Re: [PATCH V3 1/4] arm_pmu: acpi: Refactor arm_spe_acpi_register_device()

[Anshuman Khandual]

On 8/4/23 22:09, Will Deacon wrote:
> On Thu, Aug 03, 2023 at 11:43:27AM +0530, Anshuman Khandual wrote:
>>
>>
>> On 8/3/23 11:26, Anshuman Khandual wrote:
>>> +	/*
>>> +	 * Sanity check all the GICC tables for the same interrupt
>>> +	 * number. For now, only support homogeneous ACPI machines.
>>> +	 */
>>> +	for_each_possible_cpu(cpu) {
>>> +		struct acpi_madt_generic_interrupt *gicc;
>>> +
>>> +		gicc = acpi_cpu_get_madt_gicc(cpu);
>>> +		if (gicc->header.length < len)
>>> +			return gsi ? -ENXIO : 0;
>>> +
>>> +		this_gsi = parse_gsi(gicc);
>>> +		if (!this_gsi)
>>> +			return gsi ? -ENXIO : 0;
>>
>> Hello Will,
>>
>> Moved parse_gsi() return code checking to its original place just to
>> make it similar in semantics to existing 'gicc->header.length check'.
>> If 'gsi' is valid i.e atleast a single cpu has been probed, return
>> -ENXIO indicating mismatch, otherwise just return 0.
> 
> Wouldn't that still be the case without the check in this hunk? We'd run
> into the homogeneous check and return -ENXIO from there, no?
Although the return code will be the same i.e -ENXIO, but not for the same reason.

		this_gsi = parse_gsi(gicc);
		if (!this_gsi)
			return gsi ? -ENXIO : 0;

This returns 0 when IRQ could not be parsed for the first cpu, but returns -ENXIO
for subsequent cpus. Although return code -ENXIO here still indicates IRQ parsing
to have failed.

		} else if (hetid != this_hetid || gsi != this_gsi) {
			pr_warn("ACPI: %s: must be homogeneous\n", pdev->name);
			return -ENXIO;
		} 

This returns -ENXIO when there is a IRQ mismatch. But if the above check is not
there, -ENXIO return code here could not be classified into IRQ parse problem or
mismatch without looking into the IRQ value.