sys/entropy_source: add new module for entropy sources

[PeterKietzmann]

Contribution description

This PR adds an entropy_source module which acts as a place to implement different entropy sources as submodules. Currently, an ADC based source and a zero entropy source (for testing only!) are implemented. Additionally, a common component provides functionality that is usable for all. Sources can optionally enable health tests (executed during runtime) and output conditioning.

Output values can be used directly to create low-requirement seeds, or they can be used to feed more advanced "high level" entropy handlers that accumulate values of multiple entropy sources. More information can be found in the module documentation.

I have an other entropy source lying around that bases on jitters between differently clocked timers, that I will PR afterwards. As an immediate next step, I will provide a PR that adds an advanced entropy module which deals with multiple sources and can be used for cryptographic purposes (seed generation).

Open Aspects

I marked the PR as WIP because some aspects are not yet decided and implemented:

• Expose configuration options to kconfig.
• Do we want to run startup tests on module initialization?
• Do we need to return the "estimated" amount of entropy per request? The estimate would fully base on the parameters given by configuration.
• Do we want a more advanced conditioning component next to the van Neumann extractor? An alternative would be a cryptographic hash function. This might unnecessarily bloat the module for non-crypto purposes and it might be duplicate when used with cryptographic "high level" entropy implementations that include cryptographic derivation functions anyway.

Testing procedure

• Read and understand the high level documentation.
• make doc build the doxygen documentation.
• Execute the provided test on a BOARD and on native and compare the results with the expected results in the README.
• Play around with the default configuration values here and here. Re-run the test. For the ADC source, connect an exposed ADC pin to Vcc or GND and see how the error messages behave. Eventually increase the number of dumped samples, see how the output values are affected. With conditioning enabled, see how the measured runtime is affected.

Issues/PRs references

#3425, #4294, #5153

[miri64]

Seems similar to what I tried to do in #5153. Might be beneficial to have a look at @d00616's comments there.

[miri64]

Is there planned to be a way to mix entropy sources to increase the entropy? Thinking about stuff like timer (#3425), register, stack, and radio states here.

[kaspar030]

There's also #4294, whose discussion might be useful.

[PeterKietzmann]

@miri64, @kaspar030 thanks for the pointers. I've added the PRs/issues as references in the above description and I've double checked the discussions. As far as I see, all concerns that can be covered, are covered by this PR. Did you have a specific aspect in mind?

Some more comments:

• This module only provides access to different entropy sources. How PRNG seeds are generated is not in the scope of this PR and will follow afterwards.
• Mixing different entropy sources is of particular interest in crypto contexts and I will provide a solution for that. Other (lightweight) "mixers" might be implemented next to it, if needed.
• An event based entropy collector (sys: random: add entropy collector #4294) could be one other entropy source.

[miri64]

[…] Did you have a specific aspect in mind?

No, mainly wanted to point out past discussions so the insights from them don't get lost as you did not reference them before.

[leandrolanzieri]

Here are some initial comments

[PeterKietzmann]

@leandrolanzieri many thanks for your review! Please have a look at my latest commits.

[PeterKietzmann]

@leandrolanzieri I have added some more changes (as discussed offline). Would you take an other look?

[leandrolanzieri]

This is not a boolean symbol:

Suggested change

	#if IS_ACTIVE(CONFIG_ENTROPY_SOURCE_ADC_LINE)
	#if defined(CONFIG_ENTROPY_SOURCE_ADC_LINE)

[leandrolanzieri]

This one has not been addressed yet

[PeterKietzmann]

@leandrolanzieri thanks again! Please have a look at the last commits.

[PeterKietzmann]

@leandrolanzieri I've pushed a commit addressing some minor leftovers. May I squash to clean up a bit?

[leandrolanzieri]

May I squash to clean up a bit?

Go ahead

[PeterKietzmann]

It is not accurate. (i) It summarizes the methods described in [4] (14 pages research paper) and [5] (27 pages educational book chapter) in 4 short paragraphs. (ii) They do not even consider the problem we are discussing here: where does initial entropy come from and how to estimate the amount of contained entropy bits. The pure fact that you distribute entropy over pools does not relive the need for initial entropy at all.

