[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <3e08d48b-7b70-cc7f-0ec1-12ad9b1a33db@redhat.com>
Date: Thu, 24 Aug 2023 16:05:31 +0200
From: David Hildenbrand <david@...hat.com>
To: Zach O'Keefe <zokeefe@...gle.com>
Cc: linux-mm@...ck.org, Yang Shi <shy828301@...il.com>,
linux-kernel@...r.kernel.org,
Saurabh Singh Sengar <ssengar@...rosoft.com>,
Greg KH <gregkh@...uxfoundation.org>
Subject: Re: [PATCH v3] mm/thp: fix "mm: thp: kill __transhuge_page_enabled()"
On 24.08.23 15:59, Zach O'Keefe wrote:
> On Thu, Aug 24, 2023 at 12:39 AM David Hildenbrand <david@...hat.com> wrote:
>>
>> On 22.08.23 01:48, Zach O'Keefe wrote:
>>> The 6.0 commits:
>>>
>>> commit 9fec51689ff6 ("mm: thp: kill transparent_hugepage_active()")
>>> commit 7da4e2cb8b1f ("mm: thp: kill __transhuge_page_enabled()")
>>>
>>> merged "can we have THPs in this VMA?" logic that was previously done
>>> separately by fault-path, khugepaged, and smaps "THPeligible" checks.
>>>
>>> During the process, the semantics of the fault path check changed in two
>>> ways:
>>>
>>> 1) A VM_NO_KHUGEPAGED check was introduced (also added to smaps path).
>>> 2) We no longer checked if non-anonymous memory had a vm_ops->huge_fault
>>> handler that could satisfy the fault. Previously, this check had been
>>> done in create_huge_pud() and create_huge_pmd() routines, but after
>>> the changes, we never reach those routines.
>>>
>>> During the review of the above commits, it was determined that in-tree
>>> users weren't affected by the change; most notably, since the only relevant
>>> user (in terms of THP) of VM_MIXEDMAP or ->huge_fault is DAX, which is
>>> explicitly approved early in approval logic. However, there is at least
>>> one occurrence where an out-of-tree driver that used
>>> VM_HUGEPAGE|VM_MIXEDMAP with a vm_ops->huge_fault handler, was broken.
>>
>> ... so all we did is break an arbitrary out-of-tree driver? Sorry to
>> say, but why should we care?
>>
>> Is there any in-tree code affected and needs a "Fixes:" ?
>
> The in-tree code was taken care of during the rework .. but I didn't
> know about the possibility of a driver hooking in here.
And that's the problem of the driver, no? It's not the job of the kernel
developers to be aware of each out-of-tree driver to not accidentally
break something in there.
>
> I don't know what the normal policy / stance here is, but I figured
> the change was simple enough that it was worth helping out.
If you decide to be out-of-tree, then you have be prepared to only
support tested kernels and fix your driver when something changes
upstream -- like upstream developers would do for you when it would be
in-tree.
So why can't the out-of-tree driver be fixed, similarly to how we would
have fixed it if it would be in-tree?
--
Cheers,
David / dhildenb
Powered by blists - more mailing lists