platform/kernel/linux-exynos.git
9 years agoagp: change agp_free_page_array to use kvfree
Wang, Yalin [Wed, 28 Jan 2015 05:57:34 +0000 (13:57 +0800)]
agp: change agp_free_page_array to use kvfree

Change agp_free_page_array to use kvfree function,
remove the duplicated code.

Signed-off-by: Yalin Wang <yalin.wang@sonymobile.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
9 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux...
Dave Airlie [Thu, 29 Jan 2015 01:45:31 +0000 (11:45 +1000)]
Merge branch 'master' of git://git./linux/kernel/git/torvalds/linux into drm-next

This backmerges drm-fixes into drm-next mainly for the amdkfd
stuff, I'm not 100% confident, but it builds and the amdkfd
folks can fix anything up.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Conflicts:
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h

9 years agoMerge tag 'drm/panel/for-3.20-rc1' of git://anongit.freedesktop.org/tegra/linux into...
Dave Airlie [Thu, 29 Jan 2015 01:29:35 +0000 (11:29 +1000)]
Merge tag 'drm/panel/for-3.20-rc1' of git://anongit.freedesktop.org/tegra/linux into drm-next

drm/panel: Changes for v3.20-rc1

This contains the long-awaited drm_bridge series that makes Chromebooks
work for people. I had thought this would've been perfect by now, but
then I go and build test it and the first thing it does is yell about a
recursive dependency. I fixed that up because I was feeling bad for not
getting around to look at this earlier.

Biseds that there is new support for two more panels, a couple of fixup
patches to the Sharp LQ101R1SX01 dual-channel DSI panel driver and a
potential NULL pointer dereference fix.

* tag 'drm/panel/for-3.20-rc1' of git://anongit.freedesktop.org/tegra/linux: (23 commits)
  drm/bridge: dw-hdmi: Adapt to bridge API change
  drm/sti: fixup for bridge interface
  drm/bridge: dw-hdmi: Fix return error path
  drm: Check the right variable when setting formats
  Documentation: bridge: Add documentation for ps8622 DT properties
  Documentation: devicetree: Add vendor prefix for parade
  Documentation: drm: bridge: move to video/bridge
  drm/bridge: ptn3460: use gpiod interface
  drm/bridge: ptn3460: probe connector at the end of bridge attach
  drm/bridge: ptn3460: support drm_panel
  drm/exynos: dp: support drm_bridge
  drm/bridge: ptn3460: Convert to I2C driver model
  drm/bridge: make bridge registration independent of drm flow
  drm/bridge: do not pass drm_bridge_funcs to drm_bridge_init
  drm/bridge: ptn3460: Few trivial cleanups
  drm/panel: simple: Add AVIC TM070DDH03 panel support
  of: Add vendor prefix for Shanghai AVIC Optoelectronics Co., Ltd.
  drm/panel: sharp: lq101r1sx01: Remove unneeded include
  drm/panel: sharp: lq101r1sx01: Respect power timings
  drm/panel: sharp: lq101r1sx01: Add delay after display on
  ...

9 years agodrm/bridge: dw-hdmi: Adapt to bridge API change
Fabio Estevam [Tue, 27 Jan 2015 12:21:49 +0000 (10:21 -0200)]
drm/bridge: dw-hdmi: Adapt to bridge API change

Commit fbc4572e9c48e45b ("drm/bridge: make bridge registration independent of
drm flow") introduced some drm/bridge API modifications. Make the necessary
changes so that we can avoid the build breakage:

drivers/gpu/drm/bridge/dw_hdmi.c: In function 'dw_hdmi_bridge_destroy':
drivers/gpu/drm/bridge/dw_hdmi.c:1378:2: error: implicit declaration of function 'drm_bridge_cleanup' [-Werror=implicit-function-declaration]
drivers/gpu/drm/bridge/dw_hdmi.c: At top level:
drivers/gpu/drm/bridge/dw_hdmi.c:1471:2: error: unknown field 'destroy' specified in initializer
drivers/gpu/drm/bridge/dw_hdmi.c: In function 'dw_hdmi_register':
drivers/gpu/drm/bridge/dw_hdmi.c:1535:2: error: implicit declaration of function 'drm_bridge_init' [-Werror=implicit-function-declaration]

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/sti: fixup for bridge interface
Dave Airlie [Mon, 26 Jan 2015 23:37:03 +0000 (09:37 +1000)]
drm/sti: fixup for bridge interface

So sti doesn't build because the bridge interfaces changes didn't
catch up to its new DVO driver.

Now I might just carry this patch, but I might just push the
bridge pull into a side-pull until someone resolves it.

So this might not be the right solution to the problem, so
please figure it out and let me know ASAP.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/bridge: dw-hdmi: Fix return error path
Fabio Estevam [Tue, 27 Jan 2015 12:54:12 +0000 (10:54 -0200)]
drm/bridge: dw-hdmi: Fix return error path

If devm_request_threaded_irq() fails we should jump to 'err_iahb' label that
will disable the clocks that were previously enabled.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm: Check the right variable when setting formats
Dan Carpenter [Wed, 28 Jan 2015 06:43:35 +0000 (09:43 +0300)]
drm: Check the right variable when setting formats

When setting the video bus supported formats for a display device using
drm_display_info_set_bus_formats(), check for the proper variable after
duplicating memory.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoDocumentation: bridge: Add documentation for ps8622 DT properties
Ajay Kumar [Tue, 20 Jan 2015 16:38:52 +0000 (22:08 +0530)]
Documentation: bridge: Add documentation for ps8622 DT properties

Add documentation for DT properties supported by ps8622/ps8625
eDP-LVDS converter.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoDocumentation: devicetree: Add vendor prefix for parade
Ajay Kumar [Tue, 20 Jan 2015 16:38:51 +0000 (22:08 +0530)]
Documentation: devicetree: Add vendor prefix for parade

ps8622 eDP-LVDS converter bridge chip is from parade technologies

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoDocumentation: drm: bridge: move to video/bridge
Ajay Kumar [Tue, 20 Jan 2015 16:38:50 +0000 (22:08 +0530)]
Documentation: drm: bridge: move to video/bridge

Move drm/bridge documentation to video/bridge.
Also, add proper documentation for gpios used by ptn3460.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/bridge: ptn3460: use gpiod interface
Ajay Kumar [Tue, 20 Jan 2015 16:38:49 +0000 (22:08 +0530)]
drm/bridge: ptn3460: use gpiod interface

Modify driver to support gpiod interface.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/bridge: ptn3460: probe connector at the end of bridge attach
Ajay Kumar [Tue, 20 Jan 2015 16:38:48 +0000 (22:08 +0530)]
drm/bridge: ptn3460: probe connector at the end of bridge attach

Force bridge connector detection at the end of the bridge attach.
This is needed to detect the bridge connector early.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/bridge: ptn3460: support drm_panel
Ajay Kumar [Tue, 20 Jan 2015 16:38:47 +0000 (22:08 +0530)]
drm/bridge: ptn3460: support drm_panel

Add drm_panel calls to the driver to make the panel and
bridge work together in tandem.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/exynos: dp: support drm_bridge
Ajay Kumar [Tue, 20 Jan 2015 16:38:46 +0000 (22:08 +0530)]
drm/exynos: dp: support drm_bridge

Modify driver to support drm_bridge.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/bridge: ptn3460: Convert to I2C driver model
Ajay Kumar [Tue, 20 Jan 2015 16:38:45 +0000 (22:08 +0530)]
drm/bridge: ptn3460: Convert to I2C driver model

Use drm_bridge helpers to modify the driver to support I2C driver model.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
[treding@nvidia.com: remove recursive dependency on I2C]
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/bridge: make bridge registration independent of drm flow
Ajay Kumar [Tue, 20 Jan 2015 16:38:44 +0000 (22:08 +0530)]
drm/bridge: make bridge registration independent of drm flow

Currently, third party bridge drivers(ptn3460) are dependent
on the corresponding encoder driver init, since bridge driver
needs a drm_device pointer to finish drm initializations.
The encoder driver passes the drm_device pointer to the
bridge driver. Because of this dependency, third party drivers
like ptn3460 doesn't adhere to the driver model.

