Re: [PATCH -mm -V3 03/21] mm, THP, swap: Support PMD swap mapping in swap_duplicate()

[Huang\, Ying]

Daniel Jordan <daniel.m.jordan@xxxxxxxxxx> writes:

> On Tue, Jun 12, 2018 at 09:23:19AM +0800, Huang, Ying wrote:
>> Daniel Jordan <daniel.m.jordan@xxxxxxxxxx> writes:
>> >> +#else
>> >> +static inline int __swap_duplicate_cluster(swp_entry_t *entry,
>> >
>> > This doesn't need inline.
>> 
>> Why not?  This is just a one line stub.
>
> Forgot to respond to this.  The compiler will likely choose to optimize out
> calls to an empty function like this.  Checking, this is indeed what it does in
> this case on my machine, with or without inline.

Yes.  I believe a decent compiler will inline the function in any way.
And it does no harm to keep "inline" too, Yes?

> By the way, when building without CONFIG_THP_SWAP, we get
>
>   linux/mm/swapfile.c:933:13: warning: â__swap_free_clusterâ defined but not used [-Wunused-function]
>    static void __swap_free_cluster(struct swap_info_struct *si, unsigned long idx)
>                ^~~~~~~~~~~~~~~~~~~

Thanks!  I will fix this.  Don't know why 0-Day didn't catch this.

Best Regards,
Huang, Ying