[PATCH] Small (*cough*) patchset for vfs_fruit bug #12427

[Ralph Böhme]

On Sat, Feb 11, 2017 at 09:56:39PM +0100, Ralph Böhme wrote:
> On Wed, Feb 08, 2017 at 09:41:41AM +0100, Ralph Böhme wrote:
> > On Mon, Feb 06, 2017 at 07:35:27AM +0200, Uri Simchoni wrote:
> > > Continuing from 36/56 with some minor comments, then responding to last
> > > email. Everything else LGTM.
> > > 
> > > Thanks for bearing with me,
> > > Uri.
> > 
> > thanks for taking the stab! :)
> > 
> > Just a quick update, still working through your review feedback, also adding
> > caching to catia handle based ops.
> 
> ok, so after struggling with getting the recursive VFS calls and fsp recursion
> right, here's a catia name translation caching for handle based ops that finally
> passes make test, that I understand and can explain. :)
> 
> I've been banging my head against this on Wednesday together with metze and we
> couldn't figure it out. Finally came up with something under the shower this
> morning (no, really), so metze if you have the time, please take a look at the
> catia caching patch.
> 
> Full branch:
> <https://git.samba.org/?p=slow/samba.git;a=shortlog;h=refs/heads/fruit-metadata>
> 
> This is the catia patch:
> <https://git.samba.org/?p=slow/samba.git;a=commitdiff;h=0e36321c5b2d9d8f7582941b557217d528390894>
> 
> Once I have metze's blessing on this one, I'll send out a fresh patchset to Uri
> for review, alongside a diff showing what changed since the previous iteration.

updated branch:
<https://git.samba.org/?p=slow/samba.git;a=shortlog;h=refs/heads/fruit-metadata>

catia name translation cache code is new, still need metze to take a look at it:
<https://goo.gl/aVWwqc>

Looks like the cache is effective. I added some basic cache usage metrics to the
code and then copied a bunch of directories:

(gdb) p cc_stats
$1 = {created_fsp_ext = 996, created_tmp = 37, cache_hits = 13396}

I'm not planning to ship the stats code.

Cheerio!
-slow