Bug 81485

Summary: [NVE4] dmesg spam on startup
Product: xorg Reporter: Konstantin Svist <fry.kun>
Component: Driver/nouveauAssignee: Nouveau Project <nouveau>
Status: RESOLVED MOVED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium    
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
/var/log/messages from 3.15.5-200.fc20.x86_64
none
extracted messages
none
extracted dmesg none

Description Konstantin Svist 2014-07-18 08:11:32 UTC
Sager NP8268-S (Clevo P150SM-A) with dual graphics
lspci -nn
00:02.0 VGA compatible controller [0300]: Intel Corporation 4th Gen Core Processor Integrated Graphics Controller [8086:0416] (rev 06)
01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GK104M [GeForce GTX 870M] [10de:1199] (rev a1)

Had to blacklist nouveau module since it was spamming dmesg from startup:


Jul 17 05:52:48 mireille kernel: [   18.417467] nouveau E[   PFIFO][0000:01:00.0] PBDMA1: ch 0 [DRM] subc 0 mthd 0x0000 data 0x00000000
Jul 17 05:52:48 mireille kernel: [   18.417477] nouveau E[   PFIFO][0000:01:00.0] PBDMA1: GPPTR
Jul 17 05:52:48 mireille kernel: [   18.417479] nouveau E[   PFIFO][0000:01:00.0] PBDMA1: ch 0 [DRM] subc 0 mthd 0x0000 data 0x00000000
Jul 17 05:52:48 mireille kernel: [   18.417489] nouveau E[   PFIFO][0000:01:00.0] PBDMA1: GPPTR


Using Fedora 20 with default Xorg and Intel driver 2.99.911 (from git)
Comment 1 Ilia Mirkin 2014-07-18 16:03:54 UTC
Is this something new, or has it always been the case (as far as you know)? Can you try 3.16-rc5 and see if this still happens there? Assuming it does, can you provide a full dmesg log from 3.16? [Sounds like the dmesg spam is infinite, so just grab a little of it...]
Comment 2 Konstantin Svist 2014-07-18 17:08:57 UTC
I just bought this laptop; started using it yesterday, so no idea about older builds.
I got /var/log/messages from 3.15.5-200.fc20.x86_64 -- would they be useful?
Comment 3 Ilia Mirkin 2014-07-18 17:13:32 UTC
(In reply to comment #2)
> I just bought this laptop; started using it yesterday, so no idea about
> older builds.
> I got /var/log/messages from 3.15.5-200.fc20.x86_64 -- would they be useful?

Does that contain the kernel logs? If so, sure -- try to extract the kernel logs from all the other junk in there though, and limit it to just the boot in question. But really just grabbing the output of dmesg seems simpler... You can modprobe nouveau after the system has started too, btw.
Comment 4 Konstantin Svist 2014-07-18 17:29:12 UTC
Created attachment 103053 [details]
/var/log/messages from 3.15.5-200.fc20.x86_64

attaching a trimmed version from 3.15.5-200.fc20.x86_64
I've grepped out the majority of the same message, though the file is still 2.7M uncompressed

Just running "dmesg" gives me a few seconds of the spam; everything else is gone already.

Will try a build from http://koji.fedoraproject.org/koji/buildinfo?buildID=545032 later today.
Comment 5 Ilia Mirkin 2014-07-18 17:36:11 UTC
Created attachment 103055 [details]
extracted messages

OK, I've gone ahead and extracted some of the interesting stuff into the above.
Comment 6 Ilia Mirkin 2014-07-18 18:04:04 UTC
Does this card seriously have 6GB of dedicated GDDR5 VRAM? Or is it misdetected? (Hm, looks like indeed one of the options on the marketing page...)

I would not be surprised if we had some bugs for cards with > 4GB. Unfortunately we don't have a vram_limit parameter.

Could you edit drivers/gpu/drm/nouveau/core/subdev/fb/ramnvc0.c -- take a look at nvc0_ram_create_, and after the for loop that does ram->size += xyz, just do

ram->size /= 2;

which will bring it down to 3GB... No clue if it'll help, but worth a shot.
Comment 7 Konstantin Svist 2014-07-18 18:52:14 UTC
Is this just libdrm compilation section from http://nouveau.freedesktop.org/wiki/InstallNouveau/ ? Or something else?
Comment 8 Ilia Mirkin 2014-07-18 18:52:58 UTC
(In reply to comment #7)
> Is this just libdrm compilation section from
> http://nouveau.freedesktop.org/wiki/InstallNouveau/ ? Or something else?

No, in your kernel.
Comment 9 Konstantin Svist 2014-07-21 21:02:44 UTC
Created attachment 103230 [details]
extracted dmesg

Ok, as you said I added
        ram->size /=2; // KOSTYA
right after the loop
        for (part = 0; part < parts; part++) {
in function nvc0_ram_create_

Here's the dmesg from when I ran modprobe and later rmmod with this module
Comment 10 Konstantin Svist 2014-07-22 03:10:35 UTC
please let me know what other info you need or what else I should try
Comment 11 Ilia Mirkin 2014-08-21 17:55:59 UTC
Can you give this repo a shot?

http://cgit.freedesktop.org/~darktama/nouveau

It contains a few fixes for cards with "odd" memory configurations which might help your case out. You'll have to build it against 3.17-rc1 in all likelihood.

cd drm && make

should build it against the currently-running kernel and produce a new nouveau.ko.
Comment 12 Konstantin Svist 2014-08-22 01:32:26 UTC
got compilation error...

/home/kostya/projects/nouveau/darktama/drm/nouveau_drm.c: In function ‘nouveau_platform_device_create_’:
/home/kostya/projects/nouveau/darktama/drm/nouveau_drm.c:1087:2: error: implicit declaration of function ‘drm_dev_set_unique’ [-Werror=implicit-function-declaration]
  err = drm_dev_set_unique(drm, "%s", dev_name(&pdev->dev));
  ^
cc1: some warnings being treated as errors
make[2]: *** [/home/kostya/projects/nouveau/darktama/drm/nouveau_drm.o] Error 1
Comment 13 Ilia Mirkin 2014-08-22 01:41:12 UTC
It's pretty finicky about the version you're building against... are you sure you're building it against 3.17-rc1?

drm_dev_set_unique came in via ca8e2ad71013049bc88a10b11d83712bfe56cdd4 and is there starting 3.16-rc1.
Comment 14 Konstantin Svist 2014-08-22 01:53:38 UTC
Ah, i see.. I was building against my currently-running, which is 3.15.x
Comment 15 Konstantin Svist 2014-08-22 02:25:40 UTC
Ok, built against 3.17.0-0.rc1.git1.1.fc22.x86_64

now I have a problem insmod'ing it:

insmod: ERROR: could not insert module /home/kostya/projects/nouveau/darktama/drm/nouveau.ko: Unknown symbol in module
Comment 16 Ilia Mirkin 2014-08-22 02:28:06 UTC
First do

modprobe drm

Then you can do

insmod .../nouveau.ko

If it still fails, look in dmesg for what symbol is missing.
Comment 17 Konstantin Svist 2014-08-22 05:17:57 UTC
Aug 21 22:16:16 mireille kernel: nouveau: module verification failed: signature and/or  required key missing - tainting kernel
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_mmap (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_global_release (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_clean_mm (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_pool_unpopulate (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_agp_tt_create (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_manager_func (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_global_init (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_move_ttm (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_dma_tt_fini (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_device_release (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_agp_tt_populate (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_kunmap (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_del_sub_from_lru (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_agp_tt_unpopulate (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_mem_put (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_mem_space (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_device_init (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_wait (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_init_mm (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_dma_acc_size (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_tt_bind (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_evict_mm (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_kmap (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_add_to_lru (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_unref (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_mem_global_release (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_mem_global_init (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_dma_populate (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_dma_unpopulate (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_init (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_validate (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_move_accel_cleanup (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_bo_move_memcpy (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_dma_tt_init (err 0)
Aug 21 22:16:16 mireille kernel: nouveau: Unknown symbol ttm_pool_populate (err 0)
Comment 18 Ilia Mirkin 2014-08-22 05:28:28 UTC
modprobe ttm
modprobe drm_kms_helper
Comment 19 Konstantin Svist 2014-08-22 08:39:26 UTC
finally insmod'ed, thanks for your help!
unfortunately the end result is same message spam :(

I'm just glad this laptop has dual graphics and I can keep working with intel graphics for the moment
Comment 20 Konstantin Svist 2014-11-07 01:55:35 UTC
any update for this?
driver still unusable for me!
Comment 21 Ilia Mirkin 2015-10-22 07:13:13 UTC
Does this still happen with kernel 4.1 or later?
Comment 22 Martin Peres 2019-12-04 08:47:38 UTC
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/xorg/driver/xf86-video-nouveau/issues/122.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.