Hi Wolfgang,

>> We'd need to find out how long the SMSQ/E SD card driver will disable
>> the serial interrupt - if it actually does.
> Yes it does.
> What happens with data trickling in from something like a serial port, 
> is that it is put into a buffer, i.e. the slave blocks. When one slave 
> block is full (it has the same size as a sector on the card, i.e. 512 
> bytes), it is flagged as being ready to be written back to the disk (or 
> card). At the next poll interrupt the slave block then gets written out. 
> During that time interrupts are disabled (status reg = $2200). If, for 
> some reason, several slave blocks can be written out at once, they will 
> go in the same interrupt.

Thanks for the clarification. Does this mean the duration of interrupt
disable is totally unpredictable, and could even be much longer than the
polling period?

E.g. a hardware receive buffer size of 256 bytes to "survive" one
polling period would be doable.

> Question: I don't use sernet, but I do know that some use it. Has 
> anybody ever lost data transferred to the Q68 via sernet?

No, never. SERNET seems to work perfectly at 115200 Baud. For Q68,
Qzero, QIMSI connected to each other, or to Q60 or PC emulator. Which is
one reason why I did not implement RTS on the Qzero.

(As always, there were not enough dedicated pins left, so I had to
"steal" CTS from the 40 pin I/O connector of the Qzero. I want as many
pins on that connector untouched, so I hoped to avoid RTS.)

All the best
QL-Users Mailing List

Reply via email to