Re: [PATCHv2 0/4] perf, tool: Fix endian issues

[Jiri Olsa <jolsa@redhat.com>]

On Fri, May 04, 2012 at 02:59:40PM -0300, Arnaldo Carvalho de Melo wrote:
> Em Fri, May 04, 2012 at 11:40:17AM -0600, David Ahern escreveu:
> > On 5/4/12 11:11 AM, Jiri Olsa wrote:
> > >hi,
> > >sending fixies to properly handle perf.data endians.
> > >
> > >v2 changes:
> > >   - added patches 3 and 4 to handle sample_id_all header endianity
> > >
> > >Attached patches:
> > >   1/4 perf, tool: Handle different endians properly during symbol load
> > >   2/4 perf, tool: Carry perf_event_attr bitfield throught different endians
> > >   3/4 perf, tool: Handle endianity swap on sample_id_all header data
> > >   4/4 perf, tool: Fix 32 bit values endianity swap for sample_id_all header
> > 
> > Have you tested these patches by collecting data on PPC (or other
> > big endian) and analyzing on Intel x86 and vice versa?
> 
> I think so, but indeed it is good practice to show the sequence made to
> test the changes, with sample, short command output before and after.
> 
> - Arnaldo

yep, I should have mentioned that..

I did following testing with x86_64 and ppc64:

1. perf record -a -- sleep 10
2. perf report > report.origin
3. perf archive perf.data
4. copy the perf.data.tar.bz2 to a target system
5. tar xjvf perf.data.tar.bz2 -C ~/.debug
6. perf report > report.target
7. diff -u report.origin report.target

the diff should produce no output

hm, I did jsut x86_64 record and ppc64 report.. will do also
the other way around and resend with updated changelog 

thanks,
jirka