From: Dave Airlie airlied@redhat.com
Since 9eb1e57f564d4e6e10991402726cc83fe0b9172f drm/dp/mst: make sure mst_primary mstb is valid in work function
we validate the mstb structs in the work function, and doing that takes a reference. So we should never get here with the work function running using the mstb device, only if the work function hasn't run yet or is running for another mstb.
So we don't need to sync the work here, this was causing lockdep spew when the device failed during the initial EDID read.
Signed-off-by: Dave Airlie airlied@redhat.com --- drivers/gpu/drm/drm_dp_mst_topology.c | 2 -- 1 file changed, 2 deletions(-)
diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c index e23df5f..16f911c 100644 --- a/drivers/gpu/drm/drm_dp_mst_topology.c +++ b/drivers/gpu/drm/drm_dp_mst_topology.c @@ -804,8 +804,6 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref) struct drm_dp_mst_port *port, *tmp; bool wake_tx = false;
- cancel_work_sync(&mstb->mgr->work); - /* * destroy all ports - don't need lock * as there are no more references to the mst branch
On Wed, Sep 30, 2015 at 10:39:42AM +1000, Dave Airlie wrote:
From: Dave Airlie airlied@redhat.com
Since 9eb1e57f564d4e6e10991402726cc83fe0b9172f drm/dp/mst: make sure mst_primary mstb is valid in work function
we validate the mstb structs in the work function, and doing that takes a reference. So we should never get here with the work function running using the mstb device, only if the work function hasn't run yet or is running for another mstb.
So we don't need to sync the work here, this was causing lockdep spew when the device failed during the initial EDID read.
can you pls paste the lockdep spew here? I think I see the locking recursion but would like to be sure.
Signed-off-by: Dave Airlie airlied@redhat.com
drivers/gpu/drm/drm_dp_mst_topology.c | 2 -- 1 file changed, 2 deletions(-)
diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c index e23df5f..16f911c 100644 --- a/drivers/gpu/drm/drm_dp_mst_topology.c +++ b/drivers/gpu/drm/drm_dp_mst_topology.c @@ -804,8 +804,6 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref) struct drm_dp_mst_port *port, *tmp; bool wake_tx = false;
- cancel_work_sync(&mstb->mgr->work);
Without this there's no place any more where we stop this worker on teardown. Generally we need to sync with outstanding work at least in system suspend and on driver unload. I think we need a flush_work (outside of the locked region ofc!) in drm_dp_mst_topology_mgr_suspend and drm_dp_mst_topology_mgr_destroy. Might be better as a follow-up patch, but should imo be part of this.
With lockdep splat + 2x flush_work this is Reviewed-by: Daniel Vetter daniel.vetter@ffwll.ch -Daniel
/* * destroy all ports - don't need lock * as there are no more references to the mst branch -- 2.4.3
dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
On Wed, Sep 30, 2015 at 09:20:47AM +0200, Daniel Vetter wrote:
On Wed, Sep 30, 2015 at 10:39:42AM +1000, Dave Airlie wrote:
From: Dave Airlie airlied@redhat.com
Since 9eb1e57f564d4e6e10991402726cc83fe0b9172f drm/dp/mst: make sure mst_primary mstb is valid in work function
we validate the mstb structs in the work function, and doing that takes a reference. So we should never get here with the work function running using the mstb device, only if the work function hasn't run yet or is running for another mstb.
So we don't need to sync the work here, this was causing lockdep spew when the device failed during the initial EDID read.
can you pls paste the lockdep spew here? I think I see the locking recursion but would like to be sure.
Signed-off-by: Dave Airlie airlied@redhat.com
drivers/gpu/drm/drm_dp_mst_topology.c | 2 -- 1 file changed, 2 deletions(-)
diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c index e23df5f..16f911c 100644 --- a/drivers/gpu/drm/drm_dp_mst_topology.c +++ b/drivers/gpu/drm/drm_dp_mst_topology.c @@ -804,8 +804,6 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref) struct drm_dp_mst_port *port, *tmp; bool wake_tx = false;
- cancel_work_sync(&mstb->mgr->work);
Without this there's no place any more where we stop this worker on teardown. Generally we need to sync with outstanding work at least in system suspend and on driver unload. I think we need a flush_work (outside of the locked region ofc!) in drm_dp_mst_topology_mgr_suspend and drm_dp_mst_topology_mgr_destroy. Might be better as a follow-up patch, but should imo be part of this.
With lockdep splat + 2x flush_work this is Reviewed-by: Daniel Vetter daniel.vetter@ffwll.ch
*1x flush_work - the one in drm_dp_mst_topology_mgr_destroy is already there, I must have been blind ... -Daniel
dri-devel@lists.freedesktop.org
-
Daniel Vetter -
Dave Airlie