freebsd
wordexp(3): Handle ECHILD from waitpid
If the calling process has used SIG_IGN as handler or set the SA_NOCLDWAIT flag for SIGCHLD, processes will be automatically reaped on exit and calls to waitpid(3) will therefore fail with ECHILD.
We waitpid primarily to reap our child so that the caller does not have to worry about it. ECHILD indicates that there is no child to reap, so we can just treat that as a success and move on.
Thanks for the review!
It would be nice to add a variation of with_SIGCHILD_handler_test to lib/libc/tests/gen/wordexp_test.c to verify the fix works and keeps working.
Makes sense, did a simple copy-paste replacing the handler with SIG_IGN. As I had jbeich verify the original patch I never got around to spinning up a FreeBSD test environment, hopefully CI is good enough to check the test case.
(I considered adding another test for SA_NOCLDWAIT, but it should yield the exact same kernel behavior.)
@jillest, do you plan to MFC to stable/14? I'm trying to figure out how to phase out Sway workaround.
- stable/13 reaches EOL on 2026-04-30. If Sway 1.12 is out before the workaround may need to be rebased. Alternatively, the port/package can be left partially broken near EOL.
- stable/14 reaches EOL on 2028-11-30. Until then Sway may release 1.12, 1.13, 1.14, so rebasing the workaround may become tricky.
- stable/15 is scheduled for 2025-09-05 (plus 2 months to release). No need for the workaround. Unfortunately, many users may delay upgrade from 14.* releases until 15.1 is out.
@jillest, do you plan to MFC to stable/14?
Yes, to stable/14 and stable/13.