Page MenuHomeFreeBSD

rrs (Randall Stewart)
User

Projects

User Details

User Since
Jan 22 2015, 5:22 AM (333 w, 4 d)

Recent Activity

Fri, Jun 11

rrs accepted D30723: Remove debug output in RACK.
Fri, Jun 11, 6:13 PM
rrs committed R10:ba1b3e48f5be: tcp: Missing mfree in rack and bbr (authored by rrs).
tcp: Missing mfree in rack and bbr
Fri, Jun 11, 3:41 PM
rrs added a reviewer for D30727: tcp: Missing mfree in rack and bbr.: tuexen.
Fri, Jun 11, 3:05 PM
rrs requested review of D30727: tcp: Missing mfree in rack and bbr..
Fri, Jun 11, 3:05 PM
rrs added a comment to D30721: New Reno fixes.

One question, were you going to kill the printf?

Fri, Jun 11, 12:30 PM
rrs accepted D30721: New Reno fixes.
Fri, Jun 11, 12:26 PM
rrs added a comment to D30721: New Reno fixes.

And I think Richard is right here, the conditions on setting beta_ecn should be removed as well.

Fri, Jun 11, 10:38 AM
rrs requested changes to D30721: New Reno fixes.

The more I think about it.. this is incorrect.

Fri, Jun 11, 10:33 AM
rrs accepted D30721: New Reno fixes.
Fri, Jun 11, 10:18 AM

Thu, Jun 10

rrs committed R10:67e892819b26: tcp: Mbuf leak while holding a socket buffer lock. (authored by rrs).
tcp: Mbuf leak while holding a socket buffer lock.
Thu, Jun 10, 12:36 PM
rrs added a comment to D30704: tcp: Mbuf leak while holding a socket buffer lock .

Yes the pure whitespace changes are intended..

Thu, Jun 10, 12:30 PM

Wed, Jun 9

rrs updated the diff for D30704: tcp: Mbuf leak while holding a socket buffer lock .

Oh I almost forgot the other change here, i.e. the centralization
of counting bytes sent and bytes retransmitted. This is an essential
thing for hardware TLS. That way hardware TLS can have a place
they can notice a connection is retransmitting a large amount
and "remove" that connection from using hardware and instead
use software.

Wed, Jun 9, 6:18 PM
rrs added reviewers for D30704: tcp: Mbuf leak while holding a socket buffer lock : tuexen, rscheff.
Wed, Jun 9, 6:09 PM
rrs requested review of D30704: tcp: Mbuf leak while holding a socket buffer lock .
Wed, Jun 9, 6:09 PM
rrs committed R10:b45daaea95ab: tcp: LRO timestamps have lost their previous precision (authored by rrs).
tcp: LRO timestamps have lost their previous precision
Wed, Jun 9, 6:01 PM
rrs closed D30695: tcp: LRO timestamps have lost their previous precision.
Wed, Jun 9, 6:01 PM
rrs updated the diff for D30695: tcp: LRO timestamps have lost their previous precision.

This addresses Hans concern about comparing a bintime to a sbtime by just
using the bintime and converting things to pure nanoseconds for comparisions.

Wed, Jun 9, 10:59 AM

Tue, Jun 8

rrs added a comment to D30695: tcp: LRO timestamps have lost their previous precision.

Sure I can do that, I do think its safe, but will cause some inaccuracy.. let me go work on converting the other to a bintime too.

Tue, Jun 8, 7:48 PM
rrs added a comment to D30695: tcp: LRO timestamps have lost their previous precision.

I hesitated to change the flushing time to use bintime, just because it was easier not
too and in reality few drivers use this, and those that do I think less precision
is required.

Tue, Jun 8, 4:23 PM
rrs added a comment to D30695: tcp: LRO timestamps have lost their previous precision.

I tried that at First but Lawrence complained since the whole timing
structure of the sbintime is also yet another abbreviation.. so it changes
the granularity of the time (at least according to Lawrence). I did poke
into the code a bit and looked and yes, he was right .. so thats when
I elected to go with bintime.

Tue, Jun 8, 3:53 PM
rrs added reviewers for D30695: tcp: LRO timestamps have lost their previous precision: hselasky, gallatin, tuexen.
Tue, Jun 8, 2:32 PM
rrs requested review of D30695: tcp: LRO timestamps have lost their previous precision.
Tue, Jun 8, 2:32 PM
rrs closed D30627: tcp: a more proper fix for the previous ack-wark.
Tue, Jun 8, 2:30 PM

Fri, Jun 4

rrs committed R10:4747500deaaa: tcp: A better fix for the previously attempted fix of the ack-war issue with… (authored by rrs).
tcp: A better fix for the previously attempted fix of the ack-war issue with…
Fri, Jun 4, 9:29 AM

Thu, Jun 3

rrs added a reviewer for D30627: tcp: a more proper fix for the previous ack-wark: tuexen.
Thu, Jun 3, 3:24 PM
rrs requested review of D30627: tcp: a more proper fix for the previous ack-wark.
Thu, Jun 3, 3:24 PM

Tue, Jun 1

rrs closed D30497: tcp: When we have an out-of-order FIN we do want to strip off the FIN bit..
Tue, Jun 1, 12:00 PM
rrs accepted D29509: Remove unused write only "m_prev_last" field in TCP LRO code..
Tue, Jun 1, 10:45 AM

Thu, May 27

rrs committed R10:8c69d988a8d3: tcp: When we have an out-of-order FIN we do want to strip off the FIN bit. (authored by rrs).
tcp: When we have an out-of-order FIN we do want to strip off the FIN bit.
Thu, May 27, 2:52 PM
rrs added a reviewer for D30497: tcp: When we have an out-of-order FIN we do want to strip off the FIN bit.: tuexen.
Thu, May 27, 12:13 PM
rrs requested review of D30497: tcp: When we have an out-of-order FIN we do want to strip off the FIN bit..
Thu, May 27, 12:13 PM

Wed, May 26

rrs committed R10:4f3addd94be5: tcp: Add a socket option to rack so we can test various changes to the slop… (authored by rrs).
tcp: Add a socket option to rack so we can test various changes to the slop…
Wed, May 26, 10:46 AM
rrs closed D30249: Create a way to measure the impact of changing timer_slop.
Wed, May 26, 10:46 AM
rrs closed D30451: Bugs and problems even in the base stack..
Wed, May 26, 10:38 AM
rrs accepted D30460: cxgbe: set iq flags only after it has been initialized.
Wed, May 26, 10:23 AM

Tue, May 25

rrs accepted D30407: tcp: Need to take network epoch when calling tfb_tcp_fb_fini.
Tue, May 25, 5:29 PM
rrs committed R10:13c0e198ca27: tcp: Fix bugs related to the PUSH bit and rack and an ack war (authored by rrs).
tcp: Fix bugs related to the PUSH bit and rack and an ack war
Tue, May 25, 5:26 PM
rrs requested review of D30451: Bugs and problems even in the base stack..
Tue, May 25, 12:07 PM

Mon, May 24

rrs closed D30413: clone rsm needs to do the right thing with the PUSH bit.
Mon, May 24, 6:46 PM
rrs committed R10:631449d5d035: tcp: Fix an issue with the PUSH bit as well as fill in the missing mtu change… (authored by rrs).
tcp: Fix an issue with the PUSH bit as well as fill in the missing mtu change…
Mon, May 24, 6:45 PM
rrs updated the diff for D30413: clone rsm needs to do the right thing with the PUSH bit.

We are missing the mtu change call too which is why it was not acting correctly

Mon, May 24, 6:40 PM
rrs requested review of D30413: clone rsm needs to do the right thing with the PUSH bit.
Mon, May 24, 1:42 PM
rrs accepted D30249: Create a way to measure the impact of changing timer_slop.
Mon, May 24, 1:29 PM
rrs accepted D30402: Fix a RACK socket buffer lock issue.
Mon, May 24, 1:05 PM

Wed, May 19

rrs updated the diff for D30249: Create a way to measure the impact of changing timer_slop.

It makes it easier for debugging if we remove the dup rtt log in TCP_LOG_RTT and instead
put the actual slop being used by the stack. This way we can validate in the BB log
that the slop was set properly via the socket option.

Wed, May 19, 2:00 PM
rrs added a comment to D30249: Create a way to measure the impact of changing timer_slop.

These two pkt drill scripts can be used to validate that the feature is working

Wed, May 19, 12:56 PM

Tue, May 18

rrs updated the diff for D30249: Create a way to measure the impact of changing timer_slop.

I realized when we set the socket option, we also need to set the rxt timer
with the new slop if an srtt has been collected.

Tue, May 18, 4:50 PM

May 13 2021

rrs requested review of D30249: Create a way to measure the impact of changing timer_slop.
May 13 2021, 5:02 PM
rrs closed D30241: syzcaller found an interesting Invariant crash.
May 13 2021, 11:40 AM
rrs committed R10:02cffbc2507e: tcp: Incorrect KASSERT causes a panic in rack (authored by rrs).
tcp: Incorrect KASSERT causes a panic in rack
May 13 2021, 11:40 AM
rrs added a reviewer for D30241: syzcaller found an interesting Invariant crash: tuexen.
May 13 2021, 10:58 AM
rrs requested review of D30241: syzcaller found an interesting Invariant crash.
May 13 2021, 10:58 AM

May 12 2021

rrs accepted D30230: Fix retransmission timer initialisation for TCP RACK.
May 12 2021, 4:01 PM
rrs added inline comments to D30230: Fix retransmission timer initialisation for TCP RACK.
May 12 2021, 2:42 PM
rrs added inline comments to D30230: Fix retransmission timer initialisation for TCP RACK.
May 12 2021, 2:08 PM

May 11 2021

rrs closed D30213: Rack conversion in host cache needs to be conditional on the host cache being used..
May 11 2021, 12:20 PM
rrs committed R10:4b86a24a76a4: tcp: In rack, we must only convert restored rtt when the hostcache does restore… (authored by rrs).
tcp: In rack, we must only convert restored rtt when the hostcache does restore…
May 11 2021, 12:19 PM
rrs requested review of D30213: Rack conversion in host cache needs to be conditional on the host cache being used..
May 11 2021, 11:31 AM

May 10 2021

rrs accepted D28931: Basic Lost Retransmission Detection (LRD).
May 10 2021, 3:46 PM
rrs committed R10:9867224bab3f: tcp:Host cache and rack ending up with incorrect values. (authored by rrs).
tcp:Host cache and rack ending up with incorrect values.
May 10 2021, 3:32 PM
rrs closed D30172: Initial retransmit timeout improperly set.
May 10 2021, 3:31 PM
rrs updated the test plan for D30172: Initial retransmit timeout improperly set.
May 10 2021, 3:22 PM
rrs updated the diff for D30172: Initial retransmit timeout improperly set.

Address Michaels comments and further do some code reducing (create a conversion
function that all places use to get srtt/rttvar in rack format).

May 10 2021, 3:18 PM
rrs added inline comments to D30172: Initial retransmit timeout improperly set.
May 10 2021, 3:07 PM
rrs updated the diff for D30172: Initial retransmit timeout improperly set.

Turns out the problem is far deeper. There are at least
a couple of interactions here.

May 10 2021, 2:44 PM

May 8 2021

rrs requested review of D30172: Initial retransmit timeout improperly set.
May 8 2021, 8:42 AM

May 7 2021

rrs closed D30163: Ratelimit and rack.
May 7 2021, 9:35 PM
rrs committed R10:5a4333a5378f: This takes Warners suggested approach to making it so that (authored by rrs).
This takes Warners suggested approach to making it so that
May 7 2021, 9:35 PM
rrs added a comment to D30163: Ratelimit and rack.

Define a kernel that has all the TCP options to get the alternate
stacks but don't add RATELIMIT.

May 7 2021, 8:05 PM
rrs updated the diff for D30163: Ratelimit and rack.

Turns out I was wrong, we need both functions not just one.

May 7 2021, 6:33 PM
rrs updated the diff for D30163: Ratelimit and rack.

Lets get the right function here

May 7 2021, 6:29 PM
rrs updated the diff for D30163: Ratelimit and rack.

It needs to be static and match the signature duhh

May 7 2021, 6:27 PM
rrs updated the diff for D30163: Ratelimit and rack.

Lets take Warners alternative.. i.e. have the tcp_ratelimit code return 0 for the highest hw rate when
ratelimit is not in the kernel.

May 7 2021, 6:18 PM
rrs committed R10:a16cee021865: Fix a UDP tunneling issue with rack. Basically there are two (authored by rrs).
Fix a UDP tunneling issue with rack. Basically there are two
May 7 2021, 6:12 PM
rrs closed D30157: UDP tunneling issue in rack.
May 7 2021, 6:12 PM
rrs added a comment to D30163: Ratelimit and rack.

Ok thats another option too ;)

May 7 2021, 6:09 PM
rrs updated the diff for D30157: UDP tunneling issue in rack.

Michael try this with UDP port overbooking as we discussed.

May 7 2021, 5:22 PM
rrs requested review of D30163: Ratelimit and rack.
May 7 2021, 4:54 PM
rrs added a comment to D30157: UDP tunneling issue in rack.

Cool lets definitely figure out what is wrong besides what Peter found here

May 7 2021, 2:55 PM
rrs closed D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 7 2021, 2:54 PM
rrs requested review of D30157: UDP tunneling issue in rack.
May 7 2021, 8:49 AM

May 6 2021

rrs committed R10:5d8fd932e418: This brings into sync FreeBSD with the netflix versions of rack and bbr. (authored by rrs).
This brings into sync FreeBSD with the netflix versions of rack and bbr.
May 6 2021, 3:27 PM
rrs accepted D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 6 2021, 3:21 PM
rrs updated the diff for D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..

One last comment from Michael Tuexen, use %zu in printf.

May 6 2021, 3:21 PM

May 5 2021

rrs added inline comments to D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 5 2021, 6:35 PM
rrs accepted D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 5 2021, 5:50 PM
rrs updated the diff for D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..

This addresses Richards comments about mss = tcp_maxseg(). Which actually itself is
incorrect. Instead we need the fixed max segment size i.e. not including all the
dynamic things like sack since we are setting a cwnd that will survive long after
we are out of recovery.

May 5 2021, 12:26 PM
rrs added inline comments to D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 5 2021, 11:23 AM
rrs added inline comments to D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 5 2021, 11:11 AM

May 3 2021

rrs added inline comments to D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 3 2021, 7:06 PM
rrs updated the diff for D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..

gesh get the comments too silly

May 3 2021, 7:04 PM
rrs updated the diff for D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 3 2021, 6:59 PM
rrs added inline comments to D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 3 2021, 6:59 PM
rrs updated the diff for D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..

Address some of Richards comments

May 3 2021, 6:52 PM
rrs added inline comments to D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 3 2021, 6:51 PM
rrs added inline comments to D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..
May 3 2021, 4:19 PM

Apr 30 2021

rrs requested changes to D30043: tcp: Whitespace cleanup in brr and rack.
Apr 30 2021, 6:25 PM
rrs added a comment to D30043: tcp: Whitespace cleanup in brr and rack.

Please wait. I have a major set of changes coming in and this is old broken code you are playing with. I am sure
a lot of the spacings are taken care of . Please wait until D30036 lands before you do any cleanups.

Apr 30 2021, 6:25 PM
rrs abandoned D29530: Missing NULL checks for tcp_lro with compressed ack's.
Apr 30 2021, 6:23 PM
rrs added a comment to D30036: Update Rack and BBR to the latest from NF fixing LRO issues along the way..

and yes Oliver this is because when the LRO code went in I forgot to update rack_bbr_common.c. I am glad
it fixes the issue. Neither Michael or I could recreate it since we were running on VM's and we did not
realize vmware fusion does not get LRO enabled .. Drew has given me a patch for that so I should put
it in... at least on my local box ;)

Apr 30 2021, 12:49 PM