Dell XPS 15 9570
Device/Functionality | Status |
Suspend | Working |
Hibernate | Working |
Integrated Graphics | Working |
Discrete Nvidia Graphics | Modify |
Wifi | Working |
Bluetooth | Working |
rfkill | Working |
Audio | Working |
Touchpad | Working |
Touchscreen | Working |
Webcam | Working |
Card Reader | Working |
Function/Multimedia Keys | Working |
Power Management | Working |
EFI firmware updates | Working |
Fingerprint reader[broken link: invalid section] | Not working |
Contents
UEFI
Before installing it is necessary to modify some UEFI Settings. They can be accessed by pressing the F2 key repeatedly when booting.
- Change the SATA Mode from the default "RAID" to "AHCI". This will allow Linux to detect the NVME SSD. If dual booting with an existing Windows installation, Windows will not boot after the change but this can be fixed without a reinstallation.
- Change Fastboot to "Thorough" in "POST Behaviour". This prevents intermittent boot failures.
- Disable secure boot to allow Linux to boot. Alternatively, you can choose to set up Secure Boot so that it works with Linux and you can take advantage of its potential security benefits.
Installation of Arch Linux can proceed normally. Refer to the Installation guide for more information.
Power Management
Suspend
By default, the very inefficient s2idle suspend variant is incorrectly selected. This is probably due to the BIOS. The much more efficient deep variant should be selected instead:
$ cat /sys/power/mem_sleep [s2idle] deep $ echo deep|sudo tee /sys/power/mem_sleep $ cat /sys/power/mem_sleep s2idle [deep]
To make the change permanent add mem_sleep_default=deep
to your kernel parameters.
An easy way would be to add mem_sleep_default=deep
to the GRUB_CMDLINE_LINUX_DEFAULT
entry in /etc/default/grub:
GRUB_CMDLINE_LINUX_DEFAULT="mem_sleep_default=deep"
Read more regarding the sleep variants on the kernel documentation [1].
Hibernate
Works out of the Box see Power management/Suspend and hibernate
Powertop
powertop
is very efficient to manage power consumption.
Run powertop --auto-tune
and compare the Watt consumption variation (battery must be unplugged). powertop --auto-tune
can be run at every boot.
Graphics
kernel modules
The nouveau module is known to cause kernel panics and freezes on Dell XPS 15 9570.
One way to mitigate this would be by adding nomodeset
to the kernel options. However it's better to completely disable it with the blacklist method (recommended):
/etc/modprobe.d/blacklist.conf
blacklist nouveau blacklist rivafb blacklist nvidiafb blacklist rivatv blacklist nv
NVIDIA Optimus
The optimus configuration is a technology that allows an Intel integrated GPU and discrete NVIDIA GPU to be built into and accessed by a laptop. As the discret NVIDIA GPU card eats lots of power, we want to use the intergrated Intel card most of the time and activate/desactivate the NVIDIA GPU card only when required by a defined application. Due to a bug [3], the XPS 15 9570's secondary NVIDIA GPU cannot be powered down by bbswitch. However, standard Linux power management is able to properly turn off the card when the driver is unloaded.
Manually loading/unloading NVIDIA module
One solution [4] to the problem described above is to manually unload the nvidia module when not in use, which can be done as follows:
- bumblebee - The main package providing the daemon and client programs.
- nvidia - Official NVIDIA drivers for Linux.
- powertop - Tool to verify power consumption.
- unigine-valleyAUR - Graphic intensive application for testing
Make sure that bbswitch is uninstalled or at least disabled
bumblebee
configuration, in the [driver-nvidia]
section
/etc/bumblebee/bumblebee.conf
# [driver-nvidia] section Driver=nvidia PMMethod=none
Server X configuration for not auto adding gpu
/etc/X11/xorg.conf.d/01-noautogpu.conf
Section "ServerFlags" Option "AutoAddGPU" "off" EndSection
ipmi
modules are loaded together with nvidia
and block its unloading. Create the two following files to disable them:
/etc/modprobe.d/disable-ipmi.conf
install ipmi_msghandler /usr/bin/false install ipmi_devintf /usr/bin/false
/etc/modprobe.d/disable-nvidia.conf
install nvidia /bin/false
Now, add nvidia
and ipmi
in the modprobe.d blacklist to disable this functionality:
/etc/modprobe.d/blacklist.conf
blacklist nouveau blacklist rivafb blacklist nvidiafb blacklist rivatv blacklist nv blacklist nvidia blacklist nvidia-drm blacklist nvidia-modeset blacklist nvidia-uvm blacklist ipmi_msghandler blacklist ipmi_devintf
Create 2 GPU management scripts for enabling and disabling discret NVIDIA graphical card [5]:
enablegpu.sh
#!/bin/sh # allow to load nvidia module mv /etc/modprobe.d/disable-nvidia.conf /etc/modprobe.d/disable-nvidia.conf.disable # Remove NVIDIA card (currently in power/control = auto) echo -n 1 > /sys/bus/pci/devices/0000\:01\:00.0/remove sleep 1 # change PCIe power control echo -n on > /sys/bus/pci/devices/0000\:00\:01.0/power/control sleep 1 # rescan for NVIDIA card (defaults to power/control = on) echo -n 1 > /sys/bus/pci/rescan # someone said that modprobe nvidia is needed also to load nvidia, to check # modprobe nvidia
disablegpu.sh
#!/bin/sh modprobe -r nvidia_drm modprobe -r nvidia_uvm modprobe -r nvidia_modeset modprobe -r nvidia # Change NVIDIA card power control echo -n auto > /sys/bus/pci/devices/0000\:01\:00.0/power/control sleep 1 # change PCIe power control echo -n auto > /sys/bus/pci/devices/0000\:00\:01.0/power/control sleep 1 # Lock system form loading nvidia module mv /etc/modprobe.d/disable-nvidia.conf.disable /etc/modprobe.d/disable-nvidia.conf
Create service locking NVIDIA GPU on shutdown
A service which locks GPU on shutdown / restart when it is not disabled by disablegpu.sh
script is necessary. Otherwise, on next boot nvidia
will be loaded together with ipmi
modules (even if blacklisted with install
command) and it won't be possible to unload them then.
/etc/systemd/system/disable-nvidia-on-shutdown.service
[Unit] Description=Disables Nvidia GPU on OS shutdown [Service] Type=oneshot RemainAfterExit=true ExecStart=/bin/true ExecStop=/bin/bash -c "mv /etc/modprobe.d/disable-nvidia.conf.disable /etc/modprobe.d/disable-nvidia.conf || true" [Install] WantedBy=multi-user.target
Reload systemd daemons and enable the disable-nvidia-on-shutdown
service:
sudo systemctl daemon-reload sudo systemctl enable disable-nvidia-on-shutdown.service
Allow gpu to poweroff on boot
/etc/tmpfiles.d/nvidia_pm.conf
w /sys/bus/pci/devices/0000:01:00.0/power/control - - - - auto
Finally, check that everything is well configured:
- Reboot and verify that nvidia is not loaded by running:
lsmod | grep nvidia
- Should not return anything
- Disconnect / unplug charger and verify the power consumption with powertop is around 5W on idle (Dell XPS 15 4570, powertop --auto-tune previously launched, FHD screen with no touchscreen)
- Enable GPU by using the enablegpu.sh script
- Check that the GPU is loaded by using:
nvidia-smi
- If good, launch unigine-valley with optirun:
optirun unigine-valley
- unigine-valley needs a GPU to work. Should activate the fans quickly.
- Close all nvidia applications and disable gpu with the disablegpu.sh script
- Check again power consumption, it should have gone back to a similar value as before
Letting bumblebee automatically unload the kernel module
Another possible way of shutting off the secondary GPU is to use a patched version of bumblebee that automatically unloads the nvidia kernel module when the card is not in use [6]. This patch has already been merged into the development
branch of bumblebee [7] which can be obtained via bumblebee-gitAUR. It is also provided by bumblebee-forceunloadAUR, which is a modified version of bumblebee with this patch applied.
-
bumblebee-gitAUR or bumblebee-forceunloadAUR - Bumblebee with the
AlwaysUnloadKernelDriver
option - nvidia - Official NVIDIA drivers for Linux.
- mesa-demos - Mesa demos for testing device graphics capabilities
Make sure that AlwaysUnloadKernelDriver
is set to true
in the driver-nvidia
section of your bumblebee
configuration:
/etc/bumblebee/bumblebee.conf
# [driver-nvidia] section KernelDriver=nvidia PMMethod=auto AlwaysUnloadKernelDriver=true
While the above option unloads the nvidia kernel module when not in use, Xorg will still automatically discover the secondary graphics card and attempt to enable it. In order to prevent this, the option AutoAddGPU
should be disabled. However, this would lead to Xorg unable to detect the default integrated graphics card. The intel graphics card needs to be manually added to Xorg. This can be done by creating the following two configuration files in /etc/X11/xorg.conf.d
:
/etc/X11/xorg.conf.d/01-noautogpu.conf
Section "ServerFlags" Option "AutoAddGPU" "off" EndSection
/etc/X11/xorg.conf.d/20-intel.conf
Section "Device" Identifier "Intel Graphics" Driver "intel" # Uncomment line below if you are also having issues with backlight # Option "Backlight" "intel_backlight" EndSection
Set GPU power control to auto on boot so it will be powered off when the module is unloaded:
/etc/tmpfiles.d/nvidia_pm.conf
w /sys/bus/pci/devices/0000:01:00.0/power/control - - - - auto
Enable bumblebeed.service
to start on boot:
sudo systemctl enable bumblebeed.service
After restarting your system, you should be able to verify that everything is working properly:
- Verify that nvidia is not loaded on boot by running:
lsmod | grep nvidia
- Should not return anything
- Check that secondary graphics card is functioning properly with optirun:
optirun glxgears
- Don't close this window just yet, we need to check that the nvidia module is loaded correctly:
lsmod | grep nvidia
- This time, it should be able to find nvidia in the loaded modules list
- After closing
glxgears
, check again to make sure that nvidia is properly unloaded:lsmod | grep nvidia
- Should not return anything once again
- At this point, the graphics card should be working correctly. You can further check the power consumption by disconnecting the charger and using Powertop, there should be a significant power difference when
optirun glxgears
is running and when it is not.
Troubleshooting
xbacklight
To get xbacklight working and not conflicting with NVIDIA optimus:
/etc/X11/xorg.conf.d/20-intel.conf
Section "Device" Identifier "Intel Graphics" Driver "intel" Option "Backlight" "intel_backlight" EndSection
NVRM: Failed to enable MSI; falling back to PCIe virtual-wire interrupts
Sometimes it happens after suspend/resume. GPU could work fine without MSI. [8]. You could disable MSI by adding the following in /etc/modprobe.d/nvidia.conf:
options nvidia NVreg_EnableMSI=0
Built-in screen flickers or does not come on with Linux kernel 5.0.0 - 5.0.7
Some users reported that running Linux kernel 5.0.0 to 5.0.7 can cause the screen to flicker (or stay completely black) when booting up or running an X server, making the built-in display unusable (see *[9])
Currently, it seems that there are three possible workarounds :
- Downgrade to Linux 4.20.13.
- Apply Albert Astals Cid's patch on Linux kernel 5.0.x (see kernel Arch Build System).
- Install linux-lts
- upgrade to Linux 5.0.8 or higher
Lock-ups when resuming from suspend with nvidia module
If your system locks up every time you resume from suspend with the following two lines in dmesg:
[ 42.447364] pci 0000:01:00.0: Refused to change power state, currently in D3 [ 46.896493] pci 0000:01:00.0: Refused to change power state, currently in D3
you need to do the following:
Into /etc/default/grub
GRUB_CMDLINE_LINUX="nouveau.blacklist=0 acpi_osi=! acpi_osi=\"Windows 2015\" acpi_backlight=vendor mem_sleep_default=deep"
This solved the lock-ups for me. (internal nvidia bug number: 2589324, dell resolution)
Wifi and Bluetooth
These work well out of the box but you might need to update the firmware for better stability. For bluetooth, make sure you have everything installed as per the Bluetooth wiki page.
