tools lib traceevent: Get rid of malloc_or_die() in pevent_filter_alloc()

It returns NULL when allocation fails so the users should check the
return value from now on.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung.kim@lge.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386567251-22751-4-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf kvm: Make perf kvm diff support --guestmount.

In manpage of perf-kvm, --guestmount is supported by diff command, but
it does not work well.

This patch change the extend the checking in buildid-diff from
guestkallsyms or guestmodules to perf_guest. Then this checking can
cover the all cases perf kvm is used for.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Link: http://lkml.kernel.org/r/72857ed89642e0633f5e88f7e7abbc9645359e8e.1386368672.git.yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf kvm: Move code to generate filename for perf-kvm to function.

The code in builtin-kvm.c to generate filename for perf-kvm is useful to
other command such as builtin-diff.

This patch move the related code form builtin-kvm.c to util/util.c and
wrap them in a function named get_filename_for_perf_kvm.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Link: http://lkml.kernel.org/r/5e09a5c47e8a495e888cbdc65a6fafb2c950f529.1386368672.git.yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf annotate: Fix typo

A typo in comment of builtin-annotate.c about 'that'.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Link: http://lkml.kernel.org/r/46cb069a4ce21141057a07c0b50baa9968e3228c.1386629050.git.yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf archive: Remove duplicated 'runs' in man page

Two 'runs' here breaks the sentence in Description of 'perf archive'
command.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Link: http://lkml.kernel.org/r/78a15a9f4f500b6074a1e25917d6e8251f894628.1386629050.git.yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf kvm: Fix bug in 'stat report'

When we use perf kvm record-report, there is a bug in report subcommand.

Example:
# perf kvm stat record -a sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.678 MB perf.data.guest (~29641 samples) ]
# perf kvm stat report
failed to open perf.data: No such file or directory  (try 'perf record' first)
Initializing perf session failed

This bug was introduced by f5fc14124.

+       struct perf_data_file file = {
+               .path = input_name,
+               .mode = PERF_DATA_MODE_READ,
+       };

        kvm->tool = eops;
-       kvm->session = perf_session__new(kvm->file_name, O_RDONLY, 0, false,
-                                        &kvm->tool);
+       kvm->session = perf_session__new(&file, false, &kvm->tool);

It changed the path from kvm->file_name to input_name, this patch change the path back to
'kvm->file_name', then it works well.

Verification:
# perf kvm stat record -a sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.807 MB perf.data.guest (~35264 samples) ]
# perf kvm stat report

Analyze events for all VCPUs:

     VM-EXIT    Samples  Samples%     Time%   Min Time   Max Time         Avg time

       EPT_VIOLATION        200    32.79%     1.25%        0us    12064us     62.35us ( +-  96.74% )
       EPT_MISCONFIG        134    21.97%     0.21%        0us       35us     15.25us ( +-   4.14% )
       EXCEPTION_NMI         96    15.74%     0.02%        0us       11us      1.95us ( +-   9.81% )
 APIC_ACCESS         79    12.95%     0.02%        0us       13us      2.94us ( +-  11.20% )
 HLT         65    10.66%    98.47%        0us    16706us  15084.86us ( +-   1.89% )
      IO_INSTRUCTION         27     4.43%     0.02%        0us       29us      6.42us ( +-  15.53% )
  EXTERNAL_INTERRUPT          5     0.82%     0.01%        0us       77us     23.65us ( +-  57.90% )
 TPR_BELOW_THRESHOLD          4     0.66%     0.00%        0us        1us      1.22us ( +-   4.36% )

Total Samples:610, Total events handled time:995745.54us.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Link: http://lkml.kernel.org/r/1386632823-17539-1-git-send-email-yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf kvm: Introduce option -v for perf kvm command.

As there is no -v option for perf kvm, the all debug message for perf
kvm will nerver be printed out to user.

Example:
# perf kvm --guestmount /tmp/guestmount/ record -a
Not enough memory for reading perf file header

It is confusing message for newbies such as me. With this patch applied,
we can use -v option to get the detail.

Example:
# perf kvm --guestmount /tmp/guestmount/ record -a -v
Can't access file /tmp/guestmount//15069/proc/kallsyms
Not enough memory for reading perf file header

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Cc: David Ahern <dsahern@gmail.com>
Link: http://lkml.kernel.org/r/1386609311-23889-1-git-send-email-yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evlist: Fix mmap pages rounding to power of 2

'next_pow2()' only works for 'unsigned int' but the argument is
'unsigned long'.  Checking for values less than (1 << 31) ensures that
'next_pow2()' is not passed a value out of range but lets anything else
go through unvalidated.

As a result mmap_pages of zero is used e.g.

  perf record -v -m2147483649 uname
  mmap size 0B
  failed to mmap with 22 (Invalid argument)

Fixed:

  perf record -m2147483649 uname
  rounding mmap pages size to 17592186044416 bytes (4294967296 pages)
  Invalid argument for --mmap_pages/-m

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386595120-22978-4-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evlist: Fix max mmap_pages

'SIZE_MAX / page_size' is an upper limit for the maximum number of mmap
pages, not a lower limit.  Change the condition accordingly.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386595120-22978-3-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evlist: Remove unnecessary parentheses

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386595120-22978-2-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf record: Fix display of incorrect mmap pages

'mmap_pages' is 'unsigned int' not 'int' e.g.

  perf record -m2147483648 uname
  Permission error mapping pages.
  Consider increasing /proc/sys/kernel/perf_event_mlock_kb,
  or try again with a smaller value of -m/--mmap_pages.
  (current value: -2147483648)

Fixed:

  perf record -m2147483648 uname
  Permission error mapping pages.
  Consider increasing /proc/sys/kernel/perf_event_mlock_kb,
  or try again with a smaller value of -m/--mmap_pages.
  (current value: 2147483648)

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386595120-22978-5-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script: Add an option to print the source line number

Add field 'srcline' that displays the source file name and line number
associated with the sample ip.  The information displayed is the same as
from addr2line.

 $ perf script -f comm,tid,pid,time,ip,sym,dso,symoff,srcline
            grep 10701/10701 2497321.421013:  ffffffff81043ffa native_write_msr_safe+0xa ([kernel.kallsyms])
  /usr/src/debug/kernel-3.9.fc17/linux-3.9.10-100.fc17.x86_64/arch/x86/include/asm/msr.h:95
            grep 10701/10701 2497321.421984:  ffffffff8165b6b3 _raw_spin_lock+0x13 ([kernel.kallsyms])
  /usr/src/debug/kernel-3.9.fc17/linux-3.9.10-100.fc17.x86_64/arch/x86/include/asm/spinlock.h:54
            grep 10701/10701 2497321.421990:  ffffffff810b64b3 tick_sched_timer+0x53 ([kernel.kallsyms])
  /usr/src/debug/kernel-3.9.fc17/linux-3.9.10-100.fc17.x86_64/kernel/time/tick-sched.c:840
            grep 10701/10701 2497321.421992:  ffffffff8106f63f run_timer_softirq+0x2f ([kernel.kallsyms])
  /usr/src/debug/kernel-3.9.fc17/linux-3.9.10-100.fc17.x86_64/kernel/timer.c:1372

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386315778-11633-3-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script: Fix symoff printing in callchains

The address being used to calculate the offset was the memory address
but the address needed is the address mapped to the dso. i.e. the 'addr'
member of 'struct addr_location'

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: David Ahern <dsahern@gmail.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386315778-11633-2-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Report better error message on bad function args

When Jiri Olsa was writing a function callback for
scsi_trace_parse_cdb(), he thought that the traceevent library had a
bug in it because he was getting this error:

  Error: expected ')' but read ','
  Error: expected ')' but read ','
  Error: expected ')' but read ','
  Error: expected ')' but read ','

But in truth, he didn't have the write number of arguments for the
function callback, and the error was the library detecting the
discrepancy. A better error message would have prevented the confusion:

  Error: function 'scsi_trace_parse_cdb()' only expects 2 arguments but event scsi_dispatch_cmd_timeout has more
  Error: function 'scsi_trace_parse_cdb()' only expects 2 arguments but event scsi_dispatch_cmd_start has more
  Error: function 'scsi_trace_parse_cdb()' only expects 2 arguments but event scsi_dispatch_cmd_error has more
  Error: function 'scsi_trace_parse_cdb()' only expects 2 arguments but event scsi_dispatch_cmd_done has more

Or

  Error: function 'scsi_trace_parse_cdb()' expects 4 arguments but event scsi_dispatch_cmd_timeout only uses 3
  Error: function 'scsi_trace_parse_cdb()' expects 4 arguments but event scsi_dispatch_cmd_start only uses 3
  Error: function 'scsi_trace_parse_cdb()' expects 4 arguments but event scsi_dispatch_cmd_error only uses 3
  Error: function 'scsi_trace_parse_cdb()' expects 4 arguments but event scsi_dispatch_cmd_done only uses 3

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/n/tip-a4c34w62vl0diitvxb7bt3er@git.kernel.org
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf trace: Fix summary percentage when processing files

Getting a divide by 0 when events are processed from a file:

   perf trace -i perf.data -s
   ...
   dnsmasq (1684), 10 events, inf%, 0.000 msec

The problem is that the event count is not incremented as events are
processed. With this patch:

   perf trace -i perf.data -s
   ...
   dnsmasq (1684), 10 events, 8.9%, 0.000 msec

Signed-off-by: David Ahern <dsahern@gmail.com>
Link: http://lkml.kernel.org/r/1386211302-31303-4-git-send-email-dsahern@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf trace: Add support for syscalls vs raw_syscalls

Older kernels (e.g., RHEL6) do system call tracing via
syscalls:sys_{enter,exit} rather than raw_syscalls. Update perf-trace to
detect lack of raw_syscalls support and try syscalls.

Signed-off-by: David Ahern <dsahern@gmail.com>
Link: http://lkml.kernel.org/r/1386211302-31303-2-git-send-email-dsahern@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Update kvm plugin with is_writable_pte helper

Adding is_writable_pte print helper function, so the
kvmmmu:fast_page_fault print format gets resolved properly.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'kvm:*,kvmmmu:*' -a')

  --- script.kvm.old
  +++ script.kvm.new
  - qemu-system-x86  3290 [002] 10708.755312: kvmmmu:fast_page_fault: [FAILED TO PARSE] vcpu_id=0 gva=4094486080 error_code=3 sptep=0xffff88019f1e3670 old_spte=336391285 new_spte=336391287 retry=1
  + qemu-system-x86  3290 [002] 10708.755312: kvmmmu:fast_page_fault: vcpu 0 gva f40ce640 error_code P|W sptep 0xffff88019f1e3670 old 0x140cec75 new 140cec77 spurious 0 fixed 1

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-28-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Remove malloc_or_die from plugin_function.c

Removing malloc_or_die calls from plugin_function.c, replacing them and
factoring the code with standard realloc and error path.

Suggested-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-27-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Several cleanups for function plugin

Several cleanups suggested by Namhyung:

  * Remove index field from struct func_stack as it's not needed.

  * Rename get_index into add_and_get_index.

  * Use '%*X' format string capability instead of the loop

Suggested-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-26-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Use pevent_print_func_field in hrtimer_start handler

The pevent_print_func_field function encompasses all the functionality
used in the hrtimer_start handler. Change the handler to use this
function.

This also unifies the function field output with the
hrtimer_expire_entry handler.

Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-25-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Use static functions in jbd2 plugin

There's no need for following functions to be global:

  process_jbd2_dev_to_name
  process_jiffies_to_msecs

Make them static.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-24-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Remove malloc_or_die from event-plugin.c

Removing malloc_or_die calls from event-plugin.c,
replacing them with standard malloc and error path.

Suggested-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-23-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add cfg80211 plugin

Adding cfg80211 plugin.

This plugin adds handler for __le16_to_cpup function
t properly parse following tracepoint events:
  cfg80211:cfg80211_tx_mlme_mgmt
  cfg80211:cfg80211_rx_mlme_mgmt
  cfg80211:cfg80211_rx_unprot_mlme_mgmt

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'cfg80211:*' -a')

  --- script.cfg80211.old
  +++ script.cfg80211.new
  -        ifconfig  2705 [003]   662.896560: cfg80211:cfg80211_tx_mlme_mgmt: [FAILED TO PARSE] name=wlan0 ifindex=3 frame=ARRAY[c0, 00, 00, 00, 00, 3a, 98, a0, 30, 51, 10, 0b, a9, c6, f4, 74, 00, 3a, 98, a0, 30, 51, 00, 00, 03, 00]
  +        ifconfig  2705 [003]   662.896560: cfg80211:cfg80211_tx_mlme_mgmt: netdev:wlan0(3), ftype:0xc0
  -   kworker/u16:0  1697 [002]   664.808210: cfg80211:cfg80211_rx_mlme_mgmt: [FAILED TO PARSE] name=wlan0 ifindex=3 frame=ARRAY[b0, 00, da, 00, 10, 0b, a9, c6, f4, 74, d8, 24, bd, a1, 26, 31, d8, 24, bd, a1, 26, 31, 10, b7, 00, 00, 02, 00, 00, 00]
  +   kworker/u16:0  1697 [002]   664.808210: cfg80211:cfg80211_rx_mlme_mgmt: netdev:wlan0(3), ftype:0xb0

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-22-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add scsi plugin

Adding scsi plugin.

This plugin adds fields resolving functions for following tracepoint
events:

  scsi:scsi_dispatch_cmd_start
  scsi:scsi_dispatch_cmd_error
  scsi:scsi_dispatch_cmd_done
  scsi:scsi_dispatch_cmd_timeout

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'scsi:scsi_dispatch_cmd*' -a)

  -         swapper     0 [000]  6620.491019: scsi:scsi_dispatch_cmd_done: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 result=0 opcode=53 cmd_len=10 data_sglen=0 prot_sglen=0 prot_op=0 cmnd=5
  +         swapper     0 [000]  6620.491019: scsi:scsi_dispatch_cmd_done: host_no=0 channel=0 id=0 lun=0 data_sgl=0 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(SYNCHRONIZE_CACHE - raw=35 00 00 00 00 00 00 00 00 00) result=(driver=DRIVER_OK host=DID_OK message=COMMAND_COMPLETE status=SAM_STAT_GOOD)
  -     kworker/0:0 21554 [000]  6620.491126: scsi:scsi_dispatch_cmd_start: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 opcode=42 cmd_len=10 data_sglen=1 prot_sglen=0 prot_op=0 cmnd=*
  +     kworker/0:0 21554 [000]  6620.491126: scsi:scsi_dispatch_cmd_start: host_no=0 channel=0 id=0 lun=0 data_sgl=1 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(WRITE_10 lba=570899168 txlen=8 protect=0 raw=2a 00 22 07 3a e0 00 00 08 00)
  -     jbd2/dm-3-8   593 [002]  6621.607992: scsi:scsi_dispatch_cmd_error: [FAILED TO PARSE] host_no=0 channel=0 id=0 lun=0 rtn=4182 opcode=53 cmd_len=10 data_sglen=0 prot_sglen=0 prot_op=0 cmnd=5
  +     jbd2/dm-3-8   593 [002]  6621.607992: scsi:scsi_dispatch_cmd_error: host_no=0 channel=0 id=0 lun=0 data_sgl=0 prot_sgl=0 prot_op=SCSI_PROT_NORMAL cmnd=(SYNCHRONIZE_CACHE - raw=35 00 00 00 00 00 00 00 00 00) rtn=4182

NOTE I couldn't generate scsi_dispatch_cmd_timeout tracepoint,
     but it's similar to the rest, so I believe it's ok.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-21-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add xen plugin

Adding xen plugin.

