<div dir="ltr"><div class="gmail_default" style="font-family:monospace,monospace">There isn't any dark plot behind it.</div><div class="gmail_default" style="font-family:monospace,monospace"><br></div><div class="gmail_default" style="font-family:monospace,monospace">For instance, in our use case, a DPU (Display Process Unit) which has a </div><div class="gmail_default" style="font-family:monospace,monospace">blit feature (using DRM_RENDER_ALLOW) can be implemented cleanly</div><div class="gmail_default" style="font-family:monospace,monospace">in a separate driver and not being dependent on the DRM core driver. If the blit</div><div class="gmail_default" style="font-family:monospace,monospace">feature is present/enabled, we can dynamically register the ioctls at run-time.</div><div class="gmail_default" style="font-family:monospace,monospace"><br></div><div class="gmail_default" style="font-family:monospace,monospace">There are other means to mitigate this, but we thought this would beneficial </div><div class="gmail_default" style="font-family:monospace,monospace">to other drivers as well.</div><div class="gmail_default" style="font-family:monospace,monospace"><br></div><div class="gmail_default" style="font-family:monospace,monospace">Other SoC drivers like Exynos (G2D) provide this feature by inventing it's own sub-driver</div><div class="gmail_default" style="font-family:monospace,monospace">system/layer and have all the sub-drivers built-in.</div><div class="gmail_default" style="font-family:monospace,monospace"><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Sep 4, 2017 at 6:25 PM, Daniel Vetter <span dir="ltr"><<a href="mailto:daniel@ffwll.ch" target="_blank">daniel@ffwll.ch</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Mon, Sep 04, 2017 at 06:16:41PM +0300, Marius Vlad wrote:<br> > From: Marius Vlad <<a href="mailto:marius.vlad0@gmail.com">marius.vlad0@gmail.com</a>><br> ><br> > Currently driver-specific ioctls have to be declared static and are confined to<br> > DRM core driver. This patch series provides the means to remove those constrains<br> > and allow to register driver-specific ioctls dynamically by keeping a list of<br> > registered ioctls in struct drm_driver, then each component of the driver can<br> > then register its own specific ioctls using this interface.<br> ><br> > The driver must assign ioctl_register/ioctl_<wbr>deregister in<br> > its drm_driver structure in order to make use of it.<br> ><br> > While SoC drivers benefit the most from this approach (by not polluting DRM core<br> > driver and allowing sub drivers to implement and register driver-specific<br> > ioctls dynamically), further patches shows how easy is to convert drm/i915 to<br> > this approach by registering GEM and perf ioctls separately.<br> <br> </span>What exactly is the problem you're trying to solve?<br> <br> This awefully smells like some neat way to make loading driver modules for<br> blob userspace easy ... And I can't think of any other thing you could use<br> this for.<br> <br> And even for the blob userspace use case: Create a separate drm driver<br> instance, share buffers and fences with dma_buf and dma_fence, and you're<br> all good. I really have no idea what this is good for, but maybe I'm<br> missing something?<br> -Daniel<br> <div class="HOEnZb"><div class="h5"><br> ><br> > Marius Vlad (4):<br> > drm/gpu: Support registering driver-specific ioctls dynamically<br> > drm/i915: Convert i915 to use ioctl_register/ioctl_<wbr>deregister.<br> > drm/i915: Register perf_ ioctls directly in i915_perf file.<br> > drm/i915: Register GEM ioctls directly in i915_gem file.<br> ><br> > drivers/gpu/drm/drm_drv.c | 1 +<br> > drivers/gpu/drm/drm_ioctl.c | 99 ++++++++++++++++++++++++++++++<wbr>++++--<br> > drivers/gpu/drm/i915/i915_drv.<wbr>c | 107 +++++++++++++++---------------<wbr>---------<br> > drivers/gpu/drm/i915/i915_gem.<wbr>c | 52 +++++++++++++++++++<br> > drivers/gpu/drm/i915/i915_<wbr>perf.c | 21 ++++++++<br> > include/drm/drm_drv.h | 34 +++++++++++++<br> > include/drm/drm_ioctl.h | 6 +++<br> > 7 files changed, 249 insertions(+), 71 deletions(-)<br> ><br> > --<br> > 2.9.3<br> ><br> <br> </div></div><span class="HOEnZb"><font color="#888888">--<br> Daniel Vetter<br> Software Engineer, Intel Corporation<br> <a href="http://blog.ffwll.ch" rel="noreferrer" target="_blank">http://blog.ffwll.ch</a><br> </font></span></blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><font face="monospace, monospace">Marius Vlad</font></div></div> </div>