Quoting Srivatsa, Anusha (2017-09-08 18:58:11) > > > >-----Original Message----- > >From: Chris Wilson [mailto:ch...@chris-wilson.co.uk] > >Sent: Thursday, September 7, 2017 3:08 PM > >To: Tvrtko Ursulin <tvrtko.ursu...@linux.intel.com>; Srivatsa, Anusha > ><anusha.sriva...@intel.com>; intel-gfx@lists.freedesktop.org > >Cc: Sundaresan, Sujaritha <sujaritha.sundare...@intel.com> > >Subject: Re: [Intel-gfx] [PATCH 1/2] drm/i915/guc: Add GuC Load time to > >debugfs > > > >Quoting Tvrtko Ursulin (2017-09-07 09:49:02) > >> > >> On 07/09/2017 01:37, Anusha Srivatsa wrote: > >> > Calculate the time that GuC takes to load. > >> > This information could be very useful in determining if GuC is > >> > taking unreasonably long time to load in a certain platforms. > >> > >> Do we need this in debugfs or a DRM_NOTE or something would be > >> sufficient if the load time is above certain threshold? > >> > >> Also, what are the typical times here? Are jiffies precise enough? > >> Could be only 10ms granularity on some kernels. > > > >Hear, hear. Aiming for jiffie resolution sets a very low bar. > > > >Why is this important anyway? It shouldn't be on the critical path for > >bringing up > >userspace (i.e. on the synchronous portion of boot/module loading). Nor > >should > >it be delaying bringing up a screen, once the core is initialised we should > >be able > >to bring up GT and the display in parallel. > > Hi Chris, > Wont it still be of some value to know the time taken to load the firmware in > different platforms? To have some data to tell how much time it takes to load?
If it always reports 0 jiffies that is not much use. Why do you need to add code to do this since you can just log the function call timings? -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
