| Message ID | 1408693562-18068-1-git-send-email-d-gerlach@ti.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Friday 22 August 2014 01:16 PM, Dave Gerlach wrote: > From: Daniel Mack <zonque@gmail.com> > > This patch makes the edma driver resume correctly after suspend. Tested > on an AM33xx platform with cyclic audio streams and omap_hsmmc. > > All information can be reconstructed by already known runtime > information. > > As we now use some functions that were previously only used from __init > context, annotations had to be dropped. > > [nm@ti.com: added error handling for runtime + suspend_late/early_resume] > Signed-off-by: Nishanth Menon <nm@ti.com> > Signed-off-by: Daniel Mack <zonque@gmail.com> > Tested-by: Joel Fernandes <joelf@ti.com> > Acked-by: Joel Fernandes <joelf@ti.com> > [d-gerlach@ti.com: updated to remove queue_tc_mapping use] > Signed-off-by: Dave Gerlach <d-gerlach@ti.com> > --- > > Needed for am335x suspend, but never got picked up. Previously > sent here: http://marc.info/?l=linux-arm-kernel&m=138556067416051&w=2 > > arch/arm/common/edma.c | 86 ++++++++++++++++++++++++++++++++++++++++++++++++-- > 1 file changed, 84 insertions(+), 2 deletions(-) > > diff --git a/arch/arm/common/edma.c b/arch/arm/common/edma.c > index 8809917..ece1e3d 100644 > --- a/arch/arm/common/edma.c > +++ b/arch/arm/common/edma.c > @@ -244,6 +244,8 @@ struct edma { > /* list of channels with no even trigger; terminated by "-1" */ > const s8 *noevent; > > + struct edma_soc_info *info; > + > /* The edma_inuse bit for each PaRAM slot is clear unless the > * channel is in use ... by ARM or DSP, for QDMA, or whatever. > */ > @@ -295,7 +297,7 @@ static void map_dmach_queue(unsigned ctlr, unsigned ch_no, > ~(0x7 << bit), queue_no << bit); > } > > -static void __init assign_priority_to_queue(unsigned ctlr, int queue_no, > +static void assign_priority_to_queue(unsigned ctlr, int queue_no, > int priority) > { > int bit = queue_no * 4; > @@ -314,7 +316,7 @@ static void __init assign_priority_to_queue(unsigned ctlr, int queue_no, > * included in that particular EDMA variant (Eg : dm646x) > * > */ > -static void __init map_dmach_param(unsigned ctlr) > +static void map_dmach_param(unsigned ctlr) > { > int i; > for (i = 0; i < EDMA_MAX_DMACH; i++) > @@ -1791,15 +1793,95 @@ static int edma_probe(struct platform_device *pdev) > edma_write_array2(j, EDMA_DRAE, i, 1, 0x0); > edma_write_array(j, EDMA_QRAE, i, 0x0); > } > + edma_cc[j]->info = info[j]; > arch_num_cc++; > } > > return 0; > } > > +static int edma_pm_suspend(struct device *dev) > +{ > + int j, r; > + > + r = pm_runtime_get_sync(dev); > + if (r < 0) { > + dev_err(dev, "%s: get_sync returned %d\n", __func__, r); > + return r; > + } The driver currently does a pm_runtime_get_sync() once during probe. And does not do a put(). So this should actually be not required. In fact looks like this additional get() call will prevent the clock from getting disabled which is probably not what you intend. > + > + for (j = 0; j < arch_num_cc; j++) { > + struct edma *ecc = edma_cc[j]; > + > + disable_irq(ecc->irq_res_start); > + disable_irq(ecc->irq_res_end); Do we really need to disable these irqs? > + } > + > + pm_runtime_put_sync(dev); > + > + return 0; > +} > + > +static int edma_pm_resume(struct device *dev) > +{ > + int i, j, r; > + > + r = pm_runtime_get_sync(dev); > + if (r < 0) { > + dev_err(dev, "%s: get_sync returned %d\n", __func__, r); > + return r; > + } > + > + for (j = 0; j < arch_num_cc; j++) { > + struct edma *cc = edma_cc[j]; > + > + s8 (*queue_priority_mapping)[2]; > + > + queue_priority_mapping = cc->info->queue_priority_mapping; > + > + /* Event queue priority mapping */ > + for (i = 0; queue_priority_mapping[i][0] != -1; i++) > + assign_priority_to_queue(j, > + queue_priority_mapping[i][0], > + queue_priority_mapping[i][1]); > + > + /* > + * Map the channel to param entry if channel mapping logic > + * exist > + */ > + if (edma_read(j, EDMA_CCCFG) & CHMAP_EXIST) > + map_dmach_param(j); > + > + for (i = 0; i < cc->num_channels; i++) { > + if (test_bit(i, cc->edma_inuse)) { > + /* ensure access through shadow region 0 */ > + edma_or_array2(j, EDMA_DRAE, 0, i >> 5, > + BIT(i & 0x1f)); > + > + setup_dma_interrupt(i, > + cc->intr_data[i].callback, > + cc->intr_data[i].data); > + } > + } > + > + enable_irq(cc->irq_res_start); > + enable_irq(cc->irq_res_end); > + } > + > + pm_runtime_put_sync(dev); > + > + return 0; > +} > + > +static const struct dev_pm_ops edma_pm_ops = { > + .suspend_late = edma_pm_suspend, > + .resume_early = edma_pm_resume, > +}; You can use SET_LATE_SYSTEM_SLEEP_PM_OPS() as some other DMA drivers are doing too. Thanks, Sekhar -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi, On 08/26/2014 08:36 AM, Sekhar Nori wrote: > On Friday 22 August 2014 01:16 PM, Dave Gerlach wrote: Thanks for pushing that forward! >> +static int edma_pm_suspend(struct device *dev) >> +{ >> + int j, r; >> + >> + r = pm_runtime_get_sync(dev); >> + if (r < 0) { >> + dev_err(dev, "%s: get_sync returned %d\n", __func__, r); >> + return r; >> + } > > The driver currently does a pm_runtime_get_sync() once during probe. And > does not do a put(). So this should actually be not required. In fact > looks like this additional get() call will prevent the clock from > getting disabled which is probably not what you intend. Well, the pm runtime is put again ... >> + >> + for (j = 0; j < arch_num_cc; j++) { >> + struct edma *ecc = edma_cc[j]; >> + >> + disable_irq(ecc->irq_res_start); >> + disable_irq(ecc->irq_res_end); > > Do we really need to disable these irqs? > >> + } >> + >> + pm_runtime_put_sync(dev); ... here, so it's in sync and should be fine. I was also sure than when I wrote the code, disabling the interrupts during suspend was necessary, and even the only thing that has to be done at suspend time. Now that I address this again, my tests show that in can in fact be omitted. So I'll send a v9 now that has no edma_pm_suspend() at all anymore. >> +static const struct dev_pm_ops edma_pm_ops = { >> + .suspend_late = edma_pm_suspend, >> + .resume_early = edma_pm_resume, >> +}; > > You can use SET_LATE_SYSTEM_SLEEP_PM_OPS() as some other DMA drivers are > doing too. Sure, why not. Thanks, Daniel -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tuesday 26 August 2014 02:16 PM, Daniel Mack wrote: > Hi, > > On 08/26/2014 08:36 AM, Sekhar Nori wrote: >> On Friday 22 August 2014 01:16 PM, Dave Gerlach wrote: > > Thanks for pushing that forward! > >>> +static int edma_pm_suspend(struct device *dev) >>> +{ >>> + int j, r; >>> + >>> + r = pm_runtime_get_sync(dev); >>> + if (r < 0) { >>> + dev_err(dev, "%s: get_sync returned %d\n", __func__, r); >>> + return r; >>> + } >> >> The driver currently does a pm_runtime_get_sync() once during probe. And >> does not do a put(). So this should actually be not required. In fact >> looks like this additional get() call will prevent the clock from >> getting disabled which is probably not what you intend. > > Well, the pm runtime is put again ... > >>> + >>> + for (j = 0; j < arch_num_cc; j++) { >>> + struct edma *ecc = edma_cc[j]; >>> + >>> + disable_irq(ecc->irq_res_start); >>> + disable_irq(ecc->irq_res_end); >> >> Do we really need to disable these irqs? >> >>> + } >>> + >>> + pm_runtime_put_sync(dev); > > ... here, so it's in sync and should be fine. May