In this patch, we reframe the bridge registration framework
so that bridge initialization is split into 2 steps, and
bridge registration happens independent of drm flow:
--Step 1: gather all the bridge settings independent of drm and
  add the bridge onto a global list of bridges.
--Step 2: when the encoder driver is probed, call drm_bridge_attach
  for the corresponding bridge so that the bridge receives
  drm_device pointer and continues with connector and other
  drm initializations.

The old set of bridge helpers are removed, and a set of new helpers
are added to accomplish the 2 step initialization.

The bridge devices register themselves onto global list of bridges
when they get probed by calling "drm_bridge_add".

The parent encoder driver waits till the bridge is available
in the lookup table(by calling "of_drm_find_bridge") and then
continues with its initialization.

The encoder driver should also call "drm_bridge_attach" to pass
on the drm_device to the bridge object.

drm_bridge_attach inturn calls "bridge->funcs->attach" so that
bridge can continue with drm related initializations.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/bridge: do not pass drm_bridge_funcs to drm_bridge_init
Ajay Kumar [Tue, 20 Jan 2015 16:38:43 +0000 (22:08 +0530)]
drm/bridge: do not pass drm_bridge_funcs to drm_bridge_init

Assign the pointer to bridge ops structure(drm_bridge_funcs) in
the bridge driver itself, instead of passing it to drm_bridge_init.

This will allow bridge driver developer to pack bridge private
information inside the bridge object and pass only the drm-relevant
information to drm_bridge_init.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/bridge: ptn3460: Few trivial cleanups
Ajay Kumar [Tue, 20 Jan 2015 16:38:42 +0000 (22:08 +0530)]
drm/bridge: ptn3460: Few trivial cleanups

This patch does the following changes:
-- Use usleep_range instead of udelay.
-- Remove driver_private member from ptn3460 structure.
-- Make all possible functions and structures static.
-- Use dev_err for non-DRM errors.
-- Arrange header files alphabetically.
-- s/edid/EDID in all error messages.

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Acked-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Rahul Sharma <rahul.sharma@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Tested-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Tested-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/panel: simple: Add AVIC TM070DDH03 panel support
Philipp Zabel [Thu, 18 Dec 2014 15:43:43 +0000 (16:43 +0100)]
drm/panel: simple: Add AVIC TM070DDH03 panel support

The Shanghai AVIC Optoelectronics TM070DDH03 is a 7" 1024x600 TFT LCD
panel connecting to a 24-bit RGB LVDS interface.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoof: Add vendor prefix for Shanghai AVIC Optoelectronics Co., Ltd.
Philipp Zabel [Thu, 18 Dec 2014 15:43:42 +0000 (16:43 +0100)]
of: Add vendor prefix for Shanghai AVIC Optoelectronics Co., Ltd.

Shanghai AVIC Optoelectronics Co., Ltd. is a subsidiary of Tianma
Microelectronics Co., Ltd. and designs and manufactures TFT LCDs.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/panel: sharp: lq101r1sx01: Remove unneeded include
Thierry Reding [Thu, 18 Dec 2014 16:11:47 +0000 (17:11 +0100)]
drm/panel: sharp: lq101r1sx01: Remove unneeded include

Nothing in the file needs symbols from include/linux/host1x.h, so remove
the include.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/panel: sharp: lq101r1sx01: Respect power timings
Thierry Reding [Mon, 8 Dec 2014 14:38:17 +0000 (15:38 +0100)]
drm/panel: sharp: lq101r1sx01: Respect power timings

Before shutting down the display using the DCS display_off command, wait
for 4 frames according to the datasheet.

Furthermore, after enabling the power supply, the supply voltage needs
around 10 ms to settle. After that, another 120 ms is required before a
DCS exit_sleep_mode command can be sent.

While at it, no longer send the DCS soft_reset command. This is totally
unnecessary because we've just powered up the display, hence it will be
in a reset state already.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/panel: sharp: lq101r1sx01: Add delay after display on
Thierry Reding [Mon, 8 Dec 2014 14:05:48 +0000 (15:05 +0100)]
drm/panel: sharp: lq101r1sx01: Add delay after display on

After switching the display on (using the DCS display_on command), wait
for 6 frames (100ms at 60 Hz) to give the display more time to prepare.
Failing to do this results in the panel not initializing properly in a
large number of cases.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/panel: simple: Add support for Giantplus GPG482739QS5
Philipp Zabel [Wed, 19 Nov 2014 09:29:55 +0000 (10:29 +0100)]
drm/panel: simple: Add support for Giantplus GPG482739QS5

This patch adds support for the GiantPlus GPG48273QS5 4.3" WQVGA TFT LCD
panel to the simple-panel driver.

This panel is connected via a parallel bus and uses both HSYNC and
VSYNC, whose lengths are unfortunately not clearly defined. The
datasheet only specifies the front- and backporch length, but the timing
diagram suggests that both sync signals should be asserted for exactly
one clock cycle.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoof: Add vendor prefix for Giantplus Technology Co., Ltd.
Philipp Zabel [Wed, 19 Nov 2014 09:29:54 +0000 (10:29 +0100)]
of: Add vendor prefix for Giantplus Technology Co., Ltd.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/mipi-dsi: Avoid potential NULL pointer dereference
Thierry Reding [Fri, 5 Dec 2014 10:46:56 +0000 (11:46 +0100)]
drm/mipi-dsi: Avoid potential NULL pointer dereference

The mipi_dsi_packet_create() function dereferences the msg pointer
before checking that it's valid. Move the dereference down to where it
is required to avoid potentially dereferencing a NULL pointer.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Wed, 28 Jan 2015 03:12:38 +0000 (19:12 -0800)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "This feels larger than I'd like but its for three reasons.

   a) amdkfd finalising the API more, this is a new feature introduced
      last merge window, and I'd prefer to make the tweaks to the API
      before it first gets into a stable release.

   b) radeon regression required splitting an internal API to fix
      properly, so it just changed a few more lines

   c) vmwgfx fix changes a lock from a mutex->spin lock, this is fallout
      from the new sleep checking.

  Otherwise there is just some tda998x fixes"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/radeon: Remove rdev->gart.pages_addr array
  drm/radeon: Restore GART table contents after pinning it in VRAM v3
  drm/radeon: Split off gart_get_page_entry ASIC hook from set_page_entry
  drm/amdkfd: Fix bug in call to init_pipelines()
  drm/amdkfd: Fix bug in pipelines initialization
  drm/radeon: Don't increment pipe_id in kgd_init_pipeline
  drm/i2c: tda998x: set the CEC I2C address based on the slave I2C address
  drm/vmwgfx: Replace the hw mutex with a hw spinlock
  drm/amdkfd: Allow user to limit only queues per device
  drm/amdkfd: PQM handle queue creation fault
  drm: tda998x: Fix EDID read timeout on HDMI connect
  drm: tda998x: Protect the page register

9 years agoMerge tag 'topic/atomic-core-2015-01-27' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Tue, 27 Jan 2015 23:34:27 +0000 (09:34 +1000)]
Merge tag 'topic/atomic-core-2015-01-27' of git://anongit.freedesktop.org/drm-intel into drm-next

* tag 'topic/atomic-core-2015-01-27' of git://anongit.freedesktop.org/drm-intel:
  drm/atomic: Fix potential use of state after free
  drm/atomic-helper: debug output for modesets
  drm/atomic-helpers: Saner encoder/crtc callbacks
  drm/atomic-helpers: Recover full cursor plane behaviour
  drm/atomic-helper: add connector->dpms() implementation
  drm/atomic: Add drm_crtc_state->active
  drm: Add standardized boolean props
  drm/plane-helper: Fix transitional helper kerneldocs
  drm/plane-helper: Skip prepare_fb/cleanup_fb when newfb==oldfb

Conflicts:
include/drm/drm_crtc_helper.h

9 years agoMerge tag 'drm/tegra/for-3.20-rc1' of git://anongit.freedesktop.org/tegra/linux into...
Dave Airlie [Tue, 27 Jan 2015 23:27:29 +0000 (09:27 +1000)]
Merge tag 'drm/tegra/for-3.20-rc1' of git://anongit.freedesktop.org/tegra/linux into drm-next

drm/tegra: Changes for v3.20-rc1

The biggest part of these changes is the conversion to atomic mode-
setting. A lot of cleanup and demidlayering was required before the
conversion, with the result being a whole lot of changes.

Besides the atomic mode-setting support, the host1x bus now has the
proper infrastructure to support suspend/resume for child devices.

Finally, a couple of smaller cleanup patches round things off.

* tag 'drm/tegra/for-3.20-rc1' of git://anongit.freedesktop.org/tegra/linux: (54 commits)
  drm/tegra: Use correct relocation target offsets
  drm/tegra: Add minimal power management
  drm/tegra: dc: Unify enabling the display controller
  drm/tegra: Track tiling and format in plane state
  drm/tegra: Track active planes in CRTC state
  drm/tegra: Remove unused ->mode_fixup() callbacks
  drm/tegra: Atomic conversion, phase 3, step 3
  drm/tegra: Atomic conversion, phase 3, step 2
  drm/tegra: dc: Use atomic clock state in modeset
  drm/tegra: sor: Implement ->atomic_check()
  drm/tegra: hdmi: Implement ->atomic_check()
  drm/tegra: dsi: Implement ->atomic_check()
  drm/tegra: rgb: Implement ->atomic_check()
  drm/tegra: dc: Store clock setup in atomic state
  drm/tegra: Atomic conversion, phase 3, step 1
  drm/tegra: Atomic conversion, phase 2
  drm/tegra: Atomic conversion, phase 1
  drm/tegra: dc: Do not needlessly deassert reset
  drm/tegra: Output cleanup functions cannot fail
  drm/tegra: Remove remnants of the output midlayer
  ...

9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Tue, 27 Jan 2015 21:55:36 +0000 (13:55 -0800)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Don't OOPS on socket AIO, from Christoph Hellwig.

 2) Scheduled scans should be aborted upon RFKILL, from Emmanuel
    Grumbach.

 3) Fix sleep in atomic context in kvaser_usb, from Ahmed S Darwish.

 4) Fix RCU locking across copy_to_user() in bpf code, from Alexei
    Starovoitov.

 5) Lots of crash, memory leak, short TX packet et al bug fixes in
    sh_eth from Ben Hutchings.

 6) Fix memory corruption in SCTP wrt.  INIT collitions, from Daniel
    Borkmann.

 7) Fix return value logic for poll handlers in netxen, enic, and bnx2x.
    From Eric Dumazet and Govindarajulu Varadarajan.

 8) Header length calculation fix in mac80211 from Fred Chou.

 9) mv643xx_eth doesn't handle highmem correctly in non-TSO code paths.
    From Ezequiel Garcia.

10) udp_diag has bogus logic in it's hash chain skipping, copy same fix
    tcp diag used.  From Herbert Xu.

11) amd-xgbe programs wrong rx flow control register, from Thomas
    Lendacky.

12) Fix race leading to use after free in ping receive path, from Subash
    Abhinov Kasiviswanathan.

13) Cache redirect routes otherwise we can get a heavy backlog of rcu
    jobs liberating DST_NOCACHE entries.  From Hannes Frederic Sowa.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (48 commits)
  net: don't OOPS on socket aio
  stmmac: prevent probe drivers to crash kernel
  bnx2x: fix napi poll return value for repoll
  ipv6: replacing a rt6_info needs to purge possible propagated rt6_infos too
  sh_eth: Fix DMA-API usage for RX buffers
  sh_eth: Check for DMA mapping errors on transmit
  sh_eth: Ensure DMA engines are stopped before freeing buffers
  sh_eth: Remove RX overflow log messages
  ping: Fix race in free in receive path
  udp_diag: Fix socket skipping within chain
  can: kvaser_usb: Fix state handling upon BUS_ERROR events
  can: kvaser_usb: Retry the first bulk transfer on -ETIMEDOUT
  can: kvaser_usb: Send correct context to URB completion
  can: kvaser_usb: Do not sleep in atomic context
  ipv4: try to cache dst_entries which would cause a redirect
  samples: bpf: relax test_maps check
  bpf: rcu lock must not be held when calling copy_to_user()
  net: sctp: fix slab corruption from use after free on INIT collisions
  net: mv643xx_eth: Fix highmem support in non-TSO egress path
  sh_eth: Fix serialisation of interrupt disable with interrupt & NAPI handlers
  ...

9 years agonet: don't OOPS on socket aio
Christoph Hellwig [Tue, 27 Jan 2015 20:25:33 +0000 (12:25 -0800)]
net: don't OOPS on socket aio

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agostmmac: prevent probe drivers to crash kernel
Andy Shevchenko [Tue, 27 Jan 2015 16:38:03 +0000 (18:38 +0200)]
stmmac: prevent probe drivers to crash kernel

In the case when alloc_netdev fails we return NULL to a caller. But there is no
check for NULL in the probe drivers. This patch changes NULL to an error
pointer. The function description is amended to reflect what we may get
returned.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge tag 'powerpc-3.19-5' of git://git.kernel.org/pub/scm/linux/kernel/git/mpe/linux
Linus Torvalds [Tue, 27 Jan 2015 18:04:38 +0000 (10:04 -0800)]
Merge tag 'powerpc-3.19-5' of git://git./linux/kernel/git/mpe/linux

Pull powerpc fixes from Michael Ellerman:
 "Two powerpc fixes"

* tag 'powerpc-3.19-5' of git://git.kernel.org/pub/scm/linux/kernel/git/mpe/linux:
  powerpc/powernv: Restore LPCR with LPCR_PECE1 cleared
  powerpc/xmon: Fix another endiannes issue in RTAS call from xmon

9 years agoMerge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty...
Linus Torvalds [Tue, 27 Jan 2015 17:02:09 +0000 (09:02 -0800)]
Merge tag 'fixes-for-linus' of git://git./linux/kernel/git/rusty/linux

Pull one more module fix from Rusty Russell:
 "SCSI was using module_refcount() to figure out when the module was
  unloading: this broke with new atomic refcounting.  The code is still
  suspicious, but this solves the WARN_ON()"

* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux:
  scsi: always increment reference count

9 years agodrm/tegra: Use correct relocation target offsets
David Ung [Wed, 21 Jan 2015 02:37:35 +0000 (18:37 -0800)]
drm/tegra: Use correct relocation target offsets

When copying a relocation from userspace, copy the correct target
offset.

Signed-off-by: David Ung <davidu@nvidia.com>
Fixes: 961e3beae3b2 ("drm/tegra: Make job submission 64-bit safe")
Cc: stable@vger.kernel.org
[treding@nvidia.com: provide a better commit message]
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Add minimal power management
Thierry Reding [Thu, 18 Dec 2014 16:15:25 +0000 (17:15 +0100)]
drm/tegra: Add minimal power management

For now only disable the KMS hotplug polling helper logic upon suspend
and re-enable it on resume.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Mark Zhang <markz@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Unify enabling the display controller
Thierry Reding [Mon, 8 Dec 2014 15:32:47 +0000 (16:32 +0100)]
drm/tegra: dc: Unify enabling the display controller

Previously output drivers would enable continuous display mode and power
up the display controller at various points during the initialization.
This is suboptimal because it accesses display controller registers in
output drivers and duplicates a bit of code.

Move this code into the display controller driver and enable the display
controller as the final step of the ->mode_set_nofb() implementation.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Track tiling and format in plane state
Thierry Reding [Fri, 28 Nov 2014 12:14:55 +0000 (13:14 +0100)]
drm/tegra: Track tiling and format in plane state

Tracking these in the plane state allows them to be computed in the
->atomic_check() callback and reused when applying the configuration in
->atomic_update().

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Track active planes in CRTC state
Thierry Reding [Wed, 26 Nov 2014 11:28:39 +0000 (12:28 +0100)]
drm/tegra: Track active planes in CRTC state

Wrap struct drm_crtc_state in a driver-specific structure and add the
planes field which keeps track of which planes are updated or disabled
during a modeset. This allows atomic updates of the the display engine
at ->atomic_flush() time.

v2: open-code getting the state of the CRTC that the plane is being
    attached to (Daniel Vetter)

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Remove unused ->mode_fixup() callbacks
Thierry Reding [Fri, 19 Dec 2014 14:19:21 +0000 (15:19 +0100)]
drm/tegra: Remove unused ->mode_fixup() callbacks

All output drivers have now been converted to use the ->atomic_check()
callback, so the ->mode_fixup() callbacks are no longer used.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Atomic conversion, phase 3, step 3
Thierry Reding [Mon, 24 Nov 2014 16:41:23 +0000 (17:41 +0100)]
drm/tegra: Atomic conversion, phase 3, step 3

Provide a custom ->atomic_commit() implementation which supports async
commits. The generic atomic page-flip helper can use this to implement
page-flipping.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Atomic conversion, phase 3, step 2
Thierry Reding [Mon, 24 Nov 2014 16:08:20 +0000 (17:08 +0100)]
drm/tegra: Atomic conversion, phase 3, step 2

Replace drm_crtc_helper_set_config() by drm_atomic_helper_set_config().
All drivers have now been converted to use ->atomic_check() to set the
atomic state, therefore the atomic mode setting helpers can be used.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Use atomic clock state in modeset
Thierry Reding [Fri, 19 Dec 2014 14:09:16 +0000 (15:09 +0100)]
drm/tegra: dc: Use atomic clock state in modeset

All clock state is now stored in the display controller's atomic state,
so the output drivers no longer need to call back into the display
controller driver to set up the clock. This is also required to make
sure no hardware changes are made before validating a configuration.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: sor: Implement ->atomic_check()
Thierry Reding [Mon, 8 Dec 2014 16:26:46 +0000 (17:26 +0100)]
drm/tegra: sor: Implement ->atomic_check()

The implementation of the ->atomic_check() callback precomputes all
parameters to check if the given configuration can be applied. If so the
precomputed values are stored in the atomic state object for the encoder
and applied during modeset. In that way the modeset no longer needs to
perform any checking but simply program values into registers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: hdmi: Implement ->atomic_check()
Thierry Reding [Mon, 8 Dec 2014 15:33:03 +0000 (16:33 +0100)]
drm/tegra: hdmi: Implement ->atomic_check()

The implementation of the ->atomic_check() callback precomputes all
parameters to check if the given configuration can be applied. If so the
precomputed values are stored in the atomic state object for the encoder
and applied during modeset. In that way the modeset no longer needs to
perform any checking but simply program values into registers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dsi: Implement ->atomic_check()
Thierry Reding [Mon, 8 Dec 2014 15:22:28 +0000 (16:22 +0100)]
drm/tegra: dsi: Implement ->atomic_check()

The implementation of the ->atomic_check() callback precomputes all
parameters to check if the given configuration can be applied. If so the
precomputed values are stored in the atomic state object for the encoder
and applied during modeset. In that way the modeset no longer needs to
perform any checking but simply program values into registers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: rgb: Implement ->atomic_check()
Thierry Reding [Wed, 17 Dec 2014 16:04:36 +0000 (17:04 +0100)]
drm/tegra: rgb: Implement ->atomic_check()

The implementation of the ->atomic_check() callback precomputes all
parameters to check if the given configuration can be applied. If so the
precomputed values are stored in the atomic state object for the encoder
and applied during modeset. In that way the modeset no longer needs to
perform any checking but simply program values into registers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Store clock setup in atomic state
Thierry Reding [Mon, 8 Dec 2014 15:14:45 +0000 (16:14 +0100)]
drm/tegra: dc: Store clock setup in atomic state

This allows the clock setup to be separated from the clock programming
and better matches the expectations of the atomic modesetting where no
code paths must fail during modeset.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Atomic conversion, phase 3, step 1
Thierry Reding [Mon, 24 Nov 2014 16:08:06 +0000 (17:08 +0100)]
drm/tegra: Atomic conversion, phase 3, step 1

Switch out the regular plane helpers for the atomic plane helpers. Also
use the default atomic helpers to implement the ->atomic_check() and
->atomic_commit() callbacks. The driver now exclusively uses the atomic
interfaces.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Atomic conversion, phase 2
Thierry Reding [Mon, 24 Nov 2014 16:02:53 +0000 (17:02 +0100)]
drm/tegra: Atomic conversion, phase 2

Hook up the default ->reset() and ->atomic_duplicate_state() helpers.
This ensures that state objects are properly created and framebuffer
reference counts correctly maintained.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Atomic conversion, phase 1
Thierry Reding [Mon, 24 Nov 2014 15:27:13 +0000 (16:27 +0100)]
drm/tegra: Atomic conversion, phase 1

Implement initial atomic state handling. Hook up the CRTCs, planes' and
connectors' ->atomic_destroy_state() callback to ensure that the atomic
state objects don't leak.

Furthermore the CRTC now implements the ->mode_set_nofb() callback that
is used by new helpers to implement ->mode_set() and ->mode_set_base().
These new helpers also make use of the new plane helper functions which
the driver now provides.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Do not needlessly deassert reset
Thierry Reding [Fri, 16 Jan 2015 15:30:55 +0000 (16:30 +0100)]
drm/tegra: dc: Do not needlessly deassert reset

Commit 9c0127004ff4 ("drm/tegra: dc: Add powergate support") changed the
driver's ->probe() implementation to deassert the module reset, and with
there being nobody else to assert it until ->remove() there is no need
to deassert again later on.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Output cleanup functions cannot fail
Thierry Reding [Fri, 19 Dec 2014 14:55:08 +0000 (15:55 +0100)]
drm/tegra: Output cleanup functions cannot fail

The tegra_output_exit() and tegra_output_remove() functions cannot fail,
so make them return void.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Remove remnants of the output midlayer
Thierry Reding [Fri, 19 Dec 2014 14:51:35 +0000 (15:51 +0100)]
drm/tegra: Remove remnants of the output midlayer

The tegra_output midlayer is now completely gone and output drivers use
it purely as a helper library.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: debugfs cleanup cannot fail
Thierry Reding [Fri, 19 Dec 2014 14:47:30 +0000 (15:47 +0100)]
drm/tegra: debugfs cleanup cannot fail

The debugfs cleanup code never fails, so no error is returned. Therefore
the functions can all return void instead.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: sor: Demidlayer
Thierry Reding [Fri, 28 Nov 2014 14:41:34 +0000 (15:41 +0100)]
drm/tegra: sor: Demidlayer

Implement encoder and connector within the eDP driver itself using the
Tegra output helpers rather than using the Tegra output as midlayer. By
doing so one level of indirection is removed and output drivers become
more flexible while keeping the majority of the advantages provided by
the common output helpers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dsi: Demidlayer
Thierry Reding [Tue, 2 Dec 2014 16:30:23 +0000 (17:30 +0100)]
drm/tegra: dsi: Demidlayer

Implement encoder and connector within the DSI driver itself using the
Tegra output helpers rather than using the Tegra output as midlayer. By
doing so one level of indirection is removed and output drivers become
more flexible while keeping the majority of the advantages provided by
the common output helpers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: hdmi: Demidlayer
Thierry Reding [Fri, 28 Nov 2014 15:50:59 +0000 (16:50 +0100)]
drm/tegra: hdmi: Demidlayer

Implement encoder and connector within the HDMI driver itself using the
Tegra output helpers rather than using the Tegra output as midlayer. By
doing so one level of indirection is removed and output drivers become
more flexible while keeping the majority of the advantages provided by
the common output helpers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: rgb: Demidlayer
Thierry Reding [Tue, 16 Dec 2014 17:30:16 +0000 (18:30 +0100)]
drm/tegra: rgb: Demidlayer

Implement encoder and connector within the RGB driver itself using the
Tegra output helpers rather than using the Tegra output as midlayer. By
doing so one level of indirection is removed and output drivers become
more flexible while keeping the majority of the advantages provided by
the common output helpers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Add tegra_dc_setup_clock() helper
Thierry Reding [Tue, 2 Dec 2014 14:15:06 +0000 (15:15 +0100)]
drm/tegra: Add tegra_dc_setup_clock() helper

This is a small helper that performs the basic steps required by all
output drivers to prepare the display controller for use with a given
encoder.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: output: Make ->setup_clock() optional
Thierry Reding [Tue, 2 Dec 2014 14:12:28 +0000 (15:12 +0100)]
drm/tegra: output: Make ->setup_clock() optional

In order to transition output drivers to using the struct tegra_output
as a helper rather than midlayer, make this callback optional. Instead
drivers should implement the equivalent as part of ->mode_fixup(). For
the conversion to atomic modesetting a new callback ->atomic_check()
should be implemented that updates the display controller's state with
the corresponding parent clock, rate and shift clock divider.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Convert output midlayer to helpers
Thierry Reding [Fri, 28 Nov 2014 14:38:40 +0000 (15:38 +0100)]
drm/tegra: Convert output midlayer to helpers

The output layer was initially designed to help reduce the amount of
code duplicated in output drivers. An unfortunate side-effect of that
was that it turned into a midlayer and it became difficult to make the
output drivers work without bending over backwards to fit into the
midlayer.

This commit starts to convert the midlayer into a helper library by
exporting most of the common functions so that they can be used by the
output drivers directly. Doing so will allow output drivers to reuse
common code paths but more easily override them where necessary.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: No longer disable planes at CRTC disable
Thierry Reding [Mon, 17 Nov 2014 14:14:48 +0000 (15:14 +0100)]
drm/tegra: dc: No longer disable planes at CRTC disable

The DRM core should take care of disabling all unneeded planes, so there
is no need to do this explicitly.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Move tegra_drm_mode_funcs to the core
Thierry Reding [Wed, 26 Nov 2014 12:03:57 +0000 (13:03 +0100)]
drm/tegra: Move tegra_drm_mode_funcs to the core

This structure will be extended using non-framebuffer related callbacks
in subsequent patches, so it should move to a more central location.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Wait for idle when disabled
Thierry Reding [Mon, 8 Dec 2014 15:03:53 +0000 (16:03 +0100)]
drm/tegra: dc: Wait for idle when disabled

When disabling the display controller, stop it and wait for it to become
idle. Doing so ensures that no further accesses to the framebuffer occur
and the buffers can be safely unmapped or freed.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Stop CRTC at CRTC disable time
Thierry Reding [Fri, 21 Nov 2014 16:35:54 +0000 (17:35 +0100)]
drm/tegra: Stop CRTC at CRTC disable time

Previously output drivers would all stop the display controller in their
disable path. However with the transition to atomic modesetting the
display controller needs to be kept running until all planes have been
disabled so that software can properly determine (using VBLANK counts)
when it is safe to remove the framebuffers associated with the planes.

