Message ID | 20220511161602.117772-1-sireeshkodali1@gmail.com |
---|---|
Headers | show |
Series | Add support for remoteprocs on the MSM8953 platform | expand |
On 11/05/2022 19:15, Sireesh Kodali wrote: > Add support for the Audio DSP PIL found on the Qualcomm MSM8953 > platform. The same configuration is used on all SoCs based on the > MSM8953 platform (SDM450, SDA450, SDM625, SDM632, APQ8053). > > Signed-off-by: Sireesh Kodali <sireeshkodali1@gmail.com> > --- > drivers/remoteproc/qcom_q6v5_pas.c | 31 ++++++++++++++++++++++++++++++ > 1 file changed, 31 insertions(+) > > diff --git a/drivers/remoteproc/qcom_q6v5_pas.c b/drivers/remoteproc/qcom_q6v5_pas.c > index 1ae47cc153e5..4dcb714a1468 100644 > --- a/drivers/remoteproc/qcom_q6v5_pas.c > +++ b/drivers/remoteproc/qcom_q6v5_pas.c > @@ -617,7 +617,37 @@ static const struct adsp_data sm8350_adsp_resource = { > .ssctl_id = 0x14, > }; > > +static const struct adsp_data msm8953_adsp_resource = { > + .crash_reason_smem = 423, > + .firmware_name = "adsp.mdt", > + .pas_id = 1, > + .has_aggre2_clk = false, > + .auto_boot = true, > + .proxy_pd_names = (char*[]){ > + "cx", > + NULL, > + }, > + .ssr_name = "lpass", > + .sysmon_name = "adsp", > + .ssctl_id = 0x14, > +}; > + > static const struct adsp_data msm8996_adsp_resource = { > + .crash_reason_smem = 423, > + .firmware_name = "adsp.mdt", > + .pas_id = 1, > + .has_aggre2_clk = false, > + .auto_boot = true, > + .proxy_pd_names = (char*[]){ > + "cx", > + NULL, > + }, > + .ssr_name = "lpass", > + .sysmon_name = "adsp", > + .ssctl_id = 0x14, > +}; > + > +static const struct adsp_data msm8998_adsp_resource = { This was dropped in 9ae45035ba2be4117edb8fd3952c3c5b84a0b820. Please take care when rebasing your patches. > .crash_reason_smem = 423, > .firmware_name = "adsp.mdt", > .pas_id = 1, > @@ -850,6 +880,7 @@ static const struct adsp_data sdx55_mpss_resource = { > static const struct of_device_id adsp_of_match[] = { > { .compatible = "qcom,msm8974-adsp-pil", .data = &adsp_resource_init}, > { .compatible = "qcom,msm8996-adsp-pil", .data = &msm8996_adsp_resource}, > + { .compatible = "qcom,msm8953-adsp-pil", .data = &msm8953_adsp_resource}, > { .compatible = "qcom,msm8996-slpi-pil", .data = &slpi_resource_init}, > { .compatible = "qcom,msm8998-adsp-pas", .data = &msm8996_adsp_resource}, > { .compatible = "qcom,msm8998-slpi-pas", .data = &slpi_resource_init},
On 11/05/2022 18:15, Sireesh Kodali wrote: > Convert the dt-bindings from txt to YAML. This is in preparation for > including the relevant bindings for the MSM8953 platform's wcnss pil. > > Signed-off-by: Sireesh Kodali <sireeshkodali1@gmail.com> Thank you for your patch. There is something to discuss/improve. Please use existing bindings or example-schema as a starting point. Half of my review could be skipped if you just followed what we already have in the tree. Some of these qcom specific properties already exist but you decided to write them differently... please don't, rather reuse the code. (...) > + > +maintainers: > + - Bjorn Andersson <bjorn.andersson@linaro.org> > + > +description: > + This document defines the binding for a component that loads and boots > + firmware on the Qualcomm WCNSS core. > + > +properties: > + compatible: > + oneOf: > + - items: > + - enum: > + - qcom,pronto-v2-pil > + - enum: > + - qcom,pronto This does not look correct. The fallback compatible should not change. What is more, it was not documented in original binding, so this should be done in separate patch. > + - items: No need for items, it's just one item. > + - enum: > + - qcom,riva-pil > + - qcom,pronto-v1-pil > + - qcom,pronto-v2-pil > + > + reg: > + description: must specify the base address and size of the CCU, DXE and PMU > + register blocks New line after "decription:", drop "must specify" and start with capital letter. You need maxItems: 3 > + > + reg-names: > + items: > + - const: ccu > + - const: dxe > + - const: pmu > + > + interrupts-extended: > + description: > + Interrupt lines Skip description, it's obvious. It should be only "interrupts", not extended. > + minItems: 2 > + maxItems: 5 > + > + interrupt-names: > + minItems: 2 > + maxItems: 5 Names should be clearly defined. They were BTW defined in original bindings, so you should not remove them. This makes me wonder what else did you remove from original bindings... Please document all deviations from pure conversion in the commit msg. It's a second "hidden" difference. > + > + firmware-name: > + $ref: /schemas/types.yaml#/definitions/string > + description: Relative firmware image path for the WCNSS core. Defaults to > + "wcnss.mdt". Blank line after "description:". This applies to other places as well. Remove "Defailts to ..." and just add "default" schema. > + > + vddpx-supply: > + description: Reference to the PX regulator to be held on behalf of the > + booting of the WCNSS core > + > + vddmx-supply: > + description: Reference to the MX regulator to be held on behalf of the > + booting of the WCNSS core. > + > + vddcx-supply: > + description: Reference to the CX regulator to be held on behalf of the > + booting of the WCNSS core. s/Reference to the// > + > + power-domains: > + $ref: /schemas/types.yaml#/definitions/phandle-array > + description: References to the power domains that need to be held on > + behalf of the booting WCNSS core 1. Ditto. 2. No need for ref 3. maxItems > + > + power-domain-names: > + $ref: /schemas/types.yaml#/definitions/string-array No need for ref, skip description. > + description: Names of the power domains > + items: > + - const: cx > + - const: mx > + > + qcom,smem-states: > + $ref: /schemas/types.yaml#/definitions/phandle-array > + description: States used by the AP to signal the WCNSS core that it should > + shutdown > + items: > + - description: Stop the modem > + > + qcom,smem-state-names: > + $ref: /schemas/types.yaml#/definitions/string-array No need for ref. Really, it does not appear in any of existing bindings for smem-state-names, so how did you get it? > + description: The names of the state bits used for SMP2P output > + items: > + - const: stop > + > + memory-region: > + maxItems: 1 > + description: Reference to the reserved-memory for the WCNSS core > + > + smd-edge: > + type: object > + description: > + Qualcomm Shared Memory subnode which represents communication edge, > + channels and devices related to the ADSP. You should reference /schemas/soc/qcom/qcom,smd.yaml > + > + iris: Generic node name... what is "iris"? > + type: object > + description: > + The iris subnode of the WCNSS PIL is used to describe the attached rf module s/rf/RF/ > + and its resource dependencies. > + > + properties: > + compatible: > + enum: > + - qcom,wcn3620 > + - qcom,wcn3660 > + - qcom,wcn3660b > + - qcom,wcn3680 > + > + clocks: > + description: XO clock > + > + clock-names: > + items: > + - const: xo > + > + required: > + - compatible clocks and clock-names were required. Missing supplies, which were btw as well required. > + > +required: > + - compatible > + - reg > + - reg-names > + - interrupts-extended > + - interrupt-names > + - vddpx-supply > + - memory-region > + - smd-edge > + - iris > + > +additionalProperties: false > + > +if: Within allOf, please. Best regards, Krzysztof
On Wed May 11, 2022 at 10:45 PM IST, Krzysztof Kozlowski wrote: > On 11/05/2022 18:15, Sireesh Kodali wrote: > > Convert the dt-bindings from txt to YAML. This is in preparation for > > including the relevant bindings for the MSM8953 platform's wcnss pil. > > > > Signed-off-by: Sireesh Kodali <sireeshkodali1@gmail.com> > > Thank you for your patch. There is something to discuss/improve. > > Please use existing bindings or example-schema as a starting point. Half > of my review could be skipped if you just followed what we already have > in the tree. > > Some of these qcom specific properties already exist but you decided to > write them differently... please don't, rather reuse the code. > Thank you for your review, I will make the chnages as appropriate in v2. > (...) > > > + > > +maintainers: > > + - Bjorn Andersson <bjorn.andersson@linaro.org> > > + > > +description: > > + This document defines the binding for a component that loads and boots > > + firmware on the Qualcomm WCNSS core. > > + > > +properties: > > + compatible: > > + oneOf: > > + - items: > > + - enum: > > + - qcom,pronto-v2-pil > > + - enum: > > + - qcom,pronto > > This does not look correct. The fallback compatible should not change. > What is more, it was not documented in original binding, so this should > be done in separate patch. > This was not a change to