Re: [RESEND PATCH v2 1/3] x86/sev: Add new quiet parameter to snp_leak_pages() API

[Kalra, Ashish]

Hello Sean,

On 8/18/2025 4:14 PM, Sean Christopherson wrote:
> On Mon, Aug 18, 2025, Ashish Kalra wrote:
>> diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
>> index 2fbdebf79fbb..a7db96a5f56d 100644
>> --- a/arch/x86/kvm/svm/sev.c
>> +++ b/arch/x86/kvm/svm/sev.c
>> @@ -271,7 +271,7 @@ static void sev_decommission(unsigned int handle)
>>  static int kvm_rmp_make_shared(struct kvm *kvm, u64 pfn, enum pg_level level)
>>  {
>>  	if (KVM_BUG_ON(rmp_make_shared(pfn, level), kvm)) {
>> -		snp_leak_pages(pfn, page_level_size(level) >> PAGE_SHIFT);
>> +		snp_leak_pages(pfn, page_level_size(level) >> PAGE_SHIFT, false);
>>  		return -EIO;
>>  	}
>>  
>> @@ -300,7 +300,7 @@ static int snp_page_reclaim(struct kvm *kvm, u64 pfn)
>>  	data.paddr = __sme_set(pfn << PAGE_SHIFT);
>>  	rc = sev_do_cmd(SEV_CMD_SNP_PAGE_RECLAIM, &data, &fw_err);
>>  	if (KVM_BUG(rc, kvm, "Failed to reclaim PFN %llx, rc %d fw_err %d", pfn, rc, fw_err)) {
>> -		snp_leak_pages(pfn, 1);
>> +		snp_leak_pages(pfn, 1, false);
> 
> Open coded true/false literals are ugly, e.g. now I have to go look at the
> declaration (or even definition) of snp_leak_pages() to understand what %false
> controls.
> 
> Assuming "don't dump the RMP entry" is the rare case, then craft the APIs to
> reflect that, i.e. make snp_leak_pages() a wrapper for the common case.  As a
> bonus, you don't need to churn any extra code either.
> 
> void __snp_leak_pages(u64 pfn, unsigned int npages, bool dump_rmp);
> 
> static inline void snp_leak_pages(u64 pfn, unsigned int npages)
> {
> 	__snp_leak_pages(pfn, npages, true);
> }
> 
>>  		return -EIO;
>>  	}
>>  
>> diff --git a/arch/x86/virt/svm/sev.c b/arch/x86/virt/svm/sev.c
>> index 942372e69b4d..d75659859a07 100644
>> --- a/arch/x86/virt/svm/sev.c
>> +++ b/arch/x86/virt/svm/sev.c
>> @@ -1029,7 +1029,7 @@ int rmp_make_shared(u64 pfn, enum pg_level level)
>>  }
>>  EXPORT_SYMBOL_GPL(rmp_make_shared);
>>  
>> -void snp_leak_pages(u64 pfn, unsigned int npages)
>> +void snp_leak_pages(u64 pfn, unsigned int npages, bool quiet)
>>  {
>>  	struct page *page = pfn_to_page(pfn);
>>  
>> @@ -1052,7 +1052,8 @@ void snp_leak_pages(u64 pfn, unsigned int npages)
>>  		    (PageHead(page) && compound_nr(page) <= npages))
>>  			list_add_tail(&page->buddy_list, &snp_leaked_pages_list);
>>  
>> -		dump_rmpentry(pfn);
>> +		if (!quiet)
> 
> The polarity is arbitrarily odd, and "quiet" is annoyingly ambiguous and arguably
> misleading, e.g. one could expect "quiet=true" to suppress the pr_warn() too, but
> it does not.
> 
> 	pr_warn("Leaking PFN range 0x%llx-0x%llx\n", pfn, pfn + npages)
> 
> If you call it "bool dump_rmp" then it's more precise, self-explanatory, and
> doesn't need to be inverted.

Thanks, i will re-work this accordingly.

Ashish

> 
>> +			dump_rmpentry(pfn);
>>  		snp_nr_leaked_pages++;
>>  		pfn++;
>>  		page++;
>> diff --git a/drivers/crypto/ccp/sev-dev.c b/drivers/crypto/ccp/sev-dev.c
>> index 4f000dc2e639..203a43a2df63 100644
>> --- a/drivers/crypto/ccp/sev-dev.c
>> +++ b/drivers/crypto/ccp/sev-dev.c
>> @@ -408,7 +408,7 @@ static int snp_reclaim_pages(unsigned long paddr, unsigned int npages, bool lock
>>  	 * If there was a failure reclaiming the page then it is no longer safe
>>  	 * to release it back to the system; leak it instead.
>>  	 */
>> -	snp_leak_pages(__phys_to_pfn(paddr), npages - i);
>> +	snp_leak_pages(__phys_to_pfn(paddr), npages - i, false);
>>  	return ret;
>>  }
>>  
>> -- 
>> 2.34.1
>>