perf/core improvements and fixes
Subsystems (3)
Arnaldo Carvalho de Melo
maintainerHi Ingo,
Please consider pulling,
- Arnaldo
The following changes since commit 275ae411e56f8f900fa364da29c4706f9af4e1f3:
perf/x86/intel/rapl: Fix pmus free during cleanup (2016-05-25 10:56:43 +0200)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-core-for-mingo-20160525
for you to fetch changes up to 83e1e314baf9a1424bf2f50953ed7d50612763c4:
tools: Pass arg to fdarray__filter's call back function (2016-05-25 17:27:25 -0300)
----------------------------------------------------------------
perf/core improvements and fixes:
User visible/kernel ABI:
- Per event callchain limit: Recently we introduced a sysctl to tune the max-stack for all events for which callchains were requested:
$ sysctl kernel.perf_event_max_stack kernel.perf_event_max_stack = 127
Now this patch introduces a way to configure this per event, i.e. this becomes possible:
$ perf record -e sched:*/max-stack=2/ -e block:*/max-stack=10/ -a
allowing finer tuning of how much buffer space callchains use.
This uses an u16 from the reserved space at the end, leaving another u16 for future use.
There has been interest in even finer tuning, namely to control the max stack for kernel and userspace callchains separately. Further discussion is needed, we may for instance use the remaining u16 for that and when it is present, assume that the sample_max_stack introduced in this patch applies for the kernel, and the u16 left is used for limiting the userspace callchain. (Arnaldo Carvalho de Melo)
- Fix kptr_restrict=2 related 'perf record' segfault (Wang Nan)
Infrastructure;
- Adopt get_main_thread from db-export.c (Andi Kleen)
- More prep work for backward ring buffer support (Wang Nan)
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
----------------------------------------------------------------
Andi Kleen (1):
perf thread: Adopt get_main_thread from db-export.c
Arnaldo Carvalho de Melo (2):
perf core: Per event callchain limit
perf tools: Per event max-stack settings
Wang Nan (7):
perf symbols: Check kptr_restrict for root
perf record: Fix crash when kptr is restricted
perf record: Robustify perf_event__synth_time_conv()
perf evlist: Don't poll and mmap overwritable events
perf evlist: Check 'base' pointer before checking refcnt when put a mmap perf evlist: Choose correct reading direction according to evlist->backward
tools: Pass arg to fdarray__filter's call back function
include/linux/perf_event.h | 2 +- include/uapi/linux/perf_event.h | 6 +++++- kernel/bpf/stackmap.c | 2 +- kernel/events/callchain.c | 14 ++++++++++++-- kernel/events/core.c | 5 ++++- tools/lib/api/fd/array.c | 5 +++-- tools/lib/api/fd/array.h | 3 ++- tools/perf/arch/x86/util/tsc.c | 2 ++ tools/perf/builtin-record.c | 9 ++++++++- tools/perf/tests/fdarray.c | 8 ++++---- tools/perf/util/callchain.h | 1 + tools/perf/util/db-export.c | 13 +------------ tools/perf/util/event.c | 2 ++ tools/perf/util/evlist.c | 43 ++++++++++++++++++++++++++++++++--------- tools/perf/util/evlist.h | 2 ++ tools/perf/util/evsel.c | 16 +++++++++++++-- tools/perf/util/evsel.h | 2 ++ tools/perf/util/parse-events.c | 8 ++++++++ tools/perf/util/parse-events.h | 1 + tools/perf/util/parse-events.l | 1 + tools/perf/util/session.c | 2 ++ tools/perf/util/symbol.c | 16 +++++++-------- tools/perf/util/thread.c | 11 +++++++++++ tools/perf/util/thread.h | 2 ++ 24 files changed, 131 insertions(+), 45 deletions(-)
Arnaldo Carvalho de Melo
maintainerPATCH 1/10 perf symbols: Check kptr_restrict for root
From: Wang Nan <[email protected]>
If kptr_restrict is set to 2, even root is not allowed to see pointers. This patch checks kptr_restrict even if euid == 0. For root, report error if kptr_restrict is 2.
