udovdh-afk: tnuyyen has the same problem as I have, and some others:
udovdh-afk: texmode goed black when switching to it via ctrl-alt-Fx
udovdh-afk: on my crt the monitor goes asleep after a few seconds, pointing to a lack of sync?
udovdh: anything else new?
udovdh: things I could try?
udovdh: what is compiz?
udovdh: oh, a window manager
udovdh: how can it use 3d graphics hardware when that is abstracted from the gui?
Honk: the driver doesnt have 3d support, no matter who talks to it :P
udovdh: yes, but how can a windowmanager use 3d? only if there's an api available to call those functions or in our case
udovdh: when the hd driver can implement or fake the api?
mjg59`: OpenGL
udovdh: is abstraction?
mjg59`: Yes
udovdh: ok
udovdh: back via linux
libv: udovdh: is it correct that there is no edid from your monitor?
udovdh: dunno?
udovdh: the x\ctx should do ddc, don't know about anything else; do have a type number
udovdh: it is a ctx 1769se
udovdh: and does ddc
udovdh: what else can i tell?
libv: because your log doesn't show ddc at all
libv: udovdh: what does conntest tell you?
udovdh: dunno, never used that?
libv: udovdh: try it, it will look for an EDID slave
udovdh: rhd_conntest -s and then?
udovdh: what pci tag?
udovdh: of my card?
libv: lspci
udovdh: yes, ok
udovdh: ddc line[0]: slaves: 44 a0
udovdh: (yes, typing here, ssh on other screen)
udovdh: and:
libv: hrm, so why doesn't our driver accept this
udovdh: ddc: rhd_ddc_none
udovdh: dunno? can I test more?
udovdh: i have the git code around
libv: but it finds slave addresses?
udovdh: yes
udovdh: 44 a0
udovdh: if that is it
libv: egbert: any idea here?
libv: a0 is edid
udovdh: can I debug that with some i2c or whatever tool?
udovdh: to see if dcc on the crt works?
libv: udovdh: can you try running rhd_conntest several times and find out whether the bus is detected at some points or not?
udovdh: will do
libv: or whether the result always stays the same
udovdh: 5 tries: same output
libv: urgh
udovdh: maybe it is different in text mode?
udovdh: would be weird, but still
libv: udovdh: rhd_conntest is already conflicting within itself
udovdh: oh. what can i patch to test?
libv: i haven't written the relevant code, and in such a case, no human is able to blurp out a patch like that, let me look into the code first
udovdh: strace?
udovdh: it walks /sysbla/pci
libv: udovdh: whatever you're doing now, is just wasting your own time :)
udovdh: and a bunch of nanosleeps?
udovdh: no action so to say
udovdh: oh well...
libv: hrm, there is nothing different except the fact that ddcprobe has run before...
udovdh: I could try to hook up my lcd
udovdh: maybe I'll do that tomorrow, to see if it makes a difference
libv: udovdh: just hold your horses please
udovdh: done :-)
libv: udovdh: edit rhd_conntest.c
libv: go into DDCReport()
udovdh: ok
udovdh: am there
libv: copy over DDCProbe(map, 0, EDID_SLAVE);
libv: and put it one line earlier, on its own
libv: so that it reads:
udovdh: so yyp in vi
libv: Bool Chan0, Chan1, Chan2, Chan3;
libv: DDCProbe(map, 0, EDID_SLAVE);
libv: Chan0 = DDCProbe(map, 0, EDID_SLAVE);
libv: Chan1 = DDCProbe(map, 1, EDID_SLAVE);
udovdh: ah ok
udovdh: done
libv: this will point out whether running this routine before will solve the issue or not
udovdh: so build now?
libv: yes
udovdh: no difference
udovdh: well...
udovdh: it says hotplug: rhd_hpd_1
udovdh: 1...
udovdh: and the rest is ddc: rhd_ddc_0...
udovdh: is that ok?
libv: hrm
libv: so now it does report rhd_ddc_0?
