On Mon, 2021-01-04 at 14:47 +0100, Toke Høiland-Jørgensen wrote: > Lorenzo Bianconi <lorenzo.bianconi@xxxxxxxxxx> writes: > > > > Lorenzo Bianconi <lorenzo@xxxxxxxxxx> writes: > > > > > > > Introduce aql_enable node in debugfs in order to enable/disable aql. > > > > This is useful for debugging purpose. > > > > > > Don't mind having a switch, although I wonder if it would be better to > > > overload the existing debugfs file (e.g., a threshold of 0 could disable > > > everything?) so as not to clutter up debugfs too much? > > > > > > -Toke > > > > > > > You mean to consider 0 as a special value to disable aql, right? I > > would prefer to have a dedicated switch for it since I guess it is > > clearer for users (but I can live with it :) ) > > Yeah, maybe a bit clearer but at the cost of clutter. I dunno, not a > strong preference either way; I guess Johannes can make the call :) I'm not sure I care about an extra debugfs file - but I do wonder about the extra check at runtime that would basically never be true since the default is enable ... Maybe that should use a static_branch() or something? johannes