Recently, upgrade device autodetection has been added to the mvebu target.
This exposes some shortcomings of the generic export_bootdevice function,
e.g. on the Turris Omnia: export_bootdevice silently reports the root
partition to be the boot device. This makes the sysupgrade process fail at
several places.
Fix this by clearly distinguishing between /proc/cmdline arguments which
specify the boot disk, and those which specify the root partition. Only in
the latter case, strip off the partition, and do it consistently.
root=PARTUUID=<pseudo PARTUUID for MBR> (any partition) and root=/dev/*
(any partition) are accepted.
The root of the problem is that the *existing* export_bootdevice in
/lib/upgrade/common.sh behaves differently, if the kernel is booted with
root=/dev/..., or if it is booted with root=PARTUUID=...
In the former case, it reports back major/minor of the root partition,
in the latter case it reports back major/minor of the complete boot disk.
Targets, which boot with root=/dev/... *and* use export_bootdevice /
export_partdevice, have added workarounds to this behaviour, by specifying
*negative* increments to the export_partdevice function.
Consequently, those targets have to be adapted to use positive increments,
otherwise they are broken by the change to export_bootdevice.
Fixes: 4e8345ff68 ("mvebu: base-files: autodetect upgrade device")
Signed-off-by: Klaus Kudielka <klaus.kudielka@gmail.com>
Tested-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
Currently leds migration scripts in ar71xx and lantiq share a lot of
logic and introducing leds migration to another target would mean
copying this code, again. Therefore add common logic to library in
base-files package.
Suggested-by: Petr Štetiar <ynezz@true.cz>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
ucert needs to check the firmware part with metadata, but without the signature.
Use the new fwtool mode to extract that without altering the firmware image inside
the check
Signed-off-by: Felix Fietkau <nbd@nbd.name>
When config_get is called as "config_get section option" the option
is unexpectedly globbed by the shell which differs from the way options
are read to a variable with "config_get variable section option".
Add another layer of double quotes to fix it.
Signed-off-by: Günther Kelleter <guenther.kelleter@devolo.de>
Add support for passing additional parameters to mtd called during
sysupgrade. It will be required to toggle the "recovery moe" flag
supported by recent tp-link boards.
Signed-off-by: David Bauer <mail@david-bauer.net>
[split code from board support patch; add commit message]
Signed-off-by: Mathias Kresin <dev@kresin.me>
This commit allows to use non-standard UBI volume name as the rootfs
volume in sysupgrade.
ex.:
The U-Boot on Buffalo WXR-2533DHP checks existence and checksum of
"ubi_rootfs" volume when booting, so this name is required.
OpenWrt currently provides several patches:
490-ubi-auto-attach-mtd-device-named-ubi-or-data-on-boot.patch
491-ubi-auto-create-ubiblock-device-for-rootfs.patch
492-try-auto-mounting-ubi0-rootfs-in-init-do_mounts.c.patch
to facilitate ubi rootfs automount. However the upstream kernel
also supports the means of booting from a fully custom ubi
partition name and ubi volume name via bootargs/kernel's cmdline
parameters:
ubi.mtd=mtd_partition_name
ubi.block=rootfs_volume_name
root=/dev/ubiblock$X_$Y
For more information and examples visit the wiki over at linux-mtd:
<http://www.linux-mtd.infradead.org/faq/ubifs.html>
<http://www.linux-mtd.infradead.org/doc/ubi.html>
Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com> [reworded commit]
This reverts commit 41770add03.
The fwtool_check_image() procedure is used by `sysupgrade --test` which must
not alter the image under test in any way.
Currently, when the LuCI ui or any other sysupgrade wrapper first invokes
sysupgrade --test to verify the compatibility of the image and then calculates
the sha256sum over it, the resulting checksum will differ from the original
image since the test invocation will implicitely strip the metadata trailer.
To properly fix the underlying issue, the combined image checksumming code
must be modified to skip the metadata trailer.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
The return value of a package prerm script is discarded and not returned
correctly by default_prerm(). This allows other operations like service
shutdown to "leak" their return value, prompting workarounds like commit
48cfc826 which do not address the root cause.
