On Tue, Jun 23, 2020 at 12:30:16PM +0100, Robin Murphy wrote:
> On 2020-06-23 09:36, Thierry Reding wrote:
> [...]
> > > diff --git a/drivers/iommu/arm-smmu.c b/drivers/iommu/arm-smmu.c
> > > index 243bc4cb2705b..d720e1e191176 100644
> > > --- a/drivers/iommu/arm-smmu.c
> > > +++ b/drivers/iommu/arm-smmu.c
> > > @@ -673,6 +673,7 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,
> > >   	enum io_pgtable_fmt fmt;
> > >   	struct arm_smmu_domain *smmu_domain = to_smmu_domain(domain);
> > >   	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
> > > +	irqreturn_t (*context_fault)(int irq, void *dev);
> > >   	mutex_lock(&smmu_domain->init_mutex);
> > >   	if (smmu_domain->smmu)
> > > @@ -835,7 +836,9 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,
> > >   	 * handler seeing a half-initialised domain state.
> > >   	 */
> > >   	irq = smmu->irqs[smmu->num_global_irqs + cfg->irptndx];
> > > -	ret = devm_request_irq(smmu->dev, irq, arm_smmu_context_fault,
> > > +	context_fault = (smmu->impl && smmu->impl->context_fault) ?
> > > +			 smmu->impl->context_fault : arm_smmu_context_fault;
> > 
> > A simpler way might have been to assign arm_smmu_context_fault to all
> > implementations. That way we wouldn't have to perform this check here
> > and instead just always using smmu->impl->context_fault.
> 
> But smmu->impl can still be NULL...
> 
> Everything in impl, including the presence of impl itself, is optional, so
> the notion of overriding a default with the same default doesn't really make
> much sense, and would go against the pattern everywhere else.

True. I had assumed that every implementation would set smmu->impl
anyway, in which case there'd be little reason to use these default
fallbacks since each implementation could simply directly refer to the
exact implementation that it wants.

Perhaps the above could be made a bit more palatable by using a standard
if/else rather than the ternary operator? That would also more closely
match the pattern elsewhere.

Thierry