pandora-kernel.git
12 years agodrm/radeon/kms: add missing vddci setting on NI+
Alex Deucher [Mon, 25 Jul 2011 22:50:08 +0000 (18:50 -0400)]
drm/radeon/kms: add missing vddci setting on NI+

Need to add vddci setting to pm init as well as
resume.  Fixes hangs on load on some boards.

Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=38754

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon: Add a rmb() in IH processing
Benjamin Herrenschmidt [Wed, 13 Jul 2011 06:28:19 +0000 (16:28 +1000)]
drm/radeon: Add a rmb() in IH processing

We should have a read memory barrier between reading the WPTR from
memory and reading ring entries based on that value (ie, we need to
ensure both loads are done in order by the CPU).

It could be argued that the MMIO reads in r600_ack_irq() might be
enough to get that barrier but I prefer keeping an explicit one just
in case.

[airlied: fix evergreen + r/w mixup]

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon: ATOM Endian fix for atombios_crtc_program_pll()
Benjamin Herrenschmidt [Wed, 13 Jul 2011 06:28:15 +0000 (16:28 +1000)]
drm/radeon: ATOM Endian fix for atombios_crtc_program_pll()

v6 of the structure was programmed incorrectly:

  args.v6.ulCrtcPclkFreq.ulPixelClock = cpu_to_le32(clock / 10);

ulPixelClock is a 24-bit bitfield. This statement would thus
do a 32-bit swap of (clock / 10) and drop the top 8 bits which
are ... the LSB. Not what we want. Instead use masks & shifts.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon: Fix the definition of RADEON_BUF_SWAP_32BIT
Benjamin Herrenschmidt [Wed, 13 Jul 2011 06:28:26 +0000 (06:28 +0000)]
drm/radeon: Fix the definition of RADEON_BUF_SWAP_32BIT

(Note that this is duplicated under various other names such
as R600_BUF_SWAP_32BIT etc...). At least now all the definitions
agree.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon: Do an MMIO read on interrupts when not uisng MSIs
Benjamin Herrenschmidt [Wed, 13 Jul 2011 06:28:22 +0000 (06:28 +0000)]
drm/radeon: Do an MMIO read on interrupts when not uisng MSIs

When not using MSIs, there is no guarantee that DMA from the device
has been fully flushed to point where it's visible to the CPU when
taking an interrupt. To get this guarantee, we need to perform an
MMIO read from the device, which will flush all outstanding DMAs
from bridges between the device and the system.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon: Writeback endian fixes
Benjamin Herrenschmidt [Wed, 13 Jul 2011 06:28:17 +0000 (06:28 +0000)]
drm/radeon: Writeback endian fixes

The writeback ring pointer and IH ring pointer are read using le32_to_cpu
so we do not want the chip to byteswap them on big-endian.

We still want to byteswap the ring itself and the IBs, so we don't touch
that but we remove setting of the byteswap bits in CP_RB_RPTR_ADDR and
IH_CNTL.

In general, for things like that where we control all the accessors easily,
we are better off doing the swap in SW rather than HW. Paradoxally, it does
keep the code closer to x86 and avoid using poorly tested HW features.

I also changed the use of RADEON_ to R600_ in a couple of cases to be more
consistent with the surrounding code.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Michel Dänzer <michel@daenzer.net>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon: Remove a bunch of useless _iomem casts
Benjamin Herrenschmidt [Wed, 13 Jul 2011 06:28:12 +0000 (06:28 +0000)]
drm/radeon: Remove a bunch of useless _iomem casts

Just defining rdev->rmmio properly in the first place should do
the trick. In some cases, the cast were also complete dups as
the original variable was already of the right type.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/gem: add support for private objects
Alan Cox [Tue, 7 Jun 2011 13:17:51 +0000 (14:17 +0100)]
drm/gem: add support for private objects

These small changes should allow GEM to be used with non shmem objects as
well as shmem objects. In the GMA500 case it allows the base framebuffer to
appear as a GEM object and thus acquire a handle and work with KMS.

For i915 it ought to be trivial to get back the wasted memory but putting the
system fb back into stolen RAM and in general I can imagine it allowing the
use of GEM and thus KMS with all the older cards that have their framebuffer
firmly placed in video RAM.

Signed-off-by: Alan Cox <alan@linux.intel.com>
Tested-by: Rob Clark <rob@ti.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agoDRM: clean up and document parsing of video= parameter
Rolf Eike Beer [Wed, 15 Jun 2011 09:27:02 +0000 (11:27 +0200)]
DRM: clean up and document parsing of video= parameter

The video= parameter of the DRM drivers supports some additional flags that
the normal fb drivers do not have. They also allow to limit these flags to
specific outputs. Both things were previously undocumented.

Also the parsing of the line had some oddities:
-A lot of misplaced options were silently ignored or partly rejected instead
 of stopping the parsing immediately
-The 'R' option is documented to follow the 'M' option if specified. It is not
 documented that 'M' is needed to specify 'R' (also this is the case for normal
 fb drivers). In fact the code is correct for normal fb drivers but wrong for
 DRM ones.
 The old code allowed 'R' only _before_ 'M' (since it parses backwards) and only
 if 'M' is given at all which is not needed for the DRM drivers.
-the margins option ('m') was parsed but later ignored even if the later
 functions support it.
-specifying multiple enable options at the same time did not lead to an error.
-specifying something bogus for horizontal resolution (i.e. other things as
 digits) did not lead to an error but an invalid resolution was used.

If any errors are encountered the position of the faulting string is now
printed to the user and the complete mode is ignored. This gives much
more consistent error behaviour.

I also removed some useless assignments and changed the local flag variables
to be bool.

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agoDRM: Radeon: Fix section mismatch.
Ralf Baechle [Mon, 27 Jun 2011 13:40:35 +0000 (13:40 +0000)]
DRM: Radeon: Fix section mismatch.

WARNING: drivers/gpu/drm/radeon/radeon.o(.text+0x5d1fc): Section mismatch in reference from the function radeon_get_clock_info() to the function .devinit.text:radeon_read_clocks_OF()
The function radeon_get_clock_info() references
the function __devinit radeon_read_clocks_OF().
This is often because radeon_get_clock_info lacks a __devinit
annotation or the annotation of radeon_read_clocks_OF is wrong.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Cc: dri-devel@lists.freedesktop.org
Cc: linux-kernel@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm: really make debug levels match in edid failure code
Tormod Volden [Tue, 5 Jul 2011 20:12:53 +0000 (20:12 +0000)]
drm: really make debug levels match in edid failure code

Also disable the ascii dump and remove the literal printing of the
KERN_ERR macro in the log:

   [drm:drm_edid_block_valid] *ERROR* Raw EDID:
   <3>00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................

v2: Remove the trailing empty line as well.

Signed-off-by: Tormod Volden <debian.tormod@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon/kms: fix i2c map for rv250/280
Alex Deucher [Sat, 23 Jul 2011 18:02:04 +0000 (18:02 +0000)]
drm/radeon/kms: fix i2c map for rv250/280

Those chips have crt2_ddc bus.

