drm/nva3: implement support for copy engine
authorBen Skeggs <bskeggs@redhat.com>
Fri, 18 Mar 2011 00:25:59 +0000 (10:25 +1000)
committerBen Skeggs <bskeggs@redhat.com>
Mon, 16 May 2011 00:48:48 +0000 (10:48 +1000)
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
drivers/gpu/drm/nouveau/Makefile
drivers/gpu/drm/nouveau/nouveau_drv.h
drivers/gpu/drm/nouveau/nouveau_state.c
drivers/gpu/drm/nouveau/nv50_graph.c
drivers/gpu/drm/nouveau/nva3_copy.c [new file with mode: 0644]
drivers/gpu/drm/nouveau/nva3_copy.fuc [new file with mode: 0644]
drivers/gpu/drm/nouveau/nva3_copy.fuc.h [new file with mode: 0644]

index e12c97fd8db89790299836773533827a8d7af3ee..fc5702e8ce855c63c0c97cc5bcf9456ed03a60b8 100644 (file)
@@ -20,6 +20,7 @@ nouveau-y := nouveau_drv.o nouveau_state.o nouveau_channel.o nouveau_mem.o \
              nv40_graph.o nv50_graph.o nvc0_graph.o \
              nv40_grctx.o nv50_grctx.o nvc0_grctx.o \
              nv84_crypt.o \
+             nva3_copy.o \
              nv04_instmem.o nv50_instmem.o nvc0_instmem.o \
              nv50_evo.o nv50_crtc.o nv50_dac.o nv50_sor.o \
              nv50_cursor.o nv50_display.o \
index 3f90f3891f5964e351babbe38ec383dcd6b52ecd..78db2850d9ea6aafa361f211eac09d59088e71a8 100644 (file)
@@ -151,6 +151,8 @@ enum nouveau_flags {
 #define NVOBJ_ENGINE_SW                0
 #define NVOBJ_ENGINE_GR                1
 #define NVOBJ_ENGINE_CRYPT     2
+#define NVOBJ_ENGINE_COPY0     3
+#define NVOBJ_ENGINE_COPY1     4
 #define NVOBJ_ENGINE_DISPLAY   15
 #define NVOBJ_ENGINE_NR                16
 
@@ -1137,6 +1139,7 @@ extern void nv40_grctx_init(struct nouveau_grctx *);
 extern int  nv50_graph_create(struct drm_device *);
 extern int  nv50_grctx_init(struct nouveau_grctx *);
 extern struct nouveau_enum nv50_data_error_names[];
+extern int  nv50_graph_isr_chid(struct drm_device *dev, u64 inst);
 
 /* nvc0_graph.c */
 extern int  nvc0_graph_create(struct drm_device *);
@@ -1144,6 +1147,12 @@ extern int  nvc0_graph_create(struct drm_device *);
 /* nv84_crypt.c */
 extern int  nv84_crypt_create(struct drm_device *);
 
+/* nva3_copy.c */
+extern int  nva3_copy_create(struct drm_device *dev);
+
+/* nvc0_copy.c */
+extern int  nvc0_copy_create(struct drm_device *dev, int engine);
+
 /* nv04_instmem.c */
 extern int  nv04_instmem_init(struct drm_device *);
 extern void nv04_instmem_takedown(struct drm_device *);
index e50156cdcabd05cd5432b7be34cfda494a8148e5..fa1e08510eb58e274444efe21bed36ed1742a593 100644 (file)
@@ -598,6 +598,21 @@ nouveau_card_init(struct drm_device *dev)
                break;
        }
 
+       switch (dev_priv->card_type) {
+       case NV_50:
+               switch (dev_priv->chipset) {
+               case 0xa3:
+               case 0xa5:
+               case 0xa8:
+               case 0xaf:
+                       nva3_copy_create(dev);
+                       break;
+               }
+               break;
+       default:
+               break;
+       }
+
        if (!nouveau_noaccel) {
                for (e = 0; e < NVOBJ_ENGINE_NR; e++) {
                        if (dev_priv->eng[e]) {
index 5794cdef32c56f289d67f960224c692a1c6cca67..e25cbb46789a6780b5cf9611680468a835a09cff 100644 (file)
@@ -939,7 +939,7 @@ nv50_pgraph_trap_handler(struct drm_device *dev, u32 display, u64 inst, u32 chid
        return 1;
 }
 
-static int
+int
 nv50_graph_isr_chid(struct drm_device *dev, u64 inst)
 {
        struct drm_nouveau_private *dev_priv = dev->dev_private;
diff --git a/drivers/gpu/drm/nouveau/nva3_copy.c b/drivers/gpu/drm/nouveau/nva3_copy.c
new file mode 100644 (file)
index 0000000..b86820a
--- /dev/null
@@ -0,0 +1,226 @@
+/*
+ * Copyright 2011 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 <linux/firmware.h>
+#include "drmP.h"
+#include "nouveau_drv.h"
+#include "nouveau_util.h"
+#include "nouveau_vm.h"
+#include "nouveau_ramht.h"
+#include "nva3_copy.fuc.h"
+
+struct nva3_copy_engine {
+       struct nouveau_exec_engine base;
+};
+
+static int
+nva3_copy_context_new(struct nouveau_channel *chan, int engine)
+{
+       struct drm_device *dev = chan->dev;
+       struct drm_nouveau_private *dev_priv = dev->dev_private;
+       struct nouveau_gpuobj *ramin = chan->ramin;
+       struct nouveau_gpuobj *ctx = NULL;
+       int ret;
+
+       NV_DEBUG(dev, "ch%d\n", chan->id);
+
+       ret = nouveau_gpuobj_new(dev, chan, 256, 0, NVOBJ_FLAG_ZERO_ALLOC |
+                                NVOBJ_FLAG_ZERO_FREE, &ctx);
+       if (ret)
+               return ret;
+
+       nv_wo32(ramin, 0xc0, 0x00190000);
+       nv_wo32(ramin, 0xc4, ctx->vinst + ctx->size - 1);
+       nv_wo32(ramin, 0xc8, ctx->vinst);
+       nv_wo32(ramin, 0xcc, 0x00000000);
+       nv_wo32(ramin, 0xd0, 0x00000000);
+       nv_wo32(ramin, 0xd4, 0x00000000);
+       dev_priv->engine.instmem.flush(dev);
+
+       atomic_inc(&chan->vm->engref[engine]);
+       chan->engctx[engine] = ctx;
+       return 0;
+}
+
+static int
+nva3_copy_object_new(struct nouveau_channel *chan, int engine,
+                    u32 handle, u16 class)
+{
+       struct nouveau_gpuobj *ctx = chan->engctx[engine];
+
+       /* fuc engine doesn't need an object, our ramht code does.. */
+       ctx->engine = 3;
+       ctx->class  = class;
+       return nouveau_ramht_insert(chan, handle, ctx);
+}
+
+static void
+nva3_copy_context_del(struct nouveau_channel *chan, int engine)
+{
+       struct nouveau_gpuobj *ctx = chan->engctx[engine];
+       struct drm_device *dev = chan->dev;
+       u32 inst;
+
+       inst  = (chan->ramin->vinst >> 12);
+       inst |= 0x40000000;
+
+       /* disable fifo access */
+       nv_wr32(dev, 0x104048, 0x00000000);
+       /* mark channel as unloaded if it's currently active */
+       if (nv_rd32(dev, 0x104050) == inst)
+               nv_mask(dev, 0x104050, 0x40000000, 0x00000000);
+       /* mark next channel as invalid if it's about to be loaded */
+       if (nv_rd32(dev, 0x104054) == inst)
+               nv_mask(dev, 0x104054, 0x40000000, 0x00000000);
+       /* restore fifo access */
+       nv_wr32(dev, 0x104048, 0x00000003);
+
+       for (inst = 0xc0; inst <= 0xd4; inst += 4)
+               nv_wo32(chan->ramin, inst, 0x00000000);
+
+       nouveau_gpuobj_ref(NULL, &ctx);
+
+       atomic_dec(&chan->vm->engref[engine]);
+       chan->engctx[engine] = ctx;
+}
+
+static void
+nva3_copy_tlb_flush(struct drm_device *dev, int engine)
+{
+       nv50_vm_flush_engine(dev, 0x0d);
+}
+
+static int
+nva3_copy_init(struct drm_device *dev, int engine)
+{
+       int i;
+
+       nv_mask(dev, 0x000200, 0x00002000, 0x00000000);
+       nv_mask(dev, 0x000200, 0x00002000, 0x00002000);
+       nv_wr32(dev, 0x104014, 0xffffffff); /* disable all interrupts */
+
+       /* upload ucode */
+       nv_wr32(dev, 0x1041c0, 0x01000000);
+       for (i = 0; i < sizeof(nva3_pcopy_data) / 4; i++)
+               nv_wr32(dev, 0x1041c4, nva3_pcopy_data[i]);
+
+       nv_wr32(dev, 0x104180, 0x01000000);
+       for (i = 0; i < sizeof(nva3_pcopy_code) / 4; i++) {
+               if ((i & 0x3f) == 0)
+                       nv_wr32(dev, 0x104188, i >> 6);
+               nv_wr32(dev, 0x104184, nva3_pcopy_code[i]);
+       }
+
+       /* start it running */
+       nv_wr32(dev, 0x10410c, 0x00000000);
+       nv_wr32(dev, 0x104104, 0x00000000); /* ENTRY */
+       nv_wr32(dev, 0x104100, 0x00000002); /* TRIGGER */
+       return 0;
+}
+
+static int
+nva3_copy_fini(struct drm_device *dev, int engine)
+{
+       nv_mask(dev, 0x104048, 0x00000003, 0x00000000);
+
+       /* trigger fuc context unload */
+       nv_wait(dev, 0x104008, 0x0000000c, 0x00000000);
+       nv_mask(dev, 0x104054, 0x40000000, 0x00000000);
+       nv_wr32(dev, 0x104000, 0x00000008);
+       nv_wait(dev, 0x104008, 0x00000008, 0x00000000);
+
+       nv_wr32(dev, 0x104014, 0xffffffff);
+       return 0;
+}
+
+static struct nouveau_enum nva3_copy_isr_error_name[] = {
+       { 0x0001, "ILLEGAL_MTHD" },
+       { 0x0002, "INVALID_ENUM" },
+       { 0x0003, "INVALID_BITFIELD" },
+       {}
+};
+
+static void
+nva3_copy_isr(struct drm_device *dev)
+{
+       u32 dispatch = nv_rd32(dev, 0x10401c);
+       u32 stat = nv_rd32(dev, 0x104008) & dispatch & ~(dispatch >> 16);
+       u32 inst = nv_rd32(dev, 0x104050) & 0x3fffffff;
+       u32 ssta = nv_rd32(dev, 0x104040) & 0x0000ffff;
+       u32 addr = nv_rd32(dev, 0x104040) >> 16;
+       u32 mthd = (addr & 0x07ff) << 2;
+       u32 subc = (addr & 0x3800) >> 11;
+       u32 data = nv_rd32(dev, 0x104044);
+       int chid = nv50_graph_isr_chid(dev, inst);
+
+       if (stat & 0x00000040) {
+               NV_INFO(dev, "PCOPY: DISPATCH_ERROR [");
+               nouveau_enum_print(nva3_copy_isr_error_name, ssta);
+               printk("] ch %d [0x%08x] subc %d mthd 0x%04x data 0x%08x\n",
+                       chid, inst, subc, mthd, data);
+               nv_wr32(dev, 0x104004, 0x00000040);
+               stat &= ~0x00000040;
+       }
+
+       if (stat) {
+               NV_INFO(dev, "PCOPY: unhandled intr 0x%08x\n", stat);
+               nv_wr32(dev, 0x104004, stat);
+       }
+       nv50_fb_vm_trap(dev, 1);
+}
+
+static void
+nva3_copy_destroy(struct drm_device *dev, int engine)
+{
+       struct nva3_copy_engine *pcopy = nv_engine(dev, engine);
+
+       nouveau_irq_unregister(dev, 22);
+
+       NVOBJ_ENGINE_DEL(dev, COPY0);
+       kfree(pcopy);
+}
+
+int
+nva3_copy_create(struct drm_device *dev)
+{
+       struct nva3_copy_engine *pcopy;
+
+       pcopy = kzalloc(sizeof(*pcopy), GFP_KERNEL);
+       if (!pcopy)
+               return -ENOMEM;
+
+       pcopy->base.destroy = nva3_copy_destroy;
+       pcopy->base.init = nva3_copy_init;
+       pcopy->base.fini = nva3_copy_fini;
+       pcopy->base.context_new = nva3_copy_context_new;
+       pcopy->base.context_del = nva3_copy_context_del;
+       pcopy->base.object_new = nva3_copy_object_new;
+       pcopy->base.tlb_flush = nva3_copy_tlb_flush;
+
+       nouveau_irq_register(dev, 22, nva3_copy_isr);
+
+       NVOBJ_ENGINE_ADD(dev, COPY0, &pcopy->base);
+       NVOBJ_CLASS(dev, 0x85b5, COPY0);
+       return 0;
+}
diff --git a/drivers/gpu/drm/nouveau/nva3_copy.fuc b/drivers/gpu/drm/nouveau/nva3_copy.fuc
new file mode 100644 (file)
index 0000000..eaf35f8
--- /dev/null
@@ -0,0 +1,870 @@
+/* fuc microcode for copy engine on nva3- chipsets
+ *
+ * Copyright 2011 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
+ */
+
+/* To build for nva3:nvc0
+ *    m4 -DNVA3 nva3_copy.fuc | envyas -a -w -m fuc -V nva3 -o nva3_copy.fuc.h
+ *
+ * To build for nvc0-
+ *    m4 -DNVC0 nva3_copy.fuc | envyas -a -w -m fuc -V nva3 -o nvc0_copy.fuc.h
+ */
+
+ifdef(`NVA3',
+.section nva3_pcopy_data,
+.section nvc0_pcopy_data
+)
+
+ctx_object:                   .b32 0
+ifdef(`NVA3',
+ctx_dma:
+ctx_dma_query:                .b32 0
+ctx_dma_src:                  .b32 0
+ctx_dma_dst:                  .b32 0
+,)
+.equ ctx_dma_count 3
+ctx_query_address_high:       .b32 0
+ctx_query_address_low:        .b32 0
+ctx_query_counter:            .b32 0
+ctx_src_address_high:         .b32 0
+ctx_src_address_low:          .b32 0
+ctx_src_pitch:                .b32 0
+ctx_src_tile_mode:            .b32 0
+ctx_src_xsize:                .b32 0
+ctx_src_ysize:                .b32 0
+ctx_src_zsize:                .b32 0
+ctx_src_zoff:                 .b32 0
+ctx_src_xoff:                 .b32 0
+ctx_src_yoff:                 .b32 0
+ctx_src_cpp:                  .b32 0
+ctx_dst_address_high:         .b32 0
+ctx_dst_address_low:          .b32 0
+ctx_dst_pitch:                .b32 0
+ctx_dst_tile_mode:            .b32 0
+ctx_dst_xsize:                .b32 0
+ctx_dst_ysize:                .b32 0
+ctx_dst_zsize:                .b32 0
+ctx_dst_zoff:                 .b32 0
+ctx_dst_xoff:                 .b32 0
+ctx_dst_yoff:                 .b32 0
+ctx_dst_cpp:                  .b32 0
+ctx_format:                   .b32 0
+ctx_swz_const0:               .b32 0
+ctx_swz_const1:               .b32 0
+ctx_xcnt:                     .b32 0
+ctx_ycnt:                     .b32 0
+.align 256
+
+dispatch_table:
+// mthd 0x0000, NAME
+.b16 0x000 1
+.b32 ctx_object                     ~0xffffffff
+// mthd 0x0100, NOP
+.b16 0x040 1
+.b32 0x00010000 + cmd_nop           ~0xffffffff
+// mthd 0x0140, PM_TRIGGER
+.b16 0x050 1
+.b32 0x00010000 + cmd_pm_trigger    ~0xffffffff
+ifdef(`NVA3', `
+// mthd 0x0180-0x018c, DMA_
+.b16 0x060 ctx_dma_count
+dispatch_dma:
+.b32 0x00010000 + cmd_dma           ~0xffffffff
+.b32 0x00010000 + cmd_dma           ~0xffffffff
+.b32 0x00010000 + cmd_dma           ~0xffffffff
+',)
+// mthd 0x0200-0x0218, SRC_TILE
+.b16 0x80 7
+.b32 ctx_src_tile_mode              ~0x00000fff
+.b32 ctx_src_xsize                  ~0x0007ffff
+.b32 ctx_src_ysize                  ~0x00001fff
+.b32 ctx_src_zsize                  ~0x000007ff
+.b32 ctx_src_zoff                   ~0x00000fff
+.b32 ctx_src_xoff                   ~0x0007ffff
+.b32 ctx_src_yoff                   ~0x00001fff
+// mthd 0x0220-0x0238, DST_TILE
+.b16 0x88 7
+.b32 ctx_dst_tile_mode              ~0x00000fff
+.b32 ctx_dst_xsize                  ~0x0007ffff
+.b32 ctx_dst_ysize                  ~0x00001fff
+.b32 ctx_dst_zsize                  ~0x000007ff
+.b32 ctx_dst_zoff                   ~0x00000fff
+.b32 ctx_dst_xoff                   ~0x0007ffff
+.b32 ctx_dst_yoff                   ~0x00001fff
+// mthd 0x0300-0x0304, EXEC, WRCACHE_FLUSH
+.b16 0xc0 2
+.b32 0x00010000 + cmd_exec          ~0xffffffff
+.b32 0x00010000 + cmd_wrcache_flush ~0xffffffff
+// mthd 0x030c-0x0340, various stuff
+.b16 0xc3 14
+.b32 ctx_src_address_high           ~0x000000ff
+.b32 ctx_src_address_low            ~0xfffffff0
+.b32 ctx_dst_address_high           ~0x000000ff
+.b32 ctx_dst_address_low            ~0xfffffff0
+.b32 ctx_src_pitch                  ~0x0007ffff
+.b32 ctx_dst_pitch                  ~0x0007ffff
+.b32 ctx_xcnt                       ~0x0000ffff
+.b32 ctx_ycnt                       ~0x00001fff
+.b32 ctx_format                     ~0x0333ffff
+.b32 ctx_swz_const0                 ~0xffffffff
+.b32 ctx_swz_const1                 ~0xffffffff
+.b32 ctx_query_address_high         ~0x000000ff
+.b32 ctx_query_address_low          ~0xffffffff
+.b32 ctx_query_counter              ~0xffffffff
+.b16 0x800 0
+
+ifdef(`NVA3',
+.section nva3_pcopy_code,
+.section nvc0_pcopy_code
+)
+
+main:
+   clear b32 $r0
+   mov $sp $r0
+
+   // setup i0 handler and route fifo and ctxswitch to it
+   mov $r1 ih
+   mov $iv0 $r1
+   mov $r1 0x400
+   movw $r2 0xfff3
+   sethi $r2 0
+   iowr I[$r2 + 0x300] $r2
+
+   // enable interrupts
+   or $r2 0xc
+   iowr I[$r1] $r2
+   bset $flags ie0
+
+   // enable fifo access and context switching
+   mov $r1 0x1200
+   mov $r2 3
+   iowr I[$r1] $r2
+
+   // sleep forever, waking for interrupts
+   bset $flags $p0
+   spin:
+      sleep $p0
+      bra spin
+
+// i0 handler
+ih:
+   iord $r1 I[$r0 + 0x200]
+
+   and $r2 $r1 0x00000008
+   bra e ih_no_chsw
+      call chsw
+   ih_no_chsw:
+   and $r2 $r1 0x00000004
+   bra e ih_no_cmd
+      call dispatch
+
+   ih_no_cmd:
+   and $r1 $r1 0x0000000c
+   iowr I[$r0 + 0x100] $r1
+   iret
+
+// $p1 direction (0 = unload, 1 = load)
+// $r3 channel
+swctx:
+   mov $r4 0x7700
+   mov $xtargets $r4
+ifdef(`NVA3', `
+   // target 7 hardcoded to ctx dma object
+   mov $xdbase $r0
+', ` // NVC0
+   // read SCRATCH3 to decide if we are PCOPY0 or PCOPY1
+   mov $r4 0x2100
+   iord $r4 I[$r4 + 0]
+   and $r4 1
+   shl b32 $r4 4
+   add b32 $r4 0x30
+
+   // channel is in vram
+   mov $r15 0x61c
+   shl b32 $r15 6
+   mov $r5 0x114
+   iowrs I[$r15] $r5
+
+   // read 16-byte PCOPYn info, containing context pointer, from channel
+   shl b32 $r5 $r3 4
+   add b32 $r5 2
+   mov $xdbase $r5
+   mov $r5 $sp
+   // get a chunk of stack space, aligned to 256 byte boundary
+   sub b32 $r5 0x100
+   mov $r6 0xff
+   not b32 $r6
+   and $r5 $r6
+   sethi $r5 0x00020000
+   xdld $r4 $r5
+   xdwait
+   sethi $r5 0
+
+   // set context pointer, from within channel VM
+   mov $r14 0
+   iowrs I[$r15] $r14
+   ld b32 $r4 D[$r5 + 0]
+   shr b32 $r4 8
+   ld b32 $r6 D[$r5 + 4]
+   shl b32 $r6 24
+   or $r4 $r6
+   mov $xdbase $r4
+')
+   // 256-byte context, at start of data segment
+   mov b32 $r4 $r0
+   sethi $r4 0x60000
+
+   // swap!
+   bra $p1 swctx_load
+      xdst $r0 $r4
+      bra swctx_done
+   swctx_load:
+      xdld $r0 $r4
+   swctx_done:
+   xdwait
+   ret
+
+chsw:
+   // read current channel
+   mov $r2 0x1400
+   iord $r3 I[$r2]
+
+   // if it's active, unload it and return
+   xbit $r15 $r3 0x1e
+   bra e chsw_no_unload
+      bclr $flags $p1
+      call swctx
+      bclr $r3 0x1e
+      iowr I[$r2] $r3
+      mov $r4 1
+      iowr I[$r2 + 0x200] $r4
+      ret
+
+   // read next channel
+   chsw_no_unload:
+   iord $r3 I[$r2 + 0x100]
+
+   // is there a channel waiting to be loaded?
+   xbit $r13 $r3 0x1e
+   bra e chsw_finish_load
+      bset $flags $p1
+      call swctx
+ifdef(`NVA3',
+      // load dma objects back into TARGET regs
+      mov $r5 ctx_dma
+      mov $r6 ctx_dma_count
+      chsw_load_ctx_dma:
+         ld b32 $r7 D[$r5 + $r6 * 4]
+         add b32 $r8 $r6 0x180
+         shl b32 $r8 8
+         iowr I[$r8] $r7
+         sub b32 $r6 1
+         bra nc chsw_load_ctx_dma
+,)
+
+   chsw_finish_load:
+   mov $r3 2
+   iowr I[$r2 + 0x200] $r3
+   ret
+
+dispatch:
+   // read incoming fifo command
+   mov $r3 0x1900
+   iord $r2 I[$r3 + 0x100]
+   iord $r3 I[$r3 + 0x000]
+   and $r4 $r2 0x7ff
+   // $r2 will be used to store exception data
+   shl b32 $r2 0x10
+
+   // lookup method in the dispatch table, ILLEGAL_MTHD if not found
+   mov $r5 dispatch_table
+   clear b32 $r6
+   clear b32 $r7
+   dispatch_loop:
+      ld b16 $r6 D[$r5 + 0]
+      ld b16 $r7 D[$r5 + 2]
+      add b32 $r5 4
+      cmpu b32 $r4 $r6
+      bra c dispatch_illegal_mthd
+      add b32 $r7 $r6
+      cmpu b32 $r4 $r7
+      bra c dispatch_valid_mthd
+      sub b32 $r7 $r6
+      shl b32 $r7 3
+      add b32 $r5 $r7
+      bra dispatch_loop
+
+   // ensure no bits set in reserved fields, INVALID_BITFIELD
+   dispatch_valid_mthd:
+   sub b32 $r4 $r6
+   shl b32 $r4 3
+   add b32 $r4 $r5
+   ld b32 $r5 D[$r4 + 4]
+   and $r5 $r3
+   cmpu b32 $r5 0
+   bra ne dispatch_invalid_bitfield
+
+   // depending on dispatch flags: execute method, or save data as state
+   ld b16 $r5 D[$r4 + 0]
+   ld b16 $r6 D[$r4 + 2]
+   cmpu b32 $r6 0
+   bra ne dispatch_cmd
+      st b32 D[$r5] $r3
+      bra dispatch_done
+   dispatch_cmd:
+      bclr $flags $p1
+      call $r5
+      bra $p1 dispatch_error
+      bra dispatch_done
+
+   dispatch_invalid_bitfield:
+   or $r2 2
+   dispatch_illegal_mthd:
+   or $r2 1
+
+   // store exception data in SCRATCH0/SCRATCH1, signal hostirq
+   dispatch_error:
+   mov $r4 0x1000
+   iowr I[$r4 + 0x000] $r2
+   iowr I[$r4 + 0x100] $r3
+   mov $r2 0x40
+   iowr I[$r0] $r2
+   hostirq_wait:
+      iord $r2 I[$r0 + 0x200]
+      and $r2 0x40
+      cmpu b32 $r2 0
+      bra ne hostirq_wait
+
+   dispatch_done:
+   mov $r2 0x1d00
+   mov $r3 1
+   iowr I[$r2] $r3
+   ret
+
+// No-operation
+//
+// Inputs:
+//    $r1: irqh state
+//    $r2: hostirq state
+//    $r3: data
+//    $r4: dispatch table entry
+// Outputs:
+//    $r1: irqh state
+//    $p1: set on error
+//       $r2: hostirq state
+//       $r3: data
+cmd_nop:
+   ret
+
+// PM_TRIGGER
+//
+// Inputs:
+//    $r1: irqh state
+//    $r2: hostirq state
+//    $r3: data
+//    $r4: dispatch table entry
+// Outputs:
+//    $r1: irqh state
+//    $p1: set on error
+//       $r2: hostirq state
+//       $r3: data
+cmd_pm_trigger:
+   mov $r2 0x2200
+   clear b32 $r3
+   sethi $r3 0x20000
+   iowr I[$r2] $r3
+   ret
+
+ifdef(`NVA3',
+// SET_DMA_* method handler
+//
+// Inputs:
+//    $r1: irqh state
+//    $r2: hostirq state
+//    $r3: data
+//    $r4: dispatch table entry
+// Outputs:
+//    $r1: irqh state
+//    $p1: set on error
+//       $r2: hostirq state
+//       $r3: data
+cmd_dma:
+   sub b32 $r4 dispatch_dma
+   shr b32 $r4 1
+   bset $r3 0x1e
+   st b32 D[$r4 + ctx_dma] $r3
+   add b32 $r4 0x600
+   shl b32 $r4 6
+   iowr I[$r4] $r3
+   ret
+,)
+
+// Calculates the hw swizzle mask and adjusts the surface's xcnt to match
+//
+cmd_exec_set_format:
+   // zero out a chunk of the stack to store the swizzle into
+   add $sp -0x10
+   st b32 D[$sp + 0x00] $r0
+   st b32 D[$sp + 0x04] $r0
+   st b32 D[$sp + 0x08] $r0
+   st b32 D[$sp + 0x0c] $r0
+
+   // extract cpp, src_ncomp and dst_ncomp from FORMAT
+   ld b32 $r4 D[$r0 + ctx_format]
+   extr $r5 $r4 16:17
+   add b32 $r5 1
+   extr $r6 $r4 20:21
+   add b32 $r6 1
+   extr $r7 $r4 24:25
+   add b32 $r7 1
+
+   // convert FORMAT swizzle mask to hw swizzle mask
+   bclr $flags $p2
+   clear b32 $r8
+   clear b32 $r9
+   ncomp_loop:
+      and $r10 $r4 0xf
+      shr b32 $r4 4
+      clear b32 $r11
+      bpc_loop:
+         cmpu b8 $r10 4
+         bra nc cmp_c0
+            mulu $r12 $r10 $r5
+            add b32 $r12 $r11
+            bset $flags $p2
+            bra bpc_next
+         cmp_c0:
+         bra ne cmp_c1
+            mov $r12 0x10
+            add b32 $r12 $r11
+            bra bpc_next
+         cmp_c1:
+         cmpu b8 $r10 6
+         bra nc cmp_zero
+            mov $r12 0x14
+            add b32 $r12 $r11
+            bra bpc_next
+         cmp_zero:
+            mov $r12 0x80
+         bpc_next:
+         st b8 D[$sp + $r8] $r12
+         add b32 $r8 1
+         add b32 $r11 1
+         cmpu b32 $r11 $r5
+         bra c bpc_loop
+      add b32 $r9 1
+      cmpu b32 $r9 $r7
+      bra c ncomp_loop
+
+   // SRC_XCNT = (xcnt * src_cpp), or 0 if no src ref in swz (hw will hang)
+   mulu $r6 $r5
+   st b32 D[$r0 + ctx_src_cpp] $r6
+   ld b32 $r8 D[$r0 + ctx_xcnt]
+   mulu $r6 $r8
+   bra $p2 dst_xcnt
+   clear b32 $r6
+
+   dst_xcnt:
+   mulu $r7 $r5
+   st b32 D[$r0 + ctx_dst_cpp] $r7
+   mulu $r7 $r8
+
+   mov $r5 0x810
+   shl b32 $r5 6
+   iowr I[$r5 + 0x000] $r6
+   iowr I[$r5 + 0x100] $r7
+   add b32 $r5 0x800
+   ld b32 $r6 D[$r0 + ctx_dst_cpp]
+   sub b32 $r6 1
+   shl b32 $r6 8
+   ld b32 $r7 D[$r0 + ctx_src_cpp]
+   sub b32 $r7 1
+   or $r6 $r7
+   iowr I[$r5 + 0x000] $r6
+   add b32 $r5 0x100
+   ld b32 $r6 D[$sp + 0x00]
+   iowr I[$r5 + 0x000] $r6
+   ld b32 $r6 D[$sp + 0x04]
+   iowr I[$r5 + 0x100] $r6
+   ld b32 $r6 D[$sp + 0x08]
+   iowr I[$r5 + 0x200] $r6
+   ld b32 $r6 D[$sp + 0x0c]
+   iowr I[$r5 + 0x300] $r6
+   add b32 $r5 0x400
+   ld b32 $r6 D[$r0 + ctx_swz_const0]
+   iowr I[$r5 + 0x000] $r6
+   ld b32 $r6 D[$r0 + ctx_swz_const1]
+   iowr I[$r5 + 0x100] $r6
+   add $sp 0x10
+   ret
+
+// Setup to handle a tiled surface
+//
+// Calculates a number of parameters the hardware requires in order
+// to correctly handle tiling.
+//
+// Offset calculation is performed as follows (Tp/Th/Td from TILE_MODE):
+//    nTx = round_up(w * cpp, 1 << Tp) >> Tp
+//    nTy = round_up(h, 1 << Th) >> Th
+//    Txo = (x * cpp) & ((1 << Tp) - 1)
+//     Tx = (x * cpp) >> Tp
+//    Tyo = y & ((1 << Th) - 1)
+//     Ty = y >> Th
+//    Tzo = z & ((1 << Td) - 1)
+//     Tz = z >> Td
+//
+//    off  = (Tzo << Tp << Th) + (Tyo << Tp) + Txo
+//    off += ((Tz * nTy * nTx)) + (Ty * nTx) + Tx) << Td << Th << Tp;
+//
+// Inputs:
+//    $r4: hw command (0x104800)
+//    $r5: ctx offset adjustment for src/dst selection
+//    $p2: set if dst surface
+//
+cmd_exec_set_surface_tiled:
+   // translate TILE_MODE into Tp, Th, Td shift values
+   ld b32 $r7 D[$r5 + ctx_src_tile_mode]
+   extr $r9 $r7 8:11
+   extr $r8 $r7 4:7
+ifdef(`NVA3',
+   add b32 $r8 2
+,
+   add b32 $r8 3
+)
+   extr $r7 $r7 0:3
+   cmp b32 $r7 0xe
+   bra ne xtile64
+   mov $r7 4
+   bra xtileok
+   xtile64:
+   xbit $r7 $flags $p2
+   add b32 $r7 17
+   bset $r4 $r7
+   mov $r7 6
+   xtileok:
+
+   // Op = (x * cpp) & ((1 << Tp) - 1)
+   // Tx = (x * cpp) >> Tp
+   ld b32 $r10 D[$r5 + ctx_src_xoff]
+   ld b32 $r11 D[$r5 + ctx_src_cpp]
+   mulu $r10 $r11
+   mov $r11 1
+   shl b32 $r11 $r7
+   sub b32 $r11 1
+   and $r12 $r10 $r11
+   shr b32 $r10 $r7
+
+   // Tyo = y & ((1 << Th) - 1)
+   // Ty  = y >> Th
+   ld b32 $r13 D[$r5 + ctx_src_yoff]
+   mov $r14 1
+   shl b32 $r14 $r8
+   sub b32 $r14 1
+   and $r11 $r13 $r14
+   shr b32 $r13 $r8
+
+   // YTILE = ((1 << Th) << 12) | ((1 << Th) - Tyo)
+   add b32 $r14 1
+   shl b32 $r15 $r14 12
+   sub b32 $r14 $r11
+   or $r15 $r14
+   xbit $r6 $flags $p2
+   add b32 $r6 0x208
+   shl b32 $r6 8
+   iowr I[$r6 + 0x000] $r15
+
+   // Op += Tyo << Tp
+   shl b32 $r11 $r7
+   add b32 $r12 $r11
+
+   // nTx = ((w * cpp) + ((1 << Tp) - 1) >> Tp)
+   ld b32 $r15 D[$r5 + ctx_src_xsize]
+   ld b32 $r11 D[$r5 + ctx_src_cpp]
+   mulu $r15 $r11
+   mov $r11 1
+   shl b32 $r11 $r7
+   sub b32 $r11 1
+   add b32 $r15 $r11
+   shr b32 $r15 $r7
+   push $r15
+
+   // nTy = (h + ((1 << Th) - 1)) >> Th
+   ld b32 $r15 D[$r5 + ctx_src_ysize]
+   mov $r11 1
+   shl b32 $r11 $r8
+   sub b32 $r11 1
+   add b32 $r15 $r11
+   shr b32 $r15 $r8
+   push $r15
+
+   // Tys = Tp + Th
+   // CFG_YZ_TILE_SIZE = ((1 << Th) >> 2) << Td
+   add b32 $r7 $r8
+   sub b32 $r8 2
+   mov $r11 1
+   shl b32 $r11 $r8
+   shl b32 $r11 $r9
+
+   // Tzo = z & ((1 << Td) - 1)
+   // Tz  = z >> Td
+   // Op += Tzo << Tys
+   // Ts  = Tys + Td
+   ld b32 $r8 D[$r5 + ctx_src_zoff]
+   mov $r14 1
+   shl b32 $r14 $r9
+   sub b32 $r14 1
+   and $r15 $r8 $r14
+   shl b32 $r15 $r7
+   add b32 $r12 $r15
+   add b32 $r7 $r9
+   shr b32 $r8 $r9
+
+   // Ot = ((Tz * nTy * nTx) + (Ty * nTx) + Tx) << Ts
+   pop $r15
+   pop $r9
+   mulu $r13 $r9
+   add b32 $r10 $r13
+   mulu $r8 $r9
+   mulu $r8 $r15
+   add b32 $r10 $r8
+   shl b32 $r10 $r7
+
+   // PITCH = (nTx - 1) << Ts
+   sub b32 $r9 1
+   shl b32 $r9 $r7
+   iowr I[$r6 + 0x200] $r9
+
+   // SRC_ADDRESS_LOW   = (Ot + Op) & 0xffffffff
+   // CFG_ADDRESS_HIGH |= ((Ot + Op) >> 32) << 16
+   ld b32 $r7 D[$r5 + ctx_src_address_low]
+   ld b32 $r8 D[$r5 + ctx_src_address_high]
+   add b32 $r10 $r12
+   add b32 $r7 $r10
+   adc b32 $r8 0
+   shl b32 $r8 16
+   or $r8 $r11
+   sub b32 $r6 0x600
+   iowr I[$r6 + 0x000] $r7
+   add b32 $r6 0x400
+   iowr I[$r6 + 0x000] $r8
+   ret
+
+// Setup to handle a linear surface
+//
+// Nothing to see here.. Sets ADDRESS and PITCH, pretty non-exciting
+//
+cmd_exec_set_surface_linear:
+   xbit $r6 $flags $p2
+   add b32 $r6 0x202
+   shl b32 $r6 8
+   ld b32 $r7 D[$r5 + ctx_src_address_low]
+   iowr I[$r6 + 0x000] $r7
+   add b32 $r6 0x400
+   ld b32 $r7 D[$r5 + ctx_src_address_high]
+   shl b32 $r7 16
+   iowr I[$r6 + 0x000] $r7
+   add b32 $r6 0x400
+   ld b32 $r7 D[$r5 + ctx_src_pitch]
+   iowr I[$r6 + 0x000] $r7
+   ret
+
+// wait for regs to be available for use
+cmd_exec_wait:
+   push $r0
+   push $r1
+   mov $r0 0x800
+   shl b32 $r0 6
+   loop:
+      iord $r1 I[$r0]
+      and $r1 1
+      bra ne loop
+   pop $r1
+   pop $r0
+   ret
+
+cmd_exec_query:
+   // if QUERY_SHORT not set, write out { -, 0, TIME_LO, TIME_HI }
+   xbit $r4 $r3 13
+   bra ne query_counter
+      call cmd_exec_wait
+      mov $r4 0x80c
+      shl b32 $r4 6
+      ld b32 $r5 D[$r0 + ctx_query_address_low]
+      add b32 $r5 4
+      iowr I[$r4 + 0x000] $r5
+      iowr I[$r4 + 0x100] $r0
+      mov $r5 0xc
+      iowr I[$r4 + 0x200] $r5
+      add b32 $r4 0x400
+      ld b32 $r5 D[$r0 + ctx_query_address_high]
+      shl b32 $r5 16
+      iowr I[$r4 + 0x000] $r5
+      add b32 $r4 0x500
+      mov $r5 0x00000b00
+      sethi $r5 0x00010000
+      iowr I[$r4 + 0x000] $r5
+      mov $r5 0x00004040
+      shl b32 $r5 1
+      sethi $r5 0x80800000
+      iowr I[$r4 + 0x100] $r5
+      mov $r5 0x00001110
+      sethi $r5 0x13120000
+      iowr I[$r4 + 0x200] $r5
+      mov $r5 0x00001514
+      sethi $r5 0x17160000
+      iowr I[$r4 + 0x300] $r5
+      mov $r5 0x00002601
+      sethi $r5 0x00010000
+      mov $r4 0x800
+      shl b32 $r4 6
+      iowr I[$r4 + 0x000] $r5
+
+   // write COUNTER
+   query_counter:
+   call cmd_exec_wait
+   mov $r4 0x80c
+   shl b32 $r4 6
+   ld b32 $r5 D[$r0 + ctx_query_address_low]
+   iowr I[$r4 + 0x000] $r5
+   iowr I[$r4 + 0x100] $r0
+   mov $r5 0x4
+   iowr I[$r4 + 0x200] $r5
+   add b32 $r4 0x400
+   ld b32 $r5 D[$r0 + ctx_query_address_high]
+   shl b32 $r5 16
+   iowr I[$r4 + 0x000] $r5
+   add b32 $r4 0x500
+   mov $r5 0x00000300
+   iowr I[$r4 + 0x000] $r5
+   mov $r5 0x00001110
+   sethi $r5 0x13120000
+   iowr I[$r4 + 0x100] $r5
+   ld b32 $r5 D[$r0 + ctx_query_counter]
+   add b32 $r4 0x500
+   iowr I[$r4 + 0x000] $r5
+   mov $r5 0x00002601
+   sethi $r5 0x00010000
+   mov $r4 0x800
+   shl b32 $r4 6
+   iowr I[$r4 + 0x000] $r5
+   ret
+
+// Execute a copy operation
+//
+// Inputs:
+//    $r1: irqh state
+//    $r2: hostirq state
+//    $r3: data
+//       000002000 QUERY_SHORT
+//       000001000 QUERY
+//       000000100 DST_LINEAR
+//       000000010 SRC_LINEAR
+//       000000001 FORMAT
+//    $r4: dispatch table entry
+// Outputs:
+//    $r1: irqh state
+//    $p1: set on error
+//       $r2: hostirq state
+//       $r3: data
+cmd_exec:
+   call cmd_exec_wait
+
+   // if format requested, call function to calculate it, otherwise
+   // fill in cpp/xcnt for both surfaces as if (cpp == 1)
+   xbit $r15 $r3 0
+   bra e cmd_exec_no_format
+      call cmd_exec_set_format
+      mov $r4 0x200
+      bra cmd_exec_init_src_surface
+   cmd_exec_no_format:
+      mov $r6 0x810
+      shl b32 $r6 6
+      mov $r7 1
+      st b32 D[$r0 + ctx_src_cpp] $r7
+      st b32 D[$r0 + ctx_dst_cpp] $r7
+      ld b32 $r7 D[$r0 + ctx_xcnt]
+      iowr I[$r6 + 0x000] $r7
+      iowr I[$r6 + 0x100] $r7
+      clear b32 $r4
+
+   cmd_exec_init_src_surface:
+   bclr $flags $p2
+   clear b32 $r5
+   xbit $r15 $r3 4
+   bra e src_tiled
+      call cmd_exec_set_surface_linear
+      bra cmd_exec_init_dst_surface
+   src_tiled:
+      call cmd_exec_set_surface_tiled
+      bset $r4 7
+
+   cmd_exec_init_dst_surface:
+   bset $flags $p2
+   mov $r5 ctx_dst_address_high - ctx_src_address_high
+   xbit $r15 $r3 8
+   bra e dst_tiled
+      call cmd_exec_set_surface_linear
+      bra cmd_exec_kick
+   dst_tiled:
+      call cmd_exec_set_surface_tiled
+      bset $r4 8
+
+   cmd_exec_kick:
+   mov $r5 0x800
+   shl b32 $r5 6
+   ld b32 $r6 D[$r0 + ctx_ycnt]
+   iowr I[$r5 + 0x100] $r6
+   mov $r6 0x0041
+   // SRC_TARGET = 1, DST_TARGET = 2
+   sethi $r6 0x44000000
+   or $r4 $r6
+   iowr I[$r5] $r4
+
+   // if requested, queue up a QUERY write after the copy has completed
+   xbit $r15 $r3 12
+   bra e cmd_exec_done
+      call cmd_exec_query
+
+   cmd_exec_done:
+   ret
+
+// Flush write cache
+//
+// Inputs:
+//    $r1: irqh state
+//    $r2: hostirq state
+//    $r3: data
+//    $r4: dispatch table entry
+// Outputs:
+//    $r1: irqh state
+//    $p1: set on error
+//       $r2: hostirq state
+//       $r3: data
+cmd_wrcache_flush:
+   mov $r2 0x2200
+   clear b32 $r3
+   sethi $r3 0x10000
+   iowr I[$r2] $r3
+   ret
+
+.align 0x100
diff --git a/drivers/gpu/drm/nouveau/nva3_copy.fuc.h b/drivers/gpu/drm/nouveau/nva3_copy.fuc.h
new file mode 100644 (file)
index 0000000..2731de2
--- /dev/null
@@ -0,0 +1,534 @@
+uint32_t nva3_pcopy_data[] = {
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00010000,
+       0x00000000,
+       0x00000000,
+       0x00010040,
+       0x00010160,
+       0x00000000,
+       0x00010050,
+       0x00010162,
+       0x00000000,
+       0x00030060,
+       0x00010170,
+       0x00000000,
+       0x00010170,
+       0x00000000,
+       0x00010170,
+       0x00000000,
+       0x00070080,
+       0x00000028,
+       0xfffff000,
+       0x0000002c,
+       0xfff80000,
+       0x00000030,
+       0xffffe000,
+       0x00000034,
+       0xfffff800,
+       0x00000038,
+       0xfffff000,
+       0x0000003c,
+       0xfff80000,
+       0x00000040,
+       0xffffe000,
+       0x00070088,
+       0x00000054,
+       0xfffff000,
+       0x00000058,
+       0xfff80000,
+       0x0000005c,
+       0xffffe000,
+       0x00000060,
+       0xfffff800,
+       0x00000064,
+       0xfffff000,
+       0x00000068,
+       0xfff80000,
+       0x0000006c,
+       0xffffe000,
+       0x000200c0,
+       0x00010492,
+       0x00000000,
+       0x0001051b,
+       0x00000000,
+       0x000e00c3,
+       0x0000001c,
+       0xffffff00,
+       0x00000020,
+       0x0000000f,
+       0x00000048,
+       0xffffff00,
+       0x0000004c,
+       0x0000000f,
+       0x00000024,
+       0xfff80000,
+       0x00000050,
+       0xfff80000,
+       0x00000080,
+       0xffff0000,
+       0x00000084,
+       0xffffe000,
+       0x00000074,
+       0xfccc0000,
+       0x00000078,
+       0x00000000,
+       0x0000007c,
+       0x00000000,
+       0x00000010,
+       0xffffff00,
+       0x00000014,
+       0x00000000,
+       0x00000018,
+       0x00000000,
+       0x00000800,
+};
+
+uint32_t nva3_pcopy_code[] = {
+       0x04fe04bd,
+       0x3517f000,
+       0xf10010fe,
+       0xf1040017,
+       0xf0fff327,
+       0x22d00023,
+       0x0c25f0c0,
+       0xf40012d0,
+       0x17f11031,
+       0x27f01200,
+       0x0012d003,
+       0xf40031f4,
+       0x0ef40028,
+       0x8001cffd,
+       0xf40812c4,
+       0x21f4060b,
+       0x0412c472,
+       0xf4060bf4,
+       0x11c4c321,
+       0x4001d00c,
+       0x47f101f8,
+       0x4bfe7700,
+       0x0007fe00,
+       0xf00204b9,
+       0x01f40643,
+       0x0604fa09,
+       0xfa060ef4,
+       0x03f80504,
+       0x27f100f8,
+       0x23cf1400,
+       0x1e3fc800,
+       0xf4170bf4,
+       0x21f40132,
+       0x1e3af052,
+       0xf00023d0,
+       0x24d00147,
+       0xcf00f880,
+       0x3dc84023,
+       0x220bf41e,
+       0xf40131f4,
+       0x57f05221,
+       0x0367f004,
+       0xa07856bc,
+       0xb6018068,
+       0x87d00884,
+       0x0162b600,
+       0xf0f018f4,
+       0x23d00237,
+       0xf100f880,
+       0xcf190037,
+       0x33cf4032,
+       0xff24e400,
+       0x1024b607,
+       0x010057f1,
+       0x74bd64bd,
+       0x58005658,
+       0x50b60157,
+       0x0446b804,
+       0xbb4d08f4,
+       0x47b80076,
+       0x0f08f404,
+       0xb60276bb,
+       0x57bb0374,
+       0xdf0ef400,
+       0xb60246bb,
+       0x45bb0344,
+       0x01459800,
+       0xb00453fd,
+       0x1bf40054,
+       0x00455820,
+       0xb0014658,
+       0x1bf40064,
+       0x00538009,
+       0xf4300ef4,
+       0x55f90132,
+       0xf40c01f4,
+       0x25f0250e,
+       0x0125f002,
+       0x100047f1,
+       0xd00042d0,
+       0x27f04043,
+       0x0002d040,
+       0xf08002cf,
+       0x24b04024,
+       0xf71bf400,
+       0x1d0027f1,
+       0xd00137f0,
+       0x00f80023,
+       0x27f100f8,
+       0x34bd2200,
+       0xd00233f0,
+       0x00f80023,
+       0x012842b7,
+       0xf00145b6,
+       0x43801e39,
+       0x0040b701,
+       0x0644b606,
+       0xf80043d0,
+       0xf030f400,
+       0xb00001b0,
+       0x01b00101,
+       0x0301b002,
+       0xc71d0498,
+       0x50b63045,
+       0x3446c701,
+       0xc70160b6,
+       0x70b63847,
+       0x0232f401,
+       0x94bd84bd,
+       0xb60f4ac4,
+       0xb4bd0445,
+       0xf404a430,
+       0xa5ff0f18,
+       0x00cbbbc0,
+       0xf40231f4,
+       0x1bf4220e,
+       0x10c7f00c,
+       0xf400cbbb,
+       0xa430160e,
+       0x0c18f406,
+       0xbb14c7f0,
+       0x0ef400cb,
+       0x80c7f107,
+       0x01c83800,
+       0xb60180b6,
+       0xb5b801b0,
+       0xc308f404,
+       0xb80190b6,
+       0x08f40497,
+       0x0065fdb2,
+       0x98110680,
+       0x68fd2008,
+       0x0502f400,
+       0x75fd64bd,
+       0x1c078000,
+       0xf10078fd,
+       0xb6081057,
+       0x56d00654,
+       0x4057d000,
+       0x080050b7,
+       0xb61c0698,
+       0x64b60162,
+       0x11079808,
+       0xfd0172b6,
+       0x56d00567,
+       0x0050b700,
+       0x0060b401,
+       0xb40056d0,
+       0x56d00160,
+       0x0260b440,
+       0xb48056d0,
+       0x56d00360,
+       0x0050b7c0,
+       0x1e069804,
+       0x980056d0,
+       0x56d01f06,
+       0x1030f440,
+       0x579800f8,
+       0x6879c70a,
+       0xb66478c7,
+       0x77c70280,
+       0x0e76b060,
+       0xf0091bf4,
+       0x0ef40477,
+       0x027cf00f,
+       0xfd1170b6,
+       0x77f00947,
+       0x0f5a9806,
+       0xfd115b98,
+       0xb7f000ab,
+       0x04b7bb01,
+       0xff01b2b6,
+       0xa7bbc4ab,
+       0x105d9805,
+       0xbb01e7f0,
+       0xe2b604e8,
+       0xb4deff01,
+       0xb605d8bb,
+       0xef9401e0,
+       0x02ebbb0c,
+       0xf005fefd,
+       0x60b7026c,
+       0x64b60208,
+       0x006fd008,
+       0xbb04b7bb,
+       0x5f9800cb,
+       0x115b980b,
+       0xf000fbfd,
+       0xb7bb01b7,
+       0x01b2b604,
+       0xbb00fbbb,
+       0xf0f905f7,
+       0xf00c5f98,
+       0xb8bb01b7,
+       0x01b2b604,
+       0xbb00fbbb,
+       0xf0f905f8,
+       0xb60078bb,
+       0xb7f00282,
+       0x04b8bb01,
+       0x9804b9bb,
+       0xe7f00e58,
+       0x04e9bb01,
+       0xff01e2b6,
+       0xf7bbf48e,
+       0x00cfbb04,
+       0xbb0079bb,
+       0xf0fc0589,
+       0xd9fd90fc,
+       0x00adbb00,
+       0xfd0089fd,
+       0xa8bb008f,
+       0x04a7bb00,
+       0xbb0192b6,
+       0x69d00497,
+       0x08579880,
+       0xbb075898,
+       0x7abb00ac,
+       0x0081b600,
+       0xfd1084b6,
+       0x62b7058b,
+       0x67d00600,
+       0x0060b700,
+       0x0068d004,
+       0x6cf000f8,
+       0x0260b702,
+       0x0864b602,
+       0xd0085798,
+       0x60b70067,
+       0x57980400,
+       0x1074b607,
+       0xb70067d0,
+       0x98040060,
+       0x67d00957,
+       0xf900f800,
+       0xf110f900,
+       0xb6080007,
+       0x01cf0604,
+       0x0114f000,
+       0xfcfa1bf4,
+       0xf800fc10,
+       0x0d34c800,
+       0xf5701bf4,
+       0xf103ab21,
+       0xb6080c47,
+       0x05980644,
+       0x0450b605,
+       0xd00045d0,
+       0x57f04040,
+       0x8045d00c,
+       0x040040b7,
+       0xb6040598,
+       0x45d01054,
+       0x0040b700,
+       0x0057f105,
+       0x0153f00b,
+       0xf10045d0,
+       0xb6404057,
+       0x53f10154,
+       0x45d08080,
+       0x1057f140,
+       0x1253f111,
+       0x8045d013,
+       0x151457f1,
+       0x171653f1,
+       0xf1c045d0,
+       0xf0260157,
+       0x47f10153,
+       0x44b60800,
+       0x0045d006,
+       0x03ab21f5,
+       0x080c47f1,
+       0x980644b6,
+       0x45d00505,
+       0x4040d000,
+       0xd00457f0,
+       0x40b78045,
+       0x05980400,
+       0x1054b604,
+       0xb70045d0,
+       0xf1050040,
+       0xd0030057,
+       0x57f10045,
+       0x53f11110,
+       0x45d01312,
+       0x06059840,
+       0x050040b7,
+       0xf10045d0,
+       0xf0260157,
+       0x47f10153,
+       0x44b60800,
+       0x0045d006,
+       0x21f500f8,
+       0x3fc803ab,
+       0x0e0bf400,
+       0x018921f5,
+       0x020047f1,
+       0xf11e0ef4,
+       0xb6081067,
+       0x77f00664,
+       0x11078001,
+       0x981c0780,
+       0x67d02007,
+       0x4067d000,
+       0x32f444bd,
+       0xc854bd02,
+       0x0bf4043f,
+       0x8221f50a,
+       0x0a0ef403,
+       0x027621f5,
+       0xf40749f0,
+       0x57f00231,
+       0x083fc82c,
+       0xf50a0bf4,
+       0xf4038221,
+       0x21f50a0e,
+       0x49f00276,
+       0x0057f108,
+       0x0654b608,
+       0xd0210698,
+       0x67f04056,
+       0x0063f141,
+       0x0546fd44,
+       0xc80054d0,
+       0x0bf40c3f,
+       0xc521f507,
+       0xf100f803,
+       0xbd220027,
+       0x0133f034,
+       0xf80023d0,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+       0x00000000,
+};