vm_object_coalesce(): return swap reservation back if overcharged
ClosedPublic
Actions

Authored by kib on Mon, Dec 22, 12:51 PM.

Details

Reviewers

alc
markj

Commits

rG7685aaea8850: vm_object_coalesce(): return swap reservation back if overcharged

Summary

It is possible for both vm_map_insert() and vm_object_coalesce() to charge
both for the same region.  The issue is that vm_map_insert() must charge
in advance to ensure that the mapping would not exceed the swap limit,
but then the coalesce might decide to extend the object, and already
(partially) backs the mapped region.

Handle this by passing to vm_object_coalesce() exact information about
the charging mode of the extending range 'not charging', 'charged' using
flags instead of simple boolean.  In vm_object_coalesce(), detect
overcharge and undo it if needed.

Note that this relies on vm_object_coalesce() call being the last action
in vm_map_insert() before extending the previous map entry.

Reported by:    pho

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Skipped

Unit

Tests Skipped

Event Timeline

kib created this revision.Mon, Dec 22, 12:51 PM

Herald added a subscriber: imp. · View Herald TranscriptMon, Dec 22, 12:51 PM

NB: this is still being tested by Peter, I decided to go ahead and publish the review as an early exposure

kib added a parent revision: D54263: vm_object_coalesce(): do not account holes twice.Mon, Dec 22, 12:52 PM

I ran the full set of stress2 tests, without seeing any issues.

Ping? This is a follow-up fix after D54263, and I want to commit them together.

In D54338#1243187, @kib wrote:

Ping? This is a follow-up fix after D54263, and I want to commit them together.

Sorry, I am catching up on reviews today.

Does the patch need to be rebased? I can't apply it to main, the last two hunks in vm_object.c don't apply.

In D54338#1243197, @markj wrote:

In D54338#1243187, @kib wrote:

Ping? This is a follow-up fix after D54263, and I want to commit them together.

Sorry, I am catching up on reviews today.

Does the patch need to be rebased? I can't apply it to main, the last two hunks in vm_object.c don't apply.

It is on top of D54263

In D54338#1243198, @kib wrote:

In D54338#1243197, @markj wrote:

In D54338#1243187, @kib wrote:

Ping? This is a follow-up fix after D54263, and I want to commit them together.

Sorry, I am catching up on reviews today.

Does the patch need to be rebased? I can't apply it to main, the last two hunks in vm_object.c don't apply.

It is on top of D54263