ACK: [PATCH 0/2] [Trusty] (upstream) SRU: LP1616193 vhost-net:

[Dave Chiluk]

Yes I boot tested them, but as noted it's difficult to cause sufficient
fragmentation to experience the issue.

Dave.

On 08/25/2016 03:07 PM, Christopher Arges wrote:
> Clean cherry-pick, looked at history for other patches and this seems like the
> two that are necessary. You've boot tested them.
>
> --chris
>
> On Thu, Aug 25, 2016 at 02:37:14PM -0500, Dave Chiluk wrote:
>> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1616193
>>
>> The Problem: when a host machines memory becomes fragmented to the point where kmalloc order 4 fails, vhost_net_open fails causing virtual machines to start failing.
>>
>> The Fix: Adds a fall-back to vmalloc when kmalloc fails.  This is similar to what other subsystems do.  This does not relieve the sorce of the fragmentation which is believed to be dentry cache *(vfs.cache.pressure=600, helps but does not eliminate the issue).  Fix was included in 3.15 so not necessary for our other kernels.
>>
>> The first patch is required in order to fix the bug.
>>
>> The second patch was a later clean-up of the first patch, and I think it's responsible for inclusion.  However I'll leave that decision to the Team.  
>>
>> Dave.
>>
>>
>> -- 
>> kernel-team mailing list
>> kernel-team at lists.ubuntu.com
>> https://lists.ubuntu.com/mailman/listinfo/kernel-team