rx__: robbat2; so i take it you figured out the cause of the brokenness?
robbat2: not entirely, but it's the source of the problem does lie outside of X and the video drivers entirely
robbat2: in a similar vein to your suggestions, I got a friend with an X700 to throw it in an amd64 box, and livecd access showed that the rom was accessible there, then he brought it over and we put it in the G5, and the rom node was empty
robbat2: so i'm digging in drivers/pci/ of the kernel now
rx__: cool
rx__: you're a gentoo dev?
robbat2: ya
rx__: nice.. i use gentoo :)
rx__: i thought they discontinued gentoo/ppc
robbat2: what gave you that idea?
rx__: something i read..
rx__: maybe i read wrong :)
robbat2: funnily, PPC is the architecture i've had the least expose to the intimate details of previously, just because other stuff took my time
rx__: yeah i just checked.. i think i mixed it with ubuntu dropping ppc
rx__: does anyone actually know how the engine/memory clocks are calculated?
rx__: presumably there's a formula for programming the clocks xxx to get yyy mhz
olegfink: hi folks
olegfink: any info/support on tv-out for older radeons? (have 7500 with Theater 100)
mcgreg: do you know who the maintainer of the radeonhd.org page is?
daniels: michael larabel from phoronix
daniels: olegfink: 'should work'
daniels: pretty sure rt100 support has been in for ages now
olegfink: good to know, so now I needs some docs to follow
olegfink: s/needs/need
olegfink: I don't have any idea on how to do this.
olegfink: the best would be to have it be cloned, but side-by-side setup is also okay.
olegfink: I have no luck with both xrandr stuff and some xorg.conf Screen mangle
vidsan: Hi! I need help getting direct rendering for my ATI X1600 æpro 512mb PCIe card in ubuntu gutsy
vidsan: v@vidars:~$ lsmod | grep fglrx
vidsan: fglrx 1489004 36
vidsan: agpgart 35016 2 fglrx,intel_agp
vidsan: v@vidars:~$ glxinfo | grep direct
vidsan: direct rendering: No (If you want to find out why, try setting LIBGL_DEBUG=verbose)
vidsan: OpenGL renderer string: Mesa GLX Indirect
mcgreg: vidsan: sorry, you are wrong here
mcgreg: for fglrx go to #ati
vidsan: ok. thx!
olegfink: so, any gtreater ideas on how do I do this? daniels maybe?
tormod: hi, is the xorg-driver-ati ML meant for user support? Dave's blog suggests it, but mailman description is "driver maintainers"...
mcgreg: good question :)
tormod: With so much bugmail on it, I don't think many users will subscribe to it though,
robbat2: hopefully just those with enough skill to hack at bugs themselves
robbat2: rx__, the problem on my unit looks to be much more like a PPC kernel issue. thus far traced it down to pci_get_rom_size() returning 0 instead of the actual size.
alessandro: any news on radeons r500 laptop support? :-)
airlied: alessandro: no weekend + thanksgiving + lack of hw :)
airlied: I somehow left all my r5xx hw in the office....
airlied: olegfink: did the 7500 not have integrated tv-out?
airlied: olegfink: I don't think we have any plans for external tv-out chips..
alessandro: airlied: so i guess same with alex
alessandro: :-D
airlied: alessandro: well thanksgiving in the US..
alessandro: oh where is he from?
olegfink: airlied: it does have it, it's a RT100 based thingie
airlied: alessandro: somewhere east coast..
airlied: olegfink: but is it integrated with the main chip or external hw? if integrated I think it migt just work..
airlied: useing xrandr to set it up.
olegfink: xrandr says something about S-video
olegfink: maybe if you tell me what to do, I might do it.
olegfink: However, I think it's not integrated. It's Rage Theater 100, which is a separate chip
airlied: xrandr --output S-video --mode 800x600
airlied: would enasble it if integrated..
airlied: olegfink: what si the card naem?
olegfink: than it's not integrated ;)
airlied: olegfink: did that work?
olegfink: airlied: radeon 7500/7000/ve, don't know exactly
olegfink: airlied: nope, I've tried this already
airlied: lspci?
olegfink: 7000/VE
airlied: olegfink: ah definitely not integrated then.
olegfink: so no luck for me?
airlied: olegfink: hmm rv100 should be integrated..
airlied: olegfink: according to wikipedia ;-).. what is the pcid 0x5159?
airlied: and the docs I have here show it has some tv stuff..
olegfink: so?
robbat2|na: airlied, rx__, I'm close on my PPC issue. I got two other cards returning them ROM now, just not my X1900.
airlied: olegfink: I'd say ping agd5f when he comes in, he did all the tv-out support..
airlied: olegfink: I'm a tv-out newb really :)
airlied: robbat2|na: sose endianness problem in rmom code?
airlied: rom code even..
airlied: olegfink: are you doing PAL or NTSC btw?
robbat2|na: no, something more trivial
olegfink: airlied: PAL
olegfink: airlied: http://rafb.net/p/ENoSFs28.html if it's of any help
robbat2|na: enabling the PCI device before trying to use it solved my two other cards, just not the X1900. that might be endianess still, as the other two cards have x86 ROMS
robbat2|na: so you should append your 'rom' dumping instructions to "echo 1 >enable" before trying to enable and read the rom
airlied: olegfink: yes the length reading code could be mesed up..
airlied: robbat2|na: ah on most machines the primary card is enabled..
olegfink: size=128M is complete crap by the way, it's 64M
airlied: robbat2|na: as otherwise you can't see the head :)
airlied: olegfink: yeah you get two windows into the 64MB
olegfink: ah
robbat2|na: airlied, it seems if you are using an OF framebuffer, that is not true
airlied: robbat2|na: ah nice ..
robbat2|na: anyway, i'm going out to buy a new kettle to replace my one that went melty-plastic, and i'll debug more when I get back
robbat2|na: since i've got 2/3 working so far
airlied: olegfink: so thsat chip should have integraed tv-out..
robbat2|na: now just the last push
olegfink: integrated? hm
olegfink: http://gatos.sourceforge.net/theater_out.php says otherwise (from my understanding)
airlied: olegfink: http://lists.freedesktop.org/archives/xorg/2007-October/029477.html
airlied: that thread has some info in it
olegfink: aha, thanks
olegfink: airlied: ah, also how do I expilictly set a resolution for Xorg to use?
olegfink: I've written 1024x768 in all possible places but it still starts up with 1360x768
airlied: olegfink: virtual 1024 768 in the screen section
olegfink: okay, but why? :)
olegfink: where does it get its 1260x768?
airlied: olegfink: do you have a monitor plugged in? or maybe that is just the max size we pick for that card..
olegfink: sure I do :) it's an old 17" CRT CTX
airlied: olegfink: so I think it might work it out from the ddc it gives
olegfink: meh, I have Option "NoDCC" "true" ni Device section :/
olegfink: *DDC
dli: airlied, is the avivotool ready?
airlied: dli: no it'll be tomorrow, I've somehow left all my avivo hw in the office..
Agrajag-: g'day, when using xrandr to enable my second monitor, my machine totally freezes. ubuntu gutsy, using "xrandr --output DVI-0 --mode 640x480 --right-of VGA-0". Am i doing something wrong, or does gutsy just not have the newest versions of things?
airlied: Agrajag-: wshat card?
Agrajag-: my other problem is that i have my main monitor at 1280x1024, and i want the second 1024x768, but when i try to use xrandr to set that, i get "xrandr: screen cannot be larger than 2048x1536 (desired size 2304x1024)"
Agrajag-: R420 [Radeon X800 PRO/GTO]
airlied: Agrajag-: see http://www.intellinuxgraphics.org/dualhead.html
airlied: Agrajag-: for the second q
Agrajag-: ta
airlied: Agrajag-: not sure, I'd suggest trying the git master driver..
airlied: Agrajag-: I think agd5f runs ubuntu on his cards..
Agrajag-: ok, i'll see if i can get the latest git driver going then
tormod: Agrajag-: almost latest git for Ubuntu: https://wiki.ubuntu.com/XorgOnTheEdge
tilman: is the s-video output supposed to work w/ the tip of master and r400?
jahwell: hi
tilman: when i'm trying to enable the s-video output, xrandr tells me that there's no crtc set for the output
airlied: tilman: you need to disable one of other otputs
tilman: and actually setting a crtc using xrandr crashes with BadMatch
airlied: tilman: if you have 3 things plugged in
tilman: haha, i suck
airlied: tilman: I don't think wee have cloning supported properly.
tilman: airlied: i unplugged the 2nd monitor and assumed that was enough %)
tilman: but it was still enabled :)
jahwell: [I'm trying to install ati radeon 9200 SE driver anybody could help me please ^^]
airlied: jahwell: what is wrong with it?
jahwell: well
jahwell: glxinfo | grep direct: X Error of failed request: GLXBadContext :s
airlied: jahwell: what distro?
jahwell: Major opcode of failed request: 143 (GLX)
jahwell: Minor opcode of failed request: 5 (X_GLXMakeCurrent)
jahwell: Serial number of failed request: 18
jahwell: Current serial number in output stream: 18
jahwell: what's distro?
airlied: jahwell: what Linux distribution you using?
jahwell: ubuntu
jahwell: 7.04
airlied: whats wrong with the distro supplied drivers?
jahwell: i dan't know i asked people who use ubuntu they said: x.org comes with the dsitribution just select it and it will works
jahwell: but it does not
airlied: try 7.10 maybe.. the drivers come with the distro..
tormod: jahwell: there is nothing you have to install.
tormod: if I may guess, you have tried to install the proprietary drivers, and it's now messed up?
jahwell: i select x.org... [as they told me] on synaptic but nothing works
jahwell: i just select and install x.org-video-ati
airlied: jahwell: so X starts but GL is broken?
tormod: jahwell: it was installed when you first installed Ubuntu
tormod: jahwell: maybe https://help.ubuntu.com/community/RadeonDriver#head-229f59879c2a2b6c6635d1e189706d97f836b879 can be of help
jahwell: no
airlied: I don't think ou can install ubuntu without X
jahwell: and i've install xorg-driver-fglrx
airlied: jahwell: you see you need to remove that..
Magnade: airlied: you can if you use a server cd or perhaps the alternate
Magnade: airlied: desktop install of ubuntu tho you do get x without a choice
jahwell: [thanks for the link]
airlied: Magnade: I was hoping nobody would willing give a newbie a server CD :)
jahwell: i've got to go thanks for your help :-D
jahwell: nn
Agrajag-: tormod: cheers - not sure which part of the page i'm mean to follow to get an updated radeon driver though?
tormod: Agrajag-: radeon driver = -ati package
Agrajag-: ah i see ta
tormod: Agrajag-: technically the ati driver package includes radeon+r128+whatnot
Magnade: airlied: you can always hope but alas they newbies go and find these things by themselves :\
robbat2: airlied, got a moment for an atombios question? (save me digging in the source code)
airlied: robbat2: shoot.
robbat2: the v86d project was stuff to run x86 ROM emulated under other arches, is the atomBIOS x86-specific internally, or is it actually safe across arches?
airlied: robbat2: atombios is just a concept..
airlied: robbat2: there is no code in it.. just scripts..
robbat2: ah, great
airlied: robbat2: the scripts are exectuted by the driver..
robbat2: i was wondering because of the BIOS in the name ;-)
airlied: robbat2: the atombios scripts are stored in the bios :)
airlied: robbat2: and it also exeecutes them..
airlied: robbat2: at boot time
robbat2: they are more device-specific microcode then, rather than traditional x86-style bios ROMs
airlied: robbat2: of the rom is still an x86 style bios
airlied: robbat2: it just has an interpreter for the scripts
airlied: robbat2: so the BIOS and drivers use the same engines from the same source base..
airlied: robbat2: so my worry with Mac cards is that maybe Apple didn't bother using it..
airlied: robbat2: and just leave all the hardwork in the driver.
robbat2: that would suck
airlied: robbat2: well when you get a copy of the BIOS we can look inside :)
robbat2: googling I see this too: http://lists.opensuse.org/radeonhd/2007-10/msg00271.html
airlied: robbat2: must be a rom enable line..
airlied: robbat2: there is a bit in a register.
airlied: robbat2: I wonder.. grab avivotool from git://people.freedesktop.org/~airlied/radeontool.git avivo branch
airlied: and avivotool regmatch 0x4c
airlied: see if bit 2 is set
robbat2: one sec, let me get my G5 up again. I had it turned off while I was out, and I want to add another PCIe card I picked up from a friend
robbat2: just hope I don't trip my breaker
robbat2: since i'll have it maxed with PCIe now
airlied: robbat2: just let me know here I may have to got out for a while..
robbat2: Mac ATI X1900, Mac Nvidia 6600, x86 ATI X700, x86 sata_sil24 controller
robbat2: will do
Agrajag-: where can i find docs on what options i need for dual head stuff in xorg.conf? i'm trying to set up my xorg.conf statically (the newer radeon driver appears to not freeze my machine now) following that link airlied gave me earlier (http://www.intellinuxgraphics.org/dualhead.html) but can't seem to get things quite right
robbat2: airlied, no joy on that register 0x4c
robbat2: i'm trying some other ROM debug stuff per benh again now
airlied: Agrajag-: oh make sure you use radeon output names instead of Intel ones.
airlied: Agrajag-: run xrandr to get your output names
airlied: robbat2: what was in 04c?
airlied: 0x4c even.
robbat2: airlied, no output
airlied: robbat2: from avivotool? not good..
airlied: robbat2: does avivotool regs give anything?
robbat2: you said avivotool was the binary name, but the avivo branch of radeontool only gives a radeontool binary
airlied: robbat2: try make avivotool :-)
airlied: robbat2: I need to do some work on it yet..
robbat2: there is no target for avivotool in the makefile
airlied: robbat2: wierd my avivo branch I just pulled has it
robbat2: ah, stupid git
robbat2: git clone -o avivo
robbat2: with the URL on the end
airlied: ah you need to change branches
robbat2: used to give me the right branch directly
robbat2: ok, the avivo tool just skips over the X1900 card and goes to the X700 card where the ROM was readable
airlied: robbat2: care to hack in a deveice id check around 1345
robbat2: yeah, one sec
rx__: airlied; so eh.. no idea how engine/memory clocks are calculated?
airlied: rx__: for powerplay?
airlied: rx__: we get them from a table in the BIOS..
airlied: rx__: normally the bios programs them into the card
rx__: yep powerplay
rx__: well.. in theory we should be able to underclock/overclock as we see fit
rx__: do you know if there's a formula aside from using the fixed values in the bios?
airlied: rx__: it is chip specifc..
airlied: rx__: depending on how fast the attached RAM chips are..
rx__: ah.. damn
airlied: you would need datasheets for the chips etc.. the tolerances may vary etc.
rx__: okay that makes sense
airlied: the engine clocks might me more tuneable.. there are overclocking tools for windwows..
rx__: unfortunately no one released source for windows apps
airlied: rx__: you could dump the regs before/after
rx__: yeah i think i figured how to program the clocks
rx__: i guess the only way i can come up with a formula then is to grab all fixed clock values from bios then use some interpolation formula to allow user setting underclock/overclock
robbat2: # ./avivotool --debug regmatch 0x4c
robbat2: Found card 1002:7240 (30000)
robbat2: r580 (Radeon X1900 MAC)
robbat2: Radeon found. Base control address is f7fee000; base framebuffer address is efe64000.
robbat2: reading 0x4c (4c) is 00000001
robbat2: 0x4c 0x00000001 (1)
airlied: okay so it isn't disabled there..
robbat2: those addresses look weird
airlied: what bout 0x5030?
robbat2: compared to the ones in lspci
airlied: yes I thought there might be another enable..
robbat2: reading 0x05030 (5030) is b0020000
robbat2: 0x05030 0xb0020000 (-1342046208)
airlied: okay so it isn't enabled in lspci?
robbat2: Region 0: Memory at b8000000 (32-bit, prefetchable) [size=128M]
robbat2: Region 1: I/O ports at
robbat2: Region 2: Memory at b0000000 (32-bit, non-prefetchable) [size=64K]
robbat2: Expansion ROM at b0020000 [disabled] [size=128K]
airlied: so 0x5030 just mirrors the pci space.
robbat2: so b0020000 is there
robbat2: I was just wondering how it was geting the base control addr
rx__: wonders where to start with this atombios stuff
airlied: rx__: to do powerplay? first parse the atombios tables..
rx__: i'm guessing there's code already that parses them?
airlied: rx__: not for powerplay ones..
airlied: rx__: there is code in radeon_atombios.c to parse some tables..
airlied: the atombios.h has the table info in it
robbat2: airlied, http://rafb.net/p/czMRAy20.html
robbat2: that adds the ids, and fixes up the --skip handling for avivotool
rx__: airlied; thanks for the pointers :)
airlied: robbat2: thanks pushed it to repo
robbat2: i need a nap, then dinner and more hacking
Magnade: is the reason for lack of render support on r300 due to lack of docs?
airlied: Magnade: lack of time..
airlied: Magnade: I started looking at it for the 10th time this weekend..
airlied: Magnade: and actually wrote code which succeeds in crashing the card straight away..
airlied: Magnade: so another good start..
Magnade: nice...
Magnade: i doubt i got the knowledge to do the coding myself
Magnade: im willing to look at it a bit tho
airlied: Magnade: I'll probably not look at it again for ages.. the hack so far is on the r300-composite branch of git://people.freedesktop.org/~airlied/xf86-video-ati
airlied: Magnade: but it needs a lot lot more work.
Magnade: what about documentation?
Magnade: unless its in the code
airlied: Magnade: hehe.. documentation hehe..
airlied: Magnade: I would need to make code that worked to document it..
airlied: Magnade: that code is take stuff from Mesa and try to make it do what I think it should do.
airlied: Magnade: if I can get it to the point where the chip doesn't lockup thats a big win.. then I can fix the rendering..
Magnade: so more than likely pointless for me to look at it as it requires lots of knowledge of the chipset?
airlied: Magnade: it requires knowledge of how the chip is setup and of how fragment shaders work..
airlied: Magnade: which is all contained in the mesa r300 driver..
Magnade: so render uses part of the 3d system then?
airlied: Magnade: yes it just uses the 3D engine to do the operations..
Magnade: learn something everyday...
Magnade: well i got it checked out ill skim it over i doubt ill see anything but never know
Magnade: heh well over my head :\