Merge samba.org:/scratch/anton/linux-2.5

into samba.org:/scratch/anton/tmp3

arch/ppc64/boot/Makefile

@@ -98,7 +98,7 @@ $(call gz-sec, $(required)): $(obj)/kernel-%.gz: % FORCE
 $(obj)/kernel-initrd.gz: $(obj)/ramdisk.image.gz
 	cp -f $(obj)/ramdisk.image.gz $@
 
-$(call src-sec, $(required) $(initrd)): $(obj)/kernel-%.c: $(obj)/kernel-%.gz
+$(call src-sec, $(required) $(initrd)): $(obj)/kernel-%.c: $(obj)/kernel-%.gz FORCE
 	touch $@
 
 $(call obj-sec, $(required) $(initrd)): $(obj)/kernel-%.o: $(obj)/kernel-%.c FORCE

arch/ppc64/kernel/head.S

@@ -596,7 +596,7 @@ DataAccess_common:
 	cmpi    0,r22,0xc
 
 	/* Segment fault on a bolted segment. Go off and map that segment. */
-	beq     .do_stab_bolted
+	beq-	.do_stab_bolted
 stab_bolted_user_return:
 	EXCEPTION_PROLOG_COMMON
 	ld      r3,_DSISR(r1)
@@ -606,7 +606,7 @@ stab_bolted_user_return:
 	rlwinm	r4,r3,32-23,29,29	/* DSISR_STORE -> _PAGE_RW */
 	ld      r3,_DAR(r1)             /* into the hash table */
 
-	beq	2f			/* If so handle it */
+	beq+	2f			/* If so handle it */
 	li	r4,0x300                /* Trap number */
 	bl	.do_stab_SI
 	b	1f
@@ -658,7 +658,7 @@ InstructionAccess_common:
 	EXCEPTION_PROLOG_COMMON
 
 	andis.	r0,r23,0x0020		/* no ste found? */
-	beq	2f
+	beq+	2f
 	mr	r3,r22			/* SRR0 at interrupt */
 	li	r4,0x400		/* Trap number       */
 	bl	.do_stab_SI
@@ -688,7 +688,7 @@ InstructionAccessSLB_common:
 	li	r4,0x480                /* Exception vector  */
 	bl	.ste_allocate
 	or.	r3,r3,r3		/* Check return code */
-	beq     fast_exception_return   /* Return if we succeeded */
+	beq+	fast_exception_return   /* Return if we succeeded */
 
 	addi	r3,r1,STACK_FRAME_OVERHEAD
 #ifdef DO_SOFT_DISABLE
@@ -897,18 +897,8 @@ _GLOBAL(do_stab_bolted)
 
 	mfspr   r22,DSISR
 	andis.  r22,r22,0x0020
-	bne+    2f
-	ld	r22,8(r21)	/* get SRR1 */
-	andi.	r22,r22,MSR_PR	/* check if from user */
-	bne+	stab_bolted_user_return  /* from user, send the error on up */
-#if 0
-	li	r3,0
-#ifdef CONFIG_XMON
-	bl	.xmon
-#endif
-1:	b	1b
-#endif
-2:
+	beq-	stab_bolted_user_return
+
 	/* (((ea >> 28) & 0x1fff) << 15) | (ea >> 60) */
 	mfspr	r21,DAR
 	rldicl  r20,r21,36,32   /* Permits a full 32b of ESID */
@@ -1106,9 +1096,11 @@ SLB_NUM_ENTRIES = 64
 	oris    r21,r23,2048   /* valid bit    */
 	rldimi  r21,r22,0,52   /* Insert entry */
 
-	isync
+	/* 
+	 * No need for an isync before or after this slbmte. The exception
+	 * we enter with and the rfid we exit with are context synchronizing .
+	 */
 	slbmte  r20,r21
-	isync
 
 	/* All done -- return from exception. */
 	mfsprg  r20,3                   /* Load the PACA pointer  */
@@ -1861,9 +1853,9 @@ _STATIC(start_here_common)
 	li	r5,0
 	std	r0,PACAKSAVE(r13)
 
-	/* ptr to hardware interrupt stack for processor 0                */
+	/* ptr to hardware interrupt stack for boot processor */
 	LOADADDR(r3, hardware_int_paca0)
-	li      r5,0x1000
+	li      r5,PAGE_SIZE
 	sldi    r5,r5,3
 	subi    r5,r5,STACK_FRAME_OVERHEAD
 
@@ -1991,7 +1983,7 @@ ioremap_dir:
 
