openwrt/staging/blogic.git
16 years agoPCI: include linux/pm_wakeup.h for device_set_wakeup_capable
Stephen Rothwell [Thu, 10 Jul 2008 00:16:44 +0000 (02:16 +0200)]
PCI: include linux/pm_wakeup.h for device_set_wakeup_capable

drivers/pci/pci.c needs pm_wakeup.h since it uses device_set_wakup_capable().
The latter also needs to be stubbed out for !CONFIG_PM.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI PM: Fix pci_prepare_to_sleep
Rafael J. Wysocki [Sun, 13 Jul 2008 20:45:06 +0000 (22:45 +0200)]
PCI PM: Fix pci_prepare_to_sleep

The recently introduced pci_prepare_to_sleep() needs the following fix,
because there are systems which are not power manageable by ACPI (ie.
ACPI doesn't provide methods to put the device into low power states and
back), but require ACPI hooks to be executed for wake-up to work.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agox86/PCI: Fix PCI config space for domains > 0
Matthew Wilcox [Fri, 11 Jul 2008 21:21:17 +0000 (15:21 -0600)]
x86/PCI: Fix PCI config space for domains > 0

John Keller reports that PCI config space access is broken on machines
with more than one domain.  conf1 accesses only work for domain 0, so make sure
we check the domain number in the raw routines before trying conf1.

Reported-by: John Keller <jpk@sgi.com>
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoFix acpi_pm_device_sleep_wake() by providing a stub for CONFIG_PM_SLEEP=n
David Howells [Tue, 8 Jul 2008 14:00:54 +0000 (15:00 +0100)]
Fix acpi_pm_device_sleep_wake() by providing a stub for CONFIG_PM_SLEEP=n

So that one of the several config option permutations will build again.

Tested-by: Alexander Beregalov <a.beregalov@gmail.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: Simplify PCI device PM code
Rafael J. Wysocki [Mon, 7 Jul 2008 01:36:24 +0000 (03:36 +0200)]
PCI: Simplify PCI device PM code

If the offset of PCI device's PM capability in its configuration space,
the mask of states that the device supports PME# from and the D1 and D2
support bits are cached in the corresponding struct pci_dev, the PCI
device PM code can be simplified quite a bit.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI PM: Introduce pci_prepare_to_sleep and pci_back_from_sleep
Rafael J. Wysocki [Mon, 7 Jul 2008 01:35:26 +0000 (03:35 +0200)]
PCI PM: Introduce pci_prepare_to_sleep and pci_back_from_sleep

Introduce functions pci_prepare_to_sleep() and pci_back_from_sleep(),
to be used by the PCI drivers that want to place their devices into
the lowest power state appropiate for them (PCI_D3hot, if the device
is not supposed to wake up the system, or the deepest state from
which the wake-up is possible, otherwise) while the system is being
prepared to go into a sleeping state and to put them back into D0
during the subsequent transition to the working state.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI ACPI: Rework PCI handling of wake-up
Rafael J. Wysocki [Mon, 7 Jul 2008 01:34:48 +0000 (03:34 +0200)]
PCI ACPI: Rework PCI handling of wake-up

* Introduce function acpi_pm_device_sleep_wake() for enabling and
  disabling the system wake-up capability of devices that are power
  manageable by ACPI.

* Introduce function acpi_bus_can_wakeup() allowing other (dependent)
  subsystems to check if ACPI is able to enable the system wake-up
  capability of given device.

* Introduce callback .sleep_wake() in struct pci_platform_pm_ops and
  for the ACPI PCI 'driver' make it use acpi_pm_device_sleep_wake().

* Introduce callback .can_wakeup() in struct pci_platform_pm_ops and
  for the ACPI 'driver' make it use acpi_bus_can_wakeup().

* Move the PME# handlig code out of pci_enable_wake() and split it
  into two functions, pci_pme_capable() and pci_pme_active(),
  allowing the caller to check if given device is capable of
  generating PME# from given power state and to enable/disable the
  device's PME# functionality, respectively.

* Modify pci_enable_wake() to use the new ACPI callbacks and the new
  PME#-related functions.

* Drop the generic .platform_enable_wakeup() callback that is not
  used any more.

* Introduce device_set_wakeup_capable() that will set the
  power.can_wakeup flag of given device.

* Rework PCI device PM initialization so that, if given device is
  capable of generating wake-up events, either natively through the
  PME# mechanism, or with the help of the platform, its
  power.can_wakeup flag is set and its power.should_wakeup flag is
  unset as appropriate.

* Make ACPI set the power.can_wakeup flag for devices found to be
  wake-up capable by it.

* Make the ACPI wake-up code enable/disable GPEs for devices that
  have the wakeup.flags.prepared flag set (which means that their
  wake-up power has been enabled).

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoACPI: Introduce new device wakeup flag 'prepared'
Rafael J. Wysocki [Mon, 7 Jul 2008 01:34:11 +0000 (03:34 +0200)]
ACPI: Introduce new device wakeup flag 'prepared'

Introduce additional flag 'prepared' in struct acpi_device_wakeup_flags
and use it to prevent devices from being enable/disabled do wake up the
system multiple times in a row (this does not happen currently, but will
be possible after some of the following patches).

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoACPI: Introduce acpi_device_sleep_wake function
Rafael J. Wysocki [Mon, 7 Jul 2008 01:33:34 +0000 (03:33 +0200)]
ACPI: Introduce acpi_device_sleep_wake function

The currect ACPI code attempts to execute _PSW at three different
places and in one of them only it tries to execute _DSW before _PSW,
which is inconsistent with the other two cases.

Move the execution of _DSW and _PSW into a separate function called
acpi_device_sleep_wake() and call it wherever appropriate instead of
executing _DSW and/or _PSW directly.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: rework pci_set_power_state function to call platform first
Rafael J. Wysocki [Mon, 7 Jul 2008 01:32:52 +0000 (03:32 +0200)]
PCI: rework pci_set_power_state function to call platform first

Rework pci_set_power_state() so that the platform callback is
invoked before the native mechanism, if necessary.  Also, make
the function check if the device is power manageable by the
platform before invoking the platform callback.

This may matter if the device dependent on additional power
resources controlled by the platform is being put into D0, in which
case those power resources must be turned on before we attempt to
handle the device itself.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: Introduce platform_pci_power_manageable function
Rafael J. Wysocki [Mon, 7 Jul 2008 01:32:02 +0000 (03:32 +0200)]
PCI: Introduce platform_pci_power_manageable function

Introduce function pointer platform_pci_power_manageable to be used
by the platform-related code to point to a function allowing us to
check if given device is power manageable by the platform.

