Merge git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6 into for-linus

33 files changed, 612 insertions, 346 deletions

diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig
index 3aa6c821449..16be41446b5 100644
--- a/arch/ia64/Kconfig
+++ b/arch/ia64/Kconfig
@@ -19,6 +19,7 @@ config IA64
 	select HAVE_OPROFILE
 	select HAVE_KPROBES
 	select HAVE_KRETPROBES
+	select HAVE_DMA_ATTRS
 	select HAVE_KVM
 	default y
 	help
@@ -47,6 +48,9 @@ config MMU
 config SWIOTLB
        bool
 
+config IOMMU_HELPER
+       bool
+
 config GENERIC_LOCKBREAK
 	bool
 	default y
@@ -131,6 +135,7 @@ config IA64_GENERIC
 	  HP-zx1/sx1000		For HP systems
 	  HP-zx1/sx1000+swiotlb	For HP systems with (broken) DMA-constrained devices.
 	  SGI-SN2		For SGI Altix systems
+	  SGI-UV		For SGI UV systems
 	  Ski-simulator		For the HP simulator <http://www.hpl.hp.com/research/linux/ski/>
 
 	  If you don't know what to do, choose "generic".
@@ -166,6 +171,18 @@ config IA64_SGI_SN2
 	  to select this option.  If in doubt, select ia64 generic support
 	  instead.
 
+config IA64_SGI_UV`
+	bool "SGI-UV`"
+	select NUMA
+	select ACPI_NUMA
+	select SWIOTLB
+	help
+	  Selecting this option will optimize the kernel for use on UV based
+	  systems, but the resulting kernel binary will not run on other
+	  types of ia64 systems.  If you have an SGI UV system, it's safe
+	  to select this option.  If in doubt, select ia64 generic support
+	  instead.
+
 config IA64_HP_SIM
 	bool "Ski-simulator"
 	select SWIOTLB
@@ -615,7 +632,7 @@ config IRQ_PER_CPU
 	default y
 
 config IOMMU_HELPER
-	def_bool (IA64_HP_ZX1 || IA64_HP_ZX1_SWIOTLB || IA64_GENERIC)
+	def_bool (IA64_HP_ZX1 || IA64_HP_ZX1_SWIOTLB || IA64_GENERIC || SWIOTLB)
 
 source "arch/ia64/hp/sim/Kconfig"
 
diff --git a/arch/ia64/Makefile b/arch/ia64/Makefile
index ec4cca477f4..88f1a55c6c9 100644
--- a/arch/ia64/Makefile
+++ b/arch/ia64/Makefile
@@ -63,7 +63,7 @@ drivers-$(CONFIG_PCI)		+= arch/ia64/pci/
 drivers-$(CONFIG_IA64_HP_SIM)	+= arch/ia64/hp/sim/
 drivers-$(CONFIG_IA64_HP_ZX1)	+= arch/ia64/hp/common/ arch/ia64/hp/zx1/
 drivers-$(CONFIG_IA64_HP_ZX1_SWIOTLB) += arch/ia64/hp/common/ arch/ia64/hp/zx1/
-drivers-$(CONFIG_IA64_GENERIC)	+= arch/ia64/hp/common/ arch/ia64/hp/zx1/ arch/ia64/hp/sim/ arch/ia64/sn/
+drivers-$(CONFIG_IA64_GENERIC)	+= arch/ia64/hp/common/ arch/ia64/hp/zx1/ arch/ia64/hp/sim/ arch/ia64/sn/ arch/ia64/uv/
 drivers-$(CONFIG_OPROFILE)	+= arch/ia64/oprofile/
 
 boot := arch/ia64/hp/sim/boot
diff --git a/arch/ia64/hp/common/hwsw_iommu.c b/arch/ia64/hp/common/hwsw_iommu.c
index 8f6bcfe1dad..1c44ec2a1d5 100644
--- a/arch/ia64/hp/common/hwsw_iommu.c
+++ b/arch/ia64/hp/common/hwsw_iommu.c
@@ -20,10 +20,10 @@
 extern int swiotlb_late_init_with_default_size (size_t size);
 extern ia64_mv_dma_alloc_coherent	swiotlb_alloc_coherent;
 extern ia64_mv_dma_free_coherent	swiotlb_free_coherent;
-extern ia64_mv_dma_map_single		swiotlb_map_single;
-extern ia64_mv_dma_unmap_single		swiotlb_unmap_single;
-extern ia64_mv_dma_map_sg		swiotlb_map_sg;
-extern ia64_mv_dma_unmap_sg		swiotlb_unmap_sg;
+extern ia64_mv_dma_map_single_attrs	swiotlb_map_single_attrs;
+extern ia64_mv_dma_unmap_single_attrs	swiotlb_unmap_single_attrs;
+extern ia64_mv_dma_map_sg_attrs		swiotlb_map_sg_attrs;
+extern ia64_mv_dma_unmap_sg_attrs	swiotlb_unmap_sg_attrs;
 extern ia64_mv_dma_supported		swiotlb_dma_supported;
 extern ia64_mv_dma_mapping_error	swiotlb_dma_mapping_error;
 
@@ -31,19 +31,19 @@ extern ia64_mv_dma_mapping_error	swiotlb_dma_mapping_error;
 
 extern ia64_mv_dma_alloc_coherent	sba_alloc_coherent;
 extern ia64_mv_dma_free_coherent	sba_free_coherent;
-extern ia64_mv_dma_map_single		sba_map_single;
-extern ia64_mv_dma_unmap_single		sba_unmap_single;
-extern ia64_mv_dma_map_sg		sba_map_sg;
-extern ia64_mv_dma_unmap_sg		sba_unmap_sg;
+extern ia64_mv_dma_map_single_attrs	sba_map_single_attrs;
+extern ia64_mv_dma_unmap_single_attrs	sba_unmap_single_attrs;
+extern ia64_mv_dma_map_sg_attrs		sba_map_sg_attrs;
+extern ia64_mv_dma_unmap_sg_attrs	sba_unmap_sg_attrs;
 extern ia64_mv_dma_supported		sba_dma_supported;
 extern ia64_mv_dma_mapping_error	sba_dma_mapping_error;
 
 #define hwiommu_alloc_coherent		sba_alloc_coherent
 #define hwiommu_free_coherent		sba_free_coherent
-#define hwiommu_map_single		sba_map_single
-#define hwiommu_unmap_single		sba_unmap_single
-#define hwiommu_map_sg			sba_map_sg
-#define hwiommu_unmap_sg		sba_unmap_sg
+#define hwiommu_map_single_attrs	sba_map_single_attrs
+#define hwiommu_unmap_single_attrs	sba_unmap_single_attrs
+#define hwiommu_map_sg_attrs		sba_map_sg_attrs
+#define hwiommu_unmap_sg_attrs		sba_unmap_sg_attrs
 #define hwiommu_dma_supported		sba_dma_supported
 #define hwiommu_dma_mapping_error	sba_dma_mapping_error
 #define hwiommu_sync_single_for_cpu	machvec_dma_sync_single
@@ -98,41 +98,48 @@ hwsw_free_coherent (struct device *dev, size_t size, void *vaddr, dma_addr_t dma
 }
 
 dma_addr_t
-hwsw_map_single (struct device *dev, void *addr, size_t size, int dir)
+hwsw_map_single_attrs(struct device *dev, void *addr, size_t size, int dir,
+		       struct dma_attrs *attrs)
 {
 	if (use_swiotlb(dev))
-		return swiotlb_map_single(dev, addr, size, dir);
+		return swiotlb_map_single_attrs(dev, addr, size, dir, attrs);
 	else
-		return hwiommu_map_single(dev, addr, size, dir);
+		return hwiommu_map_single_attrs(dev, addr, size, dir, attrs);
 }
+EXPORT_SYMBOL(hwsw_map_single_attrs);
 
 void
-hwsw_unmap_single (struct device *dev, dma_addr_t iova, size_t size, int dir)
+hwsw_unmap_single_attrs(struct device *dev, dma_addr_t iova, size_t size,
+			 int dir, struct dma_attrs *attrs)
 {
 	if (use_swiotlb(dev))
-		return swiotlb_unmap_single(dev, iova, size, dir);
+		return swiotlb_unmap_single_attrs(dev, iova, size, dir, attrs);
 	else
-		return hwiommu_unmap_single(dev, iova, size, dir);
+		return hwiommu_unmap_single_attrs(dev, iova, size, dir, attrs);
 }
-
+EXPORT_SYMBOL(hwsw_unmap_single_attrs);
 
 int
-hwsw_map_sg (struct device *dev, struct scatterlist *sglist, int nents, int dir)
+hwsw_map_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
+		   int dir, struct dma_attrs *attrs)
 {
 	if (use_swiotlb(dev))
-		return swiotlb_map_sg(dev, sglist, nents, dir);
+		return swiotlb_map_sg_attrs(dev, sglist, nents, dir, attrs);
 	else
-		return hwiommu_map_sg(dev, sglist, nents, dir);
+		return hwiommu_map_sg_attrs(dev, sglist, nents, dir, attrs);
 }
+EXPORT_SYMBOL(hwsw_map_sg_attrs);
 
 void
-hwsw_unmap_sg (struct device *dev, struct scatterlist *sglist, int nents, int dir)
+hwsw_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
+		     int dir, struct dma_attrs *attrs)
 {
 	if (use_swiotlb(dev))
-		return swiotlb_unmap_sg(dev, sglist, nents, dir);
+		return swiotlb_unmap_sg_attrs(dev, sglist, nents, dir, attrs);
 	else
-		return hwiommu_unmap_sg(dev, sglist, nents, dir);
+		return hwiommu_unmap_sg_attrs(dev, sglist, nents, dir, attrs);
 }
+EXPORT_SYMBOL(hwsw_unmap_sg_attrs);
 
 void
 hwsw_sync_single_for_cpu (struct device *dev, dma_addr_t addr, size_t size, int dir)
@@ -185,10 +192,6 @@ hwsw_dma_mapping_error (dma_addr_t dma_addr)
 }
 
 EXPORT_SYMBOL(hwsw_dma_mapping_error);
-EXPORT_SYMBOL(hwsw_map_single);
-EXPORT_SYMBOL(hwsw_unmap_single);
-EXPORT_SYMBOL(hwsw_map_sg);
-EXPORT_SYMBOL(hwsw_unmap_sg);
 EXPORT_SYMBOL(hwsw_dma_supported);
 EXPORT_SYMBOL(hwsw_alloc_coherent);
 EXPORT_SYMBOL(hwsw_free_coherent);
diff --git a/arch/ia64/hp/common/sba_iommu.c b/arch/ia64/hp/common/sba_iommu.c
index 9409de5c944..34421aed1e2 100644
--- a/arch/ia64/hp/common/sba_iommu.c
+++ b/arch/ia64/hp/common/sba_iommu.c
@@ -899,16 +899,18 @@ sba_mark_invalid(struct ioc *ioc, dma_addr_t iova, size_t byte_cnt)
 }
 
 /**
- * sba_map_single - map one buffer and return IOVA for DMA
+ * sba_map_single_attrs - map one buffer and return IOVA for DMA
  * @dev: instance of PCI owned by the driver that's asking.
  * @addr:  driver buffer to map.
  * @size:  number of bytes to map in driver buffer.
  * @dir:  R/W or both.
+ * @attrs: optional dma attributes
  *
  * See Documentation/DMA-mapping.txt
  */
 dma_addr_t
-sba_map_single(struct device *dev, void *addr, size_t size, int dir)
+sba_map_single_attrs(struct device *dev, void *addr, size_t size, int dir,
+		     struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
 	dma_addr_t iovp;
@@ -932,7 +934,8 @@ sba_map_single(struct device *dev, void *addr, size_t size, int dir)
  		** Device is bit capable of DMA'ing to the buffer...
 		** just return the PCI address of ptr
  		*/
-		DBG_BYPASS("sba_map_single() bypass mask/addr: 0x%lx/0x%lx\n",
+		DBG_BYPASS("sba_map_single_attrs() bypass mask/addr: "
+			   "0x%lx/0x%lx\n",
 		           to_pci_dev(dev)->dma_mask, pci_addr);
 		return pci_addr;
 	}
@@ -953,7 +956,7 @@ sba_map_single(struct device *dev, void *addr, size_t size, int dir)
 
 #ifdef ASSERT_PDIR_SANITY
 	spin_lock_irqsave(&ioc->res_lock, flags);
-	if (sba_check_pdir(ioc,"Check before sba_map_single()"))
+	if (sba_check_pdir(ioc,"Check before sba_map_single_attrs()"))
 		panic("Sanity check failed");
 	spin_unlock_irqrestore(&ioc->res_lock, flags);
 #endif
@@ -982,11 +985,12 @@ sba_map_single(struct device *dev, void *addr, size_t size, int dir)
 	/* form complete address */
 #ifdef ASSERT_PDIR_SANITY
 	spin_lock_irqsave(&ioc->res_lock, flags);
-	sba_check_pdir(ioc,"Check after sba_map_single()");
+	sba_check_pdir(ioc,"Check after sba_map_single_attrs()");
 	spin_unlock_irqrestore(&ioc->res_lock, flags);
 #endif
 	return SBA_IOVA(ioc, iovp, offset);
 }
+EXPORT_SYMBOL(sba_map_single_attrs);
 
 #ifdef ENABLE_MARK_CLEAN
 static SBA_INLINE void
@@ -1013,15 +1017,17 @@ sba_mark_clean(struct ioc *ioc, dma_addr_t iova, size_t size)
 #endif
 
 /**
- * sba_unmap_single - unmap one IOVA and free resources
+ * sba_unmap_single_attrs - unmap one IOVA and free resources
  * @dev: instance of PCI owned by the driver that's asking.
  * @iova:  IOVA of driver buffer previously mapped.
  * @size:  number of bytes mapped in driver buffer.
  * @dir:  R/W or both.
+ * @attrs: optional dma attributes
  *
  * See Documentation/DMA-mapping.txt
  */
-void sba_unmap_single(struct device *dev, dma_addr_t iova, size_t size, int dir)
+void sba_unmap_single_attrs(struct device *dev, dma_addr_t iova, size_t size,
+			    int dir, struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
 #if DELAYED_RESOURCE_CNT > 0
@@ -1038,7 +1044,8 @@ void sba_unmap_single(struct device *dev, dma_addr_t iova, size_t size, int dir)
 		/*
 		** Address does not fall w/in IOVA, must be bypassing
 		*/
-		DBG_BYPASS("sba_unmap_single() bypass addr: 0x%lx\n", iova);
+		DBG_BYPASS("sba_unmap_single_atttrs() bypass addr: 0x%lx\n",
+			   iova);
 
 #ifdef ENABLE_MARK_CLEAN
 		if (dir == DMA_FROM_DEVICE) {
@@ -1087,7 +1094,7 @@ void sba_unmap_single(struct device *dev, dma_addr_t iova, size_t size, int dir)
 	spin_unlock_irqrestore(&ioc->res_lock, flags);
 #endif /* DELAYED_RESOURCE_CNT == 0 */
 }
-
+EXPORT_SYMBOL(sba_unmap_single_attrs);
 
 /**
  * sba_alloc_coherent - allocate/map shared mem for DMA
@@ -1144,7 +1151,8 @@ sba_alloc_coherent (struct device *dev, size_t size, dma_addr_t *dma_handle, gfp
 	 * If device can't bypass or bypass is disabled, pass the 32bit fake
 	 * device to map single to get an iova mapping.
 	 */
-	*dma_handle = sba_map_single(&ioc->sac_only_dev->dev, addr, size, 0);
+	*dma_handle = sba_map_single_attrs(&ioc->sac_only_dev->dev, addr,
+					   size, 0, NULL);
 
 	return addr;
 }
@@ -1161,7 +1169,7 @@ sba_alloc_coherent (struct device *dev, size_t size, dma_addr_t *dma_handle, gfp
  */
 void sba_free_coherent (struct device *dev, size_t size, void *vaddr, dma_addr_t dma_handle)
 {
-	sba_unmap_single(dev, dma_handle, size, 0);
+	sba_unmap_single_attrs(dev, dma_handle, size, 0, NULL);
 	free_pages((unsigned long) vaddr, get_order(size));
 }
 
@@ -1410,10 +1418,12 @@ sba_coalesce_chunks(struct ioc *ioc, struct device *dev,
  * @sglist:  array of buffer/length pairs
  * @nents:  number of entries in list
  * @dir:  R/W or both.
+ * @attrs: optional dma attributes
  *
  * See Documentation/DMA-mapping.txt
  */
-int sba_map_sg(struct device *dev, struct scatterlist *sglist, int nents, int dir)
+int sba_map_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
+		     int dir, struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
 	int coalesced, filled = 0;
@@ -1441,16 +1451,16 @@ int sba_map_sg(struct device *dev, struct scatterlist *sglist, int nents, int di
 	/* Fast path single entry scatterlists. */
 	if (nents == 1) {
 		sglist->dma_length = sglist->length;
-		sglist->dma_address = sba_map_single(dev, sba_sg_address(sglist), sglist->length, dir);
+		sglist->dma_address = sba_map_single_attrs(dev, sba_sg_address(sglist), sglist->length, dir, attrs);
 		return 1;
 	}
 
 #ifdef ASSERT_PDIR_SANITY
 	spin_lock_irqsave(&ioc->res_lock, flags);
-	if (sba_check_pdir(ioc,"Check before sba_map_sg()"))
+	if (sba_check_pdir(ioc,"Check before sba_map_sg_attrs()"))
 	{
 		sba_dump_sg(ioc, sglist, nents);
-		panic("Check before sba_map_sg()");
+		panic("Check before sba_map_sg_attrs()");
 	}
 	spin_unlock_irqrestore(&ioc->res_lock, flags);
 #endif
@@ -1479,10 +1489,10 @@ int sba_map_sg(struct device *dev, struct scatterlist *sglist, int nents, int di
 
 #ifdef ASSERT_PDIR_SANITY
 	spin_lock_irqsave(&ioc->res_lock, flags);
-	if (sba_check_pdir(ioc,"Check after sba_map_sg()"))
+	if (sba_check_pdir(ioc,"Check after sba_map_sg_attrs()"))
 	{
 		sba_dump_sg(ioc, sglist, nents);
-		panic("Check after sba_map_sg()\n");
+		panic("Check after sba_map_sg_attrs()\n");
 	}
 	spin_unlock_irqrestore(&ioc->res_lock, flags);
 #endif
@@ -1492,18 +1502,20 @@ int sba_map_sg(struct device *dev, struct scatterlist *sglist, int nents, int di
 
 	return filled;
 }
-
+EXPORT_SYMBOL(sba_map_sg_attrs);
 
 /**
- * sba_unmap_sg - unmap Scatter/Gather list
+ * sba_unmap_sg_attrs - unmap Scatter/Gather list
  * @dev: instance of PCI owned by the driver that's asking.
  * @sglist:  array of buffer/length pairs
  * @nents:  number of entries in list
  * @dir:  R/W or both.
+ * @attrs: optional dma attributes
  *
  * See Documentation/DMA-mapping.txt
  */
-void sba_unmap_sg (struct device *dev, struct scatterlist *sglist, int nents, int dir)
+void sba_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist,
+			int nents, int dir, struct dma_attrs *attrs)
 {
 #ifdef ASSERT_PDIR_SANITY
 	struct ioc *ioc;
@@ -1518,13 +1530,14 @@ void sba_unmap_sg (struct device *dev, struct scatterlist *sglist, int nents, in
 	ASSERT(ioc);
 
 	spin_lock_irqsave(&ioc->res_lock, flags);
-	sba_check_pdir(ioc,"Check before sba_unmap_sg()");
+	sba_check_pdir(ioc,"Check before sba_unmap_sg_attrs()");
 	spin_unlock_irqrestore(&ioc->res_lock, flags);
 #endif
 
 	while (nents && sglist->dma_length) {
 
-		sba_unmap_single(dev, sglist->dma_address, sglist->dma_length, dir);
+		sba_unmap_single_attrs(dev, sglist->dma_address,
+				       sglist->dma_length, dir, attrs);
 		sglist = sg_next(sglist);
 		nents--;
 	}
@@ -1533,11 +1546,12 @@ void sba_unmap_sg (struct device *dev, struct scatterlist *sglist, int nents, in
 
 #ifdef ASSERT_PDIR_SANITY
 	spin_lock_irqsave(&ioc->res_lock, flags);
-	sba_check_pdir(ioc,"Check after sba_unmap_sg()");
+	sba_check_pdir(ioc,"Check after sba_unmap_sg_attrs()");
 	spin_unlock_irqrestore(&ioc->res_lock, flags);
 #endif
 
 }
+EXPORT_SYMBOL(sba_unmap_sg_attrs);
 
 /**************************************************************
 *
@@ -1918,15 +1932,13 @@ static const struct file_operations ioc_fops = {
 static void __init
 ioc_proc_init(void)
 {
-	struct proc_dir_entry *dir, *entry;
+	struct proc_dir_entry *dir;
 
 	dir = proc_mkdir("bus/mckinley", NULL);
 	if (!dir)
 		return;
 
-	entry = create_proc_entry(ioc_list->name, 0, dir);
-	if (entry)
-		entry->proc_fops = &ioc_fops;
+	proc_create(ioc_list->name, 0, dir, &ioc_fops);
 }
 #endif
 
@@ -2166,10 +2178,6 @@ sba_page_override(char *str)
 __setup("sbapagesize=",sba_page_override);
 
 EXPORT_SYMBOL(sba_dma_mapping_error);
-EXPORT_SYMBOL(sba_map_single);
-EXPORT_SYMBOL(sba_unmap_single);
-EXPORT_SYMBOL(sba_map_sg);
-EXPORT_SYMBOL(sba_unmap_sg);
 EXPORT_SYMBOL(sba_dma_supported);
 EXPORT_SYMBOL(sba_alloc_coherent);
 EXPORT_SYMBOL(sba_free_coherent);
diff --git a/arch/ia64/hp/sim/simserial.c b/arch/ia64/hp/sim/simserial.c
index eb0c32a85fd..23cafc80d2a 100644
--- a/arch/ia64/hp/sim/simserial.c
+++ b/arch/ia64/hp/sim/simserial.c
@@ -210,21 +210,23 @@ static void do_softint(struct work_struct *private_)
 	printk(KERN_ERR "simserial: do_softint called\n");
 }
 
-static void rs_put_char(struct tty_struct *tty, unsigned char ch)
+static int rs_put_char(struct tty_struct *tty, unsigned char ch)
 {
 	struct async_struct *info = (struct async_struct *)tty->driver_data;
 	unsigned long flags;
 
-	if (!tty || !info->xmit.buf) return;
+	if (!tty || !info->xmit.buf)
+		return 0;
 
 	local_irq_save(flags);
 	if (CIRC_SPACE(info->xmit.head, info->xmit.tail, SERIAL_XMIT_SIZE) == 0) {
 		local_irq_restore(flags);
-		return;
+		return 0;
 	}
 	info->xmit.buf[info->xmit.head] = ch;
 	info->xmit.head = (info->xmit.head + 1) & (SERIAL_XMIT_SIZE-1);
 	local_irq_restore(flags);
+	return 1;
 }
 
 static void transmit_chars(struct async_struct *info, int *intr_done)
@@ -621,7 +623,8 @@ static void rs_close(struct tty_struct *tty, struct file * filp)
 	 * the line discipline to only process XON/XOFF characters.
 	 */
 	shutdown(info);
-	if (tty->driver->flush_buffer) tty->driver->flush_buffer(tty);
+	if (tty->ops->flush_buffer)
+		tty->ops->flush_buffer(tty);
 	if (tty->ldisc.flush_buffer) tty->ldisc.flush_buffer(tty);
 	info->event = 0;
 	info->tty = NULL;
diff --git a/arch/ia64/ia32/ia32_signal.c b/arch/ia64/ia32/ia32_signal.c
index 256a7faeda0..b763ca19ef1 100644
--- a/arch/ia64/ia32/ia32_signal.c
+++ b/arch/ia64/ia32/ia32_signal.c
@@ -463,7 +463,7 @@ sys32_sigsuspend (int history0, int history1, old_sigset_t mask)
 
 	current->state = TASK_INTERRUPTIBLE;
 	schedule();
-	set_thread_flag(TIF_RESTORE_SIGMASK);
+	set_restore_sigmask();
 	return -ERESTARTNOHAND;
 }
 
diff --git a/arch/ia64/ia32/ia32_support.c b/arch/ia64/ia32/ia32_support.c
index 896b1ebbfb2..a6965ddafc4 100644
--- a/arch/ia64/ia32/ia32_support.c
+++ b/arch/ia64/ia32/ia32_support.c
@@ -15,7 +15,6 @@
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/mm.h>
-#include <linux/personality.h>
 #include <linux/sched.h>
 
 #include <asm/intrinsics.h>
@@ -29,7 +28,6 @@
 
 extern int die_if_kernel (char *str, struct pt_regs *regs, long err);
 
-struct exec_domain ia32_exec_domain;
 struct page *ia32_shared_page[NR_CPUS];
 unsigned long *ia32_boot_gdt;
 unsigned long *cpu_gdt_table[NR_CPUS];
@@ -240,14 +238,6 @@ ia32_cpu_init (void)
 static int __init
 ia32_init (void)
 {
-	ia32_exec_domain.name = "Linux/x86";
-	ia32_exec_domain.handler = NULL;
-	ia32_exec_domain.pers_low = PER_LINUX32;
-	ia32_exec_domain.pers_high = PER_LINUX32;
-	ia32_exec_domain.signal_map = default_exec_domain.signal_map;
-	ia32_exec_domain.signal_invmap = default_exec_domain.signal_invmap;
-	register_exec_domain(&ia32_exec_domain);
-
 #if PAGE_SHIFT > IA32_PAGE_SHIFT
 	{
 		extern struct kmem_cache *ia64_partial_page_cachep;
diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c
index c7467f863c7..853d1f11be0 100644
--- a/arch/ia64/kernel/acpi.c
+++ b/arch/ia64/kernel/acpi.c
@@ -117,7 +117,10 @@ acpi_get_sysname(void)
 	if (!strcmp(hdr->oem_id, "HP")) {
 		return "hpzx1";
 	} else if (!strcmp(hdr->oem_id, "SGI")) {
-		return "sn2";
+		if (!strcmp(hdr->oem_table_id + 4, "UV"))
+			return "uv";
+		else
+			return "sn2";
 	}
 
 	return "dig";
@@ -130,6 +133,8 @@ acpi_get_sysname(void)
 	return "hpzx1_swiotlb";
 # elif defined (CONFIG_IA64_SGI_SN2)
 	return "sn2";
+# elif defined (CONFIG_IA64_SGI_UV)
+	return "uv";
 # elif defined (CONFIG_IA64_DIG)
 	return "dig";
 # else
@@ -622,6 +627,9 @@ void acpi_unregister_gsi(u32 gsi)
 	if (acpi_irq_model == ACPI_IRQ_MODEL_PLATFORM)
 		return;
 
+	if (has_8259 && gsi < 16)
+		return;
+
 	iosapic_unregister_intr(gsi);
 }
 
@@ -966,7 +974,7 @@ acpi_map_iosapics (void)
 fs_initcall(acpi_map_iosapics);
 #endif				/* CONFIG_ACPI_NUMA */
 
-int acpi_register_ioapic(acpi_handle handle, u64 phys_addr, u32 gsi_base)
+int __ref acpi_register_ioapic(acpi_handle handle, u64 phys_addr, u32 gsi_base)
 {
 	int err;
 
diff --git a/arch/ia64/kernel/asm-offsets.c b/arch/ia64/kernel/asm-offsets.c
index 230a6f92367..c64a55af9b9 100644
--- a/arch/ia64/kernel/asm-offsets.c
+++ b/arch/ia64/kernel/asm-offsets.c
@@ -9,7 +9,7 @@
 #include <linux/sched.h>
 #include <linux/pid.h>
 #include <linux/clocksource.h>
-
+#include <linux/kbuild.h>
 #include <asm-ia64/processor.h>
 #include <asm-ia64/ptrace.h>
 #include <asm-ia64/siginfo.h>
@@ -19,11 +19,6 @@
 #include "../kernel/sigframe.h"
 #include "../kernel/fsyscall_gtod_data.h"
 
-#define DEFINE(sym, val) \
-        asm volatile("\n->" #sym " %0 " #val : : "i" (val))
-
-#define BLANK() asm volatile("\n->" : : )
-
 void foo(void)
 {
 	DEFINE(IA64_TASK_SIZE, sizeof (struct task_struct));
diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S
index e49ad8c5dc6..ca2bb95726d 100644
--- a/arch/ia64/kernel/entry.S
+++ b/arch/ia64/kernel/entry.S
@@ -1156,6 +1156,9 @@ skip_rbs_switch:
 	 *	r31 = current->thread_info->flags
 	 * On exit:
 	 *	p6 = TRUE if work-pending-check needs to be redone
+	 *
+	 * Interrupts are disabled on entry, reenabled depend on work, and
+	 * disabled on exit.
 	 */
 .work_pending_syscall:
 	add r2=-8,r2
@@ -1164,16 +1167,16 @@ skip_rbs_switch:
 	st8 [r2]=r8
 	st8 [r3]=r10
 .work_pending:
-	tbit.z p6,p0=r31,TIF_NEED_RESCHED		// current_thread_info()->need_resched==0?
+	tbit.z p6,p0=r31,TIF_NEED_RESCHED	// is resched not needed?
 (p6)	br.cond.sptk.few .notify
 #ifdef CONFIG_PREEMPT
 (pKStk) dep r21=-1,r0,PREEMPT_ACTIVE_BIT,1
 	;;
 (pKStk) st4 [r20]=r21
-	ssm psr.i		// enable interrupts
 #endif
+	ssm psr.i		// enable interrupts
 	br.call.spnt.many rp=schedule
-.ret9:	cmp.eq p6,p0=r0,r0				// p6 <- 1
+.ret9:	cmp.eq p6,p0=r0,r0	// p6 <- 1 (re-check)
 	rsm psr.i		// disable interrupts
 	;;
 #ifdef CONFIG_PREEMPT
@@ -1182,13 +1185,13 @@ skip_rbs_switch:
 (pKStk)	st4 [r20]=r0		// preempt_count() <- 0
 #endif
 (pLvSys)br.cond.sptk.few  .work_pending_syscall_end
-	br.cond.sptk.many .work_processed_kernel	// re-check
+	br.cond.sptk.many .work_processed_kernel
 
 .notify:
 (pUStk)	br.call.spnt.many rp=notify_resume_user
-.ret10:	cmp.ne p6,p0=r0,r0				// p6 <- 0
+.ret10:	cmp.ne p6,p0=r0,r0	// p6 <- 0 (don't re-check)
 (pLvSys)br.cond.sptk.few  .work_pending_syscall_end
-	br.cond.sptk.many .work_processed_kernel	// don't re-check
+	br.cond.sptk.many .work_processed_kernel
 
 .work_pending_syscall_end:
 	adds r2=PT(R8)+16,r12
@@ -1196,7 +1199,7 @@ skip_rbs_switch:
 	;;
 	ld8 r8=[r2]
 	ld8 r10=[r3]
-	br.cond.sptk.many .work_processed_syscall	// re-check
+	br.cond.sptk.many .work_processed_syscall
 
 END(ia64_leave_kernel)
 
@@ -1234,9 +1237,12 @@ GLOBAL_ENTRY(ia64_invoke_schedule_tail)
 END(ia64_invoke_schedule_tail)
 
 	/*
-	 * Setup stack and call do_notify_resume_user().  Note that pSys and pNonSys need to
-	 * be set up by the caller.  We declare 8 input registers so the system call
-	 * args get preserved, in case we need to restart a system call.
+	 * Setup stack and call do_notify_resume_user(), keeping interrupts
+	 * disabled.
+	 *
+	 * Note that pSys and pNonSys need to be set up by the caller.
+	 * We declare 8 input registers so the system call args get preserved,
+	 * in case we need to restart a system call.
 	 */
 ENTRY(notify_resume_user)
 	.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
diff --git a/arch/ia64/kernel/irq.c b/arch/ia64/kernel/irq.c
index 6dee579f205..7fd18f54c05 100644
--- a/arch/ia64/kernel/irq.c
+++ b/arch/ia64/kernel/irq.c
@@ -183,10 +183,10 @@ void fixup_irqs(void)
 {
 	unsigned int irq;
 	extern void ia64_process_pending_intr(void);
-	extern void ia64_disable_timer(void);
 	extern volatile int time_keeper_id;
 
