[PATCH, i386]: Fix recent i386 testsuite failures

[Mike Stump]

On Apr 3, 2007, at 12:39 PM, Jan Hubicka wrote:
> We can easilly make always inline fucntions not to be emit with - 
> fkeep-inline functions

Yes.

> or invent new attribute for that.

Ick.

> This recalls me an issue that our inline functions are very gdb  
> unfriendly - if you try to single step any SSE program, you see  
> jumps to and from the xmmintrin.h that is not easy to follow.

Still?  :-(  If someone wanted to work on this, the tthread that  
begins at:

   http://gcc.gnu.org/ml/gcc-patches/2005-07/msg01969.html

has state on how to fix this, now, if one actually has to use the  
tools now, and how to fix it nicely, if you have copious free time.   
We've been shipping that solution for a while now, works nicely for us.

If someone waned to pre-approve that patch minus the semantics, I'd  
get the ball rolling by doing that much, to try and encourage someone  
else to finish it off the right way...