intel_iommu: Return error on machines with no PCI

[Qemu-devel] [PATCH] intel_iommu: Return error on machines with no PCI

Posted by Mohammed Gamal 6 years, 7 months ago

Starting the following command line causes a segfault
qemu-system-x86_64 -S -machine isapc,accel=kvm -device intel-iommu

This is due to the fact that the machine type 'isapc' doesn't have
a PCI bus, while intel_iommu doesn't check if the machine has PCI support
and subsequently does a null-pointer access. Intel IOMMU shouldn't even work
if the target machine doesn't have PCI.

Add a check for PCI on the given machine type and return an error if PCI
is not supported.

Signed-off-by: Mohammed Gamal <mgamal@redhat.com>
---
 hw/i386/intel_iommu.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
index 3a5bb0b..fab0b4b 100644
--- a/hw/i386/intel_iommu.c
+++ b/hw/i386/intel_iommu.c
@@ -3041,6 +3041,13 @@ static void vtd_realize(DeviceState *dev, Error **errp)
     }
 
     bus = pcms->bus;
+
+    if (!bus) {
+        error_setg(errp, "Machine-type '%s' does not support PCI",
+                   mc->name);
+        return;
+    }
+
     x86_iommu->type = TYPE_INTEL;
 
     if (!vtd_decide_config(s, errp)) {
-- 
1.8.3.1

Re: [Qemu-devel] [PATCH] intel_iommu: Return error on machines with no PCI

Posted by Thomas Huth 6 years, 7 months ago

On 14.09.2017 22:17, Mohammed Gamal wrote:
> Starting the following command line causes a segfault
> qemu-system-x86_64 -S -machine isapc,accel=kvm -device intel-iommu
> 
> This is due to the fact that the machine type 'isapc' doesn't have
> a PCI bus, while intel_iommu doesn't check if the machine has PCI support
> and subsequently does a null-pointer access. Intel IOMMU shouldn't even work
> if the target machine doesn't have PCI.
> 
> Add a check for PCI on the given machine type and return an error if PCI
> is not supported.
> 
> Signed-off-by: Mohammed Gamal <mgamal@redhat.com>
> ---
>  hw/i386/intel_iommu.c | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
> index 3a5bb0b..fab0b4b 100644
> --- a/hw/i386/intel_iommu.c
> +++ b/hw/i386/intel_iommu.c
> @@ -3041,6 +3041,13 @@ static void vtd_realize(DeviceState *dev, Error **errp)
>      }
>  
>      bus = pcms->bus;
> +
> +    if (!bus) {
> +        error_setg(errp, "Machine-type '%s' does not support PCI",
> +                   mc->name);
> +        return;
> +    }
> +
>      x86_iommu->type = TYPE_INTEL;
>  
>      if (!vtd_decide_config(s, errp)) {
> 

Patch looks basically fine to me, too, but I think I'd also rather
change the "if (!pcms)" at the beginning of the function into "if (!pcms
|| !pcms->bus)" here to use the same error message for both cases.

 Thomas