make static linking work for upgraded

Signed-off-by: John Crispin <blogic@openwrt.org>

only build upgraded on mips

Signed-off-by: John Crispin <blogic@openwrt.org>

move upgrade to its own subfolder

Signed-off-by: John Crispin <blogic@openwrt.org>

rename the folder option to path

Signed-off-by: John Crispin <blogic@openwrt.org>

procd: force -lc -lgcc to handle cross-dependencies between those two

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

Increase buffers for hotplug and cmdline for UML

Thanks to Markus Stenberg for debugging and testing.

Signed-off-by: Steven Barth <steven@midlink.org>

only add the nandupgrade method when /sbin/upgraded exists

Signed-off-by: John Crispin <blogic@openwrt.org>

add upgraded support for nand flashing

Signed-off-by: John Crispin <blogic@openwrt.org>

allow instances to report errors. if an error is set, the instance wont be started

Signed-off-by: John Crispin <blogic@openwrt.org>

service: do not emit any service events before the ubus connection is up

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: register the subscriber object before the event handler

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: fix a debug message typo

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: fix a leftover debug hack

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

0467aebb79f64d1c9d05a84367fda7c02c4ea6b8 included too much

Signed-off-by: John Crispin <blogic@openwrt.org>

wait for kmod loader to complete on first boot

Signed-off-by: John Crispin <blogic@openwrt.org>

send ubus_notify events when servers and instances change state

Signed-off-by: John Crispin <blogic@openwrt.org>

add wildcard support to trigger matching code

Signed-off-by: John Crispin <blogic@openwrt.org>

allow instances to register ubus object that should be watched

Signed-off-by: John Crispin <blogic@openwrt.org>

service: fix json object leak in validate code

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: free trigger json_script context

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: fix trigger data memleak

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: ensure that trigger timers are always cancelled before free

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: fix memleak in trigger handling

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: use avl_remove_all_elements()

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: remove unused argument to service_update()

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: fix double-free in error path

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

service: use blob_memdup()

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

hotplug: fix memleak in debug codepath

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

inittab: strip trailing whitespace (especially newline)

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

inittab: avoid character collating to work around character class escaping issue in a regex

This makes the regex work with musl

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

rcS: fix a format string bug

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

rcS: do not access ustream_buf data directly

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

instance: improve portability of setting resource limits

the data type varies for different implementations

Signed-off-by: Felix Fietkau <nbd@openwrt.org>

commit e0921ed71bab0ad7d2344774a33c22809eb10190 set the timeout globally

lets restrict it to stop scripts only

Signed-off-by: John Crispin <blogic@openwrt.org>

fix the dump logic to show validators and triggers even if no instances exist

Signed-off-by: John Crispin <blogic@openwrt.org>

force all runqueue tasks to a timeout of 15s

Signed-off-by: John Crispin <blogic@openwrt.org>

state.c: pass ubus commandline is writable buffer to service_start_early()

The service_start_early() function uses strtok() on the command line argument,
therfore we cannot pass it constant string literals directly.

Fixes OpenWrt bug #14832

Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>

procd: fix service file tracking

The instance_config_move() function in the procd instance managing code is
missing a blobmsg_list_move() call to update the tracked file items with the
newly calculated checksums in case of a file change.

Without that change, services are always reloaded after the first change to
a tracked file regardless of whether there are any subsequent changes or not.

Add a missing instance of blobmsg_list_free() as well.

Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>

fix copy paste error

Signed-off-by: John Crispin <blogic@openwrt.org>

procd: set resource limits for service instances

useful to enable core dumps or set resource limits

Signed-off-by: Ulrich Weber <uw@xyne.com>

cleanup debug level handover

Signed-off-by: John Crispin <blogic@openwrt.org>

fix rcS memory leaks and running hook

Signed-off-by: John Crispin <blogic@openwrt.org>

make ue of the md5.c inside libubox

Signed-off-by: John Crispin <blogic@openwrt.org>

fix up the order of respawn parameters for ubus

Signed-off-by: John Crispin <blogic@openwrt.org>

add a event broadcast function

Signed-off-by: John Crispin <blogic@openwrt.org>

add debug level handover between preinit and main process

Signed-off-by: John Crispin <blogic@openwrt.org>

make the service running trigger be queued directly after the service was startetd

Signed-off-by: John Crispin <blogic@openwrt.org>

add a function to add calls to the front of the rcS queue

Signed-off-by: John Crispin <blogic@openwrt.org>

procd: Allow process respawning forever

Use respawn_retry==0 as indication to respawn a process forever.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John Crispin <blogic@openwrt.org>

debloat and reorganize code
split app into procd and init binaries
remove log support, this is an external service now

Signed-off-by: John Crispin <blogic@openwrt.org>

add service_validator support

Signed-off-by: John Crispin <blogic@openwrt.org>

add data type validator

Signed-off-by: John Crispin <blogic@openwrt.org>

bind the console to /dev/null if the real console fails to come up

this was reported on irc by oc80z

Signed-off-by: John Crispin <blogic@openwrt.org>

the automatic service restart should not happen during sysupgrade

Signed-off-by: John Crispin <blogic@openwrt.org>

add hostname option to logread

Signed-off-by: John Crispin <blogic@openwrt.org>

add a prefix option for messages streamed using logread

Signed-off-by: Robin Kuck <robin@basicinside.de>

syslog: fix incorrect use of sizeof() in vsnprintf()

syslog() is a blocking call on eglibc. as procd provides the actual syslog, weneed to make sure that we do not run into a deadlock.

Signed-off-by: John Crispin <blogic@openwrt.org>
Signed-off-by: Markus Stenberg <markus.stenberg@iki.fi>

procd: Exit askfirst on read error

When running askfirst on an unused tty device askfirst starts
busylooping forever. Fix this by returning an error if we read
an EOF.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>

coldplug: remove duplicated include

The unistd.h header is inluded twice.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

coldplug: only start hotplug timer after udevtrigger finished

Currently the last event timer for hotplug is
started after the udevtrigger process is forked.

Starting the timer means that the first hotplug
event must come in before the timer expires.
During this time the kernel must load the binary
of the udevtrigger from the underlying rootfs
which can be time consuming on slow systems.

On such systems, the timer expires before the
first hotplug event happens which results in
broken behaviour. Due to the missing device nodes,
not the ubus nor the procd code works correctly in
that case.

Change the code to only start the hotplug last
event timer after the udevtrigger process finishes.
This ensures that the delay caused by accessing the
underlying filesystem does not affects the desired
behaviour.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: nuke scan_{subdir,class,block} functions

These functions are simple wrappers around the
'scan_subdir' function. Remove the wrapper code and
use the 'scan_subdir' function directly instead.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: simplify scan_class function

Remove the local directory scanning code and
use the scan_subdir function instead.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: simplify scan_block function

Remove the local directory scanning code and
use the scan_subdir function instead.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: simplify scan_subsystem function

Remove the local directory scanning code and
use the scan_subdir function instead.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: extend scan_subdir function

Add more arguments to the function to make it
usable recursively.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: remove superfluous strlcpy

Use the "/sys/class/block" string directly
for the stat call. This makes the code simpler
and reduces stack usage.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: add back directory name check into scan_subdir

It has been accidentally removed by:

commit 824a4b3da2c0be19ccf54a48dcd4c10677b82c1a
Author: Gabor Juhos <juhosg@openwrt.org>
Date:   Tue Sep 10 12:59:25 2013 +0200

    udevtrigger: reduce indent level in scan_* functions

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

fix skewed parameter order in respawn code

Signed-off-by: John Crispin <blogic@openwrt.org>

execute the "running" hook once the daemon is started

Signed-off-by: John Crispin <blogic@openwrt.org>

add a way for procd to queue async init.d calls

Signed-off-by: John Crispin <blogic@openwrt.org>

fix a bug in the trigger handling code

Signed-off-by: John Crispin <blogic@openwrt.org>

remove now unused variable

Signed-off-by: John Crispin <blogic@openwrt.org>

add code to configure respawn via ubus

Signed-off-by: John Crispin <blogic@openwrt.org>

output the env associated with an instance

Signed-off-by: John Crispin <blogic@openwrt.org>

udevtrigger: reduce indent level in scan_* functions

Return early from the scan_* functions if the
opendir call fails. This allows to reduce the
indentation level of the subsequent code by
one tab.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: rename variables in scan_subdir

There are only one variable present for a given
type. Remove the '2' suffix from the variable
names. Also rename the function argument to avoid
name collision.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: use a helper function for subdir scanning

The scan_{block,class,subsystem} functions are using
identical code for scanning. Move that into a separate
function to avoid code duplication.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: check presence of dev attribute

The presence of the 'uevent' attribute file
does not always mean that a corresponding
device node can be created under '/dev'.

For valid device nodes, a 'dev' attribute
file must be present which cntains the major
and minor numbers of the device.

Modify the code to check the presence of the
dev attribute as well. This allows to avoid
superfluous trigger_uevent calls. On a test
system, the patch reduced the number of the
triegger_uevent calls from 144 to 70.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: add device_has_attribute helper

Move the code from device_list_insert() into
a separate function. This makes it possible
to reuse the code for other attribute files.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

udevtrigger: remove trailing whitespaces

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

add a ubus handler that allows sending signals to processes

Signed-off-by: John Crispin <blogic@openwrt.org>

fix use after free bug in the trigger handling code

Signed-off-by: John Crispin <blogic@openwrt.org>

add suport for measuring process resource usage

Signed-off-by: John Crispin <blogic@openwrt.org>

fixup service_dump logic

Signed-off-by: John Crispin <blogic@openwrt.org>

trivial code cleanup

Signed-off-by: John Crispin <blogic@openwrt.org>

add respawn handling

Signed-off-by: John Crispin <blogic@openwrt.org>

logread: fix the facility name reporting

http://patchwork.openwrt.org/patch/3978/

Signed-off-by: Mike Brady<mikebrady@eircom.net>

service: terminate uloop after fork(), redirect stdin, out and err to /dev/null

hotplug: terminate uloop after fork(), redirect stdin, out and err to /dev/null

askfirst: remove superfluous 'close' calls on stdio file descriptors.

'dup2' will do it for us if necessary.

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>

Parse only init_debug option with non-empty argument.

Matching empty string makes rm_so and rm_eo of struct regmatch_t have
the same value, in this case both of them are 11. This causes a call to
atoi("\0") whose return value 0 is just the initial value of 'debug'.

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>

Error and return when failing stat('/dev/console').

Since mkdev("*", 0600) has been called in early_dev just before
early_console, no need to try checking it again.

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>

Fix event type reporting when calling 'procd_shutdown'.

Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>

early: remove superfluous close of stdio file descriptors

'man dup2' says:

  dup2() makes newfd be the copy of oldfd, closing newfd first if
  necessary, but note the following:

  *  If oldfd is not a valid file descriptor, then the call fails,
     and newfd is not closed.

  *  If oldfd is a valid file descriptor, and newfd has the same
     value as oldfd, then dup2() does nothing, and returns newfd.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

early: keep stdio files open

At the end of the 'early_console' function, the
file descriptor is closed unconditionally. This
'close' call closes the stdio files if the fd
returned by the 'open(dev/console)' call equals
with any of the STD{IN,OUT,ERR}_FILENO values.
When this happens, all subsequent accesses to
the stdio files will fail and early console
access won't work.

To avoid this, don't close the file descriptor if
that equals with any of the STD*_FILENO values.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>

the cloexec logic of the watchdog was broken

Signed-off-by: John Crispin <blogic@openwrt.org>

logread should handle return code when sending data over the net

Signed-off-by: John Crispin <blogic@openwrt.org>

fix wdt cloexec bug

Signed-off-by: John Crispin <blogic@openwrt.org>

fix up watchdog init code

Signed-off-by: John Crispin <blogic@openwrt.org>