| Age | Commit message (Collapse) | Author |
|---|
|
There are the "traffic inspecting" wait() conditions, which check an
edge to find the start of START, then wait for sample points to grab the
bit values. Bit times are sampled in their respective center, potential
glitches around sample points get ignored.
Add another independent set of wait() conditions which check _all_ edges
regardless of any data communication. This results in the most reliable
and maintainable detection of break conditions, regardless of how they
align to data frames. Break is defined as a period of low input signal
which spans at least one frame's length. Run the edge inspection after
data inspection, which results in the most appropriate annotation output
like leading data bits (of incomplete frames), frame errors (violated
STOP bit expectations), then break conditions. This approach is most
robust in the presence of incomplete input streams.
|
|
Use the already available .databits[] information which holds sample data
and bit time edge positions, and the common bitpack() routine. This shall
increase readability of the bits to value conversion.
[ best viewed with more context, like 'git diff -U5' ]
|
|
Move initialization code of protocol decoders from the constructor to a
new reset() helper method. The libsigrokdecode backend could run this
method several times to clear the decoder's internal state, before new
data from another acquisition gets fed to decode() calls.
|
|
|
|
Slightly rearrange some of the methods which are involved in UART frame
inspection. Use a consistent sequence of steps: Grab the signal's
current value, accumulate and process the information, emit respective
annotations, and advance to the next stage in the UART frame inspection.
|
|
After the decode() method got adjusted to call wait() with custom made
conditions and to check .matched[] before inspecting samples, the check
whether a bit time's sample point was reached has become obsolete.
|
|
Since either of the UART signals (RX, TX) is optional, and in the
absence of Decoder.wait() conditions that "will never match", we cannot
construct a constant layout. Instead we need to explicitly keep track of
which item in the list of wait conditions corresponds to which signal.
Once the index in the list of wait conditions is known, inspection of
samples can depend on the Decoder.matched[] attribute. Before this
change, redundant reached_bit() checks kept us from processing samples
that should not have been inspected. Tests pass before and after this
very commit.
|
|
Factor out the logic which inspects samples that were provided by the
PD version 3 query API, and dispatches their processing depending on
the progress of UART frame inspection. "Unroll" a loop over the RX and
TX signals.
This commit replaces some complicated variable assignments by easier to
verify invocations.
|
|
Adjust the UART protocol decoder, to make use of the query based API.
Have edges detected and unrelated samples skipped by common code.
This implementation keeps some redundancy in place (like checking for
having reached specific sample numbers, while the backend managed that
for us). This approach reduces the diff and shall simplify review.
Only some common checks in decode() were moved to the start of the
routine, outside of the sample inspection loop.
|
|
Rephrase the bit slot index calculation for UART frames such that it
becomes more apparent whether a start bit is involved or whether an
array index needs adjustment due to Python range semantics.
This shall improve readability, and reduce the probability of off-by-one
errors during maintenance.
|
|
When the UART frame does not contain a parity bit, then immediately
advance to reception of stop bits after all data bits were received.
This eliminates the necessity to run the parity check routine when
parity does not apply in the first place. Without this change, some
"dummy" sample needs to get inspected for correct operation of the
state machine.
|
|
|
|
Remove the FSF postal address as it might change (it did in the past).
Reference the gnu.org website instead which is more stable.
|
|
This is in almost all cases what the user will want, only rarely ASCII
(the old default) will be the more natural fit.
|
|
For 5..8 data bits the binary output will be 1 byte, for 9 data bits
it will be 2 bytes (big-endian).
This fixes bug #708.
|
|
When the start bit is not low at its sample point, then stop trying
to interpret the remaining frame -- it's already known to be invalid,
anyway.
Wait for the next start bit instead, assuming that either the falling
edge which started the inspection of the UART frame and its start bit
was a spurious glitch or that the captured signal does not communicate
at the decoder's configured bitrate.
Signed-off-by: Gerhard Sittig <gerhard.sittig@gmx.net>
|
|
Factor out the code which generates a textual representation for the
numeric values that were communicated via UART bit patterns. Make the
width of the output text depend on the number of bits in the UART frame
(five to nine) instead of assuming bytes of exactly eight bits.
Fix other minor issues while we are here: Nine bits result in a number
range of 0 to 511 (not 512). ASCII codes 30 and 31 are non-printables.
