[1/1] dma: edma: fix incorrect SG list handling

Commit Message

Sekhar Nori March 19, 2014, 5:55 a.m. UTC

The code to handle any length SG lists calls edma_resume()
even before edma_start() is called. This is incorrect
because edma_resume() enables edma events on the channel
after which CPU (in edma_start) cannot clear posted
events by writing to ECR (per the EDMA user's guide).

Because of this EDMA transfers fail to start if due
to some reason there is a pending EDMA event registered
even before EDMA transfers are started. This can happen if
an EDMA event is a byproduct of device initialization.

Fix this by calling edma_resume() only if it is not the
first batch of MAX_NR_SG elements.

Without this patch, MMC/SD fails to function on DA850 EVM
with DMA. The behaviour is triggered by specific IP and
this can explain why the issue was not reported before
(example with MMC/SD on AM335x).

Tested on DA850 EVM and AM335x EVM-SK using MMC/SD card.

Cc: [e-mail] # v3.12.x+
Cc: Joel Fernandes <[e-mail]>
Acked-by: Joel Fernandes <[e-mail]>
Tested-by: Jon Ringle <[e-mail]>
Tested-by: Alexander Holler <[e-mail]>
Reported-by: Jon Ringle <[e-mail]>
Signed-off-by: Sekhar Nori <[e-mail]>
---
Same as the version send previously (which did not make
into all mailing lists). Commit message updates with
tested-by and acks received.

 drivers/dma/edma.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

Comments

Sekhar Nori April 14, 2014, 8:31 a.m. UTC | #1

Vinod,

On Wednesday 19 March 2014 11:25 AM, Sekhar Nori wrote:
> The code to handle any length SG lists calls edma_resume()
> even before edma_start() is called. This is incorrect
> because edma_resume() enables edma events on the channel
> after which CPU (in edma_start) cannot clear posted
> events by writing to ECR (per the EDMA user's guide).
> 
> Because of this EDMA transfers fail to start if due
> to some reason there is a pending EDMA event registered
> even before EDMA transfers are started. This can happen if
> an EDMA event is a byproduct of device initialization.
> 
> Fix this by calling edma_resume() only if it is not the
> first batch of MAX_NR_SG elements.
> 
> Without this patch, MMC/SD fails to function on DA850 EVM
> with DMA. The behaviour is triggered by specific IP and
> this can explain why the issue was not reported before
> (example with MMC/SD on AM335x).
> 
> Tested on DA850 EVM and AM335x EVM-SK using MMC/SD card.
> 
> Cc: [e-mail] # v3.12.x+
> Cc: Joel Fernandes <[e-mail]>
> Acked-by: Joel Fernandes <[e-mail]>
> Tested-by: Jon Ringle <[e-mail]>
> Tested-by: Alexander Holler <[e-mail]>
> Reported-by: Jon Ringle <[e-mail]>
> Signed-off-by: Sekhar Nori <[e-mail]>

Looks like this patch is not in mainline still?

Thanks,
Sekhar

Vinod Koul April 14, 2014, 8:57 a.m. UTC | #2

On Mon, Apr 14, 2014 at 02:01:11PM +0530, Sekhar Nori wrote:
> Vinod,
> 
> On Wednesday 19 March 2014 11:25 AM, Sekhar Nori wrote:
> > The code to handle any length SG lists calls edma_resume()
> > even before edma_start() is called. This is incorrect
> > because edma_resume() enables edma events on the channel
> > after which CPU (in edma_start) cannot clear posted
> > events by writing to ECR (per the EDMA user's guide).
> > 
> > Because of this EDMA transfers fail to start if due
> > to some reason there is a pending EDMA event registered
> > even before EDMA transfers are started. This can happen if
> > an EDMA event is a byproduct of device initialization.
> > 
> > Fix this by calling edma_resume() only if it is not the
> > first batch of MAX_NR_SG elements.
> > 
> > Without this patch, MMC/SD fails to function on DA850 EVM
> > with DMA. The behaviour is triggered by specific IP and
> > this can explain why the issue was not reported before
> > (example with MMC/SD on AM335x).
> > 
> > Tested on DA850 EVM and AM335x EVM-SK using MMC/SD card.
> > 
> > Cc: [e-mail] # v3.12.x+
> > Cc: Joel Fernandes <[e-mail]>
> > Acked-by: Joel Fernandes <[e-mail]>
> > Tested-by: Jon Ringle <[e-mail]>
> > Tested-by: Alexander Holler <[e-mail]>
> > Reported-by: Jon Ringle <[e-mail]>
> > Signed-off-by: Sekhar Nori <[e-mail]>
> 
> Looks like this patch is not in mainline still?

Sorry looks like I have missed sending the email. I had applied it last week and
today rebased after rc1. It would be part of rc2...

Sekhar Nori April 14, 2014, 9:09 a.m. UTC | #3

On Monday 14 April 2014 02:27 PM, Vinod Koul wrote:
> On Mon, Apr 14, 2014 at 02:01:11PM +0530, Sekhar Nori wrote:
>> Vinod,
>>
>> On Wednesday 19 March 2014 11:25 AM, Sekhar Nori wrote:
>>> The code to handle any length SG lists calls edma_resume()
>>> even before edma_start() is called. This is incorrect
>>> because edma_resume() enables edma events on the channel
>>> after which CPU (in edma_start) cannot clear posted
>>> events by writing to ECR (per the EDMA user's guide).
>>>
>>> Because of this EDMA transfers fail to start if due
>>> to some reason there is a pending EDMA event registered
>>> even before EDMA transfers are started. This can happen if
>>> an EDMA event is a byproduct of device initialization.
>>>
>>> Fix this by calling edma_resume() only if it is not the
>>> first batch of MAX_NR_SG elements.
>>>
>>> Without this patch, MMC/SD fails to function on DA850 EVM
>>> with DMA. The behaviour is triggered by specific IP and
>>> this can explain why the issue was not reported before
>>> (example with MMC/SD on AM335x).
>>>
>>> Tested on DA850 EVM and AM335x EVM-SK using MMC/SD card.
>>>
>>> Cc: [e-mail] # v3.12.x+
>>> Cc: Joel Fernandes <[e-mail]>
>>> Acked-by: Joel Fernandes <[e-mail]>
>>> Tested-by: Jon Ringle <[e-mail]>
>>> Tested-by: Alexander Holler <[e-mail]>
>>> Reported-by: Jon Ringle <[e-mail]>
>>> Signed-off-by: Sekhar Nori <[e-mail]>
>>
>> Looks like this patch is not in mainline still?
> 
> Sorry looks like I have missed sending the email. I had applied it last week and
> today rebased after rc1. It would be part of rc2...

Thank you!

Regards,
Sekhar

Patch

diff --git a/drivers/dma/edma.c b/drivers/dma/edma.c
index cd8da45..bf5ad0f 100644
--- a/drivers/dma/edma.c
+++ b/drivers/dma/edma.c
@@ -182,11 +182,13 @@  static void edma_execute(struct edma_chan *echan)
 				  echan->ecc->dummy_slot);
 	}
 
-	edma_resume(echan->ch_num);
-
 	if (edesc->processed <= MAX_NR_SG) {
 		dev_dbg(dev, "first transfer starting %d\n", echan->ch_num);
 		edma_start(echan->ch_num);
+	} else {
+		dev_dbg(dev, "chan: %d: completed %d elements, resuming\n",
+			echan->ch_num, edesc->processed);
+		edma_resume(echan->ch_num);
 	}
 
 	/*