On Mon, Oct 29, 2018 at 03:24:29PM +0100, Daniel Vetter wrote:
> On Fri, Oct 26, 2018 at 04:35:47PM -0400, Lyude Paul wrote:
> > There has been a TODO waiting for quite a long time in
> > drm_dp_mst_topology.c:
> >
> > /* We cannot rely on port->vcpi.num_slots to update
> > * topology_state->avail_slots as the port may not exist if the parent
> > * branch device was unplugged. This should be fixed by tracking
> > * per-port slot allocation in drm_dp_mst_topology_state instead of
> > * depending on the caller to tell us how many slots to release.
> > */
> >
> > That's not the only reason we should fix this: forcing the driver to
> > track the VCPI allocations throughout a state's atomic check is
> > error prone, because it means that extra care has to be taken with the
> > order that drm_dp_atomic_find_vcpi_slots() and
> > drm_dp_atomic_release_vcpi_slots() are called in in order to ensure
> > idempotency. Currently the only driver actually using these helpers,
> > i915, doesn't even do this correctly: multiple ->best_encoder() checks
> > with i915's current implementation would not be idempotent and would
> > over-allocate VCPI slots, something I learned trying to implement
> > fallback retraining in MST.
> >
> > So: simplify this whole mess, and teach drm_dp_atomic_find_vcpi_slots()
> > and drm_dp_atomic_release_vcpi_slots() to track the VCPI allocations for
> > each port. This allows us to ensure idempotency without having to rely
> > on the driver as much. Additionally: the driver doesn't need to do any
> > kind of VCPI slot tracking anymore if it doesn't need it for it's own
> > internal state.
> >
> > Additionally; this adds a new drm_dp_mst_atomic_check() helper which
> > must be used by atomic drivers to perform validity checks for the new
> > VCPI allocations incurred by a state.
> >
> > Also: update the documentation and make it more obvious that these
> > /must/ be called by /all/ atomic drivers supporting MST.
> >
> > Changes since v1:
> > - Don't use the now-removed ->atomic_check() for private objects hook,
> > just give drivers a function to call themselves
> >
> > Signed-off-by: Lyude Paul <lyude@xxxxxxxxxx>
> > Cc: Daniel Vetter <daniel.vetter@xxxxxxxx>
> > ---
> > drivers/gpu/drm/drm_dp_mst_topology.c | 190 +++++++++++++++++++++-----
> > drivers/gpu/drm/i915/intel_display.c | 8 ++
> > drivers/gpu/drm/i915/intel_dp_mst.c | 31 +++--
> > include/drm/drm_dp_mst_helper.h | 11 +-
> > 4 files changed, 192 insertions(+), 48 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> > index 8c3cfac437f4..dcfab7536914 100644
> > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > @@ -2614,21 +2614,33 @@ static int drm_dp_init_vcpi(struct drm_dp_mst_topology_mgr *mgr,
> > }
> >
> > /**
> > - * drm_dp_atomic_find_vcpi_slots() - Find and add vcpi slots to the state
> > + * drm_dp_atomic_find_vcpi_slots() - Find and add VCPI slots to the state
> > * @state: global atomic state
> > * @mgr: MST topology manager for the port
> > * @port: port to find vcpi slots for
> > * @pbn: bandwidth required for the mode in PBN
> > *
> > + * Allocates VCPI slots to @port, replacing any previous VCPI allocations it
> > + * may have had. Any atomic drivers which support MST must call this function
> > + * in their atomic_check() handlers to change the current VCPI allocation for
>
> Maybe do a nice kerneldoc reference to the right atomic_check here.
>
> > + * the new state. After the ->atomic_check() hooks of the driver and all other
>
> This will upset the kerneldoc parser I think.
>
> > + * mode objects in the state have been called, DRM will check the final VCPI
> > + * allocations to ensure that they will fit into the available bandwidth on
> > + * the topology.
> > + *
> > + * See also: drm_dp_atomic_release_vcpi_slots()
>
> Also need to reference drm_dp_mst_atomic_check() here and that drivers
> must call it or nothing happens.
> > + *
> > * RETURNS:
> > - * Total slots in the atomic state assigned for this port or error
> > + * Total slots in the atomic state assigned for this port, or a negative error
> > + * code if the port no longer exists
> > */
> > int drm_dp_atomic_find_vcpi_slots(struct drm_atomic_state *state,
> > struct drm_dp_mst_topology_mgr *mgr,
> > struct drm_dp_mst_port *port, int pbn)
> > {
> > struct drm_dp_mst_topology_state *topology_state;
> > - int req_slots;
> > + struct drm_dp_vcpi_allocation *pos, *vcpi = NULL;
> > + int prev_slots, req_slots, ret;
> >
> > topology_state = drm_atomic_get_mst_topology_state(state, mgr);
> > if (IS_ERR(topology_state))
> > @@ -2637,20 +2649,41 @@ int drm_dp_atomic_find_vcpi_slots(struct drm_atomic_state *state,
> > port = drm_dp_get_validated_port_ref(mgr, port);
> > if (port == NULL)
> > return -EINVAL;
> > - req_slots = DIV_ROUND_UP(pbn, mgr->pbn_div);
> > - DRM_DEBUG_KMS("vcpi slots req=%d, avail=%d\n",
> > - req_slots, topology_state->avail_slots);
> >
> > - if (req_slots > topology_state->avail_slots) {
> > - drm_dp_put_port(port);
> > - return -ENOSPC;
> > + /* Find the current allocation for this port, if any */
> > + list_for_each_entry(pos, &topology_state->vcpis, next) {
> > + if (pos->port == port) {
> > + vcpi = pos;
> > + prev_slots = vcpi->vcpi;
> > + break;
> > + }
> > }
> > + if (!vcpi)
> > + prev_slots = 0;
>
> For robustness should we warn here, since drivers forgetting to release
> vcpi slots is kinda a bug? Or do we need to have this to make life easier
> for driver writers?
>
> > +
> > + req_slots = DIV_ROUND_UP(pbn, mgr->pbn_div);
> > +
> > + DRM_DEBUG_KMS("[CONNECTOR:%d:%s] [MST PORT:%p] vcpi %d -> %d\n",
> > + port->connector->base.id, port->connector->name, port,
> > + prev_slots, req_slots);
> > +
> > + /* Add the new allocation to the state */
> > + if (!vcpi) {
> > + vcpi = kzalloc(sizeof(*vcpi), GFP_KERNEL);
> > + if (!vcpi) {
> > + ret = -ENOMEM;
> > + goto out;
> > + }
> >
> > - topology_state->avail_slots -= req_slots;
> > - DRM_DEBUG_KMS("vcpi slots avail=%d", topology_state->avail_slots);
> > + vcpi->port = port;
> > + list_add(&vcpi->next, &topology_state->vcpis);
> > + }
> > + vcpi->vcpi = req_slots;
> >
> > + ret = req_slots;
> > +out:
> > drm_dp_put_port(port);
> > - return req_slots;
> > + return ret;
> > }
> > EXPORT_SYMBOL(drm_dp_atomic_find_vcpi_slots);
> >
> > @@ -2658,32 +2691,46 @@ EXPORT_SYMBOL(drm_dp_atomic_find_vcpi_slots);
> > * drm_dp_atomic_release_vcpi_slots() - Release allocated vcpi slots
> > * @state: global atomic state
> > * @mgr: MST topology manager for the port
> > - * @slots: number of vcpi slots to release
> > + *
> > + * Releases any VCPI slots that have been allocated to a port in the atomic
> > + * state. Any atomic drivers which support MST must call this function in
> > + * their connector's atomic_check() handler when the connector will no longer
> > + * have VCPI allocated (e.g. because it's CRTC was removed).
> > + *
> > + * It is OK to call this even if @port has been removed from the system, in
> > + * which case it will just amount to a no-op.
> > + *
> > + * See also: drm_dp_atomic_find_vcpi_slots()
>
> Same comments as with the _find_vcpi_slots() function.
>
> > *
> > * RETURNS:
> > - * 0 if @slots were added back to &drm_dp_mst_topology_state->avail_slots or
> > - * negative error code
> > + * 0 if all slots for this port were added back to
> > + * &drm_dp_mst_topology_state->avail_slots or negative error code
> > */
> > int drm_dp_atomic_release_vcpi_slots(struct drm_atomic_state *state,
> > struct drm_dp_mst_topology_mgr *mgr,
> > - int slots)
> > + struct drm_dp_mst_port *port)
> > {
> > struct drm_dp_mst_topology_state *topology_state;
> > + struct drm_dp_vcpi_allocation *tmp, *pos;
> >
> > topology_state = drm_atomic_get_mst_topology_state(state, mgr);
> > if (IS_ERR(topology_state))
> > return PTR_ERR(topology_state);
> >
> > - /* We cannot rely on port->vcpi.num_slots to update
> > - * topology_state->avail_slots as the port may not exist if the parent
> > - * branch device was unplugged. This should be fixed by tracking
> > - * per-port slot allocation in drm_dp_mst_topology_state instead of
> > - * depending on the caller to tell us how many slots to release.
