aboutsummaryrefslogtreecommitdiffstats
path: root/arch
diff options
context:
space:
mode:
Diffstat (limited to 'arch')
-rw-r--r--arch/powerpc/Kconfig3
-rw-r--r--arch/powerpc/boot/Makefile1
-rw-r--r--arch/powerpc/include/asm/abs_addr.h56
-rw-r--r--arch/powerpc/include/asm/debug.h2
-rw-r--r--arch/powerpc/include/asm/eeh.h136
-rw-r--r--arch/powerpc/include/asm/eeh_event.h6
-rw-r--r--arch/powerpc/include/asm/exception-64e.h6
-rw-r--r--arch/powerpc/include/asm/hvcall.h5
-rw-r--r--arch/powerpc/include/asm/hw_breakpoint.h9
-rw-r--r--arch/powerpc/include/asm/kprobes.h15
-rw-r--r--arch/powerpc/include/asm/kvm_book3s_asm.h1
-rw-r--r--arch/powerpc/include/asm/machdep.h3
-rw-r--r--arch/powerpc/include/asm/paca.h1
-rw-r--r--arch/powerpc/include/asm/pci-bridge.h2
-rw-r--r--arch/powerpc/include/asm/ppc-pci.h20
-rw-r--r--arch/powerpc/include/asm/probes.h42
-rw-r--r--arch/powerpc/include/asm/processor.h2
-rw-r--r--arch/powerpc/include/asm/reg.h49
-rw-r--r--arch/powerpc/include/asm/setup.h2
-rw-r--r--arch/powerpc/include/asm/thread_info.h4
-rw-r--r--arch/powerpc/include/asm/uprobes.h54
-rw-r--r--arch/powerpc/kernel/Makefile1
-rw-r--r--arch/powerpc/kernel/asm-offsets.c2
-rw-r--r--arch/powerpc/kernel/dma-swiotlb.c2
-rw-r--r--arch/powerpc/kernel/dma.c3
-rw-r--r--arch/powerpc/kernel/exceptions-64e.S194
-rw-r--r--arch/powerpc/kernel/fadump.c3
-rw-r--r--arch/powerpc/kernel/hw_breakpoint.c25
-rw-r--r--arch/powerpc/kernel/ibmebus.c1
-rw-r--r--arch/powerpc/kernel/machine_kexec.c14
-rw-r--r--arch/powerpc/kernel/paca.c1
-rw-r--r--arch/powerpc/kernel/pci-common.c16
-rw-r--r--arch/powerpc/kernel/process.c16
-rw-r--r--arch/powerpc/kernel/prom.c4
-rw-r--r--arch/powerpc/kernel/prom_init.c2
-rw-r--r--arch/powerpc/kernel/ptrace.c3
-rw-r--r--arch/powerpc/kernel/rtas_flash.c7
-rw-r--r--arch/powerpc/kernel/rtas_pci.c5
-rw-r--r--arch/powerpc/kernel/signal.c8
-rw-r--r--arch/powerpc/kernel/traps.c1
-rw-r--r--arch/powerpc/kernel/uprobes.c184
-rw-r--r--arch/powerpc/kernel/vdso.c4
-rw-r--r--arch/powerpc/kernel/vio.c1
-rw-r--r--arch/powerpc/kvm/book3s_hv_rmhandlers.S2
-rw-r--r--arch/powerpc/mm/fault.c1
-rw-r--r--arch/powerpc/mm/hash_native_64.c2
-rw-r--r--arch/powerpc/mm/hash_utils_64.c3
-rw-r--r--arch/powerpc/mm/init_64.c1
-rw-r--r--arch/powerpc/mm/mem.c2
-rw-r--r--arch/powerpc/mm/pgtable_64.c1
-rw-r--r--arch/powerpc/mm/stab.c3
-rw-r--r--arch/powerpc/mm/subpage-prot.c6
-rw-r--r--arch/powerpc/mm/tlb_low_64e.S18
-rw-r--r--arch/powerpc/oprofile/op_model_power4.c116
-rw-r--r--arch/powerpc/perf/core-book3s.c2
-rw-r--r--arch/powerpc/platforms/cell/beat.c4
-rw-r--r--arch/powerpc/platforms/cell/beat.h2
-rw-r--r--arch/powerpc/platforms/pasemi/iommu.c6
-rw-r--r--arch/powerpc/platforms/powernv/pci-ioda.c1
-rw-r--r--arch/powerpc/platforms/powernv/pci-p5ioc2.c1
-rw-r--r--arch/powerpc/platforms/powernv/pci.c7
-rw-r--r--arch/powerpc/platforms/ps3/setup.c10
-rw-r--r--arch/powerpc/platforms/pseries/Makefile5
-rw-r--r--arch/powerpc/platforms/pseries/eeh.c531
-rw-r--r--arch/powerpc/platforms/pseries/eeh_cache.c59
-rw-r--r--arch/powerpc/platforms/pseries/eeh_dev.c14
-rw-r--r--arch/powerpc/platforms/pseries/eeh_driver.c235
-rw-r--r--arch/powerpc/platforms/pseries/eeh_event.c54
-rw-r--r--arch/powerpc/platforms/pseries/eeh_pe.c591
-rw-r--r--arch/powerpc/platforms/pseries/eeh_pseries.c247
-rw-r--r--arch/powerpc/platforms/pseries/eeh_sysfs.c9
-rw-r--r--arch/powerpc/platforms/pseries/iommu.c12
-rw-r--r--arch/powerpc/platforms/pseries/lpar.c1
-rw-r--r--arch/powerpc/platforms/pseries/msi.c26
-rw-r--r--arch/powerpc/platforms/pseries/pci.c2
-rw-r--r--arch/powerpc/platforms/pseries/setup.c22
-rw-r--r--arch/powerpc/sysdev/dart_iommu.c7
-rw-r--r--arch/powerpc/xmon/xmon.c4
78 files changed, 1877 insertions, 1046 deletions
diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index 352f416269ce..98e513b62709 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -239,6 +239,9 @@ config PPC_OF_PLATFORM_PCI
239config ARCH_SUPPORTS_DEBUG_PAGEALLOC 239config ARCH_SUPPORTS_DEBUG_PAGEALLOC
240 def_bool y 240 def_bool y
241 241
242config ARCH_SUPPORTS_UPROBES
243 def_bool y
244
242config PPC_ADV_DEBUG_REGS 245config PPC_ADV_DEBUG_REGS
243 bool 246 bool
244 depends on 40x || BOOKE 247 depends on 40x || BOOKE
diff --git a/arch/powerpc/boot/Makefile b/arch/powerpc/boot/Makefile
index b7d833382be4..6a15c968d214 100644
--- a/arch/powerpc/boot/Makefile
+++ b/arch/powerpc/boot/Makefile
@@ -107,6 +107,7 @@ src-boot := $(addprefix $(obj)/, $(src-boot))
107obj-boot := $(addsuffix .o, $(basename $(src-boot))) 107obj-boot := $(addsuffix .o, $(basename $(src-boot)))
108obj-wlib := $(addsuffix .o, $(basename $(addprefix $(obj)/, $(src-wlib)))) 108obj-wlib := $(addsuffix .o, $(basename $(addprefix $(obj)/, $(src-wlib))))
109obj-plat := $(addsuffix .o, $(basename $(addprefix $(obj)/, $(src-plat)))) 109obj-plat := $(addsuffix .o, $(basename $(addprefix $(obj)/, $(src-plat))))
110obj-plat: $(libfdt)
110 111
111quiet_cmd_copy_zlib = COPY $@ 112quiet_cmd_copy_zlib = COPY $@
112 cmd_copy_zlib = sed "s@__used@@;s@<linux/\([^>]*\).*@\"\1\"@" $< > $@ 113 cmd_copy_zlib = sed "s@__used@@;s@<linux/\([^>]*\).*@\"\1\"@" $< > $@
diff --git a/arch/powerpc/include/asm/abs_addr.h b/arch/powerpc/include/asm/abs_addr.h
deleted file mode 100644
index 9d92ba04b033..000000000000
--- a/arch/powerpc/include/asm/abs_addr.h
+++ /dev/null
@@ -1,56 +0,0 @@
1#ifndef _ASM_POWERPC_ABS_ADDR_H
2#define _ASM_POWERPC_ABS_ADDR_H
3#ifdef __KERNEL__
4
5
6/*
7 * c 2001 PPC 64 Team, IBM Corp
8 *
9 * This program is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU General Public License
11 * as published by the Free Software Foundation; either version
12 * 2 of the License, or (at your option) any later version.
13 */
14
15#include <linux/memblock.h>
16
17#include <asm/types.h>
18#include <asm/page.h>
19#include <asm/prom.h>
20
21struct mschunks_map {
22 unsigned long num_chunks;
23 unsigned long chunk_size;
24 unsigned long chunk_shift;
25 unsigned long chunk_mask;
26 u32 *mapping;
27};
28
29extern struct mschunks_map mschunks_map;
30
31/* Chunks are 256 KB */
32#define MSCHUNKS_CHUNK_SHIFT (18)
33#define MSCHUNKS_CHUNK_SIZE (1UL << MSCHUNKS_CHUNK_SHIFT)
34#define MSCHUNKS_OFFSET_MASK (MSCHUNKS_CHUNK_SIZE - 1)
35
36static inline unsigned long chunk_to_addr(unsigned long chunk)
37{
38 return chunk << MSCHUNKS_CHUNK_SHIFT;
39}
40
41static inline unsigned long addr_to_chunk(unsigned long addr)
42{
43 return addr >> MSCHUNKS_CHUNK_SHIFT;
44}
45
46static inline unsigned long phys_to_abs(unsigned long pa)
47{
48 return pa;
49}
50
51/* Convenience macros */
52#define virt_to_abs(va) phys_to_abs(__pa(va))
53#define abs_to_virt(aa) __va(aa)
54
55#endif /* __KERNEL__ */
56#endif /* _ASM_POWERPC_ABS_ADDR_H */
diff --git a/arch/powerpc/include/asm/debug.h b/arch/powerpc/include/asm/debug.h
index 716d2f089eb6..32de2577bb6d 100644
--- a/arch/powerpc/include/asm/debug.h
+++ b/arch/powerpc/include/asm/debug.h
@@ -44,7 +44,7 @@ static inline int debugger_dabr_match(struct pt_regs *regs) { return 0; }
44static inline int debugger_fault_handler(struct pt_regs *regs) { return 0; } 44static inline int debugger_fault_handler(struct pt_regs *regs) { return 0; }
45#endif 45#endif
46 46
47extern int set_dabr(unsigned long dabr); 47extern int set_dabr(unsigned long dabr, unsigned long dabrx);
48#ifdef CONFIG_PPC_ADV_DEBUG_REGS 48#ifdef CONFIG_PPC_ADV_DEBUG_REGS
49extern void do_send_trap(struct pt_regs *regs, unsigned long address, 49extern void do_send_trap(struct pt_regs *regs, unsigned long address,
50 unsigned long error_code, int signal_code, int brkpt); 50 unsigned long error_code, int signal_code, int brkpt);
diff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h
index d60f99814ffb..58c5ee61e700 100644
--- a/arch/powerpc/include/asm/eeh.h
+++ b/arch/powerpc/include/asm/eeh.h
@@ -32,27 +32,61 @@ struct device_node;
32#ifdef CONFIG_EEH 32#ifdef CONFIG_EEH
33 33
34/* 34/*
35 * The struct is used to trace PE related EEH functionality.
36 * In theory, there will have one instance of the struct to
37 * be created against particular PE. In nature, PEs corelate
38 * to each other. the struct has to reflect that hierarchy in
39 * order to easily pick up those affected PEs when one particular
40 * PE has EEH errors.
41 *
42 * Also, one particular PE might be composed of PCI device, PCI
43 * bus and its subordinate components. The struct also need ship
44 * the information. Further more, one particular PE is only meaingful
45 * in the corresponding PHB. Therefore, the root PEs should be created
46 * against existing PHBs in on-to-one fashion.
47 */
48#define EEH_PE_PHB 1 /* PHB PE */
49#define EEH_PE_DEVICE 2 /* Device PE */
50#define EEH_PE_BUS 3 /* Bus PE */
51
52#define EEH_PE_ISOLATED (1 << 0) /* Isolated PE */
53#define EEH_PE_RECOVERING (1 << 1) /* Recovering PE */
54
55struct eeh_pe {
56 int type; /* PE type: PHB/Bus/Device */
57 int state; /* PE EEH dependent mode */
58 int config_addr; /* Traditional PCI address */
59 int addr; /* PE configuration address */
60 struct pci_controller *phb; /* Associated PHB */
61 int check_count; /* Times of ignored error */
62 int freeze_count; /* Times of froze up */
63 int false_positives; /* Times of reported #ff's */
64 struct eeh_pe *parent; /* Parent PE */
65 struct list_head child_list; /* Link PE to the child list */
66 struct list_head edevs; /* Link list of EEH devices */
67 struct list_head child; /* Child PEs */
68};
69
70#define eeh_pe_for_each_dev(pe, edev) \
71 list_for_each_entry(edev, &pe->edevs, list)
72
73/*
35 * The struct is used to trace EEH state for the associated 74 * The struct is used to trace EEH state for the associated
36 * PCI device node or PCI device. In future, it might 75 * PCI device node or PCI device. In future, it might
37 * represent PE as well so that the EEH device to form 76 * represent PE as well so that the EEH device to form
38 * another tree except the currently existing tree of PCI 77 * another tree except the currently existing tree of PCI
39 * buses and PCI devices 78 * buses and PCI devices
40 */ 79 */
41#define EEH_MODE_SUPPORTED (1<<0) /* EEH supported on the device */ 80#define EEH_DEV_IRQ_DISABLED (1<<0) /* Interrupt disabled */
42#define EEH_MODE_NOCHECK (1<<1) /* EEH check should be skipped */
43#define EEH_MODE_ISOLATED (1<<2) /* The device has been isolated */
44#define EEH_MODE_RECOVERING (1<<3) /* Recovering the device */
45#define EEH_MODE_IRQ_DISABLED (1<<4) /* Interrupt disabled */
46 81
47struct eeh_dev { 82struct eeh_dev {
48 int mode; /* EEH mode */ 83 int mode; /* EEH mode */
49 int class_code; /* Class code of the device */ 84 int class_code; /* Class code of the device */
50 int config_addr; /* Config address */ 85 int config_addr; /* Config address */
51 int pe_config_addr; /* PE config address */ 86 int pe_config_addr; /* PE config address */
52 int check_count; /* Times of ignored error */
53 int freeze_count; /* Times of froze up */
54 int false_positives; /* Times of reported #ff's */
55 u32 config_space[16]; /* Saved PCI config space */ 87 u32 config_space[16]; /* Saved PCI config space */
88 struct eeh_pe *pe; /* Associated PE */
89 struct list_head list; /* Form link list in the PE */
56 struct pci_controller *phb; /* Associated PHB */ 90 struct pci_controller *phb; /* Associated PHB */
57 struct device_node *dn; /* Associated device node */ 91 struct device_node *dn; /* Associated device node */
58 struct pci_dev *pdev; /* Associated PCI device */ 92 struct pci_dev *pdev; /* Associated PCI device */
@@ -95,19 +129,51 @@ static inline struct pci_dev *eeh_dev_to_pci_dev(struct eeh_dev *edev)
95struct eeh_ops { 129struct eeh_ops {
96 char *name; 130 char *name;
97 int (*init)(void); 131 int (*init)(void);
98 int (*set_option)(struct device_node *dn, int option); 132 void* (*of_probe)(struct device_node *dn, void *flag);
99 int (*get_pe_addr)(struct device_node *dn); 133 void* (*dev_probe)(struct pci_dev *dev, void *flag);
100 int (*get_state)(struct device_node *dn, int *state); 134 int (*set_option)(struct eeh_pe *pe, int option);
101 int (*reset)(struct device_node *dn, int option); 135 int (*get_pe_addr)(struct eeh_pe *pe);
102 int (*wait_state)(struct device_node *dn, int max_wait); 136 int (*get_state)(struct eeh_pe *pe, int *state);
103 int (*get_log)(struct device_node *dn, int severity, char *drv_log, unsigned long len); 137 int (*reset)(struct eeh_pe *pe, int option);
104 int (*configure_bridge)(struct device_node *dn); 138 int (*wait_state)(struct eeh_pe *pe, int max_wait);
139 int (*get_log)(struct eeh_pe *pe, int severity, char *drv_log, unsigned long len);
140 int (*configure_bridge)(struct eeh_pe *pe);
105 int (*read_config)(struct device_node *dn, int where, int size, u32 *val); 141 int (*read_config)(struct device_node *dn, int where, int size, u32 *val);
106 int (*write_config)(struct device_node *dn, int where, int size, u32 val); 142 int (*write_config)(struct device_node *dn, int where, int size, u32 val);
107}; 143};
108 144
109extern struct eeh_ops *eeh_ops; 145extern struct eeh_ops *eeh_ops;
110extern int eeh_subsystem_enabled; 146extern int eeh_subsystem_enabled;
147extern struct mutex eeh_mutex;
148extern int eeh_probe_mode;
149
150#define EEH_PROBE_MODE_DEV (1<<0) /* From PCI device */
151#define EEH_PROBE_MODE_DEVTREE (1<<1) /* From device tree */
152
153static inline void eeh_probe_mode_set(int flag)
154{
155 eeh_probe_mode = flag;
156}
157
158static inline int eeh_probe_mode_devtree(void)
159{
160 return (eeh_probe_mode == EEH_PROBE_MODE_DEVTREE);
161}
162
163static inline int eeh_probe_mode_dev(void)
164{
165 return (eeh_probe_mode == EEH_PROBE_MODE_DEV);
166}
167
168static inline void eeh_lock(void)
169{
170 mutex_lock(&eeh_mutex);
171}
172
173static inline void eeh_unlock(void)
174{
175 mutex_unlock(&eeh_mutex);
176}
111 177
112/* 178/*
113 * Max number of EEH freezes allowed before we consider the device 179 * Max number of EEH freezes allowed before we consider the device
@@ -115,19 +181,23 @@ extern int eeh_subsystem_enabled;
115 */ 181 */
116#define EEH_MAX_ALLOWED_FREEZES 5 182#define EEH_MAX_ALLOWED_FREEZES 5
117 183
184typedef void *(*eeh_traverse_func)(void *data, void *flag);
185int __devinit eeh_phb_pe_create(struct pci_controller *phb);
186int eeh_add_to_parent_pe(struct eeh_dev *edev);
187int eeh_rmv_from_parent_pe(struct eeh_dev *edev);
188void *eeh_pe_dev_traverse(struct eeh_pe *root,
189 eeh_traverse_func fn, void *flag);
190void eeh_pe_restore_bars(struct eeh_pe *pe);
191struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe);
192
118void * __devinit eeh_dev_init(struct device_node *dn, void *data); 193void * __devinit eeh_dev_init(struct device_node *dn, void *data);
119void __devinit eeh_dev_phb_init_dynamic(struct pci_controller *phb); 194void __devinit eeh_dev_phb_init_dynamic(struct pci_controller *phb);
120void __init eeh_dev_phb_init(void);
121void __init eeh_init(void);
122#ifdef CONFIG_PPC_PSERIES
123int __init eeh_pseries_init(void);
124#endif
125int __init eeh_ops_register(struct eeh_ops *ops); 195int __init eeh_ops_register(struct eeh_ops *ops);
126int __exit eeh_ops_unregister(const char *name); 196int __exit eeh_ops_unregister(const char *name);
127unsigned long eeh_check_failure(const volatile void __iomem *token, 197unsigned long eeh_check_failure(const volatile void __iomem *token,
128 unsigned long val); 198 unsigned long val);
129int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev); 199int eeh_dev_check_failure(struct eeh_dev *edev);
130void __init pci_addr_cache_build(void); 200void __init eeh_addr_cache_build(void);
131void eeh_add_device_tree_early(struct device_node *); 201void eeh_add_device_tree_early(struct device_node *);
132void eeh_add_device_tree_late(struct pci_bus *); 202void eeh_add_device_tree_late(struct pci_bus *);
133void eeh_remove_bus_device(struct pci_dev *); 203void eeh_remove_bus_device(struct pci_dev *);
@@ -156,34 +226,24 @@ static inline void *eeh_dev_init(struct device_node *dn, void *data)
156 226
157static inline void eeh_dev_phb_init_dynamic(struct pci_controller *phb) { } 227static inline void eeh_dev_phb_init_dynamic(struct pci_controller *phb) { }
158 228
159static inline void eeh_dev_phb_init(void) { }
160
161static inline void eeh_init(void) { }
162
163#ifdef CONFIG_PPC_PSERIES
164static inline int eeh_pseries_init(void)
165{
166 return 0;
167}
168#endif /* CONFIG_PPC_PSERIES */
169
170static inline unsigned long eeh_check_failure(const volatile void __iomem *token, unsigned long val) 229static inline unsigned long eeh_check_failure(const volatile void __iomem *token, unsigned long val)
171{ 230{
172 return val; 231 return val;
173} 232}
174 233
175static inline int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev) 234#define eeh_dev_check_failure(x) (0)
176{
177 return 0;
178}
179 235
180static inline void pci_addr_cache_build(void) { } 236static inline void eeh_addr_cache_build(void) { }
181 237
182static inline void eeh_add_device_tree_early(struct device_node *dn) { } 238static inline void eeh_add_device_tree_early(struct device_node *dn) { }
183 239
184static inline void eeh_add_device_tree_late(struct pci_bus *bus) { } 240static inline void eeh_add_device_tree_late(struct pci_bus *bus) { }
185 241
186static inline void eeh_remove_bus_device(struct pci_dev *dev) { } 242static inline void eeh_remove_bus_device(struct pci_dev *dev) { }
243
244static inline void eeh_lock(void) { }
245static inline void eeh_unlock(void) { }
246
187#define EEH_POSSIBLE_ERROR(val, type) (0) 247#define EEH_POSSIBLE_ERROR(val, type) (0)
188#define EEH_IO_ERROR_VALUE(size) (-1UL) 248#define EEH_IO_ERROR_VALUE(size) (-1UL)
189#endif /* CONFIG_EEH */ 249#endif /* CONFIG_EEH */
diff --git a/arch/powerpc/include/asm/eeh_event.h b/arch/powerpc/include/asm/eeh_event.h
index c68b012b7797..de67d830151b 100644
--- a/arch/powerpc/include/asm/eeh_event.h
+++ b/arch/powerpc/include/asm/eeh_event.h
@@ -28,11 +28,11 @@
28 */ 28 */
29struct eeh_event { 29struct eeh_event {
30 struct list_head list; /* to form event queue */ 30 struct list_head list; /* to form event queue */
31 struct eeh_dev *edev; /* EEH device */ 31 struct eeh_pe *pe; /* EEH PE */
32}; 32};
33 33
34int eeh_send_failure_event(struct eeh_dev *edev); 34int eeh_send_failure_event(struct eeh_pe *pe);
35struct eeh_dev *handle_eeh_events(struct eeh_event *); 35void eeh_handle_event(struct eeh_pe *pe);
36 36
37#endif /* __KERNEL__ */ 37#endif /* __KERNEL__ */
38#endif /* ASM_POWERPC_EEH_EVENT_H */ 38#endif /* ASM_POWERPC_EEH_EVENT_H */
diff --git a/arch/powerpc/include/asm/exception-64e.h b/arch/powerpc/include/asm/exception-64e.h
index ac13addb8495..51fa43e536b9 100644
--- a/arch/powerpc/include/asm/exception-64e.h
+++ b/arch/powerpc/include/asm/exception-64e.h
@@ -37,6 +37,7 @@
37 * critical data 37 * critical data
38 */ 38 */
39 39
40#define PACA_EXGDBELL PACA_EXGEN
40 41
41/* We are out of SPRGs so we save some things in the PACA. The normal 42/* We are out of SPRGs so we save some things in the PACA. The normal
42 * exception frame is smaller than the CRIT or MC one though 43 * exception frame is smaller than the CRIT or MC one though
@@ -45,8 +46,9 @@
45#define EX_CR (1 * 8) 46#define EX_CR (1 * 8)
46#define EX_R10 (2 * 8) 47#define EX_R10 (2 * 8)
47#define EX_R11 (3 * 8) 48#define EX_R11 (3 * 8)
48#define EX_R14 (4 * 8) 49#define EX_R13 (4 * 8)
49#define EX_R15 (5 * 8) 50#define EX_R14 (5 * 8)
51#define EX_R15 (6 * 8)
50 52
51/* 53/*
52 * The TLB miss exception uses different slots. 54 * The TLB miss exception uses different slots.
diff --git a/arch/powerpc/include/asm/hvcall.h b/arch/powerpc/include/asm/hvcall.h
index 423cf9eaf4a4..7a867065db79 100644
--- a/arch/powerpc/include/asm/hvcall.h
+++ b/arch/powerpc/include/asm/hvcall.h
@@ -152,11 +152,6 @@
152#define H_VASI_RESUMED 5 152#define H_VASI_RESUMED 5
153#define H_VASI_COMPLETED 6 153#define H_VASI_COMPLETED 6
154 154
155/* DABRX flags */
156#define H_DABRX_HYPERVISOR (1UL<<(63-61))
157#define H_DABRX_KERNEL (1UL<<(63-62))
158#define H_DABRX_USER (1UL<<(63-63))
159
160/* Each control block has to be on a 4K boundary */ 155/* Each control block has to be on a 4K boundary */
161#define H_CB_ALIGNMENT 4096 156#define H_CB_ALIGNMENT 4096
162 157
diff --git a/arch/powerpc/include/asm/hw_breakpoint.h b/arch/powerpc/include/asm/hw_breakpoint.h
index be04330af751..423424599dad 100644
--- a/arch/powerpc/include/asm/hw_breakpoint.h
+++ b/arch/powerpc/include/asm/hw_breakpoint.h
@@ -27,10 +27,11 @@
27#ifdef CONFIG_HAVE_HW_BREAKPOINT 27#ifdef CONFIG_HAVE_HW_BREAKPOINT
28 28
29struct arch_hw_breakpoint { 29struct arch_hw_breakpoint {
30 bool extraneous_interrupt;
31 u8 len; /* length of the target data symbol */
32 int type;
33 unsigned long address; 30 unsigned long address;
31 unsigned long dabrx;
32 int type;
33 u8 len; /* length of the target data symbol */
34 bool extraneous_interrupt;
34}; 35};
35 36
36#include <linux/kdebug.h> 37#include <linux/kdebug.h>
@@ -61,7 +62,7 @@ extern void ptrace_triggered(struct perf_event *bp,
61 struct perf_sample_data *data, struct pt_regs *regs); 62 struct perf_sample_data *data, struct pt_regs *regs);
62static inline void hw_breakpoint_disable(void) 63static inline void hw_breakpoint_disable(void)
63{ 64{
64 set_dabr(0); 65 set_dabr(0, 0);
65} 66}
66extern void thread_change_pc(struct task_struct *tsk, struct pt_regs *regs); 67extern void thread_change_pc(struct task_struct *tsk, struct pt_regs *regs);
67 68
diff --git a/arch/powerpc/include/asm/kprobes.h b/arch/powerpc/include/asm/kprobes.h
index be0171afdc0f..7b6feab6fd26 100644
--- a/arch/powerpc/include/asm/kprobes.h
+++ b/arch/powerpc/include/asm/kprobes.h
@@ -29,21 +29,16 @@
29#include <linux/types.h> 29#include <linux/types.h>
30#include <linux/ptrace.h> 30#include <linux/ptrace.h>
31#include <linux/percpu.h> 31#include <linux/percpu.h>
32#include <asm/probes.h>
32 33
33#define __ARCH_WANT_KPROBES_INSN_SLOT 34#define __ARCH_WANT_KPROBES_INSN_SLOT
34 35
35struct pt_regs; 36struct pt_regs;
36struct kprobe; 37struct kprobe;
37 38
38typedef unsigned int kprobe_opcode_t; 39typedef ppc_opcode_t kprobe_opcode_t;
39#define BREAKPOINT_INSTRUCTION 0x7fe00008 /* trap */
40#define MAX_INSN_SIZE 1 40#define MAX_INSN_SIZE 1
41 41
42#define IS_TW(instr) (((instr) & 0xfc0007fe) == 0x7c000008)
43#define IS_TD(instr) (((instr) & 0xfc0007fe) == 0x7c000088)
44#define IS_TDI(instr) (((instr) & 0xfc000000) == 0x08000000)
45#define IS_TWI(instr) (((instr) & 0xfc000000) == 0x0c000000)
46
47#ifdef CONFIG_PPC64 42#ifdef CONFIG_PPC64
48/* 43/*
49 * 64bit powerpc uses function descriptors. 44 * 64bit powerpc uses function descriptors.
@@ -72,12 +67,6 @@ typedef unsigned int kprobe_opcode_t;
72 addr = (kprobe_opcode_t *)kallsyms_lookup_name(dot_name); \ 67 addr = (kprobe_opcode_t *)kallsyms_lookup_name(dot_name); \
73 } \ 68 } \
74} 69}
75
76#define is_trap(instr) (IS_TW(instr) || IS_TD(instr) || \
77 IS_TWI(instr) || IS_TDI(instr))
78#else
79/* Use stock kprobe_lookup_name since ppc32 doesn't use function descriptors */
80#define is_trap(instr) (IS_TW(instr) || IS_TWI(instr))
81#endif 70#endif
82 71
83#define flush_insn_slot(p) do { } while (0) 72#define flush_insn_slot(p) do { } while (0)
diff --git a/arch/powerpc/include/asm/kvm_book3s_asm.h b/arch/powerpc/include/asm/kvm_book3s_asm.h
index bfcd00c1485d..88609b23b775 100644
--- a/arch/powerpc/include/asm/kvm_book3s_asm.h
+++ b/arch/powerpc/include/asm/kvm_book3s_asm.h
@@ -74,7 +74,6 @@ struct kvmppc_host_state {
74 ulong vmhandler; 74 ulong vmhandler;
75 ulong scratch0; 75 ulong scratch0;
76 ulong scratch1; 76 ulong scratch1;
77 ulong sprg3;
78 u8 in_guest; 77 u8 in_guest;
79 u8 restore_hid5; 78 u8 restore_hid5;
80 u8 napping; 79 u8 napping;
diff --git a/arch/powerpc/include/asm/machdep.h b/arch/powerpc/include/asm/machdep.h
index f7706d722b39..8111e1b78f7f 100644
--- a/arch/powerpc/include/asm/machdep.h
+++ b/arch/powerpc/include/asm/machdep.h
@@ -180,7 +180,8 @@ struct machdep_calls {
180 void (*enable_pmcs)(void); 180 void (*enable_pmcs)(void);
181 181
182 /* Set DABR for this platform, leave empty for default implemenation */ 182 /* Set DABR for this platform, leave empty for default implemenation */
183 int (*set_dabr)(unsigned long dabr); 183 int (*set_dabr)(unsigned long dabr,
184 unsigned long dabrx);
184 185
185#ifdef CONFIG_PPC32 /* XXX for now */ 186#ifdef CONFIG_PPC32 /* XXX for now */
186 /* A general init function, called by ppc_init in init/main.c. 187 /* A general init function, called by ppc_init in init/main.c.
diff --git a/arch/powerpc/include/asm/paca.h b/arch/powerpc/include/asm/paca.h
index daf813fea91f..7796519fd238 100644
--- a/arch/powerpc/include/asm/paca.h
+++ b/arch/powerpc/include/asm/paca.h
@@ -136,6 +136,7 @@ struct paca_struct {
136 u8 io_sync; /* writel() needs spin_unlock sync */ 136 u8 io_sync; /* writel() needs spin_unlock sync */
137 u8 irq_work_pending; /* IRQ_WORK interrupt while soft-disable */ 137 u8 irq_work_pending; /* IRQ_WORK interrupt while soft-disable */
138 u8 nap_state_lost; /* NV GPR values lost in power7_idle */ 138 u8 nap_state_lost; /* NV GPR values lost in power7_idle */
139 u64 sprg3; /* Saved user-visible sprg */
139 140
140#ifdef CONFIG_PPC_POWERNV 141#ifdef CONFIG_PPC_POWERNV
141 /* Pointer to OPAL machine check event structure set by the 142 /* Pointer to OPAL machine check event structure set by the
diff --git a/arch/powerpc/include/asm/pci-bridge.h b/arch/powerpc/include/asm/pci-bridge.h
index 8cccbee61519..973df4d9d366 100644
--- a/arch/powerpc/include/asm/pci-bridge.h
+++ b/arch/powerpc/include/asm/pci-bridge.h
@@ -184,6 +184,8 @@ static inline struct eeh_dev *of_node_to_eeh_dev(struct device_node *dn)
184{ 184{
185 return PCI_DN(dn)->edev; 185 return PCI_DN(dn)->edev;
186} 186}
187#else
188#define of_node_to_eeh_dev(x) (NULL)
187#endif 189#endif
188 190
189/** Find the bus corresponding to the indicated device node */ 191/** Find the bus corresponding to the indicated device node */
diff --git a/arch/powerpc/include/asm/ppc-pci.h b/arch/powerpc/include/asm/ppc-pci.h
index 80fa704d410f..ed57fa7920c8 100644
--- a/arch/powerpc/include/asm/ppc-pci.h
+++ b/arch/powerpc/include/asm/ppc-pci.h
@@ -47,19 +47,17 @@ extern int rtas_setup_phb(struct pci_controller *phb);
47 47
48#ifdef CONFIG_EEH 48#ifdef CONFIG_EEH
49 49
50void pci_addr_cache_build(void); 50void eeh_addr_cache_insert_dev(struct pci_dev *dev);
51void pci_addr_cache_insert_device(struct pci_dev *dev); 51void eeh_addr_cache_rmv_dev(struct pci_dev *dev);
52void pci_addr_cache_remove_device(struct pci_dev *dev); 52struct eeh_dev *eeh_addr_cache_get_dev(unsigned long addr);
53struct pci_dev *pci_addr_cache_get_device(unsigned long addr); 53void eeh_slot_error_detail(struct eeh_pe *pe, int severity);
54void eeh_slot_error_detail(struct eeh_dev *edev, int severity); 54int eeh_pci_enable(struct eeh_pe *pe, int function);
55int eeh_pci_enable(struct eeh_dev *edev, int function); 55int eeh_reset_pe(struct eeh_pe *);
56int eeh_reset_pe(struct eeh_dev *); 56void eeh_save_bars(struct eeh_dev *edev);
57void eeh_restore_bars(struct eeh_dev *);
58int rtas_write_config(struct pci_dn *, int where, int size, u32 val); 57int rtas_write_config(struct pci_dn *, int where, int size, u32 val);
59int rtas_read_config(struct pci_dn *, int where, int size, u32 *val); 58int rtas_read_config(struct pci_dn *, int where, int size, u32 *val);
60void eeh_mark_slot(struct device_node *dn, int mode_flag); 59void eeh_pe_state_mark(struct eeh_pe *pe, int state);
61void eeh_clear_slot(struct device_node *dn, int mode_flag); 60void eeh_pe_state_clear(struct eeh_pe *pe, int state);
62struct device_node *eeh_find_device_pe(struct device_node *dn);
63 61
64void eeh_sysfs_add_device(struct pci_dev *pdev); 62void eeh_sysfs_add_device(struct pci_dev *pdev);
65void eeh_sysfs_remove_device(struct pci_dev *pdev); 63void eeh_sysfs_remove_device(struct pci_dev *pdev);
diff --git a/arch/powerpc/include/asm/probes.h b/arch/powerpc/include/asm/probes.h
new file mode 100644
index 000000000000..5f1e15b68704
--- /dev/null
+++ b/arch/powerpc/include/asm/probes.h
@@ -0,0 +1,42 @@
1#ifndef _ASM_POWERPC_PROBES_H
2#define _ASM_POWERPC_PROBES_H
3#ifdef __KERNEL__
4/*
5 * Definitions common to probes files
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
20 *
21 * Copyright IBM Corporation, 2012
22 */
23#include <linux/types.h>
24
25typedef u32 ppc_opcode_t;
26#define BREAKPOINT_INSTRUCTION 0x7fe00008 /* trap */
27
28/* Trap definitions per ISA */
29#define IS_TW(instr) (((instr) & 0xfc0007fe) == 0x7c000008)
30#define IS_TD(instr) (((instr) & 0xfc0007fe) == 0x7c000088)
31#define IS_TDI(instr) (((instr) & 0xfc000000) == 0x08000000)
32#define IS_TWI(instr) (((instr) & 0xfc000000) == 0x0c000000)
33
34#ifdef CONFIG_PPC64
35#define is_trap(instr) (IS_TW(instr) || IS_TD(instr) || \
36 IS_TWI(instr) || IS_TDI(instr))
37#else
38#define is_trap(instr) (IS_TW(instr) || IS_TWI(instr))
39#endif /* CONFIG_PPC64 */
40
41#endif /* __KERNEL__ */
42#endif /* _ASM_POWERPC_PROBES_H */
diff --git a/arch/powerpc/include/asm/processor.h b/arch/powerpc/include/asm/processor.h
index 54b73a28c205..83efc6e81543 100644
--- a/arch/powerpc/include/asm/processor.h
+++ b/arch/powerpc/include/asm/processor.h
@@ -219,6 +219,8 @@ struct thread_struct {
219#endif /* CONFIG_HAVE_HW_BREAKPOINT */ 219#endif /* CONFIG_HAVE_HW_BREAKPOINT */
220#endif 220#endif
221 unsigned long dabr; /* Data address breakpoint register */ 221 unsigned long dabr; /* Data address breakpoint register */
222 unsigned long dabrx; /* ... extension */
223 unsigned long trap_nr; /* last trap # on this thread */
222#ifdef CONFIG_ALTIVEC 224#ifdef CONFIG_ALTIVEC
223 /* Complete AltiVec register set */ 225 /* Complete AltiVec register set */
224 vector128 vr[32] __attribute__((aligned(16))); 226 vector128 vr[32] __attribute__((aligned(16)));
diff --git a/arch/powerpc/include/asm/reg.h b/arch/powerpc/include/asm/reg.h
index 638608677e2a..121a90bbf778 100644
--- a/arch/powerpc/include/asm/reg.h
+++ b/arch/powerpc/include/asm/reg.h
@@ -208,6 +208,9 @@
208#define SPRN_DABRX 0x3F7 /* Data Address Breakpoint Register Extension */ 208#define SPRN_DABRX 0x3F7 /* Data Address Breakpoint Register Extension */
209#define DABRX_USER (1UL << 0) 209#define DABRX_USER (1UL << 0)
210#define DABRX_KERNEL (1UL << 1) 210#define DABRX_KERNEL (1UL << 1)
211#define DABRX_HYP (1UL << 2)
212#define DABRX_BTI (1UL << 3)
213#define DABRX_ALL (DABRX_BTI | DABRX_HYP | DABRX_KERNEL | DABRX_USER)
211#define SPRN_DAR 0x013 /* Data Address Register */ 214#define SPRN_DAR 0x013 /* Data Address Register */
212#define SPRN_DBCR 0x136 /* e300 Data Breakpoint Control Reg */ 215#define SPRN_DBCR 0x136 /* e300 Data Breakpoint Control Reg */
213#define SPRN_DSISR 0x012 /* Data Storage Interrupt Status Register */ 216#define SPRN_DSISR 0x012 /* Data Storage Interrupt Status Register */
@@ -761,7 +764,8 @@
761 * 64-bit embedded 764 * 64-bit embedded
762 * - SPRG0 generic exception scratch 765 * - SPRG0 generic exception scratch
763 * - SPRG2 TLB exception stack 766 * - SPRG2 TLB exception stack
764 * - SPRG3 CPU and NUMA node for VDSO getcpu (user visible) 767 * - SPRG3 critical exception scratch and
768 * CPU and NUMA node for VDSO getcpu (user visible)
765 * - SPRG4 unused (user visible) 769 * - SPRG4 unused (user visible)
766 * - SPRG6 TLB miss scratch (user visible, sorry !) 770 * - SPRG6 TLB miss scratch (user visible, sorry !)
767 * - SPRG7 critical exception scratch 771 * - SPRG7 critical exception scratch
@@ -858,11 +862,12 @@
858 862
859#ifdef CONFIG_PPC_BOOK3E_64 863#ifdef CONFIG_PPC_BOOK3E_64
860#define SPRN_SPRG_MC_SCRATCH SPRN_SPRG8 864#define SPRN_SPRG_MC_SCRATCH SPRN_SPRG8
861#define SPRN_SPRG_CRIT_SCRATCH SPRN_SPRG7 865#define SPRN_SPRG_CRIT_SCRATCH SPRN_SPRG3
862#define SPRN_SPRG_DBG_SCRATCH SPRN_SPRG9 866#define SPRN_SPRG_DBG_SCRATCH SPRN_SPRG9
863#define SPRN_SPRG_TLB_EXFRAME SPRN_SPRG2 867#define SPRN_SPRG_TLB_EXFRAME SPRN_SPRG2
864#define SPRN_SPRG_TLB_SCRATCH SPRN_SPRG6 868#define SPRN_SPRG_TLB_SCRATCH SPRN_SPRG6
865#define SPRN_SPRG_GEN_SCRATCH SPRN_SPRG0 869#define SPRN_SPRG_GEN_SCRATCH SPRN_SPRG0
870#define SPRN_SPRG_GDBELL_SCRATCH SPRN_SPRG_GEN_SCRATCH
866 871
867#define SET_PACA(rX) mtspr SPRN_SPRG_PACA,rX 872#define SET_PACA(rX) mtspr SPRN_SPRG_PACA,rX
868#define GET_PACA(rX) mfspr rX,SPRN_SPRG_PACA 873#define GET_PACA(rX) mfspr rX,SPRN_SPRG_PACA
@@ -937,7 +942,7 @@
937#define PVR_VER(pvr) (((pvr) >> 16) & 0xFFFF) /* Version field */ 942#define PVR_VER(pvr) (((pvr) >> 16) & 0xFFFF) /* Version field */
938#define PVR_REV(pvr) (((pvr) >> 0) & 0xFFFF) /* Revison field */ 943#define PVR_REV(pvr) (((pvr) >> 0) & 0xFFFF) /* Revison field */
939 944
940#define __is_processor(pv) (PVR_VER(mfspr(SPRN_PVR)) == (pv)) 945#define pvr_version_is(pvr) (PVR_VER(mfspr(SPRN_PVR)) == (pvr))
941 946
942/* 947/*
943 * IBM has further subdivided the standard PowerPC 16-bit version and 948 * IBM has further subdivided the standard PowerPC 16-bit version and
@@ -1002,25 +1007,25 @@
1002#define PVR_476_ISS 0x00052000 1007#define PVR_476_ISS 0x00052000
1003 1008
1004/* 64-bit processors */ 1009/* 64-bit processors */
1005/* XXX the prefix should be PVR_, we'll do a global sweep to fix it one day */ 1010#define PVR_NORTHSTAR 0x0033
1006#define PV_NORTHSTAR 0x0033 1011#define PVR_PULSAR 0x0034
1007#define PV_PULSAR 0x0034 1012#define PVR_POWER4 0x0035
1008#define PV_POWER4 0x0035 1013#define PVR_ICESTAR 0x0036
1009#define PV_ICESTAR 0x0036 1014#define PVR_SSTAR 0x0037
1010#define PV_SSTAR 0x0037 1015#define PVR_POWER4p 0x0038
1011#define PV_POWER4p 0x0038 1016#define PVR_970 0x0039
1012#define PV_970 0x0039 1017#define PVR_POWER5 0x003A
1013#define PV_POWER5 0x003A 1018#define PVR_POWER5p 0x003B
1014#define PV_POWER5p 0x003B 1019#define PVR_970FX 0x003C
1015#define PV_970FX 0x003C 1020#define PVR_POWER6 0x003E
1016#define PV_POWER6 0x003E 1021#define PVR_POWER7 0x003F
1017#define PV_POWER7 0x003F 1022#define PVR_630 0x0040
1018#define PV_630 0x0040 1023#define PVR_630p 0x0041
1019#define PV_630p 0x0041 1024#define PVR_970MP 0x0044
1020#define PV_970MP 0x0044 1025#define PVR_970GX 0x0045
1021#define PV_970GX 0x0045 1026#define PVR_POWER7p 0x004A
1022#define PV_BE 0x0070 1027#define PVR_BE 0x0070
1023#define PV_PA6T 0x0090 1028#define PVR_PA6T 0x0090
1024 1029
1025/* Macros for setting and retrieving special purpose registers */ 1030/* Macros for setting and retrieving special purpose registers */
1026#ifndef __ASSEMBLY__ 1031#ifndef __ASSEMBLY__
diff --git a/arch/powerpc/include/asm/setup.h b/arch/powerpc/include/asm/setup.h
index d084ce195fc3..8b9a306260b2 100644
--- a/arch/powerpc/include/asm/setup.h
+++ b/arch/powerpc/include/asm/setup.h
@@ -9,7 +9,7 @@ extern void ppc_printk_progress(char *s, unsigned short hex);
9extern unsigned int rtas_data; 9extern unsigned int rtas_data;
10extern int mem_init_done; /* set on boot once kmalloc can be called */ 10extern int mem_init_done; /* set on boot once kmalloc can be called */
11extern int init_bootmem_done; /* set once bootmem is available */ 11extern int init_bootmem_done; /* set once bootmem is available */
12extern phys_addr_t memory_limit; 12extern unsigned long long memory_limit;
13extern unsigned long klimit; 13extern unsigned long klimit;
14extern void *zalloc_maybe_bootmem(size_t size, gfp_t mask); 14extern void *zalloc_maybe_bootmem(size_t size, gfp_t mask);
15 15
diff --git a/arch/powerpc/include/asm/thread_info.h b/arch/powerpc/include/asm/thread_info.h
index faf93529cbf0..e942203cd4a8 100644
--- a/arch/powerpc/include/asm/thread_info.h
+++ b/arch/powerpc/include/asm/thread_info.h
@@ -102,6 +102,7 @@ static inline struct thread_info *current_thread_info(void)
102#define TIF_RESTOREALL 11 /* Restore all regs (implies NOERROR) */ 102#define TIF_RESTOREALL 11 /* Restore all regs (implies NOERROR) */
103#define TIF_NOERROR 12 /* Force successful syscall return */ 103#define TIF_NOERROR 12 /* Force successful syscall return */
104#define TIF_NOTIFY_RESUME 13 /* callback before returning to user */ 104#define TIF_NOTIFY_RESUME 13 /* callback before returning to user */
105#define TIF_UPROBE 14 /* breakpointed or single-stepping */
105#define TIF_SYSCALL_TRACEPOINT 15 /* syscall tracepoint instrumentation */ 106#define TIF_SYSCALL_TRACEPOINT 15 /* syscall tracepoint instrumentation */
106 107
107/* as above, but as bit values */ 108/* as above, but as bit values */
@@ -118,12 +119,13 @@ static inline struct thread_info *current_thread_info(void)
118#define _TIF_RESTOREALL (1<<TIF_RESTOREALL) 119#define _TIF_RESTOREALL (1<<TIF_RESTOREALL)
119#define _TIF_NOERROR (1<<TIF_NOERROR) 120#define _TIF_NOERROR (1<<TIF_NOERROR)
120#define _TIF_NOTIFY_RESUME (1<<TIF_NOTIFY_RESUME) 121#define _TIF_NOTIFY_RESUME (1<<TIF_NOTIFY_RESUME)
122#define _TIF_UPROBE (1<<TIF_UPROBE)
121#define _TIF_SYSCALL_TRACEPOINT (1<<TIF_SYSCALL_TRACEPOINT) 123#define _TIF_SYSCALL_TRACEPOINT (1<<TIF_SYSCALL_TRACEPOINT)
122#define _TIF_SYSCALL_T_OR_A (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | \ 124#define _TIF_SYSCALL_T_OR_A (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | \
123 _TIF_SECCOMP | _TIF_SYSCALL_TRACEPOINT) 125 _TIF_SECCOMP | _TIF_SYSCALL_TRACEPOINT)
124 126
125#define _TIF_USER_WORK_MASK (_TIF_SIGPENDING | _TIF_NEED_RESCHED | \ 127#define _TIF_USER_WORK_MASK (_TIF_SIGPENDING | _TIF_NEED_RESCHED | \
126 _TIF_NOTIFY_RESUME) 128 _TIF_NOTIFY_RESUME | _TIF_UPROBE)
127#define _TIF_PERSYSCALL_MASK (_TIF_RESTOREALL|_TIF_NOERROR) 129#define _TIF_PERSYSCALL_MASK (_TIF_RESTOREALL|_TIF_NOERROR)
128 130
129/* Bits in local_flags */ 131/* Bits in local_flags */
diff --git a/arch/powerpc/include/asm/uprobes.h b/arch/powerpc/include/asm/uprobes.h
new file mode 100644
index 000000000000..b532060d0916
--- /dev/null
+++ b/arch/powerpc/include/asm/uprobes.h
@@ -0,0 +1,54 @@
1#ifndef _ASM_UPROBES_H
2#define _ASM_UPROBES_H
3/*
4 * User-space Probes (UProbes) for powerpc
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
10 *
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
15 *
16 * You should have received a copy of the GNU General Public License
17 * along with this program; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
19 *
20 * Copyright IBM Corporation, 2007-2012
21 *
22 * Adapted from the x86 port by Ananth N Mavinakayanahalli <ananth@in.ibm.com>
23 */
24
25#include <linux/notifier.h>
26#include <asm/probes.h>
27
28typedef ppc_opcode_t uprobe_opcode_t;
29
30#define MAX_UINSN_BYTES 4
31#define UPROBE_XOL_SLOT_BYTES (MAX_UINSN_BYTES)
32
33/* The following alias is needed for reference from arch-agnostic code */
34#define UPROBE_SWBP_INSN BREAKPOINT_INSTRUCTION
35#define UPROBE_SWBP_INSN_SIZE 4 /* swbp insn size in bytes */
36
37struct arch_uprobe {
38 union {
39 u8 insn[MAX_UINSN_BYTES];
40 u32 ainsn;
41 };
42};
43
44struct arch_uprobe_task {
45 unsigned long saved_trap_nr;
46};
47
48extern int arch_uprobe_analyze_insn(struct arch_uprobe *aup, struct mm_struct *mm, unsigned long addr);
49extern int arch_uprobe_pre_xol(struct arch_uprobe *aup, struct pt_regs *regs);
50extern int arch_uprobe_post_xol(struct arch_uprobe *aup, struct pt_regs *regs);
51extern bool arch_uprobe_xol_was_trapped(struct task_struct *tsk);
52extern int arch_uprobe_exception_notify(struct notifier_block *self, unsigned long val, void *data);
53extern void arch_uprobe_abort_xol(struct arch_uprobe *aup, struct pt_regs *regs);
54#endif /* _ASM_UPROBES_H */
diff --git a/arch/powerpc/kernel/Makefile b/arch/powerpc/kernel/Makefile
index bb282dd81612..cde12f8a4ebc 100644
--- a/arch/powerpc/kernel/Makefile
+++ b/arch/powerpc/kernel/Makefile
@@ -96,6 +96,7 @@ obj-$(CONFIG_MODULES) += ppc_ksyms.o
96obj-$(CONFIG_BOOTX_TEXT) += btext.o 96obj-$(CONFIG_BOOTX_TEXT) += btext.o
97obj-$(CONFIG_SMP) += smp.o 97obj-$(CONFIG_SMP) += smp.o
98obj-$(CONFIG_KPROBES) += kprobes.o 98obj-$(CONFIG_KPROBES) += kprobes.o
99obj-$(CONFIG_UPROBES) += uprobes.o
99obj-$(CONFIG_PPC_UDBG_16550) += legacy_serial.o udbg_16550.o 100obj-$(CONFIG_PPC_UDBG_16550) += legacy_serial.o udbg_16550.o
100obj-$(CONFIG_STACKTRACE) += stacktrace.o 101obj-$(CONFIG_STACKTRACE) += stacktrace.o
101obj-$(CONFIG_SWIOTLB) += dma-swiotlb.o 102obj-$(CONFIG_SWIOTLB) += dma-swiotlb.o
diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
index e8995727b1c1..7523539cfe9f 100644
--- a/arch/powerpc/kernel/asm-offsets.c
+++ b/arch/powerpc/kernel/asm-offsets.c
@@ -206,6 +206,7 @@ int main(void)
206 DEFINE(PACA_SYSTEM_TIME, offsetof(struct paca_struct, system_time)); 206 DEFINE(PACA_SYSTEM_TIME, offsetof(struct paca_struct, system_time));
207 DEFINE(PACA_TRAP_SAVE, offsetof(struct paca_struct, trap_save)); 207 DEFINE(PACA_TRAP_SAVE, offsetof(struct paca_struct, trap_save));
208 DEFINE(PACA_NAPSTATELOST, offsetof(struct paca_struct, nap_state_lost)); 208 DEFINE(PACA_NAPSTATELOST, offsetof(struct paca_struct, nap_state_lost));
209 DEFINE(PACA_SPRG3, offsetof(struct paca_struct, sprg3));
209#endif /* CONFIG_PPC64 */ 210#endif /* CONFIG_PPC64 */
210 211
211 /* RTAS */ 212 /* RTAS */
@@ -534,7 +535,6 @@ int main(void)
534 HSTATE_FIELD(HSTATE_VMHANDLER, vmhandler); 535 HSTATE_FIELD(HSTATE_VMHANDLER, vmhandler);
535 HSTATE_FIELD(HSTATE_SCRATCH0, scratch0); 536 HSTATE_FIELD(HSTATE_SCRATCH0, scratch0);
536 HSTATE_FIELD(HSTATE_SCRATCH1, scratch1); 537 HSTATE_FIELD(HSTATE_SCRATCH1, scratch1);
537 HSTATE_FIELD(HSTATE_SPRG3, sprg3);
538 HSTATE_FIELD(HSTATE_IN_GUEST, in_guest); 538 HSTATE_FIELD(HSTATE_IN_GUEST, in_guest);
539 HSTATE_FIELD(HSTATE_RESTORE_HID5, restore_hid5); 539 HSTATE_FIELD(HSTATE_RESTORE_HID5, restore_hid5);
540 HSTATE_FIELD(HSTATE_NAPPING, napping); 540 HSTATE_FIELD(HSTATE_NAPPING, napping);
diff --git a/arch/powerpc/kernel/dma-swiotlb.c b/arch/powerpc/kernel/dma-swiotlb.c
index 46943651da23..a720b54b971c 100644
--- a/arch/powerpc/kernel/dma-swiotlb.c
+++ b/arch/powerpc/kernel/dma-swiotlb.c
@@ -12,6 +12,7 @@
12 */ 12 */
13 13
14#include <linux/dma-mapping.h> 14#include <linux/dma-mapping.h>
15#include <linux/memblock.h>
15#include <linux/pfn.h> 16#include <linux/pfn.h>
16#include <linux/of_platform.h> 17#include <linux/of_platform.h>
17#include <linux/platform_device.h> 18#include <linux/platform_device.h>
@@ -20,7 +21,6 @@
20#include <asm/machdep.h> 21#include <asm/machdep.h>
21#include <asm/swiotlb.h> 22#include <asm/swiotlb.h>
22#include <asm/dma.h> 23#include <asm/dma.h>
23#include <asm/abs_addr.h>
24 24
25unsigned int ppc_swiotlb_enable; 25unsigned int ppc_swiotlb_enable;
26 26
diff --git a/arch/powerpc/kernel/dma.c b/arch/powerpc/kernel/dma.c
index 355b9d84b0f8..8032b97ccdcb 100644
--- a/arch/powerpc/kernel/dma.c
+++ b/arch/powerpc/kernel/dma.c
@@ -14,7 +14,6 @@
14#include <linux/pci.h> 14#include <linux/pci.h>
15#include <asm/vio.h> 15#include <asm/vio.h>
16#include <asm/bug.h> 16#include <asm/bug.h>
17#include <asm/abs_addr.h>
18#include <asm/machdep.h> 17#include <asm/machdep.h>
19 18
20/* 19/*
@@ -50,7 +49,7 @@ void *dma_direct_alloc_coherent(struct device *dev, size_t size,
50 return NULL; 49 return NULL;
51 ret = page_address(page); 50 ret = page_address(page);
52 memset(ret, 0, size); 51 memset(ret, 0, size);
53 *dma_handle = virt_to_abs(ret) + get_dma_offset(dev); 52 *dma_handle = __pa(ret) + get_dma_offset(dev);
54 53
55 return ret; 54 return ret;
56#endif 55#endif
diff --git a/arch/powerpc/kernel/exceptions-64e.S b/arch/powerpc/kernel/exceptions-64e.S
index 98be7f0cd227..87a82fbdf05a 100644
--- a/arch/powerpc/kernel/exceptions-64e.S
+++ b/arch/powerpc/kernel/exceptions-64e.S
@@ -25,6 +25,8 @@
25#include <asm/ppc-opcode.h> 25#include <asm/ppc-opcode.h>
26#include <asm/mmu.h> 26#include <asm/mmu.h>
27#include <asm/hw_irq.h> 27#include <asm/hw_irq.h>
28#include <asm/kvm_asm.h>
29#include <asm/kvm_booke_hv_asm.h>
28 30
29/* XXX This will ultimately add space for a special exception save 31/* XXX This will ultimately add space for a special exception save
30 * structure used to save things like SRR0/SRR1, SPRGs, MAS, etc... 32 * structure used to save things like SRR0/SRR1, SPRGs, MAS, etc...
@@ -35,16 +37,18 @@
35#define SPECIAL_EXC_FRAME_SIZE INT_FRAME_SIZE 37#define SPECIAL_EXC_FRAME_SIZE INT_FRAME_SIZE
36 38
37/* Exception prolog code for all exceptions */ 39/* Exception prolog code for all exceptions */
38#define EXCEPTION_PROLOG(n, type, addition) \ 40#define EXCEPTION_PROLOG(n, intnum, type, addition) \
39 mtspr SPRN_SPRG_##type##_SCRATCH,r13; /* get spare registers */ \ 41 mtspr SPRN_SPRG_##type##_SCRATCH,r13; /* get spare registers */ \
40 mfspr r13,SPRN_SPRG_PACA; /* get PACA */ \ 42 mfspr r13,SPRN_SPRG_PACA; /* get PACA */ \
41 std r10,PACA_EX##type+EX_R10(r13); \ 43 std r10,PACA_EX##type+EX_R10(r13); \
42 std r11,PACA_EX##type+EX_R11(r13); \ 44 std r11,PACA_EX##type+EX_R11(r13); \
45 PROLOG_STORE_RESTORE_SCRATCH_##type; \
43 mfcr r10; /* save CR */ \ 46 mfcr r10; /* save CR */ \
47 mfspr r11,SPRN_##type##_SRR1;/* what are we coming from */ \
48 DO_KVM intnum,SPRN_##type##_SRR1; /* KVM hook */ \
49 stw r10,PACA_EX##type+EX_CR(r13); /* save old CR in the PACA */ \
44 addition; /* additional code for that exc. */ \ 50 addition; /* additional code for that exc. */ \
45 std r1,PACA_EX##type+EX_R1(r13); /* save old r1 in the PACA */ \ 51 std r1,PACA_EX##type+EX_R1(r13); /* save old r1 in the PACA */ \
46 stw r10,PACA_EX##type+EX_CR(r13); /* save old CR in the PACA */ \
47 mfspr r11,SPRN_##type##_SRR1;/* what are we coming from */ \
48 type##_SET_KSTACK; /* get special stack if necessary */\ 52 type##_SET_KSTACK; /* get special stack if necessary */\
49 andi. r10,r11,MSR_PR; /* save stack pointer */ \ 53 andi. r10,r11,MSR_PR; /* save stack pointer */ \
50 beq 1f; /* branch around if supervisor */ \ 54 beq 1f; /* branch around if supervisor */ \
@@ -59,6 +63,10 @@
59#define SPRN_GEN_SRR0 SPRN_SRR0 63#define SPRN_GEN_SRR0 SPRN_SRR0
60#define SPRN_GEN_SRR1 SPRN_SRR1 64#define SPRN_GEN_SRR1 SPRN_SRR1
61 65
66#define GDBELL_SET_KSTACK GEN_SET_KSTACK
67#define SPRN_GDBELL_SRR0 SPRN_GSRR0
68#define SPRN_GDBELL_SRR1 SPRN_GSRR1
69
62#define CRIT_SET_KSTACK \ 70#define CRIT_SET_KSTACK \
63 ld r1,PACA_CRIT_STACK(r13); \ 71 ld r1,PACA_CRIT_STACK(r13); \
64 subi r1,r1,SPECIAL_EXC_FRAME_SIZE; 72 subi r1,r1,SPECIAL_EXC_FRAME_SIZE;
@@ -77,29 +85,46 @@
77#define SPRN_MC_SRR0 SPRN_MCSRR0 85#define SPRN_MC_SRR0 SPRN_MCSRR0
78#define SPRN_MC_SRR1 SPRN_MCSRR1 86#define SPRN_MC_SRR1 SPRN_MCSRR1
79 87
80#define NORMAL_EXCEPTION_PROLOG(n, addition) \ 88#define NORMAL_EXCEPTION_PROLOG(n, intnum, addition) \
81 EXCEPTION_PROLOG(n, GEN, addition##_GEN(n)) 89 EXCEPTION_PROLOG(n, intnum, GEN, addition##_GEN(n))
82 90
83#define CRIT_EXCEPTION_PROLOG(n, addition) \ 91#define CRIT_EXCEPTION_PROLOG(n, intnum, addition) \
84 EXCEPTION_PROLOG(n, CRIT, addition##_CRIT(n)) 92 EXCEPTION_PROLOG(n, intnum, CRIT, addition##_CRIT(n))
85 93
86#define DBG_EXCEPTION_PROLOG(n, addition) \ 94#define DBG_EXCEPTION_PROLOG(n, intnum, addition) \
87 EXCEPTION_PROLOG(n, DBG, addition##_DBG(n)) 95 EXCEPTION_PROLOG(n, intnum, DBG, addition##_DBG(n))
88 96
89#define MC_EXCEPTION_PROLOG(n, addition) \ 97#define MC_EXCEPTION_PROLOG(n, intnum, addition) \
90 EXCEPTION_PROLOG(n, MC, addition##_MC(n)) 98 EXCEPTION_PROLOG(n, intnum, MC, addition##_MC(n))
91 99
100#define GDBELL_EXCEPTION_PROLOG(n, intnum, addition) \
101 EXCEPTION_PROLOG(n, intnum, GDBELL, addition##_GDBELL(n))
102
103/*
104 * Store user-visible scratch in PACA exception slots and restore proper value
105 */
106#define PROLOG_STORE_RESTORE_SCRATCH_GEN
107#define PROLOG_STORE_RESTORE_SCRATCH_GDBELL
108#define PROLOG_STORE_RESTORE_SCRATCH_DBG
109#define PROLOG_STORE_RESTORE_SCRATCH_MC
110
111#define PROLOG_STORE_RESTORE_SCRATCH_CRIT \
112 mfspr r10,SPRN_SPRG_CRIT_SCRATCH; /* get r13 */ \
113 std r10,PACA_EXCRIT+EX_R13(r13); \
114 ld r11,PACA_SPRG3(r13); \
115 mtspr SPRN_SPRG_CRIT_SCRATCH,r11;
92 116
93/* Variants of the "addition" argument for the prolog 117/* Variants of the "addition" argument for the prolog
94 */ 118 */
95#define PROLOG_ADDITION_NONE_GEN(n) 119#define PROLOG_ADDITION_NONE_GEN(n)
120#define PROLOG_ADDITION_NONE_GDBELL(n)
96#define PROLOG_ADDITION_NONE_CRIT(n) 121#define PROLOG_ADDITION_NONE_CRIT(n)
97#define PROLOG_ADDITION_NONE_DBG(n) 122#define PROLOG_ADDITION_NONE_DBG(n)
98#define PROLOG_ADDITION_NONE_MC(n) 123#define PROLOG_ADDITION_NONE_MC(n)
99 124
100#define PROLOG_ADDITION_MASKABLE_GEN(n) \ 125#define PROLOG_ADDITION_MASKABLE_GEN(n) \
101 lbz r11,PACASOFTIRQEN(r13); /* are irqs soft-disabled ? */ \ 126 lbz r10,PACASOFTIRQEN(r13); /* are irqs soft-disabled ? */ \
102 cmpwi cr0,r11,0; /* yes -> go out of line */ \ 127 cmpwi cr0,r10,0; /* yes -> go out of line */ \
103 beq masked_interrupt_book3e_##n 128 beq masked_interrupt_book3e_##n
104 129
105#define PROLOG_ADDITION_2REGS_GEN(n) \ 130#define PROLOG_ADDITION_2REGS_GEN(n) \
@@ -233,9 +258,9 @@ exc_##n##_bad_stack: \
2331: 2581:
234 259
235 260
236#define MASKABLE_EXCEPTION(trapnum, label, hdlr, ack) \ 261#define MASKABLE_EXCEPTION(trapnum, intnum, label, hdlr, ack) \
237 START_EXCEPTION(label); \ 262 START_EXCEPTION(label); \
238 NORMAL_EXCEPTION_PROLOG(trapnum, PROLOG_ADDITION_MASKABLE) \ 263 NORMAL_EXCEPTION_PROLOG(trapnum, intnum, PROLOG_ADDITION_MASKABLE)\
239 EXCEPTION_COMMON(trapnum, PACA_EXGEN, INTS_DISABLE) \ 264 EXCEPTION_COMMON(trapnum, PACA_EXGEN, INTS_DISABLE) \
240 ack(r8); \ 265 ack(r8); \
241 CHECK_NAPPING(); \ 266 CHECK_NAPPING(); \
@@ -286,7 +311,8 @@ interrupt_end_book3e:
286 311
287/* Critical Input Interrupt */ 312/* Critical Input Interrupt */
288 START_EXCEPTION(critical_input); 313 START_EXCEPTION(critical_input);
289 CRIT_EXCEPTION_PROLOG(0x100, PROLOG_ADDITION_NONE) 314 CRIT_EXCEPTION_PROLOG(0x100, BOOKE_INTERRUPT_CRITICAL,
315 PROLOG_ADDITION_NONE)
290// EXCEPTION_COMMON(0x100, PACA_EXCRIT, INTS_DISABLE) 316// EXCEPTION_COMMON(0x100, PACA_EXCRIT, INTS_DISABLE)
291// bl special_reg_save_crit 317// bl special_reg_save_crit
292// CHECK_NAPPING(); 318// CHECK_NAPPING();
@@ -297,7 +323,8 @@ interrupt_end_book3e:
297 323
298/* Machine Check Interrupt */ 324/* Machine Check Interrupt */
299 START_EXCEPTION(machine_check); 325 START_EXCEPTION(machine_check);
300 CRIT_EXCEPTION_PROLOG(0x200, PROLOG_ADDITION_NONE) 326 MC_EXCEPTION_PROLOG(0x200, BOOKE_INTERRUPT_MACHINE_CHECK,
327 PROLOG_ADDITION_NONE)
301// EXCEPTION_COMMON(0x200, PACA_EXMC, INTS_DISABLE) 328// EXCEPTION_COMMON(0x200, PACA_EXMC, INTS_DISABLE)
302// bl special_reg_save_mc 329// bl special_reg_save_mc
303// addi r3,r1,STACK_FRAME_OVERHEAD 330// addi r3,r1,STACK_FRAME_OVERHEAD
@@ -308,7 +335,8 @@ interrupt_end_book3e:
308 335
309/* Data Storage Interrupt */ 336/* Data Storage Interrupt */
310 START_EXCEPTION(data_storage) 337 START_EXCEPTION(data_storage)
311 NORMAL_EXCEPTION_PROLOG(0x300, PROLOG_ADDITION_2REGS) 338 NORMAL_EXCEPTION_PROLOG(0x300, BOOKE_INTERRUPT_DATA_STORAGE,
339 PROLOG_ADDITION_2REGS)
312 mfspr r14,SPRN_DEAR 340 mfspr r14,SPRN_DEAR
313 mfspr r15,SPRN_ESR 341 mfspr r15,SPRN_ESR
314 EXCEPTION_COMMON(0x300, PACA_EXGEN, INTS_DISABLE) 342 EXCEPTION_COMMON(0x300, PACA_EXGEN, INTS_DISABLE)
@@ -316,18 +344,21 @@ interrupt_end_book3e:
316 344
317/* Instruction Storage Interrupt */ 345/* Instruction Storage Interrupt */
318 START_EXCEPTION(instruction_storage); 346 START_EXCEPTION(instruction_storage);
319 NORMAL_EXCEPTION_PROLOG(0x400, PROLOG_ADDITION_2REGS) 347 NORMAL_EXCEPTION_PROLOG(0x400, BOOKE_INTERRUPT_INST_STORAGE,
348 PROLOG_ADDITION_2REGS)
320 li r15,0 349 li r15,0
321 mr r14,r10 350 mr r14,r10
322 EXCEPTION_COMMON(0x400, PACA_EXGEN, INTS_DISABLE) 351 EXCEPTION_COMMON(0x400, PACA_EXGEN, INTS_DISABLE)
323 b storage_fault_common 352 b storage_fault_common
324 353
325/* External Input Interrupt */ 354/* External Input Interrupt */
326 MASKABLE_EXCEPTION(0x500, external_input, .do_IRQ, ACK_NONE) 355 MASKABLE_EXCEPTION(0x500, BOOKE_INTERRUPT_EXTERNAL,
356 external_input, .do_IRQ, ACK_NONE)
327 357
328/* Alignment */ 358/* Alignment */
329 START_EXCEPTION(alignment); 359 START_EXCEPTION(alignment);
330 NORMAL_EXCEPTION_PROLOG(0x600, PROLOG_ADDITION_2REGS) 360 NORMAL_EXCEPTION_PROLOG(0x600, BOOKE_INTERRUPT_ALIGNMENT,
361 PROLOG_ADDITION_2REGS)
331 mfspr r14,SPRN_DEAR 362 mfspr r14,SPRN_DEAR
332 mfspr r15,SPRN_ESR 363 mfspr r15,SPRN_ESR
333 EXCEPTION_COMMON(0x600, PACA_EXGEN, INTS_KEEP) 364 EXCEPTION_COMMON(0x600, PACA_EXGEN, INTS_KEEP)
@@ -335,7 +366,8 @@ interrupt_end_book3e:
335 366
336/* Program Interrupt */ 367/* Program Interrupt */
337 START_EXCEPTION(program); 368 START_EXCEPTION(program);
338 NORMAL_EXCEPTION_PROLOG(0x700, PROLOG_ADDITION_1REG) 369 NORMAL_EXCEPTION_PROLOG(0x700, BOOKE_INTERRUPT_PROGRAM,
370 PROLOG_ADDITION_1REG)
339 mfspr r14,SPRN_ESR 371 mfspr r14,SPRN_ESR
340 EXCEPTION_COMMON(0x700, PACA_EXGEN, INTS_DISABLE) 372 EXCEPTION_COMMON(0x700, PACA_EXGEN, INTS_DISABLE)
341 std r14,_DSISR(r1) 373 std r14,_DSISR(r1)
@@ -347,7 +379,8 @@ interrupt_end_book3e:
347 379
348/* Floating Point Unavailable Interrupt */ 380/* Floating Point Unavailable Interrupt */
349 START_EXCEPTION(fp_unavailable); 381 START_EXCEPTION(fp_unavailable);
350 NORMAL_EXCEPTION_PROLOG(0x800, PROLOG_ADDITION_NONE) 382 NORMAL_EXCEPTION_PROLOG(0x800, BOOKE_INTERRUPT_FP_UNAVAIL,
383 PROLOG_ADDITION_NONE)
351 /* we can probably do a shorter exception entry for that one... */ 384 /* we can probably do a shorter exception entry for that one... */
352 EXCEPTION_COMMON(0x800, PACA_EXGEN, INTS_KEEP) 385 EXCEPTION_COMMON(0x800, PACA_EXGEN, INTS_KEEP)
353 ld r12,_MSR(r1) 386 ld r12,_MSR(r1)
@@ -362,14 +395,17 @@ interrupt_end_book3e:
362 b .ret_from_except 395 b .ret_from_except
363 396
364/* Decrementer Interrupt */ 397/* Decrementer Interrupt */
365 MASKABLE_EXCEPTION(0x900, decrementer, .timer_interrupt, ACK_DEC) 398 MASKABLE_EXCEPTION(0x900, BOOKE_INTERRUPT_DECREMENTER,
399 decrementer, .timer_interrupt, ACK_DEC)
366 400
367/* Fixed Interval Timer Interrupt */ 401/* Fixed Interval Timer Interrupt */
368 MASKABLE_EXCEPTION(0x980, fixed_interval, .unknown_exception, ACK_FIT) 402 MASKABLE_EXCEPTION(0x980, BOOKE_INTERRUPT_FIT,
403 fixed_interval, .unknown_exception, ACK_FIT)
369 404
370/* Watchdog Timer Interrupt */ 405/* Watchdog Timer Interrupt */
371 START_EXCEPTION(watchdog); 406 START_EXCEPTION(watchdog);
372 CRIT_EXCEPTION_PROLOG(0x9f0, PROLOG_ADDITION_NONE) 407 CRIT_EXCEPTION_PROLOG(0x9f0, BOOKE_INTERRUPT_WATCHDOG,
408 PROLOG_ADDITION_NONE)
373// EXCEPTION_COMMON(0x9f0, PACA_EXCRIT, INTS_DISABLE) 409// EXCEPTION_COMMON(0x9f0, PACA_EXCRIT, INTS_DISABLE)
374// bl special_reg_save_crit 410// bl special_reg_save_crit
375// CHECK_NAPPING(); 411// CHECK_NAPPING();
@@ -388,7 +424,8 @@ interrupt_end_book3e:
388 424
389/* Auxiliary Processor Unavailable Interrupt */ 425/* Auxiliary Processor Unavailable Interrupt */
390 START_EXCEPTION(ap_unavailable); 426 START_EXCEPTION(ap_unavailable);
391 NORMAL_EXCEPTION_PROLOG(0xf20, PROLOG_ADDITION_NONE) 427 NORMAL_EXCEPTION_PROLOG(0xf20, BOOKE_INTERRUPT_AP_UNAVAIL,
428 PROLOG_ADDITION_NONE)
392 EXCEPTION_COMMON(0xf20, PACA_EXGEN, INTS_DISABLE) 429 EXCEPTION_COMMON(0xf20, PACA_EXGEN, INTS_DISABLE)
393 bl .save_nvgprs 430 bl .save_nvgprs
394 addi r3,r1,STACK_FRAME_OVERHEAD 431 addi r3,r1,STACK_FRAME_OVERHEAD
@@ -397,7 +434,8 @@ interrupt_end_book3e:
397 434
398/* Debug exception as a critical interrupt*/ 435/* Debug exception as a critical interrupt*/
399 START_EXCEPTION(debug_crit); 436 START_EXCEPTION(debug_crit);
400 CRIT_EXCEPTION_PROLOG(0xd00, PROLOG_ADDITION_2REGS) 437 CRIT_EXCEPTION_PROLOG(0xd00, BOOKE_INTERRUPT_DEBUG,
438 PROLOG_ADDITION_2REGS)
401 439
402 /* 440 /*
403 * If there is a single step or branch-taken exception in an 441 * If there is a single step or branch-taken exception in an
@@ -431,7 +469,7 @@ interrupt_end_book3e:
431 mtcr r10 469 mtcr r10
432 ld r10,PACA_EXCRIT+EX_R10(r13) /* restore registers */ 470 ld r10,PACA_EXCRIT+EX_R10(r13) /* restore registers */
433 ld r11,PACA_EXCRIT+EX_R11(r13) 471 ld r11,PACA_EXCRIT+EX_R11(r13)
434 mfspr r13,SPRN_SPRG_CRIT_SCRATCH 472 ld r13,PACA_EXCRIT+EX_R13(r13)
435 rfci 473 rfci
436 474
437 /* Normal debug exception */ 475 /* Normal debug exception */
@@ -444,7 +482,7 @@ interrupt_end_book3e:
444 /* Now we mash up things to make it look like we are coming on a 482 /* Now we mash up things to make it look like we are coming on a
445 * normal exception 483 * normal exception
446 */ 484 */
447 mfspr r15,SPRN_SPRG_CRIT_SCRATCH 485 ld r15,PACA_EXCRIT+EX_R13(r13)
448 mtspr SPRN_SPRG_GEN_SCRATCH,r15 486 mtspr SPRN_SPRG_GEN_SCRATCH,r15
449 mfspr r14,SPRN_DBSR 487 mfspr r14,SPRN_DBSR
450 EXCEPTION_COMMON(0xd00, PACA_EXCRIT, INTS_DISABLE) 488 EXCEPTION_COMMON(0xd00, PACA_EXCRIT, INTS_DISABLE)
@@ -462,7 +500,8 @@ kernel_dbg_exc:
462 500
463/* Debug exception as a debug interrupt*/ 501/* Debug exception as a debug interrupt*/
464 START_EXCEPTION(debug_debug); 502 START_EXCEPTION(debug_debug);
465 DBG_EXCEPTION_PROLOG(0xd08, PROLOG_ADDITION_2REGS) 503 DBG_EXCEPTION_PROLOG(0xd00, BOOKE_INTERRUPT_DEBUG,
504 PROLOG_ADDITION_2REGS)
466 505
467 /* 506 /*
468 * If there is a single step or branch-taken exception in an 507 * If there is a single step or branch-taken exception in an
@@ -523,18 +562,21 @@ kernel_dbg_exc:
523 b .ret_from_except 562 b .ret_from_except
524 563
525 START_EXCEPTION(perfmon); 564 START_EXCEPTION(perfmon);
526 NORMAL_EXCEPTION_PROLOG(0x260, PROLOG_ADDITION_NONE) 565 NORMAL_EXCEPTION_PROLOG(0x260, BOOKE_INTERRUPT_PERFORMANCE_MONITOR,
566 PROLOG_ADDITION_NONE)
527 EXCEPTION_COMMON(0x260, PACA_EXGEN, INTS_DISABLE) 567 EXCEPTION_COMMON(0x260, PACA_EXGEN, INTS_DISABLE)
528 addi r3,r1,STACK_FRAME_OVERHEAD 568 addi r3,r1,STACK_FRAME_OVERHEAD
529 bl .performance_monitor_exception 569 bl .performance_monitor_exception
530 b .ret_from_except_lite 570 b .ret_from_except_lite
531 571
532/* Doorbell interrupt */ 572/* Doorbell interrupt */
533 MASKABLE_EXCEPTION(0x280, doorbell, .doorbell_exception, ACK_NONE) 573 MASKABLE_EXCEPTION(0x280, BOOKE_INTERRUPT_DOORBELL,
574 doorbell, .doorbell_exception, ACK_NONE)
534 575
535/* Doorbell critical Interrupt */ 576/* Doorbell critical Interrupt */
536 START_EXCEPTION(doorbell_crit); 577 START_EXCEPTION(doorbell_crit);
537 CRIT_EXCEPTION_PROLOG(0x2a0, PROLOG_ADDITION_NONE) 578 CRIT_EXCEPTION_PROLOG(0x2a0, BOOKE_INTERRUPT_DOORBELL_CRITICAL,
579 PROLOG_ADDITION_NONE)
538// EXCEPTION_COMMON(0x2a0, PACA_EXCRIT, INTS_DISABLE) 580// EXCEPTION_COMMON(0x2a0, PACA_EXCRIT, INTS_DISABLE)
539// bl special_reg_save_crit 581// bl special_reg_save_crit
540// CHECK_NAPPING(); 582// CHECK_NAPPING();
@@ -543,12 +585,24 @@ kernel_dbg_exc:
543// b ret_from_crit_except 585// b ret_from_crit_except
544 b . 586 b .
