atomize viewport

4 files changed, 126 insertions, 28 deletions

diff --git a/r600_atom.c b/r600_atom.c
index 16499bd..4f5df7b 100644
--- a/r600_atom.c
+++ b/r600_atom.c
@@ -84,17 +84,18 @@ static int r600_cb_emit(struct radeon_device *rdev,
 	cb->pkts[16] = radeon_ib_reloc(ib, cb->bo, cb->placements[0] | cb->placements[1]);
 	cb->pkts[21] = radeon_ib_reloc(ib, cb->bo, cb->placements[0] | cb->placements[1]);
 	cb->pkts[26] = radeon_ib_reloc(ib, cb->bo, cb->placements[0] | cb->placements[1]);
-	cb->pkts[271] = radeon_ib_reloc(ib, cb->vsshader1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[276] = radeon_ib_reloc(ib, cb->vsshader1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[289] = radeon_ib_reloc(ib, cb->vsshader1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[294] = radeon_ib_reloc(ib, cb->vsshader1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[310] = radeon_ib_reloc(ib, cb->psshader1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[315] = radeon_ib_reloc(ib, cb->psshader1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[358] = radeon_ib_reloc(ib, cb->vbo1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[369] = radeon_ib_reloc(ib, cb->vbo1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[376] = radeon_ib_reloc(ib, cb->vbo1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[387] = radeon_ib_reloc(ib, cb->vbo1, RADEON_GEM_DOMAIN_GTT);
-	cb->pkts[409] = radeon_ib_reloc(ib, cb->bo, cb->placements[0] | cb->placements[1]);
+	cb->pkts[259] = radeon_ib_reloc(ib, cb->vsshader1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[264] = radeon_ib_reloc(ib, cb->vsshader1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[277] = radeon_ib_reloc(ib, cb->vsshader1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[282] = radeon_ib_reloc(ib, cb->vsshader1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[298] = radeon_ib_reloc(ib, cb->psshader1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[303] = radeon_ib_reloc(ib, cb->psshader1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[346] = radeon_ib_reloc(ib, cb->vbo1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[357] = radeon_ib_reloc(ib, cb->vbo1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[364] = radeon_ib_reloc(ib, cb->vbo1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[375] = radeon_ib_reloc(ib, cb->vbo1, RADEON_GEM_DOMAIN_GTT);
+	cb->pkts[397] = radeon_ib_reloc(ib, cb->bo, cb->placements[0] | cb->placements[1]);
+
 	r = radeon_ib_copy(ib, cb->pkts, atom->npkts);
 	return r;
 }
@@ -590,6 +591,71 @@ out_err:
 }
 
