Sparc64 updates mostly build fixes:

1) Update for schedule_tail and switch_to arg changes.
2) Update for PTE highmem.
3) Add random ioctls to ioctl32 translations.
4) Kill migration IPI.
5) Fixup scheduler bitmap function and move into bitops.h

arch/sparc64/defconfig

@@ -29,6 +29,7 @@ CONFIG_BBC_I2C=m
 CONFIG_VT=y
 CONFIG_VT_CONSOLE=y
 CONFIG_SMP=y
+# CONFIG_PREEMPT is not set
 CONFIG_SPARC64=y
 CONFIG_HOTPLUG=y
 CONFIG_HAVE_DEC_LOCK=y
@@ -156,7 +157,7 @@ CONFIG_SPARCAUDIO_CS4231=m
 #
 # Block devices
 #
-CONFIG_BLK_DEV_FD=y
+# CONFIG_BLK_DEV_FD is not set
 CONFIG_BLK_DEV_LOOP=m
 CONFIG_BLK_DEV_NBD=m
 
@@ -509,12 +510,12 @@ CONFIG_WINBOND_840=m
 CONFIG_ACENIC=m
 # CONFIG_ACENIC_OMIT_TIGON_I is not set
 CONFIG_DL2K=m
+CONFIG_E1000=m
 CONFIG_MYRI_SBUS=m
 CONFIG_NS83820=m
 CONFIG_HAMACHI=m
 CONFIG_YELLOWFIN=m
 CONFIG_SK98LIN=m
-# CONFIG_TIGON3 is not set
 CONFIG_FDDI=y
 # CONFIG_DEFXX is not set
 CONFIG_SKFP=m
@@ -674,6 +675,7 @@ CONFIG_NFS_V3=y
 # CONFIG_ROOT_NFS is not set
 CONFIG_NFSD=m
 CONFIG_NFSD_V3=y
+CONFIG_NFSD_TCP=y
 CONFIG_SUNRPC=y
 CONFIG_LOCKD=y
 CONFIG_LOCKD_V4=y
@@ -743,27 +745,16 @@ CONFIG_NLS_DEFAULT="iso8859-1"
 # Sound
 #
 CONFIG_SOUND=m
-CONFIG_SOUND_BT878=m
-# CONFIG_SOUND_CMPCI is not set
-# CONFIG_SOUND_EMU10K1 is not set
-# CONFIG_MIDI_EMU10K1 is not set
-# CONFIG_SOUND_FUSION is not set
-# CONFIG_SOUND_CS4281 is not set
-# CONFIG_SOUND_ES1370 is not set
-CONFIG_SOUND_ES1371=m
-# CONFIG_SOUND_ESSSOLO1 is not set
-# CONFIG_SOUND_MAESTRO is not set
-# CONFIG_SOUND_MAESTRO3 is not set
-# CONFIG_SOUND_ICH is not set
-# CONFIG_SOUND_RME96XX is not set
-# CONFIG_SOUND_SONICVIBES is not set
-CONFIG_SOUND_TRIDENT=m
-# CONFIG_SOUND_MSNDCLAS is not set
-# CONFIG_SOUND_MSNDPIN is not set
-# CONFIG_SOUND_VIA82CXXX is not set
-# CONFIG_MIDI_VIA82CXXX is not set
-# CONFIG_SOUND_OSS is not set
-# CONFIG_SOUND_TVMIXER is not set
+
+#
+# Open Sound System
+#
+# CONFIG_SOUND_PRIME is not set
+
+#
+# Advanced Linux Sound Architecture
+#
+# CONFIG_SND is not set
 
 #
 # USB support
@@ -831,6 +822,7 @@ CONFIG_USB_STV680=m
 CONFIG_USB_VICAM=m
 CONFIG_USB_DSBR=m
 CONFIG_USB_DABUSB=m
+CONFIG_USB_KONICAWC=m
 
 #
 # USB Network adaptors

arch/sparc64/kernel/entry.S

@@ -1436,7 +1436,6 @@ ret_from_syscall:
 		 * %o7 for us.  Check performance counter stuff too.
 		 */
 		andn		%o7, _TIF_NEWCHILD, %l0
-		mov		%g5, %o0	/* 'prev' */
 		call		schedule_tail
 		 stx		%l0, [%g6 + TI_FLAGS]
 		andcc		%l0, _TIF_PERFCTR, %g0

arch/sparc64/kernel/ioctl32.c

@@ -96,6 +96,7 @@
 #include <linux/usb.h>
 #include <linux/usbdevice_fs.h>
 #include <linux/nbd.h>
+#include <linux/random.h>
 
 /* Use this to get at 32-bit user passed pointers. 
    See sys_sparc32.c for description about these. */
@@ -4527,6 +4528,13 @@ COMPATIBLE_IOCTL(WDIOC_KEEPALIVE)
 COMPATIBLE_IOCTL(WIOCSTART)
 COMPATIBLE_IOCTL(WIOCSTOP)
 COMPATIBLE_IOCTL(WIOCGSTAT)
+/* Big R */
+COMPATIBLE_IOCTL(RNDGETENTCNT)
+COMPATIBLE_IOCTL(RNDADDTOENTCNT)
+COMPATIBLE_IOCTL(RNDGETPOOL)
+COMPATIBLE_IOCTL(RNDADDENTROPY)
+COMPATIBLE_IOCTL(RNDZAPENTCNT)
+COMPATIBLE_IOCTL(RNDCLEARPOOL)
 /* Bluetooth ioctls */
 COMPATIBLE_IOCTL(HCIDEVUP)
 COMPATIBLE_IOCTL(HCIDEVDOWN)

arch/sparc64/kernel/process.c

@@ -106,9 +106,11 @@ void kpreempt_maybe(void)
 	int cpu = smp_processor_id();
 
 	if (local_irq_count(cpu) == 0 &&
-	    local_bh_count(cpu) == 0)
-		preempt_schedule();
-	current_thread_info()->preempt_count--;
+	    local_bh_count(cpu) == 0 &&
+	    test_thread_flag(TIF_NEED_RESCHED)) {
+		current->state = TASK_RUNNING;
+		schedule();
+	}
 }
 #endif
 

arch/sparc64/kernel/ptrace.c

@@ -627,9 +627,11 @@ asmlinkage void syscall_trace(void)
 	if (!(current->ptrace & PT_PTRACED))
 		return;
 	current->exit_code = SIGTRAP;
+	preempt_disable();
 	current->state = TASK_STOPPED;
 	notify_parent(current, SIGCHLD);
 	schedule();
+	preempt_enable();
 	/*
 	 * this isn't the same as continuing with a signal, but it will do
 	 * for normal use.  strace only continues with a signal if the

arch/sparc64/kernel/rtrap.S

@@ -276,8 +276,8 @@ to_kernel:
 		 add			%l5, 1, %l6
 		stw			%l6, [%g6 + TI_PRE_COUNT]
 		call			kpreempt_maybe
-		 wrpr			%g0, RTRAP_PSTATE, %pstate
-		wrpr			%g0, RTRAP_PSTATE_IRQOFF, %pstate		
+		 nop
+		ba,pt			%xcc, rtrap
 		 stw			%l5, [%g6 + TI_PRE_COUNT]
 #endif
 kern_fpucheck:	ldub			[%g6 + TI_FPDEPTH], %l5

arch/sparc64/kernel/setup.c

@@ -160,11 +160,16 @@ int prom_callback(long *args)
 			pmdp = pmd_offset(pgdp, va);
 			if (pmd_none(*pmdp))
 				goto done;
-			ptep = pte_offset(pmdp, va);
-			if (!pte_present(*ptep))
-				goto done;
+
+			/* Preemption implicitly disabled by virtue of
+			 * being called from inside OBP.
+			 */
+			ptep = pte_offset_map(pmdp, va);
+			if (pte_present(*ptep)) {
 				tte = pte_val(*ptep);
 				res = PROM_TRUE;
+			}
+			pte_unmap(ptep);
 			goto done;
 		}
 
@@ -210,11 +215,15 @@ int prom_callback(long *args)
 			pmdp = pmd_offset(pgdp, va);
 			if (pmd_none(*pmdp))
 				goto done;
-			ptep = pte_offset(pmdp, va);
-			if (!pte_present(*ptep))
-				goto done;
+
+			/* Preemption implicitly disabled by virtue of
+			 * being called from inside OBP.
+			 */
+			ptep = pte_offset_kernel(pmdp, va);
+			if (pte_present(*ptep)) {
 				tte = pte_val(*ptep);
 				res = PROM_TRUE;
+			}
 			goto done;
 		}
 
@@ -530,7 +539,7 @@ void __init setup_arch(char **cmdline_p)
 	if (!root_flags)
 		root_mountflags &= ~MS_RDONLY;
 	ROOT_DEV = to_kdev_t(root_dev);
-#ifdef CONFIG_BLK_DEV_RAM
+#ifdef CONFIG_BLK_DEV_INITRD
 	rd_image_start = ram_flags & RAMDISK_IMAGE_START_MASK;
 	rd_prompt = ((ram_flags & RAMDISK_PROMPT_FLAG) != 0);
 	rd_doload = ((ram_flags & RAMDISK_LOAD_FLAG) != 0);	

arch/sparc64/kernel/signal.c

@@ -713,9 +713,11 @@ static int do_signal(sigset_t *oldset, struct pt_regs * regs,
 
 		if ((current->ptrace & PT_PTRACED) && signr != SIGKILL) {
 			current->exit_code = signr;
+			preempt_disable();
 			current->state = TASK_STOPPED;
 			notify_parent(current, SIGCHLD);
 			schedule();
+			preempt_enable();
 			if (!(signr = current->exit_code))
 				continue;
 			current->exit_code = 0;
@@ -766,16 +768,20 @@ static int do_signal(sigset_t *oldset, struct pt_regs * regs,
 				if (is_orphaned_pgrp(current->pgrp))
 					continue;
 
-			case SIGSTOP:
-				if (current->ptrace & PT_PTRACED)
-					continue;
-				current->state = TASK_STOPPED;
+			case SIGSTOP: {
+				struct signal_struct *sig;
+
 				current->exit_code = signr;
-				if (!(current->p_pptr->sig->action[SIGCHLD-1].sa.sa_flags &
+				sig = current->p_pptr->sig;
+				preempt_disable();
+				current->state = TASK_STOPPED;
+				if (sig && !(sig->action[SIGCHLD-1].sa.sa_flags &
 				      SA_NOCLDSTOP))
 					notify_parent(current, SIGCHLD);
 				schedule();
+				preempt_enable();
 				continue;
+			}
 
 			case SIGQUIT: case SIGILL: case SIGTRAP:
 			case SIGABRT: case SIGFPE: case SIGSEGV:

arch/sparc64/kernel/signal32.c

@@ -776,7 +776,7 @@ static inline void new_setup_frame32(struct k_sigaction *ka, struct pt_regs *reg
 		unsigned long address = ((unsigned long)&(sf->insns[0]));
 		pgd_t *pgdp = pgd_offset(current->mm, address);
 		pmd_t *pmdp = pmd_offset(pgdp, address);
-		pte_t *ptep = pte_offset(pmdp, address);
+		pte_t *ptep;
 
 		regs->u_regs[UREG_I7] = (unsigned long) (&(sf->insns[0]) - 2);
 	
@@ -785,6 +785,8 @@ static inline void new_setup_frame32(struct k_sigaction *ka, struct pt_regs *reg
 		if (err)
 			goto sigsegv;
 
+		preempt_disable();
+		ptep = pte_offset_map(pmdp, address);
 		if (pte_present(*ptep)) {
 			unsigned long page = (unsigned long) page_address(pte_page(*ptep));
 
@@ -794,6 +796,8 @@ static inline void new_setup_frame32(struct k_sigaction *ka, struct pt_regs *reg
 			: : "r" (page), "r" (address & (PAGE_SIZE - 1))
 			: "memory");
 		}
+		pte_unmap(ptep);
+		preempt_enable();
 	}
 	return;
 
@@ -1225,7 +1229,7 @@ static inline void setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs
 		unsigned long address = ((unsigned long)&(sf->insns[0]));
 		pgd_t *pgdp = pgd_offset(current->mm, address);
 		pmd_t *pmdp = pmd_offset(pgdp, address);
-		pte_t *ptep = pte_offset(pmdp, address);
+		pte_t *ptep;
 
 		regs->u_regs[UREG_I7] = (unsigned long) (&(sf->insns[0]) - 2);
 	
@@ -1237,6 +1241,8 @@ static inline void setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs
 		if (err)
 			goto sigsegv;
 
+		preempt_disable();
+		ptep = pte_offset_map(pmdp, address);
 		if (pte_present(*ptep)) {
 			unsigned long page = (unsigned long) page_address(pte_page(*ptep));
 
@@ -1246,6 +1252,8 @@ static inline void setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs
 			: : "r" (page), "r" (address & (PAGE_SIZE - 1))
 			: "memory");
 		}
+		pte_unmap(ptep);
+		preempt_enable();
 	}
 	return;
 
@@ -1379,9 +1387,11 @@ int do_signal32(sigset_t *oldset, struct pt_regs * regs,
 
 		if ((current->ptrace & PT_PTRACED) && signr != SIGKILL) {
 			current->exit_code = signr;
+			preempt_disable();
 			current->state = TASK_STOPPED;
 			notify_parent(current, SIGCHLD);
 			schedule();
+			preempt_enable();
 			if (!(signr = current->exit_code))
 				continue;
 			current->exit_code = 0;
@@ -1432,17 +1442,20 @@ int do_signal32(sigset_t *oldset, struct pt_regs * regs,
 				if (is_orphaned_pgrp(current->pgrp))
 					continue;
 
-			case SIGSTOP:
-				if (current->ptrace & PT_PTRACED)
-					continue;
-				current->state = TASK_STOPPED;
+			case SIGSTOP: {
+				struct signal_struct *sig;
+
 				current->exit_code = signr;
-				if (!(current->p_pptr->sig->action[SIGCHLD-1].sa.sa_flags &
+				sig = current->p_pptr->sig;
+				preempt_disable();
+				current->state = TASK_STOPPED;
+				if (sig && !(sig->action[SIGCHLD-1].sa.sa_flags &
 				      SA_NOCLDSTOP))
 					notify_parent(current, SIGCHLD);
 				schedule();
+				preempt_enable();
 				continue;
-
+			}
 			case SIGQUIT: case SIGILL: case SIGTRAP:
 			case SIGABRT: case SIGFPE: case SIGSEGV:
 			case SIGBUS: case SIGSYS: case SIGXCPU: case SIGXFSZ:

arch/sparc64/kernel/smp.c

@@ -885,48 +885,6 @@ void smp_flush_tlb_page(struct mm_struct *mm, unsigned long page)
 	}
 }
 
-/* Process migration IPIs. */
-
-extern unsigned long xcall_migrate_task;
-
-static spinlock_t migration_lock = SPIN_LOCK_UNLOCKED;
-static task_t *new_task;
-
-void smp_migrate_task(int cpu, task_t *p)
-{
-	unsigned long mask = 1UL << cpu;
-
-	if (cpu == smp_processor_id())
-		return;
-
-	if (smp_processors_ready && (cpu_present_map & mask) != 0) {
-		u64 data0 = (((u64)&xcall_migrate_task) & 0xffffffff);
-
-		_raw_spin_lock(&migration_lock);
-		new_task = p;
-
-		if (tlb_type == spitfire)
-			spitfire_xcall_deliver(data0, 0, 0, mask);
-		else
-			cheetah_xcall_deliver(data0, 0, 0, mask);
-	}
-}
-
-/* Called at PIL level 1. */
-asmlinkage void smp_task_migration_interrupt(int irq, struct pt_regs *regs)
-{
-	task_t *p;
-
-	if (irq != PIL_MIGRATE)
-		BUG();
-
-	clear_softint(1 << irq);
-
-	p = new_task;
-	_raw_spin_unlock(&migration_lock);
-	sched_task_migrated(p);
-}
-
 /* CPU capture. */
 /* #define CAPTURE_DEBUG */
 extern unsigned long xcall_capture;

arch/sparc64/kernel/sparc64_ksyms.c

@@ -249,7 +249,7 @@ EXPORT_SYMBOL(_sigpause_common);
 /* Should really be in linux/kernel/ksyms.c */
 EXPORT_SYMBOL(dump_thread);
 EXPORT_SYMBOL(dump_fpu);
-EXPORT_SYMBOL(pte_alloc_one);
+EXPORT_SYMBOL(pte_alloc_one_kernel);
 #ifndef CONFIG_SMP
 EXPORT_SYMBOL(pgt_quicklists);
 #endif

arch/sparc64/kernel/sys_sparc32.c

@@ -50,6 +50,7 @@
 #include <linux/icmpv6.h>
 #include <linux/sysctl.h>
 #include <linux/binfmts.h>
+#include <linux/dnotify.h>
 
 #include <asm/types.h>
 #include <asm/ipc.h>
@@ -1067,16 +1068,20 @@ static long do_readv_writev32(int type, struct file *file,
 	/* First get the "struct iovec" from user memory and
 	 * verify all the pointers
 	 */
+	retval = 0;
 	if (!count)
-		return 0;
+		goto out_nofree;
+	retval = -EFAULT;
 	if (verify_area(VERIFY_READ, vector, sizeof(struct iovec32)*count))
-		return -EFAULT;
+		goto out_nofree;
+	retval = -EINVAL;
 	if (count > UIO_MAXIOV)
-		return -EINVAL;
+		goto out_nofree;
 	if (count > UIO_FASTIOV) {
+		retval = -ENOMEM;
 		iov = kmalloc(count*sizeof(struct iovec), GFP_KERNEL);
 		if (!iov)
-			return -ENOMEM;
+			goto out_nofree;
 	}
 
 	tot_len = 0;
@@ -1136,6 +1141,11 @@ static long do_readv_writev32(int type, struct file *file,
 out:
 	if (iov != iovstack)
 		kfree(iov);
+out_nofree:
+	/* VERIFY_WRITE actually means a read, as we write to user space */
+	if ((retval + (type == VERIFY_WRITE)) > 0)
+		dnotify_parent(file->f_dentry,
+			(type == VERIFY_WRITE) ? DN_MODIFY : DN_ACCESS);
 
 	return retval;
 }

arch/sparc64/kernel/ttable.S

@@ -44,11 +44,7 @@ tl0_stdfmna:	TRAP_NOSAVE(do_stdfmna)
 tl0_privact:	TRAP_NOSAVE(__do_privact)
 tl0_resv038:	BTRAP(0x38) BTRAP(0x39) BTRAP(0x3a) BTRAP(0x3b) BTRAP(0x3c) BTRAP(0x3d)
 tl0_resv03e:	BTRAP(0x3e) BTRAP(0x3f) BTRAP(0x40)
-#ifdef CONFIG_SMP
-tl0_irq1:	TRAP_IRQ(smp_task_migration_interrupt, 1)
-#else
 tl0_irq1:	BTRAP(0x41)
-#endif
 tl0_irq2:	TRAP_IRQ(handler_irq, 2)
 tl0_irq3:	TRAP_IRQ(handler_irq, 3)  TRAP_IRQ(handler_irq, 4)
 tl0_irq5:	TRAP_IRQ(handler_irq, 5)  TRAP_IRQ(handler_irq, 6)

arch/sparc64/mm/fault.c

@@ -160,10 +160,12 @@ static unsigned int get_user_insn(unsigned long tpc)
 	pmdp = pmd_offset(pgdp, tpc);
 	if (pmd_none(*pmdp))
 		goto outret;
-	ptep = pte_offset(pmdp, tpc);
+
+	/* This disables preemption for us as well. */
 	__asm__ __volatile__("rdpr %%pstate, %0" : "=r" (pstate));
 	__asm__ __volatile__("wrpr %0, %1, %%pstate"
 				: : "r" (pstate), "i" (PSTATE_IE));
+	ptep = pte_offset_map(pmdp, tpc);
 	pte = *ptep;
 	if (!pte_present(pte))
 		goto out;
@@ -177,6 +179,7 @@ static unsigned int get_user_insn(unsigned long tpc)
 			     : "r" (pa), "i" (ASI_PHYS_USE_EC));
 
 out:
+	pte_unmap(ptep);
 	__asm__ __volatile__("wrpr %0, 0x0, %%pstate" : : "r" (pstate));
 outret:
 	return insn;

arch/sparc64/mm/generic.c

@@ -101,10 +101,11 @@ static inline int io_remap_pmd_range(pmd_t * pmd, unsigned long address, unsigne
 		end = PGDIR_SIZE;
 	offset -= address;
 	do {
-		pte_t * pte = pte_alloc(current->mm, pmd, address);
+		pte_t * pte = pte_alloc_map(current->mm, pmd, address);
 		if (!pte)
 			return -ENOMEM;
 		io_remap_pte_range(pte, address, end - address, address + offset, prot, space);
+		pte_unmap(pte);
 		address = (address + PMD_SIZE) & PMD_MASK;
 		pmd++;
 	} while (address < end);

arch/sparc64/mm/init.c

@@ -65,24 +65,27 @@ struct page *mem_map_zero;
 
 int bigkernel = 0;
 
