Branch: refs/heads/main
Home: https://github.com/WebKit/WebKit
Commit: da8482b85ac12c72d6c7c08d8c7e791db6d8642e
https://github.com/WebKit/WebKit/commit/da8482b85ac12c72d6c7c08d8c7e791db6d8642e
Author: Dan Hecht <[email protected]>
Date: 2026-07-17 (Fri, 17 Jul 2026)
Changed paths:
M Source/bmalloc/libpas/src/test/AllocationZeroingTests.cpp
Log Message:
-----------
[libpas] Verify large zeroed allocations under out-of-order page faults
https://bugs.webkit.org/show_bug.cgi?id=319621
rdar://182446375
Reviewed by Keith Miller.
The >= 1 MB zeroed path leaves pages zero-fill-on-demand (madvise(MADV_ZERO)
or the mmap fallback), so a page is only observed zero when first faulted.
The other tests read and write ascending, so they only ever fault pages in
ascending order -- but a hashtable, the workload where this matters, reads
its buckets in hash order, i.e. effectively at random.
Add a test that puts dirty data under a large region (allocate, dirty,
free), reallocates the same size so the zeroed path runs over that dirty
backing, then verifies the buffer reads zero while first-faulting its pages
in ascending, descending, and deterministically shuffled orders, under both
immediate free-list reuse and forced decommit/recommit, across the
threshold and a multi-megabyte size. The reused-buffer check and the shared
reuse case gain an optional FaultOrder parameter; the generic runs keep the
default ascending scan.
* Source/bmalloc/libpas/src/test/AllocationZeroingTests.cpp:
(addAllocationZeroingTests):
Canonical link: https://commits.webkit.org/317390@main
To unsubscribe from these emails, change your notification settings at
https://github.com/WebKit/WebKit/settings/notifications
