1. 01 Jun, 2009 1 commit
  2. 31 May, 2009 1 commit
    • Ingo Molnar's avatar
      perf_counter tools: Fix unknown command help text · 27b9613b
      Ingo Molnar authored
      Arjan reported this error when entering an unknown command to perf:
      
        $ perf start
        fatal: Uh oh. Your system reports no Git commands at all.
      
      The Git code expects there to be perf-* commands - but since Perf
      is a 'pure' utility with no dash commands anymore, this old assumption
      of Git does not hold anymore. Remove that error check.
      Reported-by: default avatarArjan van de Ven <arjan@linux.intel.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: default avatarIngo Molnar <mingo@elte.hu>
      27b9613b
  3. 30 May, 2009 4 commits
    • Ingo Molnar's avatar
      perf_counter tools: Generate per command manpages (and pdf/html, etc.) · c1c2365a
      Ingo Molnar authored
      Import Git's nice .txt => {man/html/pdf} generation machinery.
      
      Fix various errors in the Documentation/perf*.txt description as well.
      
      Also fix a bug in builtin-help: we'd map 'perf help top' to 'perftop'
      if only the 'perf' binary is in the default PATH - confusing the manpage
      logic. I dont fully understand why Git did it this way - but i suppose
      it's a migration artifact from their migration from standalone git-xyz
      commands to 'git xyz' commands. The perf tools were always using the
      modern form so it's not an issue there.
      
      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: default avatarIngo Molnar <mingo@elte.hu>
      c1c2365a
    • Ingo Molnar's avatar
      perf_counter tools: Fix 'make install' · 7fbd5544
      Ingo Molnar authored
      'make install' didnt install perf itself - which needs a special
      rule to be copied to bindir.
      
      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: default avatarIngo Molnar <mingo@elte.hu>
      7fbd5544
    • Ingo Molnar's avatar
      perf_counter tools: Print 'CPU utilization factor' in builtin-stat · d7c29318
      Ingo Molnar authored
      Before:
      
       Performance counter stats for '/home/mingo/hackbench':
      
          5728.862689  task clock ticks     (msecs)
                34426  context switches     #        0.006 M/sec
                 3835  CPU migrations       #        0.001 M/sec
                18158  pagefaults           #        0.003 M/sec
          16218109156  CPU cycles           #     2830.947 M/sec
          13519616840  instructions         #     2359.913 M/sec
             55941661  cache references     #        9.765 M/sec
             23554938  cache misses         #        4.112 M/sec
      
       Wall-clock time elapsed:   528.886980 msecs
      
      After:
      
       Performance counter stats for '/home/mingo/hackbench':
      
          5845.443541  task clock ticks     #      11.886 CPU utilization factor
                38289  context switches     #       0.007 M/sec
                 4208  CPU migrations       #       0.001 M/sec
                17755  pagefaults           #       0.003 M/sec
          16664668576  CPU cycles           #    2850.882 M/sec
          13468113991  instructions         #    2304.036 M/sec
             57445468  cache references     #       9.827 M/sec
             26896502  cache misses         #       4.601 M/sec
      
       Wall-clock time elapsed:   491.802357 msecs
      
      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: default avatarIngo Molnar <mingo@elte.hu>
      d7c29318
    • Arnaldo Carvalho de Melo's avatar
      perf_counter tools: Add locking to perf top · c44613a4
      Arnaldo Carvalho de Melo authored
      perf_counter tools: Add locking to perf top
      
      We need to protect the active_symbols list as two threads change it:
      the main thread adding entries to the head and the display thread
      decaying entries from any place in the list.
      
      Also related: take a snapshot of syme->count[0] before using it to
      calculate the weight and to show the same number used in this calc when
      displaying the symbol usage.
      Reported-by: default avatarMike Galbraith <efault@gmx.de>
      Tested-by: default avatarMike Galbraith <efault@gmx.de>
      Signed-off-by: default avatarArnaldo Carvalho de Melo <acme@redhat.com>
      Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
      Cc: Paul Mackerras <paulus@samba.org>
      Cc: Steven Rostedt <rostedt@goodmis.org>
      LKML-Reference: <20090529200307.GR4747@ghostprotocols.net>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      c44613a4
  4. 29 May, 2009 34 commits