On Tue, Aug 3, 2021 at 1:43 AM Uros Bizjak <ubiz...@gmail.com> wrote:
>
> On Tue, Aug 3, 2021 at 10:15 AM Hongtao Liu <crazy...@gmail.com> wrote:
> >
> > On Tue, Aug 3, 2021 at 4:03 PM Uros Bizjak via Gcc-patches
> > <gcc-patches@gcc.gnu.org> wrote:
> > >
> > > On Mon, Aug 2, 2021 at 7:47 PM H.J. Lu <hjl.to...@gmail.com> wrote:
> > > >
> > > > In 64-bit mode, use XMM31 for scratch SSE register to avoid vzeroupper
> > > > if possible.
> > > >
> > > > gcc/
> > > >
> > > > * config/i386/i386.c (ix86_gen_scratch_sse_rtx): In 64-bit mode,
> > > > try XMM31 to avoid vzeroupper.
> > > >
> > > > gcc/testsuite/
> > > >
> > > > * gcc.target/i386/avx-vzeroupper-14.c: Pass -mno-avx512f to
> > > > disable XMM31.
> > > > * gcc.target/i386/avx-vzeroupper-15.c: Likewise.
> > > > * gcc.target/i386/pr82941-1.c: Updated. Check for vzeroupper.
> > > > * gcc.target/i386/pr82942-1.c: Likewise.
> > > > * gcc.target/i386/pr82990-1.c: Likewise.
> > > > * gcc.target/i386/pr82990-3.c: Likewise.
> > > > * gcc.target/i386/pr82990-5.c: Likewise.
> > > > * gcc.target/i386/pr100865-4b.c: Likewise.
> > > > * gcc.target/i386/pr100865-6b.c: Likewise.
> > > > * gcc.target/i386/pr100865-7b.c: Likewise.
> > > > * gcc.target/i386/pr100865-10b.c: Likewise.
> > > > * gcc.target/i386/pr100865-8b.c: Updated.
> > > > * gcc.target/i386/pr100865-9b.c: Likewise.
> > > > * gcc.target/i386/pr100865-11b.c: Likewise.
> > > > * gcc.target/i386/pr100865-12b.c: Likewise.
> > > > ---
> > > > gcc/config/i386/i386.c | 18 +++++++++++++++---
> > > > .../gcc.target/i386/avx-vzeroupper-14.c | 2 +-
> > > > .../gcc.target/i386/avx-vzeroupper-15.c | 2 +-
> > > > gcc/testsuite/gcc.target/i386/pr100865-10b.c | 1 +
> > > > gcc/testsuite/gcc.target/i386/pr100865-11b.c | 2 +-
> > > > gcc/testsuite/gcc.target/i386/pr100865-12b.c | 2 +-
> > > > gcc/testsuite/gcc.target/i386/pr100865-4b.c | 2 ++
> > > > gcc/testsuite/gcc.target/i386/pr100865-6b.c | 5 ++++-
> > > > gcc/testsuite/gcc.target/i386/pr100865-7b.c | 5 ++++-
> > > > gcc/testsuite/gcc.target/i386/pr100865-8b.c | 2 +-
> > > > gcc/testsuite/gcc.target/i386/pr100865-9b.c | 2 +-
> > > > gcc/testsuite/gcc.target/i386/pr82941-1.c | 3 ++-
> > > > gcc/testsuite/gcc.target/i386/pr82942-1.c | 3 ++-
> > > > gcc/testsuite/gcc.target/i386/pr82990-1.c | 3 ++-
> > > > gcc/testsuite/gcc.target/i386/pr82990-3.c | 3 ++-
> > > > gcc/testsuite/gcc.target/i386/pr82990-5.c | 3 ++-
> > > > 16 files changed, 42 insertions(+), 16 deletions(-)
> > > >
> > > > diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
> > > > index 842eb0e6786..ec0690876b7 100644
> > > > --- a/gcc/config/i386/i386.c
> > > > +++ b/gcc/config/i386/i386.c
> > > > @@ -23335,9 +23335,21 @@ rtx
> > > > ix86_gen_scratch_sse_rtx (machine_mode mode)
> > > > {
> > > > if (TARGET_SSE && !lra_in_progress)
> > > > - return gen_rtx_REG (mode, (TARGET_64BIT
> > > > - ? LAST_REX_SSE_REG
> > > > - : LAST_SSE_REG));
> > > > + {
> > > > + unsigned int regno;
> > > > + if (TARGET_64BIT)
> > > > + {
> > > > + /* In 64-bit mode, use XMM31 to avoid vzeroupper and always
> > > > + use XMM31 for CSE. */
> > > > + if (ix86_hard_regno_mode_ok (LAST_EXT_REX_SSE_REG, mode))
> > > > + regno = LAST_EXT_REX_SSE_REG;
> > > > + else
> > > > + regno = LAST_REX_SSE_REG;
> > > > + }
> > > > + else
> > > > + regno = LAST_SSE_REG;
> > >
> > > Assuming that ix86_hard_regno_mode_ok always returns false for XMM31
> > > in 64bit mode, we can do:
> > >
> > > /* Use XMM31 if available to avoid vzeroupper. */
> > > if (ix86_hard_regno_mode_ok (LAST_EXT_REX_SSE_REG, mode))
It doesn't work for -m32 since ix86_hard_regno_mode_ok doesn't check
for TARGET_64BIT. LAST_EXST_REX_SSE_REG is used for -m32:
$ /export/build/gnu/tools-build/gcc-gitlab-debug/build-x86_64-linux/gcc/xgcc
-B/export/build/gnu/tools-build/gcc-gitlab-debug/build-x86_64-linux/gcc/
/export/gnu/import/git/gitlab/x86-gcc/gcc/testsuite/gcc.target/i386/pr82941-1.c
-m32 -fdiagnostics-plain-output -O2 -march=skylake-avx512
-ffat-lto-objects -fno-ident -S -o pr82941-1.s
xgcc: internal compiler error: Segmentation fault signal terminated program cc1
Please submit a full bug report,
with preprocessed source if appropriate.
See <https://gcc.gnu.org/bugs/> for instructions.
$
Program received signal SIGSEGV, Segmentation fault.
0x0000000001201c64 in general_operand (
op=<error reading variable: Cannot access memory at address
0x7ffffbffeff8>, mode=<error reading variable: Cannot access memory at
address 0x7ffffbffeff4>)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/recog.c:1412
1412 {
(gdb) bt
#0 0x0000000001201c64 in general_operand (
op=<error reading variable: Cannot access memory at address
0x7ffffbffeff8>, mode=<error reading variable: Cannot access memory at
address 0x7ffffbffeff4>)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/recog.c:1412
#1 0x000000000120221e in register_operand (op=0x7ffff5c10510, mode=E_V8DFmode)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/recog.c:1559
#2 0x00000000018fcfdb in ix86_expand_vector_move (mode=E_V8DFmode,
operands=0x7ffffbfff190)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/config/i386/i386-expand.c:625
#3 0x0000000001ef9bc2 in gen_movv8df (operand0=0x7ffff5c10510,
operand1=0x7ffff6351d98)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/config/i386/sse.md:1051
#4 0x0000000000de6cfb in insn_gen_fn::operator()<rtx_def*, rtx_def*> (
this=0x2d88908 <insn_data+278376>)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/recog.h:407
#5 0x0000000000dbe4ec in emit_move_insn_1 (x=0x7ffff5c10510, y=0x7ffff6351d98)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/expr.c:3930
#6 0x0000000000dbefde in emit_move_insn (x=0x7ffff5c10510, y=0x7ffff6351d98)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/expr.c:4100
#7 0x00000000018fd036 in ix86_expand_vector_move (mode=E_V8DFmode,
operands=0x7ffffbfff420)
at /export/gnu/import/git/gitlab/x86-gcc/gcc/config/i386/i386-expand.c:629
#8 0x0000000001ef9bc2 in gen_movv8df (operand0=0x7ffff5c104e0,
> > > regno = LAST_EXST_REX_SSE_REG;
> > > else if (TARGET_64BIT)
> > > regno = LAST_EXT_REX_SSE_REG;
>
> > why? w/o avx512 xmm31 is not available.
>
> Oh, a typo, this should read LAST_REX_SSE_REG.
>
> Uros.
--
H.J.
