Note that this sysctl controls the initial value of the per socket
reordering value. It _does_ increase automatically (assuming SACK is
enabled of course)

I believe enabling SACK is not a requirement. Even with plain Reno,
reordering  is tracked in tp->reordering and the reordering event is
counted in LINUX_MIB_TCPRENOREORDER.
tcp_add_reno_sack()-->tcp_check_reno_reordering() is the code path.

Thanks,
Vijay