After this change vm_page_grab_valid() requires callers to specify ZERO if they want to fill in sparse files/objects. I audited all consumers and a couple wanted it and a couple did not. This makes valid much more consistent with the others although there are still quirks around NOWAIT/WAITFAIL that could be eliminated if necessary.
Diff Detail
Diff Detail
- Lint
Lint Passed - Unit
No Test Coverage - Build Status
Buildable 29884 Build 27704: arc lint + arc unit
Event Timeline
Comment Actions
I think that the behavior of VM_ALLOC_ZERO/has_page should be documented in the comment.