Re: Regressions (network related)

[Paul Goyette <paul%whooppee.com@localhost>]

Andreas Gustafsson wrote:
> Yes, in the case of my test runs at least.  On the monthly details
> page you can see that the number of test failures increased from
> 2 to 10 with dyoung's commits of 2011.05.03.17.44.30 through
> 2011.05.03.18.28.46, and did not decrease when in6_pcb.c 1.114 was
> committed on 2011.05.04.01.45.48

Interesting.  I've only had two successful test runs since the 17:44:30 
commit (got bit by the %x vs %zx build breaks in vtw.c).  The first of 
these was with sources from 2011-05-03 18:20:01 and had only three test 
failures, none of which were network-related as far as I can tell.

1.      lib/libc/stdlib/t_strtod : strtod_inf
2.      lib/libm/t_infinity : infinity_long_double
3.      lib/librumpclient/t_fd : sigio

The second run was _after_ the in6_pcb.c fix, with sources from 
2011-05-04 05:20:01, and it has a total of nine failures (new ones 
flagged with *)

1. *    fs/nfs/t_rquotad : get_nfs_be_1_user
2.      lib/libc/stdlib/t_strtod : strtod_inf
3. *    lib/libevent/t_event : kqueue
4. *    lib/libevent/t_event : select
5.      lib/libm/t_infinity : infinity_long_double
6.      lib/librumpclient/t_fd : sigio
7. *    lib/librumphijack/t_tcpip : http
8. *    rump/rumpnet/t_shmif : crossping
9. *    syscall/t_getrusage : getrusage_utime_back

Of the new ones, the nfs and libevent entries are common failures (there 
are some interesting timing issues in qemu on amd!)


-------------------------------------------------------------------------
| Paul Goyette     | PGP Key fingerprint:     | E-mail addresses:       |
| Customer Service | FA29 0E3B 35AF E8AE 6651 | paul at whooppee.com    |
| Network Engineer | 0786 F758 55DE 53BA 7731 | pgoyette at juniper.net |
| Kernel Developer |                          | pgoyette at netbsd.org  |
-------------------------------------------------------------------------