_buffer->commit_page)) { + /* + * If we are not in overwrite mode, + * this is easy, just stop here. + */ + if (!(buffer->flags & RB_FL_OVERWRITE)) + goto out_reset; + + ret = rb_handle_head_page(cpu_buffer, + tail_page, + next_page); + if (ret < 0) + goto out_reset; + if (ret) + goto out_again; + } else { + /* + * We need to be careful here too. The + * commit page could still be on the reader + * page. We could have a small buffer, and + * have filled up the buffer with events + * from interrupts and such, and wrapped. + * + * Note, if the tail page is also the on the + * reader_page, we let it move out. + */ + if (unlikely((cpu_buffer->commit_page != + cpu_buffer->tail_page) && + (cpu_buffer->commit_page == + cpu_buffer->reader_page))) { + local_inc(&cpu_buffer->commit_overrun); + goto out_reset; + } } } - /* - * If the tail page is still the same as what we think - * it is, then it is up to us to update the tail - * pointer. - */ - if (tail_page == cpu_buffer->tail_page) { - local_set(&next_page->write, 0); - local_set(&next_page->entries, 0); - local_set(&next_page->page->commit, 0); - cpu_buffer->tail_page = next_page; - - /* reread the time stamp */ + ret = rb_tail_page_update(cpu_buffer, tail_page, next_page); + if (ret) { + /* + * Nested commits always have zero deltas, so + * just reread the time stamp + */ *ts = rb_time_stamp(buffer, cpu_buffer->cpu); - cpu_buffer->tail_page->page->time_stamp = *ts; + next_page->page->time_stamp = *ts; } - rb_reset_tail(cpu_buffer, tail_page, tail, length); + out_again: - __raw_spin_unlock(&cpu_buffer->lock); - local_irq_restore(flags); + rb_reset_tail(cpu_buffer, tail_page, tail, length); /* fail and let the caller try again */ return ERR_PTR(-EAGAIN); @@ -1298,9 +1888,6 @@ rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer, /* reset write */ rb_reset_tail(cpu_buffer, tail_page, tail, length); - if (likely(lock_taken)) - __raw_spin_unlock(&cpu_buffer->lock); - local_irq_restore(flags); return NULL; } @@ -1317,6 +1904,9 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer, barrier(); tail_page = cpu_buffer->tail_page; write = local_add_return(length, &tail_page->write); + + /* set write to only the index of the write */ + write &= RB_WRITE_MASK; tail = write - length; /* See if we shot pass the end of this buffer page */ @@ -1361,12 +1951,16 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer, bpage = cpu_buffer->tail_page; if (bpage->page == (void *)addr && rb_page_write(bpage) == old_index) { + unsigned long write_mask = + local_read(&bpage->write) & ~RB_WRITE_MASK; /* * This is on the tail page. It is possible that * a write could come in and move the tail page * and write to the next page. That is fine * because we just shorten what is on this page. */ + old_index += write_mask; + new_index += write_mask; index = local_cmpxchg(&bpage->write, old_index, new_index); if (index == old_index) return 1; @@ -1875,9 +2469,13 @@ EXPORT_SYMBOL_GPL(ring_buffer_write); static int rb_per_cpu_empty(struct ring_buffer_per_cpu *cpu_buffer) { struct buffer_page *reader = cpu_buffer->reader_page; - struct buffer_page *head = cpu_buffer->head_page; + struct buffer_page *head = rb_set_head_page(cpu_buffer); struct buffer_page *commit = cpu_buffer->commit_page; + /* In case of error, head will be NULL */ + if (unlikely(!head)) + return 1; + return reader->read == rb_page_commit(reader) && (commit == reader || (commit == head && @@ -1968,7 +2566,7 @@ unsigned long ring_buffer_entries_cpu(struct ring_buffer *buffer, int cpu) return 0; cpu_buffer = buffer->buffers[cpu]; - ret = (local_read(&cpu_buffer->entries) - cpu_buffer->overrun) + ret = (local_read(&cpu_buffer->entries) - local_read(&cpu_buffer->overrun)) - cpu_buffer->read; return ret; @@ -1989,33 +2587,13 @@ unsigned long ring_buffer_overrun_cpu(struct ring_buffer *buffer, int cpu) return 0; cpu_buffer = buffer->buffers[cpu]; - ret = cpu_buffer->overrun; + ret = local_read(&cpu_buffer->overrun); return ret; } EXPORT_SYMBOL_GPL(ring_buffer_overrun_cpu); /** - * ring_buffer_nmi_dropped_cpu - get the number of nmis that were dropped - * @buffer: The ring buffer - * @cpu: The per CPU buffer to get the number of overruns from - */ -unsigned long ring_buffer_nmi_dropped_cpu(struct ring_buffer *buffer, int cpu) -{ - struct ring_buffer_per_cpu *cpu_buffer; - unsigned long ret; - - if (!cpumask_test_cpu(cpu, buffer->cpumask)) - return 0; - - cpu_buffer = buffer->buffers[cpu]; - ret = cpu_buffer->nmi_dropped; - - return ret; -} -EXPORT_SYMBOL_GPL(ring_buffer_nmi_dropped_cpu); - -/** * ring_buffer_commit_overrun_cpu - get the number of overruns caused by commits * @buffer: The ring buffer * @cpu: The per CPU buffer to get the number of overruns from @@ -2030,7 +2608,7 @@ ring_buffer_commit_overrun_cpu(struct ring_buffer *buffer, int cpu) return 0; cpu_buffer = buffer->buffers[cpu]; - ret = cpu_buffer->commit_overrun; + ret = local_read(&cpu_buffer->commit_overrun); return ret; } @@ -2053,7 +2631,7 @@ unsigned long ring_buffer_entries(struct ring_buffer *buffer) for_each_buffer_cpu(buffer, cpu) { cpu_buffer = buffer->buffers[cpu]; entries += (local_read(&cpu_buffer->entries) - - cpu_buffer->overrun) - cpu_buffer->read; + local_read(&cpu_buffer->overrun)) - cpu_buffer->read; } return entries; @@ -2076,7 +2654,7 @@ unsigned long ring_buffer_overruns(struct ring_buffer *buffer) /* if you care about this being correct, lock the buffer */ for_each_buffer_cpu(buffer, cpu) { cpu_buffer = buffer->buffers[cpu]; - overruns += cpu_buffer->overrun; + overruns += local_read(&cpu_buffer->overrun); } return overruns; @@ -2089,8 +2667,10 @@ static void rb_iter_reset(struct ring_buffer_iter *iter) /* Iterator usage is expected to have record disabled */ if (list_empty(&cpu_buffer->reader_page->list)) { - iter->head_page = cpu_buffer->head_page; - iter->head = cpu_buffer->head_page->read; + iter->head_page = rb_set_head_page(cpu_buffer); + if (unlikely(!iter->head_page)) + return; + iter->head = iter->head_page->read; } else { iter->head_page = cpu_buffer->reader_page; iter->head = cpu_buffer->reader_page->read; @@ -2207,6 +2787,7 @@ rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer) struct buffer_page *reader = NULL; unsigned long flags; int nr_loops = 0; + int ret; local_irq_save(flags); __raw_spin_lock(&cpu_buffer->lock); @@ -2240,30 +2821,56 @@ rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer) goto out; /* - * Splice the empty reader page into the list around the head. * Reset the reader page to size zero. */ + local_set(&cpu_buffer->reader_page->write, 0); + local_set(&cpu_buffer->reader_page->entries, 0); + local_set(&cpu_buffer->reader_page->page->commit, 0); - reader = cpu_buffer->head_page; + spin: + /* + * Splice the empty reader page into the list around the head. + */ + reader = rb_set_head_page(cpu_buffer); cpu_buffer->reader_page->list.next = reader->list.next; cpu_buffer->reader_page->list.prev = reader->list.prev; - local_set(&cpu_buffer->reader_page->write, 0); - local_set(&cpu_buffer->reader_page->entries, 0); - local_set(&cpu_buffer->reader_page->page->commit, 0); + /* + * cpu_buffer->pages just needs to point to the buffer, it + * has no specific buffer page to point to. Lets move it out + * of our way so we don't accidently swap it. + */ + cpu_buffer->pages = reader->list.prev; - /* Make the reader page now replace the head */ - reader->list.prev->next = &cpu_buffer->reader_page->list; - reader->list.next->prev = &cpu_buffer->reader_page->list; + /* The reader page will be pointing to the new head */ + rb_set_list_to_head(cpu_buffer, &cpu_buffer->reader_page->list); + + /* + * Here's the tricky part. + * + * We need to move the pointer past the header page. + * But we can only do that if a writer is not currently + * moving it. The page before the header page has the + * flag bit '1' set if it is pointing to the page we want. + * but if the writer is in the process of moving it + * than it will be '2' or already moved '0'. + */ + + ret = rb_head_page_replace(reader, cpu_buffer->reader_page); /* - * If the tail is on the reader, then we must set the head - * to the inserted page, otherwise we set it one before. + * If we did not convert it, then we must try again. */ - cpu_buffer->head_page = cpu_buffer->reader_page; + if (!ret) + goto spin; - if (cpu_buffer->commit_page != reader) - rb_inc_page(cpu_buffer, &cpu_buffer->head_page); + /* + * Yeah! We succeeded in replacing the page. + * + * Now make the new head point back to the reader page. + */ + reader->list.next->prev = &cpu_buffer->reader_page->list; + rb_inc_page(cpu_buffer, &cpu_buffer->head_page); /* Finally update the reader page to the new head */ cpu_buffer->reader_page = reader; @@ -2717,8 +3324,10 @@ EXPORT_SYMBOL_GPL(ring_buffer_size); static void rb_reset_cpu(struct ring_buffer_per_cpu *cpu_buffer) { + rb_head_page_deactivate(cpu_buffer); + cpu_buffer->head_page - = list_entry(cpu_buffer->pages.next, struct buffer_page, list); + = list_entry(cpu_buffer->pages, struct buffer_page, list); local_set(&cpu_buffer->head_page->write, 0); local_set(&cpu_buffer->head_page->entries, 0); local_set(&cpu_buffer->head_page->page->commit, 0); @@ -2734,16 +3343,17 @@ rb_reset_cpu(struct ring_buffer_per_cpu *cpu_buffer) local_set(&cpu_buffer->reader_page->page->commit, 0); cpu_buffer->reader_page->read = 0; - cpu_buffer->nmi_dropped = 0; - cpu_buffer->commit_overrun = 0; - cpu_buffer->overrun = 0; - cpu_buffer->read = 0; + local_set(&cpu_buffer->commit_overrun, 0); + local_set(&cpu_buffer->overrun, 0); local_set(&cpu_buffer->entries, 0); local_set(&cpu_buffer->committing, 0); local_set(&cpu_buffer->commits, 0); + cpu_buffer->read = 0; cpu_buffer->write_stamp = 0; cpu_buffer->read_stamp = 0; + + rb_head_page_activate(cpu_buffer); } /** @@ -3091,7 +3701,7 @@ int ring_buffer_read_page(struct ring_buffer *buffer, read = 0; } else { /* update the entry counter */ - cpu_buffer->read += local_read(&reader->entries); + cpu_buffer->read += rb_page_entries(reader); /* swap the pages */ rb_init_page(bpage); diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index c22b40f8f57..e793cda91dd 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -50,7 +50,7 @@ unsigned long __read_mostly tracing_thresh; * On boot up, the ring buffer is set to the minimum size, so that * we do not waste memory on systems that are not using tracing. */ -static int ring_buffer_expanded; +int ring_buffer_expanded; /* * We need to change this state when a selftest is running. @@ -64,7 +64,7 @@ static bool __read_mostly tracing_selftest_running; /* * If a tracer is running, we do not want to run SELFTEST. */ -static bool __read_mostly tracing_selftest_disabled; +bool __read_mostly tracing_selftest_disabled; /* For tracers that don't implement custom flags */ static struct tracer_opt dummy_tracer_opt[] = { @@ -89,7 +89,7 @@ static int dummy_set_flag(u32 old_flags, u32 bit, int set) */ static int tracing_disabled = 1; -static DEFINE_PER_CPU(local_t, ftrace_cpu_disabled); +DEFINE_PER_CPU(local_t, ftrace_cpu_disabled); static inline void ftrace_disable_cpu(void) { @@ -867,10 +867,6 @@ struct ring_buffer_event *trace_buffer_lock_reserve(struct trace_array *tr, return event; } -static void ftrace_trace_stack(struct trace_array *tr, - unsigned long flags, int skip, int pc); -static void ftrace_trace_userstack(struct trace_array *tr, - unsigned long flags, int pc); static inline void __trace_buffer_unlock_commit(struct trace_array *tr, struct ring_buffer_event *event, @@ -947,54 +943,6 @@ trace_function(struct trace_array *tr, ring_buffer_unlock_commit(tr->buffer, event); } -#ifdef CONFIG_FUNCTION_GRAPH_TRACER -static int __trace_graph_entry(struct trace_array *tr, - struct ftrace_graph_ent *trace, - unsigned long flags, - int pc) -{ - struct ftrace_event_call *call = &event_funcgraph_entry; - struct ring_buffer_event *event; - struct ftrace_graph_ent_entry *entry; - - if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled)))) - return 0; - - event = trace_buffer_lock_reserve(&global_trace, TRACE_GRAPH_ENT, - sizeof(*entry), flags, pc); - if (!event) - return 0; - entry = ring_buffer_event_data(event); - entry->graph_ent = *trace; - if (!filter_current_check_discard(call, entry, event)) - ring_buffer_unlock_commit(global_trace.buffer, event); - - return 1; -} - -static void __trace_graph_return(struct trace_array *tr, - struct ftrace_graph_ret *trace, - unsigned long flags, - int pc) -{ - struct ftrace_event_call *call = &event_funcgraph_exit; - struct ring_buffer_event *event; - struct ftrace_graph_ret_entry *entry; - - if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled)))) - return; - - event = trace_buffer_lock_reserve(&global_trace, TRACE_GRAPH_RET, - sizeof(*entry), flags, pc); - if (!event) - return; - entry = ring_buffer_event_data(event); - entry->ret = *trace; - if (!filter_current_check_discard(call, entry, event)) - ring_buffer_unlock_commit(global_trace.buffer, event); -} -#endif - void ftrace(struct trace_array *tr, struct trace_array_cpu *data, unsigned long ip, unsigned long parent_ip, unsigned long flags, @@ -1004,11 +952,11 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, ip, parent_ip, flags, pc); } +#ifdef CONFIG_STACKTRACE static void __ftrace_trace_stack(struct trace_array *tr, unsigned long flags, int skip, int pc) { -#ifdef CONFIG_STACKTRACE struct ftrace_event_call *call = &event_kernel_stack; struct ring_buffer_event *event; struct stack_entry *entry; @@ -1029,12 +977,10 @@ static void __ftrace_trace_stack(struct trace_array *tr, save_stack_trace(&trace); if (!filter_check_discard(call, entry, tr->buffer, event)) ring_buffer_unlock_commit(tr->buffer, event); -#endif } -static void ftrace_trace_stack(struct trace_array *tr, - unsigned long flags, - int skip, int pc) +void ftrace_trace_stack(struct trace_array *tr, unsigned long flags, int skip, + int pc) { if (!(trace_flags & TRACE_ITER_STACKTRACE)) return; @@ -1042,17 +988,14 @@ static void ftrace_trace_stack(struct trace_array *tr, __ftrace_trace_stack(tr, flags, skip, pc); } -void __trace_stack(struct trace_array *tr, - unsigned long flags, - int skip, int pc) +void __trace_stack(struct trace_array *tr, unsigned long flags, int skip, + int pc) { __ftrace_trace_stack(tr, flags, skip, pc); } -static void ftrace_trace_userstack(struct trace_array *tr, - unsigned long flags, int pc) +void ftrace_trace_userstack(struct trace_array *tr, unsigned long flags, int pc) { -#ifdef CONFIG_STACKTRACE struct ftrace_event_call *call = &event_user_stack; struct ring_buffer_event *event; struct userstack_entry *entry; @@ -1077,7 +1020,6 @@ static void ftrace_trace_userstack(struct trace_array *tr, save_stack_trace_user(&trace); if (!filter_check_discard(call, entry, tr->buffer, event)) ring_buffer_unlock_commit(tr->buffer, event); -#endif } #ifdef UNUSED @@ -1087,6 +1029,8 @@ static void __trace_userstack(struct trace_array *tr, unsigned long flags) } #endif /* UNUSED */ +#endif /* CONFIG_STACKTRACE */ + static void ftrace_trace_special(void *__tr, unsigned long arg1, unsigned long arg2, unsigned long arg3, @@ -1115,62 +1059,6 @@ __trace_special(void *__tr, void *__data, } void -tracing_sched_switch_trace(struct trace_array *tr, - struct task_struct *prev, - struct task_struct *next, - unsigned long flags, int pc) -{ - struct ftrace_event_call *call = &event_context_switch; - struct ring_buffer_event *event; - struct ctx_switch_entry *entry; - - event = trace_buffer_lock_reserve(tr, TRACE_CTX, - sizeof(*entry), flags, pc); - if (!event) - return; - entry = ring_buffer_event_data(event); - entry->prev_pid = prev->pid; - entry->prev_prio = prev->prio; - entry->prev_state = prev->state; - entry->next_pid = next->pid; - entry->next_prio = next->prio; - entry->next_state = next->state; - entry->next_cpu = task_cpu(next); - - if (!filter_check_discard(call, entry, tr->buffer, event)) - trace_buffer_unlock_commit(tr, event, flags, pc); -} - -void -tracing_sched_wakeup_trace(struct trace_array *tr, - struct task_struct *wakee, - struct task_struct *curr, - unsigned long flags, int pc) -{ - struct ftrace_event_call *call = &event_wakeup; - struct ring_buffer_event *event; - struct ctx_switch_entry *entry; - - event = trace_buffer_lock_reserve(tr, TRACE_WAKE, - sizeof(*entry), flags, pc); - if (!event) - return; - entry = ring_buffer_event_data(event); - entry->prev_pid = curr->pid; - entry->prev_prio = curr->prio; - entry->prev_state = curr->state; - entry->next_pid = wakee->pid; - entry->next_prio = wakee->prio; - entry->next_state = wakee->state; - entry->next_cpu = task_cpu(wakee); - - if (!filter_check_discard(call, entry, tr->buffer, event)) - ring_buffer_unlock_commit(tr->buffer, event); - ftrace_trace_stack(tr, flags, 6, pc); - ftrace_trace_userstack(tr, flags, pc); -} - -void ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) { struct trace_array *tr = &global_trace; @@ -1194,68 +1082,6 @@ ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) local_irq_restore(flags); } -#ifdef CONFIG_FUNCTION_GRAPH_TRACER -int trace_graph_entry(struct ftrace_graph_ent *trace) -{ - struct trace_array *tr = &global_trace; - struct trace_array_cpu *data; - unsigned long flags; - long disabled; - int ret; - int cpu; - int pc; - - if (!ftrace_trace_task(current)) - return 0; - - if (!ftrace_graph_addr(trace->func)) - return 0; - - local_irq_save(flags); - cpu = raw_smp_processor_id(); - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); - if (likely(disabled == 1)) { - pc = preempt_count(); - ret = __trace_graph_entry(tr, trace, flags, pc); - } else { - ret = 0; - } - /* Only do the atomic if it is not already