2005-04-17 05:20:36 +07:00
|
|
|
/* i915_drv.h -- Private header for the I915 driver -*- linux-c -*-
|
|
|
|
*/
|
2006-01-02 16:14:23 +07:00
|
|
|
/*
|
2005-06-23 19:46:46 +07:00
|
|
|
*
|
2005-04-17 05:20:36 +07:00
|
|
|
* Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
|
|
|
|
* All Rights Reserved.
|
2005-06-23 19:46:46 +07:00
|
|
|
*
|
|
|
|
* Permission is hereby granted, free of charge, to any person obtaining a
|
|
|
|
* copy of this software and associated documentation files (the
|
|
|
|
* "Software"), to deal in the Software without restriction, including
|
|
|
|
* without limitation the rights to use, copy, modify, merge, publish,
|
|
|
|
* distribute, sub license, and/or sell copies of the Software, and to
|
|
|
|
* permit persons to whom the Software is furnished to do so, subject to
|
|
|
|
* the following conditions:
|
|
|
|
*
|
|
|
|
* The above copyright notice and this permission notice (including the
|
|
|
|
* next paragraph) shall be included in all copies or substantial portions
|
|
|
|
* of the Software.
|
|
|
|
*
|
|
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
|
|
|
|
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
|
|
|
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
|
|
|
|
* IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
|
|
|
|
* ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
|
|
|
|
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
|
|
|
|
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
|
|
|
*
|
2006-01-02 16:14:23 +07:00
|
|
|
*/
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
#ifndef _I915_DRV_H_
|
|
|
|
#define _I915_DRV_H_
|
|
|
|
|
2012-12-04 04:03:14 +07:00
|
|
|
#include <uapi/drm/i915_drm.h>
|
|
|
|
|
2008-07-30 01:54:06 +07:00
|
|
|
#include "i915_reg.h"
|
DRM: i915: add mode setting support
This commit adds i915 driver support for the DRM mode setting APIs.
Currently, VGA, LVDS, SDVO DVI & VGA, TV and DVO LVDS outputs are
supported. HDMI, DisplayPort and additional SDVO output support will
follow.
Support for the mode setting code is controlled by the new 'modeset'
module option. A new config option, CONFIG_DRM_I915_KMS controls the
default behavior, and whether a PCI ID list is built into the module for
use by user level module utilities.
Note that if mode setting is enabled, user level drivers that access
display registers directly or that don't use the kernel graphics memory
manager will likely corrupt kernel graphics memory, disrupt output
configuration (possibly leading to hangs and/or blank displays), and
prevent panic/oops messages from appearing. So use caution when
enabling this code; be sure your user level code supports the new
interfaces.
A new SysRq key, 'g', provides emergency support for switching back to
the kernel's framebuffer console; which is useful for testing.
Co-authors: Dave Airlie <airlied@linux.ie>, Hong Liu <hong.liu@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2008-11-08 05:24:08 +07:00
|
|
|
#include "intel_bios.h"
|
2010-05-21 08:08:55 +07:00
|
|
|
#include "intel_ringbuffer.h"
|
2008-10-31 09:38:48 +07:00
|
|
|
#include <linux/io-mapping.h>
|
2010-07-21 05:44:45 +07:00
|
|
|
#include <linux/i2c.h>
|
2012-02-28 06:43:09 +07:00
|
|
|
#include <linux/i2c-algo-bit.h>
|
2010-08-25 03:18:41 +07:00
|
|
|
#include <drm/intel-gtt.h>
|
2011-08-12 17:11:33 +07:00
|
|
|
#include <linux/backlight.h>
|
2012-04-06 04:47:36 +07:00
|
|
|
#include <linux/intel-iommu.h>
|
2012-04-27 20:17:39 +07:00
|
|
|
#include <linux/kref.h>
|
drm/i915: irq-drive the dp aux communication
At least on the platforms that have a dp aux irq and also have it
enabled - vlvhsw should have one, too. But I don't have a machine to
test this on. Judging from docs there's no dp aux interrupt for gm45.
Also, I only have an ivb cpu edp machine, so the dp aux A code for
snb/ilk is untested.
For dpcd probing when nothing is connected it slashes about 5ms of cpu
time (cpu time is now negligible), which agrees with 3 * 5 400 usec
timeouts.
A previous version of this patch increases the time required to go
through the dp_detect cycle (which includes reading the edid) from
around 33 ms to around 40 ms. Experiments indicated that this is
purely due to the irq latency - the hw doesn't allow us to queue up
dp aux transactions and hence irq latency directly affects throughput.
gmbus is much better, there we have a 8 byte buffer, and we get the
irq once another 4 bytes can be queued up.
But by using the pm_qos interface to request the lowest possible cpu
wake-up latency this slowdown completely disappeared.
Since all our output detection logic is single-threaded with the
mode_config mutex right now anyway, I've decide not ot play fancy and
to just reuse the gmbus wait queue. But this would definitely prep the
way to run dp detection on different ports in parallel
v2: Add a timeout for dp aux transfers when using interrupts - the hw
_does_ prevent this with the hw-based 400 usec timeout, but if the
irq somehow doesn't arrive we're screwed. Lesson learned while
developing this ;-)
v3: While at it also convert the busy-loop to wait_for_atomic, so that
we don't run the risk of an infinite loop any more.
v4: Ensure we have the smallest possible irq latency by using the
pm_qos interface.
v5: Add a comment to the code to explain why we frob pm_qos. Suggested
by Chris Wilson.
v6: Disable dp irq for vlv, that's easier than trying to get at docs
and hw.
v7: Squash in a fix for Haswell that Paulo Zanoni tracked down - the
dp aux registers aren't at a fixed offset any more, but can be on the
PCH while the DP port is on the cpu die.
Reviewed-by: Imre Deak <imre.deak@intel.com> (v6)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-01 19:53:48 +07:00
|
|
|
#include <linux/pm_qos.h>
|
2008-07-30 01:54:06 +07:00
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/* General customization:
|
|
|
|
*/
|
|
|
|
|
|
|
|
#define DRIVER_AUTHOR "Tungsten Graphics, Inc."
|
|
|
|
|
|
|
|
#define DRIVER_NAME "i915"
|
|
|
|
#define DRIVER_DESC "Intel Graphics"
|
2008-07-31 02:06:12 +07:00
|
|
|
#define DRIVER_DATE "20080730"
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2008-08-26 05:11:06 +07:00
|
|
|
enum pipe {
|
2013-10-31 23:55:49 +07:00
|
|
|
INVALID_PIPE = -1,
|
2008-08-26 05:11:06 +07:00
|
|
|
PIPE_A = 0,
|
|
|
|
PIPE_B,
|
2011-02-08 03:26:52 +07:00
|
|
|
PIPE_C,
|
drm/i915: Reorganize display pipe register accesses
RFCv2: Reorganize array indexing so that full offsets can be used as
is. It makes grepping for registers in i915_reg.h much easier. Also
move offset arrays to intel_device_info.
v1: Fixed offsets for VLV, proper eDP handling
v2: Fixed BCLRPAT, PIPESRC, PIPECONF and DSP* macros.
v3: Added EDP pipe comment, removed redundant offset arrays for
MSA_MISC and DDI_FUNC_CTL.
v4: Rename patch and report object size increase.
v5: Change location of commas, add PIPE_EDP into enum pipe
v6: Insert PIPE_EDP_OFFSET into pipe offset array
v7: Set I915_MAX_PIPES back to 3, change more registers accessors
to use the new macros, get rid of _PIPE_INC and add dev_priv
as a parameter where required by the new macros.
Upcoming hardware will not have the various display pipe register
ranges evenly spaced in memory. Change register address calculations
into array lookups.
Tested on SNB, VLV, IVB, Gen2 and HSW w/eDP.
I left the UMS cruft untouched.
Size differences:
text data bss dec hex filename
596431 4634 56 601121 92c21 i915.ko (new)
593199 4634 56 597889 91f81 i915.ko (old)
Signed-off-by: Antti Koskipaa <antti.koskipaa@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-02-04 19:22:24 +07:00
|
|
|
_PIPE_EDP,
|
|
|
|
I915_MAX_PIPES = _PIPE_EDP
|
2008-08-26 05:11:06 +07:00
|
|
|
};
|
2011-02-08 03:26:52 +07:00
|
|
|
#define pipe_name(p) ((p) + 'A')
|
2008-08-26 05:11:06 +07:00
|
|
|
|
drm/i915: add TRANSCODER_EDP
Before Haswell we used to have the CPU pipes and the PCH transcoders.
We had the same amount of pipes and transcoders, and there was a 1:1
mapping between them. After Haswell what we used to call CPU pipe was
split into CPU pipe and CPU transcoder. So now we have 3 CPU pipes (A,
B and C), 4 CPU transcoders (A, B, C and EDP) and 1 PCH transcoder
(only used for VGA).
For all the outputs except for EDP we have an 1:1 mapping on the CPU
pipes and CPU transcoders, so if you're using CPU pipe A you have to
use CPU transcoder A. When have an eDP output you have to use
transcoder EDP and you can attach this CPU transcoder to any of the 3
CPU pipes. When using VGA you need to select a pair of matching CPU
pipes/transcoders (A/A, B/B, C/C) and you also need to enable/use the
PCH transcoder.
For now we're just creating the cpu_transcoder definitions and setting
cpu_transcoder to TRANSCODER_EDP on DDI eDP code, but none of the
registers was ported to use transcoder instead of pipe. The goal is to
keep the code backwards-compatible since on all cases except when
using eDP we must have pipe == cpu_transcoder.
V2: Comment the haswell_crtc_off chunk, suggested by Damien Lespiau
and Daniel Vetter.
We currently need the haswell_crtc_off chunk because TRANSCODER_EDP
can be used by any CRTC, so when you stop using it you have to stop
saying you're using it, otherwise you may have at some point 2 CRTCs
claiming they're using TRANSCODER_EDP (a disabled CRTC and an enabled
one), then the HW state readout code will get completely confused.
In other words:
Imagine the following case:
xrandr --output eDP1 --auto --crtc 0
xrandr --output eDP1 --off
xrandr --output eDP1 --auto --crtc 2
After the last command you could get a "pipe A assertion failure
(expected off, current on)" because CRTC 0 still claims it's using
TRANSCODER_EDP, so the HW state readout function will read it
(through PIPECONF) and expect it to be off, when it's actually on
because it's being used by CRTC 2.
So when we make "intel_crtc->cpu_transcoder = intel_crtc->pipe" we
make sure we're pointing to our own original CRTC which is certainly
not used by any other CRTC.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-10-25 00:59:34 +07:00
|
|
|
enum transcoder {
|
|
|
|
TRANSCODER_A = 0,
|
|
|
|
TRANSCODER_B,
|
|
|
|
TRANSCODER_C,
|
drm/i915: Reorganize display pipe register accesses
RFCv2: Reorganize array indexing so that full offsets can be used as
is. It makes grepping for registers in i915_reg.h much easier. Also
move offset arrays to intel_device_info.
v1: Fixed offsets for VLV, proper eDP handling
v2: Fixed BCLRPAT, PIPESRC, PIPECONF and DSP* macros.
v3: Added EDP pipe comment, removed redundant offset arrays for
MSA_MISC and DDI_FUNC_CTL.
v4: Rename patch and report object size increase.
v5: Change location of commas, add PIPE_EDP into enum pipe
v6: Insert PIPE_EDP_OFFSET into pipe offset array
v7: Set I915_MAX_PIPES back to 3, change more registers accessors
to use the new macros, get rid of _PIPE_INC and add dev_priv
as a parameter where required by the new macros.
Upcoming hardware will not have the various display pipe register
ranges evenly spaced in memory. Change register address calculations
into array lookups.
Tested on SNB, VLV, IVB, Gen2 and HSW w/eDP.
I left the UMS cruft untouched.
Size differences:
text data bss dec hex filename
596431 4634 56 601121 92c21 i915.ko (new)
593199 4634 56 597889 91f81 i915.ko (old)
Signed-off-by: Antti Koskipaa <antti.koskipaa@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-02-04 19:22:24 +07:00
|
|
|
TRANSCODER_EDP,
|
|
|
|
I915_MAX_TRANSCODERS
|
drm/i915: add TRANSCODER_EDP
Before Haswell we used to have the CPU pipes and the PCH transcoders.
We had the same amount of pipes and transcoders, and there was a 1:1
mapping between them. After Haswell what we used to call CPU pipe was
split into CPU pipe and CPU transcoder. So now we have 3 CPU pipes (A,
B and C), 4 CPU transcoders (A, B, C and EDP) and 1 PCH transcoder
(only used for VGA).
For all the outputs except for EDP we have an 1:1 mapping on the CPU
pipes and CPU transcoders, so if you're using CPU pipe A you have to
use CPU transcoder A. When have an eDP output you have to use
transcoder EDP and you can attach this CPU transcoder to any of the 3
CPU pipes. When using VGA you need to select a pair of matching CPU
pipes/transcoders (A/A, B/B, C/C) and you also need to enable/use the
PCH transcoder.
For now we're just creating the cpu_transcoder definitions and setting
cpu_transcoder to TRANSCODER_EDP on DDI eDP code, but none of the
registers was ported to use transcoder instead of pipe. The goal is to
keep the code backwards-compatible since on all cases except when
using eDP we must have pipe == cpu_transcoder.
V2: Comment the haswell_crtc_off chunk, suggested by Damien Lespiau
and Daniel Vetter.
We currently need the haswell_crtc_off chunk because TRANSCODER_EDP
can be used by any CRTC, so when you stop using it you have to stop
saying you're using it, otherwise you may have at some point 2 CRTCs
claiming they're using TRANSCODER_EDP (a disabled CRTC and an enabled
one), then the HW state readout code will get completely confused.
In other words:
Imagine the following case:
xrandr --output eDP1 --auto --crtc 0
xrandr --output eDP1 --off
xrandr --output eDP1 --auto --crtc 2
After the last command you could get a "pipe A assertion failure
(expected off, current on)" because CRTC 0 still claims it's using
TRANSCODER_EDP, so the HW state readout function will read it
(through PIPECONF) and expect it to be off, when it's actually on
because it's being used by CRTC 2.
So when we make "intel_crtc->cpu_transcoder = intel_crtc->pipe" we
make sure we're pointing to our own original CRTC which is certainly
not used by any other CRTC.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-10-25 00:59:34 +07:00
|
|
|
};
|
|
|
|
#define transcoder_name(t) ((t) + 'A')
|
|
|
|
|
2009-09-11 05:28:06 +07:00
|
|
|
enum plane {
|
|
|
|
PLANE_A = 0,
|
|
|
|
PLANE_B,
|
2011-02-08 03:26:52 +07:00
|
|
|
PLANE_C,
|
2009-09-11 05:28:06 +07:00
|
|
|
};
|
2011-02-08 03:26:52 +07:00
|
|
|
#define plane_name(p) ((p) + 'A')
|
2008-11-19 00:30:25 +07:00
|
|
|
|
2014-02-08 02:12:49 +07:00
|
|
|
#define sprite_name(p, s) ((p) * INTEL_INFO(dev)->num_sprites + (s) + 'A')
|
2013-04-17 21:48:51 +07:00
|
|
|
|
2012-03-29 22:32:22 +07:00
|
|
|
enum port {
|
|
|
|
PORT_A = 0,
|
|
|
|
PORT_B,
|
|
|
|
PORT_C,
|
|
|
|
PORT_D,
|
|
|
|
PORT_E,
|
|
|
|
I915_MAX_PORTS
|
|
|
|
};
|
|
|
|
#define port_name(p) ((p) + 'A')
|
|
|
|
|
2013-11-06 13:36:35 +07:00
|
|
|
#define I915_NUM_PHYS_VLV 1
|
|
|
|
|
|
|
|
enum dpio_channel {
|
|
|
|
DPIO_CH0,
|
|
|
|
DPIO_CH1
|
|
|
|
};
|
|
|
|
|
|
|
|
enum dpio_phy {
|
|
|
|
DPIO_PHY0,
|
|
|
|
DPIO_PHY1
|
|
|
|
};
|
|
|
|
|
2013-05-03 22:15:36 +07:00
|
|
|
enum intel_display_power_domain {
|
|
|
|
POWER_DOMAIN_PIPE_A,
|
|
|
|
POWER_DOMAIN_PIPE_B,
|
|
|
|
POWER_DOMAIN_PIPE_C,
|
|
|
|
POWER_DOMAIN_PIPE_A_PANEL_FITTER,
|
|
|
|
POWER_DOMAIN_PIPE_B_PANEL_FITTER,
|
|
|
|
POWER_DOMAIN_PIPE_C_PANEL_FITTER,
|
|
|
|
POWER_DOMAIN_TRANSCODER_A,
|
|
|
|
POWER_DOMAIN_TRANSCODER_B,
|
|
|
|
POWER_DOMAIN_TRANSCODER_C,
|
2013-10-16 21:25:48 +07:00
|
|
|
POWER_DOMAIN_TRANSCODER_EDP,
|
2013-09-16 21:38:30 +07:00
|
|
|
POWER_DOMAIN_VGA,
|
2013-11-25 22:15:28 +07:00
|
|
|
POWER_DOMAIN_AUDIO,
|
2013-10-25 21:36:48 +07:00
|
|
|
POWER_DOMAIN_INIT,
|
2013-10-16 21:25:49 +07:00
|
|
|
|
|
|
|
POWER_DOMAIN_NUM,
|
2013-05-03 22:15:36 +07:00
|
|
|
};
|
|
|
|
|
2013-10-16 21:25:49 +07:00
|
|
|
#define POWER_DOMAIN_MASK (BIT(POWER_DOMAIN_NUM) - 1)
|
|
|
|
|
2013-05-03 22:15:36 +07:00
|
|
|
#define POWER_DOMAIN_PIPE(pipe) ((pipe) + POWER_DOMAIN_PIPE_A)
|
|
|
|
#define POWER_DOMAIN_PIPE_PANEL_FITTER(pipe) \
|
|
|
|
((pipe) + POWER_DOMAIN_PIPE_A_PANEL_FITTER)
|
2013-10-16 21:25:48 +07:00
|
|
|
#define POWER_DOMAIN_TRANSCODER(tran) \
|
|
|
|
((tran) == TRANSCODER_EDP ? POWER_DOMAIN_TRANSCODER_EDP : \
|
|
|
|
(tran) + POWER_DOMAIN_TRANSCODER_A)
|
2013-05-03 22:15:36 +07:00
|
|
|
|
2013-10-16 21:25:49 +07:00
|
|
|
#define HSW_ALWAYS_ON_POWER_DOMAINS ( \
|
|
|
|
BIT(POWER_DOMAIN_PIPE_A) | \
|
|
|
|
BIT(POWER_DOMAIN_TRANSCODER_EDP))
|
2013-11-03 11:07:34 +07:00
|
|
|
#define BDW_ALWAYS_ON_POWER_DOMAINS ( \
|
|
|
|
BIT(POWER_DOMAIN_PIPE_A) | \
|
|
|
|
BIT(POWER_DOMAIN_TRANSCODER_EDP) | \
|
|
|
|
BIT(POWER_DOMAIN_PIPE_A_PANEL_FITTER))
|
2013-10-16 21:25:49 +07:00
|
|
|
|
2013-02-26 00:06:49 +07:00
|
|
|
enum hpd_pin {
|
|
|
|
HPD_NONE = 0,
|
|
|
|
HPD_PORT_A = HPD_NONE, /* PORT_A is internal */
|
|
|
|
HPD_TV = HPD_NONE, /* TV is known to be unreliable */
|
|
|
|
HPD_CRT,
|
|
|
|
HPD_SDVO_B,
|
|
|
|
HPD_SDVO_C,
|
|
|
|
HPD_PORT_B,
|
|
|
|
HPD_PORT_C,
|
|
|
|
HPD_PORT_D,
|
|
|
|
HPD_NUM_PINS
|
|
|
|
};
|
|
|
|
|
2012-12-03 18:49:06 +07:00
|
|
|
#define I915_GEM_GPU_DOMAINS \
|
|
|
|
(I915_GEM_DOMAIN_RENDER | \
|
|
|
|
I915_GEM_DOMAIN_SAMPLER | \
|
|
|
|
I915_GEM_DOMAIN_COMMAND | \
|
|
|
|
I915_GEM_DOMAIN_INSTRUCTION | \
|
|
|
|
I915_GEM_DOMAIN_VERTEX)
|
2010-05-22 03:26:39 +07:00
|
|
|
|
2013-03-14 04:05:41 +07:00
|
|
|
#define for_each_pipe(p) for ((p) = 0; (p) < INTEL_INFO(dev)->num_pipes; (p)++)
|
2011-02-08 03:26:52 +07:00
|
|
|
|
2012-07-05 14:50:24 +07:00
|
|
|
#define for_each_encoder_on_crtc(dev, __crtc, intel_encoder) \
|
|
|
|
list_for_each_entry((intel_encoder), &(dev)->mode_config.encoder_list, base.head) \
|
|
|
|
if ((intel_encoder)->base.crtc == (__crtc))
|
|
|
|
|
2014-02-08 03:48:15 +07:00
|
|
|
#define for_each_connector_on_encoder(dev, __encoder, intel_connector) \
|
|
|
|
list_for_each_entry((intel_connector), &(dev)->mode_config.connector_list, base.head) \
|
|
|
|
if ((intel_connector)->base.encoder == (__encoder))
|
|
|
|
|
2013-06-05 18:34:14 +07:00
|
|
|
struct drm_i915_private;
|
|
|
|
|
2013-06-05 18:34:12 +07:00
|
|
|
enum intel_dpll_id {
|
|
|
|
DPLL_ID_PRIVATE = -1, /* non-shared dpll in use */
|
|
|
|
/* real shared dpll ids must be >= 0 */
|
|
|
|
DPLL_ID_PCH_PLL_A,
|
|
|
|
DPLL_ID_PCH_PLL_B,
|
|
|
|
};
|
|
|
|
#define I915_NUM_PLLS 2
|
|
|
|
|
2013-06-05 18:34:16 +07:00
|
|
|
struct intel_dpll_hw_state {
|
2013-06-05 18:34:20 +07:00
|
|
|
uint32_t dpll;
|
2013-06-05 18:34:28 +07:00
|
|
|
uint32_t dpll_md;
|
2013-06-05 18:34:20 +07:00
|
|
|
uint32_t fp0;
|
|
|
|
uint32_t fp1;
|
2013-06-05 18:34:16 +07:00
|
|
|
};
|
|
|
|
|
2013-06-05 18:34:06 +07:00
|
|
|
struct intel_shared_dpll {
|
2012-04-20 23:11:53 +07:00
|
|
|
int refcount; /* count of number of CRTCs sharing this PLL */
|
|
|
|
int active; /* count of number of active CRTCs (i.e. DPMS on) */
|
|
|
|
bool on; /* is the PLL actually active? Disabled during modeset */
|
2013-06-05 18:34:12 +07:00
|
|
|
const char *name;
|
|
|
|
/* should match the index in the dev_priv->shared_dplls array */
|
|
|
|
enum intel_dpll_id id;
|
2013-06-05 18:34:16 +07:00
|
|
|
struct intel_dpll_hw_state hw_state;
|
2013-06-05 18:34:23 +07:00
|
|
|
void (*mode_set)(struct drm_i915_private *dev_priv,
|
|
|
|
struct intel_shared_dpll *pll);
|
2013-06-05 18:34:14 +07:00
|
|
|
void (*enable)(struct drm_i915_private *dev_priv,
|
|
|
|
struct intel_shared_dpll *pll);
|
|
|
|
void (*disable)(struct drm_i915_private *dev_priv,
|
|
|
|
struct intel_shared_dpll *pll);
|
2013-06-05 18:34:16 +07:00
|
|
|
bool (*get_hw_state)(struct drm_i915_private *dev_priv,
|
|
|
|
struct intel_shared_dpll *pll,
|
|
|
|
struct intel_dpll_hw_state *hw_state);
|
2012-04-20 23:11:53 +07:00
|
|
|
};
|
|
|
|
|
2012-11-29 21:59:36 +07:00
|
|
|
/* Used by dp and fdi links */
|
|
|
|
struct intel_link_m_n {
|
|
|
|
uint32_t tu;
|
|
|
|
uint32_t gmch_m;
|
|
|
|
uint32_t gmch_n;
|
|
|
|
uint32_t link_m;
|
|
|
|
uint32_t link_n;
|
|
|
|
};
|
|
|
|
|
|
|
|
void intel_link_compute_m_n(int bpp, int nlanes,
|
|
|
|
int pixel_clock, int link_clock,
|
|
|
|
struct intel_link_m_n *m_n);
|
|
|
|
|
2012-10-05 22:05:58 +07:00
|
|
|
struct intel_ddi_plls {
|
|
|
|
int spll_refcount;
|
|
|
|
int wrpll1_refcount;
|
|
|
|
int wrpll2_refcount;
|
|
|
|
};
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/* Interface history:
|
|
|
|
*
|
|
|
|
* 1.1: Original.
|
2006-01-02 16:14:23 +07:00
|
|
|
* 1.2: Add Power Management
|
|
|
|
* 1.3: Add vblank support
|
2006-01-25 11:31:43 +07:00
|
|
|
* 1.4: Fix cmdbuffer path, add heap destroy
|
2006-06-24 14:07:34 +07:00
|
|
|
* 1.5: Add vblank pipe configuration
|
2006-10-24 22:05:09 +07:00
|
|
|
* 1.6: - New ioctl for scheduling buffer swaps on vertical blank
|
|
|
|
* - Support vertical blank on secondary display pipe
|
2005-04-17 05:20:36 +07:00
|
|
|
*/
|
|
|
|
#define DRIVER_MAJOR 1
|
2006-10-24 22:05:09 +07:00
|
|
|
#define DRIVER_MINOR 6
|
2005-04-17 05:20:36 +07:00
|
|
|
#define DRIVER_PATCHLEVEL 0
|
|
|
|
|
2010-09-29 22:10:57 +07:00
|
|
|
#define WATCH_LISTS 0
|
2012-07-26 17:49:32 +07:00
|
|
|
#define WATCH_GTT 0
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2008-12-30 17:31:46 +07:00
|
|
|
#define I915_GEM_PHYS_CURSOR_0 1
|
|
|
|
#define I915_GEM_PHYS_CURSOR_1 2
|
|
|
|
#define I915_GEM_PHYS_OVERLAY_REGS 3
|
|
|
|
#define I915_MAX_PHYS_OBJECT (I915_GEM_PHYS_OVERLAY_REGS)
|
|
|
|
|
|
|
|
struct drm_i915_gem_phys_object {
|
|
|
|
int id;
|
|
|
|
struct page **page_list;
|
|
|
|
drm_dma_handle_t *handle;
|
2010-11-09 02:18:58 +07:00
|
|
|
struct drm_i915_gem_object *cur_obj;
|
2008-12-30 17:31:46 +07:00
|
|
|
};
|
|
|
|
|
2008-10-01 02:14:26 +07:00
|
|
|
struct opregion_header;
|
|
|
|
struct opregion_acpi;
|
|
|
|
struct opregion_swsci;
|
|
|
|
struct opregion_asle;
|
|
|
|
|
2008-08-06 01:37:25 +07:00
|
|
|
struct intel_opregion {
|
2012-04-17 04:07:42 +07:00
|
|
|
struct opregion_header __iomem *header;
|
|
|
|
struct opregion_acpi __iomem *acpi;
|
|
|
|
struct opregion_swsci __iomem *swsci;
|
2013-09-02 14:38:59 +07:00
|
|
|
u32 swsci_gbda_sub_functions;
|
|
|
|
u32 swsci_sbcb_sub_functions;
|
2012-04-17 04:07:42 +07:00
|
|
|
struct opregion_asle __iomem *asle;
|
|
|
|
void __iomem *vbt;
|
2011-01-17 02:37:30 +07:00
|
|
|
u32 __iomem *lid_state;
|
2013-10-31 23:55:48 +07:00
|
|
|
struct work_struct asle_work;
|
2008-08-06 01:37:25 +07:00
|
|
|
};
|
2010-08-19 22:09:23 +07:00
|
|
|
#define OPREGION_SIZE (8*1024)
|
2008-08-06 01:37:25 +07:00
|
|
|
|
2010-08-05 02:26:07 +07:00
|
|
|
struct intel_overlay;
|
|
|
|
struct intel_overlay_error_state;
|
|
|
|
|
2008-11-28 11:22:24 +07:00
|
|
|
struct drm_i915_master_private {
|
|
|
|
drm_local_map_t *sarea;
|
|
|
|
struct _drm_i915_sarea *sarea_priv;
|
|
|
|
};
|
2008-11-13 01:03:55 +07:00
|
|
|
#define I915_FENCE_REG_NONE -1
|
2013-04-09 17:02:47 +07:00
|
|
|
#define I915_MAX_NUM_FENCES 32
|
|
|
|
/* 32 fences + sign bit for FENCE_REG_NONE */
|
|
|
|
#define I915_MAX_NUM_FENCE_BITS 6
|
2008-11-13 01:03:55 +07:00
|
|
|
|
|
|
|
struct drm_i915_fence_reg {
|
2010-04-28 16:02:31 +07:00
|
|
|
struct list_head lru_list;
|
2010-11-12 20:53:37 +07:00
|
|
|
struct drm_i915_gem_object *obj;
|
2011-12-14 19:57:08 +07:00
|
|
|
int pin_count;
|
2008-11-13 01:03:55 +07:00
|
|
|
};
|
2008-11-28 11:22:24 +07:00
|
|
|
|
2009-05-31 16:17:17 +07:00
|
|
|
struct sdvo_device_mapping {
|
2010-09-24 18:52:03 +07:00
|
|
|
u8 initialized;
|
2009-05-31 16:17:17 +07:00
|
|
|
u8 dvo_port;
|
|
|
|
u8 slave_addr;
|
|
|
|
u8 dvo_wiring;
|
2010-09-24 18:52:03 +07:00
|
|
|
u8 i2c_pin;
|
2010-04-24 03:07:40 +07:00
|
|
|
u8 ddc_pin;
|
2009-05-31 16:17:17 +07:00
|
|
|
};
|
|
|
|
|
2010-11-21 20:12:35 +07:00
|
|
|
struct intel_display_error_state;
|
|
|
|
|
2009-06-19 06:56:52 +07:00
|
|
|
struct drm_i915_error_state {
|
2012-04-27 20:17:39 +07:00
|
|
|
struct kref ref;
|
2014-01-30 15:19:37 +07:00
|
|
|
struct timeval time;
|
|
|
|
|
|
|
|
/* Generic register state */
|
2009-06-19 06:56:52 +07:00
|
|
|
u32 eir;
|
|
|
|
u32 pgtbl_er;
|
2012-04-27 06:03:00 +07:00
|
|
|
u32 ier;
|
2012-06-05 04:42:52 +07:00
|
|
|
u32 ccid;
|
2013-01-15 19:05:55 +07:00
|
|
|
u32 derrmr;
|
|
|
|
u32 forcewake;
|
2014-01-30 15:19:37 +07:00
|
|
|
u32 error; /* gen6+ */
|
|
|
|
u32 err_int; /* gen7 */
|
|
|
|
u32 done_reg;
|
2014-01-30 15:19:39 +07:00
|
|
|
u32 gac_eco;
|
|
|
|
u32 gam_ecochk;
|
|
|
|
u32 gab_ctl;
|
|
|
|
u32 gfx_mode;
|
2014-01-30 15:19:37 +07:00
|
|
|
u32 extra_instdone[I915_NUM_INSTDONE_REG];
|
2011-02-08 03:26:52 +07:00
|
|
|
u32 pipestat[I915_MAX_PIPES];
|
2014-01-30 15:19:37 +07:00
|
|
|
u64 fence[I915_MAX_NUM_FENCES];
|
|
|
|
struct intel_overlay_error_state *overlay;
|
|
|
|
struct intel_display_error_state *display;
|
|
|
|
|
2012-02-15 18:25:37 +07:00
|
|
|
struct drm_i915_error_ring {
|
2014-01-27 20:52:34 +07:00
|
|
|
bool valid;
|
2014-01-30 15:19:38 +07:00
|
|
|
/* Software tracked state */
|
|
|
|
bool waiting;
|
|
|
|
int hangcheck_score;
|
|
|
|
enum intel_ring_hangcheck_action hangcheck_action;
|
|
|
|
int num_requests;
|
|
|
|
|
|
|
|
/* our own tracking of ring head and tail */
|
|
|
|
u32 cpu_ring_head;
|
|
|
|
u32 cpu_ring_tail;
|
|
|
|
|
|
|
|
u32 semaphore_seqno[I915_NUM_RINGS - 1];
|
|
|
|
|
|
|
|
/* Register state */
|
|
|
|
u32 tail;
|
|
|
|
u32 head;
|
|
|
|
u32 ctl;
|
|
|
|
u32 hws;
|
|
|
|
u32 ipeir;
|
|
|
|
u32 ipehr;
|
|
|
|
u32 instdone;
|
|
|
|
u32 acthd;
|
|
|
|
u32 bbstate;
|
|
|
|
u32 instpm;
|
|
|
|
u32 instps;
|
|
|
|
u32 seqno;
|
|
|
|
u64 bbaddr;
|
|
|
|
u32 fault_reg;
|
|
|
|
u32 faddr;
|
|
|
|
u32 rc_psmi; /* sleep state */
|
|
|
|
u32 semaphore_mboxes[I915_NUM_RINGS - 1];
|
|
|
|
|
2012-02-15 18:25:37 +07:00
|
|
|
struct drm_i915_error_object {
|
|
|
|
int page_count;
|
|
|
|
u32 gtt_offset;
|
|
|
|
u32 *pages[0];
|
2014-01-30 15:19:38 +07:00
|
|
|
} *ringbuffer, *batchbuffer, *ctx, *hws_page;
|
|
|
|
|
2012-02-15 18:25:37 +07:00
|
|
|
struct drm_i915_error_request {
|
|
|
|
long jiffies;
|
|
|
|
u32 seqno;
|
2012-02-15 18:25:38 +07:00
|
|
|
u32 tail;
|
2012-02-15 18:25:37 +07:00
|
|
|
} *requests;
|
2014-01-30 15:19:40 +07:00
|
|
|
|
|
|
|
struct {
|
|
|
|
u32 gfx_mode;
|
|
|
|
union {
|
|
|
|
u64 pdp[4];
|
|
|
|
u32 pp_dir_base;
|
|
|
|
};
|
|
|
|
} vm_info;
|
2012-02-15 18:25:37 +07:00
|
|
|
} ring[I915_NUM_RINGS];
|
2010-02-18 17:24:56 +07:00
|
|
|
struct drm_i915_error_buffer {
|
2011-01-10 04:07:49 +07:00
|
|
|
u32 size;
|
2010-02-18 17:24:56 +07:00
|
|
|
u32 name;
|
2012-07-20 18:41:01 +07:00
|
|
|
u32 rseqno, wseqno;
|
2010-02-18 17:24:56 +07:00
|
|
|
u32 gtt_offset;
|
|
|
|
u32 read_domains;
|
|
|
|
u32 write_domain;
|
2011-10-10 02:52:02 +07:00
|
|
|
s32 fence_reg:I915_MAX_NUM_FENCE_BITS;
|
2010-02-18 17:24:56 +07:00
|
|
|
s32 pinned:2;
|
|
|
|
u32 tiling:2;
|
|
|
|
u32 dirty:1;
|
|
|
|
u32 purgeable:1;
|
2012-02-16 17:03:29 +07:00
|
|
|
s32 ring:4;
|
2013-09-25 16:23:19 +07:00
|
|
|
u32 cache_level:3;
|
2013-08-01 07:00:15 +07:00
|
|
|
} **active_bo, **pinned_bo;
|
2014-01-30 15:19:40 +07:00
|
|
|
|
2013-08-01 07:00:15 +07:00
|
|
|
u32 *active_bo_count, *pinned_bo_count;
|
2009-06-19 06:56:52 +07:00
|
|
|
};
|
|
|
|
|
2013-11-08 21:48:56 +07:00
|
|
|
struct intel_connector;
|
2013-03-27 06:44:50 +07:00
|
|
|
struct intel_crtc_config;
|
2013-03-28 16:42:00 +07:00
|
|
|
struct intel_crtc;
|
2013-06-04 03:40:22 +07:00
|
|
|
struct intel_limit;
|
|
|
|
struct dpll;
|
2013-03-27 06:44:50 +07:00
|
|
|
|
2009-09-22 00:42:27 +07:00
|
|
|
struct drm_i915_display_funcs {
|
2010-04-23 22:17:39 +07:00
|
|
|
bool (*fbc_enabled)(struct drm_device *dev);
|
2013-12-12 22:27:40 +07:00
|
|
|
void (*enable_fbc)(struct drm_crtc *crtc);
|
2009-09-22 00:42:27 +07:00
|
|
|
void (*disable_fbc)(struct drm_device *dev);
|
|
|
|
int (*get_display_clock_speed)(struct drm_device *dev);
|
|
|
|
int (*get_fifo_size)(struct drm_device *dev, int plane);
|
2013-06-04 03:40:22 +07:00
|
|
|
/**
|
|
|
|
* find_dpll() - Find the best values for the PLL
|
|
|
|
* @limit: limits for the PLL
|
|
|
|
* @crtc: current CRTC
|
|
|
|
* @target: target frequency in kHz
|
|
|
|
* @refclk: reference clock frequency in kHz
|
|
|
|
* @match_clock: if provided, @best_clock P divider must
|
|
|
|
* match the P divider from @match_clock
|
|
|
|
* used for LVDS downclocking
|
|
|
|
* @best_clock: best PLL values found
|
|
|
|
*
|
|
|
|
* Returns true on success, false on failure.
|
|
|
|
*/
|
|
|
|
bool (*find_dpll)(const struct intel_limit *limit,
|
|
|
|
struct drm_crtc *crtc,
|
|
|
|
int target, int refclk,
|
|
|
|
struct dpll *match_clock,
|
|
|
|
struct dpll *best_clock);
|
2013-09-10 15:40:40 +07:00
|
|
|
void (*update_wm)(struct drm_crtc *crtc);
|
2013-08-07 02:24:11 +07:00
|
|
|
void (*update_sprite_wm)(struct drm_plane *plane,
|
|
|
|
struct drm_crtc *crtc,
|
2013-05-24 21:59:17 +07:00
|
|
|
uint32_t sprite_width, int pixel_size,
|
2013-07-05 15:57:13 +07:00
|
|
|
bool enable, bool scaled);
|
2012-10-26 15:58:18 +07:00
|
|
|
void (*modeset_global_resources)(struct drm_device *dev);
|
2013-03-28 16:42:00 +07:00
|
|
|
/* Returns the active state of the crtc, and if the crtc is active,
|
|
|
|
* fills out the pipe-config with the hw state. */
|
|
|
|
bool (*get_pipe_config)(struct intel_crtc *,
|
|
|
|
struct intel_crtc_config *);
|
2011-03-31 03:01:02 +07:00
|
|
|
int (*crtc_mode_set)(struct drm_crtc *crtc,
|
|
|
|
int x, int y,
|
|
|
|
struct drm_framebuffer *old_fb);
|
2012-06-30 03:39:33 +07:00
|
|
|
void (*crtc_enable)(struct drm_crtc *crtc);
|
|
|
|
void (*crtc_disable)(struct drm_crtc *crtc);
|
2012-04-20 23:11:53 +07:00
|
|
|
void (*off)(struct drm_crtc *crtc);
|
drm/i915: pass ELD to HDMI/DP audio driver
Add ELD support for Intel Eaglelake, IbexPeak/Ironlake,
SandyBridge/CougarPoint and IvyBridge/PantherPoint chips.
ELD (EDID-Like Data) describes to the HDMI/DP audio driver the audio
capabilities of the plugged monitor. It's built and passed to audio
driver in 2 steps:
(1) at get_modes time, parse EDID and save ELD to drm_connector.eld[]
(2) at mode_set time, write drm_connector.eld[] to the Transcoder's hw
ELD buffer and set the ELD_valid bit to inform HDMI/DP audio driver
This patch is tested OK on G45/HDMI, IbexPeak/HDMI and IvyBridge/HDMI+DP.
Test scheme: plug in the HDMI/DP monitor, and run
cat /proc/asound/card0/eld*
to check if the monitor name, HDMI/DP type, etc. show up correctly.
Minor imperfection: the GEN5_AUD_CNTL_ST/DIP_Port_Select field always
reads 0 (reserved). Without knowing the port number, I worked it around
by setting the ELD_valid bit for ALL the three ports. It's tested to not
be a problem, because the audio driver will find invalid ELD data and
hence rightfully abort, even when it sees the ELD_valid indicator.
Thanks to Zhenyu and Pierre-Louis for a lot of valuable help and testing.
CC: Zhao Yakui <yakui.zhao@intel.com>
CC: Wang Zhenyu <zhenyu.z.wang@intel.com>
CC: Jeremy Bush <contractfrombelow@gmail.com>
CC: Christopher White <c.white@pulseforce.com>
CC: Pierre-Louis Bossart <pierre-louis.bossart@intel.com>
CC: Paul Menzel <paulepanter@users.sourceforge.net>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
2011-09-05 13:25:34 +07:00
|
|
|
void (*write_eld)(struct drm_connector *connector,
|
2013-10-16 16:34:47 +07:00
|
|
|
struct drm_crtc *crtc,
|
|
|
|
struct drm_display_mode *mode);
|
2011-04-29 04:27:04 +07:00
|
|
|
void (*fdi_link_train)(struct drm_crtc *crtc);
|
2011-04-29 05:04:31 +07:00
|
|
|
void (*init_clock_gating)(struct drm_device *dev);
|
2011-06-16 23:19:13 +07:00
|
|
|
int (*queue_flip)(struct drm_device *dev, struct drm_crtc *crtc,
|
|
|
|
struct drm_framebuffer *fb,
|
2013-07-23 08:49:58 +07:00
|
|
|
struct drm_i915_gem_object *obj,
|
|
|
|
uint32_t flags);
|
2011-06-25 02:19:23 +07:00
|
|
|
int (*update_plane)(struct drm_crtc *crtc, struct drm_framebuffer *fb,
|
|
|
|
int x, int y);
|
2012-12-11 20:05:07 +07:00
|
|
|
void (*hpd_irq_setup)(struct drm_device *dev);
|
2009-09-22 00:42:27 +07:00
|
|
|
/* clock updates for mode set */
|
|
|
|
/* cursor updates */
|
|
|
|
/* render clock increase/decrease */
|
|
|
|
/* display clock increase/decrease */
|
|
|
|
/* pll clock increase/decrease */
|
2013-11-08 21:48:56 +07:00
|
|
|
|
|
|
|
int (*setup_backlight)(struct intel_connector *connector);
|
|
|
|
uint32_t (*get_backlight)(struct intel_connector *connector);
|
|
|
|
void (*set_backlight)(struct intel_connector *connector,
|
|
|
|
uint32_t level);
|
|
|
|
void (*disable_backlight)(struct intel_connector *connector);
|
|
|
|
void (*enable_backlight)(struct intel_connector *connector);
|
2009-09-22 00:42:27 +07:00
|
|
|
};
|
|
|
|
|
2013-07-20 02:36:52 +07:00
|
|
|
struct intel_uncore_funcs {
|
2013-11-23 16:25:42 +07:00
|
|
|
void (*force_wake_get)(struct drm_i915_private *dev_priv,
|
|
|
|
int fw_engine);
|
|
|
|
void (*force_wake_put)(struct drm_i915_private *dev_priv,
|
|
|
|
int fw_engine);
|
2013-10-05 11:22:51 +07:00
|
|
|
|
|
|
|
uint8_t (*mmio_readb)(struct drm_i915_private *dev_priv, off_t offset, bool trace);
|
|
|
|
uint16_t (*mmio_readw)(struct drm_i915_private *dev_priv, off_t offset, bool trace);
|
|
|
|
uint32_t (*mmio_readl)(struct drm_i915_private *dev_priv, off_t offset, bool trace);
|
|
|
|
uint64_t (*mmio_readq)(struct drm_i915_private *dev_priv, off_t offset, bool trace);
|
|
|
|
|
|
|
|
void (*mmio_writeb)(struct drm_i915_private *dev_priv, off_t offset,
|
|
|
|
uint8_t val, bool trace);
|
|
|
|
void (*mmio_writew)(struct drm_i915_private *dev_priv, off_t offset,
|
|
|
|
uint16_t val, bool trace);
|
|
|
|
void (*mmio_writel)(struct drm_i915_private *dev_priv, off_t offset,
|
|
|
|
uint32_t val, bool trace);
|
|
|
|
void (*mmio_writeq)(struct drm_i915_private *dev_priv, off_t offset,
|
|
|
|
uint64_t val, bool trace);
|
2012-07-02 21:51:02 +07:00
|
|
|
};
|
|
|
|
|
2013-07-20 02:36:52 +07:00
|
|
|
struct intel_uncore {
|
|
|
|
spinlock_t lock; /** lock is also taken in irq contexts. */
|
|
|
|
|
|
|
|
struct intel_uncore_funcs funcs;
|
|
|
|
|
|
|
|
unsigned fifo_count;
|
|
|
|
unsigned forcewake_count;
|
2013-08-26 19:46:09 +07:00
|
|
|
|
2013-11-23 16:25:43 +07:00
|
|
|
unsigned fw_rendercount;
|
|
|
|
unsigned fw_mediacount;
|
|
|
|
|
2013-08-26 19:46:09 +07:00
|
|
|
struct delayed_work force_wake_work;
|
2013-07-20 02:36:52 +07:00
|
|
|
};
|
|
|
|
|
2013-04-23 22:37:17 +07:00
|
|
|
#define DEV_INFO_FOR_EACH_FLAG(func, sep) \
|
|
|
|
func(is_mobile) sep \
|
|
|
|
func(is_i85x) sep \
|
|
|
|
func(is_i915g) sep \
|
|
|
|
func(is_i945gm) sep \
|
|
|
|
func(is_g33) sep \
|
|
|
|
func(need_gfx_hws) sep \
|
|
|
|
func(is_g4x) sep \
|
|
|
|
func(is_pineview) sep \
|
|
|
|
func(is_broadwater) sep \
|
|
|
|
func(is_crestline) sep \
|
|
|
|
func(is_ivybridge) sep \
|
|
|
|
func(is_valleyview) sep \
|
|
|
|
func(is_haswell) sep \
|
2013-08-24 06:00:07 +07:00
|
|
|
func(is_preliminary) sep \
|
2013-04-23 22:37:17 +07:00
|
|
|
func(has_fbc) sep \
|
|
|
|
func(has_pipe_cxsr) sep \
|
|
|
|
func(has_hotplug) sep \
|
|
|
|
func(cursor_needs_physical) sep \
|
|
|
|
func(has_overlay) sep \
|
|
|
|
func(overlay_needs_physical) sep \
|
|
|
|
func(supports_tv) sep \
|
2013-04-23 00:40:39 +07:00
|
|
|
func(has_llc) sep \
|
2013-04-23 00:40:41 +07:00
|
|
|
func(has_ddi) sep \
|
|
|
|
func(has_fpga_dbg)
|
2012-08-09 03:01:51 +07:00
|
|
|
|
2013-04-23 00:40:38 +07:00
|
|
|
#define DEFINE_FLAG(name) u8 name:1
|
|
|
|
#define SEP_SEMICOLON ;
|
2012-08-09 03:01:51 +07:00
|
|
|
|
2009-12-17 03:16:16 +07:00
|
|
|
struct intel_device_info {
|
2013-01-24 20:29:28 +07:00
|
|
|
u32 display_mmio_offset;
|
2013-03-14 04:05:41 +07:00
|
|
|
u8 num_pipes:3;
|
2014-02-08 02:12:49 +07:00
|
|
|
u8 num_sprites:2;
|
2010-08-11 15:59:24 +07:00
|
|
|
u8 gen;
|
2013-10-16 00:02:57 +07:00
|
|
|
u8 ring_mask; /* Rings supported by the HW */
|
2013-04-23 00:40:38 +07:00
|
|
|
DEV_INFO_FOR_EACH_FLAG(DEFINE_FLAG, SEP_SEMICOLON);
|
drm/i915: Reorganize display pipe register accesses
RFCv2: Reorganize array indexing so that full offsets can be used as
is. It makes grepping for registers in i915_reg.h much easier. Also
move offset arrays to intel_device_info.
v1: Fixed offsets for VLV, proper eDP handling
v2: Fixed BCLRPAT, PIPESRC, PIPECONF and DSP* macros.
v3: Added EDP pipe comment, removed redundant offset arrays for
MSA_MISC and DDI_FUNC_CTL.
v4: Rename patch and report object size increase.
v5: Change location of commas, add PIPE_EDP into enum pipe
v6: Insert PIPE_EDP_OFFSET into pipe offset array
v7: Set I915_MAX_PIPES back to 3, change more registers accessors
to use the new macros, get rid of _PIPE_INC and add dev_priv
as a parameter where required by the new macros.
Upcoming hardware will not have the various display pipe register
ranges evenly spaced in memory. Change register address calculations
into array lookups.
Tested on SNB, VLV, IVB, Gen2 and HSW w/eDP.
I left the UMS cruft untouched.
Size differences:
text data bss dec hex filename
596431 4634 56 601121 92c21 i915.ko (new)
593199 4634 56 597889 91f81 i915.ko (old)
Signed-off-by: Antti Koskipaa <antti.koskipaa@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-02-04 19:22:24 +07:00
|
|
|
/* Register offsets for the various display pipes and transcoders */
|
|
|
|
int pipe_offsets[I915_MAX_TRANSCODERS];
|
|
|
|
int trans_offsets[I915_MAX_TRANSCODERS];
|
|
|
|
int dpll_offsets[I915_MAX_PIPES];
|
|
|
|
int dpll_md_offsets[I915_MAX_PIPES];
|
|
|
|
int palette_offsets[I915_MAX_PIPES];
|
2009-12-17 03:16:16 +07:00
|
|
|
};
|
|
|
|
|
2013-04-23 00:40:38 +07:00
|
|
|
#undef DEFINE_FLAG
|
|
|
|
#undef SEP_SEMICOLON
|
|
|
|
|
2013-01-25 05:44:55 +07:00
|
|
|
enum i915_cache_level {
|
|
|
|
I915_CACHE_NONE = 0,
|
2013-08-06 19:17:02 +07:00
|
|
|
I915_CACHE_LLC, /* also used for snoopable memory on non-LLC */
|
|
|
|
I915_CACHE_L3_LLC, /* gen7+, L3 sits between the domain specifc
|
|
|
|
caches, eg sampler/render caches, and the
|
|
|
|
large Last-Level-Cache. LLC is coherent with
|
|
|
|
the CPU, but L3 is only visible to the GPU. */
|
2013-08-08 20:41:10 +07:00
|
|
|
I915_CACHE_WT, /* hsw:gt3e WriteThrough for scanouts */
|
2013-01-25 05:44:55 +07:00
|
|
|
};
|
|
|
|
|
2013-04-22 14:53:49 +07:00
|
|
|
typedef uint32_t gen6_gtt_pte_t;
|
|
|
|
|
drm/i915: Create bind/unbind abstraction for VMAs
To sum up what goes on here, we abstract the vma binding, similarly to
the previous object binding. This helps for distinguishing legacy
binding, versus modern binding. To keep the code churn as minimal as
possible, I am leaving in insert_entries(). It serves as the per
platform pte writing basically. bind_vma and insert_entries do share a
lot of similarities, and I did have designs to combine the two, but as
mentioned already... too much churn in an already massive patchset.
What follows are the 3 commits which existed discretely in the original
submissions. Upon rebasing on Broadwell support, it became clear that
separation was not good, and only made for more error prone code. Below
are the 3 commit messages with all their history.
drm/i915: Add bind/unbind object functions to VMA
drm/i915: Use the new vm [un]bind functions
drm/i915: reduce vm->insert_entries() usage
drm/i915: Add bind/unbind object functions to VMA
As we plumb the code with more VM information, it has become more
obvious that the easiest way to deal with bind and unbind is to simply
put the function pointers in the vm, and let those choose the correct
way to handle the page table updates. This change allows many places in
the code to simply be vm->bind, and not have to worry about
distinguishing PPGTT vs GGTT.
Notice that this patch has no impact on functionality. I've decided to
save the actual change until the next patch because I think it's easier
to review that way. I'm happy to squash the two, or let Daniel do it on
merge.
v2:
Make ggtt handle the quirky aliasing ppgtt
Add flags to bind object to support above
Don't ever call bind/unbind directly for PPGTT until we have real, full
PPGTT (use NULLs to assert this)
Make sure we rebind the ggtt if there already is a ggtt binding. This
happens on set cache levels.
Use VMA for bind/unbind (Daniel, Ben)
v3: Reorganize ggtt_vma_bind to be more concise and easier to read
(Ville). Change logic in unbind to only unbind ggtt when there is a
global mapping, and to remove a redundant check if the aliasing ppgtt
exists.
v4: Make the bind function a bit smarter about the cache levels to avoid
unnecessary multiple remaps. "I accept it is a wart, I think unifying
the pin_vma / bind_vma could be unified later" (Chris)
Removed the git notes, and put version info here. (Daniel)
v5: Update the comment to not suck (Chris)
v6:
Move bind/unbind to the VMA. It makes more sense in the VMA structure
(always has, but I was previously lazy). With this change, it will allow
us to keep a distinct insert_entries.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
drm/i915: Use the new vm [un]bind functions
Building on the last patch which created the new function pointers in
the VM for bind/unbind, here we actually put those new function pointers
to use.
Split out as a separate patch to aid in review. I'm fine with squashing
into the previous patch if people request it.
v2: Updated to address the smart ggtt which can do aliasing as needed
Make sure we bind to global gtt when mappable and fenceable. I thought
we could get away without this initialy, but we cannot.
v3: Make the global GTT binding explicitly use the ggtt VM for
bind_vma(). While at it, use the new ggtt_vma helper (Chris)
At this point the original mailing list thread diverges. ie.
v4^:
use target_obj instead of obj for gen6 relocate_entry
vma->bind_vma() can be called safely during pin. So simply do that
instead of the complicated conditionals.
Don't restore PPGTT bound objects on resume path
Bug fix in resume path for globally bound Bos
Properly handle secure dispatch
Rebased on vma bind/unbind conversion
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
drm/i915: reduce vm->insert_entries() usage
FKA: drm/i915: eliminate vm->insert_entries()
With bind/unbind function pointers in place, we no longer need
insert_entries. We could, and want, to remove clear_range, however it's
not totally easy at this point. Since it's used in a couple of place
still that don't only deal in objects: setup, ppgtt init, and restore
gtt mappings.
v2: Don't actually remove insert_entries, just limit its usage. It will
be useful when we introduce gen8. It will always be called from the vma
bind/unbind.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v1)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-12-07 05:10:56 +07:00
|
|
|
/**
|
|
|
|
* A VMA represents a GEM BO that is bound into an address space. Therefore, a
|
|
|
|
* VMA's presence cannot be guaranteed before binding, or after unbinding the
|
|
|
|
* object into/from the address space.
|
|
|
|
*
|
|
|
|
* To make things as simple as possible (ie. no refcounting), a VMA's lifetime
|
|
|
|
* will always be <= an objects lifetime. So object refcounting should cover us.
|
|
|
|
*/
|
|
|
|
struct i915_vma {
|
|
|
|
struct drm_mm_node node;
|
|
|
|
struct drm_i915_gem_object *obj;
|
|
|
|
struct i915_address_space *vm;
|
|
|
|
|
|
|
|
/** This object's place on the active/inactive lists */
|
|
|
|
struct list_head mm_list;
|
|
|
|
|
|
|
|
struct list_head vma_link; /* Link in the object's VMA list */
|
|
|
|
|
|
|
|
/** This vma's place in the batchbuffer or on the eviction list */
|
|
|
|
struct list_head exec_list;
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Used for performing relocations during execbuffer insertion.
|
|
|
|
*/
|
|
|
|
struct hlist_node exec_node;
|
|
|
|
unsigned long exec_handle;
|
|
|
|
struct drm_i915_gem_exec_object2 *exec_entry;
|
|
|
|
|
|
|
|
/**
|
|
|
|
* How many users have pinned this object in GTT space. The following
|
|
|
|
* users can each hold at most one reference: pwrite/pread, pin_ioctl
|
|
|
|
* (via user_pin_count), execbuffer (objects are not allowed multiple
|
|
|
|
* times for the same batchbuffer), and the framebuffer code. When
|
|
|
|
* switching/pageflipping, the framebuffer code has at most two buffers
|
|
|
|
* pinned per crtc.
|
|
|
|
*
|
|
|
|
* In the worst case this is 1 + 1 + 1 + 2*2 = 7. That would fit into 3
|
|
|
|
* bits with absolutely no headroom. So use 4 bits. */
|
|
|
|
unsigned int pin_count:4;
|
|
|
|
#define DRM_I915_GEM_OBJECT_MAX_PIN_COUNT 0xf
|
|
|
|
|
|
|
|
/** Unmap an object from an address space. This usually consists of
|
|
|
|
* setting the valid PTE entries to a reserved scratch page. */
|
|
|
|
void (*unbind_vma)(struct i915_vma *vma);
|
|
|
|
/* Map an object into an address space with the given cache flags. */
|
|
|
|
#define GLOBAL_BIND (1<<0)
|
|
|
|
void (*bind_vma)(struct i915_vma *vma,
|
|
|
|
enum i915_cache_level cache_level,
|
|
|
|
u32 flags);
|
|
|
|
};
|
|
|
|
|
2013-07-17 06:50:05 +07:00
|
|
|
struct i915_address_space {
|
2013-07-17 06:50:06 +07:00
|
|
|
struct drm_mm mm;
|
2013-07-17 06:50:05 +07:00
|
|
|
struct drm_device *dev;
|
2013-07-17 06:50:07 +07:00
|
|
|
struct list_head global_link;
|
2013-07-17 06:50:05 +07:00
|
|
|
unsigned long start; /* Start offset always 0 for dri2 */
|
|
|
|
size_t total; /* size addr space maps (ex. 2GB for ggtt) */
|
|
|
|
|
|
|
|
struct {
|
|
|
|
dma_addr_t addr;
|
|
|
|
struct page *page;
|
|
|
|
} scratch;
|
|
|
|
|
2013-07-17 06:50:08 +07:00
|
|
|
/**
|
|
|
|
* List of objects currently involved in rendering.
|
|
|
|
*
|
|
|
|
* Includes buffers having the contents of their GPU caches
|
|
|
|
* flushed, not necessarily primitives. last_rendering_seqno
|
|
|
|
* represents when the rendering involved will be completed.
|
|
|
|
*
|
|
|
|
* A reference is held on the buffer while on this list.
|
|
|
|
*/
|
|
|
|
struct list_head active_list;
|
|
|
|
|
|
|
|
/**
|
|
|
|
* LRU list of objects which are not in the ringbuffer and
|
|
|
|
* are ready to unbind, but are still in the GTT.
|
|
|
|
*
|
|
|
|
* last_rendering_seqno is 0 while an object is in this list.
|
|
|
|
*
|
|
|
|
* A reference is not held on the buffer while on this list,
|
|
|
|
* as merely being GTT-bound shouldn't prevent its being
|
|
|
|
* freed, and we'll pull it off the list in the free path.
|
|
|
|
*/
|
|
|
|
struct list_head inactive_list;
|
|
|
|
|
2013-07-17 06:50:05 +07:00
|
|
|
/* FIXME: Need a more generic return type */
|
|
|
|
gen6_gtt_pte_t (*pte_encode)(dma_addr_t addr,
|
2013-10-16 23:18:21 +07:00
|
|
|
enum i915_cache_level level,
|
|
|
|
bool valid); /* Create a valid PTE */
|
2013-07-17 06:50:05 +07:00
|
|
|
void (*clear_range)(struct i915_address_space *vm,
|
|
|
|
unsigned int first_entry,
|
2013-10-16 23:21:30 +07:00
|
|
|
unsigned int num_entries,
|
|
|
|
bool use_scratch);
|
2013-07-17 06:50:05 +07:00
|
|
|
void (*insert_entries)(struct i915_address_space *vm,
|
|
|
|
struct sg_table *st,
|
|
|
|
unsigned int first_entry,
|
|
|
|
enum i915_cache_level cache_level);
|
|
|
|
void (*cleanup)(struct i915_address_space *vm);
|
|
|
|
};
|
|
|
|
|
2013-01-18 03:45:15 +07:00
|
|
|
/* The Graphics Translation Table is the way in which GEN hardware translates a
|
|
|
|
* Graphics Virtual Address into a Physical Address. In addition to the normal
|
|
|
|
* collateral associated with any va->pa translations GEN hardware also has a
|
|
|
|
* portion of the GTT which can be mapped by the CPU and remain both coherent
|
|
|
|
* and correct (in cases like swizzling). That region is referred to as GMADR in
|
|
|
|
* the spec.
|
|
|
|
*/
|
|
|
|
struct i915_gtt {
|
2013-07-17 06:50:05 +07:00
|
|
|
struct i915_address_space base;
|
2013-01-25 04:49:57 +07:00
|
|
|
size_t stolen_size; /* Total size of stolen memory */
|
2013-01-18 03:45:15 +07:00
|
|
|
|
|
|
|
unsigned long mappable_end; /* End offset that we can CPU map */
|
|
|
|
struct io_mapping *mappable; /* Mapping to our CPU mappable region */
|
|
|
|
phys_addr_t mappable_base; /* PA of our GMADR */
|
|
|
|
|
|
|
|
/** "Graphics Stolen Memory" holds the global PTEs */
|
|
|
|
void __iomem *gsm;
|
2013-01-19 03:30:31 +07:00
|
|
|
|
|
|
|
bool do_idle_maps;
|
2013-01-25 05:44:55 +07:00
|
|
|
|
2013-06-28 06:30:23 +07:00
|
|
|
int mtrr;
|
2013-01-25 05:44:55 +07:00
|
|
|
|
|
|
|
/* global gtt ops */
|
2013-01-25 04:49:57 +07:00
|
|
|
int (*gtt_probe)(struct drm_device *dev, size_t *gtt_total,
|
2013-02-09 02:32:47 +07:00
|
|
|
size_t *stolen, phys_addr_t *mappable_base,
|
|
|
|
unsigned long *mappable_end);
|
2013-01-18 03:45:15 +07:00
|
|
|
};
|
2013-07-17 06:50:05 +07:00
|
|
|
#define gtt_total_entries(gtt) ((gtt).base.total >> PAGE_SHIFT)
|
2013-01-18 03:45:15 +07:00
|
|
|
|
2012-02-09 23:15:46 +07:00
|
|
|
struct i915_hw_ppgtt {
|
2013-07-17 06:50:05 +07:00
|
|
|
struct i915_address_space base;
|
drm/i915: Add VM to context
Pretty straightforward so far except for the bit about the refcounting.
The PPGTT will potentially be shared amongst multiple contexts. Because
contexts themselves have a refcounted lifecycle, the easiest way to
manage this will be to refcount the PPGTT. To acheive this, we piggy
back off of the existing context refcount, and will increment and
decrement the PPGTT refcount with context creation, and destruction.
To put it more clearly, if context A, and context B both use PPGTT 0, we
can't free the PPGTT until both A, and B are destroyed.
Note that because the PPGTT is permanently pinned (for now), it really
just matters for the PPGTT destruction, as opposed to making space under
memory pressure.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-12-07 05:11:15 +07:00
|
|
|
struct kref ref;
|
drm/i915: Use drm_mm for PPGTT PDEs
When PPGTT support was originally enabled, it was only designed to
support 1 PPGTT. It therefore made sense to simply hide the GGTT space
required to enable this from the drm_mm allocator.
Since we intend to support full PPGTT, which means more than 1, and they
can be created and destroyed ad hoc it will be required to use the
proper allocation techniques we already have.
The first step here is to make the existing single PPGTT use the
allocator.
The astute observer will notice that we are reserving space in the GGTT
for the PDEs for the lifetime of the address space, and would be right
to question whether or not this is a good idea. It does not make a
difference with this current patch only the aliasing PPGTT (indeed the
PDEs should still be hidden from the shrinker). For the future, we are
allocating from top to bottom to avoid using the precious "gtt
space" The GGTT space at that point should only be used for scanout, HW
contexts, ringbuffers, HWSP, PDEs, and a couple of other small buffers
(potentially) used by the kernel. Everything else should be mapped into
a PPGTT. To put the consumption in more tangible terms, it takes
approximately 4 sets of PDEs to equal one 19x10 framebuffer (with no
fancy stride or alignment constraints). 3/4 of the total [average] GGTT
can be used for PDEs, and hopefully never touch the 1/4 that the
framebuffer needs.
The astute, and persistent observer might ask about the page tables
which are also pinned for the address space. This waste is unfortunate.
We use 2MB of memory per address space. We leave wrapping the PDEs as a
real GEM object as a TODO.
v2: Align PDEs to 64b in GTT
Allocate the node dynamically so we can use drm_mm_put_block
Now tested on IGT
Allocate node at the top to avoid fragmentation (Chris)
v3: Use Chris' top down allocator
v4: Embed drm_mm_node into ppgtt struct (Jesse)
Remove hunks which didn't belong (Jesse)
v5: Don't subtract guard page since we now killed the guard page prior
to this patch. (Ben)
v6: Rebased and removed guard page stuff.
Added a chunk to the commit message
Allow adding a context to mappable region
v7: Undo v3, so we can make the drm patch last in the series
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> (v4)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
squash: drm/i915: allow PPGTT to use mappable
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-12-07 05:11:07 +07:00
|
|
|
struct drm_mm_node node;
|
2012-02-09 23:15:46 +07:00
|
|
|
unsigned num_pd_entries;
|
2013-11-05 11:47:32 +07:00
|
|
|
union {
|
|
|
|
struct page **pt_pages;
|
|
|
|
struct page *gen8_pt_pages;
|
|
|
|
};
|
|
|
|
struct page *pd_pages;
|
|
|
|
int num_pd_pages;
|
|
|
|
int num_pt_pages;
|
|
|
|
union {
|
|
|
|
uint32_t pd_offset;
|
|
|
|
dma_addr_t pd_dma_addr[4];
|
|
|
|
};
|
|
|
|
union {
|
|
|
|
dma_addr_t *pt_dma_addr;
|
|
|
|
dma_addr_t *gen8_pt_dma_addr[4];
|
|
|
|
};
|
drm/i915: Convert execbuf code to use vmas
In order to transition more of our code over to using a VMA instead of
an <OBJ, VM> pair - we must have the vma accessible at execbuf time. Up
until now, we've only had a VMA when actually binding an object.
The previous patch helped handle the distinction on bound vs. unbound.
This patch will help us catch leaks, and other issues before we actually
shuffle a bunch of stuff around.
This attempts to convert all the execbuf code to speak in vmas. Since
the execbuf code is very self contained it was a nice isolated
conversion.
The meat of the code is about turning eb_objects into eb_vma, and then
wiring up the rest of the code to use vmas instead of obj, vm pairs.
Unfortunately, to do this, we must move the exec_list link from the obj
structure. This list is reused in the eviction code, so we must also
modify the eviction code to make this work.
