Merge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6: (48 commits)
  ieee1394: raw1394: arm functions slept in atomic context
  ieee1394: sbp2: enable auto spin-up for all SBP-2 devices
  MAINTAINERS: updates to IEEE 1394 subsystem maintainership
  ieee1394: ohci1394: check for errors in suspend or resume
  set power state of firewire host during suspend
  ieee1394: ohci1394: more obvious endianess handling
  ieee1394: ohci1394: fix endianess bug in debug message
  ieee1394: sbp2: don't prefer MODE SENSE 10
  ieee1394: nodemgr: grab class.subsys.rwsem in nodemgr_resume_ne
  ieee1394: nodemgr: fix rwsem recursion
  ieee1394: sbp2: more help in Kconfig
  ieee1394: sbp2: prevent rare deadlock in shutdown
  ieee1394: sbp2: update includes
  ieee1394: sbp2: better handling of transport errors
  ieee1394: sbp2: recheck node generation in sbp2_update
  ieee1394: sbp2: safer agent reset in error handlers
  ieee1394: sbp2: handle "sbp2util_node_write_no_wait failed"
  CONFIG_PM=n slim: drivers/ieee1394/ohci1394.c
  ieee1394: safer definition of empty macros
  video1394: add poll file operation support
  ...

29 files changed, 1200 insertions, 1133 deletions

diff --git a/drivers/ieee1394/Kconfig b/drivers/ieee1394/Kconfig
index 186737539cf5..2769e505f051 100644
--- a/drivers/ieee1394/Kconfig
+++ b/drivers/ieee1394/Kconfig
@@ -120,12 +120,19 @@ config IEEE1394_VIDEO1394
 	  this option only if you have an IEEE 1394 video device connected to
 	  an OHCI-1394 card.
 
+comment "SBP-2 support (for storage devices) requires SCSI"
+	depends on IEEE1394 && SCSI=n
+
 config IEEE1394_SBP2
 	tristate "SBP-2 support (Harddisks etc.)"
 	depends on IEEE1394 && SCSI && (PCI || BROKEN)
 	help
-	  This option enables you to use SBP-2 devices connected to your IEEE
-	  1394 bus.  SBP-2 devices include harddrives and DVD devices.
+	  This option enables you to use SBP-2 devices connected to an IEEE
+	  1394 bus.  SBP-2 devices include storage devices like harddisks and
+	  DVD drives, also some other FireWire devices like scanners.
+
+	  You should also enable support for disks, CD-ROMs, etc. in the SCSI
+	  configuration section.
 
 config IEEE1394_SBP2_PHYS_DMA
 	bool "Enable replacement for physical DMA in SBP2"
diff --git a/drivers/ieee1394/csr.c b/drivers/ieee1394/csr.c
index 149573db91c5..ab0c80f61b9d 100644
--- a/drivers/ieee1394/csr.c
+++ b/drivers/ieee1394/csr.c
@@ -17,11 +17,13 @@
  *
  */
 
-#include <linux/string.h>
+#include <linux/jiffies.h>
+#include <linux/kernel.h>
 #include <linux/module.h>
 #include <linux/moduleparam.h>
 #include <linux/param.h>
 #include <linux/spinlock.h>
+#include <linux/string.h>
 
 #include "csr1212.h"
 #include "ieee1394_types.h"
@@ -149,31 +151,18 @@ static void host_reset(struct hpsb_host *host)
 
 /*
  * HI == seconds (bits 0:2)
- * LO == fraction units of 1/8000 of a second, as per 1394 (bits 19:31)
- *
- * Convert to units and then to HZ, for comparison to jiffies.
- *
- * By default this will end up being 800 units, or 100ms (125usec per
- * unit).
+ * LO == fractions of a second in units of 125usec (bits 19:31)
  *
- * NOTE: The spec says 1/8000, but also says we can compute based on 1/8192
- * like CSR specifies. Should make our math less complex.
+ * Convert SPLIT_TIMEOUT to jiffies.
+ * The default and minimum as per 1394a-2000 clause 8.3.2.2.6 is 100ms.
  */
 static inline void calculate_expire(struct csr_control *csr)
 {
-	unsigned long units;
-
-	/* Take the seconds, and convert to units */
-	units = (unsigned long)(csr->split_timeout_hi & 0x07) << 13;
-
-	/* Add in the fractional units */
-	units += (unsigned long)(csr->split_timeout_lo >> 19);
-
-	/* Convert to jiffies */
-	csr->expire = (unsigned long)(units * HZ) >> 13UL;
+	unsigned long usecs =
+		(csr->split_timeout_hi & 0x07) * USEC_PER_SEC +
+		(csr->split_timeout_lo >> 19) * 125L;
 
-	/* Just to keep from rounding low */
-	csr->expire++;
+	csr->expire = usecs_to_jiffies(usecs > 100000L ? usecs : 100000L);
 
 	HPSB_VERBOSE("CSR: setting expire to %lu, HZ=%u", csr->expire, HZ);
 }
diff --git a/drivers/ieee1394/csr.h b/drivers/ieee1394/csr.h
index ea9aa4f53ab6..f11546550d84 100644
--- a/drivers/ieee1394/csr.h
+++ b/drivers/ieee1394/csr.h
@@ -1,75 +1,73 @@
-
 #ifndef _IEEE1394_CSR_H
 #define _IEEE1394_CSR_H
 
-#ifdef CONFIG_PREEMPT
-#include <linux/sched.h>
-#endif
+#include <linux/spinlock_types.h>
 
 #include "csr1212.h"
+#include "ieee1394_types.h"
 
-#define CSR_REGISTER_BASE  0xfffff0000000ULL
+#define CSR_REGISTER_BASE		0xfffff0000000ULL
 
 /* register offsets relative to CSR_REGISTER_BASE */
-#define CSR_STATE_CLEAR           0x0
-#define CSR_STATE_SET             0x4
-#define CSR_NODE_IDS              0x8
-#define CSR_RESET_START           0xc
-#define CSR_SPLIT_TIMEOUT_HI      0x18
-#define CSR_SPLIT_TIMEOUT_LO      0x1c
-#define CSR_CYCLE_TIME            0x200
-#define CSR_BUS_TIME              0x204
-#define CSR_BUSY_TIMEOUT          0x210
-#define CSR_BUS_MANAGER_ID        0x21c
-#define CSR_BANDWIDTH_AVAILABLE   0x220
-#define CSR_CHANNELS_AVAILABLE    0x224
-#define CSR_CHANNELS_AVAILABLE_HI 0x224
-#define CSR_CHANNELS_AVAILABLE_LO 0x228
-#define CSR_BROADCAST_CHANNEL     0x234
-#define CSR_CONFIG_ROM            0x400
-#define CSR_CONFIG_ROM_END        0x800
-#define CSR_FCP_COMMAND           0xB00
-#define CSR_FCP_RESPONSE          0xD00
-#define CSR_FCP_END               0xF00
-#define CSR_TOPOLOGY_MAP          0x1000
-#define CSR_TOPOLOGY_MAP_END      0x1400
-#define CSR_SPEED_MAP             0x2000
-#define CSR_SPEED_MAP_END         0x3000
+#define CSR_STATE_CLEAR			0x0
+#define CSR_STATE_SET			0x4
+#define CSR_NODE_IDS			0x8
+#define CSR_RESET_START			0xc
+#define CSR_SPLIT_TIMEOUT_HI		0x18
+#define CSR_SPLIT_TIMEOUT_LO		0x1c
+#define CSR_CYCLE_TIME			0x200
+#define CSR_BUS_TIME			0x204
+#define CSR_BUSY_TIMEOUT		0x210
+#define CSR_BUS_MANAGER_ID		0x21c
+#define CSR_BANDWIDTH_AVAILABLE		0x220
+#define CSR_CHANNELS_AVAILABLE		0x224
+#define CSR_CHANNELS_AVAILABLE_HI	0x224
+#define CSR_CHANNELS_AVAILABLE_LO	0x228
+#define CSR_BROADCAST_CHANNEL		0x234
+#define CSR_CONFIG_ROM			0x400
+#define CSR_CONFIG_ROM_END		0x800
+#define CSR_FCP_COMMAND			0xB00
+#define CSR_FCP_RESPONSE		0xD00
+#define CSR_FCP_END			0xF00
+#define CSR_TOPOLOGY_MAP		0x1000
+#define CSR_TOPOLOGY_MAP_END		0x1400
+#define CSR_SPEED_MAP			0x2000
+#define CSR_SPEED_MAP_END		0x3000
 
 /* IEEE 1394 bus specific Configuration ROM Key IDs */
 #define IEEE1394_KV_ID_POWER_REQUIREMENTS (0x30)
 
-/* IEEE 1394 Bus Inforamation Block specifics */
+/* IEEE 1394 Bus Information Block specifics */
 #define CSR_BUS_INFO_SIZE (5 * sizeof(quadlet_t))
 
-#define CSR_IRMC_SHIFT 31
-#define CSR_CMC_SHIFT  30
-#define CSR_ISC_SHIFT  29
-#define CSR_BMC_SHIFT  28
-#define CSR_PMC_SHIFT  27
-#define CSR_CYC_CLK_ACC_SHIFT 16
-#define CSR_MAX_REC_SHIFT 12
-#define CSR_MAX_ROM_SHIFT 8
-#define CSR_GENERATION_SHIFT 4
+#define CSR_IRMC_SHIFT			31
+#define CSR_CMC_SHIFT			30
+#define CSR_ISC_SHIFT			29
+#define CSR_BMC_SHIFT			28
+#define CSR_PMC_SHIFT			27
+#define CSR_CYC_CLK_ACC_SHIFT		16
+#define CSR_MAX_REC_SHIFT		12
+#define CSR_MAX_ROM_SHIFT		8
+#define CSR_GENERATION_SHIFT		4
 
 #define CSR_SET_BUS_INFO_GENERATION(csr, gen)				\
 	((csr)->bus_info_data[2] =					\
 		cpu_to_be32((be32_to_cpu((csr)->bus_info_data[2]) &	\
-			     ~(0xf << CSR_GENERATION_SHIFT)) |          \
+			     ~(0xf << CSR_GENERATION_SHIFT)) |		\
 			    (gen) << CSR_GENERATION_SHIFT))
 
 struct csr_control {
-        spinlock_t lock;
-
-        quadlet_t state;
-        quadlet_t node_ids;
-        quadlet_t split_timeout_hi, split_timeout_lo;
-	unsigned long expire;	// Calculated from split_timeout
-        quadlet_t cycle_time;
-        quadlet_t bus_time;
-        quadlet_t bus_manager_id;
-        quadlet_t bandwidth_available;
-        quadlet_t channels_available_hi, channels_available_lo;
+	spinlock_t lock;
+
+	quadlet_t state;
+	quadlet_t node_ids;
+	quadlet_t split_timeout_hi, split_timeout_lo;
+	unsigned long expire;	/* Calculated from split_timeout */
+	quadlet_t cycle_time;
+	quadlet_t bus_time;
+	quadlet_t bus_manager_id;
+	quadlet_t bandwidth_available;
+	quadlet_t channels_available_hi, channels_available_lo;
 	quadlet_t broadcast_channel;
 
 	/* Bus Info */
@@ -84,8 +82,8 @@ struct csr_control {
 
 	struct csr1212_csr *rom;
 
-        quadlet_t topology_map[256];
-        quadlet_t speed_map[1024];
+	quadlet_t topology_map[256];
+	quadlet_t speed_map[1024];
 };
 
 extern struct csr1212_bus_ops csr_bus_ops;
@@ -93,4 +91,9 @@ extern struct csr1212_bus_ops csr_bus_ops;
 int init_csr(void);
 void cleanup_csr(void);
 
+/* hpsb_update_config_rom() is deprecated */
+struct hpsb_host;
+int hpsb_update_config_rom(struct hpsb_host *host, const quadlet_t *new_rom,
+			   size_t size, unsigned char rom_version);
+
 #endif /* _IEEE1394_CSR_H */
diff --git a/drivers/ieee1394/dma.c b/drivers/ieee1394/dma.c
index ca5167de707d..c68f328e1a29 100644
--- a/drivers/ieee1394/dma.c
+++ b/drivers/ieee1394/dma.c
@@ -7,10 +7,13 @@
  * directory of the kernel sources for details.
  */
 
+#include <linux/mm.h>
 #include <linux/module.h>
-#include <linux/vmalloc.h>
+#include <linux/pci.h>
 #include <linux/slab.h>
-#include <linux/mm.h>
+#include <linux/vmalloc.h>
+#include <asm/scatterlist.h>
+
 #include "dma.h"
 
 /* dma_prog_region */
diff --git a/drivers/ieee1394/dma.h b/drivers/ieee1394/dma.h
index 061550a6fb99..a1682aba71c7 100644
--- a/drivers/ieee1394/dma.h
+++ b/drivers/ieee1394/dma.h
@@ -10,69 +10,91 @@
 #ifndef IEEE1394_DMA_H
 #define IEEE1394_DMA_H
 
-#include <linux/pci.h>
-#include <asm/scatterlist.h>
-
-/* struct dma_prog_region
-
-   a small, physically-contiguous DMA buffer with random-access,
-   synchronous usage characteristics
-*/
-
+#include <asm/types.h>
+
+struct pci_dev;
+struct scatterlist;
+struct vm_area_struct;
+
+/**
+ * struct dma_prog_region - small contiguous DMA buffer
+ * @kvirt:    kernel virtual address
+ * @dev:      PCI device
+ * @n_pages:  number of kernel pages
+ * @bus_addr: base bus address
+ *
+ * a small, physically contiguous DMA buffer with random-access, synchronous
+ * usage characteristics
+ */
 struct dma_prog_region {
-	unsigned char    *kvirt;     /* kernel virtual address */
-	struct pci_dev   *dev;       /* PCI device */
-	unsigned int      n_pages;   /* # of kernel pages */
-	dma_addr_t        bus_addr;  /* base bus address */
+	unsigned char *kvirt;
+	struct pci_dev *dev;
+	unsigned int n_pages;
+	dma_addr_t bus_addr;
 };
 
 /* clear out all fields but do not allocate any memory */
 void dma_prog_region_init(struct dma_prog_region *prog);
-int  dma_prog_region_alloc(struct dma_prog_region *prog, unsigned long n_bytes, struct pci_dev *dev);
+int dma_prog_region_alloc(struct dma_prog_region *prog, unsigned long n_bytes,
+			  struct pci_dev *dev);
 void dma_prog_region_free(struct dma_prog_region *prog);
 
-static inline dma_addr_t dma_prog_region_offset_to_bus(struct dma_prog_region *prog, unsigned long offset)
+static inline dma_addr_t dma_prog_region_offset_to_bus(
+		struct dma_prog_region *prog, unsigned long offset)
 {
 	return prog->bus_addr + offset;
 }
 
-/* struct dma_region
-
-   a large, non-physically-contiguous DMA buffer with streaming,
-   asynchronous usage characteristics
-*/
-
+/**
+ * struct dma_region - large non-contiguous DMA buffer
+ * @virt:        kernel virtual address
+ * @dev:         PCI device
+ * @n_pages:     number of kernel pages
+ * @n_dma_pages: number of IOMMU pages
+ * @sglist:      IOMMU mapping
+ * @direction:   PCI_DMA_TODEVICE, etc.
+ *
+ * a large, non-physically-contiguous DMA buffer with streaming, asynchronous
+ * usage characteristics
+ */
 struct dma_region {
-	unsigned char      *kvirt;       /* kernel virtual address */
-	struct pci_dev     *dev;         /* PCI device */
-	unsigned int        n_pages;     /* # of kernel pages */
-	unsigned int        n_dma_pages; /* # of IOMMU pages */
-	struct scatterlist *sglist;      /* IOMMU mapping */
-	int                 direction;   /* PCI_DMA_TODEVICE, etc */
+	unsigned char *kvirt;
+	struct pci_dev *dev;
+	unsigned int n_pages;
+	unsigned int n_dma_pages;
+	struct scatterlist *sglist;
+	int direction;
 };
 
 /* clear out all fields but do not allocate anything */
 void dma_region_init(struct dma_region *dma);
 
 /* allocate the buffer and map it to the IOMMU */
-int  dma_region_alloc(struct dma_region *dma, unsigned long n_bytes, struct pci_dev *dev, int direction);
+int dma_region_alloc(struct dma_region *dma, unsigned long n_bytes,
+		     struct pci_dev *dev, int direction);
 
 /* unmap and free the buffer */
 void dma_region_free(struct dma_region *dma);
 
 /* sync the CPU's view of the buffer */
-void dma_region_sync_for_cpu(struct dma_region *dma, unsigned long offset, unsigned long len);
+void dma_region_sync_for_cpu(struct dma_region *dma, unsigned long offset,
+			     unsigned long len);
+
 /* sync the IO bus' view of the buffer */
-void dma_region_sync_for_device(struct dma_region *dma, unsigned long offset, unsigned long len);
+void dma_region_sync_for_device(struct dma_region *dma, unsigned long offset,
+				unsigned long len);
 
 /* map the buffer into a user space process */
-int  dma_region_mmap(struct dma_region *dma, struct file *file, struct vm_area_struct *vma);
+int  dma_region_mmap(struct dma_region *dma, struct file *file,
+		     struct vm_area_struct *vma);
 
 /* macro to index into a DMA region (or dma_prog_region) */
-#define dma_region_i(_dma, _type, _index) ( ((_type*) ((_dma)->kvirt)) + (_index) )
+#define dma_region_i(_dma, _type, _index) \
+	( ((_type*) ((_dma)->kvirt)) + (_index) )
 
 /* return the DMA bus address of the byte with the given offset
-   relative to the beginning of the dma_region */
-dma_addr_t dma_region_offset_to_bus(struct dma_region *dma, unsigned long offset);
+ * relative to the beginning of the dma_region */
+dma_addr_t dma_region_offset_to_bus(struct dma_region *dma,
+				    unsigned long offset);
 
 #endif /* IEEE1394_DMA_H */
diff --git a/drivers/ieee1394/dv1394-private.h b/drivers/ieee1394/dv1394-private.h
index 80b5ac7fe383..7d1d2845b420 100644
--- a/drivers/ieee1394/dv1394-private.h
+++ b/drivers/ieee1394/dv1394-private.h
@@ -460,7 +460,7 @@ struct video_card {
 	int dma_running;
 
 	/*
-	  3) the sleeping semaphore 'sem' - this is used from process context only,
+	  3) the sleeping mutex 'mtx' - this is used from process context only,
 	  to serialize various operations on the video_card. Even though only one
 	  open() is allowed, we still need to prevent multiple threads of execution
 	  from entering calls like read, write, ioctl, etc.
@@ -468,9 +468,9 @@ struct video_card {
 	  I honestly can't think of a good reason to use dv1394 from several threads
 	  at once, but we need to serialize anyway to prevent oopses =).
 
-	  NOTE: if you need both spinlock and sem, take sem first to avoid deadlock!
+	  NOTE: if you need both spinlock and mtx, take mtx first to avoid deadlock!
 	 */
-	struct semaphore sem;
+	struct mutex mtx;
 
 	/* people waiting for buffer space, please form a line here... */
 	wait_queue_head_t waitq;
diff --git a/drivers/ieee1394/dv1394.c b/drivers/ieee1394/dv1394.c
index 87532dd43374..6c72f04b2b5d 100644
--- a/drivers/ieee1394/dv1394.c
+++ b/drivers/ieee1394/dv1394.c
@@ -95,6 +95,7 @@
 #include <linux/fs.h>
 #include <linux/poll.h>
 #include <linux/smp_lock.h>
+#include <linux/mutex.h>
 #include <linux/bitops.h>
 #include <asm/byteorder.h>
 #include <asm/atomic.h>
@@ -110,15 +111,15 @@
 #include <linux/compat.h>
 #include <linux/cdev.h>
 
+#include "dv1394.h"
+#include "dv1394-private.h"
+#include "highlevel.h"
+#include "hosts.h"
 #include "ieee1394.h"
+#include "ieee1394_core.h"
+#include "ieee1394_hotplug.h"
 #include "ieee1394_types.h"
 #include "nodemgr.h"
-#include "hosts.h"
-#include "ieee1394_core.h"
-#include "highlevel.h"
-#include "dv1394.h"
-#include "dv1394-private.h"
-
 #include "ohci1394.h"
 
 /* DEBUG LEVELS:
@@ -136,13 +137,13 @@
 #if DV1394_DEBUG_LEVEL >= 2
 #define irq_printk( args... ) printk( args )
 #else
-#define irq_printk( args... )
+#define irq_printk( args... ) do {} while (0)
 #endif
 
 #if DV1394_DEBUG_LEVEL >= 1
 #define debug_printk( args... ) printk( args)
 #else
-#define debug_printk( args... )
+#define debug_printk( args... ) do {} while (0)
 #endif
 
 /* issue a dummy PCI read to force the preceding write
@@ -247,7 +248,7 @@ static void frame_delete(struct frame *f)
 
    Frame_prepare() must be called OUTSIDE the video->spinlock.
    However, frame_prepare() must still be serialized, so
-   it should be called WITH the video->sem taken.
+   it should be called WITH the video->mtx taken.
  */
 
 static void frame_prepare(struct video_card *video, unsigned int this_frame)
@@ -1271,7 +1272,7 @@ static int dv1394_mmap(struct file *file, struct vm_area_struct *vma)
 	int retval = -EINVAL;
 
 	/* serialize mmap */
-	down(&video->sem);
+	mutex_lock(&video->mtx);
 
 	if ( ! video_card_initialized(video) ) {
 		retval = do_dv1394_init_default(video);
@@ -1281,7 +1282,7 @@ static int dv1394_mmap(struct file *file, struct vm_area_struct *vma)
 
 	retval = dma_region_mmap(&video->dv_buf, file, vma);
 out:
-	up(&video->sem);
+	mutex_unlock(&video->mtx);
 	return retval;
 }
 
@@ -1337,17 +1338,17 @@ static ssize_t dv1394_write(struct file *file, const char __user *buffer, size_t
 
 	/* serialize this to prevent multi-threaded mayhem */
 	if (file->f_flags & O_NONBLOCK) {
-		if (down_trylock(&video->sem))
+		if (!mutex_trylock(&video->mtx))
 			return -EAGAIN;
 	} else {
-		if (down_interruptible(&video->sem))
+		if (mutex_lock_interruptible(&video->mtx))
 			return -ERESTARTSYS;
 	}
 
 	if ( !video_card_initialized(video) ) {
 		ret = do_dv1394_init_default(video);
 		if (ret) {
-			up(&video->sem);
+			mutex_unlock(&video->mtx);
 			return ret;
 		}
 	}
@@ -1418,7 +1419,7 @@ static ssize_t dv1394_write(struct file *file, const char __user *buffer, size_t
 
 	remove_wait_queue(&video->waitq, &wait);
 	set_current_state(TASK_RUNNING);
-	up(&video->sem);
+	mutex_unlock(&video->mtx);
 	return ret;
 }
 
@@ -1434,17 +1435,17 @@ static ssize_t dv1394_read(struct file *file,  char __user *buffer, size_t count
 
 	/* serialize this to prevent multi-threaded mayhem */
 	if (file->f_flags & O_NONBLOCK) {
-		if (down_trylock(&video->sem))
+		if (!mutex_trylock(&video->mtx))
 			return -EAGAIN;
 	} else {
-		if (down_interruptible(&video->sem))
+		if (mutex_lock_interruptible(&video->mtx))
 			return -ERESTARTSYS;
 	}
 
 	if ( !video_card_initialized(video) ) {
 		ret = do_dv1394_init_default(video);
 		if (ret) {
-			up(&video->sem);
+			mutex_unlock(&video->mtx);
 			return ret;
 		}
 		video->continuity_counter = -1;
@@ -1526,7 +1527,7 @@ static ssize_t dv1394_read(struct file *file,  char __user *buffer, size_t count
 
 	remove_wait_queue(&video->waitq, &wait);
 	set_current_state(TASK_RUNNING);
-	up(&video->sem);
+	mutex_unlock(&video->mtx);
 	return ret;
 }
 
@@ -1547,12 +1548,12 @@ static long dv1394_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 
 	/* serialize this to prevent multi-threaded mayhem */
 	if (file->f_flags & O_NONBLOCK) {
-		if (down_trylock(&video->sem)) {
+		if (!mutex_trylock(&video->mtx)) {
 			unlock_kernel();
 			return -EAGAIN;
 		}
 	} else {
-		if (down_interruptible(&video->sem)) {
+		if (mutex_lock_interruptible(&video->mtx)) {
 			unlock_kernel();
 			return -ERESTARTSYS;
 		}
@@ -1778,7 +1779,7 @@ static long dv1394_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 	}
 
  out:
-	up(&video->sem);
+	mutex_unlock(&video->mtx);
 	unlock_kernel();
 	return ret;
 }
@@ -2253,7 +2254,7 @@ static int dv1394_init(struct ti_ohci *ohci, enum pal_or_ntsc format, enum modes
 	clear_bit(0, &video->open);
 	spin_lock_init(&video->spinlock);
 	video->dma_running = 0;
-	init_MUTEX(&video->sem);
+	mutex_init(&video->mtx);
 	init_waitqueue_head(&video->waitq);
 	video->fasync = NULL;
 
diff --git a/drivers/ieee1394/eth1394.c b/drivers/ieee1394/eth1394.c
index 2d5b57be98c3..8a7b8fab6238 100644
--- a/drivers/ieee1394/eth1394.c
+++ b/drivers/ieee1394/eth1394.c
@@ -64,19 +64,19 @@
 #include <linux/ethtool.h>
 #include <asm/uaccess.h>
 #include <asm/delay.h>
-#include <asm/semaphore.h>
 #include <net/arp.h>
 
+#include "config_roms.h"
 #include "csr1212.h"
-#include "ieee1394_types.h"
+#include "eth1394.h"
+#include "highlevel.h"
+#include "ieee1394.h"
 #include "ieee1394_core.h"
+#include "ieee1394_hotplug.h"
 #include "ieee1394_transactions.h"
-#include "ieee1394.h"
-#include "highlevel.h"
+#include "ieee1394_types.h"
 #include "iso.h"
 #include "nodemgr.h"
-#include "eth1394.h"
-#include "config_roms.h"
 
 #define ETH1394_PRINT_G(level, fmt, args...) \
 	printk(level "%s: " fmt, driver_name, ## args)
diff --git a/drivers/ieee1394/highlevel.h b/drivers/ieee1394/highlevel.h
index e119fb87e5b5..50f2dd2c7e20 100644
--- a/drivers/ieee1394/highlevel.h
+++ b/drivers/ieee1394/highlevel.h
@@ -1,60 +1,61 @@
-
 #ifndef IEEE1394_HIGHLEVEL_H
 #define IEEE1394_HIGHLEVEL_H
 
+#include <linux/list.h>
+#include <linux/spinlock_types.h>
+#include <linux/types.h>
 
-struct hpsb_address_serve {
-        struct list_head host_list; /* per host list */
+struct module;
 
-        struct list_head hl_list; /* hpsb_highlevel list */
+#include "ieee1394_types.h"
 
-        struct hpsb_address_ops *op;
+struct hpsb_host;
 
+/* internal to ieee1394 core */
+struct hpsb_address_serve {
+	struct list_head host_list;	/* per host list */
+	struct list_head hl_list;	/* hpsb_highlevel list */
+	struct hpsb_address_ops *op;
 	struct hpsb_host *host;
-
-        /* first address handled and first address behind, quadlet aligned */
-        u64 start, end;
+	u64 start;	/* first address handled, quadlet aligned */
+	u64 end;	/* first address behind, quadlet aligned */
 };
 
-
-/*
- * The above structs are internal to highlevel driver handling.  Only the
- * following structures are of interest to actual highlevel drivers.
- */
+/* Only the following structures are of interest to actual highlevel drivers. */
 
 struct hpsb_highlevel {
 	struct module *owner;
 	const char *name;
 
-        /* Any of the following pointers can legally be NULL, except for
-         * iso_receive which can only be NULL when you don't request
-         * channels. */
+	/* Any of the following pointers can legally be NULL, except for
+	 * iso_receive which can only be NULL when you don't request
+	 * channels. */
 
-        /* New host initialized.  Will also be called during
-         * hpsb_register_highlevel for all hosts already installed. */
-        void (*add_host) (struct hpsb_host *host);
+	/* New host initialized.  Will also be called during
+	 * hpsb_register_highlevel for all hosts already installed. */
+	void (*add_host)(struct hpsb_host *host);
 
-        /* Host about to be removed.  Will also be called during
-         * hpsb_unregister_highlevel once for each host. */
-        void (*remove_host) (struct hpsb_host *host);
+	/* Host about to be removed.  Will also be called during
+	 * hpsb_unregister_highlevel once for each host. */
+	void (*remove_host)(struct hpsb_host *host);
 
-        /* Host experienced bus reset with possible configuration changes.
+	/* Host experienced bus reset with possible configuration changes.
 	 * Note that this one may occur during interrupt/bottom half handling.
 	 * You can not expect to be able to do stock hpsb_reads. */
-        void (*host_reset) (struct hpsb_host *host);
+	void (*host_reset)(struct hpsb_host *host);
 
-        /* An isochronous packet was received.  Channel contains the channel
-         * number for your convenience, it is also contained in the included
-         * packet header (first quadlet, CRCs are missing).  You may get called
-         * for channel/host combinations you did not request. */
-        void (*iso_receive) (struct hpsb_host *host, int channel,
-                             quadlet_t *data, size_t length);
+	/* An isochronous packet was received.  Channel contains the channel
+	 * number for your convenience, it is also contained in the included
+	 * packet header (first quadlet, CRCs are missing).  You may get called
+	 * for channel/host combinations you did not request. */
+	void (*iso_receive)(struct hpsb_host *host, int channel,
+			    quadlet_t *data, size_t length);
 
-        /* A write request was received on either the FCP_COMMAND (direction =
-         * 0) or the FCP_RESPONSE (direction = 1) register.  The cts arg
-         * contains the cts field (first byte of data). */
-        void (*fcp_request) (struct hpsb_host *host, int nodeid, int direction,
-                             int cts, u8 *data, size_t length);
+	/* A write request was received on either the FCP_COMMAND (direction =
+	 * 0) or the FCP_RESPONSE (direction = 1) register.  The cts arg
+	 * contains the cts field (first byte of data). */
+	void (*fcp_request)(struct hpsb_host *host, int nodeid, int direction,
+			    int cts, u8 *data, size_t length);
 
 	/* These are initialized by the subsystem when the
 	 * hpsb_higlevel is registered. */
@@ -67,61 +68,62 @@ struct hpsb_highlevel {
 };
 
 struct hpsb_address_ops {
-        /*
-         * Null function pointers will make the respective operation complete
-         * with RCODE_TYPE_ERROR.  Makes for easy to implement read-only
-         * registers (just leave everything but read NULL).
-         *
-         * All functions shall return appropriate IEEE 1394 rcodes.
-         */
-
-        /* These functions have to implement block reads for themselves. */
-        /* These functions either return a response code
-           or a negative number. In the first case a response will be generated; in the
-           later case, no response will be sent and the driver, that handled the request
-           will send the response itself
-        */
-        int (*read) (struct hpsb_host *host, int nodeid, quadlet_t *buffer,
-                     u64 addr, size_t length, u16 flags);
-        int (*write) (struct hpsb_host *host, int nodeid, int destid,
-		      quadlet_t *data, u64 addr, size_t length, u16 flags);
-
-        /* Lock transactions: write results of ext_tcode operation into
-         * *store. */
-        int (*lock) (struct hpsb_host *host, int nodeid, quadlet_t *store,
-                     u64 addr, quadlet_t data, quadlet_t arg, int ext_tcode, u16 flags);
-        int (*lock64) (struct hpsb_host *host, int nodeid, octlet_t *store,
-                       u64 addr, octlet_t data, octlet_t arg, int ext_tcode, u16 flags);
+	/*
+	 * Null function pointers will make the respective operation complete
+	 * with RCODE_TYPE_ERROR.  Makes for easy to implement read-only
+	 * registers (just leave everything but read NULL).
+	 *
+	 * All functions shall return appropriate IEEE 1394 rcodes.
+	 */
+
+	/* These functions have to implement block reads for themselves.
+	 *
+	 * These functions either return a response code or a negative number.
+	 * In the first case a response will be generated.  In the latter case,
+	 * no response will be sent and the driver which handled the request
+	 * will send the response itself. */
+	int (*read)(struct hpsb_host *host, int nodeid, quadlet_t *buffer,
+		    u64 addr, size_t length, u16 flags);
+	int (*write)(struct hpsb_host *host, int nodeid, int destid,
+		     quadlet_t *data, u64 addr, size_t length, u16 flags);
+
+	/* Lock transactions: write results of ext_tcode operation into
+	 * *store. */
+	int (*lock)(struct hpsb_host *host, int nodeid, quadlet_t *store,
+		    u64 addr, quadlet_t data, quadlet_t arg, int ext_tcode,
+		    u16 flags);
+	int (*lock64)(struct hpsb_host *host, int nodeid, octlet_t *store,
+		      u64 addr, octlet_t data, octlet_t arg, int ext_tcode,
+		      u16 flags);
 };
 
-
 void highlevel_add_host(struct hpsb_host *host);
 void highlevel_remove_host(struct hpsb_host *host);
 void highlevel_host_reset(struct hpsb_host *host);
 
-
-/* these functions are called to handle transactions. They are called, when
-   a packet arrives. The flags argument contains the second word of the first header
-   quadlet of the incoming packet (containing transaction label, retry code,
-   transaction code and priority). These functions either return a response code
-   or a negative number. In the first case a response will be generated; in the
-   later case, no response will be sent and the driver, that handled the request
-   will send the response itself.
-*/
-int highlevel_read(struct hpsb_host *host, int nodeid, void *data,
-                   u64 addr, unsigned int length, u16 flags);
-int highlevel_write(struct hpsb_host *host, int nodeid, int destid,
-		    void *data, u64 addr, unsigned int length, u16 flags);
+/*
+ * These functions are called to handle transactions. They are called when a
+ * packet arrives.  The flags argument contains the second word of the first
+ * header quadlet of the incoming packet (containing transaction label, retry
+ * code, transaction code and priority).  These functions either return a
+ * response code or a negative number.  In the first case a response will be
+ * generated.  In the latter case, no response will be sent and the driver which
+ * handled the request will send the response itself.
+ */
+int highlevel_read(struct hpsb_host *host, int nodeid, void *data, u64 addr,
+		   unsigned int length, u16 flags);
+int highlevel_write(struct hpsb_host *host, int nodeid, int destid, void *data,
+		    u64 addr, unsigned int length, u16 flags);
 int highlevel_lock(struct hpsb_host *host, int nodeid, quadlet_t *store,
-                   u64 addr, quadlet_t data, quadlet_t arg, int ext_tcode, u16 flags);
+		   u64 addr, quadlet_t data, quadlet_t arg, int ext_tcode,
+		   u16 flags);
 int highlevel_lock64(struct hpsb_host *host, int nodeid, octlet_t *store,
-                     u64 addr, octlet_t data, octlet_t arg, int ext_tcode, u16 flags);
+		     u64 addr, octlet_t data, octlet_t arg, int ext_tcode,
+		     u16 flags);
 
-void highlevel_iso_receive(struct hpsb_host *host, void *data,
-                           size_t length);
+void highlevel_iso_receive(struct hpsb_host *host, void *data, size_t length);
 void highlevel_fcp_request(struct hpsb_host *host, int nodeid, int direction,
-                           void *data, size_t length);
-
+			   void *data, size_t length);
 
 /*
  * Register highlevel driver.  The name pointer has to stay valid at all times
@@ -132,13 +134,15 @@ void hpsb_unregister_highlevel(struct hpsb_highlevel *hl);
 
 /*
  * Register handlers for host address spaces.  Start and end are 48 bit pointers
- * and have to be quadlet aligned (end points to the first address behind the
- * handled addresses.  This function can be called multiple times for a single
- * hpsb_highlevel to implement sparse register sets.  The requested region must
- * not overlap any previously allocated region, otherwise registering will fail.
+ * and have to be quadlet aligned.  Argument "end" points to the first address
+ * behind the handled addresses.  This function can be called multiple times for
+ * a single hpsb_highlevel to implement sparse register sets.  The requested
+ * region must not overlap any previously allocated region, otherwise
+ * registering will fail.
  *
- * It returns true for successful allocation.  There is no unregister function,
- * all address spaces are deallocated together with the hpsb_highlevel.
+ * It returns true for successful allocation.  Address spaces can be
+ * unregistered with hpsb_unregister_addrspace.  All remaining address spaces
+ * are automatically deallocated together with the hpsb_highlevel.
  */
 u64 hpsb_allocate_and_register_addrspace(struct hpsb_highlevel *hl,
 					 struct hpsb_host *host,
@@ -146,20 +150,18 @@ u64 hpsb_allocate_and_register_addrspace(struct hpsb_highlevel *hl,
 					 u64 size, u64 alignment,
 					 u64 start, u64 end);
 int hpsb_register_addrspace(struct hpsb_highlevel *hl, struct hpsb_host *host,
-                            struct hpsb_address_ops *ops, u64 start, u64 end);
-
+			    struct hpsb_address_ops *ops, u64 start, u64 end);
 int hpsb_unregister_addrspace(struct hpsb_highlevel *hl, struct hpsb_host *host,
-                              u64 start);
+			      u64 start);
 
 /*
  * Enable or disable receving a certain isochronous channel through the
  * iso_receive op.
  */
 int hpsb_listen_channel(struct hpsb_highlevel *hl, struct hpsb_host *host,
-                         unsigned int channel);
+			 unsigned int channel);
 void hpsb_unlisten_channel(struct hpsb_highlevel *hl, struct hpsb_host *host,
-                           unsigned int channel);
-
+			   unsigned int channel);
 
 /* Retrieve a hostinfo pointer bound to this driver/host */
 void *hpsb_get_hostinfo(struct hpsb_highlevel *hl, struct hpsb_host *host);
@@ -172,19 +174,24 @@ void *hpsb_create_hostinfo(struct hpsb_highlevel *hl, struct hpsb_host *host,
 void hpsb_destroy_hostinfo(struct hpsb_highlevel *hl, struct hpsb_host *host);
 
 /* Set an alternate lookup key for the hostinfo bound to this driver/host */
-void hpsb_set_hostinfo_key(struct hpsb_highlevel *hl, struct hpsb_host *host, unsigned long key);
+void hpsb_set_hostinfo_key(struct hpsb_highlevel *hl, struct hpsb_host *host,
+			   unsigned long key);
 
-/* Retrieve the alternate lookup key for the hostinfo bound to this driver/host */
-unsigned long hpsb_get_hostinfo_key(struct hpsb_highlevel *hl, struct hpsb_host *host);
+/* Retrieve the alternate lookup key for the hostinfo bound to this
+ * driver/host */
+unsigned long hpsb_get_hostinfo_key(struct hpsb_highlevel *hl,
+				    struct hpsb_host *host);
 
 /* Retrieve a hostinfo pointer bound to this driver using its alternate key */
 void *hpsb_get_hostinfo_bykey(struct hpsb_highlevel *hl, unsigned long key);
 
 /* Set the hostinfo pointer to something useful. Usually follows a call to
  * hpsb_create_hostinfo, where the size is 0. */
-int hpsb_set_hostinfo(struct hpsb_highlevel *hl, struct hpsb_host *host, void *data);
+int hpsb_set_hostinfo(struct hpsb_highlevel *hl, struct hpsb_host *host,
+		      void *data);
 
 /* Retrieve hpsb_host using a highlevel handle and a key */
-struct hpsb_host *hpsb_get_host_bykey(struct hpsb_highlevel *hl, unsigned long key);
+struct hpsb_host *hpsb_get_host_bykey(struct hpsb_highlevel *hl,
+				      unsigned long key);
 
 #endif /* IEEE1394_HIGHLEVEL_H */
diff --git a/drivers/ieee1394/hosts.c b/drivers/ieee1394/hosts.c
index 4feead4a35c5..d90a3a1898c0 100644
--- a/drivers/ieee1394/hosts.c
+++ b/drivers/ieee1394/hosts.c
@@ -90,6 +90,16 @@ static int alloc_hostnum_cb(struct hpsb_host *host, void *__data)
 	return 0;
 }
 
+/*
+ * The pending_packet_queue is special in that it's processed
+ * from hardirq context too (such as hpsb_bus_reset()). Hence
+ * split the lock class from the usual networking skb-head
+ * lock class by using a separate key for it:
+ */
+static struct lock_class_key pending_packet_queue_key;
+
+static DEFINE_MUTEX(host_num_alloc);
+
 /**
  * hpsb_alloc_host - allocate a new host controller.
  * @drv: the driver that will manage the host controller
@@ -105,16 +115,6 @@ static int alloc_hostnum_cb(struct hpsb_host *host, void *__data)
  * Return Value: a pointer to the &hpsb_host if successful, %NULL if
  * no memory was available.
  */
-static DEFINE_MUTEX(host_num_alloc);
-
-/*
- * The pending_packet_queue is special in that it's processed
- * from hardirq context too (such as hpsb_bus_reset()). Hence
- * split the lock class from the usual networking skb-head
- * lock class by using a separate key for it:
- */
-static struct lock_class_key pending_packet_queue_key;
-
 struct hpsb_host *hpsb_alloc_host(struct hpsb_host_driver *drv, size_t extra,
 				  struct device *dev)
 {
@@ -143,9 +143,6 @@ struct hpsb_host *hpsb_alloc_host(struct hpsb_host_driver *drv, size_t extra,
 	for (i = 2; i < 16; i++)
 		h->csr.gen_timestamp[i] = jiffies - 60 * HZ;
 
-	for (i = 0; i < ARRAY_SIZE(h->tpool); i++)
-		HPSB_TPOOL_INIT(&h->tpool[i]);
-
 	atomic_set(&h->generation, 0);
 
 	INIT_WORK(&h->delayed_reset, delayed_reset_bus, h);
diff --git a/drivers/ieee1394/hosts.h b/drivers/ieee1394/hosts.h
index 9ad4b2463077..bc6dbfadb891 100644
--- a/drivers/ieee1394/hosts.h
+++ b/drivers/ieee1394/hosts.h
@@ -2,17 +2,19 @@
 #define _IEEE1394_HOSTS_H
 
 #include <linux/device.h>
-#include <linux/wait.h>
 #include <linux/list.h>
-#include <linux/timer.h>
 #include <linux/skbuff.h>
+#include <linux/timer.h>
+#include <linux/types.h>
+#include <linux/workqueue.h>
+#include <asm/atomic.h>
 
-#include <asm/semaphore.h>
+struct pci_dev;
+struct module;
 
 #include "ieee1394_types.h"
 #include "csr.h"
 
-
 struct hpsb_packet;
 struct hpsb_iso;
 
@@ -33,7 +35,6 @@ struct hpsb_host {
 	int node_count;      /* number of identified nodes on this bus */
 	int selfid_count;    /* total number of SelfIDs received */
 	int nodes_active;    /* number of nodes with active link layer */
-	u8 speed[ALL_NODES]; /* speed between each node and local node */
 
 	nodeid_t node_id;    /* node ID of this host */
 	nodeid_t irm_id;     /* ID of this bus' isochronous resource manager */
@@ -53,31 +54,29 @@ struct hpsb_host {
 	int reset_retries;
 	quadlet_t *topology_map;
 	u8 *speed_map;
-	struct csr_control csr;
-
-	/* Per node tlabel pool allocation */
-	struct hpsb_tlabel_pool tpool[ALL_NODES];
 
+	int id;
 	struct hpsb_host_driver *driver;
-
 	struct pci_dev *pdev;
-
-	int id;
-
 	struct device device;
 	struct class_device class_dev;
 
 	int update_config_rom;
 	struct work_struct delayed_reset;
-
 	unsigned int config_roms;
 
 	struct list_head addr_space;
 	u64 low_addr_space;	/* upper bound of physical DMA area */
 	u64 middle_addr_space;	/* upper bound of posted write area */
-};
 
+	u8 speed[ALL_NODES];	/* speed between each node and local node */
+
+	/* per node tlabel allocation */
+	u8 next_tl[ALL_NODES];
+	struct { DECLARE_BITMAP(map, 64); } tl_pool[ALL_NODES];
 
+	struct csr_control csr;
+};
 
 enum devctl_cmd {
 	/* Host is requested to reset its bus and cancel all outstanding async
@@ -112,7 +111,7 @@ enum devctl_cmd {
 
 enum isoctl_cmd {
 	/* rawiso API - see iso.h for the meanings of these commands
-	   (they correspond exactly to the hpsb_iso_* API functions)
+	 * (they correspond exactly to the hpsb_iso_* API functions)
 	 * INIT = allocate resources
 	 * START = begin transmission/reception
 	 * STOP = halt transmission/reception
@@ -160,7 +159,8 @@ struct hpsb_host_driver {
 	/* The hardware driver may optionally support a function that is used
 	 * to set the hardware ConfigROM if the hardware supports handling
 	 * reads to the ConfigROM on its own. */
-	void (*set_hw_config_rom) (struct hpsb_host *host, quadlet_t *config_rom);
+	void (*set_hw_config_rom)(struct hpsb_host *host,
+				  quadlet_t *config_rom);
 
 	/* This function shall implement packet transmission based on
 	 * packet->type.  It shall CRC both parts of the packet (unless
@@ -170,20 +170,21 @@ struct hpsb_host_driver {
 	 * called.  Return 0 on success, negative errno on failure.
 	 * NOTE: The function must be callable in interrupt context.
 	 */
-	int (*transmit_packet) (struct hpsb_host *host,
-				struct hpsb_packet *packet);
+	int (*transmit_packet)(struct hpsb_host *host,
+			       struct hpsb_packet *packet);
 
 	/* This function requests miscellanous services from the driver, see
 	 * above for command codes and expected actions.  Return -1 for unknown
 	 * command, though that should never happen.
 	 */
-	int (*devctl) (struct hpsb_host *host, enum devctl_cmd command, int arg);
+	int (*devctl)(struct hpsb_host *host, enum devctl_cmd command, int arg);
 
 	 /* ISO transmission/reception functions. Return 0 on success, -1
 	  * (or -EXXX errno code) on failure. If the low-level driver does not
 	  * support the new ISO API, set isoctl to NULL.
 	  */
-	int (*isoctl) (struct hpsb_iso *iso, enum isoctl_cmd command, unsigned long arg);
+	int (*isoctl)(struct hpsb_iso *iso, enum isoctl_cmd command,
+		      unsigned long arg);
 
 	/* This function is mainly to redirect local CSR reads/locks to the iso
 	 * management registers (bus manager id, bandwidth available, channels
@@ -196,19 +197,11 @@ struct hpsb_host_driver {
 				 quadlet_t data, quadlet_t compare);
 };
 
-
 struct hpsb_host *hpsb_alloc_host(struct hpsb_host_driver *drv, size_t extra,
 				  struct device *dev);
 int hpsb_add_host(struct hpsb_host *host);
 void hpsb_remove_host(struct hpsb_host *h);
 
-/* The following 2 functions are deprecated and will be removed when the
- * raw1394/libraw1394 update is complete. */
-int hpsb_update_config_rom(struct hpsb_host *host,
-      const quadlet_t *new_rom, size_t size, unsigned char rom_version);
-int hpsb_get_config_rom(struct hpsb_host *host, quadlet_t *buffer,
-      size_t buffersize, size_t *rom_size, unsigned char *rom_version);
-
 /* Updates the configuration rom image of a host.  rom_version must be the
  * current version, otherwise it will fail with return value -1. If this
  * host does not support config-rom-update, it will return -EINVAL.
diff --git a/drivers/ieee1394/ieee1394-ioctl.h b/drivers/ieee1394/ieee1394-ioctl.h
index 156703986348..8f207508ed1d 100644
--- a/drivers/ieee1394/ieee1394-ioctl.h
+++ b/drivers/ieee1394/ieee1394-ioctl.h
@@ -1,5 +1,7 @@
-/* Base file for all ieee1394 ioctl's. Linux-1394 has allocated base '#'
- * with a range of 0x00-0x3f. */
+/*
+ * Base file for all ieee1394 ioctl's.
+ * Linux-1394 has allocated base '#' with a range of 0x00-0x3f.
+ */
 