This plugin adds fields resolving for following tracepoint events:

  xen:xen_mc_entry
  xen:xen_mc_extend_args

The diff of 'perf script' output generated by old and new code: (data
was generated by 'perf record -e 'xen:*' ls')

  --- script.xen.old
  +++ script.xen.new
  -         swapper     0 [002]   136.267492: xen:xen_mc_entry: [FAILED TO PARSE] op=3 nargs=2 args=ARRAY[18, 00, 00, 00, 00, 00, 00, 00, 00, e0, d4, 4b, 04, 88, ff, ff, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00]
  +         swapper     0 [002]   136.267492: xen:xen_mc_entry: op 3(stack_switch) args [18, 0, 0, 0, 0, 0]
  -            perf  1970 [008]   136.273319: xen:xen_mc_extend_args: [FAILED TO PARSE] op=1 args=16 res=1
  +            perf  1970 [008]   136.273319: xen:xen_mc_extend_args: extending op 1(mmu_update) by 16 bytes res ???

NOTE We still do not handle the 'sizeof' and fail to parse following xen
tracepoints:

  xen:xen_mmu_set_pte
  xen:xen_mmu_set_pte_atomic
  xen:xen_mmu_set_domain_pte
  xen:xen_mmu_set_pte_at
  xen:xen_mmu_set_pmd
  xen:xen_mmu_set_pud
  xen:xen_mmu_set_pgd
  xen:xen_mmu_ptep_modify_prot_start
  xen:xen_mmu_ptep_modify_prot_commit

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-20-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add function plugin

Backporting function plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds function and parent function fields resolving for
ftrace:function tracepoint event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e ftrace:function ls')

  --- script.function.old
  +++ script.function.new
  -              ls 10781 [001] 32667.291379: ftrace:function:  ffffffff811adb80 <-- ffffffff811afc48
  -              ls 10781 [001] 32667.291379: ftrace:function:  ffffffff811b35d0 <-- ffffffff811adb9b
  -              ls 10781 [001] 32667.291380: ftrace:function:  ffffffff811b3520 <-- ffffffff811b35e8
  -              ls 10781 [001] 32667.291380: ftrace:function:  ffffffff811b2720 <-- ffffffff811b3549
  -              ls 10781 [001] 32667.291381: ftrace:function:  ffffffff81297e10 <-- ffffffff811b356c
  -              ls 10781 [001] 32667.291381: ftrace:function:  ffffffff81298f40 <-- ffffffff81297e2c
  -              ls 10781 [001] 32667.291382: ftrace:function:  ffffffff81076160 <-- ffffffff811afbf0
  -              ls 10781 [001] 32667.291383: ftrace:function:  ffffffff811c3eb0 <-- ffffffff811afbfc
  -              ls 10781 [001] 32667.291383: ftrace:function:  ffffffff8164e100 <-- ffffffff811c3ed8
  -              ls 10781 [001] 32667.291384: ftrace:function:  ffffffff811a5d10 <-- ffffffff811c3f53
  -              ls 10781 [001] 32667.291384: ftrace:function:  ffffffff811e8e70 <-- ffffffff811a5d58
  -              ls 10781 [001] 32667.291385: ftrace:function:  ffffffff811f38e0 <-- ffffffff811a5d63
  -              ls 10781 [001] 32667.291385: ftrace:function:  ffffffff811a9ff0 <-- ffffffff811a5d6b
  -              ls 10781 [001] 32667.291386: ftrace:function:  ffffffff811a9fa0 <-- ffffffff811aa015
  -              ls 10781 [001] 32667.291387: ftrace:function:  ffffffff810851c0 <-- ffffffff811aa053
  -              ls 10781 [001] 32667.291387: ftrace:function:  ffffffff81090e00 <-- ffffffff81085211
  +              ls 10781 [001] 32667.291379: ftrace:function: would_dump <-- setup_new_exec
  +              ls 10781 [001] 32667.291379: ftrace:function:    inode_permission <-- would_dump
  +              ls 10781 [001] 32667.291380: ftrace:function:       __inode_permission <-- inode_permission
  +              ls 10781 [001] 32667.291380: ftrace:function:          generic_permission <-- __inode_permission
  +              ls 10781 [001] 32667.291381: ftrace:function:          security_inode_permission <-- __inode_permission
  +              ls 10781 [001] 32667.291381: ftrace:function:             cap_inode_permission <-- security_inode_permission
  +              ls 10781 [001] 32667.291382: ftrace:function: flush_signal_handlers <-- setup_new_exec
  +              ls 10781 [001] 32667.291383: ftrace:function: do_close_on_exec <-- setup_new_exec
  +              ls 10781 [001] 32667.291383: ftrace:function:    _raw_spin_lock <-- do_close_on_exec
  +              ls 10781 [001] 32667.291384: ftrace:function:    filp_close <-- do_close_on_exec
  +              ls 10781 [001] 32667.291384: ftrace:function:       dnotify_flush <-- filp_close
  +              ls 10781 [001] 32667.291385: ftrace:function:       locks_remove_posix <-- filp_close
  +              ls 10781 [001] 32667.291385: ftrace:function:       fput <-- filp_close
  +              ls 10781 [001] 32667.291386: ftrace:function:          file_sb_list_del <-- fput
  +              ls 10781 [001] 32667.291387: ftrace:function:          task_work_add <-- fput
  +              ls 10781 [001] 32667.291387: ftrace:function:             kick_process <-- task_work_add

Removing options support as it's not backported yet.
Currently this plugin supports 2 options:
 'parent' to display parent function
 'indent' to show function call indents

Enabling both of them by default.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-19-git-send-email-jolsa@redhat.com
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add sched_switch plugin

Backporting sched_switch plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds fields resolving for sched:sched_switch tracepoint
event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e sched:sched_switch -a')

  --- script.sched_switch.old
  +++ script.sched_switch.new
  -            perf   577 [002] 30965.311852: sched:sched_switch: prev_comm=perf prev_pid=577 prev_prio=120 prev_state=S ==> next_comm=swapper/2 next_pid=0 next_prio=120
  +            perf   577 [002] 30965.311852: sched:sched_switch: perf:577 [120] S ==> swapper/2:0 [120]

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-18-git-send-email-jolsa@redhat.com
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add mac80211 plugin

Backporting mac80211 plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds changed field resolving for
mac80211:drv_bss_info_changed tracepoint event.

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'mac80211:drv_bss_info_changed' -a')

  --- script.mac80211.old
  +++ script.mac80211.new
  -        ifconfig  3711 [000]  1290.446492: mac80211:drv_bss_info_changed: phy0 vif:wlan0(2) changed:0x309f
  +        ifconfig  3711 [000]  1290.446492: mac80211:drv_bss_info_changed: phy0 vif:wlan0(2)
  +                                                                 assoc:0 aid:2 cts:0 shortpre:0 shortslot:0 dtimper:1
  +                                                                 bcnint:102 assoc_cap:0x431 basic_rates:0xf enable_beacon:0
  +                                                                 ht_operation_mode:0

Omitting the mac80211:drv_config tracepoint handling because the kernel
tracepoint changed its prototype and the plugin handler is no longer
working.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-17-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add kvm plugin

Backporting kvm plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds field resolving functions for following
tracepoint events:
  kvm:kvm_exit
  kvm:kvm_emulate_insn
  kvm:kvm_nested_vmexit
  kvm:kvm_nested_vmexit_inject
  kvmmmu:kvm_mmu_get_page
  kvmmmu:kvm_mmu_sync_page
  kvmmmu:kvm_mmu_unsync_page
  kvmmmu:kvm_mmu_zap_page
  kvmmmu:kvm_mmu_prepare_zap_page

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'kvm:*,kvmmmu:*' -a')

  --- script.kvm.old
  +++ script.kvm.new
    qemu-system-x86 17414 [000]  6868.995053: kvm:kvm_exit: reason EPT_VIOLATION rip 0xfff0 info 184 0
    qemu-system-x86 17414 [000]  6868.995109: kvm:kvm_emulate_insn: f0000:c46b:e4 71 (real)
  - qemu-system-x86  3006 [002] 10562.079422: kvmmmu:kvm_mmu_get_page: [FAILED TO PARSE] mmu_valid_gen=0x2 gfn=0 role=122884 root_count=0 unsync=0 created=1
  + qemu-system-x86  3006 [002] 10562.079422: kvmmmu:kvm_mmu_get_page: new sp gfn 0 0/4 q0 direct --- !pge !nxe root 0 sync
  - qemu-system-x86  3006 [002] 10562.080502: kvmmmu:kvm_mmu_prepare_zap_page: [FAILED TO PARSE] mmu_valid_gen=0x2 gfn=0 role=122884 root_count=1 unsync=0
  + qemu-system-x86  3006 [002] 10562.080502: kvmmmu:kvm_mmu_prepare_zap_page: 0/4 q0 direct --- !pge !nxe root 1 sync
    qemu-system-x86  3290 [002] 10708.755312: kvmmmu:fast_page_fault: [FAILED TO PARSE] vcpu_id=0 gva=4094486080 error_code=3 sptep=0xffff88019f1e3670 old_spte=336391285 new_spte=336391287 retry=1
  -          insmod  2576 [001]   781.731666: kvmmmu:kvm_mmu_sync_page: [FAILED TO PARSE] mmu_valid_gen=0x1 gfn=2 role=24624 root_count=10 unsync=1
  +          insmod  2576 [001]   781.731666: kvmmmu:kvm_mmu_sync_page: 3/0 q0 --- !pge !nxe root 10 unsync
  -          insmod  2576 [001]   781.731668: kvmmmu:kvm_mmu_unsync_page: [FAILED TO PARSE] mmu_valid_gen=0x1 gfn=2 role=24624 root_count=10 unsync=1
  +          insmod  2576 [001]   781.731668: kvmmmu:kvm_mmu_unsync_page: 3/0 q0 --- !pge !nxe root 10 unsync

Note:
 - kvm_mmu_zap_page is replaced by kvm_mmu_prepare_zap_page
   in current kernel, keeping it for backward compatibility
 - some of the tracepoints keep the same output even with
   the plugin handling: kvm:kvm_exit, kvm:kvm_emulate_insn
 - the 'kvmmmu:fast_page_fault' is still broken because of
   missing is_writable_pte function and is fixed in another patch
 - ommited following tracepoints from backport because
   the output was buggy
     kvm:kvm_nested_vmexit
     kvm:kvm_nested_vmexit_inject

Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-16-git-send-email-jolsa@redhat.com
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add kmem plugin

Backporting kmem plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds call_site field resolving for following tracepoint
events:

  kmem:kfree
  kmem:kmalloc
  kmem:kmalloc_node
  kmem:kmem_cache_alloc
  kmem:kmem_cache_alloc_node
  kmem:kmem_cache_free

The diff of 'perf script' output generated by old and new code: (data
was generated by 'perf record -e 'kmem:*' -a')

  --- script.kmem.old
  +++ script.kmem.new
  -            perf 27846 [001] 29643.403319: kmem:kfree: call_site=ffffffff810e64f6 ptr=(nil)
  +            perf 27846 [001] 29643.403238: kmem:kfree: (__audit_syscall_exit+0x1f6) call_site=ffffffff810e64f6 ptr=(nil)
  -            perf 27846 [001] 29643.403337: kmem:kmem_cache_alloc: call_site=ffffffff812ff0c5 ptr=0xffff88020e155630 bytes_req=560 bytes_alloc=568 gfp_flags=GFP_KERNEL
  +            perf 27846 [001] 29643.403337: kmem:kmem_cache_alloc: (radix_tree_preload+0x35) call_site=ffffffff812ff0c5 ptr=0xffff88020e155630 bytes_req=560 bytes_alloc=568 gfp_flags=GFP_KERNEL
  -            perf 27846 [001] 29643.403342: kmem:kmem_cache_free: call_site=ffffffff8126ec61 ptr=0xffff88020dffe750
  +            perf 27846 [001] 29643.403342: kmem:kmem_cache_free: (jbd2_journal_stop+0x221) call_site=ffffffff8126ec61 ptr=0xffff88020dffe750
  -         firefox   954 [000] 29643.445477: kmem:kmem_cache_alloc_node: call_site=ffffffff8153c64e ptr=0xffff8801cecb4000 bytes_req=256 bytes_alloc=256 gfp_flags=GFP_KERNEL|GFP_REPEAT node=-1
  +         firefox   954 [000] 29643.445477: kmem:kmem_cache_alloc_node: (__alloc_skb+0x4e) call_site=ffffffff8153c64e ptr=0xffff8801cecb4000 bytes_req=256 bytes_alloc=256 gfp_flags=GFP_KERNEL|GFP_REPEAT node=-1
  -            perf 27846 [001] 29643.445510: kmem:kmalloc: call_site=ffffffff81250642 ptr=0xffff88020fd6c300 bytes_req=96 bytes_alloc=96 gfp_flags=GFP_NOFS|GFP_ZERO
  +            perf 27846 [001] 29643.445510: kmem:kmalloc: (ext4_ext_find_extent+0x362) call_site=ffffffff81250642 ptr=0xffff88020fd6c300 bytes_req=96 bytes_alloc=96 gfp_flags=GFP_NOFS|GFP_ZERO
  -               X   888 [002] 29643.445546: kmem:kmalloc_node: call_site=ffffffff8153c67e ptr=0xffff880103828e00 bytes_req=384 bytes_alloc=512 gfp_flags=GFP_KERNEL|GFP_NOWARN|GFP_REPEAT|GFP_NOMEMALLOC node=-1
  +               X   888 [002] 29643.445546: kmem:kmalloc_node: (__alloc_skb+0x7e) call_site=ffffffff8153c67e ptr=0xffff880103828e00 bytes_req=384 bytes_alloc=512 gfp_flags=GFP_KERNEL|GFP_NOWARN|GFP_REPEAT|GFP_NOMEMALLOC node=-1

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-15-git-send-email-jolsa@redhat.com
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add hrtimer plugin

