eosie: ok
zhasha: eosie: can I have it too?
soreau: eosie: If you could post it in-chan here, that would be nice :)
eosie: zhasha, soreau: http://storm.unas.cz/0001-r300g-added-support-for-3D-textures.patch
soreau: eosie: Thanks :)
eosie: the progs/trivial/*tex-3d things work, piglit/texturing/tex3d with a little modification too
zhasha: it seems kind of weird to have 3d texture mipmaps :/
zhasha: anywho~ I think our fb setup is busted because mipmaps are generated by setting some cbuf, then drawing to it. it looks like we got the rendering right but setting the cbuf just doesn't work
zhasha: I'm not sure though, it's just a guess
eosie: zhasha: well, the gpu spec says that 3d texture mipmaps are supported by the GPU, so there is no reason not to support them in the driver
spstarr: damin the GPU is HOT
spstarr: really hot
spstarr: downclocks laptop PM mode
zhasha: eosie: I didn't try with 3D textures. 2D mipmaps don't work either
eosie: zhasha: I know
zhasha: I've been trying to track down what's doing it
zhasha: look at r300_emit.c:297 and tell me that doesn't look weird
spstarr: its so hot
spstarr: I can BURN my finger?!
spstarr: uhm......
spstarr: this cannot be good
zhasha: MostAwesomeDude: when emitting a reloc, what exactly is shoved into the CS?
airlied: zhasha: two dwords, a nop header + a reloc index
zhasha: airlied: so it's a packet3?
zhasha: how exactly does that work for something like RB3D_COLOROFFSET
combr: hi all. anyone knows, what "drmRadeonCmdBuffer: -22. Kernel failed to parse or rejected command stream." mean? r600_dri in use, radeon 4670
airlied: zhasha: yeah a packet 3 nop
zhasha: it means the driver tried to write a register it's not allowed to write
airlied: it comes directly after the packet0 with the register in it
combr: zhasha: i run glxgears for test ;)
zhasha: airlied: so it becomes > packet0 header -> what dword here? -> packet3 header nop -> nothing >
spstarr: switches to intel gma for now again
mart: MostAwesomeDude, latest mesa works with egl for intel, peglgears 930fps
spstarr: laptop got to 80C almost
mart: others are also blitted to the screen but this one isn't, can't see a function that would do it either, dri isn't in a such a good shape, eagle to go
mart: i mean eglgears and such..
spstarr: not good
spstarr: does Dynamic PM work on the r6xx?
spstarr: or just the driver 'says' its turned on
eosie: zhasha: I can't see anything weird in that code, however I haven't studied it yet
combr: heh, i try ppracer and all hanged ;)
combr: please, how i to debug something?
zhasha: err my bad on the cs stuff but it looks like we have a packet0 header, then immediately after we have a packet3 header?
airlied: zhasha: no
airlied: packet 0 header + packet 0 data (maybe be 0), packet 3 header + packet 3 data == reloc index
spstarr: airlied: r6xx is working nicer, but HOT HOT HOT :(
spstarr: airlied: but it seems dangerous
eosie: spstarr: 80C is nothing, more than 100C is starting to be dangerous
spstarr: airlied: the laptop is so hot it is possible to burn my finger
spstarr: :(
spstarr: so i switched to intel gma again for now.. cool things down and the ambient temp!
zhasha: oh nevermind, I just forgot the macro
mart: spstarr, do you have any drivers that perform with intel?
zhasha: so how does the cp interpret a 0 offset for RB3D_COLOROFFSET?
spstarr: mart: just drm-next + mesa git master
spstarr: mart: but im concerned about the heat the AMD GPU is putting out
zhasha: err, package data
mart: spstarr, biggest i have is rc5 plus masters 200fps gears, that's little too low for the pseudo
mart: with my computer hardys xaa and egl's 930 is maximum, xephyr's exa 1100
spstarr: mart: i donno :)
mjg59: spstarr: You've been concerned about the heat for weeks now
spstarr: just dont want to melt plastic
mjg59: If it was going to melt, it probably would have done by now
mjg59: But yes, there's no management of the thermal policy
spstarr: mjg59: the amd gpu will downclock itself if it gets too hot?
spstarr: regardless of driver? (like intel cpus)
mjg59: No
spstarr: ouch
mjg59: Well, it might. But that's not mentioned in the docs.
spstarr: !!
mart: MostAwesomeDude, so i am temporarly thinking of putting egl out instead of gl then, but krh hardly responds what was the eagle magic he did, that library looks very sane to me
spstarr: mjg59: i will wait then switch back to amd gpu when some of the dynamic pm is started after
mart: x11 mode and console mode inited runtime with dri and kms native
mart: performance of it with some perfs and gears, have to try myself
mart: 2.6.32-rc5 didn't blit/flip galliums egl i can say only, probably pageflip isn't yet included there
mart: and a hang when exiting the client programs
simplexe: hi all
DanaG: MMmyeah, KMS+powersavings will be nice.
DanaG: Even if it's just "stay slow" power savings.
simplexe: zhasha: ]
zhasha: simplexe: yes?
simplexe: I did everything as you said did not help
eosie: zhasha: must each r300g patch go through corbin?
zhasha: not necessarily
zhasha: if it's sane I could just as easily commit it
simplexe: r300_dri.so: undefined symbol: rc_opcodes (((
zhasha: eosie: I just rebooted into my devbox so you'll have to post it again
zhasha: simplexe: did I say something? in that case, I forgot :P
simplexe: (12:27:50) zhasha: simplexe1: replace --with-state-trackers=dri,egl with --disable-gallium
simplexe: (12:28:02) zhasha: also --with-driver=dri is superfluous
simplexe: ./autogen.sh --prefix=/usr --with-dri-driverdir=/usr/lib/xorg/modules/dri --with-dri-drivers=swrast,radeon,r300 --enable-glx-tls --with-driver=dri --enable-xcb --with-state-trackers=dri,egl --disable-glut
zhasha: right, that... well, rc_opcodes should be in r300compiler
simplexe: i using latest mesa
zhasha: I don't want to get hung up on classic mesa though, as I don't know much about how it's built
simplexe: m.b. update libdrm?
zhasha: shouldn't make a difference
eosie: zhasha: what do you mean? the patch i've posted about 2 hours ago?
zhasha: simplexe: r300compiler is what you see in dri/r300/compiler
zhasha: eosie: I assumed that's what you were talking about
simplexe: zhasha: i know this
eosie: zhasha: http://storm.unas.cz/0001-r300g-added-support-for-3D-textures.patch - if you agree, please push ;)
zhasha: eosie: pushed
zhasha: there aren't, by any means, high standards for patches. as long as they don't seem to break everything else
eosie: thanks
simplexe: hmm... zasha, how to compile worked r300g?
soreau: simplexe: Compile mesa with --enable-gallium-radeon
simplexe: soreau: and ....
simplexe: how load this?
soreau: cp $PREFIX/lib/dri/radeon_dri.so $PREFIX/lib/dri/r300_dri.so then load with LIBGL_DRIVERS_DIR=$PREFIX/lib/dri glxinfo
soreau: for example
soreau: You might even want --with-state-trackers=dri,glx
simplexe: hmm... dri,glx,egl?
soreau: not sure about egl
soreau: zhasha or MostAwesomeDude might be able to clarify
simplexe: zhasha
simplexe: help pls
zhasha: simplexe: ./autogen.sh --enable-gallium-radeon --disable-gallium-intel --enable-debug --with-dri-drivers="" --with-state-trackers="dri"
simplexe: in state trackers egl works?
simplexe: ok
zhasha: cd lib/
zhasha: LD_LIBRARY_PATH=. LIBGL_DRIVERS_DIR=. LIBGL_DEBUG=verbose glxgears
simplexe: zhasha: --with-dri-driverdir such as in mesa?
zhasha: don't install gallium
zhasha: just run make on it and try it locally
simplexe: hmm...
soreau: simplexe: In other words, after you build, you can test gallium on a per-app basis without actually installing it
simplexe: installing this is careful?
zhasha: to the first question: yes
zhasha: the second: what?
soreau: You should not install it at all really as it is certainly not ready for everyday use
simplexe: heh
mart: soreau, you know Duke?
soreau: mart: Nukem?
mart: Duke` a guy named so
soreau: no, why?
