Open denysvitali opened 7 years ago
@denysvitali I started the kernel (4.17/work) now without any problems. I added all my patches and got the system start logs and the terminal command line. For further tests, i need to rebuild the rootfs.
Ps. You did not add all my patches to the kernel.
Can you push your version of the kernel?
Oh I guess I missed your linux fork, I was looking at your linux-smaug repo.
Btw, I've tried the branch w/ your patches and added some of mines, it boots but unfortunately it is still not accelerated, even with drm, mesa and wayland compiled from thierryreding's repo
@denysvitali @Samt43 Can you correct the kernel compilation script (build-linux) so that it can work in the travis-ci? This is necessary for the modules to be loaded at the time the system is started. It is necessary to include in the system the same version of the kernel that is used as a boot
Faced the problem of lack of time to build through Travis. A limit of 50 minutes is not enough for building with the kernel. But the most not logical, is that the assembled assemblies of the rootfs on the home PC do not enter the graphical shell. The same is collected on Travis, it starts normally.
You're probably missing the ./scripts/generate-rootfs
call. This scripts fixes all the permissions on the rootfs.
About the build time of Travis: I'll set up a Jenkins job on one of my own servers (probably jenkins.mastodonti.co) and give us access through the @pixelc-linux organization.
I used it(./scripts/generate-rootfs
), but instead of sudo tar -cvpzf $TOP/out/rootfs.tar.gz
,
i used sudo rsync -aAXv $TOP/out/target/$ARCH/ArchLinuxArm/* $USB_MOUNT/
maybe this is the problem. Now I'll check.
Checked. This is the problem.
@vartom Did you solve the problem you were having?
@denysvitali I do not have so much time now to check everything that I wanted.
I wanted to check again the next set, mesa and drm from @thierryreding, with the 4.15 kernel.
I can not configure the 4.17 kernel to run the nouveau driver as a module. And there are such mistakes Failed to set up IOMMU for device 57000000.gpu; retaining platform DMA ops
In the last test, I checked it. Does not load into the GUI. This makes one think of something else that the kernel or DTs are not all included for the stable operation of the GPU.
What I merged into Mesa should run fine on top of any recent release or linux-next. My for-4.17/work branch contains a bunch of patches that may be unstable.
Today's linux-next (next-20180309) should also have fixes for the green bars and things like that. I think I don't have anything else in my for-4.17/work branch that would be immediately relevant to you.
[ 118.945] Current Operating System: Linux alarm 4.15.6+ #76 SMP PREEMPT Sun Mar 4 22:54:21 +07 2018 aarch64
[ 118.945] Kernel command line: cros_secure vpr=0x08000000@0xf6800000 cros_legacy init=/init rw fbcon loglevel=0 debug earlyprintk drm.atomic=1 no_console_suspend firmware_class.path=/lib/firmware
[ 118.945] Build Date: 30 January 2018 11:53:15PM
[ 118.945]
[ 118.945] Current version of pixman: 0.34.0
[ 118.945] Before reporting problems, check http://wiki.x.org
to make sure that you have the latest version.
[ 118.945] Markers: (--) probed, (**) from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[ 118.946] (==) Log file: "/var/log/Xorg.0.log", Time: Fri Mar 9 15:34:29 2018
[ 118.947] (==) Using system config directory "/usr/share/X11/xorg.conf.d"
[ 118.949] (==) No Layout section. Using the first Screen section.
[ 118.949] (==) No screen section available. Using defaults.
[ 118.949] (**) |-->Screen "Default Screen Section" (0)
[ 118.949] (**) | |-->Monitor "<default monitor>"
[ 118.955] (==) No monitor specified for screen "Default Screen Section".
Using a default monitor configuration.
[ 118.955] (==) Automatically adding devices
[ 118.955] (==) Automatically enabling devices
[ 118.955] (==) Automatically adding GPU devices
[ 118.955] (==) Automatically binding GPU devices
[ 118.955] (==) Max clients allowed: 256, resource mask: 0x1fffff
[ 118.955] (WW) The directory "/usr/share/fonts/Type1/" does not exist.
[ 118.955] Entry deleted from font path.
[ 118.955] (WW) `fonts.dir' not found (or not valid) in "/usr/share/fonts/100dpi/".
[ 118.955] Entry deleted from font path.
[ 118.955] (Run 'mkfontdir' on "/usr/share/fonts/100dpi/").
[ 118.955] (WW) `fonts.dir' not found (or not valid) in "/usr/share/fonts/75dpi/".
[ 118.955] Entry deleted from font path.
[ 118.955] (Run 'mkfontdir' on "/usr/share/fonts/75dpi/").
[ 118.955] (==) FontPath set to:
/usr/share/fonts/misc/,
/usr/share/fonts/TTF/,
/usr/share/fonts/OTF/
[ 118.955] (==) ModulePath set to "/usr/lib/xorg/modules"
[ 118.955] (II) The server relies on udev to provide the list of input devices.
If no devices become available, reconfigure udev or disable AutoAddDevices.
[ 118.957] (II) Loader magic: 0x58510dcd28
[ 118.957] (II) Module ABI versions:
[ 118.957] X.Org ANSI C Emulation: 0.4
[ 118.958] X.Org Video Driver: 23.0
[ 118.958] X.Org XInput driver : 24.1
[ 118.958] X.Org Server Extension : 10.0
[ 118.959] (++) using VT number 2
[ 118.963] (II) systemd-logind: took control of session /org/freedesktop/login1/session/c2
[ 118.965] (II) xfree86: Adding drm device (/dev/dri/card0)
[ 118.966] (II) systemd-logind: got fd for /dev/dri/card0 226:0 fd 12 paused 0
[ 118.967] (II) xfree86: Adding drm device (/dev/dri/card1)
[ 118.968] (II) systemd-logind: got fd for /dev/dri/card1 226:1 fd 13 paused 0
[ 118.969] (II) no primary bus or device found
[ 118.969] falling back to /sys/devices/platform/50000000.host1x/drm/drm/card0
[ 118.969] (II) LoadModule: "glx"
[ 118.972] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
[ 118.987] (II) Module glx: vendor="X.Org Foundation"
[ 118.987] compiled for 1.19.6, module version = 1.0.0
[ 118.987] ABI class: X.Org Server Extension, version 10.0
[ 118.989] (==) Matched modesetting as autoconfigured driver 0
[ 118.989] (==) Matched fbdev as autoconfigured driver 1
[ 118.989] (==) Assigned the driver to the xf86ConfigLayout
[ 118.989] (II) LoadModule: "modesetting"
[ 118.991] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[ 118.996] (II) Module modesetting: vendor="X.Org Foundation"
[ 118.996] compiled for 1.19.6, module version = 1.19.6
[ 118.996] Module class: X.Org Video Driver
[ 118.996] ABI class: X.Org Video Driver, version 23.0
[ 118.996] (II) LoadModule: "fbdev"
[ 118.996] (WW) Warning, couldn't open module fbdev
[ 118.996] (II) UnloadModule: "fbdev"
[ 118.996] (II) Unloading fbdev
[ 118.996] (EE) Failed to load module "fbdev" (module does not exist, 0)
[ 118.996] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[ 118.996] (II) modeset(0): using drv /dev/dri/card0
[ 118.997] (II) modeset(G0): using drv /dev/dri/card1
[ 118.997] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support
[ 118.997] (II) modeset(0): Creating default Display subsection in Screen section
"Default Screen Section" for depth/fbbpp 24/32
[ 118.997] (==) modeset(0): Depth 24, (==) framebuffer bpp 32
[ 118.997] (==) modeset(0): RGB weight 888
[ 118.997] (==) modeset(0): Default visual is TrueColor
[ 118.997] (II) Loading sub module "glamoregl"
[ 118.997] (II) LoadModule: "glamoregl"
[ 118.997] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[ 119.009] (II) Module glamoregl: vendor="X.Org Foundation"
[ 119.009] compiled for 1.19.6, module version = 1.0.0
[ 119.009] ABI class: X.Org ANSI C Emulation, version 0.4
[ 119.009] (II) glamor: OpenGL accelerated X.org driver based.
[ 119.046] (II) glamor: EGL version 1.4 (DRI2):
[ 119.055] (II) modeset(0): glamor initialized
[ 119.055] (II) modeset(0): Output DSI-1 has no monitor section
[ 119.055] (II) modeset(0): EDID for output DSI-1
[ 119.055] (II) modeset(0): Printing probed modes for output DSI-1
[ 119.055] (II) modeset(0): Modeline "2560x1800"x60.0 304.42 2560 2640 2720 2800 1800 1804 1808 1812 (108.7 kHz)
[ 119.055] (II) modeset(0): Output DSI-1 connected
[ 119.055] (II) modeset(0): Using sloppy heuristic for initial modes
[ 119.055] (II) modeset(0): Output DSI-1 using initial mode 2560x1800 +0+0
[ 119.055] (==) modeset(0): Using gamma correction (1.0, 1.0, 1.0)
[ 119.055] (==) modeset(0): DPI set to (96, 96)
[ 119.055] (II) Loading sub module "fb"
[ 119.055] (II) LoadModule: "fb"
[ 119.056] (II) Loading /usr/lib/xorg/modules/libfb.so
[ 119.063] (II) Module fb: vendor="X.Org Foundation"
[ 119.063] compiled for 1.19.6, module version = 1.0.0
[ 119.063] ABI class: X.Org ANSI C Emulation, version 0.4
[ 119.064] (==) modeset(G0): Depth 24, (==) framebuffer bpp 32
[ 119.064] (==) modeset(G0): RGB weight 888
[ 119.064] (==) modeset(G0): Default visual is TrueColor
[ 119.064] (II) Loading sub module "glamoregl"
[ 119.064] (II) LoadModule: "glamoregl"
[ 119.064] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[ 119.064] (II) Module glamoregl: vendor="X.Org Foundation"
[ 119.064] compiled for 1.19.6, module version = 1.0.0
[ 119.064] ABI class: X.Org ANSI C Emulation, version 0.4
[ 119.064] (II) glamor: OpenGL accelerated X.org driver based.
[ 119.084] (II) glamor: EGL version 1.4 (DRI2):
[ 119.086] (II) modeset(G0): glamor initialized
[ 119.086] (==) modeset(G0): Using gamma correction (1.0, 1.0, 1.0)
[ 119.086] (==) modeset(G0): DPI set to (96, 96)
[ 119.086] (II) Loading sub module "fb"
[ 119.086] (II) LoadModule: "fb"
[ 119.087] (II) Loading /usr/lib/xorg/modules/libfb.so
[ 119.087] (II) Module fb: vendor="X.Org Foundation"
[ 119.087] compiled for 1.19.6, module version = 1.0.0
[ 119.087] ABI class: X.Org ANSI C Emulation, version 0.4
[ 119.087] (==) Depth 24 pixmap format is 32 bpp
[ 119.109] (WW) glamor256: Failed to get GLSL version
[ 119.109] (EE) modeset(G0): Failed to initialize glamor at ScreenInit() time.
[ 119.109] (EE)
Fatal server error:
[ 119.109] (EE) AddScreen/ScreenInit failed for gpu driver 0 -1
[ 119.109] (EE)
[ 119.109] (EE)
Today's assembly with a fresh mesa. the login greeted. Graphics buggy. X did not start. https://travis-ci.org/vartom/tegra-nouveau-rootfs/builds/351320319 https://github.com/vartom/tegra-nouveau-rootfs/releases/tag/untagged-06c3cfac81bb8c29d3c0
Graphics buggy in the Gnome. the mouse pointer is lost. Tried it with kernels 4.15 and 4.16 (next). X did not start. Where there are these glitches I do not know.
What about xfce4? GNOME defaults to Wayland, have you tried with GNOME on Xorg?
xfce4 I did not try, previously it worked well. GNOME on Xorg tried - does not start, I write about it.
@vartom Kernel image? Should I rebuild it from next-20180309
?
Kernel 4.15.y, nouveau driver in kernel, not modul. I think the drivers for GPU do not work. I do not see anything new from the new version of Mesa.
Can you please try to build the linux-next-20180309 and the newest version of mesa?
As Thierry said, it should work. Probably what you're doing wrong now is that you're using linux-4.15.y which doesn't include Thierry's patches.
I can't test it right now, I'll look into it ASAP, but if you have some spare time you can try as I instructed
Probably what you're doing wrong now is that you're using linux-4.15.y which doesn't include Thierry's patches.
It does not matter, I try linux-next-20180309 If the return to the modular nouveau driver load does not help, it can be said that their patches do not work with the built-in dsi panel.
@denysvitali @thierryreding i try modular nouveau driver, linux-next-20180309 and the newest version of mesa. It does not matter. X did not start. The only difference is when you start kmscube not have to write the key --device=/dev/dri/card1 Maybe this is some sort of problem when assembling Xorg?
@vartom can you provide the output of dmesg and the exact kernel configuration that you're using? According to Xorg.0.log there seem to be changes on top of next-20180309, perhaps post those as well, just to make sure they aren't potentially breaking anything.
@thierryreding kernel configuration https://github.com/vartom/linux-smaug/blob/test/arch/arm64/configs/tegra21_defconfig I also tried with this configuration https://github.com/vartom/linux-smaug/blob/test/arch/arm64/configs/defconfig I can not drop the kernel log now, not at home.
@vartom your configuration looks okay. dmesg should provide some more clues.
@thierryreding dmesg-1520793876.txt
This doesn't show anything related to nouveau. Are you sure the device is enabled in device tree and the driver is built-in or loaded?
@vartom Can you include the nouveau module in the kernel?
Set this to y
@thierryreding dmesg-1520862223.txt nouveau in the kernel
This looks all very normal, no indication that anything went wrong.
Can you list the exact versions of kernel, libdrm, mesa and X that you're using, perhaps I can reproduce somehow. Also, how do you run X, what window manager/desktop environment do you use?
@thierryreding I'm pretty sure he is using this rootfs which is our custom fork of @Gnurou work.
Vartom's rootfs is available here
When the lightdm does not load, that is, I continue to see the log instead of the desktop. I go to VT through Ctrl + alt + f4. I enter as a root. I can run a kmscube but after exiting on Ctrl + c, I do not see the text VT. I make run a startx, it gives an error Fatal server error: [ 119.109] (EE) AddScreen/ScreenInit failed for gpu driver 0 -1
and then the text VT is returned. I can again start kmscube. The only thing I do not know is why the output from kmskube does not return text to VT.
@vartom can you set the following in your defconfig:
CONFIG_TEGRA_MC=y
Without that, CONFIG_TEGRA_IOMMU_SMMU will not be enabled and that in turn will prevent the file descriptor exported from Nouveau to be imported into Tegra DRM for display. This is because Tegra DRM needs the SMMU to map non-contiguous buffers.
The Tegra Gallium driver does a bad job of reporting failure to import a file descriptor. Well... not really any job at all, it basically ignores failure to import. I'm working on a patch for that. I'll also send out a patch to make sure we have TEGRA_MC and TEGRA_IOMMU_SMMU in the default configurations, which, for some reason, isn't currently the case.
I checked CONFIG_TEGRA_MC=y and TEGRA_IOMMU_SMMU=y in .config. as I wrote earlier on kernels 4.17 and next-20180309 I see such an error
And there are such mistakes Failed to set up IOMMU for device 57000000.gpu; retaining platform DMA ops
I tried to merge the next-20180309 into my 4.15.y kernel, and this is the result. Unfortunately the next-20180309 had some troubles with our patches - therefore the panel didn't show any image.
I hope I didn't messed up anything in the merge.
Here is the Xorg.log after the nouveau module was dynamically loaded:
[ 160.612]
X.Org X Server 1.19.6
Release Date: 2017-12-20
[ 160.612] X Protocol Version 11, Revision 0
[ 160.612] Build Operating System: Linux 4.14.13-1-ARCH armv7l
[ 160.612] Current Operating System: Linux alarm 4.16.0-rc4-next-20180309+ #126 SMP PREEMPT Mon Mar 12 21:03:21 UTC 2018 aarch64
[ 160.612] Kernel command line: cros_secure vpr=0x08000000@0xf6800000 fbcon video=tegrafb debug earlyprintk earlycon=uart8250,mmio32,0x70006000 drm.atomic=1 consoleblank=0 firmware_class.path=/lib/firmware
[ 160.613] Build Date: 31 January 2018 06:57:16AM
[ 160.613]
[ 160.613] Current version of pixman: 0.34.0
[ 160.613] Before reporting problems, check http://wiki.x.org
to make sure that you have the latest version.
[ 160.613] Markers: (--) probed, (**) from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[ 160.613] (==) Log file: "/var/log/Xorg.0.log", Time: Mon Mar 12 21:07:15 2018
[ 160.614] (==) Using system config directory "/usr/share/X11/xorg.conf.d"
[ 160.614] (==) No Layout section. Using the first Screen section.
[ 160.614] (==) No screen section available. Using defaults.
[ 160.614] (**) |-->Screen "Default Screen Section" (0)
[ 160.614] (**) | |-->Monitor "<default monitor>"
[ 160.615] (==) No monitor specified for screen "Default Screen Section".
Using a default monitor configuration.
[ 160.615] (==) Automatically adding devices
[ 160.615] (==) Automatically enabling devices
[ 160.615] (==) Automatically adding GPU devices
[ 160.615] (==) Automatically binding GPU devices
[ 160.615] (==) Max clients allowed: 256, resource mask: 0x1fffff
[ 160.616] (WW) The directory "/usr/share/fonts/Type1/" does not exist.
[ 160.616] Entry deleted from font path.
[ 160.616] (WW) `fonts.dir' not found (or not valid) in "/usr/share/fonts/100dpi/".
[ 160.616] Entry deleted from font path.
[ 160.616] (Run 'mkfontdir' on "/usr/share/fonts/100dpi/").
[ 160.616] (WW) `fonts.dir' not found (or not valid) in "/usr/share/fonts/75dpi/".
[ 160.616] Entry deleted from font path.
[ 160.616] (Run 'mkfontdir' on "/usr/share/fonts/75dpi/").
[ 160.616] (==) FontPath set to:
/usr/share/fonts/misc/,
/usr/share/fonts/TTF/,
/usr/share/fonts/OTF/
[ 160.616] (==) ModulePath set to "/usr/lib/xorg/modules"
[ 160.616] (II) The server relies on udev to provide the list of input devices.
If no devices become available, reconfigure udev or disable AutoAddDevices.
[ 160.616] (II) Loader magic: 0xd17f00
[ 160.616] (II) Module ABI versions:
[ 160.616] X.Org ANSI C Emulation: 0.4
[ 160.616] X.Org Video Driver: 23.0
[ 160.616] X.Org XInput driver : 24.1
[ 160.617] X.Org Server Extension : 10.0
[ 160.618] (++) using VT number 7
[ 160.619] (II) systemd-logind: logind integration requires -keeptty and -keeptty was not provided, disabling logind integration
[ 160.622] (II) xfree86: Adding drm device (/dev/dri/card0)
[ 160.624] (II) xfree86: Adding drm device (/dev/dri/card1)
[ 160.626] (II) no primary bus or device found
[ 160.626] falling back to /sys/devices/platform/50000000.host1x/drm/drm/card0
[ 160.626] (II) LoadModule: "glx"
[ 160.627] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
[ 160.631] (II) Module glx: vendor="X.Org Foundation"
[ 160.631] compiled for 1.19.6, module version = 1.0.0
[ 160.631] ABI class: X.Org Server Extension, version 10.0
[ 160.631] (==) Matched modesetting as autoconfigured driver 0
[ 160.632] (==) Matched fbdev as autoconfigured driver 1
[ 160.632] (==) Assigned the driver to the xf86ConfigLayout
[ 160.632] (II) LoadModule: "modesetting"
[ 160.632] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[ 160.633] (II) Module modesetting: vendor="X.Org Foundation"
[ 160.633] compiled for 1.19.6, module version = 1.19.6
[ 160.633] Module class: X.Org Video Driver
[ 160.633] ABI class: X.Org Video Driver, version 23.0
[ 160.633] (II) LoadModule: "fbdev"
[ 160.634] (WW) Warning, couldn't open module fbdev
[ 160.634] (II) UnloadModule: "fbdev"
[ 160.634] (II) Unloading fbdev
[ 160.634] (EE) Failed to load module "fbdev" (module does not exist, 0)
[ 160.634] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[ 160.640] (II) modeset(0): using drv /dev/dri/card0
[ 160.641] (II) modeset(G0): using drv /dev/dri/card1
[ 160.641] (II) modeset(0): Creating default Display subsection in Screen section
"Default Screen Section" for depth/fbbpp 24/32
[ 160.641] (==) modeset(0): Depth 24, (==) framebuffer bpp 32
[ 160.642] (==) modeset(0): RGB weight 888
[ 160.642] (==) modeset(0): Default visual is TrueColor
[ 160.642] (II) Loading sub module "glamoregl"
[ 160.642] (II) LoadModule: "glamoregl"
[ 160.642] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[ 160.646] (II) Module glamoregl: vendor="X.Org Foundation"
[ 160.646] compiled for 1.19.6, module version = 1.0.0
[ 160.646] ABI class: X.Org ANSI C Emulation, version 0.4
[ 160.646] (II) glamor: OpenGL accelerated X.org driver based.
[ 160.822] (II) glamor: EGL version 1.4 (DRI2):
[ 160.822] EGL_MESA_drm_image required.
[ 160.824] (EE) modeset(0): glamor initialization failed
[ 160.824] (II) modeset(0): ShadowFB: preferred NO, enabled NO
[ 160.824] (II) modeset(0): Output DSI-1 has no monitor section
[ 160.825] (II) modeset(0): EDID for output DSI-1
[ 160.825] (II) modeset(0): Printing probed modes for output DSI-1
[ 160.825] (II) modeset(0): Modeline "2560x1800"x60.0 304.42 2560 2640 2720 2800 1800 1804 1808 1812 (108.7 kHz)
[ 160.825] (II) modeset(0): Output DSI-1 connected
[ 160.825] (II) modeset(0): Using sloppy heuristic for initial modes
[ 160.825] (II) modeset(0): Output DSI-1 using initial mode 2560x1800 +0+0
[ 160.825] (==) modeset(0): Using gamma correction (1.0, 1.0, 1.0)
[ 160.825] (==) modeset(0): DPI set to (96, 96)
[ 160.825] (II) Loading sub module "fb"
[ 160.825] (II) LoadModule: "fb"
[ 160.826] (II) Loading /usr/lib/xorg/modules/libfb.so
[ 160.827] (II) Module fb: vendor="X.Org Foundation"
[ 160.827] compiled for 1.19.6, module version = 1.0.0
[ 160.827] ABI class: X.Org ANSI C Emulation, version 0.4
[ 160.829] (==) modeset(G0): Depth 24, (==) framebuffer bpp 32
[ 160.829] (==) modeset(G0): RGB weight 888
[ 160.829] (==) modeset(G0): Default visual is TrueColor
[ 160.829] (II) Loading sub module "glamoregl"
[ 160.829] (II) LoadModule: "glamoregl"
[ 160.830] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[ 160.830] (II) Module glamoregl: vendor="X.Org Foundation"
[ 160.830] compiled for 1.19.6, module version = 1.0.0
[ 160.830] ABI class: X.Org ANSI C Emulation, version 0.4
[ 160.830] (II) glamor: OpenGL accelerated X.org driver based.
[ 160.871] (II) glamor: EGL version 1.4 (DRI2):
[ 160.881] (II) modeset(G0): glamor initialized
[ 160.882] (==) modeset(G0): Using gamma correction (1.0, 1.0, 1.0)
[ 160.882] (==) modeset(G0): DPI set to (96, 96)
[ 160.882] (II) Loading sub module "fb"
[ 160.882] (II) LoadModule: "fb"
[ 160.882] (II) Loading /usr/lib/xorg/modules/libfb.so
[ 160.883] (II) Module fb: vendor="X.Org Foundation"
[ 160.883] compiled for 1.19.6, module version = 1.0.0
[ 160.883] ABI class: X.Org ANSI C Emulation, version 0.4
[ 160.883] (==) Depth 24 pixmap format is 32 bpp
[ 160.964] (==) modeset(G0): Backing store enabled
[ 160.964] (==) modeset(G0): Silken mouse enabled
[ 160.964] (II) modeset(G0): RandR 1.2 enabled, ignore the following RandR disabled message.
[ 160.965] (==) modeset(G0): DPMS enabled
[ 160.965] (II) modeset(G0): [DRI2] Setup complete
[ 160.965] (II) modeset(G0): [DRI2] DRI driver: nouveau
[ 160.965] (II) modeset(G0): [DRI2] VDPAU driver: nouveau
[ 161.001] (==) modeset(0): Backing store enabled
[ 161.001] (==) modeset(0): Silken mouse enabled
[ 161.001] (II) modeset(0): RandR 1.2 enabled, ignore the following RandR disabled message.
[ 161.006] (==) modeset(0): DPMS enabled
[ 161.007] (--) RandR disabled
[ 161.020] (II) AIGLX: Screen 0 is not DRI2 capable
[ 161.020] (EE) AIGLX: reverting to software rendering
[ 161.025] (II) IGLX: enabled GLX_MESA_copy_sub_buffer
[ 161.027] (II) IGLX: Loaded and initialized swrast
[ 161.027] (II) GLX: Initialized DRISWRAST GL provider for screen 0
[ 161.189] (II) modeset(G0): Damage tracking initialized
[ 161.191] (II) modeset(0): Damage tracking initialized
[ 161.191] (II) modeset(0): Setting screen physical size to 677 x 476
[ 161.412] (II) config/udev: Adding input device hid-over-i2c 06CB:3370 (/dev/input/event0)
[ 161.412] (**) hid-over-i2c 06CB:3370: Applying InputClass "libinput touchscreen catchall"
[ 161.412] (II) LoadModule: "libinput"
[ 161.413] (II) Loading /usr/lib/xorg/modules/input/libinput_drv.so
[ 161.419] (II) Module libinput: vendor="X.Org Foundation"
[ 161.419] compiled for 1.19.3, module version = 0.26.0
[ 161.419] Module class: X.Org XInput Driver
[ 161.419] ABI class: X.Org XInput driver, version 24.1
[ 161.421] (II) Using input driver 'libinput' for 'hid-over-i2c 06CB:3370'
[ 161.421] (**) hid-over-i2c 06CB:3370: always reports core events
[ 161.421] (**) Option "Device" "/dev/input/event0"
[ 161.421] (**) Option "_source" "server/udev"
[ 161.425] (II) event0 - hid-over-i2c 06CB:3370: is tagged by udev as: Touchscreen
[ 161.425] (II) event0 - hid-over-i2c 06CB:3370: device is a touch device
[ 161.425] (II) event0 - hid-over-i2c 06CB:3370: device removed
[ 161.464] (**) Option "config_info" "udev:/sys/devices/platform/7000c000.i2c/i2c-0/0-0020/0018:06CB:3370.0001/input/input0/event0"
[ 161.465] (II) XINPUT: Adding extended input device "hid-over-i2c 06CB:3370" (type: TOUCHSCREEN, id 6)
[ 161.465] (**) Option "AccelerationScheme" "none"
[ 161.465] (**) hid-over-i2c 06CB:3370: (accel) selected scheme none/0
[ 161.465] (**) hid-over-i2c 06CB:3370: (accel) acceleration factor: 2.000
[ 161.465] (**) hid-over-i2c 06CB:3370: (accel) acceleration threshold: 4
[ 161.468] (II) event0 - hid-over-i2c 06CB:3370: is tagged by udev as: Touchscreen
[ 161.468] (II) event0 - hid-over-i2c 06CB:3370: device is a touch device
[ 161.472] (II) config/udev: Adding input device hid-over-i2c 06CB:3370 (/dev/input/mouse0)
[ 161.472] (II) No input driver specified, ignoring this device.
[ 161.472] (II) This device may have been added with another device file.
[ 161.477] (II) config/udev: Adding input device hid-over-i2c 06CB:3370 System Control (/dev/input/event1)
[ 161.477] (**) hid-over-i2c 06CB:3370 System Control: Applying InputClass "libinput keyboard catchall"
[ 161.477] (II) Using input driver 'libinput' for 'hid-over-i2c 06CB:3370 System Control'
[ 161.477] (**) hid-over-i2c 06CB:3370 System Control: always reports core events
[ 161.477] (**) Option "Device" "/dev/input/event1"
[ 161.477] (**) Option "_source" "server/udev"
[ 161.480] (II) event1 - hid-over-i2c 06CB:3370 System Control: is tagged by udev as: Keyboard
[ 161.480] (II) event1 - hid-over-i2c 06CB:3370 System Control: device is a keyboard
[ 161.480] (II) event1 - hid-over-i2c 06CB:3370 System Control: device removed
[ 161.516] (**) Option "config_info" "udev:/sys/devices/platform/7000c000.i2c/i2c-0/0-0020/0018:06CB:3370.0001/input/input3/event1"
[ 161.517] (II) XINPUT: Adding extended input device "hid-over-i2c 06CB:3370 System Control" (type: KEYBOARD, id 7)
[ 161.521] (II) event1 - hid-over-i2c 06CB:3370 System Control: is tagged by udev as: Keyboard
[ 161.521] (II) event1 - hid-over-i2c 06CB:3370 System Control: device is a keyboard
[ 161.524] (II) config/udev: Adding input device gpio-keys (/dev/input/event2)
[ 161.524] (**) gpio-keys: Applying InputClass "libinput keyboard catchall"
[ 161.524] (II) Using input driver 'libinput' for 'gpio-keys'
[ 161.524] (**) gpio-keys: always reports core events
[ 161.524] (**) Option "Device" "/dev/input/event2"
[ 161.524] (**) Option "_source" "server/udev"
[ 161.527] (II) event2 - gpio-keys: is tagged by udev as: Keyboard Switch
[ 161.527] (II) event2 - gpio-keys: device is a keyboard
[ 161.527] (II) event2 - gpio-keys: device is a switch device
[ 161.527] (II) event2 - gpio-keys: device is a switch device
[ 161.527] (II) event2 - gpio-keys: device removed
[ 161.544] (**) Option "config_info" "udev:/sys/devices/platform/gpio-keys/input/input6/event2"
[ 161.544] (II) XINPUT: Adding extended input device "gpio-keys" (type: KEYBOARD, id 8)
[ 161.548] (II) event2 - gpio-keys: is tagged by udev as: Keyboard Switch
[ 161.548] (II) event2 - gpio-keys: device is a keyboard
[ 161.548] (II) event2 - gpio-keys: device is a switch device
[ 161.548] (II) event2 - gpio-keys: device is a switch device
[ 162.943] (II) modeset(0): Disabling kernel dirty updates, not required.
glxinfo-4.16.0-rc4-next-20180309+.log
dmesg-2018-03-12_21:19:16-4.16.0-rc4-next-20180309+.log
Boot image - boot-linux-4.17-next-5b1c93553e12_20180312_210403
Packages:
[root@alarm ~]# pacman -Q xorg-server mesa wayland libdrm
xorg-server 1.19.6+13+gd0d1a694f-1
mesa 17.3.6-1
wayland 1.14.0-1
libdrm 2.4.91-1.1
Tested w/ GDM and LightDM. LightDM (Xorg) is okay, but it isn't accelerated.
GDM is ultra-slow, even w/ 0d (998 MHz)
set in /sys/kernel/debug/dri/129/pstate
:
[root@alarm ~]# cat /sys/kernel/debug/dri/129/pstate
01: core 76 MHz
02: core 153 MHz
03: core 230 MHz
04: core 307 MHz
05: core 384 MHz
06: core 460 MHz
07: core 537 MHz
08: core 614 MHz
09: core 691 MHz
0a: core 768 MHz
0b: core 844 MHz
0c: core 921 MHz
0d: core 998 MHz AC DC *
DC: core 998 MHz
glxgers
reports low FPS when is maximized on Xorg (around 40 FPS, the rendering is obviously done on the CPU)
@denysvitali You have the old version of Mesa library in the system! I made a patch for you to fix this situation.
[root@alarm ~]# pacman -Q xorg-server mesa wayland libdrm xorg-server 1.19.6+13+gd0d1a694f-1 mesa 17.3.6-1
@thierryreding I think the problem can be in this patch https://github.com/vartom/linux-smaug/commit/2a8102dfe0da7dbb61794e6b85dc7ac9271e5fc8 in 4.15 it is not present and I see it in the kernel log
0.190265] tegra-mc 70019000.memory-controller: no memory timings for RAM code 0 registered
[ 0.190563] iommu: Adding device reg-dummy to group 0
[ 0.190585] iommu: Adding device f66cc000.ramoops to group 1
[ 0.190627] iommu: Adding device 50000000.host1x to group 2
[ 0.190648] iommu: Adding device 54100000.tsec to group 3
[ 0.190690] iommu: Adding device 54200000.dc to group 4
[ 0.190727] iommu: Adding device 54240000.dc to group 5
[ 0.190746] iommu: Adding device 54300000.dsi to group 6
[ 0.190780] iommu: Adding device 54340000.vic to group 7
[ 0.190803] iommu: Adding device 54400000.dsi to group 8
[ 0.190822] iommu: Adding device 545c0000.dpaux to group 9
[ 0.190859] iommu: Adding device 57000000.gpu to group 10
[ 0.190879] iommu: Adding device 60005000.timer to group 11
[ 0.190898] iommu: Adding device 60007000.flow-controller to group 12
[ 0.190918] iommu: Adding device 6000d000.gpio to group 13
[ 0.190940] iommu: Adding device 60020000.dma to group 14
[ 0.190958] iommu: Adding device 70000800.apbmisc to group 15
[ 0.190976] iommu: Adding device 700008d4.pinmux to group 16
[ 0.190993] iommu: Adding device 70006000.serial to group 17
[ 0.191015] iommu: Adding device 70006200.serial to group 18
[ 0.191033] iommu: Adding device 70006300.serial to group 19
[ 0.191053] iommu: Adding device 7000c000.i2c to group 20
[ 0.191071] iommu: Adding device 7000c400.i2c to group 21
[ 0.191089] iommu: Adding device 7000c500.i2c to group 22
[ 0.191107] iommu: Adding device 7000d000.i2c to group 23
[ 0.191127] iommu: Adding device 7000d100.i2c to group 24
[ 0.191146] iommu: Adding device 7000e000.rtc to group 25
[ 0.191165] iommu: Adding device 7000e400.pmc to group 26
[ 0.191183] iommu: Adding device 7000f800.fuse to group 27
[ 0.191201] iommu: Adding device 70019000.memory-controller to group 28
[ 0.191422] iommu: Adding device 70090000.usb to group 29
[ 0.191501] iommu: Adding device 7009f000.padctl to group 30
[ 0.191607] iommu: Adding device 700b0600.sdhci to group 31
[ 0.191719] iommu: Adding device 700b0200.sdhci to group 32
[ 0.191797] iommu: Adding device 700e3000.mipi to group 33
[ 0.191862] iommu: Adding device aconnect@702c0000 to group 34
[ 0.191986] iommu: Adding device timer to group 35
[ 0.192098] iommu: Adding device 700e2000.thermal-sensor to group 36
[ 0.192162] iommu: Adding device clocks to group 37
[ 0.192226] iommu: Adding device gpio-keys to group 38
[ 0.192311] iommu: Adding device max98357a to group 39
[ 0.192377] iommu: Adding device sound to group 40
[ 0.192446] iommu: Adding device psci to group 41
[ 0.192503] iommu: Adding device regulators to group 42
[ 0.192572] iommu: Adding device regulators:regulator@0 to group 43
[ 0.192644] iommu: Adding device regulators:regulator@1 to group 44
[ 0.192710] iommu: Adding device regulators:regulator@2 to group 45
[ 0.192777] iommu: Adding device regulators:regulator@3 to group 46
[ 0.192842] iommu: Adding device regulators:regulator@4 to group 47
[ 0.192918] iommu: Adding device regulators:regulator@5 to group 48
[ 0.192989] iommu: Adding device regulators:regulator@6 to group 49
[ 0.193053] iommu: Adding device regulators:regulator@7 to group 50
[ 0.198049] of_get_named_gpiod_flags: can't parse 'gpio' property of node '/regulators/regulator@0[0]'
. . .
in 4.16 / 4.17 it is, and I see it in the kernel log
[ 0.201139] tegra-mc 70019000.memory-controller: no memory timings for RAM code 0 registered
[ 0.201460] iommu: Adding device 50000000.host1x to group 0
[ 0.201493] iommu: Adding device 54200000.dc to group 1
[ 0.201521] iommu: Adding device 54240000.dc to group 1
[ 0.201551] iommu: Adding device 54340000.vic to group 2
[ 0.201584] iommu: Adding device 57000000.gpu to group 3
[ 0.207722] cryptd: max_cpu_qlen set to 1000
. . .
