RE: [PATCH v2] dmaengine: vdma: Add 64 bit addressing support to the driver

From: Appana Durga Kedareswara Rao
Date: Fri Mar 25 2016 - 05:17:17 EST


Hi Laurent Pinchart,

> -----Original Message-----
> From: Laurent Pinchart [mailto:laurent.pinchart@xxxxxxxxxxxxxxxx]
> Sent: Monday, March 21, 2016 9:48 PM
> To: Anurag Kumar Vulisha
> Cc: Vinod Koul; Rob Herring; Pawel Moll; Mark Rutland; Ian Campbell; Kumar
> Gala; Michal Simek; Soren Brinkmann; Dan Williams; afaerber@xxxxxxx; Maxime
> Ripard; Appana Durga Kedareswara Rao; Anirudha Sarangi; Srikanth Vemula;
> devicetree@xxxxxxxxxxxxxxx; linux-arm-kernel@xxxxxxxxxxxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx; dmaengine@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH v2] dmaengine: vdma: Add 64 bit addressing support to the
> driver
>
> Hi Anurag,
>
> On Wednesday 23 Sep 2015 15:12:36 Anurag Kumar Vulisha wrote:
> > On Monday, September 21, 2015 9:27 PM Vinod Koul wrote:
> > > On Thu, Aug 27, 2015 at 09:19:18PM +0530, Anurag Kumar Vulisha wrote:
> > >> This VDMA is a soft ip, which can be programmed to support
> > >> 32 bit addressing or greater than 32 bit addressing.
> > >>
> > >> When the VDMA ip is configured for 32 bit address space the
> > >> transfer start address is specified by a single register.
> > >
> > > would be good to specfiy which one
> >
> > Will change this in v3
>
> What happened to v3 ? :-)

I have sent it today.

Thanks,
Kedar.

>
> > >> When the VDMA core is configured for an address space greater than
> > >> 32 then each start address is specified by a combination of two
> > >> registers. The first register specifies the LSB 32 bits of address,
> > >> while the next register specifies the MSB 32 bits of address.For
> > >> example,5Ch will specify the LSB 32 bits while 60h will specify the
> > >> MSB 32 bits of the first start address.So we need to program two
> > >> registers at a time.
> > >
> > > can we have spaces after full stops and commas!
> >
> > Will take care of this in v3 patch.
> >
> > >> +/* Since vdma driver is trying to write to a register offset which
> > >> +is not a
> > >> + * multiple of 64 bits(ex : 0x5c), we are writing as two separate
> > >> +32 bits
> > >> + * instead of a single 64 bit register write.
> > >> + */
> > >
> > > This is not kernel style for multi-lines, pls refer to
> > > Documentation/CodingStyle
> >
> > Will address this in v3 patch
> >
> > >> +
> > >> +static inline void vdma_desc_write_64(struct xilinx_vdma_chan
> > >> +*chan,
> > >> u32 reg,
> > >> + u32 value_lsb, u32 value_msb) {
> > >> + /* Write the lsb 32 bits*/
> > >> + writel(value_lsb, chan->xdev->regs + chan->desc_offset + reg);
> > >> +
> > >> + /* Write the msb 32 bits */
> > >> + writel(value_msb, chan->xdev->regs + chan->desc_offset + reg +
> > >> + 4);
> > >
> > > why not writeq
> >
> > We are trying to write at a register address(ex:0x5c) which is not
> > aligned on 8 bytes boundary.So if I try to use 64 bit write on
> > it,unalignment fault is getting generated.To avoid that we are using
> > two separate 32 bit writes. We had this discussion in previous
> > versions of this patch with Laurent Pinchart .I have also added this
> > exaplanation in the comments above this function.
> >
> > >> + err = of_property_read_u32(node, "xlnx,addrwidth",
> > >> + &addr_width);
> > >> +
> > >> + if (err < 0) {
> > >> + /* Setting addr_width property to default 32 bits */
> > >> + addr_width = 32;
> > >> + }
> > >
> > > braces for a single line statement! Also space is redandant before
> > > if condition
> >
> > Will take care of this in v3 patch
>
> --
> Regards,
>
> Laurent Pinchart