Fix mask computation for v6 IPs

[santoshankr]

Hey

It looks like IPv6 mask fields are not being computed correctly.
For example, _create_mask(..., 10) returns [0xff, 0x03, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], when it should be returning [0xff, 0xc0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]

Steps to reproduce:

1. Create a new chain and add a rule accepting connections from fe80::/10

table = iptc.Table6('filter')
chain = table.create_chain('cX')

rule = iptc.Rule6()
rule.src = 'fe80::/10'
rule.create_target('ACCEPT')
chain.append_rule(rule)

2. Examine installed rules.

$ sudo ip6tables -nL cX
Chain cX (0 references)
target     prot opt source               destination
ACCEPT     all      fe80::/ff03::        ::/0

Patched version installs the rule we wanted

$ sudo ip6tables -nL cX
Chain cX (0 references)
target     prot opt source               destination
ACCEPT     all      fe80::/10            ::/0

[coveralls]

Coverage remained the same at 59.975% when pulling 09bf65d on santoshankr:ipv6-mask-fix into f8d1df6 on ldx:master.

[ldx]

Can you add a test for _create_mask()? Thanks!

[ldx]

Oh right! This should've been mask.append(0xff & (2 ** plen - 1)). Your approach works too. 👍

[santoshankr]

We need the top plen bits to be 1, so I think that's 0xff ^ (2 ** (8-plen) - 1)?
This is just dropping the bottom 8-plen bits though, so the double shift seemed a bit more obvious.

[coveralls]

Coverage increased (+0.04%) to 60.017% when pulling 599d8b6 on santoshankr:ipv6-mask-fix into f8d1df6 on ldx:master.

[santoshankr]

Added a couple of test cases for _create_mask

[ldx]

Thanks!