On Tue, 2020-02-18 at 10:58 -0800, Mina Almasry wrote:
> On Mon, Feb 17, 2020 at 7:31 AM Qian Cai <cai@xxxxxx> wrote:
> >
> > The commit c32300516047 ("hugetlb_cgroup: add interface for
> > charge/uncharge hugetlb reservations") forgot to remove an unused
> > variable,
> >
> > mm/hugetlb_cgroup.c: In function 'hugetlb_cgroup_migrate':
> > mm/hugetlb_cgroup.c:777:25: warning: variable 'h_cg' set but not used
> > [-Wunused-but-set-variable]
> > struct hugetlb_cgroup *h_cg;
> > ^~~~
> >
> > Signed-off-by: Qian Cai <cai@xxxxxx>
> > ---
> > mm/hugetlb_cgroup.c | 2 --
> > 1 file changed, 2 deletions(-)
> >
> > diff --git a/mm/hugetlb_cgroup.c b/mm/hugetlb_cgroup.c
> > index ad777fecad28..8a86a2b62bef 100644
> > --- a/mm/hugetlb_cgroup.c
> > +++ b/mm/hugetlb_cgroup.c
> > @@ -774,7 +774,6 @@ void __init hugetlb_cgroup_file_init(void)
> > */
> > void hugetlb_cgroup_migrate(struct page *oldhpage, struct page *newhpage)
> > {
> > - struct hugetlb_cgroup *h_cg;
> > struct hugetlb_cgroup *h_cg_rsvd;
> > struct hstate *h = page_hstate(oldhpage);
> >
> > @@ -783,7 +782,6 @@ void hugetlb_cgroup_migrate(struct page *oldhpage, struct page *newhpage)
> >
> > VM_BUG_ON_PAGE(!PageHuge(oldhpage), oldhpage);
> > spin_lock(&hugetlb_lock);
> > - h_cg = hugetlb_cgroup_from_page(oldhpage);
> > h_cg_rsvd = hugetlb_cgroup_from_page_rsvd(oldhpage);
> > set_hugetlb_cgroup(oldhpage, NULL);
> >
> > --
> > 1.8.3.1
> >
>
> Hi Qian,
>
> Thank you very much for the fix to remove the warning, but actually
> the real fix is I'm missing a 'set_hugetlb_cgroup(newhpage, h_cg);'
> which will use the variable and set the cgroup on newhpage which is
> needed. I'll submit the proper fix.
>
> What bothers me though is that locally when I checkout the broken
> patch and try to build I don't see the warning:
>
> make -j80 mm/hugetlb_cgroup.o
> no warning.
> make -j80 mm/hugetlb_cgroup.o CFLAGS_KERNEL="-Wall"
> no warning
> make -j80 mm/hugetlb_cgroup.o CFLAGS_KERNEL="-Wunused-but-set-variable"
> I see the warning.
>
> So it seems there is a bunch of warnings I need to explicitly turn on
> otherwise I will continually submit patches that introduce warnings in
> your build. Any idea why I'm running into this? Do you also have to
> turn on these warnings manually on your make line? Is it related to
> gcc version? My gcc version is:
> gcc version 9.2.1 20190909 (Debian 9.2.1-8)
I am doing "make W=1" which will turn on those warnings. Quite noisy but you can
"grep" what you are interested in.
