[uClinux-dev] IxEthAccMACRecoveryMessageCallback

[Doug Kehn]

Hi All,

--- Doug Kehn wrote:

> Hi All,
> 
> I'm seeing the following error message coming out of
> Intel Access Library v2.4:
> 
> [error] IxEthAccMACRecoveryMessageCallback triggered
> for port ID = 0.
> 
> It appears that the Ethernet MAC is resetting; the
> question is why.  Has anyone else observed this?
> 
> 

I received the following explanation from Intel.

---8<-----

Ethernet MAC engine would perform infinite retries on
the wire on a dropped packet. The dropped packet could
be due to late collision or due to number of retries
has exceeded in the early collision. This may cause
the Ethernet MAC engine to become stuck due to its
behaviour of infinitely retrying and resulting the Tx
FIFO overflowing. When this happens, MAC will
continuously assert TX_EN signal to PHY. This will
trigger PHY to enter jabber mode and de-asert TX_EN.
But the Ethernet MAC will still be in the lockup state
resulting in Ethernet transmit services no longer
running. This lockup condition is not recoverable
unless the MAC is being reset. so this when macy
recovery will happen.  In general, excessive colisions
beyond the Ethernet retry limit is indicative of
improper network design and configuration. Customers
should take precaution to alleviate the occurances of
such excessive network congestion.

---8<-----

Regards,
...doug


      ____________________________________________________________________________________
Looking for last minute shopping deals?  
Find them fast with Yahoo! Search.  http://tools.search.yahoo.com/newsearch/category.php?category=shopping