$ lspci | grep Network 3b:00.0 Network controller: Qualcomm Atheros QCA6174 802.11ac Wireless Network Adapter (rev 32)
(please add a line here if you detect something different)
Troubleshooting
ath10k module crashes after suspend
You may notice driver crashes after suspend/resume, which for the most part does not seem to impact the running system, with coredumps similar to:
kernel: WARNING: CPU: 6 PID: 27936 at drivers/net/wireless/ath/ath10k/mac.c:5746 ath10k_bss_info_changed+0xf96/0x1120 [ath10k_core] kernel: Modules linked in: uhid algif_hash cmac rfcomm fuse ccm ipt_MASQUERADE nf_conntrack_netlink nfnetlink xfrm_user xfrm_algo iptable_nat nf_nat_ipv4> kernel: snd_hda_intel dcdbas x86_pkg_temp_thermal dell_smm_hwmon snd_hda_codec intel_powerclamp cfg80211 kvm_intel snd_hda_core input_leds snd_hwdep snd> kernel: vfio_mdev mdev vfio_iommu_type1 vfio kvm irqbypass intel_gtt i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm agpga> kernel: CPU: 6 PID: 27936 Comm: kworker/u24:42 Tainted: G U W OE 5.0.4-arch1-1-ARCH #1 kernel: Hardware name: Dell Inc. XPS 15 9570/0HWTMH, BIOS 1.8.1 02/01/2019 kernel: Workqueue: events_unbound async_run_entry_fn kernel: RIP: 0010:ath10k_bss_info_changed+0xf96/0x1120 [ath10k_core] kernel: Code: 24 8b 95 78 01 00 00 85 c0 0f 85 a7 00 00 00 89 d1 be 10 00 00 00 48 c7 c2 c0 b2 fa c0 4c 89 c7 e8 bf 68 00 00 e9 a5 f1 ff ff <0f> 0b 4c 89> kernel: RSP: 0000:ffffb7a45422fcd0 EFLAGS: 00010282 kernel: RAX: 00000000fffffffe RBX: ffff98f6d44815a0 RCX: 0000000000000000 kernel: RDX: ffff98f6d4481938 RSI: ffffb7a45422fcf0 RDI: ffff98f6d81df418 kernel: RBP: ffff98f6d81df418 R08: 0000000000000001 R09: 0000000000000000 kernel: R10: 000000000000001f R11: 0000000000000000 R12: 0000000000000020 kernel: R13: ffff98f6d81df420 R14: ffff98f6d4482598 R15: ffff98f6d44815a0 kernel: FS: 0000000000000000(0000) GS:ffff98f6dc380000(0000) knlGS:0000000000000000 179 94% kernel: process_one_work+0x1eb/0x410 kernel: worker_thread+0x2d/0x3d0 kernel: ? process_one_work+0x410/0x410 kernel: kthread+0x112/0x130 kernel: ? kthread_park+0x80/0x80 kernel: ret_from_fork+0x35/0x40 kernel: ---[ end trace 09ae3e174c178f98 ]---
Patched in some kernels and not others (which?), relevant links:
Bluetooth disappears (after suspend?)
