[PATCHES]vfs: separate fs stats from quota in vfs modules

[Volker Lendecke]

On Tue, Jan 12, 2016 at 01:55:20PM +0200, Uri Simchoni wrote:
> I'm not exactly sure what's your concern with the current implementation

"Concern" is too strong a word. It's just that I see a slight duplication
of concepts with the dfq_params struct and its members. Enums, names,
structs -- I would understand that effort if this was speed-critical
in production code that also needs properly documented parameters.
The "add get_quota imlementation" has to extend the enum and the array. If
another aspect comes along even more will be required. lp_parm_int has
its overhead, so I would not do it in a hot code path, but for test code
it would be perfectly fine.

> (whether it's simplicity/elegance or the dependence on pm_process or
> something else) but I'm beginning to like the idea of using lp_parm, albeit
> in a bit unorthodox way.
> 
> Bear in mind that I need per-folder parameters, not per-share parameters, so
> I will need to somehow encode the path name in the parameter key. It seems
> like simply concatenating the path to the parameter name would work for
> .conf files. For example to find parameter "df block size" in folder
> foo/bar, I can do lp_parm_ulonglong(snum, "dfq", "df block size/foo/bar").
> The unorthodox thing is use of '/' and '.' in the key.

We have idmap config <domainname> : * , that is also not really straight
forward :-)

Volker

-- 
SerNet GmbH, Bahnhofsallee 1b, 37081 Göttingen
phone: +49-551-370000-0, fax: +49-551-370000-9
AG Göttingen, HRB 2816, GF: Dr. Johannes Loxen
http://www.sernet.de, mailto:kontakt at sernet.de