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

into samba.org:/scratch/anton/linux-2.5_work

arch/ppc64/defconfig

@@ -487,10 +487,26 @@ CONFIG_VIOPATH=y
 #
 CONFIG_VT=y
 CONFIG_VT_CONSOLE=y
-CONFIG_SERIAL=y
-CONFIG_SERIAL_CONSOLE=y
-# CONFIG_SERIAL_EXTENDED is not set
 # CONFIG_SERIAL_NONSTANDARD is not set
+
+#
+# Serial drivers
+#
+CONFIG_SERIAL_8250=y
+CONFIG_SERIAL_8250_CONSOLE=y
+# CONFIG_SERIAL_8250_CS is not set
+# CONFIG_SERIAL_8250_EXTENDED is not set
+# CONFIG_SERIAL_8250_MANY_PORTS is not set
+# CONFIG_SERIAL_8250_SHARE_IRQ is not set
+# CONFIG_SERIAL_8250_DETECT_IRQ is not set
+# CONFIG_SERIAL_8250_MULTIPORT is not set
+# CONFIG_SERIAL_8250_RSA is not set
+
+#
+# Non-8250 serial port support
+#
+CONFIG_SERIAL_CORE=y
+CONFIG_SERIAL_CORE_CONSOLE=y
 CONFIG_UNIX98_PTYS=y
 CONFIG_UNIX98_PTY_COUNT=256
 CONFIG_HVC_CONSOLE=y

arch/ppc64/kernel/i8259.c

@@ -123,7 +123,8 @@ static void i8259_unmask_irq(unsigned int irq_nr)
 
 static void i8259_end_irq(unsigned int irq)
 {
-	if (!(irq_desc[irq].status & (IRQ_DISABLED|IRQ_INPROGRESS)))
+	if (!(irq_desc[irq].status & (IRQ_DISABLED|IRQ_INPROGRESS)) &&
+	    irq_desc[irq].action)
 		i8259_unmask_irq(irq);
 }
 

arch/ppc64/kernel/irq.c

@@ -169,10 +169,8 @@ setup_irq(unsigned int irq, struct irqaction * new)
 
 inline void synchronize_irq(unsigned int irq)
 {
-	while (irq_desc[irq].status & IRQ_INPROGRESS) {
-		barrier();
+	while (irq_desc[irq].status & IRQ_INPROGRESS)
 		cpu_relax();
-	}
 }
 
 #endif /* CONFIG_SMP */
@@ -500,7 +498,7 @@ void ppc_irq_dispatch_handler(struct pt_regs *regs, int irq)
 	 * use the action we have.
 	 */
 	action = NULL;
-	if (!(status & (IRQ_DISABLED | IRQ_INPROGRESS))) {
+	if (likely(!(status & (IRQ_DISABLED | IRQ_INPROGRESS)))) {
 		action = desc->action;
 		if (!action || !action->handler) {
 			ppc_spurious_interrupts++;
@@ -527,10 +525,9 @@ void ppc_irq_dispatch_handler(struct pt_regs *regs, int irq)
 	   a different instance of this same irq, the other processor
 	   will take care of it.
 	 */
-	if (!action)
+	if (unlikely(!action))
 		goto out;
 
-
 	/*
 	 * Edge triggered interrupts need to remember
 	 * pending events.
@@ -546,12 +543,12 @@ void ppc_irq_dispatch_handler(struct pt_regs *regs, int irq)
 		handle_irq_event(irq, regs, action);
 		spin_lock(&desc->lock);
 		
-		if (!(desc->status & IRQ_PENDING))
+		if (likely(!(desc->status & IRQ_PENDING)))
 			break;
 		desc->status &= ~IRQ_PENDING;
 	}
-	desc->status &= ~IRQ_INPROGRESS;
 out:
+	desc->status &= ~IRQ_INPROGRESS;
 	/*
 	 * The ->end() handler has to deal with interrupts which got
 	 * disabled while the handler was running.
@@ -567,7 +564,6 @@ void ppc_irq_dispatch_handler(struct pt_regs *regs, int irq)
 
 int do_IRQ(struct pt_regs *regs)
 {
-	int cpu = smp_processor_id();
 	int irq, first = 1;
 #ifdef CONFIG_PPC_ISERIES
 	struct paca_struct *lpaca;
@@ -605,7 +601,7 @@ int do_IRQ(struct pt_regs *regs)
 		ppc_spurious_interrupts++;
 #endif
 
-        irq_exit();
+	irq_exit();
 
 #ifdef CONFIG_PPC_ISERIES
 	if (lpaca->xLpPaca.xIntDword.xFields.xDecrInt) {
@@ -615,9 +611,6 @@ int do_IRQ(struct pt_regs *regs)
 	}
 #endif
 
-	if (softirq_pending(cpu))
-		do_softirq();
-
 	return 1; /* lets ret_from_int know we can do checks */
 }
 

arch/ppc64/kernel/misc.S

@@ -683,8 +683,8 @@ _GLOBAL(sys_call_table32)
 	.llong .sys32_rt_sigtimedwait
 	.llong .sys32_rt_sigqueueinfo
 	.llong .sys32_rt_sigsuspend
-	.llong .sys32_pread
-	.llong .sys32_pwrite	        /* 180 */
+	.llong .sys32_pread64
+	.llong .sys32_pwrite64	        /* 180 */
 	.llong .sys_chown
 	.llong .sys_getcwd
 	.llong .sys_capget
@@ -695,7 +695,7 @@ _GLOBAL(sys_call_table32)
 	.llong .sys_ni_syscall		/* streams2 */
 	.llong .sys32_vfork
 	.llong .sys32_getrlimit	        /* 190 */
-	.llong .sys_ni_syscall		/* 191 */	/* Unused */
+	.llong .sys32_readahead
 	.llong .sys_ni_syscall		/* 192 - reserved - mmap2 */
 	.llong .sys32_truncate64	/* 193 - truncate64 */
 	.llong .sys32_ftruncate64	/* 194 - ftruncate64 */
@@ -726,11 +726,12 @@ _GLOBAL(sys_call_table32)
 	.llong .sys_lremovexattr
 	.llong .sys_fremovexattr	/* 220 */
 	.llong .sys_futex
-	.llong .sys_ni_syscall		/* reserved for tux */
 	.llong .sys32_sched_setaffinity
 	.llong .sys32_sched_getaffinity
+	.llong .sys_ni_syscall		/* reserved for security */
+	.llong .sys_ni_syscall		/* 225 - reserved for tux */
 
-	.rept NR_syscalls-224
+	.rept NR_syscalls-225
 		.llong .sys_ni_syscall
 	.endr
 #endif
@@ -928,6 +929,7 @@ _GLOBAL(sys_call_table)
 	.llong .sys_ni_syscall		/* streams2 */
 	.llong .sys_vfork
 	.llong .sys_getrlimit	        /* 190 */
+	.llong .sys_readahead
 	.llong .sys_ni_syscall		/* 191 */	/* Unused */
 	.llong .sys_ni_syscall		/* 192 - reserved - mmap2 */
 	.llong .sys_ni_syscall		/* 193 - reserved - truncate64 */
@@ -959,10 +961,11 @@ _GLOBAL(sys_call_table)
 	.llong .sys_lremovexattr
 	.llong .sys_fremovexattr	/* 220 */
 	.llong .sys_futex
-	.llong .sys_ni_syscall		/* reserved for tux */
 	.llong .sys_sched_setaffinity
 	.llong .sys_sched_getaffinity
+	.llong .sys_ni_syscall		/* reserved for security */
+	.llong .sys_ni_syscall		/* reserved for tux */
 
-	.rept NR_syscalls-224
+	.rept NR_syscalls-225
 	.llong .sys_ni_syscall
 	.endr

arch/ppc64/kernel/open_pic.c

@@ -576,7 +576,7 @@ void openpic_request_IPIs(void)
  */
 static spinlock_t openpic_setup_lock __initdata = SPIN_LOCK_UNLOCKED;
 
-void __init do_openpic_setup_cpu(void)
+void __devinit do_openpic_setup_cpu(void)
 {
 #ifdef CONFIG_IRQ_ALL_CPUS
  	int i;

arch/ppc64/kernel/pSeries_lpar.c

@@ -155,7 +155,7 @@ long plpar_pte_protect(unsigned long flags,
 		       unsigned long ptex,
 		       unsigned long avpn)
 {
-	return plpar_hcall_norets(H_PROTECT, flags, ptex);
+	return plpar_hcall_norets(H_PROTECT, flags, ptex, avpn);
 }
 
 long plpar_tce_get(unsigned long liobn,
@@ -552,6 +552,7 @@ static long pSeries_lpar_insert_hpte(unsigned long hpte_group,
 				     int secondary, unsigned long hpteflags,
 				     int bolted, int large)
 {
+	/* XXX fix for large page */
 	unsigned long avpn = vpn >> 11;
 	unsigned long arpn = physRpn_to_absRpn(prpn);
 	unsigned long lpar_rc;
@@ -651,11 +652,10 @@ static long pSeries_lpar_hpte_updatepp(unsigned long slot, unsigned long newpp,
 				       unsigned long va, int large)
 {
 	unsigned long lpar_rc;
-	unsigned long flags;
-	flags = (newpp & 7) | H_AVPN;
-	unsigned long vpn = va >> PAGE_SHIFT;
+	unsigned long flags = (newpp & 7) | H_AVPN;
+	unsigned long avpn = va >> 23;
 
-	lpar_rc = plpar_pte_protect(flags, slot, (vpn >> 4) & ~0x7fUL);
+	lpar_rc = plpar_pte_protect(flags, slot, (avpn << 7));
 
 	if (lpar_rc == H_Not_Found) {
 		udbg_printf("updatepp missed\n");
@@ -748,18 +748,11 @@ static void pSeries_lpar_hpte_updateboltedpp(unsigned long newpp,
 static void pSeries_lpar_hpte_invalidate(unsigned long slot, unsigned long va,
 					 int large, int local)
 {
-	unsigned long vpn, avpn;
+	unsigned long avpn = va >> 23;
 	unsigned long lpar_rc;
 	unsigned long dummy1, dummy2;
 
-	if (large)
-		vpn = va >> LARGE_PAGE_SHIFT;
-	else
-		vpn = va >> PAGE_SHIFT;
-
-	avpn = vpn >> 11;
-
-	lpar_rc = plpar_pte_remove(H_AVPN, slot, (vpn >> 4) & ~0x7fUL, &dummy1,
+	lpar_rc = plpar_pte_remove(H_AVPN, slot, (avpn << 7), &dummy1,
 				   &dummy2);
 
 	if (lpar_rc == H_Not_Found) {

arch/ppc64/kernel/prom.c

@@ -144,6 +144,7 @@ static interpret_func interpret_root_props;
 #define FB_MAX	8
 #endif
 
+static int ppc64_is_smp;
 
 struct prom_t prom = {
 	0,			/* entry */

arch/ppc64/kernel/smp.c

@@ -25,8 +25,6 @@
 #include <linux/interrupt.h>
 #include <linux/kernel_stat.h>
 #include <linux/delay.h>
-#define __KERNEL_SYSCALLS__
-#include <linux/unistd.h>
 #include <linux/init.h>
 #include <linux/spinlock.h>
 #include <linux/cache.h>
@@ -53,22 +51,19 @@
 #include <asm/machdep.h>
 
 int smp_threads_ready = 0;
-volatile int smp_commenced = 0;
-int smp_tb_synchronized = 0;
 spinlock_t kernel_flag __cacheline_aligned = SPIN_LOCK_UNLOCKED;
 unsigned long cache_decay_ticks;
-static int max_cpus __initdata = NR_CPUS;
 
 /* initialised so it doesnt end up in bss */
 unsigned long cpu_online_map = 0;
 int boot_cpuid = 0;
-int ppc64_is_smp = 0;
 
-volatile unsigned long cpu_callin_map[NR_CPUS] = {0,};
+static struct smp_ops_t *smp_ops;
+
+volatile unsigned long cpu_callin_map[NR_CPUS];
 
 extern unsigned char stab_array[];
 
-int start_secondary(void *);
 extern int cpu_idle(void *unused);
 void smp_call_function_interrupt(void);
 void smp_message_pass(int target, int msg, unsigned long data, int wait);
@@ -86,7 +81,7 @@ struct xics_ipi_struct {
 
 struct xics_ipi_struct xics_ipi_message[NR_CPUS] __cacheline_aligned;
 
