Hi. <I-tell-my-life> I have a hard drive with an UDF filesystem, and something strange happened a few weeks ago: I had a directory foo, then I created a file bar at a completely different place; after that, foo was no longer a directory, it was a plain file, a hard-link to the newly created bar, except that the link count was wrong. Obviously there is a bug somewhere. Since then, I spent part of my free-time to dissect my filesystem, and I found the following strange detail. </I-tell-my-life> Some directories have eight allocated unrecorded (extent.length >> 30 == 1, according to ECMA 167, 4/14.14.1.1, page 4/46) sectors at the end. The strange thing is that some of these sectors also belong to others files or directories, as recorded sectors. Is this situation normal? (For the record, I am currently using a 2.6.14.1 kernel, but I did not see anything related in the ChangeLog of later releases.) PS: I managed to build and run fsck.udf from OpenSolaris under GNU/Linux, but it seems to be limited to version 2 records, while Linux produces both version 2 and version 3 records, so it did not help. I intend to release the tools I have written to dissect my filesystem, but they still need a lot of work. Regards, -- Nicolas George
Attachment:
signature.asc
Description: Digital signature
- Follow-Ups:
- Re: [UDF] (bogus?) overlap of directories and files + UDF oops
- From: Salah Coronya <[email protected]>
- Re: [UDF] (bogus?) overlap of directories and files + UDF oops
- Prev by Date: Re: Stradis driver conflicts with all other SAA7146 drivers
- Next by Date: Re: [ PATCH ] Longhaul - call suspend(PMSG_FREEZE) before and resume() after
- Previous by thread: [BUG] Link problem on drivers/acpi/processor_idle.c
- Next by thread: Re: [UDF] (bogus?) overlap of directories and files + UDF oops
- Index(es):
 |