04:05television7: I have an NV50 family GPU (fx1800M) with a regression
04:05television7: it used to run fine with Wayland and gnome on some kernel version from year 2022 but I updated to some year 2024 kernel and now I can’t use my computer
04:06television7: what ends up happening is it boots and I see the login screen and I can move my cursor around but nothing is clickable
04:06television7: The dmesg has some interesting messages but I can’t remember them offhand
04:06television7: anyone wanna help me out?
04:10dwfreed: If you can see everything, it doesn't really sound like this is a graphics issue
04:10soreau: well, if you click and it actually clicks but doesn't update the graphics so it seems like you're not clicking..
04:11soreau: if you type the password and try to login 'blindly' does it login at least?
04:11television7: I actually discovered this issue years ago and back then I “fixed” it by running an older kernel version
04:11television7: I’ll try that soreau, not sure
04:11soreau: have you tried starting a compositor directly from a tty without login DM?
04:12television7: how do I do that?
04:12soreau: just go to a tty, login and run a compositor like weston
04:13television7: I have Wayland and gnome, how can I get that started?
04:13soreau: no idea
04:13television7: hmm
04:13soreau: but the compositor is called mutter
04:13soreau: at least, last I heard :P
04:14television7: I’ll be back with “screenshots”
04:14soreau: weston should be available in most distro repo I think
04:14television7: I’m on Archlinux if it makes a difference
04:14tiredchiku[d]: dbus-run-session -- gnome-shell --display-server --wayland
04:15tiredchiku[d]: to run gnome from a tty
04:15television7: Oh thanks much
04:39television7: Ok I’m back
04:44television7: so, tiredchiku[d] - running that i get this : blob:https://imgur.com/5c081178-ccbc-4a63-a820-01d9bef3d8a4
04:44television7: oops...
04:44television7: https://i.imgur.com/ViI0Ljp.jpeg
04:45television7: dbus-daemon[1338]: [session uid=1000 pid=1338] Successfully activated service
04:45television7: 'org.gtk.ufs.Daemon'
04:45television7: libmutter-Message: 23:35:36.039: Running GNOME Shell (using mutter 47.3) as a Wayland display server
04:45television7: and then thats it, nothing more, its frozen like that
04:45television7: i think this is more interesting though - look at dmesg - https://i.imgur.com/42ADe94.jpeg
04:46television7: drm: failed to create ce channel, -22
04:46television7: ?
04:46television7: its not a hardware problem because downgrading to whatever kernel that was in 2022 fixes it lol
04:47television7: so, soreau i tried logging in blind and it doesn't log in
04:47television7: and i did the thing chiku said to try and that just freezes the screen at that running gnome shell message i pasted/ screenshotted above
04:52television7: for reference im currently on 6.12.4
04:52television7: and my system was working fine on kernel 6.1.2
05:01tiredchiku[d]: could you try a 6.6 lts?
05:03television: i wonder if i can do that without breaking something else in my system
05:04television: https://archlinux.org/packages/core/x86_64/linux-lts/
05:04television: is this good enough?
05:04television: 6.6.65
05:04tiredchiku[d]: yup
05:05tiredchiku[d]: if 6.6 is also fine, I'd suspect it to be a regression caused by GPS support being added
05:07asdqueerfromeu[d]: tiredchiku[d]: *GSP
05:07tiredchiku[d]: yeh, that
05:07television: whats gsp?
05:07tiredchiku[d]: race conditions got me this time
05:07television: lol race conditions on the keyboard? :>
05:07tiredchiku[d]: GSP is firmware for newer cards (Turing generation and above)
05:07tiredchiku[d]: yeah!