[maribu]

The pure fact that you distribute entropy over pools does not relive the need for initial entropy at all.

When constantly reseeding/rekeying, it does guarantee that eventually a reseed/rekeying will happen that for any given adversary contains enough entropy to counter the attack. This is one of the few approaches where you actually get a guarantee without any dubious assumptions.

But as I said, this approach will not work for long lived key material. (Said so two times before...)

[PeterKietzmann]

Why the plethora of elaborations? Regarding the API, let's wait for some more opinions. Further, I understand:

• You cannot provide a runtime estimator.
• You are bothered by the min. entropy metric. Maybe I didn't make it clear, but actually the used entropy metric refers to the minimum of all entropy estimates performed by NIST 800-90B. I can try to precise it in the documentation and macro name but this won't change the implementation.

All considerations for the accumulator should be made there while a single entropy source should provide an estimate about its contributed entropy per sample.

[tcschmidt]

This discussion seems to have gotten lost in personal opinions and scientifically unjustified claims. I suggest that we stick to the facts in this highly rigorous regime of entropy generation. Hence I propose that arguments accounted as valid remain restricted to rigorous (peer-reviewed) scientific sources. This means: No accountance of personal feelings nor of Wikipedia.

[PeterKietzmann]

@maribu, @leandrolanzieri thanks for bringing this up again. I've addressed your comments, rebased to master and added minor cleanups. Please have a look at my latest commits.

[PeterKietzmann]

Anything missing? May I squash?

[leandrolanzieri]

Anything missing? May I squash?

From my side all comments were addressed

[maribu]

Anything missing?

My comments are all addressed.

May I squash?

Sure.

[PeterKietzmann]

Squashed. CI has passed.

[miri64]

Ping @maribu @leandrolanzieri?

[leandrolanzieri]

Just a minor comment on configuration I found while testing. Please squash directly @PeterKietzmann. Otherwise I've run the test application both on native and on a nucleo-f401re and the output looks correct according to the instructions. I tested the effects of the conditioning when the line is set to fixed voltages and floating.

Keep in mind that the results are platform-specific.