Fixes:
https://bugzilla.kernel.org/show_bug.cgi?id=39672

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agoMerge branch 'drm-nouveau-next' of git://anongit.freedesktop.org/git/nouveau/linux...
Dave Airlie [Mon, 25 Jul 2011 09:15:18 +0000 (10:15 +0100)]
Merge branch 'drm-nouveau-next' of git://anongit.freedesktop.org/git/nouveau/linux-2.6 into drm-core-next

* 'drm-nouveau-next' of git://anongit.freedesktop.org/git/nouveau/linux-2.6:
  drm/nouveau/gr: disable fifo access and idle before suspend ctx unload
  drm/nouveau: pass flag to engine fini() method on suspend
  drm/nouveau: replace nv04_graph_fifo_access() use with direct reg bashing
  drm/nv40/gr: rewrite/split context takedown functions
  drm/nouveau: detect disabled device in irq handler and return IRQ_NONE
  drm/nouveau: ignore connector type when deciding digital/analog on DVI-I
  drm/nouveau: Add a quirk for Gigabyte NX86T
  drm/nouveau: do not leak in nv20_graph_create
  drm/nv50/dp: fix hack to work for macbooks booted via EFI

12 years agodrm/nouveau/gr: disable fifo access and idle before suspend ctx unload
Ben Skeggs [Wed, 20 Jul 2011 01:44:52 +0000 (11:44 +1000)]
drm/nouveau/gr: disable fifo access and idle before suspend ctx unload

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/nouveau: pass flag to engine fini() method on suspend
Ben Skeggs [Wed, 20 Jul 2011 01:22:33 +0000 (11:22 +1000)]
drm/nouveau: pass flag to engine fini() method on suspend

It may not be necessary to fail in certain cases (such as failing to idle)
on module unload, whereas on suspend it's important to ensure a consistent
state can be restored on resume.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/nouveau: replace nv04_graph_fifo_access() use with direct reg bashing
Ben Skeggs [Wed, 20 Jul 2011 01:04:46 +0000 (11:04 +1000)]
drm/nouveau: replace nv04_graph_fifo_access() use with direct reg bashing

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/nv40/gr: rewrite/split context takedown functions
Ben Skeggs [Sun, 17 Jul 2011 23:56:36 +0000 (09:56 +1000)]
drm/nv40/gr: rewrite/split context takedown functions

It's completely pointless to save the PGRAPH context when destroying a
channel, so don't bother.

This commit should also fix kernel.org bug 39422, where the DRM channel
state was incorrectly being saved because we left PGRAPH FIFO access
enabled while running the ctxprog.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/nouveau: detect disabled device in irq handler and return IRQ_NONE
Ben Skeggs [Tue, 12 Jul 2011 05:42:45 +0000 (15:42 +1000)]
drm/nouveau: detect disabled device in irq handler and return IRQ_NONE

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/nouveau: ignore connector type when deciding digital/analog on DVI-I
Ben Skeggs [Tue, 12 Jul 2011 03:32:07 +0000 (13:32 +1000)]
drm/nouveau: ignore connector type when deciding digital/analog on DVI-I

If the connector table is lying, which it often does on the boards of a
particular manufacturer, we may end up doing the wrong thing.  Listen
to the encoder table instead, it's more reliable.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/nouveau: Add a quirk for Gigabyte NX86T
Emil Velikov [Sun, 10 Jul 2011 09:59:29 +0000 (10:59 +0100)]
drm/nouveau: Add a quirk for Gigabyte NX86T

The connector table lies, the card has DVI-I not HDMI
Fixes bug https://bugs.freedesktop.org/show_bug.cgi?id=35675

v2: Mention the bugreport
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/nouveau: do not leak in nv20_graph_create
Jesper Juhl [Sun, 3 Jul 2011 23:46:53 +0000 (09:46 +1000)]
drm/nouveau: do not leak in nv20_graph_create

If we return due to an unknown chipset in
drivers/gpu/drm/nouveau/nv20_graph.c:nv20_graph_create() we'll leak the
memory allocated to 'pgraph'.

This patch should fix the leak.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Reviewed-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/nv50/dp: fix hack to work for macbooks booted via EFI
Ben Skeggs [Wed, 29 Jun 2011 23:03:36 +0000 (09:03 +1000)]
drm/nv50/dp: fix hack to work for macbooks booted via EFI

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
12 years agodrm/radeon/kms: add info query for backend map
Alex Deucher [Fri, 15 Jul 2011 19:53:52 +0000 (19:53 +0000)]
drm/radeon/kms: add info query for backend map

The 3D driver need to get the pipe to backend
map to certain things.  Add a query to get the
info.

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm: platform multi-device support
Rob Clark [Thu, 14 Jul 2011 03:12:43 +0000 (03:12 +0000)]
drm: platform multi-device support

Include the device id in the bus-id to give userspace a way to open
the correct "cardN" when there are multiple device instances.

Signed-off-by: Rob Clark <rob@ti.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon: Don't clobber error return value in page flipping cleanup paths.
Michel Dänzer [Wed, 13 Jul 2011 15:18:10 +0000 (15:18 +0000)]
drm/radeon: Don't clobber error return value in page flipping cleanup paths.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/radeon: Don't generate new fence for page flip.
Michel Dänzer [Wed, 13 Jul 2011 15:18:09 +0000 (15:18 +0000)]
drm/radeon: Don't generate new fence for page flip.

Use the fence of the new frontbuffer, if any.

Generating a new fence could cause us to wait for completely unrelated
rendering to finish before performing the flip.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agoMerge branch 'drm-intel-next' of ssh://master.kernel.org/pub/scm/linux/kernel/git...
Dave Airlie [Thu, 14 Jul 2011 05:45:23 +0000 (06:45 +0100)]
Merge branch 'drm-intel-next' of ssh:///linux/kernel/git/keithp/linux-2.6 into drm-core-next

* 'drm-intel-next' of ssh://master.kernel.org/pub/scm/linux/kernel/git/keithp/linux-2.6: (52 commits)
  drm/i915: provide module parameter description
  drm/i915: add module parameter compiler hints
  drm/i915/bios: Avoid temporary allocation whilst searching for downclock
  drm/i915: Cache GT fifo count for SandyBridge
  i915: Fix opregion notifications
  drm/i915: TVDAC_STATE_CHG does not indicate successful load-detect
  drm/i915: Select correct pipe during TV detect
  drm/i915/ringbuffer: Idling requires waiting for the ring to be empty
  Revert "drm/i915: enable rc6 by default"
  drm/i915: Clean up i915_driver_load failure path
  drm/i915: Enable i915 frame buffer compression by default
  drm/i915: Share the common work of disabling active FBC before updating
  drm/i915: Perform intel_enable_fbc() from a delayed task
  drm/i915: Disable FBC across page-flipping
  drm/i915: Set persistent-mode for ILK/SNB framebuffer compression
  drm/i915: Use of a CPU fence is mandatory to update FBC regions upon CPU writes
  drm/i915: Remove vestigial pitch from post-gen2 FBC control routines
  drm/i915: Replace direct calls to vfunc.disable_fbc with intel_disable_fbc()
  drm/i915: Only export the generic intel_disable_fbc() interface
  drm/i915: Enable GPU reset on Ivybridge.
  ...

