Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2008-1-29

Search This Log:


loswillios: airlied: google earth only works properly on my rs480 with that low_impact thing.
berVi: hey
berVi: i just wanna ask whether you are aware of that bug appearing when logging in with gdm. letters in the input field are huge, but are normal on buttons like actions, language... and also they are too big while it displays the login splash, saying "Nautilus", "volume manager" and so...
canatella: 2
canatella: woops
berVi: but after the desktop loaded in, every letter is in normal size
berVi: AFAIK this bug appeared in 6.7.194 but i cant test it right now
berVi: i mean xorg-video-ati-6.7.194
berVi: 01:00.0 VGA compatible controller: ATI Technologies Inc Radeon Mobility X700 (PCIE)
berVi: and i have debian sid, but this also appeared in slackware
berVi: i can give you a radeon bios dump too :D
berVi: http://bios.egore911.de/Mobility%20M26%20X700%20-%20RV410%20-%20Berke%20Viktor%20-%202007-06-04.rom.gz
berVi: afk
MrCooper: berVi: sounds like the X server's default DPI value isn't what you want
reaktae: why does my secondary screen go black when cloning my primary that goes fullscreen?
berVi: MrCooper: as i have already mentioned, only the input field has wrong size letters in gdm, other labels are displayed properly. and besides the gdm login and splash everywhere else the size is OK
berVi: and _only_ with xorg-video-ati-6.7.194 or newer
MrCooper: berVi: still most likely DPI related
berVi: ah then could you please mention why did it work with 6.7.193? :)
berVi: its ok it could be dpi related but its something wrong in the driver
MrCooper: it would be nice if DPI was simply about 'right' or 'wrong' :}
MrCooper: have you compared the log files between the 'good' and 'bad' driver?
berVi: i can if you wish but it will take a few minutes
berVi: :)
berVi: hmm
berVi: make needs glxint.h but configure didnt check for it (6.7.193)
berVi: but its ok now, i had to install x11proto-gl-de
berVi: v
berVi: ok now a moment i restart x
berVi: not much
berVi: http://www.stud.u-szeged.hu/Berke.Viktor.Imre/xorg.log.diff
berVi: at least not much for my knowledge
berVi: :)
berVi: but this RADEONProbeOutputModes stuff seems to be missing in .194
berVi: -(II) RADEON(0): Setting screen physical size to 338 x 211 +(II) RADEON(0): Setting screen physical size to 289 x 21
berVi: 211 vs 21 :)
berVi: this may be the reason for this problem
berVi: or i donno
agd5f: berVi: could be a bad edid from your monitor
agd5f: berVi: have you also updated your xserver?
berVi: nope
berVi: but its also in the log
berVi: -(==) Log file: "/var/log/Xorg.0.log", Time: Tue Jan 29 15:33:41 2008
berVi: eh not this
berVi: - compiled for 1.4.0.90, module version = 6.7.193
berVi: + compiled for 1.4.0.90, module version = 6.7.194
berVi: i replaced nothing else but the xorg module
berVi: i would gladly make a screenshot but its not possible in the login screen afaik :)
agd5f: berVi: did you build your own xserver or have you been using the one from your distro? They may have updated it.
berVi: no
agd5f: berVi: can you pastebin your logs?
berVi: yea sure a moment
berVi: http://pastebin.ca/878961
berVi: http://pastebin.ca/878964
berVi: i use the debian sid packages, but this problem seems to remain with any version of xorg component
berVi: s
berVi: and it also occured on slackware 12 anyway
berVi: the only thing that has an effect on it is the xf86-video-ati module
agd5f: berVi: your edid is funky for your panel
agd5f: Image Size: 289 x 21 mm
berVi: yeah
agd5f: Max H-Image Size [cm]: horiz.: 33 vert.: 21
berVi: but check out the 193 one
agd5f: yeah, that's what I'm looking at
berVi: its especially strange coz, like i told previously, nothing else is wrong. only this big font sizes in the field where i have to enter my username and password in gdm, and in the login splash where it displays what it's loading
agd5f: berVi: newer xservers have a fix for this
berVi: "xorg-server 2:1.4.1~git20080118-1"
berVi: i think its quite up-to-date
berVi: :)
agd5f: berVi: I don't know if anyone ported the fix back from master
agd5f: The difference is that .193 didn't pass the edid back to randr to set the dpi for lvds. that was fixed in .194
berVi: so what do we do now?
agd5f: but it's broken for you since your edid is crappy
berVi: ahh and what is edid anyway? :))
agd5f: berVi: you can add a DisplaySize 289 210 line to a monitor section for your LVDS
agd5f: edid is the monitor information your panel/monitor returns when queried to tell the driver about it's capabilities and size
agd5f: looks like whoever wrote your edid mixed up their units (cm/mm) for the detailed timing block
berVi: :)
berVi: cool
berVi: another good point to buy an acer laptop;)
agd5f: newer xservers (don't recall when the fix was committed) will check for mixed up unit cases and fix them
agd5f: when pasring the edid
berVi: so in theory my problems gonna cease in the future
agd5f: should. you can also use the DisplaySize line to fix it now
berVi: ok now i do a ctrl alt backspace ;)
berVi: no joy
berVi: http://www.stud.u-szeged.hu/Berke.Viktor.Imre/xorg.conf
agd5f: berVi: add Option "Monitor-LVDS" "moncsi" to your device section
agd5f: also, remove the sync ranges from your monitor section
agd5f: you shouldn't need them
agd5f: with randr you have to associate monitors with outputs
agd5f: you can also change the identifier of your monitor to Identifier "LVDS"
berVi: ok i try a moment
berVi: refresh it
berVi: is it ok now?
agd5f: yeah
berVi: ok then i try it again
berVi: now it's ok in the login but fonts got bigger in general
berVi: :)
agd5f: berVi: well, you can adjust the size to your liking
berVi: maybe i should try with the .194 values
berVi: err .193
berVi: now its fine with the values found in the .193 log
agd5f: berVi: cool
berVi: but i'm wondering when the fix you mentioned will be merged which size it will use?
agd5f: berVi: it'll use the 289 210 as that's what your edid says the size of your panel is
agd5f: which could be wrong. edid's are know to have problems
berVi: yeah i see but then where did .193 get these sizes (which look much smoother) then?
berVi: and anyway, it looked the same in .194 too, even though it reported different sizes in the log
berVi: but if i manually put the sizes in xorg.conf fonts became bigger. how's that? :))
agd5f: I think the server calculates something to try and get 96 dpi
berVi: hmm but then your previous statement could be wrong, aint it?
agd5f: ?
berVi: that it will use 289 210
berVi: coz _in theory_ .194 uses that too, but actually it doesnt
berVi: it's saying that it uses that but if i put these sizes in xorg.conf it looks different
agd5f: no. it only calculates something if the driver doesn't pass the edid to randr to set the size which was the case for .193
MrCooper: it uses 21, not 210
agd5f: or if there is no edid
berVi: uses 21 for what?
agd5f: berVi: .194 uses 289 21
agd5f: since your edid is busted
agd5f: newer xservers should see that and fix it, making it 289 210
berVi: uh-huh but i still dont understand why .194 looks good if i gave it wrong values, and looks bad if i gave it good ones. and anyway, it still works almost flawlessly if i dont gave it anything, only this gdm problem occurs
berVi: :)
agd5f: berVi: alot of desktop environments assume 96 dpi as well. the whole thing is a bit of a mess
MrCooper: because 'good' and 'bad' is really a matter of taste
MrCooper: and hence the whole DPI issue is lose-lose
berVi: well okay
berVi: but please let me show you sg
berVi: :)
berVi: http://img180.imageshack.us/img180/1308/kpernykpjy1.png
berVi: http://img441.imageshack.us/img441/8840/kpernykp3xr9.png
berVi: which do you think is the proper display?
agd5f: second one?
berVi: eh :D
agd5f: it's a matter of taste
berVi: the second one reminds me gtk1 :)
berVi: but whatever i hope the fix will be merged soon
agd5f: berVi: get out a ruler and measure your screen:)
berVi: many thanks for your assistance
berVi: :D
agd5f: berVi: np
berVi: why not..
berVi: 31x28
agd5f: looks like your edid is wrong :)
berVi: i mean 310x280
berVi: yeah and looks like .193 reports it well
berVi: i guess coz it doesnt make use of edid right?
agd5f: yup
berVi: hmm and again (sorry for my ignorance) where do that 338x211 come from (if not from edid)
berVi: ?
agd5f: berVi: the xserver calculates a size based on the resolution and assuming 96 dpi
berVi: yeah i see
berVi: okay thanks
berVi: :)
agd5f: np
berVi: hmm but the actual size is 310x208, i mistyped it hehe
berVi: nope
berVi: 332x208
berVi: :D
berVi: i think i should get some sleep ;)
berVi: anyway, im curious whether the r300 will get any faster?
berVi: i'm not a gamer but it can be quite annoying when i cant scroll with normal speed on overdecorated webpages :)
berVi: hmm i just tried it and now it seems to be a bit faster :D
berVi: i think i give enemyterritory a go
mcgreg: hmm I've been reading the thread on phoronix "AtomBIOS Radeon vs. RadeonHD Drivers?" .. and well, why exactly are there 2 drivers for actually the same things? obviously you are both working on the same things but why dont you just cooperate and make *one* driver?
rx__: differences in developer's ideologies
berVi: ok it freezes my display with .194. it loads with .197, but freezes when i try to refresh server list :D
berVi: whatever, hope it will be fine, thanks for the help, bb
MrCooper: agd5f: do you agree we should revert the default for non-v3 AGP mode to leave it unchanged?
agd5f: I agree we should change it back
MrCooper: k, feel free to beat me to it :)
agd5f: heh :)
agd5f: MrCooper: do you recall the commit sha1 offhand?
MrCooper: note that I forgot to change the manpage in the same commit
agd5f: found it
agd5f: MrCooper: pushed
MrCooper: thanks
Xoipos: It is driving me nuts that using compiz or playing warcraft 3 lockup my PC after using them for a while >.<
johnficc1: Hi I have been using the radeon driver with my ati x600 and it works with compiz but every now and then I get a complete lock up and crash. Is there a way to fix this?
pan__: how come radeon doesn't work quite well on rs480?
rx__: hmm.. driver choices.. great
rx__: :)
rx__: what is rv620/rv635?
airlied: rx__: just another chipset.. I think we need to add a bunch of pciids and let atom handle it..