[Qemu-devel] [PATCH] target-arm: Check error conditions on kvm_arm_reset_vcpu

[Qemu-devel] [PATCH] target-arm: Check error conditions on kvm_arm_reset_vcpu

[Christoffer Dall]

When resetting a VCPU we currently call both kvm_arm_vcpu_init() and
write_kvmstate_to_list(), both of which can fail, but we never check the
return value.

The only choice here is to print an error an exit if the calls fail.

Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
---
 target-arm/kvm32.c | 13 +++++++++++--
 target-arm/kvm64.c |  8 +++++++-
 2 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/target-arm/kvm32.c b/target-arm/kvm32.c
index 5ec4eb1..1fd0e83 100644
--- a/target-arm/kvm32.c
+++ b/target-arm/kvm32.c
@@ -511,9 +511,18 @@ int kvm_arch_get_registers(CPUState *cs)
 
 void kvm_arm_reset_vcpu(ARMCPU *cpu)
 {
+    int ret;
+
     /* Re-init VCPU so that all registers are set to
      * their respective reset values.
      */
-    kvm_arm_vcpu_init(CPU(cpu));
-    write_kvmstate_to_list(cpu);
+    ret = kvm_arm_vcpu_init(CPU(cpu));
+    if (ret < 0) {
+        fprintf(stderr, "kvm_arm_vcpu_init failed: %s\n", strerror(ret));
+        abort();
+    }
+    if (!write_kvmstate_to_list(cpu)) {
+        fprintf(stderr, "write_kvmstate_to_list failed\n");
+        abort();
+    }
 }
diff --git a/target-arm/kvm64.c b/target-arm/kvm64.c
index c615286..101354a 100644
--- a/target-arm/kvm64.c
+++ b/target-arm/kvm64.c
@@ -263,8 +263,14 @@ int kvm_arch_get_registers(CPUState *cs)
 
 void kvm_arm_reset_vcpu(ARMCPU *cpu)
 {
+    int ret;
+
     /* Re-init VCPU so that all registers are set to
      * their respective reset values.
      */
-    kvm_arm_vcpu_init(CPU(cpu));
+    ret = kvm_arm_vcpu_init(CPU(cpu));
+    if (ret < 0) {
+        fprintf(stderr, "kvm_arm_vcpu_init failed: %s\n", strerror(ret));
+        abort();
+    }
 }
-- 
2.1.2.330.g565301e.dirty

Re: [Qemu-devel] [PATCH] target-arm: Check error conditions on kvm_arm_reset_vcpu

[Peter Maydell]

On 3 December 2014 at 20:17, Christoffer Dall
<christoffer.dall@linaro.org> wrote:
> When resetting a VCPU we currently call both kvm_arm_vcpu_init() and
> write_kvmstate_to_list(), both of which can fail, but we never check the
> return value.
>
> The only choice here is to print an error an exit if the calls fail.

I like this patch, but it is going to conflict with a patch I'm
nearly done testing and plan to send tomorrow that unifies the
sysreg handling for 32 and 64 bit KVM/ARM. Part of what that
involves is using a single kvm_arm_reset_vcpu() in kvm.c rather
than different 32 and 64 bit versions. So you should probably
respin this patch on top of that one.

> Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
> ---
>  target-arm/kvm32.c | 13 +++++++++++--
>  target-arm/kvm64.c |  8 +++++++-
>  2 files changed, 18 insertions(+), 3 deletions(-)
>
> diff --git a/target-arm/kvm32.c b/target-arm/kvm32.c
> index 5ec4eb1..1fd0e83 100644
> --- a/target-arm/kvm32.c
> +++ b/target-arm/kvm32.c
> @@ -511,9 +511,18 @@ int kvm_arch_get_registers(CPUState *cs)
>
>  void kvm_arm_reset_vcpu(ARMCPU *cpu)
>  {
> +    int ret;
> +
>      /* Re-init VCPU so that all registers are set to
>       * their respective reset values.
>       */
> -    kvm_arm_vcpu_init(CPU(cpu));
> -    write_kvmstate_to_list(cpu);
> +    ret = kvm_arm_vcpu_init(CPU(cpu));
> +    if (ret < 0) {
> +        fprintf(stderr, "kvm_arm_vcpu_init failed: %s\n", strerror(ret));

Shouldn't this be strerror(-ret) ?

thanks
-- PMM

Re: [Qemu-devel] [PATCH] target-arm: Check error conditions on kvm_arm_reset_vcpu

[Christoffer Dall]

On Thu, Dec 4, 2014 at 8:13 PM, Peter Maydell <peter.maydell@linaro.org> wrote:
> On 3 December 2014 at 20:17, Christoffer Dall
> <christoffer.dall@linaro.org> wrote:
>> When resetting a VCPU we currently call both kvm_arm_vcpu_init() and
>> write_kvmstate_to_list(), both of which can fail, but we never check the
>> return value.
>>
>> The only choice here is to print an error an exit if the calls fail.
>
> I like this patch, but it is going to conflict with a patch I'm
> nearly done testing and plan to send tomorrow that unifies the
> sysreg handling for 32 and 64 bit KVM/ARM. Part of what that
> involves is using a single kvm_arm_reset_vcpu() in kvm.c rather
> than different 32 and 64 bit versions. So you should probably
> respin this patch on top of that one.
>

ok, will do, or you can squash this into yours, as you like.

>> Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
>> ---
>>  target-arm/kvm32.c | 13 +++++++++++--
>>  target-arm/kvm64.c |  8 +++++++-
>>  2 files changed, 18 insertions(+), 3 deletions(-)
>>
>> diff --git a/target-arm/kvm32.c b/target-arm/kvm32.c
>> index 5ec4eb1..1fd0e83 100644
>> --- a/target-arm/kvm32.c
>> +++ b/target-arm/kvm32.c
>> @@ -511,9 +511,18 @@ int kvm_arch_get_registers(CPUState *cs)
>>
>>  void kvm_arm_reset_vcpu(ARMCPU *cpu)
>>  {
>> +    int ret;
>> +
>>      /* Re-init VCPU so that all registers are set to
>>       * their respective reset values.
>>       */
>> -    kvm_arm_vcpu_init(CPU(cpu));
>> -    write_kvmstate_to_list(cpu);
>> +    ret = kvm_arm_vcpu_init(CPU(cpu));
>> +    if (ret < 0) {
>> +        fprintf(stderr, "kvm_arm_vcpu_init failed: %s\n", strerror(ret));
>
> Shouldn't this be strerror(-ret) ?
>
ah right, the init call returns a negative errno, but strerror expects
the positive one.

-Christoffer