linux-kernel - Re: [PATCH] perf documentation: Add desription forHEADER_TRACING

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] [day] [month] [year] [list]

Message-Id: <747c8b92-a342-3e0c-4232-19df02250496@linux.ibm.com>
Date:   Thu, 12 Mar 2020 12:48:31 +0100
From:   Thomas Richter <tmricht@...ux.ibm.com>
To:     Arnaldo Carvalho de Melo <arnaldo.melo@...il.com>
Cc:     linux-kernel@...r.kernel.org, linux-perf-users@...r.kernel.org,
        gor@...ux.ibm.com, sumanthk@...ux.ibm.com,
        heiko.carstens@...ibm.com
Subject: Re: [PATCH] perf documentation: Add desription forHEADER_TRACING_DATA

On 3/11/20 2:58 PM, Arnaldo Carvalho de Melo wrote:
> Em Tue, Mar 03, 2020 at 08:08:46AM +0100, Thomas Richter escreveu:
>> Add description and layout in the perf.data file for
>> the header part describing trace data used in commands
>> perf record -e XXX
>> where XXX is a probe or tracepoint.
> 
> Did you write this from scratch?  I was going thru it and couldn't find
> what that "Note member 'ftrace_count' can be zero." refers to, couldn't
> find this ftrace_counter, is this outdated?
> 
> [root@...e ~]# find /sys/kernel/debug/tracing/events/ -type f | xargs grep ftrace_count
> [root@...e ~]# find /sys/kernel/debug/tracing/events/ -name ftrace_count
> [root@...e ~]# uname -a
> Linux five 5.5.5-200.fc31.x86_64 #1 SMP Wed Feb 19 23:28:07 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
> [root@...e ~]#
> 
> The part about using a probe and then go on dissecting it is really
> nice, I'll try reproducing it soon,
> 
> Thanks,
> 

Arnaldo,
        
yes, I wrote this from scratch. I decided to invest some time into the
documentation after I have spent nearly 3 week in debugging why commands

  perf record -e probe:vfs_getname

followed by

  perf report 
  
did not print the file name on s390. Remember we talked about failing
test case 65: Check open filename arg using perf trace + vfs_getname.
(Steven Rostedt and Masami Hiramatsu helped me on this).

On ftrace_count: This is not a file available in /sys/kernel/debug/tracing.
It is a value saved in function

tracing_data_get
  record_ftrace_files
     copy_event_system --> Check for some format file in directories
                           /sys/kernel/debug/tracing/events/ftrace/*/
Depending on what probe was defined, there is no hit and zero is written
to the output.

Hope this helps.
-- 
Thomas Richter, Dept 3252, IBM s390 Linux Development, Boeblingen, Germany
--
Vorsitzender des Aufsichtsrats: Matthias Hartmann
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen / Registergericht: Amtsgericht Stuttgart, HRB 243294