Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1280701 > unrolled thread

Re: [RFC v02 01/15] dmaengine: core: Allow NULL mask pointer in __dma_device_satisfies_mask()

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [RFC v02 01/15] dmaengine: core: Allow NULL mask pointer in  __dma_device_satisfies_mask() Peter Ujfalusi <peter.ujfalusi@ti.com> - 2015-12-01 10:50 +0100
    Re: [RFC v02 01/15] dmaengine: core: Allow NULL mask pointer in __dma_device_satisfies_mask() Andy Shevchenko <andy.shevchenko@gmail.com> - 2015-12-01 14:00 +0100
      Re: [RFC v02 01/15] dmaengine: core: Allow NULL mask pointer in  __dma_device_satisfies_mask() Vinod Koul <vinod.koul@intel.com> - 2015-12-01 18:10 +0100

#1280701 — Re: [RFC v02 01/15] dmaengine: core: Allow NULL mask pointer in __dma_device_satisfies_mask()

On 11/30/2015 04:35 PM, Andy Shevchenko wrote:
> On Mon, Nov 30, 2015 at 3:45 PM, Peter Ujfalusi <peter.ujfalusi@ti.com> wrote:
>> Treat as true condition the case when the mask is NULL.
> 
> What do you think about setting some default (all "on") mask when mask
> is not supplied?

Probably rephrasing the commit message to say that when the mask is NULL it
means that the caller does not care about the capabilities of the dma device
thus return with true in such a case.

We could also drop this patch and in private_candidate() :

-	if (!__dma_device_satisfies_mask(dev, mask)) {
+	if (mask && !__dma_device_satisfies_mask(dev, mask)) {
		pr_debug("%s: wrong capabilities\n", __func__);
		return NULL;
	}


> I don't know for sure but there might be cases when you don't want
> literally *any* channel to satisfy.

Or set DMA_SLAVE only in dma_request_chan()? What happens if we have cases
when we are able to request channel for memcpy via dma_request_chan()
(dedicated memcpy channel/DMA engine?) in that case we will have the SLAVE
set, but not MEMCPY, or any other variation we do not know yet?

>>
>> Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
>> ---
>>  drivers/dma/dmaengine.c | 3 +++
>>  1 file changed, 3 insertions(+)
>>
>> diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c
>> index daf54a39bcc7..52c3eee48e2e 100644
>> --- a/drivers/dma/dmaengine.c
>> +++ b/drivers/dma/dmaengine.c
>> @@ -184,6 +184,9 @@ __dma_device_satisfies_mask(struct dma_device *device,
>>  {
>>         dma_cap_mask_t has;
>>
>> +       if (!want)
>> +               return true;
>> +
>>         bitmap_and(has.bits, want->bits, device->cap_mask.bits,
>>                 DMA_TX_TYPE_END);
>>         return bitmap_equal(want->bits, has.bits, DMA_TX_TYPE_END);
>> --
>> 2.6.3
>>
> 
> 
> 


-- 
Péter
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1280833 — Re: [RFC v02 01/15] dmaengine: core: Allow NULL mask pointer in __dma_device_satisfies_mask()

On Tue, Dec 1, 2015 at 11:47 AM, Peter Ujfalusi <peter.ujfalusi@ti.com> wrote:
> On 11/30/2015 04:35 PM, Andy Shevchenko wrote:
>> On Mon, Nov 30, 2015 at 3:45 PM, Peter Ujfalusi <peter.ujfalusi@ti.com> wrote:
>>> Treat as true condition the case when the mask is NULL.
>>
>> What do you think about setting some default (all "on") mask when mask
>> is not supplied?
>
> Probably rephrasing the commit message to say that when the mask is NULL it
> means that the caller does not care about the capabilities of the dma device
> thus return with true in such a case.
>
> We could also drop this patch and in private_candidate() :
>
> -       if (!__dma_device_satisfies_mask(dev, mask)) {
> +       if (mask && !__dma_device_satisfies_mask(dev, mask)) {
>                 pr_debug("%s: wrong capabilities\n", __func__);
>                 return NULL;
>         }

Between patch and above proposal I would choose the latter one.

>> I don't know for sure but there might be cases when you don't want
>> literally *any* channel to satisfy.
>
> Or set DMA_SLAVE only in dma_request_chan()? What happens if we have cases
> when we are able to request channel for memcpy via dma_request_chan()
> (dedicated memcpy channel/DMA engine?) in that case we will have the SLAVE
> set, but not MEMCPY, or any other variation we do not know yet?

Frankly, have no idea.

-- 
With Best Regards,
Andy Shevchenko
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1281049

On Tue, Dec 01, 2015 at 02:58:35PM +0200, Andy Shevchenko wrote:
> On Tue, Dec 1, 2015 at 11:47 AM, Peter Ujfalusi <peter.ujfalusi@ti.com> wrote:
> > On 11/30/2015 04:35 PM, Andy Shevchenko wrote:
> >> On Mon, Nov 30, 2015 at 3:45 PM, Peter Ujfalusi <peter.ujfalusi@ti.com> wrote:
> >>> Treat as true condition the case when the mask is NULL.
> >>
> >> What do you think about setting some default (all "on") mask when mask
> >> is not supplied?
> >
> > Probably rephrasing the commit message to say that when the mask is NULL it
> > means that the caller does not care about the capabilities of the dma device
> > thus return with true in such a case.
> >
> > We could also drop this patch and in private_candidate() :
> >
> > -       if (!__dma_device_satisfies_mask(dev, mask)) {
> > +       if (mask && !__dma_device_satisfies_mask(dev, mask)) {
> >                 pr_debug("%s: wrong capabilities\n", __func__);
> >                 return NULL;
> >         }
> 
> Between patch and above proposal I would choose the latter one.

Sounds better to me as well

> 
> >> I don't know for sure but there might be cases when you don't want
> >> literally *any* channel to satisfy.
> >
> > Or set DMA_SLAVE only in dma_request_chan()? What happens if we have cases
> > when we are able to request channel for memcpy via dma_request_chan()
> > (dedicated memcpy channel/DMA engine?) in that case we will have the SLAVE
> > set, but not MEMCPY, or any other variation we do not know yet?
> 
> Frankly, have no idea.

In slave cases I know that some controllers support memcpy but they are not
generic memcpy as they cannot be used for system memcpy but for 'special'
memcpy. So this can be used for memcpy as well

-- 
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web