These changes are the raw update to linux-4.4.6-rt14. Kernel sources
[kvmfornfv.git] / kernel / drivers / gpu / drm / nouveau / nvkm / engine / fifo / dmanv50.c
diff --git a/kernel/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv50.c b/kernel/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv50.c
new file mode 100644 (file)
index 0000000..6b3b15f
--- /dev/null
@@ -0,0 +1,91 @@
+/*
+ * Copyright 2012 Red Hat Inc.
+ *
+ * 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 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 COPYRIGHT HOLDER(S) OR AUTHOR(S) 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.
+ *
+ * Authors: Ben Skeggs
+ */
+#include "channv50.h"
+
+#include <core/client.h>
+#include <core/ramht.h>
+
+#include <nvif/class.h>
+#include <nvif/unpack.h>
+
+static int
+nv50_fifo_dma_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass,
+                 void *data, u32 size, struct nvkm_object **pobject)
+{
+       struct nvkm_object *parent = oclass->parent;
+       union {
+               struct nv50_channel_dma_v0 v0;
+       } *args = data;
+       struct nv50_fifo *fifo = nv50_fifo(base);
+       struct nv50_fifo_chan *chan;
+       int ret;
+
+       nvif_ioctl(parent, "create channel dma size %d\n", size);
+       if (nvif_unpack(args->v0, 0, 0, false)) {
+               nvif_ioctl(parent, "create channel dma vers %d vm %llx "
+                                  "pushbuf %llx offset %016llx\n",
+                          args->v0.version, args->v0.vm, args->v0.pushbuf,
+                          args->v0.offset);
+               if (!args->v0.pushbuf)
+                       return -EINVAL;
+       } else
+               return ret;
+
+       if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL)))
+               return -ENOMEM;
+       *pobject = &chan->base.object;
+
+       ret = nv50_fifo_chan_ctor(fifo, args->v0.vm, args->v0.pushbuf,
+                                 oclass, chan);
+       if (ret)
+               return ret;
+
+       args->v0.chid = chan->base.chid;
+
+       nvkm_kmap(chan->ramfc);
+       nvkm_wo32(chan->ramfc, 0x08, lower_32_bits(args->v0.offset));
+       nvkm_wo32(chan->ramfc, 0x0c, upper_32_bits(args->v0.offset));
+       nvkm_wo32(chan->ramfc, 0x10, lower_32_bits(args->v0.offset));
+       nvkm_wo32(chan->ramfc, 0x14, upper_32_bits(args->v0.offset));
+       nvkm_wo32(chan->ramfc, 0x3c, 0x003f6078);
+       nvkm_wo32(chan->ramfc, 0x44, 0x01003fff);
+       nvkm_wo32(chan->ramfc, 0x48, chan->base.push->node->offset >> 4);
+       nvkm_wo32(chan->ramfc, 0x4c, 0xffffffff);
+       nvkm_wo32(chan->ramfc, 0x60, 0x7fffffff);
+       nvkm_wo32(chan->ramfc, 0x78, 0x00000000);
+       nvkm_wo32(chan->ramfc, 0x7c, 0x30000001);
+       nvkm_wo32(chan->ramfc, 0x80, ((chan->ramht->bits - 9) << 27) |
+                                    (4 << 24) /* SEARCH_FULL */ |
+                                    (chan->ramht->gpuobj->node->offset >> 4));
+       nvkm_done(chan->ramfc);
+       return 0;
+}
+
+const struct nvkm_fifo_chan_oclass
+nv50_fifo_dma_oclass = {
+       .base.oclass = NV50_CHANNEL_DMA,
+       .base.minver = 0,
+       .base.maxver = 0,
+       .ctor = nv50_fifo_dma_new,
+};