drm/i915: Allow a context to define its set of engines

Over the last few years, we have debated how to extend the user API to
support an increase in the number of engines, that may be sparse and
even be heterogeneous within a class (not all video decoders created
equal). We settled on using (class, instance) tuples to identify a
specific engine, with an API for the user to construct a map of engines
to capabilities. Into this picture, we then add a challenge of virtual
engines; one user engine that maps behind the scenes to any number of
physical engines. To keep it general, we want the user to have full
control over that mapping. To that end, we allow the user to constrain a
context to define the set of engines that it can access, order fully
controlled by the user via (class, instance). With such precise control
in context setup, we can continue to use the existing execbuf uABI of
specifying a single index; only now it doesn't automagically map onto
the engines, it uses the user defined engine map from the context.

v2: Fixup freeing of local on success of get_engines()
v3: Allow empty engines[]
v4: s/nengine/num_engines/
v5: Replace 64 limit on num_engines with a note that execbuf is
currently limited to only using the first 64 engines.
v6: Actually use the engines_mutex to guard the ctx->engines.

Testcase: igt/gem_ctx_engines
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190521211134.16117-2-chris@chris-wilson.co.uk

5 files changed, 310 insertions, 13 deletions

diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index 413c4529191d..c501e2848ca0 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -90,7 +90,6 @@
 #include <drm/i915_drm.h>
 
 #include "gt/intel_lrc_reg.h"
-#include "gt/intel_workarounds.h"
 
 #include "i915_drv.h"
 #include "i915_globals.h"
@@ -141,15 +140,31 @@ static void lut_close(struct i915_gem_context *ctx)
 }
 
 static struct intel_context *
-lookup_user_engine(struct i915_gem_context *ctx, u16 class, u16 instance)
+lookup_user_engine(struct i915_gem_context *ctx,
+		   unsigned long flags,
+		   const struct i915_engine_class_instance *ci)
+#define LOOKUP_USER_INDEX BIT(0)
 {
-	struct intel_engine_cs *engine;
+	int idx;
 
-	engine = intel_engine_lookup_user(ctx->i915, class, instance);
-	if (!engine)
+	if (!!(flags & LOOKUP_USER_INDEX) != i915_gem_context_user_engines(ctx))
 		return ERR_PTR(-EINVAL);
 
-	return i915_gem_context_get_engine(ctx, engine->id);
+	if (!i915_gem_context_user_engines(ctx)) {
+		struct intel_engine_cs *engine;
+
+		engine = intel_engine_lookup_user(ctx->i915,
+						  ci->engine_class,
+						  ci->engine_instance);
+		if (!engine)
+			return ERR_PTR(-EINVAL);
+
+		idx = engine->id;
+	} else {
+		idx = ci->engine_instance;
+	}
+
+	return i915_gem_context_get_engine(ctx, idx);
 }
 
 static inline int new_hw_id(struct drm_i915_private *i915, gfp_t gfp)
@@ -257,6 +272,17 @@ static void free_engines(struct i915_gem_engines *e)
 	__free_engines(e, e->num_engines);
 }
 
+static void free_engines_rcu(struct work_struct *wrk)
+{
+	struct i915_gem_engines *e =
+		container_of(wrk, struct i915_gem_engines, rcu.work);
+	struct drm_i915_private *i915 = e->i915;
+
+	mutex_lock(&i915->drm.struct_mutex);
+	free_engines(e);
+	mutex_unlock(&i915->drm.struct_mutex);
+}
+
 static struct i915_gem_engines *default_engines(struct i915_gem_context *ctx)
 {
 	struct intel_engine_cs *engine;
@@ -1352,9 +1378,7 @@ static int set_sseu(struct i915_gem_context *ctx,
 	if (user_sseu.flags || user_sseu.rsvd)
 		return -EINVAL;
 
-	ce = lookup_user_engine(ctx,
-				user_sseu.engine.engine_class,
-				user_sseu.engine.engine_instance);
+	ce = lookup_user_engine(ctx, 0, &user_sseu.engine);
 	if (IS_ERR(ce))
 		return PTR_ERR(ce);
 
