Posts by SusisStrolch

SusisStrolch

Exactly - master doesn't build.

But at least 8.2 build works when truncating the path...

SusisStrolch

The reason seems to be the length of the PATH environment variable of my build host ().

Default, the path is ~140 chars in length:

Code

~/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/lib/jvm/jdk/bin:/usr/lib/jvm/jdk/db/bin:/usr/lib/jvm/jdk/jre/bin

When shortening the PATH to

Code

/home/strolch/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/lib/jvm/jdk/jre/bin

the error doesn't show up.

SusisStrolch

(clean) build fails when trying to cross compile for i.mx6 architecture.

Code

Buildoptions:
 ======================================================
 - CPU (ARCH):                           cortex-a9 (arm)
 - FLOAT:                                hard
 - FPU:                                  neon
 - SIMD support:                         yes
 - LTO (Link Time Optimization) support: yes
 - GOLD (Google Linker) Support:         yes
 - LLVM support:

 Graphic configuration:
 ======================================================
 - OpenGL (GLX) support (provider):      no (no)
 - OpenGLES support (provider):          yes (imx-gpu-viv)
 - uvesafb support:                      no

 Hardware decoder configuration:
 ======================================================
 - Kodi Player driver:                   libfslvpuwrap
 - VAAPI Support:
 - VDPAU Support:
...
...
[ 87%] Building CXX object build/cores/VideoPlayer/videorenderers/CMakeFiles/videorenderers.dir/FrameBufferObject.cpp.o
[ 87%] Building ASM object build/cores/VideoPlayer/videorenderers/CMakeFiles/videorenderers.dir/yuv2rgb.neon.S.o
In file included from /home/strolch/LE/build.LibreELEC-imx6.arm-9.0-devel/kodi-a8fe8e3/xbmc/cores/VideoPlayer/VideoRenderers/RenderCapture.h:117:0,
                 from /home/strolch/LE/build.LibreELEC-imx6.arm-9.0-devel/kodi-a8fe8e3/xbmc/cores/VideoPlayer/VideoRenderers/RenderCapture.cpp:21:
/home/strolch/LE/build.LibreELEC-imx6.arm-9.0-devel/kodi-a8fe8e3/xbmc/cores/VideoPlayer/../VideoPlayer/DVDCodecs/Video/DVDVideoCodecIMX.h: In function 'double recalcPts(double)':
/home/strolch/LE/build.LibreELEC-imx6.arm-9.0-devel/kodi-a8fe8e3/xbmc/cores/VideoPlayer/../VideoPlayer/DVDCodecs/Video/DVDVideoCodecIMX.h:70:26: error: 'DVD_NOPTS_VALUE' was not declared in this scope
   return (double)(pts == DVD_NOPTS_VALUE ? 0.0 : pts*1e-6);
                          ^~~~~~~~~~~~~~~
In file included from /home/strolch/LE/build.LibreELEC-imx6.arm-9.0-devel/kodi-a8fe8e3/xbmc/cores/VideoPlayer/VideoRenderers/RenderCapture.h:117:0,
                 from /home/strolch/LE/build.LibreELEC-imx6.arm-9.0-devel/kodi-a8fe8e3/xbmc/cores/VideoPlayer/VideoRenderers/RenderCapture.cpp:21:
/home/strolch/LE/build.LibreELEC-imx6.arm-9.0-devel/kodi-a8fe8e3/xbmc/cores/VideoPlayer/../VideoPlayer/DVDCodecs/Video/DVDVideoCodecIMX.h: At global scope:
/home/strolch/LE/build.LibreELEC-imx6.arm-9.0-devel/kodi-a8fe8e3/xbmc/cores/VideoPlayer/../VideoPlayer/DVDCodecs/Video/DVDVideoCodecIMX.h:302:36: error: 'DVDVideoPicture' has not been declared
   bool                  GetPicture(DVDVideoPicture *pDvdVideoPicture);
                                    ^~~~~~~~~~~~~~~

Display More

the log from failing build: - · GitHub

SusisStrolch

Same setting - so the driver doesn't set the type.

Anyway - the problem is DVI related. After connecting directly via HDMI the events don't show up anymore.