the fallback compatible. msm8916.dtsi's wcnss node has "qcom,pronto" as the compatible string, which is why this was added. It is however not documented in the txt file. Is it sufficient to add a note in the commit message, or should it be split into a separate commit? > > + - items: > > No need for items, it's just one item. > > > + - enum: > > + - qcom,riva-pil > > + - qcom,pronto-v1-pil > > + - qcom,pronto-v2-pil > > + > > + reg: > > + description: must specify the base address and size of the CCU, DXE and PMU > > + register blocks > > New line after "decription:", drop "must specify" and start with capital > letter. > > You need maxItems: 3 > Will fix in v2 > > > + > > + reg-names: > > + items: > > + - const: ccu > > + - const: dxe > > + - const: pmu > > + > > + interrupts-extended: > > + description: > > + Interrupt lines > > Skip description, it's obvious. > > It should be only "interrupts", not extended. > > > + minItems: 2 > > + maxItems: 5 > > + > > + interrupt-names: > > + minItems: 2 > > + maxItems: 5 > > Names should be clearly defined. They were BTW defined in original > bindings, so you should not remove them. This makes me wonder what else > did you remove from original bindings... > > Please document all deviations from pure conversion in the commit msg. > It's a second "hidden" difference. > Sorry, this was meant to be a pure txt->YAML conversion. The missing interrupt names was accidental, and will be fixed in v2. > > + > > + firmware-name: > > + $ref: /schemas/types.yaml#/definitions/string > > + description: Relative firmware image path for the WCNSS core. Defaults to > > + "wcnss.mdt". > > > Blank line after "description:". This applies to other places as well. > > Remove "Defailts to ..." and just add "default" schema. > Will be fixed in v2 > > + > > + vddpx-supply: > > + description: Reference to the PX regulator to be held on behalf of the > > + booting of the WCNSS core > > + > > + vddmx-supply: > > + description: Reference to the MX regulator to be held on behalf of the > > + booting of the WCNSS core. > > + > > + vddcx-supply: > > + description: Reference to the CX regulator to be held on behalf of the > > + booting of the WCNSS core. > > s/Reference to the// > > > + > > + power-domains: > > + $ref: /schemas/types.yaml#/definitions/phandle-array > > + description: References to the power domains that need to be held on > > + behalf of the booting WCNSS core > > 1. Ditto. > 2. No need for ref > 3. maxItems > > > + > > + power-domain-names: > > + $ref: /schemas/types.yaml#/definitions/string-array > > No need for ref, skip description. > > > + description: Names of the power domains > > + items: > > + - const: cx > > + - const: mx > > + > > + qcom,smem-states: > > + $ref: /schemas/types.yaml#/definitions/phandle-array > > + description: States used by the AP to signal the WCNSS core that it should > > + shutdown > > + items: > > + - description: Stop the modem > > + > > + qcom,smem-state-names: > > + $ref: /schemas/types.yaml#/definitions/string-array > > No need for ref. Really, it does not appear in any of existing bindings > for smem-state-names, so how did you get it? > The smem nodes were copied from /remoteproc/qcom,sdm845-adsp-pil.yaml > > + description: The names of the state bits used for SMP2P output > > + items: > > + - const: stop > > + > > + memory-region: > > + maxItems: 1 > > + description: Reference to the reserved-memory for the WCNSS core > > + > > + smd-edge: > > + type: object > > + description: > > + Qualcomm Shared Memory subnode which represents communication edge, > > + channels and devices related to the ADSP. > > You should reference /schemas/soc/qcom/qcom,smd.yaml Will be done in v2 > > > + > > + iris: > > Generic node name... what is "iris"? > Iris is the RF module, I'll make the description better > > + type: object > > + description: > > + The iris subnode of the WCNSS PIL is used to describe the attached rf module > > s/rf/RF/ > > > + and its resource dependencies. > > + > > + properties: > > + compatible: > > + enum: > > + - qcom,wcn3620 > > + - qcom,wcn3660 > > + - qcom,wcn3660b > > + - qcom,wcn3680 > > + > > + clocks: > > + description: XO clock > > + > > + clock-names: > > + items: > > + - const: xo > > + > > + required: > > + - compatible > > clocks and clock-names were required. > Missing supplies, which were btw as well required. > This was unintentional, it will be fixed in v2. > > + > > +required: > > + - compatible > > + - reg > > + - reg-names > > + - interrupts-extended > > + - interrupt-names > > + - vddpx-supply > > + - memory-region > > + - smd-edge > > + - iris > > + > > +additionalProperties: false > > + > > +if: > > Within allOf, please. > Will be fixed in v2 > > > Best regards, > Krzysztof Thanks again for the review, Sireesh
