NAK: vmwgfx hardware enablement patch

[Thomas Hellstrom]

On 03/10/2014 12:22 PM, Andy Whitcroft wrote:
> On Mon, Mar 10, 2014 at 11:55:48AM +0100, Thomas Hellstrom wrote:
>> On 03/10/2014 11:44 AM, Andy Whitcroft wrote:
>>> On Mon, Mar 10, 2014 at 09:54:40AM +0100, Thomas Hellstrom wrote:
>>>> On 03/04/2014 11:58 PM, Tim Gardner wrote:
>>>>> On 03/04/2014 12:57 PM, Tim Gardner wrote:
>>>>>> On 03/04/2014 08:53 AM, Tim Gardner wrote:
>>>>>>> On 03/04/2014 08:00 AM, Jakob Bornecrantz wrote:
>>>>>>>> Hello all,
>>>>>>>>
>>>>>>>> This squashed patch series implements "guest backed surfaces" in the
>>>>>>>> VMware vmwgfx kernel driver. It is used along with changes in the
>>>>>>>> mesa source tree (which you guys have already have to my knowledge
>>>>>>>> picked up since it was in the 10.1 mesa release). We would like
>>>>>>>> to have this included in the trusty release, please.
>>>>>>>>
>>>>>>>> The patch is backported from the 3.14 kernel and is rather large,
>>>>>>>> but mostly only touches the vmwgfx driver and some minor touches
>>>>>>>> on the generic ttm code. You can get it from here:
>>>>>>>>
>>>>>>>> git://people.freedesktop.org/~wallbraker/linux for-unbuntu-vmwgfx
>>>>>>>>
>>>>>>>> Comments please?
>>>>>>>>
>>>>>>>> Cheers, Jakob.
>>>>>>>>
>>>>>>>> PS: I'm sorry I don't know which tags to put in the subject,
>>>>>>>>      so I just left it as it
>>>>>>>>
>>>>>>> I prefer the individual commits, each of which must be clearly
>>>>>>> marked as
>>>>>>> a clean cherry-pick or a backport.
>>>>>>>
>>>>>>> I've reviewed the list of commits that went into the squash. With the
>>>>>>> exception of 05efb1abecce6e36457ae1a7be29ded7ac52292a "drm/ttm: ttm
>>>>>>> object security fixes for render nodes", all of the other commits only
>>>>>>> touch drm/vmwgfx. I'll assume y'all know what you're doing there.
>>>>>>>
>>>>>>> rtg
>>>>>> How about this:
>>>>>>
>>>>>> git://kernel.ubuntu.com/rtg/ubuntu-trusty.git vmwgfx
>>>>>>
>>>>>> rtg
>>>>> Never mind. I just merged these patches. They'll show up in linux
>>>>> 3.13.0-16.36
>>>>>
>>>>> rtg
>>>> Tim,
>>>>  
>>>> Thanks for this. It occured to me that there is a fix for the TTM commit
>>>> in the squash patch. It's upstream  c66f85433, and I've attached it to
>>>> this email. I can include it in the squash and resend if you prefer.
>>>>
>>>> Thanks,
>>>> Thomas
>>>>
>>>> From c66f854338253e603a4fb6817069ee23eacf0ae3 Mon Sep 17 00:00:00 2001
>>>> From: Thomas Hellstrom <thellstrom at vmware.com>
>>>> Date: Fri, 24 Jan 2014 08:49:45 +0100
>>>> Subject: [PATCH] drm/ttm: Fix TTM object open regression
>>>>
>>>> Commit drm/ttm: ttm object security fixes for render nodes introduced a
>>>> regression where, if a TTM object was opened multiple times from the same
>>>> open file, the caller would spin uninterruptibly in the kernel.
>>>>
>>>> Fix this.
>>>>
>>>> Signed-off-by: Thomas Hellstrom <thellstrom at vmware.com>
>>>> Reviewed-by: Jakob Bornecrantz <jakob at vmware.com>
>>> He pulled in the unsquashed patches so updating the squash will not
>>> help.  I have cherry-picked this into our tree as well.
>>>
>>> Has anyone been able to test this new combination at all?  (Prior to
>>> this fix.)
>>>
>>> -apw
>> Is it available in
>>
>> git://kernel.ubuntu.com/rtg/ubuntu-trusty.git vmwgfx ?
> Everything but this last cherry-pick is uploaded as 3.13.0-16.36, the
> latter will be included in the next upload and is on the master next in
> our main repo.
>
> -apw

OK, I just tested 3.13.0-16.36 with old (HW10) and new (HW11) versions,
with a new upgrade (new mesa + xorg driver) and it seems to be working OK.

Thanks,

/Thomas