]> git.ipfire.org Git - people/teissler/ipfire-2.x.git/blob - src/patches/suse-2.6.27.25/patches.xen/xen3-patch-2.6.24
projects / people / teissler / ipfire-2.x.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Updated xen patches taken from suse.
[people/teissler/ipfire-2.x.git] / src / patches / suse-2.6.27.25 / patches.xen / xen3-patch-2.6.24
1 From: www.kernel.org
2 Subject: Update to 2.6.24
3 Patch-mainline: 2.6.24
4
5 Automatically created from "patches.kernel.org/patch-2.6.24" by xen-port-patches.py
6
7 Acked-by: jbeulich@novell.com
8
9 --- sle11-2009-05-14.orig/arch/x86/Kconfig 2009-02-05 10:22:38.000000000 +0100
10 +++ sle11-2009-05-14/arch/x86/Kconfig 2009-02-16 16:18:36.000000000 +0100
11 @@ -50,15 +50,16 @@ config GENERIC_CMOS_UPDATE
12
13 config CLOCKSOURCE_WATCHDOG
14 def_bool y
15 - depends on !X86_XEN
16 + depends on !XEN
17
18 config GENERIC_CLOCKEVENTS
19 def_bool y
20 - depends on !X86_XEN
21 + depends on !XEN
22
23 config GENERIC_CLOCKEVENTS_BROADCAST
24 def_bool y
25 - depends on X86_64 || (X86_32 && X86_LOCAL_APIC && !X86_XEN)
26 + depends on X86_64 || (X86_32 && X86_LOCAL_APIC)
27 + depends on !XEN
28
29 config LOCKDEP_SUPPORT
30 def_bool y
31 @@ -211,12 +212,12 @@ config X86_TRAMPOLINE
32
33 config X86_NO_TSS
34 bool
35 - depends on X86_XEN || X86_64_XEN
36 + depends on XEN
37 default y
38
39 config X86_NO_IDT
40 bool
41 - depends on X86_XEN || X86_64_XEN
42 + depends on XEN
43 default y
44
45 config KTIME_SCALAR
46 @@ -287,6 +288,7 @@ config X86_PC
47
48 config X86_XEN
49 bool "Xen-compatible"
50 + depends on X86_32
51 select XEN
52 select X86_PAE
53 select X86_UP_APIC if !SMP && XEN_PRIVILEGED_GUEST
54 @@ -365,6 +367,7 @@ endif
55
56 config X86_64_XEN
57 bool "Enable Xen compatible kernel"
58 + depends on X86_64
59 select XEN
60 select SWIOTLB
61 help
62 @@ -417,7 +420,7 @@ config SCHED_NO_NO_OMIT_FRAME_POINTER
63
64 menuconfig PARAVIRT_GUEST
65 bool "Paravirtualized guest support"
66 - depends on !X86_XEN && !X86_64_XEN
67 + depends on !XEN
68 help
69 Say Y here to get to see options related to running Linux under
70 various hypervisors. This option alone does not add any kernel code.
71 @@ -511,7 +514,7 @@ source "arch/x86/Kconfig.cpu"
72 config HPET_TIMER
73 def_bool X86_64
74 prompt "HPET Timer Support" if X86_32
75 - depends on !X86_XEN && !X86_64_XEN
76 + depends on !XEN
77 help
78 Use the IA-PC HPET (High Precision Event Timer) to manage
79 time in preference to the PIT and RTC, if a HPET is
80 @@ -831,7 +834,7 @@ config I8K
81 config X86_REBOOTFIXUPS
82 def_bool n
83 prompt "Enable X86 board specific fixups for reboot"
84 - depends on X86_32 && !X86_XEN
85 + depends on X86_32 && !XEN
86 ---help---
87 This enables chipset and/or board specific fixups to be done
88 in order to get reboot to work correctly. This is only needed on
89 @@ -1164,7 +1167,7 @@ config X86_RESERVE_LOW_64K
90 config MATH_EMULATION
91 bool
92 prompt "Math emulation" if X86_32
93 - depends on !X86_XEN
94 + depends on !XEN
95 ---help---
96 Linux can emulate a math coprocessor (used for floating point
97 operations) if you don't have one. 486DX and Pentium processors have
98 @@ -1272,7 +1275,7 @@ config X86_PAT
99 config EFI
100 def_bool n
101 prompt "EFI runtime service support"
102 - depends on ACPI && !X86_XEN && !X86_64_XEN
103 + depends on ACPI && !XEN
104 ---help---
105 This enables the kernel to use EFI runtime services that are
106 available (such as the EFI variable services).
107 @@ -1287,7 +1290,7 @@ config EFI
108 config IRQBALANCE
109 def_bool y
110 prompt "Enable kernel irq balancing"
111 - depends on X86_32 && SMP && X86_IO_APIC && !X86_XEN
112 + depends on X86_32 && SMP && X86_IO_APIC && !XEN
113 help
114 The default yes will allow the kernel to do irq load balancing.
115 Saying no will keep the kernel from doing irq load balancing.
116 @@ -1433,7 +1436,7 @@ config PHYSICAL_START
117
118 config RELOCATABLE
119 bool "Build a relocatable kernel (EXPERIMENTAL)"
120 - depends on EXPERIMENTAL && !X86_XEN && !X86_64_XEN
121 + depends on EXPERIMENTAL && !XEN
122 help
123 This builds a kernel image that retains relocation information
124 so it can be loaded someplace besides the default 1MB.
125 @@ -1503,6 +1506,7 @@ endmenu
126 config ARCH_ENABLE_MEMORY_HOTPLUG
127 def_bool y
128 depends on X86_64 || (X86_32 && HIGHMEM)
129 + depends on !XEN
130
131 config HAVE_ARCH_EARLY_PFN_TO_NID
132 def_bool X86_64
133 @@ -1693,7 +1697,7 @@ choice
134
135 config PCI_GOBIOS
136 bool "BIOS"
137 - depends on !X86_XEN
138 + depends on !XEN
139
140 config PCI_GOMMCONFIG
141 bool "MMConfig"
142 @@ -1744,7 +1748,7 @@ config PCI_MMCONFIG
143
144 config XEN_PCIDEV_FRONTEND
145 bool "Xen PCI Frontend" if X86_64
146 - depends on PCI && ((X86_XEN && (PCI_GOXEN_FE || PCI_GOANY)) || X86_64_XEN)
147 + depends on PCI && XEN && (PCI_GOXEN_FE || PCI_GOANY || X86_64)
148 select HOTPLUG
149 default y
150 help
151 @@ -1761,6 +1765,7 @@ config XEN_PCIDEV_FE_DEBUG
152 config DMAR
153 bool "Support for DMA Remapping Devices (EXPERIMENTAL)"
154 depends on X86_64 && PCI_MSI && ACPI && EXPERIMENTAL
155 + depends on !XEN
156 help
157 DMA remapping (DMAR) devices support enables independent address
158 translations for Direct Memory Access (DMA) from devices.
159 --- sle11-2009-05-14.orig/arch/x86/Makefile 2009-02-16 16:17:21.000000000 +0100
160 +++ sle11-2009-05-14/arch/x86/Makefile 2009-02-16 16:18:36.000000000 +0100
161 @@ -191,8 +191,8 @@ PHONY += zImage bzImage vmlinuz compress
162 zdisk bzdisk fdimage fdimage144 fdimage288 isoimage install
163
164 ifdef CONFIG_XEN
165 -CPPFLAGS := -D__XEN_INTERFACE_VERSION__=$(CONFIG_XEN_INTERFACE_VERSION) \
166 - -Iinclude$(if $(KBUILD_SRC),2)/asm/mach-xen $(CPPFLAGS)
167 +KBUILD_CPPFLAGS := -D__XEN_INTERFACE_VERSION__=$(CONFIG_XEN_INTERFACE_VERSION) \
168 + -Iinclude$(if $(KBUILD_SRC),2)/asm/mach-xen $(KBUILD_CPPFLAGS)
169
170 ifdef CONFIG_X86_64
171 LDFLAGS_vmlinux := -e startup_64
172 @@ -206,6 +206,8 @@ KBUILD_IMAGE := $(boot)/vmlinuz
173
174 vmlinuz: vmlinux
175 $(Q)$(MAKE) $(build)=$(boot) $(KBUILD_IMAGE)
176 + $(Q)mkdir -p $(objtree)/arch/$(UTS_MACHINE)/boot
177 + $(Q)ln -fsn ../../x86/boot/$@ $(objtree)/arch/$(UTS_MACHINE)/boot/$@
178 else
179 # Default kernel to build
180 all: bzImage
181 --- sle11-2009-05-14.orig/arch/x86/ia32/ia32entry-xen.S 2009-02-16 16:17:21.000000000 +0100
182 +++ sle11-2009-05-14/arch/x86/ia32/ia32entry-xen.S 2009-02-16 16:18:36.000000000 +0100
183 @@ -125,20 +125,16 @@ sysenter_do_call:
184 jmp int_ret_from_sys_call
185
186 sysenter_tracesys:
187 + xchgl %r9d,%ebp
188 SAVE_REST
189 CLEAR_RREGS
190 + movq %r9,R9(%rsp)
191 movq $-ENOSYS,RAX(%rsp) /* really needed? */
192 movq %rsp,%rdi /* &pt_regs -> arg1 */
193 call syscall_trace_enter
194 LOAD_ARGS32 ARGOFFSET /* reload args from stack in case ptrace changed it */
195 RESTORE_REST
196 - movl %ebp, %ebp
197 - /* no need to do an access_ok check here because rbp has been
198 - 32bit zero extended */
199 -1: movl (%rbp),%r9d
200 - .section __ex_table,"a"
201 - .quad 1b,ia32_badarg
202 - .previous
203 + xchgl %ebp,%r9d
204 jmp sysenter_do_call
205 CFI_ENDPROC
206 ENDPROC(ia32_sysenter_target)
207 @@ -200,20 +196,17 @@ cstar_do_call:
208 jmp int_ret_from_sys_call
209
210 cstar_tracesys:
211 + xchgl %r9d,%ebp
212 SAVE_REST
213 CLEAR_RREGS
214 + movq %r9,R9(%rsp)
215 movq $-ENOSYS,RAX(%rsp) /* really needed? */
216 movq %rsp,%rdi /* &pt_regs -> arg1 */
217 call syscall_trace_enter
218 LOAD_ARGS32 ARGOFFSET /* reload args from stack in case ptrace changed it */
219 RESTORE_REST
220 + xchgl %ebp,%r9d
221 movl RSP-ARGOFFSET(%rsp), %r8d
222 - /* no need to do an access_ok check here because r8 has been
223 - 32bit zero extended */
224 -1: movl (%r8),%r9d
225 - .section __ex_table,"a"
226 - .quad 1b,ia32_badarg
227 - .previous
228 jmp cstar_do_call
229 END(ia32_cstar_target)
230
231 --- sle11-2009-05-14.orig/arch/x86/kernel/Makefile 2008-12-15 11:27:22.000000000 +0100
232 +++ sle11-2009-05-14/arch/x86/kernel/Makefile 2009-02-16 16:18:36.000000000 +0100
233 @@ -127,4 +127,4 @@ endif
234 disabled-obj-$(CONFIG_XEN) := early-quirks.o hpet.o i8253.o i8259_$(BITS).o reboot.o \
235 smpboot_$(BITS).o tsc_$(BITS).o tsc_sync.o
236 disabled-obj-$(CONFIG_XEN_UNPRIVILEGED_GUEST) += mpparse_64.o
237 -%/head_64.o %/head_64.s: $(if $(CONFIG_XEN),EXTRA_AFLAGS,dummy) :=
238 +%/head_64.o %/head_64.s: asflags-$(CONFIG_XEN) :=
239 --- sle11-2009-05-14.orig/arch/x86/kernel/acpi/sleep_32-xen.c 2009-02-16 16:17:21.000000000 +0100
240 +++ sle11-2009-05-14/arch/x86/kernel/acpi/sleep_32-xen.c 2009-02-16 16:18:36.000000000 +0100
241 @@ -90,7 +90,7 @@ __setup("acpi_sleep=", acpi_sleep_setup)
242
243 /* Ouch, we want to delete this. We already have better version in userspace, in
244 s2ram from suspend.sf.net project */
245 -static __init int reset_videomode_after_s3(struct dmi_system_id *d)
246 +static __init int reset_videomode_after_s3(const struct dmi_system_id *d)
247 {
248 acpi_realmode_flags |= 2;
249 return 0;
250 --- sle11-2009-05-14.orig/arch/x86/kernel/acpi/sleep_64-xen.c 2009-02-16 16:17:21.000000000 +0100
251 +++ sle11-2009-05-14/arch/x86/kernel/acpi/sleep_64-xen.c 2009-02-16 16:18:36.000000000 +0100
252 @@ -123,6 +123,3 @@ static int __init acpi_sleep_setup(char
253 __setup("acpi_sleep=", acpi_sleep_setup);
254 #endif /* CONFIG_ACPI_PV_SLEEP */
255
256 -void acpi_pci_link_exit(void)
257 -{
258 -}
259 --- sle11-2009-05-14.orig/arch/x86/kernel/apic_64-xen.c 2009-02-16 16:17:21.000000000 +0100
260 +++ sle11-2009-05-14/arch/x86/kernel/apic_64-xen.c 2009-02-16 16:18:36.000000000 +0100
261 @@ -63,22 +63,38 @@ int setup_profiling_timer(unsigned int m
262
263 void smp_local_timer_interrupt(void)
264 {
265 - profile_tick(CPU_PROFILING);
266 #ifndef CONFIG_XEN
267 -#ifdef CONFIG_SMP
268 - update_process_times(user_mode(get_irq_regs()));
269 -#endif
270 -#endif
271 + int cpu = smp_processor_id();
272 + struct clock_event_device *evt = &per_cpu(lapic_events, cpu);
273 +
274 /*
275 - * We take the 'long' return path, and there every subsystem
276 - * grabs the appropriate locks (kernel lock/ irq lock).
277 + * Normally we should not be here till LAPIC has been initialized but
278 + * in some cases like kdump, its possible that there is a pending LAPIC
279 + * timer interrupt from previous kernel's context and is delivered in
280 + * new kernel the moment interrupts are enabled.
281 *
282 - * We might want to decouple profiling from the 'long path',
283 - * and do the profiling totally in assembly.
284 - *
285 - * Currently this isn't too much of an issue (performance wise),
286 - * we can take more than 100K local irqs per second on a 100 MHz P5.
287 + * Interrupts are enabled early and LAPIC is setup much later, hence
288 + * its possible that when we get here evt->event_handler is NULL.
289 + * Check for event_handler being NULL and discard the interrupt as
290 + * spurious.
291 + */
292 + if (!evt->event_handler) {
293 + printk(KERN_WARNING
294 + "Spurious LAPIC timer interrupt on cpu %d\n", cpu);
295 + /* Switch it off */
296 + lapic_timer_setup(CLOCK_EVT_MODE_SHUTDOWN, evt);
297 + return;
298 + }
299 +#endif
300 +
301 + /*
302 + * the NMI deadlock-detector uses this.
303 */
304 + add_pda(apic_timer_irqs, 1);
305 +
306 +#ifndef CONFIG_XEN
307 + evt->event_handler(evt);
308 +#endif
309 }
310
311 /*
312 @@ -94,11 +110,6 @@ void smp_apic_timer_interrupt(struct pt_
313 struct pt_regs *old_regs = set_irq_regs(regs);
314
315 /*
316 - * the NMI deadlock-detector uses this.
317 - */
318 - add_pda(apic_timer_irqs, 1);
319 -
320 - /*
321 * NOTE! We'd better ACK the irq immediately,
322 * because timer handling can be slow.
323 */
324 @@ -132,6 +143,7 @@ asmlinkage void smp_spurious_interrupt(v
325 if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f)))
326 ack_APIC_irq();
327
328 + add_pda(irq_spurious_count, 1);
329 irq_exit();
330 }
331
332 --- sle11-2009-05-14.orig/arch/x86/kernel/cpu/common-xen.c 2009-02-16 16:17:21.000000000 +0100
333 +++ sle11-2009-05-14/arch/x86/kernel/cpu/common-xen.c 2009-02-16 16:18:36.000000000 +0100
334 @@ -214,7 +214,7 @@ static void __cpuinit get_cpu_vendor(str
335
336 static int __init x86_fxsr_setup(char * s)
337 {
338 - /* Tell all the other CPU's to not use it... */
339 + /* Tell all the other CPUs to not use it... */
340 disable_x86_fxsr = 1;
341
342 /*
343 --- sle11-2009-05-14.orig/arch/x86/kernel/e820_32-xen.c 2009-02-16 16:17:21.000000000 +0100
344 +++ sle11-2009-05-14/arch/x86/kernel/e820_32-xen.c 2009-02-16 16:18:36.000000000 +0100
345 @@ -52,6 +52,13 @@ struct resource code_resource = {
346 .flags = IORESOURCE_BUSY | IORESOURCE_MEM
347 };
348
349 +struct resource bss_resource = {
350 + .name = "Kernel bss",
351 + .start = 0,
352 + .end = 0,
353 + .flags = IORESOURCE_BUSY | IORESOURCE_MEM
354 +};
355 +
356 static struct resource system_rom_resource = {
357 .name = "System ROM",
358 .start = 0xf0000,
359 @@ -266,7 +273,9 @@ static struct e820map machine_e820;
360 * and also for regions reported as reserved by the e820.
361 */
362 static void __init
363 -legacy_init_iomem_resources(struct resource *code_resource, struct resource *data_resource)
364 +legacy_init_iomem_resources(struct resource *code_resource,
365 + struct resource *data_resource,
366 + struct resource *bss_resource)
367 {
368 int i;
369
370 @@ -300,9 +309,11 @@ legacy_init_iomem_resources(struct resou
371 #ifndef CONFIG_XEN
372 request_resource(res, code_resource);
373 request_resource(res, data_resource);
374 + request_resource(res, bss_resource);
375 #endif
376 #ifdef CONFIG_KEXEC
377 - request_resource(res, &crashk_res);
378 + if (crashk_res.start != crashk_res.end)
379 + request_resource(res, &crashk_res);
380 #ifdef CONFIG_XEN
381 xen_machine_kexec_register_resources(res);
382 #endif
383 @@ -329,9 +340,11 @@ static int __init request_standard_resou
384
385 printk("Setting up standard PCI resources\n");
386 if (efi_enabled)
387 - efi_initialize_iomem_resources(&code_resource, &data_resource);
388 + efi_initialize_iomem_resources(&code_resource,
389 + &data_resource, &bss_resource);
390 else
391 - legacy_init_iomem_resources(&code_resource, &data_resource);
392 + legacy_init_iomem_resources(&code_resource,
393 + &data_resource, &bss_resource);
394
395 /* EFI systems may still have VGA */
396 request_resource(&iomem_resource, &video_ram_resource);
397 @@ -774,7 +787,7 @@ void __init e820_register_memory(void)
398 #endif
399
400 /*
401 - * Search for the bigest gap in the low 32 bits of the e820
402 + * Search for the biggest gap in the low 32 bits of the e820
403 * memory space.
404 */
405 last = 0x100000000ull;
406 --- sle11-2009-05-14.orig/arch/x86/kernel/e820_64-xen.c 2009-02-16 16:17:21.000000000 +0100
407 +++ sle11-2009-05-14/arch/x86/kernel/e820_64-xen.c 2009-02-16 16:18:36.000000000 +0100
408 @@ -24,7 +24,7 @@
409 #include <asm/page.h>
410 #include <asm/e820.h>
411 #include <asm/proto.h>
412 -#include <asm/bootsetup.h>
413 +#include <asm/setup.h>
414 #include <asm/sections.h>
415 #include <xen/interface/memory.h>
416
417 @@ -51,7 +51,7 @@ unsigned long end_pfn_map;
418 */
419 static unsigned long __initdata end_user_pfn = MAXMEM>>PAGE_SHIFT;
420
421 -extern struct resource code_resource, data_resource;
422 +extern struct resource code_resource, data_resource, bss_resource;
423
424 /* Check for some hardcoded bad areas that early boot is not allowed to touch */
425 static inline int bad_addr(unsigned long *addrp, unsigned long size)
426 @@ -73,10 +73,15 @@ static inline int bad_addr(unsigned long
427
428 /* initrd */
429 #ifdef CONFIG_BLK_DEV_INITRD
430 - if (LOADER_TYPE && INITRD_START && last >= INITRD_START &&
431 - addr < INITRD_START+INITRD_SIZE) {
432 - *addrp = PAGE_ALIGN(INITRD_START + INITRD_SIZE);
433 - return 1;
434 + if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) {
435 + unsigned long ramdisk_image = boot_params.hdr.ramdisk_image;
436 + unsigned long ramdisk_size = boot_params.hdr.ramdisk_size;
437 + unsigned long ramdisk_end = ramdisk_image+ramdisk_size;
438 +
439 + if (last >= ramdisk_image && addr < ramdisk_end) {
440 + *addrp = PAGE_ALIGN(ramdisk_end);
441 + return 1;
442 + }
443 }
444 #endif
445 /* kernel code */
446 @@ -249,6 +254,7 @@ void __init e820_reserve_resources(struc
447 #ifndef CONFIG_XEN
448 request_resource(res, &code_resource);
449 request_resource(res, &data_resource);
450 + request_resource(res, &bss_resource);
451 #endif
452 #ifdef CONFIG_KEXEC
453 if (crashk_res.start != crashk_res.end)
454 @@ -650,8 +656,8 @@ void __init setup_memory_region(void)
455 * Otherwise fake a memory map; one section from 0k->640k,
456 * the next section from 1mb->appropriate_mem_k
457 */
458 - sanitize_e820_map(E820_MAP, &E820_MAP_NR);
459 - if (copy_e820_map(E820_MAP, E820_MAP_NR) < 0)
460 + sanitize_e820_map(boot_params.e820_map, &boot_params.e820_entries);
461 + if (copy_e820_map(boot_params.e820_map, boot_params.e820_entries) < 0)
462 early_panic("Cannot find a valid memory map");
463 printk(KERN_INFO "BIOS-provided physical RAM map:\n");
464 e820_print_map("BIOS-e820");
465 @@ -836,3 +842,22 @@ __init void e820_setup_gap(struct e820en
466 printk(KERN_INFO "Allocating PCI resources starting at %lx (gap: %lx:%lx)\n",
467 pci_mem_start, gapstart, gapsize);
468 }
469 +
470 +int __init arch_get_ram_range(int slot, u64 *addr, u64 *size)
471 +{
472 + int i;
473 +
474 + if (slot < 0 || slot >= e820.nr_map)
475 + return -1;
476 + for (i = slot; i < e820.nr_map; i++) {
477 + if (e820.map[i].type != E820_RAM)
478 + continue;
479 + break;
480 + }
481 + if (i == e820.nr_map || e820.map[i].addr > (max_pfn << PAGE_SHIFT))
482 + return -1;
483 + *addr = e820.map[i].addr;
484 + *size = min_t(u64, e820.map[i].size + e820.map[i].addr,
485 + max_pfn << PAGE_SHIFT) - *addr;
486 + return i + 1;
487 +}
488 --- sle11-2009-05-14.orig/arch/x86/kernel/early_printk-xen.c 2009-02-16 16:17:21.000000000 +0100
489 +++ sle11-2009-05-14/arch/x86/kernel/early_printk-xen.c 2009-02-16 16:18:36.000000000 +0100
490 @@ -6,14 +6,9 @@
491 #include <asm/io.h>
492 #include <asm/processor.h>
493 #include <asm/fcntl.h>
494 +#include <asm/setup.h>
495
496 /* Simple VGA output */
497 -
498 -#ifdef __i386__
499 -#include <asm/setup.h>
500 -#else
501 -#include <asm/bootsetup.h>
502 -#endif
503 #define VGABASE (__ISA_IO_base + 0xb8000)
504
505 #ifndef CONFIG_XEN
506 @@ -264,10 +259,10 @@ static int __init setup_early_printk(cha
507 early_console = &early_serial_console;
508 } else if (!strncmp(buf, "vga", 3)) {
509 #ifndef CONFIG_XEN
510 - && SCREEN_INFO.orig_video_isVGA == 1) {
511 - max_xpos = SCREEN_INFO.orig_video_cols;
512 - max_ypos = SCREEN_INFO.orig_video_lines;
513 - current_ypos = SCREEN_INFO.orig_y;
514 + && boot_params.screen_info.orig_video_isVGA == 1) {
515 + max_xpos = boot_params.screen_info.orig_video_cols;
516 + max_ypos = boot_params.screen_info.orig_video_lines;
517 + current_ypos = boot_params.screen_info.orig_y;
518 #endif
519 early_console = &early_vga_console;
520 } else if (!strncmp(buf, "simnow", 6)) {
521 --- sle11-2009-05-14.orig/arch/x86/kernel/entry_32-xen.S 2009-02-16 16:17:21.000000000 +0100
522 +++ sle11-2009-05-14/arch/x86/kernel/entry_32-xen.S 2009-05-14 11:18:18.000000000 +0200
523 @@ -254,6 +254,7 @@ check_userspace:
524 jb resume_kernel # not returning to v8086 or userspace
525
526 ENTRY(resume_userspace)
527 + LOCKDEP_SYS_EXIT
528 DISABLE_INTERRUPTS(CLBR_ANY) # make sure we don't miss an interrupt
529 # setting need_resched or sigpending
530 # between sampling and the iret
531 @@ -341,6 +342,7 @@ sysenter_past_esp:
532 jae syscall_badsys
533 call *sys_call_table(,%eax,4)
534 movl %eax,PT_EAX(%esp)
535 + LOCKDEP_SYS_EXIT
536 DISABLE_INTERRUPTS(CLBR_ANY)
537 TRACE_IRQS_OFF
538 movl TI_flags(%ebp), %ecx
539 @@ -406,6 +408,7 @@ syscall_call:
540 call *sys_call_table(,%eax,4)
541 movl %eax,PT_EAX(%esp) # store the return value
542 syscall_exit:
543 + LOCKDEP_SYS_EXIT
544 DISABLE_INTERRUPTS(CLBR_ANY) # make sure we don't miss an interrupt
545 # setting need_resched or sigpending
546 # between sampling and the iret
547 @@ -478,7 +481,7 @@ ldt_ss:
548 * is still available to implement the setting of the high
549 * 16-bits in the INTERRUPT_RETURN paravirt-op.
550 */
551 - cmpl $0, paravirt_ops+PARAVIRT_enabled
552 + cmpl $0, pv_info+PARAVIRT_enabled
553 jne restore_nocheck
554 #endif
555
556 @@ -540,6 +543,7 @@ work_pending:
557 jz work_notifysig
558 work_resched:
559 call schedule
560 + LOCKDEP_SYS_EXIT
561 DISABLE_INTERRUPTS(CLBR_ANY) # make sure we don't miss an interrupt
562 # setting need_resched or sigpending
563 # between sampling and the iret
564 @@ -1268,6 +1272,6 @@ ENTRY(kernel_thread_helper)
565 ENDPROC(kernel_thread_helper)
566
567 .section .rodata,"a"
568 -#include "syscall_table.S"
569 +#include "syscall_table_32.S"
570
571 syscall_table_size=(.-sys_call_table)
572 --- sle11-2009-05-14.orig/arch/x86/kernel/entry_64-xen.S 2009-02-16 16:17:21.000000000 +0100
573 +++ sle11-2009-05-14/arch/x86/kernel/entry_64-xen.S 2009-02-16 16:18:36.000000000 +0100
574 @@ -57,7 +57,7 @@
575 #include <xen/interface/arch-x86_64.h>
576 #include <xen/interface/features.h>
577
578 -#include "xen_entry.S"
579 +#include "xen_entry_64.S"
580
581 .code64
582
583 @@ -275,6 +275,7 @@ ret_from_sys_call:
584 movl $_TIF_ALLWORK_MASK,%edi
585 /* edi: flagmask */
586 sysret_check:
587 + LOCKDEP_SYS_EXIT
588 GET_THREAD_INFO(%rcx)
589 XEN_BLOCK_EVENTS(%rsi)
590 TRACE_IRQS_OFF
591 @@ -365,6 +366,7 @@ int_ret_from_sys_call:
592 movl $_TIF_ALLWORK_MASK,%edi
593 /* edi: mask to check */
594 int_with_check:
595 + LOCKDEP_SYS_EXIT_IRQ
596 GET_THREAD_INFO(%rcx)
597 movl threadinfo_flags(%rcx),%edx
598 andl %edi,%edx
599 @@ -516,11 +518,12 @@ END(stub_rt_sigreturn)
600
601 retint_check:
602 CFI_DEFAULT_STACK adj=1
603 + LOCKDEP_SYS_EXIT_IRQ
604 movl threadinfo_flags(%rcx),%edx
605 andl %edi,%edx
606 CFI_REMEMBER_STATE
607 jnz retint_careful
608 -retint_restore_args:
609 +retint_restore_args: /* return to kernel space */
610 movl EFLAGS-REST_SKIP(%rsp), %eax
611 shr $9, %eax # EAX[0] == IRET_EFLAGS.IF
612 XEN_GET_VCPU_INFO(%rsi)
613 @@ -841,7 +844,7 @@ error_call_handler:
614 movq ORIG_RAX(%rsp),%rsi # get error code
615 movq $-1,ORIG_RAX(%rsp)
616 call *%rax
617 -error_exit:
618 +error_exit:
619 RESTORE_REST
620 /* cli */
621 XEN_BLOCK_EVENTS(%rsi)
622 @@ -849,14 +852,11 @@ error_exit:
623 GET_THREAD_INFO(%rcx)
624 testb $3,CS-ARGOFFSET(%rsp)
625 jz retint_kernel
626 + LOCKDEP_SYS_EXIT_IRQ
627 movl threadinfo_flags(%rcx),%edx
628 movl $_TIF_WORK_MASK,%edi
629 andl %edi,%edx
630 jnz retint_careful
631 - /*
632 - * The iret might restore flags:
633 - */
634 - TRACE_IRQS_IRETQ
635 jmp retint_restore_args
636
637 #if 0
638 @@ -1071,7 +1071,7 @@ child_rip:
639 movq %rsi, %rdi
640 call *%rax
641 # exit
642 - xorl %edi, %edi
643 + mov %eax, %edi
644 call do_exit
645 CFI_ENDPROC
646 ENDPROC(child_rip)
647 --- sle11-2009-05-14.orig/arch/x86/kernel/genapic_64-xen.c 2008-12-15 11:27:22.000000000 +0100
648 +++ sle11-2009-05-14/arch/x86/kernel/genapic_64-xen.c 2009-02-16 16:18:36.000000000 +0100
649 @@ -24,12 +24,21 @@
650 #include <acpi/acpi_bus.h>
651 #endif
652
653 -/* which logical CPU number maps to which CPU (physical APIC ID) */
654 -u8 x86_cpu_to_apicid[NR_CPUS] __read_mostly
655 +/*
656 + * which logical CPU number maps to which CPU (physical APIC ID)
657 + *
658 + * The following static array is used during kernel startup
659 + * and the x86_cpu_to_apicid_ptr contains the address of the
660 + * array during this time. Is it zeroed when the per_cpu
661 + * data area is removed.
662 + */
663 +#ifndef CONFIG_XEN
664 +u8 x86_cpu_to_apicid_init[NR_CPUS] __initdata
665 = { [0 ... NR_CPUS-1] = BAD_APICID };
666 -EXPORT_SYMBOL(x86_cpu_to_apicid);
667 -
668 -u8 x86_cpu_to_log_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
669 +void *x86_cpu_to_apicid_ptr;
670 +#endif
671 +DEFINE_PER_CPU(u8, x86_cpu_to_apicid) = BAD_APICID;
672 +EXPORT_PER_CPU_SYMBOL(x86_cpu_to_apicid);
673
674 #ifndef CONFIG_XEN
675 struct genapic __read_mostly *genapic = &apic_flat;
676 --- sle11-2009-05-14.orig/arch/x86/kernel/head64-xen.c 2009-02-16 16:17:21.000000000 +0100
677 +++ sle11-2009-05-14/arch/x86/kernel/head64-xen.c 2009-02-16 16:18:36.000000000 +0100
678 @@ -1,5 +1,5 @@
679 /*
680 - * linux/arch/x86_64/kernel/head64.c -- prepare to run common code
681 + * prepare to run common code
682 *
683 * Copyright (C) 2000 Andrea Arcangeli <andrea@suse.de> SuSE
684 *
685 @@ -21,7 +21,6 @@
686 #include <asm/processor.h>
687 #include <asm/proto.h>
688 #include <asm/smp.h>
689 -#include <asm/bootsetup.h>
690 #include <asm/setup.h>
691 #include <asm/desc.h>
692 #include <asm/pgtable.h>
693 @@ -47,27 +46,16 @@ static void __init clear_bss(void)
694 }
695 #endif
696
697 -#define NEW_CL_POINTER 0x228 /* Relative to real mode data */
698 -#define OLD_CL_MAGIC_ADDR 0x20
699 -#define OLD_CL_MAGIC 0xA33F
700 -#define OLD_CL_OFFSET 0x22
701 -
702 static void __init copy_bootdata(char *real_mode_data)
703 {
704 #ifndef CONFIG_XEN
705 - unsigned long new_data;
706 char * command_line;
707
708 - memcpy(x86_boot_params, real_mode_data, BOOT_PARAM_SIZE);
709 - new_data = *(u32 *) (x86_boot_params + NEW_CL_POINTER);
710 - if (!new_data) {
711 - if (OLD_CL_MAGIC != *(u16 *)(real_mode_data + OLD_CL_MAGIC_ADDR)) {
712 - return;
713 - }
714 - new_data = __pa(real_mode_data) + *(u16 *)(real_mode_data + OLD_CL_OFFSET);
715 + memcpy(&boot_params, real_mode_data, sizeof boot_params);
716 + if (boot_params.hdr.cmd_line_ptr) {
717 + command_line = __va(boot_params.hdr.cmd_line_ptr);
718 + memcpy(boot_command_line, command_line, COMMAND_LINE_SIZE);
719 }
720 - command_line = __va(new_data);
721 - memcpy(boot_command_line, command_line, COMMAND_LINE_SIZE);
722 #else
723 int max_cmdline;
724
725 @@ -117,7 +105,7 @@ void __init x86_64_start_kernel(char * r
726
727 for (i = 0; i < IDT_ENTRIES; i++)
728 set_intr_gate(i, early_idt_handler);
729 - asm volatile("lidt %0" :: "m" (idt_descr));
730 + load_idt((const struct desc_ptr *)&idt_descr);
731 #endif
732
733 early_printk("Kernel alive\n");
734 --- sle11-2009-05-14.orig/arch/x86/kernel/init_task-xen.c 2009-02-16 16:17:21.000000000 +0100
735 +++ sle11-2009-05-14/arch/x86/kernel/init_task-xen.c 2009-02-16 16:18:36.000000000 +0100
736 @@ -14,11 +14,11 @@ static struct fs_struct init_fs = INIT_F
737 static struct files_struct init_files = INIT_FILES;
738 static struct signal_struct init_signals = INIT_SIGNALS(init_signals);
739 static struct sighand_struct init_sighand = INIT_SIGHAND(init_sighand);
740 -
741 +#ifdef CONFIG_X86_XEN
742 #define swapper_pg_dir ((pgd_t *)NULL)
743 +#endif
744 struct mm_struct init_mm = INIT_MM(init_mm);
745 #undef swapper_pg_dir
746 -
747 EXPORT_SYMBOL(init_mm);
748
749 /*
750 @@ -28,7 +28,7 @@ EXPORT_SYMBOL(init_mm);
751 * way process stacks are handled. This is done by having a special
752 * "init_task" linker map entry..
753 */
754 -union thread_union init_thread_union
755 +union thread_union init_thread_union
756 __attribute__((__section__(".data.init_task"))) =
757 { INIT_THREAD_INFO(init_task) };
758
759 @@ -38,14 +38,15 @@ union thread_union init_thread_union
760 * All other task structs will be allocated on slabs in fork.c
761 */
762 struct task_struct init_task = INIT_TASK(init_task);
763 -
764 EXPORT_SYMBOL(init_task);
765
766 #ifndef CONFIG_X86_NO_TSS
767 /*
768 - * per-CPU TSS segments. Threads are completely 'soft' on Linux,
769 - * no more per-task TSS's.
770 - */
771 + * no more per-task TSS's. The TSS size is kept cacheline-aligned
772 + * so they are allowed to end up in the .data.cacheline_aligned
773 + * section. Since TSS's are completely CPU-local, we want them
774 + * on exact cacheline boundaries, to eliminate cacheline ping-pong.
775 + */
776 DEFINE_PER_CPU_SHARED_ALIGNED(struct tss_struct, init_tss) = INIT_TSS;
777 #endif
778
779 --- sle11-2009-05-14.orig/arch/x86/kernel/io_apic_32-xen.c 2009-02-16 16:17:21.000000000 +0100
780 +++ sle11-2009-05-14/arch/x86/kernel/io_apic_32-xen.c 2009-02-16 16:18:36.000000000 +0100
781 @@ -427,7 +427,7 @@ static struct irq_cpu_info {
782
783 #define IRQ_ALLOWED(cpu, allowed_mask) cpu_isset(cpu, allowed_mask)
784
785 -#define CPU_TO_PACKAGEINDEX(i) (first_cpu(cpu_sibling_map[i]))
786 +#define CPU_TO_PACKAGEINDEX(i) (first_cpu(per_cpu(cpu_sibling_map, i)))
787
788 static cpumask_t balance_irq_affinity[NR_IRQS] = {
789 [0 ... NR_IRQS-1] = CPU_MASK_ALL
790 @@ -633,7 +633,7 @@ tryanotherirq:
791
792 imbalance = move_this_load;
793
794 - /* For physical_balance case, we accumlated both load
795 + /* For physical_balance case, we accumulated both load
796 * values in the one of the siblings cpu_irq[],
797 * to use the same code for physical and logical processors
798 * as much as possible.