 #ifndef __IEEE1394_IOCTL_H
 #define __IEEE1394_IOCTL_H
@@ -96,8 +98,7 @@
 	_IOW ('#', 0x27, struct raw1394_iso_packets)
 #define RAW1394_IOC_ISO_XMIT_SYNC		\
 	_IO  ('#', 0x28)
-#define RAW1394_IOC_ISO_RECV_FLUSH              \
+#define RAW1394_IOC_ISO_RECV_FLUSH		\
 	_IO  ('#', 0x29)
 
-
 #endif /* __IEEE1394_IOCTL_H */
diff --git a/drivers/ieee1394/ieee1394.h b/drivers/ieee1394/ieee1394.h
index 936d776de00a..40492074c013 100644
--- a/drivers/ieee1394/ieee1394.h
+++ b/drivers/ieee1394/ieee1394.h
@@ -5,77 +5,78 @@
 #ifndef _IEEE1394_IEEE1394_H
 #define _IEEE1394_IEEE1394_H
 
-#define TCODE_WRITEQ             0x0
-#define TCODE_WRITEB             0x1
-#define TCODE_WRITE_RESPONSE     0x2
-#define TCODE_READQ              0x4
-#define TCODE_READB              0x5
-#define TCODE_READQ_RESPONSE     0x6
-#define TCODE_READB_RESPONSE     0x7
-#define TCODE_CYCLE_START        0x8
-#define TCODE_LOCK_REQUEST       0x9
-#define TCODE_ISO_DATA           0xa
-#define TCODE_STREAM_DATA        0xa
-#define TCODE_LOCK_RESPONSE      0xb
-
-#define RCODE_COMPLETE           0x0
-#define RCODE_CONFLICT_ERROR     0x4
-#define RCODE_DATA_ERROR         0x5
-#define RCODE_TYPE_ERROR         0x6
-#define RCODE_ADDRESS_ERROR      0x7
-
-#define EXTCODE_MASK_SWAP        0x1
-#define EXTCODE_COMPARE_SWAP     0x2
-#define EXTCODE_FETCH_ADD        0x3
-#define EXTCODE_LITTLE_ADD       0x4
-#define EXTCODE_BOUNDED_ADD      0x5
-#define EXTCODE_WRAP_ADD         0x6
-
-#define ACK_COMPLETE             0x1
-#define ACK_PENDING              0x2
-#define ACK_BUSY_X               0x4
-#define ACK_BUSY_A               0x5
-#define ACK_BUSY_B               0x6
-#define ACK_TARDY                0xb
-#define ACK_CONFLICT_ERROR       0xc
-#define ACK_DATA_ERROR           0xd
-#define ACK_TYPE_ERROR           0xe
-#define ACK_ADDRESS_ERROR        0xf
+#define TCODE_WRITEQ		0x0
+#define TCODE_WRITEB		0x1
+#define TCODE_WRITE_RESPONSE	0x2
+#define TCODE_READQ		0x4
+#define TCODE_READB		0x5
+#define TCODE_READQ_RESPONSE	0x6
+#define TCODE_READB_RESPONSE	0x7
+#define TCODE_CYCLE_START	0x8
+#define TCODE_LOCK_REQUEST	0x9
+#define TCODE_ISO_DATA		0xa
+#define TCODE_STREAM_DATA	0xa
+#define TCODE_LOCK_RESPONSE	0xb
+
+#define RCODE_COMPLETE		0x0
+#define RCODE_CONFLICT_ERROR	0x4
+#define RCODE_DATA_ERROR	0x5
+#define RCODE_TYPE_ERROR	0x6
+#define RCODE_ADDRESS_ERROR	0x7
+
+#define EXTCODE_MASK_SWAP	0x1
+#define EXTCODE_COMPARE_SWAP	0x2
+#define EXTCODE_FETCH_ADD	0x3
+#define EXTCODE_LITTLE_ADD	0x4
+#define EXTCODE_BOUNDED_ADD	0x5
+#define EXTCODE_WRAP_ADD	0x6
+
+#define ACK_COMPLETE		0x1
+#define ACK_PENDING		0x2
+#define ACK_BUSY_X		0x4
+#define ACK_BUSY_A		0x5
+#define ACK_BUSY_B		0x6
+#define ACK_TARDY		0xb
+#define ACK_CONFLICT_ERROR	0xc
+#define ACK_DATA_ERROR		0xd
+#define ACK_TYPE_ERROR		0xe
+#define ACK_ADDRESS_ERROR	0xf
 
 /* Non-standard "ACK codes" for internal use */
-#define ACKX_NONE                (-1)
-#define ACKX_SEND_ERROR          (-2)
-#define ACKX_ABORTED             (-3)
-#define ACKX_TIMEOUT             (-4)
-
-
-#define IEEE1394_SPEED_100		0x00
-#define IEEE1394_SPEED_200		0x01
-#define IEEE1394_SPEED_400		0x02
-#define IEEE1394_SPEED_800		0x03
-#define IEEE1394_SPEED_1600		0x04
-#define IEEE1394_SPEED_3200		0x05
+#define ACKX_NONE		(-1)
+#define ACKX_SEND_ERROR		(-2)
+#define ACKX_ABORTED		(-3)
+#define ACKX_TIMEOUT		(-4)
+
+#define IEEE1394_SPEED_100	0x00
+#define IEEE1394_SPEED_200	0x01
+#define IEEE1394_SPEED_400	0x02
+#define IEEE1394_SPEED_800	0x03
+#define IEEE1394_SPEED_1600	0x04
+#define IEEE1394_SPEED_3200	0x05
+
 /* The current highest tested speed supported by the subsystem */
-#define IEEE1394_SPEED_MAX		IEEE1394_SPEED_800
+#define IEEE1394_SPEED_MAX	IEEE1394_SPEED_800
 
 /* Maps speed values above to a string representation */
 extern const char *hpsb_speedto_str[];
 
-
 /* 1394a cable PHY packets */
-#define SELFID_PWRCL_NO_POWER    0x0
-#define SELFID_PWRCL_PROVIDE_15W 0x1
-#define SELFID_PWRCL_PROVIDE_30W 0x2
-#define SELFID_PWRCL_PROVIDE_45W 0x3
-#define SELFID_PWRCL_USE_1W      0x4
-#define SELFID_PWRCL_USE_3W      0x5
-#define SELFID_PWRCL_USE_6W      0x6
-#define SELFID_PWRCL_USE_10W     0x7
-
-#define SELFID_PORT_CHILD        0x3
-#define SELFID_PORT_PARENT       0x2
-#define SELFID_PORT_NCONN        0x1
-#define SELFID_PORT_NONE         0x0
+#define SELFID_PWRCL_NO_POWER		0x0
+#define SELFID_PWRCL_PROVIDE_15W	0x1
+#define SELFID_PWRCL_PROVIDE_30W	0x2
+#define SELFID_PWRCL_PROVIDE_45W	0x3
+#define SELFID_PWRCL_USE_1W		0x4
+#define SELFID_PWRCL_USE_3W		0x5
+#define SELFID_PWRCL_USE_6W		0x6
+#define SELFID_PWRCL_USE_10W		0x7
+
+#define SELFID_PORT_CHILD		0x3
+#define SELFID_PORT_PARENT		0x2
+#define SELFID_PORT_NCONN		0x1
+#define SELFID_PORT_NONE		0x0
+
+#define SELFID_SPEED_UNKNOWN		0x3	/* 1394b PHY */
 
 #define PHYPACKET_LINKON			0x40000000
 #define PHYPACKET_PHYCONFIG_R			0x00800000
@@ -91,76 +92,76 @@ extern const char *hpsb_speedto_str[];
 
 #define EXTPHYPACKET_TYPEMASK			0xC0FC0000
 
-#define PHYPACKET_PORT_SHIFT     24
-#define PHYPACKET_GAPCOUNT_SHIFT 16
+#define PHYPACKET_PORT_SHIFT		24
+#define PHYPACKET_GAPCOUNT_SHIFT	16
 
 /* 1394a PHY register map bitmasks */
-#define PHY_00_PHYSICAL_ID       0xFC
-#define PHY_00_R                 0x02 /* Root */
-#define PHY_00_PS                0x01 /* Power Status*/
-#define PHY_01_RHB               0x80 /* Root Hold-Off */
-#define PHY_01_IBR               0x80 /* Initiate Bus Reset */
-#define PHY_01_GAP_COUNT         0x3F
-#define PHY_02_EXTENDED          0xE0 /* 0x7 for 1394a-compliant PHY */
-#define PHY_02_TOTAL_PORTS       0x1F
-#define PHY_03_MAX_SPEED         0xE0
-#define PHY_03_DELAY             0x0F
-#define PHY_04_LCTRL             0x80 /* Link Active Report Control */
-#define PHY_04_CONTENDER         0x40
-#define PHY_04_JITTER            0x38
-#define PHY_04_PWR_CLASS         0x07 /* Power Class */
-#define PHY_05_WATCHDOG          0x80
-#define PHY_05_ISBR              0x40 /* Initiate Short Bus Reset */
-#define PHY_05_LOOP              0x20 /* Loop Detect */
-#define PHY_05_PWR_FAIL          0x10 /* Cable Power Failure Detect */
-#define PHY_05_TIMEOUT           0x08 /* Arbitration State Machine Timeout */
-#define PHY_05_PORT_EVENT        0x04 /* Port Event Detect */
-#define PHY_05_ENAB_ACCEL        0x02 /* Enable Arbitration Acceleration */
-#define PHY_05_ENAB_MULTI        0x01 /* Ena. Multispeed Packet Concatenation */
+#define PHY_00_PHYSICAL_ID	0xFC
+#define PHY_00_R		0x02 /* Root */
+#define PHY_00_PS		0x01 /* Power Status*/
+#define PHY_01_RHB		0x80 /* Root Hold-Off */
+#define PHY_01_IBR		0x80 /* Initiate Bus Reset */
+#define PHY_01_GAP_COUNT	0x3F
+#define PHY_02_EXTENDED		0xE0 /* 0x7 for 1394a-compliant PHY */
+#define PHY_02_TOTAL_PORTS	0x1F
+#define PHY_03_MAX_SPEED	0xE0
+#define PHY_03_DELAY		0x0F
+#define PHY_04_LCTRL		0x80 /* Link Active Report Control */
+#define PHY_04_CONTENDER	0x40
+#define PHY_04_JITTER		0x38
+#define PHY_04_PWR_CLASS	0x07 /* Power Class */
+#define PHY_05_WATCHDOG		0x80
+#define PHY_05_ISBR		0x40 /* Initiate Short Bus Reset */
+#define PHY_05_LOOP		0x20 /* Loop Detect */
+#define PHY_05_PWR_FAIL		0x10 /* Cable Power Failure Detect */
+#define PHY_05_TIMEOUT		0x08 /* Arbitration State Machine Timeout */
+#define PHY_05_PORT_EVENT	0x04 /* Port Event Detect */
+#define PHY_05_ENAB_ACCEL	0x02 /* Enable Arbitration Acceleration */
+#define PHY_05_ENAB_MULTI	0x01 /* Ena. Multispeed Packet Concatenation */
 
 #include <asm/byteorder.h>
 
 #ifdef __BIG_ENDIAN_BITFIELD
 
 struct selfid {
-        u32 packet_identifier:2; /* always binary 10 */
-        u32 phy_id:6;
-        /* byte */
-        u32 extended:1; /* if true is struct ext_selfid */
-        u32 link_active:1;
-        u32 gap_count:6;
-        /* byte */
-        u32 speed:2;
-        u32 phy_delay:2;
-        u32 contender:1;
-        u32 power_class:3;
-        /* byte */
-        u32 port0:2;
-        u32 port1:2;
-        u32 port2:2;
-        u32 initiated_reset:1;
-        u32 more_packets:1;
+	u32 packet_identifier:2; /* always binary 10 */
+	u32 phy_id:6;
+	/* byte */
+	u32 extended:1; /* if true is struct ext_selfid */
+	u32 link_active:1;
+	u32 gap_count:6;
+	/* byte */
+	u32 speed:2;
+	u32 phy_delay:2;
+	u32 contender:1;
+	u32 power_class:3;
+	/* byte */
+	u32 port0:2;
+	u32 port1:2;
+	u32 port2:2;
+	u32 initiated_reset:1;
+	u32 more_packets:1;
 } __attribute__((packed));
 
 struct ext_selfid {
-        u32 packet_identifier:2; /* always binary 10 */
-        u32 phy_id:6;
-        /* byte */
-        u32 extended:1; /* if false is struct selfid */
-        u32 seq_nr:3;
-        u32 reserved:2;
-        u32 porta:2;
-        /* byte */
-        u32 portb:2;
-        u32 portc:2;
-        u32 portd:2;
-        u32 porte:2;
-        /* byte */
-        u32 portf:2;
-        u32 portg:2;
-        u32 porth:2;
-        u32 reserved2:1;
-        u32 more_packets:1;
+	u32 packet_identifier:2; /* always binary 10 */
+	u32 phy_id:6;
+	/* byte */
+	u32 extended:1; /* if false is struct selfid */
+	u32 seq_nr:3;
+	u32 reserved:2;
+	u32 porta:2;
+	/* byte */
+	u32 portb:2;
+	u32 portc:2;
+	u32 portd:2;
+	u32 porte:2;
+	/* byte */
+	u32 portf:2;
+	u32 portg:2;
+	u32 porth:2;
+	u32 reserved2:1;
+	u32 more_packets:1;
 } __attribute__((packed));
 
 #elif defined __LITTLE_ENDIAN_BITFIELD /* __BIG_ENDIAN_BITFIELD */
@@ -171,49 +172,48 @@ struct ext_selfid {
  */
 
 struct selfid {
-        u32 phy_id:6;
-        u32 packet_identifier:2; /* always binary 10 */
-        /* byte */
-        u32 gap_count:6;
-        u32 link_active:1;
-        u32 extended:1; /* if true is struct ext_selfid */
-        /* byte */
-        u32 power_class:3;
-        u32 contender:1;
-        u32 phy_delay:2;
-        u32 speed:2;
-        /* byte */
-        u32 more_packets:1;
-        u32 initiated_reset:1;
-        u32 port2:2;
-        u32 port1:2;
-        u32 port0:2;
+	u32 phy_id:6;
+	u32 packet_identifier:2; /* always binary 10 */
+	/* byte */
+	u32 gap_count:6;
+	u32 link_active:1;
+	u32 extended:1; /* if true is struct ext_selfid */
+	/* byte */
+	u32 power_class:3;
+	u32 contender:1;
+	u32 phy_delay:2;
+	u32 speed:2;
+	/* byte */
+	u32 more_packets:1;
+	u32 initiated_reset:1;
+	u32 port2:2;
+	u32 port1:2;
+	u32 port0:2;
 } __attribute__((packed));
 
 struct ext_selfid {
-        u32 phy_id:6;
-        u32 packet_identifier:2; /* always binary 10 */
-        /* byte */
-        u32 porta:2;
-        u32 reserved:2;
-        u32 seq_nr:3;
-        u32 extended:1; /* if false is struct selfid */
-        /* byte */
-        u32 porte:2;
-        u32 portd:2;
-        u32 portc:2;
-        u32 portb:2;
-        /* byte */
-        u32 more_packets:1;
-        u32 reserved2:1;
-        u32 porth:2;
-        u32 portg:2;
-        u32 portf:2;
+	u32 phy_id:6;
+	u32 packet_identifier:2; /* always binary 10 */
+	/* byte */
+	u32 porta:2;
+	u32 reserved:2;
+	u32 seq_nr:3;
+	u32 extended:1; /* if false is struct selfid */
+	/* byte */
+	u32 porte:2;
+	u32 portd:2;
+	u32 portc:2;
+	u32 portb:2;
+	/* byte */
+	u32 more_packets:1;
+	u32 reserved2:1;
+	u32 porth:2;
+	u32 portg:2;
+	u32 portf:2;
 } __attribute__((packed));
 
 #else
 #error What? PDP endian?
 #endif /* __BIG_ENDIAN_BITFIELD */
 
-
 #endif /* _IEEE1394_IEEE1394_H */
diff --git a/drivers/ieee1394/ieee1394_core.c b/drivers/ieee1394/ieee1394_core.c
index f43739c5cab2..5fccf9f7a1d2 100644
--- a/drivers/ieee1394/ieee1394_core.c
+++ b/drivers/ieee1394/ieee1394_core.c
@@ -35,7 +35,6 @@
 #include <linux/kthread.h>
 
 #include <asm/byteorder.h>
-#include <asm/semaphore.h>
 
 #include "ieee1394_types.h"
 #include "ieee1394.h"
@@ -86,7 +85,7 @@ static void dump_packet(const char *text, quadlet_t *data, int size, int speed)
 	printk("\n");
 }
 #else
-#define dump_packet(a,b,c,d)
+#define dump_packet(a,b,c,d) do {} while (0)
 #endif
 
 static void abort_requests(struct hpsb_host *host);
@@ -355,10 +354,12 @@ static void build_speed_map(struct hpsb_host *host, int nodecount)
 		}
 	}
 
+#if SELFID_SPEED_UNKNOWN != IEEE1394_SPEED_MAX
 	/* assume maximum speed for 1394b PHYs, nodemgr will correct it */
 	for (n = 0; n < nodecount; n++)
-		if (speedcap[n] == 3)
+		if (speedcap[n] == SELFID_SPEED_UNKNOWN)
 			speedcap[n] = IEEE1394_SPEED_MAX;
+#endif
 }
 
 
@@ -1169,7 +1170,7 @@ static void __exit ieee1394_cleanup(void)
 	unregister_chrdev_region(IEEE1394_CORE_DEV, 256);
 }
 
-module_init(ieee1394_init);
+fs_initcall(ieee1394_init); /* same as ohci1394 */
 module_exit(ieee1394_cleanup);
 
 /* Exported symbols */
diff --git a/drivers/ieee1394/ieee1394_core.h b/drivers/ieee1394/ieee1394_core.h
index 0ecbf335c64f..af4a78a8ef3b 100644
--- a/drivers/ieee1394/ieee1394_core.h
+++ b/drivers/ieee1394/ieee1394_core.h
@@ -1,12 +1,15 @@
-
 #ifndef _IEEE1394_CORE_H
 #define _IEEE1394_CORE_H
 
-#include <linux/slab.h>
+#include <linux/device.h>
+#include <linux/fs.h>
+#include <linux/list.h>
+#include <linux/skbuff.h>
+#include <linux/types.h>
 #include <asm/atomic.h>
-#include <asm/semaphore.h>
-#include "hosts.h"
 
+#include "hosts.h"
+#include "ieee1394_types.h"
 
 struct hpsb_packet {
 	/* This struct is basically read-only for hosts with the exception of
@@ -58,7 +61,6 @@ struct hpsb_packet {
 	size_t header_size;
 	size_t data_size;
 
-
 	struct hpsb_host *host;
 	unsigned int generation;
 
@@ -80,7 +82,7 @@ struct hpsb_packet {
 
 /* Set a task for when a packet completes */
 void hpsb_set_packet_complete_task(struct hpsb_packet *packet,
-		void (*routine)(void *), void *data);
+				   void (*routine)(void *), void *data);
 
 static inline struct hpsb_packet *driver_packet(struct list_head *l)
 {
@@ -92,7 +94,6 @@ void abort_timedouts(unsigned long __opaque);
 struct hpsb_packet *hpsb_alloc_packet(size_t data_size);
 void hpsb_free_packet(struct hpsb_packet *packet);
 
-
 /*
  * Generation counter for the complete 1394 subsystem.  Generation gets
  * incremented on every change in the subsystem (e.g. bus reset).
@@ -204,10 +205,14 @@ void hpsb_packet_received(struct hpsb_host *host, quadlet_t *data, size_t size,
 #define IEEE1394_MINOR_BLOCK_EXPERIMENTAL 15
 
 #define IEEE1394_CORE_DEV	  MKDEV(IEEE1394_MAJOR, 0)
-#define IEEE1394_RAW1394_DEV	  MKDEV(IEEE1394_MAJOR, IEEE1394_MINOR_BLOCK_RAW1394 * 16)
-#define IEEE1394_VIDEO1394_DEV	  MKDEV(IEEE1394_MAJOR, IEEE1394_MINOR_BLOCK_VIDEO1394 * 16)
-#define IEEE1394_DV1394_DEV	  MKDEV(IEEE1394_MAJOR, IEEE1394_MINOR_BLOCK_DV1394 * 16)
-#define IEEE1394_EXPERIMENTAL_DEV MKDEV(IEEE1394_MAJOR, IEEE1394_MINOR_BLOCK_EXPERIMENTAL * 16)
+#define IEEE1394_RAW1394_DEV	  MKDEV(IEEE1394_MAJOR, \
+					IEEE1394_MINOR_BLOCK_RAW1394 * 16)
+#define IEEE1394_VIDEO1394_DEV	  MKDEV(IEEE1394_MAJOR, \
+					IEEE1394_MINOR_BLOCK_VIDEO1394 * 16)
+#define IEEE1394_DV1394_DEV	  MKDEV(IEEE1394_MAJOR, \
+					IEEE1394_MINOR_BLOCK_DV1394 * 16)
+#define IEEE1394_EXPERIMENTAL_DEV MKDEV(IEEE1394_MAJOR, \
+					IEEE1394_MINOR_BLOCK_EXPERIMENTAL * 16)
 
 /* return the index (within a minor number block) of a file */
 static inline unsigned char ieee1394_file_to_instance(struct file *file)
@@ -223,4 +228,3 @@ extern struct class hpsb_host_class;
 extern struct class *hpsb_protocol_class;
 
 #endif /* _IEEE1394_CORE_H */
-
diff --git a/drivers/ieee1394/ieee1394_hotplug.h b/drivers/ieee1394/ieee1394_hotplug.h
index 5be70d31b007..dd5500ed8322 100644
--- a/drivers/ieee1394/ieee1394_hotplug.h
+++ b/drivers/ieee1394/ieee1394_hotplug.h
@@ -1,33 +1,19 @@
 #ifndef _IEEE1394_HOTPLUG_H
 #define _IEEE1394_HOTPLUG_H
 
-#include <linux/kernel.h>
-#include <linux/types.h>
-#include <linux/mod_devicetable.h>
-
 /* Unit spec id and sw version entry for some protocols */
 #define AVC_UNIT_SPEC_ID_ENTRY		0x0000A02D
 #define AVC_SW_VERSION_ENTRY		0x00010001
 #define CAMERA_UNIT_SPEC_ID_ENTRY	0x0000A02D
 #define CAMERA_SW_VERSION_ENTRY		0x00000100
 
-/* Check to make sure this all isn't already defined */
-#ifndef IEEE1394_MATCH_VENDOR_ID
-
-#define IEEE1394_MATCH_VENDOR_ID	0x0001
-#define IEEE1394_MATCH_MODEL_ID		0x0002
-#define IEEE1394_MATCH_SPECIFIER_ID	0x0004
-#define IEEE1394_MATCH_VERSION		0x0008
-
-struct ieee1394_device_id {
-	u32 match_flags;
-	u32 vendor_id;
-	u32 model_id;
-	u32 specifier_id;
-	u32 version;
-	void *driver_data;
-};
-
-#endif
+/* /include/linux/mod_devicetable.h defines:
+ *	IEEE1394_MATCH_VENDOR_ID
+ *	IEEE1394_MATCH_MODEL_ID
+ *	IEEE1394_MATCH_SPECIFIER_ID
+ *	IEEE1394_MATCH_VERSION
+ *	struct ieee1394_device_id
+ */
+#include <linux/mod_devicetable.h>
 
 #endif /* _IEEE1394_HOTPLUG_H */
diff --git a/drivers/ieee1394/ieee1394_transactions.c b/drivers/ieee1394/ieee1394_transactions.c
index a114b91d606d..0833fc9f50c4 100644
--- a/drivers/ieee1394/ieee1394_transactions.c
+++ b/drivers/ieee1394/ieee1394_transactions.c
@@ -9,19 +9,17 @@
  * directory of the kernel sources for details.
  */
 
-#include <linux/sched.h>
 #include <linux/bitops.h>
-#include <linux/smp_lock.h>
-#include <linux/interrupt.h>
+#include <linux/spinlock.h>
+#include <linux/wait.h>
 
+#include <asm/bug.h>
 #include <asm/errno.h>
 
 #include "ieee1394.h"
 #include "ieee1394_types.h"
 #include "hosts.h"
 #include "ieee1394_core.h"
-#include "highlevel.h"
-#include "nodemgr.h"
 #include "ieee1394_transactions.h"
 
 #define PREP_ASYNC_HEAD_ADDRESS(tc) \
@@ -31,6 +29,13 @@
         packet->header[1] = (packet->host->node_id << 16) | (addr >> 32); \
         packet->header[2] = addr & 0xffffffff
 
+#ifndef HPSB_DEBUG_TLABELS
+static
+#endif
+spinlock_t hpsb_tlabel_lock = SPIN_LOCK_UNLOCKED;
+
+static DECLARE_WAIT_QUEUE_HEAD(tlabel_wq);
+
 static void fill_async_readquad(struct hpsb_packet *packet, u64 addr)
 {
 	PREP_ASYNC_HEAD_ADDRESS(TCODE_READQ);
@@ -114,9 +119,41 @@ static void fill_async_stream_packet(struct hpsb_packet *packet, int length,
 	packet->tcode = TCODE_ISO_DATA;
 }
 
+/* same as hpsb_get_tlabel, except that it returns immediately */
+static int hpsb_get_tlabel_atomic(struct hpsb_packet *packet)
+{
+	unsigned long flags, *tp;
+	u8 *next;
+	int tlabel, n = NODEID_TO_NODE(packet->node_id);
+
+	/* Broadcast transactions are complete once the request has been sent.
+	 * Use the same transaction label for all broadcast transactions. */
+	if (unlikely(n == ALL_NODES)) {
+		packet->tlabel = 0;
+		return 0;
+	}
+	tp = packet->host->tl_pool[n].map;
+	next = &packet->host->next_tl[n];
+
+	spin_lock_irqsave(&hpsb_tlabel_lock, flags);
+	tlabel = find_next_zero_bit(tp, 64, *next);
+	if (tlabel > 63)
+		tlabel = find_first_zero_bit(tp, 64);
+	if (tlabel > 63) {
+		spin_unlock_irqrestore(&hpsb_tlabel_lock, flags);
+		return -EAGAIN;
+	}
+	__set_bit(tlabel, tp);
+	*next = (tlabel + 1) & 63;
+	spin_unlock_irqrestore(&hpsb_tlabel_lock, flags);
+
+	packet->tlabel = tlabel;
+	return 0;
+}
+
 /**
  * hpsb_get_tlabel - allocate a transaction label
- * @packet: the packet who's tlabel/tpool we set
+ * @packet: the packet whose tlabel and tl_pool we set
  *
  * Every asynchronous transaction on the 1394 bus needs a transaction
  * label to match the response to the request.  This label has to be
@@ -130,42 +167,25 @@ static void fill_async_stream_packet(struct hpsb_packet *packet, int length,
  * Return value: Zero on success, otherwise non-zero. A non-zero return
  * generally means there are no available tlabels. If this is called out
  * of interrupt or atomic context, then it will sleep until can return a
- * tlabel.
+ * tlabel or a signal is received.
  */
 int hpsb_get_tlabel(struct hpsb_packet *packet)
 {
-	unsigned long flags;
-	struct hpsb_tlabel_pool *tp;
-	int n = NODEID_TO_NODE(packet->node_id);
-
-	if (unlikely(n == ALL_NODES))
-		return 0;
-	tp = &packet->host->tpool[n];
-
-	if (irqs_disabled() || in_atomic()) {
-		if (down_trylock(&tp->count))
-			return 1;
-	} else {
-		down(&tp->count);
-	}
-
-	spin_lock_irqsave(&tp->lock, flags);
-
-	packet->tlabel = find_next_zero_bit(tp->pool, 64, tp->next);
-	if (packet->tlabel > 63)
-		packet->tlabel = find_first_zero_bit(tp->pool, 64);
-	tp->next = (packet->tlabel + 1) % 64;
-	/* Should _never_ happen */
-	BUG_ON(test_and_set_bit(packet->tlabel, tp->pool));
-	tp->allocations++;
-	spin_unlock_irqrestore(&tp->lock, flags);
-
-	return 0;
+	if (irqs_disabled() || in_atomic())
+		return hpsb_get_tlabel_atomic(packet);
+
+	/* NB: The macro wait_event_interruptible() is called with a condition
+	 * argument with side effect.  This is only possible because the side
+	 * effect does not occur until the condition became true, and
+	 * wait_event_interruptible() won't evaluate the condition again after
+	 * that. */
+	return wait_event_interruptible(tlabel_wq,
+					!hpsb_get_tlabel_atomic(packet));
 }
 
 /**
  * hpsb_free_tlabel - free an allocated transaction label
- * @packet: packet whos tlabel/tpool needs to be cleared
+ * @packet: packet whose tlabel and tl_pool needs to be cleared
  *
  * Frees the transaction label allocated with hpsb_get_tlabel().  The
  * tlabel has to be freed after the transaction is complete (i.e. response
@@ -176,21 +196,20 @@ int hpsb_get_tlabel(struct hpsb_packet *packet)
  */
 void hpsb_free_tlabel(struct hpsb_packet *packet)
 {
-	unsigned long flags;
-	struct hpsb_tlabel_pool *tp;
-	int n = NODEID_TO_NODE(packet->node_id);
+	unsigned long flags, *tp;
+	int tlabel, n = NODEID_TO_NODE(packet->node_id);
 
 	if (unlikely(n == ALL_NODES))
 		return;
-	tp = &packet->host->tpool[n];
+	tp = packet->host->tl_pool[n].map;
+	tlabel = packet->tlabel;
+	BUG_ON(tlabel > 63 || tlabel < 0);
 
-	BUG_ON(packet->tlabel > 63 || packet->tlabel < 0);
+	spin_lock_irqsave(&hpsb_tlabel_lock, flags);
+	BUG_ON(!__test_and_clear_bit(tlabel, tp));
+	spin_unlock_irqrestore(&hpsb_tlabel_lock, flags);
 
-	spin_lock_irqsave(&tp->lock, flags);
-	BUG_ON(!test_and_clear_bit(packet->tlabel, tp->pool));
-	spin_unlock_irqrestore(&tp->lock, flags);
-
-	up(&tp->count);
+	wake_up_interruptible(&tlabel_wq);
 }
 
 int hpsb_packet_success(struct hpsb_packet *packet)
@@ -214,7 +233,7 @@ int hpsb_packet_success(struct hpsb_packet *packet)
 				 packet->node_id);
 			return -EAGAIN;
 		}
-		HPSB_PANIC("reached unreachable code 1 in %s", __FUNCTION__);
+		BUG();
 
 	case ACK_BUSY_X:
 	case ACK_BUSY_A:
@@ -261,8 +280,7 @@ int hpsb_packet_success(struct hpsb_packet *packet)
 			 packet->ack_code, packet->node_id, packet->tcode);
 		return -EAGAIN;
 	}
-
-	HPSB_PANIC("reached unreachable code 2 in %s", __FUNCTION__);
+	BUG();
 }
 
 struct hpsb_packet *hpsb_make_readpacket(struct hpsb_host *host, nodeid_t node,
diff --git a/drivers/ieee1394/ieee1394_transactions.h b/drivers/ieee1394/ieee1394_transactions.h
index 45ba784fe6da..c1369c41469b 100644
--- a/drivers/ieee1394/ieee1394_transactions.h
+++ b/drivers/ieee1394/ieee1394_transactions.h
@@ -1,32 +1,32 @@
 #ifndef _IEEE1394_TRANSACTIONS_H
 #define _IEEE1394_TRANSACTIONS_H
 
-#include "ieee1394_core.h"
+#include <linux/types.h>
 
+#include "ieee1394_types.h"
+
+struct hpsb_packet;
+struct hpsb_host;
 
-/*
- * Get and free transaction labels.
- */
 int hpsb_get_tlabel(struct hpsb_packet *packet);
 void hpsb_free_tlabel(struct hpsb_packet *packet);
-
 struct hpsb_packet *hpsb_make_readpacket(struct hpsb_host *host, nodeid_t node,
 					 u64 addr, size_t length);
 struct hpsb_packet *hpsb_make_lockpacket(struct hpsb_host *host, nodeid_t node,
-                                         u64 addr, int extcode, quadlet_t *data,
+					 u64 addr, int extcode, quadlet_t *data,
 					 quadlet_t arg);
-struct hpsb_packet *hpsb_make_lock64packet(struct hpsb_host *host, nodeid_t node,
-                                          u64 addr, int extcode, octlet_t *data,
-					  octlet_t arg);
-struct hpsb_packet *hpsb_make_phypacket(struct hpsb_host *host,
-                                        quadlet_t data) ;
-struct hpsb_packet *hpsb_make_isopacket(struct hpsb_host *host,
-					int length, int channel,
-					int tag, int sync);
-struct hpsb_packet *hpsb_make_writepacket (struct hpsb_host *host, nodeid_t node,
-					   u64 addr, quadlet_t *buffer, size_t length);
+struct hpsb_packet *hpsb_make_lock64packet(struct hpsb_host *host,
+					   nodeid_t node, u64 addr, int extcode,
+					   octlet_t *data, octlet_t arg);
+struct hpsb_packet *hpsb_make_phypacket(struct hpsb_host *host, quadlet_t data);
+struct hpsb_packet *hpsb_make_isopacket(struct hpsb_host *host, int length,
+					int channel, int tag, int sync);
+struct hpsb_packet *hpsb_make_writepacket(struct hpsb_host *host,
+					  nodeid_t node, u64 addr,
+					  quadlet_t *buffer, size_t length);
 struct hpsb_packet *hpsb_make_streampacket(struct hpsb_host *host, u8 *buffer,
-                                           int length, int channel, int tag, int sync);
+                                           int length, int channel, int tag,
+					   int sync);
 
 /*
  * hpsb_packet_success - Make sense of the ack and reply codes and
@@ -40,9 +40,8 @@ struct hpsb_packet *hpsb_make_streampacket(struct hpsb_host *host, u8 *buffer,
  */
 int hpsb_packet_success(struct hpsb_packet *packet);
 
-
 /*
- * The generic read, write and lock functions.  All recognize the local node ID
+ * The generic read and write functions.  All recognize the local node ID
  * and act accordingly.  Read and write automatically use quadlet commands if
  * length == 4 and and block commands otherwise (however, they do not yet
  * support lengths that are not a multiple of 4).  You must explicitly specifiy
@@ -54,4 +53,8 @@ int hpsb_read(struct hpsb_host *host, nodeid_t node, unsigned int generation,
 int hpsb_write(struct hpsb_host *host, nodeid_t node, unsigned int generation,
 	       u64 addr, quadlet_t *buffer, size_t length);
 
+#ifdef HPSB_DEBUG_TLABELS
+extern spinlock_t hpsb_tlabel_lock;
+#endif
+
 #endif /* _IEEE1394_TRANSACTIONS_H */
diff --git a/drivers/ieee1394/ieee1394_types.h b/drivers/ieee1394/ieee1394_types.h
index 3165609ec1ec..9803aaa15be0 100644
--- a/drivers/ieee1394/ieee1394_types.h
+++ b/drivers/ieee1394/ieee1394_types.h
@@ -1,37 +1,11 @@
-
 #ifndef _IEEE1394_TYPES_H
 #define _IEEE1394_TYPES_H
 
 #include <linux/kernel.h>
-#include <linux/types.h>
-#include <linux/list.h>
-#include <linux/init.h>
-#include <linux/spinlock.h>
 #include <linux/string.h>
-
-#include <asm/semaphore.h>
+#include <linux/types.h>
 #include <asm/byteorder.h>
 
-
-/* Transaction Label handling */
-struct hpsb_tlabel_pool {
-	DECLARE_BITMAP(pool, 64);
-	spinlock_t lock;
-	u8 next;
-	u32 allocations;
-	struct semaphore count;
-};
-
-#define HPSB_TPOOL_INIT(_tp)			\
-do {						\
-	bitmap_zero((_tp)->pool, 64);		\
-	spin_lock_init(&(_tp)->lock);		\
-	(_tp)->next = 0;			\
-	(_tp)->allocations = 0;			\
-	sema_init(&(_tp)->count, 63);		\
-} while (0)
-
-
 typedef u32 quadlet_t;
 typedef u64 octlet_t;
 typedef u16 nodeid_t;
@@ -54,46 +28,40 @@ typedef u16 arm_length_t;
 #define NODE_BUS_ARGS(__host, __nodeid)	\
 	__host->id, NODEID_TO_NODE(__nodeid), NODEID_TO_BUS(__nodeid)
 
-#define HPSB_PRINT(level, fmt, args...) printk(level "ieee1394: " fmt "\n" , ## args)
+#define HPSB_PRINT(level, fmt, args...) \
+	printk(level "ieee1394: " fmt "\n" , ## args)
 
-#define HPSB_DEBUG(fmt, args...) HPSB_PRINT(KERN_DEBUG, fmt , ## args)
-#define HPSB_INFO(fmt, args...) HPSB_PRINT(KERN_INFO, fmt , ## args)
-#define HPSB_NOTICE(fmt, args...) HPSB_PRINT(KERN_NOTICE, fmt , ## args)
-#define HPSB_WARN(fmt, args...) HPSB_PRINT(KERN_WARNING, fmt , ## args)
-#define HPSB_ERR(fmt, args...) HPSB_PRINT(KERN_ERR, fmt , ## args)
+#define HPSB_DEBUG(fmt, args...)	HPSB_PRINT(KERN_DEBUG, fmt , ## args)
+#define HPSB_INFO(fmt, args...)		HPSB_PRINT(KERN_INFO, fmt , ## args)
+#define HPSB_NOTICE(fmt, args...)	HPSB_PRINT(KERN_NOTICE, fmt , ## args)
+#define HPSB_WARN(fmt, args...)		HPSB_PRINT(KERN_WARNING, fmt , ## args)
+#define HPSB_ERR(fmt, args...)		HPSB_PRINT(KERN_ERR, fmt , ## args)
 
 #ifdef CONFIG_IEEE1394_VERBOSEDEBUG
-#define HPSB_VERBOSE(fmt, args...) HPSB_PRINT(KERN_DEBUG, fmt , ## args)
+#define HPSB_VERBOSE(fmt, args...)	HPSB_PRINT(KERN_DEBUG, fmt , ## args)
+#define HPSB_DEBUG_TLABELS
 #else
-#define HPSB_VERBOSE(fmt, args...)
+#define HPSB_VERBOSE(fmt, args...)	do {} while (0)
 #endif
 
-#define HPSB_PANIC(fmt, args...) panic("ieee1394: " fmt "\n" , ## args)
-
-#define HPSB_TRACE() HPSB_PRINT(KERN_INFO, "TRACE - %s, %s(), line %d", __FILE__, __FUNCTION__, __LINE__)
-
-
 #ifdef __BIG_ENDIAN
 
-static __inline__ void *memcpy_le32(u32 *dest, const u32 *__src, size_t count)
+static inline void *memcpy_le32(u32 *dest, const u32 *__src, size_t count)
 {
-        void *tmp = dest;
+	void *tmp = dest;
 	u32 *src = (u32 *)__src;
 
-        count /= 4;
-
-        while (count--) {
-                *dest++ = swab32p(src++);
-        }
-
-        return tmp;
+	count /= 4;
+	while (count--)
+		*dest++ = swab32p(src++);
+	return tmp;
 }
 
 #else
 
 static __inline__ void *memcpy_le32(u32 *dest, const u32 *src, size_t count)
 {
-        return memcpy(dest, src, count);
+	return memcpy(dest, src, count);
 }
 
 #endif /* __BIG_ENDIAN */
diff --git a/drivers/ieee1394/iso.c b/drivers/ieee1394/iso.c
index f26680ebef7c..08bd15d2a7b6 100644
--- a/drivers/ieee1394/iso.c
+++ b/drivers/ieee1394/iso.c
@@ -9,8 +9,11 @@
  * directory of the kernel sources for details.
  */
 
-#include <linux/slab.h>
+#include <linux/pci.h>
 #include <linux/sched.h>
+#include <linux/slab.h>
+
+#include "hosts.h"
 #include "iso.h"
 
 void hpsb_iso_stop(struct hpsb_iso *iso)
diff --git a/drivers/ieee1394/iso.h b/drivers/ieee1394/iso.h
index 3efc60b33a88..1210a97e8685 100644
--- a/drivers/ieee1394/iso.h
+++ b/drivers/ieee1394/iso.h
@@ -12,33 +12,40 @@
 #ifndef IEEE1394_ISO_H
 #define IEEE1394_ISO_H
 
-#include "hosts.h"
+#include <linux/spinlock_types.h>
+#include <asm/atomic.h>
+#include <asm/types.h>
+
 #include "dma.h"
 
-/* high-level ISO interface */
+struct hpsb_host;
 
-/* This API sends and receives isochronous packets on a large,
-   virtually-contiguous kernel memory buffer. The buffer may be mapped
-   into a user-space process for zero-copy transmission and reception.
+/* high-level ISO interface */
 
-   There are no explicit boundaries between packets in the buffer. A
-   packet may be transmitted or received at any location. However,
-   low-level drivers may impose certain restrictions on alignment or
-   size of packets. (e.g. in OHCI no packet may cross a page boundary,
-   and packets should be quadlet-aligned)
-*/
+/*
+ * This API sends and receives isochronous packets on a large,
+ * virtually-contiguous kernel memory buffer. The buffer may be mapped
+ * into a user-space process for zero-copy transmission and reception.
+ *
+ * There are no explicit boundaries between packets in the buffer. A
+ * packet may be transmitted or received at any location. However,
+ * low-level drivers may impose certain restrictions on alignment or
+ * size of packets. (e.g. in OHCI no packet may cross a page boundary,
+ * and packets should be quadlet-aligned)
+ */
 
 /* Packet descriptor - the API maintains a ring buffer of these packet
-   descriptors in kernel memory (hpsb_iso.infos[]).  */
-
+ * descriptors in kernel memory (hpsb_iso.infos[]).  */
 struct hpsb_iso_packet_info {
 	/* offset of data payload relative to the first byte of the buffer */
 	__u32 offset;
 
-	/* length of the data payload, in bytes (not including the isochronous header) */
+	/* length of the data payload, in bytes (not including the isochronous
+	 * header) */
 	__u16 len;
 
-	/* (recv only) the cycle number (mod 8000) on which the packet was received */
+	/* (recv only) the cycle number (mod 8000) on which the packet was
+	 * received */
 	__u16 cycle;
 
 	/* (recv only) channel on which the packet was received */
@@ -48,12 +55,10 @@ struct hpsb_iso_packet_info {
 	__u8 tag;
 	__u8 sy;
 
-	/*
-	 * length in bytes of the packet including header/trailer.
-	 * MUST be at structure end, since the first part of this structure is also 
-	 * defined in raw1394.h (i.e. struct raw1394_iso_packet_info), is copied to 
-	 * userspace and is accessed there through libraw1394. 
-	 */
+	/* length in bytes of the packet including header/trailer.
+	 * MUST be at structure end, since the first part of this structure is
+	 * also defined in raw1394.h (i.e. struct raw1394_iso_packet_info), is
+	 * copied to userspace and is accessed there through libraw1394. */
 	__u16 total_len;
 };
 
@@ -75,8 +80,8 @@ struct hpsb_iso {
 	void *hostdata;
 
 	/* a function to be called (from interrupt context) after
-           outgoing packets have been sent, or incoming packets have
-           arrived */
+	 * outgoing packets have been sent, or incoming packets have
+	 * arrived */
 	void (*callback)(struct hpsb_iso*);
 
 	/* wait for buffer space */
@@ -88,7 +93,7 @@ struct hpsb_iso {
 
 
 	/* greatest # of packets between interrupts - controls
-	   the maximum latency of the buffer */
+	 * the maximum latency of the buffer */
 	int irq_interval;
 
 	/* the buffer for packet data payloads */
@@ -112,8 +117,8 @@ struct hpsb_iso {
 	int pkt_dma;
 
 	/* how many packets, starting at first_packet:
-	   (transmit) are ready to be filled with data
-	   (receive)  contain received data */
+	 * (transmit) are ready to be filled with data
+	 * (receive)  contain received data */
 	int n_ready_packets;
 
 	/* how many times the buffer has overflowed or underflowed */
@@ -134,7 +139,7 @@ struct hpsb_iso {
 	int start_cycle;
 
 	/* cycle at which next packet will be transmitted,
-	   -1 if not known */
+	 * -1 if not known */
 	int xmit_cycle;
 
 	/* ringbuffer of packet descriptors in regular kernel memory
@@ -170,25 +175,30 @@ int hpsb_iso_recv_unlisten_channel(struct hpsb_iso *iso, unsigned char channel);
 int hpsb_iso_recv_set_channel_mask(struct hpsb_iso *iso, u64 mask);
 
 /* start/stop DMA */
-int hpsb_iso_xmit_start(struct hpsb_iso *iso, int start_on_cycle, int prebuffer);
-int hpsb_iso_recv_start(struct hpsb_iso *iso, int start_on_cycle, int tag_mask, int sync);
+int hpsb_iso_xmit_start(struct hpsb_iso *iso, int start_on_cycle,
+			int prebuffer);
+int hpsb_iso_recv_start(struct hpsb_iso *iso, int start_on_cycle,
+			int tag_mask, int sync);
 void hpsb_iso_stop(struct hpsb_iso *iso);
 
 /* deallocate buffer and DMA context */
 void hpsb_iso_shutdown(struct hpsb_iso *iso);
 
-/* queue a packet for transmission. 'offset' is relative to the beginning of the
-   DMA buffer, where the packet's data payload should already have been placed */
-int hpsb_iso_xmit_queue_packet(struct hpsb_iso *iso, u32 offset, u16 len, u8 tag, u8 sy);
+/* queue a packet for transmission.
+ * 'offset' is relative to the beginning of the DMA buffer, where the packet's
+ * data payload should already have been placed. */
+int hpsb_iso_xmit_queue_packet(struct hpsb_iso *iso, u32 offset, u16 len,
+			       u8 tag, u8 sy);
 
 /* wait until all queued packets have been transmitted to the bus */
 int hpsb_iso_xmit_sync(struct hpsb_iso *iso);
 
 /* N packets have been read out of the buffer, re-use the buffer space */
-int  hpsb_iso_recv_release_packets(struct hpsb_iso *recv, unsigned int n_packets);
+int  hpsb_iso_recv_release_packets(struct hpsb_iso *recv,
+				   unsigned int n_packets);
 
 /* check for arrival of new packets immediately (even if irq_interval
-   has not yet been reached) */
+ * has not yet been reached) */
 int hpsb_iso_recv_flush(struct hpsb_iso *iso);
 
 /* returns # of packets ready to send or receive */
@@ -197,14 +207,15 @@ int hpsb_iso_n_ready(struct hpsb_iso *iso);
 /* the following are callbacks available to low-level drivers */
 
 /* call after a packet has been transmitted to the bus (interrupt context is OK)
-   'cycle' is the _exact_ cycle the packet was sent on
-   'error' should be non-zero if some sort of error occurred when sending the packet
-*/
+ * 'cycle' is the _exact_ cycle the packet was sent on
+ * 'error' should be non-zero if some sort of error occurred when sending the
+ *  packet */
 void hpsb_iso_packet_sent(struct hpsb_iso *iso, int cycle, int error);
 
 /* call after a packet has been received (interrupt context OK) */
 void hpsb_iso_packet_received(struct hpsb_iso *iso, u32 offset, u16 len,
-			      u16 total_len, u16 cycle, u8 channel, u8 tag, u8 sy);
+			      u16 total_len, u16 cycle, u8 channel, u8 tag,
+			      u8 sy);
 
 /* call to wake waiting processes after buffer space has opened up. */
 void hpsb_iso_wake(struct hpsb_iso *iso);
diff --git a/drivers/ieee1394/nodemgr.c b/drivers/ieee1394/nodemgr.c
index d541b508a159..3e7974c57443 100644
--- a/drivers/ieee1394/nodemgr.c
+++ b/drivers/ieee1394/nodemgr.c
@@ -12,26 +12,23 @@
 #include <linux/kernel.h>
 #include <linux/list.h>
 #include <linux/slab.h>
-#include <linux/smp_lock.h>
-#include <linux/interrupt.h>
-#include <linux/kmod.h>
-#include <linux/completion.h>
 #include <linux/delay.h>
-#include <linux/pci.h>
+#include <linux/kthread.h>
 #include <linux/moduleparam.h>
 #include <asm/atomic.h>
 
-#include "ieee1394_types.h"
+#include "csr.h"
+#include "highlevel.h"
+#include "hosts.h"
 #include "ieee1394.h"
 #include "ieee1394_core.h"
-#include "hosts.h"
+#include "ieee1394_hotplug.h"
+#include "ieee1394_types.h"
 #include "ieee1394_transactions.h"
-#include "highlevel.h"
-#include "csr.h"
 #include "nodemgr.h"
 
 static int ignore_drivers;
-module_param(ignore_drivers, int, 0444);
+module_param(ignore_drivers, int, S_IRUGO | S_IWUSR);
 MODULE_PARM_DESC(ignore_drivers, "Disable automatic probing for drivers.");
 
 struct nodemgr_csr_info {
@@ -71,7 +68,7 @@ static int nodemgr_check_speed(struct nodemgr_csr_info *ci, u64 addr,
 	u8 i, *speed, old_speed, good_speed;
 	int ret;
 
-	speed = ci->host->speed + NODEID_TO_NODE(ci->nodeid);
+	speed = &(ci->host->speed[NODEID_TO_NODE(ci->nodeid)]);
 	old_speed = *speed;
 	good_speed = IEEE1394_SPEED_MAX + 1;
 
@@ -161,16 +158,12 @@ static struct csr1212_bus_ops nodemgr_csr_ops = {
  * but now we are much simpler because of the LDM.
  */
 
-static DECLARE_MUTEX(nodemgr_serialize);
+static DEFINE_MUTEX(nodemgr_serialize);
 
 struct host_info {
 	struct hpsb_host *host;
 	struct list_head list;
-	struct completion exited;
-	struct semaphore reset_sem;
-	int pid;
-	char daemon_name[15];
-	int kill_me;
+	struct task_struct *thread;
 };
 
 static int nodemgr_bus_match(struct device * dev, struct device_driver * drv);
@@ -334,34 +327,44 @@ static ssize_t fw_show_ne_bus_options(struct device *dev, struct device_attribut
 static DEVICE_ATTR(bus_options,S_IRUGO,fw_show_ne_bus_options,NULL);
 
 
-/* tlabels_free, tlabels_allocations, tlabels_mask are read non-atomically
- * here, therefore displayed values may be occasionally wrong. */
-static ssize_t fw_show_ne_tlabels_free(struct device *dev, struct device_attribute *attr, char *buf)
+#ifdef HPSB_DEBUG_TLABELS
+static ssize_t fw_show_ne_tlabels_free(struct device *dev,
+				       struct device_attribute *attr, char *buf)
 {
 	struct node_entry *ne = container_of(dev, struct node_entry, device);
-	return sprintf(buf, "%d\n", 64 - bitmap_weight(ne->tpool->pool, 64));
-}
-static DEVICE_ATTR(tlabels_free,S_IRUGO,fw_show_ne_tlabels_free,NULL);
+	unsigned long flags;
+	unsigned long *tp = ne->host->tl_pool[NODEID_TO_NODE(ne->nodeid)].map;
+	int tf;
 
+	spin_lock_irqsave(&hpsb_tlabel_lock, flags);
+	tf = 64 - bitmap_weight(tp, 64);
+	spin_unlock_irqrestore(&hpsb_tlabel_lock, flags);
 
-static ssize_t fw_show_ne_tlabels_allocations(struct device *dev, struct device_attribute *attr, char *buf)
-{
-	struct node_entry *ne = container_of(dev, struct node_entry, device);
-	return sprintf(buf, "%u\n", ne->tpool->allocations);
+	return sprintf(buf, "%d\n", tf);
 }
-static DEVICE_ATTR(tlabels_allocations,S_IRUGO,fw_show_ne_tlabels_allocations,NULL);
+static DEVICE_ATTR(tlabels_free,S_IRUGO,fw_show_ne_tlabels_free,NULL);
 
 
-static ssize_t fw_show_ne_tlabels_mask(struct device *dev, struct device_attribute *attr, char *buf)
+static ssize_t fw_show_ne_tlabels_mask(struct device *dev,
+				       struct device_attribute *attr, char *buf)
 {
 	struct node_entry *ne = container_of(dev, struct node_entry, device);
+	unsigned long flags;
+	unsigned long *tp = ne->host->tl_pool[NODEID_TO_NODE(ne->nodeid)].map;
+	u64 tm;
+
+	spin_lock_irqsave(&hpsb_tlabel_lock, flags);
 #if (BITS_PER_LONG <= 32)
-	return sprintf(buf, "0x%08lx%08lx\n", ne->tpool->pool[0], ne->tpool->pool[1]);
+	tm = ((u64)tp[0] << 32) + tp[1];
 #else
-	return sprintf(buf, "0x%016lx\n", ne->tpool->pool[0]);
+	tm = tp[0];
 #endif
+	spin_unlock_irqrestore(&hpsb_tlabel_lock, flags);
+
+	return sprintf(buf, "0x%016llx\n", tm);
 }
 static DEVICE_ATTR(tlabels_mask, S_IRUGO, fw_show_ne_tlabels_mask, NULL);
+#endif /* HPSB_DEBUG_TLABELS */
 
 
 static ssize_t fw_set_ignore_driver(struct device *dev, struct device_attribute *attr, const char *buf, size_t count)
@@ -408,26 +411,11 @@ static ssize_t fw_get_destroy_node(struct bus_type *bus, char *buf)
 }
 static BUS_ATTR(destroy_node, S_IWUSR | S_IRUGO, fw_get_destroy_node, fw_set_destroy_node);
 
-static int nodemgr_rescan_bus_thread(void *__unused)
-{
-	/* No userlevel access needed */
-	daemonize("kfwrescan");
-
-	bus_rescan_devices(&ieee1394_bus_type);
-
-	return 0;
-}
 
 static ssize_t fw_set_rescan(struct bus_type *bus, const char *buf, size_t count)
 {
-	int state = simple_strtoul(buf, NULL, 10);
-
-	/* Don't wait for this, or care about errors. Root could do
-	 * something stupid and spawn this a lot of times, but that's
-	 * root's fault. */
-	if (state == 1)
-		kernel_thread(nodemgr_rescan_bus_thread, NULL, CLONE_KERNEL);
-
+	if (simple_strtoul(buf, NULL, 10) == 1)
+		bus_rescan_devices(&ieee1394_bus_type);
 	return count;
 }
 static ssize_t fw_get_rescan(struct bus_type *bus, char *buf)
@@ -483,9 +471,10 @@ static struct device_attribute *const fw_ne_attrs[] = {
 	&dev_attr_ne_vendor_id,
 	&dev_attr_ne_nodeid,
 	&dev_attr_bus_options,
+#ifdef HPSB_DEBUG_TLABELS
 	&dev_attr_tlabels_free,
-	&dev_attr_tlabels_allocations,
 	&dev_attr_tlabels_mask,
+#endif
 };
 
 
@@ -804,8 +793,6 @@ static struct node_entry *nodemgr_create_node(octlet_t guid, struct csr1212_csr
 	if (!ne)
 		return NULL;
 
-	ne->tpool = &host->tpool[nodeid & NODE_MASK];
-
 	ne->host = host;
 	ne->nodeid = nodeid;
 	ne->generation = generation;
@@ -1251,6 +1238,7 @@ static void nodemgr_node_scan_one(struct host_info *hi,
 	octlet_t guid;
 	struct csr1212_csr *csr;
 	struct nodemgr_csr_info *ci;
+	u8 *speed;
 
 	ci = kmalloc(sizeof(*ci), GFP_KERNEL);
 	if (!ci)
@@ -1259,8 +1247,12 @@ static void nodemgr_node_scan_one(struct host_info *hi,
 	ci->host = host;
 	ci->nodeid = nodeid;
 	ci->generation = generation;
-	ci->speed_unverified =
-		host->speed[NODEID_TO_NODE(nodeid)] > IEEE1394_SPEED_100;
+
+	/* Prepare for speed probe which occurs when reading the ROM */
+	speed = &(host->speed[NODEID_TO_NODE(nodeid)]);
+	if (*speed > host->csr.lnk_spd)
+		*speed = host->csr.lnk_spd;
+	ci->speed_unverified = *speed > IEEE1394_SPEED_100;
 
 	/* We need to detect when the ConfigROM's generation has changed,
 	 * so we only update the node's info when it needs to be.  */
@@ -1300,8 +1292,6 @@ static void nodemgr_node_scan_one(struct host_info *hi,
 		nodemgr_create_node(guid, csr, hi, nodeid, generation);
 	else
 		nodemgr_update_node(ne, csr, hi, nodeid, generation);
-
-	return;
 }
 
 
@@ -1326,6 +1316,7 @@ static void nodemgr_node_scan(struct host_info *hi, int generation)
 }
 
 
+/* Caller needs to hold nodemgr_ud_class.subsys.rwsem as reader. */
 static void nodemgr_suspend_ne(struct node_entry *ne)
 {
 	struct class_device *cdev;
@@ -1361,6 +1352,7 @@ static void nodemgr_resume_ne(struct node_entry *ne)
 	ne->in_limbo = 0;
 	device_remove_file(&ne->device, &dev_attr_ne_in_limbo);
 
+	down_read(&nodemgr_ud_class.subsys.rwsem);
 	down_read(&ne->device.bus->subsys.rwsem);
 	list_for_each_entry(cdev, &nodemgr_ud_class.children, node) {
 		ud = container_of(cdev, struct unit_directory, class_dev);
@@ -1372,21 +1364,21 @@ static void nodemgr_resume_ne(struct node_entry *ne)
 			ud->device.driver->resume(&ud->device);
 	}
 	up_read(&ne->device.bus->subsys.rwsem);
+	up_read(&nodemgr_ud_class.subsys.rwsem);
 
 	HPSB_DEBUG("Node resumed: ID:BUS[" NODE_BUS_FMT "]  GUID[%016Lx]",
 		   NODE_BUS_ARGS(ne->host, ne->nodeid), (unsigned long long)ne->guid);
 }
 
 
+/* Caller needs to hold nodemgr_ud_class.subsys.rwsem as reader. */
 static void nodemgr_update_pdrv(struct node_entry *ne)
 {
 	struct unit_directory *ud;
 	struct hpsb_protocol_driver *pdrv;
-	struct class *class = &nodemgr_ud_class;
 	struct class_device *cdev;
 
-	down_read(&class->subsys.rwsem);
-	list_for_each_entry(cdev, &class->children, node) {
+	list_for_each_entry(cdev, &nodemgr_ud_class.children, node) {
 		ud = container_of(cdev, struct unit_directory, class_dev);
 		if (ud->ne != ne || !ud->device.driver)
 			continue;
@@ -1399,7 +1391,6 @@ static void nodemgr_update_pdrv(struct node_entry *ne)
 			up_write(&ud->device.bus->subsys.rwsem);
 		}
 	}
-	up_read(&class->subsys.rwsem);
 }
 
 
@@ -1430,6 +1421,8 @@ static void nodemgr_irm_write_bc(struct node_entry *ne, int generation)
 }
 
 
+/* Caller needs to hold nodemgr_ud_class.subsys.rwsem as reader because the
+ * calls to nodemgr_update_pdrv() and nodemgr_suspend_ne() here require it. */
 static void nodemgr_probe_ne(struct host_info *hi, struct node_entry *ne, int generation)
 {
 	struct device *dev;
@@ -1492,9 +1485,8 @@ static void nodemgr_node_probe(struct host_info *hi, int generation)
 	/* If we had a bus reset while we were scanning the bus, it is
 	 * possible that we did not probe all nodes.  In that case, we
 	 * skip the clean up for now, since we could remove nodes that
-	 * were still on the bus.  The bus reset increased hi->reset_sem,
-	 * so there's a bus scan pending which will do the clean up
-	 * eventually.
+	 * were still on the bus.  Another bus scan is pending which will
+	 * do the clean up eventually.
 	 *
 	 * Now let's tell the bus to rescan our devices. This may seem
 	 * like overhead, but the driver-model core will only scan a
@@ -1622,41 +1614,37 @@ static int nodemgr_host_thread(void *__hi)
 {
 	struct host_info *hi = (struct host_info *)__hi;
 	struct hpsb_host *host = hi->host;
-	int reset_cycles = 0;
-
-	/* No userlevel access needed */
-	daemonize(hi->daemon_name);
+	unsigned int g, generation = get_hpsb_generation(host) - 1;
+	int i, reset_cycles = 0;
 
 	/* Setup our device-model entries */
 	nodemgr_create_host_dev_files(host);
 
-	/* Sit and wait for a signal to probe the nodes on the bus. This
-	 * happens when we get a bus reset. */
-	while (1) {
-		unsigned int generation = 0;
-		int i;
+	for (;;) {
+		/* Sleep until next bus reset */
+		set_current_state(TASK_INTERRUPTIBLE);
+		if (get_hpsb_generation(host) == generation)
+			schedule();
+		__set_current_state(TASK_RUNNING);
+
+		/* Thread may have been woken up to freeze or to exit */
+		if (try_to_freeze())
+			continue;
+		if (kthread_should_stop())
+			goto exit;
 
-		if (down_interruptible(&hi->reset_sem) ||
-		    down_interruptible(&nodemgr_serialize)) {
+		if (mutex_lock_interruptible(&nodemgr_serialize)) {
 			if (try_to_freeze())
 				continue;
-			printk("NodeMgr: received unexpected signal?!\n" );
-			break;
-		}
-
-		if (hi->kill_me) {
-			up(&nodemgr_serialize);
-			break;
+			goto exit;
 		}
 
 		/* Pause for 1/4 second in 1/16 second intervals,
 		 * to make sure things settle down. */
+		g = get_hpsb_generation(host);
 		for (i = 0; i < 4 ; i++) {
-			set_current_state(TASK_INTERRUPTIBLE);
-			if (msleep_interruptible(63)) {
-				up(&nodemgr_serialize);
-				goto caught_signal;
-			}
+			if (msleep_interruptible(63) || kthread_should_stop())
+				goto unlock_exit;
 
 			/* Now get the generation in which the node ID's we collect
 			 * are valid.  During the bus scan we will use this generation
@@ -1667,20 +1655,14 @@ static int nodemgr_host_thread(void *__hi)
 
 			/* If we get a reset before we are done waiting, then
 			 * start the the waiting over again */
-			while (!down_trylock(&hi->reset_sem))
-				i = 0;
-
-			/* Check the kill_me again */
-			if (hi->kill_me) {
-				up(&nodemgr_serialize);
-				goto caught_signal;
-			}
+			if (generation != g)
+				g = generation, i = 0;
 		}
 
 		if (!nodemgr_check_irm_capability(host, reset_cycles) ||
 		    !nodemgr_do_irm_duties(host, reset_cycles)) {
 			reset_cycles++;
-			up(&nodemgr_serialize);
+			mutex_unlock(&nodemgr_serialize);
 			continue;
 		}
 		reset_cycles = 0;
@@ -1698,13 +1680,13 @@ static int nodemgr_host_thread(void *__hi)
 		/* Update some of our sysfs symlinks */
 		nodemgr_update_host_dev_links(host);
 
-		up(&nodemgr_serialize);
+		mutex_unlock(&nodemgr_serialize);
 	}
-
-caught_signal:
+unlock_exit:
+	mutex_unlock(&nodemgr_serialize);
+exit:
 	HPSB_VERBOSE("NodeMgr: Exiting thread");
-
-	complete_and_exit(&hi->exited, 0);
+	return 0;
 }
 
 int nodemgr_for_each_host(void *__data, int (*cb)(struct hpsb_host *, void *))
@@ -1764,41 +1746,27 @@ static void nodemgr_add_host(struct hpsb_host *host)
 	struct host_info *hi;
 
 	hi = hpsb_create_hostinfo(&nodemgr_highlevel, host, sizeof(*hi));
-
 	if (!hi) {
-		HPSB_ERR ("NodeMgr: out of memory in add host");
+		HPSB_ERR("NodeMgr: out of memory in add host");
 		return;
 	}
-
 	hi->host = host;
-	init_completion(&hi->exited);
-        sema_init(&hi->reset_sem, 0);
-
-	sprintf(hi->daemon_name, "knodemgrd_%d", host->id);
-
-	hi->pid = kernel_thread(nodemgr_host_thread, hi, CLONE_KERNEL);
-
-	if (hi->pid < 0) {
-		HPSB_ERR ("NodeMgr: failed to start %s thread for %s",
-			  hi->daemon_name, host->driver->name);
+	hi->thread = kthread_run(nodemgr_host_thread, hi, "knodemgrd_%d",
+				 host->id);
+	if (IS_ERR(hi->thread)) {
+		HPSB_ERR("NodeMgr: cannot start thread for host %d", host->id);
 		hpsb_destroy_hostinfo(&nodemgr_highlevel, host);
-		return;
 	}
-
-	return;
 }
 
 static void nodemgr_host_reset(struct hpsb_host *host)
 {
 	struct host_info *hi = hpsb_get_hostinfo(&nodemgr_highlevel, host);
 
-	if (hi != NULL) {
-		HPSB_VERBOSE("NodeMgr: Processing host reset for %s", hi->daemon_name);
-		up(&hi->reset_sem);
-	} else
-		HPSB_ERR ("NodeMgr: could not process reset of unused host");
-
-	return;
+	if (hi) {
+		HPSB_VERBOSE("NodeMgr: Processing reset for host %d", host->id);
+		wake_up_process(hi->thread);
+	}
 }
 
 static void nodemgr_remove_host(struct hpsb_host *host)
@@ -1806,18 +1774,9 @@ static void nodemgr_remove_host(struct hpsb_host *host)
 	struct host_info *hi = hpsb_get_hostinfo(&nodemgr_highlevel, host);
 
 	if (hi) {
-		if (hi->pid >= 0) {
-			hi->kill_me = 1;
-			mb();
-			up(&hi->reset_sem);
-			wait_for_completion(&hi->exited);
-			nodemgr_remove_host_dev(&host->device);
-		}
-	} else
-		HPSB_ERR("NodeMgr: host %s does not exist, cannot remove",
-			 host->driver->name);
-
-	return;
+		kthread_stop(hi->thread);
+		nodemgr_remove_host_dev(&host->device);
+	}
 }
 
 static struct hpsb_highlevel nodemgr_highlevel = {
diff --git a/drivers/ieee1394/nodemgr.h b/drivers/ieee1394/nodemgr.h
index 0b26616e16c3..0e1e7d930783 100644
--- a/drivers/ieee1394/nodemgr.h
+++ b/drivers/ieee1394/nodemgr.h
@@ -21,9 +21,15 @@
 #define _IEEE1394_NODEMGR_H
 
 #include <linux/device.h>
-#include "csr1212.h"
+#include <asm/types.h>
+
 #include "ieee1394_core.h"
-#include "ieee1394_hotplug.h"
+#include "ieee1394_types.h"
+
+struct csr1212_csr;
+struct csr1212_keyval;
+struct hpsb_host;
+struct ieee1394_device_id;
 
 /* '1' '3' '9' '4' in ASCII */
 #define IEEE1394_BUSID_MAGIC	__constant_cpu_to_be32(0x31333934)
@@ -44,7 +50,6 @@ struct bus_options {
 	u16	max_rec;	/* Maximum packet size node can receive */
 };
 
-
 #define UNIT_DIRECTORY_VENDOR_ID		0x01
 #define UNIT_DIRECTORY_MODEL_ID			0x02
 #define UNIT_DIRECTORY_SPECIFIER_ID		0x04
@@ -59,8 +64,8 @@ struct bus_options {
  * unit directory for each of these protocols.
  */
 struct unit_directory {
-	struct node_entry *ne;  /* The node which this directory belongs to */
-	octlet_t address;       /* Address of the unit directory on the node */
+	struct node_entry *ne;	/* The node which this directory belongs to */
+	octlet_t address;	/* Address of the unit directory on the node */
 	u8 flags;		/* Indicates which entries were read */
 
 	quadlet_t vendor_id;
@@ -79,11 +84,10 @@ struct unit_directory {
 	int length;		/* Number of quadlets */
 
 	struct device device;
-
 	struct class_device class_dev;
 
 	struct csr1212_keyval *ud_kv;
-	u32 lun;                /* logical unit number immediate value */
+	u32 lun;		/* logical unit number immediate value */
 };
 
 struct node_entry {
@@ -103,10 +107,8 @@ struct node_entry {
 	const char *vendor_oui;
 
 	u32 capabilities;
-	struct hpsb_tlabel_pool *tpool;
 
 	struct device device;
-
 	struct class_device class_dev;
 
 	/* Means this node is not attached anymore */
@@ -153,8 +155,8 @@ static inline int hpsb_node_entry_valid(struct node_entry *ne)
 /*
  * This will fill in the given, pre-initialised hpsb_packet with the current
  * information from the node entry (host, node ID, generation number).  It will
- * return false if the node owning the GUID is not accessible (and not modify the
- * hpsb_packet) and return true otherwise.
+ * return false if the node owning the GUID is not accessible (and not modify
+ * the hpsb_packet) and return true otherwise.
  *
  * Note that packet sending may still fail in hpsb_send_packet if a bus reset
  * happens while you are trying to set up the packet (due to obsolete generation
@@ -170,16 +172,13 @@ int hpsb_node_write(struct node_entry *ne, u64 addr,
 int hpsb_node_lock(struct node_entry *ne, u64 addr,
 		   int extcode, quadlet_t *data, quadlet_t arg);
 
-
 /* Iterate the hosts, calling a given function with supplied data for each
  * host. */
 int nodemgr_for_each_host(void *__data, int (*cb)(struct hpsb_host *, void *));
 
-
 int init_ieee1394_nodemgr(void);
 void cleanup_ieee1394_nodemgr(void);
 
-
 /* The template for a host device */
 extern struct device nodemgr_dev_template_host;
 
diff --git a/drivers/ieee1394/ohci1394.c b/drivers/ieee1394/ohci1394.c
index 448df2773377..8fd0030475ba 100644
--- a/drivers/ieee1394/ohci1394.c
+++ b/drivers/ieee1394/ohci1394.c
@@ -136,7 +136,7 @@
 #define DBGMSG(fmt, args...) \
 printk(KERN_INFO "%s: fw-host%d: " fmt "\n" , OHCI1394_DRIVER_NAME, ohci->host->id , ## args)
 #else
-#define DBGMSG(fmt, args...)
+#define DBGMSG(fmt, args...) do {} while (0)
 #endif
 
 #ifdef CONFIG_IEEE1394_OHCI_DMA_DEBUG
@@ -148,8 +148,8 @@ printk(KERN_INFO "%s: fw-host%d: " fmt "\n" , OHCI1394_DRIVER_NAME, ohci->host->
 		--global_outstanding_dmas, ## args)
 static int global_outstanding_dmas = 0;
 #else
-#define OHCI_DMA_ALLOC(fmt, args...)
-#define OHCI_DMA_FREE(fmt, args...)
+#define OHCI_DMA_ALLOC(fmt, args...) do {} while (0)
+#define OHCI_DMA_FREE(fmt, args...) do {} while (0)
 #endif
 
 /* print general (card independent) information */
@@ -181,36 +181,35 @@ static int alloc_dma_trm_ctx(struct ti_ohci *ohci, struct dma_trm_ctx *d,
 static void ohci1394_pci_remove(struct pci_dev *pdev);
 
 #ifndef __LITTLE_ENDIAN
-static unsigned hdr_sizes[] =
-{
+const static size_t hdr_sizes[] = {
 	3,	/* TCODE_WRITEQ */
 	4,	/* TCODE_WRITEB */
 	3,	/* TCODE_WRITE_RESPONSE */
-	0,	/* ??? */
+	0,	/* reserved */
 	3,	/* TCODE_READQ */
 	4,	/* TCODE_READB */
 	3,	/* TCODE_READQ_RESPONSE */
 	4,	/* TCODE_READB_RESPONSE */
-	1,	/* TCODE_CYCLE_START (???) */
+	1,	/* TCODE_CYCLE_START */
 	4,	/* TCODE_LOCK_REQUEST */
 	2,	/* TCODE_ISO_DATA */
 	4,	/* TCODE_LOCK_RESPONSE */
+		/* rest is reserved or link-internal */
 };
 
-/* Swap headers */
-static inline void packet_swab(quadlet_t *data, int tcode)
+static inline void header_le32_to_cpu(quadlet_t *data, unsigned char tcode)
 {
-	size_t size = hdr_sizes[tcode];
+	size_t size;
 
-	if (tcode > TCODE_LOCK_RESPONSE || hdr_sizes[tcode] == 0)
+	if (unlikely(tcode >= ARRAY_SIZE(hdr_sizes)))
 		return;
 
+	size = hdr_sizes[tcode];
 	while (size--)
-		data[size] = swab32(data[size]);
+		data[size] = le32_to_cpu(data[size]);
 }
 #else
-/* Don't waste cycles on same sex byte swaps */
-#define packet_swab(w,x)
+#define header_le32_to_cpu(w,x) do {} while (0)
 #endif /* !LITTLE_ENDIAN */
 
 /***********************************
@@ -701,7 +700,7 @@ static void insert_packet(struct ti_ohci *ohci,
 				d->prg_cpu[idx]->data[2] = packet->header[2];
 				d->prg_cpu[idx]->data[3] = packet->header[3];
 			}
-			packet_swab(d->prg_cpu[idx]->data, packet->tcode);
+			header_le32_to_cpu(d->prg_cpu[idx]->data, packet->tcode);
                 }
 
                 if (packet->data_size) { /* block transmit */
@@ -777,7 +776,7 @@ static void insert_packet(struct ti_ohci *ohci,
                 d->prg_cpu[idx]->data[0] = packet->speed_code<<16 |
                         (packet->header[0] & 0xFFFF);
                 d->prg_cpu[idx]->data[1] = packet->header[0] & 0xFFFF0000;
-		packet_swab(d->prg_cpu[idx]->data, packet->tcode);
+		header_le32_to_cpu(d->prg_cpu[idx]->data, packet->tcode);
 
                 d->prg_cpu[idx]->begin.control =
 			cpu_to_le32(DMA_CTL_OUTPUT_MORE |
@@ -2598,8 +2597,9 @@ static const int TCODE_SIZE[16] = {20, 0, 16, -1, 16, 20, 20, 0,
  * Determine the length of a packet in the buffer
  * Optimization suggested by Pascal Drolet <pascal.drolet@informission.ca>
  */
-static __inline__ int packet_length(struct dma_rcv_ctx *d, int idx, quadlet_t *buf_ptr,
-			 int offset, unsigned char tcode, int noswap)
+static inline int packet_length(struct dma_rcv_ctx *d, int idx,
+				quadlet_t *buf_ptr, int offset,
+				unsigned char tcode, int noswap)
 {
 	int length = -1;
 
@@ -2730,7 +2730,7 @@ static void dma_rcv_tasklet (unsigned long data)
 		 * bus reset. We always ignore it.  */
 		if (tcode != OHCI1394_TCODE_PHY) {
 			if (!ohci->no_swap_incoming)
-				packet_swab(d->spb, tcode);
+				header_le32_to_cpu(d->spb, tcode);
 			DBGMSG("Packet received from node"
 				" %d ack=0x%02X spd=%d tcode=0x%X"
 				" length=%d ctx=%d tlabel=%d",
@@ -2738,7 +2738,7 @@ static void dma_rcv_tasklet (unsigned long data)
 				(cond_le32_to_cpu(d->spb[length/4-1], ohci->no_swap_incoming)>>16)&0x1f,
 				(cond_le32_to_cpu(d->spb[length/4-1], ohci->no_swap_incoming)>>21)&0x3,
 				tcode, length, d->ctx,
-				(cond_le32_to_cpu(d->spb[0], ohci->no_swap_incoming)>>10)&0x3f);
+				(d->spb[0]>>10)&0x3f);
 
 			ack = (((cond_le32_to_cpu(d->spb[length/4-1], ohci->no_swap_incoming)>>16)&0x1f)
 				== 0x11) ? 1 : 0;
@@ -3529,9 +3529,10 @@ static void ohci1394_pci_remove(struct pci_dev *pdev)
 		put_device(dev);
 }
 
-
+#ifdef CONFIG_PM
 static int ohci1394_pci_resume (struct pci_dev *pdev)
 {
+/* PowerMac resume code comes first */
 #ifdef CONFIG_PPC_PMAC
 	if (machine_is(powermac)) {
 		struct device_node *of_node;
@@ -3543,17 +3544,23 @@ static int ohci1394_pci_resume (struct pci_dev *pdev)
 	}
 #endif /* CONFIG_PPC_PMAC */
 
+	pci_set_power_state(pdev, PCI_D0);
 	pci_restore_state(pdev);
-	pci_enable_device(pdev);
-
-	return 0;
+	return pci_enable_device(pdev);
 }
 
-
 static int ohci1394_pci_suspend (struct pci_dev *pdev, pm_message_t state)
 {
-	pci_save_state(pdev);
+	int err;
+
+	err = pci_save_state(pdev);
+	if (err)
+		goto out;
+	err = pci_set_power_state(pdev, pci_choose_state(pdev, state));
+	if (err)
+		goto out;
 
+/* PowerMac suspend code comes last */
 #ifdef CONFIG_PPC_PMAC
 	if (machine_is(powermac)) {
 		struct device_node *of_node;
@@ -3563,11 +3570,11 @@ static int ohci1394_pci_suspend (struct pci_dev *pdev, pm_message_t state)
 		if (of_node)
 			pmac_call_feature(PMAC_FTR_1394_ENABLE, of_node, 0, 0);
 	}
-#endif
-
-	return 0;
+#endif /* CONFIG_PPC_PMAC */
+out:
+	return err;
 }
-
+#endif /* CONFIG_PM */
 
 #define PCI_CLASS_FIREWIRE_OHCI     ((PCI_CLASS_SERIAL_FIREWIRE << 8) | 0x10)
 
@@ -3590,8 +3597,10 @@ static struct pci_driver ohci1394_pci_driver = {
 	.id_table =	ohci1394_pci_tbl,
 	.probe =	ohci1394_pci_probe,
 	.remove =	ohci1394_pci_remove,
+#ifdef CONFIG_PM
 	.resume =	ohci1394_pci_resume,
 	.suspend =	ohci1394_pci_suspend,
+#endif
 };
 
 /***********************************
@@ -3718,5 +3727,7 @@ static int __init ohci1394_init(void)
 	return pci_register_driver(&ohci1394_pci_driver);
 }
 
-module_init(ohci1394_init);
+/* Register before most other device drivers.
+ * Useful for remote debugging via physical DMA, e.g. using firescope. */
+fs_initcall(ohci1394_init);
 module_exit(ohci1394_cleanup);
diff --git a/drivers/ieee1394/raw1394-private.h b/drivers/ieee1394/raw1394-private.h
index c93587be9cab..c7731d1bcd89 100644
--- a/drivers/ieee1394/raw1394-private.h
+++ b/drivers/ieee1394/raw1394-private.h
@@ -29,9 +29,8 @@ struct file_info {
 
         struct list_head req_pending;
         struct list_head req_complete;
-        struct semaphore complete_sem;
         spinlock_t reqlists_lock;
-        wait_queue_head_t poll_wait_complete;
+        wait_queue_head_t wait_complete;
 
         struct list_head addr_list;
 
diff --git a/drivers/ieee1394/raw1394.c b/drivers/ieee1394/raw1394.c
index 571ea68c0cf2..5ec4f5eb6b19 100644
--- a/drivers/ieee1394/raw1394.c
+++ b/drivers/ieee1394/raw1394.c
@@ -44,14 +44,15 @@
 #include <linux/compat.h>
 
 #include "csr1212.h"
+#include "highlevel.h"
+#include "hosts.h"
 #include "ieee1394.h"
-#include "ieee1394_types.h"
 #include "ieee1394_core.h"
-#include "nodemgr.h"
-#include "hosts.h"
-#include "highlevel.h"
-#include "iso.h"
+#include "ieee1394_hotplug.h"
 #include "ieee1394_transactions.h"
+#include "ieee1394_types.h"
+#include "iso.h"
+#include "nodemgr.h"
 #include "raw1394.h"
 #include "raw1394-private.h"
 
@@ -66,7 +67,7 @@
 #define DBGMSG(fmt, args...) \
 printk(KERN_INFO "raw1394:" fmt "\n" , ## args)
 #else
-#define DBGMSG(fmt, args...)
+#define DBGMSG(fmt, args...) do {} while (0)
 #endif
 
 static LIST_HEAD(host_info_list);
@@ -132,10 +133,9 @@ static void free_pending_request(struct pending_request *req)
 static void __queue_complete_req(struct pending_request *req)
 {
 	struct file_info *fi = req->file_info;
-	list_move_tail(&req->list, &fi->req_complete);
 
-	up(&fi->complete_sem);
-	wake_up_interruptible(&fi->poll_wait_complete);
+	list_move_tail(&req->list, &fi->req_complete);
+ 	wake_up(&fi->wait_complete);
 }
 
 static void queue_complete_req(struct pending_request *req)
@@ -463,13 +463,36 @@ raw1394_compat_read(const char __user *buf, struct raw1394_request *r)
 
 #endif
 
+/* get next completed request  (caller must hold fi->reqlists_lock) */
+static inline struct pending_request *__next_complete_req(struct file_info *fi)
+{
+	struct list_head *lh;
+	struct pending_request *req = NULL;
+
+	if (!list_empty(&fi->req_complete)) {
+		lh = fi->req_complete.next;
+		list_del(lh);
+		req = list_entry(lh, struct pending_request, list);
+	}
+	return req;
+}
+
+/* atomically get next completed request */
+static struct pending_request *next_complete_req(struct file_info *fi)
+{
+	unsigned long flags;
+	struct pending_request *req;
+
+	spin_lock_irqsave(&fi->reqlists_lock, flags);
+	req = __next_complete_req(fi);
+	spin_unlock_irqrestore(&fi->reqlists_lock, flags);
+	return req;
+}
 
 static ssize_t raw1394_read(struct file *file, char __user * buffer,
 			    size_t count, loff_t * offset_is_ignored)
 {
-	unsigned long flags;
 	struct file_info *fi = (struct file_info *)file->private_data;
-	struct list_head *lh;
 	struct pending_request *req;
 	ssize_t ret;
 
@@ -487,22 +510,21 @@ static ssize_t raw1394_read(struct file *file, char __user * buffer,
 	}
 
 	if (file->f_flags & O_NONBLOCK) {
-		if (down_trylock(&fi->complete_sem)) {
+		if (!(req = next_complete_req(fi)))
 			return -EAGAIN;
-		}
 	} else {
-		if (down_interruptible(&fi->complete_sem)) {
+		/*
+		 * NB: We call the macro wait_event_interruptible() with a
+		 * condition argument with side effect.  This is only possible
+		 * because the side effect does not occur until the condition
+		 * became true, and wait_event_interruptible() won't evaluate
+		 * the condition again after that.
+		 */
+		if (wait_event_interruptible(fi->wait_complete,
+					     (req = next_complete_req(fi))))
 			return -ERESTARTSYS;
-		}
 	}
 
-	spin_lock_irqsave(&fi->reqlists_lock, flags);
-	lh = fi->req_complete.next;
-	list_del(lh);
-	spin_unlock_irqrestore(&fi->reqlists_lock, flags);
-
-	req = list_entry(lh, struct pending_request, list);
-
 	if (req->req.length) {
 		if (copy_to_user(int2ptr(req->req.recvb), req->data,
 				 req->req.length)) {
@@ -1752,6 +1774,7 @@ static int arm_register(struct file_info *fi, struct pending_request *req)
 	addr->notification_options |= addr->client_transactions;
 	addr->recvb = req->req.recvb;
 	addr->rec_length = (u16) ((req->req.misc >> 16) & 0xFFFF);
+
 	spin_lock_irqsave(&host_info_lock, flags);
 	hi = find_host_info(fi->host);
 	same_host = 0;
@@ -1777,9 +1800,9 @@ static int arm_register(struct file_info *fi, struct pending_request *req)
 	}
 	if (same_host) {
 		/* addressrange occupied by same host */
+		spin_unlock_irqrestore(&host_info_lock, flags);
 		vfree(addr->addr_space_buffer);
 		kfree(addr);
-		spin_unlock_irqrestore(&host_info_lock, flags);
 		return (-EALREADY);
 	}
 	/* another host with valid address-entry containing same addressrange */
@@ -1807,6 +1830,8 @@ static int arm_register(struct file_info *fi, struct pending_request *req)
 			}
 		}
 	}
+	spin_unlock_irqrestore(&host_info_lock, flags);
+
 	if (another_host) {
 		DBGMSG("another hosts entry is valid -> SUCCESS");
 		if (copy_to_user(int2ptr(req->req.recvb),
@@ -1815,11 +1840,11 @@ static int arm_register(struct file_info *fi, struct pending_request *req)
 			       " address-range-entry is invalid -> EFAULT !!!\n");
 			vfree(addr->addr_space_buffer);
 			kfree(addr);
-			spin_unlock_irqrestore(&host_info_lock, flags);
 			return (-EFAULT);
 		}
 		free_pending_request(req);	/* immediate success or fail */
 		/* INSERT ENTRY */
+		spin_lock_irqsave(&host_info_lock, flags);
 		list_add_tail(&addr->addr_list, &fi->addr_list);
 		spin_unlock_irqrestore(&host_info_lock, flags);
 		return sizeof(struct raw1394_request);
@@ -1830,15 +1855,15 @@ static int arm_register(struct file_info *fi, struct pending_request *req)
 				    req->req.address + req->req.length);
 	if (retval) {
 		/* INSERT ENTRY */
+		spin_lock_irqsave(&host_info_lock, flags);
 		list_add_tail(&addr->addr_list, &fi->addr_list);
+		spin_unlock_irqrestore(&host_info_lock, flags);
 	} else {
 		DBGMSG("arm_register failed errno: %d \n", retval);
 		vfree(addr->addr_space_buffer);
 		kfree(addr);
-		spin_unlock_irqrestore(&host_info_lock, flags);
 		return (-EALREADY);
 	}
-	spin_unlock_irqrestore(&host_info_lock, flags);
 	free_pending_request(req);	/* immediate success or fail */
 	return sizeof(struct raw1394_request);
 }
@@ -1904,10 +1929,10 @@ static int arm_unregister(struct file_info *fi, struct pending_request *req)
 	if (another_host) {
 		DBGMSG("delete entry from list -> success");
 		list_del(&addr->addr_list);
+		spin_unlock_irqrestore(&host_info_lock, flags);
 		vfree(addr->addr_space_buffer);
 		kfree(addr);
 		free_pending_request(req);	/* immediate success or fail */
-		spin_unlock_irqrestore(&host_info_lock, flags);
 		return sizeof(struct raw1394_request);
 	}
 	retval =
@@ -1949,23 +1974,19 @@ static int arm_get_buf(struct file_info *fi, struct pending_request *req)
 		    (arm_addr->end > req->req.address)) {
 			if (req->req.address + req->req.length <= arm_addr->end) {
 				offset = req->req.address - arm_addr->start;
+				spin_unlock_irqrestore(&host_info_lock, flags);
 
 				DBGMSG
 				    ("arm_get_buf copy_to_user( %08X, %p, %u )",
 				     (u32) req->req.recvb,
 				     arm_addr->addr_space_buffer + offset,
 				     (u32) req->req.length);
-
 				if (copy_to_user
 				    (int2ptr(req->req.recvb),
 				     arm_addr->addr_space_buffer + offset,
-				     req->req.length)) {
-					spin_unlock_irqrestore(&host_info_lock,
-							       flags);
+				     req->req.length))
 					return (-EFAULT);
-				}
 
-				spin_unlock_irqrestore(&host_info_lock, flags);
 				/* We have to free the request, because we
 				 * queue no response, and therefore nobody
 				 * will free it. */
@@ -2005,24 +2026,23 @@ static int arm_set_buf(struct file_info *fi, struct pending_request *req)
 		    (arm_addr->end > req->req.address)) {
 			if (req->req.address + req->req.length <= arm_addr->end) {
 				offset = req->req.address - arm_addr->start;
+				spin_unlock_irqrestore(&host_info_lock, flags);
 
 				DBGMSG
 				    ("arm_set_buf copy_from_user( %p, %08X, %u )",
 				     arm_addr->addr_space_buffer + offset,
 				     (u32) req->req.sendb,
 				     (u32) req->req.length);
-
 				if (copy_from_user
 				    (arm_addr->addr_space_buffer + offset,
 				     int2ptr(req->req.sendb),
-				     req->req.length)) {
-					spin_unlock_irqrestore(&host_info_lock,
-							       flags);
+				     req->req.length))
 					return (-EFAULT);
-				}
 
-				spin_unlock_irqrestore(&host_info_lock, flags);
-				free_pending_request(req);	/* we have to free the request, because we queue no response, and therefore nobody will free it */
+				/* We have to free the request, because we
+				 * queue no response, and therefore nobody
+				 * will free it. */
+				free_pending_request(req);
 				return sizeof(struct raw1394_request);
 			} else {
 				DBGMSG("arm_set_buf request exceeded mapping");
@@ -2744,7 +2764,7 @@ static unsigned int raw1394_poll(struct file *file, poll_table * pt)
 	unsigned int mask = POLLOUT | POLLWRNORM;
 	unsigned long flags;
 
-	poll_wait(file, &fi->poll_wait_complete, pt);
+	poll_wait(file, &fi->wait_complete, pt);
 
 	spin_lock_irqsave(&fi->reqlists_lock, flags);
 	if (!list_empty(&fi->req_complete)) {
@@ -2769,9 +2789,8 @@ static int raw1394_open(struct inode *inode, struct file *file)
 	fi->state = opened;
 	INIT_LIST_HEAD(&fi->req_pending);
 	INIT_LIST_HEAD(&fi->req_complete);
-	sema_init(&fi->complete_sem, 0);
 	spin_lock_init(&fi->reqlists_lock);
-	init_waitqueue_head(&fi->poll_wait_complete);
+	init_waitqueue_head(&fi->wait_complete);
 	INIT_LIST_HEAD(&fi->addr_list);
 
 	file->private_data = fi;
@@ -2784,7 +2803,7 @@ static int raw1394_release(struct inode *inode, struct file *file)
 	struct file_info *fi = file->private_data;
 	struct list_head *lh;
 	struct pending_request *req;
-	int done = 0, i, fail = 0;
+	int i, fail;
 	int retval = 0;
 	struct list_head *entry;
 	struct arm_addr *addr = NULL;
@@ -2864,25 +2883,28 @@ static int raw1394_release(struct inode *inode, struct file *file)
 		       "error(s) occurred \n");
 	}
 
-	while (!done) {
+	for (;;) {
+		/* This locked section guarantees that neither
+		 * complete nor pending requests exist once i!=0 */
 		spin_lock_irqsave(&fi->reqlists_lock, flags);
-
-		while (!list_empty(&fi->req_complete)) {
-			lh = fi->req_complete.next;
-			list_del(lh);
-
-			req = list_entry(lh, struct pending_request, list);
-
+		while ((req = __next_complete_req(fi)))
 			free_pending_request(req);
-		}
-
-		if (list_empty(&fi->req_pending))
-			done = 1;
 
+		i = list_empty(&fi->req_pending);
 		spin_unlock_irqrestore(&fi->reqlists_lock, flags);
 
-		if (!done)
-			down_interruptible(&fi->complete_sem);
+		if (i)
+			break;
+		/*
+		 * Sleep until more requests can be freed.
+		 *
+		 * NB: We call the macro wait_event() with a condition argument
+		 * with side effect.  This is only possible because the side
+		 * effect does not occur until the condition became true, and
+		 * wait_event() won't evaluate the condition again after that.
+		 */
+		wait_event(fi->wait_complete, (req = next_complete_req(fi)));
+		free_pending_request(req);
 	}
 
 	/* Remove any sub-trees left by user space programs */
diff --git a/drivers/ieee1394/sbp2.c b/drivers/ieee1394/sbp2.c
index b08755e2e68f..6986ac188281 100644
--- a/drivers/ieee1394/sbp2.c
+++ b/drivers/ieee1394/sbp2.c
@@ -38,31 +38,36 @@
  *	  but the code needs additional debugging.
  */
 
+#include <linux/blkdev.h>
+#include <linux/compiler.h>
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/gfp.h>
+#include <linux/init.h>
 #include <linux/kernel.h>
 #include <linux/list.h>
-#include <linux/string.h>
-#include <linux/stringify.h>
-#include <linux/slab.h>
-#include <linux/interrupt.h>
-#include <linux/fs.h>
-#include <linux/poll.h>
 #include <linux/module.h>
 #include <linux/moduleparam.h>
-#include <linux/types.h>
-#include <linux/delay.h>
-#include <linux/sched.h>
-#include <linux/blkdev.h>
-#include <linux/smp_lock.h>
-#include <linux/init.h>
 #include <linux/pci.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+#include <linux/stat.h>
+#include <linux/string.h>
+#include <linux/stringify.h>
+#include <linux/types.h>
+#include <linux/wait.h>
 
-#include <asm/current.h>
-#include <asm/uaccess.h>
-#include <asm/io.h>
 #include <asm/byteorder.h>
-#include <asm/atomic.h>
-#include <asm/system.h>
+#include <asm/errno.h>
+#include <asm/param.h>
 #include <asm/scatterlist.h>
+#include <asm/system.h>
+#include <asm/types.h>
+
+#ifdef CONFIG_IEEE1394_SBP2_PHYS_DMA
+#include <asm/io.h> /* for bus_to_virt */
+#endif
 
 #include <scsi/scsi.h>
 #include <scsi/scsi_cmnd.h>
@@ -71,13 +76,14 @@
 #include <scsi/scsi_host.h>
 
 #include "csr1212.h"
+#include "highlevel.h"
+#include "hosts.h"
 #include "ieee1394.h"
-#include "ieee1394_types.h"
 #include "ieee1394_core.h"
-#include "nodemgr.h"
-#include "hosts.h"
-#include "highlevel.h"
+#include "ieee1394_hotplug.h"
 #include "ieee1394_transactions.h"
+#include "ieee1394_types.h"
+#include "nodemgr.h"
 #include "sbp2.h"
 
 /*
@@ -173,11 +179,6 @@ MODULE_PARM_DESC(workarounds, "Work around device bugs (default = 0"
 	", override internal blacklist = " __stringify(SBP2_WORKAROUND_OVERRIDE)
 	", or a combination)");
 
-/* legacy parameter */
-static int force_inquiry_hack;
-module_param(force_inquiry_hack, int, 0644);
-MODULE_PARM_DESC(force_inquiry_hack, "Deprecated, use 'workarounds'");
-
 /*
  * Export information about protocols/devices supported by this driver.
  */
@@ -208,9 +209,9 @@ static u32 global_outstanding_command_orbs = 0;
 #define outstanding_orb_incr global_outstanding_command_orbs++
 #define outstanding_orb_decr global_outstanding_command_orbs--
 #else
-#define SBP2_ORB_DEBUG(fmt, args...)
-#define outstanding_orb_incr
-#define outstanding_orb_decr
+#define SBP2_ORB_DEBUG(fmt, args...)	do {} while (0)
+#define outstanding_orb_incr		do {} while (0)
+#define outstanding_orb_decr		do {} while (0)
 #endif
 
 #ifdef CONFIG_IEEE1394_SBP2_DEBUG_DMA
@@ -222,8 +223,8 @@ static u32 global_outstanding_command_orbs = 0;
 		 --global_outstanding_dmas, ## args)
 static u32 global_outstanding_dmas = 0;
 #else
-#define SBP2_DMA_ALLOC(fmt, args...)
-#define SBP2_DMA_FREE(fmt, args...)
+#define SBP2_DMA_ALLOC(fmt, args...)	do {} while (0)
+#define SBP2_DMA_FREE(fmt, args...)	do {} while (0)
 #endif
 
 #if CONFIG_IEEE1394_SBP2_DEBUG >= 2
@@ -237,7 +238,7 @@ static u32 global_outstanding_dmas = 0;
 #define SBP2_NOTICE(fmt, args...)	HPSB_NOTICE("sbp2: "fmt, ## args)
 #define SBP2_WARN(fmt, args...)		HPSB_WARN("sbp2: "fmt, ## args)
 #else
-#define SBP2_DEBUG(fmt, args...)
+#define SBP2_DEBUG(fmt, args...)	do {} while (0)
 #define SBP2_INFO(fmt, args...)		HPSB_INFO("sbp2: "fmt, ## args)
 #define SBP2_NOTICE(fmt, args...)       HPSB_NOTICE("sbp2: "fmt, ## args)
 #define SBP2_WARN(fmt, args...)         HPSB_WARN("sbp2: "fmt, ## args)
@@ -356,7 +357,7 @@ static const struct {
 /*
  * Converts a buffer from be32 to cpu byte ordering. Length is in bytes.
  */
-static __inline__ void sbp2util_be32_to_cpu_buffer(void *buffer, int length)
+static inline void sbp2util_be32_to_cpu_buffer(void *buffer, int length)
 {
 	u32 *temp = buffer;
 
@@ -369,7 +370,7 @@ static __inline__ void sbp2util_be32_to_cpu_buffer(void *buffer, int length)
 /*
  * Converts a buffer from cpu to be32 byte ordering. Length is in bytes.
  */
-static __inline__ void sbp2util_cpu_to_be32_buffer(void *buffer, int length)
+static inline void sbp2util_cpu_to_be32_buffer(void *buffer, int length)
 {
 	u32 *temp = buffer;
 
@@ -380,8 +381,8 @@ static __inline__ void sbp2util_cpu_to_be32_buffer(void *buffer, int length)
 }
 #else /* BIG_ENDIAN */
 /* Why waste the cpu cycles? */
-#define sbp2util_be32_to_cpu_buffer(x,y)
-#define sbp2util_cpu_to_be32_buffer(x,y)
+#define sbp2util_be32_to_cpu_buffer(x,y) do {} while (0)
+#define sbp2util_cpu_to_be32_buffer(x,y) do {} while (0)
 #endif
 
 #ifdef CONFIG_IEEE1394_SBP2_PACKET_DUMP
@@ -417,24 +418,26 @@ static void sbp2util_packet_dump(void *buffer, int length, char *dump_name,
 	return;
 }
 #else
-#define sbp2util_packet_dump(w,x,y,z)
+#define sbp2util_packet_dump(w,x,y,z) do {} while (0)
 #endif
 
+static DECLARE_WAIT_QUEUE_HEAD(access_wq);
+
 /*
- * Goofy routine that basically does a down_timeout function.
+ * Waits for completion of an SBP-2 access request.
+ * Returns nonzero if timed out or prematurely interrupted.
  */
-static int sbp2util_down_timeout(atomic_t *done, int timeout)
+static int sbp2util_access_timeout(struct scsi_id_instance_data *scsi_id,
+				   int timeout)
 {
-	int i;
+	long leftover = wait_event_interruptible_timeout(
+				access_wq, scsi_id->access_complete, timeout);
 
-	for (i = timeout; (i > 0 && atomic_read(done) == 0); i-= HZ/10) {
-		if (msleep_interruptible(100))	/* 100ms */
-			return 1;
-	}
-	return (i > 0) ? 0 : 1;
+	scsi_id->access_complete = 0;
+	return leftover <= 0;
 }
 
-/* Free's an allocated packet */
+/* Frees an allocated packet */
 static void sbp2_free_packet(struct hpsb_packet *packet)
 {
 	hpsb_free_tlabel(packet);
@@ -468,6 +471,44 @@ static int sbp2util_node_write_no_wait(struct node_entry *ne, u64 addr,
 	return 0;
 }
 
+static void sbp2util_notify_fetch_agent(struct scsi_id_instance_data *scsi_id,
+					u64 offset, quadlet_t *data, size_t len)
+{
+	/*
+	 * There is a small window after a bus reset within which the node
+	 * entry's generation is current but the reconnect wasn't completed.
+	 */
+	if (unlikely(atomic_read(&scsi_id->state) == SBP2LU_STATE_IN_RESET))
+		return;
+
+	if (hpsb_node_write(scsi_id->ne,
+			    scsi_id->sbp2_command_block_agent_addr + offset,
+			    data, len))
+		SBP2_ERR("sbp2util_notify_fetch_agent failed.");
+	/*
+	 * Now accept new SCSI commands, unless a bus reset happended during
+	 * hpsb_node_write.
+	 */
+	if (likely(atomic_read(&scsi_id->state) != SBP2LU_STATE_IN_RESET))
+		scsi_unblock_requests(scsi_id->scsi_host);
+}
+
+static void sbp2util_write_orb_pointer(void *p)
+{
+	quadlet_t data[2];
+
+	data[0] = ORB_SET_NODE_ID(
+			((struct scsi_id_instance_data *)p)->hi->host->node_id);
+	data[1] = ((struct scsi_id_instance_data *)p)->last_orb_dma;
+	sbp2util_cpu_to_be32_buffer(data, 8);
+	sbp2util_notify_fetch_agent(p, SBP2_ORB_POINTER_OFFSET, data, 8);
+}
+
+static void sbp2util_write_doorbell(void *p)
+{
+	sbp2util_notify_fetch_agent(p, SBP2_DOORBELL_OFFSET, NULL, 4);
+}
+
 /*
  * This function is called to create a pool of command orbs used for
  * command processing. It is called when a new sbp2 device is detected.
@@ -492,7 +533,7 @@ static int sbp2util_create_command_orb_pool(struct scsi_id_instance_data *scsi_i
 		command->command_orb_dma =
 		    pci_map_single(hi->host->pdev, &command->command_orb,
 				   sizeof(struct sbp2_command_orb),
-				   PCI_DMA_BIDIRECTIONAL);
+				   PCI_DMA_TODEVICE);
 		SBP2_DMA_ALLOC("single command orb DMA");
 		command->sge_dma =
 		    pci_map_single(hi->host->pdev,
@@ -525,7 +566,7 @@ static void sbp2util_remove_command_orb_pool(struct scsi_id_instance_data *scsi_
 			/* Release our generic DMA's */
 			pci_unmap_single(host->pdev, command->command_orb_dma,
 					 sizeof(struct sbp2_command_orb),
-					 PCI_DMA_BIDIRECTIONAL);
+					 PCI_DMA_TODEVICE);
 			SBP2_DMA_FREE("single command orb DMA");
 			pci_unmap_single(host->pdev, command->sge_dma,
 					 sizeof(command->scatter_gather_element),
@@ -715,6 +756,7 @@ static int sbp2_remove(struct device *dev)
 			sbp2scsi_complete_all_commands(scsi_id, DID_NO_CONNECT);
 		/* scsi_remove_device() will trigger shutdown functions of SCSI
 		 * highlevel drivers which would deadlock if blocked. */
+		atomic_set(&scsi_id->state, SBP2LU_STATE_IN_SHUTDOWN);
 		scsi_unblock_requests(scsi_id->scsi_host);
 	}
 	sdev = scsi_id->sdev;
@@ -766,10 +808,12 @@ static int sbp2_update(struct unit_directory *ud)
 	 */
 	sbp2scsi_complete_all_commands(scsi_id, DID_BUS_BUSY);
 
-	/* Make sure we unblock requests (since this is likely after a bus
-	 * reset). */
-	scsi_unblock_requests(scsi_id->scsi_host);
-
+	/* Accept new commands unless there was another bus reset in the
+	 * meantime. */
+	if (hpsb_node_entry_valid(scsi_id->ne)) {
+		atomic_set(&scsi_id->state, SBP2LU_STATE_RUNNING);
+		scsi_unblock_requests(scsi_id->scsi_host);
+	}
 	return 0;
 }
 
@@ -794,11 +838,12 @@ static struct scsi_id_instance_data *sbp2_alloc_device(struct unit_directory *ud
 	scsi_id->speed_code = IEEE1394_SPEED_100;
 	scsi_id->max_payload_size = sbp2_speedto_max_payload[IEEE1394_SPEED_100];
 	scsi_id->status_fifo_addr = CSR1212_INVALID_ADDR_SPACE;
-	atomic_set(&scsi_id->sbp2_login_complete, 0);
 	INIT_LIST_HEAD(&scsi_id->sbp2_command_orb_inuse);
 	INIT_LIST_HEAD(&scsi_id->sbp2_command_orb_completed);
 	INIT_LIST_HEAD(&scsi_id->scsi_list);
 	spin_lock_init(&scsi_id->sbp2_command_orb_lock);
+	atomic_set(&scsi_id->state, SBP2LU_STATE_RUNNING);
+	INIT_WORK(&scsi_id->protocol_work, NULL, NULL);
 
 	ud->device.driver_data = scsi_id;
 
@@ -881,11 +926,14 @@ static void sbp2_host_reset(struct hpsb_host *host)
 	struct scsi_id_instance_data *scsi_id;
 
 	hi = hpsb_get_hostinfo(&sbp2_highlevel, host);
-
-	if (hi) {
-		list_for_each_entry(scsi_id, &hi->scsi_ids, scsi_list)
+	if (!hi)
+		return;
+	list_for_each_entry(scsi_id, &hi->scsi_ids, scsi_list)
+		if (likely(atomic_read(&scsi_id->state) !=
+			   SBP2LU_STATE_IN_SHUTDOWN)) {
+			atomic_set(&scsi_id->state, SBP2LU_STATE_IN_RESET);
 			scsi_block_requests(scsi_id->scsi_host);
-	}
+		}
 }
 
 /*
@@ -970,8 +1018,7 @@ static int sbp2_start_device(struct scsi_id_instance_data *scsi_id)
 	 * connected to the sbp2 device being removed. That host would
 	 * have a certain amount of time to relogin before the sbp2 device
 	 * allows someone else to login instead. One second makes sense. */
-	msleep_interruptible(1000);
-	if (signal_pending(current)) {
+	if (msleep_interruptible(1000)) {
 		sbp2_remove_device(scsi_id);
 		return -EINTR;
 	}
@@ -1036,7 +1083,7 @@ static void sbp2_remove_device(struct scsi_id_instance_data *scsi_id)
 		scsi_remove_host(scsi_id->scsi_host);
 		scsi_host_put(scsi_id->scsi_host);
 	}
-
+	flush_scheduled_work();
 	sbp2util_remove_command_orb_pool(scsi_id);
 
 	list_del(&scsi_id->scsi_list);
@@ -1182,17 +1229,14 @@ static int sbp2_query_logins(struct scsi_id_instance_data *scsi_id)
 			     "sbp2 query logins orb", scsi_id->query_logins_orb_dma);
 
 	memset(scsi_id->query_logins_response, 0, sizeof(struct sbp2_query_logins_response));
-	memset(&scsi_id->status_block, 0, sizeof(struct sbp2_status_block));
 
 	data[0] = ORB_SET_NODE_ID(hi->host->node_id);
 	data[1] = scsi_id->query_logins_orb_dma;
 	sbp2util_cpu_to_be32_buffer(data, 8);
 
-	atomic_set(&scsi_id->sbp2_login_complete, 0);
-
 	hpsb_node_write(scsi_id->ne, scsi_id->sbp2_management_agent_addr, data, 8);
 
-	if (sbp2util_down_timeout(&scsi_id->sbp2_login_complete, 2*HZ)) {
+	if (sbp2util_access_timeout(scsi_id, 2*HZ)) {
 		SBP2_INFO("Error querying logins to SBP-2 device - timed out");
 		return -EIO;
 	}
@@ -1202,11 +1246,8 @@ static int sbp2_query_logins(struct scsi_id_instance_data *scsi_id)
 		return -EIO;
 	}
 
-	if (STATUS_GET_RESP(scsi_id->status_block.ORB_offset_hi_misc) ||
-	    STATUS_GET_DEAD_BIT(scsi_id->status_block.ORB_offset_hi_misc) ||
-	    STATUS_GET_SBP_STATUS(scsi_id->status_block.ORB_offset_hi_misc)) {
-
-		SBP2_INFO("Error querying logins to SBP-2 device - timed out");
+	if (STATUS_TEST_RDS(scsi_id->status_block.ORB_offset_hi_misc)) {
+		SBP2_INFO("Error querying logins to SBP-2 device - failed");
 		return -EIO;
 	}
 
@@ -1278,21 +1319,18 @@ static int sbp2_login_device(struct scsi_id_instance_data *scsi_id)
 			     "sbp2 login orb", scsi_id->login_orb_dma);
 
 	memset(scsi_id->login_response, 0, sizeof(struct sbp2_login_response));
-	memset(&scsi_id->status_block, 0, sizeof(struct sbp2_status_block));
 
 	data[0] = ORB_SET_NODE_ID(hi->host->node_id);
 	data[1] = scsi_id->login_orb_dma;
 	sbp2util_cpu_to_be32_buffer(data, 8);
 
-	atomic_set(&scsi_id->sbp2_login_complete, 0);
-
 	hpsb_node_write(scsi_id->ne, scsi_id->sbp2_management_agent_addr, data, 8);
 
 	/*
 	 * Wait for login status (up to 20 seconds)...
 	 */
-	if (sbp2util_down_timeout(&scsi_id->sbp2_login_complete, 20*HZ)) {
-		SBP2_ERR("Error logging into SBP-2 device - login timed-out");
+	if (sbp2util_access_timeout(scsi_id, 20*HZ)) {
+		SBP2_ERR("Error logging into SBP-2 device - timed out");
 		return -EIO;
 	}
 
@@ -1300,18 +1338,12 @@ static int sbp2_login_device(struct scsi_id_instance_data *scsi_id)
 	 * Sanity. Make sure status returned matches login orb.
 	 */
 	if (scsi_id->status_block.ORB_offset_lo != scsi_id->login_orb_dma) {
-		SBP2_ERR("Error logging into SBP-2 device - login timed-out");
+		SBP2_ERR("Error logging into SBP-2 device - timed out");
 		return -EIO;
 	}
 
-	/*
-	 * Check status
-	 */
-	if (STATUS_GET_RESP(scsi_id->status_block.ORB_offset_hi_misc) ||
-	    STATUS_GET_DEAD_BIT(scsi_id->status_block.ORB_offset_hi_misc) ||
-	    STATUS_GET_SBP_STATUS(scsi_id->status_block.ORB_offset_hi_misc)) {
-
-		SBP2_ERR("Error logging into SBP-2 device - login failed");
+	if (STATUS_TEST_RDS(scsi_id->status_block.ORB_offset_hi_misc)) {
+		SBP2_ERR("Error logging into SBP-2 device - failed");
 		return -EIO;
 	}
 
@@ -1335,9 +1367,7 @@ static int sbp2_login_device(struct scsi_id_instance_data *scsi_id)
 	scsi_id->sbp2_command_block_agent_addr &= 0x0000ffffffffffffULL;
 
 	SBP2_INFO("Logged into SBP-2 device");
-
 	return 0;
-
 }
 
 /*
@@ -1387,21 +1417,17 @@ static int sbp2_logout_device(struct scsi_id_instance_data *scsi_id)
 	data[1] = scsi_id->logout_orb_dma;
 	sbp2util_cpu_to_be32_buffer(data, 8);
 
-	atomic_set(&scsi_id->sbp2_login_complete, 0);
-
 	error = hpsb_node_write(scsi_id->ne,
 				scsi_id->sbp2_management_agent_addr, data, 8);
 	if (error)
 		return error;
 
 	/* Wait for device to logout...1 second. */
-	if (sbp2util_down_timeout(&scsi_id->sbp2_login_complete, HZ))
+	if (sbp2util_access_timeout(scsi_id, HZ))
 		return -EIO;
 
 	SBP2_INFO("Logged out of SBP-2 device");
-
 	return 0;
-
 }
 
 /*
@@ -1445,20 +1471,10 @@ static int sbp2_reconnect_device(struct scsi_id_instance_data *scsi_id)
 	sbp2util_packet_dump(scsi_id->reconnect_orb, sizeof(struct sbp2_reconnect_orb),
 			     "sbp2 reconnect orb", scsi_id->reconnect_orb_dma);
 
-	/*
-	 * Initialize status fifo
-	 */
-	memset(&scsi_id->status_block, 0, sizeof(struct sbp2_status_block));
-
-	/*
-	 * Ok, let's write to the target's management agent register
-	 */
 	data[0] = ORB_SET_NODE_ID(hi->host->node_id);
 	data[1] = scsi_id->reconnect_orb_dma;
 	sbp2util_cpu_to_be32_buffer(data, 8);
 
-	atomic_set(&scsi_id->sbp2_login_complete, 0);
-
 	error = hpsb_node_write(scsi_id->ne,
 				scsi_id->sbp2_management_agent_addr, data, 8);
 	if (error)
@@ -1467,8 +1483,8 @@ static int sbp2_reconnect_device(struct scsi_id_instance_data *scsi_id)
 	/*
 	 * Wait for reconnect status (up to 1 second)...
 	 */
-	if (sbp2util_down_timeout(&scsi_id->sbp2_login_complete, HZ)) {
-		SBP2_ERR("Error reconnecting to SBP-2 device - reconnect timed-out");
+	if (sbp2util_access_timeout(scsi_id, HZ)) {
+		SBP2_ERR("Error reconnecting to SBP-2 device - timed out");
 		return -EIO;
 	}
 
@@ -1476,25 +1492,17 @@ static int sbp2_reconnect_device(struct scsi_id_instance_data *scsi_id)
 	 * Sanity. Make sure status returned matches reconnect orb.
 	 */
 	if (scsi_id->status_block.ORB_offset_lo != scsi_id->reconnect_orb_dma) {
-		SBP2_ERR("Error reconnecting to SBP-2 device - reconnect timed-out");
+		SBP2_ERR("Error reconnecting to SBP-2 device - timed out");
 		return -EIO;
 	}
 
-	/*
-	 * Check status
-	 */
-	if (STATUS_GET_RESP(scsi_id->status_block.ORB_offset_hi_misc) ||
-	    STATUS_GET_DEAD_BIT(scsi_id->status_block.ORB_offset_hi_misc) ||
-	    STATUS_GET_SBP_STATUS(scsi_id->status_block.ORB_offset_hi_misc)) {
-
-		SBP2_ERR("Error reconnecting to SBP-2 device - reconnect failed");
+	if (STATUS_TEST_RDS(scsi_id->status_block.ORB_offset_hi_misc)) {
+		SBP2_ERR("Error reconnecting to SBP-2 device - failed");
 		return -EIO;
 	}
 
 	HPSB_DEBUG("Reconnected to SBP-2 device");
-
 	return 0;
-
 }
 
 /*
@@ -1592,11 +1600,6 @@ static void sbp2_parse_unit_directory(struct scsi_id_instance_data *scsi_id,
 	}
 
 	workarounds = sbp2_default_workarounds;
-	if (force_inquiry_hack) {
-		SBP2_WARN("force_inquiry_hack is deprecated. "
-			  "Use parameter 'workarounds' instead.");
-		workarounds |= SBP2_WORKAROUND_INQUIRY_36;
-	}
 
 	if (!(workarounds & SBP2_WORKAROUND_OVERRIDE))
 		for (i = 0; i < ARRAY_SIZE(sbp2_workarounds_table); i++) {
@@ -1705,9 +1708,14 @@ static int sbp2_agent_reset(struct scsi_id_instance_data *scsi_id, int wait)
 	quadlet_t data;
 	u64 addr;
 	int retval;
+	unsigned long flags;
 
 	SBP2_DEBUG_ENTER();
 
+	cancel_delayed_work(&scsi_id->protocol_work);
+	if (wait)
+		flush_scheduled_work();
+
 	data = ntohl(SBP2_AGENT_RESET_DATA);
 	addr = scsi_id->sbp2_command_block_agent_addr + SBP2_AGENT_RESET_OFFSET;
 
@@ -1724,7 +1732,9 @@ static int sbp2_agent_reset(struct scsi_id_instance_data *scsi_id, int wait)
 	/*
 	 * Need to make sure orb pointer is written on next command
 	 */
+	spin_lock_irqsave(&scsi_id->sbp2_command_orb_lock, flags);
 	scsi_id->last_orb = NULL;
+	spin_unlock_irqrestore(&scsi_id->sbp2_command_orb_lock, flags);
 
 	return 0;
 }
@@ -1961,13 +1971,17 @@ static void sbp2_create_command_orb(struct scsi_id_instance_data *scsi_id,
 /*
  * This function is called in order to begin a regular SBP-2 command.
  */
-static int sbp2_link_orb_command(struct scsi_id_instance_data *scsi_id,
+static void sbp2_link_orb_command(struct scsi_id_instance_data *scsi_id,
 				 struct sbp2_command_info *command)
 {
 	struct sbp2scsi_host_info *hi = scsi_id->hi;
 	struct sbp2_command_orb *command_orb = &command->command_orb;
-	struct node_entry *ne = scsi_id->ne;
-	u64 addr;
+	struct sbp2_command_orb *last_orb;
+	dma_addr_t last_orb_dma;
+	u64 addr = scsi_id->sbp2_command_block_agent_addr;
+	quadlet_t data[2];
+	size_t length;
+	unsigned long flags;
 
 	outstanding_orb_incr;
 	SBP2_ORB_DEBUG("sending command orb %p, total orbs = %x",
@@ -1975,73 +1989,70 @@ static int sbp2_link_orb_command(struct scsi_id_instance_data *scsi_id,
 
 	pci_dma_sync_single_for_device(hi->host->pdev, command->command_orb_dma,
 				       sizeof(struct sbp2_command_orb),
-				       PCI_DMA_BIDIRECTIONAL);
+				       PCI_DMA_TODEVICE);
 	pci_dma_sync_single_for_device(hi->host->pdev, command->sge_dma,
 				       sizeof(command->scatter_gather_element),
 				       PCI_DMA_BIDIRECTIONAL);
 	/*
 	 * Check to see if there are any previous orbs to use
 	 */
-	if (scsi_id->last_orb == NULL) {
-		quadlet_t data[2];
-
+	spin_lock_irqsave(&scsi_id->sbp2_command_orb_lock, flags);
+	last_orb = scsi_id->last_orb;
+	last_orb_dma = scsi_id->last_orb_dma;
+	if (!last_orb) {
 		/*
-		 * Ok, let's write to the target's management agent register
+		 * last_orb == NULL means: We know that the target's fetch agent
+		 * is not active right now.
 		 */
-		addr = scsi_id->sbp2_command_block_agent_addr + SBP2_ORB_POINTER_OFFSET;
+		addr += SBP2_ORB_POINTER_OFFSET;
 		data[0] = ORB_SET_NODE_ID(hi->host->node_id);
 		data[1] = command->command_orb_dma;
 		sbp2util_cpu_to_be32_buffer(data, 8);
-
-		SBP2_ORB_DEBUG("write command agent, command orb %p", command_orb);
-
-		if (sbp2util_node_write_no_wait(ne, addr, data, 8) < 0) {
-			SBP2_ERR("sbp2util_node_write_no_wait failed.\n");
-			return -EIO;
-		}
-
-		SBP2_ORB_DEBUG("write command agent complete");
-
-		scsi_id->last_orb = command_orb;
-		scsi_id->last_orb_dma = command->command_orb_dma;
-
+		length = 8;
 	} else {
-		quadlet_t data;
-
 		/*
-		 * We have an orb already sent (maybe or maybe not
-		 * processed) that we can append this orb to. So do so,
-		 * and ring the doorbell. Have to be very careful
-		 * modifying these next orb pointers, as they are accessed
-		 * both by the sbp2 device and us.
+		 * last_orb != NULL means: We know that the target's fetch agent
+		 * is (very probably) not dead or in reset state right now.
+		 * We have an ORB already sent that we can append a new one to.
+		 * The target's fetch agent may or may not have read this
+		 * previous ORB yet.
 		 */
-		scsi_id->last_orb->next_ORB_lo =
-		    cpu_to_be32(command->command_orb_dma);
+		pci_dma_sync_single_for_cpu(hi->host->pdev, last_orb_dma,
+					    sizeof(struct sbp2_command_orb),
+					    PCI_DMA_TODEVICE);
+		last_orb->next_ORB_lo = cpu_to_be32(command->command_orb_dma);
+		wmb();
 		/* Tells hardware that this pointer is valid */
-		scsi_id->last_orb->next_ORB_hi = 0x0;
-		pci_dma_sync_single_for_device(hi->host->pdev,
-					       scsi_id->last_orb_dma,
+		last_orb->next_ORB_hi = 0;
+		pci_dma_sync_single_for_device(hi->host->pdev, last_orb_dma,
 					       sizeof(struct sbp2_command_orb),
-					       PCI_DMA_BIDIRECTIONAL);
+					       PCI_DMA_TODEVICE);
+		addr += SBP2_DOORBELL_OFFSET;
+		data[0] = 0;
+		length = 4;
+	}
+	scsi_id->last_orb = command_orb;
+	scsi_id->last_orb_dma = command->command_orb_dma;
+	spin_unlock_irqrestore(&scsi_id->sbp2_command_orb_lock, flags);
 
