In the switch we used to have a timer to check for link status, but
in the SPEC card the internal lm32 does that, and access to the phy
is not atomic. Assume the link is always on until we'll get notification
from lm32 to the host.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This fixes an oops when programs access ethtool methods, especially
wrn_get_drvinfo. It can be reproduced by a simple run of "irqbalance",
which I never did in my development machine.
Reported-by: Grzegorz Daniluk <grzegorz.daniluk@gmail.com>
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

EOF

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

The .wb file comes from general-cores/modules/wishbone/wb_vic, commit
6041224.  I built the header with wbgen2 by running "make vic-regs.h"
in this same directory.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

- platform_data is now wrn_drvdata, fix accordingly

- include ../spec-nic.h to have wrn_drvdata available

- #if 0 irq requesting and other detatils

- some printk statements end other irrelevant stuff
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

timestamping is disabled only temporarily, hopefully it will work later

The files come from commit "wr-switch-sw-v3.0" of wr-switch-sw
project.  They are straight copies from kernel/wr_nic (there) to
kernel/wr_nic (here).  Three have not been copied, as they don't apply
here: dmtd.c, pps.c, module.c

Later patches will customize them to they compile in this environment.
Note that as of this commit they are not compiled by the Mafile yet,
this is only a verbatim copy to easily keep track of differences.

This is an exact copy of the register definitions from the wr-switch-sw
project.  The two projects share the same WR ptp-core, so most of this
applies here as well.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>

This offset used to be mandatory, but it is not any more. The gateware
I'm using has two aliases  in the 1M address space, so offset 0 is as good.
Other gateware (like the 100Msample ADC) use the whole address space,
so after this patch the ADC driver can use the fmc accessors like other
drivers.

EOF