#include "i915_drv.h"
#include "i915_trace.h"
+#define ALL_L3_SLICES(dev) (1 << NUM_L3_SLICES(dev)) - 1
+
/* This is a HW constraint. The value below is the largest known requirement
* I've seen in a spec to date, and that was a workaround for a non-shipping
* part. It should be safe to decrease this, but it's more future proof as is.
#define GEN6_CONTEXT_ALIGN (64<<10)
#define GEN7_CONTEXT_ALIGN 4096
-static size_t get_context_alignment(struct drm_device *dev)
+static size_t get_context_alignment(struct drm_i915_private *dev_priv)
{
- if (IS_GEN6(dev))
+ if (IS_GEN6(dev_priv))
return GEN6_CONTEXT_ALIGN;
return GEN7_CONTEXT_ALIGN;
}
-static int get_context_size(struct drm_device *dev)
+static int get_context_size(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = dev->dev_private;
int ret;
u32 reg;
- switch (INTEL_INFO(dev)->gen) {
+ switch (INTEL_GEN(dev_priv)) {
case 6:
reg = I915_READ(CXT_SIZE);
ret = GEN6_CXT_TOTAL_SIZE(reg) * 64;
break;
case 7:
reg = I915_READ(GEN7_CXT_SIZE);
- if (IS_HASWELL(dev))
+ if (IS_HASWELL(dev_priv))
ret = HSW_CXT_TOTAL_SIZE;
else
ret = GEN7_CXT_TOTAL_SIZE(reg) * 64;
return ret;
}
-static void i915_gem_context_clean(struct intel_context *ctx)
+static void i915_gem_context_clean(struct i915_gem_context *ctx)
{
struct i915_hw_ppgtt *ppgtt = ctx->ppgtt;
struct i915_vma *vma, *next;
void i915_gem_context_free(struct kref *ctx_ref)
{
- struct intel_context *ctx = container_of(ctx_ref, typeof(*ctx), ref);
+ struct i915_gem_context *ctx = container_of(ctx_ref, typeof(*ctx), ref);
trace_i915_context_free(ctx);
if (ctx->legacy_hw_ctx.rcs_state)
drm_gem_object_unreference(&ctx->legacy_hw_ctx.rcs_state->base);
list_del(&ctx->link);
+
+ ida_simple_remove(&ctx->i915->context_hw_ida, ctx->hw_id);
kfree(ctx);
}
struct drm_i915_gem_object *obj;
int ret;
- obj = i915_gem_alloc_object(dev, size);
- if (obj == NULL)
- return ERR_PTR(-ENOMEM);
+ obj = i915_gem_object_create(dev, size);
+ if (IS_ERR(obj))
+ return obj;
/*
* Try to make the context utilize L3 as well as LLC.
return obj;
}
-static struct intel_context *
+static int assign_hw_id(struct drm_i915_private *dev_priv, unsigned *out)
+{
+ int ret;
+
+ ret = ida_simple_get(&dev_priv->context_hw_ida,
+ 0, MAX_CONTEXT_HW_ID, GFP_KERNEL);
+ if (ret < 0) {
+ /* Contexts are only released when no longer active.
+ * Flush any pending retires to hopefully release some
+ * stale contexts and try again.
+ */
+ i915_gem_retire_requests(dev_priv);
+ ret = ida_simple_get(&dev_priv->context_hw_ida,
+ 0, MAX_CONTEXT_HW_ID, GFP_KERNEL);
+ if (ret < 0)
+ return ret;
+ }
+
+ *out = ret;
+ return 0;
+}
+
+static struct i915_gem_context *
__create_hw_context(struct drm_device *dev,
struct drm_i915_file_private *file_priv)
{
struct drm_i915_private *dev_priv = dev->dev_private;
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
int ret;
ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
if (ctx == NULL)
return ERR_PTR(-ENOMEM);
+ ret = assign_hw_id(dev_priv, &ctx->hw_id);
+ if (ret) {
+ kfree(ctx);
+ return ERR_PTR(ret);
+ }
+
kref_init(&ctx->ref);
list_add_tail(&ctx->link, &dev_priv->context_list);
ctx->i915 = dev_priv;
/* NB: Mark all slices as needing a remap so that when the context first
* loads it will restore whatever remap state already exists. If there
* is no remap info, it will be a NOP. */
- ctx->remap_slice = (1 << NUM_L3_SLICES(dev)) - 1;
+ ctx->remap_slice = ALL_L3_SLICES(dev_priv);
ctx->hang_stats.ban_period_seconds = DRM_I915_CTX_BAN_PERIOD;
* context state of the GPU for applications that don't utilize HW contexts, as
* well as an idle case.
