From e1112b4d96859367a93468027c9635e2ac04eb3f Mon Sep 17 00:00:00 2001 From: Tom Zanussi Date: Tue, 31 Mar 2009 00:48:49 -0500 Subject: tracing/filters: add run-time field descriptions to TRACE_EVENT_FORMAT events This patch adds run-time field descriptions to all the event formats exported using TRACE_EVENT_FORMAT. It also hooks up all the tracers that use them (i.e. the tracers in the 'ftrace subsystem') so they can also have their output filtered by the event-filtering mechanism. When I was testing this, there were a couple of things that fooled me into thinking the filters weren't working, when actually they were - I'll mention them here so others don't make the same mistakes (and file bug reports. ;-) One is that some of the tracers trace multiple events e.g. the sched_switch tracer uses the context_switch and wakeup events, and if you don't set filters on all of the traced events, the unfiltered output from the events without filters on them can make it look like the filtering as a whole isn't working properly, when actually it is doing what it was asked to do - it just wasn't asked to do the right thing. The other is that for the really high-volume tracers e.g. the function tracer, the volume of filtered events can be so high that it pushes the unfiltered events out of the ring buffer before they can be read so e.g. cat'ing the trace file repeatedly shows either no output, or once in awhile some output but that isn't there the next time you read the trace, which isn't what you normally expect when reading the trace file. If you read from the trace_pipe file though, you can catch them before they disappear. Changes from v1: As suggested by Frederic Weisbecker: - get rid of externs in functions - added unlikely() to filter_check_discard() Signed-off-by: Tom Zanussi Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 026be412f35..470ad9487ec 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -185,7 +185,7 @@ void filter_free_subsystem_preds(struct event_subsystem *system) } events_for_each(call) { - if (!call->name || !call->regfunc) + if (!call->define_fields) continue; if (!strcmp(call->system, system->name)) @@ -324,7 +324,7 @@ int filter_add_subsystem_pred(struct event_subsystem *system, events_for_each(call) { int err; - if (!call->name || !call->regfunc) + if (!call->define_fields) continue; if (strcmp(call->system, system->name)) -- cgit v1.2.3 From 0a19e53c1514ad8e9c3cbab40c6c3f52c86f403d Mon Sep 17 00:00:00 2001 From: Tom Zanussi Date: Mon, 13 Apr 2009 03:17:50 -0500 Subject: tracing/filters: allow on-the-fly filter switching This patch allows event filters to be safely removed or switched on-the-fly while avoiding the use of rcu or the suspension of tracing of previous versions. It does it by adding a new filter_pred_none() predicate function which does nothing and by never deallocating either the predicates or any of the filter_pred members used in matching; the predicate lists are allocated and initialized during ftrace_event_calls initialization. Whenever a filter is removed or replaced, the filter_pred_* functions currently in use by the affected ftrace_event_call are immediately switched over to to the filter_pred_none() function, while the rest of the filter_pred members are left intact, allowing any currently executing filter_pred_* functions to finish up, using the values they're currently using. In the case of filter replacement, the new predicate values are copied into the old predicates after the above step, and the filter_pred_none() functions are replaced by the filter_pred_* functions for the new filter. In this case, it is possible though very unlikely that a previous filter_pred_* is still running even after the filter_pred_none() switch and the switch to the new filter_pred_*. In that case, however, because nothing has been deallocated in the filter_pred, the worst that can happen is that the old filter_pred_* function sees the new values and as a result produces either a false positive or a false negative, depending on the values it finds. So one downside to this method is that rarely, it can produce a bad match during the filter switch, but it should be possible to live with that, IMHO. The other downside is that at least in this patch the predicate lists are always pre-allocated, taking up memory from the start. They could probably be allocated on first-use, and de-allocated when tracing is completely stopped - if this patch makes sense, I could create another one to do that later on. Oh, and it also places a restriction on the size of __arrays in events, currently set to 128, since they can't be larger than the now embedded str_val arrays in the filter_pred struct. Signed-off-by: Tom Zanussi Acked-by: Frederic Weisbecker Cc: Steven Rostedt Cc: paulmck@linux.vnet.ibm.com LKML-Reference: <1239610670.6660.49.camel@tropicana> Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 252 +++++++++++++++++++------------------ 1 file changed, 132 insertions(+), 120 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 9f8ecca34a5..de42dad42a8 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -82,25 +82,27 @@ static int filter_pred_string(struct filter_pred *pred, void *event) return match; } +static int filter_pred_none(struct filter_pred *pred, void *event) +{ + return 0; +} + /* return 1 if event matches, 0 otherwise (discard) */ int filter_match_preds(struct ftrace_event_call *call, void *rec) { int i, matched, and_failed = 0; struct filter_pred *pred; - for (i = 0; i < MAX_FILTER_PRED; i++) { - if (call->preds[i]) { - pred = call->preds[i]; - if (and_failed && !pred->or) - continue; - matched = pred->fn(pred, rec); - if (!matched && !pred->or) { - and_failed = 1; - continue; - } else if (matched && pred->or) - return 1; - } else - break; + for (i = 0; i < call->n_preds; i++) { + pred = call->preds[i]; + if (and_failed && !pred->or) + continue; + matched = pred->fn(pred, rec); + if (!matched && !pred->or) { + and_failed = 1; + continue; + } else if (matched && pred->or) + return 1; } if (and_failed) @@ -109,31 +111,29 @@ int filter_match_preds(struct ftrace_event_call *call, void *rec) return 1; } -void filter_print_preds(struct filter_pred **preds, struct trace_seq *s) +void filter_print_preds(struct filter_pred **preds, int n_preds, + struct trace_seq *s) { char *field_name; struct filter_pred *pred; int i; - if (!preds) { + if (!n_preds) { trace_seq_printf(s, "none\n"); return; } - for (i = 0; i < MAX_FILTER_PRED; i++) { - if (preds[i]) { - pred = preds[i]; - field_name = pred->field_name; - if (i) - trace_seq_printf(s, pred->or ? "|| " : "&& "); - trace_seq_printf(s, "%s ", field_name); - trace_seq_printf(s, pred->not ? "!= " : "== "); - if (pred->str_val) - trace_seq_printf(s, "%s\n", pred->str_val); - else - trace_seq_printf(s, "%llu\n", pred->val); - } else - break; + for (i = 0; i < n_preds; i++) { + pred = preds[i]; + field_name = pred->field_name; + if (i) + trace_seq_printf(s, pred->or ? "|| " : "&& "); + trace_seq_printf(s, "%s ", field_name); + trace_seq_printf(s, pred->not ? "!= " : "== "); + if (pred->str_len) + trace_seq_printf(s, "%s\n", pred->str_val); + else + trace_seq_printf(s, "%llu\n", pred->val); } } @@ -156,20 +156,69 @@ void filter_free_pred(struct filter_pred *pred) return; kfree(pred->field_name); - kfree(pred->str_val); kfree(pred); } -void filter_free_preds(struct ftrace_event_call *call) +static void filter_clear_pred(struct filter_pred *pred) +{ + kfree(pred->field_name); + pred->field_name = NULL; + pred->str_len = 0; +} + +static int filter_set_pred(struct filter_pred *dest, + struct filter_pred *src, + filter_pred_fn_t fn) +{ + *dest = *src; + dest->field_name = kstrdup(src->field_name, GFP_KERNEL); + if (!dest->field_name) + return -ENOMEM; + dest->fn = fn; + + return 0; +} + +void filter_disable_preds(struct ftrace_event_call *call) { int i; - if (call->preds) { - for (i = 0; i < MAX_FILTER_PRED; i++) + call->n_preds = 0; + + for (i = 0; i < MAX_FILTER_PRED; i++) + call->preds[i]->fn = filter_pred_none; +} + +int init_preds(struct ftrace_event_call *call) +{ + struct filter_pred *pred; + int i; + + call->n_preds = 0; + + call->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), GFP_KERNEL); + if (!call->preds) + return -ENOMEM; + + for (i = 0; i < MAX_FILTER_PRED; i++) { + pred = kzalloc(sizeof(*pred), GFP_KERNEL); + if (!pred) + goto oom; + pred->fn = filter_pred_none; + call->preds[i] = pred; + } + + return 0; + +oom: + for (i = 0; i < MAX_FILTER_PRED; i++) { + if (call->preds[i]) filter_free_pred(call->preds[i]); - kfree(call->preds); - call->preds = NULL; } + kfree(call->preds); + call->preds = NULL; + + return -ENOMEM; } void filter_free_subsystem_preds(struct event_subsystem *system) @@ -177,11 +226,12 @@ void filter_free_subsystem_preds(struct event_subsystem *system) struct ftrace_event_call *call = __start_ftrace_events; int i; - if (system->preds) { - for (i = 0; i < MAX_FILTER_PRED; i++) + if (system->n_preds) { + for (i = 0; i < system->n_preds; i++) filter_free_pred(system->preds[i]); kfree(system->preds); system->preds = NULL; + system->n_preds = 0; } events_for_each(call) { @@ -189,33 +239,31 @@ void filter_free_subsystem_preds(struct event_subsystem *system) continue; if (!strcmp(call->system, system->name)) - filter_free_preds(call); + filter_disable_preds(call); } } static int __filter_add_pred(struct ftrace_event_call *call, - struct filter_pred *pred) + struct filter_pred *pred, + filter_pred_fn_t fn) { - int i; + int idx, err; - if (call->preds && !pred->compound) - filter_free_preds(call); + if (call->n_preds && !pred->compound) + filter_disable_preds(call); - if (!call->preds) { - call->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), - GFP_KERNEL); - if (!call->preds) - return -ENOMEM; - } + if (call->n_preds == MAX_FILTER_PRED) + return -ENOSPC; - for (i = 0; i < MAX_FILTER_PRED; i++) { - if (!call->preds[i]) { - call->preds[i] = pred; - return 0; - } - } + idx = call->n_preds; + filter_clear_pred(call->preds[idx]); + err = filter_set_pred(call->preds[idx], pred, fn); + if (err) + return err; + + call->n_preds++; - return -ENOSPC; + return 0; } static int is_string_field(const char *type) @@ -229,98 +277,66 @@ static int is_string_field(const char *type) int filter_add_pred(struct ftrace_event_call *call, struct filter_pred *pred) { struct ftrace_event_field *field; + filter_pred_fn_t fn; field = find_event_field(call, pred->field_name); if (!field) return -EINVAL; + pred->fn = filter_pred_none; pred->offset = field->offset; if (is_string_field(field->type)) { - if (!pred->str_val) + if (!pred->str_len) return -EINVAL; - pred->fn = filter_pred_string; + fn = filter_pred_string; pred->str_len = field->size; - return __filter_add_pred(call, pred); + return __filter_add_pred(call, pred, fn); } else { - if (pred->str_val) + if (pred->str_len) return -EINVAL; } switch (field->size) { case 8: - pred->fn = filter_pred_64; + fn = filter_pred_64; break; case 4: - pred->fn = filter_pred_32; + fn = filter_pred_32; break; case 2: - pred->fn = filter_pred_16; + fn = filter_pred_16; break; case 1: - pred->fn = filter_pred_8; + fn = filter_pred_8; break; default: return -EINVAL; } - return __filter_add_pred(call, pred); -} - -static struct filter_pred *copy_pred(struct filter_pred *pred) -{ - struct filter_pred *new_pred = kmalloc(sizeof(*pred), GFP_KERNEL); - if (!new_pred) - return NULL; - - memcpy(new_pred, pred, sizeof(*pred)); - - if (pred->field_name) { - new_pred->field_name = kstrdup(pred->field_name, GFP_KERNEL); - if (!new_pred->field_name) { - kfree(new_pred); - return NULL; - } - } - - if (pred->str_val) { - new_pred->str_val = kstrdup(pred->str_val, GFP_KERNEL); - if (!new_pred->str_val) { - filter_free_pred(new_pred); - return NULL; - } - } - - return new_pred; + return __filter_add_pred(call, pred, fn); } int filter_add_subsystem_pred(struct event_subsystem *system, struct filter_pred *pred) { struct ftrace_event_call *call = __start_ftrace_events; - struct filter_pred *event_pred; - int i; - if (system->preds && !pred->compound) + if (system->n_preds && !pred->compound) filter_free_subsystem_preds(system); - if (!system->preds) { + if (!system->n_preds) { system->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), GFP_KERNEL); if (!system->preds) return -ENOMEM; } - for (i = 0; i < MAX_FILTER_PRED; i++) { - if (!system->preds[i]) { - system->preds[i] = pred; - break; - } - } - - if (i == MAX_FILTER_PRED) + if (system->n_preds == MAX_FILTER_PRED) return -ENOSPC; + system->preds[system->n_preds] = pred; + events_for_each(call) { int err; @@ -333,22 +349,16 @@ int filter_add_subsystem_pred(struct event_subsystem *system, if (!find_event_field(call, pred->field_name)) continue; - event_pred = copy_pred(pred); - if (!event_pred) - goto oom; - - err = filter_add_pred(call, event_pred); - if (err) - filter_free_pred(event_pred); - if (err == -ENOMEM) - goto oom; + err = filter_add_pred(call, pred); + if (err == -ENOMEM) { + system->preds[system->n_preds] = NULL; + return err; + } } - return 0; + system->n_preds++; -oom: - system->preds[i] = NULL; - return -ENOMEM; + return 0; } int filter_parse(char **pbuf, struct filter_pred *pred) @@ -410,7 +420,8 @@ int filter_parse(char **pbuf, struct filter_pred *pred) } } - if (!val_str) { + if (!val_str || !strlen(val_str) + || strlen(val_str) >= MAX_FILTER_STR_VAL) { pred->field_name = NULL; return -EINVAL; } @@ -419,11 +430,12 @@ int filter_parse(char **pbuf, struct filter_pred *pred) if (!pred->field_name) return -ENOMEM; + pred->str_len = 0; pred->val = simple_strtoull(val_str, &tmp, 0); if (tmp == val_str) { - pred->str_val = kstrdup(val_str, GFP_KERNEL); - if (!pred->str_val) - return -ENOMEM; + strncpy(pred->str_val, val_str, MAX_FILTER_STR_VAL); + pred->str_len = strlen(val_str); + pred->str_val[pred->str_len] = '\0'; } else if (*tmp != '\0') return -EINVAL; -- cgit v1.2.3 From a59fd6027218bd7c994e39d14afe0242f895144f Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 10 Apr 2009 13:52:20 -0400 Subject: tracing/events: convert event call sites to use a link list Impact: makes it possible to define events in modules The events are created by reading down the section that they are linked in by the macros. But this is not scalable to modules. This patch converts the manipulations to use a global link list, and on boot up it adds the items in the section to the list. This change will allow modules to add their tracing events to the list as well. Note, this change alone does not permit modules to use the TRACE_EVENT macros, but the change is needed for them to eventually do so. Signed-off-by: Steven Rostedt --- kernel/trace/trace_events_filter.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index de42dad42a8..d30b06b02b4 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -223,7 +223,7 @@ oom: void filter_free_subsystem_preds(struct event_subsystem *system) { - struct ftrace_event_call *call = __start_ftrace_events; + struct ftrace_event_call *call; int i; if (system->n_preds) { @@ -234,7 +234,7 @@ void filter_free_subsystem_preds(struct event_subsystem *system) system->n_preds = 0; } - events_for_each(call) { + list_for_each_entry(call, &ftrace_events, list) { if (!call->define_fields) continue; @@ -320,7 +320,7 @@ int filter_add_pred(struct ftrace_event_call *call, struct filter_pred *pred) int filter_add_subsystem_pred(struct event_subsystem *system, struct filter_pred *pred) { - struct ftrace_event_call *call = __start_ftrace_events; + struct ftrace_event_call *call; if (system->n_preds && !pred->compound) filter_free_subsystem_preds(system); @@ -337,7 +337,7 @@ int filter_add_subsystem_pred(struct event_subsystem *system, system->preds[system->n_preds] = pred; - events_for_each(call) { + list_for_each_entry(call, &ftrace_events, list) { int err; if (!call->define_fields) -- cgit v1.2.3 From 17c873ec280a03894bc718af817f7f24fa787ae1 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 10 Apr 2009 18:12:50 -0400 Subject: tracing/events: add export symbols for trace events in modules Impact: let modules add trace events The trace event code requires some functions to be exported to allow modules to use TRACE_EVENT. This patch adds EXPORT_SYMBOL_GPL to the necessary functions. Signed-off-by: Steven Rostedt --- kernel/trace/trace_events_filter.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index d30b06b02b4..f8e5eab0424 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -110,6 +110,7 @@ int filter_match_preds(struct ftrace_event_call *call, void *rec) return 1; } +EXPORT_SYMBOL_GPL(filter_match_preds); void filter_print_preds(struct filter_pred **preds, int n_preds, struct trace_seq *s) @@ -220,6 +221,7 @@ oom: return -ENOMEM; } +EXPORT_SYMBOL_GPL(init_preds); void filter_free_subsystem_preds(struct event_subsystem *system) { -- cgit v1.2.3 From ac1adc55fc71c7515caa2eb0e63e49b3d1c6a47c Mon Sep 17 00:00:00 2001 From: Tom Zanussi Date: Fri, 17 Apr 2009 00:27:08 -0500 Subject: tracing/filters: add filter_mutex to protect filter predicates This patch adds a filter_mutex to prevent the filter predicates from being accessed concurrently by various external functions. It's based on a previous patch by Li Zefan: "[PATCH 7/7] tracing/filters: make filter preds RCU safe" v2 changes: - fixed wrong value returned in a add_subsystem_pred() failure case noticed by Li Zefan. [ Impact: fix trace filter corruption/crashes on parallel access ] Signed-off-by: Tom Zanussi Reviewed-by: Li Zefan Tested-by: Li Zefan Cc: Frederic Weisbecker Cc: Steven Rostedt Cc: paulmck@linux.vnet.ibm.com LKML-Reference: <1239946028.6639.13.camel@tropicana> Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 90 +++++++++++++++++++++++++++++--------- 1 file changed, 70 insertions(+), 20 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index f8e5eab0424..e0fcfd2a16d 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -22,10 +22,13 @@ #include #include #include +#include #include "trace.h" #include "trace_output.h" +static DEFINE_MUTEX(filter_mutex); + static int filter_pred_64(struct filter_pred *pred, void *event) { u64 *addr = (u64 *)(event + pred->offset); @@ -112,8 +115,8 @@ int filter_match_preds(struct ftrace_event_call *call, void *rec) } EXPORT_SYMBOL_GPL(filter_match_preds); -void filter_print_preds(struct filter_pred **preds, int n_preds, - struct trace_seq *s) +static void __filter_print_preds(struct filter_pred **preds, int n_preds, + struct trace_seq *s) { char *field_name; struct filter_pred *pred; @@ -138,6 +141,21 @@ void filter_print_preds(struct filter_pred **preds, int n_preds, } } +void filter_print_preds(struct ftrace_event_call *call, struct trace_seq *s) +{ + mutex_lock(&filter_mutex); + __filter_print_preds(call->preds, call->n_preds, s); + mutex_unlock(&filter_mutex); +} + +void filter_print_subsystem_preds(struct event_subsystem *system, + struct trace_seq *s) +{ + mutex_lock(&filter_mutex); + __filter_print_preds(system->preds, system->n_preds, s); + mutex_unlock(&filter_mutex); +} + static struct ftrace_event_field * find_event_field(struct ftrace_event_call *call, char *name) { @@ -180,7 +198,7 @@ static int filter_set_pred(struct filter_pred *dest, return 0; } -void filter_disable_preds(struct ftrace_event_call *call) +static void __filter_disable_preds(struct ftrace_event_call *call) { int i; @@ -190,6 +208,13 @@ void filter_disable_preds(struct ftrace_event_call *call) call->preds[i]->fn = filter_pred_none; } +void filter_disable_preds(struct ftrace_event_call *call) +{ + mutex_lock(&filter_mutex); + __filter_disable_preds(call); + mutex_unlock(&filter_mutex); +} + int init_preds(struct ftrace_event_call *call) { struct filter_pred *pred; @@ -223,7 +248,7 @@ oom: } EXPORT_SYMBOL_GPL(init_preds); -void filter_free_subsystem_preds(struct event_subsystem *system) +static void __filter_free_subsystem_preds(struct event_subsystem *system) { struct ftrace_event_call *call; int i; @@ -241,18 +266,25 @@ void filter_free_subsystem_preds(struct event_subsystem *system) continue; if (!strcmp(call->system, system->name)) - filter_disable_preds(call); + __filter_disable_preds(call); } } -static int __filter_add_pred(struct ftrace_event_call *call, - struct filter_pred *pred, - filter_pred_fn_t fn) +void filter_free_subsystem_preds(struct event_subsystem *system) +{ + mutex_lock(&filter_mutex); + __filter_free_subsystem_preds(system); + mutex_unlock(&filter_mutex); +} + +static int filter_add_pred_fn(struct ftrace_event_call *call, + struct filter_pred *pred, + filter_pred_fn_t fn) { int idx, err; if (call->n_preds && !pred->compound) - filter_disable_preds(call); + __filter_disable_preds(call); if (call->n_preds == MAX_FILTER_PRED) return -ENOSPC; @@ -276,7 +308,8 @@ static int is_string_field(const char *type) return 0; } -int filter_add_pred(struct ftrace_event_call *call, struct filter_pred *pred) +static int __filter_add_pred(struct ftrace_event_call *call, + struct filter_pred *pred) { struct ftrace_event_field *field; filter_pred_fn_t fn; @@ -293,7 +326,7 @@ int filter_add_pred(struct ftrace_event_call *call, struct filter_pred *pred) return -EINVAL; fn = filter_pred_string; pred->str_len = field->size; - return __filter_add_pred(call, pred, fn); + return filter_add_pred_fn(call, pred, fn); } else { if (pred->str_len) return -EINVAL; @@ -316,7 +349,18 @@ int filter_add_pred(struct ftrace_event_call *call, struct filter_pred *pred) return -EINVAL; } - return __filter_add_pred(call, pred, fn); + return filter_add_pred_fn(call, pred, fn); +} + +int filter_add_pred(struct ftrace_event_call *call, struct filter_pred *pred) +{ + int err; + + mutex_lock(&filter_mutex); + err = __filter_add_pred(call, pred); + mutex_unlock(&filter_mutex); + + return err; } int filter_add_subsystem_pred(struct event_subsystem *system, @@ -324,20 +368,27 @@ int filter_add_subsystem_pred(struct event_subsystem *system, { struct ftrace_event_call *call; + mutex_lock(&filter_mutex); + if (system->n_preds && !pred->compound) - filter_free_subsystem_preds(system); + __filter_free_subsystem_preds(system); if (!system->n_preds) { system->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), GFP_KERNEL); - if (!system->preds) + if (!system->preds) { + mutex_unlock(&filter_mutex); return -ENOMEM; + } } - if (system->n_preds == MAX_FILTER_PRED) + if (system->n_preds == MAX_FILTER_PRED) { + mutex_unlock(&filter_mutex); return -ENOSPC; + } system->preds[system->n_preds] = pred; + system->n_preds++; list_for_each_entry(call, &ftrace_events, list) { int err; @@ -348,17 +399,16 @@ int filter_add_subsystem_pred(struct event_subsystem *system, if (strcmp(call->system, system->name)) continue; - if (!find_event_field(call, pred->field_name)) - continue; - - err = filter_add_pred(call, pred); + err = __filter_add_pred(call, pred); if (err == -ENOMEM) { system->preds[system->n_preds] = NULL; + system->n_preds--; + mutex_unlock(&filter_mutex); return err; } } - system->n_preds++; + mutex_unlock(&filter_mutex); return 0; } -- cgit v1.2.3 From f66578a7637b87810cbb9041c4e3a77fd2fa4706 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Tue, 21 Apr 2009 17:12:11 +0800 Subject: tracing/filters: allow user-input to be integer-like string Suppose we would like to trace all tasks named '123', but this will fail: # echo 'parent_comm == 123' > events/sched/sched_process_fork/filter bash: echo: write error: Invalid argument Don't guess the type of the filter pred in filter_parse(), but instead we check it in __filter_add_pred(). [ Impact: extend allowed filter field string values ] Signed-off-by: Li Zefan Cc: Tom Zanussi Cc: Steven Rostedt Cc: Frederic Weisbecker LKML-Reference: <49ED8DEB.6000700@cn.fujitsu.com> Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 26 ++++++++++++-------------- 1 file changed, 12 insertions(+), 14 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index e0fcfd2a16d..65418288f95 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -313,6 +313,7 @@ static int __filter_add_pred(struct ftrace_event_call *call, { struct ftrace_event_field *field; filter_pred_fn_t fn; + unsigned long long val; field = find_event_field(call, pred->field_name); if (!field) @@ -322,14 +323,13 @@ static int __filter_add_pred(struct ftrace_event_call *call, pred->offset = field->offset; if (is_string_field(field->type)) { - if (!pred->str_len) - return -EINVAL; fn = filter_pred_string; pred->str_len = field->size; return filter_add_pred_fn(call, pred, fn); } else { - if (pred->str_len) + if (strict_strtoull(pred->str_val, 0, &val)) return -EINVAL; + pred->val = val; } switch (field->size) { @@ -413,12 +413,16 @@ int filter_add_subsystem_pred(struct event_subsystem *system, return 0; } +/* + * The filter format can be + * - 0, which means remove all filter preds + * - [||/&&] ==/!= + */ int filter_parse(char **pbuf, struct filter_pred *pred) { - char *tmp, *tok, *val_str = NULL; + char *tok, *val_str = NULL; int tok_n = 0; - /* field ==/!= number, or/and field ==/!= number, number */ while ((tok = strsep(pbuf, " \n"))) { if (tok_n == 0) { if (!strcmp(tok, "0")) { @@ -478,19 +482,13 @@ int filter_parse(char **pbuf, struct filter_pred *pred) return -EINVAL; } + strcpy(pred->str_val, val_str); + pred->str_len = strlen(val_str); + pred->field_name = kstrdup(pred->field_name, GFP_KERNEL); if (!pred->field_name) return -ENOMEM; - pred->str_len = 0; - pred->val = simple_strtoull(val_str, &tmp, 0); - if (tmp == val_str) { - strncpy(pred->str_val, val_str, MAX_FILTER_STR_VAL); - pred->str_len = strlen(val_str); - pred->str_val[pred->str_len] = '\0'; - } else if (*tmp != '\0') - return -EINVAL; - return 0; } -- cgit v1.2.3 From 30e673b230f9d556eb81ef68a7b1a08c8b3b142c Mon Sep 17 00:00:00 2001 From: Tom Zanussi Date: Tue, 28 Apr 2009 03:04:47 -0500 Subject: tracing/filters: move preds into event_filter object Create a new event_filter object, and move the pred-related members out of the call and subsystem objects and into the filter object - the details of the filter implementation don't need to be exposed in the call and subsystem in any case, and it will also help make the new parser implementation a little cleaner. [ Impact: refactor trace-filter code to prepare for new features ] Signed-off-by: Tom Zanussi Acked-by: Steven Rostedt Cc: fweisbec@gmail.com Cc: Li Zefan LKML-Reference: <1240905887.6416.119.camel@tropicana> Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 107 +++++++++++++++++++++++-------------- 1 file changed, 66 insertions(+), 41 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 65418288f95..1e861eca3d0 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -93,11 +93,12 @@ static int filter_pred_none(struct filter_pred *pred, void *event) /* return 1 if event matches, 0 otherwise (discard) */ int filter_match_preds(struct ftrace_event_call *call, void *rec) { + struct event_filter *filter = call->filter; int i, matched, and_failed = 0; struct filter_pred *pred; - for (i = 0; i < call->n_preds; i++) { - pred = call->preds[i]; + for (i = 0; i < filter->n_preds; i++) { + pred = filter->preds[i]; if (and_failed && !pred->or) continue; matched = pred->fn(pred, rec); @@ -115,20 +116,20 @@ int filter_match_preds(struct ftrace_event_call *call, void *rec) } EXPORT_SYMBOL_GPL(filter_match_preds); -static void __filter_print_preds(struct filter_pred **preds, int n_preds, +static void __filter_print_preds(struct event_filter *filter, struct trace_seq *s) { - char *field_name; struct filter_pred *pred; + char *field_name; int i; - if (!n_preds) { + if (!filter || !filter->n_preds) { trace_seq_printf(s, "none\n"); return; } - for (i = 0; i < n_preds; i++) { - pred = preds[i]; + for (i = 0; i < filter->n_preds; i++) { + pred = filter->preds[i]; field_name = pred->field_name; if (i) trace_seq_printf(s, pred->or ? "|| " : "&& "); @@ -144,7 +145,7 @@ static void __filter_print_preds(struct filter_pred **preds, int n_preds, void filter_print_preds(struct ftrace_event_call *call, struct trace_seq *s) { mutex_lock(&filter_mutex); - __filter_print_preds(call->preds, call->n_preds, s); + __filter_print_preds(call->filter, s); mutex_unlock(&filter_mutex); } @@ -152,7 +153,7 @@ void filter_print_subsystem_preds(struct event_subsystem *system, struct trace_seq *s) { mutex_lock(&filter_mutex); - __filter_print_preds(system->preds, system->n_preds, s); + __filter_print_preds(system->filter, s); mutex_unlock(&filter_mutex); } @@ -200,12 +201,14 @@ static int filter_set_pred(struct filter_pred *dest, static void __filter_disable_preds(struct ftrace_event_call *call) { + struct event_filter *filter = call->filter; int i; - call->n_preds = 0; + call->filter_active = 0; + filter->n_preds = 0; for (i = 0; i < MAX_FILTER_PRED; i++) - call->preds[i]->fn = filter_pred_none; + filter->preds[i]->fn = filter_pred_none; } void filter_disable_preds(struct ftrace_event_call *call) @@ -217,32 +220,39 @@ void filter_disable_preds(struct ftrace_event_call *call) int init_preds(struct ftrace_event_call *call) { + struct event_filter *filter; struct filter_pred *pred; int i; - call->n_preds = 0; - - call->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), GFP_KERNEL); - if (!call->preds) + filter = call->filter = kzalloc(sizeof(*filter), GFP_KERNEL); + if (!call->filter) return -ENOMEM; + call->filter_active = 0; + filter->n_preds = 0; + + filter->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), GFP_KERNEL); + if (!filter->preds) + goto oom; + for (i = 0; i < MAX_FILTER_PRED; i++) { pred = kzalloc(sizeof(*pred), GFP_KERNEL); if (!pred) goto oom; pred->fn = filter_pred_none; - call->preds[i] = pred; + filter->preds[i] = pred; } return 0; oom: for (i = 0; i < MAX_FILTER_PRED; i++) { - if (call->preds[i]) - filter_free_pred(call->preds[i]); + if (filter->preds[i]) + filter_free_pred(filter->preds[i]); } - kfree(call->preds); - call->preds = NULL; + kfree(filter->preds); + kfree(call->filter); + call->filter = NULL; return -ENOMEM; } @@ -250,15 +260,16 @@ EXPORT_SYMBOL_GPL(init_preds); static void __filter_free_subsystem_preds(struct event_subsystem *system) { + struct event_filter *filter = system->filter; struct ftrace_event_call *call; int i; - if (system->n_preds) { - for (i = 0; i < system->n_preds; i++) - filter_free_pred(system->preds[i]); - kfree(system->preds); - system->preds = NULL; - system->n_preds = 0; + if (filter && filter->n_preds) { + for (i = 0; i < filter->n_preds; i++) + filter_free_pred(filter->preds[i]); + kfree(filter->preds); + kfree(filter); + system->filter = NULL; } list_for_each_entry(call, &ftrace_events, list) { @@ -281,21 +292,23 @@ static int filter_add_pred_fn(struct ftrace_event_call *call, struct filter_pred *pred, filter_pred_fn_t fn) { + struct event_filter *filter = call->filter; int idx, err; - if (call->n_preds && !pred->compound) + if (filter->n_preds && !pred->compound) __filter_disable_preds(call); - if (call->n_preds == MAX_FILTER_PRED) + if (filter->n_preds == MAX_FILTER_PRED) return -ENOSPC; - idx = call->n_preds; - filter_clear_pred(call->preds[idx]); - err = filter_set_pred(call->preds[idx], pred, fn); + idx = filter->n_preds; + filter_clear_pred(filter->preds[idx]); + err = filter_set_pred(filter->preds[idx], pred, fn); if (err) return err; - call->n_preds++; + filter->n_preds++; + call->filter_active = 1; return 0; } @@ -366,29 +379,41 @@ int filter_add_pred(struct ftrace_event_call *call, struct filter_pred *pred) int filter_add_subsystem_pred(struct event_subsystem *system, struct filter_pred *pred) { + struct event_filter *filter = system->filter; struct ftrace_event_call *call; mutex_lock(&filter_mutex); - if (system->n_preds && !pred->compound) + if (filter && filter->n_preds && !pred->compound) { __filter_free_subsystem_preds(system); + filter = NULL; + } - if (!system->n_preds) { - system->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), + if (!filter) { + system->filter = kzalloc(sizeof(*filter), GFP_KERNEL); + if (!system->filter) { + mutex_unlock(&filter_mutex); + return -ENOMEM; + } + filter = system->filter; + filter->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), GFP_KERNEL); - if (!system->preds) { + + if (!filter->preds) { + kfree(system->filter); + system->filter = NULL; mutex_unlock(&filter_mutex); return -ENOMEM; } } - if (system->n_preds == MAX_FILTER_PRED) { + if (filter->n_preds == MAX_FILTER_PRED) { mutex_unlock(&filter_mutex); return -ENOSPC; } - system->preds[system->n_preds] = pred; - system->n_preds++; + filter->preds[filter->n_preds] = pred; + filter->n_preds++; list_for_each_entry(call, &ftrace_events, list) { int err; @@ -401,8 +426,8 @@ int filter_add_subsystem_pred(struct event_subsystem *system, err = __filter_add_pred(call, pred); if (err == -ENOMEM) { - system->preds[system->n_preds] = NULL; - system->n_preds--; + filter->preds[filter->n_preds] = NULL; + filter->n_preds--; mutex_unlock(&filter_mutex); return err; } -- cgit v1.2.3 From 8b3725621074040d380664964ffbc40610aef8c6 Mon Sep 17 00:00:00 2001 From: Tom Zanussi Date: Tue, 28 Apr 2009 03:04:59 -0500 Subject: tracing/filters: a better event parser Replace the current event parser hack with a better one. Filters are no longer specified predicate by predicate, but all at once and can use parens and any of the following operators: numeric fields: ==, !=, <, <=, >, >= string fields: ==, != predicates can be combined with the logical operators: &&, || examples: "common_preempt_count > 4" > filter "((sig >= 10 && sig < 15) || sig == 17) && comm != bash" > filter If there was an error, the erroneous string along with an error message can be seen by looking at the filter e.g.: ((sig >= 10 && sig < 15) || dsig == 17) && comm != bash ^ parse_error: Field not found Currently the caret for an error always appears at the beginning of the filter; a real position should be used, but the error message should be useful even without it. To clear a filter, '0' can be written to the filter file. Filters can also be set or cleared for a complete subsystem by writing the same filter as would be written to an individual event to the filter file at the root of the subsytem. Note however, that if any event in the subsystem lacks a field specified in the filter being set, the set will fail and all filters in the subsytem are automatically cleared. This change from the previous version was made because using only the fields that happen to exist for a given event would most likely result in a meaningless filter. Because the logical operators are now implemented as predicates, the maximum number of predicates in a filter was increased from 8 to 16. [ Impact: add new, extended trace-filter implementation ] Signed-off-by: Tom Zanussi Acked-by: Steven Rostedt Cc: fweisbec@gmail.com Cc: Li Zefan LKML-Reference: <1240905899.6416.121.camel@tropicana> Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 1020 ++++++++++++++++++++++++++++-------- 1 file changed, 796 insertions(+), 224 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 1e861eca3d0..f49486687ee 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -29,51 +29,130 @@ static DEFINE_MUTEX(filter_mutex); -static int filter_pred_64(struct filter_pred *pred, void *event) +enum filter_op_ids { - u64 *addr = (u64 *)(event + pred->offset); - u64 val = (u64)pred->val; - int match; - - match = (val == *addr) ^ pred->not; - - return match; -} - -static int filter_pred_32(struct filter_pred *pred, void *event) -{ - u32 *addr = (u32 *)(event + pred->offset); - u32 val = (u32)pred->val; - int match; - - match = (val == *addr) ^ pred->not; - - return match; -} - -static int filter_pred_16(struct filter_pred *pred, void *event) + OP_OR, + OP_AND, + OP_NE, + OP_EQ, + OP_LT, + OP_LE, + OP_GT, + OP_GE, + OP_NONE, + OP_OPEN_PAREN, +}; + +struct filter_op { + int id; + char *string; + int precedence; +}; + +static struct filter_op filter_ops[] = { + { OP_OR, "||", 1 }, + { OP_AND, "&&", 2 }, + { OP_NE, "!=", 4 }, + { OP_EQ, "==", 4 }, + { OP_LT, "<", 5 }, + { OP_LE, "<=", 5 }, + { OP_GT, ">", 5 }, + { OP_GE, ">=", 5 }, + { OP_NONE, "OP_NONE", 0 }, + { OP_OPEN_PAREN, "(", 0 }, +}; + +enum { + FILT_ERR_NONE, + FILT_ERR_INVALID_OP, + FILT_ERR_UNBALANCED_PAREN, + FILT_ERR_TOO_MANY_OPERANDS, + FILT_ERR_OPERAND_TOO_LONG, + FILT_ERR_FIELD_NOT_FOUND, + FILT_ERR_ILLEGAL_FIELD_OP, + FILT_ERR_ILLEGAL_INTVAL, + FILT_ERR_BAD_SUBSYS_FILTER, + FILT_ERR_TOO_MANY_PREDS, + FILT_ERR_MISSING_FIELD, + FILT_ERR_INVALID_FILTER, +}; + +static char *err_text[] = { + "No error", + "Invalid operator", + "Unbalanced parens", + "Too many operands", + "Operand too long", + "Field not found", + "Illegal operation for field type", + "Illegal integer value", + "Couldn't find or set field in one of a subsystem's events", + "Too many terms in predicate expression", + "Missing field name and/or value", + "Meaningless filter expression", +}; + +struct opstack_op { + int op; + struct list_head list; +}; + +struct postfix_elt { + int op; + char *operand; + struct list_head list; +}; + +struct filter_parse_state { + struct filter_op *ops; + struct list_head opstack; + struct list_head postfix; + int lasterr; + int lasterr_pos; + + struct { + char *string; + unsigned int cnt; + unsigned int tail; + } infix; + + struct { + char string[MAX_FILTER_STR_VAL]; + int pos; + unsigned int tail; + } operand; +}; + +DEFINE_COMPARISON_PRED(s64); +DEFINE_COMPARISON_PRED(u64); +DEFINE_COMPARISON_PRED(s32); +DEFINE_COMPARISON_PRED(u32); +DEFINE_COMPARISON_PRED(s16); +DEFINE_COMPARISON_PRED(u16); +DEFINE_COMPARISON_PRED(s8); +DEFINE_COMPARISON_PRED(u8); + +DEFINE_EQUALITY_PRED(64); +DEFINE_EQUALITY_PRED(32); +DEFINE_EQUALITY_PRED(16); +DEFINE_EQUALITY_PRED(8); + +static int filter_pred_and(struct filter_pred *pred __attribute((unused)), + void *event __attribute((unused)), + int val1, int val2) { - u16 *addr = (u16 *)(event + pred->offset); - u16 val = (u16)pred->val; - int match; - - match = (val == *addr) ^ pred->not; - - return match; + return val1 && val2; } -static int filter_pred_8(struct filter_pred *pred, void *event) +static int filter_pred_or(struct filter_pred *pred __attribute((unused)), + void *event __attribute((unused)), + int val1, int val2) { - u8 *addr = (u8 *)(event + pred->offset); - u8 val = (u8)pred->val; - int match; - - match = (val == *addr) ^ pred->not; - - return match; + return val1 || val2; } -static int filter_pred_string(struct filter_pred *pred, void *event) +static int filter_pred_string(struct filter_pred *pred, void *event, + int val1, int val2) { char *addr = (char *)(event + pred->offset); int cmp, match; @@ -85,7 +164,8 @@ static int filter_pred_string(struct filter_pred *pred, void *event) return match; } -static int filter_pred_none(struct filter_pred *pred, void *event) +static int filter_pred_none(struct filter_pred *pred, void *event, + int val1, int val2) { return 0; } @@ -94,66 +174,119 @@ static int filter_pred_none(struct filter_pred *pred, void *event) int filter_match_preds(struct ftrace_event_call *call, void *rec) { struct event_filter *filter = call->filter; - int i, matched, and_failed = 0; + int match, top = 0, val1 = 0, val2 = 0; + int stack[MAX_FILTER_PRED]; struct filter_pred *pred; + int i; for (i = 0; i < filter->n_preds; i++) { pred = filter->preds[i]; - if (and_failed && !pred->or) + if (!pred->pop_n) { + match = pred->fn(pred, rec, val1, val2); + stack[top++] = match; continue; - matched = pred->fn(pred, rec); - if (!matched && !pred->or) { - and_failed = 1; - continue; - } else if (matched && pred->or) - return 1; + } + if (pred->pop_n > top) { + WARN_ON_ONCE(1); + return 0; + } + val1 = stack[--top]; + val2 = stack[--top]; + match = pred->fn(pred, rec, val1, val2); + stack[top++] = match; } - if (and_failed) - return 0; - - return 1; + return stack[--top]; } EXPORT_SYMBOL_GPL(filter_match_preds); -static void __filter_print_preds(struct event_filter *filter, - struct trace_seq *s) +static void parse_error(struct filter_parse_state *ps, int err, int pos) { - struct filter_pred *pred; - char *field_name; - int i; + ps->lasterr = err; + ps->lasterr_pos = pos; +} - if (!filter || !filter->n_preds) { - trace_seq_printf(s, "none\n"); +static void remove_filter_string(struct event_filter *filter) +{ + kfree(filter->filter_string); + filter->filter_string = NULL; +} + +static int replace_filter_string(struct event_filter *filter, + char *filter_string) +{ + kfree(filter->filter_string); + filter->filter_string = kstrdup(filter_string, GFP_KERNEL); + if (!filter->filter_string) + return -ENOMEM; + + return 0; +} + +static int append_filter_string(struct event_filter *filter, + char *string) +{ + int newlen; + char *new_filter_string; + + BUG_ON(!filter->filter_string); + newlen = strlen(filter->filter_string) + strlen(string) + 1; + new_filter_string = kmalloc(newlen, GFP_KERNEL); + if (!new_filter_string) + return -ENOMEM; + + strcpy(new_filter_string, filter->filter_string); + strcat(new_filter_string, string); + kfree(filter->filter_string); + filter->filter_string = new_filter_string; + + return 0; +} + +static void append_filter_err(struct filter_parse_state *ps, + struct event_filter *filter) +{ + int pos = ps->lasterr_pos; + char *buf, *pbuf; + + buf = (char *)__get_free_page(GFP_TEMPORARY); + if (!buf) return; - } - for (i = 0; i < filter->n_preds; i++) { - pred = filter->preds[i]; - field_name = pred->field_name; - if (i) - trace_seq_printf(s, pred->or ? "|| " : "&& "); - trace_seq_printf(s, "%s ", field_name); - trace_seq_printf(s, pred->not ? "!= " : "== "); - if (pred->str_len) - trace_seq_printf(s, "%s\n", pred->str_val); - else - trace_seq_printf(s, "%llu\n", pred->val); - } + append_filter_string(filter, "\n"); + memset(buf, ' ', PAGE_SIZE); + if (pos > PAGE_SIZE - 128) + pos = 0; + buf[pos] = '^'; + pbuf = &buf[pos] + 1; + + sprintf(pbuf, "\nparse_error: %s\n", err_text[ps->lasterr]); + append_filter_string(filter, buf); + free_page((unsigned long) buf); } -void filter_print_preds(struct ftrace_event_call *call, struct trace_seq *s) +void print_event_filter(struct ftrace_event_call *call, struct trace_seq *s) { + struct event_filter *filter = call->filter; + mutex_lock(&filter_mutex); - __filter_print_preds(call->filter, s); + if (filter->filter_string) + trace_seq_printf(s, "%s\n", filter->filter_string); + else + trace_seq_printf(s, "none\n"); mutex_unlock(&filter_mutex); } -void filter_print_subsystem_preds(struct event_subsystem *system, +void print_subsystem_event_filter(struct event_subsystem *system, struct trace_seq *s) { + struct event_filter *filter = system->filter; + mutex_lock(&filter_mutex); - __filter_print_preds(system->filter, s); + if (filter->filter_string) + trace_seq_printf(s, "%s\n", filter->filter_string); + else + trace_seq_printf(s, "none\n"); mutex_unlock(&filter_mutex); } @@ -170,7 +303,7 @@ find_event_field(struct ftrace_event_call *call, char *name) return NULL; } -void filter_free_pred(struct filter_pred *pred) +static void filter_free_pred(struct filter_pred *pred) { if (!pred) return; @@ -191,15 +324,17 @@ static int filter_set_pred(struct filter_pred *dest, filter_pred_fn_t fn) { *dest = *src; - dest->field_name = kstrdup(src->field_name, GFP_KERNEL); - if (!dest->field_name) - return -ENOMEM; + if (src->field_name) { + dest->field_name = kstrdup(src->field_name, GFP_KERNEL); + if (!dest->field_name) + return -ENOMEM; + } dest->fn = fn; return 0; } -static void __filter_disable_preds(struct ftrace_event_call *call) +static void filter_disable_preds(struct ftrace_event_call *call) { struct event_filter *filter = call->filter; int i; @@ -211,13 +346,6 @@ static void __filter_disable_preds(struct ftrace_event_call *call) filter->preds[i]->fn = filter_pred_none; } -void filter_disable_preds(struct ftrace_event_call *call) -{ - mutex_lock(&filter_mutex); - __filter_disable_preds(call); - mutex_unlock(&filter_mutex); -} - int init_preds(struct ftrace_event_call *call) { struct event_filter *filter; @@ -258,48 +386,43 @@ oom: } EXPORT_SYMBOL_GPL(init_preds); -static void __filter_free_subsystem_preds(struct event_subsystem *system) +static void filter_free_subsystem_preds(struct event_subsystem *system) { struct event_filter *filter = system->filter; struct ftrace_event_call *call; int i; - if (filter && filter->n_preds) { + if (filter->n_preds) { for (i = 0; i < filter->n_preds; i++) filter_free_pred(filter->preds[i]); kfree(filter->preds); - kfree(filter); - system->filter = NULL; + filter->preds = NULL; + filter->n_preds = 0; } list_for_each_entry(call, &ftrace_events, list) { if (!call->define_fields) continue; - if (!strcmp(call->system, system->name)) - __filter_disable_preds(call); + if (!strcmp(call->system, system->name)) { + filter_disable_preds(call); + remove_filter_string(call->filter); + } } } -void filter_free_subsystem_preds(struct event_subsystem *system) -{ - mutex_lock(&filter_mutex); - __filter_free_subsystem_preds(system); - mutex_unlock(&filter_mutex); -} - -static int filter_add_pred_fn(struct ftrace_event_call *call, +static int filter_add_pred_fn(struct filter_parse_state *ps, + struct ftrace_event_call *call, struct filter_pred *pred, filter_pred_fn_t fn) { struct event_filter *filter = call->filter; int idx, err; - if (filter->n_preds && !pred->compound) - __filter_disable_preds(call); - - if (filter->n_preds == MAX_FILTER_PRED) + if (filter->n_preds == MAX_FILTER_PRED) { + parse_error(ps, FILT_ERR_TOO_MANY_PREDS, 0); return -ENOSPC; + } idx = filter->n_preds; filter_clear_pred(filter->preds[idx]); @@ -321,94 +444,132 @@ static int is_string_field(const char *type) return 0; } -static int __filter_add_pred(struct ftrace_event_call *call, - struct filter_pred *pred) +static int is_legal_op(struct ftrace_event_field *field, int op) +{ + if (is_string_field(field->type) && (op != OP_EQ && op != OP_NE)) + return 0; + + return 1; +} + +static filter_pred_fn_t select_comparison_fn(int op, int field_size, + int field_is_signed) +{ + filter_pred_fn_t fn = NULL; + + switch (field_size) { + case 8: + if (op == OP_EQ || op == OP_NE) + fn = filter_pred_64; + else if (field_is_signed) + fn = filter_pred_s64; + else + fn = filter_pred_u64; + break; + case 4: + if (op == OP_EQ || op == OP_NE) + fn = filter_pred_32; + else if (field_is_signed) + fn = filter_pred_s32; + else + fn = filter_pred_u32; + break; + case 2: + if (op == OP_EQ || op == OP_NE) + fn = filter_pred_16; + else if (field_is_signed) + fn = filter_pred_s16; + else + fn = filter_pred_u16; + break; + case 1: + if (op == OP_EQ || op == OP_NE) + fn = filter_pred_8; + else if (field_is_signed) + fn = filter_pred_s8; + else + fn = filter_pred_u8; + break; + } + + return fn; +} + +static int filter_add_pred(struct filter_parse_state *ps, + struct ftrace_event_call *call, + struct filter_pred *pred) { struct ftrace_event_field *field; filter_pred_fn_t fn; unsigned long long val; + pred->fn = filter_pred_none; + + if (pred->op == OP_AND) { + pred->pop_n = 2; + return filter_add_pred_fn(ps, call, pred, filter_pred_and); + } else if (pred->op == OP_OR) { + pred->pop_n = 2; + return filter_add_pred_fn(ps, call, pred, filter_pred_or); + } + field = find_event_field(call, pred->field_name); - if (!field) + if (!field) { + parse_error(ps, FILT_ERR_FIELD_NOT_FOUND, 0); return -EINVAL; + } - pred->fn = filter_pred_none; pred->offset = field->offset; + if (!is_legal_op(field, pred->op)) { + parse_error(ps, FILT_ERR_ILLEGAL_FIELD_OP, 