[Bug 29426] New: UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

Bug #: 29426
           Summary: UDF support: VOLUME_GetSuperblockSerial invalid for
                    some volume types
           Product: Wine
           Version: unspecified
          Platform: All
        OS/Version: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: kernel32
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]
    Classification: Unclassified

VOLUME_GetSuperblockSerial returns an incorrect serial for certain UDF volumes.

Volume types affected:
  * ISO9660+UDF for all revisions: Serial displays as '0104-0001'
  * UDF revision 2.50/2.60: Serial displays as '0000-0000'

Serial is correct for standalone UDF revisions 1.02/1.50/2.00/2.01

Test Case: Run 'wine cmd' to open the command line and observe the output of
the `vol` command. Output differs from `vol` run on a native windows system for
the above volume types.

History: UDF support was introduced as a patch for bug #26273
http://bugs.winehq.org/show_bug.cgi?id=26273 and is hardcoded to use the 257th
sector (Address 526336) for serial calculation.

The actual sector used for serial calculations depends on the UDF revision and
volume type. I suspect the sector may be coded somewhere in the UDF
Specifications, and if that is the case then it should be a better solution
than hardcoding.

Serial calculations:
  * Sector 257 for UDF revisions 1.02/1.50/2.00/2.01
  * Sector 260 for ISO9660+UDF revisions 1.02/1.50/2.00/2.01
  * Sector 320 for UDF revisions 2.50/2.60 with or without ISO9660

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

Vitaliy Margolen  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Platform|All                         |Other

--- Comment #1 from Vitaliy Margolen  2011-12-23 13:36:12 CST ---
Could you point to spec describing which blocks the volume serial number is
located?

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

mcfang  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #2 from mcfang  2011-12-23 22:34:09 CST ---
(In reply to comment #1)
> Could you point to spec describing which blocks the volume serial number is
> located?

I could not decipher the UDF spec to find which sector was being used for
serial calculations so I hope that someone with some experience with UDF or
ECMA-167 can shed some light.

The sectors I gave in the bug report are based on manual investigation of the
volumes.

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

--- Comment #3 from mcfang  2011-12-23 22:34:32 CST ---
(In reply to comment #1)
> Could you point to spec describing which blocks the volume serial number is
> located?

I could not decipher the UDF spec to find which sector was being used for
serial calculations so I hope that someone with some experience with UDF or
ECMA-167 can shed some light.

The sectors I gave in the bug report are based on manual investigation of the
volumes.

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

--- Comment #4 from Dmitry Timoshkov  2011-12-24 00:36:40 CST ---
Please specify 'wine --version' in the Version field above.

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

mcfang  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|unspecified                 |1.3.35

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

--- Comment #5 from Alex  2011-12-24 04:38:39 CST ---
(In reply to comment #1)
> Could you point to spec describing which blocks the volume serial number is
> located?

I'm the author of the patch "kernel32: Add UDF support"
(http://source.winehq.org/git/wine.git/commitdiff/73f4e23938142225efdc233f9db...)

I read the spec and the volume serial is not part of it.
In fact, Windows uses its own algorithm depending on the filesystem of the
volume.

See
http://social.msdn.microsoft.com/Forums/en-US/windowsopticalplatform/thread/...
Quoting the author : "This serial number is NOT part of the file system on the
disc and is NOT located on the source disc AT ALL.  It is entirely generated by
Windows."

See also the code used by wine :
http://source.winehq.org/git/wine.git/blob/HEAD:/dlls/kernel32/volume.c#l569

To find which sector was used by Microsoft :
- I made an ISO of my Sims3 DVD (a retailer/genuine one)
- I got the serial of it from a (VirtualBox'ed and genuine) XP by mounting the
ISO file
- I simply wrote a for loop and found it was the 257th sector with bytes
reversed

Note that it was the only UDF DVD I had (and still have).

If you found that this sector depends on the UDF revision, the code must be
changed.

Either you feel ready to write a patch (I can review it if you want) or not and
in this case, I can write it for you.

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

--- Comment #6 from mcfang  2011-12-28 21:00:05 CST ---
Thanks Alex!

I am also aware that the 'serial' is not in the spec, but the specific sector
used by Microsoft appears to be consistent across different UDF revisions which
is why I believe it can be mapped by understanding the UDF spec. This would
also be more robust than guessing the sector based on the UDF revision.

I'm guessing the sector used for the serial is related to the start of the data
partition. The Partition Descriptor might hold the key: it appears to give the
correct sector for all volume types except in UDF 2.50/2.60 where it gives 288
instead of 320.

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

--- Comment #7 from Alex  2012-07-22 13:14:49 CDT ---
Created attachment 41104
  --> http://bugs.winehq.org/attachment.cgi?id=41104
A simple program to compare the sector used to compute the serial and the
partition starting location

I finally managed to write a program that displays :
+ the sector used to compute the serial from GetVolumeInformation
+ and the field 'partition starting location' of the Partition Descriptor

You can compile it with mingw (i686-w64-mingw32-gcc on my computer)
And run it, on Windows, with a DVD drive in, from the command line.

The syntax is : <nameOfTheExe> <DosDriveLetter>
the last one is optional and will default to F

So give it a try with a maximum of UDF DVD and give us the results

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

Alex  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #41104|0                           |1
        is obsolete|                            |

--- Comment #8 from Alex  2012-12-02 06:14:21 CST ---
Created attachment 42691
  --> http://bugs.winehq.org/attachment.cgi?id=42691
Use the FSD sector to compute the UDF serial

In the past weeks, I did several tests and found several things :

+ windows does not compute the serial on demand but when the media is inserted. 
+ windows does a light check : it doesn't check TEA01 presence, doesn't read
every file entry but looks for the Autorun file...
+ the sector used  to compute the serial is not the partition starting location
nor dependant of the UDF revision but is the location of the FSD (file set
descriptor) location.

I'm pretty convinced of what I found but I'd like you, mcfang, to test my patch
and report here if it matches windows behavior.

The simplest way to test the serial is to use cmd (and wine cmd), then typing
e: (or the appropriate letter) to change the working directory and executing
the dir command.

Thanks for your help

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

Bruno Jesus  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #9 from Bruno Jesus  2012-12-11 17:24:21 CST ---
A patch related to this bug was commited:
http://source.winehq.org/git/wine.git/?a=commit;h=d0001e357a4e054a883a02cf97...

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

--- Comment #10 from mcfang  2012-12-18 02:39:32 CST ---
Alex, I tested your patch and it gives the expected result for all volume types
I tried.

Great work, credit to you.

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

--- Comment #11 from Alex  2012-12-18 02:50:23 CST ---
Glad to see I was right.
You can now change the status of this bug to RESOLVED.
AJ will close it Friday for wine v1.5.20 .. if we're still alive ;)

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

mcfang  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED

--- Comment #12 from mcfang  2012-12-18 02:51:59 CST ---
Fixed with submitted patch

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

Nikolay Sivov  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
      Fixed by SHA1|                            |d0001e357a4e054a883a02cf97b
                   |                            |e02db90310088

[Bug 29426] UDF support: VOLUME_GetSuperblockSerial invalid for some volume types

http://bugs.winehq.org/show_bug.cgi?id=29426

Alexandre Julliard  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #13 from Alexandre Julliard  2012-12-21 13:29:17 CST ---
Closing bugs fixed in 1.5.20.