On Wed, Nov 09, 2022 at 02:23:08PM -0800, Ben Gardon wrote: > On Mon, Nov 7, 2022 at 1:57 PM Oliver Upton <oliver.upton@xxxxxxxxx> wrote: > > > > Passing new arguments by value to the visitor callbacks is extremely > > inflexible for stuffing new parameters used by only some of the > > visitors. Use a context structure instead and pass the pointer through > > to the visitor callback. > > > > While at it, redefine the 'flags' parameter to the visitor to contain > > the bit indicating the phase of the walk. Pass the entire set of flags > > through the context structure such that the walker can communicate > > additional state to the visitor callback. > > > > No functional change intended. > > > > Signed-off-by: Oliver Upton <oliver.upton@xxxxxxxxx> > > This looks good to me. It's all fairly mechanical and I don't see any > problems. I was a little confused by the walk context flags passed via > visit, because they seem somewhat redundant if the leaf-ness can be > determined by looking at the PTE, but perhaps that's not always > possible. Some explanation is probably owed here. I think you caught the detail later on in the series, but I'm overloading flags to describe both the requested visits and some properties about the walk (i.e. a SHARED walk). I tried to leave it sufficiently generic as there will be other configuration bits we will want to stuff into a walker later on (such as TLBI and CMO elision). > Reviewed-by: Ben Gardon <bgardon@xxxxxxxxxx> Thanks! -- Best, Oliver