*/
-static struct intel_context *
+static struct i915_gem_context *
i915_gem_create_context(struct drm_device *dev,
struct drm_i915_file_private *file_priv)
{
const bool is_global_default_ctx = file_priv == NULL;
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
int ret = 0;
BUG_ON(!mutex_is_locked(&dev->struct_mutex));
* context.
*/
ret = i915_gem_obj_ggtt_pin(ctx->legacy_hw_ctx.rcs_state,
- get_context_alignment(dev), 0);
+ get_context_alignment(to_i915(dev)), 0);
if (ret) {
DRM_DEBUG_DRIVER("Couldn't pin %d\n", ret);
goto err_destroy;
return ERR_PTR(ret);
}
-static void i915_gem_context_unpin(struct intel_context *ctx,
+static void i915_gem_context_unpin(struct i915_gem_context *ctx,
struct intel_engine_cs *engine)
{
if (i915.enable_execlists) {
void i915_gem_context_reset(struct drm_device *dev)
{
struct drm_i915_private *dev_priv = dev->dev_private;
- int i;
if (i915.enable_execlists) {
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
list_for_each_entry(ctx, &dev_priv->context_list, link)
- intel_lr_context_reset(dev, ctx);
+ intel_lr_context_reset(dev_priv, ctx);
}
- for (i = 0; i < I915_NUM_RINGS; i++) {
- struct intel_engine_cs *ring = &dev_priv->ring[i];
-
- if (ring->last_context) {
- i915_gem_context_unpin(ring->last_context, ring);
- ring->last_context = NULL;
- }
- }
-
- /* Force the GPU state to be reinitialised on enabling */
- dev_priv->kernel_context->legacy_hw_ctx.initialized = false;
+ i915_gem_context_lost(dev_priv);
}
int i915_gem_context_init(struct drm_device *dev)
{
struct drm_i915_private *dev_priv = dev->dev_private;
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
/* Init should only be called once per module load. Eventually the
* restriction on the context_disabled check can be loosened. */
if (WARN_ON(dev_priv->kernel_context))
return 0;
- if (intel_vgpu_active(dev) && HAS_LOGICAL_RING_CONTEXTS(dev)) {
+ if (intel_vgpu_active(dev_priv) &&
+ HAS_LOGICAL_RING_CONTEXTS(dev_priv)) {
if (!i915.enable_execlists) {
DRM_INFO("Only EXECLIST mode is supported in vgpu.\n");
return -EINVAL;
}
}
+ /* Using the simple ida interface, the max is limited by sizeof(int) */
+ BUILD_BUG_ON(MAX_CONTEXT_HW_ID > INT_MAX);
+ ida_init(&dev_priv->context_hw_ida);
+
if (i915.enable_execlists) {
/* NB: intentionally left blank. We will allocate our own
* backing objects as we need them, thank you very much */
dev_priv->hw_context_size = 0;
- } else if (HAS_HW_CONTEXTS(dev)) {
- dev_priv->hw_context_size = round_up(get_context_size(dev), 4096);
+ } else if (HAS_HW_CONTEXTS(dev_priv)) {
+ dev_priv->hw_context_size =
+ round_up(get_context_size(dev_priv), 4096);
if (dev_priv->hw_context_size > (1<<20)) {
DRM_DEBUG_DRIVER("Disabling HW Contexts; invalid size %d\n",
dev_priv->hw_context_size);
return 0;
}
-void i915_gem_context_fini(struct drm_device *dev)
+void i915_gem_context_lost(struct drm_i915_private *dev_priv)
{
- struct drm_i915_private *dev_priv = dev->dev_private;
- struct intel_context *dctx = dev_priv->kernel_context;
- int i;
-
- if (dctx->legacy_hw_ctx.rcs_state) {
- /* The only known way to stop the gpu from accessing the hw context is
- * to reset it. Do this as the very last operation to avoid confusing
- * other code, leading to spurious errors. */
- intel_gpu_reset(dev);
-
- /* When default context is created and switched to, base object refcount
- * will be 2 (+1 from object creation and +1 from do_switch()).
- * i915_gem_context_fini() will be called after gpu_idle() has switched
- * to default context. So we need to unreference the base object once
- * to offset the do_switch part, so that i915_gem_context_unreference()
- * can then free the base object correctly. */
- WARN_ON(!dev_priv->ring[RCS].last_context);
-
- i915_gem_object_ggtt_unpin(dctx->legacy_hw_ctx.rcs_state);
- }
+ struct intel_engine_cs *engine;
- for (i = I915_NUM_RINGS; --i >= 0;) {
- struct intel_engine_cs *ring = &dev_priv->ring[i];
+ for_each_engine(engine, dev_priv) {
+ if (engine->last_context == NULL)
+ continue;
- if (ring->last_context) {
- i915_gem_context_unpin(ring->last_context, ring);
- ring->last_context = NULL;
- }
+ i915_gem_context_unpin(engine->last_context, engine);
+ engine->last_context = NULL;
}
- i915_gem_context_unreference(dctx);
- dev_priv->kernel_context = NULL;
+ /* Force the GPU state to be reinitialised on enabling */
+ dev_priv->kernel_context->legacy_hw_ctx.initialized = false;
+ dev_priv->kernel_context->remap_slice = ALL_L3_SLICES(dev_priv);
}
-int i915_gem_context_enable(struct drm_i915_gem_request *req)
+void i915_gem_context_fini(struct drm_device *dev)
{
- struct intel_engine_cs *ring = req->ring;
- int ret;
-
- if (i915.enable_execlists) {
- if (ring->init_context == NULL)
- return 0;
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ struct i915_gem_context *dctx = dev_priv->kernel_context;
- ret = ring->init_context(req);
- } else
- ret = i915_switch_context(req);
+ if (dctx->legacy_hw_ctx.rcs_state)
+ i915_gem_object_ggtt_unpin(dctx->legacy_hw_ctx.rcs_state);
- if (ret) {
- DRM_ERROR("ring init context: %d\n", ret);
- return ret;
- }
+ i915_gem_context_unreference(dctx);
+ dev_priv->kernel_context = NULL;
- return 0;
+ ida_destroy(&dev_priv->context_hw_ida);
}
static int context_idr_cleanup(int id, void *p, void *data)
{
- struct intel_context *ctx = p;
+ struct i915_gem_context *ctx = p;
i915_gem_context_unreference(ctx);
return 0;
int i915_gem_context_open(struct drm_device *dev, struct drm_file *file)
{
struct drm_i915_file_private *file_priv = file->driver_priv;
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
idr_init(&file_priv->context_idr);
idr_destroy(&file_priv->context_idr);
}
-struct intel_context *
+struct i915_gem_context *
i915_gem_context_get(struct drm_i915_file_private *file_priv, u32 id)
{
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
- ctx = (struct intel_context *)idr_find(&file_priv->context_idr, id);
+ ctx = idr_find(&file_priv->context_idr, id);
if (!ctx)
return ERR_PTR(-ENOENT);
static inline int
mi_set_context(struct drm_i915_gem_request *req, u32 hw_flags)
{
- struct intel_engine_cs *ring = req->ring;
+ struct drm_i915_private *dev_priv = req->i915;
+ struct intel_engine_cs *engine = req->engine;
u32 flags = hw_flags | MI_MM_SPACE_GTT;
const int num_rings =
/* Use an extended w/a on ivb+ if signalling from other rings */
- i915_semaphore_is_enabled(ring->dev) ?
- hweight32(INTEL_INFO(ring->dev)->ring_mask) - 1 :
+ i915_semaphore_is_enabled(dev_priv) ?
+ hweight32(INTEL_INFO(dev_priv)->ring_mask) - 1 :
0;
- int len, i, ret;
+ int len, ret;
/* w/a: If Flush TLB Invalidation Mode is enabled, driver must do a TLB
* invalidation prior to MI_SET_CONTEXT. On GEN6 we don't set the value
* explicitly, so we rely on the value at ring init, stored in
* itlb_before_ctx_switch.
*/
- if (IS_GEN6(ring->dev)) {
- ret = ring->flush(req, I915_GEM_GPU_DOMAINS, 0);
+ if (IS_GEN6(dev_priv)) {
+ ret = engine->flush(req, I915_GEM_GPU_DOMAINS, 0);
if (ret)
return ret;
}
/* These flags are for resource streamer on HSW+ */
- if (IS_HASWELL(ring->dev) || INTEL_INFO(ring->dev)->gen >= 8)
+ if (IS_HASWELL(dev_priv) || INTEL_GEN(dev_priv) >= 8)
flags |= (HSW_MI_RS_SAVE_STATE_EN | HSW_MI_RS_RESTORE_STATE_EN);
- else if (INTEL_INFO(ring->dev)->gen < 8)
+ else if (INTEL_GEN(dev_priv) < 8)
flags |= (MI_SAVE_EXT_STATE_EN | MI_RESTORE_EXT_STATE_EN);
len = 4;
- if (INTEL_INFO(ring->dev)->gen >= 7)
- len += 2 + (num_rings ? 4*num_rings + 2 : 0);
+ if (INTEL_GEN(dev_priv) >= 7)
+ len += 2 + (num_rings ? 4*num_rings + 6 : 0);
ret = intel_ring_begin(req, len);
if (ret)
return ret;
/* WaProgramMiArbOnOffAroundMiSetContext:ivb,vlv,hsw,bdw,chv */
- if (INTEL_INFO(ring->dev)->gen >= 7) {
- intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_DISABLE);
+ if (INTEL_GEN(dev_priv) >= 7) {
+ intel_ring_emit(engine, MI_ARB_ON_OFF | MI_ARB_DISABLE);
if (num_rings) {
struct intel_engine_cs *signaller;
- intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(num_rings));
- for_each_ring(signaller, to_i915(ring->dev), i) {
- if (signaller == ring)
+ intel_ring_emit(engine,
+ MI_LOAD_REGISTER_IMM(num_rings));
+ for_each_engine(signaller, dev_priv) {
+ if (signaller == engine)
continue;
- intel_ring_emit_reg(ring, RING_PSMI_CTL(signaller->mmio_base));
- intel_ring_emit(ring, _MASKED_BIT_ENABLE(GEN6_PSMI_SLEEP_MSG_DISABLE));
+ intel_ring_emit_reg(engine,
+ RING_PSMI_CTL(signaller->mmio_base));
+ intel_ring_emit(engine,
+ _MASKED_BIT_ENABLE(GEN6_PSMI_SLEEP_MSG_DISABLE));
}
}
}
- intel_ring_emit(ring, MI_NOOP);
- intel_ring_emit(ring, MI_SET_CONTEXT);
- intel_ring_emit(ring, i915_gem_obj_ggtt_offset(req->ctx->legacy_hw_ctx.rcs_state) |
+ intel_ring_emit(engine, MI_NOOP);
+ intel_ring_emit(engine, MI_SET_CONTEXT);
+ intel_ring_emit(engine,
+ i915_gem_obj_ggtt_offset(req->ctx->legacy_hw_ctx.rcs_state) |
flags);
/*
* w/a: MI_SET_CONTEXT must always be followed by MI_NOOP
* WaMiSetContext_Hang:snb,ivb,vlv
*/
- intel_ring_emit(ring, MI_NOOP);
+ intel_ring_emit(engine, MI_NOOP);
- if (INTEL_INFO(ring->dev)->gen >= 7) {
+ if (INTEL_GEN(dev_priv) >= 7) {
if (num_rings) {
struct intel_engine_cs *signaller;
+ i915_reg_t last_reg = {}; /* keep gcc quiet */
- intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(num_rings));
- for_each_ring(signaller, to_i915(ring->dev), i) {
- if (signaller == ring)
+ intel_ring_emit(engine,
+ MI_LOAD_REGISTER_IMM(num_rings));
+ for_each_engine(signaller, dev_priv) {
+ if (signaller == engine)
continue;
- intel_ring_emit_reg(ring, RING_PSMI_CTL(signaller->mmio_base));
- intel_ring_emit(ring, _MASKED_BIT_DISABLE(GEN6_PSMI_SLEEP_MSG_DISABLE));
+ last_reg = RING_PSMI_CTL(signaller->mmio_base);
+ intel_ring_emit_reg(engine, last_reg);
+ intel_ring_emit(engine,