The previous implementation skipped a significant leading digit in the
octal representation.
Signed-off-by: Gerhard Sittig <gerhard.sittig@gmx.net>
|
|
Given the generic nature of UART communication and the supported range
for the data width, "byte" may be a misleading name for the numeric
value that gets communicated in five to nine data bits. Rename the
"databyte" variable to "datavalue".
Signed-off-by: Gerhard Sittig <gerhard.sittig@gmx.net>
|
|
The UART bit information was not transmitted correctly to stacked PDs
if there was an overlap between RX and TX bytes in the data.
|
|
The previous **kwargs some PDs had is not actually ever used, so drop it.
|
|
Re-enable the fast path for identical samples but only when both
pins are waiting for the start bit. For sparse data sets (I tested
UT61E capture log) the optimization results in a >4x decode
improvement.
|
|
|
|
This is more consistent with annotation syntax and looks slightly
better in most cases.
|
|
1. Show Frame Error on the Start bit
2. Don't overwrite framing errors with (valid) start/stop bit info
|
|
At 3 samples per bit, the uart decoder took the value at the last sample
instead of the middle one. Improve calculations so that sampling is more
accurate at odd number of samples per bit.
|
|
|
|
This will allow for much simpler code in stacked PDs.
Adapt stacked PDs to new API.
|
|
|
|
Rename the old MissingDataError to the clearer ChannelError. Also, add
ChannelError in the UART decoder.
|
|
- No newlines at the end of files.
- No trailing ';' characters.
- Comparison with None: Use 'is None' or 'is not None'.
- Comparison with True/False: Use 'if cond:' or 'if not cond:'.
- Various minor whitespace fixes.
|
|
For options which only have a limited set of valid values, we don't need
to check (in the PD) whether a valid value was supplied, since the backend
can do that for us.
|
|
Also, use the "if not self.samplerate" form, which catches both the case
where self.samplerate is None, as well as the case where it is 0.
|
|
In all current PDs it is not necessary to raise an exception upon
invalid states (of the PD's state machine), since we can guarantee that
no such invalid state can ever be reached in these PDs.
|
|
|
|
Older libsigrokdecode versions are no longer able to use the current
versions of the PDs (various changes in syntax etc).
|
|
Variables of type 'struct srd_channel *' are consistently named 'pdch' to
make them easily distinguishable from libsigrok's 'struct sr_channel *'
variables that are consistently named 'ch'.
|
|
Annotation entries also consist of a tuple, not a list.
|
|
|
|
Each option consists of a dictionary with the following keys:
id The option id, which is passed in when setting a value.
desc A description of the option, suitable for display.
def The default value for this option.
values (optional) If present, a tuple containing values the option
may take. They must be of the same type as the default.
Valid types for the options are UTF-8-encoded strings, integers, and
floating point values.
|
|
|
|
|
|
Also, provide all the required annotation classes for this to work
properly.
|
|
|
|
The output type is now called OUTPUT_PYTHON, adapt all PDs to that.
|
|
This is a temporary thing, later there'll be some facility to let
frontends handle any annotations marked as "this is a number" (as opposed
to "this is a string") in a generic manner and display them in any
supported (by that frontend) format, e.g. ascii, hex, oct, decimal,
binary, big-endian vs. little-endian, and so on.
This is a fix related to #201.
|
|
|
|
The single comment re-stating the PD's name / description / purpose in
each pd.py file is not really needed, that info is available in the
Decoder class' attributes already.
|
|
The functionality of the preliminary 'uart_dump' PD is now available
in the proper 'uart' PD, via the OUTPUT_BINARY mechanism that frontends
can use to dump decoded data (in various formats) to a file, or pipe it
into other applications, and so on.
Old sigrok-cli example usage:
$ sigrok-cli -i foo.sr -P uart:rx=0:tx=1,uart_dump:filename=bootlog.txt
New sigrok-cli example usage:
$ sigrok-cli -i foo.sr -P uart:rx=0:tx=1 -B uart=rxtx > bootlog.txt
New sigrok-cli example usage (piping into other applications):
$ sigrok-cli -i foo.sr -P uart:rx=0:tx=1 -B uart=rxtx | grep "whatever"
|
|
We use the [XX] notation for non-printable characters, which is what
various other logic analyzer software packages do too, e.g. the
CWAV USBee Suite.
This fixes bug #201.
|
|
|
