moonjuice/kernel_arpi
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge tag 'perf-core-2020-08-03' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Browse Source 
Pull perf event updates from Ingo Molnar:
 "HW support updates:

   - Add uncore support for Intel Comet Lake

   - Add RAPL support for Hygon Fam18h

   - Add Intel "IIO stack to PMON mapping" support on Skylake-SP CPUs,
     which enumerates per device performance counters via sysfs and
     enables the perf stat --iiostat functionality

   - Add support for Intel "Architectural LBRs", which generalized the
     model specific LBR hardware tracing feature into a
     model-independent, architected performance monitoring feature.

     Usage is mostly seamless to tooling, as the pre-existing LBR
     features are kept, but there's a couple of advantages under the
     hood, such as faster context-switching, faster LBR reads, cleaner
     exposure of LBR features to guest kernels, etc.

     ( Since architectural LBRs are supported via XSAVE, there's related
       changes to the x86 FPU code as well. )

  ftrace/perf updates:

   - Add support to add a text poke event to record changes to kernel
     text (i.e. self-modifying code) in order to support tracers like
     Intel PT decoding through jump labels, kprobes and ftrace
     trampolines.

  Misc cleanups, smaller fixes..."

* tag 'perf-core-2020-08-03' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (47 commits)
  perf/x86/rapl: Add Hygon Fam18h RAPL support
  kprobes: Remove unnecessary module_mutex locking from kprobe_optimizer()
  x86/perf: Fix a typo
  perf: <linux/perf_event.h>: drop a duplicated word
  perf/x86/intel/lbr: Support XSAVES for arch LBR read
  perf/x86/intel/lbr: Support XSAVES/XRSTORS for LBR context switch
  x86/fpu/xstate: Add helpers for LBR dynamic supervisor feature
  x86/fpu/xstate: Support dynamic supervisor feature for LBR
  x86/fpu: Use proper mask to replace full instruction mask
  perf/x86: Remove task_ctx_size
  perf/x86/intel/lbr: Create kmem_cache for the LBR context data
  perf/core: Use kmem_cache to allocate the PMU specific data
  perf/core: Factor out functions to allocate/free the task_ctx_data
  perf/x86/intel/lbr: Support Architectural LBR
  perf/x86/intel/lbr: Factor out intel_pmu_store_lbr
  perf/x86/intel/lbr: Factor out rdlbr_all() and wrlbr_all()
  perf/x86/intel/lbr: Mark the {rd,wr}lbr_{to,from} wrappers __always_inline
  perf/x86/intel/lbr: Unify the stored format of LBR information
  perf/x86/intel/lbr: Support LBR_CTL
  perf/x86: Expose CPUID enumeration bits for arch LBR
  ...
This commit is contained in:
Linus Torvalds
2020-08-03 14:51:09 -07:00
parent 9dee86896c d903b6d029
commit b34133fec8
32 changed files with 1943 additions and 280 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
include/uapi/linux/perf_event.h
View File

@@ -383,7 +383,8 @@ struct perf_event_attr {
bpf_event : 1, /* include bpf events */
aux_output : 1, /* generate AUX records instead of events */
cgroup : 1, /* include cgroup events */
__reserved_1 : 31;
text_poke : 1, /* include text poke events */
__reserved_1 : 30;
union {
__u32 wakeup_events; /* wakeup every n events */
@@ -1041,12 +1042,35 @@ enum perf_event_type {
*/
PERF_RECORD_CGROUP = 19,
/*
* Records changes to kernel text i.e. self-modified code. 'old_len' is
* the number of old bytes, 'new_len' is the number of new bytes. Either
* 'old_len' or 'new_len' may be zero to indicate, for example, the
* addition or removal of a trampoline. 'bytes' contains the old bytes
* followed immediately by the new bytes.
*
* struct {
* struct perf_event_header header;
* u64 addr;
* u16 old_len;
* u16 new_len;
* u8 bytes[];
* struct sample_id sample_id;
* };
*/
PERF_RECORD_TEXT_POKE = 20,
PERF_RECORD_MAX, /* non-ABI */
};
enum perf_record_ksymbol_type {
PERF_RECORD_KSYMBOL_TYPE_UNKNOWN = 0,
PERF_RECORD_KSYMBOL_TYPE_BPF = 1,
/*
* Out of line code such as kprobe-replaced instructions or optimized
* kprobes or ftrace trampolines.
*/
PERF_RECORD_KSYMBOL_TYPE_OOL = 2,
PERF_RECORD_KSYMBOL_TYPE_MAX /* non-ABI */
};