0); + return -EINVAL; + } + if (is_string_field(field->type)) { fn = filter_pred_string; pred->str_len = field->size; - return filter_add_pred_fn(call, pred, fn); + if (pred->op == OP_NE) + pred->not = 1; + return filter_add_pred_fn(ps, call, pred, fn); } else { - if (strict_strtoull(pred->str_val, 0, &val)) + if (strict_strtoull(pred->str_val, 0, &val)) { + parse_error(ps, FILT_ERR_ILLEGAL_INTVAL, 0); return -EINVAL; + } pred->val = val; } - switch (field->size) { - case 8: - fn = filter_pred_64; - break; - case 4: - fn = filter_pred_32; - break; - case 2: - fn = filter_pred_16; - break; - case 1: - fn = filter_pred_8; - break; - default: + fn = select_comparison_fn(pred->op, field->size, field->is_signed); + if (!fn) { + parse_error(ps, FILT_ERR_INVALID_OP, 0); return -EINVAL; } - return filter_add_pred_fn(call, pred, fn); -} - -int filter_add_pred(struct ftrace_event_call *call, struct filter_pred *pred) -{ - int err; - - mutex_lock(&filter_mutex); - err = __filter_add_pred(call, pred); - mutex_unlock(&filter_mutex); + if (pred->op == OP_NE) + pred->not = 1; - return err; + return filter_add_pred_fn(ps, call, pred, fn); } -int filter_add_subsystem_pred(struct event_subsystem *system, - struct filter_pred *pred) +static int filter_add_subsystem_pred(struct filter_parse_state *ps, + struct event_subsystem *system, + struct filter_pred *pred, + char *filter_string) { struct event_filter *filter = system->filter; struct ftrace_event_call *call; - mutex_lock(&filter_mutex); - - if (filter && filter->n_preds && !pred->compound) { - __filter_free_subsystem_preds(system); - filter = NULL; - } - - if (!filter) { - system->filter = kzalloc(sizeof(*filter), GFP_KERNEL); - if (!system->filter) { - mutex_unlock(&filter_mutex); - return -ENOMEM; - } - filter = system->filter; + if (!filter->preds) { filter->preds = kzalloc(MAX_FILTER_PRED * sizeof(pred), GFP_KERNEL); - if (!filter->preds) { - kfree(system->filter); - system->filter = NULL; - mutex_unlock(&filter_mutex); + if (!filter->preds) return -ENOMEM; - } } if (filter->n_preds == MAX_FILTER_PRED) { - mutex_unlock(&filter_mutex); + parse_error(ps, FILT_ERR_TOO_MANY_PREDS, 0); return -ENOSPC; } @@ -424,97 +585,508 @@ int filter_add_subsystem_pred(struct event_subsystem *system, if (strcmp(call->system, system->name)) continue; - err = __filter_add_pred(call, pred); - if (err == -ENOMEM) { - filter->preds[filter->n_preds] = NULL; - filter->n_preds--; - mutex_unlock(&filter_mutex); + err = filter_add_pred(ps, call, pred); + if (err) { + filter_free_subsystem_preds(system); + parse_error(ps, FILT_ERR_BAD_SUBSYS_FILTER, 0); return err; } + replace_filter_string(call->filter, filter_string); } - mutex_unlock(&filter_mutex); + return 0; +} + +static void parse_init(struct filter_parse_state *ps, + struct filter_op *ops, + char *infix_string) +{ + memset(ps, '\0', sizeof(*ps)); + + ps->infix.string = infix_string; + ps->infix.cnt = strlen(infix_string); + ps->ops = ops; + + INIT_LIST_HEAD(&ps->opstack); + INIT_LIST_HEAD(&ps->postfix); +} + +static char infix_next(struct filter_parse_state *ps) +{ + ps->infix.cnt--; + + return ps->infix.string[ps->infix.tail++]; +} + +static char infix_peek(struct filter_parse_state *ps) +{ + if (ps->infix.tail == strlen(ps->infix.string)) + return 0; + + return ps->infix.string[ps->infix.tail]; +} + +static void infix_advance(struct filter_parse_state *ps) +{ + ps->infix.cnt--; + ps->infix.tail++; +} + +static inline int is_precedence_lower(struct filter_parse_state *ps, + int a, int b) +{ + return ps->ops[a].precedence < ps->ops[b].precedence; +} + +static inline int is_op_char(struct filter_parse_state *ps, char c) +{ + int i; + + for (i = 0; strcmp(ps->ops[i].string, "OP_NONE"); i++) { + if (ps->ops[i].string[0] == c) + return 1; + } return 0; } -/* - * The filter format can be - * - 0, which means remove all filter preds - * - [||/&&] ==/!= - */ -int filter_parse(char **pbuf, struct filter_pred *pred) -{ - char *tok, *val_str = NULL; - int tok_n = 0; - - while ((tok = strsep(pbuf, " \n"))) { - if (tok_n == 0) { - if (!strcmp(tok, "0")) { - pred->clear = 1; - return 0; - } else if (!strcmp(tok, "&&")) { - pred->or = 0; - pred->compound = 1; - } else if (!strcmp(tok, "||")) { - pred->or = 1; - pred->compound = 1; - } else - pred->field_name = tok; - tok_n = 1; - continue; +static int infix_get_op(struct filter_parse_state *ps, char firstc) +{ + char nextc = infix_peek(ps); + char opstr[3]; + int i; + + opstr[0] = firstc; + opstr[1] = nextc; + opstr[2] = '\0'; + + for (i = 0; strcmp(ps->ops[i].string, "OP_NONE"); i++) { + if (!strcmp(opstr, ps->ops[i].string)) { + infix_advance(ps); + return ps->ops[i].id; } - if (tok_n == 1) { - if (!pred->field_name) - pred->field_name = tok; - else if (!strcmp(tok, "!=")) - pred->not = 1; - else if (!strcmp(tok, "==")) - pred->not = 0; - else { - pred->field_name = NULL; + } + + opstr[1] = '\0'; + + for (i = 0; strcmp(ps->ops[i].string, "OP_NONE"); i++) { + if (!strcmp(opstr, ps->ops[i].string)) + return ps->ops[i].id; + } + + return OP_NONE; +} + +static inline void clear_operand_string(struct filter_parse_state *ps) +{ + memset(ps->operand.string, '\0', MAX_FILTER_STR_VAL); + ps->operand.tail = 0; +} + +static inline int append_operand_char(struct filter_parse_state *ps, char c) +{ + if (ps->operand.tail == MAX_FILTER_STR_VAL) + return -EINVAL; + + ps->operand.string[ps->operand.tail++] = c; + + return 0; +} + +static int filter_opstack_push(struct filter_parse_state *ps, int op) +{ + struct opstack_op *opstack_op; + + opstack_op = kmalloc(sizeof(*opstack_op), GFP_KERNEL); + if (!opstack_op) + return -ENOMEM; + + opstack_op->op = op; + list_add(&opstack_op->list, &ps->opstack); + + return 0; +} + +static int filter_opstack_empty(struct filter_parse_state *ps) +{ + return list_empty(&ps->opstack); +} + +static int filter_opstack_top(struct filter_parse_state *ps) +{ + struct opstack_op *opstack_op; + + if (filter_opstack_empty(ps)) + return OP_NONE; + + opstack_op = list_first_entry(&ps->opstack, struct opstack_op, list); + + return opstack_op->op; +} + +static int filter_opstack_pop(struct filter_parse_state *ps) +{ + struct opstack_op *opstack_op; + int op; + + if (filter_opstack_empty(ps)) + return OP_NONE; + + opstack_op = list_first_entry(&ps->opstack, struct opstack_op, list); + op = opstack_op->op; + list_del(&opstack_op->list); + + kfree(opstack_op); + + return op; +} + +static void filter_opstack_clear(struct filter_parse_state *ps) +{ + while (!filter_opstack_empty(ps)) + filter_opstack_pop(ps); +} + +static char *curr_operand(struct filter_parse_state *ps) +{ + return ps->operand.string; +} + +static int postfix_append_operand(struct filter_parse_state *ps, char *operand) +{ + struct postfix_elt *elt; + + elt = kmalloc(sizeof(*elt), GFP_KERNEL); + if (!elt) + return -ENOMEM; + + elt->op = OP_NONE; + elt->operand = kstrdup(operand, GFP_KERNEL); + if (!elt->operand) { + kfree(elt); + return -ENOMEM; + } + + list_add_tail(&elt->list, &ps->postfix); + + return 0; +} + +static int postfix_append_op(struct filter_parse_state *ps, int op) +{ + struct postfix_elt *elt; + + elt = kmalloc(sizeof(*elt), GFP_KERNEL); + if (!elt) + return -ENOMEM; + + elt->op = op; + elt->operand = NULL; + + list_add_tail(&elt->list, &ps->postfix); + + return 0; +} + +static void postfix_clear(struct filter_parse_state *ps) +{ + struct postfix_elt *elt; + + while (!list_empty(&ps->postfix)) { + elt = list_first_entry(&ps->postfix, struct postfix_elt, list); + kfree(elt->operand); + list_del(&elt->list); + } +} + +static int filter_parse(struct filter_parse_state *ps) +{ + int op, top_op; + char ch; + + while ((ch = infix_next(ps))) { + if (isspace(ch)) + continue; + + if (is_op_char(ps, ch)) { + op = infix_get_op(ps, ch); + if (op == OP_NONE) { + parse_error(ps, FILT_ERR_INVALID_OP, 0); return -EINVAL; } - tok_n = 2; + + if (strlen(curr_operand(ps))) { + postfix_append_operand(ps, curr_operand(ps)); + clear_operand_string(ps); + } + + while (!filter_opstack_empty(ps)) { + top_op = filter_opstack_top(ps); + if (!is_precedence_lower(ps, top_op, op)) { + top_op = filter_opstack_pop(ps); + postfix_append_op(ps, top_op); + continue; + } + break; + } + + filter_opstack_push(ps, op); continue; } - if (tok_n == 2) { - if (pred->compound) { - if (!strcmp(tok, "!