Hi Laurent,
On Tue, Jan 26, 2021 at 11:01 PM Laurent Pinchart
<laurent.pinchart@xxxxxxxxxxxxxxxx> wrote:
> On Mon, Jan 25, 2021 at 03:24:31PM +0100, Geert Uytterhoeven wrote:
> > The DMACs (both SYS-DMAC and RT-DMAC) on R-Car V3U differ slightly from
> > the DMACs on R-Car Gen2 and other R-Car Gen3 SoCs:
> > 1. The per-channel registers are located in a second register block.
> > Add support for mapping the second block, using the appropriate
> > offsets and stride.
> > 2. The common Channel Clear Register (DMACHCLR) was replaced by a
> > per-channel register.
> > Update rcar_dmac_chan_clear{,_all}() to handle this.
> > As rcar_dmac_init() needs to clear the status before the individual
> > channels are probed, channel index and base address initialization
> > are moved forward.
> >
> > Inspired by a patch in the BSP by Phong Hoang
> > <phong.hoang.wz@xxxxxxxxxxx>.
> >
> > Signed-off-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx>
> > --- a/drivers/dma/sh/rcar-dmac.c
> > +++ b/drivers/dma/sh/rcar-dmac.c
> > @@ -189,7 +189,8 @@ struct rcar_dmac_chan {
> > * struct rcar_dmac - R-Car Gen2 DMA Controller
> > * @engine: base DMA engine object
> > * @dev: the hardware device
> > - * @iomem: remapped I/O memory base
> > + * @dmac_base: remapped base register block
> > + * @chan_base: remapped channel register block (optional)
> > * @n_channels: number of available channels
> > * @channels: array of DMAC channels
> > * @channels_mask: bitfield of which DMA channels are managed by this driver
> > @@ -198,7 +199,8 @@ struct rcar_dmac_chan {
> > struct rcar_dmac {
> > struct dma_device engine;
> > struct device *dev;
> > - void __iomem *iomem;
> > + void __iomem *dmac_base;
> > + void __iomem *chan_base;
> >
> > unsigned int n_channels;
> > struct rcar_dmac_chan *channels;
> > @@ -339,12 +344,23 @@ static void rcar_dmac_chan_write(struct rcar_dmac_chan *chan, u32 reg, u32 data)
> > static void rcar_dmac_chan_clear(struct rcar_dmac *dmac,
> > struct rcar_dmac_chan *chan)
> > {
> > - rcar_dmac_write(dmac, RCAR_DMACHCLR, BIT(chan->index));
> > + if (dmac->chan_base)
>
> Using dmac->chan_base to check if the device is a V3U seems a bit of a
> hack (especially given that the field is otherwise unused). I'd prefer
> adding a model field to struct rcar_dmac_of_data and struct rcar_dmac.
The check is not a check for R-Car V3U in particular, but a check for
the presence of a separate register block for channel registers.
I expect to see more SoCs having this, so IMHO checking for this feature,
instead of checking a model field, makes sense.
It's indeed unused otherwise, as beyond probe(), where per-channel bases
are calculated, no access to this pointer is needed anymore, (you can
blame devm_*() for not needing the pointer ;-)
Note that a model field would be "otherwise unused", too ;-)
> > + rcar_dmac_chan_write(chan, RCAR_V3U_DMACHCLR, 1);
> > + else
> > + rcar_dmac_write(dmac, RCAR_DMACHCLR, BIT(chan->index));
> > }
> >
> > static void rcar_dmac_chan_clear_all(struct rcar_dmac *dmac)
> > {
> > - rcar_dmac_write(dmac, RCAR_DMACHCLR, dmac->channels_mask);
> > + struct rcar_dmac_chan *chan;
> > + unsigned int i;
> > +
> > + if (dmac->chan_base) {
> > + for_each_rcar_dmac_chan(i, chan, dmac)
> > + rcar_dmac_chan_write(chan, RCAR_V3U_DMACHCLR, 1);
> > + } else {
> > + rcar_dmac_write(dmac, RCAR_DMACHCLR, dmac->channels_mask);
> > + }
> > }
> >
> > /* -----------------------------------------------------------------------------
> > @@ -1744,7 +1760,6 @@ static const struct dev_pm_ops rcar_dmac_pm = {
> >
> > static int rcar_dmac_chan_probe(struct rcar_dmac *dmac,
> > struct rcar_dmac_chan *rchan,
> > - const struct rcar_dmac_of_data *data,
> > unsigned int index)
> > {
> > struct platform_device *pdev = to_platform_device(dmac->dev);
> > @@ -1753,9 +1768,6 @@ static int rcar_dmac_chan_probe(struct rcar_dmac *dmac,
> > char *irqname;
> > int ret;
> >
> > - rchan->index = index;
> > - rchan->iomem = dmac->iomem + data->chan_offset_base +
> > - data->chan_offset_stride * index;
> > rchan->mid_rid = -EINVAL;
> >
> > spin_lock_init(&rchan->lock);
> > @@ -1842,6 +1854,7 @@ static int rcar_dmac_probe(struct platform_device *pdev)
> > const struct rcar_dmac_of_data *data;
> > struct rcar_dmac_chan *chan;
> > struct dma_device *engine;
> > + void __iomem *chan_base;
> > struct rcar_dmac *dmac;
> > unsigned int i;
> > int ret;
> > @@ -1880,9 +1893,24 @@ static int rcar_dmac_probe(struct platform_device *pdev)
> > return -ENOMEM;
> >
> > /* Request resources. */
> > - dmac->iomem = devm_platform_ioremap_resource(pdev, 0);
> > - if (IS_ERR(dmac->iomem))
> > - return PTR_ERR(dmac->iomem);
> > + dmac->dmac_base = devm_platform_ioremap_resource(pdev, 0);
> > + if (IS_ERR(dmac->dmac_base))
> > + return PTR_ERR(dmac->dmac_base);
> > +
> > + if (!data->chan_offset_base) {
> > + dmac->chan_base = devm_platform_ioremap_resource(pdev, 1);
> > + if (IS_ERR(dmac->chan_base))
> > + return PTR_ERR(dmac->chan_base);
> > +
> > + chan_base = dmac->chan_base;
> > + } else {
> > + chan_base = dmac->dmac_base + data->chan_offset_base;
> > + }
> > +
> > + for_each_rcar_dmac_chan(i, chan, dmac) {
> > + chan->index = i;
>
> Now that chan->indew is set before calling rcar_dmac_chan_probe(), you
> don't have to pass the index to rcar_dmac_chan_probe() anymore.
Right, will fix.
> > + chan->iomem = chan_base + i * data->chan_offset_stride;
> > + }
> >
> > /* Enable runtime PM and initialize the device. */
> > pm_runtime_enable(&pdev->dev);
> > @@ -1929,7 +1957,7 @@ static int rcar_dmac_probe(struct platform_device *pdev)
> > INIT_LIST_HEAD(&engine->channels);
> >
> > for_each_rcar_dmac_chan(i, chan, dmac) {
> > - ret = rcar_dmac_chan_probe(dmac, chan, data, i);
> > + ret = rcar_dmac_chan_probe(dmac, chan, i);
> > if (ret < 0)
> > goto error;
> > }
Thanks for your comments!
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