simplexe: in mesa rc_opcodes error, gallium not ready for everyday use, bad days for me
bridgman: spstarr; ping
mart: soreau, hehe, no not that one, but i have playd that game, very very long time ago, school
soreau: mart: I was only joking as I do not know a Duke right off hand
mart: i think it has linux version too, but i haven't tried
soreau: mart: Why do you ask anyway?
soreau: Hi bridgman
simplexe: my child spilled juice on my computer, now I do not have two rv770 ((
bridgman: hi soreau
mart: wanted to ask what setup he had with i915, which commits and packages
soreau: bridgman: Im not typically up this late/early.. which is it for you? :)
mart: soreau, but i have seen that name logging in, sometimes, maybe he will again at some point
soreau: bridgman: You are probably the only person on irc I know that still uses dial up. Will you ever be able to upgrade? :)
mart: ehuu yuhuu, dun
bridgman: I guess I'm up late... went to bed early, woke up 1am and couldn't sleep, need to sleep again before am
bridgman: I will upgrade once I chop down enough trees
bridgman: probably next year
zhasha: eosie: I'm at a loss. Why don't mipmaps work?
eosie: zhasha: dunno
zhasha: I've been through everything but I don't understand why it doesn't work :(
ayarter: Hey all. I'm finishing up an install of debian on a sunblade 2500. I'm using the fbdev driver; but in the past i remember using the ati driver. I guess the problem has to do with PCI domains.
ayarter: Can anyone help?
eosie: zhasha: i've fixed 2 bugs related to mipmaps, but it has probably nothing to do with what you tried to accomplish
eosie: zhasha: http://storm.unas.cz/0001-r300g-fix-redefining-mipmaps-and-fetching-from-them.patch
zhasha: eosie: does progs/tests/mipmap_limits work?
eosie: zhasha: it makes piglit tests fbo-clearmipmap and tex-redefine work, please push
eosie: wait a minute
eosie: zhasha: no, it doesn't
eosie: crap
zhasha: eosie: patch pushed, and crap what?
adamyarter: well, if anyone can help, let me know. Thanks :)
eosie: zhasha: nothing
eosie: i had accidentally older version of the driver when testing mipmap_limits, so I've retested with the patch applied and it's the same
MostAwesomeDude: At a glance, why not dump width and pitch and see how far off they are?
MostAwesomeDude: Also yeah, that's not gonna work.
zhasha: MostAwesomeDude: I checked pitch, it was right
MostAwesomeDude: zhasha: But it won't always be, e.g. rectangles.
MostAwesomeDude: Oh!
MostAwesomeDude: Haha, wow.
zhasha: ...
MostAwesomeDude: It's always possible that mipmaps are broken when pitch is set...
zhasha: now I feel enlightened
MostAwesomeDude: ...because you only need to set pitch when there's rectangles...
MostAwesomeDude: ...which can't be mipmapped!
MostAwesomeDude: Man, this fever is a rush in the head.
MostAwesomeDude: I can totally detect that and write out some guard code.
zhasha: get cracking!
MostAwesomeDude: adds to todo list
MostAwesomeDude: No. Sleep, then re-read logs, *then* code. Fever coding's gotten me in trouble before.
eosie: interesting
eosie: MostAwesomeDude: as you might have noticed, the patch i've sent you is already pushed
zhasha: incidentally, did fever have anything to do with you breaking the mesa build system with automake updates?
MostAwesomeDude: eosie: I noticed, don't worry.
MostAwesomeDude: zhasha: I broke it? Link?
zhasha: you told me you broke it once
MostAwesomeDude: Oh, a while ago.
MostAwesomeDude: No, fever caused a project I did for a website to be kind of, well, funky.
ossman: ehm... glean spews out a lot of failures on the software renderer. Is that a known problem, or is something broken in my end?
ossman: pokes MostAwesomeDude
MostAwesomeDude: ?
osiris_: glisse: radeon drm module crashes when running mipmap_view on r300g http://pastebin.com/m387f1d15
osiris_: same for mipmap_limits
glisse: osiris: it's r300g issue
glisse: you render to too small buffer
glisse: the oops message after is triggered on purpose because somehow it endup calling an unemitted fence
glisse: my guess is that the cleanup path when we reject the ib is wrong somewhere for the fence
ossman: glisse, is it expected that glean fails on a lot of tests for the software renderer?
ossman: I was expecting it to succeed on everything so that I could have something to compare against when hacking r600 :/
spreeuw: [drm:radeon_cs_ioctl] *ERROR* cs->dwords too big: 25412
spreeuw: does it mean my drm is out of date?
glisse: ossman: it should work
glisse: maybe gallium software isn't that perfect yet
glisse: spreeuw: no
glisse: it means there is a bug somewhere in userspace
ossman: glisse, all programs and shaders fails miserably
osiris: can someone take a look at http://cgit.freedesktop.org/~osiris/mesa/ it's my VBOs for r300g. it works for most of the apps, but glxgears hangs the GPU (fortunately GPU reset brings it back)
ossman: seems to be rounding errors and missing alpha mostly
osiris: spreeuw: are you using r300g?
ossman: e.g:
ossman: FAILURE:
ossman: Program: SWZ test 1
ossman: Expected color: 0.5, 0.5, 0.5, 0.25
ossman: Observed color: 0.501961, 0.501961, 0.501961, 0
spreeuw: osiris: no r600 current
glisse: osiris: would be nice to capture one cs of buggy glxgears
osiris: glisse: http://pastebin.com/m4761b6c2
osiris: glisse: nevermind. I've just found it
osiris: glisse: glxgears is sending draw calls with inappopariate number of vertices for given primitive
glisse: i think mesa core should catch that
osiris: MostAwesomeDude: VBOs are working now, but certainly aren't ready for pushing upstream
osiris: MostAwesomeDude: could you take a look at blending state? it seems to be broken (e.g. progs/demos/fire)
glisse: osiris: btw which gpu ? for the gpu reset
osiris: rv380
taiu: osiris: gears has some quads with 2 verts only, classic drivers strips them, is this accounted in 300g
taiu: osiris: nevermind, finished reading scrollback
elbeardmorez: hi. do we still need to merge the drm-next code with 31.x kernels to play with the KMS features please? ..for older rv4xx chipsets.. ..i thought no but..
chithead: kms for r500 and older is in 2.6.31
redex: this happens while playing nexuiz
redex: nexuiz-linux-686-sdl: radeon_lock.c:65: radeonGetLock: Assertion `drawable != ((void *)0)' failed.
redex: Aborted
Pulsewidth: Is it possible to set modelines for kernel modesetting?
soreau: yes
Pulsewidth: Good, because my monitor's EDID "native" mode isn't the best mode
agd5f: amarks: Option "EXANoDownloadFromScreen" or Option "EXANoUploadToScreen"
MostAwesomeDude: osiris: Looks like you're making progress, I could cherry-pick a few of those.
MostAwesomeDude: I'm kind of confused on how you're going to do true VBOs for Draw though.
osiris: MostAwesomeDude: currently I'm not using draw at all
MostAwesomeDude: osiris: How will you handle SW TCL then?
osiris: MostAwesomeDude: with draw :) but I'm not interested in it for now
MostAwesomeDude: osiris: Well, I can't merge anything that breaks SW TCL more.
MostAwesomeDude: It's already kind of busted; it'd be great to have it un-busted
osiris: MostAwesomeDude: will take a look at it later
spreeuw: hmm how can I get the 3d's idea of desktop res in line with the 2d idea of res
spreeuw: after a 3d crash I'm left on a wrong 2d res
spreeuw: which I reset with xrandr
spreeuw: if I then restart a 3d game fs
spreeuw: it doesnt go full screen, but is rendered on top, smaller than the 2d desktop
spreeuw: floating above as it were
spreeuw: how can I get them in line with eachother again
spreeuw: without an xserver restart
amarks: agd5f: should AccelDFS be false when trying EXANoxxxxScreen options?
VitRuss: I have "probe of 0000:01:00.0 failed with error -22" in dmesg. And KMS seams not working on HD4550. Where is the problem?
airlied: VitRuss: AGP?
VitRuss: airlied: PCIE
airlied: does the firmware load?
VitRuss: airlied: it's enabled in config. How to check?
airlied: it should print out if it failed to load it
airlied: in the dmesg
VitRuss: airlied: http://pastebin.com/d1245a119
VitRuss: airlied: I found nothing about it
airlied: VitRuss: wierd can you boot with drm.debug=15 on the commandline/
VitRuss: airlied: ok.
