Rafał Miłecki [Thu, 5 Sep 2019 07:20:13 +0000 (09:20 +0200)]
sysupgrade: set UPGRADE_BACKUP env variable
It points to the backup file to use duing sysupgrace process. Right now
it's hardcoded to the /tmp/sysupgrade.tgz. Once all cleanups are in
place "sysupgrade" ubus method should be extended to allow passing any
custom path.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Rafał Miłecki [Thu, 5 Sep 2019 21:07:21 +0000 (23:07 +0200)]
system: fix uninitialized variables in firmware validation code
This fixes:
system.c: In function 'validate_firmware_image':
system.c:403:6: error: 'fd' may be used uninitialized in this function [-Werror=maybe-uninitialized]
if (fd >= 0) {
^
system.c:446:4: error: 'jsobj' may be used uninitialized in this function [-Werror=maybe-uninitialized]
blobmsg_add_object(&b, jsobj);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Fixes:
e990e215e8a3 ("system: add "validate_firmware_image" ubus method")
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Rafał Miłecki [Wed, 4 Sep 2019 09:06:52 +0000 (11:06 +0200)]
system: reject sysupgrade of invalid firmware images by default
This validation step can be bypassed by passing "force" argument. This
is very similar to the /sbin/sysupgrade behavior and --force.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Rafał Miłecki [Fri, 30 Aug 2019 15:46:07 +0000 (17:46 +0200)]
system: reject sysupgrade of broken firmware images
This uses recently added "validate_firmware_image" to validate passed
firmware. If it happens to be invalid and marked as impossible to force
then sysupgrade simply exits with an error.
This change is needed to avoid bricking devices with some totally broken
images.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Rafał Miłecki [Fri, 30 Aug 2019 07:28:34 +0000 (09:28 +0200)]
system: add "validate_firmware_image" ubus method
This new method allows validating firmware image (stored on a device)
using ubus. It uses new executable helper that provides detailed info
about firmware image.
The point of this method is to allow user interfaces provide various
info before starting actual upgrade process.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Rosen Penev [Mon, 26 Aug 2019 22:22:15 +0000 (15:22 -0700)]
sysupgrade: add missing _GNU_SOURCE define (FS#2469)
Fixes compilation problems:
error: implicit declaration of function 'asprintf'; did you mean 'vsprintf'?
[-Werror=implicit-function-declaration]
57 | if (asprintf(&name, "%s%s", prefix, blobmsg_name(option)) <= 0)
| ^~~~~~~~
| vsprintf
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Rafał Miłecki [Fri, 16 Aug 2019 14:14:27 +0000 (16:14 +0200)]
system: support passing "options" to the "sysupgrade" ubus method
Object passed as "options" gets translated into environment variables
UPGRADE_OPT_*. E.g.
"options": { "foo": 5 }
will result in setting UPGRADE_OPT_FOO=5.
This allows stage2 sysupgrade to get options explicitly. So far it was
guessing what to do by checking for existence of some files (e.g.
sysupgrade.tgz).
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Paul Spooren [Mon, 5 Aug 2019 16:07:45 +0000 (18:07 +0200)]
state: fix shutdown when running in a container (FS#2425)
Applies patch from @mikma [0] to fix Docker shutdown. Added detection to
state.c if running in a container or not.
Tested with a x86/64 Docker image.
Fixes FS#2425
CC: Mikael Magnusson <mikma@users.sourceforge.net>
CC: Petr Štetiar <ynezz@true.cz>
[0]: https://github.com/mikma/lxd-openwrt/blob/master/patches/procd-master/0003-docker-fix-problem-stopping-container.patch
Signed-off-by: Paul Spooren <mail@aparcar.org>
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com> [coding style aligment]
Justinas Grauslis [Mon, 8 Jul 2019 08:01:09 +0000 (11:01 +0300)]
procd: check strchr() result before using it
Subtracting some address from NULL does not necessary
results in negative value. It's lower level dependent.
In our case (IPQ4019 + Yocto + meta-openwrt) subtracting
token address from NULL strchr() return value results in
large positive number which causes out-of-boundary memory
access and eventually a segfault.
Signed-off-by: Justinas Grauslis <justinas@8devices.com>
Petr Štetiar [Thu, 30 May 2019 05:53:26 +0000 (07:53 +0200)]
container: fix .dockerenv stat check
applied check wasn't evaluating the existence of the file properly.
Fixes:
7f0f6b2a73f8 ("procd: add docker support")
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Petr Štetiar [Fri, 24 May 2019 11:04:41 +0000 (13:04 +0200)]
hotplug: improve error message during group ownership change
procd currently outputs following error messages:
procd: cannot set group dialout for /dev/ttyw8
procd: cannot set group dialout for /dev/ttyq1
procd: cannot set group dialout for /dev/ttywf
from which it's not clear where the problem is, if it's either getgrnam
or chown failing and why it's failing so this patch adds name of failed
function and its errno.
Ref: https://github.com/openwrt/openwrt/pull/1773#issuecomment-
495555284
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Paul Spooren [Mon, 27 May 2019 09:33:29 +0000 (11:33 +0200)]
procd: add docker support
detects if running in a docker container, which then requires special
treatment of mounts. OpenWrt within Docker is useful for CI testing.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Hans Dedecker [Thu, 9 May 2019 16:01:46 +0000 (18:01 +0200)]
container: include stdbool.h
Fixes following compile issue :
container.h:18:15: error: unknown type name 'bool'
static inline bool is_container() {
^~~~
make[5]: *** [CMakeFiles/init.dir/build.make:154: CMakeFiles/init.dir/initd/zram.c.o] Error 1
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Paul Spooren [Sun, 5 May 2019 16:31:41 +0000 (18:31 +0200)]
procd: detect lxc container and behave accordingly
meaning to not mount some specific parts witch cause trouble.
The patch is based on previous work of @mikma to combine OpenWrt with
lxd[0]. This patch however adds a detection copied from *virt-what* to
check /proc/1/environment for the string "container".
Thanks to @dangowrt for the cleanup.
[0]: https://github.com/containercraft/openwrt-lxd/blob/master/patches/procd-openwrt-18.06/001_lxd_no_mounts.patch
Signed-off-by: Paul Spooren <mail@aparcar.org>
Hans Dedecker [Tue, 30 Apr 2019 19:49:26 +0000 (21:49 +0200)]
instance: dump user and group as well
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Michael Heimpold [Thu, 11 Apr 2019 19:01:07 +0000 (21:01 +0200)]
service: allow setting a dedicated group id
Sometimes is desirable to run a process with a specific group id
instead of the default one which is derived from passwd entry.
However, we still want to initialize supplementary group ids
(including the default one), thus we have to store the specific
one in a dedicated structure element.
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
Michael Heimpold [Thu, 11 Apr 2019 19:01:06 +0000 (21:01 +0200)]
instance: add support for customizable syslog facilities
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
Paul Spooren [Fri, 26 Apr 2019 15:32:45 +0000 (17:32 +0200)]
procd: add SIGPWR as signal
to use procd in LXC containers they have to support SIGPWR to shutdown.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Yashavanth Chowrikoppalu [Wed, 17 Apr 2019 13:21:18 +0000 (15:21 +0200)]
procd: copy the respawn property of new instance
On updating an existing instance, the respawn property is reset
in instance_stop(). Subsequent instance_update() call does not copy
the respawn property from new instance in instance_config_move(). As
a result, the respawning does not happen, if the process is killed
externally.
Signed-off-by: Yashavanth Chowrikoppalu <Yashavanth.hsn@gmail.com>
Kristian Evensen [Sat, 6 Apr 2019 12:28:30 +0000 (14:28 +0200)]
procd: instance: Support deleting stopped instances
procd currently does not support deleting a stopped instance. The reason
is that we return in instance_stop(), if pending is set to false. This
patch adds a new function, instance_delete(), which does the necessary
clean-up of an instance. instance_delete() is called before we return in
instance_stop(), as well as when a process that should not be restarted
has exited in instance_exit().
Signed-off-by: Kristian Evensen <kristian.evensen@gmail.com>
Wojciech Dubowik [Fri, 8 Mar 2019 08:36:12 +0000 (09:36 +0100)]
procd: trace: Define syscall registers for powerpc platform
According to manpage the syscall nr is stored in r0
and return value in r3 for powerpc. Define it so we
can use seccomp and utrace on powerpc.
Signed-off-by: Wojciech Dubowik <Wojciech.Dubowik@neratec.com>
Hauke Mehrtens [Wed, 19 Dec 2018 09:51:57 +0000 (10:51 +0100)]
hotplug.c: Make sure hotplug buffer is NULL terminated
Sets the final byte explicitly to NULL because we later do string
operations on this buffer.
Fixes Coverity issue
1430926 String not null terminated
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Hans Dedecker [Fri, 23 Nov 2018 11:02:45 +0000 (12:02 +0100)]
early: set /tmp permissions explicitly
In case ramfs is used as tmpfs it creates /tmp with permissions 755
which are the the default permissions.
Therefore when mounting tmp set permissions explicitly to 1777
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Acked-by: John Crispin <john@phrozen.org>
Hans Dedecker [Sun, 4 Nov 2018 10:14:31 +0000 (11:14 +0100)]
watchdog: improve seting watchdog timeout and frequency
Due to the watchdog file descriptor check in both watchdog_timeout and
watchdog_frequency it's impossible to set the timeout/frequency via ubus
in case the watchdog was stopped.
Fix this by removing the watchdog file descriptor check in both functions
and by caching locally the set watchdog driver timeout value. The latter
will be used to set the watchdog driver timeout value in case the
watchdog is active or when the watchdog is started.
