Re: [PATCH] Revert "mm/page-writeback.c: print a warning if the vm dirtiness settings are illogical" (was: Re: [PATCH] mm: print a warning once the vm dirtiness settings is illogical)

[Jan Kara <jack@suse.cz>]

On Tue 28-11-17 18:33:02, Yafang Shao wrote:
> 2017-11-28 18:25 GMT+08:00 Jan Kara <jack@suse.cz>:
> > Hi Yafang,
> >
> > On Tue 28-11-17 11:11:40, Yafang Shao wrote:
> >> What about bellow change ?
> >> It makes the function  domain_dirty_limits() more clear.
> >> And the result will have a higher precision.
> >
> > Frankly, I don't find this any better and you've just lost the additional
> > precision of ratios computed in the "if (gdtc)" branch the multiplication by
> > PAGE_SIZE got us.
> >
> 
> What about bellow change? It won't be lost any more, becasue
> bytes and bg_bytes are both PAGE_SIZE aligned.
> 
> -       if (bytes)
> -           ratio = min(DIV_ROUND_UP(bytes, global_avail),
> -                   PAGE_SIZE);
> -       if (bg_bytes)
> -           bg_ratio = min(DIV_ROUND_UP(bg_bytes, global_avail),
> -                      PAGE_SIZE);
> +       if (bytes) {
> +           pages = DIV_ROUND_UP(bytes, PAGE_SIZE);
> +           ratio = DIV_ROUND_UP(pages * 100, global_avail);
> +
> +       }
> +
> +       if (bg_bytes) {
> +           pages = DIV_ROUND_UP(bg_bytes, PAGE_SIZE);
> +           bg_ratio = DIV_ROUND_UP(pages * 100, global_avail);
> +       }

Not better... Look, in the original code the 'ratio' and 'bg_ratio'
variables contain a number between 0 and 1 as fractions of 1/PAGE_SIZE. In
your code you have in these variables fractions of 1/100. That's certainly
less precise no matter how you get to those numbers.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>