Re: [LI] two ques

[Sharad Joshi <sharadj@xxxxxxxxxxxxxxxx>]

On Thu, 4 Nov 1999, Chetan Kumar S wrote:
+ 
+ On Thu, 4 Nov 1999, Sharad Joshi wrote:
+ 
+ > 
+ > 0.  *WHY* the *minimum* ethernet frame size is 64 bytes. Any rational  
+ >     behind this? This means that *at-least* 46 bytes of data *must* be
+ >     present in the ethernet frame. (Even if there is not much data, that
+ 
+ How is that 46 bytes free, 14 (ether header) + 20 (IP) + 8 (UDP) = 42
                             ^^^^^^^^^^^^^^^^^
The data part does not include header and trailer. With these, the minimum
ethernet frame size is 64 bytes (14 + 46 + 4)

+ filled up ofcourse assuming tcp/udp over IP, if you are looking at
+ something else??.

no yaar, this is not the case. Ethernet protocol was developed
independently of TCP/IP, and any other protocols (e.g. OSI) can also use
it as underlying protocol. So that way, it is unrelated to what the upper
layers in TCP/IP protocol suite does.

+ 
+ >     *must* be appended with zeros to make it 46 bytes).
+ > 
+ > 1.  *WHY* is a pseudo header used for calculating checksum in TCP/UDP
+ >     headers. Is this not a overhead. Any specific reasons for this? RFCs 
+ >     are silent on this.
+ 
+ This should be OK, pseudo header will take care that I will have proper
+ port numbers (also sequence numbers, ack numbers in tcp). This is more
+ serious in terms of TCP where a loss in seq number can get you no
+ where. And as you know UDP checksum is not mandatory.

Well, AFAIK, checksum is used to preserve data integrity only. Is it used
for anything else? If that is the case, what is that 'else'? And if this
is not the case, then calculating checksum only over the respective
headers, exclusive of pseudo header, shud suffice, no?

Sharad.
-=-=-=-=-=-

--------------------------------------------------------------------
The Linux India Mailing List Archives are now available.  Please search
the archive at http://lists.linux-india.org/ before posting your question
to avoid repetition and save bandwidth.