USB: skel_read really sucks royally
The read code path of the skeleton driver really sucks
- skel_read works only for devices which always send data
- the timeout comes out of thin air
- it blocks signals for the duration of the timeout
- it disallows nonblocking IO by design
This patch fixes it by using a real urb, a completion and interruptible waits.
Signed-off-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
diff --git a/drivers/usb/usb-skeleton.c b/drivers/usb/usb-skeleton.c
index 60ba631..5ffa3e2 100644
--- a/drivers/usb/usb-skeleton.c
+++ b/drivers/usb/usb-skeleton.c
@@ -52,15 +52,21 @@
struct usb_interface *interface; /* the interface for this device */
struct semaphore limit_sem; /* limiting the number of writes in progress */
struct usb_anchor submitted; /* in case we need to retract our submissions */
+ struct urb *bulk_in_urb; /* the urb to read data with */
unsigned char *bulk_in_buffer; /* the buffer to receive data */
size_t bulk_in_size; /* the size of the receive buffer */
+ size_t bulk_in_filled; /* number of bytes in the buffer */
+ size_t bulk_in_copied; /* already copied to user space */
__u8 bulk_in_endpointAddr; /* the address of the bulk in endpoint */
__u8 bulk_out_endpointAddr; /* the address of the bulk out endpoint */
int errors; /* the last request tanked */
int open_count; /* count the number of openers */
+ bool ongoing_read; /* a read is going on */
+ bool processed_urb; /* indicates we haven't processed the urb */
spinlock_t err_lock; /* lock for errors */
struct kref kref;
struct mutex io_mutex; /* synchronize I/O with disconnect */
+ struct completion bulk_in_completion; /* to wait for an ongoing read */
};
#define to_skel_dev(d) container_of(d, struct usb_skel, kref)
@@ -71,6 +77,7 @@
{
struct usb_skel *dev = to_skel_dev(kref);
+ usb_free_urb(dev->bulk_in_urb);
usb_put_dev(dev->udev);
kfree(dev->bulk_in_buffer);
kfree(dev);
@@ -174,38 +181,182 @@
return res;
}
+static void skel_read_bulk_callback(struct urb *urb)
+{
+ struct usb_skel *dev;
+
+ dev = urb->context;
+
+ spin_lock(&dev->err_lock);
+ /* sync/async unlink faults aren't errors */
+ if (urb->status) {
+ if(!(urb->status == -ENOENT ||
+ urb->status == -ECONNRESET ||
+ urb->status == -ESHUTDOWN))
+ err("%s - nonzero write bulk status received: %d",
+ __func__, urb->status);
+
+ dev->errors = urb->status;
+ } else {
+ dev->bulk_in_filled = urb->actual_length;
+ }
+ dev->ongoing_read = 0;
+ spin_unlock(&dev->err_lock);
+
+ complete(&dev->bulk_in_completion);
+}
+
+static int skel_do_read_io(struct usb_skel *dev, size_t count)
+{
+ int rv;
+
+ /* prepare a read */
+ usb_fill_bulk_urb(dev->bulk_in_urb,
+ dev->udev,
+ usb_rcvbulkpipe(dev->udev,
+ dev->bulk_in_endpointAddr),
+ dev->bulk_in_buffer,
+ min(dev->bulk_in_size, count),
+ skel_read_bulk_callback,
+ dev);
+ /* tell everybody to leave the URB alone */
+ spin_lock_irq(&dev->err_lock);
+ dev->ongoing_read = 1;
+ spin_unlock_irq(&dev->err_lock);
+
+ /* do it */
+ rv = usb_submit_urb(dev->bulk_in_urb, GFP_KERNEL);
+ if (rv < 0) {
+ err("%s - failed submitting read urb, error %d",
+ __func__, rv);
+ dev->bulk_in_filled = 0;
+ rv = (rv == -ENOMEM) ? rv : -EIO;
+ spin_lock_irq(&dev->err_lock);
+ dev->ongoing_read = 0;
+ spin_unlock_irq(&dev->err_lock);
+ }
+
+ return rv;
+}
+
static ssize_t skel_read(struct file *file, char *buffer, size_t count, loff_t *ppos)
{
struct usb_skel *dev;
- int retval;
- int bytes_read;
+ int rv;
+ bool ongoing_io;
dev = (struct usb_skel *)file->private_data;
- mutex_lock(&dev->io_mutex);
