Skip to contentSkip to conversation

FAILED: patch "[PATCH] media: uvcvideo: Fix race condition with usb_kill_urb" failed to apply to 5.4-stable tree

PATCH New [email protected] 2 participants 3 messages Mar 6, 2023 – Mar 8, 2023

Subsystems (3)
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

The patch below does not apply to the 5.4-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.4.y
git checkout FETCH_HEAD
git cherry-pick -x 619d9b710cf06f7a00a17120ca92333684ac45a8
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<[email protected]>' --in-reply-to '[email protected]' --subject-prefix 'PATCH 5.4.y' HEAD^..

Possible dependencies:

619d9b710cf0 ("media: uvcvideo: Fix race condition with usb_kill_urb") 40140eda661e ("media: uvcvideo: Implement mask for V4L2_CTRL_TYPE_MENU") adfd3910c27f ("media: uvcvideo: Remove void casting for the status endpoint") 382075604a68 ("media: uvcvideo: Limit power line control for Quanta UVC Webcam") 710871163510 ("media: uvcvideo: Add missing value for power_line_frequency") 70fa906d6fce ("media: uvcvideo: Use control names from framework") 9b31ea808a44 ("media: uvcvideo: Add support for V4L2_CTRL_TYPE_CTRL_CLASS") 866c6bdd5663 ("media: uvcvideo: refactor __uvc_ctrl_add_mapping") 9e56380ae625 ("media: uvcvideo: Rename debug functions")
ed4c5fa4d804 ("media: uvcvideo: use dev_printk() for uvc_trace()") 59e92bf62771 ("media: uvcvideo: New macro uvc_trace_cont") 69df09547e7a ("media: uvcvideo: Use dev_ printk aliases")
6f6a87eb8266 ("media: uvcvideo: Add Privacy control based on EXT_GPIO") 2886477ff987 ("media: uvcvideo: Implement UVC_EXT_GPIO_UNIT") d9c8763e6129 ("media: uvcvideo: Provide sync and async uvc_ctrl_status_event") 351509c604dc ("media: uvcvideo: Move guid to entity")
dc9455ffae02 ("media: uvcvideo: Accept invalid bFormatIndex and bFrameIndex values") b400b6f28af0 ("media: uvcvideo: Force UVC version to 1.0a for 1bcf:0b40") 8a652a17e3c0 ("media: uvcvideo: Ensure all probed info is returned to v4l2") f875bcc375c7 ("media: uvcvideo: Fix dereference of out-of-bound list iterator")

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From 619d9b710cf06f7a00a17120ca92333684ac45a8 Mon Sep 17 00:00:00 2001
From: Ricardo Ribalda <[email protected]>
Date: Thu, 5 Jan 2023 15:31:29 +0100
Subject: [PATCH] media: uvcvideo: Fix race condition with usb_kill_urb

usb_kill_urb warranties that all the handlers are finished when it returns, but does not protect against threads that might be handling asynchronously the urb.

For UVC, the function uvc_ctrl_status_event_async() takes care of control changes asynchronously.

If the code is executed in the following order:

CPU 0 CPU 1
===== =====
uvc_status_complete()
uvc_status_stop()
uvc_ctrl_status_event_work()

					uvc_status_start() -> FAIL

Then uvc_status_start will keep failing and this error will be shown:

<4>[    5.540139] URB 0000000000000000 submitted while active

drivers/usb/core/urb.c:378 usb_submit_urb+0x4c3/0x528

Let's improve the current situation, by not re-submiting the urb if we are stopping the status event. Also process the queued work (if any) during stop.

CPU 0 CPU 1
===== =====
uvc_status_complete()
uvc_status_stop() uvc_status_start()
uvc_ctrl_status_event_work() -> FAIL

Hopefully, with the usb layer protection this should be enough to cover all the cases.

