- 07 Nov, 2019 20 commits
-
-
Jin Yao authored
We have already supported the '--total-cycles' option in previous patch. It's also useful to show entries only above a threshold percent. This patch enables '--percent-limit' for not showing entries under that percent. For example: perf report --total-cycles --stdio --percent-limit 1 # To display the perf.data header info, please use --header/--header-only options. # # # Total Lost Samples: 0 # # Samples: 2M of event 'cycles' # Event count (approx.): 2753248 # # Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object # ............... .............. ........... .......... ................................................................. .................... # 26.04% 2.8M 0.40% 18 [div.c:42 -> div.c:39] div 15.17% 1.2M 0.16% 7 [random_r.c:357 -> random_r.c:380] libc-2.27.so 5.11% 402.0K 0.04% 2 [div.c:27 -> div.c:28] div 4.87% 381.6K 0.04% 2 [random.c:288 -> random.c:291] libc-2.27.so 4.53% 381.0K 0.04% 2 [div.c:40 -> div.c:40] div 3.85% 300.9K 0.02% 1 [div.c:22 -> div.c:25] div 3.08% 241.1K 0.02% 1 [rand.c:26 -> rand.c:27] libc-2.27.so 3.06% 240.0K 0.02% 1 [random.c:291 -> random.c:291] libc-2.27.so 2.78% 215.7K 0.02% 1 [random.c:298 -> random.c:298] libc-2.27.so 2.52% 198.3K 0.02% 1 [random.c:293 -> random.c:293] libc-2.27.so 2.36% 184.8K 0.02% 1 [rand.c:28 -> rand.c:28] libc-2.27.so 2.33% 180.5K 0.02% 1 [random.c:295 -> random.c:295] libc-2.27.so 2.28% 176.7K 0.02% 1 [random.c:295 -> random.c:295] libc-2.27.so 2.20% 168.8K 0.02% 1 [rand@plt+0 -> rand@plt+0] div 1.98% 158.2K 0.02% 1 [random_r.c:388 -> random_r.c:388] libc-2.27.so 1.57% 123.3K 0.02% 1 [div.c:42 -> div.c:44] div 1.44% 116.0K 0.42% 19 [random_r.c:357 -> random_r.c:394] libc-2.27.so Committer testing: From second exapmple onwards slightly edited for brevity: # perf report --total-cycles --percent-limit 2 --stdio # To display the perf.data header info, please use --header/--header-only options. # # # Total Lost Samples: 0 # # Samples: 6M of event 'cycles' # Event count (approx.): 6299936 # # Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object # ............... .............. ........... .......... ...................................................................... .................... # 2.17% 1.7M 0.08% 607 [compiler.h:199 -> common.c:221] [kernel.vmlinux] # # (Tip: Create an archive with symtabs to analyse on other machine: perf archive) # # perf report --total-cycles --percent-limit 1 --stdio # Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object 2.17% 1.7M 0.08% 607 [compiler.h:199 -> common.c:221] [kernel.vmlinux] 1.75% 1.3M 8.34% 65.5K [memset-vec-unaligned-erms.S:147 -> memset-vec-unaligned-erms.S:151] libc-2.29.so # # perf report --total-cycles --percent-limit 0.7 --stdio # Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object 2.17% 1.7M 0.08% 607 [compiler.h:199 -> common.c:221] [kernel.vmlinux] 1.75% 1.3M 8.34% 65.5K [memset-vec-unaligned-erms.S:147 -> memset-vec-unaligned-erms.S:151] libc-2.29.so 0.72% 544.5K 0.03% 230 [entry_64.S:657 -> entry_64.S:662] [kernel.vmlinux] # ------------------------------------------- It only shows the entries which 'Sampled Cycles%' > 1%. v7: --- No functional change. Only fix the conflict issue because previous patches are changed. v6: --- No functional change. Only fix the conflict issue because previous patches are changed. v5: --- No functional change. Only fix the conflict issue because previous patches are changed. v4: --- No functional change. Only fix the build issue because previous patches are changed. Signed-off-by:Jin Yao <yao.jin@linux.intel.com> Reviewed-by:
Jiri Olsa <jolsa@kernel.org> Tested-by:
Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Jin Yao <yao.jin@intel.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lore.kernel.org/lkml/20191107074719.26139-7-yao.jin@linux.intel.comSigned-off-by:
-
Jin Yao authored
It would be useful to support sorting for all blocks by the sampled cycles percent per block. This is useful to concentrate on the globally hottest blocks. This patch implements a new option "--total-cycles" which sorts all blocks by 'Sampled Cycles%'. The 'Sampled Cycles%' is the percent: percent = block sampled cycles aggregation / total sampled cycles Note that, this patch only supports "--stdio" mode. For example, # perf record -b ./div # perf report --total-cycles --stdio # To display the perf.data header info, please use --header/--header-only options. # # Total Lost Samples: 0 # # Samples: 2M of event 'cycles' # Event count (approx.): 2753248 # # Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object # ............... .............. ........... .......... ................................................ ................. # 26.04% 2.8M 0.40% 18 [div.c:42 -> div.c:39] div 15.17% 1.2M 0.16% 7 [random_r.c:357 -> random_r.c:380] libc-2.27.so 5.11% 402.0K 0.04% 2 [div.c:27 -> div.c:28] div 4.87% 381.6K 0.04% 2 [random.c:288 -> random.c:291] libc-2.27.so 4.53% 381.0K 0.04% 2 [div.c:40 -> div.c:40] div 3.85% 300.9K 0.02% 1 [div.c:22 -> div.c:25] div 3.08% 241.1K 0.02% 1 [rand.c:26 -> rand.c:27] libc-2.27.so 3.06% 240.0K 0.02% 1 [random.c:291 -> random.c:291] libc-2.27.so 2.78% 215.7K 0.02% 1 [random.c:298 -> random.c:298] libc-2.27.so 2.52% 198.3K 0.02% 1 [random.c:293 -> random.c:293] libc-2.27.so 2.36% 184.8K 0.02% 1 [rand.c:28 -> rand.c:28] libc-2.27.so 2.33% 180.5K 0.02% 1 [random.c:295 -> random.c:295] libc-2.27.so 2.28% 176.7K 0.02% 1 [random.c:295 -> random.c:295] libc-2.27.so 2.20% 168.8K 0.02% 1 [rand@plt+0 -> rand@plt+0] div 1.98% 158.2K 0.02% 1 [random_r.c:388 -> random_r.c:388] libc-2.27.so 1.57% 123.3K 0.02% 1 [div.c:42 -> div.c:44] div 1.44% 116.0K 0.42% 19 [random_r.c:357 -> random_r.c:394] libc-2.27.so 0.25% 182.5K 0.02% 1 [random_r.c:388 -> random_r.c:391] libc-2.27.so 0.00% 48 1.07% 48 [x86_pmu_enable+284 -> x86_pmu_enable+298] [kernel.kallsyms] 0.00% 74 1.64% 74 [vm_mmap_pgoff+0 -> vm_mmap_pgoff+92] [kernel.kallsyms] 0.00% 73 1.62% 73 [vm_mmap+0 -> vm_mmap+48] [kernel.kallsyms] 0.00% 63 0.69% 31 [up_write+0 -> up_write+34] [kernel.kallsyms] 0.00% 13 0.29% 13 [setup_arg_pages+396 -> setup_arg_pages+413] [kernel.kallsyms] 0.00% 3 0.07% 3 [setup_arg_pages+418 -> setup_arg_pages+450] [kernel.kallsyms] 0.00% 616 6.84% 308 [security_mmap_file+0 -> security_mmap_file+72] [kernel.kallsyms] 0.00% 23 0.51% 23 [security_mmap_file+77 -> security_mmap_file+87] [kernel.kallsyms] 0.00% 4 0.02% 1 [sched_clock+0 -> sched_clock+4] [kernel.kallsyms] 0.00% 4 0.02% 1 [sched_clock+9 -> sched_clock+12] [kernel.kallsyms] 0.00% 1 0.02% 1 [rcu_nmi_exit+0 -> rcu_nmi_exit+9] [kernel.kallsyms] Committer testing: This should provide material for hours of endless joy, both from looking for suspicious things in the implementation of this patch, such as the top one: # Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object 2.17% 1.7M 0.08% 607 [compiler.h:199 -> common.c:221] [kernel.vmlinux] As well from things that look legit: # Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object 0.16% 123.0K 0.60% 4.7K [nospec-branch.h:265 -> nospec-branch.h:278] [kernel.vmlinux] :-) Very short system wide taken branches session: # perf record -h -b Usage: perf record [<options>] [<command>] or: perf record [<options>] -- <command> [<options>] -b, --branch-any sample any taken branches # # perf record -b ^C[ perf record: Woken up 595 times to write data ] [ perf record: Captured and wrote 156.672 MB perf.data (196873 samples) ] # # perf evlist -v cycles: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CPU|PERIOD|BRANCH_STACK, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, task: 1, precise_ip: 3, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1, branch_sample_type: ANY # # perf report --total-cycles --stdio # To display the perf.data header info, please use --header/--header-only options. # # Total Lost Samples: 0 # # Samples: 6M of event 'cycles' # Event count (approx.): 6299936 # # Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object # ............... .............. ........... .......... ...................................................................... .................... # 2.17% 1.7M 0.08% 607 [compiler.h:199 -> common.c:221] [kernel.vmlinux] 1.75% 1.3M 8.34% 65.5K [memset-vec-unaligned-erms.S:147 -> memset-vec-unaligned-erms.S:151] libc-2.29.so 0.72% 544.5K 0.03% 230 [entry_64.S:657 -> entry_64.S:662] [kernel.vmlinux] 0.56% 541.8K 0.09% 672 [compiler.h:199 -> common.c:300] [kernel.vmlinux] 0.39% 293.2K 0.01% 104 [list_debug.c:43 -> list_debug.c:61] [kernel.vmlinux] 0.36% 278.6K 0.03% 272 [entry_64.S:1289 -> entry_64.S:1308] [kernel.vmlinux] 0.30% 260.8K 0.07% 564 [clear_page_64.S:47 -> clear_page_64.S:50] [kernel.vmlinux] 0.28% 215.3K 0.05% 369 [traps.c:623 -> traps.c:628] [kernel.vmlinux] 0.23% 178.1K 0.04% 278 [entry_64.S:271 -> entry_64.S:275] [kernel.vmlinux] 0.20% 152.6K 0.09% 706 [paravirt.c:177 -> paravirt.c:179] [kernel.vmlinux] 0.20% 155.8K 0.05% 373 [entry_64.S:153 -> entry_64.S:175] [kernel.vmlinux] 0.18% 136.6K 0.03% 222 [msr.h:105 -> msr.h:166] [kernel.vmlinux] 0.16% 123.0K 0.60% 4.7K [nospec-branch.h:265 -> nospec-branch.h:278] [kernel.vmlinux] 0.16% 118.3K 0.01% 44 [entry_64.S:632 -> entry_64.S:657] [kernel.vmlinux] 0.14% 104.5K 0.00% 28 [rwsem.c:1541 -> rwsem.c:1544] [kernel.vmlinux] 0.13% 99.2K 0.01% 53 [spinlock.c:150 -> spinlock.c:152] [kernel.vmlinux] 0.13% 95.5K 0.00% 35 [swap.c:456 -> swap.c:471] [kernel.vmlinux] 0.12% 96.2K 0.05% 407 [copy_user_64.S:175 -> copy_user_64.S:209] [kernel.vmlinux] 0.11% 85.9K 0.00% 31 [swap.c:400 -> page-flags.h:188] [kernel.vmlinux] 0.10% 73.0K 0.01% 52 [paravirt.h:763 -> list.h:131] [kernel.vmlinux] 0.07% 56.2K 0.03% 214 [filemap.c:1524 -> filemap.c:1557] [kernel.vmlinux] 0.07% 54.2K 0.02% 145 [memory.c:1032 -> memory.c:1049] [kernel.vmlinux] 0.07% 50.3K 0.00% 39 [mmzone.c:49 -> mmzone.c:69] [kernel.vmlinux] 0.06% 48.3K 0.01% 40 [paravirt.h:768 -> page_alloc.c:3304] [kernel.vmlinux] 0.06% 46.7K 0.02% 155 [memory.c:1032 -> memory.c:1056] [kernel.vmlinux] 0.06% 46.9K 0.01% 103 [swap.c:867 -> swap.c:902] [kernel.vmlinux] 0.06% 47.8K 0.00% 34 [entry_64.S:1201 -> entry_64.S:1202] [kernel.vmlinux] ----------------------------------------------------------- v7: --- Use use_browser in report__browse_block_hists for supporting stdio and potential tui mode. v6: --- Create report__browse_block_hists in block-info.c (codes are moved from builtin-report.c). It's called from perf_evlist__tty_browse_hists. v5: --- 1. Move all block functions to block-info.c 2. Move the code of setting ms in block hist_entry to other patch. v4: --- 1. Use new option '--total-cycles' to replace '-s total_cycles' in v3. 2. Move block info collection out of block info printing. v3: --- 1. Use common function block_info__process_sym to process the blocks per symbol. 2. Remove the nasty hack for skipping calculation of column length 3. Some minor cleanup Signed-off-by: -
Jin Yao authored
This patch provides helper routines to support new columns for block info output. The new columns are: Sampled Cycles% Sampled Cycles Avg Cycles% Avg Cycles [Program Block Range] Shared Object v5: --- 1. Move more block related functions from builtin-report.c to block-info.c 2. Set ms (map+sym) in block hist_entry. Because this info is needed for reporting the block range (i.e. source line) Committer notes: Remove unused set_fmt() function, some build were not completing with: util/block-info.c:396:20: error: unused function 'set_fmt' [-Werror,-Wunused-function] static inline void set_fmt(struct block_fmt *block_fmt, ^ 1 error generated. Signed-off-by: -
Jin Yao authored
We can get the per sample cycles by hist__account_cycles(). It's also useful to know the total cycles of all samples in order to get the cycles coverage for a single program block in further. For example: coverage = per block sampled cycles / total sampled cycles This patch creates a new argument 'total_cycles' in hist__account_cycles(), which will be added with the cycles of each sample. Signed-off-by:
-
Jin Yao authored
We have already implemented some block-info related functions. Now it's time to do some cleanup, refactoring and move the functions and structures to new block-info.h/block-info.c. v4: --- Move code for skipping column length calculation to patch: 'perf diff: Don't use hack to skip column length calculation' v3: --- 1. Rename the patch title 2. Rename from block.h/block.c to block-info.h/block-info.c 3. Move more common part to block-info, such as block_info__process_sym. 4. Remove the nasty hack for skipping calculation of column length Signed-off-by: -
Jin Yao authored
Previously we use a nasty hack to skip the hists__calc_col_len for block since this function is not very suitable for block column length calculation. This patch removes the hack code and add a check at the entry of hists__calc_col_len to skip for block case. Signed-off-by:
-
Leo Yan authored
The test case 'Read backward ring buffer' failed on 32-bit architectures which were found by LKFT perf testing. The test failed on arm32 x15 device, qemu_arm32, qemu_i386, and found intermittent failure on i386; the failure log is as below: 50: Read backward ring buffer : --- start --- test child forked, pid 510 Using CPUID GenuineIntel-6-9E-9 mmap size 1052672B mmap size 8192B Finished reading overwrite ring buffer: rewind free(): invalid next size (fast) test child interrupted ---- end ---- Read backward ring buffer: FAILED! The log hints there have issue for memory usage, thus free() reports error 'invalid next size' and directly exit for the case. Finally, this issue is root caused as out of bounds memory access for the data array 'evsel->id'. The backward ring buffer test invokes do_test() twice. 'evsel->id' is allocated at the first call with the flow: test__backward_ring_buffer() `-> do_test() `-> evlist__mmap() `-> evlist__mmap_ex() `-> perf_evsel__alloc_id() So 'evsel->id' is allocated with one item, and it will be used in function perf_evlist__id_add(): evsel->id[0] = id evsel->ids = 1 At the second call for do_test(), it skips to initialize 'evsel->id' and reuses the array which is allocated in the first call. But 'evsel->ids' contains the stale value. Thus: evsel->id[1] = id -> out of bound access evsel->ids = 2 To fix this issue, we will use evlist__open() and evlist__close() pair functions to prepare and cleanup context for evlist; so 'evsel->id' and 'evsel->ids' can be initialized properly when invoke do_test() and avoid the out of bounds memory access. Fixes: ee74701e ("perf tests: Add test to check backward ring buffer") Signed-off-by:Leo Yan <leo.yan@linaro.org> Reviewed-by:
-
Jiwei Sun authored
The patch adds a new option to limit the output file size, then based on it, we can create a wrapper of the perf command that uses the option to avoid exhausting the disk space by the unconscious user. In order to make the perf.data parsable, we just limit the sample data size, since the perf.data consists of many headers and sample data and other data, the actual size of the recorded file will bigger than the setting value. Testing it: # ./perf record -a -g --max-size=10M Couldn't synthesize bpf events. [ perf record: perf size limit reached (10249 KB), stopping session ] [ perf record: Woken up 32 times to write data ] [ perf record: Captured and wrote 10.133 MB perf.data (71964 samples) ] # ls -lh perf.data -rw------- 1 root root 11M Oct 22 14:32 perf.data # ./perf record -a -g --max-size=10K [ perf record: perf size limit reached (10 KB), stopping session ] Couldn't synthesize bpf events. [ perf record: Woken up 0 times to write data ] [ perf record: Captured and wrote 1.546 MB perf.data (69 samples) ] # ls -l perf.data -rw------- 1 root root 1626952 Oct 22 14:36 perf.data Committer notes: Fixed the build in multiple distros by using PRIu64 to print u64 struct members, fixing this: builtin-record.c: In function 'record__write': builtin-record.c:150:5: error: format '%lu' expects argument of type 'long unsigned int', but argument 3 has type 'u64' [-Werror=format=] rec->bytes_written >> 10); ^ CC /tmp/build/pe Signed-off-by:Jiwei Sun <jiwei.sun@windriver.com> Acked-by:
-
Masami Hiramatsu authored
Since debuginfo__find_probes() callback function can be called with the location which already passed, the callback function must filter out such overlapped locations. add_probe_trace_event() has already done it by commit 1a375ae7 ("perf probe: Skip same probe address for a given line"), but add_available_vars() doesn't. Thus perf probe -v shows same address repeatedly as below: # perf probe -V vfs_read:18 Available variables at vfs_read:18 @<vfs_read+217> char* buf loff_t* pos ssize_t ret struct file* file @<vfs_read+217> char* buf loff_t* pos ssize_t ret struct file* file @<vfs_read+226> char* buf loff_t* pos ssize_t ret struct file* file With this fix, perf probe -V shows it correctly: # perf probe -V vfs_read:18 Available variables at vfs_read:18 @<vfs_read+217> char* buf loff_t* pos ssize_t ret struct file* file @<vfs_read+226> char* buf loff_t* pos ssize_t ret struct file* file Fixes: cf6eb489 ("perf probe: Show accessible local variables") Signed-off-by:
Masami Hiramatsu <mhiramat@kernel.org> Tested-by:
-
Masami Hiramatsu authored
Fix to show calling lines of inlined functions (where an inline function is called). die_walk_lines() filtered out the lines inside inlined functions based on the address. However this also filtered out the lines which call those inlined functions from the target function. To solve this issue, check the call_file and call_line attributes and do not filter out if it matches to the line information. Without this fix, perf probe -L doesn't show some lines correctly. (don't see the lines after 17) # perf probe -L vfs_read <vfs_read@/home/mhiramat/ksrc/linux/fs/read_write.c:0> 0 ssize_t vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos) 1 { 2 ssize_t ret; 4 if (!(file->f_mode & FMODE_READ)) return -EBADF; 6 if (!(file->f_mode & FMODE_CAN_READ)) return -EINVAL; 8 if (unlikely(!access_ok(buf, count))) return -EFAULT; 11 ret = rw_verify_area(READ, file, pos, count); 12 if (!ret) { 13 if (count > MAX_RW_COUNT) count = MAX_RW_COUNT; 15 ret = __vfs_read(file, buf, count, pos); 16 if (ret > 0) { fsnotify_access(file); add_rchar(current, ret); } With this fix: # perf probe -L vfs_read <vfs_read@/home/mhiramat/ksrc/linux/fs/read_write.c:0> 0 ssize_t vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos) 1 { 2 ssize_t ret; 4 if (!(file->f_mode & FMODE_READ)) return -EBADF; 6 if (!(file->f_mode & FMODE_CAN_READ)) return -EINVAL; 8 if (unlikely(!access_ok(buf, count))) return -EFAULT; 11 ret = rw_verify_area(READ, file, pos, count); 12 if (!ret) { 13 if (count > MAX_RW_COUNT) count = MAX_RW_COUNT; 15 ret = __vfs_read(file, buf, count, pos); 16 if (ret > 0) { 17 fsnotify_access(file); 18 add_rchar(current, ret); } 20 inc_syscr(current); } Fixes: 4cc9cec6 ("perf probe: Introduce lines walker interface") Signed-off-by: -
Masami Hiramatsu authored
Filter out instances except for inlined_subroutine and subprogram DIE in die_walk_instances() and die_is_func_instance(). This fixes an issue that perf probe sets some probes on calling address instead of a target function itself. When perf probe walks on instances of an abstruct origin (a kind of function prototype of inlined function), die_walk_instances() can also pass a GNU_call_site (a GNU extension for call site) to callback. Since it is not an inlined instance of target function, we have to filter out when searching a probe point. Without this patch, perf probe sets probes on call site address too.This can happen on some function which is marked "inlined", but has actual symbol. (I'm not sure why GCC mark it "inlined"): # perf probe -D vfs_read p:probe/vfs_read _text+2500017 p:probe/vfs_read_1 _text+2499468 p:probe/vfs_read_2 _text+2499563 p:probe/vfs_read_3 _text+2498876 p:probe/vfs_read_4 _text+2498512 p:probe/vfs_read_5 _text+2498627 With this patch: Slightly different results, similar tho: # perf probe -D vfs_read p:probe/vfs_read _text+2498512 Committer testing: # uname -a Linux quaco 5.3.8-200.fc30.x86_64 #1 SMP Tue Oct 29 14:46:22 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux Before: # perf probe -D vfs_read p:probe/vfs_read _text+3131557 p:probe/vfs_read_1 _text+3130975 p:probe/vfs_read_2 _text+3131047 p:probe/vfs_read_3 _text+3130380 p:probe/vfs_read_4 _text+3130000 # uname -a Linux quaco 5.3.8-200.fc30.x86_64 #1 SMP Tue Oct 29 14:46:22 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux # After: # perf probe -D vfs_read p:probe/vfs_read _text+3130000 # Fixes: db0d2c64 ("perf probe: Search concrete out-of-line instances") Signed-off-by:
-
Masami Hiramatsu authored
Skip end-of-sequence and non-statement lines while walking through lines list. The "end-of-sequence" line information means: "the current address is that of the first byte after the end of a sequence of target machine instructions." (DWARF version 4 spec 6.2.2) This actually means out of scope and we can not probe on it. On the other hand, the statement lines (is_stmt) means: "the current instruction is a recommended breakpoint location. A recommended breakpoint location is intended to “represent” a line, a statement and/or a semantically distinct subpart of a statement." (DWARF version 4 spec 6.2.2) So, non-statement line info also should be skipped. These can reduce unneeded probe points and also avoid an error. E.g. without this patch: # perf probe -a "clear_tasks_mm_cpumask:1" Added new events: probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask:1) probe:clear_tasks_mm_cpumask_1 (on clear_tasks_mm_cpumask:1) probe:clear_tasks_mm_cpumask_2 (on clear_tasks_mm_cpumask:1) probe:clear_tasks_mm_cpumask_3 (on clear_tasks_mm_cpumask:1) probe:clear_tasks_mm_cpumask_4 (on clear_tasks_mm_cpumask:1) You can now use it in all perf tools, such as: perf record -e probe:clear_tasks_mm_cpumask_4 -aR sleep 1 # This puts 5 probes on one line, but acutally it's not inlined function. This is because there are many non statement instructions at the function prologue. With this patch: # perf probe -a "clear_tasks_mm_cpumask:1" Added new event: probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask:1) You can now use it in all perf tools, such as: perf record -e probe:clear_tasks_mm_cpumask -aR sleep 1 # Now perf-probe skips unneeded addresses. Committer testing: Slightly different results, but similar: Before: # uname -a Linux quaco 5.3.8-200.fc30.x86_64 #1 SMP Tue Oct 29 14:46:22 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux # # perf probe -a "clear_tasks_mm_cpumask:1" Added new events: probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask:1) probe:clear_tasks_mm_cpumask_1 (on clear_tasks_mm_cpumask:1) probe:clear_tasks_mm_cpumask_2 (on clear_tasks_mm_cpumask:1) You can now use it in all perf tools, such as: perf record -e probe:clear_tasks_mm_cpumask_2 -aR sleep 1 # After: # perf probe -a "clear_tasks_mm_cpumask:1" Added new event: probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask:1) You can now use it in all perf tools, such as: perf record -e probe:clear_tasks_mm_cpumask -aR sleep 1 # perf probe -l probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask@kernel/cpu.c) # Fixes: 4cc9cec6 ("perf probe: Introduce lines walker interface") Signed-off-by: -
Masami Hiramatsu authored
Make find_best_scope() returns innermost DIE at given address if there is no best matched scope DIE. Since Gcc sometimes generates intuitively strange line info which is out of inlined function address range, we need this fixup. Without this, sometimes perf probe failed to probe on a line inside an inlined function: # perf probe -D ksys_open:3 Failed to find scope of probe point. Error: Failed to add events. With this fix, 'perf probe' can probe it: # perf probe -D ksys_open:3 p:probe/ksys_open _text+25707308 p:probe/ksys_open_1 _text+25710596 p:probe/ksys_open_2 _text+25711114 p:probe/ksys_open_3 _text+25711343 p:probe/ksys_open_4 _text+25714058 p:probe/ksys_open_5 _text+2819653 p:probe/ksys_open_6 _text+2819701 Signed-off-by: -
Ian Rogers authored
Fix expand_tabs that copies the source lines '\0' and then appends another '\0' at a potentially out of bounds address. Signed-off-by:
Ian Rogers <irogers@google.com> Acked-by:
-
Arnaldo Carvalho de Melo authored
To reduce boilerplate in some places. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-9s1bgoxxhlnu037e1nqx0tw3@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Its sufficient to check if map->groups is NULL before using it to get ->machine value. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-utiepyiv8b1tf8f79ok9d6j8@git.kernel.orgSigned-off-by:
-
Ian Rogers authored
Add a parse_events_term deep delete function so that owned strings and arrays are freed. Signed-off-by:
-
Ian Rogers authored
Avoid a memory leak when the configuration fails. Signed-off-by:
-
Ian Rogers authored
Yyabort doesn't destruct inputs and so this must be done manually before using yyabort. Signed-off-by:
-
Ian Rogers authored
If parsing fails then destructors are ran to clean the up the stack. Rename the head union member to make the term and evlist use cases more distinct, this simplifies matching the correct destructor. Committer notes: Jiri: "Nice did not know about this.. looks like it's been in bison for some time, right?" Ian: "Looks like it wasn't in Bison 1 but in Bison 2, we're at Bison 3 and Bison 2 is > 14 years old: https://web.archive.org/web/20050924004158/http://www.gnu.org/software/bison/manual/html_mono/bison.html#Destructor-Decl" Signed-off-by:
-
- 06 Nov, 2019 20 commits
-
-
Ian Rogers authored
Make it easier to release memory associated with parse event terms by duplicating the string for the config name and ensuring the val string is a duplicate. Currently the parser may memory leak terms and this is addressed in a later patch. Signed-off-by:
-
Ian Rogers authored
Parse event error handling may overwrite one error string with another creating memory leaks. Introduce a helper routine that warns about multiple error messages as well as avoiding the memory leak. A reproduction of this problem can be seen with: perf stat -e c/c/ After this change this produces: WARNING: multiple event parsing errors event syntax error: 'c/c/' \___ unknown term valid terms: event,filter_rem,filter_opc0,edge,filter_isoc,filter_tid,filter_loc,filter_nc,inv,umask,filter_opc1,tid_en,thresh,filter_all_op,filter_not_nm,filter_state,filter_nm,config,config1,config2,name,period,percore Run 'perf list' for a list of valid events Usage: perf stat [<options>] [<command>] -e, --event <event> event selector. use 'perf list' to list available events Signed-off-by: -
Adrian Hunter authored
create_gcov (refer to the autofdo example in tools/perf/Documentation/intel-pt.txt) now needs the evsels to read the perf.data file. So don't strip them. Signed-off-by:
Adrian Hunter <adrian.hunter@intel.com> Acked-by:
-
John Garry authored
Currently when cross compiling perf tool for ARM64 on my x86 machine I get this error: arch/arm64/util/sym-handling.c:9:10: fatal error: gelf.h: No such file or directory #include <gelf.h> For the build, libelf is reported off: Auto-detecting system features: ... ... libelf: [ OFF ] Indeed, test-libelf is not built successfully: more ./build/feature/test-libelf.make.output test-libelf.c:2:10: fatal error: libelf.h: No such file or directory #include <libelf.h> ^~~~~~~~~~ compilation terminated. I have no such problems natively compiling on ARM64, and I did not previously have this issue for cross compiling. Fix by relocating the gelf.h include. Signed-off-by:John Garry <john.garry@huawei.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Will Deacon <will@kernel.org> Cc: linux-arm-kernel@lists.infradead.org Link: http://lore.kernel.org/lkml/1573045254-39833-1-git-send-email-john.garry@huawei.comSigned-off-by:
-
Jiri Olsa authored
Adding new --per-node option to aggregate counts per NUMA nodes for system-wide mode measurements. You can specify --per-node in live mode: # perf stat -a -I 1000 -e cycles --per-node # time node cpus counts unit events 1.000542550 N0 20 6,202,097 cycles 1.000542550 N1 20 639,559 cycles 2.002040063 N0 20 7,412,495 cycles 2.002040063 N1 20 2,185,577 cycles 3.003451699 N0 20 6,508,917 cycles 3.003451699 N1 20 765,607 cycles ... Or in the record/report stat session: # perf stat record -a -I 1000 -e cycles # time counts unit events 1.000536937 10,008,468 cycles 2.002090152 9,578,539 cycles 3.003625233 7,647,869 cycles 4.005135036 7,032,086 cycles ^C 4.340902364 3,923,893 cycles # perf stat report --per-node # time node cpus counts unit events 1.000536937 N0 20 9,355,086 cycles 1.000536937 N1 20 653,382 cycles 2.002090152 N0 20 7,712,838 cycles 2.002090152 N1 20 1,865,701 cycles 3.003625233 N0 20 6,604,441 cycles 3.003625233 N1 20 1,043,428 cycles 4.005135036 N0 20 6,350,522 cycles 4.005135036 N1 20 681,564 cycles 4.340902364 N0 20 3,403,188 cycles 4.340902364 N1 20 520,705 cycles Signed-off-by: -
Jiri Olsa authored
To speed up cpu to node lookup, add perf_env__numa_node(), that creates cpu array on the first lookup, that holds numa nodes for each stored cpu. Signed-off-by:
-
Haiyan Song authored
New Metrics: - DSB_Switches: fraction of cycles CPU was stalled due to switches from DSB to MITE pipeline [all] - L2_Evictions_{Silent|NonSilent}_PKI: L2 {silent|non silent} ecivtions rate per Kilo instruction [SKX+] - IpFarBranch - Instructions per Far Branch Other Enhancements & fixes: - KBLR/CFL & CLX move to separate columns (no column sharing via if #model) - Re-organized/renamed Metric Group Signed-off-by:Haiyan Song <haiyanx.song@intel.com> Reviewed-by:
Kan Liang <kan.liang@linux.intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Jin Yao <yao.jin@intel.com> Link: http://lore.kernel.org/lkml/20191030082308.10919-1-haiyanx.song@intel.comSigned-off-by:
-
Haiyan Song authored
Update CascadelakeX events to v1.05. Other changes: remove duplicated and without description events. Signed-off-by:
-
Ian Rogers authored
If event parsing fails the event list is leaked, instead splice the list onto the out result and let the caller cleanup. An example input for parse_events found by libFuzzer that reproduces this memory leak is 'm{'. Signed-off-by: -
James Clark authored
When a 'make DEBUG=1' build is done, the command parser is still built with -O6 and is hard to step through, fix it making it use -O0 in that case. Signed-off-by:
James Clark <james.clark@arm.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: nd <nd@arm.com> Link: http://lore.kernel.org/lkml/20191028113340.4282-1-james.clark@arm.com [ split from a larger patch ] Signed-off-by:
-
James Clark authored
Move EXTRA_WARNINGS and EXTRA_FLAGS to the end of the compilation line, otherwise they cannot be used to override the default values. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
To reduce boilerplate, providing a more compact form to iterate over the maps in a map_group. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-gc3go6fmdn30twusg91t2q56@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
To reduce boilerplate, provide a more compact form using an idiom present in other trees of data structures. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-59gmq4kg1r68ou1wknyjl78x@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Just like free(), return NULL in that case, will simplify the for_each_entry_safe() iterators. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-pbde2ucn49khnrebclys9pny@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
We were checking just if it was still on some rb tree, but that is not the only way that this map can still have references, map->refcnt is there exactly for this, use it. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-hany65tbeavsax7n3xvwl9pc@git.kernel.orgSigned-off-by:
-
Adrian Hunter authored
Add functions to write into the dso file data cache, but not change the file itself. Signed-off-by:
-
Adrian Hunter authored
Refactor dso_cache__read() to separate populating the cache from copying data from it. This is preparation for adding a cache "write" that will update the data in the cache. Signed-off-by:
-
Adrian Hunter authored
Add auxtrace_cache__remove(). Intel PT uses an auxtrace_cache to store the results of code-walking, so that the same block of instructions does not have to be decoded repeatedly. However, when there are text poke events, the associated cache entries need to be removed. Signed-off-by:
-
Masami Hiramatsu authored
Fix to show ranges of variables (--range and --vars option) in functions which DIE has only ranges but no entry_pc attribute. Without this fix: # perf probe --range -V clear_tasks_mm_cpumask Available variables at clear_tasks_mm_cpumask @<clear_tasks_mm_cpumask+0> (No matched variables) With this fix: # perf probe --range -V clear_tasks_mm_cpumask Available variables at clear_tasks_mm_cpumask @<clear_tasks_mm_cpumask+0> [VAL] int cpu @<clear_tasks_mm_cpumask+[0-35,317-317,2052-2059]> Committer testing: Before: [root@quaco ~]# perf probe --range -V clear_tasks_mm_cpumask Available variables at clear_tasks_mm_cpumask @<clear_tasks_mm_cpumask+0> (No matched variables) [root@quaco ~]# After: [root@quaco ~]# perf probe --range -V clear_tasks_mm_cpumask Available variables at clear_tasks_mm_cpumask @<clear_tasks_mm_cpumask+0> [VAL] int cpu @<clear_tasks_mm_cpumask+[0-23,23-105,105-106,106-106,1843-1850,1850-1862]> [root@quaco ~]# Using it: [root@quaco ~]# perf probe clear_tasks_mm_cpumask cpu Added new event: probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask with cpu) You can now use it in all perf tools, such as: perf record -e probe:clear_tasks_mm_cpumask -aR sleep 1 [root@quaco ~]# perf probe -l probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask@kernel/cpu.c with cpu) [root@quaco ~]# [root@quaco ~]# perf trace -e probe:*cpumask ^C[root@quaco ~]# Fixes: 349e8d26 ("perf probe: Add --range option to show a variable's location range") Signed-off-by: -
Masami Hiramatsu authored
Fix 'perf probe --line' option to show inlined function callsite lines even if the function DIE has only ranges. Without this: # perf probe -L amd_put_event_constraints ... 2 { 3 if (amd_has_nb(cpuc) && amd_is_nb_event(&event->hw)) __amd_put_nb_event_constraints(cpuc, event); 5 } With this patch: # perf probe -L amd_put_event_constraints ... 2 { 3 if (amd_has_nb(cpuc) && amd_is_nb_event(&event->hw)) 4 __amd_put_nb_event_constraints(cpuc, event); 5 } Committer testing: Before: [root@quaco ~]# perf probe -L amd_put_event_constraints <amd_put_event_constraints@/usr/src/debug/kernel-5.2.fc30/linux-5.2.18-200.fc30.x86_64/arch/x86/events/amd/core.c:0> 0 static void amd_put_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event) 2 { 3 if (amd_has_nb(cpuc) && amd_is_nb_event(&event->hw)) __amd_put_nb_event_constraints(cpuc, event); 5 } PMU_FORMAT_ATTR(event, "config:0-7,32-35"); PMU_FORMAT_ATTR(umask, "config:8-15" ); [root@quaco ~]# After: [root@quaco ~]# perf probe -L amd_put_event_constraints <amd_put_event_constraints@/usr/src/debug/kernel-5.2.fc30/linux-5.2.18-200.fc30.x86_64/arch/x86/events/amd/core.c:0> 0 static void amd_put_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event) 2 { 3 if (amd_has_nb(cpuc) && amd_is_nb_event(&event->hw)) 4 __amd_put_nb_event_constraints(cpuc, event); 5 } PMU_FORMAT_ATTR(event, "config:0-7,32-35"); PMU_FORMAT_ATTR(umask, "config:8-15" ); [root@quaco ~]# perf probe amd_put_event_constraints:4 Added new event: probe:amd_put_event_constraints (on amd_put_event_constraints:4) You can now use it in all perf tools, such as: perf record -e probe:amd_put_event_constraints -aR sleep 1 [root@quaco ~]# [root@quaco ~]# perf probe -l probe:amd_put_event_constraints (on amd_put_event_constraints:4@arch/x86/events/amd/core.c) probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask@kernel/cpu.c) [root@quaco ~]# Using it: [root@quaco ~]# perf trace -e probe:* ^C[root@quaco ~]# Ok, Intel system here... :-) Fixes: 4cc9cec6 ("perf probe: Introduce lines walker interface") Signed-off-by:
-
