[PATCH] missing chunk of HVM direct kernel boot patch

[PATCH] missing chunk of HVM direct kernel boot patch

[Chunyan Liu]

Found by Stefano, this chunk of the patch was never applied to
xen-unstable (commit 11dffa2359e8a2629490c14c029c7c7c777b3e47),
see http://marc.info/?l=qemu-devel&m=140471493425353&w=2.

Signed-off-by: Chunyan Liu <cyliu@suse.com>
---
 tools/libxl/libxl_dm.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c
index 3e191c3..b25b574 100644
--- a/tools/libxl/libxl_dm.c
+++ b/tools/libxl/libxl_dm.c
@@ -527,6 +527,15 @@ static char ** libxl__build_device_model_args_new(libxl__gc *gc,
     if (b_info->type == LIBXL_DOMAIN_TYPE_HVM) {
         int ioemu_nics = 0;
 
+        if (b_info->kernel)
+            flexarray_vappend(dm_args, "-kernel", b_info->kernel, NULL);
+
+        if (b_info->ramdisk)
+            flexarray_vappend(dm_args, "-initrd", b_info->ramdisk, NULL);
+
+        if (b_info->cmdline)
+            flexarray_vappend(dm_args, "-append", b_info->cmdline, NULL);
+
         if (b_info->u.hvm.serial || b_info->u.hvm.serial_list) {
             if ( b_info->u.hvm.serial && b_info->u.hvm.serial_list )
             {
-- 
1.8.4.5

Re: [PATCH] missing chunk of HVM direct kernel boot patch

[Stefano Stabellini]

On Fri, 28 Nov 2014, Chunyan Liu wrote:
> Found by Stefano, this chunk of the patch was never applied to
> xen-unstable (commit 11dffa2359e8a2629490c14c029c7c7c777b3e47),
> see http://marc.info/?l=qemu-devel&m=140471493425353&w=2.
> 
> Signed-off-by: Chunyan Liu <cyliu@suse.com>

I think it should be in 4.5: at the moment the feature is broken and
only half-applied to the tree.

Reviewed-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>


>  tools/libxl/libxl_dm.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c
> index 3e191c3..b25b574 100644
> --- a/tools/libxl/libxl_dm.c
> +++ b/tools/libxl/libxl_dm.c
> @@ -527,6 +527,15 @@ static char ** libxl__build_device_model_args_new(libxl__gc *gc,
>      if (b_info->type == LIBXL_DOMAIN_TYPE_HVM) {
>          int ioemu_nics = 0;
>  
> +        if (b_info->kernel)
> +            flexarray_vappend(dm_args, "-kernel", b_info->kernel, NULL);
> +
> +        if (b_info->ramdisk)
> +            flexarray_vappend(dm_args, "-initrd", b_info->ramdisk, NULL);
> +
> +        if (b_info->cmdline)
> +            flexarray_vappend(dm_args, "-append", b_info->cmdline, NULL);
> +
>          if (b_info->u.hvm.serial || b_info->u.hvm.serial_list) {
>              if ( b_info->u.hvm.serial && b_info->u.hvm.serial_list )
>              {
> -- 
> 1.8.4.5
> 

Re: [PATCH] missing chunk of HVM direct kernel boot patch

[Ian Campbell]

On Fri, 2014-11-28 at 13:55 +0800, Chunyan Liu wrote:
> Found by Stefano, this chunk of the patch was never applied to
> xen-unstable (commit 11dffa2359e8a2629490c14c029c7c7c777b3e47),
> see http://marc.info/?l=qemu-devel&m=140471493425353&w=2.

How strange, "git am" usually makes it pretty difficult to miss hunks.
Sorry about this.

> Signed-off-by: Chunyan Liu <cyliu@suse.com>

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

I'm afraid that despite the circumstances this still needs a release ack
from Konrad. Obviously the upside is fixing a partially implemented
feature, but I'm not sure what the downsides are.

Has this been tested with stubdoms, including when this feature is not
used? My biggest concern is that because this function is also used to
build the command line for the stubdom and the stubdom is PV and hence
has at least a ->kernel setting then this new code might break that use
case, by adding these options when they are not wanted. This path is all
a bit tangled so I'm not 100% sure if those fields are actually set or
not.

