Merge tag 'v4.10-rc5' into for-linus

Sync up with mainline to apply fixup to a commit that came through
power supply tree.

1 files changed, 62 insertions, 84 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
index 77fdd9911c3c..4c34dbc7a254 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -206,10 +206,10 @@ static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
 
 	for (i = 0; i < count; i++)
 		if (sdma && sdma->burst_nop && (i == 0))
-			amdgpu_ring_write(ring, ring->nop |
+			amdgpu_ring_write(ring, ring->funcs->nop |
 					  SDMA_NOP_COUNT(count - 1));
 		else
-			amdgpu_ring_write(ring, ring->nop);
+			amdgpu_ring_write(ring, ring->funcs->nop);
 }
 
 /**
@@ -622,7 +622,7 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib ib;
-	struct fence *f = NULL;
+	struct dma_fence *f = NULL;
 	unsigned index;
 	u32 tmp = 0;
 	u64 gpu_addr;
@@ -655,7 +655,7 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	if (r)
 		goto err1;
 
-	r = fence_wait_timeout(f, false, timeout);
+	r = dma_fence_wait_timeout(f, false, timeout);
 	if (r == 0) {
 		DRM_ERROR("amdgpu: IB test timed out\n");
 		r = -ETIMEDOUT;
@@ -675,7 +675,7 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 
 err1:
 	amdgpu_ib_free(adev, &ib, NULL);
-	fence_put(f);
+	dma_fence_put(f);
 err0:
 	amdgpu_wb_free(adev, index);
 	return r;
@@ -695,24 +695,16 @@ static void cik_sdma_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;
+	unsigned bytes = count * 8;
 
-		ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY,
-			SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
-		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;
-	}
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY,
+		SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+	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);
 }
 
 /**
@@ -720,39 +712,27 @@ static void cik_sdma_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 cik_sdma_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 cik_sdma_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe,
+				  uint64_t value, unsigned count,
+				  uint32_t incr)
 {
-	uint64_t value;
-	unsigned ndw;
+	unsigned ndw = count * 2;
 
-	while (count) {
-		ndw = count * 2;
-		if (ndw > 0xFFFFE)
-			ndw = 0xFFFFE;
-
-		/* for non-physically contiguous pages (system) */
-		ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_WRITE,
-			SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
-		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);
-		}
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_WRITE,
+		SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+	ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+	ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+	ib->ptr[ib->length_dw++] = ndw;
+	for (; ndw > 0; ndw -= 2) {
+		ib->ptr[ib->length_dw++] = lower_32_bits(value);
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		value += incr;
 	}
 }
 
@@ -768,40 +748,21 @@ static void cik_sdma_vm_write_pte(struct amdgpu_ib *ib,
  *
  * Update the page tables using sDMA (CIK).
  */
-static void cik_sdma_vm_set_pte_pde(struct amdgpu_ib *ib,
-				    uint64_t pe,
+static void cik_sdma_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_PACKET(SDMA_OPCODE_GENERATE_PTE_PDE, 0, 0);
-		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_PACKET(SDMA_OPCODE_GENERATE_PTE_PDE, 0, 0);
+	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 */
 }
 
 /**
@@ -982,11 +943,10 @@ static int cik_sdma_sw_init(void *handle)
 		ring->ring_obj = NULL;
 		sprintf(ring->name, "sdma%d", i);
 		r = amdgpu_ring_init(adev, ring, 1024,
-				     SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0), 0xf,
 				     &adev->sdma.trap_irq,
 				     (i == 0) ?
-				     AMDGPU_SDMA_IRQ_TRAP0 : AMDGPU_SDMA_IRQ_TRAP1,
-				     AMDGPU_RING_TYPE_SDMA);
+				     AMDGPU_SDMA_IRQ_TRAP0 :
+				     AMDGPU_SDMA_IRQ_TRAP1);
 		if (r)
 			return r;
 	}
@@ -1230,7 +1190,7 @@ static int cik_sdma_set_powergating_state(void *handle,
 	return 0;
 }
 
-const struct amd_ip_funcs cik_sdma_ip_funcs = {
+static const struct amd_ip_funcs cik_sdma_ip_funcs = {
 	.name = "cik_sdma",
 	.early_init = cik_sdma_early_init,
 	.late_init = NULL,
@@ -1248,10 +1208,19 @@ const struct amd_ip_funcs cik_sdma_ip_funcs = {
 };
 
 static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_SDMA,
+	.align_mask = 0xf,
+	.nop = SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0),
 	.get_rptr = cik_sdma_ring_get_rptr,
 	.get_wptr = cik_sdma_ring_get_wptr,
 	.set_wptr = cik_sdma_ring_set_wptr,
-	.parse_cs = NULL,
+	.emit_frame_size =
+		6 + /* cik_sdma_ring_emit_hdp_flush */
+		3 + /* cik_sdma_ring_emit_hdp_invalidate */
+		6 + /* cik_sdma_ring_emit_pipeline_sync */
+		12 + /* cik_sdma_ring_emit_vm_flush */
+		9 + 9 + 9, /* cik_sdma_ring_emit_fence x3 for user fence, vm fence */
+	.emit_ib_size = 7 + 4, /* cik_sdma_ring_emit_ib */
 	.emit_ib = cik_sdma_ring_emit_ib,
 	.emit_fence = cik_sdma_ring_emit_fence,
 	.emit_pipeline_sync = cik_sdma_ring_emit_pipeline_sync,
@@ -1373,3 +1342,12 @@ static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev)
 		adev->vm_manager.vm_pte_num_rings = adev->sdma.num_instances;
 	}
 }
+
+const struct amdgpu_ip_block_version cik_sdma_ip_block =
+{
+	.type = AMD_IP_BLOCK_TYPE_SDMA,
+	.major = 2,
+	.minor = 0,
+	.rev = 0,
+	.funcs = &cik_sdma_ip_funcs,
+};