12 years agodrm/i915: provide module parameter description
Ben Widawsky [Wed, 13 Jul 2011 21:38:18 +0000 (14:38 -0700)]
drm/i915: provide module parameter description

Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: add module parameter compiler hints
Ben Widawsky [Wed, 13 Jul 2011 21:38:17 +0000 (14:38 -0700)]
drm/i915: add module parameter compiler hints

Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915/bios: Avoid temporary allocation whilst searching for downclock
Chris Wilson [Wed, 13 Jul 2011 20:01:46 +0000 (21:01 +0100)]
drm/i915/bios: Avoid temporary allocation whilst searching for downclock

Alan Cox reported a missing check on the kmalloc return value for the
allocation of a temporary mode used for searching for the LVDS downlock
frequency. This allocation is roughly 200 bytes, a little too large to
friviously place on the stack. However, we can simply use the few bytes
we need stored within the original DVO timing data, skip the translation
and do the compare directly between the timing data rather than on a
mode, thus avoiding the need for any temporary allocations.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Cache GT fifo count for SandyBridge
Chris Wilson [Thu, 12 May 2011 21:17:09 +0000 (22:17 +0100)]
drm/i915: Cache GT fifo count for SandyBridge

The read back of the available FIFO entries is vital for system
stability, but extremely costly. However, we only need a guide so as to
avoid eating into the reserved entries and since we are the only
consumer we can cache the read of the count from the last write.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agoi915: Fix opregion notifications
Matthew Garrett [Tue, 12 Jul 2011 22:30:52 +0000 (18:30 -0400)]
i915: Fix opregion notifications

opregion-based platforms will send ACPI video event 0x80 for a range of
notification types for legacy compatibility. This is interpreted as a
display switch event, which may not be appropriate in the circumstances.
When we receive such an event we should make sure that the platform is
genuinely requesting a display switch before passing that event through
to userspace.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Tested-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: TVDAC_STATE_CHG does not indicate successful load-detect
Keith Packard [Fri, 13 May 2011 00:10:58 +0000 (17:10 -0700)]
drm/i915: TVDAC_STATE_CHG does not indicate successful load-detect

Do not use this bit to indicate that load detection has completed,
instead just wait for vblank, at which point the load registers will
have been updated.

Signed-off-by: Keith Packard <keithp@keithp.com>
Tested-by: Yi Sun <yi.sun@intel.com>
12 years agodrm/i915: Select correct pipe during TV detect
Keith Packard [Fri, 13 May 2011 00:10:57 +0000 (17:10 -0700)]
drm/i915: Select correct pipe during TV detect

Signed-off-by: Keith Packard <keithp@keithp.com>
Tested-by: Yi Sun <yi.sun@intel.com>
12 years agoMerge 3.0-rc7 into drm-core-next
Dave Airlie [Wed, 13 Jul 2011 07:30:11 +0000 (08:30 +0100)]
Merge 3.0-rc7 into drm-core-next

This pulls in all the drm fixes up to this point which are needed
for some -next patches to work.

12 years agoDRM: remove drm_pci_device_is_pcie
Jon Mason [Mon, 27 Jun 2011 16:07:50 +0000 (16:07 +0000)]
DRM: remove drm_pci_device_is_pcie

drm_pci_device_is_pcie duplicates the funcationality of pci_is_pcie.
Convert callers of the former to the latter.  This has the side benefit
of removing an unnecessary search in the PCI configuration space due to
using a saved PCIe capability offset.

[airlied: update for new callsite]

Signed-off-by: Jon Mason <jdmason@kudzu.us>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agoMerge branch 'drm-intel-fixes' into drm-intel-next
Keith Packard [Tue, 12 Jul 2011 17:40:25 +0000 (10:40 -0700)]
Merge branch 'drm-intel-fixes' into drm-intel-next

12 years agodrm/i915/ringbuffer: Idling requires waiting for the ring to be empty
Chris Wilson [Tue, 12 Jul 2011 17:03:29 +0000 (18:03 +0100)]
drm/i915/ringbuffer: Idling requires waiting for the ring to be empty

...which is measured by the size and not the amount of space remaining.

Waiting upon size-8, did one of two things. In the common case with more
than 8 bytes available to write into the ring, it would return
immediately. Otherwise, it would timeout given the impossible condition
of waiting for more space than is available in the ring, leading to
warnings such as:

[drm:intel_cleanup_ring_buffer] *ERROR* failed to quiesce render ring
whilst cleaning up: -16

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agoRevert "drm/i915: enable rc6 by default"
Keith Packard [Tue, 12 Jul 2011 15:49:31 +0000 (08:49 -0700)]
Revert "drm/i915: enable rc6 by default"

This reverts commit a51f7a66fb5e4af5ec4286baef940d06594b59d2.

We still have a few Ironlake and Sandybridge machines which fail when
RC6 is enabled. Better luck next release?

Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Clean up i915_driver_load failure path
Keith Packard [Sun, 10 Jul 2011 20:12:17 +0000 (13:12 -0700)]
drm/i915: Clean up i915_driver_load failure path

i915_driver_load adds a write-combining MTRR region for the GTT
aperture to improve memory speeds through the aperture. If
i915_driver_load fails after this, it would not have cleaned up the
MTRR. This shouldn't cause any problems, except for consuming an MTRR
register. Still, it's best to clean up completely in the failure path,
which is easily done by calling mtrr_del if the mtrr was successfully
allocated.

i915_driver_load calls i915_gem_load which register
i915_gem_inactive_shrink. If i915_driver_load fails after calling
i915_gem_load, the shrinker will be left registered. When called, it
will access freed memory and crash. The fix is to unregister the shrinker in the
failure path using code duplicated from i915_driver_unload.

i915_driver_load also has some incorrect gotos in the error cleanup
paths:

 * After failing to initialize the GTT (which cannot happen, btw,
   intel_gtt_get returns a fixed (non-NULL) value), it tries to
   free the uninitialized WC IO mapping. Fixed this by changing the
   target from out_iomapfree to out_rmmap

Signed-off-by: Keith Packard <keithp@keithp.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
12 years agoLinux 3.0-rc7 v3.0-rc7
Linus Torvalds [Mon, 11 Jul 2011 23:51:52 +0000 (16:51 -0700)]
Linux 3.0-rc7

12 years agoDocumentation/Changes: remove some really obsolete text
Linus Torvalds [Mon, 11 Jul 2011 23:48:38 +0000 (16:48 -0700)]
Documentation/Changes: remove some really obsolete text

That file harkens back to the days of the big 2.4 -> 2.6 version jump,
and was based even then on older versions.  Some of it is just obsolete,
and Jesper Juhl points out that it talks about kernel versions 2.6 and
should be updated to 3.0.

Remove some obsolete text, and re-phrase some other to not be 2.6-specific.

Reported-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge branch 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Mon, 11 Jul 2011 23:43:27 +0000 (16:43 -0700)]
Merge branch 'v4l_for_linus' of git://git./linux/kernel/git/mchehab/linux-2.6

* 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6:
  [media] msp3400: fill in v4l2_tuner based on vt->type field
  [media] tuner-core.c: don't change type field in g_tuner or g_frequency
  [media] cx18/ivtv: fix g_tuner support
  [media] tuner-core: power up tuner when called with s_power(1)
  [media] v4l2-ioctl.c: check for valid tuner type in S_HW_FREQ_SEEK
  [media] tuner-core: simplify the standard fixup
  [media] tuner-core/v4l2-subdev: document that the type field has to be filled in
  [media] v4l2-subdev.h: remove unused s_mode tuner op
  [media] feature-removal-schedule: change in how radio device nodes are handled
  [media] bttv: fix s_tuner for radio
  [media] pvrusb2: fix g/s_tuner support
  [media] v4l2-ioctl.c: prefill tuner type for g_frequency and g/s_tuner
  [media] tuner-core: fix tuner_resume: use t->mode instead of t->type
  [media] tuner-core: fix s_std and s_tuner

12 years agoMerge branch 'pm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/suspe...
Linus Torvalds [Mon, 11 Jul 2011 19:49:03 +0000 (12:49 -0700)]
Merge branch 'pm-fixes' of git://git./linux/kernel/git/rafael/suspend-2.6

* 'pm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/suspend-2.6:
  PM: Reintroduce dropped call to check_wakeup_irqs

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Mon, 11 Jul 2011 19:48:24 +0000 (12:48 -0700)]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
  cifs: drop spinlock before calling cifs_put_tlink
  cifs: fix expand_dfs_referral
  cifs: move bdi_setup_and_register outside of CONFIG_CIFS_DFS_UPCALL
  cifs: factor smb_vol allocation out of cifs_setup_volume_info
  cifs: have cifs_cleanup_volume_info not take a double pointer
  cifs: fix build_unc_path_to_root to account for a prefixpath
  cifs: remove bogus call to cifs_cleanup_volume_info

12 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq
Linus Torvalds [Mon, 11 Jul 2011 19:47:53 +0000 (12:47 -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 cpumask memory leak in acpi-cpufreq on cpu hotplug.

12 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mjg59/platf...
Linus Torvalds [Mon, 11 Jul 2011 19:47:09 +0000 (12:47 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/mjg59/platform-drivers-x86

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mjg59/platform-drivers-x86:
  hp-wmi: fix use after free
  dell-laptop - using buffer without mutex_lock
  Revert: "dell-laptop: Toggle the unsupported hardware killswitch"
  platform-drivers-x86: set backlight type to BACKLIGHT_PLATFORM
  thinkpad-acpi: handle HKEY 0x4010, 0x4011 events
  drivers/platform/x86: Fix memory leak
  thinkpad-acpi: handle some new HKEY 0x60xx events
  acer-wmi: fix bitwise bug when set device state
  acer-wmi: Only update rfkill status for associated hotkey events

12 years agoMerge branch 'movieboard' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394...
Linus Torvalds [Mon, 11 Jul 2011 19:46:39 +0000 (12:46 -0700)]
Merge branch 'movieboard' of git://git./linux/kernel/git/ieee1394/linux1394-2.6

* 'movieboard' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6:
  firewire: ohci: do not bind to Pinnacle cards, avert panic

12 years agoath5k: Add missing breaks in switch/case
Joe Perches [Sun, 10 Jul 2011 09:28:26 +0000 (02:28 -0700)]
ath5k: Add missing breaks in switch/case

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Pavel Roskin <proski@gnu.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoDocumentation/spinlocks.txt: Remove reference to sti()/cli()
Muthu Kumar [Mon, 11 Jul 2011 18:04:58 +0000 (11:04 -0700)]
Documentation/spinlocks.txt: Remove reference to sti()/cli()

Since we removed sti()/cli() and related, how about removing it from
Documentation/spinlocks.txt?

Signed-off-by: Muthukumar R <muthur@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agocifs: drop spinlock before calling cifs_put_tlink
Jeff Layton [Mon, 11 Jul 2011 14:16:34 +0000 (10:16 -0400)]
cifs: drop spinlock before calling cifs_put_tlink

...as that function can sleep.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
12 years agohp-wmi: fix use after free
Eric Dumazet [Mon, 11 Jul 2011 10:22:21 +0000 (12:22 +0200)]
hp-wmi: fix use after free

[  191.310008] WARNING: kmemcheck: Caught 32-bit read from freed memory (f0d25f14)
[  191.310011] c056d2f088000000105fd2f00000000050415353040000000000000000000000
[  191.310020]  i i i i f f f f f f f f f f f f f f f f f f f f f f f f f f f f
[  191.310027]                                          ^
[  191.310029]
[  191.310032] Pid: 737, comm: modprobe Not tainted 3.0.0-rc5+ #268 Hewlett-Packard HP Compaq 6005 Pro SFF PC/3047h
[  191.310036] EIP: 0060:[<f80b3104>] EFLAGS: 00010286 CPU: 0
[  191.310039] EIP is at hp_wmi_perform_query+0x104/0x150 [hp_wmi]
[  191.310041] EAX: f0d25601 EBX: f0d25f00 ECX: 000121cf EDX: 000121ce
[  191.310043] ESI: f0d25f10 EDI: f0f97ea8 EBP: f0f97ec4 ESP: c173f34c
[  191.310045]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
[  191.310046] CR0: 8005003b CR2: f540c000 CR3: 30f30000 CR4: 000006d0
[  191.310048] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
[  191.310050] DR6: ffff4ff0 DR7: 00000400
[  191.310051]  [<f80b317b>] hp_wmi_dock_state+0x2b/0x40 [hp_wmi]
[  191.310054]  [<f80b6093>] hp_wmi_init+0x93/0x1a8 [hp_wmi]
[  191.310057]  [<c10011f0>] do_one_initcall+0x30/0x170
[  191.310061]  [<c107ab9f>] sys_init_module+0xef/0x1a60
[  191.310064]  [<c149f998>] sysenter_do_call+0x12/0x28
[  191.310067]  [<ffffffff>] 0xffffffff

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
12 years agodell-laptop - using buffer without mutex_lock
Jose Alonso [Sun, 10 Jul 2011 18:46:51 +0000 (15:46 -0300)]
dell-laptop - using buffer without mutex_lock

Using buffer->output[1] without mutex_lock()

Signed-off-by: Jose Alonso <joalonsof@gmail.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
12 years agoRevert: "dell-laptop: Toggle the unsupported hardware killswitch"
Keng-Yu Lin [Mon, 27 Jun 2011 10:19:03 +0000 (11:19 +0100)]
Revert: "dell-laptop: Toggle the unsupported hardware killswitch"

This reverts commit a3d77411e8b2ad661958c1fbee65beb476ec6d70,

as it causes a mess in the wireless rfkill status on some models.
It is probably a bad idea to toggle the rfkill for all dell models
without the respect to the claim that it is hardware-controlled.

Cc: stable@kernel.org
Signed-off-by: Keng-Yu Lin <kengyu@canonical.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
12 years agoPM: Reintroduce dropped call to check_wakeup_irqs
Colin Cross [Mon, 11 Jul 2011 08:51:49 +0000 (10:51 +0200)]
PM: Reintroduce dropped call to check_wakeup_irqs

Patch 2e711c04dbbf7a7732a3f7073b1fc285d12b369d
(PM: Remove sysdev suspend, resume and shutdown operations)
deleted sysdev_suspend(), which was being relied on to call
check_wakeup_irqs() in suspend.  If check_wakeup_irqs() is not
called, wake interrupts that are pending when suspend is
entered may be lost.  It also breaks IRQCHIP_MASK_ON_SUSPEND,
which is handled in check_wakeup_irqs().

This patch adds a call to check_wakeup_irqs() in syscore_suspend(),
similar to what was deleted in sysdev_suspend().

Signed-off-by: Colin Cross <ccross@android.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
12 years ago[CPUFREQ] fix cpumask memory leak in acpi-cpufreq on cpu hotplug.
Luming Yu [Fri, 8 Jul 2011 20:37:44 +0000 (16:37 -0400)]
[CPUFREQ] fix cpumask memory leak in acpi-cpufreq on cpu hotplug.

I came across a memory leak during a cyclic cpu-online-offline test.

