NTFS IOException: Invalid magic found: 0

Project:JNode Core
Component:Code
Category:bug report
Priority:normal
Assigned:Unassigned
Status:closed
Description

I used to dismiss this error saying that it only occurred on corrupted parts of a disk, but just now we seem to have someone with a disk where every single file and directory on the disk has this error.

It may be a new version of NTFS, but I have no idea. It seems too strange a coincidence that it should be 0 every single time, when there are any number of other possible values for the magic number.

#1

A second thought... maybe this could actually be an extremely *old* version of NTFS?

#1

Daniel, did you track this down? Is it still a "live" issue, or can we close it?

#2

Never managed to track it down, no... it's a weird issue because every record in NTFS is supposed to have some kind of ASCII ID like "FILE", and yet there were clearly records pointing to "\0\0\0\0" on this one particular disk image. But I never got my hands on the image either.

Probably good to keep it around to record that it's still a problem, but the pressure on our end has let up so I probably won't be looking into it until someone else discovers it.

#3

Priority:critical» normal
Status:active» postponed

If we cannot reproduce it, this issue cannot be 'critical'. Marking as 'postponed'

#4

Status:postponed» closed

Will be eliminated by Issue 2817 as the underlying cause was that an IOException was being masked, resulting in a buffer full of zeros. After that patch it will propagate the IOException, leaving the underlying problem to be fixed (I have fixed one such problem already.)