[Intel-gfx] [PATCH v5 2/2] drm/i915: add in-kernel blitter client
Chris Wilson
chris at chris-wilson.co.uk
Tue May 28 20:52:04 UTC 2019
Quoting Matthew Auld (2019-05-28 20:57:24)
> The plan is to use the blitter engine for async object clearing when
> using local memory, but before we can move the worker to get_pages() we
> have to first tame some more of our struct_mutex usage. With this in
> mind we should be able to upstream the object clearing as some
> selftests, which should serve as a guinea pig for the ongoing locking
> rework and upcoming async get_pages() framework.
>
> Signed-off-by: Matthew Auld <matthew.auld at intel.com>
> Cc: CQ Tang <cq.tang at intel.com>
> ---
> drivers/gpu/drm/i915/Makefile | 2 +
> .../gpu/drm/i915/gem/i915_gem_client_blt.c | 306 ++++++++++++++++++
> .../gpu/drm/i915/gem/i915_gem_client_blt.h | 21 ++
> .../gpu/drm/i915/gem/i915_gem_object_blt.c | 109 +++++++
> .../gpu/drm/i915/gem/i915_gem_object_blt.h | 24 ++
> .../i915/gem/selftests/i915_gem_client_blt.c | 127 ++++++++
> .../i915/gem/selftests/i915_gem_object_blt.c | 111 +++++++
> drivers/gpu/drm/i915/gt/intel_gpu_commands.h | 1 +
> .../drm/i915/selftests/i915_live_selftests.h | 2 +
> 9 files changed, 703 insertions(+)
> create mode 100644 drivers/gpu/drm/i915/gem/i915_gem_client_blt.c
> create mode 100644 drivers/gpu/drm/i915/gem/i915_gem_client_blt.h
> create mode 100644 drivers/gpu/drm/i915/gem/i915_gem_object_blt.c
> create mode 100644 drivers/gpu/drm/i915/gem/i915_gem_object_blt.h
> create mode 100644 drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c
> create mode 100644 drivers/gpu/drm/i915/gem/selftests/i915_gem_object_blt.c
>
> diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile
> index 83588e9840f8..a7850bbffbe0 100644
> --- a/drivers/gpu/drm/i915/Makefile
> +++ b/drivers/gpu/drm/i915/Makefile
> @@ -90,6 +90,7 @@ obj-y += gem/
> gem-y += \
> gem/i915_gem_busy.o \
> gem/i915_gem_clflush.o \
> + gem/i915_gem_client_blt.o \
> gem/i915_gem_context.o \
> gem/i915_gem_dmabuf.o \
> gem/i915_gem_domain.o \
> @@ -97,6 +98,7 @@ gem-y += \
> gem/i915_gem_fence.o \
> gem/i915_gem_internal.o \
> gem/i915_gem_object.o \
> + gem/i915_gem_object_blt.o \
> gem/i915_gem_mman.o \
> gem/i915_gem_pages.o \
> gem/i915_gem_phys.o \
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c b/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c
> new file mode 100644
> index 000000000000..2d8cf29a5796
> --- /dev/null
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c
> @@ -0,0 +1,306 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright © 2019 Intel Corporation
> + */
> +#include "i915_gem_client_blt.h"
> +
> +#include "i915_gem_object_blt.h"
> +#include "intel_drv.h"
> +
> +struct i915_sleeve {
> + struct i915_vma *vma;
> + struct drm_i915_gem_object *obj;
> + struct sg_table *pages;
> + struct i915_page_sizes page_sizes;
> +};
> +
> +static int vma_set_pages(struct i915_vma *vma)
> +{
> + struct i915_sleeve *sleeve = vma->private;
> +
> + vma->pages = sleeve->pages;
> + vma->page_sizes = sleeve->page_sizes;
> +
> + return 0;
> +}
> +