Signed-off-by: Yu Luming <luming.yu@intel.com>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Dave Jones <davej@redhat.com>
12 years agoMerge branch 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groec...
Linus Torvalds [Sun, 10 Jul 2011 17:24:47 +0000 (10:24 -0700)]
Merge branch 'hwmon-for-linus' of git://git./linux/kernel/git/groeck/staging

* 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/staging:
  hwmon: (pmbus) Improve auto-detection of temperature status register
  hwmon: (lm95241) Fix negative temperature results
  hwmon: (lm95241) Fix chip detection code

12 years agohwmon: (pmbus) Improve auto-detection of temperature status register
Guenter Roeck [Sun, 3 Jul 2011 20:08:03 +0000 (13:08 -0700)]
hwmon: (pmbus) Improve auto-detection of temperature status register

It is possible that a PMBus device supports the READ_TEMPERATURE2 and/or
READ_TEMPERATURE3 registers but does not support READ_TEMPERATURE1.
Improve temperature status register detection to address this condition.

Reported-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Cc: stable@kernel.org # 2.6.39+
12 years agohwmon: (lm95241) Fix negative temperature results
Guenter Roeck [Thu, 30 Jun 2011 09:09:37 +0000 (02:09 -0700)]
hwmon: (lm95241) Fix negative temperature results

Negative temperatures were returned in degrees C instead of milli-Degrees C.
Also, negative temperatures were reported for remote temperature sensors even
if the chip was configured for positive-only results.

Fix by detecting temperature modes, and by treating negative temperatures
similar to positive temperatures, with appropriate sign extension.

Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Cc: stable@kernel.org # 2.6.30+
12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Sun, 10 Jul 2011 14:29:22 +0000 (07:29 -0700)]
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 - Fix a copmile warning
  ASoC: ak4642: fixup snd_soc_update_bits mask for PW_MGMT2
  ALSA: hda - Change all ADCs for dual-adc switching mode for Realtek
  ASoC: Manage WM8731 ACTIVE bit as a supply widget
  ASoC: Don't set invalid name string to snd_card->driver field
  ASoC: Ensure we delay long enough for WM8994 FLL to lock when starting
  ASoC: Tegra: I2S: Ensure clock is enabled when writing regs
  ASoC: Fix Blackfin I2S _pointer() implementation return in bounds values
  ASoC: tlv320aic3x: Do soft reset to codec when going to bias off state
  ASoC: tlv320aic3x: Don't sync first two registers from register cache
  audio: tlv320aic26: fix PLL register configuration

12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes...
Linus Torvalds [Sun, 10 Jul 2011 14:28:51 +0000 (07:28 -0700)]
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: conditional resource-reallocation through kernel parameter pci=realloc

12 years agoMerge branch 'fixes' of master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Sun, 10 Jul 2011 14:28:30 +0000 (07:28 -0700)]
Merge branch 'fixes' of /home/rmk/linux-2.6-arm

* 'fixes' of master.kernel.org:/home/rmk/linux-2.6-arm:
  ARM: 6994/1: smp_twd: Fix typo in 'twd_timer_rate' printing
  ARM: 6987/1: l2x0: fix disabling function to avoid deadlock
  ARM: 6966/1: ep93xx: fix inverted RTS/DTR signals on uart1
  ARM: 6980/1: mmci: use StartBitErr to detect bad connections
  ARM: 6979/1: mach-vt8500: add forgotten irq_data conversion
  ARM: move memory layout sanity checking before meminfo initialization
  ARM: 6990/1: MAINTAINERS: add entry for ARM PMU profiling and debugging
  ARM: 6989/1: perf: do not start the PMU when no events are present
  ARM: dmabounce: fix map_single() error return value

12 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Sun, 10 Jul 2011 14:27:45 +0000 (07:27 -0700)]
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/kms: clean up multiple crtc handling for evergreen+ (v2)

12 years agofirewire: ohci: do not bind to Pinnacle cards, avert panic
Stefan Richter [Sat, 9 Jul 2011 22:23:03 +0000 (00:23 +0200)]
firewire: ohci: do not bind to Pinnacle cards, avert panic

When firewire-ohci is bound to a Pinnacle MovieBoard, eventually a
"Register access failure" is logged and an interrupt storm or a kernel
panic happens.  https://bugzilla.kernel.org/show_bug.cgi?id=36622

Until this is sorted out (if that is going to succeed at all), let's
just prevent firewire-ohci from touching these devices.

Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Cc: <stable@kernel.org>
12 years agocifs: fix expand_dfs_referral
Jeff Layton [Wed, 6 Jul 2011 12:10:39 +0000 (08:10 -0400)]
cifs: fix expand_dfs_referral

Regression introduced in commit 724d9f1cfba.

Prior to that, expand_dfs_referral would regenerate the mount data string
and then call cifs_parse_mount_options to re-parse it (klunky, but it
worked). The above commit moved cifs_parse_mount_options out of cifs_mount,
so the re-parsing of the new mount options no longer occurred. Fix it by
making expand_dfs_referral re-parse the mount options.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
12 years agocifs: move bdi_setup_and_register outside of CONFIG_CIFS_DFS_UPCALL
Jeff Layton [Sat, 9 Jul 2011 16:21:07 +0000 (12:21 -0400)]
cifs: move bdi_setup_and_register outside of CONFIG_CIFS_DFS_UPCALL

This needs to be done regardless of whether that KConfig option is set
or not.

Reported-by: Sven-Haegar Koch <haegar@sdinet.de>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
12 years agoMerge branch 'fix/asoc' into for-linus
Takashi Iwai [Sat, 9 Jul 2011 09:56:43 +0000 (11:56 +0200)]
Merge branch 'fix/asoc' into for-linus

12 years agoALSA: hda - Fix a copmile warning
Takashi Iwai [Sat, 9 Jul 2011 09:55:28 +0000 (11:55 +0200)]
ALSA: hda - Fix a copmile warning

It's harmless but annyoing.
  sound/pci/hda/patch_realtek.c: In function ‘alc_cap_getput_caller’:
  sound/pci/hda/patch_realtek.c:2722:9: warning: ‘err’ may be used uninitialized in this function

Signed-off-by: Takashi Iwai <tiwai@suse.de>
12 years agoMerge branch 'for-3.0' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound...
Takashi Iwai [Sat, 9 Jul 2011 07:44:09 +0000 (09:44 +0200)]
Merge branch 'for-3.0' of git://git./linux/kernel/git/broonie/sound-2.6 into fix/asoc

12 years agoMerge branch 's5p-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 9 Jul 2011 06:28:23 +0000 (23:28 -0700)]
Merge branch 's5p-fixes-for-linus' of git://git./linux/kernel/git/kgene/linux-samsung

* 's5p-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung:
  ARM: S3C2440: fix section mismatch on mini2440
  ARM: S3C24XX: drop return codes in void function of dma.c
  ARM: S3C24XX: don't use uninitialized variable in dma.c
  ARM: EXYNOS4: Set appropriate I2C device variant
  ARM: S5PC100: Fix for compilation error
  spi/s3c64xx: Bug fix for SPI with different FIFO level
  ARM: SAMSUNG: Add tx_st_done variable
  ARM: EXYNOS4: Address a section mismatch w/ suspend issue.
  ARM: S5P: Fix bug on init of PWMTimers for HRTimer
  ARM: SAMSUNG: header file revised to prevent declaring duplicated
  ARM: EXYNOS4: fix improper gpio configuration
  ARM: EXYNOS4: Fix card detection for sdhci 0 and 2

12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lrg/voltage-2.6
Linus Torvalds [Sat, 9 Jul 2011 06:26:28 +0000 (23:26 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/lrg/voltage-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lrg/voltage-2.6:
  regulator: max8997: Fix setting inappropriate value for ramp_delay variable
  regulator: db8500-prcmu: small fixes
  regulator: max8997: remove dependency on platform_data pointer
  regulator: MAX8997: Fix for divide by zero error
  regulator: max8952 - fix wrong gpio valid check

12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs...
Linus Torvalds [Sat, 9 Jul 2011 06:25:45 +0000 (23:25 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mason/btrfs-unstable

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
  btrfs: fix oops when doing space balance
  Btrfs: don't panic if we get an error while balancing V2
  btrfs: add missing options displayed in mount output

12 years agoMAINTAINERS: update Bjorn Helgaas's email address
Bjorn Helgaas [Fri, 8 Jul 2011 22:39:48 +0000 (15:39 -0700)]
MAINTAINERS: update Bjorn Helgaas's email address

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agodrivers/leds/leds-pca9532.c: change driver name to be unique
Wolfram Sang [Fri, 8 Jul 2011 22:39:46 +0000 (15:39 -0700)]
drivers/leds/leds-pca9532.c: change driver name to be unique

This driver handles the variants pca9530-pca9533, so it chose the name
"pca953x".  However, there is a gpio driver which decided on the same
name.  As a result, those two can't be loaded at the same time.  Add a
subsystem prefix to make the driver name unique.  Device matching will not
suffer, because both are I2C drivers which match using a
i2c_device_id-table which is not altered.

Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Acked-by: Jean Delvare <khali@linux-fr.org>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/nommu.c: fix remap_pfn_range()
Bob Liu [Fri, 8 Jul 2011 22:39:46 +0000 (15:39 -0700)]
mm/nommu.c: fix remap_pfn_range()

remap_pfn_range() means map physical address pfn<<PAGE_SHIFT to user addr.

For nommu arch it's implemented by vma->vm_start = pfn << PAGE_SHIFT which
is wrong acroding the original meaning of this function.  And some driver
developer using remap_pfn_range() with correct parameter will get
unexpected result because vm_start is changed.  It should be implementd
like addr = pfn << PAGE_SHIFT but which is meanless on nommu arch, this
patch just make it simply return.

Parameter name and setting of vma->vm_flags also be fixed.

Signed-off-by: Bob Liu <lliubbo@gmail.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: David Howells <dhowells@redhat.com>
Acked-by: Greg Ungerer <gerg@uclinux.org>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: Bob Liu <lliubbo@gmail.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agow1: ds1wm: add a reset recovery parameter
Jean-François Dagenais [Fri, 8 Jul 2011 22:39:44 +0000 (15:39 -0700)]
w1: ds1wm: add a reset recovery parameter

This fixes a regression in 3.0 reported by Paul Parsons regarding the
removal of the msleep(1) in the ds1wm_reset() function:

: The linux-3.0-rc4 DS1WM 1-wire driver is logging "bus error, retrying"
: error messages on an HP iPAQ hx4700 PDA (XScale-PXA270):
:
: <snip>
: Driver for 1-wire Dallas network protocol.
: DS1WM w1 busmaster driver - (c) 2004 Szabolcs Gyurko
: 1-Wire driver for the DS2760 battery monitor  chip  - (c) 2004-2005, Szabolcs Gyurko
: ds1wm ds1wm: pass: 1 bus error, retrying
: ds1wm ds1wm: pass: 2 bus error, retrying
: ds1wm ds1wm: pass: 3 bus error, retrying
: ds1wm ds1wm: pass: 4 bus error, retrying
: ds1wm ds1wm: pass: 5 bus error, retrying
: ...
:
: The visible result is that the battery charging LED is erratic; sometimes
: it works, mostly it doesn't.
:
: The linux-2.6.39 DS1WM 1-wire driver worked OK.  I haven't tried 3.0-rc1,
: 3.0-rc2, or 3.0-rc3.

This sleep should not be required on normal circuitry provided the
pull-ups on the bus are correctly adapted to the slaves.  Unfortunately,
this is not always the case.  The sleep is restored but as a parameter to
the probe function in the pdata.

[akpm@linux-foundation.org: coding-style fixes]
Reported-by: Paul Parsons <lost.distance@yahoo.com>
Tested-by: Paul Parsons <lost.distance@yahoo.com>
Signed-off-by: Jean-François Dagenais <dagenaisj@sonatest.com>
Cc: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: fix numa scan information update to be triggered by memory event
KAMEZAWA Hiroyuki [Fri, 8 Jul 2011 22:39:43 +0000 (15:39 -0700)]
memcg: fix numa scan information update to be triggered by memory event

commit 889976dbcb12 ("memcg: reclaim memory from nodes in round-robin
order") adds an numa node round-robin for memcg.  But the information is
updated once per 10sec.

This patch changes the update trigger from jiffies to memcg's event count.
 After this patch, numa scan information will be updated when we see 1024
events of pagein/pageout under a memcg.

[akpm@linux-foundation.org: attempt to repair code layout]
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Ying Han <yinghan@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: fix reclaimable lru check in memcg
KAMEZAWA Hiroyuki [Fri, 8 Jul 2011 22:39:42 +0000 (15:39 -0700)]
memcg: fix reclaimable lru check in memcg

Now, in mem_cgroup_hierarchical_reclaim(), mem_cgroup_local_usage() is
used for checking whether the memcg contains reclaimable pages or not.  If
no pages in it, the routine skips it.

But, mem_cgroup_local_usage() contains Unevictable pages and cannot handle
"noswap" condition correctly.  This doesn't work on a swapless system.

This patch adds test_mem_cgroup_reclaimable() and replaces
mem_cgroup_local_usage().  test_mem_cgroup_reclaimable() see LRU counter
and returns correct answer to the caller.  And this new function has
"noswap" argument and can see only FILE LRU if necessary.

[akpm@linux-foundation.org: coding-style fixes]
[akpm@linux-foundation.org: fix kerneldoc layout]
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Ying Han <yinghan@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: __tlb_remove_page() check the correct batch
Shaohua Li [Fri, 8 Jul 2011 22:39:41 +0000 (15:39 -0700)]
mm: __tlb_remove_page() check the correct batch

__tlb_remove_page() switches to a new batch page, but still checks space
in the old batch.  This check always fails, and causes a forced tlb flush.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: vmscan: only read new_classzone_idx from pgdat when reclaiming successfully
Mel Gorman [Fri, 8 Jul 2011 22:39:40 +0000 (15:39 -0700)]
mm: vmscan: only read new_classzone_idx from pgdat when reclaiming successfully

During allocator-intensive workloads, kswapd will be woken frequently
causing free memory to oscillate between the high and min watermark.  This
is expected behaviour.  Unfortunately, if the highest zone is small, a
problem occurs.

When balance_pgdat() returns, it may be at a lower classzone_idx than it
started because the highest zone was unreclaimable.  Before checking if it
should go to sleep though, it checks pgdat->classzone_idx which when there
is no other activity will be MAX_NR_ZONES-1.  It interprets this as it has
been woken up while reclaiming, skips scheduling and reclaims again.  As
there is no useful reclaim work to do, it enters into a loop of shrinking
slab consuming loads of CPU until the highest zone becomes reclaimable for
a long period of time.

