iommu/mediatek-v1: Tidy up probe_finalize
Subsystems (3)
Robin Murphy
reviewerPATCH 1/1 iommu/mediatek-v1: Tidy up probe_finalize
Krzysztof points out that although the driver now supports COMPILE_TEST for other architectures, it does not build cleanly with W=1 where the stubbed-out ARM API can lead to an unused variable warning.
Since this is effectively the correct intent of the code in such cases, mark it as __maybe_unused, tidying up some cruft in the process.
---
Since I'd already got as far as thinking this through, I figure I may as well save Krzysztof the trouble of rewriting it :)
drivers/iommu/mtk_iommu_v1.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-)
drivers/iommu/mtk_iommu_v1.c +2 -6
@@ -509,14 +509,10 @@ static struct iommu_device *mtk_iommu_v1_probe_device(struct device *dev)static void mtk_iommu_v1_probe_finalize (struct device *dev) { -struct dma_iommu_mapping *mtk_mapping; -struct mtk_iommu_v1_data *data; +__maybe_unused struct mtk_iommu_v1_data *data =dev_iommu_priv_get (dev);int err; - data =dev_iommu_priv_get (dev); - mtk_mapping = data->mapping; - - err =arm_iommu_attach_device (dev, mtk_mapping); + err =arm_iommu_attach_device (dev, data->mapping);if (err)dev_err (dev,"Can't create IOMMU mapping - DMA-OPS will not work\n" ); }
Signature · 1 line
2.39.2.101.g768bb238c484.dirty
Quoted 9 lines · Robin Murphy
External email : Please do not click links or open attachments until you have verified the sender or the content.
Krzysztof points out that although the driver now supports
COMPILE_TEST
for other architectures, it does not build cleanly with W=1 where the stubbed-out ARM API can lead to an unused variable warning.
Quoted 6 lines
Since this is effectively the correct intent of the code in such cases,
mark it as__maybe_unused, tidying up some cruft in the process.Reported-by Krzysztof Kozlowski <[email protected]>Signed-off-by Robin Murphy <[email protected]>
Quoted 36 lines
---
Since I'd already got as far as thinking this through, I figure I may as well save Krzysztof the trouble of rewriting it :)
drivers/iommu/mtk_iommu_v1.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-)drivers/iommu/mtk_iommu_v1.c +2
-6@@ -509,14 +509,10 @@ static struct iommu_device *mtk_iommu_v1_probe_device(struct device *dev)static void mtk_iommu_v1_probe_finalize (struct device *dev) { -struct dma_iommu_mapping *mtk_mapping; -struct mtk_iommu_v1_data *data; +__maybe_unused struct mtk_iommu_v1_data *data =dev_iommu_priv_get (dev);int err; - data =dev_iommu_priv_get (dev); - mtk_mapping = data->mapping; - - err =arm_iommu_attach_device (dev, mtk_mapping); + err =arm_iommu_attach_device (dev, data->mapping);if (err)dev_err(dev, "Can't create IOMMU mapping - DMA-OPS
will not work\n");
}
--
2.39.2.101.g768bb238c484.dirty
Quoted 7 lines · Robin Murphy
Krzysztof points out that although the driver now supports
COMPILE_TESTfor other architectures, it does not build cleanly with W=1 where the stubbed-out ARM API can lead to an unused variable warning.Since this is effectively the correct intent of the code in such cases, mark it as
__maybe_unused, tidying up some cruft in the process.
Quoted 6 lines
Reported-by Krzysztof Kozlowski <[email protected]>Signed-off-by Robin Murphy <[email protected]>---
Since I'd already got as far as thinking this through, I figure I may as well save Krzysztof the trouble of rewriting it :)
Sure, thanks!
Best regards,
Krzysztof
Quoted 2 lines · Robin Murphy
drivers/iommu/mtk_iommu_v1.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-)
Applied, thanks Robin.