4.4's rdma plate

4.4's rdma plate

[Or Gerlitz]

Hi Doug,

As we're soon on 4.3-rc5, could you please update how things look for 4.4?

The mutlicast loopback prevention patches from Eran were intially
posted ~two months ago and went through testing by Christoph and Co.

The patches do have one new uverb entry (extended form for
ib_uverbs_create_qp) and other than that no real sensitive piece. If
you feel you need help to review that one (patch 1), I guess there
should be people on this list who can do that, and anyway, I don't
think it makes sense to delay a feature which is (1) small in volume
(2) critical to latency sensitive application for another kernel
release. The patch set also adds nice usage of the feature to the mlx4
EN driver (patches 2,3,4) and we don't want another kernel cycle to
get lost w.r.t this feature, so?

Haggai has posted the follow up patches to the demux series, which
adds name spaces support for the RDMA-CM [1] -- this is a strict
follow up which by itself went through a detailed review -- the break
up for a pre-series that does the demux refactoring didn't leave much
in the name space series, so I would say this should be ready to go
to, agree?

Matan has posted the next steps of his RoCE work, two patch series.
Please let us know where you see this standing too.

The SW-RoCE driver was posted to staging, V1 is fully operable, do you
see it as ready to be picked there?

Or.

[1] http://marc.info/?l=linux-rdma&m=144068579319679&w=2
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Or Gerlitz]

On Sun, Oct 11, 2015 at 11:58 PM, Or Gerlitz <gerlitz.or-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> Hi Doug,
> As we're soon on 4.3-rc5, could you please update how things look for 4.4?
>
> The mutlicast loopback prevention patches from Eran were intially
> posted ~two months ago and went through testing by Christoph and Co.
[...]

Hi Doug,

So we're after rc5 and not a single word from you re the plans for the
next kernel,
and your branch for that purpose contains exactly two patches... I do
see few chunks
of patches marked as under review in your patchworks, so there might
be sign for getting
some hope, is that real?

Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 982 bytes --]

On 10/14/2015 12:51 AM, Or Gerlitz wrote:
> On Sun, Oct 11, 2015 at 11:58 PM, Or Gerlitz <gerlitz.or-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
>> Hi Doug,
>> As we're soon on 4.3-rc5, could you please update how things look for 4.4?
>>
>> The mutlicast loopback prevention patches from Eran were intially
>> posted ~two months ago and went through testing by Christoph and Co.
> [...]
> 
> Hi Doug,
> 
> So we're after rc5 and not a single word from you re the plans for the
> next kernel,
> and your branch for that purpose contains exactly two patches... I do
> see few chunks
> of patches marked as under review in your patchworks, so there might
> be sign for getting
> some hope, is that real?
> 
> Or.
> 

Yes Or, I'm working on it.  A new 4.3-rc pull request should come out
today and an update on the for next area probably as well.

-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
              GPG KeyID: 0E572FDD



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]

Re: 4.4's rdma plate

[Or Gerlitz]

On 10/14/2015 3:56 PM, Doug Ledford wrote:
> Yes Or, I'm working on it.  A new 4.3-rc pull request should come out
> today and an update on the for next area probably as well.

Excellent. Re the next area, as was communicated to you multiple times, 
we really need the next bits to be in
your kernel.org tree and on one branch, else we can't really rebase 
re-spins of current/new patch series, pull
these bits into our internal build and regression systems. Also w.o 
this, you are highly exposed to
miss possible conflicts that could have caught by your branch being 
registered to linux-next and again get
those shouts from Linus when he steps on the conflicts while pulling, we 
had enough of it, really.

Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Or Gerlitz]

On Wed, Oct 14, 2015 at 3:56 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> On 10/14/2015 12:51 AM, Or Gerlitz wrote:
>> On Sun, Oct 11, 2015 at 11:58 PM, Or Gerlitz <gerlitz.or-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
>>> Hi Doug,
>>> As we're soon on 4.3-rc5, could you please update how things look for 4.4?
>>>
>>> The mutlicast loopback prevention patches from Eran were intially
>>> posted ~two months ago and went through testing by Christoph and Co.
>> [...]


>> Hi Doug,

>> So we're after rc5 and not a single word from you re the plans for the
>> next kernel, and your branch for that purpose contains exactly two patches... I do
>> see few chunks
>> of patches marked as under review in your patchworks, so there might
>> be sign for getting some hope, is that real?

> Yes Or, I'm working on it.  A new 4.3-rc pull request should come out
> today and an update on the for next area probably as well.

Doug,

So we're after rc6 and still pretty much nothing in your for-4.4
branch, are we getting there
(e.g picking something from the backlog such as the self multicast
loopback prevention and the
rdma-cm name space patches) or for 4.4 we'd be coming naked?

Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 1771 bytes --]

On 10/20/2015 12:25 PM, Or Gerlitz wrote:
> On Wed, Oct 14, 2015 at 3:56 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>> On 10/14/2015 12:51 AM, Or Gerlitz wrote:
>>> On Sun, Oct 11, 2015 at 11:58 PM, Or Gerlitz <gerlitz.or-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
>>>> Hi Doug,
>>>> As we're soon on 4.3-rc5, could you please update how things look for 4.4?
>>>>
>>>> The mutlicast loopback prevention patches from Eran were intially
>>>> posted ~two months ago and went through testing by Christoph and Co.
>>> [...]
> 
> 
>>> Hi Doug,
> 
>>> So we're after rc5 and not a single word from you re the plans for the
>>> next kernel, and your branch for that purpose contains exactly two patches... I do
>>> see few chunks
>>> of patches marked as under review in your patchworks, so there might
>>> be sign for getting some hope, is that real?
> 
>> Yes Or, I'm working on it.  A new 4.3-rc pull request should come out
>> today and an update on the for next area probably as well.
> 
> Doug,
> 
> So we're after rc6 and still pretty much nothing in your for-4.4
> branch, are we getting there
> (e.g picking something from the backlog such as the self multicast
> loopback prevention and the
> rdma-cm name space patches) or for 4.4 we'd be coming naked?
> 
> Or.
> 

Another pull request for the final 4.3 fixups will go in today.  I
prioritized those over the for-next work.

Next I'll be pulling simple for-next items, then I'll be reviewing the
RoCE GID cadche usage in verbs/cma v2 patchset, then the Add RoCE v2
support v1 patchset, then the New fast registration API v5 set.

-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
              GPG KeyID: 0E572FDD



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]

Re: 4.4's rdma plate

[Or Gerlitz]

On Tue, Oct 20, 2015 at 7:31 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> On 10/20/2015 12:25 PM, Or Gerlitz wrote:
>> On Wed, Oct 14, 2015 at 3:56 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>>> On 10/14/2015 12:51 AM, Or Gerlitz wrote:
>>>> On Sun, Oct 11, 2015 at 11:58 PM, Or Gerlitz <gerlitz.or-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
>>>>> Hi Doug,
>>>>> As we're soon on 4.3-rc5, could you please update how things look for 4.4?
>>>>>
>>>>> The mutlicast loopback prevention patches from Eran were intially
>>>>> posted ~two months ago and went through testing by Christoph and Co.
>>>> [...]
>>
>>
>>>> Hi Doug,
>>
>>>> So we're after rc5 and not a single word from you re the plans for the
>>>> next kernel, and your branch for that purpose contains exactly two patches... I do
>>>> see few chunks
>>>> of patches marked as under review in your patchworks, so there might
>>>> be sign for getting some hope, is that real?
>>
>>> Yes Or, I'm working on it.  A new 4.3-rc pull request should come out
>>> today and an update on the for next area probably as well.

>> Doug,
>> So we're after rc6 and still pretty much nothing in your for-4.4
>> branch, are we getting there
>> (e.g picking something from the backlog such as the self multicast
>> loopback prevention and the
>> rdma-cm name space patches) or for 4.4 we'd be coming naked?

> Another pull request for the final 4.3 fixups will go in today.  I
> prioritized those over the for-next work.
> Next I'll be pulling simple for-next items, then I'll be reviewing the

Sounds good as a starter

> then I'll be reviewing the
> RoCE GID cadche usage in verbs/cma v2 patchset, then the Add RoCE v2
> support v1 patchset, then the New fast registration API v5 set.

Doug, the New fast registration API v5 set has E2E community consensus,
I guess there's not much left for you to dig there, so these are good news.

As for the RoCE patches, you mentioned reviewing them, this is going to
take time, which makes sense. However, the self-multicast loopback prevention
patches are (1) sitting in the mailing list for months (2) small in
volume and very large
in benefit (3) passed testing by Christoph and Co, any reason not to
pick them prior
to review of the RoCE code.

We need to encourage developers to take part in the upstream process,
For this series
I am clueless towards the developers where they ask me what's missing?
why this isn't
moving for months?

Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Jason Gunthorpe]

On Tue, Oct 20, 2015 at 07:42:25PM +0300, Or Gerlitz wrote:

> As for the RoCE patches, you mentioned reviewing them, this is going to
> take time, which makes sense. However, the self-multicast loopback prevention
> patches are

My issue with the latest round of UAPI changes is I don't see any
Reviewed-By tag from someone at Mellanox whom I think understands our
UAPI.

UAPI changes are *NOT* something first-time list contributors should
be doing unassisted.

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Or Gerlitz]

On Tue, Oct 20, 2015 at 7:52 PM, Jason Gunthorpe wrote:
> On Tue, Oct 20, 2015 at 07:42:25PM +0300, Or Gerlitz wrote:

>> As for the RoCE patches, you mentioned reviewing them, this is going to
>> take time, which makes sense. However, the self-multicast loopback prevention
>> patches are
>
> My issue with the latest round of UAPI changes is I don't see any
> Reviewed-By tag from someone at Mellanox whom I think understands our
> UAPI.
>
> UAPI changes are *NOT* something first-time list contributors should
> be doing unassisted.

Jason, your point makes sense, when we did the internal review, I have
asked that someone with the relevant knowledge will review this part.

As I wrote below, we should encourage people to take part in upstream
development, and making such comment only after the patch sits on the
list whole two months (V1 was very minor change, V2 was re-spin) isn't
going in the encouragement direction.

There's one patch in the series that adds UAPI -- it's 1/7 "IB/core:
Extend ib_uverbs_create_qp"
https://patchwork.kernel.org/patch/7405241/

I will try to get Reviewed-by: from Haggai or Matan tomorrow, can you
also take a look today, so we don't miss 4.4 just for that?

Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Jason Gunthorpe]

On Wed, Oct 21, 2015 at 12:00:54AM +0300, Or Gerlitz wrote:

> As I wrote below, we should encourage people to take part in upstream
> development, and making such comment only after the patch sits on the
> list whole two months (V1 was very minor change, V2 was re-spin) isn't
> going in the encouragement direction.

I would say there are at least 6 people who could meaningfully put a
reviewed-by tag on that UAPI patch, why do you think none of them have
done so in two months? What is needed to get more engagement in
upstream?

> I will try to get Reviewed-by: from Haggai or Matan tomorrow, can you
> also take a look today, so we don't miss 4.4 just for that?

Unlikely, I am very busy with SC|15.

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Or Gerlitz]

> On Tue, Oct 20, 2015 at 7:31 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:

>> [...] Next I'll be pulling simple for-next items, then I'll be reviewing the
>> RoCE GID cadche usage in verbs/cma v2 patchset, then the Add RoCE v2
>> support v1 patchset, then the New fast registration API v5 set.

Doug, there are few singleton patches to mlxN drivers, just to make
sure they are not left behind (thanks)

a small mlx5 enhancement, Signed by Eli C. the driver maintainer
https://patchwork.kernel.org/patch/7454641

two small mlx4 fixes, acked by me...
https://patchwork.kernel.org/patch/7348911/
https://patchwork.kernel.org/patch/7348871/
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 1046 bytes --]

On 10/23/2015 02:46 AM, Or Gerlitz wrote:
>> On Tue, Oct 20, 2015 at 7:31 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> 
>>> [...] Next I'll be pulling simple for-next items, then I'll be reviewing the
>>> RoCE GID cadche usage in verbs/cma v2 patchset, then the Add RoCE v2
>>> support v1 patchset, then the New fast registration API v5 set.
> 
> Doug, there are few singleton patches to mlxN drivers, just to make
> sure they are not left behind (thanks)
> 
> a small mlx5 enhancement, Signed by Eli C. the driver maintainer
> https://patchwork.kernel.org/patch/7454641
> 
> two small mlx4 fixes, acked by me...
> https://patchwork.kernel.org/patch/7348911/
> https://patchwork.kernel.org/patch/7348871/

Yes, I know.  There are quite a few different things for mlx5.  I was
waiting on all of them until I knew that RoCEv2 was in as that gates
whether or not mlx5 RoCE can go in.


-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
              GPG KeyID: 0E572FDD



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]

Re: 4.4's rdma plate

[Or Gerlitz]

On Fri, Oct 23, 2015 at 5:52 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> On 10/23/2015 02:46 AM, Or Gerlitz wrote:

>> Doug, there are few singleton patches to mlxN drivers, just to make
>> sure they are not left behind (thanks)

>> a small mlx5 enhancement, Signed by Eli C. the driver maintainer
>> https://patchwork.kernel.org/patch/7454641

>> two small mlx4 fixes, acked by me...
>> https://patchwork.kernel.org/patch/7348911/
>> https://patchwork.kernel.org/patch/7348871/

> Yes, I know.  There are quite a few different things for mlx5.  I was
> waiting on all of them until I knew that RoCEv2 was in as that gates
> whether or not mlx5 RoCE can go in.

Doug,

So RoCEv2 didn't make it for 4.4, but still, the above are changes
touching driver only code and I don't see why they should be waiting
so long.

Can we finally see them in a k.o branch of yours please? either for
your 4.4-rc pull or on 4.5 branch.

Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 4.4's rdma plate

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 1718 bytes --]

On 12/08/2015 04:07 PM, Or Gerlitz wrote:
> On Fri, Oct 23, 2015 at 5:52 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>> On 10/23/2015 02:46 AM, Or Gerlitz wrote:
> 
>>> Doug, there are few singleton patches to mlxN drivers, just to make
>>> sure they are not left behind (thanks)
> 
>>> a small mlx5 enhancement, Signed by Eli C. the driver maintainer
>>> https://patchwork.kernel.org/patch/7454641
> 
>>> two small mlx4 fixes, acked by me...
>>> https://patchwork.kernel.org/patch/7348911/
>>> https://patchwork.kernel.org/patch/7348871/
> 
>> Yes, I know.  There are quite a few different things for mlx5.  I was
>> waiting on all of them until I knew that RoCEv2 was in as that gates
>> whether or not mlx5 RoCE can go in.
> 
> Doug,
> 
> So RoCEv2 didn't make it for 4.4, but still, the above are changes
> touching driver only code and I don't see why they should be waiting
> so long.

I knew some of the mlx5 changes relied on the RoCEv2 stuff (such as mlx5
RoCE support).  Others are larger in scope.  I tried to sort through the
singletons on patchworks today and saw at least one which I opted not to
take as it wasn't really a fix, it was a performance enhancement.  I'll
look over the ones you linked above to make sure I didn't miss them.

> Can we finally see them in a k.o branch of yours please? either for
> your 4.4-rc pull or on 4.5 branch.

I'm getting ready to put out the initial 4.5 branch.  I was only waiting
on getting the last few 4.4-rc patches in so I can then merge that into
the 4.5 base and go from there.


-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
              GPG KeyID: 0E572FDD



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]

Re: 4.4's rdma plate

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 961 bytes --]

On 12/08/2015 04:07 PM, Or Gerlitz wrote:
> On Fri, Oct 23, 2015 at 5:52 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>> On 10/23/2015 02:46 AM, Or Gerlitz wrote:
> 
>>> Doug, there are few singleton patches to mlxN drivers, just to make
>>> sure they are not left behind (thanks)
> 
>>> a small mlx5 enhancement, Signed by Eli C. the driver maintainer
>>> https://patchwork.kernel.org/patch/7454641

I originally had bumped this one to 4.5.  As you say, it's an
enhancement.  But, the particular issue is so simple, and can be
considered buggy behavior (hogging the CPU when you shouldn't), that
just adding a need_resched check I'll take.

>>> two small mlx4 fixes, acked by me...
>>> https://patchwork.kernel.org/patch/7348911/
>>> https://patchwork.kernel.org/patch/7348871/

I took these two.


-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
              GPG KeyID: 0E572FDD



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]