SusisStrolch

Ok, last one was a guess. udevadm shows a bit more:

Code

KERNEL[8386.314091] change   /devices/soc0/soc.0/20e0000.hdmi_video (platform)
ACTION=change
DEVPATH=/devices/soc0/soc.0/20e0000.hdmi_video
DRIVER=mxc_hdmi
EVENT=plugin
MODALIAS=of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video
OF_COMPATIBLE_0=fsl,imx6q-hdmi-video
OF_COMPATIBLE_N=1
OF_FULLNAME=/soc/hdmi_video@020e0000
OF_NAME=hdmi_video
SEQNUM=9973
SUBSYSTEM=platform

UDEV  [8386.325191] change   /devices/soc0/soc.0/20e0000.hdmi_video (platform)
ACTION=change
DEVPATH=/devices/soc0/soc.0/20e0000.hdmi_video
DRIVER=mxc_hdmi
EVENT=plugin
MODALIAS=of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video
OF_COMPATIBLE_0=fsl,imx6q-hdmi-video
OF_COMPATIBLE_N=1
OF_FULLNAME=/soc/hdmi_video@020e0000
OF_NAME=hdmi_video
SEQNUM=9973
SUBSYSTEM=platform
USEC_INITIALIZED=8386323200

Display More

So it seems to be a problem with the "plugged" state...

SusisStrolch

What does your /sys/devices/soc0/soc.0/20e0000.hdmi_video/uevent look like?

Mine contains

Code

hummingboard:~ # cat /sys/devices/soc0/soc.0/20e0000.hdmi_video/cable_state 
plugin DVI
hummingboard:~ # cat /sys/devices/soc0/soc.0/20e0000.hdmi_video/uevent 
DRIVER=mxc_hdmi
OF_NAME=hdmi_video
OF_FULLNAME=/soc/hdmi_video@020e0000
OF_COMPATIBLE_0=fsl,imx6q-hdmi-video
OF_COMPATIBLE_N=1
MODALIAS=of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video

So I suspect there's something missing when creating the udev event and the monitor is connected via DVI - because the T-field seems empty...

Seems the device type isn't set with DVI connections (excerpt from: drivers/of/device.c)

Code

/* Name & Type */
    csize = snprintf(str, len, "of:N%sT%s", dev->of_node->name,
             dev->of_node->type);

SusisStrolch

All kernels (starting from 8.0.0) show the same version - 3.14.79.

I'll investigate a bit myself to see if it only happens with DVI connection monitor. I have to admit I also changed the monitor port in the past, but didn't notice this behaviour.

In fact, now I only see

Code

Aug 11 14:27:48 hummingboard kernel: mxc_hdmi 20e0000.hdmi_video: mxc_hdmi_cable
_connected reports DVI mode
Aug 11 14:27:48 hummingboard kernel: mxc_hdmi 20e0000.hdmi_video: Same EDID
Aug 11 14:27:47 hummingboard kernel: mxc_hdmi 20e0000.hdmi_video: mxc_hdmi_cable
_connected reports DVI mode
Aug 11 14:27:47 hummingboard kernel: mxc_hdmi 20e0000.hdmi_video: Same EDID
Aug 11 14:27:46 hummingboard kernel: mxc_hdmi 20e0000.hdmi_video: mxc_hdmi_cable
_connected reports DVI mode
Aug 11 14:27:46 hummingboard kernel: mxc_hdmi 20e0000.hdmi_video: Same EDID

repeating each second.

SusisStrolch

After some iterations over SD-card and different versions I finally found the reason why it shows up.

I've changed my "uEnv.txt" in the past and added "debug progress splash=off" as additional command line parameter.

After adding the debug option it even shows up in a clean, new installation on a virgin SD-card.

But anyway - there's something wrong - either on my side (DVI connector) or with the kernel.

The irritating part is the 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video' thrown in the log file...

SusisStrolch

No, 8.01 doesn't show this behaviour

SusisStrolch

Here we go: ZgED

Unfortunately the journal size isn't large enough to record the whole boot process.

Any idea how to change the journal buffer size in live system w/o recompiling?

SusisStrolch

Just updated my hummingboard (with VDR for TV-recording) to vpeters LibreELEC-imx6.arm-8.0.2-3.14-sr image.

Starting with the 8.0.2 the systemd journal is flooded each second:

Code

Aug 10 08:23:21 hummingboard systemd-udevd[250]: seq 2155 queued, 'change' 'platform'
Aug 10 08:23:21 hummingboard kernel: mxc_hdmi 20e0000.hdmi_video: Same EDID
Aug 10 08:23:21 hummingboard kernel: mxc_hdmi 20e0000.hdmi_video: mxc_hdmi_cable_connected reports DVI mode
Aug 10 08:23:21 hummingboard systemd-udevd[250]: Validate module index
Aug 10 08:23:21 hummingboard systemd-udevd[250]: Check if link configuration needs reloading.
Aug 10 08:23:21 hummingboard systemd-udevd[250]: seq 2155 forked new worker [1445]
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: seq 2155 running
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: IMPORT builtin 'hwdb' /usr/lib/udev/rules.d/50-udev-default.rules:15
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: IMPORT builtin 'hwdb' returned non-zero
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: RUN 'kmod load $env{MODALIAS}' /usr/lib/udev/rules.d/80-drivers.rules:5
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: Execute 'load' 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: No module matches 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: passed device to netlink monitor 0xb876b478
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: seq 2155 processed
Aug 10 08:23:21 hummingboard systemd-udevd[250]: cleanup idle workers
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: Unload module index
Aug 10 08:23:21 hummingboard systemd-udevd[1445]: Unloaded link configuration context.
Aug 10 08:23:21 hummingboard systemd-udevd[250]: worker [1445] exited

