- 01 Apr, 2019 39 commits
-
-
Andi Kleen authored
Signed-off-by:
Andi Kleen <ak@linux.intel.com> Cc: Kan Liang <kan.liang@intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Link: https://lkml.kernel.org/r/20190315165219.GA21223@tassilo.jf.intel.comSigned-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Update all the Intel JSON metrics from Ahmad Yasin's TMAM 3.5 for Intel big core from Sandy Bridge to Cascade Lake. This has many improvements and new metircs - New TopDownL1_SMT group that provides a per SMT thread version of --topdown that does not require -a anymore. The drawback is increased multiplexing though since L1 TopDown does not fit into 4 generic counters anymore. - Added SMT aware versions of other metrics - Split SMT aware metrics into separate metrics to avoid unnecessary event collections - New metrics for better branch analysis: Estimated Branch_Mispredict_Costs, Instructions per taken Branch, Branch Instructions per Taken Branch, etc. - Instruction mix metrics: Instructions per load, Instructions per store, Instructions per Branch, Instructions per Call - New Cache metrics: Bandwidth to L1/L2/L3 caches. L1/L2/L3 misses per kilo instructions. memory level parallelism - New memory controller metrics: Normalized memory bandwidth in interval mode, Average memory latency, Average number of parallel read requests, - 3DXP persistent memory metrics for Cascade Lake: 3dxp read latency, 3dxp read/write bandwidth - Some other useful metrics like Instruction Level Parallelism, - Various other improvements. Not all metrics are available on all CPUs. Skylake has best coverage. Signed-off-by:
-
Alexey Budankov authored
Implement a --mmap-flush option that specifies minimal number of bytes that is extracted from mmaped kernel buffer to store into a trace. The default option value is 1 byte what means every time trace writing thread finds some new data in the mmaped buffer the data is extracted, possibly compressed and written to a trace. $ tools/perf/perf record --mmap-flush 1024 -e cycles -- matrix.gcc $ tools/perf/perf record --aio --mmap-flush 1K -e cycles -- matrix.gcc The option is independent from -z setting, doesn't vary with compression level and can serve two purposes. The first purpose is to increase the compression ratio of a trace data. Larger data chunks are compressed more effectively so the implemented option allows specifying data chunk size to compress. Also at some cases executing more write syscalls with smaller data size can take longer than executing less write syscalls with bigger data size due to syscall overhead so extracting bigger data chunks specified by the option value could additionally decrease runtime overhead. The second purpose is to avoid self monitoring live-lock issue in system wide (-a) profiling mode. Profiling in system wide mode with compression (-a -z) can additionally induce data into the kernel buffers along with the data from monitored processes. If performance data rate and volume from the monitored processes is high then trace streaming and compression activity in the tool is also high. High tool process activity can lead to subtle live-lock effect when compression of single new byte from some of mmaped kernel buffer leads to generation of the next single byte at some mmaped buffer. So perf tool process ends up in endless self monitoring. Implemented synch parameter is the mean to force data move independently from the specified flush threshold value. Despite the provided flush value the tool needs capability to unconditionally drain memory buffers, at least in the end of the collection. Committer testing: Running with the default value, i.e. as soon as there is something to read go on consuming, we first write the synthesized events, small chunks of about 128 bytes: # perf trace -m 2048 --call-graph dwarf -e write -- perf record <SNIP> 101.142 ( 0.004 ms): perf/25821 write(fd: 3</root/perf.data>, buf: 0x210db60, count: 120) = 120 __libc_write (/usr/lib64/libpthread-2.28.so) ion (/home/acme/bin/perf) record__write (inlined) process_synthesized_event (/home/acme/bin/perf) perf_tool__process_synth_event (inlined) perf_event__synthesize_mmap_events (/home/acme/bin/perf) Then we move to reading the mmap buffers consuming the events put there by the kernel perf infrastructure: 107.561 ( 0.005 ms): perf/25821 write(fd: 3</root/perf.data>, buf: 0x7f1befc02000, count: 336) = 336 __libc_write (/usr/lib64/libpthread-2.28.so) ion (/home/acme/bin/perf) record__write (inlined) record__pushfn (/home/acme/bin/perf) perf_mmap__push (/home/acme/bin/perf) record__mmap_read_evlist (inlined) record__mmap_read_all (inlined) __cmd_record (inlined) cmd_record (/home/acme/bin/perf) 12919.953 ( 0.136 ms): perf/25821 write(fd: 3</root/perf.data>, buf: 0x7f1befc83150, count: 184984) = 184984 <SNIP same backtrace as in the 107.561 timestamp> 12920.094 ( 0.155 ms): perf/25821 write(fd: 3</root/perf.data>, buf: 0x7f1befc02150, count: 261816) = 261816 <SNIP same backtrace as in the 107.561 timestamp> 12920.253 ( 0.093 ms): perf/25821 write(fd: 3</root/perf.data>, buf: 0x7f1befb81120, count: 170832) = 170832 <SNIP same backtrace as in the 107.561 timestamp> If we limit it to write only when more than 16MB are available for reading, it throttles that to a quarter of the --mmap-pages set for 'perf record', which by default get to 528384 bytes, found out using 'record -v': mmap flush: 132096 mmap size 528384B With that in place all the writes coming from record__mmap_read_evlist(), i.e. from the mmap buffers setup by the kernel perf infrastructure were at least 132096 bytes long. Trying with a bigger mmap size: perf trace -e write perf record -v -m 2048 --mmap-flush 16M 74982.928 ( 2.471 ms): perf/26500 write(fd: 3</root/perf.data>, buf: 0x7ff94a6cc000, count: 3580888) = 3580888 74985.406 ( 2.353 ms): perf/26500 write(fd: 3</root/perf.data>, buf: 0x7ff949ecb000, count: 3453256) = 3453256 74987.764 ( 2.629 ms): perf/26500 write(fd: 3</root/perf.data>, buf: 0x7ff9496ca000, count: 3859232) = 3859232 74990.399 ( 2.341 ms): perf/26500 write(fd: 3</root/perf.data>, buf: 0x7ff948ec9000, count: 3769032) = 3769032 74992.744 ( 2.064 ms): perf/26500 write(fd: 3</root/perf.data>, buf: 0x7ff9486c8000, count: 3310520) = 3310520 74994.814 ( 2.619 ms): perf/26500 write(fd: 3</root/perf.data>, buf: 0x7ff947ec7000, count: 4194688) = 4194688 74997.439 ( 2.787 ms): perf/26500 write(fd: 3</root/perf.data>, buf: 0x7ff9476c6000, count: 4029760) = 4029760 Was again limited to a quarter of the mmap size: mmap flush: 2098176 mmap size 8392704B A warning about that would be good to have but can be added later, something like: "max flush is a quarter of the mmap size, if wanting to bump the mmap flush further, bump the mmap size as well using -m/--mmap-pages" Also rename the 'sync' parameters to 'synch' to keep tools/perf building with older glibcs: cc1: warnings being treated as errors builtin-record.c: In function 'record__mmap_read_evlist': builtin-record.c:775: warning: declaration of 'sync' shadows a global declaration /usr/include/unistd.h:933: warning: shadowed declaration is here builtin-record.c: In function 'record__mmap_read_all': builtin-record.c:856: warning: declaration of 'sync' shadows a global declaration /usr/include/unistd.h:933: warning: shadowed declaration is here Signed-off-by:Alexey Budankov <alexey.budankov@linux.intel.com> Reviewed-by:
Jiri Olsa <jolsa@kernel.org> Tested-by:
-
Alexey Budankov authored
Implement libzstd feature check, NO_LIBZSTD and LIBZSTD_DIR defines to override Zstd library sources or disable the feature from the command line: $ make -C tools/perf LIBZSTD_DIR=/path/to/zstd/sources/ clean all $ make -C tools/perf NO_LIBZSTD=1 clean all Auto detection feature status is reported just before compilation starts. If your system has some version of the zstd library preinstalled then the build system finds and uses it during the build. If you still prefer to compile with some other version of zstd library you have capability to refer the compilation to that version using LIBZSTD_DIR define. Signed-off-by:
-
Tzvetomir Stoyanov authored
tools lib traceevent: Rename input arguments and local variables of libtraceevent from pevent to tep "pevent" to "tep" renaming of: - all "pevent" input arguments of libtraceevent internal functions. - all local "pevent" variables of libtraceevent. This makes the implementation consistent with the chosen naming convention, tep (trace event parser), and will avoid any confusion with the old pevent name Signed-off-by:
Tzvetomir Stoyanov <tstoyanov@vmware.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Link: http://lore.kernel.org/linux-trace-devel/20190401132111.13727-5-tstoyanov@vmware.com Link: http://lkml.kernel.org/r/20190401164344.944953447@goodmis.orgSigned-off-by:
Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by:
-
Tzvetomir Stoyanov authored
The member "pevent" of the struct tep_event_filter is renamed to "tep". This makes the struct consistent with the chosen naming convention: tep (trace event parser), instead of the old pevent. Signed-off-by:
-
Tzvetomir Stoyanov authored
The member "pevent" of the struct tep_event is renamed to "tep". This makes the struct consistent with the chosen naming convention: tep (trace event parser), instead of the old pevent. Signed-off-by:
-
Tzvetomir Stoyanov authored
Input arguments of libtraceevent APIs are renamed from "struct tep_handle *pevent" to "struct tep_handle *tep". This makes the API consistent with the chosen naming convention: tep (trace event parser), instead of the old pevent. Signed-off-by:
-
Tzvetomir Stoyanov authored
Rename some traceevent APIs for consistency: tep_pid_is_registered() to tep_is_pid_registered() tep_file_bigendian() to tep_is_file_bigendian() to make the names and return values consistent with other tep_is_... APIs tep_data_lat_fmt() to tep_data_latency_format() to make the name more descriptive tep_host_bigendian() to tep_is_bigendian() tep_set_host_bigendian() to tep_set_local_bigendian() tep_is_host_bigendian() to tep_is_local_bigendian() "host" can be confused with VMs, and "local" is about the local machine. All tep_is_..._bigendian(struct tep_handle *tep) APIs return the saved data in the tep handle, while tep_is_bigendian() returns the running machine's endianness. All tep_is_... functions are modified to return bool value, instead of int. Signed-off-by:
-
Tzvetomir Stoyanov authored
This patch removes call to exit() from tep_filter_add_filter_str(). A library function should not force the application to exit. In the current implementation tep_filter_add_filter_str() calls exit() when a special "test_filters" mode is set, used only for debugging purposes. When this mode is set and a filter is added - its string is printed to the console and exit() is called. This patch changes the logic - when in "test_filters" mode, the filter string is still printed, but the exit() is not called. It is up to the application to track when "test_filters" mode is set and to call exit, if needed. Signed-off-by:
-
Tzvetomir Stoyanov authored
This patch removes trivial filter tep APIs: enum tep_filter_trivial_type tep_filter_event_has_trivial() tep_update_trivial() tep_filter_clear_trivial() Trivial filters is an optimization, used only in the first version of KernelShark. The API is deprecated, the next KernelShark release does not use it. Signed-off-by:
-
Steven Rostedt (VMware) authored
As return is not a function we do not need parenthesis around the return value. Also, a function returning bool does not need to add !!. Signed-off-by:
-
Tzvetomir Stoyanov authored
As struct tep_handler definition is not exposed as part of libtraceevent API, its fields cannot be accessed directly by the library users. This patch implements new APIs, which can be used to access the struct tep_handler fields: tep_get_event() - retrieves an event pointer at a specific index tep_get_first_event() - is modified to use tep_get_event() tep_clear_flag() - clears a tep handle flag tep_test_flag() - test if a given flag is set tep_get_header_timestamp_size() - returns the size of the timestamp stored in the header. tep_get_cpus() - returns the number of CPUs tep_is_old_format() - returns true if data was created by an older kernel with the old data format tep_set_print_raw() - have the output print in the raw format tep_set_test_filters() - debugging utility for testing tep filters Signed-off-by: -
Tzvetomir Stoyanov authored
Fixed few coding style problems in event-parse-api.c Signed-off-by:
-
Tzvetomir Stoyanov authored
APIs descriptions should describe the purpose of the function, its parameters and return value. While working on man pages implementation, I noticed mismatches in the descriptions of few APIs. This patch changes the description of these APIs, making them consistent with the man pages: - tep_print_num_field() - tep_print_func_field() - tep_get_header_page_size() - tep_get_long_size() - tep_set_long_size() - tep_get_page_size() - tep_set_page_size() Signed-off-by:
-
Steven Rostedt (Red Hat) authored
When trace-cmd report --debug is set, show the internal ring buffer entries like time-extends and padding. This requires adding new kbuffer API to retrieve these items. Signed-off-by:
-
Tzvetomir Stoyanov authored
Existing API tep_list_events() is not thread safe, it uses the internal array sort_events to keep cache of the sorted events and reuses it. This patch implements a new API, tep_list_events_copy(), which allocates new sorted array each time it is called. It could be used when a sorted events functionality is needed in thread safe use cases. It is up to the caller to free the array. Signed-off-by:
-
Steven Rostedt (VMware) authored
The mono clocks can display in seconds instead of whole numbers: trace-cmd-521 [001] 99176715281005: sched_waking: comm=kworker/u16:2 pid=32118 prio=120 target_cpu=002 trace-cmd-521 [001] 99176715286349: sched_wake_idle_without_ipi: cpu=2 trace-cmd-521 [001] 99176715288047: sched_wakeup: kworker/u16:2:32118 [120] success=1 CPU:002 trace-cmd-521 [001] 99176715290022: sched_waking: comm=trace-cmd pid=523 prio=120 target_cpu=000 trace-cmd-521 [001] 99176715292332: sched_wake_idle_without_ipi: cpu=0 trace-cmd-521 [001] 99176715292855: sched_wakeup: trace-cmd:523 [120] success=1 CPU:000 trace-cmd-521 [001] 99176715300697: sched_stat_runtime: comm=trace-cmd pid=521 runtime=80233 [ns] vruntime=66705540554 [ns Break it up in seconds: trace-cmd-521 [001] 99176.715281: sched_waking: comm=kworker/u16:2 pid=32118 prio=120 target_cpu=002 trace-cmd-521 [001] 99176.715286: sched_wake_idle_without_ipi: cpu=2 trace-cmd-521 [001] 99176.715288: sched_wakeup: kworker/u16:2:32118 [120] success=1 CPU:002 trace-cmd-521 [001] 99176.715290: sched_waking: comm=trace-cmd pid=523 prio=120 target_cpu=000 trace-cmd-521 [001] 99176.715292: sched_wake_idle_without_ipi: cpu=0 trace-cmd-521 [001] 99176.715293: sched_wakeup: trace-cmd:523 [120] success=1 CPU:000 trace-cmd-521 [001] 99176.715301: sched_stat_runtime: comm=trace-cmd pid=521 runtime=80233 [ns] vruntime=66705540554 [ns] Signed-off-by: -
Steven Rostedt (VMware) authored
With security updates, %p in the kernel is hashed to protect true kernel locations. But trace_printk() is not allowed in production systems, and when a pointer is used, there are many times that the actual address is needed. "%px" produces the real address. But libtraceevent does not know how to handle that extension. Add it. Signed-off-by:
-
Andi Kleen authored
Add support in 'perf list' to output tool internal events, currently only 'duration_time'. Committer testing: $ perf list dur* List of pre-defined events (to be used in -e): duration_time [Tool event] Metric Groups: $ perf list sw List of pre-defined events (to be used in -e): alignment-faults [Software event] bpf-output [Software event] context-switches OR cs [Software event] cpu-clock [Software event] cpu-migrations OR migrations [Software event] dummy [Software event] emulation-faults [Software event] major-faults [Software event] minor-faults [Software event] page-faults OR faults [Software event] task-clock [Software event] duration_time [Tool event] $ perf list | grep duration duration_time [Tool event] [L1D miss outstandings duration in cycles] page walk duration are excluded in Skylake] load. EPT page walk duration are excluded in Skylake] page walk duration are excluded in Skylake] store. EPT page walk duration are excluded in Skylake] (instruction fetch) request. EPT page walk duration are excluded in instruction fetch request. EPT page walk duration are excluded in $ Signed-off-by: -
Andi Kleen authored
Implement printing the correct name for duration_time Signed-off-by:
-
Andi Kleen authored
The perf metric expression use 'duration_time' internally to normalize events. Normal 'perf stat' without -x also prints the duration time. But when using -x, the interval is not output anywhere, which is inconvenient for any post processing which often wants to normalize values to time. So implement 'duration_time' as a proper perf event that can be specified explicitely with -e. The previous implementation of 'duration_time' only worked for metric processing. This adds the concept of a tool event that is handled by the tool. On the kernel level it is still mapped to the dummy software event, but the values are not read anymore, but instead computed by the tool. Add proper plumbing to handle this in the event parser, and display it in 'perf stat'. We don't want 'duration_time' to be added up, so it's only printed for the first CPU. % perf stat -e duration_time,cycles true Performance counter stats for 'true': 555,476 ns duration_time 771,958 cycles 0.000555476 seconds time elapsed 0.000644000 seconds user 0.000000000 seconds sys Signed-off-by: -
Andi Kleen authored
This reverts e864c5ca ("perf stat: Hide internal duration_time counter") but doing it manually since the code has now moved to a different file. The next patch will properly implement duration_time as a full event, so no need to hide it anymore. Signed-off-by:
-
Thomas Richter authored
Command # perf list --long-desc pmu lists the long description of the available counters. For counter named L1D_RO_EXCL_WRITES on machine types 3906 and 3907 the long description contains the counter number 'Counter:128 Name:' prefix. This is wrong. The fix changes the description text and removes this prefix. Output before: [root@m35lp76 perf]# ./perf list --long-desc pmu ... L1D_ONDRAWER_L4_SOURCED_WRITES [A directory write to the Level-1 Data cache directory where the returned cache line was sourced from On-Drawer Level-4 cache] L1D_RO_EXCL_WRITES [Counter:128 Name:L1D_RO_EXCL_WRITES A directory write to the Level-1 Data cache where the line was originally in a Read-Only state in the cache but has been updated to be in the Exclusive state that allows stores to the cache line] ... Output after: [root@m35lp76 perf]# ./perf list --long-desc pmu ... L1D_ONDRAWER_L4_SOURCED_WRITES [A directory write to the Level-1 Data cache directory where the returned cache line was sourced from On-Drawer Level-4 cache] L1D_RO_EXCL_WRITES [L1D_RO_EXCL_WRITES A directory write to the Level-1 Data cache where the line was originally in a Read-Only state in the cache but has been updated to be in the Exclusive state that allows stores to the cache line] ... Signed-off-by:Thomas Richter <tmricht@linux.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Fixes: 109d59b9 ("perf vendor events s390: Add JSON files for IBM z14") Link: http://lkml.kernel.org/r/20190329133337.60255-1-tmricht@linux.ibm.comSigned-off-by:
-
Arnaldo Carvalho de Melo authored
When adding the 'struct namespaces_event' to event.h, referencing the 'struct perf_ns_link_info' type, we forgot to add the header where it is defined, getting that definition only by sheer luck. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Hari Bathini <hbathini@linux.vnet.ibm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Fixes: f3b3614a ("perf tools: Add PERF_RECORD_NAMESPACES to include namespaces related info") Link: https://lkml.kernel.org/n/tip-qkrld0v7boc9uabjbd8csxux@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
There is no use for what is in that file, as everything is built by the tools/perf/trace/beauty/rename_flags.sh script from the copied kernel headers, the end result being: $ cat /tmp/build/perf/trace/beauty/generated/rename_flags_array.c static const char *rename_flags[] = { [0 + 1] = "NOREPLACE", [1 + 1] = "EXCHANGE", [2 + 1] = "WHITEOUT", }; $ I.e. no use of any defines from uapi/linux/fs.h 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-lgugmfa8z4bpw5zsbuoitllb@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
The previous method, copying to the BPF stack limited us in how many bytes we could copy from strings, use a PERCPU_ARRAY map like devised by the sysdig guys[1] to copy more bytes: Before: # trace --no-inherit -e openat touch `python -c "print "$s" 'a' * 2000"` touch: cannot touch 'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa': File name too long openat(AT_FDCWD, "/etc/ld.so.cache", O_RDONLY|O_CLOEXEC) = 3 openat(AT_FDCWD, "/lib64/libc.so.6", O_RDONLY|O_CLOEXEC) = 3 openat(AT_FDCWD, "/usr/lib/locale/locale-archive", O_RDONLY|O_CLOEXEC) = 3 openat(AT_FDCWD, "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", O_CREAT|O_NOCTTY|O_NONBLOCK|O_WRONLY, S_IRUGO|S_IWUGO) = -1 ENAMETOOLONG (File name too long) openat(AT_FDCWD, "/usr/share/locale/locale.alias", O_RDONLY|O_CLOEXEC) = 3 openat(AT_FDCWD, "/usr/share/locale/en_US.UTF-8/LC_MESSAGES/coreutils.mo", O_RDONLY) = -1 ENOENT (No such file or directory) openat(AT_FDCWD, "/usr/share/locale/en_US.utf8/LC_MESSAGES/coreutils.mo", O_RDONLY) = -1 ENOENT (No such file or directory) <SNIP some openat calls> # After: [root@quaco acme]# trace --no-inherit -e openat touch `python -c "print "$s" 'a' * 2000"` <STRIP what is the same as in the 'before' part> openat(AT_FDCWD, "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", O_CREAT|O_NOCTTY|O_NONBLOC) = -1 ENAMETOOLONG (File name too long) <STRIP what is the same as in the 'before' part> If we leave something like 'perf trace -e string' to trace all syscalls with a string, and then do some 'perf top', to get some annotation for the augmented_raw_syscalls.o BPF program we get: │ → callq *ffffffffc45576d1 ▒ │ augmented_args->filename.size = probe_read_str(&augmented_args->filename.value, ▒ 0.05 │ mov %eax,0x40(%r13) Looking with pahole, expanding types, asking for hex offsets and sizes, and use of BTF type information to see what is at that 0x40 offset from %r13: # pahole -F btf -C augmented_args_filename --expand_types --hex /home/acme/git/perf/tools/perf/examples/bpf/augmented_raw_syscalls.o struct augmented_args_filename { struct syscall_enter_args { long long unsigned int common_tp_fields; /* 0 0x8 */ long int syscall_nr; /* 0x8 0x8 */ long unsigned int args[6]; /* 0x10 0x30 */ } args; /* 0 0x40 */ /* --- cacheline 1 boundary (64 bytes) --- */ struct augmented_filename { unsigned int size; /* 0x40 0x4 */ int reserved; /* 0x44 0x4 */ char value[4096]; /* 0x48 0x1000 */ } filename; /* 0x40 0x1008 */ /* size: 4168, cachelines: 66, members: 2 */ /* last cacheline: 8 bytes */ }; # Then looking if PATH_MAX leaves some signature in the tests: │ if (augmented_args->filename.size < sizeof(augmented_args->filename.value)) { ▒ │ cmp $0xfff,%rdi 0xfff == 4095 sizeof(augmented_args->filename.value) == PATH_MAX == 4096 [1] https://sysdig.com/blog/the-art-of-writing-ebpf-programs-a-primer/ Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Andrii Nakryiko <andriin@fb.com> Cc: Daniel Borkmann <borkmann@iogearbox.net> Cc: Gianluca Borello <g.borello@gmail.com> Cc: Jesper Dangaard Brouer <brouer@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Luis Cláudio Gonçalves <lclaudio@redhat.com> cc: Martin Lau <kafai@fb.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Cc: Yonghong Song <yhs@fb.com> Link: https://lkml.kernel.org/n/tip-76gce2d2ghzq537ubwhjkone@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
Gets the augmented_raw_syscalls a bit more useful as-is, add a comment stating that the intent is to have all this in a map populated by userspace via the 'syscalls' BPF map, that right now has only a flag stating if the syscall is filtered or not. With it: # grep -B1 augmented_raw ~/.perfconfig [trace] add_events = /home/acme/git/perf/tools/perf/examples/bpf/augmented_raw_syscalls.o # # perf trace -e string weechat/6001 stat("/etc/localtime", 0x7ffe22c23d10) = 0 gnome-shell/1943 openat(AT_FDCWD, "/proc/self/stat", O_RDONLY) = 81 weechat/6001 stat("/etc/localtime", 0x7ffe22c23d10) = 0 gmain/2475 inotify_add_watch(20<anon_inode:inotify>, "/home/acme/.config/firewall", 16789454) = -1 ENOENT (No such file or directory) gmain/2391 inotify_add_watch(3<anon_inode:inotify>, "", 16789454) = -1 ENOENT (No such file or directory) gmain/2391 inotify_add_watch(3<anon_inode:inotify>, "/var/cache/app-info/yaml", 16789454) = -1 ENOENT (No such file or directory) gmain/2391 inotify_add_watch(3<anon_inode:inotify>, "/var/lib/app-info/xmls", 16789454) = -1 ENOENT (No such file or directory) gmain/2391 inotify_add_watch(3<anon_inode:inotify>, "/var/lib/app-info/yaml", 16789454) = -1 ENOENT (No such file or directory) gmain/2391 inotify_add_watch(3<anon_inode:inotify>, "/usr/share/app-info/yaml", 16789454) = -1 ENOENT (No such file or directory) gmain/2391 inotify_add_watch(3<anon_inode:inotify>, "/usr/local/share/app-info/xmls", 16789454) = -1 ENOENT (No such file or directory) gmain/2391 inotify_add_watch(3<anon_inode:inotify>, "/usr/local/share/app-info/yaml", 16789454) = -1 ENOENT (No such file or directory) gmain/2391 inotify_add_watch(3<anon_inode:inotify>, "/home/acme/.local/share/app-info/yaml", 16789454) = -1 ENOENT (No such file or directory) gmain/1121 inotify_add_watch(12<anon_inode:inotify>, "/etc/NetworkManager/VPN", 16789454) = -1 ENOENT (No such file or directory) weechat/6001 stat("/etc/localtime", 0x7ffe22c23d10) = 0 gmain/2050 inotify_add_watch(8<anon_inode:inotify>, "/home/acme/~", 16789454) = -1 ENOENT (No such file or directory) gmain/2521 inotify_add_watch(6<anon_inode:inotify>, "/var/lib/fwupd/remotes.d/lvfs-testing", 16789454) = -1 ENOENT (No such file or directory) weechat/6001 stat("/etc/localtime", 0x7ffe22c23d10) = 0 DOM Worker/22714 ... [continued]: openat()) = 257 FS Broker 3982/3990 openat(AT_FDCWD, "/dev/urandom", O_RDONLY|O_CLOEXEC|O_NOCTTY) = 187 DOMCacheThread/16652 mkdir("/home/acme/.mozilla/firefox/ina67tev.default/storage/default/https+++web.whatsapp.com/cache/morgue/192", S_IRUGO|S_IXUGO|S_IWUSR) = -1 EEXIST (File exists) ^C# Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Luis Cláudio Gonçalves <lclaudio@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: https://lkml.kernel.org/n/tip-a1hxffoy8t43e0wq6bzhp23u@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
Will be used in conjunction with the change to augmented_raw_syscalls.c in the next cset that adds all syscalls with a first or second arg string. With just what we have in the syscall tracepoints we get: # perf trace -e string ls > /dev/null ? ( ): ls/22382 ... [continued]: execve()) = 0 0.043 ( 0.004 ms): ls/22382 access(filename: 0x51ad420, mode: R) = -1 ENOENT (No such file or directory) 0.051 ( 0.004 ms): ls/22382 openat(dfd: CWD, filename: 0x51aa8b3, flags: RDONLY|CLOEXEC) = 3 0.071 ( 0.004 ms): ls/22382 openat(dfd: CWD, filename: 0x51b4d00, flags: RDONLY|CLOEXEC) = 3 0.138 ( 0.009 ms): ls/22382 openat(dfd: CWD, filename: 0x51684d0, flags: RDONLY|CLOEXEC) = 3 0.192 ( 0.004 ms): ls/22382 openat(dfd: CWD, filename: 0x51689c0, flags: RDONLY|CLOEXEC) = 3 0.255 ( 0.004 ms): ls/22382 openat(dfd: CWD, filename: 0x5168eb0, flags: RDONLY|CLOEXEC) = 3 0.342 ( 0.003 ms): ls/22382 openat(dfd: CWD, filename: 0x51693a0, flags: RDONLY|CLOEXEC) = 3 0.380 ( 0.003 ms): ls/22382 openat(dfd: CWD, filename: 0x5169950, flags: RDONLY|CLOEXEC) = 3 0.670 ( 0.011 ms): ls/22382 statfs(pathname: 0x515c783, buf: 0x7fff54d75b70) = 0 0.683 ( 0.005 ms): ls/22382 statfs(pathname: 0x515c783, buf: 0x7fff54d75a60) = 0 0.725 ( 0.004 ms): ls/22382 access(filename: 0x515c7ab) = 0 0.744 ( 0.005 ms): ls/22382 openat(dfd: CWD, filename: 0x50fba20, flags: RDONLY|CLOEXEC) = 3 0.793 ( 0.004 ms): ls/22382 openat(dfd: CWD, filename: 0x9e3e8390, flags: RDONLY|CLOEXEC|DIRECTORY|NONBLOCK) = 3 0.921 ( 0.006 ms): ls/22382 openat(dfd: CWD, filename: 0x50f7d90) = 3 # If we put the vfs_getname probe point in place: # perf probe 'vfs_getname=getname_flags:73 pathname=result->name:string' Added new events: probe:vfs_getname (on getname_flags:73 with pathname=result->name:string) probe:vfs_getname_1 (on getname_flags:73 with pathname=result->name:string) You can now use it in all perf tools, such as: perf record -e probe:vfs_getname_1 -aR sleep 1 # perf trace -e string ls > /dev/null ? ( ): ls/22440 ... [continued]: execve()) = 0 0.048 ( 0.008 ms): ls/22440 access(filename: /etc/ld.so.preload, mode: R) = -1 ENOENT (No such file or directory) 0.061 ( 0.007 ms): ls/22440 openat(dfd: CWD, filename: /etc/ld.so.cache, flags: RDONLY|CLOEXEC) = 3 0.092 ( 0.008 ms): ls/22440 openat(dfd: CWD, filename: /lib64/libselinux.so.1, flags: RDONLY|CLOEXEC) = 3 0.165 ( 0.007 ms): ls/22440 openat(dfd: CWD, filename: /lib64/libcap.so.2, flags: RDONLY|CLOEXEC) = 3 0.216 ( 0.007 ms): ls/22440 openat(dfd: CWD, filename: /lib64/libc.so.6, flags: RDONLY|CLOEXEC) = 3 0.282 ( 0.007 ms): ls/22440 openat(dfd: CWD, filename: /lib64/libpcre2-8.so.0, flags: RDONLY|CLOEXEC) = 3 0.340 ( 0.007 ms): ls/22440 openat(dfd: CWD, filename: /lib64/libdl.so.2, flags: RDONLY|CLOEXEC) = 3 0.383 ( 0.007 ms): ls/22440 openat(dfd: CWD, filename: /lib64/libpthread.so.0, flags: RDONLY|CLOEXEC) = 3 0.697 ( 0.021 ms): ls/22440 statfs(pathname: /sys/fs/selinux, buf: 0x7ffee7dc9010) = 0 0.720 ( 0.007 ms): ls/22440 statfs(pathname: /sys/fs/selinux, buf: 0x7ffee7dc8f00) = 0 0.757 ( 0.007 ms): ls/22440 access(filename: /etc/selinux/config) = 0 0.779 ( 0.009 ms): ls/22440 openat(dfd: CWD, filename: /usr/lib/locale/locale-archive, flags: RDONLY|CLOEXEC) = 3 0.830 ( 0.006 ms): ls/22440 openat(dfd: CWD, filename: ., flags: RDONLY|CLOEXEC|DIRECTORY|NONBLOCK) = 3 0.958 ( 0.010 ms): ls/22440 openat(dfd: CWD, filename: /usr/lib64/gconv/gconv-modules.cache) = 3 # Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Luis Cláudio Gonçalves <lclaudio@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: https://lkml.kernel.org/n/tip-6fh1myvn7ulf4xwq9iz3o776@git.kernel.orgSigned-off-by:
-
- 29 Mar, 2019 1 commit
-
-
Thomas Gleixner authored
Merge tag 'perf-urgent-for-mingo-5.1-20190329' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/urgent Pull perf/urgent fixes from Arnaldo: Core libraries: Jiri Olsa: - Fix max perf_event_attr.precise_ip detection. Kan Liang: - Fix parser error for uncore event alias Wei Lin: - Fixup ordering of kernel maps after obtaining the main kernel map address. Intel PT: Adrian Hunter: - Fix TSC slip where A TSC packet can slip past MTC packets so that the timestamp appears to go backwards. - Fixes for exported-sql-viewer GUI conversion to python3. ARM coresight: Solomon Tan: - Fix the build by adding a missing case value for enumeration value introduced in newer library, that now is the required one. tool headers: Arnaldo Carvalho de Melo: - Syncronize kernel headers with the kernel, getting new io_uring and pidfd_send_signal syscalls so that 'perf trace' can handle them. Signed-off-by:
-
