Re: [PATCH v3 08/15] dmaengine: dw: revisit data_width property

From: Andy Shevchenko
Date: Wed Apr 13 2016 - 12:05:32 EST

Next message: kbuild test robot: "Re: [PATCH 1/2] [media] atmel-isc: add the Image Sensor Controller code"
Previous message: Fabio Estevam: "Re: [PATCH] PCI: imx6: Implement reset sequence for i.MX6+"
In reply to: Vinod Koul: "Re: [PATCH v3 08/15] dmaengine: dw: revisit data_width property"
Next in thread: Vinod Koul: "Re: [PATCH v3 08/15] dmaengine: dw: revisit data_width property"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Wed, 2016-04-13 at 21:27 +0530, Vinod Koul wrote:
> On Fri, Mar 18, 2016 at 04:24:47PM +0200, Andy Shevchenko wrote:
> >
> > There are several changes are done here:
> >
> > Â- Convert the property to be in bytes
> >
> > ÂÂÂMuch more convenient than keeping encoded value.
> >
> > Â- Use one value for all AHB masters for now
> >
> > ÂÂÂIt seems in practice we have no controllers where masters have
> > different
> > ÂÂÂdata bus width, we still might return to distinct values when
> > there is a use
> > ÂÂÂcase.
> >
> > Â- Rename data_width to data-width in the device tree bindings.
> >
> > Â- While here, replace dwc_fast_ffs() by __ffs().
> >
> > Signed-off-by: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
> > ---
> > ÂDocumentation/devicetree/bindings/dma/snps-dma.txt |ÂÂ5 ++-
> > Âarch/arc/boot/dts/abilis_tb10x.dtsiÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ|ÂÂ2 +-
> > Âarch/arm/boot/dts/spear13xx.dtsiÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ|ÂÂ4 +--
> > Âdrivers/dma/dw/core.cÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ| 40 +++------
> > -------------
> > Âdrivers/dma/dw/platform.cÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ| 10 +++---
> > Âdrivers/dma/dw/regs.hÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ|ÂÂ2 +-
> > Âinclude/linux/platform_data/dma-dw.hÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ|ÂÂ5 ++-
> > Â7 files changed, 18 insertions(+), 50 deletions(-)
> >
> > diff --git a/Documentation/devicetree/bindings/dma/snps-dma.txt
> > b/Documentation/devicetree/bindings/dma/snps-dma.txt
> > index c99c1ff..eb48229 100644
> > --- a/Documentation/devicetree/bindings/dma/snps-dma.txt
> > +++ b/Documentation/devicetree/bindings/dma/snps-dma.txt
> > @@ -13,8 +13,7 @@ Required properties:
> > Â- chan_priority: priority of channels. 0 (default): increase from
> > chan 0->n, 1:
> > ÂÂÂincrease from chan n->0
> > Â- block_size: Maximum block size supported by the controller
> > -- data_width: Maximum data width supported by hardware per AHB
> > master
> > -ÂÂ(0 - 8bits, 1 - 16bits, ..., 5 - 256bits)
> > +- data-width: Maximum data width supported by hardware (in bytes,
> > power of 2)
> You missed that binding are treated as ABI, see
> Documentation/devicetree/bindings/ABI.txt
>
> The driver should still work with older DT implementation, so you need
> to
> keep support for that in driver and hence I don't see any benfit we
> would
> get from doing both in driver!

The device tree is screwed by a process that allows to do almost
whatever you want. Here I'm trying to rectify the usage of the field.

The old is still supported and benefit is apparently in unifying
standard properties across the drivers.

>
> >
> > Â
> > Â
> > ÂOptional properties:
> > @@ -38,7 +37,7 @@ Example:
> > Â chan_allocation_order = <1>;
> > Â chan_priority = <1>;
> > Â block_size = <0xfff>;
> > - data_width = <3 3>;
> > + data-width = <8>;
> > Â };
> > Â
> > ÂDMA clients connected to the Designware DMA controller must use the
> > format
> > diff --git a/arch/arc/boot/dts/abilis_tb10x.dtsi
> > b/arch/arc/boot/dts/abilis_tb10x.dtsi
> > index cfb5052..2f53bed 100644
> > --- a/arch/arc/boot/dts/abilis_tb10x.dtsi
> > +++ b/arch/arc/boot/dts/abilis_tb10x.dtsi
> > @@ -112,7 +112,7 @@
> > Â chan_allocation_order = <0>;
> > Â chan_priority = <1>;
> > Â block_size = <0x7ff>;
> > - data_width = <2>;
> > + data-width = <4>;
> > Â clocks = <&ahb_clk>;
> > Â clock-names = "hclk";
> > Â };
> > diff --git a/arch/arm/boot/dts/spear13xx.dtsi
> > b/arch/arm/boot/dts/spear13xx.dtsi
> > index 14594ce..474b66f 100644
> > --- a/arch/arm/boot/dts/spear13xx.dtsi
> > +++ b/arch/arm/boot/dts/spear13xx.dtsi
> > @@ -117,7 +117,7 @@
> > Â chan_priority = <1>;
> > Â block_size = <0xfff>;
> > Â dma-masters = <2>;
> > - data_width = <3 3>;
> > + data-width = <8>;
> > Â };
> > Â
> > Â dma@eb000000 {
> > @@ -133,7 +133,7 @@
> > Â chan_allocation_order = <1>;
> > Â chan_priority = <1>;
> > Â block_size = <0xfff>;
> > - data_width = <3 3>;
> > + data-width = <8>;
> > Â };
> > Â
> > Â fsmc: flash@b0000000 {
> > diff --git a/drivers/dma/dw/core.c b/drivers/dma/dw/core.c
> > index 4748df9..79dbfc3 100644
> > --- a/drivers/dma/dw/core.c
> > +++ b/drivers/dma/dw/core.c
> > @@ -155,21 +155,6 @@ static void dwc_initialize(struct dw_dma_chan
> > *dwc)
> > Â
> > Â/*-----------------------------------------------------------------
> > -----*/
> > Â
> > -static inline unsigned int dwc_fast_ffs(unsigned long long v)
> > -{
> > - /*
> > - Â* We can be a lot more clever here, but this should take
> > care
> > - Â* of the most common optimization.
> > - Â*/
> > - if (!(v & 7))
> > - return 3;
> > - else if (!(v & 3))
> > - return 2;
> > - else if (!(v & 1))
> > - return 1;
> > - return 0;
> > -}
> > -
> > Âstatic inline void dwc_dump_chan_regs(struct dw_dma_chan *dwc)
> > Â{
> > Â dev_err(chan2dev(&dwc->chan),
> > @@ -703,7 +688,6 @@ dwc_prep_dma_memcpy(struct dma_chan *chan,
> > dma_addr_t dest, dma_addr_t src,
> > Â size_t offset;
> > Â unsigned int src_width;
> > Â unsigned int dst_width;
> > - unsigned int data_width;
> > Â u32 ctllo;
> > Â u8 lms = DWC_LLP_LMS(dwc->m_master);
> > Â
> > @@ -718,10 +702,7 @@ dwc_prep_dma_memcpy(struct dma_chan *chan,
> > dma_addr_t dest, dma_addr_t src,
> > Â
> > Â dwc->direction = DMA_MEM_TO_MEM;
> > Â
> > - data_width = dw->data_width[dwc->m_master];
> > -
> > - src_width = dst_width = min_t(unsigned int, data_width,
> > - ÂÂÂÂÂÂdwc_fast_ffs(src | dest |
> > len));
> > + src_width = dst_width = __ffs(dw->data_width | src | dest |
> > len);
> > Â
> > Â ctllo = DWC_DEFAULT_CTLLO(chan)
> > Â | DWC_CTLL_DST_WIDTH(dst_width)
> > @@ -785,7 +766,6 @@ dwc_prep_slave_sg(struct dma_chan *chan, struct
> > scatterlist *sgl,
> > Â dma_addr_t reg;
> > Â unsigned int reg_width;
> > Â unsigned int mem_width;
> > - unsigned int data_width;
> > Â unsigned int i;
> > Â struct scatterlist *sg;
> > Â size_t total_len = 0;
> > @@ -811,8 +791,6 @@ dwc_prep_slave_sg(struct dma_chan *chan, struct
> > scatterlist *sgl,
> > Â ctllo |= sconfig->device_fc ?
> > DWC_CTLL_FC(DW_DMA_FC_P_M2P) :
> > Â DWC_CTLL_FC(DW_DMA_FC_D_M2P);
> > Â
> > - data_width = dw->data_width[dwc->m_master];
> > -
> > Â for_each_sg(sgl, sg, sg_len, i) {
> > Â struct dw_desc *desc;
> > Â u32 len, dlen, mem;
> > @@ -820,8 +798,7 @@ dwc_prep_slave_sg(struct dma_chan *chan, struct
> > scatterlist *sgl,
> > Â mem = sg_dma_address(sg);
> > Â len = sg_dma_len(sg);
> > Â
> > - mem_width = min_t(unsigned int,
> > - ÂÂdata_width,
> > dwc_fast_ffs(mem | len));
> > + mem_width = __ffs(dw->data_width | mem |
> > len);
> > Â
> > Âslave_sg_todev_fill_desc:
> > Â desc = dwc_desc_get(dwc);
> > @@ -867,8 +844,6 @@ slave_sg_todev_fill_desc:
> > Â ctllo |= sconfig->device_fc ?
> > DWC_CTLL_FC(DW_DMA_FC_P_P2M) :
> > Â DWC_CTLL_FC(DW_DMA_FC_D_P2M);
> > Â
> > - data_width = dw->data_width[dwc->m_master];
> > -
> > Â for_each_sg(sgl, sg, sg_len, i) {
> > Â struct dw_desc *desc;
> > Â u32 len, dlen, mem;
> > @@ -876,8 +851,7 @@ slave_sg_todev_fill_desc:
> > Â mem = sg_dma_address(sg);
> > Â len = sg_dma_len(sg);
> > Â
> > - mem_width = min_t(unsigned int,
> > - ÂÂdata_width,
> > dwc_fast_ffs(mem | len));
> > + mem_width = __ffs(dw->data_width | mem |
> > len);
> > Â
> > Âslave_sg_fromdev_fill_desc:
> > Â desc = dwc_desc_get(dwc);
> > @@ -1544,10 +1518,7 @@ int dw_dma_probe(struct dw_dma_chip *chip,
> > struct dw_dma_platform_data *pdata)
> > Â /* Get hardware configuration parameters */
> > Â pdata->nr_channels = (dw_params >>
> > DW_PARAMS_NR_CHAN & 7) + 1;
> > Â pdata->nr_masters = (dw_params >>
> > DW_PARAMS_NR_MASTER & 3) + 1;
> > - for (i = 0; i < pdata->nr_masters; i++) {
> > - pdata->data_width[i] =
> > - (dw_params >>
> > DW_PARAMS_DATA_WIDTH(i) & 3) + 2;
> > - }
> > + pdata->data_width = 4 << (dw_params >>
> > DW_PARAMS_DATA_WIDTH(0) & 3);
> > Â max_blk_size = dma_readl(dw, MAX_BLK_SIZE);
> > Â
> > Â /* Fill platform data with the default values */
> > @@ -1569,8 +1540,7 @@ int dw_dma_probe(struct dw_dma_chip *chip,
> > struct dw_dma_platform_data *pdata)
> > Â
> > Â /* Get hardware configuration parameters */
> > Â dw->nr_masters = pdata->nr_masters;
> > - for (i = 0; i < dw->nr_masters; i++)
> > - dw->data_width[i] = pdata->data_width[i];
> > + dw->data_width = pdata->data_width;
> > Â
> > Â /* Calculate all channel mask before DMA setup */
> > Â dw->all_chan_mask = (1 << pdata->nr_channels) - 1;
> > diff --git a/drivers/dma/dw/platform.c b/drivers/dma/dw/platform.c
> > index 23616c5..4cf399d 100644
> > --- a/drivers/dma/dw/platform.c
> > +++ b/drivers/dma/dw/platform.c
> > @@ -102,8 +102,8 @@ dw_dma_parse_dt(struct platform_device *pdev)
> > Â{
> > Â struct device_node *np = pdev->dev.of_node;
> > Â struct dw_dma_platform_data *pdata;
> > - u32 tmp, arr[DW_DMA_MAX_NR_MASTERS];
> > Â u32 nr_channels;
> > + u32 tmp;
> > Â
> > Â if (!np) {
> > Â dev_err(&pdev->dev, "Missing DT data\n");
> > @@ -138,10 +138,10 @@ dw_dma_parse_dt(struct platform_device *pdev)
> > Â pdata->nr_masters = tmp;
> > Â }
> > Â
> > - if (!of_property_read_u32_array(np, "data_width", arr,
> > - pdata->nr_masters))
> > - for (tmp = 0; tmp < pdata->nr_masters; tmp++)
> > - pdata->data_width[tmp] = arr[tmp];
> > + if (!of_property_read_u32(np, "data-width", &tmp))
> > + pdata->data_width = tmp;
> > + else if (!of_property_read_u32(np, "data_width", &tmp))
> > + pdata->data_width = BIT(tmp & 0x07);
> > Â
> > Â return pdata;
> > Â}
> > diff --git a/drivers/dma/dw/regs.h b/drivers/dma/dw/regs.h
> > index feb3a4a..d0f9173 100644
> > --- a/drivers/dma/dw/regs.h
> > +++ b/drivers/dma/dw/regs.h
> > @@ -285,7 +285,7 @@ struct dw_dma {
> > Â
> > Â /* hardware configuration */
> > Â unsigned char nr_masters;
> > - unsigned char data_width[DW_DMA_MAX_NR_MASTE
> > RS];
> > + unsigned char data_width;
> > Â};
> > Â
> > Âstatic inline struct dw_dma_regs __iomem *__dw_regs(struct dw_dma
> > *dw)
> > diff --git a/include/linux/platform_data/dma-dw.h
> > b/include/linux/platform_data/dma-dw.h
> > index b881b97..09cf5c1 100644
> > --- a/include/linux/platform_data/dma-dw.h
> > +++ b/include/linux/platform_data/dma-dw.h
> > @@ -42,8 +42,7 @@ struct dw_dma_slave {
> > Â * @chan_priority: Set channel priority increasing from 0 to 7 or 7
> > to 0.
> > Â * @block_size: Maximum block size supported by the controller
> > Â * @nr_masters: Number of AHB masters supported by the controller
> > - * @data_width: Maximum data width supported by hardware per AHB
> > master
> > - * (0 - 8bits, 1 - 16bits, ..., 5 - 256bits)
> > + * @data_width: Maximum data width supported by hardware (in bytes,
> > power of 2)
> > Â */
> > Âstruct dw_dma_platform_data {
> > Â unsigned int nr_channels;
> > @@ -57,7 +56,7 @@ struct dw_dma_platform_data {
> > Â unsigned char chan_priority;
> > Â unsigned short block_size;
> > Â unsigned char nr_masters;
> > - unsigned char data_width[DW_DMA_MAX_NR_MASTERS];
> > + unsigned char data_width;
> > Â};
> > Â
> > Â#endif /* _PLATFORM_DATA_DMA_DW_H */
> > --Â
> > 2.7.0
> >

--
Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
Intel Finland Oy

Next message: kbuild test robot: "Re: [PATCH 1/2] [media] atmel-isc: add the Image Sensor Controller code"
Previous message: Fabio Estevam: "Re: [PATCH] PCI: imx6: Implement reset sequence for i.MX6+"
In reply to: Vinod Koul: "Re: [PATCH v3 08/15] dmaengine: dw: revisit data_width property"
Next in thread: Vinod Koul: "Re: [PATCH v3 08/15] dmaengine: dw: revisit data_width property"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]