FEC problems on MPC5200
Hi everybody,
I've got some ethernet problems with a Lite5200.
During tftp transferts, I've got a lot of timeouts. I've put a network analyzer
on my ethernet, and I can see that all is ok on the server side (all the frames
are sent). It sounds like the Lite5200 missed some frames, because it doen't send
the ACK.
Has somebody experienced same problems on a MPC5200 ?
Best regards.
Pierre
I see the same happen with my Lite5200. I get frequent timeouts when loading the kernel with TFTP.
It also happens when using the Motorola dbug boot monitor to load via TFTP. Something seems to be wrong.
This happens very rarely when I am using our custom mpc823 board to load from the same TFTP server,
but with Lite5200 it happens almost every time.
Regards,
Jon Benediktsson
Marel hf.
Quoted 2 lines · In message <[email protected]> you wrote:
I've got some ethernet problems with a Lite5200.
...
Quoted 1 line
Has somebody experienced same problems on a MPC5200 ?
Yes, we see the same problem, and "bad length" errors. It seems the problem was introduced by one of the patches lately. We're in- vestigating.
Best regards,
Wolfgang Denk
Signature · 4 lines
Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-4596-87 Fax: (+49)-8142-4596-88 Email: wd at denx.de "Summit meetings tend to be like panda matings. The expectations are always high, and the results usually disappointing." - Robert Orben
R: [U-Boot-Users] FEC problems on MPC5200
In reply to Pierre AUBERTI have the same problem with my board (using MPC8280), but I don't know if is a problem of the network (sometimes also ping fails - the switchs/routers are not configured very well, I think) and/or is a problem of driver.
Enzo
Quoted 6 lines · J?n Benediktsson
I see the same happen with my Lite5200. I get frequent timeouts when loading the kernel with TFTP.
It also happens when using the Motorola dbug boot monitor to load via TFTP. Something seems to be wrong.
I also have the problem with dbug monitor. Is there any timeouts within Linux kernel ?
Quoted 4 lines
This happens very rarely when I am using our custom mpc823 board to load from the same TFTP server,
With other boards (based on MPC860T) we don't have any timeouts.
Quoted 7 lines
but with Lite5200 it happens almost every time.
Regards,
Jon Benediktsson
Marel hf.
Regards
Pierre AUBERT <[email protected]> wrote on 19.03.2004 11:31:38:
Quoted 5 lines
Quoted 3 lines · J?n Benediktsson
I see the same happen with my Lite5200. I get frequent timeouts when loading the kernel with TFTP.
It also happens when using the Motorola dbug boot monitor to load via
TFTP.
Quoted 5 lines
Quoted 3 lines
Something seems to be wrong.
I also have the problem with dbug monitor. Is there any timeouts within Linux kernel ?
I have not observed this problem when using tftp in a Linux bash shell. By this I mean I run tftp from a bash shell, turn on trace and do a 'get' command
for a fairly large file. All the sends and acks match up and there is no observed
delay in the transfer.
Quoted 15 lines
Quoted 5 lines
This happens very rarely when I am using our custom mpc823 board to load from the same TFTP server,
With other boards (based on MPC860T) we don't have any timeouts.
Quoted 7 lines
but with Lite5200 it happens almost every time.
Regards,
Jon Benediktsson
Marel hf.Regards
Hi everybody,
Quoted 13 lines · Wolfgang Denk
Quoted 3 lines · In message <[email protected]> you wrote:
I've got some ethernet problems with a Lite5200.
...
Quoted 3 lines
Has somebody experienced same problems on a MPC5200 ?
Quoted 11 lines
Yes, we see the same problem, and "bad length" errors. It seems the problem was introduced by one of the patches lately. We're in- vestigating.
Best regards,
Wolfgang Denk
I've made a lot of tries with the MPC5200 FEC. It appears that timeouts occurs
because some ACK frames aren't sent by the MPC5200 (but all is OK in mpc5xxx_fec_send).
The problem disappears if I suppress the full duplex (in fec->eth->x_cntrl and in the reg 0 of the LXT971A). I can submit a patch.
But I have another question: if I suppress all the MII initialization sequence in cpu/mpc5xxx/fec.c, the LXT971A
reads its configuration from hardware and network is working fine at 10M and 100M on my Lite5200. The
auto-negociation works fine. Then, do we need to keep this (broken ?) init sequence ?
Is there something I have missed ?
Best regards