Re: [PATCH] libelf: Handle PVH kernels lacking ENTRY elfnote

[Jan Beulich <jbeulich@xxxxxxxx>]

On 19.10.2020 17:26, Jason Andryuk wrote:
> On Mon, Oct 19, 2020 at 3:38 AM Jan Beulich <jbeulich@xxxxxxxx> wrote:
>> On 16.10.2020 18:28, Jason Andryuk wrote:
>>> Looks like we can pass XC_DOM_PV_CONTAINER/XC_DOM_HVM_CONTAINER down
>>> into elf_xen_parse().  Then we would just validate phys_entry for HVM
>>> and virt_entry for PV.  Does that sound reasonable?
>>
>> I think so, yes. Assuming of course that you'll convert the XC_DOM_*
>> into a boolean, so that the hypervisor's use of libelf/ can also be
>> suitably adjusted.
> 
> Are you okay with:
> -int elf_xen_parse(struct elf_binary *elf,
> -                  struct elf_dom_parms *parms);
> +int elf_xen_parse_pvh(struct elf_binary *elf,
> +                      struct elf_dom_parms *parms);
> +int elf_xen_parse_pv(struct elf_binary *elf,
> +                     struct elf_dom_parms *parms);
> ?
> 
> I prefer avoiding boolean arguments since I find it helps readability.

And I view things the other way around. If readability is of
concern, how about adding an unsigned int flags parameter
and #define-ing two suitable constants to be passed?

And of course it's not me alone who needs to be okay with
whatever route you/we go.

Jan