| Message ID | 1377714003-22096-1-git-send-email-robherring2@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 28 August 2013 19:20, Rob Herring <robherring2@gmail.com> wrote: > From: Rob Herring <rob.herring@calxeda.com> > > xen_pm_init was unconditionally setting pm_power_off and arm_pm_restart > function pointers. This breaks multi-platform kernels. Make this > conditional on running as a Xen guest and make it a late_initcall to > ensure it is setup after platform code for Dom0. > > Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> > Signed-off-by: Rob Herring <rob.herring@calxeda.com> > --- > arch/arm/xen/enlighten.c | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/arch/arm/xen/enlighten.c b/arch/arm/xen/enlighten.c > index 8a6295c..13a7d1f 100644 > --- a/arch/arm/xen/enlighten.c > +++ b/arch/arm/xen/enlighten.c > @@ -273,12 +273,15 @@ core_initcall(xen_guest_init); > > static int __init xen_pm_init(void) > { > + if (!of_find_compatible_node(NULL, NULL, "xen,xen")) > + return -ENODEV; > + You should use the macro xen_domain() to check if we are running in a Xen guest. Cheers,
On Wed, 28 Aug 2013, Julien Grall wrote: > On 28 August 2013 19:20, Rob Herring <robherring2@gmail.com> wrote: > > From: Rob Herring <rob.herring@calxeda.com> > > > > xen_pm_init was unconditionally setting pm_power_off and arm_pm_restart > > function pointers. This breaks multi-platform kernels. Make this > > conditional on running as a Xen guest and make it a late_initcall to > > ensure it is setup after platform code for Dom0. > > > > Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> > > Signed-off-by: Rob Herring <rob.herring@calxeda.com> > > --- > > arch/arm/xen/enlighten.c | 5 ++++- > > 1 file changed, 4 insertions(+), 1 deletion(-) > > > > diff --git a/arch/arm/xen/enlighten.c b/arch/arm/xen/enlighten.c > > index 8a6295c..13a7d1f 100644 > > --- a/arch/arm/xen/enlighten.c > > +++ b/arch/arm/xen/enlighten.c > > @@ -273,12 +273,15 @@ core_initcall(xen_guest_init); > > > > static int __init xen_pm_init(void) > > { > > + if (!of_find_compatible_node(NULL, NULL, "xen,xen")) > > + return -ENODEV; > > + > > You should use the macro xen_domain() to check if we are running > in a Xen guest. Yep. Aside from that the patch is fine. Thanks for spotting and fixing this issue. I'll add the next version of this patch to the xen queue.
diff --git a/arch/arm/xen/enlighten.c b/arch/arm/xen/enlighten.c index 8a6295c..13a7d1f 100644 --- a/arch/arm/xen/enlighten.c +++ b/arch/arm/xen/enlighten.c @@ -273,12 +273,15 @@ core_initcall(xen_guest_init); static int __init xen_pm_init(void) { + if (!of_find_compatible_node(NULL, NULL, "xen,xen")) + return -ENODEV; + pm_power_off = xen_power_off; arm_pm_restart = xen_restart; return 0; } -subsys_initcall(xen_pm_init); +late_initcall(xen_pm_init); static irqreturn_t xen_arm_callback(int irq, void *arg) {