linux/tools/perf, branch v3.12.14

perf kvm: Fix kvm report without guestmount.

2014-02-06T19:22:18Z

commit ad85ace07a05062ef6b59c35a5e80b6eaee1eee6 upstream. Currently, if we use perf kvm --guestkallsyms --guestmodules report, we can not get the perf information from perf data file. All sample are shown as unknown. Reproducing steps: # perf kvm --guestkallsyms /tmp/kallsyms --guestmodules /tmp/modules record -a sleep 1 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.624 MB perf.data.guest (~27260 samples) ] # perf kvm --guestkallsyms /tmp/kallsyms --guestmodules /tmp/modules report |grep % 100.00% [guest/6471] [unknown] [g] 0xffffffff8164f330 This bug was introduced by 207b57926 (perf kvm: Fix regression with guest machine creation). In original code, it uses perf_session__find_machine(), it means we deliver symbol to machine which has the same pid, if no machine found, deliver it to *default* guest. But if we use perf_session__findnew_machine() here, if no machine was found, new machine with pid will be built and added. Then the default guest which with pid == 0 will never get a symbol. And because the new machine initialized here has no kernel map created, the symbol delivered to it will be marked as "unknown". This patch here is to revert commit 207b57926 and fix the SEGFAULT bug in another way. Verification steps: # ./perf kvm --guestkallsyms /home/kallsyms --guestmodules /home/modules record -a sleep 1 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.651 MB perf.data.guest (~28437 samples) ] # ./perf kvm --guestkallsyms /home/kallsyms --guestmodules /home/modules report |grep % 22.64% :6471 [guest.kernel.kallsyms] [g] update_rq_clock.part.70 19.99% :6471 [guest.kernel.kallsyms] [g] d_free 18.46% :6471 [guest.kernel.kallsyms] [g] bio_phys_segments 16.25% :6471 [guest.kernel.kallsyms] [g] dequeue_task 12.78% :6471 [guest.kernel.kallsyms] [g] __switch_to 7.91% :6471 [guest.kernel.kallsyms] [g] scheduler_tick 1.75% :6471 [guest.kernel.kallsyms] [g] native_apic_mem_write 0.21% :6471 [guest.kernel.kallsyms] [g] apic_timer_interrupt Signed-off-by: Dongsheng Yang Acked-by: David Ahern Cc: David Ahern Link: http://lkml.kernel.org/r/1387564907-3045-1-git-send-email-yangds.fnst@cn.fujitsu.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Greg Kroah-Hartman

perf tools: Synthesize anon MMAP records again

2013-12-04T19:05:10Z

commit 9d4ecc8893832337daf241236841db966fa53489 upstream. When introducing the PERF_RECORD_MMAP2 in: 5c5e854bc760 perf tools: Add attr->mmap2 support A check for the number of entries parsed by sscanf was introduced that assumed all of the 8 fields needed to be correctly parsed so that particular /proc/pid/maps line would be considered synthesizable. That broke anon records synthesizing, as it doesn't have the 'execname' field. Fix it by keeping the sscanf return check, changing it to not require that the 'execname' variable be parsed, so that the preexisting logic can kick in and set it to '//anon'. This should get things like JIT profiling working again. Signed-off-by: Don Zickus Cc: Bill Gray Cc: Jiri Olsa Cc: Joe Mario Cc: Richard Fowles Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-bo4akalno7579shpz29u867j@git.kernel.org [ commit log message is mine, dzickus reported the problem with a patch ] Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Greg Kroah-Hartman

perf tools: Remove cast of non-variadic function to variadic

2013-12-04T19:05:10Z

commit 53805eca3d89b095062c11a6798689bb0af09216 upstream. The 4fb71074a570 (perf ui/hist: Consolidate hpp helpers) cset introduced a cast of percent_color_snprintf to a function pointer type with varargs. Change percent_color_snprintf to be variadic and remove the cast. The symptom of this was all percentages being reported as 0.00% in perf report --stdio output on the armhf arch. Signed-off-by: Michael Hudson-Doyle Acked-by: Namhyung Kim Acked-by: Will Deacon Cc: Jean Pihet Cc: Jiri Olsa Cc: Will Deacon Link: http://lkml.kernel.org/r/87zjppvw7y.fsf@canonical.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Greg Kroah-Hartman

perf tools: Fixup mmap event consumption

2013-10-28T19:06:00Z

The tail position of the event buffer should only be modified after actually use that event. If not the event buffer could be invalid before use, and segment fault occurs when invoking perf top -G. Signed-off-by: Zhouyi Zhou Cc: David Ahern Cc: Zhouyi Zhou Link: http://lkml.kernel.org/r/1382600613-32177-1-git-send-email-zhouzhouyi@gmail.com [ Simplified the logic using exit gotos and renamed write_tail method to mmap_consume ] Signed-off-by: Arnaldo Carvalho de Melo

perf top: Split -G and --call-graph

2013-10-28T19:06:00Z

Splitting -G and --call-graph for record command, so we could use '-G' with no option. The '-G' option now takes NO argument and enables the configured unwind method, which is currently the frame pointers method. It will be possible to configure unwind method via config file in upcoming patches. All current '-G' arguments is overtaken by --call-graph option. NOTE: The documentation for top --call-graph option was wrongly copied from report command. Signed-off-by: Jiri Olsa Tested-by: David Ahern Tested-by: Ingo Molnar Reviewed-by: David Ahern Acked-by: Ingo Molnar Cc: Adrian Hunter Cc: Andi Kleen Cc: Corey Ashford Cc: David Ahern Cc: Ingo Molnar Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1382797536-32303-3-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo

perf record: Split -g and --call-graph

2013-10-28T19:05:59Z

Splitting -g and --call-graph for record command, so we could use '-g' with no option. The '-g' option now takes NO argument and enables the configured unwind method, which is currently the frame pointers method. It will be possible to configure unwind method via config file in upcoming patches. All current '-g' arguments is overtaken by --call-graph option. Signed-off-by: Jiri Olsa Tested-by: David Ahern Tested-by: Ingo Molnar Reviewed-by: David Ahern Acked-by: Ingo Molnar Cc: Adrian Hunter Cc: Andi Kleen Cc: Corey Ashford Cc: David Ahern Cc: Ingo Molnar Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1382797536-32303-2-git-send-email-jolsa@redhat.com [ reordered -g/--call-graph on --help and expanded the man page according to comments by David Ahern and Namhyung Kim ] Signed-off-by: Arnaldo Carvalho de Melo

perf hists: Add color overhead for stdio output buffer

2013-10-28T19:05:59Z

Following commit tightened up the buffer size for output to strict width of used format columns: 99cf666 perf hists: Fix formatting of long symbol names This works fine until you hit color overhead output which places extra bytes into output buffer. We need to account for color overhead in the output buffer. Adding maximum color byte size to the output buffer size. Signed-off-by: Jiri Olsa Cc: Adrian Hunter Cc: Corey Ashford Cc: David Ahern Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1382700293-1803-1-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo

perf tools: Fix up /proc/PID/maps parsing

2013-10-28T12:38:12Z

When introducing support for MMAP2 we considered more parts of each map representation in /proc/PID/maps, and when disabling it we forgot to reduce the number of expected parsed/assigned entries in the sscanf call, fix it to expect the right number of desired fields, 5. Reported-by: Markus Trippelsdorf Based-on-a-patch-by: Markus Trippelsdorf Cc: Adrian Hunter Cc: David Ahern Cc: Frederic Weisbecker Cc: Jiri Olsa Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-vrbo1wik997ahjzl1chm3bdm@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo

perf script python: Fix mem leak due to missing Py_DECREFs on dict entries

2013-10-24T13:16:54Z

We are using the Python scripting interface in perf to extract kernel events relevant for performance analysis of HPC codes. We noticed that the "perf script" call allocates a significant amount of memory (in the order of several 100 MiB) during it's run, e.g. 125 MiB for a 25 MiB input file: $> perf record -o perf.data -a -R -g fp \ -e power:cpu_frequency -e sched:sched_switch \ -e sched:sched_migrate_task -e sched:sched_process_exit \ -e sched:sched_process_fork -e sched:sched_process_exec \ -e cycles -m 4096 --freq 4000 $> /usr/bin/time perf script -i perf.data -s dummy_script.py 0.84user 0.13system 0:01.92elapsed 51%CPU (0avgtext+0avgdata 125532maxresident)k 73072inputs+0outputs (57major+33086minor)pagefaults 0swaps Upon further investigation using the valgrind massif tool, we noticed that Python objects that are created in trace-event-python.c via PyString_FromString*() (and their Integer and Long counterparts) are never free'd. The reason for this seem to be missing Py_DECREF calls on the objects that are returned by these functions and stored in the Python dictionaries. The Python dictionaries do not steal references (as opposed to Python tuples and lists) but instead add their own reference. Hence, the reference that is returned by these object creation functions is never released and the memory is leaked. (see [1,2]) The attached patch fixes this by wrapping all relevant calls to PyDict_SetItemString() and decrementing the reference counter immediately after the Python function call. This reduces the allocated memory to a reasonable amount: $> /usr/bin/time perf script -i perf.data -s dummy_script.py 0.73user 0.05system 0:00.79elapsed 99%CPU (0avgtext+0avgdata 49132maxresident)k 0inputs+0outputs (0major+14045minor)pagefaults 0swaps For comparison, with a 120 MiB input file the memory consumption reported by time drops from almost 600 MiB to 146 MiB. The patch has been tested using Linux 3.8.2 with Python 2.7.4 and Linux 3.11.6 with Python 2.7.5. Please let me know if you need any further information. [1] http://docs.python.org/2/c-api/tuple.html#PyTuple_SetItem [2] http://docs.python.org/2/c-api/dict.html#PyDict_SetItemString Signed-off-by: Joseph Schuchart Reviewed-by: Tom Zanussi Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Tom Zanussi Link: http://lkml.kernel.org/r/1381468543-25334-4-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo

perf: Disable PERF_RECORD_MMAP2 support

2013-10-17T19:27:14Z

For now, we disable the extended MMAP record support (MMAP2). We have identified cases where it would not report the correct mapping information, clone(VM_CLONE) but with separate pids. We will revisit the support once we find a solution for this case. The patch changes the kernel to return EINVAL if attr->mmap2 is set. The patch also modifies the perf tool to use regular PERF_RECORD_MMAP for synthetic events and it also prevents the tool from requesting attr->mmap2 mode because the kernel would reject it. The support will be revisited once the kenrel interface is updated. In V2, we reduce the patch to the strict minimum. In V3, we avoid calling perf_event_open() with mmap2 set because we know it will fail and require fallback retry. Signed-off-by: Stephane Eranian Cc: Andi Kleen Cc: David Ahern Cc: Ingo Molnar Cc: Jiri Olsa Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/20131017173215.GA8820@quad Signed-off-by: Arnaldo Carvalho de Melo