Merge with rsync://rsync.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git

Documentation/dontdiff

@@ -41,6 +41,7 @@ COPYING
 CREDITS
 CVS
 ChangeSet
+Image
 Kerntypes
 MODS.txt
 Module.symvers
@@ -103,6 +104,7 @@ logo_*.c
 logo_*_clut224.c
 logo_*_mono.c
 lxdialog
+mach-types.h
 make_times_h
 map
 maui_boot.h

Documentation/feature-removal-schedule.txt

@@ -103,11 +103,11 @@ Who:	Jody McIntyre <scjody@steamballoon.com>
 ---------------------------
 
 What:	register_serial/unregister_serial
-When:	December 2005
+When:	September 2005
 Why:	This interface does not allow serial ports to be registered against
 	a struct device, and as such does not allow correct power management
 	of such ports.  8250-based ports should use serial8250_register_port
-	and serial8250_unregister_port instead.
+	and serial8250_unregister_port, or platform devices instead.
 Who:	Russell King <rmk@arm.linux.org.uk>
 
 ---------------------------

Documentation/filesystems/ntfs.txt

@@ -21,7 +21,7 @@ Overview
 ========
 
 Linux-NTFS comes with a number of user-space programs known as ntfsprogs.
-These include mkntfs, a full-featured ntfs file system format utility,
+These include mkntfs, a full-featured ntfs filesystem format utility,
 ntfsundelete used for recovering files that were unintentionally deleted
 from an NTFS volume and ntfsresize which is used to resize an NTFS partition.
 See the web site for more information.
@@ -149,7 +149,14 @@ case_sensitive=<BOOL>	If case_sensitive is specified, treat all file names as
 			name, if it exists.  If case_sensitive, you will need
 			to provide the correct case of the short file name.
 
-errors=opt		What to do when critical file system errors are found.
+disable_sparse=<BOOL>	If disable_sparse is specified, creation of sparse
+			regions, i.e. holes, inside files is disabled for the
+			volume (for the duration of this mount only).  By
+			default, creation of sparse regions is enabled, which
+			is consistent with the behaviour of traditional Unix
+			filesystems.
+
+errors=opt		What to do when critical filesystem errors are found.
 			Following values can be used for "opt":
 			  continue: DEFAULT, try to clean-up as much as
 				    possible, e.g. marking a corrupt inode as
@@ -432,6 +439,24 @@ ChangeLog
 
 Note, a technical ChangeLog aimed at kernel hackers is in fs/ntfs/ChangeLog.
 
+2.1.23:
+	- Stamp the user space journal, aka transaction log, aka $UsnJrnl, if
+	  it is present and active thus telling Windows and applications using
+	  the transaction log that changes can have happened on the volume
+	  which are not recorded in $UsnJrnl.
+	- Detect the case when Windows has been hibernated (suspended to disk)
+	  and if this is the case do not allow (re)mounting read-write to
+	  prevent data corruption when you boot back into the suspended
+	  Windows session.
+	- Implement extension of resident files using the normal file write
+	  code paths, i.e. most very small files can be extended to be a little
+	  bit bigger but not by much.
+	- Add new mount option "disable_sparse".  (See list of mount options
+	  above for details.)
+	- Improve handling of ntfs volumes with errors and strange boot sectors
+	  in particular.
+	- Fix various bugs including a nasty deadlock that appeared in recent
+	  kernels (around 2.6.11-2.6.12 timeframe).
 2.1.22:
 	- Improve handling of ntfs volumes with errors.
 	- Fix various bugs and race conditions.

arch/arm/configs/shark_defconfig

@@ -1,22 +1,21 @@
 #
 # Automatically generated make config: don't edit
-# Linux kernel version: 2.6.12-rc1-bk2
+# Linux kernel version: 2.6.12-git3
-# Sun Mar 27 23:59:14 2005
+# Sat Jul 16 15:21:47 2005
 #
 CONFIG_ARM=y
 CONFIG_MMU=y
 CONFIG_UID16=y
 CONFIG_RWSEM_GENERIC_SPINLOCK=y
 CONFIG_GENERIC_CALIBRATE_DELAY=y
-CONFIG_GENERIC_IOMAP=y
 
 #
 # Code maturity level options
 #
 CONFIG_EXPERIMENTAL=y
-# CONFIG_CLEAN_COMPILE is not set
+CONFIG_CLEAN_COMPILE=y
-CONFIG_BROKEN=y
 CONFIG_BROKEN_ON_SMP=y
+CONFIG_INIT_ENV_ARG_LIMIT=32
 
 #
 # General setup
@@ -33,7 +32,10 @@ CONFIG_KOBJECT_UEVENT=y
 # CONFIG_IKCONFIG is not set
 # CONFIG_EMBEDDED is not set
 CONFIG_KALLSYMS=y
+# CONFIG_KALLSYMS_ALL is not set
 # CONFIG_KALLSYMS_EXTRA_PASS is not set
+CONFIG_PRINTK=y
+CONFIG_BUG=y
 CONFIG_BASE_FULL=y
 CONFIG_FUTEX=y
 CONFIG_EPOLL=y
@@ -81,6 +83,7 @@ CONFIG_ARCH_SHARK=y
 # CONFIG_ARCH_VERSATILE is not set
 # CONFIG_ARCH_IMX is not set
 # CONFIG_ARCH_H720X is not set
+# CONFIG_ARCH_AAEC2000 is not set
 
 #
 # Processor Type
@@ -103,10 +106,12 @@ CONFIG_CPU_TLB_V4WB=y
 #
 CONFIG_ISA=y
 CONFIG_ISA_DMA=y
+CONFIG_ISA_DMA_API=y
 CONFIG_PCI=y
 CONFIG_PCI_HOST_VIA82C505=y
 CONFIG_PCI_LEGACY_PROC=y
 # CONFIG_PCI_NAMES is not set
+# CONFIG_PCI_DEBUG is not set
 
 #
 # PCCARD (PCMCIA/CardBus) support
@@ -116,7 +121,9 @@ CONFIG_PCI_LEGACY_PROC=y
 #
 # Kernel Features
 #
+# CONFIG_SMP is not set
 # CONFIG_PREEMPT is not set
+# CONFIG_DISCONTIGMEM is not set
 CONFIG_LEDS=y
 CONFIG_LEDS_TIMER=y
 # CONFIG_LEDS_CPU is not set
@@ -163,6 +170,7 @@ CONFIG_BINFMT_ELF=y
 # CONFIG_STANDALONE is not set
 CONFIG_PREVENT_FIRMWARE_BUILD=y
 # CONFIG_FW_LOADER is not set
+# CONFIG_DEBUG_DRIVER is not set
 
 #
 # Memory Technology Devices (MTD)
@@ -172,8 +180,8 @@ CONFIG_PREVENT_FIRMWARE_BUILD=y
 #
 # Parallel port support
 #
-CONFIG_PARPORT=y
+CONFIG_PARPORT=m
-CONFIG_PARPORT_PC=y
+CONFIG_PARPORT_PC=m
 # CONFIG_PARPORT_SERIAL is not set
 # CONFIG_PARPORT_PC_FIFO is not set
 # CONFIG_PARPORT_PC_SUPERIO is not set
@@ -189,7 +197,6 @@ CONFIG_PARPORT_PC=y
 #
 # Block devices
 #
-# CONFIG_BLK_DEV_FD is not set
 # CONFIG_BLK_DEV_XD is not set
 # CONFIG_PARIDE is not set
 # CONFIG_BLK_CPQ_DA is not set
@@ -229,7 +236,7 @@ CONFIG_BLK_DEV_IDE=y
 # CONFIG_BLK_DEV_IDE_SATA is not set
 CONFIG_BLK_DEV_IDEDISK=y
 # CONFIG_IDEDISK_MULTI_MODE is not set
-CONFIG_BLK_DEV_IDECD=y
+CONFIG_BLK_DEV_IDECD=m
 # CONFIG_BLK_DEV_IDETAPE is not set
 CONFIG_BLK_DEV_IDEFLOPPY=y
 # CONFIG_BLK_DEV_IDESCSI is not set
@@ -261,6 +268,7 @@ CONFIG_CHR_DEV_ST=m
 CONFIG_BLK_DEV_SR=m
 # CONFIG_BLK_DEV_SR_VENDOR is not set
 CONFIG_CHR_DEV_SG=m
+# CONFIG_CHR_DEV_SCH is not set
 
 #
 # Some SCSI devices (e.g. CD jukebox) support multiple LUNs
@@ -290,17 +298,14 @@ CONFIG_CHR_DEV_SG=m
 # CONFIG_SCSI_AIC7XXX_OLD is not set
 # CONFIG_SCSI_AIC79XX is not set
 # CONFIG_SCSI_DPT_I2O is not set
-# CONFIG_SCSI_ADVANSYS is not set
 # CONFIG_SCSI_IN2000 is not set
 # CONFIG_MEGARAID_NEWGEN is not set
 # CONFIG_MEGARAID_LEGACY is not set
 # CONFIG_SCSI_SATA is not set
 # CONFIG_SCSI_BUSLOGIC is not set
-# CONFIG_SCSI_CPQFCTS is not set
 # CONFIG_SCSI_DMX3191D is not set
 # CONFIG_SCSI_DTC3280 is not set
 # CONFIG_SCSI_EATA is not set
-# CONFIG_SCSI_EATA_PIO is not set
 # CONFIG_SCSI_FUTURE_DOMAIN is not set
 # CONFIG_SCSI_GDTH is not set
 # CONFIG_SCSI_GENERIC_NCR5380 is not set
@@ -314,11 +319,8 @@ CONFIG_CHR_DEV_SG=m
 # CONFIG_SCSI_SYM53C8XX_2 is not set
 # CONFIG_SCSI_IPR is not set
 # CONFIG_SCSI_PAS16 is not set
-# CONFIG_SCSI_PCI2000 is not set
-# CONFIG_SCSI_PCI2220I is not set
 # CONFIG_SCSI_PSI240I is not set
 # CONFIG_SCSI_QLOGIC_FAS is not set
-# CONFIG_SCSI_QLOGIC_ISP is not set
 # CONFIG_SCSI_QLOGIC_FC is not set
 # CONFIG_SCSI_QLOGIC_1280 is not set
 CONFIG_SCSI_QLA2XXX=m
@@ -327,6 +329,7 @@ CONFIG_SCSI_QLA2XXX=m
 # CONFIG_SCSI_QLA2300 is not set
 # CONFIG_SCSI_QLA2322 is not set
 # CONFIG_SCSI_QLA6312 is not set
+# CONFIG_SCSI_LPFC is not set
 # CONFIG_SCSI_SYM53C416 is not set
 # CONFIG_SCSI_DC395x is not set
 # CONFIG_SCSI_DC390T is not set
@@ -344,6 +347,8 @@ CONFIG_SCSI_QLA2XXX=m
 # Fusion MPT device support
 #
 # CONFIG_FUSION is not set
+# CONFIG_FUSION_SPI is not set
+# CONFIG_FUSION_FC is not set
 
 #
 # IEEE 1394 (FireWire) support
@@ -365,7 +370,6 @@ CONFIG_NET=y
 #
 CONFIG_PACKET=y
 # CONFIG_PACKET_MMAP is not set
-# CONFIG_NETLINK_DEV is not set
 CONFIG_UNIX=y
 # CONFIG_NET_KEY is not set
 CONFIG_INET=y
@@ -380,7 +384,7 @@ CONFIG_INET=y
 # CONFIG_INET_ESP is not set
 # CONFIG_INET_IPCOMP is not set
 # CONFIG_INET_TUNNEL is not set
-# CONFIG_IP_TCPDIAG is not set
+CONFIG_IP_TCPDIAG=y
 # CONFIG_IP_TCPDIAG_IPV6 is not set
 # CONFIG_IPV6 is not set
 # CONFIG_NETFILTER is not set
@@ -439,6 +443,7 @@ CONFIG_NET_ETHERNET=y
 # CONFIG_LANCE is not set
 # CONFIG_NET_VENDOR_SMC is not set
 # CONFIG_SMC91X is not set
+# CONFIG_DM9000 is not set
 # CONFIG_NET_VENDOR_RACAL is not set
 
 #
@@ -483,9 +488,11 @@ CONFIG_CS89x0=y
 # CONFIG_HAMACHI is not set
 # CONFIG_YELLOWFIN is not set
 # CONFIG_R8169 is not set
+# CONFIG_SKGE is not set
 # CONFIG_SK98LIN is not set
 # CONFIG_VIA_VELOCITY is not set
 # CONFIG_TIGON3 is not set
+# CONFIG_BNX2 is not set
 
 #
 # Ethernet (10000 Mbit)
@@ -569,7 +576,6 @@ CONFIG_SERIO_I8042=y
 CONFIG_SERIO_LIBPS2=y
 # CONFIG_SERIO_RAW is not set
 # CONFIG_GAMEPORT is not set
-CONFIG_SOUND_GAMEPORT=y
 
 #
 # Character devices
@@ -592,6 +598,7 @@ CONFIG_SERIAL_8250_NR_UARTS=4
 #
 CONFIG_SERIAL_CORE=y
 CONFIG_SERIAL_CORE_CONSOLE=y
+# CONFIG_SERIAL_JSM is not set
 CONFIG_UNIX98_PTYS=y
 CONFIG_LEGACY_PTYS=y
 CONFIG_LEGACY_PTY_COUNT=256
@@ -653,6 +660,7 @@ CONFIG_FB_CFB_FILLRECT=y
 CONFIG_FB_CFB_COPYAREA=y
 CONFIG_FB_CFB_IMAGEBLIT=y
 CONFIG_FB_SOFT_CURSOR=y
+# CONFIG_FB_MACMODES is not set
 # CONFIG_FB_MODE_HELPERS is not set
 # CONFIG_FB_TILEBLITTING is not set
 # CONFIG_FB_CIRRUS is not set
@@ -674,7 +682,7 @@ CONFIG_FB_CYBER2000=y
 # CONFIG_FB_3DFX is not set
 # CONFIG_FB_VOODOO1 is not set
 # CONFIG_FB_TRIDENT is not set
-# CONFIG_FB_PM3 is not set
+# CONFIG_FB_S1D13XXX is not set
 # CONFIG_FB_VIRTUAL is not set
 
 #
@@ -808,7 +816,7 @@ CONFIG_DNOTIFY=y
 #
 # CD-ROM/DVD Filesystems
 #
-CONFIG_ISO9660_FS=y
+CONFIG_ISO9660_FS=m
 CONFIG_JOLIET=y
 # CONFIG_ZISOFS is not set
 # CONFIG_UDF_FS is not set
@@ -816,9 +824,9 @@ CONFIG_JOLIET=y
 #
 # DOS/FAT/NT Filesystems
 #
-CONFIG_FAT_FS=y
+CONFIG_FAT_FS=m
-CONFIG_MSDOS_FS=y
+CONFIG_MSDOS_FS=m
-CONFIG_VFAT_FS=y
+CONFIG_VFAT_FS=m
 CONFIG_FAT_DEFAULT_CODEPAGE=437
 CONFIG_FAT_DEFAULT_IOCHARSET="iso8859-1"
 # CONFIG_NTFS_FS is not set
@@ -833,7 +841,6 @@ CONFIG_DEVFS_MOUNT=y
 # CONFIG_DEVFS_DEBUG is not set
 # CONFIG_DEVPTS_FS_XATTR is not set
 # CONFIG_TMPFS is not set
-# CONFIG_HUGETLBFS is not set
 # CONFIG_HUGETLB_PAGE is not set
 CONFIG_RAMFS=y
 
@@ -857,13 +864,14 @@ CONFIG_RAMFS=y
 #
 # Network File Systems
 #
-CONFIG_NFS_FS=y
+CONFIG_NFS_FS=m
-# CONFIG_NFS_V3 is not set
+CONFIG_NFS_V3=y
 # CONFIG_NFS_V4 is not set
 # CONFIG_NFS_DIRECTIO is not set
 # CONFIG_NFSD is not set
-CONFIG_LOCKD=y
+CONFIG_LOCKD=m
-CONFIG_SUNRPC=y
+CONFIG_LOCKD_V4=y
+CONFIG_SUNRPC=m
 # CONFIG_RPCSEC_GSS_KRB5 is not set
 # CONFIG_RPCSEC_GSS_SPKM3 is not set
 # CONFIG_SMB_FS is not set
@@ -895,12 +903,12 @@ CONFIG_MSDOS_PARTITION=y
 #
 # Native Language Support
 #
-CONFIG_NLS=y
+CONFIG_NLS=m
 CONFIG_NLS_DEFAULT="iso8859-1"
-CONFIG_NLS_CODEPAGE_437=y
+CONFIG_NLS_CODEPAGE_437=m
 # CONFIG_NLS_CODEPAGE_737 is not set
 # CONFIG_NLS_CODEPAGE_775 is not set
-CONFIG_NLS_CODEPAGE_850=y
+CONFIG_NLS_CODEPAGE_850=m
 # CONFIG_NLS_CODEPAGE_852 is not set
 # CONFIG_NLS_CODEPAGE_855 is not set
 # CONFIG_NLS_CODEPAGE_857 is not set
@@ -921,7 +929,7 @@ CONFIG_NLS_CODEPAGE_850=y
 # CONFIG_NLS_CODEPAGE_1250 is not set
 # CONFIG_NLS_CODEPAGE_1251 is not set
 # CONFIG_NLS_ASCII is not set
-CONFIG_NLS_ISO8859_1=y
+CONFIG_NLS_ISO8859_1=m
 # CONFIG_NLS_ISO8859_2 is not set
 # CONFIG_NLS_ISO8859_3 is not set
 # CONFIG_NLS_ISO8859_4 is not set
@@ -945,11 +953,22 @@ CONFIG_NLS_ISO8859_1=y
 # Kernel hacking
 #
 # CONFIG_PRINTK_TIME is not set
-# CONFIG_DEBUG_KERNEL is not set
+CONFIG_DEBUG_KERNEL=y
+# CONFIG_MAGIC_SYSRQ is not set
 CONFIG_LOG_BUF_SHIFT=14
+# CONFIG_SCHEDSTATS is not set
+# CONFIG_DEBUG_SLAB is not set
+# CONFIG_DEBUG_SPINLOCK is not set
+# CONFIG_DEBUG_SPINLOCK_SLEEP is not set
+# CONFIG_DEBUG_KOBJECT is not set
 CONFIG_DEBUG_BUGVERBOSE=y
+# CONFIG_DEBUG_INFO is not set
+# CONFIG_DEBUG_FS is not set
 CONFIG_FRAME_POINTER=y
 CONFIG_DEBUG_USER=y
+# CONFIG_DEBUG_WAITQ is not set
+# CONFIG_DEBUG_ERRORS is not set
+# CONFIG_DEBUG_LL is not set
 
 #
 # Security options

arch/arm/kernel/traps.c

@@ -248,16 +248,20 @@ static DEFINE_SPINLOCK(undef_lock);
 
 void register_undef_hook(struct undef_hook *hook)
 {
-	spin_lock_irq(&undef_lock);
+	unsigned long flags;
+
+	spin_lock_irqsave(&undef_lock, flags);
 	list_add(&hook->node, &undef_hook);
-	spin_unlock_irq(&undef_lock);
+	spin_unlock_irqrestore(&undef_lock, flags);
 }
 
 void unregister_undef_hook(struct undef_hook *hook)
 {
-	spin_lock_irq(&undef_lock);
+	unsigned long flags;
+
+	spin_lock_irqsave(&undef_lock, flags);
 	list_del(&hook->node);
-	spin_unlock_irq(&undef_lock);
+	spin_unlock_irqrestore(&undef_lock, flags);
 }
 
 asmlinkage void do_undefinstr(struct pt_regs *regs)

arch/arm/lib/bitops.h

@@ -1,3 +1,33 @@
+#if __LINUX_ARM_ARCH__ >= 6
+	.macro	bitop, instr
+	mov	r2, #1
+	and	r3, r0, #7		@ Get bit offset
+	add	r1, r1, r0, lsr #3	@ Get byte offset
+	mov	r3, r2, lsl r3
+1:	ldrexb	r2, [r1]
+	\instr	r2, r2, r3
+	strexb	r0, r2, [r1]
+	cmpne	r0, #0
+	bne	1b
+	mov	pc, lr
+	.endm
+
+	.macro	testop, instr, store
+	and	r3, r0, #7		@ Get bit offset
+	mov	r2, #1
+	add	r1, r1, r0, lsr #3	@ Get byte offset
+	mov	r3, r2, lsl r3		@ create mask
+1:	ldrexb	r2, [r1]
+	ands	r0, r2, r3		@ save old value of bit
+	\instr	ip, r2, r3			@ toggle bit
+	strexb	r2, ip, [r1]
+	cmp	r2, #0
+	bne	1b
+	cmp	r0, #0
+	movne	r0, #1
+2:	mov	pc, lr
+	.endm
+#else
 	.macro	bitop, instr
 	and	r2, r0, #7
 	mov	r3, #1
@@ -31,3 +61,4 @@
 	moveq	r0, #0
 	mov	pc, lr
 	.endm
+#endif

arch/arm/lib/io-shark.c

@@ -11,73 +11,3 @@
  * it under the terms of the GNU General Public License version 2 as
  * published by the Free Software Foundation.
  */
-#include <linux/kernel.h>
-
-#include <asm/io.h>
-
-void print_warning(void)
-{
-	printk(KERN_WARNING "ins?/outs? not implemented on this architecture\n");
-}
-
-void insl(unsigned int port, void *to, int len)
-{
-	print_warning();
-}
-
-void insb(unsigned int port, void *to, int len)
-{
-	print_warning();
-}
-
-void outsl(unsigned int port, const void *from, int len)
-{
-	print_warning();
-}
-
-void outsb(unsigned int port, const void *from, int len)
-{
-	print_warning();
-}
-
-/* these should be in assembler again */
-
-/*
- * Purpose: read a block of data from a hardware register to memory.
- * Proto  : insw(int from_port, void *to, int len_in_words);
- * Proto  : inswb(int from_port, void *to, int len_in_bytes);
- * Notes  : increment to
- */
-
-void insw(unsigned int port, void *to, int len)
-{
-	int i;
-
-	for (i = 0; i < len; i++)
-		((unsigned short *) to)[i] = inw(port);
-}
-
-void inswb(unsigned int port, void *to, int len)
-{
-	insw(port, to, len >> 2);
-}
-
-/*
- * Purpose: write a block of data from memory to a hardware register.
- * Proto  : outsw(int to_reg, void *from, int len_in_words);
- * Proto  : outswb(int to_reg, void *from, int len_in_bytes);
- * Notes  : increments from
- */
-
-void outsw(unsigned int port, const void *from, int len)
-{
-	int i;
-
-	for (i = 0; i < len; i++)
-		outw(((unsigned short *) from)[i], port);
-}
-
-void outswb(unsigned int port, const void *from, int len)
-{
-	outsw(port, from, len >> 2);
-}

arch/arm/mach-shark/core.c

@@ -24,7 +24,7 @@ static struct plat_serial8250_port serial_platform_data[] = {
 		.iobase		= 0x3f8,
 		.irq		= 4,
 		.uartclk	= 1843200,
-		.regshift	= 2,
+		.regshift	= 0,
 		.iotype		= UPIO_PORT,
 		.flags		= UPF_BOOT_AUTOCONF | UPF_SKIP_TEST,
 	},
@@ -32,7 +32,7 @@ static struct plat_serial8250_port serial_platform_data[] = {
 		.iobase		= 0x2f8,
 		.irq		= 3,
 		.uartclk	= 1843200,
-		.regshift	= 2,
+		.regshift	= 0,
 		.iotype		= UPIO_PORT,
 		.flags		= UPF_BOOT_AUTOCONF | UPF_SKIP_TEST,
 	},

arch/ia64/kernel/topology.c

@@ -36,12 +36,14 @@ int arch_register_cpu(int num)
 	parent = &sysfs_nodes[cpu_to_node(num)];
 #endif /* CONFIG_NUMA */
 
+#ifdef CONFIG_ACPI_BOOT
 	/*
 	 * If CPEI cannot be re-targetted, and this is
 	 * CPEI target, then dont create the control file
 	 */
 	if (!can_cpei_retarget() && is_cpu_cpei_target(num))
 		sysfs_cpus[num].cpu.no_control = 1;
+#endif
 
 	return register_cpu(&sysfs_cpus[num].cpu, num, parent);
 }

arch/x86_64/ia32/syscall32.c

@@ -57,6 +57,7 @@ int syscall32_setup_pages(struct linux_binprm *bprm, int exstack)
 	int npages = (VSYSCALL32_END - VSYSCALL32_BASE) >> PAGE_SHIFT;
 	struct vm_area_struct *vma;
 	struct mm_struct *mm = current->mm;
+	int ret;
 
 	vma = kmem_cache_alloc(vm_area_cachep, SLAB_KERNEL);
 	if (!vma)
@@ -78,7 +79,11 @@ int syscall32_setup_pages(struct linux_binprm *bprm, int exstack)
 	vma->vm_mm = mm;
 
 	down_write(&mm->mmap_sem);
-	insert_vm_struct(mm, vma);
+	if ((ret = insert_vm_struct(mm, vma))) {
+		up_write(&mm->mmap_sem);
+		kmem_cache_free(vm_area_cachep, vma);
+		return ret;
+	}
 	mm->total_vm += npages;
 	up_write(&mm->mmap_sem);
 	return 0;

drivers/mtd/nand/nand_base.c

@@ -59,7 +59,7 @@
  *	The AG-AND chips have nice features for speed improvement,
  *	which are not supported yet. Read / program 4 pages in one go.
  *
- * $Id: nand_base.c,v 1.146 2005/06/17 15:02:06 gleixner Exp $
+ * $Id: nand_base.c,v 1.147 2005/07/15 07:18:06 gleixner Exp $
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 2 as
@@ -1410,16 +1410,6 @@ static int nand_read_oob (struct mtd_info *mtd, loff_t from, size_t len, size_t
 		this->read_buf(mtd, &buf[i], thislen);
 		i += thislen;
 
-		/* Apply delay or wait for ready/busy pin 
-		 * Do this before the AUTOINCR check, so no problems
-		 * arise if a chip which does auto increment
-		 * is marked as NOAUTOINCR by the board driver.
-		*/
-		if (!this->dev_ready) 
-			udelay (this->chip_delay);
-		else
-			nand_wait_ready(mtd);
-
 		/* Read more ? */
 		if (i < len) {
 			page++;
@@ -1432,6 +1422,16 @@ static int nand_read_oob (struct mtd_info *mtd, loff_t from, size_t len, size_t
 				this->select_chip(mtd, chipnr);
 			}
 				
+			/* Apply delay or wait for ready/busy pin 
+			 * Do this before the AUTOINCR check, so no problems
+			 * arise if a chip which does auto increment
+			 * is marked as NOAUTOINCR by the board driver.
+			 */
+			if (!this->dev_ready) 
+				udelay (this->chip_delay);
+			else
+				nand_wait_ready(mtd);
+
 			/* Check, if the chip supports auto page increment 
 			 * or if we have hit a block boundary. 
 			*/ 

drivers/mtd/nand/nand_bbt.c

@@ -6,7 +6,7 @@
  *   
  *  Copyright (C) 2004 Thomas Gleixner (tglx@linutronix.de)
  *
- * $Id: nand_bbt.c,v 1.33 2005/06/14 15:47:56 gleixner Exp $
+ * $Id: nand_bbt.c,v 1.35 2005/07/15 13:53:47 gleixner Exp $
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 2 as
@@ -109,24 +109,21 @@ static int check_pattern (uint8_t *buf, int len, int paglen, struct nand_bbt_des
 /** 
  * check_short_pattern - [GENERIC] check if a pattern is in the buffer
  * @buf:	the buffer to search
- * @len:	the length of buffer to search
- * @paglen:	the pagelength
  * @td:		search pattern descriptor
  *
  * Check for a pattern at the given place. Used to search bad block
  * tables and good / bad block identifiers. Same as check_pattern, but 
- * no optional empty check and the pattern is expected to start
+ * no optional empty check
- * at offset 0.
  *
 */
-static int check_short_pattern (uint8_t *buf, int len, int paglen, struct nand_bbt_descr *td)
+static int check_short_pattern (uint8_t *buf, struct nand_bbt_descr *td)
 {
 	int i;
 	uint8_t *p = buf;
 
 	/* Compare the pattern */
 	for (i = 0; i < td->len; i++) {
-		if (p[i] != td->pattern[i])
+		if (p[td->offs + i] != td->pattern[i])
 			return -1;
 	}
 	return 0;
@@ -337,13 +334,14 @@ static int create_bbt (struct mtd_info *mtd, uint8_t *buf, struct nand_bbt_descr
 			if (!(bd->options & NAND_BBT_SCANEMPTY)) {
 				size_t retlen;
 				
-				/* No need to read pages fully, just read required OOB bytes */
+				/* Read the full oob until read_oob is fixed to 
-				ret = mtd->read_oob(mtd, from + j * mtd->oobblock + bd->offs,
+				 * handle single byte reads for 16 bit buswidth */
-							readlen, &retlen, &buf[0]);
+				ret = mtd->read_oob(mtd, from + j * mtd->oobblock,
+							mtd->oobsize, &retlen, buf);
 				if (ret)
 					return ret;
 
-				if (check_short_pattern (&buf[j * scanlen], scanlen, mtd->oobblock, bd)) {
+				if (check_short_pattern (buf, bd)) {
 					this->bbt[i >> 3] |= 0x03 << (i & 0x6);
 					printk (KERN_WARNING "Bad eraseblock %d at 0x%08x\n", 
 						i >> 1, (unsigned int) from);

drivers/serial/crisv10.c

@@ -426,8 +426,6 @@
 static char *serial_version = "$Revision: 1.25 $";
 
 #include <linux/config.h>
-#include <linux/version.h>
-
 #include <linux/types.h>
 #include <linux/errno.h>
 #include <linux/signal.h>

drivers/serial/icom.c

@@ -25,7 +25,6 @@
 #define SERIAL_DO_RESTART
 #include <linux/module.h>
 #include <linux/config.h>
-#include <linux/version.h>
 #include <linux/kernel.h>
 #include <linux/errno.h>
 #include <linux/signal.h>

drivers/serial/jsm/jsm.h

@@ -28,7 +28,6 @@
 #define __JSM_DRIVER_H
 
 #include <linux/kernel.h>
-#include <linux/version.h>
 #include <linux/types.h>	/* To pick up the varions Linux types */
 #include <linux/tty.h>
 #include <linux/serial_core.h>

fs/jffs2/erase.c

@@ -7,7 +7,7 @@
  *
  * For licensing information, see the file 'LICENCE' in this directory.
  *
- * $Id: erase.c,v 1.76 2005/05/03 15:11:40 dedekind Exp $
+ * $Id: erase.c,v 1.80 2005/07/14 19:46:24 joern Exp $
  *
  */
 
@@ -300,100 +300,86 @@ static void jffs2_free_all_node_refs(struct jffs2_sb_info *c, struct jffs2_erase
 	jeb->last_node = NULL;
 }
 
+static int jffs2_block_check_erase(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, uint32_t *bad_offset)
+{
+	void *ebuf;
+	uint32_t ofs;
+	size_t retlen;
+	int ret = -EIO;
+	
+	ebuf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+	if (!ebuf) {
+		printk(KERN_WARNING "Failed to allocate page buffer for verifying erase at 0x%08x. Refiling\n", jeb->offset);
+		return -EAGAIN;
+	}
+
+	D1(printk(KERN_DEBUG "Verifying erase at 0x%08x\n", jeb->offset));
+
+	for (ofs = jeb->offset; ofs < jeb->offset + c->sector_size; ) {
+		uint32_t readlen = min((uint32_t)PAGE_SIZE, jeb->offset + c->sector_size - ofs);
+		int i;
+
+		*bad_offset = ofs;
+
+		ret = jffs2_flash_read(c, ofs, readlen, &retlen, ebuf);
+		if (ret) {
+			printk(KERN_WARNING "Read of newly-erased block at 0x%08x failed: %d. Putting on bad_list\n", ofs, ret);
+			goto fail;
+		}
+		if (retlen != readlen) {
+			printk(KERN_WARNING "Short read from newly-erased block at 0x%08x. Wanted %d, got %zd\n", ofs, readlen, retlen);
+			goto fail;
+		}
+		for (i=0; i<readlen; i += sizeof(unsigned long)) {
+			/* It's OK. We know it's properly aligned */
+			unsigned long *datum = ebuf + i;
+			if (*datum + 1) {
+				*bad_offset += i;
+				printk(KERN_WARNING "Newly-erased block contained word 0x%lx at offset 0x%08x\n", *datum, *bad_offset);
+				goto fail;
+			}
+		}
+		ofs += readlen;
+		cond_resched();
+	}
+	ret = 0;
+fail:
+	kfree(ebuf);
+	return ret;
+}
+
 static void jffs2_mark_erased_block(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb)
 {
 	struct jffs2_raw_node_ref *marker_ref = NULL;
-	unsigned char *ebuf;
 	size_t retlen;
 	int ret;
 	uint32_t bad_offset;
 
-	if ((!jffs2_cleanmarker_oob(c)) && (c->cleanmarker_size > 0)) {
+	switch (jffs2_block_check_erase(c, jeb, &bad_offset)) {
-		marker_ref = jffs2_alloc_raw_node_ref();
+	case -EAGAIN:	goto refile;
-		if (!marker_ref) {
+	case -EIO:	goto filebad;
-			printk(KERN_WARNING "Failed to allocate raw node ref for clean marker\n");
-			/* Stick it back on the list from whence it came and come back later */
-			jffs2_erase_pending_trigger(c);
-			spin_lock(&c->erase_completion_lock);
-			list_add(&jeb->list, &c->erase_complete_list);
-			spin_unlock(&c->erase_completion_lock);
-			return;
-		}
 	}
-	ebuf = kmalloc(PAGE_SIZE, GFP_KERNEL);
-	if (!ebuf) {
-		printk(KERN_WARNING "Failed to allocate page buffer for verifying erase at 0x%08x. Assuming it worked\n", jeb->offset);
-	} else {
-		uint32_t ofs = jeb->offset;
-
-		D1(printk(KERN_DEBUG "Verifying erase at 0x%08x\n", jeb->offset));
-		while(ofs < jeb->offset + c->sector_size) {
-			uint32_t readlen = min((uint32_t)PAGE_SIZE, jeb->offset + c->sector_size - ofs);
-			int i;
-
-			bad_offset = ofs;
-
-			ret = c->mtd->read(c->mtd, ofs, readlen, &retlen, ebuf);
-
-			if (ret) {
-				printk(KERN_WARNING "Read of newly-erased block at 0x%08x failed: %d. Putting on bad_list\n", ofs, ret);
-				goto bad;
-			}
-			if (retlen != readlen) {
-				printk(KERN_WARNING "Short read from newly-erased block at 0x%08x. Wanted %d, got %zd\n", ofs, readlen, retlen);
-				goto bad;
-			}
-			for (i=0; i<readlen; i += sizeof(unsigned long)) {
-				/* It's OK. We know it's properly aligned */
-				unsigned long datum = *(unsigned long *)(&ebuf[i]);
-				if (datum + 1) {
-					bad_offset += i;
-					printk(KERN_WARNING "Newly-erased block contained word 0x%lx at offset 0x%08x\n", datum, bad_offset);
-				bad: 
-					if ((!jffs2_cleanmarker_oob(c)) && (c->cleanmarker_size > 0))
-						jffs2_free_raw_node_ref(marker_ref);
-					kfree(ebuf);
-				bad2:
-					spin_lock(&c->erase_completion_lock);
-					/* Stick it on a list (any list) so
-					   erase_failed can take it right off
-					   again.  Silly, but shouldn't happen
-					   often. */
-					list_add(&jeb->list, &c->erasing_list);
-					spin_unlock(&c->erase_completion_lock);
-					jffs2_erase_failed(c, jeb, bad_offset);
-					return;
-				}
-			}
-			ofs += readlen;
-			cond_resched();
-		}
-		kfree(ebuf);
-	}
-
-	bad_offset = jeb->offset;
 
 	/* Write the erase complete marker */	
 	D1(printk(KERN_DEBUG "Writing erased marker to block at 0x%08x\n", jeb->offset));
-	if (jffs2_cleanmarker_oob(c)) {
+	bad_offset = jeb->offset;
 
-		if (jffs2_write_nand_cleanmarker(c, jeb))
+	/* Cleanmarker in oob area or no cleanmarker at all ? */
-			goto bad2;
+	if (jffs2_cleanmarker_oob(c) || c->cleanmarker_size == 0) {
+
+		if (jffs2_cleanmarker_oob(c)) {
+			if (jffs2_write_nand_cleanmarker(c, jeb))
+				goto filebad;
+		}
 
 		jeb->first_node = jeb->last_node = NULL;
-
 		jeb->free_size = c->sector_size;
 		jeb->used_size = 0;
 		jeb->dirty_size = 0;
 		jeb->wasted_size = 0;
-	} else if (c->cleanmarker_size == 0) {
-		jeb->first_node = jeb->last_node = NULL;
 
-		jeb->free_size = c->sector_size;
-		jeb->used_size = 0;
-		jeb->dirty_size = 0;
-		jeb->wasted_size = 0;
 	} else {
+
 		struct kvec vecs[1];
 		struct jffs2_unknown_node marker = {
 			.magic =	cpu_to_je16(JFFS2_MAGIC_BITMASK),
@@ -401,21 +387,28 @@ static void jffs2_mark_erased_block(struct jffs2_sb_info *c, struct jffs2_eraseb
 			.totlen =	cpu_to_je32(c->cleanmarker_size)
 		};
 
+		marker_ref = jffs2_alloc_raw_node_ref();
+		if (!marker_ref) {
+			printk(KERN_WARNING "Failed to allocate raw node ref for clean marker. Refiling\n");
+			goto refile;
+		}
+
 		marker.hdr_crc = cpu_to_je32(crc32(0, &marker, sizeof(struct jffs2_unknown_node)-4));
 
 		vecs[0].iov_base = (unsigned char *) &marker;
 		vecs[0].iov_len = sizeof(marker);
 		ret = jffs2_flash_direct_writev(c, vecs, 1, jeb->offset, &retlen);
 		
-		if (ret) {
+		if (ret || retlen != sizeof(marker)) {
-			printk(KERN_WARNING "Write clean marker to block at 0x%08x failed: %d\n",
+			if (ret)
-			       jeb->offset, ret);
+				printk(KERN_WARNING "Write clean marker to block at 0x%08x failed: %d\n",
-			goto bad2;
+				       jeb->offset, ret);
-		}
+			else
-		if (retlen != sizeof(marker)) {
+				printk(KERN_WARNING "Short write to newly-erased block at 0x%08x: Wanted %zd, got %zd\n",
-			printk(KERN_WARNING "Short write to newly-erased block at 0x%08x: Wanted %zd, got %zd\n",
+				       jeb->offset, sizeof(marker), retlen);
-			       jeb->offset, sizeof(marker), retlen);
+
-			goto bad2;
+			jffs2_free_raw_node_ref(marker_ref);
+			goto filebad;
 		}
 
 		marker_ref->next_in_ino = NULL;
@@ -444,5 +437,22 @@ static void jffs2_mark_erased_block(struct jffs2_sb_info *c, struct jffs2_eraseb
 	c->nr_free_blocks++;
 	spin_unlock(&c->erase_completion_lock);
 	wake_up(&c->erase_wait);
-}
+	return;
 
+filebad:
+	spin_lock(&c->erase_completion_lock);
+	/* Stick it on a list (any list) so erase_failed can take it
+	   right off again.  Silly, but shouldn't happen often. */
+	list_add(&jeb->list, &c->erasing_list);
+	spin_unlock(&c->erase_completion_lock);
+	jffs2_erase_failed(c, jeb, bad_offset);
+	return;
+
+refile:
+	/* Stick it back on the list from whence it came and come back later */
+	jffs2_erase_pending_trigger(c);
+	spin_lock(&c->erase_completion_lock);
+	list_add(&jeb->list, &c->erase_complete_list);
+	spin_unlock(&c->erase_completion_lock);
+	return;
+}

fs/ntfs/ChangeLog

@@ -1,21 +1,18 @@
 ToDo/Notes:
 	- Find and fix bugs.
-	- Checkpoint or disable the user space journal ($UsnJrnl).
 	- In between ntfs_prepare/commit_write, need exclusion between
-	  simultaneous file extensions. Need perhaps an NInoResizeUnderway()
+	  simultaneous file extensions.  This is given to us by holding i_sem
-	  flag which we can set in ntfs_prepare_write() and clear again in
+	  on the inode.  The only places in the kernel when a file is resized
-	  ntfs_commit_write(). Just have to be careful in readpage/writepage,
+	  are prepare/commit write and truncate for both of which i_sem is
-	  as well as in truncate, that we play nice... We might need to have
+	  held.  Just have to be careful in readpage/writepage and all other
-	  a data_size field in the ntfs_inode to store the real attribute
+	  helpers not running under i_sem that we play nice...
-	  length. Also need to be careful with initialized_size extention in
+	  Also need to be careful with initialized_size extention in
 	  ntfs_prepare_write. Basically, just be _very_ careful in this code...
-	  OTOH, perhaps i_sem, which is held accross generic_file_write is
+	  UPDATE: The only things that need to be checked are read/writepage
-	  sufficient for synchronisation here. We then just need to make sure
+	  which do not hold i_sem.  Note writepage cannot change i_size but it
-	  ntfs_readpage/writepage/truncate interoperate properly with us.
+	  needs to cope with a concurrent i_size change, just like readpage.
-	  UPDATE: The above is all ok as it is due to i_sem held.  The only
+	  Also both need to cope with concurrent changes to the other sizes,
-	  thing that needs to be checked is ntfs_writepage() which does not
+	  i.e. initialized/allocated/compressed size, as well.
-	  hold i_sem.  It cannot change i_size but it needs to cope with a
-	  concurrent i_size change.
 	- Implement mft.c::sync_mft_mirror_umount().  We currently will just
 	  leave the volume dirty on umount if the final iput(vol->mft_ino)
 	  causes a write of any mirrored mft records due to the mft mirror
@@ -25,12 +22,158 @@ ToDo/Notes:
 	- Enable the code for setting the NT4 compatibility flag when we start
 	  making NTFS 1.2 specific modifications.
 
-2.1.23-WIP
+2.1.23 - Implement extension of resident files and make writing safe as well as
+	 many bug fixes, cleanups, and enhancements...
 
 	- Add printk rate limiting for ntfs_warning() and ntfs_error() when
 	  compiled without debug.  This avoids a possible denial of service
 	  attack.  Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
 	  out.
+	- Fix compilation warnings on ia64.  (Randy Dunlap)
+	- Use i_size_{read,write}() instead of reading i_size by hand and cache
+	  the value where apropriate.
+	- Add size_lock to the ntfs_inode structure.  This is an rw spinlock
+	  and it locks against access to the inode sizes.  Note, ->size_lock
+	  is also accessed from irq context so you must use the _irqsave and
+	  _irqrestore lock and unlock functions, respectively.  Protect all
+	  accesses to allocated_size, initialized_size, and compressed_size.
+	- Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
+	- Implement extension of resident files in the regular file write code
+	  paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()).  At present
+	  this only works until the data attribute becomes too big for the mft
+	  record after which we abort the write returning -EOPNOTSUPP from
+	  ntfs_prepare_write().
+	- Add disable_sparse mount option together with a per volume sparse
+	  enable bit which is set appropriately and a per inode sparse disable
+	  bit which is preset on some system file inodes as appropriate.
+	- Enforce that sparse support is disabled on NTFS volumes pre 3.0.
+	- Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
+	  the creation of the unmapped runlist element for the base attribute
+	  extent.
+	- Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
+	  helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
+	  This allows us to map runlist fragments with the runlist lock already
+	  held without having to drop and reacquire it around the call.  Adapt
+	  all callers.
+	- Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
+	  runlist.  This allows us to find runlist elements with the runlist
+	  lock already held without having to drop and reacquire it around the
+	  call.  Adapt all callers.
+	- Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
+	  warning in the do_div() call on sparc32.  Thanks to Meelis Roos for
+	  the report and analysis of the warning.
+	- Fix a nasty runlist merge bug when merging two holes.
+	- Set the ntfs_inode->allocated_size to the real allocated size in the
+	  mft record for resident attributes (fs/ntfs/inode.c).
+	- Small readability cleanup to use "a" instead of "ctx->attr"
+	  everywhere (fs/ntfs/inode.c).
+	- Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
+	  definition of ntfs_export_ops from fs/ntfs/super.c to namei.c.  Also,
+	  declare ntfs_export_ops in fs/ntfs/ntfs.h.
+	- Correct sparse file handling.  The compressed values need to be
+	  checked and set in the ntfs inode as done for compressed files and
+	  the compressed size needs to be used for vfs inode->i_blocks instead
+	  of the allocated size, again, as done for compressed files.
+	- Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
+	  non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
+	- Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
+	  write code.
+	- Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
+	  dropping the read lock and taking the write lock we were not checking
+	  whether someone else did not already do the work we wanted to do.
+	- Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
+	  ntfs_attr_find_vcn_nolock() and update all callers.
+	- Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
+	- Fix sign of various error return values to be negative in
+	  fs/ntfs/lcnalloc.c.
+	- Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
+	  handle the case where an attribute is converted from resident to
+	  non-resident by a concurrent file write.
+	- Remove checks for NULL before calling kfree() since kfree() does the
+	  checking itself.  (Jesper Juhl)
+	- Some utilities modify the boot sector but do not update the checksum.
+	  Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
+	  only emit a warning when the checksum is incorrect rather than
+	  refusing the mount.  Thanks to Bernd Casimir for pointing this
+	  problem out.
+	- Update attribute definition handling.
+	- Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
+	- Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
+	- Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
+	  better code generation and one less sparse warning in fs/ntfs/aops.c.
+	- Remove spurious void pointer casts from fs/ntfs/.  (Pekka Enberg)
+	- Use C99 style structure initialization after memory allocation where
+	  possible (fs/ntfs/{attrib.c,index.c,super.c}).  Thanks to Al Viro and
+	  Pekka Enberg.
+	- Stamp the transaction log ($UsnJrnl), aka user space journal, if it
+	  is active on the volume and we are mounting read-write or remounting
+	  from read-only to read-write.
+	- Fix a bug in address space operations error recovery code paths where
+	  if the runlist was not mapped at all and a mapping error occured we
+	  would leave the runlist locked on exit to the function so that the
+	  next access to the same file would try to take the lock and deadlock.
+	- Detect the case when Windows has been suspended to disk on the volume
+	  to be mounted and if this is the case do not allow (re)mounting
+	  read-write.  This is done by parsing hiberfil.sys if present.
+	- Fix several occurences of a bug where we would perform 'var & ~const'
+	  with a 64-bit variable and a int, i.e. 32-bit, constant.  This causes
+	  the higher order 32-bits of the 64-bit variable to be zeroed.  To fix
+	  this cast the 'const' to the same 64-bit type as 'var'.
+	- Change the runlist terminator of the newly allocated cluster(s) to
+	  LCN_ENOENT in ntfs_attr_make_non_resident().  Otherwise the runlist
+	  code gets confused.
+	- Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
+	  and ntfs_mapping_pairs_build() to allow the runlist encoding to be
+	  partial which is desirable when filling holes in sparse attributes.
+	  Update all callers.
+	- Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
+	  if the requested vcn is inside it.  Otherwise we get into problems
+	  when we try to map an out of bounds vcn because we then try to map
+	  the already mapped runlist fragment which causes
+	  ntfs_mapping_pairs_decompress() to fail and return error.  Update
+	  ntfs_attr_find_vcn_nolock() accordingly.
+	- Fix a nasty deadlock that appeared in recent kernels.
+	  The situation: VFS inode X on a mounted ntfs volume is dirty.  For
+	  same inode X, the ntfs_inode is dirty and thus corresponding on-disk
+	  inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
+	  to the table of inodes, i.e. $MFT, inode 0.
+	  What happens:
+	  Process 1: sys_sync()/umount()/whatever...  calls
+	  __sync_single_inode() for $MFT -> do_writepages() -> write_page for
+	  the dirty page containing the on-disk inode X, the page is now locked
+	  -> ntfs_write_mst_block() which clears PageUptodate() on the page to
+	  prevent anyone else getting hold of it whilst it does the write out.
+	  This is necessary as the on-disk inode needs "fixups" applied before
+	  the write to disk which are removed again after the write and
+	  PageUptodate is then set again.  It then analyses the page looking
+	  for dirty on-disk inodes and when it finds one it calls
+	  ntfs_may_write_mft_record() to see if it is safe to write this
+	  on-disk inode.  This then calls ilookup5() to check if the
+	  corresponding VFS inode is in icache().  This in turn calls ifind()
+	  which waits on the inode lock via wait_on_inode whilst holding the
+	  global inode_lock.
+	  Process 2: pdflush results in a call to __sync_single_inode for the
+	  same VFS inode X on the ntfs volume.  This locks the inode (I_LOCK)
+	  then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
+	  read_cache_page() for the page (in page cache of table of inodes
+	  $MFT, inode 0) containing the on-disk inode.  This page has
+	  PageUptodate() clear because of Process 1 (see above) so
+	  read_cache_page() blocks when it tries to take the page lock for the
+	  page so it can call ntfs_read_page().
+	  Thus Process 1 is holding the page lock on the page containing the
+	  on-disk inode X and it is waiting on the inode X to be unlocked in
+	  ifind() so it can write the page out and then unlock the page.
+	  And Process 2 is holding the inode lock on inode X and is waiting for
+	  the page to be unlocked so it can call ntfs_readpage() or discover
+	  that Process 1 set PageUptodate() again and use the page.
+	  Thus we have a deadlock due to ifind() waiting on the inode lock.
+	  The solution: The fix is to use the newly introduced
+	  ilookup5_nowait() which does not wait on the inode's lock and hence
+	  avoids the deadlock.  This is safe as we do not care about the VFS
+	  inode and only use the fact that it is in the VFS inode cache and the
+	  fact that the vfs and ntfs inodes are one struct in memory to find
+	  the ntfs inode in memory if present.  Also, the ntfs inode has its
+	  own locking so it does not matter if the vfs inode is locked.
 
 2.1.22 - Many bug and race fixes and error handling improvements.
 
@@ -1037,7 +1180,7 @@ tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
 	- Further runlist merging work. (Richard Russon)
 	- Backwards compatibility for gcc-2.95. (Richard Russon)
 	- Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
-	- Convert to new file system declaration using ->ntfs_get_sb() and
+	- Convert to new filesystem declaration using ->ntfs_get_sb() and
 	  replacing ntfs_read_super() with ntfs_fill_super().
 	- Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
 	  overflow on 32-bit architectures.
@@ -1333,7 +1476,7 @@ tng-0.0.1 - The first useful version.
 	The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
 	though and it doesn't implement accesssing compressed files yet. Also,
 	accessing files with attribute list attributes is not implemented yet
-	either. But for small or simple file systems it should work and allow
+	either. But for small or simple filesystems it should work and allow
 	you to list directories, use stat on directory entries and the file
 	system, open, read, mmap and llseek around in files. A big mile stone
 	has been reached!
@@ -1341,7 +1484,7 @@ tng-0.0.1 - The first useful version.
 tng-0.0.0 - Initial version tag.
 
 	Initial driver implementation. The driver can mount and umount simple
-	NTFS file systems (i.e. ones without attribute lists in the system
+	NTFS filesystems (i.e. ones without attribute lists in the system
 	files). If the mount fails there might be problems in the error handling
 	code paths, so be warned. Otherwise it seems to be loading the system
 	files nicely and the mft record read mapping/unmapping seems to be

fs/ntfs/Makefile

@@ -6,7 +6,7 @@ ntfs-objs := aops.o attrib.o collate.o compress.o debug.o dir.o file.o \
 	     index.o inode.o mft.o mst.o namei.o runlist.o super.o sysctl.o \
 	     unistr.o upcase.o
 
-EXTRA_CFLAGS = -DNTFS_VERSION=\"2.1.22\"
+EXTRA_CFLAGS = -DNTFS_VERSION=\"2.1.23\"
 
 ifeq ($(CONFIG_NTFS_DEBUG),y)
 EXTRA_CFLAGS += -DDEBUG
@@ -15,5 +15,5 @@ endif
 ifeq ($(CONFIG_NTFS_RW),y)
 EXTRA_CFLAGS += -DNTFS_RW
 
-ntfs-objs += bitmap.o lcnalloc.o logfile.o quota.o
+ntfs-objs += bitmap.o lcnalloc.o logfile.o quota.o usnjrnl.o
 endif

fs/ntfs/aops.c

@@ -2,7 +2,7 @@
  * aops.c - NTFS kernel address space operations and page cache handling.
  *	    Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -66,19 +66,22 @@ static void ntfs_end_buffer_async_read(struct buffer_head *bh, int uptodate)
 	ni = NTFS_I(page->mapping->host);
 
 	if (likely(uptodate)) {
-		s64 file_ofs;
+		s64 file_ofs, initialized_size;
 
 		set_buffer_uptodate(bh);
 
 		file_ofs = ((s64)page->index << PAGE_CACHE_SHIFT) +
 				bh_offset(bh);
+		read_lock_irqsave(&ni->size_lock, flags);
+		initialized_size = ni->initialized_size;
+		read_unlock_irqrestore(&ni->size_lock, flags);
 		/* Check for the current buffer head overflowing. */
-		if (file_ofs + bh->b_size > ni->initialized_size) {
+		if (file_ofs + bh->b_size > initialized_size) {
 			char *addr;
 			int ofs = 0;
 
-			if (file_ofs < ni->initialized_size)
+			if (file_ofs < initialized_size)
-				ofs = ni->initialized_size - file_ofs;
+				ofs = initialized_size - file_ofs;
 			addr = kmap_atomic(page, KM_BIO_SRC_IRQ);
 			memset(addr + bh_offset(bh) + ofs, 0, bh->b_size - ofs);
 			flush_dcache_page(page);
@@ -132,7 +135,7 @@ static void ntfs_end_buffer_async_read(struct buffer_head *bh, int uptodate)
 					i * rec_size), rec_size);
 		flush_dcache_page(page);
 		kunmap_atomic(addr, KM_BIO_SRC_IRQ);
-		if (likely(!PageError(page) && page_uptodate))
+		if (likely(page_uptodate && !PageError(page)))
 			SetPageUptodate(page);
 	}
 	unlock_page(page);
@@ -168,6 +171,7 @@ static int ntfs_read_block(struct page *page)
 	runlist_element *rl;
 	struct buffer_head *bh, *head, *arr[MAX_BUF_PER_PAGE];
 	sector_t iblock, lblock, zblock;
+	unsigned long flags;
 	unsigned int blocksize, vcn_ofs;
 	int i, nr;
 	unsigned char blocksize_bits;
@@ -190,8 +194,10 @@ static int ntfs_read_block(struct page *page)
 	}
 
 	iblock = (s64)page->index << (PAGE_CACHE_SHIFT - blocksize_bits);
+	read_lock_irqsave(&ni->size_lock, flags);
 	lblock = (ni->allocated_size + blocksize - 1) >> blocksize_bits;
 	zblock = (ni->initialized_size + blocksize - 1) >> blocksize_bits;
+	read_unlock_irqrestore(&ni->size_lock, flags);
 
 	/* Loop through all the buffers in the page. */
 	rl = NULL;
@@ -258,7 +264,8 @@ static int ntfs_read_block(struct page *page)
 					goto lock_retry_remap;
 				rl = NULL;
 				lcn = err;
-			}
+			} else if (!rl)
+				up_read(&ni->runlist.lock);
 			/* Hard error, zero out region. */
 			bh->b_blocknr = -1;
 			SetPageError(page);
@@ -341,14 +348,15 @@ static int ntfs_read_block(struct page *page)
  */
 static int ntfs_readpage(struct file *file, struct page *page)
 {
-	loff_t i_size;
 	ntfs_inode *ni, *base_ni;
 	u8 *kaddr;
 	ntfs_attr_search_ctx *ctx;
 	MFT_RECORD *mrec;
+	unsigned long flags;
 	u32 attr_len;
 	int err = 0;
 
+retry_readpage:
 	BUG_ON(!PageLocked(page));
 	/*
 	 * This can potentially happen because we clear PageUptodate() during
@@ -383,9 +391,9 @@ static int ntfs_readpage(struct file *file, struct page *page)
 	 * Attribute is resident, implying it is not compressed or encrypted.
 	 * This also means the attribute is smaller than an mft record and
 	 * hence smaller than a page, so can simply zero out any pages with
-	 * index above 0.  We can also do this if the file size is 0.
+	 * index above 0.
 	 */
-	if (unlikely(page->index > 0 || !i_size_read(VFS_I(ni)))) {
+	if (unlikely(page->index > 0)) {
 		kaddr = kmap_atomic(page, KM_USER0);
 		memset(kaddr, 0, PAGE_CACHE_SIZE);
 		flush_dcache_page(page);
@@ -402,6 +410,14 @@ static int ntfs_readpage(struct file *file, struct page *page)
 		err = PTR_ERR(mrec);
 		goto err_out;
 	}
+	/*
+	 * If a parallel write made the attribute non-resident, drop the mft
+	 * record and retry the readpage.
+	 */
+	if (unlikely(NInoNonResident(ni))) {
+		unmap_mft_record(base_ni);
+		goto retry_readpage;
+	}
 	ctx = ntfs_attr_get_search_ctx(base_ni, mrec);
 	if (unlikely(!ctx)) {
 		err = -ENOMEM;
@@ -412,9 +428,10 @@ static int ntfs_readpage(struct file *file, struct page *page)
 	if (unlikely(err))
 		goto put_unm_err_out;
 	attr_len = le32_to_cpu(ctx->attr->data.resident.value_length);
-	i_size = i_size_read(VFS_I(ni));
+	read_lock_irqsave(&ni->size_lock, flags);
-	if (unlikely(attr_len > i_size))
+	if (unlikely(attr_len > ni->initialized_size))
-		attr_len = i_size;
+		attr_len = ni->initialized_size;
+	read_unlock_irqrestore(&ni->size_lock, flags);
 	kaddr = kmap_atomic(page, KM_USER0);
 	/* Copy the data to the page. */
 	memcpy(kaddr, (u8*)ctx->attr +
@@ -463,12 +480,15 @@ static int ntfs_write_block(struct page *page, struct writeback_control *wbc)
 {
 	VCN vcn;
 	LCN lcn;
+	s64 initialized_size;
+	loff_t i_size;
 	sector_t block, dblock, iblock;
 	struct inode *vi;
 	ntfs_inode *ni;
 	ntfs_volume *vol;
 	runlist_element *rl;
 	struct buffer_head *bh, *head;
+	unsigned long flags;
 	unsigned int blocksize, vcn_ofs;
 	int err;
 	BOOL need_end_writeback;
@@ -510,11 +530,16 @@ static int ntfs_write_block(struct page *page, struct writeback_control *wbc)
 	/* The first block in the page. */
 	block = (s64)page->index << (PAGE_CACHE_SHIFT - blocksize_bits);
 
