The mesh_path_node_copy() can be called like this:
mesh_path_add
 `- write_lock(&pathtbl_resize_lock); /* ! */
 `- mesh_table_grow
     `- ->copy_node
           `- mesh_path_node_copy

thus, the GFP_KERNEL is not suitable here.

The acceptable fix, I suppose, is make this allocation GPF_ATOMIC -
the mpath_node being allocated is 4 pointers, i.e. this allocation
is small enough to survive even under a moderate memory pressure.
Signed-off-by: NPavel Emelyanov <xemul@openvz.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Now - return the -ENOMEM in case kmalloc fails.
Signed-off-by: NPavel Emelyanov <xemul@openvz.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

The mesh_path_node_copy() performs kmalloc() and thus - may fail
(well, it does not now, but I'm fixing this right now). Its caller -
the mesh_table_grow() - isn't prepared for such a trick yet.

This preparation is just flush the new hash and make copy_node()
return an int value.
Signed-off-by: NPavel Emelyanov <xemul@openvz.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

(This set applies OK without the previous one of 4 patches,
 but with some fuzz in the 7th one)

The mesh_path_node_free() does so under hashwlock.

But, this one is called
1. from mesh_path_add() after an old hash is hidden and
   synchronize_rcu() is calld
2. mesh_pathtbl_unregister(), when the module is being
   unloaded and no devices exist to mess with this hash.

So, it seems to me, that simply removing the call is OK.
Signed-off-by: NPavel Emelyanov <xemul@openvz.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

There are already tree paths, that do incremental rollbacks, so
merge them together, rename labels and format the code to look a
bit nicer.

(I do not mind dropping/delaying this patch however).
Signed-off-by: NPavel Emelyanov <xemul@openvz.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This will create the following entry:
/sys/class/net/mshX
            -- boot_options
            |   |-- ...
            |   `-- channel
            ...

... which I overlooked on my previous patch.
Signed-off-by: NJavier Cardona <javier@cozybit.com>
Signed-off-by: NDavid Woodhouse <dwmw2@infradead.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Signed-off-by: NDavid Woodhouse <dwmw2@infradead.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This will create the following sysfs directories:
/sys/class/net/mshX
		...
		|-- boot_options
		|   |-- bootflag
		|   `-- boottime
		...
		|-- mesh_ie
		|   |-- capability
		|   |-- mesh_id
		|   |-- metric_id
		|   `-- protocol_id
Signed-off-by: NJavier Cardona <javier@cozybit.com>
Signed-off-by: NDavid Woodhouse <dwmw2@infradead.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This patch is based on a patch from Shailendra Govardhan and Brian Cavagnolo.
It extends the MESH_CONFIG command to configure non-volatile parameters on
libertas devices that support them (e.g. OLPC Active Antenna).

This patch only implements the driver/firmware interface.

See http://dev.laptop.org/ticket/6823 for minimal testing results and known
issues.
Signed-off-by: NJavier Cardona <javier@cozybit.com>
Signed-off-by: NDavid Woodhouse <dwmw2@infradead.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

We weren't properly handling multicast on the mesh interface. Fix that,
which involves setting up the hardware to use the union of dev->mc_list
for both eth%d and msh%d devices.

This means we can't do it directly from ->set_multicast_list() because
we'd need to lock the other device to read its list, and we can't do
that because it might deadlock. So punt the actual work to keventd.

Also, invoke the same when taking an interface down; for some reason the
core calls ->set_multicast_list while IFF_UP is still set in dev->flags
when we're taking it down, so its addresses don't get removed then.

We also convert MAC_MULTICAST_ADR to a direct command while we're at it,
removing one more entry from the big switch statement in the deprecated
lbs_prepare_and_send_command() function.
Signed-off-by: NDavid Woodhouse <dwmw2@infradead.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Using the deprecated lbs_prepare_and_send_command() function for a
command which it doesn't understand is an error; complain loudly about
it even when we're not debugging.

The mesh stats bug, where we converted MESH_ACCESS to a direct command
but accidentally missed one user which was still trying to do it through
lbs_prepare_and_send_command(), would have been caught a lot quicker if
we'd done this sooner. Such bugs aren't entirely unlikely in future too,
as we convert more code to stop using this function.
Signed-off-by: NDavid Woodhouse <dwmw2@infradead.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

With the pending removal of the tx_control structure
we can merge the RX and TX entry private data structure
in advance. This will temporarily increase the required
memory for the queue, but that overhead will only be limited.
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

the rxdesc structure is properly memsetted before passed to
the driver. This means we don't have to reinitialize the flags
and dev_flags fields in the drivers again.
This will prevent problems when the rxdone handler is adding
flags in a earlier status and will make the code look nicer
when we are adding more read attributes in the rxdone handler
in the driver.
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Split rt2x00lib_write_tx_desc() up into a TX descriptor initializor
and TX descriptor writer.

This split is required to properly allow mac80211 to move its
tx_control structure into the skb->cb array.
The rt2x00queue_create_tx_descriptor() function will read all tx control
information and convert it into a rt2x00 TX descriptor information structure.
After that function is complete, we have all information we needed from the
tx control structure and are free to start writing into the skb->cb array
for our own purposes.
rt2x00queue_write_tx_descriptor() will be in charge of really sending
the TX descriptor to the hardware and kicking the TX queue.
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

In preparation of replacing the statically allocated data DMA buffers with DMA-mapped
skb's we need to change the TXD handling of the PCI drivers, by moving the programming
of the buffer address fields to the actual TXD writing at TX time, instead of at start-up
time.
Signed-off-by: NGertjan van Wingerde <gwingerde@kpnplanet.nl>
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Due to usage of memmove() in rt2x00usb the descriptor can become
corrupted because it is being overwritten by the data part.
Overall having the descriptor in front of the frame is a bad idea,
we can however use the skb->cb array for this task, since that
contains more then enough room to hold the entire descriptor and
preserve the information long enough.

After this we can also cleanup the alignment code a bit to make it
work a bit more flexible to allow for all kinds of odd header lengths.
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

With the integration of the mac80211 multiqueue patches it has become possible that the
mac80211 layer modifies the number of TX queues that is stored inside the ieee80211_hw
structure, especially when multi-queue is not selected.

The rt2x00 drivers are not well suited to handle that situation, as they allocate the
queue structures before mac80211 has modified the number of queues it is going to use,
and also expect the number of allocated queues to match the hardware implementation.

Hence, ensure that rt2x00 maintains by itself the number of queues that the hardware
supports, and, at the same time, making is not dependent on the preservation of contents
inside a mac80211 structure.
Signed-off-by: NGertjan van Wingerde <gwingerde@kpnplanet.nl>
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Move the last remaining information details read from ieee80211_tx_control
in the drivers to the txentry_desc structure. After this we can
remove ieee80211_tx_control from the argument list for the callback function,
which makes it easier when the control information is moved into skb->cb
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

The tx_status enumeration was broken since the introduction
of rt61pci. That driver uses different values to report the
status of the tx action.
This would lead to frames that were reported as success but
actually failed to be send out, or frames that were neither
successfull or failure which were reported as failure.

Fix this by change the TX status reporting and more explicitely
check for failure or success. Note that a third possibility is
added "unknown". Not all hardware (USB) can report the actual
TX status, for rt61pci some frames will receive this status
because the TXdone handler is never called for those frames.
This unknown will now be handled as neither success or failure,
so we no longer increment the failure counter while this conclusion
could not be determined from the real status of the frame.
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Remove frame_type from skb_frame_desc and pass it
as argument to rt2x00debug_dump_frame().

Change data_len and desc_len to unsigned short
to save another 4 bytes in skb_frame_desc. Note that
this was the only location where the data_len and
desc_len was not yet treated as unsigned short.

This trim is required to help mac80211 with adding
the TX control and TX status informtation into the
skb->cb structure. When that happens, drivers will
have approximately 40 bytes left to use freely.
Signed-off-by: NIvo van Doorn <IvDoorn@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This patch contains a set of cosmetic changes to TIPC's network
topology service subsystem, including:

- updates to comments (including copyright dates)
- re-ordering structure fields to group them more logically
- removal of optional debugging code that is no longer required
- minor changes to whitespace to conform to Linux coding conventions
Signed-off-by: NAllan Stephens <allan.stephens@windriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch modifies TIPC's network topology service so that it
only requires a single reference table entry per subscriber
connection, rather than two.  This is achieved by letting the
reference to the server port communicating with the subscriber
act as the reference to the subscriber object itself.  (Since
the subscriber cannot exist without its port, and vice versa,
this dual role for the reference is perfectly natural.)  This
consolidation reduces the size of the reference table by 50%
in the default configuration.
Signed-off-by: NAllan Stephens <allan.stephens@windriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch fixes TIPC's topology server so that it does byte swapping
correctly when endianness conversion is required.  (Note: This bug only
impacted an application if it issues a subscription request to a
topology server on another node, rather than the server on it's own
node; since the topology server is normally not accessible by off-node
applications, most TIPC applications were not impacted by the bug.)
Signed-off-by: NAllan Stephens <allan.stephens@windriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch enables TIPC's topology server code to do customized
endianness conversions on a per-subscription basis.  (This
capability is needed to support the upcoming consolidation of
subscriber and subscription object references.)
Signed-off-by: NAllan Stephens <allan.stephens@windriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch enables TIPC's topology server code to do customized
overlap detection handling on a per-subscription basis.  (This
capability is needed to support the upcoming introduction of
multi-cluster TIPC networks.)
Signed-off-by: NAllan Stephens <allan.stephens@windriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Instead of assigning values for the struct cpu_reg's at runtime,
we already know these values at compile time.  Therefore, we can use
designated initializers, to initialize these structures and not have
to incur this assignment cost at run-time.
Signed-off-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

To make the bnx2 code more consistent, all instances of
RX_COPY_THRESH have been changed to BNX2_RX_COPY_THRESH.
Signed-off-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The rx_offset field is set to a constant value and initialized
only once.  By replacing all references to the rx_offset field,
we can eliminate rx_offset from the bnx2 structure.  This will
save 4 bytes for every bnx2 instance.

[Added parentheses to the definition of BNX2_RX_OFFSET, as noted
by Ben Hutchings.]
Signed-off-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add PCI recovery functions to the driver.  The initial pci state is
also saved so the the MSI state can be restored during PCI recovery.
Signed-off-by: NWendy Xiong <wendyx@us.ibm.com>
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Andy Gospodarek <andy@greyhouse.net> found that netconsole would
panic when resetting bnx2 devices.

>From Andy:
"The issue is the bnx2_set_link in bnx2_init_nic will print a link-status
message before we are fully initialized and ready to start polling.
Polling is currently disabled in this state, but since the
__LINK_STATE_RX_SCHED is overloaded to not only try and disable polling
but also to make the system aware there is something waiting to be
polled, we really have to fix this in drivers.

The problematic call is the one to netif_rx_complete as it tries to
remove an entry from the poll_list when there isn't one."

While this netconsole problem should be fixed separately, we really
should not reset the PHY when changing ring sizes, MTU, or other
similar settings.  The PHY reset causes several seconds of unnecessary
link disruptions.
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Acked-by: NAndy Gospodarek <andy@greyhouse.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Conflicts:

	drivers/net/wireless/iwlwifi/iwl-4965-rs.c
	drivers/net/wireless/rt2x00/rt61pci.c

cdebug_init() is called from kcapi_init() which is module
initialization function, so it must return negative values on errors.
Signed-off-by: NMarcin Slusarz <marcin.slusarz@gmail.com>
Acked-by: NKarsten Keil <kkeil@suse.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Time is unsigned long (except when you are in a hurry) so we need to
store rx_tmp_jif in the right sized object.
Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

With the cli/sti code sorted out we think this driver is OK for use on
SMP systems.
Acked-by: NMark Asselstine <mark.asselstine@windriver.com>
Acked-by: NKarsten Keil <kkeil@suse.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The use of cli()/sti() within the do/while was a way to ensure
interrupts were only disabled for short periods of time while the bulk
of the time interrupts were free to occur.  The use of the spin lock
has eliminated the need to play with interrupts in this way while
still allowing for IO to be protected.

The remaining 3 sti() calls seem unneeded now that at no other point
in the driver is there a call to cli().
Signed-off-by: NMark Asselstine <mark.asselstine@windriver.com>
Acked-by: NKarsten Keil <kkeil@suse.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The atm_tcp.h uses types from linux/atm.h, but does not include it.
It should also use the standard __u## types from linux/types.h rather
than the uint##_t types since the former can be found with the kernel
already.

Same goes for linux/atm.h.  The linux/socket.h include there also gets
dropped as atm.h does not actually use anything from socket.h.
Signed-off-by: NMike Frysinger <vapier@gentoo.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>