A driver overhaul on 29 Feb 2000 (!) broke locking around fiddling with 
the tx descriptor ring in start_xmit(); a follow-on "fix" removed the 
broken remnants altogether.  Here's a patch to restore proper locking in 
the function -- the complement in the interrupt handler has been correct 
all the time.

 This *may* have been the reason for the occasional confusion of the chip 
-- triggering a tx timeout followed by a chip reset sequence -- seen on 
R4k-based DECstations with the onboard Ethernet interface.  Another theory 
is the confusion is due to an unindentified problem -- perhaps a silicon 
erratum -- associated with the variation of the MT ASIC used to interface 
the R4k CPU to the rest of the system on these computers; with its 
aggressive write-back buffering the design is particularly weakly ordered 
when it comes to MMIO (in the absence of ordering barriers uncached reads 
are allowed to bypass earlier uncached writes, even if to the same 
location), which may trigger all kinds of corner cases in peripheral 
hardware as well as software.

Either way this piece of code is buggy.
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Clarify calling context and return codes of callback methods, and
add a description of the _cmsg structure and helper functions.

Impact: documentation
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>

The dereferencing of the private pointer cmsg->m in capi_cmsg2str() may
cause an Oops in case of an error, which is particularly inconvenient
as that function is typically used to format an error message. Add a
NULL pointer check to avoid this.

Impact: error handling improvement
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add kerneldoc comments for the exported funtions in capiutil.c.

Impact: documentation
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>

Change the name of the Kernel CAPI exported function capi_ctr_reseted()
to something representing its purpose better.

Impact: renaming, no functional change
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>

It is possible for tun_chr_close to race with dellink on the
a tun device.  In which case if __tun_get runs before dellink
but dellink runs before tun_chr_close calls unregister_netdevice
we will attempt to unregister the netdevice after it is already
gone.  

The two cases are already serialized on the rtnl_lock, so I have
gone for the cheap simple fix of moving rtnl_lock to cover __tun_get
in tun_chr_close.  Eliminating the possibility of the tun device
being unregistered between __tun_get and unregister_netdevice in
tun_chr_close.
Signed-off-by: Eric W. Biederman <ebiederm@aristanetworks.com>
Tested-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state)) was being hit
during e100 EEH recovery. The problem source was a napi_enable
call being made during e100_io_error_detected. Napi should remain
disabled after e100_down, and only be reenabled when the interface
is recovered.

This patch also updates e100_io_error_detected in order to make
it similar to the current versions of the error_detected callback
in drivers such as e1000e and ixgbe.
Signed-off-by: Andre Detsch <adetsch@br.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

vfree() does its own 'NULL' check, so no need for check before
calling it.
Signed-off-by: Figo.zhang <figo1802@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

If mlx4_create_eq() would fail for one of EQ's assigned for
completion handling, the code would try to free the same EQ
we failed to create.
The crash was found by Christoph Lameter
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use the right structure while incrementing the offset in tun_get_user.
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Increment the iovec base by the offset passed in for the initial
copy_to_user() in memcpy_to_iovecend().
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

I am working on enabling UFO between KVM guests using virtio-net and i have
some patches that i got working with 2.6.30-rc8. When i wanted to try them
with net-next-2.6, i noticed that virtio-net is not working with that tree.

After some debugging, it turned out to be several bugs in the recent patches
to fix aio with tun driver, specifically the following 2 commits.

http://git.kernel.org/?p=linux/kernel/git/davem/net-next-2.6.git;a=commitdiff;h=0a1ec07a67bd8b0033dace237249654d015efa21
http://git.kernel.org/?p=linux/kernel/git/davem/net-next-2.6.git;a=commitdiff;h=6f26c9a7555e5bcca3560919db9b852015077dae

Fix the call to memcpy_from_iovecend() in skb_copy_datagram_from_iovec
to pass the right iovec offset.
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

skb_dma_unmap() is quite expensive for small packets,
because we use two different cache lines from skb_shared_info.

One to access nr_frags, one to access dma_maps[0]

Instead of dma_maps being an array of MAX_SKB_FRAGS + 1 elements,
let dma_head alone in a new dma_head field, close to nr_frags,
to reduce cache lines misses.

Tested on my dev machine (bnx2 & tg3 adapters), nice speedup !
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Get rid of num_dma_maps in struct skb_shared_info, as it seems unused.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Remove the private definition of TRUE/FALSE and use the ones from
linux/stddef.h. Also remove the definition of BOOL which is not referenced
inside the driver anyway.
Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
Signed-off-by: David S. Miller <davem@davemloft.net>

This is a driver for the Micrel KS8842 ethernet switch.

