Re: Linux-Support for IBM PPC405EP's EMACs

Debbie Xu wrote:

Hi Todd,

I found some confusing code for the PPC405EP's EMACs at the Linux kernel
(2.4.21-pre2). It is common to both EMACs.

When the system is booting up, the ocp_enet_open() (in
drivers/net/ibm_ocp/ibm_ocp_enet.c) is called to open the 2 EMACs, one
at a time. However, the link status of the Ethernet ports, when it is
reported in the function mii_display_status() (in
drivers/net/ibm_ocp/ibm_ocp_phy.c),is controlled by a "link" variable in
an ocp_enet_private structure. Once a PHY is found at the probing time,
this "link" variable stays as "1" ("up" status) always. BUT, it doesn't
care if that Ethernet port is connected to any network or not. This
"link" status only reflects whether this EMAC is connected to a PHY
device or not. The definition of the ocp_enet_private structure (at
drivers/net/ibm_ocp/ibm_ocp_enet.h) doesn't tell much of the nature of
the"link" variable either. So, what shall we expect from this status
reporting? I tend to think that a link is only "up" when it's connected
on a network.

Can you share your ideas on this?

Hi, the code in question predates the recent 405EP changes, and I
haven't looked at it recently.  I have noticed that the link status for
uncabled EMACs was defaulted to "up" (and with other values such as 10BT
half-duplex).  There's a rewrite of this driver in the works so maybe
all this will change soon, otherwise it's something to be cleaned up
someday.


--
Todd


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/