[Info-vax] QIO Writes to a mailbox

[George Cornelius]

JF Mezei wrote:
> But when QIO executes, doesn't it invoke some driver code to "queue" the
> IO, and isn't it possible that for a write to memory device, that drive
> code might actually do the write to the mailbox device ?

Yes.  But you have already given a scenario where that will not work, and
it's bad form in general to rely on a 'start i/o' to actually do the i/o.
There's a discipline to using $QIO and $QIOW and it is best to follow it
in all cases instead of trying to shortcut it because you have inside
information on how a specific driver might work. Write your code as if
you are demonstrating to others how things should be done, not as a way
to demonstrate how much inside information you have.

> And if a mailbox is full, I know that QIOW will cause process to hang on
> a RWMBX, but would a QIO also result in this ?

I don't believe the process hangs if you do a $QIO, but I am not sure.

I also do not know for sure, if you leave out the IO$M_NOW modifier,
whether VMS even bothers to copy your data to nonpaged pool instead
of waiting for a rendezvous - the point where the driver can
simultaneously map both buffers and perform a direct buffer to
buffer move operation.  This adds complexity and was probably
long ago deemed not worth the effort, especially in environments
where paging and swapping were occurring.

George