Hi Mike.   In my case, I wrote the image with a Mac.  I run FreeBSD 
inside virtualbox because I no longer have FreeBSD machines at home.

You may have multiple cards with the same problem.  More comments below.

On 1/26/11 1:31 PM, Mike Tancsa wrote:
> On 1/26/2011 3:45 PM, Robert Sexton wrote:
>> I just ran into something similar with 4G cards, where the nanobsd
>> numbers for 4G sandisk cards were wrong.
>>
>> It looks like in these most recent cards, Sandisk has reserved more
>> flash for wear leveling and spare sectors,
>> so a 4G nanobsd image won't fit onto a 4G card.  I used the numbers from
>> fdisk to calculate the new device size, and used that in my nanobsd
>> script, and all was well.   You're getting errors at the 3.9G boundary.
>> You may be running off the end of the device.
> Hmmm, even if I make the media size just 2G, I still get the errors. How
> are you burning the image onto the CF ?  What values did you use to
> create the image ? Or is it CF reader dependent ?
>
>
>> On 1/24/11 6:41 PM, Mike Tancsa wrote:
>>> On the latest SANDisk batch of CFs I got, I am getting periodic errors
>>> reading from the flash on my 5501 (FreeBSD RELENG_8).  However, when I
>>> put the flash in different readers-- e.g. attached to a regular
>>> server, no errors.  On the 5501 I get periodic
>>>
>>> ad0: FAILURE - READ_DMA status=51<READY,DSC,ERROR>
>>> error=10<NID_NOT_FOUND>   LBA=7813103
>>>
<SNIP>
>>> the 2G flashes work just fine in the same box.  The only difference I
>>> can see between the 2G and 4G CF is that that atacontrol says "lba48
>>> supported".... Could this be the issue ? If so, is there a way to turn
>>> it off or work around it ?  My supplier is not carrying 2G anymore and
>>> the 4 is the same price as the 2G :(
>>>
>>>
>>>
>>> Protocol              ATA/ATAPI revision 0
>>> device model          SanDisk SDCFH-004G
>>> serial number         ACZ120910191749
>>> firmware revision     HDX 6.02
>>> cylinders             7751
>>> heads                 16
>>> sectors/track         63
>>> CFA supported
>>> lba supported         7813120 sectors
>>> lba48 supported       7813120 sectors
>>> dma supported
>>> overlap not supported

7751 * 16 * 63 = 7813008, so the system is reporting a sector count 
thats larger than that calculated from C/H/S.  If anything in the 
filesystem refers to these blocks, interesting things will happen.

I'd try 7750/16/63, or 7812000

- Robert










_______________________________________________
Soekris-tech mailing list
[email protected]
http://lists.soekris.com/mailman/listinfo/soekris-tech

Reply via email to