RFA: patch to fix bootstrap failure
Vladimir Makarov
vmakarov@redhat.com
Thu Sep 25 21:14:00 GMT 2008
Jeff Law wrote:
> Vladimir Makarov wrote:
>> The following patch fixes sporadical i686 bootstrap failure. The
>> problem was in reading garbage in an array because of a wrong
>> comparison.
>>
>> H.J. I've committed it into ira-merge-branch.
>>
>> 2008-09-24 Vladimir Makarov <vmakarov@redhat.com>
>>
>> * ira-color.c (coalesce_spill_slots): Fix typo in comparison with
>> ira_reg_equiv_len.
>>
>
>
> You rewrote part of coalesce_spill_slots in the change for PR37488 and
> inverted the condition of the first IF within the "j" loop and
> replaced "<" with ">" instead of ">=", right?
>
You probably mean 37448. Yes, that is right. That is when the typo
occurred. Using > means reading garbage right after the array end. It
is pretty rare situation when wrong reading results in bootstrap failure.
> The change for PR37488 hasn't been reviewed for the trunk yet, which
> explains why we were only seeing this failure on HJ's branch. I've
> looked at the changes for 37488 a couple times, but haven't wrapped my
> head around them completely.
> I'm going to consider this patch a revision to the changes for pr37488
> and handle them together. I'll try to finish reviewing the changes
> for 37488 today.
>
Ok, Thanks.
More information about the Gcc-patches
mailing list