+ _MASKED_BIT_DISABLE(GEN6_PSMI_SLEEP_MSG_DISABLE));
}
+
+ /* Insert a delay before the next switch! */
+ intel_ring_emit(engine,
+ MI_STORE_REGISTER_MEM |
+ MI_SRM_LRM_GLOBAL_GTT);
+ intel_ring_emit_reg(engine, last_reg);
+ intel_ring_emit(engine, engine->scratch.gtt_offset);
+ intel_ring_emit(engine, MI_NOOP);
}
- intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_ENABLE);
+ intel_ring_emit(engine, MI_ARB_ON_OFF | MI_ARB_ENABLE);
}
- intel_ring_advance(ring);
+ intel_ring_advance(engine);
return ret;
}
-static inline bool should_skip_switch(struct intel_engine_cs *ring,
- struct intel_context *from,
- struct intel_context *to)
+static int remap_l3(struct drm_i915_gem_request *req, int slice)
+{
+ u32 *remap_info = req->i915->l3_parity.remap_info[slice];
+ struct intel_engine_cs *engine = req->engine;
+ int i, ret;
+
+ if (!remap_info)
+ return 0;
+
+ ret = intel_ring_begin(req, GEN7_L3LOG_SIZE/4 * 2 + 2);
+ if (ret)
+ return ret;
+
+ /*
+ * Note: We do not worry about the concurrent register cacheline hang
+ * here because no other code should access these registers other than
+ * at initialization time.
+ */
+ intel_ring_emit(engine, MI_LOAD_REGISTER_IMM(GEN7_L3LOG_SIZE/4));
+ for (i = 0; i < GEN7_L3LOG_SIZE/4; i++) {
+ intel_ring_emit_reg(engine, GEN7_L3LOG(slice, i));
+ intel_ring_emit(engine, remap_info[i]);
+ }
+ intel_ring_emit(engine, MI_NOOP);
+ intel_ring_advance(engine);
+
+ return 0;
+}
+
+static inline bool skip_rcs_switch(struct i915_hw_ppgtt *ppgtt,
+ struct intel_engine_cs *engine,
+ struct i915_gem_context *to)
{
if (to->remap_slice)
return false;
- if (to->ppgtt && from == to &&
- !(intel_ring_flag(ring) & to->ppgtt->pd_dirty_rings))
- return true;
+ if (!to->legacy_hw_ctx.initialized)
+ return false;
- return false;
+ if (ppgtt && (intel_engine_flag(engine) & ppgtt->pd_dirty_rings))
+ return false;
+
+ return to == engine->last_context;
}
static bool
-needs_pd_load_pre(struct intel_engine_cs *ring, struct intel_context *to)
+needs_pd_load_pre(struct i915_hw_ppgtt *ppgtt,
+ struct intel_engine_cs *engine,
+ struct i915_gem_context *to)
{
- struct drm_i915_private *dev_priv = ring->dev->dev_private;
+ if (!ppgtt)
+ return false;
+
+ /* Always load the ppgtt on first use */
+ if (!engine->last_context)
+ return true;
- if (!to->ppgtt)
+ /* Same context without new entries, skip */
+ if (engine->last_context == to &&
+ !(intel_engine_flag(engine) & ppgtt->pd_dirty_rings))
return false;
- if (INTEL_INFO(ring->dev)->gen < 8)
+ if (engine->id != RCS)
return true;
- if (ring != &dev_priv->ring[RCS])
+ if (INTEL_GEN(engine->i915) < 8)
return true;
return false;
}
static bool
-needs_pd_load_post(struct intel_engine_cs *ring, struct intel_context *to,
- u32 hw_flags)
+needs_pd_load_post(struct i915_hw_ppgtt *ppgtt,
+ struct i915_gem_context *to,
+ u32 hw_flags)
{
- struct drm_i915_private *dev_priv = ring->dev->dev_private;
-
- if (!to->ppgtt)
- return false;
-
- if (!IS_GEN8(ring->dev))
+ if (!ppgtt)
return false;
- if (ring != &dev_priv->ring[RCS])
+ if (!IS_GEN8(to->i915))
return false;
if (hw_flags & MI_RESTORE_INHIBIT)
return false;
}
-static int do_switch(struct drm_i915_gem_request *req)
+static int do_rcs_switch(struct drm_i915_gem_request *req)
{
- struct intel_context *to = req->ctx;
- struct intel_engine_cs *ring = req->ring;
