glamor-egl should be replaced with xorg-server's --enable-glamor

[ojab]

Since xorg-server-1.16.0 glamor is merged to xorg-server itself and should be used instead of glamor-egl.

[ken@…]

Agreed. I note that Armin has done this in the systemd book. AFAICS, this requires libepoxy (so, we have to replace glamor-eql with libepoxy, no net loss of packages). My first attempt on a radeon failed (the server for libepoxy was offline at that time). On my intel I have done this, but I'm not sure if it is working - the intel driver configure reported

checking whether to include GLAMOR support... no

and I am not at all sure how to test if glamor from xorg-server is actually working ? I have not, for the moment, reviewed the xorg drivers in the systemd book to see if any of them have had differnet switches added.

[Armin K]

Latest intel development ddx (latest 2.99.91) supports glamor from xserver, as well as xf86-video-ati-7.4.0.

[ken@…]

I've now looked at this for my own hardware, and eventually remembered my past experience.

For intel, sna is the default. Configure has an option --enable-glamor, but this also requires the runtime AccekMethod "glamor" to actually use it. The intel devs regard glamor as experimental, and it appears to me that they now prefer sna.

For ati, glamor is non-default for anything older than South Islands. It would require enabling it in a similar way. Both my amd/ati boxes are in this category. We say "Please note that Glamor acceleration can be used with other chips as well, from the R300 onwards, but it has not been tested recently." Seems to me that there is not much point in me rebuilding to test it.

So, I still think we should use glamor from xorg-server, instead of from the old glamor-egl. But none of my hardware will be using it, so I'm reluctant to change things which I will not be testing.

[ken@…]

Following comments on -dev, I'll do this.

[ken@…]

Done, r13954.