https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87138
Bug ID: 87138 Summary: [8/9 Regression] wrong code with TI multiplication with -O -march=k8 -fno-tree-fre -mavx512bw Product: gcc Version: 9.0 Status: UNCONFIRMED Keywords: wrong-code Severity: normal Priority: P3 Component: rtl-optimization Assignee: unassigned at gcc dot gnu.org Reporter: zsojka at seznam dot cz Target Milestone: --- Host: x86_64-pc-linux-gnu Target: x86_64-pc-linux-gnu Created attachment 44620 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44620&action=edit reduced testcase Output: $ x86_64-pc-linux-gnu-gcc -O -march=k8 -fno-tree-fre -mavx512bw testcase.c $ sde64 -- ./a.out Aborted The difference between broken a good assembler is: movabs rax, 9223372036854775805 # tmp167, - mov rdx, -2 #, + mov rdx, 1 #, mov QWORD PTR [rsp], rax # %sfpD.2961, tmp167 mov QWORD PTR [rsp+8], rdx # %sfpD.2961, For some reason, 3 * 0x7FFFFFFFFFFFFFFF is computed as 0xfffffffffffffffe7ffffffffffffffd instead of 0x17ffffffffffffffd. The wrong value first appears in the .cse1 dump. $ x86_64-pc-linux-gnu-gcc -v Using built-in specs. COLLECT_GCC=/repo/gcc-trunk/binary-latest-amd64/bin/x86_64-pc-linux-gnu-gcc COLLECT_LTO_WRAPPER=/repo/gcc-trunk/binary-trunk-263946-checking-yes-rtl-df-extra-amd64/bin/../libexec/gcc/x86_64-pc-linux-gnu/9.0.0/lto-wrapper Target: x86_64-pc-linux-gnu Configured with: /repo/gcc-trunk//configure --enable-languages=c,c++ --enable-valgrind-annotations --disable-nls --enable-checking=yes,rtl,df,extra --with-cloog --with-ppl --with-isl --build=x86_64-pc-linux-gnu --host=x86_64-pc-linux-gnu --target=x86_64-pc-linux-gnu --with-ld=/usr/bin/x86_64-pc-linux-gnu-ld --with-as=/usr/bin/x86_64-pc-linux-gnu-as --disable-libstdcxx-pch --prefix=/repo/gcc-trunk//binary-trunk-263946-checking-yes-rtl-df-extra-amd64 Thread model: posix gcc version 9.0.0 20180829 (experimental) (GCC)