Bug 148250 - NTFS slightly corrupted by 2.6.16 mounted rw?
Summary: NTFS slightly corrupted by 2.6.16 mounted rw?
Status: RESOLVED INVALID
Alias: None
Product: SUSE Linux 10.1
Classification: openSUSE
Component: Installation (show other bugs)
Version: Beta 2
Hardware: x86-64 SUSE Other
: P5 - None : Normal (vote)
Target Milestone: ---
Assignee: Thomas Fehr
QA Contact: Klaus Kämpf
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-02-04 19:35 UTC by Fred Goldstein
Modified: 2006-02-28 14:56 UTC (History)
1 user (show)

See Also:
Found By: Beta-Customer
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Fred Goldstein 2006-02-04 19:35:02 UTC
I installed OpenSUSE 10.1-beta2 on my Athlon 64 system, mainly because nothing post-beta seems to support the nVidia Geforce 6150/430 combo. (Mobo: Foxconn 6150K8MA-EKRS; boot disk: Hitachi Deskstar 160 GB sata; old PATA drive also installed but not used.)  Most things work.

However, a bug (reported by someone else) mounts the NTFS partitions incorrectly, so that only root can see them.  It also mounts them rw, not ro.  It appears that the 2.6.16 kernel now has limited NTFS writing capability, but doesn't get it quite right:  After mounting one with Linux, and booting back to Windows XP, XP detects a problem and goes through chkdsk.
Comment 1 Greg Kroah-Hartman 2006-02-09 23:00:36 UTC
Not a kernel issue, but an installer issue, should not set up the fstab for these
to be mounted rw.
Comment 2 Fred Goldstein 2006-02-10 03:58:45 UTC
I first posted this in another topic but was told to post it as a kernel bug.  Yes, the installer does have a bug, but mounting it ro is more of a workaround than a fix, because the kernel appaers to think that it can write to NTFS, and does, just badly.  If it can't do it right, it shouldn't try.  Perhaps though this is not Novell's problem -- the NTFS write code comes from somewhere else, I think, though I don't know where.  Nor do I know if Linus has accepted it into his official kernel or whether Novell chose to add it.  I do appreciate the bleeding-edge nature of the 2.6.16 kernel.

So yes, for the 10.1 cycle, it is fixed by being mounted ro, but if you know who owns that code, it might be worth reporting it.
Comment 3 Thomas Fehr 2006-02-10 16:27:36 UTC
Please attach full y2log files. If I look at my code, there should be 
already an fstab option that ntfs entries get mounted readonly.
Comment 4 Szabolcs Szakacsits 2006-02-20 06:42:48 UTC
The chkdsk execution is intentionally scheduled by the NTFS kernel driver as an extra sanity check and nothing is wrong with it, except being annoying. This is the expected behaviour currently.

However if chkdsk reported any problem (you can also check this out in the application/winlogon event log) then __THAT__ would be a problem but we don't know about any. If you do then please report it to the Linux NTFS development list, linux-ntfs-dev@lists.sourceforge.net.
Comment 5 Fred Goldstein 2006-02-21 00:47:25 UTC
Thanks.  I don't know if these messages are serious enough or not to be worth worrying about; I *think* these are left over from my SuSe experiment:

Checking file system on C:
The type of the file system is NTFS.

One of your disks needs to be checked for consistency. You
may cancel the disk check, but it is strongly recommended
that you continue.
Windows will now check the disk.                         
Cleaning up 3 unused index entries from index $SII of file 0x9.
Cleaning up 3 unused index entries from index $SDH of file 0x9.
Cleaning up 3 unused security descriptors.

  20482843 KB total disk space.
Comment 6 Szabolcs Szakacsits 2006-02-21 07:15:29 UTC
These messages are absoutely innocent. Security descriptors are indexed but not reference counted and neither the Windows, nor the Linux driver remove unused ones when the last files using them get deleted. This can be useful, for example when they would be soon reused thus the recreation time can be saved (index entry lookup is cheap and easy, meanwhile removal and addition is expensive and complex).

We will document these short of, nothing-to-worry-about, informational chkdsk messages. Thanks.
Comment 7 Hendrik Vogelsang 2006-02-28 14:56:43 UTC
So no real problem here....