Re: [PATCH RFT 0/3] mm, drm: ensure .fault() does not have to be followed by .pfn_mkwrite() for write faults

"David Hildenbrand (Arm)" <[email protected]> Tue, 4 Aug 2026 14:19:28 +0200

Newsgroups	gmane.linux.kernel,gmane.comp.emulators.kvm.devel,gmane.comp.video.dri.devel,gmane.linux.kernel.mm
Message-ID	<[email protected]>

On 8/4/26 09:50, Paolo Bonzini wrote:
> On Mon, Aug 3, 2026 at 5:18 PM David Hildenbrand (Arm) <[email protected]> wrote:
>>>
>> I was wondering about the following:
>>
>> mprotect(PROT_READ)
>>
>> followed by
>>
>> mprotect(PROT_READ | PROT_WRITE)
>>
>> You'd similarly end up without the writable bit in the PTE, and apparently there is not really a way
>> to recover from this.
> 
> Why not? mprotect_fixup() calls vma_set_page_prot(), the PTE as you
> say lacks the writable bit (unless pte_dirty(pte)), and then the next
> fault calls .pfn_mkwrite().

Ah, if this works, great. I guess I was confused about your explanation about
fixup_user_fault().

So this really only about avoiding the second fault, makes sense thanks!

-- 
Cheers,

David