public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* 1GB piggys
@ 2002-05-21 11:40 Russell King
  2002-05-21 14:37 ` Kai Germaschewski
  0 siblings, 1 reply; 2+ messages in thread
From: Russell King @ 2002-05-21 11:40 UTC (permalink / raw)
  To: linux-kernel

Hi,

I'm getting reports of some people ending up with 1GB piggy objects for
gzip to compress on ARM.  While investigating this, I found the following:

00003dd4 ? __module_kernel_version
00003df4 ? __module_parm_minor
00003e04 ? __module_parm_flash_base
00003e18 ? __module_parm_flash_size
40004000 A swapper_pg_dir
40008000 ? __init_begin
40008000 ? _stext
40008000 ? stext

The __module_* stuff is from the .modinfo section of some kernel object.

Since .modinfo isn't actually used or indeed placed in the linker
script, shouldn't we be explicitly discarding it like the
.exitcall.exit, .text.exit and .data.exit sections, rather than
letting the linker apparantly pick some random memory location to
dump this section?

(Yes, its probably buggy for the linker not to moan about unrecognised
input sections...)

-- 
Russell King (rmk@arm.linux.org.uk)                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: 1GB piggys
  2002-05-21 11:40 1GB piggys Russell King
@ 2002-05-21 14:37 ` Kai Germaschewski
  0 siblings, 0 replies; 2+ messages in thread
From: Kai Germaschewski @ 2002-05-21 14:37 UTC (permalink / raw)
  To: Russell King; +Cc: linux-kernel

On Tue, 21 May 2002, Russell King wrote:

> I'm getting reports of some people ending up with 1GB piggy objects for
> gzip to compress on ARM.  While investigating this, I found the following:
> 
> 00003dd4 ? __module_kernel_version
> 00003df4 ? __module_parm_minor
> 00003e04 ? __module_parm_flash_base
> 00003e18 ? __module_parm_flash_size
> 40004000 A swapper_pg_dir
> 40008000 ? __init_begin
> 40008000 ? _stext
> 40008000 ? stext
> 
> The __module_* stuff is from the .modinfo section of some kernel object.

Well, the .modinfo section should only be generated for objects build as 
modules, not into the kernel. So it very much likes some object has been 
compiled with -DMODULE, though it gets linked into the kernel.

(Which one is it? Obviously it has MODULE_PARM(minor/flash_base/size) in 
it)

> Since .modinfo isn't actually used or indeed placed in the linker
> script, shouldn't we be explicitly discarding it like the
> .exitcall.exit, .text.exit and .data.exit sections, rather than
> letting the linker apparantly pick some random memory location to
> dump this section?

Well, discarding .modinfo would only hide errors like this, so I wouldn't 
think it's a good idea.

--Kai



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2002-05-21 14:37 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2002-05-21 11:40 1GB piggys Russell King
2002-05-21 14:37 ` Kai Germaschewski

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox