https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121224
Bug ID: 121224
Summary: auto-vectorizer generates aligned accesses to
potentially unaligned memory
Product: gcc
Version: 15.1.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: olaf.krzikalla at dlr dot de
Target Milestone: ---
```
typedef __attribute__(( aligned(32))) double aligned_double;
void add(aligned_double* a, aligned_double* b, aligned_double* c, int end, int
start)
{
for (decltype(end) i = start; i < end; ++i)
c[i] = a[i] + b[i];
}
```
Compile with `gcc 15.1 -O3 -march=core-avx2`.
The auto vectorizer generates aligned loads (`movapd`). This causes seg faults,
if `start` is not properly aligned (e.g. `start == 1`).
Runable test (on gcc trunk): https://godbolt.org/z/r71EafjK8
