aboutsummaryrefslogtreecommitdiffstats
path: root/arch/ppc64/kernel
diff options
context:
space:
mode:
Diffstat (limited to 'arch/ppc64/kernel')
-rw-r--r--arch/ppc64/kernel/Makefile16
-rw-r--r--arch/ppc64/kernel/asm-offsets.c5
-rw-r--r--arch/ppc64/kernel/bitops.c147
-rw-r--r--arch/ppc64/kernel/bpa_iic.c284
-rw-r--r--arch/ppc64/kernel/bpa_iic.h62
-rw-r--r--arch/ppc64/kernel/bpa_iommu.c381
-rw-r--r--arch/ppc64/kernel/bpa_iommu.h65
-rw-r--r--arch/ppc64/kernel/bpa_nvram.c118
-rw-r--r--arch/ppc64/kernel/bpa_setup.c141
-rw-r--r--arch/ppc64/kernel/head.S320
-rw-r--r--arch/ppc64/kernel/irq.c4
-rw-r--r--arch/ppc64/kernel/kprobes.c138
-rw-r--r--arch/ppc64/kernel/lparcfg.c8
-rw-r--r--arch/ppc64/kernel/machine_kexec.c1
-rw-r--r--arch/ppc64/kernel/pacaData.c4
-rw-r--r--arch/ppc64/kernel/pci_dn.c5
-rw-r--r--arch/ppc64/kernel/proc_ppc64.c2
-rw-r--r--arch/ppc64/kernel/prom.c94
-rw-r--r--arch/ppc64/kernel/prom_init.c4
-rw-r--r--arch/ppc64/kernel/rtas-proc.c808
-rw-r--r--arch/ppc64/kernel/rtas_flash.c725
-rw-r--r--arch/ppc64/kernel/rtasd.c527
-rw-r--r--arch/ppc64/kernel/scanlog.c3
-rw-r--r--arch/ppc64/kernel/signal.c581
-rw-r--r--arch/ppc64/kernel/smp-tbsync.c179
-rw-r--r--arch/ppc64/kernel/smp.c568
-rw-r--r--arch/ppc64/kernel/spider-pic.c191
-rw-r--r--arch/ppc64/kernel/sysfs.c1
-rw-r--r--arch/ppc64/kernel/udbg_scc.c1
29 files changed, 351 insertions, 5032 deletions
diff --git a/arch/ppc64/kernel/Makefile b/arch/ppc64/kernel/Makefile
index 327c08ce4291..c441aebe7648 100644
--- a/arch/ppc64/kernel/Makefile
+++ b/arch/ppc64/kernel/Makefile
@@ -12,8 +12,7 @@ obj-y := misc.o prom.o
12endif 12endif
13 13
14obj-y += irq.o idle.o dma.o \ 14obj-y += irq.o idle.o dma.o \
15 signal.o \ 15 align.o pacaData.o \
16 align.o bitops.o pacaData.o \
17 udbg.o ioctl32.o \ 16 udbg.o ioctl32.o \
18 rtc.o \ 17 rtc.o \
19 cpu_setup_power4.o \ 18 cpu_setup_power4.o \
@@ -29,22 +28,16 @@ ifneq ($(CONFIG_PPC_MERGE),y)
29obj-$(CONFIG_PPC_MULTIPLATFORM) += prom_init.o 28obj-$(CONFIG_PPC_MULTIPLATFORM) += prom_init.o
30endif 29endif
31 30
32obj-$(CONFIG_PPC_PSERIES) += rtasd.o udbg_16550.o 31obj-$(CONFIG_PPC_PSERIES) += udbg_16550.o
33
34obj-$(CONFIG_PPC_BPA) += bpa_setup.o bpa_iommu.o bpa_nvram.o \
35 bpa_iic.o spider-pic.o
36 32
37obj-$(CONFIG_KEXEC) += machine_kexec.o 33obj-$(CONFIG_KEXEC) += machine_kexec.o
38obj-$(CONFIG_EEH) += eeh.o 34obj-$(CONFIG_EEH) += eeh.o
39obj-$(CONFIG_PROC_FS) += proc_ppc64.o 35obj-$(CONFIG_PROC_FS) += proc_ppc64.o
40obj-$(CONFIG_RTAS_FLASH) += rtas_flash.o
41obj-$(CONFIG_SMP) += smp.o
42obj-$(CONFIG_MODULES) += module.o 36obj-$(CONFIG_MODULES) += module.o
43ifneq ($(CONFIG_PPC_MERGE),y) 37ifneq ($(CONFIG_PPC_MERGE),y)
44obj-$(CONFIG_MODULES) += ppc_ksyms.o 38obj-$(CONFIG_MODULES) += ppc_ksyms.o
45endif 39endif
46obj-$(CONFIG_PPC_RTAS) += rtas_pci.o 40obj-$(CONFIG_PPC_RTAS) += rtas_pci.o
47obj-$(CONFIG_RTAS_PROC) += rtas-proc.o
48obj-$(CONFIG_SCANLOG) += scanlog.o 41obj-$(CONFIG_SCANLOG) += scanlog.o
49obj-$(CONFIG_LPARCFG) += lparcfg.o 42obj-$(CONFIG_LPARCFG) += lparcfg.o
50obj-$(CONFIG_HVC_CONSOLE) += hvconsole.o 43obj-$(CONFIG_HVC_CONSOLE) += hvconsole.o
@@ -57,11 +50,6 @@ obj-$(CONFIG_PPC_PMAC) += udbg_scc.o
57 50
58obj-$(CONFIG_PPC_MAPLE) += udbg_16550.o 51obj-$(CONFIG_PPC_MAPLE) += udbg_16550.o
59 52
60ifdef CONFIG_SMP
61obj-$(CONFIG_PPC_PMAC) += smp-tbsync.o
62obj-$(CONFIG_PPC_MAPLE) += smp-tbsync.o
63endif
64
65obj-$(CONFIG_KPROBES) += kprobes.o 53obj-$(CONFIG_KPROBES) += kprobes.o
66 54
67CFLAGS_ioctl32.o += -Ifs/ 55CFLAGS_ioctl32.o += -Ifs/
diff --git a/arch/ppc64/kernel/asm-offsets.c b/arch/ppc64/kernel/asm-offsets.c
index 5e6046cb414e..bce9065da6cb 100644
--- a/arch/ppc64/kernel/asm-offsets.c
+++ b/arch/ppc64/kernel/asm-offsets.c
@@ -31,7 +31,7 @@
31 31
32#include <asm/paca.h> 32#include <asm/paca.h>
33#include <asm/lppaca.h> 33#include <asm/lppaca.h>
34#include <asm/iSeries/HvLpEvent.h> 34#include <asm/iseries/hv_lp_event.h>
35#include <asm/rtas.h> 35#include <asm/rtas.h>
36#include <asm/cputable.h> 36#include <asm/cputable.h>
37#include <asm/cache.h> 37#include <asm/cache.h>
@@ -93,6 +93,9 @@ int main(void)
93 DEFINE(PACASLBCACHE, offsetof(struct paca_struct, slb_cache)); 93 DEFINE(PACASLBCACHE, offsetof(struct paca_struct, slb_cache));
94 DEFINE(PACASLBCACHEPTR, offsetof(struct paca_struct, slb_cache_ptr)); 94 DEFINE(PACASLBCACHEPTR, offsetof(struct paca_struct, slb_cache_ptr));
95 DEFINE(PACACONTEXTID, offsetof(struct paca_struct, context.id)); 95 DEFINE(PACACONTEXTID, offsetof(struct paca_struct, context.id));
96#ifdef CONFIG_PPC_64K_PAGES
97 DEFINE(PACAPGDIR, offsetof(struct paca_struct, pgdir));
98#endif
96#ifdef CONFIG_HUGETLB_PAGE 99#ifdef CONFIG_HUGETLB_PAGE
97 DEFINE(PACALOWHTLBAREAS, offsetof(struct paca_struct, context.low_htlb_areas)); 100 DEFINE(PACALOWHTLBAREAS, offsetof(struct paca_struct, context.low_htlb_areas));
98 DEFINE(PACAHIGHHTLBAREAS, offsetof(struct paca_struct, context.high_htlb_areas)); 101 DEFINE(PACAHIGHHTLBAREAS, offsetof(struct paca_struct, context.high_htlb_areas));
diff --git a/arch/ppc64/kernel/bitops.c b/arch/ppc64/kernel/bitops.c
deleted file mode 100644
index ae329e8b4acb..000000000000
--- a/arch/ppc64/kernel/bitops.c
+++ /dev/null
@@ -1,147 +0,0 @@
1/*
2 * These are too big to be inlined.
3 */
4
5#include <linux/kernel.h>
6#include <linux/module.h>
7#include <linux/bitops.h>
8#include <asm/byteorder.h>
9
10unsigned long find_next_zero_bit(const unsigned long *addr, unsigned long size,
11 unsigned long offset)
12{
13 const unsigned long *p = addr + (offset >> 6);
14 unsigned long result = offset & ~63UL;
15 unsigned long tmp;
16
17 if (offset >= size)
18 return size;
19 size -= result;
20 offset &= 63UL;
21 if (offset) {
22 tmp = *(p++);
23 tmp |= ~0UL >> (64 - offset);
24 if (size < 64)
25 goto found_first;
26 if (~tmp)
27 goto found_middle;
28 size -= 64;
29 result += 64;
30 }
31 while (size & ~63UL) {
32 if (~(tmp = *(p++)))
33 goto found_middle;
34 result += 64;
35 size -= 64;
36 }
37 if (!size)
38 return result;
39 tmp = *p;
40
41found_first:
42 tmp |= ~0UL << size;
43 if (tmp == ~0UL) /* Are any bits zero? */
44 return result + size; /* Nope. */
45found_middle:
46 return result + ffz(tmp);
47}
48
49EXPORT_SYMBOL(find_next_zero_bit);
50
51unsigned long find_next_bit(const unsigned long *addr, unsigned long size,
52 unsigned long offset)
53{
54 const unsigned long *p = addr + (offset >> 6);
55 unsigned long result = offset & ~63UL;
56 unsigned long tmp;
57
58 if (offset >= size)
59 return size;
60 size -= result;
61 offset &= 63UL;
62 if (offset) {
63 tmp = *(p++);
64 tmp &= (~0UL << offset);
65 if (size < 64)
66 goto found_first;
67 if (tmp)
68 goto found_middle;
69 size -= 64;
70 result += 64;
71 }
72 while (size & ~63UL) {
73 if ((tmp = *(p++)))
74 goto found_middle;
75 result += 64;
76 size -= 64;
77 }
78 if (!size)
79 return result;
80 tmp = *p;
81
82found_first:
83 tmp &= (~0UL >> (64 - size));
84 if (tmp == 0UL) /* Are any bits set? */
85 return result + size; /* Nope. */
86found_middle:
87 return result + __ffs(tmp);
88}
89
90EXPORT_SYMBOL(find_next_bit);
91
92static inline unsigned int ext2_ilog2(unsigned int x)
93{
94 int lz;
95
96 asm("cntlzw %0,%1": "=r"(lz):"r"(x));
97 return 31 - lz;
98}
99
100static inline unsigned int ext2_ffz(unsigned int x)
101{
102 u32 rc;
103 if ((x = ~x) == 0)
104 return 32;
105 rc = ext2_ilog2(x & -x);
106 return rc;
107}
108
109unsigned long find_next_zero_le_bit(const unsigned long *addr, unsigned long size,
110 unsigned long offset)
111{
112 const unsigned int *p = ((const unsigned int *)addr) + (offset >> 5);
113 unsigned int result = offset & ~31;
114 unsigned int tmp;
115
116 if (offset >= size)
117 return size;
118 size -= result;
119 offset &= 31;
120 if (offset) {
121 tmp = cpu_to_le32p(p++);
122 tmp |= ~0U >> (32 - offset); /* bug or feature ? */
123 if (size < 32)
124 goto found_first;
125 if (tmp != ~0)
126 goto found_middle;
127 size -= 32;
128 result += 32;
129 }
130 while (size >= 32) {
131 if ((tmp = cpu_to_le32p(p++)) != ~0)
132 goto found_middle;
133 result += 32;
134 size -= 32;
135 }
136 if (!size)
137 return result;
138 tmp = cpu_to_le32p(p);
139found_first:
140 tmp |= ~0 << size;
141 if (tmp == ~0) /* Are any bits zero? */
142 return result + size; /* Nope. */
143found_middle:
144 return result + ext2_ffz(tmp);
145}
146
147EXPORT_SYMBOL(find_next_zero_le_bit);
diff --git a/arch/ppc64/kernel/bpa_iic.c b/arch/ppc64/kernel/bpa_iic.c
deleted file mode 100644
index 0aaa878e19d3..000000000000
--- a/arch/ppc64/kernel/bpa_iic.c
+++ /dev/null
@@ -1,284 +0,0 @@
1/*
2 * BPA Internal Interrupt Controller
3 *
4 * (C) Copyright IBM Deutschland Entwicklung GmbH 2005
5 *
6 * Author: Arnd Bergmann <arndb@de.ibm.com>
7 *
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2, or (at your option)
11 * any later version.
12 *
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to the Free Software
20 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
21 */
22
23#include <linux/config.h>
24#include <linux/interrupt.h>
25#include <linux/irq.h>
26#include <linux/percpu.h>
27#include <linux/types.h>
28
29#include <asm/io.h>
30#include <asm/pgtable.h>
31#include <asm/prom.h>
32#include <asm/ptrace.h>
33
34#include "bpa_iic.h"
35
36struct iic_pending_bits {
37 u32 data;
38 u8 flags;
39 u8 class;
40 u8 source;
41 u8 prio;
42};
43
44enum iic_pending_flags {
45 IIC_VALID = 0x80,
46 IIC_IPI = 0x40,
47};
48
49struct iic_regs {
50 struct iic_pending_bits pending;
51 struct iic_pending_bits pending_destr;
52 u64 generate;
53 u64 prio;
54};
55
56struct iic {
57 struct iic_regs __iomem *regs;
58};
59
60static DEFINE_PER_CPU(struct iic, iic);
61
62void iic_local_enable(void)
63{
64 out_be64(&__get_cpu_var(iic).regs->prio, 0xff);
65}
66
67void iic_local_disable(void)
68{
69 out_be64(&__get_cpu_var(iic).regs->prio, 0x0);
70}
71
72static unsigned int iic_startup(unsigned int irq)
73{
74 return 0;
75}
76
77static void iic_enable(unsigned int irq)
78{
79 iic_local_enable();
80}
81
82static void iic_disable(unsigned int irq)
83{
84}
85
86static void iic_end(unsigned int irq)
87{
88 iic_local_enable();
89}
90
91static struct hw_interrupt_type iic_pic = {
92 .typename = " BPA-IIC ",
93 .startup = iic_startup,
94 .enable = iic_enable,
95 .disable = iic_disable,
96 .end = iic_end,
97};
98
99static int iic_external_get_irq(struct iic_pending_bits pending)
100{
101 int irq;
102 unsigned char node, unit;
103
104 node = pending.source >> 4;
105 unit = pending.source & 0xf;
106 irq = -1;
107
108 /*
109 * This mapping is specific to the Broadband
110 * Engine. We might need to get the numbers
111 * from the device tree to support future CPUs.
112 */
113 switch (unit) {
114 case 0x00:
115 case 0x0b:
116 /*
117 * One of these units can be connected
118 * to an external interrupt controller.
119 */
120 if (pending.prio > 0x3f ||
121 pending.class != 2)
122 break;
123 irq = IIC_EXT_OFFSET
124 + spider_get_irq(pending.prio + node * IIC_NODE_STRIDE)
125 + node * IIC_NODE_STRIDE;
126 break;
127 case 0x01 ... 0x04:
128 case 0x07 ... 0x0a:
129 /*
130 * These units are connected to the SPEs
131 */
132 if (pending.class > 2)
133 break;
134 irq = IIC_SPE_OFFSET
135 + pending.class * IIC_CLASS_STRIDE
136 + node * IIC_NODE_STRIDE
137 + unit;
138 break;
139 }
140 if (irq == -1)
141 printk(KERN_WARNING "Unexpected interrupt class %02x, "
142 "source %02x, prio %02x, cpu %02x\n", pending.class,
143 pending.source, pending.prio, smp_processor_id());
144 return irq;
145}
146
147/* Get an IRQ number from the pending state register of the IIC */
148int iic_get_irq(struct pt_regs *regs)
149{
150 struct iic *iic;
151 int irq;
152 struct iic_pending_bits pending;
153
154 iic = &__get_cpu_var(iic);
155 *(unsigned long *) &pending =
156 in_be64((unsigned long __iomem *) &iic->regs->pending_destr);
157
158 irq = -1;
159 if (pending.flags & IIC_VALID) {
160 if (pending.flags & IIC_IPI) {
161 irq = IIC_IPI_OFFSET + (pending.prio >> 4);
162/*
163 if (irq > 0x80)
164 printk(KERN_WARNING "Unexpected IPI prio %02x"
165 "on CPU %02x\n", pending.prio,
166 smp_processor_id());
167*/
168 } else {
169 irq = iic_external_get_irq(pending);
170 }
171 }
172 return irq;
173}
174
175static struct iic_regs __iomem *find_iic(int cpu)
176{
177 struct device_node *np;
178 int nodeid = cpu / 2;
179 unsigned long regs;
180 struct iic_regs __iomem *iic_regs;
181
182 for (np = of_find_node_by_type(NULL, "cpu");
183 np;
184 np = of_find_node_by_type(np, "cpu")) {
185 if (nodeid == *(int *)get_property(np, "node-id", NULL))
186 break;
187 }
188
189 if (!np) {
190 printk(KERN_WARNING "IIC: CPU %d not found\n", cpu);
191 iic_regs = NULL;
192 } else {
193 regs = *(long *)get_property(np, "iic", NULL);
194
195 /* hack until we have decided on the devtree info */
196 regs += 0x400;
197 if (cpu & 1)
198 regs += 0x20;
199
200 printk(KERN_DEBUG "IIC for CPU %d at %lx\n", cpu, regs);
201 iic_regs = __ioremap(regs, sizeof(struct iic_regs),
202 _PAGE_NO_CACHE);
203 }
204 return iic_regs;
205}
206
207#ifdef CONFIG_SMP
208
209/* Use the highest interrupt priorities for IPI */
210static inline int iic_ipi_to_irq(int ipi)
211{
212 return IIC_IPI_OFFSET + IIC_NUM_IPIS - 1 - ipi;
213}
214
215static inline int iic_irq_to_ipi(int irq)
216{
217 return IIC_NUM_IPIS - 1 - (irq - IIC_IPI_OFFSET);
218}
219
220void iic_setup_cpu(void)
221{
222 out_be64(&__get_cpu_var(iic).regs->prio, 0xff);
223}
224
225void iic_cause_IPI(int cpu, int mesg)
226{
227 out_be64(&per_cpu(iic, cpu).regs->generate, (IIC_NUM_IPIS - 1 - mesg) << 4);
228}
229
230static irqreturn_t iic_ipi_action(int irq, void *dev_id, struct pt_regs *regs)
231{
232 smp_message_recv(iic_irq_to_ipi(irq), regs);
233 return IRQ_HANDLED;
234}
235
236static void iic_request_ipi(int ipi, const char *name)
237{
238 int irq;
239
240 irq = iic_ipi_to_irq(ipi);
241 /* IPIs are marked SA_INTERRUPT as they must run with irqs
242 * disabled */
243 get_irq_desc(irq)->handler = &iic_pic;
244 get_irq_desc(irq)->status |= IRQ_PER_CPU;
245 request_irq(irq, iic_ipi_action, SA_INTERRUPT, name, NULL);
246}
247
248void iic_request_IPIs(void)
249{
250 iic_request_ipi(PPC_MSG_CALL_FUNCTION, "IPI-call");
251 iic_request_ipi(PPC_MSG_RESCHEDULE, "IPI-resched");
252#ifdef CONFIG_DEBUGGER
253 iic_request_ipi(PPC_MSG_DEBUGGER_BREAK, "IPI-debug");
254#endif /* CONFIG_DEBUGGER */
255}
256#endif /* CONFIG_SMP */
257
258static void iic_setup_spe_handlers(void)
259{
260 int be, isrc;
261
262 /* Assume two threads per BE are present */
263 for (be=0; be < num_present_cpus() / 2; be++) {
264 for (isrc = 0; isrc < IIC_CLASS_STRIDE * 3; isrc++) {
265 int irq = IIC_NODE_STRIDE * be + IIC_SPE_OFFSET + isrc;
266 get_irq_desc(irq)->handler = &iic_pic;
267 }
268 }
269}
270
271void iic_init_IRQ(void)
272{
273 int cpu, irq_offset;
274 struct iic *iic;
275
276 irq_offset = 0;
277 for_each_cpu(cpu) {
278 iic = &per_cpu(iic, cpu);
279 iic->regs = find_iic(cpu);
280 if (iic->regs)
281 out_be64(&iic->regs->prio, 0xff);
282 }
283 iic_setup_spe_handlers();
284}
diff --git a/arch/ppc64/kernel/bpa_iic.h b/arch/ppc64/kernel/bpa_iic.h
deleted file mode 100644
index 6833c3022166..000000000000
--- a/arch/ppc64/kernel/bpa_iic.h
+++ /dev/null
@@ -1,62 +0,0 @@
1#ifndef ASM_BPA_IIC_H
2#define ASM_BPA_IIC_H
3#ifdef __KERNEL__
4/*
5 * Mapping of IIC pending bits into per-node
6 * interrupt numbers.
7 *
8 * IRQ FF CC SS PP FF CC SS PP Description
9 *
10 * 00-3f 80 02 +0 00 - 80 02 +0 3f South Bridge
11 * 00-3f 80 02 +b 00 - 80 02 +b 3f South Bridge
12 * 41-4a 80 00 +1 ** - 80 00 +a ** SPU Class 0
13 * 51-5a 80 01 +1 ** - 80 01 +a ** SPU Class 1
14 * 61-6a 80 02 +1 ** - 80 02 +a ** SPU Class 2
15 * 70-7f C0 ** ** 00 - C0 ** ** 0f IPI
16 *
17 * F flags
18 * C class
19 * S source
20 * P Priority
21 * + node number
22 * * don't care
23 *
24 * A node consists of a Broadband Engine and an optional
25 * south bridge device providing a maximum of 64 IRQs.
26 * The south bridge may be connected to either IOIF0
27 * or IOIF1.
28 * Each SPE is represented as three IRQ lines, one per
29 * interrupt class.
30 * 16 IRQ numbers are reserved for inter processor
31 * interruptions, although these are only used in the
32 * range of the first node.
33 *
34 * This scheme needs 128 IRQ numbers per BIF node ID,
35 * which means that with the total of 512 lines
36 * available, we can have a maximum of four nodes.
37 */
38
39enum {
40 IIC_EXT_OFFSET = 0x00, /* Start of south bridge IRQs */
41 IIC_NUM_EXT = 0x40, /* Number of south bridge IRQs */
42 IIC_SPE_OFFSET = 0x40, /* Start of SPE interrupts */
43 IIC_CLASS_STRIDE = 0x10, /* SPE IRQs per class */
44 IIC_IPI_OFFSET = 0x70, /* Start of IPI IRQs */
45 IIC_NUM_IPIS = 0x10, /* IRQs reserved for IPI */
46 IIC_NODE_STRIDE = 0x80, /* Total IRQs per node */
47};
48
49extern void iic_init_IRQ(void);
50extern int iic_get_irq(struct pt_regs *regs);
51extern void iic_cause_IPI(int cpu, int mesg);
52extern void iic_request_IPIs(void);
53extern void iic_setup_cpu(void);
54extern void iic_local_enable(void);
55extern void iic_local_disable(void);
56
57
58extern void spider_init_IRQ(void);
59extern int spider_get_irq(unsigned long int_pending);
60
61#endif
62#endif /* ASM_BPA_IIC_H */
diff --git a/arch/ppc64/kernel/bpa_iommu.c b/arch/ppc64/kernel/bpa_iommu.c
deleted file mode 100644
index da1b4b7a3269..000000000000
--- a/arch/ppc64/kernel/bpa_iommu.c
+++ /dev/null
@@ -1,381 +0,0 @@
1/*
2 * IOMMU implementation for Broadband Processor Architecture
3 * We just establish a linear mapping at boot by setting all the
4 * IOPT cache entries in the CPU.
5 * The mapping functions should be identical to pci_direct_iommu,
6 * except for the handling of the high order bit that is required
7 * by the Spider bridge. These should be split into a separate
8 * file at the point where we get a different bridge chip.
9 *
10 * Copyright (C) 2005 IBM Deutschland Entwicklung GmbH,
11 * Arnd Bergmann <arndb@de.ibm.com>
12 *
13 * Based on linear mapping
14 * Copyright (C) 2003 Benjamin Herrenschmidt (benh@kernel.crashing.org)
15 *
16 * This program is free software; you can redistribute it and/or
17 * modify it under the terms of the GNU General Public License
18 * as published by the Free Software Foundation; either version
19 * 2 of the License, or (at your option) any later version.
20 */
21
22#undef DEBUG
23
24#include <linux/kernel.h>
25#include <linux/pci.h>
26#include <linux/delay.h>
27#include <linux/string.h>
28#include <linux/init.h>
29#include <linux/bootmem.h>
30#include <linux/mm.h>
31#include <linux/dma-mapping.h>
32
33#include <asm/sections.h>
34#include <asm/iommu.h>
35#include <asm/io.h>
36#include <asm/prom.h>
37#include <asm/pci-bridge.h>
38#include <asm/machdep.h>
39#include <asm/pmac_feature.h>
40#include <asm/abs_addr.h>
41#include <asm/system.h>
42#include <asm/ppc-pci.h>
43
44#include "bpa_iommu.h"
45
46static inline unsigned long
47get_iopt_entry(unsigned long real_address, unsigned long ioid,
48 unsigned long prot)
49{
50 return (prot & IOPT_PROT_MASK)
51 | (IOPT_COHERENT)
52 | (IOPT_ORDER_VC)
53 | (real_address & IOPT_RPN_MASK)
54 | (ioid & IOPT_IOID_MASK);
55}
56
57typedef struct {
58 unsigned long val;
59} ioste;
60
61static inline ioste
62mk_ioste(unsigned long val)
63{
64 ioste ioste = { .val = val, };
65 return ioste;
66}
67
68static inline ioste
69get_iost_entry(unsigned long iopt_base, unsigned long io_address, unsigned page_size)
70{
71 unsigned long ps;
72 unsigned long iostep;
73 unsigned long nnpt;
74 unsigned long shift;
75
76 switch (page_size) {
77 case 0x1000000:
78 ps = IOST_PS_16M;
79 nnpt = 0; /* one page per segment */
80 shift = 5; /* segment has 16 iopt entries */
81 break;
82
83 case 0x100000:
84 ps = IOST_PS_1M;
85 nnpt = 0; /* one page per segment */
86 shift = 1; /* segment has 256 iopt entries */
87 break;
88
89 case 0x10000:
90 ps = IOST_PS_64K;
91 nnpt = 0x07; /* 8 pages per io page table */
92 shift = 0; /* all entries are used */
93 break;
94
95 case 0x1000:
96 ps = IOST_PS_4K;
97 nnpt = 0x7f; /* 128 pages per io page table */
98 shift = 0; /* all entries are used */
99 break;
100
101 default: /* not a known compile time constant */
102 {
103 /* BUILD_BUG_ON() is not usable here */
104 extern void __get_iost_entry_bad_page_size(void);
105 __get_iost_entry_bad_page_size();
106 }
107 break;
108 }
109
110 iostep = iopt_base +
111 /* need 8 bytes per iopte */
112 (((io_address / page_size * 8)
113 /* align io page tables on 4k page boundaries */
114 << shift)
115 /* nnpt+1 pages go into each iopt */
116 & ~(nnpt << 12));
117
118 nnpt++; /* this seems to work, but the documentation is not clear
119 about wether we put nnpt or nnpt-1 into the ioste bits.
120 In theory, this can't work for 4k pages. */
121 return mk_ioste(IOST_VALID_MASK
122 | (iostep & IOST_PT_BASE_MASK)
123 | ((nnpt << 5) & IOST_NNPT_MASK)
124 | (ps & IOST_PS_MASK));
125}
126
127/* compute the address of an io pte */
128static inline unsigned long
129get_ioptep(ioste iost_entry, unsigned long io_address)
130{
131 unsigned long iopt_base;
132 unsigned long page_size;
133 unsigned long page_number;
134 unsigned long iopt_offset;
135
136 iopt_base = iost_entry.val & IOST_PT_BASE_MASK;
137 page_size = iost_entry.val & IOST_PS_MASK;
138
139 /* decode page size to compute page number */
140 page_number = (io_address & 0x0fffffff) >> (10 + 2 * page_size);
141 /* page number is an offset into the io page table */
142 iopt_offset = (page_number << 3) & 0x7fff8ul;
143 return iopt_base + iopt_offset;
144}
145
146/* compute the tag field of the iopt cache entry */
147static inline unsigned long
148get_ioc_tag(ioste iost_entry, unsigned long io_address)
149{
150 unsigned long iopte = get_ioptep(iost_entry, io_address);
151
152 return IOPT_VALID_MASK
153 | ((iopte & 0x00000000000000ff8ul) >> 3)
154 | ((iopte & 0x0000003fffffc0000ul) >> 9);
155}
156
157/* compute the hashed 6 bit index for the 4-way associative pte cache */
158static inline unsigned long
159get_ioc_hash(ioste iost_entry, unsigned long io_address)
160{
161 unsigned long iopte = get_ioptep(iost_entry, io_address);
162
163 return ((iopte & 0x000000000000001f8ul) >> 3)
164 ^ ((iopte & 0x00000000000020000ul) >> 17)
165 ^ ((iopte & 0x00000000000010000ul) >> 15)
166 ^ ((iopte & 0x00000000000008000ul) >> 13)
167 ^ ((iopte & 0x00000000000004000ul) >> 11)
168 ^ ((iopte & 0x00000000000002000ul) >> 9)
169 ^ ((iopte & 0x00000000000001000ul) >> 7);
170}
171
172/* same as above, but pretend that we have a simpler 1-way associative
173 pte cache with an 8 bit index */
174static inline unsigned long
175get_ioc_hash_1way(ioste iost_entry, unsigned long io_address)
176{
177 unsigned long iopte = get_ioptep(iost_entry, io_address);
178
179 return ((iopte & 0x000000000000001f8ul) >> 3)
180 ^ ((iopte & 0x00000000000020000ul) >> 17)
181 ^ ((iopte & 0x00000000000010000ul) >> 15)
182 ^ ((iopte & 0x00000000000008000ul) >> 13)
183 ^ ((iopte & 0x00000000000004000ul) >> 11)
184 ^ ((iopte & 0x00000000000002000ul) >> 9)
185 ^ ((iopte & 0x00000000000001000ul) >> 7)
186 ^ ((iopte & 0x0000000000000c000ul) >> 8);
187}
188
189static inline ioste
190get_iost_cache(void __iomem *base, unsigned long index)
191{
192 unsigned long __iomem *p = (base + IOC_ST_CACHE_DIR);
193 return mk_ioste(in_be64(&p[index]));
194}
195
196static inline void
197set_iost_cache(void __iomem *base, unsigned long index, ioste ste)
198{
199 unsigned long __iomem *p = (base + IOC_ST_CACHE_DIR);
200 pr_debug("ioste %02lx was %016lx, store %016lx", index,
201 get_iost_cache(base, index).val, ste.val);
202 out_be64(&p[index], ste.val);
203 pr_debug(" now %016lx\n", get_iost_cache(base, index).val);
204}
205
206static inline unsigned long
207get_iopt_cache(void __iomem *base, unsigned long index, unsigned long *tag)
208{
209 unsigned long __iomem *tags = (void *)(base + IOC_PT_CACHE_DIR);
210 unsigned long __iomem *p = (void *)(base + IOC_PT_CACHE_REG);
211
212 *tag = tags[index];
213 rmb();
214 return *p;
215}
216
217static inline void
218set_iopt_cache(void __iomem *base, unsigned long index,
219 unsigned long tag, unsigned long val)
220{
221 unsigned long __iomem *tags = base + IOC_PT_CACHE_DIR;
222 unsigned long __iomem *p = base + IOC_PT_CACHE_REG;
223 pr_debug("iopt %02lx was v%016lx/t%016lx, store v%016lx/t%016lx\n",
224 index, get_iopt_cache(base, index, &oldtag), oldtag, val, tag);
225
226 out_be64(p, val);
227 out_be64(&tags[index], tag);
228}
229
230static inline void
231set_iost_origin(void __iomem *base)
232{
233 unsigned long __iomem *p = base + IOC_ST_ORIGIN;
234 unsigned long origin = IOSTO_ENABLE | IOSTO_SW;
235
236 pr_debug("iost_origin %016lx, now %016lx\n", in_be64(p), origin);
237 out_be64(p, origin);
238}
239
240static inline void
241set_iocmd_config(void __iomem *base)
242{
243 unsigned long __iomem *p = base + 0xc00;
244 unsigned long conf;
245
246 conf = in_be64(p);
247 pr_debug("iost_conf %016lx, now %016lx\n", conf, conf | IOCMD_CONF_TE);
248 out_be64(p, conf | IOCMD_CONF_TE);
249}
250
251/* FIXME: get these from the device tree */
252#define ioc_base 0x20000511000ull
253#define ioc_mmio_base 0x20000510000ull
254#define ioid 0x48a
255#define iopt_phys_offset (- 0x20000000) /* We have a 512MB offset from the SB */
256#define io_page_size 0x1000000
257
258static unsigned long map_iopt_entry(unsigned long address)
259{
260 switch (address >> 20) {
261 case 0x600:
262 address = 0x24020000000ull; /* spider i/o */
263 break;
264 default:
265 address += iopt_phys_offset;
266 break;
267 }
268
269 return get_iopt_entry(address, ioid, IOPT_PROT_RW);
270}
271
272static void iommu_bus_setup_null(struct pci_bus *b) { }
273static void iommu_dev_setup_null(struct pci_dev *d) { }
274
275/* initialize the iommu to support a simple linear mapping
276 * for each DMA window used by any device. For now, we
277 * happen to know that there is only one DMA window in use,
278 * starting at iopt_phys_offset. */
279static void bpa_map_iommu(void)
280{
281 unsigned long address;
282 void __iomem *base;
283 ioste ioste;
284 unsigned long index;
285
286 base = __ioremap(ioc_base, 0x1000, _PAGE_NO_CACHE);
287 pr_debug("%lx mapped to %p\n", ioc_base, base);
288 set_iocmd_config(base);
289 iounmap(base);
290
291 base = __ioremap(ioc_mmio_base, 0x1000, _PAGE_NO_CACHE);
292 pr_debug("%lx mapped to %p\n", ioc_mmio_base, base);
293
294 set_iost_origin(base);
295
296 for (address = 0; address < 0x100000000ul; address += io_page_size) {
297 ioste = get_iost_entry(0x10000000000ul, address, io_page_size);
298 if ((address & 0xfffffff) == 0) /* segment start */
299 set_iost_cache(base, address >> 28, ioste);
300 index = get_ioc_hash_1way(ioste, address);
301 pr_debug("addr %08lx, index %02lx, ioste %016lx\n",
302 address, index, ioste.val);
303 set_iopt_cache(base,
304 get_ioc_hash_1way(ioste, address),
305 get_ioc_tag(ioste, address),
306 map_iopt_entry(address));
307 }
308 iounmap(base);
309}
310
311
312static void *bpa_alloc_coherent(struct device *hwdev, size_t size,
313 dma_addr_t *dma_handle, gfp_t flag)
314{
315 void *ret;
316
317 ret = (void *)__get_free_pages(flag, get_order(size));
318 if (ret != NULL) {
319 memset(ret, 0, size);
320 *dma_handle = virt_to_abs(ret) | BPA_DMA_VALID;
321 }
322 return ret;
323}
324
325static void bpa_free_coherent(struct device *hwdev, size_t size,
326 void *vaddr, dma_addr_t dma_handle)
327{
328 free_pages((unsigned long)vaddr, get_order(size));
329}
330
331static dma_addr_t bpa_map_single(struct device *hwdev, void *ptr,
332 size_t size, enum dma_data_direction direction)
333{
334 return virt_to_abs(ptr) | BPA_DMA_VALID;
335}
336
337static void bpa_unmap_single(struct device *hwdev, dma_addr_t dma_addr,
338 size_t size, enum dma_data_direction direction)
339{
340}
341
342static int bpa_map_sg(struct device *hwdev, struct scatterlist *sg,
343 int nents, enum dma_data_direction direction)
344{
345 int i;
346
347 for (i = 0; i < nents; i++, sg++) {
348 sg->dma_address = (page_to_phys(sg->page) + sg->offset)
349 | BPA_DMA_VALID;
350 sg->dma_length = sg->length;
351 }
352
353 return nents;
354}
355
356static void bpa_unmap_sg(struct device *hwdev, struct scatterlist *sg,
357 int nents, enum dma_data_direction direction)
358{
359}
360
361static int bpa_dma_supported(struct device *dev, u64 mask)
362{
363 return mask < 0x100000000ull;
364}
365
366void bpa_init_iommu(void)
367{
368 bpa_map_iommu();
369
370 /* Direct I/O, IOMMU off */
371 ppc_md.iommu_dev_setup = iommu_dev_setup_null;
372 ppc_md.iommu_bus_setup = iommu_bus_setup_null;
373
374 pci_dma_ops.alloc_coherent = bpa_alloc_coherent;
375 pci_dma_ops.free_coherent = bpa_free_coherent;
376 pci_dma_ops.map_single = bpa_map_single;
377 pci_dma_ops.unmap_single = bpa_unmap_single;
378 pci_dma_ops.map_sg = bpa_map_sg;
379 pci_dma_ops.unmap_sg = bpa_unmap_sg;
380 pci_dma_ops.dma_supported = bpa_dma_supported;
381}
diff --git a/arch/ppc64/kernel/bpa_iommu.h b/arch/ppc64/kernel/bpa_iommu.h
deleted file mode 100644
index e547d77dfa04..000000000000
--- a/arch/ppc64/kernel/bpa_iommu.h
+++ /dev/null
@@ -1,65 +0,0 @@
1#ifndef BPA_IOMMU_H
2#define BPA_IOMMU_H
3
4/* some constants */
5enum {
6 /* segment table entries */
7 IOST_VALID_MASK = 0x8000000000000000ul,
8 IOST_TAG_MASK = 0x3000000000000000ul,
9 IOST_PT_BASE_MASK = 0x000003fffffff000ul,
10 IOST_NNPT_MASK = 0x0000000000000fe0ul,
11 IOST_PS_MASK = 0x000000000000000ful,
12
13 IOST_PS_4K = 0x1,
14 IOST_PS_64K = 0x3,
15 IOST_PS_1M = 0x5,
16 IOST_PS_16M = 0x7,
17
18 /* iopt tag register */
19 IOPT_VALID_MASK = 0x0000000200000000ul,
20 IOPT_TAG_MASK = 0x00000001fffffffful,
21
22 /* iopt cache register */
23 IOPT_PROT_MASK = 0xc000000000000000ul,
24 IOPT_PROT_NONE = 0x0000000000000000ul,
25 IOPT_PROT_READ = 0x4000000000000000ul,
26 IOPT_PROT_WRITE = 0x8000000000000000ul,
27 IOPT_PROT_RW = 0xc000000000000000ul,
28 IOPT_COHERENT = 0x2000000000000000ul,
29
30 IOPT_ORDER_MASK = 0x1800000000000000ul,
31 /* order access to same IOID/VC on same address */
32 IOPT_ORDER_ADDR = 0x0800000000000000ul,
33 /* similar, but only after a write access */
34 IOPT_ORDER_WRITES = 0x1000000000000000ul,
35 /* Order all accesses to same IOID/VC */
36 IOPT_ORDER_VC = 0x1800000000000000ul,
37
38 IOPT_RPN_MASK = 0x000003fffffff000ul,
39 IOPT_HINT_MASK = 0x0000000000000800ul,
40 IOPT_IOID_MASK = 0x00000000000007fful,
41
42 IOSTO_ENABLE = 0x8000000000000000ul,
43 IOSTO_ORIGIN = 0x000003fffffff000ul,
44 IOSTO_HW = 0x0000000000000800ul,
45 IOSTO_SW = 0x0000000000000400ul,
46
47 IOCMD_CONF_TE = 0x0000800000000000ul,
48
49 /* memory mapped registers */
50 IOC_PT_CACHE_DIR = 0x000,
51 IOC_ST_CACHE_DIR = 0x800,
52 IOC_PT_CACHE_REG = 0x910,
53 IOC_ST_ORIGIN = 0x918,
54 IOC_CONF = 0x930,
55
56 /* The high bit needs to be set on every DMA address,
57 only 2GB are addressable */
58 BPA_DMA_VALID = 0x80000000,
59 BPA_DMA_MASK = 0x7fffffff,
60};
61
62
63void bpa_init_iommu(void);
64
65#endif
diff --git a/arch/ppc64/kernel/bpa_nvram.c b/arch/ppc64/kernel/bpa_nvram.c
deleted file mode 100644
index 06a119cfceb5..000000000000
--- a/arch/ppc64/kernel/bpa_nvram.c
+++ /dev/null
@@ -1,118 +0,0 @@
1/*
2 * NVRAM for CPBW
3 *
4 * (C) Copyright IBM Corp. 2005
5 *
6 * Authors : Utz Bacher <utz.bacher@de.ibm.com>
7 *
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2, or (at your option)
11 * any later version.
12 *
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to the Free Software
20 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
21 */
22
23#include <linux/fs.h>
24#include <linux/init.h>
25#include <linux/kernel.h>
26#include <linux/spinlock.h>
27#include <linux/types.h>
28
29#include <asm/machdep.h>
30#include <asm/nvram.h>
31#include <asm/prom.h>
32
33static void __iomem *bpa_nvram_start;
34static long bpa_nvram_len;
35static spinlock_t bpa_nvram_lock = SPIN_LOCK_UNLOCKED;
36
37static ssize_t bpa_nvram_read(char *buf, size_t count, loff_t *index)
38{
39 unsigned long flags;
40
41 if (*index >= bpa_nvram_len)
42 return 0;
43 if (*index + count > bpa_nvram_len)
44 count = bpa_nvram_len - *index;
45
46 spin_lock_irqsave(&bpa_nvram_lock, flags);
47
48 memcpy_fromio(buf, bpa_nvram_start + *index, count);
49
50 spin_unlock_irqrestore(&bpa_nvram_lock, flags);
51
52 *index += count;
53 return count;
54}
55
56static ssize_t bpa_nvram_write(char *buf, size_t count, loff_t *index)
57{
58 unsigned long flags;
59
60 if (*index >= bpa_nvram_len)
61 return 0;
62 if (*index + count > bpa_nvram_len)
63 count = bpa_nvram_len - *index;
64
65 spin_lock_irqsave(&bpa_nvram_lock, flags);
66
67 memcpy_toio(bpa_nvram_start + *index, buf, count);
68
69 spin_unlock_irqrestore(&bpa_nvram_lock, flags);
70
71 *index += count;
72 return count;
73}
74
75static ssize_t bpa_nvram_get_size(void)
76{
77 return bpa_nvram_len;
78}
79
80int __init bpa_nvram_init(void)
81{
82 struct device_node *nvram_node;
83 unsigned long *buffer;
84 int proplen;
85 unsigned long nvram_addr;
86 int ret;
87
88 ret = -ENODEV;
89 nvram_node = of_find_node_by_type(NULL, "nvram");
90 if (!nvram_node)
91 goto out;
92
93 ret = -EIO;
94 buffer = (unsigned long *)get_property(nvram_node, "reg", &proplen);
95 if (proplen != 2*sizeof(unsigned long))
96 goto out;
97
98 ret = -ENODEV;
99 nvram_addr = buffer[0];
100 bpa_nvram_len = buffer[1];
101 if ( (!bpa_nvram_len) || (!nvram_addr) )
102 goto out;
103
104 bpa_nvram_start = ioremap(nvram_addr, bpa_nvram_len);
105 if (!bpa_nvram_start)
106 goto out;
107
108 printk(KERN_INFO "BPA NVRAM, %luk mapped to %p\n",
109 bpa_nvram_len >> 10, bpa_nvram_start);
110
111 ppc_md.nvram_read = bpa_nvram_read;
112 ppc_md.nvram_write = bpa_nvram_write;
113 ppc_md.nvram_size = bpa_nvram_get_size;
114
115out:
116 of_node_put(nvram_node);
117 return ret;
118}
diff --git a/arch/ppc64/kernel/bpa_setup.c b/arch/ppc64/kernel/bpa_setup.c
deleted file mode 100644
index c2dc8f282eb8..000000000000
--- a/arch/ppc64/kernel/bpa_setup.c
+++ /dev/null
@@ -1,141 +0,0 @@
1/*
2 * linux/arch/ppc/kernel/bpa_setup.c
3 *
4 * Copyright (C) 1995 Linus Torvalds
5 * Adapted from 'alpha' version by Gary Thomas
6 * Modified by Cort Dougan (cort@cs.nmt.edu)
7 * Modified by PPC64 Team, IBM Corp
8 * Modified by BPA Team, IBM Deutschland Entwicklung GmbH
9 *
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public License
12 * as published by the Free Software Foundation; either version
13 * 2 of the License, or (at your option) any later version.
14 */
15#undef DEBUG
16
17#include <linux/config.h>
18#include <linux/sched.h>
19#include <linux/kernel.h>
20#include <linux/mm.h>
21#include <linux/stddef.h>
22#include <linux/unistd.h>
23#include <linux/slab.h>
24#include <linux/user.h>
25#include <linux/reboot.h>
26#include <linux/init.h>
27#include <linux/delay.h>
28#include <linux/irq.h>
29#include <linux/seq_file.h>
30#include <linux/root_dev.h>
31#include <linux/console.h>
32
33#include <asm/mmu.h>
34#include <asm/processor.h>
35#include <asm/io.h>
36#include <asm/pgtable.h>
37#include <asm/prom.h>
38#include <asm/rtas.h>
39#include <asm/pci-bridge.h>
40#include <asm/iommu.h>
41#include <asm/dma.h>
42#include <asm/machdep.h>
43#include <asm/time.h>
44#include <asm/nvram.h>
45#include <asm/cputable.h>
46#include <asm/ppc-pci.h>
47#include <asm/irq.h>
48
49#include "bpa_iic.h"
50#include "bpa_iommu.h"
51
52#ifdef DEBUG
53#define DBG(fmt...) udbg_printf(fmt)
54#else
55#define DBG(fmt...)
56#endif
57
58void bpa_show_cpuinfo(struct seq_file *m)
59{
60 struct device_node *root;
61 const char *model = "";
62
63 root = of_find_node_by_path("/");
64 if (root)
65 model = get_property(root, "model", NULL);
66 seq_printf(m, "machine\t\t: BPA %s\n", model);
67 of_node_put(root);
68}
69
70static void bpa_progress(char *s, unsigned short hex)
71{
72 printk("*** %04x : %s\n", hex, s ? s : "");
73}
74
75static void __init bpa_setup_arch(void)
76{
77 ppc_md.init_IRQ = iic_init_IRQ;
78 ppc_md.get_irq = iic_get_irq;
79
80#ifdef CONFIG_SMP
81 smp_init_pSeries();
82#endif
83
84 /* init to some ~sane value until calibrate_delay() runs */
85 loops_per_jiffy = 50000000;
86
87 if (ROOT_DEV == 0) {
88 printk("No ramdisk, default root is /dev/hda2\n");
89 ROOT_DEV = Root_HDA2;
90 }
91
92 /* Find and initialize PCI host bridges */
93 init_pci_config_tokens();
94 find_and_init_phbs();
95 spider_init_IRQ();
96#ifdef CONFIG_DUMMY_CONSOLE
97 conswitchp = &dummy_con;
98#endif
99
100 bpa_nvram_init();
101}
102
103/*
104 * Early initialization. Relocation is on but do not reference unbolted pages
105 */
106static void __init bpa_init_early(void)
107{
108 DBG(" -> bpa_init_early()\n");
109
110 hpte_init_native();
111
112 bpa_init_iommu();
113
114 ppc64_interrupt_controller = IC_BPA_IIC;
115
116 DBG(" <- bpa_init_early()\n");
117}
118
119
120static int __init bpa_probe(int platform)
121{
122 if (platform != PLATFORM_BPA)
123 return 0;
124
125 return 1;
126}
127
128struct machdep_calls __initdata bpa_md = {
129 .probe = bpa_probe,
130 .setup_arch = bpa_setup_arch,
131 .init_early = bpa_init_early,
132 .show_cpuinfo = bpa_show_cpuinfo,
133 .restart = rtas_restart,
134 .power_off = rtas_power_off,
135 .halt = rtas_halt,
136 .get_boot_time = rtas_get_boot_time,
137 .get_rtc_time = rtas_get_rtc_time,
138 .set_rtc_time = rtas_set_rtc_time,
139 .calibrate_decr = generic_calibrate_decr,
140 .progress = bpa_progress,
141};
diff --git a/arch/ppc64/kernel/head.S b/arch/ppc64/kernel/head.S
index 929f9f42cf7a..9e8050ea1225 100644
--- a/arch/ppc64/kernel/head.S
+++ b/arch/ppc64/kernel/head.S
@@ -35,7 +35,7 @@
35#include <asm/cputable.h> 35#include <asm/cputable.h>
36#include <asm/setup.h> 36#include <asm/setup.h>
37#include <asm/hvcall.h> 37#include <asm/hvcall.h>
38#include <asm/iSeries/LparMap.h> 38#include <asm/iseries/lpar_map.h>
39#include <asm/thread_info.h> 39#include <asm/thread_info.h>
40 40
41#ifdef CONFIG_PPC_ISERIES 41#ifdef CONFIG_PPC_ISERIES
@@ -195,11 +195,11 @@ exception_marker:
195#define EX_R12 24 195#define EX_R12 24
196#define EX_R13 32 196#define EX_R13 32
197#define EX_SRR0 40 197#define EX_SRR0 40
198#define EX_R3 40 /* SLB miss saves R3, but not SRR0 */
199#define EX_DAR 48 198#define EX_DAR 48
200#define EX_LR 48 /* SLB miss saves LR, but not DAR */
201#define EX_DSISR 56 199#define EX_DSISR 56
202#define EX_CCR 60 200#define EX_CCR 60
201#define EX_R3 64
202#define EX_LR 72
203 203
204#define EXCEPTION_PROLOG_PSERIES(area, label) \ 204#define EXCEPTION_PROLOG_PSERIES(area, label) \
205 mfspr r13,SPRN_SPRG3; /* get paca address into r13 */ \ 205 mfspr r13,SPRN_SPRG3; /* get paca address into r13 */ \
@@ -419,17 +419,22 @@ data_access_slb_pSeries:
419 mtspr SPRN_SPRG1,r13 419 mtspr SPRN_SPRG1,r13
420 RUNLATCH_ON(r13) 420 RUNLATCH_ON(r13)
421 mfspr r13,SPRN_SPRG3 /* get paca address into r13 */ 421 mfspr r13,SPRN_SPRG3 /* get paca address into r13 */
422 std r3,PACA_EXSLB+EX_R3(r13)
423 mfspr r3,SPRN_DAR
422 std r9,PACA_EXSLB+EX_R9(r13) /* save r9 - r12 */ 424 std r9,PACA_EXSLB+EX_R9(r13) /* save r9 - r12 */
425 mfcr r9
426#ifdef __DISABLED__
427 /* Keep that around for when we re-implement dynamic VSIDs */
428 cmpdi r3,0
429 bge slb_miss_user_pseries
430#endif /* __DISABLED__ */
423 std r10,PACA_EXSLB+EX_R10(r13) 431 std r10,PACA_EXSLB+EX_R10(r13)
424 std r11,PACA_EXSLB+EX_R11(r13) 432 std r11,PACA_EXSLB+EX_R11(r13)
425 std r12,PACA_EXSLB+EX_R12(r13) 433 std r12,PACA_EXSLB+EX_R12(r13)
426 std r3,PACA_EXSLB+EX_R3(r13) 434 mfspr r10,SPRN_SPRG1
427 mfspr r9,SPRN_SPRG1 435 std r10,PACA_EXSLB+EX_R13(r13)
428 std r9,PACA_EXSLB+EX_R13(r13)
429 mfcr r9
430 mfspr r12,SPRN_SRR1 /* and SRR1 */ 436 mfspr r12,SPRN_SRR1 /* and SRR1 */
431 mfspr r3,SPRN_DAR 437 b .slb_miss_realmode /* Rel. branch works in real mode */
432 b .do_slb_miss /* Rel. branch works in real mode */
433 438
434 STD_EXCEPTION_PSERIES(0x400, instruction_access) 439 STD_EXCEPTION_PSERIES(0x400, instruction_access)
435 440
@@ -440,17 +445,22 @@ instruction_access_slb_pSeries:
440 mtspr SPRN_SPRG1,r13 445 mtspr SPRN_SPRG1,r13
441 RUNLATCH_ON(r13) 446 RUNLATCH_ON(r13)
442 mfspr r13,SPRN_SPRG3 /* get paca address into r13 */ 447 mfspr r13,SPRN_SPRG3 /* get paca address into r13 */
448 std r3,PACA_EXSLB+EX_R3(r13)
449 mfspr r3,SPRN_SRR0 /* SRR0 is faulting address */
443 std r9,PACA_EXSLB+EX_R9(r13) /* save r9 - r12 */ 450 std r9,PACA_EXSLB+EX_R9(r13) /* save r9 - r12 */
451 mfcr r9
452#ifdef __DISABLED__
453 /* Keep that around for when we re-implement dynamic VSIDs */
454 cmpdi r3,0
455 bge slb_miss_user_pseries
456#endif /* __DISABLED__ */
444 std r10,PACA_EXSLB+EX_R10(r13) 457 std r10,PACA_EXSLB+EX_R10(r13)
445 std r11,PACA_EXSLB+EX_R11(r13) 458 std r11,PACA_EXSLB+EX_R11(r13)
446 std r12,PACA_EXSLB+EX_R12(r13) 459 std r12,PACA_EXSLB+EX_R12(r13)
447 std r3,PACA_EXSLB+EX_R3(r13) 460 mfspr r10,SPRN_SPRG1
448 mfspr r9,SPRN_SPRG1 461 std r10,PACA_EXSLB+EX_R13(r13)
449 std r9,PACA_EXSLB+EX_R13(r13)
450 mfcr r9
451 mfspr r12,SPRN_SRR1 /* and SRR1 */ 462 mfspr r12,SPRN_SRR1 /* and SRR1 */
452 mfspr r3,SPRN_SRR0 /* SRR0 is faulting address */ 463 b .slb_miss_realmode /* Rel. branch works in real mode */
453 b .do_slb_miss /* Rel. branch works in real mode */
454 464
455 STD_EXCEPTION_PSERIES(0x500, hardware_interrupt) 465 STD_EXCEPTION_PSERIES(0x500, hardware_interrupt)
456 STD_EXCEPTION_PSERIES(0x600, alignment) 466 STD_EXCEPTION_PSERIES(0x600, alignment)
@@ -509,6 +519,38 @@ _GLOBAL(do_stab_bolted_pSeries)
509 EXCEPTION_PROLOG_PSERIES(PACA_EXSLB, .do_stab_bolted) 519 EXCEPTION_PROLOG_PSERIES(PACA_EXSLB, .do_stab_bolted)
510 520
511/* 521/*
522 * We have some room here we use that to put
523 * the peries slb miss user trampoline code so it's reasonably
524 * away from slb_miss_user_common to avoid problems with rfid
525 *
526 * This is used for when the SLB miss handler has to go virtual,
527 * which doesn't happen for now anymore but will once we re-implement
528 * dynamic VSIDs for shared page tables
529 */
530#ifdef __DISABLED__
531slb_miss_user_pseries:
532 std r10,PACA_EXGEN+EX_R10(r13)
533 std r11,PACA_EXGEN+EX_R11(r13)
534 std r12,PACA_EXGEN+EX_R12(r13)
535 mfspr r10,SPRG1
536 ld r11,PACA_EXSLB+EX_R9(r13)
537 ld r12,PACA_EXSLB+EX_R3(r13)
538 std r10,PACA_EXGEN+EX_R13(r13)
539 std r11,PACA_EXGEN+EX_R9(r13)
540 std r12,PACA_EXGEN+EX_R3(r13)
541 clrrdi r12,r13,32
542 mfmsr r10
543 mfspr r11,SRR0 /* save SRR0 */
544 ori r12,r12,slb_miss_user_common@l /* virt addr of handler */
545 ori r10,r10,MSR_IR|MSR_DR|MSR_RI
546 mtspr SRR0,r12
547 mfspr r12,SRR1 /* and SRR1 */
548 mtspr SRR1,r10
549 rfid
550 b . /* prevent spec. execution */
551#endif /* __DISABLED__ */
552
553/*
512 * Vectors for the FWNMI option. Share common code. 554 * Vectors for the FWNMI option. Share common code.
513 */ 555 */
514 .globl system_reset_fwnmi 556 .globl system_reset_fwnmi
@@ -559,22 +601,59 @@ END_FTR_SECTION_IFCLR(CPU_FTR_SLB)
559 .globl data_access_slb_iSeries 601 .globl data_access_slb_iSeries
560data_access_slb_iSeries: 602data_access_slb_iSeries:
561 mtspr SPRN_SPRG1,r13 /* save r13 */ 603 mtspr SPRN_SPRG1,r13 /* save r13 */
562 EXCEPTION_PROLOG_ISERIES_1(PACA_EXSLB) 604 mfspr r13,SPRN_SPRG3 /* get paca address into r13 */
563 std r3,PACA_EXSLB+EX_R3(r13) 605 std r3,PACA_EXSLB+EX_R3(r13)
564 ld r12,PACALPPACA+LPPACASRR1(r13)
565 mfspr r3,SPRN_DAR 606 mfspr r3,SPRN_DAR
566 b .do_slb_miss 607 std r9,PACA_EXSLB+EX_R9(r13)
608 mfcr r9
609#ifdef __DISABLED__
610 cmpdi r3,0
611 bge slb_miss_user_iseries
612#endif
613 std r10,PACA_EXSLB+EX_R10(r13)
614 std r11,PACA_EXSLB+EX_R11(r13)
615 std r12,PACA_EXSLB+EX_R12(r13)
616 mfspr r10,SPRN_SPRG1
617 std r10,PACA_EXSLB+EX_R13(r13)
618 ld r12,PACALPPACA+LPPACASRR1(r13);
619 b .slb_miss_realmode
567 620
568 STD_EXCEPTION_ISERIES(0x400, instruction_access, PACA_EXGEN) 621 STD_EXCEPTION_ISERIES(0x400, instruction_access, PACA_EXGEN)
569 622
570 .globl instruction_access_slb_iSeries 623 .globl instruction_access_slb_iSeries
571instruction_access_slb_iSeries: 624instruction_access_slb_iSeries:
572 mtspr SPRN_SPRG1,r13 /* save r13 */ 625 mtspr SPRN_SPRG1,r13 /* save r13 */
573 EXCEPTION_PROLOG_ISERIES_1(PACA_EXSLB) 626 mfspr r13,SPRN_SPRG3 /* get paca address into r13 */
574 std r3,PACA_EXSLB+EX_R3(r13) 627 std r3,PACA_EXSLB+EX_R3(r13)
575 ld r12,PACALPPACA+LPPACASRR1(r13) 628 ld r3,PACALPPACA+LPPACASRR0(r13) /* get SRR0 value */
576 ld r3,PACALPPACA+LPPACASRR0(r13) 629 std r9,PACA_EXSLB+EX_R9(r13)
577 b .do_slb_miss 630 mfcr r9
631#ifdef __DISABLED__
632 cmpdi r3,0
633 bge .slb_miss_user_iseries
634#endif
635 std r10,PACA_EXSLB+EX_R10(r13)
636 std r11,PACA_EXSLB+EX_R11(r13)
637 std r12,PACA_EXSLB+EX_R12(r13)
638 mfspr r10,SPRN_SPRG1
639 std r10,PACA_EXSLB+EX_R13(r13)
640 ld r12,PACALPPACA+LPPACASRR1(r13);
641 b .slb_miss_realmode
642
643#ifdef __DISABLED__
644slb_miss_user_iseries:
645 std r10,PACA_EXGEN+EX_R10(r13)
646 std r11,PACA_EXGEN+EX_R11(r13)
647 std r12,PACA_EXGEN+EX_R12(r13)
648 mfspr r10,SPRG1
649 ld r11,PACA_EXSLB+EX_R9(r13)
650 ld r12,PACA_EXSLB+EX_R3(r13)
651 std r10,PACA_EXGEN+EX_R13(r13)
652 std r11,PACA_EXGEN+EX_R9(r13)
653 std r12,PACA_EXGEN+EX_R3(r13)
654 EXCEPTION_PROLOG_ISERIES_2
655 b slb_miss_user_common
656#endif
578 657
579 MASKABLE_EXCEPTION_ISERIES(0x500, hardware_interrupt) 658 MASKABLE_EXCEPTION_ISERIES(0x500, hardware_interrupt)
580 STD_EXCEPTION_ISERIES(0x600, alignment, PACA_EXGEN) 659 STD_EXCEPTION_ISERIES(0x600, alignment, PACA_EXGEN)
@@ -809,6 +888,126 @@ instruction_access_common:
809 li r5,0x400 888 li r5,0x400
810 b .do_hash_page /* Try to handle as hpte fault */ 889 b .do_hash_page /* Try to handle as hpte fault */
811 890
891/*
892 * Here is the common SLB miss user that is used when going to virtual
893 * mode for SLB misses, that is currently not used
894 */
895#ifdef __DISABLED__
896 .align 7
897 .globl slb_miss_user_common
898slb_miss_user_common:
899 mflr r10
900 std r3,PACA_EXGEN+EX_DAR(r13)
901 stw r9,PACA_EXGEN+EX_CCR(r13)
902 std r10,PACA_EXGEN+EX_LR(r13)
903 std r11,PACA_EXGEN+EX_SRR0(r13)
904 bl .slb_allocate_user
905
906 ld r10,PACA_EXGEN+EX_LR(r13)
907 ld r3,PACA_EXGEN+EX_R3(r13)
908 lwz r9,PACA_EXGEN+EX_CCR(r13)
909 ld r11,PACA_EXGEN+EX_SRR0(r13)
910 mtlr r10
911 beq- slb_miss_fault
912
913 andi. r10,r12,MSR_RI /* check for unrecoverable exception */
914 beq- unrecov_user_slb
915 mfmsr r10
916
917.machine push
918.machine "power4"
919 mtcrf 0x80,r9
920.machine pop
921
922 clrrdi r10,r10,2 /* clear RI before setting SRR0/1 */
923 mtmsrd r10,1
924
925 mtspr SRR0,r11
926 mtspr SRR1,r12
927
928 ld r9,PACA_EXGEN+EX_R9(r13)
929 ld r10,PACA_EXGEN+EX_R10(r13)
930 ld r11,PACA_EXGEN+EX_R11(r13)
931 ld r12,PACA_EXGEN+EX_R12(r13)
932 ld r13,PACA_EXGEN+EX_R13(r13)
933 rfid
934 b .
935
936slb_miss_fault:
937 EXCEPTION_PROLOG_COMMON(0x380, PACA_EXGEN)
938 ld r4,PACA_EXGEN+EX_DAR(r13)
939 li r5,0
940 std r4,_DAR(r1)
941 std r5,_DSISR(r1)
942 b .handle_page_fault
943
944unrecov_user_slb:
945 EXCEPTION_PROLOG_COMMON(0x4200, PACA_EXGEN)
946 DISABLE_INTS
947 bl .save_nvgprs
9481: addi r3,r1,STACK_FRAME_OVERHEAD
949 bl .unrecoverable_exception
950 b 1b
951
952#endif /* __DISABLED__ */
953
954
955/*
956 * r13 points to the PACA, r9 contains the saved CR,
957 * r12 contain the saved SRR1, SRR0 is still ready for return
958 * r3 has the faulting address
959 * r9 - r13 are saved in paca->exslb.
960 * r3 is saved in paca->slb_r3
961 * We assume we aren't going to take any exceptions during this procedure.
962 */
963_GLOBAL(slb_miss_realmode)
964 mflr r10
965
966 stw r9,PACA_EXSLB+EX_CCR(r13) /* save CR in exc. frame */
967 std r10,PACA_EXSLB+EX_LR(r13) /* save LR */
968
969 bl .slb_allocate_realmode
970
971 /* All done -- return from exception. */
972
973 ld r10,PACA_EXSLB+EX_LR(r13)
974 ld r3,PACA_EXSLB+EX_R3(r13)
975 lwz r9,PACA_EXSLB+EX_CCR(r13) /* get saved CR */
976#ifdef CONFIG_PPC_ISERIES
977 ld r11,PACALPPACA+LPPACASRR0(r13) /* get SRR0 value */
978#endif /* CONFIG_PPC_ISERIES */
979
980 mtlr r10
981
982 andi. r10,r12,MSR_RI /* check for unrecoverable exception */
983 beq- unrecov_slb
984
985.machine push
986.machine "power4"
987 mtcrf 0x80,r9
988 mtcrf 0x01,r9 /* slb_allocate uses cr0 and cr7 */
989.machine pop
990
991#ifdef CONFIG_PPC_ISERIES
992 mtspr SPRN_SRR0,r11
993 mtspr SPRN_SRR1,r12
994#endif /* CONFIG_PPC_ISERIES */
995 ld r9,PACA_EXSLB+EX_R9(r13)
996 ld r10,PACA_EXSLB+EX_R10(r13)
997 ld r11,PACA_EXSLB+EX_R11(r13)
998 ld r12,PACA_EXSLB+EX_R12(r13)
999 ld r13,PACA_EXSLB+EX_R13(r13)
1000 rfid
1001 b . /* prevent speculative execution */
1002
1003unrecov_slb:
1004 EXCEPTION_PROLOG_COMMON(0x4100, PACA_EXSLB)
1005 DISABLE_INTS
1006 bl .save_nvgprs
10071: addi r3,r1,STACK_FRAME_OVERHEAD
1008 bl .unrecoverable_exception
1009 b 1b
1010
812 .align 7 1011 .align 7
813 .globl hardware_interrupt_common 1012 .globl hardware_interrupt_common
814 .globl hardware_interrupt_entry 1013 .globl hardware_interrupt_entry
@@ -1139,62 +1338,6 @@ _GLOBAL(do_stab_bolted)
1139 b . /* prevent speculative execution */ 1338 b . /* prevent speculative execution */
1140 1339
1141/* 1340/*
1142 * r13 points to the PACA, r9 contains the saved CR,
1143 * r11 and r12 contain the saved SRR0 and SRR1.
1144 * r3 has the faulting address
1145 * r9 - r13 are saved in paca->exslb.
1146 * r3 is saved in paca->slb_r3
1147 * We assume we aren't going to take any exceptions during this procedure.
1148 */
1149_GLOBAL(do_slb_miss)
1150 mflr r10
1151
1152 stw r9,PACA_EXSLB+EX_CCR(r13) /* save CR in exc. frame */
1153 std r10,PACA_EXSLB+EX_LR(r13) /* save LR */
1154
1155 bl .slb_allocate /* handle it */
1156
1157 /* All done -- return from exception. */
1158
1159 ld r10,PACA_EXSLB+EX_LR(r13)
1160 ld r3,PACA_EXSLB+EX_R3(r13)
1161 lwz r9,PACA_EXSLB+EX_CCR(r13) /* get saved CR */
1162#ifdef CONFIG_PPC_ISERIES
1163 ld r11,PACALPPACA+LPPACASRR0(r13) /* get SRR0 value */
1164#endif /* CONFIG_PPC_ISERIES */
1165
1166 mtlr r10
1167
1168 andi. r10,r12,MSR_RI /* check for unrecoverable exception */
1169 beq- unrecov_slb
1170
1171.machine push
1172.machine "power4"
1173 mtcrf 0x80,r9
1174 mtcrf 0x01,r9 /* slb_allocate uses cr0 and cr7 */
1175.machine pop
1176
1177#ifdef CONFIG_PPC_ISERIES
1178 mtspr SPRN_SRR0,r11
1179 mtspr SPRN_SRR1,r12
1180#endif /* CONFIG_PPC_ISERIES */
1181 ld r9,PACA_EXSLB+EX_R9(r13)
1182 ld r10,PACA_EXSLB+EX_R10(r13)
1183 ld r11,PACA_EXSLB+EX_R11(r13)
1184 ld r12,PACA_EXSLB+EX_R12(r13)
1185 ld r13,PACA_EXSLB+EX_R13(r13)
1186 rfid
1187 b . /* prevent speculative execution */
1188
1189unrecov_slb:
1190 EXCEPTION_PROLOG_COMMON(0x4100, PACA_EXSLB)
1191 DISABLE_INTS
1192 bl .save_nvgprs
11931: addi r3,r1,STACK_FRAME_OVERHEAD
1194 bl .unrecoverable_exception
1195 b 1b
1196
1197/*
1198 * Space for CPU0's segment table. 1341 * Space for CPU0's segment table.
1199 * 1342 *
1200 * On iSeries, the hypervisor must fill in at least one entry before 1343 * On iSeries, the hypervisor must fill in at least one entry before
@@ -1569,7 +1712,10 @@ _GLOBAL(__secondary_start)
1569#endif 1712#endif
1570 /* Initialize the first segment table (or SLB) entry */ 1713 /* Initialize the first segment table (or SLB) entry */
1571 ld r3,PACASTABVIRT(r13) /* get addr of segment table */ 1714 ld r3,PACASTABVIRT(r13) /* get addr of segment table */
1715BEGIN_FTR_SECTION
1572 bl .stab_initialize 1716 bl .stab_initialize
1717END_FTR_SECTION_IFCLR(CPU_FTR_SLB)
1718 bl .slb_initialize
1573 1719
1574 /* Initialize the kernel stack. Just a repeat for iSeries. */ 1720 /* Initialize the kernel stack. Just a repeat for iSeries. */
1575 LOADADDR(r3,current_set) 1721 LOADADDR(r3,current_set)
@@ -1914,24 +2060,6 @@ _GLOBAL(hmt_start_secondary)
1914 blr 2060 blr
1915#endif 2061#endif
1916 2062
1917#if defined(CONFIG_KEXEC) || defined(CONFIG_SMP)
1918_GLOBAL(smp_release_cpus)
1919 /* All secondary cpus are spinning on a common
1920 * spinloop, release them all now so they can start
1921 * to spin on their individual paca spinloops.
1922 * For non SMP kernels, the secondary cpus never
1923 * get out of the common spinloop.
1924 * XXX This does nothing useful on iSeries, secondaries are
1925 * already waiting on their paca.
1926 */
1927 li r3,1
1928 LOADADDR(r5,__secondary_hold_spinloop)
1929 std r3,0(r5)
1930 sync
1931 blr
1932#endif /* CONFIG_SMP */
1933
1934
1935/* 2063/*
1936 * We put a few things here that have to be page-aligned. 2064 * We put a few things here that have to be page-aligned.
1937 * This stuff goes at the beginning of the bss, which is page-aligned. 2065 * This stuff goes at the beginning of the bss, which is page-aligned.
diff --git a/arch/ppc64/kernel/irq.c b/arch/ppc64/kernel/irq.c
index f41afe545045..87474584033f 100644
--- a/arch/ppc64/kernel/irq.c
+++ b/arch/ppc64/kernel/irq.c
@@ -52,7 +52,7 @@
52#include <asm/cache.h> 52#include <asm/cache.h>
53#include <asm/prom.h> 53#include <asm/prom.h>
54#include <asm/ptrace.h> 54#include <asm/ptrace.h>
55#include <asm/iSeries/ItLpQueue.h> 55#include <asm/iseries/it_lp_queue.h>
56#include <asm/machdep.h> 56#include <asm/machdep.h>
57#include <asm/paca.h> 57#include <asm/paca.h>
58 58
@@ -392,7 +392,7 @@ int virt_irq_create_mapping(unsigned int real_irq)
392 if (ppc64_interrupt_controller == IC_OPEN_PIC) 392 if (ppc64_interrupt_controller == IC_OPEN_PIC)
393 return real_irq; /* no mapping for openpic (for now) */ 393 return real_irq; /* no mapping for openpic (for now) */
394 394
395 if (ppc64_interrupt_controller == IC_BPA_IIC) 395 if (ppc64_interrupt_controller == IC_CELL_PIC)
396 return real_irq; /* no mapping for iic either */ 396 return real_irq; /* no mapping for iic either */
397 397
398 /* don't map interrupts < MIN_VIRT_IRQ */ 398 /* don't map interrupts < MIN_VIRT_IRQ */
diff --git a/arch/ppc64/kernel/kprobes.c b/arch/ppc64/kernel/kprobes.c
index ed876a5178ae..511af54e6230 100644
--- a/arch/ppc64/kernel/kprobes.c
+++ b/arch/ppc64/kernel/kprobes.c
@@ -30,19 +30,14 @@
30#include <linux/config.h> 30#include <linux/config.h>
31#include <linux/kprobes.h> 31#include <linux/kprobes.h>
32#include <linux/ptrace.h> 32#include <linux/ptrace.h>
33#include <linux/spinlock.h>
34#include <linux/preempt.h> 33#include <linux/preempt.h>
35#include <asm/cacheflush.h> 34#include <asm/cacheflush.h>
36#include <asm/kdebug.h> 35#include <asm/kdebug.h>
37#include <asm/sstep.h> 36#include <asm/sstep.h>
38 37
39static DECLARE_MUTEX(kprobe_mutex); 38static DECLARE_MUTEX(kprobe_mutex);
40 39DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
41static struct kprobe *current_kprobe; 40DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
42static unsigned long kprobe_status, kprobe_saved_msr;
43static struct kprobe *kprobe_prev;
44static unsigned long kprobe_status_prev, kprobe_saved_msr_prev;
45static struct pt_regs jprobe_saved_regs;
46 41
47int __kprobes arch_prepare_kprobe(struct kprobe *p) 42int __kprobes arch_prepare_kprobe(struct kprobe *p)
48{ 43{
@@ -108,20 +103,28 @@ static inline void prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
108 regs->nip = (unsigned long)p->ainsn.insn; 103 regs->nip = (unsigned long)p->ainsn.insn;
109} 104}
110 105
111static inline void save_previous_kprobe(void) 106static inline void save_previous_kprobe(struct kprobe_ctlblk *kcb)
107{
108 kcb->prev_kprobe.kp = kprobe_running();
109 kcb->prev_kprobe.status = kcb->kprobe_status;
110 kcb->prev_kprobe.saved_msr = kcb->kprobe_saved_msr;
111}
112
113static inline void restore_previous_kprobe(struct kprobe_ctlblk *kcb)
112{ 114{
113 kprobe_prev = current_kprobe; 115 __get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
114 kprobe_status_prev = kprobe_status; 116 kcb->kprobe_status = kcb->prev_kprobe.status;
115 kprobe_saved_msr_prev = kprobe_saved_msr; 117 kcb->kprobe_saved_msr = kcb->prev_kprobe.saved_msr;
116} 118}
117 119
118static inline void restore_previous_kprobe(void) 120static inline void set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
121 struct kprobe_ctlblk *kcb)
119{ 122{
120 current_kprobe = kprobe_prev; 123 __get_cpu_var(current_kprobe) = p;
121 kprobe_status = kprobe_status_prev; 124 kcb->kprobe_saved_msr = regs->msr;
122 kprobe_saved_msr = kprobe_saved_msr_prev;
123} 125}
124 126
127/* Called with kretprobe_lock held */
125void __kprobes arch_prepare_kretprobe(struct kretprobe *rp, 128void __kprobes arch_prepare_kretprobe(struct kretprobe *rp,
126 struct pt_regs *regs) 129 struct pt_regs *regs)
127{ 130{
@@ -145,19 +148,24 @@ static inline int kprobe_handler(struct pt_regs *regs)
145 struct kprobe *p; 148 struct kprobe *p;
146 int ret = 0; 149 int ret = 0;
147 unsigned int *addr = (unsigned int *)regs->nip; 150 unsigned int *addr = (unsigned int *)regs->nip;
151 struct kprobe_ctlblk *kcb;
152
153 /*
154 * We don't want to be preempted for the entire
155 * duration of kprobe processing
156 */
157 preempt_disable();
158 kcb = get_kprobe_ctlblk();
148 159
149 /* Check we're not actually recursing */ 160 /* Check we're not actually recursing */
150 if (kprobe_running()) { 161 if (kprobe_running()) {
151 /* We *are* holding lock here, so this is safe.
152 Disarm the probe we just hit, and ignore it. */
153 p = get_kprobe(addr); 162 p = get_kprobe(addr);
154 if (p) { 163 if (p) {
155 kprobe_opcode_t insn = *p->ainsn.insn; 164 kprobe_opcode_t insn = *p->ainsn.insn;
156 if (kprobe_status == KPROBE_HIT_SS && 165 if (kcb->kprobe_status == KPROBE_HIT_SS &&
157 is_trap(insn)) { 166 is_trap(insn)) {
158 regs->msr &= ~MSR_SE; 167 regs->msr &= ~MSR_SE;
159 regs->msr |= kprobe_saved_msr; 168 regs->msr |= kcb->kprobe_saved_msr;
160 unlock_kprobes();
161 goto no_kprobe; 169 goto no_kprobe;
162 } 170 }
163 /* We have reentered the kprobe_handler(), since 171 /* We have reentered the kprobe_handler(), since
@@ -166,27 +174,24 @@ static inline int kprobe_handler(struct pt_regs *regs)
166 * just single step on the instruction of the new probe 174 * just single step on the instruction of the new probe
167 * without calling any user handlers. 175 * without calling any user handlers.
168 */ 176 */
169 save_previous_kprobe(); 177 save_previous_kprobe(kcb);
170 current_kprobe = p; 178 set_current_kprobe(p, regs, kcb);
171 kprobe_saved_msr = regs->msr; 179 kcb->kprobe_saved_msr = regs->msr;
172 p->nmissed++; 180 p->nmissed++;
173 prepare_singlestep(p, regs); 181 prepare_singlestep(p, regs);
174 kprobe_status = KPROBE_REENTER; 182 kcb->kprobe_status = KPROBE_REENTER;
175 return 1; 183 return 1;
176 } else { 184 } else {
177 p = current_kprobe; 185 p = __get_cpu_var(current_kprobe);
178 if (p->break_handler && p->break_handler(p, regs)) { 186 if (p->break_handler && p->break_handler(p, regs)) {
179 goto ss_probe; 187 goto ss_probe;
180 } 188 }
181 } 189 }
182 /* If it's not ours, can't be delete race, (we hold lock). */
183 goto no_kprobe; 190 goto no_kprobe;
184 } 191 }
185 192
186 lock_kprobes();
187 p = get_kprobe(addr); 193 p = get_kprobe(addr);
188 if (!p) { 194 if (!p) {
189 unlock_kprobes();
190 if (*addr != BREAKPOINT_INSTRUCTION) { 195 if (*addr != BREAKPOINT_INSTRUCTION) {
191 /* 196 /*
192 * PowerPC has multiple variants of the "trap" 197 * PowerPC has multiple variants of the "trap"
@@ -209,24 +214,19 @@ static inline int kprobe_handler(struct pt_regs *regs)
209 goto no_kprobe; 214 goto no_kprobe;
210 } 215 }
211 216
212 kprobe_status = KPROBE_HIT_ACTIVE; 217 kcb->kprobe_status = KPROBE_HIT_ACTIVE;
213 current_kprobe = p; 218 set_current_kprobe(p, regs, kcb);
214 kprobe_saved_msr = regs->msr;
215 if (p->pre_handler && p->pre_handler(p, regs)) 219 if (p->pre_handler && p->pre_handler(p, regs))
216 /* handler has already set things up, so skip ss setup */ 220 /* handler has already set things up, so skip ss setup */
217 return 1; 221 return 1;
218 222
219ss_probe: 223ss_probe:
220 prepare_singlestep(p, regs); 224 prepare_singlestep(p, regs);
221 kprobe_status = KPROBE_HIT_SS; 225 kcb->kprobe_status = KPROBE_HIT_SS;
222 /*
223 * This preempt_disable() matches the preempt_enable_no_resched()
224 * in post_kprobe_handler().
225 */
226 preempt_disable();
227 return 1; 226 return 1;
228 227
229no_kprobe: 228no_kprobe:
229 preempt_enable_no_resched();
230 return ret; 230 return ret;
231} 231}
232 232
@@ -251,9 +251,10 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
251 struct kretprobe_instance *ri = NULL; 251 struct kretprobe_instance *ri = NULL;
252 struct hlist_head *head; 252 struct hlist_head *head;
253 struct hlist_node *node, *tmp; 253 struct hlist_node *node, *tmp;
254 unsigned long orig_ret_address = 0; 254 unsigned long flags, orig_ret_address = 0;
255 unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline; 255 unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline;
256 256
257 spin_lock_irqsave(&kretprobe_lock, flags);
257 head = kretprobe_inst_table_head(current); 258 head = kretprobe_inst_table_head(current);
258 259
259 /* 260 /*
@@ -292,12 +293,14 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
292 BUG_ON(!orig_ret_address || (orig_ret_address == trampoline_address)); 293 BUG_ON(!orig_ret_address || (orig_ret_address == trampoline_address));
293 regs->nip = orig_ret_address; 294 regs->nip = orig_ret_address;
294 295
295 unlock_kprobes(); 296 reset_current_kprobe();
297 spin_unlock_irqrestore(&kretprobe_lock, flags);
298 preempt_enable_no_resched();
296 299
297 /* 300 /*
298 * By returning a non-zero value, we are telling 301 * By returning a non-zero value, we are telling
299 * kprobe_handler() that we have handled unlocking 302 * kprobe_handler() that we don't want the post_handler
300 * and re-enabling preemption. 303 * to run (and have re-enabled preemption)
301 */ 304 */
302 return 1; 305 return 1;
303} 306}
@@ -323,23 +326,26 @@ static void __kprobes resume_execution(struct kprobe *p, struct pt_regs *regs)
323 326
324static inline int post_kprobe_handler(struct pt_regs *regs) 327static inline int post_kprobe_handler(struct pt_regs *regs)
325{ 328{
326 if (!kprobe_running()) 329 struct kprobe *cur = kprobe_running();
330 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
331
332 if (!cur)
327 return 0; 333 return 0;
328 334
329 if ((kprobe_status != KPROBE_REENTER) && current_kprobe->post_handler) { 335 if ((kcb->kprobe_status != KPROBE_REENTER) && cur->post_handler) {
330 kprobe_status = KPROBE_HIT_SSDONE; 336 kcb->kprobe_status = KPROBE_HIT_SSDONE;
331 current_kprobe->post_handler(current_kprobe, regs, 0); 337 cur->post_handler(cur, regs, 0);
332 } 338 }
333 339
334 resume_execution(current_kprobe, regs); 340 resume_execution(cur, regs);
335 regs->msr |= kprobe_saved_msr; 341 regs->msr |= kcb->kprobe_saved_msr;
336 342
337 /*Restore back the original saved kprobes variables and continue. */ 343 /*Restore back the original saved kprobes variables and continue. */
338 if (kprobe_status == KPROBE_REENTER) { 344 if (kcb->kprobe_status == KPROBE_REENTER) {
339 restore_previous_kprobe(); 345 restore_previous_kprobe(kcb);
340 goto out; 346 goto out;
341 } 347 }
342 unlock_kprobes(); 348 reset_current_kprobe();
343out: 349out:
344 preempt_enable_no_resched(); 350 preempt_enable_no_resched();
345 351
@@ -354,19 +360,20 @@ out:
354 return 1; 360 return 1;
355} 361}
356 362
357/* Interrupts disabled, kprobe_lock held. */
358static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) 363static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr)
359{ 364{
360 if (current_kprobe->fault_handler 365 struct kprobe *cur = kprobe_running();
361 && current_kprobe->fault_handler(current_kprobe, regs, trapnr)) 366 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
367
368 if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr))
362 return 1; 369 return 1;
363 370
364 if (kprobe_status & KPROBE_HIT_SS) { 371 if (kcb->kprobe_status & KPROBE_HIT_SS) {
365 resume_execution(current_kprobe, regs); 372 resume_execution(cur, regs);
366 regs->msr &= ~MSR_SE; 373 regs->msr &= ~MSR_SE;
367 regs->msr |= kprobe_saved_msr; 374 regs->msr |= kcb->kprobe_saved_msr;
368 375
369 unlock_kprobes(); 376 reset_current_kprobe();
370 preempt_enable_no_resched(); 377 preempt_enable_no_resched();
371 } 378 }
372 return 0; 379 return 0;
@@ -381,11 +388,6 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
381 struct die_args *args = (struct die_args *)data; 388 struct die_args *args = (struct die_args *)data;
382 int ret = NOTIFY_DONE; 389 int ret = NOTIFY_DONE;
383 390
384 /*
385 * Interrupts are not disabled here. We need to disable
386 * preemption, because kprobe_running() uses smp_processor_id().
387 */
388 preempt_disable();
389 switch (val) { 391 switch (val) {
390 case DIE_BPT: 392 case DIE_BPT:
391 if (kprobe_handler(args->regs)) 393 if (kprobe_handler(args->regs))
@@ -396,22 +398,25 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
396 ret = NOTIFY_STOP; 398 ret = NOTIFY_STOP;
397 break; 399 break;
398 case DIE_PAGE_FAULT: 400 case DIE_PAGE_FAULT:
401 /* kprobe_running() needs smp_processor_id() */
402 preempt_disable();
399 if (kprobe_running() && 403 if (kprobe_running() &&
400 kprobe_fault_handler(args->regs, args->trapnr)) 404 kprobe_fault_handler(args->regs, args->trapnr))
401 ret = NOTIFY_STOP; 405 ret = NOTIFY_STOP;
406 preempt_enable();
402 break; 407 break;
403 default: 408 default:
404 break; 409 break;
405 } 410 }
406 preempt_enable_no_resched();
407 return ret; 411 return ret;
408} 412}
409 413
410int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) 414int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs)
411{ 415{
412 struct jprobe *jp = container_of(p, struct jprobe, kp); 416 struct jprobe *jp = container_of(p, struct jprobe, kp);
417 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
413 418
414 memcpy(&jprobe_saved_regs, regs, sizeof(struct pt_regs)); 419 memcpy(&kcb->jprobe_saved_regs, regs, sizeof(struct pt_regs));
415 420
416 /* setup return addr to the jprobe handler routine */ 421 /* setup return addr to the jprobe handler routine */
417 regs->nip = (unsigned long)(((func_descr_t *)jp->entry)->entry); 422 regs->nip = (unsigned long)(((func_descr_t *)jp->entry)->entry);
@@ -431,12 +436,15 @@ void __kprobes jprobe_return_end(void)
431 436
432int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) 437int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
433{ 438{
439 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
440
434 /* 441 /*
435 * FIXME - we should ideally be validating that we got here 'cos 442 * FIXME - we should ideally be validating that we got here 'cos
436 * of the "trap" in jprobe_return() above, before restoring the 443 * of the "trap" in jprobe_return() above, before restoring the
437 * saved regs... 444 * saved regs...
