Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6

24 files changed, 505 insertions, 302 deletions

diff --git a/arch/sparc64/defconfig b/arch/sparc64/defconfig
index 250958d1e3cb..9d4bd2229493 100644
--- a/arch/sparc64/defconfig
+++ b/arch/sparc64/defconfig
@@ -1,7 +1,7 @@
 #
 # Automatically generated make config: don't edit
-# Linux kernel version: 2.6.25-rc1
-# Sun Feb 17 22:44:12 2008
+# Linux kernel version: 2.6.25-rc3
+# Wed Mar 26 04:33:35 2008
 #
 CONFIG_SPARC=y
 CONFIG_SPARC64=y
@@ -55,9 +55,11 @@ CONFIG_POSIX_MQUEUE=y
 # CONFIG_IKCONFIG is not set
 CONFIG_LOG_BUF_SHIFT=18
 # CONFIG_CGROUPS is not set
+CONFIG_GROUP_SCHED=y
 CONFIG_FAIR_GROUP_SCHED=y
-CONFIG_FAIR_USER_SCHED=y
-# CONFIG_FAIR_CGROUP_SCHED is not set
+CONFIG_RT_GROUP_SCHED=y
+CONFIG_USER_SCHED=y
+# CONFIG_CGROUP_SCHED is not set
 CONFIG_SYSFS_DEPRECATED=y
 CONFIG_RELAY=y
 CONFIG_NAMESPACES=y
@@ -482,6 +484,7 @@ CONFIG_SCSI_LOWLEVEL=y
 # CONFIG_SCSI_IPS is not set
 # CONFIG_SCSI_INITIO is not set
 # CONFIG_SCSI_INIA100 is not set
+# CONFIG_SCSI_MVSAS is not set
 # CONFIG_SCSI_STEX is not set
 # CONFIG_SCSI_SYM53C8XX_2 is not set
 # CONFIG_SCSI_QLOGIC_1280 is not set
@@ -810,6 +813,7 @@ CONFIG_HWMON=y
 # CONFIG_SENSORS_ADM1031 is not set
 # CONFIG_SENSORS_ADM9240 is not set
 # CONFIG_SENSORS_ADT7470 is not set
+# CONFIG_SENSORS_ADT7473 is not set
 # CONFIG_SENSORS_ATXP1 is not set
 # CONFIG_SENSORS_DS1621 is not set
 # CONFIG_SENSORS_I5K_AMB is not set
diff --git a/arch/sparc64/kernel/cpu.c b/arch/sparc64/kernel/cpu.c
index dd5d28e3d798..0097c08dc600 100644
--- a/arch/sparc64/kernel/cpu.c
+++ b/arch/sparc64/kernel/cpu.c
@@ -15,6 +15,8 @@
 #include <asm/spitfire.h>
 #include <asm/oplib.h>
 
+#include "entry.h"
+
 DEFINE_PER_CPU(cpuinfo_sparc, __cpu_data) = { 0 };
 
 struct cpu_iu_info {
@@ -65,8 +67,6 @@ static struct cpu_iu_info linux_sparc_chips[] = {
 char *sparc_cpu_type;
 char *sparc_fpu_type;
 
-unsigned int fsr_storage;
-
 static void __init sun4v_cpu_probe(void)
 {
 	switch (sun4v_chip_type) {
@@ -94,8 +94,10 @@ void __init cpu_probe(void)
 	unsigned long ver, fpu_vers, manuf, impl, fprs;
 	int i;
 	
-	if (tlb_type == hypervisor)
-		return sun4v_cpu_probe();
+	if (tlb_type == hypervisor) {
+		sun4v_cpu_probe();
+		return;
+	}
 
 	fprs = fprs_read();
 	fprs_write(FPRS_FEF);
diff --git a/arch/sparc64/kernel/ds.c b/arch/sparc64/kernel/ds.c
index bd76482077be..edb74f5a1186 100644
--- a/arch/sparc64/kernel/ds.c
+++ b/arch/sparc64/kernel/ds.c
@@ -972,8 +972,7 @@ static void process_ds_work(void)
 	LIST_HEAD(todo);
 
 	spin_lock_irqsave(&ds_lock, flags);
-	list_splice(&ds_work_list, &todo);
-	INIT_LIST_HEAD(&ds_work_list);
+	list_splice_init(&ds_work_list, &todo);
 	spin_unlock_irqrestore(&ds_lock, flags);
 
 	list_for_each_entry_safe(qp, tmp, &todo, list) {
diff --git a/arch/sparc64/kernel/entry.S b/arch/sparc64/kernel/entry.S
index 6be4d2d2904e..49eca4b1cf25 100644
--- a/arch/sparc64/kernel/entry.S
+++ b/arch/sparc64/kernel/entry.S
@@ -1705,6 +1705,36 @@ __flushw_user:
 2:	retl
 	 nop
 
+	/* Flush %fp and %i7 to the stack for all register
+	 * windows active inside of the cpu.  This allows
+	 * show_stack_trace() to avoid using an expensive
+	 * 'flushw'.
+	 */
+	.globl		stack_trace_flush
+	.type		stack_trace_flush,#function
+stack_trace_flush:
+	rdpr		%pstate, %o0
+	wrpr		%o0, PSTATE_IE, %pstate
+
+	rdpr		%cwp, %g1
+	rdpr		%canrestore, %g2
+	sub		%g1, 1, %g3
+
+1:	brz,pn		%g2, 2f
+	 sub		%g2, 1, %g2
+	wrpr		%g3, %cwp
+	stx		%fp, [%sp + STACK_BIAS + RW_V9_I6]
+	stx		%i7, [%sp + STACK_BIAS + RW_V9_I7]
+	ba,pt		%xcc, 1b
+	 sub		%g3, 1, %g3
+
+2:	wrpr		%g1, %cwp
+	wrpr		%o0, %pstate
+
+	retl
+	 nop
+	.size		stack_trace_flush,.-stack_trace_flush
+
 #ifdef CONFIG_SMP
 	.globl		hard_smp_processor_id
 hard_smp_processor_id:
diff --git a/arch/sparc64/kernel/entry.h b/arch/sparc64/kernel/entry.h
new file mode 100644
index 000000000000..4a91e9c6d31b
--- /dev/null
+++ b/arch/sparc64/kernel/entry.h
@@ -0,0 +1,196 @@
+#ifndef _ENTRY_H
+#define _ENTRY_H
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/init.h>
+
+extern char *sparc_cpu_type;
+extern char *sparc_fpu_type;
+
+extern void __init per_cpu_patch(void);
+extern void __init sun4v_patch(void);
+extern void __init boot_cpu_id_too_large(int cpu);
+extern unsigned int dcache_parity_tl1_occurred;
+extern unsigned int icache_parity_tl1_occurred;
+
+extern asmlinkage void update_perfctrs(void);
+extern asmlinkage void sparc_breakpoint(struct pt_regs *regs);
+extern void timer_interrupt(int irq, struct pt_regs *regs);
+
+extern void do_notify_resume(struct pt_regs *regs,
+			     unsigned long orig_i0,
+			     int restart_syscall,
+			     unsigned long thread_info_flags);
+
+extern asmlinkage void syscall_trace(struct pt_regs *regs,
+				     int syscall_exit_p);
+
+extern void bad_trap_tl1(struct pt_regs *regs, long lvl);
+
+extern void do_fpe_common(struct pt_regs *regs);
+extern void do_fpieee(struct pt_regs *regs);
+extern void do_fpother(struct pt_regs *regs);
+extern void do_tof(struct pt_regs *regs);
+extern void do_div0(struct pt_regs *regs);
+extern void do_illegal_instruction(struct pt_regs *regs);
+extern void mem_address_unaligned(struct pt_regs *regs,
+				  unsigned long sfar,
+				  unsigned long sfsr);
+extern void sun4v_do_mna(struct pt_regs *regs,
+			 unsigned long addr,
+			 unsigned long type_ctx);
+extern void do_privop(struct pt_regs *regs);
+extern void do_privact(struct pt_regs *regs);
+extern void do_cee(struct pt_regs *regs);
+extern void do_cee_tl1(struct pt_regs *regs);
+extern void do_dae_tl1(struct pt_regs *regs);
+extern void do_iae_tl1(struct pt_regs *regs);
+extern void do_div0_tl1(struct pt_regs *regs);
+extern void do_fpdis_tl1(struct pt_regs *regs);
+extern void do_fpieee_tl1(struct pt_regs *regs);
+extern void do_fpother_tl1(struct pt_regs *regs);
+extern void do_ill_tl1(struct pt_regs *regs);
+extern void do_irq_tl1(struct pt_regs *regs);
+extern void do_lddfmna_tl1(struct pt_regs *regs);
+extern void do_stdfmna_tl1(struct pt_regs *regs);
+extern void do_paw(struct pt_regs *regs);
+extern void do_paw_tl1(struct pt_regs *regs);
+extern void do_vaw(struct pt_regs *regs);
+extern void do_vaw_tl1(struct pt_regs *regs);
+extern void do_tof_tl1(struct pt_regs *regs);
+extern void do_getpsr(struct pt_regs *regs);
+
+extern void spitfire_insn_access_exception(struct pt_regs *regs,
+					   unsigned long sfsr,
+					   unsigned long sfar);
+extern void spitfire_insn_access_exception_tl1(struct pt_regs *regs,
+					       unsigned long sfsr,
+					       unsigned long sfar);
+extern void spitfire_data_access_exception(struct pt_regs *regs,
+					   unsigned long sfsr,
+					   unsigned long sfar);
+extern void spitfire_data_access_exception_tl1(struct pt_regs *regs,
+					       unsigned long sfsr,
+					       unsigned long sfar);
+extern void spitfire_access_error(struct pt_regs *regs,
+				  unsigned long status_encoded,
+				  unsigned long afar);
+
+extern void cheetah_fecc_handler(struct pt_regs *regs,
+				 unsigned long afsr,
+				 unsigned long afar);
+extern void cheetah_cee_handler(struct pt_regs *regs,
+				unsigned long afsr,
+				unsigned long afar);
+extern void cheetah_deferred_handler(struct pt_regs *regs,
+				     unsigned long afsr,
+				     unsigned long afar);
+extern void cheetah_plus_parity_error(int type, struct pt_regs *regs);
+
+extern void sun4v_insn_access_exception(struct pt_regs *regs,
+					unsigned long addr,
+					unsigned long type_ctx);
+extern void sun4v_insn_access_exception_tl1(struct pt_regs *regs,
+					    unsigned long addr,
+					    unsigned long type_ctx);
+extern void sun4v_data_access_exception(struct pt_regs *regs,
+					unsigned long addr,
+					unsigned long type_ctx);
+extern void sun4v_data_access_exception_tl1(struct pt_regs *regs,
+					    unsigned long addr,
+					    unsigned long type_ctx);
+extern void sun4v_resum_error(struct pt_regs *regs,
+			      unsigned long offset);
+extern void sun4v_resum_overflow(struct pt_regs *regs);
+extern void sun4v_nonresum_error(struct pt_regs *regs,
+				 unsigned long offset);
+extern void sun4v_nonresum_overflow(struct pt_regs *regs);
+
+extern unsigned long sun4v_err_itlb_vaddr;
+extern unsigned long sun4v_err_itlb_ctx;
+extern unsigned long sun4v_err_itlb_pte;
+extern unsigned long sun4v_err_itlb_error;
+
+extern void sun4v_itlb_error_report(struct pt_regs *regs, int tl);
+
+extern unsigned long sun4v_err_dtlb_vaddr;
+extern unsigned long sun4v_err_dtlb_ctx;
+extern unsigned long sun4v_err_dtlb_pte;
+extern unsigned long sun4v_err_dtlb_error;
+
+extern void sun4v_dtlb_error_report(struct pt_regs *regs, int tl);
+extern void hypervisor_tlbop_error(unsigned long err,
+				   unsigned long op);
+extern void hypervisor_tlbop_error_xcall(unsigned long err,
+					 unsigned long op);
+
+/* WARNING: The error trap handlers in assembly know the precise
+ *	    layout of the following structure.
+ *
+ * C-level handlers in traps.c use this information to log the
+ * error and then determine how to recover (if possible).
+ */
+struct cheetah_err_info {
+/*0x00*/u64 afsr;
+/*0x08*/u64 afar;
+
+	/* D-cache state */
+/*0x10*/u64 dcache_data[4];	/* The actual data	*/
+/*0x30*/u64 dcache_index;	/* D-cache index	*/
+/*0x38*/u64 dcache_tag;		/* D-cache tag/valid	*/
+/*0x40*/u64 dcache_utag;	/* D-cache microtag	*/
+/*0x48*/u64 dcache_stag;	/* D-cache snooptag	*/
+
+	/* I-cache state */
+/*0x50*/u64 icache_data[8];	/* The actual insns + predecode	*/
+/*0x90*/u64 icache_index;	/* I-cache index	*/
+/*0x98*/u64 icache_tag;		/* I-cache phys tag	*/
+/*0xa0*/u64 icache_utag;	/* I-cache microtag	*/
+/*0xa8*/u64 icache_stag;	/* I-cache snooptag	*/
+/*0xb0*/u64 icache_upper;	/* I-cache upper-tag	*/
+/*0xb8*/u64 icache_lower;	/* I-cache lower-tag	*/
+
+	/* E-cache state */
+/*0xc0*/u64 ecache_data[4];	/* 32 bytes from staging registers */
+/*0xe0*/u64 ecache_index;	/* E-cache index	*/
+/*0xe8*/u64 ecache_tag;		/* E-cache tag/state	*/
+
+/*0xf0*/u64 __pad[32 - 30];
+};
+#define CHAFSR_INVALID		((u64)-1L)
+
+/* This is allocated at boot time based upon the largest hardware
+ * cpu ID in the system.  We allocate two entries per cpu, one for
+ * TL==0 logging and one for TL >= 1 logging.
+ */
+extern struct cheetah_err_info *cheetah_error_log;
+
+/* UPA nodes send interrupt packet to UltraSparc with first data reg
+ * value low 5 (7 on Starfire) bits holding the IRQ identifier being
+ * delivered.  We must translate this into a non-vector IRQ so we can
+ * set the softint on this cpu.
+ *
+ * To make processing these packets efficient and race free we use
+ * an array of irq buckets below.  The interrupt vector handler in
+ * entry.S feeds incoming packets into per-cpu pil-indexed lists.
+ *
+ * If you make changes to ino_bucket, please update hand coded assembler
+ * of the vectored interrupt trap handler(s) in entry.S and sun4v_ivec.S
+ */
+struct ino_bucket {
+/*0x00*/unsigned long __irq_chain_pa;
+
+	/* Virtual interrupt number assigned to this INO.  */
+/*0x08*/unsigned int __virt_irq;
+/*0x0c*/unsigned int __pad;
+};
+
+extern struct ino_bucket *ivector_table;
+extern unsigned long ivector_table_pa;
+
+extern void handler_irq(int irq, struct pt_regs *regs);
+extern void init_irqwork_curcpu(void);
+extern void __cpuinit sun4v_register_mondo_queues(int this_cpu);
+
+#endif /* _ENTRY_H */
diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S
index 44b105c04dd3..34f8ff57c56b 100644
--- a/arch/sparc64/kernel/head.S
+++ b/arch/sparc64/kernel/head.S
@@ -288,8 +288,12 @@ sun4v_chip_type:
 	/* Leave arg2 as-is, prom_mmu_ihandle_cache */
 	mov	-1, %l3
 	stx	%l3, [%sp + 2047 + 128 + 0x28]	! arg3: mode (-1 default)
-	sethi	%hi(8 * 1024 * 1024), %l3
-	stx	%l3, [%sp + 2047 + 128 + 0x30]	! arg4: size (8MB)
+	/* 4MB align the kernel image size. */
+	set	(_end - KERNBASE), %l3
+	set	((4 * 1024 * 1024) - 1), %l4
+	add	%l3, %l4, %l3
+	andn	%l3, %l4, %l3
+	stx	%l3, [%sp + 2047 + 128 + 0x30]	! arg4: roundup(ksize, 4MB)
 	sethi	%hi(KERNBASE), %l3
 	stx	%l3, [%sp + 2047 + 128 + 0x38]	! arg5: vaddr (KERNBASE)
 	stx	%g0, [%sp + 2047 + 128 + 0x40]	! arg6: empty
diff --git a/arch/sparc64/kernel/iommu.c b/arch/sparc64/kernel/iommu.c
index fbaab3497bfd..756fa24eeefa 100644
--- a/arch/sparc64/kernel/iommu.c
+++ b/arch/sparc64/kernel/iommu.c
@@ -516,9 +516,11 @@ static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
 	unsigned long flags, handle, prot, ctx;
 	dma_addr_t dma_next = 0, dma_addr;
 	unsigned int max_seg_size;
+	unsigned long seg_boundary_size;
 	int outcount, incount, i;
 	struct strbuf *strbuf;
 	struct iommu *iommu;
+	unsigned long base_shift;
 
 	BUG_ON(direction == DMA_NONE);
 
@@ -549,8 +551,11 @@ static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
 	outs->dma_length = 0;
 
 	max_seg_size = dma_get_max_seg_size(dev);
+	seg_boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
+				  IO_PAGE_SIZE) >> IO_PAGE_SHIFT;
+	base_shift = iommu->page_table_map_base >> IO_PAGE_SHIFT;
 	for_each_sg(sglist, s, nelems, i) {
-		unsigned long paddr, npages, entry, slen;
+		unsigned long paddr, npages, entry, out_entry = 0, slen;
 		iopte_t *base;
 
 		slen = s->length;
@@ -593,7 +598,9 @@ static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
 			 * - allocated dma_addr isn't contiguous to previous allocation
 			 */
 			if ((dma_addr != dma_next) ||
-			    (outs->dma_length + s->length > max_seg_size)) {
+			    (outs->dma_length + s->length > max_seg_size) ||
+			    (is_span_boundary(out_entry, base_shift,
+					      seg_boundary_size, outs, s))) {
 				/* Can't merge: create a new segment */
 				segstart = s;
 				outcount++;
@@ -607,6 +614,7 @@ static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
 			/* This is a new segment, fill entries */
 			outs->dma_address = dma_addr;
 			outs->dma_length = slen;
+			out_entry = entry;
 		}
 
 		/* Calculate next page pointer for contiguous check */
@@ -626,7 +634,7 @@ static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
 iommu_map_failed:
 	for_each_sg(sglist, s, nelems, i) {
 		if (s->dma_length != 0) {
-			unsigned long vaddr, npages, entry, i;
+			unsigned long vaddr, npages, entry, j;
 			iopte_t *base;
 
 			vaddr = s->dma_address & IO_PAGE_MASK;
@@ -637,8 +645,8 @@ iommu_map_failed:
 				>> IO_PAGE_SHIFT;
 			base = iommu->page_table + entry;
 
-			for (i = 0; i < npages; i++)
-				iopte_make_dummy(iommu, base + i);
+			for (j = 0; j < npages; j++)
+				iopte_make_dummy(iommu, base + j);
 
 			s->dma_address = DMA_ERROR_CODE;
 			s->dma_length = 0;
@@ -803,7 +811,7 @@ static void dma_4u_sync_sg_for_cpu(struct device *dev,
 	spin_unlock_irqrestore(&iommu->lock, flags);
 }
 
-const struct dma_ops sun4u_dma_ops = {
+static const struct dma_ops sun4u_dma_ops = {
 	.alloc_coherent		= dma_4u_alloc_coherent,
 	.free_coherent		= dma_4u_free_coherent,
 	.map_single		= dma_4u_map_single,
diff --git a/arch/sparc64/kernel/iommu_common.h b/arch/sparc64/kernel/iommu_common.h
index 0713bd58499c..f3575a614fa2 100644
--- a/arch/sparc64/kernel/iommu_common.h
+++ b/arch/sparc64/kernel/iommu_common.h
@@ -12,6 +12,7 @@
 #include <linux/mm.h>
 #include <linux/scatterlist.h>
 #include <linux/device.h>
+#include <linux/iommu-helper.h>
 
 #include <asm/iommu.h>
 #include <asm/scatterlist.h>
@@ -45,17 +46,16 @@ static inline unsigned long iommu_num_pages(unsigned long vaddr,
 	return npages;
 }
 
-static inline unsigned long calc_npages(struct scatterlist *sglist, int nelems)
+static inline int is_span_boundary(unsigned long entry,
+				   unsigned long shift,
+				   unsigned long boundary_size,
+				   struct scatterlist *outs,
+				   struct scatterlist *sg)
 {
-	unsigned long i, npages = 0;
-	struct scatterlist *sg;
+	unsigned long paddr = SG_ENT_PHYS_ADDRESS(outs);
+	int nr = iommu_num_pages(paddr, outs->dma_length + sg->length);
 
-	for_each_sg(sglist, sg, nelems, i) {
-		unsigned long paddr = SG_ENT_PHYS_ADDRESS(sg);
-		npages += iommu_num_pages(paddr, sg->length);
-	}
-
-	return npages;
+	return iommu_is_span_boundary(entry, nr, shift, boundary_size);
 }
 
 extern unsigned long iommu_range_alloc(struct device *dev,
diff --git a/arch/sparc64/kernel/irq.c b/arch/sparc64/kernel/irq.c
index 5ec06c8c7fea..eb88bd6e674e 100644
--- a/arch/sparc64/kernel/irq.c
+++ b/arch/sparc64/kernel/irq.c
@@ -44,27 +44,10 @@
 #include <asm/hypervisor.h>
 #include <asm/cacheflush.h>
 
-/* UPA nodes send interrupt packet to UltraSparc with first data reg
- * value low 5 (7 on Starfire) bits holding the IRQ identifier being
- * delivered.  We must translate this into a non-vector IRQ so we can
- * set the softint on this cpu.
- *
- * To make processing these packets efficient and race free we use
- * an array of irq buckets below.  The interrupt vector handler in
- * entry.S feeds incoming packets into per-cpu pil-indexed lists.
- *
- * If you make changes to ino_bucket, please update hand coded assembler
- * of the vectored interrupt trap handler(s) in entry.S and sun4v_ivec.S
- */
-struct ino_bucket {
-/*0x00*/unsigned long __irq_chain_pa;
-
-	/* Virtual interrupt number assigned to this INO.  */
-/*0x08*/unsigned int __virt_irq;
-/*0x0c*/unsigned int __pad;
-};
+#include "entry.h"
 
 #define NUM_IVECS	(IMAP_INR + 1)
+
 struct ino_bucket *ivector_table;
 unsigned long ivector_table_pa;
 
diff --git a/arch/sparc64/kernel/pci_sun4v.c b/arch/sparc64/kernel/pci_sun4v.c
index ddca6c6c0b49..01839706bd52 100644
--- a/arch/sparc64/kernel/pci_sun4v.c
+++ b/arch/sparc64/kernel/pci_sun4v.c
@@ -335,8 +335,10 @@ static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
 	unsigned long flags, handle, prot;
 	dma_addr_t dma_next = 0, dma_addr;
 	unsigned int max_seg_size;
+	unsigned long seg_boundary_size;
 	int outcount, incount, i;
 	struct iommu *iommu;
+	unsigned long base_shift;
 	long err;
 
 	BUG_ON(direction == DMA_NONE);
@@ -362,8 +364,11 @@ static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
 	iommu_batch_start(dev, prot, ~0UL);
 
 	max_seg_size = dma_get_max_seg_size(dev);
+	seg_boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
+				  IO_PAGE_SIZE) >> IO_PAGE_SHIFT;
+	base_shift = iommu->page_table_map_base >> IO_PAGE_SHIFT;
 	for_each_sg(sglist, s, nelems, i) {
-		unsigned long paddr, npages, entry, slen;
+		unsigned long paddr, npages, entry, out_entry = 0, slen;
 
 		slen = s->length;
 		/* Sanity check */
@@ -406,7 +411,9 @@ static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
 			 * - allocated dma_addr isn't contiguous to previous allocation
 			 */
 			if ((dma_addr != dma_next) ||
-			    (outs->dma_length + s->length > max_seg_size)) {
+			    (outs->dma_length + s->length > max_seg_size) ||
+			    (is_span_boundary(out_entry, base_shift,
+					      seg_boundary_size, outs, s))) {
 				/* Can't merge: create a new segment */
 				segstart = s;
 				outcount++;
@@ -420,6 +427,7 @@ static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
 			/* This is a new segment, fill entries */
 			outs->dma_address = dma_addr;
 			outs->dma_length = slen;
+			out_entry = entry;
 		}
 