 	.globl  hardware_int_paca0
 hardware_int_paca0:
-	.space	8*4096
+	.space	8*PAGE_SIZE
 
 /* 1 page segment table per cpu (max 48, cpu0 allocated at STAB0_PHYS_ADDR) */
 	.globl	stab_array

arch/ppc64/kernel/iSeries_setup.c

@@ -864,7 +864,7 @@ static void iSeries_setup_dprofile(void)
 {
 	if ( dprof_buffer ) {
 		unsigned i;
-		for (i=0; i<MAX_PACAS; ++i) {
+		for (i=0; i<NR_CPUS; ++i) {
 			paca[i].prof_shift = dprof_shift;
 			paca[i].prof_len = dprof_len-1;
 			paca[i].prof_buffer = dprof_buffer;

arch/ppc64/kernel/irq.c

@@ -381,22 +381,48 @@ int show_interrupts(struct seq_file *p, void *v)
 	return 0;
 }
 
-static inline void
-handle_irq_event(int irq, struct pt_regs *regs, struct irqaction *action)
+extern char *ppc_find_proc_name(unsigned *p, char *buf, unsigned buflen);
+
+static inline void handle_irq_event(int irq, struct pt_regs *regs,
+				    struct irqaction *action)
 {
 	int status = 0;
+	int retval = 0;
+	struct irqaction *first_action = action;
 
 	if (!(action->flags & SA_INTERRUPT))
 		local_irq_enable();
 
 	do {
 		status |= action->flags;
-		action->handler(irq, action->dev_id, regs);
+		retval |= action->handler(irq, action->dev_id, regs);
 		action = action->next;
 	} while (action);
 	if (status & SA_SAMPLE_RANDOM)
 		add_interrupt_randomness(irq);
 	local_irq_disable();
+	if (retval != 1) {
+		static int count = 100;
+		char name_buf[256];
+		if (count) {
+			count--;
+			if (retval) {
+				printk("irq event %d: bogus retval mask %x\n",
+					irq, retval);
+			} else {
+				printk("irq %d: nobody cared!\n", irq);
+			}
+			dump_stack();
+			printk("handlers:\n");
+			action = first_action;
+			do {
+				printk("[<%p>]", action->handler);
+				printk(" (%s)\n",
+				       ppc_find_proc_name((unsigned *)action->handler, name_buf, 256));
+				action = action->next;
+			} while (action);
+		}
+	}
 }
 
 /*
@@ -676,7 +702,7 @@ static int prof_cpu_mask_write_proc (struct file *file, const char *buffer,
 #ifdef CONFIG_PPC_ISERIES
 	{
 		unsigned i;
-		for (i=0; i<MAX_PACAS; ++i) {
+		for (i=0; i<NR_CPUS; ++i) {
 			if ( paca[i].prof_buffer && (new_value & 1) )
 				paca[i].prof_enabled = 1;
 			else

arch/ppc64/kernel/misc.S

@@ -609,7 +609,7 @@ _GLOBAL(sys_call_table32)
 	.llong .sys_ni_syscall		/* old profil syscall */
 	.llong .compat_sys_statfs
 	.llong .compat_sys_fstatfs		/* 100 */
-	.llong .sys_ioperm
+	.llong .sys_ni_syscall		/* old ioperm syscall */
 	.llong .compat_sys_socketcall
 	.llong .sys32_syslog
 	.llong .compat_sys_setitimer
@@ -852,7 +852,7 @@ _GLOBAL(sys_call_table)
 	.llong .sys_ni_syscall		/* old profil syscall holder */
 	.llong .sys_statfs
 	.llong .sys_fstatfs		/* 100 */
-	.llong .sys_ioperm
+	.llong .sys_ni_syscall		/* old ioperm syscall */
 	.llong .sys_socketcall
 	.llong .sys_syslog
 	.llong .sys_setitimer

arch/ppc64/kernel/pacaData.c

@@ -64,7 +64,7 @@ struct systemcfg *systemcfg;
 		(&paca[number].exception_stack[0]) - EXC_FRAME_SIZE,	    \
 }
 
-struct paca_struct paca[MAX_PACAS] __page_aligned = {
+struct paca_struct paca[NR_CPUS] __page_aligned = {
 #ifdef CONFIG_PPC_ISERIES
 	PACAINITDATA( 0, 1, &xItLpQueue, 0, STAB0_VIRT_ADDR),
 #else
@@ -101,6 +101,7 @@ struct paca_struct paca[MAX_PACAS] __page_aligned = {
 	PACAINITDATA(29, 0, 0, 0, 0),
 	PACAINITDATA(30, 0, 0, 0, 0),
 	PACAINITDATA(31, 0, 0, 0, 0),
+#if NR_CPUS > 32
 	PACAINITDATA(32, 0, 0, 0, 0),
 	PACAINITDATA(33, 0, 0, 0, 0),
 	PACAINITDATA(34, 0, 0, 0, 0),
@@ -116,5 +117,22 @@ struct paca_struct paca[MAX_PACAS] __page_aligned = {
 	PACAINITDATA(44, 0, 0, 0, 0),
 	PACAINITDATA(45, 0, 0, 0, 0),
 	PACAINITDATA(46, 0, 0, 0, 0),
-	PACAINITDATA(47, 0, 0, 0, 0)
+	PACAINITDATA(47, 0, 0, 0, 0),
+	PACAINITDATA(48, 0, 0, 0, 0),
+	PACAINITDATA(49, 0, 0, 0, 0),
+	PACAINITDATA(50, 0, 0, 0, 0),
+	PACAINITDATA(51, 0, 0, 0, 0),
+	PACAINITDATA(52, 0, 0, 0, 0),
+	PACAINITDATA(53, 0, 0, 0, 0),
+	PACAINITDATA(54, 0, 0, 0, 0),
+	PACAINITDATA(55, 0, 0, 0, 0),
+	PACAINITDATA(56, 0, 0, 0, 0),
+	PACAINITDATA(57, 0, 0, 0, 0),
+	PACAINITDATA(58, 0, 0, 0, 0),
+	PACAINITDATA(59, 0, 0, 0, 0),
+	PACAINITDATA(60, 0, 0, 0, 0),
+	PACAINITDATA(61, 0, 0, 0, 0),
+	PACAINITDATA(62, 0, 0, 0, 0),
+	PACAINITDATA(63, 0, 0, 0, 0),
+#endif
 };

arch/ppc64/kernel/pci.c

@@ -124,7 +124,7 @@ struct pci_dev *pci_find_dev_by_addr(unsigned long addr)
 	ioaddr = (addr > isa_io_base) ? addr - isa_io_base : 0;
 
 	pci_for_each_dev(dev) {
-		if ((dev->class >> 8) == PCI_BASE_CLASS_BRIDGE)
+		if ((dev->class >> 16) == PCI_BASE_CLASS_BRIDGE)
 			continue;
 		for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
 			unsigned long start = pci_resource_start(dev,i);

arch/ppc64/kernel/pci_dma.c

@@ -710,12 +710,6 @@ void create_tce_tables_for_busesLP(struct list_head *bus_list)
 	for (ln=bus_list->next; ln != bus_list; ln=ln->next) {
 		bus = pci_bus_b(ln);
 		busdn = PCI_GET_DN(bus);
-		/* NOTE: there should never be a window declared on a bus when
-		 * child devices also have a window.  If this should ever be
-		 * architected, we probably want children to have priority.
-		 * In reality, the PHB containing ISA has the property, but otherwise
-		 * it is the pci-bridges that have the property.
-		 */
 		dma_window = (u32 *)get_property(busdn, "ibm,dma-window", 0);
 		if (dma_window) {
 			/* Bussubno hasn't been copied yet.
@@ -724,6 +718,7 @@ void create_tce_tables_for_busesLP(struct list_head *bus_list)
 			busdn->bussubno = bus->number;
 			create_pci_bus_tce_table((unsigned long)busdn);
 		}
+		/* look for a window on a bridge even if the PHB had one */
 		create_tce_tables_for_busesLP(&bus->children);
 	}
 }

arch/ppc64/kernel/process.c

@@ -67,18 +67,34 @@ enable_kernel_fp(void)
 #endif /* CONFIG_SMP */
 }
 
-int
-dump_fpu(struct pt_regs *regs, elf_fpregset_t *fpregs)
+#ifdef CONFIG_SMP
+static void smp_unlazy_onefpu(void *arg)
 {
-	/*
-	 * XXX temporary workaround until threaded coredumps for ppc64
-	 * are implemented - Anton
-	 */
+	struct pt_regs *regs = current->thread.regs;
+
 	if (!regs)
-		return 0;
+		return;
 	if (regs->msr & MSR_FP)
 		giveup_fpu(current);
-	memcpy(fpregs, &current->thread.fpr[0], sizeof(*fpregs));
+}
+
+void dump_smp_unlazy_fpu(void)
+{
+	smp_call_function(smp_unlazy_onefpu, NULL, 1, 1);
+}
+#endif
+
+int dump_task_fpu(struct task_struct *tsk, elf_fpregset_t *fpregs)
+{
+	struct pt_regs *regs = tsk->thread.regs;
+
+	if (!regs)
+		return 0;
+	if (tsk == current && (regs->msr & MSR_FP))
+		giveup_fpu(current);
+
+	memcpy(fpregs, &tsk->thread.fpr[0], sizeof(*fpregs));
+
 	return 1;
 }
 
@@ -113,7 +129,7 @@ struct task_struct *__switch_to(struct task_struct *prev,
 }
 
 static void show_tsk_stack(struct task_struct *p, unsigned long sp);
-static char *ppc_find_proc_name(unsigned *p, char *buf, unsigned buflen);
+char *ppc_find_proc_name(unsigned *p, char *buf, unsigned buflen);
 
 void show_regs(struct pt_regs * regs)
 {
@@ -410,7 +426,7 @@ void initialize_paca_hardware_interrupt_stack(void)
 
 extern char _stext[], _etext[], __init_begin[], __init_end[];
 
-static char *ppc_find_proc_name(unsigned *p, char *buf, unsigned buflen)
+char *ppc_find_proc_name(unsigned *p, char *buf, unsigned buflen)
 {
 	unsigned long tb_flags;
 	unsigned short name_len;

arch/ppc64/kernel/prom.c

@@ -1134,7 +1134,7 @@ prom_init(unsigned long r3, unsigned long r4, unsigned long pp,
 	_prom->cpu = (int)(unsigned long)getprop_rval;
 	_xPaca[_prom->cpu].active = 1;
 #ifdef CONFIG_SMP
-	RELOC(cpu_online_map) = 1 << _prom->cpu;
+	RELOC(cpu_online_map) = 1UL << _prom->cpu;
 #endif
 	RELOC(boot_cpuid) = _prom->cpu;
 

arch/ppc64/kernel/setup.c

@@ -194,7 +194,7 @@ void setup_system(unsigned long r3, unsigned long r4, unsigned long r5,
 	printk("naca->pftSize                 = 0x%lx\n", naca->pftSize);
 	printk("naca->debug_switch            = 0x%lx\n", naca->debug_switch);
 	printk("naca->interrupt_controller    = 0x%ld\n", naca->interrupt_controller);
-	printk("systemcf                      = 0x%p\n", systemcfg);
+	printk("systemcfg                      = 0x%p\n", systemcfg);
 	printk("systemcfg->processorCount     = 0x%lx\n", systemcfg->processorCount);
 	printk("systemcfg->physicalMemorySize = 0x%lx\n", systemcfg->physicalMemorySize);
 	printk("systemcfg->dCacheL1LineSize   = 0x%x\n", systemcfg->dCacheL1LineSize);
@@ -256,7 +256,7 @@ static int show_cpuinfo(struct seq_file *m, void *v)
 		return 0;
 	}
 
-	if (!(cpu_online_map & (1<<cpu_id)))
+	if (!(cpu_online_map & (1UL << cpu_id)))
 		return 0;
 
 #ifdef CONFIG_SMP
@@ -584,7 +584,7 @@ int set_spread_lpevents( char * str )
 	/* The parameter is the number of processors to share in processing lp events */
 	unsigned long i;
 	unsigned long val = simple_strtoul( str, NULL, 0 );
-	if ( ( val > 0 ) && ( val <= MAX_PACAS ) ) {
+	if ( ( val > 0 ) && ( val <= NR_CPUS ) ) {
 		for ( i=1; i<val; ++i )
 			paca[i].lpQueuePtr = paca[0].lpQueuePtr;
 		printk("lpevent processing spread over %ld processors\n", val);

arch/ppc64/kernel/signal.c

@@ -69,13 +69,11 @@ struct rt_sigframe {
 };
 
 
-extern int do_signal(sigset_t *oldset, struct pt_regs *regs);
-
 /*
  * Atomically swap in the new signal mask, and wait for a signal.
  */
-long sys_rt_sigsuspend(sigset_t *unewset, size_t sigsetsize, int p3, int p4, int p6,
-		  int p7, struct pt_regs *regs)
+long sys_rt_sigsuspend(sigset_t *unewset, size_t sigsetsize, int p3, int p4,
+		       int p6, int p7, struct pt_regs *regs)
 {
 	sigset_t saveset, newset;
 
@@ -390,8 +388,6 @@ syscall_restart(struct pt_regs *regs, struct k_sigaction *ka)
  * want to handle. Thus you cannot kill init even with a SIGKILL even by
  * mistake.
  */
-extern int do_signal32(sigset_t *oldset, struct pt_regs *regs);
-
 int do_signal(sigset_t *oldset, struct pt_regs *regs)
 {
 	siginfo_t info;

arch/ppc64/kernel/signal32.c

@@ -563,10 +563,6 @@ long sys32_rt_sigaction(int sig, const struct sigaction32 *act,
 	return ret;
 }
 
-
-extern long sys_rt_sigprocmask(int how, sigset_t *set,
-		sigset_t *oset, size_t sigsetsize);
-
 /*
  * Note: it is necessary to treat how as an unsigned int, with the
  * corresponding cast to a signed int to insure that the proper
@@ -613,10 +609,6 @@ long sys32_rt_sigprocmask(u32 how, compat_sigset_t *set,
 	return 0;
 }
 
-
-extern long sys_rt_sigpending(sigset_t *set, size_t sigsetsize);
-
-
 long sys32_rt_sigpending(compat_sigset_t *set, compat_size_t sigsetsize)
 {
 	sigset_t s;
@@ -683,11 +675,6 @@ static int copy_siginfo_to_user32(siginfo_t32 *d, siginfo_t *s)
 	return err;
 }
 
-
-extern long sys_rt_sigtimedwait(const sigset_t *uthese,
-		siginfo_t *uinfo, const struct timespec *uts,
-		size_t sigsetsize);
-
 long sys32_rt_sigtimedwait(compat_sigset_t *uthese, siginfo_t32 *uinfo,
 		struct compat_timespec *uts, compat_size_t sigsetsize)
 {
@@ -758,9 +745,6 @@ static siginfo_t * siginfo32to64(siginfo_t *d, siginfo_t32 *s)
 	return d;
 }
 
-
-extern long sys_rt_sigqueueinfo(int pid, int sig, siginfo_t *uinfo);
-
 /*
  * Note: it is necessary to treat pid and sig as unsigned ints, with the
  * corresponding cast to a signed int to insure that the proper conversion
@@ -786,8 +770,6 @@ long sys32_rt_sigqueueinfo(u32 pid, u32 sig, siginfo_t32 *uinfo)
 	return ret;
 }
 
-extern int do_signal(sigset_t *oldset, struct pt_regs *regs);
-
 int sys32_rt_sigsuspend(compat_sigset_t* unewset, size_t sigsetsize, int p3,
 		int p4, int p6, int p7, struct pt_regs *regs)
 {

arch/ppc64/kernel/smp.c

@@ -113,7 +113,7 @@ static int smp_iSeries_numProcs(void)
 	struct ItLpPaca * lpPaca;
 
 	np = 0;
-        for (i=0; i < MAX_PACAS; ++i) {
+        for (i=0; i < NR_CPUS; ++i) {
                 lpPaca = paca[i].xLpPacaPtr;
                 if ( lpPaca->xDynProcStatus < 2 ) {
                         ++np;
@@ -128,7 +128,7 @@ static int smp_iSeries_probe(void)
 	unsigned np = 0;
 	struct ItLpPaca *lpPaca;
 
-	for (i=0; i < MAX_PACAS; ++i) {
+	for (i=0; i < NR_CPUS; ++i) {
 		lpPaca = paca[i].xLpPacaPtr;
 		if (lpPaca->xDynProcStatus < 2) {
 			paca[i].active = 1;
@@ -144,7 +144,7 @@ static void smp_iSeries_kick_cpu(int nr)
 	struct ItLpPaca * lpPaca;
 	/* Verify we have a Paca for processor nr */
 	if ( ( nr <= 0 ) ||
-	     ( nr >= MAX_PACAS ) )
+	     ( nr >= NR_CPUS ) )
 		return;
 	/* Verify that our partition has a processor nr */
 	lpPaca = paca[nr].xLpPacaPtr;
@@ -228,7 +228,7 @@ smp_kick_cpu(int nr)
 {
 	/* Verify we have a Paca for processor nr */
 	if ( ( nr <= 0 ) ||
-	     ( nr >= MAX_PACAS ) )
+	     ( nr >= NR_CPUS ) )
 		return;
 
 	/* The information for processor bringup must

arch/ppc64/kernel/stab.c

@@ -23,7 +23,8 @@
 #include <asm/pmc.h>
 
 int make_ste(unsigned long stab, unsigned long esid, unsigned long vsid);
-void make_slbe(unsigned long esid, unsigned long vsid, int large);
+void make_slbe(unsigned long esid, unsigned long vsid, int large,
+	       int kernel_segment);
 