438 */ 445 */
439 memcpy(regs, &jprobe_saved_regs, sizeof(struct pt_regs)); 446 memcpy(regs, &kcb->jprobe_saved_regs, sizeof(struct pt_regs));
447 preempt_enable_no_resched();
440 return 1; 448 return 1;
441} 449}
442 450
diff --git a/arch/ppc64/kernel/lparcfg.c b/arch/ppc64/kernel/lparcfg.c
index cae19bbd5acd..3e7b2f28ec83 100644
--- a/arch/ppc64/kernel/lparcfg.c
+++ b/arch/ppc64/kernel/lparcfg.c
@@ -26,14 +26,14 @@
26#include <linux/init.h> 26#include <linux/init.h>
27#include <linux/seq_file.h> 27#include <linux/seq_file.h>
28#include <asm/uaccess.h> 28#include <asm/uaccess.h>
29#include <asm/iSeries/HvLpConfig.h> 29#include <asm/iseries/hv_lp_config.h>
30#include <asm/lppaca.h> 30#include <asm/lppaca.h>
31#include <asm/hvcall.h> 31#include <asm/hvcall.h>
32#include <asm/firmware.h> 32#include <asm/firmware.h>
33#include <asm/rtas.h> 33#include <asm/rtas.h>
34#include <asm/system.h> 34#include <asm/system.h>
35#include <asm/time.h> 35#include <asm/time.h>
36#include <asm/iSeries/ItExtVpdPanel.h> 36#include <asm/iseries/it_exp_vpd_panel.h>
37#include <asm/prom.h> 37#include <asm/prom.h>
38 38
39#define MODULE_VERS "1.6" 39#define MODULE_VERS "1.6"
@@ -599,9 +599,7 @@ int __init lparcfg_init(void)
599void __exit lparcfg_cleanup(void) 599void __exit lparcfg_cleanup(void)
600{ 600{
601 if (proc_ppc64_lparcfg) { 601 if (proc_ppc64_lparcfg) {
602 if (proc_ppc64_lparcfg->data) { 602 kfree(proc_ppc64_lparcfg->data);
603 kfree(proc_ppc64_lparcfg->data);
604 }
605 remove_proc_entry("lparcfg", proc_ppc64_lparcfg->parent); 603 remove_proc_entry("lparcfg", proc_ppc64_lparcfg->parent);
606 } 604 }
607} 605}
diff --git a/arch/ppc64/kernel/machine_kexec.c b/arch/ppc64/kernel/machine_kexec.c
index bf7cc4f8210f..ff8679f260f3 100644
--- a/arch/ppc64/kernel/machine_kexec.c
+++ b/arch/ppc64/kernel/machine_kexec.c
@@ -244,7 +244,6 @@ static void kexec_prepare_cpus(void)
244 244
245static void kexec_prepare_cpus(void) 245static void kexec_prepare_cpus(void)
246{ 246{
247 extern void smp_release_cpus(void);
248 /* 247 /*
249 * move the secondarys to us so that we can copy 248 * move the secondarys to us so that we can copy
250 * the new kernel 0-0x100 safely 249 * the new kernel 0-0x100 safely
diff --git a/arch/ppc64/kernel/pacaData.c b/arch/ppc64/kernel/pacaData.c
index 33a2d8db3f21..3133c72b28ec 100644
--- a/arch/ppc64/kernel/pacaData.c
+++ b/arch/ppc64/kernel/pacaData.c
@@ -17,13 +17,13 @@
17#include <asm/page.h> 17#include <asm/page.h>
18 18
19#include <asm/lppaca.h> 19#include <asm/lppaca.h>
20#include <asm/iSeries/ItLpQueue.h> 20#include <asm/iseries/it_lp_queue.h>
21#include <asm/paca.h> 21#include <asm/paca.h>
22 22
23static union { 23static union {
24 struct systemcfg data; 24 struct systemcfg data;
25 u8 page[PAGE_SIZE]; 25 u8 page[PAGE_SIZE];
26} systemcfg_store __page_aligned; 26} systemcfg_store __attribute__((__section__(".data.page.aligned")));
27struct systemcfg *systemcfg = &systemcfg_store.data; 27struct systemcfg *systemcfg = &systemcfg_store.data;
28EXPORT_SYMBOL(systemcfg); 28EXPORT_SYMBOL(systemcfg);
29 29
diff --git a/arch/ppc64/kernel/pci_dn.c b/arch/ppc64/kernel/pci_dn.c
index 493bbe43f5b4..1a443a7ada4c 100644
--- a/arch/ppc64/kernel/pci_dn.c
+++ b/arch/ppc64/kernel/pci_dn.c
@@ -181,13 +181,14 @@ EXPORT_SYMBOL(fetch_dev_dn);
181static int pci_dn_reconfig_notifier(struct notifier_block *nb, unsigned long action, void *node) 181static int pci_dn_reconfig_notifier(struct notifier_block *nb, unsigned long action, void *node)
182{ 182{
183 struct device_node *np = node; 183 struct device_node *np = node;
184 struct pci_dn *pci; 184 struct pci_dn *pci = NULL;
185 int err = NOTIFY_OK; 185 int err = NOTIFY_OK;
186 186
187 switch (action) { 187 switch (action) {
188 case PSERIES_RECONFIG_ADD: 188 case PSERIES_RECONFIG_ADD:
189 pci = np->parent->data; 189 pci = np->parent->data;
190 update_dn_pci_info(np, pci->phb); 190 if (pci)
191 update_dn_pci_info(np, pci->phb);
191 break; 192 break;
192 default: 193 default:
193 err = NOTIFY_DONE; 194 err = NOTIFY_DONE;
diff --git a/arch/ppc64/kernel/proc_ppc64.c b/arch/ppc64/kernel/proc_ppc64.c
index a87c66a9652a..24e955ee9487 100644
--- a/arch/ppc64/kernel/proc_ppc64.c
+++ b/arch/ppc64/kernel/proc_ppc64.c
@@ -53,7 +53,7 @@ static int __init proc_ppc64_create(void)
53 if (!root) 53 if (!root)
54 return 1; 54 return 1;
55 55
56 if (!(systemcfg->platform & (PLATFORM_PSERIES | PLATFORM_BPA))) 56 if (!(systemcfg->platform & (PLATFORM_PSERIES | PLATFORM_CELL)))
57 return 0; 57 return 0;
58 58
59 if (!proc_mkdir("rtas", root)) 59 if (!proc_mkdir("rtas", root))
diff --git a/arch/ppc64/kernel/prom.c b/arch/ppc64/kernel/prom.c
index 97bfceb5353b..dece31e58bc4 100644
--- a/arch/ppc64/kernel/prom.c
+++ b/arch/ppc64/kernel/prom.c
@@ -635,10 +635,10 @@ static inline char *find_flat_dt_string(u32 offset)
635 * used to extract the memory informations at boot before we can 635 * used to extract the memory informations at boot before we can
636 * unflatten the tree 636 * unflatten the tree
637 */ 637 */
638static int __init scan_flat_dt(int (*it)(unsigned long node, 638int __init of_scan_flat_dt(int (*it)(unsigned long node,
639 const char *uname, int depth, 639 const char *uname, int depth,
640 void *data), 640 void *data),
641 void *data) 641 void *data)
642{ 642{
643 unsigned long p = ((unsigned long)initial_boot_params) + 643 unsigned long p = ((unsigned long)initial_boot_params) +
644 initial_boot_params->off_dt_struct; 644 initial_boot_params->off_dt_struct;
@@ -695,8 +695,8 @@ static int __init scan_flat_dt(int (*it)(unsigned long node,
695 * This function can be used within scan_flattened_dt callback to get 695 * This function can be used within scan_flattened_dt callback to get
696 * access to properties 696 * access to properties
697 */ 697 */
698static void* __init get_flat_dt_prop(unsigned long node, const char *name, 698void* __init of_get_flat_dt_prop(unsigned long node, const char *name,
699 unsigned long *size) 699 unsigned long *size)
700{ 700{
701 unsigned long p = node; 701 unsigned long p = node;
702 702
@@ -996,7 +996,7 @@ void __init unflatten_device_tree(void)
996static int __init early_init_dt_scan_cpus(unsigned long node, 996static int __init early_init_dt_scan_cpus(unsigned long node,
997 const char *uname, int depth, void *data) 997 const char *uname, int depth, void *data)
998{ 998{
999 char *type = get_flat_dt_prop(node, "device_type", NULL); 999 char *type = of_get_flat_dt_prop(node, "device_type", NULL);
1000 u32 *prop; 1000 u32 *prop;
1001 unsigned long size; 1001 unsigned long size;
1002 1002
@@ -1004,17 +1004,6 @@ static int __init early_init_dt_scan_cpus(unsigned long node,
1004 if (type == NULL || strcmp(type, "cpu") != 0) 1004 if (type == NULL || strcmp(type, "cpu") != 0)
1005 return 0; 1005 return 0;
1006 1006
1007 /* On LPAR, look for the first ibm,pft-size property for the hash table size
1008 */
1009 if (systemcfg->platform == PLATFORM_PSERIES_LPAR && ppc64_pft_size == 0) {
1010 u32 *pft_size;
1011 pft_size = (u32 *)get_flat_dt_prop(node, "ibm,pft-size", NULL);
1012 if (pft_size != NULL) {
1013 /* pft_size[0] is the NUMA CEC cookie */
1014 ppc64_pft_size = pft_size[1];
1015 }
1016 }
1017
1018 if (initial_boot_params && initial_boot_params->version >= 2) { 1007 if (initial_boot_params && initial_boot_params->version >= 2) {
1019 /* version 2 of the kexec param format adds the phys cpuid 1008 /* version 2 of the kexec param format adds the phys cpuid
1020 * of booted proc. 1009 * of booted proc.
@@ -1023,8 +1012,9 @@ static int __init early_init_dt_scan_cpus(unsigned long node,
1023 boot_cpuid = 0; 1012 boot_cpuid = 0;
1024 } else { 1013 } else {
1025 /* Check if it's the boot-cpu, set it's hw index in paca now */ 1014 /* Check if it's the boot-cpu, set it's hw index in paca now */
1026 if (get_flat_dt_prop(node, "linux,boot-cpu", NULL) != NULL) { 1015 if (of_get_flat_dt_prop(node, "linux,boot-cpu", NULL)
1027 u32 *prop = get_flat_dt_prop(node, "reg", NULL); 1016 != NULL) {
1017 u32 *prop = of_get_flat_dt_prop(node, "reg", NULL);
1028 set_hard_smp_processor_id(0, prop == NULL ? 0 : *prop); 1018 set_hard_smp_processor_id(0, prop == NULL ? 0 : *prop);
1029 boot_cpuid_phys = get_hard_smp_processor_id(0); 1019 boot_cpuid_phys = get_hard_smp_processor_id(0);
1030 } 1020 }
@@ -1032,14 +1022,14 @@ static int __init early_init_dt_scan_cpus(unsigned long node,
1032 1022
1033#ifdef CONFIG_ALTIVEC 1023#ifdef CONFIG_ALTIVEC
1034 /* Check if we have a VMX and eventually update CPU features */ 1024 /* Check if we have a VMX and eventually update CPU features */
1035 prop = (u32 *)get_flat_dt_prop(node, "ibm,vmx", NULL); 1025 prop = (u32 *)of_get_flat_dt_prop(node, "ibm,vmx", NULL);
1036 if (prop && (*prop) > 0) { 1026 if (prop && (*prop) > 0) {
1037 cur_cpu_spec->cpu_features |= CPU_FTR_ALTIVEC; 1027 cur_cpu_spec->cpu_features |= CPU_FTR_ALTIVEC;
1038 cur_cpu_spec->cpu_user_features |= PPC_FEATURE_HAS_ALTIVEC; 1028 cur_cpu_spec->cpu_user_features |= PPC_FEATURE_HAS_ALTIVEC;
1039 } 1029 }
1040 1030
1041 /* Same goes for Apple's "altivec" property */ 1031 /* Same goes for Apple's "altivec" property */
1042 prop = (u32 *)get_flat_dt_prop(node, "altivec", NULL); 1032 prop = (u32 *)of_get_flat_dt_prop(node, "altivec", NULL);
1043 if (prop) { 1033 if (prop) {
1044 cur_cpu_spec->cpu_features |= CPU_FTR_ALTIVEC; 1034 cur_cpu_spec->cpu_features |= CPU_FTR_ALTIVEC;
1045 cur_cpu_spec->cpu_user_features |= PPC_FEATURE_HAS_ALTIVEC; 1035 cur_cpu_spec->cpu_user_features |= PPC_FEATURE_HAS_ALTIVEC;
@@ -1051,7 +1041,7 @@ static int __init early_init_dt_scan_cpus(unsigned long node,
1051 * this by looking at the size of the ibm,ppc-interrupt-server#s 1041 * this by looking at the size of the ibm,ppc-interrupt-server#s
1052 * property 1042 * property
1053 */ 1043 */
1054 prop = (u32 *)get_flat_dt_prop(node, "ibm,ppc-interrupt-server#s", 1044 prop = (u32 *)of_get_flat_dt_prop(node, "ibm,ppc-interrupt-server#s",
1055 &size); 1045 &size);
1056 cur_cpu_spec->cpu_features &= ~CPU_FTR_SMT; 1046 cur_cpu_spec->cpu_features &= ~CPU_FTR_SMT;
1057 if (prop && ((size / sizeof(u32)) > 1)) 1047 if (prop && ((size / sizeof(u32)) > 1))
@@ -1072,26 +1062,26 @@ static int __init early_init_dt_scan_chosen(unsigned long node,
1072 return 0; 1062 return 0;
1073 1063
1074 /* get platform type */ 1064 /* get platform type */
1075 prop = (u32 *)get_flat_dt_prop(node, "linux,platform", NULL); 1065 prop = (u32 *)of_get_flat_dt_prop(node, "linux,platform", NULL);
1076 if (prop == NULL) 1066 if (prop == NULL)
1077 return 0; 1067 return 0;
1078 systemcfg->platform = *prop; 1068 systemcfg->platform = *prop;
1079 1069
1080 /* check if iommu is forced on or off */ 1070 /* check if iommu is forced on or off */
1081 if (get_flat_dt_prop(node, "linux,iommu-off", NULL) != NULL) 1071 if (of_get_flat_dt_prop(node, "linux,iommu-off", NULL) != NULL)
1082 iommu_is_off = 1; 1072 iommu_is_off = 1;
1083 if (get_flat_dt_prop(node, "linux,iommu-force-on", NULL) != NULL) 1073 if (of_get_flat_dt_prop(node, "linux,iommu-force-on", NULL) != NULL)
1084 iommu_force_on = 1; 1074 iommu_force_on = 1;
1085 1075
1086 prop64 = (u64*)get_flat_dt_prop(node, "linux,memory-limit", NULL); 1076 prop64 = (u64*)of_get_flat_dt_prop(node, "linux,memory-limit", NULL);
1087 if (prop64) 1077 if (prop64)
1088 memory_limit = *prop64; 1078 memory_limit = *prop64;
1089 1079
1090 prop64 = (u64*)get_flat_dt_prop(node, "linux,tce-alloc-start", NULL); 1080 prop64 = (u64*)of_get_flat_dt_prop(node, "linux,tce-alloc-start",NULL);
1091 if (prop64) 1081 if (prop64)
1092 tce_alloc_start = *prop64; 1082 tce_alloc_start = *prop64;
1093 1083
1094 prop64 = (u64*)get_flat_dt_prop(node, "linux,tce-alloc-end", NULL); 1084 prop64 = (u64*)of_get_flat_dt_prop(node, "linux,tce-alloc-end", NULL);
1095 if (prop64) 1085 if (prop64)
1096 tce_alloc_end = *prop64; 1086 tce_alloc_end = *prop64;
1097 1087
@@ -1102,9 +1092,12 @@ static int __init early_init_dt_scan_chosen(unsigned long node,
1102 { 1092 {
1103 u64 *basep, *entryp; 1093 u64 *basep, *entryp;
1104 1094
1105 basep = (u64*)get_flat_dt_prop(node, "linux,rtas-base", NULL); 1095 basep = (u64*)of_get_flat_dt_prop(node,
1106 entryp = (u64*)get_flat_dt_prop(node, "linux,rtas-entry", NULL); 1096 "linux,rtas-base", NULL);
1107 prop = (u32*)get_flat_dt_prop(node, "linux,rtas-size", NULL); 1097 entryp = (u64*)of_get_flat_dt_prop(node,
1098 "linux,rtas-entry", NULL);
1099 prop = (u32*)of_get_flat_dt_prop(node,
1100 "linux,rtas-size", NULL);
1108 if (basep && entryp && prop) { 1101 if (basep && entryp && prop) {
1109 rtas.base = *basep; 1102 rtas.base = *basep;
1110 rtas.entry = *entryp; 1103 rtas.entry = *entryp;
@@ -1125,11 +1118,11 @@ static int __init early_init_dt_scan_root(unsigned long node,
1125 if (depth != 0) 1118 if (depth != 0)
1126 return 0; 1119 return 0;
1127 1120
1128 prop = (u32 *)get_flat_dt_prop(node, "#size-cells", NULL); 1121 prop = (u32 *)of_get_flat_dt_prop(node, "#size-cells", NULL);
1129 dt_root_size_cells = (prop == NULL) ? 1 : *prop; 1122 dt_root_size_cells = (prop == NULL) ? 1 : *prop;
1130 DBG("dt_root_size_cells = %x\n", dt_root_size_cells); 1123 DBG("dt_root_size_cells = %x\n", dt_root_size_cells);
1131 1124
1132 prop = (u32 *)get_flat_dt_prop(node, "#address-cells", NULL); 1125 prop = (u32 *)of_get_flat_dt_prop(node, "#address-cells", NULL);
1133 dt_root_addr_cells = (prop == NULL) ? 2 : *prop; 1126 dt_root_addr_cells = (prop == NULL) ? 2 : *prop;
1134 DBG("dt_root_addr_cells = %x\n", dt_root_addr_cells); 1127 DBG("dt_root_addr_cells = %x\n", dt_root_addr_cells);
1135 1128
@@ -1161,7 +1154,7 @@ static unsigned long __init dt_mem_next_cell(int s, cell_t **cellp)
1161static int __init early_init_dt_scan_memory(unsigned long node, 1154static int __init early_init_dt_scan_memory(unsigned long node,
1162 const char *uname, int depth, void *data) 1155 const char *uname, int depth, void *data)
1163{ 1156{
1164 char *type = get_flat_dt_prop(node, "device_type", NULL); 1157 char *type = of_get_flat_dt_prop(node, "device_type", NULL);
1165 cell_t *reg, *endp; 1158 cell_t *reg, *endp;
1166 unsigned long l; 1159 unsigned long l;
1167 1160
@@ -1169,7 +1162,7 @@ static int __init early_init_dt_scan_memory(unsigned long node,
1169 if (type == NULL || strcmp(type, "memory") != 0) 1162 if (type == NULL || strcmp(type, "memory") != 0)
1170 return 0; 1163 return 0;
1171 1164
1172 reg = (cell_t *)get_flat_dt_prop(node, "reg", &l); 1165 reg = (cell_t *)of_get_flat_dt_prop(node, "reg", &l);
1173 if (reg == NULL) 1166 if (reg == NULL)
1174 return 0; 1167 return 0;
1175 1168
@@ -1225,19 +1218,16 @@ void __init early_init_devtree(void *params)
1225 /* Setup flat device-tree pointer */ 1218 /* Setup flat device-tree pointer */
1226 initial_boot_params = params; 1219 initial_boot_params = params;
1227 1220
1228 /* By default, hash size is not set */
1229 ppc64_pft_size = 0;
1230
1231 /* Retreive various informations from the /chosen node of the 1221 /* Retreive various informations from the /chosen node of the
1232 * device-tree, including the platform type, initrd location and 1222 * device-tree, including the platform type, initrd location and
1233 * size, TCE reserve, and more ... 1223 * size, TCE reserve, and more ...
1234 */ 1224 */
1235 scan_flat_dt(early_init_dt_scan_chosen, NULL); 1225 of_scan_flat_dt(early_init_dt_scan_chosen, NULL);
1236 1226
1237 /* Scan memory nodes and rebuild LMBs */ 1227 /* Scan memory nodes and rebuild LMBs */
1238 lmb_init(); 1228 lmb_init();
1239 scan_flat_dt(early_init_dt_scan_root, NULL); 1229 of_scan_flat_dt(early_init_dt_scan_root, NULL);
1240 scan_flat_dt(early_init_dt_scan_memory, NULL); 1230 of_scan_flat_dt(early_init_dt_scan_memory, NULL);
1241 lmb_enforce_memory_limit(memory_limit); 1231 lmb_enforce_memory_limit(memory_limit);
1242 lmb_analyze(); 1232 lmb_analyze();
1243 systemcfg->physicalMemorySize = lmb_phys_mem_size(); 1233 systemcfg->physicalMemorySize = lmb_phys_mem_size();
@@ -1253,26 +1243,8 @@ void __init early_init_devtree(void *params)
1253 /* Retreive hash table size from flattened tree plus other 1243 /* Retreive hash table size from flattened tree plus other
1254 * CPU related informations (altivec support, boot CPU ID, ...) 1244 * CPU related informations (altivec support, boot CPU ID, ...)
1255 */ 1245 */
1256 scan_flat_dt(early_init_dt_scan_cpus, NULL); 1246 of_scan_flat_dt(early_init_dt_scan_cpus, NULL);
1257
1258 /* If hash size wasn't obtained above, we calculate it now based on
1259 * the total RAM size
1260 */
1261 if (ppc64_pft_size == 0) {
1262 unsigned long rnd_mem_size, pteg_count;
1263
1264 /* round mem_size up to next power of 2 */
1265 rnd_mem_size = 1UL << __ilog2(systemcfg->physicalMemorySize);
1266 if (rnd_mem_size < systemcfg->physicalMemorySize)
1267 rnd_mem_size <<= 1;
1268
1269 /* # pages / 2 */
1270 pteg_count = max(rnd_mem_size >> (12 + 1), 1UL << 11);
1271
1272 ppc64_pft_size = __ilog2(pteg_count << 7);
1273 }
1274 1247
1275 DBG("Hash pftSize: %x\n", (int)ppc64_pft_size);
1276 DBG(" <- early_init_devtree()\n"); 1248 DBG(" <- early_init_devtree()\n");
1277} 1249}
1278 1250
diff --git a/arch/ppc64/kernel/prom_init.c b/arch/ppc64/kernel/prom_init.c
index 69924ba4d7d9..a4bbca6dbb8b 100644
--- a/arch/ppc64/kernel/prom_init.c
+++ b/arch/ppc64/kernel/prom_init.c
@@ -1939,9 +1939,9 @@ unsigned long __init prom_init(unsigned long r3, unsigned long r4, unsigned long
1939 prom_send_capabilities(); 1939 prom_send_capabilities();
1940 1940
1941 /* 1941 /*
1942 * On pSeries and BPA, copy the CPU hold code 1942 * On pSeries and Cell, copy the CPU hold code
1943 */ 1943 */
1944 if (RELOC(of_platform) & (PLATFORM_PSERIES | PLATFORM_BPA)) 1944 if (RELOC(of_platform) & (PLATFORM_PSERIES | PLATFORM_CELL))
1945 copy_and_flush(0, KERNELBASE - offset, 0x100, 0); 1945 copy_and_flush(0, KERNELBASE - offset, 0x100, 0);
1946 1946
1947 /* 1947 /*
diff --git a/arch/ppc64/kernel/rtas-proc.c b/arch/ppc64/kernel/rtas-proc.c
deleted file mode 100644
index 5bdd5b079d96..000000000000
--- a/arch/ppc64/kernel/rtas-proc.c
+++ /dev/null
@@ -1,808 +0,0 @@
1/*
2 * arch/ppc64/kernel/rtas-proc.c
3 * Copyright (C) 2000 Tilmann Bitterberg
4 * (tilmann@bitterberg.de)
5 *
6 * RTAS (Runtime Abstraction Services) stuff
7 * Intention is to provide a clean user interface
8 * to use the RTAS.
9 *
10 * TODO:
11 * Split off a header file and maybe move it to a different
12 * location. Write Documentation on what the /proc/rtas/ entries
13 * actually do.
