[PATCH] iommu/tegra241-cmdqv: do not call smp_processor_id in preemptible context

[PATCH] iommu/tegra241-cmdqv: do not call smp_processor_id in preemptible context

[Luis Claudio R. Goncalves]

With PREEMPT_RT enabled some of the calls to tegra241_cmdqv_get_cmdq()
during boot will happen in preemptible context. As this function calls
smp_processor_id(), these calls will trigger a "BUG: using smp_processor_id()
in preemptible" backtrace if DEBUG_PREEMPT is enabled.

As tegra241_cmdqv_get_cmdq() only calls smp_processor_id() to use the
CPU number as a factor to balance out traffic on cmdq usage, it is safe
to use raw_smp_processor_id() here.

Signed-off-by: Luis Claudio R. Goncalves <lgoncalv@redhat.com>
---
 drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c b/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
index fcd13d301fff6..3ea4369e838fe 100644
--- a/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
+++ b/drivers/iommu/arm/arm-smmu-v3/tegra241-cmdqv.c
@@ -339,7 +339,7 @@ tegra241_cmdqv_get_cmdq(struct arm_smmu_device *smmu,
 	 * one CPU at a time can enter the process, while the others
 	 * will be spinning at the same lock.
 	 */
-	lidx = smp_processor_id() % cmdqv->num_lvcmdqs_per_vintf;
+	lidx = raw_smp_processor_id() % cmdqv->num_lvcmdqs_per_vintf;
 	vcmdq = vintf->lvcmdqs[lidx];
 	if (!vcmdq || !READ_ONCE(vcmdq->enabled))
 		return NULL;
-- 
2.47.0

Re: [PATCH] iommu/tegra241-cmdqv: do not call smp_processor_id in preemptible context

[Sebastian Andrzej Siewior]

On 2024-11-18 21:19:28 [-0300], Luis Claudio R. Goncalves wrote:
> With PREEMPT_RT enabled some of the calls to tegra241_cmdqv_get_cmdq()
> during boot will happen in preemptible context. As this function calls
> smp_processor_id(), these calls will trigger a "BUG: using smp_processor_id()
> in preemptible" backtrace if DEBUG_PREEMPT is enabled.

If this is only on PREEMPT_RT, where is the disabled preemption coming
from on !PREEMPT_RT?

> As tegra241_cmdqv_get_cmdq() only calls smp_processor_id() to use the
> CPU number as a factor to balance out traffic on cmdq usage, it is safe
> to use raw_smp_processor_id() here.
> 
> Signed-off-by: Luis Claudio R. Goncalves <lgoncalv@redhat.com>

Sebastian

Re: [PATCH] iommu/tegra241-cmdqv: do not call smp_processor_id in preemptible context

[Luis Claudio R. Goncalves]

On Tue, Nov 19, 2024 at 08:15:10AM +0100, Sebastian Andrzej Siewior wrote:
> On 2024-11-18 21:19:28 [-0300], Luis Claudio R. Goncalves wrote:
> > With PREEMPT_RT enabled some of the calls to tegra241_cmdqv_get_cmdq()
> > during boot will happen in preemptible context. As this function calls
> > smp_processor_id(), these calls will trigger a "BUG: using smp_processor_id()
> > in preemptible" backtrace if DEBUG_PREEMPT is enabled.
> 
> If this is only on PREEMPT_RT, where is the disabled preemption coming
> from on !PREEMPT_RT?

You are correct. This is not exclusive to PREEMPT_RT, also happens with the
stock kernel. I ran more tests and this time I made sure DEBUG_PREEMPT was
enabled. That was enough to reproduce the problem.
 
> > As tegra241_cmdqv_get_cmdq() only calls smp_processor_id() to use the
> > CPU number as a factor to balance out traffic on cmdq usage, it is safe
> > to use raw_smp_processor_id() here.
> > 
> > Signed-off-by: Luis Claudio R. Goncalves <lgoncalv@redhat.com>
> 
> Sebastian
> 
---end quoted text---