Re: [Mesa-dev] Merging the Vulkan driver

[Chad Versace]

On 04/05/2016 12:30 PM, Jason Ekstrand wrote:

...

> If people step up with code review, I'm hoping that we can merge some
> time this week or next.
> 
> --Jason Ekstrand

Woo! The Vulkan code is finally merged!
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] Merging the Vulkan driver

[Eric Anholt]

Jason Ekstrand  writes:

> All,
>
> We are getting very close to being able to merge the Vulkan driver into
> mesa master.  I've got around 30 patches on the mailing list that are still
> awaiting review.  Once those get merged, the diff between the "vulkan" and
> "master" branches is basically zero except for adding new files.  The going
> plan is to get the diff between the two branches down to zero except for
> the new "src/intel" directory and minimal build system changes and then do
> an actual git merge.  I know we don't usually do merges in this project,
> but the Vulkan driver is big and has a lot of history that we would like to
> preserve.  If you're strongly opposed to doing a merge, please speak up now!
>
> Things that still need review:
>
>  1) i965 back-end patches for indirect push constants
>  2) i965 and NIR patches to add a couple new opcodes
>  3) Misc i965 and NIR patches we've been carrying around in the vulkan
> branch for a while
>  4) A patch to configure.ac to add new flags for building Vulkan drivers
> (mostly written, not yet sent).
>
> Those will all undergo the normal review process and get pushed *before* we
> merge.  That way the only thing the merge does is add a new directory and a
> three or four lines each to configure.ac and src/Makefile.am.
>
> If people step up with code review, I'm hoping that we can merge some time
> this week or next.

I think a merge makes a ton of sense for this.  Having duplicate commits
in the history sucks, but maintaining history is worth it.


signature.asc
Description: PGP signature
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] Merging the Vulkan driver

[Kristian Høgsberg]

On Thu, Apr 7, 2016 at 4:34 PM, Grazvydas Ignotas  wrote:
> Hi,
>
> On Tue, Apr 5, 2016 at 10:30 PM, Jason Ekstrand  wrote:
>> I know we don't usually do merges in this project, but
>> the Vulkan driver is big and has a lot of history that we would like to
>> preserve.  If you're strongly opposed to doing a merge, please speak up now!
>
> I fear issues like the one in bug 94406 might show up at the merge
> point and it will be basically impossible to bisect (at least I've
> failed it with bug 94406). Do you think things like that will get
> sorted by the time you get to zero diff point of existing directories?

The goal is to get to no diff outside src/intel, at which point
anything that could break GL on master should be bisectable. This
isn't so much a question about full-history merge vs squash merge
though. If we didn't merge, we'd just have a giant squashed commit
that adds the entire vulkan driver, which is even less bisectable in
case of regressions.

Kristian

> Gražvydas
> ___
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] Merging the Vulkan driver

[Jason Ekstrand]

On Thu, Apr 7, 2016 at 4:48 PM, Jason Ekstrand  wrote:

>
>
> On Thu, Apr 7, 2016 at 4:34 PM, Grazvydas Ignotas 
> wrote:
>
>> Hi,
>>
>> On Tue, Apr 5, 2016 at 10:30 PM, Jason Ekstrand 
>> wrote:
>> > I know we don't usually do merges in this project, but
>> > the Vulkan driver is big and has a lot of history that we would like to
>> > preserve.  If you're strongly opposed to doing a merge, please speak up
>> now!
>>
>> I fear issues like the one in bug 94406 might show up at the merge
>> point and it will be basically impossible to bisect (at least I've
>> failed it with bug 94406). Do you think things like that will get
>> sorted by the time you get to zero diff point of existing directories?
>>
>
> I'm going to be paying very close attention to the diff.  Thanks for
> pointing that out.
>

Also, we have a very good CI system here that I will definitely be using.
We won't be going ahead with the merge if it causes either GL or Vulkan to
fail a single test.
--Jason
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] Merging the Vulkan driver

[Jason Ekstrand]

On Thu, Apr 7, 2016 at 4:34 PM, Grazvydas Ignotas  wrote:

> Hi,
>
> On Tue, Apr 5, 2016 at 10:30 PM, Jason Ekstrand 
> wrote:
> > I know we don't usually do merges in this project, but
> > the Vulkan driver is big and has a lot of history that we would like to
> > preserve.  If you're strongly opposed to doing a merge, please speak up
> now!
>
> I fear issues like the one in bug 94406 might show up at the merge
> point and it will be basically impossible to bisect (at least I've
> failed it with bug 94406). Do you think things like that will get
> sorted by the time you get to zero diff point of existing directories?
>

I'm going to be paying very close attention to the diff.  Thanks for
pointing that out.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] Merging the Vulkan driver

[Grazvydas Ignotas]

Hi,

On Tue, Apr 5, 2016 at 10:30 PM, Jason Ekstrand  wrote:
> I know we don't usually do merges in this project, but
> the Vulkan driver is big and has a lot of history that we would like to
> preserve.  If you're strongly opposed to doing a merge, please speak up now!

I fear issues like the one in bug 94406 might show up at the merge
point and it will be basically impossible to bisect (at least I've
failed it with bug 94406). Do you think things like that will get
sorted by the time you get to zero diff point of existing directories?

Gražvydas
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] Merging the Vulkan driver

[Emil Velikov]

Hi Jason,

On 5 April 2016 at 20:30, Jason Ekstrand  wrote:
> All,
>
> We are getting very close to being able to merge the Vulkan driver into mesa
> master.  I've got around 30 patches on the mailing list that are still
> awaiting review.  Once those get merged, the diff between the "vulkan" and
> "master" branches is basically zero except for adding new files.  The going
> plan is to get the diff between the two branches down to zero except for the
> new "src/intel" directory and minimal build system changes and then do an
> actual git merge.  I know we don't usually do merges in this project, but
> the Vulkan driver is big and has a lot of history that we would like to
> preserve.  If you're strongly opposed to doing a merge, please speak up now!
>
> Things that still need review:
>
>  1) i965 back-end patches for indirect push constants
>  2) i965 and NIR patches to add a couple new opcodes
>  3) Misc i965 and NIR patches we've been carrying around in the vulkan
> branch for a while
>  4) A patch to configure.ac to add new flags for building Vulkan drivers
> (mostly written, not yet sent).
>
> Those will all undergo the normal review process and get pushed *before* we
> merge.  That way the only thing the merge does is add a new directory and a
> three or four lines each to configure.ac and src/Makefile.am.
>
> If people step up with code review, I'm hoping that we can merge some time
> this week or next.
>
I'm not sure how qualified I'll be on the i965 specifics, but I will
gladly check out the build related bits.

Just a humble suggestion - can we please keep the C++ compatible
NIR_SRC_INIT/NIR_DEST_INIT ?

Thanks
Emil
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] Merging the Vulkan driver

[Jason Ekstrand]

All,

We are getting very close to being able to merge the Vulkan driver into
mesa master.  I've got around 30 patches on the mailing list that are still
awaiting review.  Once those get merged, the diff between the "vulkan" and
"master" branches is basically zero except for adding new files.  The going
plan is to get the diff between the two branches down to zero except for
the new "src/intel" directory and minimal build system changes and then do
an actual git merge.  I know we don't usually do merges in this project,
but the Vulkan driver is big and has a lot of history that we would like to
preserve.  If you're strongly opposed to doing a merge, please speak up now!

Things that still need review:

 1) i965 back-end patches for indirect push constants
 2) i965 and NIR patches to add a couple new opcodes
 3) Misc i965 and NIR patches we've been carrying around in the vulkan
branch for a while
 4) A patch to configure.ac to add new flags for building Vulkan drivers
(mostly written, not yet sent).

Those will all undergo the normal review process and get pushed *before* we
merge.  That way the only thing the merge does is add a new directory and a
three or four lines each to configure.ac and src/Makefile.am.

If people step up with code review, I'm hoping that we can merge some time
this week or next.

--Jason Ekstrand
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev