With the first four netfs patches + 6.14-rc7 running our standard regression tests to Azure with multichannel, I saw a couple of unexpected failures. 23dbb8c5ffaf netfs: Fix netfs_unbuffered_read() to return ssize_t rather than int 78067c2bd7f9 netfs: Fix rolling_buffer_load_from_ra() to not clear mark bits 6be938f6c265 netfs: Call `invalidate_cache` only if implemented 5fb49e95f2c2 netfs: Fix collection of results during pause when collection offloaded See: http://smb311-linux-testing.southcentralus.cloudapp.azure.com/#/builders/5/builds/403 Test generic/604 +/mnt/scratch/48: No such file or directory and also test generic/738 +/data/xfstests-dev/tests/generic/738: line 31: /mnt/scratch/48: No such file or directory +/data/xfstests-dev/tests/generic/738: line 31: /mnt/scratch/48: No such file or directory +/data/xfstests-dev/tests/generic/738: line 31: /mnt/scratch/48: No such file or directory +/data/xfstests-dev/tests/generic/738: line 31: /mnt/scratch/48: No such file or directory +/data/xfstests-dev/tests/generic/738: line 31: /mnt/scratch/48: No such file or directory But with all six current netfs patches (including the two additional below): 68109110fec1 netfs: Fix wait/wake to be consistent about the waitqueue used 4f8443992c8c netfs: Fix the request's work item to not require a ref http://smb311-linux-testing.southcentralus.cloudapp.azure.com/#/builders/5/builds/404 I see a hang early on e.g. in cifs/100 then in cifs/103 (and all following cifs/105 generic/001 etc). I don't see anything useful in /var/log/messages I don't see anything suspicious in /proc/fs/cifs/DebugData but in /proc/fs/cifs/Stats I do see it shows a session reconnect, and I do see a suspicious number of open files (which oddly enough is created by an earlier test, before the hang cifs/100) Open files: 1 total (local), -2 open on server root@fedora29:~# cat /proc/fs/cifs/open_files # Version:1 # Format: # <tree id> <ses id> <persistent fid> <flags> <count> <pid> <uid> <filename> <mid> 0x5 0x792a00b20000a11 0x56ce090b 0x8001 2 5497 0 big-file 23 I will also try a more general test run to see if I can reproduce anything useful with those -- Thanks, Steve