Moving this code into the display controller's disable path also gets
rid of the duplication of this into all output drivers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Use tegra_commit_dc() in output drivers
Thierry Reding [Fri, 21 Nov 2014 16:33:33 +0000 (17:33 +0100)]
drm/tegra: Use tegra_commit_dc() in output drivers

All output drivers have open-coded variants of this function, so export
it to remove some code duplication.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/atomic: Add ->atomic_check() to encoder helpers
Thierry Reding [Wed, 3 Dec 2014 15:44:34 +0000 (16:44 +0100)]
drm/atomic: Add ->atomic_check() to encoder helpers

This callback can be used instead of the legacy ->mode_fixup() and is
passed the CRTC and connector states. It can thus use these states to
validate the modeset and cache values in the state to be used during
the actual modeset.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/plane: Add optional ->atomic_disable() callback
Thierry Reding [Thu, 20 Nov 2014 11:05:50 +0000 (12:05 +0100)]
drm/plane: Add optional ->atomic_disable() callback

In order to prevent drivers from having to perform the same checks over
and over again, add an optional ->atomic_disable callback which the core
calls under the right circumstances.

v2: pass old state and detect edges to avoid calling ->atomic_disable on
already disabled planes, remove redundant comment (Daniel Vetter)

v3: rename helper to drm_atomic_plane_disabling() to clarify that it is
checking for transitions, move helper to drm_atomic_helper.h, clarify
check for !old_state and its relation to transitional helpers

Here's an extract from some discussion rationalizing the behaviour (for
a full version, see the reference below):

    > > Hm, thinking about this some more this will result in a slight difference
    > > in behaviour, at least when drivers just use the helper ->reset functions
    > > but don't disable everything:
    > > - With transitional helpers we assume we know nothing and call
    > >   ->atomic_disable.
    > > - With atomic old_state->crtc == NULL in the same situation right after
    > >   boot-up, but we asssume the plane is really off and _dont_ call
    > >   ->atomic_disable.
    > >
    > > Should we instead check for (old_state && old_state->crtc) and state that
    > > drivers need to make sure they don't have stuff hanging around?
    >
    > I don't think we can check for old_state because otherwise this will
    > always return false, whereas we really want it to force-disable planes
    > that could be on (lacking any more accurate information). For
    > transitional helpers anyway.
    >
    > For the atomic helpers, old_state will never be NULL, but I'd assume
    > that the driver would reconstruct the current state in ->reset().

    By the way, the reason for why old_state can be NULL with transitional
    helpers is the ordering of the steps in the atomic transition. Currently
    the Tegra patches do this (based on your blog post and the Exynos proto-
    type):

        1) atomic conversion, phase 1:
           - implement ->atomic_{check,update,disable}()
           - use drm_plane_helper_{update,disable}()

        2) atomic conversion, phase 2:
           - call drm_mode_config_reset() from ->load()
           - implement ->reset()

    That's only a partial list of what's done in these steps, but that's the
    only relevant pieces for why old_state is NULL.

    What happens is that without ->reset() implemented there won't be any
    initial state, hence plane->state (the old_state here) will be NULL the
    first time atomic state is applied.

    We could of course reorder the sequence such that drivers are required
    to hook up ->reset() before they can (or at the same as they) hook up
    the transitional helpers. We could add an appropriate WARN_ON to this
    helper to make that more obvious.

    However, that will not solve the problem because it only gets rid of the
    special case. We still don't know whether old_state->crtc == NULL is the
    current state or just the initial default.

    So no matter which way we do this, I don't see a way to get away without
    requiring specific semantics from drivers. They would be that:

        - drivers recreate the correct state in ->reset() so that
          old_state->crtc != NULL if the plane is really enabled

    or

        - drivers have to ensure that the real state in fact mirrors the
          initial default as encoded in the state (plane disabled)

References: http://lists.freedesktop.org/archives/dri-devel/2015-January/075578.html
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/plane: Make ->atomic_update() mandatory
Thierry Reding [Tue, 25 Nov 2014 12:05:12 +0000 (13:05 +0100)]
drm/plane: Make ->atomic_update() mandatory

There is no use-case where it would be useful for drivers not to
implement this function and the transitional plane helpers already
require drivers to provide an implementation.

v2: add new requirement to kerneldoc

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: gem: oops in error handling
Dan Carpenter [Fri, 16 Jan 2015 10:43:42 +0000 (13:43 +0300)]
drm/tegra: gem: oops in error handling

kfree(ERR_PTR(-ENOMEM)) will not work very well.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Fix bad irqsave/restore in tegra_dc_finish_page_flip()
Dan Carpenter [Wed, 7 Jan 2015 11:01:26 +0000 (14:01 +0300)]
drm/tegra: dc: Fix bad irqsave/restore in tegra_dc_finish_page_flip()

We can't save two values to the IRQ flags at the same time so the IRQs
are not enabled at the end.  This kind of bug is easy to miss in testing
if the function is normally called with IRQs disabled so we wouldn't
enable IRQs anyway.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dsi: Adjust D-PHY timing
David Ung [Fri, 5 Dec 2014 23:30:05 +0000 (15:30 -0800)]
drm/tegra: dsi: Adjust D-PHY timing

Compliance testing shows that HS Trail is off by -12%. Increase the HS
Trail time to make this test pass.

Signed-off-by: David Ung <davidu@nvidia.com>
[treding@nvidia.com: update specification references, add comment]
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dsi: Reset across ->exit()/->init()
Thierry Reding [Mon, 24 Nov 2014 15:31:48 +0000 (16:31 +0100)]
drm/tegra: dsi: Reset across ->exit()/->init()

This allows a DRM driver unload/reload cycle to completely reset the DSI
controller and may help in situations where it's broken.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dsi: Soft-reset controller on ->disable
Thierry Reding [Mon, 24 Nov 2014 15:29:40 +0000 (16:29 +0100)]
drm/tegra: dsi: Soft-reset controller on ->disable

This reset is necessary to properly clean up the internal state of the
controller.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dsi: Registers are 32-bit
Thierry Reding [Mon, 24 Nov 2014 11:27:59 +0000 (12:27 +0100)]
drm/tegra: dsi: Registers are 32-bit

Use a sized unsigned 32-bit data type (u32) to store register contents.
The DSI registers are 32 bits wide irrespective of the architecture's
data width.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: hdmi: Registers are 32-bit
Thierry Reding [Mon, 8 Dec 2014 15:25:14 +0000 (16:25 +0100)]
drm/tegra: hdmi: Registers are 32-bit

Use a sized unsigned 32-bit data type (u32) to store register contents.
The HDMI registers are 32 bits wide irrespective of the architecture's
data width.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Return planar flag for non-YUV modes
Thierry Reding [Mon, 8 Dec 2014 14:55:28 +0000 (15:55 +0100)]
drm/tegra: dc: Return planar flag for non-YUV modes

This prevents the compiler from warning about using a variable that is
possibly uninitialized.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Describe register copies
Thierry Reding [Mon, 8 Dec 2014 14:50:04 +0000 (15:50 +0100)]
drm/tegra: dc: Describe register copies

Most of the display controller's registers are double-buffered, a few of
them are triple-buffered. The ASSEMBLY shadow copy is latched intto the
ACTIVE copy for double-buffered registers. For triple-buffered registers
the ASSEMBLY copy is first latched into the ARM copy.

