loswillios: Magnade: yes, it does disappear when switching to console and back
loswillios: Magnade: and the reappear when I hook up my extented screen with xrandr
loswillios: *they
reinouts: hi, I was wondering if anyone could help me with the question I posted yesterday. I have a laptop with a native resolution of 1400x1050, and an external monitor that does 1680x1050. I use Xorg 7.3 with the radeon driver and whereas a 1680x1050 signal is sent to the external monitor (but only when I use the MergedFB option), the actual displayed desktop is 1400x1050.
loswillios: reinouts: try the ML when you get no answer here
reinouts: loswillios: http://lists.freedesktop.org/mailman/listinfo/xorg ?
reinouts: or is there a radeon-specific list
loswillios: there's a radeon specific one
reinouts: loswillios: could you be so kind to point me to it :)
loswillios: xorg-driver-ati@lists.x.org
reinouts: thanks
agd5f: reinouts: use xrandr
agd5f: xrandr --output VGA-0 --mode 1680x1050
reinouts: agd5f: thanks, but that makes no difference
reinouts: the current resolution already is 1680x1050, according to xrandr and also according to the monitor itself
reinouts: it's just that the desktop is a stretched 1400x1050
agd5f: reinouts: got a picture?
agd5f: reinouts: It's probably just your panel if you are using clone mode
agd5f: desktop apps will generally use the xinerama info from the first "head" for sizing panels and stuff, which in clone mode may not be the biggest screen
reinouts: agd5f: what kind of picture are you asking for?
agd5f: reinouts: also if you are using ubuntu, you may be getting hit by the edid parser bug
reinouts: agd5f: opensuse 10.3
agd5f: ok
reinouts: with added xorg 7.3
agd5f: reinouts: a picture of your monitor so I can understand what you mean by "stretched"
agd5f: reinouts: also can you pastebin the output of xrandr?
reinouts: well I don't have my camera at hand, but you'll have to trust me that what's on my external monitor is identical to what's on the laptop lcd panel, and on the laptop panel the dimensions are correct
reinouts: http://papernapkin.org/pastebin/view/16462/
MrCooper: reinouts: when you say 'stretched', do you really mean 'cropped', i.e. parts not visible?
agd5f: oh. your version of randr is too old
agd5f: or your xserver/driver doesn't support 1.2
reinouts: MrCooper: what I mean is that fonts are too wide
reinouts: what looks correct in 1400x1050 is stretched in 1680x1050
reinouts: agd5f: xrandr -v reports version 1.2
agd5f: reinouts: then you are using an old version of the driver. got a xorg log?
reinouts: yes I do...
reinouts: http://papernapkin.org/pastebin/view/16460
agd5f: reinouts: yeah,m you are using an old version of radeon
reinouts: and xorg.conf is on http://papernapkin.org/pastebin/view/16459
agd5f: 6.6.193. the randr versions are 6.7.19x
reinouts: ok, given that I upgraded each and every xorg package as packaged in the opensuse 10.3 xorg 7.3 repository, what do I do to upgrade the radeon driver to the correct version
reinouts: ?
MrCooper: we don't know, not being opensuse users
reinouts: would updating to xorg 7.4 beta help?
agd5f: reinouts: not sure what they package. you could build the driver from git
agd5f: I don't know why they package 6.6.19x. that series was a dead end
Magnade: loswillios: the bug i had in mind is 13610 but its closed so perhaps you will need to open a new one
MrCooper: agd5f: btw, KDE4 seems to have a pretty nice RandR 1.2 frontend GUI
MrCooper: pretty much like the Windows driver control panels
loswillios: Magnade: I only get the 60wakups when enabling my second monitor via xrandr
loswillios: before that it's fine
loswillios: should I reopen 13610 or file a new bugreport?
MrCooper: loswillios: as it's not the same problem...
loswillios: I thought so too. ok
Magnade: loswillios: yeah yours is a bit diff bug and that one had been closed so new is prob best
loswillios: oki, filed as https://bugs.freedesktop.org/show_bug.cgi?id=14203
MrCooper: the current way of disabling vblank interrupts is just inadequate, we need to finish the DRM vblank-rework branch
Magnade: MrCooper: doesnt catch all ways they get enabled?
MrCooper: that should be fixable, but it's really pointless as the current scheme will still enable them even when it's not really needed when any 3D client (such as compiz) runs
loswillios: disabling vblanks = getting rid of the 60hz wakups?
MrCooper: right
MrCooper: the wakeups are due to vblank interrupts
loswillios: ah, so an interrupt on every vertical refresh
MrCooper: right
agd5f: MrCooper: oh cool. I should check it out
ganymede: hey, i was running into some trouble getting direct rendering on my x300. i tried building mesa and xorg-x11 from svn, but then the modules failed to load (and ABI mismatch)
ganymede: so then i tried keeping xorg from gentoo's portage (stable) and using mesa from cvs, with radeon and r300 support compiled in, but then i did not get direct rendering, and i have a xorg log if anyone needs it
ganymede: but if i'm doing it completely wrong, then i guess the log isn't useful
ganymede: okay, i'm gonna try cvs-ing more things
rx__: pastebin the log
ganymede: okay, i'm going to paste the log
ganymede: http://rafb.net/p/mGn5JH73.html
ganymede: rx__: i think the relevant lines are 714 and 717
airlied: ganymede: for some rs4xx you need Option "DRI" "On"
airlied: ganymede: and I think you need a new radeon driver.
airlied: 6.6.3 is too old.
ganymede: i never understood exactly where the radeon driver comes from. it lives inside mesa, right?
airlied: ganymede: there are 3 pieces, 2D X.org driver, drm kernel driver and 3D Mesa driver.
ganymede: i'm using a cvs version of mesa with the r300 option enabled during compile
rx__: interesting... xf86-video-ati-6.7.x is hard masked
airlied: ganymede: that won't matter if the 2D driver doesn't enable the hw..
ganymede: oh okay, so then let my cvs-ise the other two parts
ganymede: this time, i will remove ALL xorg and recompile from scratch so i don't get ABI mismatchs
ganymede: airlied: thanks for this advice, i'll start the compile now and maybe i'll back later tonight
Magnade: rx__: ? gentoo?
rx__: yeah
rx__: according to packages.g.o
ganymede: rx__: it's okay, i'll build bleeding edge from an overlay
Magnade: o_O
rx__: i've stopped caring ever since i switched to git :)