aboutsummaryrefslogtreecommitdiffstats
path: root/arch
diff options
context:
space:
mode:
Diffstat (limited to 'arch')
-rw-r--r--arch/alpha/kernel/osf_sys.c36
-rw-r--r--arch/alpha/kernel/sys_titan.c1
-rw-r--r--arch/ia64/include/asm/xen/hypercall.h2
-rw-r--r--arch/ia64/xen/suspend.c9
-rw-r--r--arch/mips/Kconfig4
-rw-r--r--arch/mips/alchemy/mtx-1/board_setup.c4
-rw-r--r--arch/mips/alchemy/mtx-1/platform.c9
-rw-r--r--arch/mips/alchemy/xxs1500/board_setup.c4
-rw-r--r--arch/mips/include/asm/perf_event.h12
-rw-r--r--arch/mips/kernel/ftrace.c179
-rw-r--r--arch/mips/kernel/perf_event.c345
-rw-r--r--arch/mips/kernel/perf_event_mipsxx.c4
-rw-r--r--arch/mips/kernel/signal.c2
-rw-r--r--arch/mips/kernel/signal32.c2
-rw-r--r--arch/mips/kernel/smp.c31
-rw-r--r--arch/mips/kernel/syscall.c5
-rw-r--r--arch/mips/kernel/vpe.c4
-rw-r--r--arch/mips/loongson/Kconfig5
-rw-r--r--arch/mips/loongson/common/cmdline.c5
-rw-r--r--arch/mips/loongson/common/machtype.c3
-rw-r--r--arch/mips/math-emu/ieee754int.h4
-rw-r--r--arch/mips/mm/init.c2
-rw-r--r--arch/mips/mm/tlbex.c2
-rw-r--r--arch/mips/pci/ops-pmcmsp.c4
-rw-r--r--arch/mips/pmc-sierra/Kconfig4
-rw-r--r--arch/mips/pmc-sierra/msp71xx/msp_time.c2
-rw-r--r--arch/mn10300/include/asm/atomic.h2
-rw-r--r--arch/mn10300/include/asm/uaccess.h5
-rw-r--r--arch/mn10300/mm/cache-inv-icache.c4
-rw-r--r--arch/parisc/hpux/sys_hpux.c65
-rw-r--r--arch/powerpc/include/asm/lppaca.h16
-rw-r--r--arch/powerpc/kernel/paca.c14
-rw-r--r--arch/powerpc/mm/numa.c3
-rw-r--r--arch/powerpc/platforms/cell/spufs/syscalls.c2
-rw-r--r--arch/powerpc/platforms/iseries/dt.c6
-rw-r--r--arch/powerpc/platforms/iseries/setup.c1
-rw-r--r--arch/um/drivers/mconsole_kern.c21
-rw-r--r--arch/x86/boot/compressed/mkpiggy.c7
-rw-r--r--arch/x86/ia32/ia32entry.S2
-rw-r--r--arch/x86/include/asm/ce4100.h6
-rw-r--r--arch/x86/include/asm/unistd_32.h4
-rw-r--r--arch/x86/include/asm/unistd_64.h4
-rw-r--r--arch/x86/include/asm/uv/uv_bau.h2
-rw-r--r--arch/x86/include/asm/xen/hypercall.h15
-rw-r--r--arch/x86/include/asm/xen/page.h47
-rw-r--r--arch/x86/include/asm/xen/pci.h8
-rw-r--r--arch/x86/kernel/check.c8
-rw-r--r--arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c2
-rw-r--r--arch/x86/kernel/syscall_table_32.S2
-rw-r--r--arch/x86/mm/fault.c14
-rw-r--r--arch/x86/mm/init_64.c6
-rw-r--r--arch/x86/mm/numa_64.c6
-rw-r--r--arch/x86/mm/pageattr.c18
-rw-r--r--arch/x86/mm/pgtable.c11
-rw-r--r--arch/x86/pci/ce4100.c7
-rw-r--r--arch/x86/pci/xen.c159
-rw-r--r--arch/x86/platform/ce4100/ce4100.c2
-rw-r--r--arch/x86/platform/uv/tlb_uv.c4
-rw-r--r--arch/x86/xen/Kconfig8
-rw-r--r--arch/x86/xen/enlighten.c8
-rw-r--r--arch/x86/xen/mmu.c82
-rw-r--r--arch/x86/xen/p2m.c330
-rw-r--r--arch/x86/xen/setup.c68
-rw-r--r--arch/x86/xen/smp.c38
-rw-r--r--arch/x86/xen/suspend.c8
-rw-r--r--arch/x86/xen/time.c4
-rw-r--r--arch/x86/xen/xen-ops.h2
67 files changed, 1120 insertions, 585 deletions
diff --git a/arch/alpha/kernel/osf_sys.c b/arch/alpha/kernel/osf_sys.c
index fe698b5045e9..376f22130791 100644
--- a/arch/alpha/kernel/osf_sys.c
+++ b/arch/alpha/kernel/osf_sys.c
@@ -230,44 +230,24 @@ linux_to_osf_statfs(struct kstatfs *linux_stat, struct osf_statfs __user *osf_st
230 return copy_to_user(osf_stat, &tmp_stat, bufsiz) ? -EFAULT : 0; 230 return copy_to_user(osf_stat, &tmp_stat, bufsiz) ? -EFAULT : 0;
231} 231}
232 232
233static int 233SYSCALL_DEFINE3(osf_statfs, const char __user *, pathname,
234do_osf_statfs(struct path *path, struct osf_statfs __user *buffer, 234 struct osf_statfs __user *, buffer, unsigned long, bufsiz)
235 unsigned long bufsiz)
236{ 235{
237 struct kstatfs linux_stat; 236 struct kstatfs linux_stat;
238 int error = vfs_statfs(path, &linux_stat); 237 int error = user_statfs(pathname, &linux_stat);
239 if (!error) 238 if (!error)
240 error = linux_to_osf_statfs(&linux_stat, buffer, bufsiz); 239 error = linux_to_osf_statfs(&linux_stat, buffer, bufsiz);
241 return error; 240 return error;
242} 241}
243 242
244SYSCALL_DEFINE3(osf_statfs, const char __user *, pathname,
245 struct osf_statfs __user *, buffer, unsigned long, bufsiz)
246{
247 struct path path;
248 int retval;
249
250 retval = user_path(pathname, &path);
251 if (!retval) {
252 retval = do_osf_statfs(&path, buffer, bufsiz);
253 path_put(&path);
254 }
255 return retval;
256}
257
258SYSCALL_DEFINE3(osf_fstatfs, unsigned long, fd, 243SYSCALL_DEFINE3(osf_fstatfs, unsigned long, fd,
259 struct osf_statfs __user *, buffer, unsigned long, bufsiz) 244 struct osf_statfs __user *, buffer, unsigned long, bufsiz)
260{ 245{
261 struct file *file; 246 struct kstatfs linux_stat;
262 int retval; 247 int error = fd_statfs(fd, &linux_stat);
263 248 if (!error)
264 retval = -EBADF; 249 error = linux_to_osf_statfs(&linux_stat, buffer, bufsiz);
265 file = fget(fd); 250 return error;
266 if (file) {
267 retval = do_osf_statfs(&file->f_path, buffer, bufsiz);
268 fput(file);
269 }
270 return retval;
271} 251}
272 252
273/* 253/*
diff --git a/arch/alpha/kernel/sys_titan.c b/arch/alpha/kernel/sys_titan.c
index f6c108a3d673..8c13a0c77830 100644
--- a/arch/alpha/kernel/sys_titan.c
+++ b/arch/alpha/kernel/sys_titan.c
@@ -149,6 +149,7 @@ static int
149titan_set_irq_affinity(struct irq_data *d, const struct cpumask *affinity, 149titan_set_irq_affinity(struct irq_data *d, const struct cpumask *affinity,
150 bool force) 150 bool force)
151{ 151{
152 unsigned int irq = d->irq;
152 spin_lock(&titan_irq_lock); 153 spin_lock(&titan_irq_lock);
153 titan_cpu_set_irq_affinity(irq - 16, *affinity); 154 titan_cpu_set_irq_affinity(irq - 16, *affinity);
154 titan_update_irq_hw(titan_cached_irq_mask); 155 titan_update_irq_hw(titan_cached_irq_mask);
diff --git a/arch/ia64/include/asm/xen/hypercall.h b/arch/ia64/include/asm/xen/hypercall.h
index 96fc62366aa4..ed28bcd5bb85 100644
--- a/arch/ia64/include/asm/xen/hypercall.h
+++ b/arch/ia64/include/asm/xen/hypercall.h
@@ -107,7 +107,7 @@ extern unsigned long __hypercall(unsigned long a1, unsigned long a2,
107static inline int 107static inline int
108xencomm_arch_hypercall_sched_op(int cmd, struct xencomm_handle *arg) 108xencomm_arch_hypercall_sched_op(int cmd, struct xencomm_handle *arg)
109{ 109{
110 return _hypercall2(int, sched_op_new, cmd, arg); 110 return _hypercall2(int, sched_op, cmd, arg);
111} 111}
112 112
113static inline long 113static inline long
diff --git a/arch/ia64/xen/suspend.c b/arch/ia64/xen/suspend.c
index fd66b048c6fa..419c8620945a 100644
--- a/arch/ia64/xen/suspend.c
+++ b/arch/ia64/xen/suspend.c
@@ -37,19 +37,14 @@ xen_mm_unpin_all(void)
37 /* nothing */ 37 /* nothing */
38} 38}
39 39
40void xen_pre_device_suspend(void)
41{
42 /* nothing */
43}
44
45void 40void
46xen_pre_suspend() 41xen_arch_pre_suspend()
47{ 42{
48 /* nothing */ 43 /* nothing */
49} 44}
50 45
51void 46void
52xen_post_suspend(int suspend_cancelled) 47xen_arch_post_suspend(int suspend_cancelled)
53{ 48{
54 if (suspend_cancelled) 49 if (suspend_cancelled)
55 return; 50 return;
diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index f5ecc0566bc2..d88983516e26 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -4,6 +4,7 @@ config MIPS
4 select HAVE_GENERIC_DMA_COHERENT 4 select HAVE_GENERIC_DMA_COHERENT
5 select HAVE_IDE 5 select HAVE_IDE
6 select HAVE_OPROFILE 6 select HAVE_OPROFILE
7 select HAVE_IRQ_WORK
7 select HAVE_PERF_EVENTS 8 select HAVE_PERF_EVENTS
8 select PERF_USE_VMALLOC 9 select PERF_USE_VMALLOC
9 select HAVE_ARCH_KGDB 10 select HAVE_ARCH_KGDB
@@ -208,6 +209,7 @@ config MACH_JZ4740
208 select ARCH_REQUIRE_GPIOLIB 209 select ARCH_REQUIRE_GPIOLIB
209 select SYS_HAS_EARLY_PRINTK 210 select SYS_HAS_EARLY_PRINTK
210 select HAVE_PWM 211 select HAVE_PWM
212 select HAVE_CLK
211 213
212config LASAT 214config LASAT
213 bool "LASAT Networks platforms" 215 bool "LASAT Networks platforms"
@@ -333,6 +335,8 @@ config PNX8550_STB810
333config PMC_MSP 335config PMC_MSP
334 bool "PMC-Sierra MSP chipsets" 336 bool "PMC-Sierra MSP chipsets"
335 depends on EXPERIMENTAL 337 depends on EXPERIMENTAL
338 select CEVT_R4K
339 select CSRC_R4K
336 select DMA_NONCOHERENT 340 select DMA_NONCOHERENT
337 select SWAP_IO_SPACE 341 select SWAP_IO_SPACE
338 select NO_EXCEPT_FILL 342 select NO_EXCEPT_FILL
diff --git a/arch/mips/alchemy/mtx-1/board_setup.c b/arch/mips/alchemy/mtx-1/board_setup.c
index 6398fa95905c..40b84b991191 100644
--- a/arch/mips/alchemy/mtx-1/board_setup.c
+++ b/arch/mips/alchemy/mtx-1/board_setup.c
@@ -54,8 +54,8 @@ int mtx1_pci_idsel(unsigned int devsel, int assert);
54 54
55static void mtx1_reset(char *c) 55static void mtx1_reset(char *c)
56{ 56{
57 /* Hit BCSR.SYSTEM_CONTROL[SW_RST] */ 57 /* Jump to the reset vector */
58 au_writel(0x00000000, 0xAE00001C); 58 __asm__ __volatile__("jr\t%0"::"r"(0xbfc00000));
59} 59}
60 60
61static void mtx1_power_off(void) 61static void mtx1_power_off(void)
diff --git a/arch/mips/alchemy/mtx-1/platform.c b/arch/mips/alchemy/mtx-1/platform.c
index e30e42add697..956f946218c5 100644
--- a/arch/mips/alchemy/mtx-1/platform.c
+++ b/arch/mips/alchemy/mtx-1/platform.c
@@ -28,6 +28,8 @@
28#include <linux/mtd/physmap.h> 28#include <linux/mtd/physmap.h>
29#include <mtd/mtd-abi.h> 29#include <mtd/mtd-abi.h>
30 30
31#include <asm/mach-au1x00/au1xxx_eth.h>
32
31static struct gpio_keys_button mtx1_gpio_button[] = { 33static struct gpio_keys_button mtx1_gpio_button[] = {
32 { 34 {
33 .gpio = 207, 35 .gpio = 207,
@@ -140,10 +142,17 @@ static struct __initdata platform_device * mtx1_devs[] = {
140 &mtx1_mtd, 142 &mtx1_mtd,
141}; 143};
142 144
145static struct au1000_eth_platform_data mtx1_au1000_eth0_pdata = {
146 .phy_search_highest_addr = 1,
147 .phy1_search_mac0 = 1,
148};
149
143static int __init mtx1_register_devices(void) 150static int __init mtx1_register_devices(void)
144{ 151{
145 int rc; 152 int rc;
146 153
154 au1xxx_override_eth_cfg(0, &mtx1_au1000_eth0_pdata);
155
147 rc = gpio_request(mtx1_gpio_button[0].gpio, 156 rc = gpio_request(mtx1_gpio_button[0].gpio,
148 mtx1_gpio_button[0].desc); 157 mtx1_gpio_button[0].desc);
149 if (rc < 0) { 158 if (rc < 0) {
diff --git a/arch/mips/alchemy/xxs1500/board_setup.c b/arch/mips/alchemy/xxs1500/board_setup.c
index b43c918925d3..80c521e5290d 100644
--- a/arch/mips/alchemy/xxs1500/board_setup.c
+++ b/arch/mips/alchemy/xxs1500/board_setup.c
@@ -36,8 +36,8 @@
36 36
37static void xxs1500_reset(char *c) 37static void xxs1500_reset(char *c)
38{ 38{
39 /* Hit BCSR.SYSTEM_CONTROL[SW_RST] */ 39 /* Jump to the reset vector */
40 au_writel(0x00000000, 0xAE00001C); 40 __asm__ __volatile__("jr\t%0"::"r"(0xbfc00000));
41} 41}
42 42
43static void xxs1500_power_off(void) 43static void xxs1500_power_off(void)
diff --git a/arch/mips/include/asm/perf_event.h b/arch/mips/include/asm/perf_event.h
index e00007cf8162..d0c77496c728 100644
--- a/arch/mips/include/asm/perf_event.h
+++ b/arch/mips/include/asm/perf_event.h
@@ -11,15 +11,5 @@
11 11
12#ifndef __MIPS_PERF_EVENT_H__ 12#ifndef __MIPS_PERF_EVENT_H__
13#define __MIPS_PERF_EVENT_H__ 13#define __MIPS_PERF_EVENT_H__
14 14/* Leave it empty here. The file is required by linux/perf_event.h */
15/*
16 * MIPS performance counters do not raise NMI upon overflow, a regular
17 * interrupt will be signaled. Hence we can do the pending perf event
18 * work at the tail of the irq handler.
19 */
20static inline void
21set_perf_event_pending(void)
22{
23}
24
25#endif /* __MIPS_PERF_EVENT_H__ */ 15#endif /* __MIPS_PERF_EVENT_H__ */
diff --git a/arch/mips/kernel/ftrace.c b/arch/mips/kernel/ftrace.c
index 5a84a1f11231..94ca2b018af7 100644
--- a/arch/mips/kernel/ftrace.c
+++ b/arch/mips/kernel/ftrace.c
@@ -17,29 +17,13 @@
17#include <asm/cacheflush.h> 17#include <asm/cacheflush.h>
18#include <asm/uasm.h> 18#include <asm/uasm.h>
19 19
20/* 20#include <asm-generic/sections.h>
21 * If the Instruction Pointer is in module space (0xc0000000), return true;
22 * otherwise, it is in kernel space (0x80000000), return false.
23 *
24 * FIXME: This will not work when the kernel space and module space are the
25 * same. If they are the same, we need to modify scripts/recordmcount.pl,
26 * ftrace_make_nop/call() and the other related parts to ensure the
27 * enabling/disabling of the calling site to _mcount is right for both kernel
28 * and module.
29 */
30
31static inline int in_module(unsigned long ip)
32{
33 return ip & 0x40000000;
34}
35 21
36#ifdef CONFIG_DYNAMIC_FTRACE 22#ifdef CONFIG_DYNAMIC_FTRACE
37 23
38#define JAL 0x0c000000 /* jump & link: ip --> ra, jump to target */ 24#define JAL 0x0c000000 /* jump & link: ip --> ra, jump to target */
39#define ADDR_MASK 0x03ffffff /* op_code|addr : 31...26|25 ....0 */ 25#define ADDR_MASK 0x03ffffff /* op_code|addr : 31...26|25 ....0 */
40 26
41#define INSN_B_1F_4 0x10000004 /* b 1f; offset = 4 */
42#define INSN_B_1F_5 0x10000005 /* b 1f; offset = 5 */
43#define INSN_NOP 0x00000000 /* nop */ 27#define INSN_NOP 0x00000000 /* nop */
44#define INSN_JAL(addr) \ 28#define INSN_JAL(addr) \
45 ((unsigned int)(JAL | (((addr) >> 2) & ADDR_MASK))) 29 ((unsigned int)(JAL | (((addr) >> 2) & ADDR_MASK)))
@@ -69,6 +53,20 @@ static inline void ftrace_dyn_arch_init_insns(void)
69#endif 53#endif
70} 54}
71 55
56/*
57 * Check if the address is in kernel space
58 *
59 * Clone core_kernel_text() from kernel/extable.c, but doesn't call
60 * init_kernel_text() for Ftrace doesn't trace functions in init sections.
61 */
62static inline int in_kernel_space(unsigned long ip)
63{
64 if (ip >= (unsigned long)_stext &&
65 ip <= (unsigned long)_etext)
66 return 1;
67 return 0;
68}
69
72static int ftrace_modify_code(unsigned long ip, unsigned int new_code) 70static int ftrace_modify_code(unsigned long ip, unsigned int new_code)
73{ 71{
74 int faulted; 72 int faulted;
@@ -84,6 +82,42 @@ static int ftrace_modify_code(unsigned long ip, unsigned int new_code)
84 return 0; 82 return 0;
85} 83}
86 84
85/*
86 * The details about the calling site of mcount on MIPS
87 *
88 * 1. For kernel:
89 *
90 * move at, ra
91 * jal _mcount --> nop
92 *
93 * 2. For modules:
94 *
95 * 2.1 For KBUILD_MCOUNT_RA_ADDRESS and CONFIG_32BIT
96 *
97 * lui v1, hi_16bit_of_mcount --> b 1f (0x10000005)
98 * addiu v1, v1, low_16bit_of_mcount
99 * move at, ra
100 * move $12, ra_address
101 * jalr v1
102 * sub sp, sp, 8
103 * 1: offset = 5 instructions
104 * 2.2 For the Other situations
105 *
106 * lui v1, hi_16bit_of_mcount --> b 1f (0x10000004)
107 * addiu v1, v1, low_16bit_of_mcount
108 * move at, ra
109 * jalr v1
110 * nop | move $12, ra_address | sub sp, sp, 8
111 * 1: offset = 4 instructions
112 */
113
114#if defined(KBUILD_MCOUNT_RA_ADDRESS) && defined(CONFIG_32BIT)
115#define MCOUNT_OFFSET_INSNS 5
116#else
117#define MCOUNT_OFFSET_INSNS 4
118#endif
119#define INSN_B_1F (0x10000000 | MCOUNT_OFFSET_INSNS)
120
87int ftrace_make_nop(struct module *mod, 121int ftrace_make_nop(struct module *mod,
88 struct dyn_ftrace *rec, unsigned long addr) 122 struct dyn_ftrace *rec, unsigned long addr)
89{ 123{
@@ -91,39 +125,11 @@ int ftrace_make_nop(struct module *mod,
91 unsigned long ip = rec->ip; 125 unsigned long ip = rec->ip;
92 126
93 /* 127 /*
94 * We have compiled module with -mlong-calls, but compiled the kernel 128 * If ip is in kernel space, no long call, otherwise, long call is
95 * without it, we need to cope with them respectively. 129 * needed.
96 */ 130 */
97 if (in_module(ip)) { 131 new = in_kernel_space(ip) ? INSN_NOP : INSN_B_1F;
98#if defined(KBUILD_MCOUNT_RA_ADDRESS) && defined(CONFIG_32BIT) 132
99 /*
100 * lui v1, hi_16bit_of_mcount --> b 1f (0x10000005)
101 * addiu v1, v1, low_16bit_of_mcount
102 * move at, ra
103 * move $12, ra_address
104 * jalr v1
105 * sub sp, sp, 8
106 * 1: offset = 5 instructions
107 */
108 new = INSN_B_1F_5;
109#else
110 /*
111 * lui v1, hi_16bit_of_mcount --> b 1f (0x10000004)
112 * addiu v1, v1, low_16bit_of_mcount
113 * move at, ra
114 * jalr v1
115 * nop | move $12, ra_address | sub sp, sp, 8
116 * 1: offset = 4 instructions
117 */
118 new = INSN_B_1F_4;
119#endif
120 } else {
121 /*
122 * move at, ra
123 * jal _mcount --> nop
124 */
125 new = INSN_NOP;
126 }
127 return ftrace_modify_code(ip, new); 133 return ftrace_modify_code(ip, new);
128} 134}
129 135
@@ -132,8 +138,8 @@ int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
132 unsigned int new; 138 unsigned int new;
133 unsigned long ip = rec->ip; 139 unsigned long ip = rec->ip;
134 140
135 /* ip, module: 0xc0000000, kernel: 0x80000000 */ 141 new = in_kernel_space(ip) ? insn_jal_ftrace_caller :
136 new = in_module(ip) ? insn_lui_v1_hi16_mcount : insn_jal_ftrace_caller; 142 insn_lui_v1_hi16_mcount;
137 143
138 return ftrace_modify_code(ip, new); 144 return ftrace_modify_code(ip, new);
139} 145}
@@ -190,29 +196,25 @@ int ftrace_disable_ftrace_graph_caller(void)
190#define S_R_SP (0xafb0 << 16) /* s{d,w} R, offset(sp) */ 196#define S_R_SP (0xafb0 << 16) /* s{d,w} R, offset(sp) */
191#define OFFSET_MASK 0xffff /* stack offset range: 0 ~ PT_SIZE */ 197#define OFFSET_MASK 0xffff /* stack offset range: 0 ~ PT_SIZE */
192 198
193unsigned long ftrace_get_parent_addr(unsigned long self_addr, 199unsigned long ftrace_get_parent_ra_addr(unsigned long self_ra, unsigned long
194 unsigned long parent, 200 old_parent_ra, unsigned long parent_ra_addr, unsigned long fp)
195 unsigned long parent_addr,
196 unsigned long fp)
197{ 201{
198 unsigned long sp, ip, ra; 202 unsigned long sp, ip, tmp;
199 unsigned int code; 203 unsigned int code;
200 int faulted; 204 int faulted;
201 205
202 /* 206 /*
203 * For module, move the ip from calling site of mcount to the 207 * For module, move the ip from the return address after the
204 * instruction "lui v1, hi_16bit_of_mcount"(offset is 20), but for 208 * instruction "lui v1, hi_16bit_of_mcount"(offset is 24), but for
205 * kernel, move to the instruction "move ra, at"(offset is 12) 209 * kernel, move after the instruction "move ra, at"(offset is 16)
206 */ 210 */
207 ip = self_addr - (in_module(self_addr) ? 20 : 12); 211 ip = self_ra - (in_kernel_space(self_ra) ? 16 : 24);
208 212
209 /* 213 /*
210 * search the text until finding the non-store instruction or "s{d,w} 214 * search the text until finding the non-store instruction or "s{d,w}
211 * ra, offset(sp)" instruction 215 * ra, offset(sp)" instruction
212 */ 216 */
213 do { 217 do {
214 ip -= 4;
215
216 /* get the code at "ip": code = *(unsigned int *)ip; */ 218 /* get the code at "ip": code = *(unsigned int *)ip; */
217 safe_load_code(code, ip, faulted); 219 safe_load_code(code, ip, faulted);
218 220
@@ -224,18 +226,20 @@ unsigned long ftrace_get_parent_addr(unsigned long self_addr,
224 * store the ra on the stack 226 * store the ra on the stack
225 */ 227 */
226 if ((code & S_R_SP) != S_R_SP) 228 if ((code & S_R_SP) != S_R_SP)
227 return parent_addr; 229 return parent_ra_addr;
228 230
229 } while (((code & S_RA_SP) != S_RA_SP)); 231 /* Move to the next instruction */
232 ip -= 4;
233 } while ((code & S_RA_SP) != S_RA_SP);
230 234
231 sp = fp + (code & OFFSET_MASK); 235 sp = fp + (code & OFFSET_MASK);
232 236
233 /* ra = *(unsigned long *)sp; */ 237 /* tmp = *(unsigned long *)sp; */
234 safe_load_stack(ra, sp, faulted); 238 safe_load_stack(tmp, sp, faulted);
235 if (unlikely(faulted)) 239 if (unlikely(faulted))
236 return 0; 240 return 0;
237 241
238 if (ra == parent) 242 if (tmp == old_parent_ra)
239 return sp; 243 return sp;
240 return 0; 244 return 0;
241} 245}
@@ -246,21 +250,21 @@ unsigned long ftrace_get_parent_addr(unsigned long self_addr,
246 * Hook the return address and push it in the stack of return addrs 250 * Hook the return address and push it in the stack of return addrs
247 * in current thread info. 251 * in current thread info.
248 */ 252 */
249void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr, 253void prepare_ftrace_return(unsigned long *parent_ra_addr, unsigned long self_ra,
250 unsigned long fp) 254 unsigned long fp)
251{ 255{
252 unsigned long old; 256 unsigned long old_parent_ra;
253 struct ftrace_graph_ent trace; 257 struct ftrace_graph_ent trace;
254 unsigned long return_hooker = (unsigned long) 258 unsigned long return_hooker = (unsigned long)
255 &return_to_handler; 259 &return_to_handler;
256 int faulted; 260 int faulted, insns;
257 261
258 if (unlikely(atomic_read(&current->tracing_graph_pause))) 262 if (unlikely(atomic_read(&current->tracing_graph_pause)))
259 return; 263 return;
260 264
261 /* 265 /*
262 * "parent" is the stack address saved the return address of the caller 266 * "parent_ra_addr" is the stack address saved the return address of
263 * of _mcount. 267 * the caller of _mcount.
264 * 268 *
265 * if the gcc < 4.5, a leaf function does not save the return address 269 * if the gcc < 4.5, a leaf function does not save the return address
266 * in the stack address, so, we "emulate" one in _mcount's stack space, 270 * in the stack address, so, we "emulate" one in _mcount's stack space,
@@ -275,37 +279,44 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr,
275 * do it in ftrace_graph_caller of mcount.S. 279 * do it in ftrace_graph_caller of mcount.S.
276 */ 280 */
277 281
278 /* old = *parent; */ 282 /* old_parent_ra = *parent_ra_addr; */
279 safe_load_stack(old, parent, faulted); 283 safe_load_stack(old_parent_ra, parent_ra_addr, faulted);
280 if (unlikely(faulted)) 284 if (unlikely(faulted))
281 goto out; 285 goto out;
282#ifndef KBUILD_MCOUNT_RA_ADDRESS 286#ifndef KBUILD_MCOUNT_RA_ADDRESS
283 parent = (unsigned long *)ftrace_get_parent_addr(self_addr, old, 287 parent_ra_addr = (unsigned long *)ftrace_get_parent_ra_addr(self_ra,
284 (unsigned long)parent, fp); 288 old_parent_ra, (unsigned long)parent_ra_addr, fp);
285 /* 289 /*
286 * If fails when getting the stack address of the non-leaf function's 290 * If fails when getting the stack address of the non-leaf function's
287 * ra, stop function graph tracer and return 291 * ra, stop function graph tracer and return
288 */ 292 */
289 if (parent == 0) 293 if (parent_ra_addr == 0)
290 goto out; 294 goto out;
291#endif 295#endif
292 /* *parent = return_hooker; */ 296 /* *parent_ra_addr = return_hooker; */
293 safe_store_stack(return_hooker, parent, faulted); 297 safe_store_stack(return_hooker, parent_ra_addr, faulted);
294 if (unlikely(faulted)) 298 if (unlikely(faulted))
295 goto out; 299 goto out;
296 300
297 if (ftrace_push_return_trace(old, self_addr, &trace.depth, fp) == 301 if (ftrace_push_return_trace(old_parent_ra, self_ra, &trace.depth, fp)
298 -EBUSY) { 302 == -EBUSY) {
299 *parent = old; 303 *parent_ra_addr = old_parent_ra;
300 return; 304 return;
301 } 305 }
302 306
303 trace.func = self_addr; 307 /*
308 * Get the recorded ip of the current mcount calling site in the
309 * __mcount_loc section, which will be used to filter the function
310 * entries configured through the tracing/set_graph_function interface.
311 */
312
313 insns = in_kernel_space(self_ra) ? 2 : MCOUNT_OFFSET_INSNS + 1;
314 trace.func = self_ra - (MCOUNT_INSN_SIZE * insns);
304 315
305 /* Only trace if the calling function expects to */ 316 /* Only trace if the calling function expects to */
306 if (!ftrace_graph_entry(&trace)) { 317 if (!ftrace_graph_entry(&trace)) {
307 current->curr_ret_stack--; 318 current->curr_ret_stack--;
308 *parent = old; 319 *parent_ra_addr = old_parent_ra;
309 } 320 }
310 return; 321 return;
311out: 322out:
diff --git a/arch/mips/kernel/perf_event.c b/arch/mips/kernel/perf_event.c
index 2b7f3f703b83..a8244854d3dc 100644
--- a/arch/mips/kernel/perf_event.c
+++ b/arch/mips/kernel/perf_event.c
@@ -161,41 +161,6 @@ mipspmu_event_set_period(struct perf_event *event,
161 return ret; 161 return ret;
162} 162}
163 163
164static int mipspmu_enable(struct perf_event *event)
165{
166 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
167 struct hw_perf_event *hwc = &event->hw;
168 int idx;
169 int err = 0;
170
171 /* To look for a free counter for this event. */
172 idx = mipspmu->alloc_counter(cpuc, hwc);
173 if (idx < 0) {
174 err = idx;
175 goto out;
176 }
177
178 /*
179 * If there is an event in the counter we are going to use then
180 * make sure it is disabled.
181 */
182 event->hw.idx = idx;
183 mipspmu->disable_event(idx);
184 cpuc->events[idx] = event;
185
186 /* Set the period for the event. */
187 mipspmu_event_set_period(event, hwc, idx);
188
189 /* Enable the event. */
190 mipspmu->enable_event(hwc, idx);
191
192 /* Propagate our changes to the userspace mapping. */
193 perf_event_update_userpage(event);
194
195out:
196 return err;
197}
198
199static void mipspmu_event_update(struct perf_event *event, 164static void mipspmu_event_update(struct perf_event *event,
200 struct hw_perf_event *hwc, 165 struct hw_perf_event *hwc,
201 int idx) 166 int idx)
@@ -204,7 +169,7 @@ static void mipspmu_event_update(struct perf_event *event,
204 unsigned long flags; 169 unsigned long flags;
205 int shift = 64 - TOTAL_BITS; 170 int shift = 64 - TOTAL_BITS;
206 s64 prev_raw_count, new_raw_count; 171 s64 prev_raw_count, new_raw_count;
207 s64 delta; 172 u64 delta;
208 173
209again: 174again:
210 prev_raw_count = local64_read(&hwc->prev_count); 175 prev_raw_count = local64_read(&hwc->prev_count);
@@ -231,32 +196,90 @@ again:
231 return; 196 return;
232} 197}
233 198
234static void mipspmu_disable(struct perf_event *event) 199static void mipspmu_start(struct perf_event *event, int flags)
200{
201 struct hw_perf_event *hwc = &event->hw;
202
203 if (!mipspmu)
204 return;
205
206 if (flags & PERF_EF_RELOAD)
207 WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
208
209 hwc->state = 0;
210
211 /* Set the period for the event. */
212 mipspmu_event_set_period(event, hwc, hwc->idx);
213
214 /* Enable the event. */
215 mipspmu->enable_event(hwc, hwc->idx);
216}
217
218static void mipspmu_stop(struct perf_event *event, int flags)
219{
220 struct hw_perf_event *hwc = &event->hw;
221
222 if (!mipspmu)
223 return;
224
225 if (!(hwc->state & PERF_HES_STOPPED)) {
226 /* We are working on a local event. */
227 mipspmu->disable_event(hwc->idx);
228 barrier();
229 mipspmu_event_update(event, hwc, hwc->idx);
230 hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE;
231 }
232}
233
234static int mipspmu_add(struct perf_event *event, int flags)
235{ 235{
236 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); 236 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
237 struct hw_perf_event *hwc = &event->hw; 237 struct hw_perf_event *hwc = &event->hw;
238 int idx = hwc->idx; 238 int idx;
239 int err = 0;
239 240
241 perf_pmu_disable(event->pmu);
240 242
241 WARN_ON(idx < 0 || idx >= mipspmu->num_counters); 243 /* To look for a free counter for this event. */
244 idx = mipspmu->alloc_counter(cpuc, hwc);
245 if (idx < 0) {
246 err = idx;
247 goto out;
248 }
242 249
243 /* We are working on a local event. */ 250 /*
251 * If there is an event in the counter we are going to use then
252 * make sure it is disabled.
253 */
254 event->hw.idx = idx;
244 mipspmu->disable_event(idx); 255 mipspmu->disable_event(idx);
256 cpuc->events[idx] = event;
245 257
246 barrier(); 258 hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
247 259 if (flags & PERF_EF_START)
248 mipspmu_event_update(event, hwc, idx); 260 mipspmu_start(event, PERF_EF_RELOAD);
249 cpuc->events[idx] = NULL;
250 clear_bit(idx, cpuc->used_mask);
251 261
262 /* Propagate our changes to the userspace mapping. */
252 perf_event_update_userpage(event); 263 perf_event_update_userpage(event);
264
265out:
266 perf_pmu_enable(event->pmu);
267 return err;
253} 268}
254 269
255static void mipspmu_unthrottle(struct perf_event *event) 270static void mipspmu_del(struct perf_event *event, int flags)
256{ 271{
272 struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
257 struct hw_perf_event *hwc = &event->hw; 273 struct hw_perf_event *hwc = &event->hw;
274 int idx = hwc->idx;
258 275
259 mipspmu->enable_event(hwc, hwc->idx); 276 WARN_ON(idx < 0 || idx >= mipspmu->num_counters);
277
278 mipspmu_stop(event, PERF_EF_UPDATE);
279 cpuc->events[idx] = NULL;
280 clear_bit(idx, cpuc->used_mask);
281
282 perf_event_update_userpage(event);
260} 283}
261 284
262static void mipspmu_read(struct perf_event *event) 285static void mipspmu_read(struct perf_event *event)
@@ -270,12 +293,17 @@ static void mipspmu_read(struct perf_event *event)
270 mipspmu_event_update(event, hwc, hwc->idx); 293 mipspmu_event_update(event, hwc, hwc->idx);
271} 294}
272 295
273static struct pmu pmu = { 296static void mipspmu_enable(struct pmu *pmu)
274 .enable = mipspmu_enable, 297{
275 .disable = mipspmu_disable, 298 if (mipspmu)
276 .unthrottle = mipspmu_unthrottle, 299 mipspmu->start();
277 .read = mipspmu_read, 300}
278}; 301
302static void mipspmu_disable(struct pmu *pmu)
303{
304 if (mipspmu)
305 mipspmu->stop();
306}
279 307
280static atomic_t active_events = ATOMIC_INIT(0); 308static atomic_t active_events = ATOMIC_INIT(0);
281static DEFINE_MUTEX(pmu_reserve_mutex); 309static DEFINE_MUTEX(pmu_reserve_mutex);
@@ -318,6 +346,82 @@ static void mipspmu_free_irq(void)
318 perf_irq = save_perf_irq; 346 perf_irq = save_perf_irq;
319} 347}
320 348
349/*
350 * mipsxx/rm9000/loongson2 have different performance counters, they have
351 * specific low-level init routines.
352 */
353static void reset_counters(void *arg);
354static int __hw_perf_event_init(struct perf_event *event);
355
356static void hw_perf_event_destroy(struct perf_event *event)
357{
358 if (atomic_dec_and_mutex_lock(&active_events,
359 &pmu_reserve_mutex)) {
360 /*
361 * We must not call the destroy function with interrupts
362 * disabled.
363 */
364 on_each_cpu(reset_counters,
365 (void *)(long)mipspmu->num_counters, 1);
366 mipspmu_free_irq();
367 mutex_unlock(&pmu_reserve_mutex);
368 }
369}
370
371static int mipspmu_event_init(struct perf_event *event)
372{
373 int err = 0;
374
375 switch (event->attr.type) {
376 case PERF_TYPE_RAW:
377 case PERF_TYPE_HARDWARE:
378 case PERF_TYPE_HW_CACHE:
379 break;
380
381 default:
382 return -ENOENT;
383 }
384
385 if (!mipspmu || event->cpu >= nr_cpumask_bits ||
386 (event->cpu >= 0 && !cpu_online(event->cpu)))
387 return -ENODEV;
388
389 if (!atomic_inc_not_zero(&active_events)) {
390 if (atomic_read(&active_events) > MIPS_MAX_HWEVENTS) {
391 atomic_dec(&active_events);
392 return -ENOSPC;
393 }
394
395 mutex_lock(&pmu_reserve_mutex);
396 if (atomic_read(&active_events) == 0)
397 err = mipspmu_get_irq();
398
399 if (!err)
400 atomic_inc(&active_events);
401 mutex_unlock(&pmu_reserve_mutex);
402 }
403
404 if (err)
405 return err;
406
407 err = __hw_perf_event_init(event);
408 if (err)
409 hw_perf_event_destroy(event);
410
411 return err;
412}
413
414static struct pmu pmu = {
415 .pmu_enable = mipspmu_enable,
416 .pmu_disable = mipspmu_disable,
417 .event_init = mipspmu_event_init,
418 .add = mipspmu_add,
419 .del = mipspmu_del,
420 .start = mipspmu_start,
421 .stop = mipspmu_stop,
422 .read = mipspmu_read,
423};
424
321static inline unsigned int 425static inline unsigned int
322mipspmu_perf_event_encode(const struct mips_perf_event *pev) 426mipspmu_perf_event_encode(const struct mips_perf_event *pev)
323{ 427{
@@ -382,8 +486,9 @@ static int validate_event(struct cpu_hw_events *cpuc,
382{ 486{
383 struct hw_perf_event fake_hwc = event->hw; 487 struct hw_perf_event fake_hwc = event->hw;
384 488
385 if (event->pmu && event->pmu != &pmu) 489 /* Allow mixed event group. So return 1 to pass validation. */
386 return 0; 490 if (event->pmu != &pmu || event->state <= PERF_EVENT_STATE_OFF)
491 return 1;
387 492
388 return mipspmu->alloc_counter(cpuc, &fake_hwc) >= 0; 493 return mipspmu->alloc_counter(cpuc, &fake_hwc) >= 0;
389} 494}
@@ -409,73 +514,6 @@ static int validate_group(struct perf_event *event)
409 return 0; 514 return 0;
410} 515}
411 516
412/*
413 * mipsxx/rm9000/loongson2 have different performance counters, they have
414 * specific low-level init routines.
415 */
416static void reset_counters(void *arg);
417static int __hw_perf_event_init(struct perf_event *event);
418
419static void hw_perf_event_destroy(struct perf_event *event)
420{
421 if (atomic_dec_and_mutex_lock(&active_events,
422 &pmu_reserve_mutex)) {
423 /*
424 * We must not call the destroy function with interrupts
425 * disabled.
426 */
427 on_each_cpu(reset_counters,
428 (void *)(long)mipspmu->num_counters, 1);
429 mipspmu_free_irq();
430 mutex_unlock(&pmu_reserve_mutex);
431 }
432}
433
434const struct pmu *hw_perf_event_init(struct perf_event *event)
435{
436 int err = 0;
437
438 if (!mipspmu || event->cpu >= nr_cpumask_bits ||
439 (event->cpu >= 0 && !cpu_online(event->cpu)))
440 return ERR_PTR(-ENODEV);
441
442 if (!atomic_inc_not_zero(&active_events)) {
443 if (atomic_read(&active_events) > MIPS_MAX_HWEVENTS) {
444 atomic_dec(&active_events);
445 return ERR_PTR(-ENOSPC);
446 }
447
448 mutex_lock(&pmu_reserve_mutex);
449 if (atomic_read(&active_events) == 0)
450 err = mipspmu_get_irq();
451
452 if (!err)
453 atomic_inc(&active_events);
454 mutex_unlock(&pmu_reserve_mutex);
455 }
456
457 if (err)
458 return ERR_PTR(err);
459
460 err = __hw_perf_event_init(event);
461 if (err)
462 hw_perf_event_destroy(event);
463
464 return err ? ERR_PTR(err) : &pmu;
465}
466
467void hw_perf_enable(void)
468{
469 if (mipspmu)
470 mipspmu->start();
471}
472
473void hw_perf_disable(void)
474{
475 if (mipspmu)
476 mipspmu->stop();
477}
478
479/* This is needed by specific irq handlers in perf_event_*.c */ 517/* This is needed by specific irq handlers in perf_event_*.c */
480static void 518static void
481handle_associated_event(struct cpu_hw_events *cpuc, 519handle_associated_event(struct cpu_hw_events *cpuc,
@@ -496,21 +534,13 @@ handle_associated_event(struct cpu_hw_events *cpuc,
496#include "perf_event_mipsxx.c" 534#include "perf_event_mipsxx.c"
497 535
498/* Callchain handling code. */ 536/* Callchain handling code. */
499static inline void
500callchain_store(struct perf_callchain_entry *entry,
501 u64 ip)
502{
503 if (entry->nr < PERF_MAX_STACK_DEPTH)
504 entry->ip[entry->nr++] = ip;
505}
506 537
507/* 538/*
508 * Leave userspace callchain empty for now. When we find a way to trace 539 * Leave userspace callchain empty for now. When we find a way to trace
509 * the user stack callchains, we add here. 540 * the user stack callchains, we add here.