Signed-off-by Wang Nan <[email protected]>
Cc Zefan Li <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/symbol.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)
tools/perf/util/symbol.c +8 -8
@@ -1933,17 +1933,17 @@ int setup_intlist(struct intlist **list, const char *list_str,static bool symbol__read_kptr_restrict (void ) {bool value = false; +FILE *fp =fopen ("/proc/sys/kernel/kptr_restrict" ,"r" ); -if (geteuid () !=0 ) { -FILE *fp =fopen ("/proc/sys/kernel/kptr_restrict" ,"r" ); -if (fp !=NULL ) { -char line[8 ]; +if (fp !=NULL ) { +char line[8 ]; -if (fgets (line,sizeof (line), fp) !=NULL ) - value =atoi (line) !=0 ; +if (fgets (line,sizeof (line), fp) !=NULL ) + value = (geteuid () !=0 ) ? + (atoi (line) !=0 ) : + (atoi (line) ==2 ); -fclose (fp); - } +fclose (fp); }return value;
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 2/10 perf record: Fix crash when kptr is restricted
In reply to Arnaldo Carvalho de MeloFrom: Wang Nan <[email protected]>
Before this patch, a simple 'perf record' could fail if kptr_restrict is set to 1 (for normal user) or 2 (for root):
# perf record ls
WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted,
check /proc/sys/kernel/kptr_restrict.Samples in kernel functions may not be resolved if a suitable vmlinux file is not found in the buildid cache or in the vmlinux path.
Samples in kernel modules won't be resolved at all.
If some relocation was applied (e.g. kexec) symbols may be misresolved even with a suitable vmlinux or kallsyms file.
Segmentation fault (core dumped)
This patch skips perf_event__synthesize_kernel_mmap() when kptr is not available.
Signed-off-by Wang Nan <[email protected]>
Fixes 45e90056904b ("perf machine: Do not bail out if not managing to read ref reloc symbol")
Cc Zefan Li <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/event.c | 2 ++ 1 file changed, 2 insertions(+)
tools/perf/util/event.c +2 -0
@@ -673,6 +673,8 @@ int perf_event__synthesize_kernel_mmap(struct perf_tool *tool,int err;union perf_event *event; +if (symbol_conf.kptr_restrict) +return -1 ;if (map ==NULL )return -1 ;
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 3/10 perf thread: Adopt get_main_thread from db-export.c
In reply to Arnaldo Carvalho de MeloFrom: Andi Kleen <[email protected]>
Move the get_main_thread function from db-export.c to thread.c so that it can be used elsewhere.
Signed-off-by Andi Kleen <[email protected]>
Cc Jiri Olsa <[email protected]>
[ Removed leftover bits from db-export.h ]
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/db-export.c | 13 +------------ tools/perf/util/thread.c | 11 +++++++++++ tools/perf/util/thread.h | 2 ++ 3 files changed, 14 insertions(+), 12 deletions(-)
tools/perf/util/db-export.c +1 -12
@@ -233,17 +233,6 @@ int db_export__symbol(struct db_export *dbe, struct symbol *sym,return 0 ; } -static struct thread *get_main_thread (struct machine *machine,struct thread *thread) -{ -if (thread->pid_ == thread->tid) -return thread__get (thread); - -if (thread->pid_ ==-1 ) -return NULL ; - -return machine__find_thread (machine, thread->pid_, thread->pid_); -} -static int db_ids_from_al (struct db_export *dbe,struct addr_location *al,u64 *dso_db_id,u64 *sym_db_id,u64 *offset) {
@@ -382,7 +371,7 @@ int db_export__sample(struct db_export *dbe, union perf_event *event,if (err)return err; - main_thread =get_main_thread (al->machine, thread); + main_thread =thread__main_thread (al->machine, thread);if (main_thread) comm =machine__thread_exec_comm (al->machine, main_thread);
tools/perf/util/thread.c +11 -0
@@ -265,3 +265,14 @@ void thread__find_cpumode_addr_location(struct thread *thread,break ; } } + +struct thread *thread__main_thread (struct machine *machine,struct thread *thread) +{ +if (thread->pid_ == thread->tid) +return thread__get (thread); + +if (thread->pid_ ==-1 ) +return NULL ; + +return machine__find_thread (machine, thread->pid_, thread->pid_); +}
tools/perf/util/thread.h +2 -0
@@ -81,6 +81,8 @@ void thread__insert_map(struct thread *thread, struct map *map);int thread__fork (struct thread *thread,struct thread *parent,u64 timestamp);size_t thread__fprintf (struct thread *thread,FILE *fp); +struct thread *thread__main_thread (struct machine *machine,struct thread *thread); +void thread__find_addr_map (struct thread *thread,u8 cpumode,enum map_type type,u64 addr,struct addr_location *al);
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 4/10 perf core: Per event callchain limit
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Additionally to being able to control the system wide maximum depth via /proc/sys/kernel/perf_event_max_stack, now we are able to ask for different depths per event, using perf_event_attr.sample_max_stack for that.
This uses an u16 hole at the end of perf_event_attr, that, when perf_event_attr.sample_type has the PERF_SAMPLE_CALLCHAIN, if sample_max_stack is zero, means use perf_event_max_stack, otherwise it'll be bounds checked under callchain_mutex.
Cc Adrian Hunter <[email protected]>
Cc Alexander Shishkin <[email protected]>
Cc Alexei Starovoitov <[email protected]>
Cc Brendan Gregg <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc He Kuang <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Milian Wolff <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Vince Weaver <[email protected]>
Cc Wang Nan <[email protected]>
Cc Zefan Li <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- include/linux/perf_event.h | 2 +- include/uapi/linux/perf_event.h | 6 +++++- kernel/bpf/stackmap.c | 2 +- kernel/events/callchain.c | 14 ++++++++++++-- kernel/events/core.c | 5 ++++- 5 files changed, 23 insertions(+), 6 deletions(-)
include/linux/perf_event.h +1 -1
@@ -1076,7 +1076,7 @@ extern void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, structextern struct perf_callchain_entry *get_perf_callchain (struct pt_regs *regs,u32 init_nr,bool kernel,bool user,u32 max_stack,bool crosstask,bool add_mark); -extern int get_callchain_buffers (void ); +extern int get_callchain_buffers (int max_stack);extern void put_callchain_buffers (void );extern int sysctl_perf_event_max_stack;
include/uapi/linux/perf_event.h +5 -1
@@ -276,6 +276,9 @@ enum perf_event_read_format {/* * Hardware event_id to monitor via a performance monitoring event: +* +* @sample_max_stack: Max number of frame pointers in a callchain, +* should be < /proc/sys/kernel/perf_event_max_stack */ struct perf_event_attr {
@@ -385,7 +388,8 @@ struct perf_event_attr {* Wakeup watermark for AUX area */ __u32 aux_watermark; -__u32 __reserved_2;/* align to __u64 */ +__u16 sample_max_stack; +__u16 __reserved_2;/* align to __u64 */ };#define perf_flags (attr) (*(&(attr)->read_format + 1))
kernel/bpf/stackmap.c +1 -1
@@ -99,7 +99,7 @@ static struct bpf_map *stack_map_alloc(union bpf_attr *attr)if (err) goto free_smap; - err =get_callchain_buffers (); + err =get_callchain_buffers (sysctl_perf_event_max_stack);if (err) goto free_smap;
kernel/events/callchain.c +12 -2
@@ -104,7 +104,7 @@ fail:return -ENOMEM ; } -int get_callchain_buffers (void ) +int get_callchain_buffers (int event_max_stack) {int err =0 ;int count;
@@ -121,6 +121,15 @@ int get_callchain_buffers(void)/* If the allocation failed, give up */ if (!callchain_cpus_entries) err = -ENOMEM ; +/* +* If requesting per event more than the global cap, +* return a different error to help userspace figure +* this out. +* +* And also do it here so that we have &callchain_mutex held. +*/ +if (event_max_stack > sysctl_perf_event_max_stack) + err = -EOVERFLOW ; goto exit; }
@@ -174,11 +183,12 @@ perf_callchain(struct perf_event *event, struct pt_regs *regs)bool user = !event->attr.exclude_callchain_user;/* Disallow cross-task user callchains. */ bool crosstask = event->ctx->task && event->ctx->task != current; +const u32 max_stack = event->attr.sample_max_stack;if (!kernel && !user)return NULL ; -return get_perf_callchain (regs,0 , kernel, user, sysctl_perf_event_max_stack, crosstask, true); +return get_perf_callchain (regs,0 , kernel, user, max_stack, crosstask, true); } struct perf_callchain_entry *
kernel/events/core.c +4 -1
@@ -8843,7 +8843,7 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu,if (!event->parent) {if (event->attr.sample_type &PERF_SAMPLE_CALLCHAIN ) { - err =get_callchain_buffers (); + err =get_callchain_buffers (attr->sample_max_stack);if (err) goto err_addr_filters; }
@@ -9165,6 +9165,9 @@ SYSCALL_DEFINE5(perf_event_open,return -EINVAL ; } +if (!attr.sample_max_stack) + attr.sample_max_stack = sysctl_perf_event_max_stack; +/* * In cgroup mode, the pid argument is used to pass the fd * opened to the cgroup directory in cgroupfs. The cpu argument
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 5/10 perf tools: Per event max-stack settings
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
The tooling counterpart, now it is possible to do:
# perf record -e sched:sched_switch/max-stack=10/ -e cycles/call-graph=dwarf,max-stack=4/ -e cpu-cycles/call-graph=dwarf,max-stack=1024/ usleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.052 MB perf.data (5 samples) ]
# perf evlist -v
sched:sched_switch: type: 2, size: 112, config: 0x110, { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|CALLCHAIN|CPU|PERIOD|RAW|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, sample_max_stack: 10
cycles/call-graph=dwarf,max-stack=4/: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|PERIOD|REGS_USER|STACK_USER|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, freq: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1, exclude_callchain_user: 1, sample_regs_user: 0xff0fff, sample_stack_user: 8192, sample_max_stack: 4
cpu-cycles/call-graph=dwarf,max-stack=1024/: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|PERIOD|REGS_USER|STACK_USER|IDENTIFIER, read_format: ID, disabled: 1, inherit: 1, freq: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1, exclude_callchain_user: 1, sample_regs_user: 0xff0fff, sample_stack_user: 8192, sample_max_stack: 1024
# Tip: use 'perf evlist --trace-fields' to show fields for tracepoint eventsUsing just /max-stack=N/ means /call-graph=fp,max-stack=N/, that should be further configurable by means of some .perfconfig knob.
Cc Adrian Hunter <[email protected]>
Cc Alexander Shishkin <[email protected]>
Cc Alexei Starovoitov <[email protected]>
Cc Brendan Gregg <[email protected]>
Cc David Ahern <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc He Kuang <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Milian Wolff <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Vince Weaver <[email protected]>
Cc Wang Nan <[email protected]>
Cc Zefan Li <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/callchain.h | 1 + tools/perf/util/evsel.c | 16 ++++++++++++++-- tools/perf/util/evsel.h | 2 ++ tools/perf/util/parse-events.c | 8 ++++++++ tools/perf/util/parse-events.h | 1 + tools/perf/util/parse-events.l | 1 + tools/perf/util/session.c | 2 ++ 7 files changed, 29 insertions(+), 2 deletions(-)
tools/perf/util/callchain.h +1 -0
@@ -94,6 +94,7 @@ struct callchain_param {enum perf_call_graph_mode record_mode;u32 dump_size;enum chain_mode mode; +u16 max_stack;u32 print_limit;double min_percent;sort_chain_func_t sort;
tools/perf/util/evsel.c +14 -2
@@ -572,6 +572,8 @@ void perf_evsel__config_callchain(struct perf_evsel *evsel,perf_evsel__set_sample_bit (evsel,CALLCHAIN ); + attr->sample_max_stack = param->max_stack; +if (param->record_mode ==CALLCHAIN_LBR ) {if (!opts->branch_stack) {if (attr->exclude_user) {
@@ -635,7 +637,8 @@ static void apply_config_terms(struct perf_evsel *evsel,struct perf_event_attr *attr = &evsel->attr;struct callchain_param param;u32 dump_size =0 ; -char *callgraph_buf =NULL ; +int max_stack =0 ; +const char *callgraph_buf =NULL ;/* callgraph default */ param.record_mode = callchain_param.record_mode;
@@ -662,6 +665,9 @@ static void apply_config_terms(struct perf_evsel *evsel,case PERF_EVSEL__CONFIG_TERM_STACK_USER : dump_size = term->val.stack_user;break ; +case PERF_EVSEL__CONFIG_TERM_MAX_STACK : + max_stack = term->val.max_stack; +break ;case PERF_EVSEL__CONFIG_TERM_INHERIT :/* * attr->inherit should has already been set by
@@ -677,7 +683,12 @@ static void apply_config_terms(struct perf_evsel *evsel, }/* User explicitly set per-event callgraph, clear the old setting and reset. */ -if ((callgraph_buf !=NULL ) || (dump_size >0 )) { +if ((callgraph_buf !=NULL ) || (dump_size >0 ) || max_stack) { +if (max_stack) { + param.max_stack = max_stack; +if (callgraph_buf ==NULL ) + callgraph_buf ="fp" ; + }/* parse callgraph parameters */ if (callgraph_buf !=NULL ) {
@@ -1329,6 +1340,7 @@ int perf_event_attr__fprintf(FILE *fp, struct perf_event_attr *attr,PRINT_ATTRf (clockid, p_signed);PRINT_ATTRf (sample_regs_intr, p_hex);PRINT_ATTRf (aux_watermark, p_unsigned); +PRINT_ATTRf (sample_max_stack, p_unsigned);return ret; }
tools/perf/util/evsel.h +2 -0
@@ -44,6 +44,7 @@ enum {PERF_EVSEL__CONFIG_TERM_CALLGRAPH ,PERF_EVSEL__CONFIG_TERM_STACK_USER ,PERF_EVSEL__CONFIG_TERM_INHERIT , +PERF_EVSEL__CONFIG_TERM_MAX_STACK , PERF_EVSEL__CONFIG_TERM_MAX, };
