[PATCH v2] irqchip/gicv3: Workaround for NVIDIA erratum T241-FABRIC-4

[Shanker Donthineni]

On 3/16/23 11:00, Marc Zyngier wrote:
> External email: Use caution opening links or attachments
> 
> 
> On 2023-03-16 15:10, Sudeep Holla wrote:
>> On Wed, Mar 15, 2023 at 07:27:14AM -0500, Shanker Donthineni wrote:
>>> Hi Marc,
>>>
>>> On 3/15/23 03:34, Marc Zyngier wrote:
>>> > Please don't duplicate existing code. There is already the required
>>> > infrastructure in drivers/firmware/smccc/soc_id.c. All you need to do
>>> > is:
>>> >
>>> > - disassociate the SMCCC probing from the device registration
>>> >
>>> > - probe the SOC_ID early
>>> >
>>> > - add accessors for the relevant data
>>> >
>>> > - select ARM_SMCCC_SOD_ID/ARM_SMCCC_DISCOVERY from the GICv3 Kconfig
>>>
>>>
>>> I have not modified soc_id.c as it expects to be loaded as a module
>>> with
>>> the use of module_init() and module_exit() functions. The exported
>>> symbols
>>> in soc_id driver cannot be accessed from the built-in code.
>>>
>>> Agree, the SOD-ID discovery code was duplicated.
>>>
>>> Please guide me if the below approach is okay?
>>>
>>> 1) Probe the SOC-ID in arm_smccc_version_init() and export two
>>> functions
>>> arm_smccc_get_soc_id_version() and arm_smccc_get_soc_id_revision().
>>>
>>> --- a/drivers/firmware/smccc/smccc.c
>>> +++ b/drivers/firmware/smccc/smccc.c
>>> @@ -17,9 +17,13 @@ static enum arm_smccc_conduit smccc_conduit =
>>> SMCCC_CONDUIT_NONE;
>>>
>>>  bool __ro_after_init smccc_trng_available = false;
>>>  u64 __ro_after_init smccc_has_sve_hint = false;
>>> +s32 __ro_after_init smccc_soc_id_version = SMCCC_RET_NOT_SUPPORTED;
>>> +s32 __ro_after_init smccc_soc_id_revision = SMCCC_RET_NOT_SUPPORTED;
>>>
>>>  void __init arm_smccc_version_init(u32 version, enum
>>> arm_smccc_conduit conduit)
>>>  {
>>> +       struct arm_smccc_res res;
>>> +
>>>         smccc_version = version;
>>>         smccc_conduit = conduit;
>>>
>>> @@ -27,6 +31,18 @@ void __init arm_smccc_version_init(u32 version,
>>> enum arm_smccc_conduit conduit)
>>>         if (IS_ENABLED(CONFIG_ARM64_SVE) &&
>>>             smccc_version >= ARM_SMCCC_VERSION_1_3)
>>>                 smccc_has_sve_hint = true;
>>> +
>>> +       if ((smccc_version >= ARM_SMCCC_VERSION_1_2) &&
>>> +           (smccc_conduit != SMCCC_CONDUIT_NONE)) {
>>> +               arm_smccc_1_1_invoke(ARM_SMCCC_ARCH_FEATURES_FUNC_ID,
>>> +                                    ARM_SMCCC_ARCH_SOC_ID, &res);
>>> +               if ((s32)res.a0 >= 0) {
>>> +                       arm_smccc_1_1_invoke(ARM_SMCCC_ARCH_SOC_ID, 0,
>>> &res);
>>> +                       smccc_soc_id_version = (s32)res.a0;
>>> +                       arm_smccc_1_1_invoke(ARM_SMCCC_ARCH_SOC_ID, 1,
>>> &res);
>>> +                       smccc_soc_id_revision = (s32)res.a0;
>>> +               }
>>> +       }
>>>  }
>>>
>>>
>>> +s32 arm_smccc_get_soc_id_version(void)
>>> +{
>>> +       return smccc_soc_id_version;
>>> +}
>>> +EXPORT_SYMBOL_GPL(arm_smccc_get_soc_id_version);
>>> +
>>> +s32 arm_smccc_get_soc_id_revision(void)
>>> +{
>>> +       return smccc_soc_id_revision;
>>> +}
>>> +EXPORT_SYMBOL_GPL(arm_smccc_get_soc_id_revision);
>>>
>>>
>>
>> Overall, it looks OK to me. However I see neither the gic nor the
>> soc_id
>> can be build as module atm. So do we really need the export symbols if
>> no
>> other modules are using it ?
> 
> It really shouldn't be exported. Having accessors should be enough.
> 

Thanks, I'll remove in v3 patch.


-Shanker