drm/radeon: re-enable PTE/PDE packet for set_page on cayman/TN

PTE/PDE doesn't support a single update (count = 1).  We had
previously disabled it since it we were hitting that case which
let to hangs.  The PTE/PDE packet is much more efficient for VM
updates where it can be used.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

2 files changed, 52 insertions, 18 deletions

diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c
index 78aead0ada7b..fd03f318cc1c 100644
--- a/drivers/gpu/drm/radeon/ni.c
+++ b/drivers/gpu/drm/radeon/ni.c
@@ -2079,28 +2079,57 @@ void cayman_vm_set_page(struct radeon_device *rdev,
                         }
                 }
         } else {
-                while (count) {
-                        ndw = count * 2;
-                        if (ndw > 0xFFFFE)
-                                ndw = 0xFFFFE;
+                if ((flags & RADEON_VM_PAGE_SYSTEM) ||
+                    (count == 1)) {
+                        while (count) {
+                                ndw = count * 2;
+                                if (ndw > 0xFFFFE)
+                                        ndw = 0xFFFFE;
+
+                                /* for non-physically contiguous pages (system) */
+                                ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, ndw);
+                                ib->ptr[ib->length_dw++] = pe;
+                                ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
+                                for (; ndw > 0; ndw -= 2, --count, pe += 8) {
+                                        if (flags & RADEON_VM_PAGE_SYSTEM) {
+                                                value = radeon_vm_map_gart(rdev, addr);
+                                                value &= 0xFFFFFFFFFFFFF000ULL;
+                                        } else if (flags & RADEON_VM_PAGE_VALID) {
+                                                value = addr;
+                                        } else {
+                                                value = 0;
+                                        }
+                                        addr += incr;
+                                        value |= r600_flags;
+                                        ib->ptr[ib->length_dw++] = value;
+                                        ib->ptr[ib->length_dw++] = upper_32_bits(value);
+                                }
+                        }
+                        while (ib->length_dw & 0x7)
+                                ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0);
+                } else {
+                        while (count) {
+                                ndw = count * 2;
+                                if (ndw > 0xFFFFE)
+                                        ndw = 0xFFFFE;
 
-                        /* for non-physically contiguous pages (system) */
-                        ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, ndw);
-                        ib->ptr[ib->length_dw++] = pe;
-                        ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
-                        for (; ndw > 0; ndw -= 2, --count, pe += 8) {
-                                if (flags & RADEON_VM_PAGE_SYSTEM) {
-                                        value = radeon_vm_map_gart(rdev, addr);
-                                        value &= 0xFFFFFFFFFFFFF000ULL;
-                                } else if (flags & RADEON_VM_PAGE_VALID) {
+                                if (flags & RADEON_VM_PAGE_VALID)
                                         value = addr;
-                                } else {
+                                else
                                         value = 0;
-                                }
-                                addr += incr;
-                                value |= r600_flags;
-                                ib->ptr[ib->length_dw++] = value;
+                                /* for physically contiguous pages (vram) */
+                                ib->ptr[ib->length_dw++] = DMA_PTE_PDE_PACKET(ndw);
+                                ib->ptr[ib->length_dw++] = pe; /* dst addr */
+                                ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
+                                ib->ptr[ib->length_dw++] = r600_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;
+                                pe += ndw * 4;
+                                addr += (ndw / 2) * incr;
+                                count -= ndw / 2;
                         }
                 }
                 while (ib->length_dw & 0x7)
diff --git a/drivers/gpu/drm/radeon/nid.h b/drivers/gpu/drm/radeon/nid.h
index 71a1709de154..e226faf16fea 100644
--- a/drivers/gpu/drm/radeon/nid.h
+++ b/drivers/gpu/drm/radeon/nid.h
@@ -684,6 +684,11 @@
                                          (((vmid) & 0xF) << 20) |       \
                                          (((n) & 0xFFFFF) << 0))
 
+#define DMA_PTE_PDE_PACKET(n)           ((2 << 28) |                    \
+                                         (1 << 26) |                    \
+                                         (1 << 21) |                    \
+                                         (((n) & 0xFFFFF) << 0))
+
 /* async DMA Packet types */
 #define DMA_PACKET_WRITE                                  0x2
 #define DMA_PACKET_COPY                                   0x3