libs/newlib-cygwin
libs
/
newlib-cygwin
mirror of git://sourceware.org/git/newlib-cygwin.git
4
0
Fork 0
Code Issues Projects Releases Wiki Activity

* fhandler_disk_file.cc (fhandler_base::fstat_helper): Try harder 

to determine remote file systems with reliable inode numbers.  Add
	longish comment.
This commit is contained in:
Corinna Vinschen 2006-01-24 12:32:33 +00:00
parent 3784b87b32
commit 26d27a276f
2 changed files with 31 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
winsup/cygwin/ChangeLog
View File

@ -1,3 +1,9 @@
2006-01-24 Corinna Vinschen <corinna@vinschen.de>
* fhandler_disk_file.cc (fhandler_base::fstat_helper): Try harder
to determine remote file systems with reliable inode numbers. Add
longish comment.
2006-01-23 Corinna Vinschen <corinna@vinschen.de> 2006-01-23 Corinna Vinschen <corinna@vinschen.de>
* fhandler_socket.cc (fhandler_socket::fixup_after_fork): Reset * fhandler_socket.cc (fhandler_socket::fixup_after_fork): Reset

28
winsup/cygwin/fhandler_disk_file.cc
View File

@ -326,14 +326,36 @@ fhandler_base::fstat_helper (struct __stat64 *buf,
case DRIVE_REMOVABLE: case DRIVE_REMOVABLE:
case DRIVE_CDROM: case DRIVE_CDROM:
case DRIVE_RAMDISK: case DRIVE_RAMDISK:
/* Temporarily enable remote drives until we find out why we disabled them
in the first place. When we find out don't forget to write a comment! */
case DRIVE_REMOTE:
/* Although the documentation indicates otherwise, it seems like /* Although the documentation indicates otherwise, it seems like
"inodes" on these devices are persistent, at least across reboots. */ "inodes" on these devices are persistent, at least across reboots. */
buf->st_ino = (((__ino64_t) nFileIndexHigh) << 32) buf->st_ino = (((__ino64_t) nFileIndexHigh) << 32)
| (__ino64_t) nFileIndexLow; | (__ino64_t) nFileIndexLow;
break; break;
case DRIVE_REMOTE:
/* From own experiments and replies from the Cygwin mailing list,
we're now trying to figure out how to determine remote file
systems which are capable of returning persistent inode
numbers. It seems that NT4 NTFS, when accessed remotly, and
some other remote file systems return unreliable values in
nFileIndex. The common factor of these unreliable remote FS
seem to be that FILE_SUPPORTS_OBJECT_IDS isn't set, even though
this should have nothing to do with inode numbers.
An exception is Samba, which seems to return valid inode numbers
without having the FILE_SUPPORTS_OBJECT_IDS flag set. So we're
testing for the flag values returned by a 3.x Samba explicitely
for now. */
#define FS_IS_SAMBA (FILE_CASE_SENSITIVE_SEARCH \
| FILE_CASE_PRESERVED_NAMES \
| FILE_PERSISTENT_ACLS)
if ((pc.fs_flags () & FILE_SUPPORTS_OBJECT_IDS)
|| pc.fs_flags () == FS_IS_SAMBA)
{
buf->st_ino = (((__ino64_t) nFileIndexHigh) << 32)
| (__ino64_t) nFileIndexLow;
break;
}
/*FALLTHRU*/
default: default:
/* Either the nFileIndex* fields are unreliable or unavailable. Use the /* Either the nFileIndex* fields are unreliable or unavailable. Use the
next best alternative. */ next best alternative. */