| Message ID | 1479304615-11360-5-git-send-email-Eugeniy.Paltsev@synopsys.com |
|---|---|
| State | New |
| Headers | show |
On Wed, 2016-11-16 at 16:56 +0300, Eugeniy Paltsev wrote: > * Rename is_private to is-private as ordered by DT policy. > The change leaves the support for the old format. > > * Add is-memcpu property, so it is possible to > enable memory-to-memory transfers support via DT. > > * Add hw-llp property, so it is possible to enable > hardware multi block transfers support via DT. > > Fix white spaces. > > Signed-off-by: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com> > --- > Documentation/devicetree/bindings/dma/snps-dma.txt | 10 ++++++++-- > 1 file changed, 8 insertions(+), 2 deletions(-) > > diff --git a/Documentation/devicetree/bindings/dma/snps-dma.txt > b/Documentation/devicetree/bindings/dma/snps-dma.txt > index 0f55832..d41d960 100644 > --- a/Documentation/devicetree/bindings/dma/snps-dma.txt > +++ b/Documentation/devicetree/bindings/dma/snps-dma.txt > @@ -20,13 +20,19 @@ Required properties: > Deprecated properties: > - data_width: Maximum data width supported by hardware per AHB master > (0 - 8bits, 1 - 16bits, ..., 5 - 256bits) > +- is_private: The device channels should be marked as private and not > for by the > + general purpose DMA channel allocator. False if not passed. This... > > > Optional properties: > - interrupt-parent: Should be the phandle for the interrupt > controller > that services interrupts for this device > -- is_private: The device channels should be marked as private and not > for by the > +- is-private: The device channels should be marked as private and not > for by the > general purpose DMA channel allocator. False if not passed. ...and this is a part of patch 1. > +- is-memcpu: The device channels do support memory-to-memory memcpy > transfers. > False > + if not passed. > +- hw-llp: Multi block transfers supported by hardware per AHB master. > + 0 (default): not supported, 1: supported. Overall, since we are going to expose some properties to the Device Tree I would really think twice about naming. Better if we reuse something existing already. So, what I can see is dmacap,private dmacap,memcpy Here is a selling point as well, i.e. standardization. 'hw-llp' sounds too tricky, perhaps 'multi-block' is better and could be re-used. > > Example: > > @@ -56,7 +62,7 @@ The four cells in order are: > 4. Peripheral master for transfers on allocated channel > > Example: > - > + No, no need to touch this.
Hi Andy, On Wed, 2016-11-16 at 17:10 +0200, Andy Shevchenko wrote: > Overall, since we are going to expose some properties to the Device > Tree > I would really think twice about naming. Better if we reuse something > existing already. > > So, what I can see is > > dmacap,private > dmacap,memcpy > > Here is a selling point as well, i.e. standardization. > As I can see these property name used only in "mv_xor" driver. And they are marked as deprecated. So, I'm not sure if I should used these names. I agree with other comments. -- Paltsev Eugeniy
On Wed, 2016-11-16 at 17:01 +0000, Eugeniy Paltsev wrote: > On Wed, 2016-11-16 at 17:10 +0200, Andy Shevchenko wrote: > > Overall, since we are going to expose some properties to the Device > > Tree > > I would really think twice about naming. Better if we reuse > > something > > existing already. > > > > So, what I can see is > > > > dmacap,private > > dmacap,memcpy > > > > Here is a selling point as well, i.e. standardization. > > > > As I can see these property name used only in "mv_xor" driver. And > they are marked as deprecated. > So, I'm not sure if I should used these names. Oh, good catch! So, then I leave this to DT experienced guys to decide. Rob?
On Wed, Nov 16, 2016 at 12:08 PM, Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote: > On Wed, 2016-11-16 at 17:01 +0000, Eugeniy Paltsev wrote: >> On Wed, 2016-11-16 at 17:10 +0200, Andy Shevchenko wrote: >> > Overall, since we are going to expose some properties to the Device >> > Tree >> > I would really think twice about naming. Better if we reuse >> > something >> > existing already. >> > >> > So, what I can see is >> > >> > dmacap,private >> > dmacap,memcpy >> > >> > Here is a selling point as well, i.e. standardization. >> > >> >> As I can see these property name used only in "mv_xor" driver. And >> they are marked as deprecated. >> So, I'm not sure if I should used these names. > > Oh, good catch! So, then I leave this to DT experienced guys to decide. > Rob? Well, maybe they were deprecated for a reason? This all seems like user configuration to me. So either they don't belong in DT or should be common if they do. Rob
diff --git a/Documentation/devicetree/bindings/dma/snps-dma.txt b/Documentation/devicetree/bindings/dma/snps-dma.txt index 0f55832..d41d960 100644 --- a/Documentation/devicetree/bindings/dma/snps-dma.txt +++ b/Documentation/devicetree/bindings/dma/snps-dma.txt @@ -20,13 +20,19 @@ Required properties: Deprecated properties: - data_width: Maximum data width supported by hardware per AHB master (0 - 8bits, 1 - 16bits, ..., 5 - 256bits) +- is_private: The device channels should be marked as private and not for by the + general purpose DMA channel allocator. False if not passed. Optional properties: - interrupt-parent: Should be the phandle for the interrupt controller that services interrupts for this device -- is_private: The device channels should be marked as private and not for by the +- is-private: The device channels should be marked as private and not for by the general purpose DMA channel allocator. False if not passed. +- is-memcpu: The device channels do support memory-to-memory transfers. False + if not passed. +- hw-llp: Multi block transfers supported by hardware per AHB master. + 0 (default): not supported, 1: supported. Example: @@ -56,7 +62,7 @@ The four cells in order are: 4. Peripheral master for transfers on allocated channel Example: - + serial@e0000000 { compatible = "arm,pl011", "arm,primecell"; reg = <0xe0000000 0x1000>;
* Rename is_private to is-private as ordered by DT policy. The change leaves the support for the old format. * Add is-memcpu property, so it is possible to enable memory-to-memory transfers support via DT. * Add hw-llp property, so it is possible to enable hardware multi block transfers support via DT. Fix white spaces. Signed-off-by: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com> --- Documentation/devicetree/bindings/dma/snps-dma.txt | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)