@@ -1002,7 +1002,13 @@ void __crypto_xor(u8 *dst, const u8 *src1, const u8 *src2, unsigned int len)
}
while (IS_ENABLED(CONFIG_64BIT) && len >= 8 && !(relalign & 7)) {
- *(u64 *)dst = *(u64 *)src1 ^ *(u64 *)src2;
+ if (IS_ENABLED(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS)) {
+ u64 l = get_unaligned((u64 *)src1) ^
+ get_unaligned((u64 *)src2);
+ put_unaligned(l, (u64 *)dst);
+ } else {
+ *(u64 *)dst = *(u64 *)src1 ^ *(u64 *)src2;
+ }
dst += 8;
src1 += 8;
src2 += 8;
@@ -1010,7 +1016,13 @@ void __crypto_xor(u8 *dst, const u8 *src1, const u8 *src2, unsigned int len)
}
while (len >= 4 && !(relalign & 3)) {
- *(u32 *)dst = *(u32 *)src1 ^ *(u32 *)src2;
+ if (IS_ENABLED(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS)) {
+ u32 l = get_unaligned((u32 *)src1) ^
+ get_unaligned((u32 *)src2);
+ put_unaligned(l, (u32 *)dst);
+ } else {
+ *(u32 *)dst = *(u32 *)src1 ^ *(u32 *)src2;
+ }
dst += 4;
src1 += 4;
src2 += 4;
@@ -1018,7 +1030,13 @@ void __crypto_xor(u8 *dst, const u8 *src1, const u8 *src2, unsigned int len)
}
while (len >= 2 && !(relalign & 1)) {
- *(u16 *)dst = *(u16 *)src1 ^ *(u16 *)src2;
+ if (IS_ENABLED(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS)) {
+ u16 l = get_unaligned((u16 *)src1) ^
+ get_unaligned((u16 *)src2);
+ put_unaligned(l, (u16 *)dst);
+ } else {
+ *(u16 *)dst = *(u16 *)src1 ^ *(u16 *)src2;
+ }
dst += 2;
src1 += 2;
src2 += 2;
@@ -13,6 +13,8 @@
#include <linux/list.h>
#include <linux/types.h>
+#include <asm/unaligned.h>
+
/*
* Maximum values for blocksize and alignmask, used to allocate
* static buffers that are big enough for any combination of
@@ -154,9 +156,11 @@ static inline void crypto_xor(u8 *dst, const u8 *src, unsigned int size)
(size % sizeof(unsigned long)) == 0) {
unsigned long *d = (unsigned long *)dst;
unsigned long *s = (unsigned long *)src;
+ unsigned long l;
while (size > 0) {
- *d++ ^= *s++;
+ l = get_unaligned(d) ^ get_unaligned(s++);
+ put_unaligned(l, d++);
size -= sizeof(unsigned long);
}
} else {
@@ -173,9 +177,12 @@ static inline void crypto_xor_cpy(u8 *dst, const u8 *src1, const u8 *src2,
unsigned long *d = (unsigned long *)dst;
unsigned long *s1 = (unsigned long *)src1;
unsigned long *s2 = (unsigned long *)src2;
+ unsigned long l;
while (size > 0) {
- *d++ = *s1++ ^ *s2++;
+ l = get_unaligned(d) ^ get_unaligned(s1++)
+ ^ get_unaligned(s2++);
+ put_unaligned(l, d++);
size -= sizeof(unsigned long);
}
} else {
Dereferencing a misaligned pointer is undefined behavior in C, and may result in codegen on architectures such as ARM that trigger alignments traps and expensive fixups in software. Instead, use the get_aligned()/put_aligned() accessors, which are cheap or even completely free when CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS=y. In the converse case, the prior alignment checks ensure that the casts are safe, and so no unaligned accessors are necessary. Signed-off-by: Ard Biesheuvel <ardb@kernel.org> --- crypto/algapi.c | 24 +++++++++++++++++++++--- include/crypto/algapi.h | 11 +++++++++-- 2 files changed, 30 insertions(+), 5 deletions(-)