14 */
15
16#include <linux/errno.h>
17#include <linux/sched.h>
18#include <linux/proc_fs.h>
19#include <linux/stat.h>
20#include <linux/ctype.h>
21#include <linux/time.h>
22#include <linux/string.h>
23#include <linux/init.h>
24#include <linux/seq_file.h>
25#include <linux/bitops.h>
26#include <linux/rtc.h>
27
28#include <asm/uaccess.h>
29#include <asm/processor.h>
30#include <asm/io.h>
31#include <asm/prom.h>
32#include <asm/rtas.h>
33#include <asm/machdep.h> /* for ppc_md */
34#include <asm/time.h>
35#include <asm/systemcfg.h>
36
37/* Token for Sensors */
38#define KEY_SWITCH 0x0001
39#define ENCLOSURE_SWITCH 0x0002
40#define THERMAL_SENSOR 0x0003
41#define LID_STATUS 0x0004
42#define POWER_SOURCE 0x0005
43#define BATTERY_VOLTAGE 0x0006
44#define BATTERY_REMAINING 0x0007
45#define BATTERY_PERCENTAGE 0x0008
46#define EPOW_SENSOR 0x0009
47#define BATTERY_CYCLESTATE 0x000a
48#define BATTERY_CHARGING 0x000b
49
50/* IBM specific sensors */
51#define IBM_SURVEILLANCE 0x2328 /* 9000 */
52#define IBM_FANRPM 0x2329 /* 9001 */
53#define IBM_VOLTAGE 0x232a /* 9002 */
54#define IBM_DRCONNECTOR 0x232b /* 9003 */
55#define IBM_POWERSUPPLY 0x232c /* 9004 */
56
57/* Status return values */
58#define SENSOR_CRITICAL_HIGH 13
59#define SENSOR_WARNING_HIGH 12
60#define SENSOR_NORMAL 11
61#define SENSOR_WARNING_LOW 10
62#define SENSOR_CRITICAL_LOW 9
63#define SENSOR_SUCCESS 0
64#define SENSOR_HW_ERROR -1
65#define SENSOR_BUSY -2
66#define SENSOR_NOT_EXIST -3
67#define SENSOR_DR_ENTITY -9000
68
69/* Location Codes */
70#define LOC_SCSI_DEV_ADDR 'A'
71#define LOC_SCSI_DEV_LOC 'B'
72#define LOC_CPU 'C'
73#define LOC_DISKETTE 'D'
74#define LOC_ETHERNET 'E'
75#define LOC_FAN 'F'
76#define LOC_GRAPHICS 'G'
77/* reserved / not used 'H' */
78#define LOC_IO_ADAPTER 'I'
79/* reserved / not used 'J' */
80#define LOC_KEYBOARD 'K'
81#define LOC_LCD 'L'
82#define LOC_MEMORY 'M'
83#define LOC_NV_MEMORY 'N'
84#define LOC_MOUSE 'O'
85#define LOC_PLANAR 'P'
86#define LOC_OTHER_IO 'Q'
87#define LOC_PARALLEL 'R'
88#define LOC_SERIAL 'S'
89#define LOC_DEAD_RING 'T'
90#define LOC_RACKMOUNTED 'U' /* for _u_nit is rack mounted */
91#define LOC_VOLTAGE 'V'
92#define LOC_SWITCH_ADAPTER 'W'
93#define LOC_OTHER 'X'
94#define LOC_FIRMWARE 'Y'
95#define LOC_SCSI 'Z'
96
97/* Tokens for indicators */
98#define TONE_FREQUENCY 0x0001 /* 0 - 1000 (HZ)*/
99#define TONE_VOLUME 0x0002 /* 0 - 100 (%) */
100#define SYSTEM_POWER_STATE 0x0003
101#define WARNING_LIGHT 0x0004
102#define DISK_ACTIVITY_LIGHT 0x0005
103#define HEX_DISPLAY_UNIT 0x0006
104#define BATTERY_WARNING_TIME 0x0007
105#define CONDITION_CYCLE_REQUEST 0x0008
106#define SURVEILLANCE_INDICATOR 0x2328 /* 9000 */
107#define DR_ACTION 0x2329 /* 9001 */
108#define DR_INDICATOR 0x232a /* 9002 */
109/* 9003 - 9004: Vendor specific */
110/* 9006 - 9999: Vendor specific */
111
112/* other */
113#define MAX_SENSORS 17 /* I only know of 17 sensors */
114#define MAX_LINELENGTH 256
115#define SENSOR_PREFIX "ibm,sensor-"
116#define cel_to_fahr(x) ((x*9/5)+32)
117
118
119/* Globals */
120static struct rtas_sensors sensors;
121static struct device_node *rtas_node = NULL;
122static unsigned long power_on_time = 0; /* Save the time the user set */
123static char progress_led[MAX_LINELENGTH];
124
125static unsigned long rtas_tone_frequency = 1000;
126static unsigned long rtas_tone_volume = 0;
127
128/* ****************STRUCTS******************************************* */
129struct individual_sensor {
130 unsigned int token;
131 unsigned int quant;
132};
133
134struct rtas_sensors {
135 struct individual_sensor sensor[MAX_SENSORS];
136 unsigned int quant;
137};
138
139/* ****************************************************************** */
140/* Declarations */
141static int ppc_rtas_sensors_show(struct seq_file *m, void *v);
142static int ppc_rtas_clock_show(struct seq_file *m, void *v);
143static ssize_t ppc_rtas_clock_write(struct file *file,
144 const char __user *buf, size_t count, loff_t *ppos);
145static int ppc_rtas_progress_show(struct seq_file *m, void *v);
146static ssize_t ppc_rtas_progress_write(struct file *file,
147 const char __user *buf, size_t count, loff_t *ppos);
148static int ppc_rtas_poweron_show(struct seq_file *m, void *v);
149static ssize_t ppc_rtas_poweron_write(struct file *file,
150 const char __user *buf, size_t count, loff_t *ppos);
151
152static ssize_t ppc_rtas_tone_freq_write(struct file *file,
153 const char __user *buf, size_t count, loff_t *ppos);
154static int ppc_rtas_tone_freq_show(struct seq_file *m, void *v);
155static ssize_t ppc_rtas_tone_volume_write(struct file *file,
156 const char __user *buf, size_t count, loff_t *ppos);
157static int ppc_rtas_tone_volume_show(struct seq_file *m, void *v);
158static int ppc_rtas_rmo_buf_show(struct seq_file *m, void *v);
159
160static int sensors_open(struct inode *inode, struct file *file)
161{
162 return single_open(file, ppc_rtas_sensors_show, NULL);
163}
164
165struct file_operations ppc_rtas_sensors_operations = {
166 .open = sensors_open,
167 .read = seq_read,
168 .llseek = seq_lseek,
169 .release = single_release,
170};
171
172static int poweron_open(struct inode *inode, struct file *file)
173{
174 return single_open(file, ppc_rtas_poweron_show, NULL);
175}
176
177struct file_operations ppc_rtas_poweron_operations = {
178 .open = poweron_open,
179 .read = seq_read,
180 .llseek = seq_lseek,
181 .write = ppc_rtas_poweron_write,
182 .release = single_release,
183};
184
185static int progress_open(struct inode *inode, struct file *file)
186{
187 return single_open(file, ppc_rtas_progress_show, NULL);
188}
189
190struct file_operations ppc_rtas_progress_operations = {
191 .open = progress_open,
192 .read = seq_read,
193 .llseek = seq_lseek,
194 .write = ppc_rtas_progress_write,
195 .release = single_release,
196};
197
198static int clock_open(struct inode *inode, struct file *file)
199{
200 return single_open(file, ppc_rtas_clock_show, NULL);
201}
202
203struct file_operations ppc_rtas_clock_operations = {
204 .open = clock_open,
205 .read = seq_read,
206 .llseek = seq_lseek,
207 .write = ppc_rtas_clock_write,
208 .release = single_release,
209};
210
211static int tone_freq_open(struct inode *inode, struct file *file)
212{
213 return single_open(file, ppc_rtas_tone_freq_show, NULL);
214}
215
216struct file_operations ppc_rtas_tone_freq_operations = {
217 .open = tone_freq_open,
218 .read = seq_read,
219 .llseek = seq_lseek,
220 .write = ppc_rtas_tone_freq_write,
221 .release = single_release,
222};
223
224static int tone_volume_open(struct inode *inode, struct file *file)
225{
226 return single_open(file, ppc_rtas_tone_volume_show, NULL);
227}
228
229struct file_operations ppc_rtas_tone_volume_operations = {
230 .open = tone_volume_open,
231 .read = seq_read,
232 .llseek = seq_lseek,
233 .write = ppc_rtas_tone_volume_write,
234 .release = single_release,
235};
236
237static int rmo_buf_open(struct inode *inode, struct file *file)
238{
239 return single_open(file, ppc_rtas_rmo_buf_show, NULL);
240}
241
242struct file_operations ppc_rtas_rmo_buf_ops = {
243 .open = rmo_buf_open,
244 .read = seq_read,
245 .llseek = seq_lseek,
246 .release = single_release,
247};
248
249static int ppc_rtas_find_all_sensors(void);
250static void ppc_rtas_process_sensor(struct seq_file *m,
251 struct individual_sensor *s, int state, int error, char *loc);
252static char *ppc_rtas_process_error(int error);
253static void get_location_code(struct seq_file *m,
254 struct individual_sensor *s, char *loc);
255static void check_location_string(struct seq_file *m, char *c);
256static void check_location(struct seq_file *m, char *c);
257
258static int __init proc_rtas_init(void)
259{
260 struct proc_dir_entry *entry;
261
262 if (!(systemcfg->platform & PLATFORM_PSERIES))
263 return 1;
264
265 rtas_node = of_find_node_by_name(NULL, "rtas");
266 if (rtas_node == NULL)
267 return 1;
268
269 entry = create_proc_entry("ppc64/rtas/progress", S_IRUGO|S_IWUSR, NULL);
270 if (entry)
271 entry->proc_fops = &ppc_rtas_progress_operations;
272
273 entry = create_proc_entry("ppc64/rtas/clock", S_IRUGO|S_IWUSR, NULL);
274 if (entry)
275 entry->proc_fops = &ppc_rtas_clock_operations;
276
277 entry = create_proc_entry("ppc64/rtas/poweron", S_IWUSR|S_IRUGO, NULL);
278 if (entry)
279 entry->proc_fops = &ppc_rtas_poweron_operations;
280
281 entry = create_proc_entry("ppc64/rtas/sensors", S_IRUGO, NULL);
282 if (entry)
283 entry->proc_fops = &ppc_rtas_sensors_operations;
284
285 entry = create_proc_entry("ppc64/rtas/frequency", S_IWUSR|S_IRUGO,
286 NULL);
287 if (entry)
288 entry->proc_fops = &ppc_rtas_tone_freq_operations;
289
290 entry = create_proc_entry("ppc64/rtas/volume", S_IWUSR|S_IRUGO, NULL);
291 if (entry)
292 entry->proc_fops = &ppc_rtas_tone_volume_operations;
293
294 entry = create_proc_entry("ppc64/rtas/rmo_buffer", S_IRUSR, NULL);
295 if (entry)
296 entry->proc_fops = &ppc_rtas_rmo_buf_ops;
297
298 return 0;
299}
300
301__initcall(proc_rtas_init);
302
303static int parse_number(const char __user *p, size_t count, unsigned long *val)
304{
305 char buf[40];
306 char *end;
307
308 if (count > 39)
309 return -EINVAL;
310
311 if (copy_from_user(buf, p, count))
312 return -EFAULT;
313
314 buf[count] = 0;
315
316 *val = simple_strtoul(buf, &end, 10);
317 if (*end && *end != '\n')
318 return -EINVAL;
319
320 return 0;
321}
322
323/* ****************************************************************** */
324/* POWER-ON-TIME */
325/* ****************************************************************** */
326static ssize_t ppc_rtas_poweron_write(struct file *file,
327 const char __user *buf, size_t count, loff_t *ppos)
328{
329 struct rtc_time tm;
330 unsigned long nowtime;
331 int error = parse_number(buf, count, &nowtime);
332 if (error)
333 return error;
334
335 power_on_time = nowtime; /* save the time */
336
337 to_tm(nowtime, &tm);
338
339 error = rtas_call(rtas_token("set-time-for-power-on"), 7, 1, NULL,
340 tm.tm_year, tm.tm_mon, tm.tm_mday,
341 tm.tm_hour, tm.tm_min, tm.tm_sec, 0 /* nano */);
342 if (error)
343 printk(KERN_WARNING "error: setting poweron time returned: %s\n",
344 ppc_rtas_process_error(error));
345 return count;
346}
347/* ****************************************************************** */
348static int ppc_rtas_poweron_show(struct seq_file *m, void *v)
349{
350 if (power_on_time == 0)
351 seq_printf(m, "Power on time not set\n");
352 else
353 seq_printf(m, "%lu\n",power_on_time);
354 return 0;
355}
356
357/* ****************************************************************** */
358/* PROGRESS */
359/* ****************************************************************** */
360static ssize_t ppc_rtas_progress_write(struct file *file,
361 const char __user *buf, size_t count, loff_t *ppos)
362{
363 unsigned long hex;
364
365 if (count >= MAX_LINELENGTH)
366 count = MAX_LINELENGTH -1;
367 if (copy_from_user(progress_led, buf, count)) { /* save the string */
368 return -EFAULT;
369 }
370 progress_led[count] = 0;
371
372 /* Lets see if the user passed hexdigits */
373 hex = simple_strtoul(progress_led, NULL, 10);
374
375 rtas_progress ((char *)progress_led, hex);
376 return count;
377
378 /* clear the line */
379 /* rtas_progress(" ", 0xffff);*/
380}
381/* ****************************************************************** */
382static int ppc_rtas_progress_show(struct seq_file *m, void *v)
383{
384 if (progress_led)
385 seq_printf(m, "%s\n", progress_led);
386 return 0;
387}
388
389/* ****************************************************************** */
390/* CLOCK */
391/* ****************************************************************** */
392static ssize_t ppc_rtas_clock_write(struct file *file,
393 const char __user *buf, size_t count, loff_t *ppos)
394{
395 struct rtc_time tm;
396 unsigned long nowtime;
397 int error = parse_number(buf, count, &nowtime);
398 if (error)
399 return error;
400
401 to_tm(nowtime, &tm);
402 error = rtas_call(rtas_token("set-time-of-day"), 7, 1, NULL,
403 tm.tm_year, tm.tm_mon, tm.tm_mday,
404 tm.tm_hour, tm.tm_min, tm.tm_sec, 0);
405 if (error)
406 printk(KERN_WARNING "error: setting the clock returned: %s\n",
407 ppc_rtas_process_error(error));
408 return count;
409}
410/* ****************************************************************** */
411static int ppc_rtas_clock_show(struct seq_file *m, void *v)
412{
413 int ret[8];
414 int error = rtas_call(rtas_token("get-time-of-day"), 0, 8, ret);
415
416 if (error) {
417 printk(KERN_WARNING "error: reading the clock returned: %s\n",
418 ppc_rtas_process_error(error));
419 seq_printf(m, "0");
420 } else {
421 unsigned int year, mon, day, hour, min, sec;
422 year = ret[0]; mon = ret[1]; day = ret[2];
423 hour = ret[3]; min = ret[4]; sec = ret[5];
424 seq_printf(m, "%lu\n",
425 mktime(year, mon, day, hour, min, sec));
426 }
427 return 0;
428}
429
430/* ****************************************************************** */
431/* SENSOR STUFF */
432/* ****************************************************************** */
433static int ppc_rtas_sensors_show(struct seq_file *m, void *v)
434{
435 int i,j;
436 int state, error;
437 int get_sensor_state = rtas_token("get-sensor-state");
438
439 seq_printf(m, "RTAS (RunTime Abstraction Services) Sensor Information\n");
440 seq_printf(m, "Sensor\t\tValue\t\tCondition\tLocation\n");
441 seq_printf(m, "********************************************************\n");
442
443 if (ppc_rtas_find_all_sensors() != 0) {
444 seq_printf(m, "\nNo sensors are available\n");
445 return 0;
446 }
447
448 for (i=0; i<sensors.quant; i++) {
449 struct individual_sensor *p = &sensors.sensor[i];
450 char rstr[64];
451 char *loc;
452 int llen, offs;
453
454 sprintf (rstr, SENSOR_PREFIX"%04d", p->token);
455 loc = (char *) get_property(rtas_node, rstr, &llen);
456
457 /* A sensor may have multiple instances */
458 for (j = 0, offs = 0; j <= p->quant; j++) {
459 error = rtas_call(get_sensor_state, 2, 2, &state,
460 p->token, j);
461
462 ppc_rtas_process_sensor(m, p, state, error, loc);
463 seq_putc(m, '\n');
464 if (loc) {
465 offs += strlen(loc) + 1;
466 loc += strlen(loc) + 1;
467 if (offs >= llen)
468 loc = NULL;
469 }
470 }
471 }
472 return 0;
473}
474
475/* ****************************************************************** */
476
477static int ppc_rtas_find_all_sensors(void)
478{
479 unsigned int *utmp;
480 int len, i;
481
482 utmp = (unsigned int *) get_property(rtas_node, "rtas-sensors", &len);
483 if (utmp == NULL) {
484 printk (KERN_ERR "error: could not get rtas-sensors\n");
485 return 1;
486 }
487
488 sensors.quant = len / 8; /* int + int */
489
490 for (i=0; i<sensors.quant; i++) {
491 sensors.sensor[i].token = *utmp++;
492 sensors.sensor[i].quant = *utmp++;
493 }
494 return 0;
495}
496
497/* ****************************************************************** */
498/*
499 * Builds a string of what rtas returned
500 */
501static char *ppc_rtas_process_error(int error)
502{
503 switch (error) {
504 case SENSOR_CRITICAL_HIGH:
505 return "(critical high)";
506 case SENSOR_WARNING_HIGH:
507 return "(warning high)";
508 case SENSOR_NORMAL:
509 return "(normal)";
510 case SENSOR_WARNING_LOW:
511 return "(warning low)";
512 case SENSOR_CRITICAL_LOW:
513 return "(critical low)";
514 case SENSOR_SUCCESS:
515 return "(read ok)";
516 case SENSOR_HW_ERROR:
517 return "(hardware error)";
518 case SENSOR_BUSY:
519 return "(busy)";
520 case SENSOR_NOT_EXIST:
521 return "(non existent)";
522 case SENSOR_DR_ENTITY:
523 return "(dr entity removed)";
524 default:
525 return "(UNKNOWN)";
526 }
527}
528
529/* ****************************************************************** */
530/*
531 * Builds a string out of what the sensor said
532 */
533
534static void ppc_rtas_process_sensor(struct seq_file *m,
535 struct individual_sensor *s, int state, int error, char *loc)
536{
537 /* Defined return vales */
538 const char * key_switch[] = { "Off\t", "Normal\t", "Secure\t",
539 "Maintenance" };
540 const char * enclosure_switch[] = { "Closed", "Open" };
541 const char * lid_status[] = { " ", "Open", "Closed" };
542 const char * power_source[] = { "AC\t", "Battery",
543 "AC & Battery" };
544 const char * battery_remaining[] = { "Very Low", "Low", "Mid", "High" };
545 const char * epow_sensor[] = {
546 "EPOW Reset", "Cooling warning", "Power warning",
547 "System shutdown", "System halt", "EPOW main enclosure",
548 "EPOW power off" };
549 const char * battery_cyclestate[] = { "None", "In progress",
550 "Requested" };
551 const char * battery_charging[] = { "Charging", "Discharching",
552 "No current flow" };
553 const char * ibm_drconnector[] = { "Empty", "Present", "Unusable",
554 "Exchange" };
555
556 int have_strings = 0;
557 int num_states = 0;
558 int temperature = 0;
559 int unknown = 0;
560
561 /* What kind of sensor do we have here? */
562
563 switch (s->token) {
564 case KEY_SWITCH:
565 seq_printf(m, "Key switch:\t");
566 num_states = sizeof(key_switch) / sizeof(char *);
567 if (state < num_states) {
568 seq_printf(m, "%s\t", key_switch[state]);
569 have_strings = 1;
570 }
571 break;
572 case ENCLOSURE_SWITCH:
573 seq_printf(m, "Enclosure switch:\t");
574 num_states = sizeof(enclosure_switch) / sizeof(char *);
575 if (state < num_states) {
576 seq_printf(m, "%s\t",
577 enclosure_switch[state]);
578 have_strings = 1;
579 }
580 break;
581 case THERMAL_SENSOR:
582 seq_printf(m, "Temp. (C/F):\t");
583 temperature = 1;
584 break;
585 case LID_STATUS:
586 seq_printf(m, "Lid status:\t");
587 num_states = sizeof(lid_status) / sizeof(char *);
588 if (state < num_states) {
589 seq_printf(m, "%s\t", lid_status[state]);
590 have_strings = 1;
591 }
592 break;
593 case POWER_SOURCE:
594 seq_printf(m, "Power source:\t");
595 num_states = sizeof(power_source) / sizeof(char *);
596 if (state < num_states) {
597 seq_printf(m, "%s\t",
598 power_source[state]);
599 have_strings = 1;
600 }
601 break;
602 case BATTERY_VOLTAGE:
603 seq_printf(m, "Battery voltage:\t");
604 break;
605 case BATTERY_REMAINING:
606 seq_printf(m, "Battery remaining:\t");
607 num_states = sizeof(battery_remaining) / sizeof(char *);
608 if (state < num_states)
609 {
610 seq_printf(m, "%s\t",
611 battery_remaining[state]);
612 have_strings = 1;
613 }
614 break;
615 case BATTERY_PERCENTAGE:
616 seq_printf(m, "Battery percentage:\t");
617 break;
618 case EPOW_SENSOR:
619 seq_printf(m, "EPOW Sensor:\t");
620 num_states = sizeof(epow_sensor) / sizeof(char *);
621 if (state < num_states) {
622 seq_printf(m, "%s\t", epow_sensor[state]);
623 have_strings = 1;
624 }
625 break;
626 case BATTERY_CYCLESTATE:
627 seq_printf(m, "Battery cyclestate:\t");
628 num_states = sizeof(battery_cyclestate) /
629 sizeof(char *);
630 if (state < num_states) {
631 seq_printf(m, "%s\t",
632 battery_cyclestate[state]);
633 have_strings = 1;
634 }
635 break;
636 case BATTERY_CHARGING:
637 seq_printf(m, "Battery Charging:\t");
638 num_states = sizeof(battery_charging) / sizeof(char *);
639 if (state < num_states) {
640 seq_printf(m, "%s\t",
641 battery_charging[state]);
642 have_strings = 1;
643 }
644 break;
645 case IBM_SURVEILLANCE:
646 seq_printf(m, "Surveillance:\t");
647 break;
648 case IBM_FANRPM:
649 seq_printf(m, "Fan (rpm):\t");
650 break;
651 case IBM_VOLTAGE:
652 seq_printf(m, "Voltage (mv):\t");
653 break;
654 case IBM_DRCONNECTOR:
655 seq_printf(m, "DR connector:\t");
656 num_states = sizeof(ibm_drconnector) / sizeof(char *);
657 if (state < num_states) {
658 seq_printf(m, "%s\t",
659 ibm_drconnector[state]);
660 have_strings = 1;
661 }
662 break;
663 case IBM_POWERSUPPLY:
664 seq_printf(m, "Powersupply:\t");
665 break;
666 default:
667 seq_printf(m, "Unknown sensor (type %d), ignoring it\n",
668 s->token);
669 unknown = 1;
670 have_strings = 1;
671 break;
672 }
673 if (have_strings == 0) {
674 if (temperature) {
675 seq_printf(m, "%4d /%4d\t", state, cel_to_fahr(state));
676 } else
677 seq_printf(m, "%10d\t", state);
678 }
679 if (unknown == 0) {
680 seq_printf(m, "%s\t", ppc_rtas_process_error(error));
681 get_location_code(m, s, loc);
682 }
683}
684
685/* ****************************************************************** */
686
687static void check_location(struct seq_file *m, char *c)
688{
689 switch (c[0]) {
690 case LOC_PLANAR:
691 seq_printf(m, "Planar #%c", c[1]);
692 break;
693 case LOC_CPU:
694 seq_printf(m, "CPU #%c", c[1]);
695 break;
696 case LOC_FAN:
697 seq_printf(m, "Fan #%c", c[1]);
698 break;
699 case LOC_RACKMOUNTED:
700 seq_printf(m, "Rack #%c", c[1]);
701 break;
702 case LOC_VOLTAGE:
703 seq_printf(m, "Voltage #%c", c[1]);
704 break;
705 case LOC_LCD:
706 seq_printf(m, "LCD #%c", c[1]);
707 break;
708 case '.':
709 seq_printf(m, "- %c", c[1]);
710 break;
711 default:
712 seq_printf(m, "Unknown location");
713 break;
714 }
715}
716
717
718/* ****************************************************************** */
719/*
720 * Format:
721 * ${LETTER}${NUMBER}[[-/]${LETTER}${NUMBER} [ ... ] ]
722 * the '.' may be an abbrevation
723 */
724static void check_location_string(struct seq_file *m, char *c)
725{
726 while (*c) {
727 if (isalpha(*c) || *c == '.')
728 check_location(m, c);
729 else if (*c == '/' || *c == '-')
730 seq_printf(m, " at ");
731 c++;
732 }
733}
734
735
736/* ****************************************************************** */
737
738static void get_location_code(struct seq_file *m, struct individual_sensor *s, char *loc)
739{
740 if (!loc || !*loc) {
741 seq_printf(m, "---");/* does not have a location */
742 } else {
743 check_location_string(m, loc);
744 }
745 seq_putc(m, ' ');
746}
747/* ****************************************************************** */
748/* INDICATORS - Tone Frequency */
749/* ****************************************************************** */
750static ssize_t ppc_rtas_tone_freq_write(struct file *file,
751 const char __user *buf, size_t count, loff_t *ppos)
752{
753 unsigned long freq;
754 int error = parse_number(buf, count, &freq);
755 if (error)
756 return error;
757
758 rtas_tone_frequency = freq; /* save it for later */
759 error = rtas_call(rtas_token("set-indicator"), 3, 1, NULL,
760 TONE_FREQUENCY, 0, freq);
761 if (error)
762 printk(KERN_WARNING "error: setting tone frequency returned: %s\n",
763 ppc_rtas_process_error(error));
764 return count;
765}
766/* ****************************************************************** */
767static int ppc_rtas_tone_freq_show(struct seq_file *m, void *v)
768{
769 seq_printf(m, "%lu\n", rtas_tone_frequency);
770 return 0;
771}
772/* ****************************************************************** */
773/* INDICATORS - Tone Volume */
774/* ****************************************************************** */
775static ssize_t ppc_rtas_tone_volume_write(struct file *file,
776 const char __user *buf, size_t count, loff_t *ppos)
777{
778 unsigned long volume;
779 int error = parse_number(buf, count, &volume);
780 if (error)
781 return error;
782
783 if (volume > 100)
784 volume = 100;
785
786 rtas_tone_volume = volume; /* save it for later */
787 error = rtas_call(rtas_token("set-indicator"), 3, 1, NULL,
788 TONE_VOLUME, 0, volume);
789 if (error)
790 printk(KERN_WARNING "error: setting tone volume returned: %s\n",
791 ppc_rtas_process_error(error));
792 return count;
793}
794/* ****************************************************************** */
795static int ppc_rtas_tone_volume_show(struct seq_file *m, void *v)
796{
797 seq_printf(m, "%lu\n", rtas_tone_volume);
798 return 0;
799}
800
801#define RMO_READ_BUF_MAX 30
802
803/* RTAS Userspace access */
804static int ppc_rtas_rmo_buf_show(struct seq_file *m, void *v)
805{
806 seq_printf(m, "%016lx %x\n", rtas_rmo_buf, RTAS_RMOBUF_MAX);
807 return 0;
808}
diff --git a/arch/ppc64/kernel/rtas_flash.c b/arch/ppc64/kernel/rtas_flash.c
deleted file mode 100644
index 923e2e201a70..000000000000
--- a/arch/ppc64/kernel/rtas_flash.c
+++ /dev/null
@@ -1,725 +0,0 @@
1/*
2 * c 2001 PPC 64 Team, IBM Corp
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License
6 * as published by the Free Software Foundation; either version
7 * 2 of the License, or (at your option) any later version.
8 *
9 * /proc/ppc64/rtas/firmware_flash interface
10 *
11 * This file implements a firmware_flash interface to pump a firmware
12 * image into the kernel. At reboot time rtas_restart() will see the
13 * firmware image and flash it as it reboots (see rtas.c).
14 */
15
16#include <linux/module.h>
17#include <linux/init.h>
18#include <linux/proc_fs.h>
19#include <asm/delay.h>
20#include <asm/uaccess.h>
21#include <asm/rtas.h>
22
23#define MODULE_VERS "1.0"
24#define MODULE_NAME "rtas_flash"
25
26#define FIRMWARE_FLASH_NAME "firmware_flash"
27#define FIRMWARE_UPDATE_NAME "firmware_update"
28#define MANAGE_FLASH_NAME "manage_flash"
29#define VALIDATE_FLASH_NAME "validate_flash"
30
31/* General RTAS Status Codes */
32#define RTAS_RC_SUCCESS 0
33#define RTAS_RC_HW_ERR -1
34#define RTAS_RC_BUSY -2
35
36/* Flash image status values */
37#define FLASH_AUTH -9002 /* RTAS Not Service Authority Partition */
38#define FLASH_NO_OP -1099 /* No operation initiated by user */
39#define FLASH_IMG_SHORT -1005 /* Flash image shorter than expected */
40#define FLASH_IMG_BAD_LEN -1004 /* Bad length value in flash list block */
41#define FLASH_IMG_NULL_DATA -1003 /* Bad data value in flash list block */
42#define FLASH_IMG_READY 0 /* Firmware img ready for flash on reboot */
43
44/* Manage image status values */
45#define MANAGE_AUTH -9002 /* RTAS Not Service Authority Partition */
46#define MANAGE_ACTIVE_ERR -9001 /* RTAS Cannot Overwrite Active Img */
47#define MANAGE_NO_OP -1099 /* No operation initiated by user */
48#define MANAGE_PARAM_ERR -3 /* RTAS Parameter Error */
49#define MANAGE_HW_ERR -1 /* RTAS Hardware Error */
50
51/* Validate image status values */
52#define VALIDATE_AUTH -9002 /* RTAS Not Service Authority Partition */
53#define VALIDATE_NO_OP -1099 /* No operation initiated by the user */
54#define VALIDATE_INCOMPLETE -1002 /* User copied < VALIDATE_BUF_SIZE */
55#define VALIDATE_READY -1001 /* Firmware image ready for validation */
56#define VALIDATE_PARAM_ERR -3 /* RTAS Parameter Error */
57#define VALIDATE_HW_ERR -1 /* RTAS Hardware Error */
58#define VALIDATE_TMP_UPDATE 0 /* Validate Return Status */
59#define VALIDATE_FLASH_AUTH 1 /* Validate Return Status */
60#define VALIDATE_INVALID_IMG 2 /* Validate Return Status */
61#define VALIDATE_CUR_UNKNOWN 3 /* Validate Return Status */
62#define VALIDATE_TMP_COMMIT_DL 4 /* Validate Return Status */
63#define VALIDATE_TMP_COMMIT 5 /* Validate Return Status */
64#define VALIDATE_TMP_UPDATE_DL 6 /* Validate Return Status */
65
66/* ibm,manage-flash-image operation tokens */
67#define RTAS_REJECT_TMP_IMG 0
68#define RTAS_COMMIT_TMP_IMG 1
69
70/* Array sizes */
71#define VALIDATE_BUF_SIZE 4096
72#define RTAS_MSG_MAXLEN 64
73
74/* Local copy of the flash block list.
75 * We only allow one open of the flash proc file and create this
76 * list as we go. This list will be put in the kernel's
77 * rtas_firmware_flash_list global var once it is fully read.
78 *
79 * For convenience as we build the list we use virtual addrs,
80 * we do not fill in the version number, and the length field
81 * is treated as the number of entries currently in the block
82 * (i.e. not a byte count). This is all fixed on release.
83 */
84
85/* Status int must be first member of struct */
86struct rtas_update_flash_t
87{
88 int status; /* Flash update status */
89 struct flash_block_list *flist; /* Local copy of flash block list */
90};
91
92/* Status int must be first member of struct */
93struct rtas_manage_flash_t
94{
95 int status; /* Returned status */
96 unsigned int op; /* Reject or commit image */
97};
98
99/* Status int must be first member of struct */
100struct rtas_validate_flash_t
101{
102 int status; /* Returned status */
103 char buf[VALIDATE_BUF_SIZE]; /* Candidate image buffer */
104 unsigned int buf_size; /* Size of image buf */
105 unsigned int update_results; /* Update results token */
106};
107
108static DEFINE_SPINLOCK(flash_file_open_lock);
109static struct proc_dir_entry *firmware_flash_pde;
110static struct proc_dir_entry *firmware_update_pde;
111static struct proc_dir_entry *validate_pde;
112static struct proc_dir_entry *manage_pde;
113
114/* Do simple sanity checks on the flash image. */
115static int flash_list_valid(struct flash_block_list *flist)
116{
117 struct flash_block_list *f;
118 int i;
119 unsigned long block_size, image_size;
120
121 /* Paranoid self test here. We also collect the image size. */
122 image_size = 0;
123 for (f = flist; f; f = f->next) {
124 for (i = 0; i < f->num_blocks; i++) {
125 if (f->blocks[i].data == NULL) {
126 return FLASH_IMG_NULL_DATA;
127 }
128 block_size = f->blocks[i].length;
129 if (block_size <= 0 || block_size > PAGE_SIZE) {
130 return FLASH_IMG_BAD_LEN;
131 }
132 image_size += block_size;
133 }
134 }
135
136 if (image_size < (256 << 10)) {
137 if (image_size < 2)
138 return FLASH_NO_OP;
139 }
140
141 printk(KERN_INFO "FLASH: flash image with %ld bytes stored for hardware flash on reboot\n", image_size);
142
143 return FLASH_IMG_READY;
144}
145
146static void free_flash_list(struct flash_block_list *f)
147{
148 struct flash_block_list *next;
149 int i;
150
151 while (f) {
152 for (i = 0; i < f->num_blocks; i++)
153 free_page((unsigned long)(f->blocks[i].data));
154 next = f->next;
155 free_page((unsigned long)f);
156 f = next;
157 }
158}
159
160static int rtas_flash_release(struct inode *inode, struct file *file)
161{
162 struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
163 struct rtas_update_flash_t *uf;
164
165 uf = (struct rtas_update_flash_t *) dp->data;
166 if (uf->flist) {
167 /* File was opened in write mode for a new flash attempt */
168 /* Clear saved list */
169 if (rtas_firmware_flash_list.next) {
170 free_flash_list(rtas_firmware_flash_list.next);
171 rtas_firmware_flash_list.next = NULL;
172 }
173
174 if (uf->status != FLASH_AUTH)
175 uf->status = flash_list_valid(uf->flist);
176
177 if (uf->status == FLASH_IMG_READY)
178 rtas_firmware_flash_list.next = uf->flist;
179 else
180 free_flash_list(uf->flist);
181
182 uf->flist = NULL;
183 }
184
185 atomic_dec(&dp->count);
186 return 0;
187}
188
189static void get_flash_status_msg(int status, char *buf)
190{
191 char *msg;
192
193 switch (status) {
194 case FLASH_AUTH:
195 msg = "error: this partition does not have service authority\n";
196 break;
197 case FLASH_NO_OP:
198 msg = "info: no firmware image for flash\n";
199 break;
200 case FLASH_IMG_SHORT:
201 msg = "error: flash image short\n";
202 break;
203 case FLASH_IMG_BAD_LEN:
204 msg = "error: internal error bad length\n";
205 break;
206 case FLASH_IMG_NULL_DATA:
207 msg = "error: internal error null data\n";
208 break;
209 case FLASH_IMG_READY:
210 msg = "ready: firmware image ready for flash on reboot\n";
211 break;
212 default:
213 sprintf(buf, "error: unexpected status value %d\n", status);
214 return;
215 }
216
217 strcpy(buf, msg);
218}
219
220/* Reading the proc file will show status (not the firmware contents) */
221static ssize_t rtas_flash_read(struct file *file, char __user *buf,
222 size_t count, loff_t *ppos)
223{
224 struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
225 struct rtas_update_flash_t *uf;
226 char msg[RTAS_MSG_MAXLEN];
227 int msglen;
228
229 uf = (struct rtas_update_flash_t *) dp->data;
230
231 if (!strcmp(dp->name, FIRMWARE_FLASH_NAME)) {
232 get_flash_status_msg(uf->status, msg);
233 } else { /* FIRMWARE_UPDATE_NAME */
234 sprintf(msg, "%d\n", uf->status);
235 }
236 msglen = strlen(msg);
237 if (msglen > count)
238 msglen = count;
239
240 if (ppos && *ppos != 0)
241 return 0; /* be cheap */
242
243 if (!access_ok(VERIFY_WRITE, buf, msglen))
244 return -EINVAL;
245
246 if (copy_to_user(buf, msg, msglen))
247 return -EFAULT;
248
249 if (ppos)
250 *ppos = msglen;
251 return msglen;
252}
253
254/* We could be much more efficient here. But to keep this function
255 * simple we allocate a page to the block list no matter how small the
256 * count is. If the system is low on memory it will be just as well
257 * that we fail....
258 */
259static ssize_t rtas_flash_write(struct file *file, const char __user *buffer,
260 size_t count, loff_t *off)
261{
262 struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
263 struct rtas_update_flash_t *uf;
264 char *p;
265 int next_free;
266 struct flash_block_list *fl;
267
268 uf = (struct rtas_update_flash_t *) dp->data;
269
270 if (uf->status == FLASH_AUTH || count == 0)
271 return count; /* discard data */
272
273 /* In the case that the image is not ready for flashing, the memory
274 * allocated for the block list will be freed upon the release of the
275 * proc file
276 */
277 if (uf->flist == NULL) {
278 uf->flist = (struct flash_block_list *) get_zeroed_page(GFP_KERNEL);
279 if (!uf->flist)
280 return -ENOMEM;
281 }
282
283 fl = uf->flist;
284 while (fl->next)
285 fl = fl->next; /* seek to last block_list for append */
286 next_free = fl->num_blocks;
287 if (next_free == FLASH_BLOCKS_PER_NODE) {
288 /* Need to allocate another block_list */
289 fl->next = (struct flash_block_list *)get_zeroed_page(GFP_KERNEL);
290 if (!fl->next)
291 return -ENOMEM;
292 fl = fl->next;
293 next_free = 0;
294 }
295
296 if (count > PAGE_SIZE)
297 count = PAGE_SIZE;
298 p = (char *)get_zeroed_page(GFP_KERNEL);
299 if (!p)
300 return -ENOMEM;
301
302 if(copy_from_user(p, buffer, count)) {
303 free_page((unsigned long)p);
304 return -EFAULT;
305 }
306 fl->blocks[next_free].data = p;
307 fl->blocks[next_free].length = count;
308 fl->num_blocks++;
309
310 return count;
311}
312
313static int rtas_excl_open(struct inode *inode, struct file *file)
314{
315 struct proc_dir_entry *dp = PDE(inode);
316
317 /* Enforce exclusive open with use count of PDE */
318 spin_lock(&flash_file_open_lock);
319 if (atomic_read(&dp->count) > 1) {
320 spin_unlock(&flash_file_open_lock);
321 return -EBUSY;
322 }
323
324 atomic_inc(&dp->count);
325 spin_unlock(&flash_file_open_lock);
326
327 return 0;
328}
329
330static int rtas_excl_release(struct inode *inode, struct file *file)
331{
332 struct proc_dir_entry *dp = PDE(inode);
333
334 atomic_dec(&dp->count);
335
336 return 0;
337}
338
339static void manage_flash(struct rtas_manage_flash_t *args_buf)
340{
341 unsigned int wait_time;
342 s32 rc;
343
344 while (1) {
345 rc = rtas_call(rtas_token("ibm,manage-flash-image"), 1,
346 1, NULL, args_buf->op);
347 if (rc == RTAS_RC_BUSY)
348 udelay(1);
349 else if (rtas_is_extended_busy(rc)) {
350 wait_time = rtas_extended_busy_delay_time(rc);
351 udelay(wait_time * 1000);
352 } else
353 break;
354 }
355
356 args_buf->status = rc;
357}
358
359static ssize_t manage_flash_read(struct file *file, char __user *buf,
360 size_t count, loff_t *ppos)
361{
362 struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
363 struct rtas_manage_flash_t *args_buf;
364 char msg[RTAS_MSG_MAXLEN];
365 int msglen;
366
367 args_buf = (struct rtas_manage_flash_t *) dp->data;
368 if (args_buf == NULL)
369 return 0;
370
371 msglen = sprintf(msg, "%d\n", args_buf->status);
372 if (msglen > count)
373 msglen = count;
374
375 if (ppos && *ppos != 0)
376 return 0; /* be cheap */
377
378 if (!access_ok(VERIFY_WRITE, buf, msglen))
379 return -EINVAL;
380
381 if (copy_to_user(buf, msg, msglen))
382 return -EFAULT;
383
384 if (ppos)
385 *ppos = msglen;
386 return msglen;
387}
388
389static ssize_t manage_flash_write(struct file *file, const char __user *buf,
390 size_t count, loff_t *off)
391{
392 struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
393 struct rtas_manage_flash_t *args_buf;
394 const char reject_str[] = "0";
395 const char commit_str[] = "1";
396 char stkbuf[10];
397 int op;
398
399 args_buf = (struct rtas_manage_flash_t *) dp->data;
400 if ((args_buf->status == MANAGE_AUTH) || (count == 0))
401 return count;
402
403 op = -1;
404 if (buf) {
405 if (count > 9) count = 9;
406 if (copy_from_user (stkbuf, buf, count)) {
407 return -EFAULT;
408 }
409 if (strncmp(stkbuf, reject_str, strlen(reject_str)) == 0)
410 op = RTAS_REJECT_TMP_IMG;
411 else if (strncmp(stkbuf, commit_str, strlen(commit_str)) == 0)
412 op = RTAS_COMMIT_TMP_IMG;
413 }
414
415 if (op == -1) /* buf is empty, or contains invalid string */
416 return -EINVAL;
417
418 args_buf->op = op;
419 manage_flash(args_buf);
420
421 return count;
422}
423
424static void validate_flash(struct rtas_validate_flash_t *args_buf)
425{
426 int token = rtas_token("ibm,validate-flash-image");
427 unsigned int wait_time;
428 int update_results;
429 s32 rc;
430
431 rc = 0;
432 while(1) {
433 spin_lock(&rtas_data_buf_lock);
434 memcpy(rtas_data_buf, args_buf->buf, VALIDATE_BUF_SIZE);
435 rc = rtas_call(token, 2, 2, &update_results,
436 (u32) __pa(rtas_data_buf), args_buf->buf_size);
437 memcpy(args_buf->buf, rtas_data_buf, VALIDATE_BUF_SIZE);
438 spin_unlock(&rtas_data_buf_lock);
439
440 if (rc == RTAS_RC_BUSY)
441 udelay(1);
442 else if (rtas_is_extended_busy(rc)) {
443 wait_time = rtas_extended_busy_delay_time(rc);
444 udelay(wait_time * 1000);
445 } else
446 break;
447 }
448
449 args_buf->status = rc;
450 args_buf->update_results = update_results;
451}
452
453static int get_validate_flash_msg(struct rtas_validate_flash_t *args_buf,
454 char *msg)
455{
456 int n;
457
458 if (args_buf->status >= VALIDATE_TMP_UPDATE) {
459 n = sprintf(msg, "%d\n", args_buf->update_results);
460 if ((args_buf->update_results >= VALIDATE_CUR_UNKNOWN) ||
461 (args_buf->update_results == VALIDATE_TMP_UPDATE))
462 n += sprintf(msg + n, "%s\n", args_buf->buf);
463 } else {
464 n = sprintf(msg, "%d\n", args_buf->status);
465 }
466 return n;
467}
468
469static ssize_t validate_flash_read(struct file *file, char __user *buf,
470 size_t count, loff_t *ppos)
471{
472 struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
473 struct rtas_validate_flash_t *args_buf;
474 char msg[RTAS_MSG_MAXLEN];
475 int msglen;
476
477 args_buf = (struct rtas_validate_flash_t *) dp->data;
478
479 if (ppos && *ppos != 0)
480 return 0; /* be cheap */
481
482 msglen = get_validate_flash_msg(args_buf, msg);
483 if (msglen > count)
484 msglen = count;
485
486 if (!access_ok(VERIFY_WRITE, buf, msglen))
487 return -EINVAL;
488
489 if (copy_to_user(buf, msg, msglen))
490 return -EFAULT;
491
492 if (ppos)
493 *ppos = msglen;
494 return msglen;
495}
496
497static ssize_t validate_flash_write(struct file *file, const char __user *buf,
498 size_t count, loff_t *off)
499{
500 struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
501 struct rtas_validate_flash_t *args_buf;
502 int rc;
503
504 args_buf = (struct rtas_validate_flash_t *) dp->data;
505
506 if (dp->data == NULL) {
507 dp->data = kmalloc(sizeof(struct rtas_validate_flash_t),
508 GFP_KERNEL);
509 if (dp->data == NULL)
510 return -ENOMEM;
511 }
512
513 /* We are only interested in the first 4K of the
514 * candidate image */
515 if ((*off >= VALIDATE_BUF_SIZE) ||
516 (args_buf->status == VALIDATE_AUTH)) {
517 *off += count;
518 return count;
519 }
520
521 if (*off + count >= VALIDATE_BUF_SIZE) {
522 count = VALIDATE_BUF_SIZE - *off;
523 args_buf->status = VALIDATE_READY;
524 } else {
525 args_buf->status = VALIDATE_INCOMPLETE;
526 }
527
528 if (!access_ok(VERIFY_READ, buf, count)) {
529 rc = -EFAULT;
530 goto done;
531 }
532 if (copy_from_user(args_buf->buf + *off, buf, count)) {
533 rc = -EFAULT;
534 goto done;
535 }
536
537 *off += count;
538 rc = count;
539done:
540 if (rc < 0) {
541 kfree(dp->data);
542 dp->data = NULL;
543 }
544 return rc;
545}
546
547static int validate_flash_release(struct inode *inode, struct file *file)
548{
549 struct proc_dir_entry *dp = PDE(file->f_dentry->d_inode);
550 struct rtas_validate_flash_t *args_buf;
551
552 args_buf = (struct rtas_validate_flash_t *) dp->data;
553
554 if (args_buf->status == VALIDATE_READY) {
555 args_buf->buf_size = VALIDATE_BUF_SIZE;
556 validate_flash(args_buf);
557 }
558
559 /* The matching atomic_inc was in rtas_excl_open() */
560 atomic_dec(&dp->count);
561
562 return 0;
563}
564
565static void remove_flash_pde(struct proc_dir_entry *dp)
566{
567 if (dp) {
568 if (dp->data != NULL)
569 kfree(dp->data);
570 dp->owner = NULL;
571 remove_proc_entry(dp->name, dp->parent);
572 }
573}
574
575static int initialize_flash_pde_data(const char *rtas_call_name,
576 size_t buf_size,
577 struct proc_dir_entry *dp)
578{
579 int *status;
580 int token;
581
582 dp->data = kmalloc(buf_size, GFP_KERNEL);
583 if (dp->data == NULL) {
584 remove_flash_pde(dp);
585 return -ENOMEM;
586 }
587
588 memset(dp->data, 0, buf_size);
589
590 /*
591 * This code assumes that the status int is the first member of the
592 * struct
593 */
594 status = (int *) dp->data;
595 token = rtas_token(rtas_call_name);
596 if (token == RTAS_UNKNOWN_SERVICE)
597 *status = FLASH_AUTH;
598 else
599 *status = FLASH_NO_OP;
600
601 return 0;
602}
603
604static struct proc_dir_entry *create_flash_pde(const char *filename,
605 struct file_operations *fops)
606{
607 struct proc_dir_entry *ent = NULL;
608
609 ent = create_proc_entry(filename, S_IRUSR | S_IWUSR, NULL);
610 if (ent != NULL) {
611 ent->nlink = 1;
612 ent->proc_fops = fops;
613 ent->owner = THIS_MODULE;
614 }
615
616 return ent;
617}
618
619static struct file_operations rtas_flash_operations = {
620 .read = rtas_flash_read,
621 .write = rtas_flash_write,
622 .open = rtas_excl_open,
623 .release = rtas_flash_release,
624};
625
626static struct file_operations manage_flash_operations = {
627 .read = manage_flash_read,
628 .write = manage_flash_write,
629 .open = rtas_excl_open,
630 .release = rtas_excl_release,
631};
632
633static struct file_operations validate_flash_operations = {
634 .read = validate_flash_read,
635 .write = validate_flash_write,
636 .open = rtas_excl_open,
637 .release = validate_flash_release,
638};
639
640int __init rtas_flash_init(void)
641{
642 int rc;
643
644 if (rtas_token("ibm,update-flash-64-and-reboot") ==
645 RTAS_UNKNOWN_SERVICE) {
646 printk(KERN_ERR "rtas_flash: no firmware flash support\n");
647 return 1;
648 }
649
650 firmware_flash_pde = create_flash_pde("ppc64/rtas/"
651 FIRMWARE_FLASH_NAME,
652 &rtas_flash_operations);
653 if (firmware_flash_pde == NULL) {
654 rc = -ENOMEM;
655 goto cleanup;
656 }
657
658 rc = initialize_flash_pde_data("ibm,update-flash-64-and-reboot",
659 sizeof(struct rtas_update_flash_t),
660 firmware_flash_pde);
661 if (rc != 0)
662 goto cleanup;
663
664 firmware_update_pde = create_flash_pde("ppc64/rtas/"
665 FIRMWARE_UPDATE_NAME,
666 &rtas_flash_operations);
667 if (firmware_update_pde == NULL) {
668 rc = -ENOMEM;
669 goto cleanup;
670 }
671
672 rc = initialize_flash_pde_data("ibm,update-flash-64-and-reboot",
673 sizeof(struct rtas_update_flash_t),
674 firmware_update_pde);
675 if (rc != 0)
676 goto cleanup;
677
678 validate_pde = create_flash_pde("ppc64/rtas/" VALIDATE_FLASH_NAME,
679 &validate_flash_operations);
680 if (validate_pde == NULL) {
681 rc = -ENOMEM;
682 goto cleanup;
683 }
684
685 rc = initialize_flash_pde_data("ibm,validate-flash-image",
686 sizeof(struct rtas_validate_flash_t),
687 validate_pde);
688 if (rc != 0)
689 goto cleanup;
690
691 manage_pde = create_flash_pde("ppc64/rtas/" MANAGE_FLASH_NAME,
692 &manage_flash_operations);
693 if (manage_pde == NULL) {
694 rc = -ENOMEM;
695 goto cleanup;
696 }
697
698 rc = initialize_flash_pde_data("ibm,manage-flash-image",
699 sizeof(struct rtas_manage_flash_t),
700 manage_pde);
701 if (rc != 0)
702 goto cleanup;
703
704 return 0;
705
706cleanup:
707 remove_flash_pde(firmware_flash_pde);
708 remove_flash_pde(firmware_update_pde);
709 remove_flash_pde(validate_pde);
710 remove_flash_pde(manage_pde);
711
712 return rc;
713}
714
715void __exit rtas_flash_cleanup(void)
716{
717 remove_flash_pde(firmware_flash_pde);
718 remove_flash_pde(firmware_update_pde);
719 remove_flash_pde(validate_pde);
720 remove_flash_pde(manage_pde);
721}
722
723module_init(rtas_flash_init);
724module_exit(rtas_flash_cleanup);
725MODULE_LICENSE("GPL");
diff --git a/arch/ppc64/kernel/rtasd.c b/arch/ppc64/kernel/rtasd.c
deleted file mode 100644
index e26b0420b6dd..000000000000
--- a/arch/ppc64/kernel/rtasd.c
+++ /dev/null
@@ -1,527 +0,0 @@
1/*
2 * Copyright (C) 2001 Anton Blanchard <anton@au.ibm.com>, IBM
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License
6 * as published by the Free Software Foundation; either version
7 * 2 of the License, or (at your option) any later version.
8 *
9 * Communication to userspace based on kernel/printk.c
10 */
11
12#include <linux/types.h>
13#include <linux/errno.h>
14#include <linux/sched.h>
15#include <linux/kernel.h>
16#include <linux/poll.h>
17#include <linux/proc_fs.h>
18#include <linux/init.h>
19#include <linux/vmalloc.h>
20#include <linux/spinlock.h>
21#include <linux/cpu.h>
22#include <linux/delay.h>
23
24#include <asm/uaccess.h>
25#include <asm/io.h>
26#include <asm/rtas.h>
27#include <asm/prom.h>
28#include <asm/nvram.h>
29#include <asm/atomic.h>
30#include <asm/systemcfg.h>
31
32#if 0
33#define DEBUG(A...) printk(KERN_ERR A)
34#else
35#define DEBUG(A...)
36#endif
37
38static DEFINE_SPINLOCK(rtasd_log_lock);
39
40DECLARE_WAIT_QUEUE_HEAD(rtas_log_wait);
41
42static char *rtas_log_buf;
43static unsigned long rtas_log_start;
44static unsigned long rtas_log_size;
45
46static int surveillance_timeout = -1;
47static unsigned int rtas_event_scan_rate;
48static unsigned int rtas_error_log_max;
49static unsigned int rtas_error_log_buffer_max;
50
51static int full_rtas_msgs = 0;
52
53extern int no_logging;
54
55volatile int error_log_cnt = 0;
56
57/*
58 * Since we use 32 bit RTAS, the physical address of this must be below
59 * 4G or else bad things happen. Allocate this in the kernel data and
60 * make it big enough.
61 */
62static unsigned char logdata[RTAS_ERROR_LOG_MAX];
63
64static int get_eventscan_parms(void);
65
66static char *rtas_type[] = {
67 "Unknown", "Retry", "TCE Error", "Internal Device Failure",
68 "Timeout", "Data Parity", "Address Parity", "Cache Parity",
69 "Address Invalid", "ECC Uncorrected", "ECC Corrupted",
70};
71
72static char *rtas_event_type(int type)
73{
74 if ((type > 0) && (type < 11))
75 return rtas_type[type];
76
77 switch (type) {
78 case RTAS_TYPE_EPOW:
79 return "EPOW";
80 case RTAS_TYPE_PLATFORM:
81 return "Platform Error";
82 case RTAS_TYPE_IO:
83 return "I/O Event";
84 case RTAS_TYPE_INFO:
85 return "Platform Information Event";
86 case RTAS_TYPE_DEALLOC:
87 return "Resource Deallocation Event";
88 case RTAS_TYPE_DUMP:
89 return "Dump Notification Event";
90 }
91
92 return rtas_type[0];
93}
94
95/* To see this info, grep RTAS /var/log/messages and each entry
96 * will be collected together with obvious begin/end.
97 * There will be a unique identifier on the begin and end lines.
98 * This will persist across reboots.
99 *
100 * format of error logs returned from RTAS:
101 * bytes (size) : contents
102 * --------------------------------------------------------
103 * 0-7 (8) : rtas_error_log
104 * 8-47 (40) : extended info
105 * 48-51 (4) : vendor id
106 * 52-1023 (vendor specific) : location code and debug data
107 */
108static void printk_log_rtas(char *buf, int len)
109{
110
111 int i,j,n = 0;
112 int perline = 16;
113 char buffer[64];
114 char * str = "RTAS event";
115
116 if (full_rtas_msgs) {
117 printk(RTAS_DEBUG "%d -------- %s begin --------\n",
118 error_log_cnt, str);
119
120 /*
121 * Print perline bytes on each line, each line will start
122 * with RTAS and a changing number, so syslogd will
123 * print lines that are otherwise the same. Separate every
124 * 4 bytes with a space.
125 */
126 for (i = 0; i < len; i++) {
127 j = i % perline;
128 if (j == 0) {
129 memset(buffer, 0, sizeof(buffer));
130 n = sprintf(buffer, "RTAS %d:", i/perline);
131 }
132
133 if ((i % 4) == 0)
134 n += sprintf(buffer+n, " ");
135
136 n += sprintf(buffer+n, "%02x", (unsigned char)buf[i]);
137
138 if (j == (perline-1))
139 printk(KERN_DEBUG "%s\n", buffer);
140 }
141 if ((i % perline) != 0)
142 printk(KERN_DEBUG "%s\n", buffer);
143
144 printk(RTAS_DEBUG "%d -------- %s end ----------\n",
145 error_log_cnt, str);
146 } else {
147 struct rtas_error_log *errlog = (struct rtas_error_log *)buf;
148
149 printk(RTAS_DEBUG "event: %d, Type: %s, Severity: %d\n",
150 error_log_cnt, rtas_event_type(errlog->type),
151 errlog->severity);
152 }
153}
154
155static int log_rtas_len(char * buf)
156{
157 int len;
158 struct rtas_error_log *err;
159
160 /* rtas fixed header */
161 len = 8;
162 err = (struct rtas_error_log *)buf;
163 if (err->extended_log_length) {
164
165 /* extended header */
166 len += err->extended_log_length;
167 }
168
169 if (rtas_error_log_max == 0) {
170 get_eventscan_parms();
171 }
172 if (len > rtas_error_log_max)
173 len = rtas_error_log_max;
174
175 return len;
176}
177
178/*
179 * First write to nvram, if fatal error, that is the only
180 * place we log the info. The error will be picked up
181 * on the next reboot by rtasd. If not fatal, run the
182 * method for the type of error. Currently, only RTAS
183 * errors have methods implemented, but in the future
184 * there might be a need to store data in nvram before a
185 * call to panic().
186 *
187 * XXX We write to nvram periodically, to indicate error has
188 * been written and sync'd, but there is a possibility
189 * that if we don't shutdown correctly, a duplicate error
190 * record will be created on next reboot.
191 */
192void pSeries_log_error(char *buf, unsigned int err_type, int fatal)
193{
194 unsigned long offset;
195 unsigned long s;
196 int len = 0;
197
198 DEBUG("logging event\n");
199 if (buf == NULL)
200 return;
201
202 spin_lock_irqsave(&rtasd_log_lock, s);
203
204 /* get length and increase count */
205 switch (err_type & ERR_TYPE_MASK) {
206 case ERR_TYPE_RTAS_LOG:
207 len = log_rtas_len(buf);
208 if (!(err_type & ERR_FLAG_BOOT))
209 error_log_cnt++;
210 break;
211 case ERR_TYPE_KERNEL_PANIC:
212 default:
213 spin_unlock_irqrestore(&rtasd_log_lock, s);
214 return;
215 }
216
217 /* Write error to NVRAM */
218 if (!no_logging && !(err_type & ERR_FLAG_BOOT))
219 nvram_write_error_log(buf, len, err_type);
220
221 /*
222 * rtas errors can occur during boot, and we do want to capture
223 * those somewhere, even if nvram isn't ready (why not?), and even
224 * if rtasd isn't ready. Put them into the boot log, at least.
225 */
226 if ((err_type & ERR_TYPE_MASK) == ERR_TYPE_RTAS_LOG)
227 printk_log_rtas(buf, len);
228
229 /* Check to see if we need to or have stopped logging */
230 if (fatal || no_logging) {
231 no_logging = 1;
232 spin_unlock_irqrestore(&rtasd_log_lock, s);
233 return;
234 }
235
236 /* call type specific method for error */
237 switch (err_type & ERR_TYPE_MASK) {
238 case ERR_TYPE_RTAS_LOG:
239 offset = rtas_error_log_buffer_max *
240 ((rtas_log_start+rtas_log_size) & LOG_NUMBER_MASK);
241
242 /* First copy over sequence number */
243 memcpy(&rtas_log_buf[offset], (void *) &error_log_cnt, sizeof(int));
244
245 /* Second copy over error log data */
246 offset += sizeof(int);
247 memcpy(&rtas_log_buf[offset], buf, len);
248
249 if (rtas_log_size < LOG_NUMBER)
250 rtas_log_size += 1;
251 else
252 rtas_log_start += 1;
253
254 spin_unlock_irqrestore(&rtasd_log_lock, s);
255 wake_up_interruptible(&rtas_log_wait);
256 break;
257 case ERR_TYPE_KERNEL_PANIC:
258 default:
259 spin_unlock_irqrestore(&rtasd_log_lock, s);
260 return;
261 }
262
263}
264
265
266static int rtas_log_open(struct inode * inode, struct file * file)
267{
268 return 0;
269}
270
271static int rtas_log_release(struct inode * inode, struct file * file)
272{
273 return 0;
274}
275
276/* This will check if all events are logged, if they are then, we
277 * know that we can safely clear the events in NVRAM.
278 * Next we'll sit and wait for something else to log.
