drm/i915: Prepare for GuC-based command submission
authorAlex Dai <yu.dai@intel.com>
Wed, 12 Aug 2015 14:43:39 +0000 (15:43 +0100)
committerDaniel Vetter <daniel.vetter@ffwll.ch>
Fri, 14 Aug 2015 16:16:41 +0000 (18:16 +0200)
This adds the first of the data structures used to communicate with the
GuC (the pool of guc_context structures).

We create a GuC-specific wrapper round the GEM object allocator as all
GEM objects shared with the GuC must be pinned into GGTT space at an
address that is NOT in the range [0..WOPCM_TOP), as that range of GGTT
addresses is not accessible to the GuC (from the GuC's point of view,
it's permanently reserved for other objects such as the BootROM & SRAM).

Later, we will need to allocate additional GuC-sharable objects for the
submission client(s) and the GuC's debug log.

v2:
    Remove redundant initialisation [Chris Wilson]
    Defer adding struct members until needed [Chris Wilson]
    Local functions should pass dev_priv rather than dev [Chris Wilson]

v5:
    Invalidate GuC TLB after allocating and pinning a new object

v6:
    Rebased

Issue: VIZ-4884
Signed-off-by: Alex Dai <yu.dai@intel.com>
Signed-off-by: Dave Gordon <david.s.gordon@intel.com>
Reviewed-by: Tom O'Rourke <Tom.O'Rourke@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
drivers/gpu/drm/i915/Makefile
drivers/gpu/drm/i915/i915_guc_submission.c [new file with mode: 0644]
drivers/gpu/drm/i915/intel_guc.h
drivers/gpu/drm/i915/intel_guc_loader.c

index cc359e0e66789731b127aef430dfce5be6deec02..ddb69f337dc6e684074ca9c405932396ac789fce 100644 (file)
@@ -41,7 +41,8 @@ i915-y += i915_cmd_parser.o \
          intel_uncore.o
 
 # general-purpose microcontroller (GuC) support
-i915-y += intel_guc_loader.o
+i915-y += intel_guc_loader.o \
+         i915_guc_submission.o
 
 # autogenerated null render state
 i915-y += intel_renderstate_gen6.o \
diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c b/drivers/gpu/drm/i915/i915_guc_submission.c
new file mode 100644 (file)
index 0000000..8ff59aa
--- /dev/null
@@ -0,0 +1,118 @@
+/*
+ * Copyright © 2014 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/circ_buf.h>
+#include "i915_drv.h"
+#include "intel_guc.h"
+
+/**
+ * gem_allocate_guc_obj() - Allocate gem object for GuC usage
+ * @dev:       drm device
+ * @size:      size of object
+ *
+ * This is a wrapper to create a gem obj. In order to use it inside GuC, the
+ * object needs to be pinned lifetime. Also we must pin it to gtt space other
+ * than [0, GUC_WOPCM_TOP) because this range is reserved inside GuC.
+ *
+ * Return:     A drm_i915_gem_object if successful, otherwise NULL.
+ */
+static struct drm_i915_gem_object *gem_allocate_guc_obj(struct drm_device *dev,
+                                                       u32 size)
+{
+       struct drm_i915_private *dev_priv = dev->dev_private;
+       struct drm_i915_gem_object *obj;
+
+       obj = i915_gem_alloc_object(dev, size);
+       if (!obj)
+               return NULL;
+
+       if (i915_gem_object_get_pages(obj)) {
+               drm_gem_object_unreference(&obj->base);
+               return NULL;
+       }
+
+       if (i915_gem_obj_ggtt_pin(obj, PAGE_SIZE,
+                       PIN_OFFSET_BIAS | GUC_WOPCM_TOP)) {
+               drm_gem_object_unreference(&obj->base);
+               return NULL;
+       }
+
+       /* Invalidate GuC TLB to let GuC take the latest updates to GTT. */
+       I915_WRITE(GEN8_GTCR, GEN8_GTCR_INVALIDATE);
+
+       return obj;
+}
+
+/**
+ * gem_release_guc_obj() - Release gem object allocated for GuC usage
+ * @obj:       gem obj to be released
+  */
+static void gem_release_guc_obj(struct drm_i915_gem_object *obj)
+{
+       if (!obj)
+               return;
+
+       if (i915_gem_obj_is_pinned(obj))
+               i915_gem_object_ggtt_unpin(obj);
+
+       drm_gem_object_unreference(&obj->base);
+}
+
+/*
+ * Set up the memory resources to be shared with the GuC.  At this point,
+ * we require just one object that can be mapped through the GGTT.
+ */
+int i915_guc_submission_init(struct drm_device *dev)
+{
+       struct drm_i915_private *dev_priv = dev->dev_private;
+       const size_t ctxsize = sizeof(struct guc_context_desc);
+       const size_t poolsize = GUC_MAX_GPU_CONTEXTS * ctxsize;
+       const size_t gemsize = round_up(poolsize, PAGE_SIZE);
+       struct intel_guc *guc = &dev_priv->guc;
+
+       if (!i915.enable_guc_submission)
+               return 0; /* not enabled  */
+
+       if (guc->ctx_pool_obj)
+               return 0; /* already allocated */
+
+       guc->ctx_pool_obj = gem_allocate_guc_obj(dev_priv->dev, gemsize);
+       if (!guc->ctx_pool_obj)
+               return -ENOMEM;
+
+       ida_init(&guc->ctx_ids);
+
+       return 0;
+}
+
+void i915_guc_submission_fini(struct drm_device *dev)
+{
+       struct drm_i915_private *dev_priv = dev->dev_private;
+       struct intel_guc *guc = &dev_priv->guc;
+
+       if (guc->ctx_pool_obj)
+               ida_destroy(&guc->ctx_ids);
+       gem_release_guc_obj(guc->ctx_pool_obj);
+       guc->ctx_pool_obj = NULL;
+}
index 2846b6dfdb8d16f1e931ea589263356424f365e1..be3cad8dad9883f234ecd66e10a3e01bff39429e 100644 (file)
@@ -56,6 +56,9 @@ struct intel_guc {
        struct intel_guc_fw guc_fw;
 
        uint32_t log_flags;
+
+       struct drm_i915_gem_object *ctx_pool_obj;
+       struct ida ctx_ids;
 };
 
 /* intel_guc_loader.c */
@@ -64,4 +67,8 @@ extern int intel_guc_ucode_load(struct drm_device *dev);
 extern void intel_guc_ucode_fini(struct drm_device *dev);
 extern const char *intel_guc_fw_status_repr(enum intel_guc_fw_status status);
 
+/* i915_guc_submission.c */
+int i915_guc_submission_init(struct drm_device *dev);
+void i915_guc_submission_fini(struct drm_device *dev);
+
 #endif
index dd62c3174a9272c2a2453433885f7561d3e3dd17..6ff7feab3c26ce0431e24d9bd54110885309b591 100644 (file)
@@ -128,6 +128,21 @@ static void set_guc_init_params(struct drm_i915_private *dev_priv)
                        i915.guc_log_level << GUC_LOG_VERBOSITY_SHIFT;
        }
 
+       /* If GuC submission is enabled, set up additional parameters here */
+       if (i915.enable_guc_submission) {
+               u32 pgs = i915_gem_obj_ggtt_offset(dev_priv->guc.ctx_pool_obj);
+               u32 ctx_in_16 = GUC_MAX_GPU_CONTEXTS / 16;
+
+               pgs >>= PAGE_SHIFT;
+               params[GUC_CTL_CTXINFO] = (pgs << GUC_CTL_BASE_ADDR_SHIFT) |
+                       (ctx_in_16 << GUC_CTL_CTXNUM_IN16_SHIFT);
+
+               params[GUC_CTL_FEATURE] |= GUC_CTL_KERNEL_SUBMISSIONS;
+
+               /* Unmask this bit to enable the GuC's internal scheduler */
+               params[GUC_CTL_FEATURE] &= ~GUC_CTL_DISABLE_SCHEDULER;
+       }
+
        I915_WRITE(SOFT_SCRATCH(0), 0);
 
        for (i = 0; i < GUC_CTL_MAX_DWORDS; i++)
@@ -360,6 +375,10 @@ int intel_guc_ucode_load(struct drm_device *dev)
                break;
        }
 
+       err = i915_guc_submission_init(dev);
+       if (err)
+               goto fail;
+
        err = guc_ucode_xfer(dev_priv);
        if (err)
                goto fail;
@@ -521,6 +540,8 @@ void intel_guc_ucode_fini(struct drm_device *dev)
        struct drm_i915_private *dev_priv = dev->dev_private;
        struct intel_guc_fw *guc_fw = &dev_priv->guc.guc_fw;
 
+       i915_guc_submission_fini(dev);
+
        if (guc_fw->guc_fw_obj)
                drm_gem_object_unreference(&guc_fw->guc_fw_obj->base);
        guc_fw->guc_fw_obj = NULL;