Re: [BUG] possible deadlock in __schedule (with reproducer available)

[Hillf Danton <hdanton@sina.com>]

On Wed, 27 Nov 2024 18:27:57 -0800 Alexei Starovoitov <alexei.starovoitov@gmail.com>
> On Wed, Nov 27, 2024 at 3:04 PM Hillf Danton <hdanton@sina.com> wrote:
> > On Tue, 26 Nov 2024 13:15:48 -0800 Andrii Nakryiko <andrii.nakryiko@gmail.com>
> > > On Mon, Nov 25, 2024 at 1:44 AM Peter Zijlstra <peterz@infradead.org> wrote:
> > > > On Mon, Nov 25, 2024 at 05:24:05AM +0000, Ruan Bonan wrote:
> > > >
> > > > > From the discussion, it appears that the root cause might involve
> > > > > specific printk or BPF operations in the given context. To clarify and
> > > > > possibly avoid similar issues in the future, are there guidelines or
> > > > > best practices for writing BPF programs/hooks that interact with
> > > > > tracepoints, especially those related to scheduler events, to prevent
> > > > > such deadlocks?
> > > >
> > > > The general guideline and recommendation for all tracepoints is to be
> > > > wait-free. Typically all tracer code should be.
> > > >
> > > > Now, BPF (users) (ab)uses tracepoints to do all sorts and takes certain
> > > > liberties with them, but it is very much at the discretion of the BPF
> > > > user.
> > >
> > > We do assume that tracepoints are just like kprobes and can run in
> > > NMI. And in this case BPF is just a vehicle to trigger a
> > > promised-to-be-wait-free strncpy_from_user_nofault(). That's as far as
> > > BPF involvement goes, we should stop discussing BPF in this context,
> > > it's misleading.
> > >
> > Given known issue, syzbot should run without bpf enabled before it is fixed
> > to avoid more useless discussing and misleading.
> 
> If you cared to read the thread it would have been obvious
> that printk is the culprit. Tell syzbot to run without printk?
>
Printk is innocent, and it makes no sense to put the gun vendor into
jail simply because bpf shoot a sheriff in the cafeteira.