From 4a234c5fae1686c8327edeaabc0f2d6fb86db470 Mon Sep 17 00:00:00 2001
From: Matthew Auld <matthew.auld@intel.com>
Date: Thu, 22 Jun 2017 10:58:36 +0100
Subject: [PATCH] drm/i915: pass the vma to insert_entries
Git-commit: 4a234c5fae1686c8327edeaabc0f2d6fb86db470
Patch-mainline: v4.14-rc1
References: FATE#322643 bsc#1055900
The vma already contains most of the information we need for insertion.
But also in preparation for supporting huge gtt pages, it would be
useful to know the details of the vma, such that we can we can easily
determine the page sizes we are allowed to use when inserting into the
48b PPGTT. This is especially true for 64K where we can't just
arbitrarily use it, since we require aligning/padding the vm space to
2M, which sometimes we can't enforce in the upper levels.
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170622095836.6800-1-matthew.auld@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Takashi Iwai <tiwai@suse.de>
---
drivers/gpu/drm/i915/i915_gem_gtt.c | 68 ++++++++++----------------
drivers/gpu/drm/i915/i915_gem_gtt.h | 3 -
drivers/gpu/drm/i915/selftests/i915_gem_gtt.c | 10 +++
drivers/gpu/drm/i915/selftests/mock_gtt.c | 3 -
4 files changed, 38 insertions(+), 46 deletions(-)
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -207,8 +207,7 @@ static int ppgtt_bind_vma(struct i915_vm
if (vma->obj->gt_ro)
pte_flags |= PTE_READ_ONLY;
- vma->vm->insert_entries(vma->vm, vma->pages, vma->node.start,
- cache_level, pte_flags);
+ vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags);
return 0;
}
@@ -907,37 +906,35 @@ gen8_ppgtt_insert_pte_entries(struct i91
}
static void gen8_ppgtt_insert_3lvl(struct i915_address_space *vm,
- struct sg_table *pages,
- u64 start,
+ struct i915_vma *vma,
enum i915_cache_level cache_level,
u32 unused)
{
- struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
+ struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vma->vm);
struct sgt_dma iter = {
- .sg = pages->sgl,
+ .sg = vma->pages->sgl,
.dma = sg_dma_address(iter.sg),
.max = iter.dma + iter.sg->length,
};
- struct gen8_insert_pte idx = gen8_insert_pte(start);
+ struct gen8_insert_pte idx = gen8_insert_pte(vma->node.start);
gen8_ppgtt_insert_pte_entries(ppgtt, &ppgtt->pdp, &iter, &idx,
cache_level);
}
static void gen8_ppgtt_insert_4lvl(struct i915_address_space *vm,
- struct sg_table *pages,
- u64 start,
+ 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 = {
- .sg = pages->sgl,
+ .sg = vma->pages->sgl,
.dma = sg_dma_address(iter.sg),
.max = iter.dma + iter.sg->length,
};
struct i915_page_directory_pointer **pdps = ppgtt->pml4.pdps;
- struct gen8_insert_pte idx = gen8_insert_pte(start);
+ 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))
@@ -1621,13 +1618,12 @@ static void gen6_ppgtt_clear_range(struc
}
static void gen6_ppgtt_insert_entries(struct i915_address_space *vm,
- struct sg_table *pages,
- u64 start,
+ 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 = start >> PAGE_SHIFT;
+ 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);
@@ -1635,7 +1631,7 @@ static void gen6_ppgtt_insert_entries(st
gen6_pte_t *vaddr;
vaddr = kmap_atomic_px(ppgtt->pd.page_table[act_pt]);
- iter.sg = pages->sgl;
+ iter.sg = vma->pages->sgl;
iter.dma = sg_dma_address(iter.sg);
iter.max = iter.dma + iter.sg->length;
do {
@@ -2090,8 +2086,7 @@ static void gen8_ggtt_insert_page(struct
}
static void gen8_ggtt_insert_entries(struct i915_address_space *vm,
- struct sg_table *st,
- u64 start,
+ struct i915_vma *vma,
enum i915_cache_level level,
u32 unused)
{
@@ -2102,8 +2097,8 @@ static void gen8_ggtt_insert_entries(str
dma_addr_t addr;
gtt_entries = (gen8_pte_t __iomem *)ggtt->gsm;
- gtt_entries += start >> PAGE_SHIFT;
- for_each_sgt_dma(addr, sgt_iter, st)
+ 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();
@@ -2137,17 +2132,16 @@ static void gen6_ggtt_insert_page(struct
* mapped BAR (dev_priv->mm.gtt->gtt).
*/
static void gen6_ggtt_insert_entries(struct i915_address_space *vm,
- struct sg_table *st,
- u64 start,
+ 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 = start >> PAGE_SHIFT;
+ unsigned int i = vma->node.start >> PAGE_SHIFT;
struct sgt_iter iter;
dma_addr_t addr;
- for_each_sgt_dma(addr, iter, st)
+ for_each_sgt_dma(addr, iter, vma->pages)
iowrite32(vm->pte_encode(addr, level, flags), &entries[i++]);
wmb();
@@ -2229,8 +2223,7 @@ static void bxt_vtd_ggtt_insert_page__BK
struct insert_entries {
struct i915_address_space *vm;
- struct sg_table *st;
- u64 start;
+ struct i915_vma *vma;
enum i915_cache_level level;
};
@@ -2238,19 +2231,18 @@ static int bxt_vtd_ggtt_insert_entries__
{
struct insert_entries *arg = _arg;
- gen8_ggtt_insert_entries(arg->vm, arg->st, arg->start, arg->level, 0);
+ 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 sg_table *st,
- u64 start,
+ struct i915_vma *vma,
enum i915_cache_level level,
u32 unused)
{
- struct insert_entries arg = { vm, st, start, level };
+ struct insert_entries arg = { vma->vm, vma, level };
stop_machine(bxt_vtd_ggtt_insert_entries__cb, &arg, NULL);
}
@@ -2316,15 +2308,15 @@ static void i915_ggtt_insert_page(struct
}
static void i915_ggtt_insert_entries(struct i915_address_space *vm,
- struct sg_table *pages,
- u64 start,
+ 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(pages, start >> PAGE_SHIFT, flags);
+ intel_gtt_insert_sg_entries(vma->pages, vma->node.start >> PAGE_SHIFT,
+ flags);
}
static void i915_ggtt_clear_range(struct i915_address_space *vm,
@@ -2353,8 +2345,7 @@ static int ggtt_bind_vma(struct i915_vma
pte_flags |= PTE_READ_ONLY;
intel_runtime_pm_get(i915);
- vma->vm->insert_entries(vma->vm, vma->pages, vma->node.start,
- cache_level, pte_flags);
+ vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags);
intel_runtime_pm_put(i915);
/*
@@ -2407,16 +2398,13 @@ static int aliasing_gtt_bind_vma(struct
goto err_pages;
}
- appgtt->base.insert_entries(&appgtt->base,
- vma->pages, vma->node.start,
- cache_level, pte_flags);
+ 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->pages, vma->node.start,
- cache_level, pte_flags);
+ vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags);
intel_runtime_pm_put(i915);
}
--- a/drivers/gpu/drm/i915/i915_gem_gtt.h
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.h
@@ -313,8 +313,7 @@ struct i915_address_space {
enum i915_cache_level cache_level,
u32 flags);
void (*insert_entries)(struct i915_address_space *vm,
- struct sg_table *st,
- u64 start,
+ struct i915_vma *vma,
enum i915_cache_level cache_level,
u32 flags);
void (*cleanup)(struct i915_address_space *vm);
--- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
@@ -197,6 +197,9 @@ static int lowlevel_hole(struct drm_i915
{
I915_RND_STATE(seed_prng);
unsigned int size;
+ struct i915_vma mock_vma;
+
+ memset(&mock_vma, 0, sizeof(struct i915_vma));
/* Keep creating larger objects until one cannot fit into the hole */
for (size = 12; (hole_end - hole_start) >> size; size++) {
@@ -255,8 +258,11 @@ static int lowlevel_hole(struct drm_i915
vm->allocate_va_range(vm, addr, BIT_ULL(size)))
break;
- vm->insert_entries(vm, obj->mm.pages, addr,
- I915_CACHE_NONE, 0);
+ mock_vma.pages = obj->mm.pages;
+ mock_vma.node.size = BIT_ULL(size);
+ mock_vma.node.start = addr;
+
+ vm->insert_entries(vm, &mock_vma, I915_CACHE_NONE, 0);
}
count = n;
--- a/drivers/gpu/drm/i915/selftests/mock_gtt.c
+++ b/drivers/gpu/drm/i915/selftests/mock_gtt.c
@@ -33,8 +33,7 @@ static void mock_insert_page(struct i915
}
static void mock_insert_entries(struct i915_address_space *vm,
- struct sg_table *st,
- u64 start,
+ struct i915_vma *vma,
enum i915_cache_level level, u32 flags)
{
}