WARNING: This patch makes an already hotly profiled path slower. The cost is
unavoidable. In reply to this mail, I will attach the extra data.
v2: Release table lock early, and two a 2 phase vma lookup to avoid
having to use a GFP_ATOMIC. (Chris)
v3: s/obj_exec_list/obj_exec_link/
Updates to address
commit 6d2b888569d366beb4be72cacfde41adee2c25e1
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Wed Aug 7 18:30:54 2013 +0100
drm/i915: List objects allocated from stolen memory in debugfs
v4: Use obj = vma->obj for neatness in some places (Chris)
need_reloc_mappable() should return false if ppgtt (Chris)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Split out prep patches. Also remove a FIXME comment which is
now taken care of.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-08-14 16:38:36 +07:00
|
|
|
|
2013-12-07 05:11:06 +07:00
|
|
|
int (*enable)(struct i915_hw_ppgtt *ppgtt);
|
2013-12-07 05:11:10 +07:00
|
|
|
int (*switch_mm)(struct i915_hw_ppgtt *ppgtt,
|
|
|
|
struct intel_ring_buffer *ring,
|
|
|
|
bool synchronous);
|
2013-12-07 05:11:29 +07:00
|
|
|
void (*debug_dump)(struct i915_hw_ppgtt *ppgtt, struct seq_file *m);
|
2012-02-09 23:15:46 +07:00
|
|
|
};
|
|
|
|
|
2013-06-12 16:35:28 +07:00
|
|
|
struct i915_ctx_hang_stats {
|
|
|
|
/* This context had batch pending when hang was declared */
|
|
|
|
unsigned batch_pending;
|
|
|
|
|
|
|
|
/* This context had batch active when hang was declared */
|
|
|
|
unsigned batch_active;
|
2013-08-30 20:19:28 +07:00
|
|
|
|
|
|
|
/* Time when this context was last blamed for a GPU reset */
|
|
|
|
unsigned long guilty_ts;
|
|
|
|
|
|
|
|
/* This context is banned to submit more work */
|
|
|
|
bool banned;
|
2013-06-12 16:35:28 +07:00
|
|
|
};
|
2012-06-05 04:42:43 +07:00
|
|
|
|
|
|
|
/* This must match up with the value previously used for execbuf2.rsvd1. */
|
|
|
|
#define DEFAULT_CONTEXT_ID 0
|
|
|
|
struct i915_hw_context {
|
2013-04-30 17:30:33 +07:00
|
|
|
struct kref ref;
|
2012-06-05 04:42:43 +07:00
|
|
|
int id;
|
2012-06-05 04:42:46 +07:00
|
|
|
bool is_initialized;
|
drm/i915: Do remaps for all contexts
On both Ivybridge and Haswell, row remapping information is saved and
restored with context. This means, we never actually properly supported
the l3 remapping because our sysfs interface is asynchronous (and not
tied to any context), and the known faulty HW would be reused by the
next context to run.
Not that due to the asynchronous nature of the sysfs entry, there is no
point modifying the registers for the existing context. Instead we set a
flag for all contexts to load the correct remapping information on the
next run. Interested clients can use debugfs to determine whether or not
the row has been remapped.
One could propose at this point that we just do the remapping in the
kernel. I guess since we have to maintain the sysfs interface anyway,
I'm not sure how useful it is, and I do like keeping the policy in
userspace; (it wasn't my original decision to make the
interface the way it is, so I'm not attached).
v2: Force a context switch when we have a remap on the next switch.
(Ville)
Don't let userspace use the interface with disabled contexts.
v3: Don't force a context switch, just let it nop
Improper context slice remap initialization, 1<<1 instead of 1<<i, but I
rewrote it to avoid a second round of confusion.
Error print moved to error path (All Ville)
Added a comment on why the slice remap initialization happens.
CC: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-19 09:03:18 +07:00
|
|
|
uint8_t remap_slice;
|
2012-06-05 04:42:43 +07:00
|
|
|
struct drm_i915_file_private *file_priv;
|
2013-12-07 05:11:02 +07:00
|
|
|
struct intel_ring_buffer *last_ring;
|
2012-06-05 04:42:43 +07:00
|
|
|
struct drm_i915_gem_object *obj;
|
2013-06-12 16:35:28 +07:00
|
|
|
struct i915_ctx_hang_stats hang_stats;
|
drm/i915: Add VM to context
Pretty straightforward so far except for the bit about the refcounting.
The PPGTT will potentially be shared amongst multiple contexts. Because
contexts themselves have a refcounted lifecycle, the easiest way to
manage this will be to refcount the PPGTT. To acheive this, we piggy
back off of the existing context refcount, and will increment and
decrement the PPGTT refcount with context creation, and destruction.
To put it more clearly, if context A, and context B both use PPGTT 0, we
can't free the PPGTT until both A, and B are destroyed.
Note that because the PPGTT is permanently pinned (for now), it really
just matters for the PPGTT destruction, as opposed to making space under
memory pressure.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-12-07 05:11:15 +07:00
|
|
|
struct i915_address_space *vm;
|
2013-09-18 11:12:45 +07:00
|
|
|
|
|
|
|
struct list_head link;
|
2012-06-05 04:42:43 +07:00
|
|
|
};
|
|
|
|
|
2013-06-28 06:30:21 +07:00
|
|
|
struct i915_fbc {
|
|
|
|
unsigned long size;
|
|
|
|
unsigned int fb_id;
|
|
|
|
enum plane plane;
|
|
|
|
int y;
|
|
|
|
|
|
|
|
struct drm_mm_node *compressed_fb;
|
|
|
|
struct drm_mm_node *compressed_llb;
|
|
|
|
|
|
|
|
struct intel_fbc_work {
|
|
|
|
struct delayed_work work;
|
|
|
|
struct drm_crtc *crtc;
|
|
|
|
struct drm_framebuffer *fb;
|
|
|
|
} *fbc_work;
|
|
|
|
|
2013-07-27 23:23:55 +07:00
|
|
|
enum no_fbc_reason {
|
|
|
|
FBC_OK, /* FBC is enabled */
|
|
|
|
FBC_UNSUPPORTED, /* FBC is not supported by this chipset */
|
2013-06-28 06:30:21 +07:00
|
|
|
FBC_NO_OUTPUT, /* no outputs enabled to compress */
|
|
|
|
FBC_STOLEN_TOO_SMALL, /* not enough space for buffers */
|
|
|
|
FBC_UNSUPPORTED_MODE, /* interlace or doublescanned mode */
|
|
|
|
FBC_MODE_TOO_LARGE, /* mode too large for compression */
|
|
|
|
FBC_BAD_PLANE, /* fbc not supported on plane */
|
|
|
|
FBC_NOT_TILED, /* buffer not tiled */
|
|
|
|
FBC_MULTIPLE_PIPES, /* more than one pipe active */
|
|
|
|
FBC_MODULE_PARAM,
|
|
|
|
FBC_CHIP_DEFAULT, /* disabled by default on this chip */
|
|
|
|
} no_fbc_reason;
|
2010-02-06 03:42:41 +07:00
|
|
|
};
|
|
|
|
|
2013-10-04 02:15:06 +07:00
|
|
|
struct i915_psr {
|
|
|
|
bool sink_support;
|
|
|
|
bool source_ok;
|
2013-07-12 04:45:00 +07:00
|
|
|
};
|
2013-06-28 06:30:21 +07:00
|
|
|
|
2010-04-07 15:15:53 +07:00
|
|
|
enum intel_pch {
|
2012-07-04 04:48:16 +07:00
|
|
|
PCH_NONE = 0, /* No PCH present */
|
2010-04-07 15:15:53 +07:00
|
|
|
PCH_IBX, /* Ibexpeak PCH */
|
|
|
|
PCH_CPT, /* Cougarpoint PCH */
|
2012-03-29 22:32:20 +07:00
|
|
|
PCH_LPT, /* Lynxpoint PCH */
|
2013-04-06 03:12:40 +07:00
|
|
|
PCH_NOP,
|
2010-04-07 15:15:53 +07:00
|
|
|
};
|
|
|
|
|
2012-12-01 21:04:24 +07:00
|
|
|
enum intel_sbi_destination {
|
|
|
|
SBI_ICLK,
|
|
|
|
SBI_MPHY,
|
|
|
|
};
|
|
|
|
|
2010-07-20 03:53:12 +07:00
|
|
|
#define QUIRK_PIPEA_FORCE (1<<0)
|
2011-07-13 04:56:22 +07:00
|
|
|
#define QUIRK_LVDS_SSC_DISABLE (1<<1)
|
2012-03-15 21:56:26 +07:00
|
|
|
#define QUIRK_INVERT_BRIGHTNESS (1<<2)
|
2010-07-20 03:53:12 +07:00
|
|
|
|
2010-03-30 12:34:14 +07:00
|
|
|
struct intel_fbdev;
|
2011-07-08 18:22:42 +07:00
|
|
|
struct intel_fbc_work;
|
2010-03-30 12:34:13 +07:00
|
|
|
|
2012-02-15 04:37:19 +07:00
|
|
|
struct intel_gmbus {
|
|
|
|
struct i2c_adapter adapter;
|
2012-11-10 22:58:21 +07:00
|
|
|
u32 force_bit;
|
2012-02-15 04:37:19 +07:00
|
|
|
u32 reg0;
|
2012-02-15 04:37:22 +07:00
|
|
|
u32 gpio_reg;
|
2012-02-28 06:43:09 +07:00
|
|
|
struct i2c_algo_bit_data bit_algo;
|
2012-02-15 04:37:19 +07:00
|
|
|
struct drm_i915_private *dev_priv;
|
|
|
|
};
|
|
|
|
|
2012-11-03 01:55:02 +07:00
|
|
|
struct i915_suspend_saved_registers {
|
2007-11-22 11:14:14 +07:00
|
|
|
u8 saveLBB;
|
|
|
|
u32 saveDSPACNTR;
|
|
|
|
u32 saveDSPBCNTR;
|
2008-05-07 09:27:53 +07:00
|
|
|
u32 saveDSPARB;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 savePIPEACONF;
|
|
|
|
u32 savePIPEBCONF;
|
|
|
|
u32 savePIPEASRC;
|
|
|
|
u32 savePIPEBSRC;
|
|
|
|
u32 saveFPA0;
|
|
|
|
u32 saveFPA1;
|
|
|
|
u32 saveDPLL_A;
|
|
|
|
u32 saveDPLL_A_MD;
|
|
|
|
u32 saveHTOTAL_A;
|
|
|
|
u32 saveHBLANK_A;
|
|
|
|
u32 saveHSYNC_A;
|
|
|
|
u32 saveVTOTAL_A;
|
|
|
|
u32 saveVBLANK_A;
|
|
|
|
u32 saveVSYNC_A;
|
|
|
|
u32 saveBCLRPAT_A;
|
2009-11-06 09:13:02 +07:00
|
|
|
u32 saveTRANSACONF;
|
2009-10-21 14:27:01 +07:00
|
|
|
u32 saveTRANS_HTOTAL_A;
|
|
|
|
u32 saveTRANS_HBLANK_A;
|
|
|
|
u32 saveTRANS_HSYNC_A;
|
|
|
|
u32 saveTRANS_VTOTAL_A;
|
|
|
|
u32 saveTRANS_VBLANK_A;
|
|
|
|
u32 saveTRANS_VSYNC_A;
|
2008-02-20 06:39:58 +07:00
|
|
|
u32 savePIPEASTAT;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveDSPASTRIDE;
|
|
|
|
u32 saveDSPASIZE;
|
|
|
|
u32 saveDSPAPOS;
|
2008-07-30 01:54:06 +07:00
|
|
|
u32 saveDSPAADDR;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveDSPASURF;
|
|
|
|
u32 saveDSPATILEOFF;
|
|
|
|
u32 savePFIT_PGM_RATIOS;
|
2009-10-15 02:33:41 +07:00
|
|
|
u32 saveBLC_HIST_CTL;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveBLC_PWM_CTL;
|
|
|
|
u32 saveBLC_PWM_CTL2;
|
2013-10-31 23:55:50 +07:00
|
|
|
u32 saveBLC_HIST_CTL_B;
|
2009-10-21 14:27:01 +07:00
|
|
|
u32 saveBLC_CPU_PWM_CTL;
|
|
|
|
u32 saveBLC_CPU_PWM_CTL2;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveFPB0;
|
|
|
|
u32 saveFPB1;
|
|
|
|
u32 saveDPLL_B;
|
|
|
|
u32 saveDPLL_B_MD;
|
|
|
|
u32 saveHTOTAL_B;
|
|
|
|
u32 saveHBLANK_B;
|
|
|
|
u32 saveHSYNC_B;
|
|
|
|
u32 saveVTOTAL_B;
|
|
|
|
u32 saveVBLANK_B;
|
|
|
|
u32 saveVSYNC_B;
|
|
|
|
u32 saveBCLRPAT_B;
|
2009-11-06 09:13:02 +07:00
|
|
|
u32 saveTRANSBCONF;
|
2009-10-21 14:27:01 +07:00
|
|
|
u32 saveTRANS_HTOTAL_B;
|
|
|
|
u32 saveTRANS_HBLANK_B;
|
|
|
|
u32 saveTRANS_HSYNC_B;
|
|
|
|
u32 saveTRANS_VTOTAL_B;
|
|
|
|
u32 saveTRANS_VBLANK_B;
|
|
|
|
u32 saveTRANS_VSYNC_B;
|
2008-02-20 06:39:58 +07:00
|
|
|
u32 savePIPEBSTAT;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveDSPBSTRIDE;
|
|
|
|
u32 saveDSPBSIZE;
|
|
|
|
u32 saveDSPBPOS;
|
2008-07-30 01:54:06 +07:00
|
|
|
u32 saveDSPBADDR;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveDSPBSURF;
|
|
|
|
u32 saveDSPBTILEOFF;
|
2008-07-30 01:54:06 +07:00
|
|
|
u32 saveVGA0;
|
|
|
|
u32 saveVGA1;
|
|
|
|
u32 saveVGA_PD;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveVGACNTRL;
|
|
|
|
u32 saveADPA;
|
|
|
|
u32 saveLVDS;
|
2008-07-30 01:54:06 +07:00
|
|
|
u32 savePP_ON_DELAYS;
|
|
|
|
u32 savePP_OFF_DELAYS;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveDVOA;
|
|
|
|
u32 saveDVOB;
|
|
|
|
u32 saveDVOC;
|
|
|
|
u32 savePP_ON;
|
|
|
|
u32 savePP_OFF;
|
|
|
|
u32 savePP_CONTROL;
|
2008-07-30 01:54:06 +07:00
|
|
|
u32 savePP_DIVISOR;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 savePFIT_CONTROL;
|
|
|
|
u32 save_palette_a[256];
|
|
|
|
u32 save_palette_b[256];
|
|
|
|
u32 saveFBC_CONTROL;
|
2008-02-20 06:39:58 +07:00
|
|
|
u32 saveIER;
|
|
|
|
u32 saveIIR;
|
|
|
|
u32 saveIMR;
|
2009-10-21 14:27:01 +07:00
|
|
|
u32 saveDEIER;
|
|
|
|
u32 saveDEIMR;
|
|
|
|
u32 saveGTIER;
|
|
|
|
u32 saveGTIMR;
|
|
|
|
u32 saveFDI_RXA_IMR;
|
|
|
|
u32 saveFDI_RXB_IMR;
|
2008-02-17 10:19:29 +07:00
|
|
|
u32 saveCACHE_MODE_0;
|
|
|
|
u32 saveMI_ARB_STATE;
|
2007-11-22 11:14:14 +07:00
|
|
|
u32 saveSWF0[16];
|
|
|
|
u32 saveSWF1[16];
|
|
|
|
u32 saveSWF2[3];
|
|
|
|
u8 saveMSR;
|
|
|
|
u8 saveSR[8];
|
2008-02-08 02:15:20 +07:00
|
|
|
u8 saveGR[25];
|
2007-11-22 11:14:14 +07:00
|
|
|
u8 saveAR_INDEX;
|
2008-05-07 09:25:46 +07:00
|
|
|
u8 saveAR[21];
|
2007-11-22 11:14:14 +07:00
|
|
|
u8 saveDACMASK;
|
2008-05-07 09:25:46 +07:00
|
|
|
u8 saveCR[37];
|
2011-10-10 02:52:02 +07:00
|
|
|
uint64_t saveFENCE[I915_MAX_NUM_FENCES];
|
2009-06-03 14:26:58 +07:00
|
|
|
u32 saveCURACNTR;
|
|
|
|
u32 saveCURAPOS;
|
|
|
|
u32 saveCURABASE;
|
|
|
|
u32 saveCURBCNTR;
|
|
|
|
u32 saveCURBPOS;
|
|
|
|
u32 saveCURBBASE;
|
|
|
|
u32 saveCURSIZE;
|
2009-04-08 06:16:42 +07:00
|
|
|
u32 saveDP_B;
|
|
|
|
u32 saveDP_C;
|
|
|
|
u32 saveDP_D;
|
|
|
|
u32 savePIPEA_GMCH_DATA_M;
|
|
|
|
u32 savePIPEB_GMCH_DATA_M;
|
|
|
|
u32 savePIPEA_GMCH_DATA_N;
|
|
|
|
u32 savePIPEB_GMCH_DATA_N;
|
|
|
|
u32 savePIPEA_DP_LINK_M;
|
|
|
|
u32 savePIPEB_DP_LINK_M;
|
|
|
|
u32 savePIPEA_DP_LINK_N;
|
|
|
|
u32 savePIPEB_DP_LINK_N;
|
2009-10-21 14:27:01 +07:00
|
|
|
u32 saveFDI_RXA_CTL;
|
|
|
|
u32 saveFDI_TXA_CTL;
|
|
|
|
u32 saveFDI_RXB_CTL;
|
|
|
|
u32 saveFDI_TXB_CTL;
|
|
|
|
u32 savePFA_CTL_1;
|
|
|
|
u32 savePFB_CTL_1;
|
|
|
|
u32 savePFA_WIN_SZ;
|
|
|
|
u32 savePFB_WIN_SZ;
|
|
|
|
u32 savePFA_WIN_POS;
|
|
|
|
u32 savePFB_WIN_POS;
|
2009-11-06 09:13:02 +07:00
|
|
|
u32 savePCH_DREF_CONTROL;
|
|
|
|
u32 saveDISP_ARB_CTL;
|
|
|
|
u32 savePIPEA_DATA_M1;
|
|
|
|
u32 savePIPEA_DATA_N1;
|
|
|
|
u32 savePIPEA_LINK_M1;
|
|
|
|
u32 savePIPEA_LINK_N1;
|
|
|
|
u32 savePIPEB_DATA_M1;
|
|
|
|
u32 savePIPEB_DATA_N1;
|
|
|
|
u32 savePIPEB_LINK_M1;
|
|
|
|
u32 savePIPEB_LINK_N1;
|
2010-02-03 01:30:47 +07:00
|
|
|
u32 saveMCHBAR_RENDER_STANDBY;
|
2011-07-27 03:53:06 +07:00
|
|
|
u32 savePCH_PORT_HOTPLUG;
|
2012-11-03 01:55:02 +07:00
|
|
|
};
|
2012-11-03 01:55:03 +07:00
|
|
|
|
|
|
|
struct intel_gen6_power_mgmt {
|
2013-07-05 04:35:28 +07:00
|
|
|
/* work and pm_iir are protected by dev_priv->irq_lock */
|
2012-11-03 01:55:03 +07:00
|
|
|
struct work_struct work;
|
|
|
|
u32 pm_iir;
|
2013-07-05 04:35:28 +07:00
|
|
|
|
2012-11-03 01:55:03 +07:00
|
|
|
u8 cur_delay;
|
|
|
|
u8 min_delay;
|
|
|
|
u8 max_delay;
|
2013-04-24 00:09:26 +07:00
|
|
|
u8 rpe_delay;
|
drm/i915: Tweak RPS thresholds to more aggressively downclock
After applying wait-boost we often find ourselves stuck at higher clocks
than required. The current threshold value requires the GPU to be
continuously and completely idle for 313ms before it is dropped by one
bin. Conversely, we require the GPU to be busy for an average of 90% over
a 84ms period before we upclock. So the current thresholds almost never
downclock the GPU, and respond very slowly to sudden demands for more
power. It is easy to observe that we currently lock into the wrong bin
and both underperform in benchmarks and consume more power than optimal
(just by repeating the task and measuring the different results).
An alternative approach, as discussed in the bspec, is to use a
continuous threshold for upclocking, and an average value for downclocking.
This is good for quickly detecting and reacting to state changes within a
frame, however it fails with the common throttling method of waiting
upon the outstanding frame - at least it is difficult to choose a
threshold that works well at 15,000fps and at 60fps. So continue to use
average busy/idle loads to determine frequency change.
v2: Use 3 power zones to keep frequencies low in steady-state mostly
idle (e.g. scrolling, interactive 2D drawing), and frequencies high
for demanding games. In between those end-states, we use a
fast-reclocking algorithm to converge more quickly on the desired bin.
v3: Bug fixes - make sure we reset adj after switching power zones.
v4: Tune - drop the continuous busy thresholds as it prevents us from
choosing the right frequency for glxgears style swap benchmarks. Instead
the goal is to be able to find the right clocks irrespective of the
wait-boost.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Stéphane Marchesin <stephane.marchesin@gmail.com>
Cc: Owen Taylor <otaylor@redhat.com>
Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com>
Cc: "Zhuang, Lena" <lena.zhuang@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 23:34:57 +07:00
|
|
|
u8 rp1_delay;
|
|
|
|
u8 rp0_delay;
|
2013-04-06 04:29:22 +07:00
|
|
|
u8 hw_max;
|
2012-11-03 01:14:00 +07:00
|
|
|
|
2014-01-27 23:05:05 +07:00
|
|
|
bool rp_up_masked;
|
|
|
|
bool rp_down_masked;
|
|
|
|
|
drm/i915: Tweak RPS thresholds to more aggressively downclock
After applying wait-boost we often find ourselves stuck at higher clocks
than required. The current threshold value requires the GPU to be
continuously and completely idle for 313ms before it is dropped by one
bin. Conversely, we require the GPU to be busy for an average of 90% over
a 84ms period before we upclock. So the current thresholds almost never
downclock the GPU, and respond very slowly to sudden demands for more
power. It is easy to observe that we currently lock into the wrong bin
and both underperform in benchmarks and consume more power than optimal
(just by repeating the task and measuring the different results).
An alternative approach, as discussed in the bspec, is to use a
continuous threshold for upclocking, and an average value for downclocking.
This is good for quickly detecting and reacting to state changes within a
frame, however it fails with the common throttling method of waiting
upon the outstanding frame - at least it is difficult to choose a
threshold that works well at 15,000fps and at 60fps. So continue to use
average busy/idle loads to determine frequency change.
v2: Use 3 power zones to keep frequencies low in steady-state mostly
idle (e.g. scrolling, interactive 2D drawing), and frequencies high
for demanding games. In between those end-states, we use a
fast-reclocking algorithm to converge more quickly on the desired bin.
v3: Bug fixes - make sure we reset adj after switching power zones.
v4: Tune - drop the continuous busy thresholds as it prevents us from
choosing the right frequency for glxgears style swap benchmarks. Instead
the goal is to be able to find the right clocks irrespective of the
wait-boost.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Stéphane Marchesin <stephane.marchesin@gmail.com>
Cc: Owen Taylor <otaylor@redhat.com>
Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com>
Cc: "Zhuang, Lena" <lena.zhuang@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 23:34:57 +07:00
|
|
|
int last_adj;
|
|
|
|
enum { LOW_POWER, BETWEEN, HIGH_POWER } power;
|
|
|
|
|
2013-10-11 03:58:50 +07:00
|
|
|
bool enabled;
|
2012-11-03 01:14:00 +07:00
|
|
|
struct delayed_work delayed_resume_work;
|
2012-11-03 01:14:01 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Protects RPS/RC6 register access and PCU communication.
|
|
|
|
* Must be taken after struct_mutex if nested.
|
|
|
|
*/
|
|
|
|
struct mutex hw_lock;
|
2012-11-03 01:55:03 +07:00
|
|
|
};
|
|
|
|
|
2012-11-30 04:18:51 +07:00
|
|
|
/* defined intel_pm.c */
|
|
|
|
extern spinlock_t mchdev_lock;
|
|
|
|
|
2012-11-03 01:55:03 +07:00
|
|
|
struct intel_ilk_power_mgmt {
|
|
|
|
u8 cur_delay;
|
|
|
|
u8 min_delay;
|
|
|
|
u8 max_delay;
|
|
|
|
u8 fmax;
|
|
|
|
u8 fstart;
|
|
|
|
|
|
|
|
u64 last_count1;
|
|
|
|
unsigned long last_time1;
|
|
|
|
unsigned long chipset_power;
|
|
|
|
u64 last_count2;
|
|
|
|
struct timespec last_time2;
|
|
|
|
unsigned long gfx_power;
|
|
|
|
u8 corr;
|
|
|
|
|
|
|
|
int c_m;
|
|
|
|
int r_t;
|
2012-11-03 01:55:04 +07:00
|
|
|
|
|
|
|
struct drm_i915_gem_object *pwrctx;
|
|
|
|
struct drm_i915_gem_object *renderctx;
|
2012-11-03 01:55:03 +07:00
|
|
|
};
|
|
|
|
|
2013-05-30 21:07:11 +07:00
|
|
|
/* Power well structure for haswell */
|
|
|
|
struct i915_power_well {
|
2013-11-25 22:15:29 +07:00
|
|
|
const char *name;
|
2013-11-25 22:15:30 +07:00
|
|
|
bool always_on;
|
2013-05-30 21:07:11 +07:00
|
|
|
/* power well enable/disable usage count */
|
|
|
|
int count;
|
2013-11-25 22:15:29 +07:00
|
|
|
unsigned long domains;
|
|
|
|
void *data;
|
|
|
|
void (*set)(struct drm_device *dev, struct i915_power_well *power_well,
|
|
|
|
bool enable);
|
|
|
|
bool (*is_enabled)(struct drm_device *dev,
|
|
|
|
struct i915_power_well *power_well);
|
2013-05-30 21:07:11 +07:00
|
|
|
};
|
|
|
|
|
2013-10-25 21:36:47 +07:00
|
|
|
struct i915_power_domains {
|
2013-10-25 21:36:48 +07:00
|
|
|
/*
|
|
|
|
* Power wells needed for initialization at driver init and suspend
|
|
|
|
* time are on. They are kept on until after the first modeset.
|
|
|
|
*/
|
|
|
|
bool init_power_on;
|
2013-11-25 22:15:29 +07:00
|
|
|
int power_well_count;
|
2013-10-25 21:36:48 +07:00
|
|
|
|
2013-10-25 21:36:47 +07:00
|
|
|
struct mutex lock;
|
2013-11-25 22:15:35 +07:00
|
|
|
int domain_use_count[POWER_DOMAIN_NUM];
|
2013-11-25 22:15:29 +07:00
|
|
|
struct i915_power_well *power_wells;
|
2013-10-25 21:36:47 +07:00
|
|
|
};
|
|
|
|
|
2012-11-03 01:55:05 +07:00
|
|
|
struct i915_dri1_state {
|
|
|
|
unsigned allow_batchbuffer : 1;
|
|
|
|
u32 __iomem *gfx_hws_cpu_addr;
|
|
|
|
|
|
|
|
unsigned int cpp;
|
|
|
|
int back_offset;
|
|
|
|
int front_offset;
|
|
|
|
int current_page;
|
|
|
|
int page_flipping;
|
|
|
|
|
|
|
|
uint32_t counter;
|
|
|
|
};
|
|
|
|
|
2013-07-09 21:51:37 +07:00
|
|
|
struct i915_ums_state {
|
|
|
|
/**
|
|
|
|
* Flag if the X Server, and thus DRM, is not currently in
|
|
|
|
* control of the device.
|
|
|
|
*
|
|
|
|
* This is set between LeaveVT and EnterVT. It needs to be
|
|
|
|
* replaced with a semaphore. It also needs to be
|
|
|
|
* transitioned away from for kernel modesetting.
|
|
|
|
*/
|
|
|
|
int mm_suspended;
|
|
|
|
};
|
|
|
|
|
2013-09-20 01:13:41 +07:00
|
|
|
#define MAX_L3_SLICES 2
|
2012-11-03 01:55:07 +07:00
|
|
|
struct intel_l3_parity {
|
2013-09-20 01:13:41 +07:00
|
|
|
u32 *remap_info[MAX_L3_SLICES];
|
2012-11-03 01:55:07 +07:00
|
|
|
struct work_struct error_work;
|
2013-09-20 01:13:41 +07:00
|
|
|
int which_slice;
|
2012-11-03 01:55:07 +07:00
|
|
|
};
|
|
|
|
|
2012-11-14 23:14:03 +07:00
|
|
|
struct i915_gem_mm {
|
|
|
|
/** Memory allocator for GTT stolen memory */
|
|
|
|
struct drm_mm stolen;
|
|
|
|
/** List of all objects in gtt_space. Used to restore gtt
|
|
|
|
* mappings on resume */
|
|
|
|
struct list_head bound_list;
|
|
|
|
/**
|
|
|
|
* List of objects which are not bound to the GTT (thus
|
|
|
|
* are idle and not used by the GPU) but still have
|
|
|
|
* (presumably uncached) pages still attached.
|
|
|
|
*/
|
|
|
|
struct list_head unbound_list;
|
|
|
|
|
|
|
|
/** Usable portion of the GTT for GEM */
|
|
|
|
unsigned long stolen_base; /* limited to low memory (32-bit) */
|
|
|
|
|
|
|
|
/** PPGTT used for aliasing the PPGTT with the GTT */
|
|
|
|
struct i915_hw_ppgtt *aliasing_ppgtt;
|
|
|
|
|
|
|
|
struct shrinker inactive_shrinker;
|
|
|
|
bool shrinker_no_lock_stealing;
|
|
|
|
|
|
|
|
/** LRU list of objects with fence regs on them. */
|
|
|
|
struct list_head fence_list;
|
|
|
|
|
|
|
|
/**
|
|
|
|
* We leave the user IRQ off as much as possible,
|
|
|
|
* but this means that requests will finish and never
|
|
|
|
* be retired once the system goes idle. Set a timer to
|
|
|
|
* fire periodically while the ring is running. When it
|
|
|
|
* fires, go retire requests.
|
|
|
|
*/
|
|
|
|
struct delayed_work retire_work;
|
|
|
|
|
drm/i915: Boost RPS frequency for CPU stalls
If we encounter a situation where the CPU blocks waiting for results
from the GPU, give the GPU a kick to boost its the frequency.
This should work to reduce user interface stalls and to quickly promote
mesa to high frequencies - but the cost is that our requested frequency
stalls high (as we do not idle for long enough before rc6 to start
reducing frequencies, nor are we aggressive at down clocking an
underused GPU). However, this should be mitigated by rc6 itself powering
off the GPU when idle, and that energy use is dependent upon the workload
of the GPU in addition to its frequency (e.g. the math or sampler
functions only consume power when used). Still, this is likely to
adversely affect light workloads.
In particular, this nearly eliminates the highly noticeable wake-up lag
in animations from idle. For example, expose or workspace transitions.
(However, given the situation where we fail to downclock, our requested
frequency is almost always the maximum, except for Baytrail where we
manually downclock upon idling. This often masks the latency of
upclocking after being idle, so animations are typically smooth - at the
cost of increased power consumption.)
Stéphane raised the concern that this will punish good applications and
reward bad applications - but due to the nature of how mesa performs its
client throttling, I believe all mesa applications will be roughly
equally affected. To address this concern, and to prevent applications
like compositors from permanently boosting the RPS state, we ratelimit the
frequency of the wait-boosts each client recieves.
Unfortunately, this techinique is ineffective with Ironlake - which also
has dynamic render power states and suffers just as dramatically. For
Ironlake, the thermal/power headroom is shared with the CPU through
Intelligent Power Sharing and the intel-ips module. This leaves us with
no GPU boost frequencies available when coming out of idle, and due to
hardware limitations we cannot change the arbitration between the CPU and
GPU quickly enough to be effective.
v2: Limit each client to receiving a single boost for each active period.
Tested by QA to only marginally increase power, and to demonstrably
increase throughput in games. No latency measurements yet.
v3: Cater for front-buffer rendering with manual throttling.
v4: Tidy up.
v5: Sadly the compositor needs frequent boosts as it may never idle, but
due to its picking mechanism (using ReadPixels) may require frequent
waits. Those waits, along with the waits for the vrefresh swap, conspire
to keep the GPU at low frequencies despite the interactive latency. To
overcome this we ditch the one-boost-per-active-period and just ratelimit
the number of wait-boosts each client can receive.
Reported-and-tested-by: Paul Neumann <paul104x@yahoo.de>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=68716
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Stéphane Marchesin <stephane.marchesin@gmail.com>
Cc: Owen Taylor <otaylor@redhat.com>
Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com>
Cc: "Zhuang, Lena" <lena.zhuang@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[danvet: No extern for function prototypes in headers.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 23:34:56 +07:00
|
|
|
/**
|
|
|
|
* When we detect an idle GPU, we want to turn on
|
|
|
|
* powersaving features. So once we see that there
|
|
|
|
* are no more requests outstanding and no more
|
|
|
|
* arrive within a small period of time, we fire
|
|
|
|
* off the idle_work.
|
|
|
|
*/
|
|
|
|
struct delayed_work idle_work;
|
|
|
|
|
2012-11-14 23:14:03 +07:00
|
|
|
/**
|
|
|
|
* Are we in a non-interruptible section of code like
|
|
|
|
* modesetting?
|
|
|
|
*/
|
|
|
|
bool interruptible;
|
|
|
|
|
|
|
|
/** Bit 6 swizzling required for X tiling */
|
|
|
|
uint32_t bit_6_swizzle_x;
|
|
|
|
/** Bit 6 swizzling required for Y tiling */
|
|
|
|
uint32_t bit_6_swizzle_y;
|
|
|
|
|
|
|
|
/* storage for physical objects */
|
|
|
|
struct drm_i915_gem_phys_object *phys_objs[I915_MAX_PHYS_OBJECT];
|
|
|
|
|
|
|
|
/* accounting, useful for userland debugging */
|
2013-07-25 03:40:23 +07:00
|
|
|
spinlock_t object_stat_lock;
|
2012-11-14 23:14:03 +07:00
|
|
|
size_t object_memory;
|
|
|
|
u32 object_count;
|
|
|
|
};
|
|
|
|
|
2013-05-23 17:55:35 +07:00
|
|
|
struct drm_i915_error_state_buf {
|
|
|
|
unsigned bytes;
|
|
|
|
unsigned size;
|
|
|
|
int err;
|
|
|
|
u8 *buf;
|
|
|
|
loff_t start;
|
|
|
|
loff_t pos;
|
|
|
|
};
|
|
|
|
|
2013-06-06 19:18:39 +07:00
|
|
|
struct i915_error_state_file_priv {
|
|
|
|
struct drm_device *dev;
|
|
|
|
struct drm_i915_error_state *error;
|
|
|
|
};
|
|
|
|
|
2012-11-14 23:14:04 +07:00
|
|
|
struct i915_gpu_error {
|
|
|
|
/* For hangcheck timer */
|
|
|
|
#define DRM_I915_HANGCHECK_PERIOD 1500 /* in ms */
|
|
|
|
#define DRM_I915_HANGCHECK_JIFFIES msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)
|
2013-08-30 20:19:28 +07:00
|
|
|
/* Hang gpu twice in this window and your context gets banned */
|
|
|
|
#define DRM_I915_CTX_BAN_PERIOD DIV_ROUND_UP(8*DRM_I915_HANGCHECK_PERIOD, 1000)
|
|
|
|
|
2012-11-14 23:14:04 +07:00
|
|
|
struct timer_list hangcheck_timer;
|
|
|
|
|
|
|
|
/* For reset and error_state handling. */
|
|
|
|
spinlock_t lock;
|
|
|
|
/* Protected by the above dev->gpu_error.lock. */
|
|
|
|
struct drm_i915_error_state *first_error;
|
|
|
|
struct work_struct work;
|
|
|
|
|
2013-09-25 23:34:55 +07:00
|
|
|
|
|
|
|
unsigned long missed_irq_rings;
|
|
|
|
|
2012-11-15 23:17:22 +07:00
|
|
|
/**
|
2013-11-12 19:44:19 +07:00
|
|
|
* State variable controlling the reset flow and count
|
2012-11-15 23:17:22 +07:00
|
|
|
*
|
2013-11-12 19:44:19 +07:00
|
|
|
* This is a counter which gets incremented when reset is triggered,
|
|
|
|
* and again when reset has been handled. So odd values (lowest bit set)
|
|
|
|
* means that reset is in progress and even values that
|
|
|
|
* (reset_counter >> 1):th reset was successfully completed.
|
|
|
|
*
|
|
|
|
* If reset is not completed succesfully, the I915_WEDGE bit is
|
|
|
|
* set meaning that hardware is terminally sour and there is no
|
|
|
|
* recovery. All waiters on the reset_queue will be woken when
|
|
|
|
* that happens.
|
|
|
|
*
|
|
|
|
* This counter is used by the wait_seqno code to notice that reset
|
|
|
|
* event happened and it needs to restart the entire ioctl (since most
|
|
|
|
* likely the seqno it waited for won't ever signal anytime soon).
|
drm/i915: create a race-free reset detection
With the previous patch the state transition handling of the reset
code itself is now (hopefully) race free and solid. But that still
leaves out everyone else - with the various lock-free wait paths
we have there's the possibility that the reset happens between the
point where we read the seqno we should wait on and the actual wait.
And if __wait_seqno then never sees the RESET_IN_PROGRESS state, we'll
happily wait for a seqno which will in all likelyhood never signal.
In practice this is not a big problem since the X server gets
constantly interrupted, and can then submit more work (hopefully) to
unblock everyone else: As soon as a new seqno write lands, all waiters
will unblock. But running the i-g-t reset testcase ZZ_hangman can
expose this race, especially on slower hw with fewer cpu cores.
Now looking forward to ARB_robustness and friends that's not the best
possible behaviour, hence this patch adds a reset_counter to be able
to detect any reset, even if a given thread never observed the
in-progress state.
The important part is to correctly order things:
- The write side needs to increment the counter after any seqno gets
reset. Hence we need to do that at the end of the reset work, and
again wake everyone up. We also need to place a barrier in between
any possible seqno changes and the counter increment, since any
unlock operations only guarantee that nothing leaks out, but not
that at later load operation gets moved ahead.
- On the read side we need to ensure that no reset can sneak in and
invalidate the seqno. In all cases we can use the one-sided barrier
that unlock operations guarantee (of the lock protecting the
respective seqno/ring pair) to ensure correct ordering. Hence it is
sufficient to place the atomic read before the mutex/spin_unlock and
no additional barriers are required.
The end-result of all this is that we need to wake up everyone twice
in a reset operation:
- First, before the reset starts, to get any lockholders of the locks,
so that the reset can proceed.
- Second, after the reset is completed, to allow waiters to properly
and reliably detect the reset condition and bail out.
I admit that this entire reset_counter thing smells a bit like
overkill, but I think it's justified since it makes it really explicit
what the bail-out condition is. And we need a reset counter anyway to
implement ARB_robustness, and imo with finer-grained locking on the
horizont this is the most resilient scheme I could think of.
v2: Drop spurious change in the wait_for_error EXIT_COND - we only
need to wait until we leave the reset-in-progress wedged state.
v3: Don't play tricks with barriers in the throttle ioctl, the
spin_unlock is barrier enough.
I've also considered using a little helper to grab the current
reset_counter, but then decided that hiding the atomic_read isn't a
great idea, since having it explicitly show up in the code is a nice
remainder to reviews to check the memory barriers.
v4: Add a comment to explain why we need to fall through in
__wait_seqno in the end variable assignments.
v5: Review from Damien:
- s/smb/smp/ in a comment
- don't increment the reset counter after we've set it to WEDGED. Now
we (again) properly wedge the gpu when the reset fails.
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-06 15:01:42 +07:00
|
|
|
*
|
|
|
|
* This is important for lock-free wait paths, where no contended lock
|
|
|
|
* naturally enforces the correct ordering between the bail-out of the
|
|
|
|
* waiter and the gpu reset work code.
|
2012-11-15 23:17:22 +07:00
|
|
|
*/
|
|
|
|
atomic_t reset_counter;
|
|
|
|
|
|
|
|
#define I915_RESET_IN_PROGRESS_FLAG 1
|
2013-11-12 19:44:19 +07:00
|
|
|
#define I915_WEDGED (1 << 31)
|
2012-11-15 23:17:22 +07:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Waitqueue to signal when the reset has completed. Used by clients
|
|
|
|
* that wait for dev_priv->mm.wedged to settle.
|
|
|
|
*/
|
|
|
|
wait_queue_head_t reset_queue;
|
2012-11-14 23:14:05 +07:00
|
|
|
|
2012-11-14 23:14:04 +07:00
|
|
|
/* For gpu hang simulation. */
|
|
|
|
unsigned int stop_rings;
|
2013-09-25 23:34:55 +07:00
|
|
|
|
|
|
|
/* For missed irq/seqno simulation. */
|
|
|
|
unsigned int test_irq_rings;
|
2012-11-14 23:14:04 +07:00
|
|
|
};
|
|
|
|
|
i915: ignore lid open event when resuming
i915 driver needs to do modeset when
1. system resumes from sleep
2. lid is opened
In PM_SUSPEND_MEM state, all the GPEs are cleared when system resumes,
thus it is the i915_resume code does the modeset rather than intel_lid_notify().
But in PM_SUSPEND_FREEZE state, this will be broken because
system is still responsive to the lid events.
1. When we close the lid in Freeze state, intel_lid_notify() sets modeset_on_lid.
2. When we reopen the lid, intel_lid_notify() will do a modeset,
before the system is resumed.
here is the error log,
[92146.548074] WARNING: at drivers/gpu/drm/i915/intel_display.c:1028 intel_wait_for_pipe_off+0x184/0x190 [i915]()
[92146.548076] Hardware name: VGN-Z540N
[92146.548078] pipe_off wait timed out
[92146.548167] Modules linked in: hid_generic usbhid hid snd_hda_codec_realtek snd_hda_intel snd_hda_codec parport_pc snd_hwdep ppdev snd_pcm_oss i915 snd_mixer_oss snd_pcm arc4 iwldvm snd_seq_dummy mac80211 snd_seq_oss snd_seq_midi fbcon tileblit font bitblit softcursor drm_kms_helper snd_rawmidi snd_seq_midi_event coretemp drm snd_seq kvm btusb bluetooth snd_timer iwlwifi pcmcia tpm_infineon i2c_algo_bit joydev snd_seq_device intel_agp cfg80211 snd intel_gtt yenta_socket pcmcia_rsrc sony_laptop agpgart microcode psmouse tpm_tis serio_raw mxm_wmi soundcore snd_page_alloc tpm acpi_cpufreq lpc_ich pcmcia_core tpm_bios mperf processor lp parport firewire_ohci firewire_core crc_itu_t sdhci_pci sdhci thermal e1000e
[92146.548173] Pid: 4304, comm: kworker/0:0 Tainted: G W 3.8.0-rc3-s0i3-v3-test+ #9
[92146.548175] Call Trace:
[92146.548189] [<c10378e2>] warn_slowpath_common+0x72/0xa0
[92146.548227] [<f86398b4>] ? intel_wait_for_pipe_off+0x184/0x190 [i915]
[92146.548263] [<f86398b4>] ? intel_wait_for_pipe_off+0x184/0x190 [i915]
[92146.548270] [<c10379b3>] warn_slowpath_fmt+0x33/0x40
[92146.548307] [<f86398b4>] intel_wait_for_pipe_off+0x184/0x190 [i915]
[92146.548344] [<f86399c2>] intel_disable_pipe+0x102/0x190 [i915]
[92146.548380] [<f8639ea4>] ? intel_disable_plane+0x64/0x80 [i915]
[92146.548417] [<f8639f7c>] i9xx_crtc_disable+0xbc/0x150 [i915]
[92146.548456] [<f863ebee>] intel_crtc_update_dpms+0x5e/0x90 [i915]
[92146.548493] [<f86437cf>] intel_modeset_setup_hw_state+0x42f/0x8f0 [i915]
[92146.548535] [<f8645b0b>] intel_lid_notify+0x9b/0xc0 [i915]
[92146.548543] [<c15610d3>] notifier_call_chain+0x43/0x60
[92146.548550] [<c105d1e1>] __blocking_notifier_call_chain+0x41/0x80
[92146.548556] [<c105d23f>] blocking_notifier_call_chain+0x1f/0x30
[92146.548563] [<c131a684>] acpi_lid_send_state+0x78/0xa4
[92146.548569] [<c131aa9e>] acpi_button_notify+0x3b/0xf1
[92146.548577] [<c12df56a>] ? acpi_os_execute+0x17/0x19
[92146.548582] [<c12e591a>] ? acpi_ec_sync_query+0xa5/0xbc
[92146.548589] [<c12e2b82>] acpi_device_notify+0x16/0x18
[92146.548595] [<c12f4904>] acpi_ev_notify_dispatch+0x38/0x4f
[92146.548600] [<c12df0e8>] acpi_os_execute_deferred+0x20/0x2b
[92146.548607] [<c1051208>] process_one_work+0x128/0x3f0
[92146.548613] [<c1564f73>] ? common_interrupt+0x33/0x38
[92146.548618] [<c104f8c0>] ? wake_up_worker+0x30/0x30
[92146.548624] [<c12df0c8>] ? acpi_os_wait_events_complete+0x1e/0x1e
[92146.548629] [<c10524f9>] worker_thread+0x119/0x3b0
[92146.548634] [<c10523e0>] ? manage_workers+0x240/0x240
[92146.548640] [<c1056e84>] kthread+0x94/0xa0
[92146.548647] [<c1060000>] ? ftrace_raw_output_sched_stat_runtime+0x70/0xf0
[92146.548652] [<c15649b7>] ret_from_kernel_thread+0x1b/0x28
[92146.548658] [<c1056df0>] ? kthread_create_on_node+0xc0/0xc0
three different modeset flags are introduced in this patch
MODESET_ON_LID_OPEN: do modeset on next lid open event
MODESET_DONE: modeset already done
MODESET_SUSPENDED: suspended, only do modeset when system is resumed
In this way,
1. when lid is closed, MODESET_ON_LID_OPEN is set so that
we'll do modeset on next lid open event.
2. when lid is opened, MODESET_DONE is set
so that duplicate lid open events will be ignored.
3. when system suspends, MODESET_SUSPENDED is set.
In this case, we will not do modeset on any lid events.
Plus, locking mechanism is also introduced to avoid racing.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-02-05 14:41:53 +07:00
|
|
|
enum modeset_restore {
|
|
|
|
MODESET_ON_LID_OPEN,
|
|
|
|
MODESET_DONE,
|
|
|
|
MODESET_SUSPENDED,
|
|
|
|
};
|
|
|
|
|
2013-09-13 03:06:24 +07:00
|
|
|
struct ddi_vbt_port_info {
|
|
|
|
uint8_t hdmi_level_shift;
|
2013-09-13 03:12:18 +07:00
|
|
|
|
|
|
|
uint8_t supports_dvi:1;
|
|
|
|
uint8_t supports_hdmi:1;
|
|
|
|
uint8_t supports_dp:1;
|
2013-09-13 03:06:24 +07:00
|
|
|
};
|
|
|
|
|
2013-05-10 06:03:18 +07:00
|
|
|
struct intel_vbt_data {
|
|
|
|
struct drm_display_mode *lfp_lvds_vbt_mode; /* if any */
|
|
|
|
struct drm_display_mode *sdvo_lvds_vbt_mode; /* if any */
|
|
|
|
|
|
|
|
/* Feature bits */
|
|
|
|
unsigned int int_tv_support:1;
|
|
|
|
unsigned int lvds_dither:1;
|
|
|
|
unsigned int lvds_vbt:1;
|
|
|
|
unsigned int int_crt_support:1;
|
|
|
|
unsigned int lvds_use_ssc:1;
|
|
|
|
unsigned int display_clock_mode:1;
|
|
|
|
unsigned int fdi_rx_polarity_inverted:1;
|
|
|
|
int lvds_ssc_freq;
|
|
|
|
unsigned int bios_lvds_val; /* initial [PCH_]LVDS reg val in VBIOS */
|
|
|
|
|
|
|
|
/* eDP */
|
|
|
|
int edp_rate;
|
|
|
|
int edp_lanes;
|
|
|
|
int edp_preemphasis;
|
|
|
|
int edp_vswing;
|
|
|
|
bool edp_initialized;
|
|
|
|
bool edp_support;
|
|
|
|
int edp_bpp;
|
|
|
|
struct edp_power_seq edp_pps;
|
|
|
|
|
2013-12-15 05:38:29 +07:00
|
|
|
struct {
|
|
|
|
u16 pwm_freq_hz;
|
|
|
|
bool active_low_pwm;
|
|
|
|
} backlight;
|
|
|
|
|
2013-08-27 19:12:25 +07:00
|
|
|
/* MIPI DSI */
|
|
|
|
struct {
|
|
|
|
u16 panel_id;
|
|
|
|
} dsi;
|
|
|
|
|
2013-05-10 06:03:18 +07:00
|
|
|
int crt_ddc_pin;
|
|
|
|
|
|
|
|
int child_dev_num;
|
2013-09-12 04:02:47 +07:00
|
|
|
union child_device_config *child_dev;
|
2013-09-13 03:06:24 +07:00
|
|
|
|
|
|
|
struct ddi_vbt_port_info ddi_port_info[I915_MAX_PORTS];
|
2013-05-10 06:03:18 +07:00
|
|
|
};
|
|
|
|
|
2013-08-07 02:24:04 +07:00
|
|
|
enum intel_ddb_partitioning {
|
|
|
|
INTEL_DDB_PART_1_2,
|
|
|
|
INTEL_DDB_PART_5_6, /* IVB+ */
|
|
|
|
};
|
|
|
|
|
2013-08-07 02:24:05 +07:00
|
|
|
struct intel_wm_level {
|
|
|
|
bool enable;
|
|
|
|
uint32_t pri_val;
|
|
|
|
uint32_t spr_val;
|
|
|
|
uint32_t cur_val;
|
|
|
|
uint32_t fbc_val;
|
|
|
|
};
|
|
|
|
|
2013-12-17 19:46:36 +07:00
|
|
|
struct ilk_wm_values {
|
2013-10-09 23:18:03 +07:00
|
|
|
uint32_t wm_pipe[3];
|
|
|
|
uint32_t wm_lp[3];
|
|
|
|
uint32_t wm_lp_spr[3];
|
|
|
|
uint32_t wm_linetime[3];
|
|
|
|
bool enable_fbc_wm;
|
|
|
|
enum intel_ddb_partitioning partitioning;
|
|
|
|
};
|
|
|
|
|
2013-08-19 23:18:09 +07:00
|
|
|
/*
|
|
|
|
* This struct tracks the state needed for the Package C8+ feature.
|
|
|
|
*
|
|
|
|
* Package states C8 and deeper are really deep PC states that can only be
|
|
|
|
* reached when all the devices on the system allow it, so even if the graphics
|
|
|
|
* device allows PC8+, it doesn't mean the system will actually get to these
|
|
|
|
* states.
|
|
|
|
*
|
|
|
|
* Our driver only allows PC8+ when all the outputs are disabled, the power well
|
|
|
|
* is disabled and the GPU is idle. When these conditions are met, we manually
|
|
|
|
* do the other conditions: disable the interrupts, clocks and switch LCPLL
|
|
|
|
* refclk to Fclk.
|
|
|
|
*
|
|
|
|
* When we really reach PC8 or deeper states (not just when we allow it) we lose
|
|
|
|
* the state of some registers, so when we come back from PC8+ we need to
|
|
|
|
* restore this state. We don't get into PC8+ if we're not in RC6, so we don't
|
|
|
|
* need to take care of the registers kept by RC6.
|
|
|
|
*
|
|
|
|
* The interrupt disabling is part of the requirements. We can only leave the
|
|
|
|
* PCH HPD interrupts enabled. If we're in PC8+ and we get another interrupt we
|
|
|
|
* can lock the machine.
|
|
|
|
*
|
|
|
|
* Ideally every piece of our code that needs PC8+ disabled would call
|
|
|
|
* hsw_disable_package_c8, which would increment disable_count and prevent the
|
|
|
|
* system from reaching PC8+. But we don't have a symmetric way to do this for
|
|
|
|
* everything, so we have the requirements_met and gpu_idle variables. When we
|
|
|
|
* switch requirements_met or gpu_idle to true we decrease disable_count, and
|
|
|
|
* increase it in the opposite case. The requirements_met variable is true when
|
|
|
|
* all the CRTCs, encoders and the power well are disabled. The gpu_idle
|
|
|
|
* variable is true when the GPU is idle.
|
|
|
|
*
|
|
|
|
* In addition to everything, we only actually enable PC8+ if disable_count
|
|
|
|
* stays at zero for at least some seconds. This is implemented with the
|
|
|
|
* enable_work variable. We do this so we don't enable/disable PC8 dozens of
|
|
|
|
* consecutive times when all screens are disabled and some background app
|
|
|
|
* queries the state of our connectors, or we have some application constantly
|
|
|
|
* waking up to use the GPU. Only after the enable_work function actually
|
|
|
|
* enables PC8+ the "enable" variable will become true, which means that it can
|
|
|
|
* be false even if disable_count is 0.
|
|
|
|
*
|
|
|
|
* The irqs_disabled variable becomes true exactly after we disable the IRQs and
|
|
|
|
* goes back to false exactly before we reenable the IRQs. We use this variable
|
|
|
|
* to check if someone is trying to enable/disable IRQs while they're supposed
|
|
|
|
* to be disabled. This shouldn't happen and we'll print some error messages in
|
|
|
|
* case it happens, but if it actually happens we'll also update the variables
|
|
|
|
* inside struct regsave so when we restore the IRQs they will contain the
|
|
|
|
* latest expected values.
|
|
|
|
*
|
|
|
|
* For more, read "Display Sequences for Package C8" on our documentation.
|
|
|
|
*/
|
|
|
|
struct i915_package_c8 {
|
|
|
|
bool requirements_met;
|
|
|
|
bool gpu_idle;
|
|
|
|
bool irqs_disabled;
|
|
|
|
/* Only true after the delayed work task actually enables it. */
|
|
|
|
bool enabled;
|
|
|
|
int disable_count;
|
|
|
|
struct mutex lock;
|
|
|
|
struct delayed_work enable_work;
|
|
|
|
|
|
|
|
struct {
|
|
|
|
uint32_t deimr;
|
|
|
|
uint32_t sdeimr;
|
|
|
|
uint32_t gtimr;
|
|
|
|
uint32_t gtier;
|
|
|
|
uint32_t gen6_pmimr;
|
|
|
|
} regsave;
|
|
|
|
};
|
|
|
|
|
2013-12-07 05:32:13 +07:00
|
|
|
struct i915_runtime_pm {
|
|
|
|
bool suspended;
|
|
|
|
};
|
|
|
|
|
2013-10-16 18:30:34 +07:00
|
|
|
enum intel_pipe_crc_source {
|
|
|
|
INTEL_PIPE_CRC_SOURCE_NONE,
|
|
|
|
INTEL_PIPE_CRC_SOURCE_PLANE1,
|
|
|
|
INTEL_PIPE_CRC_SOURCE_PLANE2,
|
|
|
|
INTEL_PIPE_CRC_SOURCE_PF,
|
2013-10-17 03:55:48 +07:00
|
|
|
INTEL_PIPE_CRC_SOURCE_PIPE,
|
2013-10-17 03:55:58 +07:00
|
|
|
/* TV/DP on pre-gen5/vlv can't use the pipe source. */
|
|
|
|
INTEL_PIPE_CRC_SOURCE_TV,
|
|
|
|
INTEL_PIPE_CRC_SOURCE_DP_B,
|
|
|
|
INTEL_PIPE_CRC_SOURCE_DP_C,
|
|
|
|
INTEL_PIPE_CRC_SOURCE_DP_D,
|
2013-11-01 16:50:20 +07:00
|
|
|
INTEL_PIPE_CRC_SOURCE_AUTO,
|
2013-10-16 18:30:34 +07:00
|
|
|
INTEL_PIPE_CRC_SOURCE_MAX,
|
|
|
|
};
|
|
|
|
|
2013-10-16 00:55:27 +07:00
|
|
|
struct intel_pipe_crc_entry {
|
2013-10-16 00:55:30 +07:00
|
|
|
uint32_t frame;
|
2013-10-16 00:55:27 +07:00
|
|
|
uint32_t crc[5];
|
|
|
|
};
|
|
|
|
|
2013-10-16 00:55:29 +07:00
|
|
|
#define INTEL_PIPE_CRC_ENTRIES_NR 128
|
2013-10-16 00:55:27 +07:00
|
|
|
struct intel_pipe_crc {
|
2013-10-21 20:29:30 +07:00
|
|
|
spinlock_t lock;
|
|
|
|
bool opened; /* exclusive access to the result file */
|
2013-10-16 00:55:34 +07:00
|
|
|
struct intel_pipe_crc_entry *entries;
|
2013-10-16 18:30:34 +07:00
|
|
|
enum intel_pipe_crc_source source;
|
2013-10-21 20:29:30 +07:00
|
|
|
int head, tail;
|
2013-10-16 00:55:40 +07:00
|
|
|
wait_queue_head_t wq;
|
2013-10-16 00:55:27 +07:00
|
|
|
};
|
|
|
|
|
2012-11-03 01:55:02 +07:00
|
|
|
typedef struct drm_i915_private {
|
|
|
|
struct drm_device *dev;
|
2012-11-15 18:32:30 +07:00
|
|
|
struct kmem_cache *slab;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
2014-02-08 02:12:48 +07:00
|
|
|
const struct intel_device_info info;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
|
|
|
int relative_constants_mode;
|
|
|
|
|
|
|
|
void __iomem *regs;
|
|
|
|
|
2013-07-20 02:36:52 +07:00
|
|
|
struct intel_uncore uncore;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
|
|
|
struct intel_gmbus gmbus[GMBUS_NUM_PORTS];
|
|
|
|
|
2012-12-01 19:53:45 +07:00
|
|
|
|
2012-11-03 01:55:02 +07:00
|
|
|
/** gmbus_mutex protects against concurrent usage of the single hw gmbus
|
|
|
|
* controller on different i2c buses. */
|
|
|
|
struct mutex gmbus_mutex;
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Base address of the gmbus and gpio block.
|
|
|
|
*/
|
|
|
|
uint32_t gpio_mmio_base;
|
|
|
|
|
2012-12-01 19:53:45 +07:00
|
|
|
wait_queue_head_t gmbus_wait_queue;
|
|
|
|
|
2012-11-03 01:55:02 +07:00
|
|
|
struct pci_dev *bridge_dev;
|
|
|
|
struct intel_ring_buffer ring[I915_NUM_RINGS];
|
2012-12-10 20:41:48 +07:00
|
|
|
uint32_t last_seqno, next_seqno;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
|
|
|
drm_dma_handle_t *status_page_dmah;
|
|
|
|
struct resource mch_res;
|
|
|
|
|
|
|
|
/* protects the irq masks */
|
|
|
|
spinlock_t irq_lock;
|
|
|
|
|
drm/i915: irq-drive the dp aux communication
At least on the platforms that have a dp aux irq and also have it
enabled - vlvhsw should have one, too. But I don't have a machine to
test this on. Judging from docs there's no dp aux interrupt for gm45.
Also, I only have an ivb cpu edp machine, so the dp aux A code for
snb/ilk is untested.
For dpcd probing when nothing is connected it slashes about 5ms of cpu
time (cpu time is now negligible), which agrees with 3 * 5 400 usec
timeouts.
A previous version of this patch increases the time required to go
through the dp_detect cycle (which includes reading the edid) from
around 33 ms to around 40 ms. Experiments indicated that this is
purely due to the irq latency - the hw doesn't allow us to queue up
dp aux transactions and hence irq latency directly affects throughput.
gmbus is much better, there we have a 8 byte buffer, and we get the
irq once another 4 bytes can be queued up.
But by using the pm_qos interface to request the lowest possible cpu
wake-up latency this slowdown completely disappeared.
Since all our output detection logic is single-threaded with the
mode_config mutex right now anyway, I've decide not ot play fancy and
to just reuse the gmbus wait queue. But this would definitely prep the
way to run dp detection on different ports in parallel
v2: Add a timeout for dp aux transfers when using interrupts - the hw
_does_ prevent this with the hw-based 400 usec timeout, but if the
irq somehow doesn't arrive we're screwed. Lesson learned while
developing this ;-)
v3: While at it also convert the busy-loop to wait_for_atomic, so that
we don't run the risk of an infinite loop any more.
v4: Ensure we have the smallest possible irq latency by using the
pm_qos interface.
v5: Add a comment to the code to explain why we frob pm_qos. Suggested
by Chris Wilson.
v6: Disable dp irq for vlv, that's easier than trying to get at docs
and hw.
v7: Squash in a fix for Haswell that Paulo Zanoni tracked down - the
dp aux registers aren't at a fixed offset any more, but can be on the
PCH while the DP port is on the cpu die.
Reviewed-by: Imre Deak <imre.deak@intel.com> (v6)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-01 19:53:48 +07:00
|
|
|
/* To control wakeup latency, e.g. for irq-driven dp aux transfers. */
|
|
|
|
struct pm_qos_request pm_qos;
|
|
|
|
|
2012-11-03 01:55:02 +07:00
|
|
|
/* DPIO indirect register protection */
|
2012-12-12 20:06:44 +07:00
|
|
|
struct mutex dpio_lock;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
|
|
|
/** Cached value of IMR to avoid reads in updating the bitfield */
|
drm/i915/bdw: Implement interrupt changes
The interrupt handling implementation remains the same as previous
generations with the 4 types of registers, status, identity, mask, and
enable. However the layout of where the bits go have changed entirely.
To address these changes, all of the interrupt vfuncs needed special
gen8 code.
The way it works is there is a top level status register now which
informs the interrupt service routine which unit caused the interrupt,
and therefore which interrupt registers to read to process the
interrupt. For display the division is quite logical, a set of interrupt
registers for each pipe, and in addition to those, a set each for "misc"
and port.
For GT the things get a bit hairy, as seen by the code. Each of the GT
units has it's own bits defined. They all look *very similar* and
resides in 16 bits of a GT register. As an example, RCS and BCS share
register 0. To compact the code a bit, at a slight expense to
complexity, this is exactly how the code works as well. 2 structures are
added to the ring buffer so that our ring buffer interrupt handling code
knows which ring shares the interrupt registers, and a shift value (ie.
the top or bottom 16 bits of the register).
The above allows us to kept the interrupt register caching scheme, the
per interrupt enables, and the code to mask and unmask interrupts
relatively clean (again at the cost of some more complexity).
Most of the GT units mentioned above are command streamers, and so the
symmetry should work quite well for even the yet to be implemented rings
which Broadwell adds.
v2: Fixes up a couple of bugs, and is more verbose about errors in the
Broadwell interrupt handler.
v3: fix DE_MISC IER offset
v4: Simplify interrupts:
I totally misread the docs the first time I implemented interrupts, and
so this should greatly simplify the mess. Unlike GEN6, we never touch
the regular mask registers in irq_get/put.
v5: Rebased on to of recent pch hotplug setup changes.
v6: Fixup on top of moving num_pipes to intel_info.
v7: Rebased on top of Egbert Eich's hpd irq handling rework. Also
wired up ibx_hpd_irq_setup for gen8.
v8: Rebase on top of Jani's asle handling rework.
v9: Rebase on top of Ben's VECS enabling for Haswell, where he
unfortunately went OCD on the gt irq #defines. Not that they're still
not yet fully consistent:
- Used the GT_RENDER_ #defines + bdw shifts.
- Dropped the shift from the L3_PARITY stuff, seemed clearer.
- s/irq_refcount/irq_refcount.gt/
v10: Squash in VECS enabling patches and the gen8_gt_irq_handler
refactoring from Zhao Yakui <yakui.zhao@intel.com>
v11: Rebase on top of the interrupt cleanups in upstream.
v12: Rebase on top of Ben's DPF changes in upstream.
v13: Drop bdw from the HAS_L3_DPF feature flag for now, it's unclear what
exactly needs to be done. Requested by Ben.
v14: Fix the patch.
- Drop the mask of reserved bits and assorted logic, it doesn't match
the spec.
- Do the posting read inconditionally instead of commenting it out.
- Add a GEN8_MASTER_IRQ_CONTROL definition and use it.
- Fix up the GEN8_PIPE interrupt defines and give the GEN8_ prefixes -
we actually will need to use them.
- Enclose macros in do {} while (0) (checkpatch).
- Clear DE_MISC interrupt bits only after having processed them.
- Fix whitespace fail (checkpatch).
- Fix overtly long lines where appropriate (checkpatch).
- Don't use typedef'ed private_t (maintainer-scripts).
- Align the function parameter list correctly.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net> (v4)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
bikeshed
2013-11-03 11:07:09 +07:00
|
|
|
union {
|
|
|
|
u32 irq_mask;
|
|
|
|
u32 de_irq_mask[I915_MAX_PIPES];
|
|
|
|
};
|
2012-11-03 01:55:02 +07:00
|
|
|
u32 gt_irq_mask;
|
2013-08-07 04:57:15 +07:00
|
|
|
u32 pm_irq_mask;
|
2014-02-10 23:42:49 +07:00
|
|
|
u32 pipestat_irq_mask[I915_MAX_PIPES];
|
2012-11-03 01:55:02 +07:00
|
|
|
|
|
|
|
struct work_struct hotplug_work;
|
2012-12-02 03:03:22 +07:00
|
|
|
bool enable_hotplug_processing;
|
2013-04-16 18:36:54 +07:00
|
|
|
struct {
|
|
|
|
unsigned long hpd_last_jiffies;
|
|
|
|
int hpd_cnt;
|
|
|
|
enum {
|
|
|
|
HPD_ENABLED = 0,
|
|
|
|
HPD_DISABLED = 1,
|
|
|
|
HPD_MARK_DISABLED = 2
|
|
|
|
} hpd_mark;
|
|
|
|
} hpd_stats[HPD_NUM_PINS];
|
2013-04-11 20:57:57 +07:00
|
|
|
u32 hpd_event_bits;
|
2013-04-16 18:36:58 +07:00
|
|
|
struct timer_list hotplug_reenable_timer;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
2013-06-28 06:30:21 +07:00
|
|
|
struct i915_fbc fbc;
|
2012-11-03 01:55:02 +07:00
|
|
|
struct intel_opregion opregion;
|
2013-05-10 06:03:18 +07:00
|
|
|
struct intel_vbt_data vbt;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
|
|
|
/* overlay */
|
|
|
|
struct intel_overlay *overlay;
|
|
|
|
|
2013-11-08 21:48:54 +07:00
|
|
|
/* backlight registers and fields in struct intel_panel */
|
|
|
|
spinlock_t backlight_lock;
|
2013-04-02 19:48:09 +07:00
|
|
|
|
2012-11-03 01:55:02 +07:00
|
|
|
/* LVDS info */
|
|
|
|
bool no_aux_handshake;
|
|
|
|
|
|
|
|
struct drm_i915_fence_reg fence_regs[I915_MAX_NUM_FENCES]; /* assume 965 */
|
|
|
|
int fence_reg_start; /* 4 if userland hasn't ioctl'd us yet */
|
|
|
|
int num_fence_regs; /* 8 on pre-965, 16 otherwise */
|
|
|
|
|
|
|
|
unsigned int fsb_freq, mem_freq, is_ddr3;
|
|
|
|
|
2013-09-02 21:22:25 +07:00
|
|
|
/**
|
|
|
|
* wq - Driver workqueue for GEM.
|
|
|
|
*
|
|
|
|
* NOTE: Work items scheduled here are not allowed to grab any modeset
|
|
|
|
* locks, for otherwise the flushing done in the pageflip code will
|
|
|
|
* result in deadlocks.
|
|
|
|
*/
|
2012-11-03 01:55:02 +07:00
|
|
|
struct workqueue_struct *wq;
|
|
|
|
|
|
|
|
/* Display functions */
|
|
|
|
struct drm_i915_display_funcs display;
|
|
|
|
|
|
|
|
/* PCH chipset type */
|
|
|
|
enum intel_pch pch_type;
|
2012-11-21 00:12:07 +07:00
|
|
|
unsigned short pch_id;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
|
|
|
unsigned long quirks;
|
|
|
|
|
i915: ignore lid open event when resuming
i915 driver needs to do modeset when
1. system resumes from sleep
2. lid is opened
In PM_SUSPEND_MEM state, all the GPEs are cleared when system resumes,
thus it is the i915_resume code does the modeset rather than intel_lid_notify().
But in PM_SUSPEND_FREEZE state, this will be broken because
system is still responsive to the lid events.
1. When we close the lid in Freeze state, intel_lid_notify() sets modeset_on_lid.
2. When we reopen the lid, intel_lid_notify() will do a modeset,
before the system is resumed.
here is the error log,
[92146.548074] WARNING: at drivers/gpu/drm/i915/intel_display.c:1028 intel_wait_for_pipe_off+0x184/0x190 [i915]()
[92146.548076] Hardware name: VGN-Z540N
[92146.548078] pipe_off wait timed out
[92146.548167] Modules linked in: hid_generic usbhid hid snd_hda_codec_realtek snd_hda_intel snd_hda_codec parport_pc snd_hwdep ppdev snd_pcm_oss i915 snd_mixer_oss snd_pcm arc4 iwldvm snd_seq_dummy mac80211 snd_seq_oss snd_seq_midi fbcon tileblit font bitblit softcursor drm_kms_helper snd_rawmidi snd_seq_midi_event coretemp drm snd_seq kvm btusb bluetooth snd_timer iwlwifi pcmcia tpm_infineon i2c_algo_bit joydev snd_seq_device intel_agp cfg80211 snd intel_gtt yenta_socket pcmcia_rsrc sony_laptop agpgart microcode psmouse tpm_tis serio_raw mxm_wmi soundcore snd_page_alloc tpm acpi_cpufreq lpc_ich pcmcia_core tpm_bios mperf processor lp parport firewire_ohci firewire_core crc_itu_t sdhci_pci sdhci thermal e1000e
[92146.548173] Pid: 4304, comm: kworker/0:0 Tainted: G W 3.8.0-rc3-s0i3-v3-test+ #9
[92146.548175] Call Trace:
[92146.548189] [<c10378e2>] warn_slowpath_common+0x72/0xa0
[92146.548227] [<f86398b4>] ? intel_wait_for_pipe_off+0x184/0x190 [i915]
[92146.548263] [<f86398b4>] ? intel_wait_for_pipe_off+0x184/0x190 [i915]
[92146.548270] [<c10379b3>] warn_slowpath_fmt+0x33/0x40
[92146.548307] [<f86398b4>] intel_wait_for_pipe_off+0x184/0x190 [i915]
[92146.548344] [<f86399c2>] intel_disable_pipe+0x102/0x190 [i915]
[92146.548380] [<f8639ea4>] ? intel_disable_plane+0x64/0x80 [i915]
[92146.548417] [<f8639f7c>] i9xx_crtc_disable+0xbc/0x150 [i915]
[92146.548456] [<f863ebee>] intel_crtc_update_dpms+0x5e/0x90 [i915]
[92146.548493] [<f86437cf>] intel_modeset_setup_hw_state+0x42f/0x8f0 [i915]
[92146.548535] [<f8645b0b>] intel_lid_notify+0x9b/0xc0 [i915]
[92146.548543] [<c15610d3>] notifier_call_chain+0x43/0x60
[92146.548550] [<c105d1e1>] __blocking_notifier_call_chain+0x41/0x80
[92146.548556] [<c105d23f>] blocking_notifier_call_chain+0x1f/0x30
[92146.548563] [<c131a684>] acpi_lid_send_state+0x78/0xa4
[92146.548569] [<c131aa9e>] acpi_button_notify+0x3b/0xf1
[92146.548577] [<c12df56a>] ? acpi_os_execute+0x17/0x19
[92146.548582] [<c12e591a>] ? acpi_ec_sync_query+0xa5/0xbc
[92146.548589] [<c12e2b82>] acpi_device_notify+0x16/0x18
[92146.548595] [<c12f4904>] acpi_ev_notify_dispatch+0x38/0x4f
[92146.548600] [<c12df0e8>] acpi_os_execute_deferred+0x20/0x2b
[92146.548607] [<c1051208>] process_one_work+0x128/0x3f0
[92146.548613] [<c1564f73>] ? common_interrupt+0x33/0x38
[92146.548618] [<c104f8c0>] ? wake_up_worker+0x30/0x30
[92146.548624] [<c12df0c8>] ? acpi_os_wait_events_complete+0x1e/0x1e
[92146.548629] [<c10524f9>] worker_thread+0x119/0x3b0
[92146.548634] [<c10523e0>] ? manage_workers+0x240/0x240
[92146.548640] [<c1056e84>] kthread+0x94/0xa0
[92146.548647] [<c1060000>] ? ftrace_raw_output_sched_stat_runtime+0x70/0xf0
[92146.548652] [<c15649b7>] ret_from_kernel_thread+0x1b/0x28
[92146.548658] [<c1056df0>] ? kthread_create_on_node+0xc0/0xc0
three different modeset flags are introduced in this patch
MODESET_ON_LID_OPEN: do modeset on next lid open event
MODESET_DONE: modeset already done
MODESET_SUSPENDED: suspended, only do modeset when system is resumed
In this way,
1. when lid is closed, MODESET_ON_LID_OPEN is set so that
we'll do modeset on next lid open event.
2. when lid is opened, MODESET_DONE is set
so that duplicate lid open events will be ignored.
3. when system suspends, MODESET_SUSPENDED is set.
In this case, we will not do modeset on any lid events.
Plus, locking mechanism is also introduced to avoid racing.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-02-05 14:41:53 +07:00
|
|
|
enum modeset_restore modeset_restore;
|
|
|
|
struct mutex modeset_restore_lock;
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2013-07-17 06:50:07 +07:00
|
|
|
struct list_head vm_list; /* Global list of all address spaces */
|
2013-07-17 06:50:05 +07:00
|
|
|
struct i915_gtt gtt; /* VMA representing the global address space */
|
2013-01-18 03:45:15 +07:00
|
|
|
|
2012-11-14 23:14:03 +07:00
|
|
|
struct i915_gem_mm mm;
|
2012-05-02 16:49:32 +07:00
|
|
|
|
|
|
|
/* Kernel Modesetting */
|
|
|
|
|
2009-05-31 16:17:17 +07:00
|
|
|
struct sdvo_device_mapping sdvo_mappings[2];
|
2009-08-18 03:31:43 +07:00
|
|
|
|
2014-02-08 02:12:52 +07:00
|
|
|
struct drm_crtc *plane_to_crtc_mapping[I915_MAX_PIPES];
|
|
|
|
struct drm_crtc *pipe_to_crtc_mapping[I915_MAX_PIPES];
|
2009-11-18 23:25:18 +07:00
|
|
|
wait_queue_head_t pending_flip_queue;
|
|
|
|
|
2013-10-22 02:04:07 +07:00
|
|
|
#ifdef CONFIG_DEBUG_FS
|
|
|
|
struct intel_pipe_crc pipe_crc[I915_MAX_PIPES];
|
|
|
|
#endif
|
|
|
|
|
2013-06-05 18:34:06 +07:00
|
|
|
int num_shared_dpll;
|
|
|
|
struct intel_shared_dpll shared_dplls[I915_NUM_PLLS];
|
2012-10-05 22:05:58 +07:00
|
|
|
struct intel_ddi_plls ddi_plls;
|
2013-11-06 13:36:35 +07:00
|
|
|
int dpio_phy_iosf_port[I915_NUM_PHYS_VLV];
|
2012-04-20 23:11:53 +07:00
|
|
|
|
2009-08-18 03:31:43 +07:00
|
|
|
/* Reclocking support */
|
|
|
|
bool render_reclock_avail;
|
|
|
|
bool lvds_downclock_avail;
|
2009-11-20 10:24:16 +07:00
|
|
|
/* indicates the reduced downclock for LVDS*/
|
|
|
|
int lvds_downclock;
|
2009-08-18 03:31:43 +07:00
|
|
|
u16 orig_clock;
|
2010-01-30 02:27:07 +07:00
|
|
|
|
2009-12-17 13:48:43 +07:00
|
|
|
bool mchbar_need_disable;
|
2010-01-30 02:27:07 +07:00
|
|
|
|
2012-11-03 01:55:07 +07:00
|
|
|
struct intel_l3_parity l3_parity;
|
|
|
|
|
2013-07-05 01:02:05 +07:00
|
|
|
/* Cannot be determined by PCIID. You must always read a register. */
|
|
|
|
size_t ellc_size;
|
|
|
|
|
2012-08-09 04:35:35 +07:00
|
|
|
/* gen6+ rps state */
|
2012-11-03 01:55:03 +07:00
|
|
|
struct intel_gen6_power_mgmt rps;
|
2012-08-09 04:35:35 +07:00
|
|
|
|
2012-08-09 04:35:39 +07:00
|
|
|
/* ilk-only ips/rps state. Everything in here is protected by the global
|
|
|
|
* mchdev_lock in intel_pm.c */
|
2012-11-03 01:55:03 +07:00
|
|
|
struct intel_ilk_power_mgmt ips;
|
2010-02-06 03:42:41 +07:00
|
|
|
|
2013-10-25 21:36:47 +07:00
|
|
|
struct i915_power_domains power_domains;
|
2013-05-30 21:07:11 +07:00
|
|
|
|
2013-10-04 02:15:06 +07:00
|
|
|
struct i915_psr psr;
|
2013-07-12 04:45:00 +07:00
|
|
|
|
2012-11-14 23:14:04 +07:00
|
|
|
struct i915_gpu_error gpu_error;
|
2010-10-01 20:57:56 +07:00
|
|
|
|
2013-05-09 00:45:13 +07:00
|
|
|
struct drm_i915_gem_object *vlv_pctx;
|
|
|
|
|
2013-10-09 14:18:51 +07:00
|
|
|
#ifdef CONFIG_DRM_I915_FBDEV
|
2010-03-30 12:34:14 +07:00
|
|
|
/* list of fbdev register on this device */
|
|
|
|
struct intel_fbdev *fbdev;
|
2013-10-09 14:18:51 +07:00
|
|
|
#endif
|
2011-02-22 05:23:52 +07:00
|
|
|
|
2012-11-03 01:13:59 +07:00
|
|
|
/*
|
|
|
|
* The console may be contended at resume, but we don't
|
|
|
|
* want it to block on it.
|
|
|
|
*/
|
|
|
|
struct work_struct console_resume_work;
|
|
|
|
|
2011-02-22 05:23:52 +07:00
|
|
|
struct drm_property *broadcast_rgb_property;
|
2011-05-13 04:17:24 +07:00
|
|
|
struct drm_property *force_audio_property;
|
2012-05-26 06:56:22 +07:00
|
|
|
|
drm/i915: preliminary context support
Very basic code for context setup/destruction in the driver.
Adds the file i915_gem_context.c This file implements HW context
support. On gen5+ a HW context consists of an opaque GPU object which is
referenced at times of context saves and restores. With RC6 enabled,
the context is also referenced as the GPU enters and exists from RC6
(GPU has it's own internal power context, except on gen5). Though
something like a context does exist for the media ring, the code only
supports contexts for the render ring.
In software, there is a distinction between contexts created by the
user, and the default HW context. The default HW context is used by GPU
clients that do not request setup of their own hardware context. The
default context's state is never restored to help prevent programming
errors. This would happen if a client ran and piggy-backed off another
clients GPU state. The default context only exists to give the GPU some
offset to load as the current to invoke a save of the context we
actually care about. In fact, the code could likely be constructed,
albeit in a more complicated fashion, to never use the default context,
though that limits the driver's ability to swap out, and/or destroy
other contexts.
All other contexts are created as a request by the GPU client. These
contexts store GPU state, and thus allow GPU clients to not re-emit
state (and potentially query certain state) at any time. The kernel
driver makes certain that the appropriate commands are inserted.
There are 4 entry points into the contexts, init, fini, open, close.
The names are self-explanatory except that init can be called during
reset, and also during pm thaw/resume. As we expect our context to be
preserved across these events, we do not reinitialize in this case.
As Adam Jackson pointed out, The cutoff of 1MB where a HW context is
considered too big is arbitrary. The reason for this is even though
context sizes are increasing with every generation, they have yet to
eclipse even 32k. If we somehow read back way more than that, it
probably means BIOS has done something strange, or we're running on a
platform that wasn't designed for this.
v2: rename load/unload to init/fini (daniel)
remove ILK support for get_size() (indirectly daniel)
add HAS_HW_CONTEXTS macro to clarify supported platforms (daniel)
added comments (Ben)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2012-06-05 04:42:42 +07:00
|
|
|
uint32_t hw_context_size;
|
2013-09-18 11:12:45 +07:00
|
|
|
struct list_head context_list;
|
2012-11-03 01:55:02 +07:00
|
|
|
|
2012-12-12 01:48:29 +07:00
|
|
|
u32 fdi_rx_config;
|
2012-12-01 21:04:26 +07:00
|
|
|
|
2012-11-03 01:55:02 +07:00
|
|
|
struct i915_suspend_saved_registers regfile;
|
2012-11-03 01:55:05 +07:00
|
|
|
|
2013-08-01 20:18:50 +07:00
|
|
|
struct {
|
|
|
|
/*
|
|
|
|
* Raw watermark latency values:
|
|
|
|
* in 0.1us units for WM0,
|
|
|
|
* in 0.5us units for WM1+.
|
|
|
|
*/
|
|
|
|
/* primary */
|
|
|
|
uint16_t pri_latency[5];
|
|
|
|
/* sprite */
|
|
|
|
uint16_t spr_latency[5];
|
|
|
|
/* cursor */
|
|
|
|
uint16_t cur_latency[5];
|
2013-10-09 23:18:03 +07:00
|
|
|
|
|
|
|
/* current hardware state */
|
2013-12-17 19:46:36 +07:00
|
|
|
struct ilk_wm_values hw;
|
2013-08-01 20:18:50 +07:00
|
|
|
} wm;
|
|
|
|
|
2013-08-19 23:18:09 +07:00
|
|
|
struct i915_package_c8 pc8;
|
|
|
|
|
2013-12-07 05:32:13 +07:00
|
|
|
struct i915_runtime_pm pm;
|
|
|
|
|
2012-11-03 01:55:05 +07:00
|
|
|
/* Old dri1 support infrastructure, beware the dragons ya fools entering
|
|
|
|
* here! */
|
|
|
|
struct i915_dri1_state dri1;
|
2013-07-09 21:51:37 +07:00
|
|
|
/* Old ums support infrastructure, same warning applies. */
|
|
|
|
struct i915_ums_state ums;
|
2005-04-17 05:20:36 +07:00
|
|
|
} drm_i915_private_t;
|
|
|
|
|
2013-08-02 00:39:55 +07:00
|
|
|
static inline struct drm_i915_private *to_i915(const struct drm_device *dev)
|
|
|
|
{
|
|
|
|
return dev->dev_private;
|
|
|
|
}
|
|
|
|
|
2012-05-11 20:29:30 +07:00
|
|
|
/* Iterate over initialised rings */
|
|
|
|
#define for_each_ring(ring__, dev_priv__, i__) \
|
|
|
|
for ((i__) = 0; (i__) < I915_NUM_RINGS; (i__)++) \
|
|
|
|
if (((ring__) = &(dev_priv__)->ring[(i__)]), intel_ring_initialized((ring__)))
|
|
|
|
|
2012-02-14 10:45:36 +07:00
|
|
|
enum hdmi_force_audio {
|
|
|
|
HDMI_AUDIO_OFF_DVI = -2, /* no aux data for HDMI-DVI converter */
|
|
|
|
HDMI_AUDIO_OFF, /* force turn off HDMI audio */
|
|
|
|
HDMI_AUDIO_AUTO, /* trust EDID */
|
|
|
|
HDMI_AUDIO_ON, /* force turn on HDMI audio */
|
|
|
|
};
|
|
|
|
|
2013-07-04 18:06:28 +07:00
|
|
|
#define I915_GTT_OFFSET_NONE ((u32)-1)
|
2012-11-15 18:32:19 +07:00
|
|
|
|
2012-06-07 21:38:42 +07:00
|
|
|
struct drm_i915_gem_object_ops {
|
|
|
|
/* Interface between the GEM object and its backing storage.
|
|
|
|
* get_pages() is called once prior to the use of the associated set
|
|
|
|
* of pages before to binding them into the GTT, and put_pages() is
|
|
|
|
* called after we no longer need them. As we expect there to be
|
|
|
|
* associated cost with migrating pages between the backing storage
|
|
|
|
* and making them available for the GPU (e.g. clflush), we may hold
|
|
|
|
* onto the pages after they are no longer referenced by the GPU
|
|
|
|
* in case they may be used again shortly (for example migrating the
|
|
|
|
* pages to a different memory domain within the GTT). put_pages()
|
|
|
|
* will therefore most likely be called when the object itself is
|
|
|
|
* being released or under memory pressure (where we attempt to
|
|
|
|
* reap pages for the shrinker).
|
|
|
|
*/
|
|
|
|
int (*get_pages)(struct drm_i915_gem_object *);
|
|
|
|
void (*put_pages)(struct drm_i915_gem_object *);
|
|
|
|
};
|
|
|
|
|
2008-07-31 02:06:12 +07:00
|
|
|
struct drm_i915_gem_object {
|
2010-04-10 02:05:07 +07:00
|
|
|
struct drm_gem_object base;
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2012-06-07 21:38:42 +07:00
|
|
|
const struct drm_i915_gem_object_ops *ops;
|
|
|
|
|
2013-07-18 02:19:03 +07:00
|
|
|
/** List of VMAs backed by this object */
|
|
|
|
struct list_head vma_list;
|
|
|
|
|
2012-11-15 18:32:21 +07:00
|
|
|
/** Stolen memory for this object, instead of being backed by shmem. */
|
|
|
|
struct drm_mm_node *stolen;
|
2013-06-01 01:28:48 +07:00
|
|
|
struct list_head global_list;
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2010-10-19 16:36:51 +07:00
|
|
|
struct list_head ring_list;
|
2013-08-14 16:38:33 +07:00
|
|
|
/** Used in execbuf to temporarily hold a ref */
|
|
|
|
struct list_head obj_exec_link;
|
2008-07-31 02:06:12 +07:00
|
|
|
|
|
|
|
/**
|
2012-07-20 18:41:02 +07:00
|
|
|
* This is set if the object is on the active lists (has pending
|
|
|
|
* rendering and so a non-zero seqno), and is not set if it i s on
|
|
|
|
* inactive (ready to be unbound) list.
|
2008-07-31 02:06:12 +07:00
|
|
|
*/
|
2011-08-17 02:34:10 +07:00
|
|
|
unsigned int active:1;
|
2008-07-31 02:06:12 +07:00
|
|
|
|
|
|
|
/**
|
|
|
|
* This is set if the object has been written to since last bound
|
|
|
|
* to the GTT
|
|
|
|
*/
|
2011-08-17 02:34:10 +07:00
|
|
|
unsigned int dirty:1;
|
2010-05-13 16:49:44 +07:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Fence register bits (if any) for this object. Will be set
|
|
|
|
* as needed when mapped into the GTT.
|
|
|
|
* Protected by dev->struct_mutex.
|
|
|
|
*/
|
2011-10-10 02:52:02 +07:00
|
|
|
signed int fence_reg:I915_MAX_NUM_FENCE_BITS;
|
2010-05-13 16:49:44 +07:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Advice: are the backing pages purgeable?
|
|
|
|
*/
|
2011-08-17 02:34:10 +07:00
|
|
|
unsigned int madv:2;
|
2010-05-13 16:49:44 +07:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Current tiling mode for the object.
|
|
|
|
*/
|
2011-08-17 02:34:10 +07:00
|
|
|
unsigned int tiling_mode:2;
|
2012-04-21 22:23:23 +07:00
|
|
|
/**
|
|
|
|
* Whether the tiling parameters for the currently associated fence
|
|
|
|
* register have changed. Note that for the purposes of tracking
|
|
|
|
* tiling changes we also treat the unfenced register, the register
|
|
|
|
* slot that the object occupies whilst it executes a fenced
|
|
|
|
* command (such as BLT on gen2/3), as a "fence".
|
|
|
|
*/
|
|
|
|
unsigned int fence_dirty:1;
|
2010-05-13 16:49:44 +07:00
|
|
|
|
2010-11-04 23:11:09 +07:00
|
|
|
/**
|
|
|
|
* Is the object at the current location in the gtt mappable and
|
|
|
|
* fenceable? Used to avoid costly recalculations.
|
|
|
|
*/
|
2011-08-17 02:34:10 +07:00
|
|
|
unsigned int map_and_fenceable:1;
|
2010-11-04 23:11:09 +07:00
|
|
|
|
2010-10-02 03:05:20 +07:00
|
|
|
/**
|
|
|
|
* Whether the current gtt mapping needs to be mappable (and isn't just
|
|
|
|
* mappable by accident). Track pin and fault separate for a more
|
|
|
|
* accurate mappable working set.
|
|
|
|
*/
|
2011-08-17 02:34:10 +07:00
|
|
|
unsigned int fault_mappable:1;
|
|
|
|
unsigned int pin_mappable:1;
|
2013-08-09 18:25:09 +07:00
|
|
|
unsigned int pin_display:1;
|
2010-10-02 03:05:20 +07:00
|
|
|
|
2010-11-12 20:53:37 +07:00
|
|
|
/*
|
|
|
|
* Is the GPU currently using a fence to access this buffer,
|
|
|
|
*/
|
|
|
|
unsigned int pending_fenced_gpu_access:1;
|
|
|
|
unsigned int fenced_gpu_access:1;
|
|
|
|
|
2013-08-08 20:41:10 +07:00
|
|
|
unsigned int cache_level:3;
|
2011-03-30 06:59:50 +07:00
|
|
|
|
2012-02-09 23:15:47 +07:00
|
|
|
unsigned int has_aliasing_ppgtt_mapping:1;
|
2012-02-16 05:50:22 +07:00
|
|
|
unsigned int has_global_gtt_mapping:1;
|
2012-06-01 21:20:22 +07:00
|
|
|
unsigned int has_dma_mapping:1;
|
2012-02-09 23:15:47 +07:00
|
|
|
|
2012-06-01 21:20:22 +07:00
|
|
|
struct sg_table *pages;
|
2012-09-05 03:02:54 +07:00
|
|
|
int pages_pin_count;
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2012-05-10 20:25:09 +07:00
|
|
|
/* prime dma-buf support */
|
2012-05-22 19:09:21 +07:00
|
|
|
void *dma_buf_vmapping;
|
|
|
|
int vmapping_count;
|
|
|
|
|
2010-11-12 20:53:37 +07:00
|
|
|
struct intel_ring_buffer *ring;
|
|
|
|
|
2012-04-17 21:31:27 +07:00
|
|
|
/** Breadcrumb of last rendering to the buffer. */
|
2012-07-20 18:41:01 +07:00
|
|
|
uint32_t last_read_seqno;
|
|
|
|
uint32_t last_write_seqno;
|
2010-11-12 20:53:37 +07:00
|
|
|
/** Breadcrumb of last fenced GPU access to the buffer. */
|
|
|
|
uint32_t last_fenced_seqno;
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2010-05-13 16:49:44 +07:00
|
|
|
/** Current tiling stride for the object, if it's tiled. */
|
2008-11-13 01:03:55 +07:00
|
|
|
uint32_t stride;
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2013-10-10 02:23:52 +07:00
|
|
|
/** References from framebuffers, locks out tiling changes. */
|
|
|
|
unsigned long framebuffer_references;
|
|
|
|
|
2009-03-13 06:56:27 +07:00
|
|
|
/** Record of address bit 17 of each page at last unbind. */
|
2010-06-06 21:40:22 +07:00
|
|
|
unsigned long *bit_17;
|
2009-03-13 06:56:27 +07:00
|
|
|
|
DRM: i915: add mode setting support
This commit adds i915 driver support for the DRM mode setting APIs.
Currently, VGA, LVDS, SDVO DVI & VGA, TV and DVO LVDS outputs are
supported. HDMI, DisplayPort and additional SDVO output support will
follow.
Support for the mode setting code is controlled by the new 'modeset'
module option. A new config option, CONFIG_DRM_I915_KMS controls the
default behavior, and whether a PCI ID list is built into the module for
use by user level module utilities.
Note that if mode setting is enabled, user level drivers that access
display registers directly or that don't use the kernel graphics memory
manager will likely corrupt kernel graphics memory, disrupt output
configuration (possibly leading to hangs and/or blank displays), and
prevent panic/oops messages from appearing. So use caution when
enabling this code; be sure your user level code supports the new
interfaces.
A new SysRq key, 'g', provides emergency support for switching back to
the kernel's framebuffer console; which is useful for testing.
Co-authors: Dave Airlie <airlied@linux.ie>, Hong Liu <hong.liu@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2008-11-08 05:24:08 +07:00
|
|
|
/** User space pin count and filp owning the pin */
|
2013-10-10 19:46:37 +07:00
|
|
|
unsigned long user_pin_count;
|
DRM: i915: add mode setting support
This commit adds i915 driver support for the DRM mode setting APIs.
Currently, VGA, LVDS, SDVO DVI & VGA, TV and DVO LVDS outputs are
supported. HDMI, DisplayPort and additional SDVO output support will
follow.
Support for the mode setting code is controlled by the new 'modeset'
module option. A new config option, CONFIG_DRM_I915_KMS controls the
default behavior, and whether a PCI ID list is built into the module for
use by user level module utilities.
Note that if mode setting is enabled, user level drivers that access
display registers directly or that don't use the kernel graphics memory
manager will likely corrupt kernel graphics memory, disrupt output
configuration (possibly leading to hangs and/or blank displays), and
prevent panic/oops messages from appearing. So use caution when
enabling this code; be sure your user level code supports the new
interfaces.
A new SysRq key, 'g', provides emergency support for switching back to
the kernel's framebuffer console; which is useful for testing.
Co-authors: Dave Airlie <airlied@linux.ie>, Hong Liu <hong.liu@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2008-11-08 05:24:08 +07:00
|
|
|
struct drm_file *pin_filp;
|
2008-12-30 17:31:46 +07:00
|
|
|
|
|
|
|
/** for phy allocated objects */
|
|
|
|
struct drm_i915_gem_phys_object *phys_obj;
|
2008-07-31 02:06:12 +07:00
|
|
|
};
|
2012-12-17 22:21:27 +07:00
|
|
|
#define to_gem_object(obj) (&((struct drm_i915_gem_object *)(obj))->base)
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2010-04-10 02:05:08 +07:00
|
|
|
#define to_intel_bo(x) container_of(x, struct drm_i915_gem_object, base)
|
2010-03-08 19:35:02 +07:00
|
|
|
|
2008-07-31 02:06:12 +07:00
|
|
|
/**
|
|
|
|
* Request queue structure.
|
|
|
|
*
|
|
|
|
* The request queue allows us to note sequence numbers that have been emitted
|
|
|
|
* and may be associated with active buffers to be retired.
|
|
|
|
*
|
|
|
|
* By keeping this list, we can avoid having to do questionable
|
|
|
|
* sequence-number comparisons on buffer last_rendering_seqnos, and associate
|
|
|
|
* an emission time with seqnos for tracking how far ahead of the GPU we are.
|
|
|
|
*/
|
|
|
|
struct drm_i915_gem_request {
|
2010-05-21 08:08:56 +07:00
|
|
|
/** On Which ring this request was generated */
|
|
|
|
struct intel_ring_buffer *ring;
|
|
|
|
|
2008-07-31 02:06:12 +07:00
|
|
|
/** GEM sequence number associated with this request. */
|
|
|
|
uint32_t seqno;
|
|
|
|
|
2013-06-12 19:01:39 +07:00
|
|
|
/** Position in the ringbuffer of the start of the request */
|
|
|
|
u32 head;
|
|
|
|
|
|
|
|
/** Position in the ringbuffer of the end of the request */
|
2012-02-15 18:25:36 +07:00
|
|
|
u32 tail;
|
|
|
|
|
2013-05-02 20:48:08 +07:00
|
|
|
/** Context related to this request */
|
|
|
|
struct i915_hw_context *ctx;
|
|
|
|
|
2013-06-12 19:01:39 +07:00
|
|
|
/** Batch buffer related to this request if any */
|
|
|
|
struct drm_i915_gem_object *batch_obj;
|
|
|
|
|
2008-07-31 02:06:12 +07:00
|
|
|
/** Time at which this request was emitted, in jiffies. */
|
|
|
|
unsigned long emitted_jiffies;
|
|
|
|
|
2009-06-03 14:27:35 +07:00
|
|
|
/** global list entry for this request */
|
2008-07-31 02:06:12 +07:00
|
|
|
struct list_head list;
|
2009-06-03 14:27:35 +07:00
|
|
|
|
2010-09-24 22:02:42 +07:00
|
|
|
struct drm_i915_file_private *file_priv;
|
2009-06-03 14:27:35 +07:00
|
|
|
/** file_priv list entry for this request */
|
|
|
|
struct list_head client_list;
|
2008-07-31 02:06:12 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
struct drm_i915_file_private {
|
drm/i915: Boost RPS frequency for CPU stalls
If we encounter a situation where the CPU blocks waiting for results
from the GPU, give the GPU a kick to boost its the frequency.
This should work to reduce user interface stalls and to quickly promote
mesa to high frequencies - but the cost is that our requested frequency
stalls high (as we do not idle for long enough before rc6 to start
reducing frequencies, nor are we aggressive at down clocking an
underused GPU). However, this should be mitigated by rc6 itself powering
off the GPU when idle, and that energy use is dependent upon the workload
of the GPU in addition to its frequency (e.g. the math or sampler
functions only consume power when used). Still, this is likely to
adversely affect light workloads.
In particular, this nearly eliminates the highly noticeable wake-up lag
in animations from idle. For example, expose or workspace transitions.
(However, given the situation where we fail to downclock, our requested
frequency is almost always the maximum, except for Baytrail where we
manually downclock upon idling. This often masks the latency of
upclocking after being idle, so animations are typically smooth - at the
cost of increased power consumption.)
Stéphane raised the concern that this will punish good applications and
reward bad applications - but due to the nature of how mesa performs its
client throttling, I believe all mesa applications will be roughly
equally affected. To address this concern, and to prevent applications
like compositors from permanently boosting the RPS state, we ratelimit the
frequency of the wait-boosts each client recieves.
Unfortunately, this techinique is ineffective with Ironlake - which also
has dynamic render power states and suffers just as dramatically. For
Ironlake, the thermal/power headroom is shared with the CPU through
Intelligent Power Sharing and the intel-ips module. This leaves us with
no GPU boost frequencies available when coming out of idle, and due to
hardware limitations we cannot change the arbitration between the CPU and
GPU quickly enough to be effective.
v2: Limit each client to receiving a single boost for each active period.
Tested by QA to only marginally increase power, and to demonstrably
increase throughput in games. No latency measurements yet.
v3: Cater for front-buffer rendering with manual throttling.
v4: Tidy up.
v5: Sadly the compositor needs frequent boosts as it may never idle, but
due to its picking mechanism (using ReadPixels) may require frequent
waits. Those waits, along with the waits for the vrefresh swap, conspire
to keep the GPU at low frequencies despite the interactive latency. To
overcome this we ditch the one-boost-per-active-period and just ratelimit
the number of wait-boosts each client can receive.
Reported-and-tested-by: Paul Neumann <paul104x@yahoo.de>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=68716
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Stéphane Marchesin <stephane.marchesin@gmail.com>
Cc: Owen Taylor <otaylor@redhat.com>
Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com>
Cc: "Zhuang, Lena" <lena.zhuang@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[danvet: No extern for function prototypes in headers.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 23:34:56 +07:00
|
|
|
struct drm_i915_private *dev_priv;
|
|
|
|
|
2008-07-31 02:06:12 +07:00
|
|
|
struct {
|
2012-11-30 03:45:06 +07:00
|
|
|
spinlock_t lock;
|
2009-06-03 14:27:35 +07:00
|
|
|
struct list_head request_list;
|
drm/i915: Boost RPS frequency for CPU stalls
If we encounter a situation where the CPU blocks waiting for results
from the GPU, give the GPU a kick to boost its the frequency.
This should work to reduce user interface stalls and to quickly promote
mesa to high frequencies - but the cost is that our requested frequency
stalls high (as we do not idle for long enough before rc6 to start
reducing frequencies, nor are we aggressive at down clocking an
underused GPU). However, this should be mitigated by rc6 itself powering
off the GPU when idle, and that energy use is dependent upon the workload
of the GPU in addition to its frequency (e.g. the math or sampler
functions only consume power when used). Still, this is likely to
adversely affect light workloads.
In particular, this nearly eliminates the highly noticeable wake-up lag
in animations from idle. For example, expose or workspace transitions.
(However, given the situation where we fail to downclock, our requested
frequency is almost always the maximum, except for Baytrail where we
manually downclock upon idling. This often masks the latency of
upclocking after being idle, so animations are typically smooth - at the
cost of increased power consumption.)
Stéphane raised the concern that this will punish good applications and
reward bad applications - but due to the nature of how mesa performs its
client throttling, I believe all mesa applications will be roughly
equally affected. To address this concern, and to prevent applications
like compositors from permanently boosting the RPS state, we ratelimit the
frequency of the wait-boosts each client recieves.
Unfortunately, this techinique is ineffective with Ironlake - which also
has dynamic render power states and suffers just as dramatically. For
Ironlake, the thermal/power headroom is shared with the CPU through
Intelligent Power Sharing and the intel-ips module. This leaves us with
no GPU boost frequencies available when coming out of idle, and due to
hardware limitations we cannot change the arbitration between the CPU and
GPU quickly enough to be effective.
v2: Limit each client to receiving a single boost for each active period.
Tested by QA to only marginally increase power, and to demonstrably
increase throughput in games. No latency measurements yet.
v3: Cater for front-buffer rendering with manual throttling.
v4: Tidy up.
v5: Sadly the compositor needs frequent boosts as it may never idle, but
due to its picking mechanism (using ReadPixels) may require frequent
waits. Those waits, along with the waits for the vrefresh swap, conspire
to keep the GPU at low frequencies despite the interactive latency. To
overcome this we ditch the one-boost-per-active-period and just ratelimit
the number of wait-boosts each client can receive.
Reported-and-tested-by: Paul Neumann <paul104x@yahoo.de>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=68716
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Stéphane Marchesin <stephane.marchesin@gmail.com>
Cc: Owen Taylor <otaylor@redhat.com>
Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com>
Cc: "Zhuang, Lena" <lena.zhuang@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[danvet: No extern for function prototypes in headers.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 23:34:56 +07:00
|
|
|
struct delayed_work idle_work;
|
2008-07-31 02:06:12 +07:00
|
|
|
} mm;
|
2012-06-05 04:42:43 +07:00
|
|
|
struct idr context_idr;
|
2013-06-12 16:35:28 +07:00
|
|
|
|
2013-12-07 05:11:19 +07:00
|
|
|
struct i915_hw_context *private_default_ctx;
|
drm/i915: Boost RPS frequency for CPU stalls
If we encounter a situation where the CPU blocks waiting for results
from the GPU, give the GPU a kick to boost its the frequency.
This should work to reduce user interface stalls and to quickly promote
mesa to high frequencies - but the cost is that our requested frequency
stalls high (as we do not idle for long enough before rc6 to start
reducing frequencies, nor are we aggressive at down clocking an
underused GPU). However, this should be mitigated by rc6 itself powering
off the GPU when idle, and that energy use is dependent upon the workload
of the GPU in addition to its frequency (e.g. the math or sampler
functions only consume power when used). Still, this is likely to
adversely affect light workloads.
In particular, this nearly eliminates the highly noticeable wake-up lag
in animations from idle. For example, expose or workspace transitions.
(However, given the situation where we fail to downclock, our requested
frequency is almost always the maximum, except for Baytrail where we
manually downclock upon idling. This often masks the latency of
upclocking after being idle, so animations are typically smooth - at the
cost of increased power consumption.)
Stéphane raised the concern that this will punish good applications and
reward bad applications - but due to the nature of how mesa performs its
client throttling, I believe all mesa applications will be roughly
equally affected. To address this concern, and to prevent applications
like compositors from permanently boosting the RPS state, we ratelimit the
frequency of the wait-boosts each client recieves.
Unfortunately, this techinique is ineffective with Ironlake - which also
has dynamic render power states and suffers just as dramatically. For
Ironlake, the thermal/power headroom is shared with the CPU through
Intelligent Power Sharing and the intel-ips module. This leaves us with
no GPU boost frequencies available when coming out of idle, and due to
hardware limitations we cannot change the arbitration between the CPU and
GPU quickly enough to be effective.
v2: Limit each client to receiving a single boost for each active period.
Tested by QA to only marginally increase power, and to demonstrably
increase throughput in games. No latency measurements yet.
v3: Cater for front-buffer rendering with manual throttling.
v4: Tidy up.
v5: Sadly the compositor needs frequent boosts as it may never idle, but
due to its picking mechanism (using ReadPixels) may require frequent
waits. Those waits, along with the waits for the vrefresh swap, conspire
to keep the GPU at low frequencies despite the interactive latency. To
overcome this we ditch the one-boost-per-active-period and just ratelimit
the number of wait-boosts each client can receive.
Reported-and-tested-by: Paul Neumann <paul104x@yahoo.de>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=68716
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Stéphane Marchesin <stephane.marchesin@gmail.com>
Cc: Owen Taylor <otaylor@redhat.com>
Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com>
Cc: "Zhuang, Lena" <lena.zhuang@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[danvet: No extern for function prototypes in headers.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 23:34:56 +07:00
|
|
|
atomic_t rps_wait_boost;
|
2008-07-31 02:06:12 +07:00
|
|
|
};
|
|
|
|
|
2014-02-08 02:12:48 +07:00
|
|
|
#define INTEL_INFO(dev) (&to_i915(dev)->info)
|
2010-11-09 16:17:32 +07:00
|
|
|
|
2013-10-04 18:53:40 +07:00
|
|
|
#define IS_I830(dev) ((dev)->pdev->device == 0x3577)
|
|
|
|
#define IS_845G(dev) ((dev)->pdev->device == 0x2562)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define IS_I85X(dev) (INTEL_INFO(dev)->is_i85x)
|
2013-10-04 18:53:40 +07:00
|
|
|
#define IS_I865G(dev) ((dev)->pdev->device == 0x2572)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define IS_I915G(dev) (INTEL_INFO(dev)->is_i915g)
|
2013-10-04 18:53:40 +07:00
|
|
|
#define IS_I915GM(dev) ((dev)->pdev->device == 0x2592)
|
|
|
|
#define IS_I945G(dev) ((dev)->pdev->device == 0x2772)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define IS_I945GM(dev) (INTEL_INFO(dev)->is_i945gm)
|
|
|
|
#define IS_BROADWATER(dev) (INTEL_INFO(dev)->is_broadwater)
|
|
|
|
#define IS_CRESTLINE(dev) (INTEL_INFO(dev)->is_crestline)
|
2013-10-04 18:53:40 +07:00
|
|
|
#define IS_GM45(dev) ((dev)->pdev->device == 0x2A42)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define IS_G4X(dev) (INTEL_INFO(dev)->is_g4x)
|
2013-10-04 18:53:40 +07:00
|
|
|
#define IS_PINEVIEW_G(dev) ((dev)->pdev->device == 0xa001)
|
|
|
|
#define IS_PINEVIEW_M(dev) ((dev)->pdev->device == 0xa011)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define IS_PINEVIEW(dev) (INTEL_INFO(dev)->is_pineview)
|
|
|
|
#define IS_G33(dev) (INTEL_INFO(dev)->is_g33)
|
2013-10-04 18:53:40 +07:00
|
|
|
#define IS_IRONLAKE_M(dev) ((dev)->pdev->device == 0x0046)
|
2011-04-29 04:33:09 +07:00
|
|
|
#define IS_IVYBRIDGE(dev) (INTEL_INFO(dev)->is_ivybridge)
|
2013-10-04 18:53:40 +07:00
|
|
|
#define IS_IVB_GT1(dev) ((dev)->pdev->device == 0x0156 || \
|
|
|
|
(dev)->pdev->device == 0x0152 || \
|
|
|
|
(dev)->pdev->device == 0x015a)
|
|
|
|
#define IS_SNB_GT1(dev) ((dev)->pdev->device == 0x0102 || \
|
|
|
|
(dev)->pdev->device == 0x0106 || \
|
|
|
|
(dev)->pdev->device == 0x010A)
|
2012-03-29 03:39:21 +07:00
|
|
|
#define IS_VALLEYVIEW(dev) (INTEL_INFO(dev)->is_valleyview)
|
2012-03-29 22:32:18 +07:00
|
|
|
#define IS_HASWELL(dev) (INTEL_INFO(dev)->is_haswell)
|
2013-11-03 11:07:31 +07:00
|
|
|
#define IS_BROADWELL(dev) (INTEL_INFO(dev)->gen == 8)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define IS_MOBILE(dev) (INTEL_INFO(dev)->is_mobile)
|
2013-08-13 00:34:08 +07:00
|
|
|
#define IS_HSW_EARLY_SDV(dev) (IS_HASWELL(dev) && \
|
2013-10-04 18:53:40 +07:00
|
|
|
((dev)->pdev->device & 0xFF00) == 0x0C00)
|
2013-11-09 01:20:06 +07:00
|
|
|
#define IS_BDW_ULT(dev) (IS_BROADWELL(dev) && \
|
|
|
|
(((dev)->pdev->device & 0xf) == 0x2 || \
|
|
|
|
((dev)->pdev->device & 0xf) == 0x6 || \
|
|
|
|
((dev)->pdev->device & 0xf) == 0xe))
|
|
|
|
#define IS_HSW_ULT(dev) (IS_HASWELL(dev) && \
|
2013-10-04 18:53:40 +07:00
|
|
|
((dev)->pdev->device & 0xFF00) == 0x0A00)
|
2013-11-09 01:20:06 +07:00
|
|
|
#define IS_ULT(dev) (IS_HSW_ULT(dev) || IS_BDW_ULT(dev))
|
2013-08-29 02:45:46 +07:00
|
|
|
#define IS_HSW_GT3(dev) (IS_HASWELL(dev) && \
|
2013-10-04 18:53:40 +07:00
|
|
|
((dev)->pdev->device & 0x00F0) == 0x0020)
|
2013-08-24 06:00:07 +07:00
|
|
|
#define IS_PRELIMINARY_HW(intel_info) ((intel_info)->is_preliminary)
|
2010-11-09 16:17:32 +07:00
|
|
|
|
2011-04-07 02:11:14 +07:00
|
|
|
/*
|
|
|
|
* The genX designation typically refers to the render engine, so render
|
|
|
|
* capability related checks should use IS_GEN, while display and other checks
|
|
|
|
* have their own (e.g. HAS_PCH_SPLIT for ILK+ display, IS_foo for particular
|
|
|
|
* chips, etc.).
|
|
|
|
*/
|
2010-11-09 16:17:32 +07:00
|
|
|
#define IS_GEN2(dev) (INTEL_INFO(dev)->gen == 2)
|
|
|
|
#define IS_GEN3(dev) (INTEL_INFO(dev)->gen == 3)
|
|
|
|
#define IS_GEN4(dev) (INTEL_INFO(dev)->gen == 4)
|
|
|
|
#define IS_GEN5(dev) (INTEL_INFO(dev)->gen == 5)
|
|
|
|
#define IS_GEN6(dev) (INTEL_INFO(dev)->gen == 6)
|
2011-04-07 02:11:14 +07:00
|
|
|
#define IS_GEN7(dev) (INTEL_INFO(dev)->gen == 7)
|
2013-11-03 11:06:59 +07:00
|
|
|
#define IS_GEN8(dev) (INTEL_INFO(dev)->gen == 8)
|
2010-11-09 16:17:32 +07:00
|
|
|
|
2013-10-16 00:02:57 +07:00
|
|
|
#define RENDER_RING (1<<RCS)
|
|
|
|
#define BSD_RING (1<<VCS)
|
|
|
|
#define BLT_RING (1<<BCS)
|
|
|
|
#define VEBOX_RING (1<<VECS)
|
|
|
|
#define HAS_BSD(dev) (INTEL_INFO(dev)->ring_mask & BSD_RING)
|
|
|
|
#define HAS_BLT(dev) (INTEL_INFO(dev)->ring_mask & BLT_RING)
|
|
|
|
#define HAS_VEBOX(dev) (INTEL_INFO(dev)->ring_mask & VEBOX_RING)
|
2012-01-17 23:43:53 +07:00
|
|
|
#define HAS_LLC(dev) (INTEL_INFO(dev)->has_llc)
|
2013-08-08 20:41:10 +07:00
|
|
|
#define HAS_WT(dev) (IS_HASWELL(dev) && to_i915(dev)->ellc_size)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define I915_NEED_GFX_HWS(dev) (INTEL_INFO(dev)->need_gfx_hws)
|
|
|
|
|
drm/i915: preliminary context support
Very basic code for context setup/destruction in the driver.
Adds the file i915_gem_context.c This file implements HW context
support. On gen5+ a HW context consists of an opaque GPU object which is
referenced at times of context saves and restores. With RC6 enabled,
the context is also referenced as the GPU enters and exists from RC6
(GPU has it's own internal power context, except on gen5). Though
something like a context does exist for the media ring, the code only
supports contexts for the render ring.
In software, there is a distinction between contexts created by the
user, and the default HW context. The default HW context is used by GPU
clients that do not request setup of their own hardware context. The
default context's state is never restored to help prevent programming
errors. This would happen if a client ran and piggy-backed off another
clients GPU state. The default context only exists to give the GPU some
offset to load as the current to invoke a save of the context we
actually care about. In fact, the code could likely be constructed,
albeit in a more complicated fashion, to never use the default context,
though that limits the driver's ability to swap out, and/or destroy
other contexts.
All other contexts are created as a request by the GPU client. These
contexts store GPU state, and thus allow GPU clients to not re-emit
state (and potentially query certain state) at any time. The kernel
driver makes certain that the appropriate commands are inserted.
There are 4 entry points into the contexts, init, fini, open, close.
The names are self-explanatory except that init can be called during
reset, and also during pm thaw/resume. As we expect our context to be
preserved across these events, we do not reinitialize in this case.
As Adam Jackson pointed out, The cutoff of 1MB where a HW context is
considered too big is arbitrary. The reason for this is even though
context sizes are increasing with every generation, they have yet to
eclipse even 32k. If we somehow read back way more than that, it
probably means BIOS has done something strange, or we're running on a
platform that wasn't designed for this.
v2: rename load/unload to init/fini (daniel)
remove ILK support for get_size() (indirectly daniel)
add HAS_HW_CONTEXTS macro to clarify supported platforms (daniel)
added comments (Ben)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2012-06-05 04:42:42 +07:00
|
|
|
#define HAS_HW_CONTEXTS(dev) (INTEL_INFO(dev)->gen >= 6)
|
2013-12-07 05:11:14 +07:00
|
|
|
#define HAS_ALIASING_PPGTT(dev) (INTEL_INFO(dev)->gen >= 6 && !IS_VALLEYVIEW(dev))
|
2014-01-28 14:07:00 +07:00
|
|
|
#define HAS_PPGTT(dev) (INTEL_INFO(dev)->gen >= 7 && !IS_VALLEYVIEW(dev) \
|
|
|
|
&& !IS_BROADWELL(dev))
|
|
|
|
#define USES_PPGTT(dev) intel_enable_ppgtt(dev, false)
|
2013-12-07 05:11:26 +07:00
|
|
|
#define USES_FULL_PPGTT(dev) intel_enable_ppgtt(dev, true)
|
2012-02-09 23:15:46 +07:00
|
|
|
|
2010-11-09 02:18:58 +07:00
|
|
|
#define HAS_OVERLAY(dev) (INTEL_INFO(dev)->has_overlay)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define OVERLAY_NEEDS_PHYSICAL(dev) (INTEL_INFO(dev)->overlay_needs_physical)
|
|
|
|
|
2012-12-17 22:21:27 +07:00
|
|
|
/* Early gen2 have a totally busted CS tlb and require pinned batches. */
|
|
|
|
#define HAS_BROKEN_CS_TLB(dev) (IS_I830(dev) || IS_845G(dev))
|
|
|
|
|
2010-11-09 16:17:32 +07:00
|
|
|
/* With the 945 and later, Y tiling got adjusted so that it was 32 128-byte
|
|
|
|
* rows, which changed the alignment requirements and fence programming.
|
|
|
|
*/
|
|
|
|
#define HAS_128_BYTE_Y_TILING(dev) (!IS_GEN2(dev) && !(IS_I915G(dev) || \
|
|
|
|
IS_I915GM(dev)))
|
|
|
|
#define SUPPORTS_DIGITAL_OUTPUTS(dev) (!IS_GEN2(dev) && !IS_PINEVIEW(dev))
|
|
|
|
#define SUPPORTS_INTEGRATED_HDMI(dev) (IS_G4X(dev) || IS_GEN5(dev))
|
|
|
|
#define SUPPORTS_INTEGRATED_DP(dev) (IS_G4X(dev) || IS_GEN5(dev))
|
|
|
|
#define SUPPORTS_TV(dev) (INTEL_INFO(dev)->supports_tv)
|
|
|
|
#define I915_HAS_HOTPLUG(dev) (INTEL_INFO(dev)->has_hotplug)
|
|
|
|
|
|
|
|
#define HAS_FW_BLC(dev) (INTEL_INFO(dev)->gen > 2)
|
|
|
|
#define HAS_PIPE_CXSR(dev) (INTEL_INFO(dev)->has_pipe_cxsr)
|
2014-01-10 14:50:12 +07:00
|
|
|
#define HAS_FBC(dev) (INTEL_INFO(dev)->has_fbc)
|
2010-11-09 16:17:32 +07:00
|
|
|
|
2013-11-03 11:07:47 +07:00
|
|
|
#define HAS_IPS(dev) (IS_ULT(dev) || IS_BROADWELL(dev))
|
2013-06-25 00:29:34 +07:00
|
|
|
|
2013-04-23 00:40:39 +07:00
|
|
|
#define HAS_DDI(dev) (INTEL_INFO(dev)->has_ddi)
|
2013-04-23 00:40:41 +07:00
|
|
|
#define HAS_FPGA_DBG_UNCLAIMED(dev) (INTEL_INFO(dev)->has_fpga_dbg)
|
2013-11-05 13:45:05 +07:00
|
|
|
#define HAS_PSR(dev) (IS_HASWELL(dev) || IS_BROADWELL(dev))
|
2013-11-19 09:32:37 +07:00
|
|
|
#define HAS_PC8(dev) (IS_HASWELL(dev)) /* XXX HSW:ULX */
|
2013-12-14 00:22:32 +07:00
|
|
|
#define HAS_RUNTIME_PM(dev) (IS_HASWELL(dev))
|
2012-11-24 00:30:39 +07:00
|
|
|
|
2012-11-21 00:12:07 +07:00
|
|
|
#define INTEL_PCH_DEVICE_ID_MASK 0xff00
|
|
|
|
#define INTEL_PCH_IBX_DEVICE_ID_TYPE 0x3b00
|
|
|
|
#define INTEL_PCH_CPT_DEVICE_ID_TYPE 0x1c00
|
|
|
|
#define INTEL_PCH_PPT_DEVICE_ID_TYPE 0x1e00
|
|
|
|
#define INTEL_PCH_LPT_DEVICE_ID_TYPE 0x8c00
|
|
|
|
#define INTEL_PCH_LPT_LP_DEVICE_ID_TYPE 0x9c00
|
|
|
|
|
2013-08-02 00:39:55 +07:00
|
|
|
#define INTEL_PCH_TYPE(dev) (to_i915(dev)->pch_type)
|
2012-03-29 22:32:20 +07:00
|
|
|
#define HAS_PCH_LPT(dev) (INTEL_PCH_TYPE(dev) == PCH_LPT)
|
2010-11-09 16:17:32 +07:00
|
|
|
#define HAS_PCH_CPT(dev) (INTEL_PCH_TYPE(dev) == PCH_CPT)
|
|
|
|
#define HAS_PCH_IBX(dev) (INTEL_PCH_TYPE(dev) == PCH_IBX)
|
2013-04-06 03:12:40 +07:00
|
|
|
#define HAS_PCH_NOP(dev) (INTEL_PCH_TYPE(dev) == PCH_NOP)
|
2012-07-04 01:57:32 +07:00
|
|
|
#define HAS_PCH_SPLIT(dev) (INTEL_PCH_TYPE(dev) != PCH_NONE)
|
2010-11-09 16:17:32 +07:00
|
|
|
|
2013-09-20 01:01:40 +07:00
|
|
|
/* DPF == dynamic parity feature */
|
|
|
|
#define HAS_L3_DPF(dev) (IS_IVYBRIDGE(dev) || IS_HASWELL(dev))
|
|
|
|
#define NUM_L3_SLICES(dev) (IS_HSW_GT3(dev) ? 2 : HAS_L3_DPF(dev))
|
2012-07-25 10:47:31 +07:00
|
|
|
|
2012-09-08 09:43:39 +07:00
|
|
|
#define GT_FREQUENCY_MULTIPLIER 50
|
|
|
|
|
2010-11-09 02:18:58 +07:00
|
|
|
#include "i915_trace.h"
|
|
|
|
|
2013-08-03 00:27:49 +07:00
|
|
|
extern const struct drm_ioctl_desc i915_ioctls[];
|
2005-09-30 15:37:36 +07:00
|
|
|
extern int i915_max_ioctl;
|
|
|
|
|
2010-02-01 12:38:10 +07:00
|
|
|
extern int i915_suspend(struct drm_device *dev, pm_message_t state);
|
|
|
|
extern int i915_resume(struct drm_device *dev);
|
2008-11-28 11:22:24 +07:00
|
|
|
extern int i915_master_create(struct drm_device *dev, struct drm_master *master);
|
|
|
|
extern void i915_master_destroy(struct drm_device *dev, struct drm_master *master);
|
|
|
|
|
2014-01-21 16:24:25 +07:00
|
|
|
/* i915_params.c */
|
|
|
|
struct i915_params {
|
|
|
|
int modeset;
|
|
|
|
int panel_ignore_lid;
|
|
|
|
unsigned int powersave;
|
|
|
|
int semaphores;
|
|
|
|
unsigned int lvds_downclock;
|
|
|
|
int lvds_channel_mode;
|
|
|
|
int panel_use_ssc;
|
|
|
|
int vbt_sdvo_panel_type;
|
|
|
|
int enable_rc6;
|
|
|
|
int enable_fbc;
|
|
|
|
int enable_ppgtt;
|
|
|
|
int enable_psr;
|
|
|
|
unsigned int preliminary_hw_support;
|
|
|
|
int disable_power_well;
|
|
|
|
int enable_ips;
|
|
|
|
int enable_pc8;
|
|
|
|
int pc8_timeout;
|
2014-02-08 02:12:53 +07:00
|
|
|
int invert_brightness;
|
|
|
|
/* leave bools at the end to not create holes */
|
|
|
|
bool enable_hangcheck;
|
|
|
|
bool fastboot;
|
2014-01-21 16:24:25 +07:00
|
|
|
bool prefault_disable;
|
|
|
|
bool reset;
|
2014-02-11 00:20:55 +07:00
|
|
|
bool disable_display;
|
2014-01-21 16:24:25 +07:00
|
|
|
};
|
|
|
|
extern struct i915_params i915 __read_mostly;
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/* i915_dma.c */
|
2012-04-27 04:28:09 +07:00
|
|
|
void i915_update_dri1_breadcrumb(struct drm_device *dev);
|
2007-07-11 12:53:27 +07:00
|
|
|
extern void i915_kernel_lost_context(struct drm_device * dev);
|
2005-11-10 18:16:34 +07:00
|
|
|
extern int i915_driver_load(struct drm_device *, unsigned long flags);
|
2007-11-22 11:14:14 +07:00
|
|
|
extern int i915_driver_unload(struct drm_device *);
|
2008-07-31 02:06:12 +07:00
|
|
|
extern int i915_driver_open(struct drm_device *dev, struct drm_file *file_priv);
|
2007-07-11 12:53:27 +07:00
|
|
|
extern void i915_driver_lastclose(struct drm_device * dev);
|
2007-08-25 17:23:09 +07:00
|
|
|
extern void i915_driver_preclose(struct drm_device *dev,
|
|
|
|
struct drm_file *file_priv);
|
2008-07-31 02:06:12 +07:00
|
|
|
extern void i915_driver_postclose(struct drm_device *dev,
|
|
|
|
struct drm_file *file_priv);
|
2007-07-11 12:53:27 +07:00
|
|
|
extern int i915_driver_device_is_agp(struct drm_device * dev);
|
2012-04-17 04:07:40 +07:00
|
|
|
#ifdef CONFIG_COMPAT
|
2006-01-02 16:14:23 +07:00
|
|
|
extern long i915_compat_ioctl(struct file *filp, unsigned int cmd,
|
|
|
|
unsigned long arg);
|
2012-04-17 04:07:40 +07:00
|
|
|
#endif
|
2008-07-31 02:06:12 +07:00
|
|
|
extern int i915_emit_box(struct drm_device *dev,
|
2010-11-30 21:10:25 +07:00
|
|
|
struct drm_clip_rect *box,
|
|
|
|
int DR1, int DR4);
|
2012-06-05 04:42:56 +07:00
|
|
|
extern int intel_gpu_reset(struct drm_device *dev);
|
2012-04-27 20:17:44 +07:00
|
|
|
extern int i915_reset(struct drm_device *dev);
|
2010-05-21 04:28:11 +07:00
|
|
|
extern unsigned long i915_chipset_val(struct drm_i915_private *dev_priv);
|
|
|
|
extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);
|
|
|
|
extern unsigned long i915_gfx_val(struct drm_i915_private *dev_priv);
|
|
|
|
extern void i915_update_gfx_val(struct drm_i915_private *dev_priv);
|
|
|
|
|
2012-11-03 01:13:59 +07:00
|
|
|
extern void intel_console_resume(struct work_struct *work);
|
2008-05-07 09:15:39 +07:00
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/* i915_irq.c */
|
2013-07-03 21:22:08 +07:00
|
|
|
void i915_queue_hangcheck(struct drm_device *dev);
|
2010-11-11 08:16:58 +07:00
|
|
|
void i915_handle_error(struct drm_device *dev, bool wedged);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2014-01-31 00:38:16 +07:00
|
|
|
void gen6_set_pm_mask(struct drm_i915_private *dev_priv, u32 pm_iir,
|
|
|
|
int new_delay);
|
2011-06-29 03:00:41 +07:00
|
|
|
extern void intel_irq_init(struct drm_device *dev);
|
2012-12-11 20:05:07 +07:00
|
|
|
extern void intel_hpd_init(struct drm_device *dev);
|
2013-07-20 02:36:52 +07:00
|
|
|
|
|
|
|
extern void intel_uncore_sanitize(struct drm_device *dev);
|
|
|
|
extern void intel_uncore_early_sanitize(struct drm_device *dev);
|
|
|
|
extern void intel_uncore_init(struct drm_device *dev);
|
|
|
|
extern void intel_uncore_check_errors(struct drm_device *dev);
|
2013-08-26 19:46:09 +07:00
|
|
|
extern void intel_uncore_fini(struct drm_device *dev);
|
2011-04-07 02:13:38 +07:00
|
|
|
|
2008-11-04 17:03:27 +07:00
|
|
|
void
|
2014-02-10 23:42:47 +07:00
|
|
|
i915_enable_pipestat(drm_i915_private_t *dev_priv, enum pipe pipe,
|
|
|
|
u32 status_mask);
|
2008-11-04 17:03:27 +07:00
|
|
|
|
|
|
|
void
|
2014-02-10 23:42:47 +07:00
|
|
|
i915_disable_pipestat(drm_i915_private_t *dev_priv, enum pipe pipe,
|
|
|
|
u32 status_mask);
|
2008-11-04 17:03:27 +07:00
|
|
|
|
2008-07-31 02:06:12 +07:00
|
|
|
/* i915_gem.c */
|
|
|
|
int i915_gem_init_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_create_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_pread_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_pwrite_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_mmap_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2008-11-13 01:03:55 +07:00
|
|
|
int i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2008-07-31 02:06:12 +07:00
|
|
|
int i915_gem_set_domain_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_sw_finish_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_execbuffer(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2009-12-18 10:05:42 +07:00
|
|
|
int i915_gem_execbuffer2(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2008-07-31 02:06:12 +07:00
|
|
|
int i915_gem_pin_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_unpin_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_busy_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2012-09-22 07:01:20 +07:00
|
|
|
int i915_gem_get_caching_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file);
|
|
|
|
int i915_gem_set_caching_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file);
|
2008-07-31 02:06:12 +07:00
|
|
|
int i915_gem_throttle_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2009-09-14 22:50:29 +07:00
|
|
|
int i915_gem_madvise_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2008-07-31 02:06:12 +07:00
|
|
|
int i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_leavevt_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_set_tiling(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
|
|
|
int i915_gem_get_tiling(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2008-10-23 11:40:13 +07:00
|
|
|
int i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2012-05-25 05:03:10 +07:00
|
|
|
int i915_gem_wait_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv);
|
2008-07-31 02:06:12 +07:00
|
|
|
void i915_gem_load(struct drm_device *dev);
|
2012-11-15 18:32:30 +07:00
|
|
|
void *i915_gem_object_alloc(struct drm_device *dev);
|
|
|
|
void i915_gem_object_free(struct drm_i915_gem_object *obj);
|
2012-06-07 21:38:42 +07:00
|
|
|
void i915_gem_object_init(struct drm_i915_gem_object *obj,
|
|
|
|
const struct drm_i915_gem_object_ops *ops);
|
2010-11-09 02:18:58 +07:00
|
|
|
struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev,
|
|
|
|
size_t size);
|
2013-12-07 05:11:26 +07:00
|
|
|
void i915_init_vm(struct drm_i915_private *dev_priv,
|
|
|
|
struct i915_address_space *vm);
|
2008-07-31 02:06:12 +07:00
|
|
|
void i915_gem_free_object(struct drm_gem_object *obj);
|
2013-07-18 02:19:03 +07:00
|
|
|
void i915_gem_vma_destroy(struct i915_vma *vma);
|
2012-11-15 18:32:30 +07:00
|
|
|
|
2014-02-14 20:01:11 +07:00
|
|
|
#define PIN_MAPPABLE 0x1
|
|
|
|
#define PIN_NONBLOCK 0x2
|
2014-02-14 20:01:12 +07:00
|
|
|
#define PIN_GLOBAL 0x4
|
2010-11-23 22:26:33 +07:00
|
|
|
int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,
|
2013-08-01 06:59:58 +07:00
|
|
|
struct i915_address_space *vm,
|
2010-11-23 22:26:33 +07:00
|
|
|
uint32_t alignment,
|
2014-02-14 20:01:11 +07:00
|
|
|
unsigned flags);
|
drm/i915: plumb VM into bind/unbind code
As alluded to in several patches, and it will be reiterated later... A
VMA is an abstraction for a GEM BO bound into an address space.
Therefore it stands to reason, that the existing bind, and unbind are
the ones which will be the most impacted. This patch implements this,
and updates all callers which weren't already updated in the series
(because it was too messy).
This patch represents the bulk of an earlier, larger patch. I've pulled
out a bunch of things by the request of Daniel. The history is preserved
for posterity with the email convention of ">" One big change from the
original patch aside from a bunch of cropping is I've created an
i915_vma_unbind() function. That is because we always have the VMA
anyway, and doing an extra lookup is useful. There is a caveat, we
retain an i915_gem_object_ggtt_unbind, for the global cases which might
not talk in VMAs.
> drm/i915: plumb VM into object operations
>
> This patch was formerly known as:
> "drm/i915: Create VMAs (part 3) - plumbing"
>
> This patch adds a VM argument, bind/unbind, and the object
> offset/size/color getters/setters. It preserves the old ggtt helper
> functions because things still need, and will continue to need them.
>
> Some code will still need to be ported over after this.
>
> v2: Fix purge to pick an object and unbind all vmas
> This was doable because of the global bound list change.
>
> v3: With the commit to actually pin/unpin pages in place, there is no
> longer a need to check if unbind succeeded before calling put_pages().
> Make put_pages only BUG() after checking pin count.
>
> v4: Rebased on top of the new hangcheck work by Mika
> plumbed eb_destroy also
> Many checkpatch related fixes
>
> v5: Very large rebase
>
> v6:
> Change BUG_ON to WARN_ON (Daniel)
> Rename vm to ggtt in preallocate stolen, since it is always ggtt when
> dealing with stolen memory. (Daniel)
> list_for_each will short-circuit already (Daniel)
> remove superflous space (Daniel)
> Use per object list of vmas (Daniel)
> Make obj_bound_any() use obj_bound for each vm (Ben)
> s/bind_to_gtt/bind_to_vm/ (Ben)
>
> Fixed up the inactive shrinker. As Daniel noticed the code could
> potentially count the same object multiple times. While it's not
> possible in the current case, since 1 object can only ever be bound into
> 1 address space thus far - we may as well try to get something more
> future proof in place now. With a prep patch before this to switch over
> to using the bound list + inactive check, we're now able to carry that
> forward for every address space an object is bound into.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: Rebase on top of the loss of "drm/i915: Cleanup more of VMA
in destroy".]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-08-01 07:00:10 +07:00
|
|
|
int __must_check i915_vma_unbind(struct i915_vma *vma);
|
2013-01-15 19:39:35 +07:00
|
|
|
int i915_gem_object_put_pages(struct drm_i915_gem_object *obj);
|
2013-12-14 00:22:31 +07:00
|
|
|
void i915_gem_release_all_mmaps(struct drm_i915_private *dev_priv);
|
2010-11-09 02:18:58 +07:00
|
|
|
void i915_gem_release_mmap(struct drm_i915_gem_object *obj);
|
2008-07-31 02:06:12 +07:00
|
|
|
void i915_gem_lastclose(struct drm_device *dev);
|
2010-09-24 22:02:42 +07:00
|
|
|
|
2012-06-07 21:38:42 +07:00
|
|
|
int __must_check i915_gem_object_get_pages(struct drm_i915_gem_object *obj);
|
2012-06-01 21:20:22 +07:00
|
|
|
static inline struct page *i915_gem_object_get_page(struct drm_i915_gem_object *obj, int n)
|
|
|
|
{
|
2013-02-19 00:28:02 +07:00
|
|
|
struct sg_page_iter sg_iter;
|
|
|
|
|
|
|
|
for_each_sg_page(obj->pages->sgl, &sg_iter, obj->pages->nents, n)
|
2013-03-26 20:14:18 +07:00
|
|
|
return sg_page_iter_page(&sg_iter);
|
2013-02-19 00:28:02 +07:00
|
|
|
|
|
|
|
return NULL;
|
2012-06-01 21:20:22 +07:00
|
|
|
}
|
2012-09-05 03:02:54 +07:00
|
|
|
static inline void i915_gem_object_pin_pages(struct drm_i915_gem_object *obj)
|
|
|
|
{
|
|
|
|
BUG_ON(obj->pages == NULL);
|
|
|
|
obj->pages_pin_count++;
|
|
|
|
}
|
|
|
|
static inline void i915_gem_object_unpin_pages(struct drm_i915_gem_object *obj)
|
|
|
|
{
|
|
|
|
BUG_ON(obj->pages_pin_count == 0);
|
|
|
|
obj->pages_pin_count--;
|
|
|
|
}
|
|
|
|
|
2010-11-26 01:00:26 +07:00
|
|
|
int __must_check i915_mutex_lock_interruptible(struct drm_device *dev);
|
2012-04-06 04:47:36 +07:00
|
|
|
int i915_gem_object_sync(struct drm_i915_gem_object *obj,
|
|
|
|
struct intel_ring_buffer *to);
|
2013-09-24 23:57:58 +07:00
|
|
|
void i915_vma_move_to_active(struct i915_vma *vma,
|
|
|
|
struct intel_ring_buffer *ring);
|
2011-02-07 09:16:14 +07:00
|
|
|
int i915_gem_dumb_create(struct drm_file *file_priv,
|
|
|
|
struct drm_device *dev,
|
|
|
|
struct drm_mode_create_dumb *args);
|
|
|
|
int i915_gem_mmap_gtt(struct drm_file *file_priv, struct drm_device *dev,
|
|
|
|
uint32_t handle, uint64_t *offset);
|
2010-09-24 22:02:42 +07:00
|
|
|
/**
|
|
|
|
* Returns true if seq1 is later than seq2.
|
|
|
|
*/
|
|
|
|
static inline bool
|
|
|
|
i915_seqno_passed(uint32_t seq1, uint32_t seq2)
|
|
|
|
{
|
|
|
|
return (int32_t)(seq1 - seq2) >= 0;
|
|
|
|
}
|
|
|
|
|
2012-12-19 16:13:08 +07:00
|
|
|
int __must_check i915_gem_get_seqno(struct drm_device *dev, u32 *seqno);
|
|
|
|
int __must_check i915_gem_set_seqno(struct drm_device *dev, u32 seqno);
|
2012-04-17 21:31:24 +07:00
|
|
|
int __must_check i915_gem_object_get_fence(struct drm_i915_gem_object *obj);
|
2010-11-10 23:40:20 +07:00
|
|
|
int __must_check i915_gem_object_put_fence(struct drm_i915_gem_object *obj);
|
2010-11-23 22:26:33 +07:00
|
|
|
|
2012-03-22 22:10:00 +07:00
|
|
|
static inline bool
|
2011-12-14 19:57:08 +07:00
|
|
|
i915_gem_object_pin_fence(struct drm_i915_gem_object *obj)
|
|
|
|
{
|
|
|
|
if (obj->fence_reg != I915_FENCE_REG_NONE) {
|
|
|
|
struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
|
|
|
|
dev_priv->fence_regs[obj->fence_reg].pin_count++;
|
2012-03-22 22:10:00 +07:00
|
|
|
return true;
|
|
|
|
} else
|
|
|
|
return false;
|
2011-12-14 19:57:08 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
i915_gem_object_unpin_fence(struct drm_i915_gem_object *obj)
|
|
|
|
{
|
|
|
|
if (obj->fence_reg != I915_FENCE_REG_NONE) {
|
|
|
|
struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
|
2013-06-12 17:29:47 +07:00
|
|
|
WARN_ON(dev_priv->fence_regs[obj->fence_reg].pin_count <= 0);
|
2011-12-14 19:57:08 +07:00
|
|
|
dev_priv->fence_regs[obj->fence_reg].pin_count--;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
drm/i915: Boost RPS frequency for CPU stalls
If we encounter a situation where the CPU blocks waiting for results
from the GPU, give the GPU a kick to boost its the frequency.
This should work to reduce user interface stalls and to quickly promote
mesa to high frequencies - but the cost is that our requested frequency
stalls high (as we do not idle for long enough before rc6 to start
reducing frequencies, nor are we aggressive at down clocking an
underused GPU). However, this should be mitigated by rc6 itself powering
off the GPU when idle, and that energy use is dependent upon the workload
of the GPU in addition to its frequency (e.g. the math or sampler
functions only consume power when used). Still, this is likely to
adversely affect light workloads.
In particular, this nearly eliminates the highly noticeable wake-up lag
in animations from idle. For example, expose or workspace transitions.
(However, given the situation where we fail to downclock, our requested
frequency is almost always the maximum, except for Baytrail where we
manually downclock upon idling. This often masks the latency of
upclocking after being idle, so animations are typically smooth - at the
cost of increased power consumption.)
Stéphane raised the concern that this will punish good applications and
reward bad applications - but due to the nature of how mesa performs its
client throttling, I believe all mesa applications will be roughly
equally affected. To address this concern, and to prevent applications
like compositors from permanently boosting the RPS state, we ratelimit the
frequency of the wait-boosts each client recieves.
Unfortunately, this techinique is ineffective with Ironlake - which also
has dynamic render power states and suffers just as dramatically. For
Ironlake, the thermal/power headroom is shared with the CPU through
Intelligent Power Sharing and the intel-ips module. This leaves us with
no GPU boost frequencies available when coming out of idle, and due to
hardware limitations we cannot change the arbitration between the CPU and
GPU quickly enough to be effective.
v2: Limit each client to receiving a single boost for each active period.
Tested by QA to only marginally increase power, and to demonstrably
increase throughput in games. No latency measurements yet.
v3: Cater for front-buffer rendering with manual throttling.
v4: Tidy up.
v5: Sadly the compositor needs frequent boosts as it may never idle, but
due to its picking mechanism (using ReadPixels) may require frequent
waits. Those waits, along with the waits for the vrefresh swap, conspire
to keep the GPU at low frequencies despite the interactive latency. To
overcome this we ditch the one-boost-per-active-period and just ratelimit
the number of wait-boosts each client can receive.
Reported-and-tested-by: Paul Neumann <paul104x@yahoo.de>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=68716
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Stéphane Marchesin <stephane.marchesin@gmail.com>
Cc: Owen Taylor <otaylor@redhat.com>
Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com>
Cc: "Zhuang, Lena" <lena.zhuang@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[danvet: No extern for function prototypes in headers.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 23:34:56 +07:00
|
|
|
bool i915_gem_retire_requests(struct drm_device *dev);
|
2012-02-15 18:25:36 +07:00
|
|
|
void i915_gem_retire_requests_ring(struct intel_ring_buffer *ring);
|
2012-11-14 23:14:05 +07:00
|
|
|
int __must_check i915_gem_check_wedge(struct i915_gpu_error *error,
|
drm/i915: non-interruptible sleeps can't handle -EAGAIN
So don't return -EAGAIN, even in the case of a gpu hang. Remap it to
-EIO instead. Note that this isn't really an issue with
interruptability, but more that we have quite a few codepaths (mostly
around kms stuff) that simply can't handle any errors and hence not
even -EAGAIN. Instead of adding proper failure paths so that we could
restart these ioctls we've opted for the cheap way out of sleeping
non-interruptibly. Which works everywhere but when the gpu dies,
which this patch fixes.
So essentially interruptible == false means 'wait for the gpu or die
trying'.'
This patch is a bit ugly because intel_ring_begin is all non-interruptible
and hence only returns -EIO. But as the comment in there says,
auditing all the callsites would be a pain.
To avoid duplicating code, reuse i915_gem_check_wedge in __wait_seqno
and intel_wait_ring_buffer. Also use the opportunity to clarify the
different cases in i915_gem_check_wedge a bit with comments.
v2: Don't access dev_priv->mm.interruptible from check_wedge - we
might not hold dev->struct_mutex, making this racy. Instead pass
interruptible in as a parameter. I've noticed this because I've hit a
BUG_ON(!mutex_is_locked) at the top of check_wedge. This has been
added in
commit b4aca0106c466b5a0329318203f65bac2d91b682
Author: Ben Widawsky <ben@bwidawsk.net>
Date: Wed Apr 25 20:50:12 2012 -0700
drm/i915: extract some common olr+wedge code
although that commit is missing any justification for this. I guess
it's just copy&paste, because the same commit add the same BUG_ON
check to check_olr, where it indeed makes sense.
But in check_wedge everything we access is protected by other means,
so this is superflous. And because it now gets in the way (we add a
new caller in __wait_seqno, which can be called without
dev->struct_mutext) let's just remove it.
v3: Group all the i915_gem_check_wedge refactoring into this patch, so
that this patch here is all about not returning -EAGAIN to callsites
that can't handle syscall restarting.
v4: Add clarification what interuptible == fales means in our code,
requested by Ben Widawsky.
v5: Fix EAGAIN mispell noticed by Chris Wilson.
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-07-05 03:54:13 +07:00
|
|
|
bool interruptible);
|
2012-11-15 23:17:22 +07:00
|
|
|
static inline bool i915_reset_in_progress(struct i915_gpu_error *error)
|
|
|
|
{
|
|
|
|
return unlikely(atomic_read(&error->reset_counter)
|
2013-11-12 19:44:19 +07:00
|
|
|
& (I915_RESET_IN_PROGRESS_FLAG | I915_WEDGED));
|
2012-11-15 23:17:22 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool i915_terminally_wedged(struct i915_gpu_error *error)
|
|
|
|
{
|
2013-11-12 19:44:19 +07:00
|
|
|
return atomic_read(&error->reset_counter) & I915_WEDGED;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline u32 i915_reset_count(struct i915_gpu_error *error)
|
|
|
|
{
|
|
|
|
return ((atomic_read(&error->reset_counter) & ~I915_WEDGED) + 1) / 2;
|
2012-11-15 23:17:22 +07:00
|
|
|
}
|
2012-02-15 18:25:36 +07:00
|
|
|
|
2010-09-30 22:53:18 +07:00
|
|
|
void i915_gem_reset(struct drm_device *dev);
|
2013-08-08 20:41:09 +07:00
|
|
|
bool i915_gem_clflush_object(struct drm_i915_gem_object *obj, bool force);
|
2011-04-14 04:04:09 +07:00
|
|
|
int __must_check i915_gem_object_finish_gpu(struct drm_i915_gem_object *obj);
|
2012-04-24 21:47:41 +07:00
|
|
|
int __must_check i915_gem_init(struct drm_device *dev);
|
2012-02-02 15:58:12 +07:00
|
|
|
int __must_check i915_gem_init_hw(struct drm_device *dev);
|
2013-09-18 11:12:44 +07:00
|
|
|
int i915_gem_l3_remap(struct intel_ring_buffer *ring, int slice);
|
2012-02-02 15:58:12 +07:00
|
|
|
void i915_gem_init_swizzling(struct drm_device *dev);
|
DRM: i915: add mode setting support
This commit adds i915 driver support for the DRM mode setting APIs.
Currently, VGA, LVDS, SDVO DVI & VGA, TV and DVO LVDS outputs are
supported. HDMI, DisplayPort and additional SDVO output support will
follow.
Support for the mode setting code is controlled by the new 'modeset'
module option. A new config option, CONFIG_DRM_I915_KMS controls the
default behavior, and whether a PCI ID list is built into the module for
use by user level module utilities.
Note that if mode setting is enabled, user level drivers that access
display registers directly or that don't use the kernel graphics memory
manager will likely corrupt kernel graphics memory, disrupt output
configuration (possibly leading to hangs and/or blank displays), and
prevent panic/oops messages from appearing. So use caution when
enabling this code; be sure your user level code supports the new
interfaces.
A new SysRq key, 'g', provides emergency support for switching back to
the kernel's framebuffer console; which is useful for testing.
Co-authors: Dave Airlie <airlied@linux.ie>, Hong Liu <hong.liu@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2008-11-08 05:24:08 +07:00
|
|
|
void i915_gem_cleanup_ringbuffer(struct drm_device *dev);
|
2012-04-27 06:02:58 +07:00
|
|
|
int __must_check i915_gpu_idle(struct drm_device *dev);
|
2013-10-16 17:50:01 +07:00
|
|
|
int __must_check i915_gem_suspend(struct drm_device *dev);
|
2013-06-12 16:35:30 +07:00
|
|
|
int __i915_add_request(struct intel_ring_buffer *ring,
|
|
|
|
struct drm_file *file,
|
2013-06-12 19:01:39 +07:00
|
|
|
struct drm_i915_gem_object *batch_obj,
|
2013-06-12 16:35:30 +07:00
|
|
|
u32 *seqno);
|
|
|
|
#define i915_add_request(ring, seqno) \
|
2013-06-18 14:29:58 +07:00
|
|
|
__i915_add_request(ring, NULL, NULL, seqno)
|
2012-05-25 05:03:11 +07:00
|
|
|
int __must_check i915_wait_seqno(struct intel_ring_buffer *ring,
|
|
|
|
uint32_t seqno);
|
2008-11-13 01:03:55 +07:00
|
|
|
int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
|
2010-11-23 22:26:33 +07:00
|
|
|
int __must_check
|
|
|
|
i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj,
|
|
|
|
bool write);
|
|
|
|
int __must_check
|
2012-03-26 15:10:27 +07:00
|
|
|
i915_gem_object_set_to_cpu_domain(struct drm_i915_gem_object *obj, bool write);
|
|
|
|
int __must_check
|
2011-04-14 15:41:17 +07:00
|
|
|
i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj,
|
|
|
|
u32 alignment,
|
2010-11-23 22:26:33 +07:00
|
|
|
struct intel_ring_buffer *pipelined);
|
2013-08-09 18:25:09 +07:00
|
|
|
void i915_gem_object_unpin_from_display_plane(struct drm_i915_gem_object *obj);
|
2008-12-30 17:31:46 +07:00
|
|
|
int i915_gem_attach_phys_object(struct drm_device *dev,
|
2010-11-09 02:18:58 +07:00
|
|
|
struct drm_i915_gem_object *obj,
|
2010-08-07 17:01:39 +07:00
|
|
|
int id,
|
|
|
|
int align);
|
2008-12-30 17:31:46 +07:00
|
|
|
void i915_gem_detach_phys_object(struct drm_device *dev,
|
2010-11-09 02:18:58 +07:00
|
|
|
struct drm_i915_gem_object *obj);
|
2008-12-30 17:31:46 +07:00
|
|
|
void i915_gem_free_all_phys_object(struct drm_device *dev);
|
drm/i915: Boost RPS frequency for CPU stalls
If we encounter a situation where the CPU blocks waiting for results
from the GPU, give the GPU a kick to boost its the frequency.
This should work to reduce user interface stalls and to quickly promote
mesa to high frequencies - but the cost is that our requested frequency
stalls high (as we do not idle for long enough before rc6 to start
reducing frequencies, nor are we aggressive at down clocking an
underused GPU). However, this should be mitigated by rc6 itself powering
off the GPU when idle, and that energy use is dependent upon the workload
of the GPU in addition to its frequency (e.g. the math or sampler
functions only consume power when used). Still, this is likely to
adversely affect light workloads.
In particular, this nearly eliminates the highly noticeable wake-up lag
in animations from idle. For example, expose or workspace transitions.
(However, given the situation where we fail to downclock, our requested
frequency is almost always the maximum, except for Baytrail where we
manually downclock upon idling. This often masks the latency of
upclocking after being idle, so animations are typically smooth - at the
cost of increased power consumption.)
Stéphane raised the concern that this will punish good applications and
reward bad applications - but due to the nature of how mesa performs its
client throttling, I believe all mesa applications will be roughly
equally affected. To address this concern, and to prevent applications
like compositors from permanently boosting the RPS state, we ratelimit the
frequency of the wait-boosts each client recieves.
Unfortunately, this techinique is ineffective with Ironlake - which also
has dynamic render power states and suffers just as dramatically. For
Ironlake, the thermal/power headroom is shared with the CPU through
Intelligent Power Sharing and the intel-ips module. This leaves us with
no GPU boost frequencies available when coming out of idle, and due to
hardware limitations we cannot change the arbitration between the CPU and
GPU quickly enough to be effective.
v2: Limit each client to receiving a single boost for each active period.
Tested by QA to only marginally increase power, and to demonstrably
increase throughput in games. No latency measurements yet.
v3: Cater for front-buffer rendering with manual throttling.
v4: Tidy up.
v5: Sadly the compositor needs frequent boosts as it may never idle, but
due to its picking mechanism (using ReadPixels) may require frequent
waits. Those waits, along with the waits for the vrefresh swap, conspire
to keep the GPU at low frequencies despite the interactive latency. To
overcome this we ditch the one-boost-per-active-period and just ratelimit
the number of wait-boosts each client can receive.
Reported-and-tested-by: Paul Neumann <paul104x@yahoo.de>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=68716
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Stéphane Marchesin <stephane.marchesin@gmail.com>
Cc: Owen Taylor <otaylor@redhat.com>
Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com>
Cc: "Zhuang, Lena" <lena.zhuang@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[danvet: No extern for function prototypes in headers.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 23:34:56 +07:00
|
|
|
int i915_gem_open(struct drm_device *dev, struct drm_file *file);
|
2010-11-09 02:18:58 +07:00
|
|
|
void i915_gem_release(struct drm_device *dev, struct drm_file *file);
|
2008-07-31 02:06:12 +07:00
|
|
|
|
2013-01-08 02:47:35 +07:00
|
|
|
uint32_t
|
|
|
|
i915_gem_get_gtt_size(struct drm_device *dev, uint32_t size, int tiling_mode);
|
2011-03-07 17:42:03 +07:00
|
|
|
uint32_t
|
2013-01-08 02:47:33 +07:00
|
|
|
i915_gem_get_gtt_alignment(struct drm_device *dev, uint32_t size,
|
|
|
|
int tiling_mode, bool fenced);
|
2011-03-07 17:42:03 +07:00
|
|
|
|
2011-04-04 15:44:39 +07:00
|
|
|
int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj,
|
|
|
|
enum i915_cache_level cache_level);
|
|
|
|
|
2012-05-10 20:25:09 +07:00
|
|
|
struct drm_gem_object *i915_gem_prime_import(struct drm_device *dev,
|
|
|
|
struct dma_buf *dma_buf);
|
|
|
|
|
|
|
|
struct dma_buf *i915_gem_prime_export(struct drm_device *dev,
|
|
|
|
struct drm_gem_object *gem_obj, int flags);
|
|
|
|
|
2013-06-12 16:15:12 +07:00
|
|
|
void i915_gem_restore_fences(struct drm_device *dev);
|
|
|
|
|
2013-08-01 06:59:56 +07:00
|
|
|
unsigned long i915_gem_obj_offset(struct drm_i915_gem_object *o,
|
|
|
|
struct i915_address_space *vm);
|
|
|
|
bool i915_gem_obj_bound_any(struct drm_i915_gem_object *o);
|
|
|
|
bool i915_gem_obj_bound(struct drm_i915_gem_object *o,
|
|
|
|
struct i915_address_space *vm);
|
|
|
|
unsigned long i915_gem_obj_size(struct drm_i915_gem_object *o,
|
|
|
|
struct i915_address_space *vm);
|
|
|
|
struct i915_vma *i915_gem_obj_to_vma(struct drm_i915_gem_object *obj,
|
|
|
|
struct i915_address_space *vm);
|
2013-08-14 16:38:35 +07:00
|
|
|
struct i915_vma *
|
|
|
|
i915_gem_obj_lookup_or_create_vma(struct drm_i915_gem_object *obj,
|
|
|
|
struct i915_address_space *vm);
|
2013-09-24 23:57:57 +07:00
|
|
|
|
|
|
|
struct i915_vma *i915_gem_obj_to_ggtt(struct drm_i915_gem_object *obj);
|
2013-12-07 05:10:55 +07:00
|
|
|
static inline bool i915_gem_obj_is_pinned(struct drm_i915_gem_object *obj) {
|
|
|
|
struct i915_vma *vma;
|
|
|
|
list_for_each_entry(vma, &obj->vma_list, vma_link)
|
|
|
|
if (vma->pin_count > 0)
|
|
|
|
return true;
|
|
|
|
return false;
|
|
|
|
}
|
2013-09-24 23:57:57 +07:00
|
|
|
|
2013-08-01 06:59:56 +07:00
|
|
|
/* Some GGTT VM helpers */
|
|
|
|
#define obj_to_ggtt(obj) \
|
|
|
|
(&((struct drm_i915_private *)(obj)->base.dev->dev_private)->gtt.base)
|
|
|
|
static inline bool i915_is_ggtt(struct i915_address_space *vm)
|
|
|
|
{
|
|
|
|
struct i915_address_space *ggtt =
|
|
|
|
&((struct drm_i915_private *)(vm)->dev->dev_private)->gtt.base;
|
|
|
|
return vm == ggtt;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool i915_gem_obj_ggtt_bound(struct drm_i915_gem_object *obj)
|
|
|
|
{
|
|
|
|
return i915_gem_obj_bound(obj, obj_to_ggtt(obj));
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline unsigned long
|
|
|
|
i915_gem_obj_ggtt_offset(struct drm_i915_gem_object *obj)
|
|
|
|
{
|
|
|
|
return i915_gem_obj_offset(obj, obj_to_ggtt(obj));
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline unsigned long
|
|
|
|
i915_gem_obj_ggtt_size(struct drm_i915_gem_object *obj)
|
|
|
|
{
|
|
|
|
return i915_gem_obj_size(obj, obj_to_ggtt(obj));
|
|
|
|
}
|
2013-08-01 06:59:58 +07:00
|
|
|
|
|
|
|
static inline int __must_check
|
|
|
|
i915_gem_obj_ggtt_pin(struct drm_i915_gem_object *obj,
|
|
|
|
uint32_t alignment,
|
2014-02-14 20:01:11 +07:00
|
|
|
unsigned flags)
|
2013-08-01 06:59:58 +07:00
|
|
|
{
|
2014-02-14 20:01:12 +07:00
|
|
|
return i915_gem_object_pin(obj, obj_to_ggtt(obj), alignment, flags | PIN_GLOBAL);
|
2013-08-01 06:59:58 +07:00
|
|
|
}
|
2013-08-01 06:59:56 +07:00
|
|
|
|
2014-02-14 20:01:19 +07:00
|
|
|
static inline int
|
|
|
|
i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj)
|
|
|
|
{
|
|
|
|
return i915_vma_unbind(i915_gem_obj_to_ggtt(obj));
|
|
|
|
}
|
|
|
|
|
|
|
|
void i915_gem_object_ggtt_unpin(struct drm_i915_gem_object *obj);
|
|
|
|
|
drm/i915: preliminary context support
Very basic code for context setup/destruction in the driver.
Adds the file i915_gem_context.c This file implements HW context
support. On gen5+ a HW context consists of an opaque GPU object which is
referenced at times of context saves and restores. With RC6 enabled,
the context is also referenced as the GPU enters and exists from RC6
(GPU has it's own internal power context, except on gen5). Though
something like a context does exist for the media ring, the code only
supports contexts for the render ring.
In software, there is a distinction between contexts created by the
user, and the default HW context. The default HW context is used by GPU
clients that do not request setup of their own hardware context. The
default context's state is never restored to help prevent programming
errors. This would happen if a client ran and piggy-backed off another
clients GPU state. The default context only exists to give the GPU some
offset to load as the current to invoke a save of the context we
actually care about. In fact, the code could likely be constructed,
albeit in a more complicated fashion, to never use the default context,
though that limits the driver's ability to swap out, and/or destroy
other contexts.
All other contexts are created as a request by the GPU client. These
contexts store GPU state, and thus allow GPU clients to not re-emit
state (and potentially query certain state) at any time. The kernel
driver makes certain that the appropriate commands are inserted.
There are 4 entry points into the contexts, init, fini, open, close.
The names are self-explanatory except that init can be called during
reset, and also during pm thaw/resume. As we expect our context to be
preserved across these events, we do not reinitialize in this case.
As Adam Jackson pointed out, The cutoff of 1MB where a HW context is
considered too big is arbitrary. The reason for this is even though
context sizes are increasing with every generation, they have yet to
eclipse even 32k. If we somehow read back way more than that, it
probably means BIOS has done something strange, or we're running on a
platform that wasn't designed for this.
v2: rename load/unload to init/fini (daniel)
remove ILK support for get_size() (indirectly daniel)
add HAS_HW_CONTEXTS macro to clarify supported platforms (daniel)
added comments (Ben)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2012-06-05 04:42:42 +07:00
|
|
|
/* i915_gem_context.c */
|
2013-12-07 05:11:19 +07:00
|
|
|
#define ctx_to_ppgtt(ctx) container_of((ctx)->vm, struct i915_hw_ppgtt, base)
|
2013-11-06 22:56:29 +07:00
|
|
|
int __must_check i915_gem_context_init(struct drm_device *dev);
|
drm/i915: preliminary context support
Very basic code for context setup/destruction in the driver.
Adds the file i915_gem_context.c This file implements HW context
support. On gen5+ a HW context consists of an opaque GPU object which is
referenced at times of context saves and restores. With RC6 enabled,
the context is also referenced as the GPU enters and exists from RC6
(GPU has it's own internal power context, except on gen5). Though
something like a context does exist for the media ring, the code only
supports contexts for the render ring.
In software, there is a distinction between contexts created by the
user, and the default HW context. The default HW context is used by GPU
clients that do not request setup of their own hardware context. The
default context's state is never restored to help prevent programming
errors. This would happen if a client ran and piggy-backed off another
clients GPU state. The default context only exists to give the GPU some
offset to load as the current to invoke a save of the context we
actually care about. In fact, the code could likely be constructed,
albeit in a more complicated fashion, to never use the default context,
though that limits the driver's ability to swap out, and/or destroy
other contexts.
All other contexts are created as a request by the GPU client. These
contexts store GPU state, and thus allow GPU clients to not re-emit
state (and potentially query certain state) at any time. The kernel
driver makes certain that the appropriate commands are inserted.
There are 4 entry points into the contexts, init, fini, open, close.
The names are self-explanatory except that init can be called during
reset, and also during pm thaw/resume. As we expect our context to be
preserved across these events, we do not reinitialize in this case.
As Adam Jackson pointed out, The cutoff of 1MB where a HW context is
considered too big is arbitrary. The reason for this is even though
context sizes are increasing with every generation, they have yet to
eclipse even 32k. If we somehow read back way more than that, it
probably means BIOS has done something strange, or we're running on a
platform that wasn't designed for this.
v2: rename load/unload to init/fini (daniel)
remove ILK support for get_size() (indirectly daniel)
add HAS_HW_CONTEXTS macro to clarify supported platforms (daniel)
added comments (Ben)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2012-06-05 04:42:42 +07:00
|
|
|
void i915_gem_context_fini(struct drm_device *dev);
|
2013-12-07 05:11:03 +07:00
|
|
|
void i915_gem_context_reset(struct drm_device *dev);
|
2013-12-07 05:10:58 +07:00
|
|
|
int i915_gem_context_open(struct drm_device *dev, struct drm_file *file);
|
drm/i915: Split context enabling from init
We **need** to do this for exactly 1 reason, because we want to embed a
PPGTT into the context, but we don't want to special case the default
context.
To achieve that, we must be able to initialize contexts after the GTT is
setup (so we can allocate and pin the default context's BO), but before
the PPGTT and rings are initialized. This is because, currently, context
initialization requires ring usage. We don't have rings until after the
GTT is setup. If we split the enabling part of context initialization,
the part requiring the ringbuffer, we can untangle this, and then later
embed the PPGTT
Incidentally this allows us to also adhere to the original design of
context init/fini in future patches: they were only ever meant to be
called at driver load and unload.
v2: Move hw_contexts_disabled test in i915_gem_context_enable() (Chris)
v3: BUG_ON after checking for disabled contexts. Or else it blows up pre
gen6 (Ben)
v4: Forward port
Modified enable for each ring, since that patch is earlier in the series
Dropped ring arg from create_default_context so it can be used by others
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-12-07 05:11:04 +07:00
|
|
|
int i915_gem_context_enable(struct drm_i915_private *dev_priv);
|
drm/i915: preliminary context support
Very basic code for context setup/destruction in the driver.
Adds the file i915_gem_context.c This file implements HW context
support. On gen5+ a HW context consists of an opaque GPU object which is
referenced at times of context saves and restores. With RC6 enabled,
the context is also referenced as the GPU enters and exists from RC6
(GPU has it's own internal power context, except on gen5). Though
something like a context does exist for the media ring, the code only
supports contexts for the render ring.
In software, there is a distinction between contexts created by the
user, and the default HW context. The default HW context is used by GPU
clients that do not request setup of their own hardware context. The
default context's state is never restored to help prevent programming
errors. This would happen if a client ran and piggy-backed off another
clients GPU state. The default context only exists to give the GPU some
offset to load as the current to invoke a save of the context we
actually care about. In fact, the code could likely be constructed,
albeit in a more complicated fashion, to never use the default context,
though that limits the driver's ability to swap out, and/or destroy
other contexts.
All other contexts are created as a request by the GPU client. These
contexts store GPU state, and thus allow GPU clients to not re-emit
state (and potentially query certain state) at any time. The kernel
driver makes certain that the appropriate commands are inserted.
There are 4 entry points into the contexts, init, fini, open, close.
The names are self-explanatory except that init can be called during
reset, and also during pm thaw/resume. As we expect our context to be
preserved across these events, we do not reinitialize in this case.
As Adam Jackson pointed out, The cutoff of 1MB where a HW context is
considered too big is arbitrary. The reason for this is even though
context sizes are increasing with every generation, they have yet to
eclipse even 32k. If we somehow read back way more than that, it
probably means BIOS has done something strange, or we're running on a
platform that wasn't designed for this.
v2: rename load/unload to init/fini (daniel)
remove ILK support for get_size() (indirectly daniel)
add HAS_HW_CONTEXTS macro to clarify supported platforms (daniel)
added comments (Ben)
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2012-06-05 04:42:42 +07:00
|
|
|
void i915_gem_context_close(struct drm_device *dev, struct drm_file *file);
|
2012-06-05 04:42:46 +07:00
|
|
|
int i915_switch_context(struct intel_ring_buffer *ring,
|
drm/i915: Get context early in execbuf
We need to have the address space when reserving space for the objects.
Since the address space and context are tied together, and reserve
occurs before context switch (for good reason), we must lookup our
context earlier in the process.
This leaves some room for optimizations where we no longer need to use
ctx_id in certain places. This will be addressed in a subsequent patch.
Important tricky bit:
Because slow relocations during execbuffer drop struct_mutex
Perhaps it would be best to acquire the reference when we get the
context, but I'll save that for another day (note I have written the
patch before, and I found the changes required to be uglier than this).
