[Info-vax] QIO Writes to a mailbox

[Jim Duff]

On 6/11/2011 5:46 AM, JF Mezei wrote:
> Jim Duff wrote:
> 
>>
>> Remind me not to hire you as a programmer, as QIO vs QIOW sure does make
>> a difference.
> 
> I am asking in the specific case of the mailbox driver which is an
  ^^^^^^^^^^^
> in-memory construct and the driver does not have to wait for some device
> to rotate to a specific location before the write can be made, or send
> the data out at low speed over a 110baud line.

You're asking?  You wrote:

"Since a write to a mailbox is nearly instantaneous if here is
sufficient buffer space to add the new message, the use of QIO or QIOW
doesn't make much of a difference from a timing point of view."

This doesn't sound like a question, but more like a statement of fact.
Unfortunately, it's completely bogus.

As a number of people have already pointed out, the Q in QIO(w) stands
for QUEUE.  There are all sorts of interesting behind-the-scenes code
happening to get the I/O into the queue (which, strangely enough, happen
if you are queuing an I/O on a mailbox, or any other high or low speed
device).  If you call QIO (the asynch version) and don't somehow
synchronise the reuse of your IOSB and buffer, anything can happen.

If you're asking questions, please append the appropriate punctuation to
them so as not to confuse things.

Jim
-- 
www.eight-cubed.com