http://en.wikipedia.org/wiki/Internet_protocol_suite - I don't understand why ICMP is considered on the Internet layer (where IP is), when 1) the ICMP headers don't contain any IP data 2) a captured ICMP packet with wireshark shows an IP layer

The purpose of the Internet layer is to select the best path through the network for packets to travel. The main protocol that functions at this layer is IP. Best path determination and packet switching occur at this layer.

The following protocols operate at the TCP/IP Internet layer:

• IP provides connectionless, best-effort delivery routing of packets. IP is not concerned with the content of the packets but looks for a path to the destination.
• Internet Control Message Protocol (ICMP) provides control and messaging capabilities.
• Address Resolution Protocol (ARP) determines the data link layer address, or MAC address, for known IP addresses.
• Reverse Address Resolution Protocol (RARP) determines the IP address for a known MAC address.
IP performs the following operations:
• Defines a packet and an addressing scheme
• Transfers data between the Internet layer and network access layer
• Routes packets to remote hosts

IP is sometimes referred to as an unreliable protocol. This does not mean that IP will not accurately deliver data across a network. IP is unreliable because it does not perform error checking and correction. That function is handled by upper layer protocols from the transport or application layers.

that doesn't answer my question.

isn't it something to do with relaying error messages.

Same layer because it relays errors with the ip connection?

I've got a sheet on it at home, sadly not at home....

I don't know why it does it that way, but from some reading apparently it simply shares a header with IP, and is located on that layer. oh well.

It would be rather difficult to ping someone if ICMP wasn't built on IP as the ICMP headers don't have a source and destination.

If you're referring to the fact that OSI specifies that IP and ICMP are in the network layer, you should remember that OSI is only a model and doesn't actually describe real life.

Then again, maybe it's there because of it's job, not based on the parent protocol.

I think because

1) it's not in the Link Layer because you can't use to to connect devices
2) it's not in the transport layer because it can't be used to transport data
3) it's not in the application layer because it's not using a transport layer to operate.