05:07television: happens to the best of us
05:10television: i remember some years ago, like ... idk, year 2016? i had another issue on this gpu that you fellas eventually resolved, it was a funky issue where in games, it would flicker between the current frame, and a frame from like 30 seconds ago, and it would flash between the two frames like 60 times a second (new, old, new, old, etc..) it took like 2
05:10television: months for it to get fixed but boy was i glad
05:11television: i appreciate y'all
05:38television: tiredchiku[d] - same issue on 6.6.65
05:38television: uname -a says Linux firearch 6.6.65-1-lts #1 SMP PREEMPT_DYNAMIC Wed, 11 Dec 2024 15:35:54 +0000 x86_64 GNU/Linux
05:43television: here's a full dmesg - https://paste.ee/p/PQNRK
05:43tiredchiku[d]: then it's likely an unrelated regression, hmm
05:46television: i can run any tests you want just guide me :>
05:51tiredchiku[d]: well, one idea would be to run each major version between 6.1 and 6.6
05:51television: i shoulda probably popped in here two years ago when this was all still fresh but you know how it goes when you put a "temporary" band-aid fix on something it becomes a "permanent" fix (holding my system back to an old kernel i mean)
05:51tiredchiku[d]: to identify which one introduced the regression
05:51television: hmm lemme see if i can try without breaking anything else
05:51tiredchiku[d]: then once we identify a major version, we'd have to identify which patch
05:52tiredchiku[d]: and then get that fixed and submit it upstream
05:52tiredchiku[d]: it's a long process e-e
05:52tiredchiku[d]: I'd have gladly done it but hardware access
05:53television: how would you go about it? i was thinking i'd download old arch packages of the kernel and install them one by one
05:53television: do you have a particular test distro you use or something
05:54tiredchiku[d]: no, I'm also on arch
05:54tiredchiku[d]: that's a valid approach, yeah, using archive.archlinux.org
05:55tiredchiku[d]: but at some point we'd have to build our own kernels to test individual commits
05:56tiredchiku[d]: fwiw, if you use discord by any chance, you can join this channel there too
05:56tiredchiku[d]: https://discord.gg/JPMsrZJW
05:56television: i've built and installed my own kernel before but that was eons ago
05:56television: you might have to refresh my memory :p
05:56television: i do use discord but not as often
05:56television: i can be found more readily on Telegram and IRC lol
05:56television: but i'll join
05:56tiredchiku[d]: all good, I can maybe even give you builds to try out once we identify the minor versions
05:57tiredchiku[d]: but first, go through 6.1, 6.2, 6.3, and so on to identify which major version introduces the regression and which is the last "good" version
06:00television: okay will do
06:01television: i bet its 6.2
06:05television: what on earth
06:06television: booting 6.2 i get mount /new_root: unknown filesystem type `ext4`
06:06television: lol
06:06unixbrain[d]: what did i do wrong? :D
06:07tiredchiku[d]: [sweating](https://cdn.discordapp.com/emojis/1233877778630447135.webp?size=48&name=sweating)
06:07television: lts boots fine but 6.2 has that dumb error
06:07tiredchiku[d]: does your initramfs have that module
06:08unixbrain[d]: how do i check?
06:08tiredchiku[d]: if using mkinitcpio, `filesystems` hook
06:08tiredchiku[d]: you can force it by adding ext4 to the MODULES array in mkinitcpio.conf
06:08unixbrain[d]: i saw -> Running build hook: [filesystems]
06:08unixbrain[d]: when installing 6.2
06:09tiredchiku[d]: and regenerating the initrd
06:09tiredchiku[d]: mmh
06:09tiredchiku[d]: unfun
06:10unixbrain[d]: modules array is empty
06:10unixbrain[d]: i'll add ext4
06:12tiredchiku[d]: could check if it exists with `ls /lib/modules/6.2-whatever/kernel/fs/ext4/ext4.ko.zst`
06:13unixbrain[d]: 532K -rw-r--r-- 1 root root 531K Feb 20 2023 /lib/modules/6.2.0-arch1-1/kernel/fs/ext4/ext4.ko.zst
06:13unixbrain[d]: well its there now if it wasn't before (i did the mkinitcpio thing)
06:14tiredchiku[d]: that dir is owned by the kernel package
06:15unixbrain[d]: https://cdn.discordapp.com/attachments/1034184951790305330/1319186420174163968/IMG_7365.HEIC?ex=67650b97&is=6763ba17&hm=5dd7f8483bcfbccd38c1d23af4e44b9273bd6097f87e673297d20a046fa43c7c&
06:16unixbrain[d]: https://cdn.discordapp.com/attachments/1034184951790305330/1319186572243107860/IMG_7365.jpg?ex=67650bbc&is=6763ba3c&hm=b7457fb1d261d5c5c647f122e0a163c2b362dea892c87741a21c80557157c7c5&
06:16unixbrain[d]: yeah... same issue.. what gives?
06:16tiredchiku[d]: pain
06:17tiredchiku[d]: are you able to manually mount your rootfs to /new_root?
06:17unixbrain[d]: like what, `mount /dev/sda1 /new_root`?
06:18unixbrain[d]: if i could do that, how do i tell it to continue booting after that
06:22unixbrain[d]: do i need the -headers package too?
06:27unixbrain[d]: tiredchiku[d]: nvm it doesn't accept keyboard input lmao
06:28unixbrain[d]: i wonder if i should make my own test images instead of this
06:28unixbrain[d]: i could use archive.archlinux.org to make any point in time i want
06:29unixbrain[d]: it'd just take longer due to all the downloaded packages and other moving parts
06:29unixbrain[d]: i do have an iODD i could use to experiment with
06:31unixbrain[d]: i'm trying 6.1.2 now just to see if that somehow works
06:32unixbrain[d]: yeah 6.1.2 works ... lol
06:36tiredchiku[d]: unixbrain[d]: yeah, just do that and then `exit` emergency shell
06:36tiredchiku[d]: unixbrain[d]: sad
06:36tiredchiku[d]: try 6.3, maybe?
06:48unixbrain[d]: okay so one thing, im on 6.1.2 and i can't get the UI to respond, but the `drm: failed to create ce channel, -22`
06:48unixbrain[d]: error is gone
06:48unixbrain[d]: so i guess this isn't 100% a nouveau issue
06:48unixbrain[d]: i wonder what else is involved then?
06:49unixbrain[d]: i guess i gotta use the archlinux archive and take it day-by-day until i find the exact change that caused it
06:59unixbrain[d]: oh wait i just remembered, my root fs is btrfs
06:59unixbrain[d]: so why was it complaining about ext4?
06:59unixbrain[d]: regardless... im still going to get to the bottom of this lol
07:04tiredchiku[d]: unixbrain[d]: might have to add that module then
07:04tiredchiku[d]: and the corresponding binary
07:41unixbrain[d]: but anyway
07:41unixbrain[d]: 6.1.2 isn't "working" as in "able to log in to the desktop"
07:41unixbrain[d]: its exhibiting the same issue as i was describing at square 1
07:41unixbrain[d]: which means i need to probably downgrade wayland too?
08:01awilfox[d]: What kernel were you using when it did work?
08:06television: 6.1.2
08:10awilfox[d]: Interesting, so perhaps it is a change in Mesa.
12:28tiredchiku[d]: if I wanted to refer to the vulkan instance inside src/nouveau/vulkan/nvkmd
12:28tiredchiku[d]: would I have to extend the struct to include it?
13:21abris: Hello, I'm reading the man page of nouveau where it says that "nouveau is an Xorg driver for NVIDIA video cards". I use gnome-shell wayland compositor with nouveau. Can you please help me understand how nouveau is X specific and how it supports wayland? Thank you for your help.
13:32awilfox[d]: nouveau is the name of a (now outdated) "DDX" or X11 device driver, but it is also the name of a Linux kernel "DRM" (Direct Rendering Manager) driver.
13:33awilfox[d]: It is the latter which is used in Wayland, and on modern X via modesetting.
14:00abris: Thank you so much, I see it now :)
14:43dwlsalmeida[d]: airlied[d]: problem is, I am not really sure this matters,
14:43dwlsalmeida[d]: things like:
14:43dwlsalmeida[d]: [0x00000002] HDR 20018080 subch 4 NINC
14:43dwlsalmeida[d]: mthd 0200 NVC5B0_SET_APPLICATION_ID
14:43dwlsalmeida[d]: .ID = H264
14:43dwlsalmeida[d]: vs
14:43dwlsalmeida[d]: [0x00000002] HDR 80038080 subch 4 IMMD
14:43dwlsalmeida[d]: mthd 0200 NVC5B0_SET_APPLICATION_ID
14:43dwlsalmeida[d]: .ID = H264
14:43dwlsalmeida[d]: It's been a while I wrote the nv_push_rs thing, but I think `NINC` and `IMMD` would be equivalent in this case?
14:44marysaka[d]: it shouldn't change much
14:44marysaka[d]: increment is just nicer if you start writing thing in sequence that save up quite a bit of bytes
14:45dwlsalmeida[d]: I don't think the increment part is ever used for the video stuff..
14:45dwlsalmeida[d]: plus vk_push_print can actually make sense of it
14:46dwlsalmeida[d]: so I assume that so can the GPU
14:46dwlsalmeida[d]: I wonder if Dave meant there's a problem in sending stuff from Rust and C in the same submission
14:46dwlsalmeida[d]: airlied[d]: as in here ^
15:50babblebones[d]: Excuse me for asking this here especially when there's open issues already but they have been spooling some time now without much hope of a resolution.
15:50babblebones[d]: Is anyone here familiar enough with the Nvidia-open kernel modules that they might be able to patch in support for, presumably, 6bpp DSC negotiation? The bigscreen beyond has a small free floating patch to do this on amdgpu but no equivalent for nvidia chips despite the majority of BSB owners being nvidia only.
15:50babblebones[d]: We have users who would gladly test and confirm patched module function if anybody here may have the time to look into this
19:26gfxstrand[d]: dwlsalmeida[d]: Yeah, that shouldn't matter as long as they decode the same.
19:30gfxstrand[d]: awilfox[d]: It's also the name of the umbrella effort to provide open Nvidia drivers in general. Even though some of the more recent efforts have dropped "nouveau" from the explicit branding (do we do branding?), they're still considered part of the broader nouveau project.
19:31gfxstrand[d]: Someone really should rework the nouveau homepage.
19:31awilfox[d]: Well, I was speaking in the context of what a (4) manpage could mean.
19:31gfxstrand[d]: Yeah...
19:32gfxstrand[d]: But I think the wiki still says that, too. 😳
19:33gfxstrand[d]: The wiki is so out of date
20:22dwlsalmeida[d]: I think it's this WFI business again
20:23dwlsalmeida[d]: even if you remove all video-related pushbufs, it still crashes, I think the video queue just dislikes this:
20:23dwlsalmeida[d]: ` result = nvk_queue_state_update(dev, &queue->state);`
20:24dwlsalmeida[d]: i.e.:
20:24dwlsalmeida[d]: static VkResult
20:24dwlsalmeida[d]: nvk_queue_submit(struct vk_queue *vk_queue,
20:24dwlsalmeida[d]: struct vk_queue_submit *submit)
20:24dwlsalmeida[d]: {
20:24dwlsalmeida[d]: struct nvk_queue *queue = container_of(vk_queue, struct nvk_queue, vk);
20:24dwlsalmeida[d]: struct nvk_device *dev = nvk_queue_device(queue);
20:24dwlsalmeida[d]: VkResult result;
20:24dwlsalmeida[d]: if (vk_queue_is_lost(&queue->vk))
20:24dwlsalmeida[d]: return VK_ERROR_DEVICE_LOST;
20:24dwlsalmeida[d]: if (submit->buffer_bind_count > 0 ||
20:24dwlsalmeida[d]: submit->image_bind_count > 0 ||
20:24dwlsalmeida[d]: submit->image_opaque_bind_count > 0) {
20:24dwlsalmeida[d]: assert(submit->command_buffer_count == 0);
20:24dwlsalmeida[d]: result = nvk_queue_submit_bind(queue, submit);
20:24dwlsalmeida[d]: if (result != VK_SUCCESS)
20:24dwlsalmeida[d]: return vk_queue_set_lost(&queue->vk, "Bind operation failed");
20:24dwlsalmeida[d]: } else {
20:24dwlsalmeida[d]: result = nvk_queue_state_update(dev, &queue->state); <----------
20:29dwlsalmeida[d]: yep, that's it
20:43dwlsalmeida[d]: Omg thank god this is working in Rust too \o/
20:43dwlsalmeida[d]: What a huge relief
22:26airlied[d]: nice!
23:37gfxstrand[d]: dwlsalmeida[d]: Makes sense! I think I modified that to only actually emit stuff if one of the 3D or compute engines is active. I'm surprised it's doing anything on video. But also, if it is doing anything on video, that something is probably wrong. 😅
23:38dwlsalmeida[d]: Gotta figure out what’s changed in nil_rs, this GOB stuff is not right again
23:40dwlsalmeida[d]: ðŸ˜
23:44dwlsalmeida[d]: This is the broken one:
23:44dwlsalmeida[d]: 21 [0x00000027] HDR 20088100 subch 4 NINC
23:44dwlsalmeida[d]: 22 mthd 0400 NVC1B5_OFFSET_IN_UPPER
23:44dwlsalmeida[d]: 23 .UPPER = (0x3f)
23:44dwlsalmeida[d]: 24 mthd 0404 NVC1B5_OFFSET_IN_LOWER
23:44dwlsalmeida[d]: 25 .VALUE = (0xfff41000)
23:44dwlsalmeida[d]: 26 mthd 0408 NVC1B5_OFFSET_OUT_UPPER
23:44dwlsalmeida[d]: 27 .UPPER = (0x3f)
23:44dwlsalmeida[d]: 28 mthd 040c NVC1B5_OFFSET_OUT_LOWER
23:44dwlsalmeida[d]: 29 .VALUE = (0xffda0000)
23:44dwlsalmeida[d]: 30 mthd 0410 NVC1B5_PITCH_IN
23:44dwlsalmeida[d]: 31 .VALUE = (0x40)
23:44dwlsalmeida[d]: 32 mthd 0414 NVC1B5_PITCH_OUT
23:44dwlsalmeida[d]: 33 .VALUE = (0x40)
23:44dwlsalmeida[d]: 34 mthd 0418 NVC1B5_LINE_LENGTH_IN
23:44dwlsalmeida[d]: 35 .VALUE = (0x20)
23:44dwlsalmeida[d]: 36 mthd 041c NVC1B5_LINE_COUNT
23:44dwlsalmeida[d]: 37 .VALUE = (0x20)
23:44dwlsalmeida[d]: 38
23:44dwlsalmeida[d]: 39 [0x00000030] HDR 200581ca subch 4 NINC
23:44dwlsalmeida[d]: 40 mthd 0728 NVC1B5_SET_SRC_BLOCK_SIZE
23:44dwlsalmeida[d]: 41 .WIDTH = ONE_GOB
23:44dwlsalmeida[d]: 42 .HEIGHT = FOUR_GOBS
23:44dwlsalmeida[d]: 43 .DEPTH = ONE_GOB
23:44dwlsalmeida[d]: 44 .GOB_HEIGHT = GOB_HEIGHT_FERMI_8
23:44dwlsalmeida[d]: 45 mthd 072c NVC1B5_SET_SRC_WIDTH
23:44dwlsalmeida[d]: 46 .V = (0x20)
23:44dwlsalmeida[d]: 47 mthd 0730 NVC1B5_SET_SRC_HEIGHT
23:44dwlsalmeida[d]: 48 .V = (0x20)
23:44dwlsalmeida[d]: 49 mthd 0734 NVC1B5_SET_SRC_DEPTH
23:44dwlsalmeida[d]: 50 .V = (0x1)
23:44dwlsalmeida[d]: 51 mthd 0738 NVC1B5_SET_SRC_LAYER
23:44dwlsalmeida[d]: 52 .V = (0x0)
23:44dwlsalmeida[d]: vs
23:44dwlsalmeida[d]: 21 [0x00000027] HDR 20088100 subch 4 NINC
23:44dwlsalmeida[d]: 22 mthd 0400 NVC1B5_OFFSET_IN_UPPER
23:44dwlsalmeida[d]: 23 .UPPER = (0x3f)
23:44dwlsalmeida[d]: 24 mthd 0404 NVC1B5_OFFSET_IN_LOWER
23:44dwlsalmeida[d]: 25 .VALUE = (0xfff41000)
23:45dwlsalmeida[d]: 26 mthd 0408 NVC1B5_OFFSET_OUT_UPPER
23:45dwlsalmeida[d]: 27 .UPPER = (0x3f)
23:45dwlsalmeida[d]: 28 mthd 040c NVC1B5_OFFSET_OUT_LOWER
23:45dwlsalmeida[d]: 29 .VALUE = (0xffda0000)
23:45dwlsalmeida[d]: 30 mthd 0410 NVC1B5_PITCH_IN
23:45dwlsalmeida[d]: 31 .VALUE = (0x40)
23:45dwlsalmeida[d]: 32 mthd 0414 NVC1B5_PITCH_OUT
23:45dwlsalmeida[d]: 33 .VALUE = (0x40)
23:45dwlsalmeida[d]: 34 mthd 0418 NVC1B5_LINE_LENGTH_IN
23:45dwlsalmeida[d]: 35 .VALUE = (0x20)
23:45dwlsalmeida[d]: 36 mthd 041c NVC1B5_LINE_COUNT
23:45dwlsalmeida[d]: 37 .VALUE = (0x20)
23:45dwlsalmeida[d]: 38
23:45dwlsalmeida[d]: 39 [0x00000030] HDR 200581ca subch 4 NINC
23:45dwlsalmeida[d]: 40 mthd 0728 NVC1B5_SET_SRC_BLOCK_SIZE
23:45dwlsalmeida[d]: 41 .WIDTH = ONE_GOB
23:45dwlsalmeida[d]: 42 .HEIGHT = FOUR_GOBS
23:45dwlsalmeida[d]: 43 .DEPTH = ONE_GOB
23:45dwlsalmeida[d]: 44 .GOB_HEIGHT = GOB_HEIGHT_FERMI_8
23:45dwlsalmeida[d]: 45 mthd 072c NVC1B5_SET_SRC_WIDTH
23:45dwlsalmeida[d]: 46 .V = (0x20)
23:45dwlsalmeida[d]: 47 mthd 0730 NVC1B5_SET_SRC_HEIGHT
23:45dwlsalmeida[d]: 48 .V = (0x20)
23:45dwlsalmeida[d]: 49 mthd 0734 NVC1B5_SET_SRC_DEPTH
23:45dwlsalmeida[d]: 50 .V = (0x1)
23:45dwlsalmeida[d]: 51 mthd 0738 NVC1B5_SET_SRC_LAYER
23:45dwlsalmeida[d]: 52 .V = (0x0)
23:46dwlsalmeida[d]: vs
23:46dwlsalmeida[d]: 0 [0x00000011] HDR 20088100 subch 4 NINC
23:46dwlsalmeida[d]: 1 mthd 0400 NVC1B5_OFFSET_IN_UPPER
23:46dwlsalmeida[d]: 2 .UPPER = (0x3f)
23:46dwlsalmeida[d]: 3 mthd 0404 NVC1B5_OFFSET_IN_LOWER
23:46dwlsalmeida[d]: 4 .VALUE = (0xfbb70000)
23:46dwlsalmeida[d]: 5 mthd 0408 NVC1B5_OFFSET_OUT_UPPER
23:46dwlsalmeida[d]: 6 .UPPER = (0x3f)
23:46dwlsalmeida[d]: 7 mthd 040c NVC1B5_OFFSET_OUT_LOWER
23:46dwlsalmeida[d]: 8 .VALUE = (0xfffd4000)
23:46dwlsalmeida[d]: 9 mthd 0410 NVC1B5_PITCH_IN
23:46dwlsalmeida[d]: 10 .VALUE = (0x40)
23:46dwlsalmeida[d]: 11 mthd 0414 NVC1B5_PITCH_OUT
23:46dwlsalmeida[d]: 12 .VALUE = (0x40)
23:46dwlsalmeida[d]: 13 mthd 0418 NVC1B5_LINE_LENGTH_IN
23:46dwlsalmeida[d]: 14 .VALUE = (0x40)
23:46dwlsalmeida[d]: 15 mthd 041c NVC1B5_LINE_COUNT
23:46dwlsalmeida[d]: 16 .VALUE = (0x40)
23:46dwlsalmeida[d]: 17
23:46dwlsalmeida[d]: 18 [0x0000001a] HDR 200581ca subch 4 NINC
23:46dwlsalmeida[d]: 19 mthd 0728 NVC1B5_SET_SRC_BLOCK_SIZE
23:46dwlsalmeida[d]: 20 .WIDTH = ONE_GOB
23:46dwlsalmeida[d]: 21 .HEIGHT = SIXTEEN_GOBS
23:46dwlsalmeida[d]: 22 .DEPTH = ONE_GOB
23:46dwlsalmeida[d]: 23 .GOB_HEIGHT = GOB_HEIGHT_FERMI_8
23:46dwlsalmeida[d]: 24 mthd 072c NVC1B5_SET_SRC_WIDTH
23:46dwlsalmeida[d]: 25 .V = (0x40)
23:46dwlsalmeida[d]: 26 mthd 0730 NVC1B5_SET_SRC_HEIGHT
23:46dwlsalmeida[d]: 27 .V = (0x40)
23:46dwlsalmeida[d]: 28 mthd 0734 NVC1B5_SET_SRC_DEPTH
23:46dwlsalmeida[d]: 29 .V = (0x1)
23:46dwlsalmeida[d]: 30 mthd 0738 NVC1B5_SET_SRC_LAYER
23:46dwlsalmeida[d]: 31 .V = (0x0)
23:46dwlsalmeida[d]: 32
23:46dwlsalmeida[d]: the `SIXTEEN_GOBS` one works..