This device requires a fundamental reset when recovering from EEH.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Clean up driver resources without touch the hardware. Add pci
save/restore state.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Call set_multi API after reset recovery.  This was exposed by tripping
tx_timeout.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add wait for NIC fifo and MGMNT fifo to empty before applying reset.
Otherwise broken frames can be processed by management processor and
cause it to hang.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The addresses were being added to the filter properly, but were not
being enabled.  This adds enable bit to filter write.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Broadcast/multicast should always be routed to the default (zeroeth) rx
ring.  Broadcast frames are already routed correctly.  This fixes
routing for multicast frames.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently frames are routed based on their type and MAC address. This
patch adds the port number on which the frame arrived to the routing.
This prevents problems in the case where both interfaces have the same MAC address in
a routing configuration.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix RX queue table size and change from random to default hash values.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Get rid of spinlock and private mutex usage for exclusive access to the
HW semaphore register.  rtnl_lock already creates exclusive access to
this register in all driver API.
Add rtnl to firmware worker threads that also use the HW semaphore register.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Check that we are not already polling firmware events before we queue the
firmware event worker, then disable firmware interrupts.
Otherwise we can queue the same event multiple times.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Move the call to rtnl_lock() to before the internal call to
ql_adapter_down()/ql_adapter_up().  This prevents collisions that can
happen when recovering from an asic error.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove the unnecessary locking around the call to ql_adapter_reset().

Sep 25 08:17:29 localhost kernel:    SOFTIRQ-ON-W at:
Sep 25 08:17:29 localhost kernel:                         [<c0000000000a2964>] .lock_acquire+0x10c/0x158
Sep 25 08:17:29 localhost kernel:                         [<c0000000004542e0>] ._spin_lock+0x34/0x58
Sep 25 08:17:29 localhost kernel:                         [<d000000006723070>] .ql_adapter_down+0x40c/0x4a0 [qlge]
Sep 25 08:17:29 localhost kernel:                         [<d0000000067256d8>] .qlge_close+0x38/0x58 [qlge]
Sep 25 08:17:29 localhost kernel:                         [<c0000000003ada6c>] .dev_close+0xdc/0x118
Sep 25 08:17:29 localhost kernel:                         [<c0000000003adb48>] .rollback_registered+0xa0/0x158
Sep 25 08:17:29 localhost kernel:                         [<c0000000003adc50>] .unregister_netdevice+0x50/0x7c
Sep 25 08:17:29 localhost kernel:                         [<c0000000003adca0>] .unregister_netdev+0x24/0x40
Sep 25 08:17:29 localhost kernel:                         [<d00000000672e0c0>] .qlge_remove+0x28/0x64 [qlge]
Sep 25 08:17:29 localhost kernel:                         [<c000000000253fdc>] .pci_device_remove+0x50/0x90
Sep 25 08:17:29 localhost kernel:                         [<c0000000002f5434>] .__device_release_driver+0x94/0xf8
Sep 25 08:17:29 localhost kernel:                         [<c0000000002f5560>] .driver_detach+0xc8/0xfc
Sep 25 08:17:29 localhost kernel:                         [<c0000000002f3fd8>] .bus_remove_driver+0xb4/0x114
Sep 25 08:17:29 localhost kernel:                         [<c0000000002f5d4c>] .driver_unregister+0x80/0xa4
Sep 25 08:17:29 localhost kernel:                         [<c00000000025421c>] .pci_unregister_driver+0x50/0xc8
Sep 25 08:17:29 localhost kernel:                         [<d00000000672e044>] .qlge_exit+0x1c/0x34 [qlge]
Sep 25 08:17:29 localhost kernel:                         [<c0000000000ac8b0>] .SyS_delete_module+0x234/0x2d0
Sep 25 08:17:29 localhost kernel:                         [<c000000000008554>] syscall_exit+0x0/0x40
Sep 25 08:17:29 localhost kernel:    INITIAL USE at:
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

ql_clear_routing_entries() takes/gives it's own hardware semaphore since
it is called from more than one place.  ql_route_initialize() should
make this call and THEN take it's own semaphore before doing it's work.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NJoe Perches <joe@perches.com>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

In a couple of cases collapse some extra code like:
   int retval = NETDEV_TX_OK;
   ...
   return retval;
into
   return NETDEV_TX_OK;
Signed-off-by: NStephen Hemminger <shemminger@vyatta.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

TX completions were running in a workqueue queued by the ISR.  This
patch moves the processing of TX completions to an existing RSS NAPI
context.
Now each irq vector runs NAPI for one RSS ring and one or more TX
completion rings.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently we downshift to MSI/Legacy if we don't get enough vectors for
cpu_count RSS rings plus cpu_count TX completion rings.  This patch
allows running MSIX with the vector count that the platform provides.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently we have three types of RX rings.

1) Default ring - services rx_ring for broadcast/multicast, handles
firmware events, and errors.

2) TX completion ring - handles only outbound completions.

3) RSS ring - handles only inbound completions.

This patch gets rid of the default ring type and moves it's functionality
into the first RSS ring.  This makes better use of MSIX vectors since
they are a limited resource on some platforms.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

PCI drivers that implement the struct pci_error_handlers' error_detected
callback should return PCI_ERS_RESULT_DISCONNECT if the state passed in is
pci_channel_io_perm_failure. This patch fixes the issue for qlge.
Signed-off-by: NDean Nelson <dnelson@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Some usage was only sizing a pointer rather than the data type.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We need to set/clear the mac address register when the link goes up/down
respectively.  Without this both ports of a 2-port device can end up
with the same mac address in a bonding scenario.
The new ql_link_on() and ql_link_off() will also be used in handling
certain firmware events.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This addes functionality to set/clear the MAC address in the hardware
when the link goes up/down.
The MAC address register is persistent across function resets. In
bonding the same address can bounce from one port to the other.  This
can cause packets to be delivered to the wrong port.
This patch clears the MAC address in the hardware when the link is down
and sets it when the link comes up.
It was found that pulling/pushing the cable from one port to another
causes the same MAC address to be in both ports.
The next patch in this series will use this functionality as well.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The caller will free acquired resouces if a failure occurs.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We were turning on the carrier without verifying the link was up.
This adds link up to the link initialize check before turning carrier
on.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Not clearing the routing bits can cause frames to erroneously get routed to
management processor.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The hardware semaphore covers the configuration register as well as the
ICB registers.  The ICB high and low regs contain the address of the
initialization control block and the config register is used to signal
the hardware that a block is ready to be downloaded.  Currently we were
only protecting the ICB regs.  This changes expands to cover the config
register as well.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

fix ETIMEOUT -> ETIMEDOUT typos
Signed-off-by: NJean Delvare <khali@linux-fr.org>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

RX buffer rings can be comprised of non-contiguous fixed
size chunks of memory. The ring is given to the hardware
as a pointer to a location that stores the location of
the queue.  If the queue is greater than 4096 bytes then
the hardware gets a list of said pointers.
This patch addes the necessary logic to generate the list if
the queue size exceeds 4096.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The alignment was on size of queue boundary, but the hardware
only requires 4-byte alignment.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently we support only PCIe NIC functions zero and one, and FCoE
functions as 3 and 4. Future configurations can mix these up in any
fashion.
This patch removes the 0-1 dependancy and allows usage of any of the 4
functions. We also find the alternate NIC function (if exist) and
determine our port number based on the comparison of the two functions:
Lower function number gets first port, higher function gets second port.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Extract either manufacturer or Bladecenter Open Fabric
Manager (BOFM) MAC address.  BOFM may indicate an
alternate MAC address.  This patch honors that request
by extracting the MAC address from a different
flash location if a flag is set.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The max frame size register is set higher than the MTU to
accomodate FCoE frames.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Not necessary if hardware supports 65536 as it's the default setting.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This is used by driver banner and ethtool info.
Signed-off-by: NRon Mercer <ron.mercer@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Followup of commits 9d21493b
and 08baf561
(net: tx scalability works : trans_start)
(net: txq_trans_update() helper)

Now that core network takes care of trans_start updates, dont do it
in drivers themselves, if possible. Multi queue drivers can
avoid one cache miss (on dev->trans_start) in their start_xmit()
handler.

Exceptions are NETIF_F_LLTX drivers (vxge & tehuti)
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace all DMA_32BIT_MASK macro with DMA_BIT_MASK(32)

Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>