https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93007
Bug ID: 93007
Summary: [10 regression] pr77698.c testcase fails due to block
commoning
Product: gcc
Version: 10.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: rtl-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: wilco at gcc dot gnu.org
Target Milestone: ---
Since r276960 we see this failure on Arm:
FAIL: gcc.dg/tree-prof/pr77698.c scan-rtl-dump-times alignments "internal loop
alignment added" 1
The issue appears to be that basic block commoning works on an unrolled loop,
which is unlikely to be beneficial for performance:
.L17:
adds r0, r0, #1
b .L27
.L6:
ldr r4, [r2, #12]
adds r0, r0, #4
ldr lr, [r1]
str lr, [r3, r4, lsl #2]
ldr r4, [r2, #12]
ldr lr, [r1]
str lr, [r3, r4, lsl #2]
ldr r4, [r2, #12]
ldr lr, [r1]
str lr, [r3, r4, lsl #2]
.L27:
ldr r4, [r2, #12]
cmp ip, r0
ldr lr, [r1]
str lr, [r3, r4, lsl #2]
bne .L6
pop {r4, pc}
The test could be easily fixed, but ensuring block commoning takes loops and
execution frequencies into account would be better overall.
