On Thu, Feb 23, 2012 at 12:15 PM, Linus Torvalds torvalds@linux-foundation.org wrote:
Sadly, this doesn't seem to make any difference to my case. My xrandr stays at 0.555s even with this patch.
Btw, profiling with call chains seems to say that it all comes from intel_sdvo_get_analog_edid() (about half from intel_sdvo_get_modes() and half from intel_sdvo_tmds_sink_detect()).
All called through drm_ioctl() -> drm_mode_getconnector() -> drm_helper_probe_single_connector_modes().
Which I guess isn't anything interesting, but that intel_sdvo_get_analog_edid() thing seems to be very much a Mac mini hack. There's a comment about that in the sources too:
/* * Mac mini hack. On this device, the DVI-I connector shares one DDC * link between analog and digital outputs. So, if the regular SDVO * DDC fails, check to see if the analog output is disconnected, in * which case we'll look there for the digital DDC data. */
and maybe that mac mini hack ends up interacting badly with something else? I'll happily test patches if people have any ideas.
Linus