101 replies to this topic

[Erik Slagter]

If it helps,
The attached file is from ATV image, OE-Alliance build using branch 4.4.

ATSC tuners seem to work correctly in this build.

 

I don't think ATV already uses udev in their release images, so that would explain it.

 

Can you check if the tuner needs some firmware to be loaded on startup? That may also be the difference.

[el bandido]

SUMMARY = "USB ATSC driver for Hauppauge 955Q WinTV-HVR Tuners"
inherit allarch

require conf/license/license-gplv2.inc

RRECOMMENDS_${PN} = " \
    kernel-module-cx231xx \
    kernel-module-cx231xx-dvb \
    kernel-module-cx25840 \
    kernel-module-lgdt3306a \
    kernel-module-si2157 \
    kernel-module-tveeprom \
    firmware-cx231xx \
    "

PV = "1.0"

ALLOW_EMPTY_${PN} = "1"


What is listed above seems all that is needed to run the 955 tuner. I do not know of
any additional firmware that needs to be loaded on startup.

As a test,
Using the same Develop image recipes that were used for the Edision, I built a vuzero4k image. This image does not have any issues when installed in the Zero4k receiver, and the Hauppauge 955 tuner is attached. Does this test suggest udev is not the problem?

[rantanplan]

next attempt to solve the problem ;-)

Try to install the firmware locally.
restart and then?

 

https://www.linuxtv...._WinTV-HVR-955Q

https://github.com/J...xx-avcore-01.fw

[el bandido]

I do not understand what to try??? Install the firmware locally???

v4l-cx231xx-avcore-01.fw Is in the /lib/firmware folder of the receiver as mentioned earlier. This is the exact same version that is in the links.

[rantanplan]

ok, then i missed it, thought it was not there.

Sorry

[Erik Slagter]

If there is a problem with missing kernel modules, then it's not only the actual building of them (which the bitbake recipe does), but also the question whether the packages containing the modules are installed (either in the image, or from the feed) and THEN also whether they are actually loaded into the kernel.

 

Please check this. If you need help, yell.

[rantanplan]

https://github.com/O...ers-meta.inc#L3

 

new trial...

try to change this line

inherit allarch

 

[el bandido]

Had to ADD that line:

require conf/license/openpli-gplv2.inc
inherit allarch

PACKAGE_ARCH := "${MACHINE_ARCH}"

PACKAGES = "${PN}"
ALLOW_EMPTY_${PN} = "1"

# We only need the packaging tasks - disable the rest
do_fetch[noexec] = "1"
do_unpack[noexec] = "1"
do_patch[noexec] = "1"
do_configure[noexec] = "1"
do_compile[noexec] = "1"
do_install[noexec] = "1"
 

 

The result is the same when compiled this way. The Hauppauge tuner is assigned to DVB 0, then there is no video file for DVB 0 which causes a crash.

 

In the rs232 debug, first this:
cx231xx 1-1.3:1.1: DVB: registering adapter 0 frontend 0 (LG Electronics LGDT3306A VSB/QAM Frontend)...

Then a bit further down in the log this:
/dev/dvb/adapter0/video0: No such file or directory
 

[rantanplan]

You have to remove the line I marked and replace it with inherit allarch.
Next comes the construction without any local.extra rules

require conf/license/openpli-gplv2.inc

inherit allarch

PACKAGES = "${PN}"
ALLOW_EMPTY_${PN} = "1"

# We only need the packaging tasks - disable the rest
do_fetch[noexec] = "1"
do_unpack[noexec] = "1"
do_patch[noexec] = "1"
do_configure[noexec] = "1"
do_compile[noexec] = "1"
do_install[noexec] = "1"

 

[el bandido]

This change did not make a difference:

require conf/license/openpli-gplv2.inc


inherit allarch

PACKAGES = "${PN}"
ALLOW_EMPTY_${PN} = "1"