On 12/05/2022 08:50, Sireesh Kodali wrote: > On Wed May 11, 2022 at 10:45 PM IST, Krzysztof Kozlowski wrote: >> On 11/05/2022 18:15, Sireesh Kodali wrote: >>> Convert the dt-bindings from txt to YAML. This is in preparation for >>> including the relevant bindings for the MSM8953 platform's wcnss pil. >>> >>> Signed-off-by: Sireesh Kodali <sireeshkodali1@gmail.com> >> >> Thank you for your patch. There is something to discuss/improve. >> >> Please use existing bindings or example-schema as a starting point. Half >> of my review could be skipped if you just followed what we already have >> in the tree. >> >> Some of these qcom specific properties already exist but you decided to >> write them differently... please don't, rather reuse the code. >> > > Thank you for your review, I will make the chnages as appropriate in v2. >> (...) >> >>> + >>> +maintainers: >>> + - Bjorn Andersson <bjorn.andersson@linaro.org> >>> + >>> +description: >>> + This document defines the binding for a component that loads and boots >>> + firmware on the Qualcomm WCNSS core. >>> + >>> +properties: >>> + compatible: >>> + oneOf: >>> + - items: >>> + - enum: >>> + - qcom,pronto-v2-pil >>> + - enum: >>> + - qcom,pronto >> >> This does not look correct. The fallback compatible should not change. >> What is more, it was not documented in original binding, so this should >> be done in separate patch. >> > > This was not a change to the fallback compatible. You made it an enum, so you expect it to use different fallback for different cases. > msm8916.dtsi's wcnss > node has "qcom,pronto" as the compatible string, which is why this was > added. It is however not documented in the txt file. Is it sufficient to > add a note in the commit message, or should it be split into a separate > commit? Please split it, assuming that fallback is correct. Maybe the fallback is wrong? > >>> + - items: >> >> No need for items, it's just one item. >> >>> + - enum: >>> + - qcom,riva-pil >>> + - qcom,pronto-v1-pil >>> + - qcom,pronto-v2-pil >>> + >>> + reg: >>> + description: must specify the base address and size of the CCU, DXE and PMU >>> + register blocks >> >> New line after "decription:", drop "must specify" and start with capital >> letter. >> >> You need maxItems: 3 >> > > Will fix in v2 >> >>> + >>> + reg-names: >>> + items: >>> + - const: ccu >>> + - const: dxe >>> + - const: pmu >>> + >>> + interrupts-extended: >>> + description: >>> + Interrupt lines >> >> Skip description, it's obvious. >> >> It should be only "interrupts", not extended. >> >>> + minItems: 2 >>> + maxItems: 5 >>> + >>> + interrupt-names: >>> + minItems: 2 >>> + maxItems: 5 >> >> Names should be clearly defined. They were BTW defined in original >> bindings, so you should not remove them. This makes me wonder what else >> did you remove from original bindings... >> >> Please document all deviations from pure conversion in the commit msg. >> It's a second "hidden" difference. >> > > Sorry, this was meant to be a pure txt->YAML conversion. The missing > interrupt names was accidental, and will be fixed in v2. >>> + >>> + firmware-name: >>> + $ref: /schemas/types.yaml#/definitions/string >>> + description: Relative firmware image path for the WCNSS core. Defaults to >>> + "wcnss.mdt". >> >> >> Blank line after "description:". This applies to other places as well. >> >> Remove "Defailts to ..." and just add "default" schema. >> > > Will be fixed in v2 >>> + >>> + vddpx-supply: >>> + description: Reference to the PX regulator to be held on behalf of the >>> + booting of the WCNSS core >>> + >>> + vddmx-supply: >>> + description: Reference to the MX regulator to be held on behalf of the >>> + booting of the WCNSS core. >>> + >>> + vddcx-supply: >>> + description: Reference to the CX regulator to be held on behalf of the >>> + booting of the WCNSS core. >> >> s/Reference to the// >> >>> + >>> + power-domains: >>> + $ref: /schemas/types.yaml#/definitions/phandle-array >>> + description: References to the power domains that need to be held on >>> + behalf of the booting WCNSS core >> >> 1. Ditto. >> 2. No need for ref >> 3. maxItems >> >>> + >>> + power-domain-names: >>> + $ref: /schemas/types.yaml#/definitions/string-array >> >> No need for ref, skip description. >> >>> + description: Names of the power domains >>> + items: >>> + - const: cx >>> + - const: mx >>> + >>> + qcom,smem-states: >>> + $ref: /schemas/types.yaml#/definitions/phandle-array >>> + description: States used by the AP to signal the WCNSS core that it should >>> + shutdown >>> + items: >>> + - description: Stop the modem >>> + >>> + qcom,smem-state-names: >>> + $ref: /schemas/types.yaml#/definitions/string-array >> >> No need for ref. Really, it does not appear in any of existing bindings >> for smem-state-names, so how did you get it? >> > > The smem nodes were copied from /remoteproc/qcom,sdm845-adsp-pil.yaml Hm, indeed, you're right. There are few files having here ref. I'll fix these. > >>> + description: The names of the state bits used for SMP2P output >>> + items: >>> + - const: stop >>> + >>> + memory-region: >>> + maxItems: 1 >>> + description: Reference to the reserved-memory for the WCNSS core >>> + >>> + smd-edge: >>> + type: object >>> + description: >>> + Qualcomm Shared Memory subnode which represents communication edge, >>> + channels and devices related to the ADSP. >> >> You should reference /schemas/soc/qcom/qcom,smd.yaml > > Will be done in v2 >> >>> + >>> + iris: >> >> Generic node name... what is "iris"? >> > Iris is the RF module, I'll make the description better RF like wifi? Then the property name should be "wifi". Best regards, Krzysztof
On 12/05/2022 08:50, Sireesh Kodali wrote: >>> + description: The names of the state bits used for SMP2P output >>> + items: >>> + - const: stop >>> + >>> + memory-region: >>> + maxItems: 1 >>> + description: Reference to the reserved-memory for the WCNSS core >>> + >>> + smd-edge: >>> + type: object >>> + description: >>> + Qualcomm Shared Memory subnode which represents communication edge, >>> + channels and devices related to the ADSP. >> >> You should reference /schemas/soc/qcom/qcom,smd.yaml It seems it is not a SMD driver so above reference is not correct. This should be probably described in its own schema, I just need to understand what's this... Best regards, Krzysztof
On Thu May 12, 2022 at 1:44 PM IST, Krzysztof Kozlowski wrote: > On 12/05/2022 08:50, Sireesh Kodali wrote: > > On Wed May 11, 2022 at 10:45 PM IST, Krzysztof Kozlowski wrote: > >> On 11/05/2022 18:15, Sireesh Kodali wrote: > >>> Convert the dt-bindings from txt to YAML. This is in preparation for > >>> including the relevant bindings for the MSM8953 platform's wcnss pil. > >>> > >>> Signed-off-by: Sireesh Kodali <sireeshkodali1@gmail.com> > >> > >> Thank you for your patch. There is something to discuss/improve. > >> > >> Please use existing bindings or example-schema as a starting point. Half > >> of my review could be skipped if you just followed what we already have > >> in the tree. > >> > >> Some of these qcom specific properties already exist but you decided to > >> write them differently... please don't, rather reuse the code. > >> > > > > Thank you for your review, I will make the chnages as appropriate in v2. > >> (...) > >> > >>> + > >>> +maintainers: > >>> + - Bjorn Andersson <bjorn.andersson@linaro.org> > >>> + > >>> +description: > >>> + This document defines the binding for a component that loads and boots > >>> + firmware on the Qualcomm WCNSS core. > >>> + > >>> +properties: > >>> + compatible: > >>> + oneOf: > >>> + - items: > >>> + - enum: > >>> + - qcom,pronto-v2-pil > >>> + - enum: > >>> + - qcom,pronto > >> > >> This does not look correct. The fallback compatible should not change. > >> What is more, it was not documented in original binding, so this should > >> be done in separate patch. > >> > > > > This was not a change to the fallback compatible. > > You made it an enum, so you expect it to use different fallback for > different cases. > > > msm8916.dtsi's wcnss > > node has "qcom,pronto" as the compatible string, which