From: Alistair Popple <apopple@nvidia.com>
To: huang ying <huang.ying.caritas@gmail.com>
Cc: linux-mm@kvack.org, akpm@linux-foundation.org,
linux-kernel@vger.kernel.org, "Sierra Guiza,
Alejandro (Alex)" <alex.sierra@amd.com>,
Felix Kuehling <Felix.Kuehling@amd.com>,
Jason Gunthorpe <jgg@nvidia.com>,
John Hubbard <jhubbard@nvidia.com>,
David Hildenbrand <david@redhat.com>,
Ralph Campbell <rcampbell@nvidia.com>,
Matthew Wilcox <willy@infradead.org>,
Karol Herbst <kherbst@redhat.com>, Lyude Paul <lyude@redhat.com>,
Ben Skeggs <bskeggs@redhat.com>,
Logan Gunthorpe <logang@deltatee.com>,
linuxram@us.ibm.com, paulus@ozlabs.org,
Peter Xu <peterx@redhat.com>
Subject: Re: [PATCH 1/2] mm/migrate_device.c: Copy pte dirty bit to page
Date: Tue, 16 Aug 2022 12:28:56 +1000 [thread overview]
Message-ID: <87r11gvrx6.fsf@nvdebian.thelocal> (raw)
In-Reply-To: <CAC=cRTON4s6LJ+09mg4SxMjNirBih3QSt6cr10gNjs3k0o+BgQ@mail.gmail.com>
huang ying <huang.ying.caritas@gmail.com> writes:
> Hi, Alistair,
>
> On Fri, Aug 12, 2022 at 1:23 PM Alistair Popple <apopple@nvidia.com> wrote:
>>
>> migrate_vma_setup() has a fast path in migrate_vma_collect_pmd() that
>> installs migration entries directly if it can lock the migrating page.
>> When removing a dirty pte the dirty bit is supposed to be carried over
>> to the underlying page to prevent it being lost.
>>
>> Currently migrate_vma_*() can only be used for private anonymous
>> mappings. That means loss of the dirty bit usually doesn't result in
>> data loss because these pages are typically not file-backed. However
>> pages may be backed by swap storage which can result in data loss if an
>> attempt is made to migrate a dirty page that doesn't yet have the
>> PageDirty flag set.
>>
>> In this case migration will fail due to unexpected references but the
>> dirty pte bit will be lost. If the page is subsequently reclaimed data
>> won't be written back to swap storage as it is considered uptodate,
>> resulting in data loss if the page is subsequently accessed.
>>
>> Prevent this by copying the dirty bit to the page when removing the pte
>> to match what try_to_migrate_one() does.
>>
>> Signed-off-by: Alistair Popple <apopple@nvidia.com>
>> Reported-by: Peter Xu <peterx@redhat.com>
>> ---
>> mm/migrate_device.c | 5 +++++
>> 1 file changed, 5 insertions(+)
>>
>> diff --git a/mm/migrate_device.c b/mm/migrate_device.c
>> index 27fb37d..d38f8a6 100644
>> --- a/mm/migrate_device.c
>> +++ b/mm/migrate_device.c
>> @@ -7,6 +7,7 @@
>> #include <linux/export.h>
>> #include <linux/memremap.h>
>> #include <linux/migrate.h>
>> +#include <linux/mm.h>
>> #include <linux/mm_inline.h>
>> #include <linux/mmu_notifier.h>
>> #include <linux/oom.h>
>> @@ -211,6 +212,10 @@ static int migrate_vma_collect_pmd(pmd_t *pmdp,
>>
>> migrate->cpages++;
>>
>> + /* Set the dirty flag on the folio now the pte is gone. */
>> + if (pte_dirty(pte))
>> + folio_mark_dirty(page_folio(page));
>> +
>
> I think that this isn't sufficient to fix all issues. Firstly, "pte"
> is assigned at the begin of the loop, before the PTE is cleared via
> ptep_clear_flush() or ptep_get_and_clear(). That is, the pte isn't
> changed atomically. Between "pte" assignment and PTE clear, the PTE
> may become dirty. That is, we need to update pte when we clear the
> PTE.
Oh good catch, thanks. Will fix.
> And I don't know why we use ptep_get_and_clear() to clear PTE if
> (!anon_exclusive). Why don't we need to flush the TLB?
We do the TLB flush at the end if anything was modified:
/* Only flush the TLB if we actually modified any entries */
if (unmapped)
flush_tlb_range(walk->vma, start, end);
Obviously I don't think that will work correctly now given we have to
read the dirty bits and clear the PTE atomically. I assume it was
originally written this way for some sort of performance reason.
- Alistair
> Best Regards,
> Huang, Ying
>
>> /* Setup special migration page table entry */
>> if (mpfn & MIGRATE_PFN_WRITE)
>> entry = make_writable_migration_entry(
>>
>> base-commit: ffcf9c5700e49c0aee42dcba9a12ba21338e8136
>> --
>> git-series 0.9.1
>>
next prev parent reply other threads:[~2022-08-16 7:06 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-12 5:22 [PATCH 1/2] mm/migrate_device.c: Copy pte dirty bit to page Alistair Popple
2022-08-12 5:22 ` [PATCH 2/2] selftests/hmm-tests: Add test for dirty bits Alistair Popple
2022-08-12 7:58 ` Mika Penttilä
2022-08-15 2:35 ` Alistair Popple
2022-08-15 3:11 ` Mika Penttilä
2022-08-15 3:21 ` Alistair Popple
2022-08-15 4:05 ` Mika Penttilä
2022-08-15 4:06 ` Mika Penttilä
2022-08-15 20:29 ` [PATCH 1/2] mm/migrate_device.c: Copy pte dirty bit to page Peter Xu
2022-08-16 0:51 ` Alistair Popple
2022-08-16 1:39 ` huang ying
2022-08-16 2:28 ` Alistair Popple [this message]
2022-08-16 6:37 ` huang ying
2022-08-17 1:27 ` Alistair Popple
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87r11gvrx6.fsf@nvdebian.thelocal \
--to=apopple@nvidia.com \
--cc=Felix.Kuehling@amd.com \
--cc=akpm@linux-foundation.org \
--cc=alex.sierra@amd.com \
--cc=bskeggs@redhat.com \
--cc=david@redhat.com \
--cc=huang.ying.caritas@gmail.com \
--cc=jgg@nvidia.com \
--cc=jhubbard@nvidia.com \
--cc=kherbst@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linuxram@us.ibm.com \
--cc=logang@deltatee.com \
--cc=lyude@redhat.com \
--cc=paulus@ozlabs.org \
--cc=peterx@redhat.com \
--cc=rcampbell@nvidia.com \
--cc=willy@infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).