Message ID | 20240918032708.14908-1-garthlei@linux.alibaba.com |
---|---|
State | New |
Headers | show |
Series | RISC-V: Allow zero operand for DI variants of vssubu.vx | expand |
LGTM, thanks :) Bohan Lei <garthlei@linux.alibaba.com> 於 2024年9月18日 週三 05:28 寫道: > The RISC-V vector machine description relies on the helper function > `sew64_scalar_helper` to emit actual insns for the DI variants of > vssub.vx and vssubu.vx. This works with vssub.vx, but can cause > problems with vssubu.vx with the scalar operand being constant zero, > because `has_vi_variant_p` returns false, and the operand will be taken > without being loaded into a reg. The attached testcases can cause an > internal compiler error as a result. > > Allowing a constant zero operand in those insns seems to be a simple > solution that only affects minimum existing code. > > gcc/ChangeLog: > > * config/riscv/vector.md: Allow zero operand for DI variants of > vssubu.vx > > gcc/testsuite/ChangeLog: > > * gcc.target/riscv/rvv/base/vssubu-1.c: New test. > * gcc.target/riscv/rvv/base/vssubu-2.c: New test. > --- > gcc/config/riscv/vector.md | 8 ++++---- > gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-1.c | 11 +++++++++++ > gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-2.c | 11 +++++++++++ > 3 files changed, 26 insertions(+), 4 deletions(-) > create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-1.c > create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-2.c > > diff --git a/gcc/config/riscv/vector.md b/gcc/config/riscv/vector.md > index d0677325ba1..66a2a477faf 100644 > --- a/gcc/config/riscv/vector.md > +++ b/gcc/config/riscv/vector.md > @@ -4400,10 +4400,10 @@ (define_insn "*pred_<optab><mode>_scalar" > (sat_int_minus_binop:VI_D > (match_operand:VI_D 3 "register_operand" " vr, vr, vr, vr") > (vec_duplicate:VI_D > - (match_operand:<VEL> 4 "register_operand" " r, r, r, > r"))) > + (match_operand:<VEL> 4 "reg_or_0_operand" " rJ, rJ, rJ, > rJ"))) > (match_operand:VI_D 2 "vector_merge_operand" " vu, 0, vu, > 0")))] > "TARGET_VECTOR" > - "v<insn>.vx\t%0,%3,%4%p1" > + "v<insn>.vx\t%0,%3,%z4%p1" > [(set_attr "type" "<int_binop_insn_type>") > (set_attr "mode" "<MODE>")]) > > @@ -4422,10 +4422,10 @@ (define_insn "*pred_<optab><mode>_extended_scalar" > (match_operand:VI_D 3 "register_operand" " vr, vr, vr, > vr") > (vec_duplicate:VI_D > (sign_extend:<VEL> > - (match_operand:<VSUBEL> 4 "register_operand" " r, r, > r, r")))) > + (match_operand:<VSUBEL> 4 "reg_or_0_operand" " rJ, rJ, > rJ, rJ")))) > (match_operand:VI_D 2 "vector_merge_operand" " vu, 0, > vu, 0")))] > "TARGET_VECTOR && !TARGET_64BIT" > - "v<insn>.vx\t%0,%3,%4%p1" > + "v<insn>.vx\t%0,%3,%z4%p1" > [(set_attr "type" "<int_binop_insn_type>") > (set_attr "mode" "<MODE>")]) > > diff --git a/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-1.c > b/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-1.c > new file mode 100644 > index 00000000000..f19b42aed04 > --- /dev/null > +++ b/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-1.c > @@ -0,0 +1,11 @@ > +/* { dg-do compile } */ > +/* { dg-options "-O2 -march=rv64gcv -mabi=lp64d" } */ > + > +#include <riscv_vector.h> > + > +vuint64m1_t test_vssubu_vx_u64m1(vuint64m1_t op1) > +{ > + return __riscv_vssubu_vx_u64m1(op1,0,0); > +} > + > +/* { dg-final { scan-assembler-not {\tvssubu} } } */ > \ No newline at end of file > diff --git a/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-2.c > b/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-2.c > new file mode 100644 > index 00000000000..cb4e4f48a9b > --- /dev/null > +++ b/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-2.c > @@ -0,0 +1,11 @@ > +/* { dg-do compile } */ > +/* { dg-options "-O2 -march=rv32gcv -mabi=ilp32d" } */ > + > +#include <riscv_vector.h> > + > +vuint64m1_t test_vssubu_vx_u64m1(vuint64m1_t op1) > +{ > + return __riscv_vssubu_vx_u64m1(op1,0,0); > +} > + > +/* { dg-final { scan-assembler-not {\tvssubu} } } */ > \ No newline at end of file > -- > 2.46.0 > >
