This is somewhaht hard to test reliably, but we'll give it a shot. Startup
a sleep(1) daemon with a hefty restart delay. In refactoring of daemon(8),
we inadvertently started dropping SIGTERMs that came in while we were
waiting to restart the child, so we employ the strategy:
- Pop the child sleep(1) first
- Wait for sleep(1) to exit (pid file truncated)
- Pop the daemon(8) with a SIGTERM
- Wait for daemon(8) to exit
The pidfile is specifically truncated outside of the event loop so that we
don't have a kqueue to catch it in the current model.
PR: 277959
Reviewed by: des, markj
Differential Revision: https://reviews.freebsd.org/D47005
We populate the kqueue with all of four kevents: three signal handlers and
one for read of the child pipe. Every time we start the child, we rebuild
this kqueue from scratch for the child and tear it down before we exit and
check if we need to restart the child. As a consequence, we effectively
drop any of the signals we're interested in between restarts.
Push the kqueue out into the daemon state to avoid losing any signal events
in the process, and reimplement the restart timer in terms of kqueue timers.
The pipe read event will be automatically deleted upon last close, which
leaves us with only the signal events that really get retained between
restarts of the child.
PR: 277959
Reviewed by: des, markj
Differential Revision: https://reviews.freebsd.org/D47004
We need to be able to test some more restart behavior that depends on
knowing specifically where we're at (inside the event loop or outside of
the event loop). Truncate the pidfile until the process is restarted to
give the test a clean marker rather than having to add arbitrary delays
and hoping for the best.
Reviewed by: des, markj
Differential Revision: https://reviews.freebsd.org/D47003
Add a new option (-C, --restart-count) to specify the maximum
number of times that the controlled process is restarted if
restart (-r) is restarted.
Reviewed by: kevans
Differential Revision: https://reviews.freebsd.org/D44944
If we're doing restarts, then we must supervise -- the 'R' case simply
got missed.
PR: 278342
Fixes: f907027b49 ("daemon: set supervise_enabled during [..]")
The __builtin_unreachable macro provided by Clang and GCC is a hint to
the compiler used for optimization. The programs work fine even if the
compiler doesn't support it. The sys/cdefs.h has had __unreachable for
9 years (commit 732b31de5d). It expands
to the builtin if it is available. In the rare case that it is
unsupported it expands to a null statement so compilation does not
fail.
Signed-off-by: Collin Funk <collin.funk1@gmail.com>
Reviewed by: imp, freebsd@igalic.co
Pull Request: https://github.com/freebsd/freebsd-src/pull/1117
Not checking for either WIFEXITED(status) or zero result results in
never finishing the loop.
PR: 277764
Reviewed by: kevans (previous version)
Discussed with: Daniel Tameling
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D44401
Re-wrap the read(2) line, and make the listen_child loop more explicit
in intent.
Reported by: kib
Fixes: 6ac7c9f06a ("daemon: remove redundant parameter from [...]")
Looping over lines in the buffer is not needed.
Same effect can be achieved by looking for the last new line.
If found the buffer is guaranteed to have one or more complete lines.
All complete lines are flushed at once with no looping.
Reviewed by: cperciva, kevans
There is no reason for a buffer in listen_child()
to be a static function variable. The buffer and
its position are parts of the daemon state and should
live together with the rest of the state variables.
Reviewed by: cperciva, kevans
Fixed narrowing conversions:
- strtol replaced with strtonum with range check
- read returns ssize_t
- kevent.data explicitly cast to int before passing into strerror
While we we're here:
- Defined and documented maximum restart delay.
- Fixed typo in a comment.
- Remove unused includes
Reviewed by: cperciva, kevans
Apply the following automated changes to try to eliminate
no-longer-needed sys/cdefs.h includes as well as now-empty
blank lines in a row.
Remove /^#if.*\n#endif.*\n#include\s+<sys/cdefs.h>.*\n/
Remove /\n+#include\s+<sys/cdefs.h>.*\n+#if.*\n#endif.*\n+/
Remove /\n+#if.*\n#endif.*\n+/
Remove /^#if.*\n#endif.*\n/
Remove /\n+#include\s+<sys/cdefs.h>\n#include\s+<sys/types.h>/
Remove /\n+#include\s+<sys/cdefs.h>\n#include\s+<sys/param.h>/
Remove /\n+#include\s+<sys/cdefs.h>\n#include\s+<sys/capsicum.h>/
Sponsored by: Netflix
The daemon utility already does its own buffering and retransmits its
child's output line by line. There's no need for stdio to add its own
buffering on top of this.
MFC after: 1 week
Sponsored by: Modirum MDPay
Reviewed by: allanjude
Differential Revision: https://reviews.freebsd.org/D42111
The SPDX folks have obsoleted the BSD-2-Clause-FreeBSD identifier. Catch
up to that fact and revert to their recommended match of BSD-2-Clause.
Discussed with: pfg
MFC After: 3 days
Sponsored by: Netflix
Simply resume waiting for events rather than exiting if we took a signal
here.
This at least fixes running programs under daemon(8) in the face of
suspend/resume, which I suspect hits us with a spurious EINTR rather
than a signal anyways.
Reported and tested by: manu
Fixes: 8935a39932 ("daemon: use kqueue for all events")
Refactor daemon to use kqueue/kevent instead of signals.
This changes allows to simplify the code in several ways:
- the execution flow is now linear, no async events.
- several variables became redundant and got removed.
- all event handling is now concentrated inside of the event loop, which
makes code reading and comprehension easier.
- new kqueuex(2) call is used for CLOEXEC, but maintained closing the
kq fd prior to execve() to ease later MFC
No UX/API changes are intended.
Reviewed by: kevans
Pull Request: https://github.com/freebsd/freebsd-src/pull/701
main() func contained both initialization and main loop logic.
This made certain operations like restarting problematic and
required dirty hacks in form of goto jumps.
This commit moves the main loop logic into daemon_eventloop(),
cleans up main, and makes restart logic clear: daemon_mainloop()
is run in a loop with a restart condition checked at the end.
Reviewed by: kevans
Pull Request: https://github.com/freebsd/freebsd-src/pull/699
Start breaking down big main()
Remove goto exit label and replace it with a function that does cleanup.
Comment re-worded by kevans@.
Pull Request: https://github.com/freebsd/freebsd-src/pull/694
The fact that most of the daemon's state is stored on the stack
of the main() makes it hard to split the logic smaller chunks.
Which in turn leads to huge main func that does a a lot of things.
struct log_params existed because some variables need to be passed
into other functions together.
This change renames struct log_params into daemon_state
and moves the rest of the variables into it. This is a necessary
preparation step for further refactroing.
Reviewed by: imp
Pull Request: https://github.com/freebsd/freebsd-src/pull/687
Now when supervsion mode has it's own variable there is really no
reason to set it separately from the rest of the variables. Move
initialization of supervise_enabled var to the argument processing
switch loop, where it belongs.
Reviewed by: kevans
Pull Request: https://github.com/freebsd/freebsd-src/pull/672
The 'restart' variable was responsible for enablement of restart
behavior and for restart delay. While it may seem convenient it
leads to cluttering the exit/restart logic
Reviewed by: kevans
Pull Request: https://github.com/freebsd/freebsd-src/pull/672
This is not a functional change.
- Clean up whitespace (spaces where there should be tabs)
- Break up lines that are longer than 80
Reviewed by: kevans
Pull Request: https://github.com/freebsd/freebsd-src/pull/672
Since struct log_params already contains logging-related
varaiables, including syslog-related, move remaining
syslog-related variables into struct log_params as well
Reviewed by: kevans
Pull Request: https://github.com/freebsd/freebsd-src/pull/669
Following style(9) and C99 recommendation use bool instead of
int for boolean operations. Also give the variable a more descriptive
name that follows boolean naming convention.
Reviewed by: kevans
Pull Request: https://github.com/freebsd/freebsd-src/pull/669
