http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55910
--- Comment #7 from Oleh Derevenko oder at eleks dot lviv.ua 2013-01-09
17:26:37 UTC ---
OK. If you are unable to track where you get size of function parameter block
from, close it. I have much other work to do, rather than trying
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55910
--- Comment #8 from Oleh Derevenko oder at eleks dot lviv.ua 2013-01-09
17:42:19 UTC ---
And go on telling your tales about numerous optimizations that affect argument
count on stack and cleanup size of final 'ret'...
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55910
Bug #: 55910
Summary: One of instances generated for a method with -O3 uses
incorrect parameter cleanup size with 'ret'
Classification: Unclassified
Product: gcc
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55910
--- Comment #3 from Oleh Derevenko oder at eleks dot lviv.ua 2013-01-08
17:51:59 UTC ---
Well, this is an optimizer issue which goes away with -O2. And optimizer is
something that is hard to predict and reproduce, when you run over an issue
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55910
--- Comment #5 from Oleh Derevenko oder at eleks dot lviv.ua 2013-01-08
18:11:42 UTC ---
IMHO, I've localized the issue clearly enough. The code that generates
instances for methods may sometimes consider some of those instances as having