Page MenuHomeFreeBSD

subr_rangeset: use pctrie_reclaim_callback in remove_all
ClosedPublic

Authored by dougm on Jun 13 2024, 4:57 PM.
Tags
None
Referenced Files
Unknown Object (File)
Fri, Apr 3, 10:12 AM
Unknown Object (File)
Sat, Mar 28, 11:32 AM
Unknown Object (File)
Fri, Mar 27, 11:08 AM
Unknown Object (File)
Sun, Mar 22, 8:05 PM
Unknown Object (File)
Jan 26 2026, 1:49 PM
Unknown Object (File)
Jan 25 2026, 4:22 PM
Unknown Object (File)
Jan 25 2026, 2:07 PM
Unknown Object (File)
Dec 27 2025, 1:27 PM
Subscribers

Details

Summary

Replace the lookup-and-remove loop in rangeset_remove_all with a call to RANGESET_PCTRIE_RECLAIM_CALLBACK, to avoid the repeated lookups.

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

dougm requested review of this revision.Jun 13 2024, 4:57 PM
dougm created this revision.

Logic looks good.

This does make me think, should we arrange to have name##_PCTRIE_RECLAIM_CALLBACK take a function pointer that is typed for the first argument? We would then need a cast to pctrie_cb_t for the calls to begin_cb/resume_cb, which I think both gcc and clang are okay with, but I am not actually sure what the standard says. That would help avoid mixing up the two void *s in the callback type and avoid the need for one of the casts in the callback.

This revision is now accepted and ready to land.Jun 13 2024, 5:39 PM

This does make me think, should we arrange to have name##_PCTRIE_RECLAIM_CALLBACK take a function pointer that is typed for the first argument? We would then need a cast to pctrie_cb_t for the calls to begin_cb/resume_cb, which I think both gcc and clang are okay with, but I am not actually sure what the standard says. That would help avoid mixing up the two void *s in the callback type and avoid the need for one of the casts in the callback.

I'm not opposed, though I'm also not sure about the standard. But I think I'll commit these bits first, and consider that next.