[Qemu-devel] [for-2.0 0/2] A15 board bugfixes

[Qemu-devel] [for-2.0 0/2] A15 board bugfixes

[Peter Maydell]

This patchset fixes a couple of issues with A15 boards.

The first is a really obvious one -- running QEMU with
'-M midway -cpu bang' segfaults.

The second one is a problem with running an A15 board
model under KVM with "-cpu host": since the 'host' CPU object
doesn't have a 'reset-cbar' QOM property we need to quietly
continue in that case rather than failing. (Ideally we'd
support lying to the guest about its CPU in the kernel,
and then we could just say "-cpu host only makes sense for
the virt machine", but we're not there right now.)

Peter Maydell (2):
  hw/arm/highbank: Don't segfault on unknown CPU names
  hw/arm/vexpress, hw/arm/highbank: Don't insist that CPU has reset-cbar
    property

 hw/arm/highbank.c | 19 ++++++++++++-------
 hw/arm/vexpress.c |  7 +++----
 2 files changed, 15 insertions(+), 11 deletions(-)

-- 
1.9.0

[Qemu-devel] [for-2.0 1/2] hw/arm/highbank: Don't segfault on unknown CPU names

[Peter Maydell]

If the user passes an unknown CPU name via the '-cpu' option, exit
with an error message rather than segfaulting.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 hw/arm/highbank.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/hw/arm/highbank.c b/hw/arm/highbank.c
index f66d57b..03545aa 100644
--- a/hw/arm/highbank.c
+++ b/hw/arm/highbank.c
@@ -233,6 +233,11 @@ static void calxeda_init(QEMUMachineInitArgs *args, enum cxmachines machine)
         ARMCPU *cpu;
         Error *err = NULL;
 
+        if (!oc) {
+            fprintf(stderr, "Unable to find CPU definition\n");
+            exit(1);
+        }
+
         cpu = ARM_CPU(object_new(object_class_get_name(oc)));
 
         object_property_set_int(OBJECT(cpu), MPCORE_PERIPHBASE, "reset-cbar",
-- 
1.9.0

Re: [Qemu-devel] [for-2.0 1/2] hw/arm/highbank: Don't segfault on unknown CPU names

[Andreas Färber]

Am 03.04.2014 15:56, schrieb Peter Maydell:
> If the user passes an unknown CPU name via the '-cpu' option, exit
> with an error message rather than segfaulting.
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
>  hw/arm/highbank.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/hw/arm/highbank.c b/hw/arm/highbank.c
> index f66d57b..03545aa 100644
> --- a/hw/arm/highbank.c
> +++ b/hw/arm/highbank.c
> @@ -233,6 +233,11 @@ static void calxeda_init(QEMUMachineInitArgs *args, enum cxmachines machine)
>          ARMCPU *cpu;
>          Error *err = NULL;
>  
> +        if (!oc) {
> +            fprintf(stderr, "Unable to find CPU definition\n");

Old habits die hard... error_report() please.

Otherwise looks good.

One side effect of the cpu_arm_init() inlining here is that it didn't
get the properties support for -cpu, I note.

Regards,
Andreas

> +            exit(1);
> +        }
> +
>          cpu = ARM_CPU(object_new(object_class_get_name(oc)));
>  
>          object_property_set_int(OBJECT(cpu), MPCORE_PERIPHBASE, "reset-cbar",
> 


-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg

[Qemu-devel] [for-2.0 2/2] hw/arm/vexpress, hw/arm/highbank: Don't insist that CPU has reset-cbar property

[Peter Maydell]

For the machine models which can have a Cortex-A15 CPU (vexpress-a15 and
midway), silently continue if the CPU object has no reset-cbar property
rather than failing. This allows these boards to be used under KVM with
the "-cpu host" option, since the 'host' CPU object has no reset-cbar
property.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 hw/arm/highbank.c | 14 +++++++-------
 hw/arm/vexpress.c |  7 +++----
 2 files changed, 10 insertions(+), 11 deletions(-)

diff --git a/hw/arm/highbank.c b/hw/arm/highbank.c
index 03545aa..83e43fd 100644
--- a/hw/arm/highbank.c
+++ b/hw/arm/highbank.c
@@ -230,6 +230,7 @@ static void calxeda_init(QEMUMachineInitArgs *args, enum cxmachines machine)
 
     for (n = 0; n < smp_cpus; n++) {
         ObjectClass *oc = cpu_class_by_name(TYPE_ARM_CPU, cpu_model);
+        Object *cpuobj;
         ARMCPU *cpu;
         Error *err = NULL;
 
@@ -238,15 +239,14 @@ static void calxeda_init(QEMUMachineInitArgs *args, enum cxmachines machine)
             exit(1);
         }
 
-        cpu = ARM_CPU(object_new(object_class_get_name(oc)));
+        cpuobj = object_new(object_class_get_name(oc));
+        cpu = ARM_CPU(cpuobj);
 
-        object_property_set_int(OBJECT(cpu), MPCORE_PERIPHBASE, "reset-cbar",
-                                &err);
-        if (err) {
-            error_report("%s", error_get_pretty(err));
-            exit(1);
+        if (object_property_find(cpuobj, "reset-cbar", NULL)) {
+            object_property_set_int(cpuobj, MPCORE_PERIPHBASE,
+                                    "reset-cbar", &error_abort);
         }
-        object_property_set_bool(OBJECT(cpu), true, "realized", &err);
+        object_property_set_bool(cpuobj, true, "realized", &err);
         if (err) {
             error_report("%s", error_get_pretty(err));
             exit(1);
diff --git a/hw/arm/vexpress.c b/hw/arm/vexpress.c
index 67628af..169eb06 100644
--- a/hw/arm/vexpress.c
+++ b/hw/arm/vexpress.c
@@ -192,10 +192,9 @@ static void init_cpus(const char *cpu_model, const char *privdev,
         Object *cpuobj = object_new(object_class_get_name(cpu_oc));
         Error *err = NULL;
 
-        object_property_set_int(cpuobj, periphbase, "reset-cbar", &err);
-        if (err) {
-            error_report("%s", error_get_pretty(err));
-            exit(1);
+        if (object_property_find(cpuobj, "reset-cbar", NULL)) {
+            object_property_set_int(cpuobj, periphbase,
+                                    "reset-cbar", &error_abort);
         }
         object_property_set_bool(cpuobj, true, "realized", &err);
         if (err) {
-- 
1.9.0

Re: [Qemu-devel] [for-2.0 0/2] A15 board bugfixes

[Rob Herring]

On Thu, Apr 3, 2014 at 8:56 AM, Peter Maydell <peter.maydell@linaro.org> wrote:
> This patchset fixes a couple of issues with A15 boards.
>
> The first is a really obvious one -- running QEMU with
> '-M midway -cpu bang' segfaults.
>
> The second one is a problem with running an A15 board
> model under KVM with "-cpu host": since the 'host' CPU object
> doesn't have a 'reset-cbar' QOM property we need to quietly
> continue in that case rather than failing. (Ideally we'd
> support lying to the guest about its CPU in the kernel,
> and then we could just say "-cpu host only makes sense for
> the virt machine", but we're not there right now.)
>
> Peter Maydell (2):
>   hw/arm/highbank: Don't segfault on unknown CPU names
>   hw/arm/vexpress, hw/arm/highbank: Don't insist that CPU has reset-cbar
>     property

For both:

Reviewed-by: Rob Herring <rob.herring@linaro.org>