aboutsummaryrefslogtreecommitdiffstats
path: root/src
Commit message (Collapse)AuthorAgeFilesLines
* Fix s6-svscanboot build with nsssLaurent Bercot7 days1-0/+2
|
* Fix s6-instance-maker invocation of s6-svscanLaurent Bercot7 days1-2/+3
|
* Define S6_FILETYPE_TEMPLATELaurent Bercot13 days2-1/+2
|
* Add s6-svc -l and reload-signalLaurent Bercot2026-06-093-11/+27
|
* Add s6-svscanboot docLaurent Bercot2026-06-081-6/+7
|
* Prepare for 2.15.1.0; add s6-svscanbootLaurent Bercot2026-06-043-1/+280
| | | | Also fix links on why.html. s6-svscanboot doc coming soon.
* s6-notify-socket-from-fd: send barrier in a separate sd_notify messageLaurent Bercot2026-05-271-17/+23
|
* And don't assume it's 0-terminatedLaurent Bercot2026-05-231-1/+1
|
* Better stralloc contract in s6_supervise_unlink_namesLaurent Bercot2026-05-231-1/+1
| | | | | | Assume sa->len is 0 at invocation of registerit(). Reset it to 0 before exiting it. So sa is always available as scratch space.
* s6-supervise: remove IOPAUSE_EXCEPT special handlingLaurent Bercot2026-05-171-2/+0
| | | | MacOS is doing stupid things again
* Prepare for 2.15.0.1; fix s6-accessrules-cdb-from-fs args numberingLaurent Bercot2026-05-151-8/+10
|
* Update max length of a file in the servicedirLaurent Bercot2026-05-071-1/+1
|
* MSG_NOSIGNAL needs posixishard.h for Solaris buildLaurent Bercot2026-04-291-0/+2
|
* Send SIGCONT after SIGKILL, workaround for stupid kernels.Laurent Bercot2026-04-281-2/+10
|
* Add flag-timeout-killpg to the list of known filesLaurent Bercot2026-04-231-0/+1
|
* Add flag-timeout-killpg for killing the whole pgroup after timeout-killLaurent Bercot2026-04-231-1/+2
|
* s6-notify-socket-from-fd: use BARRIER=1Laurent Bercot2026-04-101-21/+66
|
* Add s6-notify-fd-from-socket and s6-notify-socket-from-fdLaurent Bercot2026-04-0915-9/+290
| | | | | Also add .gitkeep files, remove old include-local/s6lockd.h, update a little bit of doc, make s6-ftrigrd pipe-only
* s6-svscan bugfix: -L impacts namemax, not maxLaurent Bercot2026-03-311-1/+1
|
* C operator precedence is hardLaurent Bercot2026-03-281-1/+1
|
* Booleans are hardLaurent Bercot2026-03-281-1/+1
|
* s6-setsid bugfix: only check ctty when actually calling tcsetpgrpLaurent Bercot2026-03-281-1/+1
|
* Merge branch '2.15.0.0'Laurent Bercot2026-03-1840-996/+825
|\
| * More fixes, ftrigr_string addition, doc updateLaurent Bercot2026-03-1811-44/+35
| |
| * s6-svscan trivial fix; s6-svunlink bugfix; start prep for 2.15.0.0Laurent Bercot2026-03-172-5/+11
| | | | | | | | | | | | | | | | s6_supervise_unlink_names can spawn a s6-ftrigrd to wait for an 'x' event. This creates a fifodir under $scandir/$name/event. But after the unlink, $scandir/$name doesn't exist anymore, so s6-ftrigrd fails to clean it up. To fix that, we subscribe with the realpath of $scandir/$name instead.
| * Refactor ftrig using sass; clean up API a bit, update its usersLaurent Bercot2026-03-1641-998/+830
| |
* | s6-svscan: remove broken optimization for inactive removalLaurent Bercot2026-02-201-11/+3
|/
* Also remove it from s6/s6.hLaurent Bercot2026-02-151-1/+0
|
* Remove mentions to s6lock from doc and code.Laurent Bercot2026-02-053-81/+0
| | | | | | s6lock has been deleted in the previous major update. There were still traces of it in the doc, and lock.h still existed, although non-functional. Clean all that up.
* Add initial cleanup to s6-svscanLaurent Bercot2026-01-241-2/+21
|
* Invoke s6-ftrigrd with a full environmentLaurent Bercot2026-01-201-2/+3
| | | | That helps when LD_LIBRARY_PATH is necessary.
* s6-svscan can be the last living piece of a deleted s6 version and still workLaurent Bercot2025-12-241-1/+1
|
* releng fixesLaurent Bercot2025-12-231-0/+1
|
* s6-softlimit: add -P option for RLIMIT_RTPRIOAlyssa Ross2025-11-091-1/+6
| | | | | | | | Bash ulimit uses -r for this, but in s6-softlimit that was already taken. I could have used -R, but Bash uses that for RLIMIT_RTTIME, so I didn't want to claim that one and create a similarly difficult decision for somebody adding that in future. I decided to use -P (for "priority") as the least bad still-available option.
* POSIX guarantees that pid == pgid for session leaders!Laurent Bercot2025-11-075-47/+23
|
* Move daemontools/runit wrapper support hereLaurent Bercot2025-11-046-0/+1039
|
* stupid s6-svc bug fixLaurent Bercot2025-11-041-1/+1
|
* Improve s6-svc; prepare for 2.14.0.0Laurent Bercot2025-11-021-12/+19
|
* s6-supervise: try getsid instead of getpgid, see what OpenBSD saysLaurent Bercot2025-10-281-2/+2
|
* s6-background-watch: don't forget selfpipe_initLaurent Bercot2025-10-271-0/+2
|
* Obviously BSD stuff needs nonposix.hLaurent Bercot2025-10-231-0/+2
|
* Revert: actual fix is in skalibsLaurent Bercot2025-10-221-1/+0
|
* Add PTHREAD_LIB to s6-setlock's dependencies, because FreeBSDLaurent Bercot2025-10-221-0/+1
|
* More keventbridge fixesLaurent Bercot2025-10-221-6/+3
|
* Fix keventbridge_start invocationLaurent Bercot2025-10-221-1/+2
|
* Change s6-background-watch to use keventbridgeLaurent Bercot2025-10-221-70/+17
|
* Support procctl in s6-background-watchLaurent Bercot2025-10-221-13/+25
|
* Fix storage of struct thinfo_s in s6-background-watchLaurent Bercot2025-10-221-7/+8
|
* Add some doc, improve s6-background-watchLaurent Bercot2025-10-222-16/+25
|
* Add s6-background-watch (untested)Laurent Bercot2025-10-202-0/+301
|