The check for Yukon lite changes was restricting itself to
rev A3. It turns out that these changes are also true on A4
and later.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jgarzik@pobox.com>

Cleanup the phy_lock deadlock because of relocking in the nway_reset path.
Reported by Francois Romieu.

Also, don't need to do irqsave/restore for blink,
just excluding bh is good enough.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jgarzik@pobox.com>

Here is a fix for a typo, thanks Eliot Dresselhaus.
Since transmitter not active when device is down, it wasn't really noticed.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jgarzik@pobox.com>

The SK-9E boards use the Marvell Yukon2 chipset which
is not supported by the skge driver. Thanks to Ralph Roesler
for noticing.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jgarzik@pobox.com>

Using Genesis board, I get harmless error reports. Rather than console
error, turn it into a error counter.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jgarzik@pobox.com>

Few more u32 vs. pm_message_t fixes.
Signed-off-by: NPavel Machek <pavel@suse.cz>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Bump version and copyright year.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Optimize the receive buffer management code to replenish the
buffers immediately (like tg3).
Signed-off-by: NStephen Hemmminger <shemminger@osdl.org>

Need to handle receive and transmit packet arbiter timeouts.
Transmit arbiter timeouts happens when Gigabit sends to 100Mbit port
on same switch and pause occurs.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Cleanup messages (for debug) about PHY interrrupts, because when
user can't get driver working that is often the problem.
Use a consistent way of enabling interrupts by port.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Rewrite the code for handling the Broadcom PHY to something that
works. Remove link polling because Broadcom and Yukon don't need it.
When I wrote initial code, didn't have a genesis chipset based
board to test, so it was a non-working guess.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Unify mapping of supported modes based on hardware.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Remove support for the non-Broadcom genesis based boards. The code
is untested, and probably won't work as is. The newer boards are all
Yukon based, and only old Genesis board I can find uses Broadcom.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Remove the bits and pieces added relating to Yukon II chipset.
The Yukon 2 will be in a separate driver.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Get rid of definitions for chip versions and PHY chips that
this driver does not support.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

The Yukon chips don't support TSO, and Yukon2 will be a separate driver.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Replace inline accessor functions for chip revision and number of ports
with simple structure members.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Rather than accessing PCI config space through MMIO space, use the
standard PCI functions.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

The inlines and macro's needed some cleanup's and fixes:
 * change name of macro SKGEMAC_REG to SK_REG to better reflect usage
   and fix comments
 * ditto for SK_GEXM_REG -> SK_XMAC_REG and SKGEGMA_REG -> SK_GMA_REG

 * change skge_gm_ to just gm_ since it is just a local function and long
   names look ugly.
 * change skge_xm_ to just xm_
 * fix xm_write32 to write as two u16's with correct byte order
 * fix xm_outaddr to correctly use offset
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Use PCI_DEVICE() macro.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Cleanup whitespace around if() and switch() and end of lines
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

	ptrdiff_t is %td, not %d
Signed-off-by: NAl Viro <viro@parcelfarce.linux.theplanet.co.uk>

Signed-off-by: NAl Viro <viro@parcelfarce.linux.theplanet.co.uk>