When I used to use smbmount, I used to get lots of "could not access" warnings when the backups I run over smb to Windows machines got to pagefile.sys, etc. That warning always let me know the offending file so I know not to worry. Now that I use mount -t cifs instead of smbmount, I get what I believe are replacement errors when it hits the same files. These errors are all "... on cifs_get_inode_info in lookup". Unfortunately, these errors do not tell me the filename like smbmount did so I'm not sure whether or not I should be worried about these errors. I propose a patch to include the filename that caused the error in the warning output so people can know whether to worry or not: instead of: ("Error 0x%x or (%d decimal) on cifs_get_inode_info in lookup", rc, rc)); do: ("Error 0x%x or (%d decimal) on cifs_get_inode_info in lookup of %s", rc, rc, full_path));
I don't know if this helps your example of open of pagefile.sys but it does make sense to dump the path name here and I have changed the code in dir.c and updated the project bk tree (will update the cogito/bit tree on kernel.org later today).
Thanks!