public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: William Lee Irwin III <wli@holomorphy.com>
To: Andrew Morton <akpm@osdl.org>, linux-kernel@vger.kernel.org
Subject: Re: [RFC] invalidate_mmap_range() misses remap_file_pages()-affected targets
Date: Sun, 12 Oct 2003 17:59:18 -0700	[thread overview]
Message-ID: <20031013005918.GD765@holomorphy.com> (raw)
In-Reply-To: <20031012195146.GB16158@holomorphy.com>

On Sun, Oct 12, 2003 at 12:51:46PM -0700, William Lee Irwin III wrote:
> +	/* PTE_FILE ptes have the same file, but pgoff can differ */
> +	if (pte_file(*pte))
> +		present = mincore_linear_page(vma, pte_to_pgoff(*pte));
> +	/* pte presence overrides the calculated offset */
> +	else if (pte_present(*pte))
> +		present = 1;
> +	/* matching offsets are the faulted in if the pte isn't set */
> +	else
> +		present = mincore_linear_page(vma, pgoff);

This is broken; _PAGE_FILE can alias other flag bits used when the page
is present, i.e. presence must always be checked first. Amended patch
below.

I noticed another disturbing oddity after the anonymous vma issue:

       The mincore function requests a vector describing which pages of a file
       are in core and can be read without disk access. The kernel will supply

Strictly interpreted, this suggests PROT_NONE should be reported as not
present by mincore() regardless of presence in the pagecache.


-- wli


diff -prauN rfp-2.6.0-test7-bk3-1/mm/mincore.c rfp-2.6.0-test7-bk3-2/mm/mincore.c
--- rfp-2.6.0-test7-bk3-1/mm/mincore.c	2003-10-08 12:24:51.000000000 -0700
+++ rfp-2.6.0-test7-bk3-2/mm/mincore.c	2003-10-12 17:43:50.000000000 -0700
@@ -22,7 +22,7 @@
  * and is up to date; i.e. that no page-in operation would be required
  * at this time if an application were to map and access this page.
  */
-static unsigned char mincore_page(struct vm_area_struct * vma,
+static unsigned char mincore_linear_page(struct vm_area_struct *vma,
 	unsigned long pgoff)
 {
 	unsigned char present = 0;
@@ -38,6 +38,68 @@ static unsigned char mincore_page(struct
 	return present;
 }
 
+static unsigned char mincore_nonlinear_page(struct vm_area_struct *vma,
+						unsigned long pgoff)
+{
+	unsigned char present = 0;
+	unsigned long vaddr;
+	pgd_t *pgd;
+	pmd_t *pmd;
+	pte_t *pte;
+
+	spin_lock(&vma->vm_mm->page_table_lock);
+	vaddr = PAGE_SIZE*(pgoff - vma->vm_pgoff) + vma->vm_start;
+	pgd = pgd_offset(vma->vm_mm, vaddr);
+	if (pgd_none(*pgd))
+		goto out;
+	else if (pgd_bad(*pgd)) {
+		pgd_ERROR(*pgd);
+		pgd_clear(pgd);
+		goto out;
+	}
+	pmd = pmd_offset(pgd, vaddr);
+	if (pmd_none(*pmd))
+		goto out;
+	else if (pmd_ERROR(*pmd)) {
+		pmd_ERROR(*pmd);
+		pmd_clear(pmd);
+		goto out;
+	}
+
+	pte = pte_offset_map(pmd, vaddr);
+
+	/* pte presence overrides the calculated offset */
+	if (pte_present(*pte))
+		present = 1;
+
+	/* PTE_FILE ptes have the same file, but pgoff can differ */
+	else if (pte_file(*pte))
+		present = mincore_linear_page(vma, pte_to_pgoff(*pte));
+
+	/* matching offsets are the faulted in if the pte isn't set */
+	else
+		present = mincore_linear_page(vma, pgoff);
+
+	pte_unmap(pte);
+out:
+	spin_unlock(&vma->vm_mm->page_table_lock);
+	return present;
+}
+
+static inline unsigned char mincore_page(struct vm_area_struct *vma,
+						unsigned long pgoff)
+{
+	unsigned char ret;
+	struct address_space *as = vma->vm_file->f_dentry->d_inode->i_mapping;
+	down(&as->i_shared_sem);
+	if (vma->vm_flags & VM_NONLINEAR)
+		ret = mincore_nonlinear_page(vma, pgoff);
+	else
+		ret = mincore_linear_page(vma, pgoff);
+	up(&as->i_shared_sem);
+	return ret;
+}
+
 static long mincore_vma(struct vm_area_struct * vma,
 	unsigned long start, unsigned long end, unsigned char __user * vec)
 {

  reply	other threads:[~2003-10-13  0:56 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-10-12  8:48 [RFC] invalidate_mmap_range() misses remap_file_pages()-affected targets William Lee Irwin III
2003-10-12 10:34 ` William Lee Irwin III
2003-10-12 11:56   ` Andrew Morton
2003-10-12 19:51     ` William Lee Irwin III
2003-10-13  0:59       ` William Lee Irwin III [this message]
2003-10-12 11:53 ` Andrew Morton
2003-10-12 19:38   ` William Lee Irwin III
2003-10-12 20:28 ` Rik van Riel
2003-10-12 21:19   ` William Lee Irwin III

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20031013005918.GD765@holomorphy.com \
    --to=wli@holomorphy.com \
    --cc=akpm@osdl.org \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox