Add basic SoC support for mediatek mt7986
Subsystems (10)
cover Add basic SoC support for mediatek mt7986
This patch adds basic SoC support for Mediatek's new 4-core SoC, MT7986, which is mainly for wifi-router application.
Sam Shih (12):
dt-bindings: clock: mediatek: document clk bindings for mediatek
mt7986 SoC
clk: mediatek: add mt7986 clock IDs
clk: mediatek: add mt7986 clock support
pinctrl: mediatek: moore: use pin number in mtk_pin_desc instead of
array index
dt-bindings: pinctrl: update bindings for MT7986 SoC
pinctrl: mediatek: add support for MT7986 SoC
dt-bindings: arm64: dts: mediatek: Add mt7986 series
dt-bindings: rng: mediatek: add mt7986 to mtk rng binding
dt-bindings: serial: Add compatible for Mediatek MT7986
dt-bindings: watchdog: Add compatible for Mediatek MT7986
arm64: dts: mediatek: add mt7986a support
arm64: dts: mediatek: add mt7986b support
.../devicetree/bindings/arm/mediatek.yaml | 8 + .../arm/mediatek/mediatek,apmixedsys.txt | 1 + .../bindings/arm/mediatek/mediatek,ethsys.txt | 1 + .../arm/mediatek/mediatek,infracfg.txt | 2 + .../arm/mediatek/mediatek,sgmiisys.txt | 2 + .../arm/mediatek/mediatek,topckgen.txt | 1 + .../bindings/pinctrl/pinctrl-mt7622.txt | 284 +++ .../devicetree/bindings/rng/mtk-rng.yaml | 1 + .../devicetree/bindings/serial/mtk-uart.txt | 1 + .../devicetree/bindings/watchdog/mtk-wdt.txt | 1 + arch/arm64/boot/dts/mediatek/Makefile | 2 + arch/arm64/boot/dts/mediatek/mt7986a-rfb.dts | 49 + arch/arm64/boot/dts/mediatek/mt7986a.dtsi | 235 +++ arch/arm64/boot/dts/mediatek/mt7986b-rfb.dts | 21 + arch/arm64/boot/dts/mediatek/mt7986b.dtsi | 235 +++ drivers/clk/mediatek/Kconfig | 17 + drivers/clk/mediatek/Makefile | 2 + drivers/clk/mediatek/clk-mt7986-eth.c | 132 ++ drivers/clk/mediatek/clk-mt7986.c | 610 ++++++ drivers/pinctrl/mediatek/Kconfig | 7 + drivers/pinctrl/mediatek/Makefile | 1 + drivers/pinctrl/mediatek/pinctrl-moore.c | 61 + drivers/pinctrl/mediatek/pinctrl-mt7986.c | 1640 +++++++++++++++++ include/dt-bindings/clock/mt7986-clk.h | 244 +++ 24 files changed, 3558 insertions(+) create mode 100644 arch/arm64/boot/dts/mediatek/mt7986a-rfb.dts create mode 100644 arch/arm64/boot/dts/mediatek/mt7986a.dtsi create mode 100644 arch/arm64/boot/dts/mediatek/mt7986b-rfb.dts create mode 100644 arch/arm64/boot/dts/mediatek/mt7986b.dtsi create mode 100644 drivers/clk/mediatek/clk-mt7986-eth.c create mode 100644 drivers/clk/mediatek/clk-mt7986.c create mode 100644 drivers/pinctrl/mediatek/pinctrl-mt7986.c create mode 100644 include/dt-bindings/clock/mt7986-clk.h
Signature · 1 line
2.29.2
PATCH 1/12 dt-bindings: clock: mediatek: document clk bindings for mediatek mt7986 SoC
This patch adds the binding documentation for topckgen, apmixedsys, infracfg, infracfg_ao, and ethernet subsystem clocks.
--- .../devicetree/bindings/arm/mediatek/mediatek,apmixedsys.txt | 1 + .../devicetree/bindings/arm/mediatek/mediatek,ethsys.txt | 1 + .../devicetree/bindings/arm/mediatek/mediatek,infracfg.txt | 2 ++ .../devicetree/bindings/arm/mediatek/mediatek,sgmiisys.txt | 2 ++ .../devicetree/bindings/arm/mediatek/mediatek,topckgen.txt | 1 + 5 files changed, 7 insertions(+)
Documentation/devicetree/bindings/arm/mediatek/mediatek,apmixedsys.txt +1 -0
@@ -14,6 +14,7 @@ Required Properties: - "mediatek,mt7622-apmixedsys" - "mediatek,mt7623-apmixedsys", "mediatek,mt2701-apmixedsys" - "mediatek,mt7629-apmixedsys" + - "mediatek,mt7986-apmixedsys" - "mediatek,mt8135-apmixedsys" - "mediatek,mt8167-apmixedsys", "syscon" - "mediatek,mt8173-apmixedsys"
Documentation/devicetree/bindings/arm/mediatek/mediatek,ethsys.txt +1 -0
@@ -10,6 +10,7 @@ Required Properties: - "mediatek,mt7622-ethsys", "syscon" - "mediatek,mt7623-ethsys", "mediatek,mt2701-ethsys", "syscon" - "mediatek,mt7629-ethsys", "syscon" + - "mediatek,mt7986-ethsys", "syscon" - #clock-cells: Must be 1 - #reset-cells: Must be 1
Documentation/devicetree/bindings/arm/mediatek/mediatek,infracfg.txt +2 -0
@@ -15,6 +15,8 @@ Required Properties: - "mediatek,mt7622-infracfg", "syscon" - "mediatek,mt7623-infracfg", "mediatek,mt2701-infracfg", "syscon" - "mediatek,mt7629-infracfg", "syscon" + - "mediatek,mt7986-infracfg", "syscon" + - "mediatek,mt7986-infracfg_ao", "syscon" - "mediatek,mt8135-infracfg", "syscon" - "mediatek,mt8167-infracfg", "syscon" - "mediatek,mt8173-infracfg", "syscon"
Documentation/devicetree/bindings/arm/mediatek/mediatek,sgmiisys.txt +2 -0
@@ -8,6 +8,8 @@ Required Properties: - compatible: Should be: - "mediatek,mt7622-sgmiisys", "syscon" - "mediatek,mt7629-sgmiisys", "syscon" + - "mediatek,mt7986-sgmiisys", "mediatek,mt7986-sgmiisys_0", "syscon" + - "mediatek,mt7986-sgmiisys", "mediatek,mt7986-sgmiisys_1", "syscon" - #clock-cells: Must be 1 The SGMIISYS controller uses the common clk binding from
Documentation/devicetree/bindings/arm/mediatek/mediatek,topckgen.txt +1 -0
@@ -14,6 +14,7 @@ Required Properties: - "mediatek,mt7622-topckgen" - "mediatek,mt7623-topckgen", "mediatek,mt2701-topckgen" - "mediatek,mt7629-topckgen" + - "mediatek,mt7986-topckgen", "syscon" - "mediatek,mt8135-topckgen" - "mediatek,mt8167-topckgen", "syscon" - "mediatek,mt8173-topckgen"
Signature · 1 line
2.29.2
Re: [PATCH 01/12] dt-bindings: clock: mediatek: document clk bindings for mediatek mt7986 SoC
Quoted 55 lines · Sam Shih
This patch adds the binding documentation for topckgen, apmixedsys, infracfg,
infracfg_ao, and ethernet subsystem clocks.Signed-off-by Sam Shih <[email protected]>--- .../devicetree/bindings/arm/mediatek/mediatek,apmixedsys.txt | 1 + .../devicetree/bindings/arm/mediatek/mediatek,ethsys.txt | 1 + .../devicetree/bindings/arm/mediatek/mediatek,infracfg.txt | 2 ++ .../devicetree/bindings/arm/mediatek/mediatek,sgmiisys.txt | 2 ++ .../devicetree/bindings/arm/mediatek/mediatek,topckgen.txt | 1 + 5 files changed, 7 insertions(+)Documentation/devicetree/bindings/arm/mediatek/mediatek,apmixedsys.txt +1
-0@@ -14,6 +14,7 @@ Required Properties: - "mediatek,mt7622-apmixedsys" - "mediatek,mt7623-apmixedsys", "mediatek,mt2701-apmixedsys" - "mediatek,mt7629-apmixedsys" + - "mediatek,mt7986-apmixedsys" - "mediatek,mt8135-apmixedsys" - "mediatek,mt8167-apmixedsys", "syscon" - "mediatek,mt8173-apmixedsys"Documentation/devicetree/bindings/arm/mediatek/mediatek,ethsys.txt +1
-0@@ -10,6 +10,7 @@ Required Properties: - "mediatek,mt7622-ethsys", "syscon" - "mediatek,mt7623-ethsys", "mediatek,mt2701-ethsys", "syscon" - "mediatek,mt7629-ethsys", "syscon" + - "mediatek,mt7986-ethsys", "syscon" - #clock-cells: Must be 1 - #reset-cells: Must be 1Documentation/devicetree/bindings/arm/mediatek/mediatek,infracfg.txt +2
-0@@ -15,6 +15,8 @@ Required Properties: - "mediatek,mt7622-infracfg", "syscon" - "mediatek,mt7623-infracfg", "mediatek,mt2701-infracfg", "syscon" - "mediatek,mt7629-infracfg", "syscon" + - "mediatek,mt7986-infracfg", "syscon" + - "mediatek,mt7986-infracfg_ao", "syscon" - "mediatek,mt8135-infracfg", "syscon" - "mediatek,mt8167-infracfg", "syscon" - "mediatek,mt8173-infracfg", "syscon"Documentation/devicetree/bindings/arm/mediatek/mediatek,sgmiisys.txt +0
-0
Documentation/devicetree/bindings/arm/mediatek/mediatek,sgmiisys.txt Required Properties: · line 12
Quoted 5 lines
Documentation/devicetree/bindings/arm/mediatek/mediatek,sgmiisys.txt +2
-0@@ -8,6 +8,8 @@ Required Properties: - compatible: Should be: - "mediatek,mt7622-sgmiisys", "syscon" - "mediatek,mt7629-sgmiisys", "syscon" + - "mediatek,mt7986-sgmiisys", "mediatek,mt7986-sgmiisys_0", "syscon" + - "mediatek,mt7986-sgmiisys", "mediatek,mt7986-sgmiisys_1", "syscon"
The order should be: most specific compatible string first, followed by fallbacks.
Furthermore, based on the driver patch and the fact that they share the same compatible string, it seems you shouldn't need to have two compatible strings for two identical hardware blocks. The need for separate entries to have different clock names is an implementation detail. Please consider using and supporting clock-output-names.
Also, please check out the MT8195 clock driver series [1]. I'm guessing a lot of the comments apply to this one as well.
Regards
ChenYu
[1] https://lore.kernel.org/linux-mediatek/[email protected]/T/#t
Quoted 21 lines
Documentation/devicetree/bindings/arm/mediatek/mediatek,sgmiisys.txt +0
-0- #clock-cells: Must be 1 The SGMIISYS controller uses the common clk binding fromDocumentation/devicetree/bindings/arm/mediatek/mediatek,topckgen.txt +1
-1@@ -14,6 +14,7 @@ Required Properties: - "mediatek,mt7622-topckgen" - "mediatek,mt7623-topckgen", "mediatek,mt2701-topckgen" - "mediatek,mt7629-topckgen" + - "mediatek,mt7986-topckgen", "syscon" - "mediatek,mt8135-topckgen" - "mediatek,mt8167-topckgen", "syscon" - "mediatek,mt8173-topckgen" --2.29.2
_______________________________________________
Linux-mediatek mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-mediatek
Re: [PATCH 01/12] dt-bindings: clock: mediatek: document clk bindings for mediatek mt7986 SoC
Hi,
Quoted 14 lines · Chen-Yu Tsai
Furthermore, based on the driver patch and the fact that they share the
same compatible string, it seems you shouldn't need to have two compatible
strings for two identical hardware blocks. The need for separate entries
to have different clock names is an implementation detail. Please consider
using and supporting clock-output-names.Also, please check out the MT8195 clock driver series [1]. I'm guessing
a lot of the comments apply to this one as well.
Quoted 7 lines
Regards
ChenYu
I have organized your comments in "Mediatek MT8195 clock support" series into the following list, reply to your here:
Quoted 2 lines
dt-binding: Move the not-to-be-exposed clock to driver directory or simply left out
Okay, thanks for your comment, I will update this in the next patch set
Quoted 2 lines
describe some of the clock relations between the various clock controllers
I have checked the files in "Documentation/devicetree/bindings/arm/mediatek", It seems that all MediaTek SoC clocks are simply described by each controller, like "mediatek,infracfg.txt" and "mediatek,topckgen.txt", and those document only include compatible strings information and examples. Can we insert the clock relationship of MT7986 directlly in common documents?
Or we should add a new "mediatek,mt7986-clock.yaml" and move compatible strings information and example to this file, and insert clock relationship descriptions to this file? Wouldn’t it be strange to skip existing files and create a new one?
Quoted 2 lines
external oscillator's case, the oscillator is described in the device tree
Yes, we have "clkxtal" in the DT, which stands for external oscillator, All clocks in apmixedsys use "clkxtal" as the parent clock
Quoted 1 line
Dual license please (and the dts files).
Okay, thanks for your comment, I will update this in the next patch set
Quoted 2 lines
Why are this and other 1:1 factor clks needed? They look like placeholders. Please remove them.
Okay, thanks for your comment, I will update this in the next patch set
Quoted 1 line
Merge duplicate parent instances
We have considered this in the MT7986 basic clock driver, but I will check again. If corrections are needed, I will make changes in the next patch set.
Quoted 1 line
Leaking
clk_dataif some function return fail
Okay, thanks for your comment, I will update this in the next patch set
Quoted 3 lines
This file contains four drivers. They do not have depend on each other, and do not need to be in the same file. Please split them into differen files and preferably different patches
Okay, thanks for your comment, I will separate those clock drivers in the next patch set
Quoted 1 line
Is there any particular reason for
arch_initcall
We have considered this in MT7986 basic clock driver, and use CLK_OF_DECLARE instead of arch_initcall.
Another question:
Should the clock patches in "Add basic SoC support for MediaTek mt7986" need to be separated into another patch series, such as MT8195 "Mediatek MT8195 clock support" ?
Regards
Sam
Re: [PATCH 01/12] dt-bindings: clock: mediatek: document clk bindings for mediatek mt7986 SoC
Quoted 26 lines · Sam Shih
Hi,
Quoted 14 lines · Chen-Yu Tsai
Furthermore, based on the driver patch and the fact that they share the
same compatible string, it seems you shouldn't need to have two compatible
strings for two identical hardware blocks. The need for separate entries
to have different clock names is an implementation detail. Please consider
using and supporting clock-output-names.Also, please check out the MT8195 clock driver series [1]. I'm guessing
a lot of the comments apply to this one as well.
Quoted 6 lines
I have organized your comments in "Mediatek MT8195 clock support" series into the following list, reply to your here:
Quoted 2 lines
dt-binding: Move the not-to-be-exposed clock to driver directory or simply left out
Okay, thanks for your comment, I will update this in the next patch set
See the following file for an example:
https://elixir.bootlin.com/linux/latest/source/drivers/clk/sunxi-ng/ccu-sun50i-a64.h
I think this is definitely optional, but it makes it safer in that other drivers would not be able to use the non-exported intermediate clocks.
Quoted 9 lines
Quoted 2 lines
describe some of the clock relations between the various clock controllers
I have checked the files in "
Documentation/devicetree/bindings/arm/mediatek", It seems that all MediaTek SoC clocks are simply described by each controller, like "mediatek,infracfg.txt" and "mediatek,topckgen.txt", and those document only include compatible strings information and examples. Can we insert the clock relationship of MT7986 directlly in common documents?
What I meant was that since each clock controller hardware block has one or many clock inputs, these should be described in the binding as required "clocks" and "clock-names" properties.
So it's not about describing the actual relationship or clock tree, but just having the inputs accurately described.
Quoted 4 lines
Or we should add a new "mediatek,mt7986-clock.yaml" and move compatible strings information and example to this file, and insert clock relationship descriptions to this file? Wouldn’t it be strange to skip existing files and create a new one?
I think that is a question for the device tree binding maintainer, Rob. At least for Mediatek stuff, there seem to be many separate files.
Quoted 4 lines
Quoted 2 lines
external oscillator's case, the oscillator is described in the device tree
Yes, we have "clkxtal" in the DT, which stands for external oscillator, All clocks in apmixedsys use "clkxtal" as the parent clock
So for the apmixedsys device node, it would at least have something like:
clocks = <&clkxtal>;
clock-names = "xtal";For topckgen, since it has xtal and some PLLs from apmixedsys as inputs:
clocks = <&clkxtal>, <&apmixedsys CLK_ID_PLLXXX>, <&apmixedsysCLK_ID_PLLYYY>;
clock-names = "xtal", "pllXXX", "pllYYY"
The above is just an example. You should adapt it to fit your hardware description. And the bindings should describe what is required. Note that the clock names used here are local to this device node. They do not need to match what the clock driver uses for the global name. So just go with something descriptive.
The point is, cross hardware block dependencies should be clearly described in the device tree, instead of implicitly buried in the clock drivers.
Quoted 6 lines
Quoted 1 line
Dual license please (and the dts files).
Okay, thanks for your comment, I will update this in the next patch set
Quoted 2 lines
Why are this and other 1:1 factor clks needed? They look like placeholders. Please remove them.
Okay, thanks for your comment, I will update this in the next patch set
Ideally the clock driver would use the device tree to get local references for this, but that is going to require some rework to Mediatek's common clock code.
Quoted 11 lines
Quoted 1 line
Merge duplicate parent instances
We have considered this in the MT7986 basic clock driver, but I will check again. If corrections are needed, I will make changes in the next patch set.
Quoted 1 line
Leaking
clk_dataif some function return failOkay, thanks for your comment, I will update this in the next patch set
Quoted 3 lines
This file contains four drivers. They do not have depend on each other, and do not need to be in the same file. Please split them into differen files and preferably different patches
Quoted 6 lines
Okay, thanks for your comment, I will separate those clock drivers in the next patch set
Quoted 1 line
Is there any particular reason for
arch_initcallWe have considered this in MT7986 basic clock driver, and use
CLK_OF_DECLAREinstead ofarch_initcall.
Having to sequence clock registration manually is likely a symptom of inadequate clock dependency handling. So if the drivers are only using global clock names to describe parents, what happens is that even if the parent isn't in the system yet, the registration is allowed to succeed. However since the parent clock isn't available yet, any calculations involving it, such as calculating clock rates, will yield invalid results, such as 0 clock rate.
Quoted 4 lines
Another question:
Should the clock patches in "Add basic SoC support for MediaTek mt7986" need to be separated into another patch series, such as MT8195 "Mediatek MT8195 clock support" ?
Nope. The MT8195 team seems to be splitting things up by module, with the device tree being its own separate module. Ideally you want to send drivers along with the related device tree changes, so people reviewing can get a sense of how things work. And if the hardware is publicly available, people can actually test the changes. We can't do that if the device tree changes aren't bundled together.
Regards
ChenYu
Re: [PATCH 01/12] dt-bindings: clock: mediatek: document clk bindings for mediatek mt7986 SoC
Hi,
Sorry for the late reply,I have prepared v2 patches, but for some of your suggestions, I think it is a bit difficult to apply to our drivers.
Quoted 21 lines · Chen-Yu Tsai
Quoted 19 lines · Sam Shih
Hi,
Quoted 15 lines · Chen-Yu Tsai
Furthermore, based on the driver patch and the fact that they share
the
same compatible string, it seems you shouldn't need to have two compatible
strings for two identical hardware blocks. The need for separate entries
to have different clock names is an implementation detail. Please consider
using and supporting clock-output-names.Also, please check out the MT8195 clock driver series [1]. I'm guessing
a lot of the comments apply to this one as well.
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Regards
ChenYu[1]
Quoted 13 lines
Quoted 6 lines
Quoted 2 lines
I have organized your comments in "Mediatek MT8195 clock support" series into the following list, reply to your here:
Quoted 6 lines
Quoted 3 lines
dt-binding: Move the not-to-be-exposed clock to driver directory or
simply left outOkay, thanks for your comment, I will update this in the next patch set
Quoted 9 lines
See the following file for an example:
drivers would not be able to use the non-exported intermediate clocks.
Thanks, I well delete some clocks that are not expected to be use in device tree.
Quoted 58 lines
Quoted 11 lines
Quoted 2 lines
describe some of the clock relations between the various clock controllers
I have checked the files in "
Documentation/devicetree/bindings/arm/mediatek", It seems that all MediaTek SoC clocks are simply described by each controller, like "mediatek,infracfg.txt" and "mediatek,topckgen.txt", and those document
only include compatible strings information and examples. Can we insert the clock relationship of MT7986 directlly in common documents?What I meant was that since each clock controller hardware block has one or many clock inputs, these should be described in the binding as required "clocks" and "clock-names" properties.
So it's not about describing the actual relationship or clock tree, but just having the inputs accurately described.
Quoted 6 lines
Or we should add a new "mediatek,mt7986-clock.yaml" and move compatible
strings information and example to this file, and insert clock relationship descriptions to this file? Wouldn’t it be strange to skip
existing files and create a new one?I think that is a question for the device tree binding maintainer, Rob.
At least for Mediatek stuff, there seem to be many separate files.Quoted 7 lines
Quoted 3 lines
external oscillator's case, the oscillator is described in the device
treeYes, we have "clkxtal" in the DT, which stands for external oscillator,
All clocks in apmixedsys use "clkxtal" as the parent clockSo for the apmixedsys device node, it would at least have something like:
clocks = <&clkxtal>; clock-names = "xtal";For topckgen, since it has xtal and some PLLs from apmixedsys as inputs:
clocks = <&clkxtal>, <&apmixedsys
CLK_ID_PLLXXX>, <&apmixedsysCLK_ID_PLLYYY>;
clock-names = "xtal", "pllXXX", "pllYYY"The above is just an example. You should adapt it to fit your hardware
description. And the bindings should describe what is required. Note that the clock names used here are local to this device node. They do not need to match what the clock driver uses for the global name. So just go with something descriptive.
Quoted 5 lines
The point is, cross hardware block dependencies should be clearly described
in the device tree, instead of implicitly buried in the clock drivers.
Make cross hardware block dependencies clearly described in the device tree seems unsuitable between "topck" and "infra" block of mt7986.
In your example, passing "clkxtal" from the device tree seems ok. Even for topckgen, passing "clkxtal", "pllXXX", "pllYYY" from the device tree and using these clocks as the parent clock of topckgen also seems to work. It is feasible because it is not much clock between these two hardware blocks.
But for the clock releationship between "topckg" and "infra", they are very complicated in hardware design. There are dozens of clocks and interact with each other. If we want to describe these relationships in the device tree, we need to use a big array inside the device tree and make device tree looks very messy. Therefore, our previous method of writing a clock driver is to use the global clock and descripbe the clock releationship inside the clock driver.
______ ________ ________
clkxtal --| | | |.x.| |
|apmix.|--| topck |.x.| infra |
| |--| |.x.|_______|
| | | | ________
|______| | |--| |
| |--| ethsys |
|_______| |________|Maybe we should keep the clock relationship in our clock driver like the previous mediatek clock drivers.
Quoted 8 lines
Quoted 8 lines
Quoted 1 line
Dual license please (and the dts files).
Okay, thanks for your comment, I will update this in the next patch set
Quoted 2 lines
Why are this and other 1:1 factor clks needed? They look like placeholders. Please remove them.
Quoted 9 lines
Quoted 2 lines
Okay, thanks for your comment, I will update this in the next patch set
Ideally the clock driver would use the device tree to get local references
for this, but that is going to require some rework to Mediatek's common
clock code.
To transfer the clock relationship through the device tree, it is necessary to make modifications to the common part of the mtk clock driver that has been merged, and may also modify other mediatek clock drivers.
Let's move to the source code:
apmixedsys {
...
}topckgen {
...
clocks = ... , <&apmixedsys NET2PLL> , ... ;
clock-names = ... , "net2pll" , ... ;
}char *parent_name;
for each clk in <device_tree_node>:
parent_name = __clk_get_name(of_clk_get(np, i))
(armada-37xx-tbg.c use this method to get global name of parent clock)
Ideally, we should use the parent_name variable above to create a clock, But mtk_fixed_factor expects input const strings
void mtk_clk_register_factors(const struct mtk_fixed_factor *clks, ...)
struct mtk_fixed_factor {
...
const char *name;
const char *parent;
...
};So we still need to use pre-defined clock name in static const clock table even we can get clock name as variable from device tree.
static const struct mtk_fixed_factor top_divs[] __initconst = {
...
FACTOR(CK_TOP_NET2_D4_D2, "net2_d4_d2", "net2pll", 1, 8),
FACTOR(CK_TOP_NET2_D3_D2, "net2_d3_d2", "net2pll", 1, 2),
...
}Quoted 42 lines
Quoted 18 lines
Quoted 1 line
Merge duplicate parent instances
We have considered this in the MT7986 basic clock driver, but I will
check again. If corrections are needed, I will make changes in the next
patch set.Quoted 1 line
Leaking
clk_dataif some function return failOkay, thanks for your comment, I will update this in the next patch set
Quoted 4 lines
This file contains four drivers. They do not have depend on each other, and do not need to be in the same file. Please split them into
differen files and preferably different patchesQuoted 8 lines
Okay, thanks for your comment, I will separate those clock drivers in
the next patch setQuoted 1 line
Is there any particular reason for
arch_initcallWe have considered this in MT7986 basic clock driver, and use
CLK_OF_DECLAREinstead ofarch_initcall.Having to sequence clock registration manually is likely a symptom of inadequate clock dependency handling. So if the drivers are only using
global clock names to describe parents, what happens is that even if the parent isn't in the system yet, the registration is allowed to succeed. However since the parent clock isn't available yet, any calculations involving it, such as calculating clock rates, will yield invalid results, such as 0 clock rate.Quoted 5 lines
Another question:
Should the clock patches in "Add basic SoC support for MediaTek mt7986"
need to be separated into another patch series, such as MT8195 "Mediatek MT8195 clock support" ?
Quoted 10 lines
Nope. The MT8195 team seems to be splitting things up by module, with the device tree being its own separate module. Ideally you want to send
drivers along with the related device tree changes, so people reviewing
can get a sense of how things work. And if the hardware is publicly available, people can actually test the changes. We can't do that if the
device tree changes aren't bundled together.
OK, I will keep clock patches and basic part patches in the same series (patches v2)
Regards,
Sam
Thanks, I will delete some clocks that are not expected to b
Re: [PATCH 01/12] dt-bindings: clock: mediatek: document clk bindings for mediatek mt7986 SoC
Quoted 150 lines · Sam Shih
Hi,
Sorry for the late reply,I have prepared v2 patches, but for some of your suggestions, I think it is a bit difficult to apply to our drivers.
Quoted 21 lines · Chen-Yu Tsai
Quoted 19 lines · Sam Shih
Hi,
Quoted 15 lines · Chen-Yu Tsai
Furthermore, based on the driver patch and the fact that they share
the
same compatible string, it seems you shouldn't need to have two compatible
strings for two identical hardware blocks. The need for separate entries
to have different clock names is an implementation detail. Please consider
using and supporting clock-output-names.Also, please check out the MT8195 clock driver series [1]. I'm guessing
a lot of the comments apply to this one as well.Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Regards
ChenYu[1]
Quoted 13 lines
Quoted 6 lines
Quoted 2 lines
I have organized your comments in "Mediatek MT8195 clock support" series into the following list, reply to your here:
Quoted 6 lines
Quoted 3 lines
dt-binding: Move the not-to-be-exposed clock to driver directory or
simply left outOkay, thanks for your comment, I will update this in the next patch set
Quoted 9 lines
See the following file for an example:
drivers would not be able to use the non-exported intermediate clocks.
Thanks, I well delete some clocks that are not expected to be use in device tree.
Quoted 58 lines
Quoted 11 lines
Quoted 2 lines
describe some of the clock relations between the various clock controllers
I have checked the files in "
Documentation/devicetree/bindings/arm/mediatek", It seems that all MediaTek SoC clocks are simply described by each controller, like "mediatek,infracfg.txt" and "mediatek,topckgen.txt", and those document
only include compatible strings information and examples. Can we insert the clock relationship of MT7986 directlly in common documents?What I meant was that since each clock controller hardware block has one or many clock inputs, these should be described in the binding as required "clocks" and "clock-names" properties.
So it's not about describing the actual relationship or clock tree, but just having the inputs accurately described.
Quoted 6 lines
Or we should add a new "mediatek,mt7986-clock.yaml" and move compatible
strings information and example to this file, and insert clock relationship descriptions to this file? Wouldn’t it be strange to skip
existing files and create a new one?I think that is a question for the device tree binding maintainer, Rob.
At least for Mediatek stuff, there seem to be many separate files.Quoted 7 lines
Quoted 3 lines
external oscillator's case, the oscillator is described in the device
treeYes, we have "clkxtal" in the DT, which stands for external oscillator,
All clocks in apmixedsys use "clkxtal" as the parent clockSo for the apmixedsys device node, it would at least have something like:
clocks = <&clkxtal>; clock-names = "xtal";For topckgen, since it has xtal and some PLLs from apmixedsys as inputs:
clocks = <&clkxtal>, <&apmixedsys
CLK_ID_PLLXXX>, <&apmixedsysCLK_ID_PLLYYY>;
clock-names = "xtal", "pllXXX", "pllYYY"The above is just an example. You should adapt it to fit your hardware
description. And the bindings should describe what is required. Note that the clock names used here are local to this device node. They do not need to match what the clock driver uses for the global name. So just go with something descriptive.Quoted 5 lines
The point is, cross hardware block dependencies should be clearly described
in the device tree, instead of implicitly buried in the clock drivers.Make cross hardware block dependencies clearly described in the device tree seems unsuitable between "topck" and "infra" block of mt7986.
In your example, passing "clkxtal" from the device tree seems ok. Even for topckgen, passing "clkxtal", "pllXXX", "pllYYY" from the device tree and using these clocks as the parent clock of topckgen also seems to work. It is feasible because it is not much clock between these two hardware blocks.
But for the clock releationship between "topckg" and "infra", they are very complicated in hardware design. There are dozens of clocks and interact with each other. If we want to describe these relationships in the device tree, we need to use a big array inside the device tree and make device tree looks very messy. Therefore, our previous method of writing a clock driver is to use the global clock and descripbe the clock releationship inside the clock driver.
______ ________ ________ clkxtal --| | | |.x.| | |apmix.|--| topck |.x.| infra | | |--| |.x.|_______| | | | | ________ |______| | |--| | | |--| ethsys | |_______| |________|
Quoted 2 lines
Maybe we should keep the clock relationship in our clock driver like the previous mediatek clock drivers.
Are you saying that some clocks in topck have inputs from infra? If so, then that's a nasty circular dependency to deal with.
And to be fair, many Mediatek device tree bindings already take a large number of clocks, so I think adding a few more isn't too bad.
Quoted 57 lines
Quoted 8 lines
Quoted 8 lines
Quoted 1 line
Dual license please (and the dts files).
Okay, thanks for your comment, I will update this in the next patch set
Quoted 2 lines
Why are this and other 1:1 factor clks needed? They look like placeholders. Please remove them.
Quoted 9 lines
Quoted 2 lines
Okay, thanks for your comment, I will update this in the next patch set
Ideally the clock driver would use the device tree to get local references
for this, but that is going to require some rework to Mediatek's common
clock code.To transfer the clock relationship through the device tree, it is necessary to make modifications to the common part of the mtk clock driver that has been merged, and may also modify other mediatek clock drivers.
Let's move to the source code:
apmixedsys { ... }topckgen { ... clocks = ... , <&apmixedsys NET2PLL> , ... ; clock-names = ... , "net2pll" , ... ; }char
*parent_name;
for each clk in <device_tree_node>:
parent_name=__clk_get_name(of_clk_get(np, i))(
armada-37xx-tbg.cuse this method to get global name of parent clock)Ideally, we should use the
parent_namevariable above to create a clock, Butmtk_fixed_factorexpects input const stringsvoid
mtk_clk_register_factors(const struct mtk_fixed_factor *clks, ...)struct mtk_fixed_factor { ... const char *name; const char *parent; ... };So we still need to use pre-defined clock name in static const clock table even we can get clock name as variable from device tree.
Quoted 6 lines
static const struct mtk_fixed_factor top_divs[] __initconst = {
...
FACTOR(CK_TOP_NET2_D4_D2, "net2_d4_d2", "net2pll", 1, 8),
FACTOR(CK_TOP_NET2_D3_D2, "net2_d3_d2", "net2pll", 1, 2),
...
}Right. I'm not asking you to do this right away. This will end up being a long migration over multiple releases. But it makes sense to get the device tree bindings sorted out first.
I believe the solution is to move to struct clk_parent_data to replace the pure strings. New internal APIs for the Mediatek clock driver would need to be added, and all the drivers slowly migrated over. Probably also a good time to migrate to clk_hw_*_register.
ChenYu
Quoted 59 lines
Quoted 42 lines
Quoted 18 lines
Quoted 1 line
Merge duplicate parent instances
We have considered this in the MT7986 basic clock driver, but I will
check again. If corrections are needed, I will make changes in the next
patch set.Quoted 1 line
Leaking
clk_dataif some function return failOkay, thanks for your comment, I will update this in the next patch set
Quoted 4 lines
This file contains four drivers. They do not have depend on each other, and do not need to be in the same file. Please split them into
differen files and preferably different patchesQuoted 8 lines
Okay, thanks for your comment, I will separate those clock drivers in
the next patch setQuoted 1 line
Is there any particular reason for
arch_initcallWe have considered this in MT7986 basic clock driver, and use
CLK_OF_DECLAREinstead ofarch_initcall.Having to sequence clock registration manually is likely a symptom of inadequate clock dependency handling. So if the drivers are only using
global clock names to describe parents, what happens is that even if the parent isn't in the system yet, the registration is allowed to succeed. However since the parent clock isn't available yet, any calculations involving it, such as calculating clock rates, will yield invalid results, such as 0 clock rate.Quoted 5 lines
Another question:
Should the clock patches in "Add basic SoC support for MediaTek mt7986"
need to be separated into another patch series, such as MT8195 "Mediatek MT8195 clock support" ?Quoted 10 lines
Nope. The MT8195 team seems to be splitting things up by module, with the device tree being its own separate module. Ideally you want to send
drivers along with the related device tree changes, so people reviewing
can get a sense of how things work. And if the hardware is publicly available, people can actually test the changes. We can't do that if the
device tree changes aren't bundled together.OK, I will keep clock patches and basic part patches in the same series (patches v2)
Regards,
Sam
Thanks, I will delete some clocks that are not expected to b
Re: [PATCH 01/12] dt-bindings: clock: mediatek: document clk bindings for mediatek mt7986 SoC
Quoted 36 lines · Chen-Yu Tsai
Quoted 34 lines · Sam Shih
Hi,
Sorry for the late reply,I have prepared v2 patches, but for some of your suggestions, I think
it is a bit difficult to apply to our drivers.Quoted 25 lines · Chen-Yu Tsai
Quoted 23 lines · Sam Shih
Hi,
Quoted 19 lines · Chen-Yu Tsai
Furthermore, based on the driver patch and the fact that they share
the
same compatible string, it seems you shouldn't need to have two
compatible
strings for two identical hardware blocks. The need for separate
entries
to have different clock names is an implementation detail. Please
consider
using and supporting clock-output-names.Also, please check out the MT8195 clock driver series [1]. I'm
guessing
a lot of the comments apply to this one as well.
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Regards
ChenYu[1]
Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
Quoted 2 lines
I have organized your comments in "Mediatek MT8195 clock support"
series into the following list, reply to your here:Quoted 4 lines
dt-binding: Move the not-to-be-exposed clock to driver directory
or
simply left out
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 3 lines
Okay, thanks for your comment, I will update this in the next patch
setSee the following file for an example:
Quoted 123 lines
Quoted 123 lines
Quoted 4 lines
drivers would not be able to use the non-exported intermediate clocks.
Thanks, I well delete some clocks that are not expected to be use in
device tree.Quoted 74 lines
Quoted 4 lines
Quoted 3 lines
describe some of the clock relations between the various clock
controllersQuoted 11 lines
I have checked the files in "
Documentation/devicetree/bindings/arm/mediatek", It seems that all
MediaTek SoC clocks are simply described by each controller, like
"mediatek,infracfg.txt" and "mediatek,topckgen.txt", and those document
only include compatible strings information and examples. Can we insert the clock relationship of MT7986 directlly in common
documents?What I meant was that since each clock controller hardware block has
one or many clock inputs, these should be described in the binding
as required "clocks" and "clock-names" properties.So it's not about describing the actual relationship or clock tree,
but just having the inputs accurately described.Quoted 7 lines
Or we should add a new "mediatek,mt7986-clock.yaml" and move compatible
strings information and example to this file, and insert clock relationship descriptions to this file? Wouldn’t it be strange to
skip
existing files and create a new one?I think that is a question for the device tree binding maintainer,
Rob.
At least for Mediatek stuff, there seem to be many separate files.Quoted 8 lines
Quoted 4 lines
external oscillator's case, the oscillator is described in the
device
treeYes, we have "clkxtal" in the DT, which stands for external oscillator,
All clocks in apmixedsys use "clkxtal" as the parent clockSo for the apmixedsys device node, it would at least have something
like:clocks = <&clkxtal>; clock-names = "xtal";For topckgen, since it has xtal and some PLLs from apmixedsys as inputs:
clocks = <&clkxtal>, <&apmixedsys
CLK_ID_PLLXXX>,
<&apmixedsysCLK_ID_PLLYYY>;
clock-names = "xtal", "pllXXX", "pllYYY"The above is just an example. You should adapt it to fit your hardware
description. And the bindings should describe what is required. Note
that the clock names used here are local to this device node. They do
not need to match what the clock driver uses for the global name. So
just go with something descriptive.Quoted 5 lines
The point is, cross hardware block dependencies should be clearly described
in the device tree, instead of implicitly buried in the clock drivers.Make cross hardware block dependencies clearly described in the device
tree seems unsuitable between "topck" and "infra" block of mt7986.In your example, passing "clkxtal" from the device tree seems ok. Even
for topckgen, passing "clkxtal", "pllXXX", "pllYYY" from the device tree and using these clocks as the parent clock of topckgen also seems
to work. It is feasible because it is not much clock between these two
hardware blocks.But for the clock releationship between "topckg" and "infra", they are
very complicated in hardware design. There are dozens of clocks and interact with each other. If we want to describe these relationships in
the device tree, we need to use a big array inside the device tree and
make device tree looks very messy. Therefore, our previous method of
writing a clock driver is to use the global clock and descripbe the clock releationship inside the clock driver.______ ________ ________ clkxtal --| | | |.x.| | |apmix.|--| topck |.x.| infra | | |--| |.x.|_______| | | | | ________ |______| | |--| | | |--| ethsys | |_______| |________|
Quoted 7 lines
Quoted 3 lines
Maybe we should keep the clock relationship in our clock driver like
the previous mediatek clock drivers.Are you saying that some clocks in topck have inputs from infra? If so, then that's a nasty circular dependency to deal with.
Sorry I didn't express it very clearly, I mean that there are many clock connections between "ckgen" and "infra" (unlike apmixedsys and ckgen). Although most of the clocks in "infra" is derived from "ckgen". But if we want to use the device tree, we will need to pass dozens of clocks.
Quoted 4 lines
And to be fair, many Mediatek device tree bindings already take a large
number of clocks, so I think adding a few more isn't too bad.
It will become something like:
clocks = <&topckgen CLK_TOP_NFI1X>,<&topckgen CLK_TOP_USB_EQ_RX250M>,<&topckgen CLK_TOP_USB_TX250M>,<&topckgen CLK_TOP_USB_LN0_CK>,<&topckgen CLK_TOP_USB_CDR_CK>,<&topckgen CLK_TOP_SPINFI_BCK>,<&topckgen CLK_TOP_I2C_BCK>,<&topckgen CLK_TOP_PEXTP_TL>,<&topckgen CLK_TOP_EMMC_250M>,<&topckgen CLK_TOP_EMMC_416M>,<&topckgen CLK_TOP_F_26M_ADC_CK>,<&topckgen CLK_TOP_SYSAXI>,<&topckgen CLK_TOP_NETSYS_WED_MCU>,<&topckgen CLK_TOP_NETSYS_2X>,<&topckgen CLK_TOP_SGM_325M>,<&topckgen CLK_TOP_A1SYS>,<&topckgen CLK_TOP_EIP_B>,<&topckgen CLK_TOP_F26M>,<&topckgen CLK_TOP_AUD_L>,<&topckgen CLK_TOP_A_TUNER>,<&topckgen CLK_TOP_U2U3_REF>,<&topckgen CLK_TOP_U2U3_SYS>,<&topckgen CLK_TOP_U2U3_XHCI>,<&topckgen CLK_TOP_AP2CNN_HOST>,<&topckgen CLK_TOP_NFI1X_SEL>,<&topckgen CLK_TOP_SPINFI_SEL>,<&topckgen CLK_TOP_SPI_SEL>,<&topckgen CLK_TOP_SPIM_MST_SEL>,<&topckgen CLK_TOP_UART_SEL>,<&topckgen CLK_TOP_PWM_SEL>,<&topckgen CLK_TOP_I2C_SEL>,<&topckgen CLK_TOP_PEXTP_TL_SEL>,<&topckgen CLK_TOP_EMMC_250M_SEL>,<&topckgen CLK_TOP_EMMC_416M_SEL>,<&topckgen CLK_TOP_F_26M_ADC_SEL>,<&topckgen CLK_TOP_DRAMC_SEL>,<&topckgen CLK_TOP_DRAMC_MD32_SEL>,<&topckgen CLK_TOP_SYSAXI_SEL>,<&topckgen CLK_TOP_SYSAPB_SEL>,<&topckgen CLK_TOP_ARM_DB_MAIN_SEL>,<&topckgen CLK_TOP_ARM_DB_JTSEL>,<&topckgen CLK_TOP_NETSYS_SEL>,<&topckgen CLK_TOP_NETSYS_500M_SEL>,<&topckgen CLK_TOP_NETSYS_MCU_SEL>,<&topckgen CLK_TOP_NETSYS_2X_SEL>,<&topckgen CLK_TOP_SGM_325M_SEL>,<&topckgen CLK_TOP_SGM_REG_SEL>,<&topckgen CLK_TOP_A1SYS_SEL>,<&topckgen CLK_TOP_CONN_MCUSYS_SEL>,<&topckgen CLK_TOP_EIP_B_SEL>,<&topckgen CLK_TOP_PCIE_PHY_SEL>,<&topckgen CLK_TOP_USB3_PHY_SEL>,<&topckgen CLK_TOP_F26M_SEL>,<&topckgen CLK_TOP_AUD_L_SEL>,<&topckgen CLK_TOP_A_TUNER_SEL>,<&topckgen CLK_TOP_U2U3_SEL>,<&topckgen CLK_TOP_U2U3_SYS_SEL>,<&topckgen CLK_TOP_U2U3_XHCI_SEL>,<&topckgen CLK_TOP_DA_U2_REFSEL>,<&topckgen CLK_TOP_DA_U2_CK_1P_SEL>,<&topckgen CLK_TOP_AP2CNN_HOST_SEL>;
I don't think this is okay in device tree...
Quoted 85 lines
Quoted 63 lines
Quoted 13 lines
Quoted 6 lines
Quoted 1 line
Dual license please (and the dts files).
Okay, thanks for your comment, I will update this in the next patch
setQuoted 6 lines
Quoted 2 lines
Why are this and other 1:1 factor clks needed? They look like placeholders. Please remove them.
Okay, thanks for your comment, I will update this in the next patch
setQuoted 6 lines
Ideally the clock driver would use the device tree to get local references
for this, but that is going to require some rework to Mediatek's common
clock code.To transfer the clock relationship through the device tree, it is necessary to make modifications to the common part of the mtk clock driver that has been merged, and may also modify other mediatek clock
drivers.Let's move to the source code:
apmixedsys { ... }topckgen { ... clocks = ... , <&apmixedsys NET2PLL> , ... ; clock-names = ... , "net2pll" , ... ; }char
*parent_name;
for each clk in <device_tree_node>:
parent_name=__clk_get_name(of_clk_get(np, i))(
armada-37xx-tbg.cuse this method to get global name of parent clock)Ideally, we should use the
parent_namevariable above to create a clock, Butmtk_fixed_factorexpects input const stringsvoid
mtk_clk_register_factors(const struct mtk_fixed_factor *clks, ...)struct mtk_fixed_factor { ... const char *name; const char *parent; ... };So we still need to use pre-defined clock name in static const clock
table even we can get clock name as variable from device tree.Quoted 6 lines
static const struct mtk_fixed_factor top_divs[] __initconst = { ... FACTOR(CK_TOP_NET2_D4_D2, "net2_d4_d2", "net2pll", 1, 8), FACTOR(CK_TOP_NET2_D3_D2, "net2_d3_d2", "net2pll", 1, 2), ... }Right. I'm not asking you to do this right away. This will end up being
a long migration over multiple releases. But it makes sense to get the
device tree bindings sorted out first.I believe the solution is to move to
struct clk_parent_datato replace
the pure strings. New internal APIs for the Mediatek clock driver would
need to be added, and all the drivers slowly migrated over. Probably also
a good time to migrate to clk_hw_*_register.
Quoted 74 lines
ChenYu
Quoted 69 lines
Quoted 50 lines
Quoted 27 lines
Quoted 1 line
Merge duplicate parent instances
We have considered this in the MT7986 basic clock driver, but I will
check again. If corrections are needed, I will make changes in the
next
patch set.Quoted 1 line
Leaking
clk_dataif some function return failOkay, thanks for your comment, I will update this in the next patch
setQuoted 6 lines
This file contains four drivers. They do not have depend on each
other, and do not need to be in the same file. Please split them
into
differen files and preferably different patchesOkay, thanks for your comment, I will separate those clock drivers
in
the next patch setQuoted 4 lines
Quoted 1 line
Is there any particular reason for
arch_initcallWe have considered this in MT7986 basic clock driver, and use
CLK_OF_DECLAREinstead ofarch_initcall.Having to sequence clock registration manually is likely a symptom of
inadequate clock dependency handling. So if the drivers are only using
global clock names to describe parents, what happens is that even if
the parent isn't in the system yet, the registration is allowed to
succeed. However since the parent clock isn't available yet, any calculations involving it, such as calculating clock rates, will yield invalid results, such as 0 clock rate.Quoted 5 lines
Another question:
Should the clock patches in "Add basic SoC support for MediaTek mt7986"
need to be separated into another patch series, such as MT8195 "Mediatek MT8195 clock support" ?Quoted 14 lines
Nope. The MT8195 team seems to be splitting things up by module, with
the device tree being its own separate module. Ideally you want to
send
drivers along with the related device tree changes, so people reviewing
can get a sense of how things work. And if the hardware is publicly
available, people can actually test the changes. We can't do that if
the
device tree changes aren't bundled together.OK, I will keep clock patches and basic part patches in the same series
(patches v2)
Regards,
Sam
Quoted 1 line
Quoted 1 line
Thanks, I will delete some clocks that are not expected to b
PATCH 2/12 clk: mediatek: add mt7986 clock IDs
In reply to Sam ShihAdd MT7986 clock dt-bindings, include topckgen, apmixedsys, infracfg, infracfg_ao, and ethernet subsystem clocks.
--- include/dt-bindings/clock/mt7986-clk.h | 244 +++++++++++++++++++++++++ 1 file changed, 244 insertions(+) create mode 100644 include/dt-bindings/clock/mt7986-clk.h
include/dt-bindings/clock/mt7986-clk.h new file +244 -0
@@ -0,0 +1,244 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* +* Copyright (c) 2021 MediaTek Inc. +* Author: Sam Shih <[email protected]> +*/ + +#ifndef _DT_BINDINGS_CLK_MT7986_H +#define _DT_BINDINGS_CLK_MT7986_H + +/* INFRACFG */ + +#define CK_INFRA_CK_F26M 0 +#define CK_INFRA_UART 1 +#define CK_INFRA_ISPI0 2 +#define CK_INFRA_I2C 3 +#define CK_INFRA_ISPI1 4 +#define CK_INFRA_PWM 5 +#define CK_INFRA_66M_MCK 6 +#define CK_INFRA_CK_F32K 7 +#define CK_INFRA_PCIE_CK 8 +#define CK_INFRA_PWM_BCK 9 +#define CK_INFRA_PWM_CK1 10 +#define CK_INFRA_PWM_CK2 11 +#define CK_INFRA_133M_HCK 12 +#define CK_INFRA_EIP_CK 13 +#define CK_INFRA_66M_PHCK 14 +#define CK_INFRA_FAUD_L_CK 15 +#define CK_INFRA_FAUD_AUD_CK 16 +#define CK_INFRA_FAUD_EG2_CK 17 +#define CK_INFRA_I2CS_CK 18 +#define CK_INFRA_MUX_UART0 19 +#define CK_INFRA_MUX_UART1 20 +#define CK_INFRA_MUX_UART2 21 +#define CK_INFRA_NFI_CK 22 +#define CK_INFRA_SPINFI_CK 23 +#define CK_INFRA_MUX_SPI0 24 +#define CK_INFRA_MUX_SPI1 25 +#define CK_INFRA_RTC_32K 26 +#define CK_INFRA_FMSDC_CK 27 +#define CK_INFRA_FMSDC_HCK_CK 28 +#define CK_INFRA_PERI_133M 29 +#define CK_INFRA_133M_PHCK 30 +#define CK_INFRA_USB_SYS_CK 31 +#define CK_INFRA_USB_CK 32 +#define CK_INFRA_USB_XHCI_CK 33 +#define CK_INFRA_PCIE_GFMUX_TL_O_PRE 34 +#define CK_INFRA_F26M_CK0 35 +#define CK_INFRA_HD_133M 36 +#define CLK_INFRA_NR_CLK 37 + +/* TOPCKGEN */ + +#define CK_TOP_CB_CKSQ_40M 0 +#define CK_TOP_CB_M_416M 1 +#define CK_TOP_CB_M_D2 2 +#define CK_TOP_CB_M_D4 3 +#define CK_TOP_CB_M_D8 4 +#define CK_TOP_M_D8_D2 5 +#define CK_TOP_M_D3_D2 6 +#define CK_TOP_CB_MM_D2 7 +#define CK_TOP_CB_MM_D4 8 +#define CK_TOP_CB_MM_D8 9 +#define CK_TOP_MM_D8_D2 10 +#define CK_TOP_MM_D3_D8 11 +#define CK_TOP_CB_U2_PHYD_CK 12 +#define CK_TOP_CB_APLL2_196M 13 +#define CK_TOP_APLL2_D4 14 +#define CK_TOP_CB_NET1_D4 15 +#define CK_TOP_CB_NET1_D5 16 +#define CK_TOP_NET1_D5_D2 17 +#define CK_TOP_NET1_D5_D4 18 +#define CK_TOP_NET1_D8_D2 19 +#define CK_TOP_NET1_D8_D4 20 +#define CK_TOP_CB_NET2_800M 21 +#define CK_TOP_CB_NET2_D4 22 +#define CK_TOP_NET2_D4_D2 23 +#define CK_TOP_NET2_D3_D2 24 +#define CK_TOP_CB_WEDMCU_760M 25 +#define CK_TOP_WEDMCU_D5_D2 26 +#define CK_TOP_CB_SGM_325M 27 +#define CK_TOP_CB_CKSQ_40M_D2 28 +#define CK_TOP_CB_RTC_32K 29 +#define CK_TOP_CB_RTC_32P7K 30 +#define CK_TOP_NFI1X 31 +#define CK_TOP_USB_EQ_RX250M 32 +#define CK_TOP_USB_TX250M 33 +#define CK_TOP_USB_LN0_CK 34 +#define CK_TOP_USB_CDR_CK 35 +#define CK_TOP_SPINFI_BCK 36 +#define CK_TOP_I2C_BCK 37 +#define CK_TOP_PEXTP_TL 38 +#define CK_TOP_EMMC_250M 39 +#define CK_TOP_EMMC_416M 40 +#define CK_TOP_F_26M_ADC_CK 41 +#define CK_TOP_SYSAXI 42 +#define CK_TOP_NETSYS_WED_MCU 43 +#define CK_TOP_NETSYS_2X 44 +#define CK_TOP_SGM_325M 45 +#define CK_TOP_A1SYS 46 +#define CK_TOP_EIP_B 47 +#define CK_TOP_F26M 48 +#define CK_TOP_AUD_L 49 +#define CK_TOP_A_TUNER 50 +#define CK_TOP_U2U3_REF 51 +#define CK_TOP_U2U3_SYS 52 +#define CK_TOP_U2U3_XHCI 53 +#define CK_TOP_AP2CNN_HOST 54 +#define CK_TOP_NFI1X_SEL 55 +#define CK_TOP_SPINFI_SEL 56 +#define CK_TOP_SPI_SEL 57 +#define CK_TOP_SPIM_MST_SEL 58 +#define CK_TOP_UART_SEL 59 +#define CK_TOP_PWM_SEL 60 +#define CK_TOP_I2C_SEL 61 +#define CK_TOP_PEXTP_TL_SEL 62 +#define CK_TOP_EMMC_250M_SEL 63 +#define CK_TOP_EMMC_416M_SEL 64 +#define CK_TOP_F_26M_ADC_SEL 65 +#define CK_TOP_DRAMC_SEL 66 +#define CK_TOP_DRAMC_MD32_SEL 67 +#define CK_TOP_SYSAXI_SEL 68 +#define CK_TOP_SYSAPB_SEL 69 +#define CK_TOP_ARM_DB_MAIN_SEL 70 +#define CK_TOP_ARM_DB_JTSEL 71 +#define CK_TOP_NETSYS_SEL 72 +#define CK_TOP_NETSYS_500M_SEL 73 +#define CK_TOP_NETSYS_MCU_SEL 74 +#define CK_TOP_NETSYS_2X_SEL 75 +#define CK_TOP_SGM_325M_SEL 76 +#define CK_TOP_SGM_REG_SEL 77 +#define CK_TOP_A1SYS_SEL 78 +#define CK_TOP_CONN_MCUSYS_SEL 79 +#define CK_TOP_EIP_B_SEL 80 +#define CK_TOP_PCIE_PHY_SEL 81 +#define CK_TOP_USB3_PHY_SEL 82 +#define CK_TOP_F26M_SEL 83 +#define CK_TOP_AUD_L_SEL 84 +#define CK_TOP_A_TUNER_SEL 85 +#define CK_TOP_U2U3_SEL 86 +#define CK_TOP_U2U3_SYS_SEL 87 +#define CK_TOP_U2U3_XHCI_SEL 88 +#define CK_TOP_DA_U2_REFSEL 89 +#define CK_TOP_DA_U2_CK_1P_SEL 90 +#define CK_TOP_AP2CNN_HOST_SEL 91 +#define CLK_TOP_NR_CLK 92 + +/* INFRACFG_AO */ + +#define CK_INFRA_UART0_SEL 0 +#define CK_INFRA_UART1_SEL 1 +#define CK_INFRA_UART2_SEL 2 +#define CK_INFRA_SPI0_SEL 3 +#define CK_INFRA_SPI1_SEL 4 +#define CK_INFRA_PWM1_SEL 5 +#define CK_INFRA_PWM2_SEL 6 +#define CK_INFRA_PWM_BSEL 7 +#define CK_INFRA_PCIE_SEL 8 +#define CK_INFRA_GPT_STA 9 +#define CK_INFRA_PWM_HCK 10 +#define CK_INFRA_PWM_STA 11 +#define CK_INFRA_PWM1_CK 12 +#define CK_INFRA_PWM2_CK 13 +#define CK_INFRA_CQ_DMA_CK 14 +#define CK_INFRA_EIP97_CK 15 +#define CK_INFRA_AUD_BUS_CK 16 +#define CK_INFRA_AUD_26M_CK 17 +#define CK_INFRA_AUD_L_CK 18 +#define CK_INFRA_AUD_AUD_CK 19 +#define CK_INFRA_AUD_EG2_CK 20 +#define CK_INFRA_DRAMC_26M_CK 21 +#define CK_INFRA_DBG_CK 22 +#define CK_INFRA_AP_DMA_CK 23 +#define CK_INFRA_SEJ_CK 24 +#define CK_INFRA_SEJ_13M_CK 25 +#define CK_INFRA_THERM_CK 26 +#define CK_INFRA_I2CO_CK 27 +#define CK_INFRA_UART0_CK 28 +#define CK_INFRA_UART1_CK 29 +#define CK_INFRA_UART2_CK 30 +#define CK_INFRA_NFI1_CK 31 +#define CK_INFRA_SPINFI1_CK 32 +#define CK_INFRA_NFI_HCK_CK 33 +#define CK_INFRA_SPI0_CK 34 +#define CK_INFRA_SPI1_CK 35 +#define CK_INFRA_SPI0_HCK_CK 36 +#define CK_INFRA_SPI1_HCK_CK 37 +#define CK_INFRA_FRTC_CK 38 +#define CK_INFRA_MSDC_CK 39 +#define CK_INFRA_MSDC_HCK_CK 40 +#define CK_INFRA_MSDC_133M_CK 41 +#define CK_INFRA_MSDC_66M_CK 42 +#define CK_INFRA_ADC_26M_CK 43 +#define CK_INFRA_ADC_FRC_CK 44 +#define CK_INFRA_FBIST2FPC_CK 45 +#define CK_INFRA_IUSB_133_CK 46 +#define CK_INFRA_IUSB_66M_CK 47 +#define CK_INFRA_IUSB_SYS_CK 48 +#define CK_INFRA_IUSB_CK 49 +#define CK_INFRA_IPCIE_CK 50 +#define CK_INFRA_IPCIE_PIPE_CK 51 +#define CK_INFRA_IPCIER_CK 52 +#define CK_INFRA_IPCIEB_CK 53 +#define CK_INFRA_TRNG_CK 54 +#define CLK_INFRA_AO_NR_CLK 55 + +/* APMIXEDSYS */ + +#define CK_APMIXED_ARMPLL 0 +#define CK_APMIXED_NET2PLL 1 +#define CK_APMIXED_MMPLL 2 +#define CK_APMIXED_SGMPLL 3 +#define CK_APMIXED_WEDMCUPLL 4 +#define CK_APMIXED_NET1PLL 5 +#define CK_APMIXED_MPLL 6 +#define CK_APMIXED_APLL2 7 +#define CLK_APMIXED_NR_CLK 8 + +/* SGMIISYS_0 */ + +#define CK_SGM0_TX_EN 0 +#define CK_SGM0_RX_EN 1 +#define CK_SGM0_CK0_EN 2 +#define CK_SGM0_CDR_CK0_EN 3 +#define CLK_SGMII0_NR_CLK 4 + +/* SGMIISYS_1 */ + +#define CK_SGM1_TX_EN 0 +#define CK_SGM1_RX_EN 1 +#define CK_SGM1_CK1_EN 2 +#define CK_SGM1_CDR_CK1_EN 3 +#define CLK_SGMII1_NR_CLK 4 + +/* ETHSYS */ + +#define CK_ETH_FE_EN 0 +#define CK_ETH_GP2_EN 1 +#define CK_ETH_GP1_EN 2 +#define CK_ETH_WOCPU1_EN 3 +#define CK_ETH_WOCPU0_EN 4 +#define CLK_ETH_NR_CLK 5 + +#endif /* _DT_BINDINGS_CLK_MT7986_H */ +
Signature · 1 line
2.29.2
Rob Herring
maintainerRe: [PATCH 02/12] clk: mediatek: add mt7986 clock IDs
Quoted 15 lines · Sam Shih
Add MT7986 clock dt-bindings, include topckgen, apmixedsys, infracfg,
infracfg_ao, and ethernet subsystem clocks.Signed-off-by Sam Shih <[email protected]>--- include/dt-bindings/clock/mt7986-clk.h | 244 +++++++++++++++++++++++++ 1 file changed, 244 insertions(+) create mode 100644 include/dt-bindings/clock/mt7986-clk.hinclude/dt-bindings/clock/mt7986-clk.h new file +0
-0
include/dt-bindings/clock/mt7986-clk.h line 1
Quoted 1 line
include/dt-bindings/clock/mt7986-clk.h new file +1
-0@@ -0,0 +1,244 @@ +/* SPDX-License-Identifier: GPL-2.0-only */
Dual license please.
This should be part of the binding patch.
Quoted 247 lines
include/dt-bindings/clock/mt7986-clk.h new file +243
-0+/* +* Copyright (c) 2021 MediaTek Inc. +* Author: Sam Shih <[email protected]> +*/ + +#ifndef _DT_BINDINGS_CLK_MT7986_H +#define _DT_BINDINGS_CLK_MT7986_H + +/* INFRACFG */ + +#define CK_INFRA_CK_F26M 0 +#define CK_INFRA_UART 1 +#define CK_INFRA_ISPI0 2 +#define CK_INFRA_I2C 3 +#define CK_INFRA_ISPI1 4 +#define CK_INFRA_PWM 5 +#define CK_INFRA_66M_MCK 6 +#define CK_INFRA_CK_F32K 7 +#define CK_INFRA_PCIE_CK 8 +#define CK_INFRA_PWM_BCK 9 +#define CK_INFRA_PWM_CK1 10 +#define CK_INFRA_PWM_CK2 11 +#define CK_INFRA_133M_HCK 12 +#define CK_INFRA_EIP_CK 13 +#define CK_INFRA_66M_PHCK 14 +#define CK_INFRA_FAUD_L_CK 15 +#define CK_INFRA_FAUD_AUD_CK 16 +#define CK_INFRA_FAUD_EG2_CK 17 +#define CK_INFRA_I2CS_CK 18 +#define CK_INFRA_MUX_UART0 19 +#define CK_INFRA_MUX_UART1 20 +#define CK_INFRA_MUX_UART2 21 +#define CK_INFRA_NFI_CK 22 +#define CK_INFRA_SPINFI_CK 23 +#define CK_INFRA_MUX_SPI0 24 +#define CK_INFRA_MUX_SPI1 25 +#define CK_INFRA_RTC_32K 26 +#define CK_INFRA_FMSDC_CK 27 +#define CK_INFRA_FMSDC_HCK_CK 28 +#define CK_INFRA_PERI_133M 29 +#define CK_INFRA_133M_PHCK 30 +#define CK_INFRA_USB_SYS_CK 31 +#define CK_INFRA_USB_CK 32 +#define CK_INFRA_USB_XHCI_CK 33 +#define CK_INFRA_PCIE_GFMUX_TL_O_PRE 34 +#define CK_INFRA_F26M_CK0 35 +#define CK_INFRA_HD_133M 36 +#define CLK_INFRA_NR_CLK 37 + +/* TOPCKGEN */ + +#define CK_TOP_CB_CKSQ_40M 0 +#define CK_TOP_CB_M_416M 1 +#define CK_TOP_CB_M_D2 2 +#define CK_TOP_CB_M_D4 3 +#define CK_TOP_CB_M_D8 4 +#define CK_TOP_M_D8_D2 5 +#define CK_TOP_M_D3_D2 6 +#define CK_TOP_CB_MM_D2 7 +#define CK_TOP_CB_MM_D4 8 +#define CK_TOP_CB_MM_D8 9 +#define CK_TOP_MM_D8_D2 10 +#define CK_TOP_MM_D3_D8 11 +#define CK_TOP_CB_U2_PHYD_CK 12 +#define CK_TOP_CB_APLL2_196M 13 +#define CK_TOP_APLL2_D4 14 +#define CK_TOP_CB_NET1_D4 15 +#define CK_TOP_CB_NET1_D5 16 +#define CK_TOP_NET1_D5_D2 17 +#define CK_TOP_NET1_D5_D4 18 +#define CK_TOP_NET1_D8_D2 19 +#define CK_TOP_NET1_D8_D4 20 +#define CK_TOP_CB_NET2_800M 21 +#define CK_TOP_CB_NET2_D4 22 +#define CK_TOP_NET2_D4_D2 23 +#define CK_TOP_NET2_D3_D2 24 +#define CK_TOP_CB_WEDMCU_760M 25 +#define CK_TOP_WEDMCU_D5_D2 26 +#define CK_TOP_CB_SGM_325M 27 +#define CK_TOP_CB_CKSQ_40M_D2 28 +#define CK_TOP_CB_RTC_32K 29 +#define CK_TOP_CB_RTC_32P7K 30 +#define CK_TOP_NFI1X 31 +#define CK_TOP_USB_EQ_RX250M 32 +#define CK_TOP_USB_TX250M 33 +#define CK_TOP_USB_LN0_CK 34 +#define CK_TOP_USB_CDR_CK 35 +#define CK_TOP_SPINFI_BCK 36 +#define CK_TOP_I2C_BCK 37 +#define CK_TOP_PEXTP_TL 38 +#define CK_TOP_EMMC_250M 39 +#define CK_TOP_EMMC_416M 40 +#define CK_TOP_F_26M_ADC_CK 41 +#define CK_TOP_SYSAXI 42 +#define CK_TOP_NETSYS_WED_MCU 43 +#define CK_TOP_NETSYS_2X 44 +#define CK_TOP_SGM_325M 45 +#define CK_TOP_A1SYS 46 +#define CK_TOP_EIP_B 47 +#define CK_TOP_F26M 48 +#define CK_TOP_AUD_L 49 +#define CK_TOP_A_TUNER 50 +#define CK_TOP_U2U3_REF 51 +#define CK_TOP_U2U3_SYS 52 +#define CK_TOP_U2U3_XHCI 53 +#define CK_TOP_AP2CNN_HOST 54 +#define CK_TOP_NFI1X_SEL 55 +#define CK_TOP_SPINFI_SEL 56 +#define CK_TOP_SPI_SEL 57 +#define CK_TOP_SPIM_MST_SEL 58 +#define CK_TOP_UART_SEL 59 +#define CK_TOP_PWM_SEL 60 +#define CK_TOP_I2C_SEL 61 +#define CK_TOP_PEXTP_TL_SEL 62 +#define CK_TOP_EMMC_250M_SEL 63 +#define CK_TOP_EMMC_416M_SEL 64 +#define CK_TOP_F_26M_ADC_SEL 65 +#define CK_TOP_DRAMC_SEL 66 +#define CK_TOP_DRAMC_MD32_SEL 67 +#define CK_TOP_SYSAXI_SEL 68 +#define CK_TOP_SYSAPB_SEL 69 +#define CK_TOP_ARM_DB_MAIN_SEL 70 +#define CK_TOP_ARM_DB_JTSEL 71 +#define CK_TOP_NETSYS_SEL 72 +#define CK_TOP_NETSYS_500M_SEL 73 +#define CK_TOP_NETSYS_MCU_SEL 74 +#define CK_TOP_NETSYS_2X_SEL 75 +#define CK_TOP_SGM_325M_SEL 76 +#define CK_TOP_SGM_REG_SEL 77 +#define CK_TOP_A1SYS_SEL 78 +#define CK_TOP_CONN_MCUSYS_SEL 79 +#define CK_TOP_EIP_B_SEL 80 +#define CK_TOP_PCIE_PHY_SEL 81 +#define CK_TOP_USB3_PHY_SEL 82 +#define CK_TOP_F26M_SEL 83 +#define CK_TOP_AUD_L_SEL 84 +#define CK_TOP_A_TUNER_SEL 85 +#define CK_TOP_U2U3_SEL 86 +#define CK_TOP_U2U3_SYS_SEL 87 +#define CK_TOP_U2U3_XHCI_SEL 88 +#define CK_TOP_DA_U2_REFSEL 89 +#define CK_TOP_DA_U2_CK_1P_SEL 90 +#define CK_TOP_AP2CNN_HOST_SEL 91 +#define CLK_TOP_NR_CLK 92 + +/* INFRACFG_AO */ + +#define CK_INFRA_UART0_SEL 0 +#define CK_INFRA_UART1_SEL 1 +#define CK_INFRA_UART2_SEL 2 +#define CK_INFRA_SPI0_SEL 3 +#define CK_INFRA_SPI1_SEL 4 +#define CK_INFRA_PWM1_SEL 5 +#define CK_INFRA_PWM2_SEL 6 +#define CK_INFRA_PWM_BSEL 7 +#define CK_INFRA_PCIE_SEL 8 +#define CK_INFRA_GPT_STA 9 +#define CK_INFRA_PWM_HCK 10 +#define CK_INFRA_PWM_STA 11 +#define CK_INFRA_PWM1_CK 12 +#define CK_INFRA_PWM2_CK 13 +#define CK_INFRA_CQ_DMA_CK 14 +#define CK_INFRA_EIP97_CK 15 +#define CK_INFRA_AUD_BUS_CK 16 +#define CK_INFRA_AUD_26M_CK 17 +#define CK_INFRA_AUD_L_CK 18 +#define CK_INFRA_AUD_AUD_CK 19 +#define CK_INFRA_AUD_EG2_CK 20 +#define CK_INFRA_DRAMC_26M_CK 21 +#define CK_INFRA_DBG_CK 22 +#define CK_INFRA_AP_DMA_CK 23 +#define CK_INFRA_SEJ_CK 24 +#define CK_INFRA_SEJ_13M_CK 25 +#define CK_INFRA_THERM_CK 26 +#define CK_INFRA_I2CO_CK 27 +#define CK_INFRA_UART0_CK 28 +#define CK_INFRA_UART1_CK 29 +#define CK_INFRA_UART2_CK 30 +#define CK_INFRA_NFI1_CK 31 +#define CK_INFRA_SPINFI1_CK 32 +#define CK_INFRA_NFI_HCK_CK 33 +#define CK_INFRA_SPI0_CK 34 +#define CK_INFRA_SPI1_CK 35 +#define CK_INFRA_SPI0_HCK_CK 36 +#define CK_INFRA_SPI1_HCK_CK 37 +#define CK_INFRA_FRTC_CK 38 +#define CK_INFRA_MSDC_CK 39 +#define CK_INFRA_MSDC_HCK_CK 40 +#define CK_INFRA_MSDC_133M_CK 41 +#define CK_INFRA_MSDC_66M_CK 42 +#define CK_INFRA_ADC_26M_CK 43 +#define CK_INFRA_ADC_FRC_CK 44 +#define CK_INFRA_FBIST2FPC_CK 45 +#define CK_INFRA_IUSB_133_CK 46 +#define CK_INFRA_IUSB_66M_CK 47 +#define CK_INFRA_IUSB_SYS_CK 48 +#define CK_INFRA_IUSB_CK 49 +#define CK_INFRA_IPCIE_CK 50 +#define CK_INFRA_IPCIE_PIPE_CK 51 +#define CK_INFRA_IPCIER_CK 52 +#define CK_INFRA_IPCIEB_CK 53 +#define CK_INFRA_TRNG_CK 54 +#define CLK_INFRA_AO_NR_CLK 55 + +/* APMIXEDSYS */ + +#define CK_APMIXED_ARMPLL 0 +#define CK_APMIXED_NET2PLL 1 +#define CK_APMIXED_MMPLL 2 +#define CK_APMIXED_SGMPLL 3 +#define CK_APMIXED_WEDMCUPLL 4 +#define CK_APMIXED_NET1PLL 5 +#define CK_APMIXED_MPLL 6 +#define CK_APMIXED_APLL2 7 +#define CLK_APMIXED_NR_CLK 8 + +/* SGMIISYS_0 */ + +#define CK_SGM0_TX_EN 0 +#define CK_SGM0_RX_EN 1 +#define CK_SGM0_CK0_EN 2 +#define CK_SGM0_CDR_CK0_EN 3 +#define CLK_SGMII0_NR_CLK 4 + +/* SGMIISYS_1 */ + +#define CK_SGM1_TX_EN 0 +#define CK_SGM1_RX_EN 1 +#define CK_SGM1_CK1_EN 2 +#define CK_SGM1_CDR_CK1_EN 3 +#define CLK_SGMII1_NR_CLK 4 + +/* ETHSYS */ + +#define CK_ETH_FE_EN 0 +#define CK_ETH_GP2_EN 1 +#define CK_ETH_GP1_EN 2 +#define CK_ETH_WOCPU1_EN 3 +#define CK_ETH_WOCPU0_EN 4 +#define CLK_ETH_NR_CLK 5 + +#endif /* _DT_BINDINGS_CLK_MT7986_H */ +Signature · 1 line
2.29.2
PATCH 3/12 clk: mediatek: add mt7986 clock support
In reply to Sam ShihAdd MT7986 clock support, include topckgen, apmixedsys, infracfg, infracfg_ao, and ethernet subsystem clocks.
--- drivers/clk/mediatek/Kconfig | 17 + drivers/clk/mediatek/Makefile | 2 + drivers/clk/mediatek/clk-mt7986-eth.c | 132 ++++++ drivers/clk/mediatek/clk-mt7986.c | 610 ++++++++++++++++++++++++++ 4 files changed, 761 insertions(+) create mode 100644 drivers/clk/mediatek/clk-mt7986-eth.c create mode 100644 drivers/clk/mediatek/clk-mt7986.c
drivers/clk/mediatek/Kconfig +17 -0
@@ -344,6 +344,23 @@ config COMMON_CLK_MT7629_HIFSYS This driver supports MediaTek MT7629 HIFSYS clocks providing to PCI-E and USB. +config COMMON_CLK_MT7986 + bool "Clock driver for MediaTek MT7986" + depends on ARCH_MEDIATEK || COMPILE_TEST + select COMMON_CLK_MEDIATEK + default ARCH_MEDIATEK + help + This driver supports MediaTek MT7986 basic clocks and clocks + required for various periperals found on MediaTek. + +config COMMON_CLK_MT7986_ETHSYS + bool "Clock driver for MediaTek MT7986 ETHSYS" + depends on COMMON_CLK_MT7986 + default COMMON_CLK_MT7986 + help + This driver add support for clocks for Ethernet and SGMII + required on MediaTek MT7986 SoC. + config COMMON_CLK_MT8135 bool "Clock driver for MediaTek MT8135" depends on (ARCH_MEDIATEK && ARM) || COMPILE_TEST
drivers/clk/mediatek/Makefile +2 -0
@@ -46,6 +46,8 @@ obj-$(CONFIG_COMMON_CLK_MT7622_AUDSYS) += clk-mt7622-aud.o obj-$(CONFIG_COMMON_CLK_MT7629) += clk-mt7629.o obj-$(CONFIG_COMMON_CLK_MT7629_ETHSYS) += clk-mt7629-eth.o obj-$(CONFIG_COMMON_CLK_MT7629_HIFSYS) += clk-mt7629-hif.o +obj-$(CONFIG_COMMON_CLK_MT7986) += clk-mt7986.o +obj-$(CONFIG_COMMON_CLK_MT7986_ETHSYS) += clk-mt7986-eth.o obj-$(CONFIG_COMMON_CLK_MT8135) += clk-mt8135.o obj-$(CONFIG_COMMON_CLK_MT8167) += clk-mt8167.o obj-$(CONFIG_COMMON_CLK_MT8167_AUDSYS) += clk-mt8167-aud.o
drivers/clk/mediatek/clk-mt7986-eth.c new file +132 -0
132 lines not on this page · raw message
drivers/clk/mediatek/clk-mt7986.c new file +610 -0
610 lines not on this page · raw message
Signature · 1 line
2.29.2
PATCH 4/12 pinctrl: mediatek: moore: use pin number in mtk_pin_desc instead of array index
In reply to Sam ShihCertain SoC are missing the middle part gpios in consecutive pins, it's better to use pin number in mtk_pin_desc instead of array index for the extensibility
--- drivers/pinctrl/mediatek/pinctrl-moore.c | 61 ++++++++++++++++++++++++ 1 file changed, 61 insertions(+)
drivers/pinctrl/mediatek/pinctrl-moore.c +61 -0
@@ -35,6 +35,19 @@ static const struct pin_config_item mtk_conf_items[] = { };#endif +static int mtk_pin_desc_lookup (struct mtk_pinctrl *hw,int pin) +{ +int idx; + +for (idx =0 ; idx < hw->soc->npins ; idx++) +if (hw->soc->pins[idx].number == pin) +break ; +if (idx < hw->soc->npins) +return idx; + +return -EINVAL ; +} +static int mtk_pinmux_set_mux (struct pinctrl_dev *pctldev,unsigned int selector,unsigned int group) {
@@ -74,6 +87,13 @@ static int mtk_pinmux_gpio_request_enable(struct pinctrl_dev *pctldev, {struct mtk_pinctrl *hw =pinctrl_dev_get_drvdata (pctldev);const struct mtk_pin_desc *desc; +int err; + + err =mtk_pin_desc_lookup (hw, pin); +if (err >=0 ) + pin = err; +else +return err; desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];
@@ -87,6 +107,13 @@ static int mtk_pinmux_gpio_set_direction(struct pinctrl_dev *pctldev, {struct mtk_pinctrl *hw =pinctrl_dev_get_drvdata (pctldev);const struct mtk_pin_desc *desc; +int err; + + err =mtk_pin_desc_lookup (hw, pin); +if (err >=0 ) + pin = err; +else +return err; desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];
@@ -102,6 +129,12 @@ static int mtk_pinconf_get(struct pinctrl_dev *pctldev,int val, val2, err, reg, ret =1 ;const struct mtk_pin_desc *desc; + err =mtk_pin_desc_lookup (hw, pin); +if (err >=0 ) + pin = err; +else +return err; + desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];switch (param) {
@@ -217,6 +250,12 @@ static int mtk_pinconf_set(struct pinctrl_dev *pctldev, unsigned int pin,u32 reg, param, arg;int cfg, err =0 ; + err =mtk_pin_desc_lookup (hw, pin); +if (err >=0 ) + pin = err; +else +return err; + desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];for (cfg =0 ; cfg < num_configs; cfg++) {
@@ -434,6 +473,12 @@ static int mtk_gpio_get(struct gpio_chip *chip, unsigned int gpio)const struct mtk_pin_desc *desc;int value, err; + err =mtk_pin_desc_lookup (hw, gpio); +if (err >=0 ) + gpio = err; +else +return err; + desc = (const struct mtk_pin_desc *)&hw->soc->pins[gpio]; err =mtk_hw_get_value (hw, desc,PINCTRL_PIN_REG_DI , &value);
@@ -447,6 +492,15 @@ static void mtk_gpio_set(struct gpio_chip *chip, unsigned int gpio, int value) {struct mtk_pinctrl *hw =gpiochip_get_data (chip);const struct mtk_pin_desc *desc; +int err; + + err =mtk_pin_desc_lookup (hw, gpio); +if (err >=0 ) { + gpio = err; + }else { +dev_err (hw->dev,"Failed to set gpio %d\n" , gpio); +return ; + } desc = (const struct mtk_pin_desc *)&hw->soc->pins[gpio];
@@ -488,6 +542,13 @@ static int mtk_gpio_set_config(struct gpio_chip *chip, unsigned int offset,struct mtk_pinctrl *hw =gpiochip_get_data (chip);const struct mtk_pin_desc *desc;u32 debounce; +int err; + + err =mtk_pin_desc_lookup (hw, offset); +if (err >=0 ) + offset = err; +else +return err; desc = (const struct mtk_pin_desc *)&hw->soc->pins[offset];
Signature · 1 line
2.29.2
Sean Wang
maintainerRe: [PATCH 04/12] pinctrl: mediatek: moore: use pin number in mtk_pin_desc instead of array index
Hi Sam,
Quoted 4 lines · Sam Shih
Certain SoC are missing the middle part gpios in consecutive pins, it's better to use pin number in
mtk_pin_descinstead of array index for the extensibility
Now the driver pin number has to be consistent with the array index because the driver would use pin number as the array index to fetch the pin descriptor.
For those missing GPIOs, we could just fill out .name in struct mtk_pin_desc as NULL to indicate the pin is unavailable for users (pin not ballout) on the certain SoC and then allow us to reuse all of the pinctrl operations with minimal modification.
Quoted 32 lines
Signed-off-by Sam Shih <[email protected]>--- drivers/pinctrl/mediatek/pinctrl-moore.c | 61 ++++++++++++++++++++++++ 1 file changed, 61 insertions(+)drivers/pinctrl/mediatek/pinctrl-moore.c +13
-0@@ -35,6 +35,19 @@ static const struct pin_config_item mtk_conf_items[] = { };#endif +static int mtk_pin_desc_lookup (struct mtk_pinctrl *hw,int pin) +{ +int idx; + +for (idx =0 ; idx < hw->soc->npins ; idx++) +if (hw->soc->pins[idx].number == pin) +break ; +if (idx < hw->soc->npins) +return idx; + +return -EINVAL ; +} +static int mtk_pinmux_set_mux (struct pinctrl_dev *pctldev,unsigned int selector,unsigned int group) {@@ -74,6 +87,13 @@ static int mtk_pinmux_gpio_request_enable(struct pinctrl_dev *pctldev, {
drivers/pinctrl/mediatek/pinctrl-moore.c static int mtk_pinmux_gpio_request_enable(struct pinctrl_dev *pctldev, · line 97
Quoted 10 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +7
-0struct mtk_pinctrl *hw =pinctrl_dev_get_drvdata (pctldev);const struct mtk_pin_desc *desc; +int err; + + err =mtk_pin_desc_lookup (hw, pin); +if (err >=0 ) + pin = err; +else +return err;
We can drop it and use the following snippet instead
desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];
/* !desc->name to show the pin is not ballout */
if (!desc->name)
return -ENOTSUPP;
Quoted 4 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +0
-0desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];@@ -87,6 +107,13 @@ static int mtk_pinmux_gpio_set_direction(struct pinctrl_dev *pctldev, {
drivers/pinctrl/mediatek/pinctrl-moore.c static int mtk_pinmux_gpio_set_direction(struct pinctrl_dev *pctldev, · line 117
Quoted 10 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +7
-0struct mtk_pinctrl *hw =pinctrl_dev_get_drvdata (pctldev);const struct mtk_pin_desc *desc; +int err; + + err =mtk_pin_desc_lookup (hw, pin); +if (err >=0 ) + pin = err; +else +return err;
Ditto
drivers/pinctrl/mediatek/pinctrl-moore.c static int mtk_pinconf_get(struct pinctrl_dev *pctldev, · line 137
Quoted 12 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +6
-0desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];@@ -102,6 +129,12 @@ static int mtk_pinconf_get(struct pinctrl_dev *pctldev,int val, val2, err, reg, ret =1 ;const struct mtk_pin_desc *desc; + err =mtk_pin_desc_lookup (hw, pin); +if (err >=0 ) + pin = err; +else +return err; +
Ditto
drivers/pinctrl/mediatek/pinctrl-moore.c static int mtk_pinconf_set(struct pinctrl_dev *pctldev, unsigned int pin, · line 258
Quoted 13 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +6
-0desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];switch (param) {@@ -217,6 +250,12 @@ static int mtk_pinconf_set(struct pinctrl_dev *pctldev, unsigned int pin,u32 reg, param, arg;int cfg, err =0 ; + err =mtk_pin_desc_lookup (hw, pin); +if (err >=0 ) + pin = err; +else +return err; +
Ditto
drivers/pinctrl/mediatek/pinctrl-moore.c static int mtk_gpio_get(struct gpio_chip *chip, unsigned int gpio) · line 481
Quoted 13 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +6
-0desc = (const struct mtk_pin_desc *)&hw->soc->pins[pin];for (cfg =0 ; cfg < num_configs; cfg++) {@@ -434,6 +473,12 @@ static int mtk_gpio_get(struct gpio_chip *chip, unsigned int gpio)const struct mtk_pin_desc *desc;int value, err; + err =mtk_pin_desc_lookup (hw, gpio); +if (err >=0 ) + gpio = err; +else +return err; +
Ditto
Quoted 7 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +0
-0desc = (const struct mtk_pin_desc *)&hw->soc->pins[gpio]; err =mtk_hw_get_value (hw, desc,PINCTRL_PIN_REG_DI , &value);@@ -447,6 +492,15 @@ static void mtk_gpio_set(struct gpio_chip *chip, unsigned int gpio, int value) {struct mtk_pinctrl *hw =gpiochip_get_data (chip);const struct mtk_pin_desc *desc;
drivers/pinctrl/mediatek/pinctrl-moore.c static void mtk_gpio_set(struct gpio_chip *chip, unsigned int gpio, int value) · line 504
Quoted 10 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +9
-0+int err; + + err =mtk_pin_desc_lookup (hw, gpio); +if (err >=0 ) { + gpio = err; + }else { +dev_err (hw->dev,"Failed to set gpio %d\n" , gpio); +return ; + }
Ditto
Quoted 4 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +0
-0desc = (const struct mtk_pin_desc *)&hw->soc->pins[gpio];@@ -488,6 +542,13 @@ static int mtk_gpio_set_config(struct gpio_chip *chip, unsigned int offset,struct mtk_pinctrl *hw =gpiochip_get_data (chip);
drivers/pinctrl/mediatek/pinctrl-moore.c static int mtk_gpio_set_config(struct gpio_chip *chip, unsigned int offset, · line 552
Quoted 10 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +7
-0const struct mtk_pin_desc *desc;u32 debounce; +int err; + + err =mtk_pin_desc_lookup (hw, offset); +if (err >=0 ) + offset = err; +else +return err;
Ditto
Quoted 5 lines
drivers/pinctrl/mediatek/pinctrl-moore.c +0
-1desc = (const struct mtk_pin_desc *)&hw->soc->pins[offset]; --2.29.2
PATCH 5/12 dt-bindings: pinctrl: update bindings for MT7986 SoC
In reply to Sam ShihThis updates bindings for MT7986 pinctrl driver. The difference of pinctrl between mt7986a and mt7986b is that pin-41 to pin-65 do not exist on mt7986b
--- .../bindings/pinctrl/pinctrl-mt7622.txt | 284 ++++++++++++++++++ 1 file changed, 284 insertions(+)
Documentation/devicetree/bindings/pinctrl/pinctrl-mt7622.txt +284 -0
@@ -4,6 +4,8 @@ Required properties for the root node: - compatible: Should be one of the following "mediatek,mt7622-pinctrl" for MT7622 SoC "mediatek,mt7629-pinctrl" for MT7629 SoC + "mediatek,mt7986a-pinctrl" for MT7986a SoC + "mediatek,mt7986b-pinctrl" for MT7986b SoC - reg: offset and length of the pinctrl space - gpio-controller: Marks the device node as a GPIO controller.