ADC pin floating with conditioning and health tests

2020-12-01 14:52:57,597 # ADC noise source entropy/sample: 65536 [2^16 * bit / sample]
2020-12-01 14:52:57,600 # ADC noise source entropy/sample: 1 [bit / sample]
2020-12-01 14:52:57,601 # 75
2020-12-01 14:52:57,602 # 47
2020-12-01 14:52:57,602 # 6c
2020-12-01 14:52:57,602 # 81
2020-12-01 14:52:57,602 # 0b
2020-12-01 14:52:57,603 # 99
2020-12-01 14:52:57,603 # b6
2020-12-01 14:52:57,604 # 76
2020-12-01 14:52:57,604 # 82
2020-12-01 14:52:57,604 # 92
2020-12-01 14:52:57,605 # 01
2020-12-01 14:52:57,605 # c8
2020-12-01 14:52:57,605 # b5
2020-12-01 14:52:57,606 # 61
2020-12-01 14:52:57,606 # db
2020-12-01 14:52:57,607 # 78
2020-12-01 14:52:57,607 # d6
2020-12-01 14:52:57,608 # 2d
2020-12-01 14:52:57,608 # d2
2020-12-01 14:52:57,608 # b1
2020-12-01 14:52:57,609 # e8
2020-12-01 14:52:57,610 # 94
2020-12-01 14:52:57,610 # f8
2020-12-01 14:52:57,611 # 87
2020-12-01 14:52:57,611 # a9
2020-12-01 14:52:57,611 # 78
2020-12-01 14:52:57,611 # 0b
2020-12-01 14:52:57,611 # b3
2020-12-01 14:52:57,612 # 35
2020-12-01 14:52:57,612 # da
2020-12-01 14:52:57,613 # 27
2020-12-01 14:52:57,614 # be
2020-12-01 14:52:57,615 # ea
2020-12-01 14:52:57,616 # 75
2020-12-01 14:52:57,616 # c2
2020-12-01 14:52:57,617 # 4d
2020-12-01 14:52:57,618 # 8f
2020-12-01 14:52:57,618 # 58
2020-12-01 14:52:57,619 # 94
2020-12-01 14:52:57,619 # 0c
2020-12-01 14:52:57,619 # b0
2020-12-01 14:52:57,619 # 81
2020-12-01 14:52:57,619 # c4
2020-12-01 14:52:57,620 # 78
2020-12-01 14:52:57,620 # 6a
2020-12-01 14:52:57,620 # 92
2020-12-01 14:52:57,620 # e0
2020-12-01 14:52:57,620 # b6
2020-12-01 14:52:57,620 # d8
2020-12-01 14:52:57,620 # 6d
2020-12-01 14:52:57,621 # 1e
2020-12-01 14:52:57,621 # 4a
2020-12-01 14:52:57,621 # a0
2020-12-01 14:52:57,621 # cc
2020-12-01 14:52:57,622 # e0
2020-12-01 14:52:57,622 # 4f
2020-12-01 14:52:57,623 # d6
2020-12-01 14:52:57,623 # 2b
2020-12-01 14:52:57,623 # 90
2020-12-01 14:52:57,624 # 32
2020-12-01 14:52:57,624 # f0
2020-12-01 14:52:57,625 # 92
2020-12-01 14:52:57,625 # 2f
2020-12-01 14:52:57,625 # 69
2020-12-01 14:52:57,630 # 1e
2020-12-01 14:52:57,631 # bc
2020-12-01 14:52:57,632 # d3
2020-12-01 14:52:57,632 # 64
2020-12-01 14:52:57,633 # 63
2020-12-01 14:52:57,633 # b9
2020-12-01 14:52:57,633 # 70
2020-12-01 14:52:57,634 # a6
2020-12-01 14:52:57,634 # 99
2020-12-01 14:52:57,634 # 47
2020-12-01 14:52:57,635 # b8
2020-12-01 14:52:57,635 # e3
2020-12-01 14:52:57,635 # 2a
2020-12-01 14:52:57,635 # b2
2020-12-01 14:52:57,635 # 0d
2020-12-01 14:52:57,635 # 97
2020-12-01 14:52:57,636 # da
2020-12-01 14:52:57,636 # 82
2020-12-01 14:52:57,636 # 5e
2020-12-01 14:52:57,636 # d4
2020-12-01 14:52:57,636 # 6e
2020-12-01 14:52:57,636 # d4
2020-12-01 14:52:57,637 # cc
2020-12-01 14:52:57,637 # d0
2020-12-01 14:52:57,637 # 2d
2020-12-01 14:52:57,637 # 9b
2020-12-01 14:52:57,637 # 34
2020-12-01 14:52:57,637 # e2
2020-12-01 14:52:57,638 # 2d
2020-12-01 14:52:57,638 # 7c
2020-12-01 14:52:57,638 # c1
2020-12-01 14:52:57,638 # 4e
2020-12-01 14:52:57,639 # c9
2020-12-01 14:52:57,639 # 66
2020-12-01 14:52:57,639 # ba
2020-12-01 14:52:57,640 # a4
2020-12-01 14:52:57,640 # ca
2020-12-01 14:52:57,640 # 07
2020-12-01 14:52:57,640 # 5b
2020-12-01 14:52:57,641 # e8
2020-12-01 14:52:57,641 # 4b
2020-12-01 14:52:57,641 # 0d
2020-12-01 14:52:57,641 # 3e
2020-12-01 14:52:57,642 # 59
2020-12-01 14:52:57,642 # 6b
2020-12-01 14:52:57,642 # 70
2020-12-01 14:52:57,643 # 2d
2020-12-01 14:52:57,643 # 72
2020-12-01 14:52:57,644 # e1
2020-12-01 14:52:57,644 # 78
2020-12-01 14:52:57,644 # 78
2020-12-01 14:52:57,644 # 4a
2020-12-01 14:52:57,644 # e9
2020-12-01 14:52:57,645 # a2
2020-12-01 14:52:57,645 # 57
2020-12-01 14:52:57,645 # ac
2020-12-01 14:52:57,645 # f8
2020-12-01 14:52:57,646 # 46
2020-12-01 14:52:57,647 # 64
2020-12-01 14:52:57,648 # 6b
2020-12-01 14:52:57,648 # 17
2020-12-01 14:52:57,648 # 86
2020-12-01 14:52:57,648 # 53
2020-12-01 14:52:57,649 # 1f
2020-12-01 14:52:57,652 # ADC noise request 64 Bytes returned: 0. Time: 4123us

When the pin is connected to GND we can observe some health test errors in the sampling and an increment in the time needed to gather an array of samples (thus also it can be seen that time is variable in this configuration):

ADC pin to GND with conditioning and health tests

2020-12-01 14:55:34,998 # ADC noise source entropy/sample: 65536 [2^16 * bit / sample]
2020-12-01 14:55:35,003 # ADC noise source entropy/sample: 1 [bit / sample]
2020-12-01 14:55:35,003 # c9
2020-12-01 14:55:35,004 # 69
2020-12-01 14:55:35,004 # 98
2020-12-01 14:55:35,005 # e9
2020-12-01 14:55:35,005 # 52
2020-12-01 14:55:35,006 # a6
2020-12-01 14:55:35,007 # 32
2020-12-01 14:55:35,007 # 1c
2020-12-01 14:55:35,007 # 3d
2020-12-01 14:55:35,007 # ab
2020-12-01 14:55:35,007 # 74
2020-12-01 14:55:35,008 # 74
2020-12-01 14:55:35,008 # e2
2020-12-01 14:55:35,010 # c5
2020-12-01 14:55:35,011 # 5c
2020-12-01 14:55:35,011 # d6
2020-12-01 14:55:35,012 # c9
2020-12-01 14:55:35,012 # 6c
2020-12-01 14:55:35,012 # b1
2020-12-01 14:55:35,012 # 8e
2020-12-01 14:55:35,012 # e5
2020-12-01 14:55:35,012 # cd
2020-12-01 14:55:35,013 # 74
2020-12-01 14:55:35,014 # 9a
2020-12-01 14:55:35,014 # 52
2020-12-01 14:55:35,014 # 73
2020-12-01 14:55:35,014 # eb
2020-12-01 14:55:35,015 # e6
2020-12-01 14:55:35,016 # 55
2020-12-01 14:55:35,019 # ADC noise single request 29/64: -6
2020-12-01 14:55:35,019 # 2c
2020-12-01 14:55:35,020 # 85
2020-12-01 14:55:35,020 # 8b
2020-12-01 14:55:35,020 # 72
2020-12-01 14:55:35,021 # 62
2020-12-01 14:55:35,021 # d6
2020-12-01 14:55:35,021 # 2d
2020-12-01 14:55:35,022 # e0
2020-12-01 14:55:35,022 # a6
2020-12-01 14:55:35,024 # 15
2020-12-01 14:55:35,024 # 4d
2020-12-01 14:55:35,024 # 86
2020-12-01 14:55:35,024 # 92
2020-12-01 14:55:35,024 # e6
2020-12-01 14:55:35,025 # 4d
2020-12-01 14:55:35,025 # a4
2020-12-01 14:55:35,026 # c3
2020-12-01 14:55:35,027 # e0
2020-12-01 14:55:35,028 # e5
2020-12-01 14:55:35,028 # 70
2020-12-01 14:55:35,029 # 56
2020-12-01 14:55:35,029 # e6
2020-12-01 14:55:35,032 # ADC noise single request 52/64: -6
2020-12-01 14:55:35,032 # 9a
2020-12-01 14:55:35,035 # ADC noise single request 54/64: -6
2020-12-01 14:55:35,035 # 38
2020-12-01 14:55:35,035 # 22
2020-12-01 14:55:35,036 # c6
2020-12-01 14:55:35,036 # 3a
2020-12-01 14:55:35,037 # 78
2020-12-01 14:55:35,037 # 38
2020-12-01 14:55:35,038 # 98
2020-12-01 14:55:35,041 # ADC noise single request 62/64: -6
2020-12-01 14:55:35,042 # ad
2020-12-01 14:55:35,048 # d3
2020-12-01 14:55:35,048 # 58
2020-12-01 14:55:35,049 # 66
2020-12-01 14:55:35,049 # 2e
2020-12-01 14:55:35,049 # 59
2020-12-01 14:55:35,049 # b6
2020-12-01 14:55:35,051 # e4
2020-12-01 14:55:35,051 # 66
2020-12-01 14:55:35,051 # a2
2020-12-01 14:55:35,052 # 57
2020-12-01 14:55:35,052 # 54
2020-12-01 14:55:35,052 # 57
2020-12-01 14:55:35,052 # e2
2020-12-01 14:55:35,052 # 2a
2020-12-01 14:55:35,052 # 9e
2020-12-01 14:55:35,052 # 85
2020-12-01 14:55:35,052 # 1b
2020-12-01 14:55:35,053 # 4e
2020-12-01 14:55:35,053 # d3
2020-12-01 14:55:35,053 # aa
2020-12-01 14:55:35,054 # e8
2020-12-01 14:55:35,054 # 91
2020-12-01 14:55:35,054 # ea
2020-12-01 14:55:35,054 # 23
2020-12-01 14:55:35,055 # 3d
2020-12-01 14:55:35,056 # b2
2020-12-01 14:55:35,056 # e1
2020-12-01 14:55:35,057 # 8c
2020-12-01 14:55:35,057 # 63
2020-12-01 14:55:35,057 # 53
2020-12-01 14:55:35,057 # 79
2020-12-01 14:55:35,058 # 1a
2020-12-01 14:55:35,058 # 3a
2020-12-01 14:55:35,058 # f5
2020-12-01 14:55:35,058 # b0
2020-12-01 14:55:35,058 # 5c
2020-12-01 14:55:35,058 # 69
2020-12-01 14:55:35,058 # c3
2020-12-01 14:55:35,059 # d1
2020-12-01 14:55:35,059 # d4
2020-12-01 14:55:35,059 # d2
2020-12-01 14:55:35,059 # e1
2020-12-01 14:55:35,060 # d1
2020-12-01 14:55:35,061 # 68
2020-12-01 14:55:35,062 # 2e
2020-12-01 14:55:35,062 # ad
2020-12-01 14:55:35,062 # 74
2020-12-01 14:55:35,062 # 2c
2020-12-01 14:55:35,062 # 39
2020-12-01 14:55:35,062 # 55
2020-12-01 14:55:35,062 # 39
2020-12-01 14:55:35,063 # 59
2020-12-01 14:55:35,063 # 53
2020-12-01 14:55:35,063 # e3
2020-12-01 14:55:35,064 # b4
2020-12-01 14:55:35,064 # 0b
2020-12-01 14:55:35,064 # ce
2020-12-01 14:55:35,065 # aa
2020-12-01 14:55:35,065 # 94
2020-12-01 14:55:35,065 # 2f
2020-12-01 14:55:35,065 # 55
2020-12-01 14:55:35,065 # 51
2020-12-01 14:55:35,065 # b3
2020-12-01 14:55:35,066 # e8
2020-12-01 14:55:35,070 # ADC noise request 64 Bytes returned: 0. Time: 6389us

Connecting the ADC to 3.3 V shows some health test errors and getting multiple samples at the end also fails due to the repetition health test failing:

ADC pin to 3.3 V with conditioning and health tests

2020-12-01 15:33:51,585 # ADC noise source entropy/sample: 1 [2^16 * bit / sample]
2020-12-01 15:33:51,590 # ADC noise source entropy/sample: 0.000015 [bit / sample]
2020-12-01 15:33:51,591 # ad
2020-12-01 15:33:51,591 # 63
2020-12-01 15:33:51,592 # 06
2020-12-01 15:33:51,593 # 8b
2020-12-01 15:33:51,594 # 9d
2020-12-01 15:33:51,594 # 09
2020-12-01 15:33:51,594 # 87
2020-12-01 15:33:51,595 # b2
2020-12-01 15:33:51,595 # 7c
2020-12-01 15:33:51,595 # 07
2020-12-01 15:33:51,596 # 1d
2020-12-01 15:33:51,596 # 9b
2020-12-01 15:33:51,596 # 2d
2020-12-01 15:33:51,596 # e2
2020-12-01 15:33:51,596 # 34
2020-12-01 15:33:51,597 # d5
2020-12-01 15:33:51,598 # c1
2020-12-01 15:33:51,598 # 23
2020-12-01 15:33:51,598 # 2e
2020-12-01 15:33:51,598 # 54
2020-12-01 15:33:51,599 # b8
2020-12-01 15:33:51,599 # 71
2020-12-01 15:33:51,600 # 3c
2020-12-01 15:33:51,600 # dc
2020-12-01 15:33:51,600 # e3
2020-12-01 15:33:51,601 # ec
2020-12-01 15:33:51,601 # d2
2020-12-01 15:33:51,602 # e1
2020-12-01 15:33:51,602 # 86
2020-12-01 15:33:51,603 # db
2020-12-01 15:33:51,603 # 2a
2020-12-01 15:33:51,603 # 8e
2020-12-01 15:33:51,604 # 5c
2020-12-01 15:33:51,604 # 78
2020-12-01 15:33:51,604 # d0
2020-12-01 15:33:51,605 # 3e
2020-12-01 15:33:51,606 # 51
2020-12-01 15:33:51,606 # 5a
2020-12-01 15:33:51,606 # a3
2020-12-01 15:33:51,607 # c2
2020-12-01 15:33:51,607 # 93
2020-12-01 15:33:51,607 # 41
2020-12-01 15:33:51,608 # cb
2020-12-01 15:33:51,608 # c5
2020-12-01 15:33:51,608 # 78
2020-12-01 15:33:51,609 # aa
2020-12-01 15:33:51,610 # f3
2020-12-01 15:33:51,611 # d2
2020-12-01 15:33:51,611 # f1
2020-12-01 15:33:51,611 # 29
2020-12-01 15:33:51,611 # 4b
2020-12-01 15:33:51,612 # 06
2020-12-01 15:33:51,615 # ADC noise single request 52/64: -3
2020-12-01 15:33:51,616 # f0
2020-12-01 15:33:51,616 # ba
2020-12-01 15:33:51,616 # aa
2020-12-01 15:33:51,616 # 2d
2020-12-01 15:33:51,617 # 4c
2020-12-01 15:33:51,617 # 74
2020-12-01 15:33:51,617 # 73
2020-12-01 15:33:51,621 # ADC noise single request 60/64: -3
2020-12-01 15:33:51,621 # 8a
2020-12-01 15:33:51,621 # 9f
2020-12-01 15:33:51,625 # ADC noise single request 63/64: -3
2020-12-01 15:33:51,630 # a3
2020-12-01 15:33:51,631 # 0d
2020-12-01 15:33:51,631 # 7a
2020-12-01 15:33:51,631 # ea
2020-12-01 15:33:51,632 # 58
2020-12-01 15:33:51,632 # b1
2020-12-01 15:33:51,632 # c3
2020-12-01 15:33:51,632 # 0b
2020-12-01 15:33:51,633 # 1d
2020-12-01 15:33:51,633 # 4f
2020-12-01 15:33:51,633 # b4
2020-12-01 15:33:51,633 # 65
2020-12-01 15:33:51,634 # c5
2020-12-01 15:33:51,634 # a3
2020-12-01 15:33:51,634 # 4a
2020-12-01 15:33:51,635 # 8f
2020-12-01 15:33:51,636 # b4
2020-12-01 15:33:51,636 # 61
2020-12-01 15:33:51,636 # c7
2020-12-01 15:33:51,637 # 1c
2020-12-01 15:33:51,637 # a6
2020-12-01 15:33:51,637 # f1
2020-12-01 15:33:51,637 # c9
2020-12-01 15:33:51,637 # 0e
2020-12-01 15:33:51,637 # 2e
2020-12-01 15:33:51,637 # d5
2020-12-01 15:33:51,638 # 95
2020-12-01 15:33:51,638 # 65
2020-12-01 15:33:51,639 # 35
2020-12-01 15:33:51,640 # 9a
2020-12-01 15:33:51,640 # 16
2020-12-01 15:33:51,640 # dc
2020-12-01 15:33:51,640 # aa
2020-12-01 15:33:51,640 # d8
2020-12-01 15:33:51,640 # 95
2020-12-01 15:33:51,640 # d2
2020-12-01 15:33:51,640 # 4d
2020-12-01 15:33:51,641 # b4
2020-12-01 15:33:51,641 # 36
2020-12-01 15:33:51,641 # f4
2020-12-01 15:33:51,641 # 70
2020-12-01 15:33:51,642 # 4c
2020-12-01 15:33:51,643 # 95
2020-12-01 15:33:51,644 # 9e
2020-12-01 15:33:51,644 # 45
2020-12-01 15:33:51,644 # b5
2020-12-01 15:33:51,644 # a2
2020-12-01 15:33:51,644 # 97
2020-12-01 15:33:51,645 # e3
2020-12-01 15:33:51,645 # 98
2020-12-01 15:33:51,645 # 72
2020-12-01 15:33:51,645 # 2c
2020-12-01 15:33:51,645 # e7
2020-12-01 15:33:51,645 # c8
2020-12-01 15:33:51,646 # c3
2020-12-01 15:33:51,646 # 15
2020-12-01 15:33:51,646 # f2
2020-12-01 15:33:51,647 # 65
2020-12-01 15:33:51,647 # d1
2020-12-01 15:33:51,647 # 15
2020-12-01 15:33:51,647 # c7
2020-12-01 15:33:51,647 # e2
2020-12-01 15:33:51,648 # f1
2020-12-01 15:33:51,648 # 64
2020-12-01 15:33:51,653 # ADC noise request 64 Bytes returned: -3. Time: 5387us

When the conditioning is removed the time to gather the array is reduced (and constant across multiple runs) but we can see that values have bias:

ADC pin to GND without conditioning and health tests

2020-12-01 15:01:34,825 # ADC noise source entropy/sample: 1 [2^16 * bit / sample]
2020-12-01 15:01:34,830 # ADC noise source entropy/sample: 0.000015 [bit / sample]
2020-12-01 15:01:34,831 # 96
2020-12-01 15:01:34,831 # 08
2020-12-01 15:01:34,831 # 00
2020-12-01 15:01:34,831 # 28
2020-12-01 15:01:34,832 # 49
2020-12-01 15:01:34,832 # 85
2020-12-01 15:01:34,832 # 85
2020-12-01 15:01:34,833 # 10
2020-12-01 15:01:34,833 # 00
2020-12-01 15:01:34,833 # 02
2020-12-01 15:01:34,834 # 90
2020-12-01 15:01:34,834 # 02
2020-12-01 15:01:34,834 # 00
2020-12-01 15:01:34,835 # 01
2020-12-01 15:01:34,835 # 81
2020-12-01 15:01:34,835 # 14
2020-12-01 15:01:34,836 # 20
2020-12-01 15:01:34,836 # 81
2020-12-01 15:01:34,836 # 52
2020-12-01 15:01:34,837 # a9
2020-12-01 15:01:34,837 # 20
2020-12-01 15:01:34,838 # a2
2020-12-01 15:01:34,838 # 42
2020-12-01 15:01:34,838 # 80
2020-12-01 15:01:34,838 # 15
2020-12-01 15:01:34,838 # 08
2020-12-01 15:01:34,839 # 49
2020-12-01 15:01:34,839 # c2
2020-12-01 15:01:34,839 # 80
2020-12-01 15:01:34,840 # 04
2020-12-01 15:01:34,841 # 02
2020-12-01 15:01:34,841 # 0c
2020-12-01 15:01:34,841 # 59
2020-12-01 15:01:34,842 # 08
2020-12-01 15:01:34,842 # 81
2020-12-01 15:01:34,843 # 90
2020-12-01 15:01:34,843 # a0
2020-12-01 15:01:34,843 # 22
2020-12-01 15:01:34,843 # 10
2020-12-01 15:01:34,843 # 22
2020-12-01 15:01:34,843 # 85
2020-12-01 15:01:34,844 # 04
2020-12-01 15:01:34,844 # 61
2020-12-01 15:01:34,844 # 05
2020-12-01 15:01:34,844 # 0a
2020-12-01 15:01:34,845 # 02
2020-12-01 15:01:34,846 # 85
2020-12-01 15:01:34,846 # 00
2020-12-01 15:01:34,846 # 06
2020-12-01 15:01:34,846 # 92
2020-12-01 15:01:34,846 # 10
2020-12-01 15:01:34,846 # 20
2020-12-01 15:01:34,847 # 50
2020-12-01 15:01:34,847 # 34
2020-12-01 15:01:34,847 # 40
2020-12-01 15:01:34,848 # 44
2020-12-01 15:01:34,848 # e5
2020-12-01 15:01:34,848 # 80
2020-12-01 15:01:34,849 # 02
2020-12-01 15:01:34,849 # 00
2020-12-01 15:01:34,849 # 90
2020-12-01 15:01:34,850 # 09
2020-12-01 15:01:34,850 # 90
2020-12-01 15:01:34,850 # 50
2020-12-01 15:01:34,852 # 4a
2020-12-01 15:01:34,854 # 00
2020-12-01 15:01:34,855 # 14
2020-12-01 15:01:34,855 # 44
2020-12-01 15:01:34,855 # 00
2020-12-01 15:01:34,856 # 20
2020-12-01 15:01:34,856 # 02
2020-12-01 15:01:34,856 # 22
2020-12-01 15:01:34,856 # 48