510 */ 541 */
511static void 542void perf_callchain_user(struct perf_callchain_entry *entry,
512perf_callchain_user(struct pt_regs *regs, 543 struct pt_regs *regs)
513 struct perf_callchain_entry *entry)
514{ 544{
515} 545}
516 546
@@ -523,23 +553,21 @@ static void save_raw_perf_callchain(struct perf_callchain_entry *entry,
523 while (!kstack_end(sp)) { 553 while (!kstack_end(sp)) {
524 addr = *sp++; 554 addr = *sp++;
525 if (__kernel_text_address(addr)) { 555 if (__kernel_text_address(addr)) {
526 callchain_store(entry, addr); 556 perf_callchain_store(entry, addr);
527 if (entry->nr >= PERF_MAX_STACK_DEPTH) 557 if (entry->nr >= PERF_MAX_STACK_DEPTH)
528 break; 558 break;
529 } 559 }
530 } 560 }
531} 561}
532 562
533static void 563void perf_callchain_kernel(struct perf_callchain_entry *entry,
534perf_callchain_kernel(struct pt_regs *regs, 564 struct pt_regs *regs)
535 struct perf_callchain_entry *entry)
536{ 565{
537 unsigned long sp = regs->regs[29]; 566 unsigned long sp = regs->regs[29];
538#ifdef CONFIG_KALLSYMS 567#ifdef CONFIG_KALLSYMS
539 unsigned long ra = regs->regs[31]; 568 unsigned long ra = regs->regs[31];
540 unsigned long pc = regs->cp0_epc; 569 unsigned long pc = regs->cp0_epc;
541 570
542 callchain_store(entry, PERF_CONTEXT_KERNEL);
543 if (raw_show_trace || !__kernel_text_address(pc)) { 571 if (raw_show_trace || !__kernel_text_address(pc)) {
544 unsigned long stack_page = 572 unsigned long stack_page =
545 (unsigned long)task_stack_page(current); 573 (unsigned long)task_stack_page(current);
@@ -549,53 +577,12 @@ perf_callchain_kernel(struct pt_regs *regs,
549 return; 577 return;
550 } 578 }
551 do { 579 do {
552 callchain_store(entry, pc); 580 perf_callchain_store(entry, pc);
553 if (entry->nr >= PERF_MAX_STACK_DEPTH) 581 if (entry->nr >= PERF_MAX_STACK_DEPTH)
554 break; 582 break;
555 pc = unwind_stack(current, &sp, pc, &ra); 583 pc = unwind_stack(current, &sp, pc, &ra);
556 } while (pc); 584 } while (pc);
557#else 585#else
558 callchain_store(entry, PERF_CONTEXT_KERNEL);
559 save_raw_perf_callchain(entry, sp); 586 save_raw_perf_callchain(entry, sp);
560#endif 587#endif
561} 588}
562
563static void
564perf_do_callchain(struct pt_regs *regs,
565 struct perf_callchain_entry *entry)
566{
567 int is_user;
568
569 if (!regs)
570 return;
571
572 is_user = user_mode(regs);
573
574 if (!current || !current->pid)
575 return;
576
577 if (is_user && current->state != TASK_RUNNING)
578 return;
579
580 if (!is_user) {
581 perf_callchain_kernel(regs, entry);
582 if (current->mm)
583 regs = task_pt_regs(current);
584 else
585 regs = NULL;
586 }
587 if (regs)
588 perf_callchain_user(regs, entry);
589}
590
591static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_irq_entry);
592
593struct perf_callchain_entry *
594perf_callchain(struct pt_regs *regs)
595{
596 struct perf_callchain_entry *entry = &__get_cpu_var(pmc_irq_entry);
597
598 entry->nr = 0;
599 perf_do_callchain(regs, entry);
600 return entry;
601}
diff --git a/arch/mips/kernel/perf_event_mipsxx.c b/arch/mips/kernel/perf_event_mipsxx.c
index 183e0d226669..d9a7db78ed62 100644
--- a/arch/mips/kernel/perf_event_mipsxx.c
+++ b/arch/mips/kernel/perf_event_mipsxx.c
@@ -696,7 +696,7 @@ static int mipsxx_pmu_handle_shared_irq(void)
696 * interrupt, not NMI. 696 * interrupt, not NMI.
697 */ 697 */
698 if (handled == IRQ_HANDLED) 698 if (handled == IRQ_HANDLED)
699 perf_event_do_pending(); 699 irq_work_run();
700 700
701#ifdef CONFIG_MIPS_MT_SMP 701#ifdef CONFIG_MIPS_MT_SMP
702 read_unlock(&pmuint_rwlock); 702 read_unlock(&pmuint_rwlock);
@@ -1045,6 +1045,8 @@ init_hw_perf_events(void)
1045 "CPU, irq %d%s\n", mipspmu->name, counters, irq, 1045 "CPU, irq %d%s\n", mipspmu->name, counters, irq,
1046 irq < 0 ? " (share with timer interrupt)" : ""); 1046 irq < 0 ? " (share with timer interrupt)" : "");
1047 1047
1048 perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
1049
1048 return 0; 1050 return 0;
1049} 1051}
1050early_initcall(init_hw_perf_events); 1052early_initcall(init_hw_perf_events);
diff --git a/arch/mips/kernel/signal.c b/arch/mips/kernel/signal.c
index 5922342bca39..dbbe0ce48d89 100644
--- a/arch/mips/kernel/signal.c
+++ b/arch/mips/kernel/signal.c
@@ -84,7 +84,7 @@ static int protected_save_fp_context(struct sigcontext __user *sc)
84 84
85static int protected_restore_fp_context(struct sigcontext __user *sc) 85static int protected_restore_fp_context(struct sigcontext __user *sc)
86{ 86{
87 int err, tmp; 87 int err, tmp __maybe_unused;
88 while (1) { 88 while (1) {
89 lock_fpu_owner(); 89 lock_fpu_owner();
90 own_fpu_inatomic(0); 90 own_fpu_inatomic(0);
diff --git a/arch/mips/kernel/signal32.c b/arch/mips/kernel/signal32.c
index a0ed0e052b2e..aae986613795 100644
--- a/arch/mips/kernel/signal32.c
+++ b/arch/mips/kernel/signal32.c
@@ -115,7 +115,7 @@ static int protected_save_fp_context32(struct sigcontext32 __user *sc)
115 115
116static int protected_restore_fp_context32(struct sigcontext32 __user *sc) 116static int protected_restore_fp_context32(struct sigcontext32 __user *sc)
117{ 117{
118 int err, tmp; 118 int err, tmp __maybe_unused;
119 while (1) { 119 while (1) {
120 lock_fpu_owner(); 120 lock_fpu_owner();
121 own_fpu_inatomic(0); 121 own_fpu_inatomic(0);
diff --git a/arch/mips/kernel/smp.c b/arch/mips/kernel/smp.c
index 383aeb95cb49..32a256101082 100644
--- a/arch/mips/kernel/smp.c
+++ b/arch/mips/kernel/smp.c
@@ -193,6 +193,22 @@ void __devinit smp_prepare_boot_cpu(void)
193 */ 193 */
194static struct task_struct *cpu_idle_thread[NR_CPUS]; 194static struct task_struct *cpu_idle_thread[NR_CPUS];
195 195
196struct create_idle {
197 struct work_struct work;
198 struct task_struct *idle;
199 struct completion done;
200 int cpu;
201};
202
203static void __cpuinit do_fork_idle(struct work_struct *work)
204{
205 struct create_idle *c_idle =
206 container_of(work, struct create_idle, work);
207
208 c_idle->idle = fork_idle(c_idle->cpu);
209 complete(&c_idle->done);
210}
211
196int __cpuinit __cpu_up(unsigned int cpu) 212int __cpuinit __cpu_up(unsigned int cpu)
197{ 213{
198 struct task_struct *idle; 214 struct task_struct *idle;
@@ -203,8 +219,19 @@ int __cpuinit __cpu_up(unsigned int cpu)
203 * Linux can schedule processes on this slave. 219 * Linux can schedule processes on this slave.
204 */ 220 */
205 if (!cpu_idle_thread[cpu]) { 221 if (!cpu_idle_thread[cpu]) {
206 idle = fork_idle(cpu); 222 /*
207 cpu_idle_thread[cpu] = idle; 223 * Schedule work item to avoid forking user task
224 * Ported from arch/x86/kernel/smpboot.c
225 */
226 struct create_idle c_idle = {
227 .cpu = cpu,
228 .done = COMPLETION_INITIALIZER_ONSTACK(c_idle.done),
229 };
230
231 INIT_WORK_ONSTACK(&c_idle.work, do_fork_idle);
232 schedule_work(&c_idle.work);
233 wait_for_completion(&c_idle.done);
234 idle = cpu_idle_thread[cpu] = c_idle.idle;
208 235
209 if (IS_ERR(idle)) 236 if (IS_ERR(idle))
210 panic(KERN_ERR "Fork failed for CPU %d", cpu); 237 panic(KERN_ERR "Fork failed for CPU %d", cpu);
diff --git a/arch/mips/kernel/syscall.c b/arch/mips/kernel/syscall.c
index 1dc6edff45e0..58beabf50b3c 100644
--- a/arch/mips/kernel/syscall.c
+++ b/arch/mips/kernel/syscall.c
@@ -383,12 +383,11 @@ save_static_function(sys_sysmips);
383static int __used noinline 383static int __used noinline
384_sys_sysmips(nabi_no_regargs struct pt_regs regs) 384_sys_sysmips(nabi_no_regargs struct pt_regs regs)
385{ 385{
386 long cmd, arg1, arg2, arg3; 386 long cmd, arg1, arg2;
387 387
388 cmd = regs.regs[4]; 388 cmd = regs.regs[4];
389 arg1 = regs.regs[5]; 389 arg1 = regs.regs[5];
390 arg2 = regs.regs[6]; 390 arg2 = regs.regs[6];
391 arg3 = regs.regs[7];
392 391
393 switch (cmd) { 392 switch (cmd) {
394 case MIPS_ATOMIC_SET: 393 case MIPS_ATOMIC_SET:
@@ -405,7 +404,7 @@ _sys_sysmips(nabi_no_regargs struct pt_regs regs)
405 if (arg1 & 2) 404 if (arg1 & 2)
406 set_thread_flag(TIF_LOGADE); 405 set_thread_flag(TIF_LOGADE);
407 else 406 else
408 clear_thread_flag(TIF_FIXADE); 407 clear_thread_flag(TIF_LOGADE);
409 408
410 return 0; 409 return 0;
411 410
diff --git a/arch/mips/kernel/vpe.c b/arch/mips/kernel/vpe.c
index 6a1fdfef8fde..ab52b7cf3b6b 100644
--- a/arch/mips/kernel/vpe.c
+++ b/arch/mips/kernel/vpe.c
@@ -148,9 +148,9 @@ struct {
148 spinlock_t tc_list_lock; 148 spinlock_t tc_list_lock;
149 struct list_head tc_list; /* Thread contexts */ 149 struct list_head tc_list; /* Thread contexts */
150} vpecontrol = { 150} vpecontrol = {
151 .vpe_list_lock = SPIN_LOCK_UNLOCKED, 151 .vpe_list_lock = __SPIN_LOCK_UNLOCKED(vpe_list_lock),
152 .vpe_list = LIST_HEAD_INIT(vpecontrol.vpe_list), 152 .vpe_list = LIST_HEAD_INIT(vpecontrol.vpe_list),
153 .tc_list_lock = SPIN_LOCK_UNLOCKED, 153 .tc_list_lock = __SPIN_LOCK_UNLOCKED(tc_list_lock),
154 .tc_list = LIST_HEAD_INIT(vpecontrol.tc_list) 154 .tc_list = LIST_HEAD_INIT(vpecontrol.tc_list)
155}; 155};
156 156
diff --git a/arch/mips/loongson/Kconfig b/arch/mips/loongson/Kconfig
index 6e1b77fec7ea..aca93eed8779 100644
--- a/arch/mips/loongson/Kconfig
+++ b/arch/mips/loongson/Kconfig
@@ -1,6 +1,7 @@
1if MACH_LOONGSON
2
1choice 3choice
2 prompt "Machine Type" 4 prompt "Machine Type"
3 depends on MACH_LOONGSON
4 5
5config LEMOTE_FULOONG2E 6config LEMOTE_FULOONG2E
6 bool "Lemote Fuloong(2e) mini-PC" 7 bool "Lemote Fuloong(2e) mini-PC"
@@ -87,3 +88,5 @@ config LOONGSON_UART_BASE
87config LOONGSON_MC146818 88config LOONGSON_MC146818
88 bool 89 bool
89 default n 90 default n
91
92endif # MACH_LOONGSON
diff --git a/arch/mips/loongson/common/cmdline.c b/arch/mips/loongson/common/cmdline.c
index 1a06defc4f7f..353e1d2e41a5 100644
--- a/arch/mips/loongson/common/cmdline.c
+++ b/arch/mips/loongson/common/cmdline.c
@@ -44,10 +44,5 @@ void __init prom_init_cmdline(void)
44 strcat(arcs_cmdline, " "); 44 strcat(arcs_cmdline, " ");
45 } 45 }
46 46
47 if ((strstr(arcs_cmdline, "console=")) == NULL)
48 strcat(arcs_cmdline, " console=ttyS0,115200");
49 if ((strstr(arcs_cmdline, "root=")) == NULL)
50 strcat(arcs_cmdline, " root=/dev/hda1");
51
52 prom_init_machtype(); 47 prom_init_machtype();
53} 48}
diff --git a/arch/mips/loongson/common/machtype.c b/arch/mips/loongson/common/machtype.c
index 81fbe6b73f91..2efd5d9dee27 100644
--- a/arch/mips/loongson/common/machtype.c
+++ b/arch/mips/loongson/common/machtype.c
@@ -41,7 +41,7 @@ void __weak __init mach_prom_init_machtype(void)
41 41
42void __init prom_init_machtype(void) 42void __init prom_init_machtype(void)
43{ 43{
44 char *p, str[MACHTYPE_LEN]; 44 char *p, str[MACHTYPE_LEN + 1];
45 int machtype = MACH_LEMOTE_FL2E; 45 int machtype = MACH_LEMOTE_FL2E;
46 46
47 mips_machtype = LOONGSON_MACHTYPE; 47 mips_machtype = LOONGSON_MACHTYPE;
@@ -53,6 +53,7 @@ void __init prom_init_machtype(void)
53 } 53 }
54 p += strlen("machtype="); 54 p += strlen("machtype=");
55 strncpy(str, p, MACHTYPE_LEN); 55 strncpy(str, p, MACHTYPE_LEN);
56 str[MACHTYPE_LEN] = '\0';
56 p = strstr(str, " "); 57 p = strstr(str, " ");
57 if (p) 58 if (p)
58 *p = '\0'; 59 *p = '\0';
diff --git a/arch/mips/math-emu/ieee754int.h b/arch/mips/math-emu/ieee754int.h
index 2701d9500959..2a7d43f4f161 100644
--- a/arch/mips/math-emu/ieee754int.h
+++ b/arch/mips/math-emu/ieee754int.h
@@ -70,7 +70,7 @@
70 70
71 71
72#define COMPXSP \ 72#define COMPXSP \
73 unsigned xm; int xe; int xs; int xc 73 unsigned xm; int xe; int xs __maybe_unused; int xc
74 74
75#define COMPYSP \ 75#define COMPYSP \
76 unsigned ym; int ye; int ys; int yc 76 unsigned ym; int ye; int ys; int yc
@@ -104,7 +104,7 @@
104 104
105 105
106#define COMPXDP \ 106#define COMPXDP \
107u64 xm; int xe; int xs; int xc 107u64 xm; int xe; int xs __maybe_unused; int xc
108 108
109#define COMPYDP \ 109#define COMPYDP \
110u64 ym; int ye; int ys; int yc 110u64 ym; int ye; int ys; int yc
diff --git a/arch/mips/mm/init.c b/arch/mips/mm/init.c
index 2efcbd24c82f..279599e9a779 100644
--- a/arch/mips/mm/init.c
+++ b/arch/mips/mm/init.c
@@ -324,7 +324,7 @@ int page_is_ram(unsigned long pagenr)
324void __init paging_init(void) 324void __init paging_init(void)
325{ 325{
326 unsigned long max_zone_pfns[MAX_NR_ZONES]; 326 unsigned long max_zone_pfns[MAX_NR_ZONES];
327 unsigned long lastpfn; 327 unsigned long lastpfn __maybe_unused;
328 328
329 pagetable_init(); 329 pagetable_init();
330 330
diff --git a/arch/mips/mm/tlbex.c b/arch/mips/mm/tlbex.c
index 083d3412d0bc..04f9e17db9d0 100644
--- a/arch/mips/mm/tlbex.c
+++ b/arch/mips/mm/tlbex.c
@@ -109,6 +109,8 @@ static bool scratchpad_available(void)
109static int scratchpad_offset(int i) 109static int scratchpad_offset(int i)
110{ 110{
111 BUG(); 111 BUG();
112 /* Really unreachable, but evidently some GCC want this. */
113 return 0;
112} 114}
113#endif 115#endif
114/* 116/*
diff --git a/arch/mips/pci/ops-pmcmsp.c b/arch/mips/pci/ops-pmcmsp.c
index b7c03d80c88c..68798f869c0f 100644
--- a/arch/mips/pci/ops-pmcmsp.c
+++ b/arch/mips/pci/ops-pmcmsp.c
@@ -308,7 +308,7 @@ static struct resource pci_mem_resource = {
308 * RETURNS: PCIBIOS_SUCCESSFUL - success 308 * RETURNS: PCIBIOS_SUCCESSFUL - success
309 * 309 *
310 ****************************************************************************/ 310 ****************************************************************************/
311static int bpci_interrupt(int irq, void *dev_id) 311static irqreturn_t bpci_interrupt(int irq, void *dev_id)
312{ 312{
313 struct msp_pci_regs *preg = (void *)PCI_BASE_REG; 313 struct msp_pci_regs *preg = (void *)PCI_BASE_REG;
314 unsigned int stat = preg->if_status; 314 unsigned int stat = preg->if_status;
@@ -326,7 +326,7 @@ static int bpci_interrupt(int irq, void *dev_id)
326 /* write to clear all asserted interrupts */ 326 /* write to clear all asserted interrupts */
327 preg->if_status = stat; 327 preg->if_status = stat;
328 328
329 return PCIBIOS_SUCCESSFUL; 329 return IRQ_HANDLED;
330} 330}
331 331
332/***************************************************************************** 332/*****************************************************************************
diff --git a/arch/mips/pmc-sierra/Kconfig b/arch/mips/pmc-sierra/Kconfig
index c139988bb85d..8d798497c614 100644
--- a/arch/mips/pmc-sierra/Kconfig
+++ b/arch/mips/pmc-sierra/Kconfig
@@ -4,15 +4,11 @@ choice
4 4
5config PMC_MSP4200_EVAL 5config PMC_MSP4200_EVAL
6 bool "PMC-Sierra MSP4200 Eval Board" 6 bool "PMC-Sierra MSP4200 Eval Board"
7 select CEVT_R4K
8 select CSRC_R4K
9 select IRQ_MSP_SLP 7 select IRQ_MSP_SLP
10 select HW_HAS_PCI 8 select HW_HAS_PCI
11 9
12config PMC_MSP4200_GW 10config PMC_MSP4200_GW
13 bool "PMC-Sierra MSP4200 VoIP Gateway" 11 bool "PMC-Sierra MSP4200 VoIP Gateway"
14 select CEVT_R4K
15 select CSRC_R4K
16 select IRQ_MSP_SLP 12 select IRQ_MSP_SLP
17 select HW_HAS_PCI 13 select HW_HAS_PCI
18 14
diff --git a/arch/mips/pmc-sierra/msp71xx/msp_time.c b/arch/mips/pmc-sierra/msp71xx/msp_time.c
index cca64e15f57f..01df84ce31e2 100644
--- a/arch/mips/pmc-sierra/msp71xx/msp_time.c
+++ b/arch/mips/pmc-sierra/msp71xx/msp_time.c
@@ -81,7 +81,7 @@ void __init plat_time_init(void)
81 mips_hpt_frequency = cpu_rate/2; 81 mips_hpt_frequency = cpu_rate/2;
82} 82}
83 83
84unsigned int __init get_c0_compare_int(void) 84unsigned int __cpuinit get_c0_compare_int(void)
85{ 85{
86 return MSP_INT_VPE0_TIMER; 86 return MSP_INT_VPE0_TIMER;
87} 87}
diff --git a/arch/mn10300/include/asm/atomic.h b/arch/mn10300/include/asm/atomic.h
index 92d2f9298e38..9d773a639513 100644
--- a/arch/mn10300/include/asm/atomic.h
+++ b/arch/mn10300/include/asm/atomic.h
@@ -139,7 +139,7 @@ static inline unsigned long __cmpxchg(volatile unsigned long *m,
139 * Atomically reads the value of @v. Note that the guaranteed 139 * Atomically reads the value of @v. Note that the guaranteed
140 * useful range of an atomic_t is only 24 bits. 140 * useful range of an atomic_t is only 24 bits.
