Hi!
Upgraded to 6.42 rc 43 and since then see that fasttrack is not operational.

Relevant setup

There's a bridge with HW acceleration, and 2 ports added to the bridge.
On firewall have 2 rules for in-interface=bridge out-interface=bridge action=fasttrack and action=accept

Worked nicely before, now on transfer between the ports, CPU load goes 100% on 50% networking and 50% firewall and transfer is about 200 mbit instead of the usual 1 gbit
device: 2011UAS-2HnD

I'm having some sort of similar scenario on my RB750Gr3, after the same RC update I get some mixed bag of performance, despite "IP -> firewall -> Connections" show my IP sessions with the fasttrack flag, I can only saturate 70% of my gigabit pppoe line, before it was saturating just fine at over 90%. The profiler shows the firewall process eating up the resources.
I disabled all firewall and routing rules except the barebone fasttrack rules and the problem is still there. I get 100% load on the second core(both threads), while the first core is at 30% load on the first thread, the second thread being inutilized.
I hope the next RC will fix this issue with fasttrack. Before this update, fasttrack worked without issues, both cores had 40-50% load while using the full bandwidth.

Edit: I fixed my fasttrack issue by disabling learning on the lan bridge on all ports. Something in the last update broke something, don't know, now I get the mac-address table in the bridge hosts section.

Issue also persists in 6.42 RC 49

*Update* - strange - downgrading setup to 6.41.2 keeps the problem. Quite sure didn't have the problem before.
Will HW reset and see if it's the config or something "leftover" as part of the downgrade

Sorry false alarm. Routing was setup wrong *Lame*