Re[2]: does 1394bus.sys serialize async access from differen t drivers?

> No, but I can run a debugger in the attached devices. I’m getting a

timeout in the following case:
-at least one isoch stream must be running all the time.
-there are two of my devices connected to one ohci controller card (If I
connect these devices to separate controllers everything seems to work fine)
-I’m doing aync reads or writes.

From the device debug output I can see:
the RQPkt Interrupt is delayed for a very long period of time (about
about half a minute or so). The residual count in the descriptor is also
not updated before I’m getting the delayed Irq.
The packet ordering seems to be ok though (no packet is missing).

For what it is worth, I have run up to 3 devices
off of a single controller and never had any difficulty. I’ve run them
such that all 3 were engaged in isochronous transmission at the same
time or, alternately, with transmission being quickly switched from
one device to another. Lots of asynchronous traffic during the
isochronous transmission. Admittedly, I haven’t done this recently so
I’ve not tested it with XP only Win2K.

Any chance that your devices become confused and fail to respond
correctly to requests when more than one is on
the bus? This is where a packet analyzer could be very useful (wish I
had one).

By the way, I’ve found a product called DevFilter
( useful for dealing with this kind of thing.
Lets you track IRPs (and the IRPs generated from them) through the
driver chain so that you can see what is going on. This tool could
possibly help you here. You can get a trial version to try it (I don’t
remember the limits placed on it).

Robert Newton

>WPP tracing is such a pain for someone who has not done much, ok…any WMI
>:-o I have some strange preproc issues to deal with.
>I can’t wait to get WPP tracing working on Win2k.

One I got it worked, I found myself couldn’t live without it-:slight_smile:

Good luck,

Calvin Guan Software Engineer
ATI Technologies Inc.