linux-next: manual merge of the akpm-current tree with the vhost tree

linux-next: manual merge of the akpm-current tree with the vhost tree

[Stephen Rothwell]

Hi Andrew,

Today's linux-next merge of the akpm-current tree got a conflict in:

  mm/balloon_compaction.c

between commit:

  0b5ffdb4f7c6 ("balloon: fix page list locking")

from the vhost tree and commit:

  0c983ec0c078 ("virtio_balloon: fix race between migration and ballooning")

from the akpm-current tree.

I fixed it up (I think that the former is a superset of the latter, so
I used that, completely reversing the latter patch) and can carry the
fix as necessary (no action is required).

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Michael S. Tsirkin]

On Fri, Jan 08, 2016 at 04:59:08PM +1100, Stephen Rothwell wrote:
> Hi Andrew,
> 
> Today's linux-next merge of the akpm-current tree got a conflict in:
> 
>   mm/balloon_compaction.c
> 
> between commit:
> 
>   0b5ffdb4f7c6 ("balloon: fix page list locking")
> 
> from the vhost tree and commit:
> 
>   0c983ec0c078 ("virtio_balloon: fix race between migration and ballooning")
> 
> from the akpm-current tree.
> 
> I fixed it up (I think that the former is a superset of the latter, so
> I used that, completely reversing the latter patch) and can carry the
> fix as necessary (no action is required).

Oops, it isn't.

Andrew, your tree seems to still carry original patch by Minchan Kim:

0c983ec0c078ed5d5c219cf97af06364e7a8e19f

I reported on list that it has a deadlock:
http://article.gmane.org/gmane.linux.kernel.stable/159208

I think it should be reverted and my patch applied - it's in my tree
but if you prefer it in yours instead, please let me know.

Thanks!

> -- 
> Cheers,
> Stephen Rothwell                    sfr@canb.auug.org.au

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Stephen Rothwell]

Hi Michael,

On Fri, 8 Jan 2016 08:46:15 +0200 "Michael S. Tsirkin" <mst@redhat.com> wrote:
>
> On Fri, Jan 08, 2016 at 04:59:08PM +1100, Stephen Rothwell wrote:
> > 
> > Today's linux-next merge of the akpm-current tree got a conflict in:
> > 
> >   mm/balloon_compaction.c
> > 
> > between commit:
> > 
> >   0b5ffdb4f7c6 ("balloon: fix page list locking")
> > 
> > from the vhost tree and commit:
> > 
> >   0c983ec0c078 ("virtio_balloon: fix race between migration and ballooning")
> > 
> > from the akpm-current tree.
> > 
> > I fixed it up (I think that the former is a superset of the latter, so
> > I used that, completely reversing the latter patch) and can carry the
> > fix as necessary (no action is required).  
> 
> Oops, it isn't.
> 
> Andrew, your tree seems to still carry original patch by Minchan Kim:
> 
> 0c983ec0c078ed5d5c219cf97af06364e7a8e19f
> 
> I reported on list that it has a deadlock:
> http://article.gmane.org/gmane.linux.kernel.stable/159208
> 
> I think it should be reverted and my patch applied - it's in my tree
> but if you prefer it in yours instead, please let me know.

That is what I effectively did in today's tree.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
\

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Michael S. Tsirkin]

On Fri, Jan 08, 2016 at 05:50:11PM +1100, Stephen Rothwell wrote:
> Hi Michael,
> 
> On Fri, 8 Jan 2016 08:46:15 +0200 "Michael S. Tsirkin" <mst@redhat.com> wrote:
> >
> > On Fri, Jan 08, 2016 at 04:59:08PM +1100, Stephen Rothwell wrote:
> > > 
> > > Today's linux-next merge of the akpm-current tree got a conflict in:
> > > 
> > >   mm/balloon_compaction.c
> > > 
> > > between commit:
> > > 
> > >   0b5ffdb4f7c6 ("balloon: fix page list locking")
> > > 
> > > from the vhost tree and commit:
> > > 
> > >   0c983ec0c078 ("virtio_balloon: fix race between migration and ballooning")
> > > 
> > > from the akpm-current tree.
> > > 
> > > I fixed it up (I think that the former is a superset of the latter, so
> > > I used that, completely reversing the latter patch) and can carry the
> > > fix as necessary (no action is required).  
> > 
> > Oops, it isn't.
> > 
> > Andrew, your tree seems to still carry original patch by Minchan Kim:
> > 
> > 0c983ec0c078ed5d5c219cf97af06364e7a8e19f
> > 
> > I reported on list that it has a deadlock:
> > http://article.gmane.org/gmane.linux.kernel.stable/159208
> > 
> > I think it should be reverted and my patch applied - it's in my tree
> > but if you prefer it in yours instead, please let me know.
> 
> That is what I effectively did in today's tree.

