Re: [libvirt] [BUG] mlock support breakage
On Wed, 15 Mar 2017 14:29:13 -0400
Luiz Capitulino <lcapitulino(a)redhat.com> wrote:
> ... we could consider <locked/> to be the explicit request
for
> setting an infinite memory locking limit and letting users set a lower
> limit with hard_limit if they want.
That's exactly how I see it! It seems we're total agreement.
Now, something has just occurred to me: shouldn't VFIO have
the same problem? It's the same hard limit that's set.
I took a look at this today. While it's the same mlock limit
that's set and while QEMU's allocations can surpass that limit,
I didn't get a crash when using VFIO. The most probable obvious
reason for this is that VFIO is probably mlocking a small region,
although I could not find where this is done in QEMU.
In that case, VFIO is not affected. This issue is specific to
<locked/>, where the 1GB limit set by libvirt conflicts with
QEMU memory needs.