2020-12-01 15:01:34,857 # 00
2020-12-01 15:01:34,858 # 20
2020-12-01 15:01:34,859 # 8a
2020-12-01 15:01:34,859 # 22
2020-12-01 15:01:34,860 # 88
2020-12-01 15:01:34,861 # 05
2020-12-01 15:01:34,861 # 05
2020-12-01 15:01:34,862 # 04
2020-12-01 15:01:34,862 # 26
2020-12-01 15:01:34,862 # 08
2020-12-01 15:01:34,863 # 10
2020-12-01 15:01:34,863 # 22
2020-12-01 15:01:34,863 # 48
2020-12-01 15:01:34,863 # 56
2020-12-01 15:01:34,863 # a8
2020-12-01 15:01:34,864 # 41
2020-12-01 15:01:34,864 # 02
2020-12-01 15:01:34,864 # 00
2020-12-01 15:01:34,864 # 08
2020-12-01 15:01:34,864 # 0a
2020-12-01 15:01:34,865 # 01
2020-12-01 15:01:34,865 # 85
2020-12-01 15:01:34,865 # 05
2020-12-01 15:01:34,865 # 18
2020-12-01 15:01:34,865 # 54
2020-12-01 15:01:34,865 # 00
2020-12-01 15:01:34,866 # 44
2020-12-01 15:01:34,866 # 40
2020-12-01 15:01:34,866 # 08
2020-12-01 15:01:34,866 # 20
2020-12-01 15:01:34,866 # 08
2020-12-01 15:01:34,866 # 20
2020-12-01 15:01:34,867 # 40
2020-12-01 15:01:34,867 # 00
2020-12-01 15:01:34,867 # 10
2020-12-01 15:01:34,867 # 03
2020-12-01 15:01:34,867 # 04
2020-12-01 15:01:34,867 # 49
2020-12-01 15:01:34,867 # 00
2020-12-01 15:01:34,868 # 0d
2020-12-01 15:01:34,868 # 48
2020-12-01 15:01:34,868 # 80
2020-12-01 15:01:34,868 # 31
2020-12-01 15:01:34,868 # 80
2020-12-01 15:01:34,868 # 41
2020-12-01 15:01:34,868 # 84
2020-12-01 15:01:34,869 # 42
2020-12-01 15:01:34,869 # 00
2020-12-01 15:01:34,869 # 4a
2020-12-01 15:01:34,869 # 12
2020-12-01 15:01:34,869 # 28
2020-12-01 15:01:34,869 # 1e
2020-12-01 15:01:34,869 # 00
2020-12-01 15:01:34,870 # 22
2020-12-01 15:01:34,871 # 41
2020-12-01 15:01:34,875 # ADC noise request 64 Bytes returned: 0. Time: 1970us

[PeterKietzmann]

@leandrolanzieri I've addressed your comment. Thanks for the intensive testing. Seems like a good collection of tests that you executed and the results are expected. They point out the properties (and problems) of ADC samples and features of this PR in terms of health testing and conditioning.

[leandrolanzieri]

The proposed entropy source, conditioning and health tests look good. Also, having the test application is a good starting point for a developer to assess if the ADC noise (and in the future others) entropy source fits its use case, with a given configuration. Documentation is good and the warnings discussed throughout the PR were included. I provided some tests results showing that the implementation is working as expected.

ACK!

[leandrolanzieri]

And GO!