> ---
>  tools/libxl/libxl_dm.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c
> index 3e191c3..b25b574 100644
> --- a/tools/libxl/libxl_dm.c
> +++ b/tools/libxl/libxl_dm.c
> @@ -527,6 +527,15 @@ static char ** libxl__build_device_model_args_new(libxl__gc *gc,
>      if (b_info->type == LIBXL_DOMAIN_TYPE_HVM) {
>          int ioemu_nics = 0;
>  
> +        if (b_info->kernel)
> +            flexarray_vappend(dm_args, "-kernel", b_info->kernel, NULL);
> +
> +        if (b_info->ramdisk)
> +            flexarray_vappend(dm_args, "-initrd", b_info->ramdisk, NULL);
> +
> +        if (b_info->cmdline)
> +            flexarray_vappend(dm_args, "-append", b_info->cmdline, NULL);
> +
>          if (b_info->u.hvm.serial || b_info->u.hvm.serial_list) {
>              if ( b_info->u.hvm.serial && b_info->u.hvm.serial_list )
>              {

答复： Re: [PATCH] missing chunk of HVM direct kernel boot patch

[Chun Yan Liu]

>>> Ian Campbell <Ian.Campbell@citrix.com> 2014-11-28 下午 20:01 >>> 
On Fri, 2014-11-28 at 13:55 +0800, Chunyan Liu wrote: 
>> Found by Stefano, this chunk of the patch was never applied to 
>> xen-unstable (commit 11dffa2359e8a2629490c14c029c7c7c777b3e47), 
>> see http://marc.info/?l=qemu-devel&m=140471493425353&w=2. 
>
> How strange, "git am" usually makes it pretty difficult to miss hunks. 
> Sorry about this. 

>> Signed-off-by: Chunyan Liu <cyliu@suse.com> 

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

> I'm afraid that despite the circumstances this still needs a release ack 
> from Konrad. Obviously the upside is fixing a partially implemented 
> feature, but I'm not sure what the downsides are. 
>
> Has this been tested with stubdoms, including when this feature is not 
> used? My biggest concern is that because this function is also used to 
> build the command line for the stubdom and the stubdom is PV and hence 
> has at least a ->kernel setting then this new code might break that use 
> case, by adding these options when they are not wanted. This path is all 
> a bit tangled so I'm not 100% sure if those fields are actually set or 
> not.



'-kernel' is only added to qemu command line under HVM case. PV would

not be affected. And only added when device model is upstream qemu, but

not old qemu-xen. About stubdom, tested before, when stubdom is using

old qemu-xen, would not be affected.

>
>> --- 
>> tools/libxl/libxl_dm.c | 9 +++++++++ 
>> 1 file changed, 9 insertions(+) 
>> 
>> diff --git a/tools/libxl/libxl_dm.c b/tools/libxl/libxl_dm.c 
>> index 3e191c3..b25b574 100644 
>> --- a/tools/libxl/libxl_dm.c 
>> +++ b/tools/libxl/libxl_dm.c 
>> @@ -527,6 +527,15 @@ static char ** libxl__build_device_model_args_new(libxl__gc *gc, 
>> if (b_info->type == LIBXL_DOMAIN_TYPE_HVM) { 
>> int ioemu_nics = 0; 
>> 
>> + if (b_info->kernel) 
>> + flexarray_vappend(dm_args, "-kernel", b_info->kernel, NULL); 
>> + 
>> + if (b_info->ramdisk) 
>> + flexarray_vappend(dm_args, "-initrd", b_info->ramdisk, NULL); 
>> + 
>> + if (b_info->cmdline) 
>> + flexarray_vappend(dm_args, "-append", b_info->cmdline, NULL); 
>> + 
>> if (b_info->u.hvm.serial || b_info->u.hvm.serial_list) { 
>> if ( b_info->u.hvm.serial && b_info->u.hvm.serial_list ) 
>> {

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: 答复： Re: [PATCH] missing chunk of HVM direct kernel boot patch

[Ian Campbell]

On Mon, 2014-12-01 at 01:21 -0700, Chun Yan Liu wrote:
> 
> >>> Ian Campbell <Ian.Campbell@citrix.com> 2014-11-28 下午 20:01 >>> 
> On Fri, 2014-11-28 at 13:55 +0800, Chunyan Liu wrote: 
> >> Found by Stefano, this chunk of the patch was never applied to 
> >> xen-unstable (commit 11dffa2359e8a2629490c14c029c7c7c777b3e47), 
> >> see http://marc.info/?l=qemu-devel&m=140471493425353&w=2. 
> >
> > How strange, "git am" usually makes it pretty difficult to miss hunks. 
> > Sorry about this. 
> 
> >> Signed-off-by: Chunyan Liu <cyliu@suse.com> 
> 
> > Acked-by: Ian Campbell <ian.campbell@citrix.com> 
> 
> > I'm afraid that despite the circumstances this still needs a release ack 
> > from Konrad. Obviously the upside is fixing a partially implemented 
> > feature, but I'm not sure what the downsides are. 
> >
> > Has this been tested with stubdoms, including when this feature is not 
> > used? My biggest concern is that because this function is also used to 
> > build the command line for the stubdom and the stubdom is PV and hence 
> > has at least a ->kernel setting then this new code might break that use 
> > case, by adding these options when they are not wanted. This path is all 
> > a bit tangled so I'm not 100% sure if those fields are actually set or 
> > not.
> 
> 
> 
> '-kernel' is only added to qemu command line under HVM case. PV would
> 
> not be affected. And only added when device model is upstream qemu, but
> 
> not old qemu-xen. About stubdom, tested before, when stubdom is using

> old qemu-xen, would not be affected.