> > - */
> > - topology_state->avail_slots += slots;
> > - DRM_DEBUG_KMS("vcpi slots released=%d, avail=%d\n",
> > - slots, topology_state->avail_slots);
> > + list_for_each_entry_safe(pos, tmp, &topology_state->vcpis, next) {
> > + if (pos->port == port) {
> > + list_del(&pos->next);
> > + DRM_DEBUG_KMS("[MST PORT:%p] vcpi %d -> 0\n",
> > + port, pos->vcpi);
> >
> > + kfree(pos);
> > + return 0;
> > + }
> > + }
> > +
> > + /* If no allocation was found, all that means is that the port was
> > + * destroyed since the last atomic commit. That's OK!
> > + */
> > return 0;
> > }
> > EXPORT_SYMBOL(drm_dp_atomic_release_vcpi_slots);
> > @@ -3112,15 +3159,50 @@ static void drm_dp_destroy_connector_work(struct work_struct *work)
> > static struct drm_private_state *
> > drm_dp_mst_duplicate_state(struct drm_private_obj *obj)
> > {
> > - struct drm_dp_mst_topology_state *state;
> > + struct drm_dp_mst_topology_state *state, *old_state =
> > + to_dp_mst_topology_state(obj->state);
> > + struct drm_dp_mst_topology_mgr *mgr = old_state->mgr;
> > + struct drm_dp_mst_port *port;
> > + struct drm_dp_vcpi_allocation *pos, *vcpi;
> >
> > - state = kmemdup(obj->state, sizeof(*state), GFP_KERNEL);
> > + state = kzalloc(sizeof(*state), GFP_KERNEL);
> > if (!state)
> > return NULL;
> >
> > __drm_atomic_helper_private_obj_duplicate_state(obj, &state->base);
> >
> > + state->mgr = mgr;
> > + INIT_LIST_HEAD(&state->vcpis);
> > +
> > + /* Copy over the VCPI allocations for ports that still exist */
> > + list_for_each_entry(pos, &old_state->vcpis, next) {
> > + port = drm_dp_get_validated_port_ref(mgr, pos->port);
> > + if (!port) {
> > + DRM_DEBUG_ATOMIC("[MST PORT:%p] is gone, vcpi %d -> 0\n",
> > + pos->port, pos->vcpi);
> > + continue;
> > + }
>
> Hm, in general we have 0 validation in the duplicate_state functions, Just
> dump copying, with minimal pointer/refcount updates.
> > +
> > + vcpi = kzalloc(sizeof(*vcpi), GFP_KERNEL);
>
> kmemdump + updating the list. Just in case we ever add more stuff here.
> Probably needs an explicit memset to avoid list debugging.
>
> At least that's the style we use everywhere else.
>
> > + if (!vcpi) {
> > + drm_dp_put_port(port);
> > + goto fail_alloc;
> > + }
> > +
> > + vcpi->port = port;
> > + vcpi->vcpi = pos->vcpi;
> > + list_add(&vcpi->next, &state->vcpis);
> > + drm_dp_put_port(port);
> > + }
> > +
> > return &state->base;
> > +
> > +fail_alloc:
> > + list_for_each_entry_safe(pos, vcpi, &state->vcpis, next)
> > + kfree(pos);
> > + kfree(state);
> > +
> > + return NULL;
> > }
> >
> > static void drm_dp_mst_destroy_state(struct drm_private_obj *obj,
> > @@ -3128,14 +3210,60 @@ static void drm_dp_mst_destroy_state(struct drm_private_obj *obj,
> > {
> > struct drm_dp_mst_topology_state *mst_state =
> > to_dp_mst_topology_state(state);
> > + struct drm_dp_vcpi_allocation *pos, *tmp;
> > +
> WARN_ON(!list_empty());
>
> I think this not being empty on mgr teardown would be a driver bug. All
> the ports/branch devices should be gone by now, and also all the modeset
> state should have been torn down, using drm_atomic_helper_shutdown() or
> friends.
Lyude pointed out on irc that I was not yet sufficiently coffee'd up when
making this comment: The destroy_state callback needs to clean up without
complaining ofc! I mixed it up somehow with drm_mode_config_cleanup,
where the WARN_ON would have been a good idea.
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
