- 28 Aug, 2015 6 commits
-
-
Arnaldo Carvalho de Melo authored
So that functions that deal primarily with an evsel to access information that concerns the whole evlist it is in. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Andy Lutomirski <luto@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/r/1440677263-21954-5-git-send-email-kan.liang@intel.comSigned-off-by:
Kan Liang <kan.liang@intel.com> Signed-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Arnaldo Carvalho de Melo authored
Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-5yqtfs728r1j1u8zmg8ufxwm@git.kernel.orgSigned-off-by:
-
Tony Jones authored
This patch creates a new script (compaction-times) to report time spent in mm compaction. It is possible to report times in nanoseconds (default) or microseconds (-u). The option -p will break down results by process id, -pv will further decompose by each compaction entry/exit. For each compaction entry/exit what is reported is controlled by the options: -t report only timing -m report migration stats -ms report migration scanner stats -fs report free scanner stats The default is to report all. Entries may be further filtered by pid, pid-range or comm (regex). The script is useful when analysing workloads that compact memory. The most common example will be THP allocations on systems with a lot of uptime that has fragmented memory. This is an example of using the script to analyse a thpscale from mmtests which deliberately fragments memory and allocates THP in 4 separate threads # Recording step, one of the following; $ perf record -e 'compaction:mm_compaction_*' ./workload # or: $ perf script record compaction-times # Reporting: basic total: 2444505743ns migration: moved=357738 failed=39275 free_scanner: scanned=2705578 isolated=387875 migration_scanner: scanned=414426 isolated=397013 # Reporting: Per task stall times $ perf script report compaction-times -- -t -p total: 2444505743ns 6384[thpscale]: 740800017ns 6385[thpscale]: 274119512ns 6386[thpscale]: 832961337ns 6383[thpscale]: 596624877ns # Reporting: Per-compaction attempts for task 6385 $ perf script report compaction-times -- -m -pv 6385 total: 274119512ns migration: moved=14893 failed=24285 6385[thpscale]: 274119512ns migration: moved=14893 failed=24285 6385[thpscale].1: 3033277ns migration: moved=511 failed=1 6385[thpscale].2: 9592094ns migration: moved=1524 failed=12 6385[thpscale].3: 2495587ns migration: moved=512 failed=0 6385[thpscale].4: 2561766ns migration: moved=512 failed=0 6385[thpscale].5: 2523521ns migration: moved=512 failed=0 ..... output continues ... Changes since v1: - report stats for isolate_migratepages and isolate_freepages (Vlastimil Babka) - refactor code to achieve above - add help text - output to stdout/stderr explicitly Signed-off-by:
Tony Jones <tonyj@suse.com> Cc: Mel Gorman <mgorman@suse.com> Cc: Vlastimil Babka <vbabka@suse.cz> Link: http://lkml.kernel.org/r/1439840932-8933-1-git-send-email-tonyj@suse.comSigned-off-by:
-
Steven Rostedt authored
It was reported that "%-8s" does not parse well when used in the printk format. The '-' is what is throwing it off. Allow that to be included. Reporter note: Example before: transhuge-stres-10730 [004] 5897.713989: mm_compaction_finished: node=0 zone=>-<8s order=-2119871790 ret= Example after: transhuge-stres-4235 [000] 453.149280: mm_compaction_finished: node=0 zone=ffffffff81815d7a order=9 ret= (I will send patches to fix the string handling in the tracepoints so it's on par with in-kernel printing via trace_pipe:) transhuge-stres-10921 [007] ...1 6307.140205: mm_compaction_finished: node=0 zone=Normal order=9 ret=partial Reported-by:
Vlastimil Babka <vbabka@suse.cz> Signed-off-by:
Steven Rostedt <rostedt@goodmis.org> Tested-by:
-
Mark Drayton authored
Sometimes when post-processing output from `perf script` one does not want to demangle C++ symbol names. Add an option to allow this. Also add --[no-]demangle-kernel to be consistent with top/report/probe. Signed-off-by:
Mark Drayton <mbd@fb.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/1440616695-32340-1-git-send-email-scientist@fb.comSigned-off-by:
Yannick Brosseau <scientist@fb.com> Signed-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo: User visible changes: - Add support for using several Intel PT features (CYC, MTC packets), the relevant documentation was updated: tools/perf/Documentation/intel-pt.txt, briefly describing those packets, its purposes, how to configure them in the event config terms and relevant external documentation for further reading. (Adrian Hunter) - Introduce support for probing at an absolute address, for user and kernel 'perf probe's, useful when one have the symbol maps on a developer machine but not on an embedded system. (Wang Nan) - Fix 'perf probe' list results when a symbol can't be found or the address is zero and when an offset is provided without a function (Wang Nan) - Do not print '0x (null)' in uprobes when offset is zero (Wang Nan) - Clear the progress bar at the end of a ordered_events flush, fixing an UI artifact when, after ordering the events the screen doesn't get completely redraw, for instance, when an error window covers just the center of the screen and waits for user input. (Arnaldo Carvalho de Melo) - Fix 'annotate' segfault by resetting the dso find_symbol cache when removing symbols. (Arnaldo Carvalho de Melo) Infrastructure changes: - Allow duplicate objects in the object list, just like it is possible to have things like this, in the kernel: drivers/Makefile:obj-$(CONFIG_PCI) += usb/ drivers/Makefile:obj-$(CONFIG_USB_GADGET) += usb/ (Jiri Olsa) - Fix Intel PT 'instructions' sample period. (Adrian Hunter) - Prevent segfault when reading probe point with absolute address. (Wang Nan) Build fixes: - Fix tarball build broken by pt/bts. (Adrian Hunter) - Remove export.h from MANIFEST, fixing the perf tarball make target. (Jiri Olsa) Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
- 26 Aug, 2015 8 commits
-
-
Wang Nan authored
When manually added uprobe point with zero address, 'uprobe_events' output '(null)' instead of 0x00000000: # echo p:probe_libc/abs_0 /path/to/lib.bin:0x0 arg1=%ax > \ /sys/kernel/debug/tracing/uprobe_events # cat /sys/kernel/debug/tracing/uprobe_events p:probe_libc/abs_0 /path/to/lib.bin:0x (null) arg1=%ax This patch fixes this behavior: # cat /sys/kernel/debug/tracing/uprobe_events p:probe_libc/abs_0 /path/to/lib.bin:0x0000000000000000 Signed-off-by:Wang Nan <wangnan0@huawei.com> Acked-by:
Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Zefan Li <lizefan@huawei.com> Cc: pi3orama@163.com Link: http://lkml.kernel.org/r/1440586666-235233-8-git-send-email-wangnan0@huawei.comSigned-off-by:
-
Wang Nan authored
It should be useful to allow 'perf probe' probe at absolute offset of a target. For example, when (u)probing at a instruction of a shared object in a embedded system where debuginfo is not avaliable but we know the offset of that instruction by manually digging. This patch enables following perf probe command syntax: # perf probe 0xffffffff811e6615 And # perf probe /lib/x86_64-linux-gnu/libc-2.19.so 0xeb860 In the above example, we don't need a anchor symbol, so it is possible to compute absolute addresses using other methods and then use 'perf probe' to create the probing points. v1 -> v2: Drop the leading '+' in cmdline; Allow uprobing at offset 0x0; Improve 'perf probe -l' result when uprobe at area without debuginfo. v2 -> v3: Split bugfix to a separated patch. Test result: # perf probe 0xffffffff8119d175 %ax # perf probe sys_write %ax # perf probe /lib64/libc-2.18.so 0x0 %ax # perf probe /lib64/libc-2.18.so 0x5 %ax # perf probe /lib64/libc-2.18.so 0xd8e40 %ax # perf probe /lib64/libc-2.18.so __write %ax # perf probe /lib64/libc-2.18.so 0xd8e49 %ax # cat /sys/kernel/debug/tracing/uprobe_events p:probe_libc/abs_0 /lib64/libc-2.18.so:0x (null) arg1=%ax p:probe_libc/abs_5 /lib64/libc-2.18.so:0x0000000000000005 arg1=%ax p:probe_libc/abs_d8e40 /lib64/libc-2.18.so:0x00000000000d8e40 arg1=%ax p:probe_libc/__write /lib64/libc-2.18.so:0x00000000000d8e40 arg1=%ax p:probe_libc/abs_d8e49 /lib64/libc-2.18.so:0x00000000000d8e49 arg1=%ax # cat /sys/kernel/debug/tracing/kprobe_events p:probe/abs_ffffffff8119d175 0xffffffff8119d175 arg1=%ax p:probe/sys_write _text+1692016 arg1=%ax # perf probe -l Failed to find debug information for address 5 probe:abs_ffffffff8119d175 (on sys_write+5 with arg1) probe:sys_write (on sys_write with arg1) probe_libc:__write (on @unix/syscall-template.S:81 in /lib64/libc-2.18.so with arg1) probe_libc:abs_0 (on 0x0 in /lib64/libc-2.18.so with arg1) probe_libc:abs_5 (on 0x5 in /lib64/libc-2.18.so with arg1) probe_libc:abs_d8e40 (on @unix/syscall-template.S:81 in /lib64/libc-2.18.so with arg1) probe_libc:abs_d8e49 (on __GI___libc_write+9 in /lib64/libc-2.18.so with arg1) Signed-off-by: -
Wang Nan authored
This patch fixes a bug that, when offset is provided but function is lost, parse_perf_probe_point() will give a "" string as function name, so the checking code at the end of parse_perf_probe_point() become useless. For example: # perf probe +0x1234 Failed to find symbol in kernel Error: Failed to add events. After this patch: # perf probe +0x1234 Semantic error :Offset requires an entry function. Error: Command Parse Error. Signed-off-by: -
Wang Nan authored
When manually added uprobe point with zero address, 'perf probe -l' reports error. For example: # echo p:probe_libc/abs_0 /path/to/lib.bin:0x0 arg1=%ax > \ /sys/kernel/debug/tracing/uprobe_events # perf probe -l Error: Failed to show event list. Probing at 0x0 is possible and useful when lib.bin is not a normal shared object but is manually mapped. However, in this case kernel report: # cat /sys/kernel/debug/tracing/uprobe_events p:probe_libc/abs_0 /path/to/lib.bin:0x (null) arg1=%ax This patch supports the above kernel output. Signed-off-by: -
