From: Eric Biggers <ebiggers@xxxxxxxxxx>
Fix crc32c-pcl-intel-asm_64.S to access 32-bit arguments as 32-bit
values instead of 64-bit, since the upper bits of the corresponding
64-bit registers are not guaranteed to be zero. Also update the type of
the length argument to be unsigned int rather than int, as the assembly
code treats it as unsigned.
Note: there haven't been any reports of this bug actually causing
incorrect behavior. Neither gcc nor clang guarantee zero-extension to
64 bits, but zero-extension is likely to happen in practice because most
instructions that operate on 32-bit registers zero-extend to 64 bits.
Signed-off-by: Eric Biggers <ebiggers@xxxxxxxxxx>
---
arch/x86/crypto/crc32c-intel_glue.c | 2 +-
arch/x86/crypto/crc32c-pcl-intel-asm_64.S | 57 +++++++++++------------
2 files changed, 27 insertions(+), 32 deletions(-)
diff --git a/arch/x86/crypto/crc32c-intel_glue.c b/arch/x86/crypto/crc32c-intel_glue.c
index feccb5254c7e5..52c5d47ef5a14 100644
--- a/arch/x86/crypto/crc32c-intel_glue.c
+++ b/arch/x86/crypto/crc32c-intel_glue.c
@@ -39,11 +39,11 @@
* size is >= 512 to account
* for fpu state save/restore overhead.
*/
#define CRC32C_PCL_BREAKEVEN 512
-asmlinkage unsigned int crc_pcl(const u8 *buffer, int len,
+asmlinkage unsigned int crc_pcl(const u8 *buffer, unsigned int len,
unsigned int crc_init);
#endif /* CONFIG_X86_64 */
static u32 crc32c_intel_le_hw_byte(u32 crc, unsigned char const *data, size_t length)
{
diff --git a/arch/x86/crypto/crc32c-pcl-intel-asm_64.S b/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
index 466cea4943963..bbf860e90951d 100644
--- a/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
+++ b/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
@@ -58,11 +58,11 @@
# Define threshold below which buffers are considered "small" and routed to
# regular CRC code that does not interleave the CRC instructions.
#define SMALL_SIZE 200
-# unsigned int crc_pcl(u8 *buffer, int len, unsigned int crc_init);
+# unsigned int crc_pcl(const u8 *buffer, unsigned int len, unsigned int crc_init);
.text
SYM_FUNC_START(crc_pcl)
#define bufp rdi
#define bufp_dw %edi
@@ -70,18 +70,15 @@ SYM_FUNC_START(crc_pcl)
#define bufp_b %dil
#define bufptmp %rcx
#define block_0 %rcx
#define block_1 %rdx
#define block_2 %r11
-#define len %rsi
-#define len_dw %esi
-#define len_w %si
-#define len_b %sil
-#define crc_init_arg %rdx
+#define len %esi
+#define crc_init_arg %edx
#define tmp %rbx
-#define crc_init %r8
-#define crc_init_dw %r8d
+#define crc_init %r8d
+#define crc_init_q %r8
#define crc1 %r9
#define crc2 %r10
pushq %rbx
pushq %rdi
@@ -105,13 +102,13 @@ SYM_FUNC_START(crc_pcl)
.Ldo_align:
#### Calculate CRC of unaligned bytes of the buffer (if any)
movq (bufptmp), tmp # load a quadward from the buffer
add %bufp, bufptmp # align buffer pointer for quadword
# processing
- sub %bufp, len # update buffer length
+ sub bufp_dw, len # update buffer length
.Lalign_loop:
- crc32b %bl, crc_init_dw # compute crc32 of 1-byte
