T`: hi, i am using fglrx now, but have no way to rotate one screen only.. is radeonhd ready for this? I was told to check back in a few months
Zajec: T`: your card is...?
T`: Zajec, hi, its a HD3450
Zajec: T`: you will need radeonhd 1.2.5 (newest version) and 2.6.30-rc1... then yes :)
T`: ah.. i dont see 1.2.5 on the website
Zajec: or you can compile drm.ko and radeon.ko for 2.6.X manually using mesa/drm
Zajec: oh, don't you?
Zajec: check first news on http://radeonhd.org/ :)
Zajec: yeah, wiki is not updated
Zajec: thanks for pointing
T`: ah great i see it!
T`: Zajec, where can i find the drm patches for kernel?
Zajec: looking for link...
Zajec: T`: you need only drm.ko and radeon.ko (instruction also mentions xf86-video-radeon git)
T`: gotcha.. thanks Zajec .. i wll try this out now
Zajec: i'll update that howto in a moment
T`: Zajec, a few suggestions: i had an issue with autogen.sh not working with AC_DISABLE_STATIC not defined error
T`: it was fixed when i installed libtool
T`: someone else had the same problem and mentioned it in this channel, i think it would be nice to have "need these packages, libtool, automake, xserver dev" in the instructions to compile
Zajec: T`: same for me, I'll add that
T`: Zajec, I can't use ATI command center for radeonhd right?
honk: of course not ;P
T`: heh, its just a X.org parsing tool
T`: xorg.conf i mean
T`: wishful thinking on my part :)
honk: what is? the catalyst control center?
honk: definitely not
T`: do i need to reboot after the drm module is installed?
honk: but you need to restart X
T`: thanks honk.. brb
T`: i have radeonhd now, but a few problems
T`: 1) No XV support
T`: 2) 2nd monitor is blurry.. its highly pixelated..
T`: X logs say i am on 1.2.5, but warn that i have fglrx kernel module loaded.. could that be the reason for this?
Zajec: T`: don't know what may cause blurry monitor... but for 1) attach your /var/log/Xorg.0.log for a good start of "debugging" :)
Zajec: T`: oh yeah
T`: sure.. lemme link this
Zajec: loaded fglrx may cause... everything :)
Zajec: T`: don't bother
T`: ok i just unloaded it.. lemme restart X
Zajec: remove fglrx first and restart your maching first (yes, full reboot)
Zajec: some registers will be still touched by fglrx if you don't reboot
Zajec: make sure fglrx won't load on system start :)
T`: i am just making sure it wont load
Zajec: blacklist, remove it, whatever :)
T`: blacklisted it
T`: yup.. brb
T`: Zajec, same problem still
Zajec: ok, now Xorg.0.log please :)
Zajec: that's *safe* log! https! :)
T`: ah wait.. i might know why
T`: i dont have dri enabled and EXA
T`: hehe yes Zajec :)
T`: is there a sample xorg.conf somewhere for dual screens?
Zajec: T`: don't know... I add second screen using xrandr every time I need it
T`: ok.. lemme just try one i guess
Zajec: "(EE) RADEONHD(0): RHDHdmiInit: unknown HDMI output type" that's weird
T`: Zajec, i have dual monitors setup via DVI and HDMI
T`: Zajec, wanna see my xrandr -q output?
Zajec: T`: hm, moment :)
Zajec: (II) RADEONHD(0): Output DVI-D_1 connected
Zajec: (II) RADEONHD(0): Output DVI-I_2/digital connected
Zajec: so you have two displays working, right?
T`: Zajec, yes, both are mirrored, but one of them is blurred
T`: its strange to describe.. its similar to zooming a small picture into a bigger one and everything is pixelated
Zajec: T`: is 1920x1200 native resolution for both your monitors?
Zajec: (I guess not) :)
Zajec: you have changes resolution on blurred monitor. for example is monitor connected to DVI-I_2/digital is blurred, then changes it's mode using xrandr
Zajec: this will be something like:
Zajec: xrandr --output DVI-I_2/digital --mode 1600x900
Zajec: of course use native resolution of your monitor, don't use my 1600x900 example :)
T`: i have 1600x1200 on both
T`: ah may be i know
T`: maximum 1920 x 1920
T`: may be thats the reason?
Neo_The_User: change virtual
T`: hmm actually that cant be it.. the screen is in mirror mode now
Zajec: Virtual bigger than monitor res is OK
T`: Zajec, sorry i meant both monitors are at resolution 1920x1200
Zajec: T`: and is that native res of that monitors?
T`: hey taht fixed it!
T`: xrandr --output DVI-I_2/digital --mode 1920x1200
T`: now everything looks crisp and nice
Zajec: good you solved that
Zajec: it's just weird radeonhd didn't use 1920x1200 by default
T`: ok i am gonna muck with this later today.. gotta go now
T`: thanks for the help Zajec!
T`: and others
Zajec: you're welcome
Neo_The_User: no problem
Zajec: [ot] how to log everything I do in console?
Zajec: there was some command...
Zajec: yeah, script creating "typescript" log file :) I know I was near :)
neosisani: hi, how can i force radeonhd NOT to load composite. I've tried commenting out line witch loads dri, setting AIGLX to false (in server layout) and AIGLX to off (in ServerFlags), i've also set composite to both false and off in Extensions, but it is still being loaded.
yangman: EXANoComposite for driver specific disable
yangman: the module itself is handled by xserver, and it's built-in
neosisani: that didn't help. Maybe my guess was wrong. Any idea on why display gets really low refresh rate, i can't play movies any more if fullscreen and cpu usage jumps to 20% whenever i change focused window? This happened after upgrade of i think kernel and udev if it matters. Had gnome upgrade a week earlier and it worked. Now i've upgraded xorg but still same problems remain.
yangman: which GPU?
yangman: and what kernel did you upgrade to?
yangman: and try 1.2.5 if you haven't already
neosisani: yangman: 2.6.29
neosisani: yangman: yup i'm using 1.2.5
neosisani: not sure about gpu, hd3300 integrated on msi ka790gx i think
yangman: pastebin your Xorg.0.log please
neosisani: ouch, hard to do inside console, one moment to switch to X
neosisaniOnX: ok, here it is http://codepad.org/7Nxlbh7O
yangman: neosisani: seems like you have DRI enabled, but missing the required modules
yangman: so, a difference in expectation? I'm assuming it had it working before
neosisaniOnX: should DRI module show in /proc if they exist? or is there some modprobe i could try?
yangman: the needed DRM modules are only in 2.6.30
yangman: you'll need to compile them manually from mesa/drm for earlier kernels
neosisaniOnX: btw, what is chance of breaking something doing this? should i postpone till exams are over or go for it?
yangman: it won't break anything
yangman: worst case, situation doesn't improve
neosisani: yangman: it works now, thanks
parry_rhd: On a 4670 I am having screen corruption issues - this is with git master
parry_rhd: Is this something known/worked upon?
parry_rhd: I am not doing 3D - just a KDE 4.2 desktop doing browsing etc.
parry_rhd: minimizing windows gives black/grey corruption pattern on the tool bars of konqueror for example
yangman: parry_rhd: similar to http://bugs.freedesktop.org/show_bug.cgi?id=20892 ?
parry_rhd: yangman: similar yes but this is on RV730 and the corrupted area in my case always has black/grey lines
parry_rhd: those are probably not significant differences
yangman: file a bug against radeonhd. try radeon as well
parry_rhd: Ok - just commented against 20892 - will try radeon and see if it has this bug as well
ootput: hello. any idea when 3d accel will be in effect for rv700 chipsets (hd4850 here)? Or, can radeonhd make use of any of the -vo's for mplayer besides x11?
airlied: ootput: you can do xv with latest bits now.
airlied: ootput: no 3D yet
ootput: i've tried -vo xv, but it says it's not supported
ootput: i'll check which version of radeonhd i have
ootput: (looking at the changelog, xv should be enabled)
yangman: ootput: for r6xx/r7xx, you need 1.2.5 along with latest DRM modules for radeon
ootput: yangman: cheers
ootput: i didn't realise there was even a 1.2.5
yangman: it was tagged yesterday
ootput: sorry if this seems rather silly, but module would be automatically used by xorg, or is that a kernel module?
yangman: kernel module
yangman: it's only in 2.6.30
ootput: ah, radeon, and drm, it would appear
yangman: or you can compile the one from mesa/drm: http://wiki.x.org/wiki/radeonhd%3Ar6xx_r7xx_branch
ootput: with 2.6.29, the git'd radeon&drm modules fail to be loaded, with alot of "Unknown symbol drm_*" errors
ootput: i've followed the instructions, and make install does not produce errors with the appropriate autogen flags for my system
ootput: FATAL: Error inserting drm (/lib/modules/2.6.29-gentoo-r1/x11-drm/drm.ko): Cannot allocate
ootput: FATAL: Error inserting radeon (/lib/modules/2.6.29-gentoo-r1/x11-drm/radeon.ko): Unknown s
ootput: ymbol in module, or unknown parameter (see dmesg)
ootput: dmesg at: http://rafb.net/p/pCpeF961.html
yangman: seems like it's not compiled against the right kernel
Shadow_Master: fI stumbled upon some odd behavior after resuming from suspend-to-disk yesterday, while having EXA and DRI enabled (for the X.org Composite extension) on an onboard "RS780M/RS780MN" chipset (as reported by lspci):
Shadow_Master: kernel: [drm] wait idle failed status : 0xA0003030 0x00000003 (repeated ad infinitum)
yangman: Shadow_Master: 1.2.4 or 1.2.5?
Shadow_Master: yangman: the latter
yangman: Shadow_Master: drm from mesa/drm or kernel?
Shadow_Master: from mesa/drm
yangman: can you give 2.6.30-rc1 a try?
yangman: mesa/drm "works", but isn't up to date
Shadow_Master: um.... not sure :-/ 2.6.30 was annoyingly unstable for me - sometimes it'd not give me control over my keyboard
ootput: yangman: i wouldn't need 2.6.30+ if i compiled from those instructions, right?
Shadow_Master: I haven't been able to reproduce it since last night either
Shadow_Master: (same kernel, same lib, same everything)
yangman: ootput: 2.6.30 includes the correct drm modules
Shadow_Master: and the server log's tail was like (WW) RADEONHD(0): DRMCPIdle: DRM CP IDLE returned BUSY! (repeated ad infinitum)
Shadow_Master: (obviously, bah)
yangman: Shadow_Master: can't really say where the fault is if it's not reproduceable. a race condition in drm is most likely, but in that case it may already be fixed for 2.6.30
Shadow_Master: right. If I stumble upon it again, I guess it'll then be time to try a prerelease kernel for the first time :)
Shadow_Master: "2.6\30 was annoyingly unstable for me" <- meant 2.6.29 , oops
ootput: yangman: 2.6.30-rc1?
ootput: i wouldn't need the -git5, would i?