 /*
  * Build an entry for the base kernel segment and put it into
@@ -45,7 +46,8 @@ void stab_initialize(unsigned long stab)
 		asm volatile("isync":::"memory");
 		asm volatile("slbmte  %0,%0"::"r" (0) : "memory");
 		asm volatile("isync; slbia; isync":::"memory");
-		make_slbe(esid, vsid, 0);
+		make_slbe(esid, vsid, 0, 1);
+		asm volatile("isync":::"memory");
 #endif
 	} else {
 		asm volatile("isync; slbia; isync":::"memory");
@@ -139,10 +141,13 @@ int make_ste(unsigned long stab, unsigned long esid, unsigned long vsid)
 
 /*
  * Create a segment buffer entry for the given esid/vsid pair.
+ *
+ * NOTE: A context syncronising instruction is required before and after
+ * this, in the common case we use exception entry and rfid.
  */
-void make_slbe(unsigned long esid, unsigned long vsid, int large)
+void make_slbe(unsigned long esid, unsigned long vsid, int large,
+	       int kernel_segment)
 {
-	int kernel_segment = 0;
 	unsigned long entry, castout_entry;
 	union {
 		unsigned long word0;
@@ -153,42 +158,15 @@ void make_slbe(unsigned long esid, unsigned long vsid, int large)
 		slb_dword1    data;
 	} vsid_data;
 
-	if (REGION_ID(esid << SID_SHIFT) >= KERNEL_REGION_ID)
-		kernel_segment = 1;
-
 	/*
-	 * Find an empty entry, if one exists.
+	 * Find an empty entry, if one exists. Must start at 0 because
+	 * we use this code to load SLB entry 0 at boot.
 	 */
 	for (entry = 0; entry < naca->slb_size; entry++) {
 		asm volatile("slbmfee  %0,%1" 
 			     : "=r" (esid_data) : "r" (entry)); 
-		if (!esid_data.data.v) {
-			/* 
-			 * Write the new SLB entry.
-			 */
-			vsid_data.word0 = 0;
-			vsid_data.data.vsid = vsid;
-			vsid_data.data.kp = 1;
-			if (large)
-				vsid_data.data.l = 1;
-			if (kernel_segment)
-				vsid_data.data.c = 1;
-
-			esid_data.word0 = 0;
-			esid_data.data.esid = esid;
-			esid_data.data.v = 1;
-			esid_data.data.index = entry;
-
-			/* slbie not needed as no previous mapping existed. */
-			/* Order update  */
-			asm volatile("isync" : : : "memory");
-			asm volatile("slbmte  %0,%1" 
-				     : : "r" (vsid_data), 
-				     "r" (esid_data)); 
-			/* Order update  */
-			asm volatile("isync" : : : "memory");
-			return;
-		}
+		if (!esid_data.data.v)
+			goto write_entry;
 	}
 
 	/*
@@ -211,13 +189,13 @@ void make_slbe(unsigned long esid, unsigned long vsid, int large)
 		if (castout_entry >= naca->slb_size)
 			castout_entry = 1; 
 		asm volatile("slbmfee  %0,%1" : "=r" (esid_data) : "r" (entry));
-	} while (esid_data.data.esid == GET_ESID((unsigned long)_get_SP()) &&
-			esid_data.data.v);
+	} while (esid_data.data.esid == GET_ESID((unsigned long)_get_SP()));
 
 	get_paca()->xStab_data.next_round_robin = castout_entry;
 
 	/* slbie not needed as the previous mapping is still valid. */
 
+write_entry:	
 	/* 
 	 * Write the new SLB entry.
 	 */
@@ -234,10 +212,11 @@ void make_slbe(unsigned long esid, unsigned long vsid, int large)
 	esid_data.data.v = 1;
 	esid_data.data.index = entry;
 
-	asm volatile("isync" : : : "memory");   /* Order update */
-	asm volatile("slbmte  %0,%1" 
-		     : : "r" (vsid_data), "r" (esid_data)); 
-	asm volatile("isync" : : : "memory" );   /* Order update */
+	/*
+	 * No need for an isync before or after this slbmte. The exception
+         * we enter with and the rfid we exit with are context synchronizing.
+	 */
+	asm volatile("slbmte  %0,%1" : : "r" (vsid_data), "r" (esid_data)); 
 }
 
 static inline void __ste_allocate(unsigned long esid, unsigned long vsid,
@@ -246,10 +225,10 @@ static inline void __ste_allocate(unsigned long esid, unsigned long vsid,
 	if (cpu_has_slb()) {
 #ifndef CONFIG_PPC_ISERIES
 		if (REGION_ID(esid << SID_SHIFT) == KERNEL_REGION_ID)
-			make_slbe(esid, vsid, 1); 
+			make_slbe(esid, vsid, 1, kernel_segment); 
 		else
 #endif
-			make_slbe(esid, vsid, 0); 
+			make_slbe(esid, vsid, 0, kernel_segment);
 	} else {
 		unsigned char top_entry, stab_entry, *segments; 
 

arch/ppc64/kernel/syscalls.c

@@ -50,11 +50,6 @@ check_bugs(void)
 {
 }
 
-int sys_ioperm(unsigned long from, unsigned long num, int on)
-{
-	return -EIO;
-}
-
 /*
  * sys_ipc() is the de-multiplexer for the SysV IPC calls..
  *

arch/ppc64/kernel/traps.c

@@ -349,9 +349,8 @@ ProgramCheckException(struct pt_regs *regs)
 void
 KernelFPUnavailableException(struct pt_regs *regs)
 {
-	printk("Illegal floating point used in kernel "
-			"(task=0x%p, pc=0x%016lx, trap=0x%08lx)\n",
-		current, regs->nip, regs->trap);
+	printk("Illegal floating point used in kernel (task=0x%p, "
+		"pc=0x%016lx, trap=0x%lx)\n", current, regs->nip, regs->trap);
 	panic("Unrecoverable FP Unavailable Exception in Kernel");
 }
 

include/asm-ppc64/elf.h

@@ -98,17 +98,38 @@ typedef elf_fpreg_t elf_fpregset_t[ELF_NFPREG];
 #define ELF_ET_DYN_BASE         (0x08000000)
 
 /* Common routine for both 32-bit and 64-bit processes */
-#define ELF_CORE_COPY_REGS(gregs, regs) ppc64_elf_core_copy_regs(gregs, regs);
-static inline void
-ppc64_elf_core_copy_regs(elf_gregset_t dstRegs, struct pt_regs* srcRegs)
+static inline void ppc64_elf_core_copy_regs(elf_gregset_t elf_regs,
+					    struct pt_regs *regs)
 {
 	int i;
+	int gprs = sizeof(struct pt_regs)/sizeof(elf_greg_t64);
 
-	int numGPRS = ((sizeof(struct pt_regs)/sizeof(elf_greg_t64)) < ELF_NGREG) ? (sizeof(struct pt_regs)/sizeof(elf_greg_t64)) : ELF_NGREG;
+	if (gprs > ELF_NGREG)
+		gprs = ELF_NGREG;
 
-	for (i=0; i < numGPRS; i++)
-		dstRegs[i] = (elf_greg_t)((elf_greg_t64 *)srcRegs)[i];
+	for (i=0; i < gprs; i++)
+		elf_regs[i] = (elf_greg_t)((elf_greg_t64 *)regs)[i];
 }
+#define ELF_CORE_COPY_REGS(gregs, regs) ppc64_elf_core_copy_regs(gregs, regs);
+
+static inline int dump_task_regs(struct task_struct *tsk,
+				 elf_gregset_t *elf_regs)
+{
+	struct pt_regs *regs = tsk->thread.regs;
+	if (regs)
+		ppc64_elf_core_copy_regs(*elf_regs, regs);
+
+	return 1;
+}
+#define ELF_CORE_COPY_TASK_REGS(tsk, elf_regs) dump_task_regs(tsk, elf_regs)
+
+extern int dump_task_fpu(struct task_struct *, elf_fpregset_t *); 
+#define ELF_CORE_COPY_FPREGS(tsk, elf_fpregs) dump_task_fpu(tsk, elf_fpregs)
+
+#ifdef CONFIG_SMP
+extern void dump_smp_unlazy_fpu(void);
+#define ELF_CORE_SYNC dump_smp_unlazy_fpu
+#endif
 
 /* This yields a mask that user programs can use to figure out what
    instruction set this cpu supports.  This could be done in userspace,

include/asm-ppc64/paca.h

@@ -33,17 +33,6 @@
 #include	<asm/mmu.h>
 #include	<asm/processor.h>
 
-/* A paca entry is required for each logical processor.  On systems
- * that support hardware multi-threading, this is equal to twice the
- * number of physical processors.  On LPAR systems, we are required
- * to have space for the maximum number of logical processors we
- * could ever possibly have.  Currently, we are limited to allocating
- * 24 processors to a partition which gives 48 logical processors on
- * an HMT box.  Therefore, we reserve this many paca entries.
- */
-#define MAX_PROCESSORS 24
-#define MAX_PACAS MAX_PROCESSORS * 2
-
 extern struct paca_struct paca[];
 register struct paca_struct *local_paca asm("r13");
 #define get_paca()	local_paca

include/asm-ppc64/pci-bridge.h

@@ -77,7 +77,7 @@ struct device_node *fetch_dev_dn(struct pci_dev *dev);
 static inline struct device_node *pci_device_to_OF_node(struct pci_dev *dev)
 {
 	struct device_node *dn = (struct device_node *)(dev->sysdata);
-	if (dn->devfn == dev->devfn && dn->busno == dev->bus->number)
+	if (dn->devfn == dev->devfn && dn->busno == (dev->bus->number&0xff))
 		return dn;	/* fast path.  sysdata is good */
 	else
 		return fetch_dev_dn(dev);

include/asm-ppc64/signal.h

@@ -2,6 +2,7 @@
 #define _ASMPPC64_SIGNAL_H
 
 #include <linux/types.h>
+#include <asm/siginfo.h>
 
 /* Avoid too many header ordering problems.  */
 struct siginfo;
@@ -72,19 +73,19 @@ typedef struct {
  * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
  * Unix names RESETHAND and NODEFER respectively.
  */
-#define SA_NOCLDSTOP	0x00000001
-#define SA_NOCLDWAIT	0x00000002
-#define SA_SIGINFO	0x00000004
-#define SA_ONSTACK	0x08000000
-#define SA_RESTART	0x10000000
-#define SA_NODEFER	0x40000000
-#define SA_RESETHAND	0x80000000
+#define SA_NOCLDSTOP	0x00000001u
+#define SA_NOCLDWAIT	0x00000002u
+#define SA_SIGINFO	0x00000004u
+#define SA_ONSTACK	0x08000000u
+#define SA_RESTART	0x10000000u
+#define SA_NODEFER	0x40000000u
+#define SA_RESETHAND	0x80000000u
 
 #define SA_NOMASK	SA_NODEFER
 #define SA_ONESHOT	SA_RESETHAND
-#define SA_INTERRUPT	0x20000000 /* dummy -- ignored */
+#define SA_INTERRUPT	0x20000000u /* dummy -- ignored */
 
-#define SA_RESTORER	0x04000000
+#define SA_RESTORER	0x04000000u
 
 /* 
  * sigaltstack controls
@@ -143,6 +144,16 @@ typedef struct sigaltstack {
 	size_t ss_size;
 } stack_t;
 
+struct pt_regs;
+struct timespec;
+extern int do_signal(sigset_t *oldset, struct pt_regs *regs);
+extern int do_signal32(sigset_t *oldset, struct pt_regs *regs);
+extern long sys_rt_sigprocmask(int how, sigset_t *set, sigset_t *oset,
+			       size_t sigsetsize);
+extern long sys_rt_sigpending(sigset_t *set, size_t sigsetsize);
+extern long sys_rt_sigtimedwait(const sigset_t *uthese, siginfo_t *uinfo,
+				const struct timespec *uts, size_t sigsetsize);
+extern long sys_rt_sigqueueinfo(int pid, int sig, siginfo_t *uinfo);
 #define ptrace_signal_deliver(regs, cookie) do { } while (0)
 
 struct pt_regs;

include/asm-ppc64/system.h

@@ -83,6 +83,7 @@ extern void show_regs(struct pt_regs * regs);
 extern void flush_instruction_cache(void);
 extern int _get_PVR(void);
 extern void giveup_fpu(struct task_struct *);
+extern void disable_kernel_fp(void);
 extern void enable_kernel_fp(void);
 extern void cvt_fd(float *from, double *to, unsigned long *fpscr);
 extern void cvt_df(double *from, float *to, unsigned long *fpscr);

include/asm-ppc64/uaccess.h

@@ -206,6 +206,7 @@ copy_from_user(void *to, const void *from, unsigned long n)
 		return __copy_tofrom_user(to, from, n);
 	if ((unsigned long)from < TASK_SIZE) {
 		over = (unsigned long)from + n - TASK_SIZE;
+		memset(to + over, 0, over);
 		return __copy_tofrom_user(to, from, n - over) + over;
 	}
 	return n;