Possibly related to the previous instability issue, sometimes the adapter seems to completely disappear. As reported here (thanks w.v.w., you can likely fix this by manually upgrading the firmware (to something newer than what's in linux-firmware:
1. Double check adapter (e.g. QCA6174, below)
$ lspci | grep Network 3b:00.0 Network controller: Qualcomm Atheros QCA6174 802.11ac Wireless Network Adapter (rev 32)
2. Confirm hardware and current firmware version (e.g. hw3.2, firmware RM.4.4.1.c3-00013-QCARMSWPZ-1, below)
$ journalctl -b | grep ath10k | egrep 'firmware|qca' kernel: ath10k_pci 0000:3b:00.0: qca6174 hw3.2 target 0x05030000 chip_id 0x00340aff sub 1a56:1535 kernel: ath10k_pci 0000:3b:00.0: firmware ver RM.4.4.1.c3-00013-QCARMSWPZ-1 api 6 features wowlan,ignore-otp,no-4addr-pad,raw-mode,mfp crc32 fc0096a8
3. Follow the instructions at https://wireless.wiki.kernel.org/en/users/Drivers/ath10k/firmware
e.g. for the above adapter, download the latest firmware from https://github.com/kvalo/ath10k-firmware/tree/master/QCA6174/hw3.0 and
$ cd /lib/firmware/ath10k/QCA6174/hw3.0 $ sudo cp firmware-6.bin firmware-6.bin.bak $ sudo cp ~/Downloads/firmware-6.bin_RM.4.4.1.c3-00013-QCARMSWPZ-1 firmware-6.bin
Either sudo rmmod ath10k_core ath10k_pci && sudo modprobe ath10k_pci
(may not work, check dmesg) or reboot.
Touchpad and Touchscreen
Wacom touchscreen and Synaptics touchpad:
$ xinput
⎡ Virtual core pointer id=2 [master pointer (3)] ⎜ ↳ Virtual core XTEST pointer id=4 [slave pointer (2)] ⎜ ↳ SYNA2393:00 06CB:7A13 Touchpad id=16 [slave pointer (2)] ⎜ ↳ Wacom HID 488F Finger id=15 [slave pointer (2)] ⎣ Virtual core keyboard id=3 [master keyboard (2)] [truncated]
Both are i2c devices:
$ udevadm info /dev/input/mouse3 # touchscreen
P: /devices/pci0000:00/0000:00:15.0/i2c_designware.0/i2c-10/i2c-WCOM488F:00/0018:056A:488F.0006/input/input47/mouse3 N: input/mouse3 L: 0 E: DEVPATH=/devices/pci0000:00/0000:00:15.0/i2c_designware.0/i2c-10/i2c-WCOM488F:00/0018:056A:488F.0006/input/input47/mouse3 E: DEVNAME=/dev/input/mouse3 E: MAJOR=13 E: MINOR=35 E: SUBSYSTEM=input E: USEC_INITIALIZED=5501073 E: ID_INPUT=1 E: ID_INPUT_TOUCHSCREEN=1 E: ID_PATH=pci-0000:00:15.0-platform-i2c_designware.0 E: ID_PATH_TAG=pci-0000_00_15_0-platform-i2c_designware_0
$ udevadm info /dev/input/mouse6 # touchpad
P: /devices/pci0000:00/0000:00:15.1/i2c_designware.1/i2c-11/i2c-SYNA2393:00/0018:06CB:7A13.0007/input/input38/mouse6 N: input/mouse6 L: 0 S: input/by-path/pci-0000:00:15.1-platform-i2c_designware.1-mouse E: DEVPATH=/devices/pci0000:00/0000:00:15.1/i2c_designware.1/i2c-11/i2c-SYNA2393:00/0018:06CB:7A13.0007/input/input38/mouse6 E: DEVNAME=/dev/input/mouse6 E: MAJOR=13 E: MINOR=38 E: SUBSYSTEM=input E: USEC_INITIALIZED=4683741 E: ID_INPUT=1 E: ID_INPUT_TOUCHPAD=1 E: ID_SERIAL=noserial E: ID_PATH=pci-0000:00:15.1-platform-i2c_designware.1 E: ID_PATH_TAG=pci-0000_00_15_1-platform-i2c_designware_1 E: DEVLINKS=/dev/input/by-path/pci-0000:00:15.1-platform-i2c_designware.1-mouse
Thunderbolt docks
TB16
TB16 works fine if either Thunderbolt security is disabled in the BIOS or using bolt to temporarily authorize or permanently enroll Thunderbolt devices with Thunderbolt security activated. Various quirks are detailed on the Dell TB16 page.