In addition when parsing the watchdog ubus call check if timeout
attribute is set so the correct value is used when doing the frequency
sanity check.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Chris Blake [Fri, 14 Sep 2018 01:37:45 +0000 (20:37 -0500)]
procd: Add cpu string to board detection
This patch is in relation to
https://github.com/openwrt/openwrt/pull/1307 to help fix architecture
detection in LuCI which was added in
https://github.com/openwrt/luci/commit/
b8f32b6da7accc672c5887e894a861de3f806cf5
Signed-off-by: Chris Blake <chrisrblake93@gmail.com>
Michal Sojka [Mon, 30 Jul 2018 07:32:19 +0000 (09:32 +0200)]
Allow disabling seccomp or changing the whitelist
Without this change, once a service is started with seccomp, it is
impossible to restart it without seccomp or change the whitelist file
name. This commit fixes that. Disabling seccomp is as easy as
commenting out the "procd_set_param seccomp" line in init.d script.
Signed-off-by: Michal Sojka <michal.sojka@cvut.cz>
Michal Sojka [Mon, 30 Jul 2018 07:31:09 +0000 (09:31 +0200)]
trace: Use properly sized type for PTRACE_GETEVENTMSG
Without this, on 64-bit systems, ptrace call corrupts memory because
it stores 64bit value to 32bit pid_t variable.
Signed-off-by: Michal Sojka <michal.sojka@cvut.cz>
John Crispin [Tue, 13 Feb 2018 15:33:48 +0000 (16:33 +0100)]
procd: fix ustream deadlock when there are 0 bytes or no newlines
Signed-off-by: John Crispin <john@phrozen.org>
Felix Fietkau [Thu, 5 Jul 2018 09:13:31 +0000 (11:13 +0200)]
procd: increase watchdog fd_buf storage size to fix gcc8 build error
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Yousong Zhou [Sun, 11 Feb 2018 09:19:11 +0000 (17:19 +0800)]
service: initialize supplementary group ids
We currently only call setgid() with user's primary group id for service
instance processes. In the case of a user belong to multiple groups,
it's natural to also explicitly initialize their supplementary group ids
Fixes FS#988
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
Yousong Zhou [Sun, 11 Feb 2018 09:17:08 +0000 (17:17 +0800)]
service: add func for string config change check
To also prepare for following commits introducing more string value
change detection
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
Rosen Penev [Tue, 23 Jan 2018 20:30:52 +0000 (12:30 -0800)]
procd: get rid of putenv usage.
setenv is prefered according to POSIX. Also allows staticly allocated strings.
Saves 200 bytes when stripped. 432 without.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Hans Dedecker [Mon, 22 Jan 2018 17:51:23 +0000 (18:51 +0100)]
trace: check asprintf() return value
Check asprintf() return value; fixes ignoring return value warnings:
warning: ignoring return value of 'asprintf', declared with attribute
warn_unused_result [-Wunused-result]
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Hans Dedecker [Mon, 22 Jan 2018 17:51:22 +0000 (18:51 +0100)]
trace: add missing limits.h include
Fixes compile issue when using glibc as INT_MAX is reported as
undeclared.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Karl Vogel [Wed, 17 Jan 2018 11:13:15 +0000 (12:13 +0100)]
procd: add timing to start/stop logging
Shows how long an initd task took, for example:
procd: stop /etc/init.d/dropbear running - took 0.088824s
procd: Update service dnsmasq
procd: Update instance dnsmasq::dnsmasq
procd: running /etc/init.d/dnsmasq running
procd: start /etc/init.d/dnsmasq running
procd: stop /etc/init.d/dnsmasq running - took 0.092586s
v2: rename variables, correct time unit
v3: use CLOCK_MONOTONIC_RAW to prevent issues with NTP and adjtime
Signed-off-by: Karl Vogel <karl.vogel@gmail.com>
Rosen Penev [Mon, 25 Dec 2017 22:14:46 +0000 (14:14 -0800)]
procd: Add %m to several functions that return errno.
Might help with debugging. No size impact.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Rosen Penev [Mon, 25 Dec 2017 22:14:45 +0000 (14:14 -0800)]
procd: Remove redundant errno variable in several printf functions.
%m, aka strerror(errno) is already used. No need to repeat it. Saves 280 bytes.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Rosen Penev [Mon, 25 Dec 2017 22:14:44 +0000 (14:14 -0800)]
procd: Replace strerror(errno) with %m.
Saves 1496 bytes from compiled size under glibc. No functional difference.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Rosen Penev [Mon, 25 Dec 2017 22:52:20 +0000 (14:52 -0800)]
procd: Fix minor null pointer dereference.
Null pointer check was placed after assigning values to c.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Nathan Hintz [Fri, 29 Dec 2017 04:48:26 +0000 (04:48 +0000)]
procd: initd: fix path allocation in early_insmod
Noticed that /tmp was not being created on /dev/zram0. This was on
ixp4xx (nslu2) using GCC 6.3 and musl.
