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@.

I've tested the driver on 11.4-STABLE, and it works reasonably well, given the limitations of the (experimental) FPGA design. Denis made a "make universe" run on CURRENT and I did on 12.1 and the current patch didn't introduce any regressions. Hence, I'm inclined to commit this to HEAD as-is, ideally with a few cosmetic / non-functional changes I suggested in most recent comments. I intend to put on record Reviewed by: zec, bz (source), rgrimes, bcr (manpage)? Any further objections from the reviewers?

rgrimes: me being essentially a manpage rookie as well, unfortunately I can't provide an advice to Denis on how proceed from here. None of the lines fill entirely or extend past 80 columns, and all sentences are terminated by line breaks. What else should we improve re. formatting? As non-native English speakers, we can also try to improve text composition and flow a bit, but I guess this is not what you had in mind?

In D1944#62124, @robak wrote:

I managed to crash freshly fetched and built 10-S today, with GENERIC+VIMAGE+RACCT while stopping a jail, attaching console screenshots.