Page MenuHomeFreeBSD

rscheff (Richard Scheffenegger)
User

Projects

User Details

User Since
Oct 18 2018, 9:44 PM (241 w, 59 m)

Recent Activity

Today

rscheff accepted D40185: cc_cubic: Use units of micro seconds (usecs) instead of ticks in rtt..

Not changing the generic rtt base stack unit / measuring, just improviing the granularity/precision of the calculations within Cubic. since this improves cubic performance, let's put this in prior 14 branching...

Thu, Jun 1, 3:30 PM

Apr 26 2023

rscheff accepted D39835: siftr: remove barely used hash generation per record.
Apr 26 2023, 6:11 PM

Apr 25 2023

rscheff accepted D39803: Change the unit of srtt and rto to usec, inspired by these in struct "tcp_info". Therefore, no need hz and tcp_rtt_scale in the headline of the log. Update the man page as well..
Apr 25 2023, 4:18 PM

Apr 24 2023

rscheff accepted D39776: Remove unused fields in siftr_stats. Thus, update the man page as well..
Apr 24 2023, 9:16 PM
rscheff accepted D39716: Correct the value of macro TF2_TCP_ACCOUNTING..
Apr 24 2023, 9:15 PM

Apr 19 2023

rscheff accepted D39693: tcp: Ending BB logs not always coming out.
Apr 19 2023, 3:53 PM
rscheff accepted D39688: netstat: cleanup.
Apr 19 2023, 10:49 AM

Apr 18 2023

rscheff accepted D39672: tcp: bbr.c is non-capable of doing ECN and sets an INP flag to fend off ECN however our syncache is not aware of that flag..
Apr 18 2023, 10:07 PM
rscheff added a comment to D39299: tcp: include RFC6675 IsLost() in pipe calculation.

tcp_compute_pipe is always controlled by mostly V_tcp_do_newsack in the output path and CC modules, and V_tcp_do_newcwv in the input path for deciding if the session is cwnd limited or not. New cwnd validation is disabled by default, newsack (formerly rfc6675_pipe) is enabled by default.

Apr 18 2023, 5:49 PM
rscheff updated the diff for D39299: tcp: include RFC6675 IsLost() in pipe calculation.
  • combine assignment and if check
Apr 18 2023, 5:43 PM

Apr 12 2023

rscheff added inline comments to D39501: TCP: Misc cleanups of tcp_subr.c.
Apr 12 2023, 6:10 PM
rscheff added a comment to D39536: net: Check per-flow priority code point.

So when I added the capability, it happened to work incidenitally one the nic drivers i had tested this on? And this would extend the proper operation to other drivers (or skip the PCP setting, when no vlan tagging capabilities exist?

Apr 12 2023, 5:24 PM
rscheff added a comment to D39501: TCP: Misc cleanups of tcp_subr.c.

Not questioning that change per se. But what you are functionally doing is not extracting the full traffic class (tclass), but extract the DSCP bits, clearing the ECN bits and sending traffic later using the same DSCP codepoint back.

Apr 12 2023, 4:07 PM

Apr 11 2023

rscheff closed D39503: use IPV6_FLOWLABEL_LEN.
Apr 11 2023, 5:24 PM
rscheff committed rG2169f71277b4: tcp: use IPV6_FLOWLABEL_LEN (authored by rscheff).
tcp: use IPV6_FLOWLABEL_LEN
Apr 11 2023, 5:24 PM
rscheff added inline comments to D39501: TCP: Misc cleanups of tcp_subr.c.
Apr 11 2023, 5:07 PM
rscheff requested review of D39503: use IPV6_FLOWLABEL_LEN.
Apr 11 2023, 2:42 PM

Apr 6 2023

rscheff added a comment to D39299: tcp: include RFC6675 IsLost() in pipe calculation.

validate that new pipe calc is gated by sysctl (to retain classical non-6675 pipe calculation if need be)

Apr 6 2023, 3:15 PM

Apr 4 2023

rscheff added a comment to D39403: Add myself (cc) as a src commiter..

These changes are already in, with D39368 - https://reviews.freebsd.org/source/src/browse/main/share/misc/committers-src.dot - not?

Apr 4 2023, 2:18 PM

Apr 1 2023

rscheff accepted D39368: Add myself (cc) as a src commiter..
Apr 1 2023, 3:23 PM

Mar 30 2023

rscheff accepted D39342: Add myself (cc) as a src commiter..
Mar 30 2023, 10:13 PM
rscheff requested changes to D39342: Add myself (cc) as a src commiter..

Yes, the entries should be sorted by mentor (1st), mentee (2nd).

Mar 30 2023, 4:21 PM

Mar 29 2023

rscheff updated the diff for D28822: Add PRR 6937bis heuristic and remove prr_conservative sysctl.
  • exclude limited transmit segments from prr_delivered.
Mar 29 2023, 5:22 AM

Mar 28 2023

rscheff requested review of D39299: tcp: include RFC6675 IsLost() in pipe calculation.
Mar 28 2023, 11:12 PM
rscheff closed D39274: tcp: send SACK rescue retransmission also mid-stream.
Mar 28 2023, 2:49 AM
rscheff committed rGf858eb916fd2: tcp: send SACK rescue retransmission also mid-stream (authored by rscheff).
tcp: send SACK rescue retransmission also mid-stream
Mar 28 2023, 2:48 AM

Mar 27 2023

rscheff added a comment to D28822: Add PRR 6937bis heuristic and remove prr_conservative sysctl.

Did more unit testing as promised. Found that the hardcoded CRB during dupack processing needs to be changed against the proper toggle (SACK_NEWLOSS or SACK_CHANGE) - to conform with RFC6937bis

Mar 27 2023, 12:52 PM
rscheff updated the diff for D28822: Add PRR 6937bis heuristic and remove prr_conservative sysctl.
  • honor proper SACK scoreboard change
Mar 27 2023, 12:47 PM

Mar 26 2023

rscheff updated the summary of D39274: tcp: send SACK rescue retransmission also mid-stream.
Mar 26 2023, 7:24 AM
rscheff updated the test plan for D39274: tcp: send SACK rescue retransmission also mid-stream.
Mar 26 2023, 7:23 AM
rscheff requested review of D39274: tcp: send SACK rescue retransmission also mid-stream.
Mar 26 2023, 7:21 AM

Mar 3 2023

rscheff committed rG9b3d6127716e: tcp: Make TCP PCAP buffer properly configurable. (authored by rscheff).
tcp: Make TCP PCAP buffer properly configurable.
Mar 3 2023, 9:29 PM
rscheff committed rG2be7e3df7275: tcp: Make TCP PCAP buffer properly configurable. (authored by rscheff).
tcp: Make TCP PCAP buffer properly configurable.
Mar 3 2023, 9:29 PM

Feb 28 2023

rscheff committed rG399a5655e6ed: tcp: Make TCP PCAP buffer properly configurable. (authored by rscheff).
tcp: Make TCP PCAP buffer properly configurable.
Feb 28 2023, 8:17 PM
rscheff closed D38824: Actually get/set the correct values when inquiring TCP_PCAP_[IN/OUT].
Feb 28 2023, 8:16 PM
rscheff updated the diff for D38824: Actually get/set the correct values when inquiring TCP_PCAP_[IN/OUT].
  • remove fat-finger line deletion
Feb 28 2023, 5:55 PM
rscheff requested review of D38824: Actually get/set the correct values when inquiring TCP_PCAP_[IN/OUT].
Feb 28 2023, 5:07 PM

Feb 26 2023

rscheff accepted D38797: New committer (src): Cheng Cui (cc).
Feb 26 2023, 9:37 PM

Feb 24 2023

rscheff committed R9:28764f8c693d: Import my keys with new expiration date (authored by rscheff).
Import my keys with new expiration date
Feb 24 2023, 1:48 PM

Feb 22 2023

rscheff accepted D38712: tcp: remove unused function prototype.
Feb 22 2023, 11:58 AM

Feb 21 2023

rscheff accepted D38710: bblog: improve timeout event.
Feb 21 2023, 8:27 PM
rscheff accepted D38701: bblog: allow logging of protocol user requests.
Feb 21 2023, 11:02 AM

Feb 15 2023

rscheff accepted D38547: tcp: allow a variable of type tt_which to be used as an index for tcptimers[].
Feb 15 2023, 3:30 PM
rscheff added a comment to D38547: tcp: allow a variable of type tt_which to be used as an index for tcptimers[].

Why would you need to remove the #ifdef TCPTIMERS when including tcp_var.h in tcp_timers.h?

Feb 15 2023, 12:15 PM

Feb 13 2023

rscheff accepted D38547: tcp: allow a variable of type tt_which to be used as an index for tcptimers[].

lgtm

Feb 13 2023, 9:59 AM

Feb 9 2023

rscheff closed D21117: Reduce size of t_rttupdated in tcpcb.
Feb 9 2023, 3:08 PM

Feb 3 2023

rscheff accepted D38353: tcp: retire net.inet.tcp.tcp_require_unique_port.

Also can not really think of any sane reason why you'd require unique ephemeral outgoing ports (outside of the local host). Quite the opposite, that restricts the total number of outgoing connections globally and not on a per-destination basis. Finally, this sysctl apparently was never mentioned anywhere (searchable) to be a workaround for any observed or perceived issue - while the original intent of the referenced patch is apparently working as expected.

Feb 3 2023, 6:32 PM

Jan 30 2023

rscheff added a comment to D28822: Add PRR 6937bis heuristic and remove prr_conservative sysctl.

Thanks Gleb! While at it, I also found that most other typedef enums use a nomenclature of a _t suffix. Also fixed up all relevant variables - haven't checked gdb to verify the proper decoding though.

Jan 30 2023, 7:00 PM
rscheff updated the diff for D28822: Add PRR 6937bis heuristic and remove prr_conservative sysctl.
  • properly define and use sackstatus_t type
Jan 30 2023, 6:57 PM

Jan 26 2023

rscheff updated the diff for D28822: Add PRR 6937bis heuristic and remove prr_conservative sysctl.
  • make the sack return status an enum for nice debugging
Jan 26 2023, 7:01 PM
rscheff committed rG18b83b626a4f: tcp: reduce the size of t_rttupdated in tcpcb (authored by rscheff).
tcp: reduce the size of t_rttupdated in tcpcb
Jan 26 2023, 5:32 PM

Jan 13 2023

rscheff requested review of D38034: Have a type-agnostic macro to determine the [u]int maximum value.
Jan 13 2023, 11:02 AM
rscheff updated the diff for D21117: Reduce size of t_rttupdated in tcpcb.
  • rebase to main
  • use static maximum value for uint8_t
Jan 13 2023, 10:39 AM

Dec 23 2022

rscheff accepted D37776: Correct some sysctls of the sysctl net.inet.tcp.bb family set..

I believe Cheng doesn't have the commit bit; I suggest we start the process of onboarding Cheng as another FBSD committer....

Dec 23 2022, 10:18 AM

Dec 15 2022

rscheff added a comment to D21117: Reduce size of t_rttupdated in tcpcb.

will use hardcode max_uint values in next rev.

Dec 15 2022, 3:03 PM

Nov 16 2022

rscheff accepted D37401: Remove t_rttbest.
Nov 16 2022, 6:30 AM

Nov 13 2022

rscheff updated the diff for D21117: Reduce size of t_rttupdated in tcpcb.
  • rebase to main
Nov 13 2022, 5:24 PM

Nov 12 2022

rscheff accepted D37375: ipfw: Spell PFIL_PASS correctly..

Wondering if a stronger type check can be enforced to find all the instances, where a magic number is used for pfil_return_t instead one of the literals...

