On Wed, 7 Feb 2024 at 18:56, R. Diez <rdiez-2006@xxxxxxx> wrote: > > Hallo Steve: > > I wonder what would happen if the SMB server said it can take a maximum of 2048 Bytes, and you insist on 4096. Would the connection still work later on? Wouldn't it be better to abort the connection with a descriptive error message? There are no servers that we would care about that have limits that low. And if there were we could just open a bug against the server to fix it server-side. If a server refuses to work with anything larger than 2048 bytes it is completely valid to just shutdown the session and refuse to connect to the server. > > You stated that this scenario is very unlikely, but my Linux client is negotiating 16580 bytes at the moment, so if PAGE_SIZE happens to be 64 KiB, that wouldn't be very unlikely any more. > > About this other change: > > if (round_up(ctx->wsize, 4096) != ctx->wsize) > cifs_dbg(VFS, "wsize should be a multiple of 4096\n") > > All my SMB connections have been automated, therefore I am very unlikely to realise of such a warning. It would also mislead people using the current Kernel versions into thinking that this limitation is there to stay. > > If the SMB client cannot really honour the user request, wouldn't it be better to fail? In any case, how about mentioning in the error or warning message that this is only a temporary limitation? > > The second version of your patch file looks like a VIM swap file. I gather you attached the wrong file. The best way to fix this is obviously to switch to Emacs. ;-) > > Regards, > rdiez > >
