Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / gpu / drm / nouveau / nvkm / engine / gr / nv20.c
diff --git a/kernel/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv20.c b/kernel/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv20.c
new file mode 100644 (file)
index 0000000..1713ffb
--- /dev/null
@@ -0,0 +1,376 @@
+#include "nv20.h"
+#include "regs.h"
+
+#include <core/client.h>
+#include <core/device.h>
+#include <core/handle.h>
+#include <engine/fifo.h>
+#include <subdev/fb.h>
+#include <subdev/timer.h>
+
+/*******************************************************************************
+ * Graphics object classes
+ ******************************************************************************/
+
+static struct nvkm_oclass
+nv20_gr_sclass[] = {
+       { 0x0012, &nv04_gr_ofuncs, NULL }, /* beta1 */
+       { 0x0019, &nv04_gr_ofuncs, NULL }, /* clip */
+       { 0x0030, &nv04_gr_ofuncs, NULL }, /* null */
+       { 0x0039, &nv04_gr_ofuncs, NULL }, /* m2mf */
+       { 0x0043, &nv04_gr_ofuncs, NULL }, /* rop */
+       { 0x0044, &nv04_gr_ofuncs, NULL }, /* patt */
+       { 0x004a, &nv04_gr_ofuncs, NULL }, /* gdi */
+       { 0x0062, &nv04_gr_ofuncs, NULL }, /* surf2d */
+       { 0x0072, &nv04_gr_ofuncs, NULL }, /* beta4 */
+       { 0x0089, &nv04_gr_ofuncs, NULL }, /* sifm */
+       { 0x008a, &nv04_gr_ofuncs, NULL }, /* ifc */
+       { 0x0096, &nv04_gr_ofuncs, NULL }, /* celcius */
+       { 0x0097, &nv04_gr_ofuncs, NULL }, /* kelvin */
+       { 0x009e, &nv04_gr_ofuncs, NULL }, /* swzsurf */
+       { 0x009f, &nv04_gr_ofuncs, NULL }, /* imageblit */
+       {},
+};
+
+/*******************************************************************************
+ * PGRAPH context
+ ******************************************************************************/
+
+static int
+nv20_gr_context_ctor(struct nvkm_object *parent, struct nvkm_object *engine,
+                    struct nvkm_oclass *oclass, void *data, u32 size,
+                    struct nvkm_object **pobject)
+{
+       struct nv20_gr_chan *chan;
+       int ret, i;
+
+       ret = nvkm_gr_context_create(parent, engine, oclass, NULL, 0x37f0,
+                                    16, NVOBJ_FLAG_ZERO_ALLOC, &chan);
+       *pobject = nv_object(chan);
+       if (ret)
+               return ret;
+
+       chan->chid = nvkm_fifo_chan(parent)->chid;
+
+       nv_wo32(chan, 0x0000, 0x00000001 | (chan->chid << 24));
+       nv_wo32(chan, 0x033c, 0xffff0000);
+       nv_wo32(chan, 0x03a0, 0x0fff0000);
+       nv_wo32(chan, 0x03a4, 0x0fff0000);
+       nv_wo32(chan, 0x047c, 0x00000101);
+       nv_wo32(chan, 0x0490, 0x00000111);
+       nv_wo32(chan, 0x04a8, 0x44400000);
+       for (i = 0x04d4; i <= 0x04e0; i += 4)
+               nv_wo32(chan, i, 0x00030303);
+       for (i = 0x04f4; i <= 0x0500; i += 4)
+               nv_wo32(chan, i, 0x00080000);
+       for (i = 0x050c; i <= 0x0518; i += 4)
+               nv_wo32(chan, i, 0x01012000);
+       for (i = 0x051c; i <= 0x0528; i += 4)
