lkml.org 
[lkml]   [2009]   [Jul]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RESEND Patch] kcore: remove its pointless size
On Fri, 17 Jul 2009 15:29:55 -0700
Andrew Morton <akpm@linux-foundation.org> wrote:

> On Thu, 2 Jul 2009 09:41:38 +0900
> KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> I think I'm about to forget about the above issues. If everyone else
> does the same, they won't get addressed. Oh well.
>
> And I still need to decide whether
> kcore-fix-proc-kcores-statst_size.patch fixes things up sufficiently
> well to justify merging it.
>

Hmm, I read fs/proc/kcore.c and feel followng, now.

- kclist doesn't handle memory hole, then, it will never be "correct" size.
For example, arch/x86/mm/init.c calls kclist_add() as following

715 kclist_add(&kcore_vmalloc, (void *)VMALLOC_START,
716 VMALLOC_END-VMALLOC_START);

Wow, extremely big anyway.

- Then, yes. Size of /proc/kcode is pointless. Anyway, what's important is
not "size", but ELF phder of kcore.

To this patch,
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

BTW, I'd like to look into handling physical memory range for /proc/kcore.
IMHO, kclist for physical memory is not necessary...it's handled by /proc/iomem.
"kdump" uses this information and it's properly maintained by memory hotplug.
I'd like to try some pathces and make kclist_add() for physical memory cleaner,
later.

Thanks,
-Kame









\
 
 \ /
  Last update: 2009-07-21 04:13    [from the cache]
©2003-2011 Jasper Spaans