The important part seems to be: checking whether ln -s works... yes checking for i686-pc-linux-gnu-gcc... /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/./gcc/xgcc -B/tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/./gcc/ -B/usr/i686-pc-linux-gnu/bin/ -B/usr/i686-pc-linux-gnu/lib/ -isystem /usr/i686-pc-linux-gnu/include -isystem /usr/i686-pc-linux-gnu/sys-include -fno-checking checking for suffix of object files... configure: error: in `/tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/i686-pc-linux-gnu/libgcc': configure: error: cannot compute suffix of object files: cannot compile See `config.log' for more details make[2]: *** [Makefile:19065: configure-stage1-target-libgcc] Error 1 make[2]: Leaving directory '/tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build' I'm not sure which config.log they mean, i got a bunch: chopin /tmp/portage/sys-devel/gcc-12.2.1_p20221008 # find -name config.log ./work/build/libcpp/config.log ./work/build/libcody/config.log ./work/build/intl/config.log ./work/build/gcc/config.log ./work/build/libdecnumber/config.log ./work/build/build-i686-pc-linux-gnu/libcpp/config.log ./work/build/build-i686-pc-linux-gnu/fixincludes/config.log ./work/build/build-i686-pc-linux-gnu/libiberty/config.log ./work/build/i686-pc-linux-gnu/libgcc/config.log ./work/build/lto-plugin/config.log ./work/build/config.log ./work/build/libbacktrace/config.log ./work/build/libiberty/config.log Reproducible: Always
Created attachment 824347 [details] build.log
Created attachment 824349 [details] /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/config.log
It's an old system. But gcc used to compile, until few years ago. Since then there was #777117, so i couldn't update gcc. This problem doesn't seem related ("CET " doesn't appear in error messages as last year). chopin /tmp/portage/sys-devel/gcc-12.2.1_p20221008 # cat /proc/cpuinfo processor : 0 vendor_id : CentaurHauls cpu family : 6 model : 9 model name : VIA Nehemiah stepping : 3 cpu MHz : 665.000 cache size : 64 KB fdiv_bug : no f00f_bug : no coma_bug : no fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu de pse tsc msr cx8 mtrr pge cmov mmx fxsr sse cpuid rng rng_en bugs : cpu_meltdown spectre_v1 spectre_v2 spec_store_bypass l1tf mds swapgs itlb_multihit mmio_unknown bogomips : 1333.33 clflush size : 32 cache_alignment : 32 address sizes : 32 bits physical, 32 bits virtual power management:
I think need the libgcc config.log but it's easier if you upload the tarball the ebuild creates for you, as it's intended for this purpose.
Created attachment 824351 [details] /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/gcc-build-logs.tar.bz2
Oh dear. configure:3792: checking for suffix of object files configure:3814: /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/./gcc/xgcc -B/tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/./gcc/ -B/usr/i686-pc-linux-gnu/bin/ -B/usr/i686-pc-linux-gnu/lib/ -isystem /usr/i686-pc-linux-gnu/include -isystem /usr/i686-pc-linux-gnu/sys-include -fno-checking -c -g -march=native -pipe -O2 conftest.c >&5 xgcc: internal compiler error: in cpu_indicator_init, at common/config/i386/cpuinfo.h:986 0x8087146 diagnostic_impl(rich_location*, diagnostic_metadata const*, int, char const*, char**, diagnostic_t) ???:0 0x8088190 internal_error(char const*, ...) ???:0 0x80883c0 fancy_abort(char const*, int, char const*) ???:0 0x8062127 cpu_indicator_init(__processor_model*, __processor_model2*, unsigned int*) ???:0 0x8062d1f host_detect_local_cpu(int, char const**) ???:0 0x8056228 eval_spec_function(char const*, char const*, char const*) ???:0 0x805644c handle_spec_function(char const*, bool*, char const*) ???:0 0x8055ab3 do_spec_1(char const*, int, char const*) ???:0 0x8056f10 process_brace_body(char const*, char const*, char const*, int, int) ???:0 0x8056d3b handle_braces(char const*) ???:0 0x8055a7d do_spec_1(char const*, int, char const*) ???:0 0x8055e9a do_spec_1(char const*, int, char const*) ???:0 0x8056f10 process_brace_body(char const*, char