Re: 4.1 regression in resizable hashtable tests

[Thomas Graf <tgraf@suug.ch>]

On 07/17/15 at 12:26pm, Phil Sutter wrote:
> On Fri, Jul 17, 2015 at 10:04:56AM +0200, Thomas Graf wrote:
> > On 07/02/15 at 10:09pm, Meelis Roos wrote:
> > > [   33.425061] Running rhashtable test nelem=8, max_sizee536, shrinking=0
> > > [   33.425154] Test 00:
> > > [   33.534470]   Adding 50000 keys
> > > [   34.743553] Info: encountered resize
> > > [   34.743698] Info: encountered resize
> > > [   34.743838] Info: encountered resize
> > > [   34.744057] Info: encountered resize
> > > [   34.744430] Info: encountered resize
> > > [   34.745139] Info: encountered resize
> > > [   34.746441] Info: encountered resize
> > > [   34.749055] Info: encountered resize
> > > [   34.754469] Info: encountered resize
> > > [   34.764836] Info: encountered resize
> > > [   34.785696] Info: encountered resize
> > > [   34.827448] Info: encountered resize
> > > [   34.896936]   Traversal complete: countedI993, nelemsP000, entriesP000, table-jumps\x12
> > > [   34.897056] Test failed: Total count mismatch ^^^
> > 
> > I do see count mismatches as well due to the design of the walker
> > which restarts and thus sees certain entries multiple times.
> > 
> > Do you have this commit as well?
> > 
> > Author: Phil Sutter <phil@nwl.cc>
> > Date:   Mon Jul 6 15:51:20 2015 +0200
> > 
> >     rhashtable: fix for resize events during table walk
> 
> Thomas, this should be resolved already. Meelis replied[1] to my patch,
> stating it fixes that problem for him. Though he's still waiting for
> your proposed patch to add a schedule() call so the kernel won't
> complain on his slow UltraSparc. :)
> 
> Cheers, Phil
> 
> [1]: http://www.spinics.net/lists/netdev/msg335767.html

OK, good to know. I've posted the schedule patch today:
https://patchwork.ozlabs.org/patch/497035/