[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #19 from Iain Sandoe iains at gcc dot gnu.org 2012-01-31 08:06:38 UTC --- x86-64-darwin10 @ r183725 bootstrapped with comment #17 #18 applied (and fails repeatably without). testing under way - do you want any more/different analysis?
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #20 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-31 08:41:53 UTC --- Author: jakub Date: Tue Jan 31 08:41:46 2012 New Revision: 183751 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=183751 Log: PR bootstrap/52041 PR bootstrap/52039 PR target/51974 * ree.c (add_removable_extension): Change def_map argument to unsigned *, store in def_map 1 + offset into *insn_list vector instead of pointers into the vector. (find_removable_extensions): Adjust caller. Modified: trunk/gcc/ChangeLog trunk/gcc/ree.c
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #21 from Richard Guenther rguenth at gcc dot gnu.org 2012-01-31 09:53:42 UTC --- (In reply to comment #18) Created attachment 26522 [details] X450 And ipa-prop doesn't seem to initialize ao.volatile_p. I think it is a bad idea not to call ao_ref_init, otherwise next time some further field is added, we'll hit this bug again. That patch is ok. We should indeed always call ao_ref_init (maybe I should move it to some place inlinable).
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 Jakub Jelinek jakub at gcc dot gnu.org changed: What|Removed |Added Status|NEW |RESOLVED Resolution||FIXED AssignedTo|unassigned at gcc dot |jakub at gcc dot gnu.org |gnu.org | --- Comment #23 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-31 10:07:25 UTC --- Fixed.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #22 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-31 10:06:55 UTC --- Author: jakub Date: Tue Jan 31 10:06:50 2012 New Revision: 183753 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=183753 Log: PR bootstrap/52041 PR bootstrap/52039 PR target/51974 * ipa-prop.c (detect_type_change_1): Call ao_ref_init. Modified: trunk/gcc/ChangeLog trunk/gcc/ipa-prop.c
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 Richard Guenther rguenth at gcc dot gnu.org changed: What|Removed |Added Target Milestone|--- |4.7.0 --- Comment #4 from Richard Guenther rguenth at gcc dot gnu.org 2012-01-30 09:27:16 UTC --- Dup of PR51974?
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #5 from Iain Sandoe iains at gcc dot gnu.org 2012-01-30 13:47:06 UTC --- I just bootstrapped r183695 (incl. Ada and Obj-C++) with --enable-checking=release on i686-darwin9, using gcc4.6 as the bootstrap compiler. What bootstrap compiler where you using?
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #6 from Dominique d'Humieres dominiq at lps dot ens.fr 2012-01-30 14:16:37 UTC --- I just bootstrapped r183695 (incl. Ada and Obj-C++) with --enable-checking=release on i686-darwin9, using gcc4.6 as the bootstrap compiler. What bootstrap compiler where you using? [macbook] gcc/p_build% gcc -v Using built-in specs. COLLECT_GCC=gcc COLLECT_LTO_WRAPPER=/opt/gcc/gcc4.7a/libexec/gcc/x86_64-apple-darwin10.8.0/4.7.0/lto-wrapper Target: x86_64-apple-darwin10.8.0 Configured with: ../_clean/configure --prefix=/opt/gcc/gcc4.7a --enable-languages=c,c++,fortran,ada,lto --with-gmp=/opt/mp --with-system-zlib --enable-checking=release --with-cloog=/opt/mp --enable-cloog-backend=isl --enable-lto Thread model: posix gcc version 4.7.0 20120117 (experimental) [trunk revision 183234] (GCC) Note that a clean bootstrap succeeded for revisions (above 183000) 183001, 183306, 183526 (patched), 183618, 183649, **183707**, and **183719**, but not for revisions 183650 and 183669. So it seems that bootstrap is back again due to some change between revision 183669 and 183695 (mostly gfortran, testsuite, ada, ... , but for r183684). I can reproduce random comparison failures as well, even with checking enabled, on openSUSE Factory x86_64. I wasn't yet able to reproduce them locally though. I'm trying again. Richard, what are random comparison failures (from PR51974)? Do the above qualify?
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 Iain Sandoe iains at gcc dot gnu.org changed: What|Removed |Added Status|UNCONFIRMED |NEW Last reconfirmed||2012-01-30 Ever Confirmed|0 |1 --- Comment #7 from Iain Sandoe iains at gcc dot gnu.org 2012-01-30 15:35:39 UTC --- @r183725 x86-64-darwin10, I see a compare fail on tree-ssa-strlen.o There seem to be real code differences doing a dissassembler dump of the stripped binaries; will try and narrow down the significant.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #8 from Dominique d'Humieres dominiq at lps dot ens.fr 2012-01-30 15:44:27 UTC --- @r183725 x86-64-darwin10, I see a compare fail on tree-ssa-strlen.o It just finished to bootstrap for me with '--enable-checking=release': [macbook] f90/bug% gfcp -v Using built-in specs. COLLECT_GCC=gfcp COLLECT_LTO_WRAPPER=/opt/gcc/gcc4.7p/libexec/gcc/x86_64-apple-darwin10.8.0/4.7.0/lto-wrapper Target: x86_64-apple-darwin10.8.0 Configured with: ../p_work/configure --prefix=/opt/gcc/gcc4.7p --enable-languages=c,c++,lto,fortran --with-gmp=/opt/mp --with-system-zlib --enable-checking=release --with-cloog=/opt/mp --enable-cloog-backend=isl --enable-lto Thread model: posix gcc version 4.7.0 20120130 (experimental) [trunk revision 183725] (GCC)
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 Jakub Jelinek jakub at gcc dot gnu.org changed: What|Removed |Added CC||jakub at gcc dot gnu.org --- Comment #9 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-30 16:16:48 UTC --- Is it -g vs. -g0 related? I.e. if you recompile tree-ssa-strlen.i with stage1 and stage2 cc1 with the same options, does it yield the same assembly? If not, stage2 might be miscompiled or similar issue, if yes, does compiling tree-ssa-strlen.i with -fcompare-debug (in addition to the standard options) yield a failure? If it is not -g vs. -g0, you could try -fdump-tree-all with both compilers and compare where the first difference appears.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #10 from Iain Sandoe iains at gcc dot gnu.org 2012-01-30 18:22:12 UTC --- the stage3 tree-ssa-strlen.o fails with -fcompare-debug failure (length). An incomplete analysis ... run out of time... The first mismatched tree dump is 18t.ssa -- with differences like # DEBUG x = y but also there are some differences in temporary var numbers. Looking at the asm, there seem to be differences like + movl al,%eax cmp $xxx, %eax and cltq in stage2 missing in stage 3 -- so size changing instructions.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #11 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-30 19:26:10 UTC --- The addition of # DEBUG lines in the dumps as well as some changes in D. numbers are expected and correct in between the dumps, but the SSA_NAME versions (_VV parts) as well as the stmt kinds, number of operands, number and order of basic blocks etc. should be the same.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #12 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-30 19:36:33 UTC --- Can't reproduce the -fcompare-debug failure with a cross from x86_64-linux to x86_64-darwin on the provided tree-ssa-strlen.i btw., though of course the configuration of the compiler could be different.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #13 from Iain Sandoe iains at gcc dot gnu.org 2012-01-30 20:31:11 UTC --- OK. I've been using the command lines from my build log, with modification to the path to account for the stage2/stage3 ... so like in stage2-gcc/ ../stage1-gcc/g++ in stage3-gcc/ ../stage3-gcc/g++ The problem is that there are so many differences between the files, that even with diff or side-by-side file displays it's hard to see which differences are significant. When I look at the 149t.optimized tree dump, and assuming that changes in D. and # DEBUG are not significant, I don't really see anything that looks problematical - so either the problem is rtl side or I need to narrow down what I'm looking for, back to the asm to try and narrow things down: there seem to be two mis-matched cases: 1. where a var is being used in a sequence of compare jump (I'd assume a switch), I the case of stage2 the asm has movzbl al, %eax cmp $xxx, %eax j in the case of stage3 , the movzbl is missing. 2. an indexed lookup from gss_for_code: in the case of stage2 there is: ..set up rdx cltq movslq (%rda, %rax,4), %rdx the (cltq) extension of reg A is missing for stage 3. both look like sign extension-ish issues. I think maybe will have to look at some rtl :(
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #14 from Iain Sandoe iains at gcc dot gnu.org 2012-01-30 20:34:08 UTC --- (In reply to comment #13) OK. I've been using the command lines from my build log, with modification to the path to account for the stage2/stage3 ... so like in stage3-gcc/ ../stage3-gcc/g++ ... typo ../stage2-gcc/g++ ... of course.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #15 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-30 21:07:57 UTC --- Can you please mail me (or upload somewhere) all the RTL logs from compilation with -fcompare-debug and -fdump-tree-optimized (both the non-gk and gk dumps)?
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #16 from Iain Sandoe iains at gcc dot gnu.org 2012-01-30 21:24:07 UTC --- (In reply to comment #15) Can you please mail me (or upload somewhere) all the RTL logs from compilation with -fcompare-debug and -fdump-tree-optimized (both the non-gk and gk dumps)? will do tomorrow - a the moment, I think scratch the comments above, ... the stage3 output is not repeatable (c.f. the bootstrap ouput) using the command lines as I described (I guess because of the differences in the paths). However, the differences seem to be more than just strings - so I need to try and understand what's happening there. [the stage2 output *is* repeatable with ../stage1-gcc/g++ ... ] === I will repeat the experiment tomorrow using the stage2 .ii file .. out of time for today.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #17 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-30 22:12:24 UTC --- Created attachment 26521 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26521 X449 The cross-compiler under valgrind actually shows some errors, I'm attaching fixes for both of the sources of the failures. This one has been introduced probably by the PR51667 changes, for object vectors storing somewhere else the pointers returned by VEC_safe_push is a bad idea, as any times a new element is pushed into the vector, the vector may have been reallocated, making all the pointers into it invalid.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #18 from Jakub Jelinek jakub at gcc dot gnu.org 2012-01-30 22:13:44 UTC --- Created attachment 26522 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26522 X450 And ipa-prop doesn't seem to initialize ao.volatile_p. I think it is a bad idea not to call ao_ref_init, otherwise next time some further field is added, we'll hit this bug again.
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #1 from Dominique d'Humieres dominiq at lps dot ens.fr 2012-01-29 14:36:29 UTC --- Created attachment 26501 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26501 Assembler for gcc/tree-ssa-strlen.c at stage 2 Assembler generated by /opt/gcc/p_build/stage1-gcc/g++ -B/opt/gcc/p_build/stage1-gcc/ -B/opt/gcc/gcc4.7p/x86_64-apple-darwin10.8.0/bin/ -nostdinc++ -B/opt/gcc/p_build/stage1-x86_64-apple-darwin10.8.0/libstdc++-v3/src/.libs -B/opt/gcc/p_build/stage1-x86_64-apple-darwin10.8.0/libstdc++-v3/libsupc++/.libs -I/opt/gcc/p_build/stage1-x86_64-apple-darwin10.8.0/libstdc++-v3/include/x86_64-apple-darwin10.8.0 -I/opt/gcc/p_build/stage1-x86_64-apple-darwin10.8.0/libstdc++-v3/include -I/opt/gcc/p_work/libstdc++-v3/libsupc++ -L/opt/gcc/p_build/stage1-x86_64-apple-darwin10.8.0/libstdc++-v3/src/.libs -L/opt/gcc/p_build/stage1-x86_64-apple-darwin10.8.0/libstdc++-v3/libsupc++/.libs -c -g -O2 -mdynamic-no-pic -gtoggle -DIN_GCC -fno-exceptions -fno-rtti -W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual -Wmissing-format-attribute -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -Werror -DHAVE_CONFIG_H -I. -I. -I../../p_work/gcc -I../../p_work/gcc/. -I../../p_work/include -I./../stage2-intl -I../../p_work/libcpp/include -I/opt/mp/include -I../../p_work/libdecnumber -I../../p_work/libdecnumber/dpd -I../stage2-libdecnumber -I/opt/mp/include -DCLOOG_INT_GMP -DCLOOG_ORG ../../p_work/gcc/tree-ssa-strlen.c -save-temps
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #2 from Dominique d'Humieres dominiq at lps dot ens.fr 2012-01-29 14:42:13 UTC --- Created attachment 26502 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26502 bzipped assembler for gcc/tree-ssa-strlen.c at stage 3 Assembler generated by /opt/gcc/p_build/stage2-gcc/g++ -B/opt/gcc/p_build/stage2-gcc/ -B/opt/gcc/gcc4.7p/x86_64-apple-darwin10.8.0/bin/ -nostdinc++ -B/opt/gcc/p_build/stage2-x86_64-apple-darwin10.8.0/libstdc++-v3/src/.libs -B/opt/gcc/p_build/stage2-x86_64-apple-darwin10.8.0/libstdc++-v3/libsupc++/.libs -I/opt/gcc/p_build/stage2-x86_64-apple-darwin10.8.0/libstdc++-v3/include/x86_64-apple-darwin10.8.0 -I/opt/gcc/p_build/stage2-x86_64-apple-darwin10.8.0/libstdc++-v3/include -I/opt/gcc/p_work/libstdc++-v3/libsupc++ -L/opt/gcc/p_build/stage2-x86_64-apple-darwin10.8.0/libstdc++-v3/src/.libs -L/opt/gcc/p_build/stage2-x86_64-apple-darwin10.8.0/libstdc++-v3/libsupc++/.libs -c -g -O2 -mdynamic-no-pic -DIN_GCC -fno-exceptions -fno-rtti -W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual -Wmissing-format-attribute -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -Werror -DHAVE_CONFIG_H -I. -I. -I../../p_work/gcc -I../../p_work/gcc/. -I../../p_work/include -I./../stage3-intl -I../../p_work/libcpp/include -I/opt/mp/include -I../../p_work/libdecnumber -I../../p_work/libdecnumber/dpd -I../stage3-libdecnumber -I/opt/mp/include -DCLOOG_INT_GMP -DCLOOG_ORG ../../p_work/gcc/tree-ssa-strlen.c -save-temps
[Bug bootstrap/52041] [4.7 Regression] Bootstrap failure at revision 183650 with --enable-checking=release
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52041 --- Comment #3 from Dominique d'Humieres dominiq at lps dot ens.fr 2012-01-29 14:44:55 UTC --- Created attachment 26503 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26503 bzipped preprocessed file for gcc/tree-ssa-strlen.c at stage 3 AFAICT the only differences between stage 2 and 3 are the obvious replacements of stage2 with stage1 and stage3 with stage2.