linux-toradex.git/drivers/s390, branch v3.6.1

s390/dasd: fix read unit address configuration loop

2012-09-17T07:58:21+00:00

Read unit address is done for all devices during online processing to read
out LCU features. This is also done after disconnect/connect a LCU.
Some older storage hardware does not provide the capability to read unit
address configuration.
This leads to a loop trying to read unit address configuration every 30
seconds. The device is still operational but logs are flooded with error
messages.

Fix the loop by recognizing a command reject saying that the suborder
for ruac is not supported.

Signed-off-by: Stefan Haberland 
Signed-off-by: Martin Schwidefsky

s390/dasd: fix pathgroup race

2012-09-17T07:58:18+00:00

If a new path is available we need to verify the path data. If it is the
first path for a device the stop bits are removed after path verification.
If a pathgroup is established we need to set system characteristics for
the lcu. Therefore I/O has to be started.
If the device is stopped the set system characteristics worker may block
the path verification worker and the device is blocked.

Turn on failfast for set system characteristics CQR to prevent a deadlock
with the path verification worker.

If a pathgroup is established on a device that is not in use trigger path
verification. Maybe we were not informed about a working path.

Signed-off-by: Stefan Haberland 
Reviewed-by: Stefan Weinhuber 
Signed-off-by: Martin Schwidefsky

s390/cio: invalidate cdev pointer before deregistration

2012-09-06T08:40:40+00:00

Make sure that the cdev pointer for IO subchannels is set to NULL when
we deregister the device (and release its last reference). This will
fix a bug were another process operates on an already freed ccw device.

Acked-by: Peter Oberparleiter 
Signed-off-by: Sebastian Ott 
Signed-off-by: Martin Schwidefsky

s390/cio: fix IO subchannel event race

2012-09-06T08:40:39+00:00

If the subchannel event function is called from IRQ context and we
observe that the subchannel in question is gone we flag the attached
device as not operational and schedule the event function to be called
again from process context where the subchannel gets deregistered.
However if the subchannel reappeared at the time the event function
gets called from process context we would do nothing and leave the
device in not operational state. Recognize this case in sch_get_action
and trigger reexamination of the subchannel/device.

Acked-by: Peter Oberparleiter 
Signed-off-by: Sebastian Ott 
Signed-off-by: Martin Schwidefsky

s390/dasd: move wake_up call

2012-09-06T08:40:38+00:00

Ensure that all work is done when the process waiting for a
dasd state change is woken up. With this change it is save
to assume that after a userspace triggered state change and
a udev settle invocation there are no unexpected users of a
dasd device.

Acked-by: Stefan Weinhuber 
Signed-off-by: Sebastian Ott 
Signed-off-by: Martin Schwidefsky

s390/dasd: fix ioctl return value

2012-08-28T08:08:31+00:00

For unimplemented ioctls the dasd driver should return -ENOTTY.

Reported-by: Wanlong Gao 
Acked-by: Stefan Weinhuber 
Signed-off-by: Heiko Carstens 
Signed-off-by: Martin Schwidefsky

s390/sclp_sdias: Add missing break and "fall through"

2012-08-08T14:32:55+00:00

If the last block of the HSA is read, EVSTATE_ALL_STORED is returned
by SCLP. Because of a missing break in the switch statement two trace
entries are written in this case: "all stored" and "part stored".

This patch adds the missing break and also adds a "fall through"
comment to improve the readability.

Signed-off-by: Michael Holzheu 
Reported-by: David A Gilbert 
Signed-off-by: Martin Schwidefsky

Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

2012-07-27T01:09:01+00:00

Pull networking updates and fixes from David Miller:

1) Reinstate the no-ref optimization for input route lookups in ipv4 to
   fix some routing cache removal perf regressions.

2) Make TCP socket pre-demux work on ipv6 side too, from Eric Dumazet.

3) Get RX hash value from correct place in be2net driver, from
   Sarveshwar Bandi.

4) Validation of FIB cached routes missing critical check, from Eric
   Dumazet.

5) EEH support in mlx4 driver, from Kleber Sacilotto de Souza.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (23 commits)
  ipv6: Early TCP socket demux
  ipv4: Fix input route performance regression.
  pch_gbe: vlan skb len fix
  pch_gbe: add extra clean tx
  pch_gbe: fix transmit watchdog timeout
  ixgbe: fix panic while dumping packets on Tx hang with IOMMU
  be2net: Fix to parse RSS hash from Receive completions correctly.
  net/mlx4_en: Limit the RFS filter IDs to be < RPS_NO_FILTER
  hyperv: Add error handling to rndis_filter_device_add()
  hyperv: Add a check for ring_size value
  ipv4: rt_cache_valid must check expired routes
  net/pch_gpe: Cannot disable ethernet autonegation
  qeth: repair crash in qeth_l3_vlan_rx_kill_vid()
  netiucv: cleanup attribute usage
  net: wiznet add missing HAS_IOMEM dependency
  be2net: Missing byteswap in be_get_fw_log_level causes oops on PowerPC
  mlx4: Add support for EEH error recovery
  cdc-ncm: tag Ericsson WWAN devices (eg F5521gw) with FLAG_WWAN
  wanmain: comparing array with NULL
  caif: fix NULL pointer check
  ...

qeth: repair crash in qeth_l3_vlan_rx_kill_vid()

2012-07-25T22:24:13+00:00

Commit efc73f4b "net: Fix memory leak - vlan_info struct" adds deletion of
VLAN 0 for devices with feature NETIF_F_HW_VLAN_FILTER. For driver
qeth these are the layer 3 devices. Usually there exists no
separate vlan net_device for VLAN 0. Thus the qeth functions
qeth_l3_free_vlan_addresses4() and qeth_l3_free_vlan_addresses6()
require an extra checking if function __vlan_find_dev_deep()
returns with a net_device.

Cc: stable@vger.kernel.org
Signed-off-by: Ursula Braun 
Signed-off-by: Frank Blaschka 
Signed-off-by: David S. Miller

netiucv: cleanup attribute usage

2012-07-25T22:24:13+00:00

Let the driver core handle device attribute creation and removal. This
will simplify the code and eliminates races between attribute
availability and userspace notification via uevents.

Signed-off-by: Sebastian Ott 
Signed-off-by: Frank Blaschka 
Acked-by: Ursula Braun 
Signed-off-by: David S. Miller