It should probably check if the device is already added, instead of blindly re-adding. :-)
I understand the drmmode part now.
It goes wrong in 3 places at the moment:
* bugs in xorg-server
- Fix 1: drm_device_keep_trying.patch (pushed to debian git, proposed on mailing list)
- Fix 2: unbinding hybrid drivers doesn't work correctly at the moment, detaching is not done correctly if it has outputs and rendering (in my local tree, need to amend the correct patch)
- Fix 3: Ignore add events for devices we already have added, instead of removing/re-adding device.
* bugs in drivers:
- all except modesetting: drmmode_init installs some handlers, and nothing cleans them up
- all: the drm fd is never closed, so reopening a device will always fail.
Plus perhaps some more bugs in some places, but those were the obvious ones I found so far.
I fixed xserver-xorg-video-ati in upstream git, other drivers should be fixable in a similar way.
It should probably check if the device is already added, instead of blindly re-adding. :-)
I understand the drmmode part now.
It goes wrong in 3 places at the moment:
* bugs in xorg-server keep_trying. patch (pushed to debian git, proposed on mailing list)
- Fix 1: drm_device_
- Fix 2: unbinding hybrid drivers doesn't work correctly at the moment, detaching is not done correctly if it has outputs and rendering (in my local tree, need to amend the correct patch)
- Fix 3: Ignore add events for devices we already have added, instead of removing/re-adding device.
* bugs in drivers:
- all except modesetting: drmmode_init installs some handlers, and nothing cleans them up
- all: the drm fd is never closed, so reopening a device will always fail.
Plus perhaps some more bugs in some places, but those were the obvious ones I found so far.
I fixed xserver- xorg-video- ati in upstream git, other drivers should be fixable in a similar way.