Re: [Xen-devel] [PATCH v6 3/7] xen/x86: populate PVHv2 Dom0 physical memory map

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 14.02.17 at 11:19, <roger.pau@xxxxxxxxxx> wrote:
> On Tue, Feb 14, 2017 at 10:10:16AM +0000, Roger Pau Monne wrote:
>> On Mon, Feb 13, 2017 at 06:53:49AM -0700, Jan Beulich wrote:
>> > >>> On 10.02.17 at 13:33, <roger.pau@xxxxxxxxxx> wrote:
>> > > --- a/xen/include/asm-x86/page.h
>> > > +++ b/xen/include/asm-x86/page.h
>> > > @@ -374,6 +374,18 @@ perms_strictly_increased(uint32_t old_flags, 
>> > > uint32_t new_flags)
>> > >      return ((of | (of ^ nf)) == nf);
>> > >  }
>> > >  
>> > > +/* Build a 32bit PSE page table using 4MB pages. */
>> > > +static inline void
>> > > +write_32bit_pse_identmap(uint32_t *l2)
>> > 
>> > Why does this need to be an inline function?
>> 
>> Given it's size and the low number of callers I though it would be better to
>> make it inline, but since this is not in any performance critical path I'm
>> going to remove the inlining, although I think the compiler is probably going
>> to do it anyway.
> 
> Right, now I remember why it needs to be inline:
> 
> xen/include/asm/page.h:379:1: error: unused function 
> 'write_32bit_pse_identmap'
>       [-Werror,-Wunused-function]
> write_32bit_pse_identmap(uint32_t *l2)

Funny - when sending the other reply a second ago I did think about
whether this needs clarification, and I decide it's obvious enough.
But it looks like I was wrong: Not inlining the function of course means
moving its definition to a C file, keeping just the declaration here.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel