[PATCH AUTOSEL 6.6 06/36] crypto: pcrypt - Fix hungtask for PADATA_RESET

[PATCH AUTOSEL 6.6 06/36] crypto: pcrypt - Fix hungtask for PADATA_RESET

[Sasha Levin]

From: Lu Jialin <lujialin4@huawei.com>

[ Upstream commit 8f4f68e788c3a7a696546291258bfa5fdb215523 ]

We found a hungtask bug in test_aead_vec_cfg as follows:

INFO: task cryptomgr_test:391009 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Call trace:
 __switch_to+0x98/0xe0
 __schedule+0x6c4/0xf40
 schedule+0xd8/0x1b4
 schedule_timeout+0x474/0x560
 wait_for_common+0x368/0x4e0
 wait_for_completion+0x20/0x30
 wait_for_completion+0x20/0x30
 test_aead_vec_cfg+0xab4/0xd50
 test_aead+0x144/0x1f0
 alg_test_aead+0xd8/0x1e0
 alg_test+0x634/0x890
 cryptomgr_test+0x40/0x70
 kthread+0x1e0/0x220
 ret_from_fork+0x10/0x18
 Kernel panic - not syncing: hung_task: blocked tasks

For padata_do_parallel, when the return err is 0 or -EBUSY, it will call
wait_for_completion(&wait->completion) in test_aead_vec_cfg. In normal
case, aead_request_complete() will be called in pcrypt_aead_serial and the
return err is 0 for padata_do_parallel. But, when pinst->flags is
PADATA_RESET, the return err is -EBUSY for padata_do_parallel, and it
won't call aead_request_complete(). Therefore, test_aead_vec_cfg will
hung at wait_for_completion(&wait->completion), which will cause
hungtask.

The problem comes as following:
(padata_do_parallel)                 |
    rcu_read_lock_bh();              |
    err = -EINVAL;                   |   (padata_replace)
                                     |     pinst->flags |= PADATA_RESET;
    err = -EBUSY                     |
    if (pinst->flags & PADATA_RESET) |
        rcu_read_unlock_bh()         |
        return err

In order to resolve the problem, we replace the return err -EBUSY with
-EAGAIN, which means parallel_data is changing, and the caller should call
it again.

v3:
remove retry and just change the return err.
v2:
introduce padata_try_do_parallel() in pcrypt_aead_encrypt and
pcrypt_aead_decrypt to solve the hungtask.

Signed-off-by: Lu Jialin <lujialin4@huawei.com>
Signed-off-by: Guo Zihua <guozihua@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 crypto/pcrypt.c | 4 ++++
 kernel/padata.c | 2 +-
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/crypto/pcrypt.c b/crypto/pcrypt.c
index 8c1d0ca412137..d0d954fe9d54f 100644
--- a/crypto/pcrypt.c
+++ b/crypto/pcrypt.c
@@ -117,6 +117,8 @@ static int pcrypt_aead_encrypt(struct aead_request *req)
 	err = padata_do_parallel(ictx->psenc, padata, &ctx->cb_cpu);
 	if (!err)
 		return -EINPROGRESS;
+	if (err == -EBUSY)
+		return -EAGAIN;
 
 	return err;
 }
@@ -164,6 +166,8 @@ static int pcrypt_aead_decrypt(struct aead_request *req)
 	err = padata_do_parallel(ictx->psdec, padata, &ctx->cb_cpu);
 	if (!err)
 		return -EINPROGRESS;
+	if (err == -EBUSY)
+		return -EAGAIN;
 
 	return err;
 }
diff --git a/kernel/padata.c b/kernel/padata.c
index 222d60195de66..81c8183f3176a 100644
--- a/kernel/padata.c
+++ b/kernel/padata.c
@@ -202,7 +202,7 @@ int padata_do_parallel(struct padata_shell *ps,
 		*cb_cpu = cpu;
 	}
 
-	err =  -EBUSY;
+	err = -EBUSY;
 	if ((pinst->flags & PADATA_RESET))
 		goto out;
 
-- 
2.42.0

[PATCH AUTOSEL 6.6 33/36] crypto: hisilicon/qm - prevent soft lockup in receive loop

[Sasha Levin]

From: Longfang Liu <liulongfang@huawei.com>

[ Upstream commit 33fc506d2ac514be1072499a263c3bff8c7c95a0 ]

In the scenario where the accelerator business is fully loaded.
When the workqueue receiving messages and performing callback
processing, there are a large number of messages that need to be
received, and there are continuously messages that have been
processed and need to be received.
This will cause the receive loop here to be locked for a long time.
This scenario will cause watchdog timeout problems on OS with kernel
preemption turned off.

The error logs:
watchdog: BUG: soft lockup - CPU#23 stuck for 23s! [kworker/u262:1:1407]
[ 1461.978428][   C23] Call trace:
[ 1461.981890][   C23]  complete+0x8c/0xf0
[ 1461.986031][   C23]  kcryptd_async_done+0x154/0x1f4 [dm_crypt]
[ 1461.992154][   C23]  sec_skcipher_callback+0x7c/0xf4 [hisi_sec2]
[ 1461.998446][   C23]  sec_req_cb+0x104/0x1f4 [hisi_sec2]
[ 1462.003950][   C23]  qm_poll_req_cb+0xcc/0x150 [hisi_qm]
[ 1462.009531][   C23]  qm_work_process+0x60/0xc0 [hisi_qm]
[ 1462.015101][   C23]  process_one_work+0x1c4/0x470
[ 1462.020052][   C23]  worker_thread+0x150/0x3c4
[ 1462.024735][   C23]  kthread+0x108/0x13c
[ 1462.028889][   C23]  ret_from_fork+0x10/0x18

Therefore, it is necessary to add an actively scheduled operation in the
while loop to prevent this problem.
After adding it, no matter whether the OS turns on or off the kernel
preemption function. Neither will cause watchdog timeout issues.

Signed-off-by: Longfang Liu <liulongfang@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/crypto/hisilicon/qm.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/crypto/hisilicon/qm.c b/drivers/crypto/hisilicon/qm.c
index a99fd589445ce..a830a2e0f4400 100644
--- a/drivers/crypto/hisilicon/qm.c
+++ b/drivers/crypto/hisilicon/qm.c
@@ -849,6 +849,8 @@ static void qm_poll_req_cb(struct hisi_qp *qp)
 		qm_db(qm, qp->qp_id, QM_DOORBELL_CMD_CQ,
 		      qp->qp_status.cq_head, 0);
 		atomic_dec(&qp->qp_status.used);
+
+		cond_resched();
 	}
 
 	/* set c_flag */
-- 
2.42.0