Merge tag 'md-3.5' of git://neil.brown.name/md

Pull md updates from NeilBrown:
 "It's been a busy cycle for md - lots of fun stuff here..  if you like
  this kind of thing :-)

  Main features:
   - RAID10 arrays can be reshaped - adding and removing devices and
     changing chunks (not 'far' array though)
   - allow RAID5 arrays to be reshaped with a backup file (not tested
     yet, but the priciple works fine for RAID10).
   - arrays can be reshaped while a bitmap is present - you no longer
     need to remove it first
   - SSSE3 support for RAID6 syndrome calculations

  and of course a number of minor fixes etc."

* tag 'md-3.5' of git://neil.brown.name/md: (56 commits)
  md/bitmap: record the space available for the bitmap in the superblock.
  md/raid10: Remove extras after reshape to smaller number of devices.
  md/raid5: improve removal of extra devices after reshape.
  md: check the return of mddev_find()
  MD RAID1: Further conditionalize 'fullsync'
  DM RAID: Use md_error() in place of simply setting Faulty bit
  DM RAID: Record and handle missing devices
  DM RAID: Set recovery flags on resume
  md/raid5: Allow reshape while a bitmap is present.
  md/raid10: resize bitmap when required during reshape.
  md: allow array to be resized while bitmap is present.
  md/bitmap: make sure reshape request are reflected in superblock.
  md/bitmap: add bitmap_resize function to allow bitmap resizing.
  md/bitmap: use DIV_ROUND_UP instead of open-code
  md/bitmap: create a 'struct bitmap_counts' substructure of 'struct bitmap'
  md/bitmap: make bitmap bitops atomic.
  md/bitmap: make _page_attr bitops atomic.
  md/bitmap: merge bitmap_file_unmap and bitmap_file_put.
  md/bitmap: remove async freeing of bitmap file.
  md/bitmap: convert some spin_lock_irqsave to spin_lock_irq
  ...

25 files changed, 3124 insertions, 868 deletions

diff --git a/arch/x86/Makefile b/arch/x86/Makefile
index dc611a40a336..1f2521434554 100644
--- a/arch/x86/Makefile
+++ b/arch/x86/Makefile
@@ -115,9 +115,10 @@ cfi-sections := $(call as-instr,.cfi_sections .debug_frame,-DCONFIG_AS_CFI_SECTI
 
 # does binutils support specific instructions?
 asinstr := $(call as-instr,fxsaveq (%rax),-DCONFIG_AS_FXSAVEQ=1)
+avx_instr := $(call as-instr,vxorps %ymm0$(comma)%ymm1$(comma)%ymm2,-DCONFIG_AS_AVX=1)
 
-KBUILD_AFLAGS += $(cfi) $(cfi-sigframe) $(cfi-sections) $(asinstr)
-KBUILD_CFLAGS += $(cfi) $(cfi-sigframe) $(cfi-sections) $(asinstr)
+KBUILD_AFLAGS += $(cfi) $(cfi-sigframe) $(cfi-sections) $(asinstr) $(avx_instr)
+KBUILD_CFLAGS += $(cfi) $(cfi-sigframe) $(cfi-sections) $(asinstr) $(avx_instr)
 
 LDFLAGS := -m elf_$(UTS_MACHINE)
 
diff --git a/arch/x86/include/asm/xor_32.h b/arch/x86/include/asm/xor_32.h
index 133b40a0f495..454570891bdc 100644
--- a/arch/x86/include/asm/xor_32.h
+++ b/arch/x86/include/asm/xor_32.h
@@ -861,6 +861,9 @@ static struct xor_block_template xor_block_pIII_sse = {
 	.do_5 = xor_sse_5,
 };
 
+/* Also try the AVX routines */
+#include "xor_avx.h"
+
 /* Also try the generic routines.  */
 #include <asm-generic/xor.h>
 
@@ -871,6 +874,7 @@ do {							\
 	xor_speed(&xor_block_8regs_p);			\
 	xor_speed(&xor_block_32regs);			\
 	xor_speed(&xor_block_32regs_p);			\
+	AVX_XOR_SPEED;					\
 	if (cpu_has_xmm)				\
 		xor_speed(&xor_block_pIII_sse);		\
 	if (cpu_has_mmx) {				\
@@ -883,6 +887,6 @@ do {							\
    We may also be able to load into the L1 only depending on how the cpu
    deals with a load to a line that is being prefetched.  */
 #define XOR_SELECT_TEMPLATE(FASTEST)			\
-	(cpu_has_xmm ? &xor_block_pIII_sse : FASTEST)
+	AVX_SELECT(cpu_has_xmm ? &xor_block_pIII_sse : FASTEST)
 
 #endif /* _ASM_X86_XOR_32_H */
diff --git a/arch/x86/include/asm/xor_64.h b/arch/x86/include/asm/xor_64.h
index 1549b5e261f6..b9b2323e90fe 100644
--- a/arch/x86/include/asm/xor_64.h
+++ b/arch/x86/include/asm/xor_64.h
@@ -347,15 +347,21 @@ static struct xor_block_template xor_block_sse = {
 	.do_5 = xor_sse_5,
 };
 
+
+/* Also try the AVX routines */
+#include "xor_avx.h"
+
 #undef XOR_TRY_TEMPLATES
 #define XOR_TRY_TEMPLATES			\
 do {						\
+	AVX_XOR_SPEED;				\
 	xor_speed(&xor_block_sse);		\
 } while (0)
 
 /* We force the use of the SSE xor block because it can write around L2.
    We may also be able to load into the L1 only depending on how the cpu
    deals with a load to a line that is being prefetched.  */
-#define XOR_SELECT_TEMPLATE(FASTEST) (&xor_block_sse)
+#define XOR_SELECT_TEMPLATE(FASTEST) \
+	AVX_SELECT(&xor_block_sse)
 
 #endif /* _ASM_X86_XOR_64_H */
diff --git a/arch/x86/include/asm/xor_avx.h b/arch/x86/include/asm/xor_avx.h
new file mode 100644
index 000000000000..2510d35f480e
--- /dev/null
+++ b/arch/x86/include/asm/xor_avx.h
@@ -0,0 +1,214 @@
+#ifndef _ASM_X86_XOR_AVX_H
+#define _ASM_X86_XOR_AVX_H
+
+/*
+ * Optimized RAID-5 checksumming functions for AVX
+ *
+ * Copyright (C) 2012 Intel Corporation
+ * Author: Jim Kukunas <james.t.kukunas@linux.intel.com>
+ *
+ * Based on Ingo Molnar and Zach Brown's respective MMX and SSE routines
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+
+#ifdef CONFIG_AS_AVX
+
+#include <linux/compiler.h>
+#include <asm/i387.h>
+
+#define ALIGN32 __aligned(32)
+
+#define YMM_SAVED_REGS 4
+
+#define YMMS_SAVE \
+do { \
+	preempt_disable(); \
+	cr0 = read_cr0(); \
+	clts(); \
+	asm volatile("vmovaps %%ymm0, %0" : "=m" (ymm_save[0]) : : "memory"); \
+	asm volatile("vmovaps %%ymm1, %0" : "=m" (ymm_save[32]) : : "memory"); \
+	asm volatile("vmovaps %%ymm2, %0" : "=m" (ymm_save[64]) : : "memory"); \
+	asm volatile("vmovaps %%ymm3, %0" : "=m" (ymm_save[96]) : : "memory"); \
+} while (0);
+
+#define YMMS_RESTORE \
+do { \
+	asm volatile("sfence" : : : "memory"); \
+	asm volatile("vmovaps %0, %%ymm3" : : "m" (ymm_save[96])); \
+	asm volatile("vmovaps %0, %%ymm2" : : "m" (ymm_save[64])); \
+	asm volatile("vmovaps %0, %%ymm1" : : "m" (ymm_save[32])); \
+	asm volatile("vmovaps %0, %%ymm0" : : "m" (ymm_save[0])); \
+	write_cr0(cr0); \
+	preempt_enable(); \
+} while (0);
+
+#define BLOCK4(i) \
+		BLOCK(32 * i, 0) \
+		BLOCK(32 * (i + 1), 1) \
+		BLOCK(32 * (i + 2), 2) \
+		BLOCK(32 * (i + 3), 3)
+
+#define BLOCK16() \
+		BLOCK4(0) \
+		BLOCK4(4) \
+		BLOCK4(8) \
+		BLOCK4(12)
+
+static void xor_avx_2(unsigned long bytes, unsigned long *p0, unsigned long *p1)
+{
+	unsigned long cr0, lines = bytes >> 9;
+	char ymm_save[32 * YMM_SAVED_REGS] ALIGN32;
+
+	YMMS_SAVE
+
+	while (lines--) {
+#undef BLOCK
+#define BLOCK(i, reg) \
+do { \
+	asm volatile("vmovdqa %0, %%ymm" #reg : : "m" (p1[i / sizeof(*p1)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm"  #reg : : \
+		"m" (p0[i / sizeof(*p0)])); \
+	asm volatile("vmovdqa %%ymm" #reg ", %0" : \
+		"=m" (p0[i / sizeof(*p0)])); \
+} while (0);
+
+		BLOCK16()
+
+		p0 = (unsigned long *)((uintptr_t)p0 + 512);
+		p1 = (unsigned long *)((uintptr_t)p1 + 512);
+	}
+
+	YMMS_RESTORE
+}
+
+static void xor_avx_3(unsigned long bytes, unsigned long *p0, unsigned long *p1,
+	unsigned long *p2)
+{
+	unsigned long cr0, lines = bytes >> 9;
+	char ymm_save[32 * YMM_SAVED_REGS] ALIGN32;
+
+	YMMS_SAVE
+
+	while (lines--) {
+#undef BLOCK
+#define BLOCK(i, reg) \
+do { \
+	asm volatile("vmovdqa %0, %%ymm" #reg : : "m" (p2[i / sizeof(*p2)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p1[i / sizeof(*p1)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p0[i / sizeof(*p0)])); \
+	asm volatile("vmovdqa %%ymm" #reg ", %0" : \
+		"=m" (p0[i / sizeof(*p0)])); \
+} while (0);
+
+		BLOCK16()
+
+		p0 = (unsigned long *)((uintptr_t)p0 + 512);
+		p1 = (unsigned long *)((uintptr_t)p1 + 512);
+		p2 = (unsigned long *)((uintptr_t)p2 + 512);
+	}
+
+	YMMS_RESTORE
+}
+
+static void xor_avx_4(unsigned long bytes, unsigned long *p0, unsigned long *p1,
+	unsigned long *p2, unsigned long *p3)
+{
+	unsigned long cr0, lines = bytes >> 9;
+	char ymm_save[32 * YMM_SAVED_REGS] ALIGN32;
+
+	YMMS_SAVE
+
+	while (lines--) {
+#undef BLOCK
+#define BLOCK(i, reg) \
+do { \
+	asm volatile("vmovdqa %0, %%ymm" #reg : : "m" (p3[i / sizeof(*p3)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p2[i / sizeof(*p2)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p1[i / sizeof(*p1)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p0[i / sizeof(*p0)])); \
+	asm volatile("vmovdqa %%ymm" #reg ", %0" : \
+		"=m" (p0[i / sizeof(*p0)])); \
+} while (0);
+
+		BLOCK16();
+
+		p0 = (unsigned long *)((uintptr_t)p0 + 512);
+		p1 = (unsigned long *)((uintptr_t)p1 + 512);
+		p2 = (unsigned long *)((uintptr_t)p2 + 512);
+		p3 = (unsigned long *)((uintptr_t)p3 + 512);
+	}
+
+	YMMS_RESTORE
+}
+
+static void xor_avx_5(unsigned long bytes, unsigned long *p0, unsigned long *p1,
+	unsigned long *p2, unsigned long *p3, unsigned long *p4)
+{
+	unsigned long cr0, lines = bytes >> 9;
+	char ymm_save[32 * YMM_SAVED_REGS] ALIGN32;
+
+	YMMS_SAVE
+
+	while (lines--) {
+#undef BLOCK
+#define BLOCK(i, reg) \
+do { \
+	asm volatile("vmovdqa %0, %%ymm" #reg : : "m" (p4[i / sizeof(*p4)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p3[i / sizeof(*p3)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p2[i / sizeof(*p2)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p1[i / sizeof(*p1)])); \
+	asm volatile("vxorps %0, %%ymm" #reg ", %%ymm" #reg : : \
+		"m" (p0[i / sizeof(*p0)])); \
+	asm volatile("vmovdqa %%ymm" #reg ", %0" : \
+		"=m" (p0[i / sizeof(*p0)])); \
+} while (0);
+
+		BLOCK16()
+
+		p0 = (unsigned long *)((uintptr_t)p0 + 512);
+		p1 = (unsigned long *)((uintptr_t)p1 + 512);
+		p2 = (unsigned long *)((uintptr_t)p2 + 512);
+		p3 = (unsigned long *)((uintptr_t)p3 + 512);
+		p4 = (unsigned long *)((uintptr_t)p4 + 512);
+	}
+
+	YMMS_RESTORE
+}
+
+static struct xor_block_template xor_block_avx = {
+	.name = "avx",
+	.do_2 = xor_avx_2,
+	.do_3 = xor_avx_3,
+	.do_4 = xor_avx_4,
+	.do_5 = xor_avx_5,
+};
+
+#define AVX_XOR_SPEED \
+do { \
+	if (cpu_has_avx) \
+		xor_speed(&xor_block_avx); \
+} while (0)
+
+#define AVX_SELECT(FASTEST) \
+	(cpu_has_avx ? &xor_block_avx : FASTEST)
+
+#else
+
+#define AVX_XOR_SPEED {}
+
+#define AVX_SELECT(FASTEST) (FASTEST)
+
+#endif
+#endif
diff --git a/crypto/xor.c b/crypto/xor.c
index 664b6dfa9e2c..65c7b416b4a3 100644
--- a/crypto/xor.c
+++ b/crypto/xor.c
@@ -21,6 +21,7 @@
 #include <linux/gfp.h>
 #include <linux/raid/xor.h>
 #include <linux/jiffies.h>
+#include <linux/preempt.h>
 #include <asm/xor.h>
 
 /* The xor routines to use.  */
@@ -63,12 +64,14 @@ static void
 do_xor_speed(struct xor_block_template *tmpl, void *b1, void *b2)
 {
 	int speed;
-	unsigned long now;
+	unsigned long now, j;
 	int i, count, max;
 
 	tmpl->next = template_list;
 	template_list = tmpl;
 
+	preempt_disable();
+
 	/*
 	 * Count the number of XORs done during a whole jiffy, and use
 	 * this to calculate the speed of checksumming.  We use a 2-page
@@ -76,9 +79,11 @@ do_xor_speed(struct xor_block_template *tmpl, void *b1, void *b2)
 	 */
 	max = 0;
 	for (i = 0; i < 5; i++) {
-		now = jiffies;
+		j = jiffies;
 		count = 0;
-		while (jiffies == now) {
+		while ((now = jiffies) == j)
+			cpu_relax();
+		while (time_before(jiffies, now + 1)) {
 			mb(); /* prevent loop optimzation */
 			tmpl->do_2(BENCH_SIZE, b1, b2);
 			mb();
@@ -89,6 +94,8 @@ do_xor_speed(struct xor_block_template *tmpl, void *b1, void *b2)
 			max = count;
 	}
 
+	preempt_enable();
+
 	speed = max * (HZ * BENCH_SIZE / 1024);
 	tmpl->speed = speed;
 
diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c
index 17e2b472e16d..15dbe03117e4 100644
--- a/drivers/md/bitmap.c
+++ b/drivers/md/bitmap.c
@@ -45,7 +45,7 @@ static inline char *bmname(struct bitmap *bitmap)
  * if we find our page, we increment the page's refcount so that it stays
  * allocated while we're using it
  */
-static int bitmap_checkpage(struct bitmap *bitmap,
+static int bitmap_checkpage(struct bitmap_counts *bitmap,
 			    unsigned long page, int create)
 __releases(bitmap->lock)
 __acquires(bitmap->lock)
@@ -76,8 +76,7 @@ __acquires(bitmap->lock)
 	spin_lock_irq(&bitmap->lock);
 
 	if (mappage == NULL) {
-		pr_debug("%s: bitmap map page allocation failed, hijacking\n",
-			 bmname(bitmap));
+		pr_debug("md/bitmap: map page allocation failed, hijacking\n");
 		/* failed - set the hijacked flag so that we can use the
 		 * pointer as a counter */
 		if (!bitmap->bp[page].map)
@@ -100,7 +99,7 @@ __acquires(bitmap->lock)
 /* if page is completely empty, put it back on the free list, or dealloc it */
 /* if page was hijacked, unmark the flag so it might get alloced next time */
 /* Note: lock should be held when calling this */
-static void bitmap_checkfree(struct bitmap *bitmap, unsigned long page)
+static void bitmap_checkfree(struct bitmap_counts *bitmap, unsigned long page)
 {
 	char *ptr;
 
@@ -130,22 +129,14 @@ static void bitmap_checkfree(struct bitmap *bitmap, unsigned long page)
  */
 
 /* IO operations when bitmap is stored near all superblocks */
-static struct page *read_sb_page(struct mddev *mddev, loff_t offset,
-				 struct page *page,
-				 unsigned long index, int size)
+static int read_sb_page(struct mddev *mddev, loff_t offset,
+			struct page *page,
+			unsigned long index, int size)
 {
 	/* choose a good rdev and read the page from there */
 
 	struct md_rdev *rdev;
 	sector_t target;
-	int did_alloc = 0;
-
-	if (!page) {
-		page = alloc_page(GFP_KERNEL);
-		if (!page)
-			return ERR_PTR(-ENOMEM);
-		did_alloc = 1;
-	}
 
 	rdev_for_each(rdev, mddev) {
 		if (! test_bit(In_sync, &rdev->flags)
@@ -158,15 +149,10 @@ static struct page *read_sb_page(struct mddev *mddev, loff_t offset,
 				 roundup(size, bdev_logical_block_size(rdev->bdev)),
 				 page, READ, true)) {
 			page->index = index;
-			attach_page_buffers(page, NULL); /* so that free_buffer will
-							  * quietly no-op */
-			return page;
+			return 0;
 		}
 	}
-	if (did_alloc)
-		put_page(page);
-	return ERR_PTR(-EIO);
-
+	return -EIO;
 }
 
 static struct md_rdev *next_active_rdev(struct md_rdev *rdev, struct mddev *mddev)
@@ -208,6 +194,7 @@ static int write_sb_page(struct bitmap *bitmap, struct page *page, int wait)
 	struct md_rdev *rdev = NULL;
 	struct block_device *bdev;
 	struct mddev *mddev = bitmap->mddev;
+	struct bitmap_storage *store = &bitmap->storage;
 
 	while ((rdev = next_active_rdev(rdev, mddev)) != NULL) {
 		int size = PAGE_SIZE;
@@ -215,9 +202,13 @@ static int write_sb_page(struct bitmap *bitmap, struct page *page, int wait)
 
 		bdev = (rdev->meta_bdev) ? rdev->meta_bdev : rdev->bdev;
 
-		if (page->index == bitmap->file_pages-1)
-			size = roundup(bitmap->last_page_size,
+		if (page->index == store->file_pages-1) {
+			int last_page_size = store->bytes & (PAGE_SIZE-1);
+			if (last_page_size == 0)
+				last_page_size = PAGE_SIZE;
+			size = roundup(last_page_size,
 				       bdev_logical_block_size(bdev));
+		}
 		/* Just make sure we aren't corrupting data or
 		 * metadata
 		 */
@@ -276,10 +267,10 @@ static void write_page(struct bitmap *bitmap, struct page *page, int wait)
 {
 	struct buffer_head *bh;
 
-	if (bitmap->file == NULL) {
+	if (bitmap->storage.file == NULL) {
 		switch (write_sb_page(bitmap, page, wait)) {
 		case -EINVAL:
-			bitmap->flags |= BITMAP_WRITE_ERROR;
+			set_bit(BITMAP_WRITE_ERROR, &bitmap->flags);
 		}
 	} else {
 
@@ -297,20 +288,16 @@ static void write_page(struct bitmap *bitmap, struct page *page, int wait)
 			wait_event(bitmap->write_wait,
 				   atomic_read(&bitmap->pending_writes)==0);
 	}
-	if (bitmap->flags & BITMAP_WRITE_ERROR)
+	if (test_bit(BITMAP_WRITE_ERROR, &bitmap->flags))
 		bitmap_file_kick(bitmap);
 }
 
 static void end_bitmap_write(struct buffer_head *bh, int uptodate)
 {
 	struct bitmap *bitmap = bh->b_private;
-	unsigned long flags;
 
-	if (!uptodate) {
-		spin_lock_irqsave(&bitmap->lock, flags);
-		bitmap->flags |= BITMAP_WRITE_ERROR;
-		spin_unlock_irqrestore(&bitmap->lock, flags);
-	}
+	if (!uptodate)
+		set_bit(BITMAP_WRITE_ERROR, &bitmap->flags);
 	if (atomic_dec_and_test(&bitmap->pending_writes))
 		wake_up(&bitmap->write_wait);
 }
@@ -325,8 +312,12 @@ __clear_page_buffers(struct page *page)
 }
 static void free_buffers(struct page *page)
 {
-	struct buffer_head *bh = page_buffers(page);
+	struct buffer_head *bh;
 
+	if (!PagePrivate(page))
+		return;
+
+	bh = page_buffers(page);
 	while (bh) {
 		struct buffer_head *next = bh->b_this_page;
 		free_buffer_head(bh);
@@ -343,11 +334,12 @@ static void free_buffers(struct page *page)
  * This usage is similar to how swap files are handled, and allows us
  * to write to a file with no concerns of memory allocation failing.
  */
-static struct page *read_page(struct file *file, unsigned long index,
-			      struct bitmap *bitmap,
-			      unsigned long count)
+static int read_page(struct file *file, unsigned long index,
+		     struct bitmap *bitmap,
+		     unsigned long count,
+		     struct page *page)
 {
-	struct page *page = NULL;
+	int ret = 0;
 	struct inode *inode = file->f_path.dentry->d_inode;
 	struct buffer_head *bh;
 	sector_t block;
@@ -355,16 +347,9 @@ static struct page *read_page(struct file *file, unsigned long index,
 	pr_debug("read bitmap file (%dB @ %llu)\n", (int)PAGE_SIZE,
 		 (unsigned long long)index << PAGE_SHIFT);
 
-	page = alloc_page(GFP_KERNEL);
-	if (!page)
-		page = ERR_PTR(-ENOMEM);
-	if (IS_ERR(page))
-		goto out;
-
 	bh = alloc_page_buffers(page, 1<<inode->i_blkbits, 0);
 	if (!bh) {
-		put_page(page);
-		page = ERR_PTR(-ENOMEM);
+		ret = -ENOMEM;
 		goto out;
 	}
 	attach_page_buffers(page, bh);
@@ -376,8 +361,7 @@ static struct page *read_page(struct file *file, unsigned long index,
 			bh->b_blocknr = bmap(inode, block);
 			if (bh->b_blocknr == 0) {
 				/* Cannot use this file! */
-				free_buffers(page);
-				page = ERR_PTR(-EINVAL);
+				ret = -EINVAL;
 				goto out;
 			}
 			bh->b_bdev = inode->i_sb->s_bdev;
@@ -400,17 +384,15 @@ static struct page *read_page(struct file *file, unsigned long index,
 
 	wait_event(bitmap->write_wait,
 		   atomic_read(&bitmap->pending_writes)==0);
-	if (bitmap->flags & BITMAP_WRITE_ERROR) {
-		free_buffers(page);
-		page = ERR_PTR(-EIO);
-	}
+	if (test_bit(BITMAP_WRITE_ERROR, &bitmap->flags))
+		ret = -EIO;
 out:
-	if (IS_ERR(page))
-		printk(KERN_ALERT "md: bitmap read error: (%dB @ %llu): %ld\n",
+	if (ret)
+		printk(KERN_ALERT "md: bitmap read error: (%dB @ %llu): %d\n",
 			(int)PAGE_SIZE,
 			(unsigned long long)index << PAGE_SHIFT,
-			PTR_ERR(page));
-	return page;
+			ret);
+	return ret;
 }
 
 /*
@@ -426,9 +408,9 @@ void bitmap_update_sb(struct bitmap *bitmap)
 		return;
 	if (bitmap->mddev->bitmap_info.external)
 		return;
-	if (!bitmap->sb_page) /* no superblock */
+	if (!bitmap->storage.sb_page) /* no superblock */
 		return;
-	sb = kmap_atomic(bitmap->sb_page);
+	sb = kmap_atomic(bitmap->storage.sb_page);
 	sb->events = cpu_to_le64(bitmap->mddev->events);
 	if (bitmap->mddev->events < bitmap->events_cleared)
 		/* rocking back to read-only */
@@ -438,8 +420,13 @@ void bitmap_update_sb(struct bitmap *bitmap)
 	/* Just in case these have been changed via sysfs: */
 	sb->daemon_sleep = cpu_to_le32(bitmap->mddev->bitmap_info.daemon_sleep/HZ);
 	sb->write_behind = cpu_to_le32(bitmap->mddev->bitmap_info.max_write_behind);
+	/* This might have been changed by a reshape */
+	sb->sync_size = cpu_to_le64(bitmap->mddev->resync_max_sectors);
+	sb->chunksize = cpu_to_le32(bitmap->mddev->bitmap_info.chunksize);
+	sb->sectors_reserved = cpu_to_le32(bitmap->mddev->
+					   bitmap_info.space);
 	kunmap_atomic(sb);
-	write_page(bitmap, bitmap->sb_page, 1);
+	write_page(bitmap, bitmap->storage.sb_page, 1);
 }
 
 /* print out the bitmap file superblock */
@@ -447,9 +434,9 @@ void bitmap_print_sb(struct bitmap *bitmap)
 {
 	bitmap_super_t *sb;
 
-	if (!bitmap || !bitmap->sb_page)
+	if (!bitmap || !bitmap->storage.sb_page)
 		return;
-	sb = kmap_atomic(bitmap->sb_page);
+	sb = kmap_atomic(bitmap->storage.sb_page);
 	printk(KERN_DEBUG "%s: bitmap file superblock:\n", bmname(bitmap));
 	printk(KERN_DEBUG "         magic: %08x\n", le32_to_cpu(sb->magic));
 	printk(KERN_DEBUG "       version: %d\n", le32_to_cpu(sb->version));
@@ -488,15 +475,15 @@ static int bitmap_new_disk_sb(struct bitmap *bitmap)
 	unsigned long chunksize, daemon_sleep, write_behind;
 	int err = -EINVAL;
 
-	bitmap->sb_page = alloc_page(GFP_KERNEL);
-	if (IS_ERR(bitmap->sb_page)) {
-		err = PTR_ERR(bitmap->sb_page);
-		bitmap->sb_page = NULL;
+	bitmap->storage.sb_page = alloc_page(GFP_KERNEL);
+	if (IS_ERR(bitmap->storage.sb_page)) {
+		err = PTR_ERR(bitmap->storage.sb_page);
+		bitmap->storage.sb_page = NULL;
 		return err;
 	}
-	bitmap->sb_page->index = 0;
+	bitmap->storage.sb_page->index = 0;
 
-	sb = kmap_atomic(bitmap->sb_page);
+	sb = kmap_atomic(bitmap->storage.sb_page);
 
 	sb->magic = cpu_to_le32(BITMAP_MAGIC);
 	sb->version = cpu_to_le32(BITMAP_MAJOR_HI);
@@ -534,8 +521,8 @@ static int bitmap_new_disk_sb(struct bitmap *bitmap)
 
 	memcpy(sb->uuid, bitmap->mddev->uuid, 16);
 
-	bitmap->flags |= BITMAP_STALE;
-	sb->state |= cpu_to_le32(BITMAP_STALE);
+	set_bit(BITMAP_STALE, &bitmap->flags);
+	sb->state = cpu_to_le32(bitmap->flags);
 	bitmap->events_cleared = bitmap->mddev->events;
 	sb->events_cleared = cpu_to_le64(bitmap->mddev->events);
 
@@ -551,31 +538,45 @@ static int bitmap_read_sb(struct bitmap *bitmap)
 	bitmap_super_t *sb;
 	unsigned long chunksize, daemon_sleep, write_behind;
 	unsigned long long events;
+	unsigned long sectors_reserved = 0;
 	int err = -EINVAL;
+	struct page *sb_page;
 
+	if (!bitmap->storage.file && !bitmap->mddev->bitmap_info.offset) {
+		chunksize = 128 * 1024 * 1024;
+		daemon_sleep = 5 * HZ;
+		write_behind = 0;
+		set_bit(BITMAP_STALE, &bitmap->flags);
+		err = 0;
+		goto out_no_sb;
+	}
 	/* page 0 is the superblock, read it... */
-	if (bitmap->file) {
-		loff_t isize = i_size_read(bitmap->file->f_mapping->host);
+	sb_page = alloc_page(GFP_KERNEL);
+	if (!sb_page)
+		return -ENOMEM;
+	bitmap->storage.sb_page = sb_page;
+
+	if (bitmap->storage.file) {
+		loff_t isize = i_size_read(bitmap->storage.file->f_mapping->host);
 		int bytes = isize > PAGE_SIZE ? PAGE_SIZE : isize;
 
-		bitmap->sb_page = read_page(bitmap->file, 0, bitmap, bytes);
+		err = read_page(bitmap->storage.file, 0,
+				bitmap, bytes, sb_page);
 	} else {
-		bitmap->sb_page = read_sb_page(bitmap->mddev,
-					       bitmap->mddev->bitmap_info.offset,
-					       NULL,
-					       0, sizeof(bitmap_super_t));
+		err = read_sb_page(bitmap->mddev,
+				   bitmap->mddev->bitmap_info.offset,
+				   sb_page,
+				   0, sizeof(bitmap_super_t));
 	}
-	if (IS_ERR(bitmap->sb_page)) {
-		err = PTR_ERR(bitmap->sb_page);
-		bitmap->sb_page = NULL;
+	if (err)
 		return err;
-	}
 
-	sb = kmap_atomic(bitmap->sb_page);
+	sb = kmap_atomic(sb_page);
 
 	chunksize = le32_to_cpu(sb->chunksize);
 	daemon_sleep = le32_to_cpu(sb->daemon_sleep) * HZ;
 	write_behind = le32_to_cpu(sb->write_behind);
+	sectors_reserved = le32_to_cpu(sb->sectors_reserved);
 
 	/* verify that the bitmap-specific fields are valid */
 	if (sb->magic != cpu_to_le32(BITMAP_MAGIC))
@@ -618,60 +619,32 @@ static int bitmap_read_sb(struct bitmap *bitmap)
 			       "-- forcing full recovery\n",
 			       bmname(bitmap), events,
 			       (unsigned long long) bitmap->mddev->events);
-			sb->state |= cpu_to_le32(BITMAP_STALE);
+			set_bit(BITMAP_STALE, &bitmap->flags);
 		}
 	}
 
 	/* assign fields using values from superblock */
-	bitmap->mddev->bitmap_info.chunksize = chunksize;
-	bitmap->mddev->bitmap_info.daemon_sleep = daemon_sleep;
-	bitmap->mddev->bitmap_info.max_write_behind = write_behind;
 	bitmap->flags |= le32_to_cpu(sb->state);
 	if (le32_to_cpu(sb->version) == BITMAP_MAJOR_HOSTENDIAN)
-		bitmap->flags |= BITMAP_HOSTENDIAN;
+		set_bit(BITMAP_HOSTENDIAN, &bitmap->flags);
 	bitmap->events_cleared = le64_to_cpu(sb->events_cleared);
-	if (bitmap->flags & BITMAP_STALE)
-		bitmap->events_cleared = bitmap->mddev->events;
 	err = 0;
 out:
 	kunmap_atomic(sb);
+out_no_sb:
+	if (test_bit(BITMAP_STALE, &bitmap->flags))
+		bitmap->events_cleared = bitmap->mddev->events;
+	bitmap->mddev->bitmap_info.chunksize = chunksize;
+	bitmap->mddev->bitmap_info.daemon_sleep = daemon_sleep;
+	bitmap->mddev->bitmap_info.max_write_behind = write_behind;
+	if (bitmap->mddev->bitmap_info.space == 0 ||
+	    bitmap->mddev->bitmap_info.space > sectors_reserved)
+		bitmap->mddev->bitmap_info.space = sectors_reserved;
 	if (err)
 		bitmap_print_sb(bitmap);
 	return err;
 }
 
-enum bitmap_mask_op {
-	MASK_SET,
-	MASK_UNSET
-};
-
-/* record the state of the bitmap in the superblock.  Return the old value */
-static int bitmap_mask_state(struct bitmap *bitmap, enum bitmap_state bits,
-			     enum bitmap_mask_op op)
-{
-	bitmap_super_t *sb;
-	int old;
-
-	if (!bitmap->sb_page) /* can't set the state */
-		return 0;
-	sb = kmap_atomic(bitmap->sb_page);
-	old = le32_to_cpu(sb->state) & bits;
-	switch (op) {
-	case MASK_SET:
-		sb->state |= cpu_to_le32(bits);
-		bitmap->flags |= bits;
-		break;
-	case MASK_UNSET:
-		sb->state &= cpu_to_le32(~bits);
-		bitmap->flags &= ~bits;
-		break;
-	default:
-		BUG();
-	}
-	kunmap_atomic(sb);
-	return old;
-}
-
 /*
  * general bitmap file operations
  */
@@ -683,17 +656,19 @@ static int bitmap_mask_state(struct bitmap *bitmap, enum bitmap_state bits,
  * file a page at a time. There's a superblock at the start of the file.
  */
 /* calculate the index of the page that contains this bit */
-static inline unsigned long file_page_index(struct bitmap *bitmap, unsigned long chunk)
+static inline unsigned long file_page_index(struct bitmap_storage *store,
+					    unsigned long chunk)
 {
-	if (!bitmap->mddev->bitmap_info.external)
+	if (store->sb_page)
 		chunk += sizeof(bitmap_super_t) << 3;
 	return chunk >> PAGE_BIT_SHIFT;
 }
 
 /* calculate the (bit) offset of this bit within a page */
-static inline unsigned long file_page_offset(struct bitmap *bitmap, unsigned long chunk)
+static inline unsigned long file_page_offset(struct bitmap_storage *store,
+					     unsigned long chunk)
 {
-	if (!bitmap->mddev->bitmap_info.external)
+	if (store->sb_page)
 		chunk += sizeof(bitmap_super_t) << 3;
 	return chunk & (PAGE_BITS - 1);
 }
@@ -705,57 +680,86 @@ static inline unsigned long file_page_offset(struct bitmap *bitmap, unsigned lon
  * 1 page (e.g., x86) or less than 1 page -- so the bitmap might start on page
  * 0 or page 1
  */
-static inline struct page *filemap_get_page(struct bitmap *bitmap,
+static inline struct page *filemap_get_page(struct bitmap_storage *store,
 					    unsigned long chunk)
 {
-	if (file_page_index(bitmap, chunk) >= bitmap->file_pages)
+	if (file_page_index(store, chunk) >= store->file_pages)
 		return NULL;
-	return bitmap->filemap[file_page_index(bitmap, chunk)
-			       - file_page_index(bitmap, 0)];
+	return store->filemap[file_page_index(store, chunk)
+			      - file_page_index(store, 0)];
 }
 
-static void bitmap_file_unmap(struct bitmap *bitmap)
+static int bitmap_storage_alloc(struct bitmap_storage *store,
+				unsigned long chunks, int with_super)
+{
+	int pnum;
+	unsigned long num_pages;
+	unsigned long bytes;
+
+	bytes = DIV_ROUND_UP(chunks, 8);
+	if (with_super)
+		bytes += sizeof(bitmap_super_t);
+
+	num_pages = DIV_ROUND_UP(bytes, PAGE_SIZE);
+
+	store->filemap = kmalloc(sizeof(struct page *)
+				 * num_pages, GFP_KERNEL);
+	if (!store->filemap)
+		return -ENOMEM;
+
+	if (with_super && !store->sb_page) {
+		store->sb_page = alloc_page(GFP_KERNEL|__GFP_ZERO);
+		if (store->sb_page == NULL)
+			return -ENOMEM;
+		store->sb_page->index = 0;
+	}
+	pnum = 0;
+	if (store->sb_page) {
+		store->filemap[0] = store->sb_page;
+		pnum = 1;
+	}
+	for ( ; pnum < num_pages; pnum++) {
+		store->filemap[pnum] = alloc_page(GFP_KERNEL|__GFP_ZERO);
+		if (!store->filemap[pnum]) {
+			store->file_pages = pnum;
+			return -ENOMEM;
+		}
+		store->filemap[pnum]->index = pnum;
+	}
+	store->file_pages = pnum;
+
+	/* We need 4 bits per page, rounded up to a multiple
+	 * of sizeof(unsigned long) */
+	store->filemap_attr = kzalloc(
+		roundup(DIV_ROUND_UP(num_pages*4, 8), sizeof(unsigned long)),
+		GFP_KERNEL);
+	if (!store->filemap_attr)
+		return -ENOMEM;
+
+	store->bytes = bytes;
+
+	return 0;
+}
+
+static void bitmap_file_unmap(struct bitmap_storage *store)
 {
 	struct page **map, *sb_page;
-	unsigned long *attr;
 	int pages;
-	unsigned long flags;
+	struct file *file;
 
-	spin_lock_irqsave(&bitmap->lock, flags);
-	map = bitmap->filemap;
-	bitmap->filemap = NULL;
-	attr = bitmap->filemap_attr;
-	bitmap->filemap_attr = NULL;
-	pages = bitmap->file_pages;
-	bitmap->file_pages = 0;
-	sb_page = bitmap->sb_page;
-	bitmap->sb_page = NULL;
-	spin_unlock_irqrestore(&bitmap->lock, flags);