Re: _primary_constraint(): start_geom is uninitialized if min_geom==0 && cylinder_size < dev->length. Debian bug #602568

[Jean-Christian de Rivaz]

Jim Meyering a écrit :
> Jean-Christian de Rivaz wrote:
> > Jim Meyering a écrit :
> > > Jean-Christian de Rivaz wrote:
[...]
> > > > address@hidden:/home/jcdr/parted-git/parted/build-O0# parted/parted
> > > > /dev/sdb print
> > > > Model: disk2go PURE II (scsi)
> > > > Disk /dev/sdb: 5243kB
> > > > Sector size (logical/physical): 512B/512B
> > > > Partition Table: msdos
> > > >
> > > > Number  Start  End     Size    Type     File system  Flags
> > > >  1      5120B  5243kB  5238kB  primary
[...]
> > > > Now I just wonder if parted should display FAT16 in the file system
> > > > field. Because fdisk do it:
> > > >
> > > > address@hidden:/home/jcdr/parted-git/parted/build-O0# fdisk -l /dev/sdb
> > > >
> > > > Disk /dev/sdb: 5 MB, 5242880 bytes
> > > > 1 heads, 10 sectors/track, 1024 cylinders
> > > > Units = cylinders of 10 * 512 = 5120 bytes
> > > > Sector size (logical/physical): 512 bytes / 512 bytes
> > > > I/O size (minimum/optimal): 512 bytes / 512 bytes
> > > > Disk identifier: 0x00000000
> > > >
> > > >    Device Boot      Start         End      Blocks   Id  System
> > > > /dev/sdb1               2        1024        5115    6  FAT16
[...]
> > > It seems that 4MiB is the minimum size of a FAT16 file system,
> > > and at 2.5MiB, yours is well under the minimum, so with the latter
> > > interpretation, parted is correct not to report it as one.
> > Sorry, I failed to find where this /deb/sdb1 partition is 2.5MiB in
> > size. Seem to be 5.1MiB (5238kB, or (1024-2+1)*512*10/2.0**20=4.99MiB
> > according to fdisk) to me ( > 4MiB ). This is also confirmed by the
> > system while mounting the /deb/sdb1 partition:
>
> Oh.  I looked at sdb1's partition size, rather than the device size.
> Thanks for the correction.  In that case, it sounds like a bug.
> Can you debug it?

I have traced the problem to the fat_boot_sector_probe_type() function 
from the file libparted/fs/fat/bootsector.c. In particular the last 
couple of lines:

        if (cluster_count > MAX_FAT12_CLUSTERS)
                return FAT_TYPE_FAT16;
        else
                return FAT_TYPE_FAT12;

And in fat.h, MAX_FAT12_CLUSTERS has value 4086. The small FAT16 
partition on my /dev/sdb1 have cluster_count = 2544, so the function 
returns FAT_TYPE_FAT12. Since parted don't support FAT12, it do not 
display this file system type. I found curious that the "File system is 
FAT12, which is unsupported." from the fat_boot_sector_analyse() is not 
displayed. I tested to change MAX_FAT12_CLUSTERS to 2048 and parted 
report a fat16 file system for my /dev/sdb1.

A this point I fail to understand why parted probe the file system type 
this way as the primary information should be the partition identifier 
from the MBR: 0x01 for FAT12 and 0x04, 0x06, 0x0E for FAT16.

If we found a way to get the partition identifier from the 
fat_boot_sector_probe_type() function, the code could be something like 
this:

     if (partition_identifier == 0x01 &&
         cluster_count <= MAX_FAT12_CLUSTERS)
                return FAT_TYPE_FAT12;
        else
                return FAT_TYPE_FAT16;

If this proposition is reasonable, my problem is how to get the 
partition_identifier.

Or should we use an other way to make the difference between FAT12 and 
FAT16 ?

Jean-Christian de Rivaz