 /*
+ * r600_vport
+ */
+struct r600_vport {
+	struct radeon_atom	atom;
+	u32			pkts[16];
+	struct r600_atoms	*atoms;
+};
+
+static void r600_vport_release(struct kref *kref)
+{
+	struct radeon_atom *atom = container_of(kref, struct radeon_atom, kref);
+	struct r600_vport *vport = container_of(atom, struct r600_vport, atom);
+
+	mutex_lock(&vport->atoms->mutex);
+	list_del_init(&vport->atom.list);
+	mutex_unlock(&vport->atoms->mutex);
+	kfree(vport);
+}
+
+static int r600_vport_create(struct radeon_device *rdev,
+				struct r600_atoms *atoms,
+				struct drm_radeon_atom *patom,
+				struct radeon_atom **atom)
+{
+	struct drm_r600_vport pvport;
+	struct r600_vport *vport;
+	int r;
+
+	vport = kmalloc(sizeof(*vport), GFP_KERNEL);
+	if (vport == NULL)
+		return -ENOMEM;
+	/* make sure structure properly initialized */
+	memset(vport, 0, sizeof(*vport));
+	r = radeon_atom_init(&vport->atom, &atoms->idr, &r600_vport_release,
+			&radeon_atom_emit_default, &r600_atom_process_default,
+			vport->pkts);
+	if (r)
+		goto out_err;
+	/* KERNEL use get user data */
+	memcpy(&pvport, (void*)(unsigned long)patom->data, sizeof(struct drm_r600_vport));
+	/* PA_SC_VPORT_ZMIN_0 */
+	vport->pkts[vport->atom.npkts++] = PKT3(PKT3_SET_CONTEXT_REG, 2);
+	vport->pkts[vport->atom.npkts++] = 0x000000B4;
+	vport->pkts[vport->atom.npkts++] = pvport.pa_sc_vport_zmin_0;
+	vport->pkts[vport->atom.npkts++] = pvport.pa_sc_vport_zmax_0;
+	/* PA_CL_VPORT_XSCALE_0 */
+	vport->pkts[vport->atom.npkts++] = PKT3(PKT3_SET_CONTEXT_REG, 6);
+	vport->pkts[vport->atom.npkts++] = 0x0000010F;
+	vport->pkts[vport->atom.npkts++] = pvport.pa_cl_vport_xscale_0;
+	vport->pkts[vport->atom.npkts++] = pvport.pa_cl_vport_xoffset_0;
+	vport->pkts[vport->atom.npkts++] = pvport.pa_cl_vport_yscale_0;
+	vport->pkts[vport->atom.npkts++] = pvport.pa_cl_vport_yoffset_0;
+	vport->pkts[vport->atom.npkts++] = pvport.pa_cl_vport_zscale_0;
+	vport->pkts[vport->atom.npkts++] = pvport.pa_cl_vport_zoffset_0;
+	*atom = &vport->atom;
+fprintf(stderr, "%s %d pkts\n", __func__, (*atom)->npkts);
+	return 0;
+out_err:
+	radeon_atom_put(&vport->atom);
+	*atom = NULL;
+	return r;
+}
+
+
+/*
  * r600 atom core functions
  */
 int r600_atom_create(struct radeon_device *rdev,
@@ -630,6 +696,13 @@ int r600_atom_create(struct radeon_device *rdev,
 		atom->type = patom->type;
 		list_add_tail(&atom->list, &atoms->tp_atoms);
 		break;
+	case R600_ATOM_VPORT:
+		r = r600_vport_create(rdev, atoms, patom, &atom);
+		if (r)
+			return r;
+		atom->type = patom->type;
+		list_add_tail(&atom->list, &atoms->vport_atoms);
+		break;
 	default:
 		dev_err(rdev->dev, "unknown R600 atom type 0x%08X\n", patom->type);
 		return -EINVAL;
@@ -723,6 +796,10 @@ int r600_batches_queue(struct radeon_device *rdev,
 					R600_ATOM_TP, &rbatch->atoms[i++]);
 	if (r)
 		goto out_err;
+	r = radeon_atom_find_locked(&atoms->vport_atoms, batch->vport_id,
+					R600_ATOM_VPORT, &rbatch->atoms[i++]);
+	if (r)
+		goto out_err;
 	r = radeon_atom_find_locked(&atoms->cb_atoms, batch->cb_id,
 					R600_ATOM_CB, &rbatch->atoms[i++]);
 	if (r)
@@ -798,11 +875,12 @@ int r600_atoms_init(struct radeon_device *rdev, struct r600_atoms *atoms)
 {
 	mutex_init(&atoms->mutex);
 	INIT_LIST_HEAD(&atoms->cb_atoms);
+	INIT_LIST_HEAD(&atoms->cb_cntl_atoms);
 	INIT_LIST_HEAD(&atoms->pa_atoms);
 	INIT_LIST_HEAD(&atoms->sq_atoms);
 	INIT_LIST_HEAD(&atoms->tp_atoms);
 	INIT_LIST_HEAD(&atoms->vgt_atoms);
-	INIT_LIST_HEAD(&atoms->cb_cntl_atoms);
+	INIT_LIST_HEAD(&atoms->vport_atoms);
 	idr_init(&atoms->idr);
 	atoms->npipes = 2;
 	atoms->nbanks = 4;
@@ -1061,18 +1139,6 @@ void r600_tflat(struct radeon_atom *atom)
     WPKT(0x00000094);
     WPKT(0x80000000);
     WPKT(0x00FA00FA);
-    WPKT(PKT3(PKT3_SET_CONTEXT_REG, 2));
-    WPKT(0x000000B4);
-    WPKT(0x00000000);
-    WPKT(0x3F800000);
-    WPKT(PKT3(PKT3_SET_CONTEXT_REG, 6));
-    WPKT(0x0000010F);
-    WPKT(0x42FA0000);
-    WPKT(0x42FA0000);
-    WPKT(0xC2FA0000);
-    WPKT(0x42FA0000);
-    WPKT(0x3F000000);
-    WPKT(0x3F000000);
     WPKT(PKT3(PKT3_SURFACE_SYNC, 3));
     WPKT(0x08000000);
     WPKT(0x00000001);
diff --git a/r600_atom_api.h b/r600_atom_api.h
index 5a309f8..a471826 100644
--- a/r600_atom_api.h
+++ b/r600_atom_api.h
@@ -31,6 +31,7 @@ struct drm_radeon_atom {
 #define R600_ATOM_TP		2
 #define R600_ATOM_PA		3
 #define R600_ATOM_CB_CNTL	4
+#define R600_ATOM_VPORT		5
 