> +static void vma_clear_pages(struct i915_vma *vma)
> +{
> + GEM_BUG_ON(!vma->pages);
> + vma->pages = NULL;
> +}
> +
> +static int vma_bind(struct i915_vma *vma,
> + enum i915_cache_level cache_level,
> + u32 flags)
> +{
> + return vma->vm->vma_ops.bind_vma(vma, cache_level, flags);
> +}
> +
> +static void vma_unbind(struct i915_vma *vma)
> +{
> + vma->vm->vma_ops.unbind_vma(vma);
> +}
> +
> +static const struct i915_vma_ops proxy_vma_ops = {
> + .set_pages = vma_set_pages,
> + .clear_pages = vma_clear_pages,
> + .bind_vma = vma_bind,
> + .unbind_vma = vma_unbind,
> +};
> +
> +static struct i915_sleeve *create_sleeve(struct i915_address_space *vm,
> + struct drm_i915_gem_object *obj,
> + struct sg_table *pages,
> + struct i915_page_sizes *page_sizes)
> +{
> + struct i915_sleeve *sleeve;
> + struct i915_vma *vma;
> + int err;
> +
> + sleeve = kzalloc(sizeof(*sleeve), GFP_KERNEL);
> + if (!sleeve)
> + return ERR_PTR(-ENOMEM);
> +
> + vma = i915_vma_instance(obj, vm, NULL);
> + if (IS_ERR(vma)) {
> + err = PTR_ERR(vma);
> + goto err_free;
> + }
> +
> + vma->private = sleeve;
> + vma->ops = &proxy_vma_ops;
> +
> + sleeve->vma = vma;
> + sleeve->obj = i915_gem_object_get(obj);
> + sleeve->pages = pages;
> + sleeve->page_sizes = *page_sizes;
> +
> + return sleeve;
> +
> +err_free:
> + kfree(sleeve);
> + return ERR_PTR(err);
> +}
> +
> +static void destroy_sleeve(struct i915_sleeve *sleeve)
> +{
> + i915_gem_object_put(sleeve->obj);
> + kfree(sleeve);
> +}
> +
> +struct clear_pages_work {
> + struct dma_fence dma;
> + struct dma_fence_cb cb;
> + struct i915_sw_fence wait;
> + struct work_struct work;
> + struct irq_work irq_work;
> + struct i915_sleeve *sleeve;
> + struct intel_context *ce;
> + u32 value;
> +};
> +
> +static const char *clear_pages_work_driver_name(struct dma_fence *fence)
> +{
> + return DRIVER_NAME;
> +}
> +
> +static const char *clear_pages_work_timeline_name(struct dma_fence *fence)
> +{
> + return "clear";
> +}
> +
> +static void clear_pages_work_release(struct dma_fence *fence)
> +{
> + struct clear_pages_work *w = container_of(fence, typeof(*w), dma);
> +
> + destroy_sleeve(w->sleeve);
> +
> + i915_sw_fence_fini(&w->wait);
> +
> + BUILD_BUG_ON(offsetof(typeof(*w), dma));
> + dma_fence_free(&w->dma);
> +}
> +
> +static const struct dma_fence_ops clear_pages_work_ops = {
> + .get_driver_name = clear_pages_work_driver_name,
> + .get_timeline_name = clear_pages_work_timeline_name,
> + .release = clear_pages_work_release,
> +};
> +
> +static void clear_pages_signal_irq_worker(struct irq_work *work)
> +{
> + struct clear_pages_work *w = container_of(work, typeof(*w), irq_work);
> +
> + dma_fence_signal(&w->dma);
> + dma_fence_put(&w->dma);
> +}
> +
> +static void clear_pages_dma_fence_cb(struct dma_fence *fence,
> + struct dma_fence_cb *cb)
> +{
> + struct clear_pages_work *w = container_of(cb, typeof(*w), cb);
> +
> + if (fence->error)
> + dma_fence_set_error(&w->dma, fence->error);
> +
> + /*
> + * Push the signalling of the fence into yet another worker to avoid
> + * the nightmare locking around the fence spinlock.
