lkml.org 
[lkml]   [2012]   [Jan]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 1/4] pagemap: avoid splitting thp when reading /proc/pid/pagemap
On Wed, Jan 04, 2012 at 03:50:42PM -0800, Andrew Morton wrote:
> On Wed, 21 Dec 2011 17:23:45 -0500
> Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> wrote:
>
> > Thp split is not necessary if we explicitly check whether pmds are
> > mapping thps or not. This patch introduces the check and the code
> > to generate pagemap entries for pmds mapping thps, which results in
> > less performance impact of pagemap on thp.
> >
> >
> > ...
>
> The type choices seem odd:
>
> > +#ifdef CONFIG_TRANSPARENT_HUGEPAGE
> > +static u64 thp_pte_to_pagemap_entry(pte_t pte, int offset)
> > +{
> > + u64 pme = 0;
>
> Why are these u64?

I guess (I just copied this type choice from other *pte_to_pagemap_entry()
type functions) it's because each entry in /proc/pid/pagemap is in fixed
sized (64 bit) format as described in the comment above pagemap_read().

> Should we have a pme_t, matching pte_t, pmd_t, etc?

Yes, it makes code's meaning clearer.

>
> > + if (pte_present(pte))
> > + pme = PM_PFRAME(pte_pfn(pte) + offset)
> > + | PM_PSHIFT(PAGE_SHIFT) | PM_PRESENT;
> > + return pme;
> > +}
> > +#else
> > +static inline u64 thp_pte_to_pagemap_entry(pte_t pte, int offset)
> > +{
> > + return 0;
> > +}
> > +#endif
> > +
> > static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
> > struct mm_walk *walk)
> > {
> > @@ -665,14 +684,34 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
> > struct pagemapread *pm = walk->private;
> > pte_t *pte;
> > int err = 0;
> > -
> > - split_huge_page_pmd(walk->mm, pmd);
> > + u64 pfn = PM_NOT_PRESENT;
>
> Again, why a u64? pfn's are usually unsigned long.

I think variable's name 'pfn' is wrong rather than type choice
because this variable stores pagemap entry which is not a pure pfn.
There's room for improvement, so I'll try it in the next turn.

Thanks,
Naoya


\
 
 \ /
  Last update: 2012-01-05 17:31    [W:0.145 / U:0.116 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site