aboutsummaryrefslogtreecommitdiffstats
path: root/arch/powerpc/kernel
diff options
context:
space:
mode:
authorPaul Mundt <lethal@linux-sh.org>2012-01-08 19:56:37 -0500
committerPaul Mundt <lethal@linux-sh.org>2012-01-08 19:56:37 -0500
commit04cf399640b7acfa9abe2eb7900cd934db8af697 (patch)
treef9a055f2f0170550f5f0b0507b06ffce8d98945d /arch/powerpc/kernel
parent17f0056e6a2f3d1818801705f5e12b71217bf4ef (diff)
parenta0e86bd4252519321b0d102dc4ed90557aa7bee9 (diff)
Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux into rmobile-latest
Conflicts: arch/arm/mach-shmobile/Makefile Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Diffstat (limited to 'arch/powerpc/kernel')
-rw-r--r--arch/powerpc/kernel/Makefile2
-rw-r--r--arch/powerpc/kernel/asm-offsets.c1
-rw-r--r--arch/powerpc/kernel/cacheinfo.c10
-rw-r--r--arch/powerpc/kernel/cpu_setup_a2.S10
-rw-r--r--arch/powerpc/kernel/cputable.c27
-rw-r--r--arch/powerpc/kernel/crash.c220
-rw-r--r--arch/powerpc/kernel/crash_dump.c4
-rw-r--r--arch/powerpc/kernel/entry_32.S15
-rw-r--r--arch/powerpc/kernel/exceptions-64s.S2
-rw-r--r--arch/powerpc/kernel/head_44x.S107
-rw-r--r--arch/powerpc/kernel/head_fsl_booke.S2
-rw-r--r--arch/powerpc/kernel/idle.c42
-rw-r--r--arch/powerpc/kernel/idle_power7.S4
-rw-r--r--arch/powerpc/kernel/irq.c28
-rw-r--r--arch/powerpc/kernel/jump_label.c2
-rw-r--r--arch/powerpc/kernel/kvm.c1
-rw-r--r--arch/powerpc/kernel/lparcfg.c2
-rw-r--r--arch/powerpc/kernel/machine_kexec.c5
-rw-r--r--arch/powerpc/kernel/misc_32.S2
-rw-r--r--arch/powerpc/kernel/pci-common.c78
-rw-r--r--arch/powerpc/kernel/pci_dn.c3
-rw-r--r--arch/powerpc/kernel/process.c46
-rw-r--r--arch/powerpc/kernel/prom.c20
-rw-r--r--arch/powerpc/kernel/prom_init.c43
-rw-r--r--arch/powerpc/kernel/reloc_32.S208
-rw-r--r--arch/powerpc/kernel/rtas_flash.c6
-rw-r--r--arch/powerpc/kernel/rtasd.c7
-rw-r--r--arch/powerpc/kernel/setup_32.c2
-rw-r--r--arch/powerpc/kernel/setup_64.c11
-rw-r--r--arch/powerpc/kernel/signal_32.c2
-rw-r--r--arch/powerpc/kernel/smp.c5
-rw-r--r--arch/powerpc/kernel/sysfs.c271
-rw-r--r--arch/powerpc/kernel/time.c105
-rw-r--r--arch/powerpc/kernel/traps.c175
-rw-r--r--arch/powerpc/kernel/vio.c1
-rw-r--r--arch/powerpc/kernel/vmlinux.lds.S8
36 files changed, 953 insertions, 524 deletions
diff --git a/arch/powerpc/kernel/Makefile b/arch/powerpc/kernel/Makefile
index ce4f7f179117..ee728e433aa2 100644
--- a/arch/powerpc/kernel/Makefile
+++ b/arch/powerpc/kernel/Makefile
@@ -85,6 +85,8 @@ extra-$(CONFIG_FSL_BOOKE) := head_fsl_booke.o
85extra-$(CONFIG_8xx) := head_8xx.o 85extra-$(CONFIG_8xx) := head_8xx.o
86extra-y += vmlinux.lds 86extra-y += vmlinux.lds
87 87
88obj-$(CONFIG_RELOCATABLE_PPC32) += reloc_32.o
89
88obj-$(CONFIG_PPC32) += entry_32.o setup_32.o 90obj-$(CONFIG_PPC32) += entry_32.o setup_32.o
89obj-$(CONFIG_PPC64) += dma-iommu.o iommu.o 91obj-$(CONFIG_PPC64) += dma-iommu.o iommu.o
90obj-$(CONFIG_KGDB) += kgdb.o 92obj-$(CONFIG_KGDB) += kgdb.o
diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
index 7c5324f1ec9c..04caee7d9bc1 100644
--- a/arch/powerpc/kernel/asm-offsets.c
+++ b/arch/powerpc/kernel/asm-offsets.c
@@ -208,6 +208,7 @@ int main(void)
208 DEFINE(PACA_USER_TIME, offsetof(struct paca_struct, user_time)); 208 DEFINE(PACA_USER_TIME, offsetof(struct paca_struct, user_time));
209 DEFINE(PACA_SYSTEM_TIME, offsetof(struct paca_struct, system_time)); 209 DEFINE(PACA_SYSTEM_TIME, offsetof(struct paca_struct, system_time));
210 DEFINE(PACA_TRAP_SAVE, offsetof(struct paca_struct, trap_save)); 210 DEFINE(PACA_TRAP_SAVE, offsetof(struct paca_struct, trap_save));
211 DEFINE(PACA_NAPSTATELOST, offsetof(struct paca_struct, nap_state_lost));
211#endif /* CONFIG_PPC64 */ 212#endif /* CONFIG_PPC64 */
212 213
213 /* RTAS */ 214 /* RTAS */
diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index a3c684b4c862..92c6b008dd2b 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -451,15 +451,15 @@ out:
451static struct cache_dir *__cpuinit cacheinfo_create_cache_dir(unsigned int cpu_id) 451static struct cache_dir *__cpuinit cacheinfo_create_cache_dir(unsigned int cpu_id)
452{ 452{
453 struct cache_dir *cache_dir; 453 struct cache_dir *cache_dir;
454 struct sys_device *sysdev; 454 struct device *dev;
455 struct kobject *kobj = NULL; 455 struct kobject *kobj = NULL;
456 456
457 sysdev = get_cpu_sysdev(cpu_id); 457 dev = get_cpu_device(cpu_id);
458 WARN_ONCE(!sysdev, "no sysdev for CPU %i\n", cpu_id); 458 WARN_ONCE(!dev, "no dev for CPU %i\n", cpu_id);
459 if (!sysdev) 459 if (!dev)
460 goto err; 460 goto err;
461 461
462 kobj = kobject_create_and_add("cache", &sysdev->kobj); 462 kobj = kobject_create_and_add("cache", &dev->kobj);
463 if (!kobj) 463 if (!kobj)
464 goto err; 464 goto err;
465 465
diff --git a/arch/powerpc/kernel/cpu_setup_a2.S b/arch/powerpc/kernel/cpu_setup_a2.S
index 7f818feaa7a5..ebc62f42a237 100644
--- a/arch/powerpc/kernel/cpu_setup_a2.S
+++ b/arch/powerpc/kernel/cpu_setup_a2.S
@@ -41,11 +41,16 @@ _GLOBAL(__setup_cpu_a2)
41 * core local but doing it always won't hurt 41 * core local but doing it always won't hurt
42 */ 42 */
43 43
44#ifdef CONFIG_PPC_WSP_COPRO 44#ifdef CONFIG_PPC_ICSWX
45 /* Make sure ACOP starts out as zero */ 45 /* Make sure ACOP starts out as zero */
46 li r3,0 46 li r3,0
47 mtspr SPRN_ACOP,r3 47 mtspr SPRN_ACOP,r3
48 48
49 /* Skip the following if we are in Guest mode */
50 mfmsr r3
51 andis. r0,r3,MSR_GS@h
52 bne _icswx_skip_guest
53
49 /* Enable icswx instruction */ 54 /* Enable icswx instruction */
50 mfspr r3,SPRN_A2_CCR2 55 mfspr r3,SPRN_A2_CCR2
51 ori r3,r3,A2_CCR2_ENABLE_ICSWX 56 ori r3,r3,A2_CCR2_ENABLE_ICSWX
@@ -54,7 +59,8 @@ _GLOBAL(__setup_cpu_a2)
54 /* Unmask all CTs in HACOP */ 59 /* Unmask all CTs in HACOP */
55 li r3,-1 60 li r3,-1
56 mtspr SPRN_HACOP,r3 61 mtspr SPRN_HACOP,r3
57#endif /* CONFIG_PPC_WSP_COPRO */ 62_icswx_skip_guest:
63#endif /* CONFIG_PPC_ICSWX */
58 64
59 /* Enable doorbell */ 65 /* Enable doorbell */
60 mfspr r3,SPRN_A2_CCR2 66 mfspr r3,SPRN_A2_CCR2
diff --git a/arch/powerpc/kernel/cputable.c b/arch/powerpc/kernel/cputable.c
index edae5bb06f1f..81db9e2a8a20 100644
--- a/arch/powerpc/kernel/cputable.c
+++ b/arch/powerpc/kernel/cputable.c
@@ -1505,6 +1505,19 @@ static struct cpu_spec __initdata cpu_specs[] = {
1505 .machine_check = machine_check_4xx, 1505 .machine_check = machine_check_4xx,
1506 .platform = "ppc405", 1506 .platform = "ppc405",
1507 }, 1507 },
1508 { /* APM8018X */
1509 .pvr_mask = 0xffff0000,
1510 .pvr_value = 0x7ff11432,
1511 .cpu_name = "APM8018X",
1512 .cpu_features = CPU_FTRS_40X,
1513 .cpu_user_features = PPC_FEATURE_32 |
1514 PPC_FEATURE_HAS_MMU | PPC_FEATURE_HAS_4xxMAC,
1515 .mmu_features = MMU_FTR_TYPE_40x,
1516 .icache_bsize = 32,
1517 .dcache_bsize = 32,
1518 .machine_check = machine_check_4xx,
1519 .platform = "ppc405",
1520 },
1508 { /* default match */ 1521 { /* default match */
1509 .pvr_mask = 0x00000000, 1522 .pvr_mask = 0x00000000,
1510 .pvr_value = 0x00000000, 1523 .pvr_value = 0x00000000,
@@ -1830,6 +1843,20 @@ static struct cpu_spec __initdata cpu_specs[] = {
1830 .machine_check = machine_check_47x, 1843 .machine_check = machine_check_47x,
1831 .platform = "ppc470", 1844 .platform = "ppc470",
1832 }, 1845 },
1846 { /* 476fpe */
1847 .pvr_mask = 0xffff0000,
1848 .pvr_value = 0x7ff50000,
1849 .cpu_name = "476fpe",
1850 .cpu_features = CPU_FTRS_47X | CPU_FTR_476_DD2,
1851 .cpu_user_features = COMMON_USER_BOOKE |
1852 PPC_FEATURE_HAS_FPU,
1853 .mmu_features = MMU_FTR_TYPE_47x |
1854 MMU_FTR_USE_TLBIVAX_BCAST | MMU_FTR_LOCK_BCAST_INVAL,
1855 .icache_bsize = 32,
1856 .dcache_bsize = 128,
1857 .machine_check = machine_check_47x,
1858 .platform = "ppc470",
1859 },
1833 { /* 476 iss */ 1860 { /* 476 iss */
1834 .pvr_mask = 0xffff0000, 1861 .pvr_mask = 0xffff0000,
1835 .pvr_value = 0x00050000, 1862 .pvr_value = 0x00050000,
diff --git a/arch/powerpc/kernel/crash.c b/arch/powerpc/kernel/crash.c
index d879809d5c45..28be3452e67a 100644
--- a/arch/powerpc/kernel/crash.c
+++ b/arch/powerpc/kernel/crash.c
@@ -10,85 +10,85 @@
10 * 10 *
11 */ 11 */
12 12
13#undef DEBUG
14
15#include <linux/kernel.h> 13#include <linux/kernel.h>
16#include <linux/smp.h> 14#include <linux/smp.h>
17#include <linux/reboot.h> 15#include <linux/reboot.h>
18#include <linux/kexec.h> 16#include <linux/kexec.h>
19#include <linux/bootmem.h>
20#include <linux/export.h> 17#include <linux/export.h>
21#include <linux/crash_dump.h> 18#include <linux/crash_dump.h>
22#include <linux/delay.h> 19#include <linux/delay.h>
23#include <linux/elf.h>
24#include <linux/elfcore.h>
25#include <linux/init.h> 20#include <linux/init.h>
26#include <linux/irq.h> 21#include <linux/irq.h>
27#include <linux/types.h> 22#include <linux/types.h>
28#include <linux/memblock.h>
29 23
30#include <asm/processor.h> 24#include <asm/processor.h>
31#include <asm/machdep.h> 25#include <asm/machdep.h>
32#include <asm/kexec.h> 26#include <asm/kexec.h>
33#include <asm/kdump.h> 27#include <asm/kdump.h>
34#include <asm/prom.h> 28#include <asm/prom.h>
35#include <asm/firmware.h>
36#include <asm/smp.h> 29#include <asm/smp.h>
37#include <asm/system.h> 30#include <asm/system.h>
38#include <asm/setjmp.h> 31#include <asm/setjmp.h>
39 32
40#ifdef DEBUG 33/*
41#include <asm/udbg.h> 34 * The primary CPU waits a while for all secondary CPUs to enter. This is to
42#define DBG(fmt...) udbg_printf(fmt) 35 * avoid sending an IPI if the secondary CPUs are entering
43#else 36 * crash_kexec_secondary on their own (eg via a system reset).
44#define DBG(fmt...) 37 *
45#endif 38 * The secondary timeout has to be longer than the primary. Both timeouts are
39 * in milliseconds.
40 */
41#define PRIMARY_TIMEOUT 500
42#define SECONDARY_TIMEOUT 1000
46 43
47/* This keeps a track of which one is crashing cpu. */ 44#define IPI_TIMEOUT 10000
45#define REAL_MODE_TIMEOUT 10000
46
47/* This keeps a track of which one is the crashing cpu. */
48int crashing_cpu = -1; 48int crashing_cpu = -1;
49static cpumask_t cpus_in_crash = CPU_MASK_NONE; 49static atomic_t cpus_in_crash;
50cpumask_t cpus_in_sr = CPU_MASK_NONE; 50static int time_to_dump;
51 51
52#define CRASH_HANDLER_MAX 3 52#define CRASH_HANDLER_MAX 3
53/* NULL terminated list of shutdown handles */ 53/* NULL terminated list of shutdown handles */
54static crash_shutdown_t crash_shutdown_handles[CRASH_HANDLER_MAX+1]; 54static crash_shutdown_t crash_shutdown_handles[CRASH_HANDLER_MAX+1];
55static DEFINE_SPINLOCK(crash_handlers_lock); 55static DEFINE_SPINLOCK(crash_handlers_lock);
56 56
57static unsigned long crash_shutdown_buf[JMP_BUF_LEN];
58static int crash_shutdown_cpu = -1;
59
60static int handle_fault(struct pt_regs *regs)
61{
62 if (crash_shutdown_cpu == smp_processor_id())
63 longjmp(crash_shutdown_buf, 1);
64 return 0;
65}
66
57#ifdef CONFIG_SMP 67#ifdef CONFIG_SMP
58static atomic_t enter_on_soft_reset = ATOMIC_INIT(0);
59 68
60void crash_ipi_callback(struct pt_regs *regs) 69void crash_ipi_callback(struct pt_regs *regs)
61{ 70{
71 static cpumask_t cpus_state_saved = CPU_MASK_NONE;
72
62 int cpu = smp_processor_id(); 73 int cpu = smp_processor_id();
63 74
64 if (!cpu_online(cpu)) 75 if (!cpu_online(cpu))
65 return; 76 return;
66 77
67 hard_irq_disable(); 78 hard_irq_disable();
68 if (!cpumask_test_cpu(cpu, &cpus_in_crash)) 79 if (!cpumask_test_cpu(cpu, &cpus_state_saved)) {
69 crash_save_cpu(regs, cpu); 80 crash_save_cpu(regs, cpu);
70 cpumask_set_cpu(cpu, &cpus_in_crash); 81 cpumask_set_cpu(cpu, &cpus_state_saved);
71
72 /*
73 * Entered via soft-reset - could be the kdump
74 * process is invoked using soft-reset or user activated
75 * it if some CPU did not respond to an IPI.
76 * For soft-reset, the secondary CPU can enter this func
77 * twice. 1 - using IPI, and 2. soft-reset.
78 * Tell the kexec CPU that entered via soft-reset and ready
79 * to go down.
80 */
81 if (cpumask_test_cpu(cpu, &cpus_in_sr)) {
82 cpumask_clear_cpu(cpu, &cpus_in_sr);
83 atomic_inc(&enter_on_soft_reset);
84 } 82 }
85 83
84 atomic_inc(&cpus_in_crash);
85 smp_mb__after_atomic_inc();
86
86 /* 87 /*
87 * Starting the kdump boot. 88 * Starting the kdump boot.
88 * This barrier is needed to make sure that all CPUs are stopped. 89 * This barrier is needed to make sure that all CPUs are stopped.
89 * If not, soft-reset will be invoked to bring other CPUs.
90 */ 90 */
91 while (!cpumask_test_cpu(crashing_cpu, &cpus_in_crash)) 91 while (!time_to_dump)
92 cpu_relax(); 92 cpu_relax();
93 93
94 if (ppc_md.kexec_cpu_down) 94 if (ppc_md.kexec_cpu_down)
@@ -103,106 +103,99 @@ void crash_ipi_callback(struct pt_regs *regs)
103 /* NOTREACHED */ 103 /* NOTREACHED */
104} 104}
105 105
106/*
107 * Wait until all CPUs are entered via soft-reset.
108 */
109static void crash_soft_reset_check(int cpu)
110{
111 unsigned int ncpus = num_online_cpus() - 1;/* Excluding the panic cpu */
112
113 cpumask_clear_cpu(cpu, &cpus_in_sr);
114 while (atomic_read(&enter_on_soft_reset) != ncpus)
115 cpu_relax();
116}
117
118
119static void crash_kexec_prepare_cpus(int cpu) 106static void crash_kexec_prepare_cpus(int cpu)
120{ 107{
121 unsigned int msecs; 108 unsigned int msecs;
122
123 unsigned int ncpus = num_online_cpus() - 1;/* Excluding the panic cpu */ 109 unsigned int ncpus = num_online_cpus() - 1;/* Excluding the panic cpu */
110 int tries = 0;
111 int (*old_handler)(struct pt_regs *regs);
112
113 printk(KERN_EMERG "Sending IPI to other CPUs\n");
124 114
125 crash_send_ipi(crash_ipi_callback); 115 crash_send_ipi(crash_ipi_callback);
126 smp_wmb(); 116 smp_wmb();
127 117
118again:
128 /* 119 /*
129 * FIXME: Until we will have the way to stop other CPUs reliably, 120 * FIXME: Until we will have the way to stop other CPUs reliably,
130 * the crash CPU will send an IPI and wait for other CPUs to 121 * the crash CPU will send an IPI and wait for other CPUs to
131 * respond. 122 * respond.
132 * Delay of at least 10 seconds.
133 */ 123 */
134 printk(KERN_EMERG "Sending IPI to other cpus...\n"); 124 msecs = IPI_TIMEOUT;
135 msecs = 10000; 125 while ((atomic_read(&cpus_in_crash) < ncpus) && (--msecs > 0))
136 while ((cpumask_weight(&cpus_in_crash) < ncpus) && (--msecs > 0)) {
137 cpu_relax();
138 mdelay(1); 126 mdelay(1);
139 }
140 127
141 /* Would it be better to replace the trap vector here? */ 128 /* Would it be better to replace the trap vector here? */
142 129
130 if (atomic_read(&cpus_in_crash) >= ncpus) {
131 printk(KERN_EMERG "IPI complete\n");
132 return;
133 }
134
135 printk(KERN_EMERG "ERROR: %d cpu(s) not responding\n",
136 ncpus - atomic_read(&cpus_in_crash));
137
143 /* 138 /*
144 * FIXME: In case if we do not get all CPUs, one possibility: ask the 139 * If we have a panic timeout set then we can't wait indefinitely
145 * user to do soft reset such that we get all. 140 * for someone to activate system reset. We also give up on the
146 * Soft-reset will be used until better mechanism is implemented. 141 * second time through if system reset fail to work.
147 */ 142 */
148 if (cpumask_weight(&cpus_in_crash) < ncpus) { 143 if ((panic_timeout > 0) || (tries > 0))
149 printk(KERN_EMERG "done waiting: %d cpu(s) not responding\n", 144 return;
150 ncpus - cpumask_weight(&cpus_in_crash)); 145
151 printk(KERN_EMERG "Activate soft-reset to stop other cpu(s)\n");
152 cpumask_clear(&cpus_in_sr);
153 atomic_set(&enter_on_soft_reset, 0);
154 while (cpumask_weight(&cpus_in_crash) < ncpus)
155 cpu_relax();
156 }
157 /* 146 /*
158 * Make sure all CPUs are entered via soft-reset if the kdump is 147 * A system reset will cause all CPUs to take an 0x100 exception.
159 * invoked using soft-reset. 148 * The primary CPU returns here via setjmp, and the secondary
149 * CPUs reexecute the crash_kexec_secondary path.
160 */ 150 */
161 if (cpumask_test_cpu(cpu, &cpus_in_sr)) 151 old_handler = __debugger;
162 crash_soft_reset_check(cpu); 152 __debugger = handle_fault;
163 /* Leave the IPI callback set */ 153 crash_shutdown_cpu = smp_processor_id();
154
155 if (setjmp(crash_shutdown_buf) == 0) {
156 printk(KERN_EMERG "Activate system reset (dumprestart) "
157 "to stop other cpu(s)\n");
158
159 /*
160 * A system reset will force all CPUs to execute the
161 * crash code again. We need to reset cpus_in_crash so we
162 * wait for everyone to do this.
163 */
164 atomic_set(&cpus_in_crash, 0);
165 smp_mb();
166
167 while (atomic_read(&cpus_in_crash) < ncpus)
168 cpu_relax();
169 }
170
171 crash_shutdown_cpu = -1;
172 __debugger = old_handler;
173
174 tries++;
175 goto again;
164} 176}
165 177
166/* 178/*
167 * This function will be called by secondary cpus or by kexec cpu 179 * This function will be called by secondary cpus.
168 * if soft-reset is activated to stop some CPUs.
169 */ 180 */
170void crash_kexec_secondary(struct pt_regs *regs) 181void crash_kexec_secondary(struct pt_regs *regs)
171{ 182{
172 int cpu = smp_processor_id();
173 unsigned long flags; 183 unsigned long flags;
174 int msecs = 5; 184 int msecs = SECONDARY_TIMEOUT;
175 185
176 local_irq_save(flags); 186 local_irq_save(flags);
177 /* Wait 5ms if the kexec CPU is not entered yet. */ 187
188 /* Wait for the primary crash CPU to signal its progress */
178 while (crashing_cpu < 0) { 189 while (crashing_cpu < 0) {
179 if (--msecs < 0) { 190 if (--msecs < 0) {
180 /* 191 /* No response, kdump image may not have been loaded */
181 * Either kdump image is not loaded or
182 * kdump process is not started - Probably xmon
183 * exited using 'x'(exit and recover) or
184 * kexec_should_crash() failed for all running tasks.
185 */
186 cpumask_clear_cpu(cpu, &cpus_in_sr);
187 local_irq_restore(flags); 192 local_irq_restore(flags);
188 return; 193 return;
189 } 194 }
195
190 mdelay(1); 196 mdelay(1);
191 cpu_relax();
192 }
193 if (cpu == crashing_cpu) {
194 /*
195 * Panic CPU will enter this func only via soft-reset.
196 * Wait until all secondary CPUs entered and
197 * then start kexec boot.
198 */
199 crash_soft_reset_check(cpu);
200 cpumask_set_cpu(crashing_cpu, &cpus_in_crash);
201 if (ppc_md.kexec_cpu_down)
202 ppc_md.kexec_cpu_down(1, 0);
203 machine_kexec(kexec_crash_image);
204 /* NOTREACHED */
205 } 197 }
198
206 crash_ipi_callback(regs); 199 crash_ipi_callback(regs);
207} 200}
208 201
@@ -211,7 +204,7 @@ void crash_kexec_secondary(struct pt_regs *regs)
211static void crash_kexec_prepare_cpus(int cpu) 204static void crash_kexec_prepare_cpus(int cpu)
212{ 205{
213 /* 206 /*
214 * move the secondarys to us so that we can copy 207 * move the secondaries to us so that we can copy
215 * the new kernel 0-0x100 safely 208 * the new kernel 0-0x100 safely
216 * 209 *
217 * do this if kexec in setup.c ? 210 * do this if kexec in setup.c ?
@@ -225,7 +218,6 @@ static void crash_kexec_prepare_cpus(int cpu)
225 218
226void crash_kexec_secondary(struct pt_regs *regs) 219void crash_kexec_secondary(struct pt_regs *regs)
227{ 220{
228 cpumask_clear(&cpus_in_sr);
229} 221}
230#endif /* CONFIG_SMP */ 222#endif /* CONFIG_SMP */
231 223
@@ -236,7 +228,7 @@ static void crash_kexec_wait_realmode(int cpu)
236 unsigned int msecs; 228 unsigned int msecs;
237 int i; 229 int i;
238 230
239 msecs = 10000; 231 msecs = REAL_MODE_TIMEOUT;
240 for (i=0; i < nr_cpu_ids && msecs > 0; i++) { 232 for (i=0; i < nr_cpu_ids && msecs > 0; i++) {
241 if (i == cpu) 233 if (i == cpu)
242 continue; 234 continue;
@@ -308,22 +300,11 @@ int crash_shutdown_unregister(crash_shutdown_t handler)
308} 300}
309EXPORT_SYMBOL(crash_shutdown_unregister); 301EXPORT_SYMBOL(crash_shutdown_unregister);
310 302
311static unsigned long crash_shutdown_buf[JMP_BUF_LEN];
312static int crash_shutdown_cpu = -1;
313
314static int handle_fault(struct pt_regs *regs)
315{
316 if (crash_shutdown_cpu == smp_processor_id())
317 longjmp(crash_shutdown_buf, 1);
318 return 0;
319}
320
321void default_machine_crash_shutdown(struct pt_regs *regs) 303void default_machine_crash_shutdown(struct pt_regs *regs)
322{ 304{
323 unsigned int i; 305 unsigned int i;
324 int (*old_handler)(struct pt_regs *regs); 306 int (*old_handler)(struct pt_regs *regs);
325 307
326
327 /* 308 /*
328 * This function is only called after the system 309 * This function is only called after the system
329 * has panicked or is otherwise in a critical state. 310 * has panicked or is otherwise in a critical state.
@@ -341,15 +322,26 @@ void default_machine_crash_shutdown(struct pt_regs *regs)
341 * such that another IPI will not be sent. 322 * such that another IPI will not be sent.
342 */ 323 */
343 crashing_cpu = smp_processor_id(); 324 crashing_cpu = smp_processor_id();
344 crash_save_cpu(regs, crashing_cpu); 325
326 /*
327 * If we came in via system reset, wait a while for the secondary
328 * CPUs to enter.
329 */
330 if (TRAP(regs) == 0x100)
331 mdelay(PRIMARY_TIMEOUT);
332
345 crash_kexec_prepare_cpus(crashing_cpu); 333 crash_kexec_prepare_cpus(crashing_cpu);
346 cpumask_set_cpu(crashing_cpu, &cpus_in_crash); 334
335 crash_save_cpu(regs, crashing_cpu);
336
337 time_to_dump = 1;
338
347 crash_kexec_wait_realmode(crashing_cpu); 339 crash_kexec_wait_realmode(crashing_cpu);
348 340
349 machine_kexec_mask_interrupts(); 341 machine_kexec_mask_interrupts();
350 342
351 /* 343 /*
352 * Call registered shutdown routines savely. Swap out 344 * Call registered shutdown routines safely. Swap out
353 * __debugger_fault_handler, and replace on exit. 345 * __debugger_fault_handler, and replace on exit.
354 */ 346 */
355 old_handler = __debugger_fault_handler; 347 old_handler = __debugger_fault_handler;
diff --git a/arch/powerpc/kernel/crash_dump.c b/arch/powerpc/kernel/crash_dump.c
index 424afb6b8fba..b3ba5163eae2 100644
--- a/arch/powerpc/kernel/crash_dump.c
+++ b/arch/powerpc/kernel/crash_dump.c
@@ -28,7 +28,7 @@
28#define DBG(fmt...) 28#define DBG(fmt...)
29#endif 29#endif
30 30
31#ifndef CONFIG_RELOCATABLE 31#ifndef CONFIG_NONSTATIC_KERNEL
32void __init reserve_kdump_trampoline(void) 32void __init reserve_kdump_trampoline(void)
33{ 33{
34 memblock_reserve(0, KDUMP_RESERVE_LIMIT); 34 memblock_reserve(0, KDUMP_RESERVE_LIMIT);
@@ -67,7 +67,7 @@ void __init setup_kdump_trampoline(void)
67 67
68 DBG(" <- setup_kdump_trampoline()\n"); 68 DBG(" <- setup_kdump_trampoline()\n");
69} 69}
70#endif /* CONFIG_RELOCATABLE */ 70#endif /* CONFIG_NONSTATIC_KERNEL */
71 71
72static int __init parse_savemaxmem(char *p) 72static int __init parse_savemaxmem(char *p)
73{ 73{
diff --git a/arch/powerpc/kernel/entry_32.S b/arch/powerpc/kernel/entry_32.S
index 56212bc0ab08..4f80cf1ce77b 100644
--- a/arch/powerpc/kernel/entry_32.S
+++ b/arch/powerpc/kernel/entry_32.S
@@ -215,7 +215,22 @@ reenable_mmu: /* re-enable mmu so we can */
215 stw r9,8(r1) 215 stw r9,8(r1)
216 stw r11,12(r1) 216 stw r11,12(r1)
217 stw r3,ORIG_GPR3(r1) 217 stw r3,ORIG_GPR3(r1)
218 /*
219 * The trace_hardirqs_off will use CALLER_ADDR0 and CALLER_ADDR1.
220 * If from user mode there is only one stack frame on the stack, and
221 * accessing CALLER_ADDR1 will cause oops. So we need create a dummy
222 * stack frame to make trace_hardirqs_off happy.
223 */
224 andi. r12,r12,MSR_PR
225 beq 11f
226 stwu r1,-16(r1)
227 bl trace_hardirqs_off
228 addi r1,r1,16
229 b 12f
230
23111:
218 bl trace_hardirqs_off 232 bl trace_hardirqs_off
23312:
219 lwz r0,GPR0(r1) 234 lwz r0,GPR0(r1)
220 lwz r3,ORIG_GPR3(r1) 235 lwz r3,ORIG_GPR3(r1)
221 lwz r4,GPR4(r1) 236 lwz r4,GPR4(r1)
diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S
index cf9c69b9189c..d4be7bb3dbdf 100644
--- a/arch/powerpc/kernel/exceptions-64s.S
+++ b/arch/powerpc/kernel/exceptions-64s.S
@@ -65,7 +65,7 @@ BEGIN_FTR_SECTION
65 lbz r0,PACAPROCSTART(r13) 65 lbz r0,PACAPROCSTART(r13)
66 cmpwi r0,0x80 66 cmpwi r0,0x80
67 bne 1f 67 bne 1f
68 li r0,0 68 li r0,1
69 stb r0,PACAPROCSTART(r13) 69 stb r0,PACAPROCSTART(r13)
70 b kvm_start_guest 70 b kvm_start_guest
711: 711:
diff --git a/arch/powerpc/kernel/head_44x.S b/arch/powerpc/kernel/head_44x.S
index b725dab0f88a..7dd2981bcc50 100644
--- a/arch/powerpc/kernel/head_44x.S
+++ b/arch/powerpc/kernel/head_44x.S
@@ -64,6 +64,35 @@ _ENTRY(_start);
64 mr r31,r3 /* save device tree ptr */ 64 mr r31,r3 /* save device tree ptr */
65 li r24,0 /* CPU number */ 65 li r24,0 /* CPU number */
66 66
67#ifdef CONFIG_RELOCATABLE
68/*
69 * Relocate ourselves to the current runtime address.
70 * This is called only by the Boot CPU.
71 * "relocate" is called with our current runtime virutal
72 * address.
73 * r21 will be loaded with the physical runtime address of _stext
74 */
75 bl 0f /* Get our runtime address */
760: mflr r21 /* Make it accessible */
77 addis r21,r21,(_stext - 0b)@ha
78 addi r21,r21,(_stext - 0b)@l /* Get our current runtime base */
79
80 /*
81 * We have the runtime (virutal) address of our base.
82 * We calculate our shift of offset from a 256M page.
83 * We could map the 256M page we belong to at PAGE_OFFSET and
84 * get going from there.
85 */
86 lis r4,KERNELBASE@h
87 ori r4,r4,KERNELBASE@l
88 rlwinm r6,r21,0,4,31 /* r6 = PHYS_START % 256M */
89 rlwinm r5,r4,0,4,31 /* r5 = KERNELBASE % 256M */
90 subf r3,r5,r6 /* r3 = r6 - r5 */
91 add r3,r4,r3 /* Required Virutal Address */
92
93 bl relocate
94#endif
95
67 bl init_cpu_state 96 bl init_cpu_state
68 97
69 /* 98 /*
@@ -88,6 +117,65 @@ _ENTRY(_start);
88 117
89#ifdef CONFIG_RELOCATABLE 118#ifdef CONFIG_RELOCATABLE
90 /* 119 /*
120 * Relocatable kernel support based on processing of dynamic
121 * relocation entries.
122 *
123 * r25 will contain RPN/ERPN for the start address of memory
124 * r21 will contain the current offset of _stext
125 */
126 lis r3,kernstart_addr@ha
127 la r3,kernstart_addr@l(r3)
128
129 /*
130 * Compute the kernstart_addr.
131 * kernstart_addr => (r6,r8)
132 * kernstart_addr & ~0xfffffff => (r6,r7)
133 */
134 rlwinm r6,r25,0,28,31 /* ERPN. Bits 32-35 of Address */
135 rlwinm r7,r25,0,0,3 /* RPN - assuming 256 MB page size */
136 rlwinm r8,r21,0,4,31 /* r8 = (_stext & 0xfffffff) */
137 or r8,r7,r8 /* Compute the lower 32bit of kernstart_addr */
138
139 /* Store kernstart_addr */
140 stw r6,0(r3) /* higher 32bit */
141 stw r8,4(r3) /* lower 32bit */
142
143 /*
144 * Compute the virt_phys_offset :
145 * virt_phys_offset = stext.run - kernstart_addr
146 *
147 * stext.run = (KERNELBASE & ~0xfffffff) + (kernstart_addr & 0xfffffff)
148 * When we relocate, we have :
149 *
150 * (kernstart_addr & 0xfffffff) = (stext.run & 0xfffffff)
151 *
152 * hence:
153 * virt_phys_offset = (KERNELBASE & ~0xfffffff) - (kernstart_addr & ~0xfffffff)
154 *
155 */
156
157 /* KERNELBASE&~0xfffffff => (r4,r5) */
158 li r4, 0 /* higer 32bit */
159 lis r5,KERNELBASE@h
160 rlwinm r5,r5,0,0,3 /* Align to 256M, lower 32bit */
161
162 /*
163 * 64bit subtraction.
164 */
165 subfc r5,r7,r5
166 subfe r4,r6,r4
167
168 /* Store virt_phys_offset */
169 lis r3,virt_phys_offset@ha
170 la r3,virt_phys_offset@l(r3)
171
172 stw r4,0(r3)
173 stw r5,4(r3)
174
175#elif defined(CONFIG_DYNAMIC_MEMSTART)
176 /*
177 * Mapping based, page aligned dynamic kernel loading.
178 *
91 * r25 will contain RPN/ERPN for the start address of memory 179 * r25 will contain RPN/ERPN for the start address of memory
92 * 180 *
93 * Add the difference between KERNELBASE and PAGE_OFFSET to the 181 * Add the difference between KERNELBASE and PAGE_OFFSET to the
@@ -732,6 +820,8 @@ _GLOBAL(init_cpu_state)
732 /* We use the PVR to differenciate 44x cores from 476 */ 820 /* We use the PVR to differenciate 44x cores from 476 */
733 mfspr r3,SPRN_PVR 821 mfspr r3,SPRN_PVR
734 srwi r3,r3,16 822 srwi r3,r3,16
823 cmplwi cr0,r3,PVR_476FPE@h
824 beq head_start_47x
735 cmplwi cr0,r3,PVR_476@h 825 cmplwi cr0,r3,PVR_476@h
736 beq head_start_47x 826 beq head_start_47x
737 cmplwi cr0,r3,PVR_476_ISS@h 827 cmplwi cr0,r3,PVR_476_ISS@h
@@ -800,12 +890,29 @@ skpinv: addi r4,r4,1 /* Increment */
800/* 890/*
801 * Configure and load pinned entry into TLB slot 63. 891 * Configure and load pinned entry into TLB slot 63.
802 */ 892 */
893#ifdef CONFIG_NONSTATIC_KERNEL
894 /*
895 * In case of a NONSTATIC_KERNEL we reuse the TLB XLAT
896 * entries of the initial mapping set by the boot loader.
