aboutsummaryrefslogtreecommitdiffstats
path: root/arch/x86
diff options
context:
space:
mode:
authorIngo Molnar <mingo@elte.hu>2009-03-03 20:29:19 -0500
committerIngo Molnar <mingo@elte.hu>2009-03-03 20:29:19 -0500
commit91d75e209bd59695f0708d66964d928d45b3b2f3 (patch)
tree32cab1359d951e4193bebb181a0f0319824a2b95 /arch/x86
parent9976b39b5031bbf76f715893cf080b6a17683881 (diff)
parent8b0e5860cb099d7958d13b00ffbc35ad02735700 (diff)
Merge branch 'x86/core' into core/percpu
Diffstat (limited to 'arch/x86')
-rw-r--r--arch/x86/Kconfig5
-rw-r--r--arch/x86/boot/compressed/Makefile21
-rw-r--r--arch/x86/boot/compressed/misc.c118
-rw-r--r--arch/x86/configs/i386_defconfig6
-rw-r--r--arch/x86/configs/x86_64_defconfig6
-rw-r--r--arch/x86/include/asm/apic.h26
-rw-r--r--arch/x86/include/asm/boot.h16
-rw-r--r--arch/x86/include/asm/fixmap.h149
-rw-r--r--arch/x86/include/asm/fixmap_32.h115
-rw-r--r--arch/x86/include/asm/fixmap_64.h79
-rw-r--r--arch/x86/include/asm/iomap.h3
-rw-r--r--arch/x86/include/asm/numa_32.h6
-rw-r--r--arch/x86/include/asm/pat.h3
-rw-r--r--arch/x86/include/asm/processor.h6
-rw-r--r--arch/x86/include/asm/seccomp_32.h6
-rw-r--r--arch/x86/include/asm/seccomp_64.h8
-rw-r--r--arch/x86/include/asm/setup.h7
-rw-r--r--arch/x86/include/asm/system.h3
-rw-r--r--arch/x86/include/asm/uaccess_64.h26
-rw-r--r--arch/x86/include/asm/uv/uv.h3
-rw-r--r--arch/x86/kernel/Makefile2
-rw-r--r--arch/x86/kernel/alternative.c6
-rw-r--r--arch/x86/kernel/apic/apic_flat_64.c2
-rw-r--r--arch/x86/kernel/apic/bigsmp_32.c41
-rw-r--r--arch/x86/kernel/apic/es7000_32.c221
-rw-r--r--arch/x86/kernel/apic/numaq_32.c12
-rw-r--r--arch/x86/kernel/apic/probe_32.c15
-rw-r--r--arch/x86/kernel/apic/probe_64.c3
-rw-r--r--arch/x86/kernel/apic/summit_32.c102
-rw-r--r--arch/x86/kernel/apic/x2apic_cluster.c1
-rw-r--r--arch/x86/kernel/apic/x2apic_phys.c1
-rw-r--r--arch/x86/kernel/apic/x2apic_uv_x.c44
-rw-r--r--arch/x86/kernel/cpu/proc.c20
-rw-r--r--arch/x86/kernel/e820.c3
-rw-r--r--arch/x86/kernel/ioport.c11
-rw-r--r--arch/x86/kernel/process.c191
-rw-r--r--arch/x86/kernel/process_32.c190
-rw-r--r--arch/x86/kernel/process_64.c188
-rw-r--r--arch/x86/kernel/ptrace.c2
-rw-r--r--arch/x86/kernel/setup.c16
-rw-r--r--arch/x86/kernel/signal.c117
-rw-r--r--arch/x86/kernel/smpboot.c19
-rw-r--r--arch/x86/kernel/traps.c46
-rw-r--r--arch/x86/kernel/vsmp_64.c12
-rw-r--r--arch/x86/mm/Makefile2
-rw-r--r--arch/x86/mm/highmem_32.c34
-rw-r--r--arch/x86/mm/init.c49
-rw-r--r--arch/x86/mm/init_32.c61
-rw-r--r--arch/x86/mm/init_64.c39
-rw-r--r--arch/x86/mm/iomap_32.c11
-rw-r--r--arch/x86/mm/memtest.c156
-rw-r--r--arch/x86/mm/numa_32.c26
-rw-r--r--arch/x86/mm/pat.c48
-rw-r--r--arch/x86/mm/pgtable.c18
-rw-r--r--arch/x86/mm/pgtable_32.c18
-rw-r--r--arch/x86/oprofile/op_model_ppro.c14
-rw-r--r--arch/x86/xen/enlighten.c3
57 files changed, 974 insertions, 1382 deletions
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 8015641478bd..f5cef3fbf9a5 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -40,6 +40,9 @@ config X86
40 select HAVE_GENERIC_DMA_COHERENT if X86_32 40 select HAVE_GENERIC_DMA_COHERENT if X86_32
41 select HAVE_EFFICIENT_UNALIGNED_ACCESS 41 select HAVE_EFFICIENT_UNALIGNED_ACCESS
42 select USER_STACKTRACE_SUPPORT 42 select USER_STACKTRACE_SUPPORT
43 select HAVE_KERNEL_GZIP
44 select HAVE_KERNEL_BZIP2
45 select HAVE_KERNEL_LZMA
43 46
44config ARCH_DEFCONFIG 47config ARCH_DEFCONFIG
45 string 48 string
@@ -1825,7 +1828,7 @@ config DMAR
1825 remapping devices. 1828 remapping devices.
1826 1829
1827config DMAR_DEFAULT_ON 1830config DMAR_DEFAULT_ON
1828 def_bool n 1831 def_bool y
1829 prompt "Enable DMA Remapping Devices by default" 1832 prompt "Enable DMA Remapping Devices by default"
1830 depends on DMAR 1833 depends on DMAR
1831 help 1834 help
diff --git a/arch/x86/boot/compressed/Makefile b/arch/x86/boot/compressed/Makefile
index 1771c804e02f..3ca4c194b8e5 100644
--- a/arch/x86/boot/compressed/Makefile
+++ b/arch/x86/boot/compressed/Makefile
@@ -4,7 +4,7 @@
4# create a compressed vmlinux image from the original vmlinux 4# create a compressed vmlinux image from the original vmlinux
5# 5#
6 6
7targets := vmlinux vmlinux.bin vmlinux.bin.gz head_$(BITS).o misc.o piggy.o 7targets := vmlinux vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2 vmlinux.bin.lzma head_$(BITS).o misc.o piggy.o
8 8
9KBUILD_CFLAGS := -m$(BITS) -D__KERNEL__ $(LINUX_INCLUDE) -O2 9KBUILD_CFLAGS := -m$(BITS) -D__KERNEL__ $(LINUX_INCLUDE) -O2
10KBUILD_CFLAGS += -fno-strict-aliasing -fPIC 10KBUILD_CFLAGS += -fno-strict-aliasing -fPIC
@@ -47,18 +47,35 @@ ifeq ($(CONFIG_X86_32),y)
47ifdef CONFIG_RELOCATABLE 47ifdef CONFIG_RELOCATABLE
48$(obj)/vmlinux.bin.gz: $(obj)/vmlinux.bin.all FORCE 48$(obj)/vmlinux.bin.gz: $(obj)/vmlinux.bin.all FORCE
49 $(call if_changed,gzip) 49 $(call if_changed,gzip)
50$(obj)/vmlinux.bin.bz2: $(obj)/vmlinux.bin.all FORCE
51 $(call if_changed,bzip2)
52$(obj)/vmlinux.bin.lzma: $(obj)/vmlinux.bin.all FORCE
53 $(call if_changed,lzma)
50else 54else
51$(obj)/vmlinux.bin.gz: $(obj)/vmlinux.bin FORCE 55$(obj)/vmlinux.bin.gz: $(obj)/vmlinux.bin FORCE
52 $(call if_changed,gzip) 56 $(call if_changed,gzip)
57$(obj)/vmlinux.bin.bz2: $(obj)/vmlinux.bin FORCE
58 $(call if_changed,bzip2)
59$(obj)/vmlinux.bin.lzma: $(obj)/vmlinux.bin FORCE
60 $(call if_changed,lzma)
53endif 61endif
54LDFLAGS_piggy.o := -r --format binary --oformat elf32-i386 -T 62LDFLAGS_piggy.o := -r --format binary --oformat elf32-i386 -T
55 63
56else 64else
65
57$(obj)/vmlinux.bin.gz: $(obj)/vmlinux.bin FORCE 66$(obj)/vmlinux.bin.gz: $(obj)/vmlinux.bin FORCE
58 $(call if_changed,gzip) 67 $(call if_changed,gzip)
68$(obj)/vmlinux.bin.bz2: $(obj)/vmlinux.bin FORCE
69 $(call if_changed,bzip2)
70$(obj)/vmlinux.bin.lzma: $(obj)/vmlinux.bin FORCE
71 $(call if_changed,lzma)
59 72
60LDFLAGS_piggy.o := -r --format binary --oformat elf64-x86-64 -T 73LDFLAGS_piggy.o := -r --format binary --oformat elf64-x86-64 -T
61endif 74endif
62 75
63$(obj)/piggy.o: $(obj)/vmlinux.scr $(obj)/vmlinux.bin.gz FORCE 76suffix_$(CONFIG_KERNEL_GZIP) = gz
77suffix_$(CONFIG_KERNEL_BZIP2) = bz2
78suffix_$(CONFIG_KERNEL_LZMA) = lzma
79
80$(obj)/piggy.o: $(obj)/vmlinux.scr $(obj)/vmlinux.bin.$(suffix_y) FORCE
64 $(call if_changed,ld) 81 $(call if_changed,ld)
diff --git a/arch/x86/boot/compressed/misc.c b/arch/x86/boot/compressed/misc.c
index da062216948a..e45be73684ff 100644
--- a/arch/x86/boot/compressed/misc.c
+++ b/arch/x86/boot/compressed/misc.c
@@ -116,71 +116,13 @@
116/* 116/*
117 * gzip declarations 117 * gzip declarations
118 */ 118 */
119
120#define OF(args) args
121#define STATIC static 119#define STATIC static
122 120
123#undef memset 121#undef memset
124#undef memcpy 122#undef memcpy
125#define memzero(s, n) memset((s), 0, (n)) 123#define memzero(s, n) memset((s), 0, (n))
126 124
127typedef unsigned char uch;
128typedef unsigned short ush;
129typedef unsigned long ulg;
130
131/*
132 * Window size must be at least 32k, and a power of two.
133 * We don't actually have a window just a huge output buffer,
134 * so we report a 2G window size, as that should always be
135 * larger than our output buffer:
136 */
137#define WSIZE 0x80000000
138
139/* Input buffer: */
140static unsigned char *inbuf;
141
142/* Sliding window buffer (and final output buffer): */
143static unsigned char *window;
144
145/* Valid bytes in inbuf: */
146static unsigned insize;
147
148/* Index of next byte to be processed in inbuf: */
149static unsigned inptr;
150
151/* Bytes in output buffer: */
152static unsigned outcnt;
153
154/* gzip flag byte */
155#define ASCII_FLAG 0x01 /* bit 0 set: file probably ASCII text */
156#define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gz file */
157#define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
158#define ORIG_NAM 0x08 /* bit 3 set: original file name present */
159#define COMMENT 0x10 /* bit 4 set: file comment present */
160#define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */
161#define RESERVED 0xC0 /* bit 6, 7: reserved */
162
163#define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf())
164
165/* Diagnostic functions */
166#ifdef DEBUG
167# define Assert(cond, msg) do { if (!(cond)) error(msg); } while (0)
168# define Trace(x) do { fprintf x; } while (0)
169# define Tracev(x) do { if (verbose) fprintf x ; } while (0)
170# define Tracevv(x) do { if (verbose > 1) fprintf x ; } while (0)
171# define Tracec(c, x) do { if (verbose && (c)) fprintf x ; } while (0)
172# define Tracecv(c, x) do { if (verbose > 1 && (c)) fprintf x ; } while (0)
173#else
174# define Assert(cond, msg)
175# define Trace(x)
176# define Tracev(x)
177# define Tracevv(x)
178# define Tracec(c, x)
179# define Tracecv(c, x)
180#endif
181 125
182static int fill_inbuf(void);
183static void flush_window(void);
184static void error(char *m); 126static void error(char *m);
185 127
186/* 128/*
@@ -189,13 +131,8 @@ static void error(char *m);
189static struct boot_params *real_mode; /* Pointer to real-mode data */ 131static struct boot_params *real_mode; /* Pointer to real-mode data */
190static int quiet; 132static int quiet;
191 133
192extern unsigned char input_data[];
193extern int input_len;
194
195static long bytes_out;
196
197static void *memset(void *s, int c, unsigned n); 134static void *memset(void *s, int c, unsigned n);
198static void *memcpy(void *dest, const void *src, unsigned n); 135void *memcpy(void *dest, const void *src, unsigned n);
199 136
200static void __putstr(int, const char *); 137static void __putstr(int, const char *);
201#define putstr(__x) __putstr(0, __x) 138#define putstr(__x) __putstr(0, __x)
@@ -213,7 +150,17 @@ static char *vidmem;
213static int vidport; 150static int vidport;
214static int lines, cols; 151static int lines, cols;
215 152
216#include "../../../../lib/inflate.c" 153#ifdef CONFIG_KERNEL_GZIP
154#include "../../../../lib/decompress_inflate.c"
155#endif
156
157#ifdef CONFIG_KERNEL_BZIP2
158#include "../../../../lib/decompress_bunzip2.c"
159#endif
160
161#ifdef CONFIG_KERNEL_LZMA
162#include "../../../../lib/decompress_unlzma.c"
163#endif
217 164
218static void scroll(void) 165static void scroll(void)
219{ 166{
@@ -282,7 +229,7 @@ static void *memset(void *s, int c, unsigned n)
282 return s; 229 return s;
283} 230}
284 231
285static void *memcpy(void *dest, const void *src, unsigned n) 232void *memcpy(void *dest, const void *src, unsigned n)
286{ 233{
287 int i; 234 int i;
288 const char *s = src; 235 const char *s = src;
@@ -293,38 +240,6 @@ static void *memcpy(void *dest, const void *src, unsigned n)
293 return dest; 240 return dest;
294} 241}
295 242
296/* ===========================================================================
297 * Fill the input buffer. This is called only when the buffer is empty
298 * and at least one byte is really needed.
299 */
300static int fill_inbuf(void)
301{
302 error("ran out of input data");
303 return 0;
304}
305
306/* ===========================================================================
307 * Write the output window window[0..outcnt-1] and update crc and bytes_out.
308 * (Used for the decompressed data only.)
309 */
310static void flush_window(void)
311{
312 /* With my window equal to my output buffer
313 * I only need to compute the crc here.
314 */
315 unsigned long c = crc; /* temporary variable */
316 unsigned n;
317 unsigned char *in, ch;
318
319 in = window;
320 for (n = 0; n < outcnt; n++) {
321 ch = *in++;
322 c = crc_32_tab[((int)c ^ ch) & 0xff] ^ (c >> 8);
323 }
324 crc = c;
325 bytes_out += (unsigned long)outcnt;
326 outcnt = 0;
327}
328 243
329static void error(char *x) 244static void error(char *x)
330{ 245{
@@ -407,12 +322,8 @@ asmlinkage void decompress_kernel(void *rmode, memptr heap,
407 lines = real_mode->screen_info.orig_video_lines; 322 lines = real_mode->screen_info.orig_video_lines;
408 cols = real_mode->screen_info.orig_video_cols; 323 cols = real_mode->screen_info.orig_video_cols;
409 324
410 window = output; /* Output buffer (Normally at 1M) */
411 free_mem_ptr = heap; /* Heap */ 325 free_mem_ptr = heap; /* Heap */
412 free_mem_end_ptr = heap + BOOT_HEAP_SIZE; 326 free_mem_end_ptr = heap + BOOT_HEAP_SIZE;
413 inbuf = input_data; /* Input buffer */
414 insize = input_len;
415 inptr = 0;
416 327
417#ifdef CONFIG_X86_64 328#ifdef CONFIG_X86_64
418 if ((unsigned long)output & (__KERNEL_ALIGN - 1)) 329 if ((unsigned long)output & (__KERNEL_ALIGN - 1))
@@ -430,10 +341,9 @@ asmlinkage void decompress_kernel(void *rmode, memptr heap,
430#endif 341#endif
431#endif 342#endif
432 343
433 makecrc();
434 if (!quiet) 344 if (!quiet)
435 putstr("\nDecompressing Linux... "); 345 putstr("\nDecompressing Linux... ");
436 gunzip(); 346 decompress(input_data, input_len, NULL, NULL, output, NULL, error);
437 parse_elf(output); 347 parse_elf(output);
438 if (!quiet) 348 if (!quiet)
439 putstr("done.\nBooting the kernel.\n"); 349 putstr("done.\nBooting the kernel.\n");
diff --git a/arch/x86/configs/i386_defconfig b/arch/x86/configs/i386_defconfig
index 5c023f6f652c..235b81d0f6f2 100644
--- a/arch/x86/configs/i386_defconfig
+++ b/arch/x86/configs/i386_defconfig
@@ -1,7 +1,7 @@
1# 1#
2# Automatically generated make config: don't edit 2# Automatically generated make config: don't edit
3# Linux kernel version: 2.6.29-rc4 3# Linux kernel version: 2.6.29-rc4
4# Thu Feb 12 12:57:57 2009 4# Tue Feb 24 15:50:58 2009
5# 5#
6# CONFIG_64BIT is not set 6# CONFIG_64BIT is not set
7CONFIG_X86_32=y 7CONFIG_X86_32=y
@@ -266,7 +266,9 @@ CONFIG_PREEMPT_VOLUNTARY=y
266CONFIG_X86_LOCAL_APIC=y 266CONFIG_X86_LOCAL_APIC=y
267CONFIG_X86_IO_APIC=y 267CONFIG_X86_IO_APIC=y
268CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS=y 268CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS=y
269# CONFIG_X86_MCE is not set 269CONFIG_X86_MCE=y
270CONFIG_X86_MCE_NONFATAL=y
271CONFIG_X86_MCE_P4THERMAL=y
270CONFIG_VM86=y 272CONFIG_VM86=y
271# CONFIG_TOSHIBA is not set 273# CONFIG_TOSHIBA is not set
272# CONFIG_I8K is not set 274# CONFIG_I8K is not set
diff --git a/arch/x86/configs/x86_64_defconfig b/arch/x86/configs/x86_64_defconfig
index 4157cc4a2bde..9fe5d212ab4c 100644
--- a/arch/x86/configs/x86_64_defconfig
+++ b/arch/x86/configs/x86_64_defconfig
@@ -1,7 +1,7 @@
1# 1#
2# Automatically generated make config: don't edit 2# Automatically generated make config: don't edit
3# Linux kernel version: 2.6.29-rc4 3# Linux kernel version: 2.6.29-rc4
4# Thu Feb 12 12:57:29 2009 4# Tue Feb 24 15:44:16 2009
5# 5#
6CONFIG_64BIT=y 6CONFIG_64BIT=y
7# CONFIG_X86_32 is not set 7# CONFIG_X86_32 is not set
@@ -266,7 +266,9 @@ CONFIG_PREEMPT_VOLUNTARY=y
266CONFIG_X86_LOCAL_APIC=y 266CONFIG_X86_LOCAL_APIC=y
267CONFIG_X86_IO_APIC=y 267CONFIG_X86_IO_APIC=y
268CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS=y 268CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS=y
269# CONFIG_X86_MCE is not set 269CONFIG_X86_MCE=y
270CONFIG_X86_MCE_INTEL=y
271CONFIG_X86_MCE_AMD=y
270# CONFIG_I8K is not set 272# CONFIG_I8K is not set
271CONFIG_MICROCODE=y 273CONFIG_MICROCODE=y
272CONFIG_MICROCODE_INTEL=y 274CONFIG_MICROCODE_INTEL=y
diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index a6208dc74633..4ef949c1972e 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -75,7 +75,14 @@ static inline void default_inquire_remote_apic(int apicid)
75#define setup_secondary_clock setup_secondary_APIC_clock 75#define setup_secondary_clock setup_secondary_APIC_clock
76#endif 76#endif
77 77
78#ifdef CONFIG_X86_VSMP
78extern int is_vsmp_box(void); 79extern int is_vsmp_box(void);
80#else
81static inline int is_vsmp_box(void)
82{
83 return 0;
84}
85#endif
79extern void xapic_wait_icr_idle(void); 86extern void xapic_wait_icr_idle(void);
80extern u32 safe_xapic_wait_icr_idle(void); 87extern u32 safe_xapic_wait_icr_idle(void);
81extern void xapic_icr_write(u32, u32); 88extern void xapic_icr_write(u32, u32);
@@ -306,7 +313,7 @@ struct apic {
306 void (*send_IPI_self)(int vector); 313 void (*send_IPI_self)(int vector);
307 314
308 /* wakeup_secondary_cpu */ 315 /* wakeup_secondary_cpu */
309 int (*wakeup_cpu)(int apicid, unsigned long start_eip); 316 int (*wakeup_secondary_cpu)(int apicid, unsigned long start_eip);
310 317
311 int trampoline_phys_low; 318 int trampoline_phys_low;
312 int trampoline_phys_high; 319 int trampoline_phys_high;
@@ -324,8 +331,21 @@ struct apic {
324 u32 (*safe_wait_icr_idle)(void); 331 u32 (*safe_wait_icr_idle)(void);
325}; 332};
326 333
334/*
335 * Pointer to the local APIC driver in use on this system (there's
336 * always just one such driver in use - the kernel decides via an
337 * early probing process which one it picks - and then sticks to it):
338 */
327extern struct apic *apic; 339extern struct apic *apic;
328 340
341/*
342 * APIC functionality to boot other CPUs - only used on SMP:
343 */
344#ifdef CONFIG_SMP
345extern atomic_t init_deasserted;
346extern int wakeup_secondary_cpu_via_nmi(int apicid, unsigned long start_eip);
347#endif
348
329static inline u32 apic_read(u32 reg) 349static inline u32 apic_read(u32 reg)
330{ 350{
331 return apic->read(reg); 351 return apic->read(reg);
@@ -384,9 +404,7 @@ static inline unsigned default_get_apic_id(unsigned long x)
384#define DEFAULT_TRAMPOLINE_PHYS_LOW 0x467 404#define DEFAULT_TRAMPOLINE_PHYS_LOW 0x467
385#define DEFAULT_TRAMPOLINE_PHYS_HIGH 0x469 405#define DEFAULT_TRAMPOLINE_PHYS_HIGH 0x469
386 406
387#ifdef CONFIG_X86_32 407#ifdef CONFIG_X86_64
388extern void es7000_update_apic_to_cluster(void);
389#else
390extern struct apic apic_flat; 408extern struct apic apic_flat;
391extern struct apic apic_physflat; 409extern struct apic apic_physflat;
392extern struct apic apic_x2apic_cluster; 410extern struct apic apic_x2apic_cluster;
diff --git a/arch/x86/include/asm/boot.h b/arch/x86/include/asm/boot.h
index dd61616cb73d..6526cf08b0e4 100644
--- a/arch/x86/include/asm/boot.h
+++ b/arch/x86/include/asm/boot.h
@@ -10,17 +10,31 @@
10#define EXTENDED_VGA 0xfffe /* 80x50 mode */ 10#define EXTENDED_VGA 0xfffe /* 80x50 mode */
11#define ASK_VGA 0xfffd /* ask for it at bootup */ 11#define ASK_VGA 0xfffd /* ask for it at bootup */
12 12
13#ifdef __KERNEL__
14
13/* Physical address where kernel should be loaded. */ 15/* Physical address where kernel should be loaded. */
14#define LOAD_PHYSICAL_ADDR ((CONFIG_PHYSICAL_START \ 16#define LOAD_PHYSICAL_ADDR ((CONFIG_PHYSICAL_START \
15 + (CONFIG_PHYSICAL_ALIGN - 1)) \ 17 + (CONFIG_PHYSICAL_ALIGN - 1)) \
16 & ~(CONFIG_PHYSICAL_ALIGN - 1)) 18 & ~(CONFIG_PHYSICAL_ALIGN - 1))
17 19
20#ifdef CONFIG_KERNEL_BZIP2
21#define BOOT_HEAP_SIZE 0x400000
22#else /* !CONFIG_KERNEL_BZIP2 */
23
18#ifdef CONFIG_X86_64 24#ifdef CONFIG_X86_64
19#define BOOT_HEAP_SIZE 0x7000 25#define BOOT_HEAP_SIZE 0x7000
20#define BOOT_STACK_SIZE 0x4000
21#else 26#else
22#define BOOT_HEAP_SIZE 0x4000 27#define BOOT_HEAP_SIZE 0x4000
28#endif
29
30#endif /* !CONFIG_KERNEL_BZIP2 */
31
32#ifdef CONFIG_X86_64
33#define BOOT_STACK_SIZE 0x4000
34#else
23#define BOOT_STACK_SIZE 0x1000 35#define BOOT_STACK_SIZE 0x1000
24#endif 36#endif
25 37
38#endif /* __KERNEL__ */
39
26#endif /* _ASM_X86_BOOT_H */ 40#endif /* _ASM_X86_BOOT_H */
diff --git a/arch/x86/include/asm/fixmap.h b/arch/x86/include/asm/fixmap.h
index 23696d44a0af..dca8f03da5b2 100644
--- a/arch/x86/include/asm/fixmap.h
+++ b/arch/x86/include/asm/fixmap.h
@@ -1,11 +1,155 @@
1/*
2 * fixmap.h: compile-time virtual memory allocation
3 *
4 * This file is subject to the terms and conditions of the GNU General Public
5 * License. See the file "COPYING" in the main directory of this archive
6 * for more details.
7 *
8 * Copyright (C) 1998 Ingo Molnar
9 *
10 * Support of BIGMEM added by Gerhard Wichert, Siemens AG, July 1999
11 * x86_32 and x86_64 integration by Gustavo F. Padovan, February 2009
12 */
13
1#ifndef _ASM_X86_FIXMAP_H 14#ifndef _ASM_X86_FIXMAP_H
2#define _ASM_X86_FIXMAP_H 15#define _ASM_X86_FIXMAP_H
3 16
17#ifndef __ASSEMBLY__
18#include <linux/kernel.h>
19#include <asm/acpi.h>
20#include <asm/apicdef.h>
21#include <asm/page.h>
22#ifdef CONFIG_X86_32
23#include <linux/threads.h>
24#include <asm/kmap_types.h>
25#else
26#include <asm/vsyscall.h>
27#ifdef CONFIG_EFI
28#include <asm/efi.h>
29#endif
30#endif
31
32/*
33 * We can't declare FIXADDR_TOP as variable for x86_64 because vsyscall
34 * uses fixmaps that relies on FIXADDR_TOP for proper address calculation.
35 * Because of this, FIXADDR_TOP x86 integration was left as later work.
36 */
37#ifdef CONFIG_X86_32
38/* used by vmalloc.c, vsyscall.lds.S.
39 *
40 * Leave one empty page between vmalloc'ed areas and
41 * the start of the fixmap.
42 */
43extern unsigned long __FIXADDR_TOP;
44#define FIXADDR_TOP ((unsigned long)__FIXADDR_TOP)
45
46#define FIXADDR_USER_START __fix_to_virt(FIX_VDSO)
47#define FIXADDR_USER_END __fix_to_virt(FIX_VDSO - 1)
48#else
49#define FIXADDR_TOP (VSYSCALL_END-PAGE_SIZE)
50
51/* Only covers 32bit vsyscalls currently. Need another set for 64bit. */
52#define FIXADDR_USER_START ((unsigned long)VSYSCALL32_VSYSCALL)
53#define FIXADDR_USER_END (FIXADDR_USER_START + PAGE_SIZE)
54#endif
55
56
57/*
58 * Here we define all the compile-time 'special' virtual
59 * addresses. The point is to have a constant address at
60 * compile time, but to set the physical address only
61 * in the boot process.
62 * for x86_32: We allocate these special addresses
63 * from the end of virtual memory (0xfffff000) backwards.
64 * Also this lets us do fail-safe vmalloc(), we
65 * can guarantee that these special addresses and
66 * vmalloc()-ed addresses never overlap.
67 *
68 * These 'compile-time allocated' memory buffers are
69 * fixed-size 4k pages (or larger if used with an increment
70 * higher than 1). Use set_fixmap(idx,phys) to associate
71 * physical memory with fixmap indices.
72 *
73 * TLB entries of such buffers will not be flushed across
74 * task switches.
75 */
76enum fixed_addresses {
4#ifdef CONFIG_X86_32 77#ifdef CONFIG_X86_32
5# include "fixmap_32.h" 78 FIX_HOLE,
79 FIX_VDSO,
6#else 80#else
7# include "fixmap_64.h" 81 VSYSCALL_LAST_PAGE,
82 VSYSCALL_FIRST_PAGE = VSYSCALL_LAST_PAGE
83 + ((VSYSCALL_END-VSYSCALL_START) >> PAGE_SHIFT) - 1,
84 VSYSCALL_HPET,
8#endif 85#endif
86 FIX_DBGP_BASE,
87 FIX_EARLYCON_MEM_BASE,
88#ifdef CONFIG_X86_LOCAL_APIC
89 FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */
90#endif
91#ifdef CONFIG_X86_IO_APIC
92 FIX_IO_APIC_BASE_0,
93 FIX_IO_APIC_BASE_END = FIX_IO_APIC_BASE_0 + MAX_IO_APICS - 1,
94#endif
95#ifdef CONFIG_X86_64
96#ifdef CONFIG_EFI
97 FIX_EFI_IO_MAP_LAST_PAGE,
98 FIX_EFI_IO_MAP_FIRST_PAGE = FIX_EFI_IO_MAP_LAST_PAGE
99 + MAX_EFI_IO_PAGES - 1,
100#endif
101#endif
102#ifdef CONFIG_X86_VISWS_APIC
103 FIX_CO_CPU, /* Cobalt timer */
104 FIX_CO_APIC, /* Cobalt APIC Redirection Table */
105 FIX_LI_PCIA, /* Lithium PCI Bridge A */
106 FIX_LI_PCIB, /* Lithium PCI Bridge B */
107#endif
108#ifdef CONFIG_X86_F00F_BUG
109 FIX_F00F_IDT, /* Virtual mapping for IDT */
110#endif
111#ifdef CONFIG_X86_CYCLONE_TIMER
112 FIX_CYCLONE_TIMER, /*cyclone timer register*/
113#endif
114#ifdef CONFIG_X86_32
115 FIX_KMAP_BEGIN, /* reserved pte's for temporary kernel mappings */
116 FIX_KMAP_END = FIX_KMAP_BEGIN+(KM_TYPE_NR*NR_CPUS)-1,
117#ifdef CONFIG_PCI_MMCONFIG
118 FIX_PCIE_MCFG,
119#endif
120#endif
121#ifdef CONFIG_PARAVIRT
122 FIX_PARAVIRT_BOOTMAP,
123#endif
124 __end_of_permanent_fixed_addresses,
125#ifdef CONFIG_PROVIDE_OHCI1394_DMA_INIT
126 FIX_OHCI1394_BASE,
127#endif
128 /*
129 * 256 temporary boot-time mappings, used by early_ioremap(),
130 * before ioremap() is functional.
131 *
132 * We round it up to the next 256 pages boundary so that we
133 * can have a single pgd entry and a single pte table:
134 */
135#define NR_FIX_BTMAPS 64
136#define FIX_BTMAPS_SLOTS 4
137 FIX_BTMAP_END = __end_of_permanent_fixed_addresses + 256 -
138 (__end_of_permanent_fixed_addresses & 255),
139 FIX_BTMAP_BEGIN = FIX_BTMAP_END + NR_FIX_BTMAPS*FIX_BTMAPS_SLOTS - 1,
140#ifdef CONFIG_X86_32
141 FIX_WP_TEST,
142#endif
143 __end_of_fixed_addresses
144};
145
146
147extern void reserve_top_address(unsigned long reserve);
148
149#define FIXADDR_SIZE (__end_of_permanent_fixed_addresses << PAGE_SHIFT)
150#define FIXADDR_BOOT_SIZE (__end_of_fixed_addresses << PAGE_SHIFT)
151#define FIXADDR_START (FIXADDR_TOP - FIXADDR_SIZE)
152#define FIXADDR_BOOT_START (FIXADDR_TOP - FIXADDR_BOOT_SIZE)
9 153
10extern int fixmaps_set; 154extern int fixmaps_set;
11 155
@@ -69,4 +213,5 @@ static inline unsigned long virt_to_fix(const unsigned long vaddr)
69 BUG_ON(vaddr >= FIXADDR_TOP || vaddr < FIXADDR_START); 213 BUG_ON(vaddr >= FIXADDR_TOP || vaddr < FIXADDR_START);
70 return __virt_to_fix(vaddr); 214 return __virt_to_fix(vaddr);
71} 215}
216#endif /* !__ASSEMBLY__ */
72#endif /* _ASM_X86_FIXMAP_H */ 217#endif /* _ASM_X86_FIXMAP_H */
diff --git a/arch/x86/include/asm/fixmap_32.h b/arch/x86/include/asm/fixmap_32.h
deleted file mode 100644
index 047d9bab2b31..000000000000
--- a/arch/x86/include/asm/fixmap_32.h
+++ /dev/null
@@ -1,115 +0,0 @@
1/*
2 * fixmap.h: compile-time virtual memory allocation
3 *
4 * This file is subject to the terms and conditions of the GNU General Public
5 * License. See the file "COPYING" in the main directory of this archive
6 * for more details.
7 *
8 * Copyright (C) 1998 Ingo Molnar
9 *
10 * Support of BIGMEM added by Gerhard Wichert, Siemens AG, July 1999
11 */
12
13#ifndef _ASM_X86_FIXMAP_32_H
14#define _ASM_X86_FIXMAP_32_H
15
16
17/* used by vmalloc.c, vsyscall.lds.S.
18 *
19 * Leave one empty page between vmalloc'ed areas and
20 * the start of the fixmap.
21 */
22extern unsigned long __FIXADDR_TOP;
23#define FIXADDR_USER_START __fix_to_virt(FIX_VDSO)
24#define FIXADDR_USER_END __fix_to_virt(FIX_VDSO - 1)
25
26#ifndef __ASSEMBLY__
27#include <linux/kernel.h>
28#include <asm/acpi.h>
29#include <asm/apicdef.h>
30#include <asm/page.h>
31#include <linux/threads.h>
32#include <asm/kmap_types.h>
33
34/*
35 * Here we define all the compile-time 'special' virtual
36 * addresses. The point is to have a constant address at
37 * compile time, but to set the physical address only
38 * in the boot process. We allocate these special addresses
39 * from the end of virtual memory (0xfffff000) backwards.
40 * Also this lets us do fail-safe vmalloc(), we
41 * can guarantee that these special addresses and
42 * vmalloc()-ed addresses never overlap.
43 *
44 * these 'compile-time allocated' memory buffers are
45 * fixed-size 4k pages. (or larger if used with an increment
46 * highger than 1) use fixmap_set(idx,phys) to associate
47 * physical memory with fixmap indices.
48 *
49 * TLB entries of such buffers will not be flushed across
50 * task switches.
51 */
52enum fixed_addresses {
53 FIX_HOLE,
54 FIX_VDSO,
55 FIX_DBGP_BASE,
56 FIX_EARLYCON_MEM_BASE,
57#ifdef CONFIG_X86_LOCAL_APIC
58 FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */
59#endif
60#ifdef CONFIG_X86_IO_APIC
61 FIX_IO_APIC_BASE_0,
62 FIX_IO_APIC_BASE_END = FIX_IO_APIC_BASE_0 + MAX_IO_APICS-1,
63#endif
64#ifdef CONFIG_X86_VISWS_APIC
65 FIX_CO_CPU, /* Cobalt timer */
66 FIX_CO_APIC, /* Cobalt APIC Redirection Table */
67 FIX_LI_PCIA, /* Lithium PCI Bridge A */
68 FIX_LI_PCIB, /* Lithium PCI Bridge B */
69#endif
70#ifdef CONFIG_X86_F00F_BUG
71 FIX_F00F_IDT, /* Virtual mapping for IDT */
72#endif
73#ifdef CONFIG_X86_CYCLONE_TIMER
74 FIX_CYCLONE_TIMER, /*cyclone timer register*/
75#endif
76 FIX_KMAP_BEGIN, /* reserved pte's for temporary kernel mappings */
77 FIX_KMAP_END = FIX_KMAP_BEGIN+(KM_TYPE_NR*NR_CPUS)-1,
78#ifdef CONFIG_PCI_MMCONFIG
79 FIX_PCIE_MCFG,
80#endif
81#ifdef CONFIG_PARAVIRT
82 FIX_PARAVIRT_BOOTMAP,
83#endif
84 __end_of_permanent_fixed_addresses,
85 /*
86 * 256 temporary boot-time mappings, used by early_ioremap(),
87 * before ioremap() is functional.
88 *
89 * We round it up to the next 256 pages boundary so that we
90 * can have a single pgd entry and a single pte table:
91 */
92#define NR_FIX_BTMAPS 64
93#define FIX_BTMAPS_SLOTS 4
94 FIX_BTMAP_END = __end_of_permanent_fixed_addresses + 256 -
95 (__end_of_permanent_fixed_addresses & 255),
96 FIX_BTMAP_BEGIN = FIX_BTMAP_END + NR_FIX_BTMAPS*FIX_BTMAPS_SLOTS - 1,
97 FIX_WP_TEST,
98#ifdef CONFIG_PROVIDE_OHCI1394_DMA_INIT
99 FIX_OHCI1394_BASE,
100#endif
101 __end_of_fixed_addresses
102};
103
104extern void reserve_top_address(unsigned long reserve);
105
106
107#define FIXADDR_TOP ((unsigned long)__FIXADDR_TOP)
108
109#define __FIXADDR_SIZE (__end_of_permanent_fixed_addresses << PAGE_SHIFT)
110#define __FIXADDR_BOOT_SIZE (__end_of_fixed_addresses << PAGE_SHIFT)
111#define FIXADDR_START (FIXADDR_TOP - __FIXADDR_SIZE)
112#define FIXADDR_BOOT_START (FIXADDR_TOP - __FIXADDR_BOOT_SIZE)
113
114#endif /* !__ASSEMBLY__ */
115#endif /* _ASM_X86_FIXMAP_32_H */
diff --git a/arch/x86/include/asm/fixmap_64.h b/arch/x86/include/asm/fixmap_64.h
deleted file mode 100644
index 298d9ba3faeb..000000000000
--- a/arch/x86/include/asm/fixmap_64.h
+++ /dev/null
@@ -1,79 +0,0 @@
1/*
2 * fixmap.h: compile-time virtual memory allocation
3 *
4 * This file is subject to the terms and conditions of the GNU General Public
5 * License. See the file "COPYING" in the main directory of this archive
6 * for more details.
7 *
8 * Copyright (C) 1998 Ingo Molnar
9 */
10
11#ifndef _ASM_X86_FIXMAP_64_H
12#define _ASM_X86_FIXMAP_64_H
13
14#include <linux/kernel.h>
15#include <asm/acpi.h>
16#include <asm/apicdef.h>
17#include <asm/page.h>
18#include <asm/vsyscall.h>
19#include <asm/efi.h>
20
21/*
22 * Here we define all the compile-time 'special' virtual
23 * addresses. The point is to have a constant address at
24 * compile time, but to set the physical address only
25 * in the boot process.
26 *
27 * These 'compile-time allocated' memory buffers are
28 * fixed-size 4k pages (or larger if used with an increment
29 * higher than 1). Use set_fixmap(idx,phys) to associate
30 * physical memory with fixmap indices.
31 *
32 * TLB entries of such buffers will not be flushed across
33 * task switches.
34 */
35
36enum fixed_addresses {
37 VSYSCALL_LAST_PAGE,
38 VSYSCALL_FIRST_PAGE = VSYSCALL_LAST_PAGE
39 + ((VSYSCALL_END-VSYSCALL_START) >> PAGE_SHIFT) - 1,
40 VSYSCALL_HPET,
41 FIX_DBGP_BASE,
42 FIX_EARLYCON_MEM_BASE,
43 FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */
44 FIX_IO_APIC_BASE_0,
45 FIX_IO_APIC_BASE_END = FIX_IO_APIC_BASE_0 + MAX_IO_APICS - 1,
46 FIX_EFI_IO_MAP_LAST_PAGE,
47 FIX_EFI_IO_MAP_FIRST_PAGE = FIX_EFI_IO_MAP_LAST_PAGE
48 + MAX_EFI_IO_PAGES - 1,
49#ifdef CONFIG_PARAVIRT
50 FIX_PARAVIRT_BOOTMAP,
51#endif
52 __end_of_permanent_fixed_addresses,
53#ifdef CONFIG_PROVIDE_OHCI1394_DMA_INIT
54 FIX_OHCI1394_BASE,
55#endif
56 /*
57 * 256 temporary boot-time mappings, used by early_ioremap(),
58 * before ioremap() is functional.
59 *
60 * We round it up to the next 256 pages boundary so that we
61 * can have a single pgd entry and a single pte table:
62 */
63#define NR_FIX_BTMAPS 64
64#define FIX_BTMAPS_SLOTS 4
65 FIX_BTMAP_END = __end_of_permanent_fixed_addresses + 256 -
66 (__end_of_permanent_fixed_addresses & 255),
67 FIX_BTMAP_BEGIN = FIX_BTMAP_END + NR_FIX_BTMAPS*FIX_BTMAPS_SLOTS - 1,
68 __end_of_fixed_addresses
69};
70
71#define FIXADDR_TOP (VSYSCALL_END-PAGE_SIZE)
72#define FIXADDR_SIZE (__end_of_fixed_addresses << PAGE_SHIFT)
73#define FIXADDR_START (FIXADDR_TOP - FIXADDR_SIZE)
74
75/* Only covers 32bit vsyscalls currently. Need another set for 64bit. */
76#define FIXADDR_USER_START ((unsigned long)VSYSCALL32_VSYSCALL)
77#define FIXADDR_USER_END (FIXADDR_USER_START + PAGE_SIZE)
78
79#endif /* _ASM_X86_FIXMAP_64_H */
diff --git a/arch/x86/include/asm/iomap.h b/arch/x86/include/asm/iomap.h
index c1f06289b14b..86af26091d6c 100644
--- a/arch/x86/include/asm/iomap.h
+++ b/arch/x86/include/asm/iomap.h
@@ -23,6 +23,9 @@
23#include <asm/pgtable.h> 23#include <asm/pgtable.h>
24#include <asm/tlbflush.h> 24#include <asm/tlbflush.h>
25 25
26int
27is_io_mapping_possible(resource_size_t base, unsigned long size);
28
26void * 29void *
27iomap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot); 30iomap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot);
28 31
diff --git a/arch/x86/include/asm/numa_32.h b/arch/x86/include/asm/numa_32.h
index e9f5db796244..a37229011b56 100644
--- a/arch/x86/include/asm/numa_32.h
+++ b/arch/x86/include/asm/numa_32.h
@@ -4,8 +4,12 @@
4extern int pxm_to_nid(int pxm); 4extern int pxm_to_nid(int pxm);
5extern void numa_remove_cpu(int cpu); 5extern void numa_remove_cpu(int cpu);
6 6
7#ifdef CONFIG_NUMA 7#ifdef CONFIG_HIGHMEM
8extern void set_highmem_pages_init(void); 8extern void set_highmem_pages_init(void);
9#else
10static inline void set_highmem_pages_init(void)
11{
12}
9#endif 13#endif
10 14
11#endif /* _ASM_X86_NUMA_32_H */ 15#endif /* _ASM_X86_NUMA_32_H */
diff --git a/arch/x86/include/asm/pat.h b/arch/x86/include/asm/pat.h
index 9709fdff6615..b0e70056838e 100644
--- a/arch/x86/include/asm/pat.h
+++ b/arch/x86/include/asm/pat.h
@@ -15,4 +15,7 @@ extern int reserve_memtype(u64 start, u64 end,
15 unsigned long req_type, unsigned long *ret_type); 15 unsigned long req_type, unsigned long *ret_type);
16extern int free_memtype(u64 start, u64 end); 16extern int free_memtype(u64 start, u64 end);
17 17
18extern int kernel_map_sync_memtype(u64 base, unsigned long size,
19 unsigned long flag);
20
18#endif /* _ASM_X86_PAT_H */ 21#endif /* _ASM_X86_PAT_H */
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index c7a98f738210..76139506c3e4 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -248,7 +248,6 @@ struct x86_hw_tss {
248#define IO_BITMAP_LONGS (IO_BITMAP_BYTES/sizeof(long)) 248#define IO_BITMAP_LONGS (IO_BITMAP_BYTES/sizeof(long))
249#define IO_BITMAP_OFFSET offsetof(struct tss_struct, io_bitmap) 249#define IO_BITMAP_OFFSET offsetof(struct tss_struct, io_bitmap)
250#define INVALID_IO_BITMAP_OFFSET 0x8000 250#define INVALID_IO_BITMAP_OFFSET 0x8000
251#define INVALID_IO_BITMAP_OFFSET_LAZY 0x9000
252 251
253struct tss_struct { 252struct tss_struct {
254 /* 253 /*
@@ -263,11 +262,6 @@ struct tss_struct {
263 * be within the limit. 262 * be within the limit.
264 */ 263 */
265 unsigned long io_bitmap[IO_BITMAP_LONGS + 1]; 264 unsigned long io_bitmap[IO_BITMAP_LONGS + 1];
266 /*
267 * Cache the current maximum and the last task that used the bitmap:
268 */
269 unsigned long io_bitmap_max;
270 struct thread_struct *io_bitmap_owner;
271 265
272 /* 266 /*
273 * .. and then another 0x100 bytes for the emergency kernel stack: 267 * .. and then another 0x100 bytes for the emergency kernel stack:
diff --git a/arch/x86/include/asm/seccomp_32.h b/arch/x86/include/asm/seccomp_32.h
index a6ad87b352c4..b811d6f5780c 100644
--- a/arch/x86/include/asm/seccomp_32.h
+++ b/arch/x86/include/asm/seccomp_32.h
@@ -1,12 +1,6 @@
1#ifndef _ASM_X86_SECCOMP_32_H 1#ifndef _ASM_X86_SECCOMP_32_H
2#define _ASM_X86_SECCOMP_32_H 2#define _ASM_X86_SECCOMP_32_H
3 3
4#include <linux/thread_info.h>
5
6#ifdef TIF_32BIT
7#error "unexpected TIF_32BIT on i386"
8#endif
9
10#include <linux/unistd.h> 4#include <linux/unistd.h>
11 5
12#define __NR_seccomp_read __NR_read 6#define __NR_seccomp_read __NR_read
diff --git a/arch/x86/include/asm/seccomp_64.h b/arch/x86/include/asm/seccomp_64.h
index 4171bb794e9e..84ec1bd161a5 100644
--- a/arch/x86/include/asm/seccomp_64.h
+++ b/arch/x86/include/asm/seccomp_64.h
@@ -1,14 +1,6 @@
1#ifndef _ASM_X86_SECCOMP_64_H 1#ifndef _ASM_X86_SECCOMP_64_H
2#define _ASM_X86_SECCOMP_64_H 2#define _ASM_X86_SECCOMP_64_H
3 3
4#include <linux/thread_info.h>
5
6#ifdef TIF_32BIT
7#error "unexpected TIF_32BIT on x86_64"
8#else
9#define TIF_32BIT TIF_IA32
10#endif
11
12#include <linux/unistd.h> 4#include <linux/unistd.h>
13#include <asm/ia32_unistd.h> 5#include <asm/ia32_unistd.h>
14 6
diff --git a/arch/x86/include/asm/setup.h b/arch/x86/include/asm/setup.h
index 66801cb72f69..05c6f6b11fd5 100644
--- a/arch/x86/include/asm/setup.h
+++ b/arch/x86/include/asm/setup.h
@@ -31,7 +31,6 @@ struct x86_quirks {
31 void (*smp_read_mpc_oem)(struct mpc_oemtable *oemtable, 31 void (*smp_read_mpc_oem)(struct mpc_oemtable *oemtable,
32 unsigned short oemsize); 32 unsigned short oemsize);
33 int (*setup_ioapic_ids)(void); 33 int (*setup_ioapic_ids)(void);
34 int (*update_apic)(void);
35}; 34};
36 35
37extern void x86_quirk_pre_intr_init(void); 36extern void x86_quirk_pre_intr_init(void);
@@ -65,7 +64,11 @@ extern void x86_quirk_time_init(void);
65#include <asm/bootparam.h> 64#include <asm/bootparam.h>
66 65
67/* Interrupt control for vSMPowered x86_64 systems */ 66/* Interrupt control for vSMPowered x86_64 systems */
67#ifdef CONFIG_X86_VSMP
68void vsmp_init(void); 68void vsmp_init(void);
69#else
70static inline void vsmp_init(void) { }
71#endif
69 72
70void setup_bios_corruption_check(void); 73void setup_bios_corruption_check(void);
71 74
@@ -77,8 +80,6 @@ static inline void visws_early_detect(void) { }
77static inline int is_visws_box(void) { return 0; } 80static inline int is_visws_box(void) { return 0; }
78#endif 81#endif
79 82
80extern int wakeup_secondary_cpu_via_nmi(int apicid, unsigned long start_eip);
81extern int wakeup_secondary_cpu_via_init(int apicid, unsigned long start_eip);
82extern struct x86_quirks *x86_quirks; 83extern struct x86_quirks *x86_quirks;
83extern unsigned long saved_video_mode; 84extern unsigned long saved_video_mode;
84 85
diff --git a/arch/x86/include/asm/system.h b/arch/x86/include/asm/system.h
index c00bfdbdd456..643c59b4bc6e 100644
--- a/arch/x86/include/asm/system.h
+++ b/arch/x86/include/asm/system.h
@@ -20,6 +20,9 @@
20struct task_struct; /* one of the stranger aspects of C forward declarations */ 20struct task_struct; /* one of the stranger aspects of C forward declarations */
21struct task_struct *__switch_to(struct task_struct *prev, 21struct task_struct *__switch_to(struct task_struct *prev,
22 struct task_struct *next); 22 struct task_struct *next);
23struct tss_struct;
24void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p,
25 struct tss_struct *tss);
23 26
24#ifdef CONFIG_X86_32 27#ifdef CONFIG_X86_32
25 28
diff --git a/arch/x86/include/asm/uaccess_64.h b/arch/x86/include/asm/uaccess_64.h
index 987a2c10fe20..8cc687326eb8 100644
--- a/arch/x86/include/asm/uaccess_64.h
+++ b/arch/x86/include/asm/uaccess_64.h
@@ -188,30 +188,18 @@ __copy_to_user_inatomic(void __user *dst, const void *src, unsigned size)
188extern long __copy_user_nocache(void *dst, const void __user *src, 188extern long __copy_user_nocache(void *dst, const void __user *src,
189 unsigned size, int zerorest); 189 unsigned size, int zerorest);
190 190
191static inline int __copy_from_user_nocache(void *dst, const void __user *src, 191static inline int
192 unsigned size) 192__copy_from_user_nocache(void *dst, const void __user *src, unsigned size)
193{ 193{
194 might_sleep(); 194 might_sleep();
195 /* 195 return __copy_user_nocache(dst, src, size, 1);
196 * In practice this limit means that large file write()s
197 * which get chunked to 4K copies get handled via
198 * non-temporal stores here. Smaller writes get handled
199 * via regular __copy_from_user():
200 */
201 if (likely(size >= PAGE_SIZE))
202 return __copy_user_nocache(dst, src, size, 1);
203 else
204 return __copy_from_user(dst, src, size);
205} 196}
206 197
207static inline int __copy_from_user_inatomic_nocache(void *dst, 198static inline int
208 const void __user *src, 199__copy_from_user_inatomic_nocache(void *dst, const void __user *src,
209 unsigned size) 200 unsigned size)
210{ 201{
211 if (likely(size >= PAGE_SIZE)) 202 return __copy_user_nocache(dst, src, size, 0);
212 return __copy_user_nocache(dst, src, size, 0);
213 else
214 return __copy_from_user_inatomic(dst, src, size);
215} 203}
216 204
217unsigned long 205unsigned long
diff --git a/arch/x86/include/asm/uv/uv.h b/arch/x86/include/asm/uv/uv.h
index 8242bf965812..c0a01b5d985b 100644
--- a/arch/x86/include/asm/uv/uv.h
+++ b/arch/x86/include/asm/uv/uv.h
@@ -12,7 +12,6 @@ extern enum uv_system_type get_uv_system_type(void);
12extern int is_uv_system(void); 12extern int is_uv_system(void);
13extern void uv_cpu_init(void); 13extern void uv_cpu_init(void);
14extern void uv_system_init(void); 14extern void uv_system_init(void);
15extern int uv_wakeup_secondary(int phys_apicid, unsigned int start_rip);
16extern const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask, 15extern const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
17 struct mm_struct *mm, 16 struct mm_struct *mm,
18 unsigned long va, 17 unsigned long va,
@@ -24,8 +23,6 @@ static inline enum uv_system_type get_uv_system_type(void) { return UV_NONE; }
24static inline int is_uv_system(void) { return 0; } 23static inline int is_uv_system(void) { return 0; }
25static inline void uv_cpu_init(void) { } 24static inline void uv_cpu_init(void) { }
26static inline void uv_system_init(void) { } 25static inline void uv_system_init(void) { }
27static inline int uv_wakeup_secondary(int phys_apicid, unsigned int start_rip)
28{ return 1; }
29static inline const struct cpumask * 26static inline const struct cpumask *
30uv_flush_tlb_others(const struct cpumask *cpumask, struct mm_struct *mm, 27uv_flush_tlb_others(const struct cpumask *cpumask, struct mm_struct *mm,
31 unsigned long va, unsigned int cpu) 28 unsigned long va, unsigned int cpu)
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index de5657c039e9..95f216bbfaf1 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -70,7 +70,7 @@ obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o
70obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o 70obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o
71obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o 71obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o
72obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o 72obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o
73obj-y += vsmp_64.o 73obj-$(CONFIG_X86_VSMP) += vsmp_64.o
74obj-$(CONFIG_KPROBES) += kprobes.o 74obj-$(CONFIG_KPROBES) += kprobes.o
75obj-$(CONFIG_MODULES) += module_$(BITS).o 75obj-$(CONFIG_MODULES) += module_$(BITS).o
76obj-$(CONFIG_EFI) += efi.o efi_$(BITS).o efi_stub_$(BITS).o 76obj-$(CONFIG_EFI) += efi.o efi_$(BITS).o efi_stub_$(BITS).o
diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c
index a84ac7b570e6..6907b8e85d52 100644
--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -498,12 +498,12 @@ void *text_poke_early(void *addr, const void *opcode, size_t len)
498 */ 498 */
499void *__kprobes text_poke(void *addr, const void *opcode, size_t len) 499void *__kprobes text_poke(void *addr, const void *opcode, size_t len)
500{ 500{
501 unsigned long flags;
502 char *vaddr; 501 char *vaddr;
503 int nr_pages = 2; 502 int nr_pages = 2;
504 struct page *pages[2]; 503 struct page *pages[2];
505 int i; 504 int i;
506 505
506 might_sleep();
507 if (!core_kernel_text((unsigned long)addr)) { 507 if (!core_kernel_text((unsigned long)addr)) {
508 pages[0] = vmalloc_to_page(addr); 508 pages[0] = vmalloc_to_page(addr);
509 pages[1] = vmalloc_to_page(addr + PAGE_SIZE); 509 pages[1] = vmalloc_to_page(addr + PAGE_SIZE);
@@ -517,9 +517,9 @@ void *__kprobes text_poke(void *addr, const void *opcode, size_t len)
517 nr_pages = 1; 517 nr_pages = 1;
518 vaddr = vmap(pages, nr_pages, VM_MAP, PAGE_KERNEL); 518 vaddr = vmap(pages, nr_pages, VM_MAP, PAGE_KERNEL);
519 BUG_ON(!vaddr); 519 BUG_ON(!vaddr);
520 local_irq_save(flags); 520 local_irq_disable();
521 memcpy(&vaddr[(unsigned long)addr & ~PAGE_MASK], opcode, len); 521 memcpy(&vaddr[(unsigned long)addr & ~PAGE_MASK], opcode, len);
522 local_irq_restore(flags); 522 local_irq_enable();
523 vunmap(vaddr); 523 vunmap(vaddr);
524 sync_core(); 524 sync_core();
525 /* Could also do a CLFLUSH here to speed up CPU recovery; but 525 /* Could also do a CLFLUSH here to speed up CPU recovery; but
diff --git a/arch/x86/kernel/apic/apic_flat_64.c b/arch/x86/kernel/apic/apic_flat_64.c
index 3b002995e145..f933822dba18 100644
--- a/arch/x86/kernel/apic/apic_flat_64.c
+++ b/arch/x86/kernel/apic/apic_flat_64.c
@@ -222,7 +222,6 @@ struct apic apic_flat = {
222 .send_IPI_all = flat_send_IPI_all, 222 .send_IPI_all = flat_send_IPI_all,
223 .send_IPI_self = apic_send_IPI_self, 223 .send_IPI_self = apic_send_IPI_self,
224 224
225 .wakeup_cpu = NULL,
226 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW, 225 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW,
227 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 226 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
228 .wait_for_init_deassert = NULL, 227 .wait_for_init_deassert = NULL,
@@ -373,7 +372,6 @@ struct apic apic_physflat = {
373 .send_IPI_all = physflat_send_IPI_all, 372 .send_IPI_all = physflat_send_IPI_all,
374 .send_IPI_self = apic_send_IPI_self, 373 .send_IPI_self = apic_send_IPI_self,
375 374
376 .wakeup_cpu = NULL,
377 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW, 375 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW,
378 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 376 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
379 .wait_for_init_deassert = NULL, 377 .wait_for_init_deassert = NULL,
diff --git a/arch/x86/kernel/apic/bigsmp_32.c b/arch/x86/kernel/apic/bigsmp_32.c
index 0b1093394fdf..d806ecaa948f 100644
--- a/arch/x86/kernel/apic/bigsmp_32.c
+++ b/arch/x86/kernel/apic/bigsmp_32.c
@@ -16,17 +16,17 @@
16#include <asm/apic.h> 16#include <asm/apic.h>
17#include <asm/ipi.h> 17#include <asm/ipi.h>
18 18
19static inline unsigned bigsmp_get_apic_id(unsigned long x) 19static unsigned bigsmp_get_apic_id(unsigned long x)
20{ 20{
21 return (x >> 24) & 0xFF; 21 return (x >> 24) & 0xFF;
22} 22}
23 23
24static inline int bigsmp_apic_id_registered(void) 24static int bigsmp_apic_id_registered(void)
25{ 25{
26 return 1; 26 return 1;
27} 27}
28 28
29static inline const cpumask_t *bigsmp_target_cpus(void) 29static const cpumask_t *bigsmp_target_cpus(void)
30{ 30{
31#ifdef CONFIG_SMP 31#ifdef CONFIG_SMP
32 return &cpu_online_map; 32 return &cpu_online_map;
@@ -35,13 +35,12 @@ static inline const cpumask_t *bigsmp_target_cpus(void)
35#endif 35#endif
36} 36}
37 37
38static inline unsigned long 38static unsigned long bigsmp_check_apicid_used(physid_mask_t bitmap, int apicid)
39bigsmp_check_apicid_used(physid_mask_t bitmap, int apicid)
40{ 39{
41 return 0; 40 return 0;
42} 41}
43 42
44static inline unsigned long bigsmp_check_apicid_present(int bit) 43static unsigned long bigsmp_check_apicid_present(int bit)
45{ 44{
46 return 1; 45 return 1;
47} 46}
@@ -64,7 +63,7 @@ static inline unsigned long calculate_ldr(int cpu)
64 * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel 63 * an APIC. See e.g. "AP-388 82489DX User's Manual" (Intel
65 * document number 292116). So here it goes... 64 * document number 292116). So here it goes...
66 */ 65 */
67static inline void bigsmp_init_apic_ldr(void) 66static void bigsmp_init_apic_ldr(void)
68{ 67{
69 unsigned long val; 68 unsigned long val;
70 int cpu = smp_processor_id(); 69 int cpu = smp_processor_id();
@@ -74,19 +73,19 @@ static inline void bigsmp_init_apic_ldr(void)
74 apic_write(APIC_LDR, val); 73 apic_write(APIC_LDR, val);
75} 74}
76 75
77static inline void bigsmp_setup_apic_routing(void) 76static void bigsmp_setup_apic_routing(void)
78{ 77{
79 printk(KERN_INFO 78 printk(KERN_INFO
80 "Enabling APIC mode: Physflat. Using %d I/O APICs\n", 79 "Enabling APIC mode: Physflat. Using %d I/O APICs\n",
81 nr_ioapics); 80 nr_ioapics);
82} 81}
83 82
84static inline int bigsmp_apicid_to_node(int logical_apicid) 83static int bigsmp_apicid_to_node(int logical_apicid)
85{ 84{
86 return apicid_2_node[hard_smp_processor_id()]; 85 return apicid_2_node[hard_smp_processor_id()];
87} 86}
88 87
89static inline int bigsmp_cpu_present_to_apicid(int mps_cpu) 88static int bigsmp_cpu_present_to_apicid(int mps_cpu)
90{ 89{
91 if (mps_cpu < nr_cpu_ids) 90 if (mps_cpu < nr_cpu_ids)
92 return (int) per_cpu(x86_bios_cpu_apicid, mps_cpu); 91 return (int) per_cpu(x86_bios_cpu_apicid, mps_cpu);
@@ -94,7 +93,7 @@ static inline int bigsmp_cpu_present_to_apicid(int mps_cpu)
94 return BAD_APICID; 93 return BAD_APICID;
95} 94}
96 95
97static inline physid_mask_t bigsmp_apicid_to_cpu_present(int phys_apicid) 96static physid_mask_t bigsmp_apicid_to_cpu_present(int phys_apicid)
98{ 97{
99 return physid_mask_of_physid(phys_apicid); 98 return physid_mask_of_physid(phys_apicid);
100} 99}
@@ -107,29 +106,24 @@ static inline int bigsmp_cpu_to_logical_apicid(int cpu)
107 return cpu_physical_id(cpu); 106 return cpu_physical_id(cpu);
108} 107}
109 108
110static inline physid_mask_t bigsmp_ioapic_phys_id_map(physid_mask_t phys_map) 109static physid_mask_t bigsmp_ioapic_phys_id_map(physid_mask_t phys_map)
111{ 110{
112 /* For clustered we don't have a good way to do this yet - hack */ 111 /* For clustered we don't have a good way to do this yet - hack */
113 return physids_promote(0xFFL); 112 return physids_promote(0xFFL);
114} 113}
115 114
116static inline void bigsmp_setup_portio_remap(void) 115static int bigsmp_check_phys_apicid_present(int boot_cpu_physical_apicid)
117{
118}
119
120static inline int bigsmp_check_phys_apicid_present(int boot_cpu_physical_apicid)
121{ 116{
122 return 1; 117 return 1;
123} 118}
124 119
125/* As we are using single CPU as destination, pick only one CPU here */ 120/* As we are using single CPU as destination, pick only one CPU here */
126static inline unsigned int bigsmp_cpu_mask_to_apicid(const cpumask_t *cpumask) 121static unsigned int bigsmp_cpu_mask_to_apicid(const cpumask_t *cpumask)
127{ 122{
128 return bigsmp_cpu_to_logical_apicid(first_cpu(*cpumask)); 123 return bigsmp_cpu_to_logical_apicid(first_cpu(*cpumask));
129} 124}
130 125
131static inline unsigned int 126static unsigned int bigsmp_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
132bigsmp_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
133 const struct cpumask *andmask) 127 const struct cpumask *andmask)
134{ 128{
135 int cpu; 129 int cpu;
@@ -148,7 +142,7 @@ bigsmp_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
148 return BAD_APICID; 142 return BAD_APICID;
149} 143}
150 144
151static inline int bigsmp_phys_pkg_id(int cpuid_apic, int index_msb) 145static int bigsmp_phys_pkg_id(int cpuid_apic, int index_msb)
152{ 146{
153 return cpuid_apic >> index_msb; 147 return cpuid_apic >> index_msb;
154} 148}
@@ -158,12 +152,12 @@ static inline void bigsmp_send_IPI_mask(const struct cpumask *mask, int vector)
158 default_send_IPI_mask_sequence_phys(mask, vector); 152 default_send_IPI_mask_sequence_phys(mask, vector);
159} 153}
160 154
161static inline void bigsmp_send_IPI_allbutself(int vector) 155static void bigsmp_send_IPI_allbutself(int vector)
162{ 156{
163 default_send_IPI_mask_allbutself_phys(cpu_online_mask, vector); 157 default_send_IPI_mask_allbutself_phys(cpu_online_mask, vector);
164} 158}
165 159
166static inline void bigsmp_send_IPI_all(int vector) 160static void bigsmp_send_IPI_all(int vector)
167{ 161{
168 bigsmp_send_IPI_mask(cpu_online_mask, vector); 162 bigsmp_send_IPI_mask(cpu_online_mask, vector);
169} 163}
@@ -256,7 +250,6 @@ struct apic apic_bigsmp = {
256 .send_IPI_all = bigsmp_send_IPI_all, 250 .send_IPI_all = bigsmp_send_IPI_all,
257 .send_IPI_self = default_send_IPI_self, 251 .send_IPI_self = default_send_IPI_self,
258 252
259 .wakeup_cpu = NULL,
260 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW, 253 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW,
261 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 254 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
262 255
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index 320f2d2e4e54..19588f2770ee 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -163,22 +163,17 @@ static int wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip)
163 return 0; 163 return 0;
164} 164}
165 165
166static int __init es7000_update_apic(void) 166static int es7000_apic_is_cluster(void)
167{ 167{
168 apic->wakeup_cpu = wakeup_secondary_cpu_via_mip;
169
170 /* MPENTIUMIII */ 168 /* MPENTIUMIII */
171 if (boot_cpu_data.x86 == 6 && 169 if (boot_cpu_data.x86 == 6 &&
172 (boot_cpu_data.x86_model >= 7 || boot_cpu_data.x86_model <= 11)) { 170 (boot_cpu_data.x86_model >= 7 || boot_cpu_data.x86_model <= 11))
173 es7000_update_apic_to_cluster(); 171 return 1;
174 apic->wait_for_init_deassert = NULL;
175 apic->wakeup_cpu = wakeup_secondary_cpu_via_mip;
176 }
177 172
178 return 0; 173 return 0;
179} 174}
180 175
181static void __init setup_unisys(void) 176static void setup_unisys(void)
182{ 177{
183 /* 178 /*
184 * Determine the generation of the ES7000 currently running. 179 * Determine the generation of the ES7000 currently running.
@@ -192,14 +187,12 @@ static void __init setup_unisys(void)
192 else 187 else
193 es7000_plat = ES7000_CLASSIC; 188 es7000_plat = ES7000_CLASSIC;
194 ioapic_renumber_irq = es7000_rename_gsi; 189 ioapic_renumber_irq = es7000_rename_gsi;
195
196 x86_quirks->update_apic = es7000_update_apic;
197} 190}
198 191
199/* 192/*
200 * Parse the OEM Table: 193 * Parse the OEM Table:
201 */ 194 */
202static int __init parse_unisys_oem(char *oemptr) 195static int parse_unisys_oem(char *oemptr)
203{ 196{
204 int i; 197 int i;
205 int success = 0; 198 int success = 0;
@@ -261,7 +254,7 @@ static int __init parse_unisys_oem(char *oemptr)
261} 254}
262 255
263#ifdef CONFIG_ACPI 256#ifdef CONFIG_ACPI
264static int __init find_unisys_acpi_oem_table(unsigned long *oem_addr) 257static int find_unisys_acpi_oem_table(unsigned long *oem_addr)
265{ 258{
266 struct acpi_table_header *header = NULL; 259 struct acpi_table_header *header = NULL;
267 struct es7000_oem_table *table; 260 struct es7000_oem_table *table;
@@ -292,7 +285,7 @@ static int __init find_unisys_acpi_oem_table(unsigned long *oem_addr)
292 return 0; 285 return 0;
293} 286}
294 287
295static void __init unmap_unisys_acpi_oem_table(unsigned long oem_addr) 288static void unmap_unisys_acpi_oem_table(unsigned long oem_addr)
296{ 289{
297 if (!oem_addr) 290 if (!oem_addr)
298 return; 291 return;
@@ -310,8 +303,10 @@ static int es7000_check_dsdt(void)
310 return 0; 303 return 0;
311} 304}
312 305
306static int es7000_acpi_ret;
307
313/* Hook from generic ACPI tables.c */ 308/* Hook from generic ACPI tables.c */
314static int __init es7000_acpi_madt_oem_check(char *oem_id, char *oem_table_id) 309static int es7000_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
315{ 310{
316 unsigned long oem_addr = 0; 311 unsigned long oem_addr = 0;
317 int check_dsdt; 312 int check_dsdt;
@@ -332,10 +327,26 @@ static int __init es7000_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
332 */ 327 */
333 unmap_unisys_acpi_oem_table(oem_addr); 328 unmap_unisys_acpi_oem_table(oem_addr);
334 } 329 }
335 return ret; 330
331 es7000_acpi_ret = ret;
332
333 return ret && !es7000_apic_is_cluster();
336} 334}
335
336static int es7000_acpi_madt_oem_check_cluster(char *oem_id, char *oem_table_id)
337{
338 int ret = es7000_acpi_ret;
339
340 return ret && es7000_apic_is_cluster();
341}
342
337#else /* !CONFIG_ACPI: */ 343#else /* !CONFIG_ACPI: */
338static int __init es7000_acpi_madt_oem_check(char *oem_id, char *oem_table_id) 344static int es7000_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
345{
346 return 0;
347}
348
349static int es7000_acpi_madt_oem_check_cluster(char *oem_id, char *oem_table_id)
339{ 350{
340 return 0; 351 return 0;
341} 352}
@@ -349,8 +360,7 @@ static void es7000_spin(int n)
349 rep_nop(); 360 rep_nop();
350} 361}
351 362
352static int __init 363static int es7000_mip_write(struct mip_reg *mip_reg)
353es7000_mip_write(struct mip_reg *mip_reg)
354{ 364{
355 int status = 0; 365 int status = 0;
356 int spin; 366 int spin;
@@ -383,7 +393,7 @@ es7000_mip_write(struct mip_reg *mip_reg)
383 return status; 393 return status;
384} 394}
385 395
386static void __init es7000_enable_apic_mode(void) 396static void es7000_enable_apic_mode(void)
387{ 397{
388 struct mip_reg es7000_mip_reg; 398 struct mip_reg es7000_mip_reg;
389 int mip_status; 399 int mip_status;
@@ -416,11 +426,8 @@ static void es7000_vector_allocation_domain(int cpu, cpumask_t *retmask)
416 426
417static void es7000_wait_for_init_deassert(atomic_t *deassert) 427static void es7000_wait_for_init_deassert(atomic_t *deassert)
418{ 428{
419#ifndef CONFIG_ES7000_CLUSTERED_APIC
420 while (!atomic_read(deassert)) 429 while (!atomic_read(deassert))
421 cpu_relax(); 430 cpu_relax();
422#endif
423 return;
424} 431}
425 432
426static unsigned int es7000_get_apic_id(unsigned long x) 433static unsigned int es7000_get_apic_id(unsigned long x)
@@ -565,72 +572,24 @@ static int es7000_check_phys_apicid_present(int cpu_physical_apicid)
565 return 1; 572 return 1;
566} 573}
567 574
568static unsigned int
569es7000_cpu_mask_to_apicid_cluster(const struct cpumask *cpumask)
570{
571 int cpus_found = 0;
572 int num_bits_set;
573 int apicid;
574 int cpu;
575
576 num_bits_set = cpumask_weight(cpumask);
577 /* Return id to all */
578 if (num_bits_set == nr_cpu_ids)
579 return 0xFF;
580 /*
581 * The cpus in the mask must all be on the apic cluster. If are not
582 * on the same apicid cluster return default value of target_cpus():
583 */
584 cpu = cpumask_first(cpumask);
585 apicid = es7000_cpu_to_logical_apicid(cpu);
586
587 while (cpus_found < num_bits_set) {
588 if (cpumask_test_cpu(cpu, cpumask)) {
589 int new_apicid = es7000_cpu_to_logical_apicid(cpu);
590
591 if (APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
592 WARN(1, "Not a valid mask!");
593
594 return 0xFF;
595 }
596 apicid = new_apicid;
597 cpus_found++;
598 }
599 cpu++;
600 }
601 return apicid;
602}
603
604static unsigned int es7000_cpu_mask_to_apicid(const cpumask_t *cpumask) 575static unsigned int es7000_cpu_mask_to_apicid(const cpumask_t *cpumask)
605{ 576{
606 int cpus_found = 0; 577 unsigned int round = 0;
607 int num_bits_set; 578 int cpu, uninitialized_var(apicid);
608 int apicid;
609 int cpu;
610 579
611 num_bits_set = cpus_weight(*cpumask);
612 /* Return id to all */
613 if (num_bits_set == nr_cpu_ids)
614 return es7000_cpu_to_logical_apicid(0);
615 /* 580 /*
616 * The cpus in the mask must all be on the apic cluster. If are not 581 * The cpus in the mask must all be on the apic cluster.
617 * on the same apicid cluster return default value of target_cpus():
618 */ 582 */
619 cpu = first_cpu(*cpumask); 583 for_each_cpu(cpu, cpumask) {
620 apicid = es7000_cpu_to_logical_apicid(cpu); 584 int new_apicid = es7000_cpu_to_logical_apicid(cpu);
621 while (cpus_found < num_bits_set) {
622 if (cpu_isset(cpu, *cpumask)) {
623 int new_apicid = es7000_cpu_to_logical_apicid(cpu);
624 585
625 if (APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) { 586 if (round && APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
626 printk("%s: Not a valid mask!\n", __func__); 587 WARN(1, "Not a valid mask!");
627 588
628 return es7000_cpu_to_logical_apicid(0); 589 return BAD_APICID;
629 }
630 apicid = new_apicid;
631 cpus_found++;
632 } 590 }
633 cpu++; 591 apicid = new_apicid;
592 round++;
634 } 593 }
635 return apicid; 594 return apicid;
636} 595}
@@ -659,37 +618,103 @@ static int es7000_phys_pkg_id(int cpuid_apic, int index_msb)
659 return cpuid_apic >> index_msb; 618 return cpuid_apic >> index_msb;
660} 619}
661 620
662void __init es7000_update_apic_to_cluster(void)
663{
664 apic->target_cpus = target_cpus_cluster;
665 apic->irq_delivery_mode = dest_LowestPrio;
666 /* logical delivery broadcast to all procs: */
667 apic->irq_dest_mode = 1;
668
669 apic->init_apic_ldr = es7000_init_apic_ldr_cluster;
670
671 apic->cpu_mask_to_apicid = es7000_cpu_mask_to_apicid_cluster;
672}
673
674static int probe_es7000(void) 621static int probe_es7000(void)
675{ 622{
676 /* probed later in mptable/ACPI hooks */ 623 /* probed later in mptable/ACPI hooks */
677 return 0; 624 return 0;
678} 625}
679 626
680static __init int 627static int es7000_mps_ret;
681es7000_mps_oem_check(struct mpc_table *mpc, char *oem, char *productid) 628static int es7000_mps_oem_check(struct mpc_table *mpc, char *oem,
629 char *productid)
682{ 630{
631 int ret = 0;
632
683 if (mpc->oemptr) { 633 if (mpc->oemptr) {
684 struct mpc_oemtable *oem_table = 634 struct mpc_oemtable *oem_table =
685 (struct mpc_oemtable *)mpc->oemptr; 635 (struct mpc_oemtable *)mpc->oemptr;
686 636
687 if (!strncmp(oem, "UNISYS", 6)) 637 if (!strncmp(oem, "UNISYS", 6))
688 return parse_unisys_oem((char *)oem_table); 638 ret = parse_unisys_oem((char *)oem_table);
689 } 639 }
690 return 0; 640
641 es7000_mps_ret = ret;
642
643 return ret && !es7000_apic_is_cluster();
691} 644}
692 645
646static int es7000_mps_oem_check_cluster(struct mpc_table *mpc, char *oem,
647 char *productid)
648{
649 int ret = es7000_mps_ret;
650
651 return ret && es7000_apic_is_cluster();
652}
653
654struct apic apic_es7000_cluster = {
655
656 .name = "es7000",
657 .probe = probe_es7000,
658 .acpi_madt_oem_check = es7000_acpi_madt_oem_check_cluster,
659 .apic_id_registered = es7000_apic_id_registered,
660
661 .irq_delivery_mode = dest_LowestPrio,
662 /* logical delivery broadcast to all procs: */
663 .irq_dest_mode = 1,
664
665 .target_cpus = target_cpus_cluster,
666 .disable_esr = 1,
667 .dest_logical = 0,
668 .check_apicid_used = es7000_check_apicid_used,
669 .check_apicid_present = es7000_check_apicid_present,
670
671 .vector_allocation_domain = es7000_vector_allocation_domain,
672 .init_apic_ldr = es7000_init_apic_ldr_cluster,
673
674 .ioapic_phys_id_map = es7000_ioapic_phys_id_map,
675 .setup_apic_routing = es7000_setup_apic_routing,
676 .multi_timer_check = NULL,
677 .apicid_to_node = es7000_apicid_to_node,
678 .cpu_to_logical_apicid = es7000_cpu_to_logical_apicid,
679 .cpu_present_to_apicid = es7000_cpu_present_to_apicid,
680 .apicid_to_cpu_present = es7000_apicid_to_cpu_present,
681 .setup_portio_remap = NULL,
682 .check_phys_apicid_present = es7000_check_phys_apicid_present,
683 .enable_apic_mode = es7000_enable_apic_mode,
684 .phys_pkg_id = es7000_phys_pkg_id,
685 .mps_oem_check = es7000_mps_oem_check_cluster,
686
687 .get_apic_id = es7000_get_apic_id,
688 .set_apic_id = NULL,
689 .apic_id_mask = 0xFF << 24,
690
691 .cpu_mask_to_apicid = es7000_cpu_mask_to_apicid,
692 .cpu_mask_to_apicid_and = es7000_cpu_mask_to_apicid_and,
693
694 .send_IPI_mask = es7000_send_IPI_mask,
695 .send_IPI_mask_allbutself = NULL,
696 .send_IPI_allbutself = es7000_send_IPI_allbutself,
697 .send_IPI_all = es7000_send_IPI_all,
698 .send_IPI_self = default_send_IPI_self,
699
700 .wakeup_secondary_cpu = wakeup_secondary_cpu_via_mip,
701
702 .trampoline_phys_low = 0x467,
703 .trampoline_phys_high = 0x469,
704
705 .wait_for_init_deassert = NULL,
706
707 /* Nothing to do for most platforms, since cleared by the INIT cycle: */
708 .smp_callin_clear_local_apic = NULL,
709 .inquire_remote_apic = default_inquire_remote_apic,
710
711 .read = native_apic_mem_read,
712 .write = native_apic_mem_write,
713 .icr_read = native_apic_icr_read,
714 .icr_write = native_apic_icr_write,
715 .wait_icr_idle = native_apic_wait_icr_idle,
716 .safe_wait_icr_idle = native_safe_apic_wait_icr_idle,
717};
693 718
694struct apic apic_es7000 = { 719struct apic apic_es7000 = {
695 720
@@ -737,8 +762,6 @@ struct apic apic_es7000 = {
737 .send_IPI_all = es7000_send_IPI_all, 762 .send_IPI_all = es7000_send_IPI_all,
738 .send_IPI_self = default_send_IPI_self, 763 .send_IPI_self = default_send_IPI_self,
739 764
740 .wakeup_cpu = NULL,
741
742 .trampoline_phys_low = 0x467, 765 .trampoline_phys_low = 0x467,
743 .trampoline_phys_high = 0x469, 766 .trampoline_phys_high = 0x469,
744 767
diff --git a/arch/x86/kernel/apic/numaq_32.c b/arch/x86/kernel/apic/numaq_32.c
index d9d6d61eed82..ba2fc6465534 100644
--- a/arch/x86/kernel/apic/numaq_32.c
+++ b/arch/x86/kernel/apic/numaq_32.c
@@ -69,7 +69,7 @@ struct mpc_trans {
69/* x86_quirks member */ 69/* x86_quirks member */
70static int mpc_record; 70static int mpc_record;
71 71
72static __cpuinitdata struct mpc_trans *translation_table[MAX_MPC_ENTRY]; 72static struct mpc_trans *translation_table[MAX_MPC_ENTRY];
73 73
74int mp_bus_id_to_node[MAX_MP_BUSSES]; 74int mp_bus_id_to_node[MAX_MP_BUSSES];
75int mp_bus_id_to_local[MAX_MP_BUSSES]; 75int mp_bus_id_to_local[MAX_MP_BUSSES];
@@ -256,13 +256,6 @@ static int __init numaq_setup_ioapic_ids(void)
256 return 1; 256 return 1;
257} 257}
258 258
259static int __init numaq_update_apic(void)
260{
261 apic->wakeup_cpu = wakeup_secondary_cpu_via_nmi;
262
263 return 0;
264}
265
266static struct x86_quirks numaq_x86_quirks __initdata = { 259static struct x86_quirks numaq_x86_quirks __initdata = {
267 .arch_pre_time_init = numaq_pre_time_init, 260 .arch_pre_time_init = numaq_pre_time_init,
268 .arch_time_init = NULL, 261 .arch_time_init = NULL,
@@ -278,7 +271,6 @@ static struct x86_quirks numaq_x86_quirks __initdata = {
278 .mpc_oem_pci_bus = mpc_oem_pci_bus, 271 .mpc_oem_pci_bus = mpc_oem_pci_bus,
279 .smp_read_mpc_oem = smp_read_mpc_oem, 272 .smp_read_mpc_oem = smp_read_mpc_oem,
280 .setup_ioapic_ids = numaq_setup_ioapic_ids, 273 .setup_ioapic_ids = numaq_setup_ioapic_ids,
281 .update_apic = numaq_update_apic,
282}; 274};
283 275
284static __init void early_check_numaq(void) 276static __init void early_check_numaq(void)
@@ -546,7 +538,7 @@ struct apic apic_numaq = {
546 .send_IPI_all = numaq_send_IPI_all, 538 .send_IPI_all = numaq_send_IPI_all,
547 .send_IPI_self = default_send_IPI_self, 539 .send_IPI_self = default_send_IPI_self,
548 540
549 .wakeup_cpu = NULL, 541 .wakeup_secondary_cpu = wakeup_secondary_cpu_via_nmi,
550 .trampoline_phys_low = NUMAQ_TRAMPOLINE_PHYS_LOW, 542 .trampoline_phys_low = NUMAQ_TRAMPOLINE_PHYS_LOW,
551 .trampoline_phys_high = NUMAQ_TRAMPOLINE_PHYS_HIGH, 543 .trampoline_phys_high = NUMAQ_TRAMPOLINE_PHYS_HIGH,
552 544
diff --git a/arch/x86/kernel/apic/probe_32.c b/arch/x86/kernel/apic/probe_32.c
index 3a730fa574bb..141c99a1c264 100644
--- a/arch/x86/kernel/apic/probe_32.c
+++ b/arch/x86/kernel/apic/probe_32.c
@@ -138,7 +138,6 @@ struct apic apic_default = {
138 .send_IPI_all = default_send_IPI_all, 138 .send_IPI_all = default_send_IPI_all,
139 .send_IPI_self = default_send_IPI_self, 139 .send_IPI_self = default_send_IPI_self,
140 140
141 .wakeup_cpu = NULL,
142 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW, 141 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW,
143 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 142 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
144 143
@@ -159,6 +158,7 @@ extern struct apic apic_numaq;
159extern struct apic apic_summit; 158extern struct apic apic_summit;
160extern struct apic apic_bigsmp; 159extern struct apic apic_bigsmp;
161extern struct apic apic_es7000; 160extern struct apic apic_es7000;
161extern struct apic apic_es7000_cluster;
162extern struct apic apic_default; 162extern struct apic apic_default;
163 163
164struct apic *apic = &apic_default; 164struct apic *apic = &apic_default;
@@ -176,6 +176,7 @@ static struct apic *apic_probe[] __initdata = {
176#endif 176#endif
177#ifdef CONFIG_X86_ES7000 177#ifdef CONFIG_X86_ES7000
178 &apic_es7000, 178 &apic_es7000,
179 &apic_es7000_cluster,
179#endif 180#endif
180 &apic_default, /* must be last */ 181 &apic_default, /* must be last */
181 NULL, 182 NULL,
@@ -197,9 +198,6 @@ static int __init parse_apic(char *arg)
197 } 198 }
198 } 199 }
199 200
200 if (x86_quirks->update_apic)
201 x86_quirks->update_apic();
202
203 /* Parsed again by __setup for debug/verbose */ 201 /* Parsed again by __setup for debug/verbose */
204 return 0; 202 return 0;
205} 203}
@@ -218,8 +216,6 @@ void __init generic_bigsmp_probe(void)
218 if (!cmdline_apic && apic == &apic_default) { 216 if (!cmdline_apic && apic == &apic_default) {
219 if (apic_bigsmp.probe()) { 217 if (apic_bigsmp.probe()) {
220 apic = &apic_bigsmp; 218 apic = &apic_bigsmp;
221 if (x86_quirks->update_apic)
222 x86_quirks->update_apic();
223 printk(KERN_INFO "Overriding APIC driver with %s\n", 219 printk(KERN_INFO "Overriding APIC driver with %s\n",
224 apic->name); 220 apic->name);
225 } 221 }
@@ -240,9 +236,6 @@ void __init generic_apic_probe(void)
240 /* Not visible without early console */ 236 /* Not visible without early console */
241 if (!apic_probe[i]) 237 if (!apic_probe[i])
242 panic("Didn't find an APIC driver"); 238 panic("Didn't find an APIC driver");
243
244 if (x86_quirks->update_apic)
245 x86_quirks->update_apic();
246 } 239 }
247 printk(KERN_INFO "Using APIC driver %s\n", apic->name); 240 printk(KERN_INFO "Using APIC driver %s\n", apic->name);
248} 241}
@@ -262,8 +255,6 @@ generic_mps_oem_check(struct mpc_table *mpc, char *oem, char *productid)
262 255
263 if (!cmdline_apic) { 256 if (!cmdline_apic) {
264 apic = apic_probe[i]; 257 apic = apic_probe[i];
265 if (x86_quirks->update_apic)
266 x86_quirks->update_apic();
267 printk(KERN_INFO "Switched to APIC driver `%s'.\n", 258 printk(KERN_INFO "Switched to APIC driver `%s'.\n",
268 apic->name); 259 apic->name);
269 } 260 }
@@ -284,8 +275,6 @@ int __init default_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
284 275
285 if (!cmdline_apic) { 276 if (!cmdline_apic) {
286 apic = apic_probe[i]; 277 apic = apic_probe[i];
287 if (x86_quirks->update_apic)
288 x86_quirks->update_apic();
289 printk(KERN_INFO "Switched to APIC driver `%s'.\n", 278 printk(KERN_INFO "Switched to APIC driver `%s'.\n",
290 apic->name); 279 apic->name);
291 } 280 }
diff --git a/arch/x86/kernel/apic/probe_64.c b/arch/x86/kernel/apic/probe_64.c
index e7c163661c77..8d7748efe6a8 100644
--- a/arch/x86/kernel/apic/probe_64.c
+++ b/arch/x86/kernel/apic/probe_64.c
@@ -68,9 +68,6 @@ void __init default_setup_apic_routing(void)
68 apic = &apic_physflat; 68 apic = &apic_physflat;
69 printk(KERN_INFO "Setting APIC routing to %s\n", apic->name); 69 printk(KERN_INFO "Setting APIC routing to %s\n", apic->name);
70 } 70 }
71
72 if (x86_quirks->update_apic)
73 x86_quirks->update_apic();
74} 71}
75 72
76/* Same for both flat and physical. */ 73/* Same for both flat and physical. */
diff --git a/arch/x86/kernel/apic/summit_32.c b/arch/x86/kernel/apic/summit_32.c
index cfe7b09015d8..aac52fa873ff 100644
--- a/arch/x86/kernel/apic/summit_32.c
+++ b/arch/x86/kernel/apic/summit_32.c
@@ -48,7 +48,7 @@
48#include <linux/gfp.h> 48#include <linux/gfp.h>
49#include <linux/smp.h> 49#include <linux/smp.h>
50 50
51static inline unsigned summit_get_apic_id(unsigned long x) 51static unsigned summit_get_apic_id(unsigned long x)
52{ 52{
53 return (x >> 24) & 0xFF; 53 return (x >> 24) & 0xFF;
54} 54}
@@ -58,7 +58,7 @@ static inline void summit_send_IPI_mask(const cpumask_t *mask, int vector)
58 default_send_IPI_mask_sequence_logical(mask, vector); 58 default_send_IPI_mask_sequence_logical(mask, vector);
59} 59}
60 60
61static inline void summit_send_IPI_allbutself(int vector) 61static void summit_send_IPI_allbutself(int vector)
62{ 62{
63 cpumask_t mask = cpu_online_map; 63 cpumask_t mask = cpu_online_map;
64 cpu_clear(smp_processor_id(), mask); 64 cpu_clear(smp_processor_id(), mask);
@@ -67,7 +67,7 @@ static inline void summit_send_IPI_allbutself(int vector)
67 summit_send_IPI_mask(&mask, vector); 67 summit_send_IPI_mask(&mask, vector);
68} 68}
69 69
70static inline void summit_send_IPI_all(int vector) 70static void summit_send_IPI_all(int vector)
71{ 71{
72 summit_send_IPI_mask(&cpu_online_map, vector); 72 summit_send_IPI_mask(&cpu_online_map, vector);
73} 73}
@@ -77,13 +77,13 @@ static inline void summit_send_IPI_all(int vector)
77extern int use_cyclone; 77extern int use_cyclone;
78 78
79#ifdef CONFIG_X86_SUMMIT_NUMA 79#ifdef CONFIG_X86_SUMMIT_NUMA
80extern void setup_summit(void); 80static void setup_summit(void);
81#else 81#else
82#define setup_summit() {} 82static inline void setup_summit(void) {}
83#endif 83#endif
84 84
85static inline int 85static int summit_mps_oem_check(struct mpc_table *mpc, char *oem,
86summit_mps_oem_check(struct mpc_table *mpc, char *oem, char *productid) 86 char *productid)
87{ 87{
88 if (!strncmp(oem, "IBM ENSW", 8) && 88 if (!strncmp(oem, "IBM ENSW", 8) &&
89 (!strncmp(productid, "VIGIL SMP", 9) 89 (!strncmp(productid, "VIGIL SMP", 9)
@@ -98,7 +98,7 @@ summit_mps_oem_check(struct mpc_table *mpc, char *oem, char *productid)
98} 98}
99 99
100/* Hook from generic ACPI tables.c */ 100/* Hook from generic ACPI tables.c */
101static inline int summit_acpi_madt_oem_check(char *oem_id, char *oem_table_id) 101static int summit_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
102{ 102{
103 if (!strncmp(oem_id, "IBM", 3) && 103 if (!strncmp(oem_id, "IBM", 3) &&
104 (!strncmp(oem_table_id, "SERVIGIL", 8) 104 (!strncmp(oem_table_id, "SERVIGIL", 8)
@@ -186,7 +186,7 @@ static inline int is_WPEG(struct rio_detail *rio){
186 186
187#define SUMMIT_APIC_DFR_VALUE (APIC_DFR_CLUSTER) 187#define SUMMIT_APIC_DFR_VALUE (APIC_DFR_CLUSTER)
188 188
189static inline const cpumask_t *summit_target_cpus(void) 189static const cpumask_t *summit_target_cpus(void)
190{ 190{
191 /* CPU_MASK_ALL (0xff) has undefined behaviour with 191 /* CPU_MASK_ALL (0xff) has undefined behaviour with
192 * dest_LowestPrio mode logical clustered apic interrupt routing 192 * dest_LowestPrio mode logical clustered apic interrupt routing
@@ -195,19 +195,18 @@ static inline const cpumask_t *summit_target_cpus(void)
195 return &cpumask_of_cpu(0); 195 return &cpumask_of_cpu(0);
196} 196}
197 197
198static inline unsigned long 198static unsigned long summit_check_apicid_used(physid_mask_t bitmap, int apicid)
199summit_check_apicid_used(physid_mask_t bitmap, int apicid)
200{ 199{
201 return 0; 200 return 0;
202} 201}
203 202
204/* we don't use the phys_cpu_present_map to indicate apicid presence */ 203/* we don't use the phys_cpu_present_map to indicate apicid presence */
205static inline unsigned long summit_check_apicid_present(int bit) 204static unsigned long summit_check_apicid_present(int bit)
206{ 205{
207 return 1; 206 return 1;
208} 207}
209 208
210static inline void summit_init_apic_ldr(void) 209static void summit_init_apic_ldr(void)
211{ 210{
212 unsigned long val, id; 211 unsigned long val, id;
213 int count = 0; 212 int count = 0;
@@ -234,18 +233,18 @@ static inline void summit_init_apic_ldr(void)
234 apic_write(APIC_LDR, val); 233 apic_write(APIC_LDR, val);
235} 234}
236 235
237static inline int summit_apic_id_registered(void) 236static int summit_apic_id_registered(void)
238{ 237{
239 return 1; 238 return 1;
240} 239}
241 240
242static inline void summit_setup_apic_routing(void) 241static void summit_setup_apic_routing(void)
243{ 242{
244 printk("Enabling APIC mode: Summit. Using %d I/O APICs\n", 243 printk("Enabling APIC mode: Summit. Using %d I/O APICs\n",
245 nr_ioapics); 244 nr_ioapics);
246} 245}
247 246
248static inline int summit_apicid_to_node(int logical_apicid) 247static int summit_apicid_to_node(int logical_apicid)
249{ 248{
250#ifdef CONFIG_SMP 249#ifdef CONFIG_SMP
251 return apicid_2_node[hard_smp_processor_id()]; 250 return apicid_2_node[hard_smp_processor_id()];
@@ -266,7 +265,7 @@ static inline int summit_cpu_to_logical_apicid(int cpu)
266#endif 265#endif
267} 266}
268 267
269static inline int summit_cpu_present_to_apicid(int mps_cpu) 268static int summit_cpu_present_to_apicid(int mps_cpu)
270{ 269{
271 if (mps_cpu < nr_cpu_ids) 270 if (mps_cpu < nr_cpu_ids)
272 return (int)per_cpu(x86_bios_cpu_apicid, mps_cpu); 271 return (int)per_cpu(x86_bios_cpu_apicid, mps_cpu);
@@ -274,64 +273,44 @@ static inline int summit_cpu_present_to_apicid(int mps_cpu)
274 return BAD_APICID; 273 return BAD_APICID;
275} 274}
276 275
277static inline physid_mask_t 276static physid_mask_t summit_ioapic_phys_id_map(physid_mask_t phys_id_map)
278summit_ioapic_phys_id_map(physid_mask_t phys_id_map)
279{ 277{
280 /* For clustered we don't have a good way to do this yet - hack */ 278 /* For clustered we don't have a good way to do this yet - hack */
281 return physids_promote(0x0F); 279 return physids_promote(0x0F);
282} 280}
283 281
284static inline physid_mask_t summit_apicid_to_cpu_present(int apicid) 282static physid_mask_t summit_apicid_to_cpu_present(int apicid)
285{ 283{
286 return physid_mask_of_physid(0); 284 return physid_mask_of_physid(0);
287} 285}
288 286
289static inline void summit_setup_portio_remap(void) 287static int summit_check_phys_apicid_present(int boot_cpu_physical_apicid)
290{
291}
292
293static inline int summit_check_phys_apicid_present(int boot_cpu_physical_apicid)
294{ 288{
295 return 1; 289 return 1;
296} 290}
297 291
298static inline unsigned int summit_cpu_mask_to_apicid(const cpumask_t *cpumask) 292static unsigned int summit_cpu_mask_to_apicid(const cpumask_t *cpumask)
299{ 293{
300 int cpus_found = 0; 294 unsigned int round = 0;
301 int num_bits_set; 295 int cpu, apicid = 0;
302 int apicid; 296
303 int cpu;
304
305 num_bits_set = cpus_weight(*cpumask);
306 /* Return id to all */
307 if (num_bits_set >= nr_cpu_ids)
308 return 0xFF;
309 /* 297 /*
310 * The cpus in the mask must all be on the apic cluster. If are not 298 * The cpus in the mask must all be on the apic cluster.
311 * on the same apicid cluster return default value of target_cpus():
312 */ 299 */
313 cpu = first_cpu(*cpumask); 300 for_each_cpu(cpu, cpumask) {
314 apicid = summit_cpu_to_logical_apicid(cpu); 301 int new_apicid = summit_cpu_to_logical_apicid(cpu);
315 302
316 while (cpus_found < num_bits_set) { 303 if (round && APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
317 if (cpu_isset(cpu, *cpumask)) { 304 printk("%s: Not a valid mask!\n", __func__);
318 int new_apicid = summit_cpu_to_logical_apicid(cpu); 305 return BAD_APICID;
319
320 if (APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
321 printk ("%s: Not a valid mask!\n", __func__);
322
323 return 0xFF;
324 }
325 apicid = apicid | new_apicid;
326 cpus_found++;
327 } 306 }
328 cpu++; 307 apicid |= new_apicid;
308 round++;
329 } 309 }
330 return apicid; 310 return apicid;
331} 311}
332 312
333static inline unsigned int 313static unsigned int summit_cpu_mask_to_apicid_and(const struct cpumask *inmask,
334summit_cpu_mask_to_apicid_and(const struct cpumask *inmask,
335 const struct cpumask *andmask) 314 const struct cpumask *andmask)
336{ 315{
337 int apicid = summit_cpu_to_logical_apicid(0); 316 int apicid = summit_cpu_to_logical_apicid(0);
@@ -356,7 +335,7 @@ summit_cpu_mask_to_apicid_and(const struct cpumask *inmask,
356 * 335 *
357 * See Intel's IA-32 SW Dev's Manual Vol2 under CPUID. 336 * See Intel's IA-32 SW Dev's Manual Vol2 under CPUID.
358 */ 337 */
359static inline int summit_phys_pkg_id(int cpuid_apic, int index_msb) 338static int summit_phys_pkg_id(int cpuid_apic, int index_msb)
360{ 339{
361 return hard_smp_processor_id() >> index_msb; 340 return hard_smp_processor_id() >> index_msb;
362} 341}
@@ -381,15 +360,15 @@ static void summit_vector_allocation_domain(int cpu, cpumask_t *retmask)
381} 360}
382 361
383#ifdef CONFIG_X86_SUMMIT_NUMA 362#ifdef CONFIG_X86_SUMMIT_NUMA
384static struct rio_table_hdr *rio_table_hdr __initdata; 363static struct rio_table_hdr *rio_table_hdr;
385static struct scal_detail *scal_devs[MAX_NUMNODES] __initdata; 364static struct scal_detail *scal_devs[MAX_NUMNODES];
386static struct rio_detail *rio_devs[MAX_NUMNODES*4] __initdata; 365static struct rio_detail *rio_devs[MAX_NUMNODES*4];
387 366
388#ifndef CONFIG_X86_NUMAQ 367#ifndef CONFIG_X86_NUMAQ
389static int mp_bus_id_to_node[MAX_MP_BUSSES] __initdata; 368static int mp_bus_id_to_node[MAX_MP_BUSSES];
390#endif 369#endif
391 370
392static int __init setup_pci_node_map_for_wpeg(int wpeg_num, int last_bus) 371static int setup_pci_node_map_for_wpeg(int wpeg_num, int last_bus)
393{ 372{
394 int twister = 0, node = 0; 373 int twister = 0, node = 0;
395 int i, bus, num_buses; 374 int i, bus, num_buses;
@@ -451,7 +430,7 @@ static int __init setup_pci_node_map_for_wpeg(int wpeg_num, int last_bus)
451 return bus; 430 return bus;
452} 431}
453 432
454static int __init build_detail_arrays(void) 433static int build_detail_arrays(void)
455{ 434{
456 unsigned long ptr; 435 unsigned long ptr;
457 int i, scal_detail_size, rio_detail_size; 436 int i, scal_detail_size, rio_detail_size;
@@ -485,7 +464,7 @@ static int __init build_detail_arrays(void)
485 return 1; 464 return 1;
486} 465}
487 466
488void __init setup_summit(void) 467void setup_summit(void)
489{ 468{
490 unsigned long ptr; 469 unsigned long ptr;
491 unsigned short offset; 470 unsigned short offset;
@@ -583,7 +562,6 @@ struct apic apic_summit = {
583 .send_IPI_all = summit_send_IPI_all, 562 .send_IPI_all = summit_send_IPI_all,
584 .send_IPI_self = default_send_IPI_self, 563 .send_IPI_self = default_send_IPI_self,
585 564
586 .wakeup_cpu = NULL,
587 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW, 565 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW,
588 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 566 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
589 567
diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index 354b9c45601d..8fb87b6dd633 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -224,7 +224,6 @@ struct apic apic_x2apic_cluster = {
224 .send_IPI_all = x2apic_send_IPI_all, 224 .send_IPI_all = x2apic_send_IPI_all,
225 .send_IPI_self = x2apic_send_IPI_self, 225 .send_IPI_self = x2apic_send_IPI_self,
226 226
227 .wakeup_cpu = NULL,
228 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW, 227 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW,
229 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 228 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
230 .wait_for_init_deassert = NULL, 229 .wait_for_init_deassert = NULL,
diff --git a/arch/x86/kernel/apic/x2apic_phys.c b/arch/x86/kernel/apic/x2apic_phys.c
index 5bcb174409bc..23625b9f98b2 100644
--- a/arch/x86/kernel/apic/x2apic_phys.c
+++ b/arch/x86/kernel/apic/x2apic_phys.c
@@ -213,7 +213,6 @@ struct apic apic_x2apic_phys = {
213 .send_IPI_all = x2apic_send_IPI_all, 213 .send_IPI_all = x2apic_send_IPI_all,
214 .send_IPI_self = x2apic_send_IPI_self, 214 .send_IPI_self = x2apic_send_IPI_self,
215 215
216 .wakeup_cpu = NULL,
217 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW, 216 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW,
218 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 217 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
219 .wait_for_init_deassert = NULL, 218 .wait_for_init_deassert = NULL,
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index 20b4ad07c3a1..1bd6da1f8fad 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -7,28 +7,28 @@
7 * 7 *
8 * Copyright (C) 2007-2008 Silicon Graphics, Inc. All rights reserved. 8 * Copyright (C) 2007-2008 Silicon Graphics, Inc. All rights reserved.
9 */ 9 */
10
11#include <linux/kernel.h>
12#include <linux/threads.h>
13#include <linux/cpu.h>
14#include <linux/cpumask.h> 10#include <linux/cpumask.h>
11#include <linux/hardirq.h>
12#include <linux/proc_fs.h>
13#include <linux/threads.h>
14#include <linux/kernel.h>
15#include <linux/module.h>
15#include <linux/string.h> 16#include <linux/string.h>
16#include <linux/ctype.h> 17#include <linux/ctype.h>
17#include <linux/init.h>
18#include <linux/sched.h> 18#include <linux/sched.h>
19#include <linux/module.h>
20#include <linux/hardirq.h>
21#include <linux/timer.h> 19#include <linux/timer.h>
22#include <linux/proc_fs.h> 20#include <linux/cpu.h>
23#include <asm/current.h> 21#include <linux/init.h>
24#include <asm/smp.h> 22
25#include <asm/apic.h>
26#include <asm/ipi.h>
27#include <asm/pgtable.h>
28#include <asm/uv/uv.h>
29#include <asm/uv/uv_mmrs.h> 23#include <asm/uv/uv_mmrs.h>
30#include <asm/uv/uv_hub.h> 24#include <asm/uv/uv_hub.h>
25#include <asm/current.h>
26#include <asm/pgtable.h>
31#include <asm/uv/bios.h> 27#include <asm/uv/bios.h>
28#include <asm/uv/uv.h>
29#include <asm/apic.h>
30#include <asm/ipi.h>
31#include <asm/smp.h>
32 32
33DEFINE_PER_CPU(int, x2apic_extra_bits); 33DEFINE_PER_CPU(int, x2apic_extra_bits);
34 34
@@ -91,24 +91,28 @@ static void uv_vector_allocation_domain(int cpu, struct cpumask *retmask)
91 cpumask_set_cpu(cpu, retmask); 91 cpumask_set_cpu(cpu, retmask);
92} 92}
93 93
94int uv_wakeup_secondary(int phys_apicid, unsigned int start_rip) 94static int uv_wakeup_secondary(int phys_apicid, unsigned long start_rip)
95{ 95{
96#ifdef CONFIG_SMP
96 unsigned long val; 97 unsigned long val;
97 int pnode; 98 int pnode;
98 99
99 pnode = uv_apicid_to_pnode(phys_apicid); 100 pnode = uv_apicid_to_pnode(phys_apicid);
100 val = (1UL << UVH_IPI_INT_SEND_SHFT) | 101 val = (1UL << UVH_IPI_INT_SEND_SHFT) |
101 (phys_apicid << UVH_IPI_INT_APIC_ID_SHFT) | 102 (phys_apicid << UVH_IPI_INT_APIC_ID_SHFT) |
102 (((long)start_rip << UVH_IPI_INT_VECTOR_SHFT) >> 12) | 103 ((start_rip << UVH_IPI_INT_VECTOR_SHFT) >> 12) |
103 APIC_DM_INIT; 104 APIC_DM_INIT;
104 uv_write_global_mmr64(pnode, UVH_IPI_INT, val); 105 uv_write_global_mmr64(pnode, UVH_IPI_INT, val);
105 mdelay(10); 106 mdelay(10);
106 107
107 val = (1UL << UVH_IPI_INT_SEND_SHFT) | 108 val = (1UL << UVH_IPI_INT_SEND_SHFT) |
108 (phys_apicid << UVH_IPI_INT_APIC_ID_SHFT) | 109 (phys_apicid << UVH_IPI_INT_APIC_ID_SHFT) |
109 (((long)start_rip << UVH_IPI_INT_VECTOR_SHFT) >> 12) | 110 ((start_rip << UVH_IPI_INT_VECTOR_SHFT) >> 12) |
110 APIC_DM_STARTUP; 111 APIC_DM_STARTUP;
111 uv_write_global_mmr64(pnode, UVH_IPI_INT, val); 112 uv_write_global_mmr64(pnode, UVH_IPI_INT, val);
113
114 atomic_set(&init_deasserted, 1);
115#endif
112 return 0; 116 return 0;
113} 117}
114 118
@@ -285,7 +289,7 @@ struct apic apic_x2apic_uv_x = {
285 .send_IPI_all = uv_send_IPI_all, 289 .send_IPI_all = uv_send_IPI_all,
286 .send_IPI_self = uv_send_IPI_self, 290 .send_IPI_self = uv_send_IPI_self,
287 291
288 .wakeup_cpu = NULL, 292 .wakeup_secondary_cpu = uv_wakeup_secondary,
289 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW, 293 .trampoline_phys_low = DEFAULT_TRAMPOLINE_PHYS_LOW,
290 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH, 294 .trampoline_phys_high = DEFAULT_TRAMPOLINE_PHYS_HIGH,
291 .wait_for_init_deassert = NULL, 295 .wait_for_init_deassert = NULL,
@@ -365,7 +369,7 @@ static __init void map_high(char *id, unsigned long base, int shift,
365 paddr = base << shift; 369 paddr = base << shift;
366 bytes = (1UL << shift) * (max_pnode + 1); 370 bytes = (1UL << shift) * (max_pnode + 1);
367 printk(KERN_INFO "UV: Map %s_HI 0x%lx - 0x%lx\n", id, paddr, 371 printk(KERN_INFO "UV: Map %s_HI 0x%lx - 0x%lx\n", id, paddr,
368 paddr + bytes); 372 paddr + bytes);
369 if (map_type == map_uc) 373 if (map_type == map_uc)
370 init_extra_mapping_uc(paddr, bytes); 374 init_extra_mapping_uc(paddr, bytes);
371 else 375 else
@@ -528,7 +532,7 @@ late_initcall(uv_init_heartbeat);
528 532
529/* 533/*
530 * Called on each cpu to initialize the per_cpu UV data area. 534 * Called on each cpu to initialize the per_cpu UV data area.
531 * ZZZ hotplug not supported yet 535 * FIXME: hotplug not supported yet
532 */ 536 */
533void __cpuinit uv_cpu_init(void) 537void __cpuinit uv_cpu_init(void)
534{ 538{
diff --git a/arch/x86/kernel/cpu/proc.c b/arch/x86/kernel/cpu/proc.c
index 01b1244ef1c0..d67e0e48bc2d 100644
--- a/arch/x86/kernel/cpu/proc.c
+++ b/arch/x86/kernel/cpu/proc.c
@@ -7,11 +7,10 @@
7/* 7/*
8 * Get CPU information for use by the procfs. 8 * Get CPU information for use by the procfs.
9 */ 9 */
10#ifdef CONFIG_X86_32
11static void show_cpuinfo_core(struct seq_file *m, struct cpuinfo_x86 *c, 10static void show_cpuinfo_core(struct seq_file *m, struct cpuinfo_x86 *c,
12 unsigned int cpu) 11 unsigned int cpu)
13{ 12{
14#ifdef CONFIG_X86_HT 13#ifdef CONFIG_SMP
15 if (c->x86_max_cores * smp_num_siblings > 1) { 14 if (c->x86_max_cores * smp_num_siblings > 1) {
16 seq_printf(m, "physical id\t: %d\n", c->phys_proc_id); 15 seq_printf(m, "physical id\t: %d\n", c->phys_proc_id);
17 seq_printf(m, "siblings\t: %d\n", 16 seq_printf(m, "siblings\t: %d\n",
@@ -24,6 +23,7 @@ static void show_cpuinfo_core(struct seq_file *m, struct cpuinfo_x86 *c,
24#endif 23#endif
25} 24}
26 25
26#ifdef CONFIG_X86_32
27static void show_cpuinfo_misc(struct seq_file *m, struct cpuinfo_x86 *c) 27static void show_cpuinfo_misc(struct seq_file *m, struct cpuinfo_x86 *c)
28{ 28{
29 /* 29 /*
@@ -50,22 +50,6 @@ static void show_cpuinfo_misc(struct seq_file *m, struct cpuinfo_x86 *c)
50 c->wp_works_ok ? "yes" : "no"); 50 c->wp_works_ok ? "yes" : "no");
51} 51}
52#else 52#else
53static void show_cpuinfo_core(struct seq_file *m, struct cpuinfo_x86 *c,
54 unsigned int cpu)
55{
56#ifdef CONFIG_SMP
57 if (c->x86_max_cores * smp_num_siblings > 1) {
58 seq_printf(m, "physical id\t: %d\n", c->phys_proc_id);
59 seq_printf(m, "siblings\t: %d\n",
60 cpus_weight(per_cpu(cpu_core_map, cpu)));
61 seq_printf(m, "core id\t\t: %d\n", c->cpu_core_id);
62 seq_printf(m, "cpu cores\t: %d\n", c->booted_cores);
63 seq_printf(m, "apicid\t\t: %d\n", c->apicid);
64 seq_printf(m, "initial apicid\t: %d\n", c->initial_apicid);
65 }
66#endif
67}
68
69static void show_cpuinfo_misc(struct seq_file *m, struct cpuinfo_x86 *c) 53static void show_cpuinfo_misc(struct seq_file *m, struct cpuinfo_x86 *c)
70{ 54{
71 seq_printf(m, 55 seq_printf(m,
diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c
index e85826829cf2..508bec1cee27 100644
--- a/arch/x86/kernel/e820.c
+++ b/arch/x86/kernel/e820.c
@@ -858,6 +858,9 @@ void __init reserve_early_overlap_ok(u64 start, u64 end, char *name)
858 */ 858 */
859void __init reserve_early(u64 start, u64 end, char *name) 859void __init reserve_early(u64 start, u64 end, char *name)
860{ 860{
861 if (start >= end)
862 return;
863
861 drop_overlaps_that_are_ok(start, end); 864 drop_overlaps_that_are_ok(start, end);
862 __reserve_early(start, end, name, 0); 865 __reserve_early(start, end, name, 0);
863} 866}
diff --git a/arch/x86/kernel/ioport.c b/arch/x86/kernel/ioport.c
index e41980a373ab..99c4d308f16b 100644
--- a/arch/x86/kernel/ioport.c
+++ b/arch/x86/kernel/ioport.c
@@ -85,19 +85,8 @@ asmlinkage long sys_ioperm(unsigned long from, unsigned long num, int turn_on)
85 85
86 t->io_bitmap_max = bytes; 86 t->io_bitmap_max = bytes;
87 87
88#ifdef CONFIG_X86_32
89 /*
90 * Sets the lazy trigger so that the next I/O operation will
91 * reload the correct bitmap.
92 * Reset the owner so that a process switch will not set
93 * tss->io_bitmap_base to IO_BITMAP_OFFSET.
94 */
95 tss->x86_tss.io_bitmap_base = INVALID_IO_BITMAP_OFFSET_LAZY;
96 tss->io_bitmap_owner = NULL;
97#else
98 /* Update the TSS: */ 88 /* Update the TSS: */
99 memcpy(tss->io_bitmap, t->io_bitmap_ptr, bytes_updated); 89 memcpy(tss->io_bitmap, t->io_bitmap_ptr, bytes_updated);
100#endif
101 90
102 put_cpu(); 91 put_cpu();
103 92
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index 87b69d4fac16..6afa5232dbb7 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -1,8 +1,8 @@
1#include <linux/errno.h> 1#include <linux/errno.h>
2#include <linux/kernel.h> 2#include <linux/kernel.h>
3#include <linux/mm.h> 3#include <linux/mm.h>
4#include <asm/idle.h>
5#include <linux/smp.h> 4#include <linux/smp.h>
5#include <linux/prctl.h>
6#include <linux/slab.h> 6#include <linux/slab.h>
7#include <linux/sched.h> 7#include <linux/sched.h>
8#include <linux/module.h> 8#include <linux/module.h>
@@ -11,6 +11,9 @@
11#include <linux/ftrace.h> 11#include <linux/ftrace.h>
12#include <asm/system.h> 12#include <asm/system.h>
13#include <asm/apic.h> 13#include <asm/apic.h>
14#include <asm/idle.h>
15#include <asm/uaccess.h>
16#include <asm/i387.h>
14 17
15unsigned long idle_halt; 18unsigned long idle_halt;
16EXPORT_SYMBOL(idle_halt); 19EXPORT_SYMBOL(idle_halt);
@@ -56,6 +59,192 @@ void arch_task_cache_init(void)
56} 59}
57 60
58/* 61/*
62 * Free current thread data structures etc..
63 */
64void exit_thread(void)
65{
66 struct task_struct *me = current;
67 struct thread_struct *t = &me->thread;
68
69 if (me->thread.io_bitmap_ptr) {
70 struct tss_struct *tss = &per_cpu(init_tss, get_cpu());
71
72 kfree(t->io_bitmap_ptr);
73 t->io_bitmap_ptr = NULL;
74 clear_thread_flag(TIF_IO_BITMAP);
75 /*
76 * Careful, clear this in the TSS too:
77 */
78 memset(tss->io_bitmap, 0xff, t->io_bitmap_max);
79 t->io_bitmap_max = 0;
80 put_cpu();
81 }
82
83 ds_exit_thread(current);
84}
85
86void flush_thread(void)
87{
88 struct task_struct *tsk = current;
89
90#ifdef CONFIG_X86_64
91 if (test_tsk_thread_flag(tsk, TIF_ABI_PENDING)) {
92 clear_tsk_thread_flag(tsk, TIF_ABI_PENDING);
93 if (test_tsk_thread_flag(tsk, TIF_IA32)) {
94 clear_tsk_thread_flag(tsk, TIF_IA32);
95 } else {
96 set_tsk_thread_flag(tsk, TIF_IA32);
97 current_thread_info()->status |= TS_COMPAT;
98 }
99 }
100#endif
101
102 clear_tsk_thread_flag(tsk, TIF_DEBUG);
103
104 tsk->thread.debugreg0 = 0;
105 tsk->thread.debugreg1 = 0;
106 tsk->thread.debugreg2 = 0;
107 tsk->thread.debugreg3 = 0;
108 tsk->thread.debugreg6 = 0;
109 tsk->thread.debugreg7 = 0;
110 memset(tsk->thread.tls_array, 0, sizeof(tsk->thread.tls_array));
111 /*
112 * Forget coprocessor state..
113 */
114 tsk->fpu_counter = 0;
115 clear_fpu(tsk);
116 clear_used_math();
117}
118
119static void hard_disable_TSC(void)
120{
121 write_cr4(read_cr4() | X86_CR4_TSD);
122}
123
124void disable_TSC(void)
125{
126 preempt_disable();
127 if (!test_and_set_thread_flag(TIF_NOTSC))
128 /*
129 * Must flip the CPU state synchronously with
130 * TIF_NOTSC in the current running context.
131 */
132 hard_disable_TSC();
133 preempt_enable();
134}
135
136static void hard_enable_TSC(void)
137{
138 write_cr4(read_cr4() & ~X86_CR4_TSD);
139}
140
141static void enable_TSC(void)
142{
143 preempt_disable();
144 if (test_and_clear_thread_flag(TIF_NOTSC))
145 /*
146 * Must flip the CPU state synchronously with
147 * TIF_NOTSC in the current running context.
148 */
149 hard_enable_TSC();
150 preempt_enable();
151}
152
153int get_tsc_mode(unsigned long adr)
154{
155 unsigned int val;
156
157 if (test_thread_flag(TIF_NOTSC))
158 val = PR_TSC_SIGSEGV;
159 else
160 val = PR_TSC_ENABLE;
161
162 return put_user(val, (unsigned int __user *)adr);
163}
164
165int set_tsc_mode(unsigned int val)
166{
167 if (val == PR_TSC_SIGSEGV)
168 disable_TSC();
169 else if (val == PR_TSC_ENABLE)
170 enable_TSC();
171 else
172 return -EINVAL;
173
174 return 0;
175}
176
177void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p,
178 struct tss_struct *tss)
179{
180 struct thread_struct *prev, *next;
181
182 prev = &prev_p->thread;
183 next = &next_p->thread;
184
185 if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) ||
186 test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR))
187 ds_switch_to(prev_p, next_p);
188 else if (next->debugctlmsr != prev->debugctlmsr)
189 update_debugctlmsr(next->debugctlmsr);
190
191 if (test_tsk_thread_flag(next_p, TIF_DEBUG)) {
192 set_debugreg(next->debugreg0, 0);
193 set_debugreg(next->debugreg1, 1);
194 set_debugreg(next->debugreg2, 2);
195 set_debugreg(next->debugreg3, 3);
196 /* no 4 and 5 */
197 set_debugreg(next->debugreg6, 6);
198 set_debugreg(next->debugreg7, 7);
199 }
200
201 if (test_tsk_thread_flag(prev_p, TIF_NOTSC) ^
202 test_tsk_thread_flag(next_p, TIF_NOTSC)) {
203 /* prev and next are different */
204 if (test_tsk_thread_flag(next_p, TIF_NOTSC))
205 hard_disable_TSC();
206 else
207 hard_enable_TSC();
208 }
209
210 if (test_tsk_thread_flag(next_p, TIF_IO_BITMAP)) {
211 /*
212 * Copy the relevant range of the IO bitmap.
213 * Normally this is 128 bytes or less:
214 */
215 memcpy(tss->io_bitmap, next->io_bitmap_ptr,
216 max(prev->io_bitmap_max, next->io_bitmap_max));
217 } else if (test_tsk_thread_flag(prev_p, TIF_IO_BITMAP)) {
218 /*
219 * Clear any possible leftover bits:
220 */
221 memset(tss->io_bitmap, 0xff, prev->io_bitmap_max);
222 }
223}
224
225int sys_fork(struct pt_regs *regs)
226{
227 return do_fork(SIGCHLD, regs->sp, regs, 0, NULL, NULL);
228}
229
230/*
231 * This is trivial, and on the face of it looks like it
232 * could equally well be done in user mode.
233 *
234 * Not so, for quite unobvious reasons - register pressure.
235 * In user mode vfork() cannot have a stack frame, and if
236 * done by calling the "clone()" system call directly, you
237 * do not have enough call-clobbered registers to hold all
238 * the information you need.
239 */
240int sys_vfork(struct pt_regs *regs)
241{
242 return do_fork(CLONE_VFORK | CLONE_VM | SIGCHLD, regs->sp, regs, 0,
243 NULL, NULL);
244}
245
246
247/*
59 * Idle related variables and functions 248 * Idle related variables and functions
60 */ 249 */
61unsigned long boot_option_idle_override = 0; 250unsigned long boot_option_idle_override = 0;
diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c
index 646da41a620a..14014d766cad 100644
--- a/arch/x86/kernel/process_32.c
+++ b/arch/x86/kernel/process_32.c
@@ -230,55 +230,6 @@ int kernel_thread(int (*fn)(void *), void *arg, unsigned long flags)
230} 230}
231EXPORT_SYMBOL(kernel_thread); 231EXPORT_SYMBOL(kernel_thread);
232 232
233/*
234 * Free current thread data structures etc..
235 */
236void exit_thread(void)
237{
238 /* The process may have allocated an io port bitmap... nuke it. */
239 if (unlikely(test_thread_flag(TIF_IO_BITMAP))) {
240 struct task_struct *tsk = current;
241 struct thread_struct *t = &tsk->thread;
242 int cpu = get_cpu();
243 struct tss_struct *tss = &per_cpu(init_tss, cpu);
244
245 kfree(t->io_bitmap_ptr);
246 t->io_bitmap_ptr = NULL;
247 clear_thread_flag(TIF_IO_BITMAP);
248 /*
249 * Careful, clear this in the TSS too:
250 */
251 memset(tss->io_bitmap, 0xff, tss->io_bitmap_max);
252 t->io_bitmap_max = 0;
253 tss->io_bitmap_owner = NULL;
254 tss->io_bitmap_max = 0;
255 tss->x86_tss.io_bitmap_base = INVALID_IO_BITMAP_OFFSET;
256 put_cpu();
257 }
258
259 ds_exit_thread(current);
260}
261
262void flush_thread(void)
263{
264 struct task_struct *tsk = current;
265
266 tsk->thread.debugreg0 = 0;
267 tsk->thread.debugreg1 = 0;
268 tsk->thread.debugreg2 = 0;
269 tsk->thread.debugreg3 = 0;
270 tsk->thread.debugreg6 = 0;
271 tsk->thread.debugreg7 = 0;
272 memset(tsk->thread.tls_array, 0, sizeof(tsk->thread.tls_array));
273 clear_tsk_thread_flag(tsk, TIF_DEBUG);
274 /*
275 * Forget coprocessor state..
276 */
277 tsk->fpu_counter = 0;
278 clear_fpu(tsk);
279 clear_used_math();
280}
281
282void release_thread(struct task_struct *dead_task) 233void release_thread(struct task_struct *dead_task)
283{ 234{
284 BUG_ON(dead_task->mm); 235 BUG_ON(dead_task->mm);
@@ -366,127 +317,6 @@ start_thread(struct pt_regs *regs, unsigned long new_ip, unsigned long new_sp)
366} 317}
367EXPORT_SYMBOL_GPL(start_thread); 318EXPORT_SYMBOL_GPL(start_thread);
368 319
369static void hard_disable_TSC(void)
370{
371 write_cr4(read_cr4() | X86_CR4_TSD);
372}
373
374void disable_TSC(void)
375{
376 preempt_disable();
377 if (!test_and_set_thread_flag(TIF_NOTSC))
378 /*
379 * Must flip the CPU state synchronously with
380 * TIF_NOTSC in the current running context.
381 */
382 hard_disable_TSC();
383 preempt_enable();
384}
385
386static void hard_enable_TSC(void)
387{
388 write_cr4(read_cr4() & ~X86_CR4_TSD);
389}
390
391static void enable_TSC(void)
392{
393 preempt_disable();
394 if (test_and_clear_thread_flag(TIF_NOTSC))
395 /*
396 * Must flip the CPU state synchronously with
397 * TIF_NOTSC in the current running context.
398 */
399 hard_enable_TSC();
400 preempt_enable();
401}
402
403int get_tsc_mode(unsigned long adr)
404{
405 unsigned int val;
406
407 if (test_thread_flag(TIF_NOTSC))
408 val = PR_TSC_SIGSEGV;
409 else
410 val = PR_TSC_ENABLE;
411
412 return put_user(val, (unsigned int __user *)adr);
413}
414
415int set_tsc_mode(unsigned int val)
416{
417 if (val == PR_TSC_SIGSEGV)
418 disable_TSC();
419 else if (val == PR_TSC_ENABLE)
420 enable_TSC();
421 else
422 return -EINVAL;
423
424 return 0;
425}
426
427static noinline void
428__switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p,
429 struct tss_struct *tss)
430{
431 struct thread_struct *prev, *next;
432
433 prev = &prev_p->thread;
434 next = &next_p->thread;
435
436 if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) ||
437 test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR))
438 ds_switch_to(prev_p, next_p);
439 else if (next->debugctlmsr != prev->debugctlmsr)
440 update_debugctlmsr(next->debugctlmsr);
441
442 if (test_tsk_thread_flag(next_p, TIF_DEBUG)) {
443 set_debugreg(next->debugreg0, 0);
444 set_debugreg(next->debugreg1, 1);
445 set_debugreg(next->debugreg2, 2);
446 set_debugreg(next->debugreg3, 3);
447 /* no 4 and 5 */
448 set_debugreg(next->debugreg6, 6);
449 set_debugreg(next->debugreg7, 7);
450 }
451
452 if (test_tsk_thread_flag(prev_p, TIF_NOTSC) ^
453 test_tsk_thread_flag(next_p, TIF_NOTSC)) {
454 /* prev and next are different */
455 if (test_tsk_thread_flag(next_p, TIF_NOTSC))
456 hard_disable_TSC();
457 else
458 hard_enable_TSC();
459 }
460
461 if (!test_tsk_thread_flag(next_p, TIF_IO_BITMAP)) {
462 /*
463 * Disable the bitmap via an invalid offset. We still cache
464 * the previous bitmap owner and the IO bitmap contents:
465 */
466 tss->x86_tss.io_bitmap_base = INVALID_IO_BITMAP_OFFSET;
467 return;
468 }
469
470 if (likely(next == tss->io_bitmap_owner)) {
471 /*
472 * Previous owner of the bitmap (hence the bitmap content)
473 * matches the next task, we dont have to do anything but
474 * to set a valid offset in the TSS:
475 */
476 tss->x86_tss.io_bitmap_base = IO_BITMAP_OFFSET;
477 return;
478 }
479 /*
480 * Lazy TSS's I/O bitmap copy. We set an invalid offset here
481 * and we let the task to get a GPF in case an I/O instruction
482 * is performed. The handler of the GPF will verify that the
483 * faulting task has a valid I/O bitmap and, it true, does the
484 * real copy and restart the instruction. This will save us
485 * redundant copies when the currently switched task does not
486 * perform any I/O during its timeslice.
487 */
488 tss->x86_tss.io_bitmap_base = INVALID_IO_BITMAP_OFFSET_LAZY;
489}
490 320
491/* 321/*
492 * switch_to(x,yn) should switch tasks from x to y. 322 * switch_to(x,yn) should switch tasks from x to y.
@@ -600,11 +430,6 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
600 return prev_p; 430 return prev_p;
601} 431}
602 432
603int sys_fork(struct pt_regs *regs)
604{
605 return do_fork(SIGCHLD, regs->sp, regs, 0, NULL, NULL);
606}
607
608int sys_clone(struct pt_regs *regs) 433int sys_clone(struct pt_regs *regs)
609{ 434{
610 unsigned long clone_flags; 435 unsigned long clone_flags;
@@ -621,21 +446,6 @@ int sys_clone(struct pt_regs *regs)
621} 446}
622 447
623/* 448/*
624 * This is trivial, and on the face of it looks like it
625 * could equally well be done in user mode.
626 *
627 * Not so, for quite unobvious reasons - register pressure.
628 * In user mode vfork() cannot have a stack frame, and if
629 * done by calling the "clone()" system call directly, you
630 * do not have enough call-clobbered registers to hold all
631 * the information you need.
632 */
633int sys_vfork(struct pt_regs *regs)
634{
635 return do_fork(CLONE_VFORK | CLONE_VM | SIGCHLD, regs->sp, regs, 0, NULL, NULL);
636}
637
638/*
639 * sys_execve() executes a new program. 449 * sys_execve() executes a new program.
640 */ 450 */
641int sys_execve(struct pt_regs *regs) 451int sys_execve(struct pt_regs *regs)
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index 836ef6575f01..abb7e6a7f0c6 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -237,61 +237,6 @@ void show_regs(struct pt_regs *regs)
237 show_trace(NULL, regs, (void *)(regs + 1), regs->bp); 237 show_trace(NULL, regs, (void *)(regs + 1), regs->bp);
238} 238}
239 239
240/*
241 * Free current thread data structures etc..
242 */
243void exit_thread(void)
244{
245 struct task_struct *me = current;
246 struct thread_struct *t = &me->thread;
247
248 if (me->thread.io_bitmap_ptr) {
249 struct tss_struct *tss = &per_cpu(init_tss, get_cpu());
250
251 kfree(t->io_bitmap_ptr);
252 t->io_bitmap_ptr = NULL;
253 clear_thread_flag(TIF_IO_BITMAP);
254 /*
255 * Careful, clear this in the TSS too:
256 */
257 memset(tss->io_bitmap, 0xff, t->io_bitmap_max);
258 t->io_bitmap_max = 0;
259 put_cpu();
260 }
261
262 ds_exit_thread(current);
263}
264
265void flush_thread(void)
266{
267 struct task_struct *tsk = current;
268
269 if (test_tsk_thread_flag(tsk, TIF_ABI_PENDING)) {
270 clear_tsk_thread_flag(tsk, TIF_ABI_PENDING);
271 if (test_tsk_thread_flag(tsk, TIF_IA32)) {
272 clear_tsk_thread_flag(tsk, TIF_IA32);
273 } else {
274 set_tsk_thread_flag(tsk, TIF_IA32);
275 current_thread_info()->status |= TS_COMPAT;
276 }
277 }
278 clear_tsk_thread_flag(tsk, TIF_DEBUG);
279
280 tsk->thread.debugreg0 = 0;
281 tsk->thread.debugreg1 = 0;
282 tsk->thread.debugreg2 = 0;
283 tsk->thread.debugreg3 = 0;
284 tsk->thread.debugreg6 = 0;
285 tsk->thread.debugreg7 = 0;
286 memset(tsk->thread.tls_array, 0, sizeof(tsk->thread.tls_array));
287 /*
288 * Forget coprocessor state..
289 */
290 tsk->fpu_counter = 0;
291 clear_fpu(tsk);
292 clear_used_math();
293}
294
295void release_thread(struct task_struct *dead_task) 240void release_thread(struct task_struct *dead_task)
296{ 241{
297 if (dead_task->mm) { 242 if (dead_task->mm) {
@@ -425,118 +370,6 @@ start_thread(struct pt_regs *regs, unsigned long new_ip, unsigned long new_sp)
425} 370}
426EXPORT_SYMBOL_GPL(start_thread); 371EXPORT_SYMBOL_GPL(start_thread);
427 372
428static void hard_disable_TSC(void)
429{
430 write_cr4(read_cr4() | X86_CR4_TSD);
431}
432
433void disable_TSC(void)
434{
435 preempt_disable();
436 if (!test_and_set_thread_flag(TIF_NOTSC))
437 /*
438 * Must flip the CPU state synchronously with
439 * TIF_NOTSC in the current running context.
440 */
441 hard_disable_TSC();
442 preempt_enable();
443}
444
445static void hard_enable_TSC(void)
446{
447 write_cr4(read_cr4() & ~X86_CR4_TSD);
448}
449
450static void enable_TSC(void)
451{
452 preempt_disable();
453 if (test_and_clear_thread_flag(TIF_NOTSC))
454 /*
455 * Must flip the CPU state synchronously with
456 * TIF_NOTSC in the current running context.
457 */
458 hard_enable_TSC();
459 preempt_enable();
460}
461
462int get_tsc_mode(unsigned long adr)
463{
464 unsigned int val;
465
466 if (test_thread_flag(TIF_NOTSC))
467 val = PR_TSC_SIGSEGV;
468 else
469 val = PR_TSC_ENABLE;
470
471 return put_user(val, (unsigned int __user *)adr);
472}
473
474int set_tsc_mode(unsigned int val)
475{
476 if (val == PR_TSC_SIGSEGV)
477 disable_TSC();
478 else if (val == PR_TSC_ENABLE)
479 enable_TSC();
480 else
481 return -EINVAL;
482
483 return 0;
484}
485
486/*
487 * This special macro can be used to load a debugging register
488 */
489#define loaddebug(thread, r) set_debugreg(thread->debugreg ## r, r)
490
491static inline void __switch_to_xtra(struct task_struct *prev_p,
492 struct task_struct *next_p,
493 struct tss_struct *tss)
494{
495 struct thread_struct *prev, *next;
496
497 prev = &prev_p->thread,
498 next = &next_p->thread;
499
500 if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) ||
501 test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR))
502 ds_switch_to(prev_p, next_p);
503 else if (next->debugctlmsr != prev->debugctlmsr)
504 update_debugctlmsr(next->debugctlmsr);
505
506 if (test_tsk_thread_flag(next_p, TIF_DEBUG)) {
507 loaddebug(next, 0);
508 loaddebug(next, 1);
509 loaddebug(next, 2);
510 loaddebug(next, 3);
511 /* no 4 and 5 */
512 loaddebug(next, 6);
513 loaddebug(next, 7);
514 }
515
516 if (test_tsk_thread_flag(prev_p, TIF_NOTSC) ^
517 test_tsk_thread_flag(next_p, TIF_NOTSC)) {
518 /* prev and next are different */
519 if (test_tsk_thread_flag(next_p, TIF_NOTSC))
520 hard_disable_TSC();
521 else
522 hard_enable_TSC();
523 }
524
525 if (test_tsk_thread_flag(next_p, TIF_IO_BITMAP)) {
526 /*
527 * Copy the relevant range of the IO bitmap.
528 * Normally this is 128 bytes or less:
529 */
530 memcpy(tss->io_bitmap, next->io_bitmap_ptr,
531 max(prev->io_bitmap_max, next->io_bitmap_max));
532 } else if (test_tsk_thread_flag(prev_p, TIF_IO_BITMAP)) {
533 /*
534 * Clear any possible leftover bits:
535 */
536 memset(tss->io_bitmap, 0xff, prev->io_bitmap_max);
537 }
538}
539
540/* 373/*
541 * switch_to(x,y) should switch tasks from x to y. 374 * switch_to(x,y) should switch tasks from x to y.
542 * 375 *
@@ -694,11 +527,6 @@ void set_personality_64bit(void)
694 current->personality &= ~READ_IMPLIES_EXEC; 527 current->personality &= ~READ_IMPLIES_EXEC;
695} 528}
696 529
697asmlinkage long sys_fork(struct pt_regs *regs)
698{
699 return do_fork(SIGCHLD, regs->sp, regs, 0, NULL, NULL);
700}
701
702asmlinkage long 530asmlinkage long
703sys_clone(unsigned long clone_flags, unsigned long newsp, 531sys_clone(unsigned long clone_flags, unsigned long newsp,
704 void __user *parent_tid, void __user *child_tid, struct pt_regs *regs) 532 void __user *parent_tid, void __user *child_tid, struct pt_regs *regs)
@@ -708,22 +536,6 @@ sys_clone(unsigned long clone_flags, unsigned long newsp,
708 return do_fork(clone_flags, newsp, regs, 0, parent_tid, child_tid); 536 return do_fork(clone_flags, newsp, regs, 0, parent_tid, child_tid);
709} 537}
710 538
711/*
712 * This is trivial, and on the face of it looks like it
713 * could equally well be done in user mode.
714 *
715 * Not so, for quite unobvious reasons - register pressure.
716 * In user mode vfork() cannot have a stack frame, and if
717 * done by calling the "clone()" system call directly, you
718 * do not have enough call-clobbered registers to hold all
719 * the information you need.
720 */
721asmlinkage long sys_vfork(struct pt_regs *regs)
722{
723 return do_fork(CLONE_VFORK | CLONE_VM | SIGCHLD, regs->sp, regs, 0,
724 NULL, NULL);
725}
726
727unsigned long get_wchan(struct task_struct *p) 539unsigned long get_wchan(struct task_struct *p)
728{ 540{
729 unsigned long stack; 541 unsigned long stack;
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index fb2159a5c817..3d9672e59c16 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -1383,7 +1383,7 @@ void send_sigtrap(struct task_struct *tsk, struct pt_regs *regs,
1383#ifdef CONFIG_X86_32 1383#ifdef CONFIG_X86_32
1384# define IS_IA32 1 1384# define IS_IA32 1
1385#elif defined CONFIG_IA32_EMULATION 1385#elif defined CONFIG_IA32_EMULATION
1386# define IS_IA32 test_thread_flag(TIF_IA32) 1386# define IS_IA32 is_compat_task()
1387#else 1387#else
1388# define IS_IA32 0 1388# define IS_IA32 0
1389#endif 1389#endif
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 5b85759e7972..4c54bc0d8ff3 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -600,19 +600,7 @@ static int __init setup_elfcorehdr(char *arg)
600early_param("elfcorehdr", setup_elfcorehdr); 600early_param("elfcorehdr", setup_elfcorehdr);
601#endif 601#endif
602 602
603static int __init default_update_apic(void) 603static struct x86_quirks default_x86_quirks __initdata;
604{
605#ifdef CONFIG_SMP
606 if (!apic->wakeup_cpu)
607 apic->wakeup_cpu = wakeup_secondary_cpu_via_init;
608#endif
609
610 return 0;
611}
612
613static struct x86_quirks default_x86_quirks __initdata = {
614 .update_apic = default_update_apic,
615};
616 604
617struct x86_quirks *x86_quirks __initdata = &default_x86_quirks; 605struct x86_quirks *x86_quirks __initdata = &default_x86_quirks;
618 606
@@ -875,9 +863,7 @@ void __init setup_arch(char **cmdline_p)
875 863
876 reserve_initrd(); 864 reserve_initrd();
877 865
878#ifdef CONFIG_X86_64
879 vsmp_init(); 866 vsmp_init();
880#endif
881 867
882 io_delay_init(); 868 io_delay_init();
883 869
diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c
index 7cdcd16885ed..d2cc6428c587 100644
--- a/arch/x86/kernel/signal.c
+++ b/arch/x86/kernel/signal.c
@@ -187,40 +187,35 @@ setup_sigcontext(struct sigcontext __user *sc, void __user *fpstate,
187/* 187/*
188 * Set up a signal frame. 188 * Set up a signal frame.
189 */ 189 */
190#ifdef CONFIG_X86_32
191static const struct {
192 u16 poplmovl;
193 u32 val;
194 u16 int80;
195} __attribute__((packed)) retcode = {
196 0xb858, /* popl %eax; movl $..., %eax */
197 __NR_sigreturn,
198 0x80cd, /* int $0x80 */
199};
200
201static const struct {
202 u8 movl;
203 u32 val;
204 u16 int80;
205 u8 pad;
206} __attribute__((packed)) rt_retcode = {
207 0xb8, /* movl $..., %eax */
208 __NR_rt_sigreturn,
209 0x80cd, /* int $0x80 */
210 0
211};
212 190
213/* 191/*
214 * Determine which stack to use.. 192 * Determine which stack to use..
215 */ 193 */
194static unsigned long align_sigframe(unsigned long sp)
195{
196#ifdef CONFIG_X86_32
197 /*
198 * Align the stack pointer according to the i386 ABI,
199 * i.e. so that on function entry ((sp + 4) & 15) == 0.
200 */
201 sp = ((sp + 4) & -16ul) - 4;
202#else /* !CONFIG_X86_32 */
203 sp = round_down(sp, 16) - 8;
204#endif
205 return sp;
206}
207
216static inline void __user * 208static inline void __user *
217get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size, 209get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size,
218 void **fpstate) 210 void __user **fpstate)
219{ 211{
220 unsigned long sp;
221
222 /* Default to using normal stack */ 212 /* Default to using normal stack */
223 sp = regs->sp; 213 unsigned long sp = regs->sp;
214
215#ifdef CONFIG_X86_64
216 /* redzone */
217 sp -= 128;
218#endif /* CONFIG_X86_64 */
224 219
225 /* 220 /*
226 * If we are on the alternate signal stack and would overflow it, don't. 221 * If we are on the alternate signal stack and would overflow it, don't.
@@ -234,30 +229,52 @@ get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size,
234 if (sas_ss_flags(sp) == 0) 229 if (sas_ss_flags(sp) == 0)
235 sp = current->sas_ss_sp + current->sas_ss_size; 230 sp = current->sas_ss_sp + current->sas_ss_size;
236 } else { 231 } else {
232#ifdef CONFIG_X86_32
237 /* This is the legacy signal stack switching. */ 233 /* This is the legacy signal stack switching. */
238 if ((regs->ss & 0xffff) != __USER_DS && 234 if ((regs->ss & 0xffff) != __USER_DS &&
239 !(ka->sa.sa_flags & SA_RESTORER) && 235 !(ka->sa.sa_flags & SA_RESTORER) &&
240 ka->sa.sa_restorer) 236 ka->sa.sa_restorer)
241 sp = (unsigned long) ka->sa.sa_restorer; 237 sp = (unsigned long) ka->sa.sa_restorer;
238#endif /* CONFIG_X86_32 */
242 } 239 }
243 240
244 if (used_math()) { 241 if (used_math()) {
245 sp = sp - sig_xstate_size; 242 sp -= sig_xstate_size;
246 *fpstate = (struct _fpstate *) sp; 243#ifdef CONFIG_X86_64
244 sp = round_down(sp, 64);
245#endif /* CONFIG_X86_64 */
246 *fpstate = (void __user *)sp;
247
247 if (save_i387_xstate(*fpstate) < 0) 248 if (save_i387_xstate(*fpstate) < 0)
248 return (void __user *)-1L; 249 return (void __user *)-1L;
249 } 250 }
250 251
251 sp -= frame_size; 252 return (void __user *)align_sigframe(sp - frame_size);
252 /*
253 * Align the stack pointer according to the i386 ABI,
254 * i.e. so that on function entry ((sp + 4) & 15) == 0.
255 */
256 sp = ((sp + 4) & -16ul) - 4;
257
258 return (void __user *) sp;
259} 253}
260 254
255#ifdef CONFIG_X86_32
256static const struct {
257 u16 poplmovl;
258 u32 val;
259 u16 int80;
260} __attribute__((packed)) retcode = {
261 0xb858, /* popl %eax; movl $..., %eax */
262 __NR_sigreturn,
263 0x80cd, /* int $0x80 */
264};
265
266static const struct {
267 u8 movl;
268 u32 val;
269 u16 int80;
270 u8 pad;
271} __attribute__((packed)) rt_retcode = {
272 0xb8, /* movl $..., %eax */
273 __NR_rt_sigreturn,
274 0x80cd, /* int $0x80 */
275 0
276};
277
261static int 278static int
262__setup_frame(int sig, struct k_sigaction *ka, sigset_t *set, 279__setup_frame(int sig, struct k_sigaction *ka, sigset_t *set,
263 struct pt_regs *regs) 280 struct pt_regs *regs)
@@ -388,24 +405,6 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
388 return 0; 405 return 0;
389} 406}
390#else /* !CONFIG_X86_32 */ 407#else /* !CONFIG_X86_32 */
391/*
392 * Determine which stack to use..
393 */
394static void __user *
395get_stack(struct k_sigaction *ka, unsigned long sp, unsigned long size)
396{
397 /* Default to using normal stack - redzone*/
398 sp -= 128;
399
400 /* This is the X/Open sanctioned signal stack switching. */
401 if (ka->sa.sa_flags & SA_ONSTACK) {
402 if (sas_ss_flags(sp) == 0)
403 sp = current->sas_ss_sp + current->sas_ss_size;
404 }
405
406 return (void __user *)round_down(sp - size, 64);
407}
408
409static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, 408static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
410 sigset_t *set, struct pt_regs *regs) 409 sigset_t *set, struct pt_regs *regs)
411{ 410{
@@ -414,15 +413,7 @@ static int __setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
414 int err = 0; 413 int err = 0;
415 struct task_struct *me = current; 414 struct task_struct *me = current;
416 415
417 if (used_math()) { 416 frame = get_sigframe(ka, regs, sizeof(struct rt_sigframe), &fp);
418 fp = get_stack(ka, regs->sp, sig_xstate_size);
419 frame = (void __user *)round_down(
420 (unsigned long)fp - sizeof(struct rt_sigframe), 16) - 8;
421
422 if (save_i387_xstate(fp) < 0)
423 return -EFAULT;
424 } else
425 frame = get_stack(ka, regs->sp, sizeof(struct rt_sigframe)) - 8;
426 417
427 if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame))) 418 if (!access_ok(VERIFY_WRITE, frame, sizeof(*frame)))
428 return -EFAULT; 419 return -EFAULT;
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 9ce666387f37..249334f5080a 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -112,7 +112,7 @@ EXPORT_PER_CPU_SYMBOL(cpu_core_map);
112DEFINE_PER_CPU_SHARED_ALIGNED(struct cpuinfo_x86, cpu_info); 112DEFINE_PER_CPU_SHARED_ALIGNED(struct cpuinfo_x86, cpu_info);
113EXPORT_PER_CPU_SYMBOL(cpu_info); 113EXPORT_PER_CPU_SYMBOL(cpu_info);
114 114
115static atomic_t init_deasserted; 115atomic_t init_deasserted;
116 116
117 117
118/* Set if we find a B stepping CPU */ 118/* Set if we find a B stepping CPU */
@@ -614,12 +614,6 @@ wakeup_secondary_cpu_via_init(int phys_apicid, unsigned long start_eip)
614 unsigned long send_status, accept_status = 0; 614 unsigned long send_status, accept_status = 0;
615 int maxlvt, num_starts, j; 615 int maxlvt, num_starts, j;
616 616
617 if (get_uv_system_type() == UV_NON_UNIQUE_APIC) {
618 send_status = uv_wakeup_secondary(phys_apicid, start_eip);
619 atomic_set(&init_deasserted, 1);
620 return send_status;
621 }
622
623 maxlvt = lapic_get_maxlvt(); 617 maxlvt = lapic_get_maxlvt();
624 618
625 /* 619 /*
@@ -748,7 +742,8 @@ static void __cpuinit do_fork_idle(struct work_struct *work)
748/* 742/*
749 * NOTE - on most systems this is a PHYSICAL apic ID, but on multiquad 743 * NOTE - on most systems this is a PHYSICAL apic ID, but on multiquad
750 * (ie clustered apic addressing mode), this is a LOGICAL apic ID. 744 * (ie clustered apic addressing mode), this is a LOGICAL apic ID.
751 * Returns zero if CPU booted OK, else error code from ->wakeup_cpu. 745 * Returns zero if CPU booted OK, else error code from
746 * ->wakeup_secondary_cpu.
752 */ 747 */
753static int __cpuinit do_boot_cpu(int apicid, int cpu) 748static int __cpuinit do_boot_cpu(int apicid, int cpu)
754{ 749{
@@ -835,9 +830,13 @@ do_rest:
835 } 830 }
836 831
837 /* 832 /*
838 * Starting actual IPI sequence... 833 * Kick the secondary CPU. Use the method in the APIC driver
834 * if it's defined - or use an INIT boot APIC message otherwise:
839 */ 835 */
840 boot_error = apic->wakeup_cpu(apicid, start_ip); 836 if (apic->wakeup_secondary_cpu)
837 boot_error = apic->wakeup_secondary_cpu(apicid, start_ip);
838 else
839 boot_error = wakeup_secondary_cpu_via_init(apicid, start_ip);
841 840
842 if (!boot_error) { 841 if (!boot_error) {
843 /* 842 /*
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index c05430ac1b44..a1d288327ff0 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -118,47 +118,6 @@ die_if_kernel(const char *str, struct pt_regs *regs, long err)
118 if (!user_mode_vm(regs)) 118 if (!user_mode_vm(regs))
119 die(str, regs, err); 119 die(str, regs, err);
120} 120}
121
122/*
123 * Perform the lazy TSS's I/O bitmap copy. If the TSS has an
124 * invalid offset set (the LAZY one) and the faulting thread has
125 * a valid I/O bitmap pointer, we copy the I/O bitmap in the TSS,
126 * we set the offset field correctly and return 1.
127 */
128static int lazy_iobitmap_copy(void)
129{
130 struct thread_struct *thread;
131 struct tss_struct *tss;
132 int cpu;
133
134 cpu = get_cpu();
135 tss = &per_cpu(init_tss, cpu);
136 thread = &current->thread;
137
138 if (tss->x86_tss.io_bitmap_base == INVALID_IO_BITMAP_OFFSET_LAZY &&
139 thread->io_bitmap_ptr) {
140 memcpy(tss->io_bitmap, thread->io_bitmap_ptr,
141 thread->io_bitmap_max);
142 /*
143 * If the previously set map was extending to higher ports
144 * than the current one, pad extra space with 0xff (no access).
145 */
146 if (thread->io_bitmap_max < tss->io_bitmap_max) {
147 memset((char *) tss->io_bitmap +
148 thread->io_bitmap_max, 0xff,
149 tss->io_bitmap_max - thread->io_bitmap_max);
150 }
151 tss->io_bitmap_max = thread->io_bitmap_max;
152 tss->x86_tss.io_bitmap_base = IO_BITMAP_OFFSET;
153 tss->io_bitmap_owner = thread;
154 put_cpu();
155
156 return 1;
157 }
158 put_cpu();
159
160 return 0;
161}
162#endif 121#endif
163 122
164static void __kprobes 123static void __kprobes
@@ -309,11 +268,6 @@ do_general_protection(struct pt_regs *regs, long error_code)
309 conditional_sti(regs); 268 conditional_sti(regs);
310 269
311#ifdef CONFIG_X86_32 270#ifdef CONFIG_X86_32
312 if (lazy_iobitmap_copy()) {
313 /* restart the faulting instruction */
314 return;
315 }
316
317 if (regs->flags & X86_VM_MASK) 271 if (regs->flags & X86_VM_MASK)
318 goto gp_in_vm86; 272 goto gp_in_vm86;
319#endif 273#endif
diff --git a/arch/x86/kernel/vsmp_64.c b/arch/x86/kernel/vsmp_64.c
index c609205df594..74de562812cc 100644
--- a/arch/x86/kernel/vsmp_64.c
+++ b/arch/x86/kernel/vsmp_64.c
@@ -22,7 +22,7 @@
22#include <asm/paravirt.h> 22#include <asm/paravirt.h>
23#include <asm/setup.h> 23#include <asm/setup.h>
24 24
25#if defined CONFIG_PCI && defined CONFIG_PARAVIRT 25#ifdef CONFIG_PARAVIRT
26/* 26/*
27 * Interrupt control on vSMPowered systems: 27 * Interrupt control on vSMPowered systems:
28 * ~AC is a shadow of IF. If IF is 'on' AC should be 'off' 28 * ~AC is a shadow of IF. If IF is 'on' AC should be 'off'
@@ -114,7 +114,6 @@ static void __init set_vsmp_pv_ops(void)
114} 114}
115#endif 115#endif
116 116
117#ifdef CONFIG_PCI
118static int is_vsmp = -1; 117static int is_vsmp = -1;
119 118
120static void __init detect_vsmp_box(void) 119static void __init detect_vsmp_box(void)
@@ -139,15 +138,6 @@ int is_vsmp_box(void)
139 return 0; 138 return 0;
140 } 139 }
141} 140}
142#else
143static void __init detect_vsmp_box(void)
144{
145}
146int is_vsmp_box(void)
147{
148 return 0;
149}
150#endif
151 141
152void __init vsmp_init(void) 142void __init vsmp_init(void)
153{ 143{
diff --git a/arch/x86/mm/Makefile b/arch/x86/mm/Makefile
index 2b938a384910..08537747cb58 100644
--- a/arch/x86/mm/Makefile
+++ b/arch/x86/mm/Makefile
@@ -1,4 +1,4 @@
1obj-y := init_$(BITS).o fault.o ioremap.o extable.o pageattr.o mmap.o \ 1obj-y := init.o init_$(BITS).o fault.o ioremap.o extable.o pageattr.o mmap.o \
2 pat.o pgtable.o gup.o 2 pat.o pgtable.o gup.o
3 3
4obj-$(CONFIG_SMP) += tlb.o 4obj-$(CONFIG_SMP) += tlb.o
diff --git a/arch/x86/mm/highmem_32.c b/arch/x86/mm/highmem_32.c
index bcc079c282dd..00f127c80b0e 100644
--- a/arch/x86/mm/highmem_32.c
+++ b/arch/x86/mm/highmem_32.c
@@ -1,5 +1,6 @@
1#include <linux/highmem.h> 1#include <linux/highmem.h>
2#include <linux/module.h> 2#include <linux/module.h>
3#include <linux/swap.h> /* for totalram_pages */
3 4
4void *kmap(struct page *page) 5void *kmap(struct page *page)
5{ 6{
@@ -156,3 +157,36 @@ EXPORT_SYMBOL(kmap);
156EXPORT_SYMBOL(kunmap); 157EXPORT_SYMBOL(kunmap);
157EXPORT_SYMBOL(kmap_atomic); 158EXPORT_SYMBOL(kmap_atomic);
158EXPORT_SYMBOL(kunmap_atomic); 159EXPORT_SYMBOL(kunmap_atomic);
160
161#ifdef CONFIG_NUMA
162void __init set_highmem_pages_init(void)
163{
164 struct zone *zone;
165 int nid;
166
167 for_each_zone(zone) {
168 unsigned long zone_start_pfn, zone_end_pfn;
169
170 if (!is_highmem(zone))
171 continue;
172
173 zone_start_pfn = zone->zone_start_pfn;
174 zone_end_pfn = zone_start_pfn + zone->spanned_pages;
175
176 nid = zone_to_nid(zone);
177 printk(KERN_INFO "Initializing %s for node %d (%08lx:%08lx)\n",
178 zone->name, nid, zone_start_pfn, zone_end_pfn);
179
180 add_highpages_with_active_regions(nid, zone_start_pfn,
181 zone_end_pfn);
182 }
183 totalram_pages += totalhigh_pages;
184}
185#else
186void __init set_highmem_pages_init(void)
187{
188 add_highpages_with_active_regions(0, highstart_pfn, highend_pfn);
189
190 totalram_pages += totalhigh_pages;
191}
192#endif /* CONFIG_NUMA */
diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c
new file mode 100644
index 000000000000..ce6a722587d8
--- /dev/null
+++ b/arch/x86/mm/init.c
@@ -0,0 +1,49 @@
1#include <linux/swap.h>
2#include <asm/cacheflush.h>
3#include <asm/page.h>
4#include <asm/sections.h>
5#include <asm/system.h>
6
7void free_init_pages(char *what, unsigned long begin, unsigned long end)
8{
9 unsigned long addr = begin;
10
11 if (addr >= end)
12 return;
13
14 /*
15 * If debugging page accesses then do not free this memory but
16 * mark them not present - any buggy init-section access will
17 * create a kernel page fault:
18 */
19#ifdef CONFIG_DEBUG_PAGEALLOC
20 printk(KERN_INFO "debug: unmapping init memory %08lx..%08lx\n",
21 begin, PAGE_ALIGN(end));
22 set_memory_np(begin, (end - begin) >> PAGE_SHIFT);
23#else
24 /*
25 * We just marked the kernel text read only above, now that
26 * we are going to free part of that, we need to make that
27 * writeable first.
28 */
29 set_memory_rw(begin, (end - begin) >> PAGE_SHIFT);
30
31 printk(KERN_INFO "Freeing %s: %luk freed\n", what, (end - begin) >> 10);
32
33 for (; addr < end; addr += PAGE_SIZE) {
34 ClearPageReserved(virt_to_page(addr));
35 init_page_count(virt_to_page(addr));
36 memset((void *)(addr & ~(PAGE_SIZE-1)),
37 POISON_FREE_INITMEM, PAGE_SIZE);
38 free_page(addr);
39 totalram_pages++;
40 }
41#endif
42}
43
44void free_initmem(void)
45{
46 free_init_pages("unused kernel memory",
47 (unsigned long)(&__init_begin),
48 (unsigned long)(&__init_end));
49}
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index ef0bb941cdf5..47df0e1bbeb9 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -50,8 +50,6 @@
50#include <asm/setup.h> 50#include <asm/setup.h>
51#include <asm/cacheflush.h> 51#include <asm/cacheflush.h>
52 52
53unsigned int __VMALLOC_RESERVE = 128 << 20;
54
55unsigned long max_low_pfn_mapped; 53unsigned long max_low_pfn_mapped;
56unsigned long max_pfn_mapped; 54unsigned long max_pfn_mapped;
57 55
@@ -486,22 +484,10 @@ void __init add_highpages_with_active_regions(int nid, unsigned long start_pfn,
486 work_with_active_regions(nid, add_highpages_work_fn, &data); 484 work_with_active_regions(nid, add_highpages_work_fn, &data);
487} 485}
488 486
489#ifndef CONFIG_NUMA
490static void __init set_highmem_pages_init(void)
491{
492 add_highpages_with_active_regions(0, highstart_pfn, highend_pfn);
493
494 totalram_pages += totalhigh_pages;
495}
496#endif /* !CONFIG_NUMA */
497
498#else 487#else
499static inline void permanent_kmaps_init(pgd_t *pgd_base) 488static inline void permanent_kmaps_init(pgd_t *pgd_base)
500{ 489{
501} 490}
502static inline void set_highmem_pages_init(void)
503{
504}
505#endif /* CONFIG_HIGHMEM */ 491#endif /* CONFIG_HIGHMEM */
506 492
507void __init native_pagetable_setup_start(pgd_t *base) 493void __init native_pagetable_setup_start(pgd_t *base)
@@ -864,10 +850,10 @@ static void __init find_early_table_space(unsigned long end, int use_pse)
864 unsigned long puds, pmds, ptes, tables, start; 850 unsigned long puds, pmds, ptes, tables, start;
865 851
866 puds = (end + PUD_SIZE - 1) >> PUD_SHIFT; 852 puds = (end + PUD_SIZE - 1) >> PUD_SHIFT;
867 tables = PAGE_ALIGN(puds * sizeof(pud_t)); 853 tables = roundup(puds * sizeof(pud_t), PAGE_SIZE);
868 854
869 pmds = (end + PMD_SIZE - 1) >> PMD_SHIFT; 855 pmds = (end + PMD_SIZE - 1) >> PMD_SHIFT;
870 tables += PAGE_ALIGN(pmds * sizeof(pmd_t)); 856 tables += roundup(pmds * sizeof(pmd_t), PAGE_SIZE);
871 857
872 if (use_pse) { 858 if (use_pse) {
873 unsigned long extra; 859 unsigned long extra;
@@ -878,10 +864,10 @@ static void __init find_early_table_space(unsigned long end, int use_pse)
878 } else 864 } else
879 ptes = (end + PAGE_SIZE - 1) >> PAGE_SHIFT; 865 ptes = (end + PAGE_SIZE - 1) >> PAGE_SHIFT;
880 866
881 tables += PAGE_ALIGN(ptes * sizeof(pte_t)); 867 tables += roundup(ptes * sizeof(pte_t), PAGE_SIZE);
882 868
883 /* for fixmap */ 869 /* for fixmap */
884 tables += PAGE_ALIGN(__end_of_fixed_addresses * sizeof(pte_t)); 870 tables += roundup(__end_of_fixed_addresses * sizeof(pte_t), PAGE_SIZE);
885 871
886 /* 872 /*
887 * RED-PEN putting page tables only on node 0 could 873 * RED-PEN putting page tables only on node 0 could
@@ -1231,45 +1217,6 @@ void mark_rodata_ro(void)
1231} 1217}
1232#endif 1218#endif
1233 1219
1234void free_init_pages(char *what, unsigned long begin, unsigned long end)
1235{
1236#ifdef CONFIG_DEBUG_PAGEALLOC
1237 /*
1238 * If debugging page accesses then do not free this memory but
1239 * mark them not present - any buggy init-section access will
1240 * create a kernel page fault:
1241 */
1242 printk(KERN_INFO "debug: unmapping init memory %08lx..%08lx\n",
1243 begin, PAGE_ALIGN(end));
1244 set_memory_np(begin, (end - begin) >> PAGE_SHIFT);
1245#else
1246 unsigned long addr;
1247
1248 /*
1249 * We just marked the kernel text read only above, now that
1250 * we are going to free part of that, we need to make that
1251 * writeable first.
1252 */
1253 set_memory_rw(begin, (end - begin) >> PAGE_SHIFT);
1254
1255 for (addr = begin; addr < end; addr += PAGE_SIZE) {
1256 ClearPageReserved(virt_to_page(addr));
1257 init_page_count(virt_to_page(addr));
1258 memset((void *)addr, POISON_FREE_INITMEM, PAGE_SIZE);
1259 free_page(addr);
1260 totalram_pages++;
1261 }
1262 printk(KERN_INFO "Freeing %s: %luk freed\n", what, (end - begin) >> 10);
1263#endif
1264}
1265
1266void free_initmem(void)
1267{
1268 free_init_pages("unused kernel memory",
1269 (unsigned long)(&__init_begin),
1270 (unsigned long)(&__init_end));
1271}
1272
1273#ifdef CONFIG_BLK_DEV_INITRD 1220#ifdef CONFIG_BLK_DEV_INITRD
1274void free_initrd_mem(unsigned long start, unsigned long end) 1221void free_initrd_mem(unsigned long start, unsigned long end)
1275{ 1222{
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 7d4e76da3368..11981fc8570a 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -748,6 +748,8 @@ unsigned long __init_refok init_memory_mapping(unsigned long start,
748 pos = start_pfn << PAGE_SHIFT; 748 pos = start_pfn << PAGE_SHIFT;
749 end_pfn = ((pos + (PMD_SIZE - 1)) >> PMD_SHIFT) 749 end_pfn = ((pos + (PMD_SIZE - 1)) >> PMD_SHIFT)
750 << (PMD_SHIFT - PAGE_SHIFT); 750 << (PMD_SHIFT - PAGE_SHIFT);
751 if (end_pfn > (end >> PAGE_SHIFT))
752 end_pfn = end >> PAGE_SHIFT;
751 if (start_pfn < end_pfn) { 753 if (start_pfn < end_pfn) {
752 nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0); 754 nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0);
753 pos = end_pfn << PAGE_SHIFT; 755 pos = end_pfn << PAGE_SHIFT;
@@ -979,43 +981,6 @@ void __init mem_init(void)
979 initsize >> 10); 981 initsize >> 10);
980} 982}
981 983
982void free_init_pages(char *what, unsigned long begin, unsigned long end)
983{
984 unsigned long addr = begin;
985
986 if (addr >= end)
987 return;
988
989 /*
990 * If debugging page accesses then do not free this memory but
991 * mark them not present - any buggy init-section access will
992 * create a kernel page fault:
993 */
994#ifdef CONFIG_DEBUG_PAGEALLOC
995 printk(KERN_INFO "debug: unmapping init memory %08lx..%08lx\n",
996 begin, PAGE_ALIGN(end));
997 set_memory_np(begin, (end - begin) >> PAGE_SHIFT);
998#else
999 printk(KERN_INFO "Freeing %s: %luk freed\n", what, (end - begin) >> 10);
1000
1001 for (; addr < end; addr += PAGE_SIZE) {
1002 ClearPageReserved(virt_to_page(addr));
1003 init_page_count(virt_to_page(addr));
1004 memset((void *)(addr & ~(PAGE_SIZE-1)),
1005 POISON_FREE_INITMEM, PAGE_SIZE);
1006 free_page(addr);
1007 totalram_pages++;
1008 }
1009#endif
1010}
1011
1012void free_initmem(void)
1013{
1014 free_init_pages("unused kernel memory",
1015 (unsigned long)(&__init_begin),
1016 (unsigned long)(&__init_end));
1017}
1018
1019#ifdef CONFIG_DEBUG_RODATA 984#ifdef CONFIG_DEBUG_RODATA
1020const int rodata_test_data = 0xC3; 985const int rodata_test_data = 0xC3;
1021EXPORT_SYMBOL_GPL(rodata_test_data); 986EXPORT_SYMBOL_GPL(rodata_test_data);
diff --git a/arch/x86/mm/iomap_32.c b/arch/x86/mm/iomap_32.c
index ca53224fc56c..04102d42ff42 100644
--- a/arch/x86/mm/iomap_32.c
+++ b/arch/x86/mm/iomap_32.c
@@ -20,6 +20,17 @@
20#include <asm/pat.h> 20#include <asm/pat.h>
21#include <linux/module.h> 21#include <linux/module.h>
22 22
23int is_io_mapping_possible(resource_size_t base, unsigned long size)
24{
25#ifndef CONFIG_X86_PAE
26 /* There is no way to map greater than 1 << 32 address without PAE */
27 if (base + size > 0x100000000ULL)
28 return 0;
29#endif
30 return 1;
31}
32EXPORT_SYMBOL_GPL(is_io_mapping_possible);
33
23/* Map 'pfn' using fixed map 'type' and protections 'prot' 34/* Map 'pfn' using fixed map 'type' and protections 'prot'
24 */ 35 */
25void * 36void *
diff --git a/arch/x86/mm/memtest.c b/arch/x86/mm/memtest.c
index 9cab18b0b857..0bcd7883d036 100644
--- a/arch/x86/mm/memtest.c
+++ b/arch/x86/mm/memtest.c
@@ -9,44 +9,44 @@
9 9
10#include <asm/e820.h> 10#include <asm/e820.h>
11 11
12static void __init memtest(unsigned long start_phys, unsigned long size, 12static u64 patterns[] __initdata = {
13 unsigned pattern) 13 0,
14 0xffffffffffffffffULL,
15 0x5555555555555555ULL,
16 0xaaaaaaaaaaaaaaaaULL,
17 0x1111111111111111ULL,
18 0x2222222222222222ULL,
19 0x4444444444444444ULL,
20 0x8888888888888888ULL,
21 0x3333333333333333ULL,
22 0x6666666666666666ULL,
23 0x9999999999999999ULL,
24 0xccccccccccccccccULL,
25 0x7777777777777777ULL,
26 0xbbbbbbbbbbbbbbbbULL,
27 0xddddddddddddddddULL,
28 0xeeeeeeeeeeeeeeeeULL,
29 0x7a6c7258554e494cULL, /* yeah ;-) */
30};
31
32static void __init reserve_bad_mem(u64 pattern, u64 start_bad, u64 end_bad)
14{ 33{
15 unsigned long i; 34 printk(KERN_INFO " %016llx bad mem addr %010llx - %010llx reserved\n",
16 unsigned long *start; 35 (unsigned long long) pattern,
17 unsigned long start_bad; 36 (unsigned long long) start_bad,
18 unsigned long last_bad; 37 (unsigned long long) end_bad);
19 unsigned long val; 38 reserve_early(start_bad, end_bad, "BAD RAM");
20 unsigned long start_phys_aligned; 39}
21 unsigned long count;
22 unsigned long incr;
23
24 switch (pattern) {
25 case 0:
26 val = 0UL;
27 break;
28 case 1:
29 val = -1UL;
30 break;
31 case 2:
32#ifdef CONFIG_X86_64
33 val = 0x5555555555555555UL;
34#else
35 val = 0x55555555UL;
36#endif
37 break;
38 case 3:
39#ifdef CONFIG_X86_64
40 val = 0xaaaaaaaaaaaaaaaaUL;
41#else
42 val = 0xaaaaaaaaUL;
43#endif
44 break;
45 default:
46 return;
47 }
48 40
49 incr = sizeof(unsigned long); 41static void __init memtest(u64 pattern, u64 start_phys, u64 size)
42{
43 u64 i, count;
44 u64 *start;
45 u64 start_bad, last_bad;
46 u64 start_phys_aligned;
47 size_t incr;
48
49 incr = sizeof(pattern);
50 start_phys_aligned = ALIGN(start_phys, incr); 50 start_phys_aligned = ALIGN(start_phys, incr);
51 count = (size - (start_phys_aligned - start_phys))/incr; 51 count = (size - (start_phys_aligned - start_phys))/incr;
52 start = __va(start_phys_aligned); 52 start = __va(start_phys_aligned);
@@ -54,25 +54,42 @@ static void __init memtest(unsigned long start_phys, unsigned long size,
54 last_bad = 0; 54 last_bad = 0;
55 55
56 for (i = 0; i < count; i++) 56 for (i = 0; i < count; i++)
57 start[i] = val; 57 start[i] = pattern;
58 for (i = 0; i < count; i++, start++, start_phys_aligned += incr) { 58 for (i = 0; i < count; i++, start++, start_phys_aligned += incr) {
59 if (*start != val) { 59 if (*start == pattern)
60 if (start_phys_aligned == last_bad + incr) { 60 continue;
61 last_bad += incr; 61 if (start_phys_aligned == last_bad + incr) {
62 } else { 62 last_bad += incr;
63 if (start_bad) { 63 continue;
64 printk(KERN_CONT "\n %016lx bad mem addr %010lx - %010lx reserved",
65 val, start_bad, last_bad + incr);
66 reserve_early(start_bad, last_bad + incr, "BAD RAM");
67 }
68 start_bad = last_bad = start_phys_aligned;
69 }
70 } 64 }
65 if (start_bad)
66 reserve_bad_mem(pattern, start_bad, last_bad + incr);
67 start_bad = last_bad = start_phys_aligned;
71 } 68 }
72 if (start_bad) { 69 if (start_bad)
73 printk(KERN_CONT "\n %016lx bad mem addr %010lx - %010lx reserved", 70 reserve_bad_mem(pattern, start_bad, last_bad + incr);
74 val, start_bad, last_bad + incr); 71}
75 reserve_early(start_bad, last_bad + incr, "BAD RAM"); 72
73static void __init do_one_pass(u64 pattern, u64 start, u64 end)
74{
75 u64 size = 0;
76
77 while (start < end) {
78 start = find_e820_area_size(start, &size, 1);
79
80 /* done ? */
81 if (start >= end)
82 break;
83 if (start + size > end)
84 size = end - start;
85
86 printk(KERN_INFO " %010llx - %010llx pattern %016llx\n",
87 (unsigned long long) start,
88 (unsigned long long) start + size,
89 (unsigned long long) cpu_to_be64(pattern));
90 memtest(pattern, start, size);
91
92 start += size;
76 } 93 }
77} 94}
78 95
@@ -90,33 +107,22 @@ early_param("memtest", parse_memtest);
90 107
91void __init early_memtest(unsigned long start, unsigned long end) 108void __init early_memtest(unsigned long start, unsigned long end)
92{ 109{
93 u64 t_start, t_size; 110 unsigned int i;
94 unsigned pattern; 111 unsigned int idx = 0;
95 112
96 if (!memtest_pattern) 113 if (!memtest_pattern)
97 return; 114 return;
98 115
99 printk(KERN_INFO "early_memtest: pattern num %d", memtest_pattern); 116 printk(KERN_INFO "early_memtest: # of tests: %d\n", memtest_pattern);
100 for (pattern = 0; pattern < memtest_pattern; pattern++) { 117 for (i = 0; i < memtest_pattern; i++) {
101 t_start = start; 118 idx = i % ARRAY_SIZE(patterns);
102 t_size = 0; 119 do_one_pass(patterns[idx], start, end);
103 while (t_start < end) { 120 }
104 t_start = find_e820_area_size(t_start, &t_size, 1);
105
106 /* done ? */
107 if (t_start >= end)
108 break;
109 if (t_start + t_size > end)
110 t_size = end - t_start;
111
112 printk(KERN_CONT "\n %010llx - %010llx pattern %d",
113 (unsigned long long)t_start,
114 (unsigned long long)t_start + t_size, pattern);
115
116 memtest(t_start, t_size, pattern);
117 121
118 t_start += t_size; 122 if (idx > 0) {
119 } 123 printk(KERN_INFO "early_memtest: wipe out "
124 "test pattern from memory\n");
125 /* additional test with pattern 0 will do this */
126 do_one_pass(0, start, end);
120 } 127 }
121 printk(KERN_CONT "\n");
122} 128}
diff --git a/arch/x86/mm/numa_32.c b/arch/x86/mm/numa_32.c
index 3957cd6d6454..451fe95a0352 100644
--- a/arch/x86/mm/numa_32.c
+++ b/arch/x86/mm/numa_32.c
@@ -423,32 +423,6 @@ void __init initmem_init(unsigned long start_pfn,
423 setup_bootmem_allocator(); 423 setup_bootmem_allocator();
424} 424}
425 425
426void __init set_highmem_pages_init(void)
427{
428#ifdef CONFIG_HIGHMEM
429 struct zone *zone;
430 int nid;
431
432 for_each_zone(zone) {
433 unsigned long zone_start_pfn, zone_end_pfn;
434
435 if (!is_highmem(zone))
436 continue;
437
438 zone_start_pfn = zone->zone_start_pfn;
439 zone_end_pfn = zone_start_pfn + zone->spanned_pages;
440
441 nid = zone_to_nid(zone);
442 printk(KERN_INFO "Initializing %s for node %d (%08lx:%08lx)\n",
443 zone->name, nid, zone_start_pfn, zone_end_pfn);
444
445 add_highpages_with_active_regions(nid, zone_start_pfn,
446 zone_end_pfn);
447 }
448 totalram_pages += totalhigh_pages;
449#endif
450}
451
452#ifdef CONFIG_MEMORY_HOTPLUG 426#ifdef CONFIG_MEMORY_HOTPLUG
453static int paddr_to_nid(u64 addr) 427static int paddr_to_nid(u64 addr)
454{ 428{
diff --git a/arch/x86/mm/pat.c b/arch/x86/mm/pat.c
index 05f9aef6818a..2ed37158012d 100644
--- a/arch/x86/mm/pat.c
+++ b/arch/x86/mm/pat.c
@@ -11,6 +11,7 @@
11#include <linux/bootmem.h> 11#include <linux/bootmem.h>
12#include <linux/debugfs.h> 12#include <linux/debugfs.h>
13#include <linux/kernel.h> 13#include <linux/kernel.h>
14#include <linux/module.h>
14#include <linux/gfp.h> 15#include <linux/gfp.h>
15#include <linux/mm.h> 16#include <linux/mm.h>
16#include <linux/fs.h> 17#include <linux/fs.h>
@@ -634,6 +635,33 @@ void unmap_devmem(unsigned long pfn, unsigned long size, pgprot_t vma_prot)
634} 635}
635 636
636/* 637/*
638 * Change the memory type for the physial address range in kernel identity
639 * mapping space if that range is a part of identity map.
640 */
641int kernel_map_sync_memtype(u64 base, unsigned long size, unsigned long flags)
642{
643 unsigned long id_sz;
644
645 if (!pat_enabled || base >= __pa(high_memory))
646 return 0;
647
648 id_sz = (__pa(high_memory) < base + size) ?
649 __pa(high_memory) - base :
650 size;
651
652 if (ioremap_change_attr((unsigned long)__va(base), id_sz, flags) < 0) {
653 printk(KERN_INFO
654 "%s:%d ioremap_change_attr failed %s "
655 "for %Lx-%Lx\n",
656 current->comm, current->pid,
657 cattr_name(flags),
658 base, (unsigned long long)(base + size));
659 return -EINVAL;
660 }
661 return 0;
662}
663
664/*
637 * Internal interface to reserve a range of physical memory with prot. 665 * Internal interface to reserve a range of physical memory with prot.
638 * Reserved non RAM regions only and after successful reserve_memtype, 666 * Reserved non RAM regions only and after successful reserve_memtype,
639 * this func also keeps identity mapping (if any) in sync with this new prot. 667 * this func also keeps identity mapping (if any) in sync with this new prot.
@@ -642,7 +670,7 @@ static int reserve_pfn_range(u64 paddr, unsigned long size, pgprot_t *vma_prot,
642 int strict_prot) 670 int strict_prot)
643{ 671{
644 int is_ram = 0; 672 int is_ram = 0;
645 int id_sz, ret; 673 int ret;
646 unsigned long flags; 674 unsigned long flags;
647 unsigned long want_flags = (pgprot_val(*vma_prot) & _PAGE_CACHE_MASK); 675 unsigned long want_flags = (pgprot_val(*vma_prot) & _PAGE_CACHE_MASK);
648 676
@@ -679,23 +707,8 @@ static int reserve_pfn_range(u64 paddr, unsigned long size, pgprot_t *vma_prot,
679 flags); 707 flags);
680 } 708 }
681 709
682 /* Need to keep identity mapping in sync */ 710 if (kernel_map_sync_memtype(paddr, size, flags) < 0) {
683 if (paddr >= __pa(high_memory))
684 return 0;
685
686 id_sz = (__pa(high_memory) < paddr + size) ?
687 __pa(high_memory) - paddr :
688 size;
689
690 if (ioremap_change_attr((unsigned long)__va(paddr), id_sz, flags) < 0) {
691 free_memtype(paddr, paddr + size); 711 free_memtype(paddr, paddr + size);
692 printk(KERN_ERR
693 "%s:%d reserve_pfn_range ioremap_change_attr failed %s "
694 "for %Lx-%Lx\n",
695 current->comm, current->pid,
696 cattr_name(flags),
697 (unsigned long long)paddr,
698 (unsigned long long)(paddr + size));
699 return -EINVAL; 712 return -EINVAL;
700 } 713 }
701 return 0; 714 return 0;
@@ -877,6 +890,7 @@ pgprot_t pgprot_writecombine(pgprot_t prot)
877 else 890 else
878 return pgprot_noncached(prot); 891 return pgprot_noncached(prot);
879} 892}
893EXPORT_SYMBOL_GPL(pgprot_writecombine);
880 894
881#if defined(CONFIG_DEBUG_FS) && defined(CONFIG_X86_PAT) 895#if defined(CONFIG_DEBUG_FS) && defined(CONFIG_X86_PAT)
882 896
diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
index 86f2ffc43c3d..5b7c7c8464fe 100644
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -313,6 +313,24 @@ int ptep_clear_flush_young(struct vm_area_struct *vma,
313 return young; 313 return young;
314} 314}
315 315
316/**
317 * reserve_top_address - reserves a hole in the top of kernel address space
318 * @reserve - size of hole to reserve
319 *
320 * Can be used to relocate the fixmap area and poke a hole in the top
321 * of kernel address space to make room for a hypervisor.
322 */
323void __init reserve_top_address(unsigned long reserve)
324{
325#ifdef CONFIG_X86_32
326 BUG_ON(fixmaps_set > 0);
327 printk(KERN_INFO "Reserving virtual address space above 0x%08x\n",
328 (int)-reserve);
329 __FIXADDR_TOP = -reserve - PAGE_SIZE;
330 __VMALLOC_RESERVE += reserve;
331#endif
332}
333
316int fixmaps_set; 334int fixmaps_set;
317 335
318void __native_set_fixmap(enum fixed_addresses idx, pte_t pte) 336void __native_set_fixmap(enum fixed_addresses idx, pte_t pte)
diff --git a/arch/x86/mm/pgtable_32.c b/arch/x86/mm/pgtable_32.c
index 0951db9ee519..f2e477c91c1b 100644
--- a/arch/x86/mm/pgtable_32.c
+++ b/arch/x86/mm/pgtable_32.c
@@ -20,6 +20,8 @@
20#include <asm/tlb.h> 20#include <asm/tlb.h>
21#include <asm/tlbflush.h> 21#include <asm/tlbflush.h>
22 22
23unsigned int __VMALLOC_RESERVE = 128 << 20;
24
23/* 25/*
24 * Associate a virtual page frame with a given physical page frame 26 * Associate a virtual page frame with a given physical page frame
25 * and protection flags for that frame. 27 * and protection flags for that frame.
@@ -97,22 +99,6 @@ void set_pmd_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags)
97unsigned long __FIXADDR_TOP = 0xfffff000; 99unsigned long __FIXADDR_TOP = 0xfffff000;
98EXPORT_SYMBOL(__FIXADDR_TOP); 100EXPORT_SYMBOL(__FIXADDR_TOP);
99 101
100/**
101 * reserve_top_address - reserves a hole in the top of kernel address space
102 * @reserve - size of hole to reserve
103 *
104 * Can be used to relocate the fixmap area and poke a hole in the top
105 * of kernel address space to make room for a hypervisor.
106 */
107void __init reserve_top_address(unsigned long reserve)
108{
109 BUG_ON(fixmaps_set > 0);
110 printk(KERN_INFO "Reserving virtual address space above 0x%08x\n",
111 (int)-reserve);
112 __FIXADDR_TOP = -reserve - PAGE_SIZE;
113 __VMALLOC_RESERVE += reserve;
114}
115
116/* 102/*
117 * vmalloc=size forces the vmalloc area to be exactly 'size' 103 * vmalloc=size forces the vmalloc area to be exactly 'size'
118 * bytes. This can be used to increase (or decrease) the 104 * bytes. This can be used to increase (or decrease) the
diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c
index e9f80c744cf3..10131fbdaada 100644
--- a/arch/x86/oprofile/op_model_ppro.c
+++ b/arch/x86/oprofile/op_model_ppro.c
@@ -78,8 +78,18 @@ static void ppro_setup_ctrs(struct op_msrs const * const msrs)
78 if (cpu_has_arch_perfmon) { 78 if (cpu_has_arch_perfmon) {
79 union cpuid10_eax eax; 79 union cpuid10_eax eax;
80 eax.full = cpuid_eax(0xa); 80 eax.full = cpuid_eax(0xa);
81 if (counter_width < eax.split.bit_width) 81
82 counter_width = eax.split.bit_width; 82 /*
83 * For Core2 (family 6, model 15), don't reset the
84 * counter width:
85 */
86 if (!(eax.split.version_id == 0 &&
87 current_cpu_data.x86 == 6 &&
88 current_cpu_data.x86_model == 15)) {
89
90 if (counter_width < eax.split.bit_width)
91 counter_width = eax.split.bit_width;
92 }
83 } 93 }
84 94
85 /* clear all counters */ 95 /* clear all counters */
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 352ea6830659..82cd39a6cbd3 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -942,6 +942,9 @@ asmlinkage void __init xen_start_kernel(void)
942 possible map and a non-dummy shared_info. */ 942 possible map and a non-dummy shared_info. */
943 per_cpu(xen_vcpu, 0) = &HYPERVISOR_shared_info->vcpu_info[0]; 943 per_cpu(xen_vcpu, 0) = &HYPERVISOR_shared_info->vcpu_info[0];
944 944
945 local_irq_disable();
946 early_boot_irqs_off();
947
945 xen_raw_console_write("mapping kernel into physical memory\n"); 948 xen_raw_console_write("mapping kernel into physical memory\n");
946 pgd = xen_setup_kernel_pagetable(pgd, xen_start_info->nr_pages); 949 pgd = xen_setup_kernel_pagetable(pgd, xen_start_info->nr_pages);
947 950