[ 0.609969] Failed to set up IOMMU for device 50000000.host1x; retaining platform DMA ops
[ 0.610920] Failed to set up IOMMU for device 54200000.dc; retaining platform DMA ops
[ 0.619128] Failed to set up IOMMU for device 54240000.dc; retaining platform DMA ops
[ 0.627469] of_get_named_gpiod_flags: can't parse 'nvidia,hpd-gpio' property of node '/host1x@50000000/dsi@54300000[0]'
[ 0.627503] tegra-dsi 54300000.dsi: Looking up avdd-dsi-csi-supply from device tree
[ 0.627521] tegra-dsi 54300000.dsi: cannot get VDD supply
[ 0.627713] tegra-dpaux 545c0000.dpaux: Looking up vdd-supply from device tree
[ 0.630309] Failed to set up IOMMU for device 54340000.vic; retaining platform DMA ops
@vartom Thanks, I've merged the PR and I'm now rebuilding the rootfs. I'll try again once it's ready 🚀
@vartom That should be unrelated. There are corresponding patches in Tegra DRM to make use of the IOMMU groups, but 4.15 should work identically. Effectively the result in programming of the SMMU and behaviour in Tegra DRM and Nouveau should be unchanged.
Meh. I'm having some troubles compiling mesa (from git) for arm64
. It fails at:
In file included from /run/media/dvitali/Shared/tegra-rootfs/out/target/aarch64/ArchLinuxArm/usr/include/libunwind.h:7:0,
from /run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.h:35,
from /run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:37:
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c: In function ‘debug_backtrace_capture’:
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:114:4: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘asm’
unw_getcontext(&context);
^
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:114:4: error: lvalue required as left operand of assignment
unw_getcontext(&context);
^
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:114:4: error: ‘unw_base’ undeclared (first use in this function); did you mean ‘unw_ctx’?
unw_getcontext(&context);
^
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:114:4: note: each undeclared identifier is reported only once for each function it appears in
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c: At top level:
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:288:0: error: unterminated #else
#if defined(HAVE_LIBUNWIND)
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:199:0: error: unterminated #else
#if defined(HAVE_LIBUNWIND)
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:39:0: error: unterminated #else
#if defined(HAVE_LIBUNWIND)
In file included from /run/media/dvitali/Shared/tegra-rootfs/out/target/aarch64/ArchLinuxArm/usr/include/libunwind.h:7:0,
from /run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.h:35,
from /run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:37:
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c: In function ‘debug_backtrace_capture’:
/run/media/dvitali/Shared/tegra-rootfs/mesa/src/gallium/auxiliary/util/u_debug_stack.c:114:4: error: invalid lvalue in asm output 0
unw_getcontext(&context);
This seems to be related to this
@vartom how did you build mesa for arm64? I'm using tegra-nouveau-rootfs with the following:
./scripts/download-gcc && ./scripts/download-rootfs && ./scripts/prepare-rootfs && ./scripts/build-drm && ./scripts/build-mesa && ./scripts/build-wayland && ./scripts/generate-rootfs
and it's failing at ./scripts/build-mesa
@denysvitali --disable-libunwind in build-mesa.
@vartom I figured it out after a couple tries https://github.com/denysvitali/tegra-rootfs-scripts/commit/0b4c661fd6661700e082ff1ffada2fc152346090 :+1:
Thank you!
@denysvitali I'm not sure that the installation of a new version of the Mesa libraries is successful. I still see(pacman -Q mesa) the old version of the Mesa libraries.
@vartom That's because the mesa package gets installed in the prepare-ArchLinuxArm
via the mesa-demos
package.
But if you run the prepare-rootfs first, and then build the mesa package, you'll end up having the compiled mesa installed, but still have the package in pacman -Q mesa
with the old version.
@denysvitali The installation of a new version of the mez libraries goes to the opt / nouveau directory. Further with this, it will be prescribed that this should be used. Now it seems to me that this does not work or does not work correctly. for example, a kmcube can call a new version and the graphical desktop(lightdm, Xorg) does not.
@denysvitali Yes, I am sure. https://github.com/vartom/tegra-nouveau-rootfs#compiling-mesa
@vartom That's just the README. The scripts aren't building mesa in /opt/nouveau...
Try adding:
echo "Running make for $PACKAGE: sysroot is $SYSROOT"
on run_make()
in env.sh
. Then build mesa.
Running make for mesa: sysroot is /run/media/dvitali/Shared/tegra-rootfs/out/target/aarch64/ArchLinuxArm
@denysvitali Pay attention to this NV_PREFIX in env.sh run_autogen in env.sh run_autogen in build-mesa, build-drm, build-kmscube, .. And look at the presence of the directory /opt/nouveau in the rootfs.
Currently Linux on Pixel C is missing GPU Acceleration, therefore the UI is laggy dmesg is available here