Re: CVS commit: src/sys/nfs

[christos%astron.com@localhost (Christos Zoulas)]

In article <1127165197.784039.921.nullmailer%yamt.dyndns.org@localhost>,
YAMAMOTO Takashi  <yamt%mwd.biglobe.ne.jp@localhost> wrote:

>> Of course it is. I need to version struct nfs_args to add those
>> and the patch is much more complex than this minimal fix for 2.x
>> and 3.x. I am planning to merge in all the fields from nfsv4 so I
>> will have to suffer only a single version bump.
>> 
>> The goal of this change was to fix am-utils (so that we were not
>> the only BSD that was broken) with minimal risk. I will add the
>> ac{reg,dir}{min,max} support when I get some spare cycles and
>> figure out what I need to add to support nfsv4.
>> 
>> christos
>
>iirc, existing practice of "noac" implies synchronous write.
>ie. it's different from acregmax=0.
>if what we need is different, it'd be better to choose
>the different name.

Yes, I am planning to keep the noac flag around to mean:

    "no attribute or data cacheing"

right now it means:

    "no attribute cacheing"

I did not document it, or allowed access to it in mount_nfs. Only
am-utils knows about it now (and am-utils only cares about attribute
cacheing). It will all be fixed eventually.

christos