-#define smp_message_pass(t,m,d,w) ppc_md.smp_message_pass((t),(m),(d),(w))
+#define smp_message_pass(t,m,d,w) smp_ops->message_pass((t),(m),(d),(w))
 
 static inline void set_tb(unsigned int upper, unsigned int lower)
 {
@@ -106,7 +101,6 @@ void iSeries_smp_message_recv( struct pt_regs * regs )
 	for ( msg = 0; msg < 4; ++msg )
 		if ( test_and_clear_bit( msg, &iSeries_smp_message[cpu] ) )
 			smp_message_recv( msg, regs );
-
 }
 
 static void smp_iSeries_message_pass(int target, int msg, unsigned long data, int wait)
@@ -127,6 +121,7 @@ static void smp_iSeries_message_pass(int target, int msg, unsigned long data, in
 	}
 }
 
+#ifdef CONFIG_PPC_ISERIES
 static int smp_iSeries_numProcs(void)
 {
 	unsigned np, i;
@@ -141,24 +136,23 @@ static int smp_iSeries_numProcs(void)
         }
 	return np;
 }
+#endif
 
-static void smp_iSeries_probe(void)
+static int smp_iSeries_probe(void)
 {
 	unsigned i;
-	unsigned np;
-	struct ItLpPaca * lpPaca;
+	unsigned np = 0;
+	struct ItLpPaca *lpPaca;
 
-	np = 0;
 	for (i=0; i < MAX_PACAS; ++i) {
 		lpPaca = paca[i].xLpPacaPtr;
-		if ( lpPaca->xDynProcStatus < 2 ) {
+		if (lpPaca->xDynProcStatus < 2) {
 			paca[i].active = 1;
 			++np;
-			paca[i].next_jiffy_update_tb = paca[0].next_jiffy_update_tb;
 		}
 	}
 
-	smp_tb_synchronized = 1;
+	return np;
 }
 
 static void smp_iSeries_kick_cpu(int nr)
@@ -194,10 +188,11 @@ static void smp_iSeries_setup_cpu(int nr)
 /* This is called very early. */
 void smp_init_iSeries(void)
 {
-	ppc_md.smp_message_pass = smp_iSeries_message_pass;
-	ppc_md.smp_probe        = smp_iSeries_probe;
-	ppc_md.smp_kick_cpu     = smp_iSeries_kick_cpu;
-	ppc_md.smp_setup_cpu    = smp_iSeries_setup_cpu;
+	smp_ops = &ppc_md.smp_ops;
+	smp_ops->message_pass = smp_iSeries_message_pass;
+	smp_ops->probe        = smp_iSeries_probe;
+	smp_ops->kick_cpu     = smp_iSeries_kick_cpu;
+	smp_ops->setup_cpu    = smp_iSeries_setup_cpu;
 #ifdef CONFIG_PPC_ISERIES
 #warning fix for iseries
 	naca->processorCount	= smp_iSeries_numProcs();
@@ -229,10 +224,20 @@ smp_openpic_message_pass(int target, int msg, unsigned long data, int wait)
 	}
 }
 
-static void smp_chrp_probe(void)
+static int smp_chrp_probe(void)
 {
-	if (ppc64_is_smp)
+	int i;
+	int nr_cpus = 0;
+
+	for (i = 0; i < NR_CPUS; i++) {
+		if (cpu_possible(i))
+			nr_cpus++;
+	}
+
+	if (nr_cpus > 1)
 		openpic_request_IPIs();
+
+	return nr_cpus;
 }
 
 static void
@@ -252,78 +257,32 @@ smp_kick_cpu(int nr)
 	/* The processor is currently spinning, waiting
 	 * for the xProcStart field to become non-zero
 	 * After we set xProcStart, the processor will
-	 * continue on to secondary_start in iSeries_head.S
+	 * continue on to secondary_start
 	 */
 	paca[nr].xProcStart = 1;
 }
 
-extern struct gettimeofday_struct do_gtod;
-
-static void smp_space_timers()
+static void smp_space_timers(unsigned int max_cpus)
 {
 	int i;
-	unsigned long offset = tb_ticks_per_jiffy / NR_CPUS;
+	unsigned long offset = tb_ticks_per_jiffy / max_cpus;
+	unsigned long previous_tb = paca[boot_cpuid].next_jiffy_update_tb;
 
-	for (i = 1; i < NR_CPUS; ++i)
-		paca[i].next_jiffy_update_tb =
-			paca[i-1].next_jiffy_update_tb + offset;
-}
-
-static void
-smp_chrp_setup_cpu(int cpu_nr)
-{
-	static atomic_t ready = ATOMIC_INIT(1);
-	static volatile int frozen = 0;
-
-	if (naca->platform == PLATFORM_PSERIES_LPAR) {
-		/* timebases already synced under the hypervisor. */
-		paca[cpu_nr].next_jiffy_update_tb = tb_last_stamp = get_tb();
-		if (cpu_nr == boot_cpuid) {
-			do_gtod.tb_orig_stamp = tb_last_stamp;
-			/* Should update do_gtod.stamp_xsec.
-			 * For now we leave it which means the time can be some
-			 * number of msecs off until someone does a settimeofday()
-			 */
-		}
-		smp_tb_synchronized = 1;
-	} else {
-		if (cpu_nr == boot_cpuid) {
-			/* wait for all the others */
-			while (atomic_read(&ready) < num_online_cpus())
-				barrier();
-			atomic_set(&ready, 1);
-			/* freeze the timebase */
-			rtas_call(rtas_token("freeze-time-base"), 0, 1, NULL);
-			mb();
-			frozen = 1;
-			set_tb(0, 0);
-			paca[boot_cpuid].next_jiffy_update_tb = 0;
-			smp_space_timers();
-			while (atomic_read(&ready) < num_online_cpus())
-				barrier();
-			/* thaw the timebase again */
-			rtas_call(rtas_token("thaw-time-base"), 0, 1, NULL);
-			mb();
-			frozen = 0;
-			tb_last_stamp = get_tb();
-			do_gtod.tb_orig_stamp = tb_last_stamp;
-			smp_tb_synchronized = 1;
-		} else {
-			atomic_inc(&ready);
-			while (!frozen)
-				barrier();
-			set_tb(0, 0);
-			mb();
-			atomic_inc(&ready);
-			while (frozen)
-				barrier();
+	for (i = 0; i < NR_CPUS; i++) {
+		if (cpu_possible(i) && i != boot_cpuid) {
+			paca[i].next_jiffy_update_tb =
+				previous_tb + offset;
+			previous_tb = paca[i].next_jiffy_update_tb;
 		}
 	}
+}
 
+static void __devinit pSeries_setup_cpu(int cpu)
+{
 	if (OpenPIC_Addr) {
 		do_openpic_setup_cpu();
 	} else {
-		if (cpu_nr != boot_cpuid)
+		if (cpu != boot_cpuid)
 			xics_setup_cpu();
 	}
 }
@@ -347,26 +306,65 @@ smp_xics_message_pass(int target, int msg, unsigned long data, int wait)
 	}
 }
 
-static void smp_xics_probe(void)
+static int smp_xics_probe(void)
 {
+	int i;
+	int nr_cpus = 0;
+
+	for (i = 0; i < NR_CPUS; i++) {
+		if (cpu_possible(i))
+			nr_cpus++;
+	}
+
+	return nr_cpus;
+}
+
+static spinlock_t timebase_lock = SPIN_LOCK_UNLOCKED;
+static unsigned long timebase = 0;
+
+static void __devinit pSeries_give_timebase(void)
+{
+	spin_lock(&timebase_lock);
+	rtas_call(rtas_token("freeze-time-base"), 0, 1, NULL);
+	timebase = get_tb();
+	spin_unlock(&timebase_lock);
+
+	while (timebase)
+		barrier();
+	rtas_call(rtas_token("thaw-time-base"), 0, 1, NULL);
+}
+
+static void __devinit pSeries_take_timebase(void)
+{
+	while (!timebase)
+		barrier();
+	spin_lock(&timebase_lock);
+	set_tb(timebase, timebase >> 32);
+	timebase = 0;
+	spin_unlock(&timebase_lock);
 }
 
 /* This is called very early */
-void smp_init_pSeries(void)
+void __init smp_init_pSeries(void)
 {
-	if(naca->interrupt_controller == IC_OPEN_PIC) {
-		ppc_md.smp_message_pass = smp_openpic_message_pass;
-		ppc_md.smp_probe        = smp_chrp_probe;
-		ppc_md.smp_kick_cpu     = smp_kick_cpu;
-		ppc_md.smp_setup_cpu    = smp_chrp_setup_cpu;
+	smp_ops = &ppc_md.smp_ops;
+
+	if (naca->interrupt_controller == IC_OPEN_PIC) {
+		smp_ops->message_pass	= smp_openpic_message_pass;
+		smp_ops->probe		= smp_chrp_probe;
 	} else {
-		ppc_md.smp_message_pass = smp_xics_message_pass;
-		ppc_md.smp_probe        = smp_xics_probe;
-		ppc_md.smp_kick_cpu     = smp_kick_cpu;
-		ppc_md.smp_setup_cpu    = smp_chrp_setup_cpu;
+		smp_ops->message_pass	= smp_xics_message_pass;
+		smp_ops->probe		= smp_xics_probe;
+	}
+
+	if (naca->platform == PLATFORM_PSERIES) {
+		smp_ops->give_timebase = pSeries_give_timebase;
+		smp_ops->take_timebase = pSeries_take_timebase;
 	}
-}
 
+	smp_ops->kick_cpu = smp_kick_cpu;
+	smp_ops->setup_cpu = pSeries_setup_cpu;
+}
 
 void smp_local_timer_interrupt(struct pt_regs * regs)
 {
@@ -469,8 +467,7 @@ static struct call_data_struct {
  * hardware interrupt handler or from a bottom half handler.
  */
 int smp_call_function (void (*func) (void *info), void *info, int nonatomic,
-			int wait)
-
+		       int wait)
 { 
 	struct call_data_struct data;
 	int ret = -1, cpus = num_online_cpus()-1;
@@ -553,38 +550,41 @@ void smp_call_function_interrupt(void)
 		atomic_inc(&call_data->finished);
 }
 
-
 extern unsigned long decr_overclock;
+extern struct gettimeofday_struct do_gtod;
 
-struct thread_struct *current_set[NR_CPUS] = {&init_thread_union, 0};
+struct thread_info *current_set[NR_CPUS];
 
-void __init smp_boot_cpus(void)
+static void __devinit smp_store_cpu_info(int id)
 {
-	int i, cpu_nr = 0;
-	struct task_struct *p;
-
-	printk("Entering SMP Mode...\n");
+	paca[id].pvr = _get_PVR();
+}
 
-        smp_store_cpu_info(boot_cpuid);
-	cpu_callin_map[boot_cpuid] = 1;
+void __init smp_prepare_cpus(unsigned int max_cpus)
+{
+	int i;
 
-	/* XXX buggy - Anton */
-	current_thread_info()->cpu = 0;
+	/* Fixup boot cpu */
+	smp_store_cpu_info(smp_processor_id());
+	cpu_callin_map[smp_processor_id()] = 1;
 
 	for (i = 0; i < NR_CPUS; i++) {
 		paca[i].prof_counter = 1;
 		paca[i].prof_multiplier = 1;
 		if (i != boot_cpuid) {
+			void *tmp;
 		        /*
 			 * the boot cpu segment table is statically 
 			 * initialized to real address 0x5000.  The
 			 * Other processor's tables are created and
 			 * initialized here.
 			 */
-			paca[i].xStab_data.virt = (unsigned long)&stab_array[PAGE_SIZE * (i-1)];
-			memset((void *)paca[i].xStab_data.virt, 0, PAGE_SIZE); 
-			paca[i].xStab_data.real = __v2a(paca[i].xStab_data.virt);
-			paca[i].default_decr = tb_ticks_per_jiffy / decr_overclock;
+			tmp = &stab_array[PAGE_SIZE * (i-1)];
+			memset(tmp, 0, PAGE_SIZE); 
+			paca[i].xStab_data.virt = (unsigned long)tmp;
+			paca[i].xStab_data.real = (unsigned long)__v2a(tmp);
+			paca[i].default_decr = tb_ticks_per_jiffy /
+				decr_overclock;
 		}
 	}
 
@@ -593,135 +593,84 @@ void __init smp_boot_cpus(void)
 	 */
 	cache_decay_ticks = HZ/100;
 
-	ppc_md.smp_probe();
+#ifndef CONFIG_PPC_ISERIES
+	paca[boot_cpuid].next_jiffy_update_tb = tb_last_stamp = get_tb();
 
-	for (i = 0; i < NR_CPUS; i++) {
-		if (paca[i].active)
-			cpu_nr++;
-	}
-	printk("Probe found %d CPUs\n", cpu_nr);
-
-#ifdef CONFIG_ISERIES
-	smp_space_timers();
+	/*
+	 * Should update do_gtod.stamp_xsec.
+	 * For now we leave it which means the time can be some
+	 * number of msecs off until someone does a settimeofday()
+	 */
+	do_gtod.tb_orig_stamp = tb_last_stamp;
 #endif
 
-	printk("Waiting for %d CPUs\n", cpu_nr-1);
-
-	for (i = 1 ; i < NR_CPUS; i++) {
-		int c;
-		struct pt_regs regs;
+	max_cpus = smp_ops->probe();
+	smp_space_timers(max_cpus);
+}
 
-		if (!paca[i].active)
-			continue;
+int __cpu_up(unsigned int cpu)
+{
+	struct pt_regs regs;
+	struct task_struct *p;
+	int c;
 
-		if (i == boot_cpuid)
-			continue;
+	/* create a process for the processor */
+	/* only regs.msr is actually used, and 0 is OK for it */
+	memset(&regs, 0, sizeof(struct pt_regs));
+	p = do_fork(CLONE_VM|CLONE_IDLETASK, 0, &regs, 0);
+	if (IS_ERR(p))
+		panic("failed fork for CPU %u: %li", cpu, PTR_ERR(p));
 
-		if (num_online_cpus() >= max_cpus)
-			break;
-
-		/* create a process for the processor */
-		/* we don't care about the values in regs since we'll
-		   never reschedule the forked task. */
-		/* We DO care about one bit in the pt_regs we
-		   pass to do_fork.  That is the MSR_FP bit in
-		   regs.msr.  If that bit is on, then do_fork
-		   (via copy_thread) will call giveup_fpu.
-		   giveup_fpu will get a pointer to our (current's)
-		   last register savearea via current->thread.regs
-		   and using that pointer will turn off the MSR_FP,
-		   MSR_FE0 and MSR_FE1 bits.  At this point, this
-		   pointer is pointing to some arbitrary point within
-		   our stack */
-
-		memset(&regs, 0, sizeof(struct pt_regs));
-
-		p = do_fork(CLONE_VM|CLONE_IDLETASK, 0, &regs, 0);
-		if (IS_ERR(p))
-			panic("failed fork for CPU %d", i);
-
-		init_idle(p, i);
-
-		unhash_process(p);
-
-		paca[i].xCurrent = (u64)p;
-		current_set[i] = p->thread_info;
-
-		/* wake up cpus */
-		ppc_md.smp_kick_cpu(i);
-
-		/*
-		 * wait to see if the cpu made a callin (is actually up).
-		 * use this value that I found through experimentation.
-		 * -- Cort
-		 */
-		for ( c = 5000; c && !cpu_callin_map[i] ; c-- ) {
-			udelay(100);
-		}
-		
-		if ( cpu_callin_map[i] )
-		{
-			printk("Processor %d found.\n", i);
-			/* this sync's the decr's -- Cort */
-		} else {
-			printk("Processor %d is stuck.\n", i);
-		}
-	}
+	init_idle(p, cpu);
+	unhash_process(p);
 
-	/* Setup boot cpu last (important) */
-	ppc_md.smp_setup_cpu(boot_cpuid);
+	paca[cpu].xCurrent = (u64)p;
+	current_set[cpu] = p->thread_info;
 
-	if (num_online_cpus() < 2) {
-	        tb_last_stamp = get_tb();
-		smp_tb_synchronized = 1;
-	}
-}
+	/* wake up cpus */
+	smp_ops->kick_cpu(cpu);
 
-void __init smp_commence(void)
-{
 	/*
-	 *	Lets the callin's below out of their loop.
+	 * wait to see if the cpu made a callin (is actually up).
+	 * use this value that I found through experimentation.
+	 * -- Cort
 	 */
-	PPCDBG(PPCDBG_SMP, "smp_commence: start\n"); 
-	wmb();
-	smp_commenced = 1;
-}
-
-void __init smp_callin(void)
-{
-	int cpu = smp_processor_id();
+	for (c = 5000; c && !cpu_callin_map[cpu]; c--)
+		udelay(100);
 
-        smp_store_cpu_info(cpu);
-	set_dec(paca[cpu].default_decr);
-	set_bit(smp_processor_id(), &cpu_online_map);
-	smp_mb();
-	cpu_callin_map[cpu] = 1;
-
-	ppc_md.smp_setup_cpu(cpu);
-	
-	while(!smp_commenced) {
-		barrier();
+	if (!cpu_callin_map[cpu]) {
+		printk("Processor %u is stuck.\n", cpu);
+		return -ENOENT;
 	}
-	local_irq_enable();
-}
 
-/* intel needs this */
-void __init initialize_secondary(void)
-{
+	printk("Processor %u found.\n", cpu);
+
+	if (smp_ops->give_timebase)
+		smp_ops->give_timebase();
+	set_bit(cpu, &cpu_online_map);
+	return 0;
 }
 
 /* Activate a secondary processor. */
-int start_secondary(void *unused)
+int __devinit start_secondary(void *unused)
 {
+	unsigned int cpu = smp_processor_id();
+
 	atomic_inc(&init_mm.mm_count);
 	current->active_mm = &init_mm;
-	smp_callin();
 
-	return cpu_idle(NULL);
-}
+	smp_store_cpu_info(cpu);
+	set_dec(paca[cpu].default_decr);
+	cpu_callin_map[cpu] = 1;
 
-void __init smp_setup(char *str, int *ints)
-{
+	smp_ops->setup_cpu(cpu);
+	if (smp_ops->take_timebase)
+		smp_ops->take_timebase();
+
+	/* XXX required? */
+	local_irq_enable();
+
+	return cpu_idle(NULL);
 }
 
 int setup_profiling_timer(unsigned int multiplier)
@@ -729,17 +678,10 @@ int setup_profiling_timer(unsigned int multiplier)
 	return 0;
 }
 
-/* this function is called for each processor
- */
-void __init smp_store_cpu_info(int id)
+void smp_cpus_done(unsigned int max_cpus)
 {
-        paca[id].pvr = _get_PVR();
-}
+	smp_ops->setup_cpu(boot_cpuid);
 
-static int __init maxcpus(char *str)
-{
-	get_option(&str, &max_cpus);
-	return 1;
+	/* XXX fix this, xics currently relies on it - Anton */
+	smp_threads_ready = 1;
 }
-
-__setup("maxcpus=", maxcpus);

arch/ppc64/kernel/sys_ppc32.c

@@ -4438,30 +4438,32 @@ asmlinkage int sys32_vm86(u32 a1, u32 a2, u32 a3, u32 a4)
 	return sys_vm86((int)a1, (int)a2, (int)a3, (int)a4);
 }
 
+extern ssize_t sys_pread64(unsigned int fd, char *buf, size_t count,
+			   loff_t pos);
 
-
-
-
-extern asmlinkage ssize_t sys_pread(unsigned int fd, char * buf,
-				    size_t count, loff_t pos);
-
-extern asmlinkage ssize_t sys_pwrite(unsigned int fd, const char * buf,
-				     size_t count, loff_t pos);
+extern ssize_t sys_pwrite64(unsigned int fd, const char *buf, size_t count,
+			    loff_t pos);
 
 typedef __kernel_ssize_t32 ssize_t32;
 
-asmlinkage ssize_t32 sys32_pread(unsigned int fd, char *ubuf,
-				 __kernel_size_t32 count, u32 reg6, u32 poshi, u32 poslo)
+ssize_t32 sys32_pread64(unsigned int fd, char *ubuf, __kernel_size_t32 count,
+			u32 reg6, u32 poshi, u32 poslo)
 {
-	return sys_pread(fd, ubuf, count, ((loff_t)AA(poshi) << 32) | AA(poslo));
+	return sys_pread64(fd, ubuf, count, ((loff_t)AA(poshi) << 32) | AA(poslo));
 }
 
-asmlinkage ssize_t32 sys32_pwrite(unsigned int fd, char *ubuf,
-				  __kernel_size_t32 count, u32 reg6 ,u32 poshi, u32 poslo)
+ssize_t32 sys32_pwrite64(unsigned int fd, char *ubuf, __kernel_size_t32 count,
+			 u32 reg6 ,u32 poshi, u32 poslo)
 {
-	return sys_pwrite(fd, ubuf, count, ((loff_t)AA(poshi) << 32) | AA(poslo));
+	return sys_pwrite64(fd, ubuf, count, ((loff_t)AA(poshi) << 32) | AA(poslo));
 }
 
+extern ssize_t sys_readahead(int fd, loff_t offset, size_t count);
+
+ssize_t32 sys32_readahead(int fd, u32 offhi, u32 offlo, s32 count)
+{
+        return sys_readahead(fd, ((loff_t)AA(offhi) << 32) | AA(offlo), count);
+}
 
 extern asmlinkage long sys_truncate(const char * path, unsigned long length);
 extern asmlinkage long sys_ftruncate(unsigned int fd, unsigned long length);

arch/ppc64/kernel/time.c

@@ -293,9 +293,6 @@ int timer_interrupt(struct pt_regs * regs)
 
 	irq_exit();
 
-	if (softirq_pending(cpu))
-		do_softirq();
-	
 	return 1;
 }
 

arch/ppc64/kernel/traps.c

@@ -128,13 +128,12 @@ void
 SystemResetException(struct pt_regs *regs)
 {
 	if (fwnmi_active) {
-		char *msg;
 		unsigned long *r3 = __va(regs->gpr[3]); /* for FWNMI debug */
 		struct rtas_error_log *errlog;
 
-		msg = "FWNMI is active with save area at %016lx\n";
-		udbg_printf(msg, r3); printk(msg, r3);
+		udbg_printf("FWNMI is active with save area at %016lx\n", r3);
 		errlog = FWNMI_get_errinfo(regs);
+		FWNMI_release_errinfo();
 	}
 
 	if (debugger)

arch/ppc64/kernel/xics.c

@@ -373,7 +373,7 @@ xics_init_IRQ( void )
 	if (naca->platform == PLATFORM_PSERIES) {
 #ifdef CONFIG_SMP
 		for (i = 0; i < NR_CPUS; ++i) {
-			if (!paca[i].active)
+			if (!cpu_possible(i))
 				continue;
 			xics_info.per_cpu[i] =
 			  __ioremap((ulong)inodes[i].addr, 

include/asm-ppc64/hardirq.h

@@ -13,30 +13,86 @@
 #include <linux/preempt.h>
 
 typedef struct {
-	unsigned long __softirq_pending;
-	unsigned long __syscall_count;
+	unsigned int __softirq_pending;
+	unsigned int __syscall_count;
 	struct task_struct * __ksoftirqd_task;
 	unsigned long idle_timestamp;
 } ____cacheline_aligned irq_cpustat_t;
 
 #include <linux/irq_cpustat.h>	/* Standard mappings for irq_cpustat_t above */
 
-#define IRQ_OFFSET 64
+/*
+ * We put the hardirq and softirq counter into the preemption
+ * counter. The bitmask has the following meaning:
+ *
+ * - bits 0-7 are the preemption count (max preemption depth: 256)
+ * - bits 8-15 are the softirq count (max # of softirqs: 256)
+ * - bits 16-24 are the hardirq count (max # of hardirqs: 512)
+ *
+ * - ( bit 26 is the PREEMPT_ACTIVE flag. )
+ *
+ * PREEMPT_MASK: 0x000000ff
+ * SOFTIRQ_MASK: 0x0000ff00
+ * HARDIRQ_MASK: 0x01ff0000
+ */
+
+#define PREEMPT_BITS	8
+#define SOFTIRQ_BITS	8
+#define HARDIRQ_BITS	9
+
+#define PREEMPT_SHIFT	0
+#define SOFTIRQ_SHIFT	(PREEMPT_SHIFT + PREEMPT_BITS)
+#define HARDIRQ_SHIFT	(SOFTIRQ_SHIFT + SOFTIRQ_BITS)
+
+#define __HARDIRQ_MASK(x)	((1UL << (x))-1)
+
+#define PREEMPT_MASK	(__HARDIRQ_MASK(PREEMPT_BITS) << PREEMPT_SHIFT)
+#define SOFTIRQ_MASK	(__HARDIRQ_MASK(SOFTIRQ_BITS) << SOFTIRQ_SHIFT)
+#define HARDIRQ_MASK	(__HARDIRQ_MASK(HARDIRQ_BITS) << HARDIRQ_SHIFT)
+
+#define hardirq_count()	(preempt_count() & HARDIRQ_MASK)
+#define softirq_count()	(preempt_count() & SOFTIRQ_MASK)
+#define irq_count()	(preempt_count() & (HARDIRQ_MASK | SOFTIRQ_MASK))
+
+#define PREEMPT_OFFSET	(1UL << PREEMPT_SHIFT)
+#define SOFTIRQ_OFFSET	(1UL << SOFTIRQ_SHIFT)
+#define HARDIRQ_OFFSET	(1UL << HARDIRQ_SHIFT)
+
+/*
+ * The hardirq mask has to be large enough to have
+ * space for potentially all IRQ sources in the system
+ * nesting on a single CPU:
+ */
+#if (1 << HARDIRQ_BITS) < NR_IRQS
+# error HARDIRQ_BITS is too low!
+#endif
 
 /*
- * Are we in an interrupt context? Either doing bottom half
- * or hardware interrupt processing?
+ * Are we doing bottom half or hardware interrupt processing?
+ * Are we in a softirq context? Interrupt context?
  */
-#define in_interrupt() \
-		((preempt_count() & ~PREEMPT_ACTIVE) >= IRQ_OFFSET)
+#define in_irq()		(hardirq_count())
+#define in_softirq()		(softirq_count())
+#define in_interrupt()		(irq_count())
 
-#define in_irq in_interrupt
 
 #define hardirq_trylock()	(!in_interrupt())
 #define hardirq_endlock()	do { } while (0)
 
-#define irq_enter()		(preempt_count() += IRQ_OFFSET)
-#define irq_exit()		(preempt_count() -= IRQ_OFFSET)
+#define irq_enter()		(preempt_count() += HARDIRQ_OFFSET)
+
+#if CONFIG_PREEMPT
+# define IRQ_EXIT_OFFSET (HARDIRQ_OFFSET-1)
+#else
+# define IRQ_EXIT_OFFSET HARDIRQ_OFFSET
+#endif
+#define irq_exit()							\
+do {									\
+		preempt_count() -= IRQ_EXIT_OFFSET;			\
+		if (!in_interrupt() && softirq_pending(smp_processor_id())) \
+			do_softirq();					\
+		preempt_enable_no_resched();				\
+} while (0)
 
 #ifndef CONFIG_SMP
 # define synchronize_irq(irq)	barrier()
@@ -45,4 +101,13 @@ typedef struct {
 #endif /* CONFIG_SMP */
 
 #endif /* __KERNEL__ */
+	
+#define show_stack(SP)				\
+do {						\
+	if (SP)					\
+		print_backtrace(SP);		\
+	else					\
+		print_backtrace(_get_SP());	\
+} while (0)
+
 #endif /* __ASM_HARDIRQ_H */

include/asm-ppc64/machdep.h

@@ -20,6 +20,17 @@ struct device_node;
 struct TceTable;
 struct rtc_time;
 
+#ifdef CONFIG_SMP
+struct smp_ops_t {
+	void  (*message_pass)(int target, int msg, unsigned long data, int wait);
+	int   (*probe)(void);
+	void  (*kick_cpu)(int nr);
+	void  (*setup_cpu)(int nr);
+	void  (*take_timebase)(void);
+	void  (*give_timebase)(void);
+};
+#endif
+
 struct machdep_calls {
 	/* High use functions in the first cachelines, low use functions
 	 * follow.  DRENG collect profile data.
@@ -58,14 +69,6 @@ struct machdep_calls {
 	void		(*tce_free_one)(struct TceTable *tbl,
 				        long tcenum);    
 
-	void		(*smp_message_pass)(int target,
-					    int msg, 
-					    unsigned long data,
-					    int wait);
-	void		(*smp_probe)(void);
-	void		(*smp_kick_cpu)(int nr);
-	void		(*smp_setup_cpu)(int nr);
-
 	void		(*setup_arch)(void);
 	/* Optional, may be NULL. */
 	void		(*setup_residual)(struct seq_file *m, int cpu_id);
@@ -145,6 +148,11 @@ struct machdep_calls {
 
 	/* this is for modules, since _machine can be a define -- Cort */
 	int ppc_machine;
+
+#ifdef CONFIG_SMP
+	/* functions for dealing with other cpus */
+	struct smp_ops_t smp_ops;
+#endif /* CONFIG_SMP */
 };
 
 extern struct machdep_calls ppc_md;

include/asm-ppc64/mmzone.h

@@ -22,7 +22,10 @@ extern plat_pg_data_t plat_node_data[];
 #define MAX_NUMNODES 4
 
 /* XXX grab this from the device tree - Anton */
-#define PHYSADDR_TO_NID(pa)		((pa) >> 36)
+#define MEMORY_ZONE_BITS	33
+#define CPU_SHIFT_BITS          1
+
+#define PHYSADDR_TO_NID(pa)		((pa) >> MEMORY_ZONE_BITS)
 #define PLAT_NODE_DATA(n)		(&plat_node_data[(n)])
 #define PLAT_NODE_DATA_STARTNR(n)	\
 	(PLAT_NODE_DATA(n)->gendata.node_start_mapnr)
@@ -84,7 +87,7 @@ extern plat_pg_data_t plat_node_data[];
 #ifdef CONFIG_NUMA
 
 /* XXX grab this from the device tree - Anton */
-#define cputonode(cpu)	((cpu) >> 3)
+#define cputonode(cpu)	((cpu) >> CPU_SHIFT_BITS)
 
 #define numa_node_id()	cputonode(smp_processor_id())
 

include/asm-ppc64/rwsem.h

@@ -57,6 +57,7 @@ struct rw_semaphore {
 extern struct rw_semaphore *rwsem_down_read_failed(struct rw_semaphore *sem);
 extern struct rw_semaphore *rwsem_down_write_failed(struct rw_semaphore *sem);
 extern struct rw_semaphore *rwsem_wake(struct rw_semaphore *sem);
+extern struct rw_semaphore *rwsem_downgrade_wake(struct rw_semaphore *sem);
 
 static inline void init_rwsem(struct rw_semaphore *sem)
 {
@@ -73,12 +74,26 @@ static inline void init_rwsem(struct rw_semaphore *sem)
  */
 static inline void __down_read(struct rw_semaphore *sem)
 {
-	if (atomic_inc_return((atomic_t *)(&sem->count)) >= 0)
+	if (atomic_inc_return((atomic_t *)(&sem->count)) > 0)
 		smp_wmb();
 	else
 		rwsem_down_read_failed(sem);
 }
 
