On Wed, May 22, 2024 at 10:07:14AM +0200, Andrew Jones wrote:
> On Tue, May 14, 2024 at 11:16:19AM GMT, Tomasz Jeznach wrote:
> ...
> > +static int riscv_iommu_bond_link(struct riscv_iommu_domain *domain,
> > + struct device *dev)
> > +{
> > + struct riscv_iommu_device *iommu = dev_to_iommu(dev);
> > + struct riscv_iommu_bond *bond;
> > + struct list_head *bonds;
> > +
> > + bond = kzalloc(sizeof(*bond), GFP_KERNEL);
> > + if (!bond)
> > + return -ENOMEM;
> > + bond->dev = dev;
> > +
> > + /*
> > + * List of devices attached to the domain is arranged based on
> > + * managed IOMMU device.
> > + */
> > +
> > + spin_lock(&domain->lock);
> > + list_for_each_rcu(bonds, &domain->bonds)
> > + if (dev_to_iommu(list_entry(bonds, struct riscv_iommu_bond, list)->dev) == iommu)
> > + break;
>
> We should wrap this list_for_each_rcu() in rcu_read_lock() and
> rcu_read_unlock().
Not quite this is the write side, it is holding the spinlock so it
doesn't need RCU. It should just call the normal list_for_each_entry()
Jason
