Re: [Xen-devel] [PATCH 2/3] x86/hvm: Rearange check_segment() to use a switch statement

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 03/07/17 13:34, Jan Beulich wrote:
>>>> On 30.06.17 at 17:04, <andrew.cooper3@xxxxxxxxxx> wrote:
>> +    case x86_seg_ds:
>> +    case x86_seg_es:
>> +        if ( (reg->attr.fields.type & 0x8) && !(reg->attr.fields.type & 
>> 0x2) )
>> +        {
>> +            gprintk(XENLOG_ERR, "Non-readable segment provided for DS or 
>> ES\n");
>> +            return -EINVAL;
>> +        }
>> +        break;
>> +
>> +    default: /* -Werror=switch */
>> +        break;
>>      }
> Perhaps better to have
>
>     default:
>         ASSERT_UNREACHABLE();
>     case x86_seg_tr:
>         break;
>
> to make more visible that it is not an oversight that especially FS
> and GS aren't being handled here? Either way
> Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>

The x86_seg_tr case exits check_segment() rather earlier.  How about

default:
    ASSERT_UNREACHABLE();
    return -EINVAL;

?

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