Re: [PATCH v2] kasan: remove kasan_find_vm_area() to prevent possible deadlock

[Marco Elver]

On Wed, 16 Jul 2025 at 11:09, Uladzislau Rezki <urezki@xxxxxxxxx> wrote:
>
> On Mon, Jul 14, 2025 at 08:27:40AM +0900, Byungchul Park wrote:
> > On Sat, Jul 12, 2025 at 03:46:10PM +0000, Yeo Reum Yun wrote:
> > > Hi ByungChul,
> > >
> > > [...]
> > > > I checked the critical section by &vn->busy.lock in find_vm_area().  The
> > > > time complextity looks O(log N).  I don't think an irq disabled section
> > > > of O(log N) is harmful.  I still think using
> > > > spin_lock_irqsave(&vn->busy.lock) can resolve this issue with no worry
> > > > of significant irq delay.  Am I missing something?
> > >
> > > I don't agree for this.
> > > since in PREEMPT_RT case, it has the same problem.
> > >
> > > In case of PREEMPT_RT, spin_lock_irqsave() becomes rt_spin_lock() which is sleepable.
> > > But, KASAN calls "rt_spin_lock()" holding raw_spin_lock_irqsave() which is definitely wrong.
> >
> > It's another issue than irq handling latency, but it's about lock usage
> > correctness.  You are right.
> >
> There is vmalloc_dump_obj() function which should be used IMO:
>
> <snip>
> pr_err("The buggy address %px belongs to a vmalloc virtual mapping, dump it...\n", addr);
> vmalloc_dump_obj(addr);
> <snip>
>
> we use trylock there to eliminate an issue if invoked from the IRQ
> context.

Something like that should work:
https://lkml.kernel.org/r/20250716152448.3877201-1-elver@xxxxxxxxxx