Message-ID: <ae53fa82-d8de-4c02-95f7-7650a03ea8e7@gmail.com>
Date: Tue, 20 May 2025 19:24:04 +0100
From: Usama Arif <usamaarif642@...il.com>
To: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>,
 David Hildenbrand <david@...hat.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
 "Liam R . Howlett" <Liam.Howlett@...cle.com>,
 Vlastimil Babka <vbabka@...e.cz>, Jann Horn <jannh@...gle.com>,
 Arnd Bergmann <arnd@...db.de>, Christian Brauner <brauner@...nel.org>,
 linux-mm@...ck.org, linux-arch@...r.kernel.org,
 linux-kernel@...r.kernel.org, SeongJae Park <sj@...nel.org>,
 Johannes Weiner <hannes@...xchg.org>, Shakeel Butt <shakeel.butt@...ux.dev>,
 Zi Yan <ziy@...dia.com>
Subject: Re: [RFC PATCH 0/5] add process_madvise() flags to modify behaviour



On 20/05/2025 18:47, Lorenzo Stoakes wrote:
> On Tue, May 20, 2025 at 05:28:35PM +0200, David Hildenbrand wrote:
>> On 19.05.25 22:52, Lorenzo Stoakes wrote:
>>> REVIEWERS NOTES:
>>> ================
>>>
>>> This is a VERY EARLY version of the idea, it's relatively untested, and I'm
>>> 'putting it out there' for feedback. Any serious version of this will add a
>>> bunch of self-tests to assert correct behaviour and I will more carefully
>>> confirm everything's working.
>>>
>>> This is based on discussion arising from Usama's series [0], SJ's input on
>>> the thread around process_madvise() behaviour [1] (and a subsequent
>>> response by me [2]) and prior discussion about a new madvise() interface
>>> [3].
>>>
>>> [0]: https://lore.kernel.org/linux-mm/20250515133519.2779639-1-usamaarif642@gmail.com/
>>> [1]: https://lore.kernel.org/linux-mm/20250517162048.36347-1-sj@kernel.org/
>>> [2]: https://lore.kernel.org/linux-mm/e3ba284c-3cb1-42c1-a0ba-9c59374d0541@lucifer.local/
>>> [3]: https://lore.kernel.org/linux-mm/c390dd7e-0770-4d29-bb0e-f410ff6678e3@lucifer.local/
>>>
>>> ================
>>>
>>> Currently, we are rather restricted in how madvise() operations
>>> proceed. While effort has been put in to expanding what process_madvise()
>>> can do (that is - unrestricted application of advice to the local process
>>> alongside recent improvements on the efficiency of TLB operations over
>>> these batvches), we are still constrained by existing madvise() limitations
>>> and default behaviours.
>>>
>>> This series makes use of the currently unused flags field in
>>> process_madvise() to provide more flexiblity.
>>>
>>
>> In general, sounds like an interesting approach.
> 
> Thanks!
> 
> If you agree this is workable, then I'll go ahead and put some more effort
> into writing tests etc. on the next respin.
> 

So the prctl and process_madvise patches both are trying to accomplish a
similar end goal.

Would it make sense to discuss what would be the best way forward before we
continue developing the solutions? If we are not at that stage and a clear
picture has not formed yet, happy to continue refining the solutions.
But just thought I would check. 

I feel like changing process_madvise which was designed to work on an array
of iovec structures to have flags to skip errors and ignore the iovec
makes it function similar to a prctl call is not the right approach.
IMHO, prctl is a more direct solution to this.

I know that Lonenzo doesn't like prctl and wants to unify this in process_madvise.
But if in the end, we want to have a THP auto way which is truly transparent,
would it not be better to just have this as prctl and not change the madvise
structure? Maybe in a few years we wont need any of this, and it will be lost
in prctl and madvise wouldn't have changed for this?

Again, this is just to have a discussion (and not an aggressive argument :)),
and would love to get feedback from everyone in the community.
If its too early to have this discussion, its completely fine and we can
still keep developing the RFCs :)

Thanks!
Usama

Powered by blists - more mailing lists