On 02.01.2025 19:33, Stefano Stabellini wrote:
> On Fri, 27 Dec 2024, Jan Beulich wrote:
>> On 19.12.2024 10:29, Sergiy Kibrik wrote:
>>> --- a/xen/common/Kconfig
>>> +++ b/xen/common/Kconfig
>>> @@ -516,4 +516,33 @@ config TRACEBUFFER
>>>       to be collected at run time for debugging or performance analysis.
>>>       Memory and execution overhead when not active is minimal.
>>>  
>>> +menu "Supported hypercall interfaces"
>>> +   visible if DOM0LESS_BOOT && EXPERT
>>> +
>>> +config SYSCTL
>>> +   bool "Enable sysctl hypercall"
>>> +   depends on !PV_SHIM_EXCLUSIVE
>>> +   default y
>>> +
>>> +config DOMCTL
>>> +   bool "Enable domctl hypercalls"
>>> +   depends on !PV_SHIM_EXCLUSIVE
>>> +   default y
>>> +
>>> +config HVM_OP
>>> +   bool "Enable HVM hypercalls"
>>> +   depends on HVM
>>> +   default y
>>> +
>>> +config PLATFORM_OP
>>> +   bool "Enable platform hypercalls"
>>> +   depends on !PV_SHIM_EXCLUSIVE
>>> +   default y
>>
>> Just to re-iterate an earlier comment: Andrew (imo validly) raised concern of
>> such negative dependencies. As said before, imo we'd better resolve that 
>> before
>> extending the issue (whether by the patch I once sent or something else is 
>> then
>> secondary).
> 
> How would you express the !PV_SHIM_EXCLUSIVE dependency without using
> negative dependencies?

By inverting the sense of the option (and renaming it), as (to a 1st 
approximation)
requested by Andrew long ago, and as proposed in [1], which I think I pointed
Sergiy at, and which continues to be lacking feedback.

Jan

[1] https://lists.xen.org/archives/html/xen-devel/2023-03/msg00040.html

Reply via email to