[SOLVED] Unable to use nvidia modeset for fast VT switching

kokoko3k · 2018-05-12 14:20:20

I'd like to use nvidia modesettings to speed-up switching to text console and xorg.
I did:

koko@slimer# cat /etc/modprobe.d/nvidia.conf 
options nvidia-drm modeset=1

koko@slimer# grep nvidia /etc/mkinitcpio.conf 
MODULES="ahci lzo nvidia nvidia_modeset nvidia_uvm nvidia_drm"
FILES="/etc/modprobe.d/modprobe.conf /etc/modprobe.d/nvidia.conf"

koko@slimer# grep nvidia /boot/syslinux/syslinux.cfg 
    APPEND root=/dev/disk/by-uuid/9cf58b0d-99a9-4643-b261-646f88c9a463 quiet ro resume=/dev/disk/by-uuid/76d65c00-5827-4707-9396-2227bf1f3bf8 pcie_aspm=force consoleblank=0 modprobe.blacklist=pata_acpi threadirqs nvidia-drm.modeset=1 nopti nospectre_v2

Then made a new initramfs and rebooted:

koko@slimer# dmesg|grep -i nvidia
    0.000000] Command line: BOOT_IMAGE=../vmlinuz-linux root=/dev/disk/by-uuid/9cf58b0d-99a9-4643-b261-646f88c9a463 quiet ro resume=/dev/disk/by-uuid/76d65c00-5827-4707-9396-2227bf1f3bf8 pcie_aspm=force consoleblank=0 modprobe.blacklist=pata_acpi threadirqs nvidia-drm.modeset=1 nopti nospectre_v2 initrd=../intel-ucode.img,../initramfs-linux.img
[    0.000000] Kernel command line: BOOT_IMAGE=../vmlinuz-linux root=/dev/disk/by-uuid/9cf58b0d-99a9-4643-b261-646f88c9a463 quiet ro resume=/dev/disk/by-uuid/76d65c00-5827-4707-9396-2227bf1f3bf8 pcie_aspm=force consoleblank=0 modprobe.blacklist=pata_acpi threadirqs nvidia-drm.modeset=1 nopti nospectre_v2 initrd=../intel-ucode.img,../initramfs-linux.img
[    1.264268] nvidia: loading out-of-tree module taints kernel.
[    1.264273] nvidia: module license 'NVIDIA' taints kernel.
[    1.272359] nvidia-nvlink: Nvlink Core is being initialized, major device number 244
[    1.272606] nvidia 0000:01:00.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=none:owns=io+mem
[    1.272681] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  390.48  Thu Mar 22 00:42:57 PDT 2018 (using threaded interrupts)
[    1.274019] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms  390.48  Wed Mar 21 23:48:34 PDT 2018
[    1.277713] nvidia-uvm: Loaded the UVM driver in 8 mode, major device number 243
[    1.282283] [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
[    1.652817] caller _nv001169rm+0xe3/0x1d0 [nvidia] mapping multiple BARs
[    1.704346] caller _nv029948rm+0x57/0x90 [nvidia] mapping multiple BARs
[    1.948551] nvidia-modeset: Allocated GPU:0 (GPU-8d5fb2f5-142c-f0e9-6aea-fa9d822a9fc7) @ PCI:0000:01:00.0
[    1.996334] [drm] Initialized nvidia-drm 0.0.0 20160202 for 0000:01:00.0 on minor 0
[    5.197694] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input13
[    5.197866] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input14
[    5.197999] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input15
[    5.198114] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input16
[    5.970061] Bad or missing usercopy whitelist? Kernel memory exposure attempt detected from SLUB object 'nvidia_stack_cache' (offset 11440, size 3)!
[    5.970095]  crc16 mbcache jbd2 fscrypto ehci_pci ehci_hcd crc32c_intel usbcore usb_common nvidia_drm(PO) drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm agpgart nvidia_uvm(PO) nvidia_modeset(PO) nvidia(PO) ipmi_devintf ipmi_msghandler sr_mod cdrom sd_mod ahci libahci libata scsi_mod
[    5.970231]  os_memcpy_to_user+0x21/0x40 [nvidia]
[    5.970348]  _nv009377rm+0xbf/0xe0 [nvidia]
[    5.970440]  ? _nv028067rm+0x79/0x90 [nvidia]
[    5.970532]  ? _nv028067rm+0x55/0x90 [nvidia]
[    5.970618]  ? _nv013694rm+0xee/0x100 [nvidia]
[    5.970704]  ? _nv015342rm+0x154/0x270 [nvidia]
[    5.970816]  ? _nv008310rm+0x134/0x1a0 [nvidia]
[    5.970928]  ? _nv008289rm+0x29c/0x2b0 [nvidia]
[    5.971041]  ? _nv001072rm+0xe/0x20 [nvidia]
[    5.971153]  ? _nv007316rm+0xd8/0x100 [nvidia]
[    5.971264]  ? _nv001171rm+0x627/0x830 [nvidia]
[    5.971376]  ? rm_ioctl+0x73/0x100 [nvidia]
[    5.971441]  ? nvidia_ioctl+0x573/0x720 [nvidia]
[    5.971507]  ? nvidia_frontend_unlocked_ioctl+0x3e/0x50 [nvidia]

...but still the text console resolution is like 640x480 and fullscreen, and VT/Xorg switch is stil very slow.
I'm not using UEFI, but CSM.

Am i doing something wrong or is this supposed to work this way?

Last edited by kokoko3k (2018-05-12 15:17:23)

V1del · 2018-05-12 14:49:49

It's supposed to work this way, it doesn't help with resolution, however there are/were cases where you couldn't switch to a VT at all without enabling modesetting. It's not going to be faster and not going to help the resolution.

kokoko3k · 2018-05-12 15:17:13

In the past, i've read that some users have had the mode changed, but not the actual resolution (hence only a part of the screen was filled by the console text), but this is not my case.
Maybe is my monitor scaling the picture.

Anyway, i'm playing with standard vga=<mode> and i found an interesting thing... if i set the mode to one that has bitdepth=32 (for dimensions >=1024x768, don't know why), then the switch VT->Xorg is really fast..
Switching from Xorg to tty is still slow, but is still something...
Thanks.

Last edited by kokoko3k (2018-05-12 15:37:06)

Arch Linux

#1 2018-05-12 14:20:20

[SOLVED] Unable to use nvidia modeset for fast VT switching

#2 2018-05-12 14:49:49

Re: [SOLVED] Unable to use nvidia modeset for fast VT switching

#3 2018-05-12 15:17:13

Re: [SOLVED] Unable to use nvidia modeset for fast VT switching

Board footer