VitRuss: airlied: [ 0.445414] [drm:radeon_driver_load_kms] *ERROR* Failed to initialize radeon, disabling IOCTL http://pastebin.com/d26e8a887
VitRuss: airlied: does CONFIG_PREVENT_FIRMWARE_BUILD=y influence?
airlied: yes that seems like the wrong thing
airlied: though not sure it matters
Ralesk: Hi... I seem to have some severe performance issues with radeon 6.12.4 running on server 1.7.0
Ralesk: qt's graphicssystem native is incredibly slow, raster is fine (faster than the previous driver I used with 1.5.0) but plasma popups still appear really slowly and drive X nuts (it'll use the CPU quite a bit until the popup appears) -- but also gtk apps seem affected and grind on my CPU more now
VitRuss: airlied: I've tried everything. Swithcing to zen4 kernel. Looked over my config again. No result.
Ralesk: hmm, on the radeon machine I have, if I only have plasma-desktop on the screen (no windows obscuring it besides conky), I get X 75% and plasma 12% and dbus-daemon 9% CPU use... if I run plasma with graphicssystem raster, plasma will use 60% and X will use like 30...
VitRuss: Could anybody help with KMS enabling? something seems to be wrong in my kernel.
Ralesk: no compositing, no anything. just sitting there eating the CPU
VitRuss: http://pastebin.com/d51572c4f here is pastebin and the most interesting [ 0.471761] radeon: probe of 0000:01:00.0 failed with error -22 (PCIE)
zcrayfish: Ralesk: anything unusual in your Xorg.0.log?
Ralesk: nothing suspicious whatsoever :/
Ralesk: ehhhhh, probably not a radeon issue, does the same with the vesa driver on the same machine, but it never happened on the old X (7.4, server 1.5.0/1.5.3 etc. etc.)
zhasha: Ralesk: try enabling compositing
Ralesk: any idea if an old kernel could be causing this? I know the intel driver keeps yapping about the lack of kernel>=2.6.28, but the other drivers don't seem to do this
zhasha: Ralesk: are you using KMS?
Ralesk: no idea at all
Ralesk: kernel 2.6.27.4 here if that helps
zhasha: then probably not
Ralesk: dig.
Ralesk: hmm, forcing EXA (XAA still seems to be the default according to the man) makes things less slow, but it still chews on the CPU quite significantly
Ralesk: now to see that thing about compositing
Ralesk: slightly less bad -- plasma only 25%, dbus ditto, X 18%
Ralesk: (this is an AthlonXP with a radeon 9600, I forgot to tell)
zhasha: Ralesk: try setting the renderer to opengl
Ralesk: flickers like mad
zhasha: Ralesk: right... forgot you don't have DRI2
Ralesk: that one depends on the kernel/KMS then?
zhasha: Ralesk: for radeon at least, it has a dependency on KMS
Ralesk: ah, gotcha
zhasha: well, technically it depends on GEM, the memory manager, which usually has KMS in its wake
Ralesk: gonna have to kick my coworker a bit more then to get the new kernel ready for the distro :))
zcrayfish: I wish I could kick my coworkers...
Ralesk: zcrayfish → heheh
zcrayfish: Could you pastebin your xorg.conf?
Ralesk: it's mostly ancient, X discards most of it -- and it works exactly the same (if I keep it on XAA) than the xserver-generated config
Ralesk: moment though
spstarr: is drm-next pending fixes?
Ralesk: http://xorg.pastebin.com/m3f4696a
eosie: zhasha: http://storm.unas.cz/0001-r300g-add-cubemap-support.patch (it works... sometimes)
eosie: zhasha: http://storm.unas.cz/0002-r300g-read-occlusion-query-results-from-both-Z-pipes.patch
eosie: zhasha: please push :)
zhasha: eosie: have you considered getting an fd.o account?
eosie: zhasha: not yet :)
zhasha: how do mipmaps fare with the cubemap patch?
eosie: as with everything else
zhasha: bad? :P
eosie: yes
zhasha: eosie: done
eosie: thanks
dmb__: what video card does the bios use when booting up your computer if you dual head?
