Re: [patch] Verify file vs. core build-id, when both available

On Wed, 2013-11-20 at 20:32 +0100, Jan Kratochvil wrote: > The problem is that: > * elfutils will find a completely different version of file with the same > name > * the file has build-id > * as the file is different the computed memory address of the build-id is > bogus > * bogus address is

[commit] [patch] link_map: Use proper bias, not l_addr

On Thu, 21 Nov 2013 15:08:07 +0100, Mark Wielaard wrote: > Looks good. Checked in: 463e983e20fdb945b11bb175ca71e23b31146d8e Thanks, Jan

[commit] [patch] Fix non-build-id core files on build-id system

On Tue, 19 Nov 2013 11:13:30 +0100, Mark Wielaard wrote: > Yes, this makes sense. This case should be handled the same as if there > was no build-id in the first place. Which your patch does. Checked in: 99a29497c673871040fed2382938eb5bd06befdd Thanks, Jan

[commit] [patch] Verify file vs. core build-id, when both available

On Thu, 21 Nov 2013 14:56:56 +0100, Mark Wielaard wrote: > Yes, looks good. Checked in: b301c2efeab0a712570bfd72ae5e07a63f1ba784 Thanks, Jan

Re: [patch] link_map: Use proper bias, not l_addr

Hi Jan, On Wed, 2013-11-20 at 20:41 +0100, Jan Kratochvil wrote: > All the test results are sure very dependent on host files version and also > their prelink status. Yes, that is somewhat unfortunate. > link_map: Use proper bias, not l_addr > > libdwfl/ > 2013-11-20 Jan Kratoc