As the sg2042 uses different address for timer and mswi of its clint
device, it should follow the aclint format. For the previous patchs,
it only use only one address for both mtime and mtimer, this is can
not be parsed by OpenSBI. To resolve this, separate these two registers
in the dtb.
Link: https://lists.infradead.org/pipermail/opensbi/2023-October/005693.html
Link: https://github.com/riscv/riscv-aclint/blob/main/riscv-aclint.adoc
This patch can be tested with upstream SBI with the following patch:
1. https://lists.infradead.org/pipermail/opensbi/2023-November/005914.html
Changed from v2:
1. Use reg-names to map the registers.
Changed from v1:
1. change the commit to address the reason for ABI change.
2. remove unnecessary link in the commit.
Inochi Amaoto (2):
dt-bindings: timer: thead,c900-aclint-mtimer: separate mtime and
mtimecmp regs
riscv: dts: sophgo: separate sg2042 mtime and mtimecmp to fit aclint
format
.../timer/thead,c900-aclint-mtimer.yaml | 9 ++-
arch/riscv/boot/dts/sophgo/sg2042.dtsi | 80 +++++++++++--------
2 files changed, 56 insertions(+), 33 deletions(-)
--
2.42.1
Change the timer layout in the dtb to fit the format that needed by
the SBI.
Signed-off-by: Inochi Amaoto <[email protected]>
Fixes: 967a94a92aaa ("riscv: dts: add initial Sophgo SG2042 SoC device tree")
---
arch/riscv/boot/dts/sophgo/sg2042.dtsi | 80 +++++++++++++++-----------
1 file changed, 48 insertions(+), 32 deletions(-)
diff --git a/arch/riscv/boot/dts/sophgo/sg2042.dtsi b/arch/riscv/boot/dts/sophgo/sg2042.dtsi
index 93256540d078..ead1cc35d88b 100644
--- a/arch/riscv/boot/dts/sophgo/sg2042.dtsi
+++ b/arch/riscv/boot/dts/sophgo/sg2042.dtsi
@@ -93,144 +93,160 @@ clint_mswi: interrupt-controller@7094000000 {
<&cpu63_intc 3>;
};
- clint_mtimer0: timer@70ac000000 {
+ clint_mtimer0: timer@70ac004000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac000000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac004000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu0_intc 7>,
<&cpu1_intc 7>,
<&cpu2_intc 7>,
<&cpu3_intc 7>;
};
- clint_mtimer1: timer@70ac010000 {
+ clint_mtimer1: timer@70ac014000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac010000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac014000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu4_intc 7>,
<&cpu5_intc 7>,
<&cpu6_intc 7>,
<&cpu7_intc 7>;
};
- clint_mtimer2: timer@70ac020000 {
+ clint_mtimer2: timer@70ac024000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac020000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac024000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu8_intc 7>,
<&cpu9_intc 7>,
<&cpu10_intc 7>,
<&cpu11_intc 7>;
};
- clint_mtimer3: timer@70ac030000 {
+ clint_mtimer3: timer@70ac034000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac030000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac034000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu12_intc 7>,
<&cpu13_intc 7>,
<&cpu14_intc 7>,
<&cpu15_intc 7>;
};
- clint_mtimer4: timer@70ac040000 {
+ clint_mtimer4: timer@70ac044000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac040000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac044000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu16_intc 7>,
<&cpu17_intc 7>,
<&cpu18_intc 7>,
<&cpu19_intc 7>;
};
- clint_mtimer5: timer@70ac050000 {
+ clint_mtimer5: timer@70ac054000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac050000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac054000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu20_intc 7>,
<&cpu21_intc 7>,
<&cpu22_intc 7>,
<&cpu23_intc 7>;
};
- clint_mtimer6: timer@70ac060000 {
