When changing the link between 100Mbps and 1Gbps in SGMII mode it was
found out that the link would stop working.  The issue is that ECNTRL[R100]
needs to be cleared when in 1Gbps mode.  Older reference manuals didn't
require the explicitly clearing but has since been found it that it is
needed.
Signed-off-by: NLi Yang <leoli@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>
Acked-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the napi api was changed to separate its 1:1 binding to the net_device
struct, the netif_rx_[prep|schedule|complete] api failed to remove the now
vestigual net_device structure parameter.  This patch cleans up that api by
properly removing it..
Signed-off-by: NNeil Horman <nhorman@tuxdriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

gfar_poll would declare polling done once the rx queue was empty,
but the tx queue could still have packets left.

Stolen mostly from the e1000 driver.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

No clean up function is executed in the interrupt context by this patch.
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Acked-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Interface name (ex. eth0) is used as the prefix for the interrupt name,
with _rx, _tx, and _er appended to distinguish multiple interrupts on
the same interface.
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Scatter Gather support in gianfar driver to handle fragmented frames on
the transmit side.
Signed-off-by: NPoonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The patch which fixed gianfar so it drops packets when it runs out
of memory left in the code which frees the skb when it drops packets.
Change the code so that we only free the skb if the new skb was successfully
created.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Whenever we want to update the status field in a BD, we usually want to
update the length field, too.  By combining them into one 32-bit field, we
reduce the number of stores to memory shared with the controller, and we
eliminate the need for order-enforcement, as the length and "READY" bit are
now updated atomically at the same time.
Signed-off-by: NDai Haruki <Dai.Haruki@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This code is based strongly on code from Dai Haruki <Dai.Haruki@freescale.com>.

The gianfar Buffer Descriptors are arranged in a circular array, the end of
which is denoted by setting the "WRAP" bit in the descriptor.  However, the
software knows the end of the ring because it knows how many descriptors are
there.  Rather than check each descriptor for whether the WRAP bit is set,
use pointer math to determine where the next BD is.  This is also useful for
when we want to look at BDs other than the very next one (for Scatter-Gather).
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

- Also, use cacheable_memzero instead of memset for performance reasons.
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Acked-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The eTSEC can prepend up to 32 bytes to a received frame, usually for the
purpose of aligning the IP address to a word boundary, so this turns it on.

While we're in there, make the handling of the pre-frame bytes (padding and
Frame Control Block) cleaner.
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Optimize the VLAN checking logic as well.
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Acked-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix some bugs in the ethtool configuration functions:
* gfar_clean_rx_ring should not be called with interrupts disabled.
* Update last transmission time to avoid tx timeout.
* Delete redundant NETIF_F_IP_CSUM check in gfar_start_xmit
* Use netif_tx_lock_bh when reconfiguring the tx csum
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Store the interrupt coalescing values in the form in which they will be
written to the interrupt coalescing registers.  This puts a little overhead
into the ethtool configuration, and takes it out of the interrupt handler
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Acked-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Does the same for the accompanying MDIO driver, and then modifies the TBI
configuration method.  The old way used fields in einfo, which no longer
exists.  The new way is to create an MDIO device-tree node for each instance
of gianfar, and create a tbi-handle property to associate ethernet controllers
with the TBI PHYs they are connected to.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

gfar_halt does everything we want to do there, including disabling
TX/RX.  It also doesn't unnecessarily enable DMA if it's already
stopped.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We weren't unmapping DMA memory, which will break when gianfar gets used
on systems with more than 32-bits of memory.  Also, it's just plain wrong.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Acked-by: NGreg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: NKay Sievers <kay.sievers@vrfy.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The generic packet receive code takes care of setting
netdev->last_rx when necessary, for the sake of the
bonding ARP monitor.

Drivers need not do it any more.

Some cases had to be skipped over because the drivers
were making use of the ->last_rx value themselves.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The link may be up already via the chip's reset strapping, or though action
of U-Boot, or from the last time the interface was brought up.  Resetting
the link causes it to go down for several seconds.  This can significantly
increase the time from power-on to DHCP completion and a device being
accessible to the network.
Signed-off-by: NTrent Piepho <tpiepho@freescale.com>
Acked-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

The init_phy() function attaches to the PHY, then configures the
SerDes<->TBI link (in SGMII mode).  The TBI is on the MDIO bus with the PHY
(sort of) and is accessed via the gianfar's MDIO registers, using the
functions gfar_local_mdio_read/write(), which don't do any locking.

The previously attached PHY will start a work-queue on a timer, and
probably an irq handler as well, which will talk to the PHY and thus use
the MDIO bus.  This uses phy_read/write(), which have locking, but not
against the gfar_local_mdio versions.

The result is that PHY code will try to use the MDIO bus at the same time
as the SerDes setup code, corrupting the transfers.

Setting up the SerDes before attaching to the PHY will insure that there is
no race between the SerDes code and *our* PHY, but doesn't fix everything.
Typically the PHYs for all gianfar devices are on the same MDIO bus, which
is associated with the first gianfar device.  This means that the first
gianfar's SerDes code could corrupt the MDIO transfers for a different
gianfar's PHY.

