| Message ID | 20180514161852.9620-1-srinivas.kandagatla@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 14-05-18, 17:18, Srinivas Kandagatla wrote: > Disabling pm runtime at probe is not sufficient to get BAM working > on remotely controller instances. pm_runtime_get_sync() would return > -EACCES in such cases. > So check if runtime pm is enabled before returning error from bam functions. > > Fixes: 5b4a68952a89 ("dmaengine: qcom: bam_dma: disable runtime pm on remote controlled") > Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org> > --- > drivers/dma/qcom/bam_dma.c | 10 +++++----- > 1 file changed, 5 insertions(+), 5 deletions(-) > > diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c > index d29275b97e84..5f4babebc508 100644 > --- a/drivers/dma/qcom/bam_dma.c > +++ b/drivers/dma/qcom/bam_dma.c > @@ -540,7 +540,7 @@ static void bam_free_chan(struct dma_chan *chan) > int ret; > > ret = pm_runtime_get_sync(bdev->dev); > - if (ret < 0) > + if (pm_runtime_enabled(bdev->dev) && ret < 0) would it make sense to first check enabled and do _get_sync() if (pm_runtime_enabled()) { ret = pm_runtime_get_sync() { ... } } thus making clear in code that we do calls only when it is enabled. Also you can add a local macro for this code and use that rather than copy pasting :)
On 17/05/18 06:21, Vinod wrote: > On 14-05-18, 17:18, Srinivas Kandagatla wrote: >> Disabling pm runtime at probe is not sufficient to get BAM working >> on remotely controller instances. pm_runtime_get_sync() would return >> -EACCES in such cases. >> So check if runtime pm is enabled before returning error from bam functions. >> >> Fixes: 5b4a68952a89 ("dmaengine: qcom: bam_dma: disable runtime pm on remote controlled") >> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org> >> --- >> drivers/dma/qcom/bam_dma.c | 10 +++++----- >> 1 file changed, 5 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c >> index d29275b97e84..5f4babebc508 100644 >> --- a/drivers/dma/qcom/bam_dma.c >> +++ b/drivers/dma/qcom/bam_dma.c >> @@ -540,7 +540,7 @@ static void bam_free_chan(struct dma_chan *chan) >> int ret; >> >> ret = pm_runtime_get_sync(bdev->dev); >> - if (ret < 0) >> + if (pm_runtime_enabled(bdev->dev) && ret < 0) > > would it make sense to first check enabled and do _get_sync() > Both of them check the disable_depth, however doing it the way you suggested would avoid some additional calls. > if (pm_runtime_enabled()) { > ret = pm_runtime_get_sync() { > ... > } > } sound like good idea! > Just noticed that some of the drivers use: err = pm_runtime_get_sync(dev); if (err < 0 && err !=-EACCES) return; others warp up this into local function. Will send a v2! thanks, srini > thus making clear in code that we do calls only when it is enabled. Also you can > add a local macro for this code and use that rather than copy pasting :) >
diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c index d29275b97e84..5f4babebc508 100644 --- a/drivers/dma/qcom/bam_dma.c +++ b/drivers/dma/qcom/bam_dma.c @@ -540,7 +540,7 @@ static void bam_free_chan(struct dma_chan *chan) int ret; ret = pm_runtime_get_sync(bdev->dev); - if (ret < 0) + if (pm_runtime_enabled(bdev->dev) && ret < 0) return; vchan_free_chan_resources(to_virt_chan(chan)); @@ -721,7 +721,7 @@ static int bam_pause(struct dma_chan *chan) int ret; ret = pm_runtime_get_sync(bdev->dev); - if (ret < 0) + if (pm_runtime_enabled(bdev->dev) && ret < 0) return ret; spin_lock_irqsave(&bchan->vc.lock, flag); @@ -747,7 +747,7 @@ static int bam_resume(struct dma_chan *chan) int ret; ret = pm_runtime_get_sync(bdev->dev); - if (ret < 0) + if (pm_runtime_enabled(bdev->dev) && ret < 0) return ret; spin_lock_irqsave(&bchan->vc.lock, flag); @@ -853,7 +853,7 @@ static irqreturn_t bam_dma_irq(int irq, void *data) tasklet_schedule(&bdev->task); ret = pm_runtime_get_sync(bdev->dev); - if (ret < 0) + if (pm_runtime_enabled(bdev->dev) && ret < 0) return ret; if (srcs & BAM_IRQ) { @@ -970,7 +970,7 @@ static void bam_start_dma(struct bam_chan *bchan) return; ret = pm_runtime_get_sync(bdev->dev); - if (ret < 0) + if (pm_runtime_enabled(bdev->dev) && ret < 0) return; while (vd && !IS_BUSY(bchan)) {
Disabling pm runtime at probe is not sufficient to get BAM working on remotely controller instances. pm_runtime_get_sync() would return -EACCES in such cases. So check if runtime pm is enabled before returning error from bam functions. Fixes: 5b4a68952a89 ("dmaengine: qcom: bam_dma: disable runtime pm on remote controlled") Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org> --- drivers/dma/qcom/bam_dma.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)