Merge tag 'gvt-fixes-2019-05-21' of https://github.com/intel/gvt-linux into drm-intel-fixes

gvt-fixes-2019-05-21

- vGPU reset fix with sane init breadcrumb (Weinan)
- Fix TRTT handling to use context state (Yan)
- Fix one error return (Dan)

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
From: Zhenyu Wang <zhenyuw@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190521062408.GH12913@zhen-hp.sh.intel.com

5 files changed, 42 insertions, 37 deletions

diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c
index ab002cfd3cab..5cb59c0b4bbe 100644
--- a/drivers/gpu/drm/i915/gvt/cmd_parser.c
+++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c
@@ -896,12 +896,16 @@ static int cmd_reg_handler(struct parser_exec_state *s,
 	}
 
 	/* TODO
-	 * Right now only scan LRI command on KBL and in inhibit context.
-	 * It's good enough to support initializing mmio by lri command in
-	 * vgpu inhibit context on KBL.
+	 * In order to let workload with inhibit context to generate
+	 * correct image data into memory, vregs values will be loaded to
+	 * hw via LRIs in the workload with inhibit context. But as
+	 * indirect context is loaded prior to LRIs in workload, we don't
+	 * want reg values specified in indirect context overwritten by
+	 * LRIs in workloads. So, when scanning an indirect context, we
+	 * update reg values in it into vregs, so LRIs in workload with
+	 * inhibit context will restore with correct values
 	 */
-	if ((IS_KABYLAKE(s->vgpu->gvt->dev_priv)
-		|| IS_COFFEELAKE(s->vgpu->gvt->dev_priv)) &&
+	if (IS_GEN(gvt->dev_priv, 9) &&
 			intel_gvt_mmio_is_in_ctx(gvt, offset) &&
 			!strncmp(cmd, "lri", 3)) {
 		intel_gvt_hypervisor_read_gpa(s->vgpu,
diff --git a/drivers/gpu/drm/i915/gvt/gtt.c b/drivers/gpu/drm/i915/gvt/gtt.c
index 08c74e65836b..244ad1729764 100644
--- a/drivers/gpu/drm/i915/gvt/gtt.c
+++ b/drivers/gpu/drm/i915/gvt/gtt.c
@@ -1076,8 +1076,10 @@ static struct intel_vgpu_ppgtt_spt *ppgtt_populate_spt_by_guest_entry(
 	} else {
 		int type = get_next_pt_type(we->type);
 
-		if (!gtt_type_is_pt(type))
+		if (!gtt_type_is_pt(type)) {
+			ret = -EINVAL;
 			goto err;
+		}
 
 		spt = ppgtt_alloc_spt_gfn(vgpu, type, ops->get_pfn(we), ips);
 		if (IS_ERR(spt)) {
diff --git a/drivers/gpu/drm/i915/gvt/handlers.c b/drivers/gpu/drm/i915/gvt/handlers.c
index 90673fca792f..e09bd6e0cc4d 100644
--- a/drivers/gpu/drm/i915/gvt/handlers.c
+++ b/drivers/gpu/drm/i915/gvt/handlers.c
@@ -1364,7 +1364,6 @@ static int dma_ctrl_write(struct intel_vgpu *vgpu, unsigned int offset,
 static int gen9_trtte_write(struct intel_vgpu *vgpu, unsigned int offset,
 		void *p_data, unsigned int bytes)
 {
-	struct drm_i915_private *dev_priv = vgpu->gvt->dev_priv;
 	u32 trtte = *(u32 *)p_data;
 
 	if ((trtte & 1) && (trtte & (1 << 1)) == 0) {
@@ -1373,11 +1372,6 @@ static int gen9_trtte_write(struct intel_vgpu *vgpu, unsigned int offset,
 		return -EINVAL;
 	}
 	write_vreg(vgpu, offset, p_data, bytes);
-	/* TRTTE is not per-context */
-
-	mmio_hw_access_pre(dev_priv);
-	I915_WRITE(_MMIO(offset), vgpu_vreg(vgpu, offset));
-	mmio_hw_access_post(dev_priv);
 
 	return 0;
 }
@@ -1385,15 +1379,6 @@ static int gen9_trtte_write(struct intel_vgpu *vgpu, unsigned int offset,
 static int gen9_trtt_chicken_write(struct intel_vgpu *vgpu, unsigned int offset,
 		void *p_data, unsigned int bytes)
 {
-	struct drm_i915_private *dev_priv = vgpu->gvt->dev_priv;
-	u32 val = *(u32 *)p_data;
-
-	if (val & 1) {
-		/* unblock hw logic */
-		mmio_hw_access_pre(dev_priv);
-		I915_WRITE(_MMIO(offset), val);
-		mmio_hw_access_post(dev_priv);
-	}
 	write_vreg(vgpu, offset, p_data, bytes);
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/gvt/mmio_context.c b/drivers/gpu/drm/i915/gvt/mmio_context.c
index edf6d646eb25..90bb3df0db50 100644
--- a/drivers/gpu/drm/i915/gvt/mmio_context.c
+++ b/drivers/gpu/drm/i915/gvt/mmio_context.c
@@ -108,12 +108,13 @@ static struct engine_mmio gen9_engine_mmio_list[] __cacheline_aligned = {
 	{RCS0, GEN9_HALF_SLICE_CHICKEN5, 0xffff, true}, /* 0xe188 */
 	{RCS0, GEN9_HALF_SLICE_CHICKEN7, 0xffff, true}, /* 0xe194 */
 	{RCS0, GEN8_ROW_CHICKEN, 0xffff, true}, /* 0xe4f0 */
-	{RCS0, TRVATTL3PTRDW(0), 0, false}, /* 0x4de0 */
-	{RCS0, TRVATTL3PTRDW(1), 0, false}, /* 0x4de4 */
-	{RCS0, TRNULLDETCT, 0, false}, /* 0x4de8 */
-	{RCS0, TRINVTILEDETCT, 0, false}, /* 0x4dec */
-	{RCS0, TRVADR, 0, false}, /* 0x4df0 */
-	{RCS0, TRTTE, 0, false}, /* 0x4df4 */
+	{RCS0, TRVATTL3PTRDW(0), 0, true}, /* 0x4de0 */
+	{RCS0, TRVATTL3PTRDW(1), 0, true}, /* 0x4de4 */
+	{RCS0, TRNULLDETCT, 0, true}, /* 0x4de8 */
+	{RCS0, TRINVTILEDETCT, 0, true}, /* 0x4dec */
+	{RCS0, TRVADR, 0, true}, /* 0x4df0 */
+	{RCS0, TRTTE, 0, true}, /* 0x4df4 */
+	{RCS0, _MMIO(0x4dfc), 0, true},
 
 	{BCS0, RING_GFX_MODE(BLT_RING_BASE), 0xffff, false}, /* 0x2229c */
 	{BCS0, RING_MI_MODE(BLT_RING_BASE), 0xffff, false}, /* 0x2209c */
@@ -392,10 +393,7 @@ static void switch_mocs(struct intel_vgpu *pre, struct intel_vgpu *next,
 	if (WARN_ON(ring_id >= ARRAY_SIZE(regs)))
 		return;
 
-	if (ring_id == RCS0 &&
-	    (IS_KABYLAKE(dev_priv) ||
-	     IS_BROXTON(dev_priv) ||
-	     IS_COFFEELAKE(dev_priv)))
+	if (ring_id == RCS0 && IS_GEN(dev_priv, 9))
 		return;
 
 	if (!pre && !gen9_render_mocs.initialized)
@@ -470,11 +468,10 @@ static void switch_mmio(struct intel_vgpu *pre,
 			continue;
 		/*
 		 * No need to do save or restore of the mmio which is in context
-		 * state image on kabylake, it's initialized by lri command and
+		 * state image on gen9, it's initialized by lri command and
 		 * save or restore with context together.
 		 */
-		if ((IS_KABYLAKE(dev_priv) || IS_BROXTON(dev_priv)
-			|| IS_COFFEELAKE(dev_priv)) && mmio->in_context)
+		if (IS_GEN(dev_priv, 9) && mmio->in_context)
 			continue;
 
 		// save
diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c
index 7c99bbc3e2b8..13632dba8b2a 100644
--- a/drivers/gpu/drm/i915/gvt/scheduler.c
+++ b/drivers/gpu/drm/i915/gvt/scheduler.c
@@ -298,12 +298,29 @@ static int copy_workload_to_ring_buffer(struct intel_vgpu_workload *workload)
 	struct i915_request *req = workload->req;
 	void *shadow_ring_buffer_va;
 	u32 *cs;
+	int err;
 
-	if ((IS_KABYLAKE(req->i915) || IS_BROXTON(req->i915)
-		|| IS_COFFEELAKE(req->i915))
-		&& is_inhibit_context(req->hw_context))
+	if (IS_GEN(req->i915, 9) && is_inhibit_context(req->hw_context))
 		intel_vgpu_restore_inhibit_context(vgpu, req);
 
+	/*
+	 * To track whether a request has started on HW, we can emit a
+	 * breadcrumb at the beginning of the request and check its
+	 * timeline's HWSP to see if the breadcrumb has advanced past the
+	 * start of this request. Actually, the request must have the
+	 * init_breadcrumb if its timeline set has_init_bread_crumb, or the
+	 * scheduler might get a wrong state of it during reset. Since the
+	 * requests from gvt always set the has_init_breadcrumb flag, here
+	 * need to do the emit_init_breadcrumb for all the requests.
+	 */
+	if (req->engine->emit_init_breadcrumb) {
+		err = req->engine->emit_init_breadcrumb(req);
+		if (err) {
+			gvt_vgpu_err("fail to emit init breadcrumb\n");
+			return err;
+		}
+	}
+
 	/* allocate shadow ring buffer */
 	cs = intel_ring_begin(workload->req, workload->rb_len / sizeof(u32));
 	if (IS_ERR(cs)) {