the-me: will radeonhd use the existing drm/radeon module later or his own one?
BigBrain: the-me: afaik, we want them to share the same drm module
BigBrain: as it would only mean duplicated code and thus less productivity
the-me: okay that's good :=)
ndim: But it is useful that radeonhd by default does not use DRM.
ndim: The DRM panics my kernel whenever I use the fast user switching.
BigBrain: hm.. is the a way to install Xorg 7.2 and Xorg svn/git at the same time? (without rendering one unusuable)
BigBrain: e.g. install 7.2 into /usr and svn/git into /usr/local
ndim: BigBrain: I would not try it any other way.
BigBrain: ndim: will it just work then, i.e. will my gdm keep on using 7.2 as it does now and will /usr/local/bin/startx run properly?
ndim: BigBrain: I'd go even further and install the Xorg svn/git stuff into $HOME/xorg-test so I can easily remove the whole thing again with rm -rf
BigBrain: ah that's a good idea, thanks!
ndim: BigBrain: I'm not sure how exactly all the stuff is built. If it installs libs into $libdir and links against them, you might have to fiddle with LD_LIBRARY_PATH unless the default build adds the proper rpath while building.
BigBrain: k
ndim: Avoiding /usr/local has the additional advantage of not being in anyone's PATH
ndim: I have to admit, though, that I have no idea whether it will actually work :)
BigBrain: it's worth a try at least ;-)
ndim: Perhaps that build-all-dependent-packages thing could make things easier.
ndim: What was it called?
BigBrain: hm... no idea
lupine: apt-get build-dep ?
lupine: ;)
ndim: It eats an XML file with all the package definitions including their deps, git/svn/etc locations, etc., and then builds/installs them in one run.
ndim: Written by James Henstridge, IIRC
lupine: not portage, then? :p
ndim: Eh.
ndim: jhbuild :)
BigBrain: http://www.x.org/wiki/Development/git <-- guess that's okay too :)
ndim: For local builds of git stuff, I have written a script which builds and installs into branch specific directories, but that is only useful for testing multiple branches of a single package.
ndim: So regardless of what branch I have checked out, 'git-amb make install' will build the current branch in _build/${branch} and install it to _install/${branch}
ndim: Reminds me... I should work a little on the successor software which can also handle bzr branches, and non-GNU buildsystems.
dvandyk: greetings
dvandyk: I got a weird resolution problem with my (new) radeonhd-capable card
dvandyk: when connected to my tv via hdmi it only shows 2 resolutions, which are both not native to the tv
bridgman: can you paste your logs somewhere ?
dvandyk: which logs do you want? Xorg.log?
bridgman: yep
bridgman: that should give some info about EDID data read from the monitor and how
bridgman: the driver decided what resolutions to expose
dvandyk: http://rafb.net/p/TVFDc343.html
bridgman: according to the log the monitor says it supports two video modes, both
bridgman: 1280x720, one with what appears to be reduced blanking
bridgman: also mentions a 640x480 vesa mode but not sure if that is from the EDID info
dvandyk: 1280x720 is definitely not display correctly
bridgman: which modes are you seeing ?
dvandyk: well, this is a 32" LCD tv. its native mode is 720p, which is not 1280x720
bridgman: not displaying correctly is not the same as "not native" though
bridgman: AFAIK 720p *is* 1280x720, isn't it ?
dvandyk: right, but 1280x720 != 720p
dvandyk: nope, it's 1360x720 or something like that
dvandyk: well, the fact is, parts of the screen are cut off
bridgman: hold on, let me check
bridgman: the monitor edid info defnitely says 1280x720
dvandyk: or you're right...
dvandyk: weird, the "ati" driver, which i previously used always reported 1360x720, but it displayed correctly
dvandyk: i'm sorry
dvandyk: so my problem is then, that 1280x720 isn't displayed correctly. all borders are cut off
bridgman: no worries; it would be interesting to see the logs from -ati, of course ;)
dvandyk: sorry, no access anymore :-/
bridgman: understood ;)
bridgman: might be that the -ati driver knows something special about that monitor
dvandyk: i just built the driver from git sources
bridgman: or, of course, you might have adjusted the monitor to display correctly with -ati ;)
dvandyk: going to test it now, biab
dvandyk: bridgman: i'm not able to adjust the monitor :-) so that's ruled out
bridgman: pity ;)
dvandyk: ok, no changes with xf86-video-radeonhd from git
dvandyk: hm, how can i check what size the virtual desktop has?
dvandyk: ahh, this is interesting (and explains why all fonts are so huge)
dvandyk: DVI-I_1/digital connected 1280x720+0+0 160mm x 90mm
dvandyk: the size definitely ain't 160mm x 90mm :-)
dvandyk: i can change that using xrandr, right?
bridgman: I think so but not sure of details without digging; ajax was explaining something
bridgman: very odd about dpi and randr a couple of days ago, so not 100% sure you can
bridgman: over-ride
bridgman: you mentioned two modes; are they 640x480 and 1280x720 or 2 different 1280x720s ?
dvandyk: 640x480 and 1280x720
bridgman: my guess is that you are running at reduced blanking but the log seems to stop
bridgman: before I can see the final decision re: resolution
bridgman: suspect you have : Modeline "1280x720"x0.0 74.25 1280 1390 1430 1650 720 725 730 750 +hsync +vsync (45.0 kHz)
bridgman: when you really want Modeline "1280x720"x0.0 74.25 1280 1720 1760 1980 720 725 730 750 +hsync +vsync (37.5 kHz)
dvandyk: ok, can you phrase that in an xrandr --addmode? ;-)
dvandyk: ah, i can show you the new logs from the git source
dvandyk: do you want to have a look at it?
bridgman: it is definitely possible to phrase that in an xrandr command but you may
bridgman: not be talking to the best person to actually *do* that ;)
bridgman: sure, please paste the other logs
dvandyk: http://rafb.net/p/SYYcb530.html
dvandyk: there is the third modeline you spoke of
dvandyk: but xrandr doesn't display it
bridgman: Two odd things with the logs - first is that the edid parsing seems to happen
bridgman: many times even though you only have one monitor, second is the following line
bridgman: (II) RADEONHD(0): Using EDID range info for horizontal sync
bridgman: first time through the above line appears, remaining times you get :
bridgman: (II) RADEONHD(0): Using hsync ranges from config file
bridgman: do you have sync ranges in the config file ? If so what are they ?
dvandyk: well, that's two logs, appended
dvandyk: lemme look
dvandyk: no sync ranges in the config file
bridgman: ok, that's wierd ;)
dvandyk: the config file is pretty empty
dvandyk: i can clear the log file and restart X if you'd like to
bridgman: at this point I think you have two options; try again with -ati and see
bridgman: what the logs say there, or wait for someone who knows the radeonhd mode
bridgman: validation code to jump in
bridgman: might want to post on the mailing list
bridgman: and include links to clean log & conf files
dvandyk: ok, -ati supports r5xx?
bridgman: yep... but didn't you say you were running with that before ?
dvandyk: on a different box
dvandyk: which had an r3xx
bridgman: ahh... and different GPU as well ?
dvandyk: jupp
bridgman: got it
bridgman: ati works with 5xx as well
bridgman: might be worth a try - or stay with radeonhd for now and post on the
bridgman: mailing list
dvandyk: compiling it already
libvde: dvandyk: send a log to the ml please.
libvde: dvandyk: unless randr is being idiotic again, our modevalidation should be correct
libvde: dvandyk: so something is going on here that is messing things up, whether it is bad config, bad edid or even a bug, that remains to be seen
libvde: bridgman: thanks.
dorino: !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
GerbilSoft: dorino this channel is for the linux RadeonHD driver, if you prefer to spam exclamation points, go to #exclamationpoints or something.
dorino: WHO CARES.
the-me: dorino, nobody cares about you? :)
dorino: THE-ME THE GAY
the-me: how evil you're, I'll laugh later if it's ok
dorino: ???????????????????????????????
libv: i really should be using only one client :)
the-me: libv, anything new with the bugreport? I'll travel the next week and so on I can't test anything because my desktop pc will stay at home :)
libvde: the-me: i talked it over with egbert, and he'll commit a fix when he has the time
the-me: ah okay, so on I think you don't need more informations?
libvde: the-me: with the bugreport and the debugging info and a partial fix, we have all the info we need :)
the-me: okay fine :)
dmb: a spammer found our channel?
dvandyk: libvde: will do
dvandyk: libvd: nvm, got it!
dvandyk: libvde: my lcd tv had overscan turned on. Setting it to "Just Scan" made everything work perfectly nice :-)