drm/nouveau/kms/nv50-: also flush fb writes when rewinding push buffer

Should hopefully fix a regression some people have been seeing since EVO
push buffers were moved to VRAM by default on Pascal GPUs.

Fixes: d00ddd9d ("drm/nouveau/kms/nv50-: allocate push buffers in vidmem on pascal")
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Cc: <stable@vger.kernel.org> # 4.19+

drivers/gpu/drm/nouveau/dispnv50/disp.c

@@ -198,6 +198,22 @@ nv50_dmac_create(struct nvif_device *device, struct nvif_object *disp,
 /******************************************************************************
  * EVO channel helpers
  *****************************************************************************/
+static void
+evo_flush(struct nv50_dmac *dmac)
+{
+	/* Push buffer fetches are not coherent with BAR1, we need to ensure
+	 * writes have been flushed right through to VRAM before writing PUT.
+	 */
+	if (dmac->push.type & NVIF_MEM_VRAM) {
+		struct nvif_device *device = dmac->base.device;
+		nvif_wr32(&device->object, 0x070000, 0x00000001);
+		nvif_msec(device, 2000,
+			if (!(nvif_rd32(&device->object, 0x070000) & 0x00000002))
+				break;
+		);
+	}
+}
+
 u32 *
 evo_wait(struct nv50_dmac *evoc, int nr)
 {
@@ -208,6 +224,7 @@ evo_wait(struct nv50_dmac *evoc, int nr)
 	mutex_lock(&dmac->lock);
 	if (put + nr >= (PAGE_SIZE / 4) - 8) {
 		dmac->ptr[put] = 0x20000000;
+		evo_flush(dmac);
 
 		nvif_wr32(&dmac->base.user, 0x0000, 0x00000000);
 		if (nvif_msec(device, 2000,
@@ -230,17 +247,7 @@ evo_kick(u32 *push, struct nv50_dmac *evoc)
 {
 	struct nv50_dmac *dmac = evoc;
 
-	/* Push buffer fetches are not coherent with BAR1, we need to ensure
-	 * writes have been flushed right through to VRAM before writing PUT.
-	 */
-	if (dmac->push.type & NVIF_MEM_VRAM) {
-		struct nvif_device *device = dmac->base.device;
-		nvif_wr32(&device->object, 0x070000, 0x00000001);
-		nvif_msec(device, 2000,
-			if (!(nvif_rd32(&device->object, 0x070000) & 0x00000002))
-				break;
-		);
-	}
+	evo_flush(dmac);
 
 	nvif_wr32(&dmac->base.user, 0x0000, (push - dmac->ptr) << 2);
 	mutex_unlock(&dmac->lock);