Yes, that is the reason. One of our generated source file is compiled with -O0 with no particular reason and no one noticed. Thanks.
Cheers, Bingfeng > -----Original Message----- > From: Richard Guenther [mailto:richard.guent...@gmail.com] > Sent: 19 May 2010 16:12 > To: Bingfeng Mei > Cc: gcc@gcc.gnu.org > Subject: Re: ICE in LTO. > > On Wed, May 19, 2010 at 4:12 PM, Bingfeng Mei > <b...@broadcom.com> wrote: > > Hello, > > I am hit by an ICE in LTO (latest GCC4.5 branch). The problem > > is that it shows up in our target/our target-specific application, > > so I cannot really reduce test case and file a bug. > > > > > > getShortVersionString.clone.0/1(-1) @0x2aa0c5fea0 (clone of > getShortVersionString/4733) availability:local 17 time, 12 > benefit 8 size, 3 benefit reachable local finalized > > called by: CoreMgr_init/109 (1.00 per call) > > calls: memcpy/4730 (1.00 per call) > > callgraph: > > ... > > > > lto1: internal compiler error: in propagate, at ipa-reference.c:1245 > > Please submit a full bug report, > > with preprocessed source if appropriate. > > See <http://gcc.gnu.org/bugs.html> for instructions. > > > > It seems that the cloned function lacks node_info as indicated > > by ip-reference.c:1245. I am not familiar with these stuff. Any > > suggestion on where I should look at? Thanks in advance. > > Are you using -flto or -fwhopr? Are you mixing -O0 and non-O0 > at compile vs. link time? Which would make it PR42402. > > You should still be able to provide a reduced testcase. > > Richard. > > > Cheers, > > Bingfeng Mei > > > >
