ZTE MC315 CDMA wireless PCMCIA data card on Linux

[Prof. Prabhat Ranjan]

Russell,
  I tried changing the baud rate from 300 to 230k in minicom and seeing if
i got a reponse to AT command. Does not seem to do anything. Is there
any better test I can do?
- Prabhat Ranjan
> On Mon, Dec 04, 2006 at 10:31:05PM -0500, Dominik Brodowski wrote:
>> Hi,
>>
>> On Thu, Nov 02, 2006 at 12:40:52PM +0530, Prof. Prabhat Ranjan wrote:
>> > Basically it is based on Qualcomm 5105/6025 platform and seems to use
>> Oxford
>> > Semiconductors OXCF950 Rev B. Here are the output from pccardctl :
>> >
>> > $MANFID - 0279,950b
>> > $FUNCID - 2
>> > $PRODID_1 - CDMA1X
>> > $PRODID_2 - CARD
>> >
>> > In the file  "linux-2.6.18.1/drivers/serial/serial_cs.c" a number of
>> > workaround for OXCF950 Rev B have been added. In addition, I have also
>> noted
>> > that in file 8250.c, 8250_pci.c and sunsu.c workarounds for this
>> paricular
>> > chip. One of this seems to be related to Baudrate miscalculation.
>> However,
>> > this workaround has not been done in serial_cs.c.
>> >
>> > I am wondering whether the card is not responding due to one of these
>> > workarounds not being applied in serial_cs.c or any other pointers one
>> may
>> > have.
>>
>> Actually, these workarounds are already executed, else the code wouldn't
>> reach the line stating the addition of ttyS3. Russell, do you have any
>> ideas
>> on what's going wrong here? Or someone (i.e. new linux-serial
>> maintainer)
>> who might be interested in this?
>
> Not sure.  Wild stab in the dark time.
>
> The 950 has a set of registers which select a clock divisor.  It appears
> that on some bluetooth cards, these registers are set at power up to be
> compatible with standard 16550A baud rate divisor numbers.
>
> However, we reset the chip, which resets the clock divisor registers
> back to zero.
>
> What Prof. Prabhat could try is communicating with the card after
> selecting a slow baud rate, and see whether that gets anywhere.
>
> Because of this clock divisor register setup, all 950 seem to require
> quite a bit of black art to get them working.  I would have dearly liked
> to have a 950 card to be used for experimenting with, but I tried for
> _years_ (literally) and never got anywhere...
>
> Consequently, I do not have any enthusiasm for 950 support.
>
> --
> Russell King
>  Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
>  maintainer of:
>