Preserve a package prerm script return value for use by default_prerm(),
sharing the corresponding code from default_postinst() for consistency.
Also use consistent code for handling of /etc/init.d/ scripts.
Run Tested on: LEDE 17.01.4 running ar71xx.
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
If I create following image:
define Device/engenius-m36
IMAGE/sysupgrade.bin := combined-image | append-metadata
endef
Sysupgrade then errors out:
Invalid image. Contents do not match checksum (image:cd285595eaf297370404ae0e2815ec1a calculated:2cf9a2286fb6b01af3ea189128017d44)
Image check 'platform_check_image' failed.
By removing the metadata from the image I get combined-image checksum
working again and sysupgrade works.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
The swconfig switch led driver has the ability to switch
between a "link, rx and/or tx" mode. However, this feature
was not implemented in uci, the led init script and
config_generate.
This patch adds a seventh parameter to the
ucidef_set_led_switch() function. The accepted values for
this parameter are: link, rx and tx.
Any permutations of these three values are supported, as
long as they are properly encased with quotes.
If the parameter is not specified it will default to "all"
(link rx tx).
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Commit 7f694582 introduced a bug where default_postinst() often fails to
remove a uci-defaults script after application, leaving it to run again
after a reboot.
(Note: commit 7f694582 also introduced FS#1021, now fixed by 73c745f6)
The subtle problem arises from the shell logical chain:
[ -f "$i" ] && . "$i" && rm -f "$i"
Most uci-defaults scripts contain a terminal 'exit 0' statement which,
when sourced, results in the logic chain exiting before executing 'rm -f'.
This was observed while testing upgrades of 'luci-app-sqm'.
The solution is to wrap the shell sourcing in a subshell relative to the
command 'rm -f':
( [ -f "$i" ] && . "$i" ) && rm -f "$i"
Revert to using 'grep' to prefilter the list of entries from the control
file, which yields the full path of uci-defaults scripts. This allows
keeping the existence check, directory change and script sourcing inside
the subshell, with the script removal correctly outside.
This approach avoids adding a second subshell only around the "." (source)
command. The change also preserves the fix FS#1021, since the full path is
used to source the script, which is POSIX-portable irrespective of PATH
variable or reference to the CWD.
Run Tested on: LEDE 17.01.4 running ar71xx, while tracing installation of
package luci-app-sqm with its associated /etc/uci-defaults/luci-sqm file.
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
Set the (sys)upgrade state when sourcing the stage2 script instead of
setting the state for each target individual.
This change fixes the, due to a missing state set, not working upgrade
led on ath79 and apm821xx.
Signed-off-by: Mathias Kresin <dev@kresin.me>
Add a function to get a mac stored as text from flash. The octets of
the mac address need to be separated by any separator supported by
macaddr_canonicalize().
Signed-off-by: Dmitry Tunin <hanipouspilot@gmail.com>
Signed-off-by: Mathias Kresin <dev@kresin.me>
Add the opening bracket right after the function name, to do it the
same way for all functions in this file.
Signed-off-by: Dmitry Tunin <hanipouspilot@gmail.com>
Signed-off-by: Mathias Kresin <dev@kresin.me>
Verify ucert signature chains in sysupgrade images in case ucert is
installed and $CHECK_IMAGE_SIGNARURE = 1.
Also make sure ucert host binary is present and generate a self-signed
ucert in case $TOPDIR/key-build.ucert is missing.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
this feature has never worked, the fw image name was not passed and the -t
parameter was missing in the tool invocation. drop the feature.
Signed-off-by: John Crispin <john@phrozen.org>
When attempting to use any of the functions in network.sh while netifd is
not started yet, the ubus interface dump query will fail with "Not found",
yielding an empty response.
Subsequently, jsonfilter is invoked with an empty string instead of a valid
JSON document, causing it to emit a second "unexpected end of data" error.
This caused the dnsmasq init script to log the following errors during
early boot on some systems:
procd: /etc/rc.d/S19dnsmasq: Command failed: Not found.
procd: /etc/rc.d/S19dnsmasq: Failed to parse json data: unexpected end of data.
