https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122343
Bug ID: 122343
Summary: Propagate volatile memory operands
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Severity: enhancement
Priority: P3
Component: rtl-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: ubizjak at gmail dot com
Target Milestone: ---
The following testcase:
--cut here--
volatile int m;
int foo (int z)
{
z *= 123;
return m + z;
}
--cut here--
currently compiles to (gcc -O2):
foo:
imull $123, %edi, %eax
movl m(%rip), %edx
addl %edx, %eax
ret
The memory operand can propagate to ADDL instruction without affecting volatile
correctness. Now that we have late combine pass, perhaps it can be taught to
carefully do this optimization.
OTOH, clang compiles the testcase to the expected:
foo:
imull $123, %edi, %eax
addl m(%rip), %eax
retq
This improvement would help to further optimize linux kernel.
