Nouveau - Dec 2023 - Thinkpad P17 gen 2 kernel 6.4 and 6.6 lack of support for nvidia GA104GLM [RTX A5000 Mobile] and missing module firmware

Howdy,

I'm trying a Thnkpad P17 gen2, the last thinkpad that still comes in
17"
4K (newer ones are 16" only, so I'm looking for other worthwhile linux
laptops with 17" or bigger LCD that also does 4K, the alienware I saw
was 18" but not 4K)

Unfortunately I seem to need the nouveau driver to turn off the nvidia
chip I don't plan on using (intel graphics is fine for me), and bios
only allows 'bybrid' or nvidia only)
On my P73, nouveau never really worked in the 3 years I've had it, but
it could at least turn off the nvidia chip. On P17gen2 it does not seem
to be able to do so.

Firmware is missing even from the latest firmware-linux-nonfree or from upstream
git https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git
sauron:~# update-initramfs -v -c -k `uname -r` 2>&1 |grep W:
W: Possible missing firmware /lib/firmware/nvidia/ga107/acr/ucode_ahesasc.bin
for module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga106/acr/ucode_ahesasc.bin
for module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga104/acr/ucode_ahesasc.bin
for module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga103/acr/ucode_ahesasc.bin
for module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga107/acr/ucode_asb.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga106/acr/ucode_asb.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga104/acr/ucode_asb.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga103/acr/ucode_asb.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga107/acr/ucode_unload.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga106/acr/ucode_unload.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga104/acr/ucode_unload.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga103/acr/ucode_unload.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga107/nvdec/scrubber.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga106/nvdec/scrubber.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga104/nvdec/scrubber.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga103/nvdec/scrubber.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga107/sec2/hs_bl_sig.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga107/sec2/sig.bin for module
nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga107/sec2/image.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga107/sec2/desc.bin for module
nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga106/sec2/hs_bl_sig.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga106/sec2/sig.bin for module
nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga106/sec2/image.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga106/sec2/desc.bin for module
nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga104/sec2/hs_bl_sig.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga104/sec2/sig.bin for module
nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga104/sec2/image.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga104/sec2/desc.bin for module
nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga103/sec2/hs_bl_sig.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga103/sec2/sig.bin for module
nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga103/sec2/image.bin for
module nouveau
W: Possible missing firmware /lib/firmware/nvidia/ga103/sec2/desc.bin for module
nouveau

During boot, the nvidia module hangs for 2mn and fails to do any work,
including being able to turn off the nvidia chip (which it could do
on P73 without otherwise ever being able to use that chip for proper
display). I want to turn off the nvidia chip so that I can get multi
hour runtime on batteries without some useless chip that is using power
for no reason.

sauron:~# lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corporation Tiger Lake-H GT1 [UHD
Graphics] (rev 01)
01:00.0 VGA compatible controller: NVIDIA Corporation GA104GLM [RTX A5000
Mobile] (rev a1)


Boot looks like this:
[    0.210932] Kernel command line:
BOOT_IMAGE=/vmlinuz-6.6.3-amd64-preempt-sysrq-20220227
root=/dev/mapper/cryptroot ro rootflags=subvol=root
cryptopts=source=/dev/nvme0n1p7,keyscript=/sbin/cryptgetpw usbcore.autosuspend=1
pcie_aspm=force resume=/dev/dm-1 thinkpad-acpi.brightness_enable=1
acpi_backlight=native nouveau.modset=0 systemd.unified_cgroup_hierarchy=0
[    3.184525] nouveau: unknown parameter 'modset' ignored
[    3.184800] nouveau: detected PR support, will not use DSM
[    3.184813] nouveau 0000:01:00.0: vgaarb: pci_notify
[    3.184816] nouveau 0000:01:00.0: runtime IRQ mapping not provided by arch
[    3.184821] nouveau 0000:01:00.0: enabling device (0000 -> 0003)
[    3.184959] nouveau 0000:01:00.0: NVIDIA GA104 (b74000a1)
[    3.295682] nouveau 0000:01:00.0: bios: version 94.04.51.00.34
[   64.941381] nouveau 0000:01:00.0: acr: firmware unavailable
[  126.381361] nouveau 0000:01:00.0: sec2: firmware unavailable
[  126.381484] nouveau 0000:01:00.0: enabling bus mastering
[  126.382073] nouveau 0000:01:00.0: fb: 16384 MiB GDDR6
[  183.695282] nouveau 0000:01:00.0: fb: VPR locked, but no scrubber binary!
[  183.702321] nouveau 0000:01:00.0: DRM: error initialising bo driver, -12
[  183.709419] nouveau: probe of 0000:01:00.0 failed with error -12
[  183.711498] nouveau 0000:01:00.0: vgaarb: pci_notify

What is the next recommended step?

Thanks,
Marc
-- 
"A mouse is a device used to point at the xterm you want to type in" -
A.S.R.
 
Home page: http://marc.merlins.org/

On Sat, 2023-12-02 at 09:13 -0800, Marc MERLIN wrote:
> [??? 3.184525] nouveau: unknown parameter 'modset' ignored
For starters, you misspelled "modeset"

On Sat, Dec 02, 2023 at 06:08:01PM +0000, Timur Tabi
wrote:
> On Sat, 2023-12-02 at 09:13 -0800, Marc MERLIN wrote:
> > [??? 3.184525] nouveau: unknown parameter 'modset' ignored
> 
> For starters, you misspelled "modeset"
That was a previous boot in dmesg where I failed to turn off the module,
but I was mostly interested in showing the errors of all the firmware
missing and nouveau failing to start, which those logs do show.

Separely, both 6.4 and 6.6 are hanging after a few hours of runtime with
networking dying or other issues that require reboot

See below

6.4:
> [55647.774842] vgaarb: client 0x00000000c24cb19e called 'target'
> [55647.774852] vgaarb: PCI:0000:00:02.0 ==> 0000:00:02.0 pdev
00000000bfa35d85
> [55647.774854] vgaarb: vgadev 000000008ea0fc7d
> [55825.318992] INFO: task NetworkManager:3372 blocked for more than 120
seconds.
> [55825.318999]       Tainted: G     U     OE     
6.4.9-amd64-preempt-sysrq-20220227 #2
> [55825.319000] "echo 0 >
/proc/sys/kernel/hung_task_timeout_secs" disables this message.
> [55825.319002] task:NetworkManager  state:D stack:0     pid:3372  ppid:1   
flags:0x00000002
> [55825.319005] Call Trace:
> [55825.319006]  <TASK>
> [55825.319009]  __schedule+0xba5/0xc17
> [55825.319015]  schedule+0x95/0xce
> [55825.319017]  schedule_preempt_disabled+0x15/0x22
> [55825.319020]  __mutex_lock.constprop.0+0x18b/0x291
> [55825.319025]  nl80211_prepare_wdev_dump+0x8b/0x19f [cfg80211
d0c23c84d531afea8d4a2711c5c3e691cbb9587f]
> [55825.319065]  nl80211_dump_station+0x49/0x1d0 [cfg80211
d0c23c84d531afea8d4a2711c5c3e691cbb9587f]
> [55825.319091]  ? __mod_lruvec_page_state+0x4c/0x86
> [55825.319093]  ? mod_lruvec_page_state.constprop.0+0x1c/0x2e
> [55825.319096]  ? __kmalloc_large_node+0xd5/0xfb
> [55825.319099]  ? __kmalloc_node_track_caller+0x5a/0xad
> [55825.319101]  ? kmalloc_reserve+0xa7/0xe2
> [55825.319104]  ? __alloc_skb+0xe9/0x148
> [55825.319106]  netlink_dump+0x143/0x2b2
> [55825.319109]  __netlink_dump_start+0x125/0x177
> [55825.319111]  genl_family_rcv_msg_dumpit+0xf1/0x110
> [55825.319114]  ? poll_freewait+0x72/0x91
> [55825.319117]  ? __pfx_genl_start+0x40/0x40
> [55825.319119]  ? __pfx_nl80211_dump_station+0x40/0x40 [cfg80211
d0c23c84d531afea8d4a2711c5c3e691cbb9587f]
> [55825.319143]  ? __pfx_genl_parallel_done+0x40/0x40
> [55825.319146]  genl_rcv_msg+0x189/0x1e2
> [55825.319148]  ? __pfx_nl80211_dump_station+0x40/0x40 [cfg80211
d0c23c84d531afea8d4a2711c5c3e691cbb9587f]
> [55825.319172]  ? __pfx_genl_rcv_msg+0x40/0x40
> [55825.319173]  netlink_rcv_skb+0x89/0xe3
> [55825.319176]  genl_rcv+0x24/0x31
> [55825.319178]  netlink_unicast+0x10e/0x1ae
> [55825.319180]  netlink_sendmsg+0x321/0x361
> [55825.319182]  sock_sendmsg_nosec+0x35/0x64
> [55825.319186]  ____sys_sendmsg+0x13e/0x1ef
> [55825.319188]  ___sys_sendmsg+0x76/0xb3
> [55825.319190]  ? __fget_light+0x41/0x50
