The reason sky2 driver was locking up on transmit on the Yukon-FE chipset
is that it was misconfiguring the internal RAM buffer so the transmitter
and receiver were sharing the same space.

The code assumed there was 16K of RAM on Yukon-FE (taken from vendor driver
sk98lin which is even more f*cked up on this). Then it assigned based on that.
The giveaway was that the registers would only hold 9bits so both RX/TX
had 0..1ff for space. It is a wonder it worked at all!

This patch addresses this, and fixes an easily reproducible hang on Transmit.
Only the Yukon-FE chip is Marvell 88E803X (10/100 only) are affected.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

During the handling of the PCI error recovery sequence, the current e1000
driver erroneously blocks a device reset for any but the first PCI
function.  It shouldn't -- this is a cut-n-paste error from a different
driver (which tolerated only one hardware reset per hardware card).
Signed-off-by: NLinas Vepstas <linas@austin.ibm.com>
Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
Acked-by: NAuke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

- move definition of 'tmc' and 'br' locals closer to usage

- handle clock_rate_calc() error

- propagate errors back to upper level open routine
Signed-off-by: NJeff Garzik <jeff@garzik.org>
Cc: Krzysztof Halasa <khc@pm.waw.pl>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

We need to specify a Versatile-specific SMC_IRQ_FLAGS value or the new
generic IRQ layer will complain thusly:

No IRQF_TRIGGER set_type function for IRQ 25 (<NULL>)
Signed-off-by: NDeepak Saxena <dsaxena@plexity.net>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Nicolas Pitre <nico@cam.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

This reverts commit 4596c75c as
requested by Olaf Hering. It causes compile errors, and says Olaf:

  "This change is also wrong, the autoloading works perfect with 2.6.18,
   no need to add random PCI ids.

   See commit a0245f7a, platform devices
   have now a modalias entry in sysfs.  The network card is not a PCI
   device."
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

When rebooting with netconsole over e100, the driver shutdown code would
deadlock with netpoll.  Reduce shutdown code to a bare minimum while retaining
WoL and suspend functionality.
Signed-off-by: NAuke Kok <auke-jan.h.kok@intel.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

As BHs are off in loopback_xmit(), preemption cannot occurs, so we can
use __get_cpu_var() instead of per_cpu() (and avoid a
preempt_enable()/preempt_disable() pair)
Signed-off-by: NEric Dumazet <dada1@cosmosbay.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The minimum tx ring size must be greater than MAX_SKB_FRAGS or 3
times that on some chips with TSO bugs.
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes the implementation of the ethtool set ring parameters for the
tg3 transmit ring.  The size of tx_pending is taken into account
before doing a netif_wake_queue.  This prevents the interface from
locking up when smaller transmit ring sizes are used.
Signed-off-by: NRanjit Manomohan <ranjitm@google.com>
Acked-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We dont need a full struct net_device_stats (currently 23 long : 184 bytes on 
x86_64) per possible CPU, but only two counters : bytes and packets

We save few CPU cycles too in loopback_xmit() not updating 4 fields, but 2. 
Signed-off-by: NEric Dumazet <dada1@cosmosbay.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NBen Collins <bcollins@ubuntu.com>

Signed-off-by: NBen Collins <ben.collins@ubuntu.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This reverts earlier change that attempted to fix flow control.
Device needs to discard pause frames, otherwise it hangs after a while.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

When using flow control, the PHY needs to accept multicast pause frames.
Without this fix, these frames were getting discarded by the PHY before
doing any flow control.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Mark version, this has been a lot of patches.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Under high load it is possible to make the receiver FIFO get overloaded.
The driver/hardware recover properly, so there is no reason to fill the log
with lots of extra messages, just update counter.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

The result of flow control negotiation should not limit the next
negotiatition. If board is plugged into an old half duplex 10Mbit port,
without pause, then replugged into a gigabit port, it should negotiate
what is desired, not inherit that last negotiation.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Don't need to reset PHY twice on startup.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

The result of duplex negotiation is avaliable in the phy status
register, so use that to simplify code and avoid rereading the PHY.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

The advertising bits (from ethtool.h) fit in 16 bits.

