safe/jmp/linux-2.6
15 years ago[S390] Remove initial kernel stack backchain initialization.
Heiko Carstens [Thu, 25 Dec 2008 12:39:21 +0000 (13:39 +0100)]
[S390] Remove initial kernel stack backchain initialization.

Early init code clears the backchain of the initial kernel stack frame.
This is not necessary since it is pre initialized with zeros. Plus it
was broken on 64 bit since it cleared only four of eight bytes.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] s390: use the new byteorder headers
Harvey Harrison [Thu, 25 Dec 2008 12:39:20 +0000 (13:39 +0100)]
[S390] s390: use the new byteorder headers

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] Add processor type march=z10 and a processor type safety check.
Martin Schwidefsky [Thu, 25 Dec 2008 12:39:19 +0000 (13:39 +0100)]
[S390] Add processor type march=z10 and a processor type safety check.

This patch adds the code generation option for IBM System z10 and
adds a check in head[31,64].S to prevents the execution of a kernel
compiled for a new processor type on an old machine.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] sclp vt220: fix compile warning
Heiko Carstens [Thu, 25 Dec 2008 12:39:18 +0000 (13:39 +0100)]
[S390] sclp vt220: fix compile warning

get rid of this one:

  CC      drivers/s390/char/sclp_vt220.o
drivers/s390/char/sclp_vt220.c:588: warning: '__sclp_vt220_flush_buffer' defined but not used

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] remove warnings with functions ending in BUG
Martin Schwidefsky [Thu, 25 Dec 2008 12:39:17 +0000 (13:39 +0100)]
[S390] remove warnings with functions ending in BUG

Functions which end in a BUG() statement and skip the return statement
cause compile warnings on s390, e.g.:

mm/bootmem.c: In function 'mark_bootmem':
mm/bootmem.c:321: warning: control reaches end of non-void function

To avoid the warning add an endless loop to the BUG() macro.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] mark disabled_wait as noreturn function
Heiko Carstens [Thu, 25 Dec 2008 12:39:16 +0000 (13:39 +0100)]
[S390] mark disabled_wait as noreturn function

disabled_wait() won't return, so add an __attribute__((noreturn)).
This will remove a false positive finding which our internal code
checker reports.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: use pim to check for multipath.
Sebastian Ott [Thu, 25 Dec 2008 12:39:15 +0000 (13:39 +0100)]
[S390] cio: use pim to check for multipath.

To check if multipath is available we count the bits set in lpm,
which could change over time (via configure [on|off] of a path).

The following patch uses the pim (which is persistent) for this
decision.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: commit all pmcw changes.
Sebastian Ott [Thu, 25 Dec 2008 12:39:14 +0000 (13:39 +0100)]
[S390] cio: commit all pmcw changes.

Sometimes we change the pmcw configuration but don't call msch
to transmit these changes to the channel subsystem.

The patch fixes this by calling cio_commit_config in such cases.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: introduce cio_commit_config
Sebastian Ott [Thu, 25 Dec 2008 12:39:13 +0000 (13:39 +0100)]
[S390] cio: introduce cio_commit_config

To change the configuration of a subchannel we alter the modifiable
bits of the subchannel's schib field and issue a modify subchannel.
There can be the case that not all changes were applied -or worse-
quietly overwritten by the hardware. With the next store subchannel
we obtain the current state of the hardware but lose our target
configuration.

With this patch we introduce a subchannel_config structure which
contains the target subchannel configuration. Additionally the msch
wrapper cio_modify is replaced with cio_commit_config which
copies the desired changes to a temporary schib. msch is then
called with the temporary schib. This schib is only written back
to the subchannel if all changes were applied.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: introduce cio_update_schib
Sebastian Ott [Thu, 25 Dec 2008 12:39:12 +0000 (13:39 +0100)]
[S390] cio: introduce cio_update_schib

There is the chance that we get condition code 0 for a stsch but
the resulting schib is not vaild. In the current code there are
2 cases:
* we do a check for validity of the schib after stsch, but at this
  time we have already stored the invaild schib in the subchannel
  structure. This may lead to problems.
* we don't do a check for validity, which is not that good either.

The patch addresses both issues by introducing the stsch wrapper
cio_update_schib which performs stsch on a local schib. This schib
is only written back to the subchannel if it's valid.

side note: For some functions (chp_events) the return codes are
different now (-ENXIO vs -ENODEV) but this shouldn't do harm
since the caller doesn't check for _specific_ errors.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: Use device_is_registered().
Cornelia Huck [Thu, 25 Dec 2008 12:39:11 +0000 (13:39 +0100)]
[S390] cio: Use device_is_registered().

Check if a ccw device is registered via device_is_registered()
and not via the old kludge of checking the membership in driver
core internal klists.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: Dont call ->release directly.
Cornelia Huck [Thu, 25 Dec 2008 12:39:10 +0000 (13:39 +0100)]
[S390] cio: Dont call ->release directly.

Just put the cdev's reference count to give up our reference.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: Dont fail probe for I/O subchannels.
Cornelia Huck [Thu, 25 Dec 2008 12:39:09 +0000 (13:39 +0100)]
[S390] cio: Dont fail probe for I/O subchannels.

If we fail the probe for an I/O subchannel, we won't be able
to unregister it again since there are no sch_event()
callbacks for unbound subchannels. Just succeed the probe in
any case and schedule unregistering the subchannel.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: Only register ccw_device for registered subchannel.
Cornelia Huck [Thu, 25 Dec 2008 12:39:08 +0000 (13:39 +0100)]
[S390] cio: Only register ccw_device for registered subchannel.

There is a race between io_subchannel_register() and
io_subchannel_sch_event() which may cause a subchannel to be
unregistered because it is no longer operational before
io_subchannel_register() had run. We need to check whether the
subchannel is still registered before the ccw device can be
registered and just bail out if it is not.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: Fix I/O subchannel refcounting.
Cornelia Huck [Thu, 25 Dec 2008 12:39:07 +0000 (13:39 +0100)]
[S390] cio: Fix I/O subchannel refcounting.

Subchannel refcounting was incorrect in some places, especially
a refcount was missing when ccw_device_call_sch_unregister()
was called and the refcount was not correctly switched after
moving devices.