is why this was > > added. It is however not documented in the txt file. Is it sufficient to > > add a note in the commit message, or should it be split into a separate > > commit? > > Please split it, assuming that fallback is correct. Maybe the fallback > is wrong? The code doesn't recognize "qcom,pronto", so perhaps the best solution is to just remove that compatible from msm8916.dtsi? > > > > >>> + - items: > >> > >> No need for items, it's just one item. > >> > >>> + - enum: > >>> + - qcom,riva-pil > >>> + - qcom,pronto-v1-pil > >>> + - qcom,pronto-v2-pil > >>> + > >>> + reg: > >>> + description: must specify the base address and size of the CCU, DXE and PMU > >>> + register blocks > >> > >> New line after "decription:", drop "must specify" and start with capital > >> letter. > >> > >> You need maxItems: 3 > >> > > > > Will fix in v2 > >> > >>> + > >>> + reg-names: > >>> + items: > >>> + - const: ccu > >>> + - const: dxe > >>> + - const: pmu > >>> + > >>> + interrupts-extended: > >>> + description: > >>> + Interrupt lines > >> > >> Skip description, it's obvious. > >> > >> It should be only "interrupts", not extended. > >> > >>> + minItems: 2 > >>> + maxItems: 5 > >>> + > >>> + interrupt-names: > >>> + minItems: 2 > >>> + maxItems: 5 > >> > >> Names should be clearly defined. They were BTW defined in original > >> bindings, so you should not remove them. This makes me wonder what else > >> did you remove from original bindings... > >> > >> Please document all deviations from pure conversion in the commit msg. > >> It's a second "hidden" difference. > >> > > > > Sorry, this was meant to be a pure txt->YAML conversion. The missing > > interrupt names was accidental, and will be fixed in v2. > >>> + > >>> + firmware-name: > >>> + $ref: /schemas/types.yaml#/definitions/string > >>> + description: Relative firmware image path for the WCNSS core. Defaults to > >>> + "wcnss.mdt". > >> > >> > >> Blank line after "description:". This applies to other places as well. > >> > >> Remove "Defailts to ..." and just add "default" schema. > >> > > > > Will be fixed in v2 > >>> + > >>> + vddpx-supply: > >>> + description: Reference to the PX regulator to be held on behalf of the > >>> + booting of the WCNSS core > >>> + > >>> + vddmx-supply: > >>> + description: Reference to the MX regulator to be held on behalf of the > >>> + booting of the WCNSS core. > >>> + > >>> + vddcx-supply: > >>> + description: Reference to the CX regulator to be held on behalf of the > >>> + booting of the WCNSS core. > >> > >> s/Reference to the// > >> > >>> + > >>> + power-domains: > >>> + $ref: /schemas/types.yaml#/definitions/phandle-array > >>> + description: References to the power domains that need to be held on > >>> + behalf of the booting WCNSS core > >> > >> 1. Ditto. > >> 2. No need for ref > >> 3. maxItems > >> > >>> + > >>> + power-domain-names: > >>> + $ref: /schemas/types.yaml#/definitions/string-array > >> > >> No need for ref, skip description. > >> > >>> + description: Names of the power domains > >>> + items: > >>> + - const: cx > >>> + - const: mx > >>> + > >>> + qcom,smem-states: > >>> + $ref: /schemas/types.yaml#/definitions/phandle-array > >>> + description: States used by the AP to signal the WCNSS core that it should > >>> + shutdown > >>> + items: > >>> + - description: Stop the modem > >>> + > >>> + qcom,smem-state-names: > >>> + $ref: /schemas/types.yaml#/definitions/string-array > >> > >> No need for ref. Really, it does not appear in any of existing bindings > >> for smem-state-names, so how did you get it? > >> > > > > The smem nodes were copied from /remoteproc/qcom,sdm845-adsp-pil.yaml > > Hm, indeed, you're right. There are few files having here ref. I'll fix > these. > > > > >>> + description: The names of the state bits used for SMP2P output > >>> + items: > >>> + - const: stop > >>> + > >>> + memory-region: > >>> + maxItems: 1 > >>> + description: Reference to the reserved-memory for the WCNSS core > >>> + > >>> + smd-edge: > >>> + type: object > >>> + description: > >>> + Qualcomm Shared Memory subnode which represents communication edge, > >>> + channels and devices related to the ADSP. > >> > >> You should reference /schemas/soc/qcom/qcom,smd.yaml > > > > Will be done in v2 > >> > >>> + > >>> + iris: > >> > >> Generic node name... what is "iris"? > >> > > Iris is the RF module, I'll make the description better > > RF like wifi? Then the property name should be "wifi". RF like wifi and bluetooth. However there are wifi and bt subnodes in the smd-edge subnode. Iris is just the antenna hardware if I understand correctly. Also this is just a documentation of the existing nodes that are present in msm8916.dtsi, but for whatever reason their documentation was missing in the txt file. Without adding this node in the YAML dtb_check fails. > > > Best regards, > Krzysztof
