On Wed, 21 May 2008 15:44:41 -0400
Benjamin Herrenschmidt [off-list ref] wrote:

On Wed, 2008-05-21 at 11:41 -0700, Andrew Morton wrote:

quoted

yup, gcc bug.  Discussed recently on lkml, "Subject: Re: huge gcc
4.1.{0,1} __weak problem".  I don't think anything ended up happening
about it though.

Hrm... do you think we should work around ? ie. move the stubs to a
separate .c file ?

istr that sticking an asm(""); in the weak function was a reliable
workaround.  If we are going to to that it should be via

/* comment goes here */
#define gcc_screws_up_weak_stuff() asm("")

but that approach didn't seem very popular.  It's a _bit_ fragile I
guess, but it's pretty easy to grep for missed workarounds.