Received-SPF: pass (google.com: domain of linux-crypto-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Robert Elliott <elliott@hpe.com>
To:     herbert@gondor.apana.org.au, davem@davemloft.net,
        tim.c.chen@linux.intel.com, linux-crypto@vger.kernel.org,
        linux-kernel@vger.kernel.org
Cc:     Robert Elliott <elliott@hpe.com>
Subject: [RFC PATCH 4/7] crypto: x86/sm3 - limit FPU preemption
Date:   Thu,  6 Oct 2022 17:31:48 -0500
Message-Id: <20221006223151.22159-5-elliott@hpe.com>
In-Reply-To: <20221006223151.22159-1-elliott@hpe.com>
References: <20221006223151.22159-1-elliott@hpe.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

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, causing:
    rcu: INFO: rcu_preempt detected expedited stalls on CPUs/tasks: {12-... } 22 jiffies s: 277 root: 0x1/.

Fixes: 930ab34d906d ("crypto: x86/sm3 - add AVX assembly implementation")
Suggested-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Robert Elliott <elliott@hpe.com>
---
 arch/x86/crypto/sm3_avx_glue.c | 28 +++++++++++++++++++++++-----
 1 file changed, 23 insertions(+), 5 deletions(-)

diff --git a/arch/x86/crypto/sm3_avx_glue.c b/arch/x86/crypto/sm3_avx_glue.c
index bfd11da956a4..adb9e55e2a16 100644
--- a/arch/x86/crypto/sm3_avx_glue.c
+++ b/arch/x86/crypto/sm3_avx_glue.c
@@ -18,6 +18,8 @@
 #include <asm/cpu_device_id.h>
 #include <asm/simd.h>
 
+#define FPU_BYTES 4096U /* avoid kernel_fpu_begin/end scheduler/rcu stalls */
+
 asmlinkage void sm3_transform_avx(struct sm3_state *state,
 			const u8 *data, int nblocks);
 
@@ -25,6 +27,7 @@ static int sm3_avx_update(struct shash_desc *desc, const u8 *data,
 			 unsigned int len)
 {
 	struct sm3_state *sctx = shash_desc_ctx(desc);
+	unsigned int chunk;
 
 	if (!crypto_simd_usable() ||
 			(sctx->count % SM3_BLOCK_SIZE) + len < SM3_BLOCK_SIZE) {
@@ -38,9 +41,14 @@ static int sm3_avx_update(struct shash_desc *desc, const u8 *data,
 	 */
 	BUILD_BUG_ON(offsetof(struct sm3_state, state) != 0);
 
-	kernel_fpu_begin();
-	sm3_base_do_update(desc, data, len, sm3_transform_avx);
-	kernel_fpu_end();
+	do {
+		chunk = min(len, FPU_BYTES);
+		len -= chunk;
+		kernel_fpu_begin();
+		sm3_base_do_update(desc, data, chunk, sm3_transform_avx);
+		kernel_fpu_end();
+		data += chunk;
+	} while (len);
 
 	return 0;
 }
@@ -48,6 +56,8 @@ static int sm3_avx_update(struct shash_desc *desc, const u8 *data,
 static int sm3_avx_finup(struct shash_desc *desc, const u8 *data,
 		      unsigned int len, u8 *out)
 {
+	unsigned int chunk;
+
 	if (!crypto_simd_usable()) {
 		struct sm3_state *sctx = shash_desc_ctx(desc);
 
@@ -58,9 +68,17 @@ static int sm3_avx_finup(struct shash_desc *desc, const u8 *data,
 		return 0;
 	}
 
+	do {
+		chunk = min(len, FPU_BYTES);
+		len -= chunk;
+		if (chunk) {
+			kernel_fpu_begin();
+			sm3_base_do_update(desc, data, chunk, sm3_transform_avx);
+			kernel_fpu_end();
+		}
+		data += chunk;
+	} while (len);
 	kernel_fpu_begin();
-	if (len)
-		sm3_base_do_update(desc, data, len, sm3_transform_avx);
 	sm3_base_do_finalize(desc, sm3_transform_avx);
 	kernel_fpu_end();
 
-- 
2.37.3