Re: [flashrom] [PATCH] Dediprog SF100 support

[Carl-Daniel Hailfinger]

Hi Stefan,

On 14.01.2010 14:08, Stefan Reinauer wrote:
> On 1/14/10 5:24 AM, Carl-Daniel Hailfinger wrote:
> > Dediprog SF100 support. Very fragile, lots of stuff guessed from logs.
> > May explode spontaneously.
> >
> > I think I know enough to actually use it to ID the chip, but that's it.
> > A USB log with which includes writing an image from the original
> > software (pretty much any image) would help clear up a few ambiguities.
> >
> > Signed-off-by: Carl-Daniel Hailfinger
>
> Awesome..!

Thanks!
> It seems to be talking to the device and get through the initialization,
> but it ends up with no "SPI" chips being "incompatible":

Actually, that's intentional. That way I could avoid hooking up the
driver in the SPI core (which would mean I'd have to create dummy
command functions) and still get initialization testing.
> MacPro:flashrom stepan$ ./flashrom -p dediprog -V
> flashrom v0.9.1-r862
> Found USB device (0483:dada).
> Sending RDID
> Receiving response: 20 20 16 10 00 00 00 00 00 00 00 00 00 00 00 00

Which flash chip is this? The response suggests a ST M25P32 flash chip,
but the extra "10" at the end looks rather strange.
> Not sure if this code is intended:
> > /* URB 9. Command Receive Device String. */
> > memset(buf, 0, sizeof(buf));
> > ret = usb_control_msg(dediprog_handle, 0xc2, 0x8, 0xff, 0xff, buf,
> > 0x10, DEFAULT_TIMEOUT);
> > if ((ret != 0x10) || !memcmp(buf, "SF100 V:2.1.1 ", 0x10)) {
>
> This would bail out if the version string is "SF100 V:2.1.1 "

Right, that's a bug. Considering that this was coded at 5am, I'm
surprised there was only one bug.
> > msg_perr("Unexpected response 3!\n");
> > return 1;
> > }
>
> Here I added this:
> > buf[0xf]=0;
> > msg_pdbg("Found a Dediprog %s\n", buf);

That potentially truncates the string. I'll use strnlen/strncpy to write
the string to a separate buffer which is big enough.
> Which got me:
>
> flashrom v0.9.1-r862
> Found USB device (0483:dada).
> Found a Dediprog SF100 V:3.1.8

Hm. The best idea might be to match against the model name and the major
version.
> Let me know if you need more information

A log with SniffUsb 2.0.0006 from
http://www.pcausa.com/Utilities/UsbSnoop/ would be very appreciated.
I wrote a tool to decode the log generated with that SniffUsb version.
Feel free to point me to a private upload via IRC (the log may contain
programmer serial numbers).

What the log should contain:
- Plugging in the SF100
- Starting the programming software
- SPI device detection/probe
- Full read
- Erase (optional)
- Full write (optional)
- No custom settings (but you can do a probe-only run with changed
settings in another file if you want to help decode a few more commands)

It would be cool if you could use a small SPI chip (4 Mbit or so)
because the logs roughly have 4x-20x the size of the flash chip.

Please do not use a SST chip (log would be at least 100x the size of the
flash chip). Pretty much all other vendors are OK.

I will post an updated patch soon.

Regards,
Carl-Daniel