On Tue, 2022-05-03 at 19:13 +0000, Robert Beckett wrote:
> refactor internal buffer backend to allocate volatile pages via
> ttm pool allocator
>
> Signed-off-by: Robert Beckett <bob.beckett@xxxxxxxxxxxxx>
> ---
> drivers/gpu/drm/i915/gem/i915_gem_internal.c | 264 ++++++++---------
> --
> drivers/gpu/drm/i915/gem/i915_gem_internal.h | 5 -
> drivers/gpu/drm/i915/gem/i915_gem_ttm.c | 12 +-
> drivers/gpu/drm/i915/gem/i915_gem_ttm.h | 12 +-
> 4 files changed, 125 insertions(+), 168 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_internal.c
> b/drivers/gpu/drm/i915/gem/i915_gem_internal.c
> index c698f95af15f..815ec9466cc0 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_internal.c
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_internal.c
> @@ -4,156 +4,119 @@
> * Copyright © 2014-2016 Intel Corporation
> */
>
> -#include <linux/scatterlist.h>
> -#include <linux/slab.h>
> -#include <linux/swiotlb.h>
> -
> +#include <drm/ttm/ttm_bo_driver.h>
> +#include <drm/ttm/ttm_placement.h>
> +#include "drm/ttm/ttm_bo_api.h"
> +#include "gem/i915_gem_internal.h"
> +#include "gem/i915_gem_region.h"
> +#include "gem/i915_gem_ttm.h"
> #include "i915_drv.h"
> -#include "i915_gem.h"
> -#include "i915_gem_internal.h"
> -#include "i915_gem_object.h"
> -#include "i915_scatterlist.h"
> -#include "i915_utils.h"
> -
> -#define QUIET (__GFP_NORETRY | __GFP_NOWARN)
> -#define MAYFAIL (__GFP_RETRY_MAYFAIL | __GFP_NOWARN)
> -
> -static void internal_free_pages(struct sg_table *st)
> -{
> - struct scatterlist *sg;
> -
> - for (sg = st->sgl; sg; sg = __sg_next(sg)) {
> - if (sg_page(sg))
> - __free_pages(sg_page(sg), get_order(sg-
> >length));
> - }
> -
> - sg_free_table(st);
> - kfree(st);
> -}
>
> -static int i915_gem_object_get_pages_internal(struct
> drm_i915_gem_object *obj)
> +static int i915_internal_get_pages(struct drm_i915_gem_object *obj)
> {
> - struct drm_i915_private *i915 = to_i915(obj->base.dev);
> - struct sg_table *st;
> - struct scatterlist *sg;
> - unsigned int sg_page_sizes;
> - unsigned int npages;
> - int max_order;
> - gfp_t gfp;
> -
> - max_order = MAX_ORDER;
> -#ifdef CONFIG_SWIOTLB
> - if (is_swiotlb_active(obj->base.dev->dev)) {
> - unsigned int max_segment;
> -
> - max_segment = swiotlb_max_segment();
> - if (max_segment) {
> - max_segment = max_t(unsigned int,
> max_segment,
> - PAGE_SIZE) >> PAGE_SHIFT;
> - max_order = min(max_order,
> ilog2(max_segment));
> - }
> + struct ttm_buffer_object *bo = i915_gem_to_ttm(obj);
> + struct ttm_operation_ctx ctx = {
> + .interruptible = true,
> + .no_wait_gpu = false,
> + };
> + struct ttm_place place = {
> + .fpfn = 0,
> + .lpfn = 0,
> + .mem_type = I915_PL_SYSTEM,
> + .flags = 0,
> + };
> + struct ttm_placement placement = {
> + .num_placement = 1,
> + .placement = &place,
> + .num_busy_placement = 0,
> + .busy_placement = NULL,
> + };
> + int ret;
> +
> + ret = ttm_bo_validate(bo, &placement, &ctx);
> + if (ret) {
> + ret = i915_ttm_err_to_gem(ret);
> + return ret;
> }
> -#endif
>
> - gfp = GFP_KERNEL | __GFP_HIGHMEM | __GFP_RECLAIMABLE;
> - if (IS_I965GM(i915) || IS_I965G(i915)) {
> - /* 965gm cannot relocate objects above 4GiB. */
> - gfp &= ~__GFP_HIGHMEM;
> - gfp |= __GFP_DMA32;
It looks like we're losing this restriction?
There is a flag to ttm_device_init() to make TTM only do __GFP_DMA32
allocations.