The supplied code is for driving the KS8842 through the Timberdale FPGA
on the Russellville board, a development board for Intel Atom CPU
in the automotive area.
Signed-off-by: Richard Röjfors <richard.rojfors.ext@mocean-labs.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

On Thu, Jun 04, 2009 at 09:06:00PM +1000, Herbert Xu wrote:
>
> tun: Optimise handling of bogus gso->hdr_len
>
> As all current versions of virtio_net generate a value for the
> header length that's too small, we should optimise this so that
> we don't copy it twice.  This can be done by ensuring that it is
> at least as large as the place where we'll write the checksum.
>
> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

With this applied we can strengthen the partial checksum check:

In skb_partial_csum_set we check to see if the checksum offset
is within the packet.  However, we really should check that it
is within the skb head as that's the only bit we can modify
without copying.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

As all current versions of virtio_net generate a value for the
header length that's too small, we should optimise this so that
we don't copy it twice.  This can be done by ensuring that it is
at least as large as the place where we'll write the checksum.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

Through a bug in the tun driver, I noticed that virtio_net is
producing bogus hdr_len values.  In particular, it only includes
the IP header in the linear area, and excludes the entire TCP
header.  This causes the TCP header to be copied twice for each
packet.  (The bug omitted the second copy :)

This patch corrects this.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

Fixes an initialization error; the chip negotiates gigabit, but
the driver mistakenly handled it as 100Mb.
Changes based on both SiS own GPL driver and forcedeth.
Hopefully should fix
	http://bugzilla.kernel.org/show_bug.cgi?id=9735
	http://bugzilla.kernel.org/show_bug.cgi?id=11149Signed-off-by: Riccardo Ghetta <birrachiara@tin.it>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Riccardo Ghetta <birrachiara@tin.it>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Riccardo Ghetta <birrachiara@tin.it>
Signed-off-by: David S. Miller <davem@davemloft.net>

Corrected dma sync handling on small packets. Should fix
http://bugzilla.kernel.org/show_bug.cgi?id=11509
Note:
While this bug was reported only on x86_64, it could have affected
any architecture.
Signed-off-by: Riccardo Ghetta <birrachiara@tin.it>
Signed-off-by: David S. Miller <davem@davemloft.net>

Some firmware mailbox commands require the firmware to
communicate with the FCoE driver running on another
PCI function.  This can potentially take several seconds.
This wait is done in process context only.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There are 8 banks of 'sub-registers' each of which are accessed
through address/data register pair.  An example would be reading
flash or the xgmac.  Accessing these require the driver to wait for
a ready bit before writing the address and then accessing the data.
This patch increases the timeout to 100us to prevent timeouts
that have been seen on some platforms.
These register are accessed in process context only.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Since an FCoE function shares a port with this NIC function, the
jumbo settings must always be in place.  This patch causes the
hardware to be set up for jumbo if it is not already done.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Each firmware mailbox command can have a different number of
valid data elements. When waiting for a mailbox command to
complete it the process passes it's element count and
waits for the completion.  It is possible that while waiting
an unrelated firmware async event (AE) can arrive.  When this
happens, the handler will over write the element count
with the value for the newly arrived AE. This can cause
the mailbox command to not get all of it's data.
This patch restores original mailbox count at the end
of the handler.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Check status on every iteration of event handler loop and exit
if an error occurred.  If an error occurred then recover process
will be queued so this loop should no continue.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The smsc911x_drv_remove() function is declared with __devexit, so the
assignment to the driver structure needs __devexit_p() wrappings to prevent
build failure when hotplug is disabled.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The smsc911x driver works fine on Blackfin systems, so add it to the arch
list in the Kconfig.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

With all the last patchsets going in for 82599 feature enablement, the
driver version needs to be increased for better identification.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

82599 has a FW running that helps manage the internal SFI PHY.  There are
also a number of pre-production samples of these SFI-based devices in the
field.  This patch adds a check to look for the firmware running on all
SFP+ based adapters, and displays a warning to the system log if a
pre-production adapter is identified.  The driver will continue loading
though, with no functional degradation.

Also remove an old function prototype from ixgbe_82599.c that isn't being
used.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch rearranges the ixgbe_ring struct to make better use of
cacheline efficiency.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch enables Flow Director's ATR functionality to the main base
driver for 82599.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Acked-by: Mallikarjuna R Chilakala <mallikarjuna.chilakala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds the functions for ixgbe to initialize Flow Director.  It
also has the function APIs to add Flow Director filters from the base
driver.  This also includes ATR, Application Targeted Routing, which is a
feature set of Flow Director.  This is the hash-based mechanism to
automatically identify flows and add filters based on the hash, and direct
the Rx of that flow back to that same CPU.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Flow Director is a Rx filter mechanism designed to match Rx flows back to
the same CPU that the flow's Tx occurred from.  This patch adds the first
piece, the defines for the hardware registers, to enable this feature in
82599 hardware.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This adds a second feature flag variable to use for future feature
expansion.  Add HW RSC to this new feature flags variable.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch moves the tx cleanup processing out of the MSI-X interrupt
processing and gives it it's own napi routine.  This allows the driver to
process TX cleanup in a polling context instead of in an interrupt context
which prevents TX from starving RX.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The v_idx value was being used as both a bitmask and an index.  This change
makes it so that the q_vector contains the index and allows for much of the
code to be simplified since disabling a q_vector involves only clearing one
bit in the interrupt bitmask.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>