Re: [PATCH v1 0/7] mm: kasan: skip KASAN for GFP_HIGHUSER
From: David Hildenbrand (Arm)
Date: Wed Nov 18 2026 - 04:46:53 EST
On 11/18/26 02:28, Jeff Xu wrote:
> On Tue, Nov 17, 2026 at 11:11 PM Tomasz Figa <tfiga@xxxxxxxxxxxx> wrote:
>>
>> On Tue, Nov 17, 2026 at 7:10 PM David Hildenbrand (Arm)
>> <david@xxxxxxxxxx> wrote:
>>>
>>> I really don't want us to spread KASAN-related GFP flags in the kernel.
>>>
>>> There must be a proper way to abstract this.
>>>
>>> Why is GFP_HIGHUSER not sufficient? We were already told that this is memory to
>>> be mapped into user space. Placing data structure on it and handing out pointers
>>> to that would already be a bit questionable.
>>>
>>> So what are the ways to avoid GFP_HIGHUSER_KASAN?
>>
>> I think the idea behind GFP_HIGHUSER_KASAN is to make it possible to
>> skip KASAN for GFP_HIGHUSER by default (for most of its users), but
>> keep it for some of the users which today use GFP_HIGHUSER for their
>> memory allocations, but could benefit from the added security. As per
>> the other patches in the series:
>>
>> pipe, fuse: use GFP_HIGHUSER_KASAN for pipe buffer pages
>> kexec: use GFP_HIGHUSER_KASAN for segment pages
>> net/rds: use GFP_HIGHUSER_KASAN for message pages
>>
>> That said, I'm not familiar with any of the code mentioned by those
>> patches, so I can't really tell how much they would benefit from KASAN
>> or if their choice of GFP_HIGHUSER to begin with is correct.
>
> Thanks David and Tomasz for the feedback.
>
> Tomasz is right about why I added GFP_HIGHUSER_KASAN in v1: out of
> caution, I audited existing GFP_HIGHUSER callers to find any that never
> map the pages to userspace and instead fill/drain them in the kernel via
> kmap_local_page() (pipe, FUSE, kexec, rds). Since this series intends to
> improve KASAN performance while preserving its security aspects, and the
> performance impact on pipe/FUSE/kexec/rds was negligible in my testing,
> v1 was optimized for minimum behavioral change.
>
> Looking closer at both of your questions, I agree that the security
> benefit of keeping those 4 under KASAN is minimal -- e.g., only linear
> overflows out of the pipe/FUSE/kexec/rds page, since none of them store
> kernel structures or pointers. That is the exact same trade-off we
> already make for GFP_HIGHUSER_MOVABLE.
Yes, that's my understanding as well.
>
> To avoid introducing GFP_HIGHUSER_KASAN, I see two options:
>
> 1) Simply keep those 4 cases under GFP_HIGHUSER (drop patches 1-4 and
> 7), reducing the series to just the GFP_HIGHUSER change and its KUnit
> test.
That's what I would look into first.
>
> 2) Switch those 4 callers to GFP_USER. Since CONFIG_KASAN_HW_TAGS is
> 64-bit arm64 only (!CONFIG_HIGHMEM), GFP_USER is identical to today's
> GFP_HIGHUSER on arm64 and all other 64-bit builds (and if 32-bit
> CONFIG_HIGHMEM=y builds still need ZONE_HIGHMEM, that could be gated
> on CONFIG_HIGHMEM, though that feels ugly at call sites).
>
> (1) seems much cleaner to me. David, would (1) work for you?
Agreed, (2) looks more hacky and I don't see the value of doing it like that
just yet.
--
Cheers,
David