> + */
> + irq_work_queue(&w->irq_work);
> +}
> +
> +static void clear_pages_worker(struct work_struct *work)
> +{
> + struct clear_pages_work *w = container_of(work, typeof(*w), work);
> + struct drm_i915_private *i915 = w->ce->gem_context->i915;
> + struct drm_i915_gem_object *obj = w->sleeve->obj;
> + struct i915_vma *vma = w->sleeve->vma;
> + struct i915_request *rq;
> + int err = w->dma.error;
> +
> + if (unlikely(err))
> + goto out_signal;
> +
> + if (obj->cache_dirty) {
> + obj->write_domain = 0;
> + if (i915_gem_object_has_struct_page(obj))
> + drm_clflush_sg(w->sleeve->pages);
> + obj->cache_dirty = false;
> + }
> +
> + /* XXX: we need to kill this */
> + mutex_lock(&i915->drm.struct_mutex);
> + err = i915_vma_pin(vma, 0, 0, PIN_USER);
Getting closer, day by day.
> + if (unlikely(err))
> + goto out_unlock;
> +
> + rq = i915_request_create(w->ce);
> + if (IS_ERR(rq)) {
> + err = PTR_ERR(rq);
> + goto out_unpin;
> + }
> +
> + /* There's no way the fence has signalled */
> + if (dma_fence_add_callback(&rq->fence, &w->cb,
> + clear_pages_dma_fence_cb))
> + GEM_BUG_ON(1);
> +
> + if (w->ce->engine->emit_init_breadcrumb) {
> + err = w->ce->engine->emit_init_breadcrumb(rq);
> + if (unlikely(err))
> + goto out_request;
> + }
> +
> + err = intel_emit_vma_fill_blt(rq, vma, w->value);
> + if (unlikely(err))
> + goto out_request;
> +
> + /* XXX: more feverish nightmares await */
> + i915_vma_lock(vma);
> + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE);
> + i915_vma_unlock(vma);
Right. That is actually going to be quite tricky. Good to have it
spelled out. How well do we detect the lock inversion if you remove the
other XXX is a good question.
In general I think it is better to the move_to_active before emitting
the commands that may write into the vma -- but it doesn't matter in
this case because you use i915_request_skip, however it keeps on
triggering alarm bells :)
> +out_request:
> + if (unlikely(err)) {
> + i915_request_skip(rq, err);
> + err = 0;
> + }
> +
> + i915_request_add(rq);
> +out_unpin:
> + i915_vma_unpin(vma);
> +out_unlock:
> + mutex_unlock(&i915->drm.struct_mutex);
> +out_signal:
> + if (unlikely(err)) {
> + dma_fence_set_error(&w->dma, err);
> + dma_fence_signal(&w->dma);
> + dma_fence_put(&w->dma);
> + }
> +}
> +
> +static int __i915_sw_fence_call
> +clear_pages_work_notify(struct i915_sw_fence *fence,
> + enum i915_sw_fence_notify state)
> +{
> + struct clear_pages_work *w = container_of(fence, typeof(*w), wait);
> +
> + switch (state) {
> + case FENCE_COMPLETE:
> + schedule_work(&w->work);
> + break;
> +
> + case FENCE_FREE:
> + dma_fence_put(&w->dma);
> + break;
> + }
> +
> + return NOTIFY_DONE;
> +}
> +
> +static DEFINE_SPINLOCK(fence_lock);
> +
> +/* XXX: better name please */
> +int i915_gem_schedule_fill_pages_blt(struct drm_i915_gem_object *obj,
> + struct intel_context *ce,
> + struct sg_table *pages,
> + struct i915_page_sizes *page_sizes,
> + u32 value)
> +{
> + struct drm_i915_private *i915 = to_i915(obj->base.dev);
> + struct i915_gem_context *ctx = ce->gem_context;
> + struct i915_address_space *vm;
> + struct clear_pages_work *work;
> + struct i915_sleeve *sleeve;
> + int err;
> +
> + vm = ctx->ppgtt ? &ctx->ppgtt->vm : &i915->ggtt.vm;
> +
> + sleeve = create_sleeve(vm, obj, pages, page_sizes);
> + if (IS_ERR(sleeve))
> + return PTR_ERR(sleeve);
> +
> + work = kmalloc(sizeof(*work), GFP_KERNEL);
> + if (!work) {
> + destroy_sleeve(sleeve);
> + return -ENOMEM;
> + }
> +
> + work->value = value;
> + work->sleeve = sleeve;
> + work->ce = ce;
> +
> + INIT_WORK(&work->work, clear_pages_worker);
> +
> + init_irq_work(&work->irq_work, clear_pages_signal_irq_worker);
> +
> + dma_fence_init(&work->dma,
> + &clear_pages_work_ops,
> + &fence_lock,
> + i915->mm.unordered_timeline,
> + 0);
> + i915_sw_fence_init(&work->wait, clear_pages_work_notify);
> +
> + i915_gem_object_lock(obj);
> + err = i915_sw_fence_await_reservation(&work->wait,
> + obj->resv, NULL,
> + true, I915_FENCE_TIMEOUT,
> + I915_FENCE_GFP);
> + if (err < 0) {
> + dma_fence_set_error(&work->dma, err);
> + } else {
> + reservation_object_add_excl_fence(obj->resv, &work->dma);
> + err = 0;
> + }
> + i915_gem_object_unlock(obj);
> +
> + dma_fence_get(&work->dma);
> + i915_sw_fence_commit(&work->wait);
> +
> + return err;
> +}
> +
> +#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
> +#include "selftests/i915_gem_client_blt.c"
> +#endif
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_client_blt.h b/drivers/gpu/drm/i915/gem/i915_gem_client_blt.h
> new file mode 100644
> index 000000000000..3dbd28c22ff5
> --- /dev/null
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_client_blt.h
> @@ -0,0 +1,21 @@
> +/* SPDX-License-Identifier: MIT */
> +/*
> + * Copyright © 2019 Intel Corporation
> + */
> +#ifndef __I915_GEM_CLIENT_BLT_H__
> +#define __I915_GEM_CLIENT_BLT_H__
> +
> +#include <linux/types.h>
> +
> +struct drm_i915_gem_object;
> +struct i915_page_sizes;
> +struct intel_context;
> +struct sg_table;
> +
> +int i915_gem_schedule_fill_pages_blt(struct drm_i915_gem_object *obj,
> + struct intel_context *ce,
> + struct sg_table *pages,
> + struct i915_page_sizes *page_sizes,
> + u32 value);
> +
> +#endif
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c
> new file mode 100644
> index 000000000000..84324b755de6
> --- /dev/null
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c
> @@ -0,0 +1,109 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright © 2019 Intel Corporation
> + */
> +
> +#include "i915_gem_object_blt.h"
> +
> +#include "i915_gem_clflush.h"
> +#include "intel_drv.h"
> +
> +int intel_emit_vma_fill_blt(struct i915_request *rq,
> + struct i915_vma *vma,
> + u32 value)
> +{
> + u32 *cs;
> +
> + cs = intel_ring_begin(rq, 8);
> + if (IS_ERR(cs))
> + return PTR_ERR(cs);
> +
> + if (INTEL_GEN(rq->i915) >= 8) {
> + *cs++ = XY_COLOR_BLT_CMD | BLT_WRITE_RGBA | (7-2);
> + *cs++ = BLT_DEPTH_32 | BLT_ROP_COLOR_COPY | PAGE_SIZE;
> + *cs++ = 0;
> + *cs++ = vma->size >> PAGE_SHIFT << 16 | PAGE_SIZE / 4;
> + *cs++ = lower_32_bits(vma->node.start);
> + *cs++ = upper_32_bits(vma->node.start);
> + *cs++ = value;
> + *cs++ = MI_NOOP;
> + } else {
> + *cs++ = XY_COLOR_BLT_CMD | BLT_WRITE_RGBA | (6-2);
> + *cs++ = BLT_DEPTH_32 | BLT_ROP_COLOR_COPY | PAGE_SIZE;
> + *cs++ = 0;
> + *cs++ = vma->size >> PAGE_SHIFT << 16 | PAGE_SIZE / 4;
> + *cs++ = vma->node.start;
> + *cs++ = value;
> + *cs++ = MI_NOOP;
> + *cs++ = MI_NOOP;
> + }
> +
> + intel_ring_advance(rq, cs);
> +
> + return 0;
> +}
> +
> +int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj,
> + struct intel_context *ce,
> + u32 value)
> +{
> + struct drm_i915_private *i915 = to_i915(obj->base.dev);
> + struct i915_gem_context *ctx = ce->gem_context;
> + struct i915_address_space *vm;
> + struct i915_request *rq;
> + struct i915_vma *vma;
> + int err;
> +
> + /* XXX: ce->vm please */
> + vm = ctx->ppgtt ? &ctx->ppgtt->vm : &i915->ggtt.vm;
> +
> + vma = i915_vma_instance(obj, vm, NULL);
> + if (IS_ERR(vma))
> + return PTR_ERR(vma);
> +
> + err = i915_vma_pin(vma, 0, 0, PIN_USER);
> + if (unlikely(err))
> + return err;
