From mboxrd@z Thu Jan  1 00:00:00 1970
From: Brad Parker <brad@heeltoe.com>
Date: Thu, 08 May 2008 16:33:55 -0400
Subject: [U-Boot-Users] fat32 w/o parition table? (a.k.a VBR instead of MBR)
Message-ID: <1998.1210278835@mini>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de


After some user complaints about not being able to fatls micro
sd cards which were formatted by Windows XP, I felt compelled to 
make the following change (to u-boot 1.1.6, which I know is ancient,
but looking at the git tree, the latest doesn't handle it either near
as I can tell.

Just curious if anyone has any thoughts.

In the latest code, it seems like checking the partition table first
is dangerous as a VBR might looks like it has a partition table (but it
would be bogus)

Some documents suggest checking the first 3 bytes of the block for the
branch instructions.  The idea being that an MBR with a valid partition
table won't have the branch.  voodoo, I know...

-brad

--- fat.c.orig  2008-05-08 16:18:40.000000000 -0400
+++ fat.c       2008-05-08 12:54:56.000000000 -0400
@@ -52,6 +52,7 @@
 #define DOS_PART_TBL_OFFSET    0x1be
 #define DOS_PART_MAGIC_OFFSET  0x1fe
 #define DOS_FS_TYPE_OFFSET     0x36
+#define DOS_FS_FAT32_TYPE_OFFSET       0x52

 int disk_read (__u32 startblock, __u32 getsize, __u8 * bufptr)
 {
@@ -85,7 +86,9 @@
                return -1;
        }

-       if(!strncmp((char *)&buffer[DOS_FS_TYPE_OFFSET],"FAT",3)) {
+       if(!strncmp((char *)&buffer[DOS_FS_TYPE_OFFSET],"FAT",3) ||
+          !strncmp((char *)&buffer[DOS_FS_FAT32_TYPE_OFFSET],"FAT32",5))
+       {
                /* ok, we assume we are on a PBR only */
                cur_part = 1;
                part_offset=0;


This lets me fatls/fatload from the fat32 file systems created by windows. 

I've done a little FAT hacking, and I think this falls under the
category of "disk with VBR instead of MBR".

-brad