| commit | a8a4fe0605e6149efb5f9051181d794572edbf27 | |
| author | Jeremy Allison <jra@samba.org> | |
| Thu, 20 May 2010 18:36:47 +0000 (20 11:36 -0700) | ||
| committer | Jeremy Allison <jra@samba.org> | |
| Thu, 20 May 2010 18:36:47 +0000 (20 11:36 -0700) | ||
| tree | 15bcb3771d87fefcd91ae26fbf713c790cbf763e | treesnapshot (tar.gz zip) |
| parent | a7cab4ea36eb21d7b72efa80a77c702bb92bec62 | commitdiff |
Fix bug #7410 - samba sends "raw" inode number as uniqueid with unix extensions.
Move to a consistent get_FileIndex() function for all inode returns,
that checks if st_dev on the file is identical to the top directory
dev_t of the exported share, and if so uses the raw 64-bit inode
number. If it isn't (we've traversed a mount point) - return what
we used to do for Windows which is the concatination of the bottom
32-bits of the inode with the 32-bit device number. We can get more
creative with this over time (hashing?) if we want as now all inode returns go
through this single function.
Jeremy.
Move to a consistent get_FileIndex() function for all inode returns,
that checks if st_dev on the file is identical to the top directory
dev_t of the exported share, and if so uses the raw 64-bit inode
number. If it isn't (we've traversed a mount point) - return what
we used to do for Windows which is the concatination of the bottom
32-bits of the inode with the 32-bit device number. We can get more
creative with this over time (hashing?) if we want as now all inode returns go
through this single function.
Jeremy.