While adapting a customers vfs module to 4.11 I noticed some regressions in the byte range locking refactoring that happened in master and 4.11.
There mainly these problems: 1. The SMB1 code has a potential race condition when a local posix lock is blocking the request. 2. The SMB2 code doesn't handle local posix locks at all. 3. The SMB1 code that queues pending locks on an fsp mixes requests and doesn't look at the byte ranges.
Created attachment 15464 [details] Patches for v4-11-test
Comment on attachment 15464 [details] Patches for v4-11-test Review granted offline, pushed to autobuild-v4-11-test.
Pushed to v4-11-test (and included in rc4). Closing out bug report. Thanks!