2008/1/19, Linus Torvalds <[EMAIL PROTECTED]>:
>
>
> On Sat, 19 Jan 2008, Anton Salikhmetov wrote:
> >
> > Before using pte_wrprotect() the vma_wrprotect() routine uses the
> > pte_offset_map_lock() macro to get the PTE and to acquire the ptl
> > spinlock. Why did you say that this code was not SMP-safe? It should
> > be atomic, I think.
>
> It's atomic WITH RESPECT TO OTHER PEOPLE WHO GET THE LOCK.
>
> Guess how much another x86 CPU cares when it sets the accessed bit in
> hardware?
Thank you very much for taking part in this discussion. Personally,
it's very important to me. But I'm not sure that I understand which
bit can be lost.
Please let me explain.
The logic for my vma_wrprotect() routine was taken from the
page_check_address() function in mm/rmap.c. Here is a code snippet of
the latter function:
pgd = pgd_offset(mm, address);
if (!pgd_present(*pgd))
return NULL;
pud = pud_offset(pgd, address);
if (!pud_present(*pud))
return NULL;
pmd = pmd_offset(pud, address);
if (!pmd_present(*pmd))
return NULL;
pte = pte_offset_map(pmd, address);
/* Make a quick check before getting the lock */
if (!pte_present(*pte)) {
pte_unmap(pte);
return NULL;
}
ptl = pte_lockptr(mm, pmd);
spin_lock(ptl);
if (pte_present(*pte) && page_to_pfn(page) == pte_pfn(*pte)) {
*ptlp = ptl;
return pte;
}
pte_unmap_unlock(pte, ptl);
The page_check_address() function is called from the
page_mkclean_one() routine as follows:
pte = page_check_address(page, mm, address, &ptl);
if (!pte)
goto out;
if (pte_dirty(*pte) || pte_write(*pte)) {
pte_t entry;
flush_cache_page(vma, address, pte_pfn(*pte));
entry = ptep_clear_flush(vma, address, pte);
entry = pte_wrprotect(entry);
entry = pte_mkclean(entry);
set_pte_at(mm, address, pte, entry);
ret = 1;
}
pte_unmap_unlock(pte, ptl);
The write-protection of the PTE is done using the pte_wrprotect()
entity. I intended to do the same during msync() with MS_ASYNC. I
understand that I'm taking a risk of looking a complete idiot now,
however I don't see any difference between the two situations.
I presumed that the code in mm/rmap.c was absolutely correct, that's
why I basically reused the design.
>
> > The POSIX standard requires the ctime and mtime stamps to be updated
> > not later than at the second call to msync() with the MS_ASYNC flag.
>
> .. and that is no excuse for bad code.
>
> Linus
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
