From: neil.armstrong@linaro.org
To: Bartosz Golaszewski <brgl@bgdev.pl>,
Thara Gopinath <thara.gopinath@gmail.com>,
Herbert Xu <herbert@gondor.apana.org.au>,
"David S. Miller" <davem@davemloft.net>,
Stanimir Varbanov <svarbanov@mm-sol.com>
Cc: linux-crypto@vger.kernel.org, linux-arm-msm@vger.kernel.org,
linux-kernel@vger.kernel.org,
Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Subject: Re: [PATCH 9/9] crypto: qce - switch to using a mutex
Date: Tue, 3 Dec 2024 14:53:21 +0100 [thread overview]
Message-ID: <d6220576-eaf5-4415-b25f-b5984255ab78@linaro.org> (raw)
In-Reply-To: <20241203-crypto-qce-refactor-v1-9-c5901d2dd45c@linaro.org>
On 03/12/2024 10:19, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>
> Having switched to workqueue from tasklet, we are no longer limited to
> atomic APIs and can now convert the spinlock to a mutex. This, along
> with the conversion from tasklet to workqueue grants us ~15% improvement
> in cryptsetup benchmarks for AES encryption.
Can you share on which platforms you did the tests and the results you got ?
>
> While at it: use guards to simplify locking code.
>
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> ---
> drivers/crypto/qce/core.c | 46 +++++++++++++++++++++-------------------------
> drivers/crypto/qce/core.h | 3 ++-
> 2 files changed, 23 insertions(+), 26 deletions(-)
>
> diff --git a/drivers/crypto/qce/core.c b/drivers/crypto/qce/core.c
> index 6de9f1e23e282..e95e84486d9ae 100644
> --- a/drivers/crypto/qce/core.c
> +++ b/drivers/crypto/qce/core.c
> @@ -3,6 +3,7 @@
> * Copyright (c) 2010-2014, The Linux Foundation. All rights reserved.
> */
>
> +#include <linux/cleanup.h>
> #include <linux/clk.h>
> #include <linux/device.h>
> #include <linux/dma-mapping.h>
> @@ -11,7 +12,6 @@
> #include <linux/module.h>
> #include <linux/mod_devicetable.h>
> #include <linux/platform_device.h>
> -#include <linux/spinlock.h>
> #include <linux/types.h>
> #include <crypto/algapi.h>
> #include <crypto/internal/hash.h>
> @@ -89,34 +89,28 @@ static int qce_handle_queue(struct qce_device *qce,
> struct crypto_async_request *req)
> {
> struct crypto_async_request *async_req, *backlog;
> - unsigned long flags;
> int ret = 0, err;
>
> - spin_lock_irqsave(&qce->lock, flags);
> + scoped_guard(mutex, &qce->lock) {
> + if (req)
> + ret = crypto_enqueue_request(&qce->queue, req);
>
> - if (req)
> - ret = crypto_enqueue_request(&qce->queue, req);
> + /* busy, do not dequeue request */
> + if (qce->req)
> + return ret;
>
> - /* busy, do not dequeue request */
> - if (qce->req) {
> - spin_unlock_irqrestore(&qce->lock, flags);
> - return ret;
> + backlog = crypto_get_backlog(&qce->queue);
> + async_req = crypto_dequeue_request(&qce->queue);
> + if (async_req)
> + qce->req = async_req;
> }
>
> - backlog = crypto_get_backlog(&qce->queue);
> - async_req = crypto_dequeue_request(&qce->queue);
> - if (async_req)
> - qce->req = async_req;
> -
> - spin_unlock_irqrestore(&qce->lock, flags);
> -
> if (!async_req)
> return ret;
>
> if (backlog) {
> - spin_lock_bh(&qce->lock);
> - crypto_request_complete(backlog, -EINPROGRESS);
> - spin_unlock_bh(&qce->lock);
> + scoped_guard(mutex, &qce->lock)
> + crypto_request_complete(backlog, -EINPROGRESS);
> }
>
> err = qce_handle_request(async_req);
> @@ -133,12 +127,11 @@ static void qce_req_done_work(struct work_struct *work)
> struct qce_device *qce = container_of(work, struct qce_device,
> done_work);
> struct crypto_async_request *req;
> - unsigned long flags;
>
> - spin_lock_irqsave(&qce->lock, flags);
> - req = qce->req;
> - qce->req = NULL;
> - spin_unlock_irqrestore(&qce->lock, flags);
> + scoped_guard(mutex, &qce->lock) {
> + req = qce->req;
> + qce->req = NULL;
> + }
>
> if (req)
> crypto_request_complete(req, qce->result);
> @@ -243,7 +236,10 @@ static int qce_crypto_probe(struct platform_device *pdev)
> if (ret)
> return ret;
>
> - spin_lock_init(&qce->lock);
> + ret = devm_mutex_init(qce->dev, &qce->lock);
> + if (ret)
> + return ret;
> +
> INIT_WORK(&qce->done_work, qce_req_done_work);
> crypto_init_queue(&qce->queue, QCE_QUEUE_LENGTH);
>
> diff --git a/drivers/crypto/qce/core.h b/drivers/crypto/qce/core.h
> index 39e75a75a4293..eb6fa7a8b64a8 100644
> --- a/drivers/crypto/qce/core.h
> +++ b/drivers/crypto/qce/core.h
> @@ -6,6 +6,7 @@
> #ifndef _CORE_H_
> #define _CORE_H_
>
> +#include <linux/mutex.h>
> #include <linux/workqueue.h>
>
> #include "dma.h"
> @@ -30,7 +31,7 @@
> */
> struct qce_device {
> struct crypto_queue queue;
> - spinlock_t lock;
> + struct mutex lock;
> struct work_struct done_work;
> struct crypto_async_request *req;
> int result;
>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
next prev parent reply other threads:[~2024-12-03 13:53 UTC|newest]
Thread overview: 35+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-12-03 9:19 [PATCH 0/9] crypto: qce - refactor the driver Bartosz Golaszewski
2024-12-03 9:19 ` [PATCH 1/9] crypto: qce - fix goto jump in error path Bartosz Golaszewski
2024-12-03 13:43 ` neil.armstrong
2024-12-03 9:19 ` [PATCH 2/9] crypto: qce - unregister previously registered algos " Bartosz Golaszewski
2024-12-03 13:55 ` neil.armstrong
2024-12-03 15:05 ` Bartosz Golaszewski
2024-12-04 10:17 ` neil.armstrong
2024-12-03 9:19 ` [PATCH 3/9] crypto: qce - remove unneeded call to icc_set_bw() " Bartosz Golaszewski
2024-12-03 13:45 ` neil.armstrong
2024-12-03 9:19 ` [PATCH 4/9] crypto: qce - shrink code with devres clk helpers Bartosz Golaszewski
2024-12-03 13:46 ` neil.armstrong
2024-12-03 18:32 ` Stephan Gerhold
2024-12-03 20:39 ` Bartosz Golaszewski
2024-12-03 9:19 ` [PATCH 5/9] crypto: qce - convert qce_dma_request() to use devres Bartosz Golaszewski
2024-12-03 13:49 ` neil.armstrong
2024-12-03 9:19 ` [PATCH 6/9] crypto: qce - make qce_register_algs() a managed interface Bartosz Golaszewski
2024-12-03 13:50 ` neil.armstrong
2024-12-03 9:19 ` [PATCH 7/9] crypto: qce - use __free() for a buffer that's always freed Bartosz Golaszewski
2024-12-03 13:52 ` neil.armstrong
2024-12-03 9:19 ` [PATCH 8/9] crypto: qce - convert tasklet to workqueue Bartosz Golaszewski
2024-12-03 13:52 ` neil.armstrong
2024-12-03 9:19 ` [PATCH 9/9] crypto: qce - switch to using a mutex Bartosz Golaszewski
2024-12-03 13:53 ` neil.armstrong [this message]
2024-12-03 15:10 ` Bartosz Golaszewski
2024-12-04 10:17 ` neil.armstrong
2025-01-18 8:06 ` Eric Biggers
2025-01-18 9:28 ` Bartosz Golaszewski
2025-01-18 17:55 ` Eric Biggers
2025-01-20 13:46 ` Bartosz Golaszewski
2025-02-20 9:14 ` Bartosz Golaszewski
2025-02-20 19:19 ` Eric Biggers
2024-12-03 17:35 ` [PATCH 0/9] crypto: qce - refactor the driver Eric Biggers
2024-12-03 18:08 ` Eric Biggers
2024-12-03 20:55 ` Bartosz Golaszewski
2024-12-14 9:27 ` Herbert Xu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=d6220576-eaf5-4415-b25f-b5984255ab78@linaro.org \
--to=neil.armstrong@linaro.org \
--cc=bartosz.golaszewski@linaro.org \
--cc=brgl@bgdev.pl \
--cc=davem@davemloft.net \
--cc=herbert@gondor.apana.org.au \
--cc=linux-arm-msm@vger.kernel.org \
--cc=linux-crypto@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=svarbanov@mm-sol.com \
--cc=thara.gopinath@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®