[PATCH] [RFC] drm/i915: make object creation avoid regions layering.

From: Dave Airlie airlied@redhat.com

This looked like indirect ptr for not much reason in the create object path, I just wonder why it couldn't be simpler like this,

The tests aren't cleaned up but this was more of is this a good idea test patch. --- drivers/gpu/drm/i915/gem/i915_gem_lmem.c | 16 ++++------- drivers/gpu/drm/i915/gem/i915_gem_region.c | 33 +++++++++------------- drivers/gpu/drm/i915/gem/i915_gem_region.h | 6 ++-- drivers/gpu/drm/i915/gem/i915_gem_shmem.c | 22 ++++++--------- drivers/gpu/drm/i915/gem/i915_gem_stolen.c | 24 ++++++---------- drivers/gpu/drm/i915/i915_gem.c | 29 ++++++++----------- drivers/gpu/drm/i915/intel_memory_region.h | 5 ---- drivers/gpu/drm/i915/intel_region_lmem.c | 1 - 8 files changed, 50 insertions(+), 86 deletions(-)

diff --git a/drivers/gpu/drm/i915/gem/i915_gem_lmem.c b/drivers/gpu/drm/i915/gem/i915_gem_lmem.c index 932ee21e6609..710fb1158904 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_lmem.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_lmem.c @@ -27,18 +27,14 @@ i915_gem_object_create_lmem(struct drm_i915_private *i915, resource_size_t size, unsigned int flags) { - return i915_gem_object_create_region(i915->mm.regions[INTEL_REGION_LMEM], - size, flags); -} - -struct drm_i915_gem_object * -__i915_gem_lmem_object_create(struct intel_memory_region *mem, - resource_size_t size, - unsigned int flags) -{ + struct intel_memory_region *mem = i915->mm.regions[INTEL_REGION_LMEM]; static struct lock_class_key lock_class; - struct drm_i915_private *i915 = mem->i915; struct drm_i915_gem_object *obj; + int ret; + + ret = i915_gem_object_pre_check(mem, &size, flags); + if (ret) + return ERR_PTR(ret);

obj = i915_gem_object_alloc(); if (!obj) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_region.c b/drivers/gpu/drm/i915/gem/i915_gem_region.c index 1515384d7e0e..0902b3790e70 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_region.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_region.c @@ -133,13 +133,12 @@ void i915_gem_object_release_memory_region(struct drm_i915_gem_object *obj) intel_memory_region_put(mem); }

-struct drm_i915_gem_object * -i915_gem_object_create_region(struct intel_memory_region *mem, - resource_size_t size, +int i915_gem_object_pre_check(struct intel_memory_region *mem, + resource_size_t *size, unsigned int flags) { - struct drm_i915_gem_object *obj;

+ GEM_BUG_ON(!is_power_of_2(mem->min_page_size)); /* * NB: Our use of resource_size_t for the size stems from using struct * resource for the mem->region. We might need to revisit this in the @@ -148,13 +147,13 @@ i915_gem_object_create_region(struct intel_memory_region *mem,

GEM_BUG_ON(flags & ~I915_BO_ALLOC_FLAGS);

- if (!mem) - return ERR_PTR(-ENODEV); + *size = round_up(*size, mem->min_page_size); + if (*size == 0) + return -EINVAL;

- size = round_up(size, mem->min_page_size); - - GEM_BUG_ON(!size); - GEM_BUG_ON(!IS_ALIGNED(size, I915_GTT_MIN_ALIGNMENT)); + /* For most of the ABI (e.g. mmap) we think in system pages */ + GEM_BUG_ON(!IS_ALIGNED(*size, PAGE_SIZE)); + GEM_BUG_ON(!IS_ALIGNED(*size, I915_GTT_MIN_ALIGNMENT));

/* * XXX: There is a prevalence of the assumption that we fit the @@ -163,15 +162,11 @@ i915_gem_object_create_region(struct intel_memory_region *mem, * spot such a local variable, please consider fixing! */

- if (size >> PAGE_SHIFT > INT_MAX) - return ERR_PTR(-E2BIG); - - if (overflows_type(size, obj->base.size)) - return ERR_PTR(-E2BIG); + if (*size >> PAGE_SHIFT > INT_MAX) + return -E2BIG;