Introduce acpi_pci_power_manageable() playing that role for ACPI.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoACPI: Introduce acpi_bus_power_manageable function
Rafael J. Wysocki [Mon, 7 Jul 2008 01:30:55 +0000 (03:30 +0200)]
ACPI: Introduce acpi_bus_power_manageable function

Introduce function acpi_bus_power_manageable() allowing other
(dependent) subsystems to check if ACPI is able to power manage given
device.  This may be useful, for example, for PCI device power
management.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: make pci_name use dev_name
Greg KH [Thu, 3 Jul 2008 16:49:39 +0000 (09:49 -0700)]
PCI: make pci_name use dev_name

Also fixes up the sparc code that was assuming this is not a constant.

Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: handle pci_name() being const
Greg Kroah-Hartman [Wed, 2 Jul 2008 20:24:49 +0000 (13:24 -0700)]
PCI: handle pci_name() being const

This changes pci_setup_device to handle pci_name() now returning a
constant string.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: add stub for pci_set_consistent_dma_mask()
Randy Dunlap [Mon, 30 Jun 2008 18:35:53 +0000 (11:35 -0700)]
PCI: add stub for pci_set_consistent_dma_mask()

When CONFIG_PCI=n, there is no stub for pci_set_consistent_dma_mask(),
so add one like other similar stubs.  Otherwise there can be build errors,
as here:

linux-next-20080630/drivers/ssb/main.c:1175: error: implicit declaration of
function 'pci_set_consistent_dma_mask'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: remove unused arch pcibios_update_resource() functions
Adrian Bunk [Tue, 17 Jun 2008 22:33:32 +0000 (01:33 +0300)]
PCI: remove unused arch pcibios_update_resource() functions

Russell King did the following back in 2003:

<--  snip  -->

    [PCI] pci-9: Kill per-architecture pcibios_update_resource()

    Kill pcibios_update_resource(), replacing it with pci_update_resource().
    pci_update_resource() uses pcibios_resource_to_bus() to convert a
    resource to a device BAR - the transformation should be exactly the
    same as the transformation used for the PCI bridges.

    pci_update_resource "knows" about 64-bit BARs, but doesn't attempt to
    set the high 32-bits to anything non-zero - currently no architecture
    attempts to do something different.  If anyone cares, please fix; I'm
    going to reflect current behaviour for the time being.

    Ivan pointed out the following architectures need to examine their
    pcibios_update_resource() implementation - they should make sure that
    this new implementation does the right thing.  #warning's have been
    added where appropriate.

        ia64
        mips
        mips64

    This cset also includes a fix for the problem reported by AKPM where
    64-bit arch compilers complain about the resource mask being placed
    in a u32.

<--  snip  -->

This patch removes the unused pcibios_update_resource() functions the
kernel gained since, from FRV, m68k, mips & sh architectures.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: David Howells <dhowells@redhat.com>
Acked-by: Greg Ungerer <gerg@uclinux.org>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: fix pci_setup_device()'s sprinting into a const buffer
David Howells [Fri, 27 Jun 2008 12:23:20 +0000 (13:23 +0100)]
PCI: fix pci_setup_device()'s sprinting into a const buffer

Make pci_setup_device() write the bus ID directly into the allotted storage,
rather than using pci_name() as the address as that now returns a const
pointer.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: Fix comment of pci_dynids
Wang Chen [Thu, 26 Jun 2008 02:50:04 +0000 (10:50 +0800)]
PCI: Fix comment of pci_dynids

struct pci_driver has no field of driver_data.
It's in pci_device_id.

Signed-off-by: Wang Chen <wangchen@cn.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: use get_service_data
Kenji Kaneshige [Thu, 26 Jun 2008 11:06:24 +0000 (20:06 +0900)]
pciehp: use get_service_data

Current pciehp driver saves its private data pointer into pci_dev
structure using pci_set_drvdata()/pci_get_drvdata(). But because
pciehp is not a pci device driver but a PCI Express service driver, it
should save its private data pointer into pcie_device structure using
set_service_data()/get_service_data().

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: remove needless command completed interrupt setting
Kenji Kaneshige [Thu, 26 Jun 2008 11:07:33 +0000 (20:07 +0900)]
pciehp: remove needless command completed interrupt setting

Currently, pciehp driver enables command completed interrupt as follows.

(1) Don't enable at initialization.
(2) Enable command completed interrupt whenever pciehp issues a
    command, if the command doesn't attempt to disable the interrupt.
(3) Disable command completed interrupt at driver unloading.

Once we enable command completed interrupt, we don't need to re-enable
it for every command. So we can simplify above steps as follows:

(1) Enable command completed interrupt at initialization.
(2) No special sequence for command completed interrupt.
(3) Disable command completed interrupt at driver unloading.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: fix interrupt initialization
Kenji Kaneshige [Fri, 20 Jun 2008 03:07:08 +0000 (12:07 +0900)]
pciehp: fix interrupt initialization

Current pciehp driver's intialization sequence is as follows:

(1) initialize controller data structure
(2) install interrupt handler
(3) enable software notification
(4) initialize controller specific slot data structure
(5) initialize generic slot data structure and register it to pci hotplug core

The interrupt handler of pciehp assumes that controller specific slot
data structure is already initialized. However, it is installed at (2)
before initializing controller specific slot data structure at
(4). Because of this, pciehp driver cannot handle the following cases
properly.

- If devices that shares IRQ with pciehp raise interrupts between (2) and (4).
- If hotplug events (e.g. MRL open) happen between (3) and (4).

We already have a workaround for this problem ("pciehp: fix NULL
dereference in interrupt handler: dbd79aed1aea2bece0bf43cc2ff3b2f9baf48a08).
But we still need fundamental fix.

This patch fix the problem by changing the initilization sequence as follows:

(1) initialize controller data structure
(2) initialize controller specific slot data structure
(3) install interrupt handler
(4) enable software notification
(5) initialize generic slot data structure and register it to pci hotplug core

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Acked-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: correct resource number in debug output
Jesse Barnes [Wed, 25 Jun 2008 23:12:25 +0000 (16:12 -0700)]
PCI: correct resource number in debug output

If pci_request_region fails, make the warning include the resource number,
not the resource number + 1.

16 years agoPCIE: port driver: use dev_printk when possible
Bjorn Helgaas [Fri, 13 Jun 2008 16:52:13 +0000 (10:52 -0600)]
PCIE: port driver: use dev_printk when possible

Convert printks to use dev_printk().

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCIE: aer: use dev_printk when possible
Bjorn Helgaas [Fri, 13 Jun 2008 16:52:12 +0000 (10:52 -0600)]
PCIE: aer: use dev_printk when possible

Convert printks to use dev_printk().

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: use dev_printk when possible
Bjorn Helgaas [Fri, 13 Jun 2008 16:52:11 +0000 (10:52 -0600)]
PCI: use dev_printk when possible

Convert printks to use dev_printk().

I converted pr_debug() to dev_dbg().  Both use KERN_DEBUG and are enabled
only when DEBUG is defined.

I converted printk(KERN_DEBUG) to dev_printk(KERN_DEBUG), not to dev_dbg(),
because dev_dbg() is only enabled when DEBUG is defined.

I converted DBG(KERN_INFO) (only in setup-bus.c) to dev_info().  The DBG()
name makes it sound like debug, but it's been enabled forever, so dev_info()
preserves the previous behavior.

I tried to make the resource assignment formats more consistent, e.g.,
  "BAR %d: got res [%#llx-%#llx] bus [%#llx-%#llx] flags %#lx\n"
instead of sometimes using "start-end" and sometimes using "size@start".
I'm not attached to one or the other; I'd just like them consistent.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoRemove unnecessary 'tmp' variable from pci_hp_register().
Kenji Kaneshige [Fri, 20 Jun 2008 02:54:06 +0000 (11:54 +0900)]
Remove unnecessary 'tmp' variable from pci_hp_register().

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: remove inline from command related functions
Kenji Kaneshige [Fri, 20 Jun 2008 03:05:52 +0000 (12:05 +0900)]
pciehp: remove inline from command related functions

The pcie_poll_cmd() and pcie_wait_cmd() are too large to be
inlined.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: change command polling frequency
Kenji Kaneshige [Fri, 20 Jun 2008 03:05:12 +0000 (12:05 +0900)]
pciehp: change command polling frequency

Change command polling frequency to 100Hz from 10Hz in order to reduce
the delay in the common case of a command completing quickly.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: cleanup pcie_poll_cmd
Kenji Kaneshige [Fri, 20 Jun 2008 03:04:33 +0000 (12:04 +0900)]
pciehp: cleanup pcie_poll_cmd

Cleanup pcie_poll_cmd(): check the slot status once before entering our
completion test loop and convert the loop to a simpler while() block.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: fix typo in hpc_release_ctlr
Kenji Kaneshige [Fri, 20 Jun 2008 03:06:24 +0000 (12:06 +0900)]
pciehp: fix typo in hpc_release_ctlr

Fix the typo in hpc_release_ctlr().

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: fixup kdoc blocks for hotplug functions
Jesse Barnes [Wed, 25 Jun 2008 22:27:34 +0000 (15:27 -0700)]
PCI: fixup kdoc blocks for hotplug functions

A few warnings snuck in as parameters were added or renamed.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agox86/PCI: fixup early quirk probing
Jesse Barnes [Mon, 16 Jun 2008 22:29:45 +0000 (15:29 -0700)]
x86/PCI: fixup early quirk probing

On x86, we do early PCI probing to apply some quirks for chipset bugs.
However, in a recent cleanup (7bcbc78dea92fdf0947fa48e248da3c993a5690f) a
thinko was introduced that causes us to probe all subfunctions of even single
function devices (a function was factored out of an inner loop and a "break"
became a "return").  Fix that up by making check_dev_quirk() return a value so
we can keep the factored code intact.

Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoSuspend-related patches for 2.6.27
Rafael J. Wysocki [Thu, 12 Jun 2008 21:24:06 +0000 (23:24 +0200)]
Suspend-related patches for 2.6.27

ACPI PM: Add possibility to change suspend sequence

There are some systems out there that don't work correctly with
our current suspend/hibernation code ordering.  Provide a workaround
for these systems allowing them to pass 'acpi_sleep=old_ordering' in
the kernel command line so that it will use the pre-ACPI 2.0 ("old")
suspend code ordering.

Unfortunately, this requires us to add a platform hook to the
resuming of devices for recovering the platform in case one of the
device drivers' .suspend() routines returns error code.  Namely,
ACPI 1.0 specifies that _PTS should be called before suspending
devices, but _WAK still should be called before resuming them in
order to undo the changes made by _PTS.  However, if there is an
error during suspending devices, they are automatically resumed
without returning control to the PM core, so the _WAK has to be
called from within device_resume() in that cases.

The patch also reorders and refactors the ACPI suspend/hibernation
code to avoid duplication as far as reasonably possible.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoMerge branch 'suspend' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Jesse Barnes [Thu, 12 Jun 2008 19:06:58 +0000 (12:06 -0700)]
Merge branch 'suspend' of git://git./linux/kernel/git/lenb/linux-acpi-2.6 into linux-next

16 years agoACPI PM: Remove obsolete Toshiba workaround
Rafael J. Wysocki [Wed, 4 Jun 2008 23:17:28 +0000 (01:17 +0200)]
ACPI PM: Remove obsolete Toshiba workaround

Remove the obsolete workaround for a Toshiba Satellite 4030cdt
S1 problem from drivers/acpi/sleep/main.c .

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoPCI ACPI: Drop the second argument of platform_pci_choose_state
Rafael J. Wysocki [Wed, 4 Jun 2008 23:16:37 +0000 (01:16 +0200)]
PCI ACPI: Drop the second argument of platform_pci_choose_state

Since the second argument of acpi_pci_choose_state() and
platform_pci_choose_state() is never used, remove it.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI PM: acpi_pm_device_sleep_state() cleanup
David Brownell [Wed, 4 Jun 2008 23:15:40 +0000 (01:15 +0200)]
ACPI PM: acpi_pm_device_sleep_state() cleanup

Get rid of a superfluous acpi_pm_device_sleep_state() parameter.  The
only legitimate value of that parameter must be derived from the first
parameter, which is what all the callers already do.  (However, this
does not address the fact that ACPI still doesn't set up those flags.)

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoPCI: ACPI PCI slot detection driver
Alex Chiang [Tue, 10 Jun 2008 21:30:42 +0000 (15:30 -0600)]
PCI: ACPI PCI slot detection driver

Detect all physical PCI slots as described by ACPI, and create entries in
/sys/bus/pci/slots/.

Not all physical slots are hotpluggable, and the acpiphp module does not
detect them.  Now we know the physical PCI geography of our system, without
caring about hotplug.

[kaneshige.kenji@jp.fujitsu.com: export-kobject_rename-for-pci_hotplug_core]
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Acked-by: Greg KH <greg@kroah.com>
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: fix build with CONFIG_DMI=n]
Signed-off-by: Alex Chiang <achiang@hp.com>
Cc: Greg KH <greg@kroah.com>
Cc: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Cc: Len Brown <lenb@kernel.org>
Acked-by: Len Brown <len.brown@intel.com>
Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: introduce pci_slot
Alex Chiang [Tue, 10 Jun 2008 21:28:50 +0000 (15:28 -0600)]
PCI: introduce pci_slot

Currently, /sys/bus/pci/slots/ only exposes hotplug attributes when a
hotplug driver is loaded, but PCI slots have attributes such as address,
speed, width, etc.  that are not related to hotplug at all.

Introduce pci_slot as the primary data structure and kobject model.
Hotplug attributes described in hotplug_slot become a secondary
structure associated with the pci_slot.

This patch only creates the infrastructure that allows the separation of
PCI slot attributes and hotplug attributes.  In this patch, the PCI
hotplug core remains the only user of this infrastructure, and thus,
/sys/bus/pci/slots/ will still only become populated when a hotplug
driver is loaded.

A later patch in this series will add a second user of this new
infrastructure and demonstrate splitting the task of exposing pci_slot
attributes from hotplug_slot attributes.

  - Make pci_slot the primary sysfs entity. hotplug_slot becomes a
    subsidiary structure.
    o pci_create_slot() creates and registers a slot with the PCI core
    o pci_slot_add_hotplug() gives it hotplug capability

  - Change the prototype of pci_hp_register() to take the bus and
    slot number (on parent bus) as parameters.

  - Remove all the ->get_address methods since this functionality is
    now handled by pci_slot directly.

[achiang@hp.com: rpaphp-correctly-pci_hp_register-for-empty-pci-slots]
Tested-by: Badari Pulavarty <pbadari@us.ibm.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: make headers_check happy]
[akpm@linux-foundation.org: nuther build fix]
[akpm@linux-foundation.org: fix typo in #include]
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Cc: Greg KH <greg@kroah.com>
Cc: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Cc: Len Brown <lenb@kernel.org>
Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: construct one fakephp slot per PCI slot
Alex Chiang [Tue, 10 Jun 2008 21:27:37 +0000 (15:27 -0600)]
PCI: construct one fakephp slot per PCI slot

Register one slot per slot, rather than one slot per function.  Change the
name of the slot to fake%d instead of the pci address.

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Cc: Greg KH <greg@kroah.com>
Cc: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Cc: Len Brown <lenb@kernel.org>
Cc: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: update location of PCI hotplug mailing lists
Jesse Barnes [Tue, 10 Jun 2008 21:20:03 +0000 (14:20 -0700)]
PCI: update location of PCI hotplug mailing lists

Just concentrate PCI traffic at linux-pci@vger rather than splitting it between
vger and a sf.net moderated (!) mailing list.

Acked-by: Matthew Wilcox <willy@linux.intel.com>
Acked-by: Scott Murray <scottm@somanetworks.com>
Acked-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Tue, 10 Jun 2008 19:34:29 +0000 (12:34 -0700)]
Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6:
  fix BLK_DEV_HD_ONLY on ARM dependencies
  ide: export ide_doubler
  palm_bk3710: add warm-plug support
  delkin_cb: add missing __init/__exit tags
  delkin_cb: add warm-plug support
  delkin_cb: use struct ide_port_info
  delkin_cb: set proper hwif->gendev.parent value
  ide: fix host drivers missing hwif->chipset initialization
  ide-generic: add missing hwif->chipset setup
  sis5513: add missing pci_enable_device() call
  MAINTAINERS: remove SIS 5513 IDE entry
  ide: remove the ide_etrax100 chipset type

16 years agoPCI: unhide the SMBus on the Compaq Deskpro EN
Krzysztof Helt [Sun, 8 Jun 2008 11:47:02 +0000 (13:47 +0200)]
PCI: unhide the SMBus on the Compaq Deskpro EN

This patch unhides the SMBus on Compaq Deskpro EN
SFF P667 with the Intel 815E chipset. Unhiding it reveals
a THMC51 hardware monitoring chip.

Jean Delvare has checked that this machine has no ACPI
magic touching the SMBus nor the hardware monitoring chip,
so this should be safe.

The patch was tested on Fedora Core 9 with 2.6.25.4 kernel.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Tested-by: Rafał Haładuda <rh1985@wp.pl>
CC: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agofix BLK_DEV_HD_ONLY on ARM dependencies
Adrian Bunk [Tue, 10 Jun 2008 18:56:38 +0000 (20:56 +0200)]
fix BLK_DEV_HD_ONLY on ARM dependencies

This patch limits BLK_DEV_HD_ONLY to the ARM platforms offering
IRQ_HARDDISK, fixing the following compile error on others:

<--  snip  -->

...
  CC      drivers/ide/legacy/hd.o
...
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ide/legacy/hd.c: In function 'hd_times_out':
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ide/legacy/hd.c:542: error: 'IRQ_HARDDISK' undeclared (first use in this function)
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ide/legacy/hd.c:542: error: (Each undeclared identifier is reported only once
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ide/legacy/hd.c:542: error: for each function it appears in.)
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ide/legacy/hd.c: In function 'do_hd_request':
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ide/legacy/hd.c:661: error: 'IRQ_HARDDISK' undeclared (first use in this function)
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ide/legacy/hd.c: In function 'hd_init':
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ide/legacy/hd.c:765: error: 'IRQ_HARDDISK' undeclared (first use in this function)
make[3]: *** [drivers/ide/legacy/hd.o] Error 1

<--  snip  -->

Reported-by: Adrian Bunk <bunk@kernel.org>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: export ide_doubler
Adrian Bunk [Tue, 10 Jun 2008 18:56:38 +0000 (20:56 +0200)]
ide: export ide_doubler

This patch fixes the following build error:

<--  snip  -->

...
  Building modules, stage 2.
  MODPOST 1204 modules
ERROR: "ide_doubler" [drivers/ide/ide-core.ko] undefined!
...
make[2]: *** [__modpost] Error 1

<--  snip  -->

Reported-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agopalm_bk3710: add warm-plug support
Bartlomiej Zolnierkiewicz [Tue, 10 Jun 2008 18:56:38 +0000 (20:56 +0200)]
palm_bk3710: add warm-plug support

Don't fail the probe if there are no devices attached to the controller.

Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agodelkin_cb: add missing __init/__exit tags
Bartlomiej Zolnierkiewicz [Tue, 10 Jun 2008 18:56:38 +0000 (20:56 +0200)]
delkin_cb: add missing __init/__exit tags

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agodelkin_cb: add warm-plug support
Bartlomiej Zolnierkiewicz [Tue, 10 Jun 2008 18:56:38 +0000 (20:56 +0200)]
delkin_cb: add warm-plug support

Don't fail the probe if there are no devices attached to the controller.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agodelkin_cb: use struct ide_port_info
Bartlomiej Zolnierkiewicz [Tue, 10 Jun 2008 18:56:37 +0000 (20:56 +0200)]
delkin_cb: use struct ide_port_info

Convert the driver to use struct ide_port_info - as a nice side-effect
this fixes racy setup of ->io_32bit/unmask settings (after ide_device_add()
call device can be already in use).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agodelkin_cb: set proper hwif->gendev.parent value
Bartlomiej Zolnierkiewicz [Tue, 10 Jun 2008 18:56:37 +0000 (20:56 +0200)]
delkin_cb: set proper hwif->gendev.parent value

hwif->dev was set too late (after ide_device_add() call)
so hwif->gendev.parent was not initialized properly.

Fix it by setting hw.dev and letting ide_init_port_hw()
do the rest.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: fix host drivers missing hwif->chipset initialization
Bartlomiej Zolnierkiewicz [Tue, 10 Jun 2008 18:56:37 +0000 (20:56 +0200)]
ide: fix host drivers missing hwif->chipset initialization

ide_find_port() now depends on ->chipset being set for occupied ide_hwifs[]
slots so all host drivers have to initialize hwif->chipset properly.

This patch fixes a regression on hosts with > 1 port or with a single port
but no devices attached to it for an affected host drivers.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-generic: add missing hwif->chipset setup
Bartlomiej Zolnierkiewicz [Tue, 10 Jun 2008 18:56:36 +0000 (20:56 +0200)]
ide-generic: add missing hwif->chipset setup

hwif->chipset need to be set properly or ide-generic driver will break once
we make a final step in fixing host drivers' dependence on ide_hwifs[].

Problem was catched early thanks to IDE tree exposure in -mm / -next trees
and reported by people listed people (thank you guys!).

Reported-by: "John Keller" <jpk@sgi.com>
Reported-by: Dmitri Vorobiev <dmitri.vorobiev@gmail.com>
Reported-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agosis5513: add missing pci_enable_device() call
Bartlomiej Zolnierkiewicz [Tue, 10 Jun 2008 18:56:36 +0000 (20:56 +0200)]
sis5513: add missing pci_enable_device() call

Cc: Riccardo Gori <goric@trivenet.it>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoMAINTAINERS: remove SIS 5513 IDE entry
Adrian Bunk [Tue, 10 Jun 2008 18:56:36 +0000 (20:56 +0200)]
MAINTAINERS: remove SIS 5513 IDE entry

- maintainer has not been active for years
- URLs no longer exist
- covered by the IDE SUBSYSTEM entry
- maintainer email bounces

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Lionel.Bouton@inet6.fr
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove the ide_etrax100 chipset type
Adrian Bunk [Tue, 10 Jun 2008 18:56:36 +0000 (20:56 +0200)]
ide: remove the ide_etrax100 chipset type

I forgot to remove the ide_etrax100 chipset type when removing the
ETRAX_IDE driver.

Reported-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoPCIe: fix 'symbol not declared' sparse warnings
Alex Chiang [Wed, 4 Jun 2008 17:39:07 +0000 (11:39 -0600)]
PCIe: fix 'symbol not declared' sparse warnings

While refreshing my physical PCI slot series against upstream, I
noticed a few simple sparse/compile warnings that were easy to
fix.

Fix the following sparse warnings in PCIe:

drivers/pci/pcie/aer/aerdrv.c:86:6: warning: symbol 'pci_no_aer'
was not declared. Should it be static?
drivers/pci/pcie/portdrv_bus.c:21:17: warning: symbol
'pcie_port_bus_type' was not declared. Should it be static?

Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: Unhide the SMBus on the Compaq Evo D510
Jean Delvare [Wed, 4 Jun 2008 11:53:31 +0000 (13:53 +0200)]
PCI: Unhide the SMBus on the Compaq Evo D510

One more machine with a hidden Intel SMBus. Unhiding it reveals a SMSC
EMC6D100 hardware monitoring chip. I have checked that this machine
has no ACPI magic touching the SMBus nor the hardware monitoring chip,
so this should be safe.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: eliminate double kfree in intel-iommu initialization
Julia Lawall [Thu, 29 May 2008 13:04:38 +0000 (15:04 +0200)]
PCI: eliminate double kfree in intel-iommu initialization

The destination of goto error also does a kfree(g_iommus), so it is not
correct to do one here.

This was found using Coccinelle (http://www.emn.fr/x-info/coccinelle/).

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: removes redundant NULL write to slot status register
Kenji Kaneshige [Mon, 2 Jun 2008 16:22:34 +0000 (09:22 -0700)]
pciehp: removes redundant NULL write to slot status register

Cleanup to remove a redundant NULL write to SLOTSTATUS.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: fixes typo in dbg_ctrl() in pciehp_hpc.c
Kenji Kaneshige [Mon, 2 Jun 2008 16:07:46 +0000 (09:07 -0700)]
pciehp: fixes typo in dbg_ctrl() in pciehp_hpc.c

Fixup a typo in dbg_ctrl(); it was fetching SLOTSTATUS twice.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoshpchp: check firmware before taking control
Kenji Kaneshige [Wed, 28 May 2008 06:01:03 +0000 (15:01 +0900)]
shpchp: check firmware before taking control

Fix the following problems of shpchp driver about getting hotplug
control from firmware.

  - The shpchp driver must not control the hotplug controller if it
    fails to get control from the firmware. But current shpchp
    controls the hotplug controller regardless the result, because it
    doesn't check the return value of get_hp_hw_control_from_firmware().

  - Current shpchp driver doesn't support _OSC.

The pciehp driver already have the code for evaluating _OSC and OSHP
and shpchp and pciehp can share it. So this patch move that code from
pciehp to acpi_pcihp.c.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: block signals while waiting for command completion
Kenji Kaneshige [Wed, 28 May 2008 05:59:44 +0000 (14:59 +0900)]
pciehp: block signals while waiting for command completion

Since we need to wait for command completion for muximum 1sec, waiting
command should not be interrupted by a signal.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: evaluate _OSC/OSHP before controller init
Kenji Kaneshige [Wed, 28 May 2008 05:57:30 +0000 (14:57 +0900)]
pciehp: evaluate _OSC/OSHP before controller init

Current pciehp evaluates _OSC/OSHP method after some controller
initialization is done. So if evaluating _OSC/OSHP is failed, we need
to cleanup already initialized data structures or hardware. This
clearly is not robust way. With this patch, _OSC/OSHP evaluation is
done first.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopciehp: remove redundant pci_dev initialization
Kenji Kaneshige [Wed, 28 May 2008 05:56:00 +0000 (14:56 +0900)]
pciehp: remove redundant pci_dev initialization

Remove the redundant initialization of pci_dev member of struct
controller in pciehp_probe(). It is initialized in pcie_init().

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI/x86: early dump pci conf space v2
Yinghai Lu [Thu, 22 May 2008 21:35:11 +0000 (14:35 -0700)]
PCI/x86: early dump pci conf space v2

Allows us to dump PCI space before any kernel changes have been made.

Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI/x86: write_pci_config_byte fix offset
Yinghai Lu [Thu, 22 May 2008 21:35:21 +0000 (14:35 -0700)]
PCI/x86: write_pci_config_byte fix offset

also add write_pci_config_16

Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoImplement new suspend and hibernation callbacks for platform busses
Rafael J. Wysocki [Tue, 20 May 2008 23:40:43 +0000 (01:40 +0200)]
Implement new suspend and hibernation callbacks for platform busses

Implement new suspend and hibernation callbacks for the platform bus
type.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Greg KH <greg@kroah.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: implement new suspend/resume callbacks
Rafael J. Wysocki [Mon, 19 May 2008 22:49:04 +0000 (00:49 +0200)]
PCI: implement new suspend/resume callbacks

Implement new suspend and hibernation callbacks for the PCI bus type.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoIntroduce new top level suspend and hibernation callbacks
Rafael J. Wysocki [Tue, 20 May 2008 21:00:01 +0000 (23:00 +0200)]
Introduce new top level suspend and hibernation callbacks

Introduce 'struct pm_ops' and 'struct pm_ext_ops' ('ext' meaning
'extended') representing suspend and hibernation operations for bus
types, device classes, device types and device drivers.

Modify the PM core to use 'struct pm_ops' and 'struct pm_ext_ops'
objects, if defined, instead of the ->suspend(), ->resume(),
->suspend_late(), and ->resume_early() callbacks (the old callbacks
will be considered as legacy and gradually phased out).

The main purpose of doing this is to separate suspend (aka S2RAM and
standby) callbacks from hibernation callbacks in such a way that the
new callbacks won't take arguments and the semantics of each of them
will be clearly specified.  This has been requested for multiple
times by many people, including Linus himself, and the reason is that
within the current scheme if ->resume() is called, for example, it's
difficult to say why it's been called (ie. is it a resume from RAM or
from hibernation or a suspend/hibernation failure etc.?).

The second purpose is to make the suspend/hibernation callbacks more
flexible so that device drivers can handle more than they can within
the current scheme.  For example, some drivers may need to prevent
new children of the device from being registered before their
->suspend() callbacks are executed or they may want to carry out some
operations requiring the availability of some other devices, not
directly bound via the parent-child relationship, in order to prepare
for the execution of ->suspend(), etc.

Ultimately, we'd like to stop using the freezing of tasks for suspend
and therefore the drivers' suspend/hibernation code will have to take
care of the handling of the user space during suspend/hibernation.
That, in turn, would be difficult within the current scheme, without
the new ->prepare() and ->complete() callbacks.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: boot parameter to avoid expansion ROM memory allocation
Gary Hade [Mon, 12 May 2008 20:57:46 +0000 (13:57 -0700)]
PCI: boot parameter to avoid expansion ROM memory allocation

Contention for scarce PCI memory resources has been growing
due to an increasing number of PCI slots in large multi-node
systems.  The kernel currently attempts by default to
allocate memory for all PCI expansion ROMs so there has
also been an increasing number of PCI memory allocation
failures seen on these systems.  This occurs because the
BIOS either (1) provides insufficient PCI memory resource
for all the expansion ROMs or (2) provides adequate PCI
memory resource for expansion ROMs but provides the
space in kernel unexpected BIOS assigned P2P non-prefetch
windows.

The resulting PCI memory allocation failures may be benign
when related to memory requests for expansion ROMs themselves
but in some cases they can occur when attempting to allocate
space for more critical BARs.  This can happen when a successful
expansion ROM allocation request consumes memory resource
that was intended for a non-ROM BAR.  We have seen this
happen during PCI hotplug of an adapter that contains a
P2P bridge where successful memory allocation for an
expansion ROM BAR on device behind the bridge consumed
memory that was intended for a non-ROM BAR on the P2P bridge.
In all cases the allocation failure messages can be very
confusing for users.

This patch provides a new 'pci=norom' kernel boot parameter
that can be used to disable the default PCI expansion ROM memory
resource allocation.  This provides a way to avoid the above
described issues on systems that do not contain PCI devices
for which drivers or user-level applications depend on the
default PCI expansion ROM memory resource allocation behavior.

Signed-off-by: Gary Hade <garyhade@us.ibm.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI/MSI: skip calling pci_find_capability from msi_set_mask_bits
Hidetoshi Seto [Mon, 19 May 2008 04:48:17 +0000 (13:48 +0900)]
PCI/MSI: skip calling pci_find_capability from msi_set_mask_bits

The position of MSI capability is already cached in the msi_desc when
we enter the msi_set_mask_bits().  Use it instead.

Signed-off-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: remove CVS keywords
Adrian Bunk [Mon, 19 May 2008 22:02:06 +0000 (01:02 +0300)]
PCI: remove CVS keywords

This patch removes CVS keywords that weren't updated for a long time
from comments.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopci-acpi: formatting cleanups for _OSC
Kenji Kaneshige [Thu, 15 May 2008 06:23:13 +0000 (15:23 +0900)]
pci-acpi: formatting cleanups for _OSC

Minor cleanup in pci-acpi.c.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Acked-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopci-acpi: remove unused variable in __pci_osc_support_set
Kenji Kaneshige [Thu, 15 May 2008 06:22:35 +0000 (15:22 +0900)]
pci-acpi: remove unused variable in __pci_osc_support_set

The 'retval' variable in __pci_osc_support_set() is no longer
used. Remove this unused variable.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Acked-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopci-acpi: add flag to indicate query had been done
Kenji Kaneshige [Thu, 15 May 2008 06:21:16 +0000 (15:21 +0900)]
pci-acpi: add flag to indicate query had been done

Current pci-acpi implementation checks osc_data->support_stat to see
if control bits had been already queried. It is not good from the
viewpoint of easy understanding. So this patch adds new 'is_queried'
flag to indicate query had been done.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Acked-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopci-acpi: use local buffer for _OSC
Kenji Kaneshige [Thu, 15 May 2008 06:20:11 +0000 (15:20 +0900)]
pci-acpi: use local buffer for _OSC

Current pci-acpi implementation uses array in osc_data directly to
evaluate _OSC. It needs to save the old data and restore it if _OSC
evaluation fails. To make it more robust, we should use local array to
evaluate _OSC.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Acked-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agopci-acpi: remove duplicate code for _OSC
Kenji Kaneshige [Thu, 15 May 2008 06:18:53 +0000 (15:18 +0900)]
pci-acpi: remove duplicate code for _OSC

Remove the duplicated code in acpi_query_osc() and acpi_run_osc().
It simplifies the code very much.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Acked-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: Disable PME during PCI scan
Zhao Yakui [Tue, 13 May 2008 03:15:05 +0000 (11:15 +0800)]
PCI: Disable PME during PCI scan

If a device supports #PME and can generate PME events from D0, we may see
superfluous events before a driver is loaded (drivers should only enable PME as
needed), preventing suspend from working if the corresponding GPE was enabled.

Likewise, if the ACPI device has the _PRW object, the _PSW/_DSW object will be
called in order to disable the wakeup functionality. But when it is allowed to
wake up the sleeping state, OSPM will enable it again.

So we should disable PME in the course of scanning PCI devices and enable it
again only when PME events are actually required to be generated from the
requested PCI state (for example, D3_hot or D3_cold).  It is also safe to
disable PME again when the PME is disabled for the PCI devices.

Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Li Shaohua <shaohua.li@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoSuspend/Resume bug in PCI layer wrt quirks
Rafael J. Wysocki [Thu, 15 May 2008 19:51:31 +0000 (21:51 +0200)]
Suspend/Resume bug in PCI layer wrt quirks

Some quirks should be called with interrupt disabled, we can't directly
call them in .resume_early. Also the patch introduces
pci_fixup_resume_early and pci_fixup_suspend, which matches current
device core callbacks (.suspend/.resume_early).

TBD: Somebody knows why we need quirk resume should double check if a
quirk should be called in resume or resume_early. I changed some per my
understanding, but can't make sure I fixed all.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agox86/PCI: janitor work in irq.c
Miklos Vajna [Tue, 13 May 2008 16:38:56 +0000 (18:38 +0200)]
x86/PCI: janitor work in irq.c

Wrapped long lines, removed trailing whitespaces, fixed case indentation
inside switch and so.

Signed-off-by: Miklos Vajna <vmiklos@frugalware.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: Replace deprecated __initcall with device_initcall.
Robert P. J. Day [Sun, 11 May 2008 20:58:53 +0000 (16:58 -0400)]
PCI: Replace deprecated __initcall with device_initcall.

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: drivers/pci/pci.c: add prototypes
Adrian Bunk [Mon, 5 May 2008 18:25:47 +0000 (21:25 +0300)]
PCI: drivers/pci/pci.c: add prototypes

This patch adds prototypes for pcibios_disable_device() and
pcibios_set_pcie_reset_state() in include/linux/pci.h

While I was at it, I also removed the unneeded "extern" from the
prototype of pcibios_add_platform_entries().

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoPCI: make {pciehp,shpchp}_slot_with_bus static
Adrian Bunk [Mon, 5 May 2008 18:21:51 +0000 (21:21 +0300)]
PCI: make {pciehp,shpchp}_slot_with_bus static

This patch makes the needlessly global {pciehp,shpchp}_slot_with_bus
static.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
16 years agoMerge branch 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Tue, 10 Jun 2008 15:00:40 +0000 (08:00 -0700)]
Merge branch 'for-linus' of git://git390.osdl.marist.edu/linux-2.6

* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
  [S390] tape_3590.c: introduce missing kfree
  [S390] Fix __ctl_load/__ctl_store inline assembly constraints
  [S390] Fix build failure in __cpu_up()
  [S390] vt220 console, initialize list head before use
  [S390] cio: Fix inverted isc priorities.
  [S390] vmemmap: fix off-by-one bug.
  [S390] cio: Fix sparse warnings in blacklist.c.
  [S390] sparsemem: use SPARSEMEM_STATIC if !64BIT.

16 years agoshm: Remove silly double assignment
Neil Horman [Tue, 10 Jun 2008 12:53:39 +0000 (08:53 -0400)]
shm: Remove silly double assignment

Found a silly double assignment of err is do_shmat.  Silly, but good to
clean up the useless code.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years ago[S390] tape_3590.c: introduce missing kfree
Julia Lawall [Tue, 10 Jun 2008 08:03:25 +0000 (10:03 +0200)]
[S390] tape_3590.c: introduce missing kfree

The semantic match that finds the problem is as follows:
(http://www.emn.fr/x-info/coccinelle/)

@r exists@
expression E,E1;
statement S;
position p1,p2,p3;
@@

E =@p1 \(kmalloc\|kcalloc\|kzalloc\)(...)
... when != E = E1
if (E == NULL || ...) S
... when != E = E1
if@p2 (...) {
 ... when != kfree(E)
 }
... when != E = E1
kfree@p3(E);

@forall@
position r.p2;
expression r.E;
int E1 != 0;
@@

* if@p2 (...) {
 ... when != kfree(E)
     when strict
return E1; }

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Fix __ctl_load/__ctl_store inline assembly constraints
Martin Schwidefsky [Tue, 10 Jun 2008 08:03:24 +0000 (10:03 +0200)]
[S390] Fix __ctl_load/__ctl_store inline assembly constraints

__ctl_load/__ctl_store are called with either an array of unsigned long or
a single unsigned long value. Add an address operator to the "m"/"=m"
contraints to make them work for unsigned long arguments as well.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Fix build failure in __cpu_up()
Segher Boessenkool [Tue, 10 Jun 2008 08:03:23 +0000 (10:03 +0200)]
[S390] Fix build failure in __cpu_up()

The first argument to __ctl_store() should be the array to store
stuff in, not just the first element of that array.  With the
current code in __cpu_up(), mainline GCC dies with an internal
compiler error.  I didn't diagnose that further, but just fixed
the kernel bug.

Signed-off-by: Segher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
16 years ago[S390] vt220 console, initialize list head before use
Carsten Otte [Tue, 10 Jun 2008 08:03:22 +0000 (10:03 +0200)]
[S390] vt220 console, initialize list head before use

This patch fixes a null pointer dereference during initialisation when no
sclp event facility is available:
sclp vt220 tty driver: could not register vt220 - sclp_register returned -5
Unable to handle kernel paging request at virtual user address 0000000000000000
Oops: 0004 [#1] PREEMPT SMP
Modules linked in:
CPU: 0 Not tainted 2.6.26-rc3-kvm-bigiron-00968-gd939e93-dirty #30
Process swapper (pid: 0, task: 0000000000600be0, ksp: 000000000064a000)
Krnl PSW : 0400000180000000 0000000000320d8c (sclp_unregister+0x48/0x8c)
           R:0 T:1 IO:0 EX:0 Key:0 M:0 W:0 P:0 AS:0 CC:0 PM:0 EA:3
Krnl GPRS: 0000000000000000 0000000000000000 0000000000630478 0700000000649c20
           0000000000000000 0000000000433060 000000000064a660 0000000002e26000
           00000000006db000 0000000000000000 0000000000a78578 0000000000649b80
           0000000000630dc0 000000000044fa20 0000000000320d76 0000000000649b80
Krnl Code: 0000000000320d7ce310c0080004       lg      %r1,8(%r12)
           0000000000320d82b9040032           lgr     %r3,%r2
           0000000000320d86c02000187b79       larl    %r2,630478
          >0000000000320d8ce34010000024       stg     %r4,0(%r1)
           0000000000320d92e31040080024       stg     %r1,8(%r4)
           0000000000320d98c01100200200       lgfi    %r1,2097664
           0000000000320d9ee310c0080024       stg     %r1,8(%r12)
           0000000000320da4c01100100100       lgfi    %r1,1048832
Call Trace:
([<0000000000320d76>] sclp_unregister+0x32/0x8c)
 [<00000000006657b4>] __sclp_vt220_cleanup+0xc4/0xe0
 [<000000000066595c>] __sclp_vt220_init+0x18c/0x1a0
 [<0000000000665aba>] sclp_vt220_con_init+0x42/0x68
 [<00000000006601ca>] console_init+0x4e/0x68
 [<000000000064acae>] start_kernel+0x3a2/0x4dc
 [<0000000000100020>] _stext+0x20/0x80
INFO: lockdep is turned off.
Last Breaking-Event-Address:
 [<000000000041f964>] _spin_lock_irqsave+0xb0/0xb4
 <4>---[ end trace 31fd0ba7d8756001 ]---

