Default to 2Mbaud for Sonoffs?

[XenoKovah]

If auto-detection for baud is currently supposed to work, then this is a bug report that it doesn't.

But I think it's not supposed to, due to it being mentioned on past tickets that differentiating the slow vs. fast devices is difficult or possibly impossible. But my request would be that we default to assuming that everyone can get the fast version of the chip these days, and that the slow version was just an anomaly from supply chain shortages during the pandemic. So my request is that sniffle_hw.py is updated to default to 2000000 for Sonoffs and let people with the slow version do the override, rather than people with the fast version.

diff --git a/python_cli/sniffle/sniffle_hw.py b/python_cli/sniffle/sniffle_hw.py
index 102b669..4474372 100644
--- a/python_cli/sniffle/sniffle_hw.py
+++ b/python_cli/sniffle/sniffle_hw.py
@@ -115,7 +115,8 @@ class SniffleHW:
                 else:
                     baud = 921600
         elif is_cp2102(serport):
-            baud = 921600
+#            baud = 921600
+            baud = 2000000

I say this just because it'd be nice if I didn't have to override the baud at all and could just have students use no-diffs Sniffle in our upcoming OST2 BLE class...

[sultanqasim]

I was contemplating doing the same. I defaulted to 921600 because as you say, I couldn't find a reliable way to distinguish between CP2102 and CP2102-based devices, so just encouraged everyone to use 921600 baud on CP2102 devices since it's generally "fast enough". However, adding a baud rate option and defaulting to 2M baud for all devices may be a better option going forward, given that the CP2102 (non-N) Sonoff dongles seem to have been a ~2022 era anomaly.

[XenoKovah]

Random question: would this change also make the auto-detection simpler? (And thus work by default again for Sonoff /dev/ttyUSB* devices.) Because I just remembered that I think with the TI dev boards I didn't need to always be passing the -s option, it just auto-detected compatible devices, but with Sonoffs I seem to always need to pass -s. If so, that seems like another good reason to do it.

[sultanqasim]

It should already be auto detecting Sonoff dongles when no -s option is specified (and use the 921600 baud rate with them). See this function:

Sniffle/python_cli/sniffle/sniffle_hw.py

Line 55 in 84b1ec0

def find_sonoff_serport():

Is there something different about your Sonoff dongles that prevents it from auto detecting them?

[XenoKovah]

Is there something different about your Sonoff dongles that prevents it from auto detecting them?

Looks like it's down to where I do the baud edit for 2Mbaud I guess:

    def __init__(self, serport=None, logger=None, timeout=None):
        baud = 2000000
        if serport is None:
            serport = find_xds110_serport()
            if serport is None:
                serport = find_sonoff_serport()
                if serport is None:
                    serport = find_catsniffer_v3_serport()
                    if serport is None:
                        raise IOError("Sniffle device not found")
                else:
#                    baud = 921600
                    baud = 2000000 # This makes auto-detection work
        elif is_cp2102(serport):
#            baud = 921600
            baud = 2000000 # I only usually change this

So in this sense, yes defaulting back to 2000000 instead of 921600 would make Sonoffs "just work" for me for autodetection too.

[mdxs]

I think that PR #102 actually proposes (amongst other changes) to add an explicit --baudrate command line argument, not sure how this exactly fits this discussion as I don't use/need it myself... but maybe something to look at in this context.

[XenoKovah]

I agree that the --baudrate option would seem to achieve my request, as long as it's paired with the removal of any straggler references to 921600 like this: https://github.com/nccgroup/Sniffle/pull/102/files#r1805402099

[sultanqasim]

Implemented with dffc49c

[XenoKovah]

Thanks for this! I love that I don't have to change sniffle_hw.py anymore!