Fix the issue by allowing the ubus query to fail with "Not found" but still
logging other failures, and by passing an empty JSON object to jsonfilter
if the interface status cache is empty.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
On bcm53xx and brcm47xx, commands are passed to default_do_upgrade that
expect the image to be passed on stdin, rather than as an argument.
Fixes: 30f61a34b4 ("base-files: always use staged sysupgrade")
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
There are several long-standing issues present in the UCI shell API as
documented in https://wiki.openwrt.org/doc/devel/config-scripting. They
relate both to high-level, user-defined callback functions used to
process UCI config files, and also to low-level functions used within
scripts generally.
The related problems have been encountered now and in the past, e.g.
https://forum.openwrt.org/viewtopic.php?id=54295, and include:
a) UCI parsing option() function and user-defined option_cb() callbacks
being erroneously called during processing of "list" config file entries;
b) normal usage of the low-level config_set() unexpectedy calling any
defined option_cb() if present; and
c) handling of the list_cb() not respecting the NO_CALLBACK variable.
Root causes include a function stack "inversion", where the low-level
config_set() function incorrectly calls the high-level option() function,
intended only for processing the "option" keyword of UCI config files.
This change addresses the inversion and other issues, making the option
handling code more consistent and smaller, and simplifying developers'
usage of UCI callbacks.
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com> [PKG_RELEASE increase]
The previous refactoring of ucidef_set_interface() removed the protocol
selection heuristic which breaks the networking defaults for the majority
of boards.
Re-add the protocol selection and rename two bad "proto" references to
the expected "protocol" value.
Fixes: 85048a9c1f ("base-files: rework _ucidef_set_interface to be more generic")
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
This is a rework of previously submitted patch reworking
ucidef_set_interface_raw [1]. Here, keep the idea but instead
make _ucidef_set_interface more generic and use it instead of
ucidef_set_interface_raw.
Also change the users like ucidef_set_interface_lan and others.
[1] https://patchwork.ozlabs.org/patch/844961/
Signed-off-by: Roman Yeryomin <roman@advem.lv>
It's intentional that some boards within a target don't have all LEDs
which are tried to be setup in a common script. Don't show a warning in
such cases.
Fixes: 4f4fc993db ("base-files: add more name source to get_dt_led helper function")
Signed-off-by: Mathias Kresin <dev@kresin.me>
Not all LED driver are using the label devicetree property for the led
name. Add support for the TI/National Semiconductor LP55xx Led Drivers,
which are using the chan-name property for the led name, as fallback.
Signed-off-by: Mathias Kresin <dev@kresin.me>
Add a fucntion to get the a binary mac address from file. Use the new
function for mtd_get_mac_binary() to limit duplicate code.
Signed-off-by: Mathias Kresin <dev@kresin.me>
Current code and also before commit da52dd0c83 was vulnerable to shell
injection using volume lables in the GPT partition table of block
devices. Given that partition names can be freely defined in GPT tables
we really shouldn't evaluate a string which is potentially crafted with
evil intentions. Hence rather use `export -n` to absorb the uevent's
variables into the environment.
Fixes commit da52dd0c83 (base-files: quote values when evaluating uevent)
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
[mschiffer@universe-factory.net: suggested export -n usage]
When sourcing /sys/class/block/*/uevent values have to be quoted as
they may contain spaces (e.g. in PARTNAME).
Fix this by pre-processing with sed before sourcing.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Add the uci option nameprefix to specifc a target netdev name. Patch the
br2684ctl code to accept and set a netdev name via commandline parameters.
It allows to use the same netdev name for ATM and PTM lines on lantiq
xdsl hardware.
Signed-off-by: Martin Schiller <ms@dev.tdt.de>
Signed-off-by: Mathis Kresin <dev@kresin.me>
Lantiq and IPQ806X (which includes IPQ40XX) both define the
same custom function {ipq806x|lantiq}_get_dt_led.
This patch moves the function into the base-file package at
lib/functions/leds.sh to make it more accessible for other
targets as well.
Cc: Mathias Kresin <dev@kresin.me>
Cc: John Crispin <john@phrozen.org>
Cc: Hannu Nyman <hannu.nyman@iki.fi>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Kernel and rootfs in a subdirectory matching the userspace boardname,
was intended to use a single sysupgrade-tar archive for multiple boards
with different kernel/rootfs images. This feature was never used.
Use the first found directory in the tar archive instead of relying on
a directory named according to the userspace boardname.
It allows to change the boardname without adding another compatibility
layer - using the nand_board_name() function - for (sub)targets using
the metadata based image validation in favour to
nand_do_platform_check().
Signed-off-by: Mathias Kresin <dev@kresin.me>
This patch fixes two issues with the current get_partitions()
function.
First: "Invalid partition table on $disk" will pop up on
legitimate images on big endian system.
This is because the little-endian representation of "55 AA" is
assumed in the context of little-endian architectures. On these
comparing it to the 16-bit word 0xAA55 does work as intented.
Whereas on big-endian systems, this would have to be 0x55AA.
This patch fixes the issue by replacing the integer conversion
and value match check with just a string comparision.
Second: The extraction of the type, start LBA and LBA num from
the partition table has the same endianness issue. This has been
fixed by using the new hex_le32_to_cpu() function. This function
will translate the stored little-endian data to the correct
byte-order if necessary.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Add the parent of the sysupgrade script to the list of pids not getting
killed
Signed-off-by: Mat Trudel <mat@geeky.net>
Signed-off-by: John Crispin <john@phrozen.org>
Shell function return code only has range [0, 255]. Other values will
be truncated, e.g. return 65536 will have the same effect as return 0
While at it, drop other "return $rc" where rc will almost always take
value 0 and whose value current callers actually do not check
Fixes FS#988
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
When we run "opkg install" on a package that installs an uci-defaults
script, functions.sh will fail to evaluate that script in its
default_postinst function.
This happens because there is no "./" present and it searches for the
file in paths specified by the PATH variable. This would work on bash,
but it will not work on ash and some other shells like sh, zsh. This
applys to the ". filename" directive used in this case.
This patch will make the path relative to the /etc/uci-defaults
directory.
Fixes: FS#1021
Signed-off-by: Marko Ratkaj <marko.ratkaj@sartura.hr>
Let the generic postinstall script invoke "kmodloader" when the just
installed package contains any /etc/module.d/ entries.
This allows us to skip the explicit "insert_module()" calls in the
package postinstall.
Due to the removed insert_module calls we do not need to assemble a
complete list of modules per package anymore, which allows for vast
simplification of the package generation code.
While we're at it, also support specifying default parameters for
modules using either the MODPARAM or MODPARAM.modulename variables
in KernelPackage.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
With failsafe disabled there is no point in early network setup. We
don't send announcement over UDP and there is no way to ssh to the
device.
A side effect of this is avoiding a possibly incorrect network config
(only with failsafe disabled). This problem is related to possible
changes made by user in /etc/config/network.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
When processes don't die on SIGKILL (usually because of kernel bugs), it's
better to give up instead of looping forever.
upgraded will trigger a reboot in this case (and if this fails, a hardware
watchdog will eventually time out and reset the system, if present).
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
For targets using the generic board detection and board specific
settings in diag.sh, the board name is still unset at the time the
set_state() provided by diag.sh is called by 10_indicate_preinit.
Change the execution order to ensure the boardname is populated before
required the first time. Do the target specific board detection as
early as possible, directly followed by the generic one to allow a
seamless switch to the generic function for populating /tmp/sysinfo/.
Signed-off-by: Mathias Kresin <dev@kresin.me>
Depending on busybox applet selection, paths of basic utiilties may differ,
and may not work as symlinks to busybox. Simply using whatever binary is
found in PATH and detecting symlinks automatically is more robust and
easier to maintain.
The list of binaries is also slightly cleaned up and duplicates are
removed.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
There already exist static assignment of uid/gid 65533 in packages feed
and we have nobody/nogroup taking 65534 as their ids. Let's change the
pid of dynamic assignment to start from 65536 so that the two assignment
scheme will not collide with each other
While at it, fix the scan command checking existence of uid/gid
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
The offset and factor are only related for LEDs which can have
different brightness values. But binary LEDs are more common and don't
require any further configuation than setting the factor to 1.
Use offset = 0 and factor = 1 in case nothing else is specified.
Signed-off-by: Mathias Kresin <dev@kresin.me>
