> -----Original Message----- > From: linux-cifs-owner@xxxxxxxxxxxxxxx <linux-cifs-owner@xxxxxxxxxxxxxxx> On > Behalf Of Aurélien Aptel > Sent: Tuesday, March 5, 2019 10:41 AM > To: Steve French <smfrench@xxxxxxxxx>; CIFS <linux-cifs@xxxxxxxxxxxxxxx> > Subject: Re: [PATCH] CIFS: fix FSCTL_SET_REPARSE_POINT SMB2_ioctl() call > > Aurélien Aptel <aaptel@xxxxxxxx> writes: > > I've tried setting MaxOutputResponse to server->maxBuf or > > Connection.MaxTransactSize but it always fails the same way... so I > > don't know. I guess it must say somewhere that ioctl() with no output > > should set MaxOutputResponse to zero? > > I've just checked what Windows does and it seems to be that as > well. Creating a symlink (mklink link target) makes an > ioctl(SET_REPARSE_POINT) with MaxOutputResponse set to 0. > > So I would keep this patch as is. Remember that in addition to the SMB server validation, the request goes to the filesystem, which has its own validation and error status returns. The STATUS_INVALID_PARAMETER can come back from any layer. If the MaxOutputResponse != 0 set_reparse_point behavior isn't covered in either MS-SMB2 or MS-FSA, I'm happy to take a doc bug. Tom.
