linux/drivers/gpu/drm/drm_fops.c, branch v4.5

Back merge tag 'v4.4-rc4' into drm-next

2015-12-08T01:04:26Z

We've picked up a few conflicts and it would be nice to resolve them before we move onwards.

drm: Fix an unwanted master inheritance v2

2015-12-04T02:28:14Z

A client calling drmSetMaster() using a file descriptor that was opened when another client was master would inherit the latter client's master object and all its authenticated clients. This is unwanted behaviour, and when this happens, instead allocate a brand new master object for the client calling drmSetMaster(). Fixes a BUG() throw in vmw_master_set(). Cc: Signed-off-by: Thomas Hellstrom Signed-off-by: Dave Airlie

drm: Serialise multiple event readers

2015-11-26T14:21:27Z

The previous patch reintroduced a race condition whereby a failure in one reader may allow a second reader to see out-of-order events. Introduce a mutex to serialise readers so that an event is completed in its entirety before another reader may process an event. The two readers may race against each other, but the events each retrieves are in the correct order. Signed-off-by: Chris Wilson Cc: Thomas Hellstrom Cc: Takashi Iwai Cc: Daniel Vetter Link: http://patchwork.freedesktop.org/patch/msgid/1448462343-2072-2-git-send-email-chris@chris-wilson.co.uk Reviewed-by: Thomas Hellstrom Signed-off-by: Daniel Vetter

drm: Drop dev->event_lock spinlock around faulting copy_to_user()

2015-11-26T14:21:06Z

In commit cdd1cf799bd24ac0a4184549601ae302267301c5 Author: Chris Wilson Date: Thu Dec 4 21:03:25 2014 +0000 drm: Make drm_read() more robust against multithreaded races I fixed the races by serialising the use of the event by extending the dev->event_lock. However, as Thomas pointed out, the copy_to_user() may fault (even the __copy_to_user_inatomic() variant used here) and calling into the driver backend with the spinlock held is bad news. Therefore we have to drop the spinlock before the copy, but that exposes us to the old race whereby a second reader could see an out-of-order event (as the first reader may claim the first request but fail to copy it back to userspace and so on returning it to the event list it will be behind the current event being copied by the second reader). Reported-by: Thomas Hellstrom Signed-off-by: Chris Wilson Cc: Thomas Hellstrom Cc: Takashi Iwai Cc: Daniel Vetter Link: http://patchwork.freedesktop.org/patch/msgid/1448462343-2072-1-git-send-email-chris@chris-wilson.co.uk Reviewed-by: Thomas Hellstrom Signed-off-by: Daniel Vetter

drm/mode: Add user blob-creation ioctl

2015-05-22T14:18:28Z

Add an ioctl which allows users to create blob properties from supplied data. Currently this only supports modes, creating a drm_display_mode from the userspace drm_mode_modeinfo. v2: Removed size/type checks. Rebased on new patches to allow error propagation from create_blob, as well as avoiding double-allocation. Signed-off-by: Daniel Stone Reviewed-by: Maarten Lankhorst Tested-by: Sean Paul Signed-off-by: Daniel Vetter

drm: simplify authentication management

2015-05-05T07:45:57Z

The magic auth tokens we have are a simple map from cyclic IDs to drm_file objects. Remove all the old bulk of code and replace it with a simple, direct IDR. The previous behavior is kept. Especially calling authmagic multiple times on the same magic results in EINVAL except on the first call. The only difference in behavior is that we never allocate IDs multiple times as long as a client has its FD open. v2: - Fix return code of GetMagic() - Use non-cyclic IDR allocator - fix off-by-one in "magic > INT_MAX" sanity check v3: - drop redundant "magic > INT_MAX" check Signed-off-by: David Herrmann Reviewed-by: Chris Wilson Signed-off-by: Daniel Vetter

drm: Make drm_read() more robust against multithreaded races

2015-01-21T13:57:03Z

The current implementation of drm_read() faces a number of issues: 1. Upon an error, it consumes the event which may lead to the client blocking. 2. Upon an error, it forgets about events already copied 3. If it fails to copy a single event with O_NONBLOCK it falls into a infinite loop of reporting EAGAIN. 3. There is a race between multiple waiters and blocking reads of the events list. Here, we inline drm_dequeue_event() into drm_read() so that we can take the spinlock around the list walking and event copying, and importantly reorder the error handling to avoid the issues above. Cc: Takashi Iwai Signed-off-by: Chris Wilson Reviewed-by: Takashi Iwai Testcase: igt/drm_read Signed-off-by: Daniel Vetter

drm: Fix memory leak at error path of drm_read()

2014-12-04T12:28:42Z

Note that the read manpages explicitly states that the read position is undefined on error. Since EFAULT is just a userspace bug we are therefore fine with just dropping the event on the floor. Signed-off-by: Takashi Iwai [danvet: Add note that just dropping the event is ok.] Signed-off-by: Daniel Vetter

drm: Implement O_NONBLOCK support on /dev/dri/cardN

2014-10-08T13:07:07Z

The implmentation is simple in the extreme: we only want to wait for events if the device was opened in blocking mode, otherwise we grab what is available and report an error if there was none. Signed-off-by: Chris Wilson Cc: dri-devel@lists.freedesktop.org Reviewed-by: Jesse Barnes Testcase: igt/kms_flip/nonblocing_read Signed-off-by: Daniel Vetter

drm/core: use helper to check driver features

2014-10-03T08:38:56Z

The patch replaces direct access to driver_features field by calls to helper function. Signed-off-by: Andrzej Hajda Reviewed-by: David Herrmann Signed-off-by: Daniel Vetter