897 * The XLAT entry is stored in r25
898 */
899
900 /* Read the XLAT entry for our current mapping */
901 tlbre r25,r23,PPC44x_TLB_XLAT
902
903 lis r3,KERNELBASE@h
904 ori r3,r3,KERNELBASE@l
905
906 /* Use our current RPN entry */
907 mr r4,r25
908#else
803 909
804 lis r3,PAGE_OFFSET@h 910 lis r3,PAGE_OFFSET@h
805 ori r3,r3,PAGE_OFFSET@l 911 ori r3,r3,PAGE_OFFSET@l
806 912
807 /* Kernel is at the base of RAM */ 913 /* Kernel is at the base of RAM */
808 li r4, 0 /* Load the kernel physical address */ 914 li r4, 0 /* Load the kernel physical address */
915#endif
809 916
810 /* Load the kernel PID = 0 */ 917 /* Load the kernel PID = 0 */
811 li r0,0 918 li r0,0
diff --git a/arch/powerpc/kernel/head_fsl_booke.S b/arch/powerpc/kernel/head_fsl_booke.S
index 9f5d210ddf3f..d5d78c4ceef6 100644
--- a/arch/powerpc/kernel/head_fsl_booke.S
+++ b/arch/powerpc/kernel/head_fsl_booke.S
@@ -197,7 +197,7 @@ _ENTRY(__early_start)
197 197
198 bl early_init 198 bl early_init
199 199
200#ifdef CONFIG_RELOCATABLE 200#ifdef CONFIG_DYNAMIC_MEMSTART
201 lis r3,kernstart_addr@ha 201 lis r3,kernstart_addr@ha
202 la r3,kernstart_addr@l(r3) 202 la r3,kernstart_addr@l(r3)
203#ifdef CONFIG_PHYS_64BIT 203#ifdef CONFIG_PHYS_64BIT
diff --git a/arch/powerpc/kernel/idle.c b/arch/powerpc/kernel/idle.c
index 39a2baa6ad58..7c66ce13da89 100644
--- a/arch/powerpc/kernel/idle.c
+++ b/arch/powerpc/kernel/idle.c
@@ -39,13 +39,23 @@
39#define cpu_should_die() 0 39#define cpu_should_die() 0
40#endif 40#endif
41 41
42unsigned long cpuidle_disable = IDLE_NO_OVERRIDE;
43EXPORT_SYMBOL(cpuidle_disable);
44
42static int __init powersave_off(char *arg) 45static int __init powersave_off(char *arg)
43{ 46{
44 ppc_md.power_save = NULL; 47 ppc_md.power_save = NULL;
48 cpuidle_disable = IDLE_POWERSAVE_OFF;
45 return 0; 49 return 0;
46} 50}
47__setup("powersave=off", powersave_off); 51__setup("powersave=off", powersave_off);
48 52
53#if defined(CONFIG_PPC_PSERIES) && defined(CONFIG_TRACEPOINTS)
54static const bool idle_uses_rcu = 1;
55#else
56static const bool idle_uses_rcu;
57#endif
58
49/* 59/*
50 * The body of the idle task. 60 * The body of the idle task.
51 */ 61 */
@@ -56,7 +66,10 @@ void cpu_idle(void)
56 66
57 set_thread_flag(TIF_POLLING_NRFLAG); 67 set_thread_flag(TIF_POLLING_NRFLAG);
58 while (1) { 68 while (1) {
59 tick_nohz_stop_sched_tick(1); 69 tick_nohz_idle_enter();
70 if (!idle_uses_rcu)
71 rcu_idle_enter();
72
60 while (!need_resched() && !cpu_should_die()) { 73 while (!need_resched() && !cpu_should_die()) {
61 ppc64_runlatch_off(); 74 ppc64_runlatch_off();
62 75
@@ -93,7 +106,9 @@ void cpu_idle(void)
93 106
94 HMT_medium(); 107 HMT_medium();
95 ppc64_runlatch_on(); 108 ppc64_runlatch_on();
96 tick_nohz_restart_sched_tick(); 109 if (!idle_uses_rcu)
110 rcu_idle_exit();
111 tick_nohz_idle_exit();
97 preempt_enable_no_resched(); 112 preempt_enable_no_resched();
98 if (cpu_should_die()) 113 if (cpu_should_die())
99 cpu_die(); 114 cpu_die();
@@ -102,6 +117,29 @@ void cpu_idle(void)
102 } 117 }
103} 118}
104 119
120
121/*
122 * cpu_idle_wait - Used to ensure that all the CPUs come out of the old
123 * idle loop and start using the new idle loop.
124 * Required while changing idle handler on SMP systems.
125 * Caller must have changed idle handler to the new value before the call.
126 * This window may be larger on shared systems.
127 */
128void cpu_idle_wait(void)
129{
130 int cpu;
131 smp_mb();
132
133 /* kick all the CPUs so that they exit out of old idle routine */
134 get_online_cpus();
135 for_each_online_cpu(cpu) {
136 if (cpu != smp_processor_id())
137 smp_send_reschedule(cpu);
138 }
139 put_online_cpus();
140}
141EXPORT_SYMBOL_GPL(cpu_idle_wait);
142
105int powersave_nap; 143int powersave_nap;
106 144
107#ifdef CONFIG_SYSCTL 145#ifdef CONFIG_SYSCTL
diff --git a/arch/powerpc/kernel/idle_power7.S b/arch/powerpc/kernel/idle_power7.S
index 3a70845a51c7..fcdff198da4b 100644
--- a/arch/powerpc/kernel/idle_power7.S
+++ b/arch/powerpc/kernel/idle_power7.S
@@ -54,6 +54,7 @@ _GLOBAL(power7_idle)
54 li r0,0 54 li r0,0
55 stb r0,PACASOFTIRQEN(r13) /* we'll hard-enable shortly */ 55 stb r0,PACASOFTIRQEN(r13) /* we'll hard-enable shortly */
56 stb r0,PACAHARDIRQEN(r13) 56 stb r0,PACAHARDIRQEN(r13)
57 stb r0,PACA_NAPSTATELOST(r13)
57 58
58 /* Continue saving state */ 59 /* Continue saving state */
59 SAVE_GPR(2, r1) 60 SAVE_GPR(2, r1)
@@ -86,6 +87,9 @@ _GLOBAL(power7_wakeup_loss)
86 rfid 87 rfid
87 88
88_GLOBAL(power7_wakeup_noloss) 89_GLOBAL(power7_wakeup_noloss)
90 lbz r0,PACA_NAPSTATELOST(r13)
91 cmpwi r0,0
92 bne .power7_wakeup_loss
89 ld r1,PACAR1(r13) 93 ld r1,PACAR1(r13)
90 ld r4,_MSR(r1) 94 ld r4,_MSR(r1)
91 ld r5,_NIP(r1) 95 ld r5,_NIP(r1)
diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
index 5c3c46948d94..701d4aceb4f4 100644
--- a/arch/powerpc/kernel/irq.c
+++ b/arch/powerpc/kernel/irq.c
@@ -115,6 +115,15 @@ static inline notrace void set_soft_enabled(unsigned long enable)
115 : : "r" (enable), "i" (offsetof(struct paca_struct, soft_enabled))); 115 : : "r" (enable), "i" (offsetof(struct paca_struct, soft_enabled)));
116} 116}
117 117
118static inline notrace void decrementer_check_overflow(void)
119{
120 u64 now = get_tb_or_rtc();
121 u64 *next_tb = &__get_cpu_var(decrementers_next_tb);
122
123 if (now >= *next_tb)
124 set_dec(1);
125}
126
118notrace void arch_local_irq_restore(unsigned long en) 127notrace void arch_local_irq_restore(unsigned long en)
119{ 128{
120 /* 129 /*
@@ -164,24 +173,21 @@ notrace void arch_local_irq_restore(unsigned long en)
164 */ 173 */
165 local_paca->hard_enabled = en; 174 local_paca->hard_enabled = en;
166 175
167#ifndef CONFIG_BOOKE 176 /*
168 /* On server, re-trigger the decrementer if it went negative since 177 * Trigger the decrementer if we have a pending event. Some processors
169 * some processors only trigger on edge transitions of the sign bit. 178 * only trigger on edge transitions of the sign bit. We might also
170 * 179 * have disabled interrupts long enough that the decrementer wrapped
171 * BookE has a level sensitive decrementer (latches in TSR) so we 180 * to positive.
172 * don't need that
173 */ 181 */
174 if ((int)mfspr(SPRN_DEC) < 0) 182 decrementer_check_overflow();
175 mtspr(SPRN_DEC, 1);
176#endif /* CONFIG_BOOKE */
177 183
178 /* 184 /*
179 * Force the delivery of pending soft-disabled interrupts on PS3. 185 * Force the delivery of pending soft-disabled interrupts on PS3.
180 * Any HV call will have this side effect. 186 * Any HV call will have this side effect.
181 */ 187 */
182 if (firmware_has_feature(FW_FEATURE_PS3_LV1)) { 188 if (firmware_has_feature(FW_FEATURE_PS3_LV1)) {
183 u64 tmp; 189 u64 tmp, tmp2;
184 lv1_get_version_info(&tmp); 190 lv1_get_version_info(&tmp, &tmp2);
185 } 191 }
186 192
187 __hard_irq_enable(); 193 __hard_irq_enable();
diff --git a/arch/powerpc/kernel/jump_label.c b/arch/powerpc/kernel/jump_label.c
index 368d158d665d..a1ed8a8c7cb4 100644
--- a/arch/powerpc/kernel/jump_label.c
+++ b/arch/powerpc/kernel/jump_label.c
@@ -11,6 +11,7 @@
11#include <linux/jump_label.h> 11#include <linux/jump_label.h>
12#include <asm/code-patching.h> 12#include <asm/code-patching.h>
13 13
14#ifdef HAVE_JUMP_LABEL
14void arch_jump_label_transform(struct jump_entry *entry, 15void arch_jump_label_transform(struct jump_entry *entry,
15 enum jump_label_type type) 16 enum jump_label_type type)
16{ 17{
@@ -21,3 +22,4 @@ void arch_jump_label_transform(struct jump_entry *entry,
21 else 22 else
22 patch_instruction(addr, PPC_INST_NOP); 23 patch_instruction(addr, PPC_INST_NOP);
23} 24}
25#endif
diff --git a/arch/powerpc/kernel/kvm.c b/arch/powerpc/kernel/kvm.c
index 35f27646c4ff..2985338d0e10 100644
--- a/arch/powerpc/kernel/kvm.c
+++ b/arch/powerpc/kernel/kvm.c
@@ -132,7 +132,6 @@ static void kvm_patch_ins_b(u32 *inst, int addr)
132 /* On relocatable kernels interrupts handlers and our code 132 /* On relocatable kernels interrupts handlers and our code
133 can be in different regions, so we don't patch them */ 133 can be in different regions, so we don't patch them */
134 134
135 extern u32 __end_interrupts;
136 if ((ulong)inst < (ulong)&__end_interrupts) 135 if ((ulong)inst < (ulong)&__end_interrupts)
137 return; 136 return;
138#endif 137#endif
diff --git a/arch/powerpc/kernel/lparcfg.c b/arch/powerpc/kernel/lparcfg.c
index 84daabe2fcba..578f35f18723 100644
--- a/arch/powerpc/kernel/lparcfg.c
+++ b/arch/powerpc/kernel/lparcfg.c
@@ -783,7 +783,7 @@ static const struct file_operations lparcfg_fops = {
783static int __init lparcfg_init(void) 783static int __init lparcfg_init(void)
784{ 784{
785 struct proc_dir_entry *ent; 785 struct proc_dir_entry *ent;
786 mode_t mode = S_IRUSR | S_IRGRP | S_IROTH; 786 umode_t mode = S_IRUSR | S_IRGRP | S_IROTH;
787 787
788 /* Allow writing if we have FW_FEATURE_SPLPAR */ 788 /* Allow writing if we have FW_FEATURE_SPLPAR */
789 if (firmware_has_feature(FW_FEATURE_SPLPAR) && 789 if (firmware_has_feature(FW_FEATURE_SPLPAR) &&
diff --git a/arch/powerpc/kernel/machine_kexec.c b/arch/powerpc/kernel/machine_kexec.c
index 9ce1672afb59..c957b1202bdc 100644
--- a/arch/powerpc/kernel/machine_kexec.c
+++ b/arch/powerpc/kernel/machine_kexec.c
@@ -107,9 +107,6 @@ void __init reserve_crashkernel(void)
107 unsigned long long crash_size, crash_base; 107 unsigned long long crash_size, crash_base;
108 int ret; 108 int ret;
109 109
110 /* this is necessary because of memblock_phys_mem_size() */
111 memblock_analyze();
112
113 /* use common parsing */ 110 /* use common parsing */
114 ret = parse_crashkernel(boot_command_line, memblock_phys_mem_size(), 111 ret = parse_crashkernel(boot_command_line, memblock_phys_mem_size(),
115 &crash_size, &crash_base); 112 &crash_size, &crash_base);
@@ -128,7 +125,7 @@ void __init reserve_crashkernel(void)
128 125
129 crash_size = resource_size(&crashk_res); 126 crash_size = resource_size(&crashk_res);
130 127
131#ifndef CONFIG_RELOCATABLE 128#ifndef CONFIG_NONSTATIC_KERNEL
132 if (crashk_res.start != KDUMP_KERNELBASE) 129 if (crashk_res.start != KDUMP_KERNELBASE)
133 printk("Crash kernel location must be 0x%x\n", 130 printk("Crash kernel location must be 0x%x\n",
134 KDUMP_KERNELBASE); 131 KDUMP_KERNELBASE);
diff --git a/arch/powerpc/kernel/misc_32.S b/arch/powerpc/kernel/misc_32.S
index f7d760ab5ca1..7cd07b42ca1a 100644
--- a/arch/powerpc/kernel/misc_32.S
+++ b/arch/powerpc/kernel/misc_32.S
@@ -738,7 +738,7 @@ relocate_new_kernel:
738 mr r5, r31 738 mr r5, r31
739 739
740 li r0, 0 740 li r0, 0
741#elif defined(CONFIG_44x) && !defined(CONFIG_47x) 741#elif defined(CONFIG_44x) && !defined(CONFIG_PPC_47x)
742 742
743/* 743/*
744 * Code for setting up 1:1 mapping for PPC440x for KEXEC 744 * Code for setting up 1:1 mapping for PPC440x for KEXEC
diff --git a/arch/powerpc/kernel/pci-common.c b/arch/powerpc/kernel/pci-common.c
index 458ed3bee663..fa4a573d6716 100644
--- a/arch/powerpc/kernel/pci-common.c
+++ b/arch/powerpc/kernel/pci-common.c
@@ -214,7 +214,7 @@ char __devinit *pcibios_setup(char *str)
214 * If the interrupt is used, then gets the interrupt line from the 214 * If the interrupt is used, then gets the interrupt line from the
215 * openfirmware and sets it in the pci_dev and pci_config line. 215 * openfirmware and sets it in the pci_dev and pci_config line.
216 */ 216 */
217int pci_read_irq_line(struct pci_dev *pci_dev) 217static int pci_read_irq_line(struct pci_dev *pci_dev)
218{ 218{
219 struct of_irq oirq; 219 struct of_irq oirq;
220 unsigned int virq; 220 unsigned int virq;
@@ -283,7 +283,6 @@ int pci_read_irq_line(struct pci_dev *pci_dev)
283 283
284 return 0; 284 return 0;
285} 285}
286EXPORT_SYMBOL(pci_read_irq_line);
287 286
288/* 287/*
289 * Platform support for /proc/bus/pci/X/Y mmap()s, 288 * Platform support for /proc/bus/pci/X/Y mmap()s,
@@ -921,18 +920,22 @@ static void __devinit pcibios_fixup_resources(struct pci_dev *dev)
921 struct resource *res = dev->resource + i; 920 struct resource *res = dev->resource + i;
922 if (!res->flags) 921 if (!res->flags)
923 continue; 922 continue;
924 /* On platforms that have PCI_PROBE_ONLY set, we don't 923
925 * consider 0 as an unassigned BAR value. It's technically 924 /* If we're going to re-assign everything, we mark all resources
926 * a valid value, but linux doesn't like it... so when we can 925 * as unset (and 0-base them). In addition, we mark BARs starting
927 * re-assign things, we do so, but if we can't, we keep it 926 * at 0 as unset as well, except if PCI_PROBE_ONLY is also set
928 * around and hope for the best... 927 * since in that case, we don't want to re-assign anything
929 */ 928 */
930 if (res->start == 0 && !pci_has_flag(PCI_PROBE_ONLY)) { 929 if (pci_has_flag(PCI_REASSIGN_ALL_RSRC) ||
931 pr_debug("PCI:%s Resource %d %016llx-%016llx [%x] is unassigned\n", 930 (res->start == 0 && !pci_has_flag(PCI_PROBE_ONLY))) {
932 pci_name(dev), i, 931 /* Only print message if not re-assigning */
933 (unsigned long long)res->start, 932 if (!pci_has_flag(PCI_REASSIGN_ALL_RSRC))
934 (unsigned long long)res->end, 933 pr_debug("PCI:%s Resource %d %016llx-%016llx [%x] "
935 (unsigned int)res->flags); 934 "is unassigned\n",
935 pci_name(dev), i,
936 (unsigned long long)res->start,
937 (unsigned long long)res->end,
938 (unsigned int)res->flags);
936 res->end -= res->start; 939 res->end -= res->start;
937 res->start = 0; 940 res->start = 0;
938 res->flags |= IORESOURCE_UNSET; 941 res->flags |= IORESOURCE_UNSET;
@@ -1042,6 +1045,16 @@ static void __devinit pcibios_fixup_bridge(struct pci_bus *bus)
1042 if (i >= 3 && bus->self->transparent) 1045 if (i >= 3 && bus->self->transparent)
1043 continue; 1046 continue;
1044 1047
1048 /* If we are going to re-assign everything, mark the resource
1049 * as unset and move it down to 0
1050 */
1051 if (pci_has_flag(PCI_REASSIGN_ALL_RSRC)) {
1052 res->flags |= IORESOURCE_UNSET;
1053 res->end -= res->start;
1054 res->start = 0;
1055 continue;
1056 }
1057
1045 pr_debug("PCI:%s Bus rsrc %d %016llx-%016llx [%x] fixup...\n", 1058 pr_debug("PCI:%s Bus rsrc %d %016llx-%016llx [%x] fixup...\n",
1046 pci_name(dev), i, 1059 pci_name(dev), i,
1047 (unsigned long long)res->start,\ 1060 (unsigned long long)res->start,\
@@ -1262,18 +1275,15 @@ void pcibios_allocate_bus_resources(struct pci_bus *bus)
1262 pci_bus_for_each_resource(bus, res, i) { 1275 pci_bus_for_each_resource(bus, res, i) {
1263 if (!res || !res->flags || res->start > res->end || res->parent) 1276 if (!res || !res->flags || res->start > res->end || res->parent)
1264 continue; 1277 continue;
1278
1279 /* If the resource was left unset at this point, we clear it */
1280 if (res->flags & IORESOURCE_UNSET)
1281 goto clear_resource;
1282
1265 if (bus->parent == NULL) 1283 if (bus->parent == NULL)
1266 pr = (res->flags & IORESOURCE_IO) ? 1284 pr = (res->flags & IORESOURCE_IO) ?
1267 &ioport_resource : &iomem_resource; 1285 &ioport_resource : &iomem_resource;
1268 else { 1286 else {
1269 /* Don't bother with non-root busses when
1270 * re-assigning all resources. We clear the
1271 * resource flags as if they were colliding
1272 * and as such ensure proper re-allocation
1273 * later.
