i386: Fix ICE during cbranchv16qi4 expansion [PR112681]

The following testcase ICEs, because cbranchv16qi4 expansion calls
ix86_expand_branch with op1 being a pre-AVX unaligned memory and
ix86_expand_branch emits a xorv16qi3 instruction without making sure
the operand predicates are satisfied.
While I could manually check if the argument (or both?) doesn't
match vector_operand predicate (apparently this one or bcst_vector_operand
is used in all integral 16+ bytes *xorv*3 instructions) force it into a
register, but as all gen_xorv*3 expanders call
ix86_expand_vector_logical_operator, it seems easier to just call that
function which ensures the right thing happens.  Calling the individual
gen_xorv*3 functions would mean ugly switch on the modes and using high
level expand_simple_binop here seems too high level to me.

2023-11-24  Jakub Jelinek  <jakub@redhat.com>

	PR target/112681
	* config/i386/i386-expand.cc (ix86_expand_branch): Use
	ix86_expand_vector_logical_operator to expand vector XOR rather than
	gen_rtx_SET on gen_rtx_XOR.

	* gcc.target/i386/sse4-pr112681.c: New test.
This commit is contained in:
Jakub Jelinek 2023-11-24 12:12:20 +01:00
parent fea27dfd22
commit 3eb9cae6d3
2 changed files with 13 additions and 1 deletions

View File

@ -2453,7 +2453,8 @@ ix86_expand_branch (enum rtx_code code, rtx op0, rtx op1, rtx label)
/* Generate XOR since we can't check that one operand is zero
vector. */
tmp = gen_reg_rtx (mode);
emit_insn (gen_rtx_SET (tmp, gen_rtx_XOR (mode, op0, op1)));
rtx ops[3] = { tmp, op0, op1 };
ix86_expand_vector_logical_operator (XOR, mode, ops);
tmp = gen_lowpart (p_mode, tmp);
emit_insn (gen_rtx_SET (gen_rtx_REG (CCZmode, FLAGS_REG),
gen_rtx_UNSPEC (CCZmode,

View File

@ -0,0 +1,11 @@
/* PR target/112681 */
/* { dg-do compile } */
/* { dg-options "-O2 -msse4 -mno-avx" } */
struct S { void *c; char d[16]; } a, b;
int
foo (void)
{
return __builtin_memcmp (a.d, b.d, sizeof (a.d)) != 0;
}