BUG? racy code at net/atm/br2684.c

Hi. I am reporting a suspected race bug at br2684_push()
in /net/atm/br2684.c.

Please examine the report and let me know your opinion.
I reported the same issue several month ago and got feedback
however, I found that the code is still not changed.

In br2684_push() accesses &brdev->brvcss without devs_lock holding at line 334.
But it seems that all accesses to &brdev->brvcss are synchronized by devs_lock.

So that br2684_push() may result race condition in concurrent execution
with other functions that manipulate &brdev->brvcss.

It seems that it is better to guard list_empty(&brdev->brvccs) by devs_lock.
Or check list_empty(&brdev->brvccs) once again after devs_lock holding.



Thank you.

Sincerely
Shin Hong