Stephen Williams wrote:
>>something else you might try: in order to re-synchronize the
>>communication you may insert a dummy read/write/read command and ignore
>>the error code if an error occurs. The first read will fail in normal
>>case (no previous write issued, the device is awaiting a new command)
>>and succeed in the unusual case when the device communication is
>>out-of-sync...
>>
>>please report your experiences and let us know whether this makes the
>>code more robust,
>>
>>Holger
>>
>>
>
>Ahhh, now we're heading outside of my current abilities, sorry but i'm
>an embedded systems Ada programmer by trade. I've never used C in
>anger and I don't know my why around the kernel / drivers, i'm much
>more at home when there isn't an OS ;-) If you could give me a bit
>more of a hint I should be able to figure it out.
>
>
Out of my mind the USB message would look somewhat like this:
unsigned char msg [64];
/* dummy command readback to get back in sync if we are
out-of-sync */
memset(msg, 0, sizeof(msg));
/* ignore error code here */
usb_bulk_write(fx2->dev, 1, msg, sizeof(msg), 200);
if ((ret = usb_bulk_read(fx2->dev, 1, msg, sizeof(msg), 200)) > 0)
/* just some verbosity for debugging... */
msg("fetched %d bogus bytes from read buffer...\n", ret);
but please check twice to be sure. Don't know for sure if the USB API
syntax is correct, please adopt in case it is not.
Holger
