[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20221006223151.22159-3-elliott@hpe.com>
Date: Thu, 6 Oct 2022 17:31:46 -0500
From: Robert Elliott <elliott@....com>
To: herbert@...dor.apana.org.au, davem@...emloft.net,
tim.c.chen@...ux.intel.com, linux-crypto@...r.kernel.org,
linux-kernel@...r.kernel.org
Cc: Robert Elliott <elliott@....com>
Subject: [RFC PATCH 2/7] crypto: x86/sha - limit FPU preemption
As done by the ECB and CBC helpers in arch/x86/crypt/ecb_cbc_helpers.h,
limit the number of bytes processed between kernel_fpu_begin() and
kernel_fpu_end() calls.
Those functions call preempt_disable() and preempt_enable(), so
the CPU core is unavailable for scheduling while running.
This leads to "rcu_preempt detected expedited stalls" with stack dumps
pointing to the optimized hash function if this module is loaded and
used a lot:
rcu: INFO: rcu_preempt detected expedited stalls on CPUs/tasks: {12-... } 22 jiffies s: 277 root: 0x1/.
For example, that can occur during boot with the stack track pointing
to the sha512-x86 function if the system set to use SHA-512 for
module signing. The call trace includes:
module_sig_check
mod_verify_sig
pkcs7_verify
pkcs7_digest
sha512_finup
sha512_base_do_update
Fixes: 66be89515888 ("crypto: sha1 - SSSE3 based SHA1 implementation for x86-64")
Fixes: 8275d1aa6422 ("crypto: sha256 - Create module providing optimized SHA256 routines using SSSE3, AVX or AVX2 instructions.")
Fixes: 87de4579f92d ("crypto: sha512 - Create module providing optimized SHA512 routines using SSSE3, AVX or AVX2 instructions.")
Fixes: aa031b8f702e ("crypto: x86/sha512 - load based on CPU features")
Suggested-by: Herbert Xu <herbert@...dor.apana.org.au>
Reviewed-by: Tim Chen <tim.c.chen@...ux.intel.com>
Signed-off-by: Robert Elliott <elliott@....com>
---
arch/x86/crypto/sha1_ssse3_glue.c | 34 +++++++++++++++++++++++-----
arch/x86/crypto/sha256_ssse3_glue.c | 35 ++++++++++++++++++++++++-----
arch/x86/crypto/sha512_ssse3_glue.c | 35 ++++++++++++++++++++++++-----
3 files changed, 89 insertions(+), 15 deletions(-)
diff --git a/arch/x86/crypto/sha1_ssse3_glue.c b/arch/x86/crypto/sha1_ssse3_glue.c
index 4430463dee62..033812989476 100644
--- a/arch/x86/crypto/sha1_ssse3_glue.c
+++ b/arch/x86/crypto/sha1_ssse3_glue.c
@@ -27,10 +27,13 @@
#include <asm/cpu_device_id.h>
#include <asm/simd.h>
+#define FPU_BYTES 4096U /* avoid kernel_fpu_begin/end scheduler/rcu stalls */
+
static int sha1_update(struct shash_desc *desc, const u8 *data,
unsigned int len, sha1_block_fn *sha1_xform)
{
struct sha1_state *sctx = shash_desc_ctx(desc);
+ unsigned int chunk;
if (!crypto_simd_usable() ||
(sctx->count % SHA1_BLOCK_SIZE) + len < SHA1_BLOCK_SIZE)
@@ -42,9 +45,18 @@ static int sha1_update(struct shash_desc *desc, const u8 *data,
*/
BUILD_BUG_ON(offsetof(struct sha1_state, state) != 0);
- kernel_fpu_begin();
- sha1_base_do_update(desc, data, len, sha1_xform);
- kernel_fpu_end();
+ do {
+ chunk = min(len, FPU_BYTES);
+ len -= chunk;
+
+ if (chunk) {
+ kernel_fpu_begin();
+ sha1_base_do_update(desc, data, chunk, sha1_xform);
+ kernel_fpu_end();
+ }
+
+ data += chunk;
+ } while (len);
return 0;
}
@@ -52,12 +64,24 @@ static int sha1_update(struct shash_desc *desc, const u8 *data,
static int sha1_finup(struct shash_desc *desc, const u8 *data,
unsigned int len, u8 *out, sha1_block_fn *sha1_xform)
{
+ unsigned int chunk;
+
if (!crypto_simd_usable())
return crypto_sha1_finup(desc, data, len, out);
+ do {
+ chunk = min(len, FPU_BYTES);
+ len -= chunk;
+
+ if (chunk) {
+ kernel_fpu_begin();
+ sha1_base_do_update(desc, data, chunk, sha1_xform);
+ kernel_fpu_end();
+ }
+ data += chunk;
+ } while (len);
+
kernel_fpu_begin();
- if (len)
- sha1_base_do_update(desc, data, len, sha1_xform);
sha1_base_do_finalize(desc, sha1_xform);
kernel_fpu_end();
diff --git a/arch/x86/crypto/sha256_ssse3_glue.c b/arch/x86/crypto/sha256_ssse3_glue.c
index e437fba0299b..99a25c238f40 100644
--- a/arch/x86/crypto/sha256_ssse3_glue.c
+++ b/arch/x86/crypto/sha256_ssse3_glue.c
@@ -40,6 +40,8 @@
#include <asm/cpu_device_id.h>
#include <asm/simd.h>
+#define FPU_BYTES 4096U /* avoid kernel_fpu_begin/end scheduler/rcu stalls */
+
asmlinkage void sha256_transform_ssse3(struct sha256_state *state,
const u8 *data, int blocks);
@@ -47,6 +49,7 @@ static int _sha256_update(struct shash_desc *desc, const u8 *data,
unsigned int len, sha256_block_fn *sha256_xform)
{
struct sha256_state *sctx = shash_desc_ctx(desc);
+ unsigned int chunk;
if (!crypto_simd_usable() ||
(sctx->count % SHA256_BLOCK_SIZE) + len < SHA256_BLOCK_SIZE)
@@ -58,9 +61,18 @@ static int _sha256_update(struct shash_desc *desc, const u8 *data,
*/
BUILD_BUG_ON(offsetof(struct sha256_state, state) != 0);
- kernel_fpu_begin();
- sha256_base_do_update(desc, data, len, sha256_xform);
- kernel_fpu_end();
+ do {
+ chunk = min(len, FPU_BYTES);
+ len -= chunk;
+
+ if (chunk) {
+ kernel_fpu_begin();
+ sha256_base_do_update(desc, data, chunk, sha256_xform);
+ kernel_fpu_end();
+ }
+
+ data += chunk;
+ } while (len);
return 0;
}
@@ -68,12 +80,25 @@ static int _sha256_update(struct shash_desc *desc, const u8 *data,
static int sha256_finup(struct shash_desc *desc, const u8 *data,
unsigned int len, u8 *out, sha256_block_fn *sha256_xform)
{
+ unsigned int chunk;
+
if (!crypto_simd_usable())
return crypto_sha256_finup(desc, data, len, out);
+ do {
+ chunk = min(len, FPU_BYTES);
+ len -= chunk;
+
+ if (chunk) {
+ kernel_fpu_begin();
+ sha256_base_do_update(desc, data, chunk, sha256_xform);
+ kernel_fpu_end();
+ }
+
+ data += chunk;
+ } while (len);
+
kernel_fpu_begin();
- if (len)
- sha256_base_do_update(desc, data, len, sha256_xform);
sha256_base_do_finalize(desc, sha256_xform);
kernel_fpu_end();
diff --git a/arch/x86/crypto/sha512_ssse3_glue.c b/arch/x86/crypto/sha512_ssse3_glue.c
index 3c19f803f288..72eee03448dc 100644
--- a/arch/x86/crypto/sha512_ssse3_glue.c
+++ b/arch/x86/crypto/sha512_ssse3_glue.c
@@ -39,6 +39,8 @@
#include <asm/cpu_device_id.h>
#include <asm/simd.h>
+#define FPU_BYTES 4096U /* avoid kernel_fpu_begin/end scheduler/rcu stalls */
+
asmlinkage void sha512_transform_ssse3(struct sha512_state *state,
const u8 *data, int blocks);
@@ -46,6 +48,7 @@ static int sha512_update(struct shash_desc *desc, const u8 *data,
unsigned int len, sha512_block_fn *sha512_xform)
{
struct sha512_state *sctx = shash_desc_ctx(desc);
+ unsigned int chunk;
if (!crypto_simd_usable() ||
(sctx->count[0] % SHA512_BLOCK_SIZE) + len < SHA512_BLOCK_SIZE)
@@ -57,9 +60,18 @@ static int sha512_update(struct shash_desc *desc, const u8 *data,
*/
BUILD_BUG_ON(offsetof(struct sha512_state, state) != 0);
- kernel_fpu_begin();
- sha512_base_do_update(desc, data, len, sha512_xform);
- kernel_fpu_end();
+ do {
+ chunk = min(len, FPU_BYTES);
+ len -= chunk;
+
+ if (chunk) {
+ kernel_fpu_begin();
+ sha512_base_do_update(desc, data, chunk, sha512_xform);
+ kernel_fpu_end();
+ }
+
+ data += chunk;
+ } while (len);
return 0;
}
@@ -67,12 +79,25 @@ static int sha512_update(struct shash_desc *desc, const u8 *data,
static int sha512_finup(struct shash_desc *desc, const u8 *data,
unsigned int len, u8 *out, sha512_block_fn *sha512_xform)
{
+ unsigned int chunk;
+
if (!crypto_simd_usable())
return crypto_sha512_finup(desc, data, len, out);
+ do {
+ chunk = min(len, FPU_BYTES);
+ len -= chunk;
+
+ if (chunk) {
+ kernel_fpu_begin();
+ sha512_base_do_update(desc, data, chunk, sha512_xform);
+ kernel_fpu_end();
+ }
+
+ data += chunk;
+ } while (len);
+
kernel_fpu_begin();
- if (len)
- sha512_base_do_update(desc, data, len, sha512_xform);
sha512_base_do_finalize(desc, sha512_xform);
kernel_fpu_end();
--
2.37.3
Powered by blists - more mailing lists