The advertisement bits for flow control are located in
different location on fiber (1000baseX)
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

When PHY is turned off on shutdown, it causes the IRQ to get stuck on.
Make sure and disable the IRQ first, and if IRQ occurs when device
is not running, don't access PHY because that will hang.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

The workaround timer is not needed in most systems with proper IRQ
routing and by perodically waking up it adds to laptop power consumption.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Some motherboards don't implement MSI correctly. The driver handles this
but the warning is too verbose and overly cautious.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

In preparation for moving check_signature, change these users from asm/io.h
to linux/io.h
Signed-off-by: NMatthew Wilcox <willy@parisc-linux.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

WARNING: drivers/net/tulip/de2104x.o - Section mismatch: reference to .init.text:de_init_one from .data.rel.local after 'de_driver' (at offset 0x20)
WARNING: drivers/net/tulip/de2104x.o - Section mismatch: reference to .exit.text:de_remove_one from .data.rel.local after 'de_driver' (at offset 0x28)
Signed-off-by: NHelge Deller <deller@parisc-linux.org>
Signed-off-by: NKyle McMartin <kyle@parisc-linux.org>
Signed-off-by: NMatthew Wilcox <matthew@wil.cx>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Adjust the pause threshold on slower systems to keep from getting overrun.
Since FIFO is 2K bytes, don't send XON pause until there is space for a full
frame.
Signed-off-by: NJeff Garzik <jeff@garzik.org>

The pci express error handling extensions don't work unless PCI access is via
mmconfig. Otherwise, all accesses to pci config registers greater than 256 fail.
Since the sky2 driver has other ways of getting to PCI config space, it works
around this short coming, but the pci_find_ext_capablity doesn't work.

This backs out commit 91aeb3ed
Go back to hardcoding, since we know where the error registers are anyway.
Fixes  http://bugzilla.kernel.org/show_bug.cgi?id=7222Signed-off-by: NStephen Hemminger <shemmminger@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Want to be able to track downstream impact of fiber related
fixes.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Do flow control negotiation properly. Don't let auto negotiation
status limit renegotiation. Separate desired pause values from
the result of auto negotiation.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Do correct mapping of pause and duplex when using 1000BaseX fiber
versions of the board.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

The PHY interrupt from the internal fiber is getting
stuck on when the link is down. Add code to handle the
transition and mask it.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Bugfix: rx descriptor release function fails to visit
the last entry while walking receive descriptor ring.
Signed-off-by: NLinas Vepstas <linas@austin.ibm.com>
Cc: James K Lewis <jklewis@us.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

The ring buffer descriptors are DMA-accessed bidirectionally,
but are not declared in this way.  Fix this.
Signed-off-by: NLinas Vepstas <linas@austin.ibm.com>
Cc: James K Lewis <jklewis@us.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Cosmetic patch: give the variable holding the numer of descriptors
a more descriptive name, so to avoid confusion.
Signed-off-by: NLinas Vepstas <linas@austin.ibm.com>
Cc: James K Lewis <jklewis@us.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

The current code attempts to start the TX dma every time a packet
is queued. This is too conservative, and wastes CPU time. This
patch changes behaviour to call the kick-dma function less often,
only when the tx queue is at risk of emptying.

This reduces cpu usage, improves performance.
Signed-off-by: NLinas Vepstas <linas@austin.ibm.com>
Cc: James K Lewis <jklewis@us.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Remove a dummy register read that is not needed.
This reduces CPU usage notably during transmit.
Signed-off-by: NLinas Vepstas <linas@austin.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: James K Lewis <jklewis@us.ibm.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

The transmit side of the spider ethernet driver currently
places locks around some very large chunks of code. This
results in a fair amount of lock contention is some cases.
This patch makes the locks much more fine-grained, protecting
only the cirtical sections. One lock is used to protect
three locations: the queue head and tail pointers, and the
queue low-watermark location.
Signed-off-by: NLinas Vepstas <linas@austin.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: James K Lewis <jklewis@us.ibm.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>