141 */ 141 */
142#define atomic_read(v) ((v)->counter) 142#define atomic_read(v) (ACCESS_ONCE((v)->counter))
143 143
144/** 144/**
145 * atomic_set - set atomic variable 145 * atomic_set - set atomic variable
diff --git a/arch/mn10300/include/asm/uaccess.h b/arch/mn10300/include/asm/uaccess.h
index 679dee0bbd08..3d6e60dad9d9 100644
--- a/arch/mn10300/include/asm/uaccess.h
+++ b/arch/mn10300/include/asm/uaccess.h
@@ -160,9 +160,10 @@ struct __large_struct { unsigned long buf[100]; };
160 160
161#define __get_user_check(x, ptr, size) \ 161#define __get_user_check(x, ptr, size) \
162({ \ 162({ \
163 const __typeof__(ptr) __guc_ptr = (ptr); \
163 int _e; \ 164 int _e; \
164 if (likely(__access_ok((unsigned long) (ptr), (size)))) \ 165 if (likely(__access_ok((unsigned long) __guc_ptr, (size)))) \
165 _e = __get_user_nocheck((x), (ptr), (size)); \ 166 _e = __get_user_nocheck((x), __guc_ptr, (size)); \
166 else { \ 167 else { \
167 _e = -EFAULT; \ 168 _e = -EFAULT; \
168 (x) = (__typeof__(x))0; \ 169 (x) = (__typeof__(x))0; \
diff --git a/arch/mn10300/mm/cache-inv-icache.c b/arch/mn10300/mm/cache-inv-icache.c
index a8933a60b2d4..a6b63dde603d 100644
--- a/arch/mn10300/mm/cache-inv-icache.c
+++ b/arch/mn10300/mm/cache-inv-icache.c
@@ -69,7 +69,7 @@ static void flush_icache_page_range(unsigned long start, unsigned long end)
69 69
70 /* invalidate the icache coverage on that region */ 70 /* invalidate the icache coverage on that region */
71 mn10300_local_icache_inv_range2(addr + off, size); 71 mn10300_local_icache_inv_range2(addr + off, size);
72 smp_cache_call(SMP_ICACHE_INV_FLUSH_RANGE, start, end); 72 smp_cache_call(SMP_ICACHE_INV_RANGE, start, end);
73} 73}
74 74
75/** 75/**
@@ -101,7 +101,7 @@ void flush_icache_range(unsigned long start, unsigned long end)
101 * directly */ 101 * directly */
102 start_page = (start >= 0x80000000UL) ? start : 0x80000000UL; 102 start_page = (start >= 0x80000000UL) ? start : 0x80000000UL;
103 mn10300_icache_inv_range(start_page, end); 103 mn10300_icache_inv_range(start_page, end);
104 smp_cache_call(SMP_ICACHE_INV_FLUSH_RANGE, start, end); 104 smp_cache_call(SMP_ICACHE_INV_RANGE, start, end);
105 if (start_page == start) 105 if (start_page == start)
106 goto done; 106 goto done;
107 end = start_page; 107 end = start_page;
diff --git a/arch/parisc/hpux/sys_hpux.c b/arch/parisc/hpux/sys_hpux.c
index 30394081d9b6..6ab9580b0b00 100644
--- a/arch/parisc/hpux/sys_hpux.c
+++ b/arch/parisc/hpux/sys_hpux.c
@@ -185,26 +185,21 @@ struct hpux_statfs {
185 int16_t f_pad; 185 int16_t f_pad;
186}; 186};
187 187
188static int do_statfs_hpux(struct path *path, struct hpux_statfs *buf) 188static int do_statfs_hpux(struct kstatfs *st, struct hpux_statfs __user *p)
189{ 189{
190 struct kstatfs st; 190 struct hpux_statfs buf;
191 int retval; 191 memset(&buf, 0, sizeof(buf));
192 192 buf.f_type = st->f_type;
193 retval = vfs_statfs(path, &st); 193 buf.f_bsize = st->f_bsize;
194 if (retval) 194 buf.f_blocks = st->f_blocks;
195 return retval; 195 buf.f_bfree = st->f_bfree;
196 196 buf.f_bavail = st->f_bavail;
197 memset(buf, 0, sizeof(*buf)); 197 buf.f_files = st->f_files;
198 buf->f_type = st.f_type; 198 buf.f_ffree = st->f_ffree;
199 buf->f_bsize = st.f_bsize; 199 buf.f_fsid[0] = st->f_fsid.val[0];
200 buf->f_blocks = st.f_blocks; 200 buf.f_fsid[1] = st->f_fsid.val[1];
201 buf->f_bfree = st.f_bfree; 201 if (copy_to_user(p, &buf, sizeof(buf)))
202 buf->f_bavail = st.f_bavail; 202 return -EFAULT;
203 buf->f_files = st.f_files;
204 buf->f_ffree = st.f_ffree;
205 buf->f_fsid[0] = st.f_fsid.val[0];
206 buf->f_fsid[1] = st.f_fsid.val[1];
207
208 return 0; 203 return 0;
209} 204}
210 205
@@ -212,35 +207,19 @@ static int do_statfs_hpux(struct path *path, struct hpux_statfs *buf)
212asmlinkage long hpux_statfs(const char __user *pathname, 207asmlinkage long hpux_statfs(const char __user *pathname,
213 struct hpux_statfs __user *buf) 208 struct hpux_statfs __user *buf)
214{ 209{
215 struct path path; 210 struct kstatfs st;
216 int error; 211 int error = user_statfs(pathname, &st);
217 212 if (!error)
218 error = user_path(pathname, &path); 213 error = do_statfs_hpux(&st, buf);
219 if (!error) {
220 struct hpux_statfs tmp;
221 error = do_statfs_hpux(&path, &tmp);
222 if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
223 error = -EFAULT;
224 path_put(&path);
225 }
226 return error; 214 return error;
227} 215}
228 216
229asmlinkage long hpux_fstatfs(unsigned int fd, struct hpux_statfs __user * buf) 217asmlinkage long hpux_fstatfs(unsigned int fd, struct hpux_statfs __user * buf)
230{ 218{
231 struct file *file; 219 struct kstatfs st;
232 struct hpux_statfs tmp; 220 int error = fd_statfs(fd, &st);
233 int error; 221 if (!error)
234 222 error = do_statfs_hpux(&st, buf);
235 error = -EBADF;
236 file = fget(fd);
237 if (!file)
238 goto out;
239 error = do_statfs_hpux(&file->f_path, &tmp);
240 if (!error && copy_to_user(buf, &tmp, sizeof(tmp)))
241 error = -EFAULT;
242 fput(file);
243 out:
244 return error; 223 return error;
245} 224}
246 225
diff --git a/arch/powerpc/include/asm/lppaca.h b/arch/powerpc/include/asm/lppaca.h
index 380d48bacd16..26b8c807f8f1 100644
--- a/arch/powerpc/include/asm/lppaca.h
+++ b/arch/powerpc/include/asm/lppaca.h
@@ -33,9 +33,25 @@
33// 33//
34//---------------------------------------------------------------------------- 34//----------------------------------------------------------------------------
35#include <linux/cache.h> 35#include <linux/cache.h>
36#include <linux/threads.h>
36#include <asm/types.h> 37#include <asm/types.h>
37#include <asm/mmu.h> 38#include <asm/mmu.h>
38 39
40/*
41 * We only have to have statically allocated lppaca structs on
42 * legacy iSeries, which supports at most 64 cpus.
43 */
44#ifdef CONFIG_PPC_ISERIES
45#if NR_CPUS < 64
46#define NR_LPPACAS NR_CPUS
47#else
48#define NR_LPPACAS 64
49#endif
50#else /* not iSeries */
51#define NR_LPPACAS 1
52#endif
53
54
39/* The Hypervisor barfs if the lppaca crosses a page boundary. A 1k 55/* The Hypervisor barfs if the lppaca crosses a page boundary. A 1k
40 * alignment is sufficient to prevent this */ 56 * alignment is sufficient to prevent this */
41struct lppaca { 57struct lppaca {
diff --git a/arch/powerpc/kernel/paca.c b/arch/powerpc/kernel/paca.c
index ebf9846f3c3b..f4adf89d7614 100644
--- a/arch/powerpc/kernel/paca.c
+++ b/arch/powerpc/kernel/paca.c
@@ -27,20 +27,6 @@ extern unsigned long __toc_start;
27#ifdef CONFIG_PPC_BOOK3S 27#ifdef CONFIG_PPC_BOOK3S
28 28
29/* 29/*
30 * We only have to have statically allocated lppaca structs on
31 * legacy iSeries, which supports at most 64 cpus.
32 */
33#ifdef CONFIG_PPC_ISERIES
34#if NR_CPUS < 64
35#define NR_LPPACAS NR_CPUS
36#else
37#define NR_LPPACAS 64
38#endif
39#else /* not iSeries */
40#define NR_LPPACAS 1
41#endif
42
43/*
44 * The structure which the hypervisor knows about - this structure 30 * The structure which the hypervisor knows about - this structure
45 * should not cross a page boundary. The vpa_init/register_vpa call 31 * should not cross a page boundary. The vpa_init/register_vpa call
46 * is now known to fail if the lppaca structure crosses a page 32 * is now known to fail if the lppaca structure crosses a page
diff --git a/arch/powerpc/mm/numa.c b/arch/powerpc/mm/numa.c
index fd4812329570..0dc95c0aa3be 100644
--- a/arch/powerpc/mm/numa.c
+++ b/arch/powerpc/mm/numa.c
@@ -1516,7 +1516,8 @@ int start_topology_update(void)
1516{ 1516{
1517 int rc = 0; 1517 int rc = 0;
1518 1518
1519 if (firmware_has_feature(FW_FEATURE_VPHN) && 1519 /* Disabled until races with load balancing are fixed */
1520 if (0 && firmware_has_feature(FW_FEATURE_VPHN) &&
1520 get_lppaca()->shared_proc) { 1521 get_lppaca()->shared_proc) {
1521 vphn_enabled = 1; 1522 vphn_enabled = 1;
1522 setup_cpu_associativity_change_counters(); 1523 setup_cpu_associativity_change_counters();
diff --git a/arch/powerpc/platforms/cell/spufs/syscalls.c b/arch/powerpc/platforms/cell/spufs/syscalls.c
index 187a7d32f86a..a3d2ce54ea2e 100644
--- a/arch/powerpc/platforms/cell/spufs/syscalls.c
+++ b/arch/powerpc/platforms/cell/spufs/syscalls.c
@@ -70,7 +70,7 @@ static long do_spu_create(const char __user *pathname, unsigned int flags,
70 if (!IS_ERR(tmp)) { 70 if (!IS_ERR(tmp)) {
71 struct nameidata nd; 71 struct nameidata nd;
72 72
73 ret = path_lookup(tmp, LOOKUP_PARENT, &nd); 73 ret = kern_path_parent(tmp, &nd);
74 if (!ret) { 74 if (!ret) {
75 nd.flags |= LOOKUP_OPEN | LOOKUP_CREATE; 75 nd.flags |= LOOKUP_OPEN | LOOKUP_CREATE;
76 ret = spufs_create(&nd, flags, mode, neighbor); 76 ret = spufs_create(&nd, flags, mode, neighbor);
diff --git a/arch/powerpc/platforms/iseries/dt.c b/arch/powerpc/platforms/iseries/dt.c
index fdb7384c0c4f..f0491cc28900 100644
--- a/arch/powerpc/platforms/iseries/dt.c
+++ b/arch/powerpc/platforms/iseries/dt.c
@@ -242,8 +242,8 @@ static void __init dt_cpus(struct iseries_flat_dt *dt)
242 pft_size[0] = 0; /* NUMA CEC cookie, 0 for non NUMA */ 242 pft_size[0] = 0; /* NUMA CEC cookie, 0 for non NUMA */
243 pft_size[1] = __ilog2(HvCallHpt_getHptPages() * HW_PAGE_SIZE); 243 pft_size[1] = __ilog2(HvCallHpt_getHptPages() * HW_PAGE_SIZE);
244 244
245 for (i = 0; i < NR_CPUS; i++) { 245 for (i = 0; i < NR_LPPACAS; i++) {
246 if (lppaca_of(i).dyn_proc_status >= 2) 246 if (lppaca[i].dyn_proc_status >= 2)
247 continue; 247 continue;
248 248
249 snprintf(p, 32 - (p - buf), "@%d", i); 249 snprintf(p, 32 - (p - buf), "@%d", i);
@@ -251,7 +251,7 @@ static void __init dt_cpus(struct iseries_flat_dt *dt)
251 251
252 dt_prop_str(dt, "device_type", device_type_cpu); 252 dt_prop_str(dt, "device_type", device_type_cpu);
253 253
254 index = lppaca_of(i).dyn_hv_phys_proc_index; 254 index = lppaca[i].dyn_hv_phys_proc_index;
255 d = &xIoHriProcessorVpd[index]; 255 d = &xIoHriProcessorVpd[index];
256 256
257 dt_prop_u32(dt, "i-cache-size", d->xInstCacheSize * 1024); 257 dt_prop_u32(dt, "i-cache-size", d->xInstCacheSize * 1024);
diff --git a/arch/powerpc/platforms/iseries/setup.c b/arch/powerpc/platforms/iseries/setup.c
index b0863410517f..2946ae10fbfd 100644
--- a/arch/powerpc/platforms/iseries/setup.c
+++ b/arch/powerpc/platforms/iseries/setup.c
@@ -680,6 +680,7 @@ void * __init iSeries_early_setup(void)
680 * on but calling this function multiple times is fine. 680 * on but calling this function multiple times is fine.
681 */ 681 */
682 identify_cpu(0, mfspr(SPRN_PVR)); 682 identify_cpu(0, mfspr(SPRN_PVR));
683 initialise_paca(&boot_paca, 0);
683 684
684 powerpc_firmware_features |= FW_FEATURE_ISERIES; 685 powerpc_firmware_features |= FW_FEATURE_ISERIES;
685 powerpc_firmware_features |= FW_FEATURE_LPAR; 686 powerpc_firmware_features |= FW_FEATURE_LPAR;
diff --git a/arch/um/drivers/mconsole_kern.c b/arch/um/drivers/mconsole_kern.c
index 975613b23dcf..c70e047eed72 100644
--- a/arch/um/drivers/mconsole_kern.c
+++ b/arch/um/drivers/mconsole_kern.c
@@ -124,35 +124,18 @@ void mconsole_log(struct mc_request *req)
124#if 0 124#if 0
125void mconsole_proc(struct mc_request *req) 125void mconsole_proc(struct mc_request *req)
126{ 126{
127 struct nameidata nd;
128 struct vfsmount *mnt = current->nsproxy->pid_ns->proc_mnt; 127 struct vfsmount *mnt = current->nsproxy->pid_ns->proc_mnt;
129 struct file *file; 128 struct file *file;
130 int n, err; 129 int n;
131 char *ptr = req->request.data, *buf; 130 char *ptr = req->request.data, *buf;
132 mm_segment_t old_fs = get_fs(); 131 mm_segment_t old_fs = get_fs();
133 132
134 ptr += strlen("proc"); 133 ptr += strlen("proc");
135 ptr = skip_spaces(ptr); 134 ptr = skip_spaces(ptr);
136 135
137 err = vfs_path_lookup(mnt->mnt_root, mnt, ptr, LOOKUP_FOLLOW, &nd); 136 file = file_open_root(mnt->mnt_root, mnt, ptr, O_RDONLY);
138 if (err) {
139 mconsole_reply(req, "Failed to look up file", 1, 0);
140 goto out;
141 }
142
143 err = may_open(&nd.path, MAY_READ, O_RDONLY);
144 if (result) {
145 mconsole_reply(req, "Failed to open file", 1, 0);
146 path_put(&nd.path);
147 goto out;
148 }
149
150 file = dentry_open(nd.path.dentry, nd.path.mnt, O_RDONLY,
151 current_cred());
152 err = PTR_ERR(file);
153 if (IS_ERR(file)) { 137 if (IS_ERR(file)) {
154 mconsole_reply(req, "Failed to open file", 1, 0); 138 mconsole_reply(req, "Failed to open file", 1, 0);
155 path_put(&nd.path);
156 goto out; 139 goto out;
157 } 140 }
158 141
diff --git a/arch/x86/boot/compressed/mkpiggy.c b/arch/x86/boot/compressed/mkpiggy.c
index 646aa78ba5fd..46a823882437 100644
--- a/arch/x86/boot/compressed/mkpiggy.c
+++ b/arch/x86/boot/compressed/mkpiggy.c
@@ -62,7 +62,12 @@ int main(int argc, char *argv[])
62 if (fseek(f, -4L, SEEK_END)) { 62 if (fseek(f, -4L, SEEK_END)) {
63 perror(argv[1]); 63 perror(argv[1]);
64 } 64 }
65 fread(&olen, sizeof olen, 1, f); 65
66 if (fread(&olen, sizeof(olen), 1, f) != 1) {
67 perror(argv[1]);
68 return 1;
69 }
70
66 ilen = ftell(f); 71 ilen = ftell(f);
67 olen = getle32(&olen); 72 olen = getle32(&olen);
68 fclose(f); 73 fclose(f);
diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S
index 518bb99c3394..98d353edfff3 100644
--- a/arch/x86/ia32/ia32entry.S
+++ b/arch/x86/ia32/ia32entry.S
@@ -851,4 +851,6 @@ ia32_sys_call_table:
851 .quad sys_fanotify_init 851 .quad sys_fanotify_init
852 .quad sys32_fanotify_mark 852 .quad sys32_fanotify_mark
853 .quad sys_prlimit64 /* 340 */ 853 .quad sys_prlimit64 /* 340 */
854 .quad sys_name_to_handle_at
855 .quad compat_sys_open_by_handle_at
854ia32_syscall_end: 856ia32_syscall_end:
diff --git a/arch/x86/include/asm/ce4100.h b/arch/x86/include/asm/ce4100.h
new file mode 100644
index 000000000000..e656ad8c0a2e
--- /dev/null
+++ b/arch/x86/include/asm/ce4100.h
@@ -0,0 +1,6 @@
1#ifndef _ASM_CE4100_H_
2#define _ASM_CE4100_H_
3
4int ce4100_pci_init(void);
5
6#endif
diff --git a/arch/x86/include/asm/unistd_32.h b/arch/x86/include/asm/unistd_32.h
index b766a5e8ba0e..f4c4973fc2ac 100644
--- a/arch/x86/include/asm/unistd_32.h
+++ b/arch/x86/include/asm/unistd_32.h
@@ -346,10 +346,12 @@
346#define __NR_fanotify_init 338 346#define __NR_fanotify_init 338
347#define __NR_fanotify_mark 339 347#define __NR_fanotify_mark 339
348#define __NR_prlimit64 340 348#define __NR_prlimit64 340
349#define __NR_name_to_handle_at 341
350#define __NR_open_by_handle_at 342
349 351
350#ifdef __KERNEL__ 352#ifdef __KERNEL__
351 353
352#define NR_syscalls 341 354#define NR_syscalls 343
353 355
354#define __ARCH_WANT_IPC_PARSE_VERSION 356#define __ARCH_WANT_IPC_PARSE_VERSION
355#define __ARCH_WANT_OLD_READDIR 357#define __ARCH_WANT_OLD_READDIR
diff --git a/arch/x86/include/asm/unistd_64.h b/arch/x86/include/asm/unistd_64.h
index 363e9b8a715b..81a3d5b70235 100644
--- a/arch/x86/include/asm/unistd_64.h
+++ b/arch/x86/include/asm/unistd_64.h
@@ -669,6 +669,10 @@ __SYSCALL(__NR_fanotify_init, sys_fanotify_init)
669__SYSCALL(__NR_fanotify_mark, sys_fanotify_mark) 669__SYSCALL(__NR_fanotify_mark, sys_fanotify_mark)
670#define __NR_prlimit64 302 670#define __NR_prlimit64 302
671__SYSCALL(__NR_prlimit64, sys_prlimit64) 671__SYSCALL(__NR_prlimit64, sys_prlimit64)
672#define __NR_name_to_handle_at 303
673__SYSCALL(__NR_name_to_handle_at, sys_name_to_handle_at)
674#define __NR_open_by_handle_at 304
675__SYSCALL(__NR_open_by_handle_at, sys_open_by_handle_at)
672 676
673#ifndef __NO_STUBS 677#ifndef __NO_STUBS
674#define __ARCH_WANT_OLD_READDIR 678#define __ARCH_WANT_OLD_READDIR
diff --git a/arch/x86/include/asm/uv/uv_bau.h b/arch/x86/include/asm/uv/uv_bau.h
index ce1d54c8a433..3e094af443c3 100644
--- a/arch/x86/include/asm/uv/uv_bau.h
+++ b/arch/x86/include/asm/uv/uv_bau.h
@@ -176,7 +176,7 @@ struct bau_msg_payload {
176struct bau_msg_header { 176struct bau_msg_header {
177 unsigned int dest_subnodeid:6; /* must be 0x10, for the LB */ 177 unsigned int dest_subnodeid:6; /* must be 0x10, for the LB */
178 /* bits 5:0 */ 178 /* bits 5:0 */
179 unsigned int base_dest_nodeid:15; /* nasid (pnode<<1) of */ 179 unsigned int base_dest_nodeid:15; /* nasid of the */
180 /* bits 20:6 */ /* first bit in uvhub map */ 180 /* bits 20:6 */ /* first bit in uvhub map */
181 unsigned int command:8; /* message type */ 181 unsigned int command:8; /* message type */
182 /* bits 28:21 */ 182 /* bits 28:21 */
diff --git a/arch/x86/include/asm/xen/hypercall.h b/arch/x86/include/asm/xen/hypercall.h
index a3c28ae4025b..8508bfe52296 100644
--- a/arch/x86/include/asm/xen/hypercall.h
+++ b/arch/x86/include/asm/xen/hypercall.h
@@ -287,7 +287,7 @@ HYPERVISOR_fpu_taskswitch(int set)
287static inline int 287static inline int
288HYPERVISOR_sched_op(int cmd, void *arg) 288HYPERVISOR_sched_op(int cmd, void *arg)
289{ 289{
290 return _hypercall2(int, sched_op_new, cmd, arg); 290 return _hypercall2(int, sched_op, cmd, arg);
291} 291}
292 292
293static inline long 293static inline long
@@ -422,10 +422,17 @@ HYPERVISOR_set_segment_base(int reg, unsigned long value)
422#endif 422#endif
423 423
424static inline int 424static inline int
425HYPERVISOR_suspend(unsigned long srec) 425HYPERVISOR_suspend(unsigned long start_info_mfn)
426{ 426{
427 return _hypercall3(int, sched_op, SCHEDOP_shutdown, 427 struct sched_shutdown r = { .reason = SHUTDOWN_suspend };
428 SHUTDOWN_suspend, srec); 428
429 /*
430 * For a PV guest the tools require that the start_info mfn be
431 * present in rdx/edx when the hypercall is made. Per the
432 * hypercall calling convention this is the third hypercall
433 * argument, which is start_info_mfn here.
434 */
435 return _hypercall3(int, sched_op, SCHEDOP_shutdown, &r, start_info_mfn);
429} 436}
430 437
431static inline int 438static inline int
diff --git a/arch/x86/include/asm/xen/page.h b/arch/x86/include/asm/xen/page.h
index f25bdf238a33..c61934fbf22a 100644
--- a/arch/x86/include/asm/xen/page.h
+++ b/arch/x86/include/asm/xen/page.h
@@ -29,8 +29,10 @@ typedef struct xpaddr {
29 29
30/**** MACHINE <-> PHYSICAL CONVERSION MACROS ****/ 30/**** MACHINE <-> PHYSICAL CONVERSION MACROS ****/
31#define INVALID_P2M_ENTRY (~0UL) 31#define INVALID_P2M_ENTRY (~0UL)
32#define FOREIGN_FRAME_BIT (1UL<<31) 32#define FOREIGN_FRAME_BIT (1UL<<(BITS_PER_LONG-1))
33#define IDENTITY_FRAME_BIT (1UL<<(BITS_PER_LONG-2))
33#define FOREIGN_FRAME(m) ((m) | FOREIGN_FRAME_BIT) 34#define FOREIGN_FRAME(m) ((m) | FOREIGN_FRAME_BIT)
35#define IDENTITY_FRAME(m) ((m) | IDENTITY_FRAME_BIT)
34 36
35/* Maximum amount of memory we can handle in a domain in pages */ 37/* Maximum amount of memory we can handle in a domain in pages */
36#define MAX_DOMAIN_PAGES \ 38#define MAX_DOMAIN_PAGES \
@@ -41,12 +43,18 @@ extern unsigned int machine_to_phys_order;
41 43
42extern unsigned long get_phys_to_machine(unsigned long pfn); 44extern unsigned long get_phys_to_machine(unsigned long pfn);
43extern bool set_phys_to_machine(unsigned long pfn, unsigned long mfn); 45extern bool set_phys_to_machine(unsigned long pfn, unsigned long mfn);
46extern bool __set_phys_to_machine(unsigned long pfn, unsigned long mfn);
47extern unsigned long set_phys_range_identity(unsigned long pfn_s,
48 unsigned long pfn_e);
44 49
45extern int m2p_add_override(unsigned long mfn, struct page *page); 50extern int m2p_add_override(unsigned long mfn, struct page *page);
46extern int m2p_remove_override(struct page *page); 51extern int m2p_remove_override(struct page *page);
47extern struct page *m2p_find_override(unsigned long mfn); 52extern struct page *m2p_find_override(unsigned long mfn);
48extern unsigned long m2p_find_override_pfn(unsigned long mfn, unsigned long pfn); 53extern unsigned long m2p_find_override_pfn(unsigned long mfn, unsigned long pfn);
49 54
55#ifdef CONFIG_XEN_DEBUG_FS
56extern int p2m_dump_show(struct seq_file *m, void *v);
57#endif
50static inline unsigned long pfn_to_mfn(unsigned long pfn) 58static inline unsigned long pfn_to_mfn(unsigned long pfn)
51{ 59{
52 unsigned long mfn; 60 unsigned long mfn;
@@ -57,7 +65,7 @@ static inline unsigned long pfn_to_mfn(unsigned long pfn)
57 mfn = get_phys_to_machine(pfn); 65 mfn = get_phys_to_machine(pfn);
58 66
59 if (mfn != INVALID_P2M_ENTRY) 67 if (mfn != INVALID_P2M_ENTRY)
60 mfn &= ~FOREIGN_FRAME_BIT; 68 mfn &= ~(FOREIGN_FRAME_BIT | IDENTITY_FRAME_BIT);
61 69
62 return mfn; 70 return mfn;
63} 71}
@@ -73,25 +81,44 @@ static inline int phys_to_machine_mapping_valid(unsigned long pfn)
73static inline unsigned long mfn_to_pfn(unsigned long mfn) 81static inline unsigned long mfn_to_pfn(unsigned long mfn)
74{ 82{
75 unsigned long pfn; 83 unsigned long pfn;
84 int ret = 0;
76 85
77 if (xen_feature(XENFEAT_auto_translated_physmap)) 86 if (xen_feature(XENFEAT_auto_translated_physmap))
78 return mfn; 87 return mfn;
79 88
89 if (unlikely((mfn >> machine_to_phys_order) != 0)) {
90 pfn = ~0;
91 goto try_override;
92 }
80 pfn = 0; 93 pfn = 0;
81 /* 94 /*
82 * The array access can fail (e.g., device space beyond end of RAM). 95 * The array access can fail (e.g., device space beyond end of RAM).
83 * In such cases it doesn't matter what we return (we return garbage), 96 * In such cases it doesn't matter what we return (we return garbage),
84 * but we must handle the fault without crashing! 97 * but we must handle the fault without crashing!
85 */ 98 */
86 __get_user(pfn, &machine_to_phys_mapping[mfn]); 99 ret = __get_user(pfn, &machine_to_phys_mapping[mfn]);
87 100try_override:
88 /* 101 /* ret might be < 0 if there are no entries in the m2p for mfn */
89 * If this appears to be a foreign mfn (because the pfn 102 if (ret < 0)
90 * doesn't map back to the mfn), then check the local override 103 pfn = ~0;
91 * table to see if there's a better pfn to use. 104 else if (get_phys_to_machine(pfn) != mfn)
105 /*
106 * If this appears to be a foreign mfn (because the pfn
107 * doesn't map back to the mfn), then check the local override
108 * table to see if there's a better pfn to use.
109 *
110 * m2p_find_override_pfn returns ~0 if it doesn't find anything.
111 */
112 pfn = m2p_find_override_pfn(mfn, ~0);
113
114 /*
115 * pfn is ~0 if there are no entries in the m2p for mfn or if the
116 * entry doesn't map back to the mfn and m2p_override doesn't have a
117 * valid entry for it.
92 */ 118 */
93 if (get_phys_to_machine(pfn) != mfn) 119 if (pfn == ~0 &&
94 pfn = m2p_find_override_pfn(mfn, pfn); 120 get_phys_to_machine(mfn) == IDENTITY_FRAME(mfn))
121 pfn = mfn;
95 122
96 return pfn; 123 return pfn;
97} 124}
diff --git a/arch/x86/include/asm/xen/pci.h b/arch/x86/include/asm/xen/pci.h
index 2329b3eaf8d3..aa8620989162 100644
--- a/arch/x86/include/asm/xen/pci.h
+++ b/arch/x86/include/asm/xen/pci.h
@@ -27,16 +27,16 @@ static inline void __init xen_setup_pirqs(void)
27 * its own functions. 27 * its own functions.
28 */ 28 */
29struct xen_pci_frontend_ops { 29struct xen_pci_frontend_ops {
30 int (*enable_msi)(struct pci_dev *dev, int **vectors); 30 int (*enable_msi)(struct pci_dev *dev, int vectors[]);
31 void (*disable_msi)(struct pci_dev *dev); 31 void (*disable_msi)(struct pci_dev *dev);
32 int (*enable_msix)(struct pci_dev *dev, int **vectors, int nvec); 32 int (*enable_msix)(struct pci_dev *dev, int vectors[], int nvec);
33 void (*disable_msix)(struct pci_dev *dev); 33 void (*disable_msix)(struct pci_dev *dev);
34}; 34};
35 35
36extern struct xen_pci_frontend_ops *xen_pci_frontend; 36extern struct xen_pci_frontend_ops *xen_pci_frontend;
37 37
38static inline int xen_pci_frontend_enable_msi(struct pci_dev *dev, 38static inline int xen_pci_frontend_enable_msi(struct pci_dev *dev,
39 int **vectors) 39 int vectors[])
40{ 40{
41 if (xen_pci_frontend && xen_pci_frontend->enable_msi) 41 if (xen_pci_frontend && xen_pci_frontend->enable_msi)
42 return xen_pci_frontend->enable_msi(dev, vectors); 42 return xen_pci_frontend->enable_msi(dev, vectors);
@@ -48,7 +48,7 @@ static inline void xen_pci_frontend_disable_msi(struct pci_dev *dev)
48 xen_pci_frontend->disable_msi(dev); 48 xen_pci_frontend->disable_msi(dev);
49} 49}
50static inline int xen_pci_frontend_enable_msix(struct pci_dev *dev, 50static inline int xen_pci_frontend_enable_msix(struct pci_dev *dev,
51 int **vectors, int nvec) 51 int vectors[], int nvec)
52{ 52{
53 if (xen_pci_frontend && xen_pci_frontend->enable_msix) 53 if (xen_pci_frontend && xen_pci_frontend->enable_msix)
54 return xen_pci_frontend->enable_msix(dev, vectors, nvec); 54 return xen_pci_frontend->enable_msix(dev, vectors, nvec);
diff --git a/arch/x86/kernel/check.c b/arch/x86/kernel/check.c
index 13a389179514..452932d34730 100644
--- a/arch/x86/kernel/check.c
+++ b/arch/x86/kernel/check.c
@@ -106,8 +106,8 @@ void __init setup_bios_corruption_check(void)
106 addr += size; 106 addr += size;
107 } 107 }
108 108
109 printk(KERN_INFO "Scanning %d areas for low memory corruption\n", 109 if (num_scan_areas)
110 num_scan_areas); 110 printk(KERN_INFO "Scanning %d areas for low memory corruption\n", num_scan_areas);
111} 111}
112 112
113 113
@@ -143,12 +143,12 @@ static void check_corruption(struct work_struct *dummy)
143{ 143{
144 check_for_bios_corruption(); 144 check_for_bios_corruption();
145 schedule_delayed_work(&bios_check_work, 145 schedule_delayed_work(&bios_check_work,
146 round_jiffies_relative(corruption_check_period*HZ)); 146 round_jiffies_relative(corruption_check_period*HZ));
147} 147}
148 148
149static int start_periodic_check_for_corruption(void) 149static int start_periodic_check_for_corruption(void)
150{ 150{
151 if (!memory_corruption_check || corruption_check_period == 0) 151 if (!num_scan_areas || !memory_corruption_check || corruption_check_period == 0)
152 return 0; 152 return 0;
153 153
154 printk(KERN_INFO "Scanning for low memory corruption every %d seconds\n", 154 printk(KERN_INFO "Scanning for low memory corruption every %d seconds\n",
diff --git a/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
index 4f6f679f2799..4a5a42b842ad 100644
--- a/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
@@ -195,7 +195,7 @@ static unsigned int pcc_get_freq(unsigned int cpu)
195cmd_incomplete: 195cmd_incomplete:
196 iowrite16(0, &pcch_hdr->status); 196 iowrite16(0, &pcch_hdr->status);
197 spin_unlock(&pcc_lock); 197 spin_unlock(&pcc_lock);
198 return -EINVAL; 198 return 0;
199} 199}
200 200
201static int pcc_cpufreq_target(struct cpufreq_policy *policy, 201static int pcc_cpufreq_target(struct cpufreq_policy *policy,
diff --git a/arch/x86/kernel/syscall_table_32.S b/arch/x86/kernel/syscall_table_32.S
index b35786dc9b8f..c314b2199efd 100644
--- a/arch/x86/kernel/syscall_table_32.S
+++ b/arch/x86/kernel/syscall_table_32.S
@@ -340,3 +340,5 @@ ENTRY(sys_call_table)
340 .long sys_fanotify_init 340 .long sys_fanotify_init
341 .long sys_fanotify_mark 341 .long sys_fanotify_mark
342 .long sys_prlimit64 /* 340 */ 342 .long sys_prlimit64 /* 340 */
343 .long sys_name_to_handle_at
344 .long sys_open_by_handle_at
diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 7d90ceb882a4..20e3f8702d1e 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -229,15 +229,14 @@ void vmalloc_sync_all(void)
229 for (address = VMALLOC_START & PMD_MASK; 229 for (address = VMALLOC_START & PMD_MASK;
230 address >= TASK_SIZE && address < FIXADDR_TOP; 230 address >= TASK_SIZE && address < FIXADDR_TOP;
231 address += PMD_SIZE) { 231 address += PMD_SIZE) {
232
233 unsigned long flags;
234 struct page *page; 232 struct page *page;
235 233
236 spin_lock_irqsave(&pgd_lock, flags); 234 spin_lock(&pgd_lock);
237 list_for_each_entry(page, &pgd_list, lru) { 235 list_for_each_entry(page, &pgd_list, lru) {
238 spinlock_t *pgt_lock; 236 spinlock_t *pgt_lock;
239 pmd_t *ret; 237 pmd_t *ret;
240 238
239 /* the pgt_lock only for Xen */
241 pgt_lock = &pgd_page_get_mm(page)->page_table_lock; 240 pgt_lock = &pgd_page_get_mm(page)->page_table_lock;
242 241
243 spin_lock(pgt_lock); 242 spin_lock(pgt_lock);
@@ -247,7 +246,7 @@ void vmalloc_sync_all(void)
247 if (!ret) 246 if (!ret)
248 break; 247 break;
249 } 248 }
250 spin_unlock_irqrestore(&pgd_lock, flags); 249 spin_unlock(&pgd_lock);
251 } 250 }
252} 251}
253 252
@@ -828,6 +827,13 @@ mm_fault_error(struct pt_regs *regs, unsigned long error_code,
828 unsigned long address, unsigned int fault) 827 unsigned long address, unsigned int fault)
829{ 828{
830 if (fault & VM_FAULT_OOM) { 829 if (fault & VM_FAULT_OOM) {
830 /* Kernel mode? Handle exceptions or die: */
831 if (!(error_code & PF_USER)) {
832 up_read(&current->mm->mmap_sem);
833 no_context(regs, error_code, address);
834 return;
835 }
836
831 out_of_memory(regs, error_code, address); 837 out_of_memory(regs, error_code, address);
832 } else { 838 } else {
833 if (fault & (VM_FAULT_SIGBUS|VM_FAULT_HWPOISON| 839 if (fault & (VM_FAULT_SIGBUS|VM_FAULT_HWPOISON|
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 71a59296af80..c14a5422e152 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -105,18 +105,18 @@ void sync_global_pgds(unsigned long start, unsigned long end)
105 105
106 for (address = start; address <= end; address += PGDIR_SIZE) { 106 for (address = start; address <= end; address += PGDIR_SIZE) {
107 const pgd_t *pgd_ref = pgd_offset_k(address); 107 const pgd_t *pgd_ref = pgd_offset_k(address);
108 unsigned long flags;
109 struct page *page; 108 struct page *page;
110 109
111 if (pgd_none(*pgd_ref)) 110 if (pgd_none(*pgd_ref))
112 continue; 111 continue;
113 112
114 spin_lock_irqsave(&pgd_lock, flags); 113 spin_lock(&pgd_lock);
115 list_for_each_entry(page, &pgd_list, lru) { 114 list_for_each_entry(page, &pgd_list, lru) {
116 pgd_t *pgd; 115 pgd_t *pgd;
117 spinlock_t *pgt_lock; 116 spinlock_t *pgt_lock;
118 117
119 pgd = (pgd_t *)page_address(page) + pgd_index(address); 118 pgd = (pgd_t *)page_address(page) + pgd_index(address);
119 /* the pgt_lock only for Xen */
120 pgt_lock = &pgd_page_get_mm(page)->page_table_lock; 120 pgt_lock = &pgd_page_get_mm(page)->page_table_lock;
121 spin_lock(pgt_lock); 121 spin_lock(pgt_lock);
122 122
@@ -128,7 +128,7 @@ void sync_global_pgds(unsigned long start, unsigned long end)
128 128
129 spin_unlock(pgt_lock); 129 spin_unlock(pgt_lock);
130 } 130 }
131 spin_unlock_irqrestore(&pgd_lock, flags); 131 spin_unlock(&pgd_lock);
132 } 132 }
133} 133}
134 134
diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c
index 95ea1551eebc..1337c51b07d7 100644
--- a/arch/x86/mm/numa_64.c
+++ b/arch/x86/mm/numa_64.c
@@ -780,11 +780,7 @@ void __cpuinit numa_add_cpu(int cpu)
780 int physnid; 780 int physnid;
781 int nid = NUMA_NO_NODE; 781 int nid = NUMA_NO_NODE;
782 782
783 apicid = early_per_cpu(x86_cpu_to_apicid, cpu); 783 nid = early_cpu_to_node(cpu);
784 if (apicid != BAD_APICID)
785 nid = apicid_to_node[apicid];
786 if (nid == NUMA_NO_NODE)
787 nid = early_cpu_to_node(cpu);
788 BUG_ON(nid == NUMA_NO_NODE || !node_online(nid)); 784 BUG_ON(nid == NUMA_NO_NODE || !node_online(nid));
789 785
790 /* 786 /*
diff --git a/arch/x86/mm/pageattr.c b/arch/x86/mm/pageattr.c
index d343b3c81f3c..90825f2eb0f4 100644
--- a/arch/x86/mm/pageattr.c
+++ b/arch/x86/mm/pageattr.c
@@ -57,12 +57,10 @@ static unsigned long direct_pages_count[PG_LEVEL_NUM];
57 57
58void update_page_count(int level, unsigned long pages) 58void update_page_count(int level, unsigned long pages)
59{ 59{
60 unsigned long flags;
61
62 /* Protect against CPA */ 60 /* Protect against CPA */
63 spin_lock_irqsave(&pgd_lock, flags); 61 spin_lock(&pgd_lock);
64 direct_pages_count[level] += pages; 62 direct_pages_count[level] += pages;
65 spin_unlock_irqrestore(&pgd_lock, flags); 63 spin_unlock(&pgd_lock);
66} 64}
67 65
68static void split_page_count(int level) 66static void split_page_count(int level)
@@ -394,7 +392,7 @@ static int
394try_preserve_large_page(pte_t *kpte, unsigned long address, 392try_preserve_large_page(pte_t *kpte, unsigned long address,
395 struct cpa_data *cpa) 393 struct cpa_data *cpa)
396{ 394{
397 unsigned long nextpage_addr, numpages, pmask, psize, flags, addr, pfn; 395 unsigned long nextpage_addr, numpages, pmask, psize, addr, pfn;
398 pte_t new_pte, old_pte, *tmp; 396 pte_t new_pte, old_pte, *tmp;
399 pgprot_t old_prot, new_prot, req_prot; 397 pgprot_t old_prot, new_prot, req_prot;
400 int i, do_split = 1; 398 int i, do_split = 1;
@@ -403,7 +401,7 @@ try_preserve_large_page(pte_t *kpte, unsigned long address,
403 if (cpa->force_split) 401 if (cpa->force_split)
404 return 1; 402 return 1;
405 403
406 spin_lock_irqsave(&pgd_lock, flags); 404 spin_lock(&pgd_lock);
407 /* 405 /*
408 * Check for races, another CPU might have split this page 406 * Check for races, another CPU might have split this page
409 * up already: 407 * up already:
@@ -498,14 +496,14 @@ try_preserve_large_page(pte_t *kpte, unsigned long address,
498 } 496 }
499 497
500out_unlock: 498out_unlock:
501 spin_unlock_irqrestore(&pgd_lock, flags); 499 spin_unlock(&pgd_lock);
502 500
503 return do_split; 501 return do_split;
504} 502}
505 503
506static int split_large_page(pte_t *kpte, unsigned long address) 504static int split_large_page(pte_t *kpte, unsigned long address)
507{ 505{
508 unsigned long flags, pfn, pfninc = 1; 506 unsigned long pfn, pfninc = 1;
509 unsigned int i, level; 507 unsigned int i, level;
510 pte_t *pbase, *tmp; 508 pte_t *pbase, *tmp;
511 pgprot_t ref_prot; 509 pgprot_t ref_prot;
@@ -519,7 +517,7 @@ static int split_large_page(pte_t *kpte, unsigned long address)
519 if (!base) 517 if (!base)
520 return -ENOMEM; 518 return -ENOMEM;
521 519
522 spin_lock_irqsave(&pgd_lock, flags); 520 spin_lock(&pgd_lock);
523 /* 521 /*
524 * Check for races, another CPU might have split this page 522 * Check for races, another CPU might have split this page
525 * up for us already: 523 * up for us already:
@@ -591,7 +589,7 @@ out_unlock:
591 */ 589 */
592 if (base) 590 if (base)
593 __free_page(base); 591 __free_page(base);
594 spin_unlock_irqrestore(&pgd_lock, flags); 592 spin_unlock(&pgd_lock);
595 593
596 return 0; 594 return 0;
597} 595}
diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
index 500242d3c96d..0113d19c8aa6 100644
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -121,14 +121,12 @@ static void pgd_ctor(struct mm_struct *mm, pgd_t *pgd)
121 121
122static void pgd_dtor(pgd_t *pgd) 122static void pgd_dtor(pgd_t *pgd)
123{ 123{
124 unsigned long flags; /* can be called from interrupt context */
125
126 if (SHARED_KERNEL_PMD) 124 if (SHARED_KERNEL_PMD)
127 return; 125 return;
128 126
129 spin_lock_irqsave(&pgd_lock, flags); 127 spin_lock(&pgd_lock);
130 pgd_list_del(pgd); 128 pgd_list_del(pgd);
131 spin_unlock_irqrestore(&pgd_lock, flags); 129 spin_unlock(&pgd_lock);
132} 130}
133 131
134/* 132/*
@@ -260,7 +258,6 @@ pgd_t *pgd_alloc(struct mm_struct *mm)
260{ 258{
261 pgd_t *pgd; 259 pgd_t *pgd;
262 pmd_t *pmds[PREALLOCATED_PMDS]; 260 pmd_t *pmds[PREALLOCATED_PMDS];
263 unsigned long flags;
264 261
265 pgd = (pgd_t *)__get_free_page(PGALLOC_GFP); 262 pgd = (pgd_t *)__get_free_page(PGALLOC_GFP);
266 263
@@ -280,12 +277,12 @@ pgd_t *pgd_alloc(struct mm_struct *mm)
280 * respect to anything walking the pgd_list, so that they 277 * respect to anything walking the pgd_list, so that they
281 * never see a partially populated pgd. 278 * never see a partially populated pgd.
