Add posix_spawnattr_{get,set}procdescp_np(3)
ClosedPublic
Actions

Authored by kib on Jan 25 2026, 7:20 PM.

Details

Reviewers

markj
asomers
emaste

Group Reviewers

capsicum

Commits

rG080d8ed7dd29: libc: add posix_spawnattr_{get,set}procdescp_np
rG74a2bf1b7a7f: libsys: add pdrfork_thread() on x86

Summary

The function set the address where to store resulting procdesc, which also explcitly request posix_spawn() to use pdfork() (really pdrfork()) instead of fork, and create procdesc for the child.

Since it is needed, also add pdrfork_thread(3) on x86.

The documentation will come later after we agree on the API. I choose this since it allows to unify use of pdrfork, so consumers still use posix_spawn instead of some other function.

An illustration for use of the all new proposed API is there https://reviews.freebsd.org/P686. I used this program to debug the code.

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Skipped

Unit

Tests Skipped

Event Timeline

kib created this revision.Jan 25 2026, 7:20 PM

Herald added a subscriber: imp. · View Herald TranscriptJan 25 2026, 7:20 PM

kib requested review of this revision.Jan 25 2026, 7:20 PM

Rename to spawnattr_setprocdescp_np. IMO pdrfork is the implementation detail there, new name is less cryptic and easier to remember.
Export pdrfork_thread(3) on x86.

Export pdrfork_thread on i386 for real

I didn't read the assembly code (I haven't had to write x86 assembly since University), but otherwise it looks good to me. And thanks for adding the P686 example, too.

lib/libc/gen/posix_spawn.c
56–57	If I'm counting right, then you should reverse the order of these two fields to eliminate alignment padding.

This revision is now accepted and ready to land.Jan 26 2026, 6:24 PM

kib marked an inline comment as done.Jan 26 2026, 7:21 PM

kib added inline comments.

lib/libc/gen/posix_spawn.c
56–57	I do not think it matters. This is an internal structure used without exposing in ABI, and we do not care about it layout. Eliminating padding is not needed.