1274 */
1275 if (pci_has_flag(PCI_REASSIGN_ALL_RSRC))
1276 goto clear_resource;
1277 pr = pci_find_parent_resource(bus->self, res); 1287 pr = pci_find_parent_resource(bus->self, res);
1278 if (pr == res) { 1288 if (pr == res) {
1279 /* this happens when the generic PCI 1289 /* this happens when the generic PCI
@@ -1304,9 +1314,9 @@ void pcibios_allocate_bus_resources(struct pci_bus *bus)
1304 if (reparent_resources(pr, res) == 0) 1314 if (reparent_resources(pr, res) == 0)
1305 continue; 1315 continue;
1306 } 1316 }
1307 printk(KERN_WARNING "PCI: Cannot allocate resource region " 1317 pr_warning("PCI: Cannot allocate resource region "
1308 "%d of PCI bridge %d, will remap\n", i, bus->number); 1318 "%d of PCI bridge %d, will remap\n", i, bus->number);
1309clear_resource: 1319 clear_resource:
1310 res->start = res->end = 0; 1320 res->start = res->end = 0;
1311 res->flags = 0; 1321 res->flags = 0;
1312 } 1322 }
@@ -1451,16 +1461,11 @@ void __init pcibios_resource_survey(void)
1451{ 1461{
1452 struct pci_bus *b; 1462 struct pci_bus *b;
1453 1463
1454 /* Allocate and assign resources. If we re-assign everything, then 1464 /* Allocate and assign resources */
1455 * we skip the allocate phase
1456 */
1457 list_for_each_entry(b, &pci_root_buses, node) 1465 list_for_each_entry(b, &pci_root_buses, node)
1458 pcibios_allocate_bus_resources(b); 1466 pcibios_allocate_bus_resources(b);
1459 1467 pcibios_allocate_resources(0);
1460 if (!pci_has_flag(PCI_REASSIGN_ALL_RSRC)) { 1468 pcibios_allocate_resources(1);
1461 pcibios_allocate_resources(0);
1462 pcibios_allocate_resources(1);
1463 }
1464 1469
1465 /* Before we start assigning unassigned resource, we try to reserve 1470 /* Before we start assigning unassigned resource, we try to reserve
1466 * the low IO area and the VGA memory area if they intersect the 1471 * the low IO area and the VGA memory area if they intersect the
@@ -1732,6 +1737,12 @@ void __devinit pcibios_scan_phb(struct pci_controller *hose)
1732 if (mode == PCI_PROBE_NORMAL) 1737 if (mode == PCI_PROBE_NORMAL)
1733 hose->last_busno = bus->subordinate = pci_scan_child_bus(bus); 1738 hose->last_busno = bus->subordinate = pci_scan_child_bus(bus);
1734 1739
1740 /* Platform gets a chance to do some global fixups before
1741 * we proceed to resource allocation
1742 */
1743 if (ppc_md.pcibios_fixup_phb)
1744 ppc_md.pcibios_fixup_phb(hose);
1745
1735 /* Configure PCI Express settings */ 1746 /* Configure PCI Express settings */
1736 if (bus && !pci_has_flag(PCI_PROBE_ONLY)) { 1747 if (bus && !pci_has_flag(PCI_PROBE_ONLY)) {
1737 struct pci_bus *child; 1748 struct pci_bus *child;
@@ -1747,10 +1758,13 @@ void __devinit pcibios_scan_phb(struct pci_controller *hose)
1747static void fixup_hide_host_resource_fsl(struct pci_dev *dev) 1758static void fixup_hide_host_resource_fsl(struct pci_dev *dev)
1748{ 1759{
1749 int i, class = dev->class >> 8; 1760 int i, class = dev->class >> 8;
1761 /* When configured as agent, programing interface = 1 */
1762 int prog_if = dev->class & 0xf;
1750 1763
1751 if ((class == PCI_CLASS_PROCESSOR_POWERPC || 1764 if ((class == PCI_CLASS_PROCESSOR_POWERPC ||
1752 class == PCI_CLASS_BRIDGE_OTHER) && 1765 class == PCI_CLASS_BRIDGE_OTHER) &&
1753 (dev->hdr_type == PCI_HEADER_TYPE_NORMAL) && 1766 (dev->hdr_type == PCI_HEADER_TYPE_NORMAL) &&
1767 (prog_if == 0) &&
1754 (dev->bus->parent == NULL)) { 1768 (dev->bus->parent == NULL)) {
1755 for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) { 1769 for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
1756 dev->resource[i].start = 0; 1770 dev->resource[i].start = 0;
diff --git a/arch/powerpc/kernel/pci_dn.c b/arch/powerpc/kernel/pci_dn.c
index 4e69deb89b37..dd9e4a04bf79 100644
--- a/arch/powerpc/kernel/pci_dn.c
+++ b/arch/powerpc/kernel/pci_dn.c
@@ -50,6 +50,9 @@ void * __devinit update_dn_pci_info(struct device_node *dn, void *data)
50 dn->data = pdn; 50 dn->data = pdn;
51 pdn->node = dn; 51 pdn->node = dn;
52 pdn->phb = phb; 52 pdn->phb = phb;
53#ifdef CONFIG_PPC_POWERNV
54 pdn->pe_number = IODA_INVALID_PE;
55#endif
53 regs = of_get_property(dn, "reg", NULL); 56 regs = of_get_property(dn, "reg", NULL);
54 if (regs) { 57 if (regs) {
55 /* First register entry is addr (00BBSS00) */ 58 /* First register entry is addr (00BBSS00) */
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 9054ca9ab4f9..ebe5766781aa 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -486,28 +486,6 @@ struct task_struct *__switch_to(struct task_struct *prev,
486 new_thread = &new->thread; 486 new_thread = &new->thread;
487 old_thread = &current->thread; 487 old_thread = &current->thread;
488 488
489#if defined(CONFIG_PPC_BOOK3E_64)
490 /* XXX Current Book3E code doesn't deal with kernel side DBCR0,
491 * we always hold the user values, so we set it now.
492 *
493 * However, we ensure the kernel MSR:DE is appropriately cleared too
494 * to avoid spurrious single step exceptions in the kernel.
495 *
496 * This will have to change to merge with the ppc32 code at some point,
497 * but I don't like much what ppc32 is doing today so there's some
498 * thinking needed there
499 */
500 if ((new_thread->dbcr0 | old_thread->dbcr0) & DBCR0_IDM) {
501 u32 dbcr0;
502
503 mtmsr(mfmsr() & ~MSR_DE);
504 isync();
505 dbcr0 = mfspr(SPRN_DBCR0);
506 dbcr0 = (dbcr0 & DBCR0_EDM) | new_thread->dbcr0;
507 mtspr(SPRN_DBCR0, dbcr0);
508 }
509#endif /* CONFIG_PPC64_BOOK3E */
510
511#ifdef CONFIG_PPC64 489#ifdef CONFIG_PPC64
512 /* 490 /*
513 * Collect processor utilization data per process 491 * Collect processor utilization data per process
@@ -606,16 +584,32 @@ static struct regbit {
606 unsigned long bit; 584 unsigned long bit;
607 const char *name; 585 const char *name;
608} msr_bits[] = { 586} msr_bits[] = {
587#if defined(CONFIG_PPC64) && !defined(CONFIG_BOOKE)
588 {MSR_SF, "SF"},
589 {MSR_HV, "HV"},
590#endif
591 {MSR_VEC, "VEC"},
592 {MSR_VSX, "VSX"},
593#ifdef CONFIG_BOOKE
594 {MSR_CE, "CE"},
595#endif
609 {MSR_EE, "EE"}, 596 {MSR_EE, "EE"},
610 {MSR_PR, "PR"}, 597 {MSR_PR, "PR"},
611 {MSR_FP, "FP"}, 598 {MSR_FP, "FP"},
612 {MSR_VEC, "VEC"},
613 {MSR_VSX, "VSX"},
614 {MSR_ME, "ME"}, 599 {MSR_ME, "ME"},
615 {MSR_CE, "CE"}, 600#ifdef CONFIG_BOOKE
616 {MSR_DE, "DE"}, 601 {MSR_DE, "DE"},
602#else
603 {MSR_SE, "SE"},
604 {MSR_BE, "BE"},
605#endif
617 {MSR_IR, "IR"}, 606 {MSR_IR, "IR"},
618 {MSR_DR, "DR"}, 607 {MSR_DR, "DR"},
608 {MSR_PMM, "PMM"},
609#ifndef CONFIG_BOOKE
610 {MSR_RI, "RI"},
611 {MSR_LE, "LE"},
612#endif
619 {0, NULL} 613 {0, NULL}
620}; 614};
621 615
@@ -657,7 +651,7 @@ void show_regs(struct pt_regs * regs)
657 if ((regs->trap != 0xc00) && cpu_has_feature(CPU_FTR_CFAR)) 651 if ((regs->trap != 0xc00) && cpu_has_feature(CPU_FTR_CFAR))
658 printk("CFAR: "REG"\n", regs->orig_gpr3); 652 printk("CFAR: "REG"\n", regs->orig_gpr3);
659 if (trap == 0x300 || trap == 0x600) 653 if (trap == 0x300 || trap == 0x600)
660#ifdef CONFIG_PPC_ADV_DEBUG_REGS 654#if defined(CONFIG_4xx) || defined(CONFIG_BOOKE)
661 printk("DEAR: "REG", ESR: "REG"\n", regs->dar, regs->dsisr); 655 printk("DEAR: "REG", ESR: "REG"\n", regs->dar, regs->dsisr);
662#else 656#else
663 printk("DAR: "REG", DSISR: %08lx\n", regs->dar, regs->dsisr); 657 printk("DAR: "REG", DSISR: %08lx\n", regs->dar, regs->dsisr);
diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index fa1235b0503b..abe405dab34d 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c
@@ -733,8 +733,6 @@ void __init early_init_devtree(void *params)
733 of_scan_flat_dt(early_init_dt_scan_chosen_ppc, cmd_line); 733 of_scan_flat_dt(early_init_dt_scan_chosen_ppc, cmd_line);
734 734
735 /* Scan memory nodes and rebuild MEMBLOCKs */ 735 /* Scan memory nodes and rebuild MEMBLOCKs */
736 memblock_init();
737
738 of_scan_flat_dt(early_init_dt_scan_root, NULL); 736 of_scan_flat_dt(early_init_dt_scan_root, NULL);
739 of_scan_flat_dt(early_init_dt_scan_memory_ppc, NULL); 737 of_scan_flat_dt(early_init_dt_scan_memory_ppc, NULL);
740 738
@@ -756,20 +754,14 @@ void __init early_init_devtree(void *params)
756 early_reserve_mem(); 754 early_reserve_mem();
757 phyp_dump_reserve_mem(); 755 phyp_dump_reserve_mem();
758 756
759 limit = memory_limit; 757 /*
760 if (! limit) { 758 * Ensure that total memory size is page-aligned, because otherwise
761 phys_addr_t memsize; 759 * mark_bootmem() gets upset.
762 760 */
763 /* Ensure that total memory size is page-aligned, because 761 limit = ALIGN(memory_limit ?: memblock_phys_mem_size(), PAGE_SIZE);
764 * otherwise mark_bootmem() gets upset. */
765 memblock_analyze();
766 memsize = memblock_phys_mem_size();
767 if ((memsize & PAGE_MASK) != memsize)
768 limit = memsize & PAGE_MASK;
769 }
770 memblock_enforce_memory_limit(limit); 762 memblock_enforce_memory_limit(limit);
771 763
772 memblock_analyze(); 764 memblock_allow_resize();
773 memblock_dump_all(); 765 memblock_dump_all();
774 766
775 DBG("Phys. mem: %llx\n", memblock_phys_mem_size()); 767 DBG("Phys. mem: %llx\n", memblock_phys_mem_size());
diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_init.c
index b4fa66127495..eca626ea3f23 100644
--- a/arch/powerpc/kernel/prom_init.c
+++ b/arch/powerpc/kernel/prom_init.c
@@ -742,7 +742,7 @@ static unsigned char ibm_architecture_vec[] = {
742 W(0xffffffff), /* virt_base */ 742 W(0xffffffff), /* virt_base */
743 W(0xffffffff), /* virt_size */ 743 W(0xffffffff), /* virt_size */
744 W(0xffffffff), /* load_base */ 744 W(0xffffffff), /* load_base */
745 W(64), /* 64MB min RMA */ 745 W(256), /* 256MB min RMA */
746 W(0xffffffff), /* full client load */ 746 W(0xffffffff), /* full client load */
747 0, /* min RMA percentage of total RAM */ 747 0, /* min RMA percentage of total RAM */
748 48, /* max log_2(hash table size) */ 748 48, /* max log_2(hash table size) */
@@ -1224,14 +1224,6 @@ static void __init prom_init_mem(void)
1224 1224
1225 RELOC(alloc_bottom) = PAGE_ALIGN((unsigned long)&RELOC(_end) + 0x4000); 1225 RELOC(alloc_bottom) = PAGE_ALIGN((unsigned long)&RELOC(_end) + 0x4000);
1226 1226
1227 /* Check if we have an initrd after the kernel, if we do move our bottom
1228 * point to after it
1229 */
1230 if (RELOC(prom_initrd_start)) {
1231 if (RELOC(prom_initrd_end) > RELOC(alloc_bottom))
1232 RELOC(alloc_bottom) = PAGE_ALIGN(RELOC(prom_initrd_end));
1233 }
1234
1235 /* 1227 /*
1236 * If prom_memory_limit is set we reduce the upper limits *except* for 1228 * If prom_memory_limit is set we reduce the upper limits *except* for
1237 * alloc_top_high. This must be the real top of RAM so we can put 1229 * alloc_top_high. This must be the real top of RAM so we can put
@@ -1269,6 +1261,15 @@ static void __init prom_init_mem(void)
1269 RELOC(alloc_top) = RELOC(rmo_top); 1261 RELOC(alloc_top) = RELOC(rmo_top);
1270 RELOC(alloc_top_high) = RELOC(ram_top); 1262 RELOC(alloc_top_high) = RELOC(ram_top);
1271 1263
1264 /*
1265 * Check if we have an initrd after the kernel but still inside
1266 * the RMO. If we do move our bottom point to after it.
1267 */
1268 if (RELOC(prom_initrd_start) &&
1269 RELOC(prom_initrd_start) < RELOC(rmo_top) &&
1270 RELOC(prom_initrd_end) > RELOC(alloc_bottom))
1271 RELOC(alloc_bottom) = PAGE_ALIGN(RELOC(prom_initrd_end));
1272
1272 prom_printf("memory layout at init:\n"); 1273 prom_printf("memory layout at init:\n");
1273 prom_printf(" memory_limit : %x (16 MB aligned)\n", RELOC(prom_memory_limit)); 1274 prom_printf(" memory_limit : %x (16 MB aligned)\n", RELOC(prom_memory_limit));
1274 prom_printf(" alloc_bottom : %x\n", RELOC(alloc_bottom)); 1275 prom_printf(" alloc_bottom : %x\n", RELOC(alloc_bottom));
@@ -1579,10 +1580,8 @@ static void __init prom_instantiate_rtas(void)
1579 return; 1580 return;
1580 1581
1581 base = alloc_down(size, PAGE_SIZE, 0); 1582 base = alloc_down(size, PAGE_SIZE, 0);
1582 if (base == 0) { 1583 if (base == 0)
1583 prom_printf("RTAS allocation failed !\n"); 1584 prom_panic("Could not allocate memory for RTAS\n");
1584 return;
1585 }
1586 1585
1587 rtas_inst = call_prom("open", 1, 1, ADDR("/rtas")); 1586 rtas_inst = call_prom("open", 1, 1, ADDR("/rtas"));
1588 if (!IHANDLE_VALID(rtas_inst)) { 1587 if (!IHANDLE_VALID(rtas_inst)) {
@@ -2081,7 +2080,7 @@ static void __init prom_check_displays(void)
2081 /* Setup a usable color table when the appropriate 2080 /* Setup a usable color table when the appropriate
2082 * method is available. Should update this to set-colors */ 2081 * method is available. Should update this to set-colors */
2083 clut = RELOC(default_colors); 2082 clut = RELOC(default_colors);
2084 for (i = 0; i < 32; i++, clut += 3) 2083 for (i = 0; i < 16; i++, clut += 3)
2085 if (prom_set_color(ih, i, clut[0], clut[1], 2084 if (prom_set_color(ih, i, clut[0], clut[1],
2086 clut[2]) != 0) 2085 clut[2]) != 0)
2087 break; 2086 break;
@@ -2846,7 +2845,7 @@ unsigned long __init prom_init(unsigned long r3, unsigned long r4,
2846 RELOC(of_platform) = prom_find_machine_type(); 2845 RELOC(of_platform) = prom_find_machine_type();
2847 prom_printf("Detected machine type: %x\n", RELOC(of_platform)); 2846 prom_printf("Detected machine type: %x\n", RELOC(of_platform));
2848 2847
2849#ifndef CONFIG_RELOCATABLE 2848#ifndef CONFIG_NONSTATIC_KERNEL
2850 /* Bail if this is a kdump kernel. */ 2849 /* Bail if this is a kdump kernel. */
2851 if (PHYSICAL_START > 0) 2850 if (PHYSICAL_START > 0)
2852 prom_panic("Error: You can't boot a kdump kernel from OF!\n"); 2851 prom_panic("Error: You can't boot a kdump kernel from OF!\n");
@@ -2971,9 +2970,11 @@ unsigned long __init prom_init(unsigned long r3, unsigned long r4,
2971 /* 2970 /*
2972 * in case stdin is USB and still active on IBM machines... 2971 * in case stdin is USB and still active on IBM machines...
2973 * Unfortunately quiesce crashes on some powermacs if we have 2972 * Unfortunately quiesce crashes on some powermacs if we have
2974 * closed stdin already (in particular the powerbook 101). 2973 * closed stdin already (in particular the powerbook 101). It
2974 * appears that the OPAL version of OFW doesn't like it either.
2975 */ 2975 */
2976 if (RELOC(of_platform) != PLATFORM_POWERMAC) 2976 if (RELOC(of_platform) != PLATFORM_POWERMAC &&
2977 RELOC(of_platform) != PLATFORM_OPAL)
2977 prom_close_stdin(); 2978 prom_close_stdin();
2978 2979
2979 /* 2980 /*
@@ -2989,8 +2990,12 @@ unsigned long __init prom_init(unsigned long r3, unsigned long r4,
2989 * is common to us and kexec 2990 * is common to us and kexec
2990 */ 2991 */
2991 hdr = RELOC(dt_header_start); 2992 hdr = RELOC(dt_header_start);
2992 prom_printf("returning from prom_init\n"); 2993
2993 prom_debug("->dt_header_start=0x%x\n", hdr); 2994 /* Don't print anything after quiesce under OPAL, it crashes OFW */
2995 if (RELOC(of_platform) != PLATFORM_OPAL) {
2996 prom_printf("returning from prom_init\n");
2997 prom_debug("->dt_header_start=0x%x\n", hdr);
2998 }
2994 2999
2995#ifdef CONFIG_PPC32 3000#ifdef CONFIG_PPC32
2996 reloc_got2(-offset); 3001 reloc_got2(-offset);
diff --git a/arch/powerpc/kernel/reloc_32.S b/arch/powerpc/kernel/reloc_32.S
new file mode 100644
index 000000000000..ef46ba6e094f
--- /dev/null
+++ b/arch/powerpc/kernel/reloc_32.S
@@ -0,0 +1,208 @@
1/*
2 * Code to process dynamic relocations for PPC32.
3 *
4 * Copyrights (C) IBM Corporation, 2011.
5 * Author: Suzuki Poulose <suzuki@in.ibm.com>
6 *
7 * - Based on ppc64 code - reloc_64.S
8 *
9 * This program is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU General Public License
11 * as published by the Free Software Foundation; either version
12 * 2 of the License, or (at your option) any later version.
