20 files changed, 119 insertions, 96 deletions

diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c
index d1264a7154a6..cb198aca9f5a 100644
--- a/drivers/gpu/drm/radeon/evergreen.c
+++ b/drivers/gpu/drm/radeon/evergreen.c
@@ -1311,7 +1311,7 @@ void evergreen_mc_program(struct radeon_device *rdev)
  */
 void evergreen_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[ib->fence->ring];
 
         /* set to DX10/11 mode */
         radeon_ring_write(cp, PACKET3(PACKET3_MODE_CONTROL, 0));
@@ -1362,7 +1362,7 @@ static int evergreen_cp_load_microcode(struct radeon_device *rdev)
 
 static int evergreen_cp_start(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r, i;
         uint32_t cp_me;
 
@@ -1428,7 +1428,7 @@ static int evergreen_cp_start(struct radeon_device *rdev)
 
 int evergreen_cp_resume(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 tmp;
         u32 rb_bufsz;
         int r;
@@ -3056,7 +3056,7 @@ restart_ih:
 
 static int evergreen_startup(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
 
         /* enable pcie gen2 link */
@@ -3168,7 +3168,7 @@ int evergreen_resume(struct radeon_device *rdev)
 
 int evergreen_suspend(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
 
         /* FIXME: we should wait for ring to be empty */
         r700_cp_stop(rdev);
@@ -3251,8 +3251,8 @@ int evergreen_init(struct radeon_device *rdev)
         if (r)
                 return r;
 
-        rdev->cp.ring_obj = NULL;
-        r600_ring_init(rdev, &rdev->cp, 1024 * 1024);
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ring_obj = NULL;
+        r600_ring_init(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX], 1024 * 1024);
 
         rdev->ih.ring_obj = NULL;
         r600_ih_ring_init(rdev, 64 * 1024);
diff --git a/drivers/gpu/drm/radeon/evergreen_blit_kms.c b/drivers/gpu/drm/radeon/evergreen_blit_kms.c
index 75d0a6f0a395..56f5d92cce24 100644
--- a/drivers/gpu/drm/radeon/evergreen_blit_kms.c
+++ b/drivers/gpu/drm/radeon/evergreen_blit_kms.c
@@ -49,7 +49,7 @@ static void
 set_render_target(struct radeon_device *rdev, int format,
                   int w, int h, u64 gpu_addr)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 cb_color_info;
         int pitch, slice;
 
@@ -88,7 +88,7 @@ cp_set_surface_sync(struct radeon_device *rdev,
                     u32 sync_type, u32 size,
                     u64 mc_addr)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 cp_coher_size;
 
         if (size == 0xffffffff)
@@ -116,7 +116,7 @@ cp_set_surface_sync(struct radeon_device *rdev,
 static void
 set_shaders(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u64 gpu_addr;
 
         /* VS */
@@ -144,7 +144,7 @@ set_shaders(struct radeon_device *rdev)
 static void
 set_vtx_resource(struct radeon_device *rdev, u64 gpu_addr)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 sq_vtx_constant_word2, sq_vtx_constant_word3;
 
         /* high addr, stride */
@@ -189,7 +189,7 @@ set_tex_resource(struct radeon_device *rdev,
                  int format, int w, int h, int pitch,
                  u64 gpu_addr, u32 size)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 sq_tex_resource_word0, sq_tex_resource_word1;
         u32 sq_tex_resource_word4, sq_tex_resource_word7;
 
@@ -230,7 +230,7 @@ static void
 set_scissors(struct radeon_device *rdev, int x1, int y1,
              int x2, int y2)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         /* workaround some hw bugs */
         if (x2 == 0)
                 x1 = 1;
@@ -261,7 +261,7 @@ set_scissors(struct radeon_device *rdev, int x1, int y1,
 static void
 draw_auto(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         radeon_ring_write(cp, PACKET3(PACKET3_SET_CONFIG_REG, 1));
         radeon_ring_write(cp, (VGT_PRIMITIVE_TYPE - PACKET3_SET_CONFIG_REG_START) >> 2);
         radeon_ring_write(cp, DI_PT_RECTLIST);
@@ -286,7 +286,7 @@ draw_auto(struct radeon_device *rdev)
 static void
 set_default_state(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 sq_config, sq_gpr_resource_mgmt_1, sq_gpr_resource_mgmt_2, sq_gpr_resource_mgmt_3;
         u32 sq_thread_resource_mgmt, sq_thread_resource_mgmt_2;
         u32 sq_stack_resource_mgmt_1, sq_stack_resource_mgmt_2, sq_stack_resource_mgmt_3;
diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c
index cc9aaeb104f5..2d809e62c4c6 100644
--- a/drivers/gpu/drm/radeon/ni.c
+++ b/drivers/gpu/drm/radeon/ni.c
@@ -1049,7 +1049,7 @@ static int cayman_cp_load_microcode(struct radeon_device *rdev)
 
 static int cayman_cp_start(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r, i;
 
         r = radeon_ring_lock(rdev, cp, 7);
@@ -1116,7 +1116,7 @@ static int cayman_cp_start(struct radeon_device *rdev)
 static void cayman_cp_fini(struct radeon_device *rdev)
 {
         cayman_cp_enable(rdev, false);
-        radeon_ring_fini(rdev, &rdev->cp);
+        radeon_ring_fini(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX]);
 }
 
 int cayman_cp_resume(struct radeon_device *rdev)
@@ -1147,7 +1147,7 @@ int cayman_cp_resume(struct radeon_device *rdev)
 
         /* ring 0 - compute and gfx */
         /* Set ring buffer size */
-        cp = &rdev->cp;
+        cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         rb_bufsz = drm_order(cp->ring_size / 8);
         tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
 #ifdef __BIG_ENDIAN
@@ -1181,7 +1181,7 @@ int cayman_cp_resume(struct radeon_device *rdev)
 
         /* ring1  - compute only */
         /* Set ring buffer size */
-        cp = &rdev->cp1;
+        cp = &rdev->cp[CAYMAN_RING_TYPE_CP1_INDEX];
         rb_bufsz = drm_order(cp->ring_size / 8);
         tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
 #ifdef __BIG_ENDIAN
@@ -1207,7 +1207,7 @@ int cayman_cp_resume(struct radeon_device *rdev)
 
         /* ring2 - compute only */
         /* Set ring buffer size */
-        cp = &rdev->cp2;
+        cp = &rdev->cp[CAYMAN_RING_TYPE_CP2_INDEX];
         rb_bufsz = drm_order(cp->ring_size / 8);
         tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
 #ifdef __BIG_ENDIAN
@@ -1233,15 +1233,15 @@ int cayman_cp_resume(struct radeon_device *rdev)
 
         /* start the rings */
         cayman_cp_start(rdev);
-        rdev->cp.ready = true;
-        rdev->cp1.ready = true;
-        rdev->cp2.ready = true;
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready = true;
+        rdev->cp[CAYMAN_RING_TYPE_CP1_INDEX].ready = true;
+        rdev->cp[CAYMAN_RING_TYPE_CP2_INDEX].ready = true;
         /* this only test cp0 */
-        r = radeon_ring_test(rdev, &rdev->cp);
+        r = radeon_ring_test(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX]);
         if (r) {
-                rdev->cp.ready = false;
-                rdev->cp1.ready = false;
-                rdev->cp2.ready = false;
+                rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready = false;
+                rdev->cp[CAYMAN_RING_TYPE_CP1_INDEX].ready = false;
+                rdev->cp[CAYMAN_RING_TYPE_CP2_INDEX].ready = false;
                 return r;
         }
 
@@ -1343,7 +1343,7 @@ int cayman_asic_reset(struct radeon_device *rdev)
 
 static int cayman_startup(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
 
         /* enable pcie gen2 link */
@@ -1438,7 +1438,7 @@ int cayman_suspend(struct radeon_device *rdev)
 {
         /* FIXME: we should wait for ring to be empty */
         cayman_cp_enable(rdev, false);
-        rdev->cp.ready = false;
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready = false;
         evergreen_irq_suspend(rdev);
         radeon_wb_disable(rdev);
         cayman_pcie_gart_disable(rdev);
@@ -1455,7 +1455,7 @@ int cayman_suspend(struct radeon_device *rdev)
  */
 int cayman_init(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
 
         /* This don't do much */
diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c
index 6c328115e66c..6ca20d7bf626 100644
--- a/drivers/gpu/drm/radeon/r100.c
+++ b/drivers/gpu/drm/radeon/r100.c
@@ -811,7 +811,7 @@ u32 r100_get_vblank_counter(struct radeon_device *rdev, int crtc)
 void r100_fence_ring_emit(struct radeon_device *rdev,
                           struct radeon_fence *fence)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[fence->ring];
 
         /* We have to make sure that caches are flushed before
          * CPU might read something from VRAM. */
@@ -849,7 +849,7 @@ int r100_copy_blit(struct radeon_device *rdev,
                    unsigned num_gpu_pages,
                    struct radeon_fence *fence)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         uint32_t cur_pages;
         uint32_t stride_bytes = RADEON_GPU_PAGE_SIZE;
         uint32_t pitch;
@@ -934,7 +934,7 @@ static int r100_cp_wait_for_idle(struct radeon_device *rdev)
 
 void r100_ring_start(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
 
         r = radeon_ring_lock(rdev, cp, 2);
@@ -1048,7 +1048,7 @@ static void r100_cp_load_microcode(struct radeon_device *rdev)
 
 int r100_cp_init(struct radeon_device *rdev, unsigned ring_size)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         unsigned rb_bufsz;
         unsigned rb_blksz;
         unsigned max_fetch;
@@ -1162,7 +1162,7 @@ void r100_cp_fini(struct radeon_device *rdev)
         }
         /* Disable ring */
         r100_cp_disable(rdev);
-        radeon_ring_fini(rdev, &rdev->cp);
+        radeon_ring_fini(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX]);
         DRM_INFO("radeon: cp finalized\n");
 }
 
@@ -1170,7 +1170,7 @@ void r100_cp_disable(struct radeon_device *rdev)
 {
         /* Disable ring */
         radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size);
-        rdev->cp.ready = false;
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready = false;
         WREG32(RADEON_CP_CSQ_MODE, 0);
         WREG32(RADEON_CP_CSQ_CNTL, 0);
         WREG32(R_000770_SCRATCH_UMSK, 0);
@@ -2587,7 +2587,7 @@ static int r100_debugfs_cp_ring_info(struct seq_file *m, void *data)
         struct drm_info_node *node = (struct drm_info_node *) m->private;
         struct drm_device *dev = node->minor->dev;
         struct radeon_device *rdev = dev->dev_private;
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         uint32_t rdp, wdp;
         unsigned count, i, j;
 
@@ -3686,7 +3686,7 @@ int r100_ring_test(struct radeon_device *rdev, struct radeon_cp *cp)
 
 void r100_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
 
         radeon_ring_write(cp, PACKET0(RADEON_CP_IB_BASE, 1));
         radeon_ring_write(cp, ib->gpu_addr);
@@ -3778,7 +3778,7 @@ void r100_mc_stop(struct radeon_device *rdev, struct r100_mc_save *save)
         /* Shutdown CP we shouldn't need to do that but better be safe than
          * sorry
          */
-        rdev->cp.ready = false;
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready = false;
         WREG32(R_000740_CP_CSQ_CNTL, 0);
 
         /* Save few CRTC registers */
diff --git a/drivers/gpu/drm/radeon/r200.c b/drivers/gpu/drm/radeon/r200.c
index d84e633f72fc..d59c727a8e0f 100644
--- a/drivers/gpu/drm/radeon/r200.c
+++ b/drivers/gpu/drm/radeon/r200.c
@@ -87,7 +87,7 @@ int r200_copy_dma(struct radeon_device *rdev,
                   unsigned num_gpu_pages,
                   struct radeon_fence *fence)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         uint32_t size;
         uint32_t cur_size;
         int i, num_loops;
diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c
index cbb62fc3f2e9..66ff35f394ce 100644
--- a/drivers/gpu/drm/radeon/r300.c
+++ b/drivers/gpu/drm/radeon/r300.c
@@ -175,7 +175,7 @@ void rv370_pcie_gart_fini(struct radeon_device *rdev)
 void r300_fence_ring_emit(struct radeon_device *rdev,
                           struct radeon_fence *fence)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[fence->ring];
 
         /* Who ever call radeon_fence_emit should call ring_lock and ask
          * for enough space (today caller are ib schedule and buffer move) */
@@ -208,7 +208,7 @@ void r300_fence_ring_emit(struct radeon_device *rdev,
 
 void r300_ring_start(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         unsigned gb_tile_config;
         int r;
 
diff --git a/drivers/gpu/drm/radeon/r420.c b/drivers/gpu/drm/radeon/r420.c
index 4c0af4955f08..62e860436999 100644
--- a/drivers/gpu/drm/radeon/r420.c
+++ b/drivers/gpu/drm/radeon/r420.c
@@ -199,7 +199,7 @@ static void r420_clock_resume(struct radeon_device *rdev)
 
 static void r420_cp_errata_init(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
 
         /* RV410 and R420 can lock up if CP DMA to host memory happens
          * while the 2D engine is busy.
@@ -217,7 +217,7 @@ static void r420_cp_errata_init(struct radeon_device *rdev)
 
 static void r420_cp_errata_fini(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
 
         /* Catch the RESYNC we dispatched all the way back,
          * at the very beginning of the CP init.
diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c
index 599753176741..aaf8cd42943e 100644
--- a/drivers/gpu/drm/radeon/r600.c
+++ b/drivers/gpu/drm/radeon/r600.c
@@ -2144,7 +2144,7 @@ static int r600_cp_load_microcode(struct radeon_device *rdev)
 
 int r600_cp_start(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
         uint32_t cp_me;
 
@@ -2174,7 +2174,7 @@ int r600_cp_start(struct radeon_device *rdev)
 
 int r600_cp_resume(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 tmp;
         u32 rb_bufsz;
         int r;
@@ -2248,7 +2248,7 @@ void r600_ring_init(struct radeon_device *rdev, struct radeon_cp *cp, unsigned r
 void r600_cp_fini(struct radeon_device *rdev)
 {
         r600_cp_stop(rdev);
-        radeon_ring_fini(rdev, &rdev->cp);
+        radeon_ring_fini(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX]);
 }
 
 
@@ -2271,7 +2271,7 @@ int r600_ring_test(struct radeon_device *rdev, struct radeon_cp *cp)
 {
         uint32_t scratch;
         uint32_t tmp = 0;
-        unsigned i;
+        unsigned i, ridx = radeon_ring_index(rdev, cp);
         int r;
 
         r = radeon_scratch_get(rdev, &scratch);
@@ -2282,7 +2282,7 @@ int r600_ring_test(struct radeon_device *rdev, struct radeon_cp *cp)
         WREG32(scratch, 0xCAFEDEAD);
         r = radeon_ring_lock(rdev, cp, 3);
         if (r) {
-                DRM_ERROR("radeon: cp failed to lock ring %p (%d).\n", cp, r);
+                DRM_ERROR("radeon: cp failed to lock ring %d (%d).\n", ridx, r);
                 radeon_scratch_free(rdev, scratch);
                 return r;
         }
@@ -2297,10 +2297,10 @@ int r600_ring_test(struct radeon_device *rdev, struct radeon_cp *cp)
                 DRM_UDELAY(1);
         }
         if (i < rdev->usec_timeout) {
-                DRM_INFO("ring test on %p succeeded in %d usecs\n", cp, i);
+                DRM_INFO("ring test on %d succeeded in %d usecs\n", ridx, i);
         } else {
-                DRM_ERROR("radeon: ring %p test failed (scratch(0x%04X)=0x%08X)\n",
-                          cp, scratch, tmp);
+                DRM_ERROR("radeon: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
+                          ridx, scratch, tmp);
                 r = -EINVAL;
         }
         radeon_scratch_free(rdev, scratch);
@@ -2310,7 +2310,7 @@ int r600_ring_test(struct radeon_device *rdev, struct radeon_cp *cp)
 void r600_fence_ring_emit(struct radeon_device *rdev,
                           struct radeon_fence *fence)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[fence->ring];
 
         if (rdev->wb.use_event) {
                 u64 addr = rdev->wb.gpu_addr + R600_WB_EVENT_OFFSET +
@@ -2420,7 +2420,7 @@ void r600_clear_surface_reg(struct radeon_device *rdev, int reg)
 
 int r600_startup(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
 
         /* enable pcie gen2 link */
@@ -2534,7 +2534,7 @@ int r600_suspend(struct radeon_device *rdev)
         r600_audio_fini(rdev);
         /* FIXME: we should wait for ring to be empty */
         r600_cp_stop(rdev);
-        rdev->cp.ready = false;
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready = false;
         r600_irq_suspend(rdev);
         radeon_wb_disable(rdev);
         r600_pcie_gart_disable(rdev);
@@ -2609,8 +2609,8 @@ int r600_init(struct radeon_device *rdev)
         if (r)
                 return r;
 
-        rdev->cp.ring_obj = NULL;
-        r600_ring_init(rdev, &rdev->cp, 1024 * 1024);
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ring_obj = NULL;
+        r600_ring_init(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX], 1024 * 1024);
 
         rdev->ih.ring_obj = NULL;
         r600_ih_ring_init(rdev, 64 * 1024);
@@ -2677,7 +2677,7 @@ void r600_fini(struct radeon_device *rdev)
  */
 void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[ib->fence->ring];
 
         /* FIXME: implement */
         radeon_ring_write(cp, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
@@ -3518,7 +3518,7 @@ static int r600_debugfs_cp_ring_info(struct seq_file *m, void *data)
         struct drm_info_node *node = (struct drm_info_node *) m->private;
         struct drm_device *dev = node->minor->dev;
         struct radeon_device *rdev = dev->dev_private;
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         unsigned count, i, j;
 
         radeon_ring_free_size(rdev, cp);
diff --git a/drivers/gpu/drm/radeon/r600_blit_kms.c b/drivers/gpu/drm/radeon/r600_blit_kms.c
index 39ae19d38c2f..62dd1c281c76 100644
--- a/drivers/gpu/drm/radeon/r600_blit_kms.c
+++ b/drivers/gpu/drm/radeon/r600_blit_kms.c
@@ -50,7 +50,7 @@ static void
 set_render_target(struct radeon_device *rdev, int format,
                   int w, int h, u64 gpu_addr)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 cb_color_info;
         int pitch, slice;
 
@@ -104,7 +104,7 @@ cp_set_surface_sync(struct radeon_device *rdev,
                     u32 sync_type, u32 size,
                     u64 mc_addr)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 cp_coher_size;
 
         if (size == 0xffffffff)
@@ -123,7 +123,7 @@ cp_set_surface_sync(struct radeon_device *rdev,
 static void
 set_shaders(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u64 gpu_addr;
         u32 sq_pgm_resources;
 
@@ -170,7 +170,7 @@ set_shaders(struct radeon_device *rdev)
 static void
 set_vtx_resource(struct radeon_device *rdev, u64 gpu_addr)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 sq_vtx_constant_word2;
 
         sq_vtx_constant_word2 = SQ_VTXC_BASE_ADDR_HI(upper_32_bits(gpu_addr) & 0xff) |
@@ -207,7 +207,7 @@ set_tex_resource(struct radeon_device *rdev,
                  int format, int w, int h, int pitch,
                  u64 gpu_addr, u32 size)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         uint32_t sq_tex_resource_word0, sq_tex_resource_word1, sq_tex_resource_word4;
 
         if (h < 1)
@@ -246,7 +246,7 @@ static void
 set_scissors(struct radeon_device *rdev, int x1, int y1,
              int x2, int y2)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         radeon_ring_write(cp, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
         radeon_ring_write(cp, (PA_SC_SCREEN_SCISSOR_TL - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2);
         radeon_ring_write(cp, (x1 << 0) | (y1 << 16));
@@ -267,7 +267,7 @@ set_scissors(struct radeon_device *rdev, int x1, int y1,
 static void
 draw_auto(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         radeon_ring_write(cp, PACKET3(PACKET3_SET_CONFIG_REG, 1));
         radeon_ring_write(cp, (VGT_PRIMITIVE_TYPE - PACKET3_SET_CONFIG_REG_OFFSET) >> 2);
         radeon_ring_write(cp, DI_PT_RECTLIST);
@@ -292,7 +292,7 @@ draw_auto(struct radeon_device *rdev)
 static void
 set_default_state(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         u32 sq_config, sq_gpr_resource_mgmt_1, sq_gpr_resource_mgmt_2;
         u32 sq_thread_resource_mgmt, sq_stack_resource_mgmt_1, sq_stack_resource_mgmt_2;
         int num_ps_gprs, num_vs_gprs, num_temp_gprs, num_gs_gprs, num_es_gprs;
@@ -687,7 +687,7 @@ static unsigned r600_blit_create_rect(unsigned num_gpu_pages,
 
 int r600_blit_prepare_copy(struct radeon_device *rdev, unsigned num_gpu_pages)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
         int ring_size;
         int num_loops = 0;
@@ -727,7 +727,7 @@ void r600_blit_done_copy(struct radeon_device *rdev, struct radeon_fence *fence)
         if (fence)
                 r = radeon_fence_emit(rdev, fence);
 
-        radeon_ring_unlock_commit(rdev, &rdev->cp);
+        radeon_ring_unlock_commit(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX]);
 }
 
 void r600_kms_blit_copy(struct radeon_device *rdev,
diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index bbe88ec3951d..76c58e9e4778 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -547,6 +547,7 @@ struct r600_ih {
         struct radeon_bo        *ring_obj;
         volatile uint32_t       *ring;
         unsigned                rptr;
+        unsigned                rptr_offs;
         unsigned                wptr;
         unsigned                wptr_old;
         unsigned                ring_size;
@@ -598,6 +599,7 @@ void radeon_ib_pool_fini(struct radeon_device *rdev);
 int radeon_ib_test(struct radeon_device *rdev);
 extern void radeon_ib_bogus_add(struct radeon_device *rdev, struct radeon_ib *ib);
 /* Ring access between begin & end cannot sleep */
+int radeon_ring_index(struct radeon_device *rdev, struct radeon_cp *cp);
 void radeon_ring_free_size(struct radeon_device *rdev, struct radeon_cp *cp);
 int radeon_ring_alloc(struct radeon_device *rdev, struct radeon_cp *cp, unsigned ndw);
 int radeon_ring_lock(struct radeon_device *rdev, struct radeon_cp *cp, unsigned ndw);
@@ -1284,9 +1286,7 @@ struct radeon_device {
         rwlock_t                        fence_lock;
         struct radeon_fence_driver      fence_drv[RADEON_NUM_RINGS];
         struct radeon_semaphore_driver  semaphore_drv;
-        struct radeon_cp                cp;
-        struct radeon_cp                cp1;
-        struct radeon_cp                cp2;
+        struct radeon_cp                cp[RADEON_NUM_RINGS];
         struct radeon_ib_pool           ib_pool;
         struct radeon_irq               irq;
         struct radeon_asic              *asic;
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index 36296ad397a1..023c156eddd0 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -718,7 +718,8 @@ int radeon_device_init(struct radeon_device *rdev,
          * can recall function without having locking issues */
         radeon_mutex_init(&rdev->cs_mutex);
         mutex_init(&rdev->ib_pool.mutex);
-        mutex_init(&rdev->cp.mutex);
+        for (i = 0; i < RADEON_NUM_RINGS; ++i)
+                mutex_init(&rdev->cp[i].mutex);
         mutex_init(&rdev->dc_hw_i2c_mutex);
         if (rdev->family >= CHIP_R600)
                 spin_lock_init(&rdev->ih.lock);
diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index 9ed0bb100bcb..86f4eeaeba66 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -84,7 +84,7 @@ int radeon_fence_emit(struct radeon_device *rdev, struct radeon_fence *fence)
                 return 0;
         }
         fence->seq = atomic_add_return(1, &rdev->fence_drv[fence->ring].seq);
-        if (!rdev->cp.ready)
+        if (!rdev->cp[fence->ring].ready)
                 /* FIXME: cp is not running assume everythings is done right
                  * away
                  */
@@ -269,7 +269,7 @@ retry:
                  * if we experiencing a lockup the value doesn't change
                  */
                 if (seq == rdev->fence_drv[fence->ring].last_seq &&
-                    radeon_gpu_is_lockup(rdev, &rdev->cp)) {
+                    radeon_gpu_is_lockup(rdev, &rdev->cp[fence->ring])) {
                         /* good news we believe it's a lockup */
                         printk(KERN_WARNING "GPU lockup (waiting for 0x%08X last fence id 0x%08X)\n",
                              fence->seq, seq);
diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c
index 136772ccfe72..1ce8fa71cf73 100644
--- a/drivers/gpu/drm/radeon/radeon_gem.c
+++ b/drivers/gpu/drm/radeon/radeon_gem.c
@@ -152,6 +152,7 @@ int radeon_gem_info_ioctl(struct drm_device *dev, void *data,
         struct radeon_device *rdev = dev->dev_private;
         struct drm_radeon_gem_info *args = data;
         struct ttm_mem_type_manager *man;
+        unsigned i;
 
         man = &rdev->mman.bdev.man[TTM_PL_VRAM];
 
@@ -161,7 +162,8 @@ int radeon_gem_info_ioctl(struct drm_device *dev, void *data,
                 args->vram_visible -= radeon_bo_size(rdev->stollen_vga_memory);
         args->vram_visible -= radeon_fbdev_total_size(rdev);
         args->gart_size = rdev->mc.gtt_size - 4096 - RADEON_IB_POOL_SIZE*64*1024;
-        args->gart_size -= rdev->cp.ring_size;
+        for(i = 0; i < RADEON_NUM_RINGS; ++i)
+                args->gart_size -= rdev->cp[i].ring_size;
         return 0;
 }
 
diff --git a/drivers/gpu/drm/radeon/radeon_pm.c b/drivers/gpu/drm/radeon/radeon_pm.c
index 73b6714d615b..50b632ac8237 100644
--- a/drivers/gpu/drm/radeon/radeon_pm.c
+++ b/drivers/gpu/drm/radeon/radeon_pm.c
@@ -252,8 +252,10 @@ static void radeon_pm_set_clocks(struct radeon_device *rdev)
 
         mutex_lock(&rdev->ddev->struct_mutex);
         mutex_lock(&rdev->vram_mutex);
-        if (rdev->cp.ring_obj)
-                mutex_lock(&rdev->cp.mutex);
+        for (i = 0; i < RADEON_NUM_RINGS; ++i) {
+                if (rdev->cp[i].ring_obj)
+                        mutex_lock(&rdev->cp[i].mutex);
+        }
 
         /* gui idle int has issues on older chips it seems */
         if (rdev->family >= CHIP_R600) {
@@ -269,11 +271,11 @@ static void radeon_pm_set_clocks(struct radeon_device *rdev)
                         radeon_irq_set(rdev);
                 }
         } else {
-                struct radeon_cp *cp = &rdev->cp;
+                struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
                 if (cp->ready) {
                         struct radeon_fence *fence;
                         radeon_ring_alloc(rdev, cp, 64);
-                        radeon_fence_create(rdev, &fence, RADEON_RING_TYPE_GFX_INDEX);
+                        radeon_fence_create(rdev, &fence, radeon_ring_index(rdev, cp));
                         radeon_fence_emit(rdev, fence);
                         radeon_ring_commit(rdev, cp);
                         radeon_fence_wait(fence, false);
@@ -309,8 +311,10 @@ static void radeon_pm_set_clocks(struct radeon_device *rdev)
 
         rdev->pm.dynpm_planned_action = DYNPM_ACTION_NONE;
 
-        if (rdev->cp.ring_obj)
-                mutex_unlock(&rdev->cp.mutex);
+        for (i = 0; i < RADEON_NUM_RINGS; ++i) {
+                if (rdev->cp[i].ring_obj)
+                        mutex_unlock(&rdev->cp[i].mutex);
+        }
         mutex_unlock(&rdev->vram_mutex);
         mutex_unlock(&rdev->ddev->struct_mutex);
 }
diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c
index 209834dbf18f..d96b13af5e9d 100644
--- a/drivers/gpu/drm/radeon/radeon_ring.c
+++ b/drivers/gpu/drm/radeon/radeon_ring.c
@@ -178,7 +178,7 @@ void radeon_ib_free(struct radeon_device *rdev, struct radeon_ib **ib)
 
 int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[ib->fence->ring];
         int r = 0;
 
         if (!ib->length_dw || !cp->ready) {
@@ -284,6 +284,21 @@ void radeon_ib_pool_fini(struct radeon_device *rdev)
 /*
  * Ring.
  */
+int radeon_ring_index(struct radeon_device *rdev, struct radeon_cp *cp)
+{
+        /* r1xx-r5xx only has CP ring */
+        if (rdev->family < CHIP_R600)
+                return RADEON_RING_TYPE_GFX_INDEX;
+
+        if (rdev->family >= CHIP_CAYMAN) {
+                if (cp == &rdev->cp[CAYMAN_RING_TYPE_CP1_INDEX])
+                        return CAYMAN_RING_TYPE_CP1_INDEX;
+                else if (cp == &rdev->cp[CAYMAN_RING_TYPE_CP2_INDEX])
+                        return CAYMAN_RING_TYPE_CP2_INDEX;
+        }
+        return RADEON_RING_TYPE_GFX_INDEX;
+}
+
 void radeon_ring_free_size(struct radeon_device *rdev, struct radeon_cp *cp)
 {
         if (rdev->wb.enabled)
@@ -312,7 +327,7 @@ int radeon_ring_alloc(struct radeon_device *rdev, struct radeon_cp *cp, unsigned
                 if (ndw < cp->ring_free_dw) {
                         break;
                 }
-                r = radeon_fence_wait_next(rdev, RADEON_RING_TYPE_GFX_INDEX);
+                r = radeon_fence_wait_next(rdev, radeon_ring_index(rdev, cp));
                 if (r)
                         return r;
         }
diff --git a/drivers/gpu/drm/radeon/radeon_semaphore.c b/drivers/gpu/drm/radeon/radeon_semaphore.c
index 064694a67824..bf4789eed0b2 100644
--- a/drivers/gpu/drm/radeon/radeon_semaphore.c
+++ b/drivers/gpu/drm/radeon/radeon_semaphore.c
@@ -121,13 +121,13 @@ int radeon_semaphore_create(struct radeon_device *rdev,
 void radeon_semaphore_emit_signal(struct radeon_device *rdev, int ring,
                                   struct radeon_semaphore *semaphore)
 {
-        radeon_semaphore_ring_emit(rdev, &rdev->cp, semaphore, false);
+        radeon_semaphore_ring_emit(rdev, &rdev->cp[ring], semaphore, false);
 }
 
 void radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring,
                                 struct radeon_semaphore *semaphore)
 {
-        radeon_semaphore_ring_emit(rdev, &rdev->cp, semaphore, true);
+        radeon_semaphore_ring_emit(rdev, &rdev->cp[ring], semaphore, true);
 }
 
 void radeon_semaphore_free(struct radeon_device *rdev,
diff --git a/drivers/gpu/drm/radeon/radeon_test.c b/drivers/gpu/drm/radeon/radeon_test.c
index ee6c160ffae9..160e7df77551 100644
--- a/drivers/gpu/drm/radeon/radeon_test.c
+++ b/drivers/gpu/drm/radeon/radeon_test.c
@@ -43,7 +43,8 @@ void radeon_test_moves(struct radeon_device *rdev)
          * (Total GTT - IB pool - writeback page - ring buffers) / test size
          */
         n = rdev->mc.gtt_size - RADEON_IB_POOL_SIZE*64*1024;
-        n -= rdev->cp.ring_size;
+        for (i = 0; i < RADEON_NUM_RINGS; ++i)
+                n -= rdev->cp[i].ring_size;
         if (rdev->wb.wb_obj)
                 n -= RADEON_GPU_PAGE_SIZE;
         if (rdev->ih.ring_obj)
diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c
index 112ecaa6362b..0be15bf38d3c 100644
--- a/drivers/gpu/drm/radeon/radeon_ttm.c
+++ b/drivers/gpu/drm/radeon/radeon_ttm.c
@@ -188,7 +188,7 @@ static void radeon_evict_flags(struct ttm_buffer_object *bo,
         rbo = container_of(bo, struct radeon_bo, tbo);
         switch (bo->mem.mem_type) {
         case TTM_PL_VRAM:
-                if (rbo->rdev->cp.ready == false)
+                if (rbo->rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready == false)
                         radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_CPU);
                 else
                         radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_GTT);
@@ -255,7 +255,7 @@ static int radeon_move_blit(struct ttm_buffer_object *bo,
                 DRM_ERROR("Unknown placement %d\n", old_mem->mem_type);
                 return -EINVAL;
         }
-        if (!rdev->cp.ready) {
+        if (!rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready) {
                 DRM_ERROR("Trying to move memory with CP turned off.\n");
                 return -EINVAL;
         }
@@ -380,7 +380,7 @@ static int radeon_bo_move(struct ttm_buffer_object *bo,
                 radeon_move_null(bo, new_mem);
                 return 0;
         }
-        if (!rdev->cp.ready || rdev->asic->copy == NULL) {
+        if (!rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready || rdev->asic->copy == NULL) {
                 /* use memcpy */
                 goto memcpy;
         }
diff --git a/drivers/gpu/drm/radeon/rv515.c b/drivers/gpu/drm/radeon/rv515.c
index 8fe13ba8143a..8a935987d022 100644
--- a/drivers/gpu/drm/radeon/rv515.c
+++ b/drivers/gpu/drm/radeon/rv515.c
@@ -55,7 +55,7 @@ void rv515_debugfs(struct radeon_device *rdev)
 
 void rv515_ring_start(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
 
         r = radeon_ring_lock(rdev, cp, 64);
diff --git a/drivers/gpu/drm/radeon/rv770.c b/drivers/gpu/drm/radeon/rv770.c
index a2c60598d0f7..9e4c0418f54d 100644
--- a/drivers/gpu/drm/radeon/rv770.c
+++ b/drivers/gpu/drm/radeon/rv770.c
@@ -357,7 +357,7 @@ static int rv770_cp_load_microcode(struct radeon_device *rdev)
 void r700_cp_fini(struct radeon_device *rdev)
 {
         r700_cp_stop(rdev);
-        radeon_ring_fini(rdev, &rdev->cp);
+        radeon_ring_fini(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX]);
 }
 
 /*
@@ -1043,7 +1043,7 @@ int rv770_mc_init(struct radeon_device *rdev)
 
 static int rv770_startup(struct radeon_device *rdev)
 {
-        struct radeon_cp *cp = &rdev->cp;
+        struct radeon_cp *cp = &rdev->cp[RADEON_RING_TYPE_GFX_INDEX];
         int r;
 
         /* enable pcie gen2 link */
@@ -1144,7 +1144,7 @@ int rv770_suspend(struct radeon_device *rdev)
         r600_audio_fini(rdev);
         /* FIXME: we should wait for ring to be empty */
         r700_cp_stop(rdev);
-        rdev->cp.ready = false;
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ready = false;
         r600_irq_suspend(rdev);
         radeon_wb_disable(rdev);
         rv770_pcie_gart_disable(rdev);
@@ -1217,8 +1217,8 @@ int rv770_init(struct radeon_device *rdev)
         if (r)
                 return r;
 
-        rdev->cp.ring_obj = NULL;
-        r600_ring_init(rdev, &rdev->cp, 1024 * 1024);
+        rdev->cp[RADEON_RING_TYPE_GFX_INDEX].ring_obj = NULL;
+        r600_ring_init(rdev, &rdev->cp[RADEON_RING_TYPE_GFX_INDEX], 1024 * 1024);
 
         rdev->ih.ring_obj = NULL;
         r600_ih_ring_init(rdev, 64 * 1024);