Fix this by establishing the following rules:
- The ccw_device obtains a reference on its parent subchannel
  when dev.parent is set and gives it up in its release
  function. This is needed because we need a parent reference
  for correct refcounting even before the ccw device is (if at
  all) registered.
- When calling device_move(), obtain a reference on the new
  subchannel before moving the ccw device and give up the
  reference on the old parent after moving. This brings the
  refcount in line with the first rule.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: Fix reference counting for online/offline.
Cornelia Huck [Thu, 25 Dec 2008 12:39:06 +0000 (13:39 +0100)]
[S390] cio: Fix reference counting for online/offline.

The current code attempts to get an extra reference count
for online devices by doing a get_device() in ccw_device_online()
and a put_device() in ccw_device_done(). However, this
- incorrectly obtains an extra reference for disconnected
  devices becoming available again (since they are already
  online)
- needs special checks for css_init_done in order to handle
  the console device
- is not obvious and
- may incorretly drop a reference count in ccw_device_done() if
  that function is called after path verification for a device
  that just became not operational.

So let's just get the reference in ccw_device_set_online() and
drop it in ccw_device_set_offline(). (Unfortunately, we still
need the special case in io_subchannel_probe().)

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: Put referernce on correct device after moving.
Cornelia Huck [Thu, 25 Dec 2008 12:39:05 +0000 (13:39 +0100)]
[S390] cio: Put referernce on correct device after moving.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: fix ccwgroup online vs. ungroup race condition
Peter Oberparleiter [Thu, 25 Dec 2008 12:39:04 +0000 (13:39 +0100)]
[S390] cio: fix ccwgroup online vs. ungroup race condition

Ensure atomicity of ungroup operation to prevent concurrent ungroup
and online processing which may lead to use-after-release situations.

Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: move irritating comment.
Sebastian Ott [Thu, 25 Dec 2008 12:39:03 +0000 (13:39 +0100)]
[S390] cio: move irritating comment.

Due to former patches a comment and device id initialization were
split from the addressed function call in io_subchannel_probe.

Move it back to where it belongs.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: update sac values
Peter Oberparleiter [Thu, 25 Dec 2008 12:39:02 +0000 (13:39 +0100)]
[S390] cio: update sac values

Values for the sac field have changed - update code accordingly.

Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cio: get rid of compile warning
Heiko Carstens [Thu, 25 Dec 2008 12:39:01 +0000 (13:39 +0100)]
[S390] cio: get rid of compile warning

Move cio_tpi() to the rest of the CONFIG_CCW_CONSOLE functions to
get rid of this one:

drivers/s390/cio/cio.c:115: warning: 'cio_tpi' defined but not used

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] remove ptrace warning on 31 bit.
Martin Schwidefsky [Thu, 25 Dec 2008 12:39:00 +0000 (13:39 +0100)]
[S390] remove ptrace warning on 31 bit.

A kernel compile on 31 bit gives the following warnings in ptrace.c:

arch/s390/kernel/ptrace.c: In function 'peek_user':
arch/s390/kernel/ptrace.c:207: warning: unused variable 'dummy'
arch/s390/kernel/ptrace.c: In function 'poke_user':
arch/s390/kernel/ptrace.c:315: warning: unused variable 'dummy'

Getting rid of the dummy variables removes the warnings.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] __page_to_pfn warnings
Martin Schwidefsky [Thu, 25 Dec 2008 12:38:59 +0000 (13:38 +0100)]
[S390] __page_to_pfn warnings

For CONFIG_SPARSEMEM_VMEMMAP=y on s390 I get warnings like

init/main.c: In function 'start_kernel':
init/main.c:641: warning: format '%08lx' expects type 'long unsigned int', but argument 2 has type 'int'

The warning can be suppressed with a cast to unsigned long in the
CONFIG_SPARSEMEM_VMEMMAP=y version of __page_to_pfn.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] iucv: Locking free version of iucv_message_(receive|send)
Hendrik Brueckner [Thu, 25 Dec 2008 12:38:58 +0000 (13:38 +0100)]
[S390] iucv: Locking free version of iucv_message_(receive|send)

Provide a locking free version of iucv_message_receive and iucv_message_send
that do not call local_bh_enable in a spin_lock_(bh|irqsave)() context.

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
15 years ago[S390] s390/hvc_console: z/VM IUCV hypervisor console support
Hendrik Brueckner [Thu, 25 Dec 2008 12:38:57 +0000 (13:38 +0100)]
[S390] s390/hvc_console: z/VM IUCV hypervisor console support

This patch introduces a new hypervisor console (HVC) back-end that provides
terminal access over the z/VM inter-user communication vehicle (IUCV).

The z/VM IUCV communication is independent of the regular tcp/ip network
and allows access even if there is no network connection between two
z/VM guest virtual machines.
The z/VM IUCV hypervisor console back-end helps the user to access a
z/VM guest virtual machine that lacks of network connectivity; and thus,
provides a "full-screen" terminal alternative to 3215/3270 terminal sessions.

Use the hvc_iucv=[0..8] kernel boot parameter to specify the number of
HVC terminals using a z/VM IUCV back-end.

A recent version of the s390-tools package is required to establish a
terminal connection to a z/VM IUCV hypervisor console back-end.

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] ftrace: function tracer backend for s390
Heiko Carstens [Thu, 25 Dec 2008 12:38:56 +0000 (13:38 +0100)]
[S390] ftrace: function tracer backend for s390

This implements just the basic function tracer (_mcount) backend for s390.
The dynamic variant will come later.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] struct device - replace bus_id with dev_name(), dev_set_name()
Kay Sievers [Thu, 25 Dec 2008 12:38:55 +0000 (13:38 +0100)]
[S390] struct device - replace bus_id with dev_name(), dev_set_name()

Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] dasd: call cleanup_cqr with request_queue_lock
Stefan Haberland [Thu, 25 Dec 2008 12:38:54 +0000 (13:38 +0100)]
[S390] dasd: call cleanup_cqr with request_queue_lock

__dasd_cleanup_cqr should be called with request_queue_lock held and
__dasd_block_process_erp with queue_lock

Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] dasd: correct sense byte condition for SIM
Stefan Haberland [Thu, 25 Dec 2008 12:38:53 +0000 (13:38 +0100)]
[S390] dasd: correct sense byte condition for SIM

SIM sense data are always 32 bit sense data so sense byte 27 bit 0
has not to be set.

Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] dasd: Use accessors instead of using driver_data directly.
Cornelia Huck [Thu, 25 Dec 2008 12:38:52 +0000 (13:38 +0100)]
[S390] dasd: Use accessors instead of using driver_data directly.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] dasd: improve dasd statistics proc interface
Stefan Haberland [Thu, 25 Dec 2008 12:38:51 +0000 (13:38 +0100)]
[S390] dasd: improve dasd statistics proc interface

For a large number of I/O requests the values were shifted binary.
The shift was not transparent for the user because the shift value
was not displayed. To make this interface more human readable the
values are shifted decimal and the scale factor is displayed.

Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] zfcp: Report microcode level through service level interface
Christof Schmitt [Thu, 25 Dec 2008 12:38:50 +0000 (13:38 +0100)]
[S390] zfcp: Report microcode level through service level interface

Register zfcp with the new /proc/service_level interface to report the
FCP microcode level. When the adapter goes offline or a channel path
disappears, zfcp unregisters, since the microcode version might change
and zfcp does not know about it.

Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] service level interface.
Martin Schwidefsky [Thu, 25 Dec 2008 12:38:49 +0000 (13:38 +0100)]
[S390] service level interface.

Add a new proc interface /proc/service_levels that allows any code
to report a relevant service level, e.g. the microcode level of
devices, the service level of the hypervisor, etc.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] qdio: fix error reporting for hipersockets
Jan Glauber [Thu, 25 Dec 2008 12:38:48 +0000 (13:38 +0100)]
[S390] qdio: fix error reporting for hipersockets

Hipersocket connections can encounter temporary busy conditions.
In case of the busy bit set we retry the SIGA operation immediatelly.
If the busy condition still persists after 100 ms we fail and report
the error to the upper layer. The second stage retry logic is removed.
In case of ongoing busy conditions the upper layer needs to reset the
connection.

The reporting of a SIGA error is now done synchronously to allow the
network driver to requeue the buffers. Also no error trace is created
for the temporary SIGA errors so the error message view is not flooded.

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] qdio: improve inbound buffer acknowledgement
Jan Glauber [Thu, 25 Dec 2008 12:38:47 +0000 (13:38 +0100)]
[S390] qdio: improve inbound buffer acknowledgement

- Use automatic acknowledgement of incoming buffers in QEBSM mode
- Move ACK for non-QEBSM mode always to the newest buffer to prevent
  a race with qdio_stop_polling
- Remove the polling spinlock, the upper layer drivers return new buffers
  in the same code path and could not run in parallel
- Don't flood the error log in case of no-target-buffer-empty
- In handle_inbound we check if we would overwrite an ACK'ed buffer, if so
  advance the pointer to the oldest ACK'ed buffer so we don't overwrite an
  empty buffer in qdio_stop_polling

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] qdio: rework debug feature logging
Jan Glauber [Thu, 25 Dec 2008 12:38:46 +0000 (13:38 +0100)]
[S390] qdio: rework debug feature logging

- make qdio_trace a per device view
- remove s390dbf exceptions
- remove CONFIG_QDIO_DEBUG, not needed anymore if we check for the level
  before calling sprintf
- use snprintf for dbf entries
- add start markers to see if the dbf view wrapped
- add a global error view for all queues

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] qdio: fix compile warning under 31 bit
Jan Glauber [Thu, 25 Dec 2008 12:38:45 +0000 (13:38 +0100)]
[S390] qdio: fix compile warning under 31 bit

The QEBSM instructions are only available for CONFIG_64BIT, they are not
used under 31 bit. Make compiler happy about the false positive:

drivers/s390/cio/qdio_main.c: In function ?qdio_inbound_q_done?:
drivers/s390/cio/qdio_main.c:532: warning: ?state? may be used uninitialized in this function

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] qdio: add eqbs/sqbs instruction counters
Jan Glauber [Thu, 25 Dec 2008 12:38:44 +0000 (13:38 +0100)]
[S390] qdio: add eqbs/sqbs instruction counters

Add counters for the eqbs and sqbs instructions that indicate how often
we issued the instructions and how often the instructions returned with
less buffers than specified.

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] qdio: fix qeth port count detection
Jan Glauber [Thu, 25 Dec 2008 12:38:43 +0000 (13:38 +0100)]
[S390] qdio: fix qeth port count detection

qeth needs to get the port count information before
qdio has allocated a page for the chsc operation.
Extend qdio_get_ssqd_desc() to store the data in the
specified structure.

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] ap: Minor code beautification.
Christian Maaser [Thu, 25 Dec 2008 12:38:42 +0000 (13:38 +0100)]
[S390] ap: Minor code beautification.

Changed some symbol names for a better and clearer code.

Signed-off-by: Christian Maaser <cmaaser@de.ibm.com>
Signed-off-by: Felix Beck <beckf@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] zcrypt: Use of Thin Interrupts
Felix Beck [Thu, 25 Dec 2008 12:38:41 +0000 (13:38 +0100)]
[S390] zcrypt: Use of Thin Interrupts

When the machine supports AP adapter interrupts polling will be
switched off at module initialization and the driver will work in
interrupt mode.

Signed-off-by: Felix Beck <felix.beck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] Move stfle to header file.
Heiko Carstens [Thu, 25 Dec 2008 12:38:40 +0000 (13:38 +0100)]
[S390] Move stfle to header file.

stfle will be needed by the ap_bus module to figure out wether the AP
queue adapter interruption facility is installed.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] convert s390 to generic IPI infrastructure
Heiko Carstens [Thu, 25 Dec 2008 12:38:39 +0000 (13:38 +0100)]
[S390] convert s390 to generic IPI infrastructure

Since etr/stp don't need the old smp_call_function semantics anymore
we can convert s390 to the generic IPI infrastructure.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] serialize stp/etr work
Martin Schwidefsky [Thu, 25 Dec 2008 12:38:38 +0000 (13:38 +0100)]
[S390] serialize stp/etr work

The work function dispatched with schedule_work() can be run twice
on different cpus because run_workqueue clears the WORK_STRUCT_PENDING
bit and then executes the function. Another cpu can call schedule_work()
again and run the work function a second time before the first call
is completed. This patch serialized the etr and stp work function with
a mutex.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] convert etr/stp to stop_machine interface
Heiko Carstens [Thu, 25 Dec 2008 12:38:37 +0000 (13:38 +0100)]
[S390] convert etr/stp to stop_machine interface

This converts the etr and stp code to the new stop_machine interface
which allows to synchronize all cpus without allocating any memory.
This way we get rid of the only reason why we haven't converted s390
to the generic IPI interface yet.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] introduce vdso on s390
Martin Schwidefsky [Thu, 25 Dec 2008 12:38:36 +0000 (13:38 +0100)]
[S390] introduce vdso on s390

Add a vdso to speed up gettimeofday and clock_getres/clock_gettime for
CLOCK_REALTIME/CLOCK_MONOTONIC.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] arch_setup_additional_pages arguments
Martin Schwidefsky [Thu, 25 Dec 2008 12:38:35 +0000 (13:38 +0100)]
[S390] arch_setup_additional_pages arguments

arch_setup_additional_pages currently gets two arguments, the binary
format descripton and an indication if the process uses an executable
stack or not. The second argument is not used by anybody, it could
be removed without replacement.

What actually does make sense is to pass an indication if the process
uses the elf interpreter or not. The glibc code will not use anything
from the vdso if the process does not use the dynamic linker, so for
statically linked binaries the architecture backend can choose not
to map the vdso.

Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] vmcp: remove BKL
Christian Borntraeger [Thu, 25 Dec 2008 12:38:34 +0000 (13:38 +0100)]
[S390] vmcp: remove BKL

The vmcp driver uses the session->mutex for concurrent access of the data
structures. Therefore, the BKL in vmcp_open does not protect against any
other function in the driver.
The BLK in vmcp_open would protect concurrent access to the module init
but all necessary steps ave finished before misc_register is called.
We can safely remove the lock_kernel from vcmp.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cpu topology: dont destroy cpu sets on topology change
Heiko Carstens [Thu, 25 Dec 2008 12:37:59 +0000 (13:37 +0100)]
[S390] cpu topology: dont destroy cpu sets on topology change

Call rebuild_sched_domains instead of arch_reinit_sched_domains if
cpu topology changes. This leaves cpu sets alone which otherwise would
be destroyed.
If and how it makes sense to define cpu sets on a virtualized
architecture is another question.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] audit: get s390 ret_from_fork in sync with other architectures
Al Viro [Thu, 25 Dec 2008 12:37:58 +0000 (13:37 +0100)]
[S390] audit: get s390 ret_from_fork in sync with other architectures

On s390 we have ret_from_fork jump not to the "do all work we
normally do on return from syscall" as on x86, ppc, etc., but to the
"do all such work except audit".  Historical reasons - the codepath
triggered when we have AUDIT process flag set is separated from the
normall one and they converge at sysc_return, which is the common
part of post-syscall work.  And does not include calling audit_syscall_exit() -
that's done in the end of sysc_tracesys path, just before that path jumps
to sysc_return.

IOW, the child returning from fork()/clone()/vfork() doesn't
call audit_syscall_exit() at all, so no matter what we do with its
audit context, we are not going to see the audit entry.

The fix is simple: have ret_from_fork go to the point just past
the call of sys_.... in the 'we have AUDIT flag set' path.  There we
have (64bit variant; for 31bit the situation is the same):
sysc_tracenogo:
        tm      __TI_flags+7(%r9),(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT)
        jz      sysc_return
        la      %r2,SP_PTREGS(%r15)     # load pt_regs
        larl    %r14,sysc_return        # return point is sysc_return
        jg      do_syscall_trace_exit
which is precisely what we need - check the flag, bugger off to sysc_return
if not set, otherwise call do_syscall_trace_exit() and bugger off to
sysc_return.  r9 has just been properly set by ret_from_fork itself,
so we are fine.

Tested on s390x, seems to work fine.  WARNING: it's been about
16 years since my last contact with 3X0 assembler[1], so additional
review would be very welcome.  I don't think I've managed to screw it
up, but...

[1] that *was* in another country and besides, the box is dead...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years ago[S390] cpu topology: fix cpu_core_map initialization
Heiko Carstens [Thu, 25 Dec 2008 12:37:57 +0000 (13:37 +0100)]
[S390] cpu topology: fix cpu_core_map initialization

Common code doesn't call arch_update_cpu_topology() anymore on
cpu hotplug. But our architecture backend relied on that in order to
update the cpu_core_map. For machines without cpu topology support
this leads uninitialized cpu_core_maps for later on added cpus.

To solve this just initialize the maps with cpu_possible_map, since
that will be always valid for machines without topology support.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
15 years agoLinux 2.6.28 v2.6.28
Linus Torvalds [Wed, 24 Dec 2008 23:26:37 +0000 (15:26 -0800)]
Linux 2.6.28

Happy holidays..

15 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Wed, 24 Dec 2008 18:24:52 +0000 (10:24 -0800)]
Merge branch 'for_linus' of git://git./linux/kernel/git/mchehab/linux-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6:
  V4L/DVB (9920): em28xx: fix NULL pointer dereference in call to VIDIOC_INT_RESET command
  V4L/DVB (9908a): MAINTAINERS: mark linux-uvc-devel as subscribers only
  V4L/DVB (9906): v4l2-compat: test for unlocked_ioctl as well.
  V4L/DVB (9885): drivers/media Kconfig's: fix bugzilla #12204
  V4L/DVB (9875): gspca - main: Fix vidioc_s_jpegcomp locking.
  V4L/DVB (9781): [PATCH] Cablestar 2 I2C retries (fix CableStar2 support)
  V4L/DVB (9780): dib0700: Stop repeating after user stops pushing button

15 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 24 Dec 2008 18:24:14 +0000 (10:24 -0800)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: disable X86_PTRACE_BTS

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Wed, 24 Dec 2008 18:23:21 +0000 (10:23 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: hda - Add missing terminators in patch_sigmatel.c

15 years agoALSA: hda - Add missing terminators in patch_sigmatel.c
Herton Ronaldo Krzesinski [Tue, 23 Dec 2008 18:53:00 +0000 (16:53 -0200)]
ALSA: hda - Add missing terminators in patch_sigmatel.c

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Cc: stable@kernel.org
Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agox86: disable X86_PTRACE_BTS
Ingo Molnar [Wed, 24 Dec 2008 09:49:51 +0000 (10:49 +0100)]
x86: disable X86_PTRACE_BTS

there's a new ptrace arch level feature in .28:

  config X86_PTRACE_BTS
  bool "Branch Trace Store"

it has broken fork() handling: the old DS area gets copied over into
a new task without clearing it.

Fixes exist but they came too late:

  c5dee61: x86, bts: memory accounting
  bf53de9: x86, bts: add fork and exit handling

and are queued up for v2.6.29. This shows that the facility is still not
tested well enough to release into a stable kernel - disable it for now and
reactivate in .29. In .29 the hardware-branch-tracer will use the DS/BTS
facilities too - hopefully resulting in better code.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoparisc: disable UP-optimized flush_tlb_mm
Kyle McMartin [Tue, 23 Dec 2008 13:44:30 +0000 (08:44 -0500)]
parisc: disable UP-optimized flush_tlb_mm

flush_tlb_mm's "optimized" uniprocessor case of allocating a new
context for userspace is exposing a race where we can suddely return
to a syscall with the protection id and space id out of sync, trapping
on the next userspace access.

Debugged-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Tested-by: Helge Deller <deller@gmx.de>
Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Wed, 24 Dec 2008 01:01:40 +0000 (17:01 -0800)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/radeon: fix correctness of irq_enabled check for radeon.

15 years agoedac: fix edac core deadlock when removing a device
Harry Ciao [Tue, 23 Dec 2008 21:57:16 +0000 (13:57 -0800)]
edac: fix edac core deadlock when removing a device

When deleting an edac device, we have to wait for its edac_dev.work to be
completed before deleting the whole edac_dev structure.  Since we have no
idea which work in current edac_poller's workqueue is the work we are
conerned about, we wait for all work in the edac_poller's workqueue to be
proceseed.  This is done via flush_cpu_workqueue() which inserts a
wq_barrier into the tail of the workqueue and then sleeping on the
completion of this wq_barrier.  The edac_poller will wake up sleepers when
it is found.

EDAC core creates only one kernel worker thread, edac_poller, to run the
works of all current edac devices.  They share the same callback function
of edac_device_workq_function(), which would grab the mutex of
device_ctls_mutex first before it checks the device.  This is exactly
where edac_poller and rmmod would have a great chance to deadlock.

In below call trace of rmmod > ... >
edac_device_del_device >
edac_device_workq_teardown > flush_workqueue > flush_cpu_workqueue,

device_ctls_mutex would have already been grabbed by
edac_device_del_device().  So, on one hand rmmod would sleep on the
completion of a wq_barrier, holding device_ctls_mutex; on the other hand
edac_poller would be blocked on the same mutex when it's running any one
of works of existing edac evices(Note, this edac_dev.work is likely to be
totally irrelevant to the one that is being removed right now)and never
would have a chance to run the work of above wq_barrier to wake rmmod up.

edac_device_workq_teardown() should not be called within the critical
region of device_ctls_mutex.  Just like is done in edac_pci_del_device()
and edac_mc_del_mc(), where edac_pci_workq_teardown() and
edac_mc_workq_teardown() are called after related mutex are released.

Moreover, an edac_dev.work should check first if it is being removed.  If
this is the case, then it should bail out immediately.  Since not all of
existing edac devices are to be removed, this "shutting flag" should be
contained to edac device being removed.  The current edac_dev.op_state can
be used to serve this purpose.

The original deadlock problem and the solution have been witnessed and
tested on actual hardware.  Without the solution, rmmod an edac driver
would result in below deadlock:

root@localhost:/root> rmmod mv64x60_edac
EDAC DEBUG: mv64x60_dma_err_remove()
EDAC DEBUG: edac_device_del_device()
EDAC DEBUG: find_edac_device_by_dev()

(hang for a moment)

INFO: task edac-poller:2030 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
edac-poller   D 00000000     0  2030      2
Call Trace:
[df159dc0] [c0071e3c] free_hot_cold_page+0x17c/0x304 (unreliable)
[df159e80] [c000a024] __switch_to+0x6c/0xa0
[df159ea0] [c03587d8] schedule+0x2f4/0x4d8
[df159f00] [c03598a8] __mutex_lock_slowpath+0xa0/0x174
[df159f40] [e1030434] edac_device_workq_function+0x28/0xd8 [edac_core]
[df159f60] [c003beb4] run_workqueue+0x114/0x218
[df159f90] [c003c674] worker_thread+0x5c/0xc8
[df159fd0] [c004106c] kthread+0x5c/0xa0
[df159ff0] [c0013538] original_kernel_thread+0x44/0x60
INFO: task rmmod:2062 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
rmmod         D 0ff2c9fc     0  2062   1839
Call Trace:
[df119c00] [c0437a74] 0xc0437a74 (unreliable)
[df119cc0] [c000a024] __switch_to+0x6c/0xa0
[df119ce0] [c03587d8] schedule+0x2f4/0x4d8
[df119d40] [c03591dc] schedule_timeout+0xb0/0xf4

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agocgroups: avoid accessing uninitialized data in failure path
Li Zefan [Tue, 23 Dec 2008 21:57:14 +0000 (13:57 -0800)]
cgroups: avoid accessing uninitialized data in failure path

If cgroup_get_rootdir() failed, free_cg_links() will be called in the
failure path, but tmp_cg_links hasn't been initialized at that time.

I introduced this bug in the 2.6.27 merge window.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Serge Hallyn <serue@us.ibm.com>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agocgroups: suppress bogus warning messages
Sharyathi Nagesh [Tue, 23 Dec 2008 21:57:12 +0000 (13:57 -0800)]
cgroups: suppress bogus warning messages

Remove spurious warning messages that are thrown onto the console during
cgroup operations.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Sharyathi Nagesh <sharyathi@in.ibm.com>
Acked-by: Serge E. Hallyn <serge@hallyn.com>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agow1: fix slave selection on big-endian systems
Evgeniy Polyakov [Tue, 23 Dec 2008 21:57:12 +0000 (13:57 -0800)]
w1: fix slave selection on big-endian systems

During test of the w1-gpio driver i found that in "w1.c:679
w1_slave_found()" the device id is converted to little-endian with
"cpu_to_le64()", but its not converted back to cpu format in "w1_io.c:293
w1_reset_select_slave()".

Based on a patch created by Andreas Hummel.

[akpm@linux-foundation.org: remove unneeded cast]
Reported-by: Andreas Hummel <andi_hummel@gmx.de>
Signed-off-by: Evgeniy Polyakov <zbr@ioremap.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agortc: rtc-isl1208: reject invalid dates
Chris Elston [Tue, 23 Dec 2008 21:57:10 +0000 (13:57 -0800)]
rtc: rtc-isl1208: reject invalid dates

This patch for the rtc-isl1208 driver makes it reject invalid dates.

Signed-off-by: Chris Elston <celston@katalix.com>
[a.zummo@towertech.it: added comment explaining the check]
Signed-off-by: Alessandro Zummo <a.zummo@towertech.it>
Cc: Hebert Valerio Riedel <hvr@gnu.org>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoV4L/DVB (9920): em28xx: fix NULL pointer dereference in call to VIDIOC_INT_RESET...
Devin Heitmueller [Wed, 17 Dec 2008 02:09:35 +0000 (23:09 -0300)]
V4L/DVB (9920): em28xx: fix NULL pointer dereference in call to VIDIOC_INT_RESET command

Fix a NULL pointer dereference that would occur if the video decoder tied to
the em28xx supports the VIDIOC_INT_RESET call (for example: the cx25840 driver)

Signed-off-by: Devin Heitmueller <dheitmueller@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
15 years agodrm/radeon: fix correctness of irq_enabled check for radeon.
Dave Airlie [Mon, 22 Dec 2008 07:11:02 +0000 (17:11 +1000)]
drm/radeon: fix correctness of irq_enabled check for radeon.

This check was introduced with the logic the wrong way around.

Fixes regression: http://bugzilla.kernel.org/show_bug.cgi?id=12216

Tested-by: François Valenduc <francois.valenduc@tvcablenet.be>
Signed-off-by: Dave Airlie <airlied@redhat.com>
15 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Mon, 22 Dec 2008 22:52:00 +0000 (14:52 -0800)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  ACPI: don't cond_resched() when irqs_disabled()
  ACPI: fix 2.6.28 acpi.debug_level regression

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Mon, 22 Dec 2008 22:40:48 +0000 (14:40 -0800)]
Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6:
  drivers/ide/{cs5530.c,sc1200.c}: Move a dereference below a NULL test

15 years agodrivers/ide/{cs5530.c,sc1200.c}: Move a dereference below a NULL test
Julia Lawall [Mon, 22 Dec 2008 22:05:06 +0000 (23:05 +0100)]
drivers/ide/{cs5530.c,sc1200.c}: Move a dereference below a NULL test

In each case, if the NULL test is necessary, then the dereference should be
moved below the NULL test.

The semantic patch that makes this change is as follows:
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@@
type T;
expression E;
identifier i,fld;
statement S;
@@

- T i = E->fld;
+ T i;
  ... when != E
      when != i
  if (E == NULL) S
+ i = E->fld;
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
15 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Mon, 22 Dec 2008 18:17:19 +0000 (10:17 -0800)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  MIPS: MIPS64R2: Fix buggy __arch_swab64
  MIPS: Fix preprocessor warnings flaged by GCC 4.4

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 22 Dec 2008 18:12:54 +0000 (10:12 -0800)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  ppp: fix segfaults introduced by netdev_priv changes
  net: Fix module refcount leak in kernel_accept()

15 years agoMIPS: MIPS64R2: Fix buggy __arch_swab64
David Daney [Wed, 17 Dec 2008 21:28:39 +0000 (13:28 -0800)]
MIPS: MIPS64R2: Fix buggy __arch_swab64

The way the code is written it was assuming dshd has the function of a
hypothetical dshw instruction ...

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Fix preprocessor warnings flaged by GCC 4.4
David Daney [Thu, 18 Dec 2008 02:19:18 +0000 (18:19 -0800)]
MIPS: Fix preprocessor warnings flaged by GCC 4.4

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoNull pointer deref with hrtimer_try_to_cancel()
Thomas Gleixner [Sat, 20 Dec 2008 20:27:34 +0000 (21:27 +0100)]
Null pointer deref with hrtimer_try_to_cancel()

Impact: Prevent kernel crash with posix timer clockid CLOCK_MONOTONIC_RAW

commit 2d42244ae71d6c7b0884b5664cf2eda30fb2ae68 (clocksource:
introduce CLOCK_MONOTONIC_RAW) introduced a new clockid, which is only
available to read out the raw not NTP adjusted system time.

The above commit did not prevent that a posix timer can be created
with that clockid. The timer_create() syscall succeeds and initializes
the timer to a non existing hrtimer base. When the timer is deleted
either by timer_delete() or by the exit() cleanup the kernel crashes.

Prevent the creation of timers for CLOCK_MONOTONIC_RAW by setting the
posix clock function to no_timer_create which returns an error code.

Reported-and-tested-by: Eric Sesterhenn <snakebyte@gmx.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh...
Linus Torvalds [Sat, 20 Dec 2008 19:07:31 +0000 (11:07 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ericvh/v9fs

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
  fs/9p: change simple_strtol to simple_strtoul
  9p: convert d_iname references to d_name.name
  9p: Remove potentially bad parameter from function entry debug print.

15 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 20 Dec 2008 19:07:18 +0000 (11:07 -0800)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: fix resume (S2R) broken by Intel microcode module, on A110L
  x86 gart: don't complain if no AMD GART found
  AMD IOMMU: panic if completion wait loop fails
  AMD IOMMU: set cmd buffer pointers to zero manually
  x86: re-enable MCE on secondary CPUS after suspend/resume
  AMD IOMMU: allocate rlookup_table with __GFP_ZERO

15 years agox86: fix resume (S2R) broken by Intel microcode module, on A110L
Dmitry Adamushko [Fri, 19 Dec 2008 23:15:24 +0000 (00:15 +0100)]
x86: fix resume (S2R) broken by Intel microcode module, on A110L

Impact: fix deadlock

This is in response to the following bug report:

Bug-Entry       : http://bugzilla.kernel.org/show_bug.cgi?id=12100
Subject         : resume (S2R) broken by Intel microcode module, on A110L
Submitter       : Andreas Mohr <andi@lisas.de>
Date            : 2008-11-25 08:48 (19 days old)
Handled-By      : Dmitry Adamushko <dmitry.adamushko@gmail.com>

[ The deadlock scenario has been discovered by Andreas Mohr ]

I think I might have a logical explanation why the system:

  (http://bugzilla.kernel.org/show_bug.cgi?id=12100)

might hang upon resuming, OTOH it should have likely hanged each and every time.

(1) possible deadlock in microcode_resume_cpu() if either 'if' section is
taken;

(2) now, I don't see it in spec. and can't experimentally verify it (newer
ucodes don't seem to be available for my Core2duo)... but logically-wise, I'd
think that when read upon resuming, the 'microcode revision' (MSR 0x8B) should
be back to its original one (we need to reload ucode anyway so it doesn't seem
logical if a cpu doesn't drop the version)... if so, the comparison with
memcmp() for the full 'struct cpu_signature' is wrong... and that's how one of
the aforementioned 'if' sections might have been triggered - leading to a
deadlock.

Obviously, in my tests I simulated loading/resuming with the ucode of the same
version (just to see that the file is loaded/re-loaded upon resuming) so this
issue has never popped up.

I'd appreciate if someone with an appropriate system might give a try to the
2nd patch (titled "fix a comparison && deadlock...").

In any case, the deadlock situation is a must-have fix.

Reported-by: Andreas Mohr <andi@lisas.de>
Signed-off-by: Dmitry Adamushko <dmitry.adamushko@gmail.com>
Tested-by: Andreas Mohr <andi@lisas.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Cc: <stable@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agofs/9p: change simple_strtol to simple_strtoul
Julia Lawall [Fri, 19 Dec 2008 22:50:22 +0000 (16:50 -0600)]
fs/9p: change simple_strtol to simple_strtoul

Since v9ses->uid is unsigned, it would seem better to use simple_strtoul that
simple_strtol.

A simplified version of the semantic patch that makes this change is as
follows: (http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@r2@
long e;
position p;
@@

e = simple_strtol@p(...)

@@
position p != r2.p;
type T;
T e;
@@

e =
- simple_strtol@p
+ simple_strtoul
  (...)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: Eric Van Hensbergen <ericvh@gmail.com>
15 years ago9p: convert d_iname references to d_name.name
Wu Fengguang [Fri, 19 Dec 2008 22:47:40 +0000 (16:47 -0600)]
9p: convert d_iname references to d_name.name

d_iname is rubbish for long file names.
Use d_name.name in printks instead.

Signed-off-by: Wu Fengguang <wfg@linux.intel.com>
Acked-by: Eric Van Hensbergen <ericvh@gmail.com>
15 years ago9p: Remove potentially bad parameter from function entry debug print.
Duane Griffin [Fri, 19 Dec 2008 22:45:21 +0000 (16:45 -0600)]
9p: Remove potentially bad parameter from function entry debug print.

Signed-off-by: Duane Griffin <duaneg@dghda.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Fri, 19 Dec 2008 19:37:23 +0000 (11:37 -0800)]
Merge git://git./linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] mpt fusion: clear list of outstanding commands on host reset
  [SCSI] scsi_lib: only call scsi_unprep_request() under queue lock
  [SCSI] ibmvstgt: move crq_queue_create to the end of initialization
  [SCSI] libiscsi REGRESSION: fix passthrough support with older iscsi tools
  [SCSI] aacraid: disable Dell Percraid quirk on Adaptec 2200S and 2120S

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Fri, 19 Dec 2008 19:37:07 +0000 (11:37 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: Fix a Oops bug in omap soc driver.
  ALSA: hda - Remove non-working headphone control for Dell laptops
  ALSA: hda - Add no-jd model for IDT 92HD73xx
  ALSA: Revert "ALSA: hda: removed unneeded hp_nid references"
  ALSA: hda - Add quirk for Dell Studio 17
  ALSA: hda - Fix silent HP output on D975

15 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Fri, 19 Dec 2008 19:36:49 +0000 (11:36 -0800)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  cciss: fix problem that deleting multiple logical drives could cause a panic

15 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Fri, 19 Dec 2008 19:36:04 +0000 (11:36 -0800)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/i915: GEM on PAE has problems - disable it for now.
  drm/i915: Don't return busy for buffers left on the flushing list.

15 years agoMerge branch 'for-linus' of git://neil.brown.name/md
Linus Torvalds [Fri, 19 Dec 2008 19:34:36 +0000 (11:34 -0800)]
Merge branch 'for-linus' of git://neil.brown.name/md

* 'for-linus' of git://neil.brown.name/md:
  md: Don't read past end of bitmap when reading bitmap.

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes...
Linus Torvalds [Fri, 19 Dec 2008 19:33:10 +0000 (11:33 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jbarnes/pci-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
  PCI hotplug: ibmphp: Fix module ref count underflow
  PCI hotplug: acpiphp wants a 64-bit _SUN
  PCI: pciehp: fix unexpected power off with pciehp_force
  PCI: fix aer resume sanity check

15 years agoMerge branch 'fix/asoc' into for-linus
Takashi Iwai [Fri, 19 Dec 2008 14:37:12 +0000 (15:37 +0100)]
Merge branch 'fix/asoc' into for-linus

15 years agoALSA: Fix a Oops bug in omap soc driver.
Stanley Miao [Fri, 19 Dec 2008 14:08:22 +0000 (22:08 +0800)]
ALSA: Fix a Oops bug in omap soc driver.

There will be a Oops or frequent underrun messages when playing music with
omap soc driver, this is because a data region is incorretly sized, other data
region will be overwriten when writing to this data region.

Signed-off-by: Stanley Miao <stanley.miao@windriver.com>
Acked-by: Jarkko Nikula <jarkko.nikula@nokia.com>
Cc: stable@kernel.org
Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoALSA: hda - Remove non-working headphone control for Dell laptops
Takashi Iwai [Fri, 19 Dec 2008 13:02:32 +0000 (14:02 +0100)]
ALSA: hda - Remove non-working headphone control for Dell laptops

The previous commit re-enabled hp_nid setup for IDT92HD73*, but
it's unneeded indeed for Dell laptops that have multiple headphones.
Setting the extra hp_nid results in a non-working "Headpohne" mixer
control.  Thus hp_nid should be 0 for these dell models.

Also, the automatic addition of hp_nid should check whether it's
a dual-HP model or not.  For dual-HPs, the pins are already checked
by the early workaround.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoACPI: don't cond_resched() when irqs_disabled()
Wu Fengguang [Wed, 26 Nov 2008 06:35:22 +0000 (14:35 +0800)]
ACPI: don't cond_resched() when irqs_disabled()

The ACPI interpreter usually runs with irqs enabled.
However, during suspend/resume it runs with
irqs disabled to evaluate _GTS/_BFS, as well as
by irqrouter_resume() which evaluates _CRS, _PRS, _SRS.

http://bugzilla.kernel.org/show_bug.cgi?id=12252

Signed-off-by: Wu Fengguang <wfg@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: fix 2.6.28 acpi.debug_level regression
Bjorn Helgaas [Thu, 13 Nov 2008 23:30:13 +0000 (17:30 -0600)]
ACPI: fix 2.6.28 acpi.debug_level regression

acpi_early_init() was changed to over-write the cmdline param,
making it really inconvenient to set debug flags at boot-time.

Also,
This sets the default level to "info", which is what all the ACPI
drivers use.  So to enable messages from drivers, you only have to
supply the "layer" (a.k.a. "component").  For non-"info" ACPI core
and ACPI interpreter messages, you have to supply both level and
layer masks, as before.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoALSA: hda - Add no-jd model for IDT 92HD73xx
Takashi Iwai [Wed, 17 Dec 2008 13:51:01 +0000 (14:51 +0100)]
ALSA: hda - Add no-jd model for IDT 92HD73xx

Added the model without the jack-detection for some desktops that
have really no jack-detection.  The recent driver caused regressions
regarding the sound output on such machines.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoALSA: Revert "ALSA: hda: removed unneeded hp_nid references"
Takashi Iwai [Fri, 19 Dec 2008 07:20:38 +0000 (08:20 +0100)]
ALSA: Revert "ALSA: hda: removed unneeded hp_nid references"

This reverts commit 07f455f779acfb3eba4921fd1399761559b10fa9.
    ALSA: hda: removed unneeded hp_nid references

    Removed unneeded hp_nid references for 92hd73xx codec family.

This caused the silent output on some Intel desktops due to missing
routing of widget 0x0a and 0x0d.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agocciss: fix problem that deleting multiple logical drives could cause a panic
Stephen M. Cameron [Thu, 18 Dec 2008 13:55:11 +0000 (14:55 +0100)]
cciss: fix problem that deleting multiple logical drives could cause a panic

Fix problem that deleting multiple logical drives could cause a panic.

It fixes a panic which can be easily reproduced in the following way: Just
create several "arrays," each with multiple logical drives via hpacucli,
then delete the first array, and it will blow up in deregister_disk(), in
the call to get_host() when it tries to dig the hba pointer out of a NULL
queue pointer.

The problem has been present since my code to make rebuild_lun_table
behave better went in.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cca.cpqcorp.net>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agoALSA: hda - Add quirk for Dell Studio 17
Joerg Schirottke [Fri, 19 Dec 2008 07:13:49 +0000 (08:13 +0100)]
ALSA: hda - Add quirk for Dell Studio 17

Added the matching model=dell-m6 for Dell Studio 17 laptop.

Signed-off-by: Joerg Schirottke <master@kanotix.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agodrm/i915: GEM on PAE has problems - disable it for now.
Dave Airlie [Fri, 19 Dec 2008 05:38:34 +0000 (15:38 +1000)]
drm/i915: GEM on PAE has problems - disable it for now.

On PAE systems, GEM allocates pages using shmem, and passes these
pages to be bound into AGP, however the AGP interfaces + the x86
set_memory interfaces all take unsigned long not dma_addr_t.

The initial fix for this was a mess, so we need to do this correctly
for 2.6.29.

Signed-off-by: Dave Airlie <airlied@redhat.com>
15 years agodrm/i915: Don't return busy for buffers left on the flushing list.
Eric Anholt [Mon, 15 Dec 2008 03:05:04 +0000 (19:05 -0800)]
drm/i915: Don't return busy for buffers left on the flushing list.

These buffers don't have active rendering still occurring to them, they just
need either a flush to be emitted or a retire_requests to occur so that we
notice they're done.  Return unbusy so that one of the two occurs.  The two
expected consumers of this interface (OpenGL and libdrm_intel BO cache) both
want this behavior.

Signed-off-by: Eric Anholt <eric@anholt.net>
Acked-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
15 years agomd: Don't read past end of bitmap when reading bitmap.
NeilBrown [Fri, 19 Dec 2008 05:25:01 +0000 (16:25 +1100)]
md: Don't read past end of bitmap when reading bitmap.

When we read the write-intent-bitmap off the device, we currently
read a whole number of pages.
When PAGE_SIZE is 4K, this works due to the alignment we enforce
on the superblock and bitmap.
When PAGE_SIZE is 64K, this case read past the end-of-device
which causes an error.

When we write the superblock, we ensure to clip the last page
to just be the required size.  Copy that code into the read path
to just read the required number of sectors.

Signed-off-by: Neil Brown <neilb@suse.de>
Cc: stable@kernel.org
15 years agoppp: fix segfaults introduced by netdev_priv changes
James Chapman [Wed, 17 Dec 2008 12:02:16 +0000 (12:02 +0000)]
ppp: fix segfaults introduced by netdev_priv changes

This patch fixes a segfault in ppp_shutdown_interface() and
ppp_destroy_interface() when a PPP connection is closed. I bisected
the problem to the following commit:

  commit c8019bf3aff653cceb64f66489fc299ee5957b57
  Author: Wang Chen <wangchen@cn.fujitsu.com>
  Date:   Thu Nov 20 04:24:17 2008 -0800

    netdevice ppp: Convert directly reference of netdev->priv

    1. Use netdev_priv(dev) to replace dev->priv.
    2. Alloc netdev's private data by alloc_netdev().

Signed-off-by: Wang Chen <wangchen@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The original ppp_generic code treated the netdev and struct ppp as
independent data structures which were freed separately. In moving the
ppp struct into the netdev, it is now possible for the private data to
be freed before the call to ppp_shutdown_interface(), which is bad.

The kfree(ppp) in ppp_destroy_interface() is also wrong; presumably
ppp hasn't worked since the above commit.

The following patch fixes both problems.

Signed-off-by: James Chapman <jchapman@katalix.com>
Reviewed-by: Wang Chen <wangchen@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: Fix module refcount leak in kernel_accept()
Wei Yongjun [Fri, 19 Dec 2008 03:35:10 +0000 (19:35 -0800)]
net: Fix module refcount leak in kernel_accept()

The kernel_accept() does not hold the module refcount of newsock->ops->owner,
so we need __module_get(newsock->ops->owner) code after call kernel_accept()
by hand.
In sunrpc, the module refcount is missing to hold. So this cause kernel panic.

Used following script to reproduct:

while [ 1 ];
do
    mount -t nfs4 192.168.0.19:/ /mnt
    touch /mnt/file
    umount /mnt
    lsmod | grep ipv6
done

This patch fixed the problem by add __module_get(newsock->ops->owner) to
kernel_accept(). So we do not need to used __module_get(newsock->ops->owner)
in every place when used kernel_accept().

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>