On Thu May 12, 2022 at 2:06 PM IST, Krzysztof Kozlowski wrote: > On 12/05/2022 08:50, Sireesh Kodali wrote: > >>> + description: The names of the state bits used for SMP2P output > >>> + items: > >>> + - const: stop > >>> + > >>> + memory-region: > >>> + maxItems: 1 > >>> + description: Reference to the reserved-memory for the WCNSS core > >>> + > >>> + smd-edge: > >>> + type: object > >>> + description: > >>> + Qualcomm Shared Memory subnode which represents communication edge, > >>> + channels and devices related to the ADSP. > >> > >> You should reference /schemas/soc/qcom/qcom,smd.yaml > > It seems it is not a SMD driver so above reference is not correct. This > should be probably described in its own schema, I just need to > understand what's this... > The smd-edge node describes the smd channels used to communicate with the remote processor. For wcnss that would be the remote proc id, and the channels for bt and wifi (both separate). There's a similar node for adsp and q6v5. > > Best regards, > Krzysztof
On Wed May 11, 2022 at 10:21 PM IST, Dmitry Baryshkov wrote: > On 11/05/2022 19:15, Sireesh Kodali wrote: > > Add support for the Audio DSP PIL found on the Qualcomm MSM8953 > > platform. The same configuration is used on all SoCs based on the > > MSM8953 platform (SDM450, SDA450, SDM625, SDM632, APQ8053). > > > > Signed-off-by: Sireesh Kodali <sireeshkodali1@gmail.com> > > --- > > drivers/remoteproc/qcom_q6v5_pas.c | 31 ++++++++++++++++++++++++++++++ > > 1 file changed, 31 insertions(+) > > > > diff --git a/drivers/remoteproc/qcom_q6v5_pas.c b/drivers/remoteproc/qcom_q6v5_pas.c > > index 1ae47cc153e5..4dcb714a1468 100644 > > --- a/drivers/remoteproc/qcom_q6v5_pas.c > > +++ b/drivers/remoteproc/qcom_q6v5_pas.c > > @@ -617,7 +617,37 @@ static const struct adsp_data sm8350_adsp_resource = { > > .ssctl_id = 0x14, > > }; > > > > +static const struct adsp_data msm8953_adsp_resource = { > > + .crash_reason_smem = 423, > > + .firmware_name = "adsp.mdt", > > + .pas_id = 1, > > + .has_aggre2_clk = false, > > + .auto_boot = true, > > + .proxy_pd_names = (char*[]){ > > + "cx", > > + NULL, > > + }, > > + .ssr_name = "lpass", > > + .sysmon_name = "adsp", > > + .ssctl_id = 0x14, > > +}; > > + > > Also it looks like this item is identical to msm8996_adsp_resources. So > you can existing structure instead. > Yes, the msm8953 item is identical to the msm8996_adsp_resource. I wasn't aware that we could re-use structs like this, I will do that in v2 of this patch > > static const struct adsp_data msm8996_adsp_resource = { > > + .crash_reason_smem = 423, > > + .firmware_name = "adsp.mdt", > > + .pas_id = 1, > > + .has_aggre2_clk = false, > > + .auto_boot = true, > > + .proxy_pd_names = (char*[]){ > > + "cx", > > + NULL, > > + }, > > + .ssr_name = "lpass", > > + .sysmon_name = "adsp", > > + .ssctl_id = 0x14, > > +}; > > + > > +static const struct adsp_data msm8998_adsp_resource = { > > .crash_reason_smem = 423, > > .firmware_name = "adsp.mdt", > > .pas_id = 1, > > @@ -850,6 +880,7 @@ static const struct adsp_data sdx55_mpss_resource = { > > static const struct of_device_id adsp_of_match[] = { > > { .compatible = "qcom,msm8974-adsp-pil", .data = &adsp_resource_init}, > > { .compatible = "qcom,msm8996-adsp-pil", .data = &msm8996_adsp_resource}, > > + { .compatible = "qcom,msm8953-adsp-pil", .data = &msm8953_adsp_resource}, > > { .compatible = "qcom,msm8996-slpi-pil", .data = &slpi_resource_init}, > > { .compatible = "qcom,msm8998-adsp-pas", .data = &msm8996_adsp_resource}, > > { .compatible = "qcom,msm8998-slpi-pas", .data = &slpi_resource_init}, > > > -- > With best wishes > Dmitry Thanks, Sireesh