> +
> + if (obj->cache_dirty & ~obj->cache_coherent) {
> + i915_gem_object_lock(obj);
> + i915_gem_clflush_object(obj, 0);
> + i915_gem_object_unlock(obj);
> + }
> +
> + rq = i915_request_create(ce);
> + if (IS_ERR(rq)) {
> + err = PTR_ERR(rq);
> + goto out_unpin;
> + }
> +
> + err = i915_request_await_object(rq, obj, true);
> + if (unlikely(err))
> + goto out_request;
> +
> + if (ce->engine->emit_init_breadcrumb) {
> + err = ce->engine->emit_init_breadcrumb(rq);
> + if (unlikely(err))
> + goto out_request;
> + }
> +
> + err = intel_emit_vma_fill_blt(rq, vma, value);
> + if (unlikely(err))
> + goto out_request;
> +
> + i915_vma_lock(vma);
> + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE);
> + i915_vma_unlock(vma);
> +out_request:
> + if (unlikely(err))
> + i915_request_skip(rq, err);
> +
> + i915_request_add(rq);
> +out_unpin:
> + i915_vma_unpin(vma);
> + return err;
> +}
> +
> +#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
> +#include "selftests/i915_gem_object_blt.c"
> +#endif
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_blt.h b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.h
> new file mode 100644
> index 000000000000..7ec7de6ac0c0
> --- /dev/null
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.h
> @@ -0,0 +1,24 @@
> +/* SPDX-License-Identifier: MIT */
> +/*
> + * Copyright © 2019 Intel Corporation
> + */
> +
> +#ifndef __I915_GEM_OBJECT_BLT_H__
> +#define __I915_GEM_OBJECT_BLT_H__
> +
> +#include <linux/types.h>
> +
> +struct drm_i915_gem_object;
> +struct intel_context;
> +struct i915_request;
> +struct i915_vma;
> +
> +int intel_emit_vma_fill_blt(struct i915_request *rq,
> + struct i915_vma *vma,
> + u32 value);
> +
> +int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj,
> + struct intel_context *ce,
> + u32 value);
> +
> +#endif
> diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c
> new file mode 100644
> index 000000000000..b650d8656d92
> --- /dev/null
> +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c
> @@ -0,0 +1,127 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright © 2019 Intel Corporation
> + */
> +
> +#include "../i915_selftest.h"
> +
> +#include "selftests/igt_flush_test.h"
> +#include "selftests/mock_drm.h"
> +#include "mock_context.h"
> +
> +static int igt_client_fill(void *arg)
> +{
> + struct intel_context *ce = arg;
> + struct drm_i915_private *i915 = ce->gem_context->i915;
> + struct drm_i915_gem_object *obj;
> + struct rnd_state prng;
> + IGT_TIMEOUT(end);
> + u32 *vaddr;
> + int err = 0;
> +
> + prandom_seed_state(&prng, i915_selftest.random_seed);
> +
> + do {
> + u32 sz = prandom_u32_state(&prng) % SZ_32M;
> + u32 val = prandom_u32_state(&prng);
> + u32 i;
> +
> + sz = round_up(sz, PAGE_SIZE);
> +
> + pr_debug("%s with sz=%x, val=%x\n", __func__, sz, val);
> +
> + obj = i915_gem_object_create_internal(i915, sz);
> + if (IS_ERR(obj)) {
> + err = PTR_ERR(obj);
> + goto err_flush;
> + }
> +
> + vaddr = i915_gem_object_pin_map(obj, I915_MAP_WB);
> + if (IS_ERR(vaddr)) {
> + err = PTR_ERR(vaddr);
> + goto err_put;
> + }
> +
> + /*
> + * XXX: The goal is move this to get_pages, so try to dirty the
> + * CPU cache first to check that we do the required clflush
> + * before scheduling the blt for !llc platforms. This matches
> + * some version of reality where at get_pages the pages
> + * themselves may not yet be coherent with the GPU(swap-in). If
> + * we are missing the flush then we should see the stale cache
> + * values after we do the set_to_cpu_domain and pick it up as a
> + * test failure.
> + */
> + memset32(vaddr, val ^ 0xdeadbeaf, obj->base.size / sizeof(u32));
> +
> + if (!(obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_WRITE))
> + obj->cache_dirty = true;
> +
> + err = i915_gem_schedule_fill_pages_blt(obj, ce, obj->mm.pages,
> + &obj->mm.page_sizes,
> + val);
> + if (err)
> + goto err_unpin;
> +
> + /*
> + * XXX: For now do the wait without the object resv lock to
> + * ensure we don't deadlock.
> + */
> + err = i915_gem_object_wait(obj,
> + I915_WAIT_INTERRUPTIBLE |
> + I915_WAIT_ALL,
> + MAX_SCHEDULE_TIMEOUT);
> + if (err)
> + goto err_unpin;
> +
> + i915_gem_object_lock(obj);
> + err = i915_gem_object_set_to_cpu_domain(obj, false);
> + i915_gem_object_unlock(obj);
> + if (err)
> + goto err_unpin;
> +
> + for (i = 0; i < obj->base.size / sizeof(u32); ++i) {
> + if (vaddr[i] != val) {
> + pr_err("vaddr[%u]=%x, expected=%x\n", i,
> + vaddr[i], val);
> + err = -EINVAL;
> + goto err_unpin;
> + }
> + }
> +
> + i915_gem_object_unpin_map(obj);
> + i915_gem_object_put(obj);
> + } while (!time_after(jiffies, end));
> +
> + goto err_flush;
> +
> +err_unpin:
> + i915_gem_object_unpin_map(obj);
> +err_put:
> + i915_gem_object_put(obj);
> +err_flush:
> + mutex_lock(&i915->drm.struct_mutex);
> + if (igt_flush_test(i915, I915_WAIT_LOCKED))
> + err = -EIO;
> + mutex_unlock(&i915->drm.struct_mutex);
> +
> + if (err == -ENOMEM)
> + err = 0;
> +
> + return err;
> +}
> +
> +int i915_gem_client_blt_live_selftests(struct drm_i915_private *i915)
> +{
> + static const struct i915_subtest tests[] = {
> + SUBTEST(igt_client_fill),
> + };
> +
> + if (i915_terminally_wedged(i915))
> + return 0;
> +
> + if (!HAS_ENGINE(i915, BCS0))
> + return 0;
> +
> + return i915_subtests(tests, i915->engine[BCS0]->kernel_context);
> +}
> diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_object_blt.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_object_blt.c
> new file mode 100644
> index 000000000000..717521c8eb0a
> --- /dev/null
> +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_object_blt.c
> @@ -0,0 +1,111 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright © 2019 Intel Corporation
> + */
> +
> +#include "../i915_selftest.h"
> +
> +#include "selftests/igt_flush_test.h"
> +#include "selftests/mock_drm.h"
> +#include "mock_context.h"
> +
> +static int igt_fill_blt(void *arg)
> +{
> + struct intel_context *ce = arg;
> + struct drm_i915_private *i915 = ce->gem_context->i915;
> + struct drm_i915_gem_object *obj;
> + struct rnd_state prng;
> + IGT_TIMEOUT(end);
> + u32 *vaddr;
> + int err = 0;
> +
> + prandom_seed_state(&prng, i915_selftest.random_seed);
> +
> + do {
> + u32 sz = prandom_u32_state(&prng) % SZ_32M;
> + u32 val = prandom_u32_state(&prng);
> + u32 i;
> +
> + sz = round_up(sz, PAGE_SIZE);
> +
> + pr_debug("%s with sz=%x, val=%x\n", __func__, sz, val);
> +
> + obj = i915_gem_object_create_internal(i915, sz);
> + if (IS_ERR(obj)) {
> + err = PTR_ERR(vaddr);
> + goto err_flush;
> + }
> +
> + vaddr = i915_gem_object_pin_map(obj, I915_MAP_WB);
> + if (IS_ERR(vaddr)) {
> + err = PTR_ERR(vaddr);
> + goto err_put;
> + }
> +
> + /*
> + * Make sure the potentially async clflush does its job, if
> + * required.
