don't assume an ipv4 header is always 20 bytes.
as reported by SecBuddyF, Tencent KeenLab:
> OpenBSD's pipex PPTP GRE packet handling uses a fixed GRE/IP header
> length check before later reading the GRE header at an offset derived
> from the actual IPv4 header length.
>
> When IPv4 options are present, the dynamic IPv4 header length can
> be larger than 20 bytes. A short GRE packet can therefore pass the
> fixed 28-byte PIPEX_IPGRE_HDRLEN check while still being too short
> for the later m_copydata(m0, hlen, sizeof(gre), &gre) read.
>
> This can trigger:
>
> panic: m_copydata: null mbuf
>
> The issue is remotely triggerable when GRE and pipex are enabled
> and inbound IPv4 GRE packets with IP options are allowed to reach
[8 lines not shown]
update to p5-Crypt-DSA-1.22, fixes key generation issue with previous
versions:
"Crypt::DSA::Util::makerandom forces the high bit of every value it
returns to obtain an exactly N-bit integer for prime search. The
signing nonce and the private key are drawn from makerandom. Because
the high bit is always set, the result is not uniform: its top bit is
fixed, producing insecure values."
Keys used to sign with an affected version should be considered
compromised and new keys should be generated.
iked: Wrong data structure zeroized
The parameter int mask was confused with the local variable struct
sockaddr_in mask4. bzero(3) was applied to mask instead of mask4.
ok markus@
