Re: [ECOS] redboot tftp not acking all packets?

["Gary Thomas" <gary at mlbassoc dot com>]

Andy Dyer said:
> Hello all:
> 	I seem to be having a problem with tftp when I
> download an unstripped elf file via tftp in redboot.
> The tftp server is a PC running cygwin, and the symptom
> I see is for every tftp transfer I do it leaves a copy
> of in.tftpd.exe running on the PC.  These seem to have a
> really long timeout, and leave the file open on the PC,
> so I have to manually kill the processes if I want to
> remove the file.
>
> I captured the packets on the wire, and it seems that
> once redboot gets what it needs from the elf file, it
> stops ack'ing the transfer, leaving the server process to
> sit there resending.
>
> What I see in the packet dump is the data packets and acks
> proceed correctly in sequence up to 149, then the tftp
> server sends 150 and redboot never ack's 150, instead
> returning to the command prompt.
>
> The full file I'm trying to load is 874787 bytes long, which
> should be 1709 tftp blocks (512 bytes/per).  If I look at the
> output of objdump -h on the elf file it looks like the file
> offset of the start of the debugging info (.stab) is 0x129f4
> or 148.9 tftp blocks worth of data.
>
> Has anyone else seen this?

Yes, but with other transports.

>
> Looking at the spec I could either force an error packet
> response or just read the extra characters and throw them away.
> Any thoughts on what's best?

Error out - this is similar to a change just committed for the
HTTP download function.




-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss