An error occurred during compilation

[matrix97317]

gfortran -O2 -Wall -frecursive -fno-optimize-sibling-calls -m64 -msse3 -mssse3 -msse4.1 -mavx -mavx2 -mavx2 -o zblat3 zblat3.o ../libopenblas_haswellp-r0.3.14.dev.a -
lm -lpthread -lgfortran -lm -lpthread -lgfortran -L/usr/lib32/../lib -L/usr/libx32/../lib -L/usr/lib/gcc/x86_64-linux-gnu/6 -L/usr/lib/gcc/x86_64-linux-gnu/6/../../../x
86_64-linux-gnu -L/usr/lib/gcc/x86_64-linux-gnu/6/../../../../lib -L/lib/x86_64-linux-gnu -L/lib/../lib -L/usr/lib/x86_64-linux-gnu -L/usr/lib/../lib -L/usr/lib/x86_64-
linux-gnu -L/usr/lib32 -L/usr/libx32 -L. -L/usr/lib/gcc/x86_64-linux-gnu/6/../../.. -lc
OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 ./sblat1

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:

Could not print backtrace: unrecognized DWARF version in .debug_info at 110
#0 0x7f9030668d1d in ???
#1 0x7f9030667f7d in ???
#2 0x7f902fdc005f in ???
#3 0x7f9030721488 in ???
#4 0x7f903072f807 in ???
#5 0x4034a8 in ???
#6 0x40183e in ???
#7 0x7f902fdad2e0 in ???
#8 0x4018b9 in ???
#9 0xffffffffffffffff in ???
Segmentation fault
Makefile:28: recipe for target 'level1' failed

---

OpenBLAS Version：0.3.13
gfortran -v
Using built-in specs.
COLLECT_GCC=gfortran
COLLECT_LTO_WRAPPER=/home/xxxxx/opt/gfortran/gcc-trunk/bin/../libexec/gcc/x86_64-pc-linux-gnu/12.0.0/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: ../gcc-trunk-source/gcc/configure --enable-languages=c,c++,fortran --enable-checking=yes,extra --enable-valgrind-annotations --disable-libstdcxx-pch --enable-libgomp --enable-lto --enable-gold --with-plugin-ld=gold --prefix=/usr/local/gcc-trunk
Thread model: posix
Supported LTO compression algorithms: zlib
gcc version 12.0.0 20210429 (experimental) [master revision e4aefface2:b286a1e29d:e4ff4ffb43d3d8520f1c106e04421f2e6a021c39] (GCC)

[martin-frbg]

This does not look familiar, but I have not tried to compile with a snapshot of gcc 12 (which I believe is in early stage 1) yet. Do you get an error when you compile with a stable release of gcc as well ?

[brada4]

0x7.... are libraries, not necessarily openblas.
0x0... adresses are syscalls

It claims DWARf version 6 binary 110, should be v5 for current releases of gdb to parse
If you could use more archaic dwarf debug info version, names of calls should tell better. So far we see 3 syscalls and 6 library calls far from the beginning it crashed accessing something OOB or null.

[martin-frbg]

gcc12 probably requires similarly recent versions of binutils and gdb... and just in case, did you use both gcc12 and gfortran12 in your build ? (mixing different versions of the two is known to fail)

[brada4]

It is practically non-debuggable backtrace.
Maybe there is some success running command in GDB

OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 gdb ./sblat1
gdb> r
gdb> bt
gdb>

As a bare minimum module+offset from memory needs to be decoded to function name from disk file, that looks close to impossible with incompatible syms lurking around....

Not certain there is a compiler bug, it may introduce more stringent "checks" around calls and reveal decade old bug in OpenBLAS, that happens all the time.

[brada4]

Cannot repeat fedora rawhide gcc 11.1 , suse tumbleweed gcc 10.3.

[martin-frbg]

Not reproduced with today's git checkouts of both gcc master and OpenBLAS develop branch on Kaby Lake (Haswell target).