Hi Linus,
Unfortunately there was a SMP race between ->read() and ->ioctl() routines
(thanks to Mark Hemment for pointing this out!). If one was inside the
->read() after the EOF check and an ioctl requests came in onother CPU
which freed the memory and reset the size then one could panic. The fi
Hi Linus,
I have just found and fixed another bug in the driver -- the
->ioctl() method would print "freed 0 bytes" when it actually freed >0, if
the update has failed previously.
The patch is cumulative, i.e. contains all the previous fixes I have sent
to you in the last few days.
Tested thoro
2 matches
Mail list logo