@@ -1379,6 +1403,217 @@ out_ce:
 	return ret;
 }
 
+struct set_engines {
+	struct i915_gem_context *ctx;
+	struct i915_gem_engines *engines;
+};
+
+static const i915_user_extension_fn set_engines__extensions[] = {
+};
+
+static int
+set_engines(struct i915_gem_context *ctx,
+	    const struct drm_i915_gem_context_param *args)
+{
+	struct i915_context_param_engines __user *user =
+		u64_to_user_ptr(args->value);
+	struct set_engines set = { .ctx = ctx };
+	unsigned int num_engines, n;
+	u64 extensions;
+	int err;
+
+	if (!args->size) { /* switch back to legacy user_ring_map */
+		if (!i915_gem_context_user_engines(ctx))
+			return 0;
+
+		set.engines = default_engines(ctx);
+		if (IS_ERR(set.engines))
+			return PTR_ERR(set.engines);
+
+		goto replace;
+	}
+
+	BUILD_BUG_ON(!IS_ALIGNED(sizeof(*user), sizeof(*user->engines)));
+	if (args->size < sizeof(*user) ||
+	    !IS_ALIGNED(args->size, sizeof(*user->engines))) {
+		DRM_DEBUG("Invalid size for engine array: %d\n",
+			  args->size);
+		return -EINVAL;
+	}
+
+	/*
+	 * Note that I915_EXEC_RING_MASK limits execbuf to only using the
+	 * first 64 engines defined here.
+	 */
+	num_engines = (args->size - sizeof(*user)) / sizeof(*user->engines);
+
+	set.engines = kmalloc(struct_size(set.engines, engines, num_engines),
+			      GFP_KERNEL);
+	if (!set.engines)
+		return -ENOMEM;
+
+	set.engines->i915 = ctx->i915;
+	for (n = 0; n < num_engines; n++) {
+		struct i915_engine_class_instance ci;
+		struct intel_engine_cs *engine;
+
+		if (copy_from_user(&ci, &user->engines[n], sizeof(ci))) {
+			__free_engines(set.engines, n);
+			return -EFAULT;
+		}
+
+		if (ci.engine_class == (u16)I915_ENGINE_CLASS_INVALID &&
+		    ci.engine_instance == (u16)I915_ENGINE_CLASS_INVALID_NONE) {
+			set.engines->engines[n] = NULL;
+			continue;
+		}
+
+		engine = intel_engine_lookup_user(ctx->i915,
+						  ci.engine_class,
+						  ci.engine_instance);
+		if (!engine) {
+			DRM_DEBUG("Invalid engine[%d]: { class:%d, instance:%d }\n",
+				  n, ci.engine_class, ci.engine_instance);
+			__free_engines(set.engines, n);
+			return -ENOENT;
+		}
+
+		set.engines->engines[n] = intel_context_create(ctx, engine);
+		if (!set.engines->engines[n]) {
+			__free_engines(set.engines, n);
+			return -ENOMEM;
+		}
+	}
+	set.engines->num_engines = num_engines;
+
+	err = -EFAULT;
+	if (!get_user(extensions, &user->extensions))
+		err = i915_user_extensions(u64_to_user_ptr(extensions),
+					   set_engines__extensions,
+					   ARRAY_SIZE(set_engines__extensions),
+					   &set);
+	if (err) {
+		free_engines(set.engines);
+		return err;
+	}
+
+replace:
+	mutex_lock(&ctx->engines_mutex);
+	if (args->size)
+		i915_gem_context_set_user_engines(ctx);
+	else
+		i915_gem_context_clear_user_engines(ctx);
+	rcu_swap_protected(ctx->engines, set.engines, 1);
+	mutex_unlock(&ctx->engines_mutex);
+
+	INIT_RCU_WORK(&set.engines->rcu, free_engines_rcu);
+	queue_rcu_work(system_wq, &set.engines->rcu);
+
+	return 0;
+}
+
+static struct i915_gem_engines *
+__copy_engines(struct i915_gem_engines *e)
+{
+	struct i915_gem_engines *copy;
+	unsigned int n;
+
+	copy = kmalloc(struct_size(e, engines, e->num_engines), GFP_KERNEL);
+	if (!copy)
+		return ERR_PTR(-ENOMEM);
+
+	copy->i915 = e->i915;
+	for (n = 0; n < e->num_engines; n++) {
+		if (e->engines[n])
+			copy->engines[n] = intel_context_get(e->engines[n]);
+		else
+			copy->engines[n] = NULL;
+	}
+	copy->num_engines = n;
+
+	return copy;
+}
+
+static int
+get_engines(struct i915_gem_context *ctx,
+	    struct drm_i915_gem_context_param *args)
+{
+	struct i915_context_param_engines __user *user;
+	struct i915_gem_engines *e;
+	size_t n, count, size;
+	int err = 0;
+
+	err = mutex_lock_interruptible(&ctx->engines_mutex);
+	if (err)
+		return err;
+
+	e = NULL;
+	if (i915_gem_context_user_engines(ctx))
+		e = __copy_engines(i915_gem_context_engines(ctx));
+	mutex_unlock(&ctx->engines_mutex);
+	if (IS_ERR_OR_NULL(e)) {
+		args->size = 0;
+		return PTR_ERR_OR_ZERO(e);
+	}
+
+	count = e->num_engines;
+
+	/* Be paranoid in case we have an impedance mismatch */
+	if (!check_struct_size(user, engines, count, &size)) {
+		err = -EINVAL;
+		goto err_free;
+	}
+	if (overflows_type(size, args->size)) {
+		err = -EINVAL;
+		goto err_free;
+	}
+
+	if (!args->size) {
+		args->size = size;
+		goto err_free;
+	}
+
+	if (args->size < size) {
+		err = -EINVAL;
+		goto err_free;
+	}
+
+	user = u64_to_user_ptr(args->value);
+	if (!access_ok(user, size)) {
+		err = -EFAULT;
+		goto err_free;
+	}
+
+	if (put_user(0, &user->extensions)) {
+		err = -EFAULT;
+		goto err_free;
+	}
+
+	for (n = 0; n < count; n++) {
+		struct i915_engine_class_instance ci = {
+			.engine_class = I915_ENGINE_CLASS_INVALID,
+			.engine_instance = I915_ENGINE_CLASS_INVALID_NONE,
+		};
+
+		if (e->engines[n]) {
+			ci.engine_class = e->engines[n]->engine->uabi_class;
+			ci.engine_instance = e->engines[n]->engine->instance;
+		}
+
+		if (copy_to_user(&user->engines[n], &ci, sizeof(ci))) {
+			err = -EFAULT;
+			goto err_free;
+		}
+	}
+
+	args->size = size;
+
+err_free:
+	INIT_RCU_WORK(&e->rcu, free_engines_rcu);
+	queue_rcu_work(system_wq, &e->rcu);
+	return err;
+}
+
 static int ctx_setparam(struct drm_i915_file_private *fpriv,
 			struct i915_gem_context *ctx,
 			struct drm_i915_gem_context_param *args)
@@ -1452,6 +1687,10 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,
 		ret = set_ppgtt(fpriv, ctx, args);
 		break;
 
+	case I915_CONTEXT_PARAM_ENGINES:
+		ret = set_engines(ctx, args);
+		break;
+
 	case I915_CONTEXT_PARAM_BAN_PERIOD:
 	default:
 		ret = -EINVAL;
@@ -1596,9 +1835,7 @@ static int get_sseu(struct i915_gem_context *ctx,
 	if (user_sseu.flags || user_sseu.rsvd)
 		return -EINVAL;
 
-	ce = lookup_user_engine(ctx,
-				user_sseu.engine.engine_class,
-				user_sseu.engine.engine_instance);
+	ce = lookup_user_engine(ctx, 0, &user_sseu.engine);
 	if (IS_ERR(ce))
 		return PTR_ERR(ce);
 
@@ -1682,6 +1919,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data,
 		ret = get_ppgtt(file_priv, ctx, args);
 		break;
 
+	case I915_CONTEXT_PARAM_ENGINES:
+		ret = get_engines(ctx, args);
+		break;
+
 	case I915_CONTEXT_PARAM_BAN_PERIOD:
 	default:
 		ret = -EINVAL;
diff --git a/drivers/gpu/drm/i915/i915_gem_context.h b/drivers/gpu/drm/i915/i915_gem_context.h
index 272e183ebc0c..9ad4a6362438 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.h
+++ b/drivers/gpu/drm/i915/i915_gem_context.h
@@ -112,6 +112,24 @@ static inline void i915_gem_context_set_force_single_submission(struct i915_gem_
 	__set_bit(CONTEXT_FORCE_SINGLE_SUBMISSION, &ctx->flags);
 }
 