Fixes e5225c820c05 ("media: uvcvideo: Send a control event when a Control Change interrupt arrives")
drivers/media/usb/uvc/uvc_ctrl.c +5 -0
index ee58f0db2763..c3aeba3fe31b 100644
@@ -6,6 +6,7 @@
  *          Laurent Pinchart ([email protected])
  */
 
+#include <asm/barrier.h>
 #include <linux/bitops.h>
 #include <linux/kernel.h>
 #include <linux/list.h>
@@ -1571,6 +1572,10 @@ static void uvc_ctrl_status_event_work(struct work_struct *work)
 
 	uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
 
+	/* The barrier is needed to synchronize with uvc_status_stop(). */
+	if (smp_load_acquire(&dev->flush_status))
+		return;
+
 	/* Resubmit the URB. */
 	w->urb->interval = dev->int_ep->desc.bInterval;
 	ret = usb_submit_urb(w->urb, GFP_KERNEL);
drivers/media/usb/uvc/uvc_status.c +37 -0
index 602830a8023e..a78a88c710e2 100644
@@ -6,6 +6,7 @@
  *          Laurent Pinchart ([email protected])
  */
 
+#include <asm/barrier.h>
 #include <linux/kernel.h>
 #include <linux/input.h>
 #include <linux/slab.h>
@@ -311,5 +312,41 @@ int uvc_status_start(struct uvc_device *dev, gfp_t flags)
 
 void uvc_status_stop(struct uvc_device *dev)
 {
+	struct uvc_ctrl_work *w = &dev->async_ctrl;
+
+	/*
+	 * Prevent the asynchronous control handler from requeing the URB. The
+	 * barrier is needed so the flush_status change is visible to other
+	 * CPUs running the asynchronous handler before usb_kill_urb() is
+	 * called below.
+	 */
+	smp_store_release(&dev->flush_status, true);
+
+	/*
+	 * Cancel any pending asynchronous work. If any status event was queued,
+	 * process it synchronously.
+	 */
+	if (cancel_work_sync(&w->work))
+		uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
+
+	/* Kill the urb. */
 	usb_kill_urb(dev->int_urb);
+
+	/*
+	 * The URB completion handler may have queued asynchronous work. This
+	 * won't resubmit the URB as flush_status is set, but it needs to be
+	 * cancelled before returning or it could then race with a future
+	 * uvc_status_start() call.
+	 */
+	if (cancel_work_sync(&w->work))
+		uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
+
+	/*
+	 * From this point, there are no events on the queue and the status URB
+	 * is dead. No events will be queued until uvc_status_start() is called.
+	 * The barrier is needed to make sure that flush_status is visible to
+	 * uvc_ctrl_status_event_work() when uvc_status_start() will be called
+	 * again.
+	 */
+	smp_store_release(&dev->flush_status, false);
 }
drivers/media/usb/uvc/uvcvideo.h +1 -0
index e85df8deb965..c0e706fcd2cb 100644
@@ -577,6 +577,7 @@ struct uvc_device {
 	struct usb_host_endpoint *int_ep;
 	struct urb *int_urb;
 	struct uvc_status *status;
+	bool flush_status;
 
 	struct input_dev *input;
 	char input_phys[64];
#1 lore raw

PATCH 1/2 media: uvcvideo: Provide sync and async uvc_ctrl_status_event

Split the functionality of void uvc_ctrl_status_event_work in two, so it can be called by functions outside interrupt context and not part of an URB.

---
 drivers/media/usb/uvc/uvc_ctrl.c   | 25 +++++++++++++++----------
 drivers/media/usb/uvc/uvc_status.c |  3 ++-
 drivers/media/usb/uvc/uvcvideo.h   |  4 +++-
 3 files changed, 20 insertions(+), 12 deletions(-)
drivers/media/usb/uvc/uvc_ctrl.c +15 -10
index 36abe47997b0..b6cadae3c187 100644
@@ -1275,17 +1275,12 @@ static void uvc_ctrl_send_slave_event(struct uvc_video_chain *chain,
 	uvc_ctrl_send_event(chain, handle, ctrl, mapping, val, changes);
 }
 
-static void uvc_ctrl_status_event_work(struct work_struct *work)
+void uvc_ctrl_status_event(struct uvc_video_chain *chain,
+			   struct uvc_control *ctrl, const u8 *data)
 {
-	struct uvc_device *dev = container_of(work, struct uvc_device,
-					      async_ctrl.work);
-	struct uvc_ctrl_work *w = &dev->async_ctrl;
-	struct uvc_video_chain *chain = w->chain;
 	struct uvc_control_mapping *mapping;
-	struct uvc_control *ctrl = w->ctrl;
 	struct uvc_fh *handle;
 	unsigned int i;
-	int ret;
 
 	mutex_lock(&chain->ctrl_mutex);
 
@@ -1293,7 +1288,7 @@ static void uvc_ctrl_status_event_work(struct work_struct *work)
 	ctrl->handle = NULL;
 
 	list_for_each_entry(mapping, &ctrl->info.mappings, list) {
-		s32 value = __uvc_ctrl_get_value(mapping, w->data);
+		s32 value = __uvc_ctrl_get_value(mapping, data);
 
 		/*
 		 * handle may be NULL here if the device sends auto-update
@@ -1312,6 +1307,16 @@ static void uvc_ctrl_status_event_work(struct work_struct *work)
 	}
 
 	mutex_unlock(&chain->ctrl_mutex);
+}
+
+static void uvc_ctrl_status_event_work(struct work_struct *work)
+{
+	struct uvc_device *dev = container_of(work, struct uvc_device,
+					      async_ctrl.work);
+	struct uvc_ctrl_work *w = &dev->async_ctrl;
+	int ret;
+
+	uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
 
 	/* Resubmit the URB. */
 	w->urb->interval = dev->int_ep->desc.bInterval;
@@ -1321,8 +1326,8 @@ static void uvc_ctrl_status_event_work(struct work_struct *work)
 			   ret);
 }
 
-bool uvc_ctrl_status_event(struct urb *urb, struct uvc_video_chain *chain,
-			   struct uvc_control *ctrl, const u8 *data)
+bool uvc_ctrl_status_event_async(struct urb *urb, struct uvc_video_chain *chain,
+				 struct uvc_control *ctrl, const u8 *data)
 {
 	struct uvc_device *dev = chain->dev;
 	struct uvc_ctrl_work *w = &dev->async_ctrl;
drivers/media/usb/uvc/uvc_status.c +2 -1
index 2bdb0ff203f8..3e26d82a906d 100644
@@ -179,7 +179,8 @@ static bool uvc_event_control(struct urb *urb,
 
 	switch (status->bAttribute) {
 	case UVC_CTRL_VALUE_CHANGE:
-		return uvc_ctrl_status_event(urb, chain, ctrl, status->bValue);
+		return uvc_ctrl_status_event_async(urb, chain, ctrl,
+						   status->bValue);
 
 	case UVC_CTRL_INFO_CHANGE:
 	case UVC_CTRL_FAILURE_CHANGE:
drivers/media/usb/uvc/uvcvideo.h +3 -1
index 5f137400bebd..fce41609e27a 100644
@@ -832,7 +832,9 @@ int uvc_ctrl_add_mapping(struct uvc_video_chain *chain,
 int uvc_ctrl_init_device(struct uvc_device *dev);
 void uvc_ctrl_cleanup_device(struct uvc_device *dev);
 int uvc_ctrl_restore_values(struct uvc_device *dev);
-bool uvc_ctrl_status_event(struct urb *urb, struct uvc_video_chain *chain,
+bool uvc_ctrl_status_event_async(struct urb *urb, struct uvc_video_chain *chain,
+				 struct uvc_control *ctrl, const u8 *data);
+void uvc_ctrl_status_event(struct uvc_video_chain *chain,
 			   struct uvc_control *ctrl, const u8 *data);
 
 int uvc_ctrl_begin(struct uvc_video_chain *chain);
Signature · 1 line
2.40.0.rc0.216.gc4246ad0f0-goog
#2 lore raw

PATCH 2/2 media: uvcvideo: Fix race condition with usb_kill_urb

usb_kill_urb warranties that all the handlers are finished when it returns, but does not protect against threads that might be handling asynchronously the urb.

For UVC, the function uvc_ctrl_status_event_async() takes care of control changes asynchronously.

If the code is executed in the following order:

CPU 0 CPU 1
===== =====
uvc_status_complete()
uvc_status_stop()
uvc_ctrl_status_event_work()

					uvc_status_start() -> FAIL

Then uvc_status_start will keep failing and this error will be shown:

<4>[    5.540139] URB 0000000000000000 submitted while active

drivers/usb/core/urb.c:378 usb_submit_urb+0x4c3/0x528

Let's improve the current situation, by not re-submiting the urb if we are stopping the status event. Also process the queued work (if any) during stop.

CPU 0 CPU 1
===== =====
uvc_status_complete()
uvc_status_stop() uvc_status_start()
uvc_ctrl_status_event_work() -> FAIL

Hopefully, with the usb layer protection this should be enough to cover all the cases.

Fixes e5225c820c05 ("media: uvcvideo: Send a control event when a Control Change interrupt arrives")
(cherry picked from commit 619d9b710cf06f7a00a17120ca92333684ac45a8)
---
 drivers/media/usb/uvc/uvc_ctrl.c   |  5 ++++
 drivers/media/usb/uvc/uvc_status.c | 37 ++++++++++++++++++++++++++++++
 drivers/media/usb/uvc/uvcvideo.h   |  1 +
 3 files changed, 43 insertions(+)
drivers/media/usb/uvc/uvc_ctrl.c +5 -0
index b6cadae3c187..e0d894c61f4b 100644
@@ -6,6 +6,7 @@
  *          Laurent Pinchart ([email protected])
  */
 
+#include <asm/barrier.h>
 #include <linux/kernel.h>
 #include <linux/list.h>
 #include <linux/module.h>
@@ -1318,6 +1319,10 @@ static void uvc_ctrl_status_event_work(struct work_struct *work)
 
 	uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
 
+	/* The barrier is needed to synchronize with uvc_status_stop(). */
+	if (smp_load_acquire(&dev->flush_status))
+		return;
+
 	/* Resubmit the URB. */
 	w->urb->interval = dev->int_ep->desc.bInterval;
 	ret = usb_submit_urb(w->urb, GFP_KERNEL);
drivers/media/usb/uvc/uvc_status.c +37 -0
index 3e26d82a906d..73725051cc16 100644
@@ -6,6 +6,7 @@
  *          Laurent Pinchart ([email protected])
  */
 
+#include <asm/barrier.h>
 #include <linux/kernel.h>
 #include <linux/input.h>
 #include <linux/slab.h>
@@ -310,5 +311,41 @@ int uvc_status_start(struct uvc_device *dev, gfp_t flags)
 
 void uvc_status_stop(struct uvc_device *dev)
 {
+	struct uvc_ctrl_work *w = &dev->async_ctrl;
+
+	/*
+	 * Prevent the asynchronous control handler from requeing the URB. The
+	 * barrier is needed so the flush_status change is visible to other
+	 * CPUs running the asynchronous handler before usb_kill_urb() is
+	 * called below.
+	 */
+	smp_store_release(&dev->flush_status, true);
+
+	/*
+	 * Cancel any pending asynchronous work. If any status event was queued,
+	 * process it synchronously.
+	 */
+	if (cancel_work_sync(&w->work))
+		uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
+
+	/* Kill the urb. */
 	usb_kill_urb(dev->int_urb);
+
+	/*
+	 * The URB completion handler may have queued asynchronous work. This
+	 * won't resubmit the URB as flush_status is set, but it needs to be
+	 * cancelled before returning or it could then race with a future
+	 * uvc_status_start() call.
+	 */
+	if (cancel_work_sync(&w->work))
+		uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
+
+	/*
+	 * From this point, there are no events on the queue and the status URB
+	 * is dead. No events will be queued until uvc_status_start() is called.
+	 * The barrier is needed to make sure that flush_status is visible to
+	 * uvc_ctrl_status_event_work() when uvc_status_start() will be called
+	 * again.
+	 */
+	smp_store_release(&dev->flush_status, false);
 }
drivers/media/usb/uvc/uvcvideo.h +1 -0
index fce41609e27a..d9e954335e75 100644
@@ -664,6 +664,7 @@ struct uvc_device {
 	struct usb_host_endpoint *int_ep;
 	struct urb *int_urb;
 	u8 *status;
+	bool flush_status;
 	struct input_dev *input;
 	char input_phys[64];
 
Signature · 1 line
2.40.0.rc0.216.gc4246ad0f0-goog