545 587
546/* Guest Doorbell */ 588/*
547 MASKABLE_EXCEPTION(0x2c0, guest_doorbell, .unknown_exception, ACK_NONE) 589 * Guest doorbell interrupt
590 * This general exception use GSRRx save/restore registers
591 */
592 START_EXCEPTION(guest_doorbell);
593 GDBELL_EXCEPTION_PROLOG(0x2c0, BOOKE_INTERRUPT_GUEST_DBELL,
594 PROLOG_ADDITION_NONE)
595 EXCEPTION_COMMON(0x2c0, PACA_EXGEN, INTS_KEEP)
596 addi r3,r1,STACK_FRAME_OVERHEAD
597 bl .save_nvgprs
598 INTS_RESTORE_HARD
599 bl .unknown_exception
600 b .ret_from_except
548 601
549/* Guest Doorbell critical Interrupt */ 602/* Guest Doorbell critical Interrupt */
550 START_EXCEPTION(guest_doorbell_crit); 603 START_EXCEPTION(guest_doorbell_crit);
551 CRIT_EXCEPTION_PROLOG(0x2e0, PROLOG_ADDITION_NONE) 604 CRIT_EXCEPTION_PROLOG(0x2e0, BOOKE_INTERRUPT_GUEST_DBELL_CRIT,
605 PROLOG_ADDITION_NONE)
552// EXCEPTION_COMMON(0x2e0, PACA_EXCRIT, INTS_DISABLE) 606// EXCEPTION_COMMON(0x2e0, PACA_EXCRIT, INTS_DISABLE)
553// bl special_reg_save_crit 607// bl special_reg_save_crit
554// CHECK_NAPPING(); 608// CHECK_NAPPING();
@@ -559,7 +613,8 @@ kernel_dbg_exc:
559 613
560/* Hypervisor call */ 614/* Hypervisor call */
561 START_EXCEPTION(hypercall); 615 START_EXCEPTION(hypercall);
562 NORMAL_EXCEPTION_PROLOG(0x310, PROLOG_ADDITION_NONE) 616 NORMAL_EXCEPTION_PROLOG(0x310, BOOKE_INTERRUPT_HV_SYSCALL,
617 PROLOG_ADDITION_NONE)
563 EXCEPTION_COMMON(0x310, PACA_EXGEN, INTS_KEEP) 618 EXCEPTION_COMMON(0x310, PACA_EXGEN, INTS_KEEP)
564 addi r3,r1,STACK_FRAME_OVERHEAD 619 addi r3,r1,STACK_FRAME_OVERHEAD
565 bl .save_nvgprs 620 bl .save_nvgprs
@@ -569,7 +624,8 @@ kernel_dbg_exc:
569 624
570/* Embedded Hypervisor priviledged */ 625/* Embedded Hypervisor priviledged */
571 START_EXCEPTION(ehpriv); 626 START_EXCEPTION(ehpriv);
572 NORMAL_EXCEPTION_PROLOG(0x320, PROLOG_ADDITION_NONE) 627 NORMAL_EXCEPTION_PROLOG(0x320, BOOKE_INTERRUPT_HV_PRIV,
628 PROLOG_ADDITION_NONE)
573 EXCEPTION_COMMON(0x320, PACA_EXGEN, INTS_KEEP) 629 EXCEPTION_COMMON(0x320, PACA_EXGEN, INTS_KEEP)
574 addi r3,r1,STACK_FRAME_OVERHEAD 630 addi r3,r1,STACK_FRAME_OVERHEAD
575 bl .save_nvgprs 631 bl .save_nvgprs
@@ -582,44 +638,42 @@ kernel_dbg_exc:
582 * accordingly and if the interrupt is level sensitive, we hard disable 638 * accordingly and if the interrupt is level sensitive, we hard disable
583 */ 639 */
584 640
641.macro masked_interrupt_book3e paca_irq full_mask
642 lbz r10,PACAIRQHAPPENED(r13)
643 ori r10,r10,\paca_irq
644 stb r10,PACAIRQHAPPENED(r13)
645
646 .if \full_mask == 1
647 rldicl r10,r11,48,1 /* clear MSR_EE */
648 rotldi r11,r10,16
649 mtspr SPRN_SRR1,r11
650 .endif
651
652 lwz r11,PACA_EXGEN+EX_CR(r13)
653 mtcr r11
654 ld r10,PACA_EXGEN+EX_R10(r13)
655 ld r11,PACA_EXGEN+EX_R11(r13)
656 mfspr r13,SPRN_SPRG_GEN_SCRATCH
657 rfi
658 b .
659.endm
660
585masked_interrupt_book3e_0x500: 661masked_interrupt_book3e_0x500:
586 /* XXX When adding support for EPR, use PACA_IRQ_EE_EDGE */ 662 // XXX When adding support for EPR, use PACA_IRQ_EE_EDGE
587 li r11,PACA_IRQ_EE 663 masked_interrupt_book3e PACA_IRQ_EE 1
588 b masked_interrupt_book3e_full_mask
589 664
590masked_interrupt_book3e_0x900: 665masked_interrupt_book3e_0x900:
591 ACK_DEC(r11); 666 ACK_DEC(r10);
592 li r11,PACA_IRQ_DEC 667 masked_interrupt_book3e PACA_IRQ_DEC 0
593 b masked_interrupt_book3e_no_mask 668
594masked_interrupt_book3e_0x980: 669masked_interrupt_book3e_0x980:
595 ACK_FIT(r11); 670 ACK_FIT(r10);
596 li r11,PACA_IRQ_DEC 671 masked_interrupt_book3e PACA_IRQ_DEC 0
597 b masked_interrupt_book3e_no_mask 672
598masked_interrupt_book3e_0x280: 673masked_interrupt_book3e_0x280:
599masked_interrupt_book3e_0x2c0: 674masked_interrupt_book3e_0x2c0:
600 li r11,PACA_IRQ_DBELL 675 masked_interrupt_book3e PACA_IRQ_DBELL 0
601 b masked_interrupt_book3e_no_mask
602 676
603masked_interrupt_book3e_no_mask:
604 mtcr r10
605 lbz r10,PACAIRQHAPPENED(r13)
606 or r10,r10,r11
607 stb r10,PACAIRQHAPPENED(r13)
608 b 1f
609masked_interrupt_book3e_full_mask:
610 mtcr r10
611 lbz r10,PACAIRQHAPPENED(r13)
612 or r10,r10,r11
613 stb r10,PACAIRQHAPPENED(r13)
614 mfspr r10,SPRN_SRR1
615 rldicl r11,r10,48,1 /* clear MSR_EE */
616 rotldi r10,r11,16
617 mtspr SPRN_SRR1,r10
6181: ld r10,PACA_EXGEN+EX_R10(r13);
619 ld r11,PACA_EXGEN+EX_R11(r13);
620 mfspr r13,SPRN_SPRG_GEN_SCRATCH;
621 rfi
622 b .
623/* 677/*
624 * Called from arch_local_irq_enable when an interrupt needs 678 * Called from arch_local_irq_enable when an interrupt needs
625 * to be resent. r3 contains either 0x500,0x900,0x260 or 0x280 679 * to be resent. r3 contains either 0x500,0x900,0x260 or 0x280
diff --git a/arch/powerpc/kernel/fadump.c b/arch/powerpc/kernel/fadump.c
index 18bdf74fa164..06c8202a69cf 100644
--- a/arch/powerpc/kernel/fadump.c
+++ b/arch/powerpc/kernel/fadump.c
@@ -289,8 +289,7 @@ int __init fadump_reserve_mem(void)
289 else 289 else
290 memory_limit = memblock_end_of_DRAM(); 290 memory_limit = memblock_end_of_DRAM();
291 printk(KERN_INFO "Adjusted memory_limit for firmware-assisted" 291 printk(KERN_INFO "Adjusted memory_limit for firmware-assisted"
292 " dump, now %#016llx\n", 292 " dump, now %#016llx\n", memory_limit);
293 (unsigned long long)memory_limit);
294 } 293 }
295 if (memory_limit) 294 if (memory_limit)
296 memory_boundary = memory_limit; 295 memory_boundary = memory_limit;
diff --git a/arch/powerpc/kernel/hw_breakpoint.c b/arch/powerpc/kernel/hw_breakpoint.c
index 956a4c496de9..a89cae481b04 100644
--- a/arch/powerpc/kernel/hw_breakpoint.c
+++ b/arch/powerpc/kernel/hw_breakpoint.c
@@ -73,7 +73,7 @@ int arch_install_hw_breakpoint(struct perf_event *bp)
73 * If so, DABR will be populated in single_step_dabr_instruction(). 73 * If so, DABR will be populated in single_step_dabr_instruction().
74 */ 74 */
75 if (current->thread.last_hit_ubp != bp) 75 if (current->thread.last_hit_ubp != bp)
76 set_dabr(info->address | info->type | DABR_TRANSLATION); 76 set_dabr(info->address | info->type | DABR_TRANSLATION, info->dabrx);
77 77
78 return 0; 78 return 0;
79} 79}
@@ -97,7 +97,7 @@ void arch_uninstall_hw_breakpoint(struct perf_event *bp)
97 } 97 }
98 98
99 *slot = NULL; 99 *slot = NULL;
100 set_dabr(0); 100 set_dabr(0, 0);
101} 101}
102 102
103/* 103/*
@@ -170,6 +170,13 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp)
170 170
171 info->address = bp->attr.bp_addr; 171 info->address = bp->attr.bp_addr;
172 info->len = bp->attr.bp_len; 172 info->len = bp->attr.bp_len;
173 info->dabrx = DABRX_ALL;
174 if (bp->attr.exclude_user)
175 info->dabrx &= ~DABRX_USER;
176 if (bp->attr.exclude_kernel)
177 info->dabrx &= ~DABRX_KERNEL;
178 if (bp->attr.exclude_hv)
179 info->dabrx &= ~DABRX_HYP;
173 180
174 /* 181 /*
175 * Since breakpoint length can be a maximum of HW_BREAKPOINT_LEN(8) 182 * Since breakpoint length can be a maximum of HW_BREAKPOINT_LEN(8)
@@ -197,7 +204,7 @@ void thread_change_pc(struct task_struct *tsk, struct pt_regs *regs)
197 204
198 info = counter_arch_bp(tsk->thread.last_hit_ubp); 205 info = counter_arch_bp(tsk->thread.last_hit_ubp);
199 regs->msr &= ~MSR_SE; 206 regs->msr &= ~MSR_SE;
200 set_dabr(info->address | info->type | DABR_TRANSLATION); 207 set_dabr(info->address | info->type | DABR_TRANSLATION, info->dabrx);
201 tsk->thread.last_hit_ubp = NULL; 208 tsk->thread.last_hit_ubp = NULL;
202} 209}
203 210
@@ -215,7 +222,7 @@ int __kprobes hw_breakpoint_handler(struct die_args *args)
215 unsigned long dar = regs->dar; 222 unsigned long dar = regs->dar;
216 223
217 /* Disable breakpoints during exception handling */ 224 /* Disable breakpoints during exception handling */
218 set_dabr(0); 225 set_dabr(0, 0);
219 226
220 /* 227 /*
221 * The counter may be concurrently released but that can only 228 * The counter may be concurrently released but that can only
@@ -281,7 +288,7 @@ int __kprobes hw_breakpoint_handler(struct die_args *args)
281 if (!info->extraneous_interrupt) 288 if (!info->extraneous_interrupt)
282 perf_bp_event(bp, regs); 289 perf_bp_event(bp, regs);
283 290
284 set_dabr(info->address | info->type | DABR_TRANSLATION); 291 set_dabr(info->address | info->type | DABR_TRANSLATION, info->dabrx);
285out: 292out:
286 rcu_read_unlock(); 293 rcu_read_unlock();
287 return rc; 294 return rc;
@@ -294,7 +301,7 @@ int __kprobes single_step_dabr_instruction(struct die_args *args)
294{ 301{
295 struct pt_regs *regs = args->regs; 302 struct pt_regs *regs = args->regs;
296 struct perf_event *bp = NULL; 303 struct perf_event *bp = NULL;
297 struct arch_hw_breakpoint *bp_info; 304 struct arch_hw_breakpoint *info;
298 305
299 bp = current->thread.last_hit_ubp; 306 bp = current->thread.last_hit_ubp;
300 /* 307 /*
@@ -304,16 +311,16 @@ int __kprobes single_step_dabr_instruction(struct die_args *args)
304 if (!bp) 311 if (!bp)
305 return NOTIFY_DONE; 312 return NOTIFY_DONE;
306 313
307 bp_info = counter_arch_bp(bp); 314 info = counter_arch_bp(bp);
308 315
309 /* 316 /*
310 * We shall invoke the user-defined callback function in the single 317 * We shall invoke the user-defined callback function in the single
311 * stepping handler to confirm to 'trigger-after-execute' semantics 318 * stepping handler to confirm to 'trigger-after-execute' semantics
312 */ 319 */
313 if (!bp_info->extraneous_interrupt) 320 if (!info->extraneous_interrupt)
314 perf_bp_event(bp, regs); 321 perf_bp_event(bp, regs);
315 322
316 set_dabr(bp_info->address | bp_info->type | DABR_TRANSLATION); 323 set_dabr(info->address | info->type | DABR_TRANSLATION, info->dabrx);
317 current->thread.last_hit_ubp = NULL; 324 current->thread.last_hit_ubp = NULL;
318 325
319 /* 326 /*
diff --git a/arch/powerpc/kernel/ibmebus.c b/arch/powerpc/kernel/ibmebus.c
index b01d14eeca8d..8220baa46faf 100644
--- a/arch/powerpc/kernel/ibmebus.c
+++ b/arch/powerpc/kernel/ibmebus.c
@@ -47,7 +47,6 @@
47#include <linux/stat.h> 47#include <linux/stat.h>
48#include <linux/of_platform.h> 48#include <linux/of_platform.h>
49#include <asm/ibmebus.h> 49#include <asm/ibmebus.h>
50#include <asm/abs_addr.h>
51 50
52static struct device ibmebus_bus_device = { /* fake "parent" device */ 51static struct device ibmebus_bus_device = { /* fake "parent" device */
53 .init_name = "ibmebus", 52 .init_name = "ibmebus",
diff --git a/arch/powerpc/kernel/machine_kexec.c b/arch/powerpc/kernel/machine_kexec.c
index 5df777794403..fa9f6c72f557 100644
--- a/arch/powerpc/kernel/machine_kexec.c
+++ b/arch/powerpc/kernel/machine_kexec.c
@@ -165,7 +165,7 @@ void __init reserve_crashkernel(void)
165 if (memory_limit && memory_limit <= crashk_res.end) { 165 if (memory_limit && memory_limit <= crashk_res.end) {
166 memory_limit = crashk_res.end + 1; 166 memory_limit = crashk_res.end + 1;
167 printk("Adjusted memory limit for crashkernel, now 0x%llx\n", 167 printk("Adjusted memory limit for crashkernel, now 0x%llx\n",
168 (unsigned long long)memory_limit); 168 memory_limit);
169 } 169 }
170 170
171 printk(KERN_INFO "Reserving %ldMB of memory at %ldMB " 171 printk(KERN_INFO "Reserving %ldMB of memory at %ldMB "
@@ -204,6 +204,12 @@ static struct property crashk_size_prop = {
204 .value = &crashk_size, 204 .value = &crashk_size,
205}; 205};
206 206
207static struct property memory_limit_prop = {
208 .name = "linux,memory-limit",
209 .length = sizeof(unsigned long long),
210 .value = &memory_limit,
211};
212
207static void __init export_crashk_values(struct device_node *node) 213static void __init export_crashk_values(struct device_node *node)
208{ 214{
209 struct property *prop; 215 struct property *prop;
@@ -223,6 +229,12 @@ static void __init export_crashk_values(struct device_node *node)
223 crashk_size = resource_size(&crashk_res); 229 crashk_size = resource_size(&crashk_res);
224 prom_add_property(node, &crashk_size_prop); 230 prom_add_property(node, &crashk_size_prop);
225 } 231 }
232
233 /*
234 * memory_limit is required by the kexec-tools to limit the
235 * crash regions to the actual memory used.
236 */
237 prom_update_property(node, &memory_limit_prop);
226} 238}
227 239
228static int __init kexec_setup(void) 240static int __init kexec_setup(void)
diff --git a/arch/powerpc/kernel/paca.c b/arch/powerpc/kernel/paca.c
index fbe1a12dc7f1..cd6da855090c 100644
--- a/arch/powerpc/kernel/paca.c
+++ b/arch/powerpc/kernel/paca.c
@@ -142,6 +142,7 @@ void __init initialise_paca(struct paca_struct *new_paca, int cpu)
142 new_paca->hw_cpu_id = 0xffff; 142 new_paca->hw_cpu_id = 0xffff;
143 new_paca->kexec_state = KEXEC_STATE_NONE; 143 new_paca->kexec_state = KEXEC_STATE_NONE;
144 new_paca->__current = &init_task; 144 new_paca->__current = &init_task;
145 new_paca->data_offset = 0xfeeeeeeeeeeeeeeeULL;
145#ifdef CONFIG_PPC_STD_MMU_64 146#ifdef CONFIG_PPC_STD_MMU_64
146 new_paca->slb_shadow_ptr = &slb_shadow[cpu]; 147 new_paca->slb_shadow_ptr = &slb_shadow[cpu];
147#endif /* CONFIG_PPC_STD_MMU_64 */ 148#endif /* CONFIG_PPC_STD_MMU_64 */
diff --git a/arch/powerpc/kernel/pci-common.c b/arch/powerpc/kernel/pci-common.c
index 43fea543d686..7f94f760dd0c 100644
--- a/arch/powerpc/kernel/pci-common.c
+++ b/arch/powerpc/kernel/pci-common.c
@@ -980,13 +980,14 @@ static void __devinit pcibios_fixup_bridge(struct pci_bus *bus)
980 if (i >= 3 && bus->self->transparent) 980 if (i >= 3 && bus->self->transparent)
981 continue; 981 continue;
982 982
983 /* If we are going to re-assign everything, mark the resource 983 /* If we're going to reassign everything, we can
984 * as unset and move it down to 0 984 * shrink the P2P resource to have size as being
985 * of 0 in order to save space.
985 */ 986 */
986 if (pci_has_flag(PCI_REASSIGN_ALL_RSRC)) { 987 if (pci_has_flag(PCI_REASSIGN_ALL_RSRC)) {
987 res->flags |= IORESOURCE_UNSET; 988 res->flags |= IORESOURCE_UNSET;
988 res->end -= res->start;
989 res->start = 0; 989 res->start = 0;
990 res->end = -1;
990 continue; 991 continue;
991 } 992 }
992 993
@@ -1248,7 +1249,14 @@ void pcibios_allocate_bus_resources(struct pci_bus *bus)
1248 pr_warning("PCI: Cannot allocate resource region " 1249 pr_warning("PCI: Cannot allocate resource region "
1249 "%d of PCI bridge %d, will remap\n", i, bus->number); 1250 "%d of PCI bridge %d, will remap\n", i, bus->number);
1250 clear_resource: 1251 clear_resource:
1251 res->start = res->end = 0; 1252 /* The resource might be figured out when doing
1253 * reassignment based on the resources required
1254 * by the downstream PCI devices. Here we set
1255 * the size of the resource to be 0 in order to
1256 * save more space.
1257 */
1258 res->start = 0;
1259 res->end = -1;
1252 res->flags = 0; 1260 res->flags = 0;
1253 } 1261 }
1254 1262
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 1a1f2ddfb581..50e504c29bb9 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -258,6 +258,7 @@ void do_send_trap(struct pt_regs *regs, unsigned long address,
258{ 258{
259 siginfo_t info; 259 siginfo_t info;
260 260
261 current->thread.trap_nr = signal_code;
261 if (notify_die(DIE_DABR_MATCH, "dabr_match", regs, error_code, 262 if (notify_die(DIE_DABR_MATCH, "dabr_match", regs, error_code,
262 11, SIGSEGV) == NOTIFY_STOP) 263 11, SIGSEGV) == NOTIFY_STOP)
263 return; 264 return;
@@ -275,6 +276,7 @@ void do_dabr(struct pt_regs *regs, unsigned long address,
275{ 276{
276 siginfo_t info; 277 siginfo_t info;
277 278
279 current->thread.trap_nr = TRAP_HWBKPT;
278 if (notify_die(DIE_DABR_MATCH, "dabr_match", regs, error_code, 280 if (notify_die(DIE_DABR_MATCH, "dabr_match", regs, error_code,
279 11, SIGSEGV) == NOTIFY_STOP) 281 11, SIGSEGV) == NOTIFY_STOP)
280 return; 282 return;
@@ -283,7 +285,7 @@ void do_dabr(struct pt_regs *regs, unsigned long address,
283 return; 285 return;
284 286
285 /* Clear the DABR */ 287 /* Clear the DABR */
286 set_dabr(0); 288 set_dabr(0, 0);
287 289
288 /* Deliver the signal to userspace */ 290 /* Deliver the signal to userspace */
289 info.si_signo = SIGTRAP; 291 info.si_signo = SIGTRAP;
@@ -364,18 +366,19 @@ static void set_debug_reg_defaults(struct thread_struct *thread)
364{ 366{
365 if (thread->dabr) { 367 if (thread->dabr) {
366 thread->dabr = 0; 368 thread->dabr = 0;
367 set_dabr(0); 369 thread->dabrx = 0;
370 set_dabr(0, 0);
368 } 371 }
369} 372}
370#endif /* !CONFIG_HAVE_HW_BREAKPOINT */ 373#endif /* !CONFIG_HAVE_HW_BREAKPOINT */
371#endif /* CONFIG_PPC_ADV_DEBUG_REGS */ 374#endif /* CONFIG_PPC_ADV_DEBUG_REGS */
372 375
373int set_dabr(unsigned long dabr) 376int set_dabr(unsigned long dabr, unsigned long dabrx)
374{ 377{
375 __get_cpu_var(current_dabr) = dabr; 378 __get_cpu_var(current_dabr) = dabr;
376 379
377 if (ppc_md.set_dabr) 380 if (ppc_md.set_dabr)
378 return ppc_md.set_dabr(dabr); 381 return ppc_md.set_dabr(dabr, dabrx);
379 382
380 /* XXX should we have a CPU_FTR_HAS_DABR ? */ 383 /* XXX should we have a CPU_FTR_HAS_DABR ? */
381#ifdef CONFIG_PPC_ADV_DEBUG_REGS 384#ifdef CONFIG_PPC_ADV_DEBUG_REGS
@@ -385,9 +388,8 @@ int set_dabr(unsigned long dabr)
385#endif 388#endif
386#elif defined(CONFIG_PPC_BOOK3S) 389#elif defined(CONFIG_PPC_BOOK3S)
387 mtspr(SPRN_DABR, dabr); 390 mtspr(SPRN_DABR, dabr);
391 mtspr(SPRN_DABRX, dabrx);
388#endif 392#endif
389
390
391 return 0; 393 return 0;
392} 394}
393 395
@@ -480,7 +482,7 @@ struct task_struct *__switch_to(struct task_struct *prev,
480 */ 482 */
481#ifndef CONFIG_HAVE_HW_BREAKPOINT 483#ifndef CONFIG_HAVE_HW_BREAKPOINT
482 if (unlikely(__get_cpu_var(current_dabr) != new->thread.dabr)) 484 if (unlikely(__get_cpu_var(current_dabr) != new->thread.dabr))
483 set_dabr(new->thread.dabr); 485 set_dabr(new->thread.dabr, new->thread.dabrx);
484#endif /* CONFIG_HAVE_HW_BREAKPOINT */ 486#endif /* CONFIG_HAVE_HW_BREAKPOINT */
485#endif 487#endif
486 488
diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index f191bf02943a..37725e86651e 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c
@@ -78,7 +78,7 @@ static int __init early_parse_mem(char *p)
78 return 1; 78 return 1;
79 79
80 memory_limit = PAGE_ALIGN(memparse(p, &p)); 80 memory_limit = PAGE_ALIGN(memparse(p, &p));
81 DBG("memory limit = 0x%llx\n", (unsigned long long)memory_limit); 81 DBG("memory limit = 0x%llx\n", memory_limit);
82 82
83 return 0; 83 return 0;
84} 84}
@@ -661,7 +661,7 @@ void __init early_init_devtree(void *params)
661 661
662 /* make sure we've parsed cmdline for mem= before this */ 662 /* make sure we've parsed cmdline for mem= before this */
663 if (memory_limit) 663 if (memory_limit)
664 first_memblock_size = min(first_memblock_size, memory_limit); 664 first_memblock_size = min_t(u64, first_memblock_size, memory_limit);
665 setup_initial_memory_limit(memstart_addr, first_memblock_size); 665 setup_initial_memory_limit(memstart_addr, first_memblock_size);
666 /* Reserve MEMBLOCK regions used by kernel, initrd, dt, etc... */ 666 /* Reserve MEMBLOCK regions used by kernel, initrd, dt, etc... */
667 memblock_reserve(PHYSICAL_START, __pa(klimit) - PHYSICAL_START); 667 memblock_reserve(PHYSICAL_START, __pa(klimit) - PHYSICAL_START);
diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_init.c
index 0794a3017b1b..ce68278a5d73 100644
--- a/arch/powerpc/kernel/prom_init.c
+++ b/arch/powerpc/kernel/prom_init.c
@@ -1691,7 +1691,7 @@ static void __init prom_initialize_tce_table(void)
1691 * else will impact performance, so we always allocate 8MB. 1691 * else will impact performance, so we always allocate 8MB.
1692 * Anton 1692 * Anton
1693 */ 1693 */
1694 if (__is_processor(PV_POWER4) || __is_processor(PV_POWER4p)) 1694 if (pvr_version_is(PVR_POWER4) || pvr_version_is(PVR_POWER4p))
1695 minsize = 8UL << 20; 1695 minsize = 8UL << 20;
1696 else 1696 else
1697 minsize = 4UL << 20; 1697 minsize = 4UL << 20;
diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
index c10fc28b9092..79d8e56470df 100644
--- a/arch/powerpc/kernel/ptrace.c
+++ b/arch/powerpc/kernel/ptrace.c
@@ -960,6 +960,7 @@ int ptrace_set_debugreg(struct task_struct *task, unsigned long addr,
960 thread->ptrace_bps[0] = bp; 960 thread->ptrace_bps[0] = bp;
961 ptrace_put_breakpoints(task); 961 ptrace_put_breakpoints(task);
962 thread->dabr = data; 962 thread->dabr = data;
963 thread->dabrx = DABRX_ALL;
963 return 0; 964 return 0;
964 } 965 }
965 966
@@ -983,6 +984,7 @@ int ptrace_set_debugreg(struct task_struct *task, unsigned long addr,
983 984
984 /* Move contents to the DABR register */ 985 /* Move contents to the DABR register */
985 task->thread.dabr = data; 986 task->thread.dabr = data;
987 task->thread.dabrx = DABRX_ALL;
986#else /* CONFIG_PPC_ADV_DEBUG_REGS */ 988#else /* CONFIG_PPC_ADV_DEBUG_REGS */
987 /* As described above, it was assumed 3 bits were passed with the data 989 /* As described above, it was assumed 3 bits were passed with the data
988 * address, but we will assume only the mode bits will be passed 990 * address, but we will assume only the mode bits will be passed
@@ -1397,6 +1399,7 @@ static long ppc_set_hwdebug(struct task_struct *child,
1397 dabr |= DABR_DATA_WRITE; 1399 dabr |= DABR_DATA_WRITE;
1398 1400
1399 child->thread.dabr = dabr; 1401 child->thread.dabr = dabr;
1402 child->thread.dabrx = DABRX_ALL;
1400 1403
1401 return 1; 1404 return 1;
1402#endif /* !CONFIG_PPC_ADV_DEBUG_DVCS */ 1405#endif /* !CONFIG_PPC_ADV_DEBUG_DVCS */
diff --git a/arch/powerpc/kernel/rtas_flash.c b/arch/powerpc/kernel/rtas_flash.c
index 2c0ee6405633..20b0120db0c3 100644
--- a/arch/powerpc/kernel/rtas_flash.c
+++ b/arch/powerpc/kernel/rtas_flash.c
@@ -21,7 +21,6 @@
21#include <asm/delay.h> 21#include <asm/delay.h>
22#include <asm/uaccess.h> 22#include <asm/uaccess.h>
23#include <asm/rtas.h> 23#include <asm/rtas.h>
24#include <asm/abs_addr.h>
25 24
26#define MODULE_VERS "1.0" 25#define MODULE_VERS "1.0"
27#define MODULE_NAME "rtas_flash" 26#define MODULE_NAME "rtas_flash"
@@ -582,7 +581,7 @@ static void rtas_flash_firmware(int reboot_type)
582 flist = (struct flash_block_list *)&rtas_data_buf[0]; 581 flist = (struct flash_block_list *)&rtas_data_buf[0];
583 flist->num_blocks = 0; 582 flist->num_blocks = 0;
584 flist->next = rtas_firmware_flash_list; 583 flist->next = rtas_firmware_flash_list;
585 rtas_block_list = virt_to_abs(flist); 584 rtas_block_list = __pa(flist);
586 if (rtas_block_list >= 4UL*1024*1024*1024) { 585 if (rtas_block_list >= 4UL*1024*1024*1024) {
587 printk(KERN_ALERT "FLASH: kernel bug...flash list header addr above 4GB\n"); 586 printk(KERN_ALERT "FLASH: kernel bug...flash list header addr above 4GB\n");
588 spin_unlock(&rtas_data_buf_lock); 587 spin_unlock(&rtas_data_buf_lock);
@@ -596,13 +595,13 @@ static void rtas_flash_firmware(int reboot_type)
596 for (f = flist; f; f = next) { 595 for (f = flist; f; f = next) {
597 /* Translate data addrs to absolute */ 596 /* Translate data addrs to absolute */
598 for (i = 0; i < f->num_blocks; i++) { 597 for (i = 0; i < f->num_blocks; i++) {
599 f->blocks[i].data = (char *)virt_to_abs(f->blocks[i].data); 598 f->blocks[i].data = (char *)__pa(f->blocks[i].data);
600 image_size += f->blocks[i].length; 599 image_size += f->blocks[i].length;
601 } 600 }
602 next = f->next; 601 next = f->next;
603 /* Don't translate NULL pointer for last entry */ 602 /* Don't translate NULL pointer for last entry */
604 if (f->next) 603 if (f->next)
605 f->next = (struct flash_block_list *)virt_to_abs(f->next); 604 f->next = (struct flash_block_list *)__pa(f->next);
606 else 605 else
607 f->next = NULL; 606 f->next = NULL;
608 /* make num_blocks into the version/length field */ 607 /* make num_blocks into the version/length field */
diff --git a/arch/powerpc/kernel/rtas_pci.c b/arch/powerpc/kernel/rtas_pci.c
index 179af906dcda..6de63e3250bb 100644
--- a/arch/powerpc/kernel/rtas_pci.c
+++ b/arch/powerpc/kernel/rtas_pci.c
@@ -81,7 +81,7 @@ int rtas_read_config(struct pci_dn *pdn, int where, int size, u32 *val)
81 return PCIBIOS_DEVICE_NOT_FOUND; 81 return PCIBIOS_DEVICE_NOT_FOUND;
82 82
83 if (returnval == EEH_IO_ERROR_VALUE(size) && 83 if (returnval == EEH_IO_ERROR_VALUE(size) &&
84 eeh_dn_check_failure (pdn->node, NULL)) 84 eeh_dev_check_failure(of_node_to_eeh_dev(pdn->node)))
85 return PCIBIOS_DEVICE_NOT_FOUND; 85 return PCIBIOS_DEVICE_NOT_FOUND;
86 86
87 return PCIBIOS_SUCCESSFUL; 87 return PCIBIOS_SUCCESSFUL;
@@ -275,9 +275,6 @@ void __init find_and_init_phbs(void)
275 of_node_put(root); 275 of_node_put(root);
276 pci_devs_phb_init(); 276 pci_devs_phb_init();
277 277
278 /* Create EEH devices for all PHBs */
279 eeh_dev_phb_init();
280
281 /* 278 /*
282 * PCI_PROBE_ONLY and PCI_REASSIGN_ALL_BUS can be set via properties 279 * PCI_PROBE_ONLY and PCI_REASSIGN_ALL_BUS can be set via properties
283 * in chosen. 280 * in chosen.
diff --git a/arch/powerpc/kernel/signal.c b/arch/powerpc/kernel/signal.c
index 5c023c9cf16e..a2dc75793bd5 100644
--- a/arch/powerpc/kernel/signal.c
+++ b/arch/powerpc/kernel/signal.c
@@ -11,6 +11,7 @@
11 11
12#include <linux/tracehook.h> 12#include <linux/tracehook.h>
13#include <linux/signal.h> 13#include <linux/signal.h>
14#include <linux/uprobes.h>
14#include <linux/key.h> 15#include <linux/key.h>
15#include <asm/hw_breakpoint.h> 16#include <asm/hw_breakpoint.h>
16#include <asm/uaccess.h> 17#include <asm/uaccess.h>
@@ -130,7 +131,7 @@ static int do_signal(struct pt_regs *regs)
130 * triggered inside the kernel. 131 * triggered inside the kernel.
131 */ 132 */
132 if (current->thread.dabr) 133 if (current->thread.dabr)
133 set_dabr(current->thread.dabr); 134 set_dabr(current->thread.dabr, current->thread.dabrx);
134#endif 135#endif
135 /* Re-enable the breakpoints for the signal stack */ 136 /* Re-enable the breakpoints for the signal stack */
136 thread_change_pc(current, regs); 137 thread_change_pc(current, regs);
@@ -157,6 +158,11 @@ static int do_signal(struct pt_regs *regs)
157 158
158void do_notify_resume(struct pt_regs *regs, unsigned long thread_info_flags) 159void do_notify_resume(struct pt_regs *regs, unsigned long thread_info_flags)
159{ 160{
161 if (thread_info_flags & _TIF_UPROBE) {
162 clear_thread_flag(TIF_UPROBE);
163 uprobe_notify_resume(regs);
164 }
165
160 if (thread_info_flags & _TIF_SIGPENDING) 166 if (thread_info_flags & _TIF_SIGPENDING)
161 do_signal(regs); 167 do_signal(regs);
162 168
diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
index ae0843fa7a61..32518401af68 100644
--- a/arch/powerpc/kernel/traps.c
+++ b/arch/powerpc/kernel/traps.c
@@ -251,6 +251,7 @@ void _exception(int signr, struct pt_regs *regs, int code, unsigned long addr)
251 if (arch_irqs_disabled() && !arch_irq_disabled_regs(regs)) 251 if (arch_irqs_disabled() && !arch_irq_disabled_regs(regs))
252 local_irq_enable(); 252 local_irq_enable();
253 253
254 current->thread.trap_nr = code;
254 memset(&info, 0, sizeof(info)); 255 memset(&info, 0, sizeof(info));
255 info.si_signo = signr; 256 info.si_signo = signr;
256 info.si_code = code; 257 info.si_code = code;
diff --git a/arch/powerpc/kernel/uprobes.c b/arch/powerpc/kernel/uprobes.c
new file mode 100644
index 000000000000..d2d46d1014f8
--- /dev/null
+++ b/arch/powerpc/kernel/uprobes.c
@@ -0,0 +1,184 @@
1/*
2 * User-space Probes (UProbes) for powerpc
3 *
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License as published by
6 * the Free Software Foundation; either version 2 of the License, or
7 * (at your option) any later version.
8 *
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
13 *
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, write to the Free Software
16 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
17 *
18 * Copyright IBM Corporation, 2007-2012
19 *
20 * Adapted from the x86 port by Ananth N Mavinakayanahalli <ananth@in.ibm.com>
21 */
22#include <linux/kernel.h>
23#include <linux/sched.h>
24#include <linux/ptrace.h>
25#include <linux/uprobes.h>
26#include <linux/uaccess.h>
27#include <linux/kdebug.h>
28
29#include <asm/sstep.h>
30
31#define UPROBE_TRAP_NR UINT_MAX
32
33/**
34 * arch_uprobe_analyze_insn
35 * @mm: the probed address space.
36 * @arch_uprobe: the probepoint information.
37 * @addr: vaddr to probe.
38 * Return 0 on success or a -ve number on error.
39 */
40int arch_uprobe_analyze_insn(struct arch_uprobe *auprobe,
41 struct mm_struct *mm, unsigned long addr)
42{
43 if (addr & 0x03)
44 return -EINVAL;
45
46 /*
47 * We currently don't support a uprobe on an already
48 * existing breakpoint instruction underneath
49 */
50 if (is_trap(auprobe->ainsn))
51 return -ENOTSUPP;
52 return 0;
53}
54
55/*
56 * arch_uprobe_pre_xol - prepare to execute out of line.
57 * @auprobe: the probepoint information.
58 * @regs: reflects the saved user state of current task.
59 */
60int arch_uprobe_pre_xol(struct arch_uprobe *auprobe, struct pt_regs *regs)
61{
62 struct arch_uprobe_task *autask = &current->utask->autask;
63
64 autask->saved_trap_nr = current->thread.trap_nr;
65 current->thread.trap_nr = UPROBE_TRAP_NR;
66 regs->nip = current->utask->xol_vaddr;
67 return 0;
68}
69
70/**
71 * uprobe_get_swbp_addr - compute address of swbp given post-swbp regs
72 * @regs: Reflects the saved state of the task after it has hit a breakpoint
73 * instruction.
74 * Return the address of the breakpoint instruction.
75 */
76unsigned long uprobe_get_swbp_addr(struct pt_regs *regs)
77{
78 return instruction_pointer(regs);
79}
80
81/*
82 * If xol insn itself traps and generates a signal (SIGILL/SIGSEGV/etc),
83 * then detect the case where a singlestepped instruction jumps back to its
84 * own address. It is assumed that anything like do_page_fault/do_trap/etc
85 * sets thread.trap_nr != UINT_MAX.
86 *
87 * arch_uprobe_pre_xol/arch_uprobe_post_xol save/restore thread.trap_nr,
88 * arch_uprobe_xol_was_trapped() simply checks that ->trap_nr is not equal to
89 * UPROBE_TRAP_NR == UINT_MAX set by arch_uprobe_pre_xol().
90 */
91bool arch_uprobe_xol_was_trapped(struct task_struct *t)
92{
93 if (t->thread.trap_nr != UPROBE_TRAP_NR)
94 return true;
95
96 return false;
97}
98
99/*
100 * Called after single-stepping. To avoid the SMP problems that can
101 * occur when we temporarily put back the original opcode to
102 * single-step, we single-stepped a copy of the instruction.
103 *
104 * This function prepares to resume execution after the single-step.
105 */
106int arch_uprobe_post_xol(struct arch_uprobe *auprobe, struct pt_regs *regs)
107{
108 struct uprobe_task *utask = current->utask;
109
110 WARN_ON_ONCE(current->thread.trap_nr != UPROBE_TRAP_NR);
111
112 current->thread.trap_nr = utask->autask.saved_trap_nr;
113
114 /*
115 * On powerpc, except for loads and stores, most instructions
116 * including ones that alter code flow (branches, calls, returns)
117 * are emulated in the kernel. We get here only if the emulation
118 * support doesn't exist and have to fix-up the next instruction
119 * to be executed.
120 */
121 regs->nip = utask->vaddr + MAX_UINSN_BYTES;
122 return 0;
123}
124
125/* callback routine for handling exceptions. */
126int arch_uprobe_exception_notify(struct notifier_block *self,
127 unsigned long val, void *data)
128{
129 struct die_args *args = data;
130 struct pt_regs *regs = args->regs;
131
132 /* regs == NULL is a kernel bug */
133 if (WARN_ON(!regs))
134 return NOTIFY_DONE;
135
136 /* We are only interested in userspace traps */
137 if (!user_mode(regs))
138 return NOTIFY_DONE;
139
140 switch (val) {
141 case DIE_BPT:
142 if (uprobe_pre_sstep_notifier(regs))
143 return NOTIFY_STOP;
144 break;
145 case DIE_SSTEP:
146 if (uprobe_post_sstep_notifier(regs))
147 return NOTIFY_STOP;
148 default:
149 break;
150 }
151 return NOTIFY_DONE;
152}
153
154/*
155 * This function gets called when XOL instruction either gets trapped or
156 * the thread has a fatal signal, so reset the instruction pointer to its
157 * probed address.
158 */
159void arch_uprobe_abort_xol(struct arch_uprobe *auprobe, struct pt_regs *regs)
160{
161 struct uprobe_task *utask = current->utask;
162
163 current->thread.trap_nr = utask->autask.saved_trap_nr;
164 instruction_pointer_set(regs, utask->vaddr);
165}
166
167/*
168 * See if the instruction can be emulated.
169 * Returns true if instruction was emulated, false otherwise.
170 */
171bool arch_uprobe_skip_sstep(struct arch_uprobe *auprobe, struct pt_regs *regs)
172{
173 int ret;
174
175 /*
176 * emulate_step() returns 1 if the insn was successfully emulated.
177 * For all other cases, we need to single-step in hardware.
178 */
179 ret = emulate_step(regs, auprobe->ainsn);
180 if (ret > 0)
181 return true;
182
183 return false;
184}
diff --git a/arch/powerpc/kernel/vdso.c b/arch/powerpc/kernel/vdso.c
index b67db22e102d..1b2076f049ce 100644
--- a/arch/powerpc/kernel/vdso.c
+++ b/arch/powerpc/kernel/vdso.c
@@ -723,9 +723,7 @@ int __cpuinit vdso_getcpu_init(void)
723 723
724 val = (cpu & 0xfff) | ((node & 0xffff) << 16); 724 val = (cpu & 0xfff) | ((node & 0xffff) << 16);
725 mtspr(SPRN_SPRG3, val); 725 mtspr(SPRN_SPRG3, val);
726#ifdef CONFIG_KVM_BOOK3S_HANDLER 726 get_paca()->sprg3 = val;
727 get_paca()->kvm_hstate.sprg3 = val;
728#endif
729 727
730 put_cpu(); 728 put_cpu();
731 729
diff --git a/arch/powerpc/kernel/vio.c b/arch/powerpc/kernel/vio.c
index 02b32216bbc3..201ba59738be 100644
--- a/arch/powerpc/kernel/vio.c
+++ b/arch/powerpc/kernel/vio.c
@@ -33,7 +33,6 @@
33#include <asm/prom.h> 33#include <asm/prom.h>
34#include <asm/firmware.h> 34#include <asm/firmware.h>
35#include <asm/tce.h> 35#include <asm/tce.h>
36#include <asm/abs_addr.h>
37#include <asm/page.h> 36#include <asm/page.h>
38#include <asm/hvcall.h> 37#include <asm/hvcall.h>
39 38
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
index 44b72feaff7d..74a24bbb9637 100644
--- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
@@ -1065,7 +1065,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_206)
1065 mtspr SPRN_DABRX,r6 1065 mtspr SPRN_DABRX,r6
1066 1066
1067 /* Restore SPRG3 */ 1067 /* Restore SPRG3 */
1068 ld r3,HSTATE_SPRG3(r13) 1068 ld r3,PACA_SPRG3(r13)
1069 mtspr SPRN_SPRG3,r3 1069 mtspr SPRN_SPRG3,r3
1070 1070
1071 /* 1071 /*
diff --git a/arch/powerpc/mm/fault.c b/arch/powerpc/mm/fault.c
index 08ffcf52a856..995f924e007f 100644
--- a/arch/powerpc/mm/fault.c
+++ b/arch/powerpc/mm/fault.c
@@ -133,6 +133,7 @@ static int do_sigbus(struct pt_regs *regs, unsigned long address)
133 up_read(&current->mm->mmap_sem); 133 up_read(&current->mm->mmap_sem);
134 134
135 if (user_mode(regs)) { 135 if (user_mode(regs)) {
136 current->thread.trap_nr = BUS_ADRERR;
136 info.si_signo = SIGBUS; 137 info.si_signo = SIGBUS;
137 info.si_errno = 0; 138 info.si_errno = 0;
138 info.si_code = BUS_ADRERR; 139 info.si_code = BUS_ADRERR;
diff --git a/arch/powerpc/mm/hash_native_64.c b/arch/powerpc/mm/hash_native_64.c
index 90039bc64119..f21e8ce8db33 100644
--- a/arch/powerpc/mm/hash_native_64.c
+++ b/arch/powerpc/mm/hash_native_64.c
@@ -14,10 +14,10 @@
14 14
15#include <linux/spinlock.h> 15#include <linux/spinlock.h>
16#include <linux/bitops.h> 16#include <linux/bitops.h>
17#include <linux/of.h>
17#include <linux/threads.h> 18#include <linux/threads.h>
18#include <linux/smp.h> 19#include <linux/smp.h>
19 20
20#include <asm/abs_addr.h>
21#include <asm/machdep.h> 21#include <asm/machdep.h>
22#include <asm/mmu.h> 22#include <asm/mmu.h>
23#include <asm/mmu_context.h> 23#include <asm/mmu_context.h>
diff --git a/arch/powerpc/mm/hash_utils_64.c b/arch/powerpc/mm/hash_utils_64.c
index 377e5cbedbbb..ba45739bdfe8 100644
--- a/arch/powerpc/mm/hash_utils_64.c
+++ b/arch/powerpc/mm/hash_utils_64.c
@@ -43,7 +43,6 @@
43#include <asm/uaccess.h> 43#include <asm/uaccess.h>
44#include <asm/machdep.h> 44#include <asm/machdep.h>
45#include <asm/prom.h> 45#include <asm/prom.h>
46#include <asm/abs_addr.h>
47#include <asm/tlbflush.h> 46#include <asm/tlbflush.h>
48#include <asm/io.h> 47#include <asm/io.h>
49#include <asm/eeh.h> 48#include <asm/eeh.h>
@@ -651,7 +650,7 @@ static void __init htab_initialize(void)
651 DBG("Hash table allocated at %lx, size: %lx\n", table, 650 DBG("Hash table allocated at %lx, size: %lx\n", table,
652 htab_size_bytes); 651 htab_size_bytes);
653 652
654 htab_address = abs_to_virt(table); 653 htab_address = __va(table);
655 654
656 /* htab absolute addr + encoded htabsize */ 655 /* htab absolute addr + encoded htabsize */
657 _SDR1 = table + __ilog2(pteg_count) - 11; 656 _SDR1 = table + __ilog2(pteg_count) - 11;
diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c
index 620b7acd2fdf..95a45293e5ac 100644
--- a/arch/powerpc/mm/init_64.c
+++ b/arch/powerpc/mm/init_64.c
@@ -62,7 +62,6 @@
62#include <asm/cputable.h> 62#include <asm/cputable.h>
63#include <asm/sections.h> 63#include <asm/sections.h>
64#include <asm/iommu.h> 64#include <asm/iommu.h>
65#include <asm/abs_addr.h>
66#include <asm/vdso.h> 65#include <asm/vdso.h>
67 66
68#include "mmu_decl.h" 67#include "mmu_decl.h"
diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
index fbdad0e3929a..44cf2b20503d 100644
--- a/arch/powerpc/mm/mem.c
+++ b/arch/powerpc/mm/mem.c
@@ -62,7 +62,7 @@
62 62
63int init_bootmem_done; 63int init_bootmem_done;
64int mem_init_done; 64int mem_init_done;
65phys_addr_t memory_limit; 65unsigned long long memory_limit;
66 66
67#ifdef CONFIG_HIGHMEM 67#ifdef CONFIG_HIGHMEM
68pte_t *kmap_pte; 68pte_t *kmap_pte;
diff --git a/arch/powerpc/mm/pgtable_64.c b/arch/powerpc/mm/pgtable_64.c
index 249a0631c4db..297d49547ea8 100644
--- a/arch/powerpc/mm/pgtable_64.c
+++ b/arch/powerpc/mm/pgtable_64.c
@@ -51,7 +51,6 @@
51#include <asm/processor.h> 51#include <asm/processor.h>
52#include <asm/cputable.h> 52#include <asm/cputable.h>
53#include <asm/sections.h> 53#include <asm/sections.h>
54#include <asm/abs_addr.h>
55#include <asm/firmware.h> 54#include <asm/firmware.h>
56 55
57#include "mmu_decl.h" 56#include "mmu_decl.h"
diff --git a/arch/powerpc/mm/stab.c b/arch/powerpc/mm/stab.c
index 9106ebb118f5..3f8efa6f2997 100644
--- a/arch/powerpc/mm/stab.c
+++ b/arch/powerpc/mm/stab.c
@@ -20,7 +20,6 @@
20#include <asm/paca.h> 20#include <asm/paca.h>
21#include <asm/cputable.h> 21#include <asm/cputable.h>
22#include <asm/prom.h> 22#include <asm/prom.h>
23#include <asm/abs_addr.h>
24 23
25struct stab_entry { 24struct stab_entry {
26 unsigned long esid_data; 25 unsigned long esid_data;
@@ -257,7 +256,7 @@ void __init stabs_alloc(void)
257 memset((void *)newstab, 0, HW_PAGE_SIZE); 256 memset((void *)newstab, 0, HW_PAGE_SIZE);
258 257
259 paca[cpu].stab_addr = newstab; 258 paca[cpu].stab_addr = newstab;
260 paca[cpu].stab_real = virt_to_abs(newstab); 259 paca[cpu].stab_real = __pa(newstab);
261 printk(KERN_INFO "Segment table for CPU %d at 0x%llx " 260 printk(KERN_INFO "Segment table for CPU %d at 0x%llx "
262 "virtual, 0x%llx absolute\n", 261 "virtual, 0x%llx absolute\n",
263 cpu, paca[cpu].stab_addr, paca[cpu].stab_real); 262 cpu, paca[cpu].stab_addr, paca[cpu].stab_real);
diff --git a/arch/powerpc/mm/subpage-prot.c b/arch/powerpc/mm/subpage-prot.c
index e4f8f1fc81a5..7c415ddde948 100644
--- a/arch/powerpc/mm/subpage-prot.c
+++ b/arch/powerpc/mm/subpage-prot.c
@@ -95,7 +95,8 @@ static void subpage_prot_clear(unsigned long addr, unsigned long len)
95 struct mm_struct *mm = current->mm; 95 struct mm_struct *mm = current->mm;
96 struct subpage_prot_table *spt = &mm->context.spt; 96 struct subpage_prot_table *spt = &mm->context.spt;
97 u32 **spm, *spp; 97 u32 **spm, *spp;
98 int i, nw; 98 unsigned long i;
99 size_t nw;
99 unsigned long next, limit; 100 unsigned long next, limit;
100 101
101 down_write(&mm->mmap_sem); 102 down_write(&mm->mmap_sem);
@@ -144,7 +145,8 @@ long sys_subpage_prot(unsigned long addr, unsigned long len, u32 __user *map)
144 struct mm_struct *mm = current->mm; 145 struct mm_struct *mm = current->mm;
145 struct subpage_prot_table *spt = &mm->context.spt; 146 struct subpage_prot_table *spt = &mm->context.spt;
146 u32 **spm, *spp; 147 u32 **spm, *spp;
147 int i, nw; 148 unsigned long i;
149 size_t nw;
148 unsigned long next, limit; 150 unsigned long next, limit;
149 int err; 151 int err;
150 152
diff --git a/arch/powerpc/mm/tlb_low_64e.S b/arch/powerpc/mm/tlb_low_64e.S
index f09d48e3268d..b4113bf86353 100644
--- a/arch/powerpc/mm/tlb_low_64e.S
+++ b/arch/powerpc/mm/tlb_low_64e.S
@@ -20,6 +20,8 @@
20#include <asm/pgtable.h> 20#include <asm/pgtable.h>
21#include <asm/exception-64e.h> 21#include <asm/exception-64e.h>
22#include <asm/ppc-opcode.h> 22#include <asm/ppc-opcode.h>
23#include <asm/kvm_asm.h>
24#include <asm/kvm_booke_hv_asm.h>
23 25
24#ifdef CONFIG_PPC_64K_PAGES 26#ifdef CONFIG_PPC_64K_PAGES
25#define VPTE_PMD_SHIFT (PTE_INDEX_SIZE+1) 27#define VPTE_PMD_SHIFT (PTE_INDEX_SIZE+1)
@@ -37,12 +39,18 @@
37 * * 39 * *
38 **********************************************************************/ 40 **********************************************************************/
39 41
40.macro tlb_prolog_bolted addr 42.macro tlb_prolog_bolted intnum addr
41 mtspr SPRN_SPRG_TLB_SCRATCH,r13 43 mtspr SPRN_SPRG_GEN_SCRATCH,r13
42 mfspr r13,SPRN_SPRG_PACA 44 mfspr r13,SPRN_SPRG_PACA
43 std r10,PACA_EXTLB+EX_TLB_R10(r13) 45 std r10,PACA_EXTLB+EX_TLB_R10(r13)
44 mfcr r10 46 mfcr r10
45 std r11,PACA_EXTLB+EX_TLB_R11(r13) 47 std r11,PACA_EXTLB+EX_TLB_R11(r13)
48#ifdef CONFIG_KVM_BOOKE_HV
49BEGIN_FTR_SECTION
50 mfspr r11, SPRN_SRR1
51END_FTR_SECTION_IFSET(CPU_FTR_EMB_HV)
52#endif
53 DO_KVM \intnum, SPRN_SRR1
46 std r16,PACA_EXTLB+EX_TLB_R16(r13) 54 std r16,PACA_EXTLB+EX_TLB_R16(r13)
47 mfspr r16,\addr /* get faulting address */ 55 mfspr r16,\addr /* get faulting address */
48 std r14,PACA_EXTLB+EX_TLB_R14(r13) 56 std r14,PACA_EXTLB+EX_TLB_R14(r13)
@@ -61,12 +69,12 @@
61 ld r15,PACA_EXTLB+EX_TLB_R15(r13) 69 ld r15,PACA_EXTLB+EX_TLB_R15(r13)
62 TLB_MISS_RESTORE_STATS_BOLTED 70 TLB_MISS_RESTORE_STATS_BOLTED
63 ld r16,PACA_EXTLB+EX_TLB_R16(r13) 71 ld r16,PACA_EXTLB+EX_TLB_R16(r13)
64 mfspr r13,SPRN_SPRG_TLB_SCRATCH 72 mfspr r13,SPRN_SPRG_GEN_SCRATCH
65.endm 73.endm
66 74
67/* Data TLB miss */ 75/* Data TLB miss */
68 START_EXCEPTION(data_tlb_miss_bolted) 76 START_EXCEPTION(data_tlb_miss_bolted)
69 tlb_prolog_bolted SPRN_DEAR 77 tlb_prolog_bolted BOOKE_INTERRUPT_DTLB_MISS SPRN_DEAR
70 78
71 /* We need _PAGE_PRESENT and _PAGE_ACCESSED set */ 79 /* We need _PAGE_PRESENT and _PAGE_ACCESSED set */
72 80
@@ -214,7 +222,7 @@ itlb_miss_fault_bolted:
214 222
215/* Instruction TLB miss */ 223/* Instruction TLB miss */
216 START_EXCEPTION(instruction_tlb_miss_bolted) 224 START_EXCEPTION(instruction_tlb_miss_bolted)
217 tlb_prolog_bolted SPRN_SRR0 225 tlb_prolog_bolted BOOKE_INTERRUPT_ITLB_MISS SPRN_SRR0
218 226
219 rldicl. r10,r16,64-PGTABLE_EADDR_SIZE,PGTABLE_EADDR_SIZE+4 227 rldicl. r10,r16,64-PGTABLE_EADDR_SIZE,PGTABLE_EADDR_SIZE+4
220 srdi r15,r16,60 /* get region */ 228 srdi r15,r16,60 /* get region */
diff --git a/arch/powerpc/oprofile/op_model_power4.c b/arch/powerpc/oprofile/op_model_power4.c
index 95ae77dec3f6..315f9495e9b2 100644
--- a/arch/powerpc/oprofile/op_model_power4.c
+++ b/arch/powerpc/oprofile/op_model_power4.c
@@ -21,6 +21,13 @@
21#include <asm/reg.h> 21#include <asm/reg.h>
22 22
23#define dbg(args...) 23#define dbg(args...)
24#define OPROFILE_PM_PMCSEL_MSK 0xffULL
25#define OPROFILE_PM_UNIT_SHIFT 60
26#define OPROFILE_PM_UNIT_MSK 0xfULL
27#define OPROFILE_MAX_PMC_NUM 3
28#define OPROFILE_PMSEL_FIELD_WIDTH 8
29#define OPROFILE_UNIT_FIELD_WIDTH 4
30#define MMCRA_SIAR_VALID_MASK 0x10000000ULL
24 31
25static unsigned long reset_value[OP_MAX_COUNTER]; 32static unsigned long reset_value[OP_MAX_COUNTER];
26 33
@@ -31,6 +38,61 @@ static int use_slot_nums;
31static u32 mmcr0_val; 38static u32 mmcr0_val;
32static u64 mmcr1_val; 39static u64 mmcr1_val;
33static u64 mmcra_val; 40static u64 mmcra_val;
41static u32 cntr_marked_events;
42
43static int power7_marked_instr_event(u64 mmcr1)
44{
45 u64 psel, unit;
46 int pmc, cntr_marked_events = 0;
47
48 /* Given the MMCR1 value, look at the field for each counter to
49 * determine if it is a marked event. Code based on the function
50 * power7_marked_instr_event() in file arch/powerpc/perf/power7-pmu.c.
51 */
52 for (pmc = 0; pmc < 4; pmc++) {
53 psel = mmcr1 & (OPROFILE_PM_PMCSEL_MSK
54 << (OPROFILE_MAX_PMC_NUM - pmc)
55 * OPROFILE_MAX_PMC_NUM);
56 psel = (psel >> ((OPROFILE_MAX_PMC_NUM - pmc)
57 * OPROFILE_PMSEL_FIELD_WIDTH)) & ~1ULL;
58 unit = mmcr1 & (OPROFILE_PM_UNIT_MSK
59 << (OPROFILE_PM_UNIT_SHIFT
60 - (pmc * OPROFILE_PMSEL_FIELD_WIDTH )));
61 unit = unit >> (OPROFILE_PM_UNIT_SHIFT
62 - (pmc * OPROFILE_PMSEL_FIELD_WIDTH));
63
64 switch (psel >> 4) {
65 case 2:
66 cntr_marked_events |= (pmc == 1 || pmc == 3) << pmc;
67 break;
68 case 3:
69 if (psel == 0x3c) {
70 cntr_marked_events |= (pmc == 0) << pmc;
71 break;
72 }
73
74 if (psel == 0x3e) {
75 cntr_marked_events |= (pmc != 1) << pmc;
76 break;
77 }
78
79 cntr_marked_events |= 1 << pmc;
80 break;
81 case 4:
82 case 5:
83 cntr_marked_events |= (unit == 0xd) << pmc;
84 break;
85 case 6:
86 if (psel == 0x64)
87 cntr_marked_events |= (pmc >= 2) << pmc;
88 break;
89 case 8:
90 cntr_marked_events |= (unit == 0xd) << pmc;
91 break;
92 }
93 }
94 return cntr_marked_events;
95}
34 96
35static int power4_reg_setup(struct op_counter_config *ctr, 97static int power4_reg_setup(struct op_counter_config *ctr,
36 struct op_system_config *sys, 98 struct op_system_config *sys,
@@ -47,6 +109,23 @@ static int power4_reg_setup(struct op_counter_config *ctr,
47 mmcr1_val = sys->mmcr1; 109 mmcr1_val = sys->mmcr1;
48 mmcra_val = sys->mmcra; 110 mmcra_val = sys->mmcra;
49 111
112 /* Power 7+ and newer architectures:
113 * Determine which counter events in the group (the group of events is
114 * specified by the bit settings in the MMCR1 register) are marked
115 * events for use in the interrupt handler. Do the calculation once
116 * before OProfile starts. Information is used in the interrupt
117 * handler. Starting with Power 7+ we only record the sample for
118 * marked events if the SIAR valid bit is set. For non marked events
119 * the sample is always recorded.
120 */
121 if (pvr_version_is(PVR_POWER7p))
122 cntr_marked_events = power7_marked_instr_event(mmcr1_val);
123 else
124 cntr_marked_events = 0; /* For older processors, set the bit map
125 * to zero so the sample will always be
126 * be recorded.
127 */
128
50 for (i = 0; i < cur_cpu_spec->num_pmcs; ++i) 129 for (i = 0; i < cur_cpu_spec->num_pmcs; ++i)
51 reset_value[i] = 0x80000000UL - ctr[i].count; 130 reset_value[i] = 0x80000000UL - ctr[i].count;
52 131
@@ -61,10 +140,10 @@ static int power4_reg_setup(struct op_counter_config *ctr,
61 else 140 else
62 mmcr0_val |= MMCR0_PROBLEM_DISABLE; 141 mmcr0_val |= MMCR0_PROBLEM_DISABLE;
63 142
64 if (__is_processor(PV_POWER4) || __is_processor(PV_POWER4p) || 143 if (pvr_version_is(PVR_POWER4) || pvr_version_is(PVR_POWER4p) ||
65 __is_processor(PV_970) || __is_processor(PV_970FX) || 144 pvr_version_is(PVR_970) || pvr_version_is(PVR_970FX) ||
66 __is_processor(PV_970MP) || __is_processor(PV_970GX) || 145 pvr_version_is(PVR_970MP) || pvr_version_is(PVR_970GX) ||
67 __is_processor(PV_POWER5) || __is_processor(PV_POWER5p)) 146 pvr_version_is(PVR_POWER5) || pvr_version_is(PVR_POWER5p))
68 use_slot_nums = 1; 147 use_slot_nums = 1;
69 148
70 return 0; 149 return 0;
@@ -84,9 +163,9 @@ extern void ppc_enable_pmcs(void);
84 */ 163 */
85static inline int mmcra_must_set_sample(void) 164static inline int mmcra_must_set_sample(void)
86{ 165{
87 if (__is_processor(PV_POWER4) || __is_processor(PV_POWER4p) || 166 if (pvr_version_is(PVR_POWER4) || pvr_version_is(PVR_POWER4p) ||
88 __is_processor(PV_970) || __is_processor(PV_970FX) || 167 pvr_version_is(PVR_970) || pvr_version_is(PVR_970FX) ||
89 __is_processor(PV_970MP) || __is_processor(PV_970GX)) 168 pvr_version_is(PVR_970MP) || pvr_version_is(PVR_970GX))
90 return 1; 169 return 1;
91 170
92 return 0; 171 return 0;
@@ -276,7 +355,7 @@ static bool pmc_overflow(unsigned long val)
276 * PMCs because a user might set a period of less than 256 and we 355 * PMCs because a user might set a period of less than 256 and we
277 * don't want to mistakenly reset them. 356 * don't want to mistakenly reset them.
278 */ 357 */
279 if (__is_processor(PV_POWER7) && ((0x80000000 - val) <= 256)) 358 if (pvr_version_is(PVR_POWER7) && ((0x80000000 - val) <= 256))
280 return true; 359 return true;
281 360
282 return false; 361 return false;
@@ -291,6 +370,7 @@ static void power4_handle_interrupt(struct pt_regs *regs,
291 int i; 370 int i;
292 unsigned int mmcr0; 371 unsigned int mmcr0;
293 unsigned long mmcra; 372 unsigned long mmcra;
373 bool siar_valid = false;
294 374
295 mmcra = mfspr(SPRN_MMCRA); 375 mmcra = mfspr(SPRN_MMCRA);
296 376
@@ -300,11 +380,29 @@ static void power4_handle_interrupt(struct pt_regs *regs,
300 /* set the PMM bit (see comment below) */ 380 /* set the PMM bit (see comment below) */
301 mtmsrd(mfmsr() | MSR_PMM); 381 mtmsrd(mfmsr() | MSR_PMM);
302 382
383 /* Check that the SIAR valid bit in MMCRA is set to 1. */
384 if ((mmcra & MMCRA_SIAR_VALID_MASK) == MMCRA_SIAR_VALID_MASK)
385 siar_valid = true;
386
303 for (i = 0; i < cur_cpu_spec->num_pmcs; ++i) { 387 for (i = 0; i < cur_cpu_spec->num_pmcs; ++i) {
304 val = classic_ctr_read(i); 388 val = classic_ctr_read(i);
305 if (pmc_overflow(val)) { 389 if (pmc_overflow(val)) {
306 if (oprofile_running && ctr[i].enabled) { 390 if (oprofile_running && ctr[i].enabled) {
307 oprofile_add_ext_sample(pc, regs, i, is_kernel); 391 /* Power 7+ and newer architectures:
392 * If the event is a marked event, then only
393 * save the sample if the SIAR valid bit is
394 * set. If the event is not marked, then
395 * always save the sample.
396 * Note, the Sample enable bit in the MMCRA
397 * register must be set to 1 if the group
398 * contains a marked event.
399 */
400 if ((siar_valid &&
401 (cntr_marked_events & (1 << i)))
402 || !(cntr_marked_events & (1 << i)))
403 oprofile_add_ext_sample(pc, regs, i,
404 is_kernel);
405
308 classic_ctr_write(i, reset_value[i]); 406 classic_ctr_write(i, reset_value[i]);
309 } else { 407 } else {
310 classic_ctr_write(i, 0); 408 classic_ctr_write(i, 0);
diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
index 7cd2dbd6e4c4..fb55da91aa45 100644
--- a/arch/powerpc/perf/core-book3s.c
+++ b/arch/powerpc/perf/core-book3s.c
@@ -1396,7 +1396,7 @@ static bool pmc_overflow(unsigned long val)
1396 * PMCs because a user might set a period of less than 256 and we 1396 * PMCs because a user might set a period of less than 256 and we
1397 * don't want to mistakenly reset them. 1397 * don't want to mistakenly reset them.
1398 */ 1398 */
1399 if (__is_processor(PV_POWER7) && ((0x80000000 - val) <= 256)) 1399 if (pvr_version_is(PVR_POWER7) && ((0x80000000 - val) <= 256))
1400 return true; 1400 return true;
1401 1401
1402 return false; 1402 return false;
diff --git a/arch/powerpc/platforms/cell/beat.c b/arch/powerpc/platforms/cell/beat.c
index 852592b2b712..affcf566d460 100644
--- a/arch/powerpc/platforms/cell/beat.c
+++ b/arch/powerpc/platforms/cell/beat.c
@@ -136,9 +136,9 @@ ssize_t beat_nvram_get_size(void)
136 return BEAT_NVRAM_SIZE; 136 return BEAT_NVRAM_SIZE;
137} 137}
138 138
139int beat_set_xdabr(unsigned long dabr) 139int beat_set_xdabr(unsigned long dabr, unsigned long dabrx)
140{ 140{
141 if (beat_set_dabr(dabr, DABRX_KERNEL | DABRX_USER)) 141 if (beat_set_dabr(dabr, dabrx))
142 return -1; 142 return -1;
143 return 0; 143 return 0;
144} 144}
diff --git a/arch/powerpc/platforms/cell/beat.h b/arch/powerpc/platforms/cell/beat.h
index 32c8efcedc80..bfcb8e351ae5 100644
--- a/arch/powerpc/platforms/cell/beat.h
+++ b/arch/powerpc/platforms/cell/beat.h
@@ -32,7 +32,7 @@ void beat_get_rtc_time(struct rtc_time *);
32ssize_t beat_nvram_get_size(void); 32ssize_t beat_nvram_get_size(void);
33ssize_t beat_nvram_read(char *, size_t, loff_t *); 33ssize_t beat_nvram_read(char *, size_t, loff_t *);
34ssize_t beat_nvram_write(char *, size_t, loff_t *); 34ssize_t beat_nvram_write(char *, size_t, loff_t *);
35int beat_set_xdabr(unsigned long); 35int beat_set_xdabr(unsigned long, unsigned long);
36void beat_power_save(void); 36void beat_power_save(void);
37void beat_kexec_cpu_down(int, int); 37void beat_kexec_cpu_down(int, int);
38 38
diff --git a/arch/powerpc/platforms/pasemi/iommu.c b/arch/powerpc/platforms/pasemi/iommu.c
index 14943ef01918..7d2d036754b5 100644
--- a/arch/powerpc/platforms/pasemi/iommu.c
+++ b/arch/powerpc/platforms/pasemi/iommu.c
@@ -19,12 +19,12 @@
19 19
20#undef DEBUG 20#undef DEBUG
21 21
22#include <linux/memblock.h>
22#include <linux/types.h> 23#include <linux/types.h>
23#include <linux/spinlock.h> 24#include <linux/spinlock.h>
24#include <linux/pci.h> 25#include <linux/pci.h>
25#include <asm/iommu.h> 26#include <asm/iommu.h>
26#include <asm/machdep.h> 27#include <asm/machdep.h>
27#include <asm/abs_addr.h>
28#include <asm/firmware.h> 28#include <asm/firmware.h>
29 29
30#define IOBMAP_PAGE_SHIFT 12 30#define IOBMAP_PAGE_SHIFT 12
@@ -99,7 +99,7 @@ static int iobmap_build(struct iommu_table *tbl, long index,
99 ip = ((u32 *)tbl->it_base) + index; 99 ip = ((u32 *)tbl->it_base) + index;
100 100
101 while (npages--) { 101 while (npages--) {
102 rpn = virt_to_abs(uaddr) >> IOBMAP_PAGE_SHIFT; 102 rpn = __pa(uaddr) >> IOBMAP_PAGE_SHIFT;
103 103
104 *(ip++) = IOBMAP_L2E_V | rpn; 104 *(ip++) = IOBMAP_L2E_V | rpn;
105 /* invalidate tlb, can be optimized more */ 105 /* invalidate tlb, can be optimized more */
@@ -258,7 +258,7 @@ void __init alloc_iobmap_l2(void)
258 return; 258 return;
259#endif 259#endif
260 /* For 2G space, 8x64 pages (2^21 bytes) is max total l2 size */ 260 /* For 2G space, 8x64 pages (2^21 bytes) is max total l2 size */
261 iob_l2_base = (u32 *)abs_to_virt(memblock_alloc_base(1UL<<21, 1UL<<21, 0x80000000)); 261 iob_l2_base = (u32 *)__va(memblock_alloc_base(1UL<<21, 1UL<<21, 0x80000000));
262 262
263 printk(KERN_INFO "IOBMAP L2 allocated at: %p\n", iob_l2_base); 263 printk(KERN_INFO "IOBMAP L2 allocated at: %p\n", iob_l2_base);
264} 264}
diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index 0e7eccc0f88d..cae7281e4e66 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c
@@ -30,7 +30,6 @@
30#include <asm/opal.h> 30#include <asm/opal.h>
31#include <asm/iommu.h> 31#include <asm/iommu.h>
32#include <asm/tce.h> 32#include <asm/tce.h>
33#include <asm/abs_addr.h>
34 33
35#include "powernv.h" 34#include "powernv.h"
36#include "pci.h" 35#include "pci.h"
diff --git a/arch/powerpc/platforms/powernv/pci-p5ioc2.c b/arch/powerpc/platforms/powernv/pci-p5ioc2.c
index 264967770c3a..6b4bef4e9d82 100644
--- a/arch/powerpc/platforms/powernv/pci-p5ioc2.c
+++ b/arch/powerpc/platforms/powernv/pci-p5ioc2.c
@@ -30,7 +30,6 @@
30#include <asm/opal.h> 30#include <asm/opal.h>
31#include <asm/iommu.h> 31#include <asm/iommu.h>
32#include <asm/tce.h> 32#include <asm/tce.h>
33#include <asm/abs_addr.h>
34 33
35#include "powernv.h" 34#include "powernv.h"
36#include "pci.h" 35#include "pci.h"
diff --git a/arch/powerpc/platforms/powernv/pci.c b/arch/powerpc/platforms/powernv/pci.c
index be3cfc5ceabb..c01688a1a741 100644
--- a/arch/powerpc/platforms/powernv/pci.c
+++ b/arch/powerpc/platforms/powernv/pci.c
@@ -30,7 +30,6 @@
30#include <asm/opal.h> 30#include <asm/opal.h>
31#include <asm/iommu.h> 31#include <asm/iommu.h>
32#include <asm/tce.h> 32#include <asm/tce.h>
33#include <asm/abs_addr.h>
34#include <asm/firmware.h> 33#include <asm/firmware.h>
35 34
36#include "powernv.h" 35#include "powernv.h"
@@ -447,6 +446,11 @@ static void pnv_tce_free(struct iommu_table *tbl, long index, long npages)
447 pnv_tce_invalidate(tbl, tces, tcep - 1); 446 pnv_tce_invalidate(tbl, tces, tcep - 1);
448} 447}
449 448
449static unsigned long pnv_tce_get(struct iommu_table *tbl, long index)
450{
451 return ((u64 *)tbl->it_base)[index - tbl->it_offset];
452}
453
450void pnv_pci_setup_iommu_table(struct iommu_table *tbl, 454void pnv_pci_setup_iommu_table(struct iommu_table *tbl,
451 void *tce_mem, u64 tce_size, 455 void *tce_mem, u64 tce_size,
452 u64 dma_offset) 456 u64 dma_offset)
@@ -597,6 +601,7 @@ void __init pnv_pci_init(void)
597 ppc_md.pci_dma_dev_setup = pnv_pci_dma_dev_setup; 601 ppc_md.pci_dma_dev_setup = pnv_pci_dma_dev_setup;
598 ppc_md.tce_build = pnv_tce_build; 602 ppc_md.tce_build = pnv_tce_build;
599 ppc_md.tce_free = pnv_tce_free; 603 ppc_md.tce_free = pnv_tce_free;
604 ppc_md.tce_get = pnv_tce_get;
600 ppc_md.pci_probe_mode = pnv_pci_probe_mode; 605 ppc_md.pci_probe_mode = pnv_pci_probe_mode;
601 set_pci_dma_ops(&dma_iommu_ops); 606 set_pci_dma_ops(&dma_iommu_ops);
602 607
diff --git a/arch/powerpc/platforms/ps3/setup.c b/arch/powerpc/platforms/ps3/setup.c
index 2d664c5a83b0..3f509f86432c 100644
--- a/arch/powerpc/platforms/ps3/setup.c
+++ b/arch/powerpc/platforms/ps3/setup.c
@@ -184,11 +184,15 @@ early_param("ps3flash", early_parse_ps3flash);
184#define prealloc_ps3flash_bounce_buffer() do { } while (0) 184#define prealloc_ps3flash_bounce_buffer() do { } while (0)
185#endif 185#endif
186 186
187static int ps3_set_dabr(unsigned long dabr) 187static int ps3_set_dabr(unsigned long dabr, unsigned long dabrx)
188{ 188{
189 enum {DABR_USER = 1, DABR_KERNEL = 2,}; 189 /* Have to set at least one bit in the DABRX */
190 if (dabrx == 0 && dabr == 0)
191 dabrx = DABRX_USER;
192 /* hypervisor only allows us to set BTI, Kernel and user */
193 dabrx &= DABRX_BTI | DABRX_KERNEL | DABRX_USER;
190 194
191 return lv1_set_dabr(dabr, DABR_KERNEL | DABR_USER) ? -1 : 0; 195 return lv1_set_dabr(dabr, dabrx) ? -1 : 0;
192} 196}
193 197
194static void __init ps3_setup_arch(void) 198static void __init ps3_setup_arch(void)
diff --git a/arch/powerpc/platforms/pseries/Makefile b/arch/powerpc/platforms/pseries/Makefile
index c222189f5bb2..890622b87c8f 100644
--- a/arch/powerpc/platforms/pseries/Makefile
+++ b/arch/powerpc/platforms/pseries/Makefile
@@ -6,8 +6,9 @@ obj-y := lpar.o hvCall.o nvram.o reconfig.o \
6 firmware.o power.o dlpar.o mobility.o 6 firmware.o power.o dlpar.o mobility.o
7obj-$(CONFIG_SMP) += smp.o 7obj-$(CONFIG_SMP) += smp.o
8obj-$(CONFIG_SCANLOG) += scanlog.o 8obj-$(CONFIG_SCANLOG) += scanlog.o
9obj-$(CONFIG_EEH) += eeh.o eeh_dev.o eeh_cache.o eeh_driver.o \ 9obj-$(CONFIG_EEH) += eeh.o eeh_pe.o eeh_dev.o eeh_cache.o \
10 eeh_event.o eeh_sysfs.o eeh_pseries.o 10 eeh_driver.o eeh_event.o eeh_sysfs.o \
11 eeh_pseries.o
11obj-$(CONFIG_KEXEC) += kexec.o 12obj-$(CONFIG_KEXEC) += kexec.o
12obj-$(CONFIG_PCI) += pci.o pci_dlpar.o 13obj-$(CONFIG_PCI) += pci.o pci_dlpar.o
13obj-$(CONFIG_PSERIES_MSI) += msi.o 14obj-$(CONFIG_PSERIES_MSI) += msi.o
diff --git a/arch/powerpc/platforms/pseries/eeh.c b/arch/powerpc/platforms/pseries/eeh.c
index ecd394cf34e6..18c168b752da 100644
--- a/arch/powerpc/platforms/pseries/eeh.c
+++ b/arch/powerpc/platforms/pseries/eeh.c
@@ -92,6 +92,20 @@ struct eeh_ops *eeh_ops = NULL;
92int eeh_subsystem_enabled; 92int eeh_subsystem_enabled;
93EXPORT_SYMBOL(eeh_subsystem_enabled); 93EXPORT_SYMBOL(eeh_subsystem_enabled);
94 94
95/*
96 * EEH probe mode support. The intention is to support multiple
97 * platforms for EEH. Some platforms like pSeries do PCI emunation
98 * based on device tree. However, other platforms like powernv probe
99 * PCI devices from hardware. The flag is used to distinguish that.
