[OE-core] [PATCH v2] mesa: Upgrade to 17.2.1 release

[Otavio Salvador]

Hello Richard,

On Fri, Sep 22, 2017 at 3:43 AM, Richard Purdie
<richard.purdie at linuxfoundation.org> wrote:
> On Thu, 2017-09-21 at 21:43 -0300, Otavio Salvador wrote:
>> On Thu, Sep 21, 2017 at 7:34 PM, Richard Purdie
>> <richard.purdie at linuxfoundation.org> wrote:
>> >
>> > On Thu, 2017-09-21 at 11:08 -0300, Otavio Salvador wrote:
>> > >
>> > > Upgrade to a new stable release and drop patches applied on
>> > > upstream.
>> > >
>> > > For a full release notes, please see:
>> > > https://mesa3d.org/relnotes/17.2.0.html
>> > > https://mesa3d.org/relnotes/17.2.1.html
>> > >
>> > > Signed-off-by: Fabio Berton <fabio.berton at ossystems.com.br>
>> > > Signed-off-by: Otavio Salvador <otavio at ossystems.com.br>
>> > > ---
>> > >
>> > > Changes in v2:
>> > >  - update to 17.2.1
>> > The trouble is this isn't just a "stable release", I didn't check
>> > the
>> > patch fully enough initially but now I see this was a 17.1.X ->
>> > 17.2.X
>> > change.
>> The commit log mentions the two release notes. The 17.1 is at 17.1.9
>> release but should not receive more fixes. It is advised to move to
>> 17.2.1 as it is the current release but it is up to you if we will
>> take this one or 17.1.9.
>
> Its not clear from the commit message that we're on 17.1.7, that 17.1.9
> (and maybe .8?) exist and that in going to 17.2.X, we're actually
> changing stable series. Release notes for 17.1.X are not mentioned.

17.2.0 implies it is a major release, does not?

...
>> That's the reason for autobuilder, right? I do not spawn all
>> configurations here as I don't own such build monster ...
>
> The autobuilder is a last final sanity check so we can test the
> proposed changes, Ross and I are not here just to queue and test and
> report back on people's patches.
>
> In this case *any* poky build of mesa is failing. That is likely
> because poky has wayland in distro features and your local
> configuration does not.

Possibly.

In fact I was testing for an i.MX machine and as it mangles the
PACKAGECONFIG the test was not showing your error, even with wayland
enabled.

>> > If people want us to be more flexible about taking "point version"
>> > changes later into the release cycle, we really need to start
>> > getting
>> > better about how these things get tested and how we refer to them
>> > in
>> > the commit messages.
>> See the commit log above. You are frustrated, and I get that, but
>> commit log mentions BOTH!
>
> The commit message implies we're upgrading along a stable release path.
> Now, I agree had I thought more carefully I'd have realised that to
> mention release notes of a .0, we must not be. You do call it a stable
> update though and it really isn't.
>
>> > [and yes, the fix is trivial and included below, that isn't really
>> > the
>> > point though]
>> Do you want a v3? or a 17.1.9? Let me know what I can help.
>
> I left it building locally overnight and the first fix I mentioned
> isn't enough, it eventually fails with:
>
> | make[4]: *** No rule to make target '/usr/share/wayland-protocols/unstable/linux-dmabuf/linux-dmabuf-unstable-v1.xml', needed by 'drivers/dri2/linux-dmabuf-unstable-v1-client-protocol.h'.  Stop.
>
> so a v3 would need to figure out and fix that. Who knows what else
> might then fail. I suspect we might have more success with 17.1.9 but
> it sounds like I'll be the one who you want to test this...

I am preparing a v3 and testing with wayland and x11 enabled. Once it
works fine I send it out.

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750