const*, char const*, int, int) ???:0 0x8056d3b handle_braces(char const*) ???:0 0x8055a7d do_spec_1(char const*, int, char const*) ???:0 0x80555e7 do_spec_1(char const*, int, char const*) ???:0 0x8055e9a do_spec_1(char const*, int, char const*) ???:0 0x8056f10 process_brace_body(char const*, char const*, char const*, int, int) ???:0 0x8056d3b handle_braces(char const*) ???:0 0x8055a7d do_spec_1(char const*, int, char const*) ???:0 Please submit a full bug report, with preprocessed source (by using -freport-bug). Please include the complete backtrace with any bug report. See <https://bugs.gentoo.org/> for instructions. configure:3818: $? = 4 configure: failed program was: | /* confdefs.h */ | #define PACKAGE_NAME "GNU C Runtime Library" | #define PACKAGE_TARNAME "libgcc" | #define PACKAGE_VERSION "1.0" | #define PACKAGE_STRING "GNU C Runtime Library 1.0" | #define PACKAGE_BUGREPORT "" | #define PACKAGE_URL "http://www.gnu.org/software/libgcc/" | /* end confdefs.h. */ | | int | main () | { | | ; | return 0; | } configure:3832: error: in `/tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/i686-pc-linux-gnu/libgcc': configure:3834: error: cannot compute suffix of object files: cannot compile See `config.log' for more details
Could you try follow https://wiki.gentoo.org/wiki/Gcc-ICE-reporting-guide?
I'm a bit worried that your existing GCC might be broken, actually. It may have been the patch which was used before.
(In reply to Sam James from comment #8) > I'm a bit worried that your existing GCC might be broken, actually. It may > have been the patch which was used before. But it's not my gcc that segfaults, it's the (stage/boostrap) one, no ? /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/./gcc/xgcc
(In reply to Sam James from comment #7) > Could you try follow https://wiki.gentoo.org/wiki/Gcc-ICE-reporting-guide? I dont have the sources anymore. I'll retry, but that will require time. Hopefully that was a spurious ram/heat stuff.
(In reply to Thomas Capricelli from comment #9) > (In reply to Sam James from comment #8) > > I'm a bit worried that your existing GCC might be broken, actually. It may > > have been the patch which was used before. > > But it's not my gcc that segfaults, it's the (stage/boostrap) one, no ? > > /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/./gcc/xgcc oh right, I'd assumed it died very early on as for some stuff, the wrappers get used even if they're actually system copies, but you're right here, it does build-and-then-use xgcc. Could you try with USE=vanilla?
It fails with the same error (i think) when trying to emerge sys-devel/gcc-11.3.1_p20221007 It's in /tmp/portage/sys-devel/gcc-11.3.1_p20221007/work/build/i686-pc-linux-gnu/libgcc There's a config.log from which i can find the bad command. And it's 100% reproducible if i type it: chopin /tmp/portage/sys-devel/gcc-11.3.1_p20221007/work/build/i686-pc-linux-gnu/libgcc # /tmp/portage/sys-devel/gcc-11.3.1_p20221007/work/build/./gcc/xgcc -B/tmp/portage/sys-devel/gcc-11.3.1_p20221007/work/build/./gcc/ -B/usr/i686-pc-linux-gnu/bin/ -B/usr/i686-pc-linux-gnu/lib/ -isystem /usr/i686-pc-linux-gnu/include -isystem /usr/i686-pc-linux-gnu/sys-include -fno-checking -c -g -march=native -pipe -O2 conftest.c xgcc: internal compiler error: in cpu_indicator_init, at common/config/i386/cpuinfo.h:934 0x8084693 diagnostic_impl(rich_location*, diagnostic_metadata const*, int, char const*, char**, diagnostic_t) ???:0 0x8085678 internal_error(char const*, ...) ???:0 0x80858a8 fancy_abort(char const*, int, char const*) ???:0 0x8061969 cpu_indicator_init(__processor_model*, __processor_model2*, unsigned int*) ???:0 0x806254b host_detect_local_cpu(int, char const**) ???:0 0x8055da1 eval_spec_function(char const*, char const*, char const*) ???:0 0x8055fc5 handle_spec_function(char const*, bool*, char const*) ???:0 0x805562c do_spec_1(char const*, int, char const*) ???:0 0x8056a89 process_brace_body(char const*, char const*, char const*, int, int) ???:0 0x80568b4 handle_braces(char const*) (...)
Fantastic (I know it's bad, but it's good too). Does it happen with USE=vanilla?
It gives exactly the same error with vanillla. chopin ~ # USE=vanilla emerge sys-devel/gcc -1 [ebuild NS ] sys-devel/gcc-12.2.1_p20221008 [9.3.0] USE="(cxx) hardened nls nptl (pie) sanitize (ssp) vanilla* -ada% (-cet) (-custom-cflags) -d -debug -doc (-fixed-point) -fortran -go -graphite -jit (-libssp) -lto (-multilib) -objc -objc++ -objc-gc -openmp (-pch) -pgo -systemtap -test -valgrind% -vtv* -zstd% (-altivec%)" (...) I can still reproduce it with cd /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/i686-pc-linux-gnu/libgcc/ /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/./gcc/xgcc -B/tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/build/./gcc/ -B/usr/i686-pc-linux-gnu/bin/ -B/usr/i686-pc-linux-gnu/lib/ -isystem /usr/i686-pc-linux-gnu/include -isystem /usr/i686-pc-linux-gnu/sys-include -fno-checking -o conftest -g -march=native -pipe -O2 conftest.c
The error message says: xgcc: internal compiler error: in cpu_indicator_init, at common/config/i386/cpuinfo.h:986 I guess it's /tmp/portage/sys-devel/gcc-12.2.1_p20221008/work/gcc-12-20221008/gcc/common/config/i386/cpuinfo.h Whose line 986 says: gcc_assert (cpu_model->__cpu_vendor < VENDOR_MAX); But if you check the code in the function "cpu_indicator_init", i can't see any way that __cpu_vendor is not within range. If i isolate this code and compile it, it works (the assert doesn't raise). If i printf cpu_model->__cpu_vendor, i got 4, which is expected (VENDOR_CENTAUR). So, globally.. i'm still lost :-(
For the record, we suspected the current gcc (9.3.0) to be kinda corrupted. So i have * installed a recent stage3/i686 in a directory on my local computer, chrooted and created a package for gcc with quickpkg * i scp it and installed gcc-11.3.0-1.xpak on the target * carefully did gcc config/source and even restarted ssh connection * tried to compile sys-devel/gcc-12.2.1_p20221008 with it It fails at the same point, with same error (at the end). the line corresponds to gcc_assert (cpu_model->__cpu_vendor < VENDOR_MAX); But I can't see any valid code bath above that could give an invalid __cpu_vendor -------------------------------------------------------------------------- xg++: internal compiler error: in cpu_indicator_init, at common/config/i386/cpuinfo.h:986 0x808790a diagnostic_impl(rich_location*, diagnostic_metadata const*, int, char const*, char**, diagnostic_t) ???:0 0x808894b internal_error(char const*, ...) ???:0 0x8088b7b fancy_abort(char const*, int, char const*) ???:0 0x80628c7 cpu_indicator_init(__processor_model*, __processor_model2*, unsigned int*) ???:0
For the record, we've narrowed it to: gcc-12 -Q -O2 -march=nehemiah --help=target -march=native # crashes gcc-12 -Q -O2 -march=nehemiah --help=target # doesn't crash Reproducible with gcc-11.3.0, but NOT with gcc-9.3.0 -> handles native without problem. The error displayed being: chopin /tmp # gcc-12 -Q -O2 -march=nehemiah --help=target -march=native gcc-12: internal compiler error: in cpu_indicator_init, at common/config/i386/cpuinfo.h:986 0xb7c0235e __libc_start_main ???:0 Please submit a full bug report, with preprocessed source (by using -freport-bug). Please include the complete backtrace with any bug report. See <https://bugs.gentoo.org/> for instructions.
I reported the bug upstream: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107364
They are reactive, and they found very quickly. It's fixed and commited. I haven't validated yet (require a new gcc compile, very complicated here). But anyway, that's definitely upstream now. Thanks a lot to Sam who helped me a lot, here and on IRC.
