From: Igor Mammedov <imammedo@redhat.com>
To: "Marc-André Lureau" <marcandre.lureau@gmail.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>, QEMU <qemu-devel@nongnu.org>
Subject: Re: [Qemu-devel] [PATCH] hostmem-memfd: honour share=on/off property
Date: Thu, 30 Aug 2018 10:20:06 +0200 [thread overview]
Message-ID: <20180830102006.5aaf9054@redhat.com> (raw)
In-Reply-To: <CAJ+F1CLniMjFc1PCYqeBSkSJJ-sCk+PKvjupdurXKez9wedA4w@mail.gmail.com>
On Wed, 29 Aug 2018 14:36:00 +0200
Marc-André Lureau <marcandre.lureau@gmail.com> wrote:
> Hi
>
> On Wed, Aug 29, 2018 at 10:50 AM Igor Mammedov <imammedo@redhat.com> wrote:
> >
> > On Tue, 28 Aug 2018 18:55:35 +0200
> > Marc-André Lureau <marcandre.lureau@gmail.com> wrote:
> >
> > > On Tue, Aug 28, 2018 at 6:54 PM Marc-André Lureau
> > > <marcandre.lureau@gmail.com> wrote:
> > > >
> > > > Hi
> > > >
> > > > On Tue, Aug 28, 2018 at 6:49 PM Igor Mammedov <imammedo@redhat.com> wrote:
> > > > >
> > > > > On Tue, 28 Aug 2018 17:38:40 +0200
> > > > > Marc-André Lureau <marcandre.lureau@redhat.com> wrote:
> > > > >
> > > > > > The share=on/off property is used to modified mmap() MAP_SHARED
> > > > > > setting. Make it on by default for convenience and compatibility
> > > > > > reasons.
> > > > > it would be nice to state reasons here.
> > > > >
> > > > > also flipping default to on is behavioral change when
> > > > > used with bind policy which could make backend initialization fail
> > > > > so CLI that used to work (if there are any users of this without shared=on)
> > > > > won't work anymore.
> > > >
> > > > memfd backend is passing shared=true to memory_region_init_ram_from_fd().
> > > >
> > > > The patch allows to tweak that (instead of ignoring the property),
> > > > while keeping the value to shared=true.
> > >
> > > keeping the default value to shared=trrue.
> > ok,
> > another question:
> > does it make sense to allow "shared=false" at all?
> > should we bail out if user asks for it?
>
> You may want to use memfd for convenience (no need to specify hugetlb
> mem-path for ex), without the share memory aspects.
ok, acked.
btw:
looking at memfd_backend_memory_alloc/file_backend_memory_alloc
both needlessly are are calling host_memory_backend_mr_inited()
which creates an illusion that alloc could be called multiple times
but it isn't, could you clean that up?
In addition a following error path needs fixing
memfd_backend_memory_alloc()
if (fd == -1) {
return;
}
it must set errp
>
> (but the patch was mostly for completeness)
>
> >
> > >
> > > >
> > > > >
> > > > > >
> > > > > > Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
> > > > > > ---
> > > > > > backends/hostmem-memfd.c | 4 +++-
> > > > > > qemu-options.hx | 4 +++-
> > > > > > 2 files changed, 6 insertions(+), 2 deletions(-)
> > > > > >
> > > > > > diff --git a/backends/hostmem-memfd.c b/backends/hostmem-memfd.c
> > > > > > index 1e20fe0ba8..abd52364db 100644
> > > > > > --- a/backends/hostmem-memfd.c
> > > > > > +++ b/backends/hostmem-memfd.c
> > > > > > @@ -59,7 +59,8 @@ memfd_backend_memory_alloc(HostMemoryBackend *backend, Error **errp)
> > > > > >
> > > > > > name = object_get_canonical_path(OBJECT(backend));
> > > > > > memory_region_init_ram_from_fd(&backend->mr, OBJECT(backend),
> > > > > > - name, backend->size, true, fd, errp);
> > > > > > + name, backend->size,
> > > > > > + backend->share, fd, errp);
> > > > > > g_free(name);
> > > > > > }
> > > > > >
> > > > > > @@ -131,6 +132,7 @@ memfd_backend_instance_init(Object *obj)
> > > > > >
> > > > > > /* default to sealed file */
> > > > > > m->seal = true;
> > > > > > + MEMORY_BACKEND(m)->share = true;
> > > > > > }
> > > > > >
> > > > > > static void
> > > > > > diff --git a/qemu-options.hx b/qemu-options.hx
> > > > > > index f14e423b7c..25c02d72b8 100644
> > > > > > --- a/qemu-options.hx
> > > > > > +++ b/qemu-options.hx
> > > > > > @@ -4029,7 +4029,7 @@ Memory backend objects offer more control than the @option{-m} option that is
> > > > > > traditionally used to define guest RAM. Please refer to
> > > > > > @option{memory-backend-file} for a description of the options.
> > > > > >
> > > > > > -@item -object memory-backend-memfd,id=@var{id},merge=@var{on|off},dump=@var{on|off},prealloc=@var{on|off},size=@var{size},host-nodes=@var{host-nodes},policy=@var{default|preferred|bind|interleave},seal=@var{on|off},hugetlb=@var{on|off},hugetlbsize=@var{size}
> > > > > > +@item -object memory-backend-memfd,id=@var{id},merge=@var{on|off},dump=@var{on|off},share=@var{on|off},prealloc=@var{on|off},size=@var{size},host-nodes=@var{host-nodes},policy=@var{default|preferred|bind|interleave},seal=@var{on|off},hugetlb=@var{on|off},hugetlbsize=@var{size}
> > > > > >
> > > > > > Creates an anonymous memory file backend object, which allows QEMU to
> > > > > > share the memory with an external process (e.g. when using
> > > > > > @@ -4051,6 +4051,8 @@ with the @option{seal} option (requires at least Linux 4.16).
> > > > > > Please refer to @option{memory-backend-file} for a description of the
> > > > > > other options.
> > > > > >
> > > > > > +The @option{share} boolean option is @var{on} by default with memfd.
> > > > > > +
> > > > > > @item -object rng-random,id=@var{id},filename=@var{/dev/random}
> > > > > >
> > > > > > Creates a random number generator backend which obtains entropy from
> > > > >
> > > > >
> > > >
> > > >
> > > > --
> > > > Marc-André Lureau
> > >
> > >
> > >
> >
>
>
next prev parent reply other threads:[~2018-08-30 8:20 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-08-28 15:38 [Qemu-devel] [PATCH] hostmem-memfd: honour share=on/off property Marc-André Lureau
2018-08-28 16:43 ` Igor Mammedov
2018-08-28 16:54 ` Marc-André Lureau
2018-08-28 16:55 ` Marc-André Lureau
2018-08-29 8:50 ` Igor Mammedov
2018-08-29 12:36 ` Marc-André Lureau
2018-08-30 8:20 ` Igor Mammedov [this message]
2018-08-30 17:10 ` Marc-André Lureau
2018-08-30 8:14 ` Igor Mammedov
2018-11-26 16:57 ` Paolo Bonzini
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180830102006.5aaf9054@redhat.com \
--to=imammedo@redhat.com \
--cc=marcandre.lureau@gmail.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.