On Sat, Jan 10, 2009 at 3:33 PM, Laurent GUERBY <laur...@guerby.net> wrote: > Hi, > > It looks like the tests for -frtl-abstract-sequences are either skipped > or xfailed for a long list of platforms: > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33642 > "unrecognizable insn for -frtl-abstract-sequences" > << > Almost any source file compiled for any of a large number of targets > (powerpc*-*, arm-eabi, hppa-linux, mipsel-linux, s390*-linux, sh*-*, sparc*-*) > with -O2 -frtl-abstract-sequences results in the following: > > bug1.c: In function 'foo': > bug1.c:8: error: unrecognizable insn: > (insn 28bug1.c:8: internal compiler error: Segmentation fault > Please submit a full bug report, > with preprocessed source if appropriate. > See <http://gcc.gnu.org/bugs.html> for instructions. > > The testcase used here: > > int foo (void) > { > return 0; > } >>> > > gcc/testsuite/gcc.c-torture/compile/pr11832.c > /* { dg-do compile } */ > /* Currently ICEs for Alpha, IA64, HPPA, MIPS, CRIS, Xtensa, PowerPC, SH and > SPARC; see PR33642. */ > /* { dg-xfail-if "PR33642" { alpha*-*-* hppa*-*-* mips*-*-* powerpc*-*-* > cris-*-* crisv32-*-* ia64-*-* xtensa*-*-* sh*-*-* sparc*-*-* s390*-*-* } { > "*" } { "" } } */ > /* Currently ICEs for (x86 && ilp32 && pic). */ > /* { dg-xfail-if "PR33642/36240" { { i?86-*-* x86_64-*-* } && { ilp32 && { ! > nonpic } } } { "*" } { "" } } */ > > > I was about to add arm*-*-* but there are other reports about this flag > (not exhaustive, just based on summary): > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30121 > "ICE on frtl-abstract-sequences" > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31549 > "Documentation for -frtl-abstract-sequences is in the wrong place" > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36240 > "PIC and -frtl-abstract-sequences" > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38694 > "gcc.c-torture/compile/pr11832.c and pr33009.c -frtl-abstract-sequences > ICE on arm-linux-gnueabi on 4.3.2 and trunk" > > What should be done about this optimization for 4.4 and 4.5? Is there > anyone trying to fix its shortcomings? > > Thanks in advance,
I will happily approve a patch to make -frtl-abstract-sequences a no-op and remove its implementation. Thanks, Richard. > Laurent > > >