-	ia64_disable_timer();
+	/* Mask ITV to disable timer */
+	ia64_set_itv(1 << 16);
 
 	/*
 	 * Find a new timesync master
diff --git a/arch/ia64/kernel/palinfo.c b/arch/ia64/kernel/palinfo.c
index 396004e8cd1..9dc00f7fe10 100644
--- a/arch/ia64/kernel/palinfo.c
+++ b/arch/ia64/kernel/palinfo.c
@@ -900,12 +900,6 @@ static void
 palinfo_smp_call(void *info)
 {
 	palinfo_smp_data_t *data = (palinfo_smp_data_t *)info;
-	if (data == NULL) {
-		printk(KERN_ERR "palinfo: data pointer is NULL\n");
-		data->ret = 0; /* no output */
-		return;
-	}
-	/* does this actual call */
 	data->ret = (*data->func)(data->page);
 }
 
@@ -1053,7 +1047,7 @@ static int __cpuinit palinfo_cpu_callback(struct notifier_block *nfb,
 	return NOTIFY_OK;
 }
 
-static struct notifier_block palinfo_cpu_notifier __cpuinitdata =
+static struct notifier_block __refdata palinfo_cpu_notifier =
 {
 	.notifier_call = palinfo_cpu_callback,
 	.priority = 0,
diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c
index c8e403752a0..71d05133f55 100644
--- a/arch/ia64/kernel/perfmon.c
+++ b/arch/ia64/kernel/perfmon.c
@@ -867,7 +867,7 @@ pfm_rvfree(void *mem, unsigned long size)
 }
 
 static pfm_context_t *
-pfm_context_alloc(void)
+pfm_context_alloc(int ctx_flags)
 {
 	pfm_context_t *ctx;
 
@@ -878,6 +878,46 @@ pfm_context_alloc(void)
 	ctx = kzalloc(sizeof(pfm_context_t), GFP_KERNEL);
 	if (ctx) {
 		DPRINT(("alloc ctx @%p\n", ctx));
+
+		/*
+		 * init context protection lock
+		 */
+		spin_lock_init(&ctx->ctx_lock);
+
+		/*
+		 * context is unloaded
+		 */
+		ctx->ctx_state = PFM_CTX_UNLOADED;
+
+		/*
+		 * initialization of context's flags
+		 */
+		ctx->ctx_fl_block       = (ctx_flags & PFM_FL_NOTIFY_BLOCK) ? 1 : 0;
+		ctx->ctx_fl_system      = (ctx_flags & PFM_FL_SYSTEM_WIDE) ? 1: 0;
+		ctx->ctx_fl_no_msg      = (ctx_flags & PFM_FL_OVFL_NO_MSG) ? 1: 0;
+		/*
+		 * will move to set properties
+		 * ctx->ctx_fl_excl_idle   = (ctx_flags & PFM_FL_EXCL_IDLE) ? 1: 0;
+		 */
+
+		/*
+		 * init restart semaphore to locked
+		 */
+		init_completion(&ctx->ctx_restart_done);
+
+		/*
+		 * activation is used in SMP only
+		 */
+		ctx->ctx_last_activation = PFM_INVALID_ACTIVATION;
+		SET_LAST_CPU(ctx, -1);
+
+		/*
+		 * initialize notification message queue
+		 */
+		ctx->ctx_msgq_head = ctx->ctx_msgq_tail = 0;
+		init_waitqueue_head(&ctx->ctx_msgq_wait);
+		init_waitqueue_head(&ctx->ctx_zombieq);
+
 	}
 	return ctx;
 }
@@ -2165,28 +2205,21 @@ static struct dentry_operations pfmfs_dentry_operations = {
 };
 
 
-static int
-pfm_alloc_fd(struct file **cfile)
+static struct file *
+pfm_alloc_file(pfm_context_t *ctx)
 {
-	int fd, ret = 0;
-	struct file *file = NULL;
-	struct inode * inode;
+	struct file *file;
+	struct inode *inode;
+	struct dentry *dentry;
 	char name[32];
 	struct qstr this;
 
-	fd = get_unused_fd();
-	if (fd < 0) return -ENFILE;
-
-	ret = -ENFILE;
-
-	file = get_empty_filp();
-	if (!file) goto out;
-
 	/*
 	 * allocate a new inode
 	 */
 	inode = new_inode(pfmfs_mnt->mnt_sb);
-	if (!inode) goto out;
+	if (!inode)
+		return ERR_PTR(-ENOMEM);
 
 	DPRINT(("new inode ino=%ld @%p\n", inode->i_ino, inode));
 
@@ -2199,59 +2232,28 @@ pfm_alloc_fd(struct file **cfile)
 	this.len  = strlen(name);
 	this.hash = inode->i_ino;
 
-	ret = -ENOMEM;
-
 	/*
 	 * allocate a new dcache entry
 	 */
-	file->f_path.dentry = d_alloc(pfmfs_mnt->mnt_sb->s_root, &this);
-	if (!file->f_path.dentry) goto out;
+	dentry = d_alloc(pfmfs_mnt->mnt_sb->s_root, &this);
+	if (!dentry) {
+		iput(inode);
+		return ERR_PTR(-ENOMEM);
+	}
 
-	file->f_path.dentry->d_op = &pfmfs_dentry_operations;
+	dentry->d_op = &pfmfs_dentry_operations;
+	d_add(dentry, inode);
 
-	d_add(file->f_path.dentry, inode);
-	file->f_path.mnt = mntget(pfmfs_mnt);
-	file->f_mapping = inode->i_mapping;
+	file = alloc_file(pfmfs_mnt, dentry, FMODE_READ, &pfm_file_ops);
+	if (!file) {
+		dput(dentry);
+		return ERR_PTR(-ENFILE);
+	}
 
-	file->f_op    = &pfm_file_ops;
-	file->f_mode  = FMODE_READ;
 	file->f_flags = O_RDONLY;
-	file->f_pos   = 0;
-
-	/*
-	 * may have to delay until context is attached?
-	 */
-	fd_install(fd, file);
-
-	/*
-	 * the file structure we will use
-	 */
-	*cfile = file;
+	file->private_data = ctx;
 
-	return fd;
-out:
-	if (file) put_filp(file);
-	put_unused_fd(fd);
-	return ret;
-}
-
-static void
-pfm_free_fd(int fd, struct file *file)
-{
-	struct files_struct *files = current->files;
-	struct fdtable *fdt;
-
-	/* 
-	 * there ie no fd_uninstall(), so we do it here
-	 */
-	spin_lock(&files->file_lock);
-	fdt = files_fdtable(files);
-	rcu_assign_pointer(fdt->fd[fd], NULL);
-	spin_unlock(&files->file_lock);
-
-	if (file)
-		put_filp(file);
-	put_unused_fd(fd);
+	return file;
 }
 
 static int
@@ -2475,6 +2477,7 @@ pfm_setup_buffer_fmt(struct task_struct *task, struct file *filp, pfm_context_t
 
 	/* link buffer format and context */
 	ctx->ctx_buf_fmt = fmt;
+	ctx->ctx_fl_is_sampling = 1; /* assume record() is defined */
 
 	/*
 	 * check if buffer format wants to use perfmon buffer allocation/mapping service
@@ -2669,78 +2672,45 @@ pfm_context_create(pfm_context_t *ctx, void *arg, int count, struct pt_regs *reg
 {
 	pfarg_context_t *req = (pfarg_context_t *)arg;
 	struct file *filp;
+	struct path path;
 	int ctx_flags;
+	int fd;
 	int ret;
 
 	/* let's check the arguments first */
 	ret = pfarg_is_sane(current, req);
-	if (ret < 0) return ret;
+	if (ret < 0)
+		return ret;
 
 	ctx_flags = req->ctx_flags;
 
 	ret = -ENOMEM;
 
-	ctx = pfm_context_alloc();
-	if (!ctx) goto error;
+	fd = get_unused_fd();
+	if (fd < 0)
+		return fd;
 
-	ret = pfm_alloc_fd(&filp);
-	if (ret < 0) goto error_file;
+	ctx = pfm_context_alloc(ctx_flags);
+	if (!ctx)
+		goto error;
 
-	req->ctx_fd = ctx->ctx_fd = ret;
+	filp = pfm_alloc_file(ctx);
+	if (IS_ERR(filp)) {
+		ret = PTR_ERR(filp);
+		goto error_file;
+	}
 
-	/*
-	 * attach context to file
-	 */
-	filp->private_data = ctx;
+	req->ctx_fd = ctx->ctx_fd = fd;
 
 	/*
 	 * does the user want to sample?
 	 */
 	if (pfm_uuid_cmp(req->ctx_smpl_buf_id, pfm_null_uuid)) {
 		ret = pfm_setup_buffer_fmt(current, filp, ctx, ctx_flags, 0, req);
-		if (ret) goto buffer_error;
+		if (ret)
+			goto buffer_error;
 	}
 
-	/*
-	 * init context protection lock
-	 */
-	spin_lock_init(&ctx->ctx_lock);
-
-	/*
-	 * context is unloaded
-	 */
-	ctx->ctx_state = PFM_CTX_UNLOADED;
-
-	/*
-	 * initialization of context's flags
-	 */
-	ctx->ctx_fl_block       = (ctx_flags & PFM_FL_NOTIFY_BLOCK) ? 1 : 0;
-	ctx->ctx_fl_system      = (ctx_flags & PFM_FL_SYSTEM_WIDE) ? 1: 0;
-	ctx->ctx_fl_is_sampling = ctx->ctx_buf_fmt ? 1 : 0; /* assume record() is defined */
-	ctx->ctx_fl_no_msg      = (ctx_flags & PFM_FL_OVFL_NO_MSG) ? 1: 0;
-	/*
-	 * will move to set properties
-	 * ctx->ctx_fl_excl_idle   = (ctx_flags & PFM_FL_EXCL_IDLE) ? 1: 0;
-	 */
-
-	/*
-	 * init restart semaphore to locked
-	 */
-	init_completion(&ctx->ctx_restart_done);
-
-	/*
-	 * activation is used in SMP only
-	 */
-	ctx->ctx_last_activation = PFM_INVALID_ACTIVATION;
-	SET_LAST_CPU(ctx, -1);
-
-	/*
-	 * initialize notification message queue
-	 */
-	ctx->ctx_msgq_head = ctx->ctx_msgq_tail = 0;
-	init_waitqueue_head(&ctx->ctx_msgq_wait);
-	init_waitqueue_head(&ctx->ctx_zombieq);
-
 	DPRINT(("ctx=%p flags=0x%x system=%d notify_block=%d excl_idle=%d no_msg=%d ctx_fd=%d \n",
 		ctx,
 		ctx_flags,
@@ -2755,10 +2725,14 @@ pfm_context_create(pfm_context_t *ctx, void *arg, int count, struct pt_regs *reg
 	 */
 	pfm_reset_pmu_state(ctx);
 
+	fd_install(fd, filp);
+
 	return 0;
 
 buffer_error:
-	pfm_free_fd(ctx->ctx_fd, filp);
+	path = filp->f_path;
+	put_filp(filp);
+	path_put(&path);
 
 	if (ctx->ctx_buf_fmt) {
 		pfm_buf_fmt_exit(ctx->ctx_buf_fmt, current, NULL, regs);
@@ -2767,6 +2741,7 @@ error_file:
 	pfm_context_free(ctx);
 
 error:
+	put_unused_fd(fd);
 	return ret;
 }
 
@@ -5038,12 +5013,13 @@ pfm_context_force_terminate(pfm_context_t *ctx, struct pt_regs *regs)
 }
 
 static int pfm_ovfl_notify_user(pfm_context_t *ctx, unsigned long ovfl_pmds);
+
  /*
   * pfm_handle_work() can be called with interrupts enabled
   * (TIF_NEED_RESCHED) or disabled. The down_interruptible
   * call may sleep, therefore we must re-enable interrupts
   * to avoid deadlocks. It is safe to do so because this function
-  * is called ONLY when returning to user level (PUStk=1), in which case
+  * is called ONLY when returning to user level (pUStk=1), in which case
   * there is no risk of kernel stack overflow due to deep
   * interrupt nesting.
   */
@@ -5059,7 +5035,8 @@ pfm_handle_work(void)
 
 	ctx = PFM_GET_CTX(current);
 	if (ctx == NULL) {
-		printk(KERN_ERR "perfmon: [%d] has no PFM context\n", task_pid_nr(current));
+		printk(KERN_ERR "perfmon: [%d] has no PFM context\n",
+			task_pid_nr(current));
 		return;
 	}
 
@@ -5083,11 +5060,12 @@ pfm_handle_work(void)
 	/*
 	 * must be done before we check for simple-reset mode
 	 */
-	if (ctx->ctx_fl_going_zombie || ctx->ctx_state == PFM_CTX_ZOMBIE) goto do_zombie;
-
+	if (ctx->ctx_fl_going_zombie || ctx->ctx_state == PFM_CTX_ZOMBIE)
+		goto do_zombie;
 
 	//if (CTX_OVFL_NOBLOCK(ctx)) goto skip_blocking;
-	if (reason == PFM_TRAP_REASON_RESET) goto skip_blocking;
+	if (reason == PFM_TRAP_REASON_RESET)
+		goto skip_blocking;
 
 	/*
 	 * restore interrupt mask to what it was on entry.
@@ -5135,7 +5113,8 @@ do_zombie:
 	/*
 	 * in case of interruption of down() we don't restart anything
 	 */
-	if (ret < 0) goto nothing_to_do;
+	if (ret < 0)
+		goto nothing_to_do;
 
 skip_blocking:
 	pfm_resume_after_ovfl(ctx, ovfl_regs, regs);
@@ -6695,16 +6674,12 @@ pfm_init(void)
 	/*
 	 * create /proc/perfmon (mostly for debugging purposes)
 	 */
- 	perfmon_dir = create_proc_entry("perfmon", S_IRUGO, NULL);
+	perfmon_dir = proc_create("perfmon", S_IRUGO, NULL, &pfm_proc_fops);
 	if (perfmon_dir == NULL) {
 		printk(KERN_ERR "perfmon: cannot create /proc entry, perfmon disabled\n");
 		pmu_conf = NULL;
 		return -1;
 	}
-  	/*
- 	 * install customized file operations for /proc/perfmon entry
- 	 */
- 	perfmon_dir->proc_fops = &pfm_proc_fops;
 
 	/*
 	 * create /proc/sys/kernel/perfmon (for debugging purposes)
diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c
index a5ea817cbcb..a3a34b4eb03 100644
--- a/arch/ia64/kernel/process.c
+++ b/arch/ia64/kernel/process.c
@@ -167,11 +167,18 @@ void tsk_clear_notify_resume(struct task_struct *tsk)
 	clear_ti_thread_flag(task_thread_info(tsk), TIF_NOTIFY_RESUME);
 }
 
+/*
+ * do_notify_resume_user():
+ *	Called from notify_resume_user at entry.S, with interrupts disabled.
+ */
 void
-do_notify_resume_user (sigset_t *unused, struct sigscratch *scr, long in_syscall)
+do_notify_resume_user(sigset_t *unused, struct sigscratch *scr, long in_syscall)
 {
 	if (fsys_mode(current, &scr->pt)) {
-		/* defer signal-handling etc. until we return to privilege-level 0.  */
+		/*
+		 * defer signal-handling etc. until we return to
+		 * privilege-level 0.
+		 */
 		if (!ia64_psr(&scr->pt)->lp)
 			ia64_psr(&scr->pt)->lp = 1;
 		return;
@@ -179,16 +186,26 @@ do_notify_resume_user (sigset_t *unused, struct sigscratch *scr, long in_syscall
 
 #ifdef CONFIG_PERFMON
 	if (current->thread.pfm_needs_checking)
+		/*
+		 * Note: pfm_handle_work() allow us to call it with interrupts
+		 * disabled, and may enable interrupts within the function.
+		 */
 		pfm_handle_work();
 #endif
 
 	/* deal with pending signal delivery */
-	if (test_thread_flag(TIF_SIGPENDING)||test_thread_flag(TIF_RESTORE_SIGMASK))
+	if (test_thread_flag(TIF_SIGPENDING)) {
+		local_irq_enable();	/* force interrupt enable */
 		ia64_do_signal(scr, in_syscall);
+	}
 
 	/* copy user rbs to kernel rbs */
-	if (unlikely(test_thread_flag(TIF_RESTORE_RSE)))
+	if (unlikely(test_thread_flag(TIF_RESTORE_RSE))) {
+		local_irq_enable();	/* force interrupt enable */
 		ia64_sync_krbs();
+	}
+
+	local_irq_disable();	/* force interrupt disable */
 }
 
 static int pal_halt        = 1;
diff --git a/arch/ia64/kernel/sal.c b/arch/ia64/kernel/sal.c
index a3022dc48ef..7e0259709c0 100644
--- a/arch/ia64/kernel/sal.c
+++ b/arch/ia64/kernel/sal.c
@@ -229,6 +229,14 @@ static void __init sal_desc_ap_wakeup(void *p) { }
  */
 static int sal_cache_flush_drops_interrupts;
 
+static int __init
+force_pal_cache_flush(char *str)
+{
+	sal_cache_flush_drops_interrupts = 1;
+	return 0;
+}
+early_param("force_pal_cache_flush", force_pal_cache_flush);
+
 void __init
 check_sal_cache_flush (void)
 {
@@ -237,6 +245,9 @@ check_sal_cache_flush (void)
 	u64 vector, cache_type = 3;
 	struct ia64_sal_retval isrv;
 
+	if (sal_cache_flush_drops_interrupts)
+		return;
+
 	cpu = get_cpu();
 	local_irq_save(flags);
 
diff --git a/arch/ia64/kernel/salinfo.c b/arch/ia64/kernel/salinfo.c
index b11bb50a197..ecb9eb78d68 100644
--- a/arch/ia64/kernel/salinfo.c
+++ b/arch/ia64/kernel/salinfo.c
@@ -648,18 +648,16 @@ salinfo_init(void)
 		if (!dir)
 			continue;
 
-		entry = create_proc_entry("event", S_IRUSR, dir);
+		entry = proc_create_data("event", S_IRUSR, dir,
+					 &salinfo_event_fops, data);
 		if (!entry)
 			continue;
-		entry->data = data;
-		entry->proc_fops = &salinfo_event_fops;
 		*sdir++ = entry;
 
-		entry = create_proc_entry("data", S_IRUSR | S_IWUSR, dir);
+		entry = proc_create_data("data", S_IRUSR | S_IWUSR, dir,
+					 &salinfo_data_fops, data);
 		if (!entry)
 			continue;
-		entry->data = data;
-		entry->proc_fops = &salinfo_data_fops;
 		*sdir++ = entry;
 
 		/* we missed any events before now */
diff --git a/arch/ia64/kernel/setup.c b/arch/ia64/kernel/setup.c
index 5015ca1275c..e9596cd0cda 100644
--- a/arch/ia64/kernel/setup.c
+++ b/arch/ia64/kernel/setup.c
@@ -239,6 +239,25 @@ __initcall(register_memory);
 
 
 #ifdef CONFIG_KEXEC
+
+/*
+ * This function checks if the reserved crashkernel is allowed on the specific
+ * IA64 machine flavour. Machines without an IO TLB use swiotlb and require
+ * some memory below 4 GB (i.e. in 32 bit area), see the implementation of
+ * lib/swiotlb.c. The hpzx1 architecture has an IO TLB but cannot use that
+ * in kdump case. See the comment in sba_init() in sba_iommu.c.
+ *
+ * So, the only machvec that really supports loading the kdump kernel
+ * over 4 GB is "sn2".
+ */
+static int __init check_crashkernel_memory(unsigned long pbase, size_t size)
+{
+	if (ia64_platform_is("sn2") || ia64_platform_is("uv"))
+		return 1;
+	else
+		return pbase < (1UL << 32);
+}
+
 static void __init setup_crashkernel(unsigned long total, int *n)
 {
 	unsigned long long base = 0, size = 0;
@@ -252,6 +271,16 @@ static void __init setup_crashkernel(unsigned long total, int *n)
 			base = kdump_find_rsvd_region(size,
 					rsvd_region, *n);
 		}
+
+		if (!check_crashkernel_memory(base, size)) {
+			pr_warning("crashkernel: There would be kdump memory "
+				"at %ld GB but this is unusable because it "
+				"must\nbe below 4 GB. Change the memory "
+				"configuration of the machine.\n",
+				(unsigned long)(base >> 30));
+			return;
+		}
+
 		if (base != ~0UL) {
 			printk(KERN_INFO "Reserving %ldMB of memory at %ldMB "
 					"for crashkernel (System RAM: %ldMB)\n",
diff --git a/arch/ia64/kernel/signal.c b/arch/ia64/kernel/signal.c
index 5740296c35a..19c5a78636f 100644
--- a/arch/ia64/kernel/signal.c
+++ b/arch/ia64/kernel/signal.c
@@ -464,7 +464,7 @@ ia64_do_signal (struct sigscratch *scr, long in_syscall)
 	if (!user_mode(&scr->pt))
 		return;
 
-	if (test_thread_flag(TIF_RESTORE_SIGMASK))
+	if (current_thread_info()->status & TS_RESTORE_SIGMASK)
 		oldset = &current->saved_sigmask;
 	else
 		oldset = &current->blocked;
@@ -530,12 +530,13 @@ ia64_do_signal (struct sigscratch *scr, long in_syscall)
 		 * continue to iterate in this loop so we can deliver the SIGSEGV...
 		 */
 		if (handle_signal(signr, &ka, &info, oldset, scr)) {
-			/* a signal was successfully delivered; the saved
+			/*
+			 * A signal was successfully delivered; the saved
 			 * sigmask will have been stored in the signal frame,
 			 * and will be restored by sigreturn, so we can simply
-			 * clear the TIF_RESTORE_SIGMASK flag */
-			if (test_thread_flag(TIF_RESTORE_SIGMASK))
-				clear_thread_flag(TIF_RESTORE_SIGMASK);
+			 * clear the TS_RESTORE_SIGMASK flag.
+			 */
+			current_thread_info()->status &= ~TS_RESTORE_SIGMASK;
 			return;
 		}
 	}
@@ -566,8 +567,8 @@ ia64_do_signal (struct sigscratch *scr, long in_syscall)
 
 	/* if there's no signal to deliver, we just put the saved sigmask
 	 * back */
-	if (test_thread_flag(TIF_RESTORE_SIGMASK)) {
-		clear_thread_flag(TIF_RESTORE_SIGMASK);
+	if (current_thread_info()->status & TS_RESTORE_SIGMASK) {
+		current_thread_info()->status &= ~TS_RESTORE_SIGMASK;
 		sigprocmask(SIG_SETMASK, &current->saved_sigmask, NULL);
 	}
 }
diff --git a/arch/ia64/kernel/smp.c b/arch/ia64/kernel/smp.c
index 9a9d4c48933..983296f1c81 100644
--- a/arch/ia64/kernel/smp.c
+++ b/arch/ia64/kernel/smp.c
@@ -98,8 +98,33 @@ unlock_ipi_calllock(void)
 	spin_unlock_irq(&call_lock);
 }
 
+static inline void
+handle_call_data(void)
+{
+	struct call_data_struct *data;
+	void (*func)(void *info);
+	void *info;
+	int wait;
+
+	/* release the 'pointer lock' */
+	data = (struct call_data_struct *)call_data;
+	func = data->func;
+	info = data->info;
+	wait = data->wait;
+
+	mb();
+	atomic_inc(&data->started);
+	/* At this point the structure may be gone unless wait is true. */
+	(*func)(info);
+
+	/* Notify the sending CPU that the task is done. */
+	mb();
+	if (wait)
+		atomic_inc(&data->finished);
+}
+
 static void
-stop_this_cpu (void)
+stop_this_cpu(void)
 {
 	/*
 	 * Remove this CPU:
@@ -138,44 +163,21 @@ handle_IPI (int irq, void *dev_id)
 			ops &= ~(1 << which);
 
 			switch (which) {
-			      case IPI_CALL_FUNC:
-			      {
-				      struct call_data_struct *data;
-				      void (*func)(void *info);
-				      void *info;
-				      int wait;
-
-				      /* release the 'pointer lock' */
-				      data = (struct call_data_struct *) call_data;
-				      func = data->func;
-				      info = data->info;
-				      wait = data->wait;
-
-				      mb();
-				      atomic_inc(&data->started);
-				      /*
-				       * At this point the structure may be gone unless
-				       * wait is true.
-				       */
-				      (*func)(info);
-
-				      /* Notify the sending CPU that the task is done.  */
-				      mb();
-				      if (wait)
-					      atomic_inc(&data->finished);
-			      }
-			      break;
-
-			      case IPI_CPU_STOP:
+			case IPI_CALL_FUNC:
+				handle_call_data();
+				break;
+
+			case IPI_CPU_STOP:
 				stop_this_cpu();
 				break;
 #ifdef CONFIG_KEXEC
-			      case IPI_KDUMP_CPU_STOP:
+			case IPI_KDUMP_CPU_STOP:
 				unw_init_running(kdump_cpu_freeze, NULL);
 				break;
 #endif
-			      default:
-				printk(KERN_CRIT "Unknown IPI on CPU %d: %lu\n", this_cpu, which);
+			default:
+				printk(KERN_CRIT "Unknown IPI on CPU %d: %lu\n",
+						this_cpu, which);
 				break;
 			}
 		} while (ops);
diff --git a/arch/ia64/kernel/smpboot.c b/arch/ia64/kernel/smpboot.c
index 16483be18c0..d7ad42b77d4 100644
--- a/arch/ia64/kernel/smpboot.c
+++ b/arch/ia64/kernel/smpboot.c
@@ -873,7 +873,8 @@ identify_siblings(struct cpuinfo_ia64 *c)
 	u16 pltid;
 	pal_logical_to_physical_t info;
 
-	if ((status = ia64_pal_logical_to_phys(-1, &info)) != PAL_STATUS_SUCCESS) {
+	status = ia64_pal_logical_to_phys(-1, &info);
+	if (status != PAL_STATUS_SUCCESS) {
 		if (status != PAL_STATUS_UNIMPLEMENTED) {
 			printk(KERN_ERR
 				"ia64_pal_logical_to_phys failed with %ld\n",
@@ -885,8 +886,13 @@ identify_siblings(struct cpuinfo_ia64 *c)
 		info.overview_cpp  = 1;
 		info.overview_tpc  = 1;
 	}
-	if ((status = ia64_sal_physical_id_info(&pltid)) != PAL_STATUS_SUCCESS) {
-		printk(KERN_ERR "ia64_sal_pltid failed with %ld\n", status);
+
+	status = ia64_sal_physical_id_info(&pltid);
+	if (status != PAL_STATUS_SUCCESS) {
+		if (status != PAL_STATUS_UNIMPLEMENTED)
+			printk(KERN_ERR
+				"ia64_sal_pltid failed with %ld\n",
+				status);
 		return;
 	}
 
diff --git a/arch/ia64/kernel/time.c b/arch/ia64/kernel/time.c
index 48e15a51782..8c73643f2d6 100644
--- a/arch/ia64/kernel/time.c
+++ b/arch/ia64/kernel/time.c
@@ -379,11 +379,6 @@ static struct irqaction timer_irqaction = {
 	.name =		"timer"
 };
 
-void __devinit ia64_disable_timer(void)
-{
-	ia64_set_itv(1 << 16);
-}
-
 void __init
 time_init (void)
 {
diff --git a/arch/ia64/kernel/topology.c b/arch/ia64/kernel/topology.c
index a2484fc1a06..26228e2d01a 100644
--- a/arch/ia64/kernel/topology.c
+++ b/arch/ia64/kernel/topology.c
@@ -27,9 +27,20 @@
 
 static struct ia64_cpu *sysfs_cpus;
 
-int arch_register_cpu(int num)
+void arch_fix_phys_package_id(int num, u32 slot)
 {
-#if defined (CONFIG_ACPI) && defined (CONFIG_HOTPLUG_CPU)
+#ifdef CONFIG_SMP
+	if (cpu_data(num)->socket_id == -1)
+		cpu_data(num)->socket_id = slot;
+#endif
+}
+EXPORT_SYMBOL_GPL(arch_fix_phys_package_id);
+
+
+#ifdef CONFIG_HOTPLUG_CPU
+int __ref arch_register_cpu(int num)
+{
+#ifdef CONFIG_ACPI
 	/*
 	 * If CPEI can be re-targetted or if this is not
 	 * CPEI target, then it is hotpluggable
@@ -38,19 +49,21 @@ int arch_register_cpu(int num)
 		sysfs_cpus[num].cpu.hotpluggable = 1;
 	map_cpu_to_node(num, node_cpuid[num].nid);
 #endif
-
 	return register_cpu(&sysfs_cpus[num].cpu, num);
 }
-
-#ifdef CONFIG_HOTPLUG_CPU
+EXPORT_SYMBOL(arch_register_cpu);
 
 void arch_unregister_cpu(int num)
 {
 	unregister_cpu(&sysfs_cpus[num].cpu);
 	unmap_cpu_from_node(num, cpu_to_node(num));
 }
-EXPORT_SYMBOL(arch_register_cpu);
 EXPORT_SYMBOL(arch_unregister_cpu);
+#else
+static int __init arch_register_cpu(int num)
+{
+	return register_cpu(&sysfs_cpus[num].cpu, num);
+}
 #endif /*CONFIG_HOTPLUG_CPU*/
 
 
diff --git a/arch/ia64/kernel/uncached.c b/arch/ia64/kernel/uncached.c
index 2a90c32024f..e77995a6e3e 100644
--- a/arch/ia64/kernel/uncached.c
+++ b/arch/ia64/kernel/uncached.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2001-2006 Silicon Graphics, Inc.  All rights reserved.
+ * Copyright (C) 2001-2008 Silicon Graphics, Inc.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms of version 2 of the GNU General Public License
@@ -177,12 +177,13 @@ failed:
  * uncached_alloc_page
  *
  * @starting_nid: node id of node to start with, or -1
+ * @n_pages: number of contiguous pages to allocate
  *
- * Allocate 1 uncached page. Allocates on the requested node. If no
- * uncached pages are available on the requested node, roundrobin starting
- * with the next higher node.
+ * Allocate the specified number of contiguous uncached pages on the
+ * the requested node. If not enough contiguous uncached pages are available
+ * on the requested node, roundrobin starting with the next higher node.
  */
-unsigned long uncached_alloc_page(int starting_nid)
+unsigned long uncached_alloc_page(int starting_nid, int n_pages)
 {
 	unsigned long uc_addr;
 	struct uncached_pool *uc_pool;
@@ -202,7 +203,8 @@ unsigned long uncached_alloc_page(int starting_nid)
 		if (uc_pool->pool == NULL)
 			continue;
 		do {
-			uc_addr = gen_pool_alloc(uc_pool->pool, PAGE_SIZE);
+			uc_addr = gen_pool_alloc(uc_pool->pool,
+						 n_pages * PAGE_SIZE);
 			if (uc_addr != 0)
 				return uc_addr;
 		} while (uncached_add_chunk(uc_pool, nid) == 0);
@@ -217,11 +219,12 @@ EXPORT_SYMBOL(uncached_alloc_page);
 /*
  * uncached_free_page
  *
- * @uc_addr: uncached address of page to free
+ * @uc_addr: uncached address of first page to free
+ * @n_pages: number of contiguous pages to free
  *
- * Free a single uncached page.
+ * Free the specified number of uncached pages.
  */
-void uncached_free_page(unsigned long uc_addr)
+void uncached_free_page(unsigned long uc_addr, int n_pages)
 {
 	int nid = paddr_to_nid(uc_addr - __IA64_UNCACHED_OFFSET);
 	struct gen_pool *pool = uncached_pools[nid].pool;
@@ -232,7 +235,7 @@ void uncached_free_page(unsigned long uc_addr)
 	if ((uc_addr & (0XFUL << 60)) != __IA64_UNCACHED_OFFSET)
 		panic("uncached_free_page invalid address %lx\n", uc_addr);
 
-	gen_pool_free(pool, uc_addr, PAGE_SIZE);
+	gen_pool_free(pool, uc_addr, n_pages * PAGE_SIZE);
 }
 EXPORT_SYMBOL(uncached_free_page);
 
diff --git a/arch/ia64/kvm/kvm-ia64.c b/arch/ia64/kvm/kvm-ia64.c
index 6df07324013..318b8110062 100644
--- a/arch/ia64/kvm/kvm-ia64.c
+++ b/arch/ia64/kvm/kvm-ia64.c
@@ -1,4 +1,3 @@
-
 /*
  * kvm_ia64.c: Basic KVM suppport On Itanium series processors
  *
@@ -431,7 +430,7 @@ int kvm_emulate_halt(struct kvm_vcpu *vcpu)
 	if (itc_diff < 0)
 		itc_diff = -itc_diff;
 
-	expires = div64_64(itc_diff, cyc_per_usec);
+	expires = div64_u64(itc_diff, cyc_per_usec);
 	kt = ktime_set(0, 1000 * expires);
 	vcpu->arch.ht_active = 1;
 	hrtimer_start(p_ht, kt, HRTIMER_MODE_ABS);
diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c
index fc6c6636ffd..200100ea761 100644
--- a/arch/ia64/mm/init.c
+++ b/arch/ia64/mm/init.c
@@ -719,3 +719,28 @@ out:
 EXPORT_SYMBOL_GPL(remove_memory);
 #endif /* CONFIG_MEMORY_HOTREMOVE */
 #endif
+
+/*
+ * Even when CONFIG_IA32_SUPPORT is not enabled it is
+ * useful to have the Linux/x86 domain registered to
+ * avoid an attempted module load when emulators call
+ * personality(PER_LINUX32). This saves several milliseconds
+ * on each such call.
+ */
+static struct exec_domain ia32_exec_domain;
+
+static int __init
+per_linux32_init(void)
+{
+	ia32_exec_domain.name = "Linux/x86";
+	ia32_exec_domain.handler = NULL;
+	ia32_exec_domain.pers_low = PER_LINUX32;
+	ia32_exec_domain.pers_high = PER_LINUX32;
+	ia32_exec_domain.signal_map = default_exec_domain.signal_map;
+	ia32_exec_domain.signal_invmap = default_exec_domain.signal_invmap;
+	register_exec_domain(&ia32_exec_domain);
+
+	return 0;
+}
+
+__initcall(per_linux32_init);
diff --git a/arch/ia64/mm/tlb.c b/arch/ia64/mm/tlb.c
index d52ec4e8340..8caf42471f0 100644
--- a/arch/ia64/mm/tlb.c
+++ b/arch/ia64/mm/tlb.c
@@ -168,7 +168,10 @@ setup_ptcg_sem(int max_purges, int nptcg_from)
 	static int firstcpu = 1;
 
 	if (toolatetochangeptcgsem) {
-		BUG_ON(max_purges < nptcg);
+		if (nptcg_from == NPTCG_FROM_PAL && max_purges == 0)
+			BUG_ON(1 < nptcg);
+		else
+			BUG_ON(max_purges < nptcg);
 		return;
 	}
 
diff --git a/arch/ia64/sn/kernel/sn2/sn2_smp.c b/arch/ia64/sn/kernel/sn2/sn2_smp.c
index dfc6bf1c7b4..49d3120415e 100644
--- a/arch/ia64/sn/kernel/sn2/sn2_smp.c
+++ b/arch/ia64/sn/kernel/sn2/sn2_smp.c
@@ -550,11 +550,12 @@ static int __init sn2_ptc_init(void)
 	if (!ia64_platform_is("sn2"))
 		return 0;
 
-	if (!(proc_sn2_ptc = create_proc_entry(PTC_BASENAME, 0444, NULL))) {
+	proc_sn2_ptc = proc_create(PTC_BASENAME, 0444,
+				   NULL, &proc_sn2_ptc_operations);
+	if (!&proc_sn2_ptc_operations) {
 		printk(KERN_ERR "unable to create %s proc entry", PTC_BASENAME);
 		return -EINVAL;
 	}
-	proc_sn2_ptc->proc_fops = &proc_sn2_ptc_operations;
 	spin_lock_init(&sn2_global_ptc_lock);
 	return 0;
 }
diff --git a/arch/ia64/sn/kernel/sn2/sn_proc_fs.c b/arch/ia64/sn/kernel/sn2/sn_proc_fs.c
index 62b3e9a496a..2526e5c783a 100644
--- a/arch/ia64/sn/kernel/sn2/sn_proc_fs.c
+++ b/arch/ia64/sn/kernel/sn2/sn_proc_fs.c
@@ -139,30 +139,21 @@ static const struct file_operations proc_sn_topo_fops = {
 void register_sn_procfs(void)
 {
 	static struct proc_dir_entry *sgi_proc_dir = NULL;
-	struct proc_dir_entry *pde;
 
 	BUG_ON(sgi_proc_dir != NULL);
 	if (!(sgi_proc_dir = proc_mkdir("sgi_sn", NULL)))
 		return;
 
-	pde = create_proc_entry("partition_id", 0444, sgi_proc_dir);
-	if (pde)
-		pde->proc_fops = &proc_partition_id_fops;
-	pde = create_proc_entry("system_serial_number", 0444, sgi_proc_dir);
-	if (pde)
-		pde->proc_fops = &proc_system_sn_fops;
-	pde = create_proc_entry("licenseID", 0444, sgi_proc_dir);
-	if (pde)
-		pde->proc_fops = &proc_license_id_fops;
-	pde = create_proc_entry("sn_force_interrupt", 0644, sgi_proc_dir);
-	if (pde)
-		pde->proc_fops = &proc_sn_force_intr_fops;
-	pde = create_proc_entry("coherence_id", 0444, sgi_proc_dir);
-	if (pde)
-		pde->proc_fops = &proc_coherence_id_fops;
-	pde = create_proc_entry("sn_topology", 0444, sgi_proc_dir);
-	if (pde)
-		pde->proc_fops = &proc_sn_topo_fops;
+	proc_create("partition_id", 0444, sgi_proc_dir,
+		    &proc_partition_id_fops);
+	proc_create("system_serial_number", 0444, sgi_proc_dir,
+		    &proc_system_sn_fops);
+	proc_create("licenseID", 0444, sgi_proc_dir, &proc_license_id_fops);
+	proc_create("sn_force_interrupt", 0644, sgi_proc_dir,
+		    &proc_sn_force_intr_fops);
+	proc_create("coherence_id", 0444, sgi_proc_dir,
+		    &proc_coherence_id_fops);
+	proc_create("sn_topology", 0444, sgi_proc_dir, &proc_sn_topo_fops);
 }
 
 #endif /* CONFIG_PROC_FS */
diff --git a/arch/ia64/sn/pci/pci_dma.c b/arch/ia64/sn/pci/pci_dma.c
index 18b94b792d5..52175af299a 100644
--- a/arch/ia64/sn/pci/pci_dma.c
+++ b/arch/ia64/sn/pci/pci_dma.c
@@ -10,6 +10,7 @@
  */
 
 #include <linux/module.h>
+#include <linux/dma-attrs.h>
 #include <asm/dma.h>
 #include <asm/sn/intr.h>
 #include <asm/sn/pcibus_provider_defs.h>
@@ -149,11 +150,12 @@ void sn_dma_free_coherent(struct device *dev, size_t size, void *cpu_addr,
 EXPORT_SYMBOL(sn_dma_free_coherent);
 
 /**
- * sn_dma_map_single - map a single page for DMA
+ * sn_dma_map_single_attrs - map a single page for DMA
  * @dev: device to map for
  * @cpu_addr: kernel virtual address of the region to map
  * @size: size of the region
  * @direction: DMA direction
+ * @attrs: optional dma attributes
  *
  * Map the region pointed to by @cpu_addr for DMA and return the
  * DMA address.
@@ -163,42 +165,59 @@ EXPORT_SYMBOL(sn_dma_free_coherent);
  * no way of saving the dmamap handle from the alloc to later free
  * (which is pretty much unacceptable).
  *
+ * mappings with the DMA_ATTR_WRITE_BARRIER get mapped with
+ * dma_map_consistent() so that writes force a flush of pending DMA.
+ * (See "SGI Altix Architecture Considerations for Linux Device Drivers",
+ * Document Number: 007-4763-001)
+ *
  * TODO: simplify our interface;
  *       figure out how to save dmamap handle so can use two step.
  */
-dma_addr_t sn_dma_map_single(struct device *dev, void *cpu_addr, size_t size,
-			     int direction)
+dma_addr_t sn_dma_map_single_attrs(struct device *dev, void *cpu_addr,
+				   size_t size, int direction,
+				   struct dma_attrs *attrs)
 {
 	dma_addr_t dma_addr;
 	unsigned long phys_addr;
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct sn_pcibus_provider *provider = SN_PCIDEV_BUSPROVIDER(pdev);
+	int dmabarr;
+
+	dmabarr = dma_get_attr(DMA_ATTR_WRITE_BARRIER, attrs);
 
 	BUG_ON(dev->bus != &pci_bus_type);
 
 	phys_addr = __pa(cpu_addr);
-	dma_addr = provider->dma_map(pdev, phys_addr, size, SN_DMA_ADDR_PHYS);
+	if (dmabarr)
+		dma_addr = provider->dma_map_consistent(pdev, phys_addr,
+							size, SN_DMA_ADDR_PHYS);
+	else
+		dma_addr = provider->dma_map(pdev, phys_addr, size,
+					     SN_DMA_ADDR_PHYS);
+
 	if (!dma_addr) {
 		printk(KERN_ERR "%s: out of ATEs\n", __func__);
 		return 0;
 	}
 	return dma_addr;
 }
-EXPORT_SYMBOL(sn_dma_map_single);
+EXPORT_SYMBOL(sn_dma_map_single_attrs);
 
 /**
- * sn_dma_unmap_single - unamp a DMA mapped page
+ * sn_dma_unmap_single_attrs - unamp a DMA mapped page
  * @dev: device to sync
  * @dma_addr: DMA address to sync
  * @size: size of region
  * @direction: DMA direction
+ * @attrs: optional dma attributes
  *
  * This routine is supposed to sync the DMA region specified
  * by @dma_handle into the coherence domain.  On SN, we're always cache
  * coherent, so we just need to free any ATEs associated with this mapping.
  */
-void sn_dma_unmap_single(struct device *dev, dma_addr_t dma_addr, size_t size,
-			 int direction)
+void sn_dma_unmap_single_attrs(struct device *dev, dma_addr_t dma_addr,
+			       size_t size, int direction,
+			       struct dma_attrs *attrs)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct sn_pcibus_provider *provider = SN_PCIDEV_BUSPROVIDER(pdev);
@@ -207,19 +226,21 @@ void sn_dma_unmap_single(struct device *dev, dma_addr_t dma_addr, size_t size,
 
 	provider->dma_unmap(pdev, dma_addr, direction);
 }
-EXPORT_SYMBOL(sn_dma_unmap_single);
+EXPORT_SYMBOL(sn_dma_unmap_single_attrs);
 
 /**
- * sn_dma_unmap_sg - unmap a DMA scatterlist
+ * sn_dma_unmap_sg_attrs - unmap a DMA scatterlist
  * @dev: device to unmap
  * @sg: scatterlist to unmap
  * @nhwentries: number of scatterlist entries
  * @direction: DMA direction
+ * @attrs: optional dma attributes
  *
  * Unmap a set of streaming mode DMA translations.
  */
-void sn_dma_unmap_sg(struct device *dev, struct scatterlist *sgl,
-		     int nhwentries, int direction)
+void sn_dma_unmap_sg_attrs(struct device *dev, struct scatterlist *sgl,
+			   int nhwentries, int direction,
+			   struct dma_attrs *attrs)
 {
 	int i;
 	struct pci_dev *pdev = to_pci_dev(dev);
@@ -234,25 +255,34 @@ void sn_dma_unmap_sg(struct device *dev, struct scatterlist *sgl,
 		sg->dma_length = 0;
 	}
 }
-EXPORT_SYMBOL(sn_dma_unmap_sg);
+EXPORT_SYMBOL(sn_dma_unmap_sg_attrs);
 
 /**
- * sn_dma_map_sg - map a scatterlist for DMA
+ * sn_dma_map_sg_attrs - map a scatterlist for DMA
  * @dev: device to map for
  * @sg: scatterlist to map
  * @nhwentries: number of entries
  * @direction: direction of the DMA transaction
+ * @attrs: optional dma attributes
+ *
+ * mappings with the DMA_ATTR_WRITE_BARRIER get mapped with
+ * dma_map_consistent() so that writes force a flush of pending DMA.
+ * (See "SGI Altix Architecture Considerations for Linux Device Drivers",
+ * Document Number: 007-4763-001)
  *
  * Maps each entry of @sg for DMA.
  */
-int sn_dma_map_sg(struct device *dev, struct scatterlist *sgl, int nhwentries,
-		  int direction)
+int sn_dma_map_sg_attrs(struct device *dev, struct scatterlist *sgl,
+			int nhwentries, int direction, struct dma_attrs *attrs)
 {
 	unsigned long phys_addr;
 	struct scatterlist *saved_sg = sgl, *sg;
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct sn_pcibus_provider *provider = SN_PCIDEV_BUSPROVIDER(pdev);
 	int i;
+	int dmabarr;
+
+	dmabarr = dma_get_attr(DMA_ATTR_WRITE_BARRIER, attrs);
 
 	BUG_ON(dev->bus != &pci_bus_type);
 
@@ -260,11 +290,19 @@ int sn_dma_map_sg(struct device *dev, struct scatterlist *sgl, int nhwentries,
 	 * Setup a DMA address for each entry in the scatterlist.
 	 */
 	for_each_sg(sgl, sg, nhwentries, i) {
+		dma_addr_t dma_addr;
 		phys_addr = SG_ENT_PHYS_ADDRESS(sg);
-		sg->dma_address = provider->dma_map(pdev,
-						    phys_addr, sg->length,
-						    SN_DMA_ADDR_PHYS);
+		if (dmabarr)
+			dma_addr = provider->dma_map_consistent(pdev,
+								phys_addr,
+								sg->length,
+								SN_DMA_ADDR_PHYS);
+		else
+			dma_addr = provider->dma_map(pdev, phys_addr,
+						     sg->length,
+						     SN_DMA_ADDR_PHYS);
 
+		sg->dma_address = dma_addr;
 		if (!sg->dma_address) {
 			printk(KERN_ERR "%s: out of ATEs\n", __func__);
 
@@ -272,7 +310,8 @@ int sn_dma_map_sg(struct device *dev, struct scatterlist *sgl, int nhwentries,
 			 * Free any successfully allocated entries.
 			 */
 			if (i > 0)
-				sn_dma_unmap_sg(dev, saved_sg, i, direction);
+				sn_dma_unmap_sg_attrs(dev, saved_sg, i,
+						      direction, attrs);
 			return 0;
 		}
 
@@ -281,7 +320,7 @@ int sn_dma_map_sg(struct device *dev, struct scatterlist *sgl, int nhwentries,
 
 	return nhwentries;
 }
-EXPORT_SYMBOL(sn_dma_map_sg);
+EXPORT_SYMBOL(sn_dma_map_sg_attrs);
 
 void sn_dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle,
 				size_t size, int direction)
diff --git a/arch/ia64/uv/Makefile b/arch/ia64/uv/Makefile
new file mode 100644
index 00000000000..aa9f91947c4
--- /dev/null
+++ b/arch/ia64/uv/Makefile
@@ -0,0 +1,12 @@
+# arch/ia64/uv/Makefile
+#
+# This file is subject to the terms and conditions of the GNU General Public
+# License.  See the file "COPYING" in the main directory of this archive
+# for more details.
+#
+# Copyright (C) 2008 Silicon Graphics, Inc.  All Rights Reserved.
+#
+# Makefile for the sn uv subplatform
+#
+
+obj-y += kernel/
diff --git a/arch/ia64/uv/kernel/Makefile b/arch/ia64/uv/kernel/Makefile
new file mode 100644
index 00000000000..8d92b4684d8
--- /dev/null
+++ b/arch/ia64/uv/kernel/Makefile
@@ -0,0 +1,13 @@
+# arch/ia64/uv/kernel/Makefile
+#
+# This file is subject to the terms and conditions of the GNU General Public
+# License.  See the file "COPYING" in the main directory of this archive
+# for more details.
+#
+# Copyright (C) 2008 Silicon Graphics, Inc.  All Rights Reserved.
+#
+
+EXTRA_CFLAGS += -Iarch/ia64/sn/include
+
+obj-y				+= setup.o
+obj-$(CONFIG_IA64_GENERIC)      += machvec.o
diff --git a/arch/ia64/uv/kernel/machvec.c b/arch/ia64/uv/kernel/machvec.c
new file mode 100644
index 00000000000..50737a9dca7
--- /dev/null
+++ b/arch/ia64/uv/kernel/machvec.c
@@ -0,0 +1,11 @@
+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (c) 2008 Silicon Graphics, Inc.  All Rights Reserved.
+ */
+
+#define MACHVEC_PLATFORM_NAME	uv
+#define MACHVEC_PLATFORM_HEADER	<asm/machvec_uv.h>
+#include <asm/machvec_init.h>
diff --git a/arch/ia64/uv/kernel/setup.c b/arch/ia64/uv/kernel/setup.c
new file mode 100644
index 00000000000..9aa743203c3
--- /dev/null
+++ b/arch/ia64/uv/kernel/setup.c
@@ -0,0 +1,98 @@
+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * SGI UV Core Functions
+ *
+ * Copyright (C) 2008 Silicon Graphics, Inc. All rights reserved.
+ */
+
+#include <linux/module.h>
+#include <linux/percpu.h>
+#include <asm/sn/simulator.h>
+#include <asm/uv/uv_mmrs.h>
+#include <asm/uv/uv_hub.h>
+
+DEFINE_PER_CPU(struct uv_hub_info_s, __uv_hub_info);
+EXPORT_PER_CPU_SYMBOL_GPL(__uv_hub_info);
+
+
+struct redir_addr {
+	unsigned long redirect;
+	unsigned long alias;
+};
+
+#define DEST_SHIFT UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_0_MMR_DEST_BASE_SHFT
+
+static __initdata struct redir_addr redir_addrs[] = {
+	{UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_0_MMR, UVH_SI_ALIAS0_OVERLAY_CONFIG},
+	{UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_1_MMR, UVH_SI_ALIAS1_OVERLAY_CONFIG},
+	{UVH_RH_GAM_ALIAS210_REDIRECT_CONFIG_2_MMR, UVH_SI_ALIAS2_OVERLAY_CONFIG},
+};
+
+static __init void get_lowmem_redirect(unsigned long *base, unsigned long *size)
+{
+	union uvh_si_alias0_overlay_config_u alias;
+	union uvh_rh_gam_alias210_redirect_config_2_mmr_u redirect;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(redir_addrs); i++) {
+		alias.v = uv_read_local_mmr(redir_addrs[i].alias);
+		if (alias.s.base == 0) {
+			*size = (1UL << alias.s.m_alias);
+			redirect.v = uv_read_local_mmr(redir_addrs[i].redirect);
+			*base = (unsigned long)redirect.s.dest_base << DEST_SHIFT;
+			return;
+		}
+	}
+	BUG();
+}
+
+void __init uv_setup(char **cmdline_p)
+{
+	union uvh_si_addr_map_config_u m_n_config;
+	union uvh_node_id_u node_id;
+	unsigned long gnode_upper;
+	int nid, cpu, m_val, n_val;
+	unsigned long mmr_base, lowmem_redir_base, lowmem_redir_size;
+
+	if (IS_MEDUSA()) {
+		lowmem_redir_base = 0;
+		lowmem_redir_size = 0;
+		node_id.v = 0;
+		m_n_config.s.m_skt = 37;
+		m_n_config.s.n_skt = 0;
+		mmr_base = 0;
+	} else {
+		get_lowmem_redirect(&lowmem_redir_base, &lowmem_redir_size);
+		node_id.v = uv_read_local_mmr(UVH_NODE_ID);
+		m_n_config.v = uv_read_local_mmr(UVH_SI_ADDR_MAP_CONFIG);
+		mmr_base =
+			uv_read_local_mmr(UVH_RH_GAM_MMR_OVERLAY_CONFIG_MMR) &
+				~UV_MMR_ENABLE;
+	}
+
+	m_val = m_n_config.s.m_skt;
+	n_val = m_n_config.s.n_skt;
+	printk(KERN_DEBUG "UV: global MMR base 0x%lx\n", mmr_base);
+
+	gnode_upper = (((unsigned long)node_id.s.node_id) &
+		       ~((1 << n_val) - 1)) << m_val;
+
+	for_each_present_cpu(cpu) {
+		nid = cpu_to_node(cpu);
+		uv_cpu_hub_info(cpu)->lowmem_remap_base = lowmem_redir_base;
+		uv_cpu_hub_info(cpu)->lowmem_remap_top =
+			lowmem_redir_base + lowmem_redir_size;
+		uv_cpu_hub_info(cpu)->m_val = m_val;
+		uv_cpu_hub_info(cpu)->n_val = m_val;
+		uv_cpu_hub_info(cpu)->pnode_mask = (1 << n_val) -1;
+		uv_cpu_hub_info(cpu)->gpa_mask = (1 << (m_val + n_val)) - 1;
+		uv_cpu_hub_info(cpu)->gnode_upper = gnode_upper;
+		uv_cpu_hub_info(cpu)->global_mmr_base = mmr_base;
+		uv_cpu_hub_info(cpu)->coherency_domain_number = 0;/* ZZZ */
+		printk(KERN_DEBUG "UV cpu %d, nid %d\n", cpu, nid);
+	}
+}
+