Created attachment 31115 [details] Repeat as ./cc1 -fpreprocessed -O2 -o dtoa.s dtoa.i Revision r204205 caused the build for cris-elf to fail as follows: /tmp/break20131030_05/gccobj/./gcc/xgcc -B/tmp/break20131030_05/gccobj/./gcc/ -nostdinc -B/tmp/break20131030_05/gccobj/cris-elf/v8/newlib/ -isystem /tmp/break20131030_05/gccobj/cris-elf/v8/newlib/targ-include -isystem /tmp/break20131030_05/gcc/newlib/libc/include -B/tmp/break20131030_05/gccobj/cris-elf/v8/libgloss/cris -L/tmp/break20131030_05/gccobj/cris-elf/v8/libgloss/libnosys -L/tmp/break20131030_05/gcc/libgloss/cris -B/tmp/break20131030_05/pre/cris-elf/bin/ -B/tmp/break20131030_05/pre/cris-elf/lib/ -isystem /tmp/break20131030_05/pre/cris-elf/include -isystem /tmp/break20131030_05/pre/cris-elf/sys-include -march=v8 -mbest-lib-options -DPACKAGE_NAME=\"newlib\" -DPACKAGE_TARNAME=\"newlib\" -DPACKAGE_VERSION=\"1.20.0\" -DPACKAGE_STRING=\"newlib\ 1.20.0\" -DPACKAGE_BUGREPORT=\"\" -DPACKAGE_URL=\"\" -I. -I/tmp/break20131030_05/gcc/newlib/libc/stdlib -DHAVE_RENAME -D_USE_WRITE -DCOMPACT_CTYPE -fno-builtin -g -O2 -march=v8 -mbest-lib-options -c -o lib_a-dtoa.o `test -f 'dtoa.c' || echo '/tmp/break20131030_05/gcc/newlib/libc/stdlib/'`dtoa.c /tmp/break20131030_05/gcc/newlib/libc/stdlib/dtoa.c: In function 'quorem': /tmp/break20131030_05/gcc/newlib/libc/stdlib/dtoa.c:142:1: error: insn does not satisfy its constraints: } ^ (insn 289 35 288 3 (set (reg/v/f:SI 171 [orig:132 S ] [132]) (reg/v/f:SI 11 r11 [orig:132 S ] [132])) 37 {*movsi_internal} (nil)) /tmp/break20131030_05/gcc/newlib/libc/stdlib/dtoa.c:142:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:411 0x8a5018 _fatal_insn(char const*, rtx_def const*, char const*, int, char const*) /tmp/break20131030_05/gcc/gcc/rtl-error.c:109 0x8a503f _fatal_insn_not_found(rtx_def const*, char const*, int, char const*) /tmp/break20131030_05/gcc/gcc/rtl-error.c:120 0x858866 reload_cse_simplify_operands /tmp/break20131030_05/gcc/gcc/postreload.c:411 0x858b44 reload_cse_simplify /tmp/break20131030_05/gcc/gcc/postreload.c:123 0x858b44 reload_cse_regs_1 /tmp/break20131030_05/gcc/gcc/postreload.c:220 0x858bd3 reload_cse_regs /tmp/break20131030_05/gcc/gcc/postreload.c:68 0x858bd3 rest_of_handle_postreload /tmp/break20131030_05/gcc/gcc/postreload.c:2332 0x858bd3 execute /tmp/break20131030_05/gcc/gcc/postreload.c:2368 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. make[8]: *** [lib_a-dtoa.o] Error 1 make[8]: Leaving directory `/tmp/break20131030_05/gccobj/cris-elf/v8/newlib/libc/stdlib' Author of commit CC:ed. Preprocessed code attached.
I'm seeing a similar error in a Solaris 11/SPARC bootstrap: libtool: compile: /var/gcc/regression/trunk/11-gcc/build/./gcc/xgcc -B/var/gcc/regression/trunk/11-gcc/build/./gcc/ -B/vol/gcc/sparc-sun-solaris2.11/bin/ -B/vol/gcc/sparc-sun-solaris2.11/lib/ -isystem /vol/gcc/sparc-sun-solaris2.11/include -isystem /vol/gcc/sparc-sun-solaris2.11/sys-include -DHAVE_CONFIG_H -I. -I/vol/gcc/src/hg/trunk/local/libgomp -I/vol/gcc/src/hg/trunk/local/libgomp/config/posix -I/vol/gcc/src/hg/trunk/local/libgomp -Wall -pthread -Werror -g -O2 -pthread -MT env.lo -MD -MP -MF .deps/env.Tpo -c /vol/gcc/src/hg/trunk/local/libgomp/env.c -fPIC -DPIC -o .libs/env.o /vol/gcc/src/hg/trunk/local/libgomp/env.c: In function 'parse_one_place': /vol/gcc/src/hg/trunk/local/libgomp/env.c:548:1: error: insn does not satisfy its constraints: } ^ (insn 1271 176 1270 17 (set (reg/v/f:SI 757 [orig:406 stridep ] [406]) (reg/v/f:SI 27 %i3 [orig:406 stridep ] [406])) 63 {*movsi_insn} (nil)) /vol/gcc/src/hg/trunk/local/libgomp/env.c:548:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:411 0xa0642f _fatal_insn(char const*, rtx_def const*, char const*, int, char const*) /vol/gcc/src/hg/trunk/local/gcc/rtl-error.c:109 0xa0649b _fatal_insn_not_found(rtx_def const*, char const*, int, char const*) /vol/gcc/src/hg/trunk/local/gcc/rtl-error.c:120 0x95d1f7 reload_cse_simplify_operands /vol/gcc/src/hg/trunk/local/gcc/postreload.c:411 0x95c32b reload_cse_simplify /vol/gcc/src/hg/trunk/local/gcc/postreload.c:123 0x95c857 reload_cse_regs_1 /vol/gcc/src/hg/trunk/local/gcc/postreload.c:220 0x95c047 reload_cse_regs /vol/gcc/src/hg/trunk/local/gcc/postreload.c:68 0x9646eb rest_of_handle_postreload /vol/gcc/src/hg/trunk/local/gcc/postreload.c:2332 0x9647d7 execute /vol/gcc/src/hg/trunk/local/gcc/postreload.c:2368 Rainer
Let me join the club with an SH failure when building libsupc++: /gcc-trunk2/libstdc++-v3/libsupc++/vmi_class_type_info.cc:68:1: error: insn does not satisfy its constraints: } ^ (insn 174 173 175 21 (set (reg:SI 7 r7) (reg/v/f:SI 260 [orig:223 src_type ] [223])) ../../../../../gcc-trunk2/libstdc++-v3/libsupc++/vmi_class_type_info.cc:58 256 {movsi_i} (nil)) ../../../../../gcc-trunk2/libstdc++-v3/libsupc++/vmi_class_type_info.cc:68:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:411 0x879ab97 _fatal_insn(char const*, rtx_def const*, char const*, int, char const*) ../../gcc-trunk2/gcc/rtl-error.c:109 0x879abd5 _fatal_insn_not_found(rtx_def const*, char const*, int, char const*) ../../gcc-trunk2/gcc/rtl-error.c:120 0x8747aba reload_cse_simplify_operands ../../gcc-trunk2/gcc/postreload.c:411 0x874830a reload_cse_simplify ../../gcc-trunk2/gcc/postreload.c:123 0x874830a reload_cse_regs_1 ../../gcc-trunk2/gcc/postreload.c:220 0x8749ac2 reload_cse_regs ../../gcc-trunk2/gcc/postreload.c:68 0x8749ac2 rest_of_handle_postreload ../../gcc-trunk2/gcc/postreload.c:2332 0x8749ac2 execute ../../gcc-trunk2/gcc/postreload.c:2368 Please submit a full bug report,
And arm-none-eabi builds in newlib. Let's join the party. ./cc1 -O2 -mthumb -O2 -g -fno-builtin -fshort-enums -S -mcpu=cortex-a9 -mfpu=neon -mfloat-abi=softfp reduced.c #>$GCCDIR/configure --target=arm-none-eabi --with-languages=c Breaks glibc builds for arm-none-linux-gnueabihf
Created attachment 31119 [details] SH testcase yep, Also fails to build the sh-none-elf newlib. Attached case to reproduce cc1 -O2 fgets.i -quiet -fno-builtin fgets.i: In function '_fgets_r': fgets.i:826:1: error: insn does not satisfy its constraints: } ^ (insn 162 163 161 4 (set (reg/v/f:SI 215 [orig:192 buf ] [192]) (reg/v/f:SI 5 r5 [orig:192 buf ] [192])) 256 {movsi_i} (nil)) fgets.i:826:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:411
I'll revert the patch momentarily. Sorry for the trouble and thanks for any testcases.
Created attachment 31121 [details] arm-none-eabi reduced testcase ./cc1 -O2 -g -mthumb -mcpu=cortex-a9 -mfpu=neon -mfloat-abi=softfp reduced.c
Author: jamborm Date: Thu Oct 31 13:39:26 2013 New Revision: 204254 URL: http://gcc.gnu.org/viewcvs?rev=204254&root=gcc&view=rev Log: 2013-10-31 Martin Jambor <mjambor@suse.cz> PR rtl-optimization/58934 Revert: 2013-10-30 Martin Jambor <mjambor@suse.cz> PR rtl-optimization/10474 * ira.c (find_moveable_pseudos): Do not calculate dominance info nor df analysis. (interesting_dest_for_shprep): New function. (split_live_ranges_for_shrink_wrap): Likewise. (ira): Calculate dominance info and df analysis. Call split_live_ranges_for_shrink_wrap. testsuite/ * gcc.dg/pr10474.c: New testcase. * gcc.dg/ira-shrinkwrap-prep-1.c: Likewise. * gcc.dg/ira-shrinkwrap-prep-2.c: Likewise. Removed: trunk/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-1.c trunk/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c trunk/gcc/testsuite/gcc.dg/pr10474.c Modified: trunk/gcc/ChangeLog trunk/gcc/ira.c trunk/gcc/testsuite/ChangeLog
It also had broken PowerPC.
Fixed as far as I'm concerned. Martin, you're assigned so I'll let you close it if you please.
Fixed.
I have re-submitted my patch in which this bug is fixed, you can find it at http://gcc.gnu.org/ml/gcc-patches/2013-11/msg00598.html I have verified the patch bootstraps on i686-linux (reported by Jakub in the mailing list), ppc64-linux (some problem reported by David in comment #8) and ia64-linux (no problem reported but anyway), however I could not try the same on Sparc or Solaris (even though failure was reported by Rainer in comment #1) because I do not have access to either of them (oh how I wish that sparcs on the compile farm came back). Similarly, I have verified that the original reported failure on cris-elf cross-compiler goes away, as does the problem on sh-none-elf cross (the one from comment #4). However, even without the fix I was not able to reproduce the failure on arm reported in comment #6. If anyone is willing to test the patch on any platform but especially on those which I could not, I'd be very grateful. Thanks.
> --- Comment #11 from Martin Jambor <jamborm at gcc dot gnu.org> --- [...] > If anyone is willing to test the patch on any platform but especially > on those which I could not, I'd be very grateful. Thanks. I've now regression-tested the patch on sparc-sun-solaris2.11. Bootstrap works and I find just a single regression: FAIL: gfortran.fortran-torture/execute/pr32604.f90 execution, -O2 -fbounds-check which is reproducible. Rainer
(In reply to ro@CeBiTec.Uni-Bielefeld.DE from comment #12) > > --- Comment #11 from Martin Jambor <jamborm at gcc dot gnu.org> --- > [...] > > If anyone is willing to test the patch on any platform but especially > > on those which I could not, I'd be very grateful. Thanks. > > I've now regression-tested the patch on sparc-sun-solaris2.11. > Bootstrap works and I find just a single regression: > > FAIL: gfortran.fortran-torture/execute/pr32604.f90 execution, -O2 > -fbounds-check > > which is reproducible. > > Rainer Thanks a lot. I committed the patch a few moments before reading your comment so I suppose trunk now suffers from this failure. If so, can you please open a bug report? I tried to reproduce it on gcc54 on the compile farm (sparc64-linux) but so far I have failed to even build the gcc itself there :-/
(In reply to ro@CeBiTec.Uni-Bielefeld.DE from comment #12) > FAIL: gfortran.fortran-torture/execute/pr32604.f90 execution, -O2 > -fbounds-check > I did not find this failure in your latest test results (http://gcc.gnu.org/ml/gcc-testresults/2013-11/msg01857.html) and so I assume it has been also fixed by any of the subsequent patches addressing other issues.
> --- Comment #14 from Martin Jambor <jamborm at gcc dot gnu.org> --- > (In reply to ro@CeBiTec.Uni-Bielefeld.DE from comment #12) >> FAIL: gfortran.fortran-torture/execute/pr32604.f90 execution, -O2 >> -fbounds-check >> > > I did not find this failure in your latest test results > (http://gcc.gnu.org/ml/gcc-testresults/2013-11/msg01857.html) and so I > assume it has been also fixed by any of the subsequent patches > addressing other issues. Right, it's gone in this weekend's bootstraps. Rainer