Re: [PATCH V0 1/2] Drivers: hv: Add new function hv_isolation_hw_cvm()

[Mukesh R <mrathor@linux.microsoft.com>]

On 11/11/26 10:09, Michael Kelley wrote:
> From: Mukesh R <mrathor@linux.microsoft.com> Sent: Tuesday, November 10, 2026 6:21 PM
>>
>> Add a new function hv_isolation_hw_cvm() for convenient check
>> if a VM is running in HW confidential mode, ie, TDX or SNP.
>>
>> Signed-off-by: Mukesh R <mrathor@linux.microsoft.com>
>> ---
>>   arch/x86/hyperv/ivm.c          | 5 +++++
>>   drivers/hv/hv_common.c         | 9 +++++++--
>>   include/asm-generic/mshyperv.h | 7 +++++++
>>   3 files changed, 19 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/x86/hyperv/ivm.c b/arch/x86/hyperv/ivm.c
>> index 2ce4dfe53472..b3340be0c43a 100644
>> --- a/arch/x86/hyperv/ivm.c
>> +++ b/arch/x86/hyperv/ivm.c
>> @@ -944,3 +944,8 @@ bool hv_isolation_type_tdx(void)
>>   {
>>   	return static_branch_unlikely(&isolation_type_tdx);
>>   }
>> +
>> +bool hv_isolation_hw_cvm(void)
>> +{
>> +	return hv_isolation_type_snp() || hv_isolation_type_tdx();
>> +}
>> diff --git a/drivers/hv/hv_common.c b/drivers/hv/hv_common.c
>> index 27b9b79d4f4c..62b52855ae69 100644
>> --- a/drivers/hv/hv_common.c
>> +++ b/drivers/hv/hv_common.c
>> @@ -528,8 +528,7 @@ int hv_common_cpu_init(unsigned int cpu)
>>   			*outputarg = (char *)mem + HV_HYP_PAGE_SIZE;
>>   		}
>>
>> -		if (!ms_hyperv.paravisor_present &&
>> -		    (hv_isolation_type_snp() || hv_isolation_type_tdx())) {
>> +		if (!ms_hyperv.paravisor_present && hv_isolation_hw_cvm()) {
> 
> Any reason not to use hv_is_isolation_supported()? Your
> new function seems conceptually equivalent, though at a
> detailed level it handles VBS isolation differently from
> hv_is_isolation_supported(). But I don't think including
> VBS isolation here is a problem since set_memory_decrypted()
> is a no-op with VBS isolation.
> 
> Also, Kameron Carr has a patch set in flight that collides
> on exactly this code.  See [1] for the patch and my comments.
> 
> Michael

Hi Michael,

The issues my patch addresses only apply to hardware isolations, like
TDX and SNP, hence I called the function: hv_isolation_hw_cvm()

IOW, hv_is_isolation_supported() cannot be used in my case.

If there is conflict, I guess, one of us will have to rebase depending
on who merges first.

Thanks,
-Mukesh


> [1] https://lore.kernel.org/linux-hyperv/MN2PR02MB6863AA6357C8E0D4881F5FEAD4712@MN2PR02MB6863.namprd02.prod.outlook.com/
> 
> 
>>   			ret = set_memory_decrypted((unsigned long)mem, pgcount);
>>   			if (ret) {
>>   				/* It may be unsafe to free 'mem' */
>> @@ -696,6 +695,12 @@ bool __weak hv_isolation_type_tdx(void)
>>   }
>>   EXPORT_SYMBOL_GPL(hv_isolation_type_tdx);
>>
>> +bool __weak hv_isolation_hw_cvm(void)
>> +{
>> +	return false;
>> +}
>> +EXPORT_SYMBOL_GPL(hv_isolation_hw_cvm);
>> +
>>   void __weak hv_setup_vmbus_handler(void (*handler)(void))
>>   {
>>   }
>> diff --git a/include/asm-generic/mshyperv.h b/include/asm-generic/mshyperv.h
>> index a3d497d99b0b..440c599784eb 100644
>> --- a/include/asm-generic/mshyperv.h
>> +++ b/include/asm-generic/mshyperv.h
>> @@ -87,6 +87,13 @@ u64 hv_do_fast_hypercall16(u16 control, u64 input1, u64
>> input2);
>>   bool hv_isolation_type_snp(void);
>>   bool hv_isolation_type_tdx(void);
>>
>> +/* This needs to be before hv_do_rep_hypercall() below */
>> +#if IS_ENABLED(CONFIG_HYPERV)
>> +bool hv_isolation_hw_cvm(void);
>> +#else
>> +static inline bool hv_isolation_hw_cvm(void) { return false; }
>> +#endif
>> +
>>   /*
>>    * On architectures where Hyper-V doesn't support AEOI (e.g., ARM64),
>>    * it doesn't provide a recommendation flag and AEOI must be disabled.
>> --
>> 2.51.2.vfs.0.1
>>