+ crc32b %bl, crc_init # compute crc32 of 1-byte
shr $8, tmp # get next byte
dec %bufp
jne .Lalign_loop
.Lproc_block:
@@ -119,19 +116,18 @@ SYM_FUNC_START(crc_pcl)
################################################################
## 2) PROCESS BLOCKS:
################################################################
## compute num of bytes to be processed
- movq len, tmp # save num bytes in tmp
- cmpq $128*24, len
+ cmp $128*24, len
jae .Lfull_block
.Lcontinue_block:
## len < 128*24
movq $2731, %rax # 2731 = ceil(2^16 / 24)
- mul len_dw
+ mul len
shrq $16, %rax
## eax contains floor(bytes / 24) = num 24-byte chunks to do
## process rax 24-byte chunks (128 >= rax >= 0)
@@ -174,21 +170,21 @@ SYM_FUNC_START(crc_pcl)
.rept 128-1
.altmacro
LABEL crc_ %i
.noaltmacro
ENDBR
- crc32q -i*8(block_0), crc_init
+ crc32q -i*8(block_0), crc_init_q
crc32q -i*8(block_1), crc1
crc32q -i*8(block_2), crc2
i=(i-1)
.endr
.altmacro
LABEL crc_ %i
.noaltmacro
ENDBR
- crc32q -i*8(block_0), crc_init
+ crc32q -i*8(block_0), crc_init_q
crc32q -i*8(block_1), crc1
# SKIP crc32 -i*8(block_2), crc2 ; Don't do this one yet
mov block_2, block_0
@@ -198,66 +194,65 @@ LABEL crc_ %i
lea (K_table-8)(%rip), %bufp # first entry is for idx 1
shlq $3, %rax # rax *= 8
pmovzxdq (%bufp,%rax), %xmm0 # 2 consts: K1:K2
leal (%eax,%eax,2), %eax # rax *= 3 (total *24)
- subq %rax, tmp # tmp -= rax*24
+ sub %eax, len # len -= rax*24
- movq crc_init, %xmm1 # CRC for block 1
+ movq crc_init_q, %xmm1 # CRC for block 1
pclmulqdq $0x00, %xmm0, %xmm1 # Multiply by K2
movq crc1, %xmm2 # CRC for block 2
pclmulqdq $0x10, %xmm0, %xmm2 # Multiply by K1
pxor %xmm2,%xmm1
movq %xmm1, %rax
xor -i*8(block_2), %rax
- mov crc2, crc_init
- crc32 %rax, crc_init
+ mov crc2, crc_init_q
+ crc32 %rax, crc_init_q
################################################################
## 5) Check for end:
################################################################
LABEL crc_ 0
ENDBR
- mov tmp, len
- cmp $128*24, tmp
+ cmp $128*24, len
jae .Lfull_block
- cmp $SMALL_SIZE, tmp
+ cmp $SMALL_SIZE, len
jae .Lcontinue_block
#######################################################################
## 6) Process any remainder without interleaving:
#######################################################################
.Lsmall:
test len, len
jz .Ldone
- mov len_dw, %eax
+ mov len, %eax
shr $3, %eax
jz .Ldo_dword
.Ldo_qwords:
- crc32q (bufptmp), crc_init
+ crc32q (bufptmp), crc_init_q
add $8, bufptmp
dec %eax
jnz .Ldo_qwords
.Ldo_dword:
- test $4, len_dw
+ test $4, len
jz .Ldo_word
- crc32l (bufptmp), crc_init_dw
+ crc32l (bufptmp), crc_init
add $4, bufptmp
.Ldo_word:
- test $2, len_dw
+ test $2, len
jz .Ldo_byte
- crc32w (bufptmp), crc_init_dw
+ crc32w (bufptmp), crc_init
add $2, bufptmp
.Ldo_byte:
- test $1, len_dw
+ test $1, len
jz .Ldone
- crc32b (bufptmp), crc_init_dw
+ crc32b (bufptmp), crc_init
.Ldone:
- movq crc_init, %rax
+ mov crc_init, %eax
popq %rsi
popq %rdi
popq %rbx
RET
SYM_FUNC_END(crc_pcl)
--
2.47.0