@@ -56,6 +57,7 @@ struct perf_evsel_config_term {bool time;char *callgraph;u64 stack_user; +int max_stack;bool inherit; } val; };
tools/perf/util/parse-events.c +8 -0
@@ -900,6 +900,7 @@ static const char *config_term_names[__PARSE_EVENTS__TERM_TYPE_NR] = { [PARSE_EVENTS__TERM_TYPE_STACKSIZE] = "stack-size", [PARSE_EVENTS__TERM_TYPE_NOINHERIT] = "no-inherit", [PARSE_EVENTS__TERM_TYPE_INHERIT] = "inherit", + [PARSE_EVENTS__TERM_TYPE_MAX_STACK] = "max-stack", };static bool config_term_shrinked;
@@ -995,6 +996,9 @@ do { \case PARSE_EVENTS__TERM_TYPE_NAME :CHECK_TYPE_VAL (STR );break ; +case PARSE_EVENTS__TERM_TYPE_MAX_STACK : +CHECK_TYPE_VAL (NUM ); +break ;default : err->str =strdup ("unknown term" ); err->idx = term->err_term;
@@ -1040,6 +1044,7 @@ static int config_term_tracepoint(struct perf_event_attr *attr,case PARSE_EVENTS__TERM_TYPE_STACKSIZE :case PARSE_EVENTS__TERM_TYPE_INHERIT :case PARSE_EVENTS__TERM_TYPE_NOINHERIT : +case PARSE_EVENTS__TERM_TYPE_MAX_STACK :return config_term_common (attr, term, err);default :if (err) {
@@ -1109,6 +1114,9 @@ do { \case PARSE_EVENTS__TERM_TYPE_NOINHERIT :ADD_CONFIG_TERM (INHERIT , inherit, term->val.num ?0 :1 );break ; +case PARSE_EVENTS__TERM_TYPE_MAX_STACK : +ADD_CONFIG_TERM (MAX_STACK , max_stack, term->val.num); +break ;default :break ; }
tools/perf/util/parse-events.h +1 -0
@@ -68,6 +68,7 @@ enum {PARSE_EVENTS__TERM_TYPE_STACKSIZE ,PARSE_EVENTS__TERM_TYPE_NOINHERIT ,PARSE_EVENTS__TERM_TYPE_INHERIT , +PARSE_EVENTS__TERM_TYPE_MAX_STACK , __PARSE_EVENTS__TERM_TYPE_NR, };
tools/perf/util/parse-events.l +1 -0
@@ -199,6 +199,7 @@ branch_type { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_BRANCH_SAMPLE_TYPE time { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_TIME); } call-graph { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_CALLGRAPH); } stack-size { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_STACKSIZE); } +max-stack { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_MAX_STACK); } inherit { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_INHERIT); } no-inherit { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_NOINHERIT); } , { return ','; }
tools/perf/util/session.c +2 -0
@@ -593,6 +593,7 @@ do { \ if (bswap_safe(f, 0)) \ attr->f = bswap_##sz(attr->f); \ } while(0) +#define bswap_field_16 (f) bswap_field(f, 16)#define bswap_field_32 (f) bswap_field(f, 32)#define bswap_field_64 (f) bswap_field(f, 64)
@@ -608,6 +609,7 @@ do { \bswap_field_64 (sample_regs_user);bswap_field_32 (sample_stack_user);bswap_field_32 (aux_watermark); +bswap_field_16 (sample_max_stack);/* * After read_format are bitfields. Check read_format because
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 6/10 perf record: Robustify perf_event__synth_time_conv()
In reply to Arnaldo Carvalho de MeloFrom: Wang Nan <[email protected]>
It is possible that all events in an evlist are overwritable. perf_event__synth_time_conv() should not crash in this case. record__pick_pc() is used to check avaliability.
Signed-off-by Wang Nan <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Zefan Li <[email protected]>
Signed-off-by He Kuang <[email protected]>
[ Split from a larger patch ]
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/arch/x86/util/tsc.c | 2 ++ tools/perf/builtin-record.c | 9 ++++++++- 2 files changed, 10 insertions(+), 1 deletion(-)
tools/perf/arch/x86/util/tsc.c +2 -0
@@ -62,6 +62,8 @@ int perf_event__synth_time_conv(const struct perf_event_mmap_page *pc,struct perf_tsc_conversion tc;int err; +if (!pc) +return 0 ; err =perf_read_tsc_conversion (pc, &tc);if (err == -EOPNOTSUPP )return 0 ;
tools/perf/builtin-record.c +8 -1
@@ -655,6 +655,13 @@ perf_event__synth_time_conv(const struct perf_event_mmap_page *pc __maybe_unusedreturn 0 ; } +static const struct perf_event_mmap_page *record__pick_pc (struct record *rec) +{ +if (rec->evlist && rec->evlist->mmap && rec->evlist->mmap[0 ].base) +return rec->evlist->mmap[0 ].base; +return NULL ; +} +static int record__synthesize (struct record *rec) {struct perf_session *session = rec->session;
@@ -692,7 +699,7 @@ static int record__synthesize(struct record *rec) } } - err =perf_event__synth_time_conv (rec->evlist->mmap[0 ].base, tool, + err =perf_event__synth_time_conv (record__pick_pc (rec), tool, process_synthesized_event, machine);if (err) goto out;
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 7/10 perf evlist: Don't poll and mmap overwritable events
In reply to Arnaldo Carvalho de MeloFrom: Wang Nan <[email protected]>
There's no need to receive events from overwritable ring buffer. Instead, perf should make them run in background until some external event of interest takes place. This patch makes ignores normal events from overwrite evlists.
Overwritable events must be mapped readonly and backward, so if evlist and evsel doesn't match (evsel->overwrite is true but either evlist is read/write or evlist is not backward, and vice versa), skip mapping it.
Signed-off-by Wang Nan <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Zefan Li <[email protected]>
Signed-off-by He Kuang <[email protected]>
[ Split from a larger patch ]
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/evlist.c | 23 +++++++++++++++++++---- 1 file changed, 19 insertions(+), 4 deletions(-)
tools/perf/util/evlist.c +19 -4
@@ -462,9 +462,9 @@ int perf_evlist__alloc_pollfd(struct perf_evlist *evlist)return 0 ; } -static int __perf_evlist__add_pollfd (struct perf_evlist *evlist,int fd,int idx) +static int __perf_evlist__add_pollfd (struct perf_evlist *evlist,int fd,int idx,short revent) { -int pos =fdarray__add (&evlist->pollfd, fd,POLLIN |POLLERR |POLLHUP ); +int pos =fdarray__add (&evlist->pollfd, fd, revent |POLLERR |POLLHUP );/* * Save the idx so that when we filter out fds POLLHUP'ed we can * close the associated evlist->mmap[] entry.
@@ -480,7 +480,7 @@ static int __perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd, int idxint perf_evlist__add_pollfd (struct perf_evlist *evlist,int fd) { -return __perf_evlist__add_pollfd (evlist, fd,-1 ); +return __perf_evlist__add_pollfd (evlist, fd,-1 ,POLLIN ); }static void perf_evlist__munmap_filtered (struct fdarray *fda,int fd)
@@ -983,15 +983,28 @@ static int __perf_evlist__mmap(struct perf_evlist *evlist, int idx,return 0 ; } +static bool +perf_evlist__should_poll (struct perf_evlist *evlist__maybe_unused , +struct perf_evsel *evsel) +{ +if (evsel->overwrite) +return false; +return true; +} +static int perf_evlist__mmap_per_evsel (struct perf_evlist *evlist,int idx,struct mmap_params *mp,int cpu,int thread,int *output) {struct perf_evsel *evsel; +int revent;evlist__for_each (evlist, evsel) {int fd; +if (evsel->overwrite != (evlist->overwrite && evlist->backward)) +continue ; +if (evsel->system_wide && thread)continue ;
@@ -1008,6 +1021,8 @@ static int perf_evlist__mmap_per_evsel(struct perf_evlist *evlist, int idx,perf_evlist__mmap_get (evlist, idx); } + revent =perf_evlist__should_poll (evlist, evsel) ?POLLIN :0 ; +/* * The system_wide flag causes a selected event to be opened * always without a pid. Consequently it will never get a
@@ -1016,7 +1031,7 @@ static int perf_evlist__mmap_per_evsel(struct perf_evlist *evlist, int idx,* Therefore don't add it for polling. */ if (!evsel->system_wide && -__perf_evlist__add_pollfd (evlist, fd, idx) <0 ) { +__perf_evlist__add_pollfd (evlist, fd, idx, revent) <0 ) {perf_evlist__mmap_put (evlist, idx);return -1 ; }
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 8/10 perf evlist: Check 'base' pointer before checking refcnt when put a mmap
In reply to Arnaldo Carvalho de MeloFrom: Wang Nan <[email protected]>
evlist->mmap[i]->refcnt could be 0 if an evlist has no evsel or if all evsels don't match the evlist during mmap. For example, when all evsels are overwritable but the evlist itself is normal. To avoid crashing, perf should check 'base' pointer before checking refcnt, and raise bug only when base is not NULL.
