Commits · d8d1656ee15d3085e0085a87e70f9093a0a102c5 · nexedi / linux

27 May, 2009 1 commit

perf report: Use hex2long instead of sscanf · d8d1656e

Arnaldo Carvalho de Melo authored May 26, 2009

Before:

[acme@emilia ~]$ perf record -o perf_report.perf perf stat perf report > /dev/null

 Performance counter stats for 'perf':

     245.414985  task clock ticks     (msecs)
              6  context switches     (events)
              6  CPU migrations       (events)
           2108  pagefaults           (events)
       37493013  CPU cycles           (events)  (scaled from 67.04%)
       13576789  instructions         (events)  (scaled from 66.76%)
          57931  cache references     (events)  (scaled from 21.96%)
          12263  cache misses         (events)  (scaled from 21.98%)

 Wall-clock time elapsed:   246.575587 msecs

[acme@emilia ~]$ perf report -i perf_report.perf | head
12.15          perf [.] 0x000000000005432a /lib64/libc-2.5.so: _IO_vfscanf_internal
 9.38          perf [k] 0xffffffff8101b1d2 intel_pmu_enable_all
 8.53          perf [.] 0x00000000000056b8 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__insert_symbol
 6.61          perf [.] 0x00000000000057cb /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__find_symbol
 5.33          perf [k] 0xffffffff811ce082 number
 4.69          perf [.] 0x0000000000034829 /lib64/libc-2.5.so: ____strtoull_l_internal
 4.48          perf [.] 0x0000000000006505 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: thread__symbol_incnew
 3.41          perf [.] 0x000000000000fce6 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: rb_insert_color
 3.20          perf [k] 0xffffffff811cfc01 vsnprintf
 2.99          perf [k] 0xffffffff811ce5e8 format_decode

After:

[acme@emilia ~]$ perf record -o perf_report.perf perf stat perf report > /dev/null

 Performance counter stats for 'perf':

     218.186805  task clock ticks     (msecs)
              4  context switches     (events)
              7  CPU migrations       (events)
           2133  pagefaults           (events)
       32735365  CPU cycles           (events)  (scaled from 67.04%)
       11952309  instructions         (events)  (scaled from 66.26%)
          50314  cache references     (events)  (scaled from 21.96%)
          13228  cache misses         (events)  (scaled from 21.98%)

 Wall-clock time elapsed:   218.810451 msecs

[acme@emilia ~]$ perf report -i perf_report.perf | head
10.68          perf [.] 0x000000000000578d /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__find_symbol
 9.62          perf [.] 0x00000000000065f7 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: thread__symbol_incnew
 9.40          perf [.] 0x00000000000056b4 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__insert_symbol
 9.19          perf [k] 0xffffffff8101b1d2 intel_pmu_enable_all
 5.13          perf [.] 0x0000000000005ec7 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex2long
 4.49          perf [k] 0xffffffff81083808 kallsyms_expand_symbol
 3.85          perf [k] 0xffffffff811ce2c1 number
 3.63          perf [.] 0x0000000000005e81 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex
 2.99          perf [.] 0x000000000000fd5b /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: rb_insert_color
 2.99          perf [k] 0xffffffff811cf251 string
[acme@emilia ~]$

[ Impact: optimization ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
LKML-Reference: <20090526222057.GI4424@ghostprotocols.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

d8d1656e

26 May, 2009 29 commits

perf report: Sort output by symbol usage · 3a4b8cc7

Arnaldo Carvalho de Melo authored May 26, 2009

[acme@emilia ~]$ perf record find / > /dev/null 2>&1
[acme@emilia ~]$ perf stat perf report | head -20
 4.95          find [k] 0xffffffff81393d65 _spin_lock
 3.89          find [.] 0x000000000000af89 /usr/bin/find: <unknown>
 2.19          find [k] 0xffffffffa00518e0 ext3fs_dirhash
 1.87          find [k] 0xffffffff810a6cea __rcu_read_lock
 1.86          find [k] 0xffffffff811c7312 _atomic_dec_and_lock
 1.86          find [.] 0x00000000000782ab /lib64/libc-2.5.so: __GI_strlen
 1.85          find [k] 0xffffffff810fedfb __kmalloc
 1.62          find [.] 0x00000000000430ff /lib64/libc-2.5.so: vfprintf
 1.59          find [k] 0xffffffff810a6d6d __rcu_read_unlock
 1.55          find [k] 0xffffffff81119395 __d_lookup
 1.39          find [.] 0x0000000000071b40 /lib64/libc-2.5.so: _int_malloc
 1.30          find [k] 0xffffffffa031c4fc nfs_do_filldir
 1.21          find [k] 0xffffffff811876a5 avc_has_perm_noaudit
 1.15          find [k] 0xffffffff810fef62 kmem_cache_alloc
 1.07          find [k] 0xffffffff811d03fb copy_user_generic_string
 1.03          find [k] 0xffffffffa0043882 ext3_htree_store_dirent
 0.99          find [k] 0xffffffff81393ebb _spin_lock_bh
 0.98          find [k] 0xffffffffa03319a2 nfs3_decode_dirent
 0.97          find [k] 0xffffffff8100bf20 system_call
 0.92          find [k] 0xffffffff8139437e _spin_unlock

 Performance counter stats for 'perf':

     244.278972  task clock ticks     (msecs)
              8  context switches     (events)
              9  CPU migrations       (events)
           2104  pagefaults           (events)
       35329669  CPU cycles           (events)  (scaled from 75.40%)
       13740366  instructions         (events)  (scaled from 75.49%)
          59073  cache references     (events)  (scaled from 24.60%)
            196  cache misses         (events)  (scaled from 24.51%)

 Wall-clock time elapsed:   246.060717 msecs

[acme@emilia ~]$
[acme@emilia ~]$ grep "model name" /proc/cpuinfo | head -1
model name	: Intel(R) Xeon(R) CPU           E5405  @ 2.00GHz
[acme@emilia ~]$ grep "model name" /proc/cpuinfo | wc -l
8
[acme@emilia ~]$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <20090526191904.GH4424@ghostprotocols.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

3a4b8cc7

perf report: More robust error handling · 6142f9ec

Peter Zijlstra authored May 26, 2009

Don't let funny events confuse us, stick to what we know and
try to find sensible data again.

If we find an unknown event, check we're still u64 aligned, and
increment by one u64. This ensures we're bound to happen upon a
valid event soon.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

6142f9ec

perf report: add more debugging · f49515b1

Ingo Molnar authored May 26, 2009

Add the offset of the file we are analyzing, and the size of the record.

In case of problems it's easier to see where the parser lost track.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

f49515b1

perf report: add counter for unknown events · 3e706114

Ingo Molnar authored May 26, 2009

Add a counter for unknown event records.

[ Impact: improve debugging ]

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

3e706114

perf report: add --dump-raw-trace option · 97b07b69

Ingo Molnar authored May 26, 2009

To help the inspection of various data files, implement an ASCII dump
method that just dumps the records as they are read in - then we exit.

[ Impact: new feature ]

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

97b07b69

perf: Don't assume /proc/kallsyms is ordered · abd54f68

Arnaldo Carvalho de Melo authored May 26, 2009

perf: Don't assume /proc/kallsyms is ordered

Since we _are_ ordering it by the symbol start, just traverse the
freshly built rbtree setting the prev->end members to curr->start - 1.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090526152134.GF4424@ghostprotocols.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

abd54f68

perf report: Fix kernel symbol resolution · 59d81029

Arnaldo Carvalho de Melo authored May 26, 2009

kallsyms have just the symbol start, so we need to read two lines
to get the len.

[ Impact: fix incorrect kernel symbol display in perf report ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

59d81029

perf report: Fix ELF symbol parsing · f17e04af

Peter Zijlstra authored May 26, 2009

[ Impact: fix DSO symbol output in perf report ]
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

f17e04af

perf top: fix typo in -d option · db20c003

Mike Galbraith authored May 26, 2009

Clean up copy/paste options parsing conversion error.

[ Impact: reactivate -d option ]
Signed-off-by: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

db20c003

perf top: Remove leftover NMI/IRQ bits · f91183fe

Mike Galbraith authored May 26, 2009

79202b removed IRQ/NMI mode selection, so remove it from
perf top as well.

[ Impact: cleanup ]
Signed-off-by: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

f91183fe

perf report: Add help/manpage · 0bec253c

Ingo Molnar authored May 26, 2009

Add a (minimal) manpage for perf report.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

0bec253c

perf record: Convert to Git option parsing · 53cb8bc2

Ingo Molnar authored May 26, 2009

Remove getopt usage and use Git's much more advanced and more compact
command option library.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

53cb8bc2

perf report: Fix segfault on unknown symbols · f3e08c53

Peter Zijlstra authored May 22, 2009

Ingo reported:

> Program received signal SIGSEGV, Segmentation fault.
> 0x0000003e25080f80 in strlen () from /lib64/libc.so.6
> Missing separate debuginfos, use: debuginfo-install elfutils.x86_64
> glibc.x86_64 zlib.x86_64
> (gdb) bt
> #0  0x0000003e25080f80 in strlen () from /lib64/libc.so.6
> #1  0x0000003e2506954e in fputs () from /lib64/libc.so.6
> #2  0x00000000004059e8 in cmd_report (argc=<value optimized out>,
>     argv=<value optimized out>) at builtin-report.c:521
> #3  0x0000000000402dad in handle_internal_command (argc=1, argv=0x7fffe1218e30)
>     at perf.c:226
> #4  0x0000000000402f6d in main (argc=1, argv=0x7fffe1218e30) at perf.c:324
> (gdb)

Signed-off-by Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

f3e08c53

perf_counter: Use rb_tree for symhists and threads in report · ce7e4365

Arnaldo Carvalho de Melo authored May 19, 2009

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

ce7e4365

perf_counter: Add our private copy of list.h · 040e6034

Arnaldo Carvalho de Melo authored May 18, 2009

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

040e6034

perf_counter: Use rb_trees in perf report · 35a50c8a

Arnaldo Carvalho de Melo authored May 18, 2009

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

35a50c8a

perf_counter: Implement dso__load using libelf · 62eb9390

Arnaldo Carvalho de Melo authored May 18, 2009

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

62eb9390

perf_counter tools: remove the standalone perf-report utility · fd4242bb

Ingo Molnar authored May 20, 2009

With a built-in 'perf report' command now available, remove the
standalone implementation for good.
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

fd4242bb

perf_counter: First part of 'perf report' conversion to C + elfutils · 8fa66bdc

Arnaldo Carvalho de Melo authored May 18, 2009

Integrate perf-report into 'perf', as builtin-report.c.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

8fa66bdc

perf top: Convert to Git option parsing · b456bae0

Ingo Molnar authored May 26, 2009

Remove getopt usage and use Git's much more advanced and more compact
command option library.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

b456bae0

perf stat: Convert to Git option parsing · 5242519b

Ingo Molnar authored May 26, 2009

Remove getopt usage and use Git's much more advanced and more compact
command option library.

Extend the event parser library with the extensions that were in
perf-stat before.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

5242519b

perf_counter tools: Librarize event string parsing · 8ad8db37

Ingo Molnar authored May 26, 2009

Extract the event string parser from builtin-record.c, and
librarize it - to be reused in other commands.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

8ad8db37

perf record: Convert to Git option parsing · 0e9b20b8

Ingo Molnar authored May 26, 2009

Remove getopt usage and use Git's much more advanced and more compact
command option library.

Git's library (util/parse-options.[ch]) constructs help texts and
error messages automatically, and has a number of other convenience
features as well.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

0e9b20b8

perf stat: Remove unused variable · 4e97ddf0

Ingo Molnar authored May 26, 2009

[ Impact: cleanup ]

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

4e97ddf0

perf record: Straighten out argv types · 69aa48ab

Ingo Molnar authored May 26, 2009

[ Impact: cleanup ]

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

69aa48ab

perf_counter: Initialize ->oncpu properly · 329d876d

Ingo Molnar authored May 26, 2009

This shouldnt matter normally (and i have not seen any
misbehavior), because active counters always have a
proper ->oncpu value - but nevertheless initialize the
field properly to -1.

[ Impact: cleanup ]

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

329d876d

perf_counter, x86: Make NMI lockups more robust · aaba9801

Ingo Molnar authored May 26, 2009

We have a debug check that detects stuck NMIs and returns with
the PMU disabled in the global ctrl MSR - but i managed to trigger
a situation where this was not enough to deassert the NMI.

So clear/reset the full PMU and keep the disable count balanced when
exiting from here. This way the box produces a debug warning but
stays up and is more debuggable.

[ Impact: in case of PMU related bugs, recover more gracefully ]

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

aaba9801

perf_counter, x86: Fix APIC NMI programming · 79202ba9

Ingo Molnar authored May 26, 2009

My Nehalem box locks up in certain situations (with an
always-asserted NMI causing a lockup) if the PMU LVT
entry is programmed between NMI and IRQ mode with a
high frequency.

Standardize exlusively on NMIs instead.

[ Impact: fix lockup ]

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

79202ba9

perf_counter: powerpc: Implement interrupt throttling · 8a7b8cb9

Paul Mackerras authored May 26, 2009

This implements interrupt throttling on powerpc.  Since we don't have
individual count enable/disable or interrupt enable/disable controls
per counter, this simply sets the hardware counter to 0, meaning that
it will not interrupt again until it has counted 2^31 counts, which
will take at least 2^30 cycles assuming a maximum of 2 counts per
cycle.  Also, we set counter->hw.period_left to the maximum possible
value (2^63 - 1), so we won't report overflows for this counter for
the forseeable future.

The unthrottle operation restores counter->hw.period_left and the
hardware counter so that we will once again report a counter overflow
after counter->hw.irq_period counts.

[ Impact: new perfcounters robustness feature on PowerPC ]
Signed-off-by: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
LKML-Reference: <18971.35823.643362.446774@cargo.ozlabs.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

8a7b8cb9

25 May, 2009 10 commits

perf_counter: fix warning & lockup · 0127c3ea

Ingo Molnar authored May 25, 2009

 - remove bogus warning
 - fix wakeup from NMI path lockup
 - also fix up whitespace noise in perf_counter.h

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525153931.703093461@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

0127c3ea

Revert "perf_counter, x86: speed up the scheduling fast-path" · 53b441a5

Ingo Molnar authored May 25, 2009

This reverts commit b68f1d2e.

It is causing problems (stuck/stuttering profiling) - when mixed
NMI and non-NMI counters are used.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525153931.703093461@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

53b441a5

perf_counter: Generic per counter interrupt throttle · a78ac325

Peter Zijlstra authored May 25, 2009

Introduce a generic per counter interrupt throttle.

This uses the perf_counter_overflow() quick disable to throttle a specific
counter when its going too fast when a pmu->unthrottle() method is provided
which can undo the quick disable.

Power needs to implement both the quick disable and the unthrottle method.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525153931.703093461@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

a78ac325

perf_counter: x86: Remove interrupt throttle · 48e22d56

Peter Zijlstra authored May 25, 2009

remove the x86 specific interrupt throttle
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525153931.616671838@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

48e22d56

perf_counter: x86: Expose INV and EDGE bits · ff99be57

Peter Zijlstra authored May 25, 2009

Expose the INV and EDGE bits of the PMU to raw configs.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525153931.494709027@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

ff99be57

perf_counter: Fix PERF_COUNTER_CONTEXT_SWITCHES for cpu counters · 10989fb2

Peter Zijlstra authored May 25, 2009

Ingo noticed that cpu counters had 0 context switches, even though
there was plenty scheduling on the cpu.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124600.419025548@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

10989fb2

perf_counter: Propagate inheritance failures down the fork() path · 6ab423e0

Peter Zijlstra authored May 25, 2009

Fail fork() when we fail inheritance for some reason (-ENOMEM most likely).
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124600.324656474@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

6ab423e0

perf_counter: Make pctrl() affect inherited counters too · 771d7cde

Peter Zijlstra authored May 25, 2009

Paul noted that the new ptcrl() didn't work on child counters.
Reported-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124600.203151469@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

771d7cde

perf_counter: Remove unused ABI bits · e527ea31

Peter Zijlstra authored May 25, 2009

extra_config_len isn't used for anything, remove it.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124600.116035832@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

e527ea31

perf_counter: Fix perf-$cmd invokation · 266dfb0b

Peter Zijlstra authored May 25, 2009

Fix:

  $ perf-top
  fatal: cannot handle -top internally
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <20090525124559.995591577@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

266dfb0b