In D21567#469977, @rscheff_gmx.at wrote:

Checked this patch with various "pure FIN after hole" and "data FIN after hole" combinations; the FIN bit becomes part of the SACK (and DSACK) blocks, as it should be when considering that FIN is part of the sequence number space.

The patch has been updated to address Richard's comments.

Address Richard's comments.

In D21567#469962, @rrs wrote:

I would like to see the Rack version of the code be identical to the tcp_input version.. i.e. it should check

if ((tlen > 0) && (tp->t_flags & TF_SACK_PERMITED))

The soon to be committed version of rack will allow a user to use the stack without sack.. so it is best to
put this in place now.

Add a check to the RACK code whether the SACK extension is enabled. This was requested by rrs@.

Is it possible to get this in head? It seems to be needed to run FreeBSD on Ampere systems...

In D21314#464437, @emaste wrote:

I tried a similar version of this change and my eMAG reports CPU 0: APM eMAG 8180 r3p2 affinity: 0 0

I used 'ampere' in the struct name and comments though.

I think not reporting APM and Ampere improves consistency. So I changed that. However, I left the apm in the structure names since it corresponds with the APM in the constant for the implementer constant and the implementer name. However, I added a comment to provide a hint that these CPUs are now managed by Ampere Computing.

Don't report APM and Ampere to be consistent with other entries. Also add comments to note that the CPU comes from Ampere even though using constants from Applied Micro.

In D20549#457756, @rscheff_gmx.at wrote:

Addressed all comments so far.

Question around the default for alpha still remains; the current default renders dctcp flows very unreactive on CE marks during the initial few windows. Linux is using MAX_ALPHA_VALUE as the inital, which IMHO would be the conservative approach...

Add Benedict for man page change.

Please update also share/man/man4/cc_dctcp.4 to describe the valid ranges of the sysctl variables and the defaults. Please fix inconsistencies like the default for dctcp_shift_g given there and in the code.

The patch doesn't apply, since the indentation is based on spaces instead of tabs. Please fix.