Page MenuHomeFreeBSD
Feed Advanced Search

Mar 24 2016

gnn closed D4306: Re-add route caching for TCP by committing rS297225: FreeBSD previously provided route caching for TCP (and UDP). Re-add.
Mar 24 2016, 7:55 AM · transport
gnn accepted D4306: Re-add route caching for TCP.
Mar 24 2016, 7:51 AM · transport
gnn added a comment to D4306: Re-add route caching for TCP.

Having applied this to HEAD I find that with the following test:

Mar 24 2016, 7:51 AM · transport
gnn added a comment to D4306: Re-add route caching for TCP.

I'll give this a benchmark in the Sentex lab.

Mar 24 2016, 6:24 AM · transport
olivier added a comment to D4306: Re-add route caching for TCP.
In D4306#121368, @mike-karels.net wrote:

This change does not affect forwarding, just TCP and UDP termination. It makes the most difference when TSO is not enabled.

Mar 24 2016, 6:19 AM · transport

Mar 19 2016

karels added a comment to D4306: Re-add route caching for TCP.

This change does not affect forwarding, just TCP and UDP termination. It makes the most difference when TSO is not enabled.

Mar 19 2016, 1:14 PM · transport
olivier added a comment to D4306: Re-add route caching for TCP.

In my standard forwarding scenario (2 static routes), I didn't see improvement neither regression with this patch (about -1% with pf or ipfw):

Mar 19 2016, 11:17 AM · transport

Mar 18 2016

karels updated the diff for D4306: Re-add route caching for TCP.

Fix user-level include of in_pcb.h
rt_gen_t was in an ifdef _KERNEL

Mar 18 2016, 1:24 AM · transport
karels added a comment to D4306: Re-add route caching for TCP.

Argh, I see what I did. I ended up with the wrong version of route.h in /usr/include.

Mar 18 2016, 1:17 AM · transport

Mar 16 2016

olivier added a comment to D4306: Re-add route caching for TCP.

I've sync my -head source to 296935, then applyed this patch without problem, but buildworld failed:

Mar 16 2016, 2:32 PM · transport
gnn added a reviewer for D4306: Re-add route caching for TCP: olivier.
Mar 16 2016, 8:02 AM · transport
gnn added a comment to D4306: Re-add route caching for TCP.

I've added Olivier as a reviewer because I'm hoping he'll test this patch and give us some updated performance numbers.

Mar 16 2016, 8:02 AM · transport
gnn updated subscribers of D4306: Re-add route caching for TCP.
Mar 16 2016, 8:01 AM · transport

Mar 14 2016

karels updated the diff for D4306: Re-add route caching for TCP.

Update to match head, fix UDP/IPv4 locking
Move definitions to match movement in header files.

Mar 14 2016, 3:02 AM · transport

Mar 9 2016

gnn accepted D4306: Re-add route caching for TCP.

4102 has already gone in. It's time to push this in as well.

Mar 9 2016, 1:49 AM · transport

Feb 15 2016

karels added a comment to D4490: inpcb rtentry/l2 prepend caching.

OK, I missed the malloc/copy in if_ethersubr.c. So there isn't a problem with pointers to freed memory, but the L2 code still has the issue as the routing code with copying the pointers to/from the route structure.

Feb 15 2016, 4:58 PM · transport

Feb 9 2016

kmacy added a comment to D4490: inpcb rtentry/l2 prepend caching.
In D4490#111292, @mike-karels.net wrote:
I'll respond to other parts soon.  But I want to correct this:

About L2 caching: this approach is very wrong. Caching a pointer and length to the L2 header allows no way to validate or invalidate the cache. What happens when the L2 entry times out and is deleted?

I invalidate all cached context. See if_llatbl.c.

This is not effective. Consider the following:

Thread A checks the routing generation number in ip_output. Thread B then increments the generation count of all the routing tables (the bigger hammer approach!)

Feb 9 2016, 5:32 AM · transport
karels added a comment to D4490: inpcb rtentry/l2 prepend caching.
I'll respond to other parts soon.  But I want to correct this:
Feb 9 2016, 2:42 AM · transport

Feb 7 2016

kmacy added a comment to D4490: inpcb rtentry/l2 prepend caching.
In D4490#111105, @mike-karels.net wrote:

About L3 caching:

I think my approach, including a "struct route" in the in_pcb, is both simpler and more correct.

I dislike building a "struct route" on the stack and then moving fields to and from that structure from the pcb. I think it is wrong to compare the route using the in_pcb destination address, as that may have changed (e.g. in the UDP case). The struct route contains the previously-looked-up destination, which is necessary.

Feb 7 2016, 9:37 PM · transport
karels added a comment to D4490: inpcb rtentry/l2 prepend caching.

About L3 caching:

Feb 7 2016, 8:16 PM · transport

Feb 4 2016

kmacy added a reviewer for D4490: inpcb rtentry/l2 prepend caching: hiren.
Feb 4 2016, 6:28 PM · transport
kmacy updated D4490: inpcb rtentry/l2 prepend caching.
Feb 4 2016, 6:27 PM · transport
kmacy added a comment to D4490: inpcb rtentry/l2 prepend caching.

Could Mike point out what his patch provides that this doesn't - apart from the lookup on bound UDP? This provides L2 caching which is important.

Feb 4 2016, 5:43 PM · transport
kmacy closed D4364: inpcb L2/L3 caching.

Closed in favor of 4490.

Feb 4 2016, 5:41 PM · transport
kmacy updated the diff for D4490: inpcb rtentry/l2 prepend caching.

Updated to (I hope) address all the comments and compile against the most recent HEAD. This has been used in ISLN's internal incast testing and I'm hoping to get LL and NFLX to test.

Feb 4 2016, 12:39 AM · transport

Feb 3 2016

kmacy added inline comments to D4490: inpcb rtentry/l2 prepend caching.
Feb 3 2016, 11:35 PM · transport

Jan 23 2016

karels updated the diff for D4306: Re-add route caching for TCP.

This revision does several things:

Jan 23 2016, 4:55 PM · transport

Dec 22 2015

bz added a comment to D4364: inpcb L2/L3 caching.

Where do we want to do the review? Here or on D4490?

Dec 22 2015, 5:12 PM · transport
bz added a reviewer for D4490: inpcb rtentry/l2 prepend caching: bz.
Dec 22 2015, 5:11 PM · transport
bz added a reviewer for D4364: inpcb L2/L3 caching: bz.
Dec 22 2015, 5:10 PM · transport

Dec 16 2015

jtl added a comment to D4490: inpcb rtentry/l2 prepend caching.

Some in-line comments from a quick pass.

Dec 16 2015, 5:20 PM · transport

Dec 15 2015

ume added inline comments to D4546: export the cc algorithm by way of netstat.
Dec 15 2015, 3:42 PM · transport
gnn accepted D4546: export the cc algorithm by way of netstat.
Dec 15 2015, 3:34 PM · transport

Dec 14 2015

kmacy retitled D4546: export the cc algorithm by way of netstat from to export the cc algorithm by way of netstat.
Dec 14 2015, 3:45 AM · transport

Dec 12 2015

melifaro added a comment to D4490: inpcb rtentry/l2 prepend caching.

Thanks for updating the patch!
Looks OK, several small comments inline

Dec 12 2015, 4:58 PM · transport

Dec 11 2015

kmacy added a comment to D4364: inpcb L2/L3 caching.

See D4490.

Dec 11 2015, 6:16 PM · transport
kmacy retitled D4490: inpcb rtentry/l2 prepend caching from to inpcb rtentry/l2 prepend caching.
Dec 11 2015, 2:47 AM · transport

Dec 8 2015

kmacy added a comment to D4364: inpcb L2/L3 caching.

I'll upload a new patch against D4102 and then Mike can do similarly then we'll have a inpcb route caching showdown.

Dec 8 2015, 1:58 AM · transport
karels added a comment to D4364: inpcb L2/L3 caching.

Randall: see also D4306 for another approach.

Dec 8 2015, 1:47 AM · transport

Dec 7 2015

rrs accepted D4364: inpcb L2/L3 caching.

I am *very* glad to see a cached route coming back in :-)

Dec 7 2015, 6:12 PM · transport

Dec 5 2015

kmacy added a comment to D4306: Re-add route caching for TCP.

I will review 4102 ASAP.

Dec 5 2015, 6:05 PM · transport
kmacy updated the diff for D4366: convert tcp timestamps to scaled sbintime.

restore 40 line context

Dec 5 2015, 6:34 AM · transport
kmacy updated the diff for D4366: convert tcp timestamps to scaled sbintime.

convert tcp_timer_activate to sbintime

Dec 5 2015, 6:33 AM · transport

Dec 4 2015

melifaro added a comment to D4306: Re-add route caching for TCP.

Meant D4102 in last comment

Dec 4 2015, 8:09 PM · transport
melifaro added a comment to D4306: Re-add route caching for TCP.

The only thing that stops projects/routing work from landing is D4102 which conflicts with this particular patch.

Dec 4 2015, 8:02 PM · transport
gnn added a comment to D4306: Re-add route caching for TCP.

I do not believe that this, or Matt's, change should make doing the right thing with the routing system any more difficult than it already is. I think it's time we pushed either this change or the associated change into the tree. The other routing work is longer term.

Dec 4 2015, 7:46 PM · transport
kbowling added a project to D4306: Re-add route caching for TCP: transport.
Dec 4 2015, 7:57 AM · transport
kmacy retitled D4366: convert tcp timestamps to scaled sbintime from to convert tcp timestamps to scaled sbintime.
Dec 4 2015, 2:04 AM · transport

Dec 3 2015

kmacy retitled D4364: inpcb L2/L3 caching from to inpcb L2/L3 caching.
Dec 3 2015, 11:41 PM · transport

Dec 1 2015

hiren added a comment to D4295: Add driver backpressure.
In D4295#90577, @kmacy wrote:
In D4295#90451, @hiren wrote:

@kmacy I don't think iflib is part of -head yet. I see this patch includes references to sys/net/iflib* which also don't exist in -head yet. Can this patch be applied cleanly to -head? if not, I'd be great to have that done.

It's not part of HEAD yet. The objective here is to elucidate the mechanism. I haven't been able to test it under real load yet. Do you really want to run with scissors? :D

Dec 1 2015, 11:22 PM · transport
sbruno accepted D4295: Add driver backpressure.

I'm adding a "looks good to me" here for the Intel Reviewers as this adds functionality in the stack and the modification to the driver is required.

Dec 1 2015, 9:11 PM · transport

Nov 30 2015

christianchristensen_gmail.com added a watcher for transport: christianchristensen_gmail.com.
Nov 30 2015, 2:04 AM
christianchristensen_gmail.com added a member for transport: christianchristensen_gmail.com.
Nov 30 2015, 1:49 AM
davide added a comment to D4292: decouple maximum clock frequency from callout scheduling granularity.

The changes look good. I'd split them in multiple patches/commit, more or less following the structure you outlined in the summary, but thanks for sending out a single one so we can have the whole picture.
IIRC there were objections to change ticks from 32-bits to 64-bits so you may want to ask feedback from a more general audience.
Other than that, I'm fine with (a splitted version of this) getting in.

Nov 30 2015, 12:13 AM · transport

Nov 28 2015

kmacy added a comment to D4295: Add driver backpressure.
In D4295#90451, @hiren wrote:

@kmacy I don't think iflib is part of -head yet. I see this patch includes references to sys/net/iflib* which also don't exist in -head yet. Can this patch be applied cleanly to -head? if not, I'd be great to have that done.

Nov 28 2015, 11:37 PM · transport

Nov 27 2015

hiren updated subscribers of D4295: Add driver backpressure.
Nov 27 2015, 7:47 AM · transport
hiren added a comment to D4295: Add driver backpressure.

@kmacy I don't think iflib is part of -head yet. I see this patch includes references to sys/net/iflib* which also don't exist in -head yet. Can this patch be applied cleanly to -head? if not, I'd be great to have that done.

Nov 27 2015, 7:47 AM · transport
hiren added a comment to D4294: modernize TCP constants.

A couple of comments in-line after a very quick glance. I'll try to take a better look next week.

Nov 27 2015, 7:34 AM · transport
kmacy updated subscribers of D4294: modernize TCP constants.
Nov 27 2015, 1:37 AM · transport
kmacy updated subscribers of D4295: Add driver backpressure.
Nov 27 2015, 1:37 AM · transport
kmacy retitled D4295: Add driver backpressure from to Add driver backpressure.
Nov 27 2015, 1:26 AM · transport
kmacy retitled D4294: modernize TCP constants from to modernize TCP constants.
Nov 27 2015, 12:47 AM · transport
kmacy updated subscribers of D4292: decouple maximum clock frequency from callout scheduling granularity.
Nov 27 2015, 12:42 AM · transport
kmacy retitled D4293: update SRTT to account for flight size from to update SRTT to account for flight size.
Nov 27 2015, 12:41 AM · transport

Nov 12 2015

eadler added a member for transport: freebsd-transport.
Nov 12 2015, 6:22 AM

Nov 11 2015

jtl added a member for transport: jtl.
Nov 11 2015, 6:57 PM

Oct 24 2015

gnn added a member for transport: bz.
Oct 24 2015, 6:43 PM

Oct 21 2015

eadler set the color for transport to Indigo.
Oct 21 2015, 5:16 AM

Oct 19 2015

gnn added a member for transport: rrs.
Oct 19 2015, 3:03 PM

Oct 16 2015

benno added a member for transport: rstone.
Oct 16 2015, 7:30 PM
jch added a member for transport: jch.
Oct 16 2015, 7:41 AM

Oct 15 2015

gnn added members for transport: hiren, jonlooney_gmail.com, rwatson, kbowling, benno, jason_eggnet.com, j-nitrology.com, scottl.
Oct 15 2015, 9:01 PM
eadler added a member for transport: gnn.
Oct 15 2015, 8:55 PM
eadler changed the edit policy for transport.
Oct 15 2015, 8:40 PM
eadler created transport.
Oct 15 2015, 8:40 PM