13 */
14
15#include <asm/ppc_asm.h>
16
17/* Dynamic section table entry tags */
18DT_RELA = 7 /* Tag for Elf32_Rela section */
19DT_RELASZ = 8 /* Size of the Rela relocs */
20DT_RELAENT = 9 /* Size of one Rela reloc entry */
21
22STN_UNDEF = 0 /* Undefined symbol index */
23STB_LOCAL = 0 /* Local binding for the symbol */
24
25R_PPC_ADDR16_LO = 4 /* Lower half of (S+A) */
26R_PPC_ADDR16_HI = 5 /* Upper half of (S+A) */
27R_PPC_ADDR16_HA = 6 /* High Adjusted (S+A) */
28R_PPC_RELATIVE = 22
29
30/*
31 * r3 = desired final address
32 */
33
34_GLOBAL(relocate)
35
36 mflr r0 /* Save our LR */
37 bl 0f /* Find our current runtime address */
380: mflr r12 /* Make it accessible */
39 mtlr r0
40
41 lwz r11, (p_dyn - 0b)(r12)
42 add r11, r11, r12 /* runtime address of .dynamic section */
43 lwz r9, (p_rela - 0b)(r12)
44 add r9, r9, r12 /* runtime address of .rela.dyn section */
45 lwz r10, (p_st - 0b)(r12)
46 add r10, r10, r12 /* runtime address of _stext section */
47 lwz r13, (p_sym - 0b)(r12)
48 add r13, r13, r12 /* runtime address of .dynsym section */
49
50 /*
51 * Scan the dynamic section for RELA, RELASZ entries
52 */
53 li r6, 0
54 li r7, 0
55 li r8, 0
561: lwz r5, 0(r11) /* ELF_Dyn.d_tag */
57 cmpwi r5, 0 /* End of ELF_Dyn[] */
58 beq eodyn
59 cmpwi r5, DT_RELA
60 bne relasz
61 lwz r7, 4(r11) /* r7 = rela.link */
62 b skip
63relasz:
64 cmpwi r5, DT_RELASZ
65 bne relaent
66 lwz r8, 4(r11) /* r8 = Total Rela relocs size */
67 b skip
68relaent:
69 cmpwi r5, DT_RELAENT
70 bne skip
71 lwz r6, 4(r11) /* r6 = Size of one Rela reloc */
72skip:
73 addi r11, r11, 8
74 b 1b
75eodyn: /* End of Dyn Table scan */
76
77 /* Check if we have found all the entries */
78 cmpwi r7, 0
79 beq done
80 cmpwi r8, 0
81 beq done
82 cmpwi r6, 0
83 beq done
84
85
86 /*
87 * Work out the current offset from the link time address of .rela
88 * section.
89 * cur_offset[r7] = rela.run[r9] - rela.link [r7]
90 * _stext.link[r12] = _stext.run[r10] - cur_offset[r7]
91 * final_offset[r3] = _stext.final[r3] - _stext.link[r12]
92 */
93 subf r7, r7, r9 /* cur_offset */
94 subf r12, r7, r10
95 subf r3, r12, r3 /* final_offset */
96
97 subf r8, r6, r8 /* relaz -= relaent */
98 /*
99 * Scan through the .rela table and process each entry
100 * r9 - points to the current .rela table entry
101 * r13 - points to the symbol table
102 */
103
104 /*
105 * Check if we have a relocation based on symbol
106 * r5 will hold the value of the symbol.
107 */
108applyrela:
109 lwz r4, 4(r9) /* r4 = rela.r_info */
110 srwi r5, r4, 8 /* ELF32_R_SYM(r_info) */
111 cmpwi r5, STN_UNDEF /* sym == STN_UNDEF ? */
112 beq get_type /* value = 0 */
113 /* Find the value of the symbol at index(r5) */
114 slwi r5, r5, 4 /* r5 = r5 * sizeof(Elf32_Sym) */
115 add r12, r13, r5 /* r12 = &__dyn_sym[Index] */
116
117 /*
118 * GNU ld has a bug, where dynamic relocs based on
119 * STB_LOCAL symbols, the value should be assumed
120 * to be zero. - Alan Modra
121 */
122 /* XXX: Do we need to check if we are using GNU ld ? */
123 lbz r5, 12(r12) /* r5 = dyn_sym[Index].st_info */
124 extrwi r5, r5, 4, 24 /* r5 = ELF32_ST_BIND(r5) */
125 cmpwi r5, STB_LOCAL /* st_value = 0, ld bug */
126 beq get_type /* We have r5 = 0 */
127 lwz r5, 4(r12) /* r5 = __dyn_sym[Index].st_value */
128
129get_type:
130 /* Load the relocation type to r4 */
131 extrwi r4, r4, 8, 24 /* r4 = ELF32_R_TYPE(r_info) = ((char*)r4)[3] */
132
133 /* R_PPC_RELATIVE */
134 cmpwi r4, R_PPC_RELATIVE
135 bne hi16
136 lwz r4, 0(r9) /* r_offset */
137 lwz r0, 8(r9) /* r_addend */
138 add r0, r0, r3 /* final addend */
139 stwx r0, r4, r7 /* memory[r4+r7]) = (u32)r0 */
140 b nxtrela /* continue */
141
142 /* R_PPC_ADDR16_HI */
143hi16:
144 cmpwi r4, R_PPC_ADDR16_HI
145 bne ha16
146 lwz r4, 0(r9) /* r_offset */
147 lwz r0, 8(r9) /* r_addend */
148 add r0, r0, r3
149 add r0, r0, r5 /* r0 = (S+A+Offset) */
150 extrwi r0, r0, 16, 0 /* r0 = (r0 >> 16) */
151 b store_half
152
153 /* R_PPC_ADDR16_HA */
154ha16:
155 cmpwi r4, R_PPC_ADDR16_HA
156 bne lo16
157 lwz r4, 0(r9) /* r_offset */
158 lwz r0, 8(r9) /* r_addend */
159 add r0, r0, r3
160 add r0, r0, r5 /* r0 = (S+A+Offset) */
161 extrwi r5, r0, 1, 16 /* Extract bit 16 */
162 extrwi r0, r0, 16, 0 /* r0 = (r0 >> 16) */
163 add r0, r0, r5 /* Add it to r0 */
164 b store_half
165
166 /* R_PPC_ADDR16_LO */
167lo16:
168 cmpwi r4, R_PPC_ADDR16_LO
169 bne nxtrela
170 lwz r4, 0(r9) /* r_offset */
171 lwz r0, 8(r9) /* r_addend */
172 add r0, r0, r3
173 add r0, r0, r5 /* r0 = (S+A+Offset) */
174 extrwi r0, r0, 16, 16 /* r0 &= 0xffff */
175 /* Fall through to */
176
177 /* Store half word */
178store_half:
179 sthx r0, r4, r7 /* memory[r4+r7] = (u16)r0 */
180
181nxtrela:
182 /*
183 * We have to flush the modified instructions to the
184 * main storage from the d-cache. And also, invalidate the
185 * cached instructions in i-cache which has been modified.
186 *
187 * We delay the sync / isync operation till the end, since
188 * we won't be executing the modified instructions until
189 * we return from here.
190 */
191 dcbst r4,r7
192 sync /* Ensure the data is flushed before icbi */
193 icbi r4,r7
194 cmpwi r8, 0 /* relasz = 0 ? */
195 ble done
196 add r9, r9, r6 /* move to next entry in the .rela table */
197 subf r8, r6, r8 /* relasz -= relaent */
198 b applyrela
199
200done:
201 sync /* Wait for the flush to finish */
202 isync /* Discard prefetched instructions */
203 blr
204
205p_dyn: .long __dynamic_start - 0b
206p_rela: .long __rela_dyn_start - 0b
207p_sym: .long __dynamic_symtab - 0b
208p_st: .long _stext - 0b
diff --git a/arch/powerpc/kernel/rtas_flash.c b/arch/powerpc/kernel/rtas_flash.c
index e037c7494fd8..4174b4b23246 100644
--- a/arch/powerpc/kernel/rtas_flash.c
+++ b/arch/powerpc/kernel/rtas_flash.c
@@ -568,6 +568,12 @@ static void rtas_flash_firmware(int reboot_type)
568 } 568 }
569 569
570 /* 570 /*
571 * Just before starting the firmware flash, cancel the event scan work
572 * to avoid any soft lockup issues.
573 */
574 rtas_cancel_event_scan();
575
576 /*
571 * NOTE: the "first" block must be under 4GB, so we create 577 * NOTE: the "first" block must be under 4GB, so we create
572 * an entry with no data blocks in the reserved buffer in 578 * an entry with no data blocks in the reserved buffer in
573 * the kernel data segment. 579 * the kernel data segment.
diff --git a/arch/powerpc/kernel/rtasd.c b/arch/powerpc/kernel/rtasd.c
index 481ef064c8f1..1045ff49cc6d 100644
--- a/arch/powerpc/kernel/rtasd.c
+++ b/arch/powerpc/kernel/rtasd.c
@@ -472,6 +472,13 @@ static void start_event_scan(void)
472 &event_scan_work, event_scan_delay); 472 &event_scan_work, event_scan_delay);
473} 473}
474 474
475/* Cancel the rtas event scan work */
476void rtas_cancel_event_scan(void)
477{
478 cancel_delayed_work_sync(&event_scan_work);
479}
480EXPORT_SYMBOL_GPL(rtas_cancel_event_scan);
481
475static int __init rtas_init(void) 482static int __init rtas_init(void)
476{ 483{
477 struct proc_dir_entry *entry; 484 struct proc_dir_entry *entry;
diff --git a/arch/powerpc/kernel/setup_32.c b/arch/powerpc/kernel/setup_32.c
index c1ce86357ecb..ac7610815113 100644
--- a/arch/powerpc/kernel/setup_32.c
+++ b/arch/powerpc/kernel/setup_32.c
@@ -107,6 +107,8 @@ notrace unsigned long __init early_init(unsigned long dt_ptr)
107 PTRRELOC(&__start___lwsync_fixup), 107 PTRRELOC(&__start___lwsync_fixup),
108 PTRRELOC(&__stop___lwsync_fixup)); 108 PTRRELOC(&__stop___lwsync_fixup));
109 109
110 do_final_fixups();
111
110 return KERNELBASE + offset; 112 return KERNELBASE + offset;
111} 113}
112 114
diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c
index 1a9dea80a69b..4cb8f1e9d044 100644
--- a/arch/powerpc/kernel/setup_64.c
+++ b/arch/powerpc/kernel/setup_64.c
@@ -35,6 +35,8 @@
35#include <linux/pci.h> 35#include <linux/pci.h>
36#include <linux/lockdep.h> 36#include <linux/lockdep.h>
37#include <linux/memblock.h> 37#include <linux/memblock.h>
38#include <linux/hugetlb.h>
39
38#include <asm/io.h> 40#include <asm/io.h>
39#include <asm/kdump.h> 41#include <asm/kdump.h>
40#include <asm/prom.h> 42#include <asm/prom.h>
@@ -64,6 +66,7 @@
64#include <asm/mmu_context.h> 66#include <asm/mmu_context.h>
65#include <asm/code-patching.h> 67#include <asm/code-patching.h>
66#include <asm/kvm_ppc.h> 68#include <asm/kvm_ppc.h>
69#include <asm/hugetlb.h>
67 70
68#include "setup.h" 71#include "setup.h"
69 72
@@ -217,6 +220,13 @@ void __init early_setup(unsigned long dt_ptr)
217 /* Initialize the hash table or TLB handling */ 220 /* Initialize the hash table or TLB handling */
218 early_init_mmu(); 221 early_init_mmu();
219 222
223 /*
224 * Reserve any gigantic pages requested on the command line.
225 * memblock needs to have been initialized by the time this is
226 * called since this will reserve memory.
227 */
228 reserve_hugetlb_gpages();
229
220 DBG(" <- early_setup()\n"); 230 DBG(" <- early_setup()\n");
221} 231}
222 232
@@ -359,6 +369,7 @@ void __init setup_system(void)
359 &__start___fw_ftr_fixup, &__stop___fw_ftr_fixup); 369 &__start___fw_ftr_fixup, &__stop___fw_ftr_fixup);
360 do_lwsync_fixups(cur_cpu_spec->cpu_features, 370 do_lwsync_fixups(cur_cpu_spec->cpu_features,
361 &__start___lwsync_fixup, &__stop___lwsync_fixup); 371 &__start___lwsync_fixup, &__stop___lwsync_fixup);
372 do_final_fixups();
362 373
363 /* 374 /*
364 * Unflatten the device-tree passed by prom_init or kexec 375 * Unflatten the device-tree passed by prom_init or kexec
diff --git a/arch/powerpc/kernel/signal_32.c b/arch/powerpc/kernel/signal_32.c
index 78b76dc54dfb..836a5a19eb2c 100644
--- a/arch/powerpc/kernel/signal_32.c
+++ b/arch/powerpc/kernel/signal_32.c
@@ -97,7 +97,7 @@ static inline int put_sigset_t(compat_sigset_t __user *uset, sigset_t *set)
97 compat_sigset_t cset; 97 compat_sigset_t cset;
98 98
99 switch (_NSIG_WORDS) { 99 switch (_NSIG_WORDS) {
100 case 4: cset.sig[5] = set->sig[3] & 0xffffffffull; 100 case 4: cset.sig[6] = set->sig[3] & 0xffffffffull;
101 cset.sig[7] = set->sig[3] >> 32; 101 cset.sig[7] = set->sig[3] >> 32;
102 case 3: cset.sig[4] = set->sig[2] & 0xffffffffull; 102 case 3: cset.sig[4] = set->sig[2] & 0xffffffffull;
103 cset.sig[5] = set->sig[2] >> 32; 103 cset.sig[5] = set->sig[2] >> 32;
diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c
index 6df70907d60a..46695febc09f 100644
--- a/arch/powerpc/kernel/smp.c
+++ b/arch/powerpc/kernel/smp.c
@@ -27,7 +27,7 @@
27#include <linux/spinlock.h> 27#include <linux/spinlock.h>
28#include <linux/cache.h> 28#include <linux/cache.h>
29#include <linux/err.h> 29#include <linux/err.h>
30#include <linux/sysdev.h> 30#include <linux/device.h>
31#include <linux/cpu.h> 31#include <linux/cpu.h>
32#include <linux/notifier.h> 32#include <linux/notifier.h>
33#include <linux/topology.h> 33#include <linux/topology.h>
@@ -187,7 +187,8 @@ int smp_request_message_ipi(int virq, int msg)
187 return 1; 187 return 1;
188 } 188 }
189#endif 189#endif
190 err = request_irq(virq, smp_ipi_action[msg], IRQF_PERCPU, 190 err = request_irq(virq, smp_ipi_action[msg],
191 IRQF_PERCPU | IRQF_NO_THREAD,
191 smp_ipi_name[msg], 0); 192 smp_ipi_name[msg], 0);
192 WARN(err < 0, "unable to request_irq %d for %s (rc %d)\n", 193 WARN(err < 0, "unable to request_irq %d for %s (rc %d)\n",
193 virq, smp_ipi_name[msg], err); 194 virq, smp_ipi_name[msg], err);
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index ce035c1905f0..883e74c0d1b3 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -1,4 +1,4 @@
1#include <linux/sysdev.h> 1#include <linux/device.h>
2#include <linux/cpu.h> 2#include <linux/cpu.h>
3#include <linux/smp.h> 3#include <linux/smp.h>
4#include <linux/percpu.h> 4#include <linux/percpu.h>
@@ -18,6 +18,7 @@
18#include <asm/machdep.h> 18#include <asm/machdep.h>
19#include <asm/smp.h> 19#include <asm/smp.h>
20#include <asm/pmc.h> 20#include <asm/pmc.h>
21#include <asm/system.h>
21 22
22#include "cacheinfo.h" 23#include "cacheinfo.h"
23 24
@@ -37,12 +38,12 @@ static DEFINE_PER_CPU(struct cpu, cpu_devices);
37/* Time in microseconds we delay before sleeping in the idle loop */ 38/* Time in microseconds we delay before sleeping in the idle loop */
38DEFINE_PER_CPU(long, smt_snooze_delay) = { 100 }; 39DEFINE_PER_CPU(long, smt_snooze_delay) = { 100 };
39 40
40static ssize_t store_smt_snooze_delay(struct sys_device *dev, 41static ssize_t store_smt_snooze_delay(struct device *dev,
41 struct sysdev_attribute *attr, 42 struct device_attribute *attr,
42 const char *buf, 43 const char *buf,
43 size_t count) 44 size_t count)
44{ 45{
45 struct cpu *cpu = container_of(dev, struct cpu, sysdev); 46 struct cpu *cpu = container_of(dev, struct cpu, dev);
46 ssize_t ret; 47 ssize_t ret;
47 long snooze; 48 long snooze;
48 49
@@ -50,21 +51,22 @@ static ssize_t store_smt_snooze_delay(struct sys_device *dev,
50 if (ret != 1) 51 if (ret != 1)
51 return -EINVAL; 52 return -EINVAL;
52 53
53 per_cpu(smt_snooze_delay, cpu->sysdev.id) = snooze; 54 per_cpu(smt_snooze_delay, cpu->dev.id) = snooze;
55 update_smt_snooze_delay(snooze);
54 56
55 return count; 57 return count;
56} 58}
57 59
58static ssize_t show_smt_snooze_delay(struct sys_device *dev, 60static ssize_t show_smt_snooze_delay(struct device *dev,
59 struct sysdev_attribute *attr, 61 struct device_attribute *attr,
60 char *buf) 62 char *buf)
61{ 63{
62 struct cpu *cpu = container_of(dev, struct cpu, sysdev); 64 struct cpu *cpu = container_of(dev, struct cpu, dev);
63 65
64 return sprintf(buf, "%ld\n", per_cpu(smt_snooze_delay, cpu->sysdev.id)); 66 return sprintf(buf, "%ld\n", per_cpu(smt_snooze_delay, cpu->dev.id));
65} 67}
66 68
67static SYSDEV_ATTR(smt_snooze_delay, 0644, show_smt_snooze_delay, 69static DEVICE_ATTR(smt_snooze_delay, 0644, show_smt_snooze_delay,
68 store_smt_snooze_delay); 70 store_smt_snooze_delay);
69 71
70static int __init setup_smt_snooze_delay(char *str) 72static int __init setup_smt_snooze_delay(char *str)
@@ -117,25 +119,25 @@ static void write_##NAME(void *val) \
117 ppc_enable_pmcs(); \ 119 ppc_enable_pmcs(); \
118 mtspr(ADDRESS, *(unsigned long *)val); \ 120 mtspr(ADDRESS, *(unsigned long *)val); \
119} \ 121} \
120static ssize_t show_##NAME(struct sys_device *dev, \ 122static ssize_t show_##NAME(struct device *dev, \
121 struct sysdev_attribute *attr, \ 123 struct device_attribute *attr, \
122 char *buf) \ 124 char *buf) \
123{ \ 125{ \
124 struct cpu *cpu = container_of(dev, struct cpu, sysdev); \ 126 struct cpu *cpu = container_of(dev, struct cpu, dev); \
125 unsigned long val; \ 127 unsigned long val; \
126 smp_call_function_single(cpu->sysdev.id, read_##NAME, &val, 1); \ 128 smp_call_function_single(cpu->dev.id, read_##NAME, &val, 1); \
127 return sprintf(buf, "%lx\n", val); \ 129 return sprintf(buf, "%lx\n", val); \
128} \ 130} \
129static ssize_t __used \ 131static ssize_t __used \
130 store_##NAME(struct sys_device *dev, struct sysdev_attribute *attr, \ 132 store_##NAME(struct device *dev, struct device_attribute *attr, \
131 const char *buf, size_t count) \ 133 const char *buf, size_t count) \
132{ \ 134{ \
133 struct cpu *cpu = container_of(dev, struct cpu, sysdev); \ 135 struct cpu *cpu = container_of(dev, struct cpu, dev); \
134 unsigned long val; \ 136 unsigned long val; \
135 int ret = sscanf(buf, "%lx", &val); \ 137 int ret = sscanf(buf, "%lx", &val); \
136 if (ret != 1) \ 138 if (ret != 1) \
137 return -EINVAL; \ 139 return -EINVAL; \
138 smp_call_function_single(cpu->sysdev.id, write_##NAME, &val, 1); \ 140 smp_call_function_single(cpu->dev.id, write_##NAME, &val, 1); \
139 return count; \ 141 return count; \
140} 142}
141 143
@@ -177,23 +179,25 @@ SYSFS_PMCSETUP(mmcra, SPRN_MMCRA);
177SYSFS_PMCSETUP(purr, SPRN_PURR); 179SYSFS_PMCSETUP(purr, SPRN_PURR);
178SYSFS_PMCSETUP(spurr, SPRN_SPURR); 180SYSFS_PMCSETUP(spurr, SPRN_SPURR);
179SYSFS_PMCSETUP(dscr, SPRN_DSCR); 181SYSFS_PMCSETUP(dscr, SPRN_DSCR);
182SYSFS_PMCSETUP(pir, SPRN_PIR);
180 183
181static SYSDEV_ATTR(mmcra, 0600, show_mmcra, store_mmcra); 184static DEVICE_ATTR(mmcra, 0600, show_mmcra, store_mmcra);
182static SYSDEV_ATTR(spurr, 0600, show_spurr, NULL); 185static DEVICE_ATTR(spurr, 0600, show_spurr, NULL);
183static SYSDEV_ATTR(dscr, 0600, show_dscr, store_dscr); 186static DEVICE_ATTR(dscr, 0600, show_dscr, store_dscr);
184static SYSDEV_ATTR(purr, 0600, show_purr, store_purr); 187static DEVICE_ATTR(purr, 0600, show_purr, store_purr);
188static DEVICE_ATTR(pir, 0400, show_pir, NULL);
185 189
186unsigned long dscr_default = 0; 190unsigned long dscr_default = 0;
187EXPORT_SYMBOL(dscr_default); 191EXPORT_SYMBOL(dscr_default);
188 192
189static ssize_t show_dscr_default(struct sysdev_class *class, 193static ssize_t show_dscr_default(struct device *dev,
190 struct sysdev_class_attribute *attr, char *buf) 194 struct device_attribute *attr, char *buf)
191{ 195{
192 return sprintf(buf, "%lx\n", dscr_default); 196 return sprintf(buf, "%lx\n", dscr_default);
193} 197}
194 198
195static ssize_t __used store_dscr_default(struct sysdev_class *class, 199static ssize_t __used store_dscr_default(struct device *dev,
196 struct sysdev_class_attribute *attr, const char *buf, 200 struct device_attribute *attr, const char *buf,
197 size_t count) 201 size_t count)
198{ 202{
199 unsigned long val; 203 unsigned long val;
@@ -207,15 +211,14 @@ static ssize_t __used store_dscr_default(struct sysdev_class *class,
207 return count; 211 return count;
208} 212}
209 213
210static SYSDEV_CLASS_ATTR(dscr_default, 0600, 214static DEVICE_ATTR(dscr_default, 0600,
211 show_dscr_default, store_dscr_default); 215 show_dscr_default, store_dscr_default);
212 216
213static void sysfs_create_dscr_default(void) 217static void sysfs_create_dscr_default(void)
214{ 218{
215 int err = 0; 219 int err = 0;
216 if (cpu_has_feature(CPU_FTR_DSCR)) 220 if (cpu_has_feature(CPU_FTR_DSCR))
217 err = sysfs_create_file(&cpu_sysdev_class.kset.kobj, 221 err = device_create_file(cpu_subsys.dev_root, &dev_attr_dscr_default);
218 &attr_dscr_default.attr);
219} 222}
220#endif /* CONFIG_PPC64 */ 223#endif /* CONFIG_PPC64 */
221 224
@@ -259,72 +262,72 @@ SYSFS_PMCSETUP(tsr3, SPRN_PA6T_TSR3);
259#endif /* HAS_PPC_PMC_PA6T */ 262#endif /* HAS_PPC_PMC_PA6T */
260 263
261#ifdef HAS_PPC_PMC_IBM 264#ifdef HAS_PPC_PMC_IBM
262static struct sysdev_attribute ibm_common_attrs[] = { 265static struct device_attribute ibm_common_attrs[] = {
263 _SYSDEV_ATTR(mmcr0, 0600, show_mmcr0, store_mmcr0), 266 __ATTR(mmcr0, 0600, show_mmcr0, store_mmcr0),
264 _SYSDEV_ATTR(mmcr1, 0600, show_mmcr1, store_mmcr1), 267 __ATTR(mmcr1, 0600, show_mmcr1, store_mmcr1),
265}; 268};
266#endif /* HAS_PPC_PMC_G4 */ 269#endif /* HAS_PPC_PMC_G4 */
267 270
268#ifdef HAS_PPC_PMC_G4 271#ifdef HAS_PPC_PMC_G4
269static struct sysdev_attribute g4_common_attrs[] = { 272static struct device_attribute g4_common_attrs[] = {
270 _SYSDEV_ATTR(mmcr0, 0600, show_mmcr0, store_mmcr0), 273 __ATTR(mmcr0, 0600, show_mmcr0, store_mmcr0),
271 _SYSDEV_ATTR(mmcr1, 0600, show_mmcr1, store_mmcr1), 274 __ATTR(mmcr1, 0600, show_mmcr1, store_mmcr1),
272 _SYSDEV_ATTR(mmcr2, 0600, show_mmcr2, store_mmcr2), 275 __ATTR(mmcr2, 0600, show_mmcr2, store_mmcr2),
273}; 276};
274#endif /* HAS_PPC_PMC_G4 */ 277#endif /* HAS_PPC_PMC_G4 */
275 278
276static struct sysdev_attribute classic_pmc_attrs[] = { 279static struct device_attribute classic_pmc_attrs[] = {
277 _SYSDEV_ATTR(pmc1, 0600, show_pmc1, store_pmc1), 280 __ATTR(pmc1, 0600, show_pmc1, store_pmc1),
278 _SYSDEV_ATTR(pmc2, 0600, show_pmc2, store_pmc2), 281 __ATTR(pmc2, 0600, show_pmc2, store_pmc2),
279 _SYSDEV_ATTR(pmc3, 0600, show_pmc3, store_pmc3), 282 __ATTR(pmc3, 0600, show_pmc3, store_pmc3),
280 _SYSDEV_ATTR(pmc4, 0600, show_pmc4, store_pmc4), 283 __ATTR(pmc4, 0600, show_pmc4, store_pmc4),
281 _SYSDEV_ATTR(pmc5, 0600, show_pmc5, store_pmc5), 284 __ATTR(pmc5, 0600, show_pmc5, store_pmc5),
282 _SYSDEV_ATTR(pmc6, 0600, show_pmc6, store_pmc6), 285 __ATTR(pmc6, 0600, show_pmc6, store_pmc6),
283#ifdef CONFIG_PPC64 286#ifdef CONFIG_PPC64
284 _SYSDEV_ATTR(pmc7, 0600, show_pmc7, store_pmc7), 287 __ATTR(pmc7, 0600, show_pmc7, store_pmc7),
285 _SYSDEV_ATTR(pmc8, 0600, show_pmc8, store_pmc8), 288 __ATTR(pmc8, 0600, show_pmc8, store_pmc8),
286#endif 289#endif
287}; 290};
288 291
289#ifdef HAS_PPC_PMC_PA6T 292#ifdef HAS_PPC_PMC_PA6T
290static struct sysdev_attribute pa6t_attrs[] = { 293static struct device_attribute pa6t_attrs[] = {
291 _SYSDEV_ATTR(mmcr0, 0600, show_mmcr0, store_mmcr0), 294 __ATTR(mmcr0, 0600, show_mmcr0, store_mmcr0),
292 _SYSDEV_ATTR(mmcr1, 0600, show_mmcr1, store_mmcr1), 295 __ATTR(mmcr1, 0600, show_mmcr1, store_mmcr1),
293 _SYSDEV_ATTR(pmc0, 0600, show_pa6t_pmc0, store_pa6t_pmc0), 296 __ATTR(pmc0, 0600, show_pa6t_pmc0, store_pa6t_pmc0),
294 _SYSDEV_ATTR(pmc1, 0600, show_pa6t_pmc1, store_pa6t_pmc1), 297 __ATTR(pmc1, 0600, show_pa6t_pmc1, store_pa6t_pmc1),
295 _SYSDEV_ATTR(pmc2, 0600, show_pa6t_pmc2, store_pa6t_pmc2), 298 __ATTR(pmc2, 0600, show_pa6t_pmc2, store_pa6t_pmc2),
296 _SYSDEV_ATTR(pmc3, 0600, show_pa6t_pmc3, store_pa6t_pmc3), 299 __ATTR(pmc3, 0600, show_pa6t_pmc3, store_pa6t_pmc3),
297 _SYSDEV_ATTR(pmc4, 0600, show_pa6t_pmc4, store_pa6t_pmc4), 300 __ATTR(pmc4, 0600, show_pa6t_pmc4, store_pa6t_pmc4),
298 _SYSDEV_ATTR(pmc5, 0600, show_pa6t_pmc5, store_pa6t_pmc5), 301 __ATTR(pmc5, 0600, show_pa6t_pmc5, store_pa6t_pmc5),
299#ifdef CONFIG_DEBUG_KERNEL 302#ifdef CONFIG_DEBUG_KERNEL
300 _SYSDEV_ATTR(hid0, 0600, show_hid0, store_hid0), 303 __ATTR(hid0, 0600, show_hid0, store_hid0),
301 _SYSDEV_ATTR(hid1, 0600, show_hid1, store_hid1), 304 __ATTR(hid1, 0600, show_hid1, store_hid1),
302 _SYSDEV_ATTR(hid4, 0600, show_hid4, store_hid4), 305 __ATTR(hid4, 0600, show_hid4, store_hid4),
303 _SYSDEV_ATTR(hid5, 0600, show_hid5, store_hid5), 306 __ATTR(hid5, 0600, show_hid5, store_hid5),
304 _SYSDEV_ATTR(ima0, 0600, show_ima0, store_ima0), 307 __ATTR(ima0, 0600, show_ima0, store_ima0),
305 _SYSDEV_ATTR(ima1, 0600, show_ima1, store_ima1), 308 __ATTR(ima1, 0600, show_ima1, store_ima1),
306 _SYSDEV_ATTR(ima2, 0600, show_ima2, store_ima2), 309 __ATTR(ima2, 0600, show_ima2, store_ima2),
307 _SYSDEV_ATTR(ima3, 0600, show_ima3, store_ima3), 310 __ATTR(ima3, 0600, show_ima3, store_ima3),
308 _SYSDEV_ATTR(ima4, 0600, show_ima4, store_ima4), 311 __ATTR(ima4, 0600, show_ima4, store_ima4),
309 _SYSDEV_ATTR(ima5, 0600, show_ima5, store_ima5), 312 __ATTR(ima5, 0600, show_ima5, store_ima5),
310 _SYSDEV_ATTR(ima6, 0600, show_ima6, store_ima6), 313 __ATTR(ima6, 0600, show_ima6, store_ima6),
311 _SYSDEV_ATTR(ima7, 0600, show_ima7, store_ima7), 314 __ATTR(ima7, 0600, show_ima7, store_ima7),
312 _SYSDEV_ATTR(ima8, 0600, show_ima8, store_ima8), 315 __ATTR(ima8, 0600, show_ima8, store_ima8),
313 _SYSDEV_ATTR(ima9, 0600, show_ima9, store_ima9), 316 __ATTR(ima9, 0600, show_ima9, store_ima9),
314 _SYSDEV_ATTR(imaat, 0600, show_imaat, store_imaat), 317 __ATTR(imaat, 0600, show_imaat, store_imaat),
315 _SYSDEV_ATTR(btcr, 0600, show_btcr, store_btcr), 318 __ATTR(btcr, 0600, show_btcr, store_btcr),
316 _SYSDEV_ATTR(pccr, 0600, show_pccr, store_pccr), 319 __ATTR(pccr, 0600, show_pccr, store_pccr),
317 _SYSDEV_ATTR(rpccr, 0600, show_rpccr, store_rpccr), 320 __ATTR(rpccr, 0600, show_rpccr, store_rpccr),
318 _SYSDEV_ATTR(der, 0600, show_der, store_der), 321 __ATTR(der, 0600, show_der, store_der),
319 _SYSDEV_ATTR(mer, 0600, show_mer, store_mer), 322 __ATTR(mer, 0600, show_mer, store_mer),
320 _SYSDEV_ATTR(ber, 0600, show_ber, store_ber), 323 __ATTR(ber, 0600, show_ber, store_ber),
321 _SYSDEV_ATTR(ier, 0600, show_ier, store_ier), 324 __ATTR(ier, 0600, show_ier, store_ier),
322 _SYSDEV_ATTR(sier, 0600, show_sier, store_sier), 325 __ATTR(sier, 0600, show_sier, store_sier),
323 _SYSDEV_ATTR(siar, 0600, show_siar, store_siar), 326 __ATTR(siar, 0600, show_siar, store_siar),
324 _SYSDEV_ATTR(tsr0, 0600, show_tsr0, store_tsr0), 327 __ATTR(tsr0, 0600, show_tsr0, store_tsr0),
325 _SYSDEV_ATTR(tsr1, 0600, show_tsr1, store_tsr1), 328 __ATTR(tsr1, 0600, show_tsr1, store_tsr1),
326 _SYSDEV_ATTR(tsr2, 0600, show_tsr2, store_tsr2), 329 __ATTR(tsr2, 0600, show_tsr2, store_tsr2),
327 _SYSDEV_ATTR(tsr3, 0600, show_tsr3, store_tsr3), 330 __ATTR(tsr3, 0600, show_tsr3, store_tsr3),
328#endif /* CONFIG_DEBUG_KERNEL */ 331#endif /* CONFIG_DEBUG_KERNEL */
329}; 332};
330#endif /* HAS_PPC_PMC_PA6T */ 333#endif /* HAS_PPC_PMC_PA6T */
@@ -333,14 +336,14 @@ static struct sysdev_attribute pa6t_attrs[] = {
333static void __cpuinit register_cpu_online(unsigned int cpu) 336static void __cpuinit register_cpu_online(unsigned int cpu)
334{ 337{
335 struct cpu *c = &per_cpu(cpu_devices, cpu); 338 struct cpu *c = &per_cpu(cpu_devices, cpu);
336 struct sys_device *s = &c->sysdev; 339 struct device *s = &c->dev;
337 struct sysdev_attribute *attrs, *pmc_attrs; 340 struct device_attribute *attrs, *pmc_attrs;
338 int i, nattrs; 341 int i, nattrs;
339 342
340#ifdef CONFIG_PPC64 343#ifdef CONFIG_PPC64
341 if (!firmware_has_feature(FW_FEATURE_ISERIES) && 344 if (!firmware_has_feature(FW_FEATURE_ISERIES) &&
342 cpu_has_feature(CPU_FTR_SMT)) 345 cpu_has_feature(CPU_FTR_SMT))
343 sysdev_create_file(s, &attr_smt_snooze_delay); 346 device_create_file(s, &dev_attr_smt_snooze_delay);
344#endif 347#endif
345 348
346 /* PMC stuff */ 349 /* PMC stuff */
@@ -348,14 +351,14 @@ static void __cpuinit register_cpu_online(unsigned int cpu)
348#ifdef HAS_PPC_PMC_IBM 351#ifdef HAS_PPC_PMC_IBM
349 case PPC_PMC_IBM: 352 case PPC_PMC_IBM:
350 attrs = ibm_common_attrs; 353 attrs = ibm_common_attrs;
351 nattrs = sizeof(ibm_common_attrs) / sizeof(struct sysdev_attribute); 354 nattrs = sizeof(ibm_common_attrs) / sizeof(struct device_attribute);
352 pmc_attrs = classic_pmc_attrs; 355 pmc_attrs = classic_pmc_attrs;
353 break; 356 break;
354#endif /* HAS_PPC_PMC_IBM */ 357#endif /* HAS_PPC_PMC_IBM */
355#ifdef HAS_PPC_PMC_G4 358#ifdef HAS_PPC_PMC_G4
356 case PPC_PMC_G4: 359 case PPC_PMC_G4:
357 attrs = g4_common_attrs; 360 attrs = g4_common_attrs;
358 nattrs = sizeof(g4_common_attrs) / sizeof(struct sysdev_attribute); 361 nattrs = sizeof(g4_common_attrs) / sizeof(struct device_attribute);
359 pmc_attrs = classic_pmc_attrs; 362 pmc_attrs = classic_pmc_attrs;
360 break; 363 break;
361#endif /* HAS_PPC_PMC_G4 */ 364#endif /* HAS_PPC_PMC_G4 */
@@ -363,7 +366,7 @@ static void __cpuinit register_cpu_online(unsigned int cpu)
363 case PPC_PMC_PA6T: 366 case PPC_PMC_PA6T:
364 /* PA Semi starts counting at PMC0 */ 367 /* PA Semi starts counting at PMC0 */
365 attrs = pa6t_attrs; 368 attrs = pa6t_attrs;
366 nattrs = sizeof(pa6t_attrs) / sizeof(struct sysdev_attribute); 369 nattrs = sizeof(pa6t_attrs) / sizeof(struct device_attribute);
367 pmc_attrs = NULL; 370 pmc_attrs = NULL;
368 break; 371 break;
369#endif /* HAS_PPC_PMC_PA6T */ 372#endif /* HAS_PPC_PMC_PA6T */
@@ -374,24 +377,27 @@ static void __cpuinit register_cpu_online(unsigned int cpu)
374 } 377 }
375 378
376 for (i = 0; i < nattrs; i++) 379 for (i = 0; i < nattrs; i++)
377 sysdev_create_file(s, &attrs[i]); 380 device_create_file(s, &attrs[i]);
378 381
379 if (pmc_attrs) 382 if (pmc_attrs)
380 for (i = 0; i < cur_cpu_spec->num_pmcs; i++) 383 for (i = 0; i < cur_cpu_spec->num_pmcs; i++)
381 sysdev_create_file(s, &pmc_attrs[i]); 384 device_create_file(s, &pmc_attrs[i]);
382 385
383#ifdef CONFIG_PPC64 386#ifdef CONFIG_PPC64
384 if (cpu_has_feature(CPU_FTR_MMCRA)) 387 if (cpu_has_feature(CPU_FTR_MMCRA))
385 sysdev_create_file(s, &attr_mmcra); 388 device_create_file(s, &dev_attr_mmcra);
386 389
387 if (cpu_has_feature(CPU_FTR_PURR)) 390 if (cpu_has_feature(CPU_FTR_PURR))
388 sysdev_create_file(s, &attr_purr); 391 device_create_file(s, &dev_attr_purr);
389 392
390 if (cpu_has_feature(CPU_FTR_SPURR)) 393 if (cpu_has_feature(CPU_FTR_SPURR))
391 sysdev_create_file(s, &attr_spurr); 394 device_create_file(s, &dev_attr_spurr);
392 395
393 if (cpu_has_feature(CPU_FTR_DSCR)) 396 if (cpu_has_feature(CPU_FTR_DSCR))
394 sysdev_create_file(s, &attr_dscr); 397 device_create_file(s, &dev_attr_dscr);
398
399 if (cpu_has_feature(CPU_FTR_PPCAS_ARCH_V2))
400 device_create_file(s, &dev_attr_pir);
395#endif /* CONFIG_PPC64 */ 401#endif /* CONFIG_PPC64 */
396 402
397 cacheinfo_cpu_online(cpu); 403 cacheinfo_cpu_online(cpu);
@@ -401,8 +407,8 @@ static void __cpuinit register_cpu_online(unsigned int cpu)
401static void unregister_cpu_online(unsigned int cpu) 407static void unregister_cpu_online(unsigned int cpu)
402{ 408{
403 struct cpu *c = &per_cpu(cpu_devices, cpu); 409 struct cpu *c = &per_cpu(cpu_devices, cpu);
404 struct sys_device *s = &c->sysdev; 410 struct device *s = &c->dev;