Nov 12 2022, 4:25 PM

Nov 10 2022

rscheff committed rG86bdbdece4ae: ipfw: Have NAT steal the TH_RES1 bit, instead of the TH_AE bit (authored by rscheff).
ipfw: Have NAT steal the TH_RES1 bit, instead of the TH_AE bit
Nov 10 2022, 3:03 PM
rscheff closed D37314: tcp: account sent/received IP ECN markings independently.
Nov 10 2022, 11:10 AM
rscheff committed rG1a70101a8700: tcp: account sent/received IP ECN markings independently (authored by rscheff).
tcp: account sent/received IP ECN markings independently
Nov 10 2022, 11:10 AM
rscheff committed rG9839a5ad3a68: ipfw: Have NAT steal the TH_RES1 bit, instead of the TH_AE bit (authored by rscheff).
ipfw: Have NAT steal the TH_RES1 bit, instead of the TH_AE bit
Nov 10 2022, 7:47 AM
rscheff committed rG02751c6321ad: ipfw: Have NAT steal the TH_RES1 bit, instead of the TH_AE bit (authored by rscheff).
ipfw: Have NAT steal the TH_RES1 bit, instead of the TH_AE bit
Nov 10 2022, 7:23 AM

Nov 9 2022

rscheff updated the diff for D23230: Implement ECN++ (draft-generalized-ecn).
  • use switch statement instead of if else
Nov 9 2022, 4:56 PM
rscheff updated the diff for D37314: tcp: account sent/received IP ECN markings independently.
  • comment typo
Nov 9 2022, 4:43 PM
rscheff updated the diff for D37314: tcp: account sent/received IP ECN markings independently.
  • fix oversights
Nov 9 2022, 4:42 PM
rscheff committed rG0b00b801493a: ipfw: Have NAT steal the TH_RES1 bit, instead of the TH_AE bit (authored by rscheff).
ipfw: Have NAT steal the TH_RES1 bit, instead of the TH_AE bit
Nov 9 2022, 4:04 PM
rscheff closed D37300: ipfw: NAT steal TH_RES1 bit, instead of TH_AE.
Nov 9 2022, 4:04 PM
rscheff updated the diff for D37314: tcp: account sent/received IP ECN markings independently.
  • remove tcps_ecn_sndce
Nov 9 2022, 3:32 PM
rscheff added inline comments to D37314: tcp: account sent/received IP ECN markings independently.
Nov 9 2022, 12:08 PM
rscheff requested review of D37314: tcp: account sent/received IP ECN markings independently.
Nov 9 2022, 11:22 AM
rscheff updated the diff for D23230: Implement ECN++ (draft-generalized-ecn).
  • rebase
  • use ect1 if session is using that
Nov 9 2022, 10:38 AM
rscheff added inline comments to D23230: Implement ECN++ (draft-generalized-ecn).
Nov 9 2022, 9:12 AM

Nov 8 2022

rscheff closed D24869: Transition DCTCP to use ECT1 marks for elegible segments.
Nov 8 2022, 6:10 PM
rscheff committed rGdc9daa04fb68: tcp: allow packets to be marked as ECT1 instead of ECT0 (authored by rscheff).
tcp: allow packets to be marked as ECT1 instead of ECT0
Nov 8 2022, 6:10 PM
rscheff updated the diff for D24869: Transition DCTCP to use ECT1 marks for elegible segments.
  • add new flag at the bottom and next bit
Nov 8 2022, 5:38 PM
rscheff updated the diff for D24869: Transition DCTCP to use ECT1 marks for elegible segments.
  • rebase to main manually
Nov 8 2022, 2:07 PM
rscheff updated the diff for D23230: Implement ECN++ (draft-generalized-ecn).
  • rebase
  • bump man date
  • remove duplicate code
Nov 8 2022, 12:11 PM

Nov 7 2022

rscheff added a comment to D37300: ipfw: NAT steal TH_RES1 bit, instead of TH_AE.

FYI - The plan is also to MFC this to Stable/13 (prior to 13.2 release) and Stable/12

Nov 7 2022, 4:09 PM
rscheff updated the summary of D37300: ipfw: NAT steal TH_RES1 bit, instead of TH_AE.
Nov 7 2022, 12:58 PM
rscheff requested review of D37300: ipfw: NAT steal TH_RES1 bit, instead of TH_AE.
Nov 7 2022, 12:53 PM
rscheff updated the diff for D36303: Implement AccECN option (w/ early assigned option number).
  • move ecn sysctls to tcp_ecn.c
  • go with AccECN options from a compile time option to a global setting
  • document the new sysctl option
  • remove outdated comment
  • add logging, and make syncache response honor most recent incoming SYN fields
  • add logging
  • more logging
  • ensure control packets have accecn opt, if enabled
  • fix packt_this_ack bug
  • bump man date
  • received accecn counter into tcpinfo
  • add more tcpinfo counters
  • make accecn option counters a struct within tcpcb, which can also be locally allocated (syncache, prior of tcpcb)
  • initialize to_ae properly
  • reference properly to counters
  • bump man page date
  • remove all logging
Nov 7 2022, 10:18 AM

Nov 6 2022

rscheff committed rG37bf391d3cee: tcp: make tcp_packets_this_ack() only visible in kernel scope (authored by rscheff).
tcp: make tcp_packets_this_ack() only visible in kernel scope
Nov 6 2022, 12:53 PM
rscheff closed D37285: tcp: Move sysctl OIDs related to ECN to tcp_ecn.c.
Nov 6 2022, 11:40 AM
rscheff committed rG004bb636ca65: tcp: Move sysctl OIDs related to ECN to tcp_ecn.c (authored by rscheff).
tcp: Move sysctl OIDs related to ECN to tcp_ecn.c
Nov 6 2022, 11:40 AM
rscheff updated the diff for D37285: tcp: Move sysctl OIDs related to ECN to tcp_ecn.c.
  • remove ecn.option (part of accecn-option)
Nov 6 2022, 11:34 AM
rscheff requested review of D37285: tcp: Move sysctl OIDs related to ECN to tcp_ecn.c.
Nov 6 2022, 11:32 AM
rscheff closed D37281: tcp: add conservative d.cep accounting algorithm.
Nov 6 2022, 11:11 AM
rscheff committed rGb1258b76435a: tcp: add conservative d.cep accounting algorithm (authored by rscheff).
tcp: add conservative d.cep accounting algorithm
Nov 6 2022, 11:11 AM
rscheff closed D37280: tcp: add AccECN CE packet counter to tcpinfo.
Nov 6 2022, 11:04 AM
rscheff committed rG22c81cc51636: tcp: add AccECN CE packet counters to tcpinfo (authored by rscheff).
tcp: add AccECN CE packet counters to tcpinfo
Nov 6 2022, 11:04 AM
rscheff updated the diff for D37280: tcp: add AccECN CE packet counter to tcpinfo.
  • add tcpi_delivered_ce for accecn and ecn (accounting received ECEs)
  • report tcpi_delivered_ce properly
Nov 6 2022, 10:54 AM

Nov 5 2022

rscheff updated the diff for D37281: tcp: add conservative d.cep accounting algorithm.
  • fix packet_this_ack bug
Nov 5 2022, 4:27 PM
rscheff requested review of D37281: tcp: add conservative d.cep accounting algorithm.
Nov 5 2022, 3:33 PM
rscheff requested review of D37280: tcp: add AccECN CE packet counter to tcpinfo.
Nov 5 2022, 10:41 AM

Nov 4 2022

rscheff committed rG3708c3d370f2: tcp: reserve tcp_info counters for AccECN (authored by rscheff).
tcp: reserve tcp_info counters for AccECN
Nov 4 2022, 9:43 AM
rscheff closed D37016: tcp: sync with latest linux tcp_info struct.
Nov 4 2022, 9:43 AM

Nov 3 2022

rscheff closed D36973: more ECN preparations for ECN++, AccECN (tcp_respond).

didn't automatically close after commit.

Nov 3 2022, 4:17 PM
rscheff updated the diff for D37016: tcp: sync with latest linux tcp_info struct.
  • only reserve the accecn related counters
Nov 3 2022, 4:11 PM

Oct 25 2022

rscheff accepted D37125: tcp: don't store VNET in every tcpcb, take it from the inpcbinfo.
Oct 25 2022, 6:35 PM