Re: [Xen-devel] [PATCH 1/2] x86/mm: Plumb a error return through {paging, hap, shadow}_vcpu_init()

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 30.01.17 at 18:26, <andrew.cooper3@xxxxxxxxxx> wrote:
> --- a/xen/arch/x86/domain.c
> +++ b/xen/arch/x86/domain.c
> @@ -384,7 +384,8 @@ int vcpu_initialise(struct vcpu *v)
>  
>      if ( !is_idle_domain(d) )
>      {
> -        paging_vcpu_init(v);
> +        if ( (rc = paging_vcpu_init(v)) )
> +            return rc;
>  
>          if ( (rc = vcpu_init_fpu(v)) != 0 )
>              return rc;

With the function so far having been impossible to fail, it was also
pretty clear that no meaningful cleanup would need doing after it.
Now that it can fail, is it really appropriate to return without
undoing its effects if vcpu_init_fpu() fails?

Jan


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