1/1 
|
Question regarding EP fifo full and NACK
by Unknown on Nov 22, 2004 |
Not available! | ||
|
I have created a new test for the USB 1.1 OpenCores testbench by
modifying the original "out0" task. I commented out the "fork" and "join" lines (558 and 621) so that all of the data is written to the fifo before anything is read out of it. With a fifo depth of 256, I would expect the test to pass until the total number of bytes written to the fifo is more than 256. With that said, when no_pack = 3 (actually 4 packets) and pl_size = 64, a total of 4 x 64 = 256 bytes will be written to the fifo. This will fill the fifo, but just to it's maximum depth. When this scenario occurs, the testbench fails, and I get the following error message: ERROR: ACK mismatch. Expected: d2, Got: 5a (2748140.0 ns) I have verified that I am indeed getting a NACK back. Shouldn't I be able to fill the fifo, as long as I don't try to overfill it, without generating a NACK? Regards, Dalton |
|||
|
Question regarding EP fifo full and NACK
by Unknown on Nov 22, 2004 |
Not available! | ||
|
Hi Dalton,
I have successfully implemented USB 1.1 in FPGA and I discovered the
same thing. My endpoint 1 OUT FIFO was 8 bytes deep. I discovered that
if I allowed the fifo_full signal to go high, the USB core didn't like
it and the windows application crashed. So I had to change the design
to make sure I read each byte as it is written.
Rahul
On Mon, 22 Nov 2004 17:14:32 -0500, Dalton Marris dmarris@charter.net> wrote:
I have created a new test for the USB 1.1 OpenCores testbench by
modifying the original "out0" task. I commented out the "fork" and
"join" lines (558 and 621) so that all of the data is written to the
fifo before anything is read out of it. With a fifo depth of 256, I
would expect the test to pass until the total number of bytes written to
the fifo is more than 256. With that said, when no_pack = 3 (actually 4
packets) and pl_size = 64, a total of 4 x 64 = 256 bytes will be written
to the fifo. This will fill the fifo, but just to it's maximum depth.
When this scenario occurs, the testbench fails, and I get the following
error message:
ERROR: ACK mismatch. Expected: d2, Got: 5a (2748140.0 ns)
I have verified that I am indeed getting a NACK back. Shouldn't I be
able to fill the fifo, as long as I don't try to overfill it, without
generating a NACK?
Regards,
Dalton
_______________________________________________
http://www.opencores.org/mailman/listinfo/usb
-- Regards, Rahul Kumar |
|||
|
Question regarding EP fifo full and NACK
by Unknown on Nov 23, 2004 |
Not available! | ||
|
hello,
excuse me, please.
i am trying to implement the USB1.1-core, but it just dont want to work..
the FPGA replys corrcectly with the device-descriptor to the PC-question..
but PC asks for more bytes, and dont accepts it :-(
did you have to make some changes to the orginial core, to get it working?
do you have a protokoll of the handshake between your devive and the PC?
could you please tell me the device-descriptor you used?
i would be very pleased for any informations or hints.
thank you very much
best regards
Suheel Georges
Zitat von Rahul Kumar kumarah@gmail.com>:
Hi Dalton,
I have successfully implemented USB 1.1 in FPGA and I discovered the
same thing. My endpoint 1 OUT FIFO was 8 bytes deep. I discovered that
if I allowed the fifo_full signal to go high, the USB core didn't like
it and the windows application crashed. So I had to change the design
to make sure I read each byte as it is written.
Rahul
On Mon, 22 Nov 2004 17:14:32 -0500, Dalton Marris dmarris@charter.net>
wrote:
> I have created a new test for the USB 1.1 OpenCores testbench by
> modifying the original "out0" task. I commented out the "fork" and
> "join" lines (558 and 621) so that all of the data is written to the
> fifo before anything is read out of it. With a fifo depth of 256, I
> would expect the test to pass until the total number of bytes written to
> the fifo is more than 256. With that said, when no_pack = 3 (actually 4
> packets) and pl_size = 64, a total of 4 x 64 = 256 bytes will be written
> to the fifo. This will fill the fifo, but just to it's maximum depth.
> When this scenario occurs, the testbench fails, and I get the following
> error message:
>
> ERROR: ACK mismatch. Expected: d2, Got: 5a (2748140.0 ns)
>
> I have verified that I am indeed getting a NACK back. Shouldn't I be
> able to fill the fifo, as long as I don't try to overfill it, without
> generating a NACK?
>
> Regards,
> Dalton
>
> _______________________________________________
> http://www.opencores.org/mailman/listinfo/usb
>
--
Regards,
Rahul Kumar
_______________________________________________
http://www.opencores.org/mailman/listinfo/usb
|
|||
1/1