799 @@ -647,7 +647,7 @@ tryanotherirq:
800 * (A+B)/2 vs B
801 */
802 load = CPU_IRQ(min_loaded) >> 1;
803 - for_each_cpu_mask(j, cpu_sibling_map[min_loaded]) {
804 + for_each_cpu_mask(j, per_cpu(cpu_sibling_map, min_loaded)) {
805 if (load > CPU_IRQ(j)) {
806 /* This won't change cpu_sibling_map[min_loaded] */
807 load = CPU_IRQ(j);
808 @@ -1018,7 +1018,7 @@ static int EISA_ELCR(unsigned int irq)
809 #define default_MCA_trigger(idx) (1)
810 #define default_MCA_polarity(idx) (0)
811
812 -static int __init MPBIOS_polarity(int idx)
813 +static int MPBIOS_polarity(int idx)
814 {
815 int bus = mp_irqs[idx].mpc_srcbus;
816 int polarity;
817 @@ -1347,6 +1347,11 @@ static void __init setup_IO_APIC_irqs(vo
818 continue;
819 }
820
821 + if (!first_notcon) {
822 + apic_printk(APIC_VERBOSE, " not connected.\n");
823 + first_notcon = 1;
824 + }
825 +
826 entry.trigger = irq_trigger(idx);
827 entry.polarity = irq_polarity(idx);
828
829 @@ -1936,13 +1941,16 @@ __setup("no_timer_check", notimercheck);
830 static int __init timer_irq_works(void)
831 {
832 unsigned long t1 = jiffies;
833 + unsigned long flags;
834
835 if (no_timer_check)
836 return 1;
837
838 + local_save_flags(flags);
839 local_irq_enable();
840 /* Let ten ticks pass... */
841 mdelay((10 * 1000) / HZ);
842 + local_irq_restore(flags);
843
844 /*
845 * Expect a few ticks at least, to be sure some possible
846 @@ -2223,6 +2231,9 @@ static inline void __init check_timer(vo
847 {
848 int apic1, pin1, apic2, pin2;
849 int vector;
850 + unsigned long flags;
851 +
852 + local_irq_save(flags);
853
854 /*
855 * get/set the timer IRQ vector:
856 @@ -2268,7 +2279,7 @@ static inline void __init check_timer(vo
857 }
858 if (disable_timer_pin_1 > 0)
859 clear_IO_APIC_pin(0, pin1);
860 - return;
861 + goto out;
862 }
863 clear_IO_APIC_pin(apic1, pin1);
864 printk(KERN_ERR "..MP-BIOS bug: 8254 timer not connected to "
865 @@ -2291,7 +2302,7 @@ static inline void __init check_timer(vo
866 if (nmi_watchdog == NMI_IO_APIC) {
867 setup_nmi();
868 }
869 - return;
870 + goto out;
871 }
872 /*
873 * Cleanup, just in case ...
874 @@ -2315,7 +2326,7 @@ static inline void __init check_timer(vo
875
876 if (timer_irq_works()) {
877 printk(" works.\n");
878 - return;
879 + goto out;
880 }
881 apic_write_around(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | vector);
882 printk(" failed.\n");
883 @@ -2331,11 +2342,13 @@ static inline void __init check_timer(vo
884
885 if (timer_irq_works()) {
886 printk(" works.\n");
887 - return;
888 + goto out;
889 }
890 printk(" failed :(.\n");
891 panic("IO-APIC + timer doesn't work! Boot with apic=debug and send a "
892 "report. Then try booting with the 'noapic' option");
893 +out:
894 + local_irq_restore(flags);
895 }
896 #else
897 int timer_uses_ioapic_pin_0 = 0;
898 @@ -2353,6 +2366,14 @@ int timer_uses_ioapic_pin_0 = 0;
899
900 void __init setup_IO_APIC(void)
901 {
902 +#ifndef CONFIG_XEN
903 + int i;
904 +
905 + /* Reserve all the system vectors. */
906 + for (i = FIRST_SYSTEM_VECTOR; i < NR_VECTORS; i++)
907 + set_bit(i, used_vectors);
908 +#endif
909 +
910 enable_IO_APIC();
911
912 if (acpi_ioapic)
913 @@ -2542,7 +2563,7 @@ void destroy_irq(unsigned int irq)
914 #endif /* CONFIG_XEN */
915
916 /*
917 - * MSI mesage composition
918 + * MSI message composition
919 */
920 #if defined(CONFIG_PCI_MSI) && !defined(CONFIG_XEN)
921 static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg)
922 @@ -2899,6 +2920,25 @@ int io_apic_set_pci_routing (int ioapic,
923 return 0;
924 }
925
926 +int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
927 +{
928 + int i;
929 +
930 + if (skip_ioapic_setup)
931 + return -1;
932 +
933 + for (i = 0; i < mp_irq_entries; i++)
934 + if (mp_irqs[i].mpc_irqtype == mp_INT &&
935 + mp_irqs[i].mpc_srcbusirq == bus_irq)
936 + break;
937 + if (i >= mp_irq_entries)
938 + return -1;
939 +
940 + *trigger = irq_trigger(i);
941 + *polarity = irq_polarity(i);
942 + return 0;
943 +}
944 +
945 #endif /* CONFIG_ACPI */
946
947 static int __init parse_disable_timer_pin_1(char *arg)
948 --- sle11-2009-05-14.orig/arch/x86/kernel/io_apic_64-xen.c 2009-02-16 16:17:21.000000000 +0100
949 +++ sle11-2009-05-14/arch/x86/kernel/io_apic_64-xen.c 2009-02-16 16:18:36.000000000 +0100
950 @@ -31,6 +31,7 @@
951 #include <linux/sysdev.h>
952 #include <linux/msi.h>
953 #include <linux/htirq.h>
954 +#include <linux/dmar.h>
955 #ifdef CONFIG_ACPI
956 #include <acpi/acpi_bus.h>
957 #endif
958 @@ -584,7 +585,7 @@ int IO_APIC_get_PCI_irq_vector(int bus,
959 #define default_PCI_trigger(idx) (1)
960 #define default_PCI_polarity(idx) (1)
961
962 -static int __init MPBIOS_polarity(int idx)
963 +static int MPBIOS_polarity(int idx)
964 {
965 int bus = mp_irqs[idx].mpc_srcbus;
966 int polarity;
967 @@ -871,6 +872,10 @@ static void __init setup_IO_APIC_irqs(vo
968 apic_printk(APIC_VERBOSE, ", %d-%d", mp_ioapics[apic].mpc_apicid, pin);
969 continue;
970 }
971 + if (!first_notcon) {
972 + apic_printk(APIC_VERBOSE, " not connected.\n");
973 + first_notcon = 1;
974 + }
975
976 irq = pin_2_irq(idx, apic, pin);
977 add_pin_to_irq(irq, apic, pin);
978 @@ -881,7 +886,7 @@ static void __init setup_IO_APIC_irqs(vo
979 }
980
981 if (!first_notcon)
982 - apic_printk(APIC_VERBOSE," not connected.\n");
983 + apic_printk(APIC_VERBOSE, " not connected.\n");
984 }
985
986 #ifndef CONFIG_XEN
987 @@ -1277,10 +1282,13 @@ void disable_IO_APIC(void)
988 static int __init timer_irq_works(void)
989 {
990 unsigned long t1 = jiffies;
991 + unsigned long flags;
992
993 + local_save_flags(flags);
994 local_irq_enable();
995 /* Let ten ticks pass... */
996 mdelay((10 * 1000) / HZ);
997 + local_irq_restore(flags);
998
999 /*
1000 * Expect a few ticks at least, to be sure some possible
1001 @@ -1655,6 +1663,9 @@ static inline void check_timer(void)
1002 {
1003 struct irq_cfg *cfg = irq_cfg + 0;
1004 int apic1, pin1, apic2, pin2;
1005 + unsigned long flags;
1006 +
1007 + local_irq_save(flags);
1008
1009 /*
1010 * get/set the timer IRQ vector:
1011 @@ -1696,7 +1707,7 @@ static inline void check_timer(void)
1012 }
1013 if (disable_timer_pin_1 > 0)
1014 clear_IO_APIC_pin(0, pin1);
1015 - return;
1016 + goto out;
1017 }
1018 clear_IO_APIC_pin(apic1, pin1);
1019 apic_printk(APIC_QUIET,KERN_ERR "..MP-BIOS bug: 8254 timer not "
1020 @@ -1718,7 +1729,7 @@ static inline void check_timer(void)
1021 if (nmi_watchdog == NMI_IO_APIC) {
1022 setup_nmi();
1023 }
1024 - return;
1025 + goto out;
1026 }
1027 /*
1028 * Cleanup, just in case ...
1029 @@ -1741,7 +1752,7 @@ static inline void check_timer(void)
1030
1031 if (timer_irq_works()) {
1032 apic_printk(APIC_VERBOSE," works.\n");
1033 - return;
1034 + goto out;
1035 }
1036 apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | cfg->vector);
1037 apic_printk(APIC_VERBOSE," failed.\n");
1038 @@ -1756,10 +1767,12 @@ static inline void check_timer(void)
1039
1040 if (timer_irq_works()) {
1041 apic_printk(APIC_VERBOSE," works.\n");
1042 - return;
1043 + goto out;
1044 }
1045 apic_printk(APIC_VERBOSE," failed :(.\n");
1046 panic("IO-APIC + timer doesn't work! Try using the 'noapic' kernel parameter\n");
1047 +out:
1048 + local_irq_restore(flags);
1049 }
1050 #else
1051 #define check_timer() ((void)0)
1052 @@ -1775,7 +1788,7 @@ __setup("no_timer_check", notimercheck);
1053
1054 /*
1055 *
1056 - * IRQ's that are handled by the PIC in the MPS IOAPIC case.
1057 + * IRQs that are handled by the PIC in the MPS IOAPIC case.
1058 * - IRQ2 is the cascade IRQ, and cannot be a io-apic IRQ.
1059 * Linux doesn't really care, as it's not actually used
1060 * for any interrupt handling anyway.
1061 @@ -1858,7 +1871,7 @@ static struct sysdev_class ioapic_sysdev
1062 static int __init ioapic_init_sysfs(void)
1063 {
1064 struct sys_device * dev;
1065 - int i, size, error = 0;
1066 + int i, size, error;
1067
1068 error = sysdev_class_register(&ioapic_sysdev_class);
1069 if (error)
1070 @@ -1867,12 +1880,11 @@ static int __init ioapic_init_sysfs(void
1071 for (i = 0; i < nr_ioapics; i++ ) {
1072 size = sizeof(struct sys_device) + nr_ioapic_registers[i]
1073 * sizeof(struct IO_APIC_route_entry);
1074 - mp_ioapic_data[i] = kmalloc(size, GFP_KERNEL);
1075 + mp_ioapic_data[i] = kzalloc(size, GFP_KERNEL);
1076 if (!mp_ioapic_data[i]) {
1077 printk(KERN_ERR "Can't suspend/resume IOAPIC %d\n", i);
1078 continue;
1079 }
1080 - memset(mp_ioapic_data[i], 0, size);
1081 dev = &mp_ioapic_data[i]->dev;
1082 dev->id = i;
1083 dev->cls = &ioapic_sysdev_class;
1084 @@ -1933,7 +1945,7 @@ void destroy_irq(unsigned int irq)
1085 #endif /* CONFIG_XEN */
1086
1087 /*
1088 - * MSI mesage composition
1089 + * MSI message composition
1090 */
1091 #if defined(CONFIG_PCI_MSI) && !defined(CONFIG_XEN)
1092 static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg)
1093 @@ -2043,8 +2055,64 @@ void arch_teardown_msi_irq(unsigned int
1094 destroy_irq(irq);
1095 }
1096
1097 -#endif /* CONFIG_PCI_MSI */
1098 +#ifdef CONFIG_DMAR
1099 +#ifdef CONFIG_SMP
1100 +static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask)
1101 +{
1102 + struct irq_cfg *cfg = irq_cfg + irq;
1103 + struct msi_msg msg;
1104 + unsigned int dest;
1105 + cpumask_t tmp;
1106 +
1107 + cpus_and(tmp, mask, cpu_online_map);
1108 + if (cpus_empty(tmp))
1109 + return;
1110 +
1111 + if (assign_irq_vector(irq, mask))
1112 + return;
1113 +
1114 + cpus_and(tmp, cfg->domain, mask);
1115 + dest = cpu_mask_to_apicid(tmp);
1116 +
1117 + dmar_msi_read(irq, &msg);
1118 +
1119 + msg.data &= ~MSI_DATA_VECTOR_MASK;
1120 + msg.data |= MSI_DATA_VECTOR(cfg->vector);
1121 + msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK;
1122 + msg.address_lo |= MSI_ADDR_DEST_ID(dest);
1123 +
1124 + dmar_msi_write(irq, &msg);
1125 + irq_desc[irq].affinity = mask;
1126 +}
1127 +#endif /* CONFIG_SMP */
1128 +
1129 +struct irq_chip dmar_msi_type = {
1130 + .name = "DMAR_MSI",
1131 + .unmask = dmar_msi_unmask,
1132 + .mask = dmar_msi_mask,
1133 + .ack = ack_apic_edge,
1134 +#ifdef CONFIG_SMP
1135 + .set_affinity = dmar_msi_set_affinity,
1136 +#endif
1137 + .retrigger = ioapic_retrigger_irq,
1138 +};
1139 +
1140 +int arch_setup_dmar_msi(unsigned int irq)
1141 +{
1142 + int ret;
1143 + struct msi_msg msg;
1144 +
1145 + ret = msi_compose_msg(NULL, irq, &msg);
1146 + if (ret < 0)
1147 + return ret;
1148 + dmar_msi_write(irq, &msg);
1149 + set_irq_chip_and_handler_name(irq, &dmar_msi_type, handle_edge_irq,
1150 + "edge");
1151 + return 0;
1152 +}
1153 +#endif
1154
1155 +#endif /* CONFIG_PCI_MSI */
1156 /*
1157 * Hypertransport interrupt support
1158 */
1159 @@ -2177,8 +2245,27 @@ int io_apic_set_pci_routing (int ioapic,
1160 return 0;
1161 }
1162
1163 -#endif /* CONFIG_ACPI */
1164
1165 +int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
1166 +{
1167 + int i;
1168 +
1169 + if (skip_ioapic_setup)
1170 + return -1;
1171 +
1172 + for (i = 0; i < mp_irq_entries; i++)
1173 + if (mp_irqs[i].mpc_irqtype == mp_INT &&
1174 + mp_irqs[i].mpc_srcbusirq == bus_irq)
1175 + break;
1176 + if (i >= mp_irq_entries)
1177 + return -1;
1178 +
1179 + *trigger = irq_trigger(i);
1180 + *polarity = irq_polarity(i);
1181 + return 0;
1182 +}
1183 +
1184 +#endif /* CONFIG_ACPI */
1185
1186 #ifndef CONFIG_XEN
1187 /*
1188 @@ -2217,3 +2304,4 @@ void __init setup_ioapic_dest(void)
1189 }
1190 #endif
1191 #endif /* !CONFIG_XEN */
1192 +
1193 --- sle11-2009-05-14.orig/arch/x86/kernel/ioport_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1194 +++ sle11-2009-05-14/arch/x86/kernel/ioport_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1195 @@ -1,6 +1,4 @@
1196 /*
1197 - * linux/arch/i386/kernel/ioport.c
1198 - *
1199 * This contains the io-permission bitmap code - written by obz, with changes
1200 * by Linus.
1201 */
1202 --- sle11-2009-05-14.orig/arch/x86/kernel/ioport_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1203 +++ sle11-2009-05-14/arch/x86/kernel/ioport_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1204 @@ -1,6 +1,4 @@
1205 /*
1206 - * linux/arch/x86_64/kernel/ioport.c
1207 - *
1208 * This contains the io-permission bitmap code - written by obz, with changes
1209 * by Linus.
1210 */
1211 --- sle11-2009-05-14.orig/arch/x86/kernel/irq_32-xen.c 2009-02-16 16:17:21.000000000 +0100
1212 +++ sle11-2009-05-14/arch/x86/kernel/irq_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1213 @@ -1,6 +1,4 @@
1214 /*
1215 - * linux/arch/i386/kernel/irq.c
1216 - *
1217 * Copyright (C) 1992, 1998 Linus Torvalds, Ingo Molnar
1218 *
1219 * This file contains the lowest level x86-specific interrupt
1220 @@ -231,8 +229,6 @@ asmlinkage void do_softirq(void)
1221
1222 local_irq_restore(flags);
1223 }
1224 -
1225 -EXPORT_SYMBOL(do_softirq);
1226 #endif
1227
1228 /*
1229 @@ -259,9 +255,17 @@ int show_interrupts(struct seq_file *p,
1230 }
1231
1232 if (i < NR_IRQS) {
1233 + unsigned any_count = 0;
1234 +
1235 spin_lock_irqsave(&irq_desc[i].lock, flags);
1236 +#ifndef CONFIG_SMP
1237 + any_count = kstat_irqs(i);
1238 +#else
1239 + for_each_online_cpu(j)
1240 + any_count |= kstat_cpu(j).irqs[i];
1241 +#endif
1242 action = irq_desc[i].action;
1243 - if (!action)
1244 + if (!action && !any_count)
1245 goto skip;
1246 seq_printf(p, "%3d: ",i);
1247 #ifndef CONFIG_SMP
1248 @@ -272,10 +276,12 @@ int show_interrupts(struct seq_file *p,
1249 #endif
1250 seq_printf(p, " %8s", irq_desc[i].chip->name);
1251 seq_printf(p, "-%-8s", irq_desc[i].name);
1252 - seq_printf(p, " %s", action->name);
1253
1254 - for (action=action->next; action; action = action->next)
1255 - seq_printf(p, ", %s", action->name);
1256 + if (action) {
1257 + seq_printf(p, " %s", action->name);
1258 + while ((action = action->next) != NULL)
1259 + seq_printf(p, ", %s", action->name);
1260 + }
1261
1262 seq_putc(p, '\n');
1263 skip:
1264 @@ -284,13 +290,46 @@ skip:
1265 seq_printf(p, "NMI: ");
1266 for_each_online_cpu(j)
1267 seq_printf(p, "%10u ", nmi_count(j));
1268 - seq_putc(p, '\n');
1269 + seq_printf(p, " Non-maskable interrupts\n");
1270 #ifdef CONFIG_X86_LOCAL_APIC
1271 seq_printf(p, "LOC: ");
1272 for_each_online_cpu(j)
1273 seq_printf(p, "%10u ",
1274 per_cpu(irq_stat,j).apic_timer_irqs);
1275 - seq_putc(p, '\n');
1276 + seq_printf(p, " Local timer interrupts\n");
1277 +#endif
1278 +#ifdef CONFIG_SMP
1279 + seq_printf(p, "RES: ");
1280 + for_each_online_cpu(j)
1281 + seq_printf(p, "%10u ",
1282 + per_cpu(irq_stat,j).irq_resched_count);
1283 + seq_printf(p, " Rescheduling interrupts\n");
1284 + seq_printf(p, "CAL: ");
1285 + for_each_online_cpu(j)
1286 + seq_printf(p, "%10u ",
1287 + per_cpu(irq_stat,j).irq_call_count);
1288 + seq_printf(p, " function call interrupts\n");
1289 +#ifndef CONFIG_XEN
1290 + seq_printf(p, "TLB: ");
1291 + for_each_online_cpu(j)
1292 + seq_printf(p, "%10u ",
1293 + per_cpu(irq_stat,j).irq_tlb_count);
1294 + seq_printf(p, " TLB shootdowns\n");
1295 +#endif
1296 +#endif
1297 +#ifdef CONFIG_X86_MCE
1298 + seq_printf(p, "TRM: ");
1299 + for_each_online_cpu(j)
1300 + seq_printf(p, "%10u ",
1301 + per_cpu(irq_stat,j).irq_thermal_count);
1302 + seq_printf(p, " Thermal event interrupts\n");
1303 +#endif
1304 +#ifdef CONFIG_X86_LOCAL_APIC
1305 + seq_printf(p, "SPU: ");
1306 + for_each_online_cpu(j)
1307 + seq_printf(p, "%10u ",
1308 + per_cpu(irq_stat,j).irq_spurious_count);
1309 + seq_printf(p, " Spurious interrupts\n");
1310 #endif
1311 seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count));
1312 #if defined(CONFIG_X86_IO_APIC)
1313 --- sle11-2009-05-14.orig/arch/x86/kernel/irq_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1314 +++ sle11-2009-05-14/arch/x86/kernel/irq_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1315 @@ -1,6 +1,4 @@
1316 /*
1317 - * linux/arch/x86_64/kernel/irq.c
1318 - *
1319 * Copyright (C) 1992, 1998 Linus Torvalds, Ingo Molnar
1320 *
1321 * This file contains the lowest level x86_64-specific interrupt
1322 @@ -64,9 +62,17 @@ int show_interrupts(struct seq_file *p,
1323 }
1324
1325 if (i < NR_IRQS) {
1326 + unsigned any_count = 0;
1327 +
1328 spin_lock_irqsave(&irq_desc[i].lock, flags);
1329 +#ifndef CONFIG_SMP
1330 + any_count = kstat_irqs(i);
1331 +#else
1332 + for_each_online_cpu(j)
1333 + any_count |= kstat_cpu(j).irqs[i];
1334 +#endif
1335 action = irq_desc[i].action;
1336 - if (!action)
1337 + if (!action && !any_count)
1338 goto skip;
1339 seq_printf(p, "%3d: ",i);
1340 #ifndef CONFIG_SMP
1341 @@ -78,9 +84,11 @@ int show_interrupts(struct seq_file *p,
1342 seq_printf(p, " %8s", irq_desc[i].chip->name);
1343 seq_printf(p, "-%-8s", irq_desc[i].name);
1344
1345 - seq_printf(p, " %s", action->name);
1346 - for (action=action->next; action; action = action->next)
1347 - seq_printf(p, ", %s", action->name);
1348 + if (action) {
1349 + seq_printf(p, " %s", action->name);
1350 + while ((action = action->next) != NULL)
1351 + seq_printf(p, ", %s", action->name);
1352 + }
1353 seq_putc(p, '\n');
1354 skip:
1355 spin_unlock_irqrestore(&irq_desc[i].lock, flags);
1356 @@ -88,12 +96,44 @@ skip:
1357 seq_printf(p, "NMI: ");
1358 for_each_online_cpu(j)
1359 seq_printf(p, "%10u ", cpu_pda(j)->__nmi_count);
1360 - seq_putc(p, '\n');
1361 + seq_printf(p, " Non-maskable interrupts\n");
1362 #ifdef CONFIG_X86_LOCAL_APIC
1363 seq_printf(p, "LOC: ");
1364 for_each_online_cpu(j)
1365 seq_printf(p, "%10u ", cpu_pda(j)->apic_timer_irqs);
1366 - seq_putc(p, '\n');
1367 + seq_printf(p, " Local timer interrupts\n");
1368 +#endif
1369 +#ifdef CONFIG_SMP
1370 + seq_printf(p, "RES: ");
1371 + for_each_online_cpu(j)
1372 + seq_printf(p, "%10u ", cpu_pda(j)->irq_resched_count);
1373 + seq_printf(p, " Rescheduling interrupts\n");
1374 + seq_printf(p, "CAL: ");
1375 + for_each_online_cpu(j)
1376 + seq_printf(p, "%10u ", cpu_pda(j)->irq_call_count);
1377 + seq_printf(p, " function call interrupts\n");
1378 +#ifndef CONFIG_XEN
1379 + seq_printf(p, "TLB: ");
1380 + for_each_online_cpu(j)
1381 + seq_printf(p, "%10u ", cpu_pda(j)->irq_tlb_count);
1382 + seq_printf(p, " TLB shootdowns\n");
1383 +#endif
1384 +#endif
1385 +#ifdef CONFIG_X86_MCE
1386 + seq_printf(p, "TRM: ");
1387 + for_each_online_cpu(j)
1388 + seq_printf(p, "%10u ", cpu_pda(j)->irq_thermal_count);
1389 + seq_printf(p, " Thermal event interrupts\n");
1390 + seq_printf(p, "THR: ");
1391 + for_each_online_cpu(j)
1392 + seq_printf(p, "%10u ", cpu_pda(j)->irq_threshold_count);
1393 + seq_printf(p, " Threshold APIC interrupts\n");
1394 +#endif
1395 +#ifdef CONFIG_X86_LOCAL_APIC
1396 + seq_printf(p, "SPU: ");
1397 + for_each_online_cpu(j)
1398 + seq_printf(p, "%10u ", cpu_pda(j)->irq_spurious_count);
1399 + seq_printf(p, " Spurious interrupts\n");
1400 #endif
1401 seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count));
1402 }
1403 @@ -211,7 +251,6 @@ asmlinkage void do_softirq(void)
1404 }
1405 local_irq_restore(flags);
1406 }
1407 -EXPORT_SYMBOL(do_softirq);
1408
1409 #ifndef CONFIG_X86_LOCAL_APIC
1410 /*
1411 --- sle11-2009-05-14.orig/arch/x86/kernel/ldt_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1412 +++ sle11-2009-05-14/arch/x86/kernel/ldt_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1413 @@ -1,6 +1,4 @@
1414 /*
1415 - * linux/arch/i386/kernel/ldt.c
1416 - *
1417 * Copyright (C) 1992 Krishna Balasubramanian and Linus Torvalds
1418 * Copyright (C) 1999 Ingo Molnar <mingo@redhat.com>
1419 */
1420 @@ -106,14 +104,14 @@ int init_new_context(struct task_struct
1421 struct mm_struct * old_mm;
1422 int retval = 0;
1423
1424 - init_MUTEX(&mm->context.sem);
1425 + mutex_init(&mm->context.lock);
1426 mm->context.size = 0;
1427 mm->context.has_foreign_mappings = 0;
1428 old_mm = current->mm;
1429 if (old_mm && old_mm->context.size > 0) {
1430 - down(&old_mm->context.sem);
1431 + mutex_lock(&old_mm->context.lock);
1432 retval = copy_ldt(&mm->context, &old_mm->context);
1433 - up(&old_mm->context.sem);
1434 + mutex_unlock(&old_mm->context.lock);
1435 }
1436 return retval;
1437 }
1438 @@ -149,7 +147,7 @@ static int read_ldt(void __user * ptr, u
1439 if (bytecount > LDT_ENTRY_SIZE*LDT_ENTRIES)
1440 bytecount = LDT_ENTRY_SIZE*LDT_ENTRIES;
1441
1442 - down(&mm->context.sem);
1443 + mutex_lock(&mm->context.lock);
1444 size = mm->context.size*LDT_ENTRY_SIZE;
1445 if (size > bytecount)
1446 size = bytecount;
1447 @@ -157,7 +155,7 @@ static int read_ldt(void __user * ptr, u
1448 err = 0;
1449 if (copy_to_user(ptr, mm->context.ldt, size))
1450 err = -EFAULT;
1451 - up(&mm->context.sem);
1452 + mutex_unlock(&mm->context.lock);
1453 if (err < 0)
1454 goto error_return;
1455 if (size != bytecount) {
1456 @@ -213,7 +211,7 @@ static int write_ldt(void __user * ptr,
1457 goto out;
1458 }
1459
1460 - down(&mm->context.sem);
1461 + mutex_lock(&mm->context.lock);
1462 if (ldt_info.entry_number >= mm->context.size) {
1463 error = alloc_ldt(&current->mm->context, ldt_info.entry_number+1, 1);
1464 if (error < 0)
1465 @@ -240,7 +238,7 @@ install:
1466 entry_1, entry_2);
1467
1468 out_unlock:
1469 - up(&mm->context.sem);
1470 + mutex_unlock(&mm->context.lock);
1471 out:
1472 return error;
1473 }
1474 --- sle11-2009-05-14.orig/arch/x86/kernel/ldt_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1475 +++ sle11-2009-05-14/arch/x86/kernel/ldt_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1476 @@ -1,6 +1,4 @@
1477 /*
1478 - * linux/arch/x86_64/kernel/ldt.c
1479 - *
1480 * Copyright (C) 1992 Krishna Balasubramanian and Linus Torvalds
1481 * Copyright (C) 1999 Ingo Molnar <mingo@redhat.com>
1482 * Copyright (C) 2002 Andi Kleen
1483 @@ -112,19 +110,14 @@ int init_new_context(struct task_struct
1484 int retval = 0;
1485
1486 memset(&mm->context, 0, sizeof(mm->context));
1487 - init_MUTEX(&mm->context.sem);
1488 + mutex_init(&mm->context.lock);
1489 old_mm = current->mm;
1490 if (old_mm)
1491 mm->context.vdso = old_mm->context.vdso;
1492 if (old_mm && old_mm->context.size > 0) {
1493 - down(&old_mm->context.sem);
1494 + mutex_lock(&old_mm->context.lock);
1495 retval = copy_ldt(&mm->context, &old_mm->context);
1496 - up(&old_mm->context.sem);
1497 - }
1498 - if (retval == 0) {
1499 - spin_lock(&mm_unpinned_lock);
1500 - list_add(&mm->context.unpinned, &mm_unpinned);
1501 - spin_unlock(&mm_unpinned_lock);
1502 + mutex_unlock(&old_mm->context.lock);
1503 }
1504 return retval;
1505 }
1506 @@ -148,11 +141,6 @@ void destroy_context(struct mm_struct *m
1507 kfree(mm->context.ldt);
1508 mm->context.size = 0;
1509 }
1510 - if (!PagePinned(virt_to_page(mm->pgd))) {
1511 - spin_lock(&mm_unpinned_lock);
1512 - list_del(&mm->context.unpinned);
1513 - spin_unlock(&mm_unpinned_lock);
1514 - }
1515 }
1516
1517 static int read_ldt(void __user * ptr, unsigned long bytecount)
1518 @@ -166,7 +154,7 @@ static int read_ldt(void __user * ptr, u
1519 if (bytecount > LDT_ENTRY_SIZE*LDT_ENTRIES)
1520 bytecount = LDT_ENTRY_SIZE*LDT_ENTRIES;
1521
1522 - down(&mm->context.sem);
1523 + mutex_lock(&mm->context.lock);
1524 size = mm->context.size*LDT_ENTRY_SIZE;
1525 if (size > bytecount)
1526 size = bytecount;
1527 @@ -174,7 +162,7 @@ static int read_ldt(void __user * ptr, u
1528 err = 0;
1529 if (copy_to_user(ptr, mm->context.ldt, size))
1530 err = -EFAULT;
1531 - up(&mm->context.sem);
1532 + mutex_unlock(&mm->context.lock);
1533 if (err < 0)
1534 goto error_return;
1535 if (size != bytecount) {
1536 @@ -227,7 +215,7 @@ static int write_ldt(void __user * ptr,
1537 goto out;
1538 }
1539
1540 - down(&mm->context.sem);
1541 + mutex_lock(&mm->context.lock);
1542 if (ldt_info.entry_number >= (unsigned)mm->context.size) {
1543 error = alloc_ldt(&current->mm->context, ldt_info.entry_number+1, 1);
1544 if (error < 0)
1545 @@ -256,7 +244,7 @@ install:
1546 error = HYPERVISOR_update_descriptor(mach_lp, (unsigned long)((entry_1 | (unsigned long) entry_2 << 32)));
1547
1548 out_unlock:
1549 - up(&mm->context.sem);
1550 + mutex_unlock(&mm->context.lock);
1551 out:
1552 return error;
1553 }
1554 --- sle11-2009-05-14.orig/arch/x86/kernel/mpparse_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1555 +++ sle11-2009-05-14/arch/x86/kernel/mpparse_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1556 @@ -1023,7 +1023,7 @@ void __init mp_config_acpi_legacy_irqs (
1557
1558 /*
1559 * Use the default configuration for the IRQs 0-15. Unless
1560 - * overriden by (MADT) interrupt source override entries.
1561 + * overridden by (MADT) interrupt source override entries.
1562 */
1563 for (i = 0; i < 16; i++) {
1564 int idx;
1565 --- sle11-2009-05-14.orig/arch/x86/kernel/mpparse_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1566 +++ sle11-2009-05-14/arch/x86/kernel/mpparse_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1567 @@ -57,6 +57,8 @@ unsigned long mp_lapic_addr = 0;
1568
1569 /* Processor that is doing the boot up */
1570 unsigned int boot_cpu_id = -1U;
1571 +EXPORT_SYMBOL(boot_cpu_id);
1572 +
1573 /* Internal processor count */
1574 unsigned int num_processors __cpuinitdata = 0;
1575
1576 @@ -87,7 +89,7 @@ static int __init mpf_checksum(unsigned
1577 }
1578
1579 #ifndef CONFIG_XEN
1580 -static void __cpuinit MP_processor_info (struct mpc_config_processor *m)
1581 +static void __cpuinit MP_processor_info(struct mpc_config_processor *m)
1582 {
1583 int cpu;
1584 cpumask_t tmp_map;
1585 @@ -124,13 +126,24 @@ static void __cpuinit MP_processor_info
1586 cpu = 0;
1587 }
1588 bios_cpu_apicid[cpu] = m->mpc_apicid;
1589 - x86_cpu_to_apicid[cpu] = m->mpc_apicid;
1590 + /*
1591 + * We get called early in the the start_kernel initialization
1592 + * process when the per_cpu data area is not yet setup, so we
1593 + * use a static array that is removed after the per_cpu data
1594 + * area is created.
1595 + */
1596 + if (x86_cpu_to_apicid_ptr) {
1597 + u8 *x86_cpu_to_apicid = (u8 *)x86_cpu_to_apicid_ptr;
1598 + x86_cpu_to_apicid[cpu] = m->mpc_apicid;
1599 + } else {
1600 + per_cpu(x86_cpu_to_apicid, cpu) = m->mpc_apicid;
1601 + }
1602
1603 cpu_set(cpu, cpu_possible_map);
1604 cpu_set(cpu, cpu_present_map);
1605 }
1606 #else
1607 -static void __cpuinit MP_processor_info (struct mpc_config_processor *m)
1608 +static void __cpuinit MP_processor_info(struct mpc_config_processor *m)
1609 {
1610 num_processors++;
1611 }
1612 --- sle11-2009-05-14.orig/arch/x86/kernel/pci-dma-xen.c 2009-02-16 16:17:21.000000000 +0100
1613 +++ sle11-2009-05-14/arch/x86/kernel/pci-dma-xen.c 2009-02-16 16:18:36.000000000 +0100
1614 @@ -13,14 +13,13 @@
1615 #include <linux/pci.h>
1616 #include <linux/module.h>
1617 #include <linux/version.h>
1618 -#include <linux/pci.h>
1619 #include <asm/io.h>
1620 #include <xen/balloon.h>
1621 #include <xen/gnttab.h>
1622 #include <asm/swiotlb.h>
1623 #include <asm/tlbflush.h>
1624 -#include <asm-i386/mach-xen/asm/swiotlb.h>
1625 -#include <asm-i386/mach-xen/asm/gnttab_dma.h>
1626 +#include <asm/swiotlb_32.h>
1627 +#include <asm/gnttab_dma.h>
1628 #include <asm/bug.h>
1629
1630 #ifdef __x86_64__
1631 @@ -106,27 +105,29 @@ int range_straddles_page_boundary(paddr_
1632 }
1633
1634 int
1635 -dma_map_sg(struct device *hwdev, struct scatterlist *sg, int nents,
1636 +dma_map_sg(struct device *hwdev, struct scatterlist *sgl, int nents,
1637 enum dma_data_direction direction)
1638 {
1639 int i, rc;
1640
1641 BUG_ON(!valid_dma_direction(direction));
1642 - WARN_ON(nents == 0 || sg[0].length == 0);
1643 + WARN_ON(nents == 0 || sgl->length == 0);
1644
1645 if (swiotlb) {
1646 - rc = swiotlb_map_sg(hwdev, sg, nents, direction);
1647 + rc = swiotlb_map_sg(hwdev, sgl, nents, direction);
1648 } else {
1649 - for (i = 0; i < nents; i++ ) {
1650 - BUG_ON(!sg[i].page);
1651 - sg[i].dma_address =
1652 - gnttab_dma_map_page(sg[i].page) + sg[i].offset;
1653 - sg[i].dma_length = sg[i].length;
1654 + struct scatterlist *sg;
1655 +
1656 + for_each_sg(sgl, sg, nents, i) {
1657 + BUG_ON(!sg_page(sg));
1658 + sg->dma_address =
1659 + gnttab_dma_map_page(sg_page(sg)) + sg->offset;
1660 + sg->dma_length = sg->length;
1661 IOMMU_BUG_ON(address_needs_mapping(
1662 - hwdev, sg[i].dma_address));
1663 + hwdev, sg->dma_address));
1664 IOMMU_BUG_ON(range_straddles_page_boundary(
1665 - page_to_pseudophys(sg[i].page) + sg[i].offset,
1666 - sg[i].length));
1667 + page_to_pseudophys(sg_page(sg)) + sg->offset,
1668 + sg->length));
1669 }
1670 rc = nents;
1671 }
1672 @@ -137,17 +138,19 @@ dma_map_sg(struct device *hwdev, struct
1673 EXPORT_SYMBOL(dma_map_sg);
1674
1675 void
1676 -dma_unmap_sg(struct device *hwdev, struct scatterlist *sg, int nents,
1677 +dma_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nents,
1678 enum dma_data_direction direction)
1679 {
1680 int i;
1681
1682 BUG_ON(!valid_dma_direction(direction));
1683 if (swiotlb)
1684 - swiotlb_unmap_sg(hwdev, sg, nents, direction);
1685 + swiotlb_unmap_sg(hwdev, sgl, nents, direction);
1686 else {
1687 - for (i = 0; i < nents; i++ )
1688 - gnttab_dma_unmap_page(sg[i].dma_address);
1689 + struct scatterlist *sg;
1690 +
1691 + for_each_sg(sgl, sg, nents, i)
1692 + gnttab_dma_unmap_page(sg->dma_address);
1693 }
1694 }
1695 EXPORT_SYMBOL(dma_unmap_sg);
1696 @@ -261,7 +264,8 @@ void dma_free_coherent(struct device *de
1697 {
1698 struct dma_coherent_mem *mem = dev ? dev->dma_mem : NULL;
1699 int order = get_order(size);
1700 -
1701 +
1702 + WARN_ON(irqs_disabled()); /* for portability */
1703 if (mem && vaddr >= mem->virt_base && vaddr < (mem->virt_base + (mem->size << PAGE_SHIFT))) {
1704 int page = (vaddr - mem->virt_base) >> PAGE_SHIFT;
1705
1706 --- sle11-2009-05-14.orig/arch/x86/kernel/process_32-xen.c 2009-02-16 16:17:21.000000000 +0100
1707 +++ sle11-2009-05-14/arch/x86/kernel/process_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1708 @@ -1,6 +1,4 @@
1709 /*
1710 - * linux/arch/i386/kernel/process.c
1711 - *
1712 * Copyright (C) 1995 Linus Torvalds
1713 *
1714 * Pentium III FXSR, SSE support
1715 @@ -190,6 +188,10 @@ void cpu_idle(void)
1716 }
1717 }
1718
1719 +static void do_nothing(void *unused)
1720 +{
1721 +}
1722 +
1723 void cpu_idle_wait(void)
1724 {
1725 unsigned int cpu, this_cpu = get_cpu();
1726 @@ -214,13 +216,20 @@ void cpu_idle_wait(void)
1727 cpu_clear(cpu, map);
1728 }
1729 cpus_and(map, map, cpu_online_map);
1730 + /*
1731 + * We waited 1 sec, if a CPU still did not call idle
1732 + * it may be because it is in idle and not waking up
1733 + * because it has nothing to do.
1734 + * Give all the remaining CPUS a kick.
1735 + */
1736 + smp_call_function_mask(map, do_nothing, 0, 0);
1737 } while (!cpus_empty(map));
1738
1739 set_cpus_allowed(current, tmp);
1740 }
1741 EXPORT_SYMBOL_GPL(cpu_idle_wait);
1742
1743 -void __devinit select_idle_routine(const struct cpuinfo_x86 *c)
1744 +void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
1745 {
1746 }
1747
1748 @@ -238,34 +247,52 @@ static int __init idle_setup(char *str)
1749 }
1750 early_param("idle", idle_setup);
1751
1752 -void show_regs(struct pt_regs * regs)
1753 +void __show_registers(struct pt_regs *regs, int all)
1754 {
1755 unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L;
1756 unsigned long d0, d1, d2, d3, d6, d7;
1757 + unsigned long esp;
1758 + unsigned short ss, gs;
1759 +
1760 + if (user_mode_vm(regs)) {
1761 + esp = regs->esp;
1762 + ss = regs->xss & 0xffff;
1763 + savesegment(gs, gs);
1764 + } else {
1765 + esp = (unsigned long) (&regs->esp);
1766 + savesegment(ss, ss);
1767 + savesegment(gs, gs);
1768 + }
1769
1770 printk("\n");
1771 - printk("Pid: %d, comm: %20s\n", current->pid, current->comm);
1772 - printk("EIP: %04x:[<%08lx>] CPU: %d\n",0xffff & regs->xcs,regs->eip, smp_processor_id());
1773 + printk("Pid: %d, comm: %s %s (%s %.*s)\n",
1774 + task_pid_nr(current), current->comm,
1775 + print_tainted(), init_utsname()->release,
1776 + (int)strcspn(init_utsname()->version, " "),
1777 + init_utsname()->version);
1778 +
1779 + printk("EIP: %04x:[<%08lx>] EFLAGS: %08lx CPU: %d\n",
1780 + 0xffff & regs->xcs, regs->eip, regs->eflags,
1781 + smp_processor_id());
1782 print_symbol("EIP is at %s\n", regs->eip);
1783
1784 - if (user_mode_vm(regs))
1785 - printk(" ESP: %04x:%08lx",0xffff & regs->xss,regs->esp);
1786 - printk(" EFLAGS: %08lx %s (%s %.*s)\n",
1787 - regs->eflags, print_tainted(), init_utsname()->release,
1788 - (int)strcspn(init_utsname()->version, " "),
1789 - init_utsname()->version);
1790 printk("EAX: %08lx EBX: %08lx ECX: %08lx EDX: %08lx\n",
1791 - regs->eax,regs->ebx,regs->ecx,regs->edx);
1792 - printk("ESI: %08lx EDI: %08lx EBP: %08lx",
1793 - regs->esi, regs->edi, regs->ebp);
1794 - printk(" DS: %04x ES: %04x FS: %04x\n",
1795 - 0xffff & regs->xds,0xffff & regs->xes, 0xffff & regs->xfs);
1796 + regs->eax, regs->ebx, regs->ecx, regs->edx);
1797 + printk("ESI: %08lx EDI: %08lx EBP: %08lx ESP: %08lx\n",
1798 + regs->esi, regs->edi, regs->ebp, esp);
1799 + printk(" DS: %04x ES: %04x FS: %04x GS: %04x SS: %04x\n",
1800 + regs->xds & 0xffff, regs->xes & 0xffff,
1801 + regs->xfs & 0xffff, gs, ss);
1802 +
1803 + if (!all)
1804 + return;
1805
1806 cr0 = read_cr0();
1807 cr2 = read_cr2();
1808 cr3 = read_cr3();
1809 cr4 = read_cr4_safe();
1810 - printk("CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n", cr0, cr2, cr3, cr4);
1811 + printk("CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n",
1812 + cr0, cr2, cr3, cr4);
1813
1814 get_debugreg(d0, 0);
1815 get_debugreg(d1, 1);
1816 @@ -273,10 +300,16 @@ void show_regs(struct pt_regs * regs)
1817 get_debugreg(d3, 3);
1818 printk("DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n",
1819 d0, d1, d2, d3);
1820 +
1821 get_debugreg(d6, 6);
1822 get_debugreg(d7, 7);
1823 - printk("DR6: %08lx DR7: %08lx\n", d6, d7);
1824 + printk("DR6: %08lx DR7: %08lx\n",
1825 + d6, d7);
1826 +}
1827
1828 +void show_regs(struct pt_regs *regs)
1829 +{
1830 + __show_registers(regs, 1);
1831 show_trace(NULL, regs, &regs->esp);
1832 }
1833
1834 --- sle11-2009-05-14.orig/arch/x86/kernel/process_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1835 +++ sle11-2009-05-14/arch/x86/kernel/process_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1836 @@ -1,6 +1,4 @@
1837 /*
1838 - * linux/arch/x86-64/kernel/process.c
1839 - *
1840 * Copyright (C) 1995 Linus Torvalds
1841 *
1842 * Pentium III FXSR, SSE support
1843 @@ -41,6 +39,7 @@
1844 #include <linux/notifier.h>
1845 #include <linux/kprobes.h>
1846 #include <linux/kdebug.h>
1847 +#include <linux/tick.h>
1848
1849 #include <asm/uaccess.h>
1850 #include <asm/pgtable.h>
1851 @@ -172,6 +171,9 @@ void cpu_idle (void)
1852
1853 if (__get_cpu_var(cpu_idle_state))
1854 __get_cpu_var(cpu_idle_state) = 0;
1855 +
1856 + tick_nohz_stop_sched_tick();
1857 +
1858 rmb();
1859 idle = xen_idle; /* no alternatives */
1860 if (cpu_is_offline(smp_processor_id()))
1861 @@ -190,12 +192,17 @@ void cpu_idle (void)
1862 __exit_idle();
1863 }
1864
1865 + tick_nohz_restart_sched_tick();
1866 preempt_enable_no_resched();
1867 schedule();
1868 preempt_disable();
1869 }
1870 }
1871
1872 +static void do_nothing(void *unused)
1873 +{
1874 +}
1875 +
1876 void cpu_idle_wait(void)
1877 {
1878 unsigned int cpu, this_cpu = get_cpu();
1879 @@ -221,6 +228,13 @@ void cpu_idle_wait(void)
1880 cpu_clear(cpu, map);
1881 }
1882 cpus_and(map, map, cpu_online_map);
1883 + /*
1884 + * We waited 1 sec, if a CPU still did not call idle
1885 + * it may be because it is in idle and not waking up
1886 + * because it has nothing to do.
1887 + * Give all the remaining CPUS a kick.
1888 + */
1889 + smp_call_function_mask(map, do_nothing, 0, 0);
1890 } while (!cpus_empty(map));
1891
1892 set_cpus_allowed(current, tmp);
1893 @@ -528,7 +542,7 @@ static inline void __switch_to_xtra(stru
1894 *
1895 * Kprobes not supported here. Set the probe on schedule instead.
1896 */
1897 -__kprobes struct task_struct *
1898 +struct task_struct *
1899 __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
1900 {
1901 struct thread_struct *prev = &prev_p->thread,
1902 --- sle11-2009-05-14.orig/arch/x86/kernel/quirks-xen.c 2008-12-15 11:27:22.000000000 +0100
1903 +++ sle11-2009-05-14/arch/x86/kernel/quirks-xen.c 2009-02-16 16:18:36.000000000 +0100
1904 @@ -41,7 +41,353 @@ static void __devinit quirk_intel_irqbal
1905 if (!(config & 0x2))
1906 pci_write_config_byte(dev, 0xf4, config);
1907 }
1908 -DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7320_MCH, quirk_intel_irqbalance);
1909 -DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7525_MCH, quirk_intel_irqbalance);
1910 -DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7520_MCH, quirk_intel_irqbalance);
1911 +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7320_MCH,
1912 + quirk_intel_irqbalance);
1913 +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7525_MCH,
1914 + quirk_intel_irqbalance);
1915 +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7520_MCH,
1916 + quirk_intel_irqbalance);
1917 +#endif
1918 +
1919 +#if defined(CONFIG_HPET_TIMER)
1920 +#include <asm/hpet.h>
1921 +
1922 +unsigned long force_hpet_address;
1923 +
1924 +static enum {
1925 + NONE_FORCE_HPET_RESUME,
1926 + OLD_ICH_FORCE_HPET_RESUME,
1927 + ICH_FORCE_HPET_RESUME,
1928 + VT8237_FORCE_HPET_RESUME,
1929 + NVIDIA_FORCE_HPET_RESUME,
1930 +} force_hpet_resume_type;
1931 +
1932 +static void __iomem *rcba_base;
1933 +
1934 +static void ich_force_hpet_resume(void)
1935 +{
1936 + u32 val;
1937 +
1938 + if (!force_hpet_address)
1939 + return;
1940 +
1941 + if (rcba_base == NULL)
1942 + BUG();
1943 +
1944 + /* read the Function Disable register, dword mode only */
1945 + val = readl(rcba_base + 0x3404);
1946 + if (!(val & 0x80)) {
1947 + /* HPET disabled in HPTC. Trying to enable */
1948 + writel(val | 0x80, rcba_base + 0x3404);
1949 + }
1950 +
1951 + val = readl(rcba_base + 0x3404);
1952 + if (!(val & 0x80))
1953 + BUG();
1954 + else
1955 + printk(KERN_DEBUG "Force enabled HPET at resume\n");
1956 +
1957 + return;
1958 +}
1959 +
1960 +static void ich_force_enable_hpet(struct pci_dev *dev)
1961 +{
1962 + u32 val;
1963 + u32 uninitialized_var(rcba);
1964 + int err = 0;
1965 +
1966 + if (hpet_address || force_hpet_address)
1967 + return;
1968 +
1969 + pci_read_config_dword(dev, 0xF0, &rcba);
1970 + rcba &= 0xFFFFC000;
1971 + if (rcba == 0) {
1972 + printk(KERN_DEBUG "RCBA disabled. Cannot force enable HPET\n");
1973 + return;
1974 + }
1975 +
1976 + /* use bits 31:14, 16 kB aligned */
1977 + rcba_base = ioremap_nocache(rcba, 0x4000);
1978 + if (rcba_base == NULL) {
1979 + printk(KERN_DEBUG "ioremap failed. Cannot force enable HPET\n");
1980 + return;
1981 + }
1982 +
1983 + /* read the Function Disable register, dword mode only */
1984 + val = readl(rcba_base + 0x3404);
1985 +
1986 + if (val & 0x80) {
1987 + /* HPET is enabled in HPTC. Just not reported by BIOS */
1988 + val = val & 0x3;
1989 + force_hpet_address = 0xFED00000 | (val << 12);
1990 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
1991 + force_hpet_address);
1992 + iounmap(rcba_base);
1993 + return;
1994 + }
1995 +
1996 + /* HPET disabled in HPTC. Trying to enable */
1997 + writel(val | 0x80, rcba_base + 0x3404);
1998 +
1999 + val = readl(rcba_base + 0x3404);
2000 + if (!(val & 0x80)) {
2001 + err = 1;
2002 + } else {
2003 + val = val & 0x3;
2004 + force_hpet_address = 0xFED00000 | (val << 12);
2005 + }
2006 +
2007 + if (err) {
2008 + force_hpet_address = 0;
2009 + iounmap(rcba_base);
2010 + printk(KERN_DEBUG "Failed to force enable HPET\n");
2011 + } else {
2012 + force_hpet_resume_type = ICH_FORCE_HPET_RESUME;
2013 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
2014 + force_hpet_address);
2015 + }
2016 +}
2017 +
2018 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ESB2_0,
2019 + ich_force_enable_hpet);
2020 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH6_1,
2021 + ich_force_enable_hpet);
2022 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_0,
2023 + ich_force_enable_hpet);
2024 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_1,
2025 + ich_force_enable_hpet);
2026 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_31,
2027 + ich_force_enable_hpet);
2028 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH8_1,
2029 + ich_force_enable_hpet);
2030 +
2031 +
2032 +static struct pci_dev *cached_dev;
2033 +
2034 +static void old_ich_force_hpet_resume(void)
2035 +{
2036 + u32 val;
2037 + u32 uninitialized_var(gen_cntl);
2038 +
2039 + if (!force_hpet_address || !cached_dev)
2040 + return;
2041 +
2042 + pci_read_config_dword(cached_dev, 0xD0, &gen_cntl);
2043 + gen_cntl &= (~(0x7 << 15));
2044 + gen_cntl |= (0x4 << 15);
2045 +
2046 + pci_write_config_dword(cached_dev, 0xD0, gen_cntl);
2047 + pci_read_config_dword(cached_dev, 0xD0, &gen_cntl);
2048 + val = gen_cntl >> 15;
2049 + val &= 0x7;
2050 + if (val == 0x4)
2051 + printk(KERN_DEBUG "Force enabled HPET at resume\n");
2052 + else
2053 + BUG();
2054 +}
2055 +
2056 +static void old_ich_force_enable_hpet(struct pci_dev *dev)
2057 +{
2058 + u32 val;
2059 + u32 uninitialized_var(gen_cntl);
2060 +
2061 + if (hpet_address || force_hpet_address)
2062 + return;
2063 +
2064 + pci_read_config_dword(dev, 0xD0, &gen_cntl);
2065 + /*
2066 + * Bit 17 is HPET enable bit.
2067 + * Bit 16:15 control the HPET base address.
2068 + */
2069 + val = gen_cntl >> 15;
2070 + val &= 0x7;
2071 + if (val & 0x4) {
2072 + val &= 0x3;
2073 + force_hpet_address = 0xFED00000 | (val << 12);
2074 + printk(KERN_DEBUG "HPET at base address 0x%lx\n",
2075 + force_hpet_address);
2076 + return;
2077 + }
2078 +
2079 + /*
2080 + * HPET is disabled. Trying enabling at FED00000 and check
2081 + * whether it sticks
2082 + */
2083 + gen_cntl &= (~(0x7 << 15));
2084 + gen_cntl |= (0x4 << 15);
2085 + pci_write_config_dword(dev, 0xD0, gen_cntl);
2086 +
2087 + pci_read_config_dword(dev, 0xD0, &gen_cntl);
2088 +
2089 + val = gen_cntl >> 15;
2090 + val &= 0x7;
2091 + if (val & 0x4) {
2092 + /* HPET is enabled in HPTC. Just not reported by BIOS */
2093 + val &= 0x3;
2094 + force_hpet_address = 0xFED00000 | (val << 12);
2095 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
2096 + force_hpet_address);
2097 + cached_dev = dev;
2098 + force_hpet_resume_type = OLD_ICH_FORCE_HPET_RESUME;
2099 + return;
2100 + }
2101 +
2102 + printk(KERN_DEBUG "Failed to force enable HPET\n");
2103 +}
2104 +
2105 +/*
2106 + * Undocumented chipset features. Make sure that the user enforced
2107 + * this.
2108 + */
2109 +static void old_ich_force_enable_hpet_user(struct pci_dev *dev)
2110 +{
2111 + if (hpet_force_user)
2112 + old_ich_force_enable_hpet(dev);
2113 +}
2114 +
2115 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801CA_0,
2116 + old_ich_force_enable_hpet_user);
2117 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801CA_12,
2118 + old_ich_force_enable_hpet_user);
2119 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801DB_0,
2120 + old_ich_force_enable_hpet_user);
2121 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801DB_12,
2122 + old_ich_force_enable_hpet_user);
2123 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801EB_0,
2124 + old_ich_force_enable_hpet);
2125 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801EB_12,
2126 + old_ich_force_enable_hpet);
2127 +
2128 +
2129 +static void vt8237_force_hpet_resume(void)
2130 +{
2131 + u32 val;
2132 +
2133 + if (!force_hpet_address || !cached_dev)
2134 + return;
2135 +
2136 + val = 0xfed00000 | 0x80;
2137 + pci_write_config_dword(cached_dev, 0x68, val);
2138 +
2139 + pci_read_config_dword(cached_dev, 0x68, &val);
2140 + if (val & 0x80)
2141 + printk(KERN_DEBUG "Force enabled HPET at resume\n");
2142 + else
2143 + BUG();
2144 +}
2145 +
2146 +static void vt8237_force_enable_hpet(struct pci_dev *dev)
2147 +{
2148 + u32 uninitialized_var(val);
2149 +
2150 + if (!hpet_force_user || hpet_address || force_hpet_address)
2151 + return;
2152 +
2153 + pci_read_config_dword(dev, 0x68, &val);
2154 + /*
2155 + * Bit 7 is HPET enable bit.
2156 + * Bit 31:10 is HPET base address (contrary to what datasheet claims)
2157 + */
2158 + if (val & 0x80) {
2159 + force_hpet_address = (val & ~0x3ff);
2160 + printk(KERN_DEBUG "HPET at base address 0x%lx\n",
2161 + force_hpet_address);
2162 + return;
2163 + }
2164 +
2165 + /*
2166 + * HPET is disabled. Trying enabling at FED00000 and check
2167 + * whether it sticks
2168 + */
2169 + val = 0xfed00000 | 0x80;
2170 + pci_write_config_dword(dev, 0x68, val);
2171 +
2172 + pci_read_config_dword(dev, 0x68, &val);
2173 + if (val & 0x80) {
2174 + force_hpet_address = (val & ~0x3ff);
2175 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
2176 + force_hpet_address);
2177 + cached_dev = dev;
2178 + force_hpet_resume_type = VT8237_FORCE_HPET_RESUME;
2179 + return;
2180 + }
2181 +
2182 + printk(KERN_DEBUG "Failed to force enable HPET\n");
2183 +}
2184 +
2185 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_VIA, PCI_DEVICE_ID_VIA_8235,
2186 + vt8237_force_enable_hpet);
2187 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_VIA, PCI_DEVICE_ID_VIA_8237,
2188 + vt8237_force_enable_hpet);
2189 +
2190 +/*
2191 + * Undocumented chipset feature taken from LinuxBIOS.
2192 + */
2193 +static void nvidia_force_hpet_resume(void)
2194 +{
2195 + pci_write_config_dword(cached_dev, 0x44, 0xfed00001);
2196 + printk(KERN_DEBUG "Force enabled HPET at resume\n");
2197 +}
2198 +
2199 +static void nvidia_force_enable_hpet(struct pci_dev *dev)
2200 +{
2201 + u32 uninitialized_var(val);
2202 +
2203 + if (!hpet_force_user || hpet_address || force_hpet_address)
2204 + return;
2205 +
2206 + pci_write_config_dword(dev, 0x44, 0xfed00001);
2207 + pci_read_config_dword(dev, 0x44, &val);
2208 + force_hpet_address = val & 0xfffffffe;
2209 + force_hpet_resume_type = NVIDIA_FORCE_HPET_RESUME;
2210 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
2211 + force_hpet_address);
2212 + cached_dev = dev;
2213 + return;
2214 +}
2215 +
2216 +/* ISA Bridges */
2217 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0050,
2218 + nvidia_force_enable_hpet);
2219 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0051,
2220 + nvidia_force_enable_hpet);
2221 +
2222 +/* LPC bridges */
2223 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0360,
2224 + nvidia_force_enable_hpet);
2225 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0361,
2226 + nvidia_force_enable_hpet);
2227 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0362,
2228 + nvidia_force_enable_hpet);
2229 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0363,
2230 + nvidia_force_enable_hpet);
2231 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0364,
2232 + nvidia_force_enable_hpet);
2233 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0365,
2234 + nvidia_force_enable_hpet);
2235 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0366,
2236 + nvidia_force_enable_hpet);
2237 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0367,
2238 + nvidia_force_enable_hpet);
2239 +
2240 +void force_hpet_resume(void)
2241 +{
2242 + switch (force_hpet_resume_type) {
2243 + case ICH_FORCE_HPET_RESUME:
2244 + return ich_force_hpet_resume();
2245 +
2246 + case OLD_ICH_FORCE_HPET_RESUME:
2247 + return old_ich_force_hpet_resume();
2248 +
2249 + case VT8237_FORCE_HPET_RESUME:
2250 + return vt8237_force_hpet_resume();
2251 +
2252 + case NVIDIA_FORCE_HPET_RESUME:
2253 + return nvidia_force_hpet_resume();
2254 +
2255 + default:
2256 + break;
2257 + }
2258 +}
2259 +
2260 #endif
2261 --- sle11-2009-05-14.orig/arch/x86/kernel/setup64-xen.c 2009-02-16 16:17:21.000000000 +0100
2262 +++ sle11-2009-05-14/arch/x86/kernel/setup64-xen.c 2009-02-16 16:18:36.000000000 +0100
2263 @@ -15,7 +15,6 @@
2264 #include <linux/bootmem.h>
2265 #include <linux/bitops.h>
2266 #include <linux/module.h>
2267 -#include <asm/bootsetup.h>
2268 #include <asm/pda.h>
2269 #include <asm/pgtable.h>
2270 #include <asm/processor.h>
2271 @@ -27,11 +26,12 @@
2272 #include <asm/percpu.h>
2273 #include <asm/proto.h>
2274 #include <asm/sections.h>
2275 +#include <asm/setup.h>
2276 #ifdef CONFIG_XEN
2277 #include <asm/hypervisor.h>
2278 #endif
2279
2280 -char x86_boot_params[BOOT_PARAM_SIZE] __initdata;
2281 +struct boot_params __initdata boot_params;
2282
2283 cpumask_t cpu_initialized __cpuinitdata = CPU_MASK_NONE;
2284
2285 @@ -159,8 +159,8 @@ static void switch_pt(void)
2286
2287 static void __cpuinit cpu_gdt_init(const struct desc_ptr *gdt_descr)
2288 {
2289 - asm volatile("lgdt %0" :: "m" (*gdt_descr));
2290 - asm volatile("lidt %0" :: "m" (idt_descr));
2291 + load_gdt(gdt_descr);
2292 + load_idt(idt_descr);
2293 }
2294 #endif
2295
2296 @@ -252,6 +252,14 @@ void __cpuinit check_efer(void)
2297
2298 unsigned long kernel_eflags;
2299
2300 +#ifndef CONFIG_X86_NO_TSS
2301 +/*
2302 + * Copies of the original ist values from the tss are only accessed during
2303 + * debugging, no special alignment required.
2304 + */
2305 +DEFINE_PER_CPU(struct orig_ist, orig_ist);
2306 +#endif
2307 +
2308 /*
2309 * cpu_init() initializes state that is per-CPU. Some data is already
2310 * initialized (naturally) in the bootstrap process, such as the GDT
2311 --- sle11-2009-05-14.orig/arch/x86/kernel/setup_32-xen.c 2009-02-16 16:17:21.000000000 +0100
2312 +++ sle11-2009-05-14/arch/x86/kernel/setup_32-xen.c 2009-02-16 16:18:36.000000000 +0100
2313 @@ -1,6 +1,4 @@
2314 /*
2315 - * linux/arch/i386/kernel/setup.c
2316 - *
2317 * Copyright (C) 1995 Linus Torvalds
2318 *
2319 * Support of BIGMEM added by Gerhard Wichert, Siemens AG, July 1999
2320 @@ -70,6 +68,7 @@
2321 #include <xen/xencons.h>
2322 #include <setup_arch.h>
2323 #include <bios_ebda.h>
2324 +#include <asm/cacheflush.h>
2325
2326 #ifdef CONFIG_XEN
2327 #include <xen/interface/kexec.h>
2328 @@ -80,13 +79,14 @@ static struct notifier_block xen_panic_b
2329 xen_panic_event, NULL, 0 /* try to go last */
2330 };
2331
2332 -int disable_pse __devinitdata = 0;
2333 +int disable_pse __cpuinitdata = 0;
2334
2335 /*
2336 * Machine setup..
2337 */
2338 extern struct resource code_resource;
2339 extern struct resource data_resource;
2340 +extern struct resource bss_resource;
2341
2342 /* cpu data as detected by the assembly code in head.S */
2343 struct cpuinfo_x86 new_cpu_data __cpuinitdata = { 0, 0, 0, 0, -1, 1, 0, 0, -1 };
2344 @@ -98,9 +98,6 @@ unsigned long mmu_cr4_features;
2345
2346 /* for MCA, but anyone else can use it if they want */
2347 unsigned int machine_id;
2348 -#ifdef CONFIG_MCA
2349 -EXPORT_SYMBOL(machine_id);
2350 -#endif
2351 unsigned int machine_submodel_id;
2352 unsigned int BIOS_revision;
2353 unsigned int mca_pentium_flag;
2354 @@ -121,7 +118,7 @@ EXPORT_SYMBOL(apm_info);
2355 struct edid_info edid_info;
2356 EXPORT_SYMBOL_GPL(edid_info);
2357 #ifndef CONFIG_XEN
2358 -#define copy_edid() (edid_info = EDID_INFO)
2359 +#define copy_edid() (edid_info = boot_params.edid_info)
2360 #endif
2361 struct ist_info ist_info;
2362 #if defined(CONFIG_X86_SPEEDSTEP_SMI) || \
2363 @@ -170,10 +167,11 @@ EXPORT_SYMBOL(edd);
2364 */
2365 static inline void copy_edd(void)
2366 {
2367 - memcpy(edd.mbr_signature, EDD_MBR_SIGNATURE, sizeof(edd.mbr_signature));
2368 - memcpy(edd.edd_info, EDD_BUF, sizeof(edd.edd_info));
2369 - edd.mbr_signature_nr = EDD_MBR_SIG_NR;
2370 - edd.edd_info_nr = EDD_NR;
2371 + memcpy(edd.mbr_signature, boot_params.edd_mbr_sig_buffer,
2372 + sizeof(edd.mbr_signature));
2373 + memcpy(edd.edd_info, boot_params.eddbuf, sizeof(edd.edd_info));
2374 + edd.mbr_signature_nr = boot_params.edd_mbr_sig_buf_entries;
2375 + edd.edd_info_nr = boot_params.eddbuf_entries;
2376 }
2377 #endif
2378 #else
2379 @@ -416,6 +414,53 @@ extern unsigned long __init setup_memory
2380 extern void zone_sizes_init(void);
2381 #endif /* !CONFIG_NEED_MULTIPLE_NODES */
2382
2383 +static inline unsigned long long get_total_mem(void)
2384 +{
2385 + unsigned long long total;
2386 +
2387 + total = max_low_pfn - min_low_pfn;
2388 +#ifdef CONFIG_HIGHMEM
2389 + total += highend_pfn - highstart_pfn;
2390 +#endif
2391 +
2392 + return total << PAGE_SHIFT;
2393 +}
2394 +
2395 +#ifdef CONFIG_KEXEC
2396 +#ifndef CONFIG_XEN
2397 +static void __init reserve_crashkernel(void)
2398 +{
2399 + unsigned long long total_mem;
2400 + unsigned long long crash_size, crash_base;
2401 + int ret;
2402 +
2403 + total_mem = get_total_mem();
2404 +
2405 + ret = parse_crashkernel(boot_command_line, total_mem,
2406 + &crash_size, &crash_base);
2407 + if (ret == 0 && crash_size > 0) {
2408 + if (crash_base > 0) {
2409 + printk(KERN_INFO "Reserving %ldMB of memory at %ldMB "
2410 + "for crashkernel (System RAM: %ldMB)\n",
2411 + (unsigned long)(crash_size >> 20),
2412 + (unsigned long)(crash_base >> 20),
2413 + (unsigned long)(total_mem >> 20));
2414 + crashk_res.start = crash_base;
2415 + crashk_res.end = crash_base + crash_size - 1;
2416 + reserve_bootmem(crash_base, crash_size);
2417 + } else
2418 + printk(KERN_INFO "crashkernel reservation failed - "
2419 + "you have to specify a base address\n");
2420 + }
2421 +}
2422 +#else
2423 +#define reserve_crashkernel xen_machine_kexec_setup_resources
2424 +#endif
2425 +#else
2426 +static inline void __init reserve_crashkernel(void)
2427 +{}
2428 +#endif
2429 +
2430 void __init setup_bootmem_allocator(void)
2431 {
2432 unsigned long bootmap_size;
2433 @@ -471,30 +516,25 @@ void __init setup_bootmem_allocator(void
2434
2435 #ifdef CONFIG_BLK_DEV_INITRD
2436 if (xen_start_info->mod_start) {
2437 - if (INITRD_START + INITRD_SIZE <= (max_low_pfn << PAGE_SHIFT)) {
2438 - /*reserve_bootmem(INITRD_START, INITRD_SIZE);*/
2439 - initrd_start = INITRD_START + PAGE_OFFSET;
2440 - initrd_end = initrd_start+INITRD_SIZE;
2441 + unsigned long ramdisk_image = __pa(xen_start_info->mod_start);
2442 + unsigned long ramdisk_size = xen_start_info->mod_len;
2443 + unsigned long ramdisk_end = ramdisk_image + ramdisk_size;
2444 + unsigned long end_of_lowmem = max_low_pfn << PAGE_SHIFT;
2445 +
2446 + if (ramdisk_end <= end_of_lowmem) {
2447 + /*reserve_bootmem(ramdisk_image, ramdisk_size);*/
2448 + initrd_start = ramdisk_image + PAGE_OFFSET;
2449 + initrd_end = initrd_start+ramdisk_size;
2450 initrd_below_start_ok = 1;
2451 - }
2452 - else {
2453 + } else {
2454 printk(KERN_ERR "initrd extends beyond end of memory "
2455 - "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2456 - INITRD_START + INITRD_SIZE,
2457 - max_low_pfn << PAGE_SHIFT);
2458 + "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2459 + ramdisk_end, end_of_lowmem);
2460 initrd_start = 0;
2461 }
2462 }
2463 #endif
2464 -#ifdef CONFIG_KEXEC
2465 -#ifdef CONFIG_XEN
2466 - xen_machine_kexec_setup_resources();
2467 -#else
2468 - if (crashk_res.start != crashk_res.end)
2469 - reserve_bootmem(crashk_res.start,
2470 - crashk_res.end - crashk_res.start + 1);
2471 -#endif
2472 -#endif
2473 + reserve_crashkernel();
2474 }
2475
2476 /*
2477 @@ -572,7 +612,8 @@ void __init setup_arch(char **cmdline_p)
2478 * the system table is valid. If not, then initialize normally.
2479 */
2480 #ifdef CONFIG_EFI
2481 - if ((LOADER_TYPE == 0x50) && EFI_SYSTAB)
2482 + if ((boot_params.hdr.type_of_loader == 0x50) &&
2483 + boot_params.efi_info.efi_systab)
2484 efi_enabled = 1;
2485 #endif
2486
2487 @@ -580,18 +621,18 @@ void __init setup_arch(char **cmdline_p)
2488 properly. Setting ROOT_DEV to default to /dev/ram0 breaks initrd.
2489 */
2490 ROOT_DEV = MKDEV(UNNAMED_MAJOR,0);
2491 - screen_info = SCREEN_INFO;
2492 + screen_info = boot_params.screen_info;
2493 copy_edid();
2494 - apm_info.bios = APM_BIOS_INFO;
2495 - ist_info = IST_INFO;
2496 - saved_videomode = VIDEO_MODE;
2497 - if( SYS_DESC_TABLE.length != 0 ) {
2498 - set_mca_bus(SYS_DESC_TABLE.table[3] & 0x2);
2499 - machine_id = SYS_DESC_TABLE.table[0];
2500 - machine_submodel_id = SYS_DESC_TABLE.table[1];
2501 - BIOS_revision = SYS_DESC_TABLE.table[2];
2502 + apm_info.bios = boot_params.apm_bios_info;
2503 + ist_info = boot_params.ist_info;
2504 + saved_videomode = boot_params.hdr.vid_mode;
2505 + if( boot_params.sys_desc_table.length != 0 ) {
2506 + set_mca_bus(boot_params.sys_desc_table.table[3] & 0x2);
2507 + machine_id = boot_params.sys_desc_table.table[0];
2508 + machine_submodel_id = boot_params.sys_desc_table.table[1];
2509 + BIOS_revision = boot_params.sys_desc_table.table[2];
2510 }
2511 - bootloader_type = LOADER_TYPE;
2512 + bootloader_type = boot_params.hdr.type_of_loader;
2513
2514 if (is_initial_xendomain()) {
2515 const struct dom0_vga_console_info *info =
2516 @@ -606,9 +647,9 @@ void __init setup_arch(char **cmdline_p)
2517 screen_info.orig_video_isVGA = 0;
2518
2519 #ifdef CONFIG_BLK_DEV_RAM
2520 - rd_image_start = RAMDISK_FLAGS & RAMDISK_IMAGE_START_MASK;
2521 - rd_prompt = ((RAMDISK_FLAGS & RAMDISK_PROMPT_FLAG) != 0);
2522 - rd_doload = ((RAMDISK_FLAGS & RAMDISK_LOAD_FLAG) != 0);
2523 + rd_image_start = boot_params.hdr.ram_size & RAMDISK_IMAGE_START_MASK;
2524 + rd_prompt = ((boot_params.hdr.ram_size & RAMDISK_PROMPT_FLAG) != 0);
2525 + rd_doload = ((boot_params.hdr.ram_size & RAMDISK_LOAD_FLAG) != 0);
2526 #endif
2527
2528 ARCH_SETUP
2529 @@ -621,7 +662,7 @@ void __init setup_arch(char **cmdline_p)
2530
2531 copy_edd();
2532
2533 - if (!MOUNT_ROOT_RDONLY)
2534 + if (!boot_params.hdr.root_flags)
2535 root_mountflags &= ~MS_RDONLY;
2536 init_mm.start_code = (unsigned long) _text;
2537 init_mm.end_code = (unsigned long) _etext;
2538 @@ -633,6 +674,8 @@ void __init setup_arch(char **cmdline_p)
2539 code_resource.end = virt_to_phys(_etext)-1;
2540 data_resource.start = virt_to_phys(_etext);
2541 data_resource.end = virt_to_phys(_edata)-1;
2542 + bss_resource.start = virt_to_phys(&__bss_start);
2543 + bss_resource.end = virt_to_phys(&__bss_stop)-1;
2544
2545 if ((i = MAX_GUEST_CMDLINE) > COMMAND_LINE_SIZE)
2546 i = COMMAND_LINE_SIZE;
2547 @@ -661,7 +704,7 @@ void __init setup_arch(char **cmdline_p)
2548 /*
2549 * NOTE: before this point _nobody_ is allowed to allocate
2550 * any memory using the bootmem allocator. Although the
2551 - * alloctor is now initialised only the first 8Mb of the kernel
2552 + * allocator is now initialised only the first 8Mb of the kernel
2553 * virtual address space has been mapped. All allocations before
2554 * paging_init() has completed must use the alloc_bootmem_low_pages()
2555 * variant (which allocates DMA'able memory) and care must be taken
2556 @@ -784,10 +827,8 @@ void __init setup_arch(char **cmdline_p)
2557 acpi_boot_table_init();
2558 #endif
2559
2560 -#ifdef CONFIG_PCI
2561 -#ifdef CONFIG_X86_IO_APIC
2562 - check_acpi_pci(); /* Checks more than just ACPI actually */
2563 -#endif
2564 +#if defined(CONFIG_PCI) && !defined(CONFIG_XEN)
2565 + early_quirks();
2566 #endif
2567
2568 #ifdef CONFIG_ACPI
2569 --- sle11-2009-05-14.orig/arch/x86/kernel/setup_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2570 +++ sle11-2009-05-14/arch/x86/kernel/setup_64-xen.c 2009-02-16 16:18:36.000000000 +0100
2571 @@ -1,10 +1,5 @@
2572 /*
2573 - * linux/arch/x86-64/kernel/setup.c
2574 - *
2575 * Copyright (C) 1995 Linus Torvalds
2576 - *
2577 - * Nov 2001 Dave Jones <davej@suse.de>
2578 - * Forked from i386 setup code.
2579 */
2580
2581 /*
2582 @@ -57,13 +52,13 @@
2583 #include <asm/dma.h>
2584 #include <asm/mpspec.h>
2585 #include <asm/mmu_context.h>
2586 -#include <asm/bootsetup.h>
2587 #include <asm/proto.h>
2588 #include <asm/setup.h>
2589 #include <asm/mach_apic.h>
2590 #include <asm/numa.h>
2591 #include <asm/sections.h>
2592 #include <asm/dmi.h>
2593 +#include <asm/cacheflush.h>
2594 #ifdef CONFIG_XEN
2595 #include <linux/percpu.h>
2596 #include <xen/interface/physdev.h>
2597 @@ -180,6 +175,12 @@ struct resource code_resource = {
2598 .end = 0,
2599 .flags = IORESOURCE_RAM,
2600 };
2601 +struct resource bss_resource = {
2602 + .name = "Kernel bss",
2603 + .start = 0,
2604 + .end = 0,
2605 + .flags = IORESOURCE_RAM,
2606 +};
2607
2608 #ifdef CONFIG_PROC_VMCORE
2609 /* elfcorehdr= specifies the location of elf core header
2610 @@ -231,10 +232,11 @@ EXPORT_SYMBOL(edd);
2611 */
2612 static inline void copy_edd(void)
2613 {
2614 - memcpy(edd.mbr_signature, EDD_MBR_SIGNATURE, sizeof(edd.mbr_signature));
2615 - memcpy(edd.edd_info, EDD_BUF, sizeof(edd.edd_info));
2616 - edd.mbr_signature_nr = EDD_MBR_SIG_NR;
2617 - edd.edd_info_nr = EDD_NR;
2618 + memcpy(edd.mbr_signature, boot_params.edd_mbr_sig_buffer,
2619 + sizeof(edd.mbr_signature));
2620 + memcpy(edd.edd_info, boot_params.eddbuf, sizeof(edd.edd_info));
2621 + edd.mbr_signature_nr = boot_params.edd_mbr_sig_buf_entries;
2622 + edd.edd_info_nr = boot_params.eddbuf_entries;
2623 }
2624 #endif
2625 #else
2626 @@ -243,6 +245,41 @@ static inline void copy_edd(void)
2627 }
2628 #endif
2629
2630 +#ifdef CONFIG_KEXEC
2631 +#ifndef CONFIG_XEN
2632 +static void __init reserve_crashkernel(void)
2633 +{
2634 + unsigned long long free_mem;
2635 + unsigned long long crash_size, crash_base;
2636 + int ret;
2637 +
2638 + free_mem = ((unsigned long long)max_low_pfn - min_low_pfn) << PAGE_SHIFT;
2639 +
2640 + ret = parse_crashkernel(boot_command_line, free_mem,
2641 + &crash_size, &crash_base);
2642 + if (ret == 0 && crash_size) {
2643 + if (crash_base > 0) {
2644 + printk(KERN_INFO "Reserving %ldMB of memory at %ldMB "
2645 + "for crashkernel (System RAM: %ldMB)\n",
2646 + (unsigned long)(crash_size >> 20),
2647 + (unsigned long)(crash_base >> 20),
2648 + (unsigned long)(free_mem >> 20));
2649 + crashk_res.start = crash_base;
2650 + crashk_res.end = crash_base + crash_size - 1;
2651 + reserve_bootmem(crash_base, crash_size);
2652 + } else
2653 + printk(KERN_INFO "crashkernel reservation failed - "
2654 + "you have to specify a base address\n");
2655 + }
2656 +}
2657 +#else
2658 +#define reserve_crashkernel xen_machine_kexec_setup_resources
2659 +#endif
2660 +#else
2661 +static inline void __init reserve_crashkernel(void)
2662 +{}
2663 +#endif
2664 +
2665 #ifndef CONFIG_XEN
2666 #define EBDA_ADDR_POINTER 0x40E
2667
2668 @@ -283,7 +320,7 @@ void __init setup_arch(char **cmdline_p)
2669 atomic_notifier_chain_register(&panic_notifier_list, &xen_panic_block);
2670
2671 ROOT_DEV = MKDEV(RAMDISK_MAJOR,0);
2672 - screen_info = SCREEN_INFO;
2673 + screen_info = boot_params.screen_info;
2674
2675 if (is_initial_xendomain()) {
2676 const struct dom0_vga_console_info *info =
2677 @@ -306,22 +343,22 @@ void __init setup_arch(char **cmdline_p)
2678 #else
2679 printk(KERN_INFO "Command line: %s\n", boot_command_line);
2680
2681 - ROOT_DEV = old_decode_dev(ORIG_ROOT_DEV);
2682 - screen_info = SCREEN_INFO;
2683 - edid_info = EDID_INFO;
2684 + ROOT_DEV = old_decode_dev(boot_params.hdr.root_dev);
2685 + screen_info = boot_params.screen_info;
2686 + edid_info = boot_params.edid_info;
2687 #endif /* !CONFIG_XEN */
2688 - saved_video_mode = SAVED_VIDEO_MODE;
2689 - bootloader_type = LOADER_TYPE;
2690 + saved_video_mode = boot_params.hdr.vid_mode;
2691 + bootloader_type = boot_params.hdr.type_of_loader;
2692
2693 #ifdef CONFIG_BLK_DEV_RAM
2694 - rd_image_start = RAMDISK_FLAGS & RAMDISK_IMAGE_START_MASK;
2695 - rd_prompt = ((RAMDISK_FLAGS & RAMDISK_PROMPT_FLAG) != 0);
2696 - rd_doload = ((RAMDISK_FLAGS & RAMDISK_LOAD_FLAG) != 0);
2697 + rd_image_start = boot_params.hdr.ram_size & RAMDISK_IMAGE_START_MASK;
2698 + rd_prompt = ((boot_params.hdr.ram_size & RAMDISK_PROMPT_FLAG) != 0);
2699 + rd_doload = ((boot_params.hdr.ram_size & RAMDISK_LOAD_FLAG) != 0);
2700 #endif
2701 setup_memory_region();
2702 copy_edd();
2703
2704 - if (!MOUNT_ROOT_RDONLY)
2705 + if (!boot_params.hdr.root_flags)
2706 root_mountflags &= ~MS_RDONLY;
2707 init_mm.start_code = (unsigned long) &_text;
2708 init_mm.end_code = (unsigned long) &_etext;
2709 @@ -332,6 +369,8 @@ void __init setup_arch(char **cmdline_p)
2710 code_resource.end = virt_to_phys(&_etext)-1;
2711 data_resource.start = virt_to_phys(&_etext);
2712 data_resource.end = virt_to_phys(&_edata)-1;
2713 + bss_resource.start = virt_to_phys(&__bss_start);
2714 + bss_resource.end = virt_to_phys(&__bss_stop)-1;
2715
2716 early_identify_cpu(&boot_cpu_data);
2717
2718 @@ -359,6 +398,11 @@ void __init setup_arch(char **cmdline_p)
2719 if (is_initial_xendomain())
2720 dmi_scan_machine();
2721
2722 +#if defined(CONFIG_SMP) && !defined(CONFIG_XEN)
2723 + /* setup to use the static apicid table during kernel startup */
2724 + x86_cpu_to_apicid_ptr = (void *)&x86_cpu_to_apicid_init;
2725 +#endif
2726 +
2727 /* How many end-of-memory variables you have, grandma! */
2728 max_low_pfn = end_pfn;
2729 max_pfn = end_pfn;
2730 @@ -423,52 +467,37 @@ void __init setup_arch(char **cmdline_p)
2731 */
2732 acpi_reserve_bootmem();
2733 #endif
2734 -#ifdef CONFIG_XEN
2735 #ifdef CONFIG_BLK_DEV_INITRD
2736 +#ifdef CONFIG_XEN
2737 if (xen_start_info->mod_start) {
2738 - if (INITRD_START + INITRD_SIZE <= (end_pfn << PAGE_SHIFT)) {
2739 - /*reserve_bootmem_generic(INITRD_START, INITRD_SIZE);*/
2740 - initrd_start = INITRD_START + PAGE_OFFSET;
2741 - initrd_end = initrd_start+INITRD_SIZE;
2742 + unsigned long ramdisk_image = __pa(xen_start_info->mod_start);
2743 + unsigned long ramdisk_size = xen_start_info->mod_len;
2744 +#else
2745 + if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) {
2746 + unsigned long ramdisk_image = boot_params.hdr.ramdisk_image;
2747 + unsigned long ramdisk_size = boot_params.hdr.ramdisk_size;
2748 +#endif
2749 + unsigned long ramdisk_end = ramdisk_image + ramdisk_size;
2750 + unsigned long end_of_mem = end_pfn << PAGE_SHIFT;
2751 +
2752 + if (ramdisk_end <= end_of_mem) {
2753 +#ifndef CONFIG_XEN
2754 + reserve_bootmem_generic(ramdisk_image, ramdisk_size);
2755 +#endif
2756 + initrd_start = ramdisk_image + PAGE_OFFSET;
2757 + initrd_end = initrd_start+ramdisk_size;
2758 +#ifdef CONFIG_XEN
2759 initrd_below_start_ok = 1;
2760 - } else {
2761 - printk(KERN_ERR "initrd extends beyond end of memory "
2762 - "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2763 - (unsigned long)(INITRD_START + INITRD_SIZE),
2764 - (unsigned long)(end_pfn << PAGE_SHIFT));
2765 - initrd_start = 0;
2766 - }
2767 - }
2768 #endif
2769 -#else /* CONFIG_XEN */
2770 -#ifdef CONFIG_BLK_DEV_INITRD
2771 - if (LOADER_TYPE && INITRD_START) {
2772 - if (INITRD_START + INITRD_SIZE <= (end_pfn << PAGE_SHIFT)) {
2773 - reserve_bootmem_generic(INITRD_START, INITRD_SIZE);
2774 - initrd_start = INITRD_START + PAGE_OFFSET;
2775 - initrd_end = initrd_start+INITRD_SIZE;
2776 - }
2777 - else {
2778 + } else {
2779 printk(KERN_ERR "initrd extends beyond end of memory "
2780 - "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2781 - (unsigned long)(INITRD_START + INITRD_SIZE),
2782 - (unsigned long)(end_pfn << PAGE_SHIFT));
2783 + "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2784 + ramdisk_end, end_of_mem);
2785 initrd_start = 0;
2786 }
2787 }
2788 #endif
2789 -#endif /* !CONFIG_XEN */
2790 -#ifdef CONFIG_KEXEC
2791 -#ifdef CONFIG_XEN
2792 - xen_machine_kexec_setup_resources();
2793 -#else
2794 - if (crashk_res.start != crashk_res.end) {
2795 - reserve_bootmem_generic(crashk_res.start,
2796 - crashk_res.end - crashk_res.start + 1);
2797 - }
2798 -#endif
2799 -#endif
2800 -
2801 + reserve_crashkernel();
2802 paging_init();
2803 #ifdef CONFIG_X86_LOCAL_APIC
2804 /*
2805 @@ -783,7 +812,7 @@ static void __init amd_detect_cmp(struct
2806 but in the same order as the HT nodeids.
2807 If that doesn't result in a usable node fall back to the
2808 path for the previous case. */
2809 - int ht_nodeid = apicid - (cpu_data[0].phys_proc_id << bits);
2810 + int ht_nodeid = apicid - (cpu_data(0).phys_proc_id << bits);
2811 if (ht_nodeid >= 0 &&
2812 apicid_to_node[ht_nodeid] != NUMA_NO_NODE)
2813 node = apicid_to_node[ht_nodeid];
2814 @@ -798,6 +827,39 @@ static void __init amd_detect_cmp(struct
2815 #endif
2816 }
2817
2818 +#define ENABLE_C1E_MASK 0x18000000
2819 +#define CPUID_PROCESSOR_SIGNATURE 1
2820 +#define CPUID_XFAM 0x0ff00000
2821 +#define CPUID_XFAM_K8 0x00000000
2822 +#define CPUID_XFAM_10H 0x00100000
2823 +#define CPUID_XFAM_11H 0x00200000
2824 +#define CPUID_XMOD 0x000f0000
2825 +#define CPUID_XMOD_REV_F 0x00040000
2826 +
2827 +#ifndef CONFIG_XEN
2828 +/* AMD systems with C1E don't have a working lAPIC timer. Check for that. */
2829 +static __cpuinit int amd_apic_timer_broken(void)
2830 +{
2831 + u32 lo, hi;
2832 + u32 eax = cpuid_eax(CPUID_PROCESSOR_SIGNATURE);
2833 + switch (eax & CPUID_XFAM) {
2834 + case CPUID_XFAM_K8:
2835 + if ((eax & CPUID_XMOD) < CPUID_XMOD_REV_F)
2836 + break;
2837 + case CPUID_XFAM_10H:
2838 + case CPUID_XFAM_11H:
2839 + rdmsr(MSR_K8_ENABLE_C1E, lo, hi);
2840 + if (lo & ENABLE_C1E_MASK)
2841 + return 1;
2842 + break;
2843 + default:
2844 + /* err on the side of caution */
2845 + return 1;
2846 + }
2847 + return 0;
2848 +}
2849 +#endif
2850 +
2851 static void __cpuinit init_amd(struct cpuinfo_x86 *c)
2852 {
2853 unsigned level;
2854 @@ -827,7 +889,7 @@ static void __cpuinit init_amd(struct cp
2855 level = cpuid_eax(1);
2856 if (c->x86 == 15 && ((level >= 0x0f48 && level < 0x0f50) || level >= 0x0f58))
2857 set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability);
2858 - if (c->x86 == 0x10)
2859 + if (c->x86 == 0x10 || c->x86 == 0x11)
2860 set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability);
2861
2862 /* Enable workaround for FXSAVE leak */
2863 @@ -869,6 +931,11 @@ static void __cpuinit init_amd(struct cp
2864 /* Family 10 doesn't support C states in MWAIT so don't use it */
2865 if (c->x86 == 0x10 && !force_mwait)
2866 clear_bit(X86_FEATURE_MWAIT, &c->x86_capability);
2867 +
2868 +#ifndef CONFIG_XEN
2869 + if (amd_apic_timer_broken())
2870 + disable_apic_timer = 1;
2871 +#endif
2872 }
2873
2874 static void __cpuinit detect_ht(struct cpuinfo_x86 *c)
2875 @@ -1179,6 +1246,7 @@ void __cpuinit print_cpu_info(struct cpu
2876 static int show_cpuinfo(struct seq_file *m, void *v)
2877 {
2878 struct cpuinfo_x86 *c = v;
2879 + int cpu = 0;
2880
2881 /*
2882 * These flag bits must match the definitions in <asm/cpufeature.h>.
2883 @@ -1188,7 +1256,7 @@ static int show_cpuinfo(struct seq_file
2884 * applications want to get the raw CPUID data, they should access
2885 * /dev/cpu/<cpu_nr>/cpuid instead.
2886 */
2887 - static char *x86_cap_flags[] = {
2888 + static const char *const x86_cap_flags[] = {
2889 /* Intel-defined */
2890 "fpu", "vme", "de", "pse", "tsc", "msr", "pae", "mce",
2891 "cx8", "apic", NULL, "sep", "mtrr", "pge", "mca", "cmov",
2892 @@ -1219,7 +1287,7 @@ static int show_cpuinfo(struct seq_file
2893 /* Intel-defined (#2) */
2894 "pni", NULL, NULL, "monitor", "ds_cpl", "vmx", "smx", "est",
2895 "tm2", "ssse3", "cid", NULL, NULL, "cx16", "xtpr", NULL,
2896 - NULL, NULL, "dca", NULL, NULL, NULL, NULL, "popcnt",
2897 + NULL, NULL, "dca", "sse4_1", "sse4_2", NULL, NULL, "popcnt",
2898 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2899
2900 /* VIA/Cyrix/Centaur-defined */
2901 @@ -1229,10 +1297,10 @@ static int show_cpuinfo(struct seq_file
2902 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2903
2904 /* AMD-defined (#2) */
2905 - "lahf_lm", "cmp_legacy", "svm", "extapic", "cr8_legacy",
2906 - "altmovcr8", "abm", "sse4a",
2907 - "misalignsse", "3dnowprefetch",
2908 - "osvw", "ibs", NULL, NULL, NULL, NULL,
2909 + "lahf_lm", "cmp_legacy", "svm", "extapic",
2910 + "cr8_legacy", "abm", "sse4a", "misalignsse",
2911 + "3dnowprefetch", "osvw", "ibs", "sse5",
2912 + "skinit", "wdt", NULL, NULL,
2913 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2914 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2915
2916 @@ -1242,7 +1310,7 @@ static int show_cpuinfo(struct seq_file
2917 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2918 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2919 };
2920 - static char *x86_power_flags[] = {
2921 + static const char *const x86_power_flags[] = {
2922 "ts", /* temperature sensor */
2923 "fid", /* frequency id control */
2924 "vid", /* voltage id control */
2925 @@ -1257,8 +1325,7 @@ static int show_cpuinfo(struct seq_file
2926
2927
2928 #ifdef CONFIG_SMP
2929 - if (!cpu_online(c-cpu_data))
2930 - return 0;
2931 + cpu = c->cpu_index;
2932 #endif
2933
2934 seq_printf(m,"processor\t: %u\n"
2935 @@ -1266,7 +1333,7 @@ static int show_cpuinfo(struct seq_file
2936 "cpu family\t: %d\n"
2937 "model\t\t: %d\n"
2938 "model name\t: %s\n",
2939 - (unsigned)(c-cpu_data),
2940 + (unsigned)cpu,
2941 c->x86_vendor_id[0] ? c->x86_vendor_id : "unknown",
2942 c->x86,
2943 (int)c->x86_model,
2944 @@ -1278,7 +1345,7 @@ static int show_cpuinfo(struct seq_file
2945 seq_printf(m, "stepping\t: unknown\n");
2946
2947 if (cpu_has(c,X86_FEATURE_TSC)) {
2948 - unsigned int freq = cpufreq_quick_get((unsigned)(c-cpu_data));
2949 + unsigned int freq = cpufreq_quick_get((unsigned)cpu);
2950 if (!freq)
2951 freq = cpu_khz;
2952 seq_printf(m, "cpu MHz\t\t: %u.%03u\n",
2953 @@ -1291,9 +1358,9 @@ static int show_cpuinfo(struct seq_file
2954
2955 #ifdef CONFIG_SMP
2956 if (smp_num_siblings * c->x86_max_cores > 1) {
2957 - int cpu = c - cpu_data;
2958 seq_printf(m, "physical id\t: %d\n", c->phys_proc_id);
2959 - seq_printf(m, "siblings\t: %d\n", cpus_weight(cpu_core_map[cpu]));
2960 + seq_printf(m, "siblings\t: %d\n",
2961 + cpus_weight(per_cpu(cpu_core_map, cpu)));
2962 seq_printf(m, "core id\t\t: %d\n", c->cpu_core_id);
2963 seq_printf(m, "cpu cores\t: %d\n", c->booted_cores);
2964 }
2965 @@ -1348,12 +1415,16 @@ static int show_cpuinfo(struct seq_file
2966
2967 static void *c_start(struct seq_file *m, loff_t *pos)
2968 {
2969 - return *pos < NR_CPUS ? cpu_data + *pos : NULL;
2970 + if (*pos == 0) /* just in case, cpu 0 is not the first */
2971 + *pos = first_cpu(cpu_online_map);
2972 + if ((*pos) < NR_CPUS && cpu_online(*pos))
2973 + return &cpu_data(*pos);
2974 + return NULL;
2975 }
2976
2977 static void *c_next(struct seq_file *m, void *v, loff_t *pos)
2978 {
2979 - ++*pos;
2980 + *pos = next_cpu(*pos, cpu_online_map);
2981 return c_start(m, pos);
2982 }
2983
2984 --- sle11-2009-05-14.orig/arch/x86/kernel/smp_32-xen.c 2009-02-16 16:17:21.000000000 +0100
2985 +++ sle11-2009-05-14/arch/x86/kernel/smp_32-xen.c 2009-02-16 16:18:36.000000000 +0100
2986 @@ -72,7 +72,7 @@
2987 *
2988 * B stepping CPUs may hang. There are hardware work arounds
2989 * for this. We warn about it in case your board doesn't have the work
2990 - * arounds. Basically thats so I can tell anyone with a B stepping
2991 + * arounds. Basically that's so I can tell anyone with a B stepping
2992 * CPU and SMP problems "tough".
2993 *
2994 * Specific items [From Pentium Processor Specification Update]
2995 @@ -241,7 +241,7 @@ void leave_mm(unsigned long cpu)
2996 * 1a1) cpu_clear(cpu, old_mm->cpu_vm_mask);
2997 * Stop ipi delivery for the old mm. This is not synchronized with
2998 * the other cpus, but smp_invalidate_interrupt ignore flush ipis
2999 - * for the wrong mm, and in the worst case we perform a superflous
3000 + * for the wrong mm, and in the worst case we perform a superfluous
3001 * tlb flush.
3002 * 1a2) set cpu_tlbstate to TLBSTATE_OK
3003 * Now the smp_invalidate_interrupt won't call leave_mm if cpu0
3004 @@ -309,6 +309,7 @@ irqreturn_t smp_invalidate_interrupt(int
3005 smp_mb__after_clear_bit();
3006 out:
3007 put_cpu_no_resched();
3008 + __get_cpu_var(irq_stat).irq_tlb_count++;
3009
3010 return IRQ_HANDLED;
3011 }
3012 @@ -580,7 +581,7 @@ static void stop_this_cpu (void * dummy)
3013 */
3014 cpu_clear(smp_processor_id(), cpu_online_map);
3015 disable_all_local_evtchn();
3016 - if (cpu_data[smp_processor_id()].hlt_works_ok)
3017 + if (cpu_data(smp_processor_id()).hlt_works_ok)
3018 for(;;) halt();
3019 for (;;);
3020 }
3021 @@ -610,6 +611,7 @@ void xen_smp_send_stop(void)
3022 */
3023 irqreturn_t smp_reschedule_interrupt(int irq, void *dev_id)
3024 {
3025 + __get_cpu_var(irq_stat).irq_resched_count++;
3026
3027 return IRQ_HANDLED;
3028 }
3029 @@ -632,6 +634,7 @@ irqreturn_t smp_call_function_interrupt(
3030 */
3031 irq_enter();
3032 (*func)(info);
3033 + __get_cpu_var(irq_stat).irq_call_count++;
3034 irq_exit();
3035
3036 if (wait) {
3037 --- sle11-2009-05-14.orig/arch/x86/kernel/smp_64-xen.c 2009-02-16 16:17:21.000000000 +0100
3038 +++ sle11-2009-05-14/arch/x86/kernel/smp_64-xen.c 2009-02-16 16:18:36.000000000 +0100
3039 @@ -167,6 +167,7 @@ asmlinkage void smp_invalidate_interrupt
3040 out:
3041 ack_APIC_irq();
3042 cpu_clear(cpu, f->flush_cpumask);
3043 + add_pda(irq_tlb_count, 1);
3044 }
3045
3046 static void flush_tlb_others(cpumask_t cpumask, struct mm_struct *mm,
3047 @@ -326,17 +327,27 @@ void unlock_ipi_call_lock(void)
3048 }
3049
3050 /*
3051 - * this function sends a 'generic call function' IPI to one other CPU
3052 - * in the system.
3053 - *
3054 - * cpu is a standard Linux logical CPU number.
3055 + * this function sends a 'generic call function' IPI to all other CPU
3056 + * of the system defined in the mask.
3057 */
3058 -static void
3059 -__smp_call_function_single(int cpu, void (*func) (void *info), void *info,
3060 - int nonatomic, int wait)
3061 +
3062 +static int
3063 +__smp_call_function_mask(cpumask_t mask,
3064 + void (*func)(void *), void *info,
3065 + int wait)
3066 {
3067 struct call_data_struct data;
3068 - int cpus = 1;
3069 + cpumask_t allbutself;
3070 + int cpus;
3071 +
3072 + allbutself = cpu_online_map;
3073 + cpu_clear(smp_processor_id(), allbutself);
3074 +
3075 + cpus_and(mask, mask, allbutself);
3076 + cpus = cpus_weight(mask);
3077 +
3078 + if (!cpus)
3079 + return 0;
3080
3081 data.func = func;
3082 data.info = info;
3083 @@ -347,19 +358,55 @@ __smp_call_function_single(int cpu, void
3084
3085 call_data = &data;
3086 wmb();
3087 - /* Send a message to all other CPUs and wait for them to respond */
3088 - send_IPI_mask(cpumask_of_cpu(cpu), CALL_FUNCTION_VECTOR);
3089 +
3090 + /* Send a message to other CPUs */
3091 + if (cpus_equal(mask, allbutself))
3092 + send_IPI_allbutself(CALL_FUNCTION_VECTOR);
3093 + else
3094 + send_IPI_mask(mask, CALL_FUNCTION_VECTOR);
3095
3096 /* Wait for response */
3097 while (atomic_read(&data.started) != cpus)
3098 cpu_relax();
3099
3100 if (!wait)
3101 - return;
3102 + return 0;
3103
3104 while (atomic_read(&data.finished) != cpus)
3105 cpu_relax();
3106 +
3107 + return 0;
3108 +}
3109 +/**
3110 + * smp_call_function_mask(): Run a function on a set of other CPUs.
3111 + * @mask: The set of cpus to run on. Must not include the current cpu.
3112 + * @func: The function to run. This must be fast and non-blocking.
3113 + * @info: An arbitrary pointer to pass to the function.
3114 + * @wait: If true, wait (atomically) until function has completed on other CPUs.
3115 + *
3116 + * Returns 0 on success, else a negative status code.
3117 + *
3118 + * If @wait is true, then returns once @func has returned; otherwise
3119 + * it returns just before the target cpu calls @func.
3120 + *
3121 + * You must not call this function with disabled interrupts or from a
3122 + * hardware interrupt handler or from a bottom half handler.
3123 + */
3124 +int smp_call_function_mask(cpumask_t mask,
3125 + void (*func)(void *), void *info,
3126 + int wait)
3127 +{
3128 + int ret;
3129 +
3130 + /* Can deadlock when called with interrupts disabled */
3131 + WARN_ON(irqs_disabled());
3132 +
3133 + spin_lock(&call_lock);
3134 + ret = __smp_call_function_mask(mask, func, info, wait);
3135 + spin_unlock(&call_lock);
3136 + return ret;
3137 }
3138 +EXPORT_SYMBOL(smp_call_function_mask);
3139
3140 /*
3141 * smp_call_function_single - Run a function on a specific CPU
3142 @@ -378,6 +425,7 @@ int smp_call_function_single (int cpu, v
3143 int nonatomic, int wait)
3144 {
3145 /* prevent preemption and reschedule on another processor */
3146 + int ret;
3147 int me = get_cpu();
3148
3149 /* Can deadlock when called with interrupts disabled */
3150 @@ -391,51 +439,14 @@ int smp_call_function_single (int cpu, v
3151 return 0;
3152 }
3153
3154 - spin_lock(&call_lock);
3155 - __smp_call_function_single(cpu, func, info, nonatomic, wait);
3156 - spin_unlock(&call_lock);
3157 + ret = smp_call_function_mask(cpumask_of_cpu(cpu), func, info, wait);
3158 +
3159 put_cpu();
3160 - return 0;
3161 + return ret;
3162 }
3163 EXPORT_SYMBOL(smp_call_function_single);
3164
3165 /*
3166 - * this function sends a 'generic call function' IPI to all other CPUs
3167 - * in the system.
3168 - */
3169 -static void __smp_call_function (void (*func) (void *info), void *info,
3170 - int nonatomic, int wait)
3171 -{
3172 - struct call_data_struct data;
3173 - int cpus = num_online_cpus()-1;
3174 -
3175 - if (!cpus)
3176 - return;
3177 -
3178 - data.func = func;
3179 - data.info = info;
3180 - atomic_set(&data.started, 0);
3181 - data.wait = wait;
3182 - if (wait)
3183 - atomic_set(&data.finished, 0);
3184 -
3185 - call_data = &data;
3186 - wmb();
3187 - /* Send a message to all other CPUs and wait for them to respond */
3188 - send_IPI_allbutself(CALL_FUNCTION_VECTOR);
3189 -
3190 - /* Wait for response */
3191 - while (atomic_read(&data.started) != cpus)
3192 - cpu_relax();
3193 -
3194 - if (!wait)
3195 - return;
3196 -
3197 - while (atomic_read(&data.finished) != cpus)
3198 - cpu_relax();
3199 -}
3200 -
3201 -/*
3202 * smp_call_function - run a function on all other CPUs.
3203 * @func: The function to run. This must be fast and non-blocking.
3204 * @info: An arbitrary pointer to pass to the function.
3205 @@ -453,10 +464,7 @@ static void __smp_call_function (void (*
3206 int smp_call_function (void (*func) (void *info), void *info, int nonatomic,
3207 int wait)
3208 {
3209 - spin_lock(&call_lock);
3210 - __smp_call_function(func,info,nonatomic,wait);
3211 - spin_unlock(&call_lock);
3212 - return 0;
3213 + return smp_call_function_mask(cpu_online_map, func, info, wait);
3214 }
3215 EXPORT_SYMBOL(smp_call_function);
3216
3217 @@ -485,7 +493,7 @@ void smp_send_stop(void)
3218 /* Don't deadlock on the call lock in panic */
3219 nolock = !spin_trylock(&call_lock);
3220 local_irq_save(flags);
3221 - __smp_call_function(stop_this_cpu, NULL, 0, 0);
3222 + __smp_call_function_mask(cpu_online_map, stop_this_cpu, NULL, 0);
3223 if (!nolock)
3224 spin_unlock(&call_lock);
3225 disable_all_local_evtchn();
3226 @@ -505,7 +513,9 @@ asmlinkage irqreturn_t smp_reschedule_in
3227 {
3228 #ifndef CONFIG_XEN
3229 ack_APIC_irq();
3230 -#else
3231 +#endif
3232 + add_pda(irq_resched_count, 1);
3233 +#ifdef CONFIG_XEN
3234 return IRQ_HANDLED;
3235 #endif
3236 }
3237 @@ -535,6 +545,7 @@ asmlinkage irqreturn_t smp_call_function
3238 exit_idle();
3239 irq_enter();
3240 (*func)(info);
3241 + add_pda(irq_call_count, 1);
3242 irq_exit();
3243 if (wait) {
3244 mb();
3245 --- sle11-2009-05-14.orig/arch/x86/kernel/time_32-xen.c 2009-03-24 10:12:09.000000000 +0100
3246 +++ sle11-2009-05-14/arch/x86/kernel/time_32-xen.c 2009-03-24 10:12:35.000000000 +0100
3247 @@ -1,6 +1,4 @@
3248 /*
3249 - * linux/arch/i386/kernel/time.c
3250 - *
3251 * Copyright (C) 1991, 1992, 1995 Linus Torvalds
3252 *
3253 * This file contains the PC-specific time handling details:
3254 @@ -74,6 +72,7 @@
3255 #include <asm/arch_hooks.h>
3256
3257 #include <xen/evtchn.h>
3258 +#include <xen/sysctl.h>
3259 #include <xen/interface/vcpu.h>
3260
3261 #include <asm/i8253.h>
3262 @@ -544,6 +543,13 @@ irqreturn_t timer_interrupt(int irq, voi
3263 struct shadow_time_info *shadow = &per_cpu(shadow_time, cpu);
3264 struct vcpu_runstate_info runstate;
3265
3266 + /* Keep nmi watchdog up to date */
3267 +#ifdef __i386__
3268 + per_cpu(irq_stat, smp_processor_id()).irq0_irqs++;
3269 +#else
3270 + add_pda(irq0_irqs, 1);
3271 +#endif
3272 +
3273 /*
3274 * Here we are in the timer irq handler. We just have irqs locally
3275 * disabled but we don't know if the timer_bh is running on the other
3276 @@ -994,7 +1000,7 @@ static int time_cpufreq_notifier(struct
3277 struct cpufreq_freqs *freq = data;
3278 struct xen_platform_op op;
3279
3280 - if (cpu_has(&cpu_data[freq->cpu], X86_FEATURE_CONSTANT_TSC))
3281 + if (cpu_has(&cpu_data(freq->cpu), X86_FEATURE_CONSTANT_TSC))
3282 return 0;
3283
3284 if (val == CPUFREQ_PRECHANGE)
3285 @@ -1032,30 +1038,33 @@ core_initcall(cpufreq_time_setup);
3286 */
3287 static ctl_table xen_subtable[] = {
3288 {
3289 - .ctl_name = 1,
3290 + .ctl_name = CTL_XEN_INDEPENDENT_WALLCLOCK,
3291 .procname = "independent_wallclock",
3292 .data = &independent_wallclock,
3293 .maxlen = sizeof(independent_wallclock),
3294 .mode = 0644,
3295 + .strategy = sysctl_data,
3296 .proc_handler = proc_dointvec
3297 },
3298 {
3299 - .ctl_name = 2,
3300 + .ctl_name = CTL_XEN_PERMITTED_CLOCK_JITTER,
3301 .procname = "permitted_clock_jitter",
3302 .data = &permitted_clock_jitter,
3303 .maxlen = sizeof(permitted_clock_jitter),
3304 .mode = 0644,
3305 + .strategy = sysctl_data,
3306 .proc_handler = proc_doulongvec_minmax
3307 },
3308 - { 0 }
3309 + { }
3310 };
3311 static ctl_table xen_table[] = {
3312 {
3313 - .ctl_name = 123,
3314 + .ctl_name = CTL_XEN,
3315 .procname = "xen",
3316 .mode = 0555,
3317 - .child = xen_subtable},
3318 - { 0 }
3319 + .child = xen_subtable
3320 + },
3321 + { }
3322 };
3323 static int __init xen_sysctl_init(void)
3324 {
3325 --- sle11-2009-05-14.orig/arch/x86/kernel/traps_32-xen.c 2009-02-16 16:17:21.000000000 +0100
3326 +++ sle11-2009-05-14/arch/x86/kernel/traps_32-xen.c 2009-02-16 16:18:36.000000000 +0100
3327 @@ -1,6 +1,4 @@
3328 /*
3329 - * linux/arch/i386/traps.c
3330 - *
3331 * Copyright (C) 1991, 1992 Linus Torvalds
3332 *
3333 * Pentium III FXSR, SSE support
3334 @@ -65,6 +63,11 @@
3335
3336 int panic_on_unrecovered_nmi;
3337
3338 +#ifndef CONFIG_XEN
3339 +DECLARE_BITMAP(used_vectors, NR_VECTORS);
3340 +EXPORT_SYMBOL_GPL(used_vectors);
3341 +#endif
3342 +
3343 asmlinkage int system_call(void);
3344
3345 /* Do we ignore FPU interrupts ? */
3346 @@ -120,7 +123,7 @@ struct stack_frame {
3347
3348 static inline unsigned long print_context_stack(struct thread_info *tinfo,
3349 unsigned long *stack, unsigned long ebp,
3350 - struct stacktrace_ops *ops, void *data)
3351 + const struct stacktrace_ops *ops, void *data)
3352 {
3353 #ifdef CONFIG_FRAME_POINTER
3354 struct stack_frame *frame = (struct stack_frame *)ebp;
3355 @@ -157,7 +160,7 @@ static inline unsigned long print_contex
3356
3357 void dump_trace(struct task_struct *task, struct pt_regs *regs,
3358 unsigned long *stack,
3359 - struct stacktrace_ops *ops, void *data)
3360 + const struct stacktrace_ops *ops, void *data)
3361 {
3362 unsigned long ebp = 0;
3363
3364 @@ -229,7 +232,7 @@ static void print_trace_address(void *da
3365 touch_nmi_watchdog();
3366 }
3367
3368 -static struct stacktrace_ops print_trace_ops = {
3369 +static const struct stacktrace_ops print_trace_ops = {
3370 .warning = print_trace_warning,
3371 .warning_symbol = print_trace_warning_symbol,
3372 .stack = print_trace_stack,
3373 @@ -288,6 +291,11 @@ void dump_stack(void)
3374 {
3375 unsigned long stack;
3376
3377 + printk("Pid: %d, comm: %.20s %s %s %.*s\n",
3378 + current->pid, current->comm, print_tainted(),
3379 + init_utsname()->release,
3380 + (int)strcspn(init_utsname()->version, " "),
3381 + init_utsname()->version);
3382 show_trace(current, NULL, &stack);
3383 }
3384
3385 @@ -296,48 +304,24 @@ EXPORT_SYMBOL(dump_stack);
3386 void show_registers(struct pt_regs *regs)
3387 {
3388 int i;
3389 - int in_kernel = 1;
3390 - unsigned long esp;
3391 - unsigned short ss, gs;
3392 -
3393 - esp = (unsigned long) (&regs->esp);
3394 - savesegment(ss, ss);
3395 - savesegment(gs, gs);
3396 - if (user_mode_vm(regs)) {
3397 - in_kernel = 0;
3398 - esp = regs->esp;
3399 - ss = regs->xss & 0xffff;
3400 - }
3401 +
3402 print_modules();
3403 - printk(KERN_EMERG "CPU: %d\n"
3404 - KERN_EMERG "EIP: %04x:[<%08lx>] %s VLI\n"
3405 - KERN_EMERG "EFLAGS: %08lx (%s %.*s)\n",
3406 - smp_processor_id(), 0xffff & regs->xcs, regs->eip,
3407 - print_tainted(), regs->eflags, init_utsname()->release,
3408 - (int)strcspn(init_utsname()->version, " "),
3409 - init_utsname()->version);
3410 - print_symbol(KERN_EMERG "EIP is at %s\n", regs->eip);
3411 - printk(KERN_EMERG "eax: %08lx ebx: %08lx ecx: %08lx edx: %08lx\n",
3412 - regs->eax, regs->ebx, regs->ecx, regs->edx);
3413 - printk(KERN_EMERG "esi: %08lx edi: %08lx ebp: %08lx esp: %08lx\n",
3414 - regs->esi, regs->edi, regs->ebp, esp);
3415 - printk(KERN_EMERG "ds: %04x es: %04x fs: %04x gs: %04x ss: %04x\n",
3416 - regs->xds & 0xffff, regs->xes & 0xffff, regs->xfs & 0xffff, gs, ss);
3417 + __show_registers(regs, 0);
3418 printk(KERN_EMERG "Process %.*s (pid: %d, ti=%p task=%p task.ti=%p)",
3419 - TASK_COMM_LEN, current->comm, current->pid,
3420 + TASK_COMM_LEN, current->comm, task_pid_nr(current),
3421 current_thread_info(), current, task_thread_info(current));
3422 /*
3423 * When in-kernel, we also print out the stack and code at the
3424 * time of the fault..
3425 */
3426 - if (in_kernel) {
3427 + if (!user_mode_vm(regs)) {
3428 u8 *eip;
3429 unsigned int code_prologue = code_bytes * 43 / 64;
3430 unsigned int code_len = code_bytes;
3431 unsigned char c;
3432
3433 printk("\n" KERN_EMERG "Stack: ");
3434 - show_stack_log_lvl(NULL, regs, (unsigned long *)esp, KERN_EMERG);
3435 + show_stack_log_lvl(NULL, regs, &regs->esp, KERN_EMERG);
3436
3437 printk(KERN_EMERG "Code: ");
3438
3439 @@ -382,11 +366,11 @@ int is_valid_bugaddr(unsigned long eip)
3440 void die(const char * str, struct pt_regs * regs, long err)
3441 {
3442 static struct {
3443 - spinlock_t lock;
3444 + raw_spinlock_t lock;
3445 u32 lock_owner;
3446 int lock_owner_depth;
3447 } die = {
3448 - .lock = __SPIN_LOCK_UNLOCKED(die.lock),
3449 + .lock = __RAW_SPIN_LOCK_UNLOCKED,
3450 .lock_owner = -1,
3451 .lock_owner_depth = 0
3452 };
3453 @@ -397,40 +381,33 @@ void die(const char * str, struct pt_reg
3454
3455 if (die.lock_owner != raw_smp_processor_id()) {
3456 console_verbose();
3457 - spin_lock_irqsave(&die.lock, flags);
3458 + raw_local_irq_save(flags);
3459 + __raw_spin_lock(&die.lock);
3460 die.lock_owner = smp_processor_id();
3461 die.lock_owner_depth = 0;
3462 bust_spinlocks(1);
3463 - }
3464 - else
3465 - local_save_flags(flags);
3466 + } else
3467 + raw_local_irq_save(flags);
3468
3469 if (++die.lock_owner_depth < 3) {
3470 - int nl = 0;
3471 unsigned long esp;
3472 unsigned short ss;
3473
3474 report_bug(regs->eip, regs);
3475
3476 - printk(KERN_EMERG "%s: %04lx [#%d]\n", str, err & 0xffff, ++die_counter);
3477 + printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff,
3478 + ++die_counter);
3479 #ifdef CONFIG_PREEMPT
3480 - printk(KERN_EMERG "PREEMPT ");
3481 - nl = 1;
3482 + printk("PREEMPT ");
3483 #endif
3484 #ifdef CONFIG_SMP
3485 - if (!nl)
3486 - printk(KERN_EMERG);
3487 printk("SMP ");
3488 - nl = 1;
3489 #endif
3490 #ifdef CONFIG_DEBUG_PAGEALLOC
3491 - if (!nl)
3492 - printk(KERN_EMERG);
3493 printk("DEBUG_PAGEALLOC");
3494 - nl = 1;
3495 #endif
3496 - if (nl)
3497 - printk("\n");
3498 + printk("\n");
3499 +
3500 if (notify_die(DIE_OOPS, str, regs, err,
3501 current->thread.trap_no, SIGSEGV) !=
3502 NOTIFY_STOP) {
3503 @@ -454,7 +431,8 @@ void die(const char * str, struct pt_reg
3504 bust_spinlocks(0);
3505 die.lock_owner = -1;
3506 add_taint(TAINT_DIE);
3507 - spin_unlock_irqrestore(&die.lock, flags);
3508 + __raw_spin_unlock(&die.lock);
3509 + raw_local_irq_restore(flags);
3510
3511 if (!regs)
3512 return;
3513 @@ -571,6 +549,7 @@ fastcall void do_##name(struct pt_regs *
3514 info.si_errno = 0; \
3515 info.si_code = sicode; \
3516 info.si_addr = (void __user *)siaddr; \
3517 + trace_hardirqs_fixup(); \
3518 if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr) \
3519 == NOTIFY_STOP) \
3520 return; \
3521 @@ -606,7 +585,7 @@ fastcall void __kprobes do_general_prote
3522 printk_ratelimit())
3523 printk(KERN_INFO
3524 "%s[%d] general protection eip:%lx esp:%lx error:%lx\n",
3525 - current->comm, current->pid,
3526 + current->comm, task_pid_nr(current),
3527 regs->eip, regs->esp, error_code);
3528
3529 force_sig(SIGSEGV, current);
3530 @@ -785,6 +764,8 @@ void restart_nmi(void)
3531 #ifdef CONFIG_KPROBES
3532 fastcall void __kprobes do_int3(struct pt_regs *regs, long error_code)
3533 {
3534 + trace_hardirqs_fixup();
3535 +
3536 if (notify_die(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP)
3537 == NOTIFY_STOP)
3538 return;
3539 @@ -822,6 +803,8 @@ fastcall void __kprobes do_debug(struct
3540 unsigned int condition;
3541 struct task_struct *tsk = current;
3542
3543 + trace_hardirqs_fixup();
3544 +
3545 get_debugreg(condition, 6);
3546
3547 if (notify_die(DIE_DEBUG, "debug", regs, condition, error_code,
3548 @@ -1084,20 +1067,6 @@ asmlinkage void math_emulate(long arg)
3549
3550 #endif /* CONFIG_MATH_EMULATION */
3551
3552 -#ifdef CONFIG_X86_F00F_BUG
3553 -void __init trap_init_f00f_bug(void)
3554 -{
3555 - __set_fixmap(FIX_F00F_IDT, __pa(&idt_table), PAGE_KERNEL_RO);
3556 -
3557 - /*
3558 - * Update the IDT descriptor and reload the IDT so that
3559 - * it uses the read-only mapped virtual address.
3560 - */
3561 - idt_descr.address = fix_to_virt(FIX_F00F_IDT);
3562 - load_idt(&idt_descr);
3563 -}
3564 -#endif
3565 -
3566
3567 /*
3568 * NB. All these are "trap gates" (i.e. events_mask isn't set) except
3569 --- sle11-2009-05-14.orig/arch/x86/kernel/traps_64-xen.c 2009-02-16 16:17:21.000000000 +0100
3570 +++ sle11-2009-05-14/arch/x86/kernel/traps_64-xen.c 2009-02-16 16:18:36.000000000 +0100
3571 @@ -1,6 +1,4 @@
3572 /*
3573 - * linux/arch/x86-64/traps.c
3574 - *
3575 * Copyright (C) 1991, 1992 Linus Torvalds
3576 * Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
3577 *
3578 @@ -33,6 +31,7 @@
3579 #include <linux/uaccess.h>
3580 #include <linux/bug.h>
3581 #include <linux/kdebug.h>
3582 +#include <linux/utsname.h>
3583
3584 #if defined(CONFIG_EDAC)
3585 #include <linux/edac.h>
3586 @@ -205,7 +204,7 @@ static unsigned long *in_exception_stack
3587 #define MSG(txt) ops->warning(data, txt)
3588
3589 /*
3590 - * x86-64 can have upto three kernel stacks:
3591 + * x86-64 can have up to three kernel stacks:
3592 * process stack
3593 * interrupt stack
3594 * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
3595 @@ -219,7 +218,7 @@ static inline int valid_stack_ptr(struct
3596
3597 void dump_trace(struct task_struct *tsk, struct pt_regs *regs,
3598 unsigned long *stack,
3599 - struct stacktrace_ops *ops, void *data)
3600 + const struct stacktrace_ops *ops, void *data)
3601 {
3602 const unsigned cpu = get_cpu();
3603 unsigned long *irqstack_end = (unsigned long*)cpu_pda(cpu)->irqstackptr;
3604 @@ -340,7 +339,7 @@ static void print_trace_address(void *da
3605 printk_address(addr);
3606 }
3607
3608 -static struct stacktrace_ops print_trace_ops = {
3609 +static const struct stacktrace_ops print_trace_ops = {
3610 .warning = print_trace_warning,
3611 .warning_symbol = print_trace_warning_symbol,
3612 .stack = print_trace_stack,
3613 @@ -404,6 +403,12 @@ void show_stack(struct task_struct *tsk,
3614 void dump_stack(void)
3615 {
3616 unsigned long dummy;
3617 +
3618 + printk("Pid: %d, comm: %.20s %s %s %.*s\n",
3619 + current->pid, current->comm, print_tainted(),
3620 + init_utsname()->release,
3621 + (int)strcspn(init_utsname()->version, " "),
3622 + init_utsname()->version);
3623 show_trace(NULL, NULL, &dummy);
3624 }
3625
3626 @@ -466,7 +471,7 @@ void out_of_line_bug(void)
3627 EXPORT_SYMBOL(out_of_line_bug);
3628 #endif
3629
3630 -static DEFINE_SPINLOCK(die_lock);
3631 +static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED;
3632 static int die_owner = -1;
3633 static unsigned int die_nest_count;
3634
3635 @@ -478,13 +483,13 @@ unsigned __kprobes long oops_begin(void)
3636 oops_enter();
3637
3638 /* racy, but better than risking deadlock. */
3639 - local_irq_save(flags);
3640 + raw_local_irq_save(flags);
3641 cpu = smp_processor_id();
3642 - if (!spin_trylock(&die_lock)) {
3643 + if (!__raw_spin_trylock(&die_lock)) {
3644 if (cpu == die_owner)
3645 /* nested oops. should stop eventually */;
3646 else
3647 - spin_lock(&die_lock);
3648 + __raw_spin_lock(&die_lock);
3649 }
3650 die_nest_count++;
3651 die_owner = cpu;
3652 @@ -498,12 +503,10 @@ void __kprobes oops_end(unsigned long fl
3653 die_owner = -1;
3654 bust_spinlocks(0);
3655 die_nest_count--;
3656 - if (die_nest_count)
3657 - /* We still own the lock */
3658 - local_irq_restore(flags);
3659 - else
3660 + if (!die_nest_count)
3661 /* Nest count reaches zero, release the lock. */
3662 - spin_unlock_irqrestore(&die_lock, flags);
3663 + __raw_spin_unlock(&die_lock);
3664 + raw_local_irq_restore(flags);
3665 if (panic_on_oops)
3666 panic("Fatal exception");
3667 oops_exit();
3668 @@ -636,6 +639,7 @@ asmlinkage void do_##name(struct pt_regs
3669 info.si_errno = 0; \
3670 info.si_code = sicode; \
3671 info.si_addr = (void __user *)siaddr; \
3672 + trace_hardirqs_fixup(); \
3673 if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr) \
3674 == NOTIFY_STOP) \
3675 return; \
3676 @@ -741,11 +745,8 @@ mem_parity_error(unsigned char reason, s
3677
3678 printk(KERN_EMERG "Dazed and confused, but trying to continue\n");
3679
3680 -#if 0 /* XEN */
3681 /* Clear and disable the memory parity error line. */
3682 - reason = (reason & 0xf) | 4;
3683 - outb(reason, 0x61);
3684 -#endif /* XEN */
3685 + clear_mem_error(reason);
3686 }
3687
3688 static __kprobes void
3689 @@ -754,14 +755,8 @@ io_check_error(unsigned char reason, str
3690 printk("NMI: IOCK error (debug interrupt?)\n");
3691 show_registers(regs);
3692
3693 -#if 0 /* XEN */
3694 /* Re-enable the IOCK line, wait for a few seconds */
3695 - reason = (reason & 0xf) | 8;
3696 - outb(reason, 0x61);
3697 - mdelay(2000);
3698 - reason &= ~8;
3699 - outb(reason, 0x61);
3700 -#endif /* XEN */
3701 + clear_io_check_error(reason);
3702 }
3703
3704 static __kprobes void
3705 @@ -821,6 +816,8 @@ asmlinkage __kprobes void default_do_nmi
3706 /* runs on IST stack. */
3707 asmlinkage void __kprobes do_int3(struct pt_regs * regs, long error_code)
3708 {
3709 + trace_hardirqs_fixup();
3710 +
3711 if (notify_die(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP) == NOTIFY_STOP) {
3712 return;
3713 }
3714 @@ -858,6 +855,8 @@ asmlinkage void __kprobes do_debug(struc
3715 struct task_struct *tsk = current;
3716 siginfo_t info;
3717
3718 + trace_hardirqs_fixup();
3719 +
3720 get_debugreg(condition, 6);
3721
3722 if (notify_die(DIE_DEBUG, "debug", regs, condition, error_code,
3723 --- sle11-2009-05-14.orig/arch/x86/kernel/vsyscall_64-xen.c 2009-02-16 16:17:21.000000000 +0100
3724 +++ sle11-2009-05-14/arch/x86/kernel/vsyscall_64-xen.c 2009-02-16 16:18:36.000000000 +0100
3725 @@ -1,6 +1,4 @@
3726 /*
3727 - * linux/arch/x86_64/kernel/vsyscall.c
3728 - *
3729 * Copyright (C) 2001 Andrea Arcangeli <andrea@suse.de> SuSE
3730 * Copyright 2003 Andi Kleen, SuSE Labs.
3731 *
3732 @@ -50,12 +48,12 @@
3733 ({unsigned long v; \
3734 extern char __vsyscall_0; \
3735 asm("" : "=r" (v) : "0" (x)); \
3736 - ((v - VSYSCALL_FIRST_PAGE) + __pa_symbol(&__vsyscall_0)); })
3737 + ((v - VSYSCALL_START) + __pa_symbol(&__vsyscall_0)); })
3738
3739 /*
3740 * vsyscall_gtod_data contains data that is :
3741 * - readonly from vsyscalls
3742 - * - writen by timer interrupt or systcl (/proc/sys/kernel/vsyscall64)
3743 + * - written by timer interrupt or systcl (/proc/sys/kernel/vsyscall64)
3744 * Try to keep this structure as small as possible to avoid cache line ping pongs
3745 */
3746 int __vgetcpu_mode __section_vgetcpu_mode;
3747 @@ -66,6 +64,16 @@ struct vsyscall_gtod_data __vsyscall_gto
3748 .sysctl_enabled = 1,
3749 };
3750
3751 +void update_vsyscall_tz(void)
3752 +{
3753 + unsigned long flags;
3754 +
3755 + write_seqlock_irqsave(&vsyscall_gtod_data.lock, flags);
3756 + /* sys_tz has changed */
3757 + vsyscall_gtod_data.sys_tz = sys_tz;
3758 + write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
3759 +}
3760 +
3761 void update_vsyscall(struct timespec *wall_time, struct clocksource *clock)
3762 {
3763 unsigned long flags;
3764 @@ -79,8 +87,6 @@ void update_vsyscall(struct timespec *wa
3765 vsyscall_gtod_data.clock.shift = clock->shift;
3766 vsyscall_gtod_data.wall_time_sec = wall_time->tv_sec;
3767 vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
3768 - vsyscall_gtod_data.sys_tz = sys_tz;
3769 - vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
3770 vsyscall_gtod_data.wall_to_monotonic = wall_to_monotonic;
3771 write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
3772 }
3773 @@ -166,7 +172,7 @@ time_t __vsyscall(1) vtime(time_t *t)
3774 if (unlikely(!__vsyscall_gtod_data.sysctl_enabled))
3775 return time_syscall(t);
3776
3777 - vgettimeofday(&tv, 0);
3778 + vgettimeofday(&tv, NULL);
3779 result = tv.tv_sec;
3780 if (t)
3781 *t = result;
3782 @@ -260,18 +266,10 @@ out:
3783 return ret;
3784 }
3785
3786 -static int vsyscall_sysctl_nostrat(ctl_table *t, int __user *name, int nlen,
3787 - void __user *oldval, size_t __user *oldlenp,
3788 - void __user *newval, size_t newlen)
3789 -{
3790 - return -ENOSYS;
3791 -}
3792 -
3793 static ctl_table kernel_table2[] = {
3794 - { .ctl_name = 99, .procname = "vsyscall64",
3795 + { .procname = "vsyscall64",
3796 .data = &vsyscall_gtod_data.sysctl_enabled, .maxlen = sizeof(int),
3797 .mode = 0644,
3798 - .strategy = vsyscall_sysctl_nostrat,
3799 .proc_handler = vsyscall_sysctl_change },
3800 {}
3801 };
3802 @@ -291,9 +289,9 @@ static void __cpuinit vsyscall_set_cpu(i
3803 unsigned long d;
3804 unsigned long node = 0;
3805 #ifdef CONFIG_NUMA
3806 - node = cpu_to_node[cpu];
3807 + node = cpu_to_node(cpu);
3808 #endif
3809 - if (cpu_has(&cpu_data[cpu], X86_FEATURE_RDTSCP))
3810 + if (cpu_has(&cpu_data(cpu), X86_FEATURE_RDTSCP))
3811 write_rdtscp_aux((node << 12) | cpu);
3812
3813 /* Store cpu number in limit so that it can be loaded quickly
3814 --- sle11-2009-05-14.orig/arch/x86/mm/fault_32-xen.c 2009-02-16 16:17:21.000000000 +0100
3815 +++ sle11-2009-05-14/arch/x86/mm/fault_32-xen.c 2009-02-16 16:18:36.000000000 +0100
3816 @@ -25,6 +25,7 @@
3817 #include <linux/kprobes.h>
3818 #include <linux/uaccess.h>
3819 #include <linux/kdebug.h>
3820 +#include <linux/kprobes.h>
3821
3822 #include <asm/system.h>
3823 #include <asm/desc.h>
3824 @@ -32,33 +33,27 @@
3825
3826 extern void die(const char *,struct pt_regs *,long);
3827
3828 -static ATOMIC_NOTIFIER_HEAD(notify_page_fault_chain);
3829 -
3830 -int register_page_fault_notifier(struct notifier_block *nb)
3831 +#ifdef CONFIG_KPROBES
3832 +static inline int notify_page_fault(struct pt_regs *regs)
3833 {
3834 - vmalloc_sync_all();
3835 - return atomic_notifier_chain_register(&notify_page_fault_chain, nb);
3836 -}
3837 -EXPORT_SYMBOL_GPL(register_page_fault_notifier);
3838 + int ret = 0;
3839
3840 -int unregister_page_fault_notifier(struct notifier_block *nb)
3841 -{
3842 - return atomic_notifier_chain_unregister(&notify_page_fault_chain, nb);
3843 -}
3844 -EXPORT_SYMBOL_GPL(unregister_page_fault_notifier);
3845 + /* kprobe_running() needs smp_processor_id() */
3846 + if (!user_mode_vm(regs)) {
3847 + preempt_disable();
3848 + if (kprobe_running() && kprobe_fault_handler(regs, 14))
3849 + ret = 1;
3850 + preempt_enable();
3851 + }
3852
3853 -static inline int notify_page_fault(struct pt_regs *regs, long err)
3854 + return ret;
3855 +}
3856 +#else
3857 +static inline int notify_page_fault(struct pt_regs *regs)
3858 {
3859 - struct die_args args = {
3860 - .regs = regs,
3861 - .str = "page fault",
3862 - .err = err,
3863 - .trapnr = 14,
3864 - .signr = SIGSEGV
3865 - };
3866 - return atomic_notifier_call_chain(&notify_page_fault_chain,
3867 - DIE_PAGE_FAULT, &args);
3868 + return 0;
3869 }
3870 +#endif
3871
3872 /*
3873 * Return EIP plus the CS segment base. The segment limit is also
3874 @@ -110,7 +105,7 @@ static inline unsigned long get_segment_
3875 LDT and other horrors are only used in user space. */
3876 if (seg & (1<<2)) {
3877 /* Must lock the LDT while reading it. */
3878 - down(&current->mm->context.sem);
3879 + mutex_lock(&current->mm->context.lock);
3880 desc = current->mm->context.ldt;
3881 desc = (void *)desc + (seg & ~7);
3882 } else {
3883 @@ -123,7 +118,7 @@ static inline unsigned long get_segment_
3884 base = get_desc_base((unsigned long *)desc);
3885
3886 if (seg & (1<<2)) {
3887 - up(&current->mm->context.sem);
3888 + mutex_unlock(&current->mm->context.lock);
3889 } else
3890 put_cpu();
3891
3892 @@ -244,7 +239,7 @@ static void dump_fault_path(unsigned lon
3893 if (mfn_to_pfn(mfn) >= highstart_pfn)
3894 return;
3895 #endif
3896 - if (p[0] & _PAGE_PRESENT) {
3897 + if ((p[0] & _PAGE_PRESENT) && !(p[0] & _PAGE_PSE)) {
3898 page = mfn_to_pfn(mfn) << PAGE_SHIFT;
3899 p = (unsigned long *) __va(page);
3900 address &= 0x001fffff;
3901 @@ -270,7 +265,8 @@ static void dump_fault_path(unsigned lon
3902 * it's allocated already.
3903 */
3904 if ((machine_to_phys(page) >> PAGE_SHIFT) < max_low_pfn
3905 - && (page & _PAGE_PRESENT)) {
3906 + && (page & _PAGE_PRESENT)
3907 + && !(page & _PAGE_PSE)) {
3908 page = machine_to_phys(page & PAGE_MASK);
3909 page = ((unsigned long *) __va(page))[(address >> PAGE_SHIFT)
3910 & (PTRS_PER_PTE - 1)];
3911 @@ -416,6 +412,11 @@ fastcall void __kprobes do_page_fault(st
3912 int write, si_code;
3913 int fault;
3914
3915 + /*
3916 + * We can fault from pretty much anywhere, with unknown IRQ state.
3917 + */
3918 + trace_hardirqs_fixup();
3919 +
3920 /* get the address */
3921 address = read_cr2();
3922
3923 @@ -453,7 +454,7 @@ fastcall void __kprobes do_page_fault(st
3924 /* Can take a spurious fault if mapping changes R/O -> R/W. */
3925 if (spurious_fault(regs, address, error_code))
3926 return;
3927 - if (notify_page_fault(regs, error_code) == NOTIFY_STOP)
3928 + if (notify_page_fault(regs))
3929 return;
3930 /*
3931 * Don't take the mm semaphore here. If we fixup a prefetch
3932 @@ -462,7 +463,7 @@ fastcall void __kprobes do_page_fault(st
3933 goto bad_area_nosemaphore;
3934 }
3935
3936 - if (notify_page_fault(regs, error_code) == NOTIFY_STOP)
3937 + if (notify_page_fault(regs))
3938 return;
3939
3940 /* It's safe to allow irq's after cr2 has been saved and the vmalloc
3941 @@ -481,7 +482,7 @@ fastcall void __kprobes do_page_fault(st
3942
3943 /* When running in the kernel we expect faults to occur only to
3944 * addresses in user space. All other faults represent errors in the
3945 - * kernel and should generate an OOPS. Unfortunatly, in the case of an
3946 + * kernel and should generate an OOPS. Unfortunately, in the case of an
3947 * erroneous fault occurring in a code path which already holds mmap_sem
3948 * we will deadlock attempting to validate the fault against the
3949 * address space. Luckily the kernel only validly references user
3950 @@ -489,7 +490,7 @@ fastcall void __kprobes do_page_fault(st
3951 * exceptions table.
3952 *
3953 * As the vast majority of faults will be valid we will only perform
3954 - * the source reference check when there is a possibilty of a deadlock.
3955 + * the source reference check when there is a possibility of a deadlock.
3956 * Attempt to lock the address space, if we cannot we then validate the
3957 * source. If this is invalid we can skip the address space check,
3958 * thus avoiding the deadlock.
3959 @@ -598,8 +599,8 @@ bad_area_nosemaphore:
3960 printk_ratelimit()) {
3961 printk("%s%s[%d]: segfault at %08lx eip %08lx "
3962 "esp %08lx error %lx\n",
3963 - tsk->pid > 1 ? KERN_INFO : KERN_EMERG,
3964 - tsk->comm, tsk->pid, address, regs->eip,
3965 + task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
3966 + tsk->comm, task_pid_nr(tsk), address, regs->eip,
3967 regs->esp, error_code);
3968 }
3969 tsk->thread.cr2 = address;
3970 @@ -664,8 +665,7 @@ no_context:
3971 printk(KERN_ALERT "BUG: unable to handle kernel paging"
3972 " request");
3973 printk(" at virtual address %08lx\n",address);
3974 - printk(KERN_ALERT " printing eip:\n");
3975 - printk("%08lx\n", regs->eip);
3976 + printk(KERN_ALERT "printing eip: %08lx\n", regs->eip);
3977 dump_fault_path(address);
3978 }
3979 tsk->thread.cr2 = address;
3980 @@ -681,14 +681,14 @@ no_context:
3981 */
3982 out_of_memory:
3983 up_read(&mm->mmap_sem);
3984 - if (is_init(tsk)) {
3985 + if (is_global_init(tsk)) {
3986 yield();
3987 down_read(&mm->mmap_sem);
3988 goto survive;
3989 }
3990 printk("VM: killing process %s\n", tsk->comm);
3991 if (error_code & 4)
3992 - do_exit(SIGKILL);
3993 + do_group_exit(SIGKILL);
3994 goto no_context;
3995
3996 do_sigbus:
3997 --- sle11-2009-05-14.orig/arch/x86/mm/fault_64-xen.c 2009-02-16 16:17:21.000000000 +0100
3998 +++ sle11-2009-05-14/arch/x86/mm/fault_64-xen.c 2009-02-16 16:18:36.000000000 +0100
3999 @@ -25,6 +25,7 @@
4000 #include <linux/kprobes.h>
4001 #include <linux/uaccess.h>
4002 #include <linux/kdebug.h>
4003 +#include <linux/kprobes.h>
4004
4005 #include <asm/system.h>
4006 #include <asm/pgalloc.h>
4007 @@ -40,34 +41,27 @@
4008 #define PF_RSVD (1<<3)
4009 #define PF_INSTR (1<<4)
4010
4011 -static ATOMIC_NOTIFIER_HEAD(notify_page_fault_chain);
4012 -
4013 -/* Hook to register for page fault notifications */
4014 -int register_page_fault_notifier(struct notifier_block *nb)
4015 +#ifdef CONFIG_KPROBES
4016 +static inline int notify_page_fault(struct pt_regs *regs)
4017 {
4018 - vmalloc_sync_all();
4019 - return atomic_notifier_chain_register(&notify_page_fault_chain, nb);
4020 -}
4021 -EXPORT_SYMBOL_GPL(register_page_fault_notifier);
4022 + int ret = 0;
4023
4024 -int unregister_page_fault_notifier(struct notifier_block *nb)
4025 -{
4026 - return atomic_notifier_chain_unregister(&notify_page_fault_chain, nb);
4027 -}
4028 -EXPORT_SYMBOL_GPL(unregister_page_fault_notifier);
4029 + /* kprobe_running() needs smp_processor_id() */
4030 + if (!user_mode(regs)) {
4031 + preempt_disable();
4032 + if (kprobe_running() && kprobe_fault_handler(regs, 14))
4033 + ret = 1;
4034 + preempt_enable();
4035 + }
4036
4037 -static inline int notify_page_fault(struct pt_regs *regs, long err)
4038 + return ret;
4039 +}
4040 +#else
4041 +static inline int notify_page_fault(struct pt_regs *regs)
4042 {
4043 - struct die_args args = {
4044 - .regs = regs,
4045 - .str = "page fault",
4046 - .err = err,
4047 - .trapnr = 14,
4048 - .signr = SIGSEGV
4049 - };
4050 - return atomic_notifier_call_chain(&notify_page_fault_chain,
4051 - DIE_PAGE_FAULT, &args);
4052 + return 0;
4053 }
4054 +#endif
4055
4056 /* Sometimes the CPU reports invalid exceptions on prefetch.
4057 Check that here and ignore.
4058 @@ -175,7 +169,7 @@ void dump_pagetable(unsigned long addres
4059 pmd = pmd_offset(pud, address);
4060 if (bad_address(pmd)) goto bad;
4061 printk("PMD %lx ", pmd_val(*pmd));
4062 - if (!pmd_present(*pmd)) goto ret;
4063 + if (!pmd_present(*pmd) || pmd_large(*pmd)) goto ret;
4064
4065 pte = pte_offset_kernel(pmd, address);
4066 if (bad_address(pte)) goto bad;
4067 @@ -294,7 +288,6 @@ static int vmalloc_fault(unsigned long a
4068 return 0;
4069 }
4070
4071 -static int page_fault_trace;
4072 int show_unhandled_signals = 1;
4073
4074
4075 @@ -371,6 +364,11 @@ asmlinkage void __kprobes do_page_fault(
4076 if (!user_mode(regs))
4077 error_code &= ~PF_USER; /* means kernel */
4078
4079 + /*
4080 + * We can fault from pretty much anywhere, with unknown IRQ state.
4081 + */
4082 + trace_hardirqs_fixup();
4083 +
4084 tsk = current;
4085 mm = tsk->mm;
4086 prefetchw(&mm->mmap_sem);
4087 @@ -408,7 +406,7 @@ asmlinkage void __kprobes do_page_fault(
4088 /* Can take a spurious fault if mapping changes R/O -> R/W. */
4089 if (spurious_fault(regs, address, error_code))
4090 return;
4091 - if (notify_page_fault(regs, error_code) == NOTIFY_STOP)
4092 + if (notify_page_fault(regs))
4093 return;
4094 /*
4095 * Don't take the mm semaphore here. If we fixup a prefetch
4096 @@ -417,16 +415,12 @@ asmlinkage void __kprobes do_page_fault(
4097 goto bad_area_nosemaphore;
4098 }
4099
4100 - if (notify_page_fault(regs, error_code) == NOTIFY_STOP)
4101 + if (notify_page_fault(regs))
4102 return;
4103
4104 if (likely(regs->eflags & X86_EFLAGS_IF))
4105 local_irq_enable();
4106
4107 - if (unlikely(page_fault_trace))
4108 - printk("pagefault rip:%lx rsp:%lx cs:%lu ss:%lu address %lx error %lx\n",
4109 - regs->rip,regs->rsp,regs->cs,regs->ss,address,error_code);
4110 -
4111 if (unlikely(error_code & PF_RSVD))
4112 pgtable_bad(address, regs, error_code);
4113
4114 @@ -447,7 +441,7 @@ asmlinkage void __kprobes do_page_fault(
4115 again:
4116 /* When running in the kernel we expect faults to occur only to
4117 * addresses in user space. All other faults represent errors in the
4118 - * kernel and should generate an OOPS. Unfortunatly, in the case of an
4119 + * kernel and should generate an OOPS. Unfortunately, in the case of an
4120 * erroneous fault occurring in a code path which already holds mmap_sem
4121 * we will deadlock attempting to validate the fault against the
4122 * address space. Luckily the kernel only validly references user
4123 @@ -455,7 +449,7 @@ asmlinkage void __kprobes do_page_fault(
4124 * exceptions table.
4125 *
4126 * As the vast majority of faults will be valid we will only perform
4127 - * the source reference check when there is a possibilty of a deadlock.
4128 + * the source reference check when there is a possibility of a deadlock.
4129 * Attempt to lock the address space, if we cannot we then validate the
4130 * source. If this is invalid we can skip the address space check,
4131 * thus avoiding the deadlock.
4132 @@ -557,7 +551,7 @@ bad_area_nosemaphore:
4133 if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
4134 printk_ratelimit()) {
4135 printk(
4136 - "%s%s[%d]: segfault at %016lx rip %016lx rsp %016lx error %lx\n",
4137 + "%s%s[%d]: segfault at %lx rip %lx rsp %lx error %lx\n",
4138 tsk->pid > 1 ? KERN_INFO : KERN_EMERG,
4139 tsk->comm, tsk->pid, address, regs->rip,
4140 regs->rsp, error_code);
4141 @@ -623,7 +617,7 @@ no_context:
4142 */
4143 out_of_memory:
4144 up_read(&mm->mmap_sem);
4145 - if (is_init(current)) {
4146 + if (is_global_init(current)) {
4147 yield();
4148 goto again;
4149 }
4150 @@ -690,10 +684,3 @@ void vmalloc_sync_all(void)
4151 BUILD_BUG_ON(!(((MODULES_END - 1) & PGDIR_MASK) ==
4152 (__START_KERNEL & PGDIR_MASK)));
4153 }
4154 -
4155 -static int __init enable_pagefaulttrace(char *str)
4156 -{
4157 - page_fault_trace = 1;
4158 - return 1;
4159 -}
4160 -__setup("pagefaulttrace", enable_pagefaulttrace);
4161 --- sle11-2009-05-14.orig/arch/x86/mm/hypervisor.c 2009-03-04 11:28:34.000000000 +0100
4162 +++ sle11-2009-05-14/arch/x86/mm/hypervisor.c 2009-05-06 10:23:43.000000000 +0200
4163 @@ -496,6 +496,9 @@ int xen_create_contiguous_region(
4164 unsigned long frame, flags;
4165 unsigned int i;
4166 int rc, success;
4167 +#ifdef CONFIG_64BIT
4168 + pte_t *ptep = NULL;
4169 +#endif
4170 struct xen_memory_exchange exchange = {
4171 .in = {
4172 .nr_extents = 1UL << order,
4173 @@ -521,6 +524,27 @@ int xen_create_contiguous_region(
4174 if (unlikely(order > MAX_CONTIG_ORDER))
4175 return -ENOMEM;
4176
4177 +#ifdef CONFIG_64BIT
4178 + if (unlikely(vstart > PAGE_OFFSET + MAXMEM)) {
4179 + unsigned int level;
4180 +
4181 + if (vstart < __START_KERNEL_map
4182 + || vstart + (PAGE_SIZE << order) > (unsigned long)_end)
4183 + return -EINVAL;
4184 + ptep = lookup_address((unsigned long)__va(__pa(vstart)),
4185 + &level);
4186 + if (ptep && pte_none(*ptep))
4187 + ptep = NULL;
4188 + if (vstart < __START_KERNEL && ptep)
4189 + return -EINVAL;
4190 + if (order > MAX_CONTIG_ORDER - 1)
4191 + return -ENOMEM;
4192 + }
4193 +#else
4194 + if (unlikely(vstart + (PAGE_SIZE << order) > (unsigned long)high_memory))
4195 + return -EINVAL;
4196 +#endif
4197 +
4198 set_xen_guest_handle(exchange.in.extent_start, in_frames);
4199 set_xen_guest_handle(exchange.out.extent_start, &out_frame);
4200
4201 @@ -533,9 +557,19 @@ int xen_create_contiguous_region(
4202 in_frames[i] = pfn_to_mfn((__pa(vstart) >> PAGE_SHIFT) + i);
4203 MULTI_update_va_mapping(cr_mcl + i, vstart + (i*PAGE_SIZE),
4204 __pte_ma(0), 0);
4205 +#ifdef CONFIG_64BIT
4206 + if (ptep)
4207 + MULTI_update_va_mapping(cr_mcl + i + (1U << order),
4208 + (unsigned long)__va(__pa(vstart)) + (i*PAGE_SIZE),
4209 + __pte_ma(0), 0);
4210 +#endif
4211 set_phys_to_machine((__pa(vstart)>>PAGE_SHIFT)+i,
4212 INVALID_P2M_ENTRY);
4213 }
4214 +#ifdef CONFIG_64BIT
4215 + if (ptep)
4216 + i += i;
4217 +#endif
4218 if (HYPERVISOR_multicall_check(cr_mcl, i, NULL))
4219 BUG();
4220
4221 @@ -569,9 +603,18 @@ int xen_create_contiguous_region(
4222 frame = success ? (out_frame + i) : in_frames[i];
4223 MULTI_update_va_mapping(cr_mcl + i, vstart + (i*PAGE_SIZE),
4224 pfn_pte_ma(frame, PAGE_KERNEL), 0);
4225 +#ifdef CONFIG_64BIT
4226 + if (ptep)
4227 + MULTI_update_va_mapping(cr_mcl + i + (1U << order),
4228 + (unsigned long)__va(__pa(vstart)) + (i*PAGE_SIZE),
4229 + pfn_pte_ma(frame, PAGE_KERNEL_RO), 0);
4230 +#endif
4231 set_phys_to_machine((__pa(vstart)>>PAGE_SHIFT)+i, frame);
4232 }
4233 -
4234 +#ifdef CONFIG_64BIT
4235 + if (ptep)
4236 + i += i;
4237 +#endif
4238 cr_mcl[i - 1].args[MULTI_UVMFLAGS_INDEX] = order
4239 ? UVMF_TLB_FLUSH|UVMF_ALL
4240 : UVMF_INVLPG|UVMF_ALL;
4241 --- sle11-2009-05-14.orig/arch/x86/mm/init_32-xen.c 2009-02-16 16:17:21.000000000 +0100
4242 +++ sle11-2009-05-14/arch/x86/mm/init_32-xen.c 2009-02-16 16:18:36.000000000 +0100
4243 @@ -94,7 +94,14 @@ static pte_t * __init one_page_table_ini
4244 #else
4245 if (!(__pmd_val(*pmd) & _PAGE_PRESENT)) {
4246 #endif
4247 - pte_t *page_table = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE);
4248 + pte_t *page_table = NULL;
4249 +
4250 +#ifdef CONFIG_DEBUG_PAGEALLOC
4251 + page_table = (pte_t *) alloc_bootmem_pages(PAGE_SIZE);
4252 +#endif
4253 + if (!page_table)
4254 + page_table =
4255 + (pte_t *)alloc_bootmem_low_pages(PAGE_SIZE);
4256
4257 paravirt_alloc_pt(&init_mm, __pa(page_table) >> PAGE_SHIFT);
4258 make_lowmem_page_readonly(page_table,
4259 @@ -102,7 +109,7 @@ static pte_t * __init one_page_table_ini
4260 set_pmd(pmd, __pmd(__pa(page_table) | _PAGE_TABLE));
4261 BUG_ON(page_table != pte_offset_kernel(pmd, 0));
4262 }
4263 -
4264 +
4265 return pte_offset_kernel(pmd, 0);
4266 }
4267
4268 @@ -360,8 +367,13 @@ extern void set_highmem_pages_init(int);
4269 static void __init set_highmem_pages_init(int bad_ppro)
4270 {
4271 int pfn;
4272 - for (pfn = highstart_pfn; pfn < highend_pfn; pfn++)
4273 - add_one_highpage_init(pfn_to_page(pfn), pfn, bad_ppro);
4274 + for (pfn = highstart_pfn; pfn < highend_pfn; pfn++) {
4275 + /*
4276 + * Holes under sparsemem might not have no mem_map[]:
4277 + */
4278 + if (pfn_valid(pfn))
4279 + add_one_highpage_init(pfn_to_page(pfn), pfn, bad_ppro);
4280 + }
4281 totalram_pages += totalhigh_pages;
4282 }
4283 #endif /* CONFIG_FLATMEM */
4284 @@ -779,35 +791,18 @@ int arch_add_memory(int nid, u64 start,
4285 return __add_pages(zone, start_pfn, nr_pages);
4286 }
4287
4288 -int remove_memory(u64 start, u64 size)
4289 -{
4290 - return -EINVAL;
4291 -}
4292 -EXPORT_SYMBOL_GPL(remove_memory);
4293 #endif
4294
4295 struct kmem_cache *pmd_cache;
4296
4297 void __init pgtable_cache_init(void)
4298 {
4299 - size_t pgd_size = PTRS_PER_PGD*sizeof(pgd_t);
4300 -
4301 - if (PTRS_PER_PMD > 1) {
4302 + if (PTRS_PER_PMD > 1)
4303 pmd_cache = kmem_cache_create("pmd",
4304 - PTRS_PER_PMD*sizeof(pmd_t),
4305 - PTRS_PER_PMD*sizeof(pmd_t),
4306 - SLAB_PANIC,
4307 - pmd_ctor);
4308 - if (!SHARED_KERNEL_PMD) {
4309 - /* If we're in PAE mode and have a non-shared
4310 - kernel pmd, then the pgd size must be a
4311 - page size. This is because the pgd_list
4312 - links through the page structure, so there
4313 - can only be one pgd per page for this to
4314 - work. */
4315 - pgd_size = PAGE_SIZE;
4316 - }
4317 - }
4318 + PTRS_PER_PMD*sizeof(pmd_t),
4319 + PTRS_PER_PMD*sizeof(pmd_t),
4320 + SLAB_PANIC,
4321 + pmd_ctor);
4322 }
4323
4324 /*
4325 --- sle11-2009-05-14.orig/arch/x86/mm/init_64-xen.c 2009-02-16 16:17:21.000000000 +0100
4326 +++ sle11-2009-05-14/arch/x86/mm/init_64-xen.c 2009-02-16 16:18:36.000000000 +0100
4327 @@ -767,7 +767,7 @@ static void xen_finish_init_mapping(void
4328 /* Setup the direct mapping of the physical memory at PAGE_OFFSET.
4329 This runs before bootmem is initialized and gets pages directly from the
4330 physical memory. To access them they are temporarily mapped. */
4331 -void __meminit init_memory_mapping(unsigned long start, unsigned long end)
4332 +void __init_refok init_memory_mapping(unsigned long start, unsigned long end)
4333 {
4334 unsigned long next;
4335
4336 @@ -901,12 +901,6 @@ error:
4337 }
4338 EXPORT_SYMBOL_GPL(arch_add_memory);
4339
4340 -int remove_memory(u64 start, u64 size)
4341 -{
4342 - return -EINVAL;
4343 -}
4344 -EXPORT_SYMBOL_GPL(remove_memory);
4345 -
4346 #if !defined(CONFIG_ACPI_NUMA) && defined(CONFIG_NUMA)
4347 int memory_add_physaddr_to_nid(u64 start)
4348 {
4349 @@ -1173,14 +1167,6 @@ int in_gate_area_no_task(unsigned long a
4350 return (addr >= VSYSCALL_START) && (addr < VSYSCALL_END);
4351 }
4352
4353 -#ifndef CONFIG_XEN
4354 -void * __init alloc_bootmem_high_node(pg_data_t *pgdat, unsigned long size)
4355 -{
4356 - return __alloc_bootmem_core(pgdat->bdata, size,
4357 - SMP_CACHE_BYTES, (4UL*1024*1024*1024), 0);
4358 -}
4359 -#endif
4360 -
4361 const char *arch_vma_name(struct vm_area_struct *vma)
4362 {
4363 if (vma->vm_mm && vma->vm_start == (long)vma->vm_mm->context.vdso)
4364 @@ -1189,3 +1175,48 @@ const char *arch_vma_name(struct vm_area
4365 return "[vsyscall]";
4366 return NULL;
4367 }
4368 +
4369 +#ifdef CONFIG_SPARSEMEM_VMEMMAP
4370 +/*
4371 + * Initialise the sparsemem vmemmap using huge-pages at the PMD level.
4372 + */
4373 +int __meminit vmemmap_populate(struct page *start_page,
4374 + unsigned long size, int node)
4375 +{
4376 + unsigned long addr = (unsigned long)start_page;
4377 + unsigned long end = (unsigned long)(start_page + size);
4378 + unsigned long next;
4379 + pgd_t *pgd;
4380 + pud_t *pud;
4381 + pmd_t *pmd;
4382 +
4383 + for (; addr < end; addr = next) {
4384 + next = pmd_addr_end(addr, end);
4385 +
4386 + pgd = vmemmap_pgd_populate(addr, node);
4387 + if (!pgd)
4388 + return -ENOMEM;
4389 + pud = vmemmap_pud_populate(pgd, addr, node);
4390 + if (!pud)
4391 + return -ENOMEM;
4392 +
4393 + pmd = pmd_offset(pud, addr);
4394 + if (pmd_none(*pmd)) {
4395 + pte_t entry;
4396 + void *p = vmemmap_alloc_block(PMD_SIZE, node);
4397 + if (!p)
4398 + return -ENOMEM;
4399 +
4400 + entry = pfn_pte(__pa(p) >> PAGE_SHIFT, PAGE_KERNEL);
4401 + mk_pte_huge(entry);
4402 + set_pmd(pmd, __pmd(pte_val(entry)));
4403 +
4404 + printk(KERN_DEBUG " [%lx-%lx] PMD ->%p on node %d\n",
4405 + addr, addr + PMD_SIZE - 1, p, node);
4406 + } else
4407 + vmemmap_verify((pte_t *)pmd, node, addr, next);
4408 + }
4409 +
4410 + return 0;
4411 +}
4412 +#endif
4413 --- sle11-2009-05-14.orig/arch/x86/mm/pageattr_64-xen.c 2009-02-16 16:17:21.000000000 +0100
4414 +++ sle11-2009-05-14/arch/x86/mm/pageattr_64-xen.c 2009-02-16 16:18:36.000000000 +0100
4415 @@ -17,9 +17,6 @@
4416 #include <asm/pgalloc.h>
4417 #include <asm/mmu_context.h>
4418
4419 -LIST_HEAD(mm_unpinned);
4420 -DEFINE_SPINLOCK(mm_unpinned_lock);
4421 -
4422 static void _pin_lock(struct mm_struct *mm, int lock) {
4423 if (lock)
4424 spin_lock(&mm->page_table_lock);
4425 @@ -81,8 +78,8 @@ static void _pin_lock(struct mm_struct *
4426 #define PIN_BATCH 8
4427 static DEFINE_PER_CPU(multicall_entry_t[PIN_BATCH], pb_mcl);
4428
4429 -static inline unsigned int mm_walk_set_prot(void *pt, pgprot_t flags,
4430 - unsigned int cpu, unsigned int seq)
4431 +static inline unsigned int pgd_walk_set_prot(void *pt, pgprot_t flags,
4432 + unsigned int cpu, unsigned int seq)
4433 {
4434 struct page *page = virt_to_page(pt);
4435 unsigned long pfn = page_to_pfn(page);
4436 @@ -100,9 +97,9 @@ static inline unsigned int mm_walk_set_p
4437 return seq;
4438 }
4439
4440 -static void mm_walk(struct mm_struct *mm, pgprot_t flags)
4441 +static void pgd_walk(pgd_t *pgd_base, pgprot_t flags)
4442 {
4443 - pgd_t *pgd;
4444 + pgd_t *pgd = pgd_base;
4445 pud_t *pud;
4446 pmd_t *pmd;
4447 pte_t *pte;
4448 @@ -110,7 +107,6 @@ static void mm_walk(struct mm_struct *mm
4449 unsigned int cpu, seq;
4450 multicall_entry_t *mcl;
4451
4452 - pgd = mm->pgd;
4453 cpu = get_cpu();
4454
4455 /*
4456 @@ -125,18 +121,18 @@ static void mm_walk(struct mm_struct *mm
4457 continue;
4458 pud = pud_offset(pgd, 0);
4459 if (PTRS_PER_PUD > 1) /* not folded */
4460 - seq = mm_walk_set_prot(pud,flags,cpu,seq);
4461 + seq = pgd_walk_set_prot(pud,flags,cpu,seq);
4462 for (u = 0; u < PTRS_PER_PUD; u++, pud++) {
4463 if (pud_none(*pud))
4464 continue;
4465 pmd = pmd_offset(pud, 0);
4466 if (PTRS_PER_PMD > 1) /* not folded */
4467 - seq = mm_walk_set_prot(pmd,flags,cpu,seq);
4468 + seq = pgd_walk_set_prot(pmd,flags,cpu,seq);
4469 for (m = 0; m < PTRS_PER_PMD; m++, pmd++) {
4470 if (pmd_none(*pmd))
4471 continue;
4472 pte = pte_offset_kernel(pmd,0);
4473 - seq = mm_walk_set_prot(pte,flags,cpu,seq);
4474 + seq = pgd_walk_set_prot(pte,flags,cpu,seq);
4475 }
4476 }
4477 }
4478 @@ -148,12 +144,12 @@ static void mm_walk(struct mm_struct *mm
4479 seq = 0;
4480 }
4481 MULTI_update_va_mapping(mcl + seq,
4482 - (unsigned long)__user_pgd(mm->pgd),
4483 - pfn_pte(virt_to_phys(__user_pgd(mm->pgd))>>PAGE_SHIFT, flags),
4484 + (unsigned long)__user_pgd(pgd_base),
4485 + pfn_pte(virt_to_phys(__user_pgd(pgd_base))>>PAGE_SHIFT, flags),
4486 0);
4487 MULTI_update_va_mapping(mcl + seq + 1,
4488 - (unsigned long)mm->pgd,
4489 - pfn_pte(virt_to_phys(mm->pgd)>>PAGE_SHIFT, flags),
4490 + (unsigned long)pgd_base,
4491 + pfn_pte(virt_to_phys(pgd_base)>>PAGE_SHIFT, flags),
4492 UVMF_TLB_FLUSH);
4493 if (unlikely(HYPERVISOR_multicall_check(mcl, seq + 2, NULL)))
4494 BUG();
4495 @@ -161,21 +157,35 @@ static void mm_walk(struct mm_struct *mm
4496 put_cpu();
4497 }
4498
4499 +static void __pgd_pin(pgd_t *pgd)
4500 +{
4501 + pgd_walk(pgd, PAGE_KERNEL_RO);
4502 + xen_pgd_pin(__pa(pgd)); /* kernel */
4503 + xen_pgd_pin(__pa(__user_pgd(pgd))); /* user */
4504 + SetPagePinned(virt_to_page(pgd));
4505 +}
4506 +
4507 +static void __pgd_unpin(pgd_t *pgd)
4508 +{
4509 + xen_pgd_unpin(__pa(pgd));
4510 + xen_pgd_unpin(__pa(__user_pgd(pgd)));
4511 + pgd_walk(pgd, PAGE_KERNEL);
4512 + ClearPagePinned(virt_to_page(pgd));
4513 +}
4514 +
4515 +void pgd_test_and_unpin(pgd_t *pgd)
4516 +{
4517 + if (PagePinned(virt_to_page(pgd)))
4518 + __pgd_unpin(pgd);
4519 +}
4520 +
4521 void mm_pin(struct mm_struct *mm)
4522 {
4523 if (xen_feature(XENFEAT_writable_page_tables))
4524 return;
4525
4526 pin_lock(mm);
4527 -
4528 - mm_walk(mm, PAGE_KERNEL_RO);
4529 - xen_pgd_pin(__pa(mm->pgd)); /* kernel */
4530 - xen_pgd_pin(__pa(__user_pgd(mm->pgd))); /* user */
4531 - SetPagePinned(virt_to_page(mm->pgd));
4532 - spin_lock(&mm_unpinned_lock);
4533 - list_del(&mm->context.unpinned);
4534 - spin_unlock(&mm_unpinned_lock);
4535 -
4536 + __pgd_pin(mm->pgd);
4537 pin_unlock(mm);
4538 }
4539
4540 @@ -185,34 +195,30 @@ void mm_unpin(struct mm_struct *mm)
4541 return;
4542
4543 pin_lock(mm);
4544 -
4545 - xen_pgd_unpin(__pa(mm->pgd));
4546 - xen_pgd_unpin(__pa(__user_pgd(mm->pgd)));
4547 - mm_walk(mm, PAGE_KERNEL);
4548 - ClearPagePinned(virt_to_page(mm->pgd));
4549 - spin_lock(&mm_unpinned_lock);
4550 - list_add(&mm->context.unpinned, &mm_unpinned);
4551 - spin_unlock(&mm_unpinned_lock);
4552 -
4553 + __pgd_unpin(mm->pgd);
4554 pin_unlock(mm);
4555 }
4556
4557 void mm_pin_all(void)
4558 {
4559 + struct page *page;
4560 + unsigned long flags;
4561 +
4562 if (xen_feature(XENFEAT_writable_page_tables))
4563 return;
4564
4565 /*
4566 - * Allow uninterrupted access to the mm_unpinned list. We don't
4567 - * actually take the mm_unpinned_lock as it is taken inside mm_pin().
4568 + * Allow uninterrupted access to the pgd_list. Also protects
4569 + * __pgd_pin() by disabling preemption.
4570 * All other CPUs must be at a safe point (e.g., in stop_machine
4571 * or offlined entirely).
4572 */
4573 - preempt_disable();
4574 - while (!list_empty(&mm_unpinned))
4575 - mm_pin(list_entry(mm_unpinned.next, struct mm_struct,
4576 - context.unpinned));
4577 - preempt_enable();
4578 + spin_lock_irqsave(&pgd_lock, flags);
4579 + list_for_each_entry(page, &pgd_list, lru) {
4580 + if (!PagePinned(page))
4581 + __pgd_pin((pgd_t *)page_address(page));
4582 + }
4583 + spin_unlock_irqrestore(&pgd_lock, flags);
4584 }
4585
4586 void arch_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
4587 @@ -331,11 +337,11 @@ static struct page *split_large_page(uns
4588 return base;
4589 }
4590
4591 -static void cache_flush_page(void *adr)
4592 +void clflush_cache_range(void *adr, int size)
4593 {
4594 int i;
4595 - for (i = 0; i < PAGE_SIZE; i += boot_cpu_data.x86_clflush_size)
4596 - asm volatile("clflush (%0)" :: "r" (adr + i));
4597 + for (i = 0; i < size; i += boot_cpu_data.x86_clflush_size)
4598 + clflush(adr+i);
4599 }
4600
4601 static void flush_kernel_map(void *arg)
4602 @@ -350,7 +356,7 @@ static void flush_kernel_map(void *arg)
4603 asm volatile("wbinvd" ::: "memory");
4604 else list_for_each_entry(pg, l, lru) {
4605 void *adr = page_address(pg);
4606 - cache_flush_page(adr);
4607 + clflush_cache_range(adr, PAGE_SIZE);
4608 }
4609 __flush_tlb_all();
4610 }
4611 @@ -418,6 +424,7 @@ __change_page_attr(unsigned long address
4612 split = split_large_page(address, prot, ref_prot2);
4613 if (!split)
4614 return -ENOMEM;
4615 + pgprot_val(ref_prot2) &= ~_PAGE_NX;
4616 set_pte(kpte, mk_pte(split, ref_prot2));
4617 kpte_page = split;
4618 }
4619 @@ -510,9 +517,14 @@ void global_flush_tlb(void)
4620 struct page *pg, *next;
4621 struct list_head l;
4622
4623 - down_read(&init_mm.mmap_sem);
4624 + /*
4625 + * Write-protect the semaphore, to exclude two contexts
4626 + * doing a list_replace_init() call in parallel and to
4627 + * exclude new additions to the deferred_pages list:
4628 + */
4629 + down_write(&init_mm.mmap_sem);
4630 list_replace_init(&deferred_pages, &l);
4631 - up_read(&init_mm.mmap_sem);
4632 + up_write(&init_mm.mmap_sem);
4633
4634 flush_map(&l);
4635
4636 --- sle11-2009-05-14.orig/arch/x86/mm/pgtable_32-xen.c 2009-02-16 16:17:21.000000000 +0100
4637 +++ sle11-2009-05-14/arch/x86/mm/pgtable_32-xen.c 2009-02-16 16:18:36.000000000 +0100
4638 @@ -6,6 +6,7 @@
4639 #include <linux/kernel.h>
4640 #include <linux/errno.h>
4641 #include <linux/mm.h>
4642 +#include <linux/nmi.h>
4643 #include <linux/swap.h>
4644 #include <linux/smp.h>
4645 #include <linux/highmem.h>
4646 @@ -46,6 +47,8 @@ void show_mem(void)
4647 for_each_online_pgdat(pgdat) {
4648 pgdat_resize_lock(pgdat, &flags);
4649 for (i = 0; i < pgdat->node_spanned_pages; ++i) {
4650 + if (unlikely(i % MAX_ORDER_NR_PAGES == 0))
4651 + touch_nmi_watchdog();
4652 page = pgdat_page_nr(pgdat, i);
4653 total++;
4654 if (PageHighMem(page))
4655 @@ -206,7 +209,7 @@ void pte_free(struct page *pte)
4656 __free_page(pte);
4657 }
4658
4659 -void pmd_ctor(void *pmd, struct kmem_cache *cache, unsigned long flags)
4660 +void pmd_ctor(struct kmem_cache *cache, void *pmd)
4661 {
4662 memset(pmd, 0, PTRS_PER_PMD*sizeof(pmd_t));
4663 }
4664 --- sle11-2009-05-14.orig/arch/x86/pci/irq-xen.c 2009-02-16 16:17:21.000000000 +0100
4665 +++ sle11-2009-05-14/arch/x86/pci/irq-xen.c 2009-02-16 16:18:36.000000000 +0100
4666 @@ -173,7 +173,7 @@ void eisa_set_level_irq(unsigned int irq
4667 }
4668
4669 /*
4670 - * Common IRQ routing practice: nybbles in config space,
4671 + * Common IRQ routing practice: nibbles in config space,
4672 * offset by some magic constant.
4673 */
4674 static unsigned int read_config_nybble(struct pci_dev *router, unsigned offset, unsigned nr)
4675 @@ -496,6 +496,26 @@ static int pirq_amd756_set(struct pci_de
4676 return 1;
4677 }
4678
4679 +/*
4680 + * PicoPower PT86C523
4681 + */
4682 +static int pirq_pico_get(struct pci_dev *router, struct pci_dev *dev, int pirq)
4683 +{
4684 + outb(0x10 + ((pirq - 1) >> 1), 0x24);
4685 + return ((pirq - 1) & 1) ? (inb(0x26) >> 4) : (inb(0x26) & 0xf);
4686 +}
4687 +
4688 +static int pirq_pico_set(struct pci_dev *router, struct pci_dev *dev, int pirq,
4689 + int irq)
4690 +{
4691 + unsigned int x;
4692 + outb(0x10 + ((pirq - 1) >> 1), 0x24);
4693 + x = inb(0x26);
4694 + x = ((pirq - 1) & 1) ? ((x & 0x0f) | (irq << 4)) : ((x & 0xf0) | (irq));
4695 + outb(x, 0x26);
4696 + return 1;
4697 +}
4698 +
4699 #ifdef CONFIG_PCI_BIOS
4700
4701 static int pirq_bios_set(struct pci_dev *router, struct pci_dev *dev, int pirq, int irq)
4702 @@ -569,7 +589,7 @@ static __init int via_router_probe(struc
4703 /* FIXME: We should move some of the quirk fixup stuff here */
4704
4705 /*
4706 - * work arounds for some buggy BIOSes
4707 + * workarounds for some buggy BIOSes
4708 */
4709 if (device == PCI_DEVICE_ID_VIA_82C586_0) {
4710 switch(router->device) {
4711 @@ -725,6 +745,24 @@ static __init int amd_router_probe(struc
4712 return 1;
4713 }
4714
4715 +static __init int pico_router_probe(struct irq_router *r, struct pci_dev *router, u16 device)
4716 +{
4717 + switch (device) {
4718 + case PCI_DEVICE_ID_PICOPOWER_PT86C523:
4719 + r->name = "PicoPower PT86C523";
4720 + r->get = pirq_pico_get;
4721 + r->set = pirq_pico_set;
4722 + return 1;
4723 +
4724 + case PCI_DEVICE_ID_PICOPOWER_PT86C523BBP:
4725 + r->name = "PicoPower PT86C523 rev. BB+";
4726 + r->get = pirq_pico_get;
4727 + r->set = pirq_pico_set;
4728 + return 1;
4729 + }
4730 + return 0;
4731 +}
4732 +
4733 static __initdata struct irq_router_handler pirq_routers[] = {
4734 { PCI_VENDOR_ID_INTEL, intel_router_probe },
4735 { PCI_VENDOR_ID_AL, ali_router_probe },
4736 @@ -736,6 +774,7 @@ static __initdata struct irq_router_hand
4737 { PCI_VENDOR_ID_VLSI, vlsi_router_probe },
4738 { PCI_VENDOR_ID_SERVERWORKS, serverworks_router_probe },
4739 { PCI_VENDOR_ID_AMD, amd_router_probe },
4740 + { PCI_VENDOR_ID_PICOPOWER, pico_router_probe },
4741 /* Someone with docs needs to add the ATI Radeon IGP */
4742 { 0, NULL }
4743 };
4744 @@ -1014,7 +1053,7 @@ static void __init pcibios_fixup_irqs(vo
4745 * Work around broken HP Pavilion Notebooks which assign USB to
4746 * IRQ 9 even though it is actually wired to IRQ 11
4747 */
4748 -static int __init fix_broken_hp_bios_irq9(struct dmi_system_id *d)
4749 +static int __init fix_broken_hp_bios_irq9(const struct dmi_system_id *d)
4750 {
4751 if (!broken_hp_bios_irq9) {
4752 broken_hp_bios_irq9 = 1;
4753 @@ -1027,7 +1066,7 @@ static int __init fix_broken_hp_bios_irq
4754 * Work around broken Acer TravelMate 360 Notebooks which assign
4755 * Cardbus to IRQ 11 even though it is actually wired to IRQ 10
4756 */
4757 -static int __init fix_acer_tm360_irqrouting(struct dmi_system_id *d)
4758 +static int __init fix_acer_tm360_irqrouting(const struct dmi_system_id *d)
4759 {
4760 if (!acer_tm360_irqrouting) {
4761 acer_tm360_irqrouting = 1;
4762 --- sle11-2009-05-14.orig/drivers/acpi/processor_idle.c 2009-02-16 15:58:14.000000000 +0100
4763 +++ sle11-2009-05-14/drivers/acpi/processor_idle.c 2009-02-16 16:18:36.000000000 +0100
4764 @@ -1743,6 +1743,13 @@ int acpi_processor_cst_has_changed(struc
4765 if (!pr->flags.power_setup_done)
4766 return -ENODEV;
4767
4768 + if (processor_pm_external()) {
4769 + acpi_processor_get_power_info(pr);
4770 + processor_notify_external(pr,
4771 + PROCESSOR_PM_CHANGE, PM_TYPE_IDLE);
4772 + return ret;
4773 + }
4774 +
4775 cpuidle_pause_and_lock();
4776 cpuidle_disable_device(&pr->power.dev);
4777 acpi_processor_get_power_info(pr);
4778 --- sle11-2009-05-14.orig/drivers/cpuidle/Kconfig 2009-05-14 10:56:29.000000000 +0200
4779 +++ sle11-2009-05-14/drivers/cpuidle/Kconfig 2009-02-16 16:18:36.000000000 +0100
4780 @@ -1,6 +1,7 @@
4781
4782 config CPU_IDLE
4783 bool "CPU idle PM support"
4784 + depends on !PROCESSOR_EXTERNAL_CONTROL
4785 default ACPI
4786 help
4787 CPU idle is a generic framework for supporting software-controlled
4788 --- sle11-2009-05-14.orig/drivers/oprofile/cpu_buffer.c 2009-02-16 16:01:39.000000000 +0100
4789 +++ sle11-2009-05-14/drivers/oprofile/cpu_buffer.c 2009-03-12 16:15:32.000000000 +0100
4790 @@ -308,6 +308,37 @@ void oprofile_add_trace(unsigned long pc
4791 }
4792
4793 #ifdef CONFIG_XEN
4794 +/*
4795 + * This is basically log_sample(b, ESCAPE_CODE, cpu_mode, CPU_TRACE_BEGIN),
4796 + * as was previously accessible through oprofile_add_pc().
4797 + */
4798 +void oprofile_add_mode(int cpu_mode)
4799 +{
4800 + struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(cpu_buffer);
4801 + struct task_struct *task;
4802 +
4803 + if (nr_available_slots(cpu_buf) < 3) {
4804 + cpu_buf->sample_lost_overflow++;
4805 + return;
4806 + }
4807 +
4808 + task = current;
4809 +
4810 + /* notice a switch from user->kernel or vice versa */
4811 + if (cpu_buf->last_cpu_mode != cpu_mode) {
4812 + cpu_buf->last_cpu_mode = cpu_mode;
4813 + add_code(cpu_buf, cpu_mode);
4814 + }
4815 +
4816 + /* notice a task switch */
4817 + if (cpu_buf->last_task != task) {
4818 + cpu_buf->last_task = task;
4819 + add_code(cpu_buf, (unsigned long)task);
4820 + }
4821 +
4822 + add_code(cpu_buf, CPU_TRACE_BEGIN);
4823 +}
4824 +
4825 int oprofile_add_domain_switch(int32_t domain_id)
4826 {
4827 struct oprofile_cpu_buffer * cpu_buf = &cpu_buffer[smp_processor_id()];
4828 --- sle11-2009-05-14.orig/drivers/pci/msi-xen.c 2008-12-15 11:27:22.000000000 +0100
4829 +++ sle11-2009-05-14/drivers/pci/msi-xen.c 2009-02-16 16:18:36.000000000 +0100
4830 @@ -264,6 +264,12 @@ static int msi_map_vector(struct pci_dev
4831 return msi_map_pirq_to_vector(dev, -1, entry_nr, table_base);
4832 }
4833
4834 +static void pci_intx_for_msi(struct pci_dev *dev, int enable)
4835 +{
4836 + if (!(dev->dev_flags & PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG))
4837 + pci_intx(dev, enable);
4838 +}
4839 +
4840 #ifdef CONFIG_PM
4841 static void __pci_restore_msi_state(struct pci_dev *dev)
4842 {
4843 @@ -272,7 +278,7 @@ static void __pci_restore_msi_state(stru
4844 if (!dev->msi_enabled)
4845 return;
4846
4847 - pci_intx(dev, 0); /* disable intx */
4848 + pci_intx_for_msi(dev, 0);
4849 msi_set_enable(dev, 0);
4850
4851 pirq = msi_map_pirq_to_vector(dev, dev->irq, 0, 0);
4852 @@ -295,7 +301,7 @@ static void __pci_restore_msix_state(str
4853 if (!dev->msix_enabled)
4854 return;
4855
4856 - pci_intx(dev, 0); /* disable intx */
4857 + pci_intx_for_msi(dev, 0);
4858 msix_set_enable(dev, 0);
4859
4860 msi_dev_entry = get_msi_dev_pirq_list(dev);
4861 @@ -348,7 +354,7 @@ static int msi_capability_init(struct pc
4862 return -EBUSY;
4863
4864 /* Set MSI enabled bits */
4865 - pci_intx(dev, 0); /* disable intx */
4866 + pci_intx_for_msi(dev, 0);
4867 msi_set_enable(dev, 1);
4868 dev->msi_enabled = 1;
4869
4870 @@ -422,7 +428,7 @@ static int msix_capability_init(struct p
4871 return avail;
4872 }
4873
4874 - pci_intx(dev, 0); /* disable intx */
4875 + pci_intx_for_msi(dev, 0);
4876 msix_set_enable(dev, 1);
4877 dev->msix_enabled = 1;
4878
4879 @@ -557,7 +563,7 @@ void pci_disable_msi(struct pci_dev* dev
4880
4881 /* Disable MSI mode */
4882 msi_set_enable(dev, 0);
4883 - pci_intx(dev, 1); /* enable intx */
4884 + pci_intx_for_msi(dev, 1);
4885 dev->msi_enabled = 0;
4886 }
4887 EXPORT_SYMBOL(pci_disable_msi);
4888 @@ -696,7 +702,7 @@ void pci_disable_msix(struct pci_dev* de
4889
4890 /* Disable MSI mode */
4891 msix_set_enable(dev, 0);
4892 - pci_intx(dev, 1); /* enable intx */
4893 + pci_intx_for_msi(dev, 1);
4894 dev->msix_enabled = 0;
4895 }
4896 EXPORT_SYMBOL(pci_disable_msix);
4897 --- sle11-2009-05-14.orig/drivers/xen/blkback/blkback.c 2009-02-16 16:17:21.000000000 +0100
4898 +++ sle11-2009-05-14/drivers/xen/blkback/blkback.c 2009-02-16 16:18:36.000000000 +0100
4899 @@ -269,13 +269,10 @@ static void __end_block_io_op(pending_re
4900 }
4901 }
4902
4903 -static int end_block_io_op(struct bio *bio, unsigned int done, int error)
4904 +static void end_block_io_op(struct bio *bio, int error)
4905 {
4906 - if (bio->bi_size != 0)
4907 - return 1;
4908 __end_block_io_op(bio->bi_private, error);
4909 bio_put(bio);
4910 - return error;
4911 }
4912
4913
4914 --- sle11-2009-05-14.orig/drivers/xen/blkfront/blkfront.c 2009-03-24 10:12:03.000000000 +0100
4915 +++ sle11-2009-05-14/drivers/xen/blkfront/blkfront.c 2009-02-16 16:18:36.000000000 +0100
4916 @@ -233,7 +233,7 @@ static int setup_blkring(struct xenbus_d
4917 SHARED_RING_INIT(sring);
4918 FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
4919
4920 - memset(info->sg, 0, sizeof(info->sg));
4921 + sg_init_table(info->sg, BLKIF_MAX_SEGMENTS_PER_REQUEST);
4922
4923 err = xenbus_grant_ring(dev, virt_to_mfn(info->ring.sring));
4924 if (err < 0) {
4925 @@ -625,9 +625,8 @@ static int blkif_queue_request(struct re
4926
4927 ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg);
4928 BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);
4929 - for (i = 0; i < ring_req->nr_segments; ++i) {
4930 - sg = info->sg + i;
4931 - buffer_mfn = page_to_phys(sg->page) >> PAGE_SHIFT;
4932 + for_each_sg(info->sg, sg, ring_req->nr_segments, i) {
4933 + buffer_mfn = page_to_phys(sg_page(sg)) >> PAGE_SHIFT;
4934 fsect = sg->offset >> 9;
4935 lsect = fsect + (sg->length >> 9) - 1;
4936 /* install a grant reference. */
4937 --- sle11-2009-05-14.orig/drivers/xen/core/firmware.c 2009-05-14 10:56:29.000000000 +0200
4938 +++ sle11-2009-05-14/drivers/xen/core/firmware.c 2009-03-25 18:10:23.000000000 +0100
4939 @@ -1,4 +1,5 @@
4940 #include <linux/kernel.h>
4941 +#include <linux/string.h>
4942 #include <linux/errno.h>
4943 #include <linux/init.h>
4944 #include <linux/edd.h>
4945 --- sle11-2009-05-14.orig/drivers/xen/core/machine_kexec.c 2009-05-14 10:56:29.000000000 +0200
4946 +++ sle11-2009-05-14/drivers/xen/core/machine_kexec.c 2009-02-17 11:46:41.000000000 +0100
4947 @@ -29,6 +29,10 @@ void __init xen_machine_kexec_setup_reso
4948 int k = 0;
4949 int rc;
4950
4951 + if (strstr(boot_command_line, "crashkernel="))
4952 + printk(KERN_WARNING "Ignoring crashkernel command line, "
4953 + "parameter will be supplied by xen\n");
4954 +
4955 if (!is_initial_xendomain())
4956 return;
4957
4958 @@ -130,6 +134,13 @@ void __init xen_machine_kexec_setup_reso
4959 xen_max_nr_phys_cpus))
4960 goto err;
4961
4962 +#ifdef CONFIG_X86
4963 + if (xen_create_contiguous_region((unsigned long)&vmcoreinfo_note,
4964 + get_order(sizeof(vmcoreinfo_note)),
4965 + BITS_PER_LONG))
4966 + goto err;
4967 +#endif
4968 +
4969 return;
4970
4971 err:
4972 @@ -205,6 +216,13 @@ NORET_TYPE void machine_kexec(struct kim
4973 panic("KEXEC_CMD_kexec hypercall should not return\n");
4974 }
4975
4976 +#ifdef CONFIG_X86
4977 +unsigned long paddr_vmcoreinfo_note(void)
4978 +{
4979 + return virt_to_machine(&vmcoreinfo_note);
4980 +}
4981 +#endif
4982 +
4983 void machine_shutdown(void)
4984 {
4985 /* do nothing */
4986 --- sle11-2009-05-14.orig/drivers/xen/core/smpboot.c 2008-12-15 11:27:22.000000000 +0100
4987 +++ sle11-2009-05-14/drivers/xen/core/smpboot.c 2009-02-16 16:18:36.000000000 +0100
4988 @@ -45,8 +45,8 @@ cpumask_t cpu_possible_map;
4989 EXPORT_SYMBOL(cpu_possible_map);
4990 cpumask_t cpu_initialized_map;
4991
4992 -struct cpuinfo_x86 cpu_data[NR_CPUS] __cacheline_aligned;
4993 -EXPORT_SYMBOL(cpu_data);
4994 +DEFINE_PER_CPU(struct cpuinfo_x86, cpu_info);
4995 +EXPORT_PER_CPU_SYMBOL(cpu_info);
4996
4997 static DEFINE_PER_CPU(int, resched_irq);
4998 static DEFINE_PER_CPU(int, callfunc_irq);
4999 @@ -55,13 +55,13 @@ static char callfunc_name[NR_CPUS][15];
5000
5001 u8 cpu_2_logical_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
5002
5003 -cpumask_t cpu_sibling_map[NR_CPUS] __cacheline_aligned;
5004 -cpumask_t cpu_core_map[NR_CPUS] __cacheline_aligned;
5005 -EXPORT_SYMBOL(cpu_core_map);
5006 +DEFINE_PER_CPU(cpumask_t, cpu_sibling_map);
5007 +DEFINE_PER_CPU(cpumask_t, cpu_core_map);
5008 +EXPORT_PER_CPU_SYMBOL(cpu_core_map);
5009
5010 #if defined(__i386__)
5011 -u8 x86_cpu_to_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = 0xff };
5012 -EXPORT_SYMBOL(x86_cpu_to_apicid);
5013 +DEFINE_PER_CPU(u8, x86_cpu_to_apicid) = BAD_APICID;
5014 +EXPORT_PER_CPU_SYMBOL(x86_cpu_to_apicid);
5015 #endif
5016
5017 void __init prefill_possible_map(void)
5018 @@ -86,25 +86,25 @@ void __init smp_alloc_memory(void)
5019 static inline void
5020 set_cpu_sibling_map(unsigned int cpu)
5021 {
5022 - cpu_data[cpu].phys_proc_id = cpu;
5023 - cpu_data[cpu].cpu_core_id = 0;
5024 + cpu_data(cpu).phys_proc_id = cpu;
5025 + cpu_data(cpu).cpu_core_id = 0;
5026
5027 - cpu_sibling_map[cpu] = cpumask_of_cpu(cpu);
5028 - cpu_core_map[cpu] = cpumask_of_cpu(cpu);
5029 + per_cpu(cpu_sibling_map, cpu) = cpumask_of_cpu(cpu);
5030 + per_cpu(cpu_core_map, cpu) = cpumask_of_cpu(cpu);
5031
5032 - cpu_data[cpu].booted_cores = 1;
5033 + cpu_data(cpu).booted_cores = 1;
5034 }
5035
5036 static void
5037 remove_siblinginfo(unsigned int cpu)
5038 {
5039 - cpu_data[cpu].phys_proc_id = BAD_APICID;
5040 - cpu_data[cpu].cpu_core_id = BAD_APICID;
5041 + cpu_data(cpu).phys_proc_id = BAD_APICID;
5042 + cpu_data(cpu).cpu_core_id = BAD_APICID;
5043
5044 - cpus_clear(cpu_sibling_map[cpu]);
5045 - cpus_clear(cpu_core_map[cpu]);
5046 + cpus_clear(per_cpu(cpu_sibling_map, cpu));
5047 + cpus_clear(per_cpu(cpu_core_map, cpu));
5048
5049 - cpu_data[cpu].booted_cores = 0;
5050 + cpu_data(cpu).booted_cores = 0;
5051 }
5052
5053 static int __cpuinit xen_smp_intr_init(unsigned int cpu)
5054 @@ -163,9 +163,9 @@ void __cpuinit cpu_bringup(void)
5055 {
5056 cpu_init();
5057 #ifdef __i386__
5058 - identify_secondary_cpu(cpu_data + smp_processor_id());
5059 + identify_secondary_cpu(&current_cpu_data);
5060 #else
5061 - identify_cpu(cpu_data + smp_processor_id());
5062 + identify_cpu(&current_cpu_data);
5063 #endif
5064 touch_softlockup_watchdog();
5065 preempt_disable();
5066 @@ -266,16 +266,16 @@ void __init smp_prepare_cpus(unsigned in
5067 if (HYPERVISOR_vcpu_op(VCPUOP_get_physid, 0, &cpu_id) == 0)
5068 apicid = xen_vcpu_physid_to_x86_apicid(cpu_id.phys_id);
5069 boot_cpu_data.apicid = apicid;
5070 - cpu_data[0] = boot_cpu_data;
5071 + cpu_data(0) = boot_cpu_data;
5072
5073 cpu_2_logical_apicid[0] = apicid;
5074 - x86_cpu_to_apicid[0] = apicid;
5075 + per_cpu(x86_cpu_to_apicid, 0) = apicid;
5076
5077 current_thread_info()->cpu = 0;
5078
5079 for (cpu = 0; cpu < NR_CPUS; cpu++) {
5080 - cpus_clear(cpu_sibling_map[cpu]);
5081 - cpus_clear(cpu_core_map[cpu]);
5082 + cpus_clear(per_cpu(cpu_sibling_map, cpu));
5083 + cpus_clear(per_cpu(cpu_core_map, cpu));
5084 }
5085
5086 set_cpu_sibling_map(0);
5087 @@ -320,11 +320,12 @@ void __init smp_prepare_cpus(unsigned in
5088 apicid = cpu;
5089 if (HYPERVISOR_vcpu_op(VCPUOP_get_physid, cpu, &cpu_id) == 0)
5090 apicid = xen_vcpu_physid_to_x86_apicid(cpu_id.phys_id);
5091 - cpu_data[cpu] = boot_cpu_data;
5092 - cpu_data[cpu].apicid = apicid;
5093 + cpu_data(cpu) = boot_cpu_data;
5094 + cpu_data(cpu).cpu_index = cpu;
5095 + cpu_data(cpu).apicid = apicid;
5096
5097 cpu_2_logical_apicid[cpu] = apicid;
5098 - x86_cpu_to_apicid[cpu] = apicid;
5099 + per_cpu(x86_cpu_to_apicid, cpu) = apicid;
5100
5101 #ifdef __x86_64__
5102 cpu_pda(cpu)->pcurrent = idle;
5103 --- sle11-2009-05-14.orig/drivers/xen/netback/loopback.c 2008-12-15 11:26:44.000000000 +0100
5104 +++ sle11-2009-05-14/drivers/xen/netback/loopback.c 2009-02-16 16:18:36.000000000 +0100
5105 @@ -285,9 +285,9 @@ static void __exit clean_loopback(int i)
5106 char dev_name[IFNAMSIZ];
5107
5108 sprintf(dev_name, "vif0.%d", i);
5109 - dev1 = dev_get_by_name(dev_name);
5110 + dev1 = dev_get_by_name(&init_net, dev_name);
5111 sprintf(dev_name, "veth%d", i);
5112 - dev2 = dev_get_by_name(dev_name);
5113 + dev2 = dev_get_by_name(&init_net, dev_name);
5114 if (dev1 && dev2) {
5115 unregister_netdev(dev2);
5116 unregister_netdev(dev1);
5117 --- sle11-2009-05-14.orig/drivers/xen/netback/netback.c 2008-12-23 09:33:22.000000000 +0100
5118 +++ sle11-2009-05-14/drivers/xen/netback/netback.c 2009-02-16 16:18:36.000000000 +0100
5119 @@ -350,8 +350,8 @@ static void xen_network_done_notify(void
5120 {
5121 static struct net_device *eth0_dev = NULL;
5122 if (unlikely(eth0_dev == NULL))
5123 - eth0_dev = __dev_get_by_name("eth0");
5124 - netif_rx_schedule(eth0_dev);
5125 + eth0_dev = __dev_get_by_name(&init_net, "eth0");
5126 + netif_rx_schedule(eth0_dev, ???);
5127 }
5128 /*
5129 * Add following to poll() function in NAPI driver (Tigon3 is example):
5130 --- sle11-2009-05-14.orig/drivers/xen/netback/xenbus.c 2009-03-04 11:25:55.000000000 +0100
5131 +++ sle11-2009-05-14/drivers/xen/netback/xenbus.c 2009-02-16 16:18:36.000000000 +0100
5132 @@ -149,12 +149,10 @@ fail:
5133 * and vif variables to the environment, for the benefit of the vif-* hotplug
5134 * scripts.
5135 */
5136 -static int netback_uevent(struct xenbus_device *xdev, char **envp,
5137 - int num_envp, char *buffer, int buffer_size)
5138 +static int netback_uevent(struct xenbus_device *xdev, struct kobj_uevent_env *env)
5139 {
5140 struct backend_info *be = xdev->dev.driver_data;
5141 netif_t *netif = be->netif;
5142 - int i = 0, length = 0;
5143 char *val;
5144
5145 DPRINTK("netback_uevent");
5146 @@ -166,15 +164,11 @@ static int netback_uevent(struct xenbus_
5147 return err;
5148 }
5149 else {
5150 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size,
5151 - &length, "script=%s", val);
5152 + add_uevent_var(env, "script=%s", val);
5153 kfree(val);
5154 }
5155
5156 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5157 - "vif=%s", netif->dev->name);
5158 -
5159 - envp[i] = NULL;
5160 + add_uevent_var(env, "vif=%s", netif->dev->name);
5161
5162 return 0;
5163 }
5164 --- sle11-2009-05-14.orig/drivers/xen/netfront/accel.c 2009-04-09 14:43:45.000000000 +0200
5165 +++ sle11-2009-05-14/drivers/xen/netfront/accel.c 2009-03-30 16:39:19.000000000 +0200
5166 @@ -313,7 +313,7 @@ accelerator_set_vif_state_hooks(struct n
5167 DPRINTK("%p\n",vif_state);
5168
5169 /* Make sure there are no data path operations going on */
5170 - netif_poll_disable(vif_state->np->netdev);
5171 + napi_disable(&vif_state->np->napi);
5172 netif_tx_lock_bh(vif_state->np->netdev);
5173
5174 accelerator = vif_state->np->accelerator;
5175 @@ -322,7 +322,7 @@ accelerator_set_vif_state_hooks(struct n
5176 spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
5177
5178 netif_tx_unlock_bh(vif_state->np->netdev);
5179 - netif_poll_enable(vif_state->np->netdev);
5180 + napi_enable(&vif_state->np->napi);
5181 }
5182
5183
5184 @@ -496,7 +496,7 @@ accelerator_remove_single_hook(struct ne
5185 unsigned long flags;
5186
5187 /* Make sure there are no data path operations going on */
5188 - netif_poll_disable(vif_state->np->netdev);
5189 + napi_disable(&vif_state->np->napi);
5190 netif_tx_lock_bh(vif_state->np->netdev);
5191
5192 spin_lock_irqsave(&accelerator->vif_states_lock, flags);
5193 @@ -512,7 +512,7 @@ accelerator_remove_single_hook(struct ne
5194 spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
5195
5196 netif_tx_unlock_bh(vif_state->np->netdev);
5197 - netif_poll_enable(vif_state->np->netdev);
5198 + napi_enable(&vif_state->np->napi);
5199 }
5200
5201
5202 --- sle11-2009-05-14.orig/drivers/xen/netfront/netfront.c 2009-03-30 16:36:30.000000000 +0200
5203 +++ sle11-2009-05-14/drivers/xen/netfront/netfront.c 2009-03-30 16:39:44.000000000 +0200
5204 @@ -626,6 +626,7 @@ static int network_open(struct net_devic
5205 struct netfront_info *np = netdev_priv(dev);
5206
5207 memset(&np->stats, 0, sizeof(np->stats));
5208 + napi_enable(&np->napi);
5209
5210 spin_lock_bh(&np->rx_lock);
5211 if (netfront_carrier_ok(np)) {
5212 @@ -634,7 +635,7 @@ static int network_open(struct net_devic
5213 if (RING_HAS_UNCONSUMED_RESPONSES(&np->rx)){
5214 netfront_accelerator_call_stop_napi_irq(np, dev);
5215
5216 - netif_rx_schedule(dev);
5217 + netif_rx_schedule(dev, &np->napi);
5218 }
5219 }
5220 spin_unlock_bh(&np->rx_lock);
5221 @@ -706,7 +707,7 @@ static void rx_refill_timeout(unsigned l
5222
5223 netfront_accelerator_call_stop_napi_irq(np, dev);
5224
5225 - netif_rx_schedule(dev);
5226 + netif_rx_schedule(dev, &np->napi);
5227 }
5228
5229 static void network_alloc_rx_buffers(struct net_device *dev)
5230 @@ -1063,7 +1064,7 @@ static irqreturn_t netif_int(int irq, vo
5231 if (RING_HAS_UNCONSUMED_RESPONSES(&np->rx)) {
5232 netfront_accelerator_call_stop_napi_irq(np, dev);
5233
5234 - netif_rx_schedule(dev);
5235 + netif_rx_schedule(dev, &np->napi);
5236 dev->last_rx = jiffies;
5237 }
5238 }
5239 @@ -1316,16 +1317,17 @@ static int xennet_set_skb_gso(struct sk_
5240 #endif
5241 }
5242
5243 -static int netif_poll(struct net_device *dev, int *pbudget)
5244 +static int netif_poll(struct napi_struct *napi, int budget)
5245 {
5246 - struct netfront_info *np = netdev_priv(dev);
5247 + struct netfront_info *np = container_of(napi, struct netfront_info, napi);
5248 + struct net_device *dev = np->netdev;
5249 struct sk_buff *skb;
5250 struct netfront_rx_info rinfo;
5251 struct netif_rx_response *rx = &rinfo.rx;
5252 struct netif_extra_info *extras = rinfo.extras;
5253 RING_IDX i, rp;
5254 struct multicall_entry *mcl;
5255 - int work_done, budget, more_to_do = 1, accel_more_to_do = 1;
5256 + int work_done, more_to_do = 1, accel_more_to_do = 1;
5257 struct sk_buff_head rxq;
5258 struct sk_buff_head errq;
5259 struct sk_buff_head tmpq;
5260 @@ -1345,8 +1347,6 @@ static int netif_poll(struct net_device
5261 skb_queue_head_init(&errq);
5262 skb_queue_head_init(&tmpq);
5263
5264 - if ((budget = *pbudget) > dev->quota)
5265 - budget = dev->quota;
5266 rp = np->rx.sring->rsp_prod;
5267 rmb(); /* Ensure we see queued responses up to 'rp'. */
5268
5269 @@ -1508,9 +1508,6 @@ err:
5270 accel_more_to_do = 0;
5271 }
5272
5273 - *pbudget -= work_done;
5274 - dev->quota -= work_done;
5275 -
5276 if (work_done < budget) {
5277 local_irq_save(flags);
5278
5279 @@ -1527,14 +1524,14 @@ err:
5280 }
5281
5282 if (!more_to_do && !accel_more_to_do)
5283 - __netif_rx_complete(dev);
5284 + __netif_rx_complete(dev, napi);
5285
5286 local_irq_restore(flags);
5287 }
5288
5289 spin_unlock(&np->rx_lock);
5290
5291 - return more_to_do | accel_more_to_do;
5292 + return work_done;
5293 }
5294
5295 static void netif_release_tx_bufs(struct netfront_info *np)
5296 @@ -1681,6 +1678,7 @@ static int network_close(struct net_devi
5297 {
5298 struct netfront_info *np = netdev_priv(dev);
5299 netif_stop_queue(np->netdev);
5300 + napi_disable(&np->napi);
5301 return 0;
5302 }
5303
5304 @@ -2088,16 +2086,14 @@ static struct net_device * __devinit cre
5305 netdev->hard_start_xmit = network_start_xmit;
5306 netdev->stop = network_close;
5307 netdev->get_stats = network_get_stats;
5308 - netdev->poll = netif_poll;
5309 + netif_napi_add(netdev, &np->napi, netif_poll, 64);
5310 netdev->set_multicast_list = network_set_multicast_list;
5311 netdev->uninit = netif_uninit;
5312 netdev->set_mac_address = xennet_set_mac_address;
5313 netdev->change_mtu = xennet_change_mtu;
5314 - netdev->weight = 64;
5315 netdev->features = NETIF_F_IP_CSUM;
5316
5317 SET_ETHTOOL_OPS(netdev, &network_ethtool_ops);
5318 - SET_MODULE_OWNER(netdev);
5319 SET_NETDEV_DEV(netdev, &dev->dev);
5320
5321 np->netdev = netdev;
5322 --- sle11-2009-05-14.orig/drivers/xen/netfront/netfront.h 2009-05-14 10:56:29.000000000 +0200
5323 +++ sle11-2009-05-14/drivers/xen/netfront/netfront.h 2009-02-16 16:18:36.000000000 +0100
5324 @@ -157,6 +157,8 @@ struct netfront_info {
5325 spinlock_t tx_lock;
5326 spinlock_t rx_lock;
5327
5328 + struct napi_struct napi;
5329 +
5330 unsigned int irq;
5331 unsigned int copying_receiver;
5332 unsigned int carrier;
5333 --- sle11-2009-05-14.orig/drivers/xen/pciback/Makefile 2009-05-14 10:56:29.000000000 +0200
5334 +++ sle11-2009-05-14/drivers/xen/pciback/Makefile 2009-02-16 16:18:36.000000000 +0100
5335 @@ -12,6 +12,4 @@ pciback-$(CONFIG_XEN_PCIDEV_BACKEND_SLOT
5336 pciback-$(CONFIG_XEN_PCIDEV_BACKEND_PASS) += passthrough.o
5337 pciback-$(CONFIG_XEN_PCIDEV_BACKEND_CONTROLLER) += controller.o
5338
5339 -ifeq ($(CONFIG_XEN_PCIDEV_BE_DEBUG),y)
5340 -EXTRA_CFLAGS += -DDEBUG
5341 -endif
5342 +ccflags-$(CONFIG_XEN_PCIDEV_BE_DEBUG) += -DDEBUG
5343 --- sle11-2009-05-14.orig/drivers/xen/pcifront/Makefile 2009-05-14 10:56:29.000000000 +0200
5344 +++ sle11-2009-05-14/drivers/xen/pcifront/Makefile 2009-02-16 16:18:36.000000000 +0100
5345 @@ -2,6 +2,4 @@ obj-y += pcifront.o
5346
5347 pcifront-y := pci_op.o xenbus.o pci.o
5348
5349 -ifeq ($(CONFIG_XEN_PCIDEV_FE_DEBUG),y)
5350 -EXTRA_CFLAGS += -DDEBUG
5351 -endif
5352 +ccflags-$(CONFIG_XEN_PCIDEV_FE_DEBUG) += -DDEBUG
5353 --- sle11-2009-05-14.orig/drivers/xen/scsiback/emulate.c 2009-05-14 10:56:29.000000000 +0200
5354 +++ sle11-2009-05-14/drivers/xen/scsiback/emulate.c 2009-02-16 16:18:36.000000000 +0100
5355 @@ -104,9 +104,10 @@ static void resp_not_supported_cmd(pendi
5356 }
5357
5358
5359 -static int __copy_to_sg(struct scatterlist *sg, unsigned int nr_sg,
5360 +static int __copy_to_sg(struct scatterlist *sgl, unsigned int nr_sg,
5361 void *buf, unsigned int buflen)
5362 {
5363 + struct scatterlist *sg;
5364 void *from = buf;
5365 void *to;
5366 unsigned int from_rest = buflen;
5367 @@ -115,8 +116,8 @@ static int __copy_to_sg(struct scatterli
5368 unsigned int i;
5369 unsigned long pfn;
5370
5371 - for (i = 0; i < nr_sg; i++) {
5372 - if (sg->page == NULL) {
5373 + for_each_sg (sgl, sg, nr_sg, i) {
5374 + if (sg_page(sg) == NULL) {
5375 printk(KERN_WARNING "%s: inconsistent length field in "
5376 "scatterlist\n", __FUNCTION__);
5377 return -ENOMEM;
5378 @@ -125,7 +126,7 @@ static int __copy_to_sg(struct scatterli
5379 to_capa = sg->length;
5380 copy_size = min_t(unsigned int, to_capa, from_rest);
5381
5382 - pfn = page_to_pfn(sg->page);
5383 + pfn = page_to_pfn(sg_page(sg));
5384 to = pfn_to_kaddr(pfn) + (sg->offset);
5385 memcpy(to, from, copy_size);
5386
5387 @@ -134,7 +135,6 @@ static int __copy_to_sg(struct scatterli
5388 return 0;
5389 }
5390
5391 - sg++;
5392 from += copy_size;
5393 }
5394
5395 @@ -143,9 +143,10 @@ static int __copy_to_sg(struct scatterli
5396 return -ENOMEM;
5397 }
5398
5399 -static int __copy_from_sg(struct scatterlist *sg, unsigned int nr_sg,
5400 +static int __copy_from_sg(struct scatterlist *sgl, unsigned int nr_sg,
5401 void *buf, unsigned int buflen)
5402 {
5403 + struct scatterlist *sg;
5404 void *from;
5405 void *to = buf;
5406 unsigned int from_rest;
5407 @@ -154,8 +155,8 @@ static int __copy_from_sg(struct scatter
5408 unsigned int i;
5409 unsigned long pfn;
5410
5411 - for (i = 0; i < nr_sg; i++) {
5412 - if (sg->page == NULL) {
5413 + for_each_sg (sgl, sg, nr_sg, i) {
5414 + if (sg_page(sg) == NULL) {
5415 printk(KERN_WARNING "%s: inconsistent length field in "
5416 "scatterlist\n", __FUNCTION__);
5417 return -ENOMEM;
5418 @@ -170,13 +171,11 @@ static int __copy_from_sg(struct scatter
5419 }
5420 copy_size = from_rest;
5421
5422 - pfn = page_to_pfn(sg->page);
5423 + pfn = page_to_pfn(sg_page(sg));
5424 from = pfn_to_kaddr(pfn) + (sg->offset);
5425 memcpy(to, from, copy_size);
5426
5427 to_capa -= copy_size;
5428 -
5429 - sg++;
5430 to += copy_size;
5431 }
5432
5433 --- sle11-2009-05-14.orig/drivers/xen/scsiback/scsiback.c 2008-12-15 11:26:44.000000000 +0100
5434 +++ sle11-2009-05-14/drivers/xen/scsiback/scsiback.c 2009-02-16 16:18:36.000000000 +0100
5435 @@ -247,6 +247,8 @@ static int scsiback_gnttab_data_map(vscs
5436 write = (data_dir == DMA_TO_DEVICE);
5437
5438 if (nr_segments) {
5439 + struct scatterlist *sg;
5440 +
5441 /* free of (sgl) in fast_flush_area()*/
5442 pending_req->sgl = kmalloc(sizeof(struct scatterlist) * nr_segments,
5443 GFP_KERNEL);
5444 @@ -255,6 +257,8 @@ static int scsiback_gnttab_data_map(vscs
5445 return -ENOMEM;
5446 }
5447
5448 + sg_init_table(pending_req->sgl, nr_segments);
5449 +
5450 for (i = 0; i < nr_segments; i++) {
5451 flags = GNTMAP_host_map;
5452 if (write)
5453 @@ -267,7 +271,7 @@ static int scsiback_gnttab_data_map(vscs
5454 err = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nr_segments);
5455 BUG_ON(err);
5456
5457 - for (i = 0; i < nr_segments; i++) {
5458 + for_each_sg (pending_req->sgl, sg, nr_segments, i) {
5459 if (unlikely(map[i].status != 0)) {
5460 printk(KERN_ERR "scsiback: invalid buffer -- could not remap it\n");
5461 map[i].handle = SCSIBACK_INVALID_HANDLE;
5462 @@ -283,15 +287,15 @@ static int scsiback_gnttab_data_map(vscs
5463 pending_req, i)) >> PAGE_SHIFT,
5464 FOREIGN_FRAME(map[i].dev_bus_addr >> PAGE_SHIFT));
5465
5466 - pending_req->sgl[i].page = virt_to_page(vaddr(pending_req, i));
5467 - pending_req->sgl[i].offset = ring_req->seg[i].offset;
5468 - pending_req->sgl[i].length = ring_req->seg[i].length;
5469 - data_len += pending_req->sgl[i].length;
5470 + sg_set_page(sg, virt_to_page(vaddr(pending_req, i)),
5471 + ring_req->seg[i].length,
5472 + ring_req->seg[i].offset);
5473 + data_len += sg->length;
5474
5475 barrier();
5476 - if (pending_req->sgl[i].offset >= PAGE_SIZE ||
5477 - pending_req->sgl[i].length > PAGE_SIZE ||
5478 - pending_req->sgl[i].offset + pending_req->sgl[i].length > PAGE_SIZE)
5479 + if (sg->offset >= PAGE_SIZE ||
5480 + sg->length > PAGE_SIZE ||
5481 + sg->offset + sg->length > PAGE_SIZE)
5482 err |= 1;
5483
5484 }
5485 @@ -320,27 +324,14 @@ static int scsiback_merge_bio(struct req
5486
5487 blk_queue_bounce(q, &bio);
5488
5489 - if (!rq->bio)
5490 - blk_rq_bio_prep(q, rq, bio);
5491 - else if (!ll_back_merge_fn(q, rq, bio))
5492 - return -EINVAL;
5493 - else {
5494 - rq->biotail->bi_next = bio;
5495 - rq->biotail = bio;
5496 - }
5497 -
5498 - return 0;
5499 + return blk_rq_append_bio(q, rq, bio);
5500 }
5501
5502
5503 /* quoted scsi_lib.c/scsi_bi_endio */
5504 -static int scsiback_bi_endio(struct bio *bio, unsigned int bytes_done, int error)
5505 +static void scsiback_bi_endio(struct bio *bio, int error)
5506 {
5507 - if (bio->bi_size)
5508 - return 1;
5509 -
5510 bio_put(bio);
5511 - return 0;
5512 }
5513
5514
5515 @@ -351,16 +342,16 @@ static int request_map_sg(struct request
5516 struct request_queue *q = rq->q;
5517 int nr_pages;
5518 unsigned int nsegs = count;
5519 -
5520 unsigned int data_len = 0, len, bytes, off;
5521 + struct scatterlist *sg;
5522 struct page *page;
5523 struct bio *bio = NULL;
5524 int i, err, nr_vecs = 0;
5525
5526 - for (i = 0; i < nsegs; i++) {
5527 - page = pending_req->sgl[i].page;
5528 - off = (unsigned int)pending_req->sgl[i].offset;
5529 - len = (unsigned int)pending_req->sgl[i].length;
5530 + for_each_sg (pending_req->sgl, sg, nsegs, i) {
5531 + page = sg_page(sg);
5532 + off = sg->offset;
5533 + len = sg->length;
5534 data_len += len;
5535
5536 nr_pages = (len + off + PAGE_SIZE - 1) >> PAGE_SHIFT;
5537 @@ -388,7 +379,7 @@ static int request_map_sg(struct request
5538 if (bio->bi_vcnt >= nr_vecs) {
5539 err = scsiback_merge_bio(rq, bio);
5540 if (err) {
5541 - bio_endio(bio, bio->bi_size, 0);
5542 + bio_endio(bio, 0);
5543 goto free_bios;
5544 }
5545 bio = NULL;
5546 @@ -411,7 +402,7 @@ free_bios:
5547 /*
5548 * call endio instead of bio_put incase it was bounced
5549 */
5550 - bio_endio(bio, bio->bi_size, 0);
5551 + bio_endio(bio, 0);
5552 }
5553
5554 return err;
5555 --- sle11-2009-05-14.orig/drivers/xen/scsifront/scsifront.c 2009-02-16 16:17:21.000000000 +0100
5556 +++ sle11-2009-05-14/drivers/xen/scsifront/scsifront.c 2009-02-16 16:18:36.000000000 +0100
5557 @@ -246,11 +246,10 @@ static int map_data_for_request(struct v
5558 {
5559 grant_ref_t gref_head;
5560 struct page *page;
5561 - int err, i, ref, ref_cnt = 0;
5562 + int err, ref, ref_cnt = 0;
5563 int write = (sc->sc_data_direction == DMA_TO_DEVICE);
5564 - int nr_pages, off, len, bytes;
5565 + unsigned int i, nr_pages, off, len, bytes;
5566 unsigned long buffer_pfn;
5567 - unsigned int data_len = 0;
5568
5569 if (sc->sc_data_direction == DMA_NONE)
5570 return 0;
5571 @@ -263,25 +262,31 @@ static int map_data_for_request(struct v
5572
5573 if (sc->use_sg) {
5574 /* quoted scsi_lib.c/scsi_req_map_sg . */
5575 - struct scatterlist *sg = (struct scatterlist *)sc->request_buffer;
5576 - nr_pages = (sc->request_bufflen + sg[0].offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
5577 + struct scatterlist *sg, *sgl = (struct scatterlist *)sc->request_buffer;
5578 + unsigned int data_len = sc->request_bufflen;
5579
5580 + nr_pages = (sc->request_bufflen + sgl->offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
5581 if (nr_pages > VSCSIIF_SG_TABLESIZE) {
5582 printk(KERN_ERR "scsifront: Unable to map request_buffer for command!\n");
5583 ref_cnt = (-E2BIG);
5584 goto big_to_sg;
5585 }
5586
5587 - for (i = 0; i < sc->use_sg; i++) {
5588 - page = sg[i].page;
5589 - off = sg[i].offset;
5590 - len = sg[i].length;
5591 - data_len += len;
5592 + for_each_sg (sgl, sg, sc->use_sg, i) {
5593 + page = sg_page(sg);
5594 + off = sg->offset;
5595 + len = sg->length;
5596
5597 buffer_pfn = page_to_phys(page) >> PAGE_SHIFT;
5598
5599 - while (len > 0) {
5600 + while (len > 0 && data_len > 0) {
5601 + /*
5602 + * sg sends a scatterlist that is larger than
5603 + * the data_len it wants transferred for certain
5604 + * IO sizes
5605 + */
5606 bytes = min_t(unsigned int, len, PAGE_SIZE - off);
5607 + bytes = min(bytes, data_len);
5608
5609 ref = gnttab_claim_grant_reference(&gref_head);
5610 BUG_ON(ref == -ENOSPC);
5611 @@ -296,6 +301,7 @@ static int map_data_for_request(struct v
5612
5613 buffer_pfn++;
5614 len -= bytes;
5615 + data_len -= bytes;
5616 off = 0;
5617 ref_cnt++;
5618 }
5619 --- sle11-2009-05-14.orig/drivers/xen/sfc_netback/accel_fwd.c 2008-12-15 11:27:22.000000000 +0100
5620 +++ sle11-2009-05-14/drivers/xen/sfc_netback/accel_fwd.c 2009-02-16 16:18:36.000000000 +0100
5621 @@ -181,10 +181,11 @@ int netback_accel_fwd_add(const __u8 *ma
5622 unsigned long flags;
5623 cuckoo_hash_mac_key key = cuckoo_mac_to_key(mac);
5624 struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
5625 + DECLARE_MAC_BUF(buf);
5626
5627 BUG_ON(fwd_priv == NULL);
5628
5629 - DPRINTK("Adding mac " MAC_FMT "\n", MAC_ARG(mac));
5630 + DPRINTK("Adding mac %s\n", print_mac(buf, mac));
5631
5632 spin_lock_irqsave(&fwd_set->fwd_lock, flags);
5633
5634 @@ -199,8 +200,8 @@ int netback_accel_fwd_add(const __u8 *ma
5635 if (cuckoo_hash_lookup(&fwd_set->fwd_hash_table,
5636 (cuckoo_hash_key *)(&key), &rc) != 0) {
5637 spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
5638 - EPRINTK("MAC address " MAC_FMT " already accelerated.\n",
5639 - MAC_ARG(mac));
5640 + EPRINTK("MAC address %s already accelerated.\n",
5641 + print_mac(buf, mac));
5642 return -EEXIST;
5643 }
5644
5645 @@ -235,8 +236,9 @@ void netback_accel_fwd_remove(const __u8
5646 unsigned long flags;
5647 cuckoo_hash_mac_key key = cuckoo_mac_to_key(mac);
5648 struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
5649 + DECLARE_MAC_BUF(buf);
5650
5651 - DPRINTK("Removing mac " MAC_FMT "\n", MAC_ARG(mac));
5652 + DPRINTK("Removing mac %s\n", print_mac(buf, mac));
5653
5654 BUG_ON(fwd_priv == NULL);
5655
5656 @@ -394,14 +396,16 @@ void netback_accel_tx_packet(struct sk_b
5657
5658 if (is_broadcast_ether_addr(skb_mac_header(skb))
5659 && packet_is_arp_reply(skb)) {
5660 + DECLARE_MAC_BUF(buf);
5661 +
5662 /*
5663 * update our fast path forwarding to reflect this
5664 * gratuitous ARP
5665 */
5666 mac = skb_mac_header(skb)+ETH_ALEN;
5667
5668 - DPRINTK("%s: found gratuitous ARP for " MAC_FMT "\n",
5669 - __FUNCTION__, MAC_ARG(mac));
5670 + DPRINTK("%s: found gratuitous ARP for %s\n",
5671 + __FUNCTION__, print_mac(buf, mac));
5672
5673 spin_lock_irqsave(&fwd_set->fwd_lock, flags);
5674 /*
5675 --- sle11-2009-05-14.orig/drivers/xen/sfc_netback/accel_msg.c 2009-05-14 10:56:29.000000000 +0200
5676 +++ sle11-2009-05-14/drivers/xen/sfc_netback/accel_msg.c 2009-02-16 16:18:36.000000000 +0100
5677 @@ -57,11 +57,11 @@ static void netback_accel_msg_tx_localma
5678 {
5679 unsigned long lock_state;
5680 struct net_accel_msg *msg;
5681 + DECLARE_MAC_BUF(buf);
5682
5683 BUG_ON(bend == NULL || mac == NULL);
5684
5685 - VPRINTK("Sending local mac message: " MAC_FMT "\n",
5686 - MAC_ARG((const char *)mac));
5687 + VPRINTK("Sending local mac message: %s\n", print_mac(buf, mac));
5688
5689 msg = net_accel_msg_start_send(bend->shared_page, &bend->to_domU,
5690 &lock_state);
5691 --- sle11-2009-05-14.orig/drivers/xen/sfc_netfront/accel_msg.c 2009-03-04 11:28:34.000000000 +0100
5692 +++ sle11-2009-05-14/drivers/xen/sfc_netfront/accel_msg.c 2009-02-16 16:18:36.000000000 +0100
5693 @@ -41,11 +41,13 @@ static void vnic_start_interrupts(netfro
5694 /* Prime our interrupt */
5695 spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
5696 if (!netfront_accel_vi_enable_interrupts(vnic)) {
5697 + struct netfront_info *np = netdev_priv(vnic->net_dev);
5698 +
5699 /* Cripes, that was quick, better pass it up */
5700 netfront_accel_disable_net_interrupts(vnic);
5701 vnic->irq_enabled = 0;
5702 NETFRONT_ACCEL_STATS_OP(vnic->stats.poll_schedule_count++);
5703 - netif_rx_schedule(vnic->net_dev);
5704 + netif_rx_schedule(vnic->net_dev, &np->napi);
5705 } else {
5706 /*
5707 * Nothing yet, make sure we get interrupts through
5708 @@ -72,6 +74,7 @@ static void vnic_stop_interrupts(netfron
5709 static void vnic_start_fastpath(netfront_accel_vnic *vnic)
5710 {
5711 struct net_device *net_dev = vnic->net_dev;
5712 + struct netfront_info *np = netdev_priv(net_dev);
5713 unsigned long flags;
5714
5715 DPRINTK("%s\n", __FUNCTION__);
5716 @@ -80,9 +83,9 @@ static void vnic_start_fastpath(netfront
5717 vnic->tx_enabled = 1;
5718 spin_unlock_irqrestore(&vnic->tx_lock, flags);
5719
5720 - netif_poll_disable(net_dev);
5721 + napi_disable(&np->napi);
5722 vnic->poll_enabled = 1;
5723 - netif_poll_enable(net_dev);
5724 + napi_enable(&np->napi);
5725
5726 vnic_start_interrupts(vnic);
5727 }
5728 @@ -114,11 +117,11 @@ void vnic_stop_fastpath(netfront_accel_v
5729 spin_unlock_irqrestore(&vnic->tx_lock, flags1);
5730
5731 /* Must prevent polls and hold lock to modify poll_enabled */
5732 - netif_poll_disable(net_dev);
5733 + napi_disable(&np->napi);
5734 spin_lock_irqsave(&vnic->irq_enabled_lock, flags1);
5735 vnic->poll_enabled = 0;
5736 spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags1);
5737 - netif_poll_enable(net_dev);
5738 + napi_enable(&np->napi);
5739 }
5740
5741
5742 @@ -326,8 +329,10 @@ static int vnic_process_localmac_msg(net
5743 cuckoo_hash_mac_key key;
5744
5745 if (msg->u.localmac.flags & NET_ACCEL_MSG_ADD) {
5746 - DPRINTK("MAC has moved, could be local: " MAC_FMT "\n",
5747 - MAC_ARG(msg->u.localmac.mac));
5748 + DECLARE_MAC_BUF(buf);
5749 +
5750 + DPRINTK("MAC has moved, could be local: %s\n",
5751 + print_mac(buf, msg->u.localmac.mac));
5752 key = cuckoo_mac_to_key(msg->u.localmac.mac);
5753 spin_lock_irqsave(&vnic->table_lock, flags);
5754 /* Try to remove it, not a big deal if not there */
5755 @@ -515,6 +520,8 @@ irqreturn_t netfront_accel_net_channel_i
5756
5757 spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
5758 if (vnic->irq_enabled) {
5759 + struct netfront_info *np = netdev_priv(net_dev);
5760 +
5761 netfront_accel_disable_net_interrupts(vnic);
5762 vnic->irq_enabled = 0;
5763 spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
5764 @@ -527,7 +534,7 @@ irqreturn_t netfront_accel_net_channel_i
5765 vnic->stats.event_count_since_irq;
5766 vnic->stats.event_count_since_irq = 0;
5767 #endif
5768 - netif_rx_schedule(net_dev);
5769 + netif_rx_schedule(net_dev, &np->napi);
5770 }
5771 else {
5772 spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
5773 --- sle11-2009-05-14.orig/drivers/xen/sfc_netfront/accel_vi.c 2009-03-30 16:36:26.000000000 +0200
5774 +++ sle11-2009-05-14/drivers/xen/sfc_netfront/accel_vi.c 2009-03-30 16:39:38.000000000 +0200
5775 @@ -641,8 +641,10 @@ netfront_accel_vi_tx_post(netfront_accel
5776 (cuckoo_hash_key *)(&key), &value);
5777
5778 if (!try_fastpath) {
5779 - VPRINTK("try fast path false for mac: " MAC_FMT "\n",
5780 - MAC_ARG(skb->data));
5781 + DECLARE_MAC_BUF(buf);
5782 +
5783 + VPRINTK("try fast path false for mac: %s\n",
5784 + print_mac(buf, skb->data));
5785
5786 return NETFRONT_ACCEL_STATUS_CANT;
5787 }
5788 @@ -768,9 +770,10 @@ static void netfront_accel_vi_rx_comple
5789 if (compare_ether_addr(skb->data, vnic->mac)) {
5790 struct iphdr *ip = (struct iphdr *)(skb->data + ETH_HLEN);
5791 u16 port;
5792 + DECLARE_MAC_BUF(buf);
5793
5794 - DPRINTK("%s: saw wrong MAC address " MAC_FMT "\n",
5795 - __FUNCTION__, MAC_ARG(skb->data));
5796 + DPRINTK("%s: saw wrong MAC address %s\n",
5797 + __FUNCTION__, print_mac(buf, skb->data));
5798
5799 if (ip->protocol == IPPROTO_TCP) {
5800 struct tcphdr *tcp = (struct tcphdr *)
5801 --- sle11-2009-05-14.orig/drivers/xen/sfc_netutil/accel_util.h 2009-05-14 10:56:29.000000000 +0200
5802 +++ sle11-2009-05-14/drivers/xen/sfc_netutil/accel_util.h 2009-02-16 16:18:36.000000000 +0100
5803 @@ -63,9 +63,6 @@
5804 DPRINTK("%s at %s:%d\n", #exp, __FILE__, __LINE__); \
5805 } while(0)
5806
5807 -#define MAC_FMT "%.2x:%.2x:%.2x:%.2x:%.2x:%.2x"
5808 -#define MAC_ARG(_mac) (_mac)[0], (_mac)[1], (_mac)[2], (_mac)[3], (_mac)[4], (_mac)[5]
5809 -
5810 #include <xen/xenbus.h>
5811
5812 /*! Map a set of pages from another domain
5813 --- sle11-2009-05-14.orig/drivers/xen/xenbus/xenbus_probe.c 2009-02-16 16:17:21.000000000 +0100
5814 +++ sle11-2009-05-14/drivers/xen/xenbus/xenbus_probe.c 2009-02-16 16:18:36.000000000 +0100
5815 @@ -174,11 +174,9 @@ static int read_backend_details(struct x
5816 }
5817
5818 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) && (defined(CONFIG_XEN) || defined(MODULE))
5819 -static int xenbus_uevent_frontend(struct device *dev, char **envp,
5820 - int num_envp, char *buffer, int buffer_size)
5821 +static int xenbus_uevent_frontend(struct device *dev, struct kobj_uevent_env *env)
5822 {
5823 struct xenbus_device *xdev;
5824 - int length = 0, i = 0;
5825
5826 if (dev == NULL)
5827 return -ENODEV;
5828 @@ -187,12 +185,9 @@ static int xenbus_uevent_frontend(struct
5829 return -ENODEV;
5830
5831 /* stuff we want to pass to /sbin/hotplug */
5832 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5833 - "XENBUS_TYPE=%s", xdev->devicetype);
5834 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5835 - "XENBUS_PATH=%s", xdev->nodename);
5836 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5837 - "MODALIAS=xen:%s", xdev->devicetype);
5838 + add_uevent_var(env, "XENBUS_TYPE=%s", xdev->devicetype);
5839 + add_uevent_var(env, "XENBUS_PATH=%s", xdev->nodename);
5840 + add_uevent_var(env, "MODALIAS=xen:%s", xdev->devicetype);
5841
5842 return 0;
5843 }
5844 --- sle11-2009-05-14.orig/drivers/xen/xenbus/xenbus_probe_backend.c 2009-02-16 16:17:21.000000000 +0100
5845 +++ sle11-2009-05-14/drivers/xen/xenbus/xenbus_probe_backend.c 2009-02-16 16:18:36.000000000 +0100
5846 @@ -60,8 +60,7 @@
5847 #include <xen/platform-compat.h>
5848 #endif
5849
5850 -static int xenbus_uevent_backend(struct device *dev, char **envp,
5851 - int num_envp, char *buffer, int buffer_size);
5852 +static int xenbus_uevent_backend(struct device *dev, struct kobj_uevent_env *env);
5853 static int xenbus_probe_backend(const char *type, const char *domid);
5854
5855 extern int read_otherend_details(struct xenbus_device *xendev,
5856 @@ -128,13 +127,10 @@ static struct xen_bus_type xenbus_backen
5857 },
5858 };
5859
5860 -static int xenbus_uevent_backend(struct device *dev, char **envp,
5861 - int num_envp, char *buffer, int buffer_size)
5862 +static int xenbus_uevent_backend(struct device *dev, struct kobj_uevent_env *env)
5863 {
5864 struct xenbus_device *xdev;
5865 struct xenbus_driver *drv;
5866 - int i = 0;
5867 - int length = 0;
5868
5869 DPRINTK("");
5870
5871 @@ -146,27 +142,16 @@ static int xenbus_uevent_backend(struct
5872 return -ENODEV;
5873
5874 /* stuff we want to pass to /sbin/hotplug */
5875 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5876 - "XENBUS_TYPE=%s", xdev->devicetype);
5877 + add_uevent_var(env, "XENBUS_TYPE=%s", xdev->devicetype);
5878
5879 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5880 - "XENBUS_PATH=%s", xdev->nodename);
5881 + add_uevent_var(env, "XENBUS_PATH=%s", xdev->nodename);
5882
5883 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5884 - "XENBUS_BASE_PATH=%s", xenbus_backend.root);
5885 -
5886 - /* terminate, set to next free slot, shrink available space */
5887 - envp[i] = NULL;
5888 - envp = &envp[i];
5889 - num_envp -= i;
5890 - buffer = &buffer[length];
5891 - buffer_size -= length;
5892 + add_uevent_var(env, "XENBUS_BASE_PATH=%s", xenbus_backend.root);
5893
5894 if (dev->driver) {
5895 drv = to_xenbus_driver(dev->driver);
5896 if (drv && drv->uevent)
5897 - return drv->uevent(xdev, envp, num_envp, buffer,
5898 - buffer_size);
5899 + return drv->uevent(xdev, env);
5900 }
5901
5902 return 0;
5903 --- sle11-2009-05-14.orig/drivers/xen/xenoprof/xenoprofile.c 2009-03-04 11:25:55.000000000 +0100
5904 +++ sle11-2009-05-14/drivers/xen/xenoprof/xenoprofile.c 2009-03-11 15:39:38.000000000 +0100
5905 @@ -29,7 +29,6 @@
5906 #include <xen/driver_util.h>
5907 #include <xen/interface/xen.h>
5908 #include <xen/interface/xenoprof.h>
5909 -#include "../../../drivers/oprofile/cpu_buffer.h"
5910 #include "../../../drivers/oprofile/event_buffer.h"
5911
5912 #define MAX_XENOPROF_SAMPLES 16
5913 @@ -142,8 +141,7 @@ static void xenoprof_add_pc(xenoprof_buf
5914 if (xenoprof_is_escape(buf, tail) &&
5915 xenoprof_get_event(buf, tail) == XENOPROF_TRACE_BEGIN) {
5916 tracing=1;
5917 - oprofile_add_pc(ESCAPE_CODE, buf->event_log[tail].mode,
5918 - CPU_TRACE_BEGIN);
5919 + oprofile_add_mode(buf->event_log[tail].mode);
5920 if (!is_passive)
5921 oprofile_samples++;
5922 else
5923 --- sle11-2009-05-14.orig/fs/xfs/linux-2.6/xfs_buf.c 2009-05-14 10:56:29.000000000 +0200
5924 +++ sle11-2009-05-14/fs/xfs/linux-2.6/xfs_buf.c 2009-02-16 16:18:36.000000000 +0100
5925 @@ -187,7 +187,7 @@ free_address(
5926 {
5927 a_list_t *aentry;
5928
5929 -#ifdef CONFIG_XEN
5930 +#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN)
5931 /*
5932 * Xen needs to be able to make sure it can get an exclusive
5933 * RO mapping of pages it wants to turn into a pagetable. If
5934 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/agp.h 2009-05-14 10:56:29.000000000 +0200
5935 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/agp.h 2009-02-16 16:18:36.000000000 +0100
5936 @@ -1,20 +1,22 @@
5937 -#ifndef AGP_H
5938 -#define AGP_H 1
5939 +#ifndef _ASM_X86_AGP_H
5940 +#define _ASM_X86_AGP_H
5941
5942 #include <asm/pgtable.h>
5943 #include <asm/cacheflush.h>
5944 #include <asm/system.h>
5945
5946 -/*
5947 - * Functions to keep the agpgart mappings coherent with the MMU.
