linux-kernel - Re: [PATCHSET RESEND 0/5] perf tools: Cleanup hist printing code (v4)

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20120908150839.GH20401@ghostprotocols.net>
Date:	Sat, 8 Sep 2012 08:08:39 -0700
From:	Arnaldo Carvalho de Melo <acme@...stprotocols.net>
To:	Namhyung Kim <namhyung@...nel.org>
Cc:	Peter Zijlstra <a.p.zijlstra@...llo.nl>,
	Paul Mackerras <paulus@...ba.org>,
	Ingo Molnar <mingo@...nel.org>,
	LKML <linux-kernel@...r.kernel.org>, Jiri Olsa <jolsa@...hat.com>
Subject: Re: [PATCHSET RESEND 0/5] perf tools: Cleanup hist printing code (v4)

Em Sat, Sep 08, 2012 at 11:00:36PM +0900, Namhyung Kim escreveu:
> On Fri, 7 Sep 2012 17:48:54 -0700, Arnaldo Carvalho de Melo wrote:
> > Please try with:

> >   perf record -a usleep 1
> >   perf record -a usleep 1
> >   perf diff

> > it will use perf.data.old and perf.data and will segfault in that branch.

> I see the problem.  I overlooked he->pair can be NULL when perf diff is
> running.  So the fix will be adding a NULL check before the line.  In
> case of NULL, it will default to 0, so no problem.

Ok, now it doesn't segfaults, but the diff results seems bogus:

Withour this patch + your fix:

[root@...dy ~]# perf diff
# Event 'cycles'
#
# Baseline  Delta          Shared Object                                  Symbol
# ........ ..........  .................  ......................................
#
     0.00%    +39.61%  [kernel.kallsyms]  [k] __alloc_pages_nodemask            
    19.82%     +6.10%  [kernel.kallsyms]  [k] intel_idle                        
     0.00%     +8.61%  [kernel.kallsyms]  [k] generic_exec_single               
     0.00%     +6.39%  [kernel.kallsyms]  [k] list_del                          
     0.00%     +6.15%  [kernel.kallsyms]  [k] nr_iowait_cpu                     
     0.00%     +5.98%  [kernel.kallsyms]  [k] selinux_inode_permission          
     0.00%     +5.23%  libc-2.12.so       [.] _dl_addr                          
     0.00%     +0.76%  [kernel.kallsyms]  [k] group_sched_in                    
     0.10%     +0.32%  [kernel.kallsyms]  [k] native_write_msr_safe             
     0.00%     +0.40%  [kernel.kallsyms]  [k] smp_call_function_single_interrupt
     0.14%     +0.19%  [kernel.kallsyms]  [k] csd_unlock                        
     0.00%     +0.20%  [kernel.kallsyms]  [k] perf_pmu_enable                   

With your patch:

[root@...dy ~]# perf diff
# Event 'cycles'
#
# Baseline   Delta       Shared Object                                  Symbol
# ........  .......  .................  ......................................
#
    39.61%  +39.61%  [kernel.kallsyms]  [k] __alloc_pages_nodemask            
    25.92%   +6.10%  [kernel.kallsyms]  [k] intel_idle                        
     8.61%   +8.61%  [kernel.kallsyms]  [k] generic_exec_single               
     6.39%   +6.39%  [kernel.kallsyms]  [k] list_del                          
     6.15%   +6.15%  [kernel.kallsyms]  [k] nr_iowait_cpu                     
     5.98%   +5.98%  [kernel.kallsyms]  [k] selinux_inode_permission          
     5.23%   +5.23%  libc-2.12.so       [.] _dl_addr                          
     0.76%   +0.76%  [kernel.kallsyms]  [k] group_sched_in                    
     0.42%   +0.32%  [kernel.kallsyms]  [k] native_write_msr_safe             
     0.40%   +0.40%  [kernel.kallsyms]  [k] smp_call_function_single_interrupt
     0.33%   +0.19%  [kernel.kallsyms]  [k] csd_unlock                        
     0.20%   +0.20%  [kernel.kallsyms]  [k] perf_pmu_enable                   
[root@...dy ~]#

The deltas are the same, but the baseline completely changed and are bogus.

So please try again and make sure that the results are kept.

- Arnaldo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/