| Message ID | 1419928907-18053-3-git-send-email-jingchang.lu@freescale.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 2014-12-30 09:41, Jingchang Lu wrote: > This adds power management suspend/resume support for > the fsl-edma driver. > > Signed-off-by: Jingchang Lu <jingchang.lu@freescale.com> > --- > drivers/dma/fsl-edma.c | 38 ++++++++++++++++++++++++++++++++++++++ > 1 file changed, 38 insertions(+) > > diff --git a/drivers/dma/fsl-edma.c b/drivers/dma/fsl-edma.c > index e0bd517..6a81699 100644 > --- a/drivers/dma/fsl-edma.c > +++ b/drivers/dma/fsl-edma.c > @@ -1017,6 +1017,43 @@ static int fsl_edma_remove(struct platform_device *pdev) > return 0; > } > > +#ifdef CONFIG_PM_SLEEP > +static int fsl_edma_pm_suspend(struct device *dev) > +{ > + struct fsl_edma_engine *fsl_edma = dev_get_drvdata(dev); > + struct fsl_edma_chan *fsl_chan; > + int i; > + > + for (i = 0; i < fsl_edma->n_chans; i++) { > + fsl_chan = &fsl_edma->chans[i]; > + fsl_edma_chan_mux(fsl_chan, 0, false); > + } > + > + return 0; > +} > + > +static int fsl_edma_pm_resume(struct device *dev) > +{ > + struct fsl_edma_engine *fsl_edma = dev_get_drvdata(dev); > + struct fsl_edma_chan *fsl_chan; > + int i; > + > + for (i = 0; i < fsl_edma->n_chans; i++) { > + fsl_chan = &fsl_edma->chans[i]; > + edma_writew(fsl_edma, 0x0, fsl_edma->membase + EDMA_TCD_CSR(i)); What is the expectation of the suspend mode, are the registers and descriptors cleared during suspend? So this makes sure we have all descriptors disabled on wake-up, right? > + /* restore the channel slave id configuration */ > + fsl_edma_chan_mux(fsl_chan, fsl_chan->slave_id, true); And here the channel gets muxed again. But the user is expected to reconfigure a channel, e.g. using dmaengine_slave_config and the dmaengine_prep*? Also and cyclic DMA which has been paused before would be interrupted, is this correct? I don't know what the common exception of the API is, I'm just asking so I now what I can expect when using the DMA driver in other device drivers... -- Stefan > + } > + > + edma_writel(fsl_edma, EDMA_CR_ERGA | EDMA_CR_ERCA, > + fsl_edma->membase + EDMA_CR); > + > + return 0; > +} > +#endif > +static > +SIMPLE_DEV_PM_OPS(fsl_edma_pm_ops, fsl_edma_pm_suspend, fsl_edma_pm_resume); > + > static const struct of_device_id fsl_edma_dt_ids[] = { > { .compatible = "fsl,vf610-edma", }, > { /* sentinel */ } > @@ -1027,6 +1064,7 @@ static struct platform_driver fsl_edma_driver = { > .driver = { > .name = "fsl-edma", > .of_match_table = fsl_edma_dt_ids, > + .pm = &fsl_edma_pm_ops, > }, > .probe = fsl_edma_probe, > .remove = fsl_edma_remove,
On Tue, Dec 30, 2014 at 04:41:47PM +0800, Jingchang Lu wrote: > This adds power management suspend/resume support for > the fsl-edma driver. > > Signed-off-by: Jingchang Lu <jingchang.lu@freescale.com> > --- > drivers/dma/fsl-edma.c | 38 ++++++++++++++++++++++++++++++++++++++ > 1 file changed, 38 insertions(+) > > diff --git a/drivers/dma/fsl-edma.c b/drivers/dma/fsl-edma.c > index e0bd517..6a81699 100644 > --- a/drivers/dma/fsl-edma.c > +++ b/drivers/dma/fsl-edma.c > @@ -1017,6 +1017,43 @@ static int fsl_edma_remove(struct platform_device *pdev) > return 0; > } > > +#ifdef CONFIG_PM_SLEEP > +static int fsl_edma_pm_suspend(struct device *dev) > +{ > + struct fsl_edma_engine *fsl_edma = dev_get_drvdata(dev); > + struct fsl_edma_chan *fsl_chan; > + int i; > + > + for (i = 0; i < fsl_edma->n_chans; i++) { > + fsl_chan = &fsl_edma->chans[i]; > + fsl_edma_chan_mux(fsl_chan, 0, false); > + } > + > + return 0; > +} > + > +static int fsl_edma_pm_resume(struct device *dev) > +{ > + struct fsl_edma_engine *fsl_edma = dev_get_drvdata(dev); > + struct fsl_edma_chan *fsl_chan; > + int i; > + > + for (i = 0; i < fsl_edma->n_chans; i++) { > + fsl_chan = &fsl_edma->chans[i]; > + edma_writew(fsl_edma, 0x0, fsl_edma->membase + EDMA_TCD_CSR(i)); > + /* restore the channel slave id configuration */ > + fsl_edma_chan_mux(fsl_chan, fsl_chan->slave_id, true); Wouldnt this be reconfigured? Should you do this for all channels or active ones. Also I dont see you pausing or suspending active channels or checking for them?
diff --git a/drivers/dma/fsl-edma.c b/drivers/dma/fsl-edma.c index e0bd517..6a81699 100644 --- a/drivers/dma/fsl-edma.c +++ b/drivers/dma/fsl-edma.c @@ -1017,6 +1017,43 @@ static int fsl_edma_remove(struct platform_device *pdev) return 0; } +#ifdef CONFIG_PM_SLEEP +static int fsl_edma_pm_suspend(struct device *dev) +{ + struct fsl_edma_engine *fsl_edma = dev_get_drvdata(dev); + struct fsl_edma_chan *fsl_chan; + int i; + + for (i = 0; i < fsl_edma->n_chans; i++) { + fsl_chan = &fsl_edma->chans[i]; + fsl_edma_chan_mux(fsl_chan, 0, false); + } + + return 0; +} + +static int fsl_edma_pm_resume(struct device *dev) +{ + struct fsl_edma_engine *fsl_edma = dev_get_drvdata(dev); + struct fsl_edma_chan *fsl_chan; + int i; + + for (i = 0; i < fsl_edma->n_chans; i++) { + fsl_chan = &fsl_edma->chans[i]; + edma_writew(fsl_edma, 0x0, fsl_edma->membase + EDMA_TCD_CSR(i)); + /* restore the channel slave id configuration */ + fsl_edma_chan_mux(fsl_chan, fsl_chan->slave_id, true); + } + + edma_writel(fsl_edma, EDMA_CR_ERGA | EDMA_CR_ERCA, + fsl_edma->membase + EDMA_CR); + + return 0; +} +#endif +static +SIMPLE_DEV_PM_OPS(fsl_edma_pm_ops, fsl_edma_pm_suspend, fsl_edma_pm_resume); + static const struct of_device_id fsl_edma_dt_ids[] = { { .compatible = "fsl,vf610-edma", }, { /* sentinel */ } @@ -1027,6 +1064,7 @@ static struct platform_driver fsl_edma_driver = { .driver = { .name = "fsl-edma", .of_match_table = fsl_edma_dt_ids, + .pm = &fsl_edma_pm_ops, }, .probe = fsl_edma_probe, .remove = fsl_edma_remove,
This adds power management suspend/resume support for the fsl-edma driver. Signed-off-by: Jingchang Lu <jingchang.lu@freescale.com> --- drivers/dma/fsl-edma.c | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+)