This adds the r600 KMS + CS support to the Linux kernel.

The r600 TTM support is quite basic and still needs more
work esp around using interrupts, but the polled fencing
should work okay for now.

Also currently TTM is using memcpy to do VRAM moves,
the code is here to use a 3D blit to do this, but
isn't fully debugged yet.

Authors:
Alex Deucher <alexdeucher@gmail.com>
Dave Airlie <airlied@redhat.com>
Jerome Glisse <jglisse@redhat.com>
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This ports the tv-out code from the DDX to KMS.

adds a radeon.tv module option, radeon.tv=0 to disable tv
Signed-off-by: NDave Airlie <airlied@redhat.com>

This adds the command stream checker for the RN50, R100 and R200 cards.

It stops any access to 3D registers on RN50, and does checks
on buffer sizes on the r100/r200 cards. It also fixes some texture
sizing checks on r300.
Signed-off-by: NDave Airlie <airlied@redhat.com>

The new code adds modes in the helper, which makes more sense
I disliked the non-driver code adding modes.
Signed-off-by: NDave Airlie <airlied@redhat.com>

Add the default mode for every output device when there
is no mode for it.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Add a function that can be used to add the default mode for the output device
without EDID.
It will add the default mode that meets with the requirements of given
hdisplay/vdisplay limit.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

When we need to add the standard timing mode, we will firstly check whether it
can be found in DMT table by comparing the hdisplay/vdisplay/vfresh_rate.
If it can't be found, then we will use the cvt/gtf to add the required mode.
If it can be found, it will be returned.

At the same time the function of drm_mode_vrefresh is also fixed. It will
return the result of actual refresh_rate plus 0.5.
For example:
When the calculated value is 84.9, then the fresh_rate is 85.
When the calculated value is 70.02, then the fresh_rate is 70.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

When we add a standard timing mode in UMS, we will first check whether it can
be found in default mode table. If it can't be found, then we will use cvt/gtf
to add the standard timing mode.
Add the default mode table so that we can check whether the given mode
can be found in the default mode table as what we have done in UMS mode.

If the status of one output device is connected but there is no EDID, it will
have no correct mode. In such case we can add some default modes for it. Of
course we only add the modes in the default modes list that visible part is not
greater than 1024x768.

The default mode is autogenerated from the DMT spec. And it is copied from
xserver/hw/xfree86/modes/xf86EdidModes.c. But the mode with reduced blank
feature is removed.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

I really don't want to have core drm module rely on CONFIG_FB,
so this is the easiest answer.
Signed-off-by: NDave Airlie <airlied@redhat.com>

Fixes up the DISCARD + 2 sided stencil in the new generator scripts.
Signed-off-by: NDave Airlie <airlied@redhat.com>

This could be used to bypass CS checks.
Signed-off-by: NDave Airlie <airlied@redhat.com>

Fixes (again) whole-system lockups due to GPU lockups.
Signed-off-by: NEric Anholt <eric@anholt.net>

According to the docs, the ringbuffer is not allowed to wrap in the middle
of an instruction.

G45 PRM, Vol 1b, p101:
  While the “free space” wrap may allow commands to be wrapped around the
  end of the Ring Buffer, the wrap should only occur between commands.
  Padding (with NOP) may be required to follow this restriction.

Do as commanded.

[Having seen bug reports where there is evidence of split commands, but
apparently the GPU has continued on merrily before a bizarre and untimely
death, this may or may not fix a few random hangs.]
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
CC: Eric Anholt <eric@anholt.net>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

mac Mini's have a single DDC line on the DVI connector, shared between the
analog link and the digital link. So, if DDC isn't detected on GPIOE (the
usual SDVO DDC link), try GPIOA (the usual VGA DDC link) when there isn't a
VGA monitor connected.
Signed-off-by: NKeith Packard <keithp@keithp.com>
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

It seems that on IGDNG the same swizzling setup always applys.
And front buffer tiling needs to set address swizzle in display
arb control too.

Fix plane tricle feed setting in v1 which should be disable bit,
and always setup address swizzle to let hardware care for buffer
tiling in all cases.
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NEric Anholt <eric@anholt.net>

And clean up a small whitespace goof-up in the same function, while
I was looking at it.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NEric Anholt <eric@anholt.net>
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>

Signed-off-by: NKyle McMartin <kyle@redhat.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

There are several sources of unnecessary power consumption on Intel
graphics systems. The first is the LVDS clock. TFTs don't suffer from
persistence issues like CRTs, and so we can reduce the LVDS refresh rate
when the screen is idle. It will be automatically upclocked when
userspace triggers graphical activity. Beyond that, we can enable memory
self refresh. This allows the memory to go into a lower power state when
the graphics are idle. Finally, we can drop some clocks on the gpu
itself. All of these things can be reenabled between frames when GPU
activity is triggered, and so there should be no user visible graphical
changes.
Signed-off-by: NJesse Barnes <jesse.barnes@intel.com>
Signed-off-by: NMatthew Garrett <mjg@redhat.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

In the event that any one of the DAC analog outputs (R,G,B) were driven
at full-scale (white video) or some analog level close to full-scale
voltage, and if the video cable were then disconnected, the analog video
voltage level would exceed the maximum electrical overstress limit of the
native (thin-oxide) transistors thus causing a long-term reliability concern.
The electrical overstress condition occurs in this particular case.

This patch address the IGD EOS (electrical overstress condition) issue.
When the EOS interrupt occurs, OS should disable DAC and then disable EOS,
then the normal hotplug operation follows.

TODO: it appears the normal unplug interrupt is missed as reported by Li Peng,
need more checks here.
Signed-off-by: NShaohua Li <shaohua.li@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Currently SDVO TV only support NTSC-M format. In this patch
we introduce PAL and SECAM formats available and create seting-format
property at init time. When user dynamically chose preferred
format by xrandr command, it will refine all modelines
provided by SDVO device, then instruct SDVO device to execute.
At the same time the property is added for SDVO-TV so that the SDVO-TV mode can be changed
by using xrandr.

https://bugs.freedesktop.org/show_bug.cgi?id=22891Signed-off-by: NMa Ling <ling.ma@intel.com>
review-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

For integrated TV there are 3 connector types: S-VIDEO, Composite and
Component(YprPb). Those tv formats whose component flag is true should
be assigned to Component connector, others are for S-VIDEO and Composite.
The patch intends to find appropriate tv format for each connector.
In such case it will return the correct modeline to user space. Otherwise
it will return the incorrect modeline when S-video/composite is connected.
Signed-off-by: NMa Ling <ling.ma@intel.com>
reviewed-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Add a debugfs file to dump the entire register range. Here we
assume that reading write-only/reserved registers won't make the chip
angry. Seems to hold true, thankfully.
Signed-off-by: NBen Gamari <bgamari.foss@gmail.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NBen Gamari <bgamari.foss@gmail.com>
[anholt: hand-applied for conflicts]
Signed-off-by: NEric Anholt <eric@anholt.net>

Remember to release the local reference if we fail to wait on
the rendering.

(Also whilst in the vicinity add some whitespace so that the phasing of
the operations is clearer.)
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NEric Anholt <eric@anholt.net>

Some i915/i945 platforms have a fairly high memory latency in certain
situations, so increase our constant a bit to avoid FIFO underruns.
The effect should be positive on other platforms as well; we'll have a
bit more insurance against a busy memory subsystem due to the extra
FIFO entries.

Fixes fdo bug #23368.  Needed for 2.6.31.
Tested-by: NSven Arvidsson <sa@whiz.se>
Tested-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NMaarten Maathuis <madman2003@gmail.com>
Signed-off-by: NDave Airlie <airlied@linux.ie>

We only want to NULL encoder->crtc when it is off.
Signed-off-by: NDave Airlie <airlied@redhat.com>

For shared tv-out and VGA encoders, we really need to know if
the encoder is just being switched off temporarily in blanking
or if we are really disabling it hard.

Also we need to try harder to disconnect encoders from unused
connectors so we can share more efficently.

(shared encoders stuff is coming in radeon tv-out support)
Signed-off-by: NDave Airlie <airlied@redhat.com>

This produces a warn on for architectures where this gets called
but we don't have a cache flushing implementation suitable.
Signed-off-by: NDave Airlie <airlied@redhat.com>

This adds code to the drm_mm to talk to debugfs, and adds
support to radeon to add the VRAM and GTT mm lists to debugfs.

I tested with spinlock debugging and it doesn't give out.
Signed-off-by: NDave Airlie <airlied@redhat.com>

Set accel to none, we really don't want anyone thinking
fb is an accel interface.
Pass pitch not depth to function for intel.
Signed-off-by: NDave Airlie <airlied@redhat.com>

airlied: fixup race against drm info by filling out
tmp before adding it to proc.
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Sometimes we can obtain the EDID with multiple blocks from the display device.
For example: HDMI monitor.
When the CEA-EDID block is detected, we should also parse the detailed timing
info from it. Otherwise we will lose some modes for the display device.

The first step is check whether the CEA EDID block is found. If it exists,
it will skip the CEA-data block and parse the detailed timing info.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Initially I always meant this code to be shared, but things
ran away from me before I got to it.

This refactors the i915 and radeon kms fbdev interaction layers
out into generic helpers + driver specific pieces.

It moves all the panic/sysrq enhancements to the core file,
and stores a linked list of kernel fbs. This could possibly be
improved to only store the fb which has fbcon on it for panics
etc.

radeon retains some specific codes used for a big endian
workaround.

changes:
fix oops in v1
fix freeing path for crtc_info
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Almost all r128's private ioctls require that the CCE state has
already been initialised.  However, most do not test that this has
been done, and will proceed to dereference a null pointer.  This may
result in a security vulnerability, since some ioctls are
unprivileged.

This adds a macro for the common initialisation test and changes all
ioctl implementations that require prior initialisation to use that
macro.

Also, r128_do_init_cce() does not test that the CCE state has not
been initialised already.  Repeated initialisation may lead to a crash
or resource leak.  This adds that test.
Signed-off-by: NBen Hutchings <ben@decadent.org.uk>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Loosely based on a patch by
Jaswinder Singh Rajput <jaswinderlinux@gmail.com>.

KMS support by Dave Airlie <airlied@redhat.com>.

For Radeon 100- to 500-series, firmware blobs look like:
    struct {
        __be32 datah;
        __be32 datal;
    } cp_ucode[256];

For Radeon 600-series, there are two separate firmware blobs:
    __be32 me_ucode[PM4_UCODE_SIZE * 3];
    __be32 pfp_ucode[PFP_UCODE_SIZE];

For Radeon 700-series, likewise:
    __be32 me_ucode[R700_PM4_UCODE_SIZE];
    __be32 pfp_ucode[R700_PFP_UCODE_SIZE];
Signed-off-by: NBen Hutchings <ben@decadent.org.uk>
Signed-off-by: NDave Airlie <airlied@redhat.com>