Backporting hrtimer plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds function field resolving for following tracepoint
events:

  timer:hrtimer_expire_entry
  timer:hrtimer_start

The diff of 'perf script' output generated by old and new code: (data
was generated by 'perf record -e 'timer:hrtimer*' -a')

  --- script.hrtimer.old
  +++ script.hrtimer.new
  -         swapper     0 [000] 27405.519092: timer:hrtimer_start: [FAILED TO PARSE] hrtimer=0xffff88021e20e800 function=0xffffffff810c0e10 expires=27398383000000 softexpires=27398383000000
  +         swapper     0 [000] 27405.519103: timer:hrtimer_start: hrtimer=0xffff88021e20e800 function=tick_sched_timer expires=27398383000000 softexpires=27398383000000
  -         swapper     0 [001] 27405.519544: timer:hrtimer_expire_entry: [FAILED TO PARSE] hrtimer=0xffff880211334058 now=27398294182491 function=0xffffffff81086f20
  +         swapper     0 [001] 27405.519544: timer:hrtimer_expire_entry: hrtimer=0xffff880211334058 now=27398294182491 function=posix_timer_fn/0x0

Check the 'function' field is translated into the function name.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-14-git-send-email-jolsa@redhat.com
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add jbd2 plugin

Backporting jbd2 plugin.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

This plugin adds field resolving functions for following tracepoint
events:

  jbd2:jbd2_checkpoint_stats
  jbd2:jbd2_run_stats

The diff of 'perf script' output generated by old and new code:
(data was generated by 'perf record -e 'jbd2:jbd2_run_stats,jbd2:jbd2_checkpoint_stats' -a')

  --- script.jbd2.old
  +++ script.jbd2.new
  -     jbd2/dm-3-8   576 [000]  2983.748423: jbd2:jbd2_checkpoint_stats: [FAILED TO PARSE] dev=265289731 tid=0x3f2bbb chp_time=0x0 forced_to_close=0 written=0 dropped=2
  +     jbd2/dm-3-8   576 [000]  2983.748423: jbd2:jbd2_checkpoint_stats: dev 253,3 tid 4139963 chp_time 0 forced_to_close 0 written 0 dropped 2
  -     jbd2/dm-3-8   576 [000]  2983.852789: jbd2:jbd2_run_stats: [FAILED TO PARSE] dev=265289731 tid=0x3f2bc0 wait=0x0 request_delay=0x0 running=0x138d locked=0x0 flushing=0x0 logging=0x68 handle_count=2995 blocks=17 blocks_logged=18
  +     jbd2/dm-3-8   576 [000]  2983.852789: jbd2:jbd2_run_stats: dev 253,3 tid 4139968 wait 0 request_delay 0 running 5005 locked 0 flushing 0 logging 104 handle_count 2995 blocks 17 blocks_logged 18

This plugin also adds jbd2_dev_to_name function, which was removed from
jdb2 tracepoints recently. Keeping it in for backward compatibility
reasons.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-13-git-send-email-jolsa@redhat.com
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Overload pr_stat traceevent print function

The traceevent lib uses pr_stat to display all standard info. It's
defined as __weak. Overloading it with perf version plugged into perf
output system logic.

Displaying the pr_stat stuff under '-v' option.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-12-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add trace-event global object for tracepoint interface

In order to get the proper plugins processing we need to use full
trace-event interface when creating tracepoint events. So far we were
using shortcut to get the parsed format.

Moving current 'event_format__new' function into trace-event object as
'trace_event__tp_format'.

This function uses properly initialized global trace-event object,
ensuring proper plugins processing.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-11-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add trace-event object

Add trace-event object to keep together 'struct pevent' object with its
loaded plugins with following interface:

int trace_event__init(struct trace_event *t);

  - Initalizes 'struct pevent' object and loads plugins for it

void trace_event__cleanup(struct trace_event *t);

  - Cleanups both 'struct pevent' and plugins

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-10-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add filename__read_str util function

Adding filename__read_str util function to read
text file and return it in the char array.

The interface is:
  int filename__read_str(const char *filename, char **buf, size_t *sizep)

  Returns 0/-1 if the read suceeded/fail respectively.

  buf  - place to store the data pointer
  size - place to store data size

v2 change:
  - better error handling suggested by Namhyung Kim.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-9-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add build and install plugins targets

Adding 'plugins' target along with the libtraceevent.a, so plugins are
built together with traceevent library.

Adding 'install-traceevent-plugins' Makefile install target, instructing
perf to install plugins into:

  $(HOME)/.traceevent/plugins
    - If installed localy under $HOME

  $(DESTDIR)/$(prefix)/$(libdir)/traceevent/plugins
    - If installed globally

Examples:
  $ make install
  ...
  $ find ~/.traceevent/plugins/
  /home/jolsa/.traceevent/plugins/
  /home/jolsa/.traceevent/plugins/plugin_mac80211.so
  /home/jolsa/.traceevent/plugins/plugin_kvm.so
  /home/jolsa/.traceevent/plugins/plugin_scsi.so
  /home/jolsa/.traceevent/plugins/plugin_sched_switch.so
  /home/jolsa/.traceevent/plugins/plugin_xen.so
  /home/jolsa/.traceevent/plugins/plugin_cfg80211.so
  /home/jolsa/.traceevent/plugins/plugin_function.so
  /home/jolsa/.traceevent/plugins/plugin_kmem.so
  /home/jolsa/.traceevent/plugins/plugin_hrtimer.so
  /home/jolsa/.traceevent/plugins/plugin_jbd2.so

  $ sudo make install DESTDIR=/opt/perf/
  ...
  $ find /opt/perf/lib64/traceevent/plugins/
  /opt/perf/lib64/traceevent/plugins/
  /opt/perf/lib64/traceevent/plugins/plugin_kvm.so
  /opt/perf/lib64/traceevent/plugins/plugin_scsi.so
  /opt/perf/lib64/traceevent/plugins/plugin_mac80211.so
  /opt/perf/lib64/traceevent/plugins/plugin_hrtimer.so
  /opt/perf/lib64/traceevent/plugins/plugin_kmem.so
  /opt/perf/lib64/traceevent/plugins/plugin_jbd2.so
  /opt/perf/lib64/traceevent/plugins/plugin_sched_switch.so
  /opt/perf/lib64/traceevent/plugins/plugin_function.so
  /opt/perf/lib64/traceevent/plugins/plugin_cfg80211.so
  /opt/perf/lib64/traceevent/plugins/plugin_xen.so

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-8-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Harmonize the install messages in lib-traceevent

Removing the 'to ...' part out of the install message, because it does
not fit to the rest of the build messages we use.

Before:
  INSTALL  plugin_hrtimer.so    to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_jbd2.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_kmem.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_kvm.so        to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_mac80211.so   to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_sched_switch.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_function.so   to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_xen.so        to      /home/jolsa/libexec/perf-core/traceevent/plugins
  INSTALL  plugin_scsi.so       to      /home/jolsa/libexec/perf-core/traceevent/plugins

Now:
  INSTALL  plugin_jbd2.so
  INSTALL  plugin_hrtimer.so
  INSTALL  plugin_kmem.so
  INSTALL  plugin_kvm.so
  INSTALL  plugin_mac80211.so
  INSTALL  plugin_sched_switch.so
  INSTALL  plugin_function.so
  INSTALL  plugin_xen.so
  INSTALL  plugin_scsi.so

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-7-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Change pevent_parse_format to include pevent handle

Changing the pevent_parse_format interface to include the pevent handle.

The goal is to always use pevent object when dealing with traceevent
library. The reason is that we might need additional processing (like
plugins), which is not possible otherwise.