405 struct sysdev_attribute *attrs, *pmc_attrs; 411 struct device_attribute *attrs, *pmc_attrs;
406 int i, nattrs; 412 int i, nattrs;
407 413
408 BUG_ON(!c->hotpluggable); 414 BUG_ON(!c->hotpluggable);
@@ -410,7 +416,7 @@ static void unregister_cpu_online(unsigned int cpu)
410#ifdef CONFIG_PPC64 416#ifdef CONFIG_PPC64
411 if (!firmware_has_feature(FW_FEATURE_ISERIES) && 417 if (!firmware_has_feature(FW_FEATURE_ISERIES) &&
412 cpu_has_feature(CPU_FTR_SMT)) 418 cpu_has_feature(CPU_FTR_SMT))
413 sysdev_remove_file(s, &attr_smt_snooze_delay); 419 device_remove_file(s, &dev_attr_smt_snooze_delay);
414#endif 420#endif
415 421
416 /* PMC stuff */ 422 /* PMC stuff */
@@ -418,14 +424,14 @@ static void unregister_cpu_online(unsigned int cpu)
418#ifdef HAS_PPC_PMC_IBM 424#ifdef HAS_PPC_PMC_IBM
419 case PPC_PMC_IBM: 425 case PPC_PMC_IBM:
420 attrs = ibm_common_attrs; 426 attrs = ibm_common_attrs;
421 nattrs = sizeof(ibm_common_attrs) / sizeof(struct sysdev_attribute); 427 nattrs = sizeof(ibm_common_attrs) / sizeof(struct device_attribute);
422 pmc_attrs = classic_pmc_attrs; 428 pmc_attrs = classic_pmc_attrs;
423 break; 429 break;
424#endif /* HAS_PPC_PMC_IBM */ 430#endif /* HAS_PPC_PMC_IBM */
425#ifdef HAS_PPC_PMC_G4 431#ifdef HAS_PPC_PMC_G4
426 case PPC_PMC_G4: 432 case PPC_PMC_G4:
427 attrs = g4_common_attrs; 433 attrs = g4_common_attrs;
428 nattrs = sizeof(g4_common_attrs) / sizeof(struct sysdev_attribute); 434 nattrs = sizeof(g4_common_attrs) / sizeof(struct device_attribute);
429 pmc_attrs = classic_pmc_attrs; 435 pmc_attrs = classic_pmc_attrs;
430 break; 436 break;
431#endif /* HAS_PPC_PMC_G4 */ 437#endif /* HAS_PPC_PMC_G4 */
@@ -433,7 +439,7 @@ static void unregister_cpu_online(unsigned int cpu)
433 case PPC_PMC_PA6T: 439 case PPC_PMC_PA6T:
434 /* PA Semi starts counting at PMC0 */ 440 /* PA Semi starts counting at PMC0 */
435 attrs = pa6t_attrs; 441 attrs = pa6t_attrs;
436 nattrs = sizeof(pa6t_attrs) / sizeof(struct sysdev_attribute); 442 nattrs = sizeof(pa6t_attrs) / sizeof(struct device_attribute);
437 pmc_attrs = NULL; 443 pmc_attrs = NULL;
438 break; 444 break;
439#endif /* HAS_PPC_PMC_PA6T */ 445#endif /* HAS_PPC_PMC_PA6T */
@@ -444,24 +450,27 @@ static void unregister_cpu_online(unsigned int cpu)
444 } 450 }
445 451
446 for (i = 0; i < nattrs; i++) 452 for (i = 0; i < nattrs; i++)
447 sysdev_remove_file(s, &attrs[i]); 453 device_remove_file(s, &attrs[i]);
448 454
449 if (pmc_attrs) 455 if (pmc_attrs)
450 for (i = 0; i < cur_cpu_spec->num_pmcs; i++) 456 for (i = 0; i < cur_cpu_spec->num_pmcs; i++)
451 sysdev_remove_file(s, &pmc_attrs[i]); 457 device_remove_file(s, &pmc_attrs[i]);
452 458
453#ifdef CONFIG_PPC64 459#ifdef CONFIG_PPC64
454 if (cpu_has_feature(CPU_FTR_MMCRA)) 460 if (cpu_has_feature(CPU_FTR_MMCRA))
455 sysdev_remove_file(s, &attr_mmcra); 461 device_remove_file(s, &dev_attr_mmcra);
456 462
457 if (cpu_has_feature(CPU_FTR_PURR)) 463 if (cpu_has_feature(CPU_FTR_PURR))
458 sysdev_remove_file(s, &attr_purr); 464 device_remove_file(s, &dev_attr_purr);
459 465
460 if (cpu_has_feature(CPU_FTR_SPURR)) 466 if (cpu_has_feature(CPU_FTR_SPURR))
461 sysdev_remove_file(s, &attr_spurr); 467 device_remove_file(s, &dev_attr_spurr);
462 468
463 if (cpu_has_feature(CPU_FTR_DSCR)) 469 if (cpu_has_feature(CPU_FTR_DSCR))
464 sysdev_remove_file(s, &attr_dscr); 470 device_remove_file(s, &dev_attr_dscr);
471
472 if (cpu_has_feature(CPU_FTR_PPCAS_ARCH_V2))
473 device_remove_file(s, &dev_attr_pir);
465#endif /* CONFIG_PPC64 */ 474#endif /* CONFIG_PPC64 */
466 475
467 cacheinfo_cpu_offline(cpu); 476 cacheinfo_cpu_offline(cpu);
@@ -513,70 +522,70 @@ static struct notifier_block __cpuinitdata sysfs_cpu_nb = {
513 522
514static DEFINE_MUTEX(cpu_mutex); 523static DEFINE_MUTEX(cpu_mutex);
515 524
516int cpu_add_sysdev_attr(struct sysdev_attribute *attr) 525int cpu_add_dev_attr(struct device_attribute *attr)
517{ 526{
518 int cpu; 527 int cpu;
519 528
520 mutex_lock(&cpu_mutex); 529 mutex_lock(&cpu_mutex);
521 530
522 for_each_possible_cpu(cpu) { 531 for_each_possible_cpu(cpu) {
523 sysdev_create_file(get_cpu_sysdev(cpu), attr); 532 device_create_file(get_cpu_device(cpu), attr);
524 } 533 }
525 534
526 mutex_unlock(&cpu_mutex); 535 mutex_unlock(&cpu_mutex);
527 return 0; 536 return 0;
528} 537}
529EXPORT_SYMBOL_GPL(cpu_add_sysdev_attr); 538EXPORT_SYMBOL_GPL(cpu_add_dev_attr);
530 539
531int cpu_add_sysdev_attr_group(struct attribute_group *attrs) 540int cpu_add_dev_attr_group(struct attribute_group *attrs)
532{ 541{
533 int cpu; 542 int cpu;
534 struct sys_device *sysdev; 543 struct device *dev;
535 int ret; 544 int ret;
536 545
537 mutex_lock(&cpu_mutex); 546 mutex_lock(&cpu_mutex);
538 547
539 for_each_possible_cpu(cpu) { 548 for_each_possible_cpu(cpu) {
540 sysdev = get_cpu_sysdev(cpu); 549 dev = get_cpu_device(cpu);
541 ret = sysfs_create_group(&sysdev->kobj, attrs); 550 ret = sysfs_create_group(&dev->kobj, attrs);
542 WARN_ON(ret != 0); 551 WARN_ON(ret != 0);
543 } 552 }
544 553
545 mutex_unlock(&cpu_mutex); 554 mutex_unlock(&cpu_mutex);
546 return 0; 555 return 0;
547} 556}
548EXPORT_SYMBOL_GPL(cpu_add_sysdev_attr_group); 557EXPORT_SYMBOL_GPL(cpu_add_dev_attr_group);
549 558
550 559
551void cpu_remove_sysdev_attr(struct sysdev_attribute *attr) 560void cpu_remove_dev_attr(struct device_attribute *attr)
552{ 561{
553 int cpu; 562 int cpu;
554 563
555 mutex_lock(&cpu_mutex); 564 mutex_lock(&cpu_mutex);
556 565
557 for_each_possible_cpu(cpu) { 566 for_each_possible_cpu(cpu) {
558 sysdev_remove_file(get_cpu_sysdev(cpu), attr); 567 device_remove_file(get_cpu_device(cpu), attr);
559 } 568 }
560 569
561 mutex_unlock(&cpu_mutex); 570 mutex_unlock(&cpu_mutex);
562} 571}
563EXPORT_SYMBOL_GPL(cpu_remove_sysdev_attr); 572EXPORT_SYMBOL_GPL(cpu_remove_dev_attr);
564 573
565void cpu_remove_sysdev_attr_group(struct attribute_group *attrs) 574void cpu_remove_dev_attr_group(struct attribute_group *attrs)
566{ 575{
567 int cpu; 576 int cpu;
568 struct sys_device *sysdev; 577 struct device *dev;
569 578
570 mutex_lock(&cpu_mutex); 579 mutex_lock(&cpu_mutex);
571 580
572 for_each_possible_cpu(cpu) { 581 for_each_possible_cpu(cpu) {
573 sysdev = get_cpu_sysdev(cpu); 582 dev = get_cpu_device(cpu);
574 sysfs_remove_group(&sysdev->kobj, attrs); 583 sysfs_remove_group(&dev->kobj, attrs);
575 } 584 }
576 585
577 mutex_unlock(&cpu_mutex); 586 mutex_unlock(&cpu_mutex);
578} 587}
579EXPORT_SYMBOL_GPL(cpu_remove_sysdev_attr_group); 588EXPORT_SYMBOL_GPL(cpu_remove_dev_attr_group);
580 589
581 590
582/* NUMA stuff */ 591/* NUMA stuff */
@@ -590,18 +599,18 @@ static void register_nodes(void)
590 register_one_node(i); 599 register_one_node(i);
591} 600}
592 601
593int sysfs_add_device_to_node(struct sys_device *dev, int nid) 602int sysfs_add_device_to_node(struct device *dev, int nid)
594{ 603{
595 struct node *node = &node_devices[nid]; 604 struct node *node = &node_devices[nid];
596 return sysfs_create_link(&node->sysdev.kobj, &dev->kobj, 605 return sysfs_create_link(&node->dev.kobj, &dev->kobj,
597 kobject_name(&dev->kobj)); 606 kobject_name(&dev->kobj));
598} 607}
599EXPORT_SYMBOL_GPL(sysfs_add_device_to_node); 608EXPORT_SYMBOL_GPL(sysfs_add_device_to_node);
600 609
601void sysfs_remove_device_from_node(struct sys_device *dev, int nid) 610void sysfs_remove_device_from_node(struct device *dev, int nid)
602{ 611{
603 struct node *node = &node_devices[nid]; 612 struct node *node = &node_devices[nid];
604 sysfs_remove_link(&node->sysdev.kobj, kobject_name(&dev->kobj)); 613 sysfs_remove_link(&node->dev.kobj, kobject_name(&dev->kobj));
605} 614}
606EXPORT_SYMBOL_GPL(sysfs_remove_device_from_node); 615EXPORT_SYMBOL_GPL(sysfs_remove_device_from_node);
607 616
@@ -614,14 +623,14 @@ static void register_nodes(void)
614#endif 623#endif
615 624
616/* Only valid if CPU is present. */ 625/* Only valid if CPU is present. */
617static ssize_t show_physical_id(struct sys_device *dev, 626static ssize_t show_physical_id(struct device *dev,
618 struct sysdev_attribute *attr, char *buf) 627 struct device_attribute *attr, char *buf)
619{ 628{
620 struct cpu *cpu = container_of(dev, struct cpu, sysdev); 629 struct cpu *cpu = container_of(dev, struct cpu, dev);
621 630
622 return sprintf(buf, "%d\n", get_hard_smp_processor_id(cpu->sysdev.id)); 631 return sprintf(buf, "%d\n", get_hard_smp_processor_id(cpu->dev.id));
623} 632}
624static SYSDEV_ATTR(physical_id, 0444, show_physical_id, NULL); 633static DEVICE_ATTR(physical_id, 0444, show_physical_id, NULL);
625 634
626static int __init topology_init(void) 635static int __init topology_init(void)
627{ 636{
@@ -646,7 +655,7 @@ static int __init topology_init(void)
646 if (cpu_online(cpu) || c->hotpluggable) { 655 if (cpu_online(cpu) || c->hotpluggable) {
647 register_cpu(c, cpu); 656 register_cpu(c, cpu);
648 657
649 sysdev_create_file(&c->sysdev, &attr_physical_id); 658 device_create_file(&c->dev, &dev_attr_physical_id);
650 } 659 }
651 660
652 if (cpu_online(cpu)) 661 if (cpu_online(cpu))
diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
index 522bb1dfc353..567dd7c3ac2a 100644
--- a/arch/powerpc/kernel/time.c
+++ b/arch/powerpc/kernel/time.c
@@ -86,8 +86,6 @@ static struct clocksource clocksource_rtc = {
86 .rating = 400, 86 .rating = 400,
87 .flags = CLOCK_SOURCE_IS_CONTINUOUS, 87 .flags = CLOCK_SOURCE_IS_CONTINUOUS,
88 .mask = CLOCKSOURCE_MASK(64), 88 .mask = CLOCKSOURCE_MASK(64),
89 .shift = 22,
90 .mult = 0, /* To be filled in */
91 .read = rtc_read, 89 .read = rtc_read,
92}; 90};
93 91
@@ -97,8 +95,6 @@ static struct clocksource clocksource_timebase = {
97 .rating = 400, 95 .rating = 400,
98 .flags = CLOCK_SOURCE_IS_CONTINUOUS, 96 .flags = CLOCK_SOURCE_IS_CONTINUOUS,
99 .mask = CLOCKSOURCE_MASK(64), 97 .mask = CLOCKSOURCE_MASK(64),
100 .shift = 22,
101 .mult = 0, /* To be filled in */
102 .read = timebase_read, 98 .read = timebase_read,
103}; 99};
104 100
@@ -110,22 +106,16 @@ static void decrementer_set_mode(enum clock_event_mode mode,
110 struct clock_event_device *dev); 106 struct clock_event_device *dev);
111 107
112static struct clock_event_device decrementer_clockevent = { 108static struct clock_event_device decrementer_clockevent = {
113 .name = "decrementer", 109 .name = "decrementer",
114 .rating = 200, 110 .rating = 200,
115 .shift = 0, /* To be filled in */ 111 .irq = 0,
116 .mult = 0, /* To be filled in */ 112 .set_next_event = decrementer_set_next_event,
117 .irq = 0, 113 .set_mode = decrementer_set_mode,
118 .set_next_event = decrementer_set_next_event, 114 .features = CLOCK_EVT_FEAT_ONESHOT,
119 .set_mode = decrementer_set_mode,
120 .features = CLOCK_EVT_FEAT_ONESHOT,
121}; 115};
122 116
123struct decrementer_clock { 117DEFINE_PER_CPU(u64, decrementers_next_tb);
124 struct clock_event_device event; 118static DEFINE_PER_CPU(struct clock_event_device, decrementers);
125 u64 next_tb;
126};
127
128static DEFINE_PER_CPU(struct decrementer_clock, decrementers);
129 119
130#ifdef CONFIG_PPC_ISERIES 120#ifdef CONFIG_PPC_ISERIES
131static unsigned long __initdata iSeries_recal_titan; 121static unsigned long __initdata iSeries_recal_titan;
@@ -168,13 +158,13 @@ EXPORT_SYMBOL_GPL(ppc_tb_freq);
168#ifdef CONFIG_VIRT_CPU_ACCOUNTING 158#ifdef CONFIG_VIRT_CPU_ACCOUNTING
169/* 159/*
170 * Factors for converting from cputime_t (timebase ticks) to 160 * Factors for converting from cputime_t (timebase ticks) to
171 * jiffies, milliseconds, seconds, and clock_t (1/USER_HZ seconds). 161 * jiffies, microseconds, seconds, and clock_t (1/USER_HZ seconds).
172 * These are all stored as 0.64 fixed-point binary fractions. 162 * These are all stored as 0.64 fixed-point binary fractions.
173 */ 163 */
174u64 __cputime_jiffies_factor; 164u64 __cputime_jiffies_factor;
175EXPORT_SYMBOL(__cputime_jiffies_factor); 165EXPORT_SYMBOL(__cputime_jiffies_factor);
176u64 __cputime_msec_factor; 166u64 __cputime_usec_factor;
177EXPORT_SYMBOL(__cputime_msec_factor); 167EXPORT_SYMBOL(__cputime_usec_factor);
178u64 __cputime_sec_factor; 168u64 __cputime_sec_factor;
179EXPORT_SYMBOL(__cputime_sec_factor); 169EXPORT_SYMBOL(__cputime_sec_factor);
180u64 __cputime_clockt_factor; 170u64 __cputime_clockt_factor;
@@ -192,8 +182,8 @@ static void calc_cputime_factors(void)
192 182
193 div128_by_32(HZ, 0, tb_ticks_per_sec, &res); 183 div128_by_32(HZ, 0, tb_ticks_per_sec, &res);
194 __cputime_jiffies_factor = res.result_low; 184 __cputime_jiffies_factor = res.result_low;
195 div128_by_32(1000, 0, tb_ticks_per_sec, &res); 185 div128_by_32(1000000, 0, tb_ticks_per_sec, &res);
196 __cputime_msec_factor = res.result_low; 186 __cputime_usec_factor = res.result_low;
197 div128_by_32(1, 0, tb_ticks_per_sec, &res); 187 div128_by_32(1, 0, tb_ticks_per_sec, &res);
198 __cputime_sec_factor = res.result_low; 188 __cputime_sec_factor = res.result_low;
199 div128_by_32(USER_HZ, 0, tb_ticks_per_sec, &res); 189 div128_by_32(USER_HZ, 0, tb_ticks_per_sec, &res);
@@ -441,7 +431,7 @@ EXPORT_SYMBOL(profile_pc);
441/* 431/*
442 * This function recalibrates the timebase based on the 49-bit time-of-day 432 * This function recalibrates the timebase based on the 49-bit time-of-day
443 * value in the Titan chip. The Titan is much more accurate than the value 433 * value in the Titan chip. The Titan is much more accurate than the value
444 * returned by the service processor for the timebase frequency. 434 * returned by the service processor for the timebase frequency.
445 */ 435 */
446 436
447static int __init iSeries_tb_recal(void) 437static int __init iSeries_tb_recal(void)
@@ -576,9 +566,8 @@ void arch_irq_work_raise(void)
576void timer_interrupt(struct pt_regs * regs) 566void timer_interrupt(struct pt_regs * regs)
577{ 567{
578 struct pt_regs *old_regs; 568 struct pt_regs *old_regs;
579 struct decrementer_clock *decrementer = &__get_cpu_var(decrementers); 569 u64 *next_tb = &__get_cpu_var(decrementers_next_tb);
580 struct clock_event_device *evt = &decrementer->event; 570 struct clock_event_device *evt = &__get_cpu_var(decrementers);
581 u64 now;
582 571
583 /* Ensure a positive value is written to the decrementer, or else 572 /* Ensure a positive value is written to the decrementer, or else
584 * some CPUs will continue to take decrementer exceptions. 573 * some CPUs will continue to take decrementer exceptions.
