Lenovo P50 (4k) Thread

[rico666]

Phew! After nearly 3 days of hard installation work, I'm finally able to write this on my new Lenovo P50 (4k display). I'd like to share some of my experiences for those who might want to install Gentoo (or linux in general) on this machine and maybe even get some answers myself to some of the still persisting issues.

The hardware is still quite new (Xeon 1505 v5, Intel P530 GFX, Nvidia Quadro M2000M, Samsung 950 NVMe) and as for basic functionality (It boots , has - most of the time - ethernet network, X works somehow and one can even hear sounds) it works quite ok.
I'm running a fairly new 4.5.2 kernel, so AFAIK much of the hardware should actually be supported already.

Booting:

I have it in a Win10/Gentoo Dual-Boot configuration and somehow I do manage to win over the horror that is EFI. You must know, my previous Thinkpad was a W500 and it worked like a charm with lilo - after I kicked grub off-disk. Now EFI is what I would call a reference example of 2nd system syndrome. Good intentions, but bloaty, quirky and - for a very special definition of engineering - over-engineered.

To install Gentoo I ended up using about 5 USB sticks. Original plan was 2: One to hold the Win10 recovery and one with a AMD64 Gentoo minimal install. After the win recovery was done and AMD64 minimal install was on the USB, I tried to boot from the Gentoo stick and ... failed.
It didn't boot in EFI (only) mode at all. It would boot in Compatibility mode, but I didn't want that as it seems EFI loaders have trouble with mixed mode OS booting.

I then put Systemrescuecd on a (3rd) USB stick and as that didn't boot also, I put Arch Linux on a (4th) USB stick and a stage3/portage on yet another (5th) stick. The Arch did boot in EFI only, so I used that to create a partition (/dev/nvme0n1p5), unpacked the stage3 and portage on it and rebooted with that being my root.

From then on it was the classical emerge orgy that is Gentoo. Some hundreds of emerged packages later, I'm writing this in an weird Google Chrome (with gargantuan font for URL, window header etc.) running in xfce4 environment on xorg - running probably the Intel P530 gfx.

I have compiled most of the packages with the basic "-O2 -pipe" CFLAGS, but right now I'm getting more bold so I've added the "-march=core-avx2" to CFLAGS. It's suboptimal, because it still doesn't enable the AVX512 optimizations, which would be with "-march=skylake-avx512" or something like that, but gcc 5.3.0 seems unaware of that.

I've set xfce font management manually to 282dpi - with the help of https://www.sven.de/dpi/, but my xorg is still telling me

Code: Select all

screen #0:
  dimensions:    3840x2160 pixels (1016x571 millimeters)
  resolution:    96x96 dots per inch

lol. No matter what I do. I mean I do have

Code: Select all

Section "Monitor"
    Identifier     "Monitor0"
    VendorName     "Unknown"
    ModelName      "Unknown"
    HorizSync       28.0 - 33.0
    VertRefresh     43.0 - 72.0
    Option         "DPMS"
    Option         "DPI" "282 x 282"
    DisplaySize    345 194
EndSection

but X still tells me on startup

Code: Select all

[     6.832] (==) intel(0): DPI set to (96, 96)
...
[     6.858] (II) intel(0): switch to mode 3840x2160@60.0 on eDP1 using pipe 0, position (0, 0), rotation normal, reflection none
[     6.858] (II) intel(0): Setting screen physical size to 1016 x 571

right. Let's take a step back and look at the hardware in general.

Code: Select all

# lspci
00:00.0 Host bridge: Intel Corporation Skylake Host Bridge/DRAM Registers (rev 07)
00:01.0 PCI bridge: Intel Corporation Skylake PCIe Controller (x16) (rev 07)
00:02.0 VGA compatible controller: Intel Corporation HD Graphics P530 (rev 06)
00:14.0 USB controller: Intel Corporation Sunrise Point-H USB 3.0 xHCI Controller (rev 31)
00:14.2 Signal processing controller: Intel Corporation Sunrise Point-H Thermal subsystem (rev 31)
00:16.0 Communication controller: Intel Corporation Sunrise Point-H CSME HECI #1 (rev 31)
00:16.3 Serial controller: Intel Corporation Sunrise Point-H KT Redirection (rev 31)
00:17.0 SATA controller: Intel Corporation Sunrise Point-H SATA controller [AHCI mode] (rev 31)
00:1c.0 PCI bridge: Intel Corporation Sunrise Point-H PCI Express Root Port #1 (rev f1)
00:1c.2 PCI bridge: Intel Corporation Sunrise Point-H PCI Express Root Port #3 (rev f1)
00:1c.4 PCI bridge: Intel Corporation Sunrise Point-H PCI Express Root Port #5 (rev f1)
00:1d.0 PCI bridge: Intel Corporation Sunrise Point-H PCI Express Root Port #9 (rev f1)
00:1d.4 PCI bridge: Intel Corporation Sunrise Point-H PCI Express Root Port #13 (rev f1)
00:1f.0 ISA bridge: Intel Corporation Sunrise Point-H LPC Controller (rev 31)
00:1f.2 Memory controller: Intel Corporation Sunrise Point-H PMC (rev 31)
00:1f.3 Audio device: Intel Corporation Sunrise Point-H HD Audio (rev 31)
00:1f.4 SMBus: Intel Corporation Sunrise Point-H SMBus (rev 31)
00:1f.6 Ethernet controller: Intel Corporation Ethernet Connection (2) I219-LM (rev 31)
01:00.0 VGA compatible controller: NVIDIA Corporation GM107GLM [Quadro M2000M] (rev a2)
04:00.0 Network controller: Intel Corporation Wireless 8260 (rev 3a)
3e:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller (rev 01)
3f:00.0 Unassigned class [ff00]: Realtek Semiconductor Co., Ltd. RTS525A PCI Express Card Reader (rev 01)

I've not managed yet to get the SD-Card reader working. Inserting a SD card does not even show up in some dmesg-event, so probably I have not even the correct driver in kernel for that.
The I219-LM ethernet works good (so far) with the e1000 driver in 4.5.2. I've had one or two hiccups where I hadn't connectivity, there was no uplink or dmesg said something of 10MBps, but I'd like to believe these were exceptions.

The builtin-USB Devices present are

Code: Select all

# lsusb
Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 001 Device 004: ID 138a:0090 Validity Sensors, Inc. 
Bus 001 Device 003: ID 04f2:b52c Chicony Electronics Co., Ltd 
Bus 001 Device 007: ID 8087:0a2b Intel Corp. 
Bus 001 Device 006: ID 0765:5010 X-Rite, Inc. 
Bus 001 Device 005: ID 058f:9540 Alcor Micro Corp. AU9540 Smartcard Reader
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

I've managed to get the builtin cam working with the regular uvcvideo driver. For fingerprint and X-Rite (color calibration) I have not high hopes at the moment, so I didn't even try.

For sound and cam to work I had to add the user to the "audio" and "video" group.

I would like to add my kernel config, Xorg config etc. as attachments but am not sure how to do that in this forum without exploding the size of a posting.


Update 2016-05-09:

* running 4.5.3 now

* Wireless (Intel 8260) works now. I had to enable the "Intel Wireless WiFi MVM Firmware support" and did download the iwlwifi-8000C-*.ucode package directly from intel, although probably just emerging "sys-kernel/linux-firmware" would have done the job too.

* The SD card reader works too:

Code: Select all

[62109.436176] mmc0: new ultra high speed SDR104 SDXC card at address 0007
[62109.438696] mmcblk0: mmc0:0007 SD64G 58.1 GiB 

and a SDXC class 10 card gives some 30MB/s

Code: Select all

# hdparm -t /dev/mmcblk0

/dev/mmcblk0:
 Timing buffered disk reads:  98 MB in  3.02 seconds =  32.43 MB/sec

* I did some battery benchmarks. Normal work, with the occasional compile, web browsing, screen full brightness on and heavy shell hacking will result in about 5-6 hours of battery life.

* A fully charged battery should be able to hold the notebook in suspend-to-ram for 120-130 hours.

* Most of the time, the sucker just doesn't spin the fans

Code: Select all

# cat /proc/acpi/ibm/fan 
status:         enabled
speed:          0
level:          auto

It doesn't need to, the machine is cool. And I'm ok with that.

* The function keys are still not working, but you can do stuff like

Code: Select all

echo 1 > /proc/acpi/ibm/kbdlight

so I guess it's only a matter of time until I figure out how to translate that to a keypress.



Rico

[Ant P.]

I have compiled most of the packages with the basic "-O2 -pipe" CFLAGS, but right now I'm getting more bold so I've added the "-march=core-avx2" to CFLAGS. It's suboptimal, because it still doesn't enable the AVX512 optimizations, which would be with "-march=skylake-avx512" or something like that, but gcc 5.3.0 seems unaware of that.

Why not -march=native?

[rico666]

Why not -march=native?

Two reasons:
First, because it seems also not to get Skylake options right with Gcc 5.3.0

Code: Select all

gcc -c -Q -march=native --help=target

yields

Code: Select all

...
  -mavx                                 [enabled]
  -mavx2                                [enabled]
  -mavx256-split-unaligned-load         [disabled]
  -mavx256-split-unaligned-store        [disabled]
  -mavx512bw                            [disabled]
  -mavx512cd                            [disabled]
  -mavx512dq                            [disabled]
  -mavx512er                            [disabled]
  -mavx512f                             [disabled]
  -mavx512ifma                          [disabled]
  -mavx512pf                            [disabled]
  -mavx512vbmi                          [disabled]
  -mavx512vl                            [disabled]
...

The avx/avx2 is ok, why it doesn't enable the avx512 is unknown to me.

Second, march=native is evil when you do distcc and I didn't want to block that path.


Rico

[alexcortes]

Hi,

X-Lite will probably work very well.
Take a look at:

Code: Select all

http://www.argyllcms.com

[MarcusXP]

hey OP, sorry to hijack your thread.. how did you configure X to work on this machine?
Are you using nvidia optimus in BIOS or discrete graphics only?
I am using discrete graphics and doesn't work with nouveau driver..

thanks a lot!

[alexcortes]

so I guess it's only a matter of time until I figure out how to translate that to a keypress.

x11-apps/xev will give you a hand, but you will need xmodmap and/or xbindkays to archive that:

https://wiki.archlinux.org/index.php/Xmodmap

https://wiki.archlinux.org/index.php/Xbindkeys

http://dev-loki.blogspot.com.br/2006/04 ... odmap.html

[MarcusXP]

Any help for my question, please ?

[Ant P.]

The avx/avx2 is ok, why it doesn't enable the avx512 is unknown to me.

Second, march=native is evil when you do distcc and I didn't want to block that path.

Oh, good points. I think march=native depends on what the kernel reports via /proc/cpuinfo...

[haarp]

I've set xfce font management manually to 282dpi - with the help of https://www.sven.de/dpi/, but my xorg is still telling me

Code: Select all

screen #0:
  dimensions:    3840x2160 pixels (1016x571 millimeters)
  resolution:    96x96 dots per inch

lol. No matter what I do. I mean I do have

Code: Select all

Section "Monitor"
    Identifier     "Monitor0"
    VendorName     "Unknown"
    ModelName      "Unknown"
    HorizSync       28.0 - 33.0
    VertRefresh     43.0 - 72.0
    Option         "DPMS"
    Option         "DPI" "282 x 282"
    DisplaySize    345 194
EndSection

but X still tells me on startup

Code: Select all

[     6.832] (==) intel(0): DPI set to (96, 96)
...
[     6.858] (II) intel(0): switch to mode 3840x2160@60.0 on eDP1 using pipe 0, position (0, 0), rotation normal, reflection none
[     6.858] (II) intel(0): Setting screen physical size to 1016 x 571

right.

This used to work correctly. Then the X devs decided to intentionally stop using the actual DPI and ignore attempts at telling it the DPI via xorg.conf. All in favor of improving compatibility with Window's "96dpi or the highway" mode.

https://bugs.freedesktop.org/show_bug.cgi?id=23705
https://bugs.freedesktop.org/show_bug.cgi?id=41115

You can tell apps to use the proper DPI for fonts in Xfce, as you're doing. If you're doing graphical work with GIMP or such however, this is insufficient. You can use xrandr --dpi 142 to set it. Use xdpyinfo | grep reso to read the currently set DPI.

[kaktuspalme]

I also bought a Lenovo P50 with a 4K screen.

