tigerchen: might it be that current git head doesn't compile? or have i done something wrong?
the-me: here all is fine, you should nopaste your compile log maybe
tigerchen: rhd_randr.c: In function ‘rhdRROutputGetProperty’:
tigerchen: rhd_randr.c:1051: error: ‘struct rhdOutput’ has no member named ‘Backlight’
the-me: tigerchen, is this maybe with this unoffical patch from bugzilla?
tigerchen: nope, just a fresh git-pull
libv: hrm... about the cursor corruption issue earlier... sadly not :(
libv: we have no real idea what is causing it either...
tigerchen: libv: is there a problem with current gt, or should i do a fresh checkout? won't compile here
libv: tigerchen: egbert is on it :)
libv: he's bent over his laptop, so i guess he will be pushing a fix any minute now :)
tigerchen: have you seen current "pc welt" there is an article titled "bye xp, hello linux" i wonder what's in there
libv: no, i barely have time to read online things anymore, let alone magazines in a language i still don't master :)
libv: i wish i was able to do both, but sadly not :(
tigerchen: hmm, i want to read that article, but i don't want to spend money on that mag...
tigerchen: that's too bad
libv: well, often times, you will see that whatever is in any magazine is too broad and not indepth enough for it to be worth the money
libv: time is often better spent online reading something there :)
tigerchen: i know. i was just wondering what that mag that i rate just a little bit over "computer bild" is writing about linux
tigerchen: well, online reading is a bit problematic if you're offline while travelling and don't have the money for a umts-flat
libv: yeah, dead tree is nice for travel
tigerchen: I'm reading a kind of confusing book right now: i always want to read more, but if i read, i can't do so for mor than about an hour, because it somehow doesn't bind me to it for a longer time
tigerchen: yawns still two more hours to go until the thing why I'm here starts
libv: tigerchen: egbert quickly pushed a fix before he ran out
libv: he was unable to buildtest it though
ciaranm: i think egbert is going to cry when he gets back
libv: so isn't building?
ciaranm: builds for me
libv: ok, and you had the build breakage before as well?
ciaranm: nope... my comment about crying was regarding the rs600 tvout weirdness i'm discussing with him via email
egbert: ciaranm: do you have any news on this?
ciaranm: egbert: just sent you an email
egbert: ciaranm: so the results are not all that bad. the patched rhd_conntest was reliable, wasn't it?
ciaranm: egbert: yep
ciaranm: at least i ran it on watch -n1 --differences=cumulative for several minutes, and it stayed consistent. whereas before it'd show the other values at least once every ten seconds
egbert: ciaranm: i would like to modify the patch so it uses the old load detection algorithm but with slightly modified settings.
ciaranm: egbert: alright
egbert: ciaranm: are you at the machine right now and can do some tests for me?
ciaranm: egbert: yep
ciaranm: i'm here for another hour or so
tigerchen: libv: *test*
tigerchen: libv: builds
egbert: ciaranm: just sent email. please put this patch on top.
egbert: tigerchen: thx for testing ;) can you check if it also works? i don't have the chance to run another laptop atm.
tigerchen: egbert: hmm, I'm at radeon atm and that works so nice, faster, with (a bit) 3D
ciaranm: egbert: can i get away with not restarting x, since it's just conntest being changed?
egbert: yes, but you may want to switch to console.
ciaranm: egbert: ok, with that patch it's back to alternating between DACA and NONE
egbert: ok, so can you try to exchange the line: RegWrite(map, offset + DACA_CONTROL1, 0x00050802);
egbert: with RegWrite(map, offset + DACA_CONTROL1, 0x000A0A02);
egbert: ciaranm: it's line 633 here.
ciaranm: egbert: still alternates with that
egbert: ciaranm: ok.
egbert: ciaranm: this is getting difficult now :/
ciaranm: i can imagine... the nastiest hardware stuff i've done is fixing up some old sbus drivers for linux 2.6... and they didn't have many registers, and the ones they did have were all explicitly documented via openpromfs
egbert: ciaranm: could you please increase the timeout 100 to 37000 in line 649?
ciaranm: mm, this is looking hopeful...
ciaranm: looks like that's worked. i'll let it run for another few minutes to be absolutely certain
egbert: ciaranm: ok, great.
egbert: ciaranm: could you restore the old setting for DACA_CONTROL1? it was: 0x00050802
ciaranm: with that restored it's still working
ciaranm: so it's just a delay issue? that'd explain the varying results
egbert: ciaranm: so we should try to reduce this nasty delay. maybe 10000?
ciaranm: 10000 looks good still
ciaranm: even 1000 appears to work
egbert: ciaranm: :)
egbert: ok, so i will give it 2000 ;)
egbert: thanks for testing! i'm going to push a fix for the driver later. let's see if this also works.
ciaranm: 150 is broken. 200 works.
ciaranm: ok, thanks. lemme know when you have things for me to test
egbert: ciaranm: ok. i think 2000 is short enough.
egbert: ciaranm: no, thank you :))
egbert: ciaranm: i've just pushed something. i need to run now but should be back later :)
ciaranm: egbert: ok, cool. i'm also just about to go out. i'll test this evening (UTCish)
Ron56: i search in the wiki but no response :s, i've installed radeonhd drivers , it works but when i do a 'glxinfo | grep rendering' it tells me : "direct rendering : No" ... is it ok ?
Ron56: sorry for my english
the-me: Ron56, look at the wiki
the-me: Ron56, radeonhd hasn't got accelerated 3d rendering, yet
Ron56: the-me, ok i didn't found the information
ndim: Meh. Still segfaulting. This is stock radeonhd (i.e. no code patches at all): https://bugzilla.redhat.com/show_bug.cgi?id=439046
egbert: ndim: do you see the segfaults, too?
ndim: egbert: No.
ndim: egbert: Otherwise, you'd have seen a patch from me.
ndim: egbert: I have a different GPU, and i386 (as opposed to his amd64)
ndim: egbert: Do you have any idea what this might be?
ndim: I hope it is not some obscure x86_64 compiler issue
Death_Syn: hmm, current git gives me no backlight on my rs690 on LVDS
Death_Syn: cehcking mailing lists for results