Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1656319 > unrolled thread

Re: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table

Started byMark Rutland <mark.rutland@arm.com>
First post2017-06-02 17:10 +0200
Last post2017-06-02 22:50 +0200
Articles 4 — 2 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match  table Mark Rutland <mark.rutland@arm.com> - 2017-06-02 17:10 +0200
    Re: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table Hoan Tran <hotran@apm.com> - 2017-06-02 19:00 +0200
      Re: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match  table Mark Rutland <mark.rutland@arm.com> - 2017-06-02 19:30 +0200
        Re: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table Hoan Tran <hotran@apm.com> - 2017-06-02 22:50 +0200

#1656319 — Re: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table

FromMark Rutland <mark.rutland@arm.com>
Date2017-06-02 17:10 +0200
SubjectRe: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table
Message-ID<tNWn8-2Pu-23@gated-at.bofh.it>
Hi Hoan,

Apologies for the last reply.

On Mon, Apr 03, 2017 at 09:47:55AM -0700, Hoan Tran wrote:
> +static const struct acpi_device_id xgene_pmu_acpi_type_match[] = {
> +	{"APMC0D5D", PMU_TYPE_L3C},
> +	{"APMC0D5E", PMU_TYPE_IOB},
> +	{"APMC0D5F", PMU_TYPE_MCB},
> +	{"APMC0D60", PMU_TYPE_MC},
> +	{},
> +};
> +
> +static const struct acpi_device_id *xgene_pmu_acpi_match_type(
> +					const struct acpi_device_id *ids,
> +					struct acpi_device *adev)
> +{
> +	const struct acpi_device_id *match_id = NULL;
> +	const struct acpi_device_id *id;
> +
> +	for (id = ids; id->id[0] || id->cls; id++) {
> +		if (!acpi_match_device_ids(adev, id))
> +			match_id = id;
> +		else if (match_id)
> +			break;
> +	}
> +
> +	return match_id;
> +}

I don't believe this look is necessary. AFAICT, acpi_match_device_ids()
already iterates over the id table it is given.

> +
>  static acpi_status acpi_pmu_dev_add(acpi_handle handle, u32 level,
>  				    void *data, void **return_value)
>  {
> +	const struct acpi_device_id *acpi_id;
>  	struct xgene_pmu *xgene_pmu = data;
>  	struct xgene_pmu_dev_ctx *ctx;
>  	struct acpi_device *adev;
> @@ -1059,17 +1085,11 @@ static acpi_status acpi_pmu_dev_add(acpi_handle handle, u32 level,
>  	if (acpi_bus_get_status(adev) || !adev->status.present)
>  		return AE_OK;
>  
> -	if (!strcmp(acpi_device_hid(adev), "APMC0D5D"))
> -		ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, PMU_TYPE_L3C);
> -	else if (!strcmp(acpi_device_hid(adev), "APMC0D5E"))
> -		ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, PMU_TYPE_IOB);
> -	else if (!strcmp(acpi_device_hid(adev), "APMC0D5F"))
> -		ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, PMU_TYPE_MCB);
> -	else if (!strcmp(acpi_device_hid(adev), "APMC0D60"))
> -		ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, PMU_TYPE_MC);
> -	else
> -		ctx = NULL;
> +	acpi_id = xgene_pmu_acpi_match_type(xgene_pmu_acpi_type_match, adev);
> +	if (!acpi_id)
> +		return AE_OK;

As above, and as I covered in my reply to v1, I think the above should
be:

	acpi_id = acpi_match_device_ids(adev, xgene_pmu_acpi_type_match);
	if (!acpi_id)
		return AE_OK;

... or am I missing something?

With that change:

Acked-by: Mark Rutland <mark.rutland@arm.com>

Thanks,
Mark.