 struct drm_r600_cb {
 	u32			pitch;
@@ -101,11 +102,24 @@ struct drm_r600_pa {
 	u32			pa_su_poly_offset_back_offset;
 };
 
+/* vport - viewport */
+struct drm_r600_vport {
+	u32			pa_sc_vport_zmin_0;
+	u32			pa_sc_vport_zmax_0;
+	u32			pa_cl_vport_xscale_0;
+	u32			pa_cl_vport_xoffset_0;
+	u32			pa_cl_vport_yscale_0;
+	u32			pa_cl_vport_yoffset_0;
+	u32			pa_cl_vport_zscale_0;
+	u32			pa_cl_vport_zoffset_0;
+};
+
 struct drm_r600_batch {
 	u32			cb_id;
 	u32			cb_cntl_id;
 	u32			pa_id;
 	u32			tp_id;
+	u32			vport_id;
 };
 
 
diff --git a/radeon_atom.h b/radeon_atom.h
index 0acea79..2491113 100644
--- a/radeon_atom.h
+++ b/radeon_atom.h
@@ -53,7 +53,7 @@ struct radeon_atom {
 };
 
 /* R600 */
-#define R600_BATCH_NATOMS	4
+#define R600_BATCH_NATOMS	5
 struct r600_batch {
 	struct list_head	list;
 	struct list_head	pre_flushes;
@@ -71,11 +71,12 @@ struct r600_batches {
 
 struct r600_atoms {
 	struct list_head	cb_atoms;
+	struct list_head	cb_cntl_atoms;
 	struct list_head	pa_atoms;
 	struct list_head	sq_atoms;
 	struct list_head	tp_atoms;
 	struct list_head	vgt_atoms;
-	struct list_head	cb_cntl_atoms;
+	struct list_head	vport_atoms;
 	struct idr		idr;
 	struct mutex		mutex;
 	struct r600_batches	batches;
diff --git a/test.c b/test.c
index 9d5a5bb..91b5442 100644
--- a/test.c
+++ b/test.c
@@ -48,9 +48,10 @@ int r600_tri_flat(struct radeon *radeon)
 {
 	struct radeon_device *rdev;
 	struct drm_r600_cb cb;
-	struct drm_r600_tp tp;
-	struct drm_r600_pa pa;
 	struct drm_r600_cb_cntl cb_cntl;
+	struct drm_r600_pa pa;
+	struct drm_r600_tp tp;
+	struct drm_r600_vport vport;
 	struct drm_r600_batch batch;
 	struct drm_radeon_atom atom;
 	int r;
@@ -146,6 +147,22 @@ int r600_tri_flat(struct radeon *radeon)
 	if (r)
 		return r;
 	batch.cb_cntl_id = atom.id;
+	/* viewport */
+	vport.pa_sc_vport_zmin_0 = 0x00000000;
+	vport.pa_sc_vport_zmax_0 = 0x3f800000;
+	vport.pa_cl_vport_xscale_0 = 0x42fa0000;
+	vport.pa_cl_vport_xoffset_0 = 0x42fa0000;
+	vport.pa_cl_vport_yscale_0 = 0xc2fa0000;
+	vport.pa_cl_vport_yoffset_0 = 0x42fa0000;
+	vport.pa_cl_vport_zscale_0 = 0x3f000000;
+	vport.pa_cl_vport_zoffset_0 = 0x3f000000;
+	atom.type = R600_ATOM_VPORT;
+	atom.id = 0;
+	atom.data = (uint64_t)(uintptr_t)&vport;
+	r = radeon_atom_create(rdev, &atom);
+	if (r)
+		return r;
+	batch.vport_id = atom.id;
 
 	r =  radeon_batches_queue(rdev, &batch);
 	if (r)