[geeks] Seagate buckles to math ignorant consumers

[Joshua Boyd]

On Fri, Nov 02, 2007 at 10:53:30AM -0400, Francois Dion wrote:

> On flash, they already are. Check df -h on a 256MB or 2GB usb stick,
> you will be unpleasantly surprised. The 2GB usb stick I'm giving away
> on my blog, with Belenix pre-loaded is showing:
> /dev/dsk/c5t0d0s0      1.8G   833M   961M    47%    /media/JD FIREFLY
> 
> and rmformat reports:
>  2. Logical Node: /dev/rdsk/c5t0d0p0
>         Physical Node: /pci at 0,0/pci1028,1c1 at 1d,7/storage at 7/disk at 0,0
>         Connected Device: LEXAR    JD FIREFLY       1100
>         Device Type: Removable
>         Bus: USB
>         Size: 1.9 GB
> 
> 
> So it is reporting maybe at most 1.9GB, but 1794MB accounted for per
> df -h. On supposedly 2GB.

The flash chips are almost certainly 2*2^30 rather than 2*10^9.  If the
device shows less available that is because either there are already bad
blocks or else because the controller is holding some blocks in reserve
to use in place of future failed blocks (or some combination of the
two).

And of course, to make life fun for me, I can by a stack of flash cards
from a single vendor and the cards will vary in capacity by a few megs.
I wanted to make lots of identical cards using dd, but i ended up giving
up on that and making a script to format the cards and then do a rsync
to get the programs onto them.