On Mon, Mar 11, 2024 at 3:31 PM Jan Čermák <sairon@xxxxxxxxx> wrote: > > Hi Paulo, Shyam, > > On 06. 03. 24 16:43, Paulo Alcantara wrote: > > > > Yes. I see a couple of issues here: > > Thanks for the review, Paulo. > > (1) cifs_chan_update_iface() seems to be called over reconnect for all > > dialect versions and servers that do not support > > FSCTL_QUERY_NETWORK_INTERFACE_INFO, so @ses->iface_count will be zero in > > some cases and then VFS message will start being flooded on dmesg. Valid point. > > > > (2) __cifs_reconnect() is queueing smb2_reconnect_server() even for SMB1 > > mounts, so smb2_reconnect() ends up being called and then call > > SMB3_request_interfaces() because SMB2_GLOBAL_CAP_MULTI_CHANNEL is mixed > > with CAP_LARGE_FILES. Excellent catch. I did not take this into account. It is very likely that this is going on here. Let me submit a patch to check the exact dialect before calling these functions to make sure we only do it for SMB3+. > > thanks for looking into this! Is there anything else you'll need from me > or to test by the users who are observing the issue, or do you have > enough information for a fix? > > Regards, > Jan Thanks for bringing this to our notice, Jan. -- Regards, Shyam
