On 12/17/14 1:12 PM, Daniel Vetter wrote:
On Tue, Dec 16, 2014 at 05:00:38PM -0200, Paulo Zanoni wrote:
2014-12-10 21:53 GMT-02:00 Todd Previte<tprevite@xxxxxxxxx>:
+ if (intel_encoder->type == INTEL_OUTPUT_DISPLAYPORT) {
+ if (intel_encoder->new_crtc) {
+ crtc_config = &intel_encoder->new_crtc->config;
+ pipe_bpp = crtc_config->pipe_bpp;
+ mode = &crtc_config->adjusted_mode;
+ hres = mode->hdisplay;
+ vres = mode->vdisplay;
+ } else if (intel_encoder->base.crtc) {
+ icrtc = to_intel_crtc(intel_encoder->base.crtc);
+ pipe_bpp = icrtc->config.pipe_bpp;
+ mode = &icrtc->config.adjusted_mode;
+ hres = mode->hdisplay;
+ vres = mode->vdisplay;
+ } else {
+ pipe_bpp = 0;
+ hres = vres = 0;
+ }
Why do you have this new_crtc vs current_crtc vs no crtc distinction
here? Is it really necessary? Shouldn't the "DP testing debugfs
protocol" establish when exactly the information should be queried,
and then give some errors in case information is being requested at
the wrong time?
Presuming adequate locking exists (haven't checked tbh) new_* pointers always
match the equivalent base pointers. new_ pointers/structures are
exclusively for modeset code (and specifically state computation). I guess
the new_crtc case needs to be dropped here.
-Daniel
I tested this again with the new_crtc case removed and it appears to
work correctly. So the new_crtc case has been removed. The changes have
been integrated into the monster patch.
-T
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
