This change builds on top if_index globalization (91f44749c6feb50f39af8805dd803e860f0418f1) which I strongly objected to, and which glebius agreed to back out as outlined in https://github.com/glebius/FreeBSD/commits/backout-ifindex, but that never happened. Hence, pls. don't proceed with this until if_index is reverted back to per-VNET state.

In D29821#677630, @zarychtam_plan-b.pwste.edu.pl wrote:

function dxr_build with caller dxr_change_rib_batch seems to be still active

In D29821#676285, @zarychtam_plan-b.pwste.edu.pl wrote:

...

I assume that it's another independent routing algo module and should be considered as full replacement, not any supplementation.

Simplify injection of artificial dependencies between successive lookups in an attempt to fool / defeat apparently nifty and mighty Ryzen's value predictors.

Switch to the new rib_walk_from KPI.

Update.

In D29588#667618, @melifaro wrote:

Do not update size of route change queue on failure.

In D29588#667323, @melifaro wrote:

Fix case with small number of updates.

In D29661#665092, @donner wrote:

May you please provide a full context diff?
I can't see, why the compiler issued a warning and how the fix will break other things (i.e. logic errors)

Expand the diff to -U30 per request from donner@.