Re: [Xen-devel] [PATCH v4] x86: relocate pvh_info

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 23.01.18 at 12:44, <wei.liu2@xxxxxxxxxx> wrote:
> On Tue, Jan 23, 2018 at 02:14:51AM -0700, Jan Beulich wrote:
>> >>> On 22.01.18 at 19:31, <wei.liu2@xxxxxxxxxx> wrote:
>> > On Mon, Jan 22, 2018 at 06:19:43PM +0000, Andrew Cooper wrote:
>> >> On 22/01/18 18:17, Wei Liu wrote:
>> >> > So you want reloc.o to contain pvh_info_reloc unconditionally?
>> >> >
>> >> > Fundamentally I don't think I care enough about all the bikeshedding so
>> >> > if Jan and you agree on this I will just make the change.
>> >> 
>> >> It wont.  The function will be dropped due to DCE, but we'll spot build
>> >> breakages far more easily.  (The important bit is that the function call
>> >> is guarded by the IS_ENABLED())
>> > 
>> > reloc.o will still have that function in non-PVH build on my machine.
>> > And that's with the following diff applied.
>> 
>> Well, DCE doesn't make any promises towards what it is able to
>> eliminate, which is why generally I prefer to help the compiler in
>> cases like the one here.
> 
> If I read this correctly, this means you prefer the ifdef CONFIG_PVH_GUEST
> version?

Well, in an ideal world I'd prefer what Andrew did suggest. If the
compiler turns out to be incapable of removing dead code
properly, I'd be slightly in favor of the #ifdef approach, but with
Andrew apparently strongly feeling the other way around, I
wouldn't mind that variant (the more that it's all .init.* contents).

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel