ISOFS corrupt filesizes

ISOFS corrupt filesizes

[Matthias Kramm]

According to the (2.4.9) MAINTAINERS-File,  ISOFS doesn't have a maintainer,
so this probably best fits in this list.

I came across a (commercial) DVD with an ISOFS Filesystem on it and filesizes
bigger than 1M.
I.e. "ls -l /mnt/cdrom/video_ts" shows

total 8153965
-r-xr-xr-x    1 root     root        34816 Feb 29  2000 video_ts.bup
-r-xr-xr-x    1 root     root        34816 Feb 29  2000 video_ts.ifo
-r-xr-xr-x    1 root     root   1201278976 Feb 29  2000 video_ts.vob <---
-r-xr-xr-x    1 root     root        18432 Feb 29  2000 vts_01_0.bup
-r-xr-xr-x    1 root     root        18432 Feb 29  2000 vts_01_0.ifo
-r-xr-xr-x    1 root     root        20480 Feb 29  2000 vts_01_0.vob
-r-xr-xr-x    1 root     root        10240 Feb 29  2000 vts_01_1.vob
-r-xr-xr-x    1 root     root        18432 Feb 29  2000 vts_02_0.bup
-r-xr-xr-x    1 root     root        18432 Feb 29  2000 vts_02_0.ifo
-r-xr-xr-x    1 root     root    967038976 Feb 29  2000 vts_02_1.vob
-r-xr-xr-x    1 root     root        96256 Feb 29  2000 vts_03_0.bup
-r-xr-xr-x    1 root     root        96256 Feb 29  2000 vts_03_0.ifo
-r-xr-xr-x    1 root     root    195196928 Feb 29  2000 vts_03_0.vob
-r-xr-xr-x    1 root     root   1073709056 Feb 29  2000 vts_03_1.vob
-r-xr-xr-x    1 root     root   1073709056 Feb 29  2000 vts_03_2.vob
-r-xr-xr-x    1 root     root   1073709056 Feb 29  2000 vts_03_3.vob
                                   (...)

Actually, that is what it _should_ show. Here is the  actual output:
(kernel versions 2.2.18 through 2.4.9)

$ ls -l /mnt/cdrom/video_ts
Warning: defective CD-ROM. Enabling "cruft" mount option.
total 142925
-r-xr-xr-x    1 root     root        34816 Feb 29  2000 video_ts.bup
-r-xr-xr-x    1 root     root        34816 Feb 29  2000 video_ts.ifo
-r-xr-xr-x    1 root     root     10096640 Feb 29  2000 video_ts.vob
-r-xr-xr-x    1 root     root        18432 Feb 29  2000 vts_01_0.bup
-r-xr-xr-x    1 root     root        18432 Feb 29  2000 vts_01_0.ifo
-r-xr-xr-x    1 root     root        20480 Feb 29  2000 vts_01_0.vob
-r-xr-xr-x    1 root     root        10240 Feb 29  2000 vts_01_1.vob
-r-xr-xr-x    1 root     root        18432 Feb 29  2000 vts_02_0.bup
-r-xr-xr-x    1 root     root        18432 Feb 29  2000 vts_02_0.ifo
-r-xr-xr-x    1 root     root     10737664 Feb 29  2000 vts_02_1.vob
-r-xr-xr-x    1 root     root        96256 Feb 29  2000 vts_03_0.bup
-r-xr-xr-x    1 root     root        96256 Feb 29  2000 vts_03_0.ifo
-r-xr-xr-x    1 root     root     10647552 Feb 29  2000 vts_03_0.vob
-r-xr-xr-x    1 root     root     16744448 Feb 29  2000 vts_03_1.vob
-r-xr-xr-x    1 root     root     16744448 Feb 29  2000 vts_03_2.vob
-r-xr-xr-x    1 root     root     16744448 Feb 29  2000 vts_03_3.vob
                                   (...)

Notice the corrupt filesizes. The culprit is obviously in the code which
"printk"s the "cruft" message above.

In 2.4.9-kernel/fs/isofs/inode.c around line 1190 it says

if ((inode->i_size < 0 || inode->i_size > 1073741824) &&
    inode->i_sb->u.isofs_sb.s_cruft == 'n') {
	printk(KERN_WARNING "Warning: defective CD-ROM.  "
	       "Enabling \"cruft\" mount option.\n");
	inode->i_sb->u.isofs_sb.s_cruft = 'y';
}

My personal guess would be that the assumption that iso files can't be
bigger than 1M unless the CD-ROM is defective is wrong.
(I don't know where the 1M comes from. 2M sounds more logical to me,
 however)
After removing the "indode->i_size > 1073741824" test, I got the correct
output for ls. Also was I able to cat (css-cat, actually) the whole 1201278976 
bytes without an error, which may lead to the assumption the file
is actually that big.
If I'm wrong and the dvd is actually broken, however, I'd like to 
suggest making the automatic cruft mount optional.

Greetings 

Matthias

Re: ISOFS corrupt filesizes

[Matthias Kramm]

> I came across a (commercial) DVD with an ISOFS Filesystem on it and filesizes
> bigger than 1M.
> (...)
> My personal guess would be that the assumption that iso files can't be
> bigger than 1M unless the CD-ROM is defective is wrong.
> (I don't know where the 1M comes from. 2M sounds more logical to me,
>  however)
> (...)

Please substitute 1M with 1G and 2M with 2G. Thank you. (My fault, sorry :) )

Greetings

Matthias

Re: ISOFS corrupt filesizes

[H. Peter Anvin]

Followup to:  <20010914145352.A9952@stud.tu-muenchen.de>
By author:    Matthias Kramm <matthias.kramm@stud.tu-muenchen.de>
In newsgroup: linux.dev.kernel
> According to the (2.4.9) MAINTAINERS-File,  ISOFS doesn't have a maintainer,
> so this probably best fits in this list.
> 
> I came across a (commercial) DVD with an ISOFS Filesystem on it and filesizes
> bigger than 1M.
>

1 GB (not MB) you mean...

> My personal guess would be that the assumption that iso files can't be
> bigger than 1M unless the CD-ROM is defective is wrong.
> (I don't know where the 1M comes from. 2M sounds more logical to me,
>  however)

1 GB comes from the fact that some old CD's actually put garbage in
the upper byte of the file size, so the test triggers if the size is
larger than any CD can be.  Unfortunately, DVDs are a lot bigger than
CDs and that assumption is no longer correct.

> After removing the "indode->i_size > 1073741824" test, I got the correct
> output for ls. Also was I able to cat (css-cat, actually) the whole 1201278976 
> bytes without an error, which may lead to the assumption the file
> is actually that big.
> If I'm wrong and the dvd is actually broken, however, I'd like to 
> suggest making the automatic cruft mount optional.

The automatic cruft mount option should probably be conditionalized
not on a fixed size, but on the actual size of the filesystem.

	-hpa

-- 
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt	<amsp@zytor.com>

Re: ISOFS corrupt filesizes

[Alan Cox]

> 1 GB comes from the fact that some old CD's actually put garbage in
> the upper byte of the file size, so the test triggers if the size is
> larger than any CD can be.  Unfortunately, DVDs are a lot bigger than
> CDs and that assumption is no longer correct.

DVD is supposed to be using 1Gb files. I don't think its a big issue as
we support UDF too

Re: ISOFS corrupt filesizes

[H. Peter Anvin]

Alan Cox wrote:
>>1 GB comes from the fact that some old CD's actually put garbage in
>>the upper byte of the file size, so the test triggers if the size is
>>larger than any CD can be.  Unfortunately, DVDs are a lot bigger than
>>CDs and that assumption is no longer correct.
>>
> 
> DVD is supposed to be using 1Gb files. I don't think its a
> big issue as we support UDF too
> 

Well, it's not all about DVDs used for video.  I think there are 
legitimate reasons to have very large files on a DVD-ROM, and they're 
likely to be encoded in iso9660 format (for maximum compatibility) as 
long as we don't have individual files > 4 GB.

	-hpa