bbswitch prevents suspend and taints kernel

[Guillaume Millet]

When initiating suspend (to RAM), the kernel gets tainted. This is a regression that appeared after upgrading from Kubuntu 13.10 to 14.04. My computer has a discrete nVidia G210M card (hybrid SLI system, on ASUS UL80VT). Here is an extract from dmesg:

PM: Preparing system for mem sleep
BUG: unable to handle kernel NULL pointer dereference at 00000000000000b8
IP: [<ffffffff8139ba5b>] pci_bus_read_config_dword+0x4b/0x90

---
ApportVersion: 2.14.1-0ubuntu3
Architecture: amd64
CurrentDesktop: KDE
DistroRelease: Ubuntu 14.04
InstallationDate: Installed on 2010-01-19 (1571 days ago)
InstallationMedia: Kubuntu 9.10 "Karmic Koala" - Release amd64 (20091027)
Package: bbswitch-dkms 0.7-2ubuntu1
PackageArchitecture: amd64
ProcVersionSignature: Ubuntu 3.13.0-26.48-generic 3.13.11
Tags: trusty
Uname: Linux 3.13.0-26-generic x86_64
UpgradeStatus: Upgraded to trusty on 2014-05-07 (2 days ago)
UserGroups: adm admin audio bumblebee cdrom dialout fuse lpadmin plugdev sambashare src users vboxusers
_MarkForUpload: True

[Guillaume Millet (guimillet)]

apport information

[Guillaume Millet (guimillet)]

apport information

[Guillaume Millet (guimillet)]

dmesg just after trying to suspend

[Guillaume Millet (guimillet)]

Still apears with linux-image-3.15.0-031500rc5-generic.

For now, I stay with 3.11 kernel from kubuntu 13.10.

[Peter Wu (lekensteyn)]

Disabling the graphics card seems to trigger an ACPI hotplug event which seems to remove the device from the system.

Could you prevent bbswitch from loading at boot (nvidia/nouveau should not load either), then report the output of the following steps including a dmesg log:

sudo modprobe bbswitch load_state=-1
lspci -nnvvd10de:
sudo tee /proc/acpi/bbswitch <<<OFF
lspci -nnvvd10de:
sudo tee /proc/acpi/bbswitch <<<ON
lspci -nnvv10de:

[Guillaume Millet (guimillet)]

$ sudo modprobe bbswitch load_state=-1
dmesg log:
[ 2780.584599] bbswitch: module verification failed: signature and/or required key missing - tainting kernel
[ 2780.584809] bbswitch: version 0.7
[ 2780.584818] bbswitch: Found integrated VGA device 0000:00:02.0: \_SB_.PCI0.VGA_
[ 2780.584830] bbswitch: Found discrete VGA device 0000:01:00.0: \_SB_.PCI0.P0P1.VGA_
[ 2780.584843] ACPI Warning: \_SB_.PCI0.P0P1.VGA_._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20131115/nsarguments-95)
[ 2780.584892] ACPI Warning: \_SB_.PCI0.P0P1.VGA_._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20131115/nsarguments-95)
[ 2780.584982] bbswitch: detected a nVidia _DSM function
[ 2780.584999] pci 0000:01:00.0: enabling device (0000 -> 0003)
[ 2780.585046] bbswitch: Succesfully loaded. Discrete card 0000:01:00.0 is on
[ 2790.596471] init: Failed to spawn nvidia-persistenced main process: unable to execute: No such file or directory

$ lspci -nnvvd10de:
01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GT218M [GeForce G210M] [10de:0a74] (rev a2) (prog-if 00 [VGA controller])
        Subsystem: ASUSTeK Computer Inc. Device [1043:1bc2]
        Control: I/O+ Mem+ BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
        Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
        Interrupt: pin A routed to IRQ 16
        Region 0: Memory at fd000000 (32-bit, non-prefetchable) [size=16M]
        Region 1: Memory at e0000000 (64-bit, prefetchable) size=256M]
        Region 3: Memory at fa000000 (64-bit, prefetchable) [size=32M]
        Region 5: I/O ports at dc00 [size=128]
        Expansion ROM at fe980000 [disabled] [size=512K]
        Capabilities: <access denied>

01:00.1 Audio device [0403]: NVIDIA Corporation High Definition Audio Controller [10de:0be3] (rev a1)
        Subsystem: ASUSTeK Computer Inc. Device [1043:1bc2]
        Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
        Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
        Latency: 0, Cache Line Size: 32 bytes
        Interrupt: pin A routed to IRQ 16
        Region 0: Memory at fe97c000 (32-bit, non-prefetchable) [size=16K]
        Capabilities: <access denied>
        Kernel driver in use: snd_hda_intel

$ sudo tee /proc/acpi/bbswitch <<<OFF
OFF

dmesg log:
[ 3058.083676] bbswitch: disabling discrete graphics
[ 3058.096117] ACPI Warning: \_SB_.PCI0.P0P1.VGA_._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20131115/nsarguments-95)
[ 3058.200492] vgaarb: device changed decodes: PCI:0000:00:02.0,olddecodes=none,decodes=io+mem:owns=none
[ 3058.312204] ------------[ cut here ]------------
[ 3058.312216] WARNING: CPU: 0 PID: 4 at /build/buildd/linux-3.13.0/fs/sysfs/group.c:214 sysfs_remove_group+0xc6/0xd0()
[ 3058.312219] sysfs group ffffffff81cabae0 not found for kobject 'acpi_video1'
[ 3058.312221] Modules linked in: bbswitch(OF) hid_generic cdc_acm usbhid hid snd_hrtimer ip6table_filter ip6_tables iptable_filter ip_tables...

[Guillaume Millet (guimillet)]

What else information is needed?

I have tried with the daily kernel 3.17.0-999.201409272205 and the tainting still occurs. The syslog trace related to suspending event is slightly different (dis_dev_get+0x15/0x40 [bbswitch] instead of pci_bus_read_config_dword+0x64/0x90 in the initial report with kernel 3.13):
[ 3342.022581] init: anacron main process (13582) killed by TERM signal
[ 3342.297888] PM: Syncing filesystems ... done.
[ 3342.489577] PM: Preparing system for mem sleep
[ 3342.489787] general protection fault: 0000 [#1] SMP
[ 3342.489988] Modules linked in: bbswitch(OE) ip6table_filter ip6_tables iptable_filter ip_tables x_tables snd_hrtimer cuse rfcomm bnep bluetooth binfmt_misc dm_crypt snd_hda_codec_hdmi arc4 ath9k ath9k_common ath9k_hw snd_hda_codec_realtek snd_hda_codec_generic ath snd_hda_intel snd_hda_controller snd_hda_codec mac80211 snd_hwdep uvcvideo snd_pcm videobuf2_vmalloc videobuf2_memops videobuf2_core v4l2_common snd_seq_midi snd_seq_midi_event kvm_intel videodev media snd_rawmidi snd_seq cfg80211 snd_seq_device kvm snd_timer snd parport_pc mxm_wmi ppdev soundcore coretemp joydev lp lpc_ich serio_raw parport mac_hid asus_laptop wmi sparse_keymap input_polldev uas usb_storage psmouse ahci libahci i915 atl1c i2c_algo_bit drm_kms_helper drm video
[ 3342.491087] CPU: 1 PID: 13498 Comm: pm-suspend Tainted: G W OE 3.17.0-999-generic #201409272205
[ 3342.491296] Hardware name: ASUSTeK Computer Inc. UL80VT /UL80VT , BIOS 214 01/17/2011
[ 3342.491505] task: ffff8800b8d1a800 ti: ffff880035860000 task.ti: ffff880035860000
[ 3342.491713] RIP: 0010:[<ffffffffc08ae035>] [<ffffffffc08ae035>] dis_dev_get+0x15/0x40 [bbswitch]
[ 3342.491928] RSP: 0018:ffff880035863d10 EFLAGS: 00010202
[ 3342.492133] RAX: 6e6f697461727564 RBX: 0000000000000003 RCX: 0000000000000002
[ 3342.492335] RDX: 0000000000000000 RSI: 0000000000000003 RDI: ffffffffc08b0000
[ 3342.492535] RBP: ffff880035863d18 R08: ffffffff814e00e0 R09: 0000000000000000
[ 3342.492737] R10: 0000000000000000 R11: 0000000fffffffe0 R12: 0000000000000000
[ 3342.492939] R13: 00000000fffffffa R14: ffffffff81cd3f90 R15: 0000000000000000
[ 3342.493142] FS: 00007f8b8e80d740(0000) GS:ffff88013fd00000(0000) knlGS:0000000000000000
[ 3342.493345] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[ 3342.493529] CR2: 00007f8b8f73cf10 CR3: 000000009c68b000 CR4: 00000000000407e0
[ 3342.493529] Stack:
[ 3342.493529] ffffffffc08ae545 ffff880035863d58 ffffffff81093f2d ffffffff81ec685c
[ 3342.493529] ffffffff81c521c0 0000000000000003 0000000000000000 00000000ffffffff
[ 3342.493529] 0000000000000000 ffff880035863da8 ffffffff81094188 ffff880035863d48
[ 3342.493529] Call Trace:
[ 3342.493529] [<ffffffffc08ae545>] ? bbswitch_pm_handler+0x55/0x70 [bbswitch]
[ 3342.493529] [<ffffffff81093f2d>] notifier_call_chain+0x4d/0x70
[ 3342.493529] [<ffffffff81094188>] __blocking_notifier_call_chain+0x58/0x80
[ 3342.493529] [<ffffffff810941c6>] blocking_notifier_call_chain+0x16/0x20
[ 3342.493529] [<ffffffff810bd5fa>] pm_notifier_call_chain+0x1a/0x30
[ 3342.493529] [<ffffffff8178aafa>] suspend_prepare+0x46/0xcd
[ 3342.493529] [<ffffffff810be83f>] enter_state+0x16f/0x250
...

[Peter Wu (lekensteyn)]

This problem may occur when your compiler (gcc) is different than the one used to compile the kernel. Have you installed an older gcc version (maybe via CUDA?). What does `gcc -v` show?

[mcsan (octagonhead)]

For me on my UL30VT with ubuntu 15.04 it was:
 gcc version 4.9.2 (Ubuntu 4.9.2-10ubuntu13)

And cat /proc/version
Linux version 3.19.0-15-generic (buildd@komainu) (gcc version 4.9.2 (Ubuntu 4.9.2-10ubuntu13) ) #15-Ubuntu SMP Thu Apr 16 23:32:37 UTC 2015

Nevertheless, I discovered and remove some old gcc versions and re-installed all dkms packages, but still have the same behaviour as detailed above. I wonder if there are any ul*0vt laptops that are still working out there?
I will try installing bbswitch on a clean install of ubuntu when I have time

[Peter Wu (lekensteyn)]

Do you have a full dmesg since the first boot? I have seen machines having issues due to hotplugging where the device is gone during suspend.

[Guillaume Millet (guimillet)]

Still facing the issue with kernel 4.2.0-23-generic, gcc version 5.2.1 20151010 (Ubuntu 5.2.1-22ubuntu2), bbswitch-dkms 0.7-2ubuntu1, no bumblebee package installed. I attach a full dmesg (from kern.log) from the boot to the freeze. After booting, I logged into tty1 and entered the following commands:
sudo service sdsm stop
sudo rmmod nouveau
sudo modprobe bbswitch load_state=0
sudo service sdsm start
Ctrl-Alt-F7 and closed the lid