[1/2] dma: axi-dmac: don't check the number of frames for alignment

["Ardelean, Alexandru" <Alex.Ardelean@analog.com>]

On Tue, 2019-02-26 at 13:31 +0530, Vinod Koul wrote:
> [External]
> 
> 
> On 26-02-19, 07:14, Ardelean, Alexandru wrote:
> > On Mon, 2019-02-25 at 12:22 +0530, Vinod Koul wrote:
> > > [External]
> > > 
> > > 
> > > On 15-02-19, 11:17, Alexandru Ardelean wrote:
> > > > Fixes commit 0e3b67b348b8 ("dmaengine: Add support for the Analog
> > > > Devices
> > > > AXI-DMAC DMA controller")
> > > 
> > > Do you mean to add a Fixes tag?
> > 
> > Yep.
> > I'm terrible at this apparently. I'll read about how to do that
> > properly.
> > Other maintainers have complained about this as well [i.e. the fact
> > that I
> > didn't add proper Fixes tags ].
> 
> So if it fixes a bug in original commit and should go immediately to
> next rc-X then fixes is mostly appropriate.. (also backporting to
> stable). In this case it doesnt seem so.
> 
> Also canonical form is
> 
> Fixes: abcdef : ("buggy patch")
> 
> and this line should be before the s-o-b line..

I just found this in the submitting-patch-docs.

> 
> > > 
> > > > For 2D transfers, there is no requirement for Y_LENGTH to be
> > > > aligned
> > > > to the bus-width (or anything). X_LENGTH is required to be aligned
> > > > though.
> > > > 
> > > > So, we shouldn't check that the number of frames is aligned.
> > > 
> > > Does this fix a bug as indicated by Fixes tag?
> > 
> > Yes
> 
> well if it is aligned it shouldn't cause break right. Yes not having
> aligned helps the driver but seems to be correct the wrong
> interpretation/implementation.. right?

I'm preparing a V2 here.
2D transfers are typically used when DMA talks to video, where Y_LENGTH is
the number of rows, which don't need to be aligned.
I'll try to make the comment more helpful.

> 
> --
> ~Vinod