This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
Re: RedBoot: '-v' option causes checksum failure when loa
On Mon, May 14, 2001 at 10:06:05AM -0600, Gary Thomas wrote:
> > My (somewhat modified) version of RedBoot has a problem with a
> > TFTP load using the '-v' option via telnet. If you telnet in
> > to the box and to a 'load -v -h 1.2.3.4 filename' you get an
> > S-Record checksum failure at a certain point in the file (it's
> > always the same address, but the failure value varies). The
> > load without the '-v' works fine, as does the same command with
> > the '-v' isssued from a serial console port.
> >
> > This is probably caused by one of my changes, but I thought I'd
> > ask if anybody else had seen this problem.
>
> Nope, I use this all the time.
I thought something like this would have been caught before now.
> What sort of changes have you made that might affect this?
I've changed a bunch of things, none of them should have
caused something like this, but I've added a lot of stuff:
* Default addresses for "fis create" taken from result of last
"load" command.
* Password for telnet connections
* Changed getc routines so that a timeout of "1 ms" returns
immediately.
* Support for gateway use by the IP stack.
* "Direct" load mode to load S-Record files via the telnet/TCP
connection. I could never get X-Modem via telnet to work, so I
use a simple stop-and-wait scheme where I ACK each S-Record
with a single newline. It's excruciatingly slow, but it
works.
* Listen on another TCP port for a propietary protocol
connection for backwards compatiblity.
* A bunch of commands for storing/retreiving configuration
parameters in an IIC EEEPROM.
One of them obviously had an un-intended side-effect. :)
--
Grant Edwards
grante@visi.com