https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123938
Paul Thomas <pault at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |pault at gcc dot gnu.org
--- Comment #7 from Paul Thomas <pault at gcc dot gnu.org> ---
Created attachment 64493
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=64493&action=edit
Regression free patch
It turns out that there was already a boolean 'finalized' that was only set in
its declaration as false. This guards the alloc comp deallocation in the second
chunk of Mikael's fix. I set 'finalized' in the first chunk, which results in a
patch that does the job and passes regression testing.
I have added a dejagnu-ified testcase.
Using parmse.finalblock.head to detect a finalized argument expression with
parmse.finalblock.head seems perfectly acceptable to me.
Andrew, as far as I am concerned, this can be pushed to mainline and backported
later, crediting Mikael as the author.
Regards
Paul