set */ - if (!test_tsk_trace_graph(current)) - set_tsk_trace_graph(current); - - atomic_dec(&data->disabled); - local_irq_restore(flags); - - return ret; -} - -void trace_graph_return(struct ftrace_graph_ret *trace) -{ - struct trace_array *tr = &global_trace; - struct trace_array_cpu *data; - unsigned long flags; - long disabled; - int cpu; - int pc; - - local_irq_save(flags); - cpu = raw_smp_processor_id(); - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); - if (likely(disabled == 1)) { - pc = preempt_count(); - __trace_graph_return(tr, trace, flags, pc); - } - if (!trace->depth) - clear_tsk_trace_graph(current); - atomic_dec(&data->disabled); - local_irq_restore(flags); -} -#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ - - /** * trace_vbprintk - write binary msg to tracing buffer * @@ -2257,8 +2083,8 @@ tracing_trace_options_read(struct file *filp, char __user *ubuf, len += 3; /* "no" and newline */ } - /* +2 for \n and \0 */ - buf = kmalloc(len + 2, GFP_KERNEL); + /* +1 for \0 */ + buf = kmalloc(len + 1, GFP_KERNEL); if (!buf) { mutex_unlock(&trace_types_lock); return -ENOMEM; @@ -2281,7 +2107,7 @@ tracing_trace_options_read(struct file *filp, char __user *ubuf, } mutex_unlock(&trace_types_lock); - WARN_ON(r >= len + 2); + WARN_ON(r >= len + 1); r = simple_read_from_buffer(ubuf, cnt, ppos, buf, r); @@ -3633,9 +3459,6 @@ tracing_stats_read(struct file *filp, char __user *ubuf, cnt = ring_buffer_commit_overrun_cpu(tr->buffer, cpu); trace_seq_printf(s, "commit overrun: %ld\n", cnt); - cnt = ring_buffer_nmi_dropped_cpu(tr->buffer, cpu); - trace_seq_printf(s, "nmi dropped: %ld\n", cnt); - count = simple_read_from_buffer(ubuf, count, ppos, s->buffer, s->len); kfree(s); @@ -4273,7 +4096,6 @@ void ftrace_dump(void) __init static int tracer_alloc_buffers(void) { - struct trace_array_cpu *data; int ring_buf_size; int i; int ret = -ENOMEM; @@ -4323,7 +4145,7 @@ __init static int tracer_alloc_buffers(void) /* Allocate the first page for all buffers */ for_each_tracing_cpu(i) { - data = global_trace.data[i] = &per_cpu(global_trace_cpu, i); + global_trace.data[i] = &per_cpu(global_trace_cpu, i); max_tr.data[i] = &per_cpu(max_data, i); } diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 8b9f4f6e955..d682357e4b1 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -467,6 +467,7 @@ void trace_function(struct trace_array *tr, void trace_graph_return(struct ftrace_graph_ret *trace); int trace_graph_entry(struct ftrace_graph_ent *trace); +void set_graph_array(struct trace_array *tr); void tracing_start_cmdline_record(void); void tracing_stop_cmdline_record(void); @@ -485,9 +486,31 @@ void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu); void update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu); -void __trace_stack(struct trace_array *tr, - unsigned long flags, - int skip, int pc); +#ifdef CONFIG_STACKTRACE +void ftrace_trace_stack(struct trace_array *tr, unsigned long flags, + int skip, int pc); + +void ftrace_trace_userstack(struct trace_array *tr, unsigned long flags, + int pc); + +void __trace_stack(struct trace_array *tr, unsigned long flags, int skip, + int pc); +#else +static inline void ftrace_trace_stack(struct trace_array *tr, + unsigned long flags, int skip, int pc) +{ +} + +static inline void ftrace_trace_userstack(struct trace_array *tr, + unsigned long flags, int pc) +{ +} + +static inline void __trace_stack(struct trace_array *tr, unsigned long flags, + int skip, int pc) +{ +} +#endif /* CONFIG_STACKTRACE */ extern cycle_t ftrace_now(int cpu); @@ -513,6 +536,10 @@ extern unsigned long ftrace_update_tot_cnt; extern int DYN_FTRACE_TEST_NAME(void); #endif +extern int ring_buffer_expanded; +extern bool tracing_selftest_disabled; +DECLARE_PER_CPU(local_t, ftrace_cpu_disabled); + #ifdef CONFIG_FTRACE_STARTUP_TEST extern int trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr); @@ -743,13 +770,15 @@ struct event_filter { int n_preds; struct filter_pred **preds; char *filter_string; + bool no_reset; }; struct event_subsystem { struct list_head list; const char *name; struct dentry *entry; - void *filter; + struct event_filter *filter; + int nr_events; }; struct filter_pred; diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index e75276a49cf..e0cbede9678 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -17,6 +17,8 @@ #include <linux/ctype.h> #include <linux/delay.h> +#include <asm/setup.h> + #include "trace_output.h" #define TRACE_SYSTEM "TRACE_SYSTEM" @@ -849,8 +851,10 @@ event_subsystem_dir(const char *name, struct dentry *d_events) /* First see if we did not already create this dir */ list_for_each_entry(system, &event_subsystems, list) { - if (strcmp(system->name, name) == 0) + if (strcmp(system->name, name) == 0) { + system->nr_events++; return system->entry; + } } /* need to create new entry */ @@ -869,6 +873,7 @@ event_subsystem_dir(const char *name, struct dentry *d_events) return d_events; } + system->nr_events = 1; system->name = kstrdup(name, GFP_KERNEL); if (!system->name) { debugfs_remove(system->entry); @@ -987,6 +992,32 @@ struct ftrace_module_file_ops { struct file_operations filter; }; +static void remove_subsystem_dir(const char *name) +{ + struct event_subsystem *system; + + if (strcmp(name, TRACE_SYSTEM) == 0) + return; + + list_for_each_entry(system, &event_subsystems, list) { + if (strcmp(system->name, name) == 0) { + if (!--system->nr_events) { + struct event_filter *filter = system->filter; + + debugfs_remove_recursive(system->entry); + list_del(&system->list); + if (filter) { + kfree(filter->filter_string); + kfree(filter); + } + kfree(system->name); + kfree(system); + } + break; + } + } +} + static struct ftrace_module_file_ops * trace_create_file_ops(struct module *mod) { @@ -1077,6 +1108,7 @@ static void trace_module_remove_events(struct module *mod) list_del(&call->list); trace_destroy_fields(call); destroy_preds(call); + remove_subsystem_dir(call->system); } } @@ -1133,6 +1165,18 @@ struct notifier_block trace_module_nb = { extern struct ftrace_event_call __start_ftrace_events[]; extern struct ftrace_event_call __stop_ftrace_events[]; +static char bootup_event_buf[COMMAND_LINE_SIZE] __initdata; + +static __init int setup_trace_event(char *str) +{ + strlcpy(bootup_event_buf, str, COMMAND_LINE_SIZE); + ring_buffer_expanded = 1; + tracing_selftest_disabled = 1; + + return 1; +} +__setup("trace_event=", setup_trace_event); + static __init int event_trace_init(void) { struct ftrace_event_call *call; @@ -1140,6 +1184,8 @@ static __init int event_trace_init(void) struct dentry *entry; struct dentry *d_events; int ret; + char *buf = bootup_event_buf; + char *token; d_tracer = tracing_init_dentry(); if (!d_tracer) @@ -1185,6 +1231,19 @@ static __init int event_trace_init(void) &ftrace_event_format_fops); } + while (true) { + token = strsep(&buf, ","); + + if (!token) + break; + if (!*token) + continue; + + ret = ftrace_set_clr_event(token, 1); + if (ret) + pr_warning("Failed to enable trace event: %s\n", token); + } + ret = register_module_notifier(&trace_module_nb); if (ret) pr_warning("Failed to register trace events module notifier\n"); @@ -1392,10 +1451,10 @@ static __init void event_trace_self_test_with_function(void) static __init int event_trace_self_tests_init(void) { - - event_trace_self_tests(); - - event_trace_self_test_with_function(); + if (!tracing_selftest_disabled) { + event_trace_self_tests(); + event_trace_self_test_with_function(); + } return 0; } diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index f32dc9d1ea7..490337abed7 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -176,11 +176,13 @@ static int filter_pred_string(struct filter_pred *pred, void *event, static int filter_pred_strloc(struct filter_pred *pred, void *event, int val1, int val2) { - unsigned short str_loc = *(unsigned short *)(event + pred->offset); + u32 str_item = *(u32 *)(event + pred->offset); + int str_loc = str_item & 0xffff; + int str_len = str_item >> 16; char *addr = (char *)(event + str_loc); int cmp, match; - cmp = strncmp(addr, pred->str_val, pred->str_len); + cmp = strncmp(addr, pred->str_val, str_len); match = (!cmp) ^ pred->not; @@ -418,24 +420,29 @@ oom: } EXPORT_SYMBOL_GPL(init_preds); -static void filter_free_subsystem_preds(struct event_subsystem *system) +enum { + FILTER_DISABLE_ALL, + FILTER_INIT_NO_RESET, + FILTER_SKIP_NO_RESET, +}; + +static void filter_free_subsystem_preds(struct event_subsystem *system, + int flag) { - struct event_filter *filter = system->filter; struct ftrace_event_call *call; - int i; - - if (filter->n_preds) { - for (i = 0; i < filter->n_preds; i++) - filter_free_pred(filter->preds[i]); - kfree(filter->preds); - filter->preds = NULL; - filter->n_preds = 0; - } list_for_each_entry(call, &ftrace_events, list) { if (!call->define_fields) continue; + if (flag == FILTER_INIT_NO_RESET) { + call->filter->no_reset = false; + continue; + } + + if (flag == FILTER_SKIP_NO_RESET && call->filter->no_reset) + continue; + if (!strcmp(call->system, system->name)) { filter_disable_preds(call); remove_filter_string(call->filter); @@ -537,7 +544,8 @@ static filter_pred_fn_t select_comparison_fn(int op, int field_size, static int filter_add_pred(struct filter_parse_state *ps, struct ftrace_event_call *call, - struct filter_pred *pred) + struct filter_pred *pred, + bool dry_run) { struct ftrace_event_field *field; filter_pred_fn_t fn; @@ -549,10 +557,12 @@ static int filter_add_pred(struct filter_parse_state *ps, if (pred->op == OP_AND) { pred->pop_n = 2; - return filter_add_pred_fn(ps, call, pred, filter_pred_and); + fn = filter_pred_and; + goto add_pred_fn; } else if (pred->op == OP_OR) { pred->pop_n = 2; - return filter_add_pred_fn(ps, call, pred, filter_pred_or); + fn = filter_pred_or; + goto add_pred_fn; } field = find_event_field(call, pred->field_name); @@ -575,9 +585,6 @@ static int filter_add_pred(struct filter_parse_state *ps, else fn = filter_pred_strloc; pred->str_len = field->size; - if (pred->op == OP_NE) - pred->not = 1; - return filter_add_pred_fn(ps, call, pred, fn); } else { if (field->is_signed) ret = strict_strtoll(pred->str_val, 0, &val); @@ -588,41 +595,33 @@ static int filter_add_pred(struct filter_parse_state *ps, return -EINVAL; } pred->val = val; - } - fn = select_comparison_fn(pred->op, field->size, field->is_signed); - if (!fn) { - parse_error(ps, FILT_ERR_INVALID_OP, 0); - return -EINVAL; + fn = select_comparison_fn(pred->op, field->size, + field->is_signed); + if (!fn) { + parse_error(ps, FILT_ERR_INVALID_OP, 0); + return -EINVAL; + } } if (pred->op == OP_NE) pred->not = 1; - return filter_add_pred_fn(ps, call, pred, fn); +add_pred_fn: + if (!dry_run) + return filter_add_pred_fn(ps, call, pred, fn); + return 0; } static int filter_add_subsystem_pred(struct filter_parse_state *ps, struct event_subsystem *system, struct filter_pred *pred, - char *filter_string) + char *filter_string, + bool dry_run) { - struct event_filter *filter = system->filter; struct ftrace_event_call *call; int err = 0; - - if (!filter->preds) { - filter->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), - GFP_KERNEL); - - if (!filter->preds) - return -ENOMEM; - } - - if (filter->n_preds == MAX_FILTER_PRED) { - parse_error(ps, FILT_ERR_TOO_MANY_PREDS, 0); - return -ENOSPC; - } + bool fail = true; list_for_each_entry(call, &ftrace_events, list) { @@ -632,19 +631,24 @@ static int filter_add_subsystem_pred(struct filter_parse_state *ps, if (strcmp(call->system, system->name)) continue; - err = filter_add_pred(ps, call, pred); - if (err) { - filter_free_subsystem_preds(system); - parse_error(ps, FILT_ERR_BAD_SUBSYS_FILTER, 0); - goto out; - } - replace_filter_string(call->filter, filter_string); + if (call->filter->no_reset) + continue; + + err = filter_add_pred(ps, call, pred, dry_run); + if (err) + call->filter->no_reset = true; + else + fail = false; + + if (!dry_run) + replace_filter_string(call->filter, filter_string); } - filter->preds[filter->n_preds] = pred; - filter->n_preds++; -out: - return err; + if (fail) { + parse_error(ps, FILT_ERR_BAD_SUBSYS_FILTER, 0); + return err; + } + return 0; } static void parse_init(struct filter_parse_state *ps, @@ -1003,12 +1007,14 @@ static int check_preds(struct filter_parse_state *ps) static int replace_preds(struct event_subsystem *system, struct ftrace_event_call *call, struct filter_parse_state *ps, - char *filter_string) + char *filter_string, + bool dry_run) { char *operand1 = NULL, *operand2 = NULL; struct filter_pred *pred; struct postfix_elt *elt; int err; + int n_preds = 0; err = check_preds(ps); if (err) @@ -1027,24 +1033,14 @@ static int replace_preds(struct event_subsystem *system, continue; } + if (n_preds++ == MAX_FILTER_PRED) { + parse_error(ps, FILT_ERR_TOO_MANY_PREDS, 0); + return -ENOSPC; + } + if (elt->op == OP_AND || elt->op == OP_OR) { pred = create_logical_pred(elt->op); - if (!pred) - return -ENOMEM; - if (call) { - err = filter_add_pred(ps, call, pred); - filter_free_pred(pred); - } else { - err = filter_add_subsystem_pred(ps, system, - pred, filter_string); - if (err) - filter_free_pred(pred); - } - if (err) - return err; - - operand1 = operand2 = NULL; - continue; + goto add_pred; } if (!operand1 || !operand2) { @@ -1053,17 +1049,15 @@ static int replace_preds(struct event_subsystem *system, } pred = create_pred(elt->op, operand1, operand2); +add_pred: if (!pred) return -ENOMEM; - if (call) { - err = filter_add_pred(ps, call, pred); - filter_free_pred(pred); - } else { + if (call) + err = filter_add_pred(ps, call, pred, false); + else err = filter_add_subsystem_pred(ps, system, pred, - filter_string); - if (err) - filter_free_pred(pred); - } + filter_string, dry_run); + filter_free_pred(pred); if (err) return err; @@ -1103,7 +1097,7 @@ int apply_event_filter(struct ftrace_event_call *call, char *filter_string) goto out; } - err = replace_preds(NULL, call, ps, filter_string); + err = replace_preds(NULL, call, ps, filter_string, false); if (err) append_filter_err(ps, call->filter); @@ -1127,7 +1121,7 @@ int apply_subsystem_event_filter(struct event_subsystem *system, mutex_lock(&event_mutex); if (!strcmp(strstrip(filter_string), "0")) { - filter_free_subsystem_preds(system); + filter_free_subsystem_preds(system, FILTER_DISABLE_ALL); remove_filter_string(system->filter); mutex_unlock(&event_mutex); return 0; @@ -1138,7 +1132,6 @@ int apply_subsystem_event_filter(struct event_subsystem *system, if (!ps) goto out_unlock; - filter_free_subsystem_preds(system); replace_filter_string(system->filter, filter_string); parse_init(ps, filter_ops, filter_string); @@ -1148,9 +1141,23 @@ int apply_subsystem_event_filter(struct event_subsystem *system, goto out; } - err = replace_preds(system, NULL, ps, filter_string); - if (err) + filter_free_subsystem_preds(system, FILTER_INIT_NO_RESET); + + /* try to see the filter can be applied to which events */ + err = replace_preds(system, NULL, ps, filter_string, true); + if (err) { append_filter_err(ps, system->filter); + goto out; + } + + filter_free_subsystem_preds(system, FILTER_SKIP_NO_RESET); + + /* really apply the filter to the events */ + err = replace_preds(system, NULL, ps, filter_string, false); + if (err) { + append_filter_err(ps, system->filter); + filter_free_subsystem_preds(system, 2); + } out: filter_opstack_clear(ps); diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index 75ef000613c..5b01b94518f 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -288,11 +288,9 @@ static int ftrace_trace_onoff_print(struct seq_file *m, unsigned long ip, struct ftrace_probe_ops *ops, void *data) { - char str[KSYM_SYMBOL_LEN]; long count = (long)data; - kallsyms_lookup(ip, NULL, NULL, NULL, str); - seq_printf(m, "%s:", str); + seq_printf(m, "%pf:", (void *)ip); if (ops == &traceon_probe_ops) seq_printf(m, "traceon"); diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 420ec348757..3f4a251b7d1 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -52,7 +52,7 @@ static struct tracer_flags tracer_flags = { .opts = trace_opts }; -/* pid on the last trace processed */ +static struct trace_array *graph_array; /* Add a function return address to the trace stack on thread info.*/ @@ -166,10 +166,121 @@ unsigned long ftrace_return_to_handler(unsigned long frame_pointer) return ret; } +static int __trace_graph_entry(struct trace_array *tr, + struct ftrace_graph_ent *trace, + unsigned long flags, + int pc) +{ + struct ftrace_event_call *call = &event_funcgraph_entry; + struct ring_buffer_event *event; + struct ftrace_graph_ent_entry *entry; + + if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled)))) + return 0; + + event = trace_buffer_lock_reserve(tr, TRACE_GRAPH_ENT, + sizeof(*entry), flags, pc); + if (!event) + return 0; + entry = ring_buffer_event_data(event); + entry->graph_ent = *trace; + if (!filter_current_check_discard(call, entry, event)) + ring_buffer_unlock_commit(tr->buffer, event); + + return 1; +} + +int trace_graph_entry(struct ftrace_graph_ent *trace) +{ + struct trace_array *tr = graph_array; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int ret; + int cpu; + int pc; + + if (unlikely(!tr)) + return 0; + + if (!ftrace_trace_task(current)) + return 0; + + if (!ftrace_graph_addr(trace->func)) + return 0; + + local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + if (likely(disabled == 1)) { + pc = preempt_count(); + ret = __trace_graph_entry(tr, trace, flags, pc); + } else { + ret = 0; + } + /* Only do the atomic if it is not already set */ + if (!test_tsk_trace_graph(current)) + set_tsk_trace_graph(current); + + atomic_dec(&data->disabled); + local_irq_restore(flags); + + return ret; +} + +static void __trace_graph_return(struct trace_array *tr, + struct ftrace_graph_ret *trace, + unsigned long flags, + int pc) +{ + struct ftrace_event_call *call = &event_funcgraph_exit; + struct ring_buffer_event *event; + struct ftrace_graph_ret_entry *entry; + + if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled)))) + return; + + event = trace_buffer_lock_reserve(tr, TRACE_GRAPH_RET, + sizeof(*entry), flags, pc); + if (!event) + return; + entry = ring_buffer_event_data(event); + entry->ret = *trace; + if (!filter_current_check_discard(call, entry, event)) + ring_buffer_unlock_commit(tr->buffer, event); +} + +void trace_graph_return(struct ftrace_graph_ret *trace) +{ + struct trace_array *tr = graph_array; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + int pc; + + local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + if (likely(disabled == 1)) { + pc = preempt_count(); + __trace_graph_return(tr, trace, flags, pc); + } + if (!trace->depth) + clear_tsk_trace_graph(current); + atomic_dec(&data->disabled); + local_irq_restore(flags); +} + static int graph_trace_init(struct trace_array *tr) { - int ret = register_ftrace_graph(&trace_graph_return, - &trace_graph_entry); + int ret; + + graph_array = tr; + ret = register_ftrace_graph(&trace_graph_return, + &trace_graph_entry); if (ret) return ret; tracing_start_cmdline_record(); @@ -177,49 +288,30 @@ static int graph_trace_init(struct trace_array *tr) return 0; } +void set_graph_array(struct trace_array *tr) +{ + graph_array = tr; +} + static void graph_trace_reset(struct trace_array *tr) { tracing_stop_cmdline_record(); unregister_ftrace_graph(); } -static inline int log10_cpu(int nb) -{ - if (nb / 100) - return 3; - if (nb / 10) - return 2; - return 1; -} +static int max_bytes_for_cpu; static enum print_line_t print_graph_cpu(struct trace_seq *s, int cpu) { - int i; int ret; - int log10_this = log10_cpu(cpu); - int log10_all = log10_cpu(cpumask_weight(cpu_online_mask)); - /* * Start with a space character - to make it stand out * to the right a bit when trace output is pasted into * email: */ - ret = trace_seq_printf(s, " "); - - /* - * 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++) { - ret = trace_seq_printf(s, " "); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - } - ret = trace_seq_printf(s, "%d) ", cpu); + ret = trace_seq_printf(s, " %*d) ", max_bytes_for_cpu, cpu); if (!ret) return TRACE_TYPE_PARTIAL_LINE; @@ -565,11 +657,7 @@ print_graph_entry_leaf(struct trace_iterator *iter, return TRACE_TYPE_PARTIAL_LINE; } - ret = seq_print_ip_sym(s, call->func, 0); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - - ret = trace_seq_printf(s, "();\n"); + ret = trace_seq_printf(s, "%pf();\n", (void *)call->func); if (!ret) return TRACE_TYPE_PARTIAL_LINE; @@ -612,11 +700,7 @@ print_graph_entry_nested(struct trace_iterator *iter, return TRACE_TYPE_PARTIAL_LINE; } - ret = seq_print_ip_sym(s, call->func, 0); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - - ret = trace_seq_printf(s, "() {\n"); + ret = trace_seq_printf(s, "%pf() {\n", (void *)call->func); if (!ret) return TRACE_TYPE_PARTIAL_LINE; @@ -934,6 +1018,8 @@ static struct tracer graph_trace __read_mostly = { static __init int init_graph_trace(void) { + max_bytes_for_cpu = snprintf(NULL, 0, "%d", nr_cpu_ids - 1); + return register_tracer(&graph_trace); } diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index a98106dd979..e1285d7b548 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -20,6 +20,34 @@ static int sched_ref; static DEFINE_MUTEX(sched_register_mutex); static int sched_stopped; + +void +tracing_sched_switch_trace(struct trace_array *tr, + struct task_struct *prev, + struct task_struct *next, + unsigned long flags, int pc) +{ + struct ftrace_event_call *call = &event_context_switch; + struct ring_buffer_event *event; + struct ctx_switch_entry *entry; + + event = trace_buffer_lock_reserve(tr, TRACE_CTX, + sizeof(*entry), flags, pc); + if (!event) + return; + entry = ring_buffer_event_data(event); + entry->prev_pid = prev->pid; + entry->prev_prio = prev->prio; + entry->prev_state = prev->state; + entry->next_pid = next->pid; + entry->next_prio = next->prio; + entry->next_state = next->state; + entry->next_cpu = task_cpu(next); + + if (!filter_check_discard(call, entry, tr->buffer, event)) + trace_buffer_unlock_commit(tr, event, flags, pc); +} + static void probe_sched_switch(struct rq *__rq, struct task_struct *prev, struct task_struct *next) @@ -49,6 +77,35 @@ probe_sched_switch(struct rq *__rq, struct task_struct *prev, local_irq_restore(flags); } +void +tracing_sched_wakeup_trace(struct trace_array *tr, + struct task_struct *wakee, + struct task_struct *curr, + unsigned long flags, int pc) +{ + struct ftrace_event_call *call = &event_wakeup; + struct ring_buffer_event *event; + struct ctx_switch_entry *entry; + + event = trace_buffer_lock_reserve(tr, TRACE_WAKE, + sizeof(*entry), flags, pc); + if (!event) + return; + entry = ring_buffer_event_data(event); + entry->prev_pid = curr->pid; + entry->prev_prio = curr->prio; + entry->prev_state = curr->state; + entry->next_pid = wakee->pid; + entry->next_prio = wakee->prio; + entry->next_state = wakee->state; + entry->next_cpu = task_cpu(wakee); + + if (!filter_check_discard(call, entry, tr->buffer, event)) + ring_buffer_unlock_commit(tr->buffer, event); + ftrace_trace_stack(tr, flags, 6, pc); + ftrace_trace_userstack(tr, flags, pc); +} + static void probe_sched_wakeup(struct rq *__rq, struct task_struct *wakee, int success) { diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 00dd6485bdd..d2cdbabb4ea 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -288,6 +288,7 @@ trace_selftest_startup_function_graph(struct tracer *trace, * to detect and recover from possible hangs */ tracing_reset_online_cpus(tr); + set_graph_array(tr); ret = register_ftrace_graph(&trace_graph_return, &trace_graph_entry_watchdog); if (ret) { diff --git a/kernel/trace/trace_stack.c b/kernel/trace/trace_stack.c index 6a2a9d484cd..0da1cff08d6 100644 --- a/kernel/trace/trace_stack.c +++ b/kernel/trace/trace_stack.c @@ -234,15 +234,8 @@ static void t_stop(struct seq_file *m, void *p) static int trace_lookup_stack(struct seq_file *m, long i) { unsigned long addr = stack_dump_trace[i]; -#ifdef CONFIG_KALLSYMS - char str[KSYM_SYMBOL_LEN]; - sprint_symbol(str, addr); - - return seq_printf(m, "%s\n", str); -#else - return seq_printf(m, "%p\n", (void*)addr); -#endif + return seq_printf(m, "%pF\n", (void *)addr); } static void print_disabled(struct seq_file *m) diff --git a/kernel/trace/trace_stat.c b/kernel/trace/trace_stat.c index aea321c82fa..07c60b09258 100644 --- a/kernel/trace/trace_stat.c +++ b/kernel/trace/trace_stat.c @@ -49,7 +49,8 @@ static struct dentry *stat_dir; * but it will at least advance closer to the next one * to be released. */ -static struct rb_node *release_next(struct rb_node *node) +static struct rb_node *release_next(struct tracer_stat *ts, + struct rb_node *node) { struct stat_node *snode; struct rb_node *parent = rb_parent(node); @@ -67,6 +68,8 @@ static struct rb_node *release_next(struct rb_node *node) parent->rb_right = NULL; snode = container_of(node, struct stat_node, node); + if (ts->stat_release) + ts->stat_release(snode->stat); kfree(snode); return parent; @@ -78,7 +81,7 @@ static void __reset_stat_session(struct stat_session *session) struct rb_node *node = session->stat_root.rb_node; while (node) - node = release_next(node); + node = release_next(session->ts, node); session->stat_root = RB_ROOT; } diff --git a/kernel/trace/trace_stat.h b/kernel/trace/trace_stat.h index f3546a2cd82..8f03914b9a6 100644 --- a/kernel/trace/trace_stat.h +++ b/kernel/trace/trace_stat.h @@ -18,6 +18,8 @@ struct tracer_stat { int (*stat_cmp)(void *p1, void *p2); /* Print a stat entry */ int (*stat_show)(struct seq_file *s, void *p); + /* Release an entry */ + void (*stat_release)(void *stat); /* Print the headers of your stat entries */ int (*stat_headers)(struct seq_file *s); }; diff --git a/kernel/trace/trace_workqueue.c b/kernel/trace/trace_workqueue.c index 97fcea4acce..40cafb07dff 100644 --- a/kernel/trace/trace_workqueue.c +++ b/kernel/trace/trace_workqueue.c @@ -9,6 +9,7 @@ #include <trace/events/workqueue.h> #include <linux/list.h> #include <linux/percpu.h> +#include <linux/kref.h> #include "trace_stat.h" #include "trace.h" @@ -16,6 +17,7 @@ /* A cpu workqueue thread */ struct cpu_workqueue_stats { struct list_head list; + struct kref kref; int cpu; pid_t pid; /* Can be inserted from interrupt or user context, need to be atomic */ @@ -39,6 +41,11 @@ struct workqueue_global_stats { static DEFINE_PER_CPU(struct workqueue_global_stats, all_workqueue_stat); #define workqueue_cpu_stat(cpu) (&per_cpu(all_workqueue_stat, cpu)) +static void cpu_workqueue_stat_free(struct kref *kref) +{ + kfree(container_of(kref, struct cpu_workqueue_stats, kref)); +} + /* Insertion of a work */ static void probe_workqueue_insertion(struct task_struct *wq_thread, @@ -96,8 +103,8 @@ static void probe_workqueue_creation(struct task_struct *wq_thread, int cpu) return; } INIT_LIST_HEAD(&cws->list); + kref_init(&cws->kref); cws->cpu = cpu; - cws->pid = wq_thread->pid; spin_lock_irqsave(&workqueue_cpu_stat(cpu)->lock, flags); @@ -118,7 +125,7 @@ static void probe_workqueue_destruction(struct task_struct *wq_thread) list) { if (node->pid == wq_thread->pid) { list_del(&node->list); - kfree(node); + kref_put(&node->kref, cpu_workqueue_stat_free); goto found; } } @@ -137,9 +144,11 @@ static struct cpu_workqueue_stats *workqueue_stat_start_cpu(int cpu) spin_lock_irqsave(&workqueue_cpu_stat(cpu)->lock, flags); - if (!list_empty(&workqueue_cpu_stat(cpu)->list)) + if (!list_empty(&workqueue_cpu_stat(cpu)->list)) { ret = list_entry(workqueue_cpu_stat(cpu)->list.next, struct cpu_workqueue_stats, list); + kref_get(&ret->kref); + } spin_unlock_irqrestore(&workqueue_cpu_stat(cpu)->lock, flags); @@ -162,9 +171,9 @@ static void *workqueue_stat_start(struct tracer_stat *trace) static void *workqueue_stat_next(void *prev, int idx) { struct cpu_workqueue_stats *prev_cws = prev; + struct cpu_workqueue_stats *ret; int cpu = prev_cws->cpu; unsigned long flags; - void *ret = NULL; spin_lock_irqsave(&workqueue_cpu_stat(cpu)->lock, flags); if (list_is_last(&prev_cws->list, &workqueue_cpu_stat(cpu)->list)) { @@ -175,11 +184,14 @@ static void *workqueue_stat_next(void *prev, int idx) return NULL; } while (!(ret = workqueue_stat_start_cpu(cpu))); return ret; + } else { + ret = list_entry(prev_cws->list.next, + struct cpu_workqueue_stats, list); + kref_get(&ret->kref); } spin_unlock_irqrestore(&workqueue_cpu_stat(cpu)->lock, flags); - return list_entry(prev_cws->list.next, struct cpu_workqueue_stats, - list); + return ret; } static int workqueue_stat_show(struct seq_file *s, void *p) @@ -203,6 +215,13 @@ static int workqueue_stat_show(struct seq_file *s, void *p) return 0; } +static void workqueue_stat_release(void *stat) +{ + struct cpu_workqueue_stats *node = stat; + + kref_put(&node->kref, cpu_workqueue_stat_free); +} + static int workqueue_stat_headers(struct seq_file *s) { seq_printf(s, "# CPU INSERTED EXECUTED NAME\n"); @@ -215,6 +234,7 @@ struct tracer_stat workqueue_stats __read_mostly = { .stat_start = workqueue_stat_start, .stat_next = workqueue_stat_next, .stat_show = workqueue_stat_show, + .stat_release = workqueue_stat_release, .stat_headers = workqueue_stat_headers }; diff --git a/scripts/recordmcount.pl b/scripts/recordmcount.pl index 911ba7ffab8..090d300d739 100755 --- a/scripts/recordmcount.pl +++ b/scripts/recordmcount.pl @@ -57,7 +57,6 @@ # call mcount (offset: 0x5) # [...] # ret -# .globl my_func # other_func: # [...] # call mcount (offset: 0x1b) |