> + if (bo->ttm && !ttm_tt_is_populated(bo->ttm)) {
> + ret = ttm_tt_populate(bo->bdev, bo->ttm, &ctx);
> + if (ret)
> + return ret;
> }
>
> -create_st:
> - st = kmalloc(sizeof(*st), GFP_KERNEL);
> - if (!st)
> - return -ENOMEM;
> + if (!i915_gem_object_has_pages(obj)) {
> + struct i915_refct_sgt *rsgt =
> + i915_ttm_resource_get_st(obj, bo->resource);
>
> - npages = obj->base.size / PAGE_SIZE;
> - if (sg_alloc_table(st, npages, GFP_KERNEL)) {
> - kfree(st);
> - return -ENOMEM;
> - }
> + if (IS_ERR(rsgt))
> + return PTR_ERR(rsgt);
>
> - sg = st->sgl;
> - st->nents = 0;
> - sg_page_sizes = 0;
> -
> - do {
> - int order = min(fls(npages) - 1, max_order);
> - struct page *page;
> -
> - do {
> - page = alloc_pages(gfp | (order ? QUIET :
> MAYFAIL),
> - order);
> - if (page)
> - break;
> - if (!order--)
> - goto err;
> -
> - /* Limit subsequent allocations as well */
> - max_order = order;
> - } while (1);
> -
> - sg_set_page(sg, page, PAGE_SIZE << order, 0);
> - sg_page_sizes |= PAGE_SIZE << order;
> - st->nents++;
> -
> - npages -= 1 << order;
> - if (!npages) {
> - sg_mark_end(sg);
> - break;
> - }
> -
> - sg = __sg_next(sg);
> - } while (1);
> -
> - if (i915_gem_gtt_prepare_pages(obj, st)) {
> - /* Failed to dma-map try again with single page sg
> segments */
> - if (get_order(st->sgl->length)) {
> - internal_free_pages(st);
> - max_order = 0;
> - goto create_st;
> - }
> - goto err;
> + GEM_BUG_ON(obj->mm.rsgt);
> + obj->mm.rsgt = rsgt;
> + __i915_gem_object_set_pages(obj, &rsgt->table,
> + i915_sg_dma_sizes(rsgt-
> >table.sgl));
> }
>
> - __i915_gem_object_set_pages(obj, st, sg_page_sizes);
> + GEM_BUG_ON(bo->ttm && ((obj->base.size >> PAGE_SHIFT) < bo-
> >ttm->num_pages));
> + i915_ttm_adjust_lru(obj);
>
> return 0;
> +}
>
> -err:
> - sg_set_page(sg, NULL, 0, 0);
> - sg_mark_end(sg);
> - internal_free_pages(st);
> +static const struct drm_i915_gem_object_ops
> i915_gem_object_internal_ops = {
> + .name = "i915_gem_object_ttm",
> + .flags = I915_GEM_OBJECT_IS_SHRINKABLE,
>
> - return -ENOMEM;
> -}
> + .get_pages = i915_internal_get_pages,
> + .put_pages = i915_ttm_put_pages,
> + .adjust_lru = i915_ttm_adjust_lru,
> + .delayed_free = i915_ttm_delayed_free,
> +};
>
> -static void i915_gem_object_put_pages_internal(struct
> drm_i915_gem_object *obj,
> - struct sg_table
> *pages)
> +void i915_ttm_internal_bo_destroy(struct ttm_buffer_object *bo)
> {
> - i915_gem_gtt_finish_pages(obj, pages);
> - internal_free_pages(pages);
> + struct drm_i915_gem_object *obj = i915_ttm_to_gem(bo);
>
> - obj->mm.dirty = false;
> + mutex_destroy(&obj->ttm.get_io_page.lock);
>
> - __start_cpu_write(obj);
> -}
> + if (obj->ttm.created) {
> + /* This releases all gem object bindings to the
> backend. */
> + __i915_gem_free_object(obj);
>
> -static const struct drm_i915_gem_object_ops
> i915_gem_object_internal_ops = {
> - .name = "i915_gem_object_internal",
> - .flags = I915_GEM_OBJECT_IS_SHRINKABLE,
> - .get_pages = i915_gem_object_get_pages_internal,
> - .put_pages = i915_gem_object_put_pages_internal,
> -};
> + call_rcu(&obj->rcu, __i915_gem_free_object_rcu);
> + } else {
> + __i915_gem_object_fini(obj);
> + }
> +}
>
> +/**
> + * i915_gem_object_create_internal: create an object with volatile
> pages
> + * @i915: the i915 device
> + * @size: the size in bytes of backing storage to allocate for the
> object
> + *
> + * Creates a new object that wraps some internal memory for private
> use.
> + * This object is not backed by swappable storage, and as such its
> contents
> + * are volatile and only valid whilst pinned. If the object is
> reaped by the
> + * shrinker, its pages and data will be discarded. Equally, it is
> not a full
> + * GEM object and so not valid for access from userspace. This makes
> it useful
> + * for hardware interfaces like ringbuffers (which are pinned from
> the time
> + * the request is written to the time the hardware stops accessing
> it), but
> + * not for contexts (which need to be preserved when not active for
> later
> + * reuse). Note that it is not cleared upon allocation.
> + */
> struct drm_i915_gem_object *
> -__i915_gem_object_create_internal(struct drm_i915_private *i915,
> - const struct
> drm_i915_gem_object_ops *ops,
> - phys_addr_t size)
> +i915_gem_object_create_internal(struct drm_i915_private *i915,
> + phys_addr_t size)
> {
> static struct lock_class_key lock_class;
> struct drm_i915_gem_object *obj;
> unsigned int cache_level;
> + struct ttm_operation_ctx ctx = {
> + .interruptible = true,
> + .no_wait_gpu = false,
> + };
> + int ret;
>
> GEM_BUG_ON(!size);
> GEM_BUG_ON(!IS_ALIGNED(size, PAGE_SIZE));
> @@ -166,45 +129,34 @@ __i915_gem_object_create_internal(struct
> drm_i915_private *i915,
> return ERR_PTR(-ENOMEM);
>
> drm_gem_private_object_init(&i915->drm, &obj->base, size);
> - i915_gem_object_init(obj, ops, &lock_class, 0);
> - obj->mem_flags |= I915_BO_FLAG_STRUCT_PAGE;
> + i915_gem_object_init(obj, &i915_gem_object_internal_ops,
> &lock_class,
> + I915_BO_ALLOC_VOLATILE);
> +
> + INIT_LIST_HEAD(&obj->mm.region_link);
> +
> + INIT_RADIX_TREE(&obj->ttm.get_io_page.radix, GFP_KERNEL |
> __GFP_NOWARN);
> + mutex_init(&obj->ttm.get_io_page.lock);
>
> - /*
> - * Mark the object as volatile, such that the pages are
> marked as
> - * dontneed whilst they are still pinned. As soon as they are
> unpinned
> - * they are allowed to be reaped by the shrinker, and the
> caller is
> - * expected to repopulate - the contents of this object are
> only valid
> - * whilst active and pinned.
> - */
> - i915_gem_object_set_volatile(obj);
> + obj->base.vma_node.driver_private = i915_gem_to_ttm(obj);
>
> + ret = ttm_bo_init_reserved(&i915->bdev, i915_gem_to_ttm(obj),
> size,
> + ttm_bo_type_kernel,
> i915_ttm_sys_placement(),
> + 0, &ctx, NULL, NULL,
> i915_ttm_internal_bo_destroy);
> + if (ret) {
> + ret = i915_ttm_err_to_gem(ret);
> + i915_gem_object_free(obj);
> + return ERR_PTR(ret);
> + }
> +
> + obj->ttm.created = true;
> obj->read_domains = I915_GEM_DOMAIN_CPU;
> obj->write_domain = I915_GEM_DOMAIN_CPU;
> -
> + obj->mem_flags &= ~I915_BO_FLAG_IOMEM;
> + obj->mem_flags |= I915_BO_FLAG_STRUCT_PAGE;
> cache_level = HAS_LLC(i915) ? I915_CACHE_LLC :
> I915_CACHE_NONE;
> i915_gem_object_set_cache_coherency(obj, cache_level);
> + i915_gem_object_unlock(obj);
>
> return obj;
> }
>
> -/**
> - * i915_gem_object_create_internal: create an object with volatile
> pages
> - * @i915: the i915 device
> - * @size: the size in bytes of backing storage to allocate for the
> object
> - *
> - * Creates a new object that wraps some internal memory for private
> use.
> - * This object is not backed by swappable storage, and as such its
> contents
> - * are volatile and only valid whilst pinned. If the object is
> reaped by the
> - * shrinker, its pages and data will be discarded. Equally, it is
> not a full
> - * GEM object and so not valid for access from userspace. This makes
> it useful
> - * for hardware interfaces like ringbuffers (which are pinned from
> the time
> - * the request is written to the time the hardware stops accessing
> it), but
> - * not for contexts (which need to be preserved when not active for
> later
> - * reuse). Note that it is not cleared upon allocation.
> - */
> -struct drm_i915_gem_object *
> -i915_gem_object_create_internal(struct drm_i915_private *i915,
> - phys_addr_t size)
> -{
> - return __i915_gem_object_create_internal(i915,
> &i915_gem_object_internal_ops, size);
While we don't have a TTM shmem backend ready yet for internal,
Did you consider setting up just yet another region,
INTEL_REGION_INTERNAL,
.class = INTEL_MEMORY_SYSTEM and
.instance = 1,
And make it create a TTM system region on integrated, and use
same region as INTEL_REGION_SMEM on dgfx.
I think ttm should automatically map that to I915_PL_SYSTEM and the
backwards mapping in i915_ttm_region() should never get called since
the object is never moved.
Then I figure it should suffice to just call
__i915_gem_ttm_object_init() and we could drop a lot of code.
/Thomas
> -}
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_internal.h
> b/drivers/gpu/drm/i915/gem/i915_gem_internal.h
> index 6664e06112fc..524e1042b20f 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_internal.h
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_internal.h
> @@ -15,9 +15,4 @@ struct drm_i915_private;
> struct drm_i915_gem_object *
> i915_gem_object_create_internal(struct drm_i915_private *i915,
> phys_addr_t size);
> -struct drm_i915_gem_object *
> -__i915_gem_object_create_internal(struct drm_i915_private *i915,
> - const struct
> drm_i915_gem_object_ops *ops,
> - phys_addr_t size);
> -
> #endif /* __I915_GEM_INTERNAL_H__ */
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
> b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
> index fdb3a1c18cb6..92195ead8c11 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
> @@ -83,7 +83,7 @@ struct ttm_placement *i915_ttm_sys_placement(void)
> return &i915_sys_placement;
> }
>
> -static int i915_ttm_err_to_gem(int err)
> +int i915_ttm_err_to_gem(int err)
> {
> /* Fastpath */
> if (likely(!err))
> @@ -745,8 +745,8 @@ struct ttm_device_funcs *i915_ttm_driver(void)
> return &i915_ttm_bo_driver;
> }
>
> -static int __i915_ttm_get_pages(struct drm_i915_gem_object *obj,
> - struct ttm_placement *placement)
> +int __i915_ttm_get_pages(struct drm_i915_gem_object *obj,
> + struct ttm_placement *placement)
> {
> struct ttm_buffer_object *bo = i915_gem_to_ttm(obj);
> struct ttm_operation_ctx ctx = {
> @@ -871,8 +871,8 @@ static int i915_ttm_migrate(struct
> drm_i915_gem_object *obj,
> return __i915_ttm_migrate(obj, mr, obj->flags);
> }
>
> -static void i915_ttm_put_pages(struct drm_i915_gem_object *obj,
> - struct sg_table *st)
> +void i915_ttm_put_pages(struct drm_i915_gem_object *obj,
> + struct sg_table *st)
> {
> /*
> * We're currently not called from a shrinker, so put_pages()
> @@ -995,7 +995,7 @@ void i915_ttm_adjust_lru(struct
> drm_i915_gem_object *obj)
> * it's not idle, and using the TTM destroyed list handling could
> help us
> * benefit from that.
> */
> -static void i915_ttm_delayed_free(struct drm_i915_gem_object *obj)
> +void i915_ttm_delayed_free(struct drm_i915_gem_object *obj)
> {
> GEM_BUG_ON(!obj->ttm.created);
>
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_ttm.h
> b/drivers/gpu/drm/i915/gem/i915_gem_ttm.h
> index 73e371aa3850..06701c46d8e2 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_ttm.h
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_ttm.h
> @@ -26,6 +26,7 @@ i915_gem_to_ttm(struct drm_i915_gem_object *obj)
> * i915 ttm gem object destructor. Internal use only.
> */
> void i915_ttm_bo_destroy(struct ttm_buffer_object *bo);
> +void i915_ttm_internal_bo_destroy(struct ttm_buffer_object *bo);
>
> /**
> * i915_ttm_to_gem - Convert a struct ttm_buffer_object to an
> embedding
> @@ -37,8 +38,10 @@ void i915_ttm_bo_destroy(struct ttm_buffer_object
> *bo);
> static inline struct drm_i915_gem_object *
> i915_ttm_to_gem(struct ttm_buffer_object *bo)
> {
> - if (bo->destroy != i915_ttm_bo_destroy)
> + if (bo->destroy != i915_ttm_bo_destroy &&
> + bo->destroy != i915_ttm_internal_bo_destroy) {
> return NULL;
> + }
>
> return container_of(bo, struct drm_i915_gem_object,
> __do_not_access);
> }
> @@ -66,6 +69,7 @@ i915_ttm_resource_get_st(struct drm_i915_gem_object
> *obj,
> struct ttm_resource *res);
>
> void i915_ttm_adjust_lru(struct drm_i915_gem_object *obj);
> +void i915_ttm_delayed_free(struct drm_i915_gem_object *obj);
>
> int i915_ttm_purge(struct drm_i915_gem_object *obj);
>
> @@ -92,4 +96,10 @@ static inline bool i915_ttm_cpu_maps_iomem(struct
> ttm_resource *mem)
> /* Once / if we support GGTT, this is also false for cached
> ttm_tts */
> return mem->mem_type != I915_PL_SYSTEM;
> }
> +
> +int __i915_ttm_get_pages(struct drm_i915_gem_object *obj,
> + struct ttm_placement *placement);
> +void i915_ttm_put_pages(struct drm_i915_gem_object *obj, struct
> sg_table *st);
> +int i915_ttm_err_to_gem(int err);
> +
> #endif
