https://gcc.gnu.org/g:d2e62fc9640620227dbb71955cb428f0ee35bba1
commit r17-4463-gd2e62fc9640620227dbb71955cb428f0ee35bba1 Author: Jeff Law <[email protected]> Date: Sat Sep 19 08:06:13 2026 -0600 [RISC-V] Fix RISC-V test expectation after recent saturation/clamping work Some of the recent work recognizing saturating the clamping resulted in vec_sat_u_sub_trunc-1-u8 starting to fail: Tests that now fail, but worked before (10 tests): unix/-march=rv64gc_zba_zbb_zbs_zicond: gcc: gcc.target/riscv/rvv/autovec/sat/vec_sat_u_sub_trunc-1-u8.c -ftree-vectorize -O2 -mrvv-vector-bits=scalable -mrvv-max-lmul=dynamic -ffast-math scan-tree-dump-times optimized ".SAT_SUB " 1 [ ... ] I bisected and compared the before/after for the identified patch (already forgot which patch it was). The code with the patch was clearly better in gimple as it eliminated another branchy sequence and replaced it with saturating arithmetic, throwing off the dump check for this test. So the right thing to do is adjust the test expectation. Interestingly enough this allows us simplify the test expectations as the test previously expected different results for -O2 vs -O3. Pushing to the trunk. gcc/testsuite * gcc.target/riscv/rvv/autovec/sat/vec_sat_u_sub_trunc-1-u8.c: Update test expectations. Diff: --- .../gcc.target/riscv/rvv/autovec/sat/vec_sat_u_sub_trunc-1-u8.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/sat/vec_sat_u_sub_trunc-1-u8.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/sat/vec_sat_u_sub_trunc-1-u8.c index c5947d92fe96..b3638bd5d5a5 100644 --- a/gcc/testsuite/gcc.target/riscv/rvv/autovec/sat/vec_sat_u_sub_trunc-1-u8.c +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/sat/vec_sat_u_sub_trunc-1-u8.c @@ -5,7 +5,6 @@ DEF_VEC_SAT_U_SUB_TRUNC_FMT_1(uint8_t, uint16_t) -/* { dg-final { scan-tree-dump-times ".SAT_SUB " 2 "optimized" { target { no-opts "-O2" } } } } */ -/* { dg-final { scan-tree-dump-times ".SAT_SUB " 1 "optimized" { target { no-opts "-O3" } } } } */ +/* { dg-final { scan-tree-dump-times ".SAT_SUB " 2 "optimized" } } */ /* { dg-final { scan-assembler-times {vssubu\.vv} 1 } } */ /* { dg-final { scan-assembler-times {vnsrl\.wi} 1 } } */
