svn commit: r270759 - in head/sys: cddl/compat/opensolaris/kern cddl/compat/opensolaris/sys cddl/contrib/opensolaris/uts/common/fs/zfs vm

[Andriy Gapon]

on 03/09/2014 17:58 Andriy Gapon said the following:
> on 03/09/2014 15:17 Steven Hartland said the following:
>> ----- Original Message ----- From: "Andriy Gapon" <avg at FreeBSD.org>
>>
>>> on 03/09/2014 11:09 Steven Hartland said the following:
>>>> I'm looking to MFC this change so wanted to check if
>>>> anyone had an final feedback / objections?
>>>
>>> I think that your changes went in a bit prematurely (little review), so perhaps
>>> MFC would be premature as well.
>>
>> Its a change which really needs to make it into 10.1 IMO
> 
> I think that this is the arguable point.
> As I've mentioned before I have not noticed, perhaps through the fault of my
> own, any reports that users need this change after Alan's pagedaemon fix(es).
> Also, there is no confirmation yet that after this change ARC does not give up
> its buffers too easily.
> 
>> due to its impact on users so I don't really want to hold
>> off too long.
>>
>> If anyone has any substantiated reason to then off course
>> I'll hold off.

Based on our parallel conversation I feel a need to clarify my position.
The commit in question has multiple changes in it:
1. removal of KVA check, which was later correctly restored for i386
2. addition of DTrace probes
3. zfs_arc_free_target check and all the support code for it

So, #1 plus later fixes is obviously correct.
#2 is useful and I like it.
#3 is what I have great doubts about.  All of what I said in the previous emails
applies to #3 exclusively.

-- 
Andriy Gapon