+static inline bool
+i915_gem_context_user_engines(const struct i915_gem_context *ctx)
+{
+	return test_bit(CONTEXT_USER_ENGINES, &ctx->flags);
+}
+
+static inline void
+i915_gem_context_set_user_engines(struct i915_gem_context *ctx)
+{
+	set_bit(CONTEXT_USER_ENGINES, &ctx->flags);
+}
+
+static inline void
+i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)
+{
+	clear_bit(CONTEXT_USER_ENGINES, &ctx->flags);
+}
+
 int __i915_gem_context_pin_hw_id(struct i915_gem_context *ctx);
 static inline int i915_gem_context_pin_hw_id(struct i915_gem_context *ctx)
 {
diff --git a/drivers/gpu/drm/i915/i915_gem_context_types.h b/drivers/gpu/drm/i915/i915_gem_context_types.h
index d5cb4f121aad..fb965ded2508 100644
--- a/drivers/gpu/drm/i915/i915_gem_context_types.h
+++ b/drivers/gpu/drm/i915/i915_gem_context_types.h
@@ -146,6 +146,7 @@ struct i915_gem_context {
 #define CONTEXT_BANNED			0
 #define CONTEXT_CLOSED			1
 #define CONTEXT_FORCE_SINGLE_SUBMISSION	2
+#define CONTEXT_USER_ENGINES		3
 
 	/**
 	 * @hw_id: - unique identifier for the context
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 679f7c1561ba..d6c5220addd0 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -2165,7 +2165,10 @@ eb_select_engine(struct i915_execbuffer *eb,
 	unsigned int idx;
 	int err;
 
-	idx = eb_select_legacy_ring(eb, file, args);
+	if (i915_gem_context_user_engines(eb->gem_context))
+		idx = args->flags & I915_EXEC_RING_MASK;
+	else
+		idx = eb_select_legacy_ring(eb, file, args);
 
 	ce = i915_gem_context_get_engine(eb->gem_context, idx);
 	if (IS_ERR(ce))
diff --git a/drivers/gpu/drm/i915/i915_utils.h b/drivers/gpu/drm/i915/i915_utils.h
index 5c94c7ab4607..e52866084891 100644
--- a/drivers/gpu/drm/i915/i915_utils.h
+++ b/drivers/gpu/drm/i915/i915_utils.h
@@ -26,6 +26,7 @@
 #define __I915_UTILS_H
 
 #include <linux/list.h>
+#include <linux/overflow.h>
 #include <linux/sched.h>
 #include <linux/types.h>
 #include <linux/workqueue.h>
@@ -78,6 +79,39 @@
 #define overflows_type(x, T) \
 	(sizeof(x) > sizeof(T) && (x) >> BITS_PER_TYPE(T))
 
+static inline bool
+__check_struct_size(size_t base, size_t arr, size_t count, size_t *size)
+{
+	size_t sz;
+
+	if (check_mul_overflow(count, arr, &sz))
+		return false;
+
+	if (check_add_overflow(sz, base, &sz))
+		return false;
+
+	*size = sz;
+	return true;
+}
+
+/**
+ * check_struct_size() - Calculate size of structure with trailing array.
+ * @p: Pointer to the structure.
+ * @member: Name of the array member.
+ * @n: Number of elements in the array.
+ * @sz: Total size of structure and array
+ *
+ * Calculates size of memory needed for structure @p followed by an
+ * array of @n @member elements, like struct_size() but reports
+ * whether it overflowed, and the resultant size in @sz
+ *
+ * Return: false if the calculation overflowed.
+ */
+#define check_struct_size(p, member, n, sz) \
+	likely(__check_struct_size(sizeof(*(p)), \
+				   sizeof(*(p)->member) + __must_be_array((p)->member), \
+				   n, sz))
+
 #define ptr_mask_bits(ptr, n) ({					\
 	unsigned long __v = (unsigned long)(ptr);			\
 	(typeof(ptr))(__v & -BIT(n));					\