Re: [PATCH 1/3] x86/mem_access: Make the mem_access ops generic

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 11/04/14 10:48, Jan Beulich wrote:
>>>> On 11.04.14 at 11:35, <andrew.cooper3@citrix.com> wrote:
>> On 11/04/14 04:49, Aravindh Puthiyaparambil wrote:
>>> --- a/xen/arch/x86/mm/mem_access.c
>>> +++ b/xen/arch/x86/mm/mem_access.c
>>> @@ -21,31 +21,98 @@
>>>   */
>>>  
>>>  
>>> +#include <xen/sched.h>
>>> +#include <xen/guest_access.h>
>>> +#include <xen/hypercall.h>
>>>  #include <asm/p2m.h>
>>>  #include <asm/mem_event.h>
>>> +#include <xsm/xsm.h>
>>>  
>>>  
>>> -int mem_access_memop(struct domain *d, xen_mem_event_op_t *meo)
>>> +#define ACCESS_op_mask  0xff
>>> +
>>> +int mem_access_memop(XEN_GUEST_HANDLE_PARAM(void) arg, unsigned long 
>> start_iter)
>>>  {
>>> -    int rc;
>>> +    long rc;
>> int function and long rc? The function wants promoting to long.
> No, what makes it out of the function fits in an "int", 

Why does that matter?  It is a recipe for accidental truncation with
future changes (and certainly not the first example in Xen).

All hypercall handlers should be returning longs, and leaving it to the
compat layer to apply correct truncation if needed.

~Andrew