linux-kernel - [PATCH 1/2] rpmsg: glink: change rx work queue type

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 7 Jun 2023 17:47:30 +0530
From:   Manikanta Mylavarapu <quic_mmanikan@...cinc.com>
To:     <agross@...nel.org>, <andersson@...nel.org>,
        <konrad.dybcio@...aro.org>, <mathieu.poirier@...aro.org>,
        <linux-arm-msm@...r.kernel.org>,
        <linux-remoteproc@...r.kernel.org>, <linux-kernel@...r.kernel.org>
CC:     <quic_srichara@...cinc.com>, <quic_sjaganat@...cinc.com>,
        <quic_kathirav@...cinc.com>, <quic_anusha@...cinc.com>,
        <quic_varada@...cinc.com>
Subject: [PATCH 1/2] rpmsg: glink: change rx work queue type

In case of heavy cpu load, rx worker thread
not able to get cpu slot because it's bound
to kernel global work queue.

Glink client drivers will return timeout
error if they didn't receive response
within stipulated duration. Changing rx
work queue type to UNBOUND workqueue
ensures rx worker thread will be executed
as early as possible.

Signed-off-by: Manikanta Mylavarapu <quic_mmanikan@...cinc.com>
---
 drivers/rpmsg/qcom_glink_native.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/rpmsg/qcom_glink_native.c b/drivers/rpmsg/qcom_glink_native.c
index 1beb40a1d3df..6f9a439e5046 100644
--- a/drivers/rpmsg/qcom_glink_native.c
+++ b/drivers/rpmsg/qcom_glink_native.c
@@ -81,6 +81,7 @@ struct glink_core_rx_intent {
  * @rx_pipe:	pipe object for receive FIFO
  * @tx_pipe:	pipe object for transmit FIFO
  * @rx_work:	worker for handling received control messages
+ * @rx_wq:	work queue of rx_work
  * @rx_lock:	protects the @rx_queue
  * @rx_queue:	queue of received control messages to be processed in @rx_work
  * @tx_lock:	synchronizes operations on the tx fifo
@@ -100,6 +101,7 @@ struct qcom_glink {
 	struct qcom_glink_pipe *tx_pipe;
 
 	struct work_struct rx_work;
+	struct workqueue_struct *rx_wq;
 	spinlock_t rx_lock;
 	struct list_head rx_queue;
 
@@ -822,7 +824,7 @@ static int qcom_glink_rx_defer(struct qcom_glink *glink, size_t extra)
 	list_add_tail(&dcmd->node, &glink->rx_queue);
 	spin_unlock(&glink->rx_lock);
 
-	schedule_work(&glink->rx_work);
+	queue_work(glink->rx_wq, &glink->rx_work);
 	qcom_glink_rx_advance(glink, sizeof(dcmd->msg) + extra);
 
 	return 0;
@@ -1665,6 +1667,8 @@ static void qcom_glink_cancel_rx_work(struct qcom_glink *glink)
 	/* cancel any pending deferred rx_work */
 	cancel_work_sync(&glink->rx_work);
 
+	destroy_workqueue(glink->rx_wq);
+
 	list_for_each_entry_safe(dcmd, tmp, &glink->rx_queue, node)
 		kfree(dcmd);
 }
@@ -1750,6 +1754,10 @@ struct qcom_glink *qcom_glink_native_probe(struct device *dev,
 	INIT_WORK(&glink->rx_work, qcom_glink_work);
 	init_waitqueue_head(&glink->tx_avail_notify);
 
+	glink->rx_wq = alloc_workqueue("glink_rx_wq", WQ_UNBOUND, 1);
+	if (!glink->rx_wq)
+		return ERR_PTR(-ENOMEM);
+
 	spin_lock_init(&glink->idr_lock);
 	idr_init(&glink->lcids);
 	idr_init(&glink->rcids);
-- 
2.17.1