-int do_check_pgt_cache(int low, int high)
-{
-        int freed = 0;
+/* XXX Tune this... */
+#define PGT_CACHE_LOW	25
+#define PGT_CACHE_HIGH	50
 
-	if (pgtable_cache_size > high) {
+void check_pgt_cache(void)
+{
+	preempt_disable();
+	if (pgtable_cache_size > PGT_CACHE_HIGH) {
 		do {
 #ifdef CONFIG_SMP
 			if (pgd_quicklist)
-				free_pgd_slow(get_pgd_fast()), freed++;
+				free_pgd_slow(get_pgd_fast());
 #endif
 			if (pte_quicklist[0])
-				free_pte_slow(pte_alloc_one_fast(NULL, 0)), freed++;
+				free_pte_slow(pte_alloc_one_fast(NULL, 0));
 			if (pte_quicklist[1])
-				free_pte_slow(pte_alloc_one_fast(NULL, 1 << (PAGE_SHIFT + 10))), freed++;
-		} while (pgtable_cache_size > low);
+				free_pte_slow(pte_alloc_one_fast(NULL, 1 << (PAGE_SHIFT + 10)));
+		} while (pgtable_cache_size > PGT_CACHE_LOW);
 	}
 #ifndef CONFIG_SMP
-        if (pgd_cache_size > high / 4) {
+        if (pgd_cache_size > PGT_CACHE_HIGH / 4) {
 		struct page *page, *page2;
                 for (page2 = NULL, page = (struct page *)pgd_quicklist; page;) {
                         if ((unsigned long)page->pprev_hash == 3) {
@@ -94,12 +97,11 @@ int do_check_pgt_cache(int low, int high)
                                 page->pprev_hash = NULL;
                                 pgd_cache_size -= 2;
                                 __free_page(page);
-                                freed++;
                                 if (page2)
                                         page = page2->next_hash;
                                 else
                                         page = (struct page *)pgd_quicklist;
-                                if (pgd_cache_size <= low / 4)
+                                if (pgd_cache_size <= PGT_CACHE_LOW / 4)
                                         break;
                                 continue;
                         }
@@ -108,7 +110,7 @@ int do_check_pgt_cache(int low, int high)
                 }
         }
 #endif
-        return freed;
+	preempt_enable();
 }
 
 #ifdef CONFIG_DEBUG_DCFLUSH
@@ -143,7 +145,7 @@ __inline__ void flush_dcache_page_impl(struct page *page)
 static __inline__ void set_dcache_dirty(struct page *page)
 {
 	unsigned long mask = smp_processor_id();
-	unsigned long non_cpu_bits = (1UL << 24UL) - 1UL;
+	unsigned long non_cpu_bits = ~((NR_CPUS - 1UL) << 24UL);
 	mask = (mask << 24) | (1UL << PG_dcache_dirty);
 	__asm__ __volatile__("1:\n\t"
 			     "ldx	[%2], %%g7\n\t"
@@ -166,6 +168,7 @@ static __inline__ void clear_dcache_dirty_cpu(struct page *page, unsigned long c
 			     "1:\n\t"
 			     "ldx	[%2], %%g7\n\t"
 			     "srlx	%%g7, 24, %%g5\n\t"
+			     "and	%%g5, %3, %%g5\n\t"
 			     "cmp	%%g5, %0\n\t"
 			     "bne,pn	%%icc, 2f\n\t"
 			     " andn	%%g7, %1, %%g5\n\t"
@@ -175,7 +178,8 @@ static __inline__ void clear_dcache_dirty_cpu(struct page *page, unsigned long c
 			     " membar	#StoreLoad | #StoreStore\n"
 			     "2:"
 			     : /* no outputs */
-			     : "r" (cpu), "r" (mask), "r" (&page->flags)
+			     : "r" (cpu), "r" (mask), "r" (&page->flags),
+			       "i" (NR_CPUS - 1UL)
 			     : "g5", "g7");
 }
 
@@ -189,7 +193,7 @@ void update_mmu_cache(struct vm_area_struct *vma, unsigned long address, pte_t p
 	if (VALID_PAGE(page) &&
 	    page->mapping &&
 	    ((pg_flags = page->flags) & (1UL << PG_dcache_dirty))) {
-		int cpu = (pg_flags >> 24);
+		int cpu = ((pg_flags >> 24) & (NR_CPUS - 1UL));
 
 		/* This is just to optimize away some function calls
 		 * in the SMP case.
@@ -212,8 +216,8 @@ void flush_dcache_page(struct page *page)
 	int dirty_cpu = dcache_dirty_cpu(page);
 
 	if (page->mapping &&
-	    page->mapping->i_mmap == NULL &&
-	    page->mapping->i_mmap_shared == NULL) {
+	    list_empty(&page->mapping->i_mmap) &&
+	    list_empty(&page->mapping->i_mmap_shared)) {
 		if (dirty) {
 			if (dirty_cpu == smp_processor_id())
 				return;
@@ -244,7 +248,7 @@ static inline void flush_cache_pte_range(struct mm_struct *mm, pmd_t *pmd, unsig
 
 	if (pmd_none(*pmd))
 		return;
-	ptep = pte_offset(pmd, address);
+	ptep = pte_offset_map(pmd, address);
 	offset = address & ~PMD_MASK;
 	if (offset + size > PMD_SIZE)
 		size = PMD_SIZE - offset;
@@ -267,6 +271,7 @@ static inline void flush_cache_pte_range(struct mm_struct *mm, pmd_t *pmd, unsig
 				flush_dcache_page_all(mm, page);
 		}
 	}
+	pte_unmap(ptep - 1);
 }
 
