Attention is currently required from: Subrata Banik, Paul Menzel, Edward O'Callaghan, Anastasia Klimchuk.
Patch set 12:Code-Review +1
4 comments:
Commit Message:
Here is the timeout calculation for any hardware sequencing operation:
Worst Case Operational Delay =
(Maximum Time consumed by a SPI operation + Any marginal
adjustment)
Timeout Recommendation for Hardware Sequencing Operation =
((Worst Case Operational Delay) * (#No. Of SPI Master - 1) +
Current Operational latency)
Assume, on Intel platform with 2 SPI master like, Host CPU and CSE, the
Timeout Calculation for SPI Write Operation would look like as below:
Maximum Time consumed by a SPI Operation = 2 seconds
(for SPI Erase Operation as per Winbond Data Sheet)
Worst Case Operational Delay = (2 seconds + 50 milliseconds) =
2 seconds 50 milliseconds
Timeout Recommendation for Hardware Seq Operation =
(2 seconds 50 milliseconds) * (2 - 1) + 15 milliseconds
(for SPI Write Operation as per Winbond Data Sheet)
= 2 seconds 65 milliseconds
Doing the calculation with these numbers can be a bit confusing. I'd
do it with the estimated worst-case numbers right away (i.e. 6 masters,
5s).
Shorting some things might make it more readable, e.g. `2.05s` instead
of `2 seconds 50 milliseconds`, also shorter identifiers and removing
unnecessary parentheses can help legibility.
File ichspi.c:
Another thing you mentioned:
Also, first calculating 2.065s and then concluding 30s probably raises more questions than are answered.
I found where 30s comes from the last paragraph: it says "defines 5 master sections ... The worst case operation ... might take ~5 seconds"
so that's 5*5=25 and taking a bit larger value 30 just in case. I hope my understanding is correct :)
It's a bit of an arbitrary choice actually. I think I mentioned 30s on
the ML because I expected up to 6 masters in the forseeable future. But
5s already has a lot of buffer. It doesn't really matter because we don't
know exact values. Chances need to be sufficiently low that the arbiter
won't delay execution any longer, and it shouldn't be too long so a
user doesn't get impatient when there's an actual hardware error.
Do you think it's fine to move this comment to commit message? And then, as you said, one line comment in the code?
Ack.
File ichspi.c:
Patch Set #12, Line 1281: operation
`operations`? because multiple masters would preform multiple operations?
Patch Set #12, Line 1282: introduces
`introduce` without s, I guess
To view, visit change 62867. To unsubscribe, or for help writing mail filters, visit settings.