Hi Kars,
On Wed, Oct 30, 2019 at 10:08 PM Kars de Jong <jongk@xxxxxxxxxxxxxx> wrote:
Um, no, that would break the FAS236. FSC is defined after FAS236. It's
import that its values is lower than FAS100A and greater or equal than
FAS236. And as I wrote in reply to Finn, I think PCSCSI falls in the
same category.
You're right -
You definitely want to add a comment like "all below use the same CONFIG3
settings", to avoid nasty surprises for future additions (if any).
Using feature bits might be even better, but may not be worthwhile, since
there seems to be only a single "esp->rev > ..." check.
No, there are several more actually:
* "esp->rev > ESP100A" which basically means "HAS_CONFIG3".
* "esp->rev < ESP236" which basically means "!HAS_FAST_CLOCK".
* "esp->rev >= FAS236" which basically means "HAS_FAST_CLOCK"
* "esp->rev >= FAS100A" which basically means "HAS_OLD_CONFIG3"
So, perhaps having feature bits is not a bad idea at all...
We need a few more feature bits (at least HAS_CONFIG2 to differentiate
between ESP100 and ESP100A in esp_get_revision()). This might get hard
in terms of test coverage, but I'd raise that as alternative to adding
FSC to the list or enum in your RFC to linux-scsi.
Cheers,
Michael
Kind regards,
Kars.