@@ -613,16 +602,9 @@ void timer_interrupt(struct pt_regs * regs)
613 get_lppaca()->int_dword.fields.decr_int = 0; 602 get_lppaca()->int_dword.fields.decr_int = 0;
614#endif 603#endif
615 604
616 now = get_tb_or_rtc(); 605 *next_tb = ~(u64)0;
617 if (now >= decrementer->next_tb) { 606 if (evt->event_handler)
618 decrementer->next_tb = ~(u64)0; 607 evt->event_handler(evt);
619 if (evt->event_handler)
620 evt->event_handler(evt);
621 } else {
622 now = decrementer->next_tb - now;
623 if (now <= DECREMENTER_MAX)
624 set_dec((int)now);
625 }
626 608
627#ifdef CONFIG_PPC_ISERIES 609#ifdef CONFIG_PPC_ISERIES
628 if (firmware_has_feature(FW_FEATURE_ISERIES) && hvlpevent_is_pending()) 610 if (firmware_has_feature(FW_FEATURE_ISERIES) && hvlpevent_is_pending())
@@ -650,9 +632,9 @@ static void generic_suspend_disable_irqs(void)
650 * with suspending. 632 * with suspending.
651 */ 633 */
652 634
653 set_dec(0x7fffffff); 635 set_dec(DECREMENTER_MAX);
654 local_irq_disable(); 636 local_irq_disable();
655 set_dec(0x7fffffff); 637 set_dec(DECREMENTER_MAX);
656} 638}
657 639
658static void generic_suspend_enable_irqs(void) 640static void generic_suspend_enable_irqs(void)
@@ -824,9 +806,8 @@ void update_vsyscall(struct timespec *wall_time, struct timespec *wtm,
824 ++vdso_data->tb_update_count; 806 ++vdso_data->tb_update_count;
825 smp_mb(); 807 smp_mb();
826 808
827 /* XXX this assumes clock->shift == 22 */ 809 /* 19342813113834067 ~= 2^(20+64) / 1e9 */
828 /* 4611686018 ~= 2^(20+64-22) / 1e9 */ 810 new_tb_to_xs = (u64) mult * (19342813113834067ULL >> clock->shift);
829 new_tb_to_xs = (u64) mult * 4611686018ULL;
830 new_stamp_xsec = (u64) wall_time->tv_nsec * XSEC_PER_SEC; 811 new_stamp_xsec = (u64) wall_time->tv_nsec * XSEC_PER_SEC;
831 do_div(new_stamp_xsec, 1000000000); 812 do_div(new_stamp_xsec, 1000000000);
832 new_stamp_xsec += (u64) wall_time->tv_sec * XSEC_PER_SEC; 813 new_stamp_xsec += (u64) wall_time->tv_sec * XSEC_PER_SEC;
@@ -877,9 +858,7 @@ static void __init clocksource_init(void)
877 else 858 else
878 clock = &clocksource_timebase; 859 clock = &clocksource_timebase;
879 860
880 clock->mult = clocksource_hz2mult(tb_ticks_per_sec, clock->shift); 861 if (clocksource_register_hz(clock, tb_ticks_per_sec)) {
881
882 if (clocksource_register(clock)) {
883 printk(KERN_ERR "clocksource: %s is already registered\n", 862 printk(KERN_ERR "clocksource: %s is already registered\n",
884 clock->name); 863 clock->name);
885 return; 864 return;
@@ -892,7 +871,7 @@ static void __init clocksource_init(void)
892static int decrementer_set_next_event(unsigned long evt, 871static int decrementer_set_next_event(unsigned long evt,
893 struct clock_event_device *dev) 872 struct clock_event_device *dev)
894{ 873{
895 __get_cpu_var(decrementers).next_tb = get_tb_or_rtc() + evt; 874 __get_cpu_var(decrementers_next_tb) = get_tb_or_rtc() + evt;
896 set_dec(evt); 875 set_dec(evt);
897 return 0; 876 return 0;
898} 877}
@@ -904,34 +883,9 @@ static void decrementer_set_mode(enum clock_event_mode mode,
904 decrementer_set_next_event(DECREMENTER_MAX, dev); 883 decrementer_set_next_event(DECREMENTER_MAX, dev);
905} 884}
906 885
907static inline uint64_t div_sc64(unsigned long ticks, unsigned long nsec,
908 int shift)
909{
910 uint64_t tmp = ((uint64_t)ticks) << shift;
911
912 do_div(tmp, nsec);
913 return tmp;
914}
915
916static void __init setup_clockevent_multiplier(unsigned long hz)
917{
918 u64 mult, shift = 32;
919
920 while (1) {
921 mult = div_sc64(hz, NSEC_PER_SEC, shift);
922 if (mult && (mult >> 32UL) == 0UL)
923 break;
924
925 shift--;
926 }
927
928 decrementer_clockevent.shift = shift;
929 decrementer_clockevent.mult = mult;
930}
931
932static void register_decrementer_clockevent(int cpu) 886static void register_decrementer_clockevent(int cpu)
933{ 887{
934 struct clock_event_device *dec = &per_cpu(decrementers, cpu).event; 888 struct clock_event_device *dec = &per_cpu(decrementers, cpu);
935 889
936 *dec = decrementer_clockevent; 890 *dec = decrementer_clockevent;
937 dec->cpumask = cpumask_of(cpu); 891 dec->cpumask = cpumask_of(cpu);
@@ -946,7 +900,8 @@ static void __init init_decrementer_clockevent(void)
946{ 900{
947 int cpu = smp_processor_id(); 901 int cpu = smp_processor_id();
948 902
949 setup_clockevent_multiplier(ppc_tb_freq); 903 clockevents_calc_mult_shift(&decrementer_clockevent, ppc_tb_freq, 4);
904
950 decrementer_clockevent.max_delta_ns = 905 decrementer_clockevent.max_delta_ns =
951 clockevent_delta2ns(DECREMENTER_MAX, &decrementer_clockevent); 906 clockevent_delta2ns(DECREMENTER_MAX, &decrementer_clockevent);
952 decrementer_clockevent.min_delta_ns = 907 decrementer_clockevent.min_delta_ns =
@@ -1014,10 +969,10 @@ void __init time_init(void)
1014 boot_tb = get_tb_or_rtc(); 969 boot_tb = get_tb_or_rtc();
1015 970
1016 /* If platform provided a timezone (pmac), we correct the time */ 971 /* If platform provided a timezone (pmac), we correct the time */
1017 if (timezone_offset) { 972 if (timezone_offset) {
1018 sys_tz.tz_minuteswest = -timezone_offset / 60; 973 sys_tz.tz_minuteswest = -timezone_offset / 60;
1019 sys_tz.tz_dsttime = 0; 974 sys_tz.tz_dsttime = 0;
1020 } 975 }
1021 976
1022 vdso_data->tb_update_count = 0; 977 vdso_data->tb_update_count = 0;
1023 vdso_data->tb_ticks_per_sec = tb_ticks_per_sec; 978 vdso_data->tb_ticks_per_sec = tb_ticks_per_sec;
diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
index 4e5908264d1a..c091527efd89 100644
--- a/arch/powerpc/kernel/traps.c
+++ b/arch/powerpc/kernel/traps.c
@@ -98,18 +98,14 @@ static void pmac_backlight_unblank(void)
98static inline void pmac_backlight_unblank(void) { } 98static inline void pmac_backlight_unblank(void) { }
99#endif 99#endif
100 100
101int die(const char *str, struct pt_regs *regs, long err) 101static arch_spinlock_t die_lock = __ARCH_SPIN_LOCK_UNLOCKED;
102static int die_owner = -1;
103static unsigned int die_nest_count;
104static int die_counter;
105
106static unsigned __kprobes long oops_begin(struct pt_regs *regs)
102{ 107{
103 static struct { 108 int cpu;
104 raw_spinlock_t lock;
105 u32 lock_owner;
106 int lock_owner_depth;
107 } die = {
108 .lock = __RAW_SPIN_LOCK_UNLOCKED(die.lock),
109 .lock_owner = -1,
110 .lock_owner_depth = 0
111 };
112 static int die_counter;
113 unsigned long flags; 109 unsigned long flags;
114 110
115 if (debugger(regs)) 111 if (debugger(regs))
@@ -117,66 +113,109 @@ int die(const char *str, struct pt_regs *regs, long err)
117 113
118 oops_enter(); 114 oops_enter();
119 115
120 if (die.lock_owner != raw_smp_processor_id()) { 116 /* racy, but better than risking deadlock. */
121 console_verbose(); 117 raw_local_irq_save(flags);
122 raw_spin_lock_irqsave(&die.lock, flags); 118 cpu = smp_processor_id();
123 die.lock_owner = smp_processor_id(); 119 if (!arch_spin_trylock(&die_lock)) {
124 die.lock_owner_depth = 0; 120 if (cpu == die_owner)
125 bust_spinlocks(1); 121 /* nested oops. should stop eventually */;
126 if (machine_is(powermac)) 122 else
127 pmac_backlight_unblank(); 123 arch_spin_lock(&die_lock);
128 } else {
129 local_save_flags(flags);
130 } 124 }
125 die_nest_count++;
126 die_owner = cpu;
127 console_verbose();
128 bust_spinlocks(1);
129 if (machine_is(powermac))
130 pmac_backlight_unblank();
131 return flags;
132}
131 133
132 if (++die.lock_owner_depth < 3) { 134static void __kprobes oops_end(unsigned long flags, struct pt_regs *regs,
133 printk("Oops: %s, sig: %ld [#%d]\n", str, err, ++die_counter); 135 int signr)
134#ifdef CONFIG_PREEMPT 136{
135 printk("PREEMPT "); 137 bust_spinlocks(0);
136#endif 138 die_owner = -1;
137#ifdef CONFIG_SMP 139 add_taint(TAINT_DIE);
138 printk("SMP NR_CPUS=%d ", NR_CPUS); 140 die_nest_count--;
139#endif 141 oops_exit();
140#ifdef CONFIG_DEBUG_PAGEALLOC 142 printk("\n");
141 printk("DEBUG_PAGEALLOC "); 143 if (!die_nest_count)
142#endif 144 /* Nest count reaches zero, release the lock. */
143#ifdef CONFIG_NUMA 145 arch_spin_unlock(&die_lock);
144 printk("NUMA "); 146 raw_local_irq_restore(flags);
145#endif
146 printk("%s\n", ppc_md.name ? ppc_md.name : "");
147 147
148 if (notify_die(DIE_OOPS, str, regs, err, 255, 148 /*
149 SIGSEGV) == NOTIFY_STOP) 149 * A system reset (0x100) is a request to dump, so we always send
150 return 1; 150 * it through the crashdump code.
151 */
152 if (kexec_should_crash(current) || (TRAP(regs) == 0x100)) {
153 crash_kexec(regs);
151 154
152 print_modules(); 155 /*
153 show_regs(regs); 156 * We aren't the primary crash CPU. We need to send it
154 } else { 157 * to a holding pattern to avoid it ending up in the panic
155 printk("Recursive die() failure, output suppressed\n"); 158 * code.
159 */
160 crash_kexec_secondary(regs);
156 } 161 }
157 162
158 bust_spinlocks(0); 163 if (!signr)
159 die.lock_owner = -1; 164 return;
160 add_taint(TAINT_DIE);
161 raw_spin_unlock_irqrestore(&die.lock, flags);
162 165
163 if (kexec_should_crash(current) || 166 /*
164 kexec_sr_activated(smp_processor_id())) 167 * While our oops output is serialised by a spinlock, output
165 crash_kexec(regs); 168 * from panic() called below can race and corrupt it. If we
166 crash_kexec_secondary(regs); 169 * know we are going to panic, delay for 1 second so we have a
170 * chance to get clean backtraces from all CPUs that are oopsing.
171 */
172 if (in_interrupt() || panic_on_oops || !current->pid ||
173 is_global_init(current)) {
174 mdelay(MSEC_PER_SEC);
175 }
167 176
168 if (in_interrupt()) 177 if (in_interrupt())
169 panic("Fatal exception in interrupt"); 178 panic("Fatal exception in interrupt");
170
171 if (panic_on_oops) 179 if (panic_on_oops)
172 panic("Fatal exception"); 180 panic("Fatal exception");
181 do_exit(signr);
182}
173 183
174 oops_exit(); 184static int __kprobes __die(const char *str, struct pt_regs *regs, long err)
175 do_exit(err); 185{
186 printk("Oops: %s, sig: %ld [#%d]\n", str, err, ++die_counter);
187#ifdef CONFIG_PREEMPT
188 printk("PREEMPT ");
189#endif
190#ifdef CONFIG_SMP
191 printk("SMP NR_CPUS=%d ", NR_CPUS);
192#endif
193#ifdef CONFIG_DEBUG_PAGEALLOC
194 printk("DEBUG_PAGEALLOC ");
195#endif
196#ifdef CONFIG_NUMA
197 printk("NUMA ");
198#endif
199 printk("%s\n", ppc_md.name ? ppc_md.name : "");
200
201 if (notify_die(DIE_OOPS, str, regs, err, 255, SIGSEGV) == NOTIFY_STOP)
202 return 1;
203
204 print_modules();
205 show_regs(regs);
176 206
177 return 0; 207 return 0;
178} 208}
179 209
210void die(const char *str, struct pt_regs *regs, long err)
211{
212 unsigned long flags = oops_begin(regs);
213
214 if (__die(str, regs, err))
215 err = 0;
216 oops_end(flags, regs, err);
217}
218
180void user_single_step_siginfo(struct task_struct *tsk, 219void user_single_step_siginfo(struct task_struct *tsk,
181 struct pt_regs *regs, siginfo_t *info) 220 struct pt_regs *regs, siginfo_t *info)
182{ 221{
@@ -195,10 +234,11 @@ void _exception(int signr, struct pt_regs *regs, int code, unsigned long addr)
195 "at %016lx nip %016lx lr %016lx code %x\n"; 234 "at %016lx nip %016lx lr %016lx code %x\n";
196 235
197 if (!user_mode(regs)) { 236 if (!user_mode(regs)) {
198 if (die("Exception in kernel mode", regs, signr)) 237 die("Exception in kernel mode", regs, signr);
199 return; 238 return;
200 } else if (show_unhandled_signals && 239 }
201 unhandled_signal(current, signr)) { 240
241 if (show_unhandled_signals && unhandled_signal(current, signr)) {
202 printk_ratelimited(regs->msr & MSR_64BIT ? fmt64 : fmt32, 242 printk_ratelimited(regs->msr & MSR_64BIT ? fmt64 : fmt32,
203 current->comm, current->pid, signr, 243 current->comm, current->pid, signr,
204 addr, regs->nip, regs->link, code); 244 addr, regs->nip, regs->link, code);
@@ -220,25 +260,8 @@ void system_reset_exception(struct pt_regs *regs)
220 return; 260 return;
221 } 261 }
222 262
223#ifdef CONFIG_KEXEC
224 cpumask_set_cpu(smp_processor_id(), &cpus_in_sr);
225#endif
226
227 die("System Reset", regs, SIGABRT); 263 die("System Reset", regs, SIGABRT);
228 264
229 /*
230 * Some CPUs when released from the debugger will execute this path.
231 * These CPUs entered the debugger via a soft-reset. If the CPU was
232 * hung before entering the debugger it will return to the hung
233 * state when exiting this function. This causes a problem in
234 * kdump since the hung CPU(s) will not respond to the IPI sent
235 * from kdump. To prevent the problem we call crash_kexec_secondary()
236 * here. If a kdump had not been initiated or we exit the debugger
237 * with the "exit and recover" command (x) crash_kexec_secondary()
238 * will return after 5ms and the CPU returns to its previous state.
239 */
240 crash_kexec_secondary(regs);
241
242 /* Must die if the interrupt is not recoverable */ 265 /* Must die if the interrupt is not recoverable */
243 if (!(regs->msr & MSR_RI)) 266 if (!(regs->msr & MSR_RI))
244 panic("Unrecoverable System Reset"); 267 panic("Unrecoverable System Reset");
@@ -1298,14 +1321,12 @@ void __kprobes DebugException(struct pt_regs *regs, unsigned long debug_status)
1298 1321
1299 if (user_mode(regs)) { 1322 if (user_mode(regs)) {
1300 current->thread.dbcr0 &= ~DBCR0_IC; 1323 current->thread.dbcr0 &= ~DBCR0_IC;
1301#ifdef CONFIG_PPC_ADV_DEBUG_REGS
1302 if (DBCR_ACTIVE_EVENTS(current->thread.dbcr0, 1324 if (DBCR_ACTIVE_EVENTS(current->thread.dbcr0,
1303 current->thread.dbcr1)) 1325 current->thread.dbcr1))
1304 regs->msr |= MSR_DE; 1326 regs->msr |= MSR_DE;
1305 else 1327 else
1306 /* Make sure the IDM bit is off */ 1328 /* Make sure the IDM bit is off */
1307 current->thread.dbcr0 &= ~DBCR0_IDM; 1329 current->thread.dbcr0 &= ~DBCR0_IDM;
1308#endif
1309 } 1330 }
1310 1331
1311 _exception(SIGTRAP, regs, TRAP_TRACE, regs->nip); 1332 _exception(SIGTRAP, regs, TRAP_TRACE, regs->nip);
diff --git a/arch/powerpc/kernel/vio.c b/arch/powerpc/kernel/vio.c
index f65af61996bd..8b086299ba25 100644
--- a/arch/powerpc/kernel/vio.c
+++ b/arch/powerpc/kernel/vio.c
@@ -1406,7 +1406,6 @@ static struct bus_type vio_bus_type = {
1406 .match = vio_bus_match, 1406 .match = vio_bus_match,
1407 .probe = vio_bus_probe, 1407 .probe = vio_bus_probe,
1408 .remove = vio_bus_remove, 1408 .remove = vio_bus_remove,
1409 .pm = GENERIC_SUBSYS_PM_OPS,
1410}; 1409};
1411 1410
1412/** 1411/**
diff --git a/arch/powerpc/kernel/vmlinux.lds.S b/arch/powerpc/kernel/vmlinux.lds.S
index 920276c0f6a1..710a54005dfb 100644
--- a/arch/powerpc/kernel/vmlinux.lds.S
+++ b/arch/powerpc/kernel/vmlinux.lds.S
@@ -170,7 +170,13 @@ SECTIONS
170 } 170 }
171#ifdef CONFIG_RELOCATABLE 171#ifdef CONFIG_RELOCATABLE
172 . = ALIGN(8); 172 . = ALIGN(8);
173 .dynsym : AT(ADDR(.dynsym) - LOAD_OFFSET) { *(.dynsym) } 173 .dynsym : AT(ADDR(.dynsym) - LOAD_OFFSET)
174 {
175#ifdef CONFIG_RELOCATABLE_PPC32
176 __dynamic_symtab = .;
177#endif
178 *(.dynsym)
179 }
174 .dynstr : AT(ADDR(.dynstr) - LOAD_OFFSET) { *(.dynstr) } 180 .dynstr : AT(ADDR(.dynstr) - LOAD_OFFSET) { *(.dynstr) }
175 .dynamic : AT(ADDR(.dynamic) - LOAD_OFFSET) 181 .dynamic : AT(ADDR(.dynamic) - LOAD_OFFSET)
176 { 182 {
generated by cgit v1.2.2 (git 2.25.0) at 2024-10-11 14:33:41 -0400