EFI firmware updates
They are 2 main ways to update efi firmware:
- through running linux session with Fwupd
- through UEFI
Fwupd
pacman -S fwupd
Then, look at:
sudo fwupdmgr get-updates sudo fwupdmgr refresh sudo fwupdmgr update
UEFI
Firmware images can be found at Dell support page as XPS_15_9570_X.Y.Z.exe
files.
In order to install:
- Download the desired firmware from section "Dell XPS 15 9570 System BIOS"
- Save it in
/boot/EFI/Dell/Bios/
(this path may vary, depending on your installation) - Reboot the system, and enter the boot menu by pressing repeatedly
F12
on Dell logo - Choose "Bios Flash Update"
- Select the file previously saved, and start the process
The process will take about five minutes, during which the system will have some reboots and push fans at maximum speed. Finally the system will reboot normally.
Thermal management
Thermal design is poor in the 9570, primarily due to higher-end chips being used inside the original system design without compensating for extra heat. This impacts numerous areas:
- Performance: at higher temperatures, CPU cores are throttled down to avoid damage. At best your system will not run as fast as it can, and at worst (and quite commonly), slower than cheaper chips and with sluggish performance.
- Battery life is unnecessarily decreased.
- System longevity: running at constantly high temperatures will negatively impact equipment lifetime.
- User discomfort: uncomfortable heat and uncomfortable fan noise.
Fortunately these can all be improved significantly to get your system running powerfully and quietly.
Diagnosis
You probably have a lot of dmesg output like this (for all CPUs), even with fairly light usage:
kernel: CPU8: Package temperature above threshold, cpu clock throttled (total events = 8451) kernel: CPU8: Package temperature/speed normal
Use lm_sensors and do some stress testing (with stress
and mprime
) to see what's happening with CPU core temperatures and fanspeed under different loads.
Undervolting
See Undervolting CPU on the wiki. Reduces heat and extends battery life.
Possible configurations for /etc/intel-undervolt.conf
:
- i9-8950HK* (last updated 2019-03-30)
undervolt 0 'CPU' -140 undervolt 1 'GPU' -75
* Tested extensively for moderate use. Was not stress tested for > 24 hours. Anecdotal reports of up to -200.
This has a more minor impact than repasting but is significantly easier to do.
Repasting & padding
Significant improvements are possible by replacing the thermal grease used by Dell with better options available from Amazon, and adding some extra thermal padding. This might sound overwhelming but is well worth the effort, especially for newer CPUs. If you can't do this yourself consider finding a shop / technician who can do this for you. See the following article as the user comments below it for more info:
- UltraBookReview: How to Fix Throttling on the Dell XPS 15 9570 / 9560
- Picture of 9570 VRMs - from comments of the above article
Results
According to the author of the UltraBookReview article above:
Undervolting seems to reduce temps at max load by 7-10C, while repasting seems to reduce temps by between 4-10C depending on your original paste job and paste used.
From this forum post, much lower/softer fan speeds were needed to maintain the same temperatures, and temperature was a few degrees lower under similar loads. Fans were on less often and for less time when they were. Throttling was less prevalent and less severe.
Tips and Tricks
Systemd doesn't wait for Network
Few months ago systemd added "after= .. .. network.target" in /usr/lib/systemd/system/systemd-user-sessions.service
This causes systemd to wait for network connection at boot, you can modify this file to remove network.target but it will be overwritten on systemd update. A better workaround is to add /etc/systemd/system/systemd-user-sessions.service with network.target removed.
/etc/systemd/system/systemd-user-sessions.service
# This file is part of systemd. # # systemd is free software; you can redistribute it and/or modify it # under the terms of the GNU Lesser General Public License as published by # the Free Software Foundation; either version 2.1 of the License, or # (at your option) any later version. [Unit] Description=Permit User Sessions Documentation=man:systemd-user-sessions.service(8) After=remote-fs.target nss-user-lookup.target [Service] Type=oneshot RemainAfterExit=yes ExecStart=/usr/lib/systemd/systemd-user-sessions start ExecStop=/usr/lib/systemd/systemd-user-sessions stop