Re: [PATCH] KVM: arm64: ptdump: Flush the last region

[Dev Jain]

On 20/07/26 2:33 pm, Marc Zyngier wrote:
> On Mon, 20 Jul 2026 09:53:29 +0100,
> Dev Jain <dev.jain@xxxxxxx> wrote:
>>
>>
>>
>> On 18/07/26 4:42 am, Wei-Lin Chang wrote:
>>> Currently the stage-2 ptdump calls note_page() at each leaf entry visit.
>>> This simply misses the output of the last region, because note_page()
>>> only dumps output when it detects a change in level/prot, or when the
>>> walk enters a next marker section. The last region in the guest IPA
>>> space with the same level/prot is not dumped since there is no change
>>> after it.
>>>
>>> To dump the final region, manually issue a note_page() call with address
>>> BIT(ia_bits) (end of guest IPA space) and level == -1 to trigger a level
>>> change after the last region. Additionally treat level == -1 as a last
>>> flushing note_page() call and avoid dumping the name of the next marker
>>> for this case.
>>>
>>> Fixes: 7c4f73548ed1 ("KVM: arm64: Register ptdump with debugfs on guest creation")
>>> Reported-by: Sashiko AI <sashiko-bot@xxxxxxxxxx>
>>> Closes: https://lore.kernel.org/kvmarm/20260630122758.891011F00A3A@xxxxxxxxxxxxxxx/
>>> Signed-off-by: Wei-Lin Chang <weilin.chang@xxxxxxx>
>>> ---
>>>  arch/arm64/kvm/ptdump.c | 8 +++++---
>>>  arch/arm64/mm/ptdump.c  | 5 +++--
>>>  2 files changed, 8 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/arch/arm64/kvm/ptdump.c b/arch/arm64/kvm/ptdump.c
>>> index c9140e22abcf..4096e4a92fae 100644
>>> --- a/arch/arm64/kvm/ptdump.c
>>> +++ b/arch/arm64/kvm/ptdump.c
>>> @@ -155,11 +155,13 @@ static int kvm_ptdump_guest_show(struct seq_file *m, void *unused)
>>>  		.seq		= m,
>>>  	};
>>>  
>>> -	write_lock(&kvm->mmu_lock);
>>> +	guard(write_lock)(&kvm->mmu_lock);
>>
>> The guard clause change has nothing to do with the fix I think? Although the change
>> is small and the fix is for a recent commit so maybe we don't care having this
>> in the same patch - not sure how strict kvm reviewers are about this : )
>>
> 
> Look at the quality spaghetti code this would otherwise result in:
> 
> 	write_lock();
> 	ret = kvm_pgtable_walk();
> 	if (ret) {
> 		write_unlock();
> 		return ret;
> 	}
> 	note_page_flush();
> 	write_unlock();
> 
> 	return 0;
> 
> How is that better? So it has *everything* to do with the fix.
> 
> And FWIW, the policy for KVM is that we don't do lock->guard
> conversions as a separate patches. Only when we need to change the
> code, and as part of the patch that changes that code.  Which is
> exactly what is happening here.

I agree, thanks.


> 
>>
>>>  	ret = kvm_pgtable_walk(mmu->pgt, 0, BIT(mmu->pgt->ia_bits), &walker);
>>> -	write_unlock(&kvm->mmu_lock);
>>> +	if (ret)
>>> +		return ret;
>>> +	note_page(&st->parser_state.ptdump, BIT(mmu->pgt->ia_bits), -1, 0);
>>>  
>>> -	return ret;
>>> +	return 0;
>>>  }
>>>  
>>>  static int kvm_ptdump_guest_open(struct inode *m, struct file *file)
>>> diff --git a/arch/arm64/mm/ptdump.c b/arch/arm64/mm/ptdump.c
>>> index ab9899ca1e5f..fed4e4407e0e 100644
>>> --- a/arch/arm64/mm/ptdump.c
>>> +++ b/arch/arm64/mm/ptdump.c
>>> @@ -194,6 +194,7 @@ void note_page(struct ptdump_state *pt_st, unsigned long addr, int level,
>>>  	struct ptdump_pg_state *st = container_of(pt_st, struct ptdump_pg_state, ptdump);
>>>  	struct ptdump_pg_level *pg_level = st->pg_level;
>>>  	static const char units[] = "KMGTPE";
>>> +	bool flush = level == -1;
>>
>> Can we do something similar to what S1 ptdump does (note_page_flush in ptdump_walk_pgd).
> 
> See Mark's email.
> 
> 	M.
>