[PATCH biosdevname]: handle dom0 on AMD systems

[PATCH biosdevname]: handle dom0 on AMD systems

[Olaf Hering]

[-- Attachment #1.1: Type: text/plain, Size: 1615 bytes --]

Starting with xen-4.7 cpuid() will return with the hypervisor bit set
in a dom0 when running on an AMD system. As a result biosdevname
thinks it runs in a guest and does nothing. Detect a dom0 by looking
into xenfs. This works with classic xenlinux based kernels and with
pvops based kernels.

Signed-off-by: Olaf Hering <olaf@aepfle.de>

---
 src/bios_dev_name.c |   31 +++++++++++++++++++++++++++++++
 1 file changed, 31 insertions(+)

--- a/src/bios_dev_name.c
+++ b/src/bios_dev_name.c
@@ -133,6 +133,31 @@ cpuid (u_int32_t eax, u_int32_t ecx)
 }
 
 /*
+  Starting with xen-4.7 cpuid will return with the hypervisor bit set
+  on AMD systems. This breaks biosdevname and network interface names.
+  Instead of relying on cpuid check for dom0 in xenfs.
+*/
+static int
+running_in_dom0(void)
+{
+    size_t len = 0;
+    char buf[16];
+    FILE *f = fopen("/proc/xen/capabilities", "r");
+
+    if (!f)
+        return 0;
+    memset(buf, 0, sizeof(buf));
+    len = fread(&buf, 1, sizeof(buf) - 1, f);
+    fclose(f);
+    while(len && --len && len < sizeof(buf)) {
+        if (buf[len] == '\n')
+            buf[len] = '\0';
+    }
+    len = !strcmp("control_d", buf);
+    return len;
+}
+
+/*
   Algorithm suggested by:
   http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1009458
 */
@@ -144,7 +171,11 @@ running_in_virtual_machine (void)
 
     ecx = cpuid (eax, ecx);
     if (ecx & 0x80000000U)
+    {
+       if (running_in_dom0())
+           return 0;
        return 1;
+    }
     return 0;
 }
 

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[-- Attachment #2: Type: text/plain, Size: 127 bytes --]

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

Re: [PATCH biosdevname]: handle dom0 on AMD systems

[Jan Beulich]

>>> On 17.08.16 at 12:30, <olaf@aepfle.de> wrote:
> @@ -144,7 +171,11 @@ running_in_virtual_machine (void)
>  
>      ecx = cpuid (eax, ecx);
>      if (ecx & 0x80000000U)
> +    {
> +       if (running_in_dom0())
> +           return 0;
>         return 1;
> +    }
>      return 0;
>  }

Isn't this backwards? Dom0 _is_ a virtual machine.

Jan


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

Re: [PATCH biosdevname]: handle dom0 on AMD systems

[Andrew Cooper]

On 17/08/16 12:51, Jan Beulich wrote:
>>>> On 17.08.16 at 12:30, <olaf@aepfle.de> wrote:
>> @@ -144,7 +171,11 @@ running_in_virtual_machine (void)
>>  
>>      ecx = cpuid (eax, ecx);
>>      if (ecx & 0x80000000U)
>> +    {
>> +       if (running_in_dom0())
>> +           return 0;
>>         return 1;
>> +    }
>>      return 0;
>>  }
> Isn't this backwards? Dom0 _is_ a virtual machine.

biosdevname uses it as a "I should silently exit and do nothing" check.

This is unhelpful behaviour as dom0 is responsible for the real hardware.

~Andrew

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

Re: [PATCH biosdevname]: handle dom0 on AMD systems

[Olaf Hering]

[-- Attachment #1.1: Type: text/plain, Size: 1806 bytes --]

Hello,

and ping.

Olaf


On Wed, Aug 17, Olaf Hering wrote:

> Starting with xen-4.7 cpuid() will return with the hypervisor bit set
> in a dom0 when running on an AMD system. As a result biosdevname
> thinks it runs in a guest and does nothing. Detect a dom0 by looking
> into xenfs. This works with classic xenlinux based kernels and with
> pvops based kernels.
> 
> Signed-off-by: Olaf Hering <olaf@aepfle.de>
> 
> ---
>  src/bios_dev_name.c |   31 +++++++++++++++++++++++++++++++
>  1 file changed, 31 insertions(+)
> 
> --- a/src/bios_dev_name.c
> +++ b/src/bios_dev_name.c
> @@ -133,6 +133,31 @@ cpuid (u_int32_t eax, u_int32_t ecx)
>  }
>  
>  /*
> +  Starting with xen-4.7 cpuid will return with the hypervisor bit set
> +  on AMD systems. This breaks biosdevname and network interface names.
> +  Instead of relying on cpuid check for dom0 in xenfs.
> +*/
> +static int
> +running_in_dom0(void)
> +{
> +    size_t len = 0;
> +    char buf[16];
> +    FILE *f = fopen("/proc/xen/capabilities", "r");
> +
> +    if (!f)
> +        return 0;
> +    memset(buf, 0, sizeof(buf));
> +    len = fread(&buf, 1, sizeof(buf) - 1, f);
> +    fclose(f);
> +    while(len && --len && len < sizeof(buf)) {
> +        if (buf[len] == '\n')
> +            buf[len] = '\0';
> +    }
> +    len = !strcmp("control_d", buf);
> +    return len;
> +}
> +
> +/*
>    Algorithm suggested by:
>    http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1009458
>  */
> @@ -144,7 +171,11 @@ running_in_virtual_machine (void)
>  
>      ecx = cpuid (eax, ecx);
>      if (ecx & 0x80000000U)
> +    {
> +       if (running_in_dom0())
> +           return 0;
>         return 1;
> +    }
>      return 0;
>  }
>  



[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 127 bytes --]

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