[Crash-utility] [RFC Patch 6/6] Crash: Recognise slim coredumps and process new elf-note sections

Dave Anderson anderson at redhat.com
Fri May 27 22:30:57 UTC 2011



Can we keep crash utility source code discussions off of LKML 
and kexec-ml please?

----- Original Message -----
>
> Now that machdep->process_elf_notes() is invoked in generic KDUMP
> processing code path, please remove the separate invocation of
> machdep->dumpfile_init() which was introduced for s390x architecture in
> dump_Elf64_Nhdr() function. The reason is, machdep->process_elf_notes()
> on s390x internally invokes machdep->dumpfile_init(). Hence we can
> safely remove it.

That's true -- but I'll let you guys (IBM) handle that...

> The function x86_process_elf_notes() is invoked through
> machdep->process_elf_notes function pointer, which makes it arch
> dependent code. How about moving this function (x86_process_elf_notes)
> to an arch dependent file say x86_common.c ? By doing so we can get
> rid of all "#ifdefs" here.
>
> Hi Dave, what do you say?

Yeah, I suppose we can create x86_common.h and x86_common.c files
to contain this kind of stuff.

In any case, let's see how this slimdump scheme pans out.
This patch was just an RFC, uses #define's that don't even
exist yet, so it's not going in any time soon.

Dave

> Mahesh J Salgaonkar




More information about the Crash-utility mailing list