cvs commit: src/sys/sys mdioctl.h src/sys/dev/md md.c src/sbin/mdconfig mdconfig.8 mdconfig.c

[Brian F. Feldman]

Poul-Henning Kamp <phk at FreeBSD.org> wrote: 
>   In addition to more closely emulating a real disk device with a
>   non-lying write-cache, this makes the writes exempt from rate-limited
>   (there to avoid starving the buffer cache) and consequently prevents
>   the deadlock.
>   
>   Unfortunately performance takes a hit.
>   
>   Add "async" option to give people who know what they are doing the
>   old behaviour.

Could you please implement an errx(3) for MD_ASYNC without MD_FORCE, or 
something of that ilk?  Users who use anything other than mdconfig(8) for 
leanring how to use mdconfig(8) could quite easily run across something like:
	"Make sure to use -o async to get better performance."
and shoot themselves in the foot, whereas if they saw:
	"Make sure to use -o force -o async..."
or
	"Make sure to use -o (dangerous|unsafe|deadlock_prone)_async..."
they would be better aware of the risks.  Just changing the option to make 
it obvious to the user that the option is not just asynchronous but also 
dangerous moves the warning documentation to a more visible location.

This is tools, not policy, but anything that, at little cost, increases
anti-foot-shooting capabilities of the tools is a Good Thing, IMHO.

-- 
Brian Fundakowski Feldman                           \'[ FreeBSD ]''''''''''\
  <> green at FreeBSD.org                               \  The Power to Serve! \
 Opinions expressed are my own.                       \,,,,,,,,,,,,,,,,,,,,,,\