Re: [edk2-devel] [PATCH v2 1/2] ArmPkg: implement EFI_MP_SERVICES_PROTOCOL based on PSCI calls

["Michael Kubacki" <mikuback@linux.microsoft.com>]

Please note that I already turned AuditOnly mode on for the EmbeddedPkg 
patch series.

On 9/7/2022 11:21 AM, Ard Biesheuvel wrote:
> On Wed, 7 Sept 2022 at 17:11, Michael Kubacki
> <mikuback@linux.microsoft.com> wrote:
>>
>> Maintainers have two options for Spell Check as described in the Spell
>> Check CI Plugin readme -
>> https://github.com/tianocore/edk2/tree/master/.pytool/Plugin/SpellCheck
>>
>> 1. Set AuditOnly to "true" - Disables all spelling errors in the package
>> from the failing the build entirely.
>>
>> 2. Extend the dictionary - Add specific words that are allowed. Many
>> packages already do this.
>>
>> 3. IgnoreStandardPaths - Do not check files by path pattern.
>>
>> As a maintainer, you can either opt out of Spell Check from the failing
>> the build entirely by enabling AuditOnly or choose a different level of
>> granularity by simply updating your package CI YAML file.
>>
> 
> Thanks for the explanation - patch coming up.
> 
> 
>> On 9/7/2022 3:35 AM, Ard Biesheuvel wrote:
>>> On Wed, 7 Sept 2022 at 06:35, Rebecca Cran <quic_rcran@quicinc.com> wrote:
>>>>
>>>> Also, CI is failing in the SpellCheck test because of the
>>>> "-mstrict-align" in
>>>> ArmPsciMpServicesDxe.inf.
>>>>
>>>> /home/vsts/work/1/s/ArmPkg/Drivers/ArmPsciMpServicesDxe/ArmPsciMpServicesDxe.inf:55:25
>>>> - Unknown word (mstrict)
>>>>
>>>> Should I add "mstrict" to pytool/Plugin/SpellCheck/cspell.base.yaml ?
>>>>
>>>
>>> This is getting ridiculous.
>>>
>>> I'm all for code quality and automation, but if t we cannot use
>>> compiler command line options that it supported since the dawn of time
>>> without bending over backwards to placate the CI machine, we are
>>> throwing out the baby with the bath water.
>>>
>>> At the very least, maintainers should be able to override the CI - I
>>> spent 2 hours yesterday preparing a submission that ultimately got
>>> rejected because MPIDR is not in the permitted word list (suggestion:
>>> use 'mpidr')
>>>
>>> In the Linux kernel, we have patch check and sparse (which is somewhat
>>> similar to ECC in a sense). Some maintainers run it, some don't, but
>>> in the end, the discretion lies with the person not the machine. This
>>> is the only way that makes sense.
>>>
>>> As the co-maintainer of ArmPkg, ArmVirtPkg, ArmPlatformPkg, OvmfPkg
>>> and EmbeddedPkg, I hereby request that the Ci restrictions are either
>>> lifted entirely for these packages, or relaxed so that a PR can be
>>> merged at my discretion, even if the CI thinks there are unresolved
>>> issues.
>>>
>>>
>>>
>>>> On 9/6/22 22:25, Rebecca Cran wrote:
>>>>> On 9/6/22 22:03, Rebecca Cran wrote:
>>>>>> diff --git a/ArmPkg/ArmPkg.dsc b/ArmPkg/ArmPkg.dsc
>>>>>> index 59fd8f295d4f..4716789402fc 100644
>>>>>> --- a/ArmPkg/ArmPkg.dsc
>>>>>> +++ b/ArmPkg/ArmPkg.dsc
>>>>>> @@ -125,6 +125,7 @@ [Components.common]
>>>>>>       ArmPkg/Drivers/CpuPei/CpuPei.inf
>>>>>>       ArmPkg/Drivers/ArmGic/ArmGicDxe.inf
>>>>>>       ArmPkg/Drivers/ArmGic/ArmGicLib.inf
>>>>>> +  ArmPkg/Drivers/ArmPsciMpServicesDxe/ArmPsciMpServicesDxe.inf
>>>>>>       ArmPkg/Drivers/GenericWatchdogDxe/GenericWatchdogDxe.inf
>>>>>>       ArmPkg/Drivers/TimerDxe/TimerDxe.inf
>>>>> This won't work, since there's no support for 32-bit ARM code.
>>>>> I'll move it into the AARCH64 section.
>>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>>
>>
>>
>> 
>>
>>