]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
powerpc/iommu: Annotate nested lock for lockdep
authorAlexey Kardashevskiy <aik@ozlabs.ru>
Mon, 1 Mar 2021 06:36:53 +0000 (17:36 +1100)
committerMichael Ellerman <mpe@ellerman.id.au>
Thu, 22 Apr 2021 15:38:04 +0000 (01:38 +1000)
commitcc7130bf119add37f36238343a593b71ef6ecc1e
tree87079ec312685b7511dcc606b0ac196f0c203790
parent4be518d838809e21354f32087aa9c26efc50b410
powerpc/iommu: Annotate nested lock for lockdep

The IOMMU table is divided into pools for concurrent mappings and each
pool has a separate spinlock. When taking the ownership of an IOMMU group
to pass through a device to a VM, we lock these spinlocks which triggers
a false negative warning in lockdep (below).

This fixes it by annotating the large pool's spinlock as a nest lock
which makes lockdep not complaining when locking nested locks if
the nest lock is locked already.

===
WARNING: possible recursive locking detected
5.11.0-le_syzkaller_a+fstn1 #100 Not tainted
--------------------------------------------
qemu-system-ppc/4129 is trying to acquire lock:
c0000000119bddb0 (&(p->lock)/1){....}-{2:2}, at: iommu_take_ownership+0xac/0x1e0

but task is already holding lock:
c0000000119bdd30 (&(p->lock)/1){....}-{2:2}, at: iommu_take_ownership+0xac/0x1e0

other info that might help us debug this:
 Possible unsafe locking scenario:

       CPU0
       ----
  lock(&(p->lock)/1);
  lock(&(p->lock)/1);
===

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20210301063653.51003-1-aik@ozlabs.ru
arch/powerpc/kernel/iommu.c