/*
 * Copyright © 2010 Daniel Vetter
 * Copyright © 2011-2014 Intel Corporation
 *
 * 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.
 *
 */

#include <linux/slab.h> /* fault-inject.h is not standalone! */

#include <linux/fault-inject.h>
#include <linux/log2.h>
#include <linux/random.h>
#include <linux/seq_file.h>
#include <linux/stop_machine.h>

#include <asm/set_memory.h>

#include <drm/drmP.h>
#include <drm/i915_drm.h>

#include "i915_drv.h"
#include "i915_vgpu.h"
#include "i915_trace.h"
#include "intel_drv.h"
#include "intel_frontbuffer.h"

#define I915_GFP_DMA (GFP_KERNEL | __GFP_HIGHMEM)

/**
 * DOC: Global GTT views
 *
 * Background and previous state
 *
 * Historically objects could exists (be bound) in global GTT space only as
 * singular instances with a view representing all of the object's backing pages
 * in a linear fashion. This view will be called a normal view.
 *
 * To support multiple views of the same object, where the number of mapped
 * pages is not equal to the backing store, or where the layout of the pages
 * is not linear, concept of a GGTT view was added.
 *
 * One example of an alternative view is a stereo display driven by a single
 * image. In this case we would have a framebuffer looking like this
 * (2x2 pages):
 *
 *    12
 *    34
 *
 * Above would represent a normal GGTT view as normally mapped for GPU or CPU
 * rendering. In contrast, fed to the display engine would be an alternative
 * view which could look something like this:
 *
 *   1212
 *   3434
 *
 * In this example both the size and layout of pages in the alternative view is
 * different from the normal view.
 *
 * Implementation and usage
 *
 * GGTT views are implemented using VMAs and are distinguished via enum
 * i915_ggtt_view_type and struct i915_ggtt_view.
 *
 * A new flavour of core GEM functions which work with GGTT bound objects were
 * added with the _ggtt_ infix, and sometimes with _view postfix to avoid
 * renaming  in large amounts of code. They take the struct i915_ggtt_view
 * parameter encapsulating all metadata required to implement a view.
 *
 * As a helper for callers which are only interested in the normal view,
 * globally const i915_ggtt_view_normal singleton instance exists. All old core
 * GEM API functions, the ones not taking the view parameter, are operating on,
 * or with the normal GGTT view.
 *
 * Code wanting to add or use a new GGTT view needs to:
 *
 * 1. Add a new enum with a suitable name.
 * 2. Extend the metadata in the i915_ggtt_view structure if required.
 * 3. Add support to i915_get_vma_pages().
 *
 * New views are required to build a scatter-gather table from within the
 * i915_get_vma_pages function. This table is stored in the vma.ggtt_view and
 * exists for the lifetime of an VMA.
 *
 * Core API is designed to have copy semantics which means that passed in
 * struct i915_ggtt_view does not need to be persistent (left around after
 * calling the core API functions).
 *
 */

static int
i915_get_ggtt_vma_pages(struct i915_vma *vma);

static void gen6_ggtt_invalidate(struct drm_i915_private *dev_priv)
{
        /* Note that as an uncached mmio write, this should flush the
         * WCB of the writes into the GGTT before it triggers the invalidate.
         */
        I915_WRITE(GFX_FLSH_CNTL_GEN6, GFX_FLSH_CNTL_EN);
}

static void guc_ggtt_invalidate(struct drm_i915_private *dev_priv)
{
        gen6_ggtt_invalidate(dev_priv);
        I915_WRITE(GEN8_GTCR, GEN8_GTCR_INVALIDATE);
}

static void gmch_ggtt_invalidate(struct drm_i915_private *dev_priv)
{
        intel_gtt_chipset_flush();
}

static inline void i915_ggtt_invalidate(struct drm_i915_private *i915)
{
        i915->ggtt.invalidate(i915);
}

int intel_sanitize_enable_ppgtt(struct drm_i915_private *dev_priv,
                                int enable_ppgtt)
{
        bool has_full_ppgtt;
        bool has_full_48bit_ppgtt;

        if (!dev_priv->info.has_aliasing_ppgtt)
                return 0;

        has_full_ppgtt = dev_priv->info.has_full_ppgtt;
        has_full_48bit_ppgtt = dev_priv->info.has_full_48bit_ppgtt;

        if (intel_vgpu_active(dev_priv)) {
                /* GVT-g has no support for 32bit ppgtt */
                has_full_ppgtt = false;
                has_full_48bit_ppgtt = intel_vgpu_has_full_48bit_ppgtt(dev_priv);
        }

        /*
         * We don't allow disabling PPGTT for gen9+ as it's a requirement for
         * execlists, the sole mechanism available to submit work.
         */
        if (enable_ppgtt == 0 && INTEL_GEN(dev_priv) < 9)
                return 0;

        if (enable_ppgtt == 1)
                return 1;

        if (enable_ppgtt == 2 && has_full_ppgtt)
                return 2;

        if (enable_ppgtt == 3 && has_full_48bit_ppgtt)
                return 3;

        /* Disable ppgtt on SNB if VT-d is on. */
        if (IS_GEN6(dev_priv) && intel_vtd_active()) {
                DRM_INFO("Disabling PPGTT because VT-d is on\n");
                return 0;
        }

        /* Early VLV doesn't have this */
        if (IS_VALLEYVIEW(dev_priv) && dev_priv->drm.pdev->revision < 0xb) {
                DRM_DEBUG_DRIVER("disabling PPGTT on pre-B3 step VLV\n");
                return 0;
        }

        if (HAS_LOGICAL_RING_CONTEXTS(dev_priv)) {
                if (has_full_48bit_ppgtt)
                        return 3;

                if (has_full_ppgtt)
                        return 2;
        }

        return 1;
}

static int ppgtt_bind_vma(struct i915_vma *vma,
                          enum i915_cache_level cache_level,
                          u32 unused)
{
        u32 pte_flags;
        int ret;

        if (!(vma->flags & I915_VMA_LOCAL_BIND)) {
                ret = vma->vm->allocate_va_range(vma->vm, vma->node.start,
                                                 vma->size);
                if (ret)
                        return ret;
        }

        /* Currently applicable only to VLV */
        pte_flags = 0;
        if (vma->obj->gt_ro)
                pte_flags |= PTE_READ_ONLY;

        vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags);

        return 0;
}

static void ppgtt_unbind_vma(struct i915_vma *vma)
{
        vma->vm->clear_range(vma->vm, vma->node.start, vma->size);
}

static int ppgtt_set_pages(struct i915_vma *vma)
{
        GEM_BUG_ON(vma->pages);

        vma->pages = vma->obj->mm.pages;

        vma->page_sizes = vma->obj->mm.page_sizes;

        return 0;
}

static void clear_pages(struct i915_vma *vma)
{
        GEM_BUG_ON(!vma->pages);

        if (vma->pages != vma->obj->mm.pages) {
                sg_free_table(vma->pages);
                kfree(vma->pages);
        }
        vma->pages = NULL;

        memset(&vma->page_sizes, 0, sizeof(vma->page_sizes));
}

static gen8_pte_t gen8_pte_encode(dma_addr_t addr,
                                  enum i915_cache_level level)
{
        gen8_pte_t pte = _PAGE_PRESENT | _PAGE_RW;
        pte |= addr;

        switch (level) {
        case I915_CACHE_NONE:
                pte |= PPAT_UNCACHED;
                break;
        case I915_CACHE_WT:
                pte |= PPAT_DISPLAY_ELLC;
                break;
        default:
                pte |= PPAT_CACHED;
                break;
        }

        return pte;
}

static gen8_pde_t gen8_pde_encode(const dma_addr_t addr,
                                  const enum i915_cache_level level)
{
        gen8_pde_t pde = _PAGE_PRESENT | _PAGE_RW;
        pde |= addr;
        if (level != I915_CACHE_NONE)
                pde |= PPAT_CACHED_PDE;
        else
                pde |= PPAT_UNCACHED;
        return pde;
}

#define gen8_pdpe_encode gen8_pde_encode
#define gen8_pml4e_encode gen8_pde_encode

static gen6_pte_t snb_pte_encode(dma_addr_t addr,
                                 enum i915_cache_level level,
                                 u32 unused)
{
        gen6_pte_t pte = GEN6_PTE_VALID;
        pte |= GEN6_PTE_ADDR_ENCODE(addr);

        switch (level) {
        case I915_CACHE_L3_LLC:
        case I915_CACHE_LLC:
                pte |= GEN6_PTE_CACHE_LLC;
                break;
        case I915_CACHE_NONE:
                pte |= GEN6_PTE_UNCACHED;
                break;
        default:
                MISSING_CASE(level);
        }

        return pte;
}

static gen6_pte_t ivb_pte_encode(dma_addr_t addr,
                                 enum i915_cache_level level,
                                 u32 unused)
{
        gen6_pte_t pte = GEN6_PTE_VALID;
        pte |= GEN6_PTE_ADDR_ENCODE(addr);

        switch (level) {
        case I915_CACHE_L3_LLC:
                pte |= GEN7_PTE_CACHE_L3_LLC;
                break;
        case I915_CACHE_LLC:
                pte |= GEN6_PTE_CACHE_LLC;
                break;
        case I915_CACHE_NONE:
                pte |= GEN6_PTE_UNCACHED;
                break;
        default:
                MISSING_CASE(level);
        }

        return pte;
}

static gen6_pte_t byt_pte_encode(dma_addr_t addr,
                                 enum i915_cache_level level,
                                 u32 flags)
{
        gen6_pte_t pte = GEN6_PTE_VALID;
        pte |= GEN6_PTE_ADDR_ENCODE(addr);

        if (!(flags & PTE_READ_ONLY))
                pte |= BYT_PTE_WRITEABLE;

        if (level != I915_CACHE_NONE)
                pte |= BYT_PTE_SNOOPED_BY_CPU_CACHES;

        return pte;
}

static gen6_pte_t hsw_pte_encode(dma_addr_t addr,
                                 enum i915_cache_level level,
                                 u32 unused)
{
        gen6_pte_t pte = GEN6_PTE_VALID;
        pte |= HSW_PTE_ADDR_ENCODE(addr);

        if (level != I915_CACHE_NONE)
                pte |= HSW_WB_LLC_AGE3;

        return pte;
}

static gen6_pte_t iris_pte_encode(dma_addr_t addr,
                                  enum i915_cache_level level,
                                  u32 unused)
{
        gen6_pte_t pte = GEN6_PTE_VALID;
        pte |= HSW_PTE_ADDR_ENCODE(addr);

        switch (level) {
        case I915_CACHE_NONE:
                break;
        case I915_CACHE_WT:
                pte |= HSW_WT_ELLC_LLC_AGE3;
                break;
        default:
                pte |= HSW_WB_ELLC_LLC_AGE3;
                break;
        }

        return pte;
}

static struct page *vm_alloc_page(struct i915_address_space *vm, gfp_t gfp)
{
        struct pagevec *pvec = &vm->free_pages;
        struct pagevec stash;

        if (I915_SELFTEST_ONLY(should_fail(&vm->fault_attr, 1)))
                i915_gem_shrink_all(vm->i915);

        if (likely(pvec->nr))
                return pvec->pages[--pvec->nr];

        if (!vm->pt_kmap_wc)
                return alloc_page(gfp);

        /* A placeholder for a specific mutex to guard the WC stash */
        lockdep_assert_held(&vm->i915->drm.struct_mutex);

        /* Look in our global stash of WC pages... */
        pvec = &vm->i915->mm.wc_stash;
        if (likely(pvec->nr))
                return pvec->pages[--pvec->nr];

        /*
         * Otherwise batch allocate pages to amoritize cost of set_pages_wc.
         *
         * We have to be careful as page allocation may trigger the shrinker
         * (via direct reclaim) which will fill up the WC stash underneath us.
         * So we add our WB pages into a temporary pvec on the stack and merge
         * them into the WC stash after all the allocations are complete.
         */
        pagevec_init(&stash);
        do {
                struct page *page;

                page = alloc_page(gfp);
                if (unlikely(!page))
                        break;

                stash.pages[stash.nr++] = page;
        } while (stash.nr < pagevec_space(pvec));

        if (stash.nr) {
                int nr = min_t(int, stash.nr, pagevec_space(pvec));
                struct page **pages = stash.pages + stash.nr - nr;

                if (nr && !set_pages_array_wc(pages, nr)) {
                        memcpy(pvec->pages + pvec->nr,
                               pages, sizeof(pages[0]) * nr);
                        pvec->nr += nr;
                        stash.nr -= nr;
                }

                pagevec_release(&stash);
        }

        return likely(pvec->nr) ? pvec->pages[--pvec->nr] : NULL;
}

static void vm_free_pages_release(struct i915_address_space *vm,
                                  bool immediate)
{
        struct pagevec *pvec = &vm->free_pages;

        GEM_BUG_ON(!pagevec_count(pvec));

        if (vm->pt_kmap_wc) {
                struct pagevec *stash = &vm->i915->mm.wc_stash;

                /* When we use WC, first fill up the global stash and then
                 * only if full immediately free the overflow.
                 */

                lockdep_assert_held(&vm->i915->drm.struct_mutex);
                if (pagevec_space(stash)) {
                        do {
                                stash->pages[stash->nr++] =
                                        pvec->pages[--pvec->nr];
                                if (!pvec->nr)
                                        return;
                        } while (pagevec_space(stash));

                        /* As we have made some room in the VM's free_pages,
                         * we can wait for it to fill again. Unless we are
                         * inside i915_address_space_fini() and must
                         * immediately release the pages!
                         */
                        if (!immediate)
                                return;
                }

                set_pages_array_wb(pvec->pages, pvec->nr);
        }

        __pagevec_release(pvec);
}

static void vm_free_page(struct i915_address_space *vm, struct page *page)
{
        /*
         * On !llc, we need to change the pages back to WB. We only do so
         * in bulk, so we rarely need to change the page attributes here,
         * but doing so requires a stop_machine() from deep inside arch/x86/mm.
         * To make detection of the possible sleep more likely, use an
         * unconditional might_sleep() for everybody.
         */
        might_sleep();
        if (!pagevec_add(&vm->free_pages, page))
                vm_free_pages_release(vm, false);
}

static int __setup_page_dma(struct i915_address_space *vm,
                            struct i915_page_dma *p,
                            gfp_t gfp)
{
        p->page = vm_alloc_page(vm, gfp | __GFP_NOWARN | __GFP_NORETRY);
        if (unlikely(!p->page))
                return -ENOMEM;

        p->daddr = dma_map_page(vm->dma, p->page, 0, PAGE_SIZE,
                                PCI_DMA_BIDIRECTIONAL);
        if (unlikely(dma_mapping_error(vm->dma, p->daddr))) {
                vm_free_page(vm, p->page);
                return -ENOMEM;
        }

        return 0;
}

static int setup_page_dma(struct i915_address_space *vm,
                          struct i915_page_dma *p)
{
        return __setup_page_dma(vm, p, I915_GFP_DMA);
}

static void cleanup_page_dma(struct i915_address_space *vm,
                             struct i915_page_dma *p)
{
        dma_unmap_page(vm->dma, p->daddr, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
        vm_free_page(vm, p->page);
}

#define kmap_atomic_px(px) kmap_atomic(px_base(px)->page)

#define setup_px(vm, px) setup_page_dma((vm), px_base(px))
#define cleanup_px(vm, px) cleanup_page_dma((vm), px_base(px))
#define fill_px(ppgtt, px, v) fill_page_dma((vm), px_base(px), (v))
#define fill32_px(ppgtt, px, v) fill_page_dma_32((vm), px_base(px), (v))

static void fill_page_dma(struct i915_address_space *vm,
                          struct i915_page_dma *p,
                          const u64 val)
{
        u64 * const vaddr = kmap_atomic(p->page);

        memset64(vaddr, val, PAGE_SIZE / sizeof(val));

        kunmap_atomic(vaddr);
}

static void fill_page_dma_32(struct i915_address_space *vm,
                             struct i915_page_dma *p,
                             const u32 v)
{
        fill_page_dma(vm, p, (u64)v << 32 | v);
}

static int
setup_scratch_page(struct i915_address_space *vm, gfp_t gfp)
{
        unsigned long size;

        /*
         * In order to utilize 64K pages for an object with a size < 2M, we will
         * need to support a 64K scratch page, given that every 16th entry for a
         * page-table operating in 64K mode must point to a properly aligned 64K
         * region, including any PTEs which happen to point to scratch.
         *
         * This is only relevant for the 48b PPGTT where we support
         * huge-gtt-pages, see also i915_vma_insert().
         *
         * TODO: we should really consider write-protecting the scratch-page and
         * sharing between ppgtt
         */
        size = I915_GTT_PAGE_SIZE_4K;
        if (i915_vm_is_48bit(vm) &&
            HAS_PAGE_SIZES(vm->i915, I915_GTT_PAGE_SIZE_64K)) {
                size = I915_GTT_PAGE_SIZE_64K;
                gfp |= __GFP_NOWARN;
        }
        gfp |= __GFP_ZERO | __GFP_RETRY_MAYFAIL;

        do {
                int order = get_order(size);
                struct page *page;
                dma_addr_t addr;

                page = alloc_pages(gfp, order);
                if (unlikely(!page))
                        goto skip;

                addr = dma_map_page(vm->dma, page, 0, size,
                                    PCI_DMA_BIDIRECTIONAL);
                if (unlikely(dma_mapping_error(vm->dma, addr)))
                        goto free_page;

                if (unlikely(!IS_ALIGNED(addr, size)))
                        goto unmap_page;

                vm->scratch_page.page = page;
                vm->scratch_page.daddr = addr;
                vm->scratch_page.order = order;
                return 0;

unmap_page:
                dma_unmap_page(vm->dma, addr, size, PCI_DMA_BIDIRECTIONAL);
free_page:
                __free_pages(page, order);
skip:
                if (size == I915_GTT_PAGE_SIZE_4K)
                        return -ENOMEM;

                size = I915_GTT_PAGE_SIZE_4K;
                gfp &= ~__GFP_NOWARN;
        } while (1);
}

static void cleanup_scratch_page(struct i915_address_space *vm)
{
        struct i915_page_dma *p = &vm->scratch_page;

        dma_unmap_page(vm->dma, p->daddr, BIT(p->order) << PAGE_SHIFT,
                       PCI_DMA_BIDIRECTIONAL);
        __free_pages(p->page, p->order);
}

static struct i915_page_table *alloc_pt(struct i915_address_space *vm)
{
        struct i915_page_table *pt;

        pt = kmalloc(sizeof(*pt), GFP_KERNEL | __GFP_NOWARN);
        if (unlikely(!pt))
                return ERR_PTR(-ENOMEM);

        if (unlikely(setup_px(vm, pt))) {
                kfree(pt);
                return ERR_PTR(-ENOMEM);
        }

        pt->used_ptes = 0;
        return pt;
}

static void free_pt(struct i915_address_space *vm, struct i915_page_table *pt)
{
        cleanup_px(vm, pt);
        kfree(pt);
}

static void gen8_initialize_pt(struct i915_address_space *vm,
                               struct i915_page_table *pt)
{
        fill_px(vm, pt,
                gen8_pte_encode(vm->scratch_page.daddr, I915_CACHE_LLC));
}

static void gen6_initialize_pt(struct i915_address_space *vm,
                               struct i915_page_table *pt)
{
        fill32_px(vm, pt,
                  vm->pte_encode(vm->scratch_page.daddr, I915_CACHE_LLC, 0));
}

static struct i915_page_directory *alloc_pd(struct i915_address_space *vm)
{
        struct i915_page_directory *pd;

        pd = kzalloc(sizeof(*pd), GFP_KERNEL | __GFP_NOWARN);
        if (unlikely(!pd))
                return ERR_PTR(-ENOMEM);

        if (unlikely(setup_px(vm, pd))) {
                kfree(pd);
                return ERR_PTR(-ENOMEM);
        }

        pd->used_pdes = 0;
        return pd;
}

static void free_pd(struct i915_address_space *vm,
                    struct i915_page_directory *pd)
{
        cleanup_px(vm, pd);
        kfree(pd);
}

static void gen8_initialize_pd(struct i915_address_space *vm,
                               struct i915_page_directory *pd)
{
        fill_px(vm, pd,
                gen8_pde_encode(px_dma(vm->scratch_pt), I915_CACHE_LLC));
        memset_p((void **)pd->page_table, vm->scratch_pt, I915_PDES);
}

static int __pdp_init(struct i915_address_space *vm,
                      struct i915_page_directory_pointer *pdp)
{
        const unsigned int pdpes = i915_pdpes_per_pdp(vm);

        pdp->page_directory = kmalloc_array(pdpes, sizeof(*pdp->page_directory),
                                            GFP_KERNEL | __GFP_NOWARN);
        if (unlikely(!pdp->page_directory))
                return -ENOMEM;

        memset_p((void **)pdp->page_directory, vm->scratch_pd, pdpes);

        return 0;
}

static void __pdp_fini(struct i915_page_directory_pointer *pdp)
{
        kfree(pdp->page_directory);
        pdp->page_directory = NULL;
}

static inline bool use_4lvl(const struct i915_address_space *vm)
{
        return i915_vm_is_48bit(vm);
}

static struct i915_page_directory_pointer *
alloc_pdp(struct i915_address_space *vm)
{
        struct i915_page_directory_pointer *pdp;
        int ret = -ENOMEM;

        GEM_BUG_ON(!use_4lvl(vm));

        pdp = kzalloc(sizeof(*pdp), GFP_KERNEL);
        if (!pdp)
                return ERR_PTR(-ENOMEM);

        ret = __pdp_init(vm, pdp);
        if (ret)
                goto fail_bitmap;

        ret = setup_px(vm, pdp);
        if (ret)
                goto fail_page_m;

        return pdp;

fail_page_m:
        __pdp_fini(pdp);
fail_bitmap:
        kfree(pdp);

        return ERR_PTR(ret);
}

static void free_pdp(struct i915_address_space *vm,
                     struct i915_page_directory_pointer *pdp)
{
        __pdp_fini(pdp);

        if (!use_4lvl(vm))
                return;

        cleanup_px(vm, pdp);
        kfree(pdp);
}

static void gen8_initialize_pdp(struct i915_address_space *vm,
                                struct i915_page_directory_pointer *pdp)
{
        gen8_ppgtt_pdpe_t scratch_pdpe;

        scratch_pdpe = gen8_pdpe_encode(px_dma(vm->scratch_pd), I915_CACHE_LLC);

        fill_px(vm, pdp, scratch_pdpe);
}

static void gen8_initialize_pml4(struct i915_address_space *vm,
                                 struct i915_pml4 *pml4)
{
        fill_px(vm, pml4,
                gen8_pml4e_encode(px_dma(vm->scratch_pdp), I915_CACHE_LLC));
        memset_p((void **)pml4->pdps, vm->scratch_pdp, GEN8_PML4ES_PER_PML4);
}

/* Broadwell Page Directory Pointer Descriptors */
static int gen8_write_pdp(struct i915_request *rq,
                          unsigned entry,
                          dma_addr_t addr)
{
        struct intel_engine_cs *engine = rq->engine;
        u32 *cs;

        BUG_ON(entry >= 4);

        cs = intel_ring_begin(rq, 6);
        if (IS_ERR(cs))
                return PTR_ERR(cs);

        *cs++ = MI_LOAD_REGISTER_IMM(1);
        *cs++ = i915_mmio_reg_offset(GEN8_RING_PDP_UDW(engine, entry));
        *cs++ = upper_32_bits(addr);
        *cs++ = MI_LOAD_REGISTER_IMM(1);
        *cs++ = i915_mmio_reg_offset(GEN8_RING_PDP_LDW(engine, entry));
        *cs++ = lower_32_bits(addr);
        intel_ring_advance(rq, cs);

        return 0;
}

static int gen8_mm_switch_3lvl(struct i915_hw_ppgtt *ppgtt,
                               struct i915_request *rq)
{
        int i, ret;

        for (i = GEN8_3LVL_PDPES - 1; i >= 0; i--) {
                const dma_addr_t pd_daddr = i915_page_dir_dma_addr(ppgtt, i);

                ret = gen8_write_pdp(rq, i, pd_daddr);
                if (ret)
                        return ret;
        }

        return 0;
}

static int gen8_mm_switch_4lvl(struct i915_hw_ppgtt *ppgtt,
                               struct i915_request *rq)
{
        return gen8_write_pdp(rq, 0, px_dma(&ppgtt->pml4));
}

/* PDE TLBs are a pain to invalidate on GEN8+. When we modify
 * the page table structures, we mark them dirty so that
 * context switching/execlist queuing code takes extra steps
 * to ensure that tlbs are flushed.
 */
static void mark_tlbs_dirty(struct i915_hw_ppgtt *ppgtt)
{
        ppgtt->pd_dirty_rings = INTEL_INFO(ppgtt->base.i915)->ring_mask;
}

/* Removes entries from a single page table, releasing it if it's empty.
 * Caller can use the return value to update higher-level entries.
 */
static bool gen8_ppgtt_clear_pt(struct i915_address_space *vm,
                                struct i915_page_table *pt,
                                u64 start, u64 length)
{
        unsigned int num_entries = gen8_pte_count(start, length);
        unsigned int pte = gen8_pte_index(start);
        unsigned int pte_end = pte + num_entries;
        const gen8_pte_t scratch_pte =
                gen8_pte_encode(vm->scratch_page.daddr, I915_CACHE_LLC);
        gen8_pte_t *vaddr;

        GEM_BUG_ON(num_entries > pt->used_ptes);

        pt->used_ptes -= num_entries;
        if (!pt->used_ptes)
                return true;

        vaddr = kmap_atomic_px(pt);
        while (pte < pte_end)
                vaddr[pte++] = scratch_pte;
        kunmap_atomic(vaddr);

        return false;
}

static void gen8_ppgtt_set_pde(struct i915_address_space *vm,
                               struct i915_page_directory *pd,
                               struct i915_page_table *pt,
                               unsigned int pde)
{
        gen8_pde_t *vaddr;

        pd->page_table[pde] = pt;

        vaddr = kmap_atomic_px(pd);
        vaddr[pde] = gen8_pde_encode(px_dma(pt), I915_CACHE_LLC);
        kunmap_atomic(vaddr);
}

static bool gen8_ppgtt_clear_pd(struct i915_address_space *vm,
                                struct i915_page_directory *pd,
                                u64 start, u64 length)
{
        struct i915_page_table *pt;
        u32 pde;

        gen8_for_each_pde(pt, pd, start, length, pde) {
                GEM_BUG_ON(pt == vm->scratch_pt);

                if (!gen8_ppgtt_clear_pt(vm, pt, start, length))
                        continue;

                gen8_ppgtt_set_pde(vm, pd, vm->scratch_pt, pde);
                GEM_BUG_ON(!pd->used_pdes);
                pd->used_pdes--;

                free_pt(vm, pt);
        }

        return !pd->used_pdes;
}

static void gen8_ppgtt_set_pdpe(struct i915_address_space *vm,
                                struct i915_page_directory_pointer *pdp,
                                struct i915_page_directory *pd,
                                unsigned int pdpe)
{
        gen8_ppgtt_pdpe_t *vaddr;

        pdp->page_directory[pdpe] = pd;
        if (!use_4lvl(vm))
                return;

        vaddr = kmap_atomic_px(pdp);
        vaddr[pdpe] = gen8_pdpe_encode(px_dma(pd), I915_CACHE_LLC);
        kunmap_atomic(vaddr);
}

/* Removes entries from a single page dir pointer, releasing it if it's empty.
 * Caller can use the return value to update higher-level entries
 */
static bool gen8_ppgtt_clear_pdp(struct i915_address_space *vm,
                                 struct i915_page_directory_pointer *pdp,
                                 u64 start, u64 length)
{
        struct i915_page_directory *pd;
        unsigned int pdpe;

        gen8_for_each_pdpe(pd, pdp, start, length, pdpe) {
                GEM_BUG_ON(pd == vm->scratch_pd);

                if (!gen8_ppgtt_clear_pd(vm, pd, start, length))
                        continue;

                gen8_ppgtt_set_pdpe(vm, pdp, vm->scratch_pd, pdpe);
                GEM_BUG_ON(!pdp->used_pdpes);
                pdp->used_pdpes--;

                free_pd(vm, pd);
        }

        return !pdp->used_pdpes;
}

static void gen8_ppgtt_clear_3lvl(struct i915_address_space *vm,
                                  u64 start, u64 length)
{
        gen8_ppgtt_clear_pdp(vm, &i915_vm_to_ppgtt(vm)->pdp, start, length);
}

static void gen8_ppgtt_set_pml4e(struct i915_pml4 *pml4,
                                 struct i915_page_directory_pointer *pdp,
                                 unsigned int pml4e)
{
        gen8_ppgtt_pml4e_t *vaddr;

        pml4->pdps[pml4e] = pdp;

        vaddr = kmap_atomic_px(pml4);
        vaddr[pml4e] = gen8_pml4e_encode(px_dma(pdp), I915_CACHE_LLC);
        kunmap_atomic(vaddr);
}

/* Removes entries from a single pml4.
 * This is the top-level structure in 4-level page tables used on gen8+.
 * Empty entries are always scratch pml4e.
 */
static void gen8_ppgtt_clear_4lvl(struct i915_address_space *vm,
                                  u64 start, u64 length)
{
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
        struct i915_pml4 *pml4 = &ppgtt->pml4;
        struct i915_page_directory_pointer *pdp;
        unsigned int pml4e;

        GEM_BUG_ON(!use_4lvl(vm));

        gen8_for_each_pml4e(pdp, pml4, start, length, pml4e) {
                GEM_BUG_ON(pdp == vm->scratch_pdp);

                if (!gen8_ppgtt_clear_pdp(vm, pdp, start, length))
                        continue;

                gen8_ppgtt_set_pml4e(pml4, vm->scratch_pdp, pml4e);

                free_pdp(vm, pdp);
        }
}

static inline struct sgt_dma {
        struct scatterlist *sg;
        dma_addr_t dma, max;
} sgt_dma(struct i915_vma *vma) {
        struct scatterlist *sg = vma->pages->sgl;
        dma_addr_t addr = sg_dma_address(sg);
        return (struct sgt_dma) { sg, addr, addr + sg->length };
}

struct gen8_insert_pte {
        u16 pml4e;
        u16 pdpe;
        u16 pde;
        u16 pte;
};

static __always_inline struct gen8_insert_pte gen8_insert_pte(u64 start)
{
        return (struct gen8_insert_pte) {
                 gen8_pml4e_index(start),
                 gen8_pdpe_index(start),
                 gen8_pde_index(start),
                 gen8_pte_index(start),
        };
}

static __always_inline bool
gen8_ppgtt_insert_pte_entries(struct i915_hw_ppgtt *ppgtt,
                              struct i915_page_directory_pointer *pdp,
                              struct sgt_dma *iter,
                              struct gen8_insert_pte *idx,
                              enum i915_cache_level cache_level)
{
        struct i915_page_directory *pd;
        const gen8_pte_t pte_encode = gen8_pte_encode(0, cache_level);
        gen8_pte_t *vaddr;
        bool ret;

        GEM_BUG_ON(idx->pdpe >= i915_pdpes_per_pdp(&ppgtt->base));
        pd = pdp->page_directory[idx->pdpe];
        vaddr = kmap_atomic_px(pd->page_table[idx->pde]);
        do {
                vaddr[idx->pte] = pte_encode | iter->dma;

                iter->dma += PAGE_SIZE;
                if (iter->dma >= iter->max) {
                        iter->sg = __sg_next(iter->sg);
                        if (!iter->sg) {
                                ret = false;
                                break;
                        }

                        iter->dma = sg_dma_address(iter->sg);
                        iter->max = iter->dma + iter->sg->length;
                }

                if (++idx->pte == GEN8_PTES) {
                        idx->pte = 0;

                        if (++idx->pde == I915_PDES) {
                                idx->pde = 0;

                                /* Limited by sg length for 3lvl */
                                if (++idx->pdpe == GEN8_PML4ES_PER_PML4) {
                                        idx->pdpe = 0;
                                        ret = true;
                                        break;
                                }

                                GEM_BUG_ON(idx->pdpe >= i915_pdpes_per_pdp(&ppgtt->base));
                                pd = pdp->page_directory[idx->pdpe];
                        }

                        kunmap_atomic(vaddr);
                        vaddr = kmap_atomic_px(pd->page_table[idx->pde]);
                }
        } while (1);
        kunmap_atomic(vaddr);

        return ret;
}

static void gen8_ppgtt_insert_3lvl(struct i915_address_space *vm,
                                   struct i915_vma *vma,
                                   enum i915_cache_level cache_level,
                                   u32 unused)
{
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
        struct sgt_dma iter = sgt_dma(vma);
        struct gen8_insert_pte idx = gen8_insert_pte(vma->node.start);

        gen8_ppgtt_insert_pte_entries(ppgtt, &ppgtt->pdp, &iter, &idx,
                                      cache_level);

        vma->page_sizes.gtt = I915_GTT_PAGE_SIZE;
}

static void gen8_ppgtt_insert_huge_entries(struct i915_vma *vma,
                                           struct i915_page_directory_pointer **pdps,
                                           struct sgt_dma *iter,
                                           enum i915_cache_level cache_level)
{
        const gen8_pte_t pte_encode = gen8_pte_encode(0, cache_level);
        u64 start = vma->node.start;
        dma_addr_t rem = iter->sg->length;

        do {
                struct gen8_insert_pte idx = gen8_insert_pte(start);
                struct i915_page_directory_pointer *pdp = pdps[idx.pml4e];
                struct i915_page_directory *pd = pdp->page_directory[idx.pdpe];
                unsigned int page_size;
                bool maybe_64K = false;
                gen8_pte_t encode = pte_encode;
                gen8_pte_t *vaddr;
                u16 index, max;

                if (vma->page_sizes.sg & I915_GTT_PAGE_SIZE_2M &&
                    IS_ALIGNED(iter->dma, I915_GTT_PAGE_SIZE_2M) &&
                    rem >= I915_GTT_PAGE_SIZE_2M && !idx.pte) {
                        index = idx.pde;
                        max = I915_PDES;
                        page_size = I915_GTT_PAGE_SIZE_2M;

                        encode |= GEN8_PDE_PS_2M;

                        vaddr = kmap_atomic_px(pd);
                } else {
                        struct i915_page_table *pt = pd->page_table[idx.pde];

                        index = idx.pte;
                        max = GEN8_PTES;
                        page_size = I915_GTT_PAGE_SIZE;

                        if (!index &&
                            vma->page_sizes.sg & I915_GTT_PAGE_SIZE_64K &&
                            IS_ALIGNED(iter->dma, I915_GTT_PAGE_SIZE_64K) &&
                            (IS_ALIGNED(rem, I915_GTT_PAGE_SIZE_64K) ||
                             rem >= (max - index) << PAGE_SHIFT))
                                maybe_64K = true;

                        vaddr = kmap_atomic_px(pt);
                }

                do {
                        GEM_BUG_ON(iter->sg->length < page_size);
                        vaddr[index++] = encode | iter->dma;

                        start += page_size;
                        iter->dma += page_size;
                        rem -= page_size;
                        if (iter->dma >= iter->max) {
                                iter->sg = __sg_next(iter->sg);
                                if (!iter->sg)
                                        break;

                                rem = iter->sg->length;
                                iter->dma = sg_dma_address(iter->sg);
                                iter->max = iter->dma + rem;

                                if (maybe_64K && index < max &&
                                    !(IS_ALIGNED(iter->dma, I915_GTT_PAGE_SIZE_64K) &&
                                      (IS_ALIGNED(rem, I915_GTT_PAGE_SIZE_64K) ||
                                       rem >= (max - index) << PAGE_SHIFT)))
                                        maybe_64K = false;

                                if (unlikely(!IS_ALIGNED(iter->dma, page_size)))
                                        break;
                        }
                } while (rem >= page_size && index < max);

                kunmap_atomic(vaddr);

                /*
                 * Is it safe to mark the 2M block as 64K? -- Either we have
                 * filled whole page-table with 64K entries, or filled part of
                 * it and have reached the end of the sg table and we have
                 * enough padding.
                 */
                if (maybe_64K &&
                    (index == max ||
                     (i915_vm_has_scratch_64K(vma->vm) &&
                      !iter->sg && IS_ALIGNED(vma->node.start +
                                              vma->node.size,
                                              I915_GTT_PAGE_SIZE_2M)))) {
                        vaddr = kmap_atomic_px(pd);
                        vaddr[idx.pde] |= GEN8_PDE_IPS_64K;
                        kunmap_atomic(vaddr);
                        page_size = I915_GTT_PAGE_SIZE_64K;
                }

                vma->page_sizes.gtt |= page_size;
        } while (iter->sg);
}

static void gen8_ppgtt_insert_4lvl(struct i915_address_space *vm,
                                   struct i915_vma *vma,
                                   enum i915_cache_level cache_level,
                                   u32 unused)
{
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
        struct sgt_dma iter = sgt_dma(vma);
        struct i915_page_directory_pointer **pdps = ppgtt->pml4.pdps;

        if (vma->page_sizes.sg > I915_GTT_PAGE_SIZE) {
                gen8_ppgtt_insert_huge_entries(vma, pdps, &iter, cache_level);
        } else {
                struct gen8_insert_pte idx = gen8_insert_pte(vma->node.start);

                while (gen8_ppgtt_insert_pte_entries(ppgtt, pdps[idx.pml4e++],
                                                     &iter, &idx, cache_level))
                        GEM_BUG_ON(idx.pml4e >= GEN8_PML4ES_PER_PML4);

                vma->page_sizes.gtt = I915_GTT_PAGE_SIZE;
        }
}

static void gen8_free_page_tables(struct i915_address_space *vm,
                                  struct i915_page_directory *pd)
{
        int i;

        if (!px_page(pd))
                return;

        for (i = 0; i < I915_PDES; i++) {
                if (pd->page_table[i] != vm->scratch_pt)
                        free_pt(vm, pd->page_table[i]);
        }
}

static int gen8_init_scratch(struct i915_address_space *vm)
{
        int ret;

        ret = setup_scratch_page(vm, I915_GFP_DMA);
        if (ret)
                return ret;

        vm->scratch_pt = alloc_pt(vm);
        if (IS_ERR(vm->scratch_pt)) {
                ret = PTR_ERR(vm->scratch_pt);
                goto free_scratch_page;
        }

        vm->scratch_pd = alloc_pd(vm);
        if (IS_ERR(vm->scratch_pd)) {
                ret = PTR_ERR(vm->scratch_pd);
                goto free_pt;
        }

        if (use_4lvl(vm)) {
                vm->scratch_pdp = alloc_pdp(vm);
                if (IS_ERR(vm->scratch_pdp)) {
                        ret = PTR_ERR(vm->scratch_pdp);
                        goto free_pd;
                }
        }

        gen8_initialize_pt(vm, vm->scratch_pt);
        gen8_initialize_pd(vm, vm->scratch_pd);
        if (use_4lvl(vm))
                gen8_initialize_pdp(vm, vm->scratch_pdp);

        return 0;

free_pd:
        free_pd(vm, vm->scratch_pd);
free_pt:
        free_pt(vm, vm->scratch_pt);
free_scratch_page:
        cleanup_scratch_page(vm);

        return ret;
}

static int gen8_ppgtt_notify_vgt(struct i915_hw_ppgtt *ppgtt, bool create)
{
        struct i915_address_space *vm = &ppgtt->base;
        struct drm_i915_private *dev_priv = vm->i915;
        enum vgt_g2v_type msg;
        int i;

        if (use_4lvl(vm)) {
                const u64 daddr = px_dma(&ppgtt->pml4);