282 */ 279 */
283 spin_lock_irqsave(&pgd_lock, flags); 280 spin_lock(&pgd_lock);
284 281
285 pgd_ctor(mm, pgd); 282 pgd_ctor(mm, pgd);
286 pgd_prepopulate_pmd(mm, pgd, pmds); 283 pgd_prepopulate_pmd(mm, pgd, pmds);
287 284
288 spin_unlock_irqrestore(&pgd_lock, flags); 285 spin_unlock(&pgd_lock);
289 286
290 return pgd; 287 return pgd;
291 288
diff --git a/arch/x86/pci/ce4100.c b/arch/x86/pci/ce4100.c
index 85b68ef5e809..9260b3eb18d4 100644
--- a/arch/x86/pci/ce4100.c
+++ b/arch/x86/pci/ce4100.c
@@ -34,6 +34,7 @@
34#include <linux/pci.h> 34#include <linux/pci.h>
35#include <linux/init.h> 35#include <linux/init.h>
36 36
37#include <asm/ce4100.h>
37#include <asm/pci_x86.h> 38#include <asm/pci_x86.h>
38 39
39struct sim_reg { 40struct sim_reg {
@@ -306,10 +307,10 @@ struct pci_raw_ops ce4100_pci_conf = {
306 .write = ce4100_conf_write, 307 .write = ce4100_conf_write,
307}; 308};
308 309
309static int __init ce4100_pci_init(void) 310int __init ce4100_pci_init(void)
310{ 311{
311 init_sim_regs(); 312 init_sim_regs();
312 raw_pci_ops = &ce4100_pci_conf; 313 raw_pci_ops = &ce4100_pci_conf;
313 return 0; 314 /* Indicate caller that it should invoke pci_legacy_init() */
315 return 1;
314} 316}
315subsys_initcall(ce4100_pci_init);
diff --git a/arch/x86/pci/xen.c b/arch/x86/pci/xen.c
index 25cd4a07d09f..8c4085a95ef1 100644
--- a/arch/x86/pci/xen.c
+++ b/arch/x86/pci/xen.c
@@ -20,7 +20,8 @@
20#include <asm/xen/pci.h> 20#include <asm/xen/pci.h>
21 21
22#ifdef CONFIG_ACPI 22#ifdef CONFIG_ACPI
23static int xen_hvm_register_pirq(u32 gsi, int triggering) 23static int acpi_register_gsi_xen_hvm(struct device *dev, u32 gsi,
24 int trigger, int polarity)
24{ 25{
25 int rc, irq; 26 int rc, irq;
26 struct physdev_map_pirq map_irq; 27 struct physdev_map_pirq map_irq;
@@ -41,7 +42,7 @@ static int xen_hvm_register_pirq(u32 gsi, int triggering)
41 return -1; 42 return -1;
42 } 43 }
43 44
44 if (triggering == ACPI_EDGE_SENSITIVE) { 45 if (trigger == ACPI_EDGE_SENSITIVE) {
45 shareable = 0; 46 shareable = 0;
46 name = "ioapic-edge"; 47 name = "ioapic-edge";
47 } else { 48 } else {
@@ -55,12 +56,6 @@ static int xen_hvm_register_pirq(u32 gsi, int triggering)
55 56
56 return irq; 57 return irq;
57} 58}
58
59static int acpi_register_gsi_xen_hvm(struct device *dev, u32 gsi,
60 int trigger, int polarity)
61{
62 return xen_hvm_register_pirq(gsi, trigger);
63}
64#endif 59#endif
65 60
66#if defined(CONFIG_PCI_MSI) 61#if defined(CONFIG_PCI_MSI)
@@ -91,7 +86,7 @@ static void xen_msi_compose_msg(struct pci_dev *pdev, unsigned int pirq,
91 86
92static int xen_hvm_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) 87static int xen_hvm_setup_msi_irqs(struct pci_dev *dev, int nvec, int type)
93{ 88{
94 int irq, pirq, ret = 0; 89 int irq, pirq;
95 struct msi_desc *msidesc; 90 struct msi_desc *msidesc;
96 struct msi_msg msg; 91 struct msi_msg msg;
97 92
@@ -99,39 +94,32 @@ static int xen_hvm_setup_msi_irqs(struct pci_dev *dev, int nvec, int type)
99 __read_msi_msg(msidesc, &msg); 94 __read_msi_msg(msidesc, &msg);
100 pirq = MSI_ADDR_EXT_DEST_ID(msg.address_hi) | 95 pirq = MSI_ADDR_EXT_DEST_ID(msg.address_hi) |
101 ((msg.address_lo >> MSI_ADDR_DEST_ID_SHIFT) & 0xff); 96 ((msg.address_lo >> MSI_ADDR_DEST_ID_SHIFT) & 0xff);
102 if (xen_irq_from_pirq(pirq) >= 0 && msg.data == XEN_PIRQ_MSI_DATA) { 97 if (msg.data != XEN_PIRQ_MSI_DATA ||
103 xen_allocate_pirq_msi((type == PCI_CAP_ID_MSIX) ? 98 xen_irq_from_pirq(pirq) < 0) {
104 "msi-x" : "msi", &irq, &pirq, XEN_ALLOC_IRQ); 99 pirq = xen_allocate_pirq_msi(dev, msidesc);
105 if (irq < 0) 100 if (pirq < 0)
106 goto error; 101 goto error;
107 ret = set_irq_msi(irq, msidesc); 102 xen_msi_compose_msg(dev, pirq, &msg);
108 if (ret < 0) 103 __write_msi_msg(msidesc, &msg);
109 goto error_while; 104 dev_dbg(&dev->dev, "xen: msi bound to pirq=%d\n", pirq);
110 printk(KERN_DEBUG "xen: msi already setup: msi --> irq=%d" 105 } else {
111 " pirq=%d\n", irq, pirq); 106 dev_dbg(&dev->dev,
112 return 0; 107 "xen: msi already bound to pirq=%d\n", pirq);
113 } 108 }
114 xen_allocate_pirq_msi((type == PCI_CAP_ID_MSIX) ? 109 irq = xen_bind_pirq_msi_to_irq(dev, msidesc, pirq, 0,
115 "msi-x" : "msi", &irq, &pirq, (XEN_ALLOC_IRQ | XEN_ALLOC_PIRQ)); 110 (type == PCI_CAP_ID_MSIX) ?
116 if (irq < 0 || pirq < 0) 111 "msi-x" : "msi");
112 if (irq < 0)
117 goto error; 113 goto error;
118 printk(KERN_DEBUG "xen: msi --> irq=%d, pirq=%d\n", irq, pirq); 114 dev_dbg(&dev->dev,
119 xen_msi_compose_msg(dev, pirq, &msg); 115 "xen: msi --> pirq=%d --> irq=%d\n", pirq, irq);
120 ret = set_irq_msi(irq, msidesc);
121 if (ret < 0)
122 goto error_while;
123 write_msi_msg(irq, &msg);
124 } 116 }
125 return 0; 117 return 0;
126 118
127error_while:
128 unbind_from_irqhandler(irq, NULL);
129error: 119error:
130 if (ret == -ENODEV) 120 dev_err(&dev->dev,
131 dev_err(&dev->dev, "Xen PCI frontend has not registered" \ 121 "Xen PCI frontend has not registered MSI/MSI-X support!\n");
132 " MSI/MSI-X support!\n"); 122 return -ENODEV;
133
134 return ret;
135} 123}
136 124
137/* 125/*
@@ -150,35 +138,26 @@ static int xen_setup_msi_irqs(struct pci_dev *dev, int nvec, int type)
150 return -ENOMEM; 138 return -ENOMEM;
151 139
152 if (type == PCI_CAP_ID_MSIX) 140 if (type == PCI_CAP_ID_MSIX)
153 ret = xen_pci_frontend_enable_msix(dev, &v, nvec); 141 ret = xen_pci_frontend_enable_msix(dev, v, nvec);
154 else 142 else
155 ret = xen_pci_frontend_enable_msi(dev, &v); 143 ret = xen_pci_frontend_enable_msi(dev, v);
156 if (ret) 144 if (ret)
157 goto error; 145 goto error;
158 i = 0; 146 i = 0;
159 list_for_each_entry(msidesc, &dev->msi_list, list) { 147 list_for_each_entry(msidesc, &dev->msi_list, list) {
160 irq = xen_allocate_pirq(v[i], 0, /* not sharable */ 148 irq = xen_bind_pirq_msi_to_irq(dev, msidesc, v[i], 0,
161 (type == PCI_CAP_ID_MSIX) ? 149 (type == PCI_CAP_ID_MSIX) ?
162 "pcifront-msi-x" : "pcifront-msi"); 150 "pcifront-msi-x" :
163 if (irq < 0) { 151 "pcifront-msi");
164 ret = -1; 152 if (irq < 0)
165 goto free; 153 goto free;
166 }
167
168 ret = set_irq_msi(irq, msidesc);
169 if (ret)
170 goto error_while;
171 i++; 154 i++;
172 } 155 }
173 kfree(v); 156 kfree(v);
174 return 0; 157 return 0;
175 158
176error_while:
177 unbind_from_irqhandler(irq, NULL);
178error: 159error:
179 if (ret == -ENODEV) 160 dev_err(&dev->dev, "Xen PCI frontend has not registered MSI/MSI-X support!\n");
180 dev_err(&dev->dev, "Xen PCI frontend has not registered" \
181 " MSI/MSI-X support!\n");
182free: 161free:
183 kfree(v); 162 kfree(v);
184 return ret; 163 return ret;
@@ -193,6 +172,9 @@ static void xen_teardown_msi_irqs(struct pci_dev *dev)
193 xen_pci_frontend_disable_msix(dev); 172 xen_pci_frontend_disable_msix(dev);
194 else 173 else
195 xen_pci_frontend_disable_msi(dev); 174 xen_pci_frontend_disable_msi(dev);
175
176 /* Free the IRQ's and the msidesc using the generic code. */
177 default_teardown_msi_irqs(dev);
196} 178}
197 179
198static void xen_teardown_msi_irq(unsigned int irq) 180static void xen_teardown_msi_irq(unsigned int irq)
@@ -200,47 +182,82 @@ static void xen_teardown_msi_irq(unsigned int irq)
200 xen_destroy_irq(irq); 182 xen_destroy_irq(irq);
201} 183}
202 184
185#ifdef CONFIG_XEN_DOM0
203static int xen_initdom_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) 186static int xen_initdom_setup_msi_irqs(struct pci_dev *dev, int nvec, int type)
204{ 187{
205 int irq, ret; 188 int ret = 0;
206 struct msi_desc *msidesc; 189 struct msi_desc *msidesc;
207 190
208 list_for_each_entry(msidesc, &dev->msi_list, list) { 191 list_for_each_entry(msidesc, &dev->msi_list, list) {
209 irq = xen_create_msi_irq(dev, msidesc, type); 192 struct physdev_map_pirq map_irq;
210 if (irq < 0)
211 return -1;
212 193
213 ret = set_irq_msi(irq, msidesc); 194 memset(&map_irq, 0, sizeof(map_irq));
214 if (ret) 195 map_irq.domid = DOMID_SELF;
215 goto error; 196 map_irq.type = MAP_PIRQ_TYPE_MSI;
216 } 197 map_irq.index = -1;
217 return 0; 198 map_irq.pirq = -1;
199 map_irq.bus = dev->bus->number;
200 map_irq.devfn = dev->devfn;
218 201
219error: 202 if (type == PCI_CAP_ID_MSIX) {
220 xen_destroy_irq(irq); 203 int pos;
204 u32 table_offset, bir;
205
206 pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
207
208 pci_read_config_dword(dev, pos + PCI_MSIX_TABLE,
209 &table_offset);
210 bir = (u8)(table_offset & PCI_MSIX_FLAGS_BIRMASK);
211
212 map_irq.table_base = pci_resource_start(dev, bir);
213 map_irq.entry_nr = msidesc->msi_attrib.entry_nr;
214 }
215
216 ret = HYPERVISOR_physdev_op(PHYSDEVOP_map_pirq, &map_irq);
217 if (ret) {
218 dev_warn(&dev->dev, "xen map irq failed %d\n", ret);
219 goto out;
220 }
221
222 ret = xen_bind_pirq_msi_to_irq(dev, msidesc,
223 map_irq.pirq, map_irq.index,
224 (type == PCI_CAP_ID_MSIX) ?
225 "msi-x" : "msi");
226 if (ret < 0)
227 goto out;
228 }
229 ret = 0;
230out:
221 return ret; 231 return ret;
222} 232}
223#endif 233#endif
234#endif
224 235
225static int xen_pcifront_enable_irq(struct pci_dev *dev) 236static int xen_pcifront_enable_irq(struct pci_dev *dev)
226{ 237{
227 int rc; 238 int rc;
228 int share = 1; 239 int share = 1;
240 u8 gsi;
229 241
230 dev_info(&dev->dev, "Xen PCI enabling IRQ: %d\n", dev->irq); 242 rc = pci_read_config_byte(dev, PCI_INTERRUPT_LINE, &gsi);
231 243 if (rc < 0) {
232 if (dev->irq < 0) 244 dev_warn(&dev->dev, "Xen PCI: failed to read interrupt line: %d\n",
233 return -EINVAL; 245 rc);
246 return rc;
247 }
234 248
235 if (dev->irq < NR_IRQS_LEGACY) 249 if (gsi < NR_IRQS_LEGACY)
236 share = 0; 250 share = 0;
237 251
238 rc = xen_allocate_pirq(dev->irq, share, "pcifront"); 252 rc = xen_allocate_pirq(gsi, share, "pcifront");
239 if (rc < 0) { 253 if (rc < 0) {
240 dev_warn(&dev->dev, "Xen PCI IRQ: %d, failed to register:%d\n", 254 dev_warn(&dev->dev, "Xen PCI: failed to register GSI%d: %d\n",
241 dev->irq, rc); 255 gsi, rc);
242 return rc; 256 return rc;
243 } 257 }
258
259 dev->irq = rc;
260 dev_info(&dev->dev, "Xen PCI mapped GSI%d to IRQ%d\n", gsi, dev->irq);
244 return 0; 261 return 0;
245} 262}
246 263
diff --git a/arch/x86/platform/ce4100/ce4100.c b/arch/x86/platform/ce4100/ce4100.c
index d2c0d51a7178..cd6f184c3b3f 100644
--- a/arch/x86/platform/ce4100/ce4100.c
+++ b/arch/x86/platform/ce4100/ce4100.c
@@ -15,6 +15,7 @@
15#include <linux/serial_reg.h> 15#include <linux/serial_reg.h>
16#include <linux/serial_8250.h> 16#include <linux/serial_8250.h>
17 17
18#include <asm/ce4100.h>
18#include <asm/setup.h> 19#include <asm/setup.h>
19#include <asm/io.h> 20#include <asm/io.h>
20 21
@@ -129,4 +130,5 @@ void __init x86_ce4100_early_setup(void)
129 x86_init.resources.probe_roms = x86_init_noop; 130 x86_init.resources.probe_roms = x86_init_noop;
130 x86_init.mpparse.get_smp_config = x86_init_uint_noop; 131 x86_init.mpparse.get_smp_config = x86_init_uint_noop;
131 x86_init.mpparse.find_smp_config = sdv_find_smp_config; 132 x86_init.mpparse.find_smp_config = sdv_find_smp_config;
133 x86_init.pci.init = ce4100_pci_init;
132} 134}
diff --git a/arch/x86/platform/uv/tlb_uv.c b/arch/x86/platform/uv/tlb_uv.c
index df58e9cad96a..a7b38d35c29a 100644
--- a/arch/x86/platform/uv/tlb_uv.c
+++ b/arch/x86/platform/uv/tlb_uv.c
@@ -1364,11 +1364,11 @@ uv_activation_descriptor_init(int node, int pnode)
1364 memset(bd2, 0, sizeof(struct bau_desc)); 1364 memset(bd2, 0, sizeof(struct bau_desc));
1365 bd2->header.sw_ack_flag = 1; 1365 bd2->header.sw_ack_flag = 1;
1366 /* 1366 /*
1367 * base_dest_nodeid is the nasid (pnode<<1) of the first uvhub 1367 * base_dest_nodeid is the nasid of the first uvhub
1368 * in the partition. The bit map will indicate uvhub numbers, 1368 * in the partition. The bit map will indicate uvhub numbers,
1369 * which are 0-N in a partition. Pnodes are unique system-wide. 1369 * which are 0-N in a partition. Pnodes are unique system-wide.
1370 */ 1370 */
1371 bd2->header.base_dest_nodeid = uv_partition_base_pnode << 1; 1371 bd2->header.base_dest_nodeid = UV_PNODE_TO_NASID(uv_partition_base_pnode);
1372 bd2->header.dest_subnodeid = 0x10; /* the LB */ 1372 bd2->header.dest_subnodeid = 0x10; /* the LB */
1373 bd2->header.command = UV_NET_ENDPOINT_INTD; 1373 bd2->header.command = UV_NET_ENDPOINT_INTD;
1374 bd2->header.int_both = 1; 1374 bd2->header.int_both = 1;
diff --git a/arch/x86/xen/Kconfig b/arch/x86/xen/Kconfig
index 5b54892e4bc3..e4343fe488ed 100644
--- a/arch/x86/xen/Kconfig
+++ b/arch/x86/xen/Kconfig
@@ -48,3 +48,11 @@ config XEN_DEBUG_FS
48 help 48 help
49 Enable statistics output and various tuning options in debugfs. 49 Enable statistics output and various tuning options in debugfs.
50 Enabling this option may incur a significant performance overhead. 50 Enabling this option may incur a significant performance overhead.
51
52config XEN_DEBUG
53 bool "Enable Xen debug checks"
54 depends on XEN
55 default n
56 help
57 Enable various WARN_ON checks in the Xen MMU code.
58 Enabling this option WILL incur a significant performance overhead.
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 50542efe45fb..49dbd78ec3cb 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1284,15 +1284,14 @@ static int init_hvm_pv_info(int *major, int *minor)
1284 1284
1285 xen_setup_features(); 1285 xen_setup_features();
1286 1286
1287 pv_info = xen_info; 1287 pv_info.name = "Xen HVM";
1288 pv_info.kernel_rpl = 0;
1289 1288
1290 xen_domain_type = XEN_HVM_DOMAIN; 1289 xen_domain_type = XEN_HVM_DOMAIN;
1291 1290
1292 return 0; 1291 return 0;
1293} 1292}
1294 1293
1295void xen_hvm_init_shared_info(void) 1294void __ref xen_hvm_init_shared_info(void)
1296{ 1295{
1297 int cpu; 1296 int cpu;
1298 struct xen_add_to_physmap xatp; 1297 struct xen_add_to_physmap xatp;
@@ -1331,6 +1330,8 @@ static int __cpuinit xen_hvm_cpu_notify(struct notifier_block *self,
1331 switch (action) { 1330 switch (action) {
1332 case CPU_UP_PREPARE: 1331 case CPU_UP_PREPARE:
1333 per_cpu(xen_vcpu, cpu) = &HYPERVISOR_shared_info->vcpu_info[cpu]; 1332 per_cpu(xen_vcpu, cpu) = &HYPERVISOR_shared_info->vcpu_info[cpu];
1333 if (xen_have_vector_callback)
1334 xen_init_lock_cpu(cpu);
1334 break; 1335 break;
1335 default: 1336 default:
1336 break; 1337 break;
@@ -1355,6 +1356,7 @@ static void __init xen_hvm_guest_init(void)
1355 1356
1356 if (xen_feature(XENFEAT_hvm_callback_vector)) 1357 if (xen_feature(XENFEAT_hvm_callback_vector))
1357 xen_have_vector_callback = 1; 1358 xen_have_vector_callback = 1;
1359 xen_hvm_smp_init();
1358 register_cpu_notifier(&xen_hvm_cpu_notifier); 1360 register_cpu_notifier(&xen_hvm_cpu_notifier);
1359 xen_unplug_emulated_devices(); 1361 xen_unplug_emulated_devices();
1360 have_vcpu_info_placement = 0; 1362 have_vcpu_info_placement = 0;
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index 5e92b61ad574..832765c0fb8c 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -46,6 +46,7 @@
46#include <linux/module.h> 46#include <linux/module.h>
47#include <linux/gfp.h> 47#include <linux/gfp.h>
48#include <linux/memblock.h> 48#include <linux/memblock.h>
49#include <linux/seq_file.h>
49 50
50#include <asm/pgtable.h> 51#include <asm/pgtable.h>
51#include <asm/tlbflush.h> 52#include <asm/tlbflush.h>
@@ -416,8 +417,12 @@ static pteval_t pte_pfn_to_mfn(pteval_t val)
416 if (val & _PAGE_PRESENT) { 417 if (val & _PAGE_PRESENT) {
417 unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT; 418 unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
418 pteval_t flags = val & PTE_FLAGS_MASK; 419 pteval_t flags = val & PTE_FLAGS_MASK;
419 unsigned long mfn = pfn_to_mfn(pfn); 420 unsigned long mfn;
420 421
422 if (!xen_feature(XENFEAT_auto_translated_physmap))
423 mfn = get_phys_to_machine(pfn);
424 else
425 mfn = pfn;
421 /* 426 /*
422 * If there's no mfn for the pfn, then just create an 427 * If there's no mfn for the pfn, then just create an
423 * empty non-present pte. Unfortunately this loses 428 * empty non-present pte. Unfortunately this loses
@@ -427,8 +432,18 @@ static pteval_t pte_pfn_to_mfn(pteval_t val)
427 if (unlikely(mfn == INVALID_P2M_ENTRY)) { 432 if (unlikely(mfn == INVALID_P2M_ENTRY)) {
428 mfn = 0; 433 mfn = 0;
429 flags = 0; 434 flags = 0;
435 } else {
436 /*
437 * Paramount to do this test _after_ the
438 * INVALID_P2M_ENTRY as INVALID_P2M_ENTRY &
439 * IDENTITY_FRAME_BIT resolves to true.
440 */
441 mfn &= ~FOREIGN_FRAME_BIT;
442 if (mfn & IDENTITY_FRAME_BIT) {
443 mfn &= ~IDENTITY_FRAME_BIT;
444 flags |= _PAGE_IOMAP;
445 }
430 } 446 }
431
432 val = ((pteval_t)mfn << PAGE_SHIFT) | flags; 447 val = ((pteval_t)mfn << PAGE_SHIFT) | flags;
433 } 448 }
434 449
@@ -532,6 +547,41 @@ pte_t xen_make_pte(pteval_t pte)
532} 547}
533PV_CALLEE_SAVE_REGS_THUNK(xen_make_pte); 548PV_CALLEE_SAVE_REGS_THUNK(xen_make_pte);
534 549
550#ifdef CONFIG_XEN_DEBUG
551pte_t xen_make_pte_debug(pteval_t pte)
552{
553 phys_addr_t addr = (pte & PTE_PFN_MASK);
554 phys_addr_t other_addr;
555 bool io_page = false;
556 pte_t _pte;
557
558 if (pte & _PAGE_IOMAP)
559 io_page = true;
560
561 _pte = xen_make_pte(pte);
562
563 if (!addr)
564 return _pte;
565
566 if (io_page &&
567 (xen_initial_domain() || addr >= ISA_END_ADDRESS)) {
568 other_addr = pfn_to_mfn(addr >> PAGE_SHIFT) << PAGE_SHIFT;
569 WARN(addr != other_addr,
570 "0x%lx is using VM_IO, but it is 0x%lx!\n",
571 (unsigned long)addr, (unsigned long)other_addr);
572 } else {
573 pteval_t iomap_set = (_pte.pte & PTE_FLAGS_MASK) & _PAGE_IOMAP;
574 other_addr = (_pte.pte & PTE_PFN_MASK);
575 WARN((addr == other_addr) && (!io_page) && (!iomap_set),
576 "0x%lx is missing VM_IO (and wasn't fixed)!\n",
577 (unsigned long)addr);
578 }
579
580 return _pte;
581}
582PV_CALLEE_SAVE_REGS_THUNK(xen_make_pte_debug);
583#endif
584
535pgd_t xen_make_pgd(pgdval_t pgd) 585pgd_t xen_make_pgd(pgdval_t pgd)
536{ 586{
537 pgd = pte_pfn_to_mfn(pgd); 587 pgd = pte_pfn_to_mfn(pgd);
@@ -986,10 +1036,9 @@ static void xen_pgd_pin(struct mm_struct *mm)
986 */ 1036 */
987void xen_mm_pin_all(void) 1037void xen_mm_pin_all(void)
988{ 1038{
989 unsigned long flags;
990 struct page *page; 1039 struct page *page;
991 1040
992 spin_lock_irqsave(&pgd_lock, flags); 1041 spin_lock(&pgd_lock);
993 1042
994 list_for_each_entry(page, &pgd_list, lru) { 1043 list_for_each_entry(page, &pgd_list, lru) {
995 if (!PagePinned(page)) { 1044 if (!PagePinned(page)) {
@@ -998,7 +1047,7 @@ void xen_mm_pin_all(void)
998 } 1047 }
999 } 1048 }
1000 1049
1001 spin_unlock_irqrestore(&pgd_lock, flags); 1050 spin_unlock(&pgd_lock);
1002} 1051}
1003 1052
1004/* 1053/*
@@ -1099,10 +1148,9 @@ static void xen_pgd_unpin(struct mm_struct *mm)
1099 */ 1148 */
1100void xen_mm_unpin_all(void) 1149void xen_mm_unpin_all(void)
1101{ 1150{
1102 unsigned long flags;
1103 struct page *page; 1151 struct page *page;
1104 1152
1105 spin_lock_irqsave(&pgd_lock, flags); 1153 spin_lock(&pgd_lock);
1106 1154
1107 list_for_each_entry(page, &pgd_list, lru) { 1155 list_for_each_entry(page, &pgd_list, lru) {
1108 if (PageSavePinned(page)) { 1156 if (PageSavePinned(page)) {
@@ -1112,7 +1160,7 @@ void xen_mm_unpin_all(void)
1112 } 1160 }
1113 } 1161 }
1114 1162
1115 spin_unlock_irqrestore(&pgd_lock, flags); 1163 spin_unlock(&pgd_lock);
1116} 1164}
1117 1165
1118void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next) 1166void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next)
@@ -1942,6 +1990,9 @@ __init void xen_ident_map_ISA(void)
1942 1990
1943static __init void xen_post_allocator_init(void) 1991static __init void xen_post_allocator_init(void)
1944{ 1992{
1993#ifdef CONFIG_XEN_DEBUG
1994 pv_mmu_ops.make_pte = PV_CALLEE_SAVE(xen_make_pte_debug);
1995#endif
1945 pv_mmu_ops.set_pte = xen_set_pte; 1996 pv_mmu_ops.set_pte = xen_set_pte;
1946 pv_mmu_ops.set_pmd = xen_set_pmd; 1997 pv_mmu_ops.set_pmd = xen_set_pmd;
1947 pv_mmu_ops.set_pud = xen_set_pud; 1998 pv_mmu_ops.set_pud = xen_set_pud;
@@ -2074,7 +2125,7 @@ static void xen_zap_pfn_range(unsigned long vaddr, unsigned int order,
2074 in_frames[i] = virt_to_mfn(vaddr); 2125 in_frames[i] = virt_to_mfn(vaddr);
2075 2126
2076 MULTI_update_va_mapping(mcs.mc, vaddr, VOID_PTE, 0); 2127 MULTI_update_va_mapping(mcs.mc, vaddr, VOID_PTE, 0);
2077 set_phys_to_machine(virt_to_pfn(vaddr), INVALID_P2M_ENTRY); 2128 __set_phys_to_machine(virt_to_pfn(vaddr), INVALID_P2M_ENTRY);
2078 2129
2079 if (out_frames) 2130 if (out_frames)
2080 out_frames[i] = virt_to_pfn(vaddr); 2131 out_frames[i] = virt_to_pfn(vaddr);
@@ -2353,6 +2404,18 @@ EXPORT_SYMBOL_GPL(xen_remap_domain_mfn_range);
2353 2404
2354#ifdef CONFIG_XEN_DEBUG_FS 2405#ifdef CONFIG_XEN_DEBUG_FS
2355 2406
2407static int p2m_dump_open(struct inode *inode, struct file *filp)
2408{
2409 return single_open(filp, p2m_dump_show, NULL);
2410}
2411
2412static const struct file_operations p2m_dump_fops = {
2413 .open = p2m_dump_open,
2414 .read = seq_read,
2415 .llseek = seq_lseek,
2416 .release = single_release,
2417};
2418
2356static struct dentry *d_mmu_debug; 2419static struct dentry *d_mmu_debug;
2357 2420
2358static int __init xen_mmu_debugfs(void) 2421static int __init xen_mmu_debugfs(void)
@@ -2408,6 +2471,7 @@ static int __init xen_mmu_debugfs(void)
2408 debugfs_create_u32("prot_commit_batched", 0444, d_mmu_debug, 2471 debugfs_create_u32("prot_commit_batched", 0444, d_mmu_debug,
2409 &mmu_stats.prot_commit_batched); 2472 &mmu_stats.prot_commit_batched);
2410 2473
2474 debugfs_create_file("p2m", 0600, d_mmu_debug, NULL, &p2m_dump_fops);
2411 return 0; 2475 return 0;
2412} 2476}
2413fs_initcall(xen_mmu_debugfs); 2477fs_initcall(xen_mmu_debugfs);
diff --git a/arch/x86/xen/p2m.c b/arch/x86/xen/p2m.c
index fd12d7ce7ff9..215a3ce61068 100644
--- a/arch/x86/xen/p2m.c
+++ b/arch/x86/xen/p2m.c
@@ -23,6 +23,129 @@
23 * P2M_PER_PAGE depends on the architecture, as a mfn is always 23 * P2M_PER_PAGE depends on the architecture, as a mfn is always
24 * unsigned long (8 bytes on 64-bit, 4 bytes on 32), leading to 24 * unsigned long (8 bytes on 64-bit, 4 bytes on 32), leading to
25 * 512 and 1024 entries respectively. 25 * 512 and 1024 entries respectively.
26 *
27 * In short, these structures contain the Machine Frame Number (MFN) of the PFN.
28 *
29 * However not all entries are filled with MFNs. Specifically for all other
30 * leaf entries, or for the top root, or middle one, for which there is a void
31 * entry, we assume it is "missing". So (for example)
32 * pfn_to_mfn(0x90909090)=INVALID_P2M_ENTRY.
33 *
34 * We also have the possibility of setting 1-1 mappings on certain regions, so
35 * that:
36 * pfn_to_mfn(0xc0000)=0xc0000
37 *
38 * The benefit of this is, that we can assume for non-RAM regions (think
39 * PCI BARs, or ACPI spaces), we can create mappings easily b/c we
40 * get the PFN value to match the MFN.
41 *
42 * For this to work efficiently we have one new page p2m_identity and
43 * allocate (via reserved_brk) any other pages we need to cover the sides
44 * (1GB or 4MB boundary violations). All entries in p2m_identity are set to
45 * INVALID_P2M_ENTRY type (Xen toolstack only recognizes that and MFNs,
46 * no other fancy value).
47 *
48 * On lookup we spot that the entry points to p2m_identity and return the
49 * identity value instead of dereferencing and returning INVALID_P2M_ENTRY.
50 * If the entry points to an allocated page, we just proceed as before and
51 * return the PFN. If the PFN has IDENTITY_FRAME_BIT set we unmask that in
52 * appropriate functions (pfn_to_mfn).
53 *
54 * The reason for having the IDENTITY_FRAME_BIT instead of just returning the
55 * PFN is that we could find ourselves where pfn_to_mfn(pfn)==pfn for a
56 * non-identity pfn. To protect ourselves against we elect to set (and get) the
57 * IDENTITY_FRAME_BIT on all identity mapped PFNs.
58 *
59 * This simplistic diagram is used to explain the more subtle piece of code.
60 * There is also a digram of the P2M at the end that can help.
61 * Imagine your E820 looking as so:
62 *
63 * 1GB 2GB
64 * /-------------------+---------\/----\ /----------\ /---+-----\
65 * | System RAM | Sys RAM ||ACPI| | reserved | | Sys RAM |
66 * \-------------------+---------/\----/ \----------/ \---+-----/
67 * ^- 1029MB ^- 2001MB
68 *
69 * [1029MB = 263424 (0x40500), 2001MB = 512256 (0x7D100),
70 * 2048MB = 524288 (0x80000)]
71 *
72 * And dom0_mem=max:3GB,1GB is passed in to the guest, meaning memory past 1GB
73 * is actually not present (would have to kick the balloon driver to put it in).
74 *
75 * When we are told to set the PFNs for identity mapping (see patch: "xen/setup:
76 * Set identity mapping for non-RAM E820 and E820 gaps.") we pass in the start
77 * of the PFN and the end PFN (263424 and 512256 respectively). The first step
78 * is to reserve_brk a top leaf page if the p2m[1] is missing. The top leaf page
79 * covers 512^2 of page estate (1GB) and in case the start or end PFN is not
80 * aligned on 512^2*PAGE_SIZE (1GB) we loop on aligned 1GB PFNs from start pfn
81 * to end pfn. We reserve_brk top leaf pages if they are missing (means they
82 * point to p2m_mid_missing).
83 *
84 * With the E820 example above, 263424 is not 1GB aligned so we allocate a
85 * reserve_brk page which will cover the PFNs estate from 0x40000 to 0x80000.
86 * Each entry in the allocate page is "missing" (points to p2m_missing).
87 *
88 * Next stage is to determine if we need to do a more granular boundary check
89 * on the 4MB (or 2MB depending on architecture) off the start and end pfn's.
90 * We check if the start pfn and end pfn violate that boundary check, and if
91 * so reserve_brk a middle (p2m[x][y]) leaf page. This way we have a much finer
92 * granularity of setting which PFNs are missing and which ones are identity.
93 * In our example 263424 and 512256 both fail the check so we reserve_brk two
94 * pages. Populate them with INVALID_P2M_ENTRY (so they both have "missing"
95 * values) and assign them to p2m[1][2] and p2m[1][488] respectively.
96 *
97 * At this point we would at minimum reserve_brk one page, but could be up to
98 * three. Each call to set_phys_range_identity has at maximum a three page
99 * cost. If we were to query the P2M at this stage, all those entries from
100 * start PFN through end PFN (so 1029MB -> 2001MB) would return
101 * INVALID_P2M_ENTRY ("missing").
102 *
103 * The next step is to walk from the start pfn to the end pfn setting
104 * the IDENTITY_FRAME_BIT on each PFN. This is done in set_phys_range_identity.
105 * If we find that the middle leaf is pointing to p2m_missing we can swap it
106 * over to p2m_identity - this way covering 4MB (or 2MB) PFN space. At this
107 * point we do not need to worry about boundary aligment (so no need to
108 * reserve_brk a middle page, figure out which PFNs are "missing" and which
109 * ones are identity), as that has been done earlier. If we find that the
110 * middle leaf is not occupied by p2m_identity or p2m_missing, we dereference
111 * that page (which covers 512 PFNs) and set the appropriate PFN with
112 * IDENTITY_FRAME_BIT. In our example 263424 and 512256 end up there, and we
113 * set from p2m[1][2][256->511] and p2m[1][488][0->256] with
114 * IDENTITY_FRAME_BIT set.
115 *
116 * All other regions that are void (or not filled) either point to p2m_missing
117 * (considered missing) or have the default value of INVALID_P2M_ENTRY (also
118 * considered missing). In our case, p2m[1][2][0->255] and p2m[1][488][257->511]
119 * contain the INVALID_P2M_ENTRY value and are considered "missing."
120 *
121 * This is what the p2m ends up looking (for the E820 above) with this
122 * fabulous drawing:
123 *
124 * p2m /--------------\
125 * /-----\ | &mfn_list[0],| /-----------------\
126 * | 0 |------>| &mfn_list[1],| /---------------\ | ~0, ~0, .. |
127 * |-----| | ..., ~0, ~0 | | ~0, ~0, [x]---+----->| IDENTITY [@256] |
128 * | 1 |---\ \--------------/ | [p2m_identity]+\ | IDENTITY [@257] |
129 * |-----| \ | [p2m_identity]+\\ | .... |
130 * | 2 |--\ \-------------------->| ... | \\ \----------------/
131 * |-----| \ \---------------/ \\
132 * | 3 |\ \ \\ p2m_identity
133 * |-----| \ \-------------------->/---------------\ /-----------------\
134 * | .. +->+ | [p2m_identity]+-->| ~0, ~0, ~0, ... |
135 * \-----/ / | [p2m_identity]+-->| ..., ~0 |
136 * / /---------------\ | .... | \-----------------/
137 * / | IDENTITY[@0] | /-+-[x], ~0, ~0.. |
138 * / | IDENTITY[@256]|<----/ \---------------/
139 * / | ~0, ~0, .... |
140 * | \---------------/
141 * |
142 * p2m_missing p2m_missing
143 * /------------------\ /------------\
144 * | [p2m_mid_missing]+---->| ~0, ~0, ~0 |
145 * | [p2m_mid_missing]+---->| ..., ~0 |
146 * \------------------/ \------------/
147 *
148 * where ~0 is INVALID_P2M_ENTRY. IDENTITY is (PFN | IDENTITY_BIT)
26 */ 149 */
27 150
28#include <linux/init.h> 151#include <linux/init.h>
@@ -30,6 +153,7 @@
30#include <linux/list.h> 153#include <linux/list.h>
31#include <linux/hash.h> 154#include <linux/hash.h>
32#include <linux/sched.h> 155#include <linux/sched.h>
156#include <linux/seq_file.h>
33 157
34#include <asm/cache.h> 158#include <asm/cache.h>
35#include <asm/setup.h> 159#include <asm/setup.h>
@@ -59,9 +183,15 @@ static RESERVE_BRK_ARRAY(unsigned long **, p2m_top, P2M_TOP_PER_PAGE);
59static RESERVE_BRK_ARRAY(unsigned long, p2m_top_mfn, P2M_TOP_PER_PAGE); 183static RESERVE_BRK_ARRAY(unsigned long, p2m_top_mfn, P2M_TOP_PER_PAGE);
60static RESERVE_BRK_ARRAY(unsigned long *, p2m_top_mfn_p, P2M_TOP_PER_PAGE); 184static RESERVE_BRK_ARRAY(unsigned long *, p2m_top_mfn_p, P2M_TOP_PER_PAGE);
61 185
186static RESERVE_BRK_ARRAY(unsigned long, p2m_identity, P2M_PER_PAGE);
187
62RESERVE_BRK(p2m_mid, PAGE_SIZE * (MAX_DOMAIN_PAGES / (P2M_PER_PAGE * P2M_MID_PER_PAGE))); 188RESERVE_BRK(p2m_mid, PAGE_SIZE * (MAX_DOMAIN_PAGES / (P2M_PER_PAGE * P2M_MID_PER_PAGE)));
63RESERVE_BRK(p2m_mid_mfn, PAGE_SIZE * (MAX_DOMAIN_PAGES / (P2M_PER_PAGE * P2M_MID_PER_PAGE))); 189RESERVE_BRK(p2m_mid_mfn, PAGE_SIZE * (MAX_DOMAIN_PAGES / (P2M_PER_PAGE * P2M_MID_PER_PAGE)));
64 190
191/* We might hit two boundary violations at the start and end, at max each
192 * boundary violation will require three middle nodes. */
193RESERVE_BRK(p2m_mid_identity, PAGE_SIZE * 2 * 3);
194
65static inline unsigned p2m_top_index(unsigned long pfn) 195static inline unsigned p2m_top_index(unsigned long pfn)
66{ 196{
67 BUG_ON(pfn >= MAX_P2M_PFN); 197 BUG_ON(pfn >= MAX_P2M_PFN);
@@ -136,7 +266,7 @@ static void p2m_init(unsigned long *p2m)
136 * - After resume we're called from within stop_machine, but the mfn 266 * - After resume we're called from within stop_machine, but the mfn
137 * tree should alreay be completely allocated. 267 * tree should alreay be completely allocated.
138 */ 268 */
139void xen_build_mfn_list_list(void) 269void __ref xen_build_mfn_list_list(void)
140{ 270{
141 unsigned long pfn; 271 unsigned long pfn;
142 272
@@ -221,6 +351,9 @@ void __init xen_build_dynamic_phys_to_machine(void)
221 p2m_top = extend_brk(PAGE_SIZE, PAGE_SIZE); 351 p2m_top = extend_brk(PAGE_SIZE, PAGE_SIZE);
222 p2m_top_init(p2m_top); 352 p2m_top_init(p2m_top);
223 353
354 p2m_identity = extend_brk(PAGE_SIZE, PAGE_SIZE);
355 p2m_init(p2m_identity);
356
224 /* 357 /*
225 * The domain builder gives us a pre-constructed p2m array in 358 * The domain builder gives us a pre-constructed p2m array in
226 * mfn_list for all the pages initially given to us, so we just 359 * mfn_list for all the pages initially given to us, so we just
@@ -266,6 +399,14 @@ unsigned long get_phys_to_machine(unsigned long pfn)
266 mididx = p2m_mid_index(pfn); 399 mididx = p2m_mid_index(pfn);
267 idx = p2m_index(pfn); 400 idx = p2m_index(pfn);
268 401
402 /*
403 * The INVALID_P2M_ENTRY is filled in both p2m_*identity
404 * and in p2m_*missing, so returning the INVALID_P2M_ENTRY
405 * would be wrong.
406 */
407 if (p2m_top[topidx][mididx] == p2m_identity)
408 return IDENTITY_FRAME(pfn);
409
269 return p2m_top[topidx][mididx][idx]; 410 return p2m_top[topidx][mididx][idx];
270} 411}
271EXPORT_SYMBOL_GPL(get_phys_to_machine); 412EXPORT_SYMBOL_GPL(get_phys_to_machine);
@@ -335,9 +476,11 @@ static bool alloc_p2m(unsigned long pfn)
335 p2m_top_mfn_p[topidx] = mid_mfn; 476 p2m_top_mfn_p[topidx] = mid_mfn;
336 } 477 }
337 478
338 if (p2m_top[topidx][mididx] == p2m_missing) { 479 if (p2m_top[topidx][mididx] == p2m_identity ||
480 p2m_top[topidx][mididx] == p2m_missing) {
339 /* p2m leaf page is missing */ 481 /* p2m leaf page is missing */
340 unsigned long *p2m; 482 unsigned long *p2m;
483 unsigned long *p2m_orig = p2m_top[topidx][mididx];
341 484
342 p2m = alloc_p2m_page(); 485 p2m = alloc_p2m_page();
343 if (!p2m) 486 if (!p2m)
@@ -345,7 +488,7 @@ static bool alloc_p2m(unsigned long pfn)
345 488
346 p2m_init(p2m); 489 p2m_init(p2m);
347 490
348 if (cmpxchg(&mid[mididx], p2m_missing, p2m) != p2m_missing) 491 if (cmpxchg(&mid[mididx], p2m_orig, p2m) != p2m_orig)
349 free_p2m_page(p2m); 492 free_p2m_page(p2m);
350 else 493 else
351 mid_mfn[mididx] = virt_to_mfn(p2m); 494 mid_mfn[mididx] = virt_to_mfn(p2m);
@@ -354,11 +497,91 @@ static bool alloc_p2m(unsigned long pfn)
354 return true; 497 return true;
355} 498}
356 499
500bool __early_alloc_p2m(unsigned long pfn)
501{
502 unsigned topidx, mididx, idx;
503
504 topidx = p2m_top_index(pfn);
505 mididx = p2m_mid_index(pfn);
506 idx = p2m_index(pfn);
507
508 /* Pfff.. No boundary cross-over, lets get out. */
509 if (!idx)
510 return false;
511
512 WARN(p2m_top[topidx][mididx] == p2m_identity,
513 "P2M[%d][%d] == IDENTITY, should be MISSING (or alloced)!\n",
514 topidx, mididx);
515
516 /*
517 * Could be done by xen_build_dynamic_phys_to_machine..
518 */
519 if (p2m_top[topidx][mididx] != p2m_missing)
520 return false;
521
522 /* Boundary cross-over for the edges: */
523 if (idx) {
524 unsigned long *p2m = extend_brk(PAGE_SIZE, PAGE_SIZE);
525
526 p2m_init(p2m);
527
528 p2m_top[topidx][mididx] = p2m;
529
530 }
531 return idx != 0;
532}
533unsigned long set_phys_range_identity(unsigned long pfn_s,
534 unsigned long pfn_e)
535{
536 unsigned long pfn;
537
538 if (unlikely(pfn_s >= MAX_P2M_PFN || pfn_e >= MAX_P2M_PFN))
539 return 0;
540
541 if (unlikely(xen_feature(XENFEAT_auto_translated_physmap)))
542 return pfn_e - pfn_s;
543
544 if (pfn_s > pfn_e)
545 return 0;
546
547 for (pfn = (pfn_s & ~(P2M_MID_PER_PAGE * P2M_PER_PAGE - 1));
548 pfn < ALIGN(pfn_e, (P2M_MID_PER_PAGE * P2M_PER_PAGE));
549 pfn += P2M_MID_PER_PAGE * P2M_PER_PAGE)
550 {
551 unsigned topidx = p2m_top_index(pfn);
552 if (p2m_top[topidx] == p2m_mid_missing) {
553 unsigned long **mid = extend_brk(PAGE_SIZE, PAGE_SIZE);
554
555 p2m_mid_init(mid);
556
557 p2m_top[topidx] = mid;
558 }
559 }
560
561 __early_alloc_p2m(pfn_s);
562 __early_alloc_p2m(pfn_e);
563
564 for (pfn = pfn_s; pfn < pfn_e; pfn++)
565 if (!__set_phys_to_machine(pfn, IDENTITY_FRAME(pfn)))
566 break;
567
568 if (!WARN((pfn - pfn_s) != (pfn_e - pfn_s),
569 "Identity mapping failed. We are %ld short of 1-1 mappings!\n",
570 (pfn_e - pfn_s) - (pfn - pfn_s)))
571 printk(KERN_DEBUG "1-1 mapping on %lx->%lx\n", pfn_s, pfn);
572
573 return pfn - pfn_s;
574}
575
357/* Try to install p2m mapping; fail if intermediate bits missing */ 576/* Try to install p2m mapping; fail if intermediate bits missing */
358bool __set_phys_to_machine(unsigned long pfn, unsigned long mfn) 577bool __set_phys_to_machine(unsigned long pfn, unsigned long mfn)
359{ 578{
360 unsigned topidx, mididx, idx; 579 unsigned topidx, mididx, idx;
361 580
581 if (unlikely(xen_feature(XENFEAT_auto_translated_physmap))) {
582 BUG_ON(pfn != mfn && mfn != INVALID_P2M_ENTRY);
583 return true;
584 }
362 if (unlikely(pfn >= MAX_P2M_PFN)) { 585 if (unlikely(pfn >= MAX_P2M_PFN)) {
363 BUG_ON(mfn != INVALID_P2M_ENTRY); 586 BUG_ON(mfn != INVALID_P2M_ENTRY);
364 return true; 587 return true;
@@ -368,6 +591,21 @@ bool __set_phys_to_machine(unsigned long pfn, unsigned long mfn)
368 mididx = p2m_mid_index(pfn); 591 mididx = p2m_mid_index(pfn);
369 idx = p2m_index(pfn); 592 idx = p2m_index(pfn);
370 593
594 /* For sparse holes were the p2m leaf has real PFN along with
595 * PCI holes, stick in the PFN as the MFN value.
596 */
597 if (mfn != INVALID_P2M_ENTRY && (mfn & IDENTITY_FRAME_BIT)) {
598 if (p2m_top[topidx][mididx] == p2m_identity)
599 return true;
600
601 /* Swap over from MISSING to IDENTITY if needed. */
602 if (p2m_top[topidx][mididx] == p2m_missing) {
603 WARN_ON(cmpxchg(&p2m_top[topidx][mididx], p2m_missing,
604 p2m_identity) != p2m_missing);
605 return true;
606 }
607 }
608
371 if (p2m_top[topidx][mididx] == p2m_missing) 609 if (p2m_top[topidx][mididx] == p2m_missing)
372 return mfn == INVALID_P2M_ENTRY; 610 return mfn == INVALID_P2M_ENTRY;
373 611
@@ -378,11 +616,6 @@ bool __set_phys_to_machine(unsigned long pfn, unsigned long mfn)
378 616
379bool set_phys_to_machine(unsigned long pfn, unsigned long mfn) 617bool set_phys_to_machine(unsigned long pfn, unsigned long mfn)
380{ 618{
381 if (unlikely(xen_feature(XENFEAT_auto_translated_physmap))) {
382 BUG_ON(pfn != mfn && mfn != INVALID_P2M_ENTRY);
383 return true;
384 }
385
386 if (unlikely(!__set_phys_to_machine(pfn, mfn))) { 619 if (unlikely(!__set_phys_to_machine(pfn, mfn))) {
387 if (!alloc_p2m(pfn)) 620 if (!alloc_p2m(pfn))
388 return false; 621 return false;
@@ -421,7 +654,7 @@ int m2p_add_override(unsigned long mfn, struct page *page)
421{ 654{
422 unsigned long flags; 655 unsigned long flags;
423 unsigned long pfn; 656 unsigned long pfn;
424 unsigned long address; 657 unsigned long uninitialized_var(address);
425 unsigned level; 658 unsigned level;
426 pte_t *ptep = NULL; 659 pte_t *ptep = NULL;
427 660
@@ -455,7 +688,7 @@ int m2p_remove_override(struct page *page)
455 unsigned long flags; 688 unsigned long flags;
456 unsigned long mfn; 689 unsigned long mfn;
457 unsigned long pfn; 690 unsigned long pfn;
458 unsigned long address; 691 unsigned long uninitialized_var(address);
459 unsigned level; 692 unsigned level;
460 pte_t *ptep = NULL; 693 pte_t *ptep = NULL;
461 694
@@ -520,3 +753,80 @@ unsigned long m2p_find_override_pfn(unsigned long mfn, unsigned long pfn)
520 return ret; 753 return ret;
521} 754}
522EXPORT_SYMBOL_GPL(m2p_find_override_pfn); 755EXPORT_SYMBOL_GPL(m2p_find_override_pfn);
756
757#ifdef CONFIG_XEN_DEBUG_FS
758
759int p2m_dump_show(struct seq_file *m, void *v)
760{
761 static const char * const level_name[] = { "top", "middle",
762 "entry", "abnormal" };
763 static const char * const type_name[] = { "identity", "missing",
764 "pfn", "abnormal"};
765#define TYPE_IDENTITY 0
766#define TYPE_MISSING 1
767#define TYPE_PFN 2
768#define TYPE_UNKNOWN 3
769 unsigned long pfn, prev_pfn_type = 0, prev_pfn_level = 0;
770 unsigned int uninitialized_var(prev_level);
771 unsigned int uninitialized_var(prev_type);
772
773 if (!p2m_top)
774 return 0;
775
776 for (pfn = 0; pfn < MAX_DOMAIN_PAGES; pfn++) {
777 unsigned topidx = p2m_top_index(pfn);
778 unsigned mididx = p2m_mid_index(pfn);
779 unsigned idx = p2m_index(pfn);
780 unsigned lvl, type;
781
782 lvl = 4;
783 type = TYPE_UNKNOWN;
784 if (p2m_top[topidx] == p2m_mid_missing) {
785 lvl = 0; type = TYPE_MISSING;
786 } else if (p2m_top[topidx] == NULL) {
787 lvl = 0; type = TYPE_UNKNOWN;
788 } else if (p2m_top[topidx][mididx] == NULL) {
789 lvl = 1; type = TYPE_UNKNOWN;
790 } else if (p2m_top[topidx][mididx] == p2m_identity) {
791 lvl = 1; type = TYPE_IDENTITY;
792 } else if (p2m_top[topidx][mididx] == p2m_missing) {
793 lvl = 1; type = TYPE_MISSING;
794 } else if (p2m_top[topidx][mididx][idx] == 0) {
795 lvl = 2; type = TYPE_UNKNOWN;
796 } else if (p2m_top[topidx][mididx][idx] == IDENTITY_FRAME(pfn)) {
797 lvl = 2; type = TYPE_IDENTITY;
798 } else if (p2m_top[topidx][mididx][idx] == INVALID_P2M_ENTRY) {
799 lvl = 2; type = TYPE_MISSING;
800 } else if (p2m_top[topidx][mididx][idx] == pfn) {
801 lvl = 2; type = TYPE_PFN;
802 } else if (p2m_top[topidx][mididx][idx] != pfn) {
803 lvl = 2; type = TYPE_PFN;
804 }
805 if (pfn == 0) {
806 prev_level = lvl;
807 prev_type = type;
808 }
809 if (pfn == MAX_DOMAIN_PAGES-1) {
810 lvl = 3;
811 type = TYPE_UNKNOWN;
812 }
813 if (prev_type != type) {
814 seq_printf(m, " [0x%lx->0x%lx] %s\n",
815 prev_pfn_type, pfn, type_name[prev_type]);
816 prev_pfn_type = pfn;
817 prev_type = type;
818 }
819 if (prev_level != lvl) {
820 seq_printf(m, " [0x%lx->0x%lx] level %s\n",
821 prev_pfn_level, pfn, level_name[prev_level]);
822 prev_pfn_level = pfn;
823 prev_level = lvl;
824 }
825 }
826 return 0;
827#undef TYPE_IDENTITY
828#undef TYPE_MISSING
829#undef TYPE_PFN
830#undef TYPE_UNKNOWN
831}
832#endif
diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c
index a8a66a50d446..fa0269a99377 100644
--- a/arch/x86/xen/setup.c
+++ b/arch/x86/xen/setup.c
@@ -52,6 +52,8 @@ phys_addr_t xen_extra_mem_start, xen_extra_mem_size;
52 52
53static __init void xen_add_extra_mem(unsigned long pages) 53static __init void xen_add_extra_mem(unsigned long pages)
54{ 54{
55 unsigned long pfn;
56
55 u64 size = (u64)pages * PAGE_SIZE; 57 u64 size = (u64)pages * PAGE_SIZE;
56 u64 extra_start = xen_extra_mem_start + xen_extra_mem_size; 58 u64 extra_start = xen_extra_mem_start + xen_extra_mem_size;
57 59
@@ -66,6 +68,9 @@ static __init void xen_add_extra_mem(unsigned long pages)
66 xen_extra_mem_size += size; 68 xen_extra_mem_size += size;
67 69
68 xen_max_p2m_pfn = PFN_DOWN(extra_start + size); 70 xen_max_p2m_pfn = PFN_DOWN(extra_start + size);
71
72 for (pfn = PFN_DOWN(extra_start); pfn <= xen_max_p2m_pfn; pfn++)
73 __set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
69} 74}
70 75
71static unsigned long __init xen_release_chunk(phys_addr_t start_addr, 76static unsigned long __init xen_release_chunk(phys_addr_t start_addr,
@@ -104,7 +109,7 @@ static unsigned long __init xen_release_chunk(phys_addr_t start_addr,
104 WARN(ret != 1, "Failed to release memory %lx-%lx err=%d\n", 109 WARN(ret != 1, "Failed to release memory %lx-%lx err=%d\n",
105 start, end, ret); 110 start, end, ret);
106 if (ret == 1) { 111 if (ret == 1) {
107 set_phys_to_machine(pfn, INVALID_P2M_ENTRY); 112 __set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
108 len++; 113 len++;
109 } 114 }
110 } 115 }
@@ -138,12 +143,55 @@ static unsigned long __init xen_return_unused_memory(unsigned long max_pfn,
138 return released; 143 return released;
139} 144}
140 145
146static unsigned long __init xen_set_identity(const struct e820entry *list,
147 ssize_t map_size)
148{
149 phys_addr_t last = xen_initial_domain() ? 0 : ISA_END_ADDRESS;
150 phys_addr_t start_pci = last;
151 const struct e820entry *entry;
152 unsigned long identity = 0;
153 int i;
154
155 for (i = 0, entry = list; i < map_size; i++, entry++) {
156 phys_addr_t start = entry->addr;
157 phys_addr_t end = start + entry->size;
158
159 if (start < last)
160 start = last;
161
162 if (end <= start)
163 continue;
164
165 /* Skip over the 1MB region. */
166 if (last > end)
167 continue;
168
169 if (entry->type == E820_RAM) {
170 if (start > start_pci)
171 identity += set_phys_range_identity(
172 PFN_UP(start_pci), PFN_DOWN(start));
173
174 /* Without saving 'last' we would gooble RAM too
175 * at the end of the loop. */
176 last = end;
177 start_pci = end;
178 continue;
179 }
180 start_pci = min(start, start_pci);
181 last = end;
182 }
183 if (last > start_pci)
184 identity += set_phys_range_identity(
185 PFN_UP(start_pci), PFN_DOWN(last));
186 return identity;
187}
141/** 188/**
142 * machine_specific_memory_setup - Hook for machine specific memory setup. 189 * machine_specific_memory_setup - Hook for machine specific memory setup.
143 **/ 190 **/
144char * __init xen_memory_setup(void) 191char * __init xen_memory_setup(void)
145{ 192{
146 static struct e820entry map[E820MAX] __initdata; 193 static struct e820entry map[E820MAX] __initdata;
194 static struct e820entry map_raw[E820MAX] __initdata;
147 195
148 unsigned long max_pfn = xen_start_info->nr_pages; 196 unsigned long max_pfn = xen_start_info->nr_pages;
149 unsigned long long mem_end; 197 unsigned long long mem_end;
@@ -151,6 +199,7 @@ char * __init xen_memory_setup(void)
151 struct xen_memory_map memmap; 199 struct xen_memory_map memmap;
152 unsigned long extra_pages = 0; 200 unsigned long extra_pages = 0;
153 unsigned long extra_limit; 201 unsigned long extra_limit;
202 unsigned long identity_pages = 0;
154 int i; 203 int i;
155 int op; 204 int op;
156 205
@@ -176,6 +225,7 @@ char * __init xen_memory_setup(void)
176 } 225 }
177 BUG_ON(rc); 226 BUG_ON(rc);
178 227
228 memcpy(map_raw, map, sizeof(map));
179 e820.nr_map = 0; 229 e820.nr_map = 0;
180 xen_extra_mem_start = mem_end; 230 xen_extra_mem_start = mem_end;
181 for (i = 0; i < memmap.nr_entries; i++) { 231 for (i = 0; i < memmap.nr_entries; i++) {
@@ -194,6 +244,15 @@ char * __init xen_memory_setup(void)
194 end -= delta; 244 end -= delta;
195 245
196 extra_pages += PFN_DOWN(delta); 246 extra_pages += PFN_DOWN(delta);
247 /*
248 * Set RAM below 4GB that is not for us to be unusable.
249 * This prevents "System RAM" address space from being
250 * used as potential resource for I/O address (happens
251 * when 'allocate_resource' is called).
252 */
253 if (delta &&
254 (xen_initial_domain() && end < 0x100000000ULL))
255 e820_add_region(end, delta, E820_UNUSABLE);
197 } 256 }
198 257
199 if (map[i].size > 0 && end > xen_extra_mem_start) 258 if (map[i].size > 0 && end > xen_extra_mem_start)
@@ -251,6 +310,13 @@ char * __init xen_memory_setup(void)
251 310
252 xen_add_extra_mem(extra_pages); 311 xen_add_extra_mem(extra_pages);
253 312
313 /*
314 * Set P2M for all non-RAM pages and E820 gaps to be identity
315 * type PFNs. We supply it with the non-sanitized version
316 * of the E820.
317 */
318 identity_pages = xen_set_identity(map_raw, memmap.nr_entries);
319 printk(KERN_INFO "Set %ld page(s) to 1-1 mapping.\n", identity_pages);
254 return "Xen"; 320 return "Xen";
255} 321}
256 322
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 72a4c7959045..30612441ed99 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -509,3 +509,41 @@ void __init xen_smp_init(void)
509 xen_fill_possible_map(); 509 xen_fill_possible_map();
510 xen_init_spinlocks(); 510 xen_init_spinlocks();
511} 511}
512
513static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
514{
515 native_smp_prepare_cpus(max_cpus);
516 WARN_ON(xen_smp_intr_init(0));
517
518 if (!xen_have_vector_callback)
519 return;
520 xen_init_lock_cpu(0);
521 xen_init_spinlocks();
522}
523
524static int __cpuinit xen_hvm_cpu_up(unsigned int cpu)
525{
526 int rc;
527 rc = native_cpu_up(cpu);
528 WARN_ON (xen_smp_intr_init(cpu));
529 return rc;
530}
531
532static void xen_hvm_cpu_die(unsigned int cpu)
533{
534 unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
535 unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
536 unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
537 unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
538 native_cpu_die(cpu);
539}
540
541void __init xen_hvm_smp_init(void)
542{
543 smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
544 smp_ops.smp_send_reschedule = xen_smp_send_reschedule;
545 smp_ops.cpu_up = xen_hvm_cpu_up;
546 smp_ops.cpu_die = xen_hvm_cpu_die;
547 smp_ops.send_call_func_ipi = xen_smp_send_call_function_ipi;
548 smp_ops.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi;
549}
diff --git a/arch/x86/xen/suspend.c b/arch/x86/xen/suspend.c
index 9bbd63a129b5..45329c8c226e 100644
--- a/arch/x86/xen/suspend.c
+++ b/arch/x86/xen/suspend.c
@@ -12,7 +12,7 @@
12#include "xen-ops.h" 12#include "xen-ops.h"
13#include "mmu.h" 13#include "mmu.h"
14 14
15void xen_pre_suspend(void) 15void xen_arch_pre_suspend(void)
16{ 16{
17 xen_start_info->store_mfn = mfn_to_pfn(xen_start_info->store_mfn); 17 xen_start_info->store_mfn = mfn_to_pfn(xen_start_info->store_mfn);
18 xen_start_info->console.domU.mfn = 18 xen_start_info->console.domU.mfn =
@@ -26,8 +26,9 @@ void xen_pre_suspend(void)
26 BUG(); 26 BUG();
27} 27}
28 28
29void xen_hvm_post_suspend(int suspend_cancelled) 29void xen_arch_hvm_post_suspend(int suspend_cancelled)
30{ 30{
31#ifdef CONFIG_XEN_PVHVM
31 int cpu; 32 int cpu;
32 xen_hvm_init_shared_info(); 33 xen_hvm_init_shared_info();
33 xen_callback_vector(); 34 xen_callback_vector();
@@ -37,9 +38,10 @@ void xen_hvm_post_suspend(int suspend_cancelled)
37 xen_setup_runstate_info(cpu); 38 xen_setup_runstate_info(cpu);
38 } 39 }
39 } 40 }
41#endif
40} 42}
41 43
42void xen_post_suspend(int suspend_cancelled) 44void xen_arch_post_suspend(int suspend_cancelled)
43{ 45{
44 xen_build_mfn_list_list(); 46 xen_build_mfn_list_list();
45 47
diff --git a/arch/x86/xen/time.c b/arch/x86/xen/time.c
index 067759e3d6a5..2e2d370a47b1 100644
--- a/arch/x86/xen/time.c
+++ b/arch/x86/xen/time.c
@@ -397,7 +397,9 @@ void xen_setup_timer(int cpu)
397 name = "<timer kasprintf failed>"; 397 name = "<timer kasprintf failed>";
398 398
399 irq = bind_virq_to_irqhandler(VIRQ_TIMER, cpu, xen_timer_interrupt, 399 irq = bind_virq_to_irqhandler(VIRQ_TIMER, cpu, xen_timer_interrupt,
400 IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING|IRQF_TIMER, 400 IRQF_DISABLED|IRQF_PERCPU|
401 IRQF_NOBALANCING|IRQF_TIMER|
402 IRQF_FORCE_RESUME,
401 name, NULL); 403 name, NULL);
402 404
403 evt = &per_cpu(xen_clock_events, cpu); 405 evt = &per_cpu(xen_clock_events, cpu);
diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h
index 9d41bf985757..3112f55638c4 100644
--- a/arch/x86/xen/xen-ops.h
+++ b/arch/x86/xen/xen-ops.h
@@ -64,10 +64,12 @@ void xen_setup_vcpu_info_placement(void);
64 64
65#ifdef CONFIG_SMP 65#ifdef CONFIG_SMP
66void xen_smp_init(void); 66void xen_smp_init(void);
67void __init xen_hvm_smp_init(void);
67 68
68extern cpumask_var_t xen_cpu_initialized_map; 69extern cpumask_var_t xen_cpu_initialized_map;
69#else 70#else
70static inline void xen_smp_init(void) {} 71static inline void xen_smp_init(void) {}
72static inline void xen_hvm_smp_init(void) {}
71#endif 73#endif
72 74
73#ifdef CONFIG_PARAVIRT_SPINLOCKS 75#ifdef CONFIG_PARAVIRT_SPINLOCKS
generated by cgit v1.2.2 (git 2.25.0) at 2025-01-09 13:24:08 -0500