https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90705
Bug ID: 90705
Summary: Suboptimal register allocation on ARM when compiling
for size
Product: gcc
Version: 9.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: fredrik.hederstie...@securitas-direct.com
Target Milestone: ---
Created attachment 46441
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=46441&action=edit
test.c
When compiling this simple example for ARM (-mcpu=cortex-m0) with gcc-9.1.0,
the code generated looks ok when use -O2, but register allocations looks weird
when compiling use -Os. Registers are pushed on stack, and code actually gets
alot bigger.
Example
int k;
int test(int i)
{
int r = 0;
for (; i >= 0; i--) {
k = i;
r += k;
}
return r;
}
Compiling gcc-9.1.0, -mcpu=cortex-m0 using -O2:
00000000 <test>:
0: 0003 movs r3, r0
2: 2000 movs r0, #0
4: 2b00 cmp r3, #0
6: db05 blt.n 14 <test+0x14>
8: 18c0 adds r0, r0, r3
a: 3b01 subs r3, #1
c: d2fc bcs.n 8 <test+0x8>
e: 2200 movs r2, #0
10: 4b01 ldr r3, [pc, #4] ; (18 <test+0x18>)
12: 601a str r2, [r3, #0]
14: 4770 bx lr
16: 46c0 nop ; (mov r8, r8)
18: 00000000 .word 0x00000000
but when compiling with same compiler with -Os:
00000000 <test>:
0: 2200 movs r2, #0
2: b530 push {r4, r5, lr}
4: 0003 movs r3, r0
6: 2501 movs r5, #1
8: 0010 movs r0, r2
a: 4906 ldr r1, [pc, #24] ; (24 <test+0x24>)
c: 680c ldr r4, [r1, #0]
e: 2b00 cmp r3, #0
10: da03 bge.n 1a <test+0x1a>
12: 2a00 cmp r2, #0
14: d000 beq.n 18 <test+0x18>
16: 600c str r4, [r1, #0]
18: bd30 pop {r4, r5, pc}
1a: 001c movs r4, r3
1c: 18c0 adds r0, r0, r3
1e: 002a movs r2, r5
20: 3b01 subs r3, #1
22: e7f4 b.n e <test+0xe>
24: 00000000 .word 0x00000000
using 2 more registers and stack, also code size significantly larger.
