Merge branch 'drm-next-4.9' of git://people.freedesktop.org/~agd5f/linux into drm...

[linux-2.6-block.git] / drivers / gpu / drm / amd / amdgpu / sdma_v2_4.c

diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
index 1351c7e834a21653a358ad3578dad91f60de9da9..9ae307505190ec0afd81bc0b049d7b2b76bff3b0 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -190,12 +190,8 @@ out:
  */
 static uint32_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)
 {
-       u32 rptr;
-
        /* XXX check if swapping is necessary on BE */
-       rptr = ring->adev->wb.wb[ring->rptr_offs] >> 2;
-
-       return rptr;
+       return ring->adev->wb.wb[ring->rptr_offs] >> 2;
 }
 
 /**
@@ -749,24 +745,16 @@ static void sdma_v2_4_vm_copy_pte(struct amdgpu_ib *ib,
                                  uint64_t pe, uint64_t src,
                                  unsigned count)
 {
-       while (count) {
-               unsigned bytes = count * 8;
-               if (bytes > 0x1FFFF8)
-                       bytes = 0x1FFFF8;
-
-               ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
-                       SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
-               ib->ptr[ib->length_dw++] = bytes;
-               ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
-               ib->ptr[ib->length_dw++] = lower_32_bits(src);
-               ib->ptr[ib->length_dw++] = upper_32_bits(src);
-               ib->ptr[ib->length_dw++] = lower_32_bits(pe);
-               ib->ptr[ib->length_dw++] = upper_32_bits(pe);
-
-               pe += bytes;
-               src += bytes;
-               count -= bytes / 8;
-       }
+       unsigned bytes = count * 8;
+
+       ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+               SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+       ib->ptr[ib->length_dw++] = bytes;
+       ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+       ib->ptr[ib->length_dw++] = lower_32_bits(src);
+       ib->ptr[ib->length_dw++] = upper_32_bits(src);
+       ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+       ib->ptr[ib->length_dw++] = upper_32_bits(pe);
 }
 
 /**
@@ -774,39 +762,27 @@ static void sdma_v2_4_vm_copy_pte(struct amdgpu_ib *ib,
  *
  * @ib: indirect buffer to fill with commands
  * @pe: addr of the page entry
- * @addr: dst addr to write into pe
+ * @value: dst addr to write into pe
  * @count: number of page entries to update
  * @incr: increase next addr by incr bytes
- * @flags: access flags
  *
  * Update PTEs by writing them manually using sDMA (CIK).
  */
-static void sdma_v2_4_vm_write_pte(struct amdgpu_ib *ib,
-                                  const dma_addr_t *pages_addr, uint64_t pe,
-                                  uint64_t addr, unsigned count,
-                                  uint32_t incr, uint32_t flags)
+static void sdma_v2_4_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+                                  uint64_t value, unsigned count,
+                                  uint32_t incr)
 {
-       uint64_t value;
-       unsigned ndw;
-
-       while (count) {
-               ndw = count * 2;
-               if (ndw > 0xFFFFE)
-                       ndw = 0xFFFFE;
-
-               /* for non-physically contiguous pages (system) */
-               ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
-                       SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
-               ib->ptr[ib->length_dw++] = pe;
-               ib->ptr[ib->length_dw++] = upper_32_bits(pe);
-               ib->ptr[ib->length_dw++] = ndw;
-               for (; ndw > 0; ndw -= 2, --count, pe += 8) {
-                       value = amdgpu_vm_map_gart(pages_addr, addr);
-                       addr += incr;
-                       value |= flags;
-                       ib->ptr[ib->length_dw++] = value;
-                       ib->ptr[ib->length_dw++] = upper_32_bits(value);
-               }
+       unsigned ndw = count * 2;
+
+       ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+               SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+       ib->ptr[ib->length_dw++] = pe;
+       ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+       ib->ptr[ib->length_dw++] = ndw;
+       for (; ndw > 0; ndw -= 2, --count, pe += 8) {
+               ib->ptr[ib->length_dw++] = lower_32_bits(value);
+               ib->ptr[ib->length_dw++] = upper_32_bits(value);
+               value += incr;
        }
 }
 
@@ -822,40 +798,21 @@ static void sdma_v2_4_vm_write_pte(struct amdgpu_ib *ib,
  *
  * Update the page tables using sDMA (CIK).
  */
-static void sdma_v2_4_vm_set_pte_pde(struct amdgpu_ib *ib,
-                                    uint64_t pe,
+static void sdma_v2_4_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe,
                                     uint64_t addr, unsigned count,
                                     uint32_t incr, uint32_t flags)
 {
-       uint64_t value;
-       unsigned ndw;
-
-       while (count) {
-               ndw = count;
-               if (ndw > 0x7FFFF)
-                       ndw = 0x7FFFF;
-
-               if (flags & AMDGPU_PTE_VALID)
-                       value = addr;
-               else
-                       value = 0;
-
-               /* for physically contiguous pages (vram) */
-               ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE);
-               ib->ptr[ib->length_dw++] = pe; /* dst addr */
-               ib->ptr[ib->length_dw++] = upper_32_bits(pe);
-               ib->ptr[ib->length_dw++] = flags; /* mask */
-               ib->ptr[ib->length_dw++] = 0;
-               ib->ptr[ib->length_dw++] = value; /* value */
-               ib->ptr[ib->length_dw++] = upper_32_bits(value);
-               ib->ptr[ib->length_dw++] = incr; /* increment size */
-               ib->ptr[ib->length_dw++] = 0;
-               ib->ptr[ib->length_dw++] = ndw; /* number of entries */
-
-               pe += ndw * 8;
-               addr += ndw * incr;
-               count -= ndw;
-       }
+       /* for physically contiguous pages (vram) */
+       ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE);
+       ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */
+       ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+       ib->ptr[ib->length_dw++] = flags; /* mask */
+       ib->ptr[ib->length_dw++] = 0;
+       ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */
+       ib->ptr[ib->length_dw++] = upper_32_bits(addr);
+       ib->ptr[ib->length_dw++] = incr; /* increment size */
+       ib->ptr[ib->length_dw++] = 0;
+       ib->ptr[ib->length_dw++] = count; /* number of entries */
 }
 
 /**
@@ -945,6 +902,22 @@ static void sdma_v2_4_ring_emit_vm_flush(struct amdgpu_ring *ring,
                          SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
 }
 
+static unsigned sdma_v2_4_ring_get_emit_ib_size(struct amdgpu_ring *ring)
+{
+       return
+               7 + 6; /* sdma_v2_4_ring_emit_ib */
+}
+
+static unsigned sdma_v2_4_ring_get_dma_frame_size(struct amdgpu_ring *ring)
+{
+       return
+               6 + /* sdma_v2_4_ring_emit_hdp_flush */
+               3 + /* sdma_v2_4_ring_emit_hdp_invalidate */
+               6 + /* sdma_v2_4_ring_emit_pipeline_sync */
+               12 + /* sdma_v2_4_ring_emit_vm_flush */
+               10 + 10 + 10; /* sdma_v2_4_ring_emit_fence x3 for user fence, vm fence */
+}
+
 static int sdma_v2_4_early_init(void *handle)
 {
        struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -1263,6 +1236,8 @@ static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
        .test_ib = sdma_v2_4_ring_test_ib,
        .insert_nop = sdma_v2_4_ring_insert_nop,
        .pad_ib = sdma_v2_4_ring_pad_ib,
+       .get_emit_ib_size = sdma_v2_4_ring_get_emit_ib_size,
+       .get_dma_frame_size = sdma_v2_4_ring_get_dma_frame_size,
 };
 
 static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev)