2023-02-27 03:26:09

by void0red

Subject: [PATCH] perf/smmuv3: add a check of devm_add_action in smmu_pmu_setup_msi

From: Kang Chen <[email protected]>

devm_add_action may fails, add a warning when it happens.

Signed-off-by: Kang Chen <[email protected]>
---
drivers/perf/arm_smmuv3_pmu.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/perf/arm_smmuv3_pmu.c b/drivers/perf/arm_smmuv3_pmu.c
index 25a269d43..307bac904 100644
--- a/drivers/perf/arm_smmuv3_pmu.c
+++ b/drivers/perf/arm_smmuv3_pmu.c
@@ -724,7 +724,11 @@ static void smmu_pmu_setup_msi(struct smmu_pmu *pmu)
pmu->irq = msi_get_virq(dev, 0);

/* Add callback to free MSIs on teardown */
- devm_add_action(dev, smmu_pmu_free_msis, dev);
+ ret = devm_add_action(dev, smmu_pmu_free_msis, dev);
+ if (ret) {
+ dev_warn(dev, "failed to add callback to free MSIs on teardown\n");
+ return;
+ }
}

static int smmu_pmu_setup_irq(struct smmu_pmu *pmu)
--
2.34.1

2023-03-27 11:32:29

by Will Deacon

Subject: Re: [PATCH] perf/smmuv3: add a check of devm_add_action in smmu_pmu_setup_msi

On Mon, Feb 27, 2023 at 11:25:57AM +0800, void0red wrote:
> From: Kang Chen <[email protected]>
>
> devm_add_action may fails, add a warning when it happens.
>
> Signed-off-by: Kang Chen <[email protected]>
> ---
> drivers/perf/arm_smmuv3_pmu.c | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/perf/arm_smmuv3_pmu.c b/drivers/perf/arm_smmuv3_pmu.c
> index 25a269d43..307bac904 100644
> --- a/drivers/perf/arm_smmuv3_pmu.c
> +++ b/drivers/perf/arm_smmuv3_pmu.c
> @@ -724,7 +724,11 @@ static void smmu_pmu_setup_msi(struct smmu_pmu *pmu)
> pmu->irq = msi_get_virq(dev, 0);
>
> /* Add callback to free MSIs on teardown */
> - devm_add_action(dev, smmu_pmu_free_msis, dev);
> + ret = devm_add_action(dev, smmu_pmu_free_msis, dev);
> + if (ret) {
> + dev_warn(dev, "failed to add callback to free MSIs on teardown\n");
> + return;
> + }

I appreciate that you're following the existing error handling in this
function, but I don't think this message is especially helpful. We'd be
better off propagating the error and failing the probe in this case, I
reckon.

Will