Cool idea.
For most tests I think it would be an advantage to use a stub
implementation of OS.Time.sleep so that the tests would run almost
instantly, regardless of time-outs. When you need to verify that the unit
under test is sleeping for the correct amount of time then you'd need to
record the calls to sleep, or something similar.
The ARP test fails for me on test_queries_retried (extra debug messages
added):
ARP: sending gratuitous from 192.168.2.222
ARP: updating 192.168.2.7 -> 00:16:3e:7b:2b:4d
ARP: transmitting probe -> 192.168.2.4
# written: 2
ff ff ff ff ff ff ff ff ff ff ff ff 08 06 00 01
08 00 06 04 00 01 ff ff ff ff ff ff c0 a8 02 de
ff ff ff ff ff ff c0 a8 02 04
ff ff ff ff ff ff ff ff ff ff ff ff 08 06 00 01
08 00 06 04 00 02 ff ff ff ff ff ff c0 a8 02 de
ff ff ff ff ff ff 00 00 00 00
Fatal error: exception OUnitTest.OUnit_failure("expected: Success but got:
Last 3 sent packets weren't ARP probes")
Raised at file "src/core/lwt.ml", line 894, characters 22-23
Called from file "src/unix/lwt_main.ml", line 34, characters 8-18
Called from file "main.ml", line 36, characters 2-28
Looks like the second packet is a gratuitous ARP.
Luke
On 12 March 2015 at 20:54, Mindy <[email protected]> wrote:
> Hi all,
>
> I've just made public `mirage-net-pcap`, a module that satisfies
> V1.NETWORK by supplying packets from a prerecorded pcap file (via crunch)
> and keeping written packets in memory for later examination. You can see
> it at https://github.com/yomimono/mirage-net-pcap , and there's an
> example unikernel using it under `arp_tester` in my example-unikernels
> repository at https://github.com/yomimono/example-unikernels .
> `mirage-net-pcap` works with both unix and xen backends, although the
> example code works only with Unix because of its use of oUnit.
>
> Right now there isn't facility for saving writes to persistent storage
> (i.e., writing to a pcap). I started writing this and then realized it was
> generally the opposite of what I actually wanted when writing testing code
> that used `mirage-net-pcap`, but I do plan on adding this in a later
> release.
>
> I don't have a fork of the mirage frontend tool that nicely sets up a
> netif in `config.ml` for the user currently, but stacks can be built on
> top of one in the normal way (i.e., as in `mirage-skeleton/ethifv4`).
>
> Bad, horrible, okay, other?
>
> -Mindy
>
> _______________________________________________
> MirageOS-devel mailing list
> [email protected]
> http://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel
>
_______________________________________________
MirageOS-devel mailing list
[email protected]
http://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel
