[U-Boot] U-boot 2010.12 Ping failed on DP83849C

[U-Boot] U-boot 2010.12 Ping failed on DP83849C

[Jerry Wang]

I have a custom MPC8313ERD board loaded with u-boot 1.1.6 and the network PHY is 
NetSemi DP83849C. The PHY info is copied from DP83865 and the network works 
fine.

Recently I am trying to upgrade u-boot to 2010.12 but I got Ping failed and host 
is not alive message when I ping the same host.
After enabling debug, it shows:
       Trying TSEC0        
        Speed: 10, half duplex
        Using TSEC0 device
        sending ARP for 0a350275
        ARP broadcast 1
        TSEC0: tsec: tx error
        ARP broadcast 2
        TSEC0: tsec: tx buffers full
        ping failed; host 10.53.2.117 is not alive

    After googling the problem, I learned this may be related to the some kind 
of bus conflict when DMA move ip packet to transmit bugger.
   My question is what caused this error in 2010.12 version and not in 1.1.6?

[U-Boot] U-boot 2010.12 Ping failed on DP83849C

[Kim Phillips]

On Fri, 6 May 2011 08:05:18 -0700
Jerry Wang <wang3522@yahoo.com> wrote:

> I have a custom MPC8313ERD board loaded with u-boot 1.1.6 and the network PHY is 
> NetSemi DP83849C. The PHY info is copied from DP83865 and the network works 
> fine.
> 
> Recently I am trying to upgrade u-boot to 2010.12 but I got Ping failed and host 
> is not alive message when I ping the same host.
> After enabling debug, it shows:
>        Trying TSEC0        
>         Speed: 10, half duplex
>         Using TSEC0 device
>         sending ARP for 0a350275
>         ARP broadcast 1
>         TSEC0: tsec: tx error
>         ARP broadcast 2
>         TSEC0: tsec: tx buffers full
>         ping failed; host 10.53.2.117 is not alive
> 
>     After googling the problem, I learned this may be related to the some kind 
> of bus conflict when DMA move ip packet to transmit bugger.
>    My question is what caused this error in 2010.12 version and not in 1.1.6?

there are almost 10,000 commits between those two revisions, not to
mention any local changes in your tree...git bisecting that might
take some time.

A more direct approach would be to try the top of tree, which has
the new phy lib that includes mdio commands for peeking and poking
phys.  You'd need to fill in the .uid, .mask, and .features values
when adding the DP83849C driver though.

Kim

diff --git a/drivers/net/phy/natsemi.c b/drivers/net/phy/natsemi.c
index ea60ac1..8900292 100644
--- a/drivers/net/phy/natsemi.c
+++ b/drivers/net/phy/natsemi.c
@@ -88,9 +88,20 @@ static struct phy_driver DP83865_driver = {
        .shutdown = &genphy_shutdown,
 };
 
+static struct phy_driver DP83849C_driver = {
+       .name = "NatSemi DP83849C",
+       .uid = 0xXXXXXXXX,
+       .mask = 0xYYYYYYYY,
+       .features = PHY_GBIT_FEATURES,
+       .config = &dp83865_config,
+       .startup = &dp83865_startup,
+       .shutdown = &genphy_shutdown,
+};
+
 int phy_natsemi_init(void)
 {
        phy_register(&DP83865_driver);
+       phy_register(&DP83849C_driver);
 
        return 0;
 }

[U-Boot] U-boot 2010.12 Ping failed on DP83849C

[Detlev Zundel]

Hi Jerry,

> I have a custom MPC8313ERD board loaded with u-boot 1.1.6 and the network PHY is 
> NetSemi DP83849C. The PHY info is copied from DP83865 and the network works 
> fine.
>
> Recently I am trying to upgrade u-boot to 2010.12 but I got Ping failed and host 
> is not alive message when I ping the same host.
> After enabling debug, it shows:
>        Trying TSEC0        
>         Speed: 10, half duplex
>         Using TSEC0 device
>         sending ARP for 0a350275
>         ARP broadcast 1
>         TSEC0: tsec: tx error
>         ARP broadcast 2
>         TSEC0: tsec: tx buffers full
>         ping failed; host 10.53.2.117 is not alive
>
>     After googling the problem, I learned this may be related to the some kind 
> of bus conflict when DMA move ip packet to transmit bugger.
>    My question is what caused this error in 2010.12 version and not in 1.1.6?

Those versions are so far away codewise, that I there surely is no easy
answer.  _If_ your code was in mainline all the way from 1.1.6 to
2010.12 you could have used "git bisect", but as this is (very likely)
not the case, this will not work for you.  So the "easiest" way is to
start debugging the problem yourself.

Best wishes
  Detlev Zundel

-- 
"Oh, didn't you know, the Lord did the original programming of the universe in
COBOL." - "That's why the world is the evil work of Satan. A true divine being
would have used Scheme."  -  "And, if so, Jesus would have been crucified on a
big lambda symbol."  -- K. Chafin, K. Schilling & D. Hanley, on comp.lang.lisp
--
DENX Software Engineering GmbH,      MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich,  Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-40 Fax: (+49)-8142-66989-80 Email: dzu at denx.de