Re: complete_all() with 5 waiters in swake_up_all_locked

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 2018-10-10 11:15:46 [+0200], Philipp K wrote:
> Hi,
Hi,

sorry for being late.

> I've been fuzzing linux-rt with syzkaller and found a way to trigger
> $title warning.
> 
> The reproducer works with a defconfig + CONFIG_NET_KEY=y on 4.18.12-rt6 kernel.

I didn't manage to reproduce this on v4.19-RT. Also that backtrace is
from v4.9. With that test case I didn't manage to end up in
cryptomgr_probe() which would be a requirement.

So. The warning says that the system managed to wake up 5 tasks in one
go, that is without re-enabling the interrupts. It has been agreed that
two wakes are okay and more should be avoided. The problem is that it
may cause latency spikes if for some reason the system wakes say 100
tasks like that.
We have one exception for the PM case because nobody does (or should do)
suspend-to-ram and run a RT task which has to meet a deadline.

So for this case, we should figure out why it managed to wake 5 tasks
and how bad can it get and can we limit it somehow.

> Philipp
> 
> PS: Liebe Grüße von Lukas Bulwahn
Ah. Greetings!

Sebastian



[Index of Archives]     [RT Stable]     [Kernel Newbies]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Samba]     [Video 4 Linux]     [Device Mapper]

  Powered by Linux