GetNetworkList() copies sa_family of ifa_addr to ifa_netmask if unset

GetNetworkList() copies sa_family of ifa_addr to ifa_netmask if unset

BUG=325910
TEST=Confirmed NetUtilTest.GetNetworkList now passes in presence of OpenVPN tunnel.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=243675

[Johnny, 2014-01-08 15:48:53]

Hi Paul,

This is a minor patch for the OpenVPN tunnel w/o an sa_family set on the
ifa_netmask issue I saw while bootstrapping.

cheers,

[Johnny, 2014-01-08 15:51:55]

Unfortunately it doesn't seem there's a way to add automated testing, without
mocking getifaddrs() use in GetNetworkList()...

[pauljensen, 2014-01-08 18:11:20]

lgtm.

I did some googleing.  It seems like many other pieces of code do this same
fixup.  Seems like it's a known bug in OpenBSD/OSX.  I see references to
<rdar://problem/5492035> but I can't seem to view that; you may have to be an
Apple employee or something.  I tried looking for it in OpenBSD source but
didn't make it too far, looks like getifaddrs() uses something like AF_NETLINK
to read some routing tables from the kernel but I couldn't see where the OpenBSD
kernel was originally setting the netmask it was giving out.

It's unfortunate that we can't easily add a direct test for this but at the same
time this is fixing a failing test so one could argue there is already a test
for this...though obviously not frequently run in a configuration that would
catch this.

[commit-bot: I haz the power, 2014-01-08 19:53:18]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jgraettinger@chromium.org/105653003/10001

[commit-bot: I haz the power, 2014-01-08 22:29:57]

Change committed as 243675