- struct drm_i915_private *dev_priv = ring->dev->dev_private;
- struct intel_context *from = ring->last_context;
- u32 hw_flags = 0;
- bool uninitialized = false;
+ struct i915_gem_context *to = req->ctx;
+ struct intel_engine_cs *engine = req->engine;
+ struct i915_hw_ppgtt *ppgtt = to->ppgtt ?: req->i915->mm.aliasing_ppgtt;
+ struct i915_gem_context *from;
+ u32 hw_flags;
int ret, i;
- if (from != NULL && ring == &dev_priv->ring[RCS]) {
- BUG_ON(from->legacy_hw_ctx.rcs_state == NULL);
- BUG_ON(!i915_gem_obj_is_pinned(from->legacy_hw_ctx.rcs_state));
- }
-
- if (should_skip_switch(ring, from, to))
+ if (skip_rcs_switch(ppgtt, engine, to))
return 0;
/* Trying to pin first makes error handling easier. */
- if (ring == &dev_priv->ring[RCS]) {
- ret = i915_gem_obj_ggtt_pin(to->legacy_hw_ctx.rcs_state,
- get_context_alignment(ring->dev), 0);
- if (ret)
- return ret;
- }
+ ret = i915_gem_obj_ggtt_pin(to->legacy_hw_ctx.rcs_state,
+ get_context_alignment(engine->i915),
+ 0);
+ if (ret)
+ return ret;
/*
* Pin can switch back to the default context if we end up calling into
* evict_everything - as a last ditch gtt defrag effort that also
* switches to the default context. Hence we need to reload from here.
+ *
+ * XXX: Doing so is painfully broken!
*/
- from = ring->last_context;
-
- if (needs_pd_load_pre(ring, to)) {
- /* Older GENs and non render rings still want the load first,
- * "PP_DCLV followed by PP_DIR_BASE register through Load
- * Register Immediate commands in Ring Buffer before submitting
- * a context."*/
- trace_switch_mm(ring, to);
- ret = to->ppgtt->switch_mm(to->ppgtt, req);
- if (ret)
- goto unpin_out;
-
- /* Doing a PD load always reloads the page dirs */
- to->ppgtt->pd_dirty_rings &= ~intel_ring_flag(ring);
- }
-
- if (ring != &dev_priv->ring[RCS]) {
- if (from)
- i915_gem_context_unreference(from);
- goto done;
- }
+ from = engine->last_context;
/*
* Clear this page out of any CPU caches for coherent swap-in/out. Note
if (ret)
goto unpin_out;
- if (!to->legacy_hw_ctx.initialized || i915_gem_context_is_default(to)) {
- hw_flags |= MI_RESTORE_INHIBIT;
+ if (needs_pd_load_pre(ppgtt, engine, to)) {
+ /* Older GENs and non render rings still want the load first,
+ * "PP_DCLV followed by PP_DIR_BASE register through Load
+ * Register Immediate commands in Ring Buffer before submitting
+ * a context."*/
+ trace_switch_mm(engine, to);
+ ret = ppgtt->switch_mm(ppgtt, req);
+ if (ret)
+ goto unpin_out;
+ }
+
+ if (!to->legacy_hw_ctx.initialized || i915_gem_context_is_default(to))
/* NB: If we inhibit the restore, the context is not allowed to
* die because future work may end up depending on valid address
* space. This means we must enforce that a page table load
* occur when this occurs. */
- } else if (to->ppgtt &&
- (intel_ring_flag(ring) & to->ppgtt->pd_dirty_rings)) {
- hw_flags |= MI_FORCE_RESTORE;
- to->ppgtt->pd_dirty_rings &= ~intel_ring_flag(ring);
- }
-
- /* We should never emit switch_mm more than once */
- WARN_ON(needs_pd_load_pre(ring, to) &&
- needs_pd_load_post(ring, to, hw_flags));
-
- ret = mi_set_context(req, hw_flags);
- if (ret)
- goto unpin_out;
-
- /* GEN8 does *not* require an explicit reload if the PDPs have been
- * setup, and we do not wish to move them.
