Re: [syzbot] [netfs?] INFO: task hung in netfs_unbuffered_write_iter

[K Prateek Nayak]

Hello Oleg,

On 3/28/2025 10:30 PM, Oleg Nesterov wrote:
> Dear syzbot,
>
> On 03/28, syzbot wrote:
> > syzbot has tested the proposed patch and the reproducer did not trigger any issue:
> >
> > Reported-by: syzbot+62262fdc0e01d99573fc@xxxxxxxxxxxxxxxxxxxxxxxxx
> > Tested-by: syzbot+62262fdc0e01d99573fc@xxxxxxxxxxxxxxxxxxxxxxxxx

Yours is the right approach. I was searching for p9_poll_mux()
which would wake the read collector to free pipe buffers but I
didn't dig enough and I was too lazy to open code it (guilty!)
If this gets picked, feel free to add:

Reviewed-and-tested-by: K Prateek Nayak <kprateek.nayak@xxxxxxx>
 
> Thanks. so the previous
>
> 	syzbot has tested the proposed patch but the reproducer is still triggering an issue:
> 	unregister_netdevice: waiting for DEV to become free
>
> 	unregister_netdevice: waiting for batadv0 to become free. Usage count = 3
>
> report suggests that upstream/master has another/unrelated issue(s).

I could not reproduce this bit on mainline at commit acb4f33713b9
("Merge tag 'm68knommu-for-v6.15' of
git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu") so I too
second that it is an unrelated issue.

> As for the patches from me or Prateek (thanks again!), I think that
> the maintainers should take a look.
>
> But at this point I am mostly confident that the bisected commit aaec5a95d5961
> ("pipe_read: don't wake up the writer if the pipe is still full") is innocent,
> it just reveals yet another problem.
>
> I guess (I hope ;) Prateek agrees.

I agree aaec5a95d5961 is not guilty! (unlike me) :)

> Oleg.

--
Thanks and Regards,
Prateek