On Sat, 05 Jun 2010, Guy Martin wrote: > > Hi Dave, > > I've stripped down gengenrtl.c to the maximum and I'm able to reproduce > the issue with the attached file. > > I've observed the following : > gcc -O2 -> Segv > gcc -O2 -fno-delayed-branch -> OK > gcc -O1 -> OK > gcc -O1 -freorder-blocks -> Segv > gcc -O1 -freorder-blocks -fno-delayed-branch -> OK > gcc -O0 -freorder-blocks -fdelayed-branch -> OK > > For reference, the build break when compiled the following way : > make -j2 LDFLAGS="-Wl,-O1" STAGE1_CFLAGS="-O" > LIBPATH=/usr/lib/gcc/hppa2.0-unknown-linux-gnu/4.5.0 BOOT_CFLAGS="-O2 > -pipe -march=2.0" bootstrap-lean I haven't had a chance to try the your stripped down gengenrtl.c but there is definitely a problem compiling the 4.5 branch with 4.4 using STAGE1_CFLAGS="-O". This didn't occur with 4.3. I get an ICE in stage1 configuration of libgcc with 4.4. At this point, I don't know the cause but this is bad as it indicates a major bug with 4.4 on hppa. -fno-reorder-blocks didn't help. As a general rule, it is not a good idea to build stage1 with optimisation. It is not tested. I get a successful bootstrap at -O0. I would guess 4.5 can probably compile 4.5 with -O, but I haven't tested it. Dave -- J. David Anglin dave.anglin@xxxxxxxxxxxxxx National Research Council of Canada (613) 990-0752 (FAX: 952-6602) -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
