14:29karolherbst: uiui, HDMI audio workaround: https://devtalk.nvidia.com/default/topic/1024022/linux/gtx-1060-no-audio-over-hdmi-only-hda-intel-detected-azalia/post/5211273/#5211273
17:31pmoreau: And back to my initial patch about handling U(S)32/64 conversions: looks like I am doing the wrong thing for signed values, so time to clean up the patch and send it to the ML once it is fixed, along with other 64-bit integer patches.
17:36|nfecteD: Is the GeForce 1030 (GP108) supposed to be supported in kernel 4.14-rc2 or is it still WIP?
17:39pmoreau: |nfecteD: modesetting for GP108 should be part of 4.14. I don’t know whether the pull request was already sent to airlied, or if it will come for a later RC>
17:40pmoreau: I don’t think NVIDIA released any firmware for the GP108, so you won’t get any hardware acceleration.
17:43|nfecteD: That explains my issue then ;P
17:43|nfecteD: thanks for the answer
17:45pmoreau: You’re welcome :-)
17:47|nfecteD: I'm guessing "display only" means no acceleration then
17:48pmoreau: modesetting only, yes
17:49pmoreau: NVIDIA hasn’t released the firmwares, I just checked. So until they do, there won’t be any acceleration available.
17:51|nfecteD: The chip was released in may i think, so who knows how long that will take
17:52|nfecteD: is it just the GP108 you are waiting for the firmware for or are the more chips?
18:00imirkin_: |nfecteD: we're waiting for pmu firmware from all chips that require signed firmware.
18:01imirkin_: without which we don't reclock, without which 'acceleration' is not really possible.
18:16dan2: i just rebooted my debian testing box and now gdm3 won't come up... can someone help me diagnose this?
18:40imirkin_: dan2: pastebin dmesg and xorg logs
18:42imirkin_: robclark: btw, i remembered - pretty sure it was lynxeye who was messing around with nvidia pcie boards on arm
18:46robclark: imirkin, I think it was a "bios" issue somehow.. we had similar issue w/ radeon card.. not *entirely* sure what bios could screw up such that option rom / GOP works (and basic nouveau stuff like reading edid worked).. I'll have to check back w/ those guys and see if they made progress
18:48imirkin_: robclark: yeah, obviously i have no clue. but in case you wanted to reach out.
18:48imirkin_: oh, i wonder if MMIO works but DMA doesn't?
18:49imirkin_: and/or the vram mapping thing
18:49imirkin_: whereas the GOP might access the vram via the peephole
18:49imirkin_: [which is mmio access to vram]
18:49imirkin_: where "mmio" stands for "BAR1", not "all memory-mapped io"
18:50imirkin_: (or is it BAR2? either way, the one with all the registers.)
18:50robclark: yeah, iirc they were suspicious about dma
18:51robclark: anyways, I think it is after their demo so not a fire anymore
18:51imirkin_: hopefully their demo wasn't "nouveau on arm64"
18:51imirkin_: but rather their crazy-bios thing
18:52dan2: imirkin: https://hastebin.com/exitaponoj.go
18:53robclark: imirkin, it was a new aarch64 motherboard w/ pci and such things... iirc 24x core aarch64 server chip..
18:53imirkin_: dan2: how is your monitor connected btw?
18:53imirkin_: robclark: fancy.
18:54imirkin_: robclark: hopefully not arm-be, otherwise they're in for a little surprise.
18:54dan2: i've got 2 external monitors connected via DP on a dock
18:54dan2: and 1 internal display
18:55dan2: but same problem happens when laptop is undocked
18:55dan2: was working fine until I rebooted ... :/
18:55imirkin_: dan2: you came in with questions the other day, didn't you? you had the monitors that had to be flipped to DP 1.1 mode?
18:55dan2: yeah that was me :)
18:55imirkin_: so what's the problem exactly?
18:56imirkin_: (nouveau decides that its console resolution should be 2560x1440 -- that seems high for a laptop panel)
18:56imirkin_: [or rather, fbcon decides this]
18:58robclark: imirkin, heheh, no, BE would be insane and I would have just told them to enjoy their llvmpipe (if it worked) ;-)
18:59dan2: that's the resolution of my external monitors. computer boots up, but gdm3 never starts. If I hit ctrl-alt-f2, I can get to a console...
19:00imirkin_: wayland or X?
19:00dan2: i was using wayland, but I have xfce/X installed too. I guess gdm3 probably uses wayland
19:00imirkin_: well, if your mesa install is at all off, none of the wayland stuff will work i think
19:01imirkin_: i have no clue how to diagnose issues in way-land.
19:01imirkin_: only in x-land
19:01dan2: heh.. i'm trying to figure out what changed. gonna look in my apt log
20:49Guepi: Hi there !
20:50Guepi: I'd like to know if the patch : [Nouveau] [PATCH v4 0/3] drm/nouveau: set DMA mask before mapping scratch page (https://lists.freedesktop.org/archives/nouveau/2016-September/026074.html) is already included in nouveau which is in Fedora 26 ?
20:51imirkin_: dunno if that got merged at all
20:52imirkin_: (not that i know that it didn't...)
20:52imirkin_: what kernel does F26 ship?
20:52Guepi: because I've got the same error (I guess)
20:53Guepi: kernel 4.12
20:53Guepi: 4.12.14-300.fc26.x86_64 to be more precise
20:53imirkin_: ah no. looks like it was merged
20:53imirkin_: definitely in that kernel
20:54Guepi: oh ok.
20:54imirkin_: could have come back though.
20:56Guepi: so, if I'm right, I should not have the same issue : [ 2.806919] nouveau 0000:01:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x0000
20:57imirkin_: that's wholly unrelated
20:57imirkin_: can you pastebin your dmesg?
20:59Guepi: here is the dmesg full
21:02imirkin_: feels like you have a variant of the ACPI issue where we don't read the bios in correctly
21:03Guepi: maybe yes. but it seems that ACPI is also well read at other moment
21:03imirkin_: [ 2.670294] ACPI: \_SB_.PCI0.PEGP.VGA_: failed to evaluate _DSM
21:03imirkin_: [ 2.670299] VGA switcheroo: detected DSM switching method \_SB_.PCI0.PEGP.VGA_ handle
21:03imirkin_: normally there's a _ROM method there as well which returns the ROM
21:03imirkin_: we read it out in a slightly out-of-spec way which normally works
21:03imirkin_: mind booting with nouveau.debug=trace ?
21:04Guepi: no problem
21:04Guepi: right now ?
21:04imirkin_: whenever you like :)
21:04imirkin_: i don't really care whether your issue gets resolved or not... you're the one who's going to have to do the work and supply additional information
21:05Guepi: and where did I get the trace ?
21:05Guepi: dmesg ?
21:05imirkin_: oh, it's just an additional kernel parameter
21:05imirkin_: it'll dump more messages into dmesg, yeah
21:06Guepi: see you soon :)
21:06imirkin_: which will hopefully help me confirm that it's indeed the issue i'm thinking of
21:06imirkin_: (or, conversely, that it's not)
21:13Guepi: I'm back
21:15Guepi: here is the dmesg with the trace https://pastebin.com/iJKd3K0t
21:16imirkin_: is there anything in any way weird about this machine that you can think of?
21:16Guepi: I've got an issue with the battery
21:17imirkin_: ok, but you didn't like install coreboot on it
21:17imirkin_: or try to replace the acpi blob
21:17imirkin_: or anything else 'odd'
21:17Guepi: no nothin like that
21:17imirkin_: that NPDE signature = "HYB$" which is ... odd
21:18Guepi: my brother got this computer before, but .. I'm pretty sure he didn't even know what acpi is
21:19imirkin_: so, it _is_ able to get a blob from ACPI, but then it says "fetch failed". with a score of 0, that means the checksum doesn't match either.
21:20imirkin_: case 0x4544504e: /* NPDE */
21:20imirkin_: so that's what it's supposed to get
21:20imirkin_: and instead it gets 0x24425948
21:20imirkin_: (which maps to "HYB$")
21:21imirkin_: although that doesn't matter
21:22imirkin_: hm. "fetch failed" means that shadow_fetch() fails??
21:23imirkin_: can you do an acpi dump
21:23Guepi: how can I do that ?
21:23imirkin_: i suspect there's a weird bug in the ACPI
21:23imirkin_: and we're tripping over it
21:23imirkin_: acpidump + acpixtract or something?
21:23imirkin_: i think acpidump is enough
21:26imirkin_: in the PCIR it says that it should be 63488 bytes, and it apparently can't read that many
21:28Guepi: here is the acpidump https://pastebin.com/E4xTBvpg
21:39imirkin_: unfortunately i don't have time to look at it right now, sorry
21:40Guepi: no problem.
21:42Guepi: what can I do ? just let that on IRC ? send an email to a mailing list ? doing nothing 'cause you send it to a friend of you :) ?
21:44imirkin_:has no friends
21:44imirkin_: like kip drordy
21:44imirkin_: anyways - file a bug at bugs.freedesktop.org, xorg -> Drivers/nouveau
21:45imirkin_: attach both a copy of that dmesg (with trace enabled) and your acpidump
21:45imirkin_: that way the info doesn't get lost
21:45Guepi: ok ! Thank you very much !
21:45Guepi: merci merci merci
22:09Guepi: imirkin, when you got friends, you could tell to some of them to have a sight at https://bugs.freedesktop.org/show_bug.cgi?id=103033 just after the movie theater :)
22:20Guepi: bye bye everybody - time to go to bed for me
23:51duttasankha: Hello Everyone! Myself and I am a researcher doing my PhD in UC Riverside. I have been working on GPU (primarily Nvidia) for quite sometime but mainly from the application perspective. Recently I started working on the security aspects of GPU which leads me to envytools. I am going through envytools docs and codes for past few months and it is really extremely interesting and completely new to me. However, sometime I feel that some help from
23:51duttasankha: the experts would have been much appreciated and so I joined the nouveau channel. I hope I would be able to get feedback from the members and in turn I would be able to contribute to the community as well. As a complete novice I might ask some very basic question and I apologize for that. Thank you everyone.
23:52imirkin_: the novice questions are the hardest to answer
23:57duttasankha: I have a query about the purpose of the codes inside the directories like demmt, hwtest, nvhw? Their description are not provided inside the git page.
23:57imirkin_: demmt is for decoding mmt logs generated by valgrind-mmt
23:58imirkin_: hwtest is for generating models of the hardware and having infra for comparing the model to the real thing
23:58imirkin_: not sure about nvhw offhand
23:58imirkin_: helpers for hwtest perhaps?
23:58imirkin_: common functions and whatnot