drm/nouveau/kms/nv50-: cursors always use core channel vram ctxdma

Ctxdmas for cursors from all heads are setup in the core channel, and due
to us tracking allocated handles per-window, we were failing with -EEXIST
on multiple-head setups trying to allocate duplicate handles.

The cursor code is hardcoded to use the core channel vram ctxdma already,
so just skip ctxdma allocation for cursor fbs to fix the issue.

Fixes: 5bca1621c0 ("drm/nouveau/kms/nv50-: move fb ctxdma tracking into windows")
Reported-by: Adam Borowski <kilobyte@angband.pl>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drivers/gpu/drm/nouveau/dispnv50/curs507a.c

@@ -132,7 +132,7 @@ curs507a_new_(const struct nv50_wimm_func *func, struct nouveau_drm *drm,
 
 	nvif_object_map(&wndw->wimm.base.user, NULL, 0);
 	wndw->immd = func;
-	wndw->ctxdma.parent = &disp->core->chan.base.user;
+	wndw->ctxdma.parent = NULL;
 	return 0;
 }
 

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

@@ -444,14 +444,17 @@ nv50_wndw_prepare_fb(struct drm_plane *plane, struct drm_plane_state *state)
 	if (ret)
 		return ret;
 
+	if (wndw->ctxdma.parent) {
 		ctxdma = nv50_wndw_ctxdma_new(wndw, fb);
 		if (IS_ERR(ctxdma)) {
 			nouveau_bo_unpin(fb->nvbo);
 			return PTR_ERR(ctxdma);
 		}
 
-	asyw->state.fence = reservation_object_get_excl_rcu(fb->nvbo->bo.resv);
 		asyw->image.handle[0] = ctxdma->object.handle;
+	}
+
+	asyw->state.fence = reservation_object_get_excl_rcu(fb->nvbo->bo.resv);
 	asyw->image.offset[0] = fb->nvbo->bo.offset;
 
 	if (wndw->func->prepare) {