+ /* if we cannot read at all, return EOF */
+ if (!dev->bulk_in_urb || !count)
+ return 0;
+
+ /* no concurrent readers */
+ rv = mutex_lock_interruptible(&dev->io_mutex);
+ if (rv < 0)
+ return rv;
+
if (!dev->interface) { /* disconnect() was called */
- retval = -ENODEV;
+ rv = -ENODEV;
goto exit;
}
- /* do a blocking bulk read to get data from the device */
- retval = usb_bulk_msg(dev->udev,
- usb_rcvbulkpipe(dev->udev, dev->bulk_in_endpointAddr),
- dev->bulk_in_buffer,
- min(dev->bulk_in_size, count),
- &bytes_read, 10000);
+ /* if IO is under way, we must not touch things */
+retry:
+ spin_lock_irq(&dev->err_lock);
+ ongoing_io = dev->ongoing_read;
+ spin_unlock_irq(&dev->err_lock);
- /* if the read was successful, copy the data to userspace */
- if (!retval) {
- if (copy_to_user(buffer, dev->bulk_in_buffer, bytes_read))
- retval = -EFAULT;
- else
- retval = bytes_read;
+ if (ongoing_io) {
+ /*
+ * IO may take forever
+ * hence wait in an interruptible state
+ */
+ rv = wait_for_completion_interruptible(&dev->bulk_in_completion);
+ if (rv < 0)
+ goto exit;
+ /*
+ * by waiting we also semiprocessed the urb
+ * we must finish now
+ */
+ dev->bulk_in_copied = 0;
+ dev->processed_urb = 1;
}
+ if (!dev->processed_urb) {
+ /*
+ * the URB hasn't been processed
+ * do it now
+ */
+ wait_for_completion(&dev->bulk_in_completion);
+ dev->bulk_in_copied = 0;
+ dev->processed_urb = 1;
+ }
+
+ /* errors must be reported */
+ if ((rv = dev->errors) < 0) {
+ /* any error is reported once */
+ dev->errors = 0;
+ /* to preserve notifications about reset */
+ rv = (rv == -EPIPE) ? rv : -EIO;
+ /* no data to deliver */
+ dev->bulk_in_filled = 0;
+ /* report it */
+ goto exit;
+ }
+
+ /*
+ * if the buffer is filled we may satisfy the read
+ * else we need to start IO
+ */
+
+ if (dev->bulk_in_filled) {
+ /* we had read data */
+ size_t available = dev->bulk_in_filled - dev->bulk_in_copied;
+ size_t chunk = min(available, count);
+
+ if (!available) {
+ /*
+ * all data has been used
+ * actual IO needs to be done
+ */
+ rv = skel_do_read_io(dev, count);
+ if (rv < 0)
+ goto exit;
+ else
+ goto retry;
+ }
+ /*
+ * data is available
+ * chunk tells us how much shall be copied
+ */
+
+ if (copy_to_user(buffer,
+ dev->bulk_in_buffer + dev->bulk_in_copied,
+ chunk))
+ rv = -EFAULT;
+ else
+ rv = chunk;
+
+ dev->bulk_in_copied += chunk;
+
+ /*
+ * if we are asked for more than we have,
+ * we start IO but don't wait
+ */
+ if (available < count)
+ skel_do_read_io(dev, count - chunk);
+ } else {
+ /* no data in the buffer */
+ rv = skel_do_read_io(dev, count);
+ if (rv < 0)
+ goto exit;
+ else
+ goto retry;
+ }
exit:
mutex_unlock(&dev->io_mutex);
- return retval;
+ return rv;
}
static void skel_write_bulk_callback(struct urb *urb)
@@ -363,6 +514,7 @@
mutex_init(&dev->io_mutex);
spin_lock_init(&dev->err_lock);
init_usb_anchor(&dev->submitted);
+ init_completion(&dev->bulk_in_completion);
dev->udev = usb_get_dev(interface_to_usbdev(interface));
dev->interface = interface;
@@ -384,6 +536,11 @@
err("Could not allocate bulk_in_buffer");
goto error;
}
+ dev->bulk_in_urb = usb_alloc_urb(0, GFP_KERNEL);
+ if (!dev->bulk_in_urb) {
+ err("Could not allocate bulk_in_urb");
+ goto error;
+ }
}
if (!dev->bulk_out_endpointAddr &&
@@ -453,6 +610,7 @@
time = usb_wait_anchor_empty_timeout(&dev->submitted, 1000);
if (!time)
usb_kill_anchored_urbs(&dev->submitted);
+ usb_kill_urb(dev->bulk_in_urb);
}
static int skel_suspend(struct usb_interface *intf, pm_message_t message)