linux-kernel - Re: [PATCH 2/2] perf: arm_dsu: Support DSU ACPI devices.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 18 Mar 2020 19:42:13 -0700
From:   Tuan Phan <tuanphan@...eremail.onmicrosoft.com>
To:     Suzuki K Poulose <suzuki.poulose@....com>
Cc:     tuanphan@...amperecomputing.com, patches@...erecomputing.com,
        will@...nel.org, mark.rutland@....com,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        sudeep.holla@....com
Subject: Re: [PATCH 2/2] perf: arm_dsu: Support DSU ACPI devices.

Thanks Suzuki and Robin for quick reply.
Please find my comments below.

> On Mar 18, 2020, at 5:45 PM, Suzuki K Poulose <suzuki.poulose@....com> wrote:
> 
> Hello,
> 
> 
> Please find my comments below.
> 
> On 03/18/2020 12:28 AM, Tuan Phan wrote:
>> Add support for probing device from ACPI node.
>> Each DSU ACPI node defines "cpus" package which
>> each element is the MPIDR of associated cpu.
>> Signed-off-by: Tuan Phan <tuanphan@...amperecomputing.com>
>> ---
>>  drivers/perf/arm_dsu_pmu.c | 53 +++++++++++++++++++++++++++++++++++++++-------
>>  1 file changed, 45 insertions(+), 8 deletions(-)
>> diff --git a/drivers/perf/arm_dsu_pmu.c b/drivers/perf/arm_dsu_pmu.c Tua
>> index 2622900..6ef762c 100644
>> --- a/drivers/perf/arm_dsu_pmu.c
>> +++ b/drivers/perf/arm_dsu_pmu.c
>> @@ -11,6 +11,7 @@
>>  #define DRVNAME		PMUNAME "_pmu"
>>  #define pr_fmt(fmt)	DRVNAME ": " fmt
>>  +#include <linux/acpi.h>
>>  #include <linux/bitmap.h>
>>  #include <linux/bitops.h>
>>  #include <linux/bug.h>
>> @@ -603,18 +604,22 @@ static struct dsu_pmu *dsu_pmu_alloc(struct platform_device *pdev)
>>  }
>>    /**
>> - * dsu_pmu_dt_get_cpus: Get the list of CPUs in the cluster.
>> + * dsu_pmu_get_cpus: Get the list of CPUs in the cluster.
>>   */
>> -static int dsu_pmu_dt_get_cpus(struct device_node *dev, cpumask_t *mask)
>> +static int dsu_pmu_get_cpus(struct platform_device *pdev)
>>  {
>> +#ifndef CONFIG_ACPI
>> +	/* Get the list of CPUs from device tree */
> 
> What if we have a kernel with both:
> 
> CONFIG_OF=y
> CONFIG_ACPI=y
> 
> and boot the kernel on a system with DT ? In other words, the decision
> to choose the DT vs ACPI must be runtime decision, not buildtime.
> 
> See drivers/hwtracing/coresight/coresight-platform.c:coresight_get_platform_data() for an example.

I will update so it can be detected in runtime.

>>  	int i = 0, n, cpu;
>>  	struct device_node *cpu_node;
>> +	struct dsu_pmu *dsu_pmu =
>> +		(struct dsu_pmu *) platform_get_drvdata(pdev);
>>  -	n = of_count_phandle_with_args(dev, "cpus", NULL);
>> +	n = of_count_phandle_with_args(pdev->dev.of_node, "cpus", NULL);
>>  	if (n <= 0)
>>  		return -ENODEV;
>>  	for (; i < n; i++) {
>> -		cpu_node = of_parse_phandle(dev, "cpus", i);
>> +		cpu_node = of_parse_phandle(pdev->dev.of_node, "cpus", i);
>>  		if (!cpu_node)
>>  			break;
>>  		cpu = of_cpu_node_to_id(cpu_node);
>> @@ -626,9 +631,33 @@ static int dsu_pmu_dt_get_cpus(struct device_node *dev, cpumask_t *mask)
>>  		 */
>>  		if (cpu < 0)
>>  			continue;
>> -		cpumask_set_cpu(cpu, mask);
>> +		cpumask_set_cpu(cpu, &dsu_pmu->associated_cpus);
>>  	}
>>  	return 0;
>> +#else /* CONFIG_ACPI */
>> +	int i, cpu, ret;
>> +	const union acpi_object *obj;
>> +	struct acpi_device *adev = ACPI_COMPANION(&pdev->dev);
>> +	struct dsu_pmu *dsu_pmu =
>> +		(struct dsu_pmu *) platform_get_drvdata(pdev);
>> +
> 
>> +	ret = acpi_dev_get_property(adev, "cpus", ACPI_TYPE_ANY, &obj);
> 
> Is the binding documented somewhere ?

=> Will add one.

> 
> 
> nit: Also, why not :
> 	ret = acpi_dev_get_propert(adev, "cpus", ACPI_TYPE_PACKAGE, &obj);
> 	if (ret < 0)
> 		return ret;
> ?
> 
> 
>> +	if (ret < 0)
>> +		return -EINVAL;
>> +
>> +	if (obj->type != ACPI_TYPE_PACKAGE)
>> +		return -EINVAL;
>> +
>> +	for (i = 0; i < obj->package.count; i++) {
> 
> 
>> +		/* Each element is the MPIDR of associated cpu */
>> +		for_each_possible_cpu(cpu) {
>> +			if (cpu_physical_id(cpu) ==
>> +				obj->package.elements[i].integer.value)
>> +				cpumask_set_cpu(cpu, &dsu_pmu->associated_cpus);
>> +		}
>> +	}
>> +	return 0;
>> +#endif
>>  }
>>  
> 
> Otherwise looks good to me.
> 
> Suzuki