To fix a 5-year-old regression, reverse changes made by commit
7ef3639 (PNP: don't fail device init if no DMA channel available).

As an example to show the problem, my sound card provides a
prioritized list of PnP "dependent sets" of requested resources:

  dependent set 0 (preferred) wants DMA 5.
  dependent set 1 (acceptable) will take DMA 5, 6, or 7.
  ...
  dependent set 4 (acceptable) doesn't request a high DMA.

If DMA 5 is not available, pnp_assign_dma has to fail on set 0 so that
pnp_auto_config_dev will move on to set 1 and get DMA 6 or 7.
Instead, pnp_assign_dma adds the resource with flags |=
IORESOURCE_DISABLED and returns success.  pnp_auto_config_dev just
sees success and therefore chooses set 0 with a disabled DMA and never
tries the sets that would have resolved the conflict.

Furthermore, this mode of "success" is unexpected and unhandled in
sound/isa/sb and probably other drivers.  sb assumes that the returned
DMA is enabled and obliviously uses the invalid DMA number.  Observed
consequences were sb successfully grabbing a DMA that was expressly
forbidden by the kernel parameter pnp_reserve_dma.

The only upside to the original change would be as a kludge for
devices that can operate in degraded mode without a DMA but that don't
provide the corresponding non-preferred dependent set.  The right
workaround for those devices is to synthesize the missing set in
quirks.c; otherwise, you're reinventing PnP fallback functionality at
the driver level for that device and all others.

Signed-off-by: David Flater <dave@flaterco.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

The patch copies the flags masked by IORESOURCE_BITS from a resource's
template.  This is necessary because the resource settings require proper
IORESOURCE_BITS which are not known during the definition of these resources
using the "/sys/bus/pnp/*/*/resources" interface.  (In fact, they should not
be set by the user as the resource templates define the proper settings.)

If the patch is not applied, the resource flags are not initialized properly
and obscure messages in the kernel log have been seen ("invalid flags").

Signed-off-by: Witold Szczeponik <Witold.Szczeponik@gmx.net>
Reviewed-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

IORESOURCE_DMA cannot be assigned without utilizing the interface
provided by CONFIG_ISA_DMA_API, specifically request_dma() and
free_dma().  Thus, there's a strict dependency on the config option and
limits IORESOURCE_DMA only to architectures that support ISA-style DMA.

ia64 is not one of those architectures, so pnp_check_dma() no longer
needs to be special-cased for that architecture.

pnp_assign_resources() will now return -EINVAL if IORESOURCE_DMA is
attempted on such a kernel.

Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files.  percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.

percpu.h -> slab.h dependency is about to be removed.  Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability.  As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.

  http://userweb.kernel.org/~tj/misc/slabh-sweep.py

The script does the followings.

* Scan files for gfp and slab usages and update includes such that
  only the necessary includes are there.  ie. if only gfp is used,
  gfp.h, if slab is used, slab.h.

* When the script inserts a new include, it looks at the include
  blocks and try to put the new include such that its order conforms
  to its surrounding.  It's put in the include block which contains
  core kernel includes, in the same order that the rest are ordered -
  alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
  doesn't seem to be any matching order.

* If the script can't find a place to put a new include (mostly
  because the file doesn't have fitting include block), it prints out
  an error message indicating which .h file needs to be added to the
  file.

The conversion was done in the following steps.

1. The initial automatic conversion of all .c files updated slightly
   over 4000 files, deleting around 700 includes and adding ~480 gfp.h
   and ~3000 slab.h inclusions.  The script emitted errors for ~400
   files.

2. Each error was manually checked.  Some didn't need the inclusion,
   some needed manual addition while adding it to implementation .h or
   embedding .c file was more appropriate for others.  This step added
   inclusions to around 150 files.

3. The script was run again and the output was compared to the edits
   from #2 to make sure no file was left behind.

4. Several build tests were done and a couple of problems were fixed.
   e.g. lib/decompress_*.c used malloc/free() wrappers around slab
   APIs requiring slab.h to be added manually.

5. The script was run on all .h files but without automatically
   editing them as sprinkling gfp.h and slab.h inclusions around .h
   files could easily lead to inclusion dependency hell.  Most gfp.h
   inclusion directives were ignored as stuff from gfp.h was usually
   wildly available and often used in preprocessor macros.  Each
   slab.h inclusion directive was examined and added manually as
   necessary.

6. percpu.h was updated not to include slab.h.

7. Build test were done on the following configurations and failures
   were fixed.  CONFIG_GCOV_KERNEL was turned off for all tests (as my
   distributed build env didn't work with gcov compiles) and a few
   more options had to be turned off depending on archs to make things
   build (like ipr on powerpc/64 which failed due to missing writeq).

