2009-09-16 03:57:34 +07:00
|
|
|
/*
|
|
|
|
* Copyright © 2009
|
|
|
|
*
|
|
|
|
* 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, sublicense,
|
|
|
|
* 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 NONINFRINGEMENT. IN NO EVENT SHALL
|
|
|
|
* THE AUTHORS OR COPYRIGHT HOLDERS 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.
|
|
|
|
*
|
|
|
|
* Authors:
|
|
|
|
* Daniel Vetter <daniel@ffwll.ch>
|
|
|
|
*
|
|
|
|
* Derived from Xorg ddx, xf86-video-intel, src/i830_video.c
|
|
|
|
*/
|
2012-10-03 00:01:07 +07:00
|
|
|
#include <drm/drmP.h>
|
|
|
|
#include <drm/i915_drm.h>
|
2009-09-16 03:57:34 +07:00
|
|
|
#include "i915_drv.h"
|
|
|
|
#include "i915_reg.h"
|
|
|
|
#include "intel_drv.h"
|
2016-08-04 22:32:35 +07:00
|
|
|
#include "intel_frontbuffer.h"
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
/* Limits for overlay size. According to intel doc, the real limits are:
|
|
|
|
* Y width: 4095, UV width (planar): 2047, Y height: 2047,
|
|
|
|
* UV width (planar): * 1023. But the xorg thinks 2048 for height and width. Use
|
|
|
|
* the mininum of both. */
|
|
|
|
#define IMAGE_MAX_WIDTH 2048
|
|
|
|
#define IMAGE_MAX_HEIGHT 2046 /* 2 * 1023 */
|
|
|
|
/* on 830 and 845 these large limits result in the card hanging */
|
|
|
|
#define IMAGE_MAX_WIDTH_LEGACY 1024
|
|
|
|
#define IMAGE_MAX_HEIGHT_LEGACY 1088
|
|
|
|
|
|
|
|
/* overlay register definitions */
|
|
|
|
/* OCMD register */
|
|
|
|
#define OCMD_TILED_SURFACE (0x1<<19)
|
|
|
|
#define OCMD_MIRROR_MASK (0x3<<17)
|
|
|
|
#define OCMD_MIRROR_MODE (0x3<<17)
|
|
|
|
#define OCMD_MIRROR_HORIZONTAL (0x1<<17)
|
|
|
|
#define OCMD_MIRROR_VERTICAL (0x2<<17)
|
|
|
|
#define OCMD_MIRROR_BOTH (0x3<<17)
|
|
|
|
#define OCMD_BYTEORDER_MASK (0x3<<14) /* zero for YUYV or FOURCC YUY2 */
|
|
|
|
#define OCMD_UV_SWAP (0x1<<14) /* YVYU */
|
|
|
|
#define OCMD_Y_SWAP (0x2<<14) /* UYVY or FOURCC UYVY */
|
|
|
|
#define OCMD_Y_AND_UV_SWAP (0x3<<14) /* VYUY */
|
|
|
|
#define OCMD_SOURCE_FORMAT_MASK (0xf<<10)
|
|
|
|
#define OCMD_RGB_888 (0x1<<10) /* not in i965 Intel docs */
|
|
|
|
#define OCMD_RGB_555 (0x2<<10) /* not in i965 Intel docs */
|
|
|
|
#define OCMD_RGB_565 (0x3<<10) /* not in i965 Intel docs */
|
|
|
|
#define OCMD_YUV_422_PACKED (0x8<<10)
|
|
|
|
#define OCMD_YUV_411_PACKED (0x9<<10) /* not in i965 Intel docs */
|
|
|
|
#define OCMD_YUV_420_PLANAR (0xc<<10)
|
|
|
|
#define OCMD_YUV_422_PLANAR (0xd<<10)
|
|
|
|
#define OCMD_YUV_410_PLANAR (0xe<<10) /* also 411 */
|
|
|
|
#define OCMD_TVSYNCFLIP_PARITY (0x1<<9)
|
|
|
|
#define OCMD_TVSYNCFLIP_ENABLE (0x1<<7)
|
2010-07-13 19:52:17 +07:00
|
|
|
#define OCMD_BUF_TYPE_MASK (0x1<<5)
|
2009-09-16 03:57:34 +07:00
|
|
|
#define OCMD_BUF_TYPE_FRAME (0x0<<5)
|
|
|
|
#define OCMD_BUF_TYPE_FIELD (0x1<<5)
|
|
|
|
#define OCMD_TEST_MODE (0x1<<4)
|
|
|
|
#define OCMD_BUFFER_SELECT (0x3<<2)
|
|
|
|
#define OCMD_BUFFER0 (0x0<<2)
|
|
|
|
#define OCMD_BUFFER1 (0x1<<2)
|
|
|
|
#define OCMD_FIELD_SELECT (0x1<<2)
|
|
|
|
#define OCMD_FIELD0 (0x0<<1)
|
|
|
|
#define OCMD_FIELD1 (0x1<<1)
|
|
|
|
#define OCMD_ENABLE (0x1<<0)
|
|
|
|
|
|
|
|
/* OCONFIG register */
|
|
|
|
#define OCONF_PIPE_MASK (0x1<<18)
|
|
|
|
#define OCONF_PIPE_A (0x0<<18)
|
|
|
|
#define OCONF_PIPE_B (0x1<<18)
|
|
|
|
#define OCONF_GAMMA2_ENABLE (0x1<<16)
|
|
|
|
#define OCONF_CSC_MODE_BT601 (0x0<<5)
|
|
|
|
#define OCONF_CSC_MODE_BT709 (0x1<<5)
|
|
|
|
#define OCONF_CSC_BYPASS (0x1<<4)
|
|
|
|
#define OCONF_CC_OUT_8BIT (0x1<<3)
|
|
|
|
#define OCONF_TEST_MODE (0x1<<2)
|
|
|
|
#define OCONF_THREE_LINE_BUFFER (0x1<<0)
|
|
|
|
#define OCONF_TWO_LINE_BUFFER (0x0<<0)
|
|
|
|
|
|
|
|
/* DCLRKM (dst-key) register */
|
|
|
|
#define DST_KEY_ENABLE (0x1<<31)
|
|
|
|
#define CLK_RGB24_MASK 0x0
|
|
|
|
#define CLK_RGB16_MASK 0x070307
|
|
|
|
#define CLK_RGB15_MASK 0x070707
|
|
|
|
#define CLK_RGB8I_MASK 0xffffff
|
|
|
|
|
|
|
|
#define RGB16_TO_COLORKEY(c) \
|
|
|
|
(((c & 0xF800) << 8) | ((c & 0x07E0) << 5) | ((c & 0x001F) << 3))
|
|
|
|
#define RGB15_TO_COLORKEY(c) \
|
|
|
|
(((c & 0x7c00) << 9) | ((c & 0x03E0) << 6) | ((c & 0x001F) << 3))
|
|
|
|
|
|
|
|
/* overlay flip addr flag */
|
|
|
|
#define OFC_UPDATE 0x1
|
|
|
|
|
|
|
|
/* polyphase filter coefficients */
|
|
|
|
#define N_HORIZ_Y_TAPS 5
|
|
|
|
#define N_VERT_Y_TAPS 3
|
|
|
|
#define N_HORIZ_UV_TAPS 3
|
|
|
|
#define N_VERT_UV_TAPS 3
|
|
|
|
#define N_PHASES 17
|
|
|
|
#define MAX_TAPS 5
|
|
|
|
|
|
|
|
/* memory bufferd overlay registers */
|
|
|
|
struct overlay_registers {
|
2011-08-17 02:34:10 +07:00
|
|
|
u32 OBUF_0Y;
|
|
|
|
u32 OBUF_1Y;
|
|
|
|
u32 OBUF_0U;
|
|
|
|
u32 OBUF_0V;
|
|
|
|
u32 OBUF_1U;
|
|
|
|
u32 OBUF_1V;
|
|
|
|
u32 OSTRIDE;
|
|
|
|
u32 YRGB_VPH;
|
|
|
|
u32 UV_VPH;
|
|
|
|
u32 HORZ_PH;
|
|
|
|
u32 INIT_PHS;
|
|
|
|
u32 DWINPOS;
|
|
|
|
u32 DWINSZ;
|
|
|
|
u32 SWIDTH;
|
|
|
|
u32 SWIDTHSW;
|
|
|
|
u32 SHEIGHT;
|
|
|
|
u32 YRGBSCALE;
|
|
|
|
u32 UVSCALE;
|
|
|
|
u32 OCLRC0;
|
|
|
|
u32 OCLRC1;
|
|
|
|
u32 DCLRKV;
|
|
|
|
u32 DCLRKM;
|
|
|
|
u32 SCLRKVH;
|
|
|
|
u32 SCLRKVL;
|
|
|
|
u32 SCLRKEN;
|
|
|
|
u32 OCONFIG;
|
|
|
|
u32 OCMD;
|
|
|
|
u32 RESERVED1; /* 0x6C */
|
|
|
|
u32 OSTART_0Y;
|
|
|
|
u32 OSTART_1Y;
|
|
|
|
u32 OSTART_0U;
|
|
|
|
u32 OSTART_0V;
|
|
|
|
u32 OSTART_1U;
|
|
|
|
u32 OSTART_1V;
|
|
|
|
u32 OTILEOFF_0Y;
|
|
|
|
u32 OTILEOFF_1Y;
|
|
|
|
u32 OTILEOFF_0U;
|
|
|
|
u32 OTILEOFF_0V;
|
|
|
|
u32 OTILEOFF_1U;
|
|
|
|
u32 OTILEOFF_1V;
|
|
|
|
u32 FASTHSCALE; /* 0xA0 */
|
|
|
|
u32 UVSCALEV; /* 0xA4 */
|
|
|
|
u32 RESERVEDC[(0x200 - 0xA8) / 4]; /* 0xA8 - 0x1FC */
|
|
|
|
u16 Y_VCOEFS[N_VERT_Y_TAPS * N_PHASES]; /* 0x200 */
|
|
|
|
u16 RESERVEDD[0x100 / 2 - N_VERT_Y_TAPS * N_PHASES];
|
|
|
|
u16 Y_HCOEFS[N_HORIZ_Y_TAPS * N_PHASES]; /* 0x300 */
|
|
|
|
u16 RESERVEDE[0x200 / 2 - N_HORIZ_Y_TAPS * N_PHASES];
|
|
|
|
u16 UV_VCOEFS[N_VERT_UV_TAPS * N_PHASES]; /* 0x500 */
|
|
|
|
u16 RESERVEDF[0x100 / 2 - N_VERT_UV_TAPS * N_PHASES];
|
|
|
|
u16 UV_HCOEFS[N_HORIZ_UV_TAPS * N_PHASES]; /* 0x600 */
|
|
|
|
u16 RESERVEDG[0x100 / 2 - N_HORIZ_UV_TAPS * N_PHASES];
|
2009-09-16 03:57:34 +07:00
|
|
|
};
|
|
|
|
|
2010-08-12 19:53:37 +07:00
|
|
|
struct intel_overlay {
|
2016-05-12 18:43:23 +07:00
|
|
|
struct drm_i915_private *i915;
|
2010-08-12 19:53:37 +07:00
|
|
|
struct intel_crtc *crtc;
|
2016-08-15 16:49:01 +07:00
|
|
|
struct i915_vma *vma;
|
|
|
|
struct i915_vma *old_vma;
|
2015-03-31 14:37:23 +07:00
|
|
|
bool active;
|
|
|
|
bool pfit_active;
|
2010-08-12 19:53:37 +07:00
|
|
|
u32 pfit_vscale_ratio; /* shifted-point number, (1<<12) == 1.0 */
|
2015-04-02 16:35:08 +07:00
|
|
|
u32 color_key:24;
|
|
|
|
u32 color_key_enabled:1;
|
2010-08-12 19:53:37 +07:00
|
|
|
u32 brightness, contrast, saturation;
|
|
|
|
u32 old_xscale, old_yscale;
|
|
|
|
/* register access */
|
|
|
|
struct drm_i915_gem_object *reg_bo;
|
2018-09-07 02:01:43 +07:00
|
|
|
struct overlay_registers __iomem *regs;
|
|
|
|
u32 flip_addr;
|
2010-08-12 19:53:37 +07:00
|
|
|
/* flip handling */
|
2016-08-04 13:52:37 +07:00
|
|
|
struct i915_gem_active last_flip;
|
2010-08-12 19:53:37 +07:00
|
|
|
};
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2016-12-08 00:28:12 +07:00
|
|
|
static void i830_overlay_clock_gating(struct drm_i915_private *dev_priv,
|
|
|
|
bool enable)
|
|
|
|
{
|
|
|
|
struct pci_dev *pdev = dev_priv->drm.pdev;
|
|
|
|
u8 val;
|
|
|
|
|
|
|
|
/* WA_OVERLAY_CLKGATE:alm */
|
|
|
|
if (enable)
|
|
|
|
I915_WRITE(DSPCLK_GATE_D, 0);
|
|
|
|
else
|
|
|
|
I915_WRITE(DSPCLK_GATE_D, OVRUNIT_CLOCK_GATE_DISABLE);
|
|
|
|
|
|
|
|
/* WA_DISABLE_L2CACHE_CLOCK_GATING:alm */
|
|
|
|
pci_bus_read_config_byte(pdev->bus,
|
|
|
|
PCI_DEVFN(0, 0), I830_CLOCK_GATE, &val);
|
|
|
|
if (enable)
|
|
|
|
val &= ~I830_L2_CACHE_CLOCK_GATE_DISABLE;
|
|
|
|
else
|
|
|
|
val |= I830_L2_CACHE_CLOCK_GATE_DISABLE;
|
|
|
|
pci_bus_write_config_byte(pdev->bus,
|
|
|
|
PCI_DEVFN(0, 0), I830_CLOCK_GATE, val);
|
|
|
|
}
|
|
|
|
|
2016-08-04 13:52:37 +07:00
|
|
|
static void intel_overlay_submit_request(struct intel_overlay *overlay,
|
2018-02-21 16:56:36 +07:00
|
|
|
struct i915_request *rq,
|
2016-08-04 13:52:37 +07:00
|
|
|
i915_gem_retire_fn retire)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2016-08-04 13:52:37 +07:00
|
|
|
GEM_BUG_ON(i915_gem_active_peek(&overlay->last_flip,
|
|
|
|
&overlay->i915->drm.struct_mutex));
|
2016-12-08 00:56:47 +07:00
|
|
|
i915_gem_active_set_retire_fn(&overlay->last_flip, retire,
|
|
|
|
&overlay->i915->drm.struct_mutex);
|
2018-02-21 16:56:36 +07:00
|
|
|
i915_gem_active_set(&overlay->last_flip, rq);
|
|
|
|
i915_request_add(rq);
|
2016-08-04 13:52:37 +07:00
|
|
|
}
|
2012-09-26 19:47:30 +07:00
|
|
|
|
2016-08-04 13:52:37 +07:00
|
|
|
static int intel_overlay_do_wait_request(struct intel_overlay *overlay,
|
2018-02-21 16:56:36 +07:00
|
|
|
struct i915_request *rq,
|
2016-08-04 13:52:37 +07:00
|
|
|
i915_gem_retire_fn retire)
|
|
|
|
{
|
2018-02-21 16:56:36 +07:00
|
|
|
intel_overlay_submit_request(overlay, rq, retire);
|
2016-08-04 13:52:37 +07:00
|
|
|
return i915_gem_active_retire(&overlay->last_flip,
|
|
|
|
&overlay->i915->drm.struct_mutex);
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
static struct i915_request *alloc_request(struct intel_overlay *overlay)
|
2016-08-03 04:50:26 +07:00
|
|
|
{
|
|
|
|
struct drm_i915_private *dev_priv = overlay->i915;
|
drm/i915: Allocate intel_engine_cs structure only for the enabled engines
With the possibility of addition of many more number of rings in future,
the drm_i915_private structure could bloat as an array, of type
intel_engine_cs, is embedded inside it.
struct intel_engine_cs engine[I915_NUM_ENGINES];
Though this is still fine as generally there is only a single instance of
drm_i915_private structure used, but not all of the possible rings would be
enabled or active on most of the platforms. Some memory can be saved by
allocating intel_engine_cs structure only for the enabled/active engines.
Currently the engine/ring ID is kept static and dev_priv->engine[] is simply
indexed using the enums defined in intel_engine_id.
To save memory and continue using the static engine/ring IDs, 'engine' is
defined as an array of pointers.
struct intel_engine_cs *engine[I915_NUM_ENGINES];
dev_priv->engine[engine_ID] will be NULL for disabled engine instances.
There is a text size reduction of 928 bytes, from 1028200 to 1027272, for
i915.o file (but for i915.ko file text size remain same as 1193131 bytes).
v2:
- Remove the engine iterator field added in drm_i915_private structure,
instead pass a local iterator variable to the for_each_engine**
macros. (Chris)
- Do away with intel_engine_initialized() and instead directly use the
NULL pointer check on engine pointer. (Chris)
v3:
- Remove for_each_engine_id() macro, as the updated macro for_each_engine()
can be used in place of it. (Chris)
- Protect the access to Render engine Fault register with a NULL check, as
engine specific init is done later in Driver load sequence.
v4:
- Use !!dev_priv->engine[VCS] style for the engine check in getparam. (Chris)
- Kill the superfluous init_engine_lists().
v5:
- Cleanup the intel_engines_init() & intel_engines_setup(), with respect to
allocation of intel_engine_cs structure. (Chris)
v6:
- Rebase.
v7:
- Optimize the for_each_engine_masked() macro. (Chris)
- Change the type of 'iter' local variable to enum intel_engine_id. (Chris)
- Rebase.
v8: Rebase.
v9: Rebase.
v10:
- For index calculation use engine ID instead of pointer based arithmetic in
intel_engine_sync_index() as engine pointers are not contiguous now (Chris)
- For appropriateness, rename local enum variable 'iter' to 'id'. (Joonas)
- Use for_each_engine macro for cleanup in intel_engines_init() and remove
check for NULL engine pointer in cleanup() routines. (Joonas)
v11: Rebase.
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1476378888-7372-1-git-send-email-akash.goel@intel.com
2016-10-14 00:14:48 +07:00
|
|
|
struct intel_engine_cs *engine = dev_priv->engine[RCS];
|
2016-08-03 04:50:26 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
return i915_request_alloc(engine, dev_priv->kernel_context);
|
2016-08-03 04:50:26 +07:00
|
|
|
}
|
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
/* overlay needs to be disable in OCMD reg */
|
|
|
|
static int intel_overlay_on(struct intel_overlay *overlay)
|
|
|
|
{
|
2016-05-12 18:43:23 +07:00
|
|
|
struct drm_i915_private *dev_priv = overlay->i915;
|
2018-02-21 16:56:36 +07:00
|
|
|
struct i915_request *rq;
|
2017-02-14 18:32:42 +07:00
|
|
|
u32 *cs;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2015-03-31 14:37:22 +07:00
|
|
|
WARN_ON(overlay->active);
|
2010-07-16 23:13:01 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
rq = alloc_request(overlay);
|
|
|
|
if (IS_ERR(rq))
|
|
|
|
return PTR_ERR(rq);
|
2010-10-27 18:45:26 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
cs = intel_ring_begin(rq, 4);
|
2017-02-14 18:32:42 +07:00
|
|
|
if (IS_ERR(cs)) {
|
2018-02-21 16:56:36 +07:00
|
|
|
i915_request_add(rq);
|
2017-02-14 18:32:42 +07:00
|
|
|
return PTR_ERR(cs);
|
2015-05-29 23:43:47 +07:00
|
|
|
}
|
|
|
|
|
2015-03-31 14:37:24 +07:00
|
|
|
overlay->active = true;
|
|
|
|
|
2016-12-08 00:28:12 +07:00
|
|
|
if (IS_I830(dev_priv))
|
|
|
|
i830_overlay_clock_gating(dev_priv, false);
|
|
|
|
|
2017-02-14 18:32:42 +07:00
|
|
|
*cs++ = MI_OVERLAY_FLIP | MI_OVERLAY_ON;
|
|
|
|
*cs++ = overlay->flip_addr | OFC_UPDATE;
|
|
|
|
*cs++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
|
|
|
|
*cs++ = MI_NOOP;
|
2018-02-21 16:56:36 +07:00
|
|
|
intel_ring_advance(rq, cs);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
return intel_overlay_do_wait_request(overlay, rq, NULL);
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
2016-12-08 00:28:06 +07:00
|
|
|
static void intel_overlay_flip_prepare(struct intel_overlay *overlay,
|
|
|
|
struct i915_vma *vma)
|
|
|
|
{
|
|
|
|
enum pipe pipe = overlay->crtc->pipe;
|
|
|
|
|
|
|
|
WARN_ON(overlay->old_vma);
|
|
|
|
|
|
|
|
i915_gem_track_fb(overlay->vma ? overlay->vma->obj : NULL,
|
|
|
|
vma ? vma->obj : NULL,
|
|
|
|
INTEL_FRONTBUFFER_OVERLAY(pipe));
|
|
|
|
|
|
|
|
intel_frontbuffer_flip_prepare(overlay->i915,
|
|
|
|
INTEL_FRONTBUFFER_OVERLAY(pipe));
|
|
|
|
|
|
|
|
overlay->old_vma = overlay->vma;
|
|
|
|
if (vma)
|
|
|
|
overlay->vma = i915_vma_get(vma);
|
|
|
|
else
|
|
|
|
overlay->vma = NULL;
|
|
|
|
}
|
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
/* overlay needs to be enabled in OCMD reg */
|
2010-08-12 18:36:12 +07:00
|
|
|
static int intel_overlay_continue(struct intel_overlay *overlay,
|
2016-12-08 00:28:06 +07:00
|
|
|
struct i915_vma *vma,
|
2010-08-12 18:36:12 +07:00
|
|
|
bool load_polyphase_filter)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2016-05-12 18:43:23 +07:00
|
|
|
struct drm_i915_private *dev_priv = overlay->i915;
|
2018-02-21 16:56:36 +07:00
|
|
|
struct i915_request *rq;
|
2009-09-16 03:57:34 +07:00
|
|
|
u32 flip_addr = overlay->flip_addr;
|
2017-02-14 18:32:42 +07:00
|
|
|
u32 tmp, *cs;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2015-03-31 14:37:22 +07:00
|
|
|
WARN_ON(!overlay->active);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
if (load_polyphase_filter)
|
|
|
|
flip_addr |= OFC_UPDATE;
|
|
|
|
|
|
|
|
/* check for underruns */
|
|
|
|
tmp = I915_READ(DOVSTA);
|
|
|
|
if (tmp & (1 << 17))
|
|
|
|
DRM_DEBUG("overlay underrun, DOVSTA: %x\n", tmp);
|
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
rq = alloc_request(overlay);
|
|
|
|
if (IS_ERR(rq))
|
|
|
|
return PTR_ERR(rq);
|
2012-09-26 19:47:30 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
cs = intel_ring_begin(rq, 2);
|
2017-02-14 18:32:42 +07:00
|
|
|
if (IS_ERR(cs)) {
|
2018-02-21 16:56:36 +07:00
|
|
|
i915_request_add(rq);
|
2017-02-14 18:32:42 +07:00
|
|
|
return PTR_ERR(cs);
|
2015-05-29 23:43:47 +07:00
|
|
|
}
|
|
|
|
|
2017-02-14 18:32:42 +07:00
|
|
|
*cs++ = MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE;
|
|
|
|
*cs++ = flip_addr;
|
2018-02-21 16:56:36 +07:00
|
|
|
intel_ring_advance(rq, cs);
|
2009-09-16 03:57:36 +07:00
|
|
|
|
2016-12-08 00:28:06 +07:00
|
|
|
intel_overlay_flip_prepare(overlay, vma);
|
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
intel_overlay_submit_request(overlay, rq, NULL);
|
2015-05-29 23:43:24 +07:00
|
|
|
|
|
|
|
return 0;
|
2009-09-16 03:57:36 +07:00
|
|
|
}
|
|
|
|
|
2016-12-08 00:28:06 +07:00
|
|
|
static void intel_overlay_release_old_vma(struct intel_overlay *overlay)
|
2009-09-16 03:57:36 +07:00
|
|
|
{
|
2016-08-15 16:49:01 +07:00
|
|
|
struct i915_vma *vma;
|
2009-09-16 03:57:36 +07:00
|
|
|
|
2016-08-15 16:49:01 +07:00
|
|
|
vma = fetch_and_zero(&overlay->old_vma);
|
|
|
|
if (WARN_ON(!vma))
|
|
|
|
return;
|
2016-08-04 13:52:37 +07:00
|
|
|
|
2016-12-08 00:28:06 +07:00
|
|
|
intel_frontbuffer_flip_complete(overlay->i915,
|
|
|
|
INTEL_FRONTBUFFER_OVERLAY(overlay->crtc->pipe));
|
2009-09-16 03:57:36 +07:00
|
|
|
|
2016-08-15 16:49:06 +07:00
|
|
|
i915_gem_object_unpin_from_display_plane(vma);
|
2016-08-15 16:49:01 +07:00
|
|
|
i915_vma_put(vma);
|
2010-08-12 20:03:48 +07:00
|
|
|
}
|
2009-09-16 03:57:37 +07:00
|
|
|
|
2016-12-08 00:28:06 +07:00
|
|
|
static void intel_overlay_release_old_vid_tail(struct i915_gem_active *active,
|
2018-02-21 16:56:36 +07:00
|
|
|
struct i915_request *rq)
|
2016-12-08 00:28:06 +07:00
|
|
|
{
|
|
|
|
struct intel_overlay *overlay =
|
|
|
|
container_of(active, typeof(*overlay), last_flip);
|
|
|
|
|
|
|
|
intel_overlay_release_old_vma(overlay);
|
|
|
|
}
|
|
|
|
|
2016-08-04 13:52:37 +07:00
|
|
|
static void intel_overlay_off_tail(struct i915_gem_active *active,
|
2018-02-21 16:56:36 +07:00
|
|
|
struct i915_request *rq)
|
2010-08-12 20:03:48 +07:00
|
|
|
{
|
2016-08-04 13:52:37 +07:00
|
|
|
struct intel_overlay *overlay =
|
|
|
|
container_of(active, typeof(*overlay), last_flip);
|
2016-12-08 00:28:12 +07:00
|
|
|
struct drm_i915_private *dev_priv = overlay->i915;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2016-12-08 00:28:06 +07:00
|
|
|
intel_overlay_release_old_vma(overlay);
|
2009-09-16 03:57:37 +07:00
|
|
|
|
2010-08-12 20:03:48 +07:00
|
|
|
overlay->crtc->overlay = NULL;
|
|
|
|
overlay->crtc = NULL;
|
2015-03-31 14:37:23 +07:00
|
|
|
overlay->active = false;
|
2016-12-08 00:28:12 +07:00
|
|
|
|
|
|
|
if (IS_I830(dev_priv))
|
|
|
|
i830_overlay_clock_gating(dev_priv, true);
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
/* overlay needs to be disabled in OCMD reg */
|
2011-02-21 21:43:56 +07:00
|
|
|
static int intel_overlay_off(struct intel_overlay *overlay)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2018-02-21 16:56:36 +07:00
|
|
|
struct i915_request *rq;
|
2017-02-14 18:32:42 +07:00
|
|
|
u32 *cs, flip_addr = overlay->flip_addr;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2015-03-31 14:37:22 +07:00
|
|
|
WARN_ON(!overlay->active);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
/* According to intel docs the overlay hw may hang (when switching
|
|
|
|
* off) without loading the filter coeffs. It is however unclear whether
|
|
|
|
* this applies to the disabling of the overlay or to the switching off
|
|
|
|
* of the hw. Do it in both cases */
|
|
|
|
flip_addr |= OFC_UPDATE;
|
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
rq = alloc_request(overlay);
|
|
|
|
if (IS_ERR(rq))
|
|
|
|
return PTR_ERR(rq);
|
2012-09-26 19:47:30 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
cs = intel_ring_begin(rq, 6);
|
2017-02-14 18:32:42 +07:00
|
|
|
if (IS_ERR(cs)) {
|
2018-02-21 16:56:36 +07:00
|
|
|
i915_request_add(rq);
|
2017-02-14 18:32:42 +07:00
|
|
|
return PTR_ERR(cs);
|
2015-05-29 23:43:47 +07:00
|
|
|
}
|
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
/* wait for overlay to go idle */
|
2017-02-14 18:32:42 +07:00
|
|
|
*cs++ = MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE;
|
|
|
|
*cs++ = flip_addr;
|
|
|
|
*cs++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
|
2016-12-23 02:52:22 +07:00
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
/* turn overlay off */
|
2017-02-14 18:32:42 +07:00
|
|
|
*cs++ = MI_OVERLAY_FLIP | MI_OVERLAY_OFF;
|
|
|
|
*cs++ = flip_addr;
|
|
|
|
*cs++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
|
2016-12-23 02:52:22 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
intel_ring_advance(rq, cs);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2016-12-08 00:28:06 +07:00
|
|
|
intel_overlay_flip_prepare(overlay, NULL);
|
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
return intel_overlay_do_wait_request(overlay, rq,
|
2016-08-04 13:52:37 +07:00
|
|
|
intel_overlay_off_tail);
|
2037-04-25 15:08:26 +07:00
|
|
|
}
|
|
|
|
|
2009-09-16 03:57:37 +07:00
|
|
|
/* recover from an interruption due to a signal
|
|
|
|
* We have to be careful not to repeat work forever an make forward progess. */
|
2011-02-21 21:43:56 +07:00
|
|
|
static int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay)
|
2009-09-16 03:57:37 +07:00
|
|
|
{
|
2016-08-04 13:52:37 +07:00
|
|
|
return i915_gem_active_retire(&overlay->last_flip,
|
|
|
|
&overlay->i915->drm.struct_mutex);
|
2009-09-16 03:57:37 +07:00
|
|
|
}
|
|
|
|
|
2009-09-16 03:57:36 +07:00
|
|
|
/* Wait for pending overlay flip and release old frame.
|
|
|
|
* Needs to be called before the overlay register are changed
|
2010-08-12 16:35:26 +07:00
|
|
|
* via intel_overlay_(un)map_regs
|
|
|
|
*/
|
2009-09-16 03:57:34 +07:00
|
|
|
static int intel_overlay_release_old_vid(struct intel_overlay *overlay)
|
|
|
|
{
|
2016-05-12 18:43:23 +07:00
|
|
|
struct drm_i915_private *dev_priv = overlay->i915;
|
2017-02-14 18:32:42 +07:00
|
|
|
u32 *cs;
|
2009-09-16 03:57:34 +07:00
|
|
|
int ret;
|
|
|
|
|
2016-07-05 16:40:23 +07:00
|
|
|
lockdep_assert_held(&dev_priv->drm.struct_mutex);
|
2014-11-26 22:07:29 +07:00
|
|
|
|
2010-08-12 18:21:54 +07:00
|
|
|
/* Only wait if there is actually an old frame to release to
|
|
|
|
* guarantee forward progress.
|
|
|
|
*/
|
2016-08-15 16:49:01 +07:00
|
|
|
if (!overlay->old_vma)
|
2009-09-16 03:57:37 +07:00
|
|
|
return 0;
|
|
|
|
|
2010-08-12 18:21:54 +07:00
|
|
|
if (I915_READ(ISR) & I915_OVERLAY_PLANE_FLIP_PENDING_INTERRUPT) {
|
|
|
|
/* synchronous slowpath */
|
2018-02-21 16:56:36 +07:00
|
|
|
struct i915_request *rq;
|
2015-05-29 23:43:47 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
rq = alloc_request(overlay);
|
|
|
|
if (IS_ERR(rq))
|
|
|
|
return PTR_ERR(rq);
|
2010-10-27 18:45:26 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
cs = intel_ring_begin(rq, 2);
|
2017-02-14 18:32:42 +07:00
|
|
|
if (IS_ERR(cs)) {
|
2018-02-21 16:56:36 +07:00
|
|
|
i915_request_add(rq);
|
2017-02-14 18:32:42 +07:00
|
|
|
return PTR_ERR(cs);
|
2015-05-29 23:43:47 +07:00
|
|
|
}
|
|
|
|
|
2017-02-14 18:32:42 +07:00
|
|
|
*cs++ = MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP;
|
|
|
|
*cs++ = MI_NOOP;
|
2018-02-21 16:56:36 +07:00
|
|
|
intel_ring_advance(rq, cs);
|
2010-08-12 18:21:54 +07:00
|
|
|
|
2018-02-21 16:56:36 +07:00
|
|
|
ret = intel_overlay_do_wait_request(overlay, rq,
|
2010-08-12 20:03:48 +07:00
|
|
|
intel_overlay_release_old_vid_tail);
|
2010-08-12 18:21:54 +07:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
2016-08-04 13:52:37 +07:00
|
|
|
} else
|
|
|
|
intel_overlay_release_old_vid_tail(&overlay->last_flip, NULL);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-11-26 22:07:29 +07:00
|
|
|
void intel_overlay_reset(struct drm_i915_private *dev_priv)
|
|
|
|
{
|
|
|
|
struct intel_overlay *overlay = dev_priv->overlay;
|
|
|
|
|
|
|
|
if (!overlay)
|
|
|
|
return;
|
|
|
|
|
|
|
|
intel_overlay_release_old_vid(overlay);
|
|
|
|
|
|
|
|
overlay->old_xscale = 0;
|
|
|
|
overlay->old_yscale = 0;
|
|
|
|
overlay->crtc = NULL;
|
|
|
|
overlay->active = false;
|
|
|
|
}
|
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
static int packed_depth_bytes(u32 format)
|
|
|
|
{
|
|
|
|
switch (format & I915_OVERLAY_DEPTH_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_YUV422:
|
|
|
|
return 4;
|
|
|
|
case I915_OVERLAY_YUV411:
|
|
|
|
/* return 6; not implemented */
|
|
|
|
default:
|
|
|
|
return -EINVAL;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int packed_width_bytes(u32 format, short width)
|
|
|
|
{
|
|
|
|
switch (format & I915_OVERLAY_DEPTH_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_YUV422:
|
|
|
|
return width << 1;
|
|
|
|
default:
|
|
|
|
return -EINVAL;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int uv_hsubsampling(u32 format)
|
|
|
|
{
|
|
|
|
switch (format & I915_OVERLAY_DEPTH_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_YUV422:
|
|
|
|
case I915_OVERLAY_YUV420:
|
|
|
|
return 2;
|
|
|
|
case I915_OVERLAY_YUV411:
|
|
|
|
case I915_OVERLAY_YUV410:
|
|
|
|
return 4;
|
|
|
|
default:
|
|
|
|
return -EINVAL;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int uv_vsubsampling(u32 format)
|
|
|
|
{
|
|
|
|
switch (format & I915_OVERLAY_DEPTH_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_YUV420:
|
|
|
|
case I915_OVERLAY_YUV410:
|
|
|
|
return 2;
|
|
|
|
case I915_OVERLAY_YUV422:
|
|
|
|
case I915_OVERLAY_YUV411:
|
|
|
|
return 1;
|
|
|
|
default:
|
|
|
|
return -EINVAL;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-05-12 18:43:23 +07:00
|
|
|
static u32 calc_swidthsw(struct drm_i915_private *dev_priv, u32 offset, u32 width)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2016-12-08 00:28:09 +07:00
|
|
|
u32 sw;
|
|
|
|
|
drm/i915: replace IS_GEN<N> with IS_GEN(..., N)
Define IS_GEN() similarly to our IS_GEN_RANGE(). but use gen instead of
gen_mask to do the comparison. Now callers can pass then gen as a parameter,
so we don't require one macro for each gen.
The following spatch was used to convert the users of these macros:
@@
expression e;
@@
(
- IS_GEN2(e)
+ IS_GEN(e, 2)
|
- IS_GEN3(e)
+ IS_GEN(e, 3)
|
- IS_GEN4(e)
+ IS_GEN(e, 4)
|
- IS_GEN5(e)
+ IS_GEN(e, 5)
|
- IS_GEN6(e)
+ IS_GEN(e, 6)
|
- IS_GEN7(e)
+ IS_GEN(e, 7)
|
- IS_GEN8(e)
+ IS_GEN(e, 8)
|
- IS_GEN9(e)
+ IS_GEN(e, 9)
|
- IS_GEN10(e)
+ IS_GEN(e, 10)
|
- IS_GEN11(e)
+ IS_GEN(e, 11)
)
v2: use IS_GEN rather than GT_GEN and compare to info.gen rather than
using the bitmask
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181212181044.15886-2-lucas.demarchi@intel.com
2018-12-13 01:10:43 +07:00
|
|
|
if (IS_GEN(dev_priv, 2))
|
2016-12-08 00:28:09 +07:00
|
|
|
sw = ALIGN((offset & 31) + width, 32);
|
|
|
|
else
|
|
|
|
sw = ALIGN((offset & 63) + width, 64);
|
|
|
|
|
|
|
|
if (sw == 0)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
return (sw - 32) >> 3;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
2016-12-08 00:28:10 +07:00
|
|
|
static const u16 y_static_hcoeffs[N_PHASES][N_HORIZ_Y_TAPS] = {
|
|
|
|
[ 0] = { 0x3000, 0xb4a0, 0x1930, 0x1920, 0xb4a0, },
|
|
|
|
[ 1] = { 0x3000, 0xb500, 0x19d0, 0x1880, 0xb440, },
|
|
|
|
[ 2] = { 0x3000, 0xb540, 0x1a88, 0x2f80, 0xb3e0, },
|
|
|
|
[ 3] = { 0x3000, 0xb580, 0x1b30, 0x2e20, 0xb380, },
|
|
|
|
[ 4] = { 0x3000, 0xb5c0, 0x1bd8, 0x2cc0, 0xb320, },
|
|
|
|
[ 5] = { 0x3020, 0xb5e0, 0x1c60, 0x2b80, 0xb2c0, },
|
|
|
|
[ 6] = { 0x3020, 0xb5e0, 0x1cf8, 0x2a20, 0xb260, },
|
|
|
|
[ 7] = { 0x3020, 0xb5e0, 0x1d80, 0x28e0, 0xb200, },
|
|
|
|
[ 8] = { 0x3020, 0xb5c0, 0x1e08, 0x3f40, 0xb1c0, },
|
|
|
|
[ 9] = { 0x3020, 0xb580, 0x1e78, 0x3ce0, 0xb160, },
|
|
|
|
[10] = { 0x3040, 0xb520, 0x1ed8, 0x3aa0, 0xb120, },
|
|
|
|
[11] = { 0x3040, 0xb4a0, 0x1f30, 0x3880, 0xb0e0, },
|
|
|
|
[12] = { 0x3040, 0xb400, 0x1f78, 0x3680, 0xb0a0, },
|
|
|
|
[13] = { 0x3020, 0xb340, 0x1fb8, 0x34a0, 0xb060, },
|
|
|
|
[14] = { 0x3020, 0xb240, 0x1fe0, 0x32e0, 0xb040, },
|
|
|
|
[15] = { 0x3020, 0xb140, 0x1ff8, 0x3160, 0xb020, },
|
|
|
|
[16] = { 0xb000, 0x3000, 0x0800, 0x3000, 0xb000, },
|
2010-08-12 15:28:50 +07:00
|
|
|
};
|
|
|
|
|
2016-12-08 00:28:10 +07:00
|
|
|
static const u16 uv_static_hcoeffs[N_PHASES][N_HORIZ_UV_TAPS] = {
|
|
|
|
[ 0] = { 0x3000, 0x1800, 0x1800, },
|
|
|
|
[ 1] = { 0xb000, 0x18d0, 0x2e60, },
|
|
|
|
[ 2] = { 0xb000, 0x1990, 0x2ce0, },
|
|
|
|
[ 3] = { 0xb020, 0x1a68, 0x2b40, },
|
|
|
|
[ 4] = { 0xb040, 0x1b20, 0x29e0, },
|
|
|
|
[ 5] = { 0xb060, 0x1bd8, 0x2880, },
|
|
|
|
[ 6] = { 0xb080, 0x1c88, 0x3e60, },
|
|
|
|
[ 7] = { 0xb0a0, 0x1d28, 0x3c00, },
|
|
|
|
[ 8] = { 0xb0c0, 0x1db8, 0x39e0, },
|
|
|
|
[ 9] = { 0xb0e0, 0x1e40, 0x37e0, },
|
|
|
|
[10] = { 0xb100, 0x1eb8, 0x3620, },
|
|
|
|
[11] = { 0xb100, 0x1f18, 0x34a0, },
|
|
|
|
[12] = { 0xb100, 0x1f68, 0x3360, },
|
|
|
|
[13] = { 0xb0e0, 0x1fa8, 0x3240, },
|
|
|
|
[14] = { 0xb0c0, 0x1fe0, 0x3140, },
|
|
|
|
[15] = { 0xb060, 0x1ff0, 0x30a0, },
|
|
|
|
[16] = { 0x3000, 0x0800, 0x3000, },
|
2010-08-12 15:28:50 +07:00
|
|
|
};
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2012-04-17 04:07:43 +07:00
|
|
|
static void update_polyphase_filter(struct overlay_registers __iomem *regs)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2012-04-17 04:07:43 +07:00
|
|
|
memcpy_toio(regs->Y_HCOEFS, y_static_hcoeffs, sizeof(y_static_hcoeffs));
|
|
|
|
memcpy_toio(regs->UV_HCOEFS, uv_static_hcoeffs,
|
|
|
|
sizeof(uv_static_hcoeffs));
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static bool update_scaling_factors(struct intel_overlay *overlay,
|
2012-04-17 04:07:43 +07:00
|
|
|
struct overlay_registers __iomem *regs,
|
2018-09-07 02:01:44 +07:00
|
|
|
struct drm_intel_overlay_put_image *params)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
|
|
|
/* fixed point with a 12 bit shift */
|
|
|
|
u32 xscale, yscale, xscale_UV, yscale_UV;
|
|
|
|
#define FP_SHIFT 12
|
|
|
|
#define FRACT_MASK 0xfff
|
|
|
|
bool scale_changed = false;
|
2018-09-07 02:01:44 +07:00
|
|
|
int uv_hscale = uv_hsubsampling(params->flags);
|
|
|
|
int uv_vscale = uv_vsubsampling(params->flags);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
if (params->dst_width > 1)
|
|
|
|
xscale = ((params->src_scan_width - 1) << FP_SHIFT) /
|
|
|
|
params->dst_width;
|
2009-09-16 03:57:34 +07:00
|
|
|
else
|
|
|
|
xscale = 1 << FP_SHIFT;
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
if (params->dst_height > 1)
|
|
|
|
yscale = ((params->src_scan_height - 1) << FP_SHIFT) /
|
|
|
|
params->dst_height;
|
2009-09-16 03:57:34 +07:00
|
|
|
else
|
|
|
|
yscale = 1 << FP_SHIFT;
|
|
|
|
|
|
|
|
/*if (params->format & I915_OVERLAY_YUV_PLANAR) {*/
|
2010-08-12 15:28:50 +07:00
|
|
|
xscale_UV = xscale/uv_hscale;
|
|
|
|
yscale_UV = yscale/uv_vscale;
|
|
|
|
/* make the Y scale to UV scale ratio an exact multiply */
|
|
|
|
xscale = xscale_UV * uv_hscale;
|
|
|
|
yscale = yscale_UV * uv_vscale;
|
2009-09-16 03:57:34 +07:00
|
|
|
/*} else {
|
2010-08-12 15:28:50 +07:00
|
|
|
xscale_UV = 0;
|
|
|
|
yscale_UV = 0;
|
|
|
|
}*/
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
if (xscale != overlay->old_xscale || yscale != overlay->old_yscale)
|
|
|
|
scale_changed = true;
|
|
|
|
overlay->old_xscale = xscale;
|
|
|
|
overlay->old_yscale = yscale;
|
|
|
|
|
2012-04-17 04:07:43 +07:00
|
|
|
iowrite32(((yscale & FRACT_MASK) << 20) |
|
|
|
|
((xscale >> FP_SHIFT) << 16) |
|
|
|
|
((xscale & FRACT_MASK) << 3),
|
|
|
|
®s->YRGBSCALE);
|
2010-08-12 15:28:50 +07:00
|
|
|
|
2012-04-17 04:07:43 +07:00
|
|
|
iowrite32(((yscale_UV & FRACT_MASK) << 20) |
|
|
|
|
((xscale_UV >> FP_SHIFT) << 16) |
|
|
|
|
((xscale_UV & FRACT_MASK) << 3),
|
|
|
|
®s->UVSCALE);
|
2010-08-12 15:28:50 +07:00
|
|
|
|
2012-04-17 04:07:43 +07:00
|
|
|
iowrite32((((yscale >> FP_SHIFT) << 16) |
|
|
|
|
((yscale_UV >> FP_SHIFT) << 0)),
|
|
|
|
®s->UVSCALEV);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
if (scale_changed)
|
|
|
|
update_polyphase_filter(regs);
|
|
|
|
|
|
|
|
return scale_changed;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void update_colorkey(struct intel_overlay *overlay,
|
2012-04-17 04:07:43 +07:00
|
|
|
struct overlay_registers __iomem *regs)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2016-12-08 00:28:11 +07:00
|
|
|
const struct intel_plane_state *state =
|
|
|
|
to_intel_plane_state(overlay->crtc->base.primary->state);
|
2009-09-16 03:57:34 +07:00
|
|
|
u32 key = overlay->color_key;
|
2016-12-08 00:28:11 +07:00
|
|
|
u32 format = 0;
|
|
|
|
u32 flags = 0;
|
2015-04-02 16:35:08 +07:00
|
|
|
|
|
|
|
if (overlay->color_key_enabled)
|
|
|
|
flags |= DST_KEY_ENABLE;
|
2010-08-12 15:30:58 +07:00
|
|
|
|
2016-12-08 00:28:11 +07:00
|
|
|
if (state->base.visible)
|
2017-01-04 17:41:10 +07:00
|
|
|
format = state->base.fb->format->format;
|
2016-12-08 00:28:11 +07:00
|
|
|
|
|
|
|
switch (format) {
|
|
|
|
case DRM_FORMAT_C8:
|
2015-04-02 16:35:08 +07:00
|
|
|
key = 0;
|
|
|
|
flags |= CLK_RGB8I_MASK;
|
2010-08-12 15:30:58 +07:00
|
|
|
break;
|
2016-12-08 00:28:11 +07:00
|
|
|
case DRM_FORMAT_XRGB1555:
|
|
|
|
key = RGB15_TO_COLORKEY(key);
|
|
|
|
flags |= CLK_RGB15_MASK;
|
2010-08-12 15:30:58 +07:00
|
|
|
break;
|
2016-12-08 00:28:11 +07:00
|
|
|
case DRM_FORMAT_RGB565:
|
|
|
|
key = RGB16_TO_COLORKEY(key);
|
|
|
|
flags |= CLK_RGB16_MASK;
|
|
|
|
break;
|
|
|
|
default:
|
2015-04-02 16:35:08 +07:00
|
|
|
flags |= CLK_RGB24_MASK;
|
2010-08-12 15:30:58 +07:00
|
|
|
break;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
2015-04-02 16:35:08 +07:00
|
|
|
|
|
|
|
iowrite32(key, ®s->DCLRKV);
|
|
|
|
iowrite32(flags, ®s->DCLRKM);
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
static u32 overlay_cmd_reg(struct drm_intel_overlay_put_image *params)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
|
|
|
u32 cmd = OCMD_ENABLE | OCMD_BUF_TYPE_FRAME | OCMD_BUFFER0;
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
if (params->flags & I915_OVERLAY_YUV_PLANAR) {
|
|
|
|
switch (params->flags & I915_OVERLAY_DEPTH_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_YUV422:
|
|
|
|
cmd |= OCMD_YUV_422_PLANAR;
|
|
|
|
break;
|
|
|
|
case I915_OVERLAY_YUV420:
|
|
|
|
cmd |= OCMD_YUV_420_PLANAR;
|
|
|
|
break;
|
|
|
|
case I915_OVERLAY_YUV411:
|
|
|
|
case I915_OVERLAY_YUV410:
|
|
|
|
cmd |= OCMD_YUV_410_PLANAR;
|
|
|
|
break;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
} else { /* YUV packed */
|
2018-09-07 02:01:44 +07:00
|
|
|
switch (params->flags & I915_OVERLAY_DEPTH_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_YUV422:
|
|
|
|
cmd |= OCMD_YUV_422_PACKED;
|
|
|
|
break;
|
|
|
|
case I915_OVERLAY_YUV411:
|
|
|
|
cmd |= OCMD_YUV_411_PACKED;
|
|
|
|
break;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
switch (params->flags & I915_OVERLAY_SWAP_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_NO_SWAP:
|
|
|
|
break;
|
|
|
|
case I915_OVERLAY_UV_SWAP:
|
|
|
|
cmd |= OCMD_UV_SWAP;
|
|
|
|
break;
|
|
|
|
case I915_OVERLAY_Y_SWAP:
|
|
|
|
cmd |= OCMD_Y_SWAP;
|
|
|
|
break;
|
|
|
|
case I915_OVERLAY_Y_AND_UV_SWAP:
|
|
|
|
cmd |= OCMD_Y_AND_UV_SWAP;
|
|
|
|
break;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return cmd;
|
|
|
|
}
|
|
|
|
|
2010-08-12 18:38:21 +07:00
|
|
|
static int intel_overlay_do_put_image(struct intel_overlay *overlay,
|
2010-11-09 02:18:58 +07:00
|
|
|
struct drm_i915_gem_object *new_bo,
|
2018-09-07 02:01:44 +07:00
|
|
|
struct drm_intel_overlay_put_image *params)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2018-09-07 02:01:43 +07:00
|
|
|
struct overlay_registers __iomem *regs = overlay->regs;
|
2016-05-12 18:43:23 +07:00
|
|
|
struct drm_i915_private *dev_priv = overlay->i915;
|
2012-04-17 04:07:43 +07:00
|
|
|
u32 swidth, swidthsw, sheight, ostride;
|
2014-06-19 04:28:09 +07:00
|
|
|
enum pipe pipe = overlay->crtc->pipe;
|
2018-09-07 02:01:43 +07:00
|
|
|
bool scale_changed = false;
|
2016-08-15 16:49:01 +07:00
|
|
|
struct i915_vma *vma;
|
2018-09-07 02:01:43 +07:00
|
|
|
int ret, tmp_width;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2016-07-05 16:40:23 +07:00
|
|
|
lockdep_assert_held(&dev_priv->drm.struct_mutex);
|
|
|
|
WARN_ON(!drm_modeset_is_locked(&dev_priv->drm.mode_config.connection_mutex));
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
ret = intel_overlay_release_old_vid(overlay);
|
|
|
|
if (ret != 0)
|
|
|
|
return ret;
|
|
|
|
|
drm/i915: More surgically unbreak the modeset vs reset deadlock
There's no reason to entirely wedge the gpu, for the minimal deadlock
bugfix we only need to unbreak/decouple the atomic commit from the gpu
reset. The simplest way to fix that is by replacing the
unconditional fence wait a the top of commit_tail by a wait which
completes either when the fences are done (normal case, or when a
reset doesn't need to touch the display state). Or when the gpu reset
needs to force-unblock all pending modeset states.
The lesser source of deadlocks is when we try to pin a new framebuffer
and run into a stall. There's a bunch of places this can happen, like
eviction, changing the caching mode, acquiring a fence on older
platforms. And we can't just break the depency loop and keep going,
the only way would be to break out and restart. But the problem with
that approach is that we must stall for the reset to complete before
we grab any locks, and with the atomic infrastructure that's a bit
tricky. The only place is the ioctl code, and we don't want to insert
code into e.g. the BUSY ioctl. Hence for that problem just create a
critical section, and if any code is in there, wedge the GPU. For the
steady-state this should never be a problem.
Note that in both cases TDR itself keeps working, so from a userspace
pov this trickery isn't observable. Users themselvs might spot a short
glitch while the rendering is catching up again, but that's still
better than pre-TDR where we've thrown away all the rendering,
including innocent batches. Also, this fixes the regression TDR
introduced of making gpu resets deadlock-prone when we do need to
touch the display.
One thing I noticed is that gpu_error.flags seems to use both our own
wait-queue in gpu_error.wait_queue, and the generic wait_on_bit
facilities. Not entirely sure why this inconsistency exists, I just
picked one style.
A possible future avenue could be to insert the gpu reset in-between
ongoing modeset changes, which would avoid the momentary glitch. But
that's a lot more work to implement in the atomic commit machinery,
and given that we only need this for pre-g4x hw, of questionable
utility just for the sake of polishing gpu reset even more on those
old boxes. It might be useful for other features though.
v2: Rebase onto 4.13 with a s/wait_queue_t/struct wait_queue_entry/.
v3: Really emabarrassing fixup, I checked the wrong bit and broke the
unbreak/wakeup logic.
v4: Also handle deadlocks in pin_to_display.
v5: Review from Michel:
- Fixup the BUILD_BUG_ON
- Don't forget about the overlay
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> (v2)
Cc: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170808080828.23650-3-daniel.vetter@ffwll.ch
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
2017-08-08 15:08:28 +07:00
|
|
|
atomic_inc(&dev_priv->gpu_error.pending_fb_pin);
|
|
|
|
|
2018-02-20 20:42:06 +07:00
|
|
|
vma = i915_gem_object_pin_to_display_plane(new_bo,
|
|
|
|
0, NULL, PIN_MAPPABLE);
|
drm/i915: More surgically unbreak the modeset vs reset deadlock
There's no reason to entirely wedge the gpu, for the minimal deadlock
bugfix we only need to unbreak/decouple the atomic commit from the gpu
reset. The simplest way to fix that is by replacing the
unconditional fence wait a the top of commit_tail by a wait which
completes either when the fences are done (normal case, or when a
reset doesn't need to touch the display state). Or when the gpu reset
needs to force-unblock all pending modeset states.
The lesser source of deadlocks is when we try to pin a new framebuffer
and run into a stall. There's a bunch of places this can happen, like
eviction, changing the caching mode, acquiring a fence on older
platforms. And we can't just break the depency loop and keep going,
the only way would be to break out and restart. But the problem with
that approach is that we must stall for the reset to complete before
we grab any locks, and with the atomic infrastructure that's a bit
tricky. The only place is the ioctl code, and we don't want to insert
code into e.g. the BUSY ioctl. Hence for that problem just create a
critical section, and if any code is in there, wedge the GPU. For the
steady-state this should never be a problem.
Note that in both cases TDR itself keeps working, so from a userspace
pov this trickery isn't observable. Users themselvs might spot a short
glitch while the rendering is catching up again, but that's still
better than pre-TDR where we've thrown away all the rendering,
including innocent batches. Also, this fixes the regression TDR
introduced of making gpu resets deadlock-prone when we do need to
touch the display.
One thing I noticed is that gpu_error.flags seems to use both our own
wait-queue in gpu_error.wait_queue, and the generic wait_on_bit
facilities. Not entirely sure why this inconsistency exists, I just
picked one style.
A possible future avenue could be to insert the gpu reset in-between
ongoing modeset changes, which would avoid the momentary glitch. But
that's a lot more work to implement in the atomic commit machinery,
and given that we only need this for pre-g4x hw, of questionable
utility just for the sake of polishing gpu reset even more on those
old boxes. It might be useful for other features though.
v2: Rebase onto 4.13 with a s/wait_queue_t/struct wait_queue_entry/.
v3: Really emabarrassing fixup, I checked the wrong bit and broke the
unbreak/wakeup logic.
v4: Also handle deadlocks in pin_to_display.
v5: Review from Michel:
- Fixup the BUILD_BUG_ON
- Don't forget about the overlay
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> (v2)
Cc: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170808080828.23650-3-daniel.vetter@ffwll.ch
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
2017-08-08 15:08:28 +07:00
|
|
|
if (IS_ERR(vma)) {
|
|
|
|
ret = PTR_ERR(vma);
|
|
|
|
goto out_pin_section;
|
|
|
|
}
|
2018-03-07 10:34:18 +07:00
|
|
|
intel_fb_obj_flush(new_bo, ORIGIN_DIRTYFB);
|
2016-08-15 16:49:01 +07:00
|
|
|
|
2016-08-18 23:17:00 +07:00
|
|
|
ret = i915_vma_put_fence(vma);
|
2010-11-10 23:40:20 +07:00
|
|
|
if (ret)
|
|
|
|
goto out_unpin;
|
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
if (!overlay->active) {
|
2012-04-17 04:07:43 +07:00
|
|
|
u32 oconfig;
|
2018-09-07 02:01:43 +07:00
|
|
|
|
2012-04-17 04:07:43 +07:00
|
|
|
oconfig = OCONF_CC_OUT_8BIT;
|
drm/i915: replace IS_GEN<N> with IS_GEN(..., N)
Define IS_GEN() similarly to our IS_GEN_RANGE(). but use gen instead of
gen_mask to do the comparison. Now callers can pass then gen as a parameter,
so we don't require one macro for each gen.
The following spatch was used to convert the users of these macros:
@@
expression e;
@@
(
- IS_GEN2(e)
+ IS_GEN(e, 2)
|
- IS_GEN3(e)
+ IS_GEN(e, 3)
|
- IS_GEN4(e)
+ IS_GEN(e, 4)
|
- IS_GEN5(e)
+ IS_GEN(e, 5)
|
- IS_GEN6(e)
+ IS_GEN(e, 6)
|
- IS_GEN7(e)
+ IS_GEN(e, 7)
|
- IS_GEN8(e)
+ IS_GEN(e, 8)
|
- IS_GEN9(e)
+ IS_GEN(e, 9)
|
- IS_GEN10(e)
+ IS_GEN(e, 10)
|
- IS_GEN11(e)
+ IS_GEN(e, 11)
)
v2: use IS_GEN rather than GT_GEN and compare to info.gen rather than
using the bitmask
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181212181044.15886-2-lucas.demarchi@intel.com
2018-12-13 01:10:43 +07:00
|
|
|
if (IS_GEN(dev_priv, 4))
|
2012-04-17 04:07:43 +07:00
|
|
|
oconfig |= OCONF_CSC_MODE_BT709;
|
2014-06-19 04:28:09 +07:00
|
|
|
oconfig |= pipe == 0 ?
|
2009-09-16 03:57:34 +07:00
|
|
|
OCONF_PIPE_A : OCONF_PIPE_B;
|
2012-04-17 04:07:43 +07:00
|
|
|
iowrite32(oconfig, ®s->OCONFIG);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
ret = intel_overlay_on(overlay);
|
|
|
|
if (ret != 0)
|
|
|
|
goto out_unpin;
|
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
iowrite32(params->dst_y << 16 | params->dst_x, ®s->DWINPOS);
|
|
|
|
iowrite32(params->dst_height << 16 | params->dst_width, ®s->DWINSZ);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
if (params->flags & I915_OVERLAY_YUV_PACKED)
|
|
|
|
tmp_width = packed_width_bytes(params->flags,
|
|
|
|
params->src_width);
|
2009-09-16 03:57:34 +07:00
|
|
|
else
|
2018-09-07 02:01:44 +07:00
|
|
|
tmp_width = params->src_width;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
swidth = params->src_width;
|
2016-05-12 18:43:23 +07:00
|
|
|
swidthsw = calc_swidthsw(dev_priv, params->offset_Y, tmp_width);
|
2018-09-07 02:01:44 +07:00
|
|
|
sheight = params->src_height;
|
2016-08-15 16:49:07 +07:00
|
|
|
iowrite32(i915_ggtt_offset(vma) + params->offset_Y, ®s->OBUF_0Y);
|
2012-04-17 04:07:43 +07:00
|
|
|
ostride = params->stride_Y;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
if (params->flags & I915_OVERLAY_YUV_PLANAR) {
|
|
|
|
int uv_hscale = uv_hsubsampling(params->flags);
|
|
|
|
int uv_vscale = uv_vsubsampling(params->flags);
|
2009-09-16 03:57:34 +07:00
|
|
|
u32 tmp_U, tmp_V;
|
2018-09-07 02:01:44 +07:00
|
|
|
|
|
|
|
swidth |= (params->src_width / uv_hscale) << 16;
|
|
|
|
sheight |= (params->src_height / uv_vscale) << 16;
|
|
|
|
|
2016-05-12 18:43:23 +07:00
|
|
|
tmp_U = calc_swidthsw(dev_priv, params->offset_U,
|
2018-09-07 02:01:44 +07:00
|
|
|
params->src_width / uv_hscale);
|
2016-05-12 18:43:23 +07:00
|
|
|
tmp_V = calc_swidthsw(dev_priv, params->offset_V,
|
2018-09-07 02:01:44 +07:00
|
|
|
params->src_width / uv_hscale);
|
|
|
|
swidthsw |= max(tmp_U, tmp_V) << 16;
|
|
|
|
|
2016-08-15 16:49:07 +07:00
|
|
|
iowrite32(i915_ggtt_offset(vma) + params->offset_U,
|
|
|
|
®s->OBUF_0U);
|
|
|
|
iowrite32(i915_ggtt_offset(vma) + params->offset_V,
|
|
|
|
®s->OBUF_0V);
|
2018-09-07 02:01:44 +07:00
|
|
|
|
2012-04-17 04:07:43 +07:00
|
|
|
ostride |= params->stride_UV << 16;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
2012-04-17 04:07:43 +07:00
|
|
|
iowrite32(swidth, ®s->SWIDTH);
|
|
|
|
iowrite32(swidthsw, ®s->SWIDTHSW);
|
|
|
|
iowrite32(sheight, ®s->SHEIGHT);
|
|
|
|
iowrite32(ostride, ®s->OSTRIDE);
|
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
scale_changed = update_scaling_factors(overlay, regs, params);
|
|
|
|
|
|
|
|
update_colorkey(overlay, regs);
|
|
|
|
|
2012-04-17 04:07:43 +07:00
|
|
|
iowrite32(overlay_cmd_reg(params), ®s->OCMD);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2016-12-08 00:28:06 +07:00
|
|
|
ret = intel_overlay_continue(overlay, vma, scale_changed);
|
2010-08-12 18:36:12 +07:00
|
|
|
if (ret)
|
|
|
|
goto out_unpin;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
out_unpin:
|
2016-08-15 16:49:06 +07:00
|
|
|
i915_gem_object_unpin_from_display_plane(vma);
|
drm/i915: More surgically unbreak the modeset vs reset deadlock
There's no reason to entirely wedge the gpu, for the minimal deadlock
bugfix we only need to unbreak/decouple the atomic commit from the gpu
reset. The simplest way to fix that is by replacing the
unconditional fence wait a the top of commit_tail by a wait which
completes either when the fences are done (normal case, or when a
reset doesn't need to touch the display state). Or when the gpu reset
needs to force-unblock all pending modeset states.
The lesser source of deadlocks is when we try to pin a new framebuffer
and run into a stall. There's a bunch of places this can happen, like
eviction, changing the caching mode, acquiring a fence on older
platforms. And we can't just break the depency loop and keep going,
the only way would be to break out and restart. But the problem with
that approach is that we must stall for the reset to complete before
we grab any locks, and with the atomic infrastructure that's a bit
tricky. The only place is the ioctl code, and we don't want to insert
code into e.g. the BUSY ioctl. Hence for that problem just create a
critical section, and if any code is in there, wedge the GPU. For the
steady-state this should never be a problem.
Note that in both cases TDR itself keeps working, so from a userspace
pov this trickery isn't observable. Users themselvs might spot a short
glitch while the rendering is catching up again, but that's still
better than pre-TDR where we've thrown away all the rendering,
including innocent batches. Also, this fixes the regression TDR
introduced of making gpu resets deadlock-prone when we do need to
touch the display.
One thing I noticed is that gpu_error.flags seems to use both our own
wait-queue in gpu_error.wait_queue, and the generic wait_on_bit
facilities. Not entirely sure why this inconsistency exists, I just
picked one style.
A possible future avenue could be to insert the gpu reset in-between
ongoing modeset changes, which would avoid the momentary glitch. But
that's a lot more work to implement in the atomic commit machinery,
and given that we only need this for pre-g4x hw, of questionable
utility just for the sake of polishing gpu reset even more on those
old boxes. It might be useful for other features though.
v2: Rebase onto 4.13 with a s/wait_queue_t/struct wait_queue_entry/.
v3: Really emabarrassing fixup, I checked the wrong bit and broke the
unbreak/wakeup logic.
v4: Also handle deadlocks in pin_to_display.
v5: Review from Michel:
- Fixup the BUILD_BUG_ON
- Don't forget about the overlay
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> (v2)
Cc: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170808080828.23650-3-daniel.vetter@ffwll.ch
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
2017-08-08 15:08:28 +07:00
|
|
|
out_pin_section:
|
|
|
|
atomic_dec(&dev_priv->gpu_error.pending_fb_pin);
|
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2011-02-21 21:43:56 +07:00
|
|
|
int intel_overlay_switch_off(struct intel_overlay *overlay)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2016-05-12 18:43:23 +07:00
|
|
|
struct drm_i915_private *dev_priv = overlay->i915;
|
2010-08-12 19:50:28 +07:00
|
|
|
int ret;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2016-07-05 16:40:23 +07:00
|
|
|
lockdep_assert_held(&dev_priv->drm.struct_mutex);
|
|
|
|
WARN_ON(!drm_modeset_is_locked(&dev_priv->drm.mode_config.connection_mutex));
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2011-02-21 21:43:56 +07:00
|
|
|
ret = intel_overlay_recover_from_interrupt(overlay);
|
2010-08-12 20:03:48 +07:00
|
|
|
if (ret != 0)
|
|
|
|
return ret;
|
2009-11-30 21:55:49 +07:00
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
if (!overlay->active)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
ret = intel_overlay_release_old_vid(overlay);
|
|
|
|
if (ret != 0)
|
|
|
|
return ret;
|
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
iowrite32(0, &overlay->regs->OCMD);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2016-08-04 13:52:37 +07:00
|
|
|
return intel_overlay_off(overlay);
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static int check_overlay_possible_on_crtc(struct intel_overlay *overlay,
|
|
|
|
struct intel_crtc *crtc)
|
|
|
|
{
|
2010-09-13 20:19:16 +07:00
|
|
|
if (!crtc->active)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
/* can't use the overlay with double wide pipe */
|
2015-01-15 19:55:25 +07:00
|
|
|
if (crtc->config->double_wide)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void update_pfit_vscale_ratio(struct intel_overlay *overlay)
|
|
|
|
{
|
2016-05-12 18:43:23 +07:00
|
|
|
struct drm_i915_private *dev_priv = overlay->i915;
|
2009-09-16 03:57:34 +07:00
|
|
|
u32 pfit_control = I915_READ(PFIT_CONTROL);
|
2010-08-12 17:15:58 +07:00
|
|
|
u32 ratio;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
/* XXX: This is not the same logic as in the xorg driver, but more in
|
2010-08-12 17:15:58 +07:00
|
|
|
* line with the intel documentation for the i965
|
|
|
|
*/
|
2016-05-12 18:43:23 +07:00
|
|
|
if (INTEL_GEN(dev_priv) >= 4) {
|
2011-08-17 02:34:10 +07:00
|
|
|
/* on i965 use the PGM reg to read out the autoscaler values */
|
2010-09-17 06:32:17 +07:00
|
|
|
ratio = I915_READ(PFIT_PGM_RATIOS) >> PFIT_VERT_SCALE_SHIFT_965;
|
|
|
|
} else {
|
2010-08-12 17:15:58 +07:00
|
|
|
if (pfit_control & VERT_AUTO_SCALE)
|
|
|
|
ratio = I915_READ(PFIT_AUTO_RATIOS);
|
2009-09-16 03:57:34 +07:00
|
|
|
else
|
2010-08-12 17:15:58 +07:00
|
|
|
ratio = I915_READ(PFIT_PGM_RATIOS);
|
|
|
|
ratio >>= PFIT_VERT_SCALE_SHIFT;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
overlay->pfit_vscale_ratio = ratio;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int check_overlay_dst(struct intel_overlay *overlay,
|
|
|
|
struct drm_intel_overlay_put_image *rec)
|
|
|
|
{
|
2016-12-08 00:28:07 +07:00
|
|
|
const struct intel_crtc_state *pipe_config =
|
|
|
|
overlay->crtc->config;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2016-12-08 00:28:07 +07:00
|
|
|
if (rec->dst_x < pipe_config->pipe_src_w &&
|
|
|
|
rec->dst_x + rec->dst_width <= pipe_config->pipe_src_w &&
|
|
|
|
rec->dst_y < pipe_config->pipe_src_h &&
|
|
|
|
rec->dst_y + rec->dst_height <= pipe_config->pipe_src_h)
|
2009-09-16 03:57:34 +07:00
|
|
|
return 0;
|
|
|
|
else
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
static int check_overlay_scaling(struct drm_intel_overlay_put_image *rec)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
|
|
|
u32 tmp;
|
|
|
|
|
|
|
|
/* downscaling limit is 8.0 */
|
2018-09-07 02:01:44 +07:00
|
|
|
tmp = ((rec->src_scan_height << 16) / rec->dst_height) >> 16;
|
2009-09-16 03:57:34 +07:00
|
|
|
if (tmp > 7)
|
|
|
|
return -EINVAL;
|
2018-09-07 02:01:44 +07:00
|
|
|
|
|
|
|
tmp = ((rec->src_scan_width << 16) / rec->dst_width) >> 16;
|
2009-09-16 03:57:34 +07:00
|
|
|
if (tmp > 7)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-05-12 18:43:23 +07:00
|
|
|
static int check_overlay_src(struct drm_i915_private *dev_priv,
|
2009-09-16 03:57:34 +07:00
|
|
|
struct drm_intel_overlay_put_image *rec,
|
2010-11-09 02:18:58 +07:00
|
|
|
struct drm_i915_gem_object *new_bo)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
|
|
|
int uv_hscale = uv_hsubsampling(rec->flags);
|
|
|
|
int uv_vscale = uv_vsubsampling(rec->flags);
|
2010-10-28 04:17:25 +07:00
|
|
|
u32 stride_mask;
|
|
|
|
int depth;
|
|
|
|
u32 tmp;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
/* check src dimensions */
|
2016-11-30 22:43:04 +07:00
|
|
|
if (IS_I845G(dev_priv) || IS_I830(dev_priv)) {
|
2010-08-12 15:28:50 +07:00
|
|
|
if (rec->src_height > IMAGE_MAX_HEIGHT_LEGACY ||
|
2010-08-12 17:29:34 +07:00
|
|
|
rec->src_width > IMAGE_MAX_WIDTH_LEGACY)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
|
|
|
} else {
|
2010-08-12 15:28:50 +07:00
|
|
|
if (rec->src_height > IMAGE_MAX_HEIGHT ||
|
2010-08-12 17:29:34 +07:00
|
|
|
rec->src_width > IMAGE_MAX_WIDTH)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
|
|
|
}
|
2010-08-12 17:29:34 +07:00
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
/* better safe than sorry, use 4 as the maximal subsampling ratio */
|
2010-08-12 15:28:50 +07:00
|
|
|
if (rec->src_height < N_VERT_Y_TAPS*4 ||
|
2010-08-12 17:29:34 +07:00
|
|
|
rec->src_width < N_HORIZ_Y_TAPS*4)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
2010-07-13 01:35:38 +07:00
|
|
|
/* check alignment constraints */
|
2009-09-16 03:57:34 +07:00
|
|
|
switch (rec->flags & I915_OVERLAY_TYPE_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_RGB:
|
|
|
|
/* not implemented */
|
|
|
|
return -EINVAL;
|
2010-08-12 17:29:34 +07:00
|
|
|
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_YUV_PACKED:
|
|
|
|
if (uv_vscale != 1)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
2010-08-12 17:29:34 +07:00
|
|
|
|
|
|
|
depth = packed_depth_bytes(rec->flags);
|
2010-08-12 15:28:50 +07:00
|
|
|
if (depth < 0)
|
|
|
|
return depth;
|
2010-08-12 17:29:34 +07:00
|
|
|
|
2010-08-12 15:28:50 +07:00
|
|
|
/* ignore UV planes */
|
|
|
|
rec->stride_UV = 0;
|
|
|
|
rec->offset_U = 0;
|
|
|
|
rec->offset_V = 0;
|
|
|
|
/* check pixel alignment */
|
|
|
|
if (rec->offset_Y % depth)
|
|
|
|
return -EINVAL;
|
|
|
|
break;
|
2010-08-12 17:29:34 +07:00
|
|
|
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_YUV_PLANAR:
|
|
|
|
if (uv_vscale < 0 || uv_hscale < 0)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
2010-08-12 15:28:50 +07:00
|
|
|
/* no offset restrictions for planar formats */
|
|
|
|
break;
|
2010-08-12 17:29:34 +07:00
|
|
|
|
2010-08-12 15:28:50 +07:00
|
|
|
default:
|
|
|
|
return -EINVAL;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
if (rec->src_width % uv_hscale)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
/* stride checking */
|
2016-11-30 22:43:04 +07:00
|
|
|
if (IS_I830(dev_priv) || IS_I845G(dev_priv))
|
2010-07-13 01:35:38 +07:00
|
|
|
stride_mask = 255;
|
|
|
|
else
|
|
|
|
stride_mask = 63;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
if (rec->stride_Y & stride_mask || rec->stride_UV & stride_mask)
|
|
|
|
return -EINVAL;
|
drm/i915: replace IS_GEN<N> with IS_GEN(..., N)
Define IS_GEN() similarly to our IS_GEN_RANGE(). but use gen instead of
gen_mask to do the comparison. Now callers can pass then gen as a parameter,
so we don't require one macro for each gen.
The following spatch was used to convert the users of these macros:
@@
expression e;
@@
(
- IS_GEN2(e)
+ IS_GEN(e, 2)
|
- IS_GEN3(e)
+ IS_GEN(e, 3)
|
- IS_GEN4(e)
+ IS_GEN(e, 4)
|
- IS_GEN5(e)
+ IS_GEN(e, 5)
|
- IS_GEN6(e)
+ IS_GEN(e, 6)
|
- IS_GEN7(e)
+ IS_GEN(e, 7)
|
- IS_GEN8(e)
+ IS_GEN(e, 8)
|
- IS_GEN9(e)
+ IS_GEN(e, 9)
|
- IS_GEN10(e)
+ IS_GEN(e, 10)
|
- IS_GEN11(e)
+ IS_GEN(e, 11)
)
v2: use IS_GEN rather than GT_GEN and compare to info.gen rather than
using the bitmask
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181212181044.15886-2-lucas.demarchi@intel.com
2018-12-13 01:10:43 +07:00
|
|
|
if (IS_GEN(dev_priv, 4) && rec->stride_Y < 512)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
tmp = (rec->flags & I915_OVERLAY_TYPE_MASK) == I915_OVERLAY_YUV_PLANAR ?
|
2010-08-12 17:29:34 +07:00
|
|
|
4096 : 8192;
|
|
|
|
if (rec->stride_Y > tmp || rec->stride_UV > 2*1024)
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
/* check buffer dimensions */
|
|
|
|
switch (rec->flags & I915_OVERLAY_TYPE_MASK) {
|
2010-08-12 15:28:50 +07:00
|
|
|
case I915_OVERLAY_RGB:
|
|
|
|
case I915_OVERLAY_YUV_PACKED:
|
|
|
|
/* always 4 Y values per depth pixels */
|
|
|
|
if (packed_width_bytes(rec->flags, rec->src_width) > rec->stride_Y)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
tmp = rec->stride_Y*rec->src_height;
|
2010-11-09 02:18:58 +07:00
|
|
|
if (rec->offset_Y + tmp > new_bo->base.size)
|
2010-08-12 15:28:50 +07:00
|
|
|
return -EINVAL;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case I915_OVERLAY_YUV_PLANAR:
|
|
|
|
if (rec->src_width > rec->stride_Y)
|
|
|
|
return -EINVAL;
|
|
|
|
if (rec->src_width/uv_hscale > rec->stride_UV)
|
|
|
|
return -EINVAL;
|
|
|
|
|
2010-08-12 17:29:34 +07:00
|
|
|
tmp = rec->stride_Y * rec->src_height;
|
2010-11-09 02:18:58 +07:00
|
|
|
if (rec->offset_Y + tmp > new_bo->base.size)
|
2010-08-12 15:28:50 +07:00
|
|
|
return -EINVAL;
|
2010-08-12 17:29:34 +07:00
|
|
|
|
|
|
|
tmp = rec->stride_UV * (rec->src_height / uv_vscale);
|
2010-11-09 02:18:58 +07:00
|
|
|
if (rec->offset_U + tmp > new_bo->base.size ||
|
|
|
|
rec->offset_V + tmp > new_bo->base.size)
|
2010-08-12 15:28:50 +07:00
|
|
|
return -EINVAL;
|
|
|
|
break;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-05-12 18:43:23 +07:00
|
|
|
int intel_overlay_put_image_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2018-09-07 02:01:44 +07:00
|
|
|
struct drm_intel_overlay_put_image *params = data;
|
2016-07-04 17:34:36 +07:00
|
|
|
struct drm_i915_private *dev_priv = to_i915(dev);
|
2009-09-16 03:57:34 +07:00
|
|
|
struct intel_overlay *overlay;
|
2014-07-18 10:30:04 +07:00
|
|
|
struct drm_crtc *drmmode_crtc;
|
2009-09-16 03:57:34 +07:00
|
|
|
struct intel_crtc *crtc;
|
2010-11-09 02:18:58 +07:00
|
|
|
struct drm_i915_gem_object *new_bo;
|
2009-09-16 03:57:34 +07:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
overlay = dev_priv->overlay;
|
|
|
|
if (!overlay) {
|
|
|
|
DRM_DEBUG("userspace bug: no overlay\n");
|
|
|
|
return -ENODEV;
|
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
if (!(params->flags & I915_OVERLAY_ENABLE)) {
|
2012-12-02 07:05:46 +07:00
|
|
|
drm_modeset_lock_all(dev);
|
2009-09-16 03:57:34 +07:00
|
|
|
mutex_lock(&dev->struct_mutex);
|
|
|
|
|
2011-02-21 21:43:56 +07:00
|
|
|
ret = intel_overlay_switch_off(overlay);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
mutex_unlock(&dev->struct_mutex);
|
2012-12-02 07:05:46 +07:00
|
|
|
drm_modeset_unlock_all(dev);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
drmmode_crtc = drm_crtc_find(dev, file_priv, params->crtc_id);
|
|
|
|
if (!drmmode_crtc)
|
|
|
|
return -ENOENT;
|
2014-07-18 10:30:04 +07:00
|
|
|
crtc = to_intel_crtc(drmmode_crtc);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
new_bo = i915_gem_object_lookup(file_priv, params->bo_handle);
|
|
|
|
if (!new_bo)
|
|
|
|
return -ENOENT;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2012-12-02 07:05:46 +07:00
|
|
|
drm_modeset_lock_all(dev);
|
2009-09-16 03:57:34 +07:00
|
|
|
mutex_lock(&dev->struct_mutex);
|
|
|
|
|
2016-08-05 16:14:23 +07:00
|
|
|
if (i915_gem_object_is_tiled(new_bo)) {
|
2014-02-14 20:06:06 +07:00
|
|
|
DRM_DEBUG_KMS("buffer used for overlay image can not be tiled\n");
|
2010-11-10 23:40:20 +07:00
|
|
|
ret = -EINVAL;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
|
2011-02-21 21:43:56 +07:00
|
|
|
ret = intel_overlay_recover_from_interrupt(overlay);
|
2010-08-12 20:03:48 +07:00
|
|
|
if (ret != 0)
|
|
|
|
goto out_unlock;
|
2009-09-16 03:57:37 +07:00
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
if (overlay->crtc != crtc) {
|
2011-02-21 21:43:56 +07:00
|
|
|
ret = intel_overlay_switch_off(overlay);
|
2009-09-16 03:57:34 +07:00
|
|
|
if (ret != 0)
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
ret = check_overlay_possible_on_crtc(overlay, crtc);
|
|
|
|
if (ret != 0)
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
overlay->crtc = crtc;
|
|
|
|
crtc->overlay = overlay;
|
|
|
|
|
2010-09-13 07:16:10 +07:00
|
|
|
/* line too wide, i.e. one-line-mode */
|
2016-12-08 00:28:07 +07:00
|
|
|
if (crtc->config->pipe_src_w > 1024 &&
|
2016-12-08 00:28:08 +07:00
|
|
|
crtc->config->gmch_pfit.control & PFIT_ENABLE) {
|
2015-03-31 14:37:23 +07:00
|
|
|
overlay->pfit_active = true;
|
2009-09-16 03:57:34 +07:00
|
|
|
update_pfit_vscale_ratio(overlay);
|
|
|
|
} else
|
2015-03-31 14:37:23 +07:00
|
|
|
overlay->pfit_active = false;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
ret = check_overlay_dst(overlay, params);
|
2009-09-16 03:57:34 +07:00
|
|
|
if (ret != 0)
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
if (overlay->pfit_active) {
|
2018-09-07 02:01:44 +07:00
|
|
|
params->dst_y = (((u32)params->dst_y << 12) /
|
2010-08-12 15:28:50 +07:00
|
|
|
overlay->pfit_vscale_ratio);
|
2009-09-16 03:57:34 +07:00
|
|
|
/* shifting right rounds downwards, so add 1 */
|
2018-09-07 02:01:44 +07:00
|
|
|
params->dst_height = (((u32)params->dst_height << 12) /
|
2010-08-12 15:28:50 +07:00
|
|
|
overlay->pfit_vscale_ratio) + 1;
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
2018-09-07 02:01:44 +07:00
|
|
|
|
|
|
|
if (params->src_scan_height > params->src_height ||
|
|
|
|
params->src_scan_width > params->src_width) {
|
2009-09-16 03:57:34 +07:00
|
|
|
ret = -EINVAL;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:44 +07:00
|
|
|
ret = check_overlay_src(dev_priv, params, new_bo);
|
2009-09-16 03:57:34 +07:00
|
|
|
if (ret != 0)
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
/* Check scaling after src size to prevent a divide-by-zero. */
|
|
|
|
ret = check_overlay_scaling(params);
|
|
|
|
if (ret != 0)
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
ret = intel_overlay_do_put_image(overlay, new_bo, params);
|
|
|
|
if (ret != 0)
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
mutex_unlock(&dev->struct_mutex);
|
2012-12-02 07:05:46 +07:00
|
|
|
drm_modeset_unlock_all(dev);
|
2016-12-08 00:28:06 +07:00
|
|
|
i915_gem_object_put(new_bo);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
out_unlock:
|
|
|
|
mutex_unlock(&dev->struct_mutex);
|
2012-12-02 07:05:46 +07:00
|
|
|
drm_modeset_unlock_all(dev);
|
2016-10-28 19:58:43 +07:00
|
|
|
i915_gem_object_put(new_bo);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void update_reg_attrs(struct intel_overlay *overlay,
|
2012-04-17 04:07:43 +07:00
|
|
|
struct overlay_registers __iomem *regs)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2012-04-17 04:07:43 +07:00
|
|
|
iowrite32((overlay->contrast << 18) | (overlay->brightness & 0xff),
|
|
|
|
®s->OCLRC0);
|
|
|
|
iowrite32(overlay->saturation, ®s->OCLRC1);
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static bool check_gamma_bounds(u32 gamma1, u32 gamma2)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
if (gamma1 & 0xff000000 || gamma2 & 0xff000000)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
for (i = 0; i < 3; i++) {
|
2010-08-12 15:28:50 +07:00
|
|
|
if (((gamma1 >> i*8) & 0xff) >= ((gamma2 >> i*8) & 0xff))
|
2009-09-16 03:57:34 +07:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool check_gamma5_errata(u32 gamma5)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < 3; i++) {
|
|
|
|
if (((gamma5 >> i*8) & 0xff) == 0x80)
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int check_gamma(struct drm_intel_overlay_attrs *attrs)
|
|
|
|
{
|
2010-08-12 15:28:50 +07:00
|
|
|
if (!check_gamma_bounds(0, attrs->gamma0) ||
|
|
|
|
!check_gamma_bounds(attrs->gamma0, attrs->gamma1) ||
|
|
|
|
!check_gamma_bounds(attrs->gamma1, attrs->gamma2) ||
|
|
|
|
!check_gamma_bounds(attrs->gamma2, attrs->gamma3) ||
|
|
|
|
!check_gamma_bounds(attrs->gamma3, attrs->gamma4) ||
|
|
|
|
!check_gamma_bounds(attrs->gamma4, attrs->gamma5) ||
|
|
|
|
!check_gamma_bounds(attrs->gamma5, 0x00ffffff))
|
2009-09-16 03:57:34 +07:00
|
|
|
return -EINVAL;
|
2010-08-12 15:28:50 +07:00
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
if (!check_gamma5_errata(attrs->gamma5))
|
|
|
|
return -EINVAL;
|
2010-08-12 15:28:50 +07:00
|
|
|
|
2009-09-16 03:57:34 +07:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-05-12 18:43:23 +07:00
|
|
|
int intel_overlay_attrs_ioctl(struct drm_device *dev, void *data,
|
|
|
|
struct drm_file *file_priv)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
|
|
|
struct drm_intel_overlay_attrs *attrs = data;
|
2016-07-04 17:34:36 +07:00
|
|
|
struct drm_i915_private *dev_priv = to_i915(dev);
|
2009-09-16 03:57:34 +07:00
|
|
|
struct intel_overlay *overlay;
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
overlay = dev_priv->overlay;
|
|
|
|
if (!overlay) {
|
|
|
|
DRM_DEBUG("userspace bug: no overlay\n");
|
|
|
|
return -ENODEV;
|
|
|
|
}
|
|
|
|
|
2012-12-02 07:05:46 +07:00
|
|
|
drm_modeset_lock_all(dev);
|
2009-09-16 03:57:34 +07:00
|
|
|
mutex_lock(&dev->struct_mutex);
|
|
|
|
|
2010-08-12 16:44:45 +07:00
|
|
|
ret = -EINVAL;
|
2009-09-16 03:57:34 +07:00
|
|
|
if (!(attrs->flags & I915_OVERLAY_UPDATE_ATTRS)) {
|
2010-08-12 16:44:45 +07:00
|
|
|
attrs->color_key = overlay->color_key;
|
2009-09-16 03:57:34 +07:00
|
|
|
attrs->brightness = overlay->brightness;
|
2010-08-12 16:44:45 +07:00
|
|
|
attrs->contrast = overlay->contrast;
|
2009-09-16 03:57:34 +07:00
|
|
|
attrs->saturation = overlay->saturation;
|
|
|
|
|
drm/i915: replace IS_GEN<N> with IS_GEN(..., N)
Define IS_GEN() similarly to our IS_GEN_RANGE(). but use gen instead of
gen_mask to do the comparison. Now callers can pass then gen as a parameter,
so we don't require one macro for each gen.
The following spatch was used to convert the users of these macros:
@@
expression e;
@@
(
- IS_GEN2(e)
+ IS_GEN(e, 2)
|
- IS_GEN3(e)
+ IS_GEN(e, 3)
|
- IS_GEN4(e)
+ IS_GEN(e, 4)
|
- IS_GEN5(e)
+ IS_GEN(e, 5)
|
- IS_GEN6(e)
+ IS_GEN(e, 6)
|
- IS_GEN7(e)
+ IS_GEN(e, 7)
|
- IS_GEN8(e)
+ IS_GEN(e, 8)
|
- IS_GEN9(e)
+ IS_GEN(e, 9)
|
- IS_GEN10(e)
+ IS_GEN(e, 10)
|
- IS_GEN11(e)
+ IS_GEN(e, 11)
)
v2: use IS_GEN rather than GT_GEN and compare to info.gen rather than
using the bitmask
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181212181044.15886-2-lucas.demarchi@intel.com
2018-12-13 01:10:43 +07:00
|
|
|
if (!IS_GEN(dev_priv, 2)) {
|
2009-09-16 03:57:34 +07:00
|
|
|
attrs->gamma0 = I915_READ(OGAMC0);
|
|
|
|
attrs->gamma1 = I915_READ(OGAMC1);
|
|
|
|
attrs->gamma2 = I915_READ(OGAMC2);
|
|
|
|
attrs->gamma3 = I915_READ(OGAMC3);
|
|
|
|
attrs->gamma4 = I915_READ(OGAMC4);
|
|
|
|
attrs->gamma5 = I915_READ(OGAMC5);
|
|
|
|
}
|
|
|
|
} else {
|
2010-08-12 16:44:45 +07:00
|
|
|
if (attrs->brightness < -128 || attrs->brightness > 127)
|
2009-09-16 03:57:34 +07:00
|
|
|
goto out_unlock;
|
2010-08-12 16:44:45 +07:00
|
|
|
if (attrs->contrast > 255)
|
2009-09-16 03:57:34 +07:00
|
|
|
goto out_unlock;
|
2010-08-12 16:44:45 +07:00
|
|
|
if (attrs->saturation > 1023)
|
2009-09-16 03:57:34 +07:00
|
|
|
goto out_unlock;
|
|
|
|
|
2010-08-12 16:44:45 +07:00
|
|
|
overlay->color_key = attrs->color_key;
|
|
|
|
overlay->brightness = attrs->brightness;
|
|
|
|
overlay->contrast = attrs->contrast;
|
|
|
|
overlay->saturation = attrs->saturation;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
update_reg_attrs(overlay, overlay->regs);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
if (attrs->flags & I915_OVERLAY_UPDATE_GAMMA) {
|
drm/i915: replace IS_GEN<N> with IS_GEN(..., N)
Define IS_GEN() similarly to our IS_GEN_RANGE(). but use gen instead of
gen_mask to do the comparison. Now callers can pass then gen as a parameter,
so we don't require one macro for each gen.
