> On Fri, 29 Aug 2008, Jan Hubicka wrote: > > > Hi, > > tonight testing on x86_64, i386 and IA-64 didn't seem to bring any new > > surprises, so I've comitted the following patch. > > I will also update changes page of 4.4. > > > > * doc/invoke.texi (-fipa-cp): Enabled by default at -O2/-Os/-O3 > > (-fipa-cp-clone): Enabled by default at -O3. > > * opts.c (decode_options): Enable ipa-cp at -O2, ipa-cp-clone at -O3; > > make ipa-cp-clone to imply ipa-cp; disable cloning at -Os. > > This broke Ada bootstrap: > > /space/rguenther/obj/./prev-gcc/xgcc -B/space/rguenther/obj/./prev-gcc/ > -B/usr/local/x86_64-unknown-linux-gnu/bin/ -c -g -O2 -gnatpg -gnata > -nostdinc -I- -I. -Iada -I../../trunk/gcc/ada > -I../../trunk/gcc/ada/gcc-interface ../../trunk/gcc/ada/sem_ch12.adb -o > ada/sem_ch12.o
Sorry fro that. I am investigating now. The same problem appeared earlier but disappeared again, it seems to be related to inlining (the ICE happens in inlining not in IPCP clonning) Looks like Ada can fold statements into constant but we don't notice that the function has address taken. Honza
