[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [libvirt] [PATCH v2] virt-host-validate: Fix warning for IOMMU detection on PPC



Hello John,

In case of PPC, IOMMU in the host kernel either has it or not compiled in.
The  /sys/kernel/iommu_groups check is good enough to verify if it was compiled with the kernel or not.

If not, then we can have a ppc specific message there:   


    virHostMsgCheck(hvname, "%s", _("if IOMMU is enabled by kernel"));
    if (sb.st_nlink <= 2) {
        if(!isPPC) {
            virHostMsgFail(level,
                           "IOMMU appears to be disabled in kernel. "
                           "Add %s to kernel cmdline arguments", bootarg);
        } else {
            virHostMsgFail(level,
                           "IOMMU capability not compiled into kernel. ");

        }
        return -1;
    }
    virHostMsgPass();
    return 0;



On Wed, Aug 23, 2017 at 7:06 PM, John Ferlan <jferlan redhat com> wrote:


On 08/17/2017 09:48 AM, Nitesh Konkar wrote:
> Fix the warning generated on PPC by virt-host-validate
> for IOMMU
>
> Signed-off-by: Nitesh Konkar <nitkon12 linux vnet ibm com>
> ---
>  tools/virt-host-validate-common.c | 10 +++++++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/tools/virt-host-validate-common.c b/tools/virt-host-validate-common.c
> index 6faed04..51fa8c0 100644
> --- a/tools/virt-host-validate-common.c
> +++ b/tools/virt-host-validate-common.c
> @@ -35,6 +35,7 @@
>  #include "virfile.h"
>  #include "virt-host-validate-common.h"
>  #include "virstring.h"
> +#include "virarch.h"
>
>  #define VIR_FROM_THIS VIR_FROM_NONE
>
> @@ -442,8 +443,7 @@ int virHostValidateIOMMU(const char *hvname,
>      virBitmapPtr flags;
>      struct stat sb;
>      const char *bootarg = NULL;
> -    bool isAMD = false, isIntel = false;
> -
> +    bool isAMD = false, isIntel = false, isPPC = false;
>      flags = virHostValidateGetCPUFlags();
>
>      if (flags && virBitmapIsBitSet(flags, VIR_HOST_VALIDATE_CPU_FLAG_VMX))
> @@ -453,9 +453,10 @@ int virHostValidateIOMMU(const char *hvname,
>
>      virBitmapFree(flags);
>
> -    virHostMsgCheck(hvname, "%s", _("for device assignment IOMMU support"));
> +    isPPC = ARCH_IS_PPC64(virArchFromHost());
>
>      if (isIntel) {
> +        virHostMsgCheck(hvname, "%s", _("for device assignment IOMMU support"));
>          if (access("/sys/firmware/acpi/tables/DMAR", F_OK) == 0) {
>              virHostMsgPass();
>              bootarg = "intel_iommu=on";
> @@ -467,6 +468,7 @@ int virHostValidateIOMMU(const char *hvname,
>              return -1;
>          }
>      } else if (isAMD) {
> +        virHostMsgCheck(hvname, "%s", _("for device assignment IOMMU support"));
>          if (access("/sys/firmware/acpi/tables/IVRS", F_OK) == 0) {
>              virHostMsgPass();
>              bootarg = "iommu=pt iommu=1";
> @@ -477,6 +479,8 @@ int virHostValidateIOMMU(const char *hvname,
>                             "hardware platform");
>              return -1;
>          }
> +    } else if (isPPC) {
> +        /* Empty Block */

So there's nothing to check at all? Perhaps elaborate in the commit
message...


Still what happens when @bootarg isn't populated and by chance we fall
into the:

if (sb.st_nlink <= 2) {
}

condition below here?  You'll get "Add <nil>..."

Or can we not get to that code?  I'm not PPC and IOMMU aware, so I'm
asking...

Tks,

John

>      } else {
>          virHostMsgFail(level,
>                         "Unknown if this platform has IOMMU support");
>


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]