[05/12] hw/xen: populate store frontend nodes with XenStore PFN/port

Commit Message

David Woodhouse Oct. 16, 2023, 3:19 p.m. UTC

From: David Woodhouse <[e-mail]>

This is kind of redundant since without being able to get these through
ome other method (HVMOP_get_param) the guest wouldn't be able to access
XenStore in order to find them. But Xen populates them, and it does
allow guests to *rebind* to the event channel port after a reset.

Signed-off-by: David Woodhouse <[e-mail]>
---
 hw/i386/kvm/xen_xenstore.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

Comments

Paul Durrant Oct. 24, 2023, 12:35 p.m. UTC | #1

On 16/10/2023 16:19, David Woodhouse wrote:
> From: David Woodhouse <[e-mail]>
> 
> This is kind of redundant since without being able to get these through
> ome other method (HVMOP_get_param) the guest wouldn't be able to access

^ typo

> XenStore in order to find them. But Xen populates them, and it does
> allow guests to *rebind* to the event channel port after a reset.
> 

... although this can also be done by querying the remote end of the 
port before reset.

> Signed-off-by: David Woodhouse <[e-mail]>
> ---
>   hw/i386/kvm/xen_xenstore.c | 10 ++++++++++
>   1 file changed, 10 insertions(+)
> 

Reviewed-by: Paul Durrant <[e-mail]>

David Woodhouse Oct. 24, 2023, 12:53 p.m. UTC | #2

On Tue, 2023-10-24 at 13:35 +0100, Paul Durrant wrote:
> On 16/10/2023 16:19, David Woodhouse wrote:
> > From: David Woodhouse <[e-mail]>
> > 
> > This is kind of redundant since without being able to get these
> > through
> > ome other method (HVMOP_get_param) the guest wouldn't be able to
> > access
> 
> ^ typo
> 
> > XenStore in order to find them. But Xen populates them, and it does
> > allow guests to *rebind* to the event channel port after a reset.
> > 
> 
> ... although this can also be done by querying the remote end of the 
> port before reset.
> 
I think you had to do that anyway; I don't think I was supposed to be
putting s->be_port in there anyway; it was supposed to be the
*frontend* port, and I've changed that in my tree.

I'll drop this whole sentence (and fix the typo).

> > Signed-off-by: David Woodhouse <[e-mail]>
> > ---
> >   hw/i386/kvm/xen_xenstore.c | 10 ++++++++++
> >   1 file changed, 10 insertions(+)
> > 
> 
> Reviewed-by: Paul Durrant <[e-mail]>
>

Patch

diff --git a/hw/i386/kvm/xen_xenstore.c b/hw/i386/kvm/xen_xenstore.c
index d2b311109b..3300e0614a 100644
--- a/hw/i386/kvm/xen_xenstore.c
+++ b/hw/i386/kvm/xen_xenstore.c
@@ -1432,6 +1432,7 @@  static void alloc_guest_port(XenXenstoreState *s)
 int xen_xenstore_reset(void)
 {
     XenXenstoreState *s = xen_xenstore_singleton;
+    GList *perms;
     int err;
 
     if (!s) {
@@ -1459,6 +1460,15 @@  int xen_xenstore_reset(void)
     }
     s->be_port = err;
 
+    /* Create frontend store nodes */
+    perms = g_list_append(NULL, xs_perm_as_string(XS_PERM_NONE, DOMID_QEMU));
+    perms = g_list_append(perms, xs_perm_as_string(XS_PERM_READ, xen_domid));
+
+    relpath_printf(s, perms, "store/ring-ref", "%lu", XEN_SPECIAL_PFN(XENSTORE));
+    relpath_printf(s, perms, "store/port", "%u", s->be_port);
+
+    g_list_free_full(perms, g_free);
+
     /*
      * We don't actually access the guest's page through the grant, because
      * this isn't real Xen, and we can just use the page we gave it in the