Hi Richard and Adam,
On Fri, 1 Sep 2006, Adam Nemet wrote:
> Richard Guenther writes:
> > So if 27226 finally is a regression the patch in there looks more complete.
> > Can you verify it works for you?
>
> That's nice. Yes it works and it is a regression with dhrystone as
> Joern has mentioned it. Thanks for noticing this.
Ok, in that case Richard's patch for middle-end/27226, as attached in
comment #13, is OK for mainline, with an appropriate ChangeLog entry,
and for the 4.1 branch after a while on mainline without problems.
Richard could you also add Adam's new MIPS testcase to the testsuite
when you commit your patch. I appreciate that you probably can't test
it, and confirm it works yourself, but its good to have a test to
ensure we don't regress again in future. If anyone can come up with
an x86 test, that'd be even better, but not required.
Thanks to you both.