ASM_OUTPUT_SECTION_NAME cleanup patch problem

[Richard Henderson]

On Thu, Jun 06, 2002 at 01:47:54PM +0100, Joern Rennecke wrote:
> If we had a scheme that drops the exception handling information &
> initialization code when there is no exception handling used anywhere
> in the program, that would be a different matter...

This is already true.

(1) .eh_frame info isn't emitted for C (or -fno-exceptions),
(2) The crtstuff initialization bits call a weak ...  

Oops, this is coff; I see your point.  Well, we could have
stubs in libgcc.a (as opposed to libgcc_eh.a) that do 
nothing.  That would reduce the overhead to perhaps low
hundereds of bytes.

Or I could just fix what I broke.  I guess I'll look at
that next week.


r~