- 14 Aug, 2014 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/probe fixes and improvements from Arnaldo Carvalho de Melo: User visible changes: * Do not show +/- callchain expansion when there are no childs (top/report) (Namhyung Kim) * Fix -z and add respective 'z' hotkey to zero samples before refresh in 'perf top' (Namhyung Kim) * Capability probing fixes, improving the detection of kernel features for non-priviledged users (Adrian Hunter) * Add beautifier for mremap flags param in 'trace' (Alex Snast) * Fix --list and --del options to show events when just uprobes is enabled (Masami Hiramatsu) * perf script: Allow callchains if any event samples them * Don't look for kernel idle symbols in all DSOs in 'perf top' (Arnaldo Carvalho de Melo) * Add cpu_startup_entry to the list of kernel idle symbols (Arnaldo Carvalho de Melo) * 'perf top' terminal output fixes (Jiri Olsa) * Fix stdin handling for 'perf kvm stat live' (Jiri Olsa) * Fix missing label symbols (Adrian Hunter) * Don't demangle C++ parameters and such by default, only in --verbose mode (Namhyung Kim) * Set proper sort__mode for the branch option (Naohiro Aota) * Check recorded kernel version when finding vmlinux (Namhyung Kim) Infrastructure changes: * More prep work for intel PT (Adrian Hunter) * Fix possible memory leaks (Namhyung Kim) * Fix a memory leak in vmlinux_path__init() (Namhyung Kim) Signed-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
- 13 Aug, 2014 39 commits
-
-
Adrian Hunter authored
Add a function to enable a specific event within a specific perf event buffer. Signed-off-by:
Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/r/1406786474-9306-14-git-send-email-adrian.hunter@intel.comSigned-off-by:
-
Adrian Hunter authored
The flags will be used to export branch type and transaction status. insn_len is preparation for pairing calls and returns because the return address equals the call address plus the instruction length (insn_len). Signed-off-by:
-
Adrian Hunter authored
Add machine__thread_exec_comm() to return the comm that matches the last exec, if the comm_exec flag is present, or the last comm otherwise. Signed-off-by:
-
Adrian Hunter authored
For grouping together all the data from a single execution, which is needed for pairing calls and returns e.g. any outstanding calls when a process exec's will never return. Signed-off-by:
-
Adrian Hunter authored
The Python script API repeatedly uses the same lines of code to get and call objects. Make that into helper functions instead. A side-effect is that some reference counting bugs disappear because the new call_object() function always decrements the reference count of 'retval'. Signed-off-by:
-
Adrian Hunter authored
perf script was not displaying callchains if any selected event did not have PERF_SAMPLE_CALLCHAIN. Change this to disable callchains only if all selected events do not have PERF_SAMPLE_CALLCHAIN. Signed-off-by:
-
Adrian Hunter authored
Add a function to peek at other events in the event stream. Signed-off-by:
-
Adrian Hunter authored
Add a function to change which event is used to track mmap, comm and task events. This is needed with Instruction Tracing because the Instruction Tracing event must come first but cannot be used for tracking because it will be disabled under some circumstances. Signed-off-by:
-
Adrian Hunter authored
Add an option to cause a selected event to be opened always without a pid when configured by perf_evsel__config(). This is needed when using the sched_switch tracepoint to follow object code execution. sched_switch occurs before the task switch and so it cannot record it in a context limited to that task. Note that also means that sched_switch is useless when capturing data per-thread, as is the 'context-switches' software event for the same reason. Signed-off-by:
-
Adrian Hunter authored
Label symbols are missing because elf_sec__is_a() fails to find the section because the section strings do not match the section headers because the sections headers are from the 'runtime' object and the sections strings are from the 'symbol source' object. Fix by getting the section strings from the 'runtime' object so that they match the section headers. Signed-off-by:
-
Namhyung Kim authored
Some C++ symbols have very long name and they make column length longer. Most of them are about parameters including templates and we can ignore such info most of time IMHO. This patch passes DMGL_NO_OPTS by default when calling bfd_demangle(). One can still see full symbols with -v/--verbose option. before: JS_CallFunctionValue(JSContext*, JSObject*, JS::Value, unsigned int, JS::Value*, JS::Value*) after: JS_CallFunctionValue Signed-off-by:
Namhyung Kim <namhyung@kernel.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung.kim@lge.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1406785662-5534-9-git-send-email-namhyung@kernel.orgSigned-off-by:
-
Namhyung Kim authored
The perf top TUI lacks 'z' key support to toggle sample zeroing. Add it. Reported-by:
Stephane Eranian <eranian@google.com> Tested-by:
-
Namhyung Kim authored
The current -z option does almost nothing. It doesn't zero the existing samples so that we can see profiles of exited process after last refresh. It seems it only affects annotation. This patch clears existing entries before processing if -z option is given. For this original decaying logic also moved before processing. Reported-by:
-
naota@elisp.net authored
When you specify "--branch-stack"("-b" for short) or "--no-branch-stack", "branch_mode" variable is set to 1 or 0 respectively. However, the code is just checking if the variable is -1 or not, ignoring "branch_mode == 1" case. Thus "perf report -b" dose not show its result with the branch sorted mode. This patch fix the problem. Signed-off-by:Naohiro Aota <naota@elisp.net> Acked-by:
-
Namhyung Kim authored
The currently when perf TUI report shows callchain, the first level chains have bogus '+' sign even though only the last one has children. Since they are on a single line of the chain, toggling intermediate entries has no effect. Fix it to show '+' sign at the last entry only. Note that non-first level callchain entries don't have this problem. Before: --------------------------------------------------------------------------- Children Self Command Shared Object Symbols - 40.70% 0.00% swapper [kernel.kallsyms] [k] cpuidle_wrap_enter + cpuidle_wrap_enter + cpuidle_enter_tk + cpuidle_idle_call + cpu_idle After: --------------------------------------------------------------------------- Children Self Command Shared Object Symbols - 40.70% 0.00% swapper [kernel.kallsyms] [k] cpuidle_wrap_enter cpuidle_wrap_enter cpuidle_enter_tk cpuidle_idle_call + cpu_idle Signed-off-by: -
Namhyung Kim authored
Currently vmlinux_path__init() only tries to find vmlinux file from current directory, /boot and some canonical directories with version number of the running kernel. This can be a problem when reporting old data recorded on a kernel version not running currently. We can use --symfs option for this but it's annoying for user to do it always. As we already have the info in the perf.data file, it can be changed to use it for the search automatically. Before: $ perf report ... # Samples: 4K of event 'cpu-clock' # Event count (approx.): 1067250000 # # Overhead Command Shared Object Symbol # ........ .......... ................. .............................. 71.87% swapper [kernel.kallsyms] [k] recover_probed_instruction After: # Overhead Command Shared Object Symbol # ........ .......... ................. .................... 71.87% swapper [kernel.kallsyms] [k] native_safe_halt This requires to change signature of symbol__init() to receive struct perf_session_env *. Reported-by:Minchan Kim <minchan@kernel.org> Signed-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
Jiri Olsa <jolsa@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Namhyung Kim <namhyung.kim@lge.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1407825645-24586-12-git-send-email-namhyung@kernel.orgSigned-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Acked-by:
-
Namhyung Kim authored
This is a preparation of fixing dso__load_kernel_sym(). It needs a session info before calling symbol__init(). Signed-off-by:
-
Namhyung Kim authored
When uname() failed, it should free vmlinux_path. Acked-by:
-
Namhyung Kim authored
Some paths in perf script don't call perf_session__delete() after creating a new session. Signed-off-by:
-
Adrian Hunter authored
When doing a system-wide trace with Intel PT, the jump label set up as a result of probing CLOEXEC gets reset while the trace is running. That causes an Intel PT decoding error because the object code (obtained from /proc/kcore) does not match the running code at that point. While we can't expect there never to be jump label changes, we can avoid cases that the perf tool itself creates. The problem is avoided by first trying a cpu-wide event (pid = -1) for probing the PERF_FLAG_FD_CLOEXEC flag and falling back to an event for the current process (pid = 0). Signed-off-by:
-
Adrian Hunter authored
Fall back to probing with the current pid if cpu-wide probing fails. This primarily affects the setting of comm_exec flag when the user is un-privileged and /proc/sys/kernel/perf_event_paranoid > 0. The change to comm_exec can be observed by using -vv with perf record and a kernel that supports comm_exec. Signed-off-by:
-
Adrian Hunter authored
When probing the kernel API the kernel should be excluded otherwise the probe will fail for users with insufficient privilege to profile the kernel. Signed-off-by:
-
Adrian Hunter authored
With /proc/sys/kernel/perf_event_paranoid set to 2, the probe of PERF_FLAG_FD_CLOEXEC would fail. Fix by excluding kernel profiling from the probe event. Signed-off-by:
-
Alex Snast authored
~/devel/kernel/tools/perf(branch:master*) » sudo ./perf trace ~/mremap_test 0.543 ( 0.003 ms): mprotect(start: 0x600000, len: 4096, prot: READ ) = 0 0.550 ( 0.003 ms): mprotect(start: 0x7f441260d000, len: 4096, prot: READ) = 0 0.561 ( 0.010 ms): munmap(addr: 0x7f44125e2000, len: 165572 ) = 0 0.595 ( 0.012 ms): mmap(len: 12288, prot: READ|WRITE, flags: SHARED|ANONYMOUS|LOCKED, fd: -1) = 0x12608000 0.603 ( 0.006 ms): mremap(addr: 0x7f4412608000, old_len: 4096, new_len: 4096, flags: MAYMOVE|FIXED, new_addr: 0x7f16da295000) = 0xda295000 0.608 ( 0.003 ms): mremap(addr: 0x7f441260a000, old_len: 4096, new_len: 4096, flags: MAYMOVE|FIXED, new_addr: 0x7f16da297000) = 0xda297000 0.612 ( 0.003 ms): mremap(addr: 0x7f4412609000, old_len: 4096, new_len: 4096, flags: MAYMOVE|FIXED, new_addr: 0x7f16da296000) = 0xda296000 0.619 ( 0.000 ms): exit_group( Signed-off-by:Alex Snast <asnast@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1407944560-26924-1-git-send-email-asnast@gmail.comSigned-off-by:
-
Masami Hiramatsu authored
Current perf probe --del doesn't work if only CONFIG_UPROBE_EVENTS=y because it aborts when it fails to open kprobe_events file before checking uprobe_events file. This fixes --del option to delete dynamic events if it can open either kprobe_events or uprobe_events. Only if it failed to open both of them, it shows an error message and aborts. Without this patch, if we run perf probe -d on the kernel configured with CONFIG_KPROBE_EVENTS=n and CONFIG_UPROBE_EVENTS=y, # perf probe -d \* kprobe_events file does not exist - please rebuild kernel with CONFIG_KPROBE_EVENTS. Error: Failed to delete events. With this patch, # perf probe -d \* Removed event: probe_perf:alloc_event Changes in v2: - Use strerror_r instead of strerror. Signed-off-by:Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Naohiro Aota <naota@elisp.net> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20140813161250.26440.24028.stgit@kbuild-fedora.novalocalSigned-off-by:
-
Masami Hiramatsu authored
Current perf probe --list doesn't work if only CONFIG_UPROBE_EVENTS=y because it aborts when it fails to open kprobe_events file before checking uprobe_events file. This fixes --list option to show dynamic events if it can open either kprobe_events or uprobe_events. Only if it failed to open both of them, it shows an error message and aborts. Without this patch, if we run perf probe -l on the kernel configured with CONFIG_KPROBE_EVENTS=n and CONFIG_UPROBE_EVENTS=y, # perf probe -l /sys/kernel/debug/tracing/kprobe_events file does not exist - please rebuild ker Error: Failed to show event list. With this patch, # perf probe -l probe_perf:alloc_event (on alloc_event@lib/traceevent/event-parse.c in /home/fedora/ksrc/linux-3/tools/perf/perf) Changes in v2: - Use strerror_r instead of strerror. Reported-by: -
Jiri Olsa authored
Currently the initial ~(ICANON | ECHO) terminal mode is not set, so we dont get stdin data until we press ENTER. Fixing this by early setting of the ~(ICANON | ECHO) mode and leaving this mode for whole life of the command, because canonical mode is not needed. Signed-off-by:
-
Jiri Olsa authored
The TUI code setup standard signals handling, while the stdio display code does not. This leads to premature termination of display thread when signal is received and leaving terminal in wrong state. Also adding terminal cleanup at the end of display thread, to ensure we get the old terminal state in case of signal interruption. Signed-off-by:
-
Jiri Olsa authored
Adding set_term_quiet_input helper to set the terminal quiet, out from 'perf top', used in following patches in 'perf kvm'. Signed-off-by:
-
Jiri Olsa authored
We create the display thread, but never join it. It gives the display thread a chance to quit and cleanup properly. Signed-off-by:
-