Ouch, looks like I can't read. Somehow I misread that and assumed the
reverse, fetched next-20160107 and saw the context there is wrong, but I
did not dig deeper.  It seems that I was just hasty, and todays tree
isn't tagged yet.

Thanks!

> -- 
> Cheers,
> Stephen Rothwell                    sfr@canb.auug.org.au
> \

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Stephen Rothwell]

Hi Michael,

On Fri, 8 Jan 2016 09:05:57 +0200 "Michael S. Tsirkin" <mst@redhat.com> wrote:
>
> Ouch, looks like I can't read. Somehow I misread that and assumed the
> reverse, fetched next-20160107 and saw the context there is wrong, but I
> did not dig deeper.  It seems that I was just hasty, and todays tree
> isn't tagged yet.

Should be there now (barring mirroring delays).
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Minchan Kim]

On Fri, Jan 08, 2016 at 08:46:15AM +0200, Michael S. Tsirkin wrote:
> On Fri, Jan 08, 2016 at 04:59:08PM +1100, Stephen Rothwell wrote:
> > Hi Andrew,
> > 
> > Today's linux-next merge of the akpm-current tree got a conflict in:
> > 
> >   mm/balloon_compaction.c
> > 
> > between commit:
> > 
> >   0b5ffdb4f7c6 ("balloon: fix page list locking")
> > 
> > from the vhost tree and commit:
> > 
> >   0c983ec0c078 ("virtio_balloon: fix race between migration and ballooning")
> > 
> > from the akpm-current tree.
> > 
> > I fixed it up (I think that the former is a superset of the latter, so
> > I used that, completely reversing the latter patch) and can carry the
> > fix as necessary (no action is required).
> 
> Oops, it isn't.
> 
> Andrew, your tree seems to still carry original patch by Minchan Kim:
> 
> 0c983ec0c078ed5d5c219cf97af06364e7a8e19f
> 
> I reported on list that it has a deadlock:
> http://article.gmane.org/gmane.linux.kernel.stable/159208

And I replied on your comment that it's not a deadlock but it seem
you lost my mail because you didn't reply to me.
Could you expalin why it's deadlock in my reply thread?

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Michael S. Tsirkin]

On Fri, Jan 08, 2016 at 04:17:06PM +0900, Minchan Kim wrote:
> On Fri, Jan 08, 2016 at 08:46:15AM +0200, Michael S. Tsirkin wrote:
> > On Fri, Jan 08, 2016 at 04:59:08PM +1100, Stephen Rothwell wrote:
> > > Hi Andrew,
> > > 
> > > Today's linux-next merge of the akpm-current tree got a conflict in:
> > > 
> > >   mm/balloon_compaction.c
> > > 
> > > between commit:
> > > 
> > >   0b5ffdb4f7c6 ("balloon: fix page list locking")
> > > 
> > > from the vhost tree and commit:
> > > 
> > >   0c983ec0c078 ("virtio_balloon: fix race between migration and ballooning")
> > > 
> > > from the akpm-current tree.
> > > 
> > > I fixed it up (I think that the former is a superset of the latter, so
> > > I used that, completely reversing the latter patch) and can carry the
> > > fix as necessary (no action is required).
> > 
> > Oops, it isn't.
> > 
> > Andrew, your tree seems to still carry original patch by Minchan Kim:
> > 
> > 0c983ec0c078ed5d5c219cf97af06364e7a8e19f
> > 
> > I reported on list that it has a deadlock:
> > http://article.gmane.org/gmane.linux.kernel.stable/159208
> 
> And I replied on your comment that it's not a deadlock but it seem
> you lost my mail because you didn't reply to me.
> Could you expalin why it's deadlock in my reply thread?

I missed your mail, sorry. I don't remember the details.  I'll think
about it some more and reply next week in any case.

Thanks!

-- 
MST

linux-next: manual merge of the akpm-current tree with the vhost tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 695 bytes --]

Hi all,

Today's linux-next merge of the akpm-current tree got a conflict in:

  drivers/virtio/virtio_balloon.c

between commit:

  b5769cdc14fc ("virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")

from the vhost tree and commits:

  b64c4d5bea98 ("virtio-balloon: pull page poisoning config out of free page hinting")
  80c03575431c ("virtio-balloon: add support for providing free page reports to host")

from the akpm-current tree.

I looked at the conflict for a while but could not easily see how to
reconcile it, so I decided to revert the vhost tree commit for today.
Some advice would be appreciated.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[David Hildenbrand]

On 10.03.20 09:02, Stephen Rothwell wrote:
> Hi all,
> 
> Today's linux-next merge of the akpm-current tree got a conflict in:
> 
>   drivers/virtio/virtio_balloon.c
> 
> between commit:
> 
>   b5769cdc14fc ("virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")
> 
> from the vhost tree and commits:
> 
>   b64c4d5bea98 ("virtio-balloon: pull page poisoning config out of free page hinting")
>   80c03575431c ("virtio-balloon: add support for providing free page reports to host")
> 
> from the akpm-current tree.
> 
> I looked at the conflict for a while but could not easily see how to
> reconcile it, so I decided to revert the vhost tree commit for today.
> Some advice would be appreciated.
> 

Yes, the free page reporting features are currently in Andrews tree and
most probably won't go via the vhost tree due to the core-mm changes.
Ideally, the VIRTIO_BALLOON_F_DEFLATE_ON_OOM fix would go in unchanged,
because some people might be interested in backporting it (it's not a
stable fix, though).

I think rebasing any way round shouldn't be too hard.

@Alex, Michael, what's your thought on this?

-- 
Thanks,

David / dhildenb

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Michael S. Tsirkin]

On Tue, Mar 10, 2020 at 09:39:21AM +0100, David Hildenbrand wrote:
> On 10.03.20 09:02, Stephen Rothwell wrote:
> > Hi all,
> > 
> > Today's linux-next merge of the akpm-current tree got a conflict in:
> > 
> >   drivers/virtio/virtio_balloon.c
> > 
> > between commit:
> > 
> >   b5769cdc14fc ("virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")
> > 
> > from the vhost tree and commits:
> > 
> >   b64c4d5bea98 ("virtio-balloon: pull page poisoning config out of free page hinting")
> >   80c03575431c ("virtio-balloon: add support for providing free page reports to host")
> > 
> > from the akpm-current tree.
> > 
> > I looked at the conflict for a while but could not easily see how to
> > reconcile it, so I decided to revert the vhost tree commit for today.
> > Some advice would be appreciated.
> > 
> 
> Yes, the free page reporting features are currently in Andrews tree and
> most probably won't go via the vhost tree due to the core-mm changes.
> Ideally, the VIRTIO_BALLOON_F_DEFLATE_ON_OOM fix would go in unchanged,
> because some people might be interested in backporting it (it's not a
> stable fix, though).
> 
> I think rebasing any way round shouldn't be too hard.
> 
> @Alex, Michael, what's your thought on this?

I propose rebasing on top of the akpm tree and merging through there.
David could you do the rebase pls? I'll review and ack.

While there, keeping virtio_balloon_unregister_shrinker and
virtio_balloon_register_shrinker around might be a good idea to
minimize backporting pain.

> -- 
> Thanks,
> 
> David / dhildenb

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[David Hildenbrand]

On 10.03.20 10:04, Michael S. Tsirkin wrote:
> On Tue, Mar 10, 2020 at 09:39:21AM +0100, David Hildenbrand wrote:
>> On 10.03.20 09:02, Stephen Rothwell wrote:
>>> Hi all,
>>>
>>> Today's linux-next merge of the akpm-current tree got a conflict in:
>>>
>>>   drivers/virtio/virtio_balloon.c
>>>
>>> between commit:
>>>
>>>   b5769cdc14fc ("virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")
>>>
>>> from the vhost tree and commits:
>>>
>>>   b64c4d5bea98 ("virtio-balloon: pull page poisoning config out of free page hinting")
>>>   80c03575431c ("virtio-balloon: add support for providing free page reports to host")
>>>
>>> from the akpm-current tree.
>>>
>>> I looked at the conflict for a while but could not easily see how to
>>> reconcile it, so I decided to revert the vhost tree commit for today.
>>> Some advice would be appreciated.
>>>
>>
>> Yes, the free page reporting features are currently in Andrews tree and
>> most probably won't go via the vhost tree due to the core-mm changes.
>> Ideally, the VIRTIO_BALLOON_F_DEFLATE_ON_OOM fix would go in unchanged,
>> because some people might be interested in backporting it (it's not a
>> stable fix, though).
>>
>> I think rebasing any way round shouldn't be too hard.
>>
>> @Alex, Michael, what's your thought on this?
> 
> I propose rebasing on top of the akpm tree and merging through there.
> David could you do the rebase pls? I'll review and ack.
> 
> While there, keeping virtio_balloon_unregister_shrinker and
> virtio_balloon_register_shrinker around might be a good idea to
> minimize backporting pain.

Will do.

-- 
Thanks,

David / dhildenb

linux-next: manual merge of the akpm-current tree with the vhost tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 814 bytes --]

Hi all,

Today's linux-next merge of the akpm-current tree got a conflict in:

  drivers/virtio/virtio_balloon.c

between commit:

  5a6b4cc5b7a1 ("virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")

from the vhost tree and commits:

  5193acb63eef ("virtio-balloon: pull page poisoning config out of free page hinting")o
  226d0484a676 ("virtio-balloon: add support for providing free page reports to host")
  49006aae9e94 ("virtio-balloon: switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")

from the akpm-current tree.

OK, this is such a mess that all I could do was to revert commit
5a6b4cc5b7a1 from the vhost tree and keep all the akpm-current tree
patches. Please sort this out before Linus gets to see it.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 978 bytes --]

Hi all,

On Mon, 30 Mar 2020 19:31:37 +1100 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>
> Today's linux-next merge of the akpm-current tree got a conflict in:
> 
>   drivers/virtio/virtio_balloon.c
> 
> between commit:
> 
>   5a6b4cc5b7a1 ("virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")
> 
> from the vhost tree and commits:
> 
>   5193acb63eef ("virtio-balloon: pull page poisoning config out of free page hinting")o
>   226d0484a676 ("virtio-balloon: add support for providing free page reports to host")
>   49006aae9e94 ("virtio-balloon: switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")
> 
> from the akpm-current tree.
> 
> OK, this is such a mess that all I could do was to revert commit
> 5a6b4cc5b7a1 from the vhost tree and keep all the akpm-current tree
> patches. Please sort this out before Linus gets to see it.

I am still reverting that commit ...

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 484 bytes --]

Re: linux-next: manual merge of the akpm-current tree with the vhost tree

[Michael S. Tsirkin]

On Tue, Apr 07, 2020 at 01:25:21PM +1000, Stephen Rothwell wrote:
> Hi all,
> 
> On Mon, 30 Mar 2020 19:31:37 +1100 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> >
> > Today's linux-next merge of the akpm-current tree got a conflict in:
> > 
> >   drivers/virtio/virtio_balloon.c
> > 
> > between commit:
> > 
> >   5a6b4cc5b7a1 ("virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")
> > 
> > from the vhost tree and commits:
> > 
> >   5193acb63eef ("virtio-balloon: pull page poisoning config out of free page hinting")o
> >   226d0484a676 ("virtio-balloon: add support for providing free page reports to host")
> >   49006aae9e94 ("virtio-balloon: switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM")
> > 
> > from the akpm-current tree.
> > 
> > OK, this is such a mess that all I could do was to revert commit
> > 5a6b4cc5b7a1 from the vhost tree and keep all the akpm-current tree
> > patches. Please sort this out before Linus gets to see it.
> 
> I am still reverting that commit ...

Oh. Thanks for reminding me. And I sent a pull with this :(
Sorry.  I pushed the revert now and sent PULL request v2.
Hope I did it in time ...

> -- 
> Cheers,
> Stephen Rothwell