 static inline void flush_cache_pmd_range(struct mm_struct *mm, pgd_t *dir, unsigned long address, unsigned long size)
@@ -389,7 +394,7 @@ unsigned long prom_virt_to_phys(unsigned long promva, int *error)
 			*error = 1;
 		return(0);
 	}
-	ptep = (pte_t *)pmd_page(*pmdp) + ((promva >> 13) & 0x3ff);
+	ptep = (pte_t *)__pmd_page(*pmdp) + ((promva >> 13) & 0x3ff);
 	if (!pte_present(*ptep)) {
 		if (error)
 			*error = 1;
@@ -466,7 +471,7 @@ static void inherit_prom_mappings(void)
 					memset(ptep, 0, BASE_PAGE_SIZE);
 					pmd_set(pmdp, ptep);
 				}
-				ptep = (pte_t *)pmd_page(*pmdp) +
+				ptep = (pte_t *)__pmd_page(*pmdp) +
 						((vaddr >> 13) & 0x3ff);
 
 				val = trans[i].data;
@@ -1133,11 +1138,20 @@ struct pgtable_cache_struct pgt_quicklists;
 #else
 #define DC_ALIAS_SHIFT	0
 #endif
-pte_t *pte_alloc_one(struct mm_struct *mm, unsigned long address)
+pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address)
 {
-	struct page *page = alloc_pages(GFP_KERNEL, DC_ALIAS_SHIFT);
-	unsigned long color = VPTE_COLOR(address);
+	struct page *page;
+	unsigned long color;
+
+	{
+		pte_t *ptep = pte_alloc_one_fast(mm, address);
+
+		if (ptep)
+			return ptep;
+	}
 
+	color = VPTE_COLOR(address);
+	page = alloc_pages(GFP_KERNEL, DC_ALIAS_SHIFT);
 	if (page) {
 		unsigned long *to_free;
 		unsigned long paddr;
@@ -1159,9 +1173,11 @@ pte_t *pte_alloc_one(struct mm_struct *mm, unsigned long address)
 
 #if (L1DCACHE_SIZE > PAGE_SIZE)			/* is there D$ aliasing problem */
 		/* Now free the other one up, adjust cache size. */
+		preempt_disable();
 		*to_free = (unsigned long) pte_quicklist[color ^ 0x1];
 		pte_quicklist[color ^ 0x1] = to_free;
 		pgtable_cache_size++;
+		preempt_enable();
 #endif
 
 		return pte;

arch/sparc64/mm/ultra.S

@@ -684,11 +684,4 @@ xcall_call_function:
 	b,pt		%xcc, rtrap_irq
 	 nop
 
-	.globl		xcall_migrate_task
-xcall_migrate_task:
-	mov		1, %g2
-	sllx		%g2, (PIL_MIGRATE), %g2
-	wr		%g2, 0x0, %set_softint
-	retry
-
 #endif /* CONFIG_SMP */

include/asm-sparc64/bitops.h

@@ -7,6 +7,7 @@
 #ifndef _SPARC64_BITOPS_H
 #define _SPARC64_BITOPS_H
 
+#include <linux/compiler.h>
 #include <asm/byteorder.h>
 
 extern long ___test_and_set_bit(unsigned long nr, volatile void *addr);
@@ -101,6 +102,23 @@ static __inline__ unsigned long __ffs(unsigned long word)
 
 #ifdef __KERNEL__
 
+/*
+ * Every architecture must define this function. It's the fastest
+ * way of searching a 140-bit bitmap where the first 100 bits are
+ * unlikely to be set. It's guaranteed that at least one of the 140
+ * bits is cleared.
+ */
+static inline int sched_find_first_bit(unsigned long *b)
+{
+	if (unlikely(b[0]))
+		return __ffs(b[0]);
+	if (unlikely(((unsigned int)b[1])))
+		return __ffs(b[1]) + 64;
+	if (b[1] >> 32)
+		return __ffs(b[1] >> 32) + 96;
+	return __ffs(b[2]) + 128;
+}
+
 /*
  * ffs: find first bit set. This is defined the same way as
  * the libc and compiler builtin ffs routines, therefore

include/asm-sparc64/mmu_context.h

@@ -27,25 +27,6 @@
 #include <asm/system.h>
 #include <asm/spitfire.h>
 
-/*
- * Every architecture must define this function. It's the fastest
- * way of searching a 168-bit bitmap where the first 128 bits are
- * unlikely to be set. It's guaranteed that at least one of the 168
- * bits is cleared.
- */
-#if MAX_RT_PRIO != 128 || MAX_PRIO != 168
-# error update this function.
-#endif
-
-static inline int sched_find_first_bit(unsigned long *b)
-{
-	if (unlikely(b[0]))
-		return __ffs(b[0]);
-	if (unlikely(b[1]))
-		return __ffs(b[1]) + 64;
-	return __ffs(b[2]) + 128;
-}
-
 static inline void enter_lazy_tlb(struct mm_struct *mm, struct task_struct *tsk, unsigned cpu)
 {
 }

include/asm-sparc64/pgalloc.h

@@ -10,95 +10,6 @@
 #include <asm/spitfire.h>
 #include <asm/pgtable.h>
 
-/* Cache and TLB flush operations. */
-
-/* These are the same regardless of whether this is an SMP kernel or not. */
-#define flush_cache_mm(__mm) \
-	do { if ((__mm) == current->mm) flushw_user(); } while(0)
-extern void flush_cache_range(struct vm_area_struct *, unsigned long, unsigned long);
-#define flush_cache_page(vma, page) \
-	flush_cache_mm((vma)->vm_mm)
-
-/* This is unnecessary on the SpitFire since D-CACHE is write-through. */
-#define flush_page_to_ram(page)			do { } while (0)
-
-/* 
- * On spitfire, the icache doesn't snoop local stores and we don't
- * use block commit stores (which invalidate icache lines) during
- * module load, so we need this.
- */
-extern void flush_icache_range(unsigned long start, unsigned long end);
-
-extern void __flush_dcache_page(void *addr, int flush_icache);
-extern void __flush_icache_page(unsigned long);
-extern void flush_dcache_page_impl(struct page *page);
-#ifdef CONFIG_SMP
-extern void smp_flush_dcache_page_impl(struct page *page, int cpu);
-extern void flush_dcache_page_all(struct mm_struct *mm, struct page *page);
-#else
-#define smp_flush_dcache_page_impl(page,cpu) flush_dcache_page_impl(page)
-#define flush_dcache_page_all(mm,page) flush_dcache_page_impl(page)
-#endif
-
-extern void flush_dcache_page(struct page *page);
-
-extern void __flush_dcache_range(unsigned long start, unsigned long end);
-
-extern void __flush_cache_all(void);
-
-extern void __flush_tlb_all(void);
-extern void __flush_tlb_mm(unsigned long context, unsigned long r);
-extern void __flush_tlb_range(unsigned long context, unsigned long start,
-			      unsigned long r, unsigned long end,
-			      unsigned long pgsz, unsigned long size);
-extern void __flush_tlb_page(unsigned long context, unsigned long page, unsigned long r);
-
-#ifndef CONFIG_SMP
-
-#define flush_cache_all()	__flush_cache_all()
-#define flush_tlb_all()		__flush_tlb_all()
-
-#define flush_tlb_mm(__mm) \
-do { if(CTX_VALID((__mm)->context)) \
-	__flush_tlb_mm(CTX_HWBITS((__mm)->context), SECONDARY_CONTEXT); \
-} while(0)
-
-#define flush_tlb_range(__vma, start, end) \
-do { if(CTX_VALID((__vma)->vm_mm->context)) { \
-	unsigned long __start = (start)&PAGE_MASK; \
-	unsigned long __end = PAGE_ALIGN(end); \
-	__flush_tlb_range(CTX_HWBITS((__vma)->vm_mm->context), __start, \
-			  SECONDARY_CONTEXT, __end, PAGE_SIZE, \
-			  (__end - __start)); \
-     } \
-} while(0)
-
-#define flush_tlb_page(vma, page) \
-do { struct mm_struct *__mm = (vma)->vm_mm; \
-     if(CTX_VALID(__mm->context)) \
-	__flush_tlb_page(CTX_HWBITS(__mm->context), (page)&PAGE_MASK, \
-			 SECONDARY_CONTEXT); \
-} while(0)
-
-#else /* CONFIG_SMP */
-
-extern void smp_flush_cache_all(void);
-extern void smp_flush_tlb_all(void);
-extern void smp_flush_tlb_mm(struct mm_struct *mm);
-extern void smp_flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
-				unsigned long end);
-extern void smp_flush_tlb_page(struct mm_struct *mm, unsigned long page);
-
-#define flush_cache_all()	smp_flush_cache_all()
-#define flush_tlb_all()		smp_flush_tlb_all()
-#define flush_tlb_mm(mm)	smp_flush_tlb_mm(mm)
-#define flush_tlb_range(vma, start, end) \
-	smp_flush_tlb_range(vma, start, end)
-#define flush_tlb_page(vma, page) \
-	smp_flush_tlb_page((vma)->vm_mm, page)
-
-#endif /* ! CONFIG_SMP */
-
 #define VPTE_BASE_SPITFIRE	0xfffffffe00000000
 #if 1
 #define VPTE_BASE_CHEETAH	VPTE_BASE_SPITFIRE
@@ -106,7 +17,7 @@ extern void smp_flush_tlb_page(struct mm_struct *mm, unsigned long page);
 #define VPTE_BASE_CHEETAH	0xffe0000000000000
 #endif
 
-extern __inline__ void flush_tlb_pgtables(struct mm_struct *mm, unsigned long start,
+static __inline__ void flush_tlb_pgtables(struct mm_struct *mm, unsigned long start,
 					  unsigned long end)
 {
 	/* Note the signed type.  */
@@ -154,7 +65,7 @@ extern struct pgtable_cache_struct {
 
 #ifndef CONFIG_SMP
 
-extern __inline__ void free_pgd_fast(pgd_t *pgd)
+static __inline__ void free_pgd_fast(pgd_t *pgd)
 {
 	struct page *page = virt_to_page(pgd);
 
@@ -169,7 +80,7 @@ extern __inline__ void free_pgd_fast(pgd_t *pgd)
 	preempt_enable();
 }
 
-extern __inline__ pgd_t *get_pgd_fast(void)
+static __inline__ pgd_t *get_pgd_fast(void)
 {
         struct page *ret;
 
@@ -212,7 +123,7 @@ extern __inline__ pgd_t *get_pgd_fast(void)
 
 #else /* CONFIG_SMP */
 
-extern __inline__ void free_pgd_fast(pgd_t *pgd)
+static __inline__ void free_pgd_fast(pgd_t *pgd)
 {
 	preempt_disable();
 	*(unsigned long *)pgd = (unsigned long) pgd_quicklist;
@@ -221,7 +132,7 @@ extern __inline__ void free_pgd_fast(pgd_t *pgd)
 	preempt_enable();
 }
 
-extern __inline__ pgd_t *get_pgd_fast(void)
+static __inline__ pgd_t *get_pgd_fast(void)
 {
 	unsigned long *ret;
 
@@ -240,7 +151,7 @@ extern __inline__ pgd_t *get_pgd_fast(void)
 	return (pgd_t *)ret;
 }
 
-extern __inline__ void free_pgd_slow(pgd_t *pgd)
+static __inline__ void free_pgd_slow(pgd_t *pgd)
 {
 	free_page((unsigned long)pgd);
 }
@@ -257,23 +168,15 @@ extern __inline__ void free_pgd_slow(pgd_t *pgd)
 
 #define pgd_populate(MM, PGD, PMD)	pgd_set(PGD, PMD)
 
-extern __inline__ pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
-{
-	pmd_t *pmd = (pmd_t *)__get_free_page(GFP_KERNEL);
-	if (pmd)
-		memset(pmd, 0, PAGE_SIZE);
-	return pmd;
-}
-
-extern __inline__ pmd_t *pmd_alloc_one_fast(struct mm_struct *mm, unsigned long address)
+static __inline__ pmd_t *pmd_alloc_one_fast(struct mm_struct *mm, unsigned long address)
 {
 	unsigned long *ret;
 	int color = 0;
 
+	preempt_disable();
 	if (pte_quicklist[color] == NULL)
 		color = 1;
 
-	preempt_disable();
 	if((ret = (unsigned long *)pte_quicklist[color]) != NULL) {
 		pte_quicklist[color] = (unsigned long *)(*ret);
 		ret[0] = 0;
@@ -284,7 +187,20 @@ extern __inline__ pmd_t *pmd_alloc_one_fast(struct mm_struct *mm, unsigned long
 	return (pmd_t *)ret;
 }
 
-extern __inline__ void free_pmd_fast(pmd_t *pmd)
+static __inline__ pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
+{
+	pmd_t *pmd;
+
+	pmd = pmd_alloc_one_fast(mm, address);
+	if (!pmd) {
+		pmd = (pmd_t *)__get_free_page(GFP_KERNEL);
+		if (pmd)
+			memset(pmd, 0, PAGE_SIZE);
+	}
+	return pmd;
+}
+
+static __inline__ void free_pmd_fast(pmd_t *pmd)
 {
 	unsigned long color = DCACHE_COLOR((unsigned long)pmd);
 
@@ -295,16 +211,19 @@ extern __inline__ void free_pmd_fast(pmd_t *pmd)
 	preempt_enable();
 }
 
-extern __inline__ void free_pmd_slow(pmd_t *pmd)
+static __inline__ void free_pmd_slow(pmd_t *pmd)
 {
 	free_page((unsigned long)pmd);
 }
 
-#define pmd_populate(MM, PMD, PTE)	pmd_set(PMD, PTE)
+#define pmd_populate_kernel(MM, PMD, PTE)	pmd_set(PMD, PTE)
+#define pmd_populate(MM,PMD,PTE_PAGE)		\
+	pmd_populate_kernel(MM,PMD,page_address(PTE_PAGE))
 
