[flashrom] [PATCH] Fix the delay loop

[Maciej Pijanka]

On Mon, 29 Mar 2010, Carl-Daniel Hailfinger wrote:

> [Side note: This is not the big timing code rewrite I originally wanted
> to post. I decided to focus on readable code instead of clever code. My
> other rewrite is available on request if anyone is interested.]
> 
> The current delay loop calculation is still from revision 1 of flashrom,
> and since then it had a logic bug which caused all delays to be twice as
> long as intended. Fix the delay duration.
> 
> Protect against delay loop overflows.
> 
> Detect a non-working delay loop.
> 
> Change the delay loop itself to ensure clever compiler optimizers won't
> eliminate it (as happens with clang/llvm in the current code). Some
> people suggested machine-specific asm, but the empty asm statement with
> the loop counter as register/memory input has the benefit of being
> perfectly cross-platform and working in gcc and clang.
> 
> If time goes backwards (catastrophical NTP time difference, manual time
> change), timing measurements were shot because the new-old time
> subtraction yielded negative numbers which weren't handled correctly
> because the variable is unsigned. Simply return 1 microsecond timing in
> that case.
> 
> If time goes forward too fast, pick the biggest possible timing
> measurement with a guaranteed overflow avoidance for all timing
> calculations.

Maybe in case of leap seconds or time changing simply detect, and either fail
and restart whole timing calculation, from scratch (possibly noting attempt
number). If after some fixed number of retries you can't get reliable result.
simply fail?

> Check four times if the calculated timing is at most 10% too fast. This
> addresses OS scheduler interactions, e.g. being scheduled out during
> measurement which inflates measurements.

this makes sense but if we want that good results, maybe odd number of repeats,
leave off minimum and maximum ones, then calculate over remaining (and minimum
sane number of remaining atempts is imo 5, thus seven is needed to make sufficient
number of iterations).


Anyway, imo code is good and readable and gives more acurate results on my machines.

Acked-by: Maciej Pijanka <maciej.pijanka at gmail.com>


babol:~/flashrom# ./flashrom -V 2>&1 |grep loop
Calibrating delay loop... 167M loops per second, 10 myus = 12 us, 100 myus = 103 us, 1000 myus = 1003 us, 10000 myus = 10054 us, OK.
babol:~/flashrom# ./flashrom -V 2>&1 |grep loop
Calibrating delay loop... 164M loops per second, 10 myus = 12 us, 100 myus = 101 us, 1000 myus = 1031 us, 10000 myus = 9877 us, OK.
babol:~/flashrom# ./flashrom -V -p nic3com 2>&1 |grep loop
Calibrating delay loop... 167M loops per second, 10 myus = 12 us, 100 myus = 102 us, 1000 myus = 1034 us, 10000 myus = 10052 us, OK.
babol:~/flashrom# ./flashrom -V -p nic3com 2>&1 |grep loop
Calibrating delay loop... 167M loops per second, 10 myus = 11 us, 100 myus = 103 us, 1000 myus = 1035 us, 10000 myus = 10052 us, OK.

-- 
Maciej Pijanka
I don't fear computers, I fear lack of them -- Isaac Asimov