The allocation should be using the length of the passed string
(module path), not the size of the pointer to the string.
Signed-off-by: Nathan Hintz <nlhintz@hotmail.com>
Rosen Penev [Tue, 7 Nov 2017 20:05:12 +0000 (12:05 -0800)]
procd: Remove unnecessary memset calls.
Changes allocation to calloc and {} as needed.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Pierre Lebleu [Fri, 10 Nov 2017 10:04:47 +0000 (11:04 +0100)]
service: add data within the service itself
It gives the ability to create firewall data within the
service itself.
Signed-off-by: Pierre Lebleu <pme.lebleu@gmail.com>
Pierre Lebleu [Fri, 10 Nov 2017 09:59:35 +0000 (10:59 +0100)]
service: fix calls to blobmsg_parse()
We should use blobmsg_data() rather than blob_data() and
blobmsg_data_len() rather than blob_len().
Signed-off-by: Pierre Lebleu <pme.lebleu@gmail.com>
John Crispin [Mon, 6 Nov 2017 07:39:09 +0000 (08:39 +0100)]
procd: lower the logging threshold
Signed-off-by: John Crispin <john@phrozen.org>
Michal Sojka [Fri, 3 Nov 2017 21:31:41 +0000 (22:31 +0100)]
service: Start services normally when seccomp is disabled
When service init file declares seccomp support (procd_set_param seccomp),
but procd is compiled without seccomp support, the service should be
started normally, because seccomp-trace and utrace are not available.
Older procd versions decided about whether to start a service in
seccomp sandbox or not based on existence of seccomp whitelist in the
filesystem. This was recently removed (
c8faedc "Do not disable seccomp
when configuration is not found", 2017-09-12) because it could be easy
for attackers to disable seccomp support. This changes is a follow-up
to the mentioned commit. With it, procd decides about whether to use
seccomp sandbox based only on compile-time configuration.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Tested-by: Hans Dedecker <dedeckeh@gmail.com>
John Crispin [Thu, 26 Oct 2017 07:39:52 +0000 (09:39 +0200)]
procd: add missing new lines inside debug code
Signed-off-by: John Crispin <john@phrozen.org>
Pierre Lebleu [Wed, 25 Oct 2017 15:43:16 +0000 (17:43 +0200)]
service: fix service_handle_event array
The array concerning the "config.change" is __VALIDATE_MAX length,
not __DATA_MAX length.
Signed-off-by: Pierre Lebleu <pme.lebleu@gmail.com>
pme.lebleu@gmail.com [Wed, 4 Oct 2017 11:14:16 +0000 (13:14 +0200)]
service: fix SERVICE_ATTR_NAME usage in service_handle_set
The SERVICE_SET_NAME belongs to enum SERVICE_SET_XXXX, not SERVICE_ATTR_NAME.
Signed-off-by: Pierre Lebleu <pme.lebleu@gmail.com>
Karl Palsson [Mon, 25 Sep 2017 12:28:16 +0000 (12:28 +0000)]
instance: properly compare and reload respawn config
respawn configuration variables were not checked for changes, and were
not copied to new instances.
Signed-off-by: Karl Palsson <karlp@etactica.com>
Michal Sojka [Wed, 27 Sep 2017 08:59:54 +0000 (10:59 +0200)]
utrace: Start the tracee only after uloop initialization
Without this, early ptrace stops can be missed because they can happen
before the call to uloop_init().
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:49 +0000 (13:12 +0200)]
utrace: Switch all logging to ulog
This unifies all logs messages produced by utrace and removes
duplicated functionality.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Sun, 24 Sep 2017 14:38:21 +0000 (16:38 +0200)]
utrace: Support non-contiguous syscall numbers
ARM architecture does not have its system call numbers contiguous. So
far, utrace ignored the non-contiguous system calls, but it makes it
difficult to setup seccomp whitelists. This patch adds support for
these extra out-of-range syscalls.
It extends the generated file syscall_names.h to include a few
functions. Now, for ARM this file looks like:
#include <asm/unistd.h>
static const char *__syscall_names[] = {
[280] = "waitid",
[148] = "fdatasync",
...