-extern pte_t *pte_alloc_one(struct mm_struct *mm, unsigned long address);
+extern pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address);
+#define pte_alloc_one(MM,ADDR)	virt_to_page(pte_alloc_one_kernel(MM,ADDR))
 
-extern __inline__ pte_t *pte_alloc_one_fast(struct mm_struct *mm, unsigned long address)
+static __inline__ pte_t *pte_alloc_one_fast(struct mm_struct *mm, unsigned long address)
 {
 	unsigned long color = VPTE_COLOR(address);
 	unsigned long *ret;
@@ -319,7 +238,7 @@ extern __inline__ pte_t *pte_alloc_one_fast(struct mm_struct *mm, unsigned long
 	return (pte_t *)ret;
 }
 
-extern __inline__ void free_pte_fast(pte_t *pte)
+static __inline__ void free_pte_fast(pte_t *pte)
 {
 	unsigned long color = DCACHE_COLOR((unsigned long)pte);
 
@@ -330,16 +249,15 @@ extern __inline__ void free_pte_fast(pte_t *pte)
 	preempt_enable();
 }
 
-extern __inline__ void free_pte_slow(pte_t *pte)
+static __inline__ void free_pte_slow(pte_t *pte)
 {
 	free_page((unsigned long)pte);
 }
 
-#define pte_free(pte)		free_pte_fast(pte)
+#define pte_free_kernel(pte)	free_pte_fast(pte)
+#define pte_free(pte)		free_pte_fast(page_address(pte))
 #define pmd_free(pmd)		free_pmd_fast(pmd)
 #define pgd_free(pgd)		free_pgd_fast(pgd)
 #define pgd_alloc(mm)		get_pgd_fast()
 
-extern int do_check_pgt_cache(int, int);
-
 #endif /* _SPARC64_PGALLOC_H */

include/asm-sparc64/pgtable.h

@@ -36,6 +36,95 @@
 #define LOW_OBP_ADDRESS		0x00000000f0000000
 #define HI_OBP_ADDRESS		0x0000000100000000
 
+#ifndef __ASSEMBLY__
+
+/* Cache and TLB flush operations. */
+
+/* These are the same regardless of whether this is an SMP kernel or not. */
+#define flush_cache_mm(__mm) \
+	do { if ((__mm) == current->mm) flushw_user(); } while(0)
+extern void flush_cache_range(struct vm_area_struct *, unsigned long, unsigned long);
+#define flush_cache_page(vma, page) \
+	flush_cache_mm((vma)->vm_mm)
+
+
+/* 
+ * On spitfire, the icache doesn't snoop local stores and we don't
+ * use block commit stores (which invalidate icache lines) during
+ * module load, so we need this.
+ */
+extern void flush_icache_range(unsigned long start, unsigned long end);
+
+extern void __flush_dcache_page(void *addr, int flush_icache);
+extern void __flush_icache_page(unsigned long);
+extern void flush_dcache_page_impl(struct page *page);
+#ifdef CONFIG_SMP
+extern void smp_flush_dcache_page_impl(struct page *page, int cpu);
+extern void flush_dcache_page_all(struct mm_struct *mm, struct page *page);
+#else
+#define smp_flush_dcache_page_impl(page,cpu) flush_dcache_page_impl(page)
+#define flush_dcache_page_all(mm,page) flush_dcache_page_impl(page)
+#endif
+
+extern void __flush_dcache_range(unsigned long start, unsigned long end);
+
+extern void __flush_cache_all(void);
+
+extern void __flush_tlb_all(void);
+extern void __flush_tlb_mm(unsigned long context, unsigned long r);
+extern void __flush_tlb_range(unsigned long context, unsigned long start,
+			      unsigned long r, unsigned long end,
+			      unsigned long pgsz, unsigned long size);
+extern void __flush_tlb_page(unsigned long context, unsigned long page, unsigned long r);
+
+#ifndef CONFIG_SMP
+
+#define flush_cache_all()	__flush_cache_all()
+#define flush_tlb_all()		__flush_tlb_all()
+
+#define flush_tlb_mm(__mm) \
+do { if(CTX_VALID((__mm)->context)) \
+	__flush_tlb_mm(CTX_HWBITS((__mm)->context), SECONDARY_CONTEXT); \
+} while(0)
+
+#define flush_tlb_range(__vma, start, end) \
+do { if(CTX_VALID((__vma)->vm_mm->context)) { \
+	unsigned long __start = (start)&PAGE_MASK; \
+	unsigned long __end = PAGE_ALIGN(end); \
+	__flush_tlb_range(CTX_HWBITS((__vma)->vm_mm->context), __start, \
+			  SECONDARY_CONTEXT, __end, PAGE_SIZE, \
+			  (__end - __start)); \
+     } \
+} while(0)
+
+#define flush_tlb_page(vma, page) \
+do { struct mm_struct *__mm = (vma)->vm_mm; \
+     if(CTX_VALID(__mm->context)) \
+	__flush_tlb_page(CTX_HWBITS(__mm->context), (page)&PAGE_MASK, \
+			 SECONDARY_CONTEXT); \
+} while(0)
+
+#else /* CONFIG_SMP */
+
+extern void smp_flush_cache_all(void);
+extern void smp_flush_tlb_all(void);
+extern void smp_flush_tlb_mm(struct mm_struct *mm);
+extern void smp_flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
+				unsigned long end);
+extern void smp_flush_tlb_page(struct mm_struct *mm, unsigned long page);
+
+#define flush_cache_all()	smp_flush_cache_all()
+#define flush_tlb_all()		smp_flush_tlb_all()
+#define flush_tlb_mm(mm)	smp_flush_tlb_mm(mm)
+#define flush_tlb_range(vma, start, end) \
+	smp_flush_tlb_range(vma, start, end)
+#define flush_tlb_page(vma, page) \
+	smp_flush_tlb_page((vma)->vm_mm, page)
+
+#endif /* ! CONFIG_SMP */
+
+#endif /* ! __ASSEMBLY__ */
+
 /* XXX All of this needs to be rethought so we can take advantage
  * XXX cheetah's full 64-bit virtual address space, ie. no more hole
  * XXX in the middle like on spitfire. -DaveM
@@ -215,7 +304,8 @@ extern inline pte_t pte_modify(pte_t orig_pte, pgprot_t new_prot)
 	(pmd_val(*(pmdp)) = (__pa((unsigned long) (ptep)) >> 11UL))
 #define pgd_set(pgdp, pmdp)	\
 	(pgd_val(*(pgdp)) = (__pa((unsigned long) (pmdp)) >> 11UL))
-#define pmd_page(pmd)			((unsigned long) __va((pmd_val(pmd)<<11UL)))
+#define __pmd_page(pmd)			((unsigned long) __va((pmd_val(pmd)<<11UL)))
+#define pmd_page(pmd) 			virt_to_page((void *)__pmd_page(pmd))
 #define pgd_page(pgd)			((unsigned long) __va((pgd_val(pgd)<<11UL)))
 #define pte_none(pte) 			(!pte_val(pte))
 #define pte_present(pte)		(pte_val(pte) & _PAGE_PRESENT)
@@ -264,8 +354,13 @@ extern inline pte_t pte_modify(pte_t orig_pte, pgprot_t new_prot)
 					((address >> PMD_SHIFT) & (REAL_PTRS_PER_PMD-1)))
 
 /* Find an entry in the third-level page table.. */
-#define pte_offset(dir, address)	((pte_t *) pmd_page(*(dir)) + \
+#define __pte_offset(dir, address)	((pte_t *) __pmd_page(*(dir)) + \
 					((address >> PAGE_SHIFT) & (PTRS_PER_PTE - 1)))
+#define pte_offset_kernel		__pte_offset
+#define pte_offset_map			__pte_offset
+#define pte_offset_map_nested		__pte_offset
+#define pte_unmap(pte)			do { } while (0)
+#define pte_unmap_nested(pte)		do { } while (0)
 
 extern pgd_t swapper_pg_dir[1];
 
@@ -312,10 +407,10 @@ sun4u_get_pte (unsigned long addr)
 		return addr & _PAGE_PADDR;
 	if ((addr >= LOW_OBP_ADDRESS) && (addr < HI_OBP_ADDRESS))
 		return prom_virt_to_phys(addr, 0);
-	pgdp = pgd_offset_k (addr);
-	pmdp = pmd_offset (pgdp, addr);
-	ptep = pte_offset (pmdp, addr);
-	return pte_val (*ptep) & _PAGE_PADDR;
+	pgdp = pgd_offset_k(addr);
+	pmdp = pmd_offset(pgdp, addr);
+	ptep = pte_offset_kernel(pmdp, addr);
+	return pte_val(*ptep) & _PAGE_PADDR;
 }
 
 extern __inline__ unsigned long
@@ -350,11 +445,18 @@ extern int io_remap_page_range(struct vm_area_struct *vma, unsigned long from, u
 extern unsigned long get_fb_unmapped_area(struct file *filp, unsigned long, unsigned long, unsigned long, unsigned long);
 #define HAVE_ARCH_FB_UNMAPPED_AREA
 
-#endif /* !(__ASSEMBLY__) */
-
 /*
  * No page table caches to initialise
  */
 #define pgtable_cache_init()	do { } while (0)
 
+extern void check_pgt_cache(void);
+
+extern void flush_dcache_page(struct page *page);
+
+/* This is unnecessary on the SpitFire since D-CACHE is write-through. */
+#define flush_page_to_ram(page)			do { } while (0)
+
+#endif /* !(__ASSEMBLY__) */
+
 #endif /* !(_SPARC64_PGTABLE_H) */

include/asm-sparc64/pil.h

@@ -12,10 +12,11 @@
  * XXX happily sit at the same PIL.  We would then need only two
  * XXX PILs, one for devices and one for the CPU local timer tick.
  */
-#define PIL_MIGRATE	1
+/* None currently allocated, '1' is available for use. */
+#define PIL_SMP_1	1
 
 #ifndef __ASSEMBLY__
-#define PIL_RESERVED(PIL)	((PIL) == PIL_MIGRATE)
+#define PIL_RESERVED(PIL)	((PIL) == PIL_SMP_1)
 #endif
 
 #endif /* !(_SPARC64_PIL_H) */

include/asm-sparc64/system.h

@@ -172,7 +172,7 @@ if ((PREV)->thread.smp_lock_count) {					\
 	 * not preserve it's value.  Hairy, but it lets us remove 2 loads
 	 * and 2 stores in this critical code path.  -DaveM
 	 */
-#define switch_to(prev, next, last)						\
+#define switch_to(prev, next)							\
 do {	CHECK_LOCKS(prev);							\
 	if (test_thread_flag(TIF_PERFCTR)) {					\
 		unsigned long __tmp;						\
@@ -193,16 +193,16 @@ do {	CHECK_LOCKS(prev);							\
 	"stx	%%i6, [%%sp + 2047 + 0x70]\n\t"					\
 	"stx	%%i7, [%%sp + 2047 + 0x78]\n\t"					\
 	"rdpr	%%wstate, %%o5\n\t"						\
-	"stx	%%o6, [%%g6 + %3]\n\t"						\
-	"stb	%%o5, [%%g6 + %2]\n\t"						\
+	"stx	%%o6, [%%g6 + %2]\n\t"						\
+	"stb	%%o5, [%%g6 + %1]\n\t"						\
 	"rdpr	%%cwp, %%o5\n\t"						\
-	"stb	%%o5, [%%g6 + %5]\n\t"						\
-	"mov	%1, %%g6\n\t"							\
-	"ldub	[%1 + %5], %%g1\n\t"						\
+	"stb	%%o5, [%%g6 + %4]\n\t"						\
+	"mov	%0, %%g6\n\t"							\
+	"ldub	[%0 + %4], %%g1\n\t"						\
 	"wrpr	%%g1, %%cwp\n\t"						\
-	"ldx	[%%g6 + %3], %%o6\n\t"						\
-	"ldub	[%%g6 + %2], %%o5\n\t"						\
-	"ldx	[%%g6 + %4], %%o7\n\t"						\
+	"ldx	[%%g6 + %2], %%o6\n\t"						\
+	"ldub	[%%g6 + %1], %%o5\n\t"						\
+	"ldx	[%%g6 + %3], %%o7\n\t"						\
 	"mov	%%g6, %%l2\n\t"							\
 	"wrpr	%%o5, 0x0, %%wstate\n\t"					\
 	"ldx	[%%sp + 2047 + 0x70], %%i6\n\t"					\
@@ -210,13 +210,13 @@ do {	CHECK_LOCKS(prev);							\
 	"wrpr	%%g0, 0x94, %%pstate\n\t"					\
 	"mov	%%l2, %%g6\n\t"							\
 	"wrpr	%%g0, 0x96, %%pstate\n\t"					\
-	"andcc	%%o7, %6, %%g0\n\t"						\
+	"andcc	%%o7, %5, %%g0\n\t"						\
 	"bne,pn	%%icc, ret_from_syscall\n\t"					\
-	" ldx	[%%g5 + %7], %0\n\t"						\
-	: "=&r" (last)								\
+	" nop\n\t"								\
+	: /* no outputs */							\
 	: "r" (next->thread_info),						\
 	  "i" (TI_WSTATE), "i" (TI_KSP), "i" (TI_FLAGS), "i" (TI_CWP),		\
-	  "i" (_TIF_NEWCHILD), "i" (TI_TASK)					\
+	  "i" (_TIF_NEWCHILD)							\
 	: "cc", "g1", "g2", "g3", "g5", "g7",					\
 	  "l2", "l3", "l4", "l5", "l6", "l7",					\
 	  "i0", "i1", "i2", "i3", "i4", "i5",					\