[PATCH 4/5] DMA: sun6i: Add driver for the Allwinner A31 DMA controller

[Maxime Ripard]

On Fri, Feb 28, 2014 at 03:33:11PM +0200, Andy Shevchenko wrote:
> On Fri, 2014-02-28 at 11:36 +0100, Maxime Ripard wrote:
> > Hi Andy,
> > 
> > On Tue, Feb 25, 2014 at 01:28:15PM +0200, Andy Shevchenko wrote:
> > > > +static irqreturn_t sun6i_dma_interrupt(int irq, void *dev_id)
> > > > +{
> > > > +	struct sun6i_dma_dev *sdev = (struct sun6i_dma_dev *)dev_id;
> > > > +	struct sun6i_vchan *vchan;
> > > > +	struct sun6i_pchan *pchan;
> > > > +	int i, j, ret = 0;
> > > > +	u32 status;
> > > > +
> > > > +	for (i = 0; i < 2; i++) {
> > > > +		status = readl(sdev->base + DMA_IRQ_STAT(i));
> > > > +		if (!status) {
> > > > +			ret |= IRQ_NONE;
> > > 
> > > Maybe move this to definition block.
> > > 
> > > > +			continue;
> > > > +		}
> > > > +
> > > > +		dev_dbg(sdev->slave.dev, "DMA irq status %s: 0x%x\n",
> > > > +			i ? "high" : "low", status);
> > > > +
> > > > +		writel(status, sdev->base + DMA_IRQ_STAT(i));
> > > > +
> > > > +		for (j = 0; (j < 8) && status; j++) {
> > > > +			if (status & DMA_IRQ_QUEUE) {
> > > > +				pchan = sdev->pchans + j;
> > > > +				vchan = pchan->vchan;
> > > > +
> > > > +				if (vchan) {
> > > > +					unsigned long flags;
> > > > +
> > > > +					spin_lock_irqsave(&vchan->vc.lock,
> > > > +							  flags);
> > > > +					vchan_cookie_complete(&pchan->desc->vd);
> > > > +					pchan->done = pchan->desc;
> > > > +					spin_unlock_irqrestore(&vchan->vc.lock,
> > > > +							       flags);
> > > > +				}
> > > > +			}
> > > > +
> > > > +			status = status >> 4;
> > > > +		}
> > > > +
> > > > +		ret |= IRQ_HANDLED;
> > > 
> > > In case one is handled, another is not, what you have to do?
> > 
> > The interrupt status is split across two registers. In the case where
> > one of the two register reports an interrupt, we still have to handle
> > our interrupt, we actually did, so we have to return IRQ_HANDLED.
> 
> You removed the code below this assignment, but if I remember correctly
> you check for exact value there.
> 
> In case of one is not handled and the other is handled you will have ret
> = IRQ_HANDLED | IRQ_NONE. Thus, your following code will not be
> executed. Is it by design?

The code that got removed was

if (ret == IRQ_HANDLED)
   tasklet_schedule()

return ret;

The only thing that wouldn't have been executed if we had no
interrupts to report was the tasklet_schedule.

This has lightly changed though in the v2, thanks to your comments.

I don't have the | anymore, and call tasklet_schedule directly in the
loop.

I'll send the v2 in a short while.

Maxime

-- 
Maxime Ripard, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20140228/da5afbc1/attachment.sig>