=")) - pred->not = 1; - else if (!strcmp(tok, "==")) - pred->not = 0; - else { - pred->field_name = NULL; - return -EINVAL; - } - } else { - val_str = tok; - break; /* done */ + + if (ch == '(') { + filter_opstack_push(ps, OP_OPEN_PAREN); + continue; + } + + if (ch == ')') { + if (strlen(curr_operand(ps))) { + postfix_append_operand(ps, curr_operand(ps)); + clear_operand_string(ps); + } + + top_op = filter_opstack_pop(ps); + while (top_op != OP_NONE) { + if (top_op == OP_OPEN_PAREN) + break; + postfix_append_op(ps, top_op); + top_op = filter_opstack_pop(ps); + } + if (top_op == OP_NONE) { + parse_error(ps, FILT_ERR_UNBALANCED_PAREN, 0); + return -EINVAL; } - tok_n = 3; continue; } - if (tok_n == 3) { - val_str = tok; - break; /* done */ + if (append_operand_char(ps, ch)) { + parse_error(ps, FILT_ERR_OPERAND_TOO_LONG, 0); + return -EINVAL; + } + } + + if (strlen(curr_operand(ps))) + postfix_append_operand(ps, curr_operand(ps)); + + while (!filter_opstack_empty(ps)) { + top_op = filter_opstack_pop(ps); + if (top_op == OP_NONE) + break; + if (top_op == OP_OPEN_PAREN) { + parse_error(ps, FILT_ERR_UNBALANCED_PAREN, 0); + return -EINVAL; + } + postfix_append_op(ps, top_op); + } + + return 0; +} + +static struct filter_pred *create_pred(int op, char *operand1, char *operand2) +{ + struct filter_pred *pred; + + pred = kzalloc(sizeof(*pred), GFP_KERNEL); + if (!pred) + return NULL; + + pred->field_name = kstrdup(operand1, GFP_KERNEL); + if (!pred->field_name) { + kfree(pred); + return NULL; + } + + strcpy(pred->str_val, operand2); + pred->str_len = strlen(operand2); + + pred->op = op; + + return pred; +} + +static struct filter_pred *create_logical_pred(int op) +{ + struct filter_pred *pred; + + pred = kzalloc(sizeof(*pred), GFP_KERNEL); + if (!pred) + return NULL; + + pred->op = op; + + return pred; +} + +static int check_preds(struct filter_parse_state *ps) +{ + int n_normal_preds = 0, n_logical_preds = 0; + struct postfix_elt *elt; + + list_for_each_entry(elt, &ps->postfix, list) { + if (elt->op == OP_NONE) + continue; + + if (elt->op == OP_AND || elt->op == OP_OR) { + n_logical_preds++; + continue; } + n_normal_preds++; } - if (!val_str || !strlen(val_str) - || strlen(val_str) >= MAX_FILTER_STR_VAL) { - pred->field_name = NULL; + if (!n_normal_preds || n_logical_preds >= n_normal_preds) { + parse_error(ps, FILT_ERR_INVALID_FILTER, 0); return -EINVAL; } - strcpy(pred->str_val, val_str); - pred->str_len = strlen(val_str); + return 0; +} - pred->field_name = kstrdup(pred->field_name, GFP_KERNEL); - if (!pred->field_name) - return -ENOMEM; +static int replace_preds(struct event_subsystem *system, + struct ftrace_event_call *call, + struct filter_parse_state *ps, + char *filter_string) +{ + char *operand1 = NULL, *operand2 = NULL; + struct filter_pred *pred; + struct postfix_elt *elt; + int err; + + err = check_preds(ps); + if (err) + return err; + + list_for_each_entry(elt, &ps->postfix, list) { + if (elt->op == OP_NONE) { + if (!operand1) + operand1 = elt->operand; + else if (!operand2) + operand2 = elt->operand; + else { + parse_error(ps, FILT_ERR_TOO_MANY_OPERANDS, 0); + return -EINVAL; + } + continue; + } + + if (elt->op == OP_AND || elt->op == OP_OR) { + pred = create_logical_pred(elt->op); + 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) + return err; + + operand1 = operand2 = NULL; + continue; + } + + if (!operand1 || !operand2) { + parse_error(ps, FILT_ERR_MISSING_FIELD, 0); + return -EINVAL; + } + + pred = create_pred(elt->op, operand1, operand2); + 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) + return err; + + operand1 = operand2 = NULL; + } return 0; } +int apply_event_filter(struct ftrace_event_call *call, char *filter_string) +{ + int err; + + struct filter_parse_state *ps; + + mutex_lock(&filter_mutex); + + if (!strcmp(strstrip(filter_string), "0")) { + filter_disable_preds(call); + remove_filter_string(call->filter); + mutex_unlock(&filter_mutex); + return 0; + } + + ps = kzalloc(sizeof(*ps), GFP_KERNEL); + if (!ps) + return -ENOMEM; + + filter_disable_preds(call); + replace_filter_string(call->filter, filter_string); + + parse_init(ps, filter_ops, filter_string); + err = filter_parse(ps); + if (err) { + append_filter_err(ps, call->filter); + goto out; + } + + err = replace_preds(NULL, call, ps, filter_string); + if (err) + append_filter_err(ps, call->filter); + +out: + filter_opstack_clear(ps); + postfix_clear(ps); + kfree(ps); + + mutex_unlock(&filter_mutex); + + return err; +} + +int apply_subsystem_event_filter(struct event_subsystem *system, + char *filter_string) +{ + int err; + + struct filter_parse_state *ps; + + mutex_lock(&filter_mutex); + + if (!strcmp(strstrip(filter_string), "0")) { + filter_free_subsystem_preds(system); + remove_filter_string(system->filter); + mutex_unlock(&filter_mutex); + return 0; + } + + ps = kzalloc(sizeof(*ps), GFP_KERNEL); + if (!ps) + return -ENOMEM; + + filter_free_subsystem_preds(system); + replace_filter_string(system->filter, filter_string); + + parse_init(ps, filter_ops, filter_string); + err = filter_parse(ps); + if (err) { + append_filter_err(ps, system->filter); + goto out; + } + + err = replace_preds(system, NULL, ps, filter_string); + if (err) + append_filter_err(ps, system->filter); + +out: + filter_opstack_clear(ps); + postfix_clear(ps); + kfree(ps); + + mutex_unlock(&filter_mutex); + + return err; +} -- cgit v1.2.3 From 2df75e415709ad12862028916c772c1f377f6a7c Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Wed, 6 May 2009 10:33:04 +0800 Subject: tracing/events: fix memory leak when unloading module When unloading a module, memory allocated by init_preds() and trace_define_field() is not freed. [ Impact: fix memory leak ] Signed-off-by: Li Zefan Acked-by: Frederic Weisbecker Acked-by: Steven Rostedt Cc: Tom Zanussi LKML-Reference: <4A00F6E0.3040503@cn.fujitsu.com> Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 22 +++++++++++++++------- 1 file changed, 15 insertions(+), 7 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index f49486687ee..ce07b818671 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -346,6 +346,20 @@ static void filter_disable_preds(struct ftrace_event_call *call) filter->preds[i]->fn = filter_pred_none; } +void destroy_preds(struct ftrace_event_call *call) +{ + struct event_filter *filter = call->filter; + int i; + + for (i = 0; i < MAX_FILTER_PRED; i++) { + if (filter->preds[i]) + filter_free_pred(filter->preds[i]); + } + kfree(filter->preds); + kfree(filter); + call->filter = NULL; +} + int init_preds(struct ftrace_event_call *call) { struct event_filter *filter; @@ -374,13 +388,7 @@ int init_preds(struct ftrace_event_call *call) return 0; oom: - for (i = 0; i < MAX_FILTER_PRED; i++) { - if (filter->preds[i]) - filter_free_pred(filter->preds[i]); - } - kfree(filter->preds); - kfree(call->filter); - call->filter = NULL; + destroy_preds(call); return -ENOMEM; } -- cgit v1.2.3 From 20c8928abe70e204bd077ab6cfe23002d7788983 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Wed, 6 May 2009 10:33:45 +0800 Subject: tracing/events: fix concurrent access to ftrace_events list A module will add/remove its trace events when it gets loaded/unloaded, so the ftrace_events list is not "const", and concurrent access needs to be protected. This patch thus fixes races between loading/unloding modules and read 'available_events' or read/write 'set_event', etc. Below shows how to reproduce the race: # for ((; ;)) { cat /mnt/tracing/available_events; } > /dev/null & # for ((; ;)) { insmod trace-events-sample.ko; rmmod sample; } & After a while: BUG: unable to handle kernel paging request at 0010011c IP: [] t_next+0x1b/0x2d ... Call Trace: [] ? seq_read+0x217/0x30d [] ? seq_read+0x0/0x30d [] ? vfs_read+0x8f/0x136 [] ? sys_read+0x40/0x65 [] ? sysenter_do_call+0x12/0x36 [ Impact: fix races when concurrent accessing ftrace_events list ] Signed-off-by: Li Zefan Acked-by: Steven Rostedt Acked-by: Frederic Weisbecker Cc: Tom Zanussi Cc: Peter Zijlstra LKML-Reference: <4A00F709.3080800@cn.fujitsu.com> Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index ce07b818671..7ac69108527 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -408,6 +408,7 @@ static void filter_free_subsystem_preds(struct event_subsystem *system) filter->n_preds = 0; } + mutex_lock(&event_mutex); list_for_each_entry(call, &ftrace_events, list) { if (!call->define_fields) continue; @@ -417,6 +418,7 @@ static void filter_free_subsystem_preds(struct event_subsystem *system) remove_filter_string(call->filter); } } + mutex_unlock(&event_mutex); } static int filter_add_pred_fn(struct filter_parse_state *ps, @@ -567,6 +569,7 @@ static int filter_add_subsystem_pred(struct filter_parse_state *ps, { 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), @@ -584,8 +587,8 @@ static int filter_add_subsystem_pred(struct filter_parse_state *ps, filter->preds[filter->n_preds] = pred; filter->n_preds++; + mutex_lock(&event_mutex); list_for_each_entry(call, &ftrace_events, list) { - int err; if (!call->define_fields) continue; @@ -597,12 +600,13 @@ static int filter_add_subsystem_pred(struct filter_parse_state *ps, if (err) { filter_free_subsystem_preds(system); parse_error(ps, FILT_ERR_BAD_SUBSYS_FILTER, 0); - return err; + break; } replace_filter_string(call->filter, filter_string); } + mutex_unlock(&event_mutex); - return 0; + return err; } static void parse_init(struct filter_parse_state *ps, -- cgit v1.2.3 From e8808c1019b048a43686dbd25c188a035842c2e2 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sun, 3 May 2009 02:48:52 +0200 Subject: tracing/filters: support for filters of dynamic sized arrays Currently the filtering infrastructure supports well the numeric types and fixed sized array types. But the recently added __string() field uses a specific indirect offset mechanism which requires a specific predicate. Until now it wasn't supported. This patch adds this support and implies very few changes, only a new predicate is needed, the management of this specific field can be done through the usual string helpers in the filtering infrastructure. [ Impact: support all kinds of strings in the tracing filters ] Cc: Tom Zanussi Cc: Steven Rostedt Cc: Li Zefan Cc: Zhaolei Signed-off-by: Frederic Weisbecker --- kernel/trace/trace_events_filter.c | 44 +++++++++++++++++++++++++++++++++++--- 1 file changed, 41 insertions(+), 3 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 7ac69108527..01c76eb3e16 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -151,6 +151,7 @@ static int filter_pred_or(struct filter_pred *pred __attribute((unused)), return val1 || val2; } +/* Filter predicate for fixed sized arrays of characters */ static int filter_pred_string(struct filter_pred *pred, void *event, int val1, int val2) { @@ -164,6 +165,30 @@ static int filter_pred_string(struct filter_pred *pred, void *event, return match; } +/* + * Filter predicate for dynamic sized arrays of characters. + * These are implemented through a list of strings at the end + * of the entry. + * Also each of these strings have a field in the entry which + * contains its offset from the beginning of the entry. + * We have then first to get this field, dereference it + * and add it to the address of the entry, and at last we have + * the address of the string. + */ +static int filter_pred_strloc(struct filter_pred *pred, void *event, + int val1, int val2) +{ + int str_loc = *(int *)(event + pred->offset); + char *addr = (char *)(event + str_loc); + int cmp, match; + + cmp = strncmp(addr, pred->str_val, pred->str_len); + + match = (!cmp) ^ pred->not; + + return match; +} + static int filter_pred_none(struct filter_pred *pred, void *event, int val1, int val2) { @@ -446,10 +471,18 @@ static int filter_add_pred_fn(struct filter_parse_state *ps, return 0; } +enum { + FILTER_STATIC_STRING = 1, + FILTER_DYN_STRING +}; + static int is_string_field(const char *type) { if (strchr(type, '[') && strstr(type, "char")) - return 1; + return FILTER_STATIC_STRING; + + if (!strcmp(type, "__str_loc")) + return FILTER_DYN_STRING; return 0; } @@ -512,6 +545,7 @@ static int filter_add_pred(struct filter_parse_state *ps, struct ftrace_event_field *field; filter_pred_fn_t fn; unsigned long long val; + int string_type; pred->fn = filter_pred_none; @@ -536,8 +570,12 @@ static int filter_add_pred(struct filter_parse_state *ps, return -EINVAL; } - if (is_string_field(field->type)) { - fn = filter_pred_string; + string_type = is_string_field(field->type); + if (string_type) { + if (string_type == FILTER_STATIC_STRING) + fn = filter_pred_string; + else + fn = filter_pred_strloc; pred->str_len = field->size; if (pred->op == OP_NE) pred->not = 1; -- cgit v1.2.3 From 5928c3cc0ffcb6894bbab6be591b7ae1786b2d87 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sun, 3 May 2009 03:03:57 +0200 Subject: tracing/filters: support for operator reserved characters in strings When we set a filter for an event, such as: echo "name == my_lock_name" > \ /debug/tracing/events/lockdep/lock_acquired/filter then the following order of token type is parsed: - space - operator - parentheses - operand Because the operators and parentheses have a higher precedence than the operand characters, which is normal, then we can't use any string containing such special characters: ()=<>!&| To get this support and also avoid ambiguous intepretation from the parser or the human, we can do it using double quotes so that we keep the usual languages habits. Then after this patch you can still declare string condition like before: echo name == myname But if you want to compare against a string containing an operator character, you can use double quotes: echo 'name == "&myname"' Don't forget to include the whole expression into single quotes or the double ones will be eaten by echo. [ Impact: support strings with special characters for tracing filters ] Cc: Tom Zanussi Cc: Steven Rostedt Cc: Li Zefan Cc: Zhaolei Signed-off-by: Frederic Weisbecker --- kernel/trace/trace_events_filter.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 01c76eb3e16..8c62e5bdff0 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -851,10 +851,19 @@ static void postfix_clear(struct filter_parse_state *ps) static int filter_parse(struct filter_parse_state *ps) { + int in_string = 0; int op, top_op; char ch; while ((ch = infix_next(ps))) { + if (ch == '"') { + in_string ^= 1; + continue; + } + + if (in_string) + goto parse_operand; + if (isspace(ch)) continue; @@ -908,6 +917,7 @@ static int filter_parse(struct filter_parse_state *ps) } continue; } +parse_operand: if (append_operand_char(ps, ch)) { parse_error(ps, FILT_ERR_OPERAND_TOO_LONG, 0); return -EINVAL; -- cgit v1.2.3 From d94fc523f3c35bd8013f04827e94756cbc0212f4 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Thu, 7 May 2009 15:11:15 +0800 Subject: tracing/events: fix concurrent access to ftrace_events list, fix In filter_add_subsystem_pred() we should release event_mutex before calling filter_free_subsystem_preds(), since both functions hold event_mutex. [ Impact: fix deadlock when writing invalid pred into subsystem filter ] Signed-off-by: Li Zefan Cc: tzanussi@gmail.com Cc: a.p.zijlstra@chello.nl Cc: fweisbec@gmail.com Cc: rostedt@goodmis.org LKML-Reference: <4A028993.7020509@cn.fujitsu.com> Signed-off-by: Ingo Molnar --- kernel/trace/trace_events_filter.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 8c62e5bdff0..85ad6a8939a 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -636,14 +636,15 @@ static int filter_add_subsystem_pred(struct filter_parse_state *ps, err = filter_add_pred(ps, call, pred); if (err) { + mutex_unlock(&event_mutex); filter_free_subsystem_preds(system); parse_error(ps, FILT_ERR_BAD_SUBSYS_FILTER, 0); - break; + goto out; } replace_filter_string(call->filter, filter_string); } mutex_unlock(&event_mutex); - +out: return err; } -- cgit v1.2.3 From 8cd995b6deedf98b7694ed32a786ee7f793d1eec Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Fri, 15 May 2009 11:07:27 +0800 Subject: tracing/filters: add missing unlock in a failure path [ Impact: fix deadlock in a rare case we fail to allocate memory ] Signed-off-by: Li Zefan LKML-Reference: <4A0CDC6F.7070200@cn.fujitsu.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace_events_filter.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 85ad6a8939a..22c29984fe0 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -1079,9 +1079,10 @@ int apply_event_filter(struct ftrace_event_call *call, char *filter_string) return 0; } + err = -ENOMEM; ps = kzalloc(sizeof(*ps), GFP_KERNEL); if (!ps) - return -ENOMEM; + goto out_unlock; filter_disable_preds(call); replace_filter_string(call->filter, filter_string); @@ -1101,7 +1102,7 @@ out: filter_opstack_clear(ps); postfix_clear(ps); kfree(ps); - +out_unlock: mutex_unlock(&filter_mutex); return err; @@ -1123,9 +1124,10 @@ int apply_subsystem_event_filter(struct event_subsystem *system, return 0; } + err = -ENOMEM; ps = kzalloc(sizeof(*ps), GFP_KERNEL); if (!ps) - return -ENOMEM; + goto out_unlock; filter_free_subsystem_preds(system); replace_filter_string(system->filter, filter_string); @@ -1145,7 +1147,7 @@ out: filter_opstack_clear(ps); postfix_clear(ps); kfree(ps); - +out_unlock: mutex_unlock(&filter_mutex); return err; -- cgit v1.2.3 From 5872144f64b34a5942f6b4acedc90b02de72c58b Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Fri, 15 May 2009 11:07:56 +0800 Subject: tracing/filters: fix off-by-one bug We should leave the last slot for the ending '\0'. [ Impact: fix possible crash when the length of an operand is 128 ] Signed-off-by: Li Zefan LKML-Reference: <4A0CDC8C.30602@cn.fujitsu.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace_events_filter.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 22c29984fe0..a7430b16d24 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -736,7 +736,7 @@ static inline void clear_operand_string(struct filter_parse_state *ps) static inline int append_operand_char(struct filter_parse_state *ps, char c) { - if (ps->operand.tail == MAX_FILTER_STR_VAL) + if (ps->operand.tail == MAX_FILTER_STR_VAL - 1) return -EINVAL; ps->operand.string[ps->operand.tail++] = c; -- cgit v1.2.3 From 7fcb7c472f455d1711eb5a7633204dba8800a6d6 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Mon, 1 Jun 2009 15:35:46 +0800 Subject: tracing/events: introduce __dynamic_array() __string() is limited: - it's a char array, but we may want to define array with other types - a source string should be available, but we may just know the string size We introduce __dynamic_array() to break those limitations, and __string() becomes a wrapper of it. As a side effect, now __get_str() can be used in TP_fast_assign but not only TP_print. Take XFS for example, we have the string length in the dirent, but the string itself is not NULL-terminated, so __dynamic_array() can be used: TRACE_EVENT(xfs_dir2, TP_PROTO(struct xfs_da_args *args), TP_ARGS(args), TP_STRUCT__entry( __field(int, namelen) __dynamic_array(char, name, args->namelen + 1) ... ), TP_fast_assign( char *name = __get_str(name); if (args->namelen) memcpy(name, args->name, args->namelen); name[args->namelen] = '\0'; __entry->namelen = args->namelen; ), TP_printk("name %.*s namelen %d", __entry->namelen ? __get_str(name) : NULL __entry->namelen) ); [ Impact: allow defining dynamic size arrays ] Signed-off-by: Li Zefan LKML-Reference: <4A2384D2.3080403@cn.fujitsu.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace_events_filter.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel/trace/trace_events_filter.c') diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index a7430b16d24..db6e54bdb59 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -478,12 +478,12 @@ enum { static int is_string_field(const char *type) { + if (strstr(type, "__data_loc") && strstr(type, "char")) + return FILTER_DYN_STRING; + if (strchr(type, '[') && strstr(type, "char")) return FILTER_STATIC_STRING; - if (!strcmp(type, "__str_loc")) - return FILTER_DYN_STRING; - return 0; } -- cgit v1.2.3