+ clint_mtimer6: timer@70ac064000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac060000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac064000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu24_intc 7>,
<&cpu25_intc 7>,
<&cpu26_intc 7>,
<&cpu27_intc 7>;
};
- clint_mtimer7: timer@70ac070000 {
+ clint_mtimer7: timer@70ac074000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac070000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac074000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu28_intc 7>,
<&cpu29_intc 7>,
<&cpu30_intc 7>,
<&cpu31_intc 7>;
};
- clint_mtimer8: timer@70ac080000 {
+ clint_mtimer8: timer@70ac084000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac080000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac084000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu32_intc 7>,
<&cpu33_intc 7>,
<&cpu34_intc 7>,
<&cpu35_intc 7>;
};
- clint_mtimer9: timer@70ac090000 {
+ clint_mtimer9: timer@70ac094000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac090000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac094000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu36_intc 7>,
<&cpu37_intc 7>,
<&cpu38_intc 7>,
<&cpu39_intc 7>;
};
- clint_mtimer10: timer@70ac0a0000 {
+ clint_mtimer10: timer@70ac0a4000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac0a0000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac0a4000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu40_intc 7>,
<&cpu41_intc 7>,
<&cpu42_intc 7>,
<&cpu43_intc 7>;
};
- clint_mtimer11: timer@70ac0b0000 {
+ clint_mtimer11: timer@70ac0b4000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac0b0000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac0b4000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu44_intc 7>,
<&cpu45_intc 7>,
<&cpu46_intc 7>,
<&cpu47_intc 7>;
};
- clint_mtimer12: timer@70ac0c0000 {
+ clint_mtimer12: timer@70ac0c4000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac0c0000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac0c4000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu48_intc 7>,
<&cpu49_intc 7>,
<&cpu50_intc 7>,
<&cpu51_intc 7>;
};
- clint_mtimer13: timer@70ac0d0000 {
+ clint_mtimer13: timer@70ac0d4000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac0d0000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac0d4000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu52_intc 7>,
<&cpu53_intc 7>,
<&cpu54_intc 7>,
<&cpu55_intc 7>;
};
- clint_mtimer14: timer@70ac0e0000 {
+ clint_mtimer14: timer@70ac0e4000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac0e0000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac0e4000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu56_intc 7>,
<&cpu57_intc 7>,
<&cpu58_intc 7>,
<&cpu59_intc 7>;
};
- clint_mtimer15: timer@70ac0f0000 {
+ clint_mtimer15: timer@70ac0f4000 {
compatible = "sophgo,sg2042-aclint-mtimer", "thead,c900-aclint-mtimer";
- reg = <0x00000070 0xac0f0000 0x00000000 0x00007ff8>;
+ reg = <0x00000070 0xac0f4000 0x00000000 0x0000c000>;
+ reg-names = "mtimecmp";
interrupts-extended = <&cpu60_intc 7>,
<&cpu61_intc 7>,
<&cpu62_intc 7>,
--
2.42.1
The timer registers of aclint don't follow the clint layout and can
be mapped on any different offset. As sg2042 uses separated timer
and mswi for its clint, it should follow the aclint spec and have
separated registers.
The previous patch introduced a new type of T-HEAD aclint timer which
has clint timer layout. Although it has the clint timer layout, it
should follow the aclint spec and uses the separated mtime and mtimecmp
regs. So a ABI change is needed to make the timer fit the aclint spec.
To make T-HEAD aclint timer more closer to the aclint spec, use
regs-names to represent the mtimecmp register, which can avoid hack
for unsupport mtime register of T-HEAD aclint timer.