hummingboard:~/.config # journalctl -b|grep 'No module matches'
Aug 10 10:41:24 hummingboard systemd-udevd[10107]: No module matches 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'
Aug 10 10:41:25 hummingboard systemd-udevd[10109]: No module matches 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'
Aug 10 10:41:26 hummingboard systemd-udevd[10110]: No module matches 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'
Aug 10 10:41:27 hummingboard systemd-udevd[10111]: No module matches 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'
Aug 10 10:41:28 hummingboard systemd-udevd[10112]: No module matches 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'
Aug 10 10:41:29 hummingboard systemd-udevd[10113]: No module matches 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'
Aug 10 10:41:30 hummingboard systemd-udevd[10114]: No module matches 'of:Nhdmi_videoT<NULL>Cfsl,imx6q-hdmi-video'

Display More

Because my SAT receiver (DVBSky S960 DVB-S2) doesn't work anymore with newer kernels I'm bound to the 3.14-sr.

Here's a part of the boot log:

Code

U-Boot SPL 2013.10-rc4 (Nov 17 2014 - 09:28:07)
Boot Device: SD1


U-Boot 2013.10-rc4 (Dec 31 2016 - 15:01:43)

CPU:   Freescale i.MX6Q rev1.5 at 792 MHz
Reset cause: POR
Board: MX6-HummingBoard
DRAM:  1 GiB
MMC:   FSL_SDHC: 0
*** Warning - bad CRC, using default environment

In:    serial
Out:   serial
Err:   serial
Net:   FEC [PRIME]
(Re)start USB...
USB0:   USB EHCI 1.00
scanning bus 0 for devices... Interface 0 has too many endpoints!
3 USB Device(s) found
       scanning usb for storage devices... 0 Storage Device(s) found
       scanning usb for ethernet devices... 0 Ethernet Device(s) found
Hit any key to stop autoboot:  0 
mmc0 is current device
** Unable to read file /boot.scr **
703 bytes read in 12 ms (56.6 KiB/s)
Importing environment from mmc0 ...
6567352 bytes read in 306 ms (20.5 MiB/s)
Booting from mmc ...
38636 bytes read in 18 ms (2 MiB/s)
Loaded imx6q-hummingboard.dtb
Booting KERNEL
Kernel image @ 0x10800000 [ 0x000000 - 0x6435b8 ]
## Flattened Device Tree blob at 18000000
   Booting using the fdt blob at 0x18000000
   Using Device Tree in place at 18000000, end 1800c6eb

Starting kernel ...

[    0.000000] Booting Linux on physical CPU 0x0
[    0.000000] Linux version 3.14.79 (vpeter@pc) (gcc version 6.2.0 (GCC) ) #1 SMP Fri May 26 11:32:03 CEST 2017
[    0.000000] CPU: ARMv7 Processor [412fc09a] revision 10 (ARMv7), cr=10c53c7d
[    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
[    0.000000] Machine model: SolidRun HummingBoard Dual/Quad
[    0.000000] cma: CMA: reserved 320 MiB at 2a000000
[    0.000000] Memory policy: Data cache writealloc
[    0.000000] On node 0 totalpages: 262144
[    0.000000] free_area_init_node: node 0, pgdat c0aea7c0, node_mem_map eef57000
[    0.000000]   DMA zone: 1520 pages used for memmap
[    0.000000]   DMA zone: 0 pages reserved
[    0.000000]   DMA zone: 194560 pages, LIFO batch:31
[    0.000000]   HighMem zone: 528 pages used for memmap
[    0.000000]   HighMem zone: 67584 pages, LIFO batch:15
[    0.000000] PERCPU: Embedded 7 pages/cpu @eef33000 s6848 r8192 d13632 u32768
[    0.000000] pcpu-alloc: s6848 r8192 d13632 u32768 alloc=8*4096
[    0.000000] pcpu-alloc: [0] 0 [0] 1 
[    0.000000] Built 1 zonelists in Zone order, mobility grouping on.  Total pages: 260624
[    0.000000] Kernel command line: root=/dev/ram0 rdinit=/init noram usbcore.autosuspend=-1 coherent_pool=2M console=ttymxc0,115200 kgdboc=ttymxc0,115200 video=mxcfb0:dev=hdmi,1920x1080M@50,if=RGB24,bpp=32 netboot=nfs ip=192.168.254.26:192.168.254.20:192.168.254.1:255.255.255.0:hummingboard boot=/dev/mmcblk0p1 disk=NFS=192.168.254.20:/volume1/nfsboot/hummingboard/storage,rsize=131072,wsize=131072,timeo=15 debug progress splash=off

Display More

SusisStrolch · Build problems (and solutions) SusisStrolch December 16, 2016 at 12:16 PM

By using GCC 6.2 to build master (also 7.90.x and 7.02) I stumbled because of some non C++11 conform code fragments in Kodi.
I fixed them by the following patch:

Diff

diff --git a/config/optimize b/config/optimize
index 2c395a5..8f5b6cf 100644
--- a/config/optimize
+++ b/config/optimize
@@ -38,6 +38,10 @@ HOST_LIBDIR="$ROOT/$TOOLCHAIN/lib"
 HOST_CFLAGS="$HOST_CFLAGS -Wno-format-security"
 HOST_CXXFLAGS="$HOST_CXXFLAGS -Wno-format-security"


+# work around C++11 warning
+TARGET_CXXFLAGS="$TARGET_CXXFLAGS -Wno-literal-suffix"
+HOST_CXXFLAGS="$HOST_CXXFLAGS -Wno-literal-suffix"
+
 # add distro specific library dirs
   # ubuntu/debian specific "multiarch support"
     FAMILY_TRIPLET=$(echo $HOST_NAME | sed -e "s,$(uname -m),$(uname -i),")

Display More

Another very interesting problem showed up during further make (cmake):

Code

make[3]: Entering directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
[  1%] Built target compileinfo
[  1%] Built target wrapper
make[3]: Entering directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
[  1%] Built target crossguid
make[3]: Leaving directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
/bin/sh: 1: PATH=/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/toolchain/bin:/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/toolchain/bin:/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/toolchain/sbin:~/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/usr/lib/jvm/java-8-oracle/bin:/usr/lib/jvm/java-8-oracle/db/bin:/usr/lib/jvm/java-8-oracle/jre/bin:/usr/lib/jvm/java-8-oracle/bin:/usr/lib/jvm/java-8-oracle/db/bin:/usr/lib/jvm/java-8-oracle/jre/bin: not found
make[3]: *** [CMakeFiles/libcpluff.dir/build.make:121: build/cpluff/src/libcpluff-stamp/libcpluff-autoreconf] Error 127
make[3]: Leaving directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
make[3]: Entering directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
make[3]: Entering directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
make[2]: *** [CMakeFiles/Makefile2:214: CMakeFiles/libcpluff.dir/all] Error 2
make[2]: *** Waiting for unfinished jobs....
make[3]: Leaving directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
make[3]: Leaving directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
make[3]: Leaving directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
make[3]: Leaving directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'
make[3]: Leaving directory '/home/strolch/Development/LibreELEC.tv/build.LibreELEC-imx6.arm-8.0-devel/kodi-17.0-beta6-654f1b7/.armv7a-libreelec-linux-gnueabi'

Display More

I finally found out that my PATH was too large - it looks like there's some limitation around 256 bytes in cmake(??) during invokation of external commands.
After removing the multipe defined JAVA path fragments and therefore shrinking the PATH size the build worked without further problems.

Posts by SusisStrolch

[solved] master (fb748cf) - imx6 - build failure when compiling kodi because of PATH length

[solved] master (fb748cf) - imx6 - build failure when compiling kodi because of PATH length

[solved] master (fb748cf) - imx6 - build failure when compiling kodi because of PATH length

SolidRun hummingboard - since 8.0.2-3.14-sr journal flooded with "of:Nhdmi_video"

SolidRun hummingboard - since 8.0.2-3.14-sr journal flooded with "of:Nhdmi_video"

SolidRun hummingboard - since 8.0.2-3.14-sr journal flooded with "of:Nhdmi_video"

SolidRun hummingboard - since 8.0.2-3.14-sr journal flooded with "of:Nhdmi_video"

SolidRun hummingboard - since 8.0.2-3.14-sr journal flooded with "of:Nhdmi_video"

SolidRun hummingboard - since 8.0.2-3.14-sr journal flooded with "of:Nhdmi_video"

SolidRun hummingboard - since 8.0.2-3.14-sr journal flooded with "of:Nhdmi_video"

SolidRun hummingboard - since 8.0.2-3.14-sr journal flooded with "of:Nhdmi_video"

Build problems (and solutions)