Fix USB console hang due to non-atomic URB allocation This is intended to fix a problem seen with the Amtel sam-ba tool by Alexander Stein (alexander.stein@xxxxxxxxxxxxxxxxxxxxx). It appears when using an A91 controller. He bisected it to commit: 8e8dce065088833fc418bfa5fbf035cb0726c04c: USB: use kfifo to buffer usb-generic serial writes The current patch fixes a problem when using a USB serial device as the kernel console and as /dev/console. The URB allocation is not atomic and an URB can be doubly allocated, leading to a continual rejection of URB submissions. The fix puts all pieces of the URB allocation in the same spinlock-protected section of code. Having said that... This fix was developed because, after Alexander's email, I took a look at my USB console and found that it, too, was experiencing a hang. I assumed that this hang was the same as the one Alexander is seeing and developed this fix. It's a good fix for *a* hang, but after thinking about this a bit, I'm not sure this is a fix for *Alexander's* hang. Signed-off-by: <dvomlehn@xxxxxxxxx> --- drivers/usb/serial/generic.c | 5 ++++- 1 files changed, 4 insertions(+), 1 deletions(-) diff --git a/drivers/usb/serial/generic.c b/drivers/usb/serial/generic.c index a817ced..7085809 100644 --- a/drivers/usb/serial/generic.c +++ b/drivers/usb/serial/generic.c @@ -199,6 +199,9 @@ retry: } i = (int)find_first_bit(&port->write_urbs_free, ARRAY_SIZE(port->write_urbs)); + if (i == ARRAY_SIZE(port->write_urbs)) + return 0; + clear_bit(i, &port->write_urbs_free); spin_unlock_irqrestore(&port->lock, flags); urb = port->write_urbs[i]; @@ -213,9 +216,9 @@ retry: dev_err(&port->dev, "%s - error submitting urb: %d\n", __func__, result); clear_bit_unlock(USB_SERIAL_WRITE_BUSY, &port->flags); + set_bit(i, &port->write_urbs_free); return result; } - clear_bit(i, &port->write_urbs_free); spin_lock_irqsave(&port->lock, flags); port->tx_bytes += count; -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
