Age | Commit message (Collapse) | Author |
|
Since oe-core commit 3150253898ba ("systemd-conf: simplify
creation of machine-specific configuration") systemd-conf no
longer unpacks the systemd sources. Hence the patch which patches
units/systemd-udevd.service.in does no longer apply.
Apply the patch in the systemd recipe instead.
The systemd-udevd.service file is only deployed in a package
built from the systemd recipe anyway, hence this also makes sure
that the patched file gets actually deployed.
This essentially reverts commit 118f6b1b17b6 ("systemd: follow
recipe split which added systemd-conf").
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
|
|
Upstream has changes in the context lines.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
The patch is not obsolete AFAIKT.
i.e. the configuration file should NOT make mounts by udevd private or
other processes will not be able to see the mount.
This reverts commit 3640a12c85df4ac180e38f19038072df6b1d6d55.
|
|
That patch has been merged to systemd-conf.
Signed-off-by: Ming Liu <liu.ming50@gmail.com>
|
|
Or else the changes for systemd-conf recipe will be missed.
Signed-off-by: Ming Liu <liu.ming50@gmail.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
The systemd configuration has been split from the systemd recipe.
Thus we need to move our changes to the systemd configuration.
While at it drop the kmod packageconfig. This was only needed with angstrom.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
These machines use an old Nvidia provided downstream kernel 3.1.10. They no
longer much the way more recent sysroot components, e.g. libc, xserver et. al.
The maintance burden to keep the userspace components in their old version
becomes simply to high. Keep using the rocko based 2.8 BSP for these machines.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
This starts a service which inserts kernel modules on boot.
Updates in meta-angstrom removed this default packageconfig.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Keywords in udevd.service.in were changed upstream.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
OE-core dropped to do this for whatever reasons.
Compare with openembedded core, commit 6c1765a05c1321f08f3d1fb521dfe6b07bc8e92
* Drop mount propagation patch, it only happens with libseccomp, OE doesnt
enable it
This reverts commit a8ee603737444c74db61d0a705b08e9a1dea5158.
The patch is forward ported to systemd 234.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Use '${COMMON_LICENSE_DIR} for license files
fix the warning:
| WARNING: packagegroup-xfce-base-1.0-r5 do_populate_lic:
| ${COREBASE}/LICENSE is not a valid license file, please use
|'${COMMON_LICENSE_DIR}/MIT' for a MIT License file in LIC_FILES_CHKSUM.
|This will become an error in the future
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Configure the read_ahead cache size to optimal value, for the
mmcblk devices (eMMC and SD card).
The cache values are module specific, tests were run on the pre-built
image (BSP V2.7 Beta 1). The measurement of read speed for a range of
cache sizes (16kB to 256MB) was made using hdparm -t.
Signed-off-by: Leonardo Graboski Veiga <leogveiga@gmail.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
In case the resizefs.sh script ends quite fast, the system might still
be in boot sequence. Removing a systemd service can be problematic
in that case: It seems that systemd looses the state of .device units,
and some services depend on such units (namely serial consoles such as
serial-getty@ttymxc0.service). The following sequence shows this problem:
Jan 12 00:16:45 apalis-imx6 sh[399]: Removed /etc/systemd/system/multi-user.target.wants/resizefs.service.
Jan 12 00:16:45 apalis-imx6 systemd[1]: Reloading.
Jan 12 00:16:45 apalis-imx6 root[704]: resizing mmcblk0p2 finished, new size is 3755M
...
Jan 19 00:04:08 apalis-imx6 systemd[1]: dev-ttymxc0.device: Job dev-ttymxc0.device/start timed out.
Jan 19 00:04:08 apalis-imx6 systemd[1]: Timed out waiting for device dev-ttymxc0.device.
Jan 19 00:04:08 apalis-imx6 systemd[1]: Dependency failed for Serial Getty on ttymxc0.
(the time has been updated between this to events, but that does not
influence the issue)
Using --no-reload avoids the "Reloading." message above and seems to not
cause such issues anymore.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
The script used for all machines (e.g. T30) and the mx6 specific
script are the very same scripts. Drop the mx6 specific script,
with that the recipe will use the general scripts for all machines.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
The BSP should automount for all users.
The Demo Software Layers cares for RNDIS.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
|
|
oe-core moved the mountpoint used to /run/media/<drive>.
In our setup we don't see issues with /media being read only
when udev starts mounting.
In our environment with graphical desktop udisks will later take over
the automount job and udisks currently still uses /media,
so keep this consistent between mounting at boottime and mounting at runtime.
Refer to openembedded-core commits:
acfe3014d41de5e87cdbc58d0396349c6b9c3ffd
62ae16c40252f39ba28e072218d67f47b26b3535
|
|
|
|
Non graphical images will never execute graphical.target services.
|
|
With fido e2fsprogs did split out the package e2fsprogs-resize2fs.
RRECOMMEND it so that it gets picked up during rootfs creation.
|
|
We noticed that our latest move to using systemd/networkd leaves the
rndis0 interface down. Manually bringing the link up solved the issue.
As a workaround add upping the link to our start-rndis.sh script.
|
|
This reverts commit bd2d32bf3a8460c87849e75ad01e11450d8cda10.
Remove the backport, is part of fido.
|
|
Angstom Dizzy now uses systemd V218 so there is no need to build
a component seperately anymore.
This reverts commit 16b8b392ad048566729f30d7c3887d1bfdf21bd4.
|
|
This changes the minimum required Linux version from 3.7 back to 3.0.
[YOCTO #7409]
Signed-off-by: Jonathan Liu <net147@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
backported from openembedded-core,
commit 31163abc7bedd477c1aad10e6d1f47dbee194513
|
|
This reverts commit 3cf2edb10f5e8ac60db7ca2e71d1e754a177d08c.
Conflicts:
recipes-core/systemd/systemd_%.bbappend
|
|
Since we use networkd now for RNDIS network interface setup now,
we can remove the custom part which starts up the RNDIS device and
sets up a DHCP server.
|
|
Since Connman is better suited for more advanced Wireless solutions,
we keep Connman for the main network configuration manager. However,
for the simple RNDIS debug/test configuration, Connman is not well
suited since Connman tries to maintain one connection only at any
given time.
This change enables networkd and adds a simple standard configuration
for USB/RNDIS network devices.
|
|
When the link goes down, the DHCP service has been disabled, while
there is no corresponding functionality which enables the service
when the link goes up again. This fix is taken from systemd 217.
|
|
|
|
The SUMMARY variable gives a short description of the package (<72 chars).
A missing DESCRIPTION is automatically set to the content of SUMMARY.
|
|
|
|
- Make sure leases file is not persistent (in /var/run now)
This allows to serve the same IP to the generated (and hence
different IP addresses)
- Start RNDIS service after the network.target
- Tegra: Add extra service for udhcpd. When running dhcpd directly
from udev, the daemon will be killed after a while
|
|
meta-angstrom now overrides the effect of the patch and provides the
journald.conffile in full.
To override a setting either create the file with your settings in
meta-toradex/recipes-core/systemd/systemd/journald.conf or
edit meta-angstrom/recipes-core/systemd/systemd/journald.conf.
|
|
|
|
Disable USB RNDIS systemd service for now due to possible race upon
first boot:
[ OK ] Started Load/Configure the kernel driver providing the RNDIS
gadget.
=> system hang
|
|
|
|
|
|
ifconfig does no longer understand 192.168.11.2/24 and its
use is deprecated, so instead of changing to another way of
stating the subnetmask use /sbin/ip.
|
|
i.MX6 and Vybrid modules use the g_ether driver for RNDIS functionality.
This commit uses a systemd service to load the kernel module, configures the new
usb0 NIC and starts a dhcp server to serve an IP address to a USB host.
The Tegras use the android gadget driver. The configuration remains unchanged.
|
|
Patch pending in meta-angstrom
|
|
|
|
|
|
|
|
This uses the % wildcard for recipe matching.
remove the deprecated PRINC variable while at it.
|
|
|