]> asedeno.scripts.mit.edu Git - linux.git/commit
r8152: Re-order napi_disable in rtl8152_close
authorPrashant Malani <pmalani@chromium.org>
Wed, 20 Nov 2019 19:40:21 +0000 (11:40 -0800)
committerDavid S. Miller <davem@davemloft.net>
Wed, 20 Nov 2019 20:48:13 +0000 (12:48 -0800)
commit84811412464d66fcf73661a279aa1d7985166495
tree9929dd11897149a19c3cf6ec9b04d2d888e4e28e
parentb172845a4090e52f233e45801ce7f44ecc4cc955
r8152: Re-order napi_disable in rtl8152_close

Both rtl_work_func_t() and rtl8152_close() call napi_disable().
Since the two calls aren't protected by a lock, if the close
function starts executing before the work function, we can get into a
situation where the napi_disable() function is called twice in
succession (first by rtl8152_close(), then by set_carrier()).

In such a situation, the second call would loop indefinitely, since
rtl8152_close() doesn't call napi_enable() to clear the NAPI_STATE_SCHED
bit.

The rtl8152_close() function in turn issues a
cancel_delayed_work_sync(), and so it would wait indefinitely for the
rtl_work_func_t() to complete. Since rtl8152_close() is called by a
process holding rtnl_lock() which is requested by other processes, this
eventually leads to a system deadlock and crash.

Re-order the napi_disable() call to occur after the work function
disabling and urb cancellation calls are issued.

Change-Id: I6ef0b703fc214998a037a68f722f784e1d07815e
Reported-by: http://crbug.com/1017928
Signed-off-by: Prashant Malani <pmalani@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/usb/r8152.c