There are two problems here.  1) If the returned classzone or order is
lower, it'll continue reclaiming without scheduling.  2) if the highest
zone was marked unreclaimable but balance_pgdat() returns immediately at
DEF_PRIORITY, the new lower classzone is not communicated back to kswapd()
for sleeping.

This patch does two things that are related.  If the end_zone is
unreclaimable, this information is communicated back.  Second, if the
classzone or order was reduced due to failing to reclaim, new information
is not read from pgdat and instead an attempt is made to go to sleep.  Due
to this, it is also necessary that pgdat->classzone_idx be initialised
each time to pgdat->nr_zones - 1 to avoid re-reads being interpreted as
wakeups.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reported-by: Pádraig Brady <P@draigBrady.com>
Tested-by: Pádraig Brady <P@draigBrady.com>
Tested-by: Andrew Lutomirski <luto@mit.edu>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: vmscan: evaluate the watermarks against the correct classzone
Mel Gorman [Fri, 8 Jul 2011 22:39:39 +0000 (15:39 -0700)]
mm: vmscan: evaluate the watermarks against the correct classzone

When deciding if kswapd is sleeping prematurely, the classzone is taken
into account but this is different to what balance_pgdat() and the
allocator are doing.  Specifically, the DMA zone will be checked based on
the classzone used when waking kswapd which could be for a GFP_KERNEL or
GFP_HIGHMEM request.  The lowmem reserve limit kicks in, the watermark is
not met and kswapd thinks it's sleeping prematurely keeping kswapd awake in
error.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reported-by: Pádraig Brady <P@draigBrady.com>
Tested-by: Pádraig Brady <P@draigBrady.com>
Tested-by: Andrew Lutomirski <luto@mit.edu>
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: vmscan: do not apply pressure to slab if we are not applying pressure to zone
Mel Gorman [Fri, 8 Jul 2011 22:39:38 +0000 (15:39 -0700)]
mm: vmscan: do not apply pressure to slab if we are not applying pressure to zone

During allocator-intensive workloads, kswapd will be woken frequently
causing free memory to oscillate between the high and min watermark.  This
is expected behaviour.

When kswapd applies pressure to zones during node balancing, it checks if
the zone is above a high+balance_gap threshold.  If it is, it does not
apply pressure but it unconditionally shrinks slab on a global basis which
is excessive.  In the event kswapd is being kept awake due to a high small
unreclaimable zone, it skips zone shrinking but still calls shrink_slab().

Once pressure has been applied, the check for zone being unreclaimable is
being made before the check is made if all_unreclaimable should be set.
This miss of unreclaimable can cause has_under_min_watermark_zone to be
set due to an unreclaimable zone preventing kswapd backing off on
congestion_wait().

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reported-by: Pádraig Brady <P@draigBrady.com>
Tested-by: Pádraig Brady <P@draigBrady.com>
Tested-by: Andrew Lutomirski <luto@mit.edu>
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: vmscan: correct check for kswapd sleeping in sleeping_prematurely
Mel Gorman [Fri, 8 Jul 2011 22:39:36 +0000 (15:39 -0700)]
mm: vmscan: correct check for kswapd sleeping in sleeping_prematurely

During allocator-intensive workloads, kswapd will be woken frequently
causing free memory to oscillate between the high and min watermark.  This
is expected behaviour.  Unfortunately, if the highest zone is small, a
problem occurs.

This seems to happen most with recent sandybridge laptops but it's
probably a co-incidence as some of these laptops just happen to have a
small Normal zone.  The reproduction case is almost always during copying
large files that kswapd pegs at 100% CPU until the file is deleted or
cache is dropped.

The problem is mostly down to sleeping_prematurely() keeping kswapd awake
when the highest zone is small and unreclaimable and compounded by the
fact we shrink slabs even when not shrinking zones causing a lot of time
to be spent in shrinkers and a lot of memory to be reclaimed.

Patch 1 corrects sleeping_prematurely to check the zones matching
the classzone_idx instead of all zones.

Patch 2 avoids shrinking slab when we are not shrinking a zone.

Patch 3 notes that sleeping_prematurely is checking lower zones against
a high classzone which is not what allocators or balance_pgdat()
is doing leading to an artifical belief that kswapd should be
still awake.

Patch 4 notes that when balance_pgdat() gives up on a high zone that the
decision is not communicated to sleeping_prematurely()

This problem affects 2.6.38.8 for certain and is expected to affect 2.6.39
and 3.0-rc4 as well.  If accepted, they need to go to -stable to be picked
up by distros and this series is against 3.0-rc4.  I've cc'd people that
reported similar problems recently to see if they still suffer from the
problem and if this fixes it.

This patch: correct the check for kswapd sleeping in sleeping_prematurely()

During allocator-intensive workloads, kswapd will be woken frequently
causing free memory to oscillate between the high and min watermark.  This
is expected behaviour.

A problem occurs if the highest zone is small.  balance_pgdat() only
considers unreclaimable zones when priority is DEF_PRIORITY but
sleeping_prematurely considers all zones.  It's possible for this sequence
to occur

  1. kswapd wakes up and enters balance_pgdat()
  2. At DEF_PRIORITY, marks highest zone unreclaimable
  3. At DEF_PRIORITY-1, ignores highest zone setting end_zone
  4. At DEF_PRIORITY-1, calls shrink_slab freeing memory from
        highest zone, clearing all_unreclaimable. Highest zone
        is still unbalanced
  5. kswapd returns and calls sleeping_prematurely
  6. sleeping_prematurely looks at *all* zones, not just the ones
     being considered by balance_pgdat. The highest small zone
     has all_unreclaimable cleared but the zone is not
     balanced. all_zones_ok is false so kswapd stays awake

This patch corrects the behaviour of sleeping_prematurely to check the
zones balance_pgdat() checked.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reported-by: Pádraig Brady <P@draigBrady.com>
Tested-by: Pádraig Brady <P@draigBrady.com>
Tested-by: Andrew Lutomirski <luto@mit.edu>
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agohwmon: (lm95241) Fix chip detection code
Guenter Roeck [Mon, 27 Jun 2011 18:22:46 +0000 (11:22 -0700)]
hwmon: (lm95241) Fix chip detection code

The LM95241 driver accepts every chip ID equal to or larger than 0xA4 as its
own, and other chips such as LM95245 use chip IDs in the accepted ID range.
This results in false chip detection.

Fix problem by accepting only the known LM95241 chip ID.

Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Cc: stable@kernel.org # 2.6.30+
12 years agoPCI: conditional resource-reallocation through kernel parameter pci=realloc
Ram Pai [Thu, 7 Jul 2011 18:19:10 +0000 (11:19 -0700)]
PCI: conditional resource-reallocation through kernel parameter pci=realloc

Multiple attempts to dynamically reallocate pci resources have
unfortunately lead to regressions. Though we continue to fix the
regressions and fine tune the dynamic-reallocation behavior, we have not
reached a acceptable state yet.

This patch provides a interim solution. It disables dynamic reallocation
by default, but adds the ability to enable it through pci=realloc kernel
command line parameter.

Tested-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Ram Pai <linuxram@us.ibm.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
12 years agoregulator: max8997: Fix setting inappropriate value for ramp_delay variable
Donggeun Kim [Thu, 7 Jul 2011 07:51:56 +0000 (16:51 +0900)]
regulator: max8997: Fix setting inappropriate value for ramp_delay variable

The ramp_delay variable can be set lower than the desired value.
This patch fixes it.

Signed-off-by: Donggeun Kim <dg77.kim@samsung.com>
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Signed-off-by: KyungMin Park <kyungmin.park@samsung.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
12 years agoregulator: db8500-prcmu: small fixes
Axel Lin [Wed, 6 Jul 2011 03:41:12 +0000 (11:41 +0800)]
regulator: db8500-prcmu: small fixes

Small cleanups for better readability.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
12 years agoregulator: max8997: remove dependency on platform_data pointer
MyungJoo Ham [Mon, 20 Jun 2011 08:30:17 +0000 (17:30 +0900)]
regulator: max8997: remove dependency on platform_data pointer

The platform_data (pdata) may be pointing to __initdata section, which
may be free'd from the memory. The dependency on pdata in non-init
functions is removed in this patch to allow platform to declare
__initdata for platform data.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
12 years agoregulator: MAX8997: Fix for divide by zero error
Tushar Behera [Tue, 21 Jun 2011 03:38:59 +0000 (09:08 +0530)]
regulator: MAX8997: Fix for divide by zero error

Currently, ramp_delay variable is used uninitialzed in
max8997_set_voltage_ldobuck which gets called through
regulator_register calls.

To fix the problem, in max8997_pmic_probe, ramp_delay initialization
code is moved before calls to regulator_register.

Cc: Liam Girdwood <lrg@ti.com>
Cc: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: MyungJoo Ham <myungjoo.ham@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Samuel Ortiz <sameo@linux.intel.com>
Signed-off-by: Tushar Behera <tushar.behera@linaro.org>
Acked-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
12 years agoregulator: max8952 - fix wrong gpio valid check
Jonghwan Choi [Wed, 15 Jun 2011 08:54:12 +0000 (17:54 +0900)]
regulator: max8952 - fix wrong gpio valid check

Signed-off-by: Jonghwan Choi <jhbird.choi@samsung.com>
Acked-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
12 years agodrm/i915: Enable i915 frame buffer compression by default
Keith Packard [Fri, 8 Jul 2011 17:29:42 +0000 (10:29 -0700)]
drm/i915: Enable i915 frame buffer compression by default

We'll try again with the new fixes. Prepare to see this reverted when
we get regression reports...

Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/radeon/kms: clean up multiple crtc handling for evergreen+ (v2)
Alex Deucher [Fri, 8 Jul 2011 15:44:56 +0000 (11:44 -0400)]
drm/radeon/kms: clean up multiple crtc handling for evergreen+ (v2)

evergreen+ asics have 2-6 crtcs.  Don't access crtc registers
for crtc regs that don't exist as they have very high latency
and may cause problems on some asics.  The previous code missed
a few cases and was not fine grained enough (missed the 4 crtc
case for example).

Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=38800

v2: fix typo noticed by Chris Bandy <cbandy@jbandy.com>

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Reviewed-by: Michel Dänzer <michel@daenzer.net>
Tested-by: Simon Farnsworth <simon.farnsworth@onelan.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agodrm/i915: Share the common work of disabling active FBC before updating
Chris Wilson [Fri, 8 Jul 2011 11:22:43 +0000 (12:22 +0100)]
drm/i915: Share the common work of disabling active FBC before updating

Upon review, all path share the same dependencies for updating the
registers and so we can benefit from sharing the code and checking
early.

This removes the unsightly intel_wait_for_vblank() from the lowlevel
functions and upon further analysis the only path that will require a
wait is if we are performing an instantaneous transition between two
valid FBC configurations. The page-flip path itself will have disabled
FBC registers and will have waited for at least one vblank before
finishing the flip and attempting to re-enable FBC. This wait can be
accomplished simply by delaying the enable until after we are sure that
a vblank will have passed, which we are already doing to make sure that
the display is settled before enabling FBC.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Perform intel_enable_fbc() from a delayed task
Chris Wilson [Fri, 8 Jul 2011 11:22:42 +0000 (12:22 +0100)]
drm/i915: Perform intel_enable_fbc() from a delayed task

In order to accommodate the requirements of re-enabling FBC after
page-flipping, but to avoid doing so and incurring the cost of a wait
for vblank in the middle of a page-flip sequence, we defer the actual
enablement by 50ms. If any request to disable FBC arrive within that
interval, the enablement is cancelled and we are saved from blocking on
the wait.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Disable FBC across page-flipping
Chris Wilson [Fri, 8 Jul 2011 11:22:41 +0000 (12:22 +0100)]
drm/i915: Disable FBC across page-flipping

Page-flipping updates the scanout address, nukes the FBC compressed
image and so forces an FBC update so that the displayed image remains
consistent. However, page-flipping does not update the FBC registers
themselves, which remain pointing to both the old address and the old
CPU fence. Future updates to the new front-buffer (scanout) are then
undetected!

This first approach to demonstrate the issue and highlight the fix,
simply disables FBC upon page-flip (a recompression will be forced on
every flip so FBC becomes immaterial) and then re-enables FBC in the
page-flip finish work function, so that the FBC registers are now
pointing to the new framebuffer and front-buffer rendering works once
more.

Ideally, we want to only re-enable FBC after page-flipping is complete,
as otherwise we are just wasting cycles and power (with needless
recompression) whilst the page-flipping application is still running.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=33487
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Set persistent-mode for ILK/SNB framebuffer compression
Chris Wilson [Fri, 8 Jul 2011 11:22:40 +0000 (12:22 +0100)]
drm/i915: Set persistent-mode for ILK/SNB framebuffer compression

Persistent mode is intended for use with front-buffer rendering, such as
X, where it is necessary to detect writes to the scanout either by the
GPU or through the CPU's fence, and recompress the dirty regions on the
fly. (By comparison to the back-buffer rendering, the scanout is always
recompressed after a page-flip.)

References: https://bugs.freedesktop.org/show_bug.cgi?id=33487
References: https://bugs.freedesktop.org/show_bug.cgi?id=31742
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Use of a CPU fence is mandatory to update FBC regions upon CPU writes
Chris Wilson [Fri, 8 Jul 2011 11:22:39 +0000 (12:22 +0100)]
drm/i915: Use of a CPU fence is mandatory to update FBC regions upon CPU writes

...and this requirement is enforced by intel_update_fbc() so we can
remove the later check from g4x_enable_fbc() and ironlake_enable_fbc().

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Remove vestigial pitch from post-gen2 FBC control routines
Chris Wilson [Fri, 8 Jul 2011 11:22:38 +0000 (12:22 +0100)]
drm/i915: Remove vestigial pitch from post-gen2 FBC control routines

The cfb_pitch was only used for 8xx_enable_fbc(), every later routine
was just overwriting the value with itself thanks to a copy'n'paste
error.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Replace direct calls to vfunc.disable_fbc with intel_disable_fbc()
Chris Wilson [Fri, 8 Jul 2011 11:22:37 +0000 (12:22 +0100)]
drm/i915: Replace direct calls to vfunc.disable_fbc with intel_disable_fbc()

...to ensure that any pending FBC enable tasklet is cancelled.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Keith Packard <keithp@keithp.com>
12 years agodrm/i915: Only export the generic intel_disable_fbc() interface
Chris Wilson [Fri, 8 Jul 2011 11:22:36 +0000 (12:22 +0100)]
drm/i915: Only export the generic intel_disable_fbc() interface

As the enable/disable routines will be gain additional complexity in
future patches, it is necessary that all callers do not bypass the
generic interface by calling into the chipset routines directly. to do
this we make the chipset routines static, so there is no choice.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Keith Packard <keithp@keithp.com>