Stress testing of multichannel reconnect in Microsoft uncovered a few bugs in multichannel reconnect codepath (particularly when reconnects happen in parallel). I have fixed those in the following commits: https://github.com/sprasad-microsoft/smb3-kernel-client/commit/a87bb308b7d126b522d4390dbf37f63e743133ac.patch https://github.com/sprasad-microsoft/smb3-kernel-client/commit/0a95e989840d2617940ad4670c52b43646b491ad.patch https://github.com/sprasad-microsoft/smb3-kernel-client/commit/5640b4cd537b70a690e3c1a6aa22afced71c8c87.patch Also fixed a couple of regressions: https://github.com/sprasad-microsoft/smb3-kernel-client/commit/30a563a20e00f99899b703f32feb30793b04bfcb.patch https://github.com/sprasad-microsoft/smb3-kernel-client/commit/0c1c85c0720ef8fa3bac3819315e8d9311926915.patch All the above are critical fixes that we should probably CC stable. And a minor fix: https://github.com/sprasad-microsoft/smb3-kernel-client/commit/be94b055601c48f4156e48da7b282d51970bad07.patch I've tried to break up the changes into separate patches. Please review. -- Regards, Shyam
