diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2015-06-22 15:19:21 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2015-06-22 15:19:21 -0700 |
commit | c58267e9fa7b0345dd9006939254701e3622ca6a (patch) | |
tree | 9a96adabaa1c61ecbb9e5e8653d5085dad27ef07 /include | |
parent | 1bf7067c6e173dc10411704db48338ed69c05565 (diff) | |
parent | a9a3cd900fbbcbf837d65653105e7bfc583ced09 (diff) | |
download | linux-c58267e9fa7b0345dd9006939254701e3622ca6a.tar.bz2 |
Merge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf updates from Ingo Molnar:
"Kernel side changes mostly consist of work on x86 PMU drivers:
- x86 Intel PT (hardware CPU tracer) improvements (Alexander
Shishkin)
- x86 Intel CQM (cache quality monitoring) improvements (Thomas
Gleixner)
- x86 Intel PEBSv3 support (Peter Zijlstra)
- x86 Intel PEBS interrupt batching support for lower overhead
sampling (Zheng Yan, Kan Liang)
- x86 PMU scheduler fixes and improvements (Peter Zijlstra)
There's too many tooling improvements to list them all - here are a
few select highlights:
'perf bench':
- Introduce new 'perf bench futex' benchmark: 'wake-parallel', to
measure parallel waker threads generating contention for kernel
locks (hb->lock). (Davidlohr Bueso)
'perf top', 'perf report':
- Allow disabling/enabling events dynamicaly in 'perf top':
a 'perf top' session can instantly become a 'perf report'
one, i.e. going from dynamic analysis to a static one,
returning to a dynamic one is possible, to toogle the
modes, just press 'f' to 'freeze/unfreeze' the sampling. (Arnaldo Carvalho de Melo)
- Make Ctrl-C stop processing on TUI, allowing interrupting the load of big
perf.data files (Namhyung Kim)
'perf probe': (Masami Hiramatsu)
- Support glob wildcards for function name
- Support $params special probe argument: Collect all function arguments
- Make --line checks validate C-style function name.
- Add --no-inlines option to avoid searching inline functions
- Greatly speed up 'perf probe --list' by caching debuginfo.
- Improve --filter support for 'perf probe', allowing using its arguments
on other commands, as --add, --del, etc.
'perf sched':
- Add option in 'perf sched' to merge like comms to lat output (Josef Bacik)
Plus tons of infrastructure work - in particular preparation for
upcoming threaded perf report support, but also lots of other work -
and fixes and other improvements. See (much) more details in the
shortlog and in the git log"
* 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (305 commits)
perf tools: Configurable per thread proc map processing time out
perf tools: Add time out to force stop proc map processing
perf report: Fix sort__sym_cmp to also compare end of symbol
perf hists browser: React to unassigned hotkey pressing
perf top: Tell the user how to unfreeze events after pressing 'f'
perf hists browser: Honour the help line provided by builtin-{top,report}.c
perf hists browser: Do not exit when 'f' is pressed in 'report' mode
perf top: Replace CTRL+z with 'f' as hotkey for enable/disable events
perf annotate: Rename source_line_percent to source_line_samples
perf annotate: Display total number of samples with --show-total-period
perf tools: Ensure thread-stack is flushed
perf top: Allow disabling/enabling events dynamicly
perf evlist: Add toggle_enable() method
perf trace: Fix race condition at the end of started workloads
perf probe: Speed up perf probe --list by caching debuginfo
perf probe: Show usage even if the last event is skipped
perf tools: Move libtraceevent dynamic list to separated LDFLAGS variable
perf tools: Fix a problem when opening old perf.data with different byte order
perf tools: Ignore .config-detected in .gitignore
perf probe: Fix to return error if no probe is added
...