+	read_lock_irqsave(&ni->size_lock, flags);
+	i_size = i_size_read(vi);
+	initialized_size = ni->initialized_size;
+	read_unlock_irqrestore(&ni->size_lock, flags);
+
 	/* The first out of bounds block for the data size. */
-	dblock = (vi->i_size + blocksize - 1) >> blocksize_bits;
+	dblock = (i_size + blocksize - 1) >> blocksize_bits;
 
 	/* The last (fully or partially) initialized block. */
-	iblock = ni->initialized_size >> blocksize_bits;
+	iblock = initialized_size >> blocksize_bits;
 
 	/*
 	 * Be very careful.  We have no exclusion from __set_page_dirty_buffers
@@ -559,7 +584,7 @@ static int ntfs_write_block(struct page *page, struct writeback_control *wbc)
 
 		/* Make sure we have enough initialized size. */
 		if (unlikely((block >= iblock) &&
-				(ni->initialized_size < vi->i_size))) {
+				(initialized_size < i_size))) {
 			/*
 			 * If this page is fully outside initialized size, zero
 			 * out all pages between the current initialized size
@@ -666,7 +691,8 @@ static int ntfs_write_block(struct page *page, struct writeback_control *wbc)
 				goto lock_retry_remap;
 			rl = NULL;
 			lcn = err;
-		}
+		} else if (!rl)
+			up_read(&ni->runlist.lock);
 		/* Failed to map the buffer, even after retrying. */
 		bh->b_blocknr = -1;
 		ntfs_error(vol->sb, "Failed to write to inode 0x%lx, "
@@ -801,17 +827,15 @@ static int ntfs_write_mst_block(struct page *page,
 	ntfs_inode *ni = NTFS_I(vi);
 	ntfs_volume *vol = ni->vol;
 	u8 *kaddr;
-	unsigned char bh_size_bits = vi->i_blkbits;
-	unsigned int bh_size = 1 << bh_size_bits;
 	unsigned int rec_size = ni->itype.index.block_size;
 	ntfs_inode *locked_nis[PAGE_CACHE_SIZE / rec_size];
 	struct buffer_head *bh, *head, *tbh, *rec_start_bh;
-	int max_bhs = PAGE_CACHE_SIZE / bh_size;
+	struct buffer_head *bhs[MAX_BUF_PER_PAGE];
-	struct buffer_head *bhs[max_bhs];
 	runlist_element *rl;
-	int i, nr_locked_nis, nr_recs, nr_bhs, bhs_per_rec, err, err2;
+	int i, nr_locked_nis, nr_recs, nr_bhs, max_bhs, bhs_per_rec, err, err2;
-	unsigned rec_size_bits;
+	unsigned bh_size, rec_size_bits;
 	BOOL sync, is_mft, page_is_dirty, rec_is_dirty;
+	unsigned char bh_size_bits;
 
 	ntfs_debug("Entering for inode 0x%lx, attribute type 0x%x, page index "
 			"0x%lx.", vi->i_ino, ni->type, page->index);
@@ -826,7 +850,11 @@ static int ntfs_write_mst_block(struct page *page,
 	 */
 	BUG_ON(!(is_mft || S_ISDIR(vi->i_mode) ||
 			(NInoAttr(ni) && ni->type == AT_INDEX_ALLOCATION)));
+	bh_size_bits = vi->i_blkbits;
+	bh_size = 1 << bh_size_bits;
+	max_bhs = PAGE_CACHE_SIZE / bh_size;
 	BUG_ON(!max_bhs);
+	BUG_ON(max_bhs > MAX_BUF_PER_PAGE);
 
 	/* Were we called for sync purposes? */
 	sync = (wbc->sync_mode == WB_SYNC_ALL);
@@ -846,7 +874,7 @@ static int ntfs_write_mst_block(struct page *page,
 			(PAGE_CACHE_SHIFT - bh_size_bits);
 
 	/* The first out of bounds block for the data size. */
-	dblock = (vi->i_size + bh_size - 1) >> bh_size_bits;
+	dblock = (i_size_read(vi) + bh_size - 1) >> bh_size_bits;
 
 	rl = NULL;
 	err = err2 = nr_bhs = nr_recs = nr_locked_nis = 0;
@@ -858,6 +886,7 @@ static int ntfs_write_mst_block(struct page *page,
 		if (likely(block < rec_block)) {
 			if (unlikely(block >= dblock)) {
 				clear_buffer_dirty(bh);
+				set_buffer_uptodate(bh);
 				continue;
 			}
 			/*
@@ -938,8 +967,11 @@ static int ntfs_write_mst_block(struct page *page,
 					if (err2 == -ENOMEM)
 						page_is_dirty = TRUE;
 					lcn = err2;
-				} else
+				} else {
 					err2 = -EIO;
+					if (!rl)
+						up_read(&ni->runlist.lock);
+				}
 				/* Hard error.  Abort writing this record. */
 				if (!err || err == -ENOMEM)
 					err = err2;
@@ -949,7 +981,8 @@ static int ntfs_write_mst_block(struct page *page,
 						"attribute type 0x%x) because "
 						"its location on disk could "
 						"not be determined (error "
-						"code %lli).", (s64)block <<
+						"code %lli).",
+						(long long)block <<
 						bh_size_bits >>
 						vol->mft_record_size_bits,
 						ni->mft_no, ni->type,
@@ -1223,19 +1256,17 @@ static int ntfs_write_mst_block(struct page *page,
 static int ntfs_writepage(struct page *page, struct writeback_control *wbc)
 {
 	loff_t i_size;
-	struct inode *vi;
+	struct inode *vi = page->mapping->host;
-	ntfs_inode *ni, *base_ni;
+	ntfs_inode *base_ni = NULL, *ni = NTFS_I(vi);
 	char *kaddr;
-	ntfs_attr_search_ctx *ctx;
+	ntfs_attr_search_ctx *ctx = NULL;
-	MFT_RECORD *m;
+	MFT_RECORD *m = NULL;
 	u32 attr_len;
 	int err;
 
+retry_writepage:
 	BUG_ON(!PageLocked(page));
-
-	vi = page->mapping->host;
 	i_size = i_size_read(vi);
-
 	/* Is the page fully outside i_size? (truncate in progress) */
 	if (unlikely(page->index >= (i_size + PAGE_CACHE_SIZE - 1) >>
 			PAGE_CACHE_SHIFT)) {
@@ -1248,8 +1279,6 @@ static int ntfs_writepage(struct page *page, struct writeback_control *wbc)
 		ntfs_debug("Write outside i_size - truncated?");
 		return 0;
 	}
-	ni = NTFS_I(vi);
-
 	/* NInoNonResident() == NInoIndexAllocPresent() */
 	if (NInoNonResident(ni)) {
 		/*
@@ -1326,6 +1355,14 @@ static int ntfs_writepage(struct page *page, struct writeback_control *wbc)
 		ctx = NULL;
 		goto err_out;
 	}
+	/*
+	 * If a parallel write made the attribute non-resident, drop the mft
+	 * record and retry the writepage.
+	 */
+	if (unlikely(NInoNonResident(ni))) {
+		unmap_mft_record(base_ni);
+		goto retry_writepage;
+	}
 	ctx = ntfs_attr_get_search_ctx(base_ni, m);
 	if (unlikely(!ctx)) {
 		err = -ENOMEM;
@@ -1367,15 +1404,12 @@ static int ntfs_writepage(struct page *page, struct writeback_control *wbc)
 	 */
 
 	attr_len = le32_to_cpu(ctx->attr->data.resident.value_length);
-	i_size = i_size_read(VFS_I(ni));
+	i_size = i_size_read(vi);
-	kaddr = kmap_atomic(page, KM_USER0);
 	if (unlikely(attr_len > i_size)) {
-		/* Zero out of bounds area in the mft record. */
-		memset((u8*)ctx->attr + le16_to_cpu(
-				ctx->attr->data.resident.value_offset) +
-				i_size, 0, attr_len - i_size);
 		attr_len = i_size;
+		ctx->attr->data.resident.value_length = cpu_to_le32(attr_len);
 	}
+	kaddr = kmap_atomic(page, KM_USER0);
 	/* Copy the data from the page to the mft record. */
 	memcpy((u8*)ctx->attr +
 			le16_to_cpu(ctx->attr->data.resident.value_offset),
@@ -1405,8 +1439,10 @@ static int ntfs_writepage(struct page *page, struct writeback_control *wbc)
 		err = 0;
 	} else {
 		ntfs_error(vi->i_sb, "Resident attribute write failed with "
-				"error %i.  Setting page error flag.", err);
+				"error %i.", err);
 		SetPageError(page);
+		NVolSetErrors(ni->vol);
+		make_bad_inode(vi);
 	}
 	unlock_page(page);
 	if (ctx)
@@ -1425,12 +1461,15 @@ static int ntfs_prepare_nonresident_write(struct page *page,
 {
 	VCN vcn;
 	LCN lcn;
+	s64 initialized_size;
+	loff_t i_size;
 	sector_t block, ablock, iblock;
 	struct inode *vi;
 	ntfs_inode *ni;
 	ntfs_volume *vol;
 	runlist_element *rl;
 	struct buffer_head *bh, *head, *wait[2], **wait_bh = wait;
+	unsigned long flags;
 	unsigned int vcn_ofs, block_start, block_end, blocksize;
 	int err;
 	BOOL is_retry;
@@ -1462,16 +1501,20 @@ static int ntfs_prepare_nonresident_write(struct page *page,
 	/* The first block in the page. */
 	block = (s64)page->index << (PAGE_CACHE_SHIFT - blocksize_bits);
 
+	read_lock_irqsave(&ni->size_lock, flags);
 	/*
-	 * The first out of bounds block for the allocated size. No need to
+	 * The first out of bounds block for the allocated size.  No need to
 	 * round up as allocated_size is in multiples of cluster size and the
 	 * minimum cluster size is 512 bytes, which is equal to the smallest
 	 * blocksize.
 	 */
 	ablock = ni->allocated_size >> blocksize_bits;
+	i_size = i_size_read(vi);
+	initialized_size = ni->initialized_size;
+	read_unlock_irqrestore(&ni->size_lock, flags);
 
 	/* The last (fully or partially) initialized block. */
-	iblock = ni->initialized_size >> blocksize_bits;
+	iblock = initialized_size >> blocksize_bits;
 
 	/* Loop through all the buffers in the page. */
 	block_start = 0;
@@ -1518,7 +1561,7 @@ static int ntfs_prepare_nonresident_write(struct page *page,
 		 * request, i.e. block < ablock is true.
 		 */
 		if (unlikely((block >= iblock) &&
-				(ni->initialized_size < vi->i_size))) {
+				(initialized_size < i_size))) {
 			/*
 			 * If this page is fully outside initialized size, zero
 			 * out all pages between the current initialized size
@@ -1622,6 +1665,8 @@ static int ntfs_prepare_nonresident_write(struct page *page,
 							"not supported yet. "
 							"Sorry.");
 					err = -EOPNOTSUPP;
+					if (!rl)
+						up_read(&ni->runlist.lock);
 					goto err_out;
 				} else if (!is_retry &&
 						lcn == LCN_RL_NOT_MAPPED) {
@@ -1636,7 +1681,8 @@ static int ntfs_prepare_nonresident_write(struct page *page,
 						goto lock_retry_remap;
 					rl = NULL;
 					lcn = err;
-				}
+				} else if (!rl)
+					up_read(&ni->runlist.lock);
 				/*
 				 * Failed to map the buffer, even after
 				 * retrying.
@@ -1797,6 +1843,7 @@ static int ntfs_prepare_write(struct file *file, struct page *page,
 		unsigned from, unsigned to)
 {
 	s64 new_size;
+	loff_t i_size;
 	struct inode *vi = page->mapping->host;
 	ntfs_inode *base_ni = NULL, *ni = NTFS_I(vi);
 	ntfs_volume *vol = ni->vol;
@@ -1868,14 +1915,8 @@ static int ntfs_prepare_write(struct file *file, struct page *page,
 	BUG_ON(page_has_buffers(page));
 	new_size = ((s64)page->index << PAGE_CACHE_SHIFT) + to;
 	/* If we do not need to resize the attribute allocation we are done. */
-	if (new_size <= vi->i_size)
+	if (new_size <= i_size_read(vi))
 		goto done;
-
-	// FIXME: We abort for now as this code is not safe.
-	ntfs_error(vi->i_sb, "Changing the file size is not supported yet.  "
-			"Sorry.");
-	return -EOPNOTSUPP;
-
 	/* Map, pin, and lock the (base) mft record. */
 	if (!NInoAttr(ni))
 		base_ni = ni;
@@ -1904,7 +1945,15 @@ static int ntfs_prepare_write(struct file *file, struct page *page,
 	a = ctx->attr;
 	/* The total length of the attribute value. */
 	attr_len = le32_to_cpu(a->data.resident.value_length);
-	BUG_ON(vi->i_size != attr_len);
+	/* Fix an eventual previous failure of ntfs_commit_write(). */
+	i_size = i_size_read(vi);
+	if (unlikely(attr_len > i_size)) {
+		attr_len = i_size;
+		a->data.resident.value_length = cpu_to_le32(attr_len);
+	}
+	/* If we do not need to resize the attribute allocation we are done. */
+	if (new_size <= attr_len)
+		goto done_unm;
 	/* Check if new size is allowed in $AttrDef. */
 	err = ntfs_attr_size_bounds_check(vol, ni->type, new_size);
 	if (unlikely(err)) {
@@ -1962,6 +2011,7 @@ static int ntfs_prepare_write(struct file *file, struct page *page,
 	}
 	flush_dcache_mft_record_page(ctx->ntfs_ino);
 	mark_mft_record_dirty(ctx->ntfs_ino);
+done_unm:
 	ntfs_attr_put_search_ctx(ctx);
 	unmap_mft_record(base_ni);
 	/*
@@ -2047,7 +2097,7 @@ static int ntfs_commit_nonresident_write(struct page *page,
 	 * now we know ntfs_prepare_write() would have failed in the write
 	 * exceeds i_size case, so this will never trigger which is fine.
 	 */
-	if (pos > vi->i_size) {
+	if (pos > i_size_read(vi)) {
 		ntfs_error(vi->i_sb, "Writing beyond the existing file size is "
 				"not supported yet.  Sorry.");
 		return -EOPNOTSUPP;
@@ -2183,9 +2233,13 @@ static int ntfs_commit_write(struct file *file, struct page *page,
 	}
 	kunmap_atomic(kaddr, KM_USER0);
 	/* Update i_size if necessary. */
-	if (vi->i_size < attr_len) {
+	if (i_size_read(vi) < attr_len) {
+		unsigned long flags;
+
+		write_lock_irqsave(&ni->size_lock, flags);
 		ni->allocated_size = ni->initialized_size = attr_len;
 		i_size_write(vi, attr_len);
+		write_unlock_irqrestore(&ni->size_lock, flags);
 	}
 	/* Mark the mft record dirty, so it gets written back. */
 	flush_dcache_mft_record_page(ctx->ntfs_ino);

fs/ntfs/attrib.c

@@ -1,7 +1,7 @@
 /**
  * attrib.c - NTFS attribute operations.  Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -21,14 +21,88 @@
  */
 
 #include <linux/buffer_head.h>
+#include <linux/swap.h>
 
 #include "attrib.h"
 #include "debug.h"
 #include "layout.h"
+#include "lcnalloc.h"
+#include "malloc.h"
 #include "mft.h"
 #include "ntfs.h"
 #include "types.h"
 
+/**
+ * ntfs_map_runlist_nolock - map (a part of) a runlist of an ntfs inode
+ * @ni:		ntfs inode for which to map (part of) a runlist
+ * @vcn:	map runlist part containing this vcn
+ *
+ * Map the part of a runlist containing the @vcn of the ntfs inode @ni.
+ *
+ * Return 0 on success and -errno on error.  There is one special error code
+ * which is not an error as such.  This is -ENOENT.  It means that @vcn is out
+ * of bounds of the runlist.
+ *
+ * Locking: - The runlist must be locked for writing.
+ *	    - This function modifies the runlist.
+ */
+int ntfs_map_runlist_nolock(ntfs_inode *ni, VCN vcn)
+{
+	VCN end_vcn;
+	ntfs_inode *base_ni;
+	MFT_RECORD *m;
+	ATTR_RECORD *a;
+	ntfs_attr_search_ctx *ctx;
+	runlist_element *rl;
+	int err = 0;
+
+	ntfs_debug("Mapping runlist part containing vcn 0x%llx.",
+			(unsigned long long)vcn);
+	if (!NInoAttr(ni))
+		base_ni = ni;
+	else
+		base_ni = ni->ext.base_ntfs_ino;
+	m = map_mft_record(base_ni);
+	if (IS_ERR(m))
+		return PTR_ERR(m);
+	ctx = ntfs_attr_get_search_ctx(base_ni, m);
+	if (unlikely(!ctx)) {
+		err = -ENOMEM;
+		goto err_out;
+	}
+	err = ntfs_attr_lookup(ni->type, ni->name, ni->name_len,
+			CASE_SENSITIVE, vcn, NULL, 0, ctx);
+	if (unlikely(err)) {
+		if (err == -ENOENT)
+			err = -EIO;
+		goto err_out;
+	}
+	a = ctx->attr;
+	/*
+	 * Only decompress the mapping pairs if @vcn is inside it.  Otherwise
+	 * we get into problems when we try to map an out of bounds vcn because
+	 * we then try to map the already mapped runlist fragment and
+	 * ntfs_mapping_pairs_decompress() fails.
+	 */
+	end_vcn = sle64_to_cpu(a->data.non_resident.highest_vcn) + 1;
+	if (unlikely(!a->data.non_resident.lowest_vcn && end_vcn <= 1))
+		end_vcn = ni->allocated_size >> ni->vol->cluster_size_bits;
+	if (unlikely(vcn >= end_vcn)) {
+		err = -ENOENT;
+		goto err_out;
+	}
+	rl = ntfs_mapping_pairs_decompress(ni->vol, a, ni->runlist.rl);
+	if (IS_ERR(rl))
+		err = PTR_ERR(rl);
+	else
+		ni->runlist.rl = rl;
+err_out:
+	if (likely(ctx))
+		ntfs_attr_put_search_ctx(ctx);
+	unmap_mft_record(base_ni);
+	return err;
+}
+
 /**
  * ntfs_map_runlist - map (a part of) a runlist of an ntfs inode
  * @ni:		ntfs inode for which to map (part of) a runlist
@@ -36,73 +110,128 @@
  *
  * Map the part of a runlist containing the @vcn of the ntfs inode @ni.
  *
- * Return 0 on success and -errno on error.
+ * Return 0 on success and -errno on error.  There is one special error code
+ * which is not an error as such.  This is -ENOENT.  It means that @vcn is out
+ * of bounds of the runlist.
  *
  * Locking: - The runlist must be unlocked on entry and is unlocked on return.
- *	    - This function takes the lock for writing and modifies the runlist.
+ *	    - This function takes the runlist lock for writing and modifies the
+ *	      runlist.
  */
 int ntfs_map_runlist(ntfs_inode *ni, VCN vcn)
 {
-	ntfs_inode *base_ni;
-	ntfs_attr_search_ctx *ctx;
-	MFT_RECORD *mrec;
 	int err = 0;
 
-	ntfs_debug("Mapping runlist part containing vcn 0x%llx.",
-			(unsigned long long)vcn);
-
-	if (!NInoAttr(ni))
-		base_ni = ni;
-	else
-		base_ni = ni->ext.base_ntfs_ino;
-
-	mrec = map_mft_record(base_ni);
-	if (IS_ERR(mrec))
-		return PTR_ERR(mrec);
-	ctx = ntfs_attr_get_search_ctx(base_ni, mrec);
-	if (unlikely(!ctx)) {
-		err = -ENOMEM;
-		goto err_out;
-	}
-	err = ntfs_attr_lookup(ni->type, ni->name, ni->name_len,
-			CASE_SENSITIVE, vcn, NULL, 0, ctx);
-	if (unlikely(err))
-		goto put_err_out;
-
 	down_write(&ni->runlist.lock);
 	/* Make sure someone else didn't do the work while we were sleeping. */
 	if (likely(ntfs_rl_vcn_to_lcn(ni->runlist.rl, vcn) <=
-			LCN_RL_NOT_MAPPED)) {
+			LCN_RL_NOT_MAPPED))
-		runlist_element *rl;
+		err = ntfs_map_runlist_nolock(ni, vcn);
-
-		rl = ntfs_mapping_pairs_decompress(ni->vol, ctx->attr,
-				ni->runlist.rl);
-		if (IS_ERR(rl))
-			err = PTR_ERR(rl);
-		else
-			ni->runlist.rl = rl;
-	}
 	up_write(&ni->runlist.lock);
-
-put_err_out:
-	ntfs_attr_put_search_ctx(ctx);
-err_out:
-	unmap_mft_record(base_ni);
 	return err;
 }
 
 /**
- * ntfs_find_vcn - find a vcn in the runlist described by an ntfs inode
+ * ntfs_attr_vcn_to_lcn_nolock - convert a vcn into a lcn given an ntfs inode
- * @ni:		ntfs inode describing the runlist to search
+ * @ni:			ntfs inode of the attribute whose runlist to search
- * @vcn:	vcn to find
+ * @vcn:		vcn to convert
- * @need_write:	if false, lock for reading and if true, lock for writing
+ * @write_locked:	true if the runlist is locked for writing
+ *
+ * Find the virtual cluster number @vcn in the runlist of the ntfs attribute
+ * described by the ntfs inode @ni and return the corresponding logical cluster
+ * number (lcn).
+ *
+ * If the @vcn is not mapped yet, the attempt is made to map the attribute
+ * extent containing the @vcn and the vcn to lcn conversion is retried.
+ *
+ * If @write_locked is true the caller has locked the runlist for writing and
+ * if false for reading.
+ *
+ * Since lcns must be >= 0, we use negative return codes with special meaning:
+ *
+ * Return code	Meaning / Description
+ * ==========================================
+ *  LCN_HOLE	Hole / not allocated on disk.
+ *  LCN_ENOENT	There is no such vcn in the runlist, i.e. @vcn is out of bounds.
+ *  LCN_ENOMEM	Not enough memory to map runlist.
+ *  LCN_EIO	Critical error (runlist/file is corrupt, i/o error, etc).
+ *
+ * Locking: - The runlist must be locked on entry and is left locked on return.
+ *	    - If @write_locked is FALSE, i.e. the runlist is locked for reading,
+ *	      the lock may be dropped inside the function so you cannot rely on
+ *	      the runlist still being the same when this function returns.
+ */
+LCN ntfs_attr_vcn_to_lcn_nolock(ntfs_inode *ni, const VCN vcn,
+		const BOOL write_locked)
+{
+	LCN lcn;
+	BOOL is_retry = FALSE;
+
+	ntfs_debug("Entering for i_ino 0x%lx, vcn 0x%llx, %s_locked.",
+			ni->mft_no, (unsigned long long)vcn,
+			write_locked ? "write" : "read");
+	BUG_ON(!ni);
+	BUG_ON(!NInoNonResident(ni));
+	BUG_ON(vcn < 0);
+retry_remap:
+	/* Convert vcn to lcn.  If that fails map the runlist and retry once. */
+	lcn = ntfs_rl_vcn_to_lcn(ni->runlist.rl, vcn);
+	if (likely(lcn >= LCN_HOLE)) {
+		ntfs_debug("Done, lcn 0x%llx.", (long long)lcn);
+		return lcn;
+	}
+	if (lcn != LCN_RL_NOT_MAPPED) {
+		if (lcn != LCN_ENOENT)
+			lcn = LCN_EIO;
+	} else if (!is_retry) {
+		int err;
+
+		if (!write_locked) {
+			up_read(&ni->runlist.lock);
+			down_write(&ni->runlist.lock);
+			if (unlikely(ntfs_rl_vcn_to_lcn(ni->runlist.rl, vcn) !=
+					LCN_RL_NOT_MAPPED)) {
+				up_write(&ni->runlist.lock);
+				down_read(&ni->runlist.lock);
+				goto retry_remap;
+			}
+		}
+		err = ntfs_map_runlist_nolock(ni, vcn);
+		if (!write_locked) {
+			up_write(&ni->runlist.lock);
+			down_read(&ni->runlist.lock);
+		}
+		if (likely(!err)) {
+			is_retry = TRUE;
+			goto retry_remap;
+		}
+		if (err == -ENOENT)
+			lcn = LCN_ENOENT;
+		else if (err == -ENOMEM)
+			lcn = LCN_ENOMEM;
+		else
+			lcn = LCN_EIO;
+	}
+	if (lcn != LCN_ENOENT)
+		ntfs_error(ni->vol->sb, "Failed with error code %lli.",
+				(long long)lcn);
+	return lcn;
+}
+
+/**
+ * ntfs_attr_find_vcn_nolock - find a vcn in the runlist of an ntfs inode
+ * @ni:			ntfs inode describing the runlist to search
+ * @vcn:		vcn to find
+ * @write_locked:	true if the runlist is locked for writing
  *
  * Find the virtual cluster number @vcn in the runlist described by the ntfs
  * inode @ni and return the address of the runlist element containing the @vcn.
- * The runlist is left locked and the caller has to unlock it.  If @need_write
+ *
- * is true, the runlist is locked for writing and if @need_write is false, the
+ * If the @vcn is not mapped yet, the attempt is made to map the attribute
- * runlist is locked for reading.  In the error case, the runlist is not left
+ * extent containing the @vcn and the vcn to lcn conversion is retried.
- * locked.
+ *
+ * If @write_locked is true the caller has locked the runlist for writing and
+ * if false for reading.
  *
  * Note you need to distinguish between the lcn of the returned runlist element
  * being >= 0 and LCN_HOLE.  In the later case you have to return zeroes on
@@ -118,34 +247,29 @@ int ntfs_map_runlist(ntfs_inode *ni, VCN vcn)
  *	-ENOMEM - Not enough memory to map runlist.
  *	-EIO	- Critical error (runlist/file is corrupt, i/o error, etc).
  *
- * Locking: - The runlist must be unlocked on entry.
+ * Locking: - The runlist must be locked on entry and is left locked on return.
- *	    - On failing return, the runlist is unlocked.
+ *	    - If @write_locked is FALSE, i.e. the runlist is locked for reading,
- *	    - On successful return, the runlist is locked.  If @need_write us
+ *	      the lock may be dropped inside the function so you cannot rely on
- *	      true, it is locked for writing.  Otherwise is is locked for
+ *	      the runlist still being the same when this function returns.
- *	      reading.
  */
-runlist_element *ntfs_find_vcn(ntfs_inode *ni, const VCN vcn,
+runlist_element *ntfs_attr_find_vcn_nolock(ntfs_inode *ni, const VCN vcn,
-		const BOOL need_write)
+		const BOOL write_locked)
 {
 	runlist_element *rl;
 	int err = 0;
 	BOOL is_retry = FALSE;
 
-	ntfs_debug("Entering for i_ino 0x%lx, vcn 0x%llx, lock for %sing.",
+	ntfs_debug("Entering for i_ino 0x%lx, vcn 0x%llx, %s_locked.",
 			ni->mft_no, (unsigned long long)vcn,
-			!need_write ? "read" : "writ");
+			write_locked ? "write" : "read");
 	BUG_ON(!ni);
 	BUG_ON(!NInoNonResident(ni));
 	BUG_ON(vcn < 0);
-lock_retry_remap:
+retry_remap:
-	if (!need_write)
-		down_read(&ni->runlist.lock);
-	else
-		down_write(&ni->runlist.lock);
 	rl = ni->runlist.rl;
 	if (likely(rl && vcn >= rl[0].vcn)) {
 		while (likely(rl->length)) {
-			if (likely(vcn < rl[1].vcn)) {
+			if (unlikely(vcn < rl[1].vcn)) {
 				if (likely(rl->lcn >= LCN_HOLE)) {
 					ntfs_debug("Done.");
 					return rl;
@@ -161,30 +285,41 @@ runlist_element *ntfs_find_vcn(ntfs_inode *ni, const VCN vcn,
 				err = -EIO;
 		}
 	}
-	if (!need_write)
-		up_read(&ni->runlist.lock);
-	else
-		up_write(&ni->runlist.lock);
 	if (!err && !is_retry) {
 		/*
 		 * The @vcn is in an unmapped region, map the runlist and
 		 * retry.
 		 */
-		err = ntfs_map_runlist(ni, vcn);
+		if (!write_locked) {
+			up_read(&ni->runlist.lock);
+			down_write(&ni->runlist.lock);
+			if (unlikely(ntfs_rl_vcn_to_lcn(ni->runlist.rl, vcn) !=
+					LCN_RL_NOT_MAPPED)) {
+				up_write(&ni->runlist.lock);
+				down_read(&ni->runlist.lock);
+				goto retry_remap;
+			}
+		}
+		err = ntfs_map_runlist_nolock(ni, vcn);
+		if (!write_locked) {
+			up_write(&ni->runlist.lock);
+			down_read(&ni->runlist.lock);
+		}
 		if (likely(!err)) {
 			is_retry = TRUE;
-			goto lock_retry_remap;
+			goto retry_remap;
 		}
 		/*
-		 * -EINVAL and -ENOENT coming from a failed mapping attempt are
+		 * -EINVAL coming from a failed mapping attempt is equivalent
-		 * equivalent to i/o errors for us as they should not happen in
+		 * to i/o error for us as it should not happen in our code
-		 * our code paths.
+		 * paths.
 		 */
-		if (err == -EINVAL || err == -ENOENT)
+		if (err == -EINVAL)
 			err = -EIO;
 	} else if (!err)
 		err = -EIO;
-	ntfs_error(ni->vol->sb, "Failed with error code %i.", err);
+	if (err != -ENOENT)
+		ntfs_error(ni->vol->sb, "Failed with error code %i.", err);
 	return ERR_PTR(err);
 }
 
