https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124043
Torbjorn SVENSSON <azoff at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |rearnsha at gcc dot gnu.org
--- Comment #1 from Torbjorn SVENSSON <azoff at gcc dot gnu.org> ---
Researching this a bit further gave me the information that lane0 is different
from lane1 and it's up to the compiler to pick either the pattern used for
lane0 or lane1.
Maybe the tests should be updated like this?
diff --git a/gcc/testsuite/gcc.target/arm/crypto-vsha1cq_u32.c
b/gcc/testsuite/gcc.target/arm/crypto-vsha1cq_u32.c
index 0cadd19c4dc..c35b162d23a 100644
--- a/gcc/testsuite/gcc.target/arm/crypto-vsha1cq_u32.c
+++ b/gcc/testsuite/gcc.target/arm/crypto-vsha1cq_u32.c
@@ -31,5 +31,6 @@ uint32_t foo (void)
TEST_SHA1C_VEC_SELECT (GET_LANE)
/* { dg-final { scan-assembler-times {sha1c.32\tq[0-9]+, q[0-9]+} 5 } } */
-/* { dg-final { scan-assembler-times {vdup.32\tq[0-9]+, r[0-9]+} 4 } } */
+/* { dg-final { scan-assembler-times {vdup.32\tq[0-9]+, r[0-9]+} 3 } } */
+/* { dg-final { scan-assembler-times {vld1.32\t\{d[0-9]+\[\], d[0-9]+\[\]\},
\[sp\]} 1 } } */
/* { dg-final { scan-assembler-times {vmov.32\tr[0-9]+, d[0-9]+\[[0-9]+\]+} 3
} } */
As I only test a very limited number of arm targets, maybe someone with wider
knowledge knows if this test, in current form, passes for any arm target.
