This is wrong because both routines can be called from within do_file_page,
which is called when !pte_present(pte) && !pte_none(pte) && pte_file(pte). I.e.
the pte is not zeroed, so it has been used, but the page has been swapped out,
or .

Actually, in that situation ->populate is called with nonblock == 0, so only
install_page can be called there. If ->populate fails, the faulting process
will get a SIGBUS; and this *could* happen.

The other caller is sys_remap_file_page. In that situation, we fail if
!(vma->vm_flags & VM_SHARED), so we don't call ->populate.

Actually, however, we could be called inside mmap for MAP_POPULATE. However,
even in that case, old mappings have already been unmapped away.

And actually, in Ingo's patch things were different, because if
remap_file_pages was called only to change protections, without creating
non-linear mappings, we didn't punt even if the VM was not shared. So,
currently MAP_POPULATE will fail (though without returning any error) if the VMA
is private and not shared, without a real reason to do so.

But again, after mmap'ing an area, even with MAP_POPULATE, old mappings should
have been cleared, so what are we bothering about here?

diff -puN mm/fremap.c~remap-file-pages-prot-2.6.13-rc3-A2 mm/fremap.c
--- linux-2.6.git/mm/fremap.c~remap-file-pages-prot-2.6.13-rc3-A2	2005-08-03 12:41:36.000000000 +0200
+++ linux-2.6.git-paolo/mm/fremap.c	2005-08-03 12:41:36.000000000 +0200
@@ -90,6 +90,14 @@ int install_page(struct mm_struct *mm, s
 	if (!page->mapping || page->index >= size)
 		goto err_unlock;
 
+	/*
+	 * Only install a new page for a non-shared mapping if it's
+	 * not existent yet:
+	 */
+	err = -EEXIST;
+	if (!pte_none(*pte) && !(vma->vm_flags & VM_SHARED))
+		goto err_unlock;
+
 	zap_pte(mm, vma, addr, pte);
 
 	inc_mm_counter(mm,rss);
@@ -136,6 +146,13 @@ int install_file_pte(struct mm_struct *m
 	err = 0;
 	if (linear && pte_none(*pte))
 		goto err_unlock;
+	/*
+	 * Only install a new page for a non-shared mapping if it's
+	 * not existent yet:
+	 */
+	err = -EEXIST;
+	if (!pte_none(*pte) && !(vma->vm_flags & VM_SHARED))
+		goto err_unlock;
 
 	zap_pte(mm, vma, addr, pte);
 


# vim: tw=80