On Tue, Jun 18, 2019 at 10:37 AM Todd Kjos <tkjos@xxxxxxxxxx> wrote: > > On Tue, Jun 18, 2019 at 5:18 AM Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote: > > > > It's weird that that binder_alloc_copy_from_buffer() is a void function. > > It would be easier to do the error handling at that point, instead of in > > the callers. It feels like we keep hitting similar bugs to this. I took your advice. Fix posted: https://lkml.org/lkml/2019/6/28/857 -Todd > > The idea is that if it is an error that the user can cause, it is > checked by the caller of binder_alloc_copy_from_buffer(). Most uses > are kernel cases where the expected alignments should be fine and it's > a BUG if they are not. > > Admittedly, a few cases (like this one) have slipped through since > they cannot happen in Android (syzkaller has been very useful to find > our bad assumptions). > > -Todd > > > > > regards, > > dan carpenter > > _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
