>> Have you examined whether "docker" supports POSIX advisory locks and sync operations correctly? > I must say I haven't. That being said, it seems like SQLite is using docker in the context of OSS-fuzz. And this OSS-Fuzz thing is multithreaded / multiple connections to the same database file, or just a single connection to a single database file? I would point out that SQLite3 works perfectly with no locking whatsoever in a "single connection to a single database" scenario even in the face of defective hardware (ie, that do not handle sync operations reliably) and filesystems (that do not support locking reliably) -- so comparisons of "multithreaded apples with mutilple stems" to "single-threaded apples with one stem" are not like comparisons and no inference can be drawn from the latter with respect to the former that involve "multiple stems" or "multiple cores" in the same apple. Of course, if OSS-Fuzz *did* do testing of multi-threaded access by multiple connections to the same on-disk database that was hosted on an "overlay" filesystem, then it might be a valid comparison. Of course, according to the OSS-Fuzz docs there can be no comparable comparison since OSS-Fuzz does not allow access to "files" on the overlay filesystem (only /tmp) and the documentation says nothing about whether it tests multiple access via either multiple threads or multiple connection access to the same shared database. The documentation only speaks to single thread access using a single connection to a non-multiple-access database.