                I915_WRITE(vgtif_reg(pdp[0].lo), lower_32_bits(daddr));
                I915_WRITE(vgtif_reg(pdp[0].hi), upper_32_bits(daddr));

                msg = (create ? VGT_G2V_PPGTT_L4_PAGE_TABLE_CREATE :
                                VGT_G2V_PPGTT_L4_PAGE_TABLE_DESTROY);
        } else {
                for (i = 0; i < GEN8_3LVL_PDPES; i++) {
                        const u64 daddr = i915_page_dir_dma_addr(ppgtt, i);

                        I915_WRITE(vgtif_reg(pdp[i].lo), lower_32_bits(daddr));
                        I915_WRITE(vgtif_reg(pdp[i].hi), upper_32_bits(daddr));
                }

                msg = (create ? VGT_G2V_PPGTT_L3_PAGE_TABLE_CREATE :
                                VGT_G2V_PPGTT_L3_PAGE_TABLE_DESTROY);
        }

        I915_WRITE(vgtif_reg(g2v_notify), msg);

        return 0;
}

static void gen8_free_scratch(struct i915_address_space *vm)
{
        if (use_4lvl(vm))
                free_pdp(vm, vm->scratch_pdp);
        free_pd(vm, vm->scratch_pd);
        free_pt(vm, vm->scratch_pt);
        cleanup_scratch_page(vm);
}

static void gen8_ppgtt_cleanup_3lvl(struct i915_address_space *vm,
                                    struct i915_page_directory_pointer *pdp)
{
        const unsigned int pdpes = i915_pdpes_per_pdp(vm);
        int i;

        for (i = 0; i < pdpes; i++) {
                if (pdp->page_directory[i] == vm->scratch_pd)
                        continue;

                gen8_free_page_tables(vm, pdp->page_directory[i]);
                free_pd(vm, pdp->page_directory[i]);
        }

        free_pdp(vm, pdp);
}

static void gen8_ppgtt_cleanup_4lvl(struct i915_hw_ppgtt *ppgtt)
{
        int i;

        for (i = 0; i < GEN8_PML4ES_PER_PML4; i++) {
                if (ppgtt->pml4.pdps[i] == ppgtt->base.scratch_pdp)
                        continue;

                gen8_ppgtt_cleanup_3lvl(&ppgtt->base, ppgtt->pml4.pdps[i]);
        }

        cleanup_px(&ppgtt->base, &ppgtt->pml4);
}

static void gen8_ppgtt_cleanup(struct i915_address_space *vm)
{
        struct drm_i915_private *dev_priv = vm->i915;
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);

        if (intel_vgpu_active(dev_priv))
                gen8_ppgtt_notify_vgt(ppgtt, false);

        if (use_4lvl(vm))
                gen8_ppgtt_cleanup_4lvl(ppgtt);
        else
                gen8_ppgtt_cleanup_3lvl(&ppgtt->base, &ppgtt->pdp);

        gen8_free_scratch(vm);
}

static int gen8_ppgtt_alloc_pd(struct i915_address_space *vm,
                               struct i915_page_directory *pd,
                               u64 start, u64 length)
{
        struct i915_page_table *pt;
        u64 from = start;
        unsigned int pde;

        gen8_for_each_pde(pt, pd, start, length, pde) {
                int count = gen8_pte_count(start, length);

                if (pt == vm->scratch_pt) {
                        pd->used_pdes++;

                        pt = alloc_pt(vm);
                        if (IS_ERR(pt)) {
                                pd->used_pdes--;
                                goto unwind;
                        }

                        if (count < GEN8_PTES || intel_vgpu_active(vm->i915))
                                gen8_initialize_pt(vm, pt);

                        gen8_ppgtt_set_pde(vm, pd, pt, pde);
                        GEM_BUG_ON(pd->used_pdes > I915_PDES);
                }

                pt->used_ptes += count;
        }
        return 0;

unwind:
        gen8_ppgtt_clear_pd(vm, pd, from, start - from);
        return -ENOMEM;
}

static int gen8_ppgtt_alloc_pdp(struct i915_address_space *vm,
                                struct i915_page_directory_pointer *pdp,
                                u64 start, u64 length)
{
        struct i915_page_directory *pd;
        u64 from = start;
        unsigned int pdpe;
        int ret;

        gen8_for_each_pdpe(pd, pdp, start, length, pdpe) {
                if (pd == vm->scratch_pd) {
                        pdp->used_pdpes++;

                        pd = alloc_pd(vm);
                        if (IS_ERR(pd)) {
                                pdp->used_pdpes--;
                                goto unwind;
                        }

                        gen8_initialize_pd(vm, pd);
                        gen8_ppgtt_set_pdpe(vm, pdp, pd, pdpe);
                        GEM_BUG_ON(pdp->used_pdpes > i915_pdpes_per_pdp(vm));

                        mark_tlbs_dirty(i915_vm_to_ppgtt(vm));
                }

                ret = gen8_ppgtt_alloc_pd(vm, pd, start, length);
                if (unlikely(ret))
                        goto unwind_pd;
        }

        return 0;

unwind_pd:
        if (!pd->used_pdes) {
                gen8_ppgtt_set_pdpe(vm, pdp, vm->scratch_pd, pdpe);
                GEM_BUG_ON(!pdp->used_pdpes);
                pdp->used_pdpes--;
                free_pd(vm, pd);
        }
unwind:
        gen8_ppgtt_clear_pdp(vm, pdp, from, start - from);
        return -ENOMEM;
}

static int gen8_ppgtt_alloc_3lvl(struct i915_address_space *vm,
                                 u64 start, u64 length)
{
        return gen8_ppgtt_alloc_pdp(vm,
                                    &i915_vm_to_ppgtt(vm)->pdp, start, length);
}

static int gen8_ppgtt_alloc_4lvl(struct i915_address_space *vm,
                                 u64 start, u64 length)
{
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
        struct i915_pml4 *pml4 = &ppgtt->pml4;
        struct i915_page_directory_pointer *pdp;
        u64 from = start;
        u32 pml4e;
        int ret;

        gen8_for_each_pml4e(pdp, pml4, start, length, pml4e) {
                if (pml4->pdps[pml4e] == vm->scratch_pdp) {
                        pdp = alloc_pdp(vm);
                        if (IS_ERR(pdp))
                                goto unwind;

                        gen8_initialize_pdp(vm, pdp);
                        gen8_ppgtt_set_pml4e(pml4, pdp, pml4e);
                }

                ret = gen8_ppgtt_alloc_pdp(vm, pdp, start, length);
                if (unlikely(ret))
                        goto unwind_pdp;
        }

        return 0;

unwind_pdp:
        if (!pdp->used_pdpes) {
                gen8_ppgtt_set_pml4e(pml4, vm->scratch_pdp, pml4e);
                free_pdp(vm, pdp);
        }
unwind:
        gen8_ppgtt_clear_4lvl(vm, from, start - from);
        return -ENOMEM;
}

static void gen8_dump_pdp(struct i915_hw_ppgtt *ppgtt,
                          struct i915_page_directory_pointer *pdp,
                          u64 start, u64 length,
                          gen8_pte_t scratch_pte,
                          struct seq_file *m)
{
        struct i915_address_space *vm = &ppgtt->base;
        struct i915_page_directory *pd;
        u32 pdpe;

        gen8_for_each_pdpe(pd, pdp, start, length, pdpe) {
                struct i915_page_table *pt;
                u64 pd_len = length;
                u64 pd_start = start;
                u32 pde;

                if (pdp->page_directory[pdpe] == ppgtt->base.scratch_pd)
                        continue;

                seq_printf(m, "\tPDPE #%d\n", pdpe);
                gen8_for_each_pde(pt, pd, pd_start, pd_len, pde) {
                        u32 pte;
                        gen8_pte_t *pt_vaddr;

                        if (pd->page_table[pde] == ppgtt->base.scratch_pt)
                                continue;

                        pt_vaddr = kmap_atomic_px(pt);
                        for (pte = 0; pte < GEN8_PTES; pte += 4) {
                                u64 va = (pdpe << GEN8_PDPE_SHIFT |
                                          pde << GEN8_PDE_SHIFT |
                                          pte << GEN8_PTE_SHIFT);
                                int i;
                                bool found = false;

                                for (i = 0; i < 4; i++)
                                        if (pt_vaddr[pte + i] != scratch_pte)
                                                found = true;
                                if (!found)
                                        continue;

                                seq_printf(m, "\t\t0x%llx [%03d,%03d,%04d]: =", va, pdpe, pde, pte);
                                for (i = 0; i < 4; i++) {
                                        if (pt_vaddr[pte + i] != scratch_pte)
                                                seq_printf(m, " %llx", pt_vaddr[pte + i]);
                                        else
                                                seq_puts(m, "  SCRATCH ");
                                }
                                seq_puts(m, "\n");
                        }
                        kunmap_atomic(pt_vaddr);
                }
        }
}

static void gen8_dump_ppgtt(struct i915_hw_ppgtt *ppgtt, struct seq_file *m)
{
        struct i915_address_space *vm = &ppgtt->base;
        const gen8_pte_t scratch_pte =
                gen8_pte_encode(vm->scratch_page.daddr, I915_CACHE_LLC);
        u64 start = 0, length = ppgtt->base.total;

        if (use_4lvl(vm)) {
                u64 pml4e;
                struct i915_pml4 *pml4 = &ppgtt->pml4;
                struct i915_page_directory_pointer *pdp;

                gen8_for_each_pml4e(pdp, pml4, start, length, pml4e) {
                        if (pml4->pdps[pml4e] == ppgtt->base.scratch_pdp)
                                continue;

                        seq_printf(m, "    PML4E #%llu\n", pml4e);
                        gen8_dump_pdp(ppgtt, pdp, start, length, scratch_pte, m);
                }
        } else {
                gen8_dump_pdp(ppgtt, &ppgtt->pdp, start, length, scratch_pte, m);
        }
}

static int gen8_preallocate_top_level_pdp(struct i915_hw_ppgtt *ppgtt)
{
        struct i915_address_space *vm = &ppgtt->base;
        struct i915_page_directory_pointer *pdp = &ppgtt->pdp;
        struct i915_page_directory *pd;
        u64 start = 0, length = ppgtt->base.total;
        u64 from = start;
        unsigned int pdpe;

        gen8_for_each_pdpe(pd, pdp, start, length, pdpe) {
                pd = alloc_pd(vm);
                if (IS_ERR(pd))
                        goto unwind;

                gen8_initialize_pd(vm, pd);
                gen8_ppgtt_set_pdpe(vm, pdp, pd, pdpe);
                pdp->used_pdpes++;
        }

        pdp->used_pdpes++; /* never remove */
        return 0;

unwind:
        start -= from;
        gen8_for_each_pdpe(pd, pdp, from, start, pdpe) {
                gen8_ppgtt_set_pdpe(vm, pdp, vm->scratch_pd, pdpe);
                free_pd(vm, pd);
        }
        pdp->used_pdpes = 0;
        return -ENOMEM;
}

/*
 * GEN8 legacy ppgtt programming is accomplished through a max 4 PDP registers
 * with a net effect resembling a 2-level page table in normal x86 terms. Each
 * PDP represents 1GB of memory 4 * 512 * 512 * 4096 = 4GB legacy 32b address
 * space.
 *
 */
static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
{
        struct i915_address_space *vm = &ppgtt->base;
        struct drm_i915_private *dev_priv = vm->i915;
        int ret;

        ppgtt->base.total = USES_FULL_48BIT_PPGTT(dev_priv) ?
                1ULL << 48 :
                1ULL << 32;

        /* There are only few exceptions for gen >=6. chv and bxt.
         * And we are not sure about the latter so play safe for now.
         */
        if (IS_CHERRYVIEW(dev_priv) || IS_BROXTON(dev_priv))
                ppgtt->base.pt_kmap_wc = true;

        ret = gen8_init_scratch(&ppgtt->base);
        if (ret) {
                ppgtt->base.total = 0;
                return ret;
        }

        if (use_4lvl(vm)) {
                ret = setup_px(&ppgtt->base, &ppgtt->pml4);
                if (ret)
                        goto free_scratch;

                gen8_initialize_pml4(&ppgtt->base, &ppgtt->pml4);

                ppgtt->switch_mm = gen8_mm_switch_4lvl;
                ppgtt->base.allocate_va_range = gen8_ppgtt_alloc_4lvl;
                ppgtt->base.insert_entries = gen8_ppgtt_insert_4lvl;
                ppgtt->base.clear_range = gen8_ppgtt_clear_4lvl;
        } else {
                ret = __pdp_init(&ppgtt->base, &ppgtt->pdp);
                if (ret)
                        goto free_scratch;

                if (intel_vgpu_active(dev_priv)) {
                        ret = gen8_preallocate_top_level_pdp(ppgtt);
                        if (ret) {
                                __pdp_fini(&ppgtt->pdp);
                                goto free_scratch;
                        }
                }

                ppgtt->switch_mm = gen8_mm_switch_3lvl;
                ppgtt->base.allocate_va_range = gen8_ppgtt_alloc_3lvl;
                ppgtt->base.insert_entries = gen8_ppgtt_insert_3lvl;
                ppgtt->base.clear_range = gen8_ppgtt_clear_3lvl;
        }

        if (intel_vgpu_active(dev_priv))
                gen8_ppgtt_notify_vgt(ppgtt, true);

        ppgtt->base.cleanup = gen8_ppgtt_cleanup;
        ppgtt->base.unbind_vma = ppgtt_unbind_vma;
        ppgtt->base.bind_vma = ppgtt_bind_vma;
        ppgtt->base.set_pages = ppgtt_set_pages;
        ppgtt->base.clear_pages = clear_pages;
        ppgtt->debug_dump = gen8_dump_ppgtt;

        return 0;

free_scratch:
        gen8_free_scratch(&ppgtt->base);
        return ret;
}

static void gen6_dump_ppgtt(struct i915_hw_ppgtt *ppgtt, struct seq_file *m)
{
        struct i915_address_space *vm = &ppgtt->base;
        struct i915_page_table *unused;
        gen6_pte_t scratch_pte;
        u32 pd_entry, pte, pde;
        u32 start = 0, length = ppgtt->base.total;

        scratch_pte = vm->pte_encode(vm->scratch_page.daddr,
                                     I915_CACHE_LLC, 0);

        gen6_for_each_pde(unused, &ppgtt->pd, start, length, pde) {
                u32 expected;
                gen6_pte_t *pt_vaddr;
                const dma_addr_t pt_addr = px_dma(ppgtt->pd.page_table[pde]);
                pd_entry = readl(ppgtt->pd_addr + pde);
                expected = (GEN6_PDE_ADDR_ENCODE(pt_addr) | GEN6_PDE_VALID);

                if (pd_entry != expected)
                        seq_printf(m, "\tPDE #%d mismatch: Actual PDE: %x Expected PDE: %x\n",
                                   pde,
                                   pd_entry,
                                   expected);
                seq_printf(m, "\tPDE: %x\n", pd_entry);

                pt_vaddr = kmap_atomic_px(ppgtt->pd.page_table[pde]);

                for (pte = 0; pte < GEN6_PTES; pte+=4) {
                        unsigned long va =
                                (pde * PAGE_SIZE * GEN6_PTES) +
                                (pte * PAGE_SIZE);
                        int i;
                        bool found = false;
                        for (i = 0; i < 4; i++)
                                if (pt_vaddr[pte + i] != scratch_pte)
                                        found = true;
                        if (!found)
                                continue;

                        seq_printf(m, "\t\t0x%lx [%03d,%04d]: =", va, pde, pte);
                        for (i = 0; i < 4; i++) {
                                if (pt_vaddr[pte + i] != scratch_pte)
                                        seq_printf(m, " %08x", pt_vaddr[pte + i]);
                                else
                                        seq_puts(m, "  SCRATCH ");
                        }
                        seq_puts(m, "\n");
                }
                kunmap_atomic(pt_vaddr);
        }
}

/* Write pde (index) from the page directory @pd to the page table @pt */
static inline void gen6_write_pde(const struct i915_hw_ppgtt *ppgtt,
                                  const unsigned int pde,
                                  const struct i915_page_table *pt)
{
        /* Caller needs to make sure the write completes if necessary */
        writel_relaxed(GEN6_PDE_ADDR_ENCODE(px_dma(pt)) | GEN6_PDE_VALID,
                       ppgtt->pd_addr + pde);
}

/* Write all the page tables found in the ppgtt structure to incrementing page
 * directories. */
static void gen6_write_page_range(struct i915_hw_ppgtt *ppgtt,
                                  u32 start, u32 length)
{
        struct i915_page_table *pt;
        unsigned int pde;

        gen6_for_each_pde(pt, &ppgtt->pd, start, length, pde)
                gen6_write_pde(ppgtt, pde, pt);

        mark_tlbs_dirty(ppgtt);
        wmb();
}

static inline u32 get_pd_offset(struct i915_hw_ppgtt *ppgtt)
{
        GEM_BUG_ON(ppgtt->pd.base.ggtt_offset & 0x3f);
        return ppgtt->pd.base.ggtt_offset << 10;
}

static int hsw_mm_switch(struct i915_hw_ppgtt *ppgtt,
                         struct i915_request *rq)
{
        struct intel_engine_cs *engine = rq->engine;
        u32 *cs;

        /* NB: TLBs must be flushed and invalidated before a switch */
        cs = intel_ring_begin(rq, 6);
        if (IS_ERR(cs))
                return PTR_ERR(cs);

        *cs++ = MI_LOAD_REGISTER_IMM(2);
        *cs++ = i915_mmio_reg_offset(RING_PP_DIR_DCLV(engine));
        *cs++ = PP_DIR_DCLV_2G;
        *cs++ = i915_mmio_reg_offset(RING_PP_DIR_BASE(engine));
        *cs++ = get_pd_offset(ppgtt);
        *cs++ = MI_NOOP;
        intel_ring_advance(rq, cs);

        return 0;
}

static int gen7_mm_switch(struct i915_hw_ppgtt *ppgtt,
                          struct i915_request *rq)
{
        struct intel_engine_cs *engine = rq->engine;
        u32 *cs;

        /* NB: TLBs must be flushed and invalidated before a switch */
        cs = intel_ring_begin(rq, 6);
        if (IS_ERR(cs))
                return PTR_ERR(cs);

        *cs++ = MI_LOAD_REGISTER_IMM(2);
        *cs++ = i915_mmio_reg_offset(RING_PP_DIR_DCLV(engine));
        *cs++ = PP_DIR_DCLV_2G;
        *cs++ = i915_mmio_reg_offset(RING_PP_DIR_BASE(engine));
        *cs++ = get_pd_offset(ppgtt);
        *cs++ = MI_NOOP;
        intel_ring_advance(rq, cs);

        return 0;
}

static int gen6_mm_switch(struct i915_hw_ppgtt *ppgtt,
                          struct i915_request *rq)
{
        struct intel_engine_cs *engine = rq->engine;
        struct drm_i915_private *dev_priv = rq->i915;

        I915_WRITE(RING_PP_DIR_DCLV(engine), PP_DIR_DCLV_2G);
        I915_WRITE(RING_PP_DIR_BASE(engine), get_pd_offset(ppgtt));
        return 0;
}

static void gen8_ppgtt_enable(struct drm_i915_private *dev_priv)
{
        struct intel_engine_cs *engine;
        enum intel_engine_id id;

        for_each_engine(engine, dev_priv, id) {
                u32 four_level = USES_FULL_48BIT_PPGTT(dev_priv) ?
                                 GEN8_GFX_PPGTT_48B : 0;
                I915_WRITE(RING_MODE_GEN7(engine),
                           _MASKED_BIT_ENABLE(GFX_PPGTT_ENABLE | four_level));
        }
}

static void gen7_ppgtt_enable(struct drm_i915_private *dev_priv)
{
        struct intel_engine_cs *engine;
        u32 ecochk, ecobits;
        enum intel_engine_id id;

        ecobits = I915_READ(GAC_ECO_BITS);
        I915_WRITE(GAC_ECO_BITS, ecobits | ECOBITS_PPGTT_CACHE64B);

        ecochk = I915_READ(GAM_ECOCHK);
        if (IS_HASWELL(dev_priv)) {
                ecochk |= ECOCHK_PPGTT_WB_HSW;
        } else {
                ecochk |= ECOCHK_PPGTT_LLC_IVB;
                ecochk &= ~ECOCHK_PPGTT_GFDT_IVB;
        }
        I915_WRITE(GAM_ECOCHK, ecochk);

        for_each_engine(engine, dev_priv, id) {
                /* GFX_MODE is per-ring on gen7+ */
                I915_WRITE(RING_MODE_GEN7(engine),
                           _MASKED_BIT_ENABLE(GFX_PPGTT_ENABLE));
        }
}

static void gen6_ppgtt_enable(struct drm_i915_private *dev_priv)
{
        u32 ecochk, gab_ctl, ecobits;

        ecobits = I915_READ(GAC_ECO_BITS);
        I915_WRITE(GAC_ECO_BITS, ecobits | ECOBITS_SNB_BIT |
                   ECOBITS_PPGTT_CACHE64B);

        gab_ctl = I915_READ(GAB_CTL);
        I915_WRITE(GAB_CTL, gab_ctl | GAB_CTL_CONT_AFTER_PAGEFAULT);

        ecochk = I915_READ(GAM_ECOCHK);
        I915_WRITE(GAM_ECOCHK, ecochk | ECOCHK_SNB_BIT | ECOCHK_PPGTT_CACHE64B);

        I915_WRITE(GFX_MODE, _MASKED_BIT_ENABLE(GFX_PPGTT_ENABLE));
}

/* PPGTT support for Sandybdrige/Gen6 and later */
static void gen6_ppgtt_clear_range(struct i915_address_space *vm,
                                   u64 start, u64 length)
{
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
        unsigned int first_entry = start >> PAGE_SHIFT;
        unsigned int pde = first_entry / GEN6_PTES;
        unsigned int pte = first_entry % GEN6_PTES;
        unsigned int num_entries = length >> PAGE_SHIFT;
        gen6_pte_t scratch_pte =
                vm->pte_encode(vm->scratch_page.daddr, I915_CACHE_LLC, 0);

        while (num_entries) {
                struct i915_page_table *pt = ppgtt->pd.page_table[pde++];
                unsigned int end = min(pte + num_entries, GEN6_PTES);
                gen6_pte_t *vaddr;

                num_entries -= end - pte;

                /* Note that the hw doesn't support removing PDE on the fly
                 * (they are cached inside the context with no means to
                 * invalidate the cache), so we can only reset the PTE
                 * entries back to scratch.
                 */

                vaddr = kmap_atomic_px(pt);
                do {
                        vaddr[pte++] = scratch_pte;
                } while (pte < end);
                kunmap_atomic(vaddr);

                pte = 0;
        }
}

static void gen6_ppgtt_insert_entries(struct i915_address_space *vm,
                                      struct i915_vma *vma,
                                      enum i915_cache_level cache_level,
                                      u32 flags)
{
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
        unsigned first_entry = vma->node.start >> PAGE_SHIFT;
        unsigned act_pt = first_entry / GEN6_PTES;
        unsigned act_pte = first_entry % GEN6_PTES;
        const u32 pte_encode = vm->pte_encode(0, cache_level, flags);
        struct sgt_dma iter = sgt_dma(vma);
        gen6_pte_t *vaddr;

        vaddr = kmap_atomic_px(ppgtt->pd.page_table[act_pt]);
        do {
                vaddr[act_pte] = pte_encode | GEN6_PTE_ADDR_ENCODE(iter.dma);

                iter.dma += PAGE_SIZE;
                if (iter.dma == iter.max) {
                        iter.sg = __sg_next(iter.sg);
                        if (!iter.sg)
                                break;

