Paths

Table of Contentst

Differential D47047

vm_object: Add wrapper for bumping generation counter
AbandonedPublic
Actions

Authored by bnovkov on Thu, Oct 10, 9:39 PM.

Tags

None

Referenced Files

	F101981454: D47047.diff
	Wed, Nov 6, 3:27 AM

	Unknown Object (File)
	Mon, Nov 4, 7:01 AM

	Unknown Object (File)
	Thu, Oct 31, 9:49 AM

	Unknown Object (File)
	Thu, Oct 24, 5:08 PM

	Unknown Object (File)
	Mon, Oct 21, 9:58 AM

	Unknown Object (File)
	Sun, Oct 20, 1:13 AM

	Unknown Object (File)
	Sat, Oct 19, 11:33 PM

	Unknown Object (File)
	Sat, Oct 19, 10:43 PM

View All 11 Files

Subscribers

Details

Reviewers

markj
alc
kib

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Skipped

Unit

Tests Skipped

Build Status

Buildable 59893
Build 56778: arc lint + arc unit

Event Timeline

bnovkov created this revision.Thu, Oct 10, 9:39 PM

Herald added a subscriber: imp. · View Herald TranscriptThu, Oct 10, 9:39 PM

bnovkov requested review of this revision.Thu, Oct 10, 9:39 PM

bnovkov added a parent revision: D47046: vm_radix: Add iter lookup_le interface.

Harbormaster completed remote builds in B59893: Diff 144609.Thu, Oct 10, 9:39 PM

bnovkov added a child revision: D47048: vm_reserv: Introduce vm_reserv_alloc_npages.Thu, Oct 10, 9:39 PM

bnovkov added a child revision: D47050: vm_page: Add batch page allocation routine.Thu, Oct 10, 9:41 PM

bnovkov removed a parent revision: D47046: vm_radix: Add iter lookup_le interface.

bnovkov removed a child revision: D47048: vm_reserv: Introduce vm_reserv_alloc_npages.

markj added inline comments.Thu, Oct 10, 10:06 PM

sys/vm/vm_object.c
2344	Where is this used? I can't find it in the rest of the patch stack. Why is it ever necessary to bump the counter by more than 1? Consumers (should) only care about whether it changed relative to a previously loaded value, and not about the magnitude of the difference.

bnovkov abandoned this revision.Fri, Oct 11, 3:03 PM

bnovkov added inline comments.

sys/vm/vm_object.c
2344	Where is this used? I can't find it in the rest of the patch stack. It is used in the `vm_radixdone`'s batch counterpart. Why is it ever necessary to bump the counter by more than 1? Consumers (should) only care about whether it changed relative to a previously loaded value, and not about the magnitude of the difference. I agree with you on this, but I eventually opted for replicating the behavior of calling `vm_page_alloc` in a loop, just to avoid inadvertently breaking something unrelated to this change. However, seeing as you pretty much confirmed that this does not matter I'll go ahead and drop the review, thanks!

Revision Contents
Changeset List

Path

Size

sys/

vm/

1 line

7 lines

Diff 144609

sys/vm/vm_object.h

Loading...

sys/vm/vm_object.c

Loading...