Xen devel - Jul 2012 - [PATCH] libxl: fix vif.ifname when used with stub device model

# HG changeset patch
# User Ian Campbell <ian.campbell@citrix.com>
# Date 1342792265 -3600
# Node ID 79cfa1892a5d55f2c137da6d4f2d5f261b47db26
# Parent  9cdc74869620ed5e9b4f379d83963755f4c45e01
libxl: fix vif.ifname when used with stub device model.

Currently the same libxl_device_nic is used to create both the HVM domain and
its stub domain''s NICs. This means that if a vifname is provided both
the HVM
domains PV NIC and the stub domains PV NIC will get the same name and the
DM''s NIC will fail to be attached.

Instead launder the libxl_device_nic to add the TAP_DEVICE_SUFFIX
("-emu").
This is a bit of a misnomer, since the device is actually PV, but it is used to
"back" the emulated device in the stub domain and this naming scheme
is
consistent with the non-stub case and is known to work e.g. with our hotplug
scripts.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>

diff -r 9cdc74869620 -r 79cfa1892a5d tools/libxl/libxl_dm.c
--- a/tools/libxl/libxl_dm.c	Fri Jul 20 12:03:49 2012 +0100
+++ b/tools/libxl/libxl_dm.c	Fri Jul 20 14:51:05 2012 +0100
@@ -616,6 +616,24 @@ static char ** libxl__build_device_model
     }
 }
 
+static void libxl__dm_nics_from_hvm_guest_config(libxl__gc *gc,
+                                    libxl_domain_config * const guest_config,
+                                    libxl_domain_config *dm_config)
+{
+    int i, nr = guest_config->num_nics;
+
+    GCNEW_ARRAY(dm_config->nics, nr);
+
+    for (i=0; i<nr; i++) {
+        dm_config->nics[i] = guest_config->nics[i];
+        if (dm_config->nics[i].ifname)
+            dm_config->nics[i].ifname = GCSPRINTF("%s"
TAP_DEVICE_SUFFIX,
+                                                 
dm_config->nics[i].ifname);
+    }
+
+    dm_config->num_nics = nr;
+}
+
 static int libxl__vfb_and_vkb_from_hvm_guest_config(libxl__gc *gc,
                                         const libxl_domain_config
*guest_config,
                                         libxl_device_vfb *vfb,
@@ -764,8 +782,7 @@ void libxl__spawn_stub_dm(libxl__egc *eg
     dm_config->disks = guest_config->disks;
     dm_config->num_disks = guest_config->num_disks;
 
-    dm_config->nics = guest_config->nics;
-    dm_config->num_nics = guest_config->num_nics;
+    libxl__dm_nics_from_hvm_guest_config(gc, guest_config, dm_config);
 
     dm_config->c_info.run_hotplug_scripts         
guest_config->c_info.run_hotplug_scripts;

Ian Campbell writes ("[PATCH] libxl: fix vif.ifname when used with stub
device model"):
> libxl: fix vif.ifname when used with stub device model.
> 
> Currently the same libxl_device_nic is used to create both the HVM domain
and
> its stub domain''s NICs. This means that if a vifname is provided
both the HVM
> domains PV NIC and the stub domains PV NIC will get the same name and the
> DM''s NIC will fail to be attached.
> 
> Instead launder the libxl_device_nic to add the TAP_DEVICE_SUFFIX
("-emu").
> This is a bit of a misnomer, since the device is actually PV, but it is
used to
> "back" the emulated device in the stub domain and this naming
scheme is
> consistent with the non-stub case and is known to work e.g. with our
hotplug
> scripts.
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

as far as it goes.

However I think this is related to Roger''s:
   [PATCH v9 12/17] libxl: set correct nic type depending on the guest

Shouldn''t the type field for the nics to be given to the stub dm be
forced to VIF ?

Ian.

On Fri, 2012-07-20 at 15:06 +0100, Ian Jackson wrote:
> Ian Campbell writes ("[PATCH] libxl: fix vif.ifname when used with
stub device model"):
> > libxl: fix vif.ifname when used with stub device model.
> > 
> > Currently the same libxl_device_nic is used to create both the HVM
domain and
> > its stub domain''s NICs. This means that if a vifname is
provided both the HVM
> > domains PV NIC and the stub domains PV NIC will get the same name and
the
> > DM''s NIC will fail to be attached.
> > 
> > Instead launder the libxl_device_nic to add the TAP_DEVICE_SUFFIX
("-emu").
> > This is a bit of a misnomer, since the device is actually PV, but it
is used to
> > "back" the emulated device in the stub domain and this
naming scheme is
> > consistent with the non-stub case and is known to work e.g. with our
hotplug
> > scripts.
> 
> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
> 
> as far as it goes.
> 
> However I think this is related to Roger''s:
>    [PATCH v9 12/17] libxl: set correct nic type depending on the guest
> 
> Shouldn''t the type field for the nics to be given to the stub dm
be
> forced to VIF ?
I should imagine so, although it''s somewhat orthogonal other than both
involve frobbing the DM''s NIC config.

I presume that in practice PV guests can''t have an EMU path and
therefore the two NIC types collapse into the same thing, although
there''s no harm being explicit I suppose.

On Fri, 2012-07-20 at 15:17 +0100, Ian Campbell wrote:
> I presume that in practice PV guests can''t have an EMU path and
> therefore the two NIC types collapse into the same thing,
This happens in libxl__device_nic_setdefaults BTW.

Ian.

On Fri, 2012-07-20 at 15:06 +0100, Ian Jackson wrote:
> Ian Campbell writes ("[PATCH] libxl: fix vif.ifname when used with
stub device model"):
> > libxl: fix vif.ifname when used with stub device model.
> > 
> > Currently the same libxl_device_nic is used to create both the HVM
domain and
> > its stub domain''s NICs. This means that if a vifname is
provided both the HVM
> > domains PV NIC and the stub domains PV NIC will get the same name and
the
> > DM''s NIC will fail to be attached.
> > 
> > Instead launder the libxl_device_nic to add the TAP_DEVICE_SUFFIX
("-emu").
> > This is a bit of a misnomer, since the device is actually PV, but it
is used to
> > "back" the emulated device in the stub domain and this
naming scheme is
> > consistent with the non-stub case and is known to work e.g. with our
hotplug
> > scripts.
> 
> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Applied, thanks.