On Thu, 2014-08-14 at 17:45 -0500, Bruce Dubbs wrote: > Ken Moffat wrote: > > As of xorg-server-1.16, glamor is included in xorg-server. Ticket > > #5347 was created for this, but nobody apart from me (and Armin, for > > a clarification which led me to an optional --enable-glamor in the > > intel driver) has commented, and I have expressed my unwillingness > > to touch it because I won't be using it - changing things I don't > > use, and which don't get tested, is a recipe for pain in the future. > > > > Since we are getting close to 7.6, it would be nice to resolve this > > one way or the other. If people are willing to review this, and if > > my systems are usable (I'll probably be offline from time to time in > > the next few days, rebuilding my RAID array and then expanding it), > > I would be willing to touch this. > > > > First, a question - does anyone reading this USE the radeonsi > > xorg driver ? > > Not me. The last ATI card I used was on a 386 long retired. I believe > it was a Rage 128. > > > My current understanding of glamor is: > > > > 1. it was developed initially by intel, but intel devs now prefer > > sna and regard glamor as experimental - it needs to be enabled at > > compile time, and also an xorg.conf entry at runtime to use it. > > OTOH, for newer drivers (such as radeon South Islands) it will speed > > up driver creation. > > > > 2. for radeon South Islands and newer (radeonsi) it is *required*. > > I do not have that hardware. For earlier radeon chipsets (r300 and > > newer) it is available, but not the default : it requires another > > --enable-glamor at compile time, and a similar xorg.conf entry. For > > South Islands, --enable-glamor is required, but not the xorg.conf > > entry. We use a phrase like "not recently tested" for glamor on > > non-radeonsi radeons, I see no reason to change that. > > > > 3. glamor used to be a separate package (glamor-egl), which is what > > we still have in trunk. Although the 1.16 server includes its own > > glamor [ with --enable-glamor ], it requires libepoxy (at least in > > my experience, but if I was doing this I would be happier if someone > > else confirmed that). So, dropping glamor-egl is replaced by adding > > libepoxy. > > My thought is to not --enable-glamor, but explain it in the comments. > I've got no problem with dropping glamor-egl and adding libepoxy (as > built, not checked) as long as it gets done in the next week or so. > > > 4. Armin has put the internal xorg-server glamor, and libepoxy, into > > the systemd book. To me, that appears to be the way to go. He > > marked libepoxy as "recommended" which initially puzzled me (is > > there maybe an option to build without it?), but I now regard that > > as a technically correct description for the specified configure > > options. If I was doing this in trunk, I think I would probably use > > --enable-glamor with an explanation that it is required for > > radeonsi, and for libepoxy I would add a modification " (Required > > only for glamor)". > > I'd suggest putting it as optional and specify '(required if using > glamor)' in the dependencies section of the xorg-server. > > It's doesn't seem to make sense for people like me with intel or nvidia > video cards. > > I could be wrong about all of this, but that's my current interpretation. > > -- Bruce >
Hello, For me I made the mistake of actually using the glamour included in the latest xorg server on my intel laptop. I created the xorg.conf and rebooted. In my case it slowed everything down. It caused the mouse to lag and the actual screen refresh rate was slowed down so much you could almost see it creating the images pixel by pixel. Note that this is on an i686 laptop, and an elderly one at that. Needless to say I quickly recompiled without it and deleted the xorg.conf file and sanity was once again returned to my computer. Regards, Christopher. -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
