Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 8 Sep 2014 12:01:31 +0300
From:      Konstantin Belousov <kostikbel@gmail.com>
To:        Jean-S??bastien P??dron <dumbbell@FreeBSD.org>
Cc:        freebsd-x11@FreeBSD.org
Subject:   Re: [CFT] i915: add HW context support (for Mesa 9.2+)
Message-ID:  <20140908090130.GY2737@kib.kiev.ua>
In-Reply-To: <540C1CA8.603@FreeBSD.org>
References:  <540C1CA8.603@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help

[-- Attachment #1 --]
On Sun, Sep 07, 2014 at 10:51:52AM +0200, Jean-S??bastien P??dron wrote:
> Hello!
> 
> Mesa 9.2+ requires "hardware context support" [1] in the i915 driver to
> work on Intel GPUs found in Sandy Bridge and later. Our driver lacks
> this feature, that's why we're stuck with Mesa 9.1 in the ports tree.
> 
> If you have one of those GPUs and you try Mesa 10.2 for instance
> (available from the xorg-dev's experimental branch [2]), it will
> complain with the following message:
> 
>   # glxinfo
>   name of display: :0.0
>   Gen6+ requires Kernel 3.6 or later.
>   Assertion failed: (ctx->Version > 0), (...)
>   Abort (core dumped)
> 
> I prepared a patch to add this feature to our i915 driver; the goal is
> to have it in 10.1-RELEASE:
> https://wiki.freebsd.org/Graphics/Add%20HW%20context%20support%20to%20i915
> 
> The article above explains how to test the patch. We're interested in
> success/failure reports for any GPU, not just Intel ones. So please test
> it! :)
> 
> Thank you!
> 
> [1]
> https://bwidawsk.net/blog/index.php/2013/01/i915-hardware-contexts-and-some-bits-about-batchbuffers/
> [2] https://wiki.freebsd.org/Graphics#experimental
> 

Please do not add dependency on OFED to the drm2.  It seems you need
idr, and all previous idr uses were replaced by unr(9) for drm2, please
do not break this.  It seems that unr(9) interface would need to be
extended to provide analog of idr_get_new_above().

What is the revision which you used for import of i915_gem_context.c ?
There were some very subtle semantical changes in the global gtt
mappings functions, esp. WRT interaction with lowmem conditions.  I am
not sure if it is applicable, and it would get me some time to really
remember the details there.

The i915 reset rework looks fine, I have it imported in my WIP branch,
but I think it will be merged without troubles.

Why do you need the changes to struct eb_objects ? If this is due to
OFED, please do not.

[-- Attachment #2 --]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=fzpE
-----END PGP SIGNATURE-----

Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20140908090130.GY2737>