Messages in this thread | | | From | "Kirill A. Shutemov" <> | Subject | Re: [PATCH 8/9] mm: implement split page table lock for PMD level | Date | Fri, 13 Sep 2013 17:25:43 +0300 (EEST) |
| |
Peter Zijlstra wrote: > On Fri, Sep 13, 2013 at 04:06:15PM +0300, Kirill A. Shutemov wrote: > > +#if USE_SPLIT_PMD_PTLOCKS > > + > > +static inline void pgtable_pmd_page_ctor(struct page *page) > > +{ > > + spin_lock_init(&page->ptl); > > +#ifdef CONFIG_TRANSPARENT_HUGEPAGE > > + page->pmd_huge_pte = NULL; > > +#endif > > +} > > + > > +static inline void pgtable_pmd_page_dtor(struct page *page) > > +{ > > +#ifdef CONFIG_TRANSPARENT_HUGEPAGE > > + VM_BUG_ON(page->pmd_huge_pte); > > +#endif > > +} > > + > > +#define pmd_huge_pte(mm, pmd) (virt_to_page(pmd)->pmd_huge_pte) > > + > > +#else > > So on -rt we have the problem that spinlock_t is rather huge (its a > rtmutex) so instead of blowing up the pageframe like that we treat > page->pte as a pointer and allocate the spinlock. > > Since allocations could fail the above ctor path gets 'interesting'. > > It would be good if new code could assume the ctor could fail so we > don't have to replicate that horror-show.
Okay, I'll rework this.
-- Kirill A. Shutemov
| |