famicom: lo
famicom: (EE) RADEONHD(0): Cannot map connectors on an unknown card!
famicom: I was told by the internet that i should come here and bitch about it
famicom: can i just edit the source and addmy own PCIID
SunTzuTech: famicom: so if you compiled your own radeonhd source, cd utils/conntest, and do an rhd_conntest 0x:yy.z for the graphics device shown by lspci
SunTzuTech: do that for each connection (dvi, vga, etc)
famicom: ok
famicom: If i do so
famicom: where can i submit the results?
SunTzuTech: you might want to dump the bios using rhd_conntest -d 0x:yy.z
SunTzuTech: to radeonhd@suse.something. cr*p
SunTzuTech: lemme find it
famicom: okiez
yangman: before you do that: driver version number, hardware model, and what's the exact problem?
yangman: a single line of log isn't exactly helpful
famicom: 02:00.0 VGA compatible controller: ATI Technologies Inc RV610 video device [Radeon HD 2400 PRO] (prog-if 00 [VGA controller])
famicom: let me just find the version
famicom: ii xserver-xorg-video-radeonhd 0.0.1+git20070918-1ubuntu1 X.Org X server -- AMD/ATI r5xx, r6xx display
yangman: 20070918? that's ancient
yangman: try upgrading to the latest version available first
famicom: ok
famicom: Are there any debs out ther
yangman: afaik, it should be in one of the repositories
SunTzuTech: lol
yangman: I'm not familiar with ubuntu, though
famicom: and wow
famicom: i just realised
famicom: that IS ancient
SunTzuTech: It compiles really easily on ubuntu
famicom: yeah, but i prefer to keep my little box to be nice and clean
SunTzuTech: I have bog stock ubuntu 7.10 and used the package manager to get git
SunTzuTech: installed radeonhd in /usr/local/..... and use a symlink to switch it back and forth...
famicom: thats one way to do it
famicom: Or, i could just put some *effort* into it and package it
famicom: but that would require me to have an attention span longer than 20 seconds
famicom: and i'm all out of weed for that matter
famicom: humm
famicom: still no luck
SunTzuTech: 2400 should be doable out of radeonhd-1.2.1 or git
SunTzuTech: if you do git, it's a ./autogen.sh (though you need pciutils to get lspci and libpciutils) make, make install
famicom: hold on
famicom: let me smoke a bowl first
famicom: this is too boring
famicom: SunTzuTech
famicom: after i dumped the bios
famicom: what do you want me to do
SunTzuTech: assuming the new driver doesn't work, send it to radeonhd@opensuse.org
famicom: ok
famicom: so
SunTzuTech: no point in shipping the bios if the compiled driver works now, though you may have to create a new xorg.conf file...
famicom: yeah
famicom: after autogen make make install
famicom: ok
famicom: it is still broken
famicom: I'll send the files and what not
SunTzuTech: an xorg.0.log file might be good too. started with -logverbose 7
famicom: blegh
famicom: linux REALLY needs a standardized log format
Arrow: Hi all
joecool: er, probably known... but the radeonhd driver causes GPU to not fully sleep when system is sleeping
joecool: hence battery drains in a couple days (rather then a couple weeks)
joecool: not a huge issue though, scaling isn't implemented yet, that's more important to be done i think
yangman: joecool: that's the first I've heard that. you may want to file a bug
yangman: suspend is done by the kernel, so radeonhd shouldn't cause any issues
yangman: if it's leaving the chip in a state that the kernel can't properly suspend it, then there's a problem
airlied: yangman: the kernel doesn't suspend gpus.
airlied: yangman: that would require a kernel driver.
yangman: airlied: hm... I've always figured there was some sort of generic way to suspend GPUs. I reckon X has hooks for suspend, then?
airlied: yangman: X just switches VT to text mode.
airlied: yangman: and then Linux just does suspend and on resume any of 3 things happens.
airlied: all equally hacky.
airlied: there is no generic way.
yangman: okay, so, how does the GPU actually get the signal to enter suspend?
airlied: yangman: it doesn't we might do PCI suspend settings.
airlied: but I'm not 100% sure on that.
yangman: hm...
dmb: airlied: why is suspend so hard on linux?
dmb: in general
airlied: dmb: because manufacturers don't tell us anything.
airlied: dmb: for GPUs because not having kernel drivers makes it a big set of hoops to jump through
dmb: oh :/