100 * In addition, struct eeh_ops::probe would be invoked for particular
101 * OF node or PCI device so that the corresponding PE would be created
102 * there.
103 */
104int eeh_probe_mode;
105
106/* Global EEH mutex */
107DEFINE_MUTEX(eeh_mutex);
108
95/* Lock to avoid races due to multiple reports of an error */ 109/* Lock to avoid races due to multiple reports of an error */
96static DEFINE_RAW_SPINLOCK(confirm_error_lock); 110static DEFINE_RAW_SPINLOCK(confirm_error_lock);
97 111
@@ -204,22 +218,12 @@ static size_t eeh_gather_pci_data(struct eeh_dev *edev, char * buf, size_t len)
204 } 218 }
205 } 219 }
206 220
207 /* Gather status on devices under the bridge */
208 if (dev->class >> 16 == PCI_BASE_CLASS_BRIDGE) {
209 struct device_node *child;
210
211 for_each_child_of_node(dn, child) {
212 if (of_node_to_eeh_dev(child))
213 n += eeh_gather_pci_data(of_node_to_eeh_dev(child), buf+n, len-n);
214 }
215 }
216
217 return n; 221 return n;
218} 222}
219 223
220/** 224/**
221 * eeh_slot_error_detail - Generate combined log including driver log and error log 225 * eeh_slot_error_detail - Generate combined log including driver log and error log
222 * @edev: device to report error log for 226 * @pe: EEH PE
223 * @severity: temporary or permanent error log 227 * @severity: temporary or permanent error log
224 * 228 *
225 * This routine should be called to generate the combined log, which 229 * This routine should be called to generate the combined log, which
@@ -227,17 +231,22 @@ static size_t eeh_gather_pci_data(struct eeh_dev *edev, char * buf, size_t len)
227 * out from the config space of the corresponding PCI device, while 231 * out from the config space of the corresponding PCI device, while
228 * the error log is fetched through platform dependent function call. 232 * the error log is fetched through platform dependent function call.
229 */ 233 */
230void eeh_slot_error_detail(struct eeh_dev *edev, int severity) 234void eeh_slot_error_detail(struct eeh_pe *pe, int severity)
231{ 235{
232 size_t loglen = 0; 236 size_t loglen = 0;
233 pci_regs_buf[0] = 0; 237 struct eeh_dev *edev;
234 238
235 eeh_pci_enable(edev, EEH_OPT_THAW_MMIO); 239 eeh_pci_enable(pe, EEH_OPT_THAW_MMIO);
236 eeh_ops->configure_bridge(eeh_dev_to_of_node(edev)); 240 eeh_ops->configure_bridge(pe);
237 eeh_restore_bars(edev); 241 eeh_pe_restore_bars(pe);
238 loglen = eeh_gather_pci_data(edev, pci_regs_buf, EEH_PCI_REGS_LOG_LEN);
239 242
240 eeh_ops->get_log(eeh_dev_to_of_node(edev), severity, pci_regs_buf, loglen); 243 pci_regs_buf[0] = 0;
244 eeh_pe_for_each_dev(pe, edev) {
245 loglen += eeh_gather_pci_data(edev, pci_regs_buf,
246 EEH_PCI_REGS_LOG_LEN);
247 }
248
249 eeh_ops->get_log(pe, severity, pci_regs_buf, loglen);
241} 250}
242 251
243/** 252/**
@@ -261,126 +270,8 @@ static inline unsigned long eeh_token_to_phys(unsigned long token)
261} 270}
262 271
263/** 272/**
264 * eeh_find_device_pe - Retrieve the PE for the given device 273 * eeh_dev_check_failure - Check if all 1's data is due to EEH slot freeze
265 * @dn: device node 274 * @edev: eeh device
266 *
267 * Return the PE under which this device lies
268 */
269struct device_node *eeh_find_device_pe(struct device_node *dn)
270{
271 while (dn->parent && of_node_to_eeh_dev(dn->parent) &&
272 (of_node_to_eeh_dev(dn->parent)->mode & EEH_MODE_SUPPORTED)) {
273 dn = dn->parent;
274 }
275 return dn;
276}
277
278/**
279 * __eeh_mark_slot - Mark all child devices as failed
280 * @parent: parent device
281 * @mode_flag: failure flag
282 *
283 * Mark all devices that are children of this device as failed.
284 * Mark the device driver too, so that it can see the failure
285 * immediately; this is critical, since some drivers poll
286 * status registers in interrupts ... If a driver is polling,
287 * and the slot is frozen, then the driver can deadlock in
288 * an interrupt context, which is bad.
289 */
290static void __eeh_mark_slot(struct device_node *parent, int mode_flag)
291{
292 struct device_node *dn;
293
294 for_each_child_of_node(parent, dn) {
295 if (of_node_to_eeh_dev(dn)) {
296 /* Mark the pci device driver too */
297 struct pci_dev *dev = of_node_to_eeh_dev(dn)->pdev;
298
299 of_node_to_eeh_dev(dn)->mode |= mode_flag;
300
301 if (dev && dev->driver)
302 dev->error_state = pci_channel_io_frozen;
303
304 __eeh_mark_slot(dn, mode_flag);
305 }
306 }
307}
308
309/**
310 * eeh_mark_slot - Mark the indicated device and its children as failed
311 * @dn: parent device
312 * @mode_flag: failure flag
313 *
314 * Mark the indicated device and its child devices as failed.
315 * The device drivers are marked as failed as well.
316 */
317void eeh_mark_slot(struct device_node *dn, int mode_flag)
318{
319 struct pci_dev *dev;
320 dn = eeh_find_device_pe(dn);
321
322 /* Back up one, since config addrs might be shared */
323 if (!pcibios_find_pci_bus(dn) && of_node_to_eeh_dev(dn->parent))
324 dn = dn->parent;
325
326 of_node_to_eeh_dev(dn)->mode |= mode_flag;
327
328 /* Mark the pci device too */
329 dev = of_node_to_eeh_dev(dn)->pdev;
330 if (dev)
331 dev->error_state = pci_channel_io_frozen;
332
333 __eeh_mark_slot(dn, mode_flag);
334}
335
336/**
337 * __eeh_clear_slot - Clear failure flag for the child devices
338 * @parent: parent device
339 * @mode_flag: flag to be cleared
340 *
341 * Clear failure flag for the child devices.
342 */
343static void __eeh_clear_slot(struct device_node *parent, int mode_flag)
344{
345 struct device_node *dn;
346
347 for_each_child_of_node(parent, dn) {
348 if (of_node_to_eeh_dev(dn)) {
349 of_node_to_eeh_dev(dn)->mode &= ~mode_flag;
350 of_node_to_eeh_dev(dn)->check_count = 0;
351 __eeh_clear_slot(dn, mode_flag);
352 }
353 }
354}
355
356/**
357 * eeh_clear_slot - Clear failure flag for the indicated device and its children
358 * @dn: parent device
359 * @mode_flag: flag to be cleared
360 *
361 * Clear failure flag for the indicated device and its children.
362 */
363void eeh_clear_slot(struct device_node *dn, int mode_flag)
364{
365 unsigned long flags;
366 raw_spin_lock_irqsave(&confirm_error_lock, flags);
367
368 dn = eeh_find_device_pe(dn);
369
370 /* Back up one, since config addrs might be shared */
371 if (!pcibios_find_pci_bus(dn) && of_node_to_eeh_dev(dn->parent))
372 dn = dn->parent;
373
374 of_node_to_eeh_dev(dn)->mode &= ~mode_flag;
375 of_node_to_eeh_dev(dn)->check_count = 0;
376 __eeh_clear_slot(dn, mode_flag);
377 raw_spin_unlock_irqrestore(&confirm_error_lock, flags);
378}
379
380/**
381 * eeh_dn_check_failure - Check if all 1's data is due to EEH slot freeze
382 * @dn: device node
383 * @dev: pci device, if known
384 * 275 *
385 * Check for an EEH failure for the given device node. Call this 276 * Check for an EEH failure for the given device node. Call this
386 * routine if the result of a read was all 0xff's and you want to 277 * routine if the result of a read was all 0xff's and you want to
@@ -392,11 +283,13 @@ void eeh_clear_slot(struct device_node *dn, int mode_flag)
392 * 283 *
393 * It is safe to call this routine in an interrupt context. 284 * It is safe to call this routine in an interrupt context.
394 */ 285 */
395int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev) 286int eeh_dev_check_failure(struct eeh_dev *edev)
396{ 287{
397 int ret; 288 int ret;
398 unsigned long flags; 289 unsigned long flags;
399 struct eeh_dev *edev; 290 struct device_node *dn;
291 struct pci_dev *dev;
292 struct eeh_pe *pe;
400 int rc = 0; 293 int rc = 0;
401 const char *location; 294 const char *location;
402 295
@@ -405,23 +298,23 @@ int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev)
405 if (!eeh_subsystem_enabled) 298 if (!eeh_subsystem_enabled)
406 return 0; 299 return 0;
407 300
408 if (!dn) { 301 if (!edev) {
409 eeh_stats.no_dn++; 302 eeh_stats.no_dn++;
410 return 0; 303 return 0;
411 } 304 }
412 dn = eeh_find_device_pe(dn); 305 dn = eeh_dev_to_of_node(edev);
413 edev = of_node_to_eeh_dev(dn); 306 dev = eeh_dev_to_pci_dev(edev);
307 pe = edev->pe;
414 308
415 /* Access to IO BARs might get this far and still not want checking. */ 309 /* Access to IO BARs might get this far and still not want checking. */
416 if (!(edev->mode & EEH_MODE_SUPPORTED) || 310 if (!pe) {
417 edev->mode & EEH_MODE_NOCHECK) {
418 eeh_stats.ignored_check++; 311 eeh_stats.ignored_check++;
419 pr_debug("EEH: Ignored check (%x) for %s %s\n", 312 pr_debug("EEH: Ignored check for %s %s\n",
420 edev->mode, eeh_pci_name(dev), dn->full_name); 313 eeh_pci_name(dev), dn->full_name);
421 return 0; 314 return 0;
422 } 315 }
423 316
424 if (!edev->config_addr && !edev->pe_config_addr) { 317 if (!pe->addr && !pe->config_addr) {
425 eeh_stats.no_cfg_addr++; 318 eeh_stats.no_cfg_addr++;
426 return 0; 319 return 0;
427 } 320 }
@@ -434,13 +327,13 @@ int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev)
434 */ 327 */
435 raw_spin_lock_irqsave(&confirm_error_lock, flags); 328 raw_spin_lock_irqsave(&confirm_error_lock, flags);
436 rc = 1; 329 rc = 1;
437 if (edev->mode & EEH_MODE_ISOLATED) { 330 if (pe->state & EEH_PE_ISOLATED) {
438 edev->check_count++; 331 pe->check_count++;
439 if (edev->check_count % EEH_MAX_FAILS == 0) { 332 if (pe->check_count % EEH_MAX_FAILS == 0) {
440 location = of_get_property(dn, "ibm,loc-code", NULL); 333 location = of_get_property(dn, "ibm,loc-code", NULL);
441 printk(KERN_ERR "EEH: %d reads ignored for recovering device at " 334 printk(KERN_ERR "EEH: %d reads ignored for recovering device at "
442 "location=%s driver=%s pci addr=%s\n", 335 "location=%s driver=%s pci addr=%s\n",
443 edev->check_count, location, 336 pe->check_count, location,
444 eeh_driver_name(dev), eeh_pci_name(dev)); 337 eeh_driver_name(dev), eeh_pci_name(dev));
445 printk(KERN_ERR "EEH: Might be infinite loop in %s driver\n", 338 printk(KERN_ERR "EEH: Might be infinite loop in %s driver\n",
446 eeh_driver_name(dev)); 339 eeh_driver_name(dev));
@@ -456,7 +349,7 @@ int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev)
456 * function zero of a multi-function device. 349 * function zero of a multi-function device.
457 * In any case they must share a common PHB. 350 * In any case they must share a common PHB.
458 */ 351 */
459 ret = eeh_ops->get_state(dn, NULL); 352 ret = eeh_ops->get_state(pe, NULL);
460 353
461 /* Note that config-io to empty slots may fail; 354 /* Note that config-io to empty slots may fail;
462 * they are empty when they don't have children. 355 * they are empty when they don't have children.
@@ -469,7 +362,7 @@ int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev)
469 (ret & (EEH_STATE_MMIO_ACTIVE | EEH_STATE_DMA_ACTIVE)) == 362 (ret & (EEH_STATE_MMIO_ACTIVE | EEH_STATE_DMA_ACTIVE)) ==
470 (EEH_STATE_MMIO_ACTIVE | EEH_STATE_DMA_ACTIVE)) { 363 (EEH_STATE_MMIO_ACTIVE | EEH_STATE_DMA_ACTIVE)) {
471 eeh_stats.false_positives++; 364 eeh_stats.false_positives++;
472 edev->false_positives ++; 365 pe->false_positives++;
473 rc = 0; 366 rc = 0;
474 goto dn_unlock; 367 goto dn_unlock;
475 } 368 }
@@ -480,10 +373,10 @@ int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev)
480 * with other functions on this device, and functions under 373 * with other functions on this device, and functions under
481 * bridges. 374 * bridges.
482 */ 375 */
483 eeh_mark_slot(dn, EEH_MODE_ISOLATED); 376 eeh_pe_state_mark(pe, EEH_PE_ISOLATED);
484 raw_spin_unlock_irqrestore(&confirm_error_lock, flags); 377 raw_spin_unlock_irqrestore(&confirm_error_lock, flags);
485 378
486 eeh_send_failure_event(edev); 379 eeh_send_failure_event(pe);
487 380
488 /* Most EEH events are due to device driver bugs. Having 381 /* Most EEH events are due to device driver bugs. Having
489 * a stack trace will help the device-driver authors figure 382 * a stack trace will help the device-driver authors figure
@@ -497,7 +390,7 @@ dn_unlock:
497 return rc; 390 return rc;
498} 391}
499 392
500EXPORT_SYMBOL_GPL(eeh_dn_check_failure); 393EXPORT_SYMBOL_GPL(eeh_dev_check_failure);
501 394
502/** 395/**
503 * eeh_check_failure - Check if all 1's data is due to EEH slot freeze 396 * eeh_check_failure - Check if all 1's data is due to EEH slot freeze
@@ -514,21 +407,19 @@ EXPORT_SYMBOL_GPL(eeh_dn_check_failure);
514unsigned long eeh_check_failure(const volatile void __iomem *token, unsigned long val) 407unsigned long eeh_check_failure(const volatile void __iomem *token, unsigned long val)
515{ 408{
516 unsigned long addr; 409 unsigned long addr;
517 struct pci_dev *dev; 410 struct eeh_dev *edev;
518 struct device_node *dn;
519 411
520 /* Finding the phys addr + pci device; this is pretty quick. */ 412 /* Finding the phys addr + pci device; this is pretty quick. */
521 addr = eeh_token_to_phys((unsigned long __force) token); 413 addr = eeh_token_to_phys((unsigned long __force) token);
522 dev = pci_addr_cache_get_device(addr); 414 edev = eeh_addr_cache_get_dev(addr);
523 if (!dev) { 415 if (!edev) {
524 eeh_stats.no_device++; 416 eeh_stats.no_device++;
525 return val; 417 return val;
526 } 418 }
527 419
528 dn = pci_device_to_OF_node(dev); 420 eeh_dev_check_failure(edev);
529 eeh_dn_check_failure(dn, dev);
530 421
531 pci_dev_put(dev); 422 pci_dev_put(eeh_dev_to_pci_dev(edev));
532 return val; 423 return val;
533} 424}
534 425
@@ -537,23 +428,22 @@ EXPORT_SYMBOL(eeh_check_failure);
537 428
538/** 429/**
539 * eeh_pci_enable - Enable MMIO or DMA transfers for this slot 430 * eeh_pci_enable - Enable MMIO or DMA transfers for this slot
540 * @edev: pci device node 431 * @pe: EEH PE
541 * 432 *
542 * This routine should be called to reenable frozen MMIO or DMA 433 * This routine should be called to reenable frozen MMIO or DMA
543 * so that it would work correctly again. It's useful while doing 434 * so that it would work correctly again. It's useful while doing
544 * recovery or log collection on the indicated device. 435 * recovery or log collection on the indicated device.
545 */ 436 */
546int eeh_pci_enable(struct eeh_dev *edev, int function) 437int eeh_pci_enable(struct eeh_pe *pe, int function)
547{ 438{
548 int rc; 439 int rc;
549 struct device_node *dn = eeh_dev_to_of_node(edev);
550 440
551 rc = eeh_ops->set_option(dn, function); 441 rc = eeh_ops->set_option(pe, function);
552 if (rc) 442 if (rc)
553 printk(KERN_WARNING "EEH: Unexpected state change %d, err=%d dn=%s\n", 443 pr_warning("%s: Unexpected state change %d on PHB#%d-PE#%x, err=%d\n",
554 function, rc, dn->full_name); 444 __func__, function, pe->phb->global_number, pe->addr, rc);
555 445
556 rc = eeh_ops->wait_state(dn, PCI_BUS_RESET_WAIT_MSEC); 446 rc = eeh_ops->wait_state(pe, PCI_BUS_RESET_WAIT_MSEC);
557 if (rc > 0 && (rc & EEH_STATE_MMIO_ENABLED) && 447 if (rc > 0 && (rc & EEH_STATE_MMIO_ENABLED) &&
558 (function == EEH_OPT_THAW_MMIO)) 448 (function == EEH_OPT_THAW_MMIO))
559 return 0; 449 return 0;
@@ -571,17 +461,24 @@ int eeh_pci_enable(struct eeh_dev *edev, int function)
571 */ 461 */
572int pcibios_set_pcie_reset_state(struct pci_dev *dev, enum pcie_reset_state state) 462int pcibios_set_pcie_reset_state(struct pci_dev *dev, enum pcie_reset_state state)
573{ 463{
574 struct device_node *dn = pci_device_to_OF_node(dev); 464 struct eeh_dev *edev = pci_dev_to_eeh_dev(dev);
465 struct eeh_pe *pe = edev->pe;
466
467 if (!pe) {
468 pr_err("%s: No PE found on PCI device %s\n",
469 __func__, pci_name(dev));
470 return -EINVAL;
471 }
575 472
576 switch (state) { 473 switch (state) {
577 case pcie_deassert_reset: 474 case pcie_deassert_reset:
578 eeh_ops->reset(dn, EEH_RESET_DEACTIVATE); 475 eeh_ops->reset(pe, EEH_RESET_DEACTIVATE);
579 break; 476 break;
580 case pcie_hot_reset: 477 case pcie_hot_reset:
581 eeh_ops->reset(dn, EEH_RESET_HOT); 478 eeh_ops->reset(pe, EEH_RESET_HOT);
582 break; 479 break;
583 case pcie_warm_reset: 480 case pcie_warm_reset:
584 eeh_ops->reset(dn, EEH_RESET_FUNDAMENTAL); 481 eeh_ops->reset(pe, EEH_RESET_FUNDAMENTAL);
585 break; 482 break;
586 default: 483 default:
587 return -EINVAL; 484 return -EINVAL;
@@ -591,66 +488,37 @@ int pcibios_set_pcie_reset_state(struct pci_dev *dev, enum pcie_reset_state stat
591} 488}
592 489
593/** 490/**
594 * __eeh_set_pe_freset - Check the required reset for child devices 491 * eeh_set_pe_freset - Check the required reset for the indicated device
595 * @parent: parent device 492 * @data: EEH device
596 * @freset: return value 493 * @flag: return value
597 *
598 * Each device might have its preferred reset type: fundamental or
599 * hot reset. The routine is used to collect the information from
600 * the child devices so that they could be reset accordingly.
601 */
602void __eeh_set_pe_freset(struct device_node *parent, unsigned int *freset)
603{
604 struct device_node *dn;
605
606 for_each_child_of_node(parent, dn) {
607 if (of_node_to_eeh_dev(dn)) {
608 struct pci_dev *dev = of_node_to_eeh_dev(dn)->pdev;
609
610 if (dev && dev->driver)
611 *freset |= dev->needs_freset;
612
613 __eeh_set_pe_freset(dn, freset);
614 }
615 }
616}
617
618/**
619 * eeh_set_pe_freset - Check the required reset for the indicated device and its children
620 * @dn: parent device
621 * @freset: return value
622 * 494 *
623 * Each device might have its preferred reset type: fundamental or 495 * Each device might have its preferred reset type: fundamental or
624 * hot reset. The routine is used to collected the information for 496 * hot reset. The routine is used to collected the information for
625 * the indicated device and its children so that the bunch of the 497 * the indicated device and its children so that the bunch of the
626 * devices could be reset properly. 498 * devices could be reset properly.
627 */ 499 */
628void eeh_set_pe_freset(struct device_node *dn, unsigned int *freset) 500static void *eeh_set_dev_freset(void *data, void *flag)
629{ 501{
630 struct pci_dev *dev; 502 struct pci_dev *dev;
631 dn = eeh_find_device_pe(dn); 503 unsigned int *freset = (unsigned int *)flag;
632 504 struct eeh_dev *edev = (struct eeh_dev *)data;
633 /* Back up one, since config addrs might be shared */
634 if (!pcibios_find_pci_bus(dn) && of_node_to_eeh_dev(dn->parent))
635 dn = dn->parent;
636 505
637 dev = of_node_to_eeh_dev(dn)->pdev; 506 dev = eeh_dev_to_pci_dev(edev);
638 if (dev) 507 if (dev)
639 *freset |= dev->needs_freset; 508 *freset |= dev->needs_freset;
640 509
641 __eeh_set_pe_freset(dn, freset); 510 return NULL;
642} 511}
643 512
644/** 513/**
645 * eeh_reset_pe_once - Assert the pci #RST line for 1/4 second 514 * eeh_reset_pe_once - Assert the pci #RST line for 1/4 second
646 * @edev: pci device node to be reset. 515 * @pe: EEH PE
647 * 516 *
648 * Assert the PCI #RST line for 1/4 second. 517 * Assert the PCI #RST line for 1/4 second.
649 */ 518 */
650static void eeh_reset_pe_once(struct eeh_dev *edev) 519static void eeh_reset_pe_once(struct eeh_pe *pe)
651{ 520{
652 unsigned int freset = 0; 521 unsigned int freset = 0;
653 struct device_node *dn = eeh_dev_to_of_node(edev);
654 522
655 /* Determine type of EEH reset required for 523 /* Determine type of EEH reset required for
656 * Partitionable Endpoint, a hot-reset (1) 524 * Partitionable Endpoint, a hot-reset (1)
@@ -658,12 +526,12 @@ static void eeh_reset_pe_once(struct eeh_dev *edev)
658 * A fundamental reset required by any device under 526 * A fundamental reset required by any device under
659 * Partitionable Endpoint trumps hot-reset. 527 * Partitionable Endpoint trumps hot-reset.
660 */ 528 */
661 eeh_set_pe_freset(dn, &freset); 529 eeh_pe_dev_traverse(pe, eeh_set_dev_freset, &freset);
662 530
663 if (freset) 531 if (freset)
664 eeh_ops->reset(dn, EEH_RESET_FUNDAMENTAL); 532 eeh_ops->reset(pe, EEH_RESET_FUNDAMENTAL);
665 else 533 else
666 eeh_ops->reset(dn, EEH_RESET_HOT); 534 eeh_ops->reset(pe, EEH_RESET_HOT);
667 535
668 /* The PCI bus requires that the reset be held high for at least 536 /* The PCI bus requires that the reset be held high for at least
669 * a 100 milliseconds. We wait a bit longer 'just in case'. 537 * a 100 milliseconds. We wait a bit longer 'just in case'.
@@ -675,9 +543,9 @@ static void eeh_reset_pe_once(struct eeh_dev *edev)
675 * pci slot reset line is dropped. Make sure we don't miss 543 * pci slot reset line is dropped. Make sure we don't miss
676 * these, and clear the flag now. 544 * these, and clear the flag now.
677 */ 545 */
678 eeh_clear_slot(dn, EEH_MODE_ISOLATED); 546 eeh_pe_state_clear(pe, EEH_PE_ISOLATED);
679 547
680 eeh_ops->reset(dn, EEH_RESET_DEACTIVATE); 548 eeh_ops->reset(pe, EEH_RESET_DEACTIVATE);
681 549
682 /* After a PCI slot has been reset, the PCI Express spec requires 550 /* After a PCI slot has been reset, the PCI Express spec requires
683 * a 1.5 second idle time for the bus to stabilize, before starting 551 * a 1.5 second idle time for the bus to stabilize, before starting
@@ -689,116 +557,36 @@ static void eeh_reset_pe_once(struct eeh_dev *edev)
689 557
690/** 558/**
691 * eeh_reset_pe - Reset the indicated PE 559 * eeh_reset_pe - Reset the indicated PE
692 * @edev: PCI device associated EEH device 560 * @pe: EEH PE
693 * 561 *
694 * This routine should be called to reset indicated device, including 562 * This routine should be called to reset indicated device, including
695 * PE. A PE might include multiple PCI devices and sometimes PCI bridges 563 * PE. A PE might include multiple PCI devices and sometimes PCI bridges
696 * might be involved as well. 564 * might be involved as well.
697 */ 565 */
698int eeh_reset_pe(struct eeh_dev *edev) 566int eeh_reset_pe(struct eeh_pe *pe)
699{ 567{
700 int i, rc; 568 int i, rc;
701 struct device_node *dn = eeh_dev_to_of_node(edev);
702 569
703 /* Take three shots at resetting the bus */ 570 /* Take three shots at resetting the bus */
704 for (i=0; i<3; i++) { 571 for (i=0; i<3; i++) {
705 eeh_reset_pe_once(edev); 572 eeh_reset_pe_once(pe);
706 573
707 rc = eeh_ops->wait_state(dn, PCI_BUS_RESET_WAIT_MSEC); 574 rc = eeh_ops->wait_state(pe, PCI_BUS_RESET_WAIT_MSEC);
708 if (rc == (EEH_STATE_MMIO_ACTIVE | EEH_STATE_DMA_ACTIVE)) 575 if (rc == (EEH_STATE_MMIO_ACTIVE | EEH_STATE_DMA_ACTIVE))
709 return 0; 576 return 0;
710 577
711 if (rc < 0) { 578 if (rc < 0) {
712 printk(KERN_ERR "EEH: unrecoverable slot failure %s\n", 579 pr_err("%s: Unrecoverable slot failure on PHB#%d-PE#%x",
713 dn->full_name); 580 __func__, pe->phb->global_number, pe->addr);
714 return -1; 581 return -1;
715 } 582 }
716 printk(KERN_ERR "EEH: bus reset %d failed on slot %s, rc=%d\n", 583 pr_err("EEH: bus reset %d failed on PHB#%d-PE#%x, rc=%d\n",
717 i+1, dn->full_name, rc); 584 i+1, pe->phb->global_number, pe->addr, rc);
718 } 585 }
719 586
720 return -1; 587 return -1;
721} 588}
722 589
723/** Save and restore of PCI BARs
724 *
725 * Although firmware will set up BARs during boot, it doesn't
726 * set up device BAR's after a device reset, although it will,
727 * if requested, set up bridge configuration. Thus, we need to
728 * configure the PCI devices ourselves.
729 */
730
731/**
732 * eeh_restore_one_device_bars - Restore the Base Address Registers for one device
733 * @edev: PCI device associated EEH device
734 *
735 * Loads the PCI configuration space base address registers,
736 * the expansion ROM base address, the latency timer, and etc.
737 * from the saved values in the device node.
738 */
739static inline void eeh_restore_one_device_bars(struct eeh_dev *edev)
740{
741 int i;
742 u32 cmd;
743 struct device_node *dn = eeh_dev_to_of_node(edev);
744
745 if (!edev->phb)
746 return;
747
748 for (i=4; i<10; i++) {
749 eeh_ops->write_config(dn, i*4, 4, edev->config_space[i]);
750 }
751
752 /* 12 == Expansion ROM Address */
753 eeh_ops->write_config(dn, 12*4, 4, edev->config_space[12]);
754
755#define BYTE_SWAP(OFF) (8*((OFF)/4)+3-(OFF))
756#define SAVED_BYTE(OFF) (((u8 *)(edev->config_space))[BYTE_SWAP(OFF)])
757
758 eeh_ops->write_config(dn, PCI_CACHE_LINE_SIZE, 1,
759 SAVED_BYTE(PCI_CACHE_LINE_SIZE));
760
761 eeh_ops->write_config(dn, PCI_LATENCY_TIMER, 1,
762 SAVED_BYTE(PCI_LATENCY_TIMER));
763
764 /* max latency, min grant, interrupt pin and line */
765 eeh_ops->write_config(dn, 15*4, 4, edev->config_space[15]);
766
767 /* Restore PERR & SERR bits, some devices require it,
768 * don't touch the other command bits
769 */
770 eeh_ops->read_config(dn, PCI_COMMAND, 4, &cmd);
771 if (edev->config_space[1] & PCI_COMMAND_PARITY)
772 cmd |= PCI_COMMAND_PARITY;
773 else
774 cmd &= ~PCI_COMMAND_PARITY;
775 if (edev->config_space[1] & PCI_COMMAND_SERR)
776 cmd |= PCI_COMMAND_SERR;
777 else
778 cmd &= ~PCI_COMMAND_SERR;
779 eeh_ops->write_config(dn, PCI_COMMAND, 4, cmd);
780}
781
782/**
783 * eeh_restore_bars - Restore the PCI config space info
784 * @edev: EEH device
785 *
786 * This routine performs a recursive walk to the children
787 * of this device as well.
788 */
789void eeh_restore_bars(struct eeh_dev *edev)
790{
791 struct device_node *dn;
792 if (!edev)
793 return;
794
795 if ((edev->mode & EEH_MODE_SUPPORTED) && !IS_BRIDGE(edev->class_code))
796 eeh_restore_one_device_bars(edev);
797
798 for_each_child_of_node(eeh_dev_to_of_node(edev), dn)
799 eeh_restore_bars(of_node_to_eeh_dev(dn));
800}
801
802/** 590/**
803 * eeh_save_bars - Save device bars 591 * eeh_save_bars - Save device bars
804 * @edev: PCI device associated EEH device 592 * @edev: PCI device associated EEH device
@@ -808,7 +596,7 @@ void eeh_restore_bars(struct eeh_dev *edev)
808 * PCI devices are added individually; but, for the restore, 596 * PCI devices are added individually; but, for the restore,
809 * an entire slot is reset at a time. 597 * an entire slot is reset at a time.
810 */ 598 */
811static void eeh_save_bars(struct eeh_dev *edev) 599void eeh_save_bars(struct eeh_dev *edev)
812{ 600{
813 int i; 601 int i;
814 struct device_node *dn; 602 struct device_node *dn;
@@ -822,102 +610,6 @@ static void eeh_save_bars(struct eeh_dev *edev)
822} 610}
823 611
824/** 612/**
825 * eeh_early_enable - Early enable EEH on the indicated device
826 * @dn: device node
827 * @data: BUID
828 *
829 * Enable EEH functionality on the specified PCI device. The function
830 * is expected to be called before real PCI probing is done. However,
831 * the PHBs have been initialized at this point.
832 */
833static void *eeh_early_enable(struct device_node *dn, void *data)
834{
835 int ret;
836 const u32 *class_code = of_get_property(dn, "class-code", NULL);
837 const u32 *vendor_id = of_get_property(dn, "vendor-id", NULL);
838 const u32 *device_id = of_get_property(dn, "device-id", NULL);
839 const u32 *regs;
840 int enable;
841 struct eeh_dev *edev = of_node_to_eeh_dev(dn);
842
843 edev->class_code = 0;
844 edev->mode = 0;
845 edev->check_count = 0;
846 edev->freeze_count = 0;
847 edev->false_positives = 0;
848
849 if (!of_device_is_available(dn))
850 return NULL;
851
852 /* Ignore bad nodes. */
853 if (!class_code || !vendor_id || !device_id)
854 return NULL;
855
856 /* There is nothing to check on PCI to ISA bridges */
857 if (dn->type && !strcmp(dn->type, "isa")) {
858 edev->mode |= EEH_MODE_NOCHECK;
859 return NULL;
860 }
861 edev->class_code = *class_code;
862
863 /* Ok... see if this device supports EEH. Some do, some don't,
864 * and the only way to find out is to check each and every one.
865 */
866 regs = of_get_property(dn, "reg", NULL);
867 if (regs) {
868 /* First register entry is addr (00BBSS00) */
869 /* Try to enable eeh */
870 ret = eeh_ops->set_option(dn, EEH_OPT_ENABLE);
871
872 enable = 0;
873 if (ret == 0) {
874 edev->config_addr = regs[0];
875
876 /* If the newer, better, ibm,get-config-addr-info is supported,
877 * then use that instead.
878 */
879 edev->pe_config_addr = eeh_ops->get_pe_addr(dn);
880
881 /* Some older systems (Power4) allow the
882 * ibm,set-eeh-option call to succeed even on nodes
883 * where EEH is not supported. Verify support
884 * explicitly.
885 */
886 ret = eeh_ops->get_state(dn, NULL);
887 if (ret > 0 && ret != EEH_STATE_NOT_SUPPORT)
888 enable = 1;
889 }
890
891 if (enable) {
892 eeh_subsystem_enabled = 1;
893 edev->mode |= EEH_MODE_SUPPORTED;
894
895 pr_debug("EEH: %s: eeh enabled, config=%x pe_config=%x\n",
896 dn->full_name, edev->config_addr,
897 edev->pe_config_addr);
898 } else {
899
900 /* This device doesn't support EEH, but it may have an
901 * EEH parent, in which case we mark it as supported.
902 */
903 if (dn->parent && of_node_to_eeh_dev(dn->parent) &&
904 (of_node_to_eeh_dev(dn->parent)->mode & EEH_MODE_SUPPORTED)) {
905 /* Parent supports EEH. */
906 edev->mode |= EEH_MODE_SUPPORTED;
907 edev->config_addr = of_node_to_eeh_dev(dn->parent)->config_addr;
908 return NULL;
909 }
910 }
911 } else {
912 printk(KERN_WARNING "EEH: %s: unable to get reg property.\n",
913 dn->full_name);
914 }
915
916 eeh_save_bars(edev);
917 return NULL;
918}
919
920/**
921 * eeh_ops_register - Register platform dependent EEH operations 613 * eeh_ops_register - Register platform dependent EEH operations
922 * @ops: platform dependent EEH operations 614 * @ops: platform dependent EEH operations
923 * 615 *
@@ -982,7 +674,7 @@ int __exit eeh_ops_unregister(const char *name)
982 * Even if force-off is set, the EEH hardware is still enabled, so that 674 * Even if force-off is set, the EEH hardware is still enabled, so that
983 * newer systems can boot. 675 * newer systems can boot.
984 */ 676 */
985void __init eeh_init(void) 677static int __init eeh_init(void)
986{ 678{
987 struct pci_controller *hose, *tmp; 679 struct pci_controller *hose, *tmp;
988 struct device_node *phb; 680 struct device_node *phb;
@@ -992,27 +684,34 @@ void __init eeh_init(void)
992 if (!eeh_ops) { 684 if (!eeh_ops) {
993 pr_warning("%s: Platform EEH operation not found\n", 685 pr_warning("%s: Platform EEH operation not found\n",
994 __func__); 686 __func__);
995 return; 687 return -EEXIST;
996 } else if ((ret = eeh_ops->init())) { 688 } else if ((ret = eeh_ops->init())) {
997 pr_warning("%s: Failed to call platform init function (%d)\n", 689 pr_warning("%s: Failed to call platform init function (%d)\n",
998 __func__, ret); 690 __func__, ret);
999 return; 691 return ret;
1000 } 692 }
1001 693
1002 raw_spin_lock_init(&confirm_error_lock); 694 raw_spin_lock_init(&confirm_error_lock);
1003 695
1004 /* Enable EEH for all adapters */ 696 /* Enable EEH for all adapters */
1005 list_for_each_entry_safe(hose, tmp, &hose_list, list_node) { 697 if (eeh_probe_mode_devtree()) {
1006 phb = hose->dn; 698 list_for_each_entry_safe(hose, tmp,
1007 traverse_pci_devices(phb, eeh_early_enable, NULL); 699 &hose_list, list_node) {
700 phb = hose->dn;
701 traverse_pci_devices(phb, eeh_ops->of_probe, NULL);
702 }
1008 } 703 }
1009 704
1010 if (eeh_subsystem_enabled) 705 if (eeh_subsystem_enabled)
1011 printk(KERN_INFO "EEH: PCI Enhanced I/O Error Handling Enabled\n"); 706 pr_info("EEH: PCI Enhanced I/O Error Handling Enabled\n");
1012 else 707 else
1013 printk(KERN_WARNING "EEH: No capable adapters found\n"); 708 pr_warning("EEH: No capable adapters found\n");
709
710 return ret;
1014} 711}
1015 712
713core_initcall_sync(eeh_init);
714
1016/** 715/**
1017 * eeh_add_device_early - Enable EEH for the indicated device_node 716 * eeh_add_device_early - Enable EEH for the indicated device_node
1018 * @dn: device node for which to set up EEH 717 * @dn: device node for which to set up EEH
@@ -1037,7 +736,8 @@ static void eeh_add_device_early(struct device_node *dn)
1037 if (NULL == phb || 0 == phb->buid) 736 if (NULL == phb || 0 == phb->buid)
1038 return; 737 return;
1039 738
1040 eeh_early_enable(dn, NULL); 739 /* FIXME: hotplug support on POWERNV */
740 eeh_ops->of_probe(dn, NULL);
1041} 741}
1042 742
1043/** 743/**
@@ -1087,7 +787,7 @@ static void eeh_add_device_late(struct pci_dev *dev)
1087 edev->pdev = dev; 787 edev->pdev = dev;
1088 dev->dev.archdata.edev = edev; 788 dev->dev.archdata.edev = edev;
1089 789
1090 pci_addr_cache_insert_device(dev); 790 eeh_addr_cache_insert_dev(dev);
1091 eeh_sysfs_add_device(dev); 791 eeh_sysfs_add_device(dev);
1092} 792}
1093 793
@@ -1143,7 +843,8 @@ static void eeh_remove_device(struct pci_dev *dev)
1143 dev->dev.archdata.edev = NULL; 843 dev->dev.archdata.edev = NULL;
1144 pci_dev_put(dev); 844 pci_dev_put(dev);
1145 845
1146 pci_addr_cache_remove_device(dev); 846 eeh_rmv_from_parent_pe(edev);
847 eeh_addr_cache_rmv_dev(dev);
1147 eeh_sysfs_remove_device(dev); 848 eeh_sysfs_remove_device(dev);
1148} 849}
1149 850
diff --git a/arch/powerpc/platforms/pseries/eeh_cache.c b/arch/powerpc/platforms/pseries/eeh_cache.c
index e5ae1c687c66..5a4c87903057 100644
--- a/arch/powerpc/platforms/pseries/eeh_cache.c
+++ b/arch/powerpc/platforms/pseries/eeh_cache.c
@@ -50,6 +50,7 @@ struct pci_io_addr_range {
50 struct rb_node rb_node; 50 struct rb_node rb_node;
51 unsigned long addr_lo; 51 unsigned long addr_lo;
52 unsigned long addr_hi; 52 unsigned long addr_hi;
53 struct eeh_dev *edev;
53 struct pci_dev *pcidev; 54 struct pci_dev *pcidev;
54 unsigned int flags; 55 unsigned int flags;
55}; 56};
@@ -59,7 +60,7 @@ static struct pci_io_addr_cache {
59 spinlock_t piar_lock; 60 spinlock_t piar_lock;
60} pci_io_addr_cache_root; 61} pci_io_addr_cache_root;
61 62
62static inline struct pci_dev *__pci_addr_cache_get_device(unsigned long addr) 63static inline struct eeh_dev *__eeh_addr_cache_get_device(unsigned long addr)
63{ 64{
64 struct rb_node *n = pci_io_addr_cache_root.rb_root.rb_node; 65 struct rb_node *n = pci_io_addr_cache_root.rb_root.rb_node;
65 66
@@ -74,7 +75,7 @@ static inline struct pci_dev *__pci_addr_cache_get_device(unsigned long addr)
74 n = n->rb_right; 75 n = n->rb_right;
75 } else { 76 } else {
76 pci_dev_get(piar->pcidev); 77 pci_dev_get(piar->pcidev);
77 return piar->pcidev; 78 return piar->edev;
78 } 79 }
79 } 80 }
80 } 81 }
@@ -83,7 +84,7 @@ static inline struct pci_dev *__pci_addr_cache_get_device(unsigned long addr)
83} 84}
84 85
85/** 86/**
86 * pci_addr_cache_get_device - Get device, given only address 87 * eeh_addr_cache_get_dev - Get device, given only address
87 * @addr: mmio (PIO) phys address or i/o port number 88 * @addr: mmio (PIO) phys address or i/o port number
88 * 89 *
89 * Given an mmio phys address, or a port number, find a pci device 90 * Given an mmio phys address, or a port number, find a pci device
@@ -92,15 +93,15 @@ static inline struct pci_dev *__pci_addr_cache_get_device(unsigned long addr)
92 * from zero (that is, they do *not* have pci_io_addr added in). 93 * from zero (that is, they do *not* have pci_io_addr added in).
93 * It is safe to call this function within an interrupt. 94 * It is safe to call this function within an interrupt.
94 */ 95 */
95struct pci_dev *pci_addr_cache_get_device(unsigned long addr) 96struct eeh_dev *eeh_addr_cache_get_dev(unsigned long addr)
96{ 97{
97 struct pci_dev *dev; 98 struct eeh_dev *edev;
98 unsigned long flags; 99 unsigned long flags;
99 100
100 spin_lock_irqsave(&pci_io_addr_cache_root.piar_lock, flags); 101 spin_lock_irqsave(&pci_io_addr_cache_root.piar_lock, flags);
101 dev = __pci_addr_cache_get_device(addr); 102 edev = __eeh_addr_cache_get_device(addr);
102 spin_unlock_irqrestore(&pci_io_addr_cache_root.piar_lock, flags); 103 spin_unlock_irqrestore(&pci_io_addr_cache_root.piar_lock, flags);
103 return dev; 104 return edev;
104} 105}
105 106
106#ifdef DEBUG 107#ifdef DEBUG
@@ -108,7 +109,7 @@ struct pci_dev *pci_addr_cache_get_device(unsigned long addr)
108 * Handy-dandy debug print routine, does nothing more 109 * Handy-dandy debug print routine, does nothing more
109 * than print out the contents of our addr cache. 110 * than print out the contents of our addr cache.
110 */ 111 */
111static void pci_addr_cache_print(struct pci_io_addr_cache *cache) 112static void eeh_addr_cache_print(struct pci_io_addr_cache *cache)
112{ 113{
113 struct rb_node *n; 114 struct rb_node *n;
114 int cnt = 0; 115 int cnt = 0;
@@ -117,7 +118,7 @@ static void pci_addr_cache_print(struct pci_io_addr_cache *cache)
117 while (n) { 118 while (n) {
118 struct pci_io_addr_range *piar; 119 struct pci_io_addr_range *piar;
119 piar = rb_entry(n, struct pci_io_addr_range, rb_node); 120 piar = rb_entry(n, struct pci_io_addr_range, rb_node);
120 printk(KERN_DEBUG "PCI: %s addr range %d [%lx-%lx]: %s\n", 121 pr_debug("PCI: %s addr range %d [%lx-%lx]: %s\n",
121 (piar->flags & IORESOURCE_IO) ? "i/o" : "mem", cnt, 122 (piar->flags & IORESOURCE_IO) ? "i/o" : "mem", cnt,
122 piar->addr_lo, piar->addr_hi, pci_name(piar->pcidev)); 123 piar->addr_lo, piar->addr_hi, pci_name(piar->pcidev));
123 cnt++; 124 cnt++;
@@ -128,7 +129,7 @@ static void pci_addr_cache_print(struct pci_io_addr_cache *cache)
128 129
129/* Insert address range into the rb tree. */ 130/* Insert address range into the rb tree. */
130static struct pci_io_addr_range * 131static struct pci_io_addr_range *
131pci_addr_cache_insert(struct pci_dev *dev, unsigned long alo, 132eeh_addr_cache_insert(struct pci_dev *dev, unsigned long alo,
132 unsigned long ahi, unsigned int flags) 133 unsigned long ahi, unsigned int flags)
133{ 134{
134 struct rb_node **p = &pci_io_addr_cache_root.rb_root.rb_node; 135 struct rb_node **p = &pci_io_addr_cache_root.rb_root.rb_node;
@@ -146,23 +147,24 @@ pci_addr_cache_insert(struct pci_dev *dev, unsigned long alo,
146 } else { 147 } else {
147 if (dev != piar->pcidev || 148 if (dev != piar->pcidev ||
148 alo != piar->addr_lo || ahi != piar->addr_hi) { 149 alo != piar->addr_lo || ahi != piar->addr_hi) {
149 printk(KERN_WARNING "PIAR: overlapping address range\n"); 150 pr_warning("PIAR: overlapping address range\n");
150 } 151 }
151 return piar; 152 return piar;
152 } 153 }
153 } 154 }
154 piar = kmalloc(sizeof(struct pci_io_addr_range), GFP_ATOMIC); 155 piar = kzalloc(sizeof(struct pci_io_addr_range), GFP_ATOMIC);
155 if (!piar) 156 if (!piar)
156 return NULL; 157 return NULL;
157 158
158 pci_dev_get(dev); 159 pci_dev_get(dev);
159 piar->addr_lo = alo; 160 piar->addr_lo = alo;
160 piar->addr_hi = ahi; 161 piar->addr_hi = ahi;
162 piar->edev = pci_dev_to_eeh_dev(dev);
161 piar->pcidev = dev; 163 piar->pcidev = dev;
162 piar->flags = flags; 164 piar->flags = flags;
163 165
164#ifdef DEBUG 166#ifdef DEBUG
165 printk(KERN_DEBUG "PIAR: insert range=[%lx:%lx] dev=%s\n", 167 pr_debug("PIAR: insert range=[%lx:%lx] dev=%s\n",
166 alo, ahi, pci_name(dev)); 168 alo, ahi, pci_name(dev));
167#endif 169#endif
168 170
@@ -172,7 +174,7 @@ pci_addr_cache_insert(struct pci_dev *dev, unsigned long alo,
172 return piar; 174 return piar;
173} 175}
174 176
175static void __pci_addr_cache_insert_device(struct pci_dev *dev) 177static void __eeh_addr_cache_insert_dev(struct pci_dev *dev)
176{ 178{
177 struct device_node *dn; 179 struct device_node *dn;
178 struct eeh_dev *edev; 180 struct eeh_dev *edev;
@@ -180,7 +182,7 @@ static void __pci_addr_cache_insert_device(struct pci_dev *dev)
180 182
181 dn = pci_device_to_OF_node(dev); 183 dn = pci_device_to_OF_node(dev);
182 if (!dn) { 184 if (!dn) {
183 printk(KERN_WARNING "PCI: no pci dn found for dev=%s\n", pci_name(dev)); 185 pr_warning("PCI: no pci dn found for dev=%s\n", pci_name(dev));
184 return; 186 return;
185 } 187 }
186 188
@@ -192,8 +194,7 @@ static void __pci_addr_cache_insert_device(struct pci_dev *dev)
192 } 194 }
193 195
194 /* Skip any devices for which EEH is not enabled. */ 196 /* Skip any devices for which EEH is not enabled. */
195 if (!(edev->mode & EEH_MODE_SUPPORTED) || 197 if (!edev->pe) {
196 edev->mode & EEH_MODE_NOCHECK) {
197#ifdef DEBUG 198#ifdef DEBUG
198 pr_info("PCI: skip building address cache for=%s - %s\n", 199 pr_info("PCI: skip building address cache for=%s - %s\n",
199 pci_name(dev), dn->full_name); 200 pci_name(dev), dn->full_name);
@@ -212,19 +213,19 @@ static void __pci_addr_cache_insert_device(struct pci_dev *dev)
212 continue; 213 continue;
213 if (start == 0 || ~start == 0 || end == 0 || ~end == 0) 214 if (start == 0 || ~start == 0 || end == 0 || ~end == 0)
214 continue; 215 continue;
215 pci_addr_cache_insert(dev, start, end, flags); 216 eeh_addr_cache_insert(dev, start, end, flags);
216 } 217 }
217} 218}
218 219
219/** 220/**
220 * pci_addr_cache_insert_device - Add a device to the address cache 221 * eeh_addr_cache_insert_dev - Add a device to the address cache
221 * @dev: PCI device whose I/O addresses we are interested in. 222 * @dev: PCI device whose I/O addresses we are interested in.
222 * 223 *
223 * In order to support the fast lookup of devices based on addresses, 224 * In order to support the fast lookup of devices based on addresses,
224 * we maintain a cache of devices that can be quickly searched. 225 * we maintain a cache of devices that can be quickly searched.
225 * This routine adds a device to that cache. 226 * This routine adds a device to that cache.
226 */ 227 */
227void pci_addr_cache_insert_device(struct pci_dev *dev) 228void eeh_addr_cache_insert_dev(struct pci_dev *dev)
228{ 229{
229 unsigned long flags; 230 unsigned long flags;
230 231
@@ -233,11 +234,11 @@ void pci_addr_cache_insert_device(struct pci_dev *dev)
233 return; 234 return;
234 235
235 spin_lock_irqsave(&pci_io_addr_cache_root.piar_lock, flags); 236 spin_lock_irqsave(&pci_io_addr_cache_root.piar_lock, flags);
236 __pci_addr_cache_insert_device(dev); 237 __eeh_addr_cache_insert_dev(dev);
237 spin_unlock_irqrestore(&pci_io_addr_cache_root.piar_lock, flags); 238 spin_unlock_irqrestore(&pci_io_addr_cache_root.piar_lock, flags);
238} 239}
239 240
240static inline void __pci_addr_cache_remove_device(struct pci_dev *dev) 241static inline void __eeh_addr_cache_rmv_dev(struct pci_dev *dev)
241{ 242{
242 struct rb_node *n; 243 struct rb_node *n;
243 244
@@ -258,7 +259,7 @@ restart:
258} 259}
259 260
260/** 261/**
261 * pci_addr_cache_remove_device - remove pci device from addr cache 262 * eeh_addr_cache_rmv_dev - remove pci device from addr cache
262 * @dev: device to remove 263 * @dev: device to remove
263 * 264 *
264 * Remove a device from the addr-cache tree. 265 * Remove a device from the addr-cache tree.
@@ -266,17 +267,17 @@ restart:
266 * the tree multiple times (once per resource). 267 * the tree multiple times (once per resource).
267 * But so what; device removal doesn't need to be that fast. 268 * But so what; device removal doesn't need to be that fast.
268 */ 269 */
269void pci_addr_cache_remove_device(struct pci_dev *dev) 270void eeh_addr_cache_rmv_dev(struct pci_dev *dev)
270{ 271{
271 unsigned long flags; 272 unsigned long flags;
272 273
273 spin_lock_irqsave(&pci_io_addr_cache_root.piar_lock, flags); 274 spin_lock_irqsave(&pci_io_addr_cache_root.piar_lock, flags);
274 __pci_addr_cache_remove_device(dev); 275 __eeh_addr_cache_rmv_dev(dev);
275 spin_unlock_irqrestore(&pci_io_addr_cache_root.piar_lock, flags); 276 spin_unlock_irqrestore(&pci_io_addr_cache_root.piar_lock, flags);
276} 277}
277 278
278/** 279/**
279 * pci_addr_cache_build - Build a cache of I/O addresses 280 * eeh_addr_cache_build - Build a cache of I/O addresses
280 * 281 *
281 * Build a cache of pci i/o addresses. This cache will be used to 282 * Build a cache of pci i/o addresses. This cache will be used to
282 * find the pci device that corresponds to a given address. 283 * find the pci device that corresponds to a given address.
@@ -284,7 +285,7 @@ void pci_addr_cache_remove_device(struct pci_dev *dev)
284 * Must be run late in boot process, after the pci controllers 285 * Must be run late in boot process, after the pci controllers
285 * have been scanned for devices (after all device resources are known). 286 * have been scanned for devices (after all device resources are known).
286 */ 287 */
287void __init pci_addr_cache_build(void) 288void __init eeh_addr_cache_build(void)
288{ 289{
289 struct device_node *dn; 290 struct device_node *dn;
290 struct eeh_dev *edev; 291 struct eeh_dev *edev;
@@ -293,7 +294,7 @@ void __init pci_addr_cache_build(void)
293 spin_lock_init(&pci_io_addr_cache_root.piar_lock); 294 spin_lock_init(&pci_io_addr_cache_root.piar_lock);
294 295
295 for_each_pci_dev(dev) { 296 for_each_pci_dev(dev) {
296 pci_addr_cache_insert_device(dev); 297 eeh_addr_cache_insert_dev(dev);
297 298
298 dn = pci_device_to_OF_node(dev); 299 dn = pci_device_to_OF_node(dev);
299 if (!dn) 300 if (!dn)
@@ -312,7 +313,7 @@ void __init pci_addr_cache_build(void)
312 313
313#ifdef DEBUG 314#ifdef DEBUG
314 /* Verify tree built up above, echo back the list of addrs. */ 315 /* Verify tree built up above, echo back the list of addrs. */
315 pci_addr_cache_print(&pci_io_addr_cache_root); 316 eeh_addr_cache_print(&pci_io_addr_cache_root);
316#endif 317#endif
317} 318}
318 319
diff --git a/arch/powerpc/platforms/pseries/eeh_dev.c b/arch/powerpc/platforms/pseries/eeh_dev.c
index c4507d095900..66442341d3a6 100644
--- a/arch/powerpc/platforms/pseries/eeh_dev.c
+++ b/arch/powerpc/platforms/pseries/eeh_dev.c
@@ -55,7 +55,7 @@ void * __devinit eeh_dev_init(struct device_node *dn, void *data)
55 struct eeh_dev *edev; 55 struct eeh_dev *edev;
56 56
57 /* Allocate EEH device */ 57 /* Allocate EEH device */
58 edev = zalloc_maybe_bootmem(sizeof(*edev), GFP_KERNEL); 58 edev = kzalloc(sizeof(*edev), GFP_KERNEL);
59 if (!edev) { 59 if (!edev) {
60 pr_warning("%s: out of memory\n", __func__); 60 pr_warning("%s: out of memory\n", __func__);
61 return NULL; 61 return NULL;
@@ -65,6 +65,7 @@ void * __devinit eeh_dev_init(struct device_node *dn, void *data)
65 PCI_DN(dn)->edev = edev; 65 PCI_DN(dn)->edev = edev;
66 edev->dn = dn; 66 edev->dn = dn;
67 edev->phb = phb; 67 edev->phb = phb;
68 INIT_LIST_HEAD(&edev->list);
68 69
69 return NULL; 70 return NULL;
70} 71}
@@ -80,6 +81,9 @@ void __devinit eeh_dev_phb_init_dynamic(struct pci_controller *phb)
80{ 81{
81 struct device_node *dn = phb->dn; 82 struct device_node *dn = phb->dn;
82 83
84 /* EEH PE for PHB */
85 eeh_phb_pe_create(phb);
86
83 /* EEH device for PHB */ 87 /* EEH device for PHB */
84 eeh_dev_init(dn, phb); 88 eeh_dev_init(dn, phb);
85 89
@@ -93,10 +97,16 @@ void __devinit eeh_dev_phb_init_dynamic(struct pci_controller *phb)
93 * Scan all the existing PHBs and create EEH devices for their OF 97 * Scan all the existing PHBs and create EEH devices for their OF
94 * nodes and their children OF nodes 98 * nodes and their children OF nodes
95 */ 99 */
96void __init eeh_dev_phb_init(void) 100static int __init eeh_dev_phb_init(void)
97{ 101{
98 struct pci_controller *phb, *tmp; 102 struct pci_controller *phb, *tmp;
99 103
100 list_for_each_entry_safe(phb, tmp, &hose_list, list_node) 104 list_for_each_entry_safe(phb, tmp, &hose_list, list_node)
101 eeh_dev_phb_init_dynamic(phb); 105 eeh_dev_phb_init_dynamic(phb);
106
107 pr_info("EEH: devices created\n");
108
109 return 0;
102} 110}
111
112core_initcall(eeh_dev_phb_init);
diff --git a/arch/powerpc/platforms/pseries/eeh_driver.c b/arch/powerpc/platforms/pseries/eeh_driver.c
index baf92cd9dfab..8370ce7d5931 100644
--- a/arch/powerpc/platforms/pseries/eeh_driver.c
+++ b/arch/powerpc/platforms/pseries/eeh_driver.c
@@ -93,7 +93,7 @@ static void eeh_disable_irq(struct pci_dev *dev)
93 if (!irq_has_action(dev->irq)) 93 if (!irq_has_action(dev->irq))
94 return; 94 return;
95 95
96 edev->mode |= EEH_MODE_IRQ_DISABLED; 96 edev->mode |= EEH_DEV_IRQ_DISABLED;
97 disable_irq_nosync(dev->irq); 97 disable_irq_nosync(dev->irq);
98} 98}
99 99
@@ -108,36 +108,43 @@ static void eeh_enable_irq(struct pci_dev *dev)
108{ 108{
109 struct eeh_dev *edev = pci_dev_to_eeh_dev(dev); 109 struct eeh_dev *edev = pci_dev_to_eeh_dev(dev);
110 110
111 if ((edev->mode) & EEH_MODE_IRQ_DISABLED) { 111 if ((edev->mode) & EEH_DEV_IRQ_DISABLED) {
112 edev->mode &= ~EEH_MODE_IRQ_DISABLED; 112 edev->mode &= ~EEH_DEV_IRQ_DISABLED;
113 enable_irq(dev->irq); 113 enable_irq(dev->irq);
114 } 114 }
115} 115}
116 116
117/** 117/**
118 * eeh_report_error - Report pci error to each device driver 118 * eeh_report_error - Report pci error to each device driver
119 * @dev: PCI device 119 * @data: eeh device
120 * @userdata: return value 120 * @userdata: return value
121 * 121 *
122 * Report an EEH error to each device driver, collect up and 122 * Report an EEH error to each device driver, collect up and
123 * merge the device driver responses. Cumulative response 123 * merge the device driver responses. Cumulative response
124 * passed back in "userdata". 124 * passed back in "userdata".
125 */ 125 */
126static int eeh_report_error(struct pci_dev *dev, void *userdata) 126static void *eeh_report_error(void *data, void *userdata)
127{ 127{
128 struct eeh_dev *edev = (struct eeh_dev *)data;
129 struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
128 enum pci_ers_result rc, *res = userdata; 130 enum pci_ers_result rc, *res = userdata;
129 struct pci_driver *driver = dev->driver; 131 struct pci_driver *driver = dev->driver;
130 132
133 /* We might not have the associated PCI device,
134 * then we should continue for next one.
135 */
136 if (!dev) return NULL;
137
131 dev->error_state = pci_channel_io_frozen; 138 dev->error_state = pci_channel_io_frozen;
132 139
133 if (!driver) 140 if (!driver)
134 return 0; 141 return NULL;
135 142
136 eeh_disable_irq(dev); 143 eeh_disable_irq(dev);
137 144
138 if (!driver->err_handler || 145 if (!driver->err_handler ||
139 !driver->err_handler->error_detected) 146 !driver->err_handler->error_detected)
140 return 0; 147 return NULL;
141 148
142 rc = driver->err_handler->error_detected(dev, pci_channel_io_frozen); 149 rc = driver->err_handler->error_detected(dev, pci_channel_io_frozen);
143 150
@@ -145,27 +152,31 @@ static int eeh_report_error(struct pci_dev *dev, void *userdata)
145 if (rc == PCI_ERS_RESULT_NEED_RESET) *res = rc; 152 if (rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
146 if (*res == PCI_ERS_RESULT_NONE) *res = rc; 153 if (*res == PCI_ERS_RESULT_NONE) *res = rc;
147 154
148 return 0; 155 return NULL;
149} 156}
150 157
151/** 158/**
152 * eeh_report_mmio_enabled - Tell drivers that MMIO has been enabled 159 * eeh_report_mmio_enabled - Tell drivers that MMIO has been enabled
153 * @dev: PCI device 160 * @data: eeh device
154 * @userdata: return value 161 * @userdata: return value
155 * 162 *
156 * Tells each device driver that IO ports, MMIO and config space I/O 163 * Tells each device driver that IO ports, MMIO and config space I/O
157 * are now enabled. Collects up and merges the device driver responses. 164 * are now enabled. Collects up and merges the device driver responses.
158 * Cumulative response passed back in "userdata". 165 * Cumulative response passed back in "userdata".
159 */ 166 */
160static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata) 167static void *eeh_report_mmio_enabled(void *data, void *userdata)
161{ 168{
169 struct eeh_dev *edev = (struct eeh_dev *)data;
170 struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
162 enum pci_ers_result rc, *res = userdata; 171 enum pci_ers_result rc, *res = userdata;
163 struct pci_driver *driver = dev->driver; 172 struct pci_driver *driver;
164 173
165 if (!driver || 174 if (!dev) return NULL;
175
176 if (!(driver = dev->driver) ||
166 !driver->err_handler || 177 !driver->err_handler ||
167 !driver->err_handler->mmio_enabled) 178 !driver->err_handler->mmio_enabled)
168 return 0; 179 return NULL;
169 180
170 rc = driver->err_handler->mmio_enabled(dev); 181 rc = driver->err_handler->mmio_enabled(dev);
171 182
@@ -173,12 +184,12 @@ static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata)
173 if (rc == PCI_ERS_RESULT_NEED_RESET) *res = rc; 184 if (rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
174 if (*res == PCI_ERS_RESULT_NONE) *res = rc; 185 if (*res == PCI_ERS_RESULT_NONE) *res = rc;
175 186
176 return 0; 187 return NULL;
177} 188}
178 189
179/** 190/**
180 * eeh_report_reset - Tell device that slot has been reset 191 * eeh_report_reset - Tell device that slot has been reset
181 * @dev: PCI device 192 * @data: eeh device
182 * @userdata: return value 193 * @userdata: return value
183 * 194 *
184 * This routine must be called while EEH tries to reset particular 195 * This routine must be called while EEH tries to reset particular
@@ -186,13 +197,15 @@ static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata)
186 * some actions, usually to save data the driver needs so that the 197 * some actions, usually to save data the driver needs so that the
187 * driver can work again while the device is recovered. 198 * driver can work again while the device is recovered.
188 */ 199 */
189static int eeh_report_reset(struct pci_dev *dev, void *userdata) 200static void *eeh_report_reset(void *data, void *userdata)
190{ 201{
202 struct eeh_dev *edev = (struct eeh_dev *)data;
203 struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
191 enum pci_ers_result rc, *res = userdata; 204 enum pci_ers_result rc, *res = userdata;
192 struct pci_driver *driver = dev->driver; 205 struct pci_driver *driver;
193 206
194 if (!driver) 207 if (!dev || !(driver = dev->driver))
195 return 0; 208 return NULL;
196 209
197 dev->error_state = pci_channel_io_normal; 210 dev->error_state = pci_channel_io_normal;
198 211
@@ -200,7 +213,7 @@ static int eeh_report_reset(struct pci_dev *dev, void *userdata)
200 213
201 if (!driver->err_handler || 214 if (!driver->err_handler ||
202 !driver->err_handler->slot_reset) 215 !driver->err_handler->slot_reset)
203 return 0; 216 return NULL;
204 217
205 rc = driver->err_handler->slot_reset(dev); 218 rc = driver->err_handler->slot_reset(dev);
206 if ((*res == PCI_ERS_RESULT_NONE) || 219 if ((*res == PCI_ERS_RESULT_NONE) ||
@@ -208,82 +221,89 @@ static int eeh_report_reset(struct pci_dev *dev, void *userdata)
208 if (*res == PCI_ERS_RESULT_DISCONNECT && 221 if (*res == PCI_ERS_RESULT_DISCONNECT &&
209 rc == PCI_ERS_RESULT_NEED_RESET) *res = rc; 222 rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
210 223
211 return 0; 224 return NULL;
212} 225}
213 226
214/** 227/**
215 * eeh_report_resume - Tell device to resume normal operations 228 * eeh_report_resume - Tell device to resume normal operations
216 * @dev: PCI device 229 * @data: eeh device
217 * @userdata: return value 230 * @userdata: return value
218 * 231 *
219 * This routine must be called to notify the device driver that it 232 * This routine must be called to notify the device driver that it
220 * could resume so that the device driver can do some initialization 233 * could resume so that the device driver can do some initialization
221 * to make the recovered device work again. 234 * to make the recovered device work again.
222 */ 235 */
223static int eeh_report_resume(struct pci_dev *dev, void *userdata) 236static void *eeh_report_resume(void *data, void *userdata)
224{ 237{
225 struct pci_driver *driver = dev->driver; 238 struct eeh_dev *edev = (struct eeh_dev *)data;
239 struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
240 struct pci_driver *driver;
241
242 if (!dev) return NULL;
226 243
227 dev->error_state = pci_channel_io_normal; 244 dev->error_state = pci_channel_io_normal;
228 245
229 if (!driver) 246 if (!(driver = dev->driver))
230 return 0; 247 return NULL;
231 248
232 eeh_enable_irq(dev); 249 eeh_enable_irq(dev);
233 250
234 if (!driver->err_handler || 251 if (!driver->err_handler ||
235 !driver->err_handler->resume) 252 !driver->err_handler->resume)
236 return 0; 253 return NULL;
237 254
238 driver->err_handler->resume(dev); 255 driver->err_handler->resume(dev);
239 256
240 return 0; 257 return NULL;
241} 258}
242 259
243/** 260/**
244 * eeh_report_failure - Tell device driver that device is dead. 261 * eeh_report_failure - Tell device driver that device is dead.
245 * @dev: PCI device 262 * @data: eeh device
246 * @userdata: return value 263 * @userdata: return value
247 * 264 *
248 * This informs the device driver that the device is permanently 265 * This informs the device driver that the device is permanently
249 * dead, and that no further recovery attempts will be made on it. 266 * dead, and that no further recovery attempts will be made on it.
250 */ 267 */
251static int eeh_report_failure(struct pci_dev *dev, void *userdata) 268static void *eeh_report_failure(void *data, void *userdata)
252{ 269{
253 struct pci_driver *driver = dev->driver; 270 struct eeh_dev *edev = (struct eeh_dev *)data;
271 struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
272 struct pci_driver *driver;
273
274 if (!dev) return NULL;
254 275
255 dev->error_state = pci_channel_io_perm_failure; 276 dev->error_state = pci_channel_io_perm_failure;
256 277
257 if (!driver) 278 if (!(driver = dev->driver))
258 return 0; 279 return NULL;
259 280
260 eeh_disable_irq(dev); 281 eeh_disable_irq(dev);
261 282
262 if (!driver->err_handler || 283 if (!driver->err_handler ||
263 !driver->err_handler->error_detected) 284 !driver->err_handler->error_detected)
264 return 0; 285 return NULL;
265 286
266 driver->err_handler->error_detected(dev, pci_channel_io_perm_failure); 287 driver->err_handler->error_detected(dev, pci_channel_io_perm_failure);
267 288
268 return 0; 289 return NULL;
269} 290}
270 291
271/** 292/**
272 * eeh_reset_device - Perform actual reset of a pci slot 293 * eeh_reset_device - Perform actual reset of a pci slot
273 * @edev: PE associated EEH device 294 * @pe: EEH PE
274 * @bus: PCI bus corresponding to the isolcated slot 295 * @bus: PCI bus corresponding to the isolcated slot
275 * 296 *
276 * This routine must be called to do reset on the indicated PE. 297 * This routine must be called to do reset on the indicated PE.
277 * During the reset, udev might be invoked because those affected 298 * During the reset, udev might be invoked because those affected
278 * PCI devices will be removed and then added. 299 * PCI devices will be removed and then added.
279 */ 300 */
280static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus) 301static int eeh_reset_device(struct eeh_pe *pe, struct pci_bus *bus)
281{ 302{
282 struct device_node *dn;
283 int cnt, rc; 303 int cnt, rc;
284 304
285 /* pcibios will clear the counter; save the value */ 305 /* pcibios will clear the counter; save the value */
286 cnt = edev->freeze_count; 306 cnt = pe->freeze_count;
287 307
288 if (bus) 308 if (bus)
289 pcibios_remove_pci_devices(bus); 309 pcibios_remove_pci_devices(bus);
@@ -292,25 +312,13 @@ static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
292 * Reconfigure bridges and devices. Don't try to bring the system 312 * Reconfigure bridges and devices. Don't try to bring the system
293 * up if the reset failed for some reason. 313 * up if the reset failed for some reason.
294 */ 314 */
295 rc = eeh_reset_pe(edev); 315 rc = eeh_reset_pe(pe);
296 if (rc) 316 if (rc)
297 return rc; 317 return rc;
298 318
299 /* Walk over all functions on this device. */ 319 /* Restore PE */
300 dn = eeh_dev_to_of_node(edev); 320 eeh_ops->configure_bridge(pe);
301 if (!pcibios_find_pci_bus(dn) && of_node_to_eeh_dev(dn->parent)) 321 eeh_pe_restore_bars(pe);
302 dn = dn->parent->child;
303
304 while (dn) {
305 struct eeh_dev *pedev = of_node_to_eeh_dev(dn);
306
307 /* On Power4, always true because eeh_pe_config_addr=0 */
308 if (edev->pe_config_addr == pedev->pe_config_addr) {
309 eeh_ops->configure_bridge(dn);
310 eeh_restore_bars(pedev);
311 }
312 dn = dn->sibling;
313 }
314 322
315 /* Give the system 5 seconds to finish running the user-space 323 /* Give the system 5 seconds to finish running the user-space
316 * hotplug shutdown scripts, e.g. ifdown for ethernet. Yes, 324 * hotplug shutdown scripts, e.g. ifdown for ethernet. Yes,
@@ -322,7 +330,7 @@ static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
322 ssleep(5); 330 ssleep(5);
323 pcibios_add_pci_devices(bus); 331 pcibios_add_pci_devices(bus);
324 } 332 }
325 edev->freeze_count = cnt; 333 pe->freeze_count = cnt;
326 334
327 return 0; 335 return 0;
328} 336}
@@ -334,7 +342,7 @@ static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
334 342
335/** 343/**
336 * eeh_handle_event - Reset a PCI device after hard lockup. 344 * eeh_handle_event - Reset a PCI device after hard lockup.
337 * @event: EEH event 345 * @pe: EEH PE
338 * 346 *
339 * While PHB detects address or data parity errors on particular PCI 347 * While PHB detects address or data parity errors on particular PCI
340 * slot, the associated PE will be frozen. Besides, DMA's occurring 348 * slot, the associated PE will be frozen. Besides, DMA's occurring
@@ -349,69 +357,24 @@ static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
349 * drivers (which cause a second set of hotplug events to go out to 357 * drivers (which cause a second set of hotplug events to go out to
350 * userspace). 358 * userspace).
351 */ 359 */
352struct eeh_dev *handle_eeh_events(struct eeh_event *event) 360void eeh_handle_event(struct eeh_pe *pe)
353{ 361{
354 struct device_node *frozen_dn;
355 struct eeh_dev *frozen_edev;
356 struct pci_bus *frozen_bus; 362 struct pci_bus *frozen_bus;
357 int rc = 0; 363 int rc = 0;
358 enum pci_ers_result result = PCI_ERS_RESULT_NONE; 364 enum pci_ers_result result = PCI_ERS_RESULT_NONE;
359 const char *location, *pci_str, *drv_str, *bus_pci_str, *bus_drv_str;
360
361 frozen_dn = eeh_find_device_pe(eeh_dev_to_of_node(event->edev));
362 if (!frozen_dn) {
363 location = of_get_property(eeh_dev_to_of_node(event->edev), "ibm,loc-code", NULL);
364 location = location ? location : "unknown";
365 printk(KERN_ERR "EEH: Error: Cannot find partition endpoint "
366 "for location=%s pci addr=%s\n",
367 location, eeh_pci_name(eeh_dev_to_pci_dev(event->edev)));
368 return NULL;
369 }
370
371 frozen_bus = pcibios_find_pci_bus(frozen_dn);
372 location = of_get_property(frozen_dn, "ibm,loc-code", NULL);
373 location = location ? location : "unknown";
374
375 /* There are two different styles for coming up with the PE.
376 * In the old style, it was the highest EEH-capable device
377 * which was always an EADS pci bridge. In the new style,
378 * there might not be any EADS bridges, and even when there are,
379 * the firmware marks them as "EEH incapable". So another
380 * two-step is needed to find the pci bus..
381 */
382 if (!frozen_bus)
383 frozen_bus = pcibios_find_pci_bus(frozen_dn->parent);
384 365
366 frozen_bus = eeh_pe_bus_get(pe);
385 if (!frozen_bus) { 367 if (!frozen_bus) {
386 printk(KERN_ERR "EEH: Cannot find PCI bus " 368 pr_err("%s: Cannot find PCI bus for PHB#%d-PE#%x\n",
387 "for location=%s dn=%s\n", 369 __func__, pe->phb->global_number, pe->addr);
388 location, frozen_dn->full_name); 370 return;
389 return NULL;
390 } 371 }
391 372
392 frozen_edev = of_node_to_eeh_dev(frozen_dn); 373 pe->freeze_count++;
393 frozen_edev->freeze_count++; 374 if (pe->freeze_count > EEH_MAX_ALLOWED_FREEZES)
394 pci_str = eeh_pci_name(eeh_dev_to_pci_dev(event->edev));
395 drv_str = eeh_pcid_name(eeh_dev_to_pci_dev(event->edev));
396
397 if (frozen_edev->freeze_count > EEH_MAX_ALLOWED_FREEZES)
398 goto excess_failures; 375 goto excess_failures;
399 376 pr_warning("EEH: This PCI device has failed %d times in the last hour\n",
400 printk(KERN_WARNING 377 pe->freeze_count);
401 "EEH: This PCI device has failed %d times in the last hour:\n",
402 frozen_edev->freeze_count);
403
404 if (frozen_edev->pdev) {
405 bus_pci_str = pci_name(frozen_edev->pdev);
406 bus_drv_str = eeh_pcid_name(frozen_edev->pdev);
407 printk(KERN_WARNING
408 "EEH: Bus location=%s driver=%s pci addr=%s\n",
409 location, bus_drv_str, bus_pci_str);
410 }
411
412 printk(KERN_WARNING
413 "EEH: Device location=%s driver=%s pci addr=%s\n",
414 location, drv_str, pci_str);
415 378
416 /* Walk the various device drivers attached to this slot through 379 /* Walk the various device drivers attached to this slot through
417 * a reset sequence, giving each an opportunity to do what it needs 380 * a reset sequence, giving each an opportunity to do what it needs
@@ -419,12 +382,12 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
419 * status ... if any child can't handle the reset, then the entire 382 * status ... if any child can't handle the reset, then the entire
420 * slot is dlpar removed and added. 383 * slot is dlpar removed and added.
421 */ 384 */
422 pci_walk_bus(frozen_bus, eeh_report_error, &result); 385 eeh_pe_dev_traverse(pe, eeh_report_error, &result);
423 386
424 /* Get the current PCI slot state. This can take a long time, 387 /* Get the current PCI slot state. This can take a long time,
425 * sometimes over 3 seconds for certain systems. 388 * sometimes over 3 seconds for certain systems.
426 */ 389 */
427 rc = eeh_ops->wait_state(eeh_dev_to_of_node(frozen_edev), MAX_WAIT_FOR_RECOVERY*1000); 390 rc = eeh_ops->wait_state(pe, MAX_WAIT_FOR_RECOVERY*1000);
428 if (rc < 0 || rc == EEH_STATE_NOT_SUPPORT) { 391 if (rc < 0 || rc == EEH_STATE_NOT_SUPPORT) {
429 printk(KERN_WARNING "EEH: Permanent failure\n"); 392 printk(KERN_WARNING "EEH: Permanent failure\n");
430 goto hard_fail; 393 goto hard_fail;
@@ -434,14 +397,14 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
434 * don't post the error log until after all dev drivers 397 * don't post the error log until after all dev drivers
435 * have been informed. 398 * have been informed.
436 */ 399 */
437 eeh_slot_error_detail(frozen_edev, EEH_LOG_TEMP); 400 eeh_slot_error_detail(pe, EEH_LOG_TEMP);
438 401
439 /* If all device drivers were EEH-unaware, then shut 402 /* If all device drivers were EEH-unaware, then shut
440 * down all of the device drivers, and hope they 403 * down all of the device drivers, and hope they
441 * go down willingly, without panicing the system. 404 * go down willingly, without panicing the system.
442 */ 405 */
443 if (result == PCI_ERS_RESULT_NONE) { 406 if (result == PCI_ERS_RESULT_NONE) {
444 rc = eeh_reset_device(frozen_edev, frozen_bus); 407 rc = eeh_reset_device(pe, frozen_bus);
445 if (rc) { 408 if (rc) {
446 printk(KERN_WARNING "EEH: Unable to reset, rc=%d\n", rc); 409 printk(KERN_WARNING "EEH: Unable to reset, rc=%d\n", rc);
447 goto hard_fail; 410 goto hard_fail;
@@ -450,7 +413,7 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
450 413
451 /* If all devices reported they can proceed, then re-enable MMIO */ 414 /* If all devices reported they can proceed, then re-enable MMIO */
452 if (result == PCI_ERS_RESULT_CAN_RECOVER) { 415 if (result == PCI_ERS_RESULT_CAN_RECOVER) {
453 rc = eeh_pci_enable(frozen_edev, EEH_OPT_THAW_MMIO); 416 rc = eeh_pci_enable(pe, EEH_OPT_THAW_MMIO);
454 417
455 if (rc < 0) 418 if (rc < 0)
456 goto hard_fail; 419 goto hard_fail;
@@ -458,13 +421,13 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
458 result = PCI_ERS_RESULT_NEED_RESET; 421 result = PCI_ERS_RESULT_NEED_RESET;
459 } else { 422 } else {
460 result = PCI_ERS_RESULT_NONE; 423 result = PCI_ERS_RESULT_NONE;
461 pci_walk_bus(frozen_bus, eeh_report_mmio_enabled, &result); 424 eeh_pe_dev_traverse(pe, eeh_report_mmio_enabled, &result);
462 } 425 }
463 } 426 }
464 427
465 /* If all devices reported they can proceed, then re-enable DMA */ 428 /* If all devices reported they can proceed, then re-enable DMA */
466 if (result == PCI_ERS_RESULT_CAN_RECOVER) { 429 if (result == PCI_ERS_RESULT_CAN_RECOVER) {
467 rc = eeh_pci_enable(frozen_edev, EEH_OPT_THAW_DMA); 430 rc = eeh_pci_enable(pe, EEH_OPT_THAW_DMA);
468 431
469 if (rc < 0) 432 if (rc < 0)
470 goto hard_fail; 433 goto hard_fail;
@@ -482,13 +445,13 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
482 445
483 /* If any device called out for a reset, then reset the slot */ 446 /* If any device called out for a reset, then reset the slot */
484 if (result == PCI_ERS_RESULT_NEED_RESET) { 447 if (result == PCI_ERS_RESULT_NEED_RESET) {
485 rc = eeh_reset_device(frozen_edev, NULL); 448 rc = eeh_reset_device(pe, NULL);
486 if (rc) { 449 if (rc) {
487 printk(KERN_WARNING "EEH: Cannot reset, rc=%d\n", rc); 450 printk(KERN_WARNING "EEH: Cannot reset, rc=%d\n", rc);
488 goto hard_fail; 451 goto hard_fail;
489 } 452 }
490 result = PCI_ERS_RESULT_NONE; 453 result = PCI_ERS_RESULT_NONE;
491 pci_walk_bus(frozen_bus, eeh_report_reset, &result); 454 eeh_pe_dev_traverse(pe, eeh_report_reset, &result);
492 } 455 }
493 456
494 /* All devices should claim they have recovered by now. */ 457 /* All devices should claim they have recovered by now. */
@@ -499,9 +462,9 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
499 } 462 }
500 463
501 /* Tell all device drivers that they can resume operations */ 464 /* Tell all device drivers that they can resume operations */
502 pci_walk_bus(frozen_bus, eeh_report_resume, NULL); 465 eeh_pe_dev_traverse(pe, eeh_report_resume, NULL);
503 466
504 return frozen_edev; 467 return;
505 468
506excess_failures: 469excess_failures:
507 /* 470 /*
@@ -509,30 +472,26 @@ excess_failures:
509 * are due to poorly seated PCI cards. Only 10% or so are 472 * are due to poorly seated PCI cards. Only 10% or so are
510 * due to actual, failed cards. 473 * due to actual, failed cards.
511 */ 474 */
512 printk(KERN_ERR 475 pr_err("EEH: PHB#%d-PE#%x has failed %d times in the\n"
513 "EEH: PCI device at location=%s driver=%s pci addr=%s\n" 476 "last hour and has been permanently disabled.\n"
514 "has failed %d times in the last hour " 477 "Please try reseating or replacing it.\n",
515 "and has been permanently disabled.\n" 478 pe->phb->global_number, pe->addr,
516 "Please try reseating this device or replacing it.\n", 479 pe->freeze_count);
517 location, drv_str, pci_str, frozen_edev->freeze_count);
518 goto perm_error; 480 goto perm_error;
519 481
520hard_fail: 482hard_fail:
521 printk(KERN_ERR 483 pr_err("EEH: Unable to recover from failure from PHB#%d-PE#%x.\n"
522 "EEH: Unable to recover from failure of PCI device " 484 "Please try reseating or replacing it\n",
523 "at location=%s driver=%s pci addr=%s\n" 485 pe->phb->global_number, pe->addr);
524 "Please try reseating this device or replacing it.\n",
525 location, drv_str, pci_str);
526 486
527perm_error: 487perm_error:
528 eeh_slot_error_detail(frozen_edev, EEH_LOG_PERM); 488 eeh_slot_error_detail(pe, EEH_LOG_PERM);
529 489
530 /* Notify all devices that they're about to go down. */ 490 /* Notify all devices that they're about to go down. */
531 pci_walk_bus(frozen_bus, eeh_report_failure, NULL); 491 eeh_pe_dev_traverse(pe, eeh_report_failure, NULL);
532 492
533 /* Shut down the device drivers for good. */ 493 /* Shut down the device drivers for good. */
534 pcibios_remove_pci_devices(frozen_bus); 494 if (frozen_bus)
535 495 pcibios_remove_pci_devices(frozen_bus);
536 return NULL;
537} 496}
538 497
diff --git a/arch/powerpc/platforms/pseries/eeh_event.c b/arch/powerpc/platforms/pseries/eeh_event.c
index fb506317ebb0..51faaac8abe6 100644
--- a/arch/powerpc/platforms/pseries/eeh_event.c
+++ b/arch/powerpc/platforms/pseries/eeh_event.c
@@ -57,7 +57,7 @@ static int eeh_event_handler(void * dummy)
57{ 57{
58 unsigned long flags; 58 unsigned long flags;
59 struct eeh_event *event; 59 struct eeh_event *event;
60 struct eeh_dev *edev; 60 struct eeh_pe *pe;
61 61
62 set_task_comm(current, "eehd"); 62 set_task_comm(current, "eehd");
63 63
@@ -76,28 +76,23 @@ static int eeh_event_handler(void * dummy)
76 76
77 /* Serialize processing of EEH events */ 77 /* Serialize processing of EEH events */
78 mutex_lock(&eeh_event_mutex); 78 mutex_lock(&eeh_event_mutex);
79 edev = event->edev; 79 pe = event->pe;
80 eeh_mark_slot(eeh_dev_to_of_node(edev), EEH_MODE_RECOVERING); 80 eeh_pe_state_mark(pe, EEH_PE_RECOVERING);
81 81 pr_info("EEH: Detected PCI bus error on PHB#%d-PE#%x\n",
82 printk(KERN_INFO "EEH: Detected PCI bus error on device %s\n", 82 pe->phb->global_number, pe->addr);
83 eeh_pci_name(edev->pdev));
84 83
85 set_current_state(TASK_INTERRUPTIBLE); /* Don't add to load average */ 84 set_current_state(TASK_INTERRUPTIBLE); /* Don't add to load average */
86 edev = handle_eeh_events(event); 85 eeh_handle_event(pe);
87 86 eeh_pe_state_clear(pe, EEH_PE_RECOVERING);
88 if (edev) {
89 eeh_clear_slot(eeh_dev_to_of_node(edev), EEH_MODE_RECOVERING);
90 pci_dev_put(edev->pdev);
91 }
92 87
93 kfree(event); 88 kfree(event);
94 mutex_unlock(&eeh_event_mutex); 89 mutex_unlock(&eeh_event_mutex);
95 90
96 /* If there are no new errors after an hour, clear the counter. */ 91 /* If there are no new errors after an hour, clear the counter. */
97 if (edev && edev->freeze_count>0) { 92 if (pe && pe->freeze_count > 0) {
98 msleep_interruptible(3600*1000); 93 msleep_interruptible(3600*1000);
99 if (edev->freeze_count>0) 94 if (pe->freeze_count > 0)
100 edev->freeze_count--; 95 pe->freeze_count--;
101 96
102 } 97 }
103 98
@@ -119,36 +114,23 @@ static void eeh_thread_launcher(struct work_struct *dummy)
119 114
120/** 115/**
121 * eeh_send_failure_event - Generate a PCI error event 116 * eeh_send_failure_event - Generate a PCI error event
122 * @edev: EEH device 117 * @pe: EEH PE
123 * 118 *
124 * This routine can be called within an interrupt context; 119 * This routine can be called within an interrupt context;
125 * the actual event will be delivered in a normal context 120 * the actual event will be delivered in a normal context
126 * (from a workqueue). 121 * (from a workqueue).
127 */ 122 */
128int eeh_send_failure_event(struct eeh_dev *edev) 123int eeh_send_failure_event(struct eeh_pe *pe)
129{ 124{
130 unsigned long flags; 125 unsigned long flags;
131 struct eeh_event *event; 126 struct eeh_event *event;
132 struct device_node *dn = eeh_dev_to_of_node(edev);
133 const char *location;
134
135 if (!mem_init_done) {
136 printk(KERN_ERR "EEH: event during early boot not handled\n");
137 location = of_get_property(dn, "ibm,loc-code", NULL);
138 printk(KERN_ERR "EEH: device node = %s\n", dn->full_name);
139 printk(KERN_ERR "EEH: PCI location = %s\n", location);
140 return 1;
141 }
142 event = kmalloc(sizeof(*event), GFP_ATOMIC);
143 if (event == NULL) {
144 printk(KERN_ERR "EEH: out of memory, event not handled\n");
145 return 1;
146 }
147
148 if (edev->pdev)
149 pci_dev_get(edev->pdev);
150 127
151 event->edev = edev; 128 event = kzalloc(sizeof(*event), GFP_ATOMIC);
129 if (!event) {
130 pr_err("EEH: out of memory, event not handled\n");
131 return -ENOMEM;
132 }
133 event->pe = pe;
152 134
153 /* We may or may not be called in an interrupt context */ 135 /* We may or may not be called in an interrupt context */
154 spin_lock_irqsave(&eeh_eventlist_lock, flags); 136 spin_lock_irqsave(&eeh_eventlist_lock, flags);
diff --git a/arch/powerpc/platforms/pseries/eeh_pe.c b/arch/powerpc/platforms/pseries/eeh_pe.c
new file mode 100644
index 000000000000..904123c7657b
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/eeh_pe.c
@@ -0,0 +1,591 @@
1/*
2 * The file intends to implement PE based on the information from
3 * platforms. Basically, there have 3 types of PEs: PHB/Bus/Device.
4 * All the PEs should be organized as hierarchy tree. The first level
5 * of the tree will be associated to existing PHBs since the particular
6 * PE is only meaningful in one PHB domain.
7 *
8 * Copyright Benjamin Herrenschmidt & Gavin Shan, IBM Corporation 2012.
9 *
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
14 *
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
19 *
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
23 */
24
25#include <linux/export.h>
26#include <linux/gfp.h>
27#include <linux/init.h>
28#include <linux/kernel.h>
29#include <linux/pci.h>
30#include <linux/string.h>
31
32#include <asm/pci-bridge.h>
33#include <asm/ppc-pci.h>
34
35static LIST_HEAD(eeh_phb_pe);
36
37/**
38 * eeh_pe_alloc - Allocate PE
39 * @phb: PCI controller
40 * @type: PE type
41 *
42 * Allocate PE instance dynamically.
43 */
44static struct eeh_pe *eeh_pe_alloc(struct pci_controller *phb, int type)
45{
46 struct eeh_pe *pe;
47
48 /* Allocate PHB PE */
49 pe = kzalloc(sizeof(struct eeh_pe), GFP_KERNEL);
50 if (!pe) return NULL;
51
52 /* Initialize PHB PE */
53 pe->type = type;
54 pe->phb = phb;
55 INIT_LIST_HEAD(&pe->child_list);
56 INIT_LIST_HEAD(&pe->child);
57 INIT_LIST_HEAD(&pe->edevs);
58
59 return pe;
60}
61
62/**
63 * eeh_phb_pe_create - Create PHB PE
64 * @phb: PCI controller
65 *
66 * The function should be called while the PHB is detected during
67 * system boot or PCI hotplug in order to create PHB PE.
68 */
69int __devinit eeh_phb_pe_create(struct pci_controller *phb)
70{
71 struct eeh_pe *pe;
72
73 /* Allocate PHB PE */
74 pe = eeh_pe_alloc(phb, EEH_PE_PHB);
75 if (!pe) {
76 pr_err("%s: out of memory!\n", __func__);
77 return -ENOMEM;
78 }
79
80 /* Put it into the list */
81 eeh_lock();
82 list_add_tail(&pe->child, &eeh_phb_pe);
83 eeh_unlock();
84
85 pr_debug("EEH: Add PE for PHB#%d\n", phb->global_number);
86
87 return 0;
88}
89
90/**
91 * eeh_phb_pe_get - Retrieve PHB PE based on the given PHB
92 * @phb: PCI controller
93 *
94 * The overall PEs form hierarchy tree. The first layer of the
95 * hierarchy tree is composed of PHB PEs. The function is used
96 * to retrieve the corresponding PHB PE according to the given PHB.
97 */
98static struct eeh_pe *eeh_phb_pe_get(struct pci_controller *phb)
99{
100 struct eeh_pe *pe;
101
102 eeh_lock();
103
104 list_for_each_entry(pe, &eeh_phb_pe, child) {
105 /*
106 * Actually, we needn't check the type since
107 * the PE for PHB has been determined when that
108 * was created.
109 */
110 if (pe->type == EEH_PE_PHB &&
111 pe->phb == phb) {
112 eeh_unlock();
113 return pe;
114 }
115 }
116
117 eeh_unlock();
118
119 return NULL;
120}
121
122/**
123 * eeh_pe_next - Retrieve the next PE in the tree
124 * @pe: current PE
125 * @root: root PE
126 *
127 * The function is used to retrieve the next PE in the
128 * hierarchy PE tree.
129 */
130static struct eeh_pe *eeh_pe_next(struct eeh_pe *pe,
131 struct eeh_pe *root)
132{
133 struct list_head *next = pe->child_list.next;
134
135 if (next == &pe->child_list) {
136 while (1) {
137 if (pe == root)
138 return NULL;
139 next = pe->child.next;
140 if (next != &pe->parent->child_list)
141 break;
142 pe = pe->parent;
143 }
144 }
145
146 return list_entry(next, struct eeh_pe, child);
147}
148
149/**
150 * eeh_pe_traverse - Traverse PEs in the specified PHB
151 * @root: root PE
152 * @fn: callback
153 * @flag: extra parameter to callback
154 *
155 * The function is used to traverse the specified PE and its
156 * child PEs. The traversing is to be terminated once the
157 * callback returns something other than NULL, or no more PEs
158 * to be traversed.
159 */
160static void *eeh_pe_traverse(struct eeh_pe *root,
161 eeh_traverse_func fn, void *flag)
162{
163 struct eeh_pe *pe;
164 void *ret;
165
166 for (pe = root; pe; pe = eeh_pe_next(pe, root)) {
167 ret = fn(pe, flag);
168 if (ret) return ret;
169 }
170
171 return NULL;
172}
173
174/**
175 * eeh_pe_dev_traverse - Traverse the devices from the PE
176 * @root: EEH PE
177 * @fn: function callback
178 * @flag: extra parameter to callback
179 *
180 * The function is used to traverse the devices of the specified
181 * PE and its child PEs.
182 */
183void *eeh_pe_dev_traverse(struct eeh_pe *root,
184 eeh_traverse_func fn, void *flag)
185{
186 struct eeh_pe *pe;
187 struct eeh_dev *edev;
188 void *ret;
189
190 if (!root) {
191 pr_warning("%s: Invalid PE %p\n", __func__, root);
192 return NULL;
193 }
194
195 /* Traverse root PE */
196 for (pe = root; pe; pe = eeh_pe_next(pe, root)) {
197 eeh_pe_for_each_dev(pe, edev) {
198 ret = fn(edev, flag);
199 if (ret) return ret;
200 }
201 }
202
203 return NULL;
204}
205
206/**
207 * __eeh_pe_get - Check the PE address
208 * @data: EEH PE
209 * @flag: EEH device
210 *
211 * For one particular PE, it can be identified by PE address
212 * or tranditional BDF address. BDF address is composed of
213 * Bus/Device/Function number. The extra data referred by flag
214 * indicates which type of address should be used.
215 */
216static void *__eeh_pe_get(void *data, void *flag)
217{
218 struct eeh_pe *pe = (struct eeh_pe *)data;
219 struct eeh_dev *edev = (struct eeh_dev *)flag;
220
221 /* Unexpected PHB PE */
222 if (pe->type == EEH_PE_PHB)
223 return NULL;
224
225 /* We prefer PE address */
226 if (edev->pe_config_addr &&
227 (edev->pe_config_addr == pe->addr))
228 return pe;
229
230 /* Try BDF address */
231 if (edev->pe_config_addr &&
232 (edev->config_addr == pe->config_addr))
233 return pe;
234
235 return NULL;
236}
237
238/**
239 * eeh_pe_get - Search PE based on the given address
240 * @edev: EEH device
241 *
242 * Search the corresponding PE based on the specified address which
243 * is included in the eeh device. The function is used to check if
244 * the associated PE has been created against the PE address. It's
245 * notable that the PE address has 2 format: traditional PE address
246 * which is composed of PCI bus/device/function number, or unified
247 * PE address.
248 */
249static struct eeh_pe *eeh_pe_get(struct eeh_dev *edev)
250{
251 struct eeh_pe *root = eeh_phb_pe_get(edev->phb);
252 struct eeh_pe *pe;
253
254 eeh_lock();
255 pe = eeh_pe_traverse(root, __eeh_pe_get, edev);
256 eeh_unlock();
257
258 return pe;
259}
260
261/**
262 * eeh_pe_get_parent - Retrieve the parent PE
263 * @edev: EEH device
264 *
265 * The whole PEs existing in the system are organized as hierarchy
266 * tree. The function is used to retrieve the parent PE according
267 * to the parent EEH device.
268 */
269static struct eeh_pe *eeh_pe_get_parent(struct eeh_dev *edev)
270{
271 struct device_node *dn;
272 struct eeh_dev *parent;
273
274 /*
275 * It might have the case for the indirect parent
276 * EEH device already having associated PE, but
277 * the direct parent EEH device doesn't have yet.
278 */
279 dn = edev->dn->parent;
280 while (dn) {
281 /* We're poking out of PCI territory */
282 if (!PCI_DN(dn)) return NULL;
283
284 parent = of_node_to_eeh_dev(dn);
285 /* We're poking out of PCI territory */
286 if (!parent) return NULL;
287
288 if (parent->pe)
289 return parent->pe;
290
291 dn = dn->parent;
292 }
293
294 return NULL;
295}
296
297/**
298 * eeh_add_to_parent_pe - Add EEH device to parent PE
299 * @edev: EEH device
300 *
301 * Add EEH device to the parent PE. If the parent PE already
302 * exists, the PE type will be changed to EEH_PE_BUS. Otherwise,
303 * we have to create new PE to hold the EEH device and the new
304 * PE will be linked to its parent PE as well.
305 */
306int eeh_add_to_parent_pe(struct eeh_dev *edev)
307{
308 struct eeh_pe *pe, *parent;
309
310 /*
311 * Search the PE has been existing or not according
312 * to the PE address. If that has been existing, the
313 * PE should be composed of PCI bus and its subordinate
314 * components.
315 */
316 pe = eeh_pe_get(edev);
317 if (pe) {
318 if (!edev->pe_config_addr) {
319 pr_err("%s: PE with addr 0x%x already exists\n",
320 __func__, edev->config_addr);
321 return -EEXIST;
322 }
323
324 /* Mark the PE as type of PCI bus */
325 pe->type = EEH_PE_BUS;
326 edev->pe = pe;
327
328 /* Put the edev to PE */
329 list_add_tail(&edev->list, &pe->edevs);
330 pr_debug("EEH: Add %s to Bus PE#%x\n",
331 edev->dn->full_name, pe->addr);
332
333 return 0;
334 }
335
336 /* Create a new EEH PE */
337 pe = eeh_pe_alloc(edev->phb, EEH_PE_DEVICE);
338 if (!pe) {
339 pr_err("%s: out of memory!\n", __func__);
340 return -ENOMEM;
341 }
342 pe->addr = edev->pe_config_addr;
343 pe->config_addr = edev->config_addr;
344
345 /*
346 * Put the new EEH PE into hierarchy tree. If the parent
347 * can't be found, the newly created PE will be attached
348 * to PHB directly. Otherwise, we have to associate the
349 * PE with its parent.
350 */
351 parent = eeh_pe_get_parent(edev);
352 if (!parent) {
353 parent = eeh_phb_pe_get(edev->phb);
354 if (!parent) {
355 pr_err("%s: No PHB PE is found (PHB Domain=%d)\n",
356 __func__, edev->phb->global_number);
357 edev->pe = NULL;
358 kfree(pe);
359 return -EEXIST;
360 }
361 }
362 pe->parent = parent;
363
364 /*
365 * Put the newly created PE into the child list and
366 * link the EEH device accordingly.
367 */
368 list_add_tail(&pe->child, &parent->child_list);
369 list_add_tail(&edev->list, &pe->edevs);
370 edev->pe = pe;
371 pr_debug("EEH: Add %s to Device PE#%x, Parent PE#%x\n",
372 edev->dn->full_name, pe->addr, pe->parent->addr);
373
374 return 0;
375}
376
377/**
378 * eeh_rmv_from_parent_pe - Remove one EEH device from the associated PE
379 * @edev: EEH device
380 *
381 * The PE hierarchy tree might be changed when doing PCI hotplug.
382 * Also, the PCI devices or buses could be removed from the system
383 * during EEH recovery. So we have to call the function remove the
384 * corresponding PE accordingly if necessary.
385 */
386int eeh_rmv_from_parent_pe(struct eeh_dev *edev)
387{
388 struct eeh_pe *pe, *parent;
389
390 if (!edev->pe) {
391 pr_warning("%s: No PE found for EEH device %s\n",
392 __func__, edev->dn->full_name);
393 return -EEXIST;
394 }
395
396 /* Remove the EEH device */
397 pe = edev->pe;
398 edev->pe = NULL;
399 list_del(&edev->list);
400
401 /*
402 * Check if the parent PE includes any EEH devices.
403 * If not, we should delete that. Also, we should
404 * delete the parent PE if it doesn't have associated
405 * child PEs and EEH devices.
406 */
407 while (1) {
408 parent = pe->parent;
409 if (pe->type == EEH_PE_PHB)
410 break;
411
412 if (list_empty(&pe->edevs) &&
413 list_empty(&pe->child_list)) {
414 list_del(&pe->child);
415 kfree(pe);
416 }
417
418 pe = parent;
419 }
420
421 return 0;
422}
423
424/**
425 * __eeh_pe_state_mark - Mark the state for the PE
426 * @data: EEH PE
427 * @flag: state
428 *
429 * The function is used to mark the indicated state for the given
430 * PE. Also, the associated PCI devices will be put into IO frozen
431 * state as well.
432 */
433static void *__eeh_pe_state_mark(void *data, void *flag)
434{
435 struct eeh_pe *pe = (struct eeh_pe *)data;
436 int state = *((int *)flag);
437 struct eeh_dev *tmp;
438 struct pci_dev *pdev;
439
440 /*
441 * Mark the PE with the indicated state. Also,
442 * the associated PCI device will be put into
443 * I/O frozen state to avoid I/O accesses from
444 * the PCI device driver.
445 */
446 pe->state |= state;
447 eeh_pe_for_each_dev(pe, tmp) {
448 pdev = eeh_dev_to_pci_dev(tmp);
449 if (pdev)
450 pdev->error_state = pci_channel_io_frozen;
451 }
452
453 return NULL;
454}
455
456/**
457 * eeh_pe_state_mark - Mark specified state for PE and its associated device
458 * @pe: EEH PE
459 *
460 * EEH error affects the current PE and its child PEs. The function
461 * is used to mark appropriate state for the affected PEs and the
462 * associated devices.
463 */
464void eeh_pe_state_mark(struct eeh_pe *pe, int state)
465{
466 eeh_pe_traverse(pe, __eeh_pe_state_mark, &state);
467}
468
469/**
470 * __eeh_pe_state_clear - Clear state for the PE
471 * @data: EEH PE
472 * @flag: state
473 *
474 * The function is used to clear the indicated state from the
475 * given PE. Besides, we also clear the check count of the PE
476 * as well.
477 */
478static void *__eeh_pe_state_clear(void *data, void *flag)
479{
480 struct eeh_pe *pe = (struct eeh_pe *)data;
481 int state = *((int *)flag);
482
483 pe->state &= ~state;
484 pe->check_count = 0;
485
486 return NULL;
487}
488
489/**
490 * eeh_pe_state_clear - Clear state for the PE and its children
491 * @pe: PE
492 * @state: state to be cleared
493 *
494 * When the PE and its children has been recovered from error,
495 * we need clear the error state for that. The function is used
496 * for the purpose.
497 */
498void eeh_pe_state_clear(struct eeh_pe *pe, int state)
499{
500 eeh_pe_traverse(pe, __eeh_pe_state_clear, &state);
501}
502
503/**
504 * eeh_restore_one_device_bars - Restore the Base Address Registers for one device
505 * @data: EEH device
506 * @flag: Unused
507 *
508 * Loads the PCI configuration space base address registers,
509 * the expansion ROM base address, the latency timer, and etc.
510 * from the saved values in the device node.
511 */
512static void *eeh_restore_one_device_bars(void *data, void *flag)
513{
514 int i;
515 u32 cmd;
516 struct eeh_dev *edev = (struct eeh_dev *)data;
517 struct device_node *dn = eeh_dev_to_of_node(edev);
518
519 for (i = 4; i < 10; i++)
520 eeh_ops->write_config(dn, i*4, 4, edev->config_space[i]);
521 /* 12 == Expansion ROM Address */
522 eeh_ops->write_config(dn, 12*4, 4, edev->config_space[12]);
523
524#define BYTE_SWAP(OFF) (8*((OFF)/4)+3-(OFF))
525#define SAVED_BYTE(OFF) (((u8 *)(edev->config_space))[BYTE_SWAP(OFF)])
526
527 eeh_ops->write_config(dn, PCI_CACHE_LINE_SIZE, 1,
528 SAVED_BYTE(PCI_CACHE_LINE_SIZE));
529 eeh_ops->write_config(dn, PCI_LATENCY_TIMER, 1,
530 SAVED_BYTE(PCI_LATENCY_TIMER));
531
532 /* max latency, min grant, interrupt pin and line */
533 eeh_ops->write_config(dn, 15*4, 4, edev->config_space[15]);
534
535 /*
536 * Restore PERR & SERR bits, some devices require it,
537 * don't touch the other command bits
538 */
539 eeh_ops->read_config(dn, PCI_COMMAND, 4, &cmd);
540 if (edev->config_space[1] & PCI_COMMAND_PARITY)
541 cmd |= PCI_COMMAND_PARITY;
542 else
543 cmd &= ~PCI_COMMAND_PARITY;
544 if (edev->config_space[1] & PCI_COMMAND_SERR)
545 cmd |= PCI_COMMAND_SERR;
546 else
547 cmd &= ~PCI_COMMAND_SERR;
548 eeh_ops->write_config(dn, PCI_COMMAND, 4, cmd);
549
550 return NULL;
551}
552
553/**
554 * eeh_pe_restore_bars - Restore the PCI config space info
555 * @pe: EEH PE
556 *
557 * This routine performs a recursive walk to the children
558 * of this device as well.
559 */
560void eeh_pe_restore_bars(struct eeh_pe *pe)
561{
562 eeh_pe_dev_traverse(pe, eeh_restore_one_device_bars, NULL);
563}
564
565/**
566 * eeh_pe_bus_get - Retrieve PCI bus according to the given PE
567 * @pe: EEH PE
568 *
569 * Retrieve the PCI bus according to the given PE. Basically,
570 * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the
571 * primary PCI bus will be retrieved. The parent bus will be
572 * returned for BUS PE. However, we don't have associated PCI
573 * bus for DEVICE PE.
574 */
575struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe)
576{
577 struct pci_bus *bus = NULL;
578 struct eeh_dev *edev;
579 struct pci_dev *pdev;
580
581 if (pe->type == EEH_PE_PHB) {
582 bus = pe->phb->bus;
583 } else if (pe->type == EEH_PE_BUS) {
584 edev = list_first_entry(&pe->edevs, struct eeh_dev, list);
585 pdev = eeh_dev_to_pci_dev(edev);
586 if (pdev)
587 bus = pdev->bus;
588 }
589
590 return bus;
591}
diff --git a/arch/powerpc/platforms/pseries/eeh_pseries.c b/arch/powerpc/platforms/pseries/eeh_pseries.c
index c33360ec4f4f..19506f935737 100644
--- a/arch/powerpc/platforms/pseries/eeh_pseries.c
+++ b/arch/powerpc/platforms/pseries/eeh_pseries.c
@@ -129,27 +129,117 @@ static int pseries_eeh_init(void)
129 eeh_error_buf_size = RTAS_ERROR_LOG_MAX; 129 eeh_error_buf_size = RTAS_ERROR_LOG_MAX;
130 } 130 }
131 131
132 /* Set EEH probe mode */
133 eeh_probe_mode_set(EEH_PROBE_MODE_DEVTREE);
134
132 return 0; 135 return 0;
133} 136}
134 137
135/** 138/**
139 * pseries_eeh_of_probe - EEH probe on the given device
140 * @dn: OF node
141 * @flag: Unused
142 *
143 * When EEH module is installed during system boot, all PCI devices
144 * are checked one by one to see if it supports EEH. The function
145 * is introduced for the purpose.
146 */
147static void *pseries_eeh_of_probe(struct device_node *dn, void *flag)
148{
149 struct eeh_dev *edev;
150 struct eeh_pe pe;
151 const u32 *class_code, *vendor_id, *device_id;
152 const u32 *regs;
153 int enable = 0;
154 int ret;
155
156 /* Retrieve OF node and eeh device */
157 edev = of_node_to_eeh_dev(dn);
158 if (!of_device_is_available(dn))
159 return NULL;
160
161 /* Retrieve class/vendor/device IDs */
162 class_code = of_get_property(dn, "class-code", NULL);
163 vendor_id = of_get_property(dn, "vendor-id", NULL);
164 device_id = of_get_property(dn, "device-id", NULL);
165
166 /* Skip for bad OF node or PCI-ISA bridge */
167 if (!class_code || !vendor_id || !device_id)
168 return NULL;
169 if (dn->type && !strcmp(dn->type, "isa"))
170 return NULL;
171
172 /* Update class code and mode of eeh device */
173 edev->class_code = *class_code;
174 edev->mode = 0;
175
176 /* Retrieve the device address */
177 regs = of_get_property(dn, "reg", NULL);
178 if (!regs) {
179 pr_warning("%s: OF node property %s::reg not found\n",
180 __func__, dn->full_name);
181 return NULL;
182 }
183
184 /* Initialize the fake PE */
185 memset(&pe, 0, sizeof(struct eeh_pe));
186 pe.phb = edev->phb;
187 pe.config_addr = regs[0];
188
189 /* Enable EEH on the device */
190 ret = eeh_ops->set_option(&pe, EEH_OPT_ENABLE);
191 if (!ret) {
192 edev->config_addr = regs[0];
193 /* Retrieve PE address */
194 edev->pe_config_addr = eeh_ops->get_pe_addr(&pe);
195 pe.addr = edev->pe_config_addr;
196
197 /* Some older systems (Power4) allow the ibm,set-eeh-option
198 * call to succeed even on nodes where EEH is not supported.
199 * Verify support explicitly.
200 */
201 ret = eeh_ops->get_state(&pe, NULL);
202 if (ret > 0 && ret != EEH_STATE_NOT_SUPPORT)
203 enable = 1;
204
205 if (enable) {
206 eeh_subsystem_enabled = 1;
207 eeh_add_to_parent_pe(edev);
208
209 pr_debug("%s: EEH enabled on %s PHB#%d-PE#%x, config addr#%x\n",
210 __func__, dn->full_name, pe.phb->global_number,
211 pe.addr, pe.config_addr);
212 } else if (dn->parent && of_node_to_eeh_dev(dn->parent) &&
213 (of_node_to_eeh_dev(dn->parent))->pe) {
214 /* This device doesn't support EEH, but it may have an
215 * EEH parent, in which case we mark it as supported.
216 */
217 edev->config_addr = of_node_to_eeh_dev(dn->parent)->config_addr;
218 edev->pe_config_addr = of_node_to_eeh_dev(dn->parent)->pe_config_addr;
219 eeh_add_to_parent_pe(edev);
220 }
221 }
222
223 /* Save memory bars */
224 eeh_save_bars(edev);
225
226 return NULL;
227}
228
229/**
136 * pseries_eeh_set_option - Initialize EEH or MMIO/DMA reenable 230 * pseries_eeh_set_option - Initialize EEH or MMIO/DMA reenable
137 * @dn: device node 231 * @pe: EEH PE
138 * @option: operation to be issued 232 * @option: operation to be issued
139 * 233 *
140 * The function is used to control the EEH functionality globally. 234 * The function is used to control the EEH functionality globally.
141 * Currently, following options are support according to PAPR: 235 * Currently, following options are support according to PAPR:
142 * Enable EEH, Disable EEH, Enable MMIO and Enable DMA 236 * Enable EEH, Disable EEH, Enable MMIO and Enable DMA
143 */ 237 */
144static int pseries_eeh_set_option(struct device_node *dn, int option) 238static int pseries_eeh_set_option(struct eeh_pe *pe, int option)
145{ 239{
146 int ret = 0; 240 int ret = 0;
147 struct eeh_dev *edev;
148 const u32 *reg;
149 int config_addr; 241 int config_addr;
150 242
151 edev = of_node_to_eeh_dev(dn);
152
153 /* 243 /*
154 * When we're enabling or disabling EEH functioality on 244 * When we're enabling or disabling EEH functioality on
155 * the particular PE, the PE config address is possibly 245 * the particular PE, the PE config address is possibly
@@ -159,15 +249,11 @@ static int pseries_eeh_set_option(struct device_node *dn, int option)
159 switch (option) { 249 switch (option) {
160 case EEH_OPT_DISABLE: 250 case EEH_OPT_DISABLE:
161 case EEH_OPT_ENABLE: 251 case EEH_OPT_ENABLE:
162 reg = of_get_property(dn, "reg", NULL);
163 config_addr = reg[0];
164 break;
165
166 case EEH_OPT_THAW_MMIO: 252 case EEH_OPT_THAW_MMIO:
167 case EEH_OPT_THAW_DMA: 253 case EEH_OPT_THAW_DMA:
168 config_addr = edev->config_addr; 254 config_addr = pe->config_addr;
169 if (edev->pe_config_addr) 255 if (pe->addr)
170 config_addr = edev->pe_config_addr; 256 config_addr = pe->addr;
171 break; 257 break;
172 258
173 default: 259 default:
@@ -177,15 +263,15 @@ static int pseries_eeh_set_option(struct device_node *dn, int option)
177 } 263 }
178 264
179 ret = rtas_call(ibm_set_eeh_option, 4, 1, NULL, 265 ret = rtas_call(ibm_set_eeh_option, 4, 1, NULL,
180 config_addr, BUID_HI(edev->phb->buid), 266 config_addr, BUID_HI(pe->phb->buid),
181 BUID_LO(edev->phb->buid), option); 267 BUID_LO(pe->phb->buid), option);
182 268
183 return ret; 269 return ret;
184} 270}
185 271
186/** 272/**
187 * pseries_eeh_get_pe_addr - Retrieve PE address 273 * pseries_eeh_get_pe_addr - Retrieve PE address
188 * @dn: device node 274 * @pe: EEH PE
189 * 275 *
190 * Retrieve the assocated PE address. Actually, there're 2 RTAS 276 * Retrieve the assocated PE address. Actually, there're 2 RTAS
191 * function calls dedicated for the purpose. We need implement 277 * function calls dedicated for the purpose. We need implement
@@ -196,14 +282,11 @@ static int pseries_eeh_set_option(struct device_node *dn, int option)
196 * It's notable that zero'ed return value means invalid PE config 282 * It's notable that zero'ed return value means invalid PE config
197 * address. 283 * address.