On 9/18/24 1:41 AM, Kito Cheng wrote: > LGTM, thanks :) Agreed and pushed. > > > Bohan Lei <garthlei@linux.alibaba.com > <mailto:garthlei@linux.alibaba.com>> 於 2024年9月18日 週三 05:28 寫道: > > The RISC-V vector machine description relies on the helper function > `sew64_scalar_helper` to emit actual insns for the DI variants of > vssub.vx and vssubu.vx. This works with vssub.vx, but can cause > problems with vssubu.vx with the scalar operand being constant zero, > because `has_vi_variant_p` returns false, and the operand will be taken > without being loaded into a reg. The attached testcases can cause an > internal compiler error as a result. > > Allowing a constant zero operand in those insns seems to be a simple > solution that only affects minimum existing code. > > gcc/ChangeLog: > > * config/riscv/vector.md: Allow zero operand for DI variants of > vssubu.vx > > gcc/testsuite/ChangeLog: > > * gcc.target/riscv/rvv/base/vssubu-1.c: New test. > * gcc.target/riscv/rvv/base/vssubu-2.c: New test. > ---
diff --git a/gcc/config/riscv/vector.md b/gcc/config/riscv/vector.md index d0677325ba1..66a2a477faf 100644 --- a/gcc/config/riscv/vector.md +++ b/gcc/config/riscv/vector.md @@ -4400,10 +4400,10 @@ (define_insn "*pred_<optab><mode>_scalar" (sat_int_minus_binop:VI_D (match_operand:VI_D 3 "register_operand" " vr, vr, vr, vr") (vec_duplicate:VI_D - (match_operand:<VEL> 4 "register_operand" " r, r, r, r"))) + (match_operand:<VEL> 4 "reg_or_0_operand" " rJ, rJ, rJ, rJ"))) (match_operand:VI_D 2 "vector_merge_operand" " vu, 0, vu, 0")))] "TARGET_VECTOR" - "v<insn>.vx\t%0,%3,%4%p1" + "v<insn>.vx\t%0,%3,%z4%p1" [(set_attr "type" "<int_binop_insn_type>") (set_attr "mode" "<MODE>")]) @@ -4422,10 +4422,10 @@ (define_insn "*pred_<optab><mode>_extended_scalar" (match_operand:VI_D 3 "register_operand" " vr, vr, vr, vr") (vec_duplicate:VI_D (sign_extend:<VEL> - (match_operand:<VSUBEL> 4 "register_operand" " r, r, r, r")))) + (match_operand:<VSUBEL> 4 "reg_or_0_operand" " rJ, rJ, rJ, rJ")))) (match_operand:VI_D 2 "vector_merge_operand" " vu, 0, vu, 0")))] "TARGET_VECTOR && !TARGET_64BIT" - "v<insn>.vx\t%0,%3,%4%p1" + "v<insn>.vx\t%0,%3,%z4%p1" [(set_attr "type" "<int_binop_insn_type>") (set_attr "mode" "<MODE>")]) diff --git a/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-1.c b/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-1.c new file mode 100644 index 00000000000..f19b42aed04 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-1.c @@ -0,0 +1,11 @@ +/* { dg-do compile } */ +/* { dg-options "-O2 -march=rv64gcv -mabi=lp64d" } */ + +#include <riscv_vector.h> + +vuint64m1_t test_vssubu_vx_u64m1(vuint64m1_t op1) +{ + return __riscv_vssubu_vx_u64m1(op1,0,0); +} + +/* { dg-final { scan-assembler-not {\tvssubu} } } */ \ No newline at end of file diff --git a/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-2.c b/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-2.c new file mode 100644 index 00000000000..cb4e4f48a9b --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/base/vssubu-2.c @@ -0,0 +1,11 @@ +/* { dg-do compile } */ +/* { dg-options "-O2 -march=rv32gcv -mabi=ilp32d" } */ + +#include <riscv_vector.h> + +vuint64m1_t test_vssubu_vx_u64m1(vuint64m1_t op1) +{ + return __riscv_vssubu_vx_u64m1(op1,0,0); +} + +/* { dg-final { scan-assembler-not {\tvssubu} } } */ \ No newline at end of file