I meant to send this to the newlib list.
Apparently yesterday was a bad day for sending emails correctly. :(
Andrew
On 19/03/2025 15:04, Andrew Stubbs wrote:
Since January, GCC has been miscompiling Newlib libm on AMD GCN due to
undefined behaviour in the RESIZE_VECTOR macro. It was "working" but expanding
the size of a vector would no longer zero the additional lanes, as it expected.
See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119325
---
newlib/libm/machine/amdgcn/amdgcn_veclib.h | 17 +++++++++++++++--
1 file changed, 15 insertions(+), 2 deletions(-)
diff --git a/newlib/libm/machine/amdgcn/amdgcn_veclib.h
b/newlib/libm/machine/amdgcn/amdgcn_veclib.h
index bd67740ac..9e9d3ebf0 100644
--- a/newlib/libm/machine/amdgcn/amdgcn_veclib.h
+++ b/newlib/libm/machine/amdgcn/amdgcn_veclib.h
@@ -85,8 +85,21 @@ typedef union {
#define RESIZE_VECTOR(to_t, from) \
({ \
- __auto_type __from = (from); \
- *((to_t *) &__from); \
+ to_t __to; \
+ if (VECTOR_WIDTH (to_t) < VECTOR_WIDTH (__typeof (from))) \
+ asm ("; no-op cast %0" : "=v"(__to) : "0"(from)); \
+ else \
+ { \
+ unsigned long __mask = -1L; \
+ int lanes = VECTOR_WIDTH (__typeof (from)); \
+ __mask <<= lanes; \
+ __builtin_choose_expr ( \
+ V_SF_SI_P (to_t), \
+ ({asm ("v_mov_b32 %0, 0" : "=v"(__to) : "0"(from), "e"(__mask));}), \
+ ({asm ("v_mov_b32 %H0, 0\n\t" \
+ "v_mov_b32 %L0, 0" : "=v"(__to) : "0"(from), "e"(__mask));})); \
+ } \
+ __to; \
})
/* Bit-wise cast vector FROM to type TO_T. */