                        iter.dma = sg_dma_address(iter.sg);
                        iter.max = iter.dma + iter.sg->length;
                }

                if (++act_pte == GEN6_PTES) {
                        kunmap_atomic(vaddr);
                        vaddr = kmap_atomic_px(ppgtt->pd.page_table[++act_pt]);
                        act_pte = 0;
                }
        } while (1);
        kunmap_atomic(vaddr);

        vma->page_sizes.gtt = I915_GTT_PAGE_SIZE;
}

static int gen6_alloc_va_range(struct i915_address_space *vm,
                               u64 start, u64 length)
{
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
        struct i915_page_table *pt;
        u64 from = start;
        unsigned int pde;
        bool flush = false;

        gen6_for_each_pde(pt, &ppgtt->pd, start, length, pde) {
                if (pt == vm->scratch_pt) {
                        pt = alloc_pt(vm);
                        if (IS_ERR(pt))
                                goto unwind_out;

                        gen6_initialize_pt(vm, pt);
                        ppgtt->pd.page_table[pde] = pt;
                        gen6_write_pde(ppgtt, pde, pt);
                        flush = true;
                }
        }

        if (flush) {
                mark_tlbs_dirty(ppgtt);
                wmb();
        }

        return 0;

unwind_out:
        gen6_ppgtt_clear_range(vm, from, start);
        return -ENOMEM;
}

static int gen6_init_scratch(struct i915_address_space *vm)
{
        int ret;

        ret = setup_scratch_page(vm, I915_GFP_DMA);
        if (ret)
                return ret;

        vm->scratch_pt = alloc_pt(vm);
        if (IS_ERR(vm->scratch_pt)) {
                cleanup_scratch_page(vm);
                return PTR_ERR(vm->scratch_pt);
        }

        gen6_initialize_pt(vm, vm->scratch_pt);

        return 0;
}

static void gen6_free_scratch(struct i915_address_space *vm)
{
        free_pt(vm, vm->scratch_pt);
        cleanup_scratch_page(vm);
}

static void gen6_ppgtt_cleanup(struct i915_address_space *vm)
{
        struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
        struct i915_page_directory *pd = &ppgtt->pd;
        struct i915_page_table *pt;
        u32 pde;

        drm_mm_remove_node(&ppgtt->node);

        gen6_for_all_pdes(pt, pd, pde)
                if (pt != vm->scratch_pt)
                        free_pt(vm, pt);

        gen6_free_scratch(vm);
}

static int gen6_ppgtt_allocate_page_directories(struct i915_hw_ppgtt *ppgtt)
{
        struct i915_address_space *vm = &ppgtt->base;
        struct drm_i915_private *dev_priv = ppgtt->base.i915;
        struct i915_ggtt *ggtt = &dev_priv->ggtt;
        int ret;

        /* PPGTT PDEs reside in the GGTT and consists of 512 entries. The
         * allocator works in address space sizes, so it's multiplied by page

         * size. We allocate at the top of the GTT to avoid fragmentation.
         */
        BUG_ON(!drm_mm_initialized(&ggtt->base.mm));

        ret = gen6_init_scratch(vm);
        if (ret)
                return ret;

        ret = i915_gem_gtt_insert(&ggtt->base, &ppgtt->node,
                                  GEN6_PD_SIZE, GEN6_PD_ALIGN,
                                  I915_COLOR_UNEVICTABLE,
                                  0, ggtt->base.total,
                                  PIN_HIGH);
        if (ret)
                goto err_out;

        if (ppgtt->node.start < ggtt->mappable_end)
                DRM_DEBUG("Forced to use aperture for PDEs\n");

        ppgtt->pd.base.ggtt_offset =
                ppgtt->node.start / PAGE_SIZE * sizeof(gen6_pte_t);

        ppgtt->pd_addr = (gen6_pte_t __iomem *)ggtt->gsm +
                ppgtt->pd.base.ggtt_offset / sizeof(gen6_pte_t);

        return 0;

err_out:
        gen6_free_scratch(vm);
        return ret;
}

static int gen6_ppgtt_alloc(struct i915_hw_ppgtt *ppgtt)
{
        return gen6_ppgtt_allocate_page_directories(ppgtt);
}

static void gen6_scratch_va_range(struct i915_hw_ppgtt *ppgtt,
                                  u64 start, u64 length)
{
        struct i915_page_table *unused;
        u32 pde;

        gen6_for_each_pde(unused, &ppgtt->pd, start, length, pde)
                ppgtt->pd.page_table[pde] = ppgtt->base.scratch_pt;
}

static int gen6_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
{
        struct drm_i915_private *dev_priv = ppgtt->base.i915;
        struct i915_ggtt *ggtt = &dev_priv->ggtt;
        int ret;

        ppgtt->base.pte_encode = ggtt->base.pte_encode;
        if (intel_vgpu_active(dev_priv) || IS_GEN6(dev_priv))
                ppgtt->switch_mm = gen6_mm_switch;
        else if (IS_HASWELL(dev_priv))
                ppgtt->switch_mm = hsw_mm_switch;
        else if (IS_GEN7(dev_priv))
                ppgtt->switch_mm = gen7_mm_switch;
        else
                BUG();

        ret = gen6_ppgtt_alloc(ppgtt);
        if (ret)
                return ret;

        ppgtt->base.total = I915_PDES * GEN6_PTES * PAGE_SIZE;

        gen6_scratch_va_range(ppgtt, 0, ppgtt->base.total);
        gen6_write_page_range(ppgtt, 0, ppgtt->base.total);

        ret = gen6_alloc_va_range(&ppgtt->base, 0, ppgtt->base.total);
        if (ret) {
                gen6_ppgtt_cleanup(&ppgtt->base);
                return ret;
        }

        ppgtt->base.clear_range = gen6_ppgtt_clear_range;
        ppgtt->base.insert_entries = gen6_ppgtt_insert_entries;
        ppgtt->base.unbind_vma = ppgtt_unbind_vma;
        ppgtt->base.bind_vma = ppgtt_bind_vma;
        ppgtt->base.set_pages = ppgtt_set_pages;
        ppgtt->base.clear_pages = clear_pages;
        ppgtt->base.cleanup = gen6_ppgtt_cleanup;
        ppgtt->debug_dump = gen6_dump_ppgtt;

        DRM_DEBUG_DRIVER("Allocated pde space (%lldM) at GTT entry: %llx\n",
                         ppgtt->node.size >> 20,
                         ppgtt->node.start / PAGE_SIZE);

        DRM_DEBUG_DRIVER("Adding PPGTT at offset %x\n",
                         ppgtt->pd.base.ggtt_offset << 10);

        return 0;
}

static int __hw_ppgtt_init(struct i915_hw_ppgtt *ppgtt,
                           struct drm_i915_private *dev_priv)
{
        ppgtt->base.i915 = dev_priv;
        ppgtt->base.dma = &dev_priv->drm.pdev->dev;

        if (INTEL_GEN(dev_priv) < 8)
                return gen6_ppgtt_init(ppgtt);
        else
                return gen8_ppgtt_init(ppgtt);
}

static void i915_address_space_init(struct i915_address_space *vm,
                                    struct drm_i915_private *dev_priv,
                                    const char *name)
{
        i915_gem_timeline_init(dev_priv, &vm->timeline, name);

        drm_mm_init(&vm->mm, 0, vm->total);
        vm->mm.head_node.color = I915_COLOR_UNEVICTABLE;

        INIT_LIST_HEAD(&vm->active_list);
        INIT_LIST_HEAD(&vm->inactive_list);
        INIT_LIST_HEAD(&vm->unbound_list);

        list_add_tail(&vm->global_link, &dev_priv->vm_list);
        pagevec_init(&vm->free_pages);
}

static void i915_address_space_fini(struct i915_address_space *vm)
{
        if (pagevec_count(&vm->free_pages))
                vm_free_pages_release(vm, true);

        i915_gem_timeline_fini(&vm->timeline);
        drm_mm_takedown(&vm->mm);
        list_del(&vm->global_link);
}

static void gtt_write_workarounds(struct drm_i915_private *dev_priv)
{
        /* This function is for gtt related workarounds. This function is
         * called on driver load and after a GPU reset, so you can place
         * workarounds here even if they get overwritten by GPU reset.
         */
        /* WaIncreaseDefaultTLBEntries:chv,bdw,skl,bxt,kbl,glk,cfl,cnl */
        if (IS_BROADWELL(dev_priv))
                I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN8_L3_LRA_1_GPGPU_DEFAULT_VALUE_BDW);
        else if (IS_CHERRYVIEW(dev_priv))
                I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN8_L3_LRA_1_GPGPU_DEFAULT_VALUE_CHV);
        else if (IS_GEN9_BC(dev_priv) || IS_GEN10(dev_priv))
                I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN9_L3_LRA_1_GPGPU_DEFAULT_VALUE_SKL);
        else if (IS_GEN9_LP(dev_priv))
                I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN9_L3_LRA_1_GPGPU_DEFAULT_VALUE_BXT);

        /*
         * To support 64K PTEs we need to first enable the use of the
         * Intermediate-Page-Size(IPS) bit of the PDE field via some magical
         * mmio, otherwise the page-walker will simply ignore the IPS bit. This
         * shouldn't be needed after GEN10.
         *
         * 64K pages were first introduced from BDW+, although technically they
         * only *work* from gen9+. For pre-BDW we instead have the option for
         * 32K pages, but we don't currently have any support for it in our
         * driver.
         */
        if (HAS_PAGE_SIZES(dev_priv, I915_GTT_PAGE_SIZE_64K) &&
            INTEL_GEN(dev_priv) <= 10)
                I915_WRITE(GEN8_GAMW_ECO_DEV_RW_IA,
                           I915_READ(GEN8_GAMW_ECO_DEV_RW_IA) |
                           GAMW_ECO_ENABLE_64K_IPS_FIELD);
}

int i915_ppgtt_init_hw(struct drm_i915_private *dev_priv)
{
        gtt_write_workarounds(dev_priv);

        /* In the case of execlists, PPGTT is enabled by the context descriptor
         * and the PDPs are contained within the context itself.  We don't
         * need to do anything here. */
        if (HAS_LOGICAL_RING_CONTEXTS(dev_priv))
                return 0;

        if (!USES_PPGTT(dev_priv))
                return 0;

        if (IS_GEN6(dev_priv))
                gen6_ppgtt_enable(dev_priv);
        else if (IS_GEN7(dev_priv))
                gen7_ppgtt_enable(dev_priv);
        else if (INTEL_GEN(dev_priv) >= 8)
                gen8_ppgtt_enable(dev_priv);
        else
                MISSING_CASE(INTEL_GEN(dev_priv));

        return 0;
}

struct i915_hw_ppgtt *
i915_ppgtt_create(struct drm_i915_private *dev_priv,
                  struct drm_i915_file_private *fpriv,
                  const char *name)
{
        struct i915_hw_ppgtt *ppgtt;
        int ret;

        ppgtt = kzalloc(sizeof(*ppgtt), GFP_KERNEL);
        if (!ppgtt)
                return ERR_PTR(-ENOMEM);

        ret = __hw_ppgtt_init(ppgtt, dev_priv);
        if (ret) {
                kfree(ppgtt);
                return ERR_PTR(ret);
        }

        kref_init(&ppgtt->ref);
        i915_address_space_init(&ppgtt->base, dev_priv, name);
        ppgtt->base.file = fpriv;

        trace_i915_ppgtt_create(&ppgtt->base);

        return ppgtt;
}

void i915_ppgtt_close(struct i915_address_space *vm)
{
        struct list_head *phases[] = {
                &vm->active_list,
                &vm->inactive_list,
                &vm->unbound_list,
                NULL,
        }, **phase;

        GEM_BUG_ON(vm->closed);
        vm->closed = true;

        for (phase = phases; *phase; phase++) {
                struct i915_vma *vma, *vn;

                list_for_each_entry_safe(vma, vn, *phase, vm_link)
                        if (!i915_vma_is_closed(vma))
                                i915_vma_close(vma);
        }
}

void i915_ppgtt_release(struct kref *kref)
{
        struct i915_hw_ppgtt *ppgtt =
                container_of(kref, struct i915_hw_ppgtt, ref);

        trace_i915_ppgtt_release(&ppgtt->base);

        /* vmas should already be unbound and destroyed */
        GEM_BUG_ON(!list_empty(&ppgtt->base.active_list));
        GEM_BUG_ON(!list_empty(&ppgtt->base.inactive_list));
        GEM_BUG_ON(!list_empty(&ppgtt->base.unbound_list));

        ppgtt->base.cleanup(&ppgtt->base);
        i915_address_space_fini(&ppgtt->base);
        kfree(ppgtt);
}

/* Certain Gen5 chipsets require require idling the GPU before
 * unmapping anything from the GTT when VT-d is enabled.
 */
static bool needs_idle_maps(struct drm_i915_private *dev_priv)
{
        /* Query intel_iommu to see if we need the workaround. Presumably that
         * was loaded first.
         */
        return IS_GEN5(dev_priv) && IS_MOBILE(dev_priv) && intel_vtd_active();
}

static void gen6_check_and_clear_faults(struct drm_i915_private *dev_priv)
{
        struct intel_engine_cs *engine;
        enum intel_engine_id id;
        u32 fault;

        for_each_engine(engine, dev_priv, id) {
                fault = I915_READ(RING_FAULT_REG(engine));
                if (fault & RING_FAULT_VALID) {
                        DRM_DEBUG_DRIVER("Unexpected fault\n"
                                         "\tAddr: 0x%08lx\n"
                                         "\tAddress space: %s\n"
                                         "\tSource ID: %d\n"
                                         "\tType: %d\n",
                                         fault & PAGE_MASK,
                                         fault & RING_FAULT_GTTSEL_MASK ? "GGTT" : "PPGTT",
                                         RING_FAULT_SRCID(fault),
                                         RING_FAULT_FAULT_TYPE(fault));
                        I915_WRITE(RING_FAULT_REG(engine),
                                   fault & ~RING_FAULT_VALID);
                }
        }

        POSTING_READ(RING_FAULT_REG(dev_priv->engine[RCS]));
}

static void gen8_check_and_clear_faults(struct drm_i915_private *dev_priv)
{
        u32 fault = I915_READ(GEN8_RING_FAULT_REG);

        if (fault & RING_FAULT_VALID) {
                u32 fault_data0, fault_data1;
                u64 fault_addr;

                fault_data0 = I915_READ(GEN8_FAULT_TLB_DATA0);
                fault_data1 = I915_READ(GEN8_FAULT_TLB_DATA1);
                fault_addr = ((u64)(fault_data1 & FAULT_VA_HIGH_BITS) << 44) |
                             ((u64)fault_data0 << 12);

                DRM_DEBUG_DRIVER("Unexpected fault\n"
                                 "\tAddr: 0x%08x_%08x\n"
                                 "\tAddress space: %s\n"
                                 "\tEngine ID: %d\n"
                                 "\tSource ID: %d\n"
                                 "\tType: %d\n",
                                 upper_32_bits(fault_addr),
                                 lower_32_bits(fault_addr),
                                 fault_data1 & FAULT_GTT_SEL ? "GGTT" : "PPGTT",
                                 GEN8_RING_FAULT_ENGINE_ID(fault),
                                 RING_FAULT_SRCID(fault),
                                 RING_FAULT_FAULT_TYPE(fault));
                I915_WRITE(GEN8_RING_FAULT_REG,
                           fault & ~RING_FAULT_VALID);
        }

        POSTING_READ(GEN8_RING_FAULT_REG);
}

void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
{
        /* From GEN8 onwards we only have one 'All Engine Fault Register' */
        if (INTEL_GEN(dev_priv) >= 8)
                gen8_check_and_clear_faults(dev_priv);
        else if (INTEL_GEN(dev_priv) >= 6)
                gen6_check_and_clear_faults(dev_priv);
        else
                return;
}

void i915_gem_suspend_gtt_mappings(struct drm_i915_private *dev_priv)
{
        struct i915_ggtt *ggtt = &dev_priv->ggtt;

        /* Don't bother messing with faults pre GEN6 as we have little
         * documentation supporting that it's a good idea.
         */
        if (INTEL_GEN(dev_priv) < 6)
                return;

        i915_check_and_clear_faults(dev_priv);

        ggtt->base.clear_range(&ggtt->base, 0, ggtt->base.total);

        i915_ggtt_invalidate(dev_priv);
}

int i915_gem_gtt_prepare_pages(struct drm_i915_gem_object *obj,
                               struct sg_table *pages)
{
        do {
                if (dma_map_sg_attrs(&obj->base.dev->pdev->dev,
                                     pages->sgl, pages->nents,
                                     PCI_DMA_BIDIRECTIONAL,
                                     DMA_ATTR_NO_WARN))
                        return 0;

                /* If the DMA remap fails, one cause can be that we have
                 * too many objects pinned in a small remapping table,
                 * such as swiotlb. Incrementally purge all other objects and
                 * try again - if there are no more pages to remove from
                 * the DMA remapper, i915_gem_shrink will return 0.
                 */
                GEM_BUG_ON(obj->mm.pages == pages);
        } while (i915_gem_shrink(to_i915(obj->base.dev),
                                 obj->base.size >> PAGE_SHIFT, NULL,
                                 I915_SHRINK_BOUND |
                                 I915_SHRINK_UNBOUND |
                                 I915_SHRINK_ACTIVE));

        return -ENOSPC;
}

static void gen8_set_pte(void __iomem *addr, gen8_pte_t pte)
{
        writeq(pte, addr);
}

static void gen8_ggtt_insert_page(struct i915_address_space *vm,
                                  dma_addr_t addr,
                                  u64 offset,
                                  enum i915_cache_level level,
                                  u32 unused)
{
        struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm);
        gen8_pte_t __iomem *pte =
                (gen8_pte_t __iomem *)ggtt->gsm + (offset >> PAGE_SHIFT);

        gen8_set_pte(pte, gen8_pte_encode(addr, level));

        ggtt->invalidate(vm->i915);
}

static void gen8_ggtt_insert_entries(struct i915_address_space *vm,
                                     struct i915_vma *vma,
                                     enum i915_cache_level level,
                                     u32 unused)
{
        struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm);
        struct sgt_iter sgt_iter;
        gen8_pte_t __iomem *gtt_entries;
        const gen8_pte_t pte_encode = gen8_pte_encode(0, level);
        dma_addr_t addr;

        gtt_entries = (gen8_pte_t __iomem *)ggtt->gsm;
        gtt_entries += vma->node.start >> PAGE_SHIFT;
        for_each_sgt_dma(addr, sgt_iter, vma->pages)
                gen8_set_pte(gtt_entries++, pte_encode | addr);

        wmb();

        /* This next bit makes the above posting read even more important. We
         * want to flush the TLBs only after we're certain all the PTE updates
         * have finished.
         */
        ggtt->invalidate(vm->i915);
}

static void gen6_ggtt_insert_page(struct i915_address_space *vm,
                                  dma_addr_t addr,
                                  u64 offset,
                                  enum i915_cache_level level,
                                  u32 flags)
{
        struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm);
        gen6_pte_t __iomem *pte =
                (gen6_pte_t __iomem *)ggtt->gsm + (offset >> PAGE_SHIFT);

        iowrite32(vm->pte_encode(addr, level, flags), pte);

        ggtt->invalidate(vm->i915);
}

/*
 * Binds an object into the global gtt with the specified cache level. The object
 * will be accessible to the GPU via commands whose operands reference offsets
 * within the global GTT as well as accessible by the GPU through the GMADR
 * mapped BAR (dev_priv->mm.gtt->gtt).
 */
static void gen6_ggtt_insert_entries(struct i915_address_space *vm,
                                     struct i915_vma *vma,
                                     enum i915_cache_level level,
                                     u32 flags)
{
        struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm);
        gen6_pte_t __iomem *entries = (gen6_pte_t __iomem *)ggtt->gsm;
        unsigned int i = vma->node.start >> PAGE_SHIFT;
        struct sgt_iter iter;
        dma_addr_t addr;
        for_each_sgt_dma(addr, iter, vma->pages)
                iowrite32(vm->pte_encode(addr, level, flags), &entries[i++]);
        wmb();

        /* This next bit makes the above posting read even more important. We
         * want to flush the TLBs only after we're certain all the PTE updates
         * have finished.
         */
        ggtt->invalidate(vm->i915);
}

static void nop_clear_range(struct i915_address_space *vm,
                            u64 start, u64 length)
{
}

static void gen8_ggtt_clear_range(struct i915_address_space *vm,
                                  u64 start, u64 length)
{
        struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm);
        unsigned first_entry = start >> PAGE_SHIFT;
        unsigned num_entries = length >> PAGE_SHIFT;
        const gen8_pte_t scratch_pte =
                gen8_pte_encode(vm->scratch_page.daddr, I915_CACHE_LLC);
        gen8_pte_t __iomem *gtt_base =
                (gen8_pte_t __iomem *)ggtt->gsm + first_entry;
        const int max_entries = ggtt_total_entries(ggtt) - first_entry;
        int i;

        if (WARN(num_entries > max_entries,
                 "First entry = %d; Num entries = %d (max=%d)\n",
                 first_entry, num_entries, max_entries))
                num_entries = max_entries;

        for (i = 0; i < num_entries; i++)
                gen8_set_pte(&gtt_base[i], scratch_pte);
}

static void bxt_vtd_ggtt_wa(struct i915_address_space *vm)
{
        struct drm_i915_private *dev_priv = vm->i915;

        /*
         * Make sure the internal GAM fifo has been cleared of all GTT
         * writes before exiting stop_machine(). This guarantees that
         * any aperture accesses waiting to start in another process
         * cannot back up behind the GTT writes causing a hang.
         * The register can be any arbitrary GAM register.
         */
        POSTING_READ(GFX_FLSH_CNTL_GEN6);
}

struct insert_page {
        struct i915_address_space *vm;
        dma_addr_t addr;
        u64 offset;
        enum i915_cache_level level;
};

static int bxt_vtd_ggtt_insert_page__cb(void *_arg)
{
        struct insert_page *arg = _arg;

        gen8_ggtt_insert_page(arg->vm, arg->addr, arg->offset, arg->level, 0);
        bxt_vtd_ggtt_wa(arg->vm);

        return 0;
}

static void bxt_vtd_ggtt_insert_page__BKL(struct i915_address_space *vm,
                                          dma_addr_t addr,
                                          u64 offset,
                                          enum i915_cache_level level,
                                          u32 unused)
{
        struct insert_page arg = { vm, addr, offset, level };

        stop_machine(bxt_vtd_ggtt_insert_page__cb, &arg, NULL);
}

struct insert_entries {
        struct i915_address_space *vm;
        struct i915_vma *vma;
        enum i915_cache_level level;
};

static int bxt_vtd_ggtt_insert_entries__cb(void *_arg)
{
        struct insert_entries *arg = _arg;

        gen8_ggtt_insert_entries(arg->vm, arg->vma, arg->level, 0);
        bxt_vtd_ggtt_wa(arg->vm);

        return 0;
}

static void bxt_vtd_ggtt_insert_entries__BKL(struct i915_address_space *vm,
                                             struct i915_vma *vma,
                                             enum i915_cache_level level,
                                             u32 unused)
{
        struct insert_entries arg = { vm, vma, level };

        stop_machine(bxt_vtd_ggtt_insert_entries__cb, &arg, NULL);
}

struct clear_range {
        struct i915_address_space *vm;
        u64 start;
        u64 length;
};

static int bxt_vtd_ggtt_clear_range__cb(void *_arg)
{
        struct clear_range *arg = _arg;

        gen8_ggtt_clear_range(arg->vm, arg->start, arg->length);
        bxt_vtd_ggtt_wa(arg->vm);

        return 0;
}

static void bxt_vtd_ggtt_clear_range__BKL(struct i915_address_space *vm,
                                          u64 start,
                                          u64 length)
{
        struct clear_range arg = { vm, start, length };

        stop_machine(bxt_vtd_ggtt_clear_range__cb, &arg, NULL);
}

static void gen6_ggtt_clear_range(struct i915_address_space *vm,
                                  u64 start, u64 length)
{
        struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm);
        unsigned first_entry = start >> PAGE_SHIFT;
        unsigned num_entries = length >> PAGE_SHIFT;
        gen6_pte_t scratch_pte, __iomem *gtt_base =
                (gen6_pte_t __iomem *)ggtt->gsm + first_entry;
        const int max_entries = ggtt_total_entries(ggtt) - first_entry;
        int i;

        if (WARN(num_entries > max_entries,
                 "First entry = %d; Num entries = %d (max=%d)\n",
                 first_entry, num_entries, max_entries))
                num_entries = max_entries;

        scratch_pte = vm->pte_encode(vm->scratch_page.daddr,
                                     I915_CACHE_LLC, 0);

        for (i = 0; i < num_entries; i++)
                iowrite32(scratch_pte, &gtt_base[i]);
}

static void i915_ggtt_insert_page(struct i915_address_space *vm,
                                  dma_addr_t addr,
                                  u64 offset,
                                  enum i915_cache_level cache_level,
                                  u32 unused)
{
        unsigned int flags = (cache_level == I915_CACHE_NONE) ?
                AGP_USER_MEMORY : AGP_USER_CACHED_MEMORY;

        intel_gtt_insert_page(addr, offset >> PAGE_SHIFT, flags);
}

static void i915_ggtt_insert_entries(struct i915_address_space *vm,
                                     struct i915_vma *vma,
                                     enum i915_cache_level cache_level,
                                     u32 unused)
{
        unsigned int flags = (cache_level == I915_CACHE_NONE) ?
                AGP_USER_MEMORY : AGP_USER_CACHED_MEMORY;

        intel_gtt_insert_sg_entries(vma->pages, vma->node.start >> PAGE_SHIFT,
                                    flags);
}

static void i915_ggtt_clear_range(struct i915_address_space *vm,
                                  u64 start, u64 length)
{
        intel_gtt_clear_range(start >> PAGE_SHIFT, length >> PAGE_SHIFT);
}

static int ggtt_bind_vma(struct i915_vma *vma,
                         enum i915_cache_level cache_level,
                         u32 flags)
{
        struct drm_i915_private *i915 = vma->vm->i915;
        struct drm_i915_gem_object *obj = vma->obj;
        u32 pte_flags;

        /* Currently applicable only to VLV */
        pte_flags = 0;
        if (obj->gt_ro)
                pte_flags |= PTE_READ_ONLY;

        intel_runtime_pm_get(i915);
        vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags);
        intel_runtime_pm_put(i915);

        vma->page_sizes.gtt = I915_GTT_PAGE_SIZE;

        /*
         * Without aliasing PPGTT there's no difference between
         * GLOBAL/LOCAL_BIND, it's all the same ptes. Hence unconditionally
         * upgrade to both bound if we bind either to avoid double-binding.
         */
        vma->flags |= I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND;

        return 0;
}

static void ggtt_unbind_vma(struct i915_vma *vma)
{
        struct drm_i915_private *i915 = vma->vm->i915;

        intel_runtime_pm_get(i915);
        vma->vm->clear_range(vma->vm, vma->node.start, vma->size);
        intel_runtime_pm_put(i915);
}

static int aliasing_gtt_bind_vma(struct i915_vma *vma,
                                 enum i915_cache_level cache_level,
                                 u32 flags)
{
        struct drm_i915_private *i915 = vma->vm->i915;
        u32 pte_flags;
        int ret;

        /* Currently applicable only to VLV */
        pte_flags = 0;
        if (vma->obj->gt_ro)
                pte_flags |= PTE_READ_ONLY;

        if (flags & I915_VMA_LOCAL_BIND) {
                struct i915_hw_ppgtt *appgtt = i915->mm.aliasing_ppgtt;

                if (!(vma->flags & I915_VMA_LOCAL_BIND) &&
                    appgtt->base.allocate_va_range) {
                        ret = appgtt->base.allocate_va_range(&appgtt->base,
                                                             vma->node.start,
                                                             vma->size);
                        if (ret)
                                return ret;
                }

                appgtt->base.insert_entries(&appgtt->base, vma, cache_level,
                                            pte_flags);
        }

        if (flags & I915_VMA_GLOBAL_BIND) {
                intel_runtime_pm_get(i915);
                vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags);
                intel_runtime_pm_put(i915);
        }

        return 0;
}

static void aliasing_gtt_unbind_vma(struct i915_vma *vma)
{
        struct drm_i915_private *i915 = vma->vm->i915;

        if (vma->flags & I915_VMA_GLOBAL_BIND) {
                intel_runtime_pm_get(i915);
                vma->vm->clear_range(vma->vm, vma->node.start, vma->size);
                intel_runtime_pm_put(i915);
        }

        if (vma->flags & I915_VMA_LOCAL_BIND) {
                struct i915_address_space *vm = &i915->mm.aliasing_ppgtt->base;

                vm->clear_range(vm, vma->node.start, vma->size);
        }
}

void i915_gem_gtt_finish_pages(struct drm_i915_gem_object *obj,
                               struct sg_table *pages)
{
        struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
        struct device *kdev = &dev_priv->drm.pdev->dev;
        struct i915_ggtt *ggtt = &dev_priv->ggtt;

        if (unlikely(ggtt->do_idle_maps)) {
                if (i915_gem_wait_for_idle(dev_priv, 0)) {
                        DRM_ERROR("Failed to wait for idle; VT'd may hang.\n");
                        /* Wait a bit, in hopes it avoids the hang */
                        udelay(10);
                }
        }

        dma_unmap_sg(kdev, pages->sgl, pages->nents, PCI_DMA_BIDIRECTIONAL);
}

static int ggtt_set_pages(struct i915_vma *vma)
{
        int ret;

        GEM_BUG_ON(vma->pages);

        ret = i915_get_ggtt_vma_pages(vma);
        if (ret)
                return ret;

        vma->page_sizes = vma->obj->mm.page_sizes;

        return 0;
}

static void i915_gtt_color_adjust(const struct drm_mm_node *node,
                                  unsigned long color,
                                  u64 *start,
                                  u64 *end)
{
        if (node->allocated && node->color != color)
                *start += I915_GTT_PAGE_SIZE;

        /* Also leave a space between the unallocated reserved node after the
         * GTT and any objects within the GTT, i.e. we use the color adjustment
         * to insert a guard page to prevent prefetches crossing over the
         * GTT boundary.
         */
        node = list_next_entry(node, node_list);
        if (node->color != color)
                *end -= I915_GTT_PAGE_SIZE;
}

int i915_gem_init_aliasing_ppgtt(struct drm_i915_private *i915)
{
        struct i915_ggtt *ggtt = &i915->ggtt;
        struct i915_hw_ppgtt *ppgtt;
        int err;

        ppgtt = i915_ppgtt_create(i915, ERR_PTR(-EPERM), "[alias]");
        if (IS_ERR(ppgtt))
                return PTR_ERR(ppgtt);

        if (WARN_ON(ppgtt->base.total < ggtt->base.total)) {
                err = -ENODEV;
                goto err_ppgtt;
        }

        if (ppgtt->base.allocate_va_range) {
                /* Note we only pre-allocate as far as the end of the global
                 * GTT. On 48b / 4-level page-tables, the difference is very,
                 * very significant! We have to preallocate as GVT/vgpu does
                 * not like the page directory disappearing.
                 */
                err = ppgtt->base.allocate_va_range(&ppgtt->base,
                                                    0, ggtt->base.total);
                if (err)
                        goto err_ppgtt;
        }

        i915->mm.aliasing_ppgtt = ppgtt;

        GEM_BUG_ON(ggtt->base.bind_vma != ggtt_bind_vma);
        ggtt->base.bind_vma = aliasing_gtt_bind_vma;

        GEM_BUG_ON(ggtt->base.unbind_vma != ggtt_unbind_vma);
        ggtt->base.unbind_vma = aliasing_gtt_unbind_vma;

        return 0;

err_ppgtt:
        i915_ppgtt_put(ppgtt);
        return err;
}

void i915_gem_fini_aliasing_ppgtt(struct drm_i915_private *i915)
{
        struct i915_ggtt *ggtt = &i915->ggtt;
        struct i915_hw_ppgtt *ppgtt;

        ppgtt = fetch_and_zero(&i915->mm.aliasing_ppgtt);
        if (!ppgtt)
                return;

        i915_ppgtt_put(ppgtt);

        ggtt->base.bind_vma = ggtt_bind_vma;
        ggtt->base.unbind_vma = ggtt_unbind_vma;
}

int i915_gem_init_ggtt(struct drm_i915_private *dev_priv)
{
        /* Let GEM Manage all of the aperture.
         *
         * However, leave one page at the end still bound to the scratch page.
         * There are a number of places where the hardware apparently prefetches
         * past the end of the object, and we've seen multiple hangs with the
         * GPU head pointer stuck in a batchbuffer bound at the last page of the
         * aperture.  One page should be enough to keep any prefetching inside
         * of the aperture.
         */
        struct i915_ggtt *ggtt = &dev_priv->ggtt;
        unsigned long hole_start, hole_end;
        struct drm_mm_node *entry;
        int ret;

        ret = intel_vgt_balloon(dev_priv);
        if (ret)
                return ret;

        /* Reserve a mappable slot for our lockless error capture */
        ret = drm_mm_insert_node_in_range(&ggtt->base.mm, &ggtt->error_capture,
                                          PAGE_SIZE, 0, I915_COLOR_UNEVICTABLE,
                                          0, ggtt->mappable_end,
                                          DRM_MM_INSERT_LOW);
        if (ret)
                return ret;

        /* Clear any non-preallocated blocks */
        drm_mm_for_each_hole(entry, &ggtt->base.mm, hole_start, hole_end) {
                DRM_DEBUG_KMS("clearing unused GTT space: [%lx, %lx]\n",
                              hole_start, hole_end);
                ggtt->base.clear_range(&ggtt->base, hole_start,
                                       hole_end - hole_start);
        }

        /* And finally clear the reserved guard page */
        ggtt->base.clear_range(&ggtt->base,
                               ggtt->base.total - PAGE_SIZE, PAGE_SIZE);

        if (USES_PPGTT(dev_priv) && !USES_FULL_PPGTT(dev_priv)) {
                ret = i915_gem_init_aliasing_ppgtt(dev_priv);
                if (ret)
                        goto err;
        }

        return 0;

err:
        drm_mm_remove_node(&ggtt->error_capture);
        return ret;
}

/**
 * i915_ggtt_cleanup_hw - Clean up GGTT hardware initialization
 * @dev_priv: i915 device
 */
void i915_ggtt_cleanup_hw(struct drm_i915_private *dev_priv)
{
        struct i915_ggtt *ggtt = &dev_priv->ggtt;
        struct i915_vma *vma, *vn;
        struct pagevec *pvec;

        ggtt->base.closed = true;

        mutex_lock(&dev_priv->drm.struct_mutex);
        GEM_BUG_ON(!list_empty(&ggtt->base.active_list));
        list_for_each_entry_safe(vma, vn, &ggtt->base.inactive_list, vm_link)
                WARN_ON(i915_vma_unbind(vma));
        mutex_unlock(&dev_priv->drm.struct_mutex);

        i915_gem_cleanup_stolen(&dev_priv->drm);

        mutex_lock(&dev_priv->drm.struct_mutex);
        i915_gem_fini_aliasing_ppgtt(dev_priv);

        if (drm_mm_node_allocated(&ggtt->error_capture))
                drm_mm_remove_node(&ggtt->error_capture);

        if (drm_mm_initialized(&ggtt->base.mm)) {
                intel_vgt_deballoon(dev_priv);
                i915_address_space_fini(&ggtt->base);
        }

        ggtt->base.cleanup(&ggtt->base);

        pvec = &dev_priv->mm.wc_stash;
        if (pvec->nr) {
                set_pages_array_wb(pvec->pages, pvec->nr);
                __pagevec_release(pvec);
        }

        mutex_unlock(&dev_priv->drm.struct_mutex);

        arch_phys_wc_del(ggtt->mtrr);
        io_mapping_fini(&ggtt->iomap);
}

static unsigned int gen6_get_total_gtt_size(u16 snb_gmch_ctl)
{
        snb_gmch_ctl >>= SNB_GMCH_GGMS_SHIFT;
        snb_gmch_ctl &= SNB_GMCH_GGMS_MASK;
        return snb_gmch_ctl << 20;
}

static unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
{
        bdw_gmch_ctl >>= BDW_GMCH_GGMS_SHIFT;
        bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
        if (bdw_gmch_ctl)
                bdw_gmch_ctl = 1 << bdw_gmch_ctl;

#ifdef CONFIG_X86_32
        /* Limit 32b platforms to a 2GB GGTT: 4 << 20 / pte size * PAGE_SIZE */
        if (bdw_gmch_ctl > 4)
                bdw_gmch_ctl = 4;
#endif

        return bdw_gmch_ctl << 20;
}

static unsigned int chv_get_total_gtt_size(u16 gmch_ctrl)
{
        gmch_ctrl >>= SNB_GMCH_GGMS_SHIFT;
        gmch_ctrl &= SNB_GMCH_GGMS_MASK;

        if (gmch_ctrl)
                return 1 << (20 + gmch_ctrl);

        return 0;
}

static int ggtt_probe_common(struct i915_ggtt *ggtt, u64 size)
{
        struct drm_i915_private *dev_priv = ggtt->base.i915;
        struct pci_dev *pdev = dev_priv->drm.pdev;
        phys_addr_t phys_addr;
        int ret;

        /* For Modern GENs the PTEs and register space are split in the BAR */
        phys_addr = pci_resource_start(pdev, 0) + pci_resource_len(pdev, 0) / 2;

        /*
         * On BXT+/CNL+ writes larger than 64 bit to the GTT pagetable range
         * will be dropped. For WC mappings in general we have 64 byte burst
         * writes when the WC buffer is flushed, so we can't use it, but have to
         * resort to an uncached mapping. The WC issue is easily caught by the
         * readback check when writing GTT PTE entries.
         */
        if (IS_GEN9_LP(dev_priv) || INTEL_GEN(dev_priv) >= 10)
                ggtt->gsm = ioremap_nocache(phys_addr, size);
        else
                ggtt->gsm = ioremap_wc(phys_addr, size);
        if (!ggtt->gsm) {
                DRM_ERROR("Failed to map the ggtt page table\n");
                return -ENOMEM;