On 4/14/2022 2:39 AM, Dmitry Osipenko wrote:
External email: Use caution opening links or attachments
On 4/13/22 19:17, Ashish Mhetre wrote:
On 4/13/2022 7:34 PM, Dmitry Osipenko wrote:
External email: Use caution opening links or attachments
On 4/13/22 16:37, Rob Herring wrote:
On Wed, Apr 13, 2022 at 03:10:11PM +0530, Ashish Mhetre wrote:
From tegra186 onwards, memory controller support multiple channels.
Reg items are updated with address and size of these channels.
Tegra186 has overall 5 memory controller channels. Tegra194 and
tegra234
have overall 17 memory controller channels each.
There is 1 reg item for memory controller stream-id registers.
So update the reg maxItems to 18 in tegra186 devicetree documentation.
Also update validation for reg-names added for these corresponding reg
items.
Somehow your subject should indicate this is for Tegra.
Signed-off-by: Ashish Mhetre <amhetre@xxxxxxxxxx>
---
.../nvidia,tegra186-mc.yaml | 80
+++++++++++++++++--
1 file changed, 74 insertions(+), 6 deletions(-)
diff --git
a/Documentation/devicetree/bindings/memory-controllers/nvidia,tegra186-mc.yaml
b/Documentation/devicetree/bindings/memory-controllers/nvidia,tegra186-mc.yaml
index 13c4c82fd0d3..c7cfa6c2cd81 100644
---
a/Documentation/devicetree/bindings/memory-controllers/nvidia,tegra186-mc.yaml
+++
b/Documentation/devicetree/bindings/memory-controllers/nvidia,tegra186-mc.yaml
@@ -34,8 +34,12 @@ properties:
- nvidia,tegra234-mc
reg:
- minItems: 1
- maxItems: 3
+ minItems: 6
You just broke current users.
+ maxItems: 18
+
+ reg-names:
+ minItems: 6
+ maxItems: 18
interrupts:
items:
@@ -142,7 +146,18 @@ allOf:
then:
properties:
reg:
- maxItems: 1
+ maxItems: 6
+ description: 5 memory controller channels and 1 for
stream-id registers
+
+ reg-names:
+ maxItems: 6
+ items:
+ - const: sid
+ - const: broadcast
+ - const: ch0
+ - const: ch1
+ - const: ch2
+ - const: ch3
- if:
properties:
@@ -151,7 +166,30 @@ allOf:
then:
properties:
reg:
- minItems: 3
+ minItems: 18
+ description: 17 memory controller channels and 1 for
stream-id registers
+
+ reg-names:
+ minItems: 18
+ items:
+ - const: sid
+ - const: broadcast
+ - const: ch0
+ - const: ch1
+ - const: ch2
+ - const: ch3
+ - const: ch4
+ - const: ch5
+ - const: ch6
+ - const: ch7
+ - const: ch8
+ - const: ch9
+ - const: ch10
+ - const: ch11
+ - const: ch12
+ - const: ch13
+ - const: ch14
+ - const: ch15
- if:
properties:
@@ -160,13 +198,37 @@ allOf:
then:
properties:
reg:
- minItems: 3
+ minItems: 18
+ description: 17 memory controller channels and 1 for
stream-id registers
+
+ reg-names:
+ minItems: 18
+ items:
+ - const: sid
+ - const: broadcast
+ - const: ch0
+ - const: ch1
+ - const: ch2
+ - const: ch3
+ - const: ch4
+ - const: ch5
+ - const: ch6
+ - const: ch7
+ - const: ch8
+ - const: ch9
+ - const: ch10
+ - const: ch11
+ - const: ch12
+ - const: ch13
+ - const: ch14
+ - const: ch15
additionalProperties: false
required:
- compatible
- reg
+ - reg-names
New, added properties cannot be required. That's an ABI break.
- interrupts
- "#address-cells"
- "#size-cells"
@@ -182,7 +244,13 @@ examples:
memory-controller@2c00000 {
compatible = "nvidia,tegra186-mc";
- reg = <0x0 0x02c00000 0x0 0xb0000>;
+ reg = <0x0 0x02c00000 0x0 0x10000>, /* MC-SID */
+ <0x0 0x02c10000 0x0 0x10000>, /* Broadcast
channel */
+ <0x0 0x02c20000 0x0 0x10000>, /* MC0 */
+ <0x0 0x02c30000 0x0 0x10000>, /* MC1 */
+ <0x0 0x02c40000 0x0 0x10000>, /* MC2 */
+ <0x0 0x02c50000 0x0 0x10000>; /* MC3 */
+ reg-names = "sid", "broadcast", "ch0", "ch1", "ch2",
"ch3";
interrupts = <GIC_SPI 223 IRQ_TYPE_LEVEL_HIGH>;
#address-cells = <2>;
--
2.17.1
Oh, wait.. I didn't notice that the new reg ranges are only splitting up
the old ranges. Previously it appeared to me that these are the new
ranges.
> Ashish, in this case you don't need to change the regs in the DT at
all.
Instead, you need to specify the per-channel reg-base offsets in the
driver code.
Yes, it's kind of splitting up the old ranges and straight forward for
Tegra186. But on Tegra194 and Tegra234 the old address is not in single
range. It's already split across 3 ranges. We have to choose right range
and add channel offsets to that range in order to read interrupts.
So I went with the approach of splitting the regs in DT itself as per
the channels because that way they can be mapped in a single loop and
used easily.
If we want to specify per-channel reg-base offsets then that would be
per-SOC. Also we would need to choose correct reg-range for Tegra194 and
Tegra234 and have a way to maintain offsets of channels from those
respective reg-ranges.
That is not nice too. Should be better to switch to the new DT scheme,
since those channels weren't used by older kernels. It's okay to change
the binding ABI in this case then, driver will continue to work for the
older dtbs.
So the current DTS and binding changes are fine?
Have you tested driver using the older dtbs?
Yes, the driver is tested with old dtb and it's working fine.
