Fix aarch64 assembly for bignum multiplication
Add memory constraints to the aarch64 inline assembly in MULADDC_STOP.
This fixes an issue where Clang 12 and 13 were generating
non-functional code on aarch64 platforms. See #4962, #4943
for further details.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
diff --git a/library/bn_mul.h b/library/bn_mul.h
index 328e765..b71ddd8 100644
--- a/library/bn_mul.h
+++ b/library/bn_mul.h
@@ -224,7 +224,7 @@
"adcq %%rdx, %%rcx\n" \
"addq $8, %%rdi\n"
-#define MULADDC_STOP \
+#define MULADDC_STOP \
: "+c" (c), "+D" (d), "+S" (s), "+m" (*(uint64_t (*)[16]) d) \
: "b" (b), "m" (*(const uint64_t (*)[16]) s) \
: "rax", "rdx", "r8" \
@@ -240,18 +240,18 @@
#define MULADDC_CORE \
"ldr x4, [%2], #8 \n\t" \
"ldr x5, [%1] \n\t" \
- "mul x6, x4, %3 \n\t" \
- "umulh x7, x4, %3 \n\t" \
+ "mul x6, x4, %4 \n\t" \
+ "umulh x7, x4, %4 \n\t" \
"adds x5, x5, x6 \n\t" \
"adc x7, x7, xzr \n\t" \
"adds x5, x5, %0 \n\t" \
"adc %0, x7, xzr \n\t" \
"str x5, [%1], #8 \n\t"
-#define MULADDC_STOP \
- : "+r" (c), "+r" (d), "+r" (s) \
- : "r" (b) \
- : "x4", "x5", "x6", "x7", "cc" \
+#define MULADDC_STOP \
+ : "+r" (c), "+r" (d), "+r" (s), "+m" (*(uint64_t (*)[16]) d) \
+ : "r" (b), "m" (*(const uint64_t (*)[16]) s) \
+ : "x4", "x5", "x6", "x7", "cc" \
);
#endif /* Aarch64 */