Minki/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2024-12-28 05:41:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

crypto: arm64/crc-t10dif - move NEON yield to C code

Browse Source 
Instead of yielding from the bowels of the asm routine if a reschedule
is needed, divide up the input into 4 KB chunks in the C glue. This
simplifies the code substantially, and avoids scheduling out the task
with the asm routine on the call stack, which is undesirable from a
CFI/instrumentation point of view.

Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This commit is contained in:
Ard Biesheuvel 2021-02-03 12:36:25 +01:00 committed by Herbert Xu
parent f0070f4a79
commit fc754c024a
2 changed files with 35 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
arch/arm64/crypto/crct10dif-ce-core.S
View File

@ -68,10 +68,10 @@
.text
.arch armv8-a+crypto
init_crc .req w19
buf .req x20
len .req x21
fold_consts_ptr .req x22
init_crc .req w0
buf .req x1
len .req x2
fold_consts_ptr .req x3
fold_consts .req v10
@ -257,12 +257,6 @@ CPU_LE( ext v12.16b, v12.16b, v12.16b, #8 )
.endm
.macro crc_t10dif_pmull, p
frame_push 4, 128
mov init_crc, w0
mov buf, x1
mov len, x2
__pmull_init_\p
// For sizes less than 256 bytes, we can't fold 128 bytes at a time.
@ -317,26 +311,7 @@ CPU_LE( ext v7.16b, v7.16b, v7.16b, #8 )
fold_32_bytes \p, v6, v7
subs len, len, #128
b.lt .Lfold_128_bytes_loop_done_\@
if_will_cond_yield_neon
stp q0, q1, [sp, #.Lframe_local_offset]
stp q2, q3, [sp, #.Lframe_local_offset + 32]
stp q4, q5, [sp, #.Lframe_local_offset + 64]
stp q6, q7, [sp, #.Lframe_local_offset + 96]
do_cond_yield_neon
ldp q0, q1, [sp, #.Lframe_local_offset]
ldp q2, q3, [sp, #.Lframe_local_offset + 32]
ldp q4, q5, [sp, #.Lframe_local_offset + 64]
ldp q6, q7, [sp, #.Lframe_local_offset + 96]
ld1 {fold_consts.2d}, [fold_consts_ptr]
__pmull_init_\p
__pmull_pre_\p fold_consts
endif_yield_neon
b .Lfold_128_bytes_loop_\@
.Lfold_128_bytes_loop_done_\@:
b.ge .Lfold_128_bytes_loop_\@
// Now fold the 112 bytes in v0-v6 into the 16 bytes in v7.
@ -453,7 +428,9 @@ CPU_LE( ext v0.16b, v0.16b, v0.16b, #8 )
// Final CRC value (x^16 * M(x)) mod G(x) is in low 16 bits of v0.
umov w0, v0.h[0]
frame_pop
.ifc \p, p8
ldp x29, x30, [sp], #16
.endif
ret
.Lless_than_256_bytes_\@:
@ -489,7 +466,9 @@ CPU_LE( ext v7.16b, v7.16b, v7.16b, #8 )
// Assumes len >= 16.
//
SYM_FUNC_START(crc_t10dif_pmull_p8)
crc_t10dif_pmull p8
stp x29, x30, [sp, #-16]!
mov x29, sp
crc_t10dif_pmull p8
SYM_FUNC_END(crc_t10dif_pmull_p8)
.align 5

30
arch/arm64/crypto/crct10dif-ce-glue.c
View File

@ -37,9 +37,18 @@ static int crct10dif_update_pmull_p8(struct shash_desc *desc, const u8 *data,
u16 *crc = shash_desc_ctx(desc);
if (length >= CRC_T10DIF_PMULL_CHUNK_SIZE && crypto_simd_usable()) {
kernel_neon_begin();
*crc = crc_t10dif_pmull_p8(*crc, data, length);
kernel_neon_end();
do {
unsigned int chunk = length;
if (chunk > SZ_4K + CRC_T10DIF_PMULL_CHUNK_SIZE)
chunk = SZ_4K;
kernel_neon_begin();
*crc = crc_t10dif_pmull_p8(*crc, data, chunk);
kernel_neon_end();
data += chunk;
length -= chunk;
} while (length);
} else {
*crc = crc_t10dif_generic(*crc, data, length);
}
@ -53,9 +62,18 @@ static int crct10dif_update_pmull_p64(struct shash_desc *desc, const u8 *data,
u16 *crc = shash_desc_ctx(desc);
if (length >= CRC_T10DIF_PMULL_CHUNK_SIZE && crypto_simd_usable()) {
kernel_neon_begin();
*crc = crc_t10dif_pmull_p64(*crc, data, length);
kernel_neon_end();
do {
unsigned int chunk = length;
if (chunk > SZ_4K + CRC_T10DIF_PMULL_CHUNK_SIZE)
chunk = SZ_4K;
kernel_neon_begin();
*crc = crc_t10dif_pmull_p64(*crc, data, chunk);
kernel_neon_end();
data += chunk;
length -= chunk;
} while (length);
} else {
*crc = crc_t10dif_generic(*crc, data, length);
}