Signed-off-by Wang Nan <[email protected]>
Cc He Kuang <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Zefan Li <[email protected]>
[ Renamed 'mmap' variable, it is reserved in old distros such as Ubuntu 12.04, breaking the build ]
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/evlist.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
tools/perf/util/evlist.c +4 -2
@@ -856,9 +856,11 @@ static void perf_evlist__mmap_get(struct perf_evlist *evlist, int idx)static void perf_evlist__mmap_put (struct perf_evlist *evlist,int idx) { -BUG_ON (atomic_read (&evlist->mmap[idx].refcnt) ==0 ); +struct perf_mmap *md = &evlist->mmap[idx]; + +BUG_ON (md->base &&atomic_read (&md->refcnt) ==0 ); -if (atomic_dec_and_test (&evlist->mmap[idx].refcnt)) +if (atomic_dec_and_test (&md->refcnt))__perf_evlist__munmap (evlist, idx); }
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 9/10 perf evlist: Choose correct reading direction according to evlist->backward
In reply to Arnaldo Carvalho de MeloFrom: Wang Nan <[email protected]>
Now we have evlist->backward to indicate the mmap direction. Make perf_evlist__mmap_read() choose right direction automatically.
Signed-off-by Wang Nan <[email protected]>
Cc He Kuang <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Zefan Li <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/evlist.c | 9 ++++++++- tools/perf/util/evlist.h | 2 ++ 2 files changed, 10 insertions(+), 1 deletion(-)
tools/perf/util/evlist.c +8 -1
@@ -777,7 +777,7 @@ broken_event:return event; } -union perf_event *perf_evlist__mmap_read (struct perf_evlist *evlist,int idx) +union perf_event *perf_evlist__mmap_read_forward (struct perf_evlist *evlist,int idx) {struct perf_mmap *md = &evlist->mmap[idx];u64 head;
@@ -832,6 +832,13 @@ perf_evlist__mmap_read_backward(struct perf_evlist *evlist, int idx)return perf_mmap__read (md, false, start, end, &md->prev); } +union perf_event *perf_evlist__mmap_read (struct perf_evlist *evlist,int idx) +{ +if (!evlist->backward) +return perf_evlist__mmap_read_forward (evlist, idx); +return perf_evlist__mmap_read_backward (evlist, idx); +} +void perf_evlist__mmap_read_catchup (struct perf_evlist *evlist,int idx) {struct perf_mmap *md = &evlist->mmap[idx];
tools/perf/util/evlist.h +2 -0
@@ -131,6 +131,8 @@ struct perf_sample_id *perf_evlist__id2sid(struct perf_evlist *evlist, u64 id);union perf_event *perf_evlist__mmap_read (struct perf_evlist *evlist,int idx); +union perf_event *perf_evlist__mmap_read_forward (struct perf_evlist *evlist, +int idx);union perf_event *perf_evlist__mmap_read_backward (struct perf_evlist *evlist,int idx);void perf_evlist__mmap_read_catchup (struct perf_evlist *evlist,int idx);
Signature · 1 line
2.5.5
Arnaldo Carvalho de Melo
maintainerPATCH 10/10 tools: Pass arg to fdarray__filter's call back function
In reply to Arnaldo Carvalho de MeloFrom: Wang Nan <[email protected]>
Before this patch there's no way to pass arguments to fdarray__filter's call back function.
This improvement will be used by 'perf record' to support unmapping ring buffer for both main evlist and overwrite evlist. Without this patch there's no way to track overwrite evlist from 'struct fdarray'.
Signed-off-by Wang Nan <[email protected]>
Cc He Kuang <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Zefan Li <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/lib/api/fd/array.c | 5 +++-- tools/lib/api/fd/array.h | 3 ++- tools/perf/tests/fdarray.c | 8 ++++---- tools/perf/util/evlist.c | 5 +++-- 4 files changed, 12 insertions(+), 9 deletions(-)
tools/lib/api/fd/array.c +3 -2
@@ -85,7 +85,8 @@ int fdarray__add(struct fdarray *fda, int fd, short revents) }int fdarray__filter (struct fdarray *fda,short revents, -void (*entry_destructor)(struct fdarray *fda,int fd)) +void (*entry_destructor)(struct fdarray *fda,int fd,void *arg), +void *arg) {int fd, nr =0 ;
@@ -95,7 +96,7 @@ int fdarray__filter(struct fdarray *fda, short revents,for (fd =0 ; fd < fda->nr; ++fd) {if (fda->entries[fd].revents & revents) {if (entry_destructor) -entry_destructor (fda, fd); +entry_destructor (fda, fd, arg);continue ; }
tools/lib/api/fd/array.h +2 -1
@@ -34,7 +34,8 @@ void fdarray__delete(struct fdarray *fda);int fdarray__add (struct fdarray *fda,int fd,short revents);int fdarray__poll (struct fdarray *fda,int timeout);int fdarray__filter (struct fdarray *fda,short revents, -void (*entry_destructor)(struct fdarray *fda,int fd)); +void (*entry_destructor)(struct fdarray *fda,int fd,void *arg), +void *arg);int fdarray__grow (struct fdarray *fda,int extra);int fdarray__fprintf (struct fdarray *fda,FILE *fp);
tools/perf/tests/fdarray.c +4 -4
@@ -36,7 +36,7 @@ int test__fdarray__filter(int subtest __maybe_unused) }fdarray__init_revents (fda,POLLIN ); - nr_fds =fdarray__filter (fda,POLLHUP ,NULL ); + nr_fds =fdarray__filter (fda,POLLHUP ,NULL ,NULL );if (nr_fds != fda->nr_alloc) {pr_debug ("\nfdarray__filter()=%d != %d shouldn't have filtered anything" , nr_fds, fda->nr_alloc);
@@ -44,7 +44,7 @@ int test__fdarray__filter(int subtest __maybe_unused) }fdarray__init_revents (fda,POLLHUP ); - nr_fds =fdarray__filter (fda,POLLHUP ,NULL ); + nr_fds =fdarray__filter (fda,POLLHUP ,NULL ,NULL );if (nr_fds !=0 ) {pr_debug ("\nfdarray__filter()=%d != %d, should have filtered all fds" , nr_fds, fda->nr_alloc);
@@ -57,7 +57,7 @@ int test__fdarray__filter(int subtest __maybe_unused)pr_debug ("\nfiltering all but fda->entries[2]:" );fdarray__fprintf_prefix (fda,"before" , stderr); - nr_fds =fdarray__filter (fda,POLLHUP ,NULL ); + nr_fds =fdarray__filter (fda,POLLHUP ,NULL ,NULL );fdarray__fprintf_prefix (fda," after" , stderr);if (nr_fds !=1 ) {pr_debug ("\nfdarray__filter()=%d != 1, should have left just one event" , nr_fds);
@@ -78,7 +78,7 @@ int test__fdarray__filter(int subtest __maybe_unused)pr_debug ("\nfiltering all but (fda->entries[0], fda->entries[3]):" );fdarray__fprintf_prefix (fda,"before" , stderr); - nr_fds =fdarray__filter (fda,POLLHUP ,NULL ); + nr_fds =fdarray__filter (fda,POLLHUP ,NULL ,NULL );fdarray__fprintf_prefix (fda," after" , stderr);if (nr_fds !=2 ) { pr_debug("\nfdarray__filter()=%d != 2, should have left just two events",
tools/perf/util/evlist.c +3 -2
@@ -483,7 +483,8 @@ int perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd)return __perf_evlist__add_pollfd (evlist, fd,-1 ,POLLIN ); } -static void perf_evlist__munmap_filtered (struct fdarray *fda,int fd) +static void perf_evlist__munmap_filtered (struct fdarray *fda,int fd, +void *arg__maybe_unused ) { struct perf_evlist *evlist = container_of(fda, struct perf_evlist, pollfd);
@@ -493,7 +494,7 @@ static void perf_evlist__munmap_filtered(struct fdarray *fda, int fd)int perf_evlist__filter_pollfd (struct perf_evlist *evlist,short revents_and_mask) {return fdarray__filter (&evlist->pollfd, revents_and_mask, - perf_evlist__munmap_filtered); + perf_evlist__munmap_filtered,NULL ); }int perf_evlist__poll (struct perf_evlist *evlist,int timeout)
Signature · 1 line
2.5.5