Latching into the ACTIVE copy happens immediately if the controller is
inactive. Otherwise the latching happens on the next frame boundary. The
latching of the ASSEMBLY into the ARM copy happens immediately. Latching
is controlled by a set of *_ACT_REQ and *_UPDATE bits in the
DC_CMD_STATE_CONTROL register.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: dc: Initialize border color
Thierry Reding [Mon, 8 Dec 2014 14:45:39 +0000 (15:45 +0100)]
drm/tegra: dc: Initialize border color

Tegra114 and earlier support specifying the color of the border (i.e.
the active area of the screen that is not covered by any of the overlay
windows). By default this is set to a light blue, so set it to black to
comply with the requirements set by atomic modesetting.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Check for NULL pointer instead of IS_ERR()
Dan Carpenter [Thu, 4 Dec 2014 11:00:35 +0000 (14:00 +0300)]
drm/tegra: Check for NULL pointer instead of IS_ERR()

iommu_domain_alloc() returns NULL on error, it never returns error
pointers.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: plane: Use proper possible_crtcs mask
Thierry Reding [Tue, 16 Dec 2014 17:04:08 +0000 (18:04 +0100)]
drm/tegra: plane: Use proper possible_crtcs mask

The possible_crtcs mask needs to be a mask of CRTC indices. There is no
guarantee that the DRM indices match the hardware pipe number, so the
mask must be computed from the CRTC index.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: Remove redundant zeroing out of memory
Thierry Reding [Wed, 10 Dec 2014 10:41:05 +0000 (11:41 +0100)]
drm/tegra: Remove redundant zeroing out of memory

The DRM core now zeroes out the memory associated with CRTC, encoder and
connector objects upon cleanup, so there's no need to explicitly do that
in drivers anymore.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/tegra: gem: Use iommu_map_sg()
Thierry Reding [Wed, 17 Dec 2014 15:46:37 +0000 (16:46 +0100)]
drm/tegra: gem: Use iommu_map_sg()

The iommu_map_sg() function is now available in the IOMMU API, so drop
the open-coded variant.

Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agogpu: host1x: Provide a proper struct bus_type
Thierry Reding [Thu, 18 Dec 2014 14:29:14 +0000 (15:29 +0100)]
gpu: host1x: Provide a proper struct bus_type

Previously the struct bus_type exported by the host1x infrastructure was
only a very basic skeleton. Turn that implementation into a more full-
fledged bus to support proper probe ordering and power management.

Note that the bus infrastructure needs to be available before any of the
drivers can be registered. This is automatically ensured if all drivers
are built as loadable modules (via symbol dependencies). If all drivers
are built-in there are no such guarantees and the link order determines
the initcall ordering. Adjust drivers/gpu/Makefile to make sure that the
host1x bus infrastructure is initialized prior to any of its users (only
drm/tegra currently).

v2: Fix building host1x and tegra-drm as modules
Reported-by: Dave Airlie <airlied@gmail.com>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Mark Zhang <markz@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agodrm/atomic: Fix potential use of state after free
Ander Conselvan de Oliveira [Fri, 23 Jan 2015 07:27:59 +0000 (09:27 +0200)]
drm/atomic: Fix potential use of state after free

The atomic helpers rely on drm_atomic_state_clear() to reset an atomic
state if a retry is needed due to the w/w mutexes. The subsequent calls
to drm_atomic_get_{crtc,plane,...}_state() would then return the stale
pointers in state->{crtc,plane,...}_states.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/atomic-helper: debug output for modesets
Daniel Vetter [Thu, 22 Jan 2015 15:36:25 +0000 (16:36 +0100)]
drm/atomic-helper: debug output for modesets

With the combination of ->enable and ->active it's a bit complicated
to follow what exactly is going on sometimes within a full modeset.
Add debug output to make this all traceable.

Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/atomic-helpers: Saner encoder/crtc callbacks
Daniel Vetter [Thu, 22 Jan 2015 15:36:24 +0000 (16:36 +0100)]
drm/atomic-helpers: Saner encoder/crtc callbacks

For historical reasons going all the way back to how the Xrandr code
was implemented the semantics of the callbacks used to enable/disable
crtcs and encoders are ... interesting.

But with atomic helpers all that complexity has been binned, with only
a well-defined on/off action left. Unfortunately the names stuck.

Let's fix that by adding enable/disable hooks every, make them the
preferred variant for atomic and update documentations.

Later on we add debug warnings when drivers have deprecated hooks. But
while everything is in-flight with lots of drivers converting to
atomic that's a bit too much - better wait for things to settle a bit
first.

v2: Fix kerneldoc, reported by Wu Fengguang.

Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/atomic-helpers: Recover full cursor plane behaviour
Daniel Vetter [Thu, 22 Jan 2015 15:36:23 +0000 (16:36 +0100)]
drm/atomic-helpers: Recover full cursor plane behaviour

Cursor plane updates have historically been fully async and mutliple
updates batched together for the next vsync. And userspace relies upon
that. Since implementing a full queue of async atomic updates is a bit
of work lets just recover the cursor specific behaviour with a hint
flag and some hacks to drop the vblank wait.

v2: Fix kerneldoc, reported by Wu Fengguang.

Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/atomic-helper: add connector->dpms() implementation
Daniel Vetter [Thu, 22 Jan 2015 17:53:05 +0000 (18:53 +0100)]
drm/atomic-helper: add connector->dpms() implementation

This builds on top of the crtc->active infrastructure to implement
legacy DPMS. My choice of semantics is somewhat arbitrary, but the
entire pipe is enabled as along as one output is still enabled.

Of course it also clamps everything that's not ON to OFF.

v2: Fix spelling in one comment.

v3: Don't do an async commit (Thierry)

v4: Dan Carpenter noticed missing error case handling.

Cc: Thierry Reding <thierry.reding@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Tested-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/atomic: Add drm_crtc_state->active
Daniel Vetter [Thu, 22 Jan 2015 15:36:21 +0000 (16:36 +0100)]
drm/atomic: Add drm_crtc_state->active

This is the infrastructure for DPMS ported to the atomic world.
Fundamental changes compare to legacy DPMS are:

- No more per-connector dpms state, instead there's just one per each
  display pipeline. So if you clone either you have to unclone first
  if you only want to switch off one screen, or you just switch of
  everything (like all desktops do). This massively reduces complexity
  for cloning since now there's no more half-enabled cloned configs to
  consider.

- Only on/off, dpms standby/suspend are as dead as real CRTs. Again
  reduces complexity a lot.

Now especially for backwards compat the really important part for dpms
support is that dpms on always succeeds (except for hw death and
unplugged cables ofc). Which means everything that could fail (like
configuration checking, resources assignments and buffer management)
must be done irrespective from ->active. ->active is really only a
toggle to change the hardware state. More precisely:

- Drivers MUST NOT look at ->active in their ->atomic_check callbacks.
  Changes to ->active MUST always suceed if nothing else changes.

- Drivers using the atomic helpers MUST NOT look at ->active anywhere,
  period. The helpers will take care of calling the respective
  enable/modeset/disable hooks as necessary. As before the helpers
  will carefully keep track of the state and not call any hooks
  unecessarily, so still no double-disables or enables like with crtc
  helpers.

- ->mode_set hooks are only called when the mode or output
  configuration changes, not for changes in ->active state.

- Drivers which reconstruct the state objects in their ->reset hooks
  or through some other hw state readout infrastructure must ensure
  that ->active reflects actual hw state.

This just implements the core bits and helper logic, a subsequent
patch will implement the helper code to implement legacy dpms with
this.

v2: Rebase on top of the drm ioctl work:
- Move crtc checks to the core check function.
- Also check for ->active_changed when deciding whether a modeset
  might happen (for the ALLOW_MODESET mode).
- Expose the ->active state with an atomic prop.

v3: Review from Rob
- Spelling fix in comment.
- Extract needs_modeset helper to consolidate the ->mode_changed ||
  ->active_changed checks.

v4: Fixup fumble between crtc->state and crtc_state.

Cc: Rob Clark <robdclark@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Tested-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm: Add standardized boolean props
Daniel Vetter [Wed, 21 Jan 2015 07:47:38 +0000 (08:47 +0100)]
drm: Add standardized boolean props

Not a new type exposed to userspace, just a standard way to create
them since between range, bitmask and enum there's 3 different ways to
pull out a boolean prop.

Also add the kerneldoc for the recently added new prop types, which
Rob forgot all about.

v2: Fixup kerneldoc, spotted by Rob.

Cc: Rob Clark <robdclark@gmail.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm: Add rotation value to plane state
Matt Roper [Thu, 22 Jan 2015 00:35:40 +0000 (16:35 -0800)]
drm: Add rotation value to plane state

The rotation property is shared by multiple drivers, so it makes sense
to store the rotation value (for atomic-converted drivers) in the common
plane state so that core code can eventually access it as well.

Cc: dri-devel@lists.freedesktop.org
Suggested-by: Daniel Vetter <daniel@ffwll.ch>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
9 years agobnx2x: fix napi poll return value for repoll
Govindarajulu Varadarajan [Sun, 25 Jan 2015 10:39:23 +0000 (16:09 +0530)]
bnx2x: fix napi poll return value for repoll

With the commit d75b1ade567ffab ("net: less interrupt masking in NAPI") napi
repoll is done only when work_done == budget. When in busy_poll is we return 0
in napi_poll. We should return budget.

Signed-off-by: Govindarajulu Varadarajan <_govind@gmx.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec
David S. Miller [Tue, 27 Jan 2015 08:28:38 +0000 (00:28 -0800)]
Merge branch 'master' of git://git./linux/kernel/git/klassert/ipsec

Steffen Klassert says:

====================
ipsec 2015-01-26

Just two small fixes for _decode_session6() where we
might decode to wrong header information in some rare
situations.

Please pull or let me know if there are problems.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoipv6: replacing a rt6_info needs to purge possible propagated rt6_infos too
Hannes Frederic Sowa [Mon, 26 Jan 2015 14:11:17 +0000 (15:11 +0100)]
ipv6: replacing a rt6_info needs to purge possible propagated rt6_infos too

Lubomir Rintel reported that during replacing a route the interface
reference counter isn't correctly decremented.

To quote bug <https://bugzilla.kernel.org/show_bug.cgi?id=91941>:
| [root@rhel7-5 lkundrak]# sh -x lal
| + ip link add dev0 type dummy
| + ip link set dev0 up
| + ip link add dev1 type dummy
| + ip link set dev1 up
| + ip addr add 2001:db8:8086::2/64 dev dev0
| + ip route add 2001:db8:8086::/48 dev dev0 proto static metric 20
| + ip route add 2001:db8:8088::/48 dev dev1 proto static metric 10
| + ip route replace 2001:db8:8086::/48 dev dev1 proto static metric 20
| + ip link del dev0 type dummy
| Message from syslogd@rhel7-5 at Jan 23 10:54:41 ...
|  kernel:unregister_netdevice: waiting for dev0 to become free. Usage count = 2
|
| Message from syslogd@rhel7-5 at Jan 23 10:54:51 ...
|  kernel:unregister_netdevice: waiting for dev0 to become free. Usage count = 2

During replacement of a rt6_info we must walk all parent nodes and check
if the to be replaced rt6_info got propagated. If so, replace it with
an alive one.

Fixes: 4a287eba2de3957 ("IPv6 routing, NLM_F_* flag support: REPLACE and EXCL flags support, warn about missing CREATE flag")
Reported-by: Lubomir Rintel <lkundrak@v3.sk>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Tested-by: Lubomir Rintel <lkundrak@v3.sk>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'sh_eth'
David S. Miller [Tue, 27 Jan 2015 08:18:57 +0000 (00:18 -0800)]
Merge branch 'sh_eth'

Ben Hutchings says:

====================
Fixes for sh_eth #3

I'm continuing review and testing of Ethernet support on the R-Car H2
chip.  This series fixes the last of the more serious issues I've found.

These are not tested on any of the other supported chips.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agosh_eth: Fix DMA-API usage for RX buffers
Ben Hutchings [Tue, 27 Jan 2015 00:50:24 +0000 (00:50 +0000)]
sh_eth: Fix DMA-API usage for RX buffers

- Use the return value of dma_map_single(), rather than calling
  virt_to_page() separately
- Check for mapping failue
- Call dma_unmap_single() rather than dma_sync_single_for_cpu()

Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agosh_eth: Check for DMA mapping errors on transmit
Ben Hutchings [Tue, 27 Jan 2015 00:49:47 +0000 (00:49 +0000)]
sh_eth: Check for DMA mapping errors on transmit

dma_map_single() may fail if an IOMMU or swiotlb is in use, so
we need to check for this.

Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agosh_eth: Ensure DMA engines are stopped before freeing buffers
Ben Hutchings [Tue, 27 Jan 2015 00:49:32 +0000 (00:49 +0000)]
sh_eth: Ensure DMA engines are stopped before freeing buffers

Currently we try to clear EDRRR and EDTRR and immediately continue to
free buffers.  This is unsafe because:

- In general, register writes are not serialised with DMA, so we still
  have to wait for DMA to complete somehow
- The R8A7790 (R-Car H2) manual states that the TX running flag cannot
  be cleared by writing to EDTRR
- The same manual states that clearing the RX running flag only stops
  RX DMA at the next packet boundary

I applied this patch to the driver to detect DMA writes to freed
buffers:

> --- a/drivers/net/ethernet/renesas/sh_eth.c
> +++ b/drivers/net/ethernet/renesas/sh_eth.c
> @@ -1098,7 +1098,14 @@ static void sh_eth_ring_free(struct net_device *ndev)
>   /* Free Rx skb ringbuffer */
>   if (mdp->rx_skbuff) {
>   for (i = 0; i < mdp->num_rx_ring; i++)
> + memcpy(mdp->rx_skbuff[i]->data,
> +        "Hello, world", 12);
> + msleep(100);
> + for (i = 0; i < mdp->num_rx_ring; i++) {
> + WARN_ON(memcmp(mdp->rx_skbuff[i]->data,
> +        "Hello, world", 12));
>   dev_kfree_skb(mdp->rx_skbuff[i]);
> + }
>   }
>   kfree(mdp->rx_skbuff);
>   mdp->rx_skbuff = NULL;

then ran the loop:

    while ethtool -G eth0 rx 128 ; ethtool -G eth0 rx 64; do echo -n .; done

and 'ping -f' toward the sh_eth port from another machine.  The
warning fired several times a minute.

To fix these issues:

- Deactivate all TX descriptors rather than writing to EDTRR
- As there seems to be no way of telling when RX DMA is stopped,
  perform a soft reset to ensure that both DMA enginess are stopped
- To reduce the possibility of the reset truncating a transmitted
  frame, disable egress and wait a reasonable time to reach a
  packet boundary before resetting
- Update statistics before resetting

(The 'reasonable time' does not allow for CS/CD in half-duplex
mode, but half-duplex no longer seems reasonable!)

Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>