Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

RadeonHD IRC Logs For 2008-11-01

Search This Log:

Jin^eLD: hi
Jin^eLD: I found exactly one google post regarding this error message: "rhdAtomGetDDCIndex: GPIO_DDC Index 4 exceeds maximum 4" and it did not offer a solution :> any hints on what to do?
MostAwesomeDude: Try radeon? :3
Jin^eLD: what?
MostAwesomeDude: radeon, from the xf86-video-ati package. :3
Jin^eLD: oh
Jin^eLD: thx
MostAwesomeDude: I have no idea about your error, but somebody will come by and see it eventually, I bet.
Jin^eLD: theone I have is a mobility radeon hd 3500 btw (or was it 3400?)
MostAwesomeDude: Hm.
MostAwesomeDude: Might be a card quirk, or it might be a driver bug. Hard to say for sure.
Jin^eLD: the funny thing is - I had it working today
MostAwesomeDude: Hm.
Jin^eLD: I installed F9, but then I screwed up at some point with yum, and decided to reinstall clean again
MostAwesomeDude: Ah.
Jin^eLD: and after that - does not work anymore
MostAwesomeDude: And I bet that radeonhd got updated. :3
Jin^eLD: not at first
Jin^eLD: but indeed, let me double check that
Jin^eLD: I do not think that I can figure this out
Jin^eLD: I mean, if the previous install had a different version or not
Jin^eLD: fglrx worked out, not nice but it works
Jin^eLD: had to downgrade X11
NIIcK: hello
NIIcK: am i in the right place to ask some questions about the radeonhd driver?
the-me: NIIcK, sure
NIIcK: cheers
NIIcK: I have an Asus A7D laptop with an ATi Mobility Radeon x700 video card
NIIcK: am running FC9.92
NIIcK: x64
the-me: libv, hm are there any similar reports where radeonhd may damaged the GPU (for example x1250 mobility)?
NIIcK: i've installed the latest radeonhd drivers to get hardware acceleration but when i set the radeonhd driver in xorg.conf and restart X the screen goes black
NIIcK: and it does not come out of that :) the system doesn hang since i can reboot it with ctrl+alt+del
NIIcK: i've tried to install the latest catalyst ... doesn't work
NIIcK: and with the default radeon driver i get no hardware acceleration
NIIcK: any ideas would be appreciated
NIIcK: thank you
uspjoe: NIIck: Hardware ex. is under development, but you can use ati driver http://ati.amd.com/support/drivers/linux64/linux64-radeon.html
uspjoe: NIIck, I see x700 ist RV410 / M26 chip (http://dri.freedesktop.org/wiki/ATIRadeon), and this chip is in radeon driver not in radeonhd
g3d: hello, does "DONE" in XRandR for R700 for radeonhd mean that I can rotate the screen now?
g3d: its not working for me...
rindolf: Hi all.
tomsh: i have installed xorg 7.4 from debian experimental repository
tomsh: i have a Radeon HD 3450
tomsh: but my lcd at start remain in standby mode (only if use radeonhd driver)
tomsh: if i use radeon driver works?
tomsh: works!
tomsh: any suggestion?
andy_123: hi all, can anyone tell me how to get xvideo working or how to see if it is enabled at all?
purplep: Anyone knowledgable about getting radeonhd to initialize a secondary video card?
purplep: I'm having a problem that I filed as
purplep: The basic issue is that by default radeonhd can't find the second card's BIOS
purplep: I get the error "RADEONHD(1): rhdAtomGetTables: No AtomBios signature found"
Griffon26: egbert_: have you had a chance to take a look at the log I sent to the list? (Asus EAH3870 X2, MC not idle on mode switch)
egbert_: Griffon26: i'm just about to push a change which might address this issue.
egbert: Griffon26: done. please give it a try.
purplep: egbert: I'm trying to figure out a problem with initializing a secondary card. You wrote something about this in the IRC logs from 2008-10-16:
purplep: "about initializing a secondary card: you need to do some voodoo to make the PCIROM visible."
purplep: "i think i've described on the wiki page what to do"
purplep: But I couldn't find anything on or its subpages, do you a pointer to the info?
egbert: purplep: there are basically 2 things to do:
egbert: boot with pci=rom
egbert: then do: echo 1 > /sys/bus/pci/device//rom
purplep: egbert: Yeah, I tried that, I found the info in a bugreport on bugs.freedesktop.com. But that didn't help me.
egbert: there is a report in bugzilla where this appearantly didn't work either.
egbert: right.
purplep: That's me, assuming you me the bug that I reopened yesterday
egbert: i haven't tried all to recently.
purplep: After booting with pci=rom, the rom is still disabled
egbert: the dual card support seems to horrifically broken.
purplep: And the echo command didn't activate it
purplep: :-(
egbert: but i don't think it's a driver issue.
purplep: You mean it's xorg-x11-server?
egbert: part of it may be related to the libpciaccess desaster.
purplep: I guess I'm not familiar with that disaster.
egbert: i cannot really try as i don't have a box with 2 peg slots.
purplep: I actually tried manually enabling the PCI ROM with setpci, then radeonhd found the ROM but complained there was no videoRAM
egbert: purplep: what do you get when you do the steps described?>
purplep: The two steps you described made no difference in the behavior of radeonhd
egbert: purplep: did you report this bug i've just closed?
egbert: the other setpci command is wrong.
egbert: this enables the card.
purplep: I reopend bug
egbert: then the driver doesn't know that it should call InitASIC
Griffon26: egbert: ok, thanks, I'll try it tomorrow as soon as I have access to that computer again.
purplep: Yes, let me read your comments
purplep: So radeonhd can post the secondary card by calling InitASIC?
egbert: purplep: actually libpciaccess should take care of reading the pci ROM without requiring the echo > /sys/bus... stuff.
purplep: Is it possible the libpciaccess changes have caused radeonhd to be confused and never call InitASIC?
egbert: that's also possible.
purplep: Just to clarify my bug report filed yesterday, I only tried the setpci "magic" after a normal boot, and a boot with the "pci=rom" and "echo" command didn't work.
egbert: purplep: have you tried without the setpci .. COMMAND=3 ?
purplep: So to try to get this working, I should look at why libpciaccess isn't probably enabling the card, and why radeonhd isn't calling InitASIC?
purplep: Yes, yes, that was the first thing I tried:
purplep: (1) Normal boot, dual card xorg.conf gives "rhdAtomGetTables: No AtomBios signature found"
egbert: this 'echo 1 > ...' is a magic thing to enable reading the ROM.
purplep: (2) Boot with "pci=rom", and "echo 1 > /sys/bus/pci/device//rom" still gives "rhdAtomGetTables: No AtomBios signature found"
egbert: after that reading out this /sys/... file should just work.
purplep: After doing the "echo", lspci still says the ROM is disabled.
egbert: i think you should be able to verify this with dd
purplep: I tried that, dd gives 0 bytes
egbert: hrm. so it's broken in the kernel?
purplep: If you say so, I don't know who is supposed to do it. :-)
purplep: This is kernel
purplep: what is the "pci=rom" part supposed to do?
purplep: All I notice is that it changes the address of the ROM slightly, but it is still disabled.
purplep: I ran into a report that I believed said somehow had to setpci COMMAND=2.2 in order to get the /sys filesystem working, does that make any sense?
egbert: it's supposed to make the rom readable thru the /sys/bus/.../rom file.
purplep: (not sure if I remembered that right)
egbert: without it this file will not be there.
purplep: That's odd, because my current boot is without "pci=rom", and the /sys/bus/pci/../rom file is still there
egbert: purplep: if you do the pci=rom and the echo thing and then etpci -s 02:00.0 ROM_ADDRESS=fbdc0001
egbert: what do you get then?
egbert: purplep: ok. so the kernel changed again.
purplep: You mean in terms of reading the rom?
egbert: yes.
egbert: try dd.
egbert: but don't do setpci -s 02:00.0 COMMAND=3
purplep: I currently haven't booted with "pci=rom", but let me try the "echo" and the "setpci ROM_ADDRESS" and then dd.
egbert: ok.
purplep: No joy.
purplep: I tried the "echo 1", and checked dd--0 bytes. Then I did the "setpci ROM_ADDRESS" and checked dd--0 bytes.
purplep: Should I try the COMMAND=2.2 idea?
purplep: As you can tell, I don't really know anything about PCI.
egbert: purplep: yes, please try.
purplep: OK, that's "setpci -s 02:00.0 COMMAND=2.2"?
purplep: That gives me 'setpci: Invalid value "2.2"'
purplep: Is it supposed to be 2:2?
egbert: purplep: if you do COMMAND=3 you can read it?
purplep: OK, COMMAND=2:2 worked, after that dd works, and the copy of the ROM I get matches the copy I got after doing the COMMAND=3
purplep: So should I try radeonhd now, with the hope that it can now find the BIOS, and since I haven't actually enabled the card, it will realize it needs to do InitASIC?
purplep: I'm going to give it a try, worst case I'll have to reboot so I may be a little while before returning.
purplep: OK, that didn't work, radeonhd behaved the same as when I had done the "COMMAND=3", that is it failed with "No Video RAM detected."
purplep: I tried it with "IsolateDevice" for the secondary card only, since I'm currently running X on the primary card to do this. Does that matter, do I need to exit X and try it with a dual card xorg.conf?
egbert: purplep: if you do COMMAND=2?
purplep: Should I do that now, even though I've already done COMMAND=2:2, or do I need to reboot and try COMMAND=2 fresh?
egbert: purplep: sorry, i overlooked that you did COMMAND=2:2 already.
purplep: What is the difference between COMMAND=2 and COMMAND=2:2?
egbert: purplep: COMMAND=2 will only enable MEM not IO
purplep: That's odd, I did the COMMAND=2:2, and lspci -v does show the memory enabled, but shows: "I/O ports at d000 [disabled] [size=256]"
egbert: purplep: not exactly sure. possibly it only writes bit 1
purplep: Oh, the MEM vs IO is the difference between COMMAND=2 and COMMAND=3
purplep: That makes sense.
egbert: purplep: yes. if this is still detected as a primary card we need to force asicinit in the driver.
purplep: Well, the test I just did was with IsolateDevice, so it was the only card X saw, is that a problem?
purplep: I can do a test with a dual card xorg.conf, I just have to leave the IRC chat since I'm using an X client for IRC
purplep: How does radeonhd decide if a card is primary or secondary?
purplep: Is libpciaccess supposed to do its part before radeonhd gets to "Getting BIOS copy"?
purplep: I'm just trying to figure out (a) what needs to be done PCI wise and (b) who is supposed to do it :-)
egbert: purplep: the problem is that the behavior of so many components change all the time.
purplep: components = software components within linux?
purplep: libpciaccess changed, and the kernel changed?
egbert: could be either one.
egbert: well, i could use a vram check to see if the card is posted.
egbert: maybe i'm going to do this.
purplep: If you do that, I will certainly give it a try.
purplep: I'm going to try the dual card xorg.conf with the COMMAND=2 on a fresh boot.
purplep: Unfortunately, I have to go now, what is the best way to followup with you egbert, is it via reopening the bug ?
egbert: purplep: i will add this eventually. it's not really a bug.
egbert: or someone elses bug i need to work around.
egbert: purplep: best you check the commit logs going by.
purplep: So you don't want me to reopen .
purplep: OK, if the COMMAND=2 doesn't work for me, I'll try looking at where radeonhd does InitASIC and see if I can hack it a little. :-)
purplep: Thanks again for you help
purplep: bye
purplep: egbert: if you have anything to add, I'll check the IRC logs when I return.
purplep: bye
Jin^eLD: guys, what is the suggestion driver version / xorg version combo for the Mobilitiy Radeon HD 3400 Series?
Jin^eLD: xorg-x11-drv-radeonhd-1.2.3-1.3.20081029git does not work for me :P
Jin^eLD: lower left part of the screen starts getting white
Jin^eLD: then I kill X
foxbuntu: Hi all, I am having problems using the radeonHD driver with my HD3200/780G card, all video rendering is very slow
MostAwesomeDude: foxbuntu: Neither radeon nor radeonhd support acceleration for newer cards yet.
foxbuntu: MostAwesomeDude, ahh...well that would explain things
foxbuntu: MostAwesomeDude, thanks
Jin^eLD: MostAwesomeDude: is mobility hd 3400 a "newer" one or not? and should it work without acceleration?
MostAwesomeDude: Jin^eLD: Yes, yes.
MostAwesomeDude: HD 2000 and newer have no acceleration on either driver.
Jin^eLD: well, for me it does not work at all, unfortunately
Jin^eLD: and fglrx seems to be eating quite some cpu time in Xorg
Design_Stance: hi all
Design_Stance: anyone here?
Design_Stance: um
dthsqd: Hello
dthsqd: I would like to debug an issue with radeonhd
dthsqd: Any way I can have it print more output to Xorg.0.log?
dthsqd: I have xf86-video-radeonhd 1.3.2, but Xorg is not vanilla (it is Gentoo's Xorg, probably patched).
dthsqd: X just crashed on me a few seconds after my last message (hence the pidgin logoff).
dthsqd: Though it happens only when starting firefox
dthsqd: thunderbird too...