Re: [PATCH] staging: android: lowmemorykiller.c: Fix checkpatch warning

[Arve HjÃnnevÃg]

On Thu, Sep 1, 2016 at 11:42 AM, Joe Perches <joe@xxxxxxxxxxx> wrote:
> On Thu, 2016-09-01 at 20:33 +0200, Greg Kroah-Hartman wrote:
>> On Thu, Sep 01, 2016 at 11:19:41AM -0700, Joe Perches wrote:
>> > On Thu, 2016-09-01 at 17:36 +0200, Greg Kroah-Hartman wrote:
>> > > On Thu, Aug 25, 2016 at 06:26:48PM +0300, Andrey Utkin wrote:
>> > > > On Thu, Aug 25, 2016 at 11:10:25AM -0400, Anson Jacob wrote:
>> > > > > Fix checkpatch.pl 'line over 80 characters' warning
> []
>> > > > > diff --git a/drivers/staging/android/lowmemorykiller.c b/drivers/staging/android/lowmemorykiller.c
> []
>> > > > > @@ -92,8 +92,8 @@ static unsigned long lowmem_scan(struct shrinker *s, struct shrink_control *sc)
>> > > > >       int array_size = ARRAY_SIZE(lowmem_adj);
>> > > > >       int other_free = global_page_state(NR_FREE_PAGES) - totalreserve_pages;
>> > > > >       int other_file = global_node_page_state(NR_FILE_PAGES) -
>> > > > > -                                             global_node_page_state(NR_SHMEM) -
>> > > > > -                                             total_swapcache_pages();
>> > > > > +                             global_node_page_state(NR_SHMEM) -
>> > > > > +                             total_swapcache_pages();
>> > > > >
>> > > > >       if (lowmem_adj_size < array_size)
>> > > > >               array_size = lowmem_adj_size;
>> > > > Indeed original alignment on opening brace here doesn't make sense.
>> > > > I don't mind if your patch gets accepted, but out of curiosity I have opened
>> > > > this file in vim with kernel coding style plugin loaded
>> > > > (https://github.com/vivien/vim-linux-coding-style), selected the affected
>> > > > lines, and reindented the selection by pressing 'gq'. It ended up with this,
>> > > > being both shorter and simpler:
>> > > >
>> > > > diff --git a/drivers/staging/android/lowmemorykiller.c b/drivers/staging/android/lowmemorykiller.c
> []
>> > > > @@ -92,8 +92,7 @@ static unsigned long lowmem_scan(struct shrinker *s, struct shrink_control *sc)
>> > > >         int array_size = ARRAY_SIZE(lowmem_adj);
>> > > >         int other_free = global_page_state(NR_FREE_PAGES) - totalreserve_pages;
>> > > >         int other_file = global_node_page_state(NR_FILE_PAGES) -
>> > > > -                                               global_node_page_state(NR_SHMEM) -
>> > > > -                                               total_swapcache_pages();
>> > > > +               global_node_page_state(NR_SHMEM) - total_swapcache_pages();
>> > > The original patch here is nicer to read, don't you think?
>> > Strictly, these should probably be unsigned long instead of int
>> > as that's the return type of global_[node_]page_state().
>> >
>> > Maybe something like the below, but then maybe other_free/file
>> > should become unsigned long too.
> []
>> > diff --git a/drivers/staging/android/lowmemorykiller.c b/drivers/staging/android/lowmemorykiller.c
> []
>> > @@ -90,10 +90,11 @@ static unsigned long lowmem_scan(struct shrinker *s, struct shrink_control *sc)
>> >     int selected_tasksize = 0;
>> >     short selected_oom_score_adj;
>> >     int array_size = ARRAY_SIZE(lowmem_adj);
>> > -   int other_free = global_page_state(NR_FREE_PAGES) - totalreserve_pages;
>> > -   int other_file = global_node_page_state(NR_FILE_PAGES) -
>> > -                                           global_node_page_state(NR_SHMEM) -
>> > -                                           total_swapcache_pages();
>> > +   unsigned long nr_free = global_page_state(NR_FREE_PAGES);
>> > +   unsigned long nr_file = global_node_page_state(NR_FILE_PAGES);
>> > +   unsigned long nr_shmem = global_node_page_state(NR_SHMEM);
>> > +   int other_free = nr_free - totalreserve_pages;
>> > +   int other_file = nr_file - nr_shmem - total_swapcache_pages();
>> >
>> >     if (lowmem_adj_size < array_size)
>> >             array_size = lowmem_adj_size;
>> I like it, much easier to follow.  Care to resend it in a patch form I
>> can apply it in?
>
> I think the android folk should figure out what the right thing
> to do with the int/unsigned long calculation is first.
>
> Arve?  Riley?
>

I think the end result is the same if you store it in an int. If you
store the result in a long it will work better, but I don't know of
anyone using lowmemorykilller on a system with more than 8TB of
memory.

-- 
Arve HjÃnnevÃg