udovdh: yes
udovdh: but the 1 is for the port where the crt is detected?
libv: uncomment that line you just added
udovdh: comment you mean?
libv: this is an io pin, attached to the dvi hotplug detect pin
libv: err, yes, comment, sorry :)
udovdh: done
libv: try again
udovdh: no change
libv: so it reports ddc_0 now?
libv: have you rebuilt rhd_conntest in between?
udovdh: i did make it in between the edits
udovdh: it reports:
udovdh: hotplug: rhd_hpd_1
udovdh: ddc: rhd_ddc_none
libv: so now it reports none again
udovdh: ddc line[0]: slaves: 44 a0
libv: ignore HPD
libv: we are trying to follow rhd_ddc_
udovdh: ok
libv: please verify again that with the change, it reliably (so try multiple times in a row) returns0, and without the change, it reliably returns none
udovdh: will do
udovdh: it might be a timing issue?
libv: it looks more like an inialisation issue
udovdh: 5 times with change and 5 times without:
udovdh: consistent results
libv: udovdh: ok, can you send in a copy of your bios, you can create it with rhd_conntest -d
udovdh: will do
udovdh: the same email as before?
libv: this is an issue for egbert to solve, i don't know whether he will have enough from the results we are seeing here, so maybe if this isn't enough, he might see something useful in there
libv: yup
libv: just reply to the same thread
udovdh: sent
udovdh: fresh email, sorry
udovdh: maybe it is a bios issue I gather from your output?
libv: np
udovdh: in that area I have a problem:
udovdh: gigabyte expects win32 to update the bios on the card
libv: well, it is an issue with initialising for a new ddc channel
udovdh: I asked them for a non-windoze bios flasher for the card
udovdh: no reply yet
udovdh: we leave it here I understand?
udovdh: then I'll be back tomorrow
udovdh: later
libv: cya
dto: hello folks.
dto: i have an HD2600PRO (a cool one with the Zalman passive cooling)
dto: and I am on debian, using a patched kernel (with the realtime patches from kernel.org)
dto: what's the best way to get at least decent 2D for now?
dto: i'll try the radeon hd driver
dto: airlied: ok, i checked radeonhd out of git, did autogen.sh, and got http://paste.lisp.org/display/52273
dli: dto, try: apt-get build-dep
dto: dli: for what package
dto: i tried it for xorg
dli: dto, try, apt-get build-dep xf86-video-ati
dto: E: Unable to find a source package for xf86-video-ati
dto:
dto: hmm
dto: this seemed to work: apt-get build-dep xserver-xorg-video-ati
dto:
dto: didn't help autogen.sh
dli: dto, right, debian names
dto: is current git just broken? should i try a released tarball?
dto: i have an HD2600PRO
dto: how good is the 2d performance btw?
dli: dto, try a debian package instead
dto: hmm
dli: dto, http://packages.debian.org/sid/x11/xserver-xorg-video-radeonhd
dli: dto, backport this one
dto: how do I backport something? just apt-get source and then rebuild?
dli: dto, http://www.phoronix.com/forums/showpost.php?p=19575&postcount=37
dto: dli: that looks good. let me try...
dto: btw, thanks for your help everyone
dto: error on that page: " apt-get devscripts fakeroot pbuilder" should really be "apt-get install devscripts ..."
dto: dli: the last step fails: E: Build-Depends dependency for xserver-xorg-video-radeonhd cannot be satisfied because no available versions of package xserver-xorg-dev can satisfy version requirements
dto: i'm just going to upgrade to sid
dto: i've wanted to anyway
dli: dto, you didn't do the sed step
dto: oh.
dto: hold on
dto: i just pasted that into the terminal, maybe it didn't get through
dto: ah, working now
dto: hold on
dto: the debuild -us -uc fails: http://paste.lisp.org/display/52273#1
dto: oh, debuild -S
dto: but that fails because of "secret key not available"
dto: hmm, ok now i'm doing pbuilder create
dto: debuild -S -us -uc worked
dto: ok... i've upgraded to sid