The issue is caused by a list_empty() check in __sclp_vt220_cleanup, which
usually fails on non-initialized list heads that contain {NULL,NULL} instead.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] cio: Fix inverted isc priorities.
Cornelia Huck [Tue, 10 Jun 2008 08:03:21 +0000 (10:03 +0200)]
[S390] cio: Fix inverted isc priorities.

Priorities for I/O interruption subclasses range from 0 (highest)
to 7 (lowest). Unfortunately, the console has been using isc 7
instead of an isc with a higher priority than regular I/O
subchannels (which use 3). Fix this by making the console use
isc 1.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] vmemmap: fix off-by-one bug.
Heiko Carstens [Tue, 10 Jun 2008 08:03:20 +0000 (10:03 +0200)]
[S390] vmemmap: fix off-by-one bug.

If a memory range is supposed to be added to the 1:1 mapping and it
ends just below the maximum supported physical address it won't
succeed. This is because a test doesn't consider that the end address
is 1 smaller than start + size.
Fix the comparison.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] cio: Fix sparse warnings in blacklist.c.
Cornelia Huck [Tue, 10 Jun 2008 08:03:19 +0000 (10:03 +0200)]
[S390] cio: Fix sparse warnings in blacklist.c.

sparse complains about signedness:

drivers/s390/cio/blacklist.c:132:28: warning: incorrect type in
argument 2 (different signedness)
drivers/s390/cio/blacklist.c:132:28:    expected unsigned int *val
drivers/s390/cio/blacklist.c:132:28:    got int *cssid
drivers/s390/cio/blacklist.c:136:28: warning: incorrect type in
argument 2 (different signedness)
drivers/s390/cio/blacklist.c:136:28:    expected unsigned int *val
drivers/s390/cio/blacklist.c:136:28:    got int *ssid
drivers/s390/cio/blacklist.c:140:28: warning: incorrect type in
argument 2 (different signedness)
drivers/s390/cio/blacklist.c:140:28:    expected unsigned int *val
drivers/s390/cio/blacklist.c:140:28:    got int *devno

cssid, ssid and devno are of course unsigned, so let's make the
variables unsigned as well.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] sparsemem: use SPARSEMEM_STATIC if !64BIT.
Heiko Carstens [Tue, 10 Jun 2008 08:03:18 +0000 (10:03 +0200)]
[S390] sparsemem: use SPARSEMEM_STATIC if !64BIT.

In case of !64BIT kernel we end up with a zero sized mem_section array.
This happens because NR_MEM_SECTIONS is smaller than SECTIONS_PER_ROOT
but we have:

#define NR_SECTION_ROOTS (NR_MEM_SECTIONS / SECTIONS_PER_ROOT)

and

struct mem_section *mem_section[NR_SECTION_ROOTS];

So fix this by selecting SPARSEMEM_STATIC which makes sure
that SECTIONS_PER_ROOT is 1.

Cc: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years agoMAINTAINERS: update PPPoE maintainer address
Chris Wright [Mon, 9 Jun 2008 23:07:28 +0000 (16:07 -0700)]
MAINTAINERS: update PPPoE maintainer address

Cc: Michal Ostrowski <mostrows@earthlink.net>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Tue, 10 Jun 2008 02:28:54 +0000 (19:28 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/core: Remove IB_DEVICE_SEND_W_INV capability flag
  IB/umem: Avoid sign problems when demoting npages to integer

16 years agoMerge git://git.infradead.org/users/cbou/battery-2.6.26
Linus Torvalds [Tue, 10 Jun 2008 02:28:31 +0000 (19:28 -0700)]
Merge git://git.infradead.org/users/cbou/battery-2.6.26

* git://git.infradead.org/users/cbou/battery-2.6.26:
  power_supply: Fix race in power_supply_uevent

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney...
Linus Torvalds [Tue, 10 Jun 2008 02:27:52 +0000 (19:27 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/cooloney/blackfin-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney/blackfin-2.6:
  Blackfin serial driver: fix up tty core set_ldisc API change breakage bug
  Blackfin arch: protect only the SPI bus controller with CONFIG_SPI_BFIN
  Blackfin arch: fixup warnings with the new cplb saved values
  Blackfin Serial Driver: Clean up BF54x macro in blackfin UART driver.

16 years agopower_supply: Fix race in power_supply_uevent
Dmitry Baryshkov [Sun, 8 Jun 2008 15:43:42 +0000 (19:43 +0400)]
power_supply: Fix race in power_supply_uevent

Commit 54d29ad33e3483bcc7ca433a21cf294854e5154a (Power Supply: fix race
in device_create) introduced a race in power_supply_uevent. Previously it
checked that power_supply is available by checking for dev->driver_data.
But now dev->driver_data is set before power_supply->dev is initialised.

Signed-off-by: Dmitry Baryshkov <dbaryshkov@gmail.com>
Signed-off-by: Anton Vorontsov <cbouatmailru@gmail.com>
16 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq
Linus Torvalds [Mon, 9 Jun 2008 18:27:55 +0000 (11:27 -0700)]
Merge branch 'fixes' of git://git./linux/kernel/git/davej/cpufreq

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
  [CPUFREQ] Fix format string bug.

16 years ago[CPUFREQ] Fix format string bug.
Chris Wright [Sat, 7 Jun 2008 04:26:02 +0000 (21:26 -0700)]
[CPUFREQ] Fix format string bug.

Format string bug.  Not exploitable, as this is only writable by root,
but worth fixing all the same.

Spotted-by: Ilja van Sprundel <ilja@netric.org>
Signed-off-by: Dave Jones <davej@redhat.com>