RPI3 swap experiments

[Mark Millard]

On 2018-Jul-31, at 4:19 PM, bob prohaska <fbsd AT www.zefox.net> wrote:

> On Tue, Jul 31, 2018 at 01:49:57PM -0700, Mark Millard wrote:
>> On 2018-Jul-31, at 12:10 PM, bob prohaska <fbsd at www.zefox.net> wrote:
>> 
>> 
>>> Some weeks (months?) ago there was a thread about swap being broken. Was
>>> that in any way related to what I'm seeing?
>> 
>> There was some ZFS context stuff that seemed to be independent of
>> UFS stuff relative to memory use. I continue to see reports tied
>> to ZFS contexts.
>> 
>> But I'm not sure if this is in any way related to what you are
>> calling "swap being broken". I do not remember anything about
>> swap being directly broken for swap partitions. (Swap files are
>> a different issue and are problematical.)
>> 
> 
> The thread I'm referring to starts with 
> https://lists.freebsd.org/pipermail/freebsd-current/2018-June/069728.html and titled
> " swapping is completely broken in -CURRENT r334649? "
> 
> There's no mention whether ARM is affected, just "certain hardware". 
> It looks like a bug was introduced in r329882 and fixed in r334752.
> Workarounds suggested  include 
> sysctl vm.pageout_update_period=0
> and
> sysctl vm.pageout_update_period=1000. 
> 
> Maybe the first will turn OOMA off, and the second delay it? 
> On  RPi the value is 600. The one correspondent who tried 1000
> said it didn't help, maybe zero is a better guess. For me it's
> a shot in total darkness either way 8-) 

In that thread is the message:

https://lists.freebsd.org/pipermail/freebsd-current/2018-June/069835.html

that has a patch for dumping out more information to demsg before
it does the kill:

diff --git a/sys/vm/vm_pageout.c b/sys/vm/vm_pageout.c
index 264c98203c51..9c7ebcf451ec 100644
--- a/sys/vm/vm_pageout.c
+++ b/sys/vm/vm_pageout.c
@@ -1670,6 +1670,8 @@ vm_pageout_mightbe_oom(struct vm_domain *vmd, int page_shortage,
 	 * start OOM.  Initiate the selection and signaling of the
 	 * victim.
 	 */
+	printf("v_free_count: %u, v_inactive_count: %u\n",
+	    vmd->vmd_free_count, vmd->vmd_pagequeues[PQ_INACTIVE].pq_cnt);
 	vm_pageout_oom(VM_OOM_MEM);
 
 	/*


Mark Johnston also mentions "sysctl vm" output as proving more
contextual information.



===
Mark Millard
marklmi at yahoo.com
( dsl-only.net went
away in early 2018-Mar)