Note that since we currently access everything via context id, and not
the data structure this is fine, though not desirable. The next change
attempts to get the context only once via the context ID idr lookup, and
as such, the following can happen:
CTX-A is created, refcount = 1
CTX-A execbuf, mutex dropped
close IOCTL called on CTX-A, refcount = 0
CTX-A resumes in execbuf.
v2: Rebased on top of
commit b6359918b885da7c7b58c050674278dbd06020ab
Author: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Date: Wed Oct 30 15:44:16 2013 +0200
drm/i915: add i915_get_reset_stats_ioctl
v3: Rebased on top of
commit 25b3dfc87bff80317d67ddd2cd4cfb91e6fe7d79
Author: Mika Westerberg <mika.westerberg@linux.intel.com>
Date: Tue Nov 12 11:57:30 2013 +0200
Author: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Date: Tue Nov 26 16:14:33 2013 +0200
drm/i915: check context reset stats before relocations
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-12-07 05:11:21 +07:00
|
|
|
struct drm_file *file, struct i915_hw_context *to);
|
|
|
|
struct i915_hw_context *
|
|
|
|
i915_gem_context_get(struct drm_i915_file_private *file_priv, u32 id);
|
2013-04-30 17:30:33 +07:00
|
|
|
void i915_gem_context_free(struct kref *ctx_ref);
|
|
|
|
static inline void i915_gem_context_reference(struct i915_hw_context *ctx)
|
|
|
|
{
|
2013-12-07 05:11:20 +07:00
|
|
|
if (ctx->obj && HAS_HW_CONTEXTS(ctx->obj->base.dev))
|
|
|
|
kref_get(&ctx->ref);
|
2013-04-30 17:30:33 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void i915_gem_context_unreference(struct i915_hw_context *ctx)
|
|
|
|
{
|
2013-12-07 05:11:20 +07:00
|
|
|
if (ctx->obj && HAS_HW_CONTEXTS(ctx->obj->base.dev))
|
|
|
|
kref_put(&ctx->ref, i915_gem_context_free);
|
2013-04-30 17:30:33 +07:00
|
|
|
}
|
|
|
|
|
2014-01-30 21:05:48 +07:00
|
|
|
static inline bool i915_gem_context_is_default(const struct i915_hw_context *c)
|
|
|
|
{
|
|
|
|
return c->id == DEFAULT_CONTEXT_ID;
|
|
|
|
}
|
|
|
|
|
2012-06-05 04:42:54 +07:00
|
|
|
int i915_gem_context_create_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file);
|
|
|
|
int i915_gem_context_destroy_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file);
|
2012-05-10 20:25:09 +07:00
|
|
|
|
2013-12-07 05:11:23 +07:00
|
|
|
/* i915_gem_evict.c */
|
|
|
|
int __must_check i915_gem_evict_something(struct drm_device *dev,
|
|
|
|
struct i915_address_space *vm,
|
|
|
|
int min_size,
|
|
|
|
unsigned alignment,
|
|
|
|
unsigned cache_level,
|
2014-02-14 20:01:11 +07:00
|
|
|
unsigned flags);
|
2013-12-07 05:11:23 +07:00
|
|
|
int i915_gem_evict_vm(struct i915_address_space *vm, bool do_idle);
|
|
|
|
int i915_gem_evict_everything(struct drm_device *dev);
|
2012-02-09 23:15:46 +07:00
|
|
|
|
2010-11-06 04:23:30 +07:00
|
|
|
/* i915_gem_gtt.c */
|
2013-10-16 23:21:30 +07:00
|
|
|
void i915_check_and_clear_faults(struct drm_device *dev);
|
|
|
|
void i915_gem_suspend_gtt_mappings(struct drm_device *dev);
|
2010-11-06 04:23:30 +07:00
|
|
|
void i915_gem_restore_gtt_mappings(struct drm_device *dev);
|
2012-02-16 05:50:21 +07:00
|
|
|
int __must_check i915_gem_gtt_prepare_object(struct drm_i915_gem_object *obj);
|
|
|
|
void i915_gem_gtt_finish_object(struct drm_i915_gem_object *obj);
|
2012-12-19 01:31:25 +07:00
|
|
|
void i915_gem_init_global_gtt(struct drm_device *dev);
|
|
|
|
void i915_gem_setup_global_gtt(struct drm_device *dev, unsigned long start,
|
|
|
|
unsigned long mappable_end, unsigned long end);
|
2012-11-05 00:21:27 +07:00
|
|
|
int i915_gem_gtt_init(struct drm_device *dev);
|
2012-11-16 03:06:09 +07:00
|
|
|
static inline void i915_gem_chipset_flush(struct drm_device *dev)
|
2012-11-05 00:21:27 +07:00
|
|
|
{
|
|
|
|
if (INTEL_INFO(dev)->gen < 6)
|
|
|
|
intel_gtt_chipset_flush();
|
|
|
|
}
|
2013-12-07 05:11:14 +07:00
|
|
|
int i915_gem_init_ppgtt(struct drm_device *dev, struct i915_hw_ppgtt *ppgtt);
|
|
|
|
static inline bool intel_enable_ppgtt(struct drm_device *dev, bool full)
|
|
|
|
{
|
2014-01-21 16:24:25 +07:00
|
|
|
if (i915.enable_ppgtt == 0 || !HAS_ALIASING_PPGTT(dev))
|
2013-12-07 05:11:14 +07:00
|
|
|
return false;
|
2012-11-05 00:21:27 +07:00
|
|
|
|
2014-01-21 16:24:25 +07:00
|
|
|
if (i915.enable_ppgtt == 1 && full)
|
2013-12-07 05:11:26 +07:00
|
|
|
return false;
|
2010-11-06 04:23:30 +07:00
|
|
|
|
2013-12-07 05:11:14 +07:00
|
|
|
#ifdef CONFIG_INTEL_IOMMU
|
|
|
|
/* Disable ppgtt on SNB if VT-d is on. */
|
|
|
|
if (INTEL_INFO(dev)->gen == 6 && intel_iommu_gfx_mapped) {
|
|
|
|
DRM_INFO("Disabling PPGTT because VT-d is on\n");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2013-12-07 05:11:26 +07:00
|
|
|
if (full)
|
|
|
|
return HAS_PPGTT(dev);
|
|
|
|
else
|
|
|
|
return HAS_ALIASING_PPGTT(dev);
|
2013-12-07 05:11:14 +07:00
|
|
|
}
|
|
|
|
|
drm/i915: Add VM to context
Pretty straightforward so far except for the bit about the refcounting.
The PPGTT will potentially be shared amongst multiple contexts. Because
contexts themselves have a refcounted lifecycle, the easiest way to
manage this will be to refcount the PPGTT. To acheive this, we piggy
back off of the existing context refcount, and will increment and
decrement the PPGTT refcount with context creation, and destruction.
To put it more clearly, if context A, and context B both use PPGTT 0, we
can't free the PPGTT until both A, and B are destroyed.
Note that because the PPGTT is permanently pinned (for now), it really
just matters for the PPGTT destruction, as opposed to making space under
memory pressure.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-12-07 05:11:15 +07:00
|
|
|
static inline void ppgtt_release(struct kref *kref)
|
|
|
|
{
|
|
|
|
struct i915_hw_ppgtt *ppgtt = container_of(kref, struct i915_hw_ppgtt, ref);
|
2013-12-07 05:11:23 +07:00
|
|
|
struct drm_device *dev = ppgtt->base.dev;
|
|
|
|
struct drm_i915_private *dev_priv = dev->dev_private;
|
|
|
|
struct i915_address_space *vm = &ppgtt->base;
|
|
|
|
|
|
|
|
if (ppgtt == dev_priv->mm.aliasing_ppgtt ||
|
|
|
|
(list_empty(&vm->active_list) && list_empty(&vm->inactive_list))) {
|
|
|
|
ppgtt->base.cleanup(&ppgtt->base);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Make sure vmas are unbound before we take down the drm_mm
|
|
|
|
*
|
|
|
|
* FIXME: Proper refcounting should take care of this, this shouldn't be
|
|
|
|
* needed at all.
|
|
|
|
*/
|
|
|
|
if (!list_empty(&vm->active_list)) {
|
|
|
|
struct i915_vma *vma;
|
|
|
|
|
|
|
|
list_for_each_entry(vma, &vm->active_list, mm_list)
|
|
|
|
if (WARN_ON(list_empty(&vma->vma_link) ||
|
|
|
|
list_is_singular(&vma->vma_link)))
|
|
|
|
break;
|
|
|
|
|
|
|
|
i915_gem_evict_vm(&ppgtt->base, true);
|
|
|
|
} else {
|
|
|
|
i915_gem_retire_requests(dev);
|
|
|
|
i915_gem_evict_vm(&ppgtt->base, false);
|
|
|
|
}
|
drm/i915: Add VM to context
Pretty straightforward so far except for the bit about the refcounting.
The PPGTT will potentially be shared amongst multiple contexts. Because
contexts themselves have a refcounted lifecycle, the easiest way to
manage this will be to refcount the PPGTT. To acheive this, we piggy
back off of the existing context refcount, and will increment and
decrement the PPGTT refcount with context creation, and destruction.
To put it more clearly, if context A, and context B both use PPGTT 0, we
can't free the PPGTT until both A, and B are destroyed.
Note that because the PPGTT is permanently pinned (for now), it really
just matters for the PPGTT destruction, as opposed to making space under
memory pressure.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-12-07 05:11:15 +07:00
|
|
|
|
|
|
|
ppgtt->base.cleanup(&ppgtt->base);
|
|
|
|
}
|
2010-08-07 17:01:23 +07:00
|
|
|
|
2012-04-24 21:47:39 +07:00
|
|
|
/* i915_gem_stolen.c */
|
|
|
|
int i915_gem_init_stolen(struct drm_device *dev);
|
2012-11-15 18:32:20 +07:00
|
|
|
int i915_gem_stolen_setup_compression(struct drm_device *dev, int size);
|
|
|
|
void i915_gem_stolen_cleanup_compression(struct drm_device *dev);
|
2012-04-24 21:47:39 +07:00
|
|
|
void i915_gem_cleanup_stolen(struct drm_device *dev);
|
2012-11-15 18:32:26 +07:00
|
|
|
struct drm_i915_gem_object *
|
|
|
|
i915_gem_object_create_stolen(struct drm_device *dev, u32 size);
|
2013-02-20 04:31:37 +07:00
|
|
|
struct drm_i915_gem_object *
|
|
|
|
i915_gem_object_create_stolen_for_preallocated(struct drm_device *dev,
|
|
|
|
u32 stolen_offset,
|
|
|
|
u32 gtt_offset,
|
|
|
|
u32 size);
|
2012-11-15 18:32:26 +07:00
|
|
|
void i915_gem_object_release_stolen(struct drm_i915_gem_object *obj);
|
2012-04-24 21:47:39 +07:00
|
|
|
|
2008-07-31 02:06:12 +07:00
|
|
|
/* i915_gem_tiling.c */
|
2013-08-02 00:39:55 +07:00
|
|
|
static inline bool i915_gem_object_needs_bit17_swizzle(struct drm_i915_gem_object *obj)
|
2012-12-04 04:03:14 +07:00
|
|
|
{
|
|
|
|
drm_i915_private_t *dev_priv = obj->base.dev->dev_private;
|
|
|
|
|
|
|
|
return dev_priv->mm.bit_6_swizzle_x == I915_BIT_6_SWIZZLE_9_10_17 &&
|
|
|
|
obj->tiling_mode != I915_TILING_NONE;
|
|
|
|
}
|
|
|
|
|
2008-07-31 02:06:12 +07:00
|
|
|
void i915_gem_detect_bit_6_swizzle(struct drm_device *dev);
|
2010-11-09 02:18:58 +07:00
|
|
|
void i915_gem_object_do_bit_17_swizzle(struct drm_i915_gem_object *obj);
|
|
|
|
void i915_gem_object_save_bit_17_swizzle(struct drm_i915_gem_object *obj);
|
2008-07-31 02:06:12 +07:00
|
|
|
|
|
|
|
/* i915_gem_debug.c */
|
2010-09-29 22:10:57 +07:00
|
|
|
#if WATCH_LISTS
|
|
|
|
int i915_verify_lists(struct drm_device *dev);
|
2008-07-31 02:06:12 +07:00
|
|
|
#else
|
2010-09-29 22:10:57 +07:00
|
|
|
#define i915_verify_lists(dev) 0
|
2008-07-31 02:06:12 +07:00
|
|
|
#endif
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2009-02-18 08:08:50 +07:00
|
|
|
/* i915_debugfs.c */
|
2009-07-02 09:26:52 +07:00
|
|
|
int i915_debugfs_init(struct drm_minor *minor);
|
|
|
|
void i915_debugfs_cleanup(struct drm_minor *minor);
|
2013-10-16 16:49:58 +07:00
|
|
|
#ifdef CONFIG_DEBUG_FS
|
2013-10-16 00:55:40 +07:00
|
|
|
void intel_display_crc_init(struct drm_device *dev);
|
|
|
|
#else
|
2013-10-16 16:49:58 +07:00
|
|
|
static inline void intel_display_crc_init(struct drm_device *dev) {}
|
2013-10-16 00:55:40 +07:00
|
|
|
#endif
|
2013-07-12 20:50:57 +07:00
|
|
|
|
|
|
|
/* i915_gpu_error.c */
|
2013-05-23 17:55:35 +07:00
|
|
|
__printf(2, 3)
|
|
|
|
void i915_error_printf(struct drm_i915_error_state_buf *e, const char *f, ...);
|
2013-06-06 19:18:39 +07:00
|
|
|
int i915_error_state_to_str(struct drm_i915_error_state_buf *estr,
|
|
|
|
const struct i915_error_state_file_priv *error);
|
2013-06-06 19:18:41 +07:00
|
|
|
int i915_error_state_buf_init(struct drm_i915_error_state_buf *eb,
|
|
|
|
size_t count, loff_t pos);
|
|
|
|
static inline void i915_error_state_buf_release(
|
|
|
|
struct drm_i915_error_state_buf *eb)
|
|
|
|
{
|
|
|
|
kfree(eb->buf);
|
|
|
|
}
|
2013-07-12 20:50:57 +07:00
|
|
|
void i915_capture_error_state(struct drm_device *dev);
|
|
|
|
void i915_error_state_get(struct drm_device *dev,
|
|
|
|
struct i915_error_state_file_priv *error_priv);
|
|
|
|
void i915_error_state_put(struct i915_error_state_file_priv *error_priv);
|
|
|
|
void i915_destroy_error_state(struct drm_device *dev);
|
|
|
|
|
|
|
|
void i915_get_extra_instdone(struct drm_device *dev, uint32_t *instdone);
|
|
|
|
const char *i915_cache_level_str(int type);
|
2009-02-18 08:08:50 +07:00
|
|
|
|
2008-08-26 05:11:06 +07:00
|
|
|
/* i915_suspend.c */
|
|
|
|
extern int i915_save_state(struct drm_device *dev);
|
|
|
|
extern int i915_restore_state(struct drm_device *dev);
|
2008-10-01 02:14:26 +07:00
|
|
|
|
2013-01-25 23:53:20 +07:00
|
|
|
/* i915_ums.c */
|
|
|
|
void i915_save_display_reg(struct drm_device *dev);
|
|
|
|
void i915_restore_display_reg(struct drm_device *dev);
|
2008-08-26 05:11:06 +07:00
|
|
|
|
2012-04-11 11:17:01 +07:00
|
|
|
/* i915_sysfs.c */
|
|
|
|
void i915_setup_sysfs(struct drm_device *dev_priv);
|
|
|
|
void i915_teardown_sysfs(struct drm_device *dev_priv);
|
|
|
|
|
2010-07-21 05:44:45 +07:00
|
|
|
/* intel_i2c.c */
|
|
|
|
extern int intel_setup_gmbus(struct drm_device *dev);
|
|
|
|
extern void intel_teardown_gmbus(struct drm_device *dev);
|
2013-05-06 19:52:08 +07:00
|
|
|
static inline bool intel_gmbus_is_port_valid(unsigned port)
|
2012-03-28 01:36:14 +07:00
|
|
|
{
|
2012-03-28 01:36:15 +07:00
|
|
|
return (port >= GMBUS_PORT_SSC && port <= GMBUS_PORT_DPD);
|
2012-03-28 01:36:14 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
extern struct i2c_adapter *intel_gmbus_get_adapter(
|
|
|
|
struct drm_i915_private *dev_priv, unsigned port);
|
2010-09-24 18:52:03 +07:00
|
|
|
extern void intel_gmbus_set_speed(struct i2c_adapter *adapter, int speed);
|
|
|
|
extern void intel_gmbus_force_bit(struct i2c_adapter *adapter, bool force_bit);
|
2013-05-06 19:52:08 +07:00
|
|
|
static inline bool intel_gmbus_is_forced_bit(struct i2c_adapter *adapter)
|
2010-09-28 22:41:32 +07:00
|
|
|
{
|
|
|
|
return container_of(adapter, struct intel_gmbus, adapter)->force_bit;
|
|
|
|
}
|
2010-07-21 05:44:45 +07:00
|
|
|
extern void intel_i2c_reset(struct drm_device *dev);
|
|
|
|
|
2010-08-24 15:02:58 +07:00
|
|
|
/* intel_opregion.c */
|
2013-08-30 23:40:30 +07:00
|
|
|
struct intel_encoder;
|
2010-08-19 22:09:23 +07:00
|
|
|
extern int intel_opregion_setup(struct drm_device *dev);
|
|
|
|
#ifdef CONFIG_ACPI
|
|
|
|
extern void intel_opregion_init(struct drm_device *dev);
|
|
|
|
extern void intel_opregion_fini(struct drm_device *dev);
|
2010-08-24 15:02:58 +07:00
|
|
|
extern void intel_opregion_asle_intr(struct drm_device *dev);
|
2013-08-30 23:40:30 +07:00
|
|
|
extern int intel_opregion_notify_encoder(struct intel_encoder *intel_encoder,
|
|
|
|
bool enable);
|
2013-08-30 23:40:31 +07:00
|
|
|
extern int intel_opregion_notify_adapter(struct drm_device *dev,
|
|
|
|
pci_power_t state);
|
2008-10-25 04:18:10 +07:00
|
|
|
#else
|
2010-08-19 22:09:23 +07:00
|
|
|
static inline void intel_opregion_init(struct drm_device *dev) { return; }
|
|
|
|
static inline void intel_opregion_fini(struct drm_device *dev) { return; }
|
2010-08-24 15:02:58 +07:00
|
|
|
static inline void intel_opregion_asle_intr(struct drm_device *dev) { return; }
|
2013-08-30 23:40:30 +07:00
|
|
|
static inline int
|
|
|
|
intel_opregion_notify_encoder(struct intel_encoder *intel_encoder, bool enable)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2013-08-30 23:40:31 +07:00
|
|
|
static inline int
|
|
|
|
intel_opregion_notify_adapter(struct drm_device *dev, pci_power_t state)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2008-10-25 04:18:10 +07:00
|
|
|
#endif
|
2008-08-06 01:37:25 +07:00
|
|
|
|
2010-10-08 06:01:13 +07:00
|
|
|
/* intel_acpi.c */
|
|
|
|
#ifdef CONFIG_ACPI
|
|
|
|
extern void intel_register_dsm_handler(void);
|
|
|
|
extern void intel_unregister_dsm_handler(void);
|
|
|
|
#else
|
|
|
|
static inline void intel_register_dsm_handler(void) { return; }
|
|
|
|
static inline void intel_unregister_dsm_handler(void) { return; }
|
|
|
|
#endif /* CONFIG_ACPI */
|
|
|
|
|
DRM: i915: add mode setting support
This commit adds i915 driver support for the DRM mode setting APIs.
Currently, VGA, LVDS, SDVO DVI & VGA, TV and DVO LVDS outputs are
supported. HDMI, DisplayPort and additional SDVO output support will
follow.
Support for the mode setting code is controlled by the new 'modeset'
module option. A new config option, CONFIG_DRM_I915_KMS controls the
default behavior, and whether a PCI ID list is built into the module for
use by user level module utilities.
Note that if mode setting is enabled, user level drivers that access
display registers directly or that don't use the kernel graphics memory
manager will likely corrupt kernel graphics memory, disrupt output
configuration (possibly leading to hangs and/or blank displays), and
prevent panic/oops messages from appearing. So use caution when
enabling this code; be sure your user level code supports the new
interfaces.
A new SysRq key, 'g', provides emergency support for switching back to
the kernel's framebuffer console; which is useful for testing.
Co-authors: Dave Airlie <airlied@linux.ie>, Hong Liu <hong.liu@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2008-11-08 05:24:08 +07:00
|
|
|
/* modesetting */
|
2012-04-10 20:50:11 +07:00
|
|
|
extern void intel_modeset_init_hw(struct drm_device *dev);
|
2013-04-17 18:04:50 +07:00
|
|
|
extern void intel_modeset_suspend_hw(struct drm_device *dev);
|
DRM: i915: add mode setting support
This commit adds i915 driver support for the DRM mode setting APIs.
Currently, VGA, LVDS, SDVO DVI & VGA, TV and DVO LVDS outputs are
supported. HDMI, DisplayPort and additional SDVO output support will
follow.
Support for the mode setting code is controlled by the new 'modeset'
module option. A new config option, CONFIG_DRM_I915_KMS controls the
default behavior, and whether a PCI ID list is built into the module for
use by user level module utilities.
Note that if mode setting is enabled, user level drivers that access
display registers directly or that don't use the kernel graphics memory
manager will likely corrupt kernel graphics memory, disrupt output
configuration (possibly leading to hangs and/or blank displays), and
prevent panic/oops messages from appearing. So use caution when
enabling this code; be sure your user level code supports the new
interfaces.
A new SysRq key, 'g', provides emergency support for switching back to
the kernel's framebuffer console; which is useful for testing.
Co-authors: Dave Airlie <airlied@linux.ie>, Hong Liu <hong.liu@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2008-11-08 05:24:08 +07:00
|
|
|
extern void intel_modeset_init(struct drm_device *dev);
|
2011-03-29 16:40:27 +07:00
|
|
|
extern void intel_modeset_gem_init(struct drm_device *dev);
|
DRM: i915: add mode setting support
This commit adds i915 driver support for the DRM mode setting APIs.
Currently, VGA, LVDS, SDVO DVI & VGA, TV and DVO LVDS outputs are
supported. HDMI, DisplayPort and additional SDVO output support will
follow.
Support for the mode setting code is controlled by the new 'modeset'
module option. A new config option, CONFIG_DRM_I915_KMS controls the
default behavior, and whether a PCI ID list is built into the module for
use by user level module utilities.
Note that if mode setting is enabled, user level drivers that access
display registers directly or that don't use the kernel graphics memory
manager will likely corrupt kernel graphics memory, disrupt output
configuration (possibly leading to hangs and/or blank displays), and
prevent panic/oops messages from appearing. So use caution when
enabling this code; be sure your user level code supports the new
interfaces.
A new SysRq key, 'g', provides emergency support for switching back to
the kernel's framebuffer console; which is useful for testing.
Co-authors: Dave Airlie <airlied@linux.ie>, Hong Liu <hong.liu@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2008-11-08 05:24:08 +07:00
|
|
|
extern void intel_modeset_cleanup(struct drm_device *dev);
|
2014-02-11 22:12:48 +07:00
|
|
|
extern void intel_connector_unregister(struct intel_connector *);
|
2009-09-21 11:33:58 +07:00
|
|
|
extern int intel_modeset_vga_set_state(struct drm_device *dev, bool state);
|
2012-11-24 00:16:34 +07:00
|
|
|
extern void intel_modeset_setup_hw_state(struct drm_device *dev,
|
|
|
|
bool force_restore);
|
2013-01-25 23:53:21 +07:00
|
|
|
extern void i915_redisable_vga(struct drm_device *dev);
|
2010-04-23 22:17:39 +07:00
|
|
|
extern bool intel_fbc_enabled(struct drm_device *dev);
|
2011-07-08 18:22:36 +07:00
|
|
|
extern void intel_disable_fbc(struct drm_device *dev);
|
2010-05-21 04:28:11 +07:00
|
|
|
extern bool ironlake_set_drps(struct drm_device *dev, u8 val);
|
2012-12-01 21:04:25 +07:00
|
|
|
extern void intel_init_pch_refclk(struct drm_device *dev);
|
2010-12-18 05:19:02 +07:00
|
|
|
extern void gen6_set_rps(struct drm_device *dev, u8 val);
|
2013-04-18 05:54:58 +07:00
|
|
|
extern void valleyview_set_rps(struct drm_device *dev, u8 val);
|
|
|
|
extern int valleyview_rps_max_freq(struct drm_i915_private *dev_priv);
|
|
|
|
extern int valleyview_rps_min_freq(struct drm_i915_private *dev_priv);
|
2011-08-17 02:34:10 +07:00
|
|
|
extern void intel_detect_pch(struct drm_device *dev);
|
|
|
|
extern int intel_trans_dp_port_sel(struct drm_crtc *crtc);
|
2012-04-11 11:17:01 +07:00
|
|
|
extern int intel_enable_rc6(const struct drm_device *dev);
|
2010-04-07 15:15:53 +07:00
|
|
|
|
2012-04-06 04:47:36 +07:00
|
|
|
extern bool i915_semaphore_is_enabled(struct drm_device *dev);
|
2012-07-13 01:01:05 +07:00
|
|
|
int i915_reg_read_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file);
|
2013-10-30 20:44:16 +07:00
|
|
|
int i915_get_reset_stats_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file);
|
2012-03-29 03:39:37 +07:00
|
|
|
|
2010-08-05 02:26:07 +07:00
|
|
|
/* overlay */
|
|
|
|
extern struct intel_overlay_error_state *intel_overlay_capture_error_state(struct drm_device *dev);
|
2013-05-23 17:55:35 +07:00
|
|
|
extern void intel_overlay_print_error_state(struct drm_i915_error_state_buf *e,
|
|
|
|
struct intel_overlay_error_state *error);
|
2010-11-21 20:12:35 +07:00
|
|
|
|
|
|
|
extern struct intel_display_error_state *intel_display_capture_error_state(struct drm_device *dev);
|
2013-05-23 17:55:35 +07:00
|
|
|
extern void intel_display_print_error_state(struct drm_i915_error_state_buf *e,
|
2010-11-21 20:12:35 +07:00
|
|
|
struct drm_device *dev,
|
|
|
|
struct intel_display_error_state *error);
|
2010-08-05 02:26:07 +07:00
|
|
|
|
2011-04-26 01:22:22 +07:00
|
|
|
/* On SNB platform, before reading ring registers forcewake bit
|
|
|
|
* must be set to prevent GT core from power down and stale values being
|
|
|
|
* returned.
|
|
|
|
*/
|
2013-11-23 16:25:42 +07:00
|
|
|
void gen6_gt_force_wake_get(struct drm_i915_private *dev_priv, int fw_engine);
|
|
|
|
void gen6_gt_force_wake_put(struct drm_i915_private *dev_priv, int fw_engine);
|
2011-04-26 01:22:22 +07:00
|
|
|
|
2012-09-27 00:34:00 +07:00
|
|
|
int sandybridge_pcode_read(struct drm_i915_private *dev_priv, u8 mbox, u32 *val);
|
|
|
|
int sandybridge_pcode_write(struct drm_i915_private *dev_priv, u8 mbox, u32 val);
|
2013-05-22 19:36:16 +07:00
|
|
|
|
|
|
|
/* intel_sideband.c */
|
2013-05-22 19:36:20 +07:00
|
|
|
u32 vlv_punit_read(struct drm_i915_private *dev_priv, u8 addr);
|
|
|
|
void vlv_punit_write(struct drm_i915_private *dev_priv, u8 addr, u32 val);
|
|
|
|
u32 vlv_nc_read(struct drm_i915_private *dev_priv, u8 addr);
|
2013-08-27 19:12:14 +07:00
|
|
|
u32 vlv_gpio_nc_read(struct drm_i915_private *dev_priv, u32 reg);
|
|
|
|
void vlv_gpio_nc_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
|
|
|
|
u32 vlv_cck_read(struct drm_i915_private *dev_priv, u32 reg);
|
|
|
|
void vlv_cck_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
|
|
|
|
u32 vlv_ccu_read(struct drm_i915_private *dev_priv, u32 reg);
|
|
|
|
void vlv_ccu_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
|
2013-11-05 02:52:44 +07:00
|
|
|
u32 vlv_bunit_read(struct drm_i915_private *dev_priv, u32 reg);
|
|
|
|
void vlv_bunit_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
|
2013-08-27 19:12:14 +07:00
|
|
|
u32 vlv_gps_core_read(struct drm_i915_private *dev_priv, u32 reg);
|
|
|
|
void vlv_gps_core_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
|
2013-09-05 19:41:49 +07:00
|
|
|
u32 vlv_dpio_read(struct drm_i915_private *dev_priv, enum pipe pipe, int reg);
|
|
|
|
void vlv_dpio_write(struct drm_i915_private *dev_priv, enum pipe pipe, int reg, u32 val);
|
2013-05-22 19:36:16 +07:00
|
|
|
u32 intel_sbi_read(struct drm_i915_private *dev_priv, u16 reg,
|
|
|
|
enum intel_sbi_destination destination);
|
|
|
|
void intel_sbi_write(struct drm_i915_private *dev_priv, u16 reg, u32 value,
|
|
|
|
enum intel_sbi_destination destination);
|
2013-12-10 13:44:55 +07:00
|
|
|
u32 vlv_flisdsi_read(struct drm_i915_private *dev_priv, u32 reg);
|
|
|
|
void vlv_flisdsi_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
|
2013-04-18 05:54:58 +07:00
|
|
|
|
2013-11-06 03:42:29 +07:00
|
|
|
int vlv_gpu_freq(struct drm_i915_private *dev_priv, int val);
|
|
|
|
int vlv_freq_opcode(struct drm_i915_private *dev_priv, int val);
|
2012-09-27 00:34:00 +07:00
|
|
|
|
2013-11-23 16:25:43 +07:00
|
|
|
void vlv_force_wake_get(struct drm_i915_private *dev_priv, int fw_engine);
|
|
|
|
void vlv_force_wake_put(struct drm_i915_private *dev_priv, int fw_engine);
|
|
|
|
|
|
|
|
#define FORCEWAKE_VLV_RENDER_RANGE_OFFSET(reg) \
|
|
|
|
(((reg) >= 0x2000 && (reg) < 0x4000) ||\
|
|
|
|
((reg) >= 0x5000 && (reg) < 0x8000) ||\
|
|
|
|
((reg) >= 0xB000 && (reg) < 0x12000) ||\
|
|
|
|
((reg) >= 0x2E000 && (reg) < 0x30000))
|
|
|
|
|
|
|
|
#define FORCEWAKE_VLV_MEDIA_RANGE_OFFSET(reg)\
|
|
|
|
(((reg) >= 0x12000 && (reg) < 0x14000) ||\
|
|
|
|
((reg) >= 0x22000 && (reg) < 0x24000) ||\
|
|
|
|
((reg) >= 0x30000 && (reg) < 0x40000))
|
|
|
|
|
2013-11-23 16:25:42 +07:00
|
|
|
#define FORCEWAKE_RENDER (1 << 0)
|
|
|
|
#define FORCEWAKE_MEDIA (1 << 1)
|
|
|
|
#define FORCEWAKE_ALL (FORCEWAKE_RENDER | FORCEWAKE_MEDIA)
|
|
|
|
|
|
|
|
|
2013-10-05 11:22:51 +07:00
|
|
|
#define I915_READ8(reg) dev_priv->uncore.funcs.mmio_readb(dev_priv, (reg), true)
|
|
|
|
#define I915_WRITE8(reg, val) dev_priv->uncore.funcs.mmio_writeb(dev_priv, (reg), (val), true)
|
|
|
|
|
|
|
|
#define I915_READ16(reg) dev_priv->uncore.funcs.mmio_readw(dev_priv, (reg), true)
|
|
|
|
#define I915_WRITE16(reg, val) dev_priv->uncore.funcs.mmio_writew(dev_priv, (reg), (val), true)
|
|
|
|
#define I915_READ16_NOTRACE(reg) dev_priv->uncore.funcs.mmio_readw(dev_priv, (reg), false)
|
|
|
|
#define I915_WRITE16_NOTRACE(reg, val) dev_priv->uncore.funcs.mmio_writew(dev_priv, (reg), (val), false)
|
|
|
|
|
|
|
|
#define I915_READ(reg) dev_priv->uncore.funcs.mmio_readl(dev_priv, (reg), true)
|
|
|
|
#define I915_WRITE(reg, val) dev_priv->uncore.funcs.mmio_writel(dev_priv, (reg), (val), true)
|
|
|
|
#define I915_READ_NOTRACE(reg) dev_priv->uncore.funcs.mmio_readl(dev_priv, (reg), false)
|
|
|
|
#define I915_WRITE_NOTRACE(reg, val) dev_priv->uncore.funcs.mmio_writel(dev_priv, (reg), (val), false)
|
|
|
|
|
|
|
|
#define I915_WRITE64(reg, val) dev_priv->uncore.funcs.mmio_writeq(dev_priv, (reg), (val), true)
|
|
|
|
#define I915_READ64(reg) dev_priv->uncore.funcs.mmio_readq(dev_priv, (reg), true)
|
2010-11-09 16:17:32 +07:00
|
|
|
|
|
|
|
#define POSTING_READ(reg) (void)I915_READ_NOTRACE(reg)
|
|
|
|
#define POSTING_READ16(reg) (void)I915_READ16_NOTRACE(reg)
|
|
|
|
|
2013-01-17 21:31:29 +07:00
|
|
|
/* "Broadcast RGB" property */
|
|
|
|
#define INTEL_BROADCAST_RGB_AUTO 0
|
|
|
|
#define INTEL_BROADCAST_RGB_FULL 1
|
|
|
|
#define INTEL_BROADCAST_RGB_LIMITED 2
|
2010-11-08 16:09:41 +07:00
|
|
|
|
2013-01-26 02:44:46 +07:00
|
|
|
static inline uint32_t i915_vgacntrl_reg(struct drm_device *dev)
|
|
|
|
{
|
|
|
|
if (HAS_PCH_SPLIT(dev))
|
|
|
|
return CPU_VGACNTRL;
|
|
|
|
else if (IS_VALLEYVIEW(dev))
|
|
|
|
return VLV_VGACNTRL;
|
|
|
|
else
|
|
|
|
return VGACNTRL;
|
|
|
|
}
|
|
|
|
|
2013-02-22 21:12:51 +07:00
|
|
|
static inline void __user *to_user_ptr(u64 address)
|
|
|
|
{
|
|
|
|
return (void __user *)(uintptr_t)address;
|
|
|
|
}
|
|
|
|
|
2013-05-22 00:03:17 +07:00
|
|
|
static inline unsigned long msecs_to_jiffies_timeout(const unsigned int m)
|
|
|
|
{
|
|
|
|
unsigned long j = msecs_to_jiffies(m);
|
|
|
|
|
|
|
|
return min_t(unsigned long, MAX_JIFFY_OFFSET, j + 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline unsigned long
|
|
|
|
timespec_to_jiffies_timeout(const struct timespec *value)
|
|
|
|
{
|
|
|
|
unsigned long j = timespec_to_jiffies(value);
|
|
|
|
|
|
|
|
return min_t(unsigned long, MAX_JIFFY_OFFSET, j + 1);
|
|
|
|
}
|
|
|
|
|
2013-12-19 23:29:40 +07:00
|
|
|
/*
|
|
|
|
* If you need to wait X milliseconds between events A and B, but event B
|
|
|
|
* doesn't happen exactly after event A, you record the timestamp (jiffies) of
|
|
|
|
* when event A happened, then just before event B you call this function and
|
|
|
|
* pass the timestamp as the first argument, and X as the second argument.
|
|
|
|
*/
|
|
|
|
static inline void
|
|
|
|
wait_remaining_ms_from_jiffies(unsigned long timestamp_jiffies, int to_wait_ms)
|
|
|
|
{
|
2014-01-29 18:25:40 +07:00
|
|
|
unsigned long target_jiffies, tmp_jiffies, remaining_jiffies;
|
2013-12-19 23:29:40 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Don't re-read the value of "jiffies" every time since it may change
|
|
|
|
* behind our back and break the math.
|
|
|
|
*/
|
|
|
|
tmp_jiffies = jiffies;
|
|
|
|
target_jiffies = timestamp_jiffies +
|
|
|
|
msecs_to_jiffies_timeout(to_wait_ms);
|
|
|
|
|
|
|
|
if (time_after(target_jiffies, tmp_jiffies)) {
|
2014-01-29 18:25:40 +07:00
|
|
|
remaining_jiffies = target_jiffies - tmp_jiffies;
|
|
|
|
while (remaining_jiffies)
|
|
|
|
remaining_jiffies =
|
|
|
|
schedule_timeout_uninterruptible(remaining_jiffies);
|
2013-12-19 23:29:40 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
#endif
|