lkml.org 
[lkml]   [2008]   [Jan]   [21]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH -v7 2/2] Update ctime and mtime for memory-mapped files
2008/1/22, Linus Torvalds <torvalds@linux-foundation.org>:
>
>
> On Tue, 22 Jan 2008, Anton Salikhmetov wrote:
> >
> > /*
> > + * Scan the PTEs for pages belonging to the VMA and mark them read-only.
> > + * It will force a pagefault on the next write access.
> > + */
> > +static void vma_wrprotect(struct vm_area_struct *vma)
> > +{
> > + unsigned long addr;
> > +
> > + for (addr = vma->vm_start; addr < vma->vm_end; addr += PAGE_SIZE) {
> > + spinlock_t *ptl;
> > + pgd_t *pgd = pgd_offset(vma->vm_mm, addr);
> > + pud_t *pud = pud_offset(pgd, addr);
> > + pmd_t *pmd = pmd_offset(pud, addr);
> > + pte_t *pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
>
> This is extremely expensive over bigger areas, especially sparsely mapped
> ones (it does all the lookups for all four levels over and over and over
> again for eachg page).
>
> I think Peter Zijlstra posted a version that uses the regular kind of
> nested loop (with inline functions to keep the thing nice and clean),
> which gets rid of that.

Thanks for your feedback, Linus!

I will use Peter Zijlstra's version of such an operation in my next
patch series.

>
> [ The sad/funny part is that this is all how we *used* to do msync(), back
> in the days: we're literally going back to the "pre-cleanup" logic. See
> commit 204ec841fbea3e5138168edbc3a76d46747cc987: "mm: msync() cleanup"
> for details ]
>
> Quite frankly, I really think you might be better off just doing a
>
> git revert 204ec841fbea3e5138168edbc3a76d46747cc987
>
> and working from there! I just checked, and it still reverts cleanly, and
> you'd end up with a nice code-base that (a) has gotten years of testing
> and (b) already has the looping-over-the-pagetables code.
>
> Linus
>


\
 
 \ /
  Last update: 2008-01-22 03:41    [W:0.355 / U:0.084 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site