------- Comment #2 from hubicka at ucw dot cz 2010-04-30 08:45 -------
Subject: Re: [4.6 Regression] FAIL:
gcc.dg/lto/20090126-1 c_lto_20090126-1_0.o-c_lto_20090126-1_0.o
Hi,
the problem here is that the testcase combine -O0 module into -O3 module. This
results in optimization summaries to be missing and thus ipa-pure-const reads
in nothing and later ICE on uninitialized datastructure. I sent patch for this
that I need to modify, just did not get into it. WHOPR needs significant
reorganization (well pretty much rewrite of all WHOPR specific bits) to be
correct and it is difficult to do so without some fallout
- fixing one problem enabling a transform usually uncovers new one. The
testcase should ICE on release branch alsi when -fwhopr is replaced by -flto
so it is a bug in LTO implementation overall. Trying to optimize -O0 compiled
modules is in general not very sane anyway.
But it should be attributed to the change enabling ipa-pure-const at -fwhopr.
I will handle this over weekend.
Honza
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43853
