usb-storage: fix freezing of the scanning thread

This patch (as1521b) fixes the interaction between usb-storage's
scanning thread and the freezer.  The current implementation has a
race: If the device is unplugged shortly after being plugged in and
just as a system sleep begins, the scanning thread may get frozen
before the khubd task.  Khubd won't be able to freeze until the
disconnect processing is complete, and the disconnect processing can't
proceed until the scanning thread finishes, so the sleep transition
will fail.

The implementation in the 3.2 kernel suffers from an additional
problem.  There the scanning thread calls set_freezable_with_signal(),
and the signals sent by the freezer will mess up the thread's I/O
delays, which are all interruptible.

The solution to both problems is the same: Replace the kernel thread
used for scanning with a delayed-work routine on the system freezable
work queue.  Freezable work queues have the nice property that you can
cancel a work item even while the work queue is frozen, and no signals
are needed.

The 3.2 version of this patch solves the problem in Bugzilla #42730.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
CC: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

2 files changed, 35 insertions, 62 deletions

diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c
index 3dd7da9fd504..db51ba16dc07 100644
--- a/drivers/usb/storage/usb.c
+++ b/drivers/usb/storage/usb.c
@@ -788,15 +788,19 @@ static void quiesce_and_remove_host(struct us_data *us)
 	struct Scsi_Host *host = us_to_host(us);
 
 	/* If the device is really gone, cut short reset delays */
-	if (us->pusb_dev->state == USB_STATE_NOTATTACHED)
+	if (us->pusb_dev->state == USB_STATE_NOTATTACHED) {
 		set_bit(US_FLIDX_DISCONNECTING, &us->dflags);
+		wake_up(&us->delay_wait);
+	}
 
-	/* Prevent SCSI-scanning (if it hasn't started yet)
-	 * and wait for the SCSI-scanning thread to stop.
+	/* Prevent SCSI scanning (if it hasn't started yet)
+	 * or wait for the SCSI-scanning routine to stop.
 	 */
-	set_bit(US_FLIDX_DONT_SCAN, &us->dflags);
-	wake_up(&us->delay_wait);
-	wait_for_completion(&us->scanning_done);
+	cancel_delayed_work_sync(&us->scan_dwork);
+
+	/* Balance autopm calls if scanning was cancelled */
+	if (test_bit(US_FLIDX_SCAN_PENDING, &us->dflags))
+		usb_autopm_put_interface_no_suspend(us->pusb_intf);
 
 	/* Removing the host will perform an orderly shutdown: caches
 	 * synchronized, disks spun down, etc.
@@ -823,53 +827,28 @@ static void release_everything(struct us_data *us)
 	scsi_host_put(us_to_host(us));
 }
 
-/* Thread to carry out delayed SCSI-device scanning */
-static int usb_stor_scan_thread(void * __us)
+/* Delayed-work routine to carry out SCSI-device scanning */
+static void usb_stor_scan_dwork(struct work_struct *work)
 {
-	struct us_data *us = (struct us_data *)__us;
+	struct us_data *us = container_of(work, struct us_data,
+			scan_dwork.work);
 	struct device *dev = &us->pusb_intf->dev;
 
-	dev_dbg(dev, "device found\n");
-
-	set_freezable();
+	dev_dbg(dev, "starting scan\n");
 
-	/*
-	 * Wait for the timeout to expire or for a disconnect
-	 *
-	 * We can't freeze in this thread or we risk causing khubd to
-	 * fail to freeze, but we can't be non-freezable either. Nor can
-	 * khubd freeze while waiting for scanning to complete as it may
-	 * hold the device lock, causing a hang when suspending devices.
-	 * So instead of using wait_event_freezable(), explicitly test
-	 * for (DONT_SCAN || freezing) in interruptible wait and proceed
-	 * if any of DONT_SCAN, freezing or timeout has happened.
-	 */
-	if (delay_use > 0) {
-		dev_dbg(dev, "waiting for device to settle "
-				"before scanning\n");
-		wait_event_interruptible_timeout(us->delay_wait,
-				test_bit(US_FLIDX_DONT_SCAN, &us->dflags) ||
-				freezing(current), delay_use * HZ);
+	/* For bulk-only devices, determine the max LUN value */
+	if (us->protocol == USB_PR_BULK && !(us->fflags & US_FL_SINGLE_LUN)) {
+		mutex_lock(&us->dev_mutex);
+		us->max_lun = usb_stor_Bulk_max_lun(us);
+		mutex_unlock(&us->dev_mutex);
 	}
+	scsi_scan_host(us_to_host(us));
+	dev_dbg(dev, "scan complete\n");
 
-	/* If the device is still connected, perform the scanning */
-	if (!test_bit(US_FLIDX_DONT_SCAN, &us->dflags)) {
-
-		/* For bulk-only devices, determine the max LUN value */
-		if (us->protocol == USB_PR_BULK &&
-				!(us->fflags & US_FL_SINGLE_LUN)) {
-			mutex_lock(&us->dev_mutex);
-			us->max_lun = usb_stor_Bulk_max_lun(us);
-			mutex_unlock(&us->dev_mutex);
-		}
-		scsi_scan_host(us_to_host(us));
-		dev_dbg(dev, "scan complete\n");
-
-		/* Should we unbind if no devices were detected? */
-	}
+	/* Should we unbind if no devices were detected? */
 
 	usb_autopm_put_interface(us->pusb_intf);
-	complete_and_exit(&us->scanning_done, 0);
+	clear_bit(US_FLIDX_SCAN_PENDING, &us->dflags);
 }
 
 static unsigned int usb_stor_sg_tablesize(struct usb_interface *intf)
@@ -916,7 +895,7 @@ int usb_stor_probe1(struct us_data **pus,
 	init_completion(&us->cmnd_ready);
 	init_completion(&(us->notify));
 	init_waitqueue_head(&us->delay_wait);
-	init_completion(&us->scanning_done);
+	INIT_DELAYED_WORK(&us->scan_dwork, usb_stor_scan_dwork);
 
 	/* Associate the us_data structure with the USB device */
 	result = associate_dev(us, intf);
@@ -947,7 +926,6 @@ EXPORT_SYMBOL_GPL(usb_stor_probe1);
 /* Second part of general USB mass-storage probing */
 int usb_stor_probe2(struct us_data *us)
 {
-	struct task_struct *th;
 	int result;
 	struct device *dev = &us->pusb_intf->dev;
 
@@ -988,20 +966,14 @@ int usb_stor_probe2(struct us_data *us)
 		goto BadDevice;
 	}
 
-	/* Start up the thread for delayed SCSI-device scanning */
-	th = kthread_create(usb_stor_scan_thread, us, "usb-stor-scan");
-	if (IS_ERR(th)) {
-		dev_warn(dev,
-				"Unable to start the device-scanning thread\n");
-		complete(&us->scanning_done);
-		quiesce_and_remove_host(us);
-		result = PTR_ERR(th);
-		goto BadDevice;
-	}
-
+	/* Submit the delayed_work for SCSI-device scanning */
 	usb_autopm_get_interface_no_resume(us->pusb_intf);
-	wake_up_process(th);
+	set_bit(US_FLIDX_SCAN_PENDING, &us->dflags);
 
+	if (delay_use > 0)
+		dev_dbg(dev, "waiting for device to settle before scanning\n");
+	queue_delayed_work(system_freezable_wq, &us->scan_dwork,
+			delay_use * HZ);
 	return 0;
 
 	/* We come here if there are any problems */
diff --git a/drivers/usb/storage/usb.h b/drivers/usb/storage/usb.h
index 7b0f2113632e..75f70f04f37b 100644
--- a/drivers/usb/storage/usb.h
+++ b/drivers/usb/storage/usb.h
@@ -47,6 +47,7 @@
 #include <linux/blkdev.h>
 #include <linux/completion.h>
 #include <linux/mutex.h>
+#include <linux/workqueue.h>
 #include <scsi/scsi_host.h>
 
 struct us_data;
@@ -72,7 +73,7 @@ struct us_unusual_dev {
 #define US_FLIDX_DISCONNECTING	3	/* disconnect in progress   */
 #define US_FLIDX_RESETTING	4	/* device reset in progress */
 #define US_FLIDX_TIMED_OUT	5	/* SCSI midlayer timed out  */
-#define US_FLIDX_DONT_SCAN	6	/* don't scan (disconnect)  */
+#define US_FLIDX_SCAN_PENDING	6	/* scanning not yet done    */
 #define US_FLIDX_REDO_READ10	7	/* redo READ(10) command    */
 #define US_FLIDX_READ10_WORKED	8	/* previous READ(10) succeeded */
 
@@ -147,8 +148,8 @@ struct us_data {
 	/* mutual exclusion and synchronization structures */
 	struct completion	cmnd_ready;	 /* to sleep thread on	    */
 	struct completion	notify;		 /* thread begin/end	    */
-	wait_queue_head_t	delay_wait;	 /* wait during scan, reset */
-	struct completion	scanning_done;	 /* wait for scan thread    */
+	wait_queue_head_t	delay_wait;	 /* wait during reset	    */
+	struct delayed_work	scan_dwork;	 /* for async scanning      */
 
 	/* subdriver information */
 	void			*extra;		 /* Any extra data          */