be I am missing something but because of the pm_runtime_get_sync() in probe() usage count is already 1 when suspend() is called. The pm_runtime_get_sync() in this function makes it 2 and therefore pm_runtime_put_sync() returns immediately because the usage count is greater that 0 after decrementing by 1. That means the module's clocks is not disabled after suspend() is finished. > > I was also sure than when I wrote the code, disabling the interrupts > during suspend was necessary, and even the only thing that has to be > done at suspend time. Now that I address this again, my tests show that > in can in fact be omitted. Thanks! Regards, Sekhar -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/arch/arm/common/edma.c b/arch/arm/common/edma.c index 8809917..ece1e3d 100644 --- a/arch/arm/common/edma.c +++ b/arch/arm/common/edma.c @@ -244,6 +244,8 @@ struct edma { /* list of channels with no even trigger; terminated by "-1" */ const s8 *noevent; + struct edma_soc_info *info; + /* The edma_inuse bit for each PaRAM slot is clear unless the * channel is in use ... by ARM or DSP, for QDMA, or whatever. */ @@ -295,7 +297,7 @@ static void map_dmach_queue(unsigned ctlr, unsigned ch_no, ~(0x7 << bit), queue_no << bit); } -static void __init assign_priority_to_queue(unsigned ctlr, int queue_no, +static void assign_priority_to_queue(unsigned ctlr, int queue_no, int priority) { int bit = queue_no * 4; @@ -314,7 +316,7 @@ static void __init assign_priority_to_queue(unsigned ctlr, int queue_no, * included in that particular EDMA variant (Eg : dm646x) * */ -static void __init map_dmach_param(unsigned ctlr) +static void map_dmach_param(unsigned ctlr) { int i; for (i = 0; i < EDMA_MAX_DMACH; i++) @@ -1791,15 +1793,95 @@ static int edma_probe(struct platform_device *pdev) edma_write_array2(j, EDMA_DRAE, i, 1, 0x0); edma_write_array(j, EDMA_QRAE, i, 0x0); } + edma_cc[j]->info = info[j]; arch_num_cc++; } return 0; } +static int edma_pm_suspend(struct device *dev) +{ + int j, r; + + r = pm_runtime_get_sync(dev); + if (r < 0) { + dev_err(dev, "%s: get_sync returned %d\n", __func__, r); + return r; + } + + for (j = 0; j < arch_num_cc; j++) { + struct edma *ecc = edma_cc[j]; + + disable_irq(ecc->irq_res_start); + disable_irq(ecc->irq_res_end); + } + + pm_runtime_put_sync(dev); + + return 0; +} + +static int edma_pm_resume(struct device *dev) +{ + int i, j, r; + + r = pm_runtime_get_sync(dev); + if (r < 0) { + dev_err(dev, "%s: get_sync returned %d\n", __func__, r); + return r; + } + + for (j = 0; j < arch_num_cc; j++) { + struct edma *cc = edma_cc[j]; + + s8 (*queue_priority_mapping)[2]; + + queue_priority_mapping = cc->info->queue_priority_mapping; + + /* Event queue priority mapping */ + for (i = 0; queue_priority_mapping[i][0] != -1; i++) + assign_priority_to_queue(j, + queue_priority_mapping[i][0], + queue_priority_mapping[i][1]); + + /* + * Map the channel to param entry if channel mapping logic + * exist + */ + if (edma_read(j, EDMA_CCCFG) & CHMAP_EXIST) + map_dmach_param(j); + + for (i = 0; i < cc->num_channels; i++) { + if (test_bit(i, cc->edma_inuse)) { + /* ensure access through shadow region 0 */ + edma_or_array2(j, EDMA_DRAE, 0, i >> 5, + BIT(i & 0x1f)); + + setup_dma_interrupt(i, + cc->intr_data[i].callback, + cc->intr_data[i].data); + } + } + + enable_irq(cc->irq_res_start); + enable_irq(cc->irq_res_end); + } + + pm_runtime_put_sync(dev); + + return 0; +} + +static const struct dev_pm_ops edma_pm_ops = { + .suspend_late = edma_pm_suspend, + .resume_early = edma_pm_resume, +}; + static struct platform_driver edma_driver = { .driver = { .name = "edma", + .pm = &edma_pm_ops, .of_match_table = edma_of_ids, }, .probe = edma_probe,