+static inline int __down_read_trylock(struct rw_semaphore *sem)
+{
+	int tmp;
+
+	while ((tmp = sem->count) >= 0) {
+		if (tmp == cmpxchg(&sem->count, tmp,
+				   tmp + RWSEM_ACTIVE_READ_BIAS)) {
+			smp_wmb();
+			return 1;
+		}
+	}
+	return 0;
+}
+
 /*
  * lock for writing
  */
@@ -94,6 +109,16 @@ static inline void __down_write(struct rw_semaphore *sem)
 		rwsem_down_write_failed(sem);
 }
 
+static inline int __down_write_trylock(struct rw_semaphore *sem)
+{
+	int tmp;
+
+	tmp = cmpxchg(&sem->count, RWSEM_UNLOCKED_VALUE,
+		      RWSEM_ACTIVE_WRITE_BIAS);
+	smp_wmb();
+	return tmp == RWSEM_UNLOCKED_VALUE;
+}
+
 /*
  * unlock after reading
  */
@@ -126,6 +151,19 @@ static inline void rwsem_atomic_add(int delta, struct rw_semaphore *sem)
 	atomic_add(delta, (atomic_t *)(&sem->count));
 }
 
+/*
+ * downgrade write lock to read lock
+ */
+static inline void __downgrade_write(struct rw_semaphore *sem)
+{
+	int tmp;
+
+	smp_wmb();
+	tmp = atomic_add_return(-RWSEM_WAITING_BIAS, (atomic_t *)(&sem->count));
+	if (tmp < 0)
+		rwsem_downgrade_wake(sem);
+}
+
 /*
  * implement exchange and add functionality
  */

include/asm-ppc64/smp.h

@@ -30,7 +30,6 @@
 extern unsigned long cpu_online_map;
 
 extern void smp_message_pass(int target, int msg, unsigned long data, int wait);
-extern void smp_store_cpu_info(int id);
 extern void smp_send_tlb_invalidate(int);
 extern void smp_send_xmon_break(int cpu);
 struct pt_regs;

include/asm-ppc64/softirq.h

@@ -12,24 +12,16 @@
 #include <asm/hardirq.h>
 
 #define local_bh_disable() \
-		do { preempt_count() += IRQ_OFFSET; barrier(); } while (0)
-
+		do { preempt_count() += SOFTIRQ_OFFSET; barrier(); } while (0)
 #define __local_bh_enable() \
-		do { barrier(); preempt_count() -= IRQ_OFFSET; } while (0)
+		do { barrier(); preempt_count() -= SOFTIRQ_OFFSET; } while (0)
 
-#define local_bh_enable() \
-do { \
-	if (unlikely((preempt_count() == IRQ_OFFSET) && \
-		     softirq_pending(smp_processor_id()))) { \
-		__local_bh_enable(); \
-		do_softirq(); \
-		preempt_check_resched(); \
-	} else { \
-		__local_bh_enable(); \
-		preempt_check_resched(); \
-	} \
+#define local_bh_enable()						\
+do {									\
+	__local_bh_enable();						\
+	if (unlikely(!in_interrupt() && softirq_pending(smp_processor_id()))) \
+		do_softirq();						\
+	preempt_check_resched();					\
 } while (0)
 
-#define in_softirq() in_interrupt()
-
 #endif	/* __ASM_SOFTIRQ_H */

include/asm-ppc64/system.h

@@ -88,26 +88,18 @@ struct task_struct;
 extern void __switch_to(struct task_struct *, struct task_struct *);
 #define switch_to(prev, next, last) __switch_to((prev), (next))
 
-#define prepare_arch_schedule(prev)		do { } while(0)
-#define finish_arch_schedule(prev)		do { } while(0)
-#define prepare_arch_switch(rq)			do { } while(0)
-#define finish_arch_switch(rq)			spin_unlock_irq(&(rq)->lock)
-
 struct thread_struct;
 extern void _switch(struct thread_struct *prev, struct thread_struct *next);
 
 struct pt_regs;
 extern void dump_regs(struct pt_regs *);
 
-#ifndef CONFIG_SMP
-
-#define cli()	local_irq_disable()
-#define sti()	local_irq_enable()
-#define save_flags(flags)	local_save_flags(flags)
-#define restore_flags(flags)	local_irq_restore(flags)
-#define save_and_cli(flags)	local_irq_save(flags)
-
-#endif /* !CONFIG_SMP */
+#define irqs_disabled()				\
+({						\
+	unsigned long flags;			\
+	local_save_flags(flags);		\
+	!(flags & MSR_EE);			\
+})
 
 static __inline__ int __is_processor(unsigned long pv)
 {

include/asm-ppc64/thread_info.h

@@ -27,6 +27,8 @@ struct thread_info {
 
 /*
  * macros/functions for gaining access to the thread information structure
+ *
+ * preempt_count needs to be 1 initially, until the scheduler is functional.
  */
 #define INIT_THREAD_INFO(tsk)			\
 {						\
@@ -34,6 +36,7 @@ struct thread_info {
 	exec_domain:	&default_exec_domain,	\
 	flags:		0,			\
 	cpu:		0,			\
+	preempt_count:	1,			\
 }
 
 #define init_thread_info	(init_thread_union.thread_info)

include/asm-ppc64/unistd.h

@@ -200,6 +200,7 @@
 #define __NR_putpmsg		188	/* some people actually want streams */
 #define __NR_vfork		189
 #define __NR_ugetrlimit		190	/* SuS compliant getrlimit */
+#define __NR_readahead		191
 #define __NR_mmap2		192
 #define __NR_truncate64		193
 #define __NR_ftruncate64	194
@@ -230,15 +231,10 @@
 #define __NR_lremovexattr	219
 #define __NR_fremovexattr	220
 #define __NR_futex		221
-#define __NR_tux		222
-#define __NR_sched_setaffinity	223     
-#define __NR_sched_getaffinity	224
-
-#if 0
-/* Remind paulus to add these into ppc32 */
-__NR_security
-__NR_readahead
-#endif
+#define __NR_sched_setaffinity	222     
+#define __NR_sched_getaffinity	223
+#define __NR_security		224
+#define __NR_tuxcall		225
 
 #define __NR(n)	#n