On Tue, Dec 12, 2017 at 06:11:09PM +0100, Thomas Gleixner wrote: > > > > I wonder if anything would break if we made sys_timer_create() return -EINVAL > > for unrecognized values of sigev_notify? That's what it *should* do, but it > > seems to be the classic "unchecked flags" bug, yet again... > > So this is the 5th time this comes up and I sent a patch the first time. No > answer since than just more repeating reports. > > https://marc.info/?l=linux-kernel&m=151204669103208&w=2 > > Thanks, > > tglx Are you expecting an answer from syzbot? It is just a bot so it cannot review or apply patches. There is a way to ask it to test a patch (see https://github.com/google/syzkaller/blob/master/docs/syzbot.md) though it only works for bugs with reproducers, and when fixing bugs I've personally found it easier (and often necessary to debug the problem in the first place) to just run the reproducer myself. Note that syzbot *did* provide a C reproducer for this report, and also a C reproducer for the original report after a short delay, so you could run those as well as my simplified reproducer to verify the bug is fixed -- though given my investigation your patch very likely does fix the bug. Keep in mind that you can't expect that a human "behind the scenes" will quickly step in and help review/test the fixes for syzbot bugs either. Ideally that *would* happen, but there are simply too many open bugs (200+) and syzbot is operated primarily by just one person, Dmitry Vyukov. I've been helping out with some bugs I am interested in and I will get to as many as I can but there are far too many for even 2 people. So the community needs to help out. Again, see the docs at https://github.com/google/syzkaller/blob/master/docs/syzbot.md for how to communicate with the bot and tell it when a bug is a dup, when a bug is fixed, or how to test a patch, etc. Eric