+	SBP2_ORB_DEBUG("write to %s register, command orb %p",
+			last_orb ? "DOORBELL" : "ORB_POINTER", command_orb);
+	if (sbp2util_node_write_no_wait(scsi_id->ne, addr, data, length)) {
 		/*
-		 * Ring the doorbell
+		 * sbp2util_node_write_no_wait failed. We certainly ran out
+		 * of transaction labels, perhaps just because there were no
+		 * context switches which gave khpsbpkt a chance to collect
+		 * free tlabels. Try again in non-atomic context. If necessary,
+		 * the workqueue job will sleep to guaranteedly get a tlabel.
+		 * We do not accept new commands until the job is over.
 		 */
-		data = cpu_to_be32(command->command_orb_dma);
-		addr = scsi_id->sbp2_command_block_agent_addr + SBP2_DOORBELL_OFFSET;
-
-		SBP2_ORB_DEBUG("ring doorbell, command orb %p", command_orb);
-
-		if (sbp2util_node_write_no_wait(ne, addr, &data, 4) < 0) {
-			SBP2_ERR("sbp2util_node_write_no_wait failed");
-			return -EIO;
-		}
-
-		scsi_id->last_orb = command_orb;
-		scsi_id->last_orb_dma = command->command_orb_dma;
-
+		scsi_block_requests(scsi_id->scsi_host);
+		PREPARE_WORK(&scsi_id->protocol_work,
+			     last_orb ? sbp2util_write_doorbell:
+					sbp2util_write_orb_pointer,
+			     scsi_id);
+		schedule_work(&scsi_id->protocol_work);
 	}
-	return 0;
 }
 
 /*
@@ -2078,11 +2089,6 @@ static int sbp2_send_command(struct scsi_id_instance_data *scsi_id,
 			     "sbp2 command orb", command->command_orb_dma);
 
 	/*
-	 * Initialize status fifo
-	 */
-	memset(&scsi_id->status_block, 0, sizeof(struct sbp2_status_block));
-
-	/*
 	 * Link up the orb, and ring the doorbell if needed
 	 */
 	sbp2_link_orb_command(scsi_id, command);
@@ -2123,12 +2129,14 @@ static unsigned int sbp2_status_to_sense_data(unchar *sbp2_status, unchar *sense
 /*
  * This function deals with status writes from the SBP-2 device
  */
-static int sbp2_handle_status_write(struct hpsb_host *host, int nodeid, int destid,
-				    quadlet_t *data, u64 addr, size_t length, u16 fl)
+static int sbp2_handle_status_write(struct hpsb_host *host, int nodeid,
+				    int destid, quadlet_t *data, u64 addr,
+				    size_t length, u16 fl)
 {
 	struct sbp2scsi_host_info *hi;
 	struct scsi_id_instance_data *scsi_id = NULL, *scsi_id_tmp;
 	struct scsi_cmnd *SCpnt = NULL;
+	struct sbp2_status_block *sb;
 	u32 scsi_status = SBP2_SCSI_STATUS_GOOD;
 	struct sbp2_command_info *command;
 	unsigned long flags;
@@ -2137,18 +2145,19 @@ static int sbp2_handle_status_write(struct hpsb_host *host, int nodeid, int dest
 
 	sbp2util_packet_dump(data, length, "sbp2 status write by device", (u32)addr);
 
-	if (!host) {
+	if (unlikely(length < 8 || length > sizeof(struct sbp2_status_block))) {
+		SBP2_ERR("Wrong size of status block");
+		return RCODE_ADDRESS_ERROR;
+	}
+	if (unlikely(!host)) {
 		SBP2_ERR("host is NULL - this is bad!");
 		return RCODE_ADDRESS_ERROR;
 	}
-
 	hi = hpsb_get_hostinfo(&sbp2_highlevel, host);
-
-	if (!hi) {
+	if (unlikely(!hi)) {
 		SBP2_ERR("host info is NULL - this is bad!");
 		return RCODE_ADDRESS_ERROR;
 	}
-
 	/*
 	 * Find our scsi_id structure by looking at the status fifo address
 	 * written to by the sbp2 device.
@@ -2160,32 +2169,35 @@ static int sbp2_handle_status_write(struct hpsb_host *host, int nodeid, int dest
 			break;
 		}
 	}
-
-	if (!scsi_id) {
+	if (unlikely(!scsi_id)) {
 		SBP2_ERR("scsi_id is NULL - device is gone?");
 		return RCODE_ADDRESS_ERROR;
 	}
 
 	/*
-	 * Put response into scsi_id status fifo...
+	 * Put response into scsi_id status fifo buffer. The first two bytes
+	 * come in big endian bit order. Often the target writes only a
+	 * truncated status block, minimally the first two quadlets. The rest
+	 * is implied to be zeros.
 	 */
-	memcpy(&scsi_id->status_block, data, length);
+	sb = &scsi_id->status_block;
+	memset(sb->command_set_dependent, 0, sizeof(sb->command_set_dependent));
+	memcpy(sb, data, length);
+	sbp2util_be32_to_cpu_buffer(sb, 8);
 
 	/*
-	 * Byte swap first two quadlets (8 bytes) of status for processing
+	 * Ignore unsolicited status. Handle command ORB status.
 	 */
-	sbp2util_be32_to_cpu_buffer(&scsi_id->status_block, 8);
-
-	/*
-	 * Handle command ORB status here if necessary. First, need to match status with command.
-	 */
-	command = sbp2util_find_command_for_orb(scsi_id, scsi_id->status_block.ORB_offset_lo);
+	if (unlikely(STATUS_GET_SRC(sb->ORB_offset_hi_misc) == 2))
+		command = NULL;
+	else
+		command = sbp2util_find_command_for_orb(scsi_id,
+							sb->ORB_offset_lo);
 	if (command) {
-
 		SBP2_DEBUG("Found status for command ORB");
 		pci_dma_sync_single_for_cpu(hi->host->pdev, command->command_orb_dma,
 					    sizeof(struct sbp2_command_orb),
-					    PCI_DMA_BIDIRECTIONAL);
+					    PCI_DMA_TODEVICE);
 		pci_dma_sync_single_for_cpu(hi->host->pdev, command->sge_dma,
 					    sizeof(command->scatter_gather_element),
 					    PCI_DMA_BIDIRECTIONAL);
@@ -2194,7 +2206,12 @@ static int sbp2_handle_status_write(struct hpsb_host *host, int nodeid, int dest
 		outstanding_orb_decr;
 
 		/*
-		 * Matched status with command, now grab scsi command pointers and check status
+		 * Matched status with command, now grab scsi command pointers
+		 * and check status.
+		 */
+		/*
+		 * FIXME: If the src field in the status is 1, the ORB DMA must
+		 * not be reused until status for a subsequent ORB is received.
 		 */
 		SCpnt = command->Current_SCpnt;
 		spin_lock_irqsave(&scsi_id->sbp2_command_orb_lock, flags);
@@ -2202,61 +2219,64 @@ static int sbp2_handle_status_write(struct hpsb_host *host, int nodeid, int dest
 		spin_unlock_irqrestore(&scsi_id->sbp2_command_orb_lock, flags);
 
 		if (SCpnt) {
-
+			u32 h = sb->ORB_offset_hi_misc;
+			u32 r = STATUS_GET_RESP(h);
+
+			if (r != RESP_STATUS_REQUEST_COMPLETE) {
+				SBP2_WARN("resp 0x%x, sbp_status 0x%x",
+					  r, STATUS_GET_SBP_STATUS(h));
+				scsi_status =
+					r == RESP_STATUS_TRANSPORT_FAILURE ?
+					SBP2_SCSI_STATUS_BUSY :
+					SBP2_SCSI_STATUS_COMMAND_TERMINATED;
+			}
 			/*
-			 * See if the target stored any scsi status information
+			 * See if the target stored any scsi status information.
 			 */
-			if (STATUS_GET_LENGTH(scsi_id->status_block.ORB_offset_hi_misc) > 1) {
-				/*
-				 * Translate SBP-2 status to SCSI sense data
-				 */
+			if (STATUS_GET_LEN(h) > 1) {
 				SBP2_DEBUG("CHECK CONDITION");
-				scsi_status = sbp2_status_to_sense_data((unchar *)&scsi_id->status_block, SCpnt->sense_buffer);
+				scsi_status = sbp2_status_to_sense_data(
+					(unchar *)sb, SCpnt->sense_buffer);
 			}
-
 			/*
-			 * Check to see if the dead bit is set. If so, we'll have to initiate
-			 * a fetch agent reset.
+			 * Check to see if the dead bit is set. If so, we'll
+			 * have to initiate a fetch agent reset.
 			 */
-			if (STATUS_GET_DEAD_BIT(scsi_id->status_block.ORB_offset_hi_misc)) {
-
-				/*
-				 * Initiate a fetch agent reset.
-				 */
-				SBP2_DEBUG("Dead bit set - initiating fetch agent reset");
+			if (STATUS_TEST_DEAD(h)) {
+				SBP2_DEBUG("Dead bit set - "
+					   "initiating fetch agent reset");
                                 sbp2_agent_reset(scsi_id, 0);
 			}
-
 			SBP2_ORB_DEBUG("completing command orb %p", &command->command_orb);
 		}
 
 		/*
-		 * Check here to see if there are no commands in-use. If there are none, we can
-		 * null out last orb so that next time around we write directly to the orb pointer...
-		 * Quick start saves one 1394 bus transaction.
+		 * Check here to see if there are no commands in-use. If there
+		 * are none, we know that the fetch agent left the active state
+		 * _and_ that we did not reactivate it yet. Therefore clear
+		 * last_orb so that next time we write directly to the
+		 * ORB_POINTER register. That way the fetch agent does not need
+		 * to refetch the next_ORB.
 		 */
 		spin_lock_irqsave(&scsi_id->sbp2_command_orb_lock, flags);
-		if (list_empty(&scsi_id->sbp2_command_orb_inuse)) {
+		if (list_empty(&scsi_id->sbp2_command_orb_inuse))
 			scsi_id->last_orb = NULL;
-		}
 		spin_unlock_irqrestore(&scsi_id->sbp2_command_orb_lock, flags);
 
 	} else {
-
 		/*
 		 * It's probably a login/logout/reconnect status.
 		 */
-		if ((scsi_id->login_orb_dma == scsi_id->status_block.ORB_offset_lo) ||
-		    (scsi_id->query_logins_orb_dma == scsi_id->status_block.ORB_offset_lo) ||
-		    (scsi_id->reconnect_orb_dma == scsi_id->status_block.ORB_offset_lo) ||
-		    (scsi_id->logout_orb_dma == scsi_id->status_block.ORB_offset_lo)) {
-			atomic_set(&scsi_id->sbp2_login_complete, 1);
+		if ((sb->ORB_offset_lo == scsi_id->reconnect_orb_dma) ||
+		    (sb->ORB_offset_lo == scsi_id->login_orb_dma) ||
+		    (sb->ORB_offset_lo == scsi_id->query_logins_orb_dma) ||
+		    (sb->ORB_offset_lo == scsi_id->logout_orb_dma)) {
+			scsi_id->access_complete = 1;
+			wake_up_interruptible(&access_wq);
 		}
 	}
 
 	if (SCpnt) {
-
-		/* Complete the SCSI command. */
 		SBP2_DEBUG("Completing SCSI command");
 		sbp2scsi_complete_command(scsi_id, scsi_status, SCpnt,
 					  command->Current_done);
@@ -2372,7 +2392,7 @@ static void sbp2scsi_complete_all_commands(struct scsi_id_instance_data *scsi_id
 		command = list_entry(lh, struct sbp2_command_info, list);
 		pci_dma_sync_single_for_cpu(hi->host->pdev, command->command_orb_dma,
 					    sizeof(struct sbp2_command_orb),
-					    PCI_DMA_BIDIRECTIONAL);
+					    PCI_DMA_TODEVICE);
 		pci_dma_sync_single_for_cpu(hi->host->pdev, command->sge_dma,
 					    sizeof(command->scatter_gather_element),
 					    PCI_DMA_BIDIRECTIONAL);
@@ -2495,6 +2515,7 @@ static int sbp2scsi_slave_alloc(struct scsi_device *sdev)
 		(struct scsi_id_instance_data *)sdev->host->hostdata[0];
 
 	scsi_id->sdev = sdev;
+	sdev->allow_restart = 1;
 
 	if (scsi_id->workarounds & SBP2_WORKAROUND_INQUIRY_36)
 		sdev->inquiry_len = 36;
@@ -2508,16 +2529,12 @@ static int sbp2scsi_slave_configure(struct scsi_device *sdev)
 
 	blk_queue_dma_alignment(sdev->request_queue, (512 - 1));
 	sdev->use_10_for_rw = 1;
-	sdev->use_10_for_ms = 1;
 
 	if (sdev->type == TYPE_DISK &&
 	    scsi_id->workarounds & SBP2_WORKAROUND_MODE_SENSE_8)
 		sdev->skip_ms_page_8 = 1;
 	if (scsi_id->workarounds & SBP2_WORKAROUND_FIX_CAPACITY)
 		sdev->fix_capacity = 1;
-	if (scsi_id->ne->guid_vendor_id == 0x0010b9 && /* Maxtor's OUI */
-	    (sdev->type == TYPE_DISK || sdev->type == TYPE_RBC))
-		sdev->allow_restart = 1;
 	return 0;
 }
 
@@ -2555,7 +2572,7 @@ static int sbp2scsi_abort(struct scsi_cmnd *SCpnt)
 			pci_dma_sync_single_for_cpu(hi->host->pdev,
 						    command->command_orb_dma,
 						    sizeof(struct sbp2_command_orb),
-						    PCI_DMA_BIDIRECTIONAL);
+						    PCI_DMA_TODEVICE);
 			pci_dma_sync_single_for_cpu(hi->host->pdev,
 						    command->sge_dma,
 						    sizeof(command->scatter_gather_element),
@@ -2571,7 +2588,7 @@ static int sbp2scsi_abort(struct scsi_cmnd *SCpnt)
 		/*
 		 * Initiate a fetch agent reset.
 		 */
-		sbp2_agent_reset(scsi_id, 0);
+		sbp2_agent_reset(scsi_id, 1);
 		sbp2scsi_complete_all_commands(scsi_id, DID_BUS_BUSY);
 	}
 
@@ -2590,7 +2607,7 @@ static int sbp2scsi_reset(struct scsi_cmnd *SCpnt)
 
 	if (sbp2util_node_is_available(scsi_id)) {
 		SBP2_ERR("Generating sbp2 fetch agent reset");
-		sbp2_agent_reset(scsi_id, 0);
+		sbp2_agent_reset(scsi_id, 1);
 	}
 
 	return SUCCESS;
diff --git a/drivers/ieee1394/sbp2.h b/drivers/ieee1394/sbp2.h
index b22ce1aa8fe4..abbe48e646c3 100644
--- a/drivers/ieee1394/sbp2.h
+++ b/drivers/ieee1394/sbp2.h
@@ -46,8 +46,8 @@
 #define ORB_SET_DIRECTION(value)		((value & 0x1) << 27)
 
 struct sbp2_command_orb {
-	volatile u32 next_ORB_hi;
-	volatile u32 next_ORB_lo;
+	u32 next_ORB_hi;
+	u32 next_ORB_lo;
 	u32 data_descriptor_hi;
 	u32 data_descriptor_lo;
 	u32 misc;
@@ -180,12 +180,14 @@ struct sbp2_unrestricted_page_table {
 
 #define SBP2_SCSI_STATUS_SELECTION_TIMEOUT	0xff
 
-#define STATUS_GET_ORB_OFFSET_HI(value)         (value & 0xffff)
-#define STATUS_GET_SBP_STATUS(value)            ((value >> 16) & 0xff)
-#define STATUS_GET_LENGTH(value)                ((value >> 24) & 0x7)
-#define STATUS_GET_DEAD_BIT(value)              ((value >> 27) & 0x1)
-#define STATUS_GET_RESP(value)                  ((value >> 28) & 0x3)
-#define STATUS_GET_SRC(value)                   ((value >> 30) & 0x3)
+#define STATUS_GET_SRC(value)			(((value) >> 30) & 0x3)
+#define STATUS_GET_RESP(value)			(((value) >> 28) & 0x3)
+#define STATUS_GET_LEN(value)			(((value) >> 24) & 0x7)
+#define STATUS_GET_SBP_STATUS(value)		(((value) >> 16) & 0xff)
+#define STATUS_GET_ORB_OFFSET_HI(value)		((value) & 0x0000ffff)
+#define STATUS_TEST_DEAD(value)			((value) & 0x08000000)
+/* test 'resp' | 'dead' | 'sbp2_status' */
+#define STATUS_TEST_RDS(value)			((value) & 0x38ff0000)
 
 struct sbp2_status_block {
 	u32 ORB_offset_hi_misc;
@@ -318,9 +320,9 @@ struct scsi_id_instance_data {
 	u64 status_fifo_addr;
 
 	/*
-	 * Variable used for logins, reconnects, logouts, query logins
+	 * Waitqueue flag for logins, reconnects, logouts, query logins
 	 */
-	atomic_t sbp2_login_complete;
+	int access_complete:1;
 
 	/*
 	 * Pool of command orbs, so we can have more than overlapped command per id
@@ -344,6 +346,16 @@ struct scsi_id_instance_data {
 
 	/* Device specific workarounds/brokeness */
 	unsigned workarounds;
+
+	atomic_t state;
+	struct work_struct protocol_work;
+};
+
+/* For use in scsi_id_instance_data.state */
+enum sbp2lu_state_types {
+	SBP2LU_STATE_RUNNING,		/* all normal */
+	SBP2LU_STATE_IN_RESET,		/* between bus reset and reconnect */
+	SBP2LU_STATE_IN_SHUTDOWN	/* when sbp2_remove was called */
 };
 
 /* Sbp2 host data structure (one per IEEE1394 host) */
@@ -390,11 +402,6 @@ static int sbp2_logout_device(struct scsi_id_instance_data *scsi_id);
 static int sbp2_handle_status_write(struct hpsb_host *host, int nodeid, int destid,
 				    quadlet_t *data, u64 addr, size_t length, u16 flags);
 static int sbp2_agent_reset(struct scsi_id_instance_data *scsi_id, int wait);
-static int sbp2_link_orb_command(struct scsi_id_instance_data *scsi_id,
-				 struct sbp2_command_info *command);
-static int sbp2_send_command(struct scsi_id_instance_data *scsi_id,
-			     struct scsi_cmnd *SCpnt,
-			     void (*done)(struct scsi_cmnd *));
 static unsigned int sbp2_status_to_sense_data(unchar *sbp2_status,
 					      unchar *sense_data);
 static void sbp2_parse_unit_directory(struct scsi_id_instance_data *scsi_id,
diff --git a/drivers/ieee1394/video1394.c b/drivers/ieee1394/video1394.c
index c6e3f02bc6d7..9bc65059cc69 100644
--- a/drivers/ieee1394/video1394.c
+++ b/drivers/ieee1394/video1394.c
@@ -49,16 +49,16 @@
 #include <linux/compat.h>
 #include <linux/cdev.h>
 
-#include "ieee1394.h"
-#include "ieee1394_types.h"
+#include "dma.h"
+#include "highlevel.h"
 #include "hosts.h"
+#include "ieee1394.h"
 #include "ieee1394_core.h"
-#include "highlevel.h"
-#include "video1394.h"
+#include "ieee1394_hotplug.h"
+#include "ieee1394_types.h"
 #include "nodemgr.h"
-#include "dma.h"
-
 #include "ohci1394.h"
+#include "video1394.h"
 
 #define ISO_CHANNELS 64
 
@@ -129,7 +129,7 @@ struct file_ctx {
 #define DBGMSG(card, fmt, args...) \
 printk(KERN_INFO "video1394_%d: " fmt "\n" , card , ## args)
 #else
-#define DBGMSG(card, fmt, args...)
+#define DBGMSG(card, fmt, args...) do {} while (0)
 #endif
 
 /* print general (card independent) information */
@@ -1181,7 +1181,8 @@ static int video1394_mmap(struct file *file, struct vm_area_struct *vma)
 
 	lock_kernel();
 	if (ctx->current_ctx == NULL) {
-		PRINT(KERN_ERR, ctx->ohci->host->id, "Current iso context not set");
+		PRINT(KERN_ERR, ctx->ohci->host->id,
+				"Current iso context not set");
 	} else
 		res = dma_region_mmap(&ctx->current_ctx->dma, file, vma);
 	unlock_kernel();
@@ -1189,6 +1190,40 @@ static int video1394_mmap(struct file *file, struct vm_area_struct *vma)
 	return res;
 }
 
+static unsigned int video1394_poll(struct file *file, poll_table *pt)
+{
+	struct file_ctx *ctx;
+	unsigned int mask = 0;
+	unsigned long flags;
+	struct dma_iso_ctx *d;
+	int i;
+
+	lock_kernel();
+	ctx = file->private_data;
+	d = ctx->current_ctx;
+	if (d == NULL) {
+		PRINT(KERN_ERR, ctx->ohci->host->id,
+				"Current iso context not set");
+		mask = POLLERR;
+		goto done;
+	}
+
+	poll_wait(file, &d->waitq, pt);
+
+	spin_lock_irqsave(&d->lock, flags);
+	for (i = 0; i < d->num_desc; i++) {
+		if (d->buffer_status[i] == VIDEO1394_BUFFER_READY) {
+			mask |= POLLIN | POLLRDNORM;
+			break;
+		}
+	}
+	spin_unlock_irqrestore(&d->lock, flags);
+done:
+	unlock_kernel();
+
+	return mask;
+}
+
 static int video1394_open(struct inode *inode, struct file *file)
 {
 	int i = ieee1394_file_to_instance(file);
@@ -1257,6 +1292,7 @@ static struct file_operations video1394_fops=
 #ifdef CONFIG_COMPAT
 	.compat_ioctl = video1394_compat_ioctl,
 #endif
+	.poll =		video1394_poll,
 	.mmap =		video1394_mmap,
 	.open =		video1394_open,
 	.release =	video1394_release