On 05/08/13 18:53, Dan Carpenter wrote:
On Mon, Aug 05, 2013 at 06:40:13PM +0100, Rupesh Gujare wrote:
This patch fixes crash issue when there is quick cycle of
de-enumeration & enumeration due to loss of wireless link.
It is found that sometimes new device (or coming back device)
returns very fast, even before USB core read out hub status,
resulting in allocation of same port, which results in unstable
system & crash.
Above issue is resolved by making sure that we always assign
new port to new device, making sure that USB core reads correct
hub status.
This feels like papering over the problem. Surely the real fix
would be to improve the reference counting.
This patch is probably effective but it makes the code more subtle
and it shows that we don't really understand what we are doing with
regards to reference counting.
Probably this is easier way to fix issue, since we don't have reference
count for ports & we rely on flags to check port status.
Any suggestions are appreciated.
--
Regards,
Rupesh Gujare
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
