airlied: umtc: cd /sys/bus/pci/devices/
airlied: umtc: in theory rom will contain something..
umtc: what is /sys?
airlied: umtc: sysfs..
airlied: umtc: a kernel filesystem exposing the entrails of the Linux kernel...
umtc: i'm not using linux
airlied: umtc: *BSD?
umtc: yes. is there something like a c program equivalent?
airlied: umtc: you can use /dev/mem I think
airlied: dd if=/dev/mem of=vgabios.bin skip=1536 count=128
umtc: R300 Infineon 9500 Pro BIOS - P/N 113-A05602-101
umtc: thank you for your assistance
umtc: do you think he'd want it at the kemper.freedesktop.org email address?
airlied: umtc: agd5f@gmail.com
airlied: oops..
airlied: alexdeucher@gmail.com
rx__: :)
paran: any radeon developers around?
loswillios: I read somewhere that compiz/AIGLX doesn't work with XPress 200M cards due to "Memory issues.
loswillios: does that still apply?
loswillios: it works nice with composite extension though. I get all the shiny transparency from xfce ☺
MrCooper: paran: just ask your actual question
paran: MrCooper: did that 8 hour ago and was recomended to wait and see when some devs where around
MrCooper: k, but actual questions are more likely to get a response than meta questions
paran: yeah, I know
paran: the short version: laptop CPU never goes lower than C2 sleep after suspend/resume, restarting the X-server resolves the problem, I am guessing some register gets messed up
paran: but I have really no idea at all how to proceed with debugging, was hoping to get some pointers here :)
reaktae: HELLO!!!!!!
reaktae: i need some serious help! ...probably just a few minutes for a proper geek though...
glisse: reaktae: just fire your question and you will see if anyone can help
reaktae: lspci tells me my card is ATI Technologies Inc RS300M AGP [Radeon Mobility 9100IGP].
reaktae: i've tried every possible driver i could come across but none of them worked for me...
reaktae: which one should i go with... ??
glisse: reaktae: its ati one
reaktae: #ati tells me it's radeon...
reaktae: in the meantime... doesn't ati only "wrap" radeon?
glisse: ati wrap radeon
glisse: reaktae: what doesn't work ?
reaktae: when i go with anything else bu vesa my ubuntu always ends up booting up in "Failsafe"
reaktae: tried reinstalling, even other distros act this way
reaktae: .. that's WITH "failsafe" drivers
reaktae: What difference does 'ati' hold over 'radeon'.... configuration-vise
reaktae: wise
glisse: nothings
glisse: reaktae: well xorg log of failed radeon run would be helpfull
reaktae: ...reproducing...
reaktae: bb
loswillios: nope, compiz-fusion still hardlocks on my XPress 200M
reakta1: glisse: http://paste.ubuntu-nl.org/52575/
glisse: reakta1: this is xorg log of failsaf
reakta1: which doesn't exist ... i deleted it before the reboot...
reakta1: unless i'm totally off... isn't it /var/log/xorg.0.log?
reakta1: sorry - i'm a lamer pretty much!
glisse: reakta1: well as failsafe overwrite it is should be somethings like Xorg.0.log.old
reakta1: that one says "(++) Using config file: "/etc/X11/xorg.conf.failsafe"" in it as well... and there no other ones...
glisse: reakta1: ubuntu failsafe things suxx
reakta1: PLUS the fact that xorg.conf.failsafe DOES NOT EXIST....
reakta1: what do i do?
glisse: in /etc/gdm/gdm.conf comment out line about failsafe
glisse: then restart gdm with radeon as driver
glisse: save Xorg.0.log
glisse: reenable failsafe
glisse: and voila
reakta1: momento por favor..
reakta1: brb
reaktae: glisse: ended up not booting here and since i'm a lamer it took me a while to recover from that... :) sorry about that...
reaktae: anyway> http://paste.ubuntu-nl.org/52584/
reaktae: what is a DotClock Modeline anyway?
glisse: reaktae: mv /etc/X11/xorg.conf /etc/X11/xorg.conf.old
glisse: and restart gdm
reaktae: so now that i don't have an xorg.conf it seems to load 'ati' and my real screen resolution is 1280x800 (desired) but system thinks it's 640x480
glisse: reaktae: do you got xorg log handy ?
reaktae: http://paste.ubuntu-nl.org/52586/
glisse: reaktae: well according to log it set 1280x800
glisse: what does xrandr say ?
reaktae: Screen 0: minimum 320 x 200, current 1280 x 800, maximum 1280 x 1200
reaktae: VGA-0 disconnected (normal left inverted right)
reaktae: LVDS connected 1280x800+0+0 (normal left inverted right) 0mm x 0mm
reaktae: 1280x800 60.0*+ 60.0
reaktae: 1280x768 60.0
reaktae: 1024x768 60.0
reaktae: 800x600 60.3
reaktae: 640x480 59.9
reaktae: S-video disconnected (normal left inverted right)
reaktae: sorry about that... should have pasted another bin, huh?
glisse: the * denote the actual mode
glisse: well better to pastebin or ubuntubin as you might get kick for flood
glisse: :)
reaktae: ... don't want that ...
glisse: so what make you think that you are in 640x480 ?
reaktae: i know i'm in 1280x640 - it's my "Screens and Graphics" UI that says a whole lot of nonsense...
reaktae: for instance: Graphics card (ATI Radeon (fglrx)) Driver: ati
reaktae: ...and then the resolution....
glisse: reaktae: well likely ubuntu bug as log & xranr say that you are in 1280x800
reaktae: must be.. whatever i try to change through this ui - it only messes up my xorg.conf ... if it does anything.... most of the times it just closes without a change to my settings...
glisse: reaktae: sadly the whole ui stuff to setup graphic isn't yet good enough
glisse: well it mostly due to us not providing for long time proper way to dynamicly configure things
glisse: :)
reaktae: :)
glisse: next ubuntu should have a better ui on top of xrandr 1.2
reaktae: can't wait...
reaktae: what do i do until then?
glisse: reaktae: in the meantime you can use xrandr command directly tjere is a bunch of tutorial on the net
reaktae: what should i put into my xorg.conf... radeon?
reaktae: or ati?
reaktae: :)
glisse: radeon is good enough
reaktae: alright
glisse: just let your xorg.conf as it is now and use xrandr :)
reaktae: i'll give that a go...
reaktae: thanks so much for your help glisse!
glisse: you welcome
reaktae: you JUST PAID IT FORWARD!
reaktae: and if you ever need any help with windows registry - let me know :)
glisse: i stay away of windows
glisse: i don't have anywindows anymore since years now
glisse: i am just missing good old gta games and things like that
glisse: :)
reaktae: i'm trying not to either... and the main reason why i'm so desperate to hook this driver up is to play some nfs via cedega or something....
reaktae: BTW - DOES THE RADEON provide me with 3d accel of sorts?
glisse: reaktae: it should if glxinfo say direct rendering yes
reaktae: it used to but just now it dumps the core instead......
reaktae: anyway - sorry to ask again but what should i have in my xorg.conf?
reaktae: i don't rally have one, remember? should i copy the *.old back ?
glisse: not the old one got bad modeline
glisse: pastebin output og LIBGL_DEBUG=verbose glxinfo
reaktae: sorry mate - don't can't seem to be able to get that...
reaktae: how do i set the verbose?
glisse: just type of copy: LIBGL_DEBUG=verbose glxinfo
glisse: in xterminal
reaktae: ih my, an I lost or what?! it still says "If you want to find out why, try setting LIBGL_DEBUG=verbose" Hope this is it though... http://paste.ubuntu-nl.org/52592/
glisse: reaktae: force reinstallation of a package name mesa-dri somethings
glisse: under debian this is libgl1-mesa-dri
reaktae: done that...
reaktae: no more segmentation fault...
glisse: glxinfo says direct rendering yes ?
glisse: if so you are done
reaktae: sorry to still be a pain but no... it still says NO...
reaktae: http://paste.ubuntu-nl.org/52600/
glisse: ls of /usr/lib/dri ?
reaktae: no such f o d
glisse: dpkg -l | grep dri
reaktae: got it sorted - my bad... reinstalled a wrong package...
reaktae: glxinfo now says direct rendering: YES
reaktae: so now i tweak the xrandr?
glisse: well only if you want to change resolution
glisse: xrandr is mostly for changing video mode or when you connect another monitor
glisse: you can configure it with xrandr without restarting xserver
reaktae: i see... how do i put my xorg together then?
glisse: well just forget your xorg.conf it isn't used anymore
glisse: well its used but we prefer to avoid using it
reaktae: hmm... i'm so lost... is my graphics auto-detected now?
glisse: reaktae: well not for the proper driver but everythings else yes
glisse: xorg.conf is a things of the past
glisse: we want to kill it
glisse: sooner the better
reaktae: sounds GOOOD to me...
reaktae: so am i all done here doctor?
glisse: yup
glisse: up and running
reaktae: that's cool...
reaktae: thanks so much!
glisse: yoou welcome
reaktae: you have a nice day/night... wherever you're from....
reaktae: can i donate somewhere or something?
reaktae: man i really appreciate this!
glisse: well just help other if they ever face what you have been through :)
reaktae: sure will! i might even remember most of this... :)
reaktae: take care!
reaktae: hi AGAIN !
reaktae: is there a way to "clone" my laptop's widescreen display and scale it to an external 4:3 using xrandr?
loswillios: reaktae: yes
reaktae: that's good :)
reaktae: thanks
reaktae: loswillios: how would i do it please?
loswillios: I'm using it to have an extended desktop (with xrandr --output VGA-0 --mode 1440x900 --right-of LVDS), no idea about cloning. I suggest reading 'man xrandr'
reaktae: couldn't find it there..
loswillios: --same-as is what you looking for I guess
reaktae: xrandr --output VGA-0 --same-as LVDS ?
agd5f: reaktae: you need ati git master for scaling on panels
agd5f: currently all we support are centered and full scaling modes
agd5f: see xrandr --verbose for the list of output attributes
agd5f: xrandr --output LVDS --set panel_scaler full or center to switch the scaling mode
reaktae: i keep getting "X Error of failed request: BadName (named color or font does not exist)"
reaktae: Major opcode of failed request: 156 (RANDR)
reakta1: deep-frozen after that
reakta1: how do i solve this "X Error of failed request: BadName (named color or font does not exist)" ?
reakta1: when trying to scale my output as full
Kris_Kelvin: hello, anybody got working pal tv-out with radeon 9200 pro (rv280)?
agd5f: Kris_Kelvin: sure
Kris_Kelvin: well, i think i found your blog... so it is fixed only in git and not in 6.7.197 ?
agd5f: yes git only
Kris_Kelvin: ok, so i'll try the git version now. thx
agd5f: sure
umtc: agd5f, was my bios dump acceptable?
agd5f: umtc: yeah looks fine, doesn't have the table I was hoping for though
umtc: the dump could very well have been truncated, i don't know how long it is supposed to be
agd5f: no, I think it just doesn't have it. my laptop has a similar pattern
umtc: does the table have any magic byte pattern or string that i could search for?
agd5f: umtc: one thing you can try is adjusting the tvdac bgadj and dacadj values
agd5f: bits 19:16 and 20:23 in tv_dac_cntl register
agd5f: umtc: take a look at RADEONGetTVDacAdjInfo() in radeon_output.c
agd5f: and try alternate values than what the table specifies
agd5f: see if that helps and let me know which, if any, values help
umtc: i'll go in the other room and start walking the table now
agd5f: umtc: you can try any values for those bytes
umtc: agd5f, i went through the table with no success. i set the root window to 0xfffff8 as a helpful indicator
agd5f: umtc: you can use values other than what is in the table
agd5f: umtc: 0x00340000 for example
umtc: should some of them work or disable tvout? as far as i could see, there was no change at all
agd5f: all should work more or less, it controls how washed out or not the image is
umtc: such as radeon_output->tv_dac_adj = 0x00780000; correct?
umtc: i didn't compare any image quality, i just had the root window set to #FFFFF8 and looked for colored static
umtc: what's minimum/maximum value?
agd5f: 0x00000000 to 0x00ff0000
agd5f: like I said it's two bytes, bits 16:23
umtc: all values in table and ff did not clear the static
agd5f: ok. it's probably not that then
umtc: how did you want me to adjust the tvdac bgadj/dacadj?
paran: tries one more time :)
paran: Anybody who have any tips for debugging radeon suspend/resume problems?
paran: After resume the processor never goes deeper than C2 sleep, unless I restart X, or start a new X-server on another VT.
paran: Ubuntu 7.10 with custom 2.6.24-rc8 and x11-video-ati 6.7.197.
glisse: paran: well dump of register might help
glisse: dump before resume want this doesn't work
glisse: and dump after when it does
paran: what utility can get the dump?
glisse: paran: git clone git://people.freedesktop.org/~glisse/radeondump
glisse: i can't remember the state of the dump for radeon
glisse: i really should clean it up
glisse: paran: you need cmake & libpciaccess
paran: exactly the kind of tool I needed, thanks :)
glisse: paran: instruction is easy to build: cmake . && make
glisse: then ./radeondump -d bad (before suspend when this doesn't work
glisse: ./radeondump -d good after suspedn when this work
glisse: btw this might somethings else than reg like the drm not properly setting vblank interupt or somethings like that
paran: yes, of course. could probably be lots of things, however I don't think it is vlank, I am not seeing any interrupts.
paran: with older drivers and kernel I got the vblank issue after resume, but that is gone now
glisse: paran: well as soon as you got dump i can take a look and help you spot if there is anyguilty reg
paran: glisse: http://paran.mooo.com/slask/irc/radeon/2008-01-19/
paran: glisse: unfortunately I have to go away a while now, but please tell me if you see anything weird. I will check the irc backlog in a few hours.
glisse: paran: ok
glisse: paran: create a file fixc2 which content is what is at http://www.pastebin.ca/864903
glisse: paran: then cold boot to be in the situation where you see no c2
glisse: paran: then ./radeondump -r fixc2 (file you just created)
glisse: paran: if it fix the issue then report it here
glisse: this simply disable second TMDS
agd5f: glisse: BTW, there's a bug in radeondump. pll registers increment by 1 not 4 since they are indirect
agd5f: paran: what version of the radeon driver are you using? there was a bug a while back which left interrrupts enabled all the time after VT switches
paran: agd5f: good timing, just sat down at the computer again :)
paran: glisse: that didn't help
paran: agd5f: I use 6.7.197
ajax: agd5f: i'm reasonably sure, but just to check: max tiled surface size on r500+ is 8k, right?
agd5f: ajax: need to double check on r5xx. I think it may be 4k
ajax: it might be 3968 like r300.
agd5f: right. ought to test that. let me look
ajax: i'm trying to get dual 30" working. it's been a day or two, but i think i concluded that even on one head, as soon as i took the virtual X beyond 3968 things went wacko.
agd5f: ajax: probably the same as r3xx then
airlied: ajax: damn shatter workin yet :)
ajax: airlied: no kidding. would be weird if shatter actually made performance _better_ because you could fit in a tiled surface that way.
paran: agd5f: I had the same issue with ubuntus default (6.7.195), but there the cause for not sleeping is more obvious, I get a interrupt wakeup every vblank
agd5f: paran: I don't recall when I committed teh fix, before or after .197
airlied: ajax: yeah it probably would, we should work out 3D at LCA..
paran: agd5f: if you look at the url I posted earlier I have some powertop output there, with 6.7.197 there are no wakeups but still only c2 after resume
paran: agd5f: I can try latest git
agd5f: paran: do that
ajax: airlied: srsly. i think i've got an rv515 and an rs690 laptop i can bring.
ajax: rs690 should be entertaining
Konekcioni_Kabal: Can somebody help me about x1950 drivers, please ?
Milanche: please somebody
Milanche: omg
Milanche: somebody please help me !!!
Milanche: asap !!!
agd5f: ajax: r5xx is the same as r3/4xx for tile surface setup
agd5f: I just checked
agd5f: ajax, airlied: what's shatter?
paran: agd5f: same behaviour using latest git
agd5f: paran: weird. Not sure what would be the problem off hand
airlied: agd5f: ajax has been trying to get multiple crtc scanout going without zaphod/xinerama layering..
ajax: agd5f: code name i've been using for "allow more than one allocation for the root window"
ajax: tied right into randr so that you can say what the CRTC size limits _actually_ are, and provide a callback or two for allocating them.
agd5f: ajax: sweet. rock on
ajax: will be fun to integrate that with DRI. swapbuffers gets slightly more complicated...
agd5f: heh... yeah
paran: agd5f: ok. good that I have a work around at least, just not very convenient.
paran: agd5f: let me know if you think of something else I can try
agd5f: paran: will do