The lock used by phy_read/write() is contained in the mii_bus structure,
which is pointed to by the PHY.  This is difficult to access from the
gianfar drivers, as there is no link between a gianfar device and the
mii_bus which shares the same MDIO registers.  As far as the device layer
and drivers are concerned they are two unrelated devices (which happen to
share registers).

Generally all gianfar devices' PHYs will be on the bus associated with the
first gianfar.  But this might not be the case, so simply locking the
gianfar's PHY's mii bus might not lock the mii bus that the SerDes setup
code is going to use.

We solve this by having the code that creates the gianfar platform device
look in the device tree for an mdio device that shares the gianfar's
registers.  If one is found the ID of its platform device is saved in the
gianfar's platform data.

A new function in the gianfar mii code, gfar_get_miibus(), can use the bus
ID to search through the platform devices for a gianfar_mdio device with
the right ID.  The platform device's driver data is the mii_bus structure,
which the SerDes setup code can use to lock the current bus.
Signed-off-by: NTrent Piepho <tpiepho@freescale.com>
CC: Andy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

This converts pretty much everything to print_mac. There were
a few things that had conflicts which I have just dropped for
now, no harm done.

I've built an allyesconfig with this and looked at the files
that weren't built very carefully, but it's a huge patch.
Signed-off-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

platform_get_irq*() returns on -ENXIO when the resource cannot be
found, but this remains unnoticed if stored in an unsigned.
Signed-off-by: NRoel Kluin <roel.kluin@gmail.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

The device's carrier status is controlled via the functions
netif_carrier_on() and netif_carrier_off().  These set or clear a bit
indicating the carrier (aka lower level link) is down, and if the state
changed, they fire off a routing netlink event.

Add a call to netif_carrier_off() before register_netdev() so that the
newly created device will be set to carrier down.  Then when the carrier
comes up for the first time, a netlink event will be generated, as the
carrier changed from down to up.  Otherwise the initial carrier up will
appear to be changing the status from up to up, and so no event is
generated since that's not a change.
Signed-off-by: NTrent Piepho <tpiepho@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

I got the following backtrace while network was unavailble:

|NETDEV WATCHDOG: eth0: transmit timed out
|BUG: sleeping function called from invalid context at /home/bigeasy/git/linux-2.6-powerpc/kernel/mutex.c:87
|in_atomic():1, irqs_disabled():0
|Call Trace:
|[c0383d90] [c0006dd8] show_stack+0x48/0x184 (unreliable)
|[c0383db0] [c001e938] __might_sleep+0xe0/0xf4
|[c0383dc0] [c025a43c] mutex_lock+0x24/0x3c
|[c0383de0] [c019005c] phy_stop+0x20/0x70
|[c0383df0] [c018d4ec] stop_gfar+0x28/0xf4
|[c0383e10] [c018e8c4] gfar_timeout+0x30/0x60
|[c0383e20] [c01fe7c0] dev_watchdog+0xa8/0x144
|[c0383e30] [c002f93c] run_timer_softirq+0x148/0x1c8
|[c0383e60] [c002b084] __do_softirq+0x5c/0xc4
|[c0383e80] [c00046fc] do_softirq+0x3c/0x54
|[c0383e90] [c002ac60] irq_exit+0x3c/0x5c
|[c0383ea0] [c000b378] timer_interrupt+0xe0/0xf8
|[c0383ec0] [c000e5ac] ret_from_except+0x0/0x18
|[c0383f80] [c000804c] cpu_idle+0xcc/0xdc
|[c0383fa0] [c025c07c] etext+0x7c/0x90
|[c0383fc0] [c0338960] start_kernel+0x294/0x2a8
|[c0383ff0] [c00003dc] skpinv+0x304/0x340
|------------[ cut here ]------------

The phylock was once a spinlock but got changed into a mutex via
commit 35b5f6b1 aka [PHYLIB: Locking fixes for PHY I/O potentially sleeping]
Signed-off-by: NSebastian Siewior <bigeasy@linutronix.de>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

gfar_halt() was factored out into halting and disabling by commit
d87eb127, as the suspend() method
only wants to do the former.  However, the call to gfar_halt_nodisable()
from gfar_halt() apparently got lost during the patch respin process.

This adds it back.
Signed-off-by: NScott Wood <scottwood@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

The commit that made the CONFIG_GFAR_NAPI code unconditional was
included at the same time as a new CONFIG_GFAR_NAPI user, resulting
in these bugus #ifdef's.
Reported-by: NRobert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

If the net queue has not been started, we'll get this nice oops
and non-working ethernet:

PHY: 0:01 - Link is Up - 1000/Full
------------[ cut here ]------------
kernel BUG at net/core/dev.c:1328!
Oops: Exception in kernel mode, sig: 5 [#1]
MPC837x RDB
Modules linked in:
NIP: c02544a0 LR: c01a17d0 CTR: c01a16ac
REGS: cf837e40 TRAP: 0700   Not tainted  (2.6.26-05253-g14b395e3)
MSR: 00021032 <ME,IR,DR>  CR: 22042044  XER: 00000000
TASK = cf819400[5] 'events/0' THREAD: cf836000
GPR00: c01a17d0 cf837ef0 cf819400 c03d8d08 cf8469a0 00000064 00000000 00000000
GPR08: c03d8d08 00000001 00000001 cf899ba0 22044044 00000000 0fffd000 00000000
GPR16: 0fff3028 0fff6cf0 00000000 0fff8390 0ff494a0 00000004 00000000 00000000
GPR24: c0361a00 00001058 cf9f6600 00009032 cf846800 cf846b80 00000001 00000014
NIP [c02544a0] __netif_schedule+0x28/0x8c
LR [c01a17d0] adjust_link+0x124/0x1cc
Call Trace:
[cf837ef0] [c03fb3a0] 0xc03fb3a0 (unreliable)
[cf837f10] [c01a17d0] adjust_link+0x124/0x1cc
[cf837f40] [c01a8e28] phy_state_machine+0x2e0/0x448
[cf837f60] [c0040254] run_workqueue+0xc8/0x168
[cf837f90] [c00408d8] worker_thread+0x70/0xd0
[cf837fd0] [c0044630] kthread+0x48/0x84
[cf837ff0] [c0012610] kernel_thread+0x44/0x60
Instruction dump:
7c0803a6 4e800020 3d20c03e 9421ffe0 7c0802a6 7c681b78 39298d08 7c694a78
7d290034 90010024 bfa10014 5529d97e <0f090000> 39600002 38030024 7d200028
---[ end trace 13dfd73ee42d0c30 ]---

Since the driver is using phylib (which is doing netif_carrier_on/off()),
we should simply remove netif_tx_schedule_all() from adjust_link().
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

They logically all want to trigger a schedule for all device
TX queues.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NScott Wood <scottwood@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Compile-tested only.
Signed-off-by: NFrancois Romieu <romieu@fr.zoreil.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

The loop that unmaps all of the TX Buffer Descriptors never actually
moves the txbd pointer, so we were just repeatedly unmapping the first one.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

During sparse cleanup, found a locking bug.  Some of the sysfs functions were
acquiring a lock, and then returning in the event of an error.  We rearrange
the code so that the lock is released in error conditions, too.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

TBIPA needs to be set to a value (on connected MDIO buses) that
doesn't conflict with PHYs on the bus.  By hardcoding it to 0x1f,
we were preventing boards with PHYs at 0x1f from working properly.
Instead, scan the bus when it comes up, and find an address that
doesn't have a PHY on it.  The TBI PHY configuration code then
trusts that the value in TBIPA is either safe, or doesn't matter
(ie - it's not an active bus with other PHYs).
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

gianfar was unable to handle failed skb allocation for rx buffers, so
we were spinning until it succeeded.  Actually, it was worse--we were
spinning for a long time, and then silently failing.  Instead, we take
Stephen Hemminger's suggestion to try the allocation earlier, and drop the
packet if it failed.

We also make a couple of tweaks to how buffer descriptors are set up.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

Since 43cc71ee, the platform modalias is
prefixed with "platform:".  Add MODULE_ALIAS() to the hotpluggable network
platform drivers, to re-enable auto loading.

NOTE: didn't change drivers/net/fs_enet/fs_enet-main.c "old binding" support.
That looks problematic in the first place (it even uses the ancient "struct
device_driver" binding scheme for platform_bus!) and I suspect it will vanish
soonish when arch/powerpc rules the world.  Also, drivers/net/ne.c would have
needed more thought to sort out.

[akpm@linux-foundation.org: fix sgiseeq.c]
[dbrownell@users.sourceforge.net: more drivers, registration fixes]
Signed-off-by: NKay Sievers <kay.sievers@vrfy.org>
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: Scott Wood <scottwood@freescale.com>
Cc: Vitaly Bordug <vitb@kernel.crashing.org>
Cc: Dale Farnsworth <dale@farnsworth.org>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Andrew Victor <andrew@sanpeople.com>
Cc: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

Poll the completed TX frames in gfar_poll().  This prevents the tx
completion interrupt from interfering with processing of received
frames.

We also disable hardware rx coalescing when NAPI is enabled.
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

- Fix Rx/Tx HW interrupt coalescing counter reset logic. Disabling
is required before resetting the counter.

- Update the Default both Rx and Tx coalescing timer
threshold. Formerly 4 is set which is equal to 1.5 frame at the line
rate of 1GbE interface, and it doesn't match to the coalescing frame
count which is set to 16. Threashold 21 is matched to frame count 16.
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

If the LAST bit is not set in the RxBD, it's possible we're processing
an incomplete frame, which is bad.  While we're at it, add a constant
for the error bitmask, so the whole if-clause fits on one line,
and is more legible.
Signed-off-by: NDai Haruki <dai.haruki@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>