| Summary: | two copies of a constant in two different registers | ||
|---|---|---|---|
| Product: | gcc | Reporter: | Andrew Pinski <pinskia> |
| Component: | target | Assignee: | Not yet assigned to anyone <unassigned> |
| Status: | RESOLVED WORKSFORME | ||
| Severity: | normal | CC: | gcc-bugs, pawel_sikora, steven |
| Priority: | P3 | Keywords: | missed-optimization |
| Version: | 4.1.0 | ||
| Target Milestone: | --- | ||
| Host: | Target: | i686-pc-linux-gnu | |
| Build: | Known to work: | ||
| Known to fail: | Last reconfirmed: | 2007-07-01 00:38:11 | |
| Bug Depends on: | |||
| Bug Blocks: | 16996 | ||
on x86-64 I get:
f: movl i.0(%rip), %eax
testl %eax, %eax
jne .L2
movb $2, %al
movl $2, i.0(%rip)
.L2: rep ; ret
$ pr24647.c.t97.final_cleanup
f() {
static int i;
int i1;
static int i;
<bb 0>:
i1 = i;
if (i1 == 0) goto <L0>; else goto <L1>;
<L0>:;
i = 2;
i1 = 2;
<L1>:;
return i1;
}
(In reply to comment #1) > on x86-64 I get: That is what I get on x86 also now: f: movl i.1523, %eax pushl %ebp movl %esp, %ebp testl %eax, %eax jne .L2 movl $2, i.1523 movb $2, %al .L2: popl %ebp ret But just change "== 0" to "!= 0" and you get the problem still. Also this is a code size issue for -Os as movl with a constant takes up too much space. Trunk today:
f():
movl f()::i, %eax
testl %eax, %eax
jne .L1 # or je for the != case
movl $2, f()::i
movl $2, %eax
.L1:
ret
|
Take the following code: int f(void) { static int i; int i1; i1 = i; if (i1 == 0) i = i1 = 2; return i1; } ---- Currently we generate: f: movl i.1285, %eax pushl %ebp movl %esp, %ebp testl %eax, %eax jne .L7 movl $2, %ecx movl $2, %eax movl %ecx, i.1285 .L7: popl %ebp ret Why is 2 in two different registers (I think the issue here is really a target issue)? But should postreload CSE detect this or postreload GCSE?