This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: ppc glibc 2.3 failure


Dale Johannesen <dalej@apple.com> writes:

> On Monday, October 7, 2002, at 04:06  PM, Aldy Hernandez wrote:
> 
> > Hi guys.
> >
> > The following distilled case from glibc from CVS...
> >
> > ...So what we have is rs6000_stack_info behaving different in
> > subsequent
> > runs, and I've traced it to rs6000_ra_ever_killed() giving different
> > results.
> 
> This should not happen.  As the comments in
> rs6000_ra_ever_killed indicate, its expectation is that code (in
> particular,
> copies into LR) added in the prolog have a REG_MAYBE_DEAD note, to
> prevent
> this problem. 

That's not right.  REG_MAYBE_DEAD should only be applied to those
stores in the prologue that might actually become dead, not to every
store or every insn.

The proper way to test if an instruction is in the prologue is to use
the function named "prologue_epilogue_contains".

-- 
- Geoffrey Keating <geoffk@geoffk.org>


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]