[252] = "epoll_wait",
[74] = "sethostname",
};
static inline const char *syscall_name(unsigned i) {
if (i < ARRAY_SIZE(__syscall_names))
return __syscall_names[i];
switch (i) {
case 0x0f0001: return "breakpoint";
case 0x0f0003: return "usr26";
case 0x0f0004: return "usr32";
case 0x0f0005: return "set_tls";
case 0x0f0002: return "cacheflush";
default: return (void*)0;
}
}
static inline int syscall_index(unsigned i) {
if (i < ARRAY_SIZE(__syscall_names))
return i;
switch (i) {
case 0x0f0001: return ARRAY_SIZE(__syscall_names) + 0;
case 0x0f0003: return ARRAY_SIZE(__syscall_names) + 1;
case 0x0f0004: return ARRAY_SIZE(__syscall_names) + 2;
case 0x0f0005: return ARRAY_SIZE(__syscall_names) + 3;
case 0x0f0002: return ARRAY_SIZE(__syscall_names) + 4;
default: return -1;
}
}
static inline int syscall_index_to_number(unsigned i) {
if (i < ARRAY_SIZE(__syscall_names))
return i;
switch (i) {
case ARRAY_SIZE(__syscall_names) + 0: return 0x0f0001;
case ARRAY_SIZE(__syscall_names) + 1: return 0x0f0003;
case ARRAY_SIZE(__syscall_names) + 2: return 0x0f0004;
case ARRAY_SIZE(__syscall_names) + 3: return 0x0f0005;
case ARRAY_SIZE(__syscall_names) + 4: return 0x0f0002;
default: return -1;
}
}
#define SYSCALL_COUNT (ARRAY_SIZE(__syscall_names) + 5)
For x86, which does not have extra syscalls, the file looks this way:
#include <asm/unistd.h>
static const char *__syscall_names[] = {
[247] = "waitid",
[75] = "fdatasync",
...
[232] = "epoll_wait",
[170] = "sethostname",
};
static inline const char *syscall_name(unsigned i) {
if (i < ARRAY_SIZE(__syscall_names))
return __syscall_names[i];
switch (i) {
default: return (void*)0;
}
}
static inline int syscall_index(unsigned i) {
if (i < ARRAY_SIZE(__syscall_names))
return i;
switch (i) {
default: return -1;
}
}
static inline int syscall_index_to_number(unsigned i) {
if (i < ARRAY_SIZE(__syscall_names))
return i;
switch (i) {
default: return -1;
}
}
#define SYSCALL_COUNT (ARRAY_SIZE(__syscall_names) + 0)
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:47 +0000 (13:12 +0200)]
utrace: Forward SIGTERM to the traced process
When a service is started with "/etc/init.d/<service> trace" or when
it has seccomp enabled (i.e. runs under seccomp-trace), stopping the
service with "/etc/init.d/<service> stop" stops only the tracer. The
service itself continue executing. This patch ensures that the service
is terminated as well.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:46 +0000 (13:12 +0200)]
utrace: Report ptrace errors
Michal Sojka [Tue, 12 Sep 2017 11:12:45 +0000 (13:12 +0200)]
seccomp: Improve error message
Print "SECCOMP_FILE not specified" instead of "failed to load (null)".
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:44 +0000 (13:12 +0200)]
preload-seccomp: Use proper log level for error messages
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:43 +0000 (13:12 +0200)]
Start seccomp-enabled services via seccomp-trace
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Sun, 24 Sep 2017 22:56:33 +0000 (00:56 +0200)]
seccomp: Log seccomp violations with utrace
Older kernel version shipped by LEDE/OpenWrt contained patch
target/linux/generic/patches-3.18/999-seccomp_log.patch that logged
seccomp violations. For some reason, newer kernels do not have this
patch. Without this kind of logging, it is very hard to setup seccomp
whitelist properly, so this commit modifies utrace to serve as a
logger for seccomp violations.
With this patch, when utrace is executed via seccomp-trace symlink, it
does not trace normal syscalls but only seccomp violations and logs
them to syslog. For example:
seccomp-trace: uci[3955] tried to call non-whitelisted syscall: ftruncate64 (see /etc/seccomp/myservice.json)
Compared to the kernel-based logging, this approach gives users more
information - which json whitelist needs to be extended. This is
especially useful for services, which fork many diverse children such
as shell scripts.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:41 +0000 (13:12 +0200)]
utrace: Use PTHREAD_SEIZE instead of PTHREAD_TRACEME
This makes it easier to handle initial ptrace-stops (after
fork/clone/...), because we don't need to distinguish whether SIGSTOP
is from user or from ptrace. Also execve() does not deliver an extra
SIGTRAP, which we would have to handle.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:40 +0000 (13:12 +0200)]
utrace: Deliver signals to traced processes
Without this change, traced processes do not receive any signal,
because all the signals are "eaten" by the tracer.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:39 +0000 (13:12 +0200)]
utrace: Support tracing multi-threaded processes and vfork
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:38 +0000 (13:12 +0200)]
utrace: Trace processes across forks
Without this change, utrace can trace only a single process. When the
process forks, syscalls of its children do not appear in utrace
output. This is a problem, because seccomp filters are inherited by
children and therefore filters generated by utrace may lack legitimate
syscalls.
This commit enables utrace to trace processes across forks. The
functionality can be demonstrated by the following examples:
utrace /bin/touch /tmp/xxx
produces:
{
"whitelist": [
"rt_sigaction",
"rt_sigreturn",
"exit",
"getuid",
"exit_group",
"utimensat"
],
"policy": 1
}
The command:
utrace /bin/sh -c 'touch /tmp/xxx'
without this commit produces:
{
"whitelist": [
"stat",
"rt_sigaction",
"rt_sigprocmask",
"rt_sigreturn",
"getpid",
"fork",
"exit",
"wait4",
"uname",
"getcwd",
"getuid",
"getppid",
"exit_group"
],
"policy": 1
}
but with this commit, the output is the following:
{
"whitelist": [
"read",
"open",
"close",
"stat",
"fstat",
"mmap",
"mprotect",
"rt_sigaction",
"rt_sigprocmask",
"rt_sigreturn",
"getpid",
"fork",
"execve",
"exit",
"wait4",
"uname",
"fcntl",
"getcwd",
"getuid",
"getppid",
"arch_prctl",
"gettid",
"set_tid_address",
"exit_group",
"utimensat"
],
"policy": 1
}
Note that in addition to utimensat syscall from touch, this output
contains more syscalls than what is in the union of single-process sh
and touch traces. The reason is that single-process traces do not
include syscalls from dynamic linker (due to preload trick), but the
trace of forked processes includes the dynamic linker syscalls. This
is unavoidable, because dynamic linker of the forked processes will be
subject to seccomp filters of the parent process.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:37 +0000 (13:12 +0200)]
utrace: Sort syscalls by number of invocations
seccomp and service jailing announce email [1] mentioned that "utrace
tool will sort the syscalls by the number of invocations". The code
did not do that until this commit.
[1] https://lists.openwrt.org/pipermail/openwrt-devel/2015-March/032197.html
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:36 +0000 (13:12 +0200)]
Update trace attribute
When a service is started for the first time without trace
attribute (e.g. during boot), then it was impossible to restart it in
tracing mode (/etc/init.d/service trace). This is fixed here.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:35 +0000 (13:12 +0200)]
Do not disable seccomp when configuration is not found
Previously, when seccomp configuration file for a service was not
found, the service was started without seccomp. I consider this
potential attack vector.
With this change, procd starts the service as if the configuration
existed but the service fails in libpreload-seccomp.so, because the
configuration cannot be loaded. This is announced in the syslog.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:34 +0000 (13:12 +0200)]
utrace: Fix off-by-one errors
This fixes two errors:
1) memcpy() copies envc elements starting from index 1, so the number
of elements in target array should be envc + 1. But only envc was
allocated.
2) If original environment envp is empty, i.e. it contains only a NULL
element, the while loop misses it.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Tue, 12 Sep 2017 11:12:33 +0000 (13:12 +0200)]
utrace: Fix environment initialization
We want to copy the existing environment instead of the new one to
itself. Other bugs in this code are fixed in the next commit.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
John Crispin [Tue, 22 Aug 2017 14:43:54 +0000 (16:43 +0200)]
procd: send ubus notify when a config file has changed
Signed-off-by: John Crispin <john@phrozen.org>
Hans Dedecker [Tue, 8 Aug 2017 07:49:26 +0000 (09:49 +0200)]
watchdog: fix inline watchdog_get_magicclose function prototype
Fix procd compilation issue which occurs when DISABLE_INIT is set as the
inline watchdog_get_magicclose function prototype was wrong.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Michal Sojka [Thu, 6 Jul 2017 20:25:16 +0000 (22:25 +0200)]
procd: Do not leak pipe file descriptors to children
Without this change, a process started by procd can have access to
stdout/err of processes started by procd before.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Matthias Schiffer [Thu, 20 Jul 2017 19:35:14 +0000 (21:35 +0200)]
system: return ubus error when sysupgrade_exec_upgraded() has failed
When sysupgrade_exec_upgraded() is successful, it will replace procd with
upgraded, so it won't ever return (making the ubus call hang forever, which
is fine as stage2 will kill all processes anyways).
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Florian Fainelli [Sat, 15 Jul 2017 17:59:51 +0000 (10:59 -0700)]
upgraded: Check chroot() return value
Check the chroot() return value, fixes unused return value warnings/errors:
procd-2017-06-22-
e5e99c46/upgraded/upgraded.c:78:8: error: ignoring return
value of 'chroot', declared with attribute warn_unused_result
[-Werror=unused-result] chroot(".");
Fixes:
056d8ddda8d5 ("upgraded: link dynamically, chroot during exec")
Reviewed-by: Matthias Schiffer <mschiffer@universe-factory.net>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Florian Fainelli [Sat, 15 Jul 2017 17:51:20 +0000 (10:51 -0700)]
init: Check chroot return value in sysupgrade_exec_upgraded()
chroot() can fail and its return value should be checked against, in that case
do an exit() since this is a fatal condition that we cannot recover from.
Fixes:
63789e51ed91 ("init: add support for sysupgrades triggered from preinit")
Reviewed-by: Matthias Schiffer <mschiffer@universe-factory.net>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Matthias Schiffer [Thu, 13 Jul 2017 11:18:00 +0000 (13:18 +0200)]
upgraded: improve error handling
* exit with code 1 instead of unusual -1 in the parent process
* exit using _exit() when child exec fails
* fix fork/exec error messages
* only uloop_run() after successful child fork (uloop_end() before
uloop_run() doesn't have any effect, so uloop_run() would hang forever)
* minor code cleanup
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Matthias Schiffer [Wed, 12 Jul 2017 22:04:49 +0000 (00:04 +0200)]
upgraded: register stage2 process in uloop as intended
Make the process callback effective, so an exit of state2 will trigger a
reboot.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Hans Dedecker [Mon, 19 Jun 2017 09:14:27 +0000 (11:14 +0200)]
watchdog: add support for starting/stopping kernel watchdog
Extend the ubus watchdog cmd with the parameter magicclose; when set and
in case the stopped parameter is enabled the kernel watchdog will be stopped
by first sending the magic character 'V' followed by a close of the watchdog
fd. In case stopped is set to disabled the watchdog fd will be created again.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Acked-by: John Crispin <john@phrozen.org>
Daniel Golle [Sun, 18 Jun 2017 22:53:00 +0000 (00:53 +0200)]
system: introduce new attribute board_name
Instead of modifying the already defined 'model' attribute rather
introduce a new attribute board_name to return either
/tmp/sysinfo/board_name or /proc/device-tree/compatible.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Hans Dedecker [Fri, 16 Jun 2017 14:09:01 +0000 (16:09 +0200)]
preinit: define _GNU_SOURCE
In uClibc getdelim is only defined if _GNU_SOURCE is defined; fix
compilation issue by defining _GNU_SOURCE
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Florian Fainelli [Wed, 31 May 2017 21:00:34 +0000 (14:00 -0700)]
upgraded: cmake: Find and include uloop.h
Add a CMake FIND_PATH and INCLUDE_DIRECTORIES searching for
libubox/uloop.h. Some external toolchains which do not include standard
locations would fail to find the header otherwise.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Alexander Couzens [Sun, 11 Jun 2017 09:52:47 +0000 (11:52 +0200)]
hotplug: fix a memory leak in handle_button_complete()
Found-by: Coverity Scan #1412460
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Alexander Couzens [Sun, 11 Jun 2017 09:49:00 +0000 (11:49 +0200)]
service/service_stopped(): fix a use-after-free
Found-by: Coverity Scan #1412548
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Daniel Golle [Thu, 8 Jun 2017 20:21:27 +0000 (22:21 +0200)]
system: return legacy board name
The system board call returns the verbose model string instead of the
board name. Unfortunately we have not yet estabkushed clear convention
regarding device-tree 'compatible' or 'model' property and sysupgrade
image name (same accounts for scraping /proc/cpuinfo on legacy
targets). This is odd as the idea was to return information needed to
identify the right sysupgrade image. On most targets we got a large
shell-script which translates either /proc/cpuinfo or the 'model'
property back into the board name used here. Hence
/tmp/sysinfo/board_name should be returned instead of
/tmp/sysinfo/model for the board ubus call as well.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Matthias Schiffer [Tue, 30 May 2017 05:23:57 +0000 (07:23 +0200)]
upgraded: define __GNU_SOURCE
It is required on non-musl libcs for O_DIRECTORY.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Matthias Schiffer [Mon, 29 May 2017 21:39:53 +0000 (23:39 +0200)]
rcS: add missing fcntl.h include
Fixes:
d42b21ed9305 "procd/rcS: Use /dev/null as stdin"
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Matthias Schiffer [Sun, 23 Apr 2017 22:40:27 +0000 (00:40 +0200)]
init: add support for sysupgrades triggered from preinit
This will allow to add support for sysupgrades via upgraded from failsafe
mode.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Matthias Schiffer [Sun, 23 Apr 2017 17:06:12 +0000 (19:06 +0200)]
Remove code that has become unnecessary after sysupgrade changes
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Matthias Schiffer [Sun, 23 Apr 2017 17:04:25 +0000 (19:04 +0200)]
upgraded: add support for passing a "command" argument to stage2
This allows us to make use of upgraded in "snapshot convert" as well.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Matthias Schiffer [Sun, 23 Apr 2017 00:28:13 +0000 (02:28 +0200)]
upgraded: link dynamically, chroot during exec
The chroot ensures we don't reference anything on the rootfs and is
reverted after the upgraded exec. While we're at it, also improve error
handling a bit.
This change also required changes to sysupgrade, as the dynamically linked
version is expected at a different location, and libraries need to be made
available.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Matthias Schiffer [Fri, 21 Apr 2017 18:06:59 +0000 (20:06 +0200)]
system: always support staged sysupgrade
In preparation for switching all targets to the staged sysupgrade
mechanism, upgraded is always built, and the "nandupgrade" ubus method is
renamed to "sysupgrade".
To make the migration easier, support for the old name "nandupgrade" and
the "upgrade" method that will become unused with the staged sysupgrade is
retained for now.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Michal Sojka [Sun, 30 Apr 2017 07:51:20 +0000 (09:51 +0200)]
procd/rcS: Use /dev/null as stdin
This change ensures that /etc/init.d/* scripts are started with
/dev/null as stdin. This is useful in cases where an init.d script
reads (e.g. by mistake) from stdin, which a user can perceive as if
some characters typed into shell on serial console are "eaten" by
something else (i.e. by the init.d script running on background). This
is very annoying, because each character needs to be pressed several
times before it appears on the screen.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
John Crispin [Wed, 5 Apr 2017 06:58:22 +0000 (08:58 +0200)]
service/instance: add an auto start option
this allows us to register services with procd but not auto start them. An
additional ubus call is required to start the service.
Signed-off-by: John Crispin <john@phrozen.org>
Michal Sojka [Sat, 11 Mar 2017 00:48:16 +0000 (01:48 +0100)]
procd: Log initscript output prefixed with script name
It helps with debugging of initscript problems.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Sun, 12 Mar 2017 01:11:39 +0000 (02:11 +0100)]
procd: Don't use syslog before its initialization
When procd starts a rcS script, it captures its stdout and stderr and
logs them via syslog(). The problem with that is that the rest of
procd code uses ulog rather than syslog() directly and ulog_open()
doesn't call openlog() immediately, but only after something is logged
with ulog(). This lazy calling of openlog() can result in the
following unwanted behavior:
1) When rcS's stdout/err is logged via syslog(), the log identifier is
not set yet (due to openlog() not called) and so the log message
lacks information about source.
2) procd can also log stdout/err from services. When a message from a
service needs to be logged, ulog_open() is called to change the log
identifier to match the service name/PID. After logging the service
messages, ulog_open() is called again the change the identifier
back to "procd". The lazy call to openlog() means that the messages
logged directly with syslog() will be logged with the
identification of the previously logged service and not of the rcS
script that produced the message.
Both problems are fixed by replacing direct call to syslog() with
ULOG_NOTE, which automatically calls openlog() if needed.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Michal Sojka [Sat, 11 Mar 2017 00:48:14 +0000 (01:48 +0100)]
procd: Add missing \n in debug message
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Alin Năstac [Mon, 27 Feb 2017 09:08:34 +0000 (10:08 +0100)]
procd: service gets deleted when its last instance is freed
This fixes the following regression introduced in commit
961dc692aff7457f874bce61f8e766514edcf794:
1) reboot using the following configuration
root@OpenWrt:~# uci show system.ntp
system.ntp=timeserver
system.ntp.enable_server='0'
system.ntp.use_dhcp='1'
system.ntp.dhcp_interface='wan'
root@OpenWrt:~# uci show network.wan
network.wan=interface
network.wan.proto='dhcp'
network.wan.ifname='eth4'
network.wan.reqopts='1 3 6 15 33 42 51 121 249'
2) if obtained DHCP lease has an option 42 sysntpd service will have an
instance
3) run "ifup wan"
4) although the same DHCP lease was obtained, sysntpd would be stopped
Because sysntpd service is deleted when last instance is freed, its triggers
will also be released. Without these triggers in place, sysntpd will not be
reloaded when a new DHCP lease containing option 42 will be received.
Signed-off-by: Alin Nastac <alin.nastac@gmail.com>
Nathan Hintz [Mon, 20 Feb 2017 20:39:11 +0000 (20:39 +0000)]
procd: update modprobe path
Commit
81aeba9b7f619ee1af1a64f355ae8001fa147d03 in LEDE source.git moved
modprobe to the "/sbin" directory. Update procd with the new path.
Signed-off-by: Nathan Hintz <nlhintz@hotmail.com>
Jurgen Van Ham [Fri, 10 Feb 2017 06:55:23 +0000 (07:55 +0100)]
procd: add cancel_timeout on rc scripts when a runtime_timeout is specified
Enable procd to restrict the execution time of a rc scripts during shutdown, even
when a script ignores the SIGTERM signal by insisting after 10s with a SIGKILL.
Signed-off-by: Jurgen Van Ham <juvanham.tc@gmail.com>
Alin Năstac [Thu, 9 Feb 2017 10:02:53 +0000 (11:02 +0100)]
procd: stop service using SIGKILL if SIGTERM failed to do so
SIGKILL is sent if instance process is still running after
<term_timeout> seconds after SIGTERM has been sent. To prevent
another daemon process being launched before old process dies,
the instance is kept until SIGCHLD confirms that service has
been stopped.
Signed-off-by: Alin Nastac <alin.nastac@gmail.com>
Felix Fietkau [Sat, 11 Feb 2017 14:43:36 +0000 (15:43 +0100)]
add missing includes
Including sys/sysmacros.h is now necessary for makedev() on glibc 2.25.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
John Crispin [Wed, 8 Feb 2017 11:02:10 +0000 (12:02 +0100)]
ujail: fix signal forwarding
Signed-off-by: John Crispin <john@phrozen.org>