aboutsummaryrefslogtreecommitdiff
path: root/tools
AgeCommit message (Collapse)AuthorFilesLines
2016-05-06perf trace: Do not print raw args list for syscalls with no argsArnaldo Carvalho de Melo1-1/+6
The test to check if the arg format had been read from the syscall:sys_enter_name/format file was looking at the list of non-commom fields, and if that is empty, it would think it had failed to read it, because it doesn't exist, for instance, for the clone() syscall. So instead before dumping the raw syscall args list check IS_ERR(sc->tp_format), if that is true, then an attempt was made to read the format file and failed, in which case dump the raw arg list values. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf evlist: Rename variable in perf_mmap__read()Wang Nan1-14/+15
In perf_mmap__read(), give better names to pointers. Original name 'old' and 'head' directly related to pointers in ring buffer control page. For backward ring buffer, the meaning of 'head' point is not 'the first byte of free space', but 'the first byte of the last record'. To reduce confusion, rename 'old' to 'start', 'head' to 'end'. 'start' -> 'end' is the direction the records should be read from. Change parameter order. Change 'overwrite' to 'check_messup'. When reading from 'head', no need to check messup for for backward ring buffer. Signed-off-by: Wang Nan <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Zefan Li <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf evlist: Extract perf_mmap__read()Wang Nan1-15/+24
Extract event reader from perf_evlist__mmap_read() to perf__mmap_read(). Future commit will feed it with manually computed 'head' and 'old' pointers. Signed-off-by: Wang Nan <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Zefan Li <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf symbols: Fix kallsyms perf test on ppc64leNaveen N. Rao5-15/+31
ppc64le functions have a Global Entry Point (GEP) and a Local Entry Point (LEP). While placing a probe, we always prefer the LEP since it catches function calls through both the GEP and the LEP. In order to do this, we fixup the function entry points during elf symbol table lookup to point to the LEPs. This works, but breaks 'perf test kallsyms' since the symbols loaded from the symbol table (pointing to the LEP) do not match the symbols in kallsyms. To fix this, we do not adjust all the symbols during symbol table load. Instead, we note down st_other in a newly introduced arch-specific member of perf symbol structure, and later use this to adjust the probe trace point. Reported-by: Michael Ellerman <[email protected]> Signed-off-by: Naveen N. Rao <[email protected]> Acked-by: Ananth N Mavinakayanahalli <[email protected]> Acked-by: Balbir Singh <[email protected]> Cc: Mark Wielaard <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Thiago Jung Bauermann <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/6be7c2b17e370100c2f79dd444509df7929bdd3e.1460451721.git.naveen.n.rao@linux.vnet.ibm.com Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf powerpc: Fix kprobe and kretprobe handling with kallsyms on ppc64leNaveen N. Rao1-5/+14
So far, we used to treat probe point offsets as being offset from the LEP. However, userspace applications (objdump/readelf) always show disassembly and offsets from the function GEP. This is confusing to the user as we will end up probing at an address different from what the user expects when looking at the function disassembly with readelf/objdump. Fix this by changing how we modify probe address with perf. If only the function name is provided, we assume the user needs the LEP. Otherwise, if an offset is specified, we assume that the user knows the exact address to probe based on function disassembly, and so we just place the probe from the GEP offset. Finally, kretprobe was also broken with kallsyms as we were trying to specify an offset. This patch also fixes that issue. Reported-by: Ananth N Mavinakayanahalli <[email protected]> Signed-off-by: Naveen N. Rao <[email protected]> Acked-by: Balbir Singh <[email protected]> Cc: Mark Wielaard <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Michael Ellerman <[email protected]> Cc: Thiago Jung Bauermann <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/75df860aad8216bf4b9bcd10c6351ecc0e3dee54.1460451721.git.naveen.n.rao@linux.vnet.ibm.com Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf hists: Move sort__has_comm into struct perf_hpp_listJiri Olsa4-5/+6
Now we have sort dimensions private for struct hists, we need to make dimension booleans hists specific as well. Moving sort__has_comm into struct perf_hpp_list. Signed-off-by: Jiri Olsa <[email protected]> Cc: David Ahern <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf hists: Move sort__has_thread into struct perf_hpp_listJiri Olsa4-9/+8
Now we have sort dimensions private for struct hists, we need to make dimension booleans hists specific as well. Moving sort__has_thread into struct perf_hpp_list. Signed-off-by: Jiri Olsa <[email protected]> Cc: David Ahern <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf hists: Move sort__has_socket into struct perf_hpp_listJiri Olsa5-6/+5
Now we have sort dimensions private for struct hists, we need to make dimension booleans hists specific as well. Moving sort__has_socket into struct perf_hpp_list. Signed-off-by: Jiri Olsa <[email protected]> Cc: David Ahern <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf hists: Move sort__has_dso into struct perf_hpp_listJiri Olsa4-9/+8
Now we have sort dimensions private for struct hists, we need to make dimension booleans hists specific as well. Moving sort__has_dso into struct perf_hpp_list. Signed-off-by: Jiri Olsa <[email protected]> Cc: David Ahern <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf hists: Move sort__has_sym into struct perf_hpp_listJiri Olsa8-14/+13
Now we have sort dimensions private for struct hists, we need to make dimension booleans hists specific as well. Moving sort__has_sym into struct perf_hpp_list. Signed-off-by: Jiri Olsa <[email protected]> Cc: David Ahern <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf hists: Move sort__has_parent into struct perf_hpp_listJiri Olsa6-7/+6
Now we have sort dimensions private for struct hists, we need to make dimension booleans hists specific as well. Moving sort__has_parent into struct perf_hpp_list. Signed-off-by: Jiri Olsa <[email protected]> Cc: David Ahern <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf hists: Move sort__need_collapse into struct perf_hpp_listJiri Olsa11-20/+22
Now we have sort dimensions private for struct hists, we need to make dimension booleans hists specific as well. Moving sort__need_collapse into struct perf_hpp_list. Adding hists__has macro to easily access this info perf struct hists object. Signed-off-by: Jiri Olsa <[email protected]> Cc: David Ahern <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf tools powerpc: Add support for generating bpf prologueNaveen N. Rao2-12/+29
Generalize existing macros to serve the purpose. Signed-off-by: Naveen N. Rao <[email protected]> Cc: Ian Munsie <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Michael Ellerman <[email protected]> Cc: Wang Nan <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/1462461799-17518-1-git-send-email-naveen.n.rao@linux.vnet.ibm.com Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf trace: Do not show the runtime_ms for a thread when not collecting itArnaldo Carvalho de Melo1-1/+5
That field is only updated when we use the "sched:sched_stat_runtime" tracepoint, and that is only done so far when we use the '--stat' command line option, without it we get just zeros, confusing the users: Without this patch: # trace -a -s sleep 1 <SNIP> qemu-system-x86 (9931), 468 events, 9.6%, 0.000 msec syscall calls total min avg max stddev (msec) (msec) (msec) (msec) (%) ---------- ------ --------- --------- --------- --------- ------ ppoll 98 982.374 0.000 10.024 29.983 12.65% write 34 0.401 0.005 0.012 0.027 5.49% ioctl 102 0.347 0.002 0.003 0.007 3.08% firefox (10871), 1856 events, 38.2%, 0.000 msec (msec) (msec) (msec) (msec) (%) ---------- ------ --------- --------- --------- --------- ------ poll 395 934.873 0.000 2.367 17.120 11.51% recvmsg 395 0.988 0.001 0.003 0.021 4.20% read 106 0.460 0.002 0.004 0.007 3.17% futex 24 0.108 0.001 0.004 0.010 10.05% mmap 2 0.041 0.016 0.021 0.026 23.92% write 6 0.027 0.004 0.004 0.005 2.52% After this patch that ', 0.000 msecs' gets suppressed when --stat is not in use. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf trace: Sort syscalls stats by msecs in --summaryArnaldo Carvalho de Melo1-10/+22
# trace -a -s sleep 1 <SNIP> Xorg (1965), 788 events, 19.0%, 0.000 msec syscall calls total min avg max stddev (msec) (msec) (msec) (msec) (%) --------------- -------- --------- --------- --------- --------- ------ select 89 731.038 0.000 8.214 175.218 36.71% ioctl 22 0.661 0.010 0.030 0.072 10.43% writev 42 0.253 0.002 0.006 0.011 5.94% recvmsg 60 0.185 0.001 0.003 0.009 5.90% setitimer 60 0.127 0.001 0.002 0.006 6.14% read 52 0.102 0.001 0.002 0.005 8.55% rt_sigprocmask 45 0.092 0.001 0.002 0.023 23.65% poll 12 0.021 0.001 0.002 0.003 7.21% epoll_wait 12 0.019 0.001 0.002 0.002 2.71% firefox (10871), 1080 events, 26.1%, 0.000 msec syscall calls total min avg max stddev (msec) (msec) (msec) (msec) (%) --------------- -------- --------- --------- --------- --------- ------ poll 240 979.562 0.000 4.082 17.132 11.33% recvmsg 240 0.532 0.001 0.002 0.007 3.69% read 60 0.303 0.003 0.005 0.029 8.50% Suggested-by: Milian Wolff <[email protected]> Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf trace: Sort summary output by number of eventsArnaldo Carvalho de Melo1-21/+28
# trace -a -s sleep 1 |& grep events | tail gmain (1733), 34 events, 1.0%, 0.000 msec hexchat (9765), 46 events, 1.4%, 0.000 msec ssh (11109), 80 events, 2.4%, 0.000 msec sleep (32631), 81 events, 2.4%, 0.000 msec qemu-system-x86 (10021), 272 events, 8.2%, 0.000 msec Xorg (1965), 322 events, 9.7%, 0.000 msec SoftwareVsyncTh (10922), 366 events, 11.1%, 0.000 msec gnome-shell (2231), 446 events, 13.5%, 0.000 msec qemu-system-x86 (9931), 468 events, 14.1%, 0.000 msec firefox (10871), 1098 events, 33.2%, 0.000 msec [root@jouet ~]# Suggested-by: Milian Wolff <[email protected]> Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf tools: Add template for generating rbtree resort classArnaldo Carvalho de Melo1-0/+149
Sometimes we want to sort an existing rbtree by a different key, introduce a template for that, that needs only to be provided the rbtree root and the number of entries in it. To do that a new rbtree will be created with extra space for each entry, where possibly pre-calculated keys will be stored to be used in the resort process and also later, when using the newly sorted rbtree. Please check the following two changesets to see it in use for resorting stats for threads and its syscalls in 'perf trace --summary'. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05perf machine: Introduce number of threads memberArnaldo Carvalho de Melo2-1/+7
To be used, for instance, for pre-allocating an rb_tree array for sorting by other keys besides the current pid one. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-05-05Merge branch 'perf/urgent' into perf/core, to pick up fixesIngo Molnar5-35/+311
Signed-off-by: Ingo Molnar <[email protected]>
2016-05-05Merge tag 'v4.6-rc6' into x86/asm, to refresh the treeIngo Molnar6-36/+312
Signed-off-by: Ingo Molnar <[email protected]>
2016-05-04signals/sigaltstack: Change SS_AUTODISARM to (1U << 31)Andy Lutomirski1-1/+1
Using bit 4 divides the space of available bits strangely. Use bit 31 instead so that we have a better chance of keeping flag and mode bits separate in the long run. Signed-off-by: Andy Lutomirski <[email protected]> Cc: Al Viro <[email protected]> Cc: Aleksa Sarai <[email protected]> Cc: Amanieu d'Antras <[email protected]> Cc: Andrea Arcangeli <[email protected]> Cc: Andrew Morton <[email protected]> Cc: Andy Lutomirski <[email protected]> Cc: Borislav Petkov <[email protected]> Cc: Brian Gerst <[email protected]> Cc: Denys Vlasenko <[email protected]> Cc: Eric W. Biederman <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: H. Peter Anvin <[email protected]> Cc: Heinrich Schuchardt <[email protected]> Cc: Jason Low <[email protected]> Cc: Josh Triplett <[email protected]> Cc: Konstantin Khlebnikov <[email protected]> Cc: Linus Torvalds <[email protected]> Cc: Oleg Nesterov <[email protected]> Cc: Palmer Dabbelt <[email protected]> Cc: Paul Moore <[email protected]> Cc: Pavel Emelyanov <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Richard Weinberger <[email protected]> Cc: Sasha Levin <[email protected]> Cc: Shuah Khan <[email protected]> Cc: Stas Sergeev <[email protected]> Cc: Tejun Heo <[email protected]> Cc: Thomas Gleixner <[email protected]> Cc: Vladimir Davydov <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/bb996508a600af14b406810c3d58fe0e0d0afe0d.1462296606.git.luto@kernel.org Signed-off-by: Ingo Molnar <[email protected]>
2016-05-04signals/sigaltstack: Report current flag bits in sigaltstack()Andy Lutomirski1-3/+16
sigaltstack()'s reported previous state uses a somewhat odd convention, but the concept of flag bits is new, and we can do the flag bits sensibly. Specifically, let's just report them directly. This will allow saving and restoring the sigaltstack state using sigaltstack() to work correctly. Signed-off-by: Andy Lutomirski <[email protected]> Cc: Al Viro <[email protected]> Cc: Amanieu d'Antras <[email protected]> Cc: Andrew Morton <[email protected]> Cc: Andy Lutomirski <[email protected]> Cc: Borislav Petkov <[email protected]> Cc: Brian Gerst <[email protected]> Cc: Denys Vlasenko <[email protected]> Cc: H. Peter Anvin <[email protected]> Cc: Linus Torvalds <[email protected]> Cc: Michal Hocko <[email protected]> Cc: Oleg Nesterov <[email protected]> Cc: Pavel Emelyanov <[email protected]> Cc: Peter Zijlstra (Intel) <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Richard Weinberger <[email protected]> Cc: Sasha Levin <[email protected]> Cc: Shuah Khan <[email protected]> Cc: Stas Sergeev <[email protected]> Cc: Thomas Gleixner <[email protected]> Cc: Vladimir Davydov <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/94b291ec9fd47741a9264851e316e158ded0b00d.1462296606.git.luto@kernel.org Signed-off-by: Ingo Molnar <[email protected]>
2016-05-04selftests/sigaltstack: Fix the sigaltstack test on old kernelsAndy Lutomirski1-7/+14
The handling for old kernels was wrong, resulting in a segfault. Fix it. Reported-by: Ingo Molnar <[email protected]> Signed-off-by: Andy Lutomirski <[email protected]> Cc: Al Viro <[email protected]> Cc: Andrew Morton <[email protected]> Cc: Andy Lutomirski <[email protected]> Cc: Borislav Petkov <[email protected]> Cc: Brian Gerst <[email protected]> Cc: Denys Vlasenko <[email protected]> Cc: H. Peter Anvin <[email protected]> Cc: Linus Torvalds <[email protected]> Cc: Oleg Nesterov <[email protected]> Cc: Pavel Emelyanov <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Shuah Khan <[email protected]> Cc: Stas Sergeev <[email protected]> Cc: Thomas Gleixner <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/f3e739bf435beeaecbd5f038f1359d2eac6d1e63.1462296606.git.luto@kernel.org Signed-off-by: Ingo Molnar <[email protected]>
2016-05-03selftests/sigaltstack: Add new testcase for ↵Stas Sergeev3-0/+165
sigaltstack(SS_ONSTACK|SS_AUTODISARM) This patch adds the test case for SS_AUTODISARM flag. The test-case tries to set SS_AUTODISARM flag and checks if the nested signal corrupts the stack after swapcontext(). Signed-off-by: Stas Sergeev <[email protected]> Cc: Al Viro <[email protected]> Cc: Andrew Morton <[email protected]> Cc: Andy Lutomirski <[email protected]> Cc: Borislav Petkov <[email protected]> Cc: Brian Gerst <[email protected]> Cc: Denys Vlasenko <[email protected]> Cc: H. Peter Anvin <[email protected]> Cc: Linus Torvalds <[email protected]> Cc: Oleg Nesterov <[email protected]> Cc: Pavel Emelyanov <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Shuah Khan <[email protected]> Cc: Thomas Gleixner <[email protected]> Cc: [email protected] Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2016-04-29selftests/x86/ldt_gdt: Test set_thread_area() deletion of an active segmentAndy Lutomirski1-0/+250
Now that set_thread_area() is supposed to give deterministic behavior when it modifies in-use segments, test it. Signed-off-by: Andy Lutomirski <[email protected]> Cc: Andy Lutomirski <[email protected]> Cc: Borislav Petkov <[email protected]> Cc: Brian Gerst <[email protected]> Cc: Denys Vlasenko <[email protected]> Cc: H. Peter Anvin <[email protected]> Cc: Linus Torvalds <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Thomas Gleixner <[email protected]> Link: http://lkml.kernel.org/r/f2bc11af1ee1a0f815ed910840cbdba06b640a20.1461698311.git.luto@kernel.org Signed-off-by: Ingo Molnar <[email protected]>
2016-04-28perf tests: Do not use sizeof on pointer typeVaishali Thakkar1-1/+1
Using sizeof on a malloced pointer type will return the wordsize which can often cause one to allocate a buffer much smaller than it is needed. So, here do not use sizeof on pointer type. Note that this has no effect on runtime because 'dsos' is a pointer to a pointer. Problem found using Coccinelle. Signed-off-by: Vaishali Thakkar <[email protected]> Cc: Alexander Shishkin <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf trace: Move msg_flags beautifier to tools/perf/trace/beauty/Arnaldo Carvalho de Melo2-61/+63
To reduce the size of builtin-trace.c. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf record: Generate tracking events for process forked by perfWang Nan1-1/+30
With 'perf record --switch-output' without -a, record__synthesize() in record__switch_output() won't generate tracking events because there's no thread_map in evlist. Which causes newly created perf.data doesn't contain map and comm information. This patch creates a fake thread_map and directly call perf_event__synthesize_thread_map() for those events. Signed-off-by: Wang Nan <[email protected]> Tested-by: Arnaldo Carvalho de Melo <[email protected]> Cc: Adrian Hunter <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Zefan Li <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: He Kuang <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf record: Disable buildid cache options by default in switch output modeWang Nan2-2/+30
The cost of buildid cache processing is high: reading all events in output perf.data, opening each elf file to read buildids then copying them into ~/.debug directory. In switch output mode, these heavy works block perf from receiving perf events for too long. Enable no-buildid and no-buildid-cache by default if --switch-output is provided. Still allow user use --no-no-buildid to explicitly enable buildid in this case. Signed-off-by: Wang Nan <[email protected]> Cc: Adrian Hunter <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Zefan Li <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: He Kuang <[email protected]> [ Updated man page ] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf record: Force enable --timestamp-filename when --switch-output is providedWang Nan2-0/+8
Without this patch, the last output doesn't have timestamp appended if --timestamp-filename is not explicitly provided. For example: # perf record -a --switch-output & [1] 11224 # kill -s SIGUSR2 11224 [ perf record: dump data: Woken up 1 times ] # [ perf record: Dump perf.data.2015122622372823 ] # fg perf record -a --switch-output ^C[ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.027 MB perf.data (540 samples) ] # ls -l total 836 -rw------- 1 root root 33256 Dec 26 22:37 perf.data <---- *Odd* -rw------- 1 root root 817156 Dec 26 22:37 perf.data.2015122622372823 Signed-off-by: Wang Nan <[email protected]> Tested-by: Arnaldo Carvalho de Melo <[email protected]> Cc: Adrian Hunter <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Zefan Li <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: He Kuang <[email protected]> [ Updated man page, that also got an entry for --timestamp-filename ] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf record: Split output into multiple files via '--switch-output'Wang Nan2-2/+46
Allow 'perf record' to split its output into multiple files. For example: # ~/perf record -a --timestamp-filename --switch-output & [1] 10763 # kill -s SIGUSR2 10763 [ perf record: dump data: Woken up 1 times ] # [ perf record: Dump perf.data.2015122622314468 ] # kill -s SIGUSR2 10763 [ perf record: dump data: Woken up 1 times ] # [ perf record: Dump perf.data.2015122622314762 ] # kill -s SIGUSR2 10763 [ perf record: dump data: Woken up 1 times ] #[ perf record: Dump perf.data.2015122622315171 ] # fg perf record -a --timestamp-filename --switch-output ^C[ perf record: Woken up 1 times to write data ] [ perf record: Dump perf.data.2015122622315513 ] [ perf record: Captured and wrote 0.014 MB perf.data.<timestamp> (296 samples) ] # ls -l total 920 -rw------- 1 root root 797692 Dec 26 22:31 perf.data.2015122622314468 -rw------- 1 root root 59960 Dec 26 22:31 perf.data.2015122622314762 -rw------- 1 root root 59912 Dec 26 22:31 perf.data.2015122622315171 -rw------- 1 root root 19220 Dec 26 22:31 perf.data.2015122622315513 Signed-off-by: Wang Nan <[email protected]> Tested-by: Arnaldo Carvalho de Melo <[email protected]> Cc: Adrian Hunter <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Zefan Li <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: He Kuang <[email protected]> [ Added man page entry, used the re-synthesize patch in this series as a fixup ] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf tools: Derive trigger class from auxtrace_snapshotWang Nan1-53/+20
auxtrace_snapshot_state matches the trigger model. Use trigger to implement it. auxtrace_snapshot_state and auxtrace_snapshot_err are absorbed. Signed-off-by: Wang Nan <[email protected]> Acked-by: Adrian Hunter <[email protected]> Cc: He Kuang <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Zefan Li <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf tools: Introduce trigger classWang Nan1-0/+94
Use 'trigger' to model operations which need to be executed when an event (a signal, for example) is observed. States and transits: OFF--(on)--> READY --(hit)--> HIT ^ | | (ready) | | \_____________/ is_hit and is_ready are two key functions to query the state of a trigger. is_hit means the event already happen; is_ready means the trigger is waiting for the event. Signed-off-by: Wang Nan <[email protected]> Acked-by: Adrian Hunter <[email protected]> Cc: He Kuang <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Zefan Li <[email protected]> Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf probe: Use strbuf for making stringsMasami Hiramatsu3-169/+93
Replace many fixed-length char array with strbuf to stringify perf_probe_event and probe_trace_event etc. Signed-off-by: Masami Hiramatsu <[email protected]> Signed-off-by: Masami Hiramatsu <[email protected]> Cc: Ananth N Mavinakayanahalli <[email protected]> Cc: Hemant Kumar <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/20160427183713.23446.97377.stgit@devbox Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf evsel: Remove two extraneous ending newlines in open_strerror()Arnaldo Carvalho de Melo1-2/+2
The error messages returned by this method should not have an ending newline, fix the two cases where it was. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28perf evsel: Handle ENOMEM for perf_event_max_stack + PERF_SAMPLE_CALLCHAINArnaldo Carvalho de Melo1-0/+8
When the kernel allows tweaking perf_event_max_stack and the event being setup has PERF_SAMPLE_CALLCHAIN in its perf_event_attr.sample_type, tell the user that tweaking /proc/sys/kernel/perf_event_max_stack may solve the problem. Before: # echo 32000 > /proc/sys/kernel/perf_event_max_stack # perf record -g usleep 1 Error: The sys_perf_event_open() syscall returned with 12 (Cannot allocate memory) for event (cycles:ppp). /bin/dmesg may provide additional information. No CONFIG_PERF_EVENTS=y kernel support configured? # After: # echo 64000 > /proc/sys/kernel/perf_event_max_stack # perf record -g usleep 1 Error: Not enough memory to setup event with callchain. Hint: Try tweaking /proc/sys/kernel/perf_event_max_stack Hint: Current value: 64000 # Suggested-by: David Ahern <[email protected]> Cc: Adrian Hunter <[email protected]> Cc: Brendan Gregg <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28bpf tools: Fix syscall argumentFlorian Fainelli1-1/+1
Coverity flagged this under CID 1354884 as a sizeof mismatch, it turns out that the argument "attr" passed to syscall should have been a pointer to attr in the first place. Reported-by: coverity (CID 1354884) Signed-off-by: Florian Fainelli <[email protected]> Acked-by: Alexei Starovoitov <[email protected]> Acked-by: Wang Nan <[email protected]> Cc: Jiri Olsa <[email protected]> Fixes: 8f9e05fb298f ("perf tools: Fix PowerPC native building") Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-28bpf tools: Remove expression with no effectFlorian Fainelli1-1/+0
Assigning "attr" to "attr" does not have any effect, but was caught by Coverity, so let's remove this. Reported-by: coverity (CID 1354720) Signed-off-by: Florian Fainelli <[email protected]> Tested-by: Wang Nan <[email protected]> Acked-by: Alexei Starovoitov <[email protected]> Cc: Jiri Olsa <[email protected]> Fixes: 1b76c13e4b36 ("bpf tools: Introduce 'bpf' library and add bpf feature check") Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-27Merge branch 'for-mingo' of ↵Ingo Molnar13-21/+476
git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu into core/rcu Pull RCU updates from Paul E. McKenney: * Documentation updates, including fixes to the design-level requirements documentation and a fixed version of the design-level data-structure documentation. These fixes include removing cartoons and getting rid of the html/htmlx duplication. * Further improvements to the new-age expedited grace periods. * Miscellaneous fixes. * Torture-test changes, including a new rcuperf module for measuring RCU grace-period performance and scalability, which is useful for the expedited-grace-period changes. Signed-off-by: Ingo Molnar <[email protected]>
2016-04-27perf tools: Set the maximum allowed stack from ↵Arnaldo Carvalho de Melo16-18/+28
/proc/sys/kernel/perf_event_max_stack There is an upper limit to what tooling considers a valid callchain, and it was tied to the hardcoded value in the kernel, PERF_MAX_STACK_DEPTH (127), now that this can be tuned via a sysctl, make it read it and use that as the upper limit, falling back to PERF_MAX_STACK_DEPTH for kernels where this sysctl isn't present. Cc: Adrian Hunter <[email protected]> Cc: Brendan Gregg <[email protected]> Cc: David Ahern <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf bench: Remove one more die() callArnaldo Carvalho de Melo1-7/+15
Propagate the error instead. Cc: David Ahern <[email protected]> Cc: Hitoshi Mitake <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf tools: Update x86's syscall_64.tbl, adding preadv2 & pwritev2Arnaldo Carvalho de Melo1-0/+2
Introduced in commit 4babf2c5efb7 ("x86: wire up preadv2 and pwritev2"). This will make 'perf trace' aware of them. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf probe: Fix module probe issue if no dwarf supportRavi Bangoria1-3/+73
Perf is not able to register probe in kernel module when dwarf supprt is not there(and so it goes for symtab). Perf passes full path of module where only module name is required which is causing the problem. This patch fixes this issue. Before applying patch: $ dpkg -s libdw-dev dpkg-query: package 'libdw-dev' is not installed and no information is... $ sudo ./perf probe -m /linux/samples/kprobes/kprobe_example.ko kprobe_init Added new event: probe:kprobe_init (on kprobe_init in /linux/samples/kprobes/kprobe_example.ko) You can now use it in all perf tools, such as: perf record -e probe:kprobe_init -aR sleep 1 $ sudo cat /sys/kernel/debug/tracing/kprobe_events p:probe/kprobe_init /linux/samples/kprobes/kprobe_example.ko:kprobe_init $ sudo ./perf record -a -e probe:kprobe_init [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.105 MB perf.data ] $ sudo ./perf script # No output here After applying patch: $ sudo ./perf probe -m /linux/samples/kprobes/kprobe_example.ko kprobe_init Added new event: probe:kprobe_init (on kprobe_init in kprobe_example) You can now use it in all perf tools, such as: perf record -e probe:kprobe_init -aR sleep 1 $ sudo cat /sys/kernel/debug/tracing/kprobe_events p:probe/kprobe_init kprobe_example:kprobe_init $ sudo ./perf record -a -e probe:kprobe_init [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.105 MB perf.data (2 samples) ] $ sudo ./perf script insmod 13990 [002] 5961.216833: probe:kprobe_init: ... insmod 13995 [002] 5962.889384: probe:kprobe_init: ... Signed-off-by: Ravi Bangoria <[email protected]> Acked-by: Masami Hiramatsu <[email protected]> Cc: Alexander Shishkin <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Naveen N. Rao <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Srikar Dronamraju <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/r/1461680741-12517-1-git-send-email-ravi.bangoria@linux.vnet.ibm.com Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf probe: Fix offline module name missmatch issueRavi Bangoria1-14/+5
Perf can add a probe on kernel module which has not been loaded yet. The current implementation finds the module name from path. But if the filename is different from the actual module name then perf fails to register a probe while loading module because of mismatch in the names. For example, samples/kobject/kobject-example.ko is loaded as kobject_example. Before applying patch: $ sudo ./perf probe -m /linux/samples/kobject/kobject-example.ko foo_show Added new event: probe:foo_show (on foo_show in kobject-example) You can now use it in all perf tools, such as: perf record -e probe:foo_show -aR sleep 1 $ cat /sys/kernel/debug/tracing/kprobe_events p:probe/foo_show kobject-example:foo_show $ insmod kobject-example.ko $ lsmod Module Size Used by kobject_example 16384 0 Generate read to /sys/kernel/kobject_example/foo while recording data with below command $ sudo ./perf record -e probe:foo_show -a [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.093 MB perf.data ] $./perf report --stdio -F overhead,comm,dso,sym Error: The perf.data.old file has no samples! After applying patch: $ sudo ./perf probe -m /linux/samples/kobject/kobject-example.ko foo_show Added new event: probe:foo_show (on foo_show in kobject_example) You can now use it in all perf tools, such as: perf record -e probe:foo_show -aR sleep 1 $ sudo cat /sys/kernel/debug/tracing/kprobe_events p:probe/foo_show kobject_example:foo_show $ insmod kobject-example.ko $ lsmod Module Size Used by kobject_example 16384 0 Generate read to /sys/kernel/kobject_example/foo while recording data with below command $ sudo ./perf record -e probe:foo_show -a [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.097 MB perf.data (8 samples) ] $ sudo ./perf report --stdio -F overhead,comm,dso,sym ... # Samples: 8 of event 'probe:foo_show' # Event count (approx.): 8 # # Overhead Command Shared Object Symbol # ........ ....... ................. ............ # 100.00% cat [kobject_example] [k] foo_show Signed-off-by: Ravi Bangoria <[email protected]> Acked-by: Masami Hiramatsu <[email protected]> Cc: Alexander Shishkin <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Naveen N. Rao <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Srikar Dronamraju <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/r/1461680741-12517-2-git-send-email-ravi.bangoria@linux.vnet.ibm.com Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf trace: Read thread's COMM from /proc when not setArnaldo Carvalho de Melo1-1/+4
We get notifications for threads that gets created while we're tracing, but for preexisting threads we may end not having synthesized them, like when tracing a 'perf trace' session that will use '--pid' to trace some other thread. And besides we should probably stop synthesizing those records and instead read thread information in a lazy way, i.e. just when we need, like done in this patch: Now the 'pid_t' argument in 'perf_event_open' gets translated to a COMM: # perf trace -e perf_event_open perf stat -e cycles -p 31601 0.027 ( 0.027 ms): perf/23393 perf_event_open(attr_uptr: 0x2fdd0d8, pid: 31601 (abrt-dump-journ), cpu: -1, group_fd: -1, flags: FD_CLOEXEC) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ = 3 ^C And in other syscalls containing pid_t without thread->comm_set at the time of the formatting. Cc: Adrian Hunter <[email protected]> Cc: Borislav Petkov <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf thread: Introduce method to set comm from /proc/pid/selfArnaldo Carvalho de Melo2-0/+21
Will be used for lazy comm loading in 'perf trace'. Cc: Adrian Hunter <[email protected]> Cc: Borislav Petkov <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26tools lib api fs: Add helper to read string from procfs fileArnaldo Carvalho de Melo2-0/+15
To read things like /proc/self/comm. Cc: Adrian Hunter <[email protected]> Cc: Borislav Petkov <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf trace: Do not beautify the 'pid' parameter as a simple integerArnaldo Carvalho de Melo1-2/+1
Leave it alone so that it ends up assigned to SCA_PID via its type, 'pid_t', that will look up the pid on the machine thread rb_tree and possibly find its COMM. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Milian Wolff <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf trace: Move perf_flags beautifier to tools/perf/trace/beauty/Arnaldo Carvalho de Melo2-44/+44
To reduce the size of builtin-trace.c. Cc: Adrian Hunter <[email protected]> Cc: David Ahern <[email protected]> Cc: Jiri Olsa <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Wang Nan <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2016-04-26perf probe: Set default kprobe group name if it is not givenMasami Hiramatsu1-3/+7
Set kprobe group name as "probe" if it is not given. Signed-off-by: Masami Hiramatsu <[email protected]> Signed-off-by: Masami Hiramatsu <[email protected]> Cc: Ananth N Mavinakayanahalli <[email protected]> Cc: Hemant Kumar <[email protected]> Cc: Namhyung Kim <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/20160426090413.11891.95640.stgit@devbox Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
Powered by cgit, Gruvboxed by Blaster4385.