mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Hugh Dickins <hugh@veritas.com>
To: Andrew Morton <akpm@osdl.org>
Cc: linux-kernel@vger.kernel.org
Subject: [PATCH 11/21] mm: move_page_tables by extents
Date: Sun, 25 Sep 2005 16:57:10 +0100 (BST)	[thread overview]
Message-ID: <Pine.LNX.4.61.0509251656310.3490@goblin.wat.veritas.com> (raw)
In-Reply-To: <Pine.LNX.4.61.0509251644100.3490@goblin.wat.veritas.com>

Speeding up mremap's moving of ptes has never been a priority, but the
locking will get more complicated shortly, and is already too baroque.

Scrap the current one-by-one moving, do an extent at a time: curtailed
by end of src and dst pmds (have to use PMD_SIZE: the way pmd_addr_end
gets elided doesn't match this usage), and by latency considerations.

One nice property of the old method is lost: it never allocated a page
table unless absolutely necessary, so you could free empty page tables
by mremapping to and fro.  Whereas this way, it allocates a dst table
wherever there was a src table.  I keep diving in to reinstate the old
behaviour, then come out preferring not to clutter how it now is.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
---

 mm/mremap.c |  166 +++++++++++++++++++++++++-----------------------------------
 1 files changed, 71 insertions(+), 95 deletions(-)

--- mm10/mm/mremap.c	2005-09-24 19:27:33.000000000 +0100
+++ mm11/mm/mremap.c	2005-09-24 19:28:42.000000000 +0100
@@ -22,40 +22,15 @@
 #include <asm/cacheflush.h>
 #include <asm/tlbflush.h>
 
-static pte_t *get_one_pte_map_nested(struct mm_struct *mm, unsigned long addr)
-{
-	pgd_t *pgd;
-	pud_t *pud;
-	pmd_t *pmd;
-	pte_t *pte = NULL;
-
-	pgd = pgd_offset(mm, addr);
-	if (pgd_none_or_clear_bad(pgd))
-		goto end;
-
-	pud = pud_offset(pgd, addr);
-	if (pud_none_or_clear_bad(pud))
-		goto end;
-
-	pmd = pmd_offset(pud, addr);
-	if (pmd_none_or_clear_bad(pmd))
-		goto end;
-
-	pte = pte_offset_map_nested(pmd, addr);
-	if (pte_none(*pte)) {
-		pte_unmap_nested(pte);
-		pte = NULL;
-	}
-end:
-	return pte;
-}
-
-static pte_t *get_one_pte_map(struct mm_struct *mm, unsigned long addr)
+static pmd_t *get_old_pmd(struct mm_struct *mm, unsigned long addr)
 {
 	pgd_t *pgd;
 	pud_t *pud;
 	pmd_t *pmd;
 
+	/*
+	 * We don't need page_table_lock: we have mmap_sem exclusively.
+	 */
 	pgd = pgd_offset(mm, addr);
 	if (pgd_none_or_clear_bad(pgd))
 		return NULL;
@@ -68,35 +43,48 @@ static pte_t *get_one_pte_map(struct mm_
 	if (pmd_none_or_clear_bad(pmd))
 		return NULL;
 
-	return pte_offset_map(pmd, addr);
+	return pmd;
 }
 
-static inline pte_t *alloc_one_pte_map(struct mm_struct *mm, unsigned long addr)
+static pmd_t *alloc_new_pmd(struct mm_struct *mm, unsigned long addr)
 {
 	pgd_t *pgd;
 	pud_t *pud;
-	pmd_t *pmd;
-	pte_t *pte = NULL;
+	pmd_t *pmd = NULL;
+	pte_t *pte;
 
+	/*
+	 * We do need page_table_lock: because allocators expect that.
+	 */
+	spin_lock(&mm->page_table_lock);
 	pgd = pgd_offset(mm, addr);
-
 	pud = pud_alloc(mm, pgd, addr);
 	if (!pud)
-		return NULL;
+		goto out;
+
 	pmd = pmd_alloc(mm, pud, addr);
-	if (pmd)
-		pte = pte_alloc_map(mm, pmd, addr);
-	return pte;
+	if (!pmd)
+		goto out;
+
+	pte = pte_alloc_map(mm, pmd, addr);
+	if (!pte) {
+		pmd = NULL;
+		goto out;
+	}
+	pte_unmap(pte);
+out:
+	spin_unlock(&mm->page_table_lock);
+	return pmd;
 }
 
-static int
-move_one_page(struct vm_area_struct *vma, unsigned long old_addr,
-		struct vm_area_struct *new_vma, unsigned long new_addr)
+static void move_ptes(struct vm_area_struct *vma, pmd_t *old_pmd,
+		unsigned long old_addr, unsigned long old_end,
+		struct vm_area_struct *new_vma, pmd_t *new_pmd,
+		unsigned long new_addr)
 {
 	struct address_space *mapping = NULL;
 	struct mm_struct *mm = vma->vm_mm;
-	int error = 0;
-	pte_t *src, *dst;
+	pte_t *old_pte, *new_pte, pte;
 
 	if (vma->vm_file) {
 		/*
@@ -111,74 +99,62 @@ move_one_page(struct vm_area_struct *vma
 		    new_vma->vm_truncate_count != vma->vm_truncate_count)
 			new_vma->vm_truncate_count = 0;
 	}
+
 	spin_lock(&mm->page_table_lock);
+	old_pte = pte_offset_map(old_pmd, old_addr);
+	new_pte = pte_offset_map_nested(new_pmd, new_addr);
 
-	src = get_one_pte_map_nested(mm, old_addr);
-	if (src) {
-		/*
-		 * Look to see whether alloc_one_pte_map needs to perform a
-		 * memory allocation.  If it does then we need to drop the
-		 * atomic kmap
-		 */
-		dst = get_one_pte_map(mm, new_addr);
-		if (unlikely(!dst)) {
-			pte_unmap_nested(src);
-			if (mapping)
-				spin_unlock(&mapping->i_mmap_lock);
-			dst = alloc_one_pte_map(mm, new_addr);
-			if (mapping && !spin_trylock(&mapping->i_mmap_lock)) {
-				spin_unlock(&mm->page_table_lock);
-				spin_lock(&mapping->i_mmap_lock);
-				spin_lock(&mm->page_table_lock);
-			}
-			src = get_one_pte_map_nested(mm, old_addr);
-		}
-		/*
-		 * Since alloc_one_pte_map can drop and re-acquire
-		 * page_table_lock, we should re-check the src entry...
-		 */
-		if (src) {
-			if (dst) {
-				pte_t pte;
-				pte = ptep_clear_flush(vma, old_addr, src);
-
-				/* ZERO_PAGE can be dependant on virtual addr */
-				pte = move_pte(pte, new_vma->vm_page_prot,
-							old_addr, new_addr);
-				set_pte_at(mm, new_addr, dst, pte);
-			} else
-				error = -ENOMEM;
-			pte_unmap_nested(src);
-		}
-		if (dst)
-			pte_unmap(dst);
+	for (; old_addr < old_end; old_pte++, old_addr += PAGE_SIZE,
+				   new_pte++, new_addr += PAGE_SIZE) {
+		if (pte_none(*old_pte))
+			continue;
+		pte = ptep_clear_flush(vma, old_addr, old_pte);
+		/* ZERO_PAGE can be dependant on virtual addr */
+		pte = move_pte(pte, new_vma->vm_page_prot, old_addr, new_addr);
+		set_pte_at(mm, new_addr, new_pte, pte);
 	}
+
+	pte_unmap_nested(new_pte - 1);
+	pte_unmap(old_pte - 1);
 	spin_unlock(&mm->page_table_lock);
 	if (mapping)
 		spin_unlock(&mapping->i_mmap_lock);
-	return error;
 }
 
+#define LATENCY_LIMIT	(64 * PAGE_SIZE)
+
 static unsigned long move_page_tables(struct vm_area_struct *vma,
 		unsigned long old_addr, struct vm_area_struct *new_vma,
 		unsigned long new_addr, unsigned long len)
 {
-	unsigned long offset;
+	unsigned long extent, next, old_end;
+	pmd_t *old_pmd, *new_pmd;
 
-	flush_cache_range(vma, old_addr, old_addr + len);
+	old_end = old_addr + len;
+	flush_cache_range(vma, old_addr, old_end);
 
-	/*
-	 * This is not the clever way to do this, but we're taking the
-	 * easy way out on the assumption that most remappings will be
-	 * only a few pages.. This also makes error recovery easier.
-	 */
-	for (offset = 0; offset < len; offset += PAGE_SIZE) {
-		if (move_one_page(vma, old_addr + offset,
-				new_vma, new_addr + offset) < 0)
-			break;
+	for (; old_addr < old_end; old_addr += extent, new_addr += extent) {
 		cond_resched();
+		next = (old_addr + PMD_SIZE) & PMD_MASK;
+		if (next - 1 > old_end)
+			next = old_end;
+		extent = next - old_addr;
+		old_pmd = get_old_pmd(vma->vm_mm, old_addr);
+		if (!old_pmd)
+			continue;
+		new_pmd = alloc_new_pmd(vma->vm_mm, new_addr);
+		if (!new_pmd)
+			break;
+		next = (new_addr + PMD_SIZE) & PMD_MASK;
+		if (extent > next - new_addr)
+			extent = next - new_addr;
+		if (extent > LATENCY_LIMIT)
+			extent = LATENCY_LIMIT;
+		move_ptes(vma, old_pmd, old_addr, old_addr + extent,
+				new_vma, new_pmd, new_addr);
 	}
-	return offset;
+
+	return len + old_addr - old_end;	/* how much done */
 }
 
 static unsigned long move_vma(struct vm_area_struct *vma,

  parent reply	other threads:[~2005-09-25 15:57 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-09-25 15:46 [PATCH 00/21] mm: page fault scalability prep Hugh Dickins
2005-09-25 15:47 ` [PATCH 01/21] mm: hugetlb truncation fixes Hugh Dickins
2005-09-25 15:48 ` [PATCH 02/21] mm: copy_pte_range progress fix Hugh Dickins
2005-09-25 15:49 ` [PATCH 03/21] mm: msync_pte_range progress Hugh Dickins
2005-09-25 15:49 ` [PATCH 04/21] mm: zap_pte_range dont dirty anon Hugh Dickins
2005-09-25 22:26   ` Andrew Morton
2005-09-26  6:02     ` Hugh Dickins
2005-09-26  6:14       ` Andrew Morton
2005-09-26  7:20         ` Hugh Dickins
2005-09-25 15:51 ` [PATCH 05/21] mm: anon is already wrprotected Hugh Dickins
2005-09-25 15:52 ` [PATCH 06/21] mm: vm_stat_account unshackled Hugh Dickins
2005-09-25 15:53 ` [PATCH 07/21] mm: remove_vma_list consolidation Hugh Dickins
2005-09-25 15:53 ` [PATCH 08/21] mm: unlink_file_vma, remove_vma Hugh Dickins
2005-09-25 15:54 ` [PATCH 09/21] mm: exit_mmap need not reset Hugh Dickins
2005-09-25 15:56 ` [PATCH 10/21] mm: page fault handlers tidyup Hugh Dickins
2005-09-25 15:57 ` Hugh Dickins [this message]
2005-09-25 15:59 ` [PATCH 12/21] mm: tlb_gather_mmu get_cpu_var Hugh Dickins
2005-09-25 16:01 ` [PATCH 13/21] mm: tlb_is_full_mm was obscure Hugh Dickins
2005-09-25 16:03 ` [PATCH 14/21] mm: tlb_finish_mmu forget rss Hugh Dickins
2005-09-25 16:06 ` [PATCH 15/21] mm: mm_init set_mm_counters Hugh Dickins
2005-09-25 16:07 ` [PATCH 16/21] mm: rss = file_rss + anon_rss Hugh Dickins
2005-09-25 16:08 ` [PATCH 17/21] mm: batch updating mm_counters Hugh Dickins
2005-09-26  7:25   ` Nick Piggin
2005-09-26  8:42     ` Hugh Dickins
2005-09-25 16:09 ` [PATCH 18/21] mm: dup_mmap use oldmm more Hugh Dickins
2005-09-25 16:10 ` [PATCH 19/21] mm: dup_mmap down new mmap_sem Hugh Dickins
2005-09-25 16:11 ` [PATCH 20/21] mm: sh64 hugetlbpage.c Hugh Dickins
2005-09-29  7:00   ` Paul Mundt
2005-09-25 16:15 ` [PATCH 21/21] mm: m68k kill stram swap Hugh Dickins
2005-09-28  0:05 ` [PATCH 00/21] mm: page fault scalability prep Christoph Lameter

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=Pine.LNX.4.61.0509251656310.3490@goblin.wat.veritas.com \
    --to=hugh@veritas.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

all inboxes | Powered by JetHome®