Wang Nan authored
'perf probe -l' reports error if it is unable find symbol through address. Here is an example. # echo 'p:probe_libc/abs_5 /lib64/libc.so.6:0x5' > /sys/kernel/debug/tracing/uprobe_events # cat /sys/kernel/debug/tracing/uprobe_events p:probe_libc/abs_5 /lib64/libc.so.6:0x0000000000000005 # perf probe -l Error: Failed to show event list Also, this situation triggers a logical inconsistency in convert_to_perf_probe_point() that, it returns ENOMEM but actually it never try strdup(). This patch removes !tp->module && !is_kprobe condition, so it always uses address to build function name if symbol not found. Test result: # perf probe -l probe_libc:abs_5 (on 0x5 in /lib64/libc.so.6) Signed-off-by: -
Jiri Olsa authored
It's sometimes useful to specify the object affiliation to multiple config options like: libperf-$(CONFIG_X86) += tsc.o libperf-$(CONFIG_AUXTRACE) += tsc.o while the object itself is linked only once. Adding the support for this and ignoring duplicate objects in the object list. Suggested-by:
Arnaldo Carvalho de Melo <acme@kernel.org> Signed-off-by:
Jiri Olsa <jolsa@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> Link: http://lkml.kernel.org/r/20150826130103.GF22670@krava.redhat.comSigned-off-by:
-
Jiri Olsa authored
We don't carry an export.h wrapper anymore, remove it from the MANIFEST file to avoid breaking the make perf-tar targets. Signed-off-by:
-
Wang Nan authored
'perf probe -l' panic if there is a manually inserted probing point with absolute address. For example: # echo 'p:probe/abs_ffffffff811e6615 0xffffffff811e6615' > /sys/kernel/debug/tracing/kprobe_events # perf probe -l Segmentation fault (core dumped) This patch fix this problem by considering the situation that "tp->symbol == NULL" in find_perf_probe_point_from_dwarf() and find_perf_probe_point_from_map(). After this patch: # perf probe -l probe:abs_ffffffff811e6615 (on SyS_write+5@fs/read_write.c) And when debug info is missing: # rm -rf ~/.debug # mv /lib/modules/4.2.0-rc1+/build/vmlinux /lib/modules/4.2.0-rc1+/build/vmlinux.bak # perf probe -l probe:abs_ffffffff811e6615 (on sys_write+5) Signed-off-by:
-
- 24 Aug, 2015 13 commits
-
-
Adrian Hunter authored
Update Intel PT documentation to describe new features. Signed-off-by:
Adrian Hunter <adrian.hunter@intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Link: http://lkml.kernel.org/r/1437150840-31811-26-git-send-email-adrian.hunter@intel.comSigned-off-by:
-
Adrian Hunter authored
A TRACESTOP packet is produced when an Intel PT trace enters a defined region of the address space at which point the tracing stops. This patch just adds decoder support. Support for specifying TRACESTOP regions is left until later. For details refer to the June 2015 or later Intel 64 and IA-32 Architectures SDM Chapter 36 Intel Processor Trace. Signed-off-by:
-
Adrian Hunter authored
CYC packets are a new Intel PT feature. CYC packets provide even finer grain timestamp information than MTC and TSC packets. A CYC packet contains the number of CPU cycles since the last CYC packet. Unlike MTC and TSC packets, CYC packets are only sent when another packet is also sent. Support for this feature is indicated by: /sys/bus/event_source/devices/intel_pt/caps/psb_cyc which contains "1" if the feature is supported and "0" otherwise. CYC packets can be requested using a PMU config term e.g. perf record -e intel_pt/cyc/u sleep 1 The frequency of CYC packets can also be specified. e.g. perf record -e intel_pt/cyc,cyc_thresh=2/u sleep 1 CYC packets are not requested by default. Valid cyc_thresh values are given by: /sys/bus/event_source/devices/intel_pt/caps/cycle_thresholds which contains a hexadecimal value, the bits of which represent valid values e.g. bit 2 set means value 2 is valid. The value represents the minimum number of CPU cycles that must have passed before a CYC packet can be sent. The number of CPU cycles is: 2 ^ (value - 1) e.g. value 4 means 8 CPU cycles must pass before a CYC packet can be sent. Note a CYC packet is still only sent when another packet is sent, not at, e.g. every 8 CPU cycles. If an invalid value is entered, the error message will give a list of valid values e.g. $ perf record -e intel_pt/cyc,cyc_thresh=15/u uname Invalid cyc_thresh for intel_pt. Valid values are: 0-12 tools/perf/Documentation/intel-pt.txt is updated in a later patch as there are a number of new features being added. For more information refer to the June 2015 or later Intel 64 and IA-32 Architectures SDM Chapter 36 Intel Processor Trace. Signed-off-by: -
Adrian Hunter authored
CYC packets provide even finer grain timestamp information than MTC and TSC packets. A CYC packet contains the number of CPU cycles since the last CYC packet. This patch just adds decoder support. The CPU frequency can be related to TSC using the Maximum Non-Turbo Ratio in combination with the CBR (core-to-bus ratio) packet. However more accuracy is achieved by simply interpolating the number of cycles between other timing packets like MTC or TSC. This patch takes the latter approach. Support for a default value and validation of values is provided by a later patch. Also documentation is updated in a separate patch. For details refer to the June 2015 or later Intel 64 and IA-32 Architectures SDM Chapter 36 Intel Processor Trace. Signed-off-by:
-
Adrian Hunter authored
MTC packets are a new Intel PT feature. MTC packets provide finer grain timestamp information than TSC packets. Support for this feature is indicated by: /sys/bus/event_source/devices/intel_pt/caps/mtc which contains "1" if the feature is supported and "0" otherwise. MTC packets can be requested using a PMU config term e.g. perf record -e intel_pt/mtc/u sleep 1 The frequency of MTC packets can also be specified. e.g. perf record -e intel_pt/mtc,mtc_period=2/u sleep 1 The default value is 3 or the nearest lower value that is supported. 0 is always supported. Valid values are given by: /sys/bus/event_source/devices/intel_pt/caps/mtc_periods which contains a hexadecimal value, the bits of which represent valid values e.g. bit 2 set means value 2 is valid. The value is converted to the MTC frequency as: CTC-frequency / (2 ^ value) e.g. value 3 means one eighth of CTC-frequency Where CTC is the hardware crystal clock, the frequency of which can be related to TSC via values provided in cpuid leaf 0x15. If an invalid value is entered, the error message will give a list of valid values e.g. $ perf record -e intel_pt/mtc_period=15/u uname Invalid mtc_period for intel_pt. Valid values are: 0,3,6,9 tools/perf/Documentation/intel-pt.txt is updated in a later patch as there are a number of new features being added. For more information refer to the June 2015 or later Intel 64 and IA-32 Architectures SDM Chapter 36 Intel Processor Trace. Signed-off-by:
-
Adrian Hunter authored
MTC packets provide finer grain timestamp information than TSC packets. MTC packets record time using the hardware crystal clock (CTC) which is related to TSC packets using a TMA packet. This patch just adds decoder support. Support for a default value and validation of values is provided by a later patch. Also documentation is updated in a separate patch. For details refer to the June 2015 or later Intel 64 and IA-32 Architectures SDM Chapter 36 Intel Processor Trace. Signed-off-by:
-
Adrian Hunter authored
Record additional information in the AUXTRACE_INFO event in preparation for decoding MTC and CYC packets. Pass the information to the decoder. The AUXTRACE_INFO record can be extended by using the size to indicate the presence of new members. The additional information includes PMU config bit positions and the TSC to CTC (hardware crystal clock) ratio needed to decode MTC packets. Signed-off-by:
-
Adrian Hunter authored
New features have been added to Intel PT which include a number of new packet definitions. This patch adds packet definitions for new packets: TMA, MTC, CYC, VMCS, TRACESTOP and MNT. Also another bit in PIP is defined. This patch only adds support for the definitions. Later patches add support for decoding TMA, MTC, CYC and TRACESTOP which is where those packets are explained. VMCS and the newly defined bit in PIP are used with virtualization which is not supported yet. MNT is a maintenance packet which the decoder should ignore. For details, refer to the June 2015 or later Intel 64 and IA-32 Architectures SDM Chapter 36 Intel Processor Trace. Signed-off-by:
-
Adrian Hunter authored
The PSB packet is a synchronization packet that provides a starting point for decoding or recovery from errors. This patch adds support for a new Intel PT feature that allows the frequency of PSB packets to be specified. Support for this feature is indicated by /sys/bus/event_source/devices/intel_pt/caps/psb_cyc which contains "1" if the feature is supported and "0" otherwise. The PSB period can be specified as a PMU config term e.g. perf record -e intel_pt/psb_period=2/u sleep 1 The default value is 3 or the nearest lower value that is supported. 0 is always supported. Valid values are given by: /sys/bus/event_source/devices/intel_pt/caps/psb_periods which contains a hexadecimal value, the bits of which represent valid values e.g. bit 2 set means value 2 is valid. The value is converted to the approximate number of trace bytes between PSB packets as: 2 ^ (value + 11) e.g. value 3 means 16KiB bytes between PSBs If an invalid value is entered, the error message will give a list of valid values e.g. $ perf record -e intel_pt/psb_period=15/u uname Invalid psb_period for intel_pt. Valid values are: 0-5 tools/perf/Documentation/intel-pt.txt is updated in a later patch as there are a number of new features being added. For more information about PSB periods refer to the Intel 64 and IA-32 Architectures SDM Chapter 36 Intel Processor Trace from June 2015 or later. Signed-off-by:
-
Adrian Hunter authored
The period on synthesized 'instructions' samples was being set to a fixed value, whereas the correct value is the number of instructions since the last sample, which is a value that the decoder can provide. So do it that way. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
We were depending on the next screen operation after a flush() being one that would redraw the whole screen so that the progress bar would be overwritten, when that didn't happen a screen artifact of, say, a error dialog window would be overlaid on top of the progress bar, fix it by calling ui_browser__finish(), that now has a TUI implementation. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-el0fyw6duemnx62lydjzhs8c@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
So that we can erase the progress bar after we're done with it, avoiding things like: ------------------------------------------------------------------- ┌─Error:──────────────────────────────────────────────────────┐ │Can't annotate unmapped_area_topdown: │ │ │ │No vmlinux file with build id a826726b5ddacfab1f0bade868f1a79│ │was found in the path. │ │ │ │Note that annotation using /proc/kcore requires CAP_SYS_RAWIO│ ┌Processin│ │──┐ │ │Please use: │ │ └─────────│ │──┘ │ perf buildid-cache -vu vmlinux │ │ │ │or: │ │ │ │ --vmlinux vmlinux │ │ │ │ │ │Press any key... │ └─────────────────────────────────────────────────────────────┘ Can't annotate unmapped_area_topdown: ------------------------------------------------------------------- I.e. that finished progress bar behind the error window. It is not a problem when we end up redrawing the whole screen, but its ugly when we present such error windows, provide a TUI method so that code like the above may avoid this situation, as will be done with the annotation code in the next cset. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-qvktnojzwwe37pweging058t@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
The 'annotate' tool does some filtering in the entries in a DSO but forgot to reset the cache done in dso__find_symbol(), cauxing a SEGV: [root@zoo ~]# perf annotate netlink_poll perf: Segmentation fault -------- backtrace -------- perf[0x526ceb] /lib64/libc.so.6(+0x34960)[0x7faedfbe0960] perf(rb_erase+0x223)[0x499d63] perf[0x4213e9] perf[0x4bc123] perf[0x4bc621] perf[0x4bf26b] perf[0x4bc855] perf(perf_session__process_events+0x340)[0x4bddc0] perf(cmd_annotate+0x6bb)[0x421b5b] perf[0x479063] perf(main+0x60a)[0x42098a] /lib64/libc.so.6(__libc_start_main+0xf0)[0x7faedfbcbfe0] perf[0x420aa9] [0x0] [root@zoo ~]# Fix it by reseting the find cache when removing symbols. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Fixes: b685ac22 ("perf symbols: Add front end cache for DSO symbol lookup") Link: http://lkml.kernel.org/n/tip-b2y9x46y0t8yem1ive41zqyp@git.kernel.orgSigned-off-by:
-
- 22 Aug, 2015 2 commits
-
-
Adrian Hunter authored
Fix some include paths and add missing inat_types.h. Reported-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo: User visible changes: - Fix segfault using 'perf script --show-mmap-events', affects only current perf/core. (Adrian Hunter) - /proc/kcore requires CAP_SYS_RAWIO message too noisy, make it debug only. (Adrian Hunter) - Fix Intel PT timestamp handling. (Adrian Hunter) - Add Intel BTS support, with a call-graph script to show it and PT in use in a GUI using 'perf script' python scripting with postgresql and Qt. (Adrian Hunter) - Add checks for returned EVENT_ERROR type in libtraceevent, fixing a bug that surfaced on arm64 systems. (Dean Nelson) - Fallback to using kallsyms when libdw fails to handle a vmlinux file, that can happen, for instance, when perf is statically linked and then libdw fails to load libebl_{arch}.so. (Wang Nan) Infrastructure changes: - Initialize reference counts in map__clone(). (Arnaldo Carvalho de Melo) Signed-off-by:
-
- 21 Aug, 2015 10 commits
-
-
Wang Nan authored
A problem can occur in a statically linked perf when vmlinux can be found: # perf probe --add sys_epoll_pwait probe-definition(0): sys_epoll_pwait symbol:sys_epoll_pwait file:(null) line:0 offset:0 return:0 lazy:(null) 0 arguments Looking at the vmlinux_path (7 entries long) Using /lib/modules/4.2.0-rc1+/build/vmlinux for symbols Open Debuginfo file: /lib/modules/4.2.0-rc1+/build/vmlinux Try to find probe point from debuginfo. Symbol sys_epoll_pwait address found : ffffffff8122bd40 Matched function: SyS_epoll_pwait Failed to get call frame on 0xffffffff8122bd40 An error occurred in debuginfo analysis (-2). Error: Failed to add events. Reason: No such file or directory (Code: -2) The reason is caused by libdw that, if libdw is statically linked, it can't load libebl_{arch}.so reliable. In this case it is still possible to get the address from /proc/kalksyms. However, perf tries that only when libdw returns -EBADF. This patch gives it another chance to utilize symbol table, even if libdw returns an error code other than -EBADF. After applying this patch: # perf probe -nv --add sys_epoll_pwait probe-definition(0): sys_epoll_pwait symbol:sys_epoll_pwait file:(null) line:0 offset:0 return:0 lazy:(null) 0 arguments Looking at the vmlinux_path (7 entries long) Using /lib/modules/4.2.0-rc1+/build/vmlinux for symbols Open Debuginfo file: /lib/modules/4.2.0-rc1+/build/vmlinux Try to find probe point from debuginfo. Symbol sys_epoll_pwait address found : ffffffff8122bd40 Matched function: SyS_epoll_pwait Failed to get call frame on 0xffffffff8122bd40 An error occurred in debuginfo analysis (-2). Trying to use symbols. Opening /sys/kernel/debug/tracing/kprobe_events write=1 Added new event: Writing event: p:probe/sys_epoll_pwait _text+2276672 probe:sys_epoll_pwait (on sys_epoll_pwait) You can now use it in all perf tools, such as: perf record -e probe:sys_epoll_pwait -aR sleep 1 Although libdw returns an error (Failed to get call frame), perf tries symbol table and finally gets correct address. Signed-off-by: -
Arnaldo Carvalho de Melo authored
Map clone was written before we introduced reference counts for maps and dsos, so all that was needed was just a copy and then we would insert it into the new map_groups instance. Fix it by, after copying, initializing the map->refcnt, grabbing a struct dso refcount and resetting pointers that may be used to determine if a map, when deleted, is in a rb_tree. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-pd4mr80o5b9gvk50iineacec@git.kernel.orgSigned-off-by:
-
Adrian Hunter authored
Add a script to produce a call-graph from data exported to a postgresql database and derived from a processor trace event like intel_pt or intel_bts. Refer to comments in the scripts call-graph-from-postgresql.py and export-to-postgresql.py for more details on how to set up the environment, install the required packages, etc. Committer note: From the scripts, for convenience while reading 'git log': An example of using this script with Intel PT: $ perf record -e intel_pt//u ls $ perf script -s ~/libexec/perf-core/scripts/python/export-to-postgresql.py pt_example branches calls 2015-05-29 12:49:23.464364 Creating database... 2015-05-29 12:49:26.281717 Writing to intermediate files... 2015-05-29 12:49:27.190383 Copying to database... 2015-05-29 12:49:28.140451 Removing intermediate files... 2015-05-29 12:49:28.147451 Adding primary keys 2015-05-29 12:49:28.655683 Adding foreign keys 2015-05-29 12:49:29.365350 Done $ python tools/perf/scripts/python/call-graph-from-postgresql.py pt_example # The result is a GUI window with a tree representing a context-sensitive # call-graph. Expanding a couple of levels of the tree and adjusting column # widths to suit will display something like: Call Graph: pt_example Call Path |Object |Count|Time(ns)|Time(%)|Branch Count|Branch Count(%) v- ls v- 2638:2638 v- _start ld-2.19.so 1 10074071 100.0 211135 100.0 |- unknown unknown 1 13198 0.1 1 0.0 >- _dl_start ld-2.19.so 1 1400980 13.9 19637 9.3 >- _d_linit_internal ld-2.19.so 1 448152 4.4 11094 5.3 v-__libc_start_main@plt ls 1 8211741 81.5 180397 85.4 >- _dl_fixup ld-2.19.so 1 7607 0.1 108 0.1 >- __cxa_atexit libc-2.19.so 1 11737 0.1 10 0.0 >- __libc_csu_init ls 1 10354 0.1 10 0.0 |- _setjmp libc-2.19.so 1 0 0.0 4 0.0 v- main ls 1 8182043 99.6 180254 99.9 Signed-off-by: -
Adrian Hunter authored
perf script, report and inject all have the same itrace options. Put them into an asciidoc include file. Signed-off-by:
-
Adrian Hunter authored
Intel BTS support fits within the new auxtrace infrastructure. Recording is supporting by identifying the Intel BTS PMU, parsing options and setting up events. Decoding is supported by queuing up trace data by thread and then decoding synchronously delivering synthesized event samples into the session processing for tools to consume. Committer note: E.g: [root@felicio ~]# perf record --per-thread -e intel_bts// ls anaconda-ks.cfg apctest.output bin kernel-rt-3.10.0-298.rt56.171.el7.x86_64.rpm libexec lock_page.bpf.c perf.data perf.data.old [ perf record: Woken up 3 times to write data ] [ perf record: Captured and wrote 4.367 MB perf.data ] [root@felicio ~]# perf evlist -v intel_bts//: type: 6, size: 112, { sample_period, sample_freq }: 1, sample_type: IP|TID|IDENTIFIER, read_format: ID, disabled: 1, enable_on_exec: 1, sample_id_all: 1, exclude_guest: 1 dummy:u: type: 1, size: 112, config: 0x9, { sample_period, sample_freq }: 1, sample_type: IP|TID|IDENTIFIER, read_format: ID, disabled: 1, exclude_kernel: 1, exclude_hv: 1, mmap: 1, comm: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, mmap2: 1, comm_exec: 1 [root@felicio ~]# perf script # the navigate in the pager to some interesting place: ls 1843 1 branches: ffffffff810a60cb flush_signal_handlers ([kernel.kallsyms]) => ffffffff8121a522 setup_new_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8121a529 setup_new_exec ([kernel.kallsyms]) => ffffffff8122fa30 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fa5d do_close_on_exec ([kernel.kallsyms]) => ffffffff81767ae0 _raw_spin_lock ([kernel.kallsyms]) ls 1843 1 branches: ffffffff81767af4 _raw_spin_lock ([kernel.kallsyms]) => ffffffff8122fa62 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fa8e do_close_on_exec ([kernel.kallsyms]) => ffffffff8122faf0 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122faf7 do_close_on_exec ([kernel.kallsyms]) => ffffffff8122fa8b do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fa8e do_close_on_exec ([kernel.kallsyms]) => ffffffff8122faf0 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122faf7 do_close_on_exec ([kernel.kallsyms]) => ffffffff8122fa8b do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fa8e do_close_on_exec ([kernel.kallsyms]) => ffffffff8122faf0 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122faf7 do_close_on_exec ([kernel.kallsyms]) => ffffffff8122fa8b do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fa8e do_close_on_exec ([kernel.kallsyms]) => ffffffff8122faf0 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122faf7 do_close_on_exec ([kernel.kallsyms]) => ffffffff8122fa8b do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fa8e do_close_on_exec ([kernel.kallsyms]) => ffffffff8122faf0 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122faf7 do_close_on_exec ([kernel.kallsyms]) => ffffffff8122fa8b do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fa8e do_close_on_exec ([kernel.kallsyms]) => ffffffff8122faf0 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122faf7 do_close_on_exec ([kernel.kallsyms]) => ffffffff8122fa8b do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fac9 do_close_on_exec ([kernel.kallsyms]) => ffffffff8122fad2 do_close_on_exec ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8122fadd do_close_on_exec ([kernel.kallsyms]) => ffffffff8120fc80 filp_close ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8120fcaf filp_close ([kernel.kallsyms]) => ffffffff8120fcb6 filp_close ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8120fcc2 filp_close ([kernel.kallsyms]) => ffffffff812547f0 dnotify_flush ([kernel.kallsyms]) ls 1843 1 branches: ffffffff81254823 dnotify_flush ([kernel.kallsyms]) => ffffffff8120fcc7 filp_close ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8120fccd filp_close ([kernel.kallsyms]) => ffffffff81261790 locks_remove_posix ([kernel.kallsyms]) ls 1843 1 branches: ffffffff812617a3 locks_remove_posix ([kernel.kallsyms]) => ffffffff812617b9 locks_remove_posix ([kernel.kallsyms]) ls 1843 1 branches: ffffffff812617b9 locks_remove_posix ([kernel.kallsyms]) => ffffffff8120fcd2 filp_close ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8120fcd5 filp_close ([kernel.kallsyms]) => ffffffff812142c0 fput ([kernel.kallsyms]) ls 1843 1 branches: ffffffff812142d6 fput ([kernel.kallsyms]) => ffffffff812142df fput ([kernel.kallsyms]) ls 1843 1 branches: ffffffff8121430c fput ([kernel.kallsyms]) => ffffffff810b6580 task_work_add ([kernel.kallsyms]) ls 1843 1 branches: ffffffff810b65ad task_work_add ([kernel.kallsyms]) => ffffffff810b65b1 task_work_add ([kernel.kallsyms]) ls 1843 1 branches: ffffffff810b65c1 task_work_add ([kernel.kallsyms]) => ffffffff810bc710 kick_process ([kernel.kallsyms]) ls 1843 1 branches: ffffffff810bc725 kick_process ([kernel.kallsyms]) => ffffffff810bc742 kick_process ([kernel.kallsyms]) ls 1843 1 branches: ffffffff810bc742 kick_process ([kernel.kallsyms]) => ffffffff810b65c6 task_work_add ([kernel.kallsyms]) ls 1843 1 branches: ffffffff810b65c9 task_work_add ([kernel.kallsyms]) => ffffffff81214311 fput ([kernel.kallsyms]) Signed-off-by: -
Dean Nelson authored
Running the following perf-stat command on an arm64 system produces the following result... [root@aarch64 ~]# perf stat -e kmem:mm_page_alloc -a sleep 1 Warning: [kmem:mm_page_alloc] function sizeof not defined Warning: Error: expected type 4 but read 0 Segmentation fault [root@aarch64 ~]# The second warning was a result of the first warning not stopping processing after it detected the issue. That is, code that found the issue reported the first problem, but because it did not exit out of the functions smoothly, it caused the other warning to appear and not only that, it later caused the SIGSEGV. Signed-off-by:Dean Nelson <dnelson@redhat.com> Reviewed-by:
Namhyung Kim <namhyung@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20150820151632.13927.13791.email-sent-by-dnelson@tealSigned-off-by:
-
Adrian Hunter authored
Events that don't sample the timestamp have a timestamp value of -1. Intel PT processing wasn't taking that into account. This is particularly noticeable with Intel BTS because timestamps are not requested by default. Then, if the conversion of -1 to TSC results in a small number, the processing is unaffected. However if the conversion results in a big number, then the data is processed prematurely before relevant sideband data like mmap events, which in turn results in samples with unknown dsos. Commiter note: Since BTS wasn't upstream, I split the patch to fold the BTS part with the patch introducing it, to avoid having this bug in the commit history. PT was already upstream, so this patch contains that part. Signed-off-by:
-
Adrian Hunter authored
The "/proc/kcore requires CAP_SYS_RAWIO" message comes up all the time for 'perf script' if vmlinux is not found and the user isn't root, even when the kernel is not being traced and even though the message is only really relevant for annotation. Change it to pr_debug and instead put a note in the message displayed if annotation is not possible. Also, the file being accessed might not be /proc/kcore. Tools can be directed to a different location using the --kallsyms option in which case kcore is expected to be in the same directory. Adjust the message so it is not misleading in that case. Signed-off-by:
-
Adrian Hunter authored
Patch "perf script: Don't assume evsel position of tracking events" changed 'perf script' to use 'perf_evlist__id2evsel()'. That results in a segfault if there is more than 1 event and there are synthesized mmap events e.g. $ perf record -e cycles,instructions -p$$ sleep 1 $ perf script --show-mmap-events Segmentation fault (core dumped) That happens because these synthesized events have an 'id' of zero which does not match any 'evsel'. Currently, these synthesized events use the sample type of the first evsel. Change 'perf_evlist__id2evsel()' to reflect that which also makes it consistent with 'perf_evlist__event2evsel()'. Signed-off-by:
-
Ingo Molnar authored
The new MSR PMU driver made use of rdtsc() which does not exist (yet) in this tree: arch/x86/kernel/cpu/perf_event_msr.c:91:3: error: implicit declaration of function 'rdtsc' Use the old rdtscll() primitive for now. Reported-by:
kbuild test robot <fengguang.wu@intel.com> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by:
-
- 20 Aug, 2015 1 commit
-
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo: - Support Intel PT in several tools, enabling the use of the processor trace feature introduced in Intel Broadwell processors: (Adrian Hunter) # dmesg | grep Performance # [0.188477] Performance Events: PEBS fmt2+, 16-deep LBR, Broadwell events, full-width counters, Intel PMU driver. # perf record -e intel_pt//u -a sleep 1 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.216 MB perf.data ] # perf script # then navigate in the tool output to some area, like this one: 184 1030 dl_main (/usr/lib64/ld-2.17.so) => 7f21ba661440 dl_main (/usr/lib64/ld-2.17.so) 185 1457 dl_main (/usr/lib64/ld-2.17.so) => 7f21ba669f10 _dl_new_object (/usr/lib64/ld-2.17.so) 186 9f37 _dl_new_object (/usr/lib64/ld-2.17.so) => 7f21ba677b90 strlen (/usr/lib64/ld-2.17.so) 187 7ba3 strlen (/usr/lib64/ld-2.17.so) => 7f21ba677c75 strlen (/usr/lib64/ld-2.17.so) 188 7c78 strlen (/usr/lib64/ld-2.17.so) => 7f21ba669f3c _dl_new_object (/usr/lib64/ld-2.17.so) 189 9f8a _dl_new_object (/usr/lib64/ld-2.17.so) => 7f21ba65fab0 calloc@plt (/usr/lib64/ld-2.17.so) 190 fab0 calloc@plt (/usr/lib64/ld-2.17.so) => 7f21ba675e70 calloc (/usr/lib64/ld-2.17.so) 191 5e87 calloc (/usr/lib64/ld-2.17.so) => 7f21ba65fa90 malloc@plt (/usr/lib64/ld-2.17.so) 192 fa90 malloc@plt (/usr/lib64/ld-2.17.so) => 7f21ba675e60 malloc (/usr/lib64/ld-2.17.so) 193 5e68 malloc (/usr/lib64/ld-2.17.so) => 7f21ba65fa80 __libc_memalign@plt (/usr/lib64/ld-2.17.so) 194 fa80 __libc_memalign@plt (/usr/lib64/ld-2.17.so) => 7f21ba675d50 __libc_memalign (/usr/lib64/ld-2.17.so) 195 5d63 __libc_memalign (/usr/lib64/ld-2.17.so) => 7f21ba675e20 __libc_memalign (/usr/lib64/ld-2.17.so) 196 5e40 __libc_memalign (/usr/lib64/ld-2.17.so) => 7f21ba675d73 __libc_memalign (/usr/lib64/ld-2.17.so) 197 5d97 __libc_memalign (/usr/lib64/ld-2.17.so) => 7f21ba675e18 __libc_memalign (/usr/lib64/ld-2.17.so) 198 5e1e __libc_memalign (/usr/lib64/ld-2.17.so) => 7f21ba675df9 __libc_memalign (/usr/lib64/ld-2.17.so) 199 5e10 __libc_memalign (/usr/lib64/ld-2.17.so) => 7f21ba669f8f _dl_new_object (/usr/lib64/ld-2.17.so) 200 9fc2 _dl_new_object (/usr/lib64/ld-2.17.so) => 7f21ba678e70 memcpy (/usr/lib64/ld-2.17.so) 201 8e8c memcpy (/usr/lib64/ld-2.17.so) => 7f21ba678ea0 memcpy (/usr/lib64/ld-2.17.so) - Fix annotation of vdso (Adrian Hunter) - Fix DWARF callchains in 'perf script' (Jiri Olsa) - Fix adding probes in kernel syscalls and listing which variables can be collected at kernel syscall function lines (Masami Hiramatsu) Build Fixes: - Fix 32-bit compilation error in util/annotate.c (Adrian Hunter) - Support static linking with libdw on Fedora 22 (Andi Kleen) Infrastructure changes: - Add a helper function to probe whether cpu-wide tracing is possible (Adrian Hunter) - Move vfs_getname storage to per thread area in 'perf trace' (Arnaldo Carvalho de Melo) Signed-off-by:
-