- obj = mem->ops->create_object(mem, size, flags); - if (!IS_ERR(obj)) - trace_i915_gem_object_create(obj); + if (overflows_type(*size, size_t)) + return -E2BIG;

- return obj; + return 0; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_region.h b/drivers/gpu/drm/i915/gem/i915_gem_region.h index f2ff6f8bff74..584f3e91060c 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_region.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_region.h @@ -21,9 +21,7 @@ void i915_gem_object_init_memory_region(struct drm_i915_gem_object *obj, unsigned long flags); void i915_gem_object_release_memory_region(struct drm_i915_gem_object *obj);

-struct drm_i915_gem_object * -i915_gem_object_create_region(struct intel_memory_region *mem, - resource_size_t size, +int i915_gem_object_pre_check(struct intel_memory_region *mem, + resource_size_t *size, unsigned int flags); - #endif diff --git a/drivers/gpu/drm/i915/gem/i915_gem_shmem.c b/drivers/gpu/drm/i915/gem/i915_gem_shmem.c index 38113d3c0138..a190933399d4 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_shmem.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_shmem.c @@ -464,19 +464,22 @@ static int __create_shmem(struct drm_i915_private *i915, return 0; }

-static struct drm_i915_gem_object * -create_shmem(struct intel_memory_region *mem, - resource_size_t size, - unsigned int flags) +struct drm_i915_gem_object * +i915_gem_object_create_shmem(struct drm_i915_private *i915, + resource_size_t size) { + struct intel_memory_region *mem = i915->mm.regions[INTEL_REGION_SMEM]; static struct lock_class_key lock_class; - struct drm_i915_private *i915 = mem->i915; struct drm_i915_gem_object *obj; struct address_space *mapping; unsigned int cache_level; gfp_t mask; int ret; + int flags = 0;

+ ret = i915_gem_object_pre_check(mem, &size, flags); + if (ret) + return ERR_PTR(ret); obj = i915_gem_object_alloc(); if (!obj) return ERR_PTR(-ENOMEM); @@ -529,14 +532,6 @@ create_shmem(struct intel_memory_region *mem, return ERR_PTR(ret); }

-struct drm_i915_gem_object * -i915_gem_object_create_shmem(struct drm_i915_private *i915, - resource_size_t size) -{ - return i915_gem_object_create_region(i915->mm.regions[INTEL_REGION_SMEM], - size, 0); -} - /* Allocate a new GEM object and fill it with the supplied data */ struct drm_i915_gem_object * i915_gem_object_create_shmem_from_data(struct drm_i915_private *dev_priv, @@ -611,7 +606,6 @@ static void release_shmem(struct intel_memory_region *mem) static const struct intel_memory_region_ops shmem_region_ops = { .init = init_shmem, .release = release_shmem, - .create_object = create_shmem, };

struct intel_memory_region *i915_gem_shmem_setup(struct drm_i915_private *i915) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_stolen.c b/drivers/gpu/drm/i915/gem/i915_gem_stolen.c index e0f21f12d3ce..1dd4ccd020b1 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_stolen.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_stolen.c @@ -607,21 +607,22 @@ __i915_gem_object_create_stolen(struct intel_memory_region *mem, return ERR_PTR(err); }

-static struct drm_i915_gem_object * -_i915_gem_object_create_stolen(struct intel_memory_region *mem, - resource_size_t size, - unsigned int flags) +struct drm_i915_gem_object * +i915_gem_object_create_stolen(struct drm_i915_private *i915, + resource_size_t size) { - struct drm_i915_private *i915 = mem->i915; + struct intel_memory_region *mem = i915->mm.regions[INTEL_REGION_STOLEN]; struct drm_i915_gem_object *obj; struct drm_mm_node *stolen; int ret; + int flags = I915_BO_ALLOC_CONTIGUOUS;

if (!drm_mm_initialized(&i915->mm.stolen)) return ERR_PTR(-ENODEV);

- if (size == 0) - return ERR_PTR(-EINVAL); + ret = i915_gem_object_pre_check(mem, &size, flags); + if (ret) + return ERR_PTR(ret);

stolen = kzalloc(sizeof(*stolen), GFP_KERNEL); if (!stolen) @@ -646,14 +647,6 @@ _i915_gem_object_create_stolen(struct intel_memory_region *mem, return obj; }

-struct drm_i915_gem_object * -i915_gem_object_create_stolen(struct drm_i915_private *i915, - resource_size_t size) -{ - return i915_gem_object_create_region(i915->mm.regions[INTEL_REGION_STOLEN], - size, I915_BO_ALLOC_CONTIGUOUS); -} - static int init_stolen(struct intel_memory_region *mem) { intel_memory_region_set_name(mem, "stolen"); @@ -673,7 +666,6 @@ static void release_stolen(struct intel_memory_region *mem) static const struct intel_memory_region_ops i915_region_stolen_ops = { .init = init_stolen, .release = release_stolen, - .create_object = _i915_gem_object_create_stolen, };

struct intel_memory_region *i915_gem_stolen_setup(struct drm_i915_private *i915) diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 9aa3066cb75d..1096c1818a66 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -45,6 +45,7 @@ #include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_mman.h" #include "gem/i915_gem_region.h" +#include "gem/i915_gem_lmem.h" #include "gt/intel_engine_user.h" #include "gt/intel_gt.h" #include "gt/intel_gt_pm.h" @@ -205,7 +206,8 @@ i915_gem_phys_pwrite(struct drm_i915_gem_object *obj,

static int i915_gem_create(struct drm_file *file, - struct intel_memory_region *mr, + struct drm_i915_private *i915, + enum intel_memory_type mem_type, u64 *size_p, u32 *handle_p) { @@ -214,16 +216,12 @@ i915_gem_create(struct drm_file *file, u64 size; int ret;

- GEM_BUG_ON(!is_power_of_2(mr->min_page_size)); - size = round_up(*size_p, mr->min_page_size); - if (size == 0) - return -EINVAL; - - /* For most of the ABI (e.g. mmap) we think in system pages */ - GEM_BUG_ON(!IS_ALIGNED(size, PAGE_SIZE)); - - /* Allocate the new object */ - obj = i915_gem_object_create_region(mr, size, 0); + if (mem_type == INTEL_MEMORY_LOCAL) + obj = i915_gem_object_create_lmem(i915, size, 0); + else if (mem_type == INTEL_MEMORY_SYSTEM) + obj = i915_gem_object_create_shmem(i915, size); + else + obj = ERR_PTR(-ENODEV); if (IS_ERR(obj)) return PTR_ERR(obj);

@@ -278,9 +276,8 @@ i915_gem_dumb_create(struct drm_file *file, if (HAS_LMEM(to_i915(dev))) mem_type = INTEL_MEMORY_LOCAL;

- return i915_gem_create(file, - intel_memory_region_by_type(to_i915(dev), - mem_type), + return i915_gem_create(file, to_i915(dev), + mem_type, &args->size, &args->handle); }

@@ -299,9 +296,7 @@ i915_gem_create_ioctl(struct drm_device *dev, void *data,

i915_gem_flush_free_objects(i915);

- return i915_gem_create(file, - intel_memory_region_by_type(i915, - INTEL_MEMORY_SYSTEM), + return i915_gem_create(file, i915, INTEL_MEMORY_SYSTEM, &args->size, &args->handle); }

diff --git a/drivers/gpu/drm/i915/intel_memory_region.h b/drivers/gpu/drm/i915/intel_memory_region.h index 232490d89a83..b2db98a89795 100644 --- a/drivers/gpu/drm/i915/intel_memory_region.h +++ b/drivers/gpu/drm/i915/intel_memory_region.h @@ -61,11 +61,6 @@ struct intel_memory_region_ops {

int (*init)(struct intel_memory_region *mem); void (*release)(struct intel_memory_region *mem); - - struct drm_i915_gem_object * - (*create_object)(struct intel_memory_region *mem, - resource_size_t size, - unsigned int flags); };

struct intel_memory_region { diff --git a/drivers/gpu/drm/i915/intel_region_lmem.c b/drivers/gpu/drm/i915/intel_region_lmem.c index 40d8f1a95df6..e1b14add006b 100644 --- a/drivers/gpu/drm/i915/intel_region_lmem.c +++ b/drivers/gpu/drm/i915/intel_region_lmem.c @@ -98,7 +98,6 @@ region_lmem_init(struct intel_memory_region *mem) const struct intel_memory_region_ops intel_region_lmem_ops = { .init = region_lmem_init, .release = region_lmem_release, - .create_object = __i915_gem_lmem_object_create, };

struct intel_memory_region *