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]
Message-Id: <20240403-uvc_request_length_by_interval-v4-5-ca22f334226e@pengutronix.de>
Date: Tue, 13 Aug 2024 11:09:29 +0200
From: Michael Grzeschik <m.grzeschik@...gutronix.de>
To: Laurent Pinchart <laurent.pinchart@...asonboard.com>, 
 Daniel Scally <dan.scally@...asonboard.com>, 
 Greg Kroah-Hartman <gregkh@...uxfoundation.org>, 
 Avichal Rakesh <arakesh@...gle.com>
Cc: linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org, 
 Michael Grzeschik <m.grzeschik@...gutronix.de>
Subject: [PATCH v4 05/10] usb: gadget: uvc: remove
 uvc_video_ep_queue_initial_requests

By moving the buffer enqueing to an extra worker and not
enqueueing the each request by its corresponding complete
interrupt, we can remove the initial amount of zero length requests.

As soon as real data is available the isoc queue will be filled
as much as possible.

Signed-off-by: Michael Grzeschik <m.grzeschik@...gutronix.de>

---
v3 -> v4: -
v1 -> v3: new patch
---
 drivers/usb/gadget/function/uvc_video.c | 46 ---------------------------------
 1 file changed, 46 deletions(-)

diff --git a/drivers/usb/gadget/function/uvc_video.c b/drivers/usb/gadget/function/uvc_video.c
index a51e95e3b717c..259920ae36843 100644
--- a/drivers/usb/gadget/function/uvc_video.c
+++ b/drivers/usb/gadget/function/uvc_video.c
@@ -325,50 +325,6 @@ static int uvcg_video_usb_req_queue(struct uvc_video *video,
 	return 0;
 }
 
-/*
- * Must only be called from uvcg_video_enable - since after that we only want to
- * queue requests to the endpoint from the uvc_video_complete complete handler.
- * This function is needed in order to 'kick start' the flow of requests from
- * gadget driver to the usb controller.
- */
-static void uvc_video_ep_queue_initial_requests(struct uvc_video *video)
-{
-	struct usb_request *req = NULL;
-	unsigned long flags = 0;
-	unsigned int count = 0;
-	int ret = 0;
-
-	/*
-	 * We only queue half of the free list since we still want to have
-	 * some free usb_requests in the free list for the video_pump async_wq
-	 * thread to encode uvc buffers into. Otherwise we could get into a
-	 * situation where the free list does not have any usb requests to
-	 * encode into - we always end up queueing 0 length requests to the
-	 * end point.
-	 */
-	unsigned int half_list_size = video->uvc_num_requests / 2;
-
-	spin_lock_irqsave(&video->req_lock, flags);
-	/*
-	 * Take these requests off the free list and queue them all to the
-	 * endpoint. Since we queue 0 length requests with the req_lock held,
-	 * there isn't any 'data' race involved here with the complete handler.
-	 */
-	while (count < half_list_size) {
-		req = list_first_entry(&video->req_free, struct usb_request,
-					list);
-		list_del(&req->list);
-		req->length = 0;
-		ret = uvcg_video_ep_queue(video, req);
-		if (ret < 0) {
-			uvcg_queue_cancel(&video->queue, 0);
-			break;
-		}
-		count++;
-	}
-	spin_unlock_irqrestore(&video->req_lock, flags);
-}
-
 static void
 uvc_video_complete(struct usb_ep *ep, struct usb_request *req)
 {
@@ -760,8 +716,6 @@ int uvcg_video_enable(struct uvc_video *video)
 
 	video->req_int_count = 0;
 
-	uvc_video_ep_queue_initial_requests(video);
-
 	return ret;
 }
 

-- 
2.39.2


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