198 */ 284 */
199static int pseries_eeh_get_pe_addr(struct device_node *dn) 285static int pseries_eeh_get_pe_addr(struct eeh_pe *pe)
200{ 286{
201 struct eeh_dev *edev;
202 int ret = 0; 287 int ret = 0;
203 int rets[3]; 288 int rets[3];
204 289
205 edev = of_node_to_eeh_dev(dn);
206
207 if (ibm_get_config_addr_info2 != RTAS_UNKNOWN_SERVICE) { 290 if (ibm_get_config_addr_info2 != RTAS_UNKNOWN_SERVICE) {
208 /* 291 /*
209 * First of all, we need to make sure there has one PE 292 * First of all, we need to make sure there has one PE
@@ -211,18 +294,18 @@ static int pseries_eeh_get_pe_addr(struct device_node *dn)
211 * meaningless. 294 * meaningless.
212 */ 295 */
213 ret = rtas_call(ibm_get_config_addr_info2, 4, 2, rets, 296 ret = rtas_call(ibm_get_config_addr_info2, 4, 2, rets,
214 edev->config_addr, BUID_HI(edev->phb->buid), 297 pe->config_addr, BUID_HI(pe->phb->buid),
215 BUID_LO(edev->phb->buid), 1); 298 BUID_LO(pe->phb->buid), 1);
216 if (ret || (rets[0] == 0)) 299 if (ret || (rets[0] == 0))
217 return 0; 300 return 0;
218 301
219 /* Retrieve the associated PE config address */ 302 /* Retrieve the associated PE config address */
220 ret = rtas_call(ibm_get_config_addr_info2, 4, 2, rets, 303 ret = rtas_call(ibm_get_config_addr_info2, 4, 2, rets,
221 edev->config_addr, BUID_HI(edev->phb->buid), 304 pe->config_addr, BUID_HI(pe->phb->buid),
222 BUID_LO(edev->phb->buid), 0); 305 BUID_LO(pe->phb->buid), 0);
223 if (ret) { 306 if (ret) {
224 pr_warning("%s: Failed to get PE address for %s\n", 307 pr_warning("%s: Failed to get address for PHB#%d-PE#%x\n",
225 __func__, dn->full_name); 308 __func__, pe->phb->global_number, pe->config_addr);
226 return 0; 309 return 0;
227 } 310 }
228 311
@@ -231,11 +314,11 @@ static int pseries_eeh_get_pe_addr(struct device_node *dn)
231 314
232 if (ibm_get_config_addr_info != RTAS_UNKNOWN_SERVICE) { 315 if (ibm_get_config_addr_info != RTAS_UNKNOWN_SERVICE) {
233 ret = rtas_call(ibm_get_config_addr_info, 4, 2, rets, 316 ret = rtas_call(ibm_get_config_addr_info, 4, 2, rets,
234 edev->config_addr, BUID_HI(edev->phb->buid), 317 pe->config_addr, BUID_HI(pe->phb->buid),
235 BUID_LO(edev->phb->buid), 0); 318 BUID_LO(pe->phb->buid), 0);
236 if (ret) { 319 if (ret) {
237 pr_warning("%s: Failed to get PE address for %s\n", 320 pr_warning("%s: Failed to get address for PHB#%d-PE#%x\n",
238 __func__, dn->full_name); 321 __func__, pe->phb->global_number, pe->config_addr);
239 return 0; 322 return 0;
240 } 323 }
241 324
@@ -247,7 +330,7 @@ static int pseries_eeh_get_pe_addr(struct device_node *dn)
247 330
248/** 331/**
249 * pseries_eeh_get_state - Retrieve PE state 332 * pseries_eeh_get_state - Retrieve PE state
250 * @dn: PE associated device node 333 * @pe: EEH PE
251 * @state: return value 334 * @state: return value
252 * 335 *
253 * Retrieve the state of the specified PE. On RTAS compliant 336 * Retrieve the state of the specified PE. On RTAS compliant
@@ -258,30 +341,28 @@ static int pseries_eeh_get_pe_addr(struct device_node *dn)
258 * RTAS calls for the purpose, we need to try the new one and back 341 * RTAS calls for the purpose, we need to try the new one and back
259 * to the old one if the new one couldn't work properly. 342 * to the old one if the new one couldn't work properly.
260 */ 343 */
261static int pseries_eeh_get_state(struct device_node *dn, int *state) 344static int pseries_eeh_get_state(struct eeh_pe *pe, int *state)
262{ 345{
263 struct eeh_dev *edev;
264 int config_addr; 346 int config_addr;
265 int ret; 347 int ret;
266 int rets[4]; 348 int rets[4];
267 int result; 349 int result;
268 350
269 /* Figure out PE config address if possible */ 351 /* Figure out PE config address if possible */
270 edev = of_node_to_eeh_dev(dn); 352 config_addr = pe->config_addr;
271 config_addr = edev->config_addr; 353 if (pe->addr)
272 if (edev->pe_config_addr) 354 config_addr = pe->addr;
273 config_addr = edev->pe_config_addr;
274 355
275 if (ibm_read_slot_reset_state2 != RTAS_UNKNOWN_SERVICE) { 356 if (ibm_read_slot_reset_state2 != RTAS_UNKNOWN_SERVICE) {
276 ret = rtas_call(ibm_read_slot_reset_state2, 3, 4, rets, 357 ret = rtas_call(ibm_read_slot_reset_state2, 3, 4, rets,
277 config_addr, BUID_HI(edev->phb->buid), 358 config_addr, BUID_HI(pe->phb->buid),
278 BUID_LO(edev->phb->buid)); 359 BUID_LO(pe->phb->buid));
279 } else if (ibm_read_slot_reset_state != RTAS_UNKNOWN_SERVICE) { 360 } else if (ibm_read_slot_reset_state != RTAS_UNKNOWN_SERVICE) {
280 /* Fake PE unavailable info */ 361 /* Fake PE unavailable info */
281 rets[2] = 0; 362 rets[2] = 0;
282 ret = rtas_call(ibm_read_slot_reset_state, 3, 3, rets, 363 ret = rtas_call(ibm_read_slot_reset_state, 3, 3, rets,
283 config_addr, BUID_HI(edev->phb->buid), 364 config_addr, BUID_HI(pe->phb->buid),
284 BUID_LO(edev->phb->buid)); 365 BUID_LO(pe->phb->buid));
285 } else { 366 } else {
286 return EEH_STATE_NOT_SUPPORT; 367 return EEH_STATE_NOT_SUPPORT;
287 } 368 }
@@ -333,34 +414,32 @@ static int pseries_eeh_get_state(struct device_node *dn, int *state)
333 414
334/** 415/**
335 * pseries_eeh_reset - Reset the specified PE 416 * pseries_eeh_reset - Reset the specified PE
336 * @dn: PE associated device node 417 * @pe: EEH PE
337 * @option: reset option 418 * @option: reset option
338 * 419 *
339 * Reset the specified PE 420 * Reset the specified PE
340 */ 421 */
341static int pseries_eeh_reset(struct device_node *dn, int option) 422static int pseries_eeh_reset(struct eeh_pe *pe, int option)
342{ 423{
343 struct eeh_dev *edev;
344 int config_addr; 424 int config_addr;
345 int ret; 425 int ret;
346 426
347 /* Figure out PE address */ 427 /* Figure out PE address */
348 edev = of_node_to_eeh_dev(dn); 428 config_addr = pe->config_addr;
349 config_addr = edev->config_addr; 429 if (pe->addr)
350 if (edev->pe_config_addr) 430 config_addr = pe->addr;
351 config_addr = edev->pe_config_addr;
352 431
353 /* Reset PE through RTAS call */ 432 /* Reset PE through RTAS call */
354 ret = rtas_call(ibm_set_slot_reset, 4, 1, NULL, 433 ret = rtas_call(ibm_set_slot_reset, 4, 1, NULL,
355 config_addr, BUID_HI(edev->phb->buid), 434 config_addr, BUID_HI(pe->phb->buid),
356 BUID_LO(edev->phb->buid), option); 435 BUID_LO(pe->phb->buid), option);
357 436
358 /* If fundamental-reset not supported, try hot-reset */ 437 /* If fundamental-reset not supported, try hot-reset */
359 if (option == EEH_RESET_FUNDAMENTAL && 438 if (option == EEH_RESET_FUNDAMENTAL &&
360 ret == -8) { 439 ret == -8) {
361 ret = rtas_call(ibm_set_slot_reset, 4, 1, NULL, 440 ret = rtas_call(ibm_set_slot_reset, 4, 1, NULL,
362 config_addr, BUID_HI(edev->phb->buid), 441 config_addr, BUID_HI(pe->phb->buid),
363 BUID_LO(edev->phb->buid), EEH_RESET_HOT); 442 BUID_LO(pe->phb->buid), EEH_RESET_HOT);
364 } 443 }
365 444
366 return ret; 445 return ret;
@@ -368,13 +447,13 @@ static int pseries_eeh_reset(struct device_node *dn, int option)
368 447
369/** 448/**
370 * pseries_eeh_wait_state - Wait for PE state 449 * pseries_eeh_wait_state - Wait for PE state
371 * @dn: PE associated device node 450 * @pe: EEH PE
372 * @max_wait: maximal period in microsecond 451 * @max_wait: maximal period in microsecond
373 * 452 *
374 * Wait for the state of associated PE. It might take some time 453 * Wait for the state of associated PE. It might take some time
375 * to retrieve the PE's state. 454 * to retrieve the PE's state.
376 */ 455 */
377static int pseries_eeh_wait_state(struct device_node *dn, int max_wait) 456static int pseries_eeh_wait_state(struct eeh_pe *pe, int max_wait)
378{ 457{
379 int ret; 458 int ret;
380 int mwait; 459 int mwait;
@@ -391,7 +470,7 @@ static int pseries_eeh_wait_state(struct device_node *dn, int max_wait)
391#define EEH_STATE_MAX_WAIT_TIME (300 * 1000) 470#define EEH_STATE_MAX_WAIT_TIME (300 * 1000)
392 471
393 while (1) { 472 while (1) {
394 ret = pseries_eeh_get_state(dn, &mwait); 473 ret = pseries_eeh_get_state(pe, &mwait);
395 474
396 /* 475 /*
397 * If the PE's state is temporarily unavailable, 476 * If the PE's state is temporarily unavailable,
@@ -426,7 +505,7 @@ static int pseries_eeh_wait_state(struct device_node *dn, int max_wait)
426 505
427/** 506/**
428 * pseries_eeh_get_log - Retrieve error log 507 * pseries_eeh_get_log - Retrieve error log
429 * @dn: device node 508 * @pe: EEH PE
430 * @severity: temporary or permanent error log 509 * @severity: temporary or permanent error log
431 * @drv_log: driver log to be combined with retrieved error log 510 * @drv_log: driver log to be combined with retrieved error log
432 * @len: length of driver log 511 * @len: length of driver log
@@ -435,24 +514,22 @@ static int pseries_eeh_wait_state(struct device_node *dn, int max_wait)
435 * Actually, the error will be retrieved through the dedicated 514 * Actually, the error will be retrieved through the dedicated
436 * RTAS call. 515 * RTAS call.
437 */ 516 */
438static int pseries_eeh_get_log(struct device_node *dn, int severity, char *drv_log, unsigned long len) 517static int pseries_eeh_get_log(struct eeh_pe *pe, int severity, char *drv_log, unsigned long len)
439{ 518{
440 struct eeh_dev *edev;
441 int config_addr; 519 int config_addr;
442 unsigned long flags; 520 unsigned long flags;
443 int ret; 521 int ret;
444 522
445 edev = of_node_to_eeh_dev(dn);
446 spin_lock_irqsave(&slot_errbuf_lock, flags); 523 spin_lock_irqsave(&slot_errbuf_lock, flags);
447 memset(slot_errbuf, 0, eeh_error_buf_size); 524 memset(slot_errbuf, 0, eeh_error_buf_size);
448 525
449 /* Figure out the PE address */ 526 /* Figure out the PE address */
450 config_addr = edev->config_addr; 527 config_addr = pe->config_addr;
451 if (edev->pe_config_addr) 528 if (pe->addr)
452 config_addr = edev->pe_config_addr; 529 config_addr = pe->addr;
453 530
454 ret = rtas_call(ibm_slot_error_detail, 8, 1, NULL, config_addr, 531 ret = rtas_call(ibm_slot_error_detail, 8, 1, NULL, config_addr,
455 BUID_HI(edev->phb->buid), BUID_LO(edev->phb->buid), 532 BUID_HI(pe->phb->buid), BUID_LO(pe->phb->buid),
456 virt_to_phys(drv_log), len, 533 virt_to_phys(drv_log), len,
457 virt_to_phys(slot_errbuf), eeh_error_buf_size, 534 virt_to_phys(slot_errbuf), eeh_error_buf_size,
458 severity); 535 severity);
@@ -465,40 +542,38 @@ static int pseries_eeh_get_log(struct device_node *dn, int severity, char *drv_l
465 542
466/** 543/**
467 * pseries_eeh_configure_bridge - Configure PCI bridges in the indicated PE 544 * pseries_eeh_configure_bridge - Configure PCI bridges in the indicated PE
468 * @dn: PE associated device node 545 * @pe: EEH PE
469 * 546 *
470 * The function will be called to reconfigure the bridges included 547 * The function will be called to reconfigure the bridges included
471 * in the specified PE so that the mulfunctional PE would be recovered 548 * in the specified PE so that the mulfunctional PE would be recovered
472 * again. 549 * again.
473 */ 550 */
474static int pseries_eeh_configure_bridge(struct device_node *dn) 551static int pseries_eeh_configure_bridge(struct eeh_pe *pe)
475{ 552{
476 struct eeh_dev *edev;
477 int config_addr; 553 int config_addr;
478 int ret; 554 int ret;
479 555
480 /* Figure out the PE address */ 556 /* Figure out the PE address */
481 edev = of_node_to_eeh_dev(dn); 557 config_addr = pe->config_addr;
482 config_addr = edev->config_addr; 558 if (pe->addr)
483 if (edev->pe_config_addr) 559 config_addr = pe->addr;
484 config_addr = edev->pe_config_addr;
485 560
486 /* Use new configure-pe function, if supported */ 561 /* Use new configure-pe function, if supported */
487 if (ibm_configure_pe != RTAS_UNKNOWN_SERVICE) { 562 if (ibm_configure_pe != RTAS_UNKNOWN_SERVICE) {
488 ret = rtas_call(ibm_configure_pe, 3, 1, NULL, 563 ret = rtas_call(ibm_configure_pe, 3, 1, NULL,
489 config_addr, BUID_HI(edev->phb->buid), 564 config_addr, BUID_HI(pe->phb->buid),
490 BUID_LO(edev->phb->buid)); 565 BUID_LO(pe->phb->buid));
491 } else if (ibm_configure_bridge != RTAS_UNKNOWN_SERVICE) { 566 } else if (ibm_configure_bridge != RTAS_UNKNOWN_SERVICE) {
492 ret = rtas_call(ibm_configure_bridge, 3, 1, NULL, 567 ret = rtas_call(ibm_configure_bridge, 3, 1, NULL,
493 config_addr, BUID_HI(edev->phb->buid), 568 config_addr, BUID_HI(pe->phb->buid),
494 BUID_LO(edev->phb->buid)); 569 BUID_LO(pe->phb->buid));
495 } else { 570 } else {
496 return -EFAULT; 571 return -EFAULT;
497 } 572 }
498 573
499 if (ret) 574 if (ret)
500 pr_warning("%s: Unable to configure bridge %d for %s\n", 575 pr_warning("%s: Unable to configure bridge PHB#%d-PE#%x (%d)\n",
501 __func__, ret, dn->full_name); 576 __func__, pe->phb->global_number, pe->addr, ret);
502 577
503 return ret; 578 return ret;
504} 579}
@@ -542,6 +617,8 @@ static int pseries_eeh_write_config(struct device_node *dn, int where, int size,
542static struct eeh_ops pseries_eeh_ops = { 617static struct eeh_ops pseries_eeh_ops = {
543 .name = "pseries", 618 .name = "pseries",
544 .init = pseries_eeh_init, 619 .init = pseries_eeh_init,
620 .of_probe = pseries_eeh_of_probe,
621 .dev_probe = NULL,
545 .set_option = pseries_eeh_set_option, 622 .set_option = pseries_eeh_set_option,
546 .get_pe_addr = pseries_eeh_get_pe_addr, 623 .get_pe_addr = pseries_eeh_get_pe_addr,
547 .get_state = pseries_eeh_get_state, 624 .get_state = pseries_eeh_get_state,
@@ -559,7 +636,21 @@ static struct eeh_ops pseries_eeh_ops = {
559 * EEH initialization on pseries platform. This function should be 636 * EEH initialization on pseries platform. This function should be
560 * called before any EEH related functions. 637 * called before any EEH related functions.
561 */ 638 */
562int __init eeh_pseries_init(void) 639static int __init eeh_pseries_init(void)
563{ 640{
564 return eeh_ops_register(&pseries_eeh_ops); 641 int ret = -EINVAL;
642
643 if (!machine_is(pseries))
644 return ret;
645
646 ret = eeh_ops_register(&pseries_eeh_ops);
647 if (!ret)
648 pr_info("EEH: pSeries platform initialized\n");
649 else
650 pr_info("EEH: pSeries platform initialization failure (%d)\n",
651 ret);
652
653 return ret;
565} 654}
655
656early_initcall(eeh_pseries_init);
diff --git a/arch/powerpc/platforms/pseries/eeh_sysfs.c b/arch/powerpc/platforms/pseries/eeh_sysfs.c
index 243b3510d70f..d37708360f2e 100644
--- a/arch/powerpc/platforms/pseries/eeh_sysfs.c
+++ b/arch/powerpc/platforms/pseries/eeh_sysfs.c
@@ -53,9 +53,6 @@ static DEVICE_ATTR(_name, S_IRUGO, eeh_show_##_name, NULL);
53EEH_SHOW_ATTR(eeh_mode, mode, "0x%x"); 53EEH_SHOW_ATTR(eeh_mode, mode, "0x%x");
54EEH_SHOW_ATTR(eeh_config_addr, config_addr, "0x%x"); 54EEH_SHOW_ATTR(eeh_config_addr, config_addr, "0x%x");
55EEH_SHOW_ATTR(eeh_pe_config_addr, pe_config_addr, "0x%x"); 55EEH_SHOW_ATTR(eeh_pe_config_addr, pe_config_addr, "0x%x");
56EEH_SHOW_ATTR(eeh_check_count, check_count, "%d" );
57EEH_SHOW_ATTR(eeh_freeze_count, freeze_count, "%d" );
58EEH_SHOW_ATTR(eeh_false_positives, false_positives, "%d" );
59 56
60void eeh_sysfs_add_device(struct pci_dev *pdev) 57void eeh_sysfs_add_device(struct pci_dev *pdev)
61{ 58{
@@ -64,9 +61,6 @@ void eeh_sysfs_add_device(struct pci_dev *pdev)
64 rc += device_create_file(&pdev->dev, &dev_attr_eeh_mode); 61 rc += device_create_file(&pdev->dev, &dev_attr_eeh_mode);
65 rc += device_create_file(&pdev->dev, &dev_attr_eeh_config_addr); 62 rc += device_create_file(&pdev->dev, &dev_attr_eeh_config_addr);
66 rc += device_create_file(&pdev->dev, &dev_attr_eeh_pe_config_addr); 63 rc += device_create_file(&pdev->dev, &dev_attr_eeh_pe_config_addr);
67 rc += device_create_file(&pdev->dev, &dev_attr_eeh_check_count);
68 rc += device_create_file(&pdev->dev, &dev_attr_eeh_false_positives);
69 rc += device_create_file(&pdev->dev, &dev_attr_eeh_freeze_count);
70 64
71 if (rc) 65 if (rc)
72 printk(KERN_WARNING "EEH: Unable to create sysfs entries\n"); 66 printk(KERN_WARNING "EEH: Unable to create sysfs entries\n");
@@ -77,8 +71,5 @@ void eeh_sysfs_remove_device(struct pci_dev *pdev)
77 device_remove_file(&pdev->dev, &dev_attr_eeh_mode); 71 device_remove_file(&pdev->dev, &dev_attr_eeh_mode);
78 device_remove_file(&pdev->dev, &dev_attr_eeh_config_addr); 72 device_remove_file(&pdev->dev, &dev_attr_eeh_config_addr);
79 device_remove_file(&pdev->dev, &dev_attr_eeh_pe_config_addr); 73 device_remove_file(&pdev->dev, &dev_attr_eeh_pe_config_addr);
80 device_remove_file(&pdev->dev, &dev_attr_eeh_check_count);
81 device_remove_file(&pdev->dev, &dev_attr_eeh_false_positives);
82 device_remove_file(&pdev->dev, &dev_attr_eeh_freeze_count);
83} 74}
84 75
diff --git a/arch/powerpc/platforms/pseries/iommu.c b/arch/powerpc/platforms/pseries/iommu.c
index bca220f2873c..6153eea27ce7 100644
--- a/arch/powerpc/platforms/pseries/iommu.c
+++ b/arch/powerpc/platforms/pseries/iommu.c
@@ -28,6 +28,7 @@
28#include <linux/types.h> 28#include <linux/types.h>
29#include <linux/slab.h> 29#include <linux/slab.h>
30#include <linux/mm.h> 30#include <linux/mm.h>
31#include <linux/memblock.h>
31#include <linux/spinlock.h> 32#include <linux/spinlock.h>
32#include <linux/sched.h> /* for show_stack */ 33#include <linux/sched.h> /* for show_stack */
33#include <linux/string.h> 34#include <linux/string.h>
@@ -41,7 +42,6 @@
41#include <asm/iommu.h> 42#include <asm/iommu.h>
42#include <asm/pci-bridge.h> 43#include <asm/pci-bridge.h>
43#include <asm/machdep.h> 44#include <asm/machdep.h>
44#include <asm/abs_addr.h>
45#include <asm/pSeries_reconfig.h> 45#include <asm/pSeries_reconfig.h>
46#include <asm/firmware.h> 46#include <asm/firmware.h>
47#include <asm/tce.h> 47#include <asm/tce.h>
@@ -99,7 +99,7 @@ static int tce_build_pSeries(struct iommu_table *tbl, long index,
99 99
100 while (npages--) { 100 while (npages--) {
101 /* can't move this out since we might cross MEMBLOCK boundary */ 101 /* can't move this out since we might cross MEMBLOCK boundary */
102 rpn = (virt_to_abs(uaddr)) >> TCE_SHIFT; 102 rpn = __pa(uaddr) >> TCE_SHIFT;
103 *tcep = proto_tce | (rpn & TCE_RPN_MASK) << TCE_RPN_SHIFT; 103 *tcep = proto_tce | (rpn & TCE_RPN_MASK) << TCE_RPN_SHIFT;
104 104
105 uaddr += TCE_PAGE_SIZE; 105 uaddr += TCE_PAGE_SIZE;
@@ -148,7 +148,7 @@ static int tce_build_pSeriesLP(struct iommu_table *tbl, long tcenum,
148 int ret = 0; 148 int ret = 0;
149 long tcenum_start = tcenum, npages_start = npages; 149 long tcenum_start = tcenum, npages_start = npages;
150 150
151 rpn = (virt_to_abs(uaddr)) >> TCE_SHIFT; 151 rpn = __pa(uaddr) >> TCE_SHIFT;
152 proto_tce = TCE_PCI_READ; 152 proto_tce = TCE_PCI_READ;
153 if (direction != DMA_TO_DEVICE) 153 if (direction != DMA_TO_DEVICE)
154 proto_tce |= TCE_PCI_WRITE; 154 proto_tce |= TCE_PCI_WRITE;
@@ -217,7 +217,7 @@ static int tce_buildmulti_pSeriesLP(struct iommu_table *tbl, long tcenum,
217 __get_cpu_var(tce_page) = tcep; 217 __get_cpu_var(tce_page) = tcep;
218 } 218 }
219 219
220 rpn = (virt_to_abs(uaddr)) >> TCE_SHIFT; 220 rpn = __pa(uaddr) >> TCE_SHIFT;
221 proto_tce = TCE_PCI_READ; 221 proto_tce = TCE_PCI_READ;
222 if (direction != DMA_TO_DEVICE) 222 if (direction != DMA_TO_DEVICE)
223 proto_tce |= TCE_PCI_WRITE; 223 proto_tce |= TCE_PCI_WRITE;
@@ -237,7 +237,7 @@ static int tce_buildmulti_pSeriesLP(struct iommu_table *tbl, long tcenum,
237 237
238 rc = plpar_tce_put_indirect((u64)tbl->it_index, 238 rc = plpar_tce_put_indirect((u64)tbl->it_index,
239 (u64)tcenum << 12, 239 (u64)tcenum << 12,
240 (u64)virt_to_abs(tcep), 240 (u64)__pa(tcep),
241 limit); 241 limit);
242 242
243 npages -= limit; 243 npages -= limit;
@@ -441,7 +441,7 @@ static int tce_setrange_multi_pSeriesLP(unsigned long start_pfn,
441 441
442 rc = plpar_tce_put_indirect(liobn, 442 rc = plpar_tce_put_indirect(liobn,
443 dma_offset, 443 dma_offset,
444 (u64)virt_to_abs(tcep), 444 (u64)__pa(tcep),
445 limit); 445 limit);
446 446
447 num_tce -= limit; 447 num_tce -= limit;
diff --git a/arch/powerpc/platforms/pseries/lpar.c b/arch/powerpc/platforms/pseries/lpar.c
index 5f3ef876ded2..177055d0186b 100644
--- a/arch/powerpc/platforms/pseries/lpar.c
+++ b/arch/powerpc/platforms/pseries/lpar.c
@@ -31,7 +31,6 @@
31#include <asm/page.h> 31#include <asm/page.h>
32#include <asm/pgtable.h> 32#include <asm/pgtable.h>
33#include <asm/machdep.h> 33#include <asm/machdep.h>
34#include <asm/abs_addr.h>
35#include <asm/mmu_context.h> 34#include <asm/mmu_context.h>
36#include <asm/iommu.h> 35#include <asm/iommu.h>
37#include <asm/tlbflush.h> 36#include <asm/tlbflush.h>
diff --git a/arch/powerpc/platforms/pseries/msi.c b/arch/powerpc/platforms/pseries/msi.c
index 109fdb75578d..d19f4977c834 100644
--- a/arch/powerpc/platforms/pseries/msi.c
+++ b/arch/powerpc/platforms/pseries/msi.c
@@ -210,6 +210,7 @@ static struct device_node *find_pe_total_msi(struct pci_dev *dev, int *total)
210static struct device_node *find_pe_dn(struct pci_dev *dev, int *total) 210static struct device_node *find_pe_dn(struct pci_dev *dev, int *total)
211{ 211{
212 struct device_node *dn; 212 struct device_node *dn;
213 struct eeh_dev *edev;
213 214
214 /* Found our PE and assume 8 at that point. */ 215 /* Found our PE and assume 8 at that point. */
215 216
@@ -217,7 +218,10 @@ static struct device_node *find_pe_dn(struct pci_dev *dev, int *total)
217 if (!dn) 218 if (!dn)
218 return NULL; 219 return NULL;
219 220
220 dn = eeh_find_device_pe(dn); 221 /* Get the top level device in the PE */
222 edev = of_node_to_eeh_dev(dn);
223 edev = list_first_entry(&edev->pe->edevs, struct eeh_dev, list);
224 dn = eeh_dev_to_of_node(edev);
221 if (!dn) 225 if (!dn)
222 return NULL; 226 return NULL;
223 227
@@ -387,12 +391,13 @@ static int check_msix_entries(struct pci_dev *pdev)
387 return 0; 391 return 0;
388} 392}
389 393
390static int rtas_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type) 394static int rtas_setup_msi_irqs(struct pci_dev *pdev, int nvec_in, int type)
391{ 395{
392 struct pci_dn *pdn; 396 struct pci_dn *pdn;
393 int hwirq, virq, i, rc; 397 int hwirq, virq, i, rc;
394 struct msi_desc *entry; 398 struct msi_desc *entry;
395 struct msi_msg msg; 399 struct msi_msg msg;
400 int nvec = nvec_in;
396 401
397 pdn = get_pdn(pdev); 402 pdn = get_pdn(pdev);
398 if (!pdn) 403 if (!pdn)
@@ -402,10 +407,23 @@ static int rtas_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type)
402 return -EINVAL; 407 return -EINVAL;
403 408
404 /* 409 /*
410 * Firmware currently refuse any non power of two allocation
411 * so we round up if the quota will allow it.
412 */
413 if (type == PCI_CAP_ID_MSIX) {
414 int m = roundup_pow_of_two(nvec);
415 int quota = msi_quota_for_device(pdev, m);
416
417 if (quota >= m)
418 nvec = m;
419 }
420
421 /*
405 * Try the new more explicit firmware interface, if that fails fall 422 * Try the new more explicit firmware interface, if that fails fall
406 * back to the old interface. The old interface is known to never 423 * back to the old interface. The old interface is known to never
407 * return MSI-Xs. 424 * return MSI-Xs.
408 */ 425 */
426again:
409 if (type == PCI_CAP_ID_MSI) { 427 if (type == PCI_CAP_ID_MSI) {
410 rc = rtas_change_msi(pdn, RTAS_CHANGE_MSI_FN, nvec); 428 rc = rtas_change_msi(pdn, RTAS_CHANGE_MSI_FN, nvec);
411 429
@@ -417,6 +435,10 @@ static int rtas_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type)
417 rc = rtas_change_msi(pdn, RTAS_CHANGE_MSIX_FN, nvec); 435 rc = rtas_change_msi(pdn, RTAS_CHANGE_MSIX_FN, nvec);
418 436
419 if (rc != nvec) { 437 if (rc != nvec) {
438 if (nvec != nvec_in) {
439 nvec = nvec_in;
440 goto again;
441 }
420 pr_debug("rtas_msi: rtas_change_msi() failed\n"); 442 pr_debug("rtas_msi: rtas_change_msi() failed\n");
421 return rc; 443 return rc;
422 } 444 }
diff --git a/arch/powerpc/platforms/pseries/pci.c b/arch/powerpc/platforms/pseries/pci.c
index 2c6ded29f73d..56b864d777ee 100644
--- a/arch/powerpc/platforms/pseries/pci.c
+++ b/arch/powerpc/platforms/pseries/pci.c
@@ -73,7 +73,7 @@ void __init pSeries_final_fixup(void)
73{ 73{
74 pSeries_request_regions(); 74 pSeries_request_regions();
75 75
76 pci_addr_cache_build(); 76 eeh_addr_cache_build();
77} 77}
78 78
79/* 79/*
diff --git a/arch/powerpc/platforms/pseries/setup.c b/arch/powerpc/platforms/pseries/setup.c
index 51ecac920dd8..e3cb7ae61658 100644
--- a/arch/powerpc/platforms/pseries/setup.c
+++ b/arch/powerpc/platforms/pseries/setup.c
@@ -388,10 +388,8 @@ static void __init pSeries_setup_arch(void)
388 388
389 /* Find and initialize PCI host bridges */ 389 /* Find and initialize PCI host bridges */
390 init_pci_config_tokens(); 390 init_pci_config_tokens();
391 eeh_pseries_init();
392 find_and_init_phbs(); 391 find_and_init_phbs();
393 pSeries_reconfig_notifier_register(&pci_dn_reconfig_nb); 392 pSeries_reconfig_notifier_register(&pci_dn_reconfig_nb);
394 eeh_init();
395 393
396 pSeries_nvram_init(); 394 pSeries_nvram_init();
397 395
@@ -416,16 +414,20 @@ static int __init pSeries_init_panel(void)
416} 414}
417machine_arch_initcall(pseries, pSeries_init_panel); 415machine_arch_initcall(pseries, pSeries_init_panel);
418 416
419static int pseries_set_dabr(unsigned long dabr) 417static int pseries_set_dabr(unsigned long dabr, unsigned long dabrx)
420{ 418{
421 return plpar_hcall_norets(H_SET_DABR, dabr); 419 return plpar_hcall_norets(H_SET_DABR, dabr);
422} 420}
423 421
424static int pseries_set_xdabr(unsigned long dabr) 422static int pseries_set_xdabr(unsigned long dabr, unsigned long dabrx)
425{ 423{
426 /* We want to catch accesses from kernel and userspace */ 424 /* Have to set at least one bit in the DABRX according to PAPR */
427 return plpar_hcall_norets(H_SET_XDABR, dabr, 425 if (dabrx == 0 && dabr == 0)
428 H_DABRX_KERNEL | H_DABRX_USER); 426 dabrx = DABRX_USER;
427 /* PAPR says we can only set kernel and user bits */
428 dabrx &= DABRX_KERNEL | DABRX_USER;
429
430 return plpar_hcall_norets(H_SET_XDABR, dabr, dabrx);
429} 431}
430 432
431#define CMO_CHARACTERISTICS_TOKEN 44 433#define CMO_CHARACTERISTICS_TOKEN 44
@@ -529,10 +531,10 @@ static void __init pSeries_init_early(void)
529 if (firmware_has_feature(FW_FEATURE_LPAR)) 531 if (firmware_has_feature(FW_FEATURE_LPAR))
530 hvc_vio_init_early(); 532 hvc_vio_init_early();
531#endif 533#endif
532 if (firmware_has_feature(FW_FEATURE_DABR)) 534 if (firmware_has_feature(FW_FEATURE_XDABR))
533 ppc_md.set_dabr = pseries_set_dabr;
534 else if (firmware_has_feature(FW_FEATURE_XDABR))
535 ppc_md.set_dabr = pseries_set_xdabr; 535 ppc_md.set_dabr = pseries_set_xdabr;
536 else if (firmware_has_feature(FW_FEATURE_DABR))
537 ppc_md.set_dabr = pseries_set_dabr;
536 538
537 pSeries_cmo_feature_init(); 539 pSeries_cmo_feature_init();
538 iommu_init_early_pSeries(); 540 iommu_init_early_pSeries();
diff --git a/arch/powerpc/sysdev/dart_iommu.c b/arch/powerpc/sysdev/dart_iommu.c
index 4f2680f431b5..8ef63a01e345 100644
--- a/arch/powerpc/sysdev/dart_iommu.c
+++ b/arch/powerpc/sysdev/dart_iommu.c
@@ -43,7 +43,6 @@
43#include <asm/iommu.h> 43#include <asm/iommu.h>
44#include <asm/pci-bridge.h> 44#include <asm/pci-bridge.h>
45#include <asm/machdep.h> 45#include <asm/machdep.h>
46#include <asm/abs_addr.h>
47#include <asm/cacheflush.h> 46#include <asm/cacheflush.h>
48#include <asm/ppc-pci.h> 47#include <asm/ppc-pci.h>
49 48
@@ -167,7 +166,7 @@ static int dart_build(struct iommu_table *tbl, long index,
167 */ 166 */
168 l = npages; 167 l = npages;
169 while (l--) { 168 while (l--) {
170 rpn = virt_to_abs(uaddr) >> DART_PAGE_SHIFT; 169 rpn = __pa(uaddr) >> DART_PAGE_SHIFT;
171 170
172 *(dp++) = DARTMAP_VALID | (rpn & DARTMAP_RPNMASK); 171 *(dp++) = DARTMAP_VALID | (rpn & DARTMAP_RPNMASK);
173 172
@@ -244,7 +243,7 @@ static int __init dart_init(struct device_node *dart_node)
244 panic("DART: Cannot map registers!"); 243 panic("DART: Cannot map registers!");
245 244
246 /* Map in DART table */ 245 /* Map in DART table */
247 dart_vbase = ioremap(virt_to_abs(dart_tablebase), dart_tablesize); 246 dart_vbase = ioremap(__pa(dart_tablebase), dart_tablesize);
248 247
249 /* Fill initial table */ 248 /* Fill initial table */
250 for (i = 0; i < dart_tablesize/4; i++) 249 for (i = 0; i < dart_tablesize/4; i++)
@@ -463,7 +462,7 @@ void __init alloc_dart_table(void)
463 * will blow up an entire large page anyway in the kernel mapping 462 * will blow up an entire large page anyway in the kernel mapping
464 */ 463 */
465 dart_tablebase = (unsigned long) 464 dart_tablebase = (unsigned long)
466 abs_to_virt(memblock_alloc_base(1UL<<24, 1UL<<24, 0x80000000L)); 465 __va(memblock_alloc_base(1UL<<24, 1UL<<24, 0x80000000L));
467 466
468 printk(KERN_INFO "DART table allocated at: %lx\n", dart_tablebase); 467 printk(KERN_INFO "DART table allocated at: %lx\n", dart_tablebase);
469} 468}
diff --git a/arch/powerpc/xmon/xmon.c b/arch/powerpc/xmon/xmon.c
index 9b49c65ee7a4..987f441525cb 100644
--- a/arch/powerpc/xmon/xmon.c
+++ b/arch/powerpc/xmon/xmon.c
@@ -740,7 +740,7 @@ static void insert_bpts(void)
740static void insert_cpu_bpts(void) 740static void insert_cpu_bpts(void)
741{ 741{
742 if (dabr.enabled) 742 if (dabr.enabled)
743 set_dabr(dabr.address | (dabr.enabled & 7)); 743 set_dabr(dabr.address | (dabr.enabled & 7), DABRX_ALL);
744 if (iabr && cpu_has_feature(CPU_FTR_IABR)) 744 if (iabr && cpu_has_feature(CPU_FTR_IABR))
745 mtspr(SPRN_IABR, iabr->address 745 mtspr(SPRN_IABR, iabr->address
746 | (iabr->enabled & (BP_IABR|BP_IABR_TE))); 746 | (iabr->enabled & (BP_IABR|BP_IABR_TE)));
@@ -768,7 +768,7 @@ static void remove_bpts(void)
768 768
769static void remove_cpu_bpts(void) 769static void remove_cpu_bpts(void)
770{ 770{
771 set_dabr(0); 771 set_dabr(0, 0);
772 if (cpu_has_feature(CPU_FTR_IABR)) 772 if (cpu_has_feature(CPU_FTR_IABR))
773 mtspr(SPRN_IABR, 0); 773 mtspr(SPRN_IABR, 0);
774} 774}
generated by cgit v1.2.2 (git 2.25.0) at 2025-02-05 20:27:57 -0500