[Babel-users] resuming 1.8.3 deployment
Dave Taht
dave.taht at gmail.com
Wed Oct 31 17:11:41 GMT 2018
ok, so, patching the xroute-nlogn branch with the attached patch,
(essentially the same as christofs), and running with a mix of
unicast and multicast routers doesn't crash.
Carnage still eventually ensues with injecting 8192 (* 2) routes from
two places, but I think replacing the existing linked list with a
binary tree in the resend routines will lick this. Possibly also
limiting garbage collection attempts in resend_expired as well.
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds seconds calls ms/call ms/call name
80.97 57.77 57.77 938622 0.06 0.06 find_resend
3.74 60.44 2.67 159556840 0.00 0.00 timeval_minus_msec
3.67 63.06 2.62 4980064 0.00 0.00 find_route_slot
1.35 64.02 0.96 6907 0.14 0.14 route_stream
1.12 64.82 0.80 48779070 0.00 0.00 route_stream_next
0.78 65.38 0.56 211642614 0.00 0.00 resend_expired
0.70 65.88 0.50 27750 0.02 0.13 recompute_resend_time
0.60 66.31 0.43 50625680 0.00 0.00 chg_route_metric
0.57 66.72 0.41 4733838 0.00 0.00 really_buffer_update
0.57 67.13 0.41 4544 0.09 3.00 update_neighbour_metric
0.49 67.48 0.35 51329597 0.00 0.00 prefix_cmp
0.34 67.72 0.24 450100 0.00 0.00 netlink_read
0.34 67.96 0.24 8705959 0.00 0.00 roughly
0.25 68.14 0.18 246825 0.00 0.00 kernel_route
0.22 68.30 0.16 timeval_minus
0.18 68.42 0.13 17959773 0.00 0.00 timeval_add_msec
0.17 68.54 0.12 792003 0.00 0.00 really_send_update
0.16 68.66 0.12 4939380 0.00 0.00 route_expired
0.15 68.77 0.11 68291064 0.00 0.00 is_default
0.15 68.88 0.11 998027 0.00 0.00 find_route
0.15 68.99 0.11 431076 0.00 0.00 netlink_talk
0.14 69.09 0.10 9417335 0.00 0.00 route_smoothed_metric
0.14 69.19 0.10 2897419 0.00 0.00 update_route_metric
0.14 69.29 0.10 544065 0.00 0.02 send_update
0.14 69.39 0.10 178975 0.00 0.00 find_source
0.14 69.49 0.10 86 1.16 32.92 conflict_solution
0.11 69.57 0.08 498328 0.00 0.00 find_best_route
0.10 69.64 0.07 14280435 0.00 0.00 accumulate_short
0.10 69.71 0.07 4875431 0.00 0.00 accumulate_bytes
:
On Wed, Oct 31, 2018 at 9:32 AM Dave Taht <dave.taht at gmail.com> wrote:
>
> I merged the first two against the nlogn branch (it looked like
> juliusz had fixed the dref), and am testing it now, using a mixture of
> unicast and multicast. )
>
> however, after a few minutes after injecting rtod -r 8192 from two places...
>
> It looks like it crashing on the same kind of dref on a different
> path. Your patch was different, so I'll try that in a sec.
>
> [Thread debugging using libthread_db enabled]
> Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
> Core was generated by `./babeld -L /tmp/babel.log -l -g 33123 -S
> /var/lib/babeld/state eno1 wlp2s0 wlx'.
> Program terminated with signal SIGSEGV, Segmentation fault.
> #0 0x000000000040c21c in start_message (buf=buf at entry=0x14af6c8,
> type=type at entry=10, len=len at entry=22) at message.c:957
> 957 buf->buf[buf->len++] = type;
> (gdb) up
> #1 0x000000000040c5b0 in send_multihop_request (buf=0x14af6c8,
> prefix=0x14cefc8 "\374\346\022\302ؑ\b\201", plen=<optimized out>,
> src_prefix=0x14cefd9 "", src_plen=<optimized out>, seqno=<optimized out>,
> id=0x14cefc0 "\002%\220\377\376\301bR\374\346\022\302ؑ\b\201",
> hop_count=127) at message.c:1809
> 1809 start_message(buf, MESSAGE_MH_REQUEST, len);
> (gdb) bt
> #0 0x000000000040c21c in start_message (buf=buf at entry=0x14af6c8,
> type=type at entry=10, len=len at entry=22) at message.c:957
> #1 0x000000000040c5b0 in send_multihop_request (buf=0x14af6c8,
> prefix=0x14cefc8 "\374\346\022\302ؑ\b\201", plen=<optimized out>,
> src_prefix=0x14cefd9 "", src_plen=<optimized out>, seqno=<optimized out>,
> id=0x14cefc0 "\002%\220\377\376\301bR\374\346\022\302ؑ\b\201",
> hop_count=127) at message.c:1809
> #2 0x000000000040e019 in send_request_resend (
> prefix=0x14cefc8 "\374\346\022\302ؑ\b\201", plen=64 '@',
> src_prefix=0x14cefd9 "", src_plen=0 '\000', seqno=<optimized out>,
> id=0x14cefc0 "\002%\220\377\376\301bR\374\346\022\302ؑ\b\201")
> at message.c:1907
> #3 0x000000000040a836 in flush_route (route=route at entry=0x1628180)
> at route.c:300
> #4 0x000000000040b076 in expire_routes () at route.c:1216
> #5 0x0000000000403121 in main (argc=<optimized out>, argv=<optimized out>)
> at babeld.c:751
>
> On Tue, Oct 30, 2018 at 12:03 PM Christof Schulze
> <christof.schulze at gmx.net> wrote:
> >
> > On Thu, Oct 25, 2018 at 09:26:51PM +0200, Juliusz Chroboczek wrote:
> > >> Also it would be really helpful if that remaining PR (prefsrc) could be
> > >> considered.
> > Sure! I attached the two patches on this mail.
> >
> > Patch 0001 removes outdated information from the manpage
> > Patch 0002 adds prefsrc capabilities
> > Patch 0003 fixes a build issue in rfc6126bis in message.c
> >
> >
> >
> > viele Grüße
> >
> > Christof
> >
> >
> >
> > >
> > >Christof, are you willing to send me a clean patch against the rfc6126bis
> > >branch, or shall I do the merge myself?
> > >
> > >-- Juliusz
> >
> > --
> > () ascii ribbon campaign - against html e-mail
> > /\ against proprietary attachments
> >
> > _______________________________________________
> > Babel-users mailing list
> > Babel-users at alioth-lists.debian.net
> > https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/babel-users
>
>
>
> --
>
> Dave Täht
> CTO, TekLibre, LLC
> http://www.teklibre.com
> Tel: 1-831-205-9740
--
Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-205-9740
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-yet-another-fix-for-interface-deref.patch
Type: text/x-patch
Size: 1204 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/babel-users/attachments/20181031/df3a3590/attachment.bin>
More information about the Babel-users
mailing list