279 */
280static ssize_t rtas_log_read(struct file * file, char __user * buf,
281 size_t count, loff_t *ppos)
282{
283 int error;
284 char *tmp;
285 unsigned long s;
286 unsigned long offset;
287
288 if (!buf || count < rtas_error_log_buffer_max)
289 return -EINVAL;
290
291 count = rtas_error_log_buffer_max;
292
293 if (!access_ok(VERIFY_WRITE, buf, count))
294 return -EFAULT;
295
296 tmp = kmalloc(count, GFP_KERNEL);
297 if (!tmp)
298 return -ENOMEM;
299
300
301 spin_lock_irqsave(&rtasd_log_lock, s);
302 /* if it's 0, then we know we got the last one (the one in NVRAM) */
303 if (rtas_log_size == 0 && !no_logging)
304 nvram_clear_error_log();
305 spin_unlock_irqrestore(&rtasd_log_lock, s);
306
307
308 error = wait_event_interruptible(rtas_log_wait, rtas_log_size);
309 if (error)
310 goto out;
311
312 spin_lock_irqsave(&rtasd_log_lock, s);
313 offset = rtas_error_log_buffer_max * (rtas_log_start & LOG_NUMBER_MASK);
314 memcpy(tmp, &rtas_log_buf[offset], count);
315
316 rtas_log_start += 1;
317 rtas_log_size -= 1;
318 spin_unlock_irqrestore(&rtasd_log_lock, s);
319
320 error = copy_to_user(buf, tmp, count) ? -EFAULT : count;
321out:
322 kfree(tmp);
323 return error;
324}
325
326static unsigned int rtas_log_poll(struct file *file, poll_table * wait)
327{
328 poll_wait(file, &rtas_log_wait, wait);
329 if (rtas_log_size)
330 return POLLIN | POLLRDNORM;
331 return 0;
332}
333
334struct file_operations proc_rtas_log_operations = {
335 .read = rtas_log_read,
336 .poll = rtas_log_poll,
337 .open = rtas_log_open,
338 .release = rtas_log_release,
339};
340
341static int enable_surveillance(int timeout)
342{
343 int error;
344
345 error = rtas_set_indicator(SURVEILLANCE_TOKEN, 0, timeout);
346
347 if (error == 0)
348 return 0;
349
350 if (error == -EINVAL) {
351 printk(KERN_INFO "rtasd: surveillance not supported\n");
352 return 0;
353 }
354
355 printk(KERN_ERR "rtasd: could not update surveillance\n");
356 return -1;
357}
358
359static int get_eventscan_parms(void)
360{
361 struct device_node *node;
362 int *ip;
363
364 node = of_find_node_by_path("/rtas");
365
366 ip = (int *)get_property(node, "rtas-event-scan-rate", NULL);
367 if (ip == NULL) {
368 printk(KERN_ERR "rtasd: no rtas-event-scan-rate\n");
369 of_node_put(node);
370 return -1;
371 }
372 rtas_event_scan_rate = *ip;
373 DEBUG("rtas-event-scan-rate %d\n", rtas_event_scan_rate);
374
375 /* Make room for the sequence number */
376 rtas_error_log_max = rtas_get_error_log_max();
377 rtas_error_log_buffer_max = rtas_error_log_max + sizeof(int);
378
379 of_node_put(node);
380
381 return 0;
382}
383
384static void do_event_scan(int event_scan)
385{
386 int error;
387 do {
388 memset(logdata, 0, rtas_error_log_max);
389 error = rtas_call(event_scan, 4, 1, NULL,
390 RTAS_EVENT_SCAN_ALL_EVENTS, 0,
391 __pa(logdata), rtas_error_log_max);
392 if (error == -1) {
393 printk(KERN_ERR "event-scan failed\n");
394 break;
395 }
396
397 if (error == 0)
398 pSeries_log_error(logdata, ERR_TYPE_RTAS_LOG, 0);
399
400 } while(error == 0);
401}
402
403static void do_event_scan_all_cpus(long delay)
404{
405 int cpu;
406
407 lock_cpu_hotplug();
408 cpu = first_cpu(cpu_online_map);
409 for (;;) {
410 set_cpus_allowed(current, cpumask_of_cpu(cpu));
411 do_event_scan(rtas_token("event-scan"));
412 set_cpus_allowed(current, CPU_MASK_ALL);
413
414 /* Drop hotplug lock, and sleep for the specified delay */
415 unlock_cpu_hotplug();
416 msleep_interruptible(delay);
417 lock_cpu_hotplug();
418
419 cpu = next_cpu(cpu, cpu_online_map);
420 if (cpu == NR_CPUS)
421 break;
422 }
423 unlock_cpu_hotplug();
424}
425
426static int rtasd(void *unused)
427{
428 unsigned int err_type;
429 int event_scan = rtas_token("event-scan");
430 int rc;
431
432 daemonize("rtasd");
433
434 if (event_scan == RTAS_UNKNOWN_SERVICE || get_eventscan_parms() == -1)
435 goto error;
436
437 rtas_log_buf = vmalloc(rtas_error_log_buffer_max*LOG_NUMBER);
438 if (!rtas_log_buf) {
439 printk(KERN_ERR "rtasd: no memory\n");
440 goto error;
441 }
442
443 printk(KERN_INFO "RTAS daemon started\n");
444
445 DEBUG("will sleep for %d milliseconds\n", (30000/rtas_event_scan_rate));
446
447 /* See if we have any error stored in NVRAM */
448 memset(logdata, 0, rtas_error_log_max);
449
450 rc = nvram_read_error_log(logdata, rtas_error_log_max, &err_type);
451
452 /* We can use rtas_log_buf now */
453 no_logging = 0;
454
455 if (!rc) {
456 if (err_type != ERR_FLAG_ALREADY_LOGGED) {
457 pSeries_log_error(logdata, err_type | ERR_FLAG_BOOT, 0);
458 }
459 }
460
461 /* First pass. */
462 do_event_scan_all_cpus(1000);
463
464 if (surveillance_timeout != -1) {
465 DEBUG("enabling surveillance\n");
466 enable_surveillance(surveillance_timeout);
467 DEBUG("surveillance enabled\n");
468 }
469
470 /* Delay should be at least one second since some
471 * machines have problems if we call event-scan too
472 * quickly. */
473 for (;;)
474 do_event_scan_all_cpus(30000/rtas_event_scan_rate);
475
476error:
477 /* Should delete proc entries */
478 return -EINVAL;
479}
480
481static int __init rtas_init(void)
482{
483 struct proc_dir_entry *entry;
484
485 /* No RTAS, only warn if we are on a pSeries box */
486 if (rtas_token("event-scan") == RTAS_UNKNOWN_SERVICE) {
487 if (systemcfg->platform & PLATFORM_PSERIES)
488 printk(KERN_INFO "rtasd: no event-scan on system\n");
489 return 1;
490 }
491
492 entry = create_proc_entry("ppc64/rtas/error_log", S_IRUSR, NULL);
493 if (entry)
494 entry->proc_fops = &proc_rtas_log_operations;
495 else
496 printk(KERN_ERR "Failed to create error_log proc entry\n");
497
498 if (kernel_thread(rtasd, NULL, CLONE_FS) < 0)
499 printk(KERN_ERR "Failed to start RTAS daemon\n");
500
501 return 0;
502}
503
504static int __init surveillance_setup(char *str)
505{
506 int i;
507
508 if (get_option(&str,&i)) {
509 if (i >= 0 && i <= 255)
510 surveillance_timeout = i;
511 }
512
513 return 1;
514}
515
516static int __init rtasmsgs_setup(char *str)
517{
518 if (strcmp(str, "on") == 0)
519 full_rtas_msgs = 1;
520 else if (strcmp(str, "off") == 0)
521 full_rtas_msgs = 0;
522
523 return 1;
524}
525__initcall(rtas_init);
526__setup("surveillance=", surveillance_setup);
527__setup("rtasmsgs=", rtasmsgs_setup);
diff --git a/arch/ppc64/kernel/scanlog.c b/arch/ppc64/kernel/scanlog.c
index 215bf8900304..2edc947f7c44 100644
--- a/arch/ppc64/kernel/scanlog.c
+++ b/arch/ppc64/kernel/scanlog.c
@@ -225,8 +225,7 @@ int __init scanlog_init(void)
225void __exit scanlog_cleanup(void) 225void __exit scanlog_cleanup(void)
226{ 226{
227 if (proc_ppc64_scan_log_dump) { 227 if (proc_ppc64_scan_log_dump) {
228 if (proc_ppc64_scan_log_dump->data) 228 kfree(proc_ppc64_scan_log_dump->data);
229 kfree(proc_ppc64_scan_log_dump->data);
230 remove_proc_entry("scan-log-dump", proc_ppc64_scan_log_dump->parent); 229 remove_proc_entry("scan-log-dump", proc_ppc64_scan_log_dump->parent);
231 } 230 }
232} 231}
diff --git a/arch/ppc64/kernel/signal.c b/arch/ppc64/kernel/signal.c
deleted file mode 100644
index ec9d0984b6a0..000000000000
--- a/arch/ppc64/kernel/signal.c
+++ /dev/null
@@ -1,581 +0,0 @@
1/*
2 * linux/arch/ppc64/kernel/signal.c
3 *
4 * PowerPC version
5 * Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
6 *
7 * Derived from "arch/i386/kernel/signal.c"
8 * Copyright (C) 1991, 1992 Linus Torvalds
9 * 1997-11-28 Modified for POSIX.1b signals by Richard Henderson
10 *
11 * This program is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU General Public License
13 * as published by the Free Software Foundation; either version
14 * 2 of the License, or (at your option) any later version.
15 */
16
17#include <linux/config.h>
18#include <linux/sched.h>
19#include <linux/mm.h>
20#include <linux/smp.h>
21#include <linux/smp_lock.h>
22#include <linux/kernel.h>
23#include <linux/signal.h>
24#include <linux/errno.h>
25#include <linux/wait.h>
26#include <linux/unistd.h>
27#include <linux/stddef.h>
28#include <linux/elf.h>
29#include <linux/ptrace.h>
30#include <linux/module.h>
31
32#include <asm/sigcontext.h>
33#include <asm/ucontext.h>
34#include <asm/uaccess.h>
35#include <asm/pgtable.h>
36#include <asm/ppcdebug.h>
37#include <asm/unistd.h>
38#include <asm/cacheflush.h>
39#include <asm/vdso.h>
40
41#define DEBUG_SIG 0
42
43#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
44
45#define GP_REGS_SIZE min(sizeof(elf_gregset_t), sizeof(struct pt_regs))
46#define FP_REGS_SIZE sizeof(elf_fpregset_t)
47
48#define TRAMP_TRACEBACK 3
49#define TRAMP_SIZE 6
50
51/*
52 * When we have signals to deliver, we set up on the user stack,
53 * going down from the original stack pointer:
54 * 1) a rt_sigframe struct which contains the ucontext
55 * 2) a gap of __SIGNAL_FRAMESIZE bytes which acts as a dummy caller
56 * frame for the signal handler.
57 */
58
59struct rt_sigframe {
60 /* sys_rt_sigreturn requires the ucontext be the first field */
61 struct ucontext uc;
62 unsigned long _unused[2];
63 unsigned int tramp[TRAMP_SIZE];
64 struct siginfo *pinfo;
65 void *puc;
66 struct siginfo info;
67 /* 64 bit ABI allows for 288 bytes below sp before decrementing it. */
68 char abigap[288];
69} __attribute__ ((aligned (16)));
70
71
72/*
73 * Atomically swap in the new signal mask, and wait for a signal.
74 */
75long sys_rt_sigsuspend(sigset_t __user *unewset, size_t sigsetsize, int p3, int p4,
76 int p6, int p7, struct pt_regs *regs)
77{
78 sigset_t saveset, newset;
79
80 /* XXX: Don't preclude handling different sized sigset_t's. */
81 if (sigsetsize != sizeof(sigset_t))
82 return -EINVAL;
83
84 if (copy_from_user(&newset, unewset, sizeof(newset)))
85 return -EFAULT;
86 sigdelsetmask(&newset, ~_BLOCKABLE);
87
88 spin_lock_irq(&current->sighand->siglock);
89 saveset = current->blocked;
90 current->blocked = newset;
91 recalc_sigpending();
92 spin_unlock_irq(&current->sighand->siglock);
93
94 regs->result = -EINTR;
95 regs->gpr[3] = EINTR;
96 regs->ccr |= 0x10000000;
97 while (1) {
98 current->state = TASK_INTERRUPTIBLE;
99 schedule();
100 if (do_signal(&saveset, regs))
101 return 0;
102 }
103}
104
105long sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss, unsigned long r5,
106 unsigned long r6, unsigned long r7, unsigned long r8,
107 struct pt_regs *regs)
108{
109 return do_sigaltstack(uss, uoss, regs->gpr[1]);
110}
111
112
113/*
114 * Set up the sigcontext for the signal frame.
115 */
116
117static long setup_sigcontext(struct sigcontext __user *sc, struct pt_regs *regs,
118 int signr, sigset_t *set, unsigned long handler)
119{
120 /* When CONFIG_ALTIVEC is set, we _always_ setup v_regs even if the
121 * process never used altivec yet (MSR_VEC is zero in pt_regs of
122 * the context). This is very important because we must ensure we
123 * don't lose the VRSAVE content that may have been set prior to
124 * the process doing its first vector operation
125 * Userland shall check AT_HWCAP to know wether it can rely on the
126 * v_regs pointer or not
127 */
128#ifdef CONFIG_ALTIVEC
129 elf_vrreg_t __user *v_regs = (elf_vrreg_t __user *)(((unsigned long)sc->vmx_reserve + 15) & ~0xful);
130#endif
131 long err = 0;
132
133 flush_fp_to_thread(current);
134
135 /* Make sure signal doesn't get spurrious FP exceptions */
136 current->thread.fpscr.val = 0;
137
138#ifdef CONFIG_ALTIVEC
139 err |= __put_user(v_regs, &sc->v_regs);
140
141 /* save altivec registers */
142 if (current->thread.used_vr) {
143 flush_altivec_to_thread(current);
144 /* Copy 33 vec registers (vr0..31 and vscr) to the stack */
145 err |= __copy_to_user(v_regs, current->thread.vr, 33 * sizeof(vector128));
146 /* set MSR_VEC in the MSR value in the frame to indicate that sc->v_reg)
147 * contains valid data.
148 */
149 regs->msr |= MSR_VEC;
150 }
151 /* We always copy to/from vrsave, it's 0 if we don't have or don't
152 * use altivec.
153 */
154 err |= __put_user(current->thread.vrsave, (u32 __user *)&v_regs[33]);
155#else /* CONFIG_ALTIVEC */
156 err |= __put_user(0, &sc->v_regs);
157#endif /* CONFIG_ALTIVEC */
158 err |= __put_user(&sc->gp_regs, &sc->regs);
159 err |= __copy_to_user(&sc->gp_regs, regs, GP_REGS_SIZE);
160 err |= __copy_to_user(&sc->fp_regs, &current->thread.fpr, FP_REGS_SIZE);
161 err |= __put_user(signr, &sc->signal);
162 err |= __put_user(handler, &sc->handler);
163 if (set != NULL)
164 err |= __put_user(set->sig[0], &sc->oldmask);
165
166 return err;
167}
168
169/*
170 * Restore the sigcontext from the signal frame.
171 */
172
173static long restore_sigcontext(struct pt_regs *regs, sigset_t *set, int sig,
174 struct sigcontext __user *sc)
175{
176#ifdef CONFIG_ALTIVEC
177 elf_vrreg_t __user *v_regs;
178#endif
179 unsigned long err = 0;
180 unsigned long save_r13 = 0;
181 elf_greg_t *gregs = (elf_greg_t *)regs;
182#ifdef CONFIG_ALTIVEC
183 unsigned long msr;
184#endif
185 int i;
186
187 /* If this is not a signal return, we preserve the TLS in r13 */
188 if (!sig)
189 save_r13 = regs->gpr[13];
190
191 /* copy everything before MSR */
192 err |= __copy_from_user(regs, &sc->gp_regs,
193 PT_MSR*sizeof(unsigned long));
194
195 /* skip MSR and SOFTE */
196 for (i = PT_MSR+1; i <= PT_RESULT; i++) {
197 if (i == PT_SOFTE)
198 continue;
199 err |= __get_user(gregs[i], &sc->gp_regs[i]);
200 }
201
202 if (!sig)
203 regs->gpr[13] = save_r13;
204 err |= __copy_from_user(&current->thread.fpr, &sc->fp_regs, FP_REGS_SIZE);
205 if (set != NULL)
206 err |= __get_user(set->sig[0], &sc->oldmask);
207
208#ifdef CONFIG_ALTIVEC
209 err |= __get_user(v_regs, &sc->v_regs);
210 err |= __get_user(msr, &sc->gp_regs[PT_MSR]);
211 if (err)
212 return err;
213 /* Copy 33 vec registers (vr0..31 and vscr) from the stack */
214 if (v_regs != 0 && (msr & MSR_VEC) != 0)
215 err |= __copy_from_user(current->thread.vr, v_regs,
216 33 * sizeof(vector128));
217 else if (current->thread.used_vr)
218 memset(current->thread.vr, 0, 33 * sizeof(vector128));
219 /* Always get VRSAVE back */
220 if (v_regs != 0)
221 err |= __get_user(current->thread.vrsave, (u32 __user *)&v_regs[33]);
222 else
223 current->thread.vrsave = 0;
224#endif /* CONFIG_ALTIVEC */
225
226#ifndef CONFIG_SMP
227 preempt_disable();
228 if (last_task_used_math == current)
229 last_task_used_math = NULL;
230 if (last_task_used_altivec == current)
231 last_task_used_altivec = NULL;
232 preempt_enable();
233#endif
234 /* Force reload of FP/VEC */
235 regs->msr &= ~(MSR_FP | MSR_FE0 | MSR_FE1 | MSR_VEC);
236
237 return err;
238}
239
240/*
241 * Allocate space for the signal frame
242 */
243static inline void __user * get_sigframe(struct k_sigaction *ka, struct pt_regs *regs,
244 size_t frame_size)
245{
246 unsigned long newsp;
247
248 /* Default to using normal stack */
249 newsp = regs->gpr[1];
250
251 if (ka->sa.sa_flags & SA_ONSTACK) {
252 if (! on_sig_stack(regs->gpr[1]))
253 newsp = (current->sas_ss_sp + current->sas_ss_size);
254 }
255
256 return (void __user *)((newsp - frame_size) & -16ul);
257}
258
259/*
260 * Setup the trampoline code on the stack
261 */
262static long setup_trampoline(unsigned int syscall, unsigned int __user *tramp)
263{
264 int i;
265 long err = 0;
266
267 /* addi r1, r1, __SIGNAL_FRAMESIZE # Pop the dummy stackframe */
268 err |= __put_user(0x38210000UL | (__SIGNAL_FRAMESIZE & 0xffff), &tramp[0]);
269 /* li r0, __NR_[rt_]sigreturn| */
270 err |= __put_user(0x38000000UL | (syscall & 0xffff), &tramp[1]);
271 /* sc */
272 err |= __put_user(0x44000002UL, &tramp[2]);
273
274 /* Minimal traceback info */
275 for (i=TRAMP_TRACEBACK; i < TRAMP_SIZE ;i++)
276 err |= __put_user(0, &tramp[i]);
277
278 if (!err)
279 flush_icache_range((unsigned long) &tramp[0],
280 (unsigned long) &tramp[TRAMP_SIZE]);
281
282 return err;
283}
284
285/*
286 * Restore the user process's signal mask (also used by signal32.c)
287 */
288void restore_sigmask(sigset_t *set)
289{
290 sigdelsetmask(set, ~_BLOCKABLE);
291 spin_lock_irq(&current->sighand->siglock);
292 current->blocked = *set;
293 recalc_sigpending();
294 spin_unlock_irq(&current->sighand->siglock);
295}
296
297
298/*
299 * Handle {get,set,swap}_context operations
300 */
301int sys_swapcontext(struct ucontext __user *old_ctx,
302 struct ucontext __user *new_ctx,
303 long ctx_size, long r6, long r7, long r8, struct pt_regs *regs)
304{
305 unsigned char tmp;
306 sigset_t set;
307
308 /* Context size is for future use. Right now, we only make sure
309 * we are passed something we understand
310 */
311 if (ctx_size < sizeof(struct ucontext))
312 return -EINVAL;
313
314 if (old_ctx != NULL) {
315 if (!access_ok(VERIFY_WRITE, old_ctx, sizeof(*old_ctx))
316 || setup_sigcontext(&old_ctx->uc_mcontext, regs, 0, NULL, 0)
317 || __copy_to_user(&old_ctx->uc_sigmask,
318 &current->blocked, sizeof(sigset_t)))
319 return -EFAULT;
320 }
321 if (new_ctx == NULL)
322 return 0;
323 if (!access_ok(VERIFY_READ, new_ctx, sizeof(*new_ctx))
324 || __get_user(tmp, (u8 __user *) new_ctx)
325 || __get_user(tmp, (u8 __user *) (new_ctx + 1) - 1))
326 return -EFAULT;
327
328 /*
329 * If we get a fault copying the context into the kernel's
330 * image of the user's registers, we can't just return -EFAULT
331 * because the user's registers will be corrupted. For instance
332 * the NIP value may have been updated but not some of the
333 * other registers. Given that we have done the access_ok
334 * and successfully read the first and last bytes of the region
335 * above, this should only happen in an out-of-memory situation
336 * or if another thread unmaps the region containing the context.
337 * We kill the task with a SIGSEGV in this situation.
338 */
339
340 if (__copy_from_user(&set, &new_ctx->uc_sigmask, sizeof(set)))
341 do_exit(SIGSEGV);
342 restore_sigmask(&set);
343 if (restore_sigcontext(regs, NULL, 0, &new_ctx->uc_mcontext))
344 do_exit(SIGSEGV);
345
346 /* This returns like rt_sigreturn */
347 return 0;
348}
349
350
351/*
352 * Do a signal return; undo the signal stack.
353 */
354
355int sys_rt_sigreturn(unsigned long r3, unsigned long r4, unsigned long r5,
356 unsigned long r6, unsigned long r7, unsigned long r8,
357 struct pt_regs *regs)
358{
359 struct ucontext __user *uc = (struct ucontext __user *)regs->gpr[1];
360 sigset_t set;
361
362 /* Always make any pending restarted system calls return -EINTR */
363 current_thread_info()->restart_block.fn = do_no_restart_syscall;
364
365 if (!access_ok(VERIFY_READ, uc, sizeof(*uc)))
366 goto badframe;
367
368 if (__copy_from_user(&set, &uc->uc_sigmask, sizeof(set)))
369 goto badframe;
370 restore_sigmask(&set);
371 if (restore_sigcontext(regs, NULL, 1, &uc->uc_mcontext))
372 goto badframe;
373
374 /* do_sigaltstack expects a __user pointer and won't modify
375 * what's in there anyway
376 */
377 do_sigaltstack(&uc->uc_stack, NULL, regs->gpr[1]);
378
379 return regs->result;
380
381badframe:
382#if DEBUG_SIG
383 printk("badframe in sys_rt_sigreturn, regs=%p uc=%p &uc->uc_mcontext=%p\n",
384 regs, uc, &uc->uc_mcontext);
385#endif
386 force_sig(SIGSEGV, current);
387 return 0;
388}
389
390static int setup_rt_frame(int signr, struct k_sigaction *ka, siginfo_t *info,
391 sigset_t *set, struct pt_regs *regs)
392{
393 /* Handler is *really* a pointer to the function descriptor for
394 * the signal routine. The first entry in the function
395 * descriptor is the entry address of signal and the second
396 * entry is the TOC value we need to use.
397 */
398 func_descr_t __user *funct_desc_ptr;
399 struct rt_sigframe __user *frame;
400 unsigned long newsp = 0;
401 long err = 0;
402
403 frame = get_sigframe(ka, regs, sizeof(*frame));
404
405 if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame)))
406 goto badframe;
407
408 err |= __put_user(&frame->info, &frame->pinfo);
409 err |= __put_user(&frame->uc, &frame->puc);
410 err |= copy_siginfo_to_user(&frame->info, info);
411 if (err)
412 goto badframe;
413
414 /* Create the ucontext. */
415 err |= __put_user(0, &frame->uc.uc_flags);
416 err |= __put_user(0, &frame->uc.uc_link);
417 err |= __put_user(current->sas_ss_sp, &frame->uc.uc_stack.ss_sp);
418 err |= __put_user(sas_ss_flags(regs->gpr[1]),
419 &frame->uc.uc_stack.ss_flags);
420 err |= __put_user(current->sas_ss_size, &frame->uc.uc_stack.ss_size);
421 err |= setup_sigcontext(&frame->uc.uc_mcontext, regs, signr, NULL,
422 (unsigned long)ka->sa.sa_handler);
423 err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set));
424 if (err)
425 goto badframe;
426
427 /* Set up to return from userspace. */
428 if (vdso64_rt_sigtramp && current->thread.vdso_base) {
429 regs->link = current->thread.vdso_base + vdso64_rt_sigtramp;
430 } else {
431 err |= setup_trampoline(__NR_rt_sigreturn, &frame->tramp[0]);
432 if (err)
433 goto badframe;
434 regs->link = (unsigned long) &frame->tramp[0];
435 }
436 funct_desc_ptr = (func_descr_t __user *) ka->sa.sa_handler;
437
438 /* Allocate a dummy caller frame for the signal handler. */
439 newsp = (unsigned long)frame - __SIGNAL_FRAMESIZE;
440 err |= put_user(regs->gpr[1], (unsigned long __user *)newsp);
441
442 /* Set up "regs" so we "return" to the signal handler. */
443 err |= get_user(regs->nip, &funct_desc_ptr->entry);
444 regs->gpr[1] = newsp;
445 err |= get_user(regs->gpr[2], &funct_desc_ptr->toc);
446 regs->gpr[3] = signr;
447 regs->result = 0;
448 if (ka->sa.sa_flags & SA_SIGINFO) {
449 err |= get_user(regs->gpr[4], (unsigned long __user *)&frame->pinfo);
450 err |= get_user(regs->gpr[5], (unsigned long __user *)&frame->puc);
451 regs->gpr[6] = (unsigned long) frame;
452 } else {
453 regs->gpr[4] = (unsigned long)&frame->uc.uc_mcontext;
454 }
455 if (err)
456 goto badframe;
457
458 if (test_thread_flag(TIF_SINGLESTEP))
459 ptrace_notify(SIGTRAP);
460
461 return 1;
462
463badframe:
464#if DEBUG_SIG
465 printk("badframe in setup_rt_frame, regs=%p frame=%p newsp=%lx\n",
466 regs, frame, newsp);
467#endif
468 force_sigsegv(signr, current);
469 return 0;
470}
471
472
473/*
474 * OK, we're invoking a handler
475 */
476static int handle_signal(unsigned long sig, struct k_sigaction *ka,
477 siginfo_t *info, sigset_t *oldset, struct pt_regs *regs)
478{
479 int ret;
480
481 /* Set up Signal Frame */
482 ret = setup_rt_frame(sig, ka, info, oldset, regs);
483
484 if (ret) {
485 spin_lock_irq(&current->sighand->siglock);
486 sigorsets(&current->blocked, &current->blocked, &ka->sa.sa_mask);
487 if (!(ka->sa.sa_flags & SA_NODEFER))
488 sigaddset(&current->blocked,sig);
489 recalc_sigpending();
490 spin_unlock_irq(&current->sighand->siglock);
491 }
492
493 return ret;
494}
495
496static inline void syscall_restart(struct pt_regs *regs, struct k_sigaction *ka)
497{
498 switch ((int)regs->result) {
499 case -ERESTART_RESTARTBLOCK:
500 case -ERESTARTNOHAND:
501 /* ERESTARTNOHAND means that the syscall should only be
502 * restarted if there was no handler for the signal, and since
503 * we only get here if there is a handler, we dont restart.
504 */
505 regs->result = -EINTR;
506 break;
507 case -ERESTARTSYS:
508 /* ERESTARTSYS means to restart the syscall if there is no
509 * handler or the handler was registered with SA_RESTART
510 */
511 if (!(ka->sa.sa_flags & SA_RESTART)) {
512 regs->result = -EINTR;
513 break;
514 }
515 /* fallthrough */
516 case -ERESTARTNOINTR:
517 /* ERESTARTNOINTR means that the syscall should be
518 * called again after the signal handler returns.
519 */
520 regs->gpr[3] = regs->orig_gpr3;
521 regs->nip -= 4;
522 regs->result = 0;
523 break;
524 }
525}
526
527/*
528 * Note that 'init' is a special process: it doesn't get signals it doesn't
529 * want to handle. Thus you cannot kill init even with a SIGKILL even by
530 * mistake.
531 */
532int do_signal(sigset_t *oldset, struct pt_regs *regs)
533{
534 siginfo_t info;
535 int signr;
536 struct k_sigaction ka;
537
538 /*
539 * If the current thread is 32 bit - invoke the
540 * 32 bit signal handling code
541 */
542 if (test_thread_flag(TIF_32BIT))
543 return do_signal32(oldset, regs);
544
545 if (!oldset)
546 oldset = &current->blocked;
547
548 signr = get_signal_to_deliver(&info, &ka, regs, NULL);
549 if (signr > 0) {
550 /* Whee! Actually deliver the signal. */
551 if (TRAP(regs) == 0x0C00)
552 syscall_restart(regs, &ka);
553
554 /*
555 * Reenable the DABR before delivering the signal to
556 * user space. The DABR will have been cleared if it
557 * triggered inside the kernel.
558 */
559 if (current->thread.dabr)
560 set_dabr(current->thread.dabr);
561
562 return handle_signal(signr, &ka, &info, oldset, regs);
563 }
564
565 if (TRAP(regs) == 0x0C00) { /* System Call! */
566 if ((int)regs->result == -ERESTARTNOHAND ||
567 (int)regs->result == -ERESTARTSYS ||
568 (int)regs->result == -ERESTARTNOINTR) {
569 regs->gpr[3] = regs->orig_gpr3;
570 regs->nip -= 4; /* Back up & retry system call */
571 regs->result = 0;
572 } else if ((int)regs->result == -ERESTART_RESTARTBLOCK) {
573 regs->gpr[0] = __NR_restart_syscall;
574 regs->nip -= 4;
575 regs->result = 0;
576 }
577 }
578
579 return 0;
580}
581EXPORT_SYMBOL(do_signal);
diff --git a/arch/ppc64/kernel/smp-tbsync.c b/arch/ppc64/kernel/smp-tbsync.c
deleted file mode 100644
index 7d8ec9996b3e..000000000000
--- a/arch/ppc64/kernel/smp-tbsync.c
+++ /dev/null
@@ -1,179 +0,0 @@
1/*
2 * Smp timebase synchronization for ppc.
3 *
4 * Copyright (C) 2003 Samuel Rydh (samuel@ibrium.se)
5 *
6 */
7
8#include <linux/config.h>
9#include <linux/kernel.h>
10#include <linux/sched.h>
11#include <linux/smp.h>
12#include <linux/unistd.h>
13#include <linux/init.h>
14#include <asm/atomic.h>
15#include <asm/smp.h>
16#include <asm/time.h>
17
18#define NUM_ITER 300
19
20enum {
21 kExit=0, kSetAndTest, kTest
22};
23
24static struct {
25 volatile long tb;
26 volatile long mark;
27 volatile int cmd;
28 volatile int handshake;
29 int filler[3];
30
31 volatile int ack;
32 int filler2[7];
33
34 volatile int race_result;
35} *tbsync;
36
37static volatile int running;
38
39static void __devinit
40enter_contest( long mark, long add )
41{
42 while( (long)(mftb() - mark) < 0 )
43 tbsync->race_result = add;
44}
45
46void __devinit
47smp_generic_take_timebase( void )
48{
49 int cmd;
50 long tb;
51
52 local_irq_disable();
53 while( !running )
54 ;
55 rmb();
56
57 for( ;; ) {
58 tbsync->ack = 1;
59 while( !tbsync->handshake )
60 ;
61 rmb();
62
63 cmd = tbsync->cmd;
64 tb = tbsync->tb;
65 tbsync->ack = 0;
66 if( cmd == kExit )
67 return;
68
69 if( cmd == kSetAndTest ) {
70 while( tbsync->handshake )
71 ;
72 asm volatile ("mttbl %0" :: "r" (tb & 0xfffffffful) );
73 asm volatile ("mttbu %0" :: "r" (tb >> 32) );
74 } else {
75 while( tbsync->handshake )
76 ;
77 }
78 enter_contest( tbsync->mark, -1 );
79 }
80 local_irq_enable();
81}
82
83static int __devinit
84start_contest( int cmd, long offset, long num )
85{
86 int i, score=0;
87 long tb, mark;
88
89 tbsync->cmd = cmd;
90
91 local_irq_disable();
92 for( i=-3; i<num; ) {
93 tb = (long)mftb() + 400;
94 tbsync->tb = tb + offset;
95 tbsync->mark = mark = tb + 400;
96
97 wmb();
98
99 tbsync->handshake = 1;
100 while( tbsync->ack )
101 ;
102
103 while( (long)(mftb() - tb) <= 0 )
104 ;
105 tbsync->handshake = 0;
106 enter_contest( mark, 1 );
107
108 while( !tbsync->ack )
109 ;
110
111 if ((tbsync->tb ^ (long)mftb()) & 0x8000000000000000ul)
112 continue;
113 if( i++ > 0 )
114 score += tbsync->race_result;
115 }
116 local_irq_enable();
117 return score;
118}
119
120void __devinit
121smp_generic_give_timebase( void )
122{
123 int i, score, score2, old, min=0, max=5000, offset=1000;
124
125 printk("Synchronizing timebase\n");
126
127 /* if this fails then this kernel won't work anyway... */
128 tbsync = kmalloc( sizeof(*tbsync), GFP_KERNEL );
129 memset( tbsync, 0, sizeof(*tbsync) );
130 mb();
131 running = 1;
132
133 while( !tbsync->ack )
134 ;
135
136 printk("Got ack\n");
137
138 /* binary search */
139 for( old=-1 ; old != offset ; offset=(min+max)/2 ) {
140 score = start_contest( kSetAndTest, offset, NUM_ITER );
141
142 printk("score %d, offset %d\n", score, offset );
143
144 if( score > 0 )
145 max = offset;
146 else
147 min = offset;
148 old = offset;
149 }
150 score = start_contest( kSetAndTest, min, NUM_ITER );
151 score2 = start_contest( kSetAndTest, max, NUM_ITER );
152
153 printk( "Min %d (score %d), Max %d (score %d)\n", min, score, max, score2 );
154 score = abs( score );
155 score2 = abs( score2 );
156 offset = (score < score2) ? min : max;
157
158 /* guard against inaccurate mttb */
159 for( i=0; i<10; i++ ) {
160 start_contest( kSetAndTest, offset, NUM_ITER/10 );
161
162 if( (score2=start_contest(kTest, offset, NUM_ITER)) < 0 )
163 score2 = -score2;
164 if( score2 <= score || score2 < 20 )
165 break;
166 }
167 printk("Final offset: %d (%d/%d)\n", offset, score2, NUM_ITER );
168
169 /* exiting */
170 tbsync->cmd = kExit;
171 wmb();
172 tbsync->handshake = 1;
173 while( tbsync->ack )
174 ;
175 tbsync->handshake = 0;
176 kfree( tbsync );
177 tbsync = NULL;
178 running = 0;
179}
diff --git a/arch/ppc64/kernel/smp.c b/arch/ppc64/kernel/smp.c
deleted file mode 100644
index 017c12919832..000000000000
--- a/arch/ppc64/kernel/smp.c
+++ /dev/null
@@ -1,568 +0,0 @@
1/*
2 * SMP support for ppc.
3 *
4 * Written by Cort Dougan (cort@cs.nmt.edu) borrowing a great
5 * deal of code from the sparc and intel versions.
6 *
7 * Copyright (C) 1999 Cort Dougan <cort@cs.nmt.edu>
8 *
9 * PowerPC-64 Support added by Dave Engebretsen, Peter Bergner, and
10 * Mike Corrigan {engebret|bergner|mikec}@us.ibm.com
11 *
12 * This program is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU General Public License
14 * as published by the Free Software Foundation; either version
15 * 2 of the License, or (at your option) any later version.
16 */
17
18#undef DEBUG
19
20#include <linux/config.h>
21#include <linux/kernel.h>
22#include <linux/module.h>
23#include <linux/sched.h>
24#include <linux/smp.h>
25#include <linux/interrupt.h>
26#include <linux/delay.h>
27#include <linux/init.h>
28#include <linux/spinlock.h>
29#include <linux/cache.h>
30#include <linux/err.h>
31#include <linux/sysdev.h>
32#include <linux/cpu.h>
33#include <linux/notifier.h>
34
35#include <asm/ptrace.h>
36#include <asm/atomic.h>
37#include <asm/irq.h>
38#include <asm/page.h>
39#include <asm/pgtable.h>
40#include <asm/prom.h>
41#include <asm/smp.h>
42#include <asm/paca.h>
43#include <asm/time.h>
44#include <asm/machdep.h>
45#include <asm/cputable.h>
46#include <asm/system.h>
47#include <asm/abs_addr.h>
48#include <asm/mpic.h>
49
50#ifdef DEBUG
51#define DBG(fmt...) udbg_printf(fmt)
52#else
53#define DBG(fmt...)
54#endif
55
56cpumask_t cpu_possible_map = CPU_MASK_NONE;
57cpumask_t cpu_online_map = CPU_MASK_NONE;
58cpumask_t cpu_sibling_map[NR_CPUS] = { [0 ... NR_CPUS-1] = CPU_MASK_NONE };
59
60EXPORT_SYMBOL(cpu_online_map);
61EXPORT_SYMBOL(cpu_possible_map);
62
63struct smp_ops_t *smp_ops;
64
65static volatile unsigned int cpu_callin_map[NR_CPUS];
66
67void smp_call_function_interrupt(void);
68
69int smt_enabled_at_boot = 1;
70
71#ifdef CONFIG_MPIC
72int __init smp_mpic_probe(void)
73{
74 int nr_cpus;
75
76 DBG("smp_mpic_probe()...\n");
77
78 nr_cpus = cpus_weight(cpu_possible_map);
79
80 DBG("nr_cpus: %d\n", nr_cpus);
81
82 if (nr_cpus > 1)
83 mpic_request_ipis();
84
85 return nr_cpus;
86}
87
88void __devinit smp_mpic_setup_cpu(int cpu)
89{
90 mpic_setup_this_cpu();
91}
92
93void __devinit smp_generic_kick_cpu(int nr)
94{
95 BUG_ON(nr < 0 || nr >= NR_CPUS);
96
97 /*
98 * The processor is currently spinning, waiting for the
99 * cpu_start field to become non-zero After we set cpu_start,
100 * the processor will continue on to secondary_start
101 */
102 paca[nr].cpu_start = 1;
103 smp_mb();
104}
105
106#endif /* CONFIG_MPIC */
107
108void smp_message_recv(int msg, struct pt_regs *regs)
109{
110 switch(msg) {
111 case PPC_MSG_CALL_FUNCTION:
112 smp_call_function_interrupt();
113 break;
114 case PPC_MSG_RESCHEDULE:
115 /* XXX Do we have to do this? */
116 set_need_resched();
117 break;
118#if 0
119 case PPC_MSG_MIGRATE_TASK:
120 /* spare */
121 break;
122#endif
123#ifdef CONFIG_DEBUGGER
124 case PPC_MSG_DEBUGGER_BREAK:
125 debugger_ipi(regs);
126 break;
127#endif
128 default:
129 printk("SMP %d: smp_message_recv(): unknown msg %d\n",
130 smp_processor_id(), msg);
131 break;
132 }
133}
134
135void smp_send_reschedule(int cpu)
136{
137 smp_ops->message_pass(cpu, PPC_MSG_RESCHEDULE);
138}
139
140#ifdef CONFIG_DEBUGGER
141void smp_send_debugger_break(int cpu)
142{
143 smp_ops->message_pass(cpu, PPC_MSG_DEBUGGER_BREAK);
144}
145#endif
146
147static void stop_this_cpu(void *dummy)
148{
149 local_irq_disable();
150 while (1)
151 ;
152}
153
154void smp_send_stop(void)
155{
156 smp_call_function(stop_this_cpu, NULL, 1, 0);
157}
158
159/*
160 * Structure and data for smp_call_function(). This is designed to minimise
161 * static memory requirements. It also looks cleaner.
162 * Stolen from the i386 version.
163 */
164static __cacheline_aligned_in_smp DEFINE_SPINLOCK(call_lock);
165
166static struct call_data_struct {
167 void (*func) (void *info);
168 void *info;
169 atomic_t started;
170 atomic_t finished;
171 int wait;
172} *call_data;
173
174/* delay of at least 8 seconds on 1GHz cpu */
175#define SMP_CALL_TIMEOUT (1UL << (30 + 3))
176
177/*
178 * This function sends a 'generic call function' IPI to all other CPUs
179 * in the system.
180 *
181 * [SUMMARY] Run a function on all other CPUs.
182 * <func> The function to run. This must be fast and non-blocking.
183 * <info> An arbitrary pointer to pass to the function.
184 * <nonatomic> currently unused.
185 * <wait> If true, wait (atomically) until function has completed on other CPUs.
186 * [RETURNS] 0 on success, else a negative status code. Does not return until
187 * remote CPUs are nearly ready to execute <<func>> or are or have executed.
188 *
189 * You must not call this function with disabled interrupts or from a
190 * hardware interrupt handler or from a bottom half handler.
191 */
192int smp_call_function (void (*func) (void *info), void *info, int nonatomic,
193 int wait)
194{
195 struct call_data_struct data;
196 int ret = -1, cpus;
197 unsigned long timeout;
198
199 /* Can deadlock when called with interrupts disabled */
200 WARN_ON(irqs_disabled());
201
202 data.func = func;
203 data.info = info;
204 atomic_set(&data.started, 0);
205 data.wait = wait;
206 if (wait)
207 atomic_set(&data.finished, 0);
208
209 spin_lock(&call_lock);
210 /* Must grab online cpu count with preempt disabled, otherwise
211 * it can change. */
212 cpus = num_online_cpus() - 1;
213 if (!cpus) {
214 ret = 0;
215 goto out;
216 }
217
218 call_data = &data;
219 smp_wmb();
220 /* Send a message to all other CPUs and wait for them to respond */
221 smp_ops->message_pass(MSG_ALL_BUT_SELF, PPC_MSG_CALL_FUNCTION);
222
223 /* Wait for response */
224 timeout = SMP_CALL_TIMEOUT;
225 while (atomic_read(&data.started) != cpus) {
226 HMT_low();
227 if (--timeout == 0) {
228 printk("smp_call_function on cpu %d: other cpus not "
229 "responding (%d)\n", smp_processor_id(),
230 atomic_read(&data.started));
231 debugger(NULL);
232 goto out;
233 }
234 }
235
236 if (wait) {
237 timeout = SMP_CALL_TIMEOUT;
238 while (atomic_read(&data.finished) != cpus) {
239 HMT_low();
240 if (--timeout == 0) {
241 printk("smp_call_function on cpu %d: other "
242 "cpus not finishing (%d/%d)\n",
243 smp_processor_id(),
244 atomic_read(&data.finished),
245 atomic_read(&data.started));
246 debugger(NULL);
247 goto out;
248 }
249 }
250 }
251
252 ret = 0;
253
254out:
255 call_data = NULL;
256 HMT_medium();
257 spin_unlock(&call_lock);
258 return ret;
259}
260
261EXPORT_SYMBOL(smp_call_function);
262
263void smp_call_function_interrupt(void)
264{
265 void (*func) (void *info);
266 void *info;
267 int wait;
268
269 /* call_data will be NULL if the sender timed out while
270 * waiting on us to receive the call.
271 */
272 if (!call_data)
273 return;
274
275 func = call_data->func;
276 info = call_data->info;
277 wait = call_data->wait;
278
279 if (!wait)
280 smp_mb__before_atomic_inc();
281
282 /*
283 * Notify initiating CPU that I've grabbed the data and am
284 * about to execute the function
285 */
286 atomic_inc(&call_data->started);
287 /*
288 * At this point the info structure may be out of scope unless wait==1
289 */
290 (*func)(info);
291 if (wait) {
292 smp_mb__before_atomic_inc();
293 atomic_inc(&call_data->finished);
294 }
295}
296
297extern struct gettimeofday_struct do_gtod;
298
299struct thread_info *current_set[NR_CPUS];
300
301DECLARE_PER_CPU(unsigned int, pvr);
302
303static void __devinit smp_store_cpu_info(int id)
304{
305 per_cpu(pvr, id) = mfspr(SPRN_PVR);
306}
307
308static void __init smp_create_idle(unsigned int cpu)
309{
310 struct task_struct *p;
311
312 /* create a process for the processor */
313 p = fork_idle(cpu);
314 if (IS_ERR(p))
315 panic("failed fork for CPU %u: %li", cpu, PTR_ERR(p));
316 paca[cpu].__current = p;
317 current_set[cpu] = p->thread_info;
318}
319
320void __init smp_prepare_cpus(unsigned int max_cpus)
321{
322 unsigned int cpu;
323
324 DBG("smp_prepare_cpus\n");
325
326 /*
327 * setup_cpu may need to be called on the boot cpu. We havent
328 * spun any cpus up but lets be paranoid.
329 */
330 BUG_ON(boot_cpuid != smp_processor_id());
331
332 /* Fixup boot cpu */
333 smp_store_cpu_info(boot_cpuid);
334 cpu_callin_map[boot_cpuid] = 1;
335
336#ifndef CONFIG_PPC_ISERIES
337 paca[boot_cpuid].next_jiffy_update_tb = tb_last_stamp = get_tb();
338
339 /*
340 * Should update do_gtod.stamp_xsec.
341 * For now we leave it which means the time can be some
342 * number of msecs off until someone does a settimeofday()
343 */
344 do_gtod.varp->tb_orig_stamp = tb_last_stamp;
345 systemcfg->tb_orig_stamp = tb_last_stamp;
346#endif
347
348 max_cpus = smp_ops->probe();
349
350 smp_space_timers(max_cpus);
351
352 for_each_cpu(cpu)
353 if (cpu != boot_cpuid)
354 smp_create_idle(cpu);
355}
356
357void __devinit smp_prepare_boot_cpu(void)
358{
359 BUG_ON(smp_processor_id() != boot_cpuid);
360
361 cpu_set(boot_cpuid, cpu_online_map);
362
363 paca[boot_cpuid].__current = current;
364 current_set[boot_cpuid] = current->thread_info;
365}
366
367#ifdef CONFIG_HOTPLUG_CPU
368/* State of each CPU during hotplug phases */
369DEFINE_PER_CPU(int, cpu_state) = { 0 };
370
371int generic_cpu_disable(void)
372{
373 unsigned int cpu = smp_processor_id();
374
375 if (cpu == boot_cpuid)
376 return -EBUSY;
377
378 systemcfg->processorCount--;
379 cpu_clear(cpu, cpu_online_map);
380 fixup_irqs(cpu_online_map);
381 return 0;
382}
383
384int generic_cpu_enable(unsigned int cpu)
385{
386 /* Do the normal bootup if we haven't
387 * already bootstrapped. */
388 if (system_state != SYSTEM_RUNNING)
389 return -ENOSYS;
390
391 /* get the target out of it's holding state */
392 per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
393 smp_wmb();
394
395 while (!cpu_online(cpu))
396 cpu_relax();
397
398 fixup_irqs(cpu_online_map);
399 /* counter the irq disable in fixup_irqs */
400 local_irq_enable();
401 return 0;
402}
403
404void generic_cpu_die(unsigned int cpu)
405{
406 int i;
407
408 for (i = 0; i < 100; i++) {
409 smp_rmb();
410 if (per_cpu(cpu_state, cpu) == CPU_DEAD)
411 return;
412 msleep(100);
413 }
414 printk(KERN_ERR "CPU%d didn't die...\n", cpu);
415}
416
417void generic_mach_cpu_die(void)
418{
419 unsigned int cpu;
420
421 local_irq_disable();
422 cpu = smp_processor_id();
423 printk(KERN_DEBUG "CPU%d offline\n", cpu);
424 __get_cpu_var(cpu_state) = CPU_DEAD;
425 smp_wmb();
426 while (__get_cpu_var(cpu_state) != CPU_UP_PREPARE)
427 cpu_relax();
428
429 flush_tlb_pending();
430 cpu_set(cpu, cpu_online_map);
431 local_irq_enable();
432}
433#endif
434
435static int __devinit cpu_enable(unsigned int cpu)
436{
437 if (smp_ops->cpu_enable)
438 return smp_ops->cpu_enable(cpu);
439
440 return -ENOSYS;
441}
442
443int __devinit __cpu_up(unsigned int cpu)
444{
445 int c;
446
447 if (!cpu_enable(cpu))
448 return 0;
449
450 if (smp_ops->cpu_bootable && !smp_ops->cpu_bootable(cpu))
451 return -EINVAL;
452
453 paca[cpu].default_decr = tb_ticks_per_jiffy;
454
455 /* Make sure callin-map entry is 0 (can be leftover a CPU
456 * hotplug
457 */
458 cpu_callin_map[cpu] = 0;
459
460 /* The information for processor bringup must
461 * be written out to main store before we release
462 * the processor.
463 */
464 smp_mb();
465
466 /* wake up cpus */
467 DBG("smp: kicking cpu %d\n", cpu);
468 smp_ops->kick_cpu(cpu);
469
470 /*
471 * wait to see if the cpu made a callin (is actually up).
472 * use this value that I found through experimentation.
473 * -- Cort
474 */
475 if (system_state < SYSTEM_RUNNING)
476 for (c = 5000; c && !cpu_callin_map[cpu]; c--)
477 udelay(100);
478#ifdef CONFIG_HOTPLUG_CPU
479 else
480 /*
481 * CPUs can take much longer to come up in the
482 * hotplug case. Wait five seconds.
483 */
484 for (c = 25; c && !cpu_callin_map[cpu]; c--) {
485 msleep(200);
486 }
487#endif
488
489 if (!cpu_callin_map[cpu]) {
490 printk("Processor %u is stuck.\n", cpu);
491 return -ENOENT;
492 }
493
494 printk("Processor %u found.\n", cpu);
495
496 if (smp_ops->give_timebase)
497 smp_ops->give_timebase();
498
499 /* Wait until cpu puts itself in the online map */
500 while (!cpu_online(cpu))
501 cpu_relax();
502
503 return 0;
504}
505
506
507/* Activate a secondary processor. */
508int __devinit start_secondary(void *unused)
509{
510 unsigned int cpu = smp_processor_id();
511
512 atomic_inc(&init_mm.mm_count);
513 current->active_mm = &init_mm;
514
515 smp_store_cpu_info(cpu);
516 set_dec(paca[cpu].default_decr);
517 cpu_callin_map[cpu] = 1;
518
519 smp_ops->setup_cpu(cpu);
520 if (smp_ops->take_timebase)
521 smp_ops->take_timebase();
522
523 spin_lock(&call_lock);
524 cpu_set(cpu, cpu_online_map);
525 spin_unlock(&call_lock);
526
527 local_irq_enable();
528
529 cpu_idle();
530 return 0;
531}
532
533int setup_profiling_timer(unsigned int multiplier)
534{
535 return 0;
536}
537
538void __init smp_cpus_done(unsigned int max_cpus)
539{
540 cpumask_t old_mask;
541
542 /* We want the setup_cpu() here to be called from CPU 0, but our
543 * init thread may have been "borrowed" by another CPU in the meantime
544 * se we pin us down to CPU 0 for a short while
545 */
546 old_mask = current->cpus_allowed;
547 set_cpus_allowed(current, cpumask_of_cpu(boot_cpuid));
548
549 smp_ops->setup_cpu(boot_cpuid);
550
551 set_cpus_allowed(current, old_mask);
552}
553
554#ifdef CONFIG_HOTPLUG_CPU
555int __cpu_disable(void)
556{
557 if (smp_ops->cpu_disable)
558 return smp_ops->cpu_disable();
559
560 return -ENOSYS;
561}
562
563void __cpu_die(unsigned int cpu)
564{
565 if (smp_ops->cpu_die)
566 smp_ops->cpu_die(cpu);
567}
568#endif
diff --git a/arch/ppc64/kernel/spider-pic.c b/arch/ppc64/kernel/spider-pic.c
deleted file mode 100644
index d5c9a02fb119..000000000000
--- a/arch/ppc64/kernel/spider-pic.c
+++ /dev/null
@@ -1,191 +0,0 @@
1/*
2 * External Interrupt Controller on Spider South Bridge
3 *
4 * (C) Copyright IBM Deutschland Entwicklung GmbH 2005
5 *
6 * Author: Arnd Bergmann <arndb@de.ibm.com>
7 *
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2, or (at your option)
11 * any later version.
12 *
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to the Free Software
20 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
21 */
22
23#include <linux/interrupt.h>
24#include <linux/irq.h>
25
26#include <asm/pgtable.h>
27#include <asm/prom.h>
28#include <asm/io.h>
29
30#include "bpa_iic.h"
31
32/* register layout taken from Spider spec, table 7.4-4 */
33enum {
34 TIR_DEN = 0x004, /* Detection Enable Register */
35 TIR_MSK = 0x084, /* Mask Level Register */
36 TIR_EDC = 0x0c0, /* Edge Detection Clear Register */
37 TIR_PNDA = 0x100, /* Pending Register A */
38 TIR_PNDB = 0x104, /* Pending Register B */
39 TIR_CS = 0x144, /* Current Status Register */
40 TIR_LCSA = 0x150, /* Level Current Status Register A */
41 TIR_LCSB = 0x154, /* Level Current Status Register B */
42 TIR_LCSC = 0x158, /* Level Current Status Register C */
43 TIR_LCSD = 0x15c, /* Level Current Status Register D */
44 TIR_CFGA = 0x200, /* Setting Register A0 */
45 TIR_CFGB = 0x204, /* Setting Register B0 */
46 /* 0x208 ... 0x3ff Setting Register An/Bn */
47 TIR_PPNDA = 0x400, /* Packet Pending Register A */
48 TIR_PPNDB = 0x404, /* Packet Pending Register B */
49 TIR_PIERA = 0x408, /* Packet Output Error Register A */
50 TIR_PIERB = 0x40c, /* Packet Output Error Register B */
51 TIR_PIEN = 0x444, /* Packet Output Enable Register */
52 TIR_PIPND = 0x454, /* Packet Output Pending Register */
53 TIRDID = 0x484, /* Spider Device ID Register */
54 REISTIM = 0x500, /* Reissue Command Timeout Time Setting */
55 REISTIMEN = 0x504, /* Reissue Command Timeout Setting */
56 REISWAITEN = 0x508, /* Reissue Wait Control*/
57};
58
59static void __iomem *spider_pics[4];
60
61static void __iomem *spider_get_pic(int irq)
62{
63 int node = irq / IIC_NODE_STRIDE;
64 irq %= IIC_NODE_STRIDE;
65
66 if (irq >= IIC_EXT_OFFSET &&
67 irq < IIC_EXT_OFFSET + IIC_NUM_EXT &&
68 spider_pics)
69 return spider_pics[node];
70 return NULL;
71}
72
73static int spider_get_nr(unsigned int irq)
74{
75 return (irq % IIC_NODE_STRIDE) - IIC_EXT_OFFSET;
76}
77
78static void __iomem *spider_get_irq_config(int irq)
79{
80 void __iomem *pic;
81 pic = spider_get_pic(irq);
82 return pic + TIR_CFGA + 8 * spider_get_nr(irq);
83}
84
85static void spider_enable_irq(unsigned int irq)
86{
87 void __iomem *cfg = spider_get_irq_config(irq);
88 irq = spider_get_nr(irq);
89
90 out_be32(cfg, in_be32(cfg) | 0x3107000eu);
91 out_be32(cfg + 4, in_be32(cfg + 4) | 0x00020000u | irq);
92}
93
94static void spider_disable_irq(unsigned int irq)
95{
96 void __iomem *cfg = spider_get_irq_config(irq);
97 irq = spider_get_nr(irq);
98
99 out_be32(cfg, in_be32(cfg) & ~0x30000000u);
100}
101
102static unsigned int spider_startup_irq(unsigned int irq)
103{
104 spider_enable_irq(irq);
105 return 0;
106}
107
108static void spider_shutdown_irq(unsigned int irq)
109{
110 spider_disable_irq(irq);
111}
112
113static void spider_end_irq(unsigned int irq)
114{
115 spider_enable_irq(irq);
116}
117
118static void spider_ack_irq(unsigned int irq)
119{
120 spider_disable_irq(irq);
121 iic_local_enable();
122}
123
124static struct hw_interrupt_type spider_pic = {
125 .typename = " SPIDER ",
126 .startup = spider_startup_irq,
127 .shutdown = spider_shutdown_irq,
128 .enable = spider_enable_irq,
129 .disable = spider_disable_irq,
130 .ack = spider_ack_irq,
131 .end = spider_end_irq,
132};
133
134
135int spider_get_irq(unsigned long int_pending)
136{
137 void __iomem *regs = spider_get_pic(int_pending);
138 unsigned long cs;
139 int irq;
140
141 cs = in_be32(regs + TIR_CS);
142
143 irq = cs >> 24;
144 if (irq != 63)
145 return irq;
146
147 return -1;
148}
149
150void spider_init_IRQ(void)
151{
152 int node;
153 struct device_node *dn;
154 unsigned int *property;
155 long spiderpic;
156 int n;
157
158/* FIXME: detect multiple PICs as soon as the device tree has them */
159 for (node = 0; node < 1; node++) {
160 dn = of_find_node_by_path("/");
161 n = prom_n_addr_cells(dn);
162 property = (unsigned int *) get_property(dn,
163 "platform-spider-pic", NULL);
164
165 if (!property)
166 continue;
167 for (spiderpic = 0; n > 0; --n)
168 spiderpic = (spiderpic << 32) + *property++;
169 printk(KERN_DEBUG "SPIDER addr: %lx\n", spiderpic);
170 spider_pics[node] = __ioremap(spiderpic, 0x800, _PAGE_NO_CACHE);
171 for (n = 0; n < IIC_NUM_EXT; n++) {
172 int irq = n + IIC_EXT_OFFSET + node * IIC_NODE_STRIDE;
173 get_irq_desc(irq)->handler = &spider_pic;
174
175 /* do not mask any interrupts because of level */
176 out_be32(spider_pics[node] + TIR_MSK, 0x0);
177
178 /* disable edge detection clear */
179 /* out_be32(spider_pics[node] + TIR_EDC, 0x0); */
180
181 /* enable interrupt packets to be output */
182 out_be32(spider_pics[node] + TIR_PIEN,
183 in_be32(spider_pics[node] + TIR_PIEN) | 0x1);
184
185 /* Enable the interrupt detection enable bit. Do this last! */
186 out_be32(spider_pics[node] + TIR_DEN,
187 in_be32(spider_pics[node] +TIR_DEN) | 0x1);
188
189 }
190 }
191}
diff --git a/arch/ppc64/kernel/sysfs.c b/arch/ppc64/kernel/sysfs.c
index 6654b350979c..e99ec62c2c52 100644
--- a/arch/ppc64/kernel/sysfs.c
+++ b/arch/ppc64/kernel/sysfs.c
@@ -20,6 +20,7 @@
20#include <asm/paca.h> 20#include <asm/paca.h>
21#include <asm/lppaca.h> 21#include <asm/lppaca.h>
22#include <asm/machdep.h> 22#include <asm/machdep.h>
23#include <asm/smp.h>
23 24
24static DEFINE_PER_CPU(struct cpu, cpu_devices); 25static DEFINE_PER_CPU(struct cpu, cpu_devices);
25 26
diff --git a/arch/ppc64/kernel/udbg_scc.c b/arch/ppc64/kernel/udbg_scc.c
index c47fd6c63531..820c53551507 100644
--- a/arch/ppc64/kernel/udbg_scc.c
+++ b/arch/ppc64/kernel/udbg_scc.c
@@ -12,7 +12,6 @@
12#include <linux/types.h> 12#include <linux/types.h>
13#include <asm/udbg.h> 13#include <asm/udbg.h>
14#include <asm/processor.h> 14#include <asm/processor.h>
15#include <asm/naca.h>
16#include <asm/io.h> 15#include <asm/io.h>
17#include <asm/prom.h> 16#include <asm/prom.h>
18#include <asm/pmac_feature.h> 17#include <asm/pmac_feature.h>