On 12/05/2022 11:32, Sireesh Kodali wrote: >>>>> + - enum: >>>>> + - qcom,pronto-v2-pil >>>>> + - enum: >>>>> + - qcom,pronto >>>> >>>> This does not look correct. The fallback compatible should not change. >>>> What is more, it was not documented in original binding, so this should >>>> be done in separate patch. >>>> >>> >>> This was not a change to the fallback compatible. >> >> You made it an enum, so you expect it to use different fallback for >> different cases. >> >>> msm8916.dtsi's wcnss >>> node has "qcom,pronto" as the compatible string, which is why this was >>> added. It is however not documented in the txt file. Is it sufficient to >>> add a note in the commit message, or should it be split into a separate >>> commit? >> >> Please split it, assuming that fallback is correct. Maybe the fallback >> is wrong? > > The code doesn't recognize "qcom,pronto", so perhaps the best solution > is to just remove that compatible from msm8916.dtsi? Eh, I don't know. You need to check, maybe also in downstream sources. (...) >>>> >>>>> + >>>>> + iris: >>>> >>>> Generic node name... what is "iris"? >>>> >>> Iris is the RF module, I'll make the description better >> >> RF like wifi? Then the property name should be "wifi". > > RF like wifi and bluetooth. However there are wifi and bt subnodes in > the smd-edge subnode. Iris is just the antenna hardware if I understand > correctly. Also this is just a documentation of the existing nodes that > are present in msm8916.dtsi, but for whatever reason their documentation > was missing in the txt file. Without adding this node in the YAML > dtb_check fails. It seems commit fd52bdae9ab0 ("wcn36xx: Disable 5GHz for wcn3620") added usage of "iris" property but did not document it in the bindings. You can fix it by documenting (separate patch) existing practice or document with changing the node name. I am not sure if it is worth the effort, so just new patch please. Best regards, Krzysztof
On Thu May 12, 2022 at 4:32 PM IST, Krzysztof Kozlowski wrote: > On 12/05/2022 11:32, Sireesh Kodali wrote: > >>>>> + - enum: > >>>>> + - qcom,pronto-v2-pil > >>>>> + - enum: > >>>>> + - qcom,pronto > >>>> > >>>> This does not look correct. The fallback compatible should not change. > >>>> What is more, it was not documented in original binding, so this should > >>>> be done in separate patch. > >>>> > >>> > >>> This was not a change to the fallback compatible. > >> > >> You made it an enum, so you expect it to use different fallback for > >> different cases. > >> > >>> msm8916.dtsi's wcnss > >>> node has "qcom,pronto" as the compatible string, which is why this was > >>> added. It is however not documented in the txt file. Is it sufficient to > >>> add a note in the commit message, or should it be split into a separate > >>> commit? > >> > >> Please split it, assuming that fallback is correct. Maybe the fallback > >> is wrong? > > > > The code doesn't recognize "qcom,pronto", so perhaps the best solution > > is to just remove that compatible from msm8916.dtsi? > > Eh, I don't know. You need to check, maybe also in downstream sources. > I just checked, it seems "qcom,pronto" is used by the wcnss driver in /net. So both "qcom,pronto-v2-pil" and "qcom,pronto" need to be present, but the latter wasn't documented. > (...) > > >>>> > >>>>> + > >>>>> + iris: > >>>> > >>>> Generic node name... what is "iris"? > >>>> > >>> Iris is the RF module, I'll make the description better > >> > >> RF like wifi? Then the property name should be "wifi". > > > > RF like wifi and bluetooth. However there are wifi and bt subnodes in > > the smd-edge subnode. Iris is just the antenna hardware if I understand > > correctly. Also this is just a documentation of the existing nodes that > > are present in msm8916.dtsi, but for whatever reason their documentation > > was missing in the txt file. Without adding this node in the YAML > > dtb_check fails. > > It seems commit fd52bdae9ab0 ("wcn36xx: Disable 5GHz for wcn3620") > added usage of "iris" property but did not document it in the bindings. > > You can fix it by documenting (separate patch) existing practice or > document with changing the node name. I am not sure if it is worth the > effort, so just new patch please. > I'll make a 2 separate patches, documenting the extra "qcom,pronto" compatible, and the iris subnode. Thanks, Sireesh > Best regards, > Krzysztof