I have problems with suspending the notebook or waking up from suspend.
As soon as I wake up the notebook only the hdd led starts blinking but nothing else happens.

I have to power down the notebook by holding the power button.

No one suffering from this problem?

I have the newest UEFI Bios (1.29) and I'm running kernel 4.7.0.
I read elsewhere that boot parameter intel_pstate=no_hwp should help, but nothing changes.
When I set acpi=off to the boot parameter the notebook even doesn't boot completely.

It doesn't matter how I suspend the notebook, by lid closing or via systemctl-suspend it just doesn't work.

For the moment I have to use windows on this notebook but I really hate that.


I configured the kernel using genkernel and made some tweeks by menuconfig (iwlwifi firmware etc.).
Maybe someone could give me the kernel config?

[tamiko]

I also bought a Lenovo P50 with a 4K screen.

[...]

I have to power down the notebook by holding the power button.

No one suffering from this problem?

This is very likely an issue with a kernel lacking drivers for all relevant hardware. I had the same issue with a Lenovo T540p (wakeup from suspend freezed the system) and back then a missing driver for the TPM chip was the problem.
I can confirm that suspend/resume works on a p50. I have the following tpm related configuration options enabled (for the p50):

Code: Select all

CONFIG_TCG_TPM=y
CONFIG_TCG_TIS=y

It is also advisable to set the TPM mode to 1.2 within the BIOS/UEFI (because 2.0 isn't really supported by applications yet).

[kaktuspalme]

Code: Select all

CONFIG_TCG_TPM=y
CONFIG_TCG_TIS=y

Thank you very much tamiko. The tpm driver completely solved my problem.

[tamiko]

I started writing up a short wiki entry for the P50:
https://wiki.gentoo.org/wiki/Lenovo_ThinkPad_P50

[tamiko]

[...]

The avx/avx2 is ok, why it doesn't enable the avx512 is unknown to me.

Rico

Because your processor simply doesn't support avx512. For the skylake architecture introduction of avx512 is expected to be with the "Purley" architecture for Skylake Xeons.

[rico666]

The avx/avx2 is ok, why it doesn't enable the avx512 is unknown to me.

Because your processor simply doesn't support avx512. For the skylake architecture introduction of avx512 is expected to be with the "Purley" architecture for Skylake Xeons.

Confirmed. Unfortunately I thought by having a "Skylake Xeon" I'm in the AVX512 game as Intel presented it like that. Well...

Several months in, I'm trying to work out a good working Hybrid graphics configuration and a good OpenCL development environment, but it's not going well.

When working with the notebook in GPU "discrete mode" (set in BIOS), all was ok. Now I am

* hybrid graphics
* kernel 4.8.4 and bleeding edge nvidia-drivers 375

things are gloom. Nvidia OpenGL causes segfaults all the time, but I'll address that later. Right now, I seem to not even be able to get the P530 graphics working ok.
I had a look at https://wiki.gentoo.org/wiki/Lenovo_ThinkPad_P50 (which I hopefully can contribute to at some point), but e.g. video output with fullscreen is not possible

I'm pretty sure I do have quite recent versions:

Code: Select all

# eix -c intel
[I] dev-util/intel-ocl-sdk (4.4.0.117-r1@10/29/2016): Intel's implementation of the OpenCL standard
[I] sys-firmware/intel-microcode (20160714@10/29/2016): Intel IA32/IA64 microcode update data
[I] x11-apps/intel-gpu-tools (1.15@10/29/2016): Intel GPU userland tools
[I] x11-drivers/xf86-video-intel (2.99.917_p20161024@10/29/2016): X.Org driver for Intel cards
[I] x11-libs/libva-intel-driver (1.7.2@10/29/2016): HW video decode support for Intel integrated graphics
[I] sys-kernel/linux-firmware (99999999@10/28/2016): Linux firmware files

Xorg setting for intel as recommended all over the place:

Code: Select all

Section "Device"
    Option "AccelMethod" "uxa" 
    Identifier "modesetting"
    Driver "modesetting"
EndSection

But xvinfo sees no adapter:

Code: Select all

$ xvinfo 
X-Video Extension version 2.2
screen #0
 no adaptors present

And thus mplayer experience is poor also:

Code: Select all

...
[vdpau] Error when calling vdp_device_create_x11: 1
...

Rico