On 11/22/25 15:02, Haochen Jiang wrote:
On Linux/x86_64,
bd6d591799385b326b3589071b8cfa0bb61a9d19 is the first bad commit
commit bd6d591799385b326b3589071b8cfa0bb61a9d19
Author: Sandra Loosemore <[email protected]>
Date: Thu Nov 20 21:45:11 2025 +0000
OpenMP: C/C++ common testcases for "omp begin declare variant"
caused
FAIL: c-c++-common/gomp/delim-declare-variant-6.c scan-tree-dump gimple
"f3\\.ompvariant.*arch \\(.x86_64.\\)"
FAIL: c-c++-common/gomp/delim-declare-variant-6.c scan-tree-dump gimple
"f3\\.ompvariant.*kind \\(.host.\\)"
FAIL: c-c++-common/gomp/delim-declare-variant-6.c -std=c++17 scan-tree-dump gimple
"f3\\.ompvariant.*arch \\(.x86_64.\\)"
FAIL: c-c++-common/gomp/delim-declare-variant-6.c -std=c++17 scan-tree-dump gimple
"f3\\.ompvariant.*kind \\(.host.\\)"
FAIL: c-c++-common/gomp/delim-declare-variant-6.c -std=c++26 scan-tree-dump gimple
"f3\\.ompvariant.*arch \\(.x86_64.\\)"
FAIL: c-c++-common/gomp/delim-declare-variant-6.c -std=c++26 scan-tree-dump gimple
"f3\\.ompvariant.*kind \\(.host.\\)"
FAIL: c-c++-common/gomp/delim-declare-variant-6.c -std=c++98 scan-tree-dump gimple
"f3\\.ompvariant.*arch \\(.x86_64.\\)"
FAIL: c-c++-common/gomp/delim-declare-variant-6.c -std=c++98 scan-tree-dump gimple
"f3\\.ompvariant.*kind \\(.host.\\)"
with GCC configured with
../../gcc/configure
--prefix=/export/users3/haochenj/src/gcc-bisect/master/master/r16-5512/usr
--enable-clocale=gnu --with-system-zlib --with-demangler-in-ld
--with-fpmath=sse --enable-languages=c,c++,fortran --enable-cet --without-isl
--enable-libmpx x86_64-linux --disable-bootstrap
To reproduce:
$ cd {build_dir}/gcc && make check
RUNTESTFLAGS="gomp.exp=c-c++-common/gomp/delim-declare-variant-6.c
--target_board='unix{-m32}'"
$ cd {build_dir}/gcc && make check
RUNTESTFLAGS="gomp.exp=c-c++-common/gomp/delim-declare-variant-6.c
--target_board='unix{-m32\ -march=cascadelake}'"
Sigh, I keep forgetting that -m32 causes "match
(device={arch("x86_64")})" to fail, even if gcc was configured for
x86_64 instead of i?86. Reading the code for
ix86_omp_device_kind_arch_isa() I see that "x86" matches regardless of
whether it's 32 or 64 bit, so I'll push a patch shortly to use that in
the testcase.
-Sandra
