Hi!
This is a similar problem as PR52086, but back then when looking
at whether other peepholes might need similar treatment I didn't think
about the possibility that addqi_1 pattern would allow non-q constraint
register. Apparently it does in some of the alternatives, the insn is then
emitted widened as addl etc., but for addqi_2 it isn't possible (it adds to
memory).
Fixed thusly, bootstrapped/regtested on i686-linux, ok for trunk/4.7?
2012-05-15 Jakub Jelinek <ja...@redhat.com>
PR target/53358
* config/i386/i386.md (*addqi_2 peephole with QImode addition): Check
that operands[2] is either immediate, or q_regs_operand.
* gcc.dg/pr53358.c: New test.
--- gcc/config/i386/i386.md.jj 2012-05-02 09:42:33.000000000 +0200
+++ gcc/config/i386/i386.md 2012-05-15 10:48:19.197020618 +0200
@@ -17209,6 +17209,9 @@ (define_peephole2
"(TARGET_READ_MODIFY_WRITE || optimize_insn_for_size_p ())
&& peep2_reg_dead_p (4, operands[0])
&& !reg_overlap_mentioned_p (operands[0], operands[1])
+ && (<MODE>mode != QImode
+ || immediate_operand (operands[2], QImode)
+ || q_regs_operand (operands[2], QImode))
&& ix86_match_ccmode (peep2_next_insn (3),
(GET_CODE (operands[3]) == PLUS
|| GET_CODE (operands[3]) == MINUS)
--- gcc/testsuite/gcc.dg/pr53358.c.jj 2012-05-15 11:54:40.680128157 +0200
+++ gcc/testsuite/gcc.dg/pr53358.c 2012-05-15 11:54:47.698086318 +0200
@@ -0,0 +1,22 @@
+/* PR target/53358 */
+/* { dg-do compile } */
+/* { dg-options "-O2" } */
+/* { dg-additional-options "-fpic" { target fpic } } */
+/* { dg-additional-options "-mtune=pentium4" { target { { i?86-*-* x86_64-*-*
} && ia32 } } } */
+
+struct S { unsigned char s, t[17]; };
+int bar (void);
+
+void
+foo (struct S *x)
+{
+ unsigned char i, z;
+ if (bar ())
+ {
+ z = bar ();
+ bar ();
+ x->s += z;
+ for (i = 0; i < x->s; i++)
+ x->t[i] = bar ();
+ }
+}
Jakub