 		/* Calculate next page pointer for contiguous check */
diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c
index e116e38b160e..acf8c5250aa9 100644
--- a/arch/sparc64/kernel/process.c
+++ b/arch/sparc64/kernel/process.c
@@ -731,9 +731,6 @@ asmlinkage int sparc_execve(struct pt_regs *regs)
 		current_thread_info()->xfsr[0] = 0;
 		current_thread_info()->fpsaved[0] = 0;
 		regs->tstate &= ~TSTATE_PEF;
-		task_lock(current);
-		current->ptrace &= ~PT_DTRACE;
-		task_unlock(current);
 	}
 out:
 	return error;
diff --git a/arch/sparc64/kernel/ptrace.c b/arch/sparc64/kernel/ptrace.c
index 9a1ba1fe859d..aaae865e7932 100644
--- a/arch/sparc64/kernel/ptrace.c
+++ b/arch/sparc64/kernel/ptrace.c
@@ -35,6 +35,9 @@
 #include <asm/spitfire.h>
 #include <asm/page.h>
 #include <asm/cpudata.h>
+#include <asm/cacheflush.h>
+
+#include "entry.h"
 
 /* #define ALLOW_INIT_TRACING */
 
@@ -67,6 +70,8 @@ void flush_ptrace_access(struct vm_area_struct *vma, struct page *page,
 	if (tlb_type == hypervisor)
 		return;
 
+	preempt_disable();
+
 #ifdef DCACHE_ALIASING_POSSIBLE
 	/* If bit 13 of the kernel address we used to access the
 	 * user page is the same as the virtual address that page
@@ -105,6 +110,8 @@ void flush_ptrace_access(struct vm_area_struct *vma, struct page *page,
 		for (; start < end; start += icache_line_size)
 			flushi(start);
 	}
+
+	preempt_enable();
 }
 
 enum sparc_regset {
@@ -382,6 +389,7 @@ static const struct user_regset_view user_sparc64_view = {
 	.regsets = sparc64_regsets, .n = ARRAY_SIZE(sparc64_regsets)
 };
 
+#ifdef CONFIG_COMPAT
 static int genregs32_get(struct task_struct *target,
 			 const struct user_regset *regset,
 			 unsigned int pos, unsigned int count,
@@ -676,14 +684,18 @@ static const struct user_regset_view user_sparc32_view = {
 	.name = "sparc", .e_machine = EM_SPARC,
 	.regsets = sparc32_regsets, .n = ARRAY_SIZE(sparc32_regsets)
 };
+#endif /* CONFIG_COMPAT */
 
 const struct user_regset_view *task_user_regset_view(struct task_struct *task)
 {
+#ifdef CONFIG_COMPAT
 	if (test_tsk_thread_flag(task, TIF_32BIT))
 		return &user_sparc32_view;
+#endif
 	return &user_sparc64_view;
 }
 
+#ifdef CONFIG_COMPAT
 struct compat_fps {
 	unsigned int regs[32];
 	unsigned int fsr;
@@ -798,6 +810,7 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request,
 
 	return ret;
 }
+#endif /* CONFIG_COMPAT */
 
 struct fps {
 	unsigned int regs[64];
@@ -807,11 +820,14 @@ struct fps {
 long arch_ptrace(struct task_struct *child, long request, long addr, long data)
 {
 	const struct user_regset_view *view = task_user_regset_view(child);
-	struct pt_regs __user *pregs = (struct pt_regs __user *) addr;
 	unsigned long addr2 = task_pt_regs(current)->u_regs[UREG_I4];
-	struct fps __user *fps = (struct fps __user *) addr;
+	struct pt_regs __user *pregs;
+	struct fps __user *fps;
 	int ret;
 
+	pregs = (struct pt_regs __user *) (unsigned long) addr;
+	fps = (struct fps __user *) (unsigned long) addr;
+
 	switch (request) {
 	case PTRACE_PEEKUSR:
 		ret = (addr != 0) ? -EIO : 0;
diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c
index d036dbe72864..6acb4c51cfe4 100644
--- a/arch/sparc64/kernel/setup.c
+++ b/arch/sparc64/kernel/setup.c
@@ -51,6 +51,8 @@
 #include <net/ipconfig.h>
 #endif
 
+#include "entry.h"
+
 /* Used to synchronize accesses to NatSemi SUPER I/O chip configure
  * operations in asm/ns87303.h
  */
@@ -335,9 +337,6 @@ void __init setup_arch(char **cmdline_p)
 
 /* BUFFER is PAGE_SIZE bytes long. */
 
-extern char *sparc_cpu_type;
-extern char *sparc_fpu_type;
-
 extern void smp_info(struct seq_file *);
 extern void smp_bogo(struct seq_file *);
 extern void mmu_info(struct seq_file *);
diff --git a/arch/sparc64/kernel/signal.c b/arch/sparc64/kernel/signal.c
index fb13775b3682..94a9d64208ee 100644
--- a/arch/sparc64/kernel/signal.c
+++ b/arch/sparc64/kernel/signal.c
@@ -32,6 +32,9 @@
 #include <asm/siginfo.h>
 #include <asm/visasm.h>
 
+#include "entry.h"
+#include "systbls.h"
+
 #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
 
 /* {set, get}context() needed for 64-bit SparcLinux userland. */
diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c
index cc454731d879..59f020d69d4c 100644
--- a/arch/sparc64/kernel/smp.c
+++ b/arch/sparc64/kernel/smp.c
@@ -1,6 +1,6 @@
 /* smp.c: Sparc64 SMP support.
  *
- * Copyright (C) 1997, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1997, 2007, 2008 David S. Miller (davem@davemloft.net)
  */
 
 #include <linux/module.h>
@@ -30,6 +30,7 @@
 #include <asm/cpudata.h>
 #include <asm/hvtramp.h>
 #include <asm/io.h>
+#include <asm/timer.h>
 
 #include <asm/irq.h>
 #include <asm/irq_regs.h>
@@ -284,14 +285,17 @@ static void ldom_startcpu_cpuid(unsigned int cpu, unsigned long thread_reg)
 {
 	extern unsigned long sparc64_ttable_tl0;
 	extern unsigned long kern_locked_tte_data;
-	extern int bigkernel;
 	struct hvtramp_descr *hdesc;
 	unsigned long trampoline_ra;
 	struct trap_per_cpu *tb;
 	u64 tte_vaddr, tte_data;
 	unsigned long hv_err;
+	int i;
 
-	hdesc = kzalloc(sizeof(*hdesc), GFP_KERNEL);
+	hdesc = kzalloc(sizeof(*hdesc) +
+			(sizeof(struct hvtramp_mapping) *
+			 num_kernel_image_mappings - 1),
+			GFP_KERNEL);
 	if (!hdesc) {
 		printk(KERN_ERR "ldom_startcpu_cpuid: Cannot allocate "
 		       "hvtramp_descr.\n");
@@ -299,7 +303,7 @@ static void ldom_startcpu_cpuid(unsigned int cpu, unsigned long thread_reg)
 	}
 
 	hdesc->cpu = cpu;
-	hdesc->num_mappings = (bigkernel ? 2 : 1);
+	hdesc->num_mappings = num_kernel_image_mappings;
 
 	tb = &trap_block[cpu];
 	tb->hdesc = hdesc;
@@ -312,13 +316,11 @@ static void ldom_startcpu_cpuid(unsigned int cpu, unsigned long thread_reg)
 	tte_vaddr = (unsigned long) KERNBASE;
 	tte_data = kern_locked_tte_data;
 
-	hdesc->maps[0].vaddr = tte_vaddr;
-	hdesc->maps[0].tte   = tte_data;
-	if (bigkernel) {
+	for (i = 0; i < hdesc->num_mappings; i++) {
+		hdesc->maps[i].vaddr = tte_vaddr;
+		hdesc->maps[i].tte   = tte_data;
 		tte_vaddr += 0x400000;
 		tte_data  += 0x400000;
-		hdesc->maps[1].vaddr = tte_vaddr;
-		hdesc->maps[1].tte   = tte_data;
 	}
 
 	trampoline_ra = kimage_addr_to_ra(hv_cpu_startup);
diff --git a/arch/sparc64/kernel/stacktrace.c b/arch/sparc64/kernel/stacktrace.c
index 47f92a59be18..84d39e873e88 100644
--- a/arch/sparc64/kernel/stacktrace.c
+++ b/arch/sparc64/kernel/stacktrace.c
@@ -2,13 +2,15 @@
 #include <linux/stacktrace.h>
 #include <linux/thread_info.h>
 #include <asm/ptrace.h>
+#include <asm/stacktrace.h>
 
 void save_stack_trace(struct stack_trace *trace)
 {
 	unsigned long ksp, fp, thread_base;
 	struct thread_info *tp = task_thread_info(current);
 
-	flushw_all();
+	stack_trace_flush();
+
 	__asm__ __volatile__(
 		"mov	%%fp, %0"
 		: "=r" (ksp)
diff --git a/arch/sparc64/kernel/sys_sparc.c b/arch/sparc64/kernel/sys_sparc.c
index 134d801579f9..f952745d0f3d 100644
--- a/arch/sparc64/kernel/sys_sparc.c
+++ b/arch/sparc64/kernel/sys_sparc.c
@@ -1,5 +1,4 @@
-/* $Id: sys_sparc.c,v 1.57 2002/02/09 19:49:30 davem Exp $
- * linux/arch/sparc64/kernel/sys_sparc.c
+/* linux/arch/sparc64/kernel/sys_sparc.c
  *
  * This file contains various random system calls that
  * have a non-standard calling sequence on the Linux/sparc
@@ -30,6 +29,9 @@
 #include <asm/perfctr.h>
 #include <asm/unistd.h>
 
+#include "entry.h"
+#include "systbls.h"
+
 /* #define DEBUG_UNIMP_SYSCALL */
 
 asmlinkage unsigned long sys_getpagesize(void)
@@ -445,7 +447,8 @@ asmlinkage long sys_ipc(unsigned int call, int first, unsigned long second,
 			goto out;
 		case SEMTIMEDOP:
 			err = sys_semtimedop(first, ptr, (unsigned)second,
-				(const struct timespec __user *) fifth);
+				(const struct timespec __user *)
+					     (unsigned long) fifth);
 			goto out;
 		case SEMGET:
 			err = sys_semget(first, (int)second, (int)third);
@@ -788,7 +791,7 @@ asmlinkage long sys_utrap_install(utrap_entry_t type,
 	} else {
 		if ((utrap_handler_t)current_thread_info()->utraps[type] != new_p &&
 		    current_thread_info()->utraps[0] > 1) {
-			long *p = current_thread_info()->utraps;
+			unsigned long *p = current_thread_info()->utraps;
 
 			current_thread_info()->utraps =
 				kmalloc((UT_TRAP_INSTRUCTION_31+1)*sizeof(long),
@@ -816,7 +819,8 @@ asmlinkage long sys_utrap_install(utrap_entry_t type,
 	return 0;
 }
 
-long sparc_memory_ordering(unsigned long model, struct pt_regs *regs)
+asmlinkage long sparc_memory_ordering(unsigned long model,
+				      struct pt_regs *regs)
 {
 	if (model >= 3)
 		return -EINVAL;
diff --git a/arch/sparc64/kernel/sys_sparc32.c b/arch/sparc64/kernel/sys_sparc32.c
index deaba2bd0535..2455fa498876 100644
--- a/arch/sparc64/kernel/sys_sparc32.c
+++ b/arch/sparc64/kernel/sys_sparc32.c
@@ -678,9 +678,6 @@ asmlinkage long sparc32_execve(struct pt_regs *regs)
 		current_thread_info()->xfsr[0] = 0;
 		current_thread_info()->fpsaved[0] = 0;
 		regs->tstate &= ~TSTATE_PEF;
-		task_lock(current);
-		current->ptrace &= ~PT_DTRACE;
-		task_unlock(current);
 	}
 out:
 	return error;
diff --git a/arch/sparc64/kernel/systbls.h b/arch/sparc64/kernel/systbls.h
new file mode 100644
index 000000000000..8a0d20a35d0c
--- /dev/null
+++ b/arch/sparc64/kernel/systbls.h
@@ -0,0 +1,53 @@
+#ifndef _SYSTBLS_H
+#define _SYSTBLS_H
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/utsname.h>
+#include <asm/utrap.h>
+#include <asm/signal.h>
+
+extern asmlinkage unsigned long sys_getpagesize(void);
+extern asmlinkage unsigned long sparc_brk(unsigned long brk);
+extern asmlinkage long sparc_pipe(struct pt_regs *regs);
+extern asmlinkage long sys_ipc(unsigned int call, int first,
+			       unsigned long second,
+			       unsigned long third,
+			       void __user *ptr, long fifth);
+extern asmlinkage long sparc64_newuname(struct new_utsname __user *name);
+extern asmlinkage long sparc64_personality(unsigned long personality);
+extern asmlinkage unsigned long sys_mmap(unsigned long addr, unsigned long len,
+					 unsigned long prot, unsigned long flags,
+					 unsigned long fd, unsigned long off);
+extern asmlinkage long sys64_munmap(unsigned long addr, size_t len);
+extern asmlinkage unsigned long sys64_mremap(unsigned long addr,
+					     unsigned long old_len,
+					     unsigned long new_len,
+					     unsigned long flags,
+					     unsigned long new_addr);
+extern asmlinkage unsigned long c_sys_nis_syscall(struct pt_regs *regs);
+extern asmlinkage long sys_getdomainname(char __user *name, int len);
+extern asmlinkage long solaris_syscall(struct pt_regs *regs);
+extern asmlinkage long sunos_syscall(struct pt_regs *regs);
+extern asmlinkage long sys_utrap_install(utrap_entry_t type,
+					 utrap_handler_t new_p,
+					 utrap_handler_t new_d,
+					 utrap_handler_t __user *old_p,
+					 utrap_handler_t __user *old_d);
+extern asmlinkage long sparc_memory_ordering(unsigned long model,
+					     struct pt_regs *regs);
+extern asmlinkage long sys_rt_sigaction(int sig,
+					const struct sigaction __user *act,
+					struct sigaction __user *oact,
+					void __user *restorer,
+					size_t sigsetsize);
+extern asmlinkage long sys_perfctr(int opcode, unsigned long arg0,
+				   unsigned long arg1, unsigned long arg2);
+
+extern asmlinkage void sparc64_set_context(struct pt_regs *regs);
+extern asmlinkage void sparc64_get_context(struct pt_regs *regs);
+extern asmlinkage long sys_sigpause(unsigned int set);
+extern asmlinkage long sys_sigsuspend(old_sigset_t set);
+extern void do_rt_sigreturn(struct pt_regs *regs);
+
+#endif /* _SYSTBLS_H */
diff --git a/arch/sparc64/kernel/time.c b/arch/sparc64/kernel/time.c
index d204f1ab1d4c..e5d238970c7e 100644
--- a/arch/sparc64/kernel/time.c
+++ b/arch/sparc64/kernel/time.c
@@ -1,7 +1,6 @@
-/* $Id: time.c,v 1.42 2002/01/23 14:33:55 davem Exp $
- * time.c: UltraSparc timer and TOD clock support.
+/* time.c: UltraSparc timer and TOD clock support.
  *
- * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1997, 2008 David S. Miller (davem@davemloft.net)
  * Copyright (C) 1998 Eddie C. Dost   (ecd@skynet.be)
  *
  * Based largely on code which is:
@@ -48,6 +47,8 @@
 #include <asm/uaccess.h>
 #include <asm/irq_regs.h>
 
+#include "entry.h"
+
 DEFINE_SPINLOCK(mostek_lock);
 DEFINE_SPINLOCK(rtc_lock);
 void __iomem *mstk48t02_regs = NULL;
@@ -508,6 +509,37 @@ static int __init has_low_battery(void)
 	return (data1 == data2);	/* Was the write blocked? */
 }
 
+static void __init mostek_set_system_time(void __iomem *mregs)
+{
+	unsigned int year, mon, day, hour, min, sec;
+	u8 tmp;
+
+	spin_lock_irq(&mostek_lock);
+
+	/* Traditional Mostek chip. */
+	tmp = mostek_read(mregs + MOSTEK_CREG);
+	tmp |= MSTK_CREG_READ;
+	mostek_write(mregs + MOSTEK_CREG, tmp);
+
+	sec = MSTK_REG_SEC(mregs);
+	min = MSTK_REG_MIN(mregs);
+	hour = MSTK_REG_HOUR(mregs);
+	day = MSTK_REG_DOM(mregs);
+	mon = MSTK_REG_MONTH(mregs);
+	year = MSTK_CVT_YEAR( MSTK_REG_YEAR(mregs) );
+
+	xtime.tv_sec = mktime(year, mon, day, hour, min, sec);
+	xtime.tv_nsec = (INITIAL_JIFFIES % HZ) * (NSEC_PER_SEC / HZ);
+	set_normalized_timespec(&wall_to_monotonic,
+ 	                        -xtime.tv_sec, -xtime.tv_nsec);
+
+	tmp = mostek_read(mregs + MOSTEK_CREG);
+	tmp &= ~MSTK_CREG_READ;
+	mostek_write(mregs + MOSTEK_CREG, tmp);
+
+	spin_unlock_irq(&mostek_lock);
+}
+
 /* Probe for the real time clock chip. */
 static void __init set_system_time(void)
 {
@@ -520,7 +552,6 @@ static void __init set_system_time(void)
 	unsigned long dregs = 0UL;
 	void __iomem *bregs = 0UL;
 #endif
-	u8 tmp;
 
 	if (!mregs && !dregs && !bregs) {
 		prom_printf("Something wrong, clock regs not mapped yet.\n");
@@ -528,20 +559,11 @@ static void __init set_system_time(void)
 	}		
 
 	if (mregs) {
-		spin_lock_irq(&mostek_lock);
-
-		/* Traditional Mostek chip. */
-		tmp = mostek_read(mregs + MOSTEK_CREG);
-		tmp |= MSTK_CREG_READ;
-		mostek_write(mregs + MOSTEK_CREG, tmp);
+		mostek_set_system_time(mregs);
+		return;
+	}
 
-		sec = MSTK_REG_SEC(mregs);
-		min = MSTK_REG_MIN(mregs);
-		hour = MSTK_REG_HOUR(mregs);
-		day = MSTK_REG_DOM(mregs);
-		mon = MSTK_REG_MONTH(mregs);
-		year = MSTK_CVT_YEAR( MSTK_REG_YEAR(mregs) );
-	} else if (bregs) {
+	if (bregs) {
 		unsigned char val = readb(bregs + 0x0e);
 		unsigned int century;
 
@@ -596,14 +618,6 @@ static void __init set_system_time(void)
 	xtime.tv_nsec = (INITIAL_JIFFIES % HZ) * (NSEC_PER_SEC / HZ);
 	set_normalized_timespec(&wall_to_monotonic,
  	                        -xtime.tv_sec, -xtime.tv_nsec);
-
-	if (mregs) {
-		tmp = mostek_read(mregs + MOSTEK_CREG);
-		tmp &= ~MSTK_CREG_READ;
-		mostek_write(mregs + MOSTEK_CREG, tmp);
-
-		spin_unlock_irq(&mostek_lock);
-	}
 }
 
 /* davem suggests we keep this within the 4M locked kernel image */
@@ -1027,7 +1041,7 @@ void __init time_init(void)
 	setup_clockevent_multiplier(clock);
 
 	sparc64_clockevent.max_delta_ns =
-		clockevent_delta2ns(0x7fffffffffffffff, &sparc64_clockevent);
+		clockevent_delta2ns(0x7fffffffffffffffUL, &sparc64_clockevent);
 	sparc64_clockevent.min_delta_ns =
 		clockevent_delta2ns(0xF, &sparc64_clockevent);
 
diff --git a/arch/sparc64/kernel/trampoline.S b/arch/sparc64/kernel/trampoline.S
index 4ae2e525d68b..56ff55211341 100644
--- a/arch/sparc64/kernel/trampoline.S
+++ b/arch/sparc64/kernel/trampoline.S
@@ -105,7 +105,7 @@ startup_continue:
 	wr		%g2, 0, %tick_cmpr
 
 	/* Call OBP by hand to lock KERNBASE into i/d tlbs.
-	 * We lock 2 consequetive entries if we are 'bigkernel'.
+	 * We lock 'num_kernel_image_mappings' consequetive entries.
 	 */
 	sethi		%hi(prom_entry_lock), %g2
 1:	ldstub		[%g2 + %lo(prom_entry_lock)], %g1
@@ -119,6 +119,29 @@ startup_continue:
 	add		%l2, -(192 + 128), %sp
 	flushw
 
+	/* Setup the loop variables:
+	 * %l3: VADDR base
+	 * %l4: TTE base
+	 * %l5: Loop iterator, iterates from 0 to 'num_kernel_image_mappings'
+	 * %l6: Number of TTE entries to map
+	 * %l7: Highest TTE entry number, we count down
+	 */
+	sethi		%hi(KERNBASE), %l3
+	sethi		%hi(kern_locked_tte_data), %l4
+	ldx		[%l4 + %lo(kern_locked_tte_data)], %l4
+	clr		%l5
+	sethi		%hi(num_kernel_image_mappings), %l6
+	lduw		[%l6 + %lo(num_kernel_image_mappings)], %l6
+	add		%l6, 1, %l6
+
+	mov		15, %l7
+	BRANCH_IF_ANY_CHEETAH(g1,g5,2f)
+
+	mov		63, %l7
+2:
+
+3:
+	/* Lock into I-MMU */
 	sethi		%hi(call_method), %g2
 	or		%g2, %lo(call_method), %g2
 	stx		%g2, [%sp + 2047 + 128 + 0x00]
@@ -132,63 +155,26 @@ startup_continue:
 	sethi		%hi(prom_mmu_ihandle_cache), %g2
 	lduw		[%g2 + %lo(prom_mmu_ihandle_cache)], %g2
 	stx		%g2, [%sp + 2047 + 128 + 0x20]
-	sethi		%hi(KERNBASE), %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x28]
-	sethi		%hi(kern_locked_tte_data), %g2
-	ldx		[%g2 + %lo(kern_locked_tte_data)], %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x30]
-
-	mov		15, %g2
-	BRANCH_IF_ANY_CHEETAH(g1,g5,1f)
 
-	mov		63, %g2
-1:
-	stx		%g2, [%sp + 2047 + 128 + 0x38]
-	sethi		%hi(p1275buf), %g2
-	or		%g2, %lo(p1275buf), %g2
-	ldx		[%g2 + 0x08], %o1
-	call		%o1
-	 add		%sp, (2047 + 128), %o0
+	/* Each TTE maps 4MB, convert index to offset.  */
+	sllx		%l5, 22, %g1
 
-	sethi		%hi(bigkernel), %g2
-	lduw		[%g2 + %lo(bigkernel)], %g2
-	brz,pt		%g2, do_dtlb
-	 nop
+	add		%l3, %g1, %g2
+	stx		%g2, [%sp + 2047 + 128 + 0x28]	! VADDR
+	add		%l4, %g1, %g2
+	stx		%g2, [%sp + 2047 + 128 + 0x30]	! TTE
 
-	sethi		%hi(call_method), %g2
-	or		%g2, %lo(call_method), %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x00]
-	mov		5, %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x08]
-	mov		1, %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x10]
-	sethi		%hi(itlb_load), %g2
-	or		%g2, %lo(itlb_load), %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x18]
-	sethi		%hi(prom_mmu_ihandle_cache), %g2
-	lduw		[%g2 + %lo(prom_mmu_ihandle_cache)], %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x20]
-	sethi		%hi(KERNBASE + 0x400000), %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x28]
-	sethi		%hi(kern_locked_tte_data), %g2
-	ldx		[%g2 + %lo(kern_locked_tte_data)], %g2
-	sethi		%hi(0x400000), %g1
-	add		%g2, %g1, %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x30]
-
-	mov		14, %g2
-	BRANCH_IF_ANY_CHEETAH(g1,g5,1f)
-
-	mov		62, %g2
-1:
+	/* TTE index is highest minus loop index.  */
+	sub		%l7, %l5, %g2
 	stx		%g2, [%sp + 2047 + 128 + 0x38]
+
 	sethi		%hi(p1275buf), %g2
 	or		%g2, %lo(p1275buf), %g2
 	ldx		[%g2 + 0x08], %o1
 	call		%o1
 	 add		%sp, (2047 + 128), %o0
 
-do_dtlb:
+	/* Lock into D-MMU */
 	sethi		%hi(call_method), %g2
 	or		%g2, %lo(call_method), %g2
 	stx		%g2, [%sp + 2047 + 128 + 0x00]
@@ -202,65 +188,30 @@ do_dtlb:
 	sethi		%hi(prom_mmu_ihandle_cache), %g2
 	lduw		[%g2 + %lo(prom_mmu_ihandle_cache)], %g2
 	stx		%g2, [%sp + 2047 + 128 + 0x20]
-	sethi		%hi(KERNBASE), %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x28]
-	sethi		%hi(kern_locked_tte_data), %g2
-	ldx		[%g2 + %lo(kern_locked_tte_data)], %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x30]
 
-	mov		15, %g2
-	BRANCH_IF_ANY_CHEETAH(g1,g5,1f)
+	/* Each TTE maps 4MB, convert index to offset.  */
+	sllx		%l5, 22, %g1
 
-	mov		63, %g2
-1:
+	add		%l3, %g1, %g2
+	stx		%g2, [%sp + 2047 + 128 + 0x28]	! VADDR
+	add		%l4, %g1, %g2
+	stx		%g2, [%sp + 2047 + 128 + 0x30]	! TTE
 
+	/* TTE index is highest minus loop index.  */
+	sub		%l7, %l5, %g2
 	stx		%g2, [%sp + 2047 + 128 + 0x38]
+
 	sethi		%hi(p1275buf), %g2
 	or		%g2, %lo(p1275buf), %g2
 	ldx		[%g2 + 0x08], %o1
 	call		%o1
 	 add		%sp, (2047 + 128), %o0
 
-	sethi		%hi(bigkernel), %g2
-	lduw		[%g2 + %lo(bigkernel)], %g2
-	brz,pt		%g2, do_unlock
+	add		%l5, 1, %l5
+	cmp		%l5, %l6
+	bne,pt		%xcc, 3b
 	 nop
 
-	sethi		%hi(call_method), %g2
-	or		%g2, %lo(call_method), %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x00]
-	mov		5, %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x08]
-	mov		1, %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x10]
-	sethi		%hi(dtlb_load), %g2
-	or		%g2, %lo(dtlb_load), %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x18]
-	sethi		%hi(prom_mmu_ihandle_cache), %g2
-	lduw		[%g2 + %lo(prom_mmu_ihandle_cache)], %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x20]
-	sethi		%hi(KERNBASE + 0x400000), %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x28]
-	sethi		%hi(kern_locked_tte_data), %g2
-	ldx		[%g2 + %lo(kern_locked_tte_data)], %g2
-	sethi		%hi(0x400000), %g1
-	add		%g2, %g1, %g2
-	stx		%g2, [%sp + 2047 + 128 + 0x30]
-
-	mov		14, %g2
-	BRANCH_IF_ANY_CHEETAH(g1,g5,1f)
-
-	mov		62, %g2
-1:
-
-	stx		%g2, [%sp + 2047 + 128 + 0x38]
-	sethi		%hi(p1275buf), %g2
-	or		%g2, %lo(p1275buf), %g2
-	ldx		[%g2 + 0x08], %o1
-	call		%o1
-	 add		%sp, (2047 + 128), %o0
-
-do_unlock:
 	sethi		%hi(prom_entry_lock), %g2
 	stb		%g0, [%g2 + %lo(prom_entry_lock)]
 	membar		#StoreStore | #StoreLoad
@@ -269,47 +220,36 @@ do_unlock:
 	 nop
 
 niagara_lock_tlb:
+	sethi		%hi(KERNBASE), %l3
+	sethi		%hi(kern_locked_tte_data), %l4
+	ldx		[%l4 + %lo(kern_locked_tte_data)], %l4
+	clr		%l5
+	sethi		%hi(num_kernel_image_mappings), %l6
+	lduw		[%l6 + %lo(num_kernel_image_mappings)], %l6
+	add		%l6, 1, %l6
+
+1:
 	mov		HV_FAST_MMU_MAP_PERM_ADDR, %o5
-	sethi		%hi(KERNBASE), %o0
+	sllx		%l5, 22, %g2
+	add		%l3, %g2, %o0
 	clr		%o1
-	sethi		%hi(kern_locked_tte_data), %o2
-	ldx		[%o2 + %lo(kern_locked_tte_data)], %o2
+	add		%l4, %g2, %o2
 	mov		HV_MMU_IMMU, %o3
 	ta		HV_FAST_TRAP
 
 	mov		HV_FAST_MMU_MAP_PERM_ADDR, %o5
-	sethi		%hi(KERNBASE), %o0
+	sllx		%l5, 22, %g2
+	add		%l3, %g2, %o0
 	clr		%o1
-	sethi		%hi(kern_locked_tte_data), %o2
-	ldx		[%o2 + %lo(kern_locked_tte_data)], %o2
+	add		%l4, %g2, %o2
 	mov		HV_MMU_DMMU, %o3
 	ta		HV_FAST_TRAP
 
-	sethi		%hi(bigkernel), %g2
-	lduw		[%g2 + %lo(bigkernel)], %g2
-	brz,pt		%g2, after_lock_tlb
+	add		%l5, 1, %l5
+	cmp		%l5, %l6
+	bne,pt		%xcc, 1b
 	 nop
 
-	mov		HV_FAST_MMU_MAP_PERM_ADDR, %o5
-	sethi		%hi(KERNBASE + 0x400000), %o0
-	clr		%o1
-	sethi		%hi(kern_locked_tte_data), %o2
-	ldx		[%o2 + %lo(kern_locked_tte_data)], %o2
-	sethi		%hi(0x400000), %o3
-	add		%o2, %o3, %o2
-	mov		HV_MMU_IMMU, %o3
-	ta		HV_FAST_TRAP
-
-	mov		HV_FAST_MMU_MAP_PERM_ADDR, %o5
-	sethi		%hi(KERNBASE + 0x400000), %o0
-	clr		%o1
-	sethi		%hi(kern_locked_tte_data), %o2
-	ldx		[%o2 + %lo(kern_locked_tte_data)], %o2
-	sethi		%hi(0x400000), %o3
-	add		%o2, %o3, %o2
-	mov		HV_MMU_DMMU, %o3
-	ta		HV_FAST_TRAP
-
 after_lock_tlb:
 	wrpr		%g0, (PSTATE_PRIV | PSTATE_PEF), %pstate
 	wr		%g0, 0, %fprs
diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c
index 007f5317c0de..96da847023f3 100644
--- a/arch/sparc64/kernel/traps.c
+++ b/arch/sparc64/kernel/traps.c
@@ -42,6 +42,7 @@
 #endif
 #include <asm/prom.h>
 
+#include "entry.h"
 
 /* When an irrecoverable trap occurs at tl > 0, the trap entry
  * code logs the trap state registers at every level in the trap
@@ -77,11 +78,6 @@ static void dump_tl1_traplog(struct tl1_traplog *p)
 	}
 }
 
-void do_call_debug(struct pt_regs *regs) 
-{ 
-	notify_die(DIE_CALL, "debug call", regs, 0, 255, SIGINT); 
-}
-
 void bad_trap(struct pt_regs *regs, long lvl)
 {
 	char buffer[32];
@@ -550,41 +546,6 @@ static unsigned long ecache_flush_physbase;
 static unsigned long ecache_flush_linesize;
 static unsigned long ecache_flush_size;
 
-/* WARNING: The error trap handlers in assembly know the precise
- *	    layout of the following structure.
- *
- * C-level handlers below use this information to log the error
- * and then determine how to recover (if possible).
- */
-struct cheetah_err_info {
-/*0x00*/u64 afsr;
-/*0x08*/u64 afar;
-
-	/* D-cache state */
-/*0x10*/u64 dcache_data[4];	/* The actual data	*/
-/*0x30*/u64 dcache_index;	/* D-cache index	*/
-/*0x38*/u64 dcache_tag;		/* D-cache tag/valid	*/
-/*0x40*/u64 dcache_utag;	/* D-cache microtag	*/
-/*0x48*/u64 dcache_stag;	/* D-cache snooptag	*/
-
-	/* I-cache state */
-/*0x50*/u64 icache_data[8];	/* The actual insns + predecode	*/
-/*0x90*/u64 icache_index;	/* I-cache index	*/
-/*0x98*/u64 icache_tag;		/* I-cache phys tag	*/
-/*0xa0*/u64 icache_utag;	/* I-cache microtag	*/
-/*0xa8*/u64 icache_stag;	/* I-cache snooptag	*/
-/*0xb0*/u64 icache_upper;	/* I-cache upper-tag	*/
-/*0xb8*/u64 icache_lower;	/* I-cache lower-tag	*/
-
-	/* E-cache state */
-/*0xc0*/u64 ecache_data[4];	/* 32 bytes from staging registers */
-/*0xe0*/u64 ecache_index;	/* E-cache index	*/
-/*0xe8*/u64 ecache_tag;		/* E-cache tag/state	*/
-
-/*0xf0*/u64 __pad[32 - 30];
-};
-#define CHAFSR_INVALID		((u64)-1L)
-
 /* This table is ordered in priority of errors and matches the
  * AFAR overwrite policy as well.
  */
@@ -758,10 +719,6 @@ static struct afsr_error_table __jalapeno_error_table[] = {
 static struct afsr_error_table *cheetah_error_table;
 static unsigned long cheetah_afsr_errors;
 
-/* This is allocated at boot time based upon the largest hardware
- * cpu ID in the system.  We allocate two entries per cpu, one for
- * TL==0 logging and one for TL >= 1 logging.
- */
 struct cheetah_err_info *cheetah_error_log;
 
 static inline struct cheetah_err_info *cheetah_get_error_log(unsigned long afsr)
@@ -2102,7 +2059,7 @@ void do_div0(struct pt_regs *regs)
 	force_sig_info(SIGFPE, &info, current);
 }
 
-void instruction_dump (unsigned int *pc)
+static void instruction_dump(unsigned int *pc)
 {
 	int i;
 
@@ -2115,7 +2072,7 @@ void instruction_dump (unsigned int *pc)
 	printk("\n");
 }
 
-static void user_instruction_dump (unsigned int __user *pc)
+static void user_instruction_dump(unsigned int __user *pc)
 {
 	int i;
 	unsigned int buf[9];
diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c
index b5c30416fdac..f37078d96407 100644
--- a/arch/sparc64/mm/init.c
+++ b/arch/sparc64/mm/init.c
@@ -46,6 +46,7 @@
 #include <asm/prom.h>
 #include <asm/sstate.h>
 #include <asm/mdesc.h>
+#include <asm/cpudata.h>
 
 #define MAX_PHYS_ADDRESS	(1UL << 42UL)
 #define KPTE_BITMAP_CHUNK_SZ	(256UL * 1024UL * 1024UL)
@@ -166,7 +167,7 @@ unsigned long sparc64_kern_pri_context __read_mostly;
 unsigned long sparc64_kern_pri_nuc_bits __read_mostly;
 unsigned long sparc64_kern_sec_context __read_mostly;
 
-int bigkernel = 0;
+int num_kernel_image_mappings;
 
 #ifdef CONFIG_DEBUG_DCFLUSH
 atomic_t dcpage_flushes = ATOMIC_INIT(0);
@@ -572,7 +573,7 @@ static unsigned long kern_large_tte(unsigned long paddr);
 static void __init remap_kernel(void)
 {
 	unsigned long phys_page, tte_vaddr, tte_data;
-	int tlb_ent = sparc64_highest_locked_tlbent();
+	int i, tlb_ent = sparc64_highest_locked_tlbent();
 
 	tte_vaddr = (unsigned long) KERNBASE;
 	phys_page = (prom_boot_mapping_phys_low >> 22UL) << 22UL;
@@ -582,27 +583,20 @@ static void __init remap_kernel(void)
 
 	/* Now lock us into the TLBs via Hypervisor or OBP. */
 	if (tlb_type == hypervisor) {
-		hypervisor_tlb_lock(tte_vaddr, tte_data, HV_MMU_DMMU);
-		hypervisor_tlb_lock(tte_vaddr, tte_data, HV_MMU_IMMU);
-		if (bigkernel) {
-			tte_vaddr += 0x400000;
-			tte_data += 0x400000;
+		for (i = 0; i < num_kernel_image_mappings; i++) {
 			hypervisor_tlb_lock(tte_vaddr, tte_data, HV_MMU_DMMU);
 			hypervisor_tlb_lock(tte_vaddr, tte_data, HV_MMU_IMMU);
+			tte_vaddr += 0x400000;
+			tte_data += 0x400000;
 		}
 	} else {
-		prom_dtlb_load(tlb_ent, tte_data, tte_vaddr);
-		prom_itlb_load(tlb_ent, tte_data, tte_vaddr);
-		if (bigkernel) {
-			tlb_ent -= 1;
-			prom_dtlb_load(tlb_ent,
-				       tte_data + 0x400000, 
-				       tte_vaddr + 0x400000);
-			prom_itlb_load(tlb_ent,
-				       tte_data + 0x400000, 
-				       tte_vaddr + 0x400000);
+		for (i = 0; i < num_kernel_image_mappings; i++) {
+			prom_dtlb_load(tlb_ent - i, tte_data, tte_vaddr);
+			prom_itlb_load(tlb_ent - i, tte_data, tte_vaddr);
+			tte_vaddr += 0x400000;
+			tte_data += 0x400000;
 		}
-		sparc64_highest_unlocked_tlb_ent = tlb_ent - 1;
+		sparc64_highest_unlocked_tlb_ent = tlb_ent - i;
 	}
 	if (tlb_type == cheetah_plus) {
 		sparc64_kern_pri_context = (CTX_CHEETAH_PLUS_CTX0 |
@@ -1280,10 +1274,6 @@ void __cpuinit sun4v_ktsb_register(void)
 
 /* paging_init() sets up the page tables */
 
-extern void cheetah_ecache_flush_init(void);
-extern void sun4v_patch_tlb_handlers(void);
-
-extern void cpu_probe(void);
 extern void central_probe(void);
 
 static unsigned long last_valid_pfn;
@@ -1352,12 +1342,9 @@ void __init paging_init(void)
 	shift = kern_base + PAGE_OFFSET - ((unsigned long)KERNBASE);
 
 	real_end = (unsigned long)_end;
-	if ((real_end > ((unsigned long)KERNBASE + 0x400000)))
-		bigkernel = 1;
-	if ((real_end > ((unsigned long)KERNBASE + 0x800000))) {
-		prom_printf("paging_init: Kernel > 8MB, too large.\n");
-		prom_halt();
-	}
+	num_kernel_image_mappings = DIV_ROUND_UP(real_end - KERNBASE, 1 << 22);
+	printk("Kernel: Using %d locked TLB entries for main kernel image.\n",
+	       num_kernel_image_mappings);
 
 	/* Set kernel pgd to upper alias so physical page computations
 	 * work.
diff --git a/arch/sparc64/mm/tlb.c b/arch/sparc64/mm/tlb.c
index 3f10fc921b00..ae24919cba7c 100644
--- a/arch/sparc64/mm/tlb.c
+++ b/arch/sparc64/mm/tlb.c
@@ -23,9 +23,7 @@ DEFINE_PER_CPU(struct mmu_gather, mmu_gathers) = { 0, };
 
 void flush_tlb_pending(void)
 {
-	struct mmu_gather *mp = &__get_cpu_var(mmu_gathers);
-
-	preempt_disable();
+	struct mmu_gather *mp = &get_cpu_var(mmu_gathers);
 
 	if (mp->tlb_nr) {
 		flush_tsb_user(mp);
@@ -42,7 +40,7 @@ void flush_tlb_pending(void)
 		mp->tlb_nr = 0;
 	}
 
-	preempt_enable();
+	put_cpu_var(mmu_gathers);
 }
 
 void tlb_batch_add(struct mm_struct *mm, unsigned long vaddr, pte_t *ptep, pte_t orig)