Diffstat (limited to 'include')
-rw-r--r-- | include/linux/perf_event.h | 42 | ||||
-rw-r--r-- | include/linux/sched.h | 7 | ||||
-rw-r--r-- | include/uapi/linux/perf_event.h | 18 |
3 files changed, 63 insertions, 4 deletions
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index d8a82a89f35a..a204d5266f5f 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -120,7 +120,7 @@ struct hw_perf_event { }; struct { /* intel_cqm */ int cqm_state; - int cqm_rmid; + u32 cqm_rmid; struct list_head cqm_events_entry; struct list_head cqm_groups_entry; struct list_head cqm_group_entry; @@ -730,6 +730,22 @@ extern int perf_event_overflow(struct perf_event *event, struct perf_sample_data *data, struct pt_regs *regs); +extern void perf_event_output(struct perf_event *event, + struct perf_sample_data *data, + struct pt_regs *regs); + +extern void +perf_event_header__init_id(struct perf_event_header *header, + struct perf_sample_data *data, + struct perf_event *event); +extern void +perf_event__output_id_sample(struct perf_event *event, + struct perf_output_handle *handle, + struct perf_sample_data *sample); + +extern void +perf_log_lost_samples(struct perf_event *event, u64 lost); + static inline bool is_sampling_event(struct perf_event *event) { return event->attr.sample_period != 0; @@ -794,11 +810,33 @@ perf_sw_event_sched(u32 event_id, u64 nr, u64 addr) extern struct static_key_deferred perf_sched_events; +static __always_inline bool +perf_sw_migrate_enabled(void) +{ + if (static_key_false(&perf_swevent_enabled[PERF_COUNT_SW_CPU_MIGRATIONS])) + return true; + return false; +} + +static inline void perf_event_task_migrate(struct task_struct *task) +{ + if (perf_sw_migrate_enabled()) + task->sched_migrated = 1; +} + static inline void perf_event_task_sched_in(struct task_struct *prev, struct task_struct *task) { if (static_key_false(&perf_sched_events.key)) __perf_event_task_sched_in(prev, task); + + if (perf_sw_migrate_enabled() && task->sched_migrated) { + struct pt_regs *regs = this_cpu_ptr(&__perf_regs[0]); + + perf_fetch_caller_regs(regs); + ___perf_sw_event(PERF_COUNT_SW_CPU_MIGRATIONS, 1, regs, 0); + task->sched_migrated = 0; + } } static inline void perf_event_task_sched_out(struct task_struct *prev, @@ -921,6 +959,8 @@ perf_aux_output_skip(struct perf_output_handle *handle, static inline void * perf_get_aux(struct perf_output_handle *handle) { return NULL; } static inline void +perf_event_task_migrate(struct task_struct *task) { } +static inline void perf_event_task_sched_in(struct task_struct *prev, struct task_struct *task) { } static inline void diff --git a/include/linux/sched.h b/include/linux/sched.h index 8ca95f6a9395..659f5729cacc 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1357,9 +1357,6 @@ struct task_struct { #endif struct mm_struct *mm, *active_mm; -#ifdef CONFIG_COMPAT_BRK - unsigned brk_randomized:1; -#endif /* per-thread vma caching */ u32 vmacache_seqnum; struct vm_area_struct *vmacache[VMACACHE_SIZE]; @@ -1382,10 +1379,14 @@ struct task_struct { /* Revert to default priority/policy when forking */ unsigned sched_reset_on_fork:1; unsigned sched_contributes_to_load:1; + unsigned sched_migrated:1; #ifdef CONFIG_MEMCG_KMEM unsigned memcg_kmem_skip_account:1; #endif +#ifdef CONFIG_COMPAT_BRK + unsigned brk_randomized:1; +#endif unsigned long atomic_flags; /* Flags needing atomic access. */ diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h index 309211b3eb67..d97f84c080da 100644 --- a/include/uapi/linux/perf_event.h +++ b/include/uapi/linux/perf_event.h @@ -167,6 +167,7 @@ enum perf_branch_sample_type_shift { PERF_SAMPLE_BRANCH_COND_SHIFT = 10, /* conditional branches */ PERF_SAMPLE_BRANCH_CALL_STACK_SHIFT = 11, /* call/ret stack */ + PERF_SAMPLE_BRANCH_IND_JUMP_SHIFT = 12, /* indirect jumps */ PERF_SAMPLE_BRANCH_MAX_SHIFT /* non-ABI */ }; @@ -186,6 +187,7 @@ enum perf_branch_sample_type { PERF_SAMPLE_BRANCH_COND = 1U << PERF_SAMPLE_BRANCH_COND_SHIFT, PERF_SAMPLE_BRANCH_CALL_STACK = 1U << PERF_SAMPLE_BRANCH_CALL_STACK_SHIFT, + PERF_SAMPLE_BRANCH_IND_JUMP = 1U << PERF_SAMPLE_BRANCH_IND_JUMP_SHIFT, PERF_SAMPLE_BRANCH_MAX = 1U << PERF_SAMPLE_BRANCH_MAX_SHIFT, }; @@ -564,6 +566,10 @@ struct perf_event_mmap_page { #define PERF_RECORD_MISC_GUEST_USER (5 << 0) /* + * Indicates that /proc/PID/maps parsing are truncated by time out. + */ +#define PERF_RECORD_MISC_PROC_MAP_PARSE_TIMEOUT (1 << 12) +/* * PERF_RECORD_MISC_MMAP_DATA and PERF_RECORD_MISC_COMM_EXEC are used on * different events so can reuse the same bit position. */ @@ -800,6 +806,18 @@ enum perf_event_type { */ PERF_RECORD_ITRACE_START = 12, + /* + * Records the dropped/lost sample number. + * + * struct { + * struct perf_event_header header; + * + * u64 lost; + * struct sample_id sample_id; + * }; + */ + PERF_RECORD_LOST_SAMPLES = 13, + PERF_RECORD_MAX, /* non-ABI */ }; |