i915_gem_tiling.c revision 1450
1450N/A/*
1450N/A * Copyright (c) 2006, 2013, Oracle and/or its affiliates. All rights reserved.
1450N/A */
1450N/A
1450N/A/*
1450N/A * Copyright (c) 2009, 2013, Intel Corporation.
1450N/A * All Rights Reserved.
1450N/A *
1450N/A * Permission is hereby granted, free of charge, to any person obtaining a
1450N/A * copy of this software and associated documentation files (the "Software"),
1450N/A * to deal in the Software without restriction, including without limitation
1450N/A * the rights to use, copy, modify, merge, publish, distribute, sublicense,
1450N/A * and/or sell copies of the Software, and to permit persons to whom the
1450N/A * Software is furnished to do so, subject to the following conditions:
1450N/A *
1450N/A * The above copyright notice and this permission notice (including the next
1450N/A * paragraph) shall be included in all copies or substantial portions of the
1450N/A * Software.
1450N/A *
1450N/A * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
1450N/A * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
1450N/A * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
1450N/A * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
1450N/A * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
1450N/A * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
1450N/A * IN THE SOFTWARE.
1450N/A *
1450N/A * Authors:
1450N/A * Eric Anholt <eric@anholt.net>
1450N/A *
1450N/A */
1450N/A
1450N/A#include "drmP.h"
1450N/A#include "drm.h"
1450N/A#include "i915_drm.h"
1450N/A#include "i915_drv.h"
1450N/A
1450N/A/** @file i915_gem_tiling.c
1450N/A *
1450N/A * Support for managing tiling state of buffer objects.
1450N/A *
1450N/A * The idea behind tiling is to increase cache hit rates by rearranging
1450N/A * pixel data so that a group of pixel accesses are in the same cacheline.
1450N/A * Performance improvement from doing this on the back/depth buffer are on
1450N/A * the order of 30%.
1450N/A *
1450N/A * Intel architectures make this somewhat more complicated, though, by
1450N/A * adjustments made to addressing of data when the memory is in interleaved
1450N/A * mode (matched pairs of DIMMS) to improve memory bandwidth.
1450N/A * For interleaved memory, the CPU sends every sequential 64 bytes
1450N/A * to an alternate memory channel so it can get the bandwidth from both.
1450N/A *
1450N/A * The GPU also rearranges its accesses for increased bandwidth to interleaved
1450N/A * memory, and it matches what the CPU does for non-tiled. However, when tiled
1450N/A * it does it a little differently, since one walks addresses not just in the
1450N/A * X direction but also Y. So, along with alternating channels when bit
1450N/A * 6 of the address flips, it also alternates when other bits flip -- Bits 9
1450N/A * (every 512 bytes, an X tile scanline) and 10 (every two X tile scanlines)
1450N/A * are common to both the 915 and 965-class hardware.
1450N/A *
1450N/A * The CPU also sometimes XORs in higher bits as well, to improve
1450N/A * bandwidth doing strided access like we do so frequently in graphics. This
1450N/A * is called "Channel XOR Randomization" in the MCH documentation. The result
1450N/A * is that the CPU is XORing in either bit 11 or bit 17 to bit 6 of its address
1450N/A * decode.
1450N/A *
1450N/A * All of this bit 6 XORing has an effect on our memory management,
1450N/A * as we need to make sure that the 3d driver can correctly address object
1450N/A * contents.
1450N/A *
1450N/A * If we don't have interleaved memory, all tiling is safe and no swizzling is
1450N/A * required.
1450N/A *
1450N/A * When bit 17 is XORed in, we simply refuse to tile at all. Bit
1450N/A * 17 is not just a page offset, so as we page an objet out and back in,
1450N/A * individual pages in it will have different bit 17 addresses, resulting in
1450N/A * each 64 bytes being swapped with its neighbor!
1450N/A *
1450N/A * Otherwise, if interleaved, we have to tell the 3d driver what the address
1450N/A * swizzling it needs to do is, since it's writing with the CPU to the pages
1450N/A * (bit 6 and potentially bit 11 XORed in), and the GPU is reading from the
1450N/A * pages (bit 6, 9, and 10 XORed in), resulting in a cumulative bit swizzling
1450N/A * required by the CPU of XORing in bit 6, 9, 10, and potentially 11, in order
1450N/A * to match what the GPU expects.
1450N/A */
1450N/A
1450N/A/**
1450N/A * Detects bit 6 swizzling of address lookup between IGD access and CPU
1450N/A * access through main memory.
1450N/A */
1450N/Avoid
1450N/Ai915_gem_detect_bit_6_swizzle(struct drm_device *dev)
1450N/A{
1450N/A drm_i915_private_t *dev_priv = dev->dev_private;
1450N/A uint32_t swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN;
1450N/A uint32_t swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN;
1450N/A
1450N/A if (IS_VALLEYVIEW(dev)) {
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_NONE;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_NONE;
1450N/A } else if (INTEL_INFO(dev)->gen >= 6) {
1450N/A uint32_t dimm_c0, dimm_c1;
1450N/A dimm_c0 = I915_READ(MAD_DIMM_C0);
1450N/A dimm_c1 = I915_READ(MAD_DIMM_C1);
1450N/A dimm_c0 &= MAD_DIMM_A_SIZE_MASK | MAD_DIMM_B_SIZE_MASK;
1450N/A dimm_c1 &= MAD_DIMM_A_SIZE_MASK | MAD_DIMM_B_SIZE_MASK;
1450N/A /* Enable swizzling when the channels are populated with
1450N/A * identically sized dimms. We don't need to check the 3rd
1450N/A * channel because no cpu with gpu attached ships in that
1450N/A * configuration. Also, swizzling only makes sense for 2
1450N/A * channels anyway. */
1450N/A if (dimm_c0 == dimm_c1) {
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_9_10;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_9;
1450N/A } else {
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_NONE;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_NONE;
1450N/A }
1450N/A } else if (IS_GEN5(dev)) {
1450N/A /* On IRONLAKE whatever DRAM config, GPU always do
1450N/A * same swizzling setup.
1450N/A */
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_9_10;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_9;
1450N/A } else if (IS_GEN2(dev)) {
1450N/A /* As far as we know, the 865 doesn't have these bit 6
1450N/A * swizzling issues.
1450N/A */
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_NONE;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_NONE;
1450N/A } else if (IS_MOBILE(dev) || (IS_GEN3(dev) && !IS_G33(dev))) {
1450N/A uint32_t dcc;
1450N/A
1450N/A /* On mobile 9xx chipsets, channel interleave by the CPU is
1450N/A * determined by DCC. For single-channel, neither the CPU
1450N/A * nor the GPU do swizzling. For dual channel interleaved,
1450N/A * the GPU's interleave is bit 9 and 10 for X tiled, and bit
1450N/A * 9 for Y tiled. The CPU's interleave is independent, and
1450N/A * can be based on either bit 11 (haven't seen this yet) or
1450N/A * bit 17 (common).
1450N/A */
1450N/A dcc = I915_READ(DCC);
1450N/A switch (dcc & DCC_ADDRESSING_MODE_MASK) {
1450N/A case DCC_ADDRESSING_MODE_SINGLE_CHANNEL:
1450N/A case DCC_ADDRESSING_MODE_DUAL_CHANNEL_ASYMMETRIC:
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_NONE;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_NONE;
1450N/A break;
1450N/A case DCC_ADDRESSING_MODE_DUAL_CHANNEL_INTERLEAVED:
1450N/A if (dcc & DCC_CHANNEL_XOR_DISABLE) {
1450N/A /* This is the base swizzling by the GPU for
1450N/A * tiled buffers.
1450N/A */
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_9_10;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_9;
1450N/A } else if ((dcc & DCC_CHANNEL_XOR_BIT_17) == 0) {
1450N/A /* Work around: for fixing 965GM flickering issue on OpenArena */
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_9_10_11;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_9_11;
1450N/A } else {
1450N/A /* Bit 17 swizzling by the CPU in addition. */
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_9_10_17;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_9_17;
1450N/A }
1450N/A break;
1450N/A }
1450N/A if (dcc == 0xffffffff) {
1450N/A DRM_ERROR("Couldn't read from MCHBAR. "
1450N/A "Disabling tiling.\n");
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN;
1450N/A }
1450N/A } else {
1450N/A /* The 965, G33, and newer, have a very flexible memory
1450N/A * configuration. It will enable dual-channel mode
1450N/A * (interleaving) on as much memory as it can, and the GPU
1450N/A * will additionally sometimes enable different bit 6
1450N/A * swizzling for tiled objects from the CPU.
1450N/A *
1450N/A * Here's what I found on the G965:
1450N/A * slot fill memory size swizzling
1450N/A * 0A 0B 1A 1B 1-ch 2-ch
1450N/A * 512 0 0 0 512 0 O
1450N/A * 512 0 512 0 16 1008 X
1450N/A * 512 0 0 512 16 1008 X
1450N/A * 0 512 0 512 16 1008 X
1450N/A * 1024 1024 1024 0 2048 1024 O
1450N/A *
1450N/A * We could probably detect this based on either the DRB
1450N/A * matching, which was the case for the swizzling required in
1450N/A * the table above, or from the 1-ch value being less than
1450N/A * the minimum size of a rank.
1450N/A */
1450N/A if (I915_READ16(C0DRB3) != I915_READ16(C1DRB3)) {
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_NONE;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_NONE;
1450N/A } else {
1450N/A swizzle_x = I915_BIT_6_SWIZZLE_9_10;
1450N/A swizzle_y = I915_BIT_6_SWIZZLE_9;
1450N/A }
1450N/A }
1450N/A
1450N/A dev_priv->mm.bit_6_swizzle_x = swizzle_x;
1450N/A dev_priv->mm.bit_6_swizzle_y = swizzle_y;
1450N/A}
1450N/A
1450N/A/* Check pitch constriants for all chips & tiling formats */
1450N/Astatic bool
1450N/Ai915_tiling_ok(struct drm_device *dev, int stride, int size, int tiling_mode)
1450N/A{
1450N/A int tile_width;
1450N/A
1450N/A /* Linear is always fine */
1450N/A if (tiling_mode == I915_TILING_NONE)
1450N/A return true;
1450N/A
1450N/A if (IS_GEN2(dev) ||
1450N/A (tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev)))
1450N/A tile_width = 128;
1450N/A else
1450N/A tile_width = 512;
1450N/A
1450N/A /* check maximum stride & object size */
1450N/A /* i965+ stores the end address of the gtt mapping in the fence
1450N/A * reg, so dont bother to check the size */
1450N/A if (INTEL_INFO(dev)->gen >= 7) {
1450N/A if (stride / 128 > GEN7_FENCE_MAX_PITCH_VAL)
1450N/A return false;
1450N/A } else if (INTEL_INFO(dev)->gen >= 4) {
1450N/A if (stride / 128 > I965_FENCE_MAX_PITCH_VAL)
1450N/A return false;
1450N/A } else {
1450N/A if (stride > 8192)
1450N/A return false;
1450N/A
1450N/A if (IS_GEN3(dev)) {
1450N/A if (size > I830_FENCE_MAX_SIZE_VAL << 20)
1450N/A return false;
1450N/A } else {
1450N/A if (size > I830_FENCE_MAX_SIZE_VAL << 19)
1450N/A return false;
1450N/A }
1450N/A }
1450N/A
1450N/A if (stride < tile_width)
1450N/A return false;
1450N/A
1450N/A /* 965+ just needs multiples of tile width */
1450N/A if (INTEL_INFO(dev)->gen >= 4) {
1450N/A if (stride & (tile_width - 1))
1450N/A return false;
1450N/A return true;
1450N/A }
1450N/A
1450N/A /* Pre-965 needs power of two tile widths */
1450N/A if (stride & (stride - 1))
1450N/A return false;
1450N/A
1450N/A return true;
1450N/A}
1450N/A
1450N/A/* Is the current GTT allocation valid for the change in tiling? */
1450N/Astatic bool
1450N/Ai915_gem_object_fence_ok(struct drm_i915_gem_object *obj, int tiling_mode)
1450N/A{
1450N/A u32 size;
1450N/A
1450N/A if (tiling_mode == I915_TILING_NONE)
1450N/A return true;
1450N/A
1450N/A if (INTEL_INFO(obj->base.dev)->gen >= 4)
1450N/A return true;
1450N/A
1450N/A if (INTEL_INFO(obj->base.dev)->gen == 3) {
1450N/A if (obj->gtt_offset & ~I915_FENCE_START_MASK)
1450N/A return false;
1450N/A } else {
1450N/A if (obj->gtt_offset & ~I830_FENCE_START_MASK)
1450N/A return false;
1450N/A }
1450N/A
1450N/A size = i915_gem_get_gtt_size(obj->base.dev, obj->base.size, tiling_mode);
1450N/A if (obj->gtt_space->size != size)
1450N/A return false;
1450N/A
1450N/A if (obj->gtt_offset & (size - 1))
1450N/A return false;
1450N/A
1450N/A return true;
1450N/A}
1450N/A
1450N/A/**
1450N/A * Sets the tiling mode of an object, returning the required swizzling of
1450N/A * bit 6 of addresses in the object.
1450N/A */
1450N/Aint
1450N/A/* LINTED */
1450N/Ai915_gem_set_tiling(DRM_IOCTL_ARGS)
1450N/A{
1450N/A struct drm_i915_gem_set_tiling *args = data;
1450N/A drm_i915_private_t *dev_priv = dev->dev_private;
1450N/A struct drm_i915_gem_object *obj;
1450N/A int ret = 0;
1450N/A
1450N/A obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
1450N/A if (&obj->base == NULL)
1450N/A return -ENOENT;
1450N/A
1450N/A if (!i915_tiling_ok(dev,
1450N/A args->stride, obj->base.size, args->tiling_mode)) {
1450N/A drm_gem_object_unreference_unlocked(&obj->base);
1450N/A return -EINVAL;
1450N/A }
1450N/A
1450N/A if (obj->pin_count) {
1450N/A drm_gem_object_unreference_unlocked(&obj->base);
1450N/A return -EBUSY;
1450N/A }
1450N/A
1450N/A if (args->tiling_mode == I915_TILING_NONE) {
1450N/A args->swizzle_mode = I915_BIT_6_SWIZZLE_NONE;
1450N/A args->stride = 0;
1450N/A } else {
1450N/A if (args->tiling_mode == I915_TILING_X)
1450N/A args->swizzle_mode = dev_priv->mm.bit_6_swizzle_x;
1450N/A else
1450N/A args->swizzle_mode = dev_priv->mm.bit_6_swizzle_y;
1450N/A
1450N/A /* Hide bit 17 swizzling from the user. This prevents old Mesa
1450N/A * from aborting the application on sw fallbacks to bit 17,
1450N/A * and we use the pread/pwrite bit17 paths to swizzle for it.
1450N/A * If there was a user that was relying on the swizzle
1450N/A * information for drm_intel_bo_map()ed reads/writes this would
1450N/A * break it, but we don't have any of those.
1450N/A */
1450N/A if (args->swizzle_mode == I915_BIT_6_SWIZZLE_9_17)
1450N/A args->swizzle_mode = I915_BIT_6_SWIZZLE_9;
1450N/A if (args->swizzle_mode == I915_BIT_6_SWIZZLE_9_10_17)
1450N/A args->swizzle_mode = I915_BIT_6_SWIZZLE_9_10;
1450N/A
1450N/A /* If we can't handle the swizzling, make it untiled. */
1450N/A if (args->swizzle_mode == I915_BIT_6_SWIZZLE_UNKNOWN) {
1450N/A args->tiling_mode = I915_TILING_NONE;
1450N/A args->swizzle_mode = I915_BIT_6_SWIZZLE_NONE;
1450N/A args->stride = 0;
1450N/A }
1450N/A }
1450N/A
1450N/A mutex_lock(&dev->struct_mutex);
1450N/A if (args->tiling_mode != obj->tiling_mode ||
1450N/A args->stride != obj->stride) {
1450N/A /* We need to rebind the object if its current allocation
1450N/A * no longer meets the alignment restrictions for its new
1450N/A * tiling mode. Otherwise we can just leave it alone, but
1450N/A * need to ensure that any fence register is cleared.
1450N/A * the next fenced (either through the GTT or by the BLT unit
1450N/A * on older GPUs) access.
1450N/A *
1450N/A * After updating the tiling parameters, we then flag whether
1450N/A * we need to update an associated fence register. Note this
1450N/A * has to also include the unfenced register the GPU uses
1450N/A * whilst executing a fenced command for an untiled object.
1450N/A */
1450N/A
1450N/A obj->map_and_fenceable =
1450N/A obj->gtt_space == NULL ||
1450N/A (obj->gtt_offset + obj->base.size <= dev_priv->gtt.mappable_end &&
1450N/A i915_gem_object_fence_ok(obj, args->tiling_mode));
1450N/A
1450N/A /* Rebind if we need a change of alignment */
1450N/A if (!obj->map_and_fenceable) {
1450N/A u32 unfenced_alignment =
1450N/A i915_gem_get_gtt_alignment(dev, obj->base.size,
1450N/A args->tiling_mode,
1450N/A false);
1450N/A if (obj->gtt_offset & (unfenced_alignment - 1))
1450N/A ret = i915_gem_object_unbind(obj, 1);
1450N/A }
1450N/A
1450N/A if (ret == 0) {
1450N/A obj->fence_dirty =
1450N/A obj->fenced_gpu_access ||
1450N/A obj->fence_reg != I915_FENCE_REG_NONE;
1450N/A
1450N/A obj->tiling_mode = args->tiling_mode;
1450N/A obj->stride = args->stride;
1450N/A
1450N/A /* Force the fence to be reacquired for GTT access */
1450N/A i915_gem_release_mmap(obj);
1450N/A }
1450N/A }
1450N/A /* we have to maintain this existing ABI... */
1450N/A args->stride = obj->stride;
1450N/A args->tiling_mode = obj->tiling_mode;
1450N/A
1450N/A /* Try to preallocate memory required to save swizzling on put-pages */
1450N/A if (i915_gem_object_needs_bit17_swizzle(obj)) {
1450N/A if (obj->bit_17 == NULL) {
1450N/A obj->bit_17 = kmalloc(BITS_TO_LONGS(obj->base.size >> PAGE_SHIFT) *
1450N/A sizeof(long), GFP_KERNEL);
1450N/A }
1450N/A } else {
1450N/A if (obj->bit_17 != NULL) {
1450N/A kfree(obj->bit_17, BITS_TO_LONGS(obj->base.size >> PAGE_SHIFT) *
1450N/A sizeof(long));
1450N/A obj->bit_17 = NULL;
1450N/A }
1450N/A }
1450N/A
1450N/A drm_gem_object_unreference(&obj->base);
1450N/A mutex_unlock(&dev->struct_mutex);
1450N/A
1450N/A return ret;
1450N/A}
1450N/A
1450N/A/**
1450N/A * Returns the current tiling mode and required bit 6 swizzling for the object.
1450N/A */
1450N/Aint
1450N/A/* LINTED */
1450N/Ai915_gem_get_tiling(DRM_IOCTL_ARGS)
1450N/A{
1450N/A struct drm_i915_gem_get_tiling *args = data;
1450N/A drm_i915_private_t *dev_priv = dev->dev_private;
1450N/A struct drm_i915_gem_object *obj;
1450N/A
1450N/A obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
1450N/A if (&obj->base == NULL)
1450N/A return -ENOENT;
1450N/A
1450N/A mutex_lock(&dev->struct_mutex);
1450N/A
1450N/A args->tiling_mode = obj->tiling_mode;
1450N/A switch (obj->tiling_mode) {
1450N/A case I915_TILING_X:
1450N/A args->swizzle_mode = dev_priv->mm.bit_6_swizzle_x;
1450N/A break;
1450N/A case I915_TILING_Y:
1450N/A args->swizzle_mode = dev_priv->mm.bit_6_swizzle_y;
1450N/A break;
1450N/A case I915_TILING_NONE:
1450N/A args->swizzle_mode = I915_BIT_6_SWIZZLE_NONE;
1450N/A break;
1450N/A default:
1450N/A DRM_ERROR("unknown tiling mode\n");
1450N/A }
1450N/A
1450N/A /* Hide bit 17 from the user -- see comment in i915_gem_set_tiling */
1450N/A if (args->swizzle_mode == I915_BIT_6_SWIZZLE_9_17)
1450N/A args->swizzle_mode = I915_BIT_6_SWIZZLE_9;
1450N/A if (args->swizzle_mode == I915_BIT_6_SWIZZLE_9_10_17)
1450N/A args->swizzle_mode = I915_BIT_6_SWIZZLE_9_10;
1450N/A
1450N/A drm_gem_object_unreference(&obj->base);
1450N/A mutex_unlock(&dev->struct_mutex);
1450N/A
1450N/A return 0;
1450N/A}
1450N/A
1450N/A/**
1450N/A * Swap every 64 bytes of this page around, to account for it having a new
1450N/A * bit 17 of its physical address and therefore being interpreted differently
1450N/A * by the GPU.
1450N/A */
1450N/Astatic void
1450N/Ai915_gem_swizzle_page(caddr_t va)
1450N/A{
1450N/A char temp[64];
1450N/A char *vaddr;
1450N/A int i;
1450N/A
1450N/A vaddr = (char *)(uintptr_t) va;
1450N/A
1450N/A for (i = 0; i < PAGE_SIZE; i += 128) {
1450N/A (void) memcpy(temp, &vaddr[i], 64);
1450N/A (void) memcpy(&vaddr[i], &vaddr[i + 64], 64);
1450N/A (void) memcpy(&vaddr[i + 64], temp, 64);
1450N/A }
1450N/A
1450N/A}
1450N/A
1450N/Avoid
1450N/Ai915_gem_object_do_bit_17_swizzle(struct drm_i915_gem_object *obj)
1450N/A{
1450N/A int page_count = obj->base.size >> PAGE_SHIFT;
1450N/A int i;
1450N/A
1450N/A if (obj->bit_17 == NULL)
1450N/A return;
1450N/A
1450N/A for (i = 0; i < page_count; i++) {
1450N/A char new_bit_17 = (char)(page_to_phys(obj->page_list[i]) >> 17);
1450N/A if ((new_bit_17 & 0x1) !=
1450N/A (test_bit(i, obj->bit_17) != 0)) {
1450N/A i915_gem_swizzle_page(obj->page_list[i]);
1450N/A }
1450N/A }
1450N/A}
1450N/A
1450N/Avoid
1450N/Ai915_gem_object_save_bit_17_swizzle(struct drm_i915_gem_object *obj)
1450N/A{
1450N/A int page_count = obj->base.size >> PAGE_SHIFT;
1450N/A int i;
1450N/A
1450N/A if (obj->bit_17 == NULL) {
1450N/A obj->bit_17 = kmalloc(BITS_TO_LONGS(page_count) *
1450N/A sizeof(long), GFP_KERNEL);
1450N/A if (obj->bit_17 == NULL) {
1450N/A DRM_ERROR("Failed to allocate memory for bit 17 "
1450N/A "record\n");
1450N/A return;
1450N/A }
1450N/A }
1450N/A
1450N/A for (i = 0; i < page_count; i++) {
1450N/A if (page_to_phys(obj->page_list[i]) & (1 << 17))
1450N/A set_bit(i, obj->bit_17);
1450N/A else
1450N/A clear_bit(i, obj->bit_17);
1450N/A }
1450N/A}