- */
- if (needs_pd_load_post(ring, to, hw_flags)) {
- trace_switch_mm(ring, to);
- ret = to->ppgtt->switch_mm(to->ppgtt, req);
- /* The hardware context switch is emitted, but we haven't
- * actually changed the state - so it's probably safe to bail
- * here. Still, let the user know something dangerous has
- * happened.
- */
- if (ret) {
- DRM_ERROR("Failed to change address space on context switch\n");
- goto unpin_out;
- }
- }
-
- for (i = 0; i < MAX_L3_SLICES; i++) {
- if (!(to->remap_slice & (1<<i)))
- continue;
-
- ret = i915_gem_l3_remap(req, i);
- /* If it failed, try again next round */
+ hw_flags = MI_RESTORE_INHIBIT;
+ else if (ppgtt && intel_engine_flag(engine) & ppgtt->pd_dirty_rings)
+ hw_flags = MI_FORCE_RESTORE;
+ else
+ hw_flags = 0;
+
+ if (to != from || (hw_flags & MI_FORCE_RESTORE)) {
+ ret = mi_set_context(req, hw_flags);
if (ret)
- DRM_DEBUG_DRIVER("L3 remapping failed\n");
- else
- to->remap_slice &= ~(1<<i);
+ goto unpin_out;
}
/* The backing object for the context is done after switching to the
i915_gem_object_ggtt_unpin(from->legacy_hw_ctx.rcs_state);
i915_gem_context_unreference(from);
}
+ i915_gem_context_reference(to);
+ engine->last_context = to;
+
+ /* GEN8 does *not* require an explicit reload if the PDPs have been
+ * setup, and we do not wish to move them.
+ */
+ if (needs_pd_load_post(ppgtt, to, hw_flags)) {
+ trace_switch_mm(engine, to);
+ ret = ppgtt->switch_mm(ppgtt, req);
+ /* The hardware context switch is emitted, but we haven't
+ * actually changed the state - so it's probably safe to bail
+ * here. Still, let the user know something dangerous has
+ * happened.
