On Tue, Jan 30, 2018 at 11:57:49AM +0000, Chris Wilson wrote: > Quoting Imre Deak (2018-01-30 11:47:11) > > These two functions are very similar so simplify things by removing the > > duplication. > > > > Add a seperate sleeping poll timeout parameter, useful for longer polls > > like the CDCLK change on BXT/GLK. The next patch will take that into use. > > > > While at it document snb_pcode_request() and clean up a bit the > > error/debug prints. Other than that no functional changes. > > In my patches to do the same (and move it to intel_sideband.c) I kept > the sandybridge_pcode_read/sandybridge_pcode_write functions to both > take the sb_lock and to provide imo clearer debug messages. > https://patchwork.freedesktop.org/series/36469/ Ah didn't notice it, will drop mine. Does it make sense to pass the fast/slow timeouts to sandybridge_pcode_read/write? Imo it'd document things better and could avoid the long atomic poll on BXT/GLK. --Imre _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