   * x86 and x86_64 UP and SMP allmodconfig and a custom test config.
   * powerpc and powerpc64 SMP allmodconfig
   * sparc and sparc64 SMP allmodconfig
   * ia64 SMP allmodconfig
   * s390 SMP allmodconfig
   * alpha SMP allmodconfig
   * um on x86_64 SMP allmodconfig

8. percpu.h modifications were reverted so that it could be applied as
   a separate patch and serve as bisection point.

Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.

Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>

pnp_dbg() is equivalent to dev_dbg() except that we can turn it
on at boot-time with the "pnp.debug" kernel parameter, so we don't
have to build a new kernel image.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>

ISAPNP, PNPBIOS, and ACPI describe the "possible resource settings" of
a device, i.e., the possibilities an OS bus driver has when it assigns
I/O port, MMIO, and other resources to the device.

PNP used to maintain this "possible resource setting" information in
one independent option structure and a list of dependent option
structures for each device.  Each of these option structures had lists
of I/O, memory, IRQ, and DMA resources, for example:

  dev
    independent options
      ind-io0  -> ind-io1  ...
      ind-mem0 -> ind-mem1 ...
      ...
    dependent option set 0
      dep0-io0  -> dep0-io1  ...
      dep0-mem0 -> dep0-mem1 ...
      ...
    dependent option set 1
      dep1-io0  -> dep1-io1  ...
      dep1-mem0 -> dep1-mem1 ...
      ...
    ...

This data structure was designed for ISAPNP, where the OS configures
device resource settings by writing directly to configuration
registers.  The OS can write the registers in arbitrary order much
like it writes PCI BARs.

However, for PNPBIOS and ACPI devices, the OS uses firmware interfaces
that perform device configuration, and it is important to pass the
desired settings to those interfaces in the correct order.  The OS
learns the correct order by using firmware interfaces that return the
"current resource settings" and "possible resource settings," but the
option structures above doesn't store the ordering information.

This patch replaces the independent and dependent lists with a single
list of options.  For example, a device might have possible resource
settings like this:

  dev
    options
      ind-io0 -> dep0-io0 -> dep1->io0 -> ind-io1 ...

All the possible settings are in the same list, in the order they
come from the firmware "possible resource settings" list.  Each entry
is tagged with an independent/dependent flag.  Dependent entries also
have a "set number" and an optional priority value.  All dependent
entries must be assigned from the same set.  For example, the OS can
use all the entries from dependent set 0, or all the entries from
dependent set 1, but it cannot mix entries from set 0 with entries
from set 1.

Prior to this patch PNP didn't keep track of the order of this list,
and it assigned all independent options first, then all dependent
ones.  Using the example above, that resulted in a "desired
configuration" list like this:

  ind->io0 -> ind->io1 -> depN-io0 ...

instead of the list the firmware expects, which looks like this:

  ind->io0 -> depN-io0 -> ind-io1 ...

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Rene Herman <rene.herman@gmail.com>
Signed-off-by: Len Brown <len.brown@intel.com>

This patch adds an IORESOURCE_IRQ_OPTIONAL flag for use when
assigning resources to a device.  If the flag is set and we are
unable to assign an IRQ to the device, we can leave the IRQ
disabled but allow the overall resource allocation to succeed.

Some devices request an IRQ, but can run without an IRQ
(possibly with degraded performance).  This flag lets us run
the device without the IRQ instead of just leaving the
device disabled.

This is a reimplementation of this previous change by Rene
Herman <rene.herman@gmail.com>:
    http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=3b73a223661ed137c5d3d2635f954382e94f5a43

I reimplemented this for two reasons:
    - to prepare for converting all resource options into a single linked
      list, as opposed to the per-resource-type lists we have now, and
    - to preserve the order and number of resource options.

In PNPBIOS and ACPI, we configure a device by giving firmware a
list of resource assignments.  It is important that this list
has exactly the same number of resources, in the same order,
as the "template" list we got from the firmware in the first
place.

The problem of a sound card MPU401 being left disabled for want of
an IRQ was reported by Uwe Bugla <uwe.bugla@gmx.de>.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Rene Herman <rene.herman@gmail.com>
Signed-off-by: Len Brown <len.brown@intel.com>

pnp_assign_resources() is static and the only caller checks
pnp_can_configure() before calling it, so no need to do it
again.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Rene Herman <rene.herman@gmail.com>
Signed-off-by: Len Brown <len.brown@intel.com>

This patch doesn't change any behavior; it just makes the return
values more conventional.

This changes pnp_assign_dma() from a void function to one that
returns an int, just like the other assignment functions.  For
now, at least, pnp_assign_dma() always returns 0 (success), so
it appears to never fail, just like before.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Rene Herman <rene.herman@gmail.com>
Signed-off-by: Len Brown <len.brown@intel.com>

When we fail to assign an I/O or MEM resource, include the min/max
in the debug output to help match it with the options.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Rene Herman <rene.herman@gmail.com>
Signed-off-by: Len Brown <len.brown@intel.com>