On 12/3/19 12:54 PM, Eric Blake wrote:
On 12/3/19 11:52 AM, Vladimir Sementsov-Ogievskiy wrote:
It's just a memory leak, but it's a regression in 4.2.
Should we take it into 4.2?
Sorry, I was on holiday and then jury service, so I missed any chance at
getting this into -rc3. The memory leak only happens on failure, and
you'd have to be pretty desperate to purposefully attempt to open a lot
of NBD devices where you know you'll get a failure just to trigger
enough of a leak to cause the OOM-killer to target qemu. So I'm fine if
this is deferred to 5.0, and just cc's qemu-stable (now done).
I'll queue this through my NBD tree for 5.0.
Actually, given the review comments on 1/2, we'll probably be better off
with a v4 for the series.
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3226
Virtualization: qemu.org | libvirt.org
