hash(9): introduce hashalloc()/hashfree() KPI
AcceptedPublic
Actions

Authored by glebius on Tue, Mar 17, 7:27 PM.

Details

Reviewers

gallatin
pouria

Group Reviewers

Src Committers
Doc Committers

Summary

This is a more extendable version than traditional hashinit(9). It allows
different kinds of slot headers with optional locks.

Implement traditional hashinit()/hashdestroy() on top of it.

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Skipped

Unit

Tests Skipped

Build Status

Buildable 71489
Build 68372: arc lint + arc unit

Event Timeline

glebius created this revision.Tue, Mar 17, 7:27 PM

Herald added subscribers: ziaee, olce, imp. · View Herald TranscriptTue, Mar 17, 7:27 PM

glebius requested review of this revision.Tue, Mar 17, 7:27 PM

Harbormaster completed remote builds in B71468: Diff 173822.Tue, Mar 17, 7:27 PM

glebius added a reviewer: Src Committers.Tue, Mar 17, 7:27 PM

glebius added a reviewer: Doc Committers.

Last minute change before sharing of course had a small mistake :)

Harbormaster completed remote builds in B71469: Diff 173823.Tue, Mar 17, 7:48 PM

This is really clever. What about other lock types (rm / rms, sx, etc)?

I'll see if rmlock(9) and sx(9) can be added and update.

Do you need to validate the enum value is know?

Can the lock and list types be automatically detected with _Generic somehow?

In D55904#1279057, @imp wrote:

Do you need to validate the enum value is know?

Do you mean if somebody passes a numeric instead of one of enum values? I think our compiler flags will forbid that.

Can the lock and list types be automatically detected with _Generic somehow?

I don't see a way to use _Generic facing outside of the KPI. It might be possible to make some use of it internally, I'll look into that.

Add sx(9) and rmlock(9) support.

Harbormaster completed remote builds in B71474: Diff 173831.Tue, Mar 17, 11:58 PM

In D55904#1279059, @glebius wrote:

In D55904#1279057, @imp wrote:

Do you need to validate the enum value is know?

Do you mean if somebody passes a numeric instead of one of enum values? I think our compiler flags will forbid that.

I'm thinking default: would make sure.

Can the lock and list types be automatically detected with _Generic somehow?

I don't see a way to use _Generic facing outside of the KPI. It might be possible to make some use of it internally, I'll look into that.

Sure.

Some additions to the manual page.

Harbormaster completed remote builds in B71489: Diff 173860.Wed, Mar 18, 4:58 PM

gallatin accepted this revision.Wed, Mar 18, 5:28 PM

This revision is now accepted and ready to land.Wed, Mar 18, 5:28 PM

pouria added a subscriber: pouria.Thu, Mar 19, 3:59 PM

pouria added inline comments.Thu, Mar 19, 9:31 PM

sys/kern/subr_hash.c
160–162	Why <= then >>=? why not just <?

pouria added inline comments.Thu, Mar 19, 9:51 PM

share/man/man9/hashinit.9
47–52	Are we going to remove `hashinit_flags` and `hashdestroy` as well? We just updated them to use the new KPI, and we inform users that we plan to obsolete them at the same time. If that's not the case, IMHO, we should include them in the new manual. Otherwise, we should put a deprecation warning above their functions.
sys/kern/subr_hash.c
180	style?
284	style?