Ah yes, I'd forgotten we didn't have upstream stubdom yet, obviously any
issues here will become apparent whenever that gets implemented.

Ian.



_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: 答复： Re: [PATCH] missing chunk of HVM direct kernel boot patch

[Konrad Rzeszutek Wilk]

On Mon, Dec 01, 2014 at 09:21:24AM +0000, Ian Campbell wrote:
> On Mon, 2014-12-01 at 01:21 -0700, Chun Yan Liu wrote:
> > 
> > >>> Ian Campbell <Ian.Campbell@citrix.com> 2014-11-28 下午 20:01 >>> 
> > On Fri, 2014-11-28 at 13:55 +0800, Chunyan Liu wrote: 
> > >> Found by Stefano, this chunk of the patch was never applied to 
> > >> xen-unstable (commit 11dffa2359e8a2629490c14c029c7c7c777b3e47), 
> > >> see http://marc.info/?l=qemu-devel&m=140471493425353&w=2. 
> > >
> > > How strange, "git am" usually makes it pretty difficult to miss hunks. 
> > > Sorry about this. 
> > 
> > >> Signed-off-by: Chunyan Liu <cyliu@suse.com> 
> > 
> > > Acked-by: Ian Campbell <ian.campbell@citrix.com> 
> > 
> > > I'm afraid that despite the circumstances this still needs a release ack 
> > > from Konrad. Obviously the upside is fixing a partially implemented 
> > > feature, but I'm not sure what the downsides are. 
> > >
> > > Has this been tested with stubdoms, including when this feature is not 
> > > used? My biggest concern is that because this function is also used to 
> > > build the command line for the stubdom and the stubdom is PV and hence 
> > > has at least a ->kernel setting then this new code might break that use 
> > > case, by adding these options when they are not wanted. This path is all 
> > > a bit tangled so I'm not 100% sure if those fields are actually set or 
> > > not.
> > 
> > 
> > 
> > '-kernel' is only added to qemu command line under HVM case. PV would
> > 
> > not be affected. And only added when device model is upstream qemu, but
> > 
> > not old qemu-xen. About stubdom, tested before, when stubdom is using
> 
> > old qemu-xen, would not be affected.
> 
> Ah yes, I'd forgotten we didn't have upstream stubdom yet, obviously any
> issues here will become apparent whenever that gets implemented.

<nods>

Release-Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> 
> Ian.
> 
> 

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: 答复： Re: [PATCH] missing chunk of HVM direct kernel boot patch

[Ian Campbell]

On Mon, 2014-12-01 at 16:51 -0500, Konrad Rzeszutek Wilk wrote:
> On Mon, Dec 01, 2014 at 09:21:24AM +0000, Ian Campbell wrote:
> > On Mon, 2014-12-01 at 01:21 -0700, Chun Yan Liu wrote:
> > > 
> > > >>> Ian Campbell <Ian.Campbell@citrix.com> 2014-11-28 下午 20:01 >>> 
> > > On Fri, 2014-11-28 at 13:55 +0800, Chunyan Liu wrote: 
> > > >> Found by Stefano, this chunk of the patch was never applied to 
> > > >> xen-unstable (commit 11dffa2359e8a2629490c14c029c7c7c777b3e47), 
> > > >> see http://marc.info/?l=qemu-devel&m=140471493425353&w=2. 
> > > >
> > > > How strange, "git am" usually makes it pretty difficult to miss hunks. 
> > > > Sorry about this. 
> > > 
> > > >> Signed-off-by: Chunyan Liu <cyliu@suse.com> 
> > > 
> > > > Acked-by: Ian Campbell <ian.campbell@citrix.com> 
> > > 
> > > > I'm afraid that despite the circumstances this still needs a release ack 
> > > > from Konrad. Obviously the upside is fixing a partially implemented 
> > > > feature, but I'm not sure what the downsides are. 
> > > >
> > > > Has this been tested with stubdoms, including when this feature is not 
> > > > used? My biggest concern is that because this function is also used to 
> > > > build the command line for the stubdom and the stubdom is PV and hence 
> > > > has at least a ->kernel setting then this new code might break that use 
> > > > case, by adding these options when they are not wanted. This path is all 
> > > > a bit tangled so I'm not 100% sure if those fields are actually set or 
> > > > not.
> > > 
> > > 
> > > 
> > > '-kernel' is only added to qemu command line under HVM case. PV would
> > > 
> > > not be affected. And only added when device model is upstream qemu, but
> > > 
> > > not old qemu-xen. About stubdom, tested before, when stubdom is using
> > 
> > > old qemu-xen, would not be affected.
> > 
> > Ah yes, I'd forgotten we didn't have upstream stubdom yet, obviously any
> > issues here will become apparent whenever that gets implemented.
> 
> <nods>
> 
> Release-Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

Applied, thanks.

> > 
> > Ian.
> > 
> > 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel



_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel