https://bugs.llvm.org/show_bug.cgi?id=49708
Bug ID: 49708
Summary: std::memcpy with incorrect align attribute on
arguments introduces undefined behavior
Product: clang
Version: trunk
Hardware: All
OS: All
Status: NEW
Severity: normal
Priority: P
Component: LLVM Codegen
Assignee: [email protected]
Reporter: [email protected]
CC: [email protected], [email protected],
[email protected]
Example (https://clang.godbolt.org/z/7r8qndWzG):
#include <cstring>
void f(int* src, int* dst) {
std::memcpy(src, dst, sizeof(int));
}
void g(char* a, char *b) {
f((int*)a, (int*)b);
}
lowers to: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %6, i8* align 4 %8,
i64 4, i1 false)
The "align 4" in the pointer arguments of memcpy is incorrect.
No C++ code loads/stores from the src and dst pointers at type "int", and IIUC
std::memcpy(void*,void*,size) access its arguments at type "unsigned char*", so
TBAA does not allow inferring that the pointers are 4 byte aligned in the
memcpy.
As a consequence, the memcpy loads/stores with an incorrect alignment, and the
behavior is undefined.
--
You are receiving this mail because:
You are on the CC list for the bug._______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
