Re: [PATCH] x86/pv: Drop HYPERVISOR_COMPAT_VIRT_START()

[Jan Beulich <jbeulich@xxxxxxxx>]

On 27.04.2021 20:05, Andrew Cooper wrote:
> On 27/04/2021 14:18, Jan Beulich wrote:
>> On 27.04.2021 15:02, Andrew Cooper wrote:
>>> --- a/xen/include/asm-x86/config.h
>>> +++ b/xen/include/asm-x86/config.h
>>> @@ -254,21 +254,16 @@ extern unsigned char boot_edid_info[128];
>>>  
>>>  /* This is not a fixed value, just a lower limit. */
>>>  #define __HYPERVISOR_COMPAT_VIRT_START 0xF5800000
>>> -#define HYPERVISOR_COMPAT_VIRT_START(d) ((d)->arch.hv_compat_vstart)
>>> -
>>> -#else /* !CONFIG_PV32 */
>>> -
>>> -#define HYPERVISOR_COMPAT_VIRT_START(d) ((void)(d), 0)
>> ... this to allow things to build in the absence of the actual struct
>> member.
> 
> Hmm - I really should have got this change in earlier, shouldn't I...
> 
> For this example you've pointed out, feeding 0 into the limit
> calculation is not a correct thing to do in the slightest.

Does it actually get fed into any such calculation? From looking
around yesterday as well as from when I made that change over
half a year ago I seem to recall that all potentially problematic
uses are already suitably guarded.

Jan