mikkoc: is radeonhd from git master supposed to work with drm in 2.6.32-rc3? r600 chipset
mikkoc: i get [drm:r600_do_init_cp] *ERROR* Need gart offset from userspace
brbrbr: warmly greeet everyone
amarsh04: builds radeon 1.30 source
amarsh04: radeonhd rather
amarsh04: still only 2 fps on etracer though
amarsh04: with the rs780 and radeonhd 1.3.0, is there any way to avoid:
amarsh04: (EE) AIGLX error: dlopen of /usr/lib/dri/r600_dri.so failed (/usr/lib/dri/r600_d
amarsh04: ri.so: cannot open shared object file: No such file or directory)
amarsh04: (EE) AIGLX: reverting to software rendering
amarsh04: (II) AIGLX: Loaded and initialized /usr/lib/dri/swrast_dri.so
chithead: amarsh04: if you want r600 dri, you need to compile mesa with r600 support
amarsh04: thanks chithead
amarsh04: strange, Debian has mesa 7.6.1 but r600 support is not compiled in
amarsh04: rebuilding mesa with r600 support...
amarsh04: pardon the flood, any ideas on this error when building mesa with r600 support:
amarsh04: r600_cmdbuf.c: In function ‘r600_cs_emit’:
amarsh04: r600_cmdbuf.c:331: error: storage size of ‘cs_cmd’ isn’t known
amarsh04: r600_cmdbuf.c:332: error: array type has incomplete element type
amarsh04: r600_cmdbuf.c:353: error: ‘RADEON_CHUNK_ID_IB’ undeclared (first use in this function)
amarsh04: r600_cmdbuf.c:353: error: (Each undeclared identifier is reported only once
amarsh04: r600_cmdbuf.c:353: error: for each function it appears in.)
amarsh04: r600_cmdbuf.c:358: error: ‘RADEON_CHUNK_ID_RELOCS’ undeclared (first use in this function)
amarsh04: r600_cmdbuf.c:373: error: ‘DRM_RADEON_CS’ undeclared (first use in this function)
amarsh04: r600_cmdbuf.c:332: warning: unused variable ‘cs_chunk’
amarsh04: r600_cmdbuf.c:331: warning: unused variable ‘cs_cmd’
mikkoc: is radeonhd supposed to work with drm from 2.6.32-rc3 on r600?
mikkoc: I get [drm:r600_do_init_cp] *ERROR* Need gart offset from userspace
legume: mikkoc: it won't work with kms you need radeon for that. Should work with kms disabled.
mikkoc: legume: kms disabled
legume: mikkoc: Not sure then, I haven't tried it with recent kernels.
amarsh04: ah, installing mesa as per http://www.x.org/wiki/radeonhd:experimental_3D seems to help
amarsh04: ok, slightly further ahead, get AIGLX: Screen 0 is not DRI2 capable, 2d stuff comes up fine, but glxgears fails on startup and etracer gives black screen
amarsh04: also has trouble with non-existant DVI-D monitor:
amarsh04: (II) RADEONHD(0): EDID for output DVI-D_1
amarsh04: [mi] EQ overflowing. The server is probably stuck in an infinite loop.
karlis-repsons: Can anyone post a link with some comparisions of radeon and catalyst performances on some recent Radeon HD card(s)?
edman007: karlis-repsons, well tremulous gets over 30fps on my system, and ut2k4 is too slow but runs
edman007: but fglrx won't run on my system, so i can't tell you how it compares
karlis-repsons: And things like crashes frequency and power consumption variations with those divers?
bridgman: karlis-repsons; there's a recent Phoronix article comparing performance
bridgman: mostly 2d though, not much 3d
karlis-repsons: bridgman, perhaps you have the link?
bridgman: looking for it
bridgman: http://www.phoronix.com/scan.php?page=article&item=amd_r600_r700_2d&num=1
karlis-repsons: bridgman, Thanks!
legume: mikkoc: I just tried hd with current gits - drm-next,mesa,mesa/drm (lib) and ddx and it works OK with modeset=0
mikkoc: legume: r600?
legume: mikkoc: yes
legume: mikkoc: For me boot option radeon.modeset=X doesn't work, so it may be worth checking kms is really disabled.
mikkoc: i googled that error and it's related to rs780, which i also have, but the message is old
mikkoc: legume it is, im sure
legume: mikkoc: Ahh, I guess it's rs780 then.
mikkoc: http://lists.opensuse.org/radeonhd/2009-01/msg00198.html
mikkoc: its old
mikkoc: but i guess it hasnt been fixed yet :/
legume: mikkoc: I would have thought so - of course I am using drm-next branch, so not actually testing the same kernel as you.
edman007: karlis-repsons, to me it seems a bit unstable, but not significantly worse than what i had with fglrx in the past
edman007: and i'm running git
karlis-repsons: Anyone knows about *multiseat* X mode with radeon{,hd}? I had it with nvidia, but it was crashing after hibernation, so I have up then...
chithead: karlis-repsons: what is multiseat x mode? presently you have to either start several x servers on different cards, or start xephyr instances on different heads of one server
karlis-repsons: Yes, the same. I was starting several X servers.
karlis-repsons: I'm interested in setting that up again with hibernation possible...
rozzo: hello
rozzo: i have a big problem with my radeonhd
rozzo: driver
rozzo: can anyone help me?
bridgman: rozzo; what is the problem ?
rozzo: the log of Xorg say this
rozzo: (WW) RADEONHD(0): DRMCPIdle: DRM CP IDLE returned BUSY!
rozzo: and the xorg freeze
bridgman: I've seen that reported a few times, not sure if there is a resolution yet
rozzo: only mouse moving
rozzo: mmmm
rozzo: ok
bridgman: let me do some searching on the mailing list
rozzo: i have search and recompile but with nothing result
bridgman: IIRC the most common discussion involves someone asking if the radeon driver has the same problem then never hearing back
bridgman: which doesn't help the devs much ;)
rozzo: bridgman, i have a ati radeon Xpress 1270 the radeonhd is the right driver?
bridgman: yep, either driver should be fine
bridgman: are you running the latest version of radeonhd ? there have been a lot of changes recently so first step should be picking up latest from git
bridgman: a 1.30 release came out yesterday
bridgman: maybe use the pastebin site (or similar) to post your xorg log and dmesg output
rozzo: yes
rozzo: the latest
rozzo: bridgman, ok
rozzo: bridgman, my xorg configuration
rozzo: http://nopaste.com/p/aUejKk1Aj
bridgman: it's really the xorg log and dmesg output we need
rozzo: ok but now i have reboot and i have try to disable acpi
rozzo: bridgman, http://nopaste.com/p/aknxEMc6u
rozzo: my xorg.log
rozzo: http://nopaste.com/p/afVkfkzRW
rozzo: my dmesg
bridgman: ok, there's one questionable message in dmesg output, don't remember seeing this before (but that could just mean I haven't looked at an rs690/740 dmesg before)
bridgman: [drm:radeon_set_igpgart] *ERROR* Unable to use IGP GART table size 32768
rozzo: ok
rozzo: now?
rozzo: what i have to do?
bridgman: do you still get the problem when Option DRI is set to 0 ?
bridgman: something like Option "DRI" "no" in device section
rozzo: mmm
rozzo: now try
bridgman: I didn't actually see that "BUSY" message in your xorg log, should I ?
rozzo: i reboot the machine
rozzo: but if you want i reenable the error
rozzo: whait
rozzo: wait*
rozzo: in dmesg see this
rozzo: [drm] Setting GART location based on new memory map
rozzo: [drm:radeon_set_igpgart] *ERROR* Unable to use IGP GART table size 32768
rozzo: [drm] Loading RS690/RS740 Microcode
rozzo: [drm] Num pipes: 1
rozzo: [drm] writeback test succeeded in 1 usecs
rozzo: i try to reproduce the problem
bridgman: yeah, that's the message I posted above
bridgman: not sure if it's normal or not though
rozzo: bridgman, i can't reproduce, seems random
rozzo: but i think is a blackgreen acpi
rozzo: i suspect
rozzo: now i try to disable acpi for a few of days
rozzo: and retry
rozzo: i understand is a 3d problem
rozzo: bridgman,
rozzo: now crash zorg
rozzo: prepare to get Xorg log
rozzo: bridgman, thanke this http://nopaste.com/p/aT06OZOOc
rozzo: the xorg.log
rozzo: and take this dmesg
rozzo: bridgman, this is dmesg
rozzo: http://nopaste.com/p/ah8t0ULcab
rozzo: bridgman, are you live?
bridgman: yeah, just reading the log
rozzo: bridgman, there are news, now i re see dmesg and get this
bridgman: you mentioned "reenabling the problem", how were you doing that ?
rozzo: http://nopaste.com/p/aELUR8zkU
rozzo: see the last line
rozzo: after i killing X
rozzo: mmm how i reproduce the problem?
rozzo: i start youtube video , mplayer and muc more
rozzo: i close and reopen
rozzo: when i run a youtube video the xorg crash
rozzo: sorry for my bad english
bridgman: no problem, you're doing fine
bridgman: can you try with Option "DRI" set to "0" ?
bridgman: it'll be slow, just want to see if it crashes
rozzo: ok i try now reboot the machines, i hope to solve this problem for all user
rozzo: :D
rozzo: bridgman, i put Option DRM 0
rozzo: in the device section
rozzo: is rigth?
rozzo: or DRI?
rozzo: Option "DRM" "0"
legume: rozzo: it's DRI
rozzo: ok in the device section or module?
legume: rozzo: device section
rozzo: ok
rozzo: now init 5
rozzo: and try
rozzo: direct render yes
rozzo: software rasterize
rozzo: is right?
bridgman: sounds right so far
rozzo: gh now is very slow
rozzo: gh
bridgman: what is gh ?
rozzo: is a noise
rozzo: mmm similar to burp
rozzo: heheheh
bridgman: ahh, not a program ;)
bridgman: anyways, can you run like that for a while and see if it crashes, then post to the radeonhd mailing list with the results
bridgman: doesn't seem like any of the devs are on right now, so mailing list is best option
rozzo: mmm i think is a devs can try to debug in live this proble , i think is a best solution
float: im trying to compile, but when compiling rhd_acpi.o it fails: rhd_acpi.c: In function ‘rhdDoBacklight’:
rozzo: but i try without acpi
float: rhd_acpi.c:70: error: dereferencing pointer to incomplete type
rozzo: the error seem referred to blackligth
rozzo: and happen whe dri is enable
rozzo: i remember
bridgman: rozzo; the difficulty is making the problem happen on the developer's system
bridgman: usually it works on the dev's system and doesn't work on yours so debugging is hard
rozzo: yes
float: tried on both latest from git and 1.3.0
rozzo: infact i think is a cool solution to connect with ssh on my pc
rozzo: and try
legume: rozzo: Maybe you could look in bios and see if there is a setting for video size - it it's small set it bigger.
rozzo: legume, are you italian?
rozzo: legume, you think is a problem resolution?
rozzo: if i set the modes the problem go away?
legume: rozzo: no English - just something to try, I haven't got a clue if it will help the gart error or not.
rozzo: now i have set modes and i re enable drm
rozzo: now try to reproduce error
rozzo: legume, if you interest i have a log of gart
legume: rozzo: I saw the error - I don't know if changing video mem size in bios will fix it or not. It's just something you could test.
rozzo: legume, what the correct dimension?
rozzo: 64?
legume: rozzo: Don't know really - if you have enough RAM try 128 or 256.
legume: rozzo: You can always change it back - I was just thinking it may be set to something too small by default.
rozzo: legume, i remember to set video memory in xorg, is helpness?
legume: rozzo: I wouldn't bother with xorg.conf - the error was in dmesg - so before X started.
rozzo: ha yes
rozzo: true.... but i try to set 32 mb and nothing
legume: rozzo: I would remove that from xorg.conf in case it causes problems.
rozzo: legume,
rozzo: i don't know how to set this
rozzo: i try to switch to 256 to 32
rozzo: but the problem remaining
legume: rozzo: You mean you rebooted and went into bios setup and changed it to 256?
rozzo: no
rozzo: i reboot and change to 256 to 32 mb
rozzo: shaed memory
legume: rozzo: If it was already 256 and you made it smaller, I would change it back to 256. Then as it still doesn't work file a bug.
rozzo: this is a bug?
legume: DRI not working is a bug.
legume: well - maybe a bug - the gart error also.
rozzo: mmm i have a radeonhd 1.3.0
rozzo: but if i launch glxgears
rozzo: is work vell
legume: it will still work using software, but more complicated 3D and video will be too slow.
rozzo: legume, if i set a modes and the bug not reapperar
rozzo: peappear i reatry to use 1.2.5
legume: rozzo: I don't know what you mean by "set a modes "
rozzo: the mode set of xorg
rozzo: modes "1440x900"
legume: rozzo: Hmm strange, I haven't got a clue what's going on.
rozzo: in the device section you can set the default mode
rozzo: ops not in device section
rozzo: screen
rozzo: section
legume: rozzo: Yea - I just don't know why that fixes your problem.
legume: Anyway I've got to go now - bye.
rozzo: thanks
rozzo: at all
rozzo: bye