# We only need the packaging tasks - disable the rest
do_fetch[noexec] = "1"
do_unpack[noexec] = "1"
do_patch[noexec] = "1"
do_configure[noexec] = "1"
do_compile[noexec] = "1"
do_install[noexec] = "1"

[el bandido]

Go back to mdev and it will run correctly with the Hauppauge atsc tuner loading.
I had a copy of the busybox folder in recipes-core from the old Zeus branch (before it switched to develop).
I deleted the busybox folder and replaced it with the old Zeus busybox folder. Then I changed one line in openpli.conf file to this:
VIRTUAL-RUNTIME_dev_manager = "busybox-mdev"

The Hauppauge usb tuners that I have tested will load correctly and is usable with the above changes.  So rolling back to mdev works...

[rantanplan]

Yes...

https://stackoverflo...instead-of-udev

I also think that busybox-mdev has to be reactivated to manage the virtual devices.
So it seems in the OE Alliance Branch 4.4.

[WanWizard]

If a fixed device table works, those devices can also be created using udev.

[athoik]

Strange that it tries /dev/dvb/adapter0/video0.

It happens with other usb sticks as well?

Edit. Now I saw the mdev posts above. Ok another issue with udev

Edited by athoik, 29 September 2020 - 17:58.

[Erik Slagter]

It's not really an udev issue, it's more something like a workaround made for mdev that doesn't work in udev because udev works differently. It's not udev's fault 

 

Please tell me what device nodes are missing (or even better, if possible, where they need to have an symlink) and I'll add them.

Edited by Erik Slagter, 2 October 2020 - 09:13.

[el bandido]

The  localextra.rules in post#16 of this thread seem to work better. I get options for more tuners when this file is installed in the build. (file attached)
 

Thanks rantanplan for finding this issue.

Attached Files

•  localextra.zip   808bytes   5 downloads

Edited by el bandido, 2 October 2020 - 14:47.

[Erik Slagter]

# There are a number of modifiers that are allowed to be used in some
# of the different fields. They provide the following subsitutions:
#
# %n the "kernel number" of the device.
#    For example, 'sda3' has a "kernel number" of '3'
# %e the smallest number for that name which does not matches an existing node
# %k the kernel name for the device
# %M the kernel major number for the device
# %m the kernel minor number for the device
# %b the bus id for the device
# %c the string returned by the PROGRAM
# %s{filename} the content of a sysfs attribute
# %% the '%' char itself

# The first rtc device is symlinked to /dev/rtc
KERNEL=="rtc0", SYMLINK+="rtc"

# Legacy paths for dbox devices
KERNEL=="oled0", SYMLINK+="dbox/oled0"
KERNEL=="fp0", SYMLINK+="dbox/fp0"
KERNEL=="lcd0", SYMLINK+="dbox/lcd0"
KERNEL=="vtuner0", SYMLINK+="misc/vtuner0"
KERNEL=="vtuner1", SYMLINK+="misc/vtuner1"
KERNEL=="vtuner2", SYMLINK+="misc/vtuner2"
KERNEL=="vtuner3", SYMLINK+="misc/vtuner3"
KERNEL=="vtuner4", SYMLINK+="misc/vtuner4"
KERNEL=="vtuner5", SYMLINK+="misc/vtuner5"
KERNEL=="vtuner6", SYMLINK+="misc/vtuner6"
KERNEL=="vtuner7", SYMLINK+="misc/vtuner7"
KERNEL=="vtuner8", SYMLINK+="misc/vtuner8"
KERNEL=="vtuner9", SYMLINK+="misc/vtuner9"
KERNEL=="vtuner10", SYMLINK+="misc/vtuner10"
KERNEL=="vtuner11", SYMLINK+="misc/vtuner11"
KERNEL=="pvr", SYMLINK+="misc/pvr"
KERNEL=="i2c-[0-9]*", SYMLINK+="i2c/%n"

# The first framebuffer is symlinked to /dev/fb
KERNEL=="fb0", SYMLINK+="fb/0"

# Make all input devices read-write to the input group
SUBSYSTEM=="input", GROUP="input", MODE="660"

AFAIK all of these are already in, so I don't see what makes the difference?

[Erik Slagter]

Yes...

https://stackoverflo...instead-of-udev

I also think that busybox-mdev has to be reactivated to manage the virtual devices.
So it seems in the OE Alliance Branch 4.4.

 

This will not fix anything at all.

[el bandido]

# There are a number of modifiers that are allowed to be used in some
# of the different fields. They provide the following subsitutions:
#
# %n the "kernel number" of the device.
#    For example, 'sda3' has a "kernel number" of '3'
# %e the smallest number for that name which does not matches an existing node
# %k the kernel name for the device
# %M the kernel major number for the device
# %m the kernel minor number for the device
# %b the bus id for the device
# %c the string returned by the PROGRAM
# %s{filename} the content of a sysfs attribute
# %% the '%' char itself

# The first rtc device is symlinked to /dev/rtc
KERNEL=="rtc0", SYMLINK+="rtc"

# Legacy paths for dbox devices
KERNEL=="oled0", SYMLINK+="dbox/oled0"
KERNEL=="fp0", SYMLINK+="dbox/fp0"
KERNEL=="lcd0", SYMLINK+="dbox/lcd0"
KERNEL=="vtuner0", SYMLINK+="misc/vtuner0"
KERNEL=="vtuner1", SYMLINK+="misc/vtuner1"
KERNEL=="vtuner2", SYMLINK+="misc/vtuner2"
KERNEL=="vtuner3", SYMLINK+="misc/vtuner3"
KERNEL=="vtuner4", SYMLINK+="misc/vtuner4"
KERNEL=="vtuner5", SYMLINK+="misc/vtuner5"
KERNEL=="vtuner6", SYMLINK+="misc/vtuner6"
KERNEL=="vtuner7", SYMLINK+="misc/vtuner7"
KERNEL=="vtuner8", SYMLINK+="misc/vtuner8"
KERNEL=="vtuner9", SYMLINK+="misc/vtuner9"
KERNEL=="vtuner10", SYMLINK+="misc/vtuner10"
KERNEL=="vtuner11", SYMLINK+="misc/vtuner11"
KERNEL=="pvr", SYMLINK+="misc/pvr"
KERNEL=="i2c-[0-9]*", SYMLINK+="i2c/%n"

# The first framebuffer is symlinked to /dev/fb
KERNEL=="fb0", SYMLINK+="fb/0"

# Make all input devices read-write to the input group
SUBSYSTEM=="input", GROUP="input", MODE="660"

AFAIK all of these are already in, so I don't see what makes the difference?

 

The current localextra.rules only allows for 4 vtuners, and the Edision MIO can do a lot more than that.
https://github.com/O...ocalextra.rules

[rantanplan]

Perhaps the tuner assignment can also be displayed freely variable instead of the static assignment.

 

 

KERNEL=="vtuner0", SYMLINK+="misc/vtuner0"
KERNEL=="vtuner1", SYMLINK+="misc/vtuner1"
KERNEL=="vtuner2", SYMLINK+="misc/vtuner2"
KERNEL=="vtuner3", SYMLINK+="misc/vtuner3"
KERNEL=="vtuner4", SYMLINK+="misc/vtuner4"
KERNEL=="vtuner5", SYMLINK+="misc/vtuner5"
KERNEL=="vtuner6", SYMLINK+="misc/vtuner6"
KERNEL=="vtuner7", SYMLINK+="misc/vtuner7"
KERNEL=="vtuner8", SYMLINK+="misc/vtuner8"
KERNEL=="vtuner9", SYMLINK+="misc/vtuner9"
KERNEL=="vtuner10", SYMLINK+="misc/vtuner10"
KERNEL=="vtuner11", SYMLINK+="misc/vtuner11"

change with

KERNEL=="vtuner%n", SYMLINK+="misc/vtuner%n"

I don't know if that will work.
just an idea