Re: [PATCH 4/6] mshv: Get the vmm capabilities offered by the hypervisor

[Nuno Das Neves <nunodasneves@linux.microsoft.com>]

On 9/5/2025 12:03 PM, Anirudh Rayabharam wrote:
> On Thu, Aug 28, 2025 at 05:43:48PM -0700, Nuno Das Neves wrote:
>> From: Purna Pavan Chandra Aekkaladevi <paekkaladevi@linux.microsoft.com>
>>
>> Some newer hypervisor APIs are gated by feature bits in the so-called
>> "vmm capabilities" partition property. Store the capabilities on
>> mshv_root module init, using HVCALL_GET_PARTITION_PROPERTY_EX.
>>
>> Signed-off-by: Purna Pavan Chandra Aekkaladevi <paekkaladevi@linux.microsoft.com>
>> Signed-off-by: Nuno Das Neves <nunodasneves@linux.microsoft.com>
>> ---
>>  drivers/hv/mshv_root.h      |  1 +
>>  drivers/hv/mshv_root_main.c | 28 ++++++++++++++++++++++++++++
>>  2 files changed, 29 insertions(+)
>>
>> diff --git a/drivers/hv/mshv_root.h b/drivers/hv/mshv_root.h
>> index 4aeb03bea6b6..0cb1e2589fe1 100644
>> --- a/drivers/hv/mshv_root.h
>> +++ b/drivers/hv/mshv_root.h
>> @@ -178,6 +178,7 @@ struct mshv_root {
>>  	struct hv_synic_pages __percpu *synic_pages;
>>  	spinlock_t pt_ht_lock;
>>  	DECLARE_HASHTABLE(pt_htable, MSHV_PARTITIONS_HASH_BITS);
>> +	struct hv_partition_property_vmm_capabilities vmm_caps;
>>  };
>>  
>>  /*
>> diff --git a/drivers/hv/mshv_root_main.c b/drivers/hv/mshv_root_main.c
>> index 56ababab57ce..29f61ecc9771 100644
>> --- a/drivers/hv/mshv_root_main.c
>> +++ b/drivers/hv/mshv_root_main.c
>> @@ -2327,6 +2327,28 @@ static int __init mshv_root_partition_init(struct device *dev)
>>  	return err;
>>  }
>>  
>> +static int mshv_init_vmm_caps(struct device *dev)
>> +{
>> +	int ret;
>> +
>> +	memset(&mshv_root.vmm_caps, 0, sizeof(mshv_root.vmm_caps));
>> +	ret = hv_call_get_partition_property_ex(HV_PARTITION_ID_SELF,
>> +						HV_PARTITION_PROPERTY_VMM_CAPABILITIES,
>> +						0, &mshv_root.vmm_caps,
>> +						sizeof(mshv_root.vmm_caps));
>> +
>> +	/*
>> +	 * HV_PARTITION_PROPERTY_VMM_CAPABILITIES is not supported in
>> +	 * older hyperv. Ignore the -EIO error code.
>> +	 */
>> +	if (ret && ret != -EIO)
>> +		return ret;
>> +
>> +	dev_dbg(dev, "vmm_caps=0x%llx\n", mshv_root.vmm_caps.as_uint64[0]);
>> +
>> +	return 0;
>> +}
>> +
>>  static int __init mshv_parent_partition_init(void)
>>  {
>>  	int ret;
>> @@ -2377,6 +2399,12 @@ static int __init mshv_parent_partition_init(void)
>>  	if (ret)
>>  		goto remove_cpu_state;
>>  
>> +	ret = mshv_init_vmm_caps(dev);
>> +	if (ret) {
>> +		dev_err(dev, "Failed to get VMM capabilities\n");
>> +		goto exit_partition;
> 
> Should this really be treated as a failure here? We could still offer
> /dev/mshv albeit with potentially limited capabilities.
> 

I agree. It can continue with vmm_caps set to zero, and print an error.

The check for -EIO is meant to achieve something similar but it's kind of
unreliable since so many HV_STATUS's map to EIO. So this idea is better,
thanks.

Nuno

> Thanks,
> Anirudh.
> 
>> +	}
>> +
>>  	ret = mshv_irqfd_wq_init();
>>  	if (ret)
>>  		goto exit_partition;
>> -- 
>> 2.34.1
>>