Signed-off-by: Inochi Amaoto <[email protected]>
Fixes: 4734449f7311 ("dt-bindings: timer: Add Sophgo sg2042 CLINT timer")
Link: https://lists.infradead.org/pipermail/opensbi/2023-October/005693.html
Link: https://github.com/riscv/riscv-aclint/blob/main/riscv-aclint.adoc
---
.../bindings/timer/thead,c900-aclint-mtimer.yaml | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/Documentation/devicetree/bindings/timer/thead,c900-aclint-mtimer.yaml b/Documentation/devicetree/bindings/timer/thead,c900-aclint-mtimer.yaml
index fbd235650e52..2e92bcdeb423 100644
--- a/Documentation/devicetree/bindings/timer/thead,c900-aclint-mtimer.yaml
+++ b/Documentation/devicetree/bindings/timer/thead,c900-aclint-mtimer.yaml
@@ -17,7 +17,12 @@ properties:
- const: thead,c900-aclint-mtimer
reg:
- maxItems: 1
+ items:
+ - description: MTIMECMP Registers
+
+ reg-names:
+ items:
+ - const: mtimecmp
interrupts-extended:
minItems: 1
@@ -28,6 +33,7 @@ additionalProperties: false
required:
- compatible
- reg
+ - reg-names
- interrupts-extended
examples:
@@ -39,5 +45,6 @@ examples:
<&cpu3intc 7>,
<&cpu4intc 7>;
reg = <0xac000000 0x00010000>;
+ reg-names = "mtimecmp";
};
...
--
2.42.1
On 17/11/2023 06:07, Inochi Amaoto wrote:
> The timer registers of aclint don't follow the clint layout and can
> be mapped on any different offset. As sg2042 uses separated timer
> and mswi for its clint, it should follow the aclint spec and have
> separated registers.
>
> The previous patch introduced a new type of T-HEAD aclint timer which
> has clint timer layout. Although it has the clint timer layout, it
> should follow the aclint spec and uses the separated mtime and mtimecmp
> regs. So a ABI change is needed to make the timer fit the aclint spec.
>
> To make T-HEAD aclint timer more closer to the aclint spec, use
> regs-names to represent the mtimecmp register, which can avoid hack
> for unsupport mtime register of T-HEAD aclint timer.
>
I don't understand this reasoning. You had one entry, you still have one
entry. Adding reg-names (not regs-names) does not change it.
Best regards,
Krzysztof
>
>On 17/11/2023 06:07, Inochi Amaoto wrote:
>> The timer registers of aclint don't follow the clint layout and can
>> be mapped on any different offset. As sg2042 uses separated timer
>> and mswi for its clint, it should follow the aclint spec and have
>> separated registers.
>>
>> The previous patch introduced a new type of T-HEAD aclint timer which
>> has clint timer layout. Although it has the clint timer layout, it
>> should follow the aclint spec and uses the separated mtime and mtimecmp
>> regs. So a ABI change is needed to make the timer fit the aclint spec.
>>
>> To make T-HEAD aclint timer more closer to the aclint spec, use
>> regs-names to represent the mtimecmp register, which can avoid hack
>> for unsupport mtime register of T-HEAD aclint timer.
>>
>
>I don't understand this reasoning. You had one entry, you still have one
>entry. Adding reg-names (not regs-names) does not change it.
>
If no "reg-names", all the register of ACLINT should be defined. However,
T-HEAD aclint timer of sg2042 only supports mtimecmp register. If no extra
prompt is provided for the SBI, it will fail to recognize aclint timer
registers when parsing the aclint node with one reg entry.
There is another way to avoid this by using an empty entry to identify
unsupported mtime, but Conor have already rejected this. See [1].
Link: https://lore.kernel.org/all/20231114-skedaddle-precinct-66c8897227bb@squawk/ [1]
On Fri, Nov 17, 2023 at 08:30:21PM +0800, Inochi Amaoto wrote:
> >
> >On 17/11/2023 06:07, Inochi Amaoto wrote:
> >> The timer registers of aclint don't follow the clint layout and can
> >> be mapped on any different offset. As sg2042 uses separated timer
> >> and mswi for its clint, it should follow the aclint spec and have
> >> separated registers.
> >>
> >> The previous patch introduced a new type of T-HEAD aclint timer which
> >> has clint timer layout. Although it has the clint timer layout, it
> >> should follow the aclint spec and uses the separated mtime and mtimecmp
> >> regs. So a ABI change is needed to make the timer fit the aclint spec.