+               nv_wo32(chan, i, 0x000105b8);
+       for (i = 0x052c; i <= 0x0538; i += 4)
+               nv_wo32(chan, i, 0x00080008);
+       for (i = 0x055c; i <= 0x0598; i += 4)
+               nv_wo32(chan, i, 0x07ff0000);
+       nv_wo32(chan, 0x05a4, 0x4b7fffff);
+       nv_wo32(chan, 0x05fc, 0x00000001);
+       nv_wo32(chan, 0x0604, 0x00004000);
+       nv_wo32(chan, 0x0610, 0x00000001);
+       nv_wo32(chan, 0x0618, 0x00040000);
+       nv_wo32(chan, 0x061c, 0x00010000);
+       for (i = 0x1c1c; i <= 0x248c; i += 16) {
+               nv_wo32(chan, (i + 0), 0x10700ff9);
+               nv_wo32(chan, (i + 4), 0x0436086c);
+               nv_wo32(chan, (i + 8), 0x000c001b);
+       }
+       nv_wo32(chan, 0x281c, 0x3f800000);
+       nv_wo32(chan, 0x2830, 0x3f800000);
+       nv_wo32(chan, 0x285c, 0x40000000);
+       nv_wo32(chan, 0x2860, 0x3f800000);
+       nv_wo32(chan, 0x2864, 0x3f000000);
+       nv_wo32(chan, 0x286c, 0x40000000);
+       nv_wo32(chan, 0x2870, 0x3f800000);
+       nv_wo32(chan, 0x2878, 0xbf800000);
+       nv_wo32(chan, 0x2880, 0xbf800000);
+       nv_wo32(chan, 0x34a4, 0x000fe000);
+       nv_wo32(chan, 0x3530, 0x000003f8);
+       nv_wo32(chan, 0x3540, 0x002fe000);
+       for (i = 0x355c; i <= 0x3578; i += 4)
+               nv_wo32(chan, i, 0x001c527c);
+       return 0;
+}
+
+int
+nv20_gr_context_init(struct nvkm_object *object)
+{
+       struct nv20_gr_priv *priv = (void *)object->engine;
+       struct nv20_gr_chan *chan = (void *)object;
+       int ret;
+
+       ret = nvkm_gr_context_init(&chan->base);
+       if (ret)
+               return ret;
+
+       nv_wo32(priv->ctxtab, chan->chid * 4, nv_gpuobj(chan)->addr >> 4);
+       return 0;
+}
+
+int
+nv20_gr_context_fini(struct nvkm_object *object, bool suspend)
+{
+       struct nv20_gr_priv *priv = (void *)object->engine;
+       struct nv20_gr_chan *chan = (void *)object;
+       int chid = -1;
+
+       nv_mask(priv, 0x400720, 0x00000001, 0x00000000);
+       if (nv_rd32(priv, 0x400144) & 0x00010000)
+               chid = (nv_rd32(priv, 0x400148) & 0x1f000000) >> 24;
+       if (chan->chid == chid) {
+               nv_wr32(priv, 0x400784, nv_gpuobj(chan)->addr >> 4);
+               nv_wr32(priv, 0x400788, 0x00000002);
+               nv_wait(priv, 0x400700, 0xffffffff, 0x00000000);
+               nv_wr32(priv, 0x400144, 0x10000000);
+               nv_mask(priv, 0x400148, 0xff000000, 0x1f000000);
+       }
+       nv_mask(priv, 0x400720, 0x00000001, 0x00000001);
+
+       nv_wo32(priv->ctxtab, chan->chid * 4, 0x00000000);
+       return nvkm_gr_context_fini(&chan->base, suspend);
+}
+
+static struct nvkm_oclass
+nv20_gr_cclass = {
+       .handle = NV_ENGCTX(GR, 0x20),
+       .ofuncs = &(struct nvkm_ofuncs) {
+               .ctor = nv20_gr_context_ctor,
+               .dtor = _nvkm_gr_context_dtor,
+               .init = nv20_gr_context_init,
+               .fini = nv20_gr_context_fini,
+               .rd32 = _nvkm_gr_context_rd32,
+               .wr32 = _nvkm_gr_context_wr32,
+       },
+};
+
+/*******************************************************************************
+ * PGRAPH engine/subdev functions
+ ******************************************************************************/
+
+void
+nv20_gr_tile_prog(struct nvkm_engine *engine, int i)
+{
+       struct nvkm_fb_tile *tile = &nvkm_fb(engine)->tile.region[i];
+       struct nvkm_fifo *pfifo = nvkm_fifo(engine);
+       struct nv20_gr_priv *priv = (void *)engine;
+       unsigned long flags;
+
+       pfifo->pause(pfifo, &flags);
+       nv04_gr_idle(priv);
+
+       nv_wr32(priv, NV20_PGRAPH_TLIMIT(i), tile->limit);
+       nv_wr32(priv, NV20_PGRAPH_TSIZE(i), tile->pitch);
+       nv_wr32(priv, NV20_PGRAPH_TILE(i), tile->addr);
+
+       nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00EA0030 + 4 * i);
+       nv_wr32(priv, NV10_PGRAPH_RDI_DATA, tile->limit);
+       nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00EA0050 + 4 * i);
+       nv_wr32(priv, NV10_PGRAPH_RDI_DATA, tile->pitch);
+       nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00EA0010 + 4 * i);
+       nv_wr32(priv, NV10_PGRAPH_RDI_DATA, tile->addr);
+
+       if (nv_device(engine)->chipset != 0x34) {
+               nv_wr32(priv, NV20_PGRAPH_ZCOMP(i), tile->zcomp);
+               nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00ea0090 + 4 * i);
+               nv_wr32(priv, NV10_PGRAPH_RDI_DATA, tile->zcomp);
+       }
+
+       pfifo->start(pfifo, &flags);
+}
+
+void
+nv20_gr_intr(struct nvkm_subdev *subdev)
+{
+       struct nvkm_engine *engine = nv_engine(subdev);
+       struct nvkm_object *engctx;
+       struct nvkm_handle *handle;
+       struct nv20_gr_priv *priv = (void *)subdev;
+       u32 stat = nv_rd32(priv, NV03_PGRAPH_INTR);
+       u32 nsource = nv_rd32(priv, NV03_PGRAPH_NSOURCE);
+       u32 nstatus = nv_rd32(priv, NV03_PGRAPH_NSTATUS);
+       u32 addr = nv_rd32(priv, NV04_PGRAPH_TRAPPED_ADDR);
+       u32 chid = (addr & 0x01f00000) >> 20;
+       u32 subc = (addr & 0x00070000) >> 16;
+       u32 mthd = (addr & 0x00001ffc);
+       u32 data = nv_rd32(priv, NV04_PGRAPH_TRAPPED_DATA);
+       u32 class = nv_rd32(priv, 0x400160 + subc * 4) & 0xfff;
+       u32 show = stat;
+
+       engctx = nvkm_engctx_get(engine, chid);
+       if (stat & NV_PGRAPH_INTR_ERROR) {
+               if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) {
+                       handle = nvkm_handle_get_class(engctx, class);
+                       if (handle && !nv_call(handle->object, mthd, data))
+                               show &= ~NV_PGRAPH_INTR_ERROR;
+                       nvkm_handle_put(handle);
+               }
+       }
+
+       nv_wr32(priv, NV03_PGRAPH_INTR, stat);
+       nv_wr32(priv, NV04_PGRAPH_FIFO, 0x00000001);
+
+       if (show) {
+               nv_error(priv, "%s", "");
+               nvkm_bitfield_print(nv10_gr_intr_name, show);
+               pr_cont(" nsource:");
+               nvkm_bitfield_print(nv04_gr_nsource, nsource);
+               pr_cont(" nstatus:");
+               nvkm_bitfield_print(nv10_gr_nstatus, nstatus);
+               pr_cont("\n");
+               nv_error(priv,
+                        "ch %d [%s] subc %d class 0x%04x mthd 0x%04x data 0x%08x\n",
+                        chid, nvkm_client_name(engctx), subc, class, mthd,
+                        data);
+       }
+
+       nvkm_engctx_put(engctx);
+}
+
+static int
+nv20_gr_ctor(struct nvkm_object *parent, struct nvkm_object *engine,
+            struct nvkm_oclass *oclass, void *data, u32 size,
+            struct nvkm_object **pobject)
+{
+       struct nv20_gr_priv *priv;
+       int ret;
+
+       ret = nvkm_gr_create(parent, engine, oclass, true, &priv);
+       *pobject = nv_object(priv);
+       if (ret)
+               return ret;
+
+       ret = nvkm_gpuobj_new(nv_object(priv), NULL, 32 * 4, 16,
+                             NVOBJ_FLAG_ZERO_ALLOC, &priv->ctxtab);
+       if (ret)
+               return ret;
+
+       nv_subdev(priv)->unit = 0x00001000;
+       nv_subdev(priv)->intr = nv20_gr_intr;
+       nv_engine(priv)->cclass = &nv20_gr_cclass;
+       nv_engine(priv)->sclass = nv20_gr_sclass;
+       nv_engine(priv)->tile_prog = nv20_gr_tile_prog;
+       return 0;
+}
+
+void
+nv20_gr_dtor(struct nvkm_object *object)
+{
+       struct nv20_gr_priv *priv = (void *)object;
+       nvkm_gpuobj_ref(NULL, &priv->ctxtab);
+       nvkm_gr_destroy(&priv->base);
+}
+
+int
+nv20_gr_init(struct nvkm_object *object)
+{
+       struct nvkm_engine *engine = nv_engine(object);
+       struct nv20_gr_priv *priv = (void *)engine;
+       struct nvkm_fb *pfb = nvkm_fb(object);
+       u32 tmp, vramsz;
+       int ret, i;
+
+       ret = nvkm_gr_init(&priv->base);
+       if (ret)
+               return ret;
+
+       nv_wr32(priv, NV20_PGRAPH_CHANNEL_CTX_TABLE, priv->ctxtab->addr >> 4);
+
+       if (nv_device(priv)->chipset == 0x20) {
+               nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x003d0000);
+               for (i = 0; i < 15; i++)
+                       nv_wr32(priv, NV10_PGRAPH_RDI_DATA, 0x00000000);
+               nv_wait(priv, 0x400700, 0xffffffff, 0x00000000);
+       } else {
+               nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x02c80000);
+               for (i = 0; i < 32; i++)
+                       nv_wr32(priv, NV10_PGRAPH_RDI_DATA, 0x00000000);
+               nv_wait(priv, 0x400700, 0xffffffff, 0x00000000);
+       }
+
+       nv_wr32(priv, NV03_PGRAPH_INTR   , 0xFFFFFFFF);
+       nv_wr32(priv, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF);
+
+       nv_wr32(priv, NV04_PGRAPH_DEBUG_0, 0xFFFFFFFF);
+       nv_wr32(priv, NV04_PGRAPH_DEBUG_0, 0x00000000);
+       nv_wr32(priv, NV04_PGRAPH_DEBUG_1, 0x00118700);
+       nv_wr32(priv, NV04_PGRAPH_DEBUG_3, 0xF3CE0475); /* 0x4 = auto ctx switch */
+       nv_wr32(priv, NV10_PGRAPH_DEBUG_4, 0x00000000);
+       nv_wr32(priv, 0x40009C           , 0x00000040);
+
+       if (nv_device(priv)->chipset >= 0x25) {
+               nv_wr32(priv, 0x400890, 0x00a8cfff);
+               nv_wr32(priv, 0x400610, 0x304B1FB6);
+               nv_wr32(priv, 0x400B80, 0x1cbd3883);
+               nv_wr32(priv, 0x400B84, 0x44000000);
+               nv_wr32(priv, 0x400098, 0x40000080);
+               nv_wr32(priv, 0x400B88, 0x000000ff);
+
+       } else {
+               nv_wr32(priv, 0x400880, 0x0008c7df);
+               nv_wr32(priv, 0x400094, 0x00000005);
+               nv_wr32(priv, 0x400B80, 0x45eae20e);
+               nv_wr32(priv, 0x400B84, 0x24000000);
+               nv_wr32(priv, 0x400098, 0x00000040);
+               nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00E00038);
+               nv_wr32(priv, NV10_PGRAPH_RDI_DATA , 0x00000030);
+               nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00E10038);
+               nv_wr32(priv, NV10_PGRAPH_RDI_DATA , 0x00000030);
+       }
+
+       /* Turn all the tiling regions off. */
+       for (i = 0; i < pfb->tile.regions; i++)
+               engine->tile_prog(engine, i);
+
+       nv_wr32(priv, 0x4009a0, nv_rd32(priv, 0x100324));
+       nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00EA000C);
+       nv_wr32(priv, NV10_PGRAPH_RDI_DATA, nv_rd32(priv, 0x100324));
+
+       nv_wr32(priv, NV10_PGRAPH_CTX_CONTROL, 0x10000100);
+       nv_wr32(priv, NV10_PGRAPH_STATE      , 0xFFFFFFFF);
+
+       tmp = nv_rd32(priv, NV10_PGRAPH_SURFACE) & 0x0007ff00;
+       nv_wr32(priv, NV10_PGRAPH_SURFACE, tmp);
+       tmp = nv_rd32(priv, NV10_PGRAPH_SURFACE) | 0x00020100;
+       nv_wr32(priv, NV10_PGRAPH_SURFACE, tmp);
+
+       /* begin RAM config */
+       vramsz = nv_device_resource_len(nv_device(priv), 0) - 1;
+       nv_wr32(priv, 0x4009A4, nv_rd32(priv, 0x100200));
+       nv_wr32(priv, 0x4009A8, nv_rd32(priv, 0x100204));
+       nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00EA0000);
+       nv_wr32(priv, NV10_PGRAPH_RDI_DATA , nv_rd32(priv, 0x100200));
+       nv_wr32(priv, NV10_PGRAPH_RDI_INDEX, 0x00EA0004);
+       nv_wr32(priv, NV10_PGRAPH_RDI_DATA , nv_rd32(priv, 0x100204));
+       nv_wr32(priv, 0x400820, 0);
+       nv_wr32(priv, 0x400824, 0);
+       nv_wr32(priv, 0x400864, vramsz - 1);
+       nv_wr32(priv, 0x400868, vramsz - 1);
+
+       /* interesting.. the below overwrites some of the tile setup above.. */
+       nv_wr32(priv, 0x400B20, 0x00000000);
+       nv_wr32(priv, 0x400B04, 0xFFFFFFFF);
+
+       nv_wr32(priv, NV03_PGRAPH_ABS_UCLIP_XMIN, 0);
+       nv_wr32(priv, NV03_PGRAPH_ABS_UCLIP_YMIN, 0);
+       nv_wr32(priv, NV03_PGRAPH_ABS_UCLIP_XMAX, 0x7fff);
+       nv_wr32(priv, NV03_PGRAPH_ABS_UCLIP_YMAX, 0x7fff);
+       return 0;
+}
+
+struct nvkm_oclass
+nv20_gr_oclass = {
+       .handle = NV_ENGINE(GR, 0x20),
+       .ofuncs = &(struct nvkm_ofuncs) {
+               .ctor = nv20_gr_ctor,
+               .dtor = nv20_gr_dtor,
+               .init = nv20_gr_init,
+               .fini = _nvkm_gr_fini,
+       },
+};