kern/proc: expose reaper PID and subtree root in struct kinfo_proc

[rikka.goering_outlook.de]

Summary

Expose process reaper metadata through struct kinfo_proc so userland can reconstruct reaper hierarchies from kern.proc.all without adding a new procctl(2) operation.
The patch adds two pid_t fields consuming the two remaining KI_NSPARE_INT slots (KI_NSPARE_INT 2->0), without changing sizeof(struct kinfo_proc):

• ki_reaper: PID of the owning reaper process
• ki_reapsubtree: PID of the direct child of the reaper that roots the subtree the process belongs to

Updated components:

• sys/sys/user.h: add the new kinfo_proc fields
• sys/kern/kern_proc.c: populate the fields in fill_kinfo_proc_pgrp()
• lib/libkvm/kvm_proc.c: update crash dump consumers
• sys/compat/freebsd32/freebsd32.h and freebsd32_kinfo_proc_out(): update 32-bit compat layout and conversion

The implementation follows the existing kern.proc.all snapshot model. The values are populated while the sysctl path holds proctree_lock shared, and NULL reaper pointers fall back to NO_PID.
Tested locally with a small userland program covering baseline state, PROC_REAP_ACQUIRE, direct-child subtree tracking, and reparenting after PROC_REAP_RELEASE.

Bugzilla PR: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=293871

[kib]

sys/kern/kern_proc.c

1238

Is it ever possible that p_reaper == NULL?

sys/sys/user.h

190–192

KI_NSPARE_INT is zero, and the field does not make sense anymore. Also I suspect that the member array with zero length is not valid C, but accepted by both gcc and clang.

Consuming the last int spare makes me nervous. I think we should carve some space from ki_sparestrings for at least one spare int.

[rikka.goering_outlook.de]

kern/proc: address review feedback on kinfo_proc reaper fields

• Remove unnecessary NULL guard around p_reaper in fill_kinfo_proc_pgrp(); grep confirms p_reaper is always non-NULL for any live process
• Restore KI_NSPARE_INT to 2 by carving 8 bytes from ki_sparestrings (46 -> 38), avoiding a zero-length array and preserving spare room for future growth

[kib]

lib/libkvm/kvm_proc.c

369

p_reaper should be not NULL even in core dumps

sys/sys/user.h

190–192

I would move ki_spareints[] up, right after ki_sparestrings[]

[rikka.goering_outlook.de]

kern/proc: address review feedback on kinfo_proc reaper fields

• Removed leftover NULL guard around p_reaper in lib/libkvm/kvm_proc.c
• Re-ordered ki_spareints position in struct kinfo_proc  and in struct kinfo_proc32

[kib]

lib/libkvm/kvm_proc.c

369

Please move the rproc declaration to around line 123

sys/kern/kern_proc.c

1238

It seems that the indent is incorrect?

[rikka.goering_outlook.de]

kern/proc: address review feedback on kinfo_proc reaper fields

• Fixed indentation error in sys/kern/kern_proc.c

[kib]

lib/libkvm/kvm_proc.c

369

Please fix this. Then mail me the output from the 'git format-patch -1' with the proper author set.

[rikka.goering_outlook.de]

kern/proc: expose reaper PID and subtree root in struct kinfo_proc

• moved rproc struct declaration to similar struct definitions in kvm_proc.c