>  
> +	ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, (u32)acpi_id->driver_data);
>  	if (!ctx)
>  		return AE_OK;
>  
> -- 
> 1.9.1
>

[toc] | [next] | [standalone]

#1656394 — Re: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table

FromHoan Tran <hotran@apm.com>
Date2017-06-02 19:00 +0200
SubjectRe: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table
Message-ID<tNY5z-3Ka-1@gated-at.bofh.it>
In reply to#1656319
Hi Mark

On Fri, Jun 2, 2017 at 7:59 AM, Mark Rutland <mark.rutland@arm.com> wrote:
> Hi Hoan,
>
> Apologies for the last reply.
>
> On Mon, Apr 03, 2017 at 09:47:55AM -0700, Hoan Tran wrote:
>> +static const struct acpi_device_id xgene_pmu_acpi_type_match[] = {
>> +     {"APMC0D5D", PMU_TYPE_L3C},
>> +     {"APMC0D5E", PMU_TYPE_IOB},
>> +     {"APMC0D5F", PMU_TYPE_MCB},
>> +     {"APMC0D60", PMU_TYPE_MC},
>> +     {},
>> +};
>> +
>> +static const struct acpi_device_id *xgene_pmu_acpi_match_type(
>> +                                     const struct acpi_device_id *ids,
>> +                                     struct acpi_device *adev)
>> +{
>> +     const struct acpi_device_id *match_id = NULL;
>> +     const struct acpi_device_id *id;
>> +
>> +     for (id = ids; id->id[0] || id->cls; id++) {
>> +             if (!acpi_match_device_ids(adev, id))
>> +                     match_id = id;
>> +             else if (match_id)
>> +                     break;
>> +     }
>> +
>> +     return match_id;
>> +}
>
> I don't believe this look is necessary. AFAICT, acpi_match_device_ids()
> already iterates over the id table it is given.

The acpi_match_device_ids() function just returns if a device ID is
available on the given list. It does not return the first matching ID.
That's the reason I created this function to find the first matching ID.

>
>> +
>>  static acpi_status acpi_pmu_dev_add(acpi_handle handle, u32 level,
>>                                   void *data, void **return_value)
>>  {
>> +     const struct acpi_device_id *acpi_id;
>>       struct xgene_pmu *xgene_pmu = data;
>>       struct xgene_pmu_dev_ctx *ctx;
>>       struct acpi_device *adev;
>> @@ -1059,17 +1085,11 @@ static acpi_status acpi_pmu_dev_add(acpi_handle handle, u32 level,
>>       if (acpi_bus_get_status(adev) || !adev->status.present)
>>               return AE_OK;
>>
>> -     if (!strcmp(acpi_device_hid(adev), "APMC0D5D"))
>> -             ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, PMU_TYPE_L3C);
>> -     else if (!strcmp(acpi_device_hid(adev), "APMC0D5E"))
>> -             ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, PMU_TYPE_IOB);
>> -     else if (!strcmp(acpi_device_hid(adev), "APMC0D5F"))
>> -             ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, PMU_TYPE_MCB);
>> -     else if (!strcmp(acpi_device_hid(adev), "APMC0D60"))
>> -             ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, PMU_TYPE_MC);
>> -     else
>> -             ctx = NULL;
>> +     acpi_id = xgene_pmu_acpi_match_type(xgene_pmu_acpi_type_match, adev);
>> +     if (!acpi_id)
>> +             return AE_OK;
>
> As above, and as I covered in my reply to v1, I think the above should
> be:
>
>         acpi_id = acpi_match_device_ids(adev, xgene_pmu_acpi_type_match);
>         if (!acpi_id)
>                 return AE_OK;
>
> ... or am I missing something?

The same above.

Thanks
Hoan

>
> With that change:
>
> Acked-by: Mark Rutland <mark.rutland@arm.com>
>
> Thanks,
> Mark.
>
>>
>> +     ctx = acpi_get_pmu_hw_inf(xgene_pmu, adev, (u32)acpi_id->driver_data);
>>       if (!ctx)
>>               return AE_OK;
>>
>> --
>> 1.9.1
>>

[toc] | [prev] | [next] | [standalone]

#1656404

FromMark Rutland <mark.rutland@arm.com>
Date2017-06-02 19:30 +0200
Message-ID<tNYyC-49F-19@gated-at.bofh.it>
In reply to#1656394
On Fri, Jun 02, 2017 at 09:54:32AM -0700, Hoan Tran wrote:
> On Fri, Jun 2, 2017 at 7:59 AM, Mark Rutland <mark.rutland@arm.com> wrote:
> > On Mon, Apr 03, 2017 at 09:47:55AM -0700, Hoan Tran wrote:
> >> +static const struct acpi_device_id *xgene_pmu_acpi_match_type(
> >> +                                     const struct acpi_device_id *ids,
> >> +                                     struct acpi_device *adev)
> >> +{
> >> +     const struct acpi_device_id *match_id = NULL;
> >> +     const struct acpi_device_id *id;
> >> +
> >> +     for (id = ids; id->id[0] || id->cls; id++) {
> >> +             if (!acpi_match_device_ids(adev, id))
> >> +                     match_id = id;
> >> +             else if (match_id)
> >> +                     break;
> >> +     }
> >> +
> >> +     return match_id;
> >> +}
> >
> > I don't believe this look is necessary. AFAICT, acpi_match_device_ids()
> > already iterates over the id table it is given.
> 
> The acpi_match_device_ids() function just returns if a device ID is
> available on the given list. It does not return the first matching ID.
> That's the reason I created this function to find the first matching ID.

Ah, I see. Thanks for correcting me!

Can we use acpi_match_device(ids, &adev->dev), or is that the wrong dev?

Thanks,
Mark.

[toc] | [prev] | [next] | [standalone]

#1656541 — Re: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table

FromHoan Tran <hotran@apm.com>
Date2017-06-02 22:50 +0200
SubjectRe: [PATCH v2 1/3] perf: xgene: Parse PMU subnode from the match table
Message-ID<tO1G9-621-5@gated-at.bofh.it>
In reply to#1656404
Hi Mark,

On Fri, Jun 2, 2017 at 10:23 AM, Mark Rutland <mark.rutland@arm.com> wrote:
> On Fri, Jun 02, 2017 at 09:54:32AM -0700, Hoan Tran wrote:
>> On Fri, Jun 2, 2017 at 7:59 AM, Mark Rutland <mark.rutland@arm.com> wrote:
>> > On Mon, Apr 03, 2017 at 09:47:55AM -0700, Hoan Tran wrote:
>> >> +static const struct acpi_device_id *xgene_pmu_acpi_match_type(
>> >> +                                     const struct acpi_device_id *ids,
>> >> +                                     struct acpi_device *adev)
>> >> +{
>> >> +     const struct acpi_device_id *match_id = NULL;
>> >> +     const struct acpi_device_id *id;
>> >> +
>> >> +     for (id = ids; id->id[0] || id->cls; id++) {
>> >> +             if (!acpi_match_device_ids(adev, id))
>> >> +                     match_id = id;
>> >> +             else if (match_id)
>> >> +                     break;
>> >> +     }
>> >> +
>> >> +     return match_id;
>> >> +}
>> >
>> > I don't believe this look is necessary. AFAICT, acpi_match_device_ids()
>> > already iterates over the id table it is given.
>>
>> The acpi_match_device_ids() function just returns if a device ID is
>> available on the given list. It does not return the first matching ID.
>> That's the reason I created this function to find the first matching ID.
>
> Ah, I see. Thanks for correcting me!
>
> Can we use acpi_match_device(ids, &adev->dev), or is that the wrong dev?

They are subnode device, so they don't have full dev. Because of that,
acpi_match_device doesn't work.

Thanks
Hoan

>
> Thanks,
> Mark.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web