Re: kern/54553: Successful TFTP boot ends with error packet

[Valery Ushakov <uwe%stderr.spb.ru@localhost>]

The following reply was made to PR kern/54553; it has been noted by GNATS.

From: Valery Ushakov <uwe%stderr.spb.ru@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/54553: Successful TFTP boot ends with error packet
Date: Tue, 17 Sep 2019 13:52:06 +0300

 On Tue, Sep 17, 2019 at 09:00:01 +0000, Andreas Gustafsson wrote:
 
 > length 65: 10.169.0.2.2707 > 10.169.0.1.69:  23 RRQ "netbsd-install" octet
 [...]
 > length 558: 10.169.0.1.65405 > 10.169.0.2.2707:  516 DATA block 49138
 > length 60: 10.169.0.2.2707 > 10.169.0.1.65405:  4 ACK block 49138
 > length 558: 10.169.0.1.65405 > 10.169.0.2.2707:  516 DATA block 49139
 > length 60: 10.169.0.2.2707 > 10.169.0.1.65405:  5 ERROR ENOSPACE ""
 > 
 > Although this is probably harmless in itself, it makes it harder to
 > debug netboot failures by distracting attention from the actual cause.
 
 What is the actual size of netbsd-install?  It looks like the client
 decided it doesn't need the rest of the file (e.g. it's done reading
 ELF code/data segments and doesn't need e.g. symbol table or strings,
 etc) and terminates the transfer early, telling the sender the sender
 doesn't need to worry and resend the last packet after timeouts.
 
 I.e. it looks perfectly fine and courteous.  If that error packet
 bother you you can compile with -DTFTP_NOTERMINATE, but consider that
 the alternative will be seeing the server resending the last unacked
 packet.
 
 -uwe