On Tue, Dec 25, 2018 at 06:09:19PM +0800, Jason Wang wrote:
On 2018/12/25 äå2:12, Michael S. Tsirkin wrote:It's nice but not great. E.g. adaptive polling would be
On Mon, Dec 24, 2018 at 04:32:39PM +0800, Jason Wang wrote:If we can reduce the latency that's will be more than what we expect.
On 2018/12/14 äå8:33, Michael S. Tsirkin wrote:Do we care a lot about the light load? How would you benchmark it?
On Fri, Dec 14, 2018 at 11:42:18AM +0800, Jason Wang wrote:unsafe variants can only work if you can batch userspace access. This is not
On 2018/12/13 äå11:27, Michael S. Tsirkin wrote:I mean the next barrier people decide to put into userspace
On Thu, Dec 13, 2018 at 06:10:19PM +0800, Jason Wang wrote:I don't get here, do you mean spec barriers?
Hi:Userspace accesses through remapping tricks and next time there's a need
This series tries to access virtqueue metadata through kernel virtual
address instead of copy_user() friends since they had too much
overheads like checks, spec barriers or even hardware feature
toggling.
for a new barrier we are left to figure it out by ourselves.
memory accesses.
It's completely unnecessary forIt's defence in depth. Take a look at the commit that added them.
vhost which is kernel thread.
And yes quite possibly in most cases we actually have a spec
barrier in the validation phase. If we do let's use the
unsafe variants so they can be found.
necessarily the case for light load.
1 byte TCP_RR shows 1.5%-2% improvement.
a better approach to work on latency imho.
Some people use file backed pages for vms.It doesn't matter really, if you dirty pages behind the MM backWe only pin metadata pages, I don't believe they will be used by any DMA.And even if you're right, vhost is not theFor sure. But if one can get by without get user pages, one
only place, there's lots of vmap() based accessing in kernel.
really should. Witness recently uncovered mess with file
backed storage.
the problem is there.
Ok, but the usual case is anonymous pages, do we use file backed pages for
user of vhost?
Nothing prevents them from using vhost as well.
And even if we use sometime, according to the pointer it'sExcept it's not broken if we don't to gup + write.
not something that can fix, RFC has been posted to solve this issue.
Thanks
So yea, wait for rfc to be merged.