On Thu, 23 Jan 2003 14:06:36 -0700
Steven Dake <[EMAIL PROTECTED]> wrote:
>
> I cant speak about OS/X, but I have crashed windows several times
> (BSOD) while hot removing a USB SCSI CDROM.
You can crash 98 and NT by simply removing a CD without telling them...
Hi there
I now found out that in order to get high speeds with bulk transfer you
need to submit lots of data at a time and you also need to submit lots
of urbs at the same time.
So I'm thinking use the next field and make a circle que with some urbs.
Ok so each time the complete method is calle
Oliver:
In case my message from last Friday fell through the cracks, I'm resending
it. This patch should do what you want:
Calls to usb_unlink_urb() while the completion handler is running
won't do anything until the handler has returned.
Synchronous calls to usb_unlink
Heinrich du Toit wrote:
So I'm thinking use the next field and make a circle que with some urbs.
Don't use urb->next, it never worked for anything except ISO and
is gone in 2.5 anyway ... since it was always possible to just
have completion handlers resubmit, if you want "ring" behavior.
Ok.
On Sat, Jan 25, 2003 at 01:09:28PM -0800, Matthew Dharm wrote:
> This patch does the following:
> (o) Add comments showing what needs to be done to complete the hot-unplug
> system.
> (o) Add a BUG_ON() for (what is now) a critical failure case.
> (o) Make certain that a debug print happens eve
They should go on top of what I sent you before.
Didn't you get the patch I sent which should make things work for you?
You need (in order):
(o) clean 2.5.59
(o) the patches that didn't work for you
(o) the patch that fixes that
(o) the set of patches we're talking about now
Matt
On Mon, Jan 27
I've have written a driver for Creative's extigy usb sound card, based on
the generic audio.c driver. The card uses usb interrupts to signal events
such as button presses on the unit and MIDI messages.
When the host driver gets the interrupt, the event code can be read by
using the GET_MEM reque
David Brownell wrote:
Heinrich du Toit wrote:
So I'm thinking use the next field and make a circle que with some urbs.
Don't use urb->next, it never worked for anything except ISO and
is gone in 2.5 anyway ... since it was always possible to just
have completion handlers resubmit, if you wan
I really need to get 12Mbit/s out of this thing.
Easy. Do your math, with the variables being more
or less
- buffer size for each queue entry
- irq latency (can't be decreased below 1 uframe)
- queue depth
- 13 bulk packets of 512 bytes per uframe
- 8000 uframes/sec
What's the transfer
