On Wed, Jan 29, 2014 at 03:01:47PM +0100, David Herrmann wrote: > Hi > > I was looking into our minor-allocation code and it has one major drawback: > char-dev minor-IDs are unreliable. The <id>+128 hacks we use in user-space to > calculate render-node IDs based on the original card just does not work. > > Instead of allocating dummy IDs for each driver, I went ahead and tried to fix > it properly. So this series changes the minor-ID management to just allocate a > single dev->minor_base ID instead of one ID per "drm_minor". Now we can use > this > base to calculate the correct offset minor-ID for each existing DRM-minor. > > While at it, I introduced drm-refcounts to make minor-lookup independent of > drm_global_mutex. This is still not finished (and dev->open_count still > exists) > but I already have the next patches waiting here. > > Comments welcome! > > Branch is also available here: > http://cgit.freedesktop.org/~dvdhrm/linux/log/?h=minor > If someone could pull this into their tree and push to Fengguang's > test-framework, I'd appreciate it a lot! I'm still waiting for a reply from > him.
The series: Tested-by: Thierry Reding <treding at nvidia.com> -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 836 bytes Desc: not available URL: <http://lists.freedesktop.org/archives/dri-devel/attachments/20140221/fa247aa0/attachment-0001.pgp>
