This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: question about spill failure - I smell a wumpus :-)
- To: apl at alum dot mit dot edu
- Subject: Re: question about spill failure - I smell a wumpus :-)
- From: Joern Rennecke <amylaar at redhat dot com>
- Date: Fri, 22 Jun 2001 21:28:15 +0100 (BST)
- Cc: gcc at gcc dot gnu dot org, amylaar at redhat dot com
> I can't think of ANYTHING where the register allocation should
> remember any state between functions (other than global registers,
> which we're not using....) I suppose I could start up TWO copies of
> gdb and parallel step my way thru the reload pass. But I wonder if
> anyone has any cleverer ideas?
It's usually quicker if you go backwards and / or do a binary search.
I.e. check if the reloads for insn 30 match in both compilations;
if yes, go forward and find why the allocation failed - there would
probably be a register that was available for one function ordering
but not the other, or something like that...
if, on the other hand, the reloads are different, pick the first
reload that is different, and find out why it was made this way only
for one function ordering...