Re: [PATCH v4 00/12] KVM/hostmem: Support init-shared guest-memfd as VM backends

[Peter Xu <peterx@redhat.com>]

On Sun, Aug 23, 2026 at 10:52:13AM -0500, Michael Roth wrote:
> On Fri, Aug 21, 2026 at 10:18:26AM -0400, Peter Xu wrote:
> > On Wed, Aug 12, 2026 at 03:16:38PM -0500, Michael Roth wrote:
> > > v1: https://lore.kernel.org/r/20251023185913.2923322-1-peterx@redhat.com
> > > v2: https://lore.kernel.org/r/20251119172913.577392-1-peterx@redhat.com
> > > v3: https://lore.kernel.org/r/20251215205203.1185099-1-peterx@redhat.com/
> > > v4:
> > > - Picked up v3 series from Peter Xu and rebased on 11.1.0
> > > - Collected pending Reviewed-by's
> > > - Per-patch changes:
> > >   [01/12] Fix blank space (Xiaoyao)
> > >           Fix commit message typos
> > >   [02/12] Drop duplicated error-handling for capability checks
> > >   [05/12] Fix up commit message, disambigurate "in-place" terminology (Xiaoyao)
> > >   [07/12] Fix up commit message, disambigurate "in-place" terminology (Xiaoyao)
> > >   [08/12] Disallow 'seal' and 'hugetlb' options based on guest-memfd=on
> > >           Fix minor typo in comment
> > >   [11/12] Fix up typo in g_test_skip() reason
> > >           Pending: Move kvm_guest_memfd_init_shared_supported() to
> > >             migration-util.c (Fabiano)
> > > 
> > > This patchset is also available at:
> > > 
> > >   https://github.com/amdese/qemu/commits/gmem-shared-mem-v4
> > > 
> > > and is based on top of v11.1.0
> > 
> > Should I just pick this up?  I'll touch up the commit log that Phil
> > requested.  Objections welcomed before next week.
> 
> Hi Peter,

Hi, Michael,

> 
> No objection here, although I'll be posting new version of the in-place
> conversion support later this week (~Wed-ish) and was planning to an
> updated version of this series roughly the same time so which ever way
> is best for you.
> 
> If you do post one though there's also the missing 'return false' and
> dropping the seal=off enforcement that Daniel noticed before we started
> talking about introspection fun.

I didn't mean to repost this series, I meant to pick it up for a pull.
Sorry if it was confusing..

I want to make sure balls are rolling.  From our side, we weren't rush on
this one to land, as we'll need to wait for 1G to settle anyway, but if we
have others (like your in-place series upcoming) pending, I think we should
just land it sooner.

You're right, we need to fix things, I noticed it right after I sent the
email and then I found CI / unit test failing.. but I haven't got a chance
to say..  If you repost please check migration-test with --full, otherwise
the gmemfd test will be skipped.

I have two fixups queued locally, I'll attach it here, please use it
whatever way you like, I'll wait for your repost.  NOTE: I touched up the
hugetlb failure message when I fixed it up myself, but since you'll repost
you can decide; please just treat it as a comment then, I just think the
new err message is less strong because we will support it.

Thanks,

===8<===


From 0a0dd0f808d4f668ac099945ab35a981e1a07bfb Mon Sep 17 00:00:00 2001
From: Peter Xu <peterx@redhat.com>
Date: Fri, 21 Aug 2026 12:16:15 -0400
Subject: [PATCH 1/2] fixup! hostmem: Support fully shared guest memfd to back
 a VM

Signed-off-by: Peter Xu <peterx@redhat.com>
---
 backends/hostmem-memfd.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/backends/hostmem-memfd.c b/backends/hostmem-memfd.c
index fbe65b00be..6576331441 100644
--- a/backends/hostmem-memfd.c
+++ b/backends/hostmem-memfd.c
@@ -57,14 +57,16 @@ memfd_backend_memory_alloc(HostMemoryBackend *backend, Error **errp)
     }
 
     if (m->guest_memfd) {
+        /*
+         * NOTE: guest-memfd ignores seal=on/off because it always
+         * implicitly seals the FD by definition.
+         */
         if (!backend->share) {
             error_setg(errp, "guest-memfd=on must be used with share=on");
             return false;
-        } else if (m->seal) {
-            error_setg(errp, "guest-memfd=on must be used with seal=off");
-            return false;
         } else if (m->hugetlb) {
-            error_setg(errp, "guest-memfd=on must be used with hugetlb=off");
+            error_setg(errp, "guest-memfd=on doesn't support hugetlb=on yet");
+            return false;
         }
 
         fd = kvm_create_guest_memfd(backend->size,
-- 
2.54.0



From 66ac7b3f2268430106bfe8034ef8460215e9db91 Mon Sep 17 00:00:00 2001
From: Peter Xu <peterx@redhat.com>
Date: Fri, 21 Aug 2026 13:00:09 -0400
Subject: [PATCH 2/2] fixup! tests/migration-test: Add a precopy test for
 guest-memfd

Signed-off-by: Peter Xu <peterx@redhat.com>
---
 tests/qtest/migration/precopy-tests.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/tests/qtest/migration/precopy-tests.c b/tests/qtest/migration/precopy-tests.c
index 8146f37d61..d57ffcac23 100644
--- a/tests/qtest/migration/precopy-tests.c
+++ b/tests/qtest/migration/precopy-tests.c
@@ -184,7 +184,6 @@ static void test_precopy_tcp_plain(char *name, MigrateCommon *args)
 
 static void test_precopy_tcp_plain_gmemfd(char *name, MigrateCommon *args)
 {
-    args->uri = "tcp:127.0.0.1:0";
     args->start.mem_type = MEM_TYPE_GUEST_MEMFD;
 
     test_precopy_common(args);
-- 
2.54.0


-- 
Peter Xu