> + */
> + memset32(vaddr, val ^ 0xdeadbeaf, obj->base.size / sizeof(u32));
> +
> + if (!(obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_WRITE))
> + obj->cache_dirty = true;
> +
> +
> + mutex_lock(&i915->drm.struct_mutex);
> + err = i915_gem_object_fill_blt(obj, ce, val);
> + mutex_unlock(&i915->drm.struct_mutex);
> + if (err)
> + goto err_unpin;
> +
> + i915_gem_object_lock(obj);
> + err = i915_gem_object_set_to_cpu_domain(obj, false);
> + i915_gem_object_unlock(obj);
> + if (err)
> + goto err_unpin;
> +
> + for (i = 0; i < obj->base.size / sizeof(u32); ++i) {
> + if (vaddr[i] != val) {
> + pr_err("vaddr[%u]=%x, expected=%x\n", i,
> + vaddr[i], val);
> + err = -EINVAL;
> + goto err_unpin;
> + }
> + }
> +
> + i915_gem_object_unpin_map(obj);
> + i915_gem_object_put(obj);
> + } while (!time_after(jiffies, end));
> +
> + goto err_flush;
> +
> +err_unpin:
> + i915_gem_object_unpin_map(obj);
> +err_put:
> + i915_gem_object_put(obj);
> +err_flush:
> + mutex_lock(&i915->drm.struct_mutex);
> + if (igt_flush_test(i915, I915_WAIT_LOCKED))
> + err = -EIO;
> + mutex_unlock(&i915->drm.struct_mutex);
> +
> + if (err == -ENOMEM)
> + err = 0;
> +
> + return err;
> +}
> +
> +int i915_gem_object_blt_live_selftests(struct drm_i915_private *i915)
> +{
> + static const struct i915_subtest tests[] = {
> + SUBTEST(igt_fill_blt),
> + };
> +
> + if (i915_terminally_wedged(i915))
> + return 0;
> +
> + if (!HAS_ENGINE(i915, BCS0))
> + return 0;
> +
> + return i915_subtests(tests, i915->engine[BCS0]->kernel_context);
> +}
> diff --git a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> index a34ece53a771..7e95827b0726 100644
> --- a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> +++ b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> @@ -180,6 +180,7 @@
> #define GFX_OP_DRAWRECT_INFO_I965 ((0x7900<<16)|0x2)
>
> #define COLOR_BLT_CMD (2<<29 | 0x40<<22 | (5-2))
> +#define XY_COLOR_BLT_CMD (2<<29 | 0x50<<22)
> #define SRC_COPY_BLT_CMD ((2<<29)|(0x43<<22)|4)
> #define XY_SRC_COPY_BLT_CMD ((2<<29)|(0x53<<22)|6)
> #define XY_MONO_SRC_COPY_IMM_BLT ((2<<29)|(0x71<<22)|5)
> diff --git a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
> index 9bda36a598b3..d5dc4427d664 100644
> --- a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
> +++ b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
> @@ -25,6 +25,8 @@ selftest(gem, i915_gem_live_selftests)
> selftest(evict, i915_gem_evict_live_selftests)
> selftest(hugepages, i915_gem_huge_page_live_selftests)
> selftest(contexts, i915_gem_context_live_selftests)
> +selftest(blt, i915_gem_object_blt_live_selftests)
> +selftest(client, i915_gem_client_blt_live_selftests)
> selftest(reset, intel_reset_live_selftests)
> selftest(hangcheck, intel_hangcheck_live_selftests)
> selftest(execlists, intel_execlists_live_selftests)
Tons of XXX to be worked on, but that is the point of the patch: to
start spelling out the work that needs to be done for this approach to
even work. And there's plenty more layers we need to add tests as we
remove the lock impediments.
Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>
-Chris
More information about the Intel-gfx
mailing list