sgcb: if your mobo is like mine, it will have a primary pcie slot and a secondary pcie slot, bootup messages only display on whichever card is in the primary slot
sgcb: there is an option in the bios to switch between a pcie card or a pci card but that's about it, no pcie0 or pcie1 option for me
adamyarter: Hey all. I'm having trouble with PCI domains in X. Is anyone familiar with such a thing?
airlied: adamyarter: familiar with trouble with them, not sure about making them work ;-)
adamyarter: airlined: I remember getting it to work once on this machine, but I had to mess with it and screw it up.
adamyarter: I'm using fbdev atm, but its ehh. lol
airlied: adamyarter: some sort of powerpc?
adamyarter: Sunblade 2500.
airlied: ah sparc the second guess ;-)
adamyarter: yeah. What a pain. lol
adamyarter: I honestly remember using a newer xorg and it working. Debian'[s a little behind though
airlied: adamyarter: yes the debian stuff is probably a bit old if I had to guess
airlied: esp if Debian stable, though I'm not sure I've heard many reports of it working with any releases ;-)
amarks: wouldn't want to be at Sun right now lol
adamyarter: God, no. I'd be terrified.
adamyarter: airlined: Its Debian stable. lol
amarks: and intel want to go after mainframe with x86, sparc is under pressure
adamyarter: I think sparc is probably gonna be toast soon here.
zcrayfish: May it rest in peace. :-(
amarks: unfortunately sun doesn't have a lot of distinctive value now, dtrace isn't gonna cut it
adamyarter: sparc l
adamyarter: pitnumbers x86 here.
adamyarter: *outnumbers rather.
amarks: i used to love sun, but that was in the heyday lol
adamyarter: dtrace is awesome. But yeah, I think linux/x86 has got it these days.
amarks: i threw out a ultra 10 the other day
adamyarter: amarks? Sad day. I'm thinking about tossing my sparcstations, but iono
amarks: i wanted to keep it, but really it was a waste of space
adamyarter: those things are pretty massive
amarks: would cost more in ebay fees to sell it lol
adamyarter: Oh god. Probably $60 to ship.
amarks: rubbish bin $0
amarks: i still have solaris CDs
adamyarter: So, I'm using xorg 1.73 apparently. Do newer versions of xorg have better luck with pci domains?
amarks: oh the memories
adamyarter: Oh yeah? Solaris 8 or something?
amarks: i even have 2.6
adamyarter: oh wow.
amarks: xorg 7.3?
adamyarter: My Ultra 2 and dell server both run 9.
adamyarter: I dunno. the package name is xserver-xorg 1.73+20
adamyarter: I'm just coming off of Solaris, so this is all a royal pain in the a**
amarks: well that's not a standard numbering convention
adamyarter: abd fbdev sucks at 1280x1024.
adamyarter: hrm, lemme checkj
adamyarter: X.Org X Server 1.4.2
adamyarter: Release Date: 11 June 2008
airlied: yeah that is old X server
adamyarter: figures. I suppose I'm off compiling, eh?
amarks: i'm surprised it hasn't disintegrated
airlied: adamyarter: you could pull stuff from unstable
airlied: or experinemtanl
airlied: debian stable is only "stable" when it works
adamyarter: yeah. I remember experimental working
adamyarter: but then the package manager would break the machine once a week.
airlied: you can fix what bits come from where, but I'm no Debian person
adamyarter: I could backport it I suppose, but I don't know a thing about it.
adamyarter: I just needed to get to something a little more home computer desktop friendly.
amarks: wonders what debian's definition of stable is
airlied: amarks: pretty much what works or doesn't work won't change over the lifetime
amarks: if they wait around for little to no bugs, by that time the world has moved on and they are on unmaintained code
adamyarter: amarks: I've noticewd.
adamyarter: I don't have much other option though, except for Gentoo.
adamyarter: And I have a strange feeling that by the time I'm done compiling and fixing the distro, I'll be single.
airlied: adamyarter: just use experimental or unstable X/kernel packages
amarks: don't worry sparc will love you forever
amarks: it's called the gentoo holiday
adamyarter: wouldn't I need to upgrade the whole distribution to do that?
amarks: set off your compile, go overseas, come back, hopefully all done
adamyarter: amarks:And on a 2x1.6 ghz sparc, it might be a long holiday.
adamyarter: I'd love opensolaris, but I need to setup an x86 install server for it? No thats.
adamyarter: thanks rather.
airlied: adamyarter: I don't think so, though you'd yhave to ask some Debian person
amarks: adamyarter: perhaps spend $500 or so, get an x86?
amarks: and more time for your other half
adamyarter: well, I have an x86 machine. Its just stupid that opensolaris sparc requires an x86 install server with opensolaris
adamyarter: so I'd literally need to setup an install server inside of a virtual machine, read the 300 page manual to do that, and then setup this machine.
airlied: wow that sucks
adamyarter: And save the install server's virtual drive, in case I need to ever reinstall this machine.
airlied: agd5f: btw linux always disable msi on rs400/rs480 chipsets
amarks: can it be any more difficult?
adamyarter: God, I don't know.
amarks: keep your sanity. use the sunblade as a footrest
adamyarter: I just about feel that way.
adamyarter: Actually, once xorg is fixed, my machine will be just fine. ]
adamyarter: but fbdev is so slow.....
adamyarter: so I must apparantly compile from source in debian
adamyarter: to avoid package dependancy issues. cute.
amarks: well i guess that is always an option
airlied: yeah normally I'd just go to unstable ;-)
amarks: i've used gentoo for a few years now, i used to use fedora but got stuck in rpm hell, i swore never to touch it again purely on package management issues
adamyarter: how is gentoo?
dmb__: there is yum
adamyarter: Thats another option of mine.
dmb__: don't dis fedora in front of airlied :)
amarks: i'm sorry to diss it, but it was my experience
adamyarter: I'm really concerned about the setup of gentoo more than anything.
adamyarter: the install manual looks pretty daunting, but that's how these thinsg always are.
amarks: yum didn't solve anything
amarks: adamyarter: gentoo install is well documented if you wish to go down that path
airlied: most ppl dis Fedora from like FC1 or 2 days
adamyarter: amarks: How complicated are we talking?
amarks: it was FC3 IIRC
airlied: 99% of rpm depends hell were people mixing/matching random repos
amarks: package rollback was a nightmare
airlied: rollback is always a mess no matter what distro, esp if the package changes config file formats etc
amarks: for the majority, i am sure it is fine, it just didn't provide enough flexibility for me - some parts i want stable, some i want bleeding edge
adamyarter: amarks:I have no problem spending a day or two with it, but anymore wouldn't be acceptable.
amarks: and i didn't want it to upgrade half of my system to move one package to a dev version, when mostly that was unnecessary
amarks: adamyarter: gentoo is a knife, it can cut well, but can also cut you if you aren't careful
airlied: most distros fail to do that sort of things unless you build yourself
amarks: adamyarter: gentoo is slow and painful to install, though i found the investment worth it
amarks: adamyarter: most don't want to invest the time upfront
amarks: airlied: true
amarks: airlied: fedora is a good out-of-the-box solution
airlied: you either end up with 15 version of openssl installed and other things like it
airlied: or build from src
adam_yarter: sorry. Anyhow, how long do you think that would take?
amarks: if you follow normal install steps, it doesn't take too long, but really it comes down to your machine speed. a slow machine is going to require patience whereas a core i7 slices through the compiles extremely fast
amarks: i've never tried gentoo on sparc though
adam_yarter: amarks: How involved would you say the install is?
amarks: it's the opposite of fedora :)
amarks: you do everything
amarks: there is a graphic installer nowadays, never tried it though
amarks: probably not as good as fedora's
adam_yarter: is there? lol. The geek in me says 'yeah, do it'.
airlied: would just go to Debian unstable, it would be a lot less painful
adam_yarter: The part of me that says 'you have friends, and a job that you don't want to lose' says 'no.'
amarks: airlied is right
amarks: gentoo is the painful path, there is no denying that
amarks: especially on a slow machine
adam_yarter: airlined: I'm gonna try to backport the package.
adam_yarter: If that's a no go, I'll just go that route.
adam_yarter: lol, my 1 mb DSL line is gonna love it.
adam_yarter: amarks: a dual 1.6 ghz sparc would be slow.
adam_yarter: some sick part of me loves this machine though.
amarks: you are alone in that i can assure you :)
adam_yarter: why do you say that? lol
zcrayfish: Hey, I'd love to get my hands on a SPARC. :-p