Re: + binfmt-introduce-coredump-parameter-structure.patch added to -mm tree

[Masami Hiramatsu <mhiramat@redhat.com>]

Oleg Nesterov wrote:
> On 11/26, Masami Hiramatsu wrote:
>>
>> $ grep -r DUMP_WRITE arch/*/include
>> arch/ia64/include/asm/elf.h:            DUMP_WRITE(&phdr, sizeof(phdr));       \
>> arch/ia64/include/asm/elf.h:                    DUMP_WRITE((void *) gate_phdrs[\
>
> arch/um/sys-i386/asm/elf.h uses DUMP_WRITE() too.
>
>> Oops, certainly, that's a problem.
>> IMHO, we should not do like that, all parameter required by a macro should be
>> specified explicitly, since it reduces readability so much...
>> I think we'd better make those macros inline function, check it's return value
>> for error handling.
>
> Agreed, DUMP_WRITE() in its current form should die. Not only
> it has implicit parameter, it does "goto" from the macro body
> and it has multiple definitions withing the same file.
>
> But perhaps this needs a separate patch? It is not trivial to kill
> DUMP_WRITE(), you can fix this patch if you change DUMP_WRITE()
> to use cprm->file instead of file.

Hmm, certainly, that should be separated. So I just change DUMP_WRITE()
to use cprm->limit/file.

Thank you,

-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com