Must every packet have a creating socket? (was Re: Does a forwarded packet has a local socket with it?)

Must every packet have a creating socket? (was Re: Does a forwarded packet has a local socket with it?)

[Park Lee]

On Tue, 19 Apr 2005 at 08:26, Neil Horman wrote:
> Not sure what your asking here.  Asking "Does these 
> two sockets belong to local socket?" is a pretty 
> meaningless question, as the terminology is 
> nonsensical. All sockets have a protocol associated 
> with them (except raw sockets, but thats another 
> topic).  When you create a socket you associate it 
> with a protocol family (AF_INET, AF_INET6, 
> AF_NETLINK, AF_APPLETALK, etc), 
> a connection type (SOCK_STREAM for connection 
> oriented  protocols, SOCK_DGRAM for connection-less 
> protocols), and a protocol (IPPROTO_TCP, 
> IPPROTO_UDP, etc).  Depending on the family, type 
> and protocol you select, you can talk to different 
> systems/services.
>
> Does that answer your question?

Thanks a lot.

Can I think that every packet (e.g. IP packet) must
have a corresponding creating socket? (i.e. Must every
packet be created by a socket?) 
Is there any other way to originate a packet?

Thank you.

Best Regards,
Park Lee


		
__________________________________ 
Do you Yahoo!? 
Make Yahoo! your home page 
http://www.yahoo.com/r/hs

Re: Must every packet have a creating socket? (was Re: Does a forwarded packet has a local socket with it?)

[Rick Jones]

> Can I think that every packet (e.g. IP packet) must
> have a corresponding creating socket? (i.e. Must every
> packet be created by a socket?) 

No.  ICMP messages come to mind - although I _suppose_ that since those are in 
response to other traffic, you could claim it was in response to something sent 
from a "socket" or "endpoint" - depends on how far away you consider it to still 
be  from a socket.

> Is there any other way to originate a packet?


> 
> Thank you.
> 
> Best Regards,
> Park Lee
> 
> 
> 		
> __________________________________ 
> Do you Yahoo!? 
> Make Yahoo! your home page 
> http://www.yahoo.com/r/hs

Re: Must every packet have a creating socket?

[Andi Kleen]

Rick Jones <rick.jones2@hp.com> writes:

>> Can I think that every packet (e.g. IP packet) must
>> have a corresponding creating socket? (i.e. Must every
>> packet be created by a socket?)
>
> No.  ICMP messages come to mind - although I _suppose_ that since
> those are in response to other traffic, you could claim it was in
> response to something sent from a "socket" or "endpoint" - depends on
> how far away you consider it to still be  from a socket.

Actually Linux has kernel private sockets for ICMP.
Very old Linux didnt, but it required ugly special cases
in the transmit path, so it was removed.

-Andi

Re: Must every packet have a creating socket? (was Re: Does a forwarded packet has a local socket with it?)

[Park Lee]

On Wed, 20 Apr 2005 at 18:20, Rick Jones wrote:
>> On Wed, 20 Apr 2005 at 10:29, Park Lee wrote:
>> Can I think that every packet (e.g. IP packet) must
>> have a corresponding creating socket? (i.e. Must 
>> every packet be created by a socket?) 
>
> No.  ICMP messages come to mind - although I 
> _suppose_ that since those are in response to other 
> traffic, you could claim it was in response to 
> something sent from a "socket" or "endpoint" - 
> depends on how far away you consider it to still 
> be  from a socket.

    But as I know, The Linux network component creates
two special purpose sockets for use by the AF_INET
protocol family. The tcp socket is used to send resets
when a TCP packet is rejected, since there may be no
local socket corresponding to the packet. The icmp
socket is used to send ICMP messages.
    Then, ICMP echo replies are asociated with the
special kernel socket.
    So, I still think that every packet (e.g. IP
packet) must have a corresponding creating socket. 
    Am I right?

    Thank you.

Best Regards,
Park Lee


		
__________________________________ 
Do you Yahoo!? 
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/