> [55825.319193]  ? do_epoll_wait+0x49b/0x4d4
> [55825.319196]  ? __pfx_pollwake+0x40/0x40
> [55825.319198]  ? __rseq_handle_notify_resume+0x2a0/0x4bd
> [55825.319200]  ? __fget+0x38/0x47
> [55825.319202]  __sys_sendmsg+0x60/0x97
> [55825.319204]  do_syscall_64+0x7e/0xa7
> [55825.319208]  ? syscall_exit_to_user_mode+0x18/0x27
> [55825.319210]  ? __task_pid_nr_ns+0x5f/0x6d
> [55825.319213]  ? syscall_exit_to_user_mode+0x18/0x27
> [55825.319214]  ? do_syscall_64+0x9d/0xa7
> [55825.319216]  ? do_syscall_64+0x9d/0xa7
> [55825.319218]  ? do_syscall_64+0x9d/0xa7
> [55825.319220]  ? do_syscall_64+0x9d/0xa7
> [55825.319222]  entry_SYSCALL_64_after_hwframe+0x77/0xe1
> [55825.319224] RIP: 0033:0x7f1fdc79e9bd
> [55825.319226] RSP: 002b:00007ffeb6460900 EFLAGS: 00000293 ORIG_RAX:
000000000000002e
> [55825.319228] RAX: ffffffffffffffda RBX: 000055e0a9ce1d90 RCX:
00007f1fdc79e9bd
> [55825.319229] RDX: 0000000000000000 RSI: 00007ffeb6460950 RDI:
000000000000000b
> [55825.319230] RBP: 00007ffeb6460950 R08: 0000000000000000 R09:
0000000000000300
> [55825.319231] R10: 0000000000000000 R11: 0000000000000293 R12:
00007ffeb6460a30
> [55825.319232] R13: 00007f1fd0038690 R14: 00007ffeb6460c60 R15:
000055e0aa210400
> [55825.319234]  </TASK>
6.6.3:
[  443.613095] BTRFS info (device dm-2): scrub: started on devid 1
[  484.778344] INFO: task kworker/2:1:106 blocked for more than 120 seconds.
[  484.778352]       Tainted: G     U            
6.6.3-amd64-preempt-sysrq-20220227 #4
[  484.778353] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  484.778354] task:kworker/2:1     state:D stack:0     pid:106   ppid:2     
flags:0x00004000
[  484.778358] Workqueue: ipv6_addrconf addrconf_verify_work
[  484.778365] Call Trace:
[  484.778367]  <TASK>
[  484.778369]  __schedule+0xba0/0xc05
[  484.778373]  schedule+0x95/0xce
[  484.778375]  schedule_preempt_disabled+0x15/0x22
[  484.778376]  __mutex_lock.constprop.0+0x18b/0x291
[  484.778379]  addrconf_verify_work+0xe/0x20
[  484.778381]  process_scheduled_works+0x1da/0x2e0
[  484.778385]  worker_thread+0x1ca/0x224
[  484.778388]  ? __pfx_worker_thread+0x40/0x40
[  484.778390]  kthread+0xe9/0xf4
[  484.778393]  ? __pfx_kthread+0x40/0x40
[  484.778395]  ret_from_fork+0x21/0x38
[  484.778397]  ? __pfx_kthread+0x40/0x40
[  484.778399]  ret_from_fork_asm+0x1b/0x80
[  484.778403]  </TASK>
[  484.778409] INFO: task kworker/4:2:388 blocked for more than 120 seconds.
[  484.778410]       Tainted: G     U            
6.6.3-amd64-preempt-sysrq-20220227 #4
[  484.778411] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  484.778412] task:kworker/4:2     state:D stack:0     pid:388   ppid:2     
flags:0x00004000
[  484.778415] Workqueue: ipv6_addrconf addrconf_verify_work
[  484.778417] Call Trace:
[  484.778418]  <TASK>
[  484.778420]  __schedule+0xba0/0xc05
[  484.778422]  schedule+0x95/0xce
[  484.778423]  schedule_preempt_disabled+0x15/0x22
[  484.778425]  __mutex_lock.constprop.0+0x18b/0x291
[  484.778427]  addrconf_verify_work+0xe/0x20
[  484.778429]  process_scheduled_works+0x1da/0x2e0
[  484.778431]  worker_thread+0x1ca/0x224
[  484.778433]  ? __pfx_worker_thread+0x40/0x40
[  484.778436]  kthread+0xe9/0xf4
[  484.778437]  ? __pfx_kthread+0x40/0x40
[  484.778439]  ret_from_fork+0x21/0x38
[  484.778440]  ? __pfx_kthread+0x40/0x40
[  484.778442]  ret_from_fork_asm+0x1b/0x80
[  484.778444]  </TASK>
[  484.778468] INFO: task NetworkManager:3372 blocked for more than 120 seconds.
[  484.778469]       Tainted: G     U            
6.6.3-amd64-preempt-sysrq-20220227 #4
[  484.778470] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  484.778471] task:NetworkManager  state:D stack:0     pid:3372  ppid:1     
flags:0x00000002
[  484.778473] Call Trace:
[  484.778474]  <TASK>
[  484.778475]  __schedule+0xba0/0xc05
[  484.778477]  schedule+0x95/0xce
[  484.778478]  schedule_preempt_disabled+0x15/0x22
[  484.778480]  __mutex_lock.constprop.0+0x18b/0x291
[  484.778483]  nl80211_prepare_wdev_dump+0x8b/0x19f [cfg80211
443448ea372df5c5c09782a3fb412c115f1aa45a]
[  484.778526]  nl80211_dump_station+0x49/0x1d0 [cfg80211
443448ea372df5c5c09782a3fb412c115f1aa45a]
[  484.778555]  ? __alloc_pages+0x131/0x1e8
[  484.778558]  ? __mod_lruvec_page_state+0x4c/0x86
[  484.778561]  ? mod_lruvec_page_state.constprop.0+0x1c/0x2e
[  484.778564]  ? __kmalloc_large_node+0xd5/0xfb
[  484.778566]  ? __alloc_skb+0xad/0x148
[  484.778569]  ? __kmalloc_node_track_caller+0x5b/0xb2
[  484.778571]  ? kmalloc_reserve+0xab/0xe6
[  484.778573]  genl_dumpit+0x32/0x4d
[  484.778576]  netlink_dump+0x143/0x2b2
[  484.778579]  __netlink_dump_start+0x145/0x197
[  484.778583]  genl_family_rcv_msg_dumpit+0xa3/0xd1
[  484.778585]  ? __pfx_genl_start+0x40/0x40
[  484.778586]  ? __pfx_genl_dumpit+0x40/0x40
[  484.778588]  ? __pfx_genl_done+0x40/0x40
[  484.778589]  genl_rcv_msg+0x1a0/0x1f2
[  484.778591]  ? __pfx_nl80211_dump_station+0x40/0x40 [cfg80211
443448ea372df5c5c09782a3fb412c115f1aa45a]
[  484.778619]  ? __pfx_genl_rcv_msg+0x40/0x40
[  484.778620]  netlink_rcv_skb+0x89/0xe3
[  484.778623]  genl_rcv+0x24/0x31
[  484.778625]  netlink_unicast+0x114/0x1b4
[  484.778627]  netlink_sendmsg+0x321/0x361
[  484.778630]  sock_sendmsg_nosec+0x46/0x70
[  484.778633]  ____sys_sendmsg+0x144/0x1f5
[  484.778635]  ___sys_sendmsg+0x76/0xb3
[  484.778637]  ? __fget+0x38/0x47
[  484.778640]  ? __fget_light+0x41/0x50
[  484.778642]  ? do_epoll_wait+0x49e/0x4d7
[  484.778645]  ? __pfx_pollwake+0x40/0x40
[  484.778647]  ? __fget+0x38/0x47
[  484.778649]  __sys_sendmsg+0x60/0x97
[  484.778652]  do_syscall_64+0x7e/0xa7
[  484.778655]  ? do_syscall_64+0x9d/0xa7
[  484.778656]  ? do_syscall_64+0x9d/0xa7
[  484.778657]  entry_SYSCALL_64_after_hwframe+0x6e/0xd8
[  484.778661] RIP: 0033:0x7fad667fe9bd
[  484.778663] RSP: 002b:00007ffd9d5a4f00 EFLAGS: 00000293 ORIG_RAX:
000000000000002e
[  484.778665] RAX: ffffffffffffffda RBX: 000055d7ea4dad90 RCX: 00007fad667fe9bd
[  484.778666] RDX: 0000000000000000 RSI: 00007ffd9d5a4f50 RDI: 000000000000000b
[  484.778667] RBP: 00007ffd9d5a4f50 R08: 0000000000000000 R09: 0000000000000300
[  484.778669] R10: 0000000000000000 R11: 0000000000000293 R12: 00007ffd9d5a5030
[  484.778670] R13: 000055d7ea9f9050 R14: 00007ffd9d5a5260 R15: 000055d7eaa03050
[  484.778671]  </TASK>


-- 
"A mouse is a device used to point at the xterm you want to type in" -
A.S.R.
 
Home page: http://marc.merlins.org/                       | PGP 7F55D5F27AAF9D08

On 12/02/23 Marc MERLIN wrote:
> Howdy,
Howdy!
> I'm trying a Thnkpad P17 gen2, the last thinkpad that still comes in
17"
> 4K (newer ones are 16" only, so I'm looking for other worthwhile
linux
> laptops with 17" or bigger LCD that also does 4K, the alienware I saw
> was 18" but not 4K)
>
> Unfortunately I seem to need the nouveau driver to turn off the nvidia
> chip I don't plan on using (intel graphics is fine for me), and bios
> only allows 'bybrid' or nvidia only)
> On my P73, nouveau never really worked in the 3 years I've had it, but
> it could at least turn off the nvidia chip. On P17gen2 it does not seem
> to be able to do so.
At the moment you'd have to use the proprietary Nvidia driver for
graphics support. But there are, and have been for a long time, ways to
disable the additional dedicated graphics device completely and save
power, which is nice thing on a laptop...
> sauron:~# lspci | grep VGA
> 00:02.0 VGA compatible controller: Intel Corporation Tiger Lake-H GT1 [UHD
Graphics] (rev 01)
> 01:00.0 VGA compatible controller: NVIDIA Corporation GA104GLM [RTX A5000
Mobile] (rev a1)
Note the Nvidia card's PCI address is 01:00.0...
> What is the next recommended step?
STEP #1: disable nouveau by blacklisting the module.

There's more than one way to do this:

* Add it to /etc/modprobe.d/<someconfigfilename>.conf
E.g. /etc/modprobe.d/blacklist-nouveau.conf, run in a root shell (if the
file doesn't already exist!):
echo "blacklist nouveau" > /etc/modprobe.d/blacklist-nouveau.conf

* Add a kernel command line parameter: modprobe.blacklist=nouveau
How you do this depends on which Linux (distribution) you're running.
E.g. GRUB's command line may be used, if GRUB /is/ used, or dracut and
so on...

STEP #2: you could power down the PCI device (only after you've disabled
the driver in step #1).

Try it out first by disabling the PCI device you noted above on a
running system (as root!), e.g. like this:

echo 1 > /sys/bus/pci/devices/0000\:01\:00.0/remove

If that works, you'd do something like adding a new udev rule in e.g.
/etc/udev/rules.d/00-remove-nvidia.rules with contents of the sort:
ACTION=="add", SUBSYSTEM=="pci",
ATTR{vendor}=="0x10de",
ATTR{class}=="0x03[0-9]*", ATTR{power/control}="auto",
ATTR{remove}="1"

Take a full example from the Arch Wiki:
https://wiki.archlinux.org/title/Hybrid_graphics

Other resouces:
https://github.com/bayasdev/nvidia-gpu-off
https://unix.stackexchange.com/questions/702774/how-to-disable-pcie-device-at-boot
> Thanks,
> Marc
Welcome! Hope this helps, and I also hope it's not too late. I just saw
your posting and thought, better late than never...
Linux User #330250