@@ -455,6 +457,288 @@ Valid values for groups are: "wf0_5g" "wifi" 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 +== Valid values for pins, function and groups on MT7986a == + +Valid values for pins are: +pins can be referenced via the pin names as the below table shown and the +related physical number is also put ahead of those names which helps cross +references to pins between groups to know whether pins assignment conflict +happens among devices try to acquire those available pins. + + Pin #: Valid values for pins + ----------------------------- + PIN 0: "SYS_WATCHDOG" + PIN 1: "WF2G_LED" + PIN 2: "WF5G_LED" + PIN 3: "I2C_SCL" + PIN 4: "I2C_SDA" + PIN 5: "GPIO_0" + PIN 6: "GPIO_1" + PIN 7: "GPIO_2" + PIN 8: "GPIO_3" + PIN 9: "GPIO_4" + PIN 10: "GPIO_5" + PIN 11: "GPIO_6" + PIN 12: "GPIO_7" + PIN 13: "GPIO_8" + PIN 14: "GPIO_9" + PIN 15: "GPIO_10" + PIN 16: "GPIO_11" + PIN 17: "GPIO_12" + PIN 18: "GPIO_13" + PIN 19: "GPIO_14" + PIN 20: "GPIO_15" + PIN 21: "PWM0" + PIN 22: "PWM1" + PIN 23: "SPI0_CLK" + PIN 24: "SPI0_MOSI" + PIN 25: "SPI0_MISO" + PIN 26: "SPI0_CS" + PIN 27: "SPI0_HOLD" + PIN 28: "SPI0_WP" + PIN 29: "SPI1_CLK" + PIN 30: "SPI1_MOSI" + PIN 31: "SPI1_MISO" + PIN 32: "SPI1_CS" + PIN 33: "SPI2_CLK" + PIN 34: "SPI2_MOSI" + PIN 35: "SPI2_MISO" + PIN 36: "SPI2_CS" + PIN 37: "SPI2_HOLD" + PIN 38: "SPI2_WP" + PIN 39: "UART0_RXD" + PIN 40: "UART0_TXD" + PIN 41: "PCIE_PERESET_N" + PIN 42: "UART1_RXD" + PIN 43: "UART1_TXD" + PIN 44: "UART1_CTS" + PIN 45: "UART1_RTS" + PIN 46: "UART2_RXD" + PIN 47: "UART2_TXD" + PIN 48: "UART2_CTS" + PIN 49: "UART2_RTS" + PIN 50: "EMMC_DATA_0" + PIN 51: "EMMC_DATA_1" + PIN 52: "EMMC_DATA_2" + PIN 53: "EMMC_DATA_3" + PIN 54: "EMMC_DATA_4" + PIN 55: "EMMC_DATA_5" + PIN 56: "EMMC_DATA_6" + PIN 57: "EMMC_DATA_7" + PIN 58: "EMMC_CMD" + PIN 59: "EMMC_CK" + PIN 60: "EMMC_DSL" + PIN 61: "EMMC_RSTB" + PIN 62: "PCM_DTX" + PIN 63: "PCM_DRX" + PIN 64: "PCM_CLK" + PIN 65: "PCM_FS" + PIN 66: "MT7531_INT" + PIN 67: "SMI_MDC" + PIN 68: "SMI_MDIO" + PIN 69: "WF0_DIG_RESETB" + PIN 70: "WF0_CBA_RESETB" + PIN 71: "WF0_XO_REQ" + PIN 72: "WF0_TOP_CLK" + PIN 73: "WF0_TOP_DATA" + PIN 74: "WF0_HB1" + PIN 75: "WF0_HB2" + PIN 76: "WF0_HB3" + PIN 77: "WF0_HB4" + PIN 78: "WF0_HB0" + PIN 79: "WF0_HB0_B" + PIN 80: "WF0_HB5" + PIN 81: "WF0_HB6" + PIN 82: "WF0_HB7" + PIN 83: "WF0_HB8" + PIN 84: "WF0_HB9" + PIN 85: "WF0_HB10" + PIN 86: "WF1_DIG_RESETB" + PIN 87: "WF1_CBA_RESETB" + PIN 88: "WF1_XO_REQ" + PIN 89: "WF1_TOP_CLK" + PIN 90: "WF1_TOP_DATA" + PIN 91: "WF1_HB1" + PIN 92: "WF1_HB2" + PIN 93: "WF1_HB3" + PIN 94: "WF1_HB4" + PIN 95: "WF1_HB0" + PIN 96: "WF1_HB0_B" + PIN 97: "WF1_HB5" + PIN 98: "WF1_HB6" + PIN 99: "WF1_HB7" + PIN 100: "WF1_HB8" + +Valid values for function are: + "audio, "emmc", "eth", "i2c", "wifi", "led", "flash", "pcie", + "pwm", "spi", "uart", "watchdog" + +Valid values for groups are: +additional data is put followingly with valid value allowing us to know which +applicable function and which relevant pins (in pin#) are able applied for that +group. + + Valid value function pins (in pin#) + ------------------------------------------------------------------------- + "watchdog" "watchdog" 0 + "wifi_led" "led" 1, 2 + "i2c" "i2c" 3, 4 + "uart1_0" "uart" 7, 8, 9, 10 + "pcie_clk" "pcie" 9 + "pcie_wake" "pcie" 10 + "spi1_0" "spi" 11, 12, 13, 14 + "pwm1_1" "pwm" 20, + "pwm0" "pwm" 21, + "pwm1_0" "pwm" 22, + "snfi" "flash" 23, 24, 25, 26, 27, 28 + "spi1_2" "spi" 29, 30, 31, 32 + "emmc_45" "emmc" 22, 23, 24, 25, 26, 27, + 28, 29, 30, 31, 32 + "spi1_1" "spi" 23, 24, 25, 26 + "uart1_2" "uart" 29, 30, 31, 32 + "uart1_1" "uart" 23, 24, 25, 26 + "uart2_0" "uart" 29, 30, 31, 32 + "spi0" "spi" 33, 34, 35, 36 + "spi0_wp_hold" "spi" 37, 38 + "uart1_3_rx_tx" "uart" 35, 36 + "uart1_3_cts_rts" "uart" 37, 38 + "uart2_1" "uart" 33, 34, 35, 36 + "spi1_3" "spi" 33, 34, 35, 36 + "uart0" "uart" 39, 40 + "pcie_pereset" "pcie" 41 + "uart1" "uart" 42, 43, 44, 45 + "uart2" "uart" 46, 47, 48, 49 + "emmc_51" "emmc" 50, 51, 52, 53, 54, 55, + 56, 57, 57, 59, 60, 61 + "pcm" "audio" 62, 63, 64, 65 + "i2s" "audio" 62, 63, 64, 65 + "switch_int" "eth" 66 + "mdc_mdio" "eth" 67 + +== Valid values for pins, function and groups on MT7986b == + +Valid values for pins are: +pins can be referenced via the pin names as the below table shown and the +related physical number is also put ahead of those names which helps cross +references to pins between groups to know whether pins assignment conflict +happens among devices try to acquire those available pins. + + Pin #: Valid values for pins + ----------------------------- + PIN 0: "SYS_WATCHDOG" + PIN 1: "WF2G_LED" + PIN 2: "WF5G_LED" + PIN 3: "I2C_SCL" + PIN 4: "I2C_SDA" + PIN 5: "GPIO_0" + PIN 6: "GPIO_1" + PIN 7: "GPIO_2" + PIN 8: "GPIO_3" + PIN 9: "GPIO_4" + PIN 10: "GPIO_5" + PIN 11: "GPIO_6" + PIN 12: "GPIO_7" + PIN 13: "GPIO_8" + PIN 14: "GPIO_9" + PIN 15: "GPIO_10" + PIN 16: "GPIO_11" + PIN 17: "GPIO_12" + PIN 18: "GPIO_13" + PIN 19: "GPIO_14" + PIN 20: "GPIO_15" + PIN 21: "PWM0" + PIN 22: "PWM1" + PIN 23: "SPI0_CLK" + PIN 24: "SPI0_MOSI" + PIN 25: "SPI0_MISO" + PIN 26: "SPI0_CS" + PIN 27: "SPI0_HOLD" + PIN 28: "SPI0_WP" + PIN 29: "SPI1_CLK" + PIN 30: "SPI1_MOSI" + PIN 31: "SPI1_MISO" + PIN 32: "SPI1_CS" + PIN 33: "SPI2_CLK" + PIN 34: "SPI2_MOSI" + PIN 35: "SPI2_MISO" + PIN 36: "SPI2_CS" + PIN 37: "SPI2_HOLD" + PIN 38: "SPI2_WP" + PIN 39: "UART0_RXD" + PIN 40: "UART0_TXD" + PIN 66: "MT7531_INT" + PIN 67: "SMI_MDC" + PIN 68: "SMI_MDIO" + PIN 69: "WF0_DIG_RESETB" + PIN 70: "WF0_CBA_RESETB" + PIN 71: "WF0_XO_REQ" + PIN 72: "WF0_TOP_CLK" + PIN 73: "WF0_TOP_DATA" + PIN 74: "WF0_HB1" + PIN 75: "WF0_HB2" + PIN 76: "WF0_HB3" + PIN 77: "WF0_HB4" + PIN 78: "WF0_HB0" + PIN 79: "WF0_HB0_B" + PIN 80: "WF0_HB5" + PIN 81: "WF0_HB6" + PIN 82: "WF0_HB7" + PIN 83: "WF0_HB8" + PIN 84: "WF0_HB9" + PIN 85: "WF0_HB10" + PIN 86: "WF1_DIG_RESETB" + PIN 87: "WF1_CBA_RESETB" + PIN 88: "WF1_XO_REQ" + PIN 89: "WF1_TOP_CLK" + PIN 90: "WF1_TOP_DATA" + PIN 91: "WF1_HB1" + PIN 92: "WF1_HB2" + PIN 93: "WF1_HB3" + PIN 94: "WF1_HB4" + PIN 95: "WF1_HB0" + PIN 96: "WF1_HB0_B" + PIN 97: "WF1_HB5" + PIN 98: "WF1_HB6" + PIN 99: "WF1_HB7" + PIN 100: "WF1_HB8" + +Valid values for function are: + "emmc", "eth", "i2c", "wifi", "led", "flash", "pwm", "spi", + "uart", "watchdog" + +Valid values for groups are: +additional data is put followingly with valid value allowing us to know which +applicable function and which relevant pins (in pin#) are able applied for that +group. + + Valid value function pins (in pin#) + ------------------------------------------------------------------------- + "watchdog" "watchdog" 0 + "wifi_led" "led" 1, 2 + "i2c" "i2c" 3, 4 + "uart1_0" "uart" 7, 8, 9, 10 + "spi1_0" "spi" 11, 12, 13, 14 + "pwm1_1" "pwm" 20, + "pwm0" "pwm" 21, + "pwm1_0" "pwm" 22, + "snfi" "flash" 23, 24, 25, 26, 27, 28 + "spi1_2" "spi" 29, 30, 31, 32 + "emmc_45" "emmc" 22, 23, 24, 25, 26, 27, + 28, 29, 30, 31, 32 + "spi1_1" "spi" 23, 24, 25, 26 + "uart1_2" "uart" 29, 30, 31, 32 + "uart1_1" "uart" 23, 24, 25, 26 + "uart2_0" "uart" 29, 30, 31, 32 + "spi0" "spi" 33, 34, 35, 36 + "spi0_wp_hold" "spi" 37, 38 + "uart1_3_rx_tx" "uart" 35, 36 + "uart1_3_cts_rts" "uart" 37, 38 + "uart2_1" "uart" 33, 34, 35, 36 + "spi1_3" "spi" 33, 34, 35, 36 + "switch_int" "eth" 66 + "mdc_mdio" "eth" 67 + + Example: pio: pinctrl@10211000 {
Signature · 1 line
2.29.2
Rob Herring
maintainerRe: [PATCH 05/12] dt-bindings: pinctrl: update bindings for MT7986 SoC
Quoted 3 lines · Sam Shih
This updates bindings for MT7986 pinctrl driver. The difference of pinctrl between mt7986a and mt7986b is that pin-41 to pin-65 do not exist on mt7986b
Sounds like you could use the same compatible string. Wouldn't the DT for mt7986b simply not have any config for those pins?
Quoted 5 lines
Signed-off-by Sam Shih <[email protected]>--- .../bindings/pinctrl/pinctrl-mt7622.txt | 284 ++++++++++++++++++ 1 file changed, 284 insertions(+)
This is a big change. Please convert to a schema. It probably should be a separate schema.
PATCH 6/12 pinctrl: mediatek: add support for MT7986 SoC
In reply to Sam ShihThis commit includes pinctrl driver for Mediatek MT7986
The difference of pinctrl between mt7986a and mt7986b is that pin-41 to pin-65 do not exist on mt7986b
--- drivers/pinctrl/mediatek/Kconfig | 7 + drivers/pinctrl/mediatek/Makefile | 1 + drivers/pinctrl/mediatek/pinctrl-mt7986.c | 1640 +++++++++++++++++++++ 3 files changed, 1648 insertions(+) create mode 100644 drivers/pinctrl/mediatek/pinctrl-mt7986.c
drivers/pinctrl/mediatek/Kconfig +7 -0
@@ -119,6 +119,13 @@ config PINCTRL_MT7622 default ARM64 && ARCH_MEDIATEK select PINCTRL_MTK_MOORE +config PINCTRL_MT7986 + bool "Mediatek MT7986 pin control" + depends on OF + depends on ARM64 || COMPILE_TEST + default ARM64 && ARCH_MEDIATEK + select PINCTRL_MTK_MOORE + config PINCTRL_MT8167 bool "Mediatek MT8167 pin control" depends on OF
drivers/pinctrl/mediatek/Makefile +1 -0
@@ -17,6 +17,7 @@ obj-$(CONFIG_PINCTRL_MT6797) += pinctrl-mt6797.o obj-$(CONFIG_PINCTRL_MT7622) += pinctrl-mt7622.o obj-$(CONFIG_PINCTRL_MT7623) += pinctrl-mt7623.o obj-$(CONFIG_PINCTRL_MT7629) += pinctrl-mt7629.o +obj-$(CONFIG_PINCTRL_MT7986) += pinctrl-mt7986.o obj-$(CONFIG_PINCTRL_MT8167) += pinctrl-mt8167.o obj-$(CONFIG_PINCTRL_MT8173) += pinctrl-mt8173.o obj-$(CONFIG_PINCTRL_MT8183) += pinctrl-mt8183.o
drivers/pinctrl/mediatek/pinctrl-mt7986.c new file +1640 -0
1,640 lines not on this page · raw message
Signature · 1 line
2.29.2
kernel test robot
build botRe: [PATCH 06/12] pinctrl: mediatek: add support for MT7986 SoC
Thank you for the patch! Yet something to improve:
Hi Sam,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on robh/for-next]
[also build test ERROR on pinctrl/devel clk/clk-next v5.14-rc3 next-20210723] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]
url: https://github.com/0day-ci/linux/commits/Sam-Shih/Add-basic-SoC-support-for-mediatek-mt7986/20210726-152017 base: https://git.kernel.org/pub/scm/linux/kernel/git/robh/linux.git for-next config: arc-allyesconfig (attached as .config) compiler: arceb-elf-gcc (GCC) 10.3.0
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# https://github.com/0day-ci/linux/commit/a6906cf97b80e3911cf37680351c9aa40354b76f
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Sam-Shih/Add-basic-SoC-support-for-mediatek-mt7986/20210726-152017 git checkout a6906cf97b80e3911cf37680351c9aa40354b76f
# save the attached .config to linux build tree
mkdir build_dir
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-10.3.0 make.cross O=build_dir ARCH=arc SHELL=/bin/bash
If you fix the issue, kindly add following tag as appropriate
All errors (new ones prefixed by >>):
>> drivers/pinctrl/mediatek/pinctrl-mt7986.c:62: error: Cannot parse enum!
vim +62 drivers/pinctrl/mediatek/pinctrl-mt7986.c
12
13 #define PIN_FIELD_BASE(_s_pin, _e_pin, _i_base, _s_addr, _x_addrs, _s_bit, \
14 _x_bits) \
15 PIN_FIELD_CALC(_s_pin, _e_pin, _i_base, _s_addr, _x_addrs, _s_bit, \
16 _x_bits, 32, 0)
17
18 /**
19 * enum - Locking variants of the iocfg bases
20 *
21 * MT7986 have multiple bases to program pin configuration listed as the below:
22 * iocfg_rt:0x11c30000, iocfg_rb:0x11c40000, iocfg_lt:0x11e20000,
23 * iocfg_lb:0x11e30000, iocfg_tr:0x11f00000, iocfg_tl:0x11f10000,
24 * _i_based could be used to indicate what base the pin should be mapped into.
25 *
26 * Each iocfg register base control different group of pads on the SoC
27 *
28 *
29 * chip carrier
30 *
31 * A B C D E F G H
32 * +------------------------+
33 * 8 | o o o o o o o o |
34 * 7 | o o o o o o o o |
35 * 6 | o o o o o o o o |
36 * 5 | o o o o o o o o |
37 * 4 | o o o o o o o o |
38 * 3 | o o o o o o o o |
39 * 2 | o o o o o o o o |
40 * 1 | o o o o o o o o |
41 * +------------------------+
42 *
43 * inside Chip carrier
44 *
45 * A B C D E F G H
46 * +------------------------+
47 * 8 | |
48 * 7 | TL TR |
49 * 6 | +---------+ |
50 * 5 | LT | | RT |
51 * 4 | | | |
52 * 3 | LB | | RB |
53 * 2 | +---------+ |
54 * 1 | |
55 * +------------------------+
56 *
57 */
58
59 static const char *const mt7986_pinctrl_register_base_names[] = {
60 "gpio_base", "iocfg_rt_base", "iocfg_rb_base", "iocfg_lt_base",
61 "iocfg_lb_base", "iocfg_tr_base", "iocfg_tl_base",
> 62 };
63 ---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org
PATCH 7/12 dt-bindings: arm64: dts: mediatek: Add mt7986 series
In reply to Sam ShihMT7986 is Mediatek's new 4-core SoC, which is mainly for wifi-router application. The difference between mt7986a and mt7986b is that some pins do not exist on mt7986b.
--- Documentation/devicetree/bindings/arm/mediatek.yaml | 8 ++++++++ 1 file changed, 8 insertions(+)
Documentation/devicetree/bindings/arm/mediatek.yaml +8 -0
@@ -76,6 +76,14 @@ properties: - enum: - mediatek,mt7629-rfb - const: mediatek,mt7629 + - items: + - enum: + - mediatek,mt7986a-rfb + - const: mediatek,mt7986a + - items: + - enum: + - mediatek,mt7986b-rfb + - const: mediatek,mt7986b - items: - enum: - mediatek,mt8127-moose
Signature · 1 line
2.29.2
Re: [PATCH 07/12] dt-bindings: arm64: dts: mediatek: Add mt7986 series
Quoted 9 lines · Sam Shih
MT7986 is Mediatek's new 4-core SoC, which is mainly for wifi-router application. The difference between mt7986a and mt7986b is that some pins do not exist on mt7986b.
Signed-off-by Sam Shih <[email protected]>--- Documentation/devicetree/bindings/arm/mediatek.yaml | 8 ++++++++ 1 file changed, 8 insertions(+)
PATCH 8/12 dt-bindings: rng: mediatek: add mt7986 to mtk rng binding
In reply to Sam ShihAdd RNG binding for MT7986 SoC.
--- Documentation/devicetree/bindings/rng/mtk-rng.yaml | 1 + 1 file changed, 1 insertion(+)
Documentation/devicetree/bindings/rng/mtk-rng.yaml +1 -0
@@ -21,6 +21,7 @@ properties: - enum: - mediatek,mt7622-rng - mediatek,mt7629-rng + - mediatek,mt7986-rng - mediatek,mt8365-rng - mediatek,mt8516-rng - const: mediatek,mt7623-rng
Signature · 1 line
2.29.2
Re: [PATCH 08/12] dt-bindings: rng: mediatek: add mt7986 to mtk rng binding
Quoted 7 lines · Sam Shih
Add RNG binding for MT7986 SoC.
Signed-off-by Sam Shih <[email protected]>--- Documentation/devicetree/bindings/rng/mtk-rng.yaml | 1 + 1 file changed, 1 insertion(+)
Applied, thanks!
PATCH 9/12 dt-bindings: serial: Add compatible for Mediatek MT7986
In reply to Sam ShihThis commit adds dt-binding documentation of uart for Mediatek MT7986 SoC Platform.
--- Documentation/devicetree/bindings/serial/mtk-uart.txt | 1 + 1 file changed, 1 insertion(+)
Documentation/devicetree/bindings/serial/mtk-uart.txt +1 -0
@@ -15,6 +15,7 @@ Required properties: * "mediatek,mt7622-uart" for MT7622 compatible UARTS * "mediatek,mt7623-uart" for MT7623 compatible UARTS * "mediatek,mt7629-uart" for MT7629 compatible UARTS + * "mediatek,mt7986-uart", "mediatek,mt6577-uart" for MT7986 compatible UARTS * "mediatek,mt8127-uart" for MT8127 compatible UARTS * "mediatek,mt8135-uart" for MT8135 compatible UARTS * "mediatek,mt8173-uart" for MT8173 compatible UARTS
Signature · 1 line
2.29.2
PATCH 10/12 dt-bindings: watchdog: Add compatible for Mediatek MT7986
In reply to Sam ShihThis commit adds dt-binding documentation of watchdog for Mediatek MT7986 SoC Platform.
--- Documentation/devicetree/bindings/watchdog/mtk-wdt.txt | 1 + 1 file changed, 1 insertion(+)
Documentation/devicetree/bindings/watchdog/mtk-wdt.txt +1 -0
@@ -13,6 +13,7 @@ Required properties: "mediatek,mt7622-wdt", "mediatek,mt6589-wdt": for MT7622 "mediatek,mt7623-wdt", "mediatek,mt6589-wdt": for MT7623 "mediatek,mt7629-wdt", "mediatek,mt6589-wdt": for MT7629 + "mediatek,mt7986-wdt", "mediatek,mt6589-wdt": for MT7986 "mediatek,mt8183-wdt": for MT8183 "mediatek,mt8516-wdt", "mediatek,mt6589-wdt": for MT8516 "mediatek,mt8192-wdt": for MT8192
Signature · 1 line
2.29.2
Re: [PATCH 10/12] dt-bindings: watchdog: Add compatible for Mediatek MT7986
Quoted 8 lines · Sam Shih
This commit adds dt-binding documentation of watchdog for Mediatek MT7986 SoC Platform.
Signed-off-by Sam Shih <[email protected]>--- Documentation/devicetree/bindings/watchdog/mtk-wdt.txt | 1 + 1 file changed, 1 insertion(+)
Re: [PATCH 10/12] dt-bindings: watchdog: Add compatible for Mediatek MT7986
In reply to Sam ShihQuoted 5 lines · Sam Shih
This commit adds dt-binding documentation of watchdog for Mediatek MT7986 SoC Platform.
Signed-off-by Sam Shih <[email protected]>Acked-by Rob Herring <[email protected]>
Quoted 16 lines
--- Documentation/devicetree/bindings/watchdog/mtk-wdt.txt | 1 + 1 file changed, 1 insertion(+)Documentation/devicetree/bindings/watchdog/mtk-wdt.txt +1
-0@@ -13,6 +13,7 @@ Required properties: "mediatek,mt7622-wdt", "mediatek,mt6589-wdt": for MT7622 "mediatek,mt7623-wdt", "mediatek,mt6589-wdt": for MT7623 "mediatek,mt7629-wdt", "mediatek,mt6589-wdt": for MT7629 + "mediatek,mt7986-wdt", "mediatek,mt6589-wdt": for MT7986 "mediatek,mt8183-wdt": for MT8183 "mediatek,mt8516-wdt", "mediatek,mt6589-wdt": for MT8516 "mediatek,mt8192-wdt": for MT8192
PATCH 11/12 arm64: dts: mediatek: add mt7986a support
In reply to Sam ShihAdd basic chip support for Mediatek mt7986a, include uart nodes with correct clocks, rng node with correct clock, watchdog node and mt7986a pinctrl node.
Add cpu node, timer node, gic node, psci and reserved-memory node for ARM Trusted Firmware,
Add clock controller nodes, include 40M clock source, topckgen, infracfg, infracfg_ao (always on), apmixedsys and ethernet subsystem.
--- arch/arm64/boot/dts/mediatek/Makefile | 1 + arch/arm64/boot/dts/mediatek/mt7986a-rfb.dts | 49 ++++ arch/arm64/boot/dts/mediatek/mt7986a.dtsi | 235 +++++++++++++++++++ 3 files changed, 285 insertions(+) create mode 100644 arch/arm64/boot/dts/mediatek/mt7986a-rfb.dts create mode 100644 arch/arm64/boot/dts/mediatek/mt7986a.dtsi
arch/arm64/boot/dts/mediatek/Makefile +1 -0
@@ -7,6 +7,7 @@ dtb-$(CONFIG_ARCH_MEDIATEK) += mt6797-evb.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt6797-x20-dev.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt7622-rfb1.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt7622-bananapi-bpi-r64.dtb +dtb-$(CONFIG_ARCH_MEDIATEK) += mt7986a-rfb.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt8167-pumpkin.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt8173-elm.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt8173-elm-hana.dtb
arch/arm64/boot/dts/mediatek/mt7986a-rfb.dts new file +49 -0
49 lines not on this page · raw message
arch/arm64/boot/dts/mediatek/mt7986a.dtsi new file +235 -0
235 lines not on this page · raw message
Signature · 1 line
2.29.2
PATCH 12/12 arm64: dts: mediatek: add mt7986b support
In reply to Sam ShihAdd basic chip support for Mediatek mt7986b, include uart nodes with correct clocks, rng node with correct clock, watchdog node and mt7986b pinctrl node.
Add cpu node, timer node, gic node, psci and reserved-memory node for ARM Trusted Firmware,
Add clock controller nodes, include 40M clock source, topckgen, infracfg, infracfg_ao (always on), apmixedsys and ethernet subsystem.
--- arch/arm64/boot/dts/mediatek/Makefile | 1 + arch/arm64/boot/dts/mediatek/mt7986b-rfb.dts | 21 ++ arch/arm64/boot/dts/mediatek/mt7986b.dtsi | 235 +++++++++++++++++++ 3 files changed, 257 insertions(+) create mode 100644 arch/arm64/boot/dts/mediatek/mt7986b-rfb.dts create mode 100644 arch/arm64/boot/dts/mediatek/mt7986b.dtsi
arch/arm64/boot/dts/mediatek/Makefile +1 -0
@@ -8,6 +8,7 @@ dtb-$(CONFIG_ARCH_MEDIATEK) += mt6797-x20-dev.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt7622-rfb1.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt7622-bananapi-bpi-r64.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt7986a-rfb.dtb +dtb-$(CONFIG_ARCH_MEDIATEK) += mt7986b-rfb.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt8167-pumpkin.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt8173-elm.dtb dtb-$(CONFIG_ARCH_MEDIATEK) += mt8173-elm-hana.dtb
arch/arm64/boot/dts/mediatek/mt7986b-rfb.dts new file +21 -0
@@ -0,0 +1,21 @@ +// SPDX-License-Identifier: (GPL-2.0 OR MIT) +/* + * Copyright (C) 2021 MediaTek Inc. + * Author: Sam.Shih <[email protected]> + */ + +/dts-v1/; +#include "mt7986b.dtsi" + +/ { + model = "MediaTek MT7986b RFB"; + compatible = "mediatek,mt7986b-rfb"; + chosen { + bootargs = "console=ttyS0,115200n1 loglevel=8 \ + earlycon=uart8250,mmio32,0x11002000"; + }; +}; + +&uart0 { + status = "okay"; +};
arch/arm64/boot/dts/mediatek/mt7986b.dtsi new file +235 -0
235 lines not on this page · raw message
Signature · 1 line
2.29.2