[PATCH hyperv-next v2 3/4] arch: hyperv: Get/set SynIC synth.registers via paravisor

[Roman Kisel]

On 5/12/2025 2:39 AM, ALOK TIWARI wrote:
> 
> 
> On 12-05-2025 04:37, Roman Kisel wrote:
>> +/*
>> + * Not every paravisor supports getting SynIC registers, and
>> + * this function may fail. The caller has to make sure that this 
>> function
>> + * runs on the CPU of interest.
>> + */
> 
> Title and Intent: Clearly state the purpose of the function in the first 
> sentence
> /*
>   * Attempt to get the SynIC register value.
>   *
>   * Not all paravisors support reading SynIC registers, so this function
>   * may fail. The caller must ensure that it is executed on the target
>   * CPU.
>   *
>   * Returns: The SynIC register value or ~0ULL on failure.
>   * Sets err to -ENODEV if the provided register is not a valid SynIC
>   * MSR.
>   */
> 
>> +u64 hv_pv_get_synic_register(unsigned int reg, int *err)
>> +{
>> +    if (!hv_is_synic_msr(reg)) {
>> +        *err = -ENODEV;
>> +        return !0ULL;
>> +    }
>> +    return native_read_msr_safe(reg, err);
>> +}
>> +EXPORT_SYMBOL_GPL(hv_pv_get_synic_register);
>> +
>> +/*
>> + * Not every paravisor supports setting SynIC registers, and
>> + * this function may fail. The caller has to make sure that this 
>> function
>> + * runs on the CPU of interest.
>> + */
> 
> ditto.
> 
>> +int hv_pv_set_synic_register(unsigned int reg, u64 val)
>> +{
>> +    if (!hv_is_synic_msr(reg))
>> +        return -ENODEV;
>> +    return wrmsrl_safe(reg, val);
>> +}
>> +EXPORT_SYMBOL_GPL(hv_pv_set_synic_register);
> 

Indeed, I wrote a bit of a novel in the comments which might be
distracting and making it harder to find the point :)

Ought to be more conscious of the reader's perhaps constrained
time budget. I'll restructure that as you suggested!

> Thanks,
> Alok
> 

-- 
Thank you,
Roman