+ */
+ if (ret)
+ return ret;
+ }
- uninitialized = !to->legacy_hw_ctx.initialized;
- to->legacy_hw_ctx.initialized = true;
+ if (ppgtt)
+ ppgtt->pd_dirty_rings &= ~intel_engine_flag(engine);
-done:
- i915_gem_context_reference(to);
- ring->last_context = to;
+ for (i = 0; i < MAX_L3_SLICES; i++) {
+ if (!(to->remap_slice & (1<<i)))
+ continue;
- if (uninitialized) {
- if (ring->init_context) {
- ret = ring->init_context(req);
+ ret = remap_l3(req, i);
+ if (ret)
+ return ret;
+
+ to->remap_slice &= ~(1<<i);
+ }
+
+ if (!to->legacy_hw_ctx.initialized) {
+ if (engine->init_context) {
+ ret = engine->init_context(req);
if (ret)
- DRM_ERROR("ring init context: %d\n", ret);
+ return ret;
}
+ to->legacy_hw_ctx.initialized = true;
}
return 0;
unpin_out:
- if (ring->id == RCS)
- i915_gem_object_ggtt_unpin(to->legacy_hw_ctx.rcs_state);
+ i915_gem_object_ggtt_unpin(to->legacy_hw_ctx.rcs_state);
return ret;
}
*/
int i915_switch_context(struct drm_i915_gem_request *req)
{
- struct intel_engine_cs *ring = req->ring;
- struct drm_i915_private *dev_priv = ring->dev->dev_private;
+ struct intel_engine_cs *engine = req->engine;
+ struct drm_i915_private *dev_priv = req->i915;
WARN_ON(i915.enable_execlists);
WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex));
- if (req->ctx->legacy_hw_ctx.rcs_state == NULL) { /* We have the fake context */
- if (req->ctx != ring->last_context) {
- i915_gem_context_reference(req->ctx);
- if (ring->last_context)
- i915_gem_context_unreference(ring->last_context);
- ring->last_context = req->ctx;
+ if (engine->id != RCS ||
+ req->ctx->legacy_hw_ctx.rcs_state == NULL) {
+ struct i915_gem_context *to = req->ctx;
+ struct i915_hw_ppgtt *ppgtt =
+ to->ppgtt ?: req->i915->mm.aliasing_ppgtt;
+
+ if (needs_pd_load_pre(ppgtt, engine, to)) {
+ int ret;
+
+ trace_switch_mm(engine, to);
+ ret = ppgtt->switch_mm(ppgtt, req);
+ if (ret)
+ return ret;
+
+ ppgtt->pd_dirty_rings &= ~intel_engine_flag(engine);
+ }
+
+ if (to != engine->last_context) {
+ i915_gem_context_reference(to);
+ if (engine->last_context)
+ i915_gem_context_unreference(engine->last_context);
+ engine->last_context = to;
}
+
return 0;
}
- return do_switch(req);
+ return do_rcs_switch(req);
}
static bool contexts_enabled(struct drm_device *dev)
{
struct drm_i915_gem_context_create *args = data;
struct drm_i915_file_private *file_priv = file->driver_priv;
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
int ret;
if (!contexts_enabled(dev))
{
struct drm_i915_gem_context_destroy *args = data;
struct drm_i915_file_private *file_priv = file->driver_priv;
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
int ret;
if (args->pad != 0)
{
struct drm_i915_file_private *file_priv = file->driver_priv;
struct drm_i915_gem_context_param *args = data;
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
int ret;
ret = i915_mutex_lock_interruptible(dev);
else if (to_i915(dev)->mm.aliasing_ppgtt)
args->value = to_i915(dev)->mm.aliasing_ppgtt->base.total;
else
- args->value = to_i915(dev)->gtt.base.total;
+ args->value = to_i915(dev)->ggtt.base.total;
break;
default:
ret = -EINVAL;
{
struct drm_i915_file_private *file_priv = file->driver_priv;
struct drm_i915_gem_context_param *args = data;
- struct intel_context *ctx;
+ struct i915_gem_context *ctx;
int ret;
ret = i915_mutex_lock_interruptible(dev);
return ret;
}
+
+int i915_gem_context_reset_stats_ioctl(struct drm_device *dev,
+ void *data, struct drm_file *file)
+{
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_reset_stats *args = data;
+ struct i915_ctx_hang_stats *hs;
+ struct i915_gem_context *ctx;
+ int ret;
+
+ if (args->flags || args->pad)
+ return -EINVAL;
+
+ if (args->ctx_id == DEFAULT_CONTEXT_HANDLE && !capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ ret = i915_mutex_lock_interruptible(dev);
+ if (ret)
+ return ret;
+
+ ctx = i915_gem_context_get(file->driver_priv, args->ctx_id);
+ if (IS_ERR(ctx)) {
+ mutex_unlock(&dev->struct_mutex);
+ return PTR_ERR(ctx);
+ }
+ hs = &ctx->hang_stats;
+
+ if (capable(CAP_SYS_ADMIN))
+ args->reset_count = i915_reset_count(&dev_priv->gpu_error);
+ else
+ args->reset_count = 0;
+
+ args->batch_active = hs->batch_active;
+ args->batch_pending = hs->batch_pending;
+
+ mutex_unlock(&dev->struct_mutex);
+
+ return 0;
+}