@@ -870,15 +1005,14 @@ int ntfs_attr_lookup(const ATTR_TYPE type, const ntfschar *name,
 static inline void ntfs_attr_init_search_ctx(ntfs_attr_search_ctx *ctx,
 		ntfs_inode *ni, MFT_RECORD *mrec)
 {
-	ctx->mrec = mrec;
+	*ctx = (ntfs_attr_search_ctx) {
-	/* Sanity checks are performed elsewhere. */
+		.mrec = mrec,
-	ctx->attr = (ATTR_RECORD*)((u8*)mrec + le16_to_cpu(mrec->attrs_offset));
+		/* Sanity checks are performed elsewhere. */
-	ctx->is_first = TRUE;
+		.attr = (ATTR_RECORD*)((u8*)mrec +
-	ctx->ntfs_ino = ni;
+				le16_to_cpu(mrec->attrs_offset)),
-	ctx->al_entry = NULL;
+		.is_first = TRUE,
-	ctx->base_ntfs_ino = NULL;
+		.ntfs_ino = ni,
-	ctx->base_mrec = NULL;
+	};
-	ctx->base_attr = NULL;
 }
 
 /**
@@ -945,6 +1079,8 @@ void ntfs_attr_put_search_ctx(ntfs_attr_search_ctx *ctx)
 	return;
 }
 
+#ifdef NTFS_RW
+
 /**
  * ntfs_attr_find_in_attrdef - find an attribute in the $AttrDef system file
  * @vol:	ntfs volume to which the attribute belongs
@@ -1024,27 +1160,21 @@ int ntfs_attr_size_bounds_check(const ntfs_volume *vol, const ATTR_TYPE type,
  * Check whether the attribute of @type on the ntfs volume @vol is allowed to
  * be non-resident.  This information is obtained from $AttrDef system file.
  *
- * Return 0 if the attribute is allowed to be non-resident, -EPERM if not, or
+ * Return 0 if the attribute is allowed to be non-resident, -EPERM if not, and
  * -ENOENT if the attribute is not listed in $AttrDef.
  */
 int ntfs_attr_can_be_non_resident(const ntfs_volume *vol, const ATTR_TYPE type)
 {
 	ATTR_DEF *ad;
 
-	/*
-	 * $DATA is always allowed to be non-resident even if $AttrDef does not
-	 * specify this in the flags of the $DATA attribute definition record.
-	 */
-	if (type == AT_DATA)
-		return 0;
 	/* Find the attribute definition record in $AttrDef. */
 	ad = ntfs_attr_find_in_attrdef(vol, type);
 	if (unlikely(!ad))
 		return -ENOENT;
 	/* Check the flags and return the result. */
-	if (ad->flags & CAN_BE_NON_RESIDENT)
+	if (ad->flags & ATTR_DEF_RESIDENT)
-		return 0;
+		return -EPERM;
-	return -EPERM;
+	return 0;
 }
 
 /**
@@ -1067,9 +1197,9 @@ int ntfs_attr_can_be_non_resident(const ntfs_volume *vol, const ATTR_TYPE type)
  */
 int ntfs_attr_can_be_resident(const ntfs_volume *vol, const ATTR_TYPE type)
 {
-	if (type != AT_INDEX_ALLOCATION && type != AT_EA)
+	if (type == AT_INDEX_ALLOCATION || type == AT_EA)
-		return 0;
+		return -EPERM;
-	return -EPERM;
+	return 0;
 }
 
 /**
@@ -1116,6 +1246,328 @@ int ntfs_attr_record_resize(MFT_RECORD *m, ATTR_RECORD *a, u32 new_size)
 	return 0;
 }
 
+/**
+ * ntfs_attr_make_non_resident - convert a resident to a non-resident attribute
+ * @ni:		ntfs inode describing the attribute to convert
+ *
+ * Convert the resident ntfs attribute described by the ntfs inode @ni to a
+ * non-resident one.
+ *
+ * Return 0 on success and -errno on error.  The following error return codes
+ * are defined:
+ *	-EPERM	- The attribute is not allowed to be non-resident.
+ *	-ENOMEM	- Not enough memory.
+ *	-ENOSPC	- Not enough disk space.
+ *	-EINVAL	- Attribute not defined on the volume.
+ *	-EIO	- I/o error or other error.
+ * Note that -ENOSPC is also returned in the case that there is not enough
+ * space in the mft record to do the conversion.  This can happen when the mft
+ * record is already very full.  The caller is responsible for trying to make
+ * space in the mft record and trying again.  FIXME: Do we need a separate
+ * error return code for this kind of -ENOSPC or is it always worth trying
+ * again in case the attribute may then fit in a resident state so no need to
+ * make it non-resident at all?  Ho-hum...  (AIA)
+ *
+ * NOTE to self: No changes in the attribute list are required to move from
+ *		 a resident to a non-resident attribute.
+ *
+ * Locking: - The caller must hold i_sem on the inode.
+ */
+int ntfs_attr_make_non_resident(ntfs_inode *ni)
+{
+	s64 new_size;
+	struct inode *vi = VFS_I(ni);
+	ntfs_volume *vol = ni->vol;
+	ntfs_inode *base_ni;
+	MFT_RECORD *m;
+	ATTR_RECORD *a;
+	ntfs_attr_search_ctx *ctx;
+	struct page *page;
+	runlist_element *rl;
+	u8 *kaddr;
+	unsigned long flags;
+	int mp_size, mp_ofs, name_ofs, arec_size, err, err2;
+	u32 attr_size;
+	u8 old_res_attr_flags;
+
+	/* Check that the attribute is allowed to be non-resident. */
+	err = ntfs_attr_can_be_non_resident(vol, ni->type);
+	if (unlikely(err)) {
+		if (err == -EPERM)
+			ntfs_debug("Attribute is not allowed to be "
+					"non-resident.");
+		else
+			ntfs_debug("Attribute not defined on the NTFS "
+					"volume!");
+		return err;
+	}
+	/*
+	 * The size needs to be aligned to a cluster boundary for allocation
+	 * purposes.
+	 */
+	new_size = (i_size_read(vi) + vol->cluster_size - 1) &
+			~(vol->cluster_size - 1);
+	if (new_size > 0) {
+		runlist_element *rl2;
+
+		/*
+		 * Will need the page later and since the page lock nests
+		 * outside all ntfs locks, we need to get the page now.
+		 */
+		page = find_or_create_page(vi->i_mapping, 0,
+				mapping_gfp_mask(vi->i_mapping));
+		if (unlikely(!page))
+			return -ENOMEM;
+		/* Start by allocating clusters to hold the attribute value. */
+		rl = ntfs_cluster_alloc(vol, 0, new_size >>
+				vol->cluster_size_bits, -1, DATA_ZONE);
+		if (IS_ERR(rl)) {
+			err = PTR_ERR(rl);
+			ntfs_debug("Failed to allocate cluster%s, error code "
+					"%i.", (new_size >>
+					vol->cluster_size_bits) > 1 ? "s" : "",
+					err);
+			goto page_err_out;
+		}
+		/* Change the runlist terminator to LCN_ENOENT. */
+		rl2 = rl;
+		while (rl2->length)
+			rl2++;
+		BUG_ON(rl2->lcn != LCN_RL_NOT_MAPPED);
+		rl2->lcn = LCN_ENOENT;
+	} else {
+		rl = NULL;
+		page = NULL;
+	}
+	/* Determine the size of the mapping pairs array. */
+	mp_size = ntfs_get_size_for_mapping_pairs(vol, rl, 0, -1);
+	if (unlikely(mp_size < 0)) {
+		err = mp_size;
+		ntfs_debug("Failed to get size for mapping pairs array, error "
+				"code %i.", err);
+		goto rl_err_out;
+	}
+	down_write(&ni->runlist.lock);
+	if (!NInoAttr(ni))
+		base_ni = ni;
+	else
+		base_ni = ni->ext.base_ntfs_ino;
+	m = map_mft_record(base_ni);
+	if (IS_ERR(m)) {
+		err = PTR_ERR(m);
+		m = NULL;
+		ctx = NULL;
+		goto err_out;
+	}
+	ctx = ntfs_attr_get_search_ctx(base_ni, m);
+	if (unlikely(!ctx)) {
+		err = -ENOMEM;
+		goto err_out;
+	}
+	err = ntfs_attr_lookup(ni->type, ni->name, ni->name_len,
+			CASE_SENSITIVE, 0, NULL, 0, ctx);
+	if (unlikely(err)) {
+		if (err == -ENOENT)
+			err = -EIO;
+		goto err_out;
+	}
+	m = ctx->mrec;
+	a = ctx->attr;
+	BUG_ON(NInoNonResident(ni));
+	BUG_ON(a->non_resident);
+	/*
+	 * Calculate new offsets for the name and the mapping pairs array.
+	 * We assume the attribute is not compressed or sparse.
+	 */
+	name_ofs = (offsetof(ATTR_REC,
+			data.non_resident.compressed_size) + 7) & ~7;
+	mp_ofs = (name_ofs + a->name_length * sizeof(ntfschar) + 7) & ~7;
+	/*
+	 * Determine the size of the resident part of the now non-resident
+	 * attribute record.
+	 */
+	arec_size = (mp_ofs + mp_size + 7) & ~7;
+	/*
+	 * If the page is not uptodate bring it uptodate by copying from the
+	 * attribute value.
+	 */
+	attr_size = le32_to_cpu(a->data.resident.value_length);
+	BUG_ON(attr_size != i_size_read(vi));
+	if (page && !PageUptodate(page)) {
+		kaddr = kmap_atomic(page, KM_USER0);
+		memcpy(kaddr, (u8*)a +
+				le16_to_cpu(a->data.resident.value_offset),
+				attr_size);
+		memset(kaddr + attr_size, 0, PAGE_CACHE_SIZE - attr_size);
+		kunmap_atomic(kaddr, KM_USER0);
+		flush_dcache_page(page);
+		SetPageUptodate(page);
+	}
+	/* Backup the attribute flag. */
+	old_res_attr_flags = a->data.resident.flags;
+	/* Resize the resident part of the attribute record. */
+	err = ntfs_attr_record_resize(m, a, arec_size);
+	if (unlikely(err))
+		goto err_out;
+	/*
+	 * Convert the resident part of the attribute record to describe a
+	 * non-resident attribute.
+	 */
+	a->non_resident = 1;
+	/* Move the attribute name if it exists and update the offset. */
+	if (a->name_length)
+		memmove((u8*)a + name_ofs, (u8*)a + le16_to_cpu(a->name_offset),
+				a->name_length * sizeof(ntfschar));
+	a->name_offset = cpu_to_le16(name_ofs);
+	/*
+	 * FIXME: For now just clear all of these as we do not support them
+	 * when writing.
+	 */
+	a->flags &= cpu_to_le16(0xffff & ~le16_to_cpu(ATTR_IS_SPARSE |
+			ATTR_IS_ENCRYPTED | ATTR_COMPRESSION_MASK));
+	/* Setup the fields specific to non-resident attributes. */
+	a->data.non_resident.lowest_vcn = 0;
+	a->data.non_resident.highest_vcn = cpu_to_sle64((new_size - 1) >>
+			vol->cluster_size_bits);
+	a->data.non_resident.mapping_pairs_offset = cpu_to_le16(mp_ofs);
+	a->data.non_resident.compression_unit = 0;
+	memset(&a->data.non_resident.reserved, 0,
+			sizeof(a->data.non_resident.reserved));
+	a->data.non_resident.allocated_size = cpu_to_sle64(new_size);
+	a->data.non_resident.data_size =
+			a->data.non_resident.initialized_size =
+			cpu_to_sle64(attr_size);
+	/* Generate the mapping pairs array into the attribute record. */
+	err = ntfs_mapping_pairs_build(vol, (u8*)a + mp_ofs,
+			arec_size - mp_ofs, rl, 0, -1, NULL);
+	if (unlikely(err)) {
+		ntfs_debug("Failed to build mapping pairs, error code %i.",
+				err);
+		goto undo_err_out;
+	}
+	/* Setup the in-memory attribute structure to be non-resident. */
+	/*
+	 * FIXME: For now just clear all of these as we do not support them
+	 * when writing.
+	 */
+	NInoClearSparse(ni);
+	NInoClearEncrypted(ni);
+	NInoClearCompressed(ni);
+	ni->runlist.rl = rl;
+	write_lock_irqsave(&ni->size_lock, flags);
+	ni->allocated_size = new_size;
+	write_unlock_irqrestore(&ni->size_lock, flags);
+	/*
+	 * This needs to be last since the address space operations ->readpage
+	 * and ->writepage can run concurrently with us as they are not
+	 * serialized on i_sem.  Note, we are not allowed to fail once we flip
+	 * this switch, which is another reason to do this last.
+	 */
+	NInoSetNonResident(ni);
+	/* Mark the mft record dirty, so it gets written back. */
+	flush_dcache_mft_record_page(ctx->ntfs_ino);
+	mark_mft_record_dirty(ctx->ntfs_ino);
+	ntfs_attr_put_search_ctx(ctx);
+	unmap_mft_record(base_ni);
+	up_write(&ni->runlist.lock);
+	if (page) {
+		set_page_dirty(page);
+		unlock_page(page);
+		mark_page_accessed(page);
+		page_cache_release(page);
+	}
+	ntfs_debug("Done.");
+	return 0;
+undo_err_out:
+	/* Convert the attribute back into a resident attribute. */
+	a->non_resident = 0;
+	/* Move the attribute name if it exists and update the offset. */
+	name_ofs = (offsetof(ATTR_RECORD, data.resident.reserved) +
+			sizeof(a->data.resident.reserved) + 7) & ~7;
+	if (a->name_length)
+		memmove((u8*)a + name_ofs, (u8*)a + le16_to_cpu(a->name_offset),
+				a->name_length * sizeof(ntfschar));
+	mp_ofs = (name_ofs + a->name_length * sizeof(ntfschar) + 7) & ~7;
+	a->name_offset = cpu_to_le16(name_ofs);
+	arec_size = (mp_ofs + attr_size + 7) & ~7;
+	/* Resize the resident part of the attribute record. */
+	err2 = ntfs_attr_record_resize(m, a, arec_size);
+	if (unlikely(err2)) {
+		/*
+		 * This cannot happen (well if memory corruption is at work it
+		 * could happen in theory), but deal with it as well as we can.
+		 * If the old size is too small, truncate the attribute,
+		 * otherwise simply give it a larger allocated size.
+		 * FIXME: Should check whether chkdsk complains when the
+		 * allocated size is much bigger than the resident value size.
+		 */
+		arec_size = le32_to_cpu(a->length);
+		if ((mp_ofs + attr_size) > arec_size) {
+			err2 = attr_size;
+			attr_size = arec_size - mp_ofs;
+			ntfs_error(vol->sb, "Failed to undo partial resident "
+					"to non-resident attribute "
+					"conversion.  Truncating inode 0x%lx, "
+					"attribute type 0x%x from %i bytes to "
+					"%i bytes to maintain metadata "
+					"consistency.  THIS MEANS YOU ARE "
+					"LOSING %i BYTES DATA FROM THIS %s.",
+					vi->i_ino,
+					(unsigned)le32_to_cpu(ni->type),
+					err2, attr_size, err2 - attr_size,
+					((ni->type == AT_DATA) &&
+					!ni->name_len) ? "FILE": "ATTRIBUTE");
+			write_lock_irqsave(&ni->size_lock, flags);
+			ni->initialized_size = attr_size;
+			i_size_write(vi, attr_size);
+			write_unlock_irqrestore(&ni->size_lock, flags);
+		}
+	}
+	/* Setup the fields specific to resident attributes. */
+	a->data.resident.value_length = cpu_to_le32(attr_size);
+	a->data.resident.value_offset = cpu_to_le16(mp_ofs);
+	a->data.resident.flags = old_res_attr_flags;
+	memset(&a->data.resident.reserved, 0,
+			sizeof(a->data.resident.reserved));
+	/* Copy the data from the page back to the attribute value. */
+	if (page) {
+		kaddr = kmap_atomic(page, KM_USER0);
+		memcpy((u8*)a + mp_ofs, kaddr, attr_size);
+		kunmap_atomic(kaddr, KM_USER0);
+	}
+	/* Setup the allocated size in the ntfs inode in case it changed. */
+	write_lock_irqsave(&ni->size_lock, flags);
+	ni->allocated_size = arec_size - mp_ofs;
+	write_unlock_irqrestore(&ni->size_lock, flags);
+	/* Mark the mft record dirty, so it gets written back. */
+	flush_dcache_mft_record_page(ctx->ntfs_ino);
+	mark_mft_record_dirty(ctx->ntfs_ino);
+err_out:
+	if (ctx)
+		ntfs_attr_put_search_ctx(ctx);
+	if (m)
+		unmap_mft_record(base_ni);
+	ni->runlist.rl = NULL;
+	up_write(&ni->runlist.lock);
+rl_err_out:
+	if (rl) {
+		if (ntfs_cluster_free_from_rl(vol, rl) < 0) {
+			ntfs_error(vol->sb, "Failed to release allocated "
+					"cluster(s) in error code path.  Run "
+					"chkdsk to recover the lost "
+					"cluster(s).");
+			NVolSetErrors(vol);
+		}
+		ntfs_free(rl);
+page_err_out:
+		unlock_page(page);
+		page_cache_release(page);
+	}
+	if (err == -EINVAL)
+		err = -EIO;
+	return err;
+}
+
 /**
  * ntfs_attr_set - fill (a part of) an attribute with a byte
  * @ni:		ntfs inode describing the attribute to fill
@@ -1127,6 +1579,10 @@ int ntfs_attr_record_resize(MFT_RECORD *m, ATTR_RECORD *a, u32 new_size)
  * byte offset @ofs inside the attribute with the constant byte @val.
  *
  * This function is effectively like memset() applied to an ntfs attribute.
+ * Note thie function actually only operates on the page cache pages belonging
+ * to the ntfs attribute and it marks them dirty after doing the memset().
+ * Thus it relies on the vm dirty page write code paths to cause the modified
+ * pages to be written to the mft record/disk.
  *
  * Return 0 on success and -errno on error.  An error code of -ESPIPE means
  * that @ofs + @cnt were outside the end of the attribute and no write was
@@ -1155,7 +1611,7 @@ int ntfs_attr_set(ntfs_inode *ni, const s64 ofs, const s64 cnt, const u8 val)
 	end = ofs + cnt;
 	end_ofs = end & ~PAGE_CACHE_MASK;
 	/* If the end is outside the inode size return -ESPIPE. */
-	if (unlikely(end > VFS_I(ni)->i_size)) {
+	if (unlikely(end > i_size_read(VFS_I(ni)))) {
 		ntfs_error(vol->sb, "Request exceeds end of attribute.");
 		return -ESPIPE;
 	}
@@ -1256,3 +1712,5 @@ int ntfs_attr_set(ntfs_inode *ni, const s64 ofs, const s64 cnt, const u8 val)
 	ntfs_debug("Done.");
 	return 0;
 }
+
+#endif /* NTFS_RW */

fs/ntfs/attrib.h

@@ -2,7 +2,7 @@
  * attrib.h - Defines for attribute handling in NTFS Linux kernel driver.
  *	      Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -60,10 +60,14 @@ typedef struct {
 	ATTR_RECORD *base_attr;
 } ntfs_attr_search_ctx;
 
+extern int ntfs_map_runlist_nolock(ntfs_inode *ni, VCN vcn);
 extern int ntfs_map_runlist(ntfs_inode *ni, VCN vcn);
 
-extern runlist_element *ntfs_find_vcn(ntfs_inode *ni, const VCN vcn,
+extern LCN ntfs_attr_vcn_to_lcn_nolock(ntfs_inode *ni, const VCN vcn,
-		const BOOL need_write);
+		const BOOL write_locked);
+
+extern runlist_element *ntfs_attr_find_vcn_nolock(ntfs_inode *ni,
+		const VCN vcn, const BOOL write_locked);
 
 int ntfs_attr_lookup(const ATTR_TYPE type, const ntfschar *name,
 		const u32 name_len, const IGNORE_CASE_BOOL ic,
@@ -85,6 +89,8 @@ extern ntfs_attr_search_ctx *ntfs_attr_get_search_ctx(ntfs_inode *ni,
 		MFT_RECORD *mrec);
 extern void ntfs_attr_put_search_ctx(ntfs_attr_search_ctx *ctx);
 
+#ifdef NTFS_RW
+
 extern int ntfs_attr_size_bounds_check(const ntfs_volume *vol,
 		const ATTR_TYPE type, const s64 size);
 extern int ntfs_attr_can_be_non_resident(const ntfs_volume *vol,
@@ -94,7 +100,11 @@ extern int ntfs_attr_can_be_resident(const ntfs_volume *vol,
 
 extern int ntfs_attr_record_resize(MFT_RECORD *m, ATTR_RECORD *a, u32 new_size);
 
+extern int ntfs_attr_make_non_resident(ntfs_inode *ni);
+
 extern int ntfs_attr_set(ntfs_inode *ni, const s64 ofs, const s64 cnt,
 		const u8 val);
 
+#endif /* NTFS_RW */
+
 #endif /* _LINUX_NTFS_ATTRIB_H */

fs/ntfs/compress.c

@@ -96,13 +96,14 @@ void free_compression_buffers(void)
 /**
  * zero_partial_compressed_page - zero out of bounds compressed page region
  */
-static void zero_partial_compressed_page(ntfs_inode *ni, struct page *page)
+static void zero_partial_compressed_page(struct page *page,
+		const s64 initialized_size)
 {
 	u8 *kp = page_address(page);
 	unsigned int kp_ofs;
 
 	ntfs_debug("Zeroing page region outside initialized size.");
-	if (((s64)page->index << PAGE_CACHE_SHIFT) >= ni->initialized_size) {
+	if (((s64)page->index << PAGE_CACHE_SHIFT) >= initialized_size) {
 		/*
 		 * FIXME: Using clear_page() will become wrong when we get
 		 * PAGE_CACHE_SIZE != PAGE_SIZE but for now there is no problem.
@@ -110,7 +111,7 @@ static void zero_partial_compressed_page(ntfs_inode *ni, struct page *page)
 		clear_page(kp);
 		return;
 	}
-	kp_ofs = ni->initialized_size & ~PAGE_CACHE_MASK;
+	kp_ofs = initialized_size & ~PAGE_CACHE_MASK;
 	memset(kp + kp_ofs, 0, PAGE_CACHE_SIZE - kp_ofs);
 	return;
 }
@@ -118,12 +119,12 @@ static void zero_partial_compressed_page(ntfs_inode *ni, struct page *page)
 /**
  * handle_bounds_compressed_page - test for&handle out of bounds compressed page
  */
-static inline void handle_bounds_compressed_page(ntfs_inode *ni,
+static inline void handle_bounds_compressed_page(struct page *page,
-		struct page *page)
+		const loff_t i_size, const s64 initialized_size)
 {
-	if ((page->index >= (ni->initialized_size >> PAGE_CACHE_SHIFT)) &&
+	if ((page->index >= (initialized_size >> PAGE_CACHE_SHIFT)) &&
-			(ni->initialized_size < VFS_I(ni)->i_size))
+			(initialized_size < i_size))
-		zero_partial_compressed_page(ni, page);
+		zero_partial_compressed_page(page, initialized_size);
 	return;
 }
 
@@ -138,6 +139,8 @@ static inline void handle_bounds_compressed_page(ntfs_inode *ni,
  * @xpage_done:		set to 1 if xpage was completed successfully (IN/OUT)
  * @cb_start:		compression block to decompress (IN)
  * @cb_size:		size of compression block @cb_start in bytes (IN)
+ * @i_size:		file size when we started the read (IN)
+ * @initialized_size:	initialized file size when we started the read (IN)
  *
  * The caller must have disabled preemption. ntfs_decompress() reenables it when
  * the critical section is finished.
@@ -165,7 +168,8 @@ static inline void handle_bounds_compressed_page(ntfs_inode *ni,
 static int ntfs_decompress(struct page *dest_pages[], int *dest_index,
 		int *dest_ofs, const int dest_max_index, const int dest_max_ofs,
 		const int xpage, char *xpage_done, u8 *const cb_start,
-		const u32 cb_size)
+		const u32 cb_size, const loff_t i_size,
+		const s64 initialized_size)
 {
 	/*
 	 * Pointers into the compressed data, i.e. the compression block (cb),
@@ -219,9 +223,6 @@ static int ntfs_decompress(struct page *dest_pages[], int *dest_index,
 		spin_unlock(&ntfs_cb_lock);
 		/* Second stage: finalize completed pages. */
 		if (nr_completed_pages > 0) {
-			struct page *page = dest_pages[completed_pages[0]];
-			ntfs_inode *ni = NTFS_I(page->mapping->host);
-
 			for (i = 0; i < nr_completed_pages; i++) {
 				int di = completed_pages[i];
 
@@ -230,7 +231,8 @@ static int ntfs_decompress(struct page *dest_pages[], int *dest_index,
 				 * If we are outside the initialized size, zero
 				 * the out of bounds page range.
 				 */
-				handle_bounds_compressed_page(ni, dp);
+				handle_bounds_compressed_page(dp, i_size,
+						initialized_size);
 				flush_dcache_page(dp);
 				kunmap(dp);
 				SetPageUptodate(dp);
@@ -478,12 +480,14 @@ static int ntfs_decompress(struct page *dest_pages[], int *dest_index,
  */
 int ntfs_read_compressed_block(struct page *page)
 {
+	loff_t i_size;
+	s64 initialized_size;
 	struct address_space *mapping = page->mapping;
 	ntfs_inode *ni = NTFS_I(mapping->host);
 	ntfs_volume *vol = ni->vol;
 	struct super_block *sb = vol->sb;
 	runlist_element *rl;
-	unsigned long block_size = sb->s_blocksize;
+	unsigned long flags, block_size = sb->s_blocksize;
 	unsigned char block_size_bits = sb->s_blocksize_bits;
 	u8 *cb, *cb_pos, *cb_end;
 	struct buffer_head **bhs;
@@ -552,8 +556,12 @@ int ntfs_read_compressed_block(struct page *page)
 	 * The remaining pages need to be allocated and inserted into the page
 	 * cache, alignment guarantees keep all the below much simpler. (-8
 	 */
-	max_page = ((VFS_I(ni)->i_size + PAGE_CACHE_SIZE - 1) >>
+	read_lock_irqsave(&ni->size_lock, flags);
-			PAGE_CACHE_SHIFT) - offset;
+	i_size = i_size_read(VFS_I(ni));
+	initialized_size = ni->initialized_size;
+	read_unlock_irqrestore(&ni->size_lock, flags);
+	max_page = ((i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) -
+			offset;
 	if (nr_pages < max_page)
 		max_page = nr_pages;
 	for (i = 0; i < max_page; i++, offset++) {
@@ -824,7 +832,8 @@ int ntfs_read_compressed_block(struct page *page)
 				 * If we are outside the initialized size, zero
 				 * the out of bounds page range.
 				 */
-				handle_bounds_compressed_page(ni, page);
+				handle_bounds_compressed_page(page, i_size,
+						initialized_size);
 				flush_dcache_page(page);
 				kunmap(page);
 				SetPageUptodate(page);
@@ -847,7 +856,8 @@ int ntfs_read_compressed_block(struct page *page)
 		ntfs_debug("Found compressed compression block.");
 		err = ntfs_decompress(pages, &cur_page, &cur_ofs,
 				cb_max_page, cb_max_ofs, xpage, &xpage_done,
-				cb_pos,	cb_size - (cb_pos - cb));
+				cb_pos,	cb_size - (cb_pos - cb), i_size,
+				initialized_size);
 		/*
 		 * We can sleep from now on, lock already dropped by
 		 * ntfs_decompress().

fs/ntfs/debug.c

@@ -164,14 +164,17 @@ void ntfs_debug_dump_runlist(const runlist_element *rl)
 			if (index > -LCN_ENOENT - 1)
 				index = 3;
 			printk(KERN_DEBUG "%-16Lx %s %-16Lx%s\n",
-					(rl + i)->vcn, lcn_str[index],
+					(long long)(rl + i)->vcn, lcn_str[index],
-					(rl + i)->length, (rl + i)->length ?
+					(long long)(rl + i)->length,
-					"" : " (runlist end)");
+					(rl + i)->length ? "" :
+						" (runlist end)");
 		} else
 			printk(KERN_DEBUG "%-16Lx %-16Lx  %-16Lx%s\n",
-					(rl + i)->vcn, (rl + i)->lcn,
+					(long long)(rl + i)->vcn,
-					(rl + i)->length, (rl + i)->length ?
+					(long long)(rl + i)->lcn,
-					"" : " (runlist end)");
+					(long long)(rl + i)->length,
+					(rl + i)->length ? "" :
+						" (runlist end)");
 		if (!(rl + i)->length)
 			break;
 	}

fs/ntfs/dir.c

@@ -1,7 +1,7 @@
 /**
  * dir.c - NTFS kernel directory operations. Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -183,8 +183,7 @@ MFT_REF ntfs_lookup_inode_by_name(ntfs_inode *dir_ni, const ntfschar *uname,
 				name->len = 0;
 				*res = name;
 			} else {
-				if (name)
+				kfree(name);
-					kfree(name);
 				*res = NULL;
 			}
 			mref = le64_to_cpu(ie->data.dir.indexed_file);
@@ -444,8 +443,7 @@ MFT_REF ntfs_lookup_inode_by_name(ntfs_inode *dir_ni, const ntfschar *uname,
 				name->len = 0;
 				*res = name;
 			} else {
-				if (name)
+				kfree(name);
-					kfree(name);
 				*res = NULL;
 			}
 			mref = le64_to_cpu(ie->data.dir.indexed_file);
@@ -610,7 +608,7 @@ MFT_REF ntfs_lookup_inode_by_name(ntfs_inode *dir_ni, const ntfschar *uname,
 // TODO: (AIA)
 // The algorithm embedded in this code will be required for the time when we
 // want to support adding of entries to directories, where we require correct
-// collation of file names in order not to cause corruption of the file system.
+// collation of file names in order not to cause corruption of the filesystem.
 
 /**
  * ntfs_lookup_inode_by_name - find an inode in a directory given its name
@@ -1101,7 +1099,7 @@ static inline int ntfs_filldir(ntfs_volume *vol, loff_t fpos,
 static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
 {
 	s64 ia_pos, ia_start, prev_ia_pos, bmp_pos;
-	loff_t fpos;
+	loff_t fpos, i_size;
 	struct inode *bmp_vi, *vdir = filp->f_dentry->d_inode;
 	struct super_block *sb = vdir->i_sb;
 	ntfs_inode *ndir = NTFS_I(vdir);
@@ -1122,7 +1120,8 @@ static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
 			vdir->i_ino, fpos);
 	rc = err = 0;
 	/* Are we at end of dir yet? */
-	if (fpos >= vdir->i_size + vol->mft_record_size)
+	i_size = i_size_read(vdir);
+	if (fpos >= i_size + vol->mft_record_size)
 		goto done;
 	/* Emulate . and .. for all directories. */
 	if (!fpos) {
@@ -1264,7 +1263,7 @@ static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
 	bmp_mapping = bmp_vi->i_mapping;
 	/* Get the starting bitmap bit position and sanity check it. */
 	bmp_pos = ia_pos >> ndir->itype.index.block_size_bits;
-	if (unlikely(bmp_pos >> 3 >= bmp_vi->i_size)) {
+	if (unlikely(bmp_pos >> 3 >= i_size_read(bmp_vi))) {
 		ntfs_error(sb, "Current index allocation position exceeds "
 				"index bitmap size.");
 		goto err_out;
@@ -1301,7 +1300,7 @@ static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
 			goto get_next_bmp_page;
 		}
 		/* If we have reached the end of the bitmap, we are done. */
-		if (unlikely(((bmp_pos + cur_bmp_pos) >> 3) >= vdir->i_size))
+		if (unlikely(((bmp_pos + cur_bmp_pos) >> 3) >= i_size))
 			goto unm_EOD;
 		ia_pos = (bmp_pos + cur_bmp_pos) <<
 				ndir->itype.index.block_size_bits;
@@ -1309,7 +1308,8 @@ static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
 	ntfs_debug("Handling index buffer 0x%llx.",
 			(unsigned long long)bmp_pos + cur_bmp_pos);
 	/* If the current index buffer is in the same page we reuse the page. */
-	if ((prev_ia_pos & PAGE_CACHE_MASK) != (ia_pos & PAGE_CACHE_MASK)) {
+	if ((prev_ia_pos & (s64)PAGE_CACHE_MASK) !=
+			(ia_pos & (s64)PAGE_CACHE_MASK)) {
 		prev_ia_pos = ia_pos;
 		if (likely(ia_page != NULL)) {
 			unlock_page(ia_page);
@@ -1441,7 +1441,7 @@ static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
 	ntfs_unmap_page(bmp_page);
 EOD:
 	/* We are finished, set fpos to EOD. */
-	fpos = vdir->i_size + vol->mft_record_size;
+	fpos = i_size + vol->mft_record_size;
 abort:
 	kfree(name);
 done:
@@ -1461,10 +1461,8 @@ static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
 		unlock_page(ia_page);
 		ntfs_unmap_page(ia_page);
 	}
-	if (ir)
+	kfree(ir);
-		kfree(ir);
+	kfree(name);
-	if (name)
-		kfree(name);
 	if (ctx)
 		ntfs_attr_put_search_ctx(ctx);
 	if (m)
@@ -1495,7 +1493,7 @@ static int ntfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
 static int ntfs_dir_open(struct inode *vi, struct file *filp)
 {
 	if (sizeof(unsigned long) < 8) {
-		if (vi->i_size > MAX_LFS_FILESIZE)
+		if (i_size_read(vi) > MAX_LFS_FILESIZE)
 			return -EFBIG;
 	}
 	return 0;

fs/ntfs/file.c

@@ -47,7 +47,7 @@
 static int ntfs_file_open(struct inode *vi, struct file *filp)
 {
 	if (sizeof(unsigned long) < 8) {
-		if (vi->i_size > MAX_LFS_FILESIZE)
+		if (i_size_read(vi) > MAX_LFS_FILESIZE)
 			return -EFBIG;
 	}
 	return generic_file_open(vi, filp);

fs/ntfs/index.c

@@ -1,7 +1,7 @@
 /*
  * index.c - NTFS kernel index handling.  Part of the Linux-NTFS project.
  *
- * Copyright (c) 2004 Anton Altaparmakov
+ * Copyright (c) 2004-2005 Anton Altaparmakov
  *
  * This program/include file is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License as published
@@ -39,18 +39,8 @@ ntfs_index_context *ntfs_index_ctx_get(ntfs_inode *idx_ni)
 	ntfs_index_context *ictx;
 
 	ictx = kmem_cache_alloc(ntfs_index_ctx_cache, SLAB_NOFS);
-	if (ictx) {
+	if (ictx)
-		ictx->idx_ni = idx_ni;
+		*ictx = (ntfs_index_context){ .idx_ni = idx_ni };
-		ictx->entry = NULL;
-		ictx->data = NULL;
-		ictx->data_len = 0;
-		ictx->is_in_root = 0;
-		ictx->ir = NULL;
-		ictx->actx = NULL;
-		ictx->base_ni = NULL;
-		ictx->ia = NULL;
-		ictx->page = NULL;
-	}
 	return ictx;
 }
 

fs/ntfs/inode.h

@@ -2,7 +2,7 @@
  * inode.h - Defines for inode structures NTFS Linux kernel driver. Part of
  *	     the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -44,6 +44,7 @@ typedef struct _ntfs_inode ntfs_inode;
  * fields already provided in the VFS inode.
  */
 struct _ntfs_inode {
+	rwlock_t size_lock;	/* Lock serializing access to inode sizes. */
 	s64 initialized_size;	/* Copy from the attribute record. */
 	s64 allocated_size;	/* Copy from the attribute record. */
 	unsigned long state;	/* NTFS specific flags describing this inode.
@@ -109,7 +110,7 @@ struct _ntfs_inode {
 			u8 block_size_bits; 	/* Log2 of the above. */
 			u8 vcn_size_bits;	/* Log2 of the above. */
 		} index;
-		struct { /* It is a compressed file or an attribute inode. */
+		struct { /* It is a compressed/sparse file/attribute inode. */
 			s64 size;		/* Copy of compressed_size from
 						   $DATA. */
 			u32 block_size;		/* Size of a compression block
@@ -165,6 +166,7 @@ typedef enum {
 	NI_Sparse,		/* 1: Unnamed data attr is sparse (f).
 				   1: Create sparse files by default (d).
 				   1: Attribute is sparse (a). */
+	NI_SparseDisabled,	/* 1: May not create sparse regions. */
 	NI_TruncateFailed,	/* 1: Last ntfs_truncate() call failed. */
 } ntfs_inode_state_bits;
 
@@ -217,6 +219,7 @@ NINO_FNS(IndexAllocPresent)
 NINO_FNS(Compressed)
 NINO_FNS(Encrypted)
 NINO_FNS(Sparse)
+NINO_FNS(SparseDisabled)
 NINO_FNS(TruncateFailed)
 
 /*

fs/ntfs/layout.h

@@ -2,7 +2,7 @@
  * layout.h - All NTFS associated on-disk structures. Part of the Linux-NTFS
  *	      project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -547,26 +547,44 @@ enum {
 	COLLATION_NTOFS_ULONG		= const_cpu_to_le32(0x10),
 	COLLATION_NTOFS_SID		= const_cpu_to_le32(0x11),
 	COLLATION_NTOFS_SECURITY_HASH	= const_cpu_to_le32(0x12),
-	COLLATION_NTOFS_ULONGS		= const_cpu_to_le32(0x13)
+	COLLATION_NTOFS_ULONGS		= const_cpu_to_le32(0x13),
 };
 
 typedef le32 COLLATION_RULE;
 
 /*
  * The flags (32-bit) describing attribute properties in the attribute
- * definition structure.  FIXME: This information is from Regis's information
+ * definition structure.  FIXME: This information is based on Regis's
- * and, according to him, it is not certain and probably incomplete.
+ * information and, according to him, it is not certain and probably
- * The INDEXABLE flag is fairly certainly correct as only the file name
+ * incomplete.  The INDEXABLE flag is fairly certainly correct as only the file
- * attribute has this flag set and this is the only attribute indexed in NT4.
+ * name attribute has this flag set and this is the only attribute indexed in
+ * NT4.
  */
 enum {
-	INDEXABLE	    = const_cpu_to_le32(0x02), /* Attribute can be
+	ATTR_DEF_INDEXABLE	= const_cpu_to_le32(0x02), /* Attribute can be
-							  indexed. */
+					indexed. */
-	NEED_TO_REGENERATE  = const_cpu_to_le32(0x40), /* Need to regenerate
+	ATTR_DEF_MULTIPLE	= const_cpu_to_le32(0x04), /* Attribute type
-							  during regeneration
+					can be present multiple times in the
-							  phase. */
+					mft records of an inode. */
-	CAN_BE_NON_RESIDENT = const_cpu_to_le32(0x80), /* Attribute can be
+	ATTR_DEF_NOT_ZERO	= const_cpu_to_le32(0x08), /* Attribute value
-							  non-resident. */
+					must contain at least one non-zero
+					byte. */
+	ATTR_DEF_INDEXED_UNIQUE	= const_cpu_to_le32(0x10), /* Attribute must be
+					indexed and the attribute value must be
+					unique for the attribute type in all of
+					the mft records of an inode. */
+	ATTR_DEF_NAMED_UNIQUE	= const_cpu_to_le32(0x20), /* Attribute must be
+					named and the name must be unique for
+					the attribute type in all of the mft
+					records of an inode. */
+	ATTR_DEF_RESIDENT	= const_cpu_to_le32(0x40), /* Attribute must be
+					resident. */
+	ATTR_DEF_ALWAYS_LOG	= const_cpu_to_le32(0x80), /* Always log
+					modifications to this attribute,
+					regardless of whether it is resident or
+					non-resident.  Without this, only log
+					modifications if the attribute is
+					resident. */
 };
 
 typedef le32 ATTR_DEF_FLAGS;
@@ -749,10 +767,11 @@ typedef struct {
 				record header aligned to 8-byte boundary. */
 /* 34*/			u8 compression_unit; /* The compression unit expressed
 				as the log to the base 2 of the number of
-				clusters in a compression unit. 0 means not
+				clusters in a compression unit.  0 means not
-				compressed. (This effectively limits the
+				compressed.  (This effectively limits the
 				compression unit size to be a power of two
-				clusters.) WinNT4 only uses a value of 4. */
+				clusters.)  WinNT4 only uses a value of 4.
+				Sparse files also have this set to 4. */
 /* 35*/			u8 reserved[5];		/* Align to 8-byte boundary. */
 /* The sizes below are only used when lowest_vcn is zero, as otherwise it would
    be difficult to keep them up-to-date.*/
@@ -772,10 +791,10 @@ typedef struct {
 				data_size. */
 /* sizeof(uncompressed attr) = 64*/
 /* 64*/			sle64 compressed_size;	/* Byte size of the attribute
-				value after compression. Only present when
+				value after compression.  Only present when
-				compressed. Always is a multiple of the
+				compressed or sparse.  Always is a multiple of
-				cluster size. Represents the actual amount of
+				the cluster size.  Represents the actual amount
-				disk space being used on the disk. */
+				of disk space being used on the disk. */
 /* sizeof(compressed attr) = 72*/
 		} __attribute__ ((__packed__)) non_resident;
 	} __attribute__ ((__packed__)) data;
@@ -834,7 +853,7 @@ enum {
 	/* Note, this is a copy of the corresponding bit from the mft record,
 	   telling us whether this file has a view index present (eg. object id
 	   index, quota index, one of the security indexes or the encrypting
-	   file system related indexes). */
+	   filesystem related indexes). */
 };
 
 typedef le32 FILE_ATTR_FLAGS;
@@ -917,20 +936,12 @@ typedef struct {
 		/* 56*/	le64 quota_charged;	/* Byte size of the charge to
 				the quota for all streams of the file. Note: Is
 				zero if quotas are disabled. */
-		/* 64*/	le64 usn;		/* Last update sequence number
+		/* 64*/	leUSN usn;		/* Last update sequence number
-				of the file. This is a direct index into the
+				of the file.  This is a direct index into the
-				change (aka usn) journal file. It is zero if
+				transaction log file ($UsnJrnl).  It is zero if
-				the usn journal is disabled.
+				the usn journal is disabled or this file has
-				NOTE: To disable the journal need to delete
+				not been subject to logging yet.  See usnjrnl.h
-				the journal file itself and to then walk the
+				for details. */
-				whole mft and set all Usn entries in all mft
-				records to zero! (This can take a while!)
-				The journal is FILE_Extend/$UsnJrnl. Win2k
-				will recreate the journal and initiate
-				logging if necessary when mounting the
-				partition. This, in contrast to disabling the
-				journal is a very fast process, so the user
-				won't even notice it. */
 		} __attribute__ ((__packed__)) v3;
 	/* sizeof() = 72 bytes (NTFS 3.x) */
 	} __attribute__ ((__packed__)) ver;
@@ -1893,7 +1904,7 @@ enum {
 	VOLUME_FLAGS_MASK		= const_cpu_to_le16(0x803f),
 
 	/* To make our life easier when checking if we must mount read-only. */
-	VOLUME_MUST_MOUNT_RO_MASK	= const_cpu_to_le16(0x8037),
+	VOLUME_MUST_MOUNT_RO_MASK	= const_cpu_to_le16(0x8027),
 } __attribute__ ((__packed__));
 
 typedef le16 VOLUME_FLAGS;

fs/ntfs/lcnalloc.c

@@ -1,7 +1,7 @@
 /*
  * lcnalloc.c - Cluster (de)allocation code.  Part of the Linux-NTFS project.
  *
- * Copyright (c) 2004 Anton Altaparmakov
+ * Copyright (c) 2004-2005 Anton Altaparmakov
  *
  * This program/include file is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License as published
@@ -60,7 +60,7 @@ int ntfs_cluster_free_from_rl_nolock(ntfs_volume *vol,
 		if (rl->lcn < 0)
 			continue;
 		err = ntfs_bitmap_clear_run(lcnbmp_vi, rl->lcn, rl->length);
-		if (unlikely(err && (!ret || ret == ENOMEM) && ret != err))
+		if (unlikely(err && (!ret || ret == -ENOMEM) && ret != err))
 			ret = err;
 	}
 	ntfs_debug("Done.");
@@ -140,6 +140,7 @@ runlist_element *ntfs_cluster_alloc(ntfs_volume *vol, const VCN start_vcn,
 	LCN zone_start, zone_end, bmp_pos, bmp_initial_pos, last_read_pos, lcn;
 	LCN prev_lcn = 0, prev_run_len = 0, mft_zone_size;
 	s64 clusters;
+	loff_t i_size;
 	struct inode *lcnbmp_vi;
 	runlist_element *rl = NULL;
 	struct address_space *mapping;
@@ -249,6 +250,7 @@ runlist_element *ntfs_cluster_alloc(ntfs_volume *vol, const VCN start_vcn,
 	clusters = count;
 	rlpos = rlsize = 0;
 	mapping = lcnbmp_vi->i_mapping;
+	i_size = i_size_read(lcnbmp_vi);
 	while (1) {
 		ntfs_debug("Start of outer while loop: done_zones 0x%x, "
 				"search_zone %i, pass %i, zone_start 0x%llx, "
@@ -263,7 +265,7 @@ runlist_element *ntfs_cluster_alloc(ntfs_volume *vol, const VCN start_vcn,
 		last_read_pos = bmp_pos >> 3;
 		ntfs_debug("last_read_pos 0x%llx.",
 				(unsigned long long)last_read_pos);
-		if (last_read_pos > lcnbmp_vi->i_size) {
+		if (last_read_pos > i_size) {
 			ntfs_debug("End of attribute reached.  "
 					"Skipping to zone_pass_done.");
 			goto zone_pass_done;
@@ -287,11 +289,11 @@ runlist_element *ntfs_cluster_alloc(ntfs_volume *vol, const VCN start_vcn,
 		buf_size = last_read_pos & ~PAGE_CACHE_MASK;
 		buf = page_address(page) + buf_size;
 		buf_size = PAGE_CACHE_SIZE - buf_size;
-		if (unlikely(last_read_pos + buf_size > lcnbmp_vi->i_size))
+		if (unlikely(last_read_pos + buf_size > i_size))
-			buf_size = lcnbmp_vi->i_size - last_read_pos;
+			buf_size = i_size - last_read_pos;
 		buf_size <<= 3;
 		lcn = bmp_pos & 7;
-		bmp_pos &= ~7;
+		bmp_pos &= ~(LCN)7;
 		ntfs_debug("Before inner while loop: buf_size %i, lcn 0x%llx, "
 				"bmp_pos 0x%llx, need_writeback %i.", buf_size,
 				(unsigned long long)lcn,
@@ -309,7 +311,7 @@ runlist_element *ntfs_cluster_alloc(ntfs_volume *vol, const VCN start_vcn,
 					(unsigned int)*byte);
 			/* Skip full bytes. */
 			if (*byte == 0xff) {
-				lcn = (lcn + 8) & ~7;
+				lcn = (lcn + 8) & ~(LCN)7;
 				ntfs_debug("Continuing while loop 1.");
 				continue;
 			}
@@ -691,7 +693,7 @@ switch_to_data1_zone:		search_zone = 2;
 		if (zone == MFT_ZONE || mft_zone_size <= 0) {
 			ntfs_debug("No free clusters left, going to out.");
 			/* Really no more space left on device. */
-			err = ENOSPC;
+			err = -ENOSPC;
 			goto out;
 		} /* zone == DATA_ZONE && mft_zone_size > 0 */
 		ntfs_debug("Shrinking mft zone.");
@@ -755,13 +757,13 @@ switch_to_data1_zone:		search_zone = 2;
 	if (rl) {
 		int err2;
 
-		if (err == ENOSPC)
+		if (err == -ENOSPC)
 			ntfs_debug("Not enough space to complete allocation, "
-					"err ENOSPC, first free lcn 0x%llx, "
+					"err -ENOSPC, first free lcn 0x%llx, "
 					"could allocate up to 0x%llx "
 					"clusters.",
 					(unsigned long long)rl[0].lcn,
-					(unsigned long long)count - clusters);
+					(unsigned long long)(count - clusters));
 		/* Deallocate all allocated clusters. */
 		ntfs_debug("Attempting rollback...");
 		err2 = ntfs_cluster_free_from_rl_nolock(vol, rl);
@@ -773,10 +775,10 @@ switch_to_data1_zone:		search_zone = 2;
 		}
 		/* Free the runlist. */
 		ntfs_free(rl);
-	} else if (err == ENOSPC)
+	} else if (err == -ENOSPC)
-		ntfs_debug("No space left at all, err = ENOSPC, "
+		ntfs_debug("No space left at all, err = -ENOSPC, first free "
-				"first free lcn = 0x%llx.",
+				"lcn = 0x%llx.",
-				(unsigned long long)vol->data1_zone_pos);
+				(long long)vol->data1_zone_pos);
 	up_write(&vol->lcnbmp_lock);
 	return ERR_PTR(err);
 }
@@ -846,8 +848,8 @@ s64 __ntfs_cluster_free(struct inode *vi, const VCN start_vcn, s64 count,
 
 	total_freed = real_freed = 0;
 
-	/* This returns with ni->runlist locked for reading on success. */
+	down_read(&ni->runlist.lock);
-	rl = ntfs_find_vcn(ni, start_vcn, FALSE);
+	rl = ntfs_attr_find_vcn_nolock(ni, start_vcn, FALSE);
 	if (IS_ERR(rl)) {
 		if (!is_rollback)
 			ntfs_error(vol->sb, "Failed to find first runlist "
@@ -861,7 +863,7 @@ s64 __ntfs_cluster_free(struct inode *vi, const VCN start_vcn, s64 count,
 			ntfs_error(vol->sb, "First runlist element has "
 					"invalid lcn, aborting.");
 		err = -EIO;
-		goto unl_err_out;
+		goto err_out;
 	}
 	/* Find the starting cluster inside the run that needs freeing. */
 	delta = start_vcn - rl->vcn;
@@ -879,7 +881,7 @@ s64 __ntfs_cluster_free(struct inode *vi, const VCN start_vcn, s64 count,
 			if (!is_rollback)
 				ntfs_error(vol->sb, "Failed to clear first run "
 						"(error %i), aborting.", err);
-			goto unl_err_out;
+			goto err_out;
 		}
 		/* We have freed @to_free real clusters. */
 		real_freed = to_free;
@@ -899,30 +901,15 @@ s64 __ntfs_cluster_free(struct inode *vi, const VCN start_vcn, s64 count,
 		if (unlikely(rl->lcn < LCN_HOLE)) {
 			VCN vcn;
 
-			/*
+			/* Attempt to map runlist. */
-			 * Attempt to map runlist, dropping runlist lock for
-			 * the duration.
-			 */
 			vcn = rl->vcn;
-			up_read(&ni->runlist.lock);
+			rl = ntfs_attr_find_vcn_nolock(ni, vcn, FALSE);
-			err = ntfs_map_runlist(ni, vcn);
-			if (err) {
-				if (!is_rollback)
-					ntfs_error(vol->sb, "Failed to map "
-							"runlist fragment.");
-				if (err == -EINVAL || err == -ENOENT)
-					err = -EIO;
-				goto err_out;
-			}
-			/*
-			 * This returns with ni->runlist locked for reading on
-			 * success.
-			 */
-			rl = ntfs_find_vcn(ni, vcn, FALSE);
 			if (IS_ERR(rl)) {
 				err = PTR_ERR(rl);
 				if (!is_rollback)
-					ntfs_error(vol->sb, "Failed to find "
+					ntfs_error(vol->sb, "Failed to map "
+							"runlist fragment or "
+							"failed to find "
 							"subsequent runlist "
 							"element.");
 				goto err_out;
@@ -935,7 +922,7 @@ s64 __ntfs_cluster_free(struct inode *vi, const VCN start_vcn, s64 count,
 							(unsigned long long)
 							rl->lcn);
 				err = -EIO;
-				goto unl_err_out;
+				goto err_out;
 			}
 		}
 		/* The number of clusters in this run that need freeing. */
@@ -951,7 +938,7 @@ s64 __ntfs_cluster_free(struct inode *vi, const VCN start_vcn, s64 count,
 				if (!is_rollback)
 					ntfs_error(vol->sb, "Failed to clear "
 							"subsequent run.");
-				goto unl_err_out;
+				goto err_out;
 			}
 			/* We have freed @to_free real clusters. */
 			real_freed += to_free;
@@ -972,9 +959,8 @@ s64 __ntfs_cluster_free(struct inode *vi, const VCN start_vcn, s64 count,
 	/* We are done.  Return the number of actually freed clusters. */
 	ntfs_debug("Done.");
 	return real_freed;
-unl_err_out:
-	up_read(&ni->runlist.lock);
 err_out:
+	up_read(&ni->runlist.lock);
 	if (is_rollback)
 		return err;
 	/* If no real clusters were freed, no need to rollback. */

fs/ntfs/logfile.c

@@ -1,7 +1,7 @@
 /*
  * logfile.c - NTFS kernel journal handling. Part of the Linux-NTFS project.
  *
- * Copyright (c) 2002-2004 Anton Altaparmakov
+ * Copyright (c) 2002-2005 Anton Altaparmakov
  *
  * This program/include file is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License as published
@@ -410,7 +410,7 @@ static BOOL ntfs_check_and_load_restart_page(struct inode *vi,
 }
 
 /**
- * ntfs_ckeck_logfile - check in the journal if the volume is consistent
+ * ntfs_check_logfile - check the journal for consistency
  * @log_vi:	struct inode of loaded journal $LogFile to check
  *
  * Check the $LogFile journal for consistency and return TRUE if it is
@@ -443,7 +443,7 @@ BOOL ntfs_check_logfile(struct inode *log_vi)
 	/* An empty $LogFile must have been clean before it got emptied. */
 	if (NVolLogFileEmpty(vol))
 		goto is_empty;
-	size = log_vi->i_size;
+	size = i_size_read(log_vi);
 	/* Make sure the file doesn't exceed the maximum allowed size. */
 	if (size > MaxLogFileSize)
 		size = MaxLogFileSize;
@@ -464,7 +464,7 @@ BOOL ntfs_check_logfile(struct inode *log_vi)
 	 * optimize log_page_size and log_page_bits into constants.
 	 */
 	log_page_bits = generic_ffs(log_page_size) - 1;
-	size &= ~(log_page_size - 1);
+	size &= ~(s64)(log_page_size - 1);
 	/*
 	 * Ensure the log file is big enough to store at least the two restart
 	 * pages and the minimum number of log record pages.
@@ -689,7 +689,8 @@ BOOL ntfs_empty_logfile(struct inode *log_vi)
 	if (!NVolLogFileEmpty(vol)) {
 		int err;
 		
-		err = ntfs_attr_set(NTFS_I(log_vi), 0, log_vi->i_size, 0xff);
+		err = ntfs_attr_set(NTFS_I(log_vi), 0, i_size_read(log_vi),
+				0xff);
 		if (unlikely(err)) {
 			ntfs_error(vol->sb, "Failed to fill $LogFile with "
 					"0xff bytes (error code %i).", err);

fs/ntfs/mft.c

@@ -1,7 +1,7 @@
 /**
  * mft.c - NTFS kernel mft record operations. Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -45,6 +45,7 @@
  */
 static inline MFT_RECORD *map_mft_record_page(ntfs_inode *ni)
 {
+	loff_t i_size;
 	ntfs_volume *vol = ni->vol;
 	struct inode *mft_vi = vol->mft_ino;
 	struct page *page;
@@ -60,13 +61,14 @@ static inline MFT_RECORD *map_mft_record_page(ntfs_inode *ni)
 	index = ni->mft_no << vol->mft_record_size_bits >> PAGE_CACHE_SHIFT;
 	ofs = (ni->mft_no << vol->mft_record_size_bits) & ~PAGE_CACHE_MASK;
 
+	i_size = i_size_read(mft_vi);
 	/* The maximum valid index into the page cache for $MFT's data. */
-	end_index = mft_vi->i_size >> PAGE_CACHE_SHIFT;
+	end_index = i_size >> PAGE_CACHE_SHIFT;
 
 	/* If the wanted index is out of bounds the mft record doesn't exist. */
 	if (unlikely(index >= end_index)) {
-		if (index > end_index || (mft_vi->i_size & ~PAGE_CACHE_MASK) <
+		if (index > end_index || (i_size & ~PAGE_CACHE_MASK) < ofs +
-				ofs + vol->mft_record_size) {
+				vol->mft_record_size) {
 			page = ERR_PTR(-ENOENT);
 			ntfs_error(vol->sb, "Attemt to read mft record 0x%lx, "
 					"which is beyond the end of the mft.  "
@@ -285,7 +287,7 @@ MFT_RECORD *map_extent_mft_record(ntfs_inode *base_ni, MFT_REF mref,
 			}
 			unmap_mft_record(ni);
 			ntfs_error(base_ni->vol->sb, "Found stale extent mft "
-					"reference! Corrupt file system. "
+					"reference! Corrupt filesystem. "
 					"Run chkdsk.");
 			return ERR_PTR(-EIO);
 		}
@@ -316,7 +318,7 @@ MFT_RECORD *map_extent_mft_record(ntfs_inode *base_ni, MFT_REF mref,
 	/* Verify the sequence number if it is present. */
 	if (seq_no && (le16_to_cpu(m->sequence_number) != seq_no)) {
 		ntfs_error(base_ni->vol->sb, "Found stale extent mft "
-				"reference! Corrupt file system. Run chkdsk.");
+				"reference! Corrupt filesystem. Run chkdsk.");
 		destroy_ni = TRUE;
 		m = ERR_PTR(-EIO);
 		goto unm_err_out;
@@ -946,20 +948,23 @@ BOOL ntfs_may_write_mft_record(ntfs_volume *vol, const unsigned long mft_no,
 	na.name_len = 0;
 	na.type = AT_UNUSED;
 	/*
-	 * For inode 0, i.e. $MFT itself, we cannot use ilookup5() from here or
+	 * Optimize inode 0, i.e. $MFT itself, since we have it in memory and
-	 * we deadlock because the inode is already locked by the kernel
+	 * we get here for it rather often.
-	 * (fs/fs-writeback.c::__sync_single_inode()) and ilookup5() waits
-	 * until the inode is unlocked before returning it and it never gets
-	 * unlocked because ntfs_should_write_mft_record() never returns.  )-:
-	 * Fortunately, we have inode 0 pinned in icache for the duration of
-	 * the mount so we can access it directly.
 	 */
 	if (!mft_no) {
 		/* Balance the below iput(). */
 		vi = igrab(mft_vi);
 		BUG_ON(vi != mft_vi);
-	} else
+	} else {
-		vi = ilookup5(sb, mft_no, (test_t)ntfs_test_inode, &na);
+		/*
+		 * Have to use ilookup5_nowait() since ilookup5() waits for the
+		 * inode lock which causes ntfs to deadlock when a concurrent
+		 * inode write via the inode dirty code paths and the page
+		 * dirty code path of the inode dirty code path when writing
+		 * $MFT occurs.
+		 */
+		vi = ilookup5_nowait(sb, mft_no, (test_t)ntfs_test_inode, &na);
+	}
 	if (vi) {
 		ntfs_debug("Base inode 0x%lx is in icache.", mft_no);
 		/* The inode is in icache. */
@@ -1014,7 +1019,13 @@ BOOL ntfs_may_write_mft_record(ntfs_volume *vol, const unsigned long mft_no,
 	na.mft_no = MREF_LE(m->base_mft_record);
 	ntfs_debug("Mft record 0x%lx is an extent record.  Looking for base "
 			"inode 0x%lx in icache.", mft_no, na.mft_no);
-	vi = ilookup5(sb, na.mft_no, (test_t)ntfs_test_inode, &na);
+	if (!na.mft_no) {
+		/* Balance the below iput(). */
+		vi = igrab(mft_vi);
+		BUG_ON(vi != mft_vi);
+	} else
+		vi = ilookup5_nowait(sb, na.mft_no, (test_t)ntfs_test_inode,
+				&na);
 	if (!vi) {
 		/*
 		 * The base inode is not in icache, write this extent mft
@@ -1121,6 +1132,7 @@ static int ntfs_mft_bitmap_find_and_alloc_free_rec_nolock(ntfs_volume *vol,
 		ntfs_inode *base_ni)
 {
 	s64 pass_end, ll, data_pos, pass_start, ofs, bit;
+	unsigned long flags;
 	struct address_space *mftbmp_mapping;
 	u8 *buf, *byte;
 	struct page *page;
@@ -1134,9 +1146,13 @@ static int ntfs_mft_bitmap_find_and_alloc_free_rec_nolock(ntfs_volume *vol,
 	 * Set the end of the pass making sure we do not overflow the mft
 	 * bitmap.
 	 */
+	read_lock_irqsave(&NTFS_I(vol->mft_ino)->size_lock, flags);
 	pass_end = NTFS_I(vol->mft_ino)->allocated_size >>
 			vol->mft_record_size_bits;
+	read_unlock_irqrestore(&NTFS_I(vol->mft_ino)->size_lock, flags);
+	read_lock_irqsave(&NTFS_I(vol->mftbmp_ino)->size_lock, flags);
 	ll = NTFS_I(vol->mftbmp_ino)->initialized_size << 3;
+	read_unlock_irqrestore(&NTFS_I(vol->mftbmp_ino)->size_lock, flags);
 	if (pass_end > ll)
 		pass_end = ll;
 	pass = 1;
@@ -1263,6 +1279,7 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 {
 	LCN lcn;
 	s64 ll;
+	unsigned long flags;
 	struct page *page;
 	ntfs_inode *mft_ni, *mftbmp_ni;
 	runlist_element *rl, *rl2 = NULL;
@@ -1284,17 +1301,20 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 	/*
 	 * Determine the last lcn of the mft bitmap.  The allocated size of the
 	 * mft bitmap cannot be zero so we are ok to do this.
-	 * ntfs_find_vcn() returns the runlist locked on success.
 	 */
-	rl = ntfs_find_vcn(mftbmp_ni, (mftbmp_ni->allocated_size - 1) >>
+	down_write(&mftbmp_ni->runlist.lock);
-			vol->cluster_size_bits, TRUE);
+	read_lock_irqsave(&mftbmp_ni->size_lock, flags);
+	ll = mftbmp_ni->allocated_size;
+	read_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
+	rl = ntfs_attr_find_vcn_nolock(mftbmp_ni,
+			(ll - 1) >> vol->cluster_size_bits, TRUE);
 	if (unlikely(IS_ERR(rl) || !rl->length || rl->lcn < 0)) {
+		up_write(&mftbmp_ni->runlist.lock);
 		ntfs_error(vol->sb, "Failed to determine last allocated "
 				"cluster of mft bitmap attribute.");
-		if (!IS_ERR(rl)) {
+		if (!IS_ERR(rl))
-			up_write(&mftbmp_ni->runlist.lock);
 			ret = -EIO;
-		} else
+		else
 			ret = PTR_ERR(rl);
 		return ret;
 	}
@@ -1396,7 +1416,7 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 	BUG_ON(ll < rl2->vcn);
 	BUG_ON(ll >= rl2->vcn + rl2->length);
 	/* Get the size for the new mapping pairs array for this extent. */
-	mp_size = ntfs_get_size_for_mapping_pairs(vol, rl2, ll);
+	mp_size = ntfs_get_size_for_mapping_pairs(vol, rl2, ll, -1);
 	if (unlikely(mp_size <= 0)) {
 		ntfs_error(vol->sb, "Get size for mapping pairs failed for "
 				"mft bitmap attribute extent.");
@@ -1418,6 +1438,8 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 		// TODO: Deal with this by moving this extent to a new mft
 		// record or by starting a new extent in a new mft record or by
 		// moving other attributes out of this mft record.
+		// Note: It will need to be a special mft record and if none of
+		// those are available it gets rather complicated...
 		ntfs_error(vol->sb, "Not enough space in this mft record to "
 				"accomodate extended mft bitmap attribute "
 				"extent.  Cannot handle this yet.");
@@ -1428,7 +1450,7 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 	/* Generate the mapping pairs array directly into the attr record. */
 	ret = ntfs_mapping_pairs_build(vol, (u8*)a +
 			le16_to_cpu(a->data.non_resident.mapping_pairs_offset),
-			mp_size, rl2, ll, NULL);
+			mp_size, rl2, ll, -1, NULL);
 	if (unlikely(ret)) {
 		ntfs_error(vol->sb, "Failed to build mapping pairs array for "
 				"mft bitmap attribute.");
@@ -1458,9 +1480,11 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 		}
 		a = ctx->attr;
 	}
+	write_lock_irqsave(&mftbmp_ni->size_lock, flags);
 	mftbmp_ni->allocated_size += vol->cluster_size;
 	a->data.non_resident.allocated_size =
 			cpu_to_sle64(mftbmp_ni->allocated_size);
+	write_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
 	/* Ensure the changes make it to disk. */
 	flush_dcache_mft_record_page(ctx->ntfs_ino);
 	mark_mft_record_dirty(ctx->ntfs_ino);
@@ -1476,7 +1500,9 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 			0, ctx)) {
 		ntfs_error(vol->sb, "Failed to find last attribute extent of "
 				"mft bitmap attribute.%s", es);
+		write_lock_irqsave(&mftbmp_ni->size_lock, flags);
 		mftbmp_ni->allocated_size += vol->cluster_size;
+		write_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
 		ntfs_attr_put_search_ctx(ctx);
 		unmap_mft_record(mft_ni);
 		up_write(&mftbmp_ni->runlist.lock);
@@ -1512,7 +1538,7 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 				a->data.non_resident.mapping_pairs_offset),
 				old_alen - le16_to_cpu(
 				a->data.non_resident.mapping_pairs_offset),
-				rl2, ll, NULL)) {
+				rl2, ll, -1, NULL)) {
 			ntfs_error(vol->sb, "Failed to restore mapping pairs "
 					"array.%s", es);
 			NVolSetErrors(vol);
@@ -1550,6 +1576,7 @@ static int ntfs_mft_bitmap_extend_allocation_nolock(ntfs_volume *vol)
 static int ntfs_mft_bitmap_extend_initialized_nolock(ntfs_volume *vol)
 {
 	s64 old_data_size, old_initialized_size;
+	unsigned long flags;
 	struct inode *mftbmp_vi;
 	ntfs_inode *mft_ni, *mftbmp_ni;
 	ntfs_attr_search_ctx *ctx;
@@ -1583,7 +1610,8 @@ static int ntfs_mft_bitmap_extend_initialized_nolock(ntfs_volume *vol)
 		goto put_err_out;
 	}
 	a = ctx->attr;
-	old_data_size = mftbmp_vi->i_size;
+	write_lock_irqsave(&mftbmp_ni->size_lock, flags);
+	old_data_size = i_size_read(mftbmp_vi);
 	old_initialized_size = mftbmp_ni->initialized_size;
 	/*
 	 * We can simply update the initialized_size before filling the space
@@ -1593,11 +1621,12 @@ static int ntfs_mft_bitmap_extend_initialized_nolock(ntfs_volume *vol)
 	mftbmp_ni->initialized_size += 8;
 	a->data.non_resident.initialized_size =
 			cpu_to_sle64(mftbmp_ni->initialized_size);
-	if (mftbmp_ni->initialized_size > mftbmp_vi->i_size) {
+	if (mftbmp_ni->initialized_size > old_data_size) {
-		mftbmp_vi->i_size = mftbmp_ni->initialized_size;
+		i_size_write(mftbmp_vi, mftbmp_ni->initialized_size);
 		a->data.non_resident.data_size =
-				cpu_to_sle64(mftbmp_vi->i_size);
+				cpu_to_sle64(mftbmp_ni->initialized_size);
 	}
+	write_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
 	/* Ensure the changes make it to disk. */
 	flush_dcache_mft_record_page(ctx->ntfs_ino);
 	mark_mft_record_dirty(ctx->ntfs_ino);
@@ -1636,22 +1665,28 @@ static int ntfs_mft_bitmap_extend_initialized_nolock(ntfs_volume *vol)
 		goto err_out;
 	}
 	a = ctx->attr;
+	write_lock_irqsave(&mftbmp_ni->size_lock, flags);
 	mftbmp_ni->initialized_size = old_initialized_size;
 	a->data.non_resident.initialized_size =
 			cpu_to_sle64(old_initialized_size);
-	if (mftbmp_vi->i_size != old_data_size) {
+	if (i_size_read(mftbmp_vi) != old_data_size) {
-		mftbmp_vi->i_size = old_data_size;
+		i_size_write(mftbmp_vi, old_data_size);
 		a->data.non_resident.data_size = cpu_to_sle64(old_data_size);
 	}
+	write_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
 	flush_dcache_mft_record_page(ctx->ntfs_ino);
 	mark_mft_record_dirty(ctx->ntfs_ino);
 	ntfs_attr_put_search_ctx(ctx);
 	unmap_mft_record(mft_ni);
+#ifdef DEBUG
+	read_lock_irqsave(&mftbmp_ni->size_lock, flags);
 	ntfs_debug("Restored status of mftbmp: allocated_size 0x%llx, "
 			"data_size 0x%llx, initialized_size 0x%llx.",
 			(long long)mftbmp_ni->allocated_size,
-			(long long)mftbmp_vi->i_size,
+			(long long)i_size_read(mftbmp_vi),
 			(long long)mftbmp_ni->initialized_size);
+	read_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
+#endif /* DEBUG */
 err_out:
 	return ret;
 }
@@ -1679,7 +1714,8 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 {
 	LCN lcn;
 	VCN old_last_vcn;
-	s64 min_nr, nr, ll = 0;
+	s64 min_nr, nr, ll;
+	unsigned long flags;
 	ntfs_inode *mft_ni;
 	runlist_element *rl, *rl2;
 	ntfs_attr_search_ctx *ctx = NULL;
@@ -1695,23 +1731,25 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 	 * Determine the preferred allocation location, i.e. the last lcn of
 	 * the mft data attribute.  The allocated size of the mft data
 	 * attribute cannot be zero so we are ok to do this.
-	 * ntfs_find_vcn() returns the runlist locked on success.
 	 */
-	rl = ntfs_find_vcn(mft_ni, (mft_ni->allocated_size - 1) >>
+	down_write(&mft_ni->runlist.lock);
-			vol->cluster_size_bits, TRUE);
+	read_lock_irqsave(&mft_ni->size_lock, flags);
+	ll = mft_ni->allocated_size;
+	read_unlock_irqrestore(&mft_ni->size_lock, flags);
+	rl = ntfs_attr_find_vcn_nolock(mft_ni,
+			(ll - 1) >> vol->cluster_size_bits, TRUE);
 	if (unlikely(IS_ERR(rl) || !rl->length || rl->lcn < 0)) {
+		up_write(&mft_ni->runlist.lock);
 		ntfs_error(vol->sb, "Failed to determine last allocated "
 				"cluster of mft data attribute.");
-		if (!IS_ERR(rl)) {
+		if (!IS_ERR(rl))
-			up_write(&mft_ni->runlist.lock);
 			ret = -EIO;
-		} else
+		else
 			ret = PTR_ERR(rl);
 		return ret;
 	}
 	lcn = rl->lcn + rl->length;
-	ntfs_debug("Last lcn of mft data attribute is 0x%llx.",
+	ntfs_debug("Last lcn of mft data attribute is 0x%llx.", (long long)lcn);
-			(long long)lcn);
 	/* Minimum allocation is one mft record worth of clusters. */
 	min_nr = vol->mft_record_size >> vol->cluster_size_bits;
 	if (!min_nr)
@@ -1721,12 +1759,13 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 	if (!nr)
 		nr = min_nr;
 	/* Ensure we do not go above 2^32-1 mft records. */
-	if (unlikely((mft_ni->allocated_size +
+	read_lock_irqsave(&mft_ni->size_lock, flags);
-			(nr << vol->cluster_size_bits)) >>
+	ll = mft_ni->allocated_size;
+	read_unlock_irqrestore(&mft_ni->size_lock, flags);
+	if (unlikely((ll + (nr << vol->cluster_size_bits)) >>
 			vol->mft_record_size_bits >= (1ll << 32))) {
 		nr = min_nr;
-		if (unlikely((mft_ni->allocated_size +
+		if (unlikely((ll + (nr << vol->cluster_size_bits)) >>
-				(nr << vol->cluster_size_bits)) >>
 				vol->mft_record_size_bits >= (1ll << 32))) {
 			ntfs_warning(vol->sb, "Cannot allocate mft record "
 					"because the maximum number of inodes "
@@ -1772,7 +1811,7 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 		return PTR_ERR(rl);
 	}
 	mft_ni->runlist.rl = rl;
-	ntfs_debug("Allocated %lli clusters.", nr);
+	ntfs_debug("Allocated %lli clusters.", (long long)nr);
 	/* Find the last run in the new runlist. */
 	for (; rl[1].length; rl++)
 		;
@@ -1808,7 +1847,7 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 	BUG_ON(ll < rl2->vcn);
 	BUG_ON(ll >= rl2->vcn + rl2->length);
 	/* Get the size for the new mapping pairs array for this extent. */
-	mp_size = ntfs_get_size_for_mapping_pairs(vol, rl2, ll);
+	mp_size = ntfs_get_size_for_mapping_pairs(vol, rl2, ll, -1);
 	if (unlikely(mp_size <= 0)) {
 		ntfs_error(vol->sb, "Get size for mapping pairs failed for "
 				"mft data attribute extent.");
@@ -1832,7 +1871,11 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 		// moving other attributes out of this mft record.
 		// Note: Use the special reserved mft records and ensure that
 		// this extent is not required to find the mft record in
-		// question.
+		// question.  If no free special records left we would need to
+		// move an existing record away, insert ours in its place, and
+		// then place the moved record into the newly allocated space
+		// and we would then need to update all references to this mft
+		// record appropriately.  This is rather complicated...
 		ntfs_error(vol->sb, "Not enough space in this mft record to "
 				"accomodate extended mft data attribute "
 				"extent.  Cannot handle this yet.");
@@ -1843,7 +1886,7 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 	/* Generate the mapping pairs array directly into the attr record. */
 	ret = ntfs_mapping_pairs_build(vol, (u8*)a +
 			le16_to_cpu(a->data.non_resident.mapping_pairs_offset),
-			mp_size, rl2, ll, NULL);
+			mp_size, rl2, ll, -1, NULL);
 	if (unlikely(ret)) {
 		ntfs_error(vol->sb, "Failed to build mapping pairs array of "
 				"mft data attribute.");
@@ -1875,9 +1918,11 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 		}
 		a = ctx->attr;
 	}
+	write_lock_irqsave(&mft_ni->size_lock, flags);
 	mft_ni->allocated_size += nr << vol->cluster_size_bits;
 	a->data.non_resident.allocated_size =
 			cpu_to_sle64(mft_ni->allocated_size);
+	write_unlock_irqrestore(&mft_ni->size_lock, flags);
 	/* Ensure the changes make it to disk. */
 	flush_dcache_mft_record_page(ctx->ntfs_ino);
 	mark_mft_record_dirty(ctx->ntfs_ino);
@@ -1892,7 +1937,9 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 			CASE_SENSITIVE, rl[1].vcn, NULL, 0, ctx)) {
 		ntfs_error(vol->sb, "Failed to find last attribute extent of "
 				"mft data attribute.%s", es);
+		write_lock_irqsave(&mft_ni->size_lock, flags);
 		mft_ni->allocated_size += nr << vol->cluster_size_bits;
+		write_unlock_irqrestore(&mft_ni->size_lock, flags);
 		ntfs_attr_put_search_ctx(ctx);
 		unmap_mft_record(mft_ni);
 		up_write(&mft_ni->runlist.lock);
@@ -1921,7 +1968,7 @@ static int ntfs_mft_data_extend_allocation_nolock(ntfs_volume *vol)
 				a->data.non_resident.mapping_pairs_offset),
 				old_alen - le16_to_cpu(
 				a->data.non_resident.mapping_pairs_offset),
-				rl2, ll, NULL)) {
+				rl2, ll, -1, NULL)) {
 			ntfs_error(vol->sb, "Failed to restore mapping pairs "
 					"array.%s", es);
 			NVolSetErrors(vol);
@@ -1991,7 +2038,7 @@ static int ntfs_mft_record_layout(const ntfs_volume *vol, const s64 mft_no,
 				"reports this as corruption, please email "
 				"linux-ntfs-dev@lists.sourceforge.net stating "
 				"that you saw this message and that the "
-				"modified file system created was corrupt.  "
+				"modified filesystem created was corrupt.  "
 				"Thank you.");
 	}
 	/* Set the update sequence number to 1. */
@@ -2036,6 +2083,7 @@ static int ntfs_mft_record_layout(const ntfs_volume *vol, const s64 mft_no,
  */
 static int ntfs_mft_record_format(const ntfs_volume *vol, const s64 mft_no)
 {
+	loff_t i_size;
 	struct inode *mft_vi = vol->mft_ino;
 	struct page *page;
 	MFT_RECORD *m;
@@ -2051,10 +2099,11 @@ static int ntfs_mft_record_format(const ntfs_volume *vol, const s64 mft_no)
 	index = mft_no << vol->mft_record_size_bits >> PAGE_CACHE_SHIFT;
 	ofs = (mft_no << vol->mft_record_size_bits) & ~PAGE_CACHE_MASK;
 	/* The maximum valid index into the page cache for $MFT's data. */
-	end_index = mft_vi->i_size >> PAGE_CACHE_SHIFT;
+	i_size = i_size_read(mft_vi);
+	end_index = i_size >> PAGE_CACHE_SHIFT;
 	if (unlikely(index >= end_index)) {
 		if (unlikely(index > end_index || ofs + vol->mft_record_size >=
-				(mft_vi->i_size & ~PAGE_CACHE_MASK))) {
+				(i_size & ~PAGE_CACHE_MASK))) {
 			ntfs_error(vol->sb, "Tried to format non-existing mft "
 					"record 0x%llx.", (long long)mft_no);
 			return -ENOENT;
@@ -2188,6 +2237,7 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 		ntfs_inode *base_ni, MFT_RECORD **mrec)
 {
 	s64 ll, bit, old_data_initialized, old_data_size;
+	unsigned long flags;
 	struct inode *vi;
 	struct page *page;
 	ntfs_inode *mft_ni, *mftbmp_ni, *ni;
@@ -2237,9 +2287,13 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 	 * the first 24 mft records as they are special and whilst they may not
 	 * be in use, we do not allocate from them.
 	 */
+	read_lock_irqsave(&mft_ni->size_lock, flags);
 	ll = mft_ni->initialized_size >> vol->mft_record_size_bits;
-	if (mftbmp_ni->initialized_size << 3 > ll &&
+	read_unlock_irqrestore(&mft_ni->size_lock, flags);
-			mftbmp_ni->initialized_size > 3) {
+	read_lock_irqsave(&mftbmp_ni->size_lock, flags);
+	old_data_initialized = mftbmp_ni->initialized_size;
+	read_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
+	if (old_data_initialized << 3 > ll && old_data_initialized > 3) {
 		bit = ll;
 		if (bit < 24)
 			bit = 24;
@@ -2254,15 +2308,18 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 	 * mft record that we can allocate.
 	 * Note: The smallest mft record we allocate is mft record 24.
 	 */
-	bit = mftbmp_ni->initialized_size << 3;
+	bit = old_data_initialized << 3;
 	if (unlikely(bit >= (1ll << 32)))
 		goto max_err_out;
+	read_lock_irqsave(&mftbmp_ni->size_lock, flags);
+	old_data_size = mftbmp_ni->allocated_size;
 	ntfs_debug("Status of mftbmp before extension: allocated_size 0x%llx, "
 			"data_size 0x%llx, initialized_size 0x%llx.",
-			(long long)mftbmp_ni->allocated_size,
+			(long long)old_data_size,
-			(long long)vol->mftbmp_ino->i_size,
+			(long long)i_size_read(vol->mftbmp_ino),
-			(long long)mftbmp_ni->initialized_size);
+			(long long)old_data_initialized);
-	if (mftbmp_ni->initialized_size + 8 > mftbmp_ni->allocated_size) {
+	read_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
+	if (old_data_initialized + 8 > old_data_size) {
 		/* Need to extend bitmap by one more cluster. */
 		ntfs_debug("mftbmp: initialized_size + 8 > allocated_size.");
 		err = ntfs_mft_bitmap_extend_allocation_nolock(vol);
@@ -2270,12 +2327,16 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 			up_write(&vol->mftbmp_lock);
 			goto err_out;
 		}
+#ifdef DEBUG
+		read_lock_irqsave(&mftbmp_ni->size_lock, flags);
 		ntfs_debug("Status of mftbmp after allocation extension: "
 				"allocated_size 0x%llx, data_size 0x%llx, "
 				"initialized_size 0x%llx.",
 				(long long)mftbmp_ni->allocated_size,
-				(long long)vol->mftbmp_ino->i_size,
+				(long long)i_size_read(vol->mftbmp_ino),
 				(long long)mftbmp_ni->initialized_size);
+		read_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
+#endif /* DEBUG */
 	}
 	/*
 	 * We now have sufficient allocated space, extend the initialized_size
@@ -2287,12 +2348,16 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 		up_write(&vol->mftbmp_lock);
 		goto err_out;
 	}
+#ifdef DEBUG
+	read_lock_irqsave(&mftbmp_ni->size_lock, flags);
 	ntfs_debug("Status of mftbmp after initialized extention: "
 			"allocated_size 0x%llx, data_size 0x%llx, "
 			"initialized_size 0x%llx.",
 			(long long)mftbmp_ni->allocated_size,
-			(long long)vol->mftbmp_ino->i_size,
+			(long long)i_size_read(vol->mftbmp_ino),
 			(long long)mftbmp_ni->initialized_size);
+	read_unlock_irqrestore(&mftbmp_ni->size_lock, flags);
+#endif /* DEBUG */
 	ntfs_debug("Found free record (#3), bit 0x%llx.", (long long)bit);
 found_free_rec:
 	/* @bit is the found free mft record, allocate it in the mft bitmap. */
@@ -2314,7 +2379,10 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 	 * parallel allocation could allocate the same mft record as this one.
 	 */
 	ll = (bit + 1) << vol->mft_record_size_bits;
-	if (ll <= mft_ni->initialized_size) {
+	read_lock_irqsave(&mft_ni->size_lock, flags);
+	old_data_initialized = mft_ni->initialized_size;
+	read_unlock_irqrestore(&mft_ni->size_lock, flags);
+	if (ll <= old_data_initialized) {
 		ntfs_debug("Allocated mft record already initialized.");
 		goto mft_rec_already_initialized;
 	}
@@ -2325,26 +2393,30 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 	 * actually traversed more than once when a freshly formatted volume is
 	 * first written to so it optimizes away nicely in the common case.
 	 */
+	read_lock_irqsave(&mft_ni->size_lock, flags);
 	ntfs_debug("Status of mft data before extension: "
 			"allocated_size 0x%llx, data_size 0x%llx, "
 			"initialized_size 0x%llx.",
 			(long long)mft_ni->allocated_size,
-			(long long)vol->mft_ino->i_size,
+			(long long)i_size_read(vol->mft_ino),
 			(long long)mft_ni->initialized_size);
 	while (ll > mft_ni->allocated_size) {
+		read_unlock_irqrestore(&mft_ni->size_lock, flags);
 		err = ntfs_mft_data_extend_allocation_nolock(vol);
 		if (unlikely(err)) {
 			ntfs_error(vol->sb, "Failed to extend mft data "
 					"allocation.");
 			goto undo_mftbmp_alloc_nolock;
 		}
+		read_lock_irqsave(&mft_ni->size_lock, flags);
 		ntfs_debug("Status of mft data after allocation extension: "
 				"allocated_size 0x%llx, data_size 0x%llx, "
 				"initialized_size 0x%llx.",
 				(long long)mft_ni->allocated_size,
-				(long long)vol->mft_ino->i_size,
+				(long long)i_size_read(vol->mft_ino),
 				(long long)mft_ni->initialized_size);
 	}
+	read_unlock_irqrestore(&mft_ni->size_lock, flags);
 	/*
 	 * Extend mft data initialized size (and data size of course) to reach
 	 * the allocated mft record, formatting the mft records allong the way.
@@ -2352,6 +2424,7 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 	 * needed by ntfs_mft_record_format().  We will update the attribute
 	 * record itself in one fell swoop later on.
 	 */
+	write_lock_irqsave(&mft_ni->size_lock, flags);
 	old_data_initialized = mft_ni->initialized_size;
 	old_data_size = vol->mft_ino->i_size;
 	while (ll > mft_ni->initialized_size) {
@@ -2360,8 +2433,9 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 		new_initialized_size = mft_ni->initialized_size +
 				vol->mft_record_size;
 		mft_no = mft_ni->initialized_size >> vol->mft_record_size_bits;
-		if (new_initialized_size > vol->mft_ino->i_size)
+		if (new_initialized_size > i_size_read(vol->mft_ino))
-			vol->mft_ino->i_size = new_initialized_size;
+			i_size_write(vol->mft_ino, new_initialized_size);
+		write_unlock_irqrestore(&mft_ni->size_lock, flags);
 		ntfs_debug("Initializing mft record 0x%llx.",
 				(long long)mft_no);
 		err = ntfs_mft_record_format(vol, mft_no);
@@ -2369,8 +2443,10 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 			ntfs_error(vol->sb, "Failed to format mft record.");
 			goto undo_data_init;
 		}
+		write_lock_irqsave(&mft_ni->size_lock, flags);
 		mft_ni->initialized_size = new_initialized_size;
 	}
+	write_unlock_irqrestore(&mft_ni->size_lock, flags);
 	record_formatted = TRUE;
 	/* Update the mft data attribute record to reflect the new sizes. */
 	m = map_mft_record(mft_ni);
@@ -2396,22 +2472,27 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 		goto undo_data_init;
 	}
 	a = ctx->attr;
+	read_lock_irqsave(&mft_ni->size_lock, flags);
 	a->data.non_resident.initialized_size =
 			cpu_to_sle64(mft_ni->initialized_size);
-	a->data.non_resident.data_size = cpu_to_sle64(vol->mft_ino->i_size);
+	a->data.non_resident.data_size =
+			cpu_to_sle64(i_size_read(vol->mft_ino));
+	read_unlock_irqrestore(&mft_ni->size_lock, flags);
 	/* Ensure the changes make it to disk. */
 	flush_dcache_mft_record_page(ctx->ntfs_ino);
 	mark_mft_record_dirty(ctx->ntfs_ino);
 	ntfs_attr_put_search_ctx(ctx);
 	unmap_mft_record(mft_ni);
+	read_lock_irqsave(&mft_ni->size_lock, flags);
 	ntfs_debug("Status of mft data after mft record initialization: "
 			"allocated_size 0x%llx, data_size 0x%llx, "
 			"initialized_size 0x%llx.",
 			(long long)mft_ni->allocated_size,
-			(long long)vol->mft_ino->i_size,
+			(long long)i_size_read(vol->mft_ino),
 			(long long)mft_ni->initialized_size);
-	BUG_ON(vol->mft_ino->i_size > mft_ni->allocated_size);
+	BUG_ON(i_size_read(vol->mft_ino) > mft_ni->allocated_size);
-	BUG_ON(mft_ni->initialized_size > vol->mft_ino->i_size);
+	BUG_ON(mft_ni->initialized_size > i_size_read(vol->mft_ino));
+	read_unlock_irqrestore(&mft_ni->size_lock, flags);
 mft_rec_already_initialized:
 	/*
 	 * We can finally drop the mft bitmap lock as the mft data attribute
@@ -2652,8 +2733,10 @@ ntfs_inode *ntfs_mft_record_alloc(ntfs_volume *vol, const int mode,
 	*mrec = m;
 	return ni;
 undo_data_init:
+	write_lock_irqsave(&mft_ni->size_lock, flags);
 	mft_ni->initialized_size = old_data_initialized;
-	vol->mft_ino->i_size = old_data_size;
+	i_size_write(vol->mft_ino, old_data_size);
+	write_unlock_irqrestore(&mft_ni->size_lock, flags);
 	goto undo_mftbmp_alloc_nolock;
 undo_mftbmp_alloc:
 	down_write(&vol->mftbmp_lock);

fs/ntfs/namei.c

@@ -153,8 +153,7 @@ static struct dentry *ntfs_lookup(struct inode *dir_ino, struct dentry *dent,
 			ntfs_error(vol->sb, "ntfs_iget(0x%lx) failed with "
 					"error code %li.", dent_ino,
 					PTR_ERR(dent_inode));
-		if (name)
+		kfree(name);
-			kfree(name);
 		/* Return the error code. */
 		return (struct dentry *)dent_inode;
 	}
@@ -380,7 +379,7 @@ struct inode_operations ntfs_dir_inode_ops = {
  * Return the dentry of the parent directory on success or the error code on
  * error (IS_ERR() is true).
  */
-struct dentry *ntfs_get_parent(struct dentry *child_dent)
+static struct dentry *ntfs_get_parent(struct dentry *child_dent)
 {
 	struct inode *vi = child_dent->d_inode;
 	ntfs_inode *ni = NTFS_I(vi);
@@ -465,7 +464,7 @@ struct dentry *ntfs_get_parent(struct dentry *child_dent)
  *
  * Return the dentry on success or the error code on error (IS_ERR() is true).
  */
-struct dentry *ntfs_get_dentry(struct super_block *sb, void *fh)
+static struct dentry *ntfs_get_dentry(struct super_block *sb, void *fh)
 {
 	struct inode *vi;
 	struct dentry *dent;
@@ -496,3 +495,30 @@ struct dentry *ntfs_get_dentry(struct super_block *sb, void *fh)
 	ntfs_debug("Done for inode 0x%lx, generation 0x%x.", ino, gen);
 	return dent;
 }
+
+/**
+ * Export operations allowing NFS exporting of mounted NTFS partitions.
+ *
+ * We use the default ->decode_fh() and ->encode_fh() for now.  Note that they
+ * use 32 bits to store the inode number which is an unsigned long so on 64-bit
+ * architectures is usually 64 bits so it would all fail horribly on huge
+ * volumes.  I guess we need to define our own encode and decode fh functions
+ * that store 64-bit inode numbers at some point but for now we will ignore the
+ * problem...
+ *
+ * We also use the default ->get_name() helper (used by ->decode_fh() via
+ * fs/exportfs/expfs.c::find_exported_dentry()) as that is completely fs
+ * independent.
+ *
+ * The default ->get_parent() just returns -EACCES so we have to provide our
+ * own and the default ->get_dentry() is incompatible with NTFS due to not
+ * allowing the inode number 0 which is used in NTFS for the system file $MFT
+ * and due to using iget() whereas NTFS needs ntfs_iget().
+ */
+struct export_operations ntfs_export_ops = {
+	.get_parent	= ntfs_get_parent,	/* Find the parent of a given
+						   directory. */
+	.get_dentry	= ntfs_get_dentry,	/* Find a dentry for the inode
+						   given a file handle
+						   sub-fragment. */
+};

fs/ntfs/ntfs.h

@@ -2,7 +2,7 @@
  * ntfs.h - Defines for NTFS Linux kernel driver. Part of the Linux-NTFS
  *	    project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (C) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -31,6 +31,7 @@
 #include <linux/fs.h>
 #include <linux/nls.h>
 #include <linux/smp.h>
+#include <linux/pagemap.h>
 
 #include "types.h"
 #include "volume.h"
@@ -41,6 +42,9 @@ typedef enum {
 	NTFS_BLOCK_SIZE_BITS	= 9,
 	NTFS_SB_MAGIC		= 0x5346544e,	/* 'NTFS' */
 	NTFS_MAX_NAME_LEN	= 255,
+	NTFS_MAX_ATTR_NAME_LEN	= 255,
+	NTFS_MAX_CLUSTER_SIZE	= 64 * 1024,	/* 64kiB */
+	NTFS_MAX_PAGES_PER_CLUSTER = NTFS_MAX_CLUSTER_SIZE / PAGE_CACHE_SIZE,
 } NTFS_CONSTANTS;
 
 /* Global variables. */
@@ -65,6 +69,8 @@ extern struct inode_operations ntfs_dir_inode_ops;
 extern struct  file_operations ntfs_empty_file_ops;
 extern struct inode_operations ntfs_empty_inode_ops;
 
+extern struct export_operations ntfs_export_ops;
+
 /**
  * NTFS_SB - return the ntfs volume given a vfs super block
  * @sb:		VFS super block

fs/ntfs/runlist.c

@@ -1,7 +1,7 @@
 /**
  * runlist.c - NTFS runlist handling code.  Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -59,7 +59,7 @@ static inline void ntfs_rl_mc(runlist_element *dstbase, int dst,
  *
  * As the runlists grow, more memory will be required.  To prevent the
  * kernel having to allocate and reallocate large numbers of small bits of
- * memory, this function returns and entire page of memory.
+ * memory, this function returns an entire page of memory.
  *
  * It is up to the caller to serialize access to the runlist @rl.
  *
@@ -113,8 +113,11 @@ static inline BOOL ntfs_are_rl_mergeable(runlist_element *dst,
 	BUG_ON(!dst);
 	BUG_ON(!src);
 
-	if ((dst->lcn < 0) || (src->lcn < 0))     /* Are we merging holes? */
+	if ((dst->lcn < 0) || (src->lcn < 0)) {   /* Are we merging holes? */
+		if (dst->lcn == LCN_HOLE && src->lcn == LCN_HOLE)
+			return TRUE;
 		return FALSE;
+	}
 	if ((dst->lcn + dst->length) != src->lcn) /* Are the runs contiguous? */
 		return FALSE;
 	if ((dst->vcn + dst->length) != src->vcn) /* Are the runs misaligned? */
@@ -855,30 +858,42 @@ runlist_element *ntfs_mapping_pairs_decompress(const ntfs_volume *vol,
 	if (!attr->data.non_resident.lowest_vcn) {
 		VCN max_cluster;
 
-		max_cluster = (sle64_to_cpu(
+		max_cluster = ((sle64_to_cpu(
 				attr->data.non_resident.allocated_size) +
 				vol->cluster_size - 1) >>
-				vol->cluster_size_bits;
+				vol->cluster_size_bits) - 1;
 		/*
-		 * If there is a difference between the highest_vcn and the
+		 * A highest_vcn of zero means this is a single extent
-		 * highest cluster, the runlist is either corrupt or, more
+		 * attribute so simply terminate the runlist with LCN_ENOENT).
-		 * likely, there are more extents following this one.
 		 */
-		if (deltaxcn < --max_cluster) {
+		if (deltaxcn) {
-			ntfs_debug("More extents to follow; deltaxcn = 0x%llx, "
+			/*
-					"max_cluster = 0x%llx",
+			 * If there is a difference between the highest_vcn and
-					(unsigned long long)deltaxcn,
+			 * the highest cluster, the runlist is either corrupt
-					(unsigned long long)max_cluster);
+			 * or, more likely, there are more extents following
-			rl[rlpos].vcn = vcn;
+			 * this one.
-			vcn += rl[rlpos].length = max_cluster - deltaxcn;
+			 */
-			rl[rlpos].lcn = LCN_RL_NOT_MAPPED;
+			if (deltaxcn < max_cluster) {
-			rlpos++;
+				ntfs_debug("More extents to follow; deltaxcn "
-		} else if (unlikely(deltaxcn > max_cluster)) {
+						"= 0x%llx, max_cluster = "
-			ntfs_error(vol->sb, "Corrupt attribute. deltaxcn = "
+						"0x%llx",
-					"0x%llx, max_cluster = 0x%llx",
+						(unsigned long long)deltaxcn,
-					(unsigned long long)deltaxcn,
+						(unsigned long long)
-					(unsigned long long)max_cluster);
+						max_cluster);
-			goto mpa_err;
+				rl[rlpos].vcn = vcn;
+				vcn += rl[rlpos].length = max_cluster -
+						deltaxcn;
+				rl[rlpos].lcn = LCN_RL_NOT_MAPPED;
+				rlpos++;
+			} else if (unlikely(deltaxcn > max_cluster)) {
+				ntfs_error(vol->sb, "Corrupt attribute.  "
+						"deltaxcn = 0x%llx, "
+						"max_cluster = 0x%llx",
+						(unsigned long long)deltaxcn,
+						(unsigned long long)
+						max_cluster);
+				goto mpa_err;
+			}
 		}
 		rl[rlpos].lcn = LCN_ENOENT;
 	} else /* Not the base extent. There may be more extents to follow. */
@@ -918,17 +933,18 @@ runlist_element *ntfs_mapping_pairs_decompress(const ntfs_volume *vol,
  *
  * It is up to the caller to serialize access to the runlist @rl.
  *
- * Since lcns must be >= 0, we use negative return values with special meaning:
+ * Since lcns must be >= 0, we use negative return codes with special meaning:
  *
- * Return value			Meaning / Description
+ * Return code		Meaning / Description
  * ==================================================
- *  -1 = LCN_HOLE		Hole / not allocated on disk.
+ *  LCN_HOLE		Hole / not allocated on disk.
- *  -2 = LCN_RL_NOT_MAPPED	This is part of the runlist which has not been
+ *  LCN_RL_NOT_MAPPED	This is part of the runlist which has not been
- *				inserted into the runlist yet.
+ *			inserted into the runlist yet.
- *  -3 = LCN_ENOENT		There is no such vcn in the attribute.
+ *  LCN_ENOENT		There is no such vcn in the attribute.
  *
  * Locking: - The caller must have locked the runlist (for reading or writing).
- *	    - This function does not touch the lock.
+ *	    - This function does not touch the lock, nor does it modify the
+ *	      runlist.
  */
 LCN ntfs_rl_vcn_to_lcn(const runlist_element *rl, const VCN vcn)
 {
@@ -964,6 +980,39 @@ LCN ntfs_rl_vcn_to_lcn(const runlist_element *rl, const VCN vcn)
 	return LCN_ENOENT;
 }
 
+#ifdef NTFS_RW
+
+/**
+ * ntfs_rl_find_vcn_nolock - find a vcn in a runlist
+ * @rl:		runlist to search
+ * @vcn:	vcn to find
+ *
+ * Find the virtual cluster number @vcn in the runlist @rl and return the
+ * address of the runlist element containing the @vcn on success.
+ *
+ * Return NULL if @rl is NULL or @vcn is in an unmapped part/out of bounds of
+ * the runlist.
+ *
+ * Locking: The runlist must be locked on entry.
+ */
+runlist_element *ntfs_rl_find_vcn_nolock(runlist_element *rl, const VCN vcn)
+{
+	BUG_ON(vcn < 0);
+	if (unlikely(!rl || vcn < rl[0].vcn))
+		return NULL;
+	while (likely(rl->length)) {
+		if (unlikely(vcn < rl[1].vcn)) {
+			if (likely(rl->lcn >= LCN_HOLE))
+				return rl;
+			return NULL;
+		}
+		rl++;
+	}
+	if (likely(rl->lcn == LCN_ENOENT))
+		return rl;
+	return NULL;
+}
+
 /**
  * ntfs_get_nr_significant_bytes - get number of bytes needed to store a number
  * @n:		number for which to get the number of bytes for
@@ -999,10 +1048,17 @@ static inline int ntfs_get_nr_significant_bytes(const s64 n)
  * ntfs_get_size_for_mapping_pairs - get bytes needed for mapping pairs array
  * @vol:	ntfs volume (needed for the ntfs version)
  * @rl:		locked runlist to determine the size of the mapping pairs of
- * @start_vcn:	vcn at which to start the mapping pairs array
+ * @first_vcn:	first vcn which to include in the mapping pairs array
+ * @last_vcn:	last vcn which to include in the mapping pairs array
  *
  * Walk the locked runlist @rl and calculate the size in bytes of the mapping
- * pairs array corresponding to the runlist @rl, starting at vcn @start_vcn.
+ * pairs array corresponding to the runlist @rl, starting at vcn @first_vcn and
+ * finishing with vcn @last_vcn.
+ *
+ * A @last_vcn of -1 means end of runlist and in that case the size of the
+ * mapping pairs array corresponding to the runlist starting at vcn @first_vcn
+ * and finishing at the end of the runlist is determined.
+ *
  * This for example allows us to allocate a buffer of the right size when
  * building the mapping pairs array.
  *
@@ -1018,34 +1074,50 @@ static inline int ntfs_get_nr_significant_bytes(const s64 n)
  *	    remains locked throughout, and is left locked upon return.
  */
 int ntfs_get_size_for_mapping_pairs(const ntfs_volume *vol,
-		const runlist_element *rl, const VCN start_vcn)
+		const runlist_element *rl, const VCN first_vcn,
+		const VCN last_vcn)
 {
 	LCN prev_lcn;
 	int rls;
+	BOOL the_end = FALSE;
 
-	BUG_ON(start_vcn < 0);
+	BUG_ON(first_vcn < 0);
+	BUG_ON(last_vcn < -1);
+	BUG_ON(last_vcn >= 0 && first_vcn > last_vcn);
 	if (!rl) {
-		BUG_ON(start_vcn);
+		BUG_ON(first_vcn);
+		BUG_ON(last_vcn > 0);
 		return 1;
 	}
-	/* Skip to runlist element containing @start_vcn. */
+	/* Skip to runlist element containing @first_vcn. */
-	while (rl->length && start_vcn >= rl[1].vcn)
+	while (rl->length && first_vcn >= rl[1].vcn)
 		rl++;
-	if ((!rl->length && start_vcn > rl->vcn) || start_vcn < rl->vcn)
+	if (unlikely((!rl->length && first_vcn > rl->vcn) ||
+			first_vcn < rl->vcn))
 		return -EINVAL;
 	prev_lcn = 0;
 	/* Always need the termining zero byte. */
 	rls = 1;
 	/* Do the first partial run if present. */
-	if (start_vcn > rl->vcn) {
+	if (first_vcn > rl->vcn) {
-		s64 delta;
+		s64 delta, length = rl->length;
 
 		/* We know rl->length != 0 already. */
-		if (rl->length < 0 || rl->lcn < LCN_HOLE)
+		if (unlikely(length < 0 || rl->lcn < LCN_HOLE))
 			goto err_out;
-		delta = start_vcn - rl->vcn;
+		/*
+		 * If @stop_vcn is given and finishes inside this run, cap the
+		 * run length.
+		 */
+		if (unlikely(last_vcn >= 0 && rl[1].vcn > last_vcn)) {
+			s64 s1 = last_vcn + 1;
+			if (unlikely(rl[1].vcn > s1))
+				length = s1 - rl->vcn;
+			the_end = TRUE;
+		}
+		delta = first_vcn - rl->vcn;
 		/* Header byte + length. */
-		rls += 1 + ntfs_get_nr_significant_bytes(rl->length - delta);
+		rls += 1 + ntfs_get_nr_significant_bytes(length - delta);
 		/*
 		 * If the logical cluster number (lcn) denotes a hole and we
 		 * are on NTFS 3.0+, we don't store it at all, i.e. we need
@@ -1053,9 +1125,9 @@ int ntfs_get_size_for_mapping_pairs(const ntfs_volume *vol,
 		 * Note: this assumes that on NTFS 1.2-, holes are stored with
 		 * an lcn of -1 and not a delta_lcn of -1 (unless both are -1).
 		 */
-		if (rl->lcn >= 0 || vol->major_ver < 3) {
+		if (likely(rl->lcn >= 0 || vol->major_ver < 3)) {
 			prev_lcn = rl->lcn;
-			if (rl->lcn >= 0)
+			if (likely(rl->lcn >= 0))
 				prev_lcn += delta;
 			/* Change in lcn. */
 			rls += ntfs_get_nr_significant_bytes(prev_lcn);
@@ -1064,11 +1136,23 @@ int ntfs_get_size_for_mapping_pairs(const ntfs_volume *vol,
 		rl++;
 	}
 	/* Do the full runs. */
-	for (; rl->length; rl++) {
+	for (; rl->length && !the_end; rl++) {
-		if (rl->length < 0 || rl->lcn < LCN_HOLE)
+		s64 length = rl->length;
+
+		if (unlikely(length < 0 || rl->lcn < LCN_HOLE))
 			goto err_out;
+		/*
+		 * If @stop_vcn is given and finishes inside this run, cap the
+		 * run length.
+		 */
+		if (unlikely(last_vcn >= 0 && rl[1].vcn > last_vcn)) {
+			s64 s1 = last_vcn + 1;
+			if (unlikely(rl[1].vcn > s1))
+				length = s1 - rl->vcn;
+			the_end = TRUE;
+		}
 		/* Header byte + length. */
-		rls += 1 + ntfs_get_nr_significant_bytes(rl->length);
+		rls += 1 + ntfs_get_nr_significant_bytes(length);
 		/*
 		 * If the logical cluster number (lcn) denotes a hole and we
 		 * are on NTFS 3.0+, we don't store it at all, i.e. we need
@@ -1076,7 +1160,7 @@ int ntfs_get_size_for_mapping_pairs(const ntfs_volume *vol,
 		 * Note: this assumes that on NTFS 1.2-, holes are stored with
 		 * an lcn of -1 and not a delta_lcn of -1 (unless both are -1).
 		 */
-		if (rl->lcn >= 0 || vol->major_ver < 3) {
+		if (likely(rl->lcn >= 0 || vol->major_ver < 3)) {
 			/* Change in lcn. */
 			rls += ntfs_get_nr_significant_bytes(rl->lcn -
 					prev_lcn);
@@ -1119,7 +1203,7 @@ static inline int ntfs_write_significant_bytes(s8 *dst, const s8 *dst_max,
 
 	i = 0;
 	do {
-		if (dst > dst_max)
+		if (unlikely(dst > dst_max))
 			goto err_out;
 		*dst++ = l & 0xffll;
 		l >>= 8;
@@ -1128,12 +1212,12 @@ static inline int ntfs_write_significant_bytes(s8 *dst, const s8 *dst_max,
 	j = (n >> 8 * (i - 1)) & 0xff;
 	/* If the sign bit is wrong, we need an extra byte. */
 	if (n < 0 && j >= 0) {
-		if (dst > dst_max)
+		if (unlikely(dst > dst_max))
 			goto err_out;
 		i++;
 		*dst = (s8)-1;
 	} else if (n > 0 && j < 0) {
-		if (dst > dst_max)
+		if (unlikely(dst > dst_max))
 			goto err_out;
 		i++;
 		*dst = (s8)0;
@@ -1149,13 +1233,18 @@ static inline int ntfs_write_significant_bytes(s8 *dst, const s8 *dst_max,
  * @dst:	destination buffer to which to write the mapping pairs array
  * @dst_len:	size of destination buffer @dst in bytes
  * @rl:		locked runlist for which to build the mapping pairs array
- * @start_vcn:	vcn at which to start the mapping pairs array
+ * @first_vcn:	first vcn which to include in the mapping pairs array
+ * @last_vcn:	last vcn which to include in the mapping pairs array
  * @stop_vcn:	first vcn outside destination buffer on success or -ENOSPC
  *
  * Create the mapping pairs array from the locked runlist @rl, starting at vcn
- * @start_vcn and save the array in @dst.  @dst_len is the size of @dst in
+ * @first_vcn and finishing with vcn @last_vcn and save the array in @dst.
- * bytes and it should be at least equal to the value obtained by calling
+ * @dst_len is the size of @dst in bytes and it should be at least equal to the
- * ntfs_get_size_for_mapping_pairs().
+ * value obtained by calling ntfs_get_size_for_mapping_pairs().
+ *
+ * A @last_vcn of -1 means end of runlist and in that case the mapping pairs
+ * array corresponding to the runlist starting at vcn @first_vcn and finishing
+ * at the end of the runlist is created.
  *
  * If @rl is NULL, just write a single terminator byte to @dst.
  *
@@ -1164,7 +1253,7 @@ static inline int ntfs_write_significant_bytes(s8 *dst, const s8 *dst_max,
  * been filled with all the mapping pairs that will fit, thus it can be treated
  * as partial success, in that a new attribute extent needs to be created or
  * the next extent has to be used and the mapping pairs build has to be
- * continued with @start_vcn set to *@stop_vcn.
+ * continued with @first_vcn set to *@stop_vcn.
  *
  * Return 0 on success and -errno on error.  The following error codes are
  * defined:
@@ -1178,27 +1267,32 @@ static inline int ntfs_write_significant_bytes(s8 *dst, const s8 *dst_max,
  */
 int ntfs_mapping_pairs_build(const ntfs_volume *vol, s8 *dst,
 		const int dst_len, const runlist_element *rl,
-		const VCN start_vcn, VCN *const stop_vcn)
+		const VCN first_vcn, const VCN last_vcn, VCN *const stop_vcn)
 {
 	LCN prev_lcn;
 	s8 *dst_max, *dst_next;
 	int err = -ENOSPC;
+	BOOL the_end = FALSE;
 	s8 len_len, lcn_len;
 
-	BUG_ON(start_vcn < 0);
+	BUG_ON(first_vcn < 0);
+	BUG_ON(last_vcn < -1);
+	BUG_ON(last_vcn >= 0 && first_vcn > last_vcn);
 	BUG_ON(dst_len < 1);
 	if (!rl) {
-		BUG_ON(start_vcn);
+		BUG_ON(first_vcn);
+		BUG_ON(last_vcn > 0);
 		if (stop_vcn)
 			*stop_vcn = 0;
 		/* Terminator byte. */
 		*dst = 0;
 		return 0;
 	}
-	/* Skip to runlist element containing @start_vcn. */
+	/* Skip to runlist element containing @first_vcn. */
-	while (rl->length && start_vcn >= rl[1].vcn)
+	while (rl->length && first_vcn >= rl[1].vcn)
 		rl++;
-	if ((!rl->length && start_vcn > rl->vcn) || start_vcn < rl->vcn)
+	if (unlikely((!rl->length && first_vcn > rl->vcn) ||
+			first_vcn < rl->vcn))
 		return -EINVAL;
 	/*
 	 * @dst_max is used for bounds checking in
@@ -1207,17 +1301,27 @@ int ntfs_mapping_pairs_build(const ntfs_volume *vol, s8 *dst,
 	dst_max = dst + dst_len - 1;
 	prev_lcn = 0;
 	/* Do the first partial run if present. */
-	if (start_vcn > rl->vcn) {
+	if (first_vcn > rl->vcn) {
-		s64 delta;
+		s64 delta, length = rl->length;
 
 		/* We know rl->length != 0 already. */
-		if (rl->length < 0 || rl->lcn < LCN_HOLE)
+		if (unlikely(length < 0 || rl->lcn < LCN_HOLE))
 			goto err_out;
-		delta = start_vcn - rl->vcn;
+		/*
+		 * If @stop_vcn is given and finishes inside this run, cap the
+		 * run length.
+		 */
+		if (unlikely(last_vcn >= 0 && rl[1].vcn > last_vcn)) {
+			s64 s1 = last_vcn + 1;
+			if (unlikely(rl[1].vcn > s1))
+				length = s1 - rl->vcn;
+			the_end = TRUE;
+		}
+		delta = first_vcn - rl->vcn;
 		/* Write length. */
 		len_len = ntfs_write_significant_bytes(dst + 1, dst_max,
-				rl->length - delta);
+				length - delta);
-		if (len_len < 0)
+		if (unlikely(len_len < 0))
 			goto size_err;
 		/*
 		 * If the logical cluster number (lcn) denotes a hole and we
@@ -1228,19 +1332,19 @@ int ntfs_mapping_pairs_build(const ntfs_volume *vol, s8 *dst,
 		 * case on NT4. - We assume that we just need to write the lcn
 		 * change until someone tells us otherwise... (AIA)
 		 */
-		if (rl->lcn >= 0 || vol->major_ver < 3) {
+		if (likely(rl->lcn >= 0 || vol->major_ver < 3)) {
 			prev_lcn = rl->lcn;
-			if (rl->lcn >= 0)
+			if (likely(rl->lcn >= 0))
 				prev_lcn += delta;
 			/* Write change in lcn. */
 			lcn_len = ntfs_write_significant_bytes(dst + 1 +
 					len_len, dst_max, prev_lcn);
-			if (lcn_len < 0)
+			if (unlikely(lcn_len < 0))
 				goto size_err;
 		} else
 			lcn_len = 0;
 		dst_next = dst + len_len + lcn_len + 1;
-		if (dst_next > dst_max)
+		if (unlikely(dst_next > dst_max))
 			goto size_err;
 		/* Update header byte. */
 		*dst = lcn_len << 4 | len_len;
@@ -1250,13 +1354,25 @@ int ntfs_mapping_pairs_build(const ntfs_volume *vol, s8 *dst,
 		rl++;
 	}
 	/* Do the full runs. */
-	for (; rl->length; rl++) {
+	for (; rl->length && !the_end; rl++) {
-		if (rl->length < 0 || rl->lcn < LCN_HOLE)
+		s64 length = rl->length;
+
+		if (unlikely(length < 0 || rl->lcn < LCN_HOLE))
 			goto err_out;
+		/*
+		 * If @stop_vcn is given and finishes inside this run, cap the
+		 * run length.
+		 */
+		if (unlikely(last_vcn >= 0 && rl[1].vcn > last_vcn)) {
+			s64 s1 = last_vcn + 1;
+			if (unlikely(rl[1].vcn > s1))
+				length = s1 - rl->vcn;
+			the_end = TRUE;
+		}
 		/* Write length. */
 		len_len = ntfs_write_significant_bytes(dst + 1, dst_max,
-				rl->length);
+				length);
-		if (len_len < 0)
+		if (unlikely(len_len < 0))
 			goto size_err;
 		/*
 		 * If the logical cluster number (lcn) denotes a hole and we
@@ -1267,17 +1383,17 @@ int ntfs_mapping_pairs_build(const ntfs_volume *vol, s8 *dst,
 		 * case on NT4. - We assume that we just need to write the lcn
 		 * change until someone tells us otherwise... (AIA)
 		 */
-		if (rl->lcn >= 0 || vol->major_ver < 3) {
+		if (likely(rl->lcn >= 0 || vol->major_ver < 3)) {
 			/* Write change in lcn. */
 			lcn_len = ntfs_write_significant_bytes(dst + 1 +
 					len_len, dst_max, rl->lcn - prev_lcn);
-			if (lcn_len < 0)
+			if (unlikely(lcn_len < 0))
 				goto size_err;
 			prev_lcn = rl->lcn;
 		} else
 			lcn_len = 0;
 		dst_next = dst + len_len + lcn_len + 1;
-		if (dst_next > dst_max)
+		if (unlikely(dst_next > dst_max))
 			goto size_err;
 		/* Update header byte. */
 		*dst = lcn_len << 4 | len_len;
@@ -1436,3 +1552,5 @@ int ntfs_rl_truncate_nolock(const ntfs_volume *vol, runlist *const runlist,
 	ntfs_debug("Done.");
 	return 0;
 }
+
+#endif /* NTFS_RW */

fs/ntfs/runlist.h

@@ -2,7 +2,7 @@
  * runlist.h - Defines for runlist handling in NTFS Linux kernel driver.
  *	       Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -66,6 +66,8 @@ typedef enum {
 	LCN_HOLE		= -1,	/* Keep this as highest value or die! */
 	LCN_RL_NOT_MAPPED	= -2,
 	LCN_ENOENT		= -3,
+	LCN_ENOMEM		= -4,
+	LCN_EIO			= -5,
 } LCN_SPECIAL_VALUES;
 
 extern runlist_element *ntfs_runlists_merge(runlist_element *drl,
@@ -76,14 +78,22 @@ extern runlist_element *ntfs_mapping_pairs_decompress(const ntfs_volume *vol,
 
 extern LCN ntfs_rl_vcn_to_lcn(const runlist_element *rl, const VCN vcn);
 
+#ifdef NTFS_RW
+
+extern runlist_element *ntfs_rl_find_vcn_nolock(runlist_element *rl,
+		const VCN vcn);
+
 extern int ntfs_get_size_for_mapping_pairs(const ntfs_volume *vol,
-		const runlist_element *rl, const VCN start_vcn);
+		const runlist_element *rl, const VCN first_vcn,
+		const VCN last_vcn);
 
 extern int ntfs_mapping_pairs_build(const ntfs_volume *vol, s8 *dst,
 		const int dst_len, const runlist_element *rl,
-		const VCN start_vcn, VCN *const stop_vcn);
+		const VCN first_vcn, const VCN last_vcn, VCN *const stop_vcn);
 
 extern int ntfs_rl_truncate_nolock(const ntfs_volume *vol,
 		runlist *const runlist, const s64 new_length);
 
+#endif /* NTFS_RW */
+
 #endif /* _LINUX_NTFS_RUNLIST_H */

fs/ntfs/sysctl.c

@@ -3,7 +3,7 @@
  *	      the Linux-NTFS project. Adapted from the old NTFS driver,
  *	      Copyright (C) 1997 Martin von Löwis, Régis Duchesne
  *
- * Copyright (c) 2002-2004 Anton Altaparmakov
+ * Copyright (c) 2002-2005 Anton Altaparmakov
  *
  * This program/include file is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License as published
@@ -67,7 +67,7 @@ int ntfs_sysctl(int add)
 			return -ENOMEM;
 #ifdef CONFIG_PROC_FS
 		/*
-		 * If the proc file system is in use and we are a module, need
+		 * If the proc filesystem is in use and we are a module, need
 		 * to set the owner of our proc entry to our module. In the
 		 * non-modular case, THIS_MODULE is NULL, so this is ok.
 		 */

fs/ntfs/time.h

@@ -1,7 +1,7 @@
 /*
  * time.h - NTFS time conversion functions.  Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  *
  * This program/include file is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License as published
@@ -87,7 +87,7 @@ static inline struct timespec ntfs2utc(const sle64 time)
 	struct timespec ts;
 
 	/* Subtract the NTFS time offset. */
-	s64 t = sle64_to_cpu(time) - NTFS_TIME_OFFSET;
+	u64 t = (u64)(sle64_to_cpu(time) - NTFS_TIME_OFFSET);
 	/*
 	 * Convert the time to 1-second intervals and the remainder to
 	 * 1-nano-second intervals.

fs/ntfs/types.h

@@ -2,7 +2,7 @@
  * types.h - Defines for NTFS Linux kernel driver specific types.
  *	     Part of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  *
  * This program/include file is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License as published
@@ -53,6 +53,14 @@ typedef sle64 leLCN;
 typedef s64 LSN;
 typedef sle64 leLSN;
 
+/*
+ * The NTFS transaction log $UsnJrnl uses usn which are signed 64-bit values.
+ * We define our own type USN, to allow for type checking and better code
+ * readability.
+ */
+typedef s64 USN;
+typedef sle64 leUSN;
+
 typedef enum {
 	FALSE = 0,
 	TRUE = 1

fs/ntfs/unistr.c

@@ -264,7 +264,7 @@ int ntfs_nlstoucs(const ntfs_volume *vol, const char *ins,
 
 	/* We don't trust outside sources. */
 	if (ins) {
-		ucs = (ntfschar*)kmem_cache_alloc(ntfs_name_cache, SLAB_NOFS);
+		ucs = kmem_cache_alloc(ntfs_name_cache, SLAB_NOFS);
 		if (ucs) {
 			for (i = o = 0; i < ins_len; i += wc_len) {
 				wc_len = nls->char2uni(ins + i, ins_len - i,

fs/ntfs/usnjrnl.c

@@ -0,0 +1,84 @@
+/*
+ * usnjrnl.h - NTFS kernel transaction log ($UsnJrnl) handling.  Part of the
+ *	       Linux-NTFS project.
+ *
+ * Copyright (c) 2005 Anton Altaparmakov
+ *
+ * This program/include file is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as published
+ * by the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program/include file is distributed in the hope that it will be
+ * useful, but WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program (in the main directory of the Linux-NTFS
+ * distribution in the file COPYING); if not, write to the Free Software
+ * Foundation,Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifdef NTFS_RW
+
+#include <linux/fs.h>
+#include <linux/highmem.h>
+#include <linux/mm.h>
+
+#include "aops.h"
+#include "debug.h"
+#include "endian.h"
+#include "time.h"
+#include "types.h"
+#include "usnjrnl.h"
+#include "volume.h"
+
+/**
+ * ntfs_stamp_usnjrnl - stamp the transaction log ($UsnJrnl) on an ntfs volume
+ * @vol:	ntfs volume on which to stamp the transaction log
+ *
+ * Stamp the transaction log ($UsnJrnl) on the ntfs volume @vol and return
+ * TRUE on success and FALSE on error.
+ *
+ * This function assumes that the transaction log has already been loaded and
+ * consistency checked by a call to fs/ntfs/super.c::load_and_init_usnjrnl().
+ */
+BOOL ntfs_stamp_usnjrnl(ntfs_volume *vol)
+{
+	ntfs_debug("Entering.");
+	if (likely(!NVolUsnJrnlStamped(vol))) {
+		sle64 stamp;
+		struct page *page;
+		USN_HEADER *uh;
+
+		page = ntfs_map_page(vol->usnjrnl_max_ino->i_mapping, 0);
+		if (IS_ERR(page)) {
+			ntfs_error(vol->sb, "Failed to read from "
+					"$UsnJrnl/$DATA/$Max attribute.");
+			return FALSE;
+		}
+		uh = (USN_HEADER*)page_address(page);
+		stamp = get_current_ntfs_time();
+		ntfs_debug("Stamping transaction log ($UsnJrnl): old "
+				"journal_id 0x%llx, old lowest_valid_usn "
+				"0x%llx, new journal_id 0x%llx, new "
+				"lowest_valid_usn 0x%llx.",
+				(long long)sle64_to_cpu(uh->journal_id),
+				(long long)sle64_to_cpu(uh->lowest_valid_usn),
+				(long long)sle64_to_cpu(stamp),
+				i_size_read(vol->usnjrnl_j_ino));
+		uh->lowest_valid_usn =
+				cpu_to_sle64(i_size_read(vol->usnjrnl_j_ino));
+		uh->journal_id = stamp;
+		flush_dcache_page(page);
+		set_page_dirty(page);
+		ntfs_unmap_page(page);
+		/* Set the flag so we do not have to do it again on remount. */
+		NVolSetUsnJrnlStamped(vol);
+	}
+	ntfs_debug("Done.");
+	return TRUE;
+}
+
+#endif /* NTFS_RW */

fs/ntfs/usnjrnl.h

@@ -0,0 +1,205 @@
+/*
+ * usnjrnl.h - Defines for NTFS kernel transaction log ($UsnJrnl) handling.
+ *	       Part of the Linux-NTFS project.
+ *
+ * Copyright (c) 2005 Anton Altaparmakov
+ *
+ * This program/include file is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as published
+ * by the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program/include file is distributed in the hope that it will be
+ * useful, but WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program (in the main directory of the Linux-NTFS
+ * distribution in the file COPYING); if not, write to the Free Software
+ * Foundation,Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LINUX_NTFS_USNJRNL_H
+#define _LINUX_NTFS_USNJRNL_H
+
+#ifdef NTFS_RW
+
+#include "types.h"
+#include "endian.h"
+#include "layout.h"
+#include "volume.h"
+
+/*
+ * Transaction log ($UsnJrnl) organization:
+ *
+ * The transaction log records whenever a file is modified in any way.  So for
+ * example it will record that file "blah" was written to at a particular time
+ * but not what was written.  If will record that a file was deleted or
+ * created, that a file was truncated, etc.  See below for all the reason
+ * codes used.
+ *
+ * The transaction log is in the $Extend directory which is in the root
+ * directory of each volume.  If it is not present it means transaction
+ * logging is disabled.  If it is present it means transaction logging is
+ * either enabled or in the process of being disabled in which case we can
+ * ignore it as it will go away as soon as Windows gets its hands on it.
+ *
+ * To determine whether the transaction logging is enabled or in the process
+ * of being disabled, need to check the volume flags in the
+ * $VOLUME_INFORMATION attribute in the $Volume system file (which is present
+ * in the root directory and has a fixed mft record number, see layout.h).
+ * If the flag VOLUME_DELETE_USN_UNDERWAY is set it means the transaction log
+ * is in the process of being disabled and if this flag is clear it means the
+ * transaction log is enabled.
+ *
+ * The transaction log consists of two parts; the $DATA/$Max attribute as well
+ * as the $DATA/$J attribute.  $Max is a header describing the transaction
+ * log whilst $J is the transaction log data itself as a sequence of variable
+ * sized USN_RECORDs (see below for all the structures).
+ *
+ * We do not care about transaction logging at this point in time but we still
+ * need to let windows know that the transaction log is out of date.  To do
+ * this we need to stamp the transaction log.  This involves setting the
+ * lowest_valid_usn field in the $DATA/$Max attribute to the usn to be used
+ * for the next added USN_RECORD to the $DATA/$J attribute as well as
+ * generating a new journal_id in $DATA/$Max.
+ *
+ * The journal_id is as of the current version (2.0) of the transaction log
+ * simply the 64-bit timestamp of when the journal was either created or last
+ * stamped.
+ *
+ * To determine the next usn there are two ways.  The first is to parse
+ * $DATA/$J and to find the last USN_RECORD in it and to add its record_length
+ * to its usn (which is the byte offset in the $DATA/$J attribute).  The
+ * second is simply to take the data size of the attribute.  Since the usns
+ * are simply byte offsets into $DATA/$J, this is exactly the next usn.  For
+ * obvious reasons we use the second method as it is much simpler and faster.
+ *
+ * As an aside, note that to actually disable the transaction log, one would
+ * need to set the VOLUME_DELETE_USN_UNDERWAY flag (see above), then go
+ * through all the mft records on the volume and set the usn field in their
+ * $STANDARD_INFORMATION attribute to zero.  Once that is done, one would need
+ * to delete the transaction log file, i.e. \$Extent\$UsnJrnl, and finally,
+ * one would need to clear the VOLUME_DELETE_USN_UNDERWAY flag.
+ *
+ * Note that if a volume is unmounted whilst the transaction log is being
+ * disabled, the process will continue the next time the volume is mounted.
+ * This is why we can safely mount read-write when we see a transaction log
+ * in the process of being deleted.
+ */
+
+/* Some $UsnJrnl related constants. */
+#define UsnJrnlMajorVer		2
+#define UsnJrnlMinorVer		0
+
+/*
+ * $DATA/$Max attribute.  This is (always?) resident and has a fixed size of
+ * 32 bytes.  It contains the header describing the transaction log.
+ */
+typedef struct {
+/*Ofs*/
+/*   0*/sle64 maximum_size;	/* The maximum on-disk size of the $DATA/$J
+				   attribute. */
+/*   8*/sle64 allocation_delta;	/* Number of bytes by which to increase the
+				   size of the $DATA/$J attribute. */
+/*0x10*/sle64 journal_id;	/* Current id of the transaction log. */
+/*0x18*/leUSN lowest_valid_usn;	/* Lowest valid usn in $DATA/$J for the
+				   current journal_id. */
+/* sizeof() = 32 (0x20) bytes */
+} __attribute__ ((__packed__)) USN_HEADER;
+
+/*
+ * Reason flags (32-bit).  Cumulative flags describing the change(s) to the
+ * file since it was last opened.  I think the names speak for themselves but
+ * if you disagree check out the descriptions in the Linux NTFS project NTFS
+ * documentation: http://linux-ntfs.sourceforge.net/ntfs/files/usnjrnl.html
+ */
+enum {
+	USN_REASON_DATA_OVERWRITE	= const_cpu_to_le32(0x00000001),
+	USN_REASON_DATA_EXTEND		= const_cpu_to_le32(0x00000002),
+	USN_REASON_DATA_TRUNCATION	= const_cpu_to_le32(0x00000004),
+	USN_REASON_NAMED_DATA_OVERWRITE	= const_cpu_to_le32(0x00000010),
+	USN_REASON_NAMED_DATA_EXTEND	= const_cpu_to_le32(0x00000020),
+	USN_REASON_NAMED_DATA_TRUNCATION= const_cpu_to_le32(0x00000040),
+	USN_REASON_FILE_CREATE		= const_cpu_to_le32(0x00000100),
+	USN_REASON_FILE_DELETE		= const_cpu_to_le32(0x00000200),
+	USN_REASON_EA_CHANGE		= const_cpu_to_le32(0x00000400),
+	USN_REASON_SECURITY_CHANGE	= const_cpu_to_le32(0x00000800),
+	USN_REASON_RENAME_OLD_NAME	= const_cpu_to_le32(0x00001000),
+	USN_REASON_RENAME_NEW_NAME	= const_cpu_to_le32(0x00002000),
+	USN_REASON_INDEXABLE_CHANGE	= const_cpu_to_le32(0x00004000),
+	USN_REASON_BASIC_INFO_CHANGE	= const_cpu_to_le32(0x00008000),
+	USN_REASON_HARD_LINK_CHANGE	= const_cpu_to_le32(0x00010000),
+	USN_REASON_COMPRESSION_CHANGE	= const_cpu_to_le32(0x00020000),
+	USN_REASON_ENCRYPTION_CHANGE	= const_cpu_to_le32(0x00040000),
+	USN_REASON_OBJECT_ID_CHANGE	= const_cpu_to_le32(0x00080000),
+	USN_REASON_REPARSE_POINT_CHANGE	= const_cpu_to_le32(0x00100000),
+	USN_REASON_STREAM_CHANGE	= const_cpu_to_le32(0x00200000),
+	USN_REASON_CLOSE		= const_cpu_to_le32(0x80000000),
+};
+
+typedef le32 USN_REASON_FLAGS;
+
+/*
+ * Source info flags (32-bit).  Information about the source of the change(s)
+ * to the file.  For detailed descriptions of what these mean, see the Linux
+ * NTFS project NTFS documentation:
+ *	http://linux-ntfs.sourceforge.net/ntfs/files/usnjrnl.html
+ */
+enum {
+	USN_SOURCE_DATA_MANAGEMENT	  = const_cpu_to_le32(0x00000001),
+	USN_SOURCE_AUXILIARY_DATA	  = const_cpu_to_le32(0x00000002),
+	USN_SOURCE_REPLICATION_MANAGEMENT = const_cpu_to_le32(0x00000004),
+};
+
+typedef le32 USN_SOURCE_INFO_FLAGS;
+
+/*
+ * $DATA/$J attribute.  This is always non-resident, is marked as sparse, and
+ * is of variabled size.  It consists of a sequence of variable size
+ * USN_RECORDS.  The minimum allocated_size is allocation_delta as
+ * specified in $DATA/$Max.  When the maximum_size specified in $DATA/$Max is
+ * exceeded by more than allocation_delta bytes, allocation_delta bytes are
+ * allocated and appended to the $DATA/$J attribute and an equal number of
+ * bytes at the beginning of the attribute are freed and made sparse.  Note the
+ * making sparse only happens at volume checkpoints and hence the actual
+ * $DATA/$J size can exceed maximum_size + allocation_delta temporarily.
+ */
+typedef struct {
+/*Ofs*/
+/*   0*/le32 length;		/* Byte size of this record (8-byte
+				   aligned). */
+/*   4*/le16 major_ver;		/* Major version of the transaction log used
+				   for this record. */
+/*   6*/le16 minor_ver;		/* Minor version of the transaction log used
+				   for this record. */
+/*   8*/leMFT_REF mft_reference;/* The mft reference of the file (or
+				   directory) described by this record. */
+/*0x10*/leMFT_REF parent_directory;/* The mft reference of the parent
+				   directory of the file described by this
+				   record. */
+/*0x18*/leUSN usn;		/* The usn of this record.  Equals the offset
+				   within the $DATA/$J attribute. */
+/*0x20*/sle64 time;		/* Time when this record was created. */
+/*0x28*/USN_REASON_FLAGS reason;/* Reason flags (see above). */
+/*0x2c*/USN_SOURCE_INFO_FLAGS source_info;/* Source info flags (see above). */
+/*0x30*/le32 security_id;	/* File security_id copied from
+				   $STANDARD_INFORMATION. */
+/*0x34*/FILE_ATTR_FLAGS file_attributes;	/* File attributes copied from
+				   $STANDARD_INFORMATION or $FILE_NAME (not
+				   sure which). */
+/*0x38*/le16 file_name_size;	/* Size of the file name in bytes. */
+/*0x3a*/le16 file_name_offset;	/* Offset to the file name in bytes from the
+				   start of this record. */
+/*0x3c*/ntfschar file_name[0];	/* Use when creating only.  When reading use
+				   file_name_offset to determine the location
+				   of the name. */
+/* sizeof() = 60 (0x3c) bytes */
+} __attribute__ ((__packed__)) USN_RECORD;
+
+extern BOOL ntfs_stamp_usnjrnl(ntfs_volume *vol);
+
+#endif /* NTFS_RW */
+
+#endif /* _LINUX_NTFS_USNJRNL_H */

fs/ntfs/volume.h

@@ -2,7 +2,7 @@
  * volume.h - Defines for volume structures in NTFS Linux kernel driver. Part
  *	      of the Linux-NTFS project.
  *
- * Copyright (c) 2001-2004 Anton Altaparmakov
+ * Copyright (c) 2001-2005 Anton Altaparmakov
  * Copyright (c) 2002 Richard Russon
  *
  * This program/include file is free software; you can redistribute it and/or
@@ -54,7 +54,7 @@ typedef struct {
 	mode_t dmask;			/* The mask for directory
 					   permissions. */
 	u8 mft_zone_multiplier;		/* Initial mft zone multiplier. */
-	u8 on_errors;			/* What to do on file system errors. */
+	u8 on_errors;			/* What to do on filesystem errors. */
 	/* NTFS bootsector provided information. */
 	u16 sector_size;		/* in bytes */
 	u8 sector_size_bits;		/* log2(sector_size) */
@@ -125,6 +125,10 @@ typedef struct {
 	/* $Quota stuff is NTFS3.0+ specific.  Unused/NULL otherwise. */
 	struct inode *quota_ino;	/* The VFS inode of $Quota. */
 	struct inode *quota_q_ino;	/* Attribute inode for $Quota/$Q. */
+	/* $UsnJrnl stuff is NTFS3.0+ specific.  Unused/NULL otherwise. */
+	struct inode *usnjrnl_ino;	/* The VFS inode of $UsnJrnl. */
+	struct inode *usnjrnl_max_ino;	/* Attribute inode for $UsnJrnl/$Max. */
+	struct inode *usnjrnl_j_ino;	/* Attribute inode for $UsnJrnl/$J. */
 #endif /* NTFS_RW */
 	struct nls_table *nls_map;
 } ntfs_volume;
@@ -141,6 +145,8 @@ typedef enum {
 				      file names in WIN32 namespace. */
 	NV_LogFileEmpty,	/* 1: $LogFile journal is empty. */
 	NV_QuotaOutOfDate,	/* 1: $Quota is out of date. */
+	NV_UsnJrnlStamped,	/* 1: $UsnJrnl has been stamped. */
+	NV_SparseEnabled,	/* 1: May create sparse files. */
 } ntfs_volume_flags;
 
 /*
@@ -167,5 +173,7 @@ NVOL_FNS(ShowSystemFiles)
 NVOL_FNS(CaseSensitive)
 NVOL_FNS(LogFileEmpty)
 NVOL_FNS(QuotaOutOfDate)
+NVOL_FNS(UsnJrnlStamped)
+NVOL_FNS(SparseEnabled)
 
 #endif /* _LINUX_NTFS_VOLUME_H */

include/asm-arm/arch-shark/io.h

@@ -21,38 +21,8 @@
  */
 #define __PORT_PCIO(x)	(!((x) & 0x80000000))
 
-/*
+#define __io(a)                 ((void __iomem *)(PCIO_BASE + (a)))
- * Dynamic IO functions - let the compiler
- * optimize the expressions
- */
-#define DECLARE_DYN_OUT(fnsuffix,instr)						\
-static inline void __out##fnsuffix (unsigned int value, unsigned int port)	\
-{										\
-	unsigned long temp;							\
-	__asm__ __volatile__(							\
-	"tst	%2, #0x80000000\n\t"						\
-	"mov	%0, %4\n\t"							\
-	"addeq	%0, %0, %3\n\t"							\
-	"str" instr "	%1, [%0, %2]	@ out" #fnsuffix			\
-	: "=&r" (temp)								\
-	: "r" (value), "r" (port), "Ir" (PCIO_BASE - IO_BASE), "Ir" (IO_BASE)	\
-	: "cc");								\
-}
 
-#define DECLARE_DYN_IN(sz,fnsuffix,instr)					\
-static inline unsigned sz __in##fnsuffix (unsigned int port)		\
-{										\
-	unsigned long temp, value;						\
-	__asm__ __volatile__(							\
-	"tst	%2, #0x80000000\n\t"						\
-	"mov	%0, %4\n\t"							\
-	"addeq	%0, %0, %3\n\t"							\
-	"ldr" instr "	%1, [%0, %2]	@ in" #fnsuffix				\
-	: "=&r" (temp), "=r" (value)						\
-	: "r" (port), "Ir" (PCIO_BASE - IO_BASE), "Ir" (IO_BASE)		\
-	: "cc");								\
-	return (unsigned sz)value;						\
-}
 
 static inline unsigned int __ioaddr (unsigned int port)			\
 {										\
@@ -62,123 +32,8 @@ static inline unsigned int __ioaddr (unsigned int port)			\
 		return (unsigned int)(IO_BASE + (port));			\
 }
 
-#define DECLARE_IO(sz,fnsuffix,instr)	\
-	DECLARE_DYN_OUT(fnsuffix,instr)	\
-	DECLARE_DYN_IN(sz,fnsuffix,instr)
-
-DECLARE_IO(char,b,"b")
-DECLARE_IO(short,w,"h")
-DECLARE_IO(long,l,"")
-
-#undef DECLARE_IO
-#undef DECLARE_DYN_OUT
-#undef DECLARE_DYN_IN
-
-/*
- * Constant address IO functions
- *
- * These have to be macros for the 'J' constraint to work -
- * +/-4096 immediate operand.
- */
-#define __outbc(value,port)							\
-({										\
-	if (__PORT_PCIO((port)))						\
-		__asm__ __volatile__(						\
-		"strb	%0, [%1, %2]		@ outbc"			\
-		: : "r" (value), "r" (PCIO_BASE), "Jr" (port));		\
-	else									\
-		__asm__ __volatile__(						\
-		"strb	%0, [%1, %2]		@ outbc"			\
-		: : "r" (value), "r" (IO_BASE), "r" (port));		\
-})
-
-#define __inbc(port)								\
-({										\
-	unsigned char result;                                                   \
-	if (__PORT_PCIO((port)))						\
-		__asm__ __volatile__(						\
-		"ldrb	%0, [%1, %2]		@ inbc"				\
-		: "=r" (result) : "r" (PCIO_BASE), "Jr" (port));		\
-	else									\
-		__asm__ __volatile__(						\
-		"ldrb	%0, [%1, %2]		@ inbc"				\
-		: "=r" (result) : "r" (IO_BASE), "r" (port));		\
-	result;									\
-})
-
-#define __outwc(value,port)							\
-({										\
-	unsigned long v = value;						\
-	if (__PORT_PCIO((port)))						\
-		__asm__ __volatile__(						\
-		"strh	%0, [%1, %2]		@ outwc"			\
-		: : "r" (v|v<<16), "r" (PCIO_BASE), "Jr" (port));	\
-	else									\
-		__asm__ __volatile__(						\
-		"strh	%0, [%1, %2]		@ outwc"			\
-		: : "r" (v|v<<16), "r" (IO_BASE), "r" (port));		\
-})
-
-#define __inwc(port)								\
-({										\
-	unsigned short result;							\
-	if (__PORT_PCIO((port)))						\
-		__asm__ __volatile__(						\
-		"ldrh	%0, [%1, %2]		@ inwc"				\
-		: "=r" (result) : "r" (PCIO_BASE), "Jr" (port));		\
-	else									\
-		__asm__ __volatile__(						\
-		"ldrh	%0, [%1, %2]		@ inwc"				\
-		: "=r" (result) : "r" (IO_BASE), "r" (port));		\
-	result & 0xffff;							\
-})
-
-#define __outlc(value,port)								\
-({										\
-	unsigned long v = value;						\
-	if (__PORT_PCIO((port)))						\
-		__asm__ __volatile__(						\
-		"str	%0, [%1, %2]		@ outlc"			\
-		: : "r" (v), "r" (PCIO_BASE), "Jr" (port));		\
-	else									\
-		__asm__ __volatile__(						\
-		"str	%0, [%1, %2]		@ outlc"			\
-		: : "r" (v), "r" (IO_BASE), "r" (port));			\
-})
-
-#define __inlc(port)								\
-({										\
-	unsigned long result;							\
-	if (__PORT_PCIO((port)))						\
-		__asm__ __volatile__(						\
-		"ldr	%0, [%1, %2]		@ inlc"				\
-		: "=r" (result) : "r" (PCIO_BASE), "Jr" (port));		\
-	else									\
-		__asm__ __volatile__(						\
-		"ldr	%0, [%1, %2]		@ inlc"				\
-		: "=r" (result) : "r" (IO_BASE), "r" (port));		\
-	result;									\
-})
-
-#define __ioaddrc(port)								\
-({										\
-	unsigned long addr;							\
-	if (__PORT_PCIO((port)))						\
-		addr = PCIO_BASE + (port);				\
-	else									\
-		addr = IO_BASE + (port);					\
-	addr;									\
-})
-
 #define __mem_pci(addr) (addr)
 
-#define inb(p)	 	(__builtin_constant_p((p)) ? __inbc(p)    : __inb(p))
-#define inw(p)	 	(__builtin_constant_p((p)) ? __inwc(p)    : __inw(p))
-#define inl(p)	 	(__builtin_constant_p((p)) ? __inlc(p)    : __inl(p))
-#define outb(v,p)	(__builtin_constant_p((p)) ? __outbc(v,p) : __outb(v,p))
-#define outw(v,p)	(__builtin_constant_p((p)) ? __outwc(v,p) : __outw(v,p))
-#define outl(v,p)	(__builtin_constant_p((p)) ? __outlc(v,p) : __outl(v,p))
-
 /*
  * Translated address IO functions
  *

include/linux/serial.h

@@ -174,9 +174,11 @@ struct serial_icounter_struct {
 
 
 #ifdef __KERNEL__
+#include <linux/compiler.h>
+
 /* Export to allow PCMCIA to use this - Dave Hinds */
-extern int register_serial(struct serial_struct *req);
+extern int __deprecated register_serial(struct serial_struct *req);
-extern void unregister_serial(int line);
+extern void __deprecated unregister_serial(int line);
 
 /* Allow architectures to override entries in serial8250_ports[] at run time: */
 struct uart_port;	/* forward declaration */

include/linux/serialP.h

@@ -19,7 +19,6 @@
  * For definitions of the flags field, see tty.h
  */
 
-#include <linux/version.h>
 #include <linux/config.h>
 #include <linux/termios.h>
 #include <linux/workqueue.h>

include/linux/serial_core.h

@@ -122,6 +122,7 @@
 #ifdef __KERNEL__
 
 #include <linux/config.h>
+#include <linux/compiler.h>
 #include <linux/interrupt.h>
 #include <linux/circ_buf.h>
 #include <linux/spinlock.h>
@@ -359,8 +360,8 @@ struct tty_driver *uart_console_device(struct console *co, int *index);
  */
 int uart_register_driver(struct uart_driver *uart);
 void uart_unregister_driver(struct uart_driver *uart);
-void uart_unregister_port(struct uart_driver *reg, int line);
+void __deprecated uart_unregister_port(struct uart_driver *reg, int line);
-int uart_register_port(struct uart_driver *reg, struct uart_port *port);
+int __deprecated uart_register_port(struct uart_driver *reg, struct uart_port *port);
 int uart_add_one_port(struct uart_driver *reg, struct uart_port *port);
 int uart_remove_one_port(struct uart_driver *reg, struct uart_port *port);
 int uart_match_port(struct uart_port *port1, struct uart_port *port2);