[PATCH] Omit frame pointer and fix %ebp by default on x86 (take 3)

Jakub Jelinek jakub@redhat.com
Thu Aug 19 19:32:00 GMT 2004


On Thu, Aug 19, 2004 at 07:20:29PM +0100, Andrew Haley wrote:
>  > 	Note that certain compiler optimizations may interfere with
>  > 	obtaining a valid backtrace.
> 
> Ah, okay.  You have a very strong argument here.  Perhaps some rather
> stronger wording is required in glibc documentation, but clearly
> that's a red flag for anyone who wants to look.
> 
> Fair enough.  I suppose we should change backtrace() in glibc so that
> it can work in the same way on x86 as x86_64.

But it can only work the same way as on x86-64 if
-fasynchronous-unwind-tables (or -funwind-tables, but the latter
for some strange reason seems to generate often bigger .eh_frame)
is on by default.

I hope I'll be able to get exact data on binary and shared library sizes
of all distribution packages soon.

	Jakub



More information about the Gcc-patches mailing list