Patches follow to make this happen completely.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-6-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add traceevent_host_bigendian function

Adding traceevent_host_bigendian function to get host endianity. It's
used in following patches.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-5-git-send-email-jolsa@redhat.com
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add plugin build support

Backporting missing pieces of plugin building infrastructure:

  - Adding Makefile 'plugins' target to build all
    defined plugins

  - Adding Makefile 'install_plugins' target as 'install_lib'
    target dependency

  - Link plugin objects with shared object building

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

Plugins are by default installed into following locations:

  '$(HOME)/.traceevent/plugins'
     - If we are installing under $(HOME)

  '$(prefix)/lib/traceevent/plugins'
     - Otherwise

This path is propagated to the plugin object as a plugins search path.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-4-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Add plugin support

Backporting plugin support for traceevent lib.

Backported from Steven Rostedt's trace-cmd repo (HEAD 0f2c2fb):
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/trace-cmd.git

It's now possible to use following interface to load plugins
(shared objects) to enhance pevent object functionality.

The plugin interface/hooks are as follows:
(taken from event-parse.h comments)

- 'pevent_plugin_loader' (required)
    The function name to initialized the plugin.

    int pevent_plugin_loader(struct pevent *pevent)

- 'pevent_plugin_unloader' (optional)
    The function called just before unloading

    int pevent_plugin_unloader(void)

- 'pevent_plugin_options'  (optional)
    Plugin options that can be set before loading

    struct plugin_option pevent_plugin_options[] = {
       {
               .name = "option-name",
               .plugin_alias = "overide-file-name", (optional)
               .description = "description of option to show users",
       },
       {
               .name = NULL,
       },
    };

    Array must end with .name = NULL;

    The plugin_alias (below) can be used to give a shorter
    name to access the variable. Useful if a plugin handles
    more than one event.

    NOTE options support is not backported yet.

- 'pevent_plugin_alias' (optional)
    The name to use for finding options (uses filename if not defined)

New traceevent functions are added to search and load
available plugins:

  struct plugin_list*
  traceevent_load_plugins(struct pevent *pevent)
    - loads plusing for 'struct pevent' object and returns
      loaded plugins list

  void traceevent_unload_plugins(struct plugin_list *plugin_list);
    - unload plugin list

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-3-git-send-email-jolsa@redhat.com
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script: Do not call perf_event__preprocess_sample() twice)

The perf_event__preprocess_sample() function is called in
process_sample_event().  Instead of calling it again in
perf_evsel__print_ip(), pass through the resultant addr_location.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/529F3944.9050007@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf symbols: Fix random fd closing with no libelf

When built without libelf, perf tools was failing to initialize a file
descriptor, but nevertheless closing it.  That sometimes resulted in the
output being truncated because the stdout file descriptor got closed.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386166981-30197-1-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf kvm: Update the 'record' man page entry for new --guest/--host behavior

As we have changed the default behavior of 'perf kvm' to --guest
enabled, the parts of the man page that covers the 'record' subcommand
are outdated.

This patch updates it to show the correct output with
--host/--guest/neither/both of them.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/3a3a9c1e05acb5a274d1d8369db5a4c6467d6276.1386197481.git.yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf kvm: Fix spurious '=' use in man page

As option --host and --guest request no input for it, there should not
be a '=' after them in the man page sources.

And --output expects a filename as the input, so there should be a '='
after it.

This patch removes the needless '=' after --guest and --host, and adds a
'=' after --output in perf-kvm.txt.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/6124d9eb10a3f1f6b399d1db660110bc7a60fd6b.1386197481.git.yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf kvm: Add more detail about buildid-list in man page

As the buildid is read from /sys/kernel/notes, then if we use perf kvm
buildid-list with a perf data file captured by perf kvm record with
--guestkallsyms and --guestmodules, there is no result in output.

This patch add a explanation about it and add a limit of using perf kvm
buildid-list.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/d605a805486340b53bc261aa64d7632ad0a8cf53.1386197481.git.yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf target: Move the checking of which map function to call into function.

Check for cpu_map__dummy_new() or cpu_map__new() to be called in
perf_evlist__create_maps() is more complicated.

This patch moves the checking work into target.h, combining two
conditions and making perf_evlist__create_maps() more readable.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/b8c41f1fd2c4f0df71eb7b19aea74fb64d46cdda.1386197481.git.yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Remove condition in machine__get_kernel_start_addr.

In machine__get_kernel_start_addr, the code, which is using
machine->root_dir to build filename, works for both host and guests
initialized from guestmount, as root_dir is set to "" for the host
machine in the machine__init() function.

So this patch remove the branch for machine__is_host.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/0a81645dd0b384a12cb4f962cf193ef8c3ce2010.1386197481.git.yangds.fnst@cn.fujitsu.com
[ Clarified changeset mentioning root_dir setup in machine__init() ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Remove stackprotector feature check

We use -fstack-protector-all option to enable stack protecting for all
available functions. There's no reason for enabling -Wstack-protector to
get warning for unprotected functions.

Removing stackprotector feature check which was used to enable the
-Wstack-protector option.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1386076182-14484-2-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Do not disable source line lookup just because of 1 failure

Looking up an ip's source file name and line number does not succeed
always.  Current logic disables the lookup for a dso entirely on any
failure.  Change it so that disabling never happens if there has ever
been a successful lookup for that dso but disable if the first 123
lookups fail.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386055390-13757-8-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf symbols: Retain symbol source file name to lookup source line numbers

Currently, lookup of an ip's source file name and line number is done
using the dso file name.

Instead retain the file name used to lookup the dso's symbols and use
that.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386055390-13757-6-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf symbols: Retain bfd reference to lookup source line numbers

Closng and re-opening for every lookup when using libbfd to lookup
source file name and line number is very very slow.  Instead keep the
reference on struct dso.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386055390-13757-5-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Use asprintf instead of malloc plus snprintf

The asprintf library function is equivalent to malloc plus snprintf so
use it because it is simpler.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1386055390-13757-4-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core

Pull perf/tools improvements and fixes from Arnaldo Carvalho de Melo:

  * Honour -m option in 'trace', the tool was offering the option to
    set the mmap size, but wasn't using it when doing the actual mmap
    on the events file descriptors, fix from Jiri Olsa.

  * Correct the message in feature-libnuma checking, swowing the right
    devel package names for various distros, from Dongsheng Yang.

  * Polish 'readn' function and introduce its counterpart, 'writen', from
    Jiri Olsa.

  * Start moving timechart state from global variables to a 'perf_tool' derived
    'timechart' struct.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

perf trace: Honour -m option

Currently trace command supports '-m' option, but does not honours its
value and keeps the default.

Changing the perf_evlist__mmap function call to use the '-m' configured
value.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: David Ahern <dsahern@gmail.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1385657842-8914-1-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Include test-stackprotector-all.c in test-all

We're missing test-stackprotector-all.c in test-all check, adding it.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/1385638408-23519-3-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Correct the message in feature-libnuma checking.

The package required for numa is named numactl-devel in Fedora or RHEL,
and libnuma-devel in OpenSuSE, and libnuma-dev in Ubuntu.

This patch corrects the package name in warning message in
feature-libnuma checking.

Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Acked-by: Peter Hurley <peter@hurleysoftware.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Link: http://lkml.kernel.org/r/1385998008-6851-1-git-send-email-yangds.fnst@cn.fujitsu.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Move wake_events list to 'struct timechart'

Removing another global variable.

Cc: 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: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stanislav Fomichev <stfomichev@yandex-team.ru>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-14rpuci11l2s0o01yta87kxe@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Move power_events list to 'struct timechart'

Removing another global variable.

Cc: 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: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stanislav Fomichev <stfomichev@yandex-team.ru>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-2akef3p9caau56itf5mugd2b@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Move all_data per_pid list to 'struct timechart'

Removing another global variable.

This one tho would be better done by using the machine infrastructure,
searching for the 'struct thread' with a pid, then using thread->priv,
etc.

TODO list material for now.

Cc: 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: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stanislav Fomichev <stfomichev@yandex-team.ru>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-yyfpudgjvr6mev4bue9u72a2@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Introduce tool struct

To avoid having all those global variables and to use the interface to
event processing that is based on passing a 'perf_tool' struct that
should be embedded in a per tool specific struct passed to all the
sample processing callbacks.

There are some more globals to move, next patches will do it.

Cc: 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: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stanislav Fomichev <stfomichev@yandex-team.ru>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-0iah65pq796ezbk5u1lzwy1k@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add perf_data_file__write interface

Adding perf_data_file__write interface to centralize output to files.
The function prototype is:

  ssize_t perf_data_file__write(struct perf_data_file *file,
                                void *buf, size_t size);

Returns number of bytes written or -1 in case of error.

NOTE: Also indenting 'struct perf_data_file' members, no functional
      change done.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1385634619-8129-6-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add writen function

Adding 'writen' function as a synchronous wrapper for write syscall with
following prototype:

  ssize_t writen(int fd, void *buf, size_t n)

Returns the number of bytes written on success or -1 in case of err.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Requested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1385634619-8129-5-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Fine tune readn function

Added a 'left' variable to make the flow clearer, and added a debug
check for the return value - returning 'n' is more obvious.

Added small comment for readn.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Original-patch-by: Ingo Molnar <mingo@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1385634619-8129-4-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Use correct return type for readn function

Changing readn function return type to ssize_t because read returns
ssize_t not int.

Changing callers holding variable types as well.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1385634619-8129-3-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf record: Unify data output code into perf_record__write function

Unifying current 2 data output functions do_write_output and
write_output into single one perf_record__write.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1385634619-8129-2-git-send-email-jolsa@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

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:

  * Make per-cpu mmaps the default in 'perf record', from Adrian Hunter.

  * Default -t/--thread 'perf record' option to no inheritance,
    from Adrian Hunter.

  * Make 'perf top -g' refer to callchains, for consistency with other tools,
    from David Ahern.

  * Skip ignored symbols while printing callchain, from David Ahern.

  * Print callchains and symbols if they exist in 'perf script',
    from David Ahern.

  * Remove thread summary coloring in 'perf trace', from Pekka Enberg.

  * zsh completion support, from Ramkumar Ramachandra.

  * 'perf timechart' improvements, including backtrace support,
    from Stanislav Fomichev.

  * Fix using kcore files stored in the buildid cache when doing report/annotate
    in non-live sessions, from Adrian Hunter

  * Minor 'timechart' cleanups.

  * Fix tags/TAGS targets rebuilding, from Jiri Olsa.

  * Add options to show comm, fork, exit and mmap PERF_RECORD_ events in
    'perf script', from Namhyung Kim.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

perf tools: Fix tags/TAGS targets rebuilding

Once the tags/TAGS file is generated it's never rebuilt until it's
removed by hand.

The reason is that the Makefile does not treat tags/TAGS as targets but
as files and thus won't rebuilt them once they are in place.

Adding PHONY tags/TAGS targets into Makefile.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20131126125412.GJ1267@krava.brq.redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Remove misplaced __maybe_unused

The 'event' parameter _is_ used.

Cc: 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: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stanislav Fomichev <stfomichev@yandex-team.ru>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-`ranpwd
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Remove some needless struct forward declarations

Cc: 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: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stanislav Fomichev <stfomichev@yandex-team.ru>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-jomi6mjv5zi9vsn4vmih5xps@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: dynamically determine event fields offset

Since b000c8065a92 "tracing: Remove the extra 4 bytes of padding in
events" removed padding bytes, perf timechart got out of sync with the
kernel's trace_entry structure.

Convert perf timechart to use dynamic fields offsets (via
perf_evsel__intval) not relying on a hardcoded copy of fields layout
from the kernel.

Signed-off-by: Stanislav Fomichev <stfomichev@yandex-team.ru>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Chia-I Wu <olvaffe@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/20131127104459.GB3309@stfomichev-desktop
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf symbols: Fix not finding kcore in buildid cache

The logic was not looking in the buildid cache for kcore if the host
kernel buildid did not match the recorded kernel buildid.

This affects the non-live case i.e. the kernel has changed and we are
looking at a special copy of kcore that we placed in the buildid cache
(using "perf buildid-cache -v -k /proc/kcore") when the data was
recorded.

After this fix kernel symbols get resolved/annotated correctly.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1385471964-4037-1-git-send-email-adrian.hunter@intel.com
[ Added further explanation extracted from conversation between Ingo & Adrian on lkml ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script: Print mmap[2] events also

If --show-mmap-events option is given, also print internal MMAP and
MMAP2 events.  It would be helpful for debugging.

  $ perf script --show-mmap-events
  ...
           sleep  9486 [009] 3350640.335531: PERF_RECORD_MMAP 9486/9486: [0x400000(0x6000) @ 0]: x /usr/bin/sleep
           sleep  9486 [009] 3350640.335542: PERF_RECORD_MMAP 9486/9486: [0x3153a00000(0x223000) @ 0]: x /usr/lib64/ld-2.17.so
           sleep  9486 [009] 3350640.335553: PERF_RECORD_MMAP 9486/9486: [0x7fff8b5fe000(0x2000) @ 0x7fff8b5fe000]: x [vdso]
           sleep  9486 [009] 3350640.335643: PERF_RECORD_MMAP 9486/9486: [0x3153e00000(0x3c0000) @ 0]: x /usr/lib64/libc-2.17.so

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Suggested-by: Frederic Weisbecker <fweisbec@gmail.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@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/1385456066-26592-1-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script: Print comm, fork and exit events also

If --show-task-events option is given, also print internal COMM, FORK
and EXIT events.  It would be helpful for debugging.

  $ perf script --show-task-events
  ...
         swapper     0 [009] 3350640.335261: sched:sched_switch: prev_comm=swapper/9
           sleep  9486 [009] 3350640.335509: PERF_RECORD_COMM: sleep:9486
           sleep  9486 [009] 3350640.335806: sched:sched_stat_runtime: comm=sleep pid=9486
         firefox  2635 [003] 3350641.275896: PERF_RECORD_FORK(2635:9487):(2635:2635)
         firefox  2635 [003] 3350641.275896: sched:sched_process_fork: comm=firefox pid=2635
           sleep  9486 [009] 3350641.336009: PERF_RECORD_EXIT(9486:9486):(9486:9486)

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Suggested-by: Frederic Weisbecker <fweisbec@gmail.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@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/1385455873-25865-1-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script: Print callchains and symbols if they exist

The intent of perf-script is to dump the events and information in the
file. H/W, S/W and raw events all dump callchains if they are present;
might as well make that the default for tracepoints too.

v2: Only add options for sym, dso and ip if callchains are present

Signed-off-by: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/1384920457-5986-1-git-send-email-dsahern@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Export setup_list

Used in upcoming patches (perf sched timehist command).

Signed-off-by: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1384806771-2945-6-git-send-email-dsahern@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf thread: Move comm_list check into function

Signed-off-by: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1384806771-2945-5-git-send-email-dsahern@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf symbols: Move idle syms check from top to generic function

Allows list of idle symbols to be leveraged by other commands, such as
the upcoming timehist command.

Signed-off-by: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1384806771-2945-3-git-send-email-dsahern@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evsel: Skip ignored symbols while printing callchain

Allows a command to have a symbol_filter controlled by the user to skip
certain functions in a backtrace. One example is to allow the user to
reduce repeating patterns like:

    do_select  core_sys_select  sys_select

to just sys_select when dumping callchains, consuming less real estate
on the screen while still conveying the essential message - the process
is in a select call.

This option is leveraged by the upcoming timehist command.

Signed-off-by: David Ahern <dsahern@gmail.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1384806771-2945-2-git-send-email-dsahern@gmail.com
[ Checked if al.sym is NULL before touching al.sym->ignored, as noted by Adrian Hunter ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Add backtrace support

Add -g flag to `perf timechart record` which saves callchain info in the
perf.data.

When generating SVG, add backtrace information to the figure details, so
now it's possible to see which code path woke up the task and why some
task went to sleep.

Signed-off-by: Stanislav Fomichev <stfomichev@yandex-team.ru>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1383323151-19810-8-git-send-email-stfomichev@yandex-team.ru
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Add support for -P and -T in timechart recording

If we don't want either power or task events we may use -T or -P with
the `perf timechart record` command to filter out events while recording
to keep perf.data small.

Signed-off-by: Stanislav Fomichev <stfomichev@yandex-team.ru>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1383323151-19810-7-git-send-email-stfomichev@yandex-team.ru
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Group figures and add title with details

Add titles to figures so we can run SVG interactively in Firefox and
check event details in the tooltips.

This also aids exploring SVG with Inkscape because when user clicks on
one part of logical figure, all parts are selected.

It's also possible to read titles with Inkscape in the object details.

Signed-off-by: Stanislav Fomichev <stfomichev@yandex-team.ru>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1383323151-19810-6-git-send-email-stfomichev@yandex-team.ru
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Add support for displaying only tasks related data

In order to make SVG smaller and faster to browse add possibility to
switch off power related information with -T switch.

Signed-off-by: Stanislav Fomichev <stfomichev@yandex-team.ru>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1383323151-19810-5-git-send-email-stfomichev@yandex-team.ru
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Use proc_num to implement --power-only

Don't use special flag to indicate power-only mode, just set proc_num to
0.

Signed-off-by: Stanislav Fomichev <stfomichev@yandex-team.ru>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1383323151-19810-4-git-send-email-stfomichev@yandex-team.ru
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Add option to limit number of tasks

Add -n option to specify min. number of tasks to print.

Signed-off-by: Stanislav Fomichev <stfomichev@yandex-team.ru>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1383323151-19810-3-git-send-email-stfomichev@yandex-team.ru
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf timechart: Always try to print at least 15 tasks

Always try to print at least 15 tasks no matter how long they run.

Signed-off-by: Stanislav Fomichev <stfomichev@yandex-team.ru>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1383323151-19810-2-git-send-email-stfomichev@yandex-team.ru
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf record: Default -t option to no inheritance

The change to per-cpu mmaps causes the -p, -t and -u options now to have
inheritance enabled by default.  Change that back to no inheritance but
for the -t option only.

Requested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1384768557-23331-5-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add option macro OPT_BOOLEAN_SET

OPT_BOOLEAN_SET records whether a boolean option was set by the user.

That information can be used to change the default value for the option
after the options have been parsed.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1384768557-23331-4-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Allow '--inherit' as the negation of '--no-inherit'

Long options can be negated by prefixing them with 'no-'.  However
options that already start with 'no-', such as '--no-inherit' result in
ugly double 'no's.

Avoid that by accepting that the removal of 'no-' also negates the long
option.

Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1384768557-23331-3-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf record: Make per-cpu mmaps the default.

This affects the -p, -t and -u options that previously defaulted to
per-thread mmaps.

Consequently add an option to select per-thread mmaps to support the old
behaviour.

Note that per-thread can be used with a workload-only (i.e. none of -p,
-t, -u, -a or -C is selected) to get a per-thread mmap with no
inheritance.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/5286271D.3020808@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script: Move evname print code to process_event()

The print_sample_start() will be reused by other printing routine for
internal events like COMM, FORK and EXIT from next patch.  And because
they're not tied to a specific event, move the evname print code to its
caller.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Reviewed-by: David Ahern <dsahern@gmail.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
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/1384752894-10974-1-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf completion: Rename file to reflect zsh support

Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Link: http://lkml.kernel.org/r/1384704807-15779-6-git-send-email-artagnon@gmail.com
[ Fix 'make install' target ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf completion: Introduce zsh support

__perfcomp(), __perfcomp_colon(), and _perf() have to be overridden.
Inspired by the way the git.git completion system is structured.

Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/r/1384704807-15779-5-git-send-email-artagnon@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf completion: Factor out call to __ltrim_colon_completions

In our sole callsite, __ltrim_colon_completions is called after
__perfcomp, to modify the COMPREPLY set by the invocation.

This is problematic, because in the zsh equivalent (using compset/
compadd), we'll have to generate completions in one-shot.

So factor out this entire callsite into a special override'able
__perfcomp_colon function; we will override it when introducing zsh
support.

Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Link: http://lkml.kernel.org/r/1384704807-15779-4-git-send-email-artagnon@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf completion: Factor out compgen stuff

compgen is a bash-builtin; factor out the invocations into a separate
function to give us a chance to override it with a zsh equivalent in
future patches.

Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Link: http://lkml.kernel.org/r/1384704807-15779-3-git-send-email-artagnon@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf completion: Introduce a layer of indirection

Define the variables cur, words, cword, and prev outside the main
completion function so that we have a chance to override it when we
introduce zsh support.

Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Link: http://lkml.kernel.org/r/1384704807-15779-2-git-send-email-artagnon@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf top: Make -g refer to callchains

In most commands -g is used for callchains. Make perf-top follow suit.
Move group to just --group with no short cut making it similar to
perf-record.

Signed-off-by: David Ahern <dsahern@gmail.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/r/1384487490-6865-1-git-send-email-dsahern@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf trace: Remove thread summary coloring

Thread summary line coloring looks ugly.  It doesn't add much value so
remove coloring completely.

Signed-off-by: Pekka Enberg <penberg@kernel.org>
Acked-by: David Ahern <dsahern@gmail.com>
Link: http://lkml.kernel.org/r/1384447410-1771-1-git-send-email-penberg@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools lib traceevent: Use helper trace-seq in print functions like kernel does

Jiri Olsa reported that his plugin for scsi was chopping off part of the
output. Investigating this, I found that Jiri used the same functions as
what is in the kernel, which adds the following:

trace_seq_putc(p, 0);

This adds a '\0' to the output string. The reason this works in the
kernel is that the "p" that is passed to the function helper is a
temporary trace_seq. But in the libtraceevent library, it's the pointer
to the trace_seq used to output. By adding the '\0', it truncates the
line and nothing added after that will be printed.

We can solve this in two ways. One is to have the helper functions for
the library not add the unnecessary '\0'. The other is to change the
library to also use a helper trace_seq structure that gets copied to the
main trace_seq just like the kernel does.

The latter allows the helper functions in the plugins to be the same as
the kernel, which is the better solution.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Reported-by: Jiri Olsa <jolsa@redhat.com>
Tested-by: Jiri Olsa <jolsa@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/r/20131119182937.401668e3@gandalf.local.home
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>