| Message ID | 20240906191438.4104329-2-quic_msavaliy@quicinc.com |
|---|---|
| State | Changes Requested |
| Delegated to: | Andi Shyti |
| Headers | show |
| Series | Enable shared SE support over I2C | expand |
On 06/09/2024 21:14, Mukesh Kumar Savaliya wrote: > Adds qcom,shared-se flag usage. Use this when particular I2C serial > controller needs to be shared between two subsystems. <form letter> Please use scripts/get_maintainers.pl to get a list of necessary people and lists to CC (and consider --no-git-fallback argument). It might happen, that command when run on an older kernel, gives you outdated entries. Therefore please be sure you base your patches on recent Linux kernel. Tools like b4 or scripts/get_maintainer.pl provide you proper list of people, so fix your workflow. Tools might also fail if you work on some ancient tree (don't, instead use mainline) or work on fork of kernel (don't, instead use mainline). Just use b4 and everything should be fine, although remember about `b4 prep --auto-to-cc` if you added new patches to the patchset. </form letter> You already got this comment, so how many times it has to be repeated? Your process is just wrong if you do not use the tools for this. > > SE = Serial Engine, meant for I2C controller here. > TRE = Transfer Ring Element, refers to Queued Descriptor. > > Example : > Two clients from different SS can share an I2C SE for same slave device What is SS? > OR their owned slave devices. > Assume I2C Slave EEPROM device connected with I2C controller. > Each client from ADSP SS and APPS Linux SS can perform i2c transactions. > This gets serialized by lock TRE + DMA Transfers + Unlock TRE at HW level. > > Signed-off-by: Mukesh Kumar Savaliya <quic_msavaliy@quicinc.com> > --- > Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml > index 9f66a3bb1f80..ae423127f736 100644 > --- a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml > +++ b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml > @@ -60,6 +60,10 @@ properties: > power-domains: > maxItems: 1 > > + qcom,shared-se: > + description: True if I2C needs to be shared between two or more subsystems. What is a subsystem? With commit msg I still do not understand this. Maybe presence of hwlock defines it anyway, so this is redundant? Best regards, Krzysztof
Thanks Krzysztof. On 9/7/2024 2:34 PM, Krzysztof Kozlowski wrote: > On 06/09/2024 21:14, Mukesh Kumar Savaliya wrote: >> Adds qcom,shared-se flag usage. Use this when particular I2C serial >> controller needs to be shared between two subsystems. > > <form letter> > Please use scripts/get_maintainers.pl to get a list of necessary people > and lists to CC (and consider --no-git-fallback argument). It might > happen, that command when run on an older kernel, gives you outdated > entries. Therefore please be sure you base your patches on recent Linux > kernel. > > Tools like b4 or scripts/get_maintainer.pl provide you proper list of > people, so fix your workflow. Tools might also fail if you work on some > ancient tree (don't, instead use mainline) or work on fork of kernel > (don't, instead use mainline). Just use b4 and everything should be > fine, although remember about `b4 prep --auto-to-cc` if you added new > patches to the patchset. > </form letter> > > You already got this comment, so how many times it has to be repeated? > Your process is just wrong if you do not use the tools for this. > Sorry, I was already using scripts/get_maintainer.pl but i kept everyone into To list (That's my mistake here). I shall keep maintainers in TO list and rest in CC list. Question: With <Form Letter> , are you asking to add letter in this first patch ? I have cover letter, but it will get removed when patch gets merged. Please help suggest and clarify. > >> >> SE = Serial Engine, meant for I2C controller here. >> TRE = Transfer Ring Element, refers to Queued Descriptor. >> >> Example : >> Two clients from different SS can share an I2C SE for same slave device > > What is SS? > SS = Subsystem (EE - Execution Environment, can be Apps processor/TZ/Modem/ADSP etc). Let me add this too in next patch. >> OR their owned slave devices. >> Assume I2C Slave EEPROM device connected with I2C controller. >> Each client from ADSP SS and APPS Linux SS can perform i2c transactions. >> This gets serialized by lock TRE + DMA Transfers + Unlock TRE at HW level. >> >> Signed-off-by: Mukesh Kumar Savaliya <quic_msavaliy@quicinc.com> >> --- >> Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml | 4 ++++ >> 1 file changed, 4 insertions(+) >> >> diff --git a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >> index 9f66a3bb1f80..ae423127f736 100644 >> --- a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >> +++ b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >> @@ -60,6 +60,10 @@ properties: >> power-domains: >> maxItems: 1 >> >> + qcom,shared-se: >> + description: True if I2C needs to be shared between two or more subsystems. > > What is a subsystem? With commit msg I still do not understand this. SS = Subsystem (EE - Execution Environment, can be Apps processor/TZ/Modem/ADSP etc). Let me add EE too with full form. > Maybe presence of hwlock defines it anyway, so this is redundant? No, this flag is required. As hwlock comes into picture if this flag is defined. So flag is acting as a condition to take hwlock TRE descriptor(transfer ring element). Hope i could answer your query. > Best regards, > Krzysztof >
On 10/09/2024 11:09, Mukesh Kumar Savaliya wrote: > Thanks Krzysztof. > > On 9/7/2024 2:34 PM, Krzysztof Kozlowski wrote: >> On 06/09/2024 21:14, Mukesh Kumar Savaliya wrote: >>> Adds qcom,shared-se flag usage. Use this when particular I2C serial >>> controller needs to be shared between two subsystems. >> >> <form letter> >> Please use scripts/get_maintainers.pl to get a list of necessary people >> and lists to CC (and consider --no-git-fallback argument). It might >> happen, that command when run on an older kernel, gives you outdated >> entries. Therefore please be sure you base your patches on recent Linux >> kernel. >> >> Tools like b4 or scripts/get_maintainer.pl provide you proper list of >> people, so fix your workflow. Tools might also fail if you work on some >> ancient tree (don't, instead use mainline) or work on fork of kernel >> (don't, instead use mainline). Just use b4 and everything should be >> fine, although remember about `b4 prep --auto-to-cc` if you added new >> patches to the patchset. >> </form letter> >> >> You already got this comment, so how many times it has to be repeated? >> Your process is just wrong if you do not use the tools for this. >> > Sorry, I was already using scripts/get_maintainer.pl but i kept everyone > into To list (That's my mistake here). I shall keep maintainers in TO > list and rest in CC list. No, To or Cc does not matter. Your list is just incomplete. > > Question: With <Form Letter> , are you asking to add letter in this > first patch ? I have cover letter, but it will get removed when patch > gets merged. Please help suggest and clarify. No, it's just template. Form letter... I am just bored to repeat the same comment. >> >>> >>> SE = Serial Engine, meant for I2C controller here. >>> TRE = Transfer Ring Element, refers to Queued Descriptor. >>> >>> Example : >>> Two clients from different SS can share an I2C SE for same slave device >> >> What is SS? >> > SS = Subsystem (EE - Execution Environment, can be Apps > processor/TZ/Modem/ADSP etc). Let me add this too in next patch. Yes, please explain in the binding itself. >>> OR their owned slave devices. >>> Assume I2C Slave EEPROM device connected with I2C controller. >>> Each client from ADSP SS and APPS Linux SS can perform i2c transactions. >>> This gets serialized by lock TRE + DMA Transfers + Unlock TRE at HW level. >>> >>> Signed-off-by: Mukesh Kumar Savaliya <quic_msavaliy@quicinc.com> >>> --- >>> Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml | 4 ++++ >>> 1 file changed, 4 insertions(+) >>> >>> diff --git a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >>> index 9f66a3bb1f80..ae423127f736 100644 >>> --- a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >>> +++ b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >>> @@ -60,6 +60,10 @@ properties: >>> power-domains: >>> maxItems: 1 >>> >>> + qcom,shared-se: >>> + description: True if I2C needs to be shared between two or more subsystems. >> >> What is a subsystem? With commit msg I still do not understand this. > SS = Subsystem (EE - Execution Environment, can be Apps > processor/TZ/Modem/ADSP etc). Let me add EE too with full form. >> Maybe presence of hwlock defines it anyway, so this is redundant? > No, this flag is required. As hwlock comes into picture if this flag is Flag is required? By what? Sorry, you push your downstream solution to us. > defined. So flag is acting as a condition to take hwlock TRE > descriptor(transfer ring element). Hope i could answer your query. Hm, not sure, maybe indeed hwlock would not be enough. However I think existing binding misses hwlock property. Best regards, Krzysztof
Thanks For your reviews, On 9/10/2024 3:24 PM, Krzysztof Kozlowski wrote: > On 10/09/2024 11:09, Mukesh Kumar Savaliya wrote: >> Thanks Krzysztof. >> >> On 9/7/2024 2:34 PM, Krzysztof Kozlowski wrote: >>> On 06/09/2024 21:14, Mukesh Kumar Savaliya wrote: >>>> Adds qcom,shared-se flag usage. Use this when particular I2C serial >>>> controller needs to be shared between two subsystems. >>> >>> <form letter> >>> Please use scripts/get_maintainers.pl to get a list of necessary people >>> and lists to CC (and consider --no-git-fallback argument). It might >>> happen, that command when run on an older kernel, gives you outdated >>> entries. Therefore please be sure you base your patches on recent Linux >>> kernel. >>> >>> Tools like b4 or scripts/get_maintainer.pl provide you proper list of >>> people, so fix your workflow. Tools might also fail if you work on some >>> ancient tree (don't, instead use mainline) or work on fork of kernel >>> (don't, instead use mainline). Just use b4 and everything should be >>> fine, although remember about `b4 prep --auto-to-cc` if you added new >>> patches to the patchset. >>> </form letter> >>> >>> You already got this comment, so how many times it has to be repeated? >>> Your process is just wrong if you do not use the tools for this. >>> >> Sorry, I was already using scripts/get_maintainer.pl but i kept everyone >> into To list (That's my mistake here). I shall keep maintainers in TO >> list and rest in CC list. > > No, To or Cc does not matter. Your list is just incomplete. > Got it, sorry for the trouble. It seems i missed below 3 names adding into reviewers by copy paste mistake. I hope this makes it complete now and will add them in V3. Bryan O'Donoghue" <bryan.odonoghue@linaro.org> Krzysztof Kozlowski <krzk+dt@kernel.org> Rob Herring <robh@kernel.org> >> >> Question: With <Form Letter> , are you asking to add letter in this >> first patch ? I have cover letter, but it will get removed when patch >> gets merged. Please help suggest and clarify. > > No, it's just template. Form letter... I am just bored to repeat the > same comment. > Sorry for that. I hope i could catch now as per above missing list. >>> >>>> >>>> SE = Serial Engine, meant for I2C controller here. >>>> TRE = Transfer Ring Element, refers to Queued Descriptor. >>>> >>>> Example : >>>> Two clients from different SS can share an I2C SE for same slave device >>> >>> What is SS? >>> >> SS = Subsystem (EE - Execution Environment, can be Apps >> processor/TZ/Modem/ADSP etc). Let me add this too in next patch. > > Yes, please explain in the binding itself. > ok, Sure. >>>> OR their owned slave devices. >>>> Assume I2C Slave EEPROM device connected with I2C controller. >>>> Each client from ADSP SS and APPS Linux SS can perform i2c transactions. >>>> This gets serialized by lock TRE + DMA Transfers + Unlock TRE at HW level. >>>> >>>> Signed-off-by: Mukesh Kumar Savaliya <quic_msavaliy@quicinc.com> >>>> --- >>>> Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml | 4 ++++ >>>> 1 file changed, 4 insertions(+) >>>> >>>> diff --git a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >>>> index 9f66a3bb1f80..ae423127f736 100644 >>>> --- a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >>>> +++ b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml >>>> @@ -60,6 +60,10 @@ properties: >>>> power-domains: >>>> maxItems: 1 >>>> >>>> + qcom,shared-se: >>>> + description: True if I2C needs to be shared between two or more subsystems. >>> >>> What is a subsystem? With commit msg I still do not understand this. >> SS = Subsystem (EE - Execution Environment, can be Apps >> processor/TZ/Modem/ADSP etc). Let me add EE too with full form. >>> Maybe presence of hwlock defines it anyway, so this is redundant? >> No, this flag is required. As hwlock comes into picture if this flag is > > Flag is required? By what? Sorry, you push your downstream solution to us. > Let me explain, Using this flag to take hwlock via TRE @ [PATCH v2 2/4] We need this to lock SE protecting from other SS transfers until unlocked. Hence shared-se flag becomes a decision marker. drivers/dma/qcom/gpi.c => gpi_create_i2c_tre() + /* create lock tre for first tranfser */ + if (i2c->shared_se && i2c->first_msg) { Question: what exactly you mean "Maybe presence of hwlock defines it anyway" ? I am open to consider all upstream solutions, trying to understand your suggestions and comments. >> defined. So flag is acting as a condition to take hwlock TRE >> descriptor(transfer ring element). Hope i could answer your query. > > Hm, not sure, maybe indeed hwlock would not be enough. However I think > existing binding misses hwlock property. > Let me clarify, you may help suggest further. hwlock is a descriptor bit(TRE_I2C_LOCK). "However I think existing binding misses hwlock property" Where shall i keep this hwlock property? what's the usage ? > Best regards, > Krzysztof >
diff --git a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml index 9f66a3bb1f80..ae423127f736 100644 --- a/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml +++ b/Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml @@ -60,6 +60,10 @@ properties: power-domains: maxItems: 1 + qcom,shared-se: + description: True if I2C needs to be shared between two or more subsystems. + type: boolean + reg: maxItems: 1
Adds qcom,shared-se flag usage. Use this when particular I2C serial controller needs to be shared between two subsystems. SE = Serial Engine, meant for I2C controller here. TRE = Transfer Ring Element, refers to Queued Descriptor. Example : Two clients from different SS can share an I2C SE for same slave device OR their owned slave devices. Assume I2C Slave EEPROM device connected with I2C controller. Each client from ADSP SS and APPS Linux SS can perform i2c transactions. This gets serialized by lock TRE + DMA Transfers + Unlock TRE at HW level. Signed-off-by: Mukesh Kumar Savaliya <quic_msavaliy@quicinc.com> --- Documentation/devicetree/bindings/i2c/qcom,i2c-geni-qcom.yaml | 4 ++++ 1 file changed, 4 insertions(+)