> >>
> >> To make T-HEAD aclint timer more closer to the aclint spec, use
> >> regs-names to represent the mtimecmp register, which can avoid hack
> >> for unsupport mtime register of T-HEAD aclint timer.
> >>
> >
> >I don't understand this reasoning. You had one entry, you still have one
> >entry. Adding reg-names (not regs-names) does not change it.
> >
>
> If no "reg-names", all the register of ACLINT should be defined. However,
> T-HEAD aclint timer of sg2042 only supports mtimecmp register. If no extra
> prompt is provided for the SBI, it will fail to recognize aclint timer
> registers when parsing the aclint node with one reg entry.
>
> There is another way to avoid this by using an empty entry to identify
> unsupported mtime, but Conor have already rejected this. See [1].
>
> Link: https://lore.kernel.org/all/20231114-skedaddle-precinct-66c8897227bb@squawk/ [1]
Perhaps you misunderstood my suggestion. I was looking for _both_
registers to be defined in the binding as well as adding reg-names as a
required property. Doing what you have here might work for your use
case, but does not make sense from a bindings point of view as there is
no way to describe the mtime register, should it exist in another SoC.
Cheers,
Conor
>
>On Fri, Nov 17, 2023 at 08:30:21PM +0800, Inochi Amaoto wrote:
>>>
>>> On 17/11/2023 06:07, Inochi Amaoto wrote:
>>>> The timer registers of aclint don't follow the clint layout and can
>>>> be mapped on any different offset. As sg2042 uses separated timer
>>>> and mswi for its clint, it should follow the aclint spec and have
>>>> separated registers.
>>>>
>>>> The previous patch introduced a new type of T-HEAD aclint timer which
>>>> has clint timer layout. Although it has the clint timer layout, it
>>>> should follow the aclint spec and uses the separated mtime and mtimecmp
>>>> regs. So a ABI change is needed to make the timer fit the aclint spec.
>>>>
>>>> To make T-HEAD aclint timer more closer to the aclint spec, use
>>>> regs-names to represent the mtimecmp register, which can avoid hack
>>>> for unsupport mtime register of T-HEAD aclint timer.
>>>>
>>>
>>> I don't understand this reasoning. You had one entry, you still have one
>>> entry. Adding reg-names (not regs-names) does not change it.
>>>
>>
>> If no "reg-names", all the register of ACLINT should be defined. However,
>> T-HEAD aclint timer of sg2042 only supports mtimecmp register. If no extra
>> prompt is provided for the SBI, it will fail to recognize aclint timer
>> registers when parsing the aclint node with one reg entry.
>>
>> There is another way to avoid this by using an empty entry to identify
>> unsupported mtime, but Conor have already rejected this. See [1].
>>
>> Link: https://lore.kernel.org/all/20231114-skedaddle-precinct-66c8897227bb@squawk/ [1]
>
>Perhaps you misunderstood my suggestion. I was looking for _both_
>registers to be defined in the binding as well as adding reg-names as a
>required property. Doing what you have here might work for your use
>case, but does not make sense from a bindings point of view as there is
>no way to describe the mtime register, should it exist in another SoC.
>
Thanks for your clarification. If I understand you correctly, the binding
should have all registers that ACLINT has. But for specific use case,
it should only contain supported registers and omit unsupported. Please
correct me if I misunderstood. Thanks.
>Cheers,
>Conor
>
> Thanks for your clarification. If I understand you correctly, the binding
> should have all registers that ACLINT has. But for specific use case,
> it should only contain supported registers and omit unsupported. Please
> correct me if I misunderstood. Thanks.
Yes.
>> Thanks for your clarification. If I understand you correctly, the binding
>> should have all registers that ACLINT has. But for specific use case,
>> it should only contain supported registers and omit unsupported. Please
>> correct me if I misunderstood. Thanks.
>
>Yes.
>
Thanks, I will prepare a new binding for this.