The following spatch was used to convert the users of these macros:
@@
expression e;
@@
(
- IS_GEN2(e)
+ IS_GEN(e, 2)
|
- IS_GEN3(e)
+ IS_GEN(e, 3)
|
- IS_GEN4(e)
+ IS_GEN(e, 4)
|
- IS_GEN5(e)
+ IS_GEN(e, 5)
|
- IS_GEN6(e)
+ IS_GEN(e, 6)
|
- IS_GEN7(e)
+ IS_GEN(e, 7)
|
- IS_GEN8(e)
+ IS_GEN(e, 8)
|
- IS_GEN9(e)
+ IS_GEN(e, 9)
|
- IS_GEN10(e)
+ IS_GEN(e, 10)
|
- IS_GEN11(e)
+ IS_GEN(e, 11)
)
v2: use IS_GEN rather than GT_GEN and compare to info.gen rather than
using the bitmask
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181212181044.15886-2-lucas.demarchi@intel.com
2018-12-13 01:10:43 +07:00
|
|
|
if (IS_GEN(dev_priv, 2))
|
2009-09-16 03:57:34 +07:00
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
if (overlay->active) {
|
|
|
|
ret = -EBUSY;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = check_gamma(attrs);
|
2010-08-12 16:44:45 +07:00
|
|
|
if (ret)
|
2009-09-16 03:57:34 +07:00
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
I915_WRITE(OGAMC0, attrs->gamma0);
|
|
|
|
I915_WRITE(OGAMC1, attrs->gamma1);
|
|
|
|
I915_WRITE(OGAMC2, attrs->gamma2);
|
|
|
|
I915_WRITE(OGAMC3, attrs->gamma3);
|
|
|
|
I915_WRITE(OGAMC4, attrs->gamma4);
|
|
|
|
I915_WRITE(OGAMC5, attrs->gamma5);
|
|
|
|
}
|
|
|
|
}
|
2015-04-02 16:35:08 +07:00
|
|
|
overlay->color_key_enabled = (attrs->flags & I915_OVERLAY_DISABLE_DEST_COLORKEY) == 0;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2010-08-12 16:44:45 +07:00
|
|
|
ret = 0;
|
2009-09-16 03:57:34 +07:00
|
|
|
out_unlock:
|
|
|
|
mutex_unlock(&dev->struct_mutex);
|
2012-12-02 07:05:46 +07:00
|
|
|
drm_modeset_unlock_all(dev);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
static int get_registers(struct intel_overlay *overlay, bool use_phys)
|
|
|
|
{
|
|
|
|
struct drm_i915_gem_object *obj;
|
|
|
|
struct i915_vma *vma;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
obj = i915_gem_object_create_stolen(overlay->i915, PAGE_SIZE);
|
|
|
|
if (obj == NULL)
|
|
|
|
obj = i915_gem_object_create_internal(overlay->i915, PAGE_SIZE);
|
|
|
|
if (IS_ERR(obj))
|
|
|
|
return PTR_ERR(obj);
|
|
|
|
|
|
|
|
vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, PIN_MAPPABLE);
|
|
|
|
if (IS_ERR(vma)) {
|
|
|
|
err = PTR_ERR(vma);
|
|
|
|
goto err_put_bo;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (use_phys)
|
|
|
|
overlay->flip_addr = sg_dma_address(obj->mm.pages->sgl);
|
|
|
|
else
|
|
|
|
overlay->flip_addr = i915_ggtt_offset(vma);
|
|
|
|
overlay->regs = i915_vma_pin_iomap(vma);
|
|
|
|
i915_vma_unpin(vma);
|
|
|
|
|
|
|
|
if (IS_ERR(overlay->regs)) {
|
|
|
|
err = PTR_ERR(overlay->regs);
|
|
|
|
goto err_put_bo;
|
|
|
|
}
|
|
|
|
|
|
|
|
overlay->reg_bo = obj;
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
err_put_bo:
|
|
|
|
i915_gem_object_put(obj);
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2018-11-08 07:16:47 +07:00
|
|
|
void intel_overlay_setup(struct drm_i915_private *dev_priv)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
|
|
|
struct intel_overlay *overlay;
|
|
|
|
int ret;
|
|
|
|
|
2016-05-12 18:43:23 +07:00
|
|
|
if (!HAS_OVERLAY(dev_priv))
|
2009-09-16 03:57:34 +07:00
|
|
|
return;
|
|
|
|
|
2013-09-19 17:18:32 +07:00
|
|
|
overlay = kzalloc(sizeof(*overlay), GFP_KERNEL);
|
2009-09-16 03:57:34 +07:00
|
|
|
if (!overlay)
|
|
|
|
return;
|
2011-06-28 17:27:47 +07:00
|
|
|
|
2016-05-12 18:43:23 +07:00
|
|
|
overlay->i915 = dev_priv;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
overlay->color_key = 0x0101fe;
|
2015-04-02 16:35:08 +07:00
|
|
|
overlay->color_key_enabled = true;
|
2009-09-16 03:57:34 +07:00
|
|
|
overlay->brightness = -19;
|
|
|
|
overlay->contrast = 75;
|
|
|
|
overlay->saturation = 146;
|
|
|
|
|
2016-12-21 21:45:47 +07:00
|
|
|
init_request_active(&overlay->last_flip, NULL);
|
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
mutex_lock(&dev_priv->drm.struct_mutex);
|
|
|
|
|
|
|
|
ret = get_registers(overlay, OVERLAY_NEEDS_PHYSICAL(dev_priv));
|
|
|
|
if (ret)
|
|
|
|
goto out_free;
|
|
|
|
|
|
|
|
ret = i915_gem_object_set_to_gtt_domain(overlay->reg_bo, true);
|
|
|
|
if (ret)
|
|
|
|
goto out_reg_bo;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
mutex_unlock(&dev_priv->drm.struct_mutex);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
memset_io(overlay->regs, 0, sizeof(struct overlay_registers));
|
|
|
|
update_polyphase_filter(overlay->regs);
|
|
|
|
update_reg_attrs(overlay, overlay->regs);
|
2009-09-16 03:57:34 +07:00
|
|
|
|
|
|
|
dev_priv->overlay = overlay;
|
2018-09-07 02:01:43 +07:00
|
|
|
DRM_INFO("Initialized overlay support.\n");
|
2009-09-16 03:57:34 +07:00
|
|
|
return;
|
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
out_reg_bo:
|
|
|
|
i915_gem_object_put(overlay->reg_bo);
|
2009-09-16 03:57:34 +07:00
|
|
|
out_free:
|
2016-07-05 16:40:23 +07:00
|
|
|
mutex_unlock(&dev_priv->drm.struct_mutex);
|
2009-09-16 03:57:34 +07:00
|
|
|
kfree(overlay);
|
|
|
|
}
|
|
|
|
|
2018-11-08 07:16:47 +07:00
|
|
|
void intel_overlay_cleanup(struct drm_i915_private *dev_priv)
|
2009-09-16 03:57:34 +07:00
|
|
|
{
|
2018-09-07 02:01:43 +07:00
|
|
|
struct intel_overlay *overlay;
|
|
|
|
|
|
|
|
overlay = fetch_and_zero(&dev_priv->overlay);
|
|
|
|
if (!overlay)
|
2010-08-12 16:50:36 +07:00
|
|
|
return;
|
2009-09-16 03:57:34 +07:00
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
/*
|
|
|
|
* The bo's should be free'd by the generic code already.
|
2010-08-12 16:50:36 +07:00
|
|
|
* Furthermore modesetting teardown happens beforehand so the
|
2018-09-07 02:01:43 +07:00
|
|
|
* hardware should be off already.
|
|
|
|
*/
|
|
|
|
WARN_ON(overlay->active);
|
|
|
|
|
|
|
|
i915_gem_object_put(overlay->reg_bo);
|
2010-08-12 16:50:36 +07:00
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
kfree(overlay);
|
2009-09-16 03:57:34 +07:00
|
|
|
}
|
2010-08-05 02:26:07 +07:00
|
|
|
|
2016-10-12 16:05:18 +07:00
|
|
|
#if IS_ENABLED(CONFIG_DRM_I915_CAPTURE_ERROR)
|
|
|
|
|
2010-08-05 02:26:07 +07:00
|
|
|
struct intel_overlay_error_state {
|
|
|
|
struct overlay_registers regs;
|
|
|
|
unsigned long base;
|
|
|
|
u32 dovsta;
|
|
|
|
u32 isr;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct intel_overlay_error_state *
|
2016-05-06 21:40:21 +07:00
|
|
|
intel_overlay_capture_error_state(struct drm_i915_private *dev_priv)
|
2010-08-05 02:26:07 +07:00
|
|
|
{
|
|
|
|
struct intel_overlay *overlay = dev_priv->overlay;
|
|
|
|
struct intel_overlay_error_state *error;
|
|
|
|
|
|
|
|
if (!overlay || !overlay->active)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
error = kmalloc(sizeof(*error), GFP_ATOMIC);
|
|
|
|
if (error == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
error->dovsta = I915_READ(DOVSTA);
|
|
|
|
error->isr = I915_READ(ISR);
|
2016-04-28 15:56:36 +07:00
|
|
|
error->base = overlay->flip_addr;
|
2010-08-05 02:26:07 +07:00
|
|
|
|
2018-09-07 02:01:43 +07:00
|
|
|
memcpy_fromio(&error->regs, overlay->regs, sizeof(error->regs));
|
2010-08-05 02:26:07 +07:00
|
|
|
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
2013-05-23 17:55:35 +07:00
|
|
|
intel_overlay_print_error_state(struct drm_i915_error_state_buf *m,
|
|
|
|
struct intel_overlay_error_state *error)
|
2010-08-05 02:26:07 +07:00
|
|
|
{
|
2013-05-23 17:55:35 +07:00
|
|
|
i915_error_printf(m, "Overlay, status: 0x%08x, interrupt: 0x%08x\n",
|
|
|
|
error->dovsta, error->isr);
|
|
|
|
i915_error_printf(m, " Register file at 0x%08lx:\n",
|
|
|
|
error->base);
|
2010-08-05 02:26:07 +07:00
|
|
|
|
2013-05-23 17:55:35 +07:00
|
|
|
#define P(x) i915_error_printf(m, " " #x ": 0x%08x\n", error->regs.x)
|
2010-08-05 02:26:07 +07:00
|
|
|
P(OBUF_0Y);
|
|
|
|
P(OBUF_1Y);
|
|
|
|
P(OBUF_0U);
|
|
|
|
P(OBUF_0V);
|
|
|
|
P(OBUF_1U);
|
|
|
|
P(OBUF_1V);
|
|
|
|
P(OSTRIDE);
|
|
|
|
P(YRGB_VPH);
|
|
|
|
P(UV_VPH);
|
|
|
|
P(HORZ_PH);
|
|
|
|
P(INIT_PHS);
|
|
|
|
P(DWINPOS);
|
|
|
|
P(DWINSZ);
|
|
|
|
P(SWIDTH);
|
|
|
|
P(SWIDTHSW);
|
|
|
|
P(SHEIGHT);
|
|
|
|
P(YRGBSCALE);
|
|
|
|
P(UVSCALE);
|
|
|
|
P(OCLRC0);
|
|
|
|
P(OCLRC1);
|
|
|
|
P(DCLRKV);
|
|
|
|
P(DCLRKM);
|
|
|
|
P(SCLRKVH);
|
|
|
|
P(SCLRKVL);
|
|
|
|
P(SCLRKEN);
|
|
|
|
P(OCONFIG);
|
|
|
|
P(OCMD);
|
|
|
|
P(OSTART_0Y);
|
|
|
|
P(OSTART_1Y);
|
|
|
|
P(OSTART_0U);
|
|
|
|
P(OSTART_0V);
|
|
|
|
P(OSTART_1U);
|
|
|
|
P(OSTART_1V);
|
|
|
|
P(OTILEOFF_0Y);
|
|
|
|
P(OTILEOFF_1Y);
|
|
|
|
P(OTILEOFF_0U);
|
|
|
|
P(OTILEOFF_0V);
|
|
|
|
P(OTILEOFF_1U);
|
|
|
|
P(OTILEOFF_1V);
|
|
|
|
P(FASTHSCALE);
|
|
|
|
P(UVSCALEV);
|
|
|
|
#undef P
|
|
|
|
}
|
2016-10-12 16:05:18 +07:00
|
|
|
|
|
|
|
#endif
|