With r199422 gcc.dg/tls/emutls-2.c started ICEing with: $SOURCE/gcc/testsuite/gcc.dg/tls/emutls-2.c: In function '_GLOBAL__sub_I_65535_0_emutls_2.c': $SOURCE/gcc/testsuite/gcc.dg/tls/emutls-2.c:8:1: internal compiler error: Segmentation fault 0x916255 crash_signal $SOURCE/gcc/toplev.c:333 0x6575b6 dom_info_available_p(cdi_direction) $SOURCE/gcc/dominance.c:1482 0x657fd3 free_dominance_info(cdi_direction) $SOURCE/gcc/dominance.c:685 0x86d429 do_per_function $SOURCE/gcc/passes.c:1719 Please submit a full bug report, with preprocessed source if appropriate. Please include the complete backtrace with any bug report. See <http://gcc.gnu.org/bugs.html> for instructions. compiler exited with status 1 on arm-none-eabi. Cross compiler configured with: --with-fpu=neon-vfpv4 --with-float=hard --with-arch=armv7-a
It looks similar to pr57467.
Fails also on x86_64-apple-darwin10.
Confirming; cris-elf too (r199421:199433], being an emutls target.
Seems to have been fixed with the "Symtab cleanups 2/17 - merge alias code" fix (r199577) or possibly with the same fix as for PR57467 (r199582).
(In reply to Hans-Peter Nilsson from comment #4) > Seems to have been fixed with the "Symtab cleanups 2/17 - merge alias code" > fix (r199577) or possibly with the same fix as for PR57467 (r199582). Fixed on arm-none-eabi it seems. Dominique, can you confirm that x86_64-apple-darwin10 is fixed as well?
Also fails on Solaris 9 with Sun as, will check if current mainline works again. Rainer
> --- Comment #6 from Rainer Orth <ro at gcc dot gnu.org> --- > Also fails on Solaris 9 with Sun as, will check if current mainline works > again. Both i386-pc-solaris2.9 and sparc-sun-solaris2.9 bootstraps have now completed and as of r199603, all emutls related failures are gone. Thanks. Rainer
Fixed on trunk.