scottl at FreeBSD.org
Mon Oct 11 15:26:29 PDT 2004
Matthew Jacob wrote:
> No point in a giant-free SIM until CAM itself can cope.
> The isp driver has all the stuff in place to be Giant-Free, and might
> even work, but there's no point if you have to transition in and out of
> a Giant dependent subsystem.
> And the answer to the next question is: "No, I don't know who *really*
> is working on a Giant-Free CAM".
I am, but I got sidetracked by these pesky release engineering tasks
that people seem to care about =-) There is a branch called 'camlock'
in perforce that has the beginnings of it. The big work so far in there
has been to take the device probe out of the camisr context so that
locking the rest of CAM is easier.
More information about the freebsd-current