Simplify handling of superpages in pmap_clear_modify()
ClosedPublic
Actions

Authored by alc on Jul 22 2019, 3:19 PM.

Details

Reviewers

kib
markj
manu

Commits

rS350335: Simplify the handling of superpages in pmap_clear_modify(). Specifically,

Summary

If a demotion succeeds, then all of the 4KB page mappings within the superpage-sized region must be valid, so there is no point in testing the validity of the 4KB page mapping that is going to be write protected.

Deindent the code.

Diff Detail

Lint

Lint Skipped

Unit

Tests Skipped

Event Timeline

alc created this revision.Jul 22 2019, 3:19 PM

Herald added subscribers: emaste, andrew. · View Herald TranscriptJul 22 2019, 3:19 PM

Peter, could you please test this patch on amd64 and i386.

Sure, happy to.

I finished the full stress2 tests on amd64 without any issues.

On i386 I ran into this problem, which seems unrelated to me. Unfortunately I did not get a dump as the debugger was looping.
I made a short attempt to recreate the problem, but without any luck.
Here's the partial console log: https://people.freebsd.org/~pho/stress/log/alc010.txt
I still need to run 170 tests on i386 to complete a full test.

In D21027#456866, @pho wrote:

I finished the full stress2 tests on amd64 without any issues.

Thank you.

On i386 I ran into this problem, which seems unrelated to me. Unfortunately I did not get a dump as the debugger was looping.
I made a short attempt to recreate the problem, but without any luck.
Here's the partial console log: https://people.freebsd.org/~pho/stress/log/alc010.txt
I still need to run 170 tests on i386 to complete a full test.

I agree that that problem seems unrelated.

The i386 tests finished with one further problem and an incomplete report: https://people.freebsd.org/~pho/stress/log/alc011.txt

In D21027#457081, @pho wrote:

The i386 tests finished with one further problem and an incomplete report: https://people.freebsd.org/~pho/stress/log/alc011.txt

That also seems unrelated. Thanks.

alc added reviewers: kib, markj.Jul 25 2019, 4:57 AM

kib accepted this revision.Jul 25 2019, 7:26 AM

This revision is now accepted and ready to land.Jul 25 2019, 7:26 AM

markj accepted this revision.Jul 25 2019, 3:30 PM

Closed by commit rS350335: Simplify the handling of superpages in pmap_clear_modify(). Specifically, (authored by alc). · Explain WhyJul 25 2019, 10:03 PM

This revision was automatically updated to reflect the committed changes.

alc added a commit: rS350335: Simplify the handling of superpages in pmap_clear_modify(). Specifically,.

Herald added a reviewer: manu. · View Herald TranscriptJul 25 2019, 10:03 PM

Herald added a subscriber: imp. · View Herald Transcript

hi,
I stopped by this code review while I was troubleshooting things for PR 242137.

as far as we understood, the issue I reported on that bug happens only if I use a particular hardware, but others may also be affected (not sure though).

should some of you guys can drop a quick look on that PR would be awesome; I'm pretty much interested on getting a better picture of all that.

In D21027#506785, @egypcio wrote:

hi,
I stopped by this code review while I was troubleshooting things for PR 242137.

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=242137#c106

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=242137#c121 (*)

as far as we understood, the issue I reported on that bug happens only if I use a particular hardware, but others may also be affected (not sure though).

should some of you guys can drop a quick look on that PR would be awesome; I'm pretty much interested on getting a better picture of all that.

The thread in that PR is quite long and hard to understand. Can you provide a brief summary? From comment 126 it sounds like you are able to trigger some ntpd problem when both r352517 and r352519 are applied, but not when either commit is reverted?