Re: [avr-gcc-list] Re: [avr-libc-dev] [bug #29774] prologue/epilogue stack pointer manipulation not interrupt safe in XMega

[Georg Lay]

Galen Seitz schrieb:

> I think my message to the list was the last mention of it.
> <http://lists.nongnu.org/archive/html/avr-libc-dev/2010-05/msg00002.html>
> 
> It certainly seems like a serious bug, but I'm puzzled as to why it's
> occurring.  The gcc code I looked at appeared to be correct.

(correct CCs in this post)

There are several places that handle SP adjustment:

avr.c: output_movhi, expand_prologue, expand_epilogue
avr.md: movhi_sp_r_irq_off, movhi_sp_r_irq_on (both generated in
epilogue/prologue)
libgcc.S: __prologue_saves__, __epilogue_restores__

For changes in libgcc.S to take effect, appropriate builtin-defines must
be provided so that the right multilib variants will be generated.

As far I can see from the gcc sources, it knows nothing about xmega.
So I guess it is still good practice to have a bunch of patches floating
around that fix this or that. I must admit that I lost track of all the
patches; what they do, what they don't do, where they are incorporated
and where they aren't...