05:28head8debian: So, im considering switching to arch, but im wondering if my card will be supported in it. Im currently in debian 8, and my card is not supported. It is a GEForce 7300 LE.( yes its old). Does a newer version of noveau support it?
05:28head8debian: Its probably a popular question. :)
05:29head8debian: No im not going to dissapear... :p
05:48pmoreau: head8debian: Which kernel version are you running in Debian 8 and what kind of issues are you experiencing?
06:49head8debian: pmoreau: Linux debian 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt9-3~deb8u1 (2015-04-24) x86_64 GNU/Linux
06:51head8debian: Without the Nvidia 304XX drivers it won't even boot. To install them i have to use the nomodeset option.
06:52head8debian: Did yo uget both lines? Our internet went down temporarly.
06:52head8debian: Debian wheezy ( kernel 3.2) worked great, untill i ran a game oru something then i had a GPU lockup.
06:53head8debian: Um, did you get those 4 lines? Stinking internet conection. :p
06:54head8debian: Also in wheezy the screen colors seemed duller.
06:54whompy: You can check http://people.freedesktop.org/~cbrill/dri-log/index.php to confirm the messages went through if curious.
06:55whompy: head8debian: Probably more of a distribution channel question, but the proprietary drivers typically work well in Arch from past experience. I have not used the 304 series, however.
06:56pmoreau: head8debian: Sorry, was afk.
06:58pmoreau: You could try one of the images here (https://nouveau.pmoreau.org) to find out if the card works better with a recent kernel, which should answer your question about Arch (they usually get the latest kernel about a week after its release)
07:00pmoreau: If you have some time, you could use netconsole to capture boot messages on another machine so we can try to debug that, or even better (but more time consuming), try to find the faulty commit by using git bisect.
07:02Karlton: 3.16... is that even a kernel that gets long term support? I thought it reached EOL
07:03Karlton: why would Debian use an EOL kernel?
09:17head8debian: I would first like to see if it can be solved in debian. ;) Im considering Arch...
09:17head8debian: Ask them why they still support 3.2 then in Wheezy? ;)
09:17head8debian: Its all with stability...
09:18head8debian: Is there a list of know supported cards?
09:32pmoreau: head8debian: Not that I know of, and it would be easier to do a list of known unsupported cards. Usually a family or chipset is well supported, just some really specific cards of those well supported families / chipsets do not want to work.
09:33pmoreau: There is a features matrix per chipset, but not per card.
09:54pmoreau: Was there some planned maintenance on freedesktop.org?
09:57pmoreau: RSpliet: NIce, your reclocking patches for NVA0 were merged. :)
10:24dardevelin: pmoreau, ping heyia remember me from yesterday ?
10:24dardevelin: I have built 4.1-rc6 nothing, I have built the 4.4 and nothing
10:24dardevelin: I then cloned the nouveau 2.6 switched to linux-4.1 branch
10:24dardevelin: but instead it built 3.18 or something
10:24dardevelin: which does remove the error
10:29pmoreau: I do :)
10:30pmoreau: Hum... So the 4.1 rebranded as 3.18 did remove the error, but not the other versions? Hum...
10:31pmoreau: dardevelin: How did you managed to build 4.4?!!
10:31pmoreau: Are you coming from the future? :D
10:33dardevelin: well it probably is actually 4.0.4 (I assume) is the one from kernel.org
10:33pmoreau: That would make more sense :)
10:33dardevelin: pmoreau, in fact this is very odd I have the logs
10:33dardevelin: one second so i can re link to you
10:35dardevelin: if I am not mistaken the reported memory is being 2GB instead of 4GB (let me confirm) I was kinda sleepy yesterday
10:35dardevelin: oh no never mind, i think rc6 did that while 3.18-nouveau still has the wrong vram
10:36dardevelin: but the SCHED message was 'decreased or hidden'
10:36dardevelin: btw the debian pastes are dmesg results (if it was not obvious)
10:36pmoreau: 3.18 still has the wrong ram, indeed
10:36dardevelin: in fact never mind. the SCHED message still there
10:36pmoreau: well, debian pastes do not want to load
10:36dardevelin:kills himself for missing it yesterday and making a big mess
10:37dardevelin: I can copy the text and repaste
10:37pmoreau: Don't worry
10:37pmoreau: The server doesn't load
10:37pmoreau: Not your particular paste
10:37dardevelin: i will paste in another one
10:37dardevelin: as it works here
10:37pmoreau: Are you able to connect to the Nouveau wiki?
10:38dardevelin: let me see
10:38dardevelin: this ? http://nouveau.freedesktop.org/wiki/ if so yes i can
10:39pmoreau: Oh, syslog_3.18nouveau.txt is actually 4.0.4!
10:40dardevelin: but i am pretty sure it is 3.18nouveau
10:41dardevelin: if the contents say otherwise it is because the naming of the kernel while building named it 3.18 and not actually being
10:42pmoreau: [ 0.000000] Linux version 4.0.4-customkernel
10:42pmoreau: And the the paste.zfg is 3.18.0-rc6, which I guess is 4.1-rc6 then, right?
10:42dardevelin: ok you know what I still have them built I will install them boot onto each and get the link
10:43pmoreau: Which would make sense as the RAM amount is correct
10:43pmoreau: I'll check for the other errors, you probably don't need to reboot
10:44pmoreau: dardevelin: You have a problem with the external firmware you're supplying
10:44pmoreau: [ 10.235415] nouveau 0000:01:00.0: Direct firmware load for nouveau/nv117_fuc409c failed with error -2
10:44pmoreau: [ 10.235419] nouveau 0000:01:00.0: Direct firmware load for nouveau/fuc409c failed with error -2
10:44dardevelin: well I have supplied firmware available in 'non-free'
10:44dardevelin: i think
10:45dardevelin: since during the install I have provided the external usb device with the packages for the non-free firmware
10:45pmoreau: Did you follow the wiki? (I have no idea how you are supposed to do that)
10:45dardevelin: I have not followed the wiki i followed advice i got here
10:46dardevelin: that new kernel should do the trick as the errors were already fixed
10:46dardevelin: so i did that first
10:46pmoreau: Ok, but if you tell Nouveau to use external firmware and do not provide them, it might be angry :D
10:47dardevelin: ok. but how do i check If i am providing them. remember I have provided firmware (and this assumption comes from the simple fact that I have given debian access to it during install)
10:48dardevelin: maybe debian actually only used the wifi and realtek firmware (as it's the only ones that are mentioned during the install)
10:48pmoreau: I *think* it has to be present in initramfs, but let me find the wiki page
10:50dardevelin: this talks about extraction from the blob
10:51dardevelin: i guess i might have to read it all to understand 'where is the blob'
10:51dardevelin: so that i can extract
10:51pmoreau: blob == Nvidia proprietary driver
10:52dardevelin: ok so gtx850m would be 340.32 still ?
10:52pmoreau: I'm not 100% sure how you are supposed to extract the firmware for GM107
10:52pmoreau: iirc Ilia's script isn't working, or maybe it's for GM20x
10:52pmoreau: I'll check
10:53dardevelin: i appreciate that pmoreau
10:53huehner: i think gm20x was the ones being more strict for signing but not 100% either
10:54pmoreau: GM20x definitely needs signed firmware, while GM10x doesn't (or at least not for PGRAPH)
10:55pmoreau: But I think even for GM107 they changed where they were storing the firmwares inside the blob, and Ilia couldn't find the new location
10:55pmoreau: Or rather, then changed that for new blob versions
10:56huehner: pmoreau: any news on them providing firmware topic? i think i did not see any after the initial 'signing required' heads-up mail
10:56pmoreau: gnurou: Was talking a few weeks ago that he was going to investigate that
10:57pmoreau: But otherwise, no progress
10:57dardevelin: i am currently downloading the envytools
11:00huehner: pmoreau: thx
11:00huehner: pmoreau: any pratical improvement on gm206 with bens work for 4.1 over 4.0? i seem some commits but unclear if/what those provides in practice over 4.0
11:02dardevelin: no special tricks for building envytools right ?
11:02dardevelin: could not find FLEX_EXECUTABLE ?
11:03pmoreau: huehner: I don't know... You should ask him directly :)
11:03pmoreau: dardevelin: Well, just run cmake and install the packages that are missing I would say
11:03huehner: pmoreau: fair enough ;)
11:04dardevelin: what is flex though ?
11:04huehner: dardevelin: kind of code generator for parsing things
11:04huehner: dardevelin: probably envytools has definition and at compile time it runs it to generate c code which later is compiled
11:05pmoreau: https://bugs.freedesktop.org/show_bug.cgi?id=89558 they aren't talking about firmware, but it is being loaded. Not sure where it comes from
11:05dardevelin: oh like bison ?
11:05huehner: dardevelin: yes , both needed in fact
11:05huehner: dardevelin: check the README.rst in the repo
11:05huehner: dardevelin: it points to those 2 + a few more things needed for compiling
11:06pmoreau: dardevelin: You should add yourself to the bug report, and you could ask how to get the firmware
11:07dardevelin: cmake libxml2 flex bison and pkg-config
11:09Karlton: even if Nvidia releases the firmware, how would we address the issue of running a Nvidia card without blobs?
11:10dardevelin:guesses rewriting the blobs ?
11:10Karlton: because wouldn't the RE firmware need to be signed?
11:10pmoreau: If NVidia releases the firmware, we would use their
11:11dardevelin: uh libxml2 has a bug
11:11dardevelin: not sure if this will affect envytools
11:11dardevelin: arsing Found/Fixed information... Done
11:11dardevelin: grave bugs of libxml2 (2.9.1+dfsg1-5 → 2.9.2+dfsg1-3) <Forwarded>
11:11dardevelin: b1 - #766884 - libxml2: "validity error : ID ... already defined" errors with xmllint --noent
11:12pmoreau: Karlton: But you're right, the RE'ed firmware would need to be signed.
11:13dardevelin: I guess I will have to delay this process a little longer until libxml2 is patched/fixed as at the moment it renders unusable
11:13pmoreau: No problem
11:13dardevelin: in other words will try see if I can patch libxml2 or something
11:14dardevelin:hates his life (prefers this any day to web development) something I should be doing right now
11:14Karlton: pmoreau: ugh that is going to make using free software in the future even harder :(
11:14dardevelin: Karlton, i believe that's exactly the point/idea
11:15dardevelin: and eventually just have everything 'sas'
11:15dardevelin: you will pay a monthly membership and you will get a laptop (chromebook style) will alls apps in the 'cloud' and they just magically update and it will look like tv channels subscription
11:16dardevelin: a package that has it all. you just pay a single 'medium-to-large' fee
11:16dardevelin: and every time someone builds a 'traditional' desktop tool as a web tool this gets closer
11:17pmoreau: Karlton: Well, the firmware will be less free, but not necessarily the software, except if you consider firmware as software
11:18pmoreau: (As long as the signed firmware are released.... Which is... not the case yet...)
11:18huehner: and maybe we can use parts of it i.e. reclocking calling it instead of re-ing? but just hoping here...
11:19Karlton: ethically speaking, non-free firmware is no different
11:19dardevelin: pmoreau, i am giving it a shot if it breaks it breaks
11:19dardevelin: built envytools
11:20dardevelin: Karlton, the line is very blurry. is a transistor 'free or non free' having the schematics or not will never help you 'read/inspect' how it is actually made
11:21Karlton: dardevelin: yeah hardware is different when it is not designed to be chagned
11:22dardevelin: Karlton, so in other words. think of firmware like little cables that connect 2 bits of hardware
11:22dardevelin: even though it may have 2 little different combos is just that. and to work it always needs to be connected
11:23dardevelin: so even though you can go and plug your own cables there, there are no _real_ differences between your cable and the original one. to which point is it really software ?
11:23dardevelin: anyways I assume this is not the place for this discussion to happen... sorry about that guys
11:24Karlton: yes, perhaps code for microcontrollers is not the same
11:24dardevelin: that's the thing. arguably micro controllers have more flexibility then a lot of firmware out there for specific hardware.
11:25dardevelin: pmoreau, i managed to build envytools
11:26dardevelin: so I am guessing I download the proprietary driver and run that fancy command
11:26dardevelin: and the drive id want to take is 340.
11:26pmoreau: Most likely not
11:27pmoreau: Except if they support GM107 but I doubt that
11:27dardevelin: so 337 ?
11:27pmoreau: I don't know by heart which chipsets are supported for every blob version :D
11:28Karlton: dardevelin: nonetheless, it should be published considering how complex they are becoming and should be able to place it even if it is an exact copy
11:28dardevelin: pmoreau, sorry. just trying to get you to hold my hand
11:28pmoreau: You could check if 340.xx supports GM107, or ask on the bug report and hope someone will answer (most likely Ilia would)
11:29pmoreau: dardevelin: Except in that case I don't know much more than you do :/
11:31dardevelin: pmoreau, alright I will hold it for a little longer. thank you very much
11:32dardevelin: I really do need assistance in this process as I am not comfortable at all
11:32dardevelin:is already in a process 'with asus' for windows refund. consumer protection i don't want to f up irreversibly
11:33dardevelin: they will laugh if i f up before getting my refund.
11:33dardevelin: win8.1 is super tricky to reject
12:09mlankhorst: how do you object then? :p
12:39head8debian: pamaury: Is there a noveau list of non-supported card?
12:40tobijk: head8debian: no card is supported, cards may only work ;-)
12:41head8debian: I know. ;) But there are ones that usually work and ones that don't...
12:41head8debian: I remember reading a list somewhere, but i cannot find it.
12:43tobijk: head8debian: if thee is a list somewhere about which card works and which does not, that is outdated anyway, nobody really cares for these lists
13:00pmoreau: head8debian: NVE6/7 have a high probability of not working well with Nouveau, otherwise, apart from Maxwell, not sure
13:05tobijk: pmoreau: hmm nve7 is fine in my eyes :D
13:07crazedpsyc: I'm one of the many affected by this :(
13:08crazedpsyc: head8debian: this is probably what you were thinking of: http://nouveau.freedesktop.org/wiki/FeatureMatrix/
13:08tobijk: crazedpsyc: well many nve7's are working well, dadly not all
13:13pmoreau: tobijk: I was thinking of the many bugs on NVE7, but I still know that some may work fine, hence my "high probability of not working", not "it won't work" :)
13:14tobijk: pmoreau: the ddr3 ones work, the gddr5 ones dont, thats how i see it
13:15pmoreau: I didn't know or forget that only GDDR5 ones were affected
13:16tobijk: we had several problems with the ddr3 ones as well but thats fine now (afar from a new little bug in 4.1-rc which i havent yet gotten to reproduce reliably)
16:17newb123: Is there a quick way to just recompile nouveau kernel module for a particular version of ubuntu? I keep on running into a vga_switcheroo_fini_domain_pm_ops implicit-function-declaration error.
16:56moochmcgee: does anybody know what the subsystem ids for a riva 128 are?
16:56moochmcgee: win9x drivers seem to check this and fail
23:15Guest72155: rats.. sorry, I didn't tell nickserv my name fast enough :/
23:17Guest72155: I have a gtx 970, and I'm trying to get nouveau working on it.
23:19Guest72155: I'm having network issues, please bear with me.
23:19Guest72155: anyways, the xorg log files says that KMS is not loaded and lsmod says nouveau is not loaded
23:19Guest72155: if I try to modprobe nouveau, I get an immediate black screen
23:20Guest72155: I don't think the machine is hung or anything, I think it just turned off the video
23:20Guest72155: any ideas?
23:21sri: much better.
23:43huehner: sri: what kernel version? how if your monitor connected?
23:44huehner: sri: what is dmesg output of the kernel when it tries to load nouveau? (note you need to check via maybe via ssh or so) if you have just 'black screen'