[c-nsp] corrupted IOS image when transferring using

[Tim Franklin]

>  That is because it is really hard to convice (original) tftp to go
>  over 32MB. According to the RFC1350 tftp sends the file in 512 byte
>  blocks and there is a two byte field for numbering the blocks. So, if
>  I used bc correctly, the maximum number of bytes without the counter
>  wrapping around is 33553920. 32 MB is 33554432 bytes, 512 less than
>  the maximum. So the maximum size would be 32MB + 511 bytes (not 512
>  due to the end of file detection in tftp).

It's 32M - 1 byte.  64K sequence numbers at 0.5K data per packet is 32M,
except for the last packet which is a byte short for EOF, as you say.

It's even worse when someone's coded your tftpd with sequence number as
two-byte signed *cough*Solaris*cough*.

>  The extensions provided in RFCs 2347 - 2349 may help the matters but
>  I've no idea how widely they are supported. In any case, ftp, rcp and
>  scp are far superior to tftp (except in protocol overhead).

I'm a *big* fan of 'copy http: flash:' where supported - firewall guys
everywhere are amenable to port 80 so it Just Works.

Regards,
Tim.

-- 
____________   Tim Franklin                 e: tim at colt.net 
\C/\O/\L/\T/   Product Engineering Manager  w: www.colt.net 
 V  V  V  V    Managed Data Services        t: +44 20 7863 5714 
                                            f: +44 20 7863 5876