5948 - * The GART gives the CPU a physical alias of pages in memory. The alias region is
5949 - * mapped uncacheable. Make sure there are no conflicting mappings
5950 - * with different cachability attributes for the same page. This avoids
5951 - * data corruption on some CPUs.
5952 +/*
5953 + * Functions to keep the agpgart mappings coherent with the MMU. The
5954 + * GART gives the CPU a physical alias of pages in memory. The alias
5955 + * region is mapped uncacheable. Make sure there are no conflicting
5956 + * mappings with different cachability attributes for the same
5957 + * page. This avoids data corruption on some CPUs.
5958 */
5959
5960 -/* Caller's responsibility to call global_flush_tlb() for
5961 - * performance reasons */
5962 +/*
5963 + * Caller's responsibility to call global_flush_tlb() for performance
5964 + * reasons
5965 + */
5966 #define map_page_into_agp(page) ( \
5967 xen_create_contiguous_region((unsigned long)page_address(page), 0, 32) \
5968 ?: change_page_attr(page, 1, PAGE_KERNEL_NOCACHE))
5969 @@ -24,9 +26,11 @@
5970 change_page_attr(page, 1, PAGE_KERNEL))
5971 #define flush_agp_mappings() global_flush_tlb()
5972
5973 -/* Could use CLFLUSH here if the cpu supports it. But then it would
5974 - need to be called for each cacheline of the whole page so it may not be
5975 - worth it. Would need a page for it. */
5976 +/*
5977 + * Could use CLFLUSH here if the cpu supports it. But then it would
5978 + * need to be called for each cacheline of the whole page so it may
5979 + * not be worth it. Would need a page for it.
5980 + */
5981 #define flush_agp_cache() wbinvd()
5982
5983 /* Convert a physical address to an address suitable for the GART. */
5984 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
5985 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/desc.h 2009-02-16 16:18:36.000000000 +0100
5986 @@ -0,0 +1,5 @@
5987 +#ifdef CONFIG_X86_32
5988 +# include "desc_32.h"
5989 +#else
5990 +# include "desc_64.h"
5991 +#endif
5992 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/desc_64.h 2008-12-15 11:27:22.000000000 +0100
5993 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/desc_64.h 2009-02-16 16:18:36.000000000 +0100
5994 @@ -34,6 +34,18 @@ static inline void clear_LDT(void)
5995 put_cpu();
5996 }
5997
5998 +#ifndef CONFIG_X86_NO_TSS
5999 +static inline unsigned long __store_tr(void)
6000 +{
6001 + unsigned long tr;
6002 +
6003 + asm volatile ("str %w0":"=r" (tr));
6004 + return tr;
6005 +}
6006 +
6007 +#define store_tr(tr) (tr) = __store_tr()
6008 +#endif
6009 +
6010 /*
6011 * This is the ldt that every process will get unless we need
6012 * something other than this.
6013 @@ -47,6 +59,18 @@ extern struct desc_ptr cpu_gdt_descr[];
6014 /* the cpu gdt accessor */
6015 #define cpu_gdt(_cpu) ((struct desc_struct *)cpu_gdt_descr[_cpu].address)
6016
6017 +#ifndef CONFIG_XEN
6018 +static inline void load_gdt(const struct desc_ptr *ptr)
6019 +{
6020 + asm volatile("lgdt %w0"::"m" (*ptr));
6021 +}
6022 +
6023 +static inline void store_gdt(struct desc_ptr *ptr)
6024 +{
6025 + asm("sgdt %w0":"=m" (*ptr));
6026 +}
6027 +#endif
6028 +
6029 static inline void _set_gate(void *adr, unsigned type, unsigned long func, unsigned dpl, unsigned ist)
6030 {
6031 struct gate_struct s;
6032 @@ -87,6 +111,16 @@ static inline void set_system_gate_ist(i
6033 {
6034 _set_gate(&idt_table[nr], GATE_INTERRUPT, (unsigned long) func, 3, ist);
6035 }
6036 +
6037 +static inline void load_idt(const struct desc_ptr *ptr)
6038 +{
6039 + asm volatile("lidt %w0"::"m" (*ptr));
6040 +}
6041 +
6042 +static inline void store_idt(struct desc_ptr *dtr)
6043 +{
6044 + asm("sidt %w0":"=m" (*dtr));
6045 +}
6046 #endif
6047
6048 static inline void set_tssldt_descriptor(void *ptr, unsigned long tss, unsigned type,
6049 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
6050 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/dma-mapping.h 2009-02-16 16:18:36.000000000 +0100
6051 @@ -0,0 +1,5 @@
6052 +#ifdef CONFIG_X86_32
6053 +# include "dma-mapping_32.h"
6054 +#else
6055 +# include "dma-mapping_64.h"
6056 +#endif
6057 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/dma-mapping_32.h 2008-12-15 11:26:44.000000000 +0100
6058 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/dma-mapping_32.h 2009-02-16 16:18:36.000000000 +0100
6059 @@ -7,9 +7,9 @@
6060 */
6061
6062 #include <linux/mm.h>
6063 +#include <linux/scatterlist.h>
6064 #include <asm/cache.h>
6065 #include <asm/io.h>
6066 -#include <asm/scatterlist.h>
6067 #include <asm/swiotlb.h>
6068
6069 static inline int
6070 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/dma-mapping_64.h 2008-12-15 11:27:22.000000000 +0100
6071 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/dma-mapping_64.h 2009-02-16 16:18:36.000000000 +0100
6072 @@ -6,8 +6,7 @@
6073 * documentation.
6074 */
6075
6076 -
6077 -#include <asm/scatterlist.h>
6078 +#include <linux/scatterlist.h>
6079 #include <asm/io.h>
6080
6081 struct dma_mapping_ops {
6082 @@ -203,4 +202,4 @@ extern int panic_on_overflow;
6083
6084 #endif /* _X8664_DMA_MAPPING_H */
6085
6086 -#include <asm-i386/mach-xen/asm/dma-mapping.h>
6087 +#include "dma-mapping_32.h"
6088 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
6089 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/fixmap.h 2009-02-16 16:18:36.000000000 +0100
6090 @@ -0,0 +1,5 @@
6091 +#ifdef CONFIG_X86_32
6092 +# include "fixmap_32.h"
6093 +#else
6094 +# include "fixmap_64.h"
6095 +#endif
6096 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
6097 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/hypercall.h 2009-02-16 16:18:36.000000000 +0100
6098 @@ -0,0 +1,404 @@
6099 +/******************************************************************************
6100 + * hypercall.h
6101 + *
6102 + * Linux-specific hypervisor handling.
6103 + *
6104 + * Copyright (c) 2002-2004, K A Fraser
6105 + *
6106 + * 64-bit updates:
6107 + * Benjamin Liu <benjamin.liu@intel.com>
6108 + * Jun Nakajima <jun.nakajima@intel.com>
6109 + *
6110 + * This program is free software; you can redistribute it and/or
6111 + * modify it under the terms of the GNU General Public License version 2
6112 + * as published by the Free Software Foundation; or, when distributed
6113 + * separately from the Linux kernel or incorporated into other
6114 + * software packages, subject to the following license:
6115 + *
6116 + * Permission is hereby granted, free of charge, to any person obtaining a copy
6117 + * of this source file (the "Software"), to deal in the Software without
6118 + * restriction, including without limitation the rights to use, copy, modify,
6119 + * merge, publish, distribute, sublicense, and/or sell copies of the Software,
6120 + * and to permit persons to whom the Software is furnished to do so, subject to
6121 + * the following conditions:
6122 + *
6123 + * The above copyright notice and this permission notice shall be included in
6124 + * all copies or substantial portions of the Software.
6125 + *
6126 + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
6127 + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
6128 + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
6129 + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
6130 + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
6131 + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
6132 + * IN THE SOFTWARE.
6133 + */
6134 +
6135 +#ifndef __HYPERCALL_H__
6136 +#define __HYPERCALL_H__
6137 +
6138 +#ifndef __HYPERVISOR_H__
6139 +# error "please don't include this file directly"
6140 +#endif
6141 +
6142 +#if CONFIG_XEN_COMPAT <= 0x030002
6143 +# include <linux/string.h> /* memcpy() */
6144 +#endif
6145 +
6146 +#ifdef CONFIG_XEN
6147 +#define HYPERCALL_ASM_OPERAND "%c"
6148 +#define HYPERCALL_LOCATION(op) (hypercall_page + (op) * 32)
6149 +#define HYPERCALL_C_OPERAND(name) "i" (HYPERCALL_LOCATION(__HYPERVISOR_##name))
6150 +#else
6151 +#define HYPERCALL_ASM_OPERAND "*%"
6152 +#define HYPERCALL_LOCATION(op) (hypercall_stubs + (op) * 32)
6153 +#define HYPERCALL_C_OPERAND(name) "g" (HYPERCALL_LOCATION(__HYPERVISOR_##name))
6154 +#endif
6155 +
6156 +#define HYPERCALL_ARG(arg, n) \
6157 + register typeof((arg)+0) __arg##n asm(HYPERCALL_arg##n) = (arg)
6158 +
6159 +#define _hypercall0(type, name) \
6160 +({ \
6161 + type __res; \
6162 + asm volatile ( \
6163 + "call " HYPERCALL_ASM_OPERAND "1" \
6164 + : "=a" (__res) \
6165 + : HYPERCALL_C_OPERAND(name) \
6166 + : "memory" ); \
6167 + __res; \
6168 +})
6169 +
6170 +#define _hypercall1(type, name, arg) \
6171 +({ \
6172 + type __res; \
6173 + HYPERCALL_ARG(arg, 1); \
6174 + asm volatile ( \
6175 + "call " HYPERCALL_ASM_OPERAND "2" \
6176 + : "=a" (__res), "+r" (__arg1) \
6177 + : HYPERCALL_C_OPERAND(name) \
6178 + : "memory" ); \
6179 + __res; \
6180 +})
6181 +
6182 +#define _hypercall2(type, name, a1, a2) \
6183 +({ \
6184 + type __res; \
6185 + HYPERCALL_ARG(a1, 1); \
6186 + HYPERCALL_ARG(a2, 2); \
6187 + asm volatile ( \
6188 + "call " HYPERCALL_ASM_OPERAND "3" \
6189 + : "=a" (__res), "+r" (__arg1), "+r" (__arg2) \
6190 + : HYPERCALL_C_OPERAND(name) \
6191 + : "memory" ); \
6192 + __res; \
6193 +})
6194 +
6195 +#define _hypercall3(type, name, a1, a2, a3) \
6196 +({ \
6197 + type __res; \
6198 + HYPERCALL_ARG(a1, 1); \
6199 + HYPERCALL_ARG(a2, 2); \
6200 + HYPERCALL_ARG(a3, 3); \
6201 + asm volatile ( \
6202 + "call " HYPERCALL_ASM_OPERAND "4" \
6203 + : "=a" (__res), "+r" (__arg1), \
6204 + "+r" (__arg2), "+r" (__arg3) \
6205 + : HYPERCALL_C_OPERAND(name) \
6206 + : "memory" ); \
6207 + __res; \
6208 +})
6209 +
6210 +#define _hypercall4(type, name, a1, a2, a3, a4) \
6211 +({ \
6212 + type __res; \
6213 + HYPERCALL_ARG(a1, 1); \
6214 + HYPERCALL_ARG(a2, 2); \
6215 + HYPERCALL_ARG(a3, 3); \
6216 + HYPERCALL_ARG(a4, 4); \
6217 + asm volatile ( \
6218 + "call " HYPERCALL_ASM_OPERAND "5" \
6219 + : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
6220 + "+r" (__arg3), "+r" (__arg4) \
6221 + : HYPERCALL_C_OPERAND(name) \
6222 + : "memory" ); \
6223 + __res; \
6224 +})
6225 +
6226 +#define _hypercall5(type, name, a1, a2, a3, a4, a5) \
6227 +({ \
6228 + type __res; \
6229 + HYPERCALL_ARG(a1, 1); \
6230 + HYPERCALL_ARG(a2, 2); \
6231 + HYPERCALL_ARG(a3, 3); \
6232 + HYPERCALL_ARG(a4, 4); \
6233 + HYPERCALL_ARG(a5, 5); \
6234 + asm volatile ( \
6235 + "call " HYPERCALL_ASM_OPERAND "6" \
6236 + : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
6237 + "+r" (__arg3), "+r" (__arg4), "+r" (__arg5) \
6238 + : HYPERCALL_C_OPERAND(name) \
6239 + : "memory" ); \
6240 + __res; \
6241 +})
6242 +
6243 +#define _hypercall(type, op, a1, a2, a3, a4, a5) \
6244 +({ \
6245 + type __res; \
6246 + HYPERCALL_ARG(a1, 1); \
6247 + HYPERCALL_ARG(a2, 2); \
6248 + HYPERCALL_ARG(a3, 3); \
6249 + HYPERCALL_ARG(a4, 4); \
6250 + HYPERCALL_ARG(a5, 5); \
6251 + asm volatile ( \
6252 + "call *%6" \
6253 + : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
6254 + "+r" (__arg3), "+r" (__arg4), "+r" (__arg5) \
6255 + : "g" (HYPERCALL_LOCATION(op)) \
6256 + : "memory" ); \
6257 + __res; \
6258 +})
6259 +
6260 +#ifdef CONFIG_X86_32
6261 +# include "hypercall_32.h"
6262 +#else
6263 +# include "hypercall_64.h"
6264 +#endif
6265 +
6266 +static inline int __must_check
6267 +HYPERVISOR_set_trap_table(
6268 + const trap_info_t *table)
6269 +{
6270 + return _hypercall1(int, set_trap_table, table);
6271 +}
6272 +
6273 +static inline int __must_check
6274 +HYPERVISOR_mmu_update(
6275 + mmu_update_t *req, unsigned int count, unsigned int *success_count,
6276 + domid_t domid)
6277 +{
6278 + if (arch_use_lazy_mmu_mode())
6279 + return xen_multi_mmu_update(req, count, success_count, domid);
6280 + return _hypercall4(int, mmu_update, req, count, success_count, domid);
6281 +}
6282 +
6283 +static inline int __must_check
6284 +HYPERVISOR_mmuext_op(
6285 + struct mmuext_op *op, unsigned int count, unsigned int *success_count,
6286 + domid_t domid)
6287 +{
6288 + if (arch_use_lazy_mmu_mode())
6289 + return xen_multi_mmuext_op(op, count, success_count, domid);
6290 + return _hypercall4(int, mmuext_op, op, count, success_count, domid);
6291 +}
6292 +
6293 +static inline int __must_check
6294 +HYPERVISOR_set_gdt(
6295 + unsigned long *frame_list, unsigned int entries)
6296 +{
6297 + return _hypercall2(int, set_gdt, frame_list, entries);
6298 +}
6299 +
6300 +static inline int __must_check
6301 +HYPERVISOR_stack_switch(
6302 + unsigned long ss, unsigned long esp)
6303 +{
6304 + return _hypercall2(int, stack_switch, ss, esp);
6305 +}
6306 +
6307 +static inline int
6308 +HYPERVISOR_fpu_taskswitch(
6309 + int set)
6310 +{
6311 + return _hypercall1(int, fpu_taskswitch, set);
6312 +}
6313 +
6314 +#if CONFIG_XEN_COMPAT <= 0x030002
6315 +static inline int __must_check
6316 +HYPERVISOR_sched_op_compat(
6317 + int cmd, unsigned long arg)
6318 +{
6319 + return _hypercall2(int, sched_op_compat, cmd, arg);
6320 +}
6321 +#endif
6322 +
6323 +static inline int __must_check
6324 +HYPERVISOR_sched_op(
6325 + int cmd, void *arg)
6326 +{
6327 + return _hypercall2(int, sched_op, cmd, arg);
6328 +}
6329 +
6330 +static inline int __must_check
6331 +HYPERVISOR_platform_op(
6332 + struct xen_platform_op *platform_op)
6333 +{
6334 + platform_op->interface_version = XENPF_INTERFACE_VERSION;
6335 + return _hypercall1(int, platform_op, platform_op);
6336 +}
6337 +
6338 +static inline int __must_check
6339 +HYPERVISOR_set_debugreg(
6340 + unsigned int reg, unsigned long value)
6341 +{
6342 + return _hypercall2(int, set_debugreg, reg, value);
6343 +}
6344 +
6345 +static inline unsigned long __must_check
6346 +HYPERVISOR_get_debugreg(
6347 + unsigned int reg)
6348 +{
6349 + return _hypercall1(unsigned long, get_debugreg, reg);
6350 +}
6351 +
6352 +static inline int __must_check
6353 +HYPERVISOR_memory_op(
6354 + unsigned int cmd, void *arg)
6355 +{
6356 + if (arch_use_lazy_mmu_mode())
6357 + xen_multicall_flush(false);
6358 + return _hypercall2(int, memory_op, cmd, arg);
6359 +}
6360 +
6361 +static inline int __must_check
6362 +HYPERVISOR_multicall(
6363 + multicall_entry_t *call_list, unsigned int nr_calls)
6364 +{
6365 + return _hypercall2(int, multicall, call_list, nr_calls);
6366 +}
6367 +
6368 +static inline int __must_check
6369 +HYPERVISOR_event_channel_op(
6370 + int cmd, void *arg)
6371 +{
6372 + int rc = _hypercall2(int, event_channel_op, cmd, arg);
6373 +
6374 +#if CONFIG_XEN_COMPAT <= 0x030002
6375 + if (unlikely(rc == -ENOSYS)) {
6376 + struct evtchn_op op;
6377 + op.cmd = cmd;
6378 + memcpy(&op.u, arg, sizeof(op.u));
6379 + rc = _hypercall1(int, event_channel_op_compat, &op);
6380 + memcpy(arg, &op.u, sizeof(op.u));
6381 + }
6382 +#endif
6383 +
6384 + return rc;
6385 +}
6386 +
6387 +static inline int __must_check
6388 +HYPERVISOR_xen_version(
6389 + int cmd, void *arg)
6390 +{
6391 + return _hypercall2(int, xen_version, cmd, arg);
6392 +}
6393 +
6394 +static inline int __must_check
6395 +HYPERVISOR_console_io(
6396 + int cmd, unsigned int count, char *str)
6397 +{
6398 + return _hypercall3(int, console_io, cmd, count, str);
6399 +}
6400 +
6401 +static inline int __must_check
6402 +HYPERVISOR_physdev_op(
6403 + int cmd, void *arg)
6404 +{
6405 + int rc = _hypercall2(int, physdev_op, cmd, arg);
6406 +
6407 +#if CONFIG_XEN_COMPAT <= 0x030002
6408 + if (unlikely(rc == -ENOSYS)) {
6409 + struct physdev_op op;
6410 + op.cmd = cmd;
6411 + memcpy(&op.u, arg, sizeof(op.u));
6412 + rc = _hypercall1(int, physdev_op_compat, &op);
6413 + memcpy(arg, &op.u, sizeof(op.u));
6414 + }
6415 +#endif
6416 +
6417 + return rc;
6418 +}
6419 +
6420 +static inline int __must_check
6421 +HYPERVISOR_grant_table_op(
6422 + unsigned int cmd, void *uop, unsigned int count)
6423 +{
6424 + if (arch_use_lazy_mmu_mode())
6425 + xen_multicall_flush(false);
6426 + return _hypercall3(int, grant_table_op, cmd, uop, count);
6427 +}
6428 +
6429 +static inline int __must_check
6430 +HYPERVISOR_vm_assist(
6431 + unsigned int cmd, unsigned int type)
6432 +{
6433 + return _hypercall2(int, vm_assist, cmd, type);
6434 +}
6435 +
6436 +static inline int __must_check
6437 +HYPERVISOR_vcpu_op(
6438 + int cmd, unsigned int vcpuid, void *extra_args)
6439 +{
6440 + return _hypercall3(int, vcpu_op, cmd, vcpuid, extra_args);
6441 +}
6442 +
6443 +static inline int __must_check
6444 +HYPERVISOR_suspend(
6445 + unsigned long srec)
6446 +{
6447 + struct sched_shutdown sched_shutdown = {
6448 + .reason = SHUTDOWN_suspend
6449 + };
6450 +
6451 + int rc = _hypercall3(int, sched_op, SCHEDOP_shutdown,
6452 + &sched_shutdown, srec);
6453 +
6454 +#if CONFIG_XEN_COMPAT <= 0x030002
6455 + if (rc == -ENOSYS)
6456 + rc = _hypercall3(int, sched_op_compat, SCHEDOP_shutdown,
6457 + SHUTDOWN_suspend, srec);
6458 +#endif
6459 +
6460 + return rc;
6461 +}
6462 +
6463 +#if CONFIG_XEN_COMPAT <= 0x030002
6464 +static inline int
6465 +HYPERVISOR_nmi_op(
6466 + unsigned long op, void *arg)
6467 +{
6468 + return _hypercall2(int, nmi_op, op, arg);
6469 +}
6470 +#endif
6471 +
6472 +#ifndef CONFIG_XEN
6473 +static inline unsigned long __must_check
6474 +HYPERVISOR_hvm_op(
6475 + int op, void *arg)
6476 +{
6477 + return _hypercall2(unsigned long, hvm_op, op, arg);
6478 +}
6479 +#endif
6480 +
6481 +static inline int __must_check
6482 +HYPERVISOR_callback_op(
6483 + int cmd, const void *arg)
6484 +{
6485 + return _hypercall2(int, callback_op, cmd, arg);
6486 +}
6487 +
6488 +static inline int __must_check
6489 +HYPERVISOR_xenoprof_op(
6490 + int op, void *arg)
6491 +{
6492 + return _hypercall2(int, xenoprof_op, op, arg);
6493 +}
6494 +
6495 +static inline int __must_check
6496 +HYPERVISOR_kexec_op(
6497 + unsigned long op, void *args)
6498 +{
6499 + return _hypercall2(int, kexec_op, op, args);
6500 +}
6501 +
6502 +#endif /* __HYPERCALL_H__ */
6503 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/hypercall_32.h 2009-03-04 11:28:34.000000000 +0100
6504 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/hypercall_32.h 2009-02-16 16:18:36.000000000 +0100
6505 @@ -1,191 +1,10 @@
6506 -/******************************************************************************
6507 - * hypercall.h
6508 - *
6509 - * Linux-specific hypervisor handling.
6510 - *
6511 - * Copyright (c) 2002-2004, K A Fraser
6512 - *
6513 - * This program is free software; you can redistribute it and/or
6514 - * modify it under the terms of the GNU General Public License version 2
6515 - * as published by the Free Software Foundation; or, when distributed
6516 - * separately from the Linux kernel or incorporated into other
6517 - * software packages, subject to the following license:
6518 - *
6519 - * Permission is hereby granted, free of charge, to any person obtaining a copy
6520 - * of this source file (the "Software"), to deal in the Software without
6521 - * restriction, including without limitation the rights to use, copy, modify,
6522 - * merge, publish, distribute, sublicense, and/or sell copies of the Software,
6523 - * and to permit persons to whom the Software is furnished to do so, subject to
6524 - * the following conditions:
6525 - *
6526 - * The above copyright notice and this permission notice shall be included in
6527 - * all copies or substantial portions of the Software.
6528 - *
6529 - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
6530 - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
6531 - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
6532 - * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
6533 - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
6534 - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
6535 - * IN THE SOFTWARE.
6536 - */
6537 -
6538 -#ifndef __HYPERCALL_H__
6539 -#define __HYPERCALL_H__
6540 -
6541 -#include <linux/string.h> /* memcpy() */
6542 -#include <linux/stringify.h>
6543 -
6544 -#ifndef __HYPERVISOR_H__
6545 -# error "please don't include this file directly"
6546 -#endif
6547 -
6548 -#ifdef CONFIG_XEN
6549 -#define HYPERCALL_STR(name) \
6550 - "call hypercall_page + ("__stringify(__HYPERVISOR_##name)" * 32)"
6551 -#else
6552 -#define HYPERCALL_STR(name) \
6553 - "mov hypercall_stubs,%%eax; " \
6554 - "add $("__stringify(__HYPERVISOR_##name)" * 32),%%eax; "\
6555 - "call *%%eax"
6556 -#endif
6557 -
6558 -#define _hypercall0(type, name) \
6559 -({ \
6560 - type __res; \
6561 - asm volatile ( \
6562 - HYPERCALL_STR(name) \
6563 - : "=a" (__res) \
6564 - : \
6565 - : "memory" ); \
6566 - __res; \
6567 -})
6568 -
6569 -#define _hypercall1(type, name, a1) \
6570 -({ \
6571 - type __res; \
6572 - long __ign1; \
6573 - asm volatile ( \
6574 - HYPERCALL_STR(name) \
6575 - : "=a" (__res), "=b" (__ign1) \
6576 - : "1" ((long)(a1)) \
6577 - : "memory" ); \
6578 - __res; \
6579 -})
6580 -
6581 -#define _hypercall2(type, name, a1, a2) \
6582 -({ \
6583 - type __res; \
6584 - long __ign1, __ign2; \
6585 - asm volatile ( \
6586 - HYPERCALL_STR(name) \
6587 - : "=a" (__res), "=b" (__ign1), "=c" (__ign2) \
6588 - : "1" ((long)(a1)), "2" ((long)(a2)) \
6589 - : "memory" ); \
6590 - __res; \
6591 -})
6592 -
6593 -#define _hypercall3(type, name, a1, a2, a3) \
6594 -({ \
6595 - type __res; \
6596 - long __ign1, __ign2, __ign3; \
6597 - asm volatile ( \
6598 - HYPERCALL_STR(name) \
6599 - : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
6600 - "=d" (__ign3) \
6601 - : "1" ((long)(a1)), "2" ((long)(a2)), \
6602 - "3" ((long)(a3)) \
6603 - : "memory" ); \
6604 - __res; \
6605 -})
6606 -
6607 -#define _hypercall4(type, name, a1, a2, a3, a4) \
6608 -({ \
6609 - type __res; \
6610 - long __ign1, __ign2, __ign3, __ign4; \
6611 - asm volatile ( \
6612 - HYPERCALL_STR(name) \
6613 - : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
6614 - "=d" (__ign3), "=S" (__ign4) \
6615 - : "1" ((long)(a1)), "2" ((long)(a2)), \
6616 - "3" ((long)(a3)), "4" ((long)(a4)) \
6617 - : "memory" ); \
6618 - __res; \
6619 -})
6620 -
6621 -#define _hypercall5(type, name, a1, a2, a3, a4, a5) \
6622 -({ \
6623 - type __res; \
6624 - long __ign1, __ign2, __ign3, __ign4, __ign5; \
6625 - asm volatile ( \
6626 - HYPERCALL_STR(name) \
6627 - : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
6628 - "=d" (__ign3), "=S" (__ign4), "=D" (__ign5) \
6629 - : "1" ((long)(a1)), "2" ((long)(a2)), \
6630 - "3" ((long)(a3)), "4" ((long)(a4)), \
6631 - "5" ((long)(a5)) \
6632 - : "memory" ); \
6633 - __res; \
6634 -})
6635 -
6636 -#define _hypercall(type, op, a1, a2, a3, a4, a5) \
6637 -({ \
6638 - type __res; \
6639 - register typeof((a1)+0) __arg1 asm("ebx") = (a1); \
6640 - register typeof((a2)+0) __arg2 asm("ecx") = (a2); \
6641 - register typeof((a3)+0) __arg3 asm("edx") = (a3); \
6642 - register typeof((a4)+0) __arg4 asm("esi") = (a4); \
6643 - register typeof((a5)+0) __arg5 asm("edi") = (a5); \
6644 - asm volatile ( \
6645 - "call *%6" \
6646 - : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
6647 - "+r" (__arg3), "+r" (__arg4), "+r" (__arg5) \
6648 - : "0" (hypercall_page + (op) * 32) \
6649 - : "memory" ); \
6650 - __res; \
6651 -})
6652 -
6653 -static inline int __must_check
6654 -HYPERVISOR_set_trap_table(
6655 - const trap_info_t *table)
6656 -{
6657 - return _hypercall1(int, set_trap_table, table);
6658 -}
6659 -
6660 -static inline int __must_check
6661 -HYPERVISOR_mmu_update(
6662 - mmu_update_t *req, unsigned int count, unsigned int *success_count,
6663 - domid_t domid)
6664 -{
6665 - if (arch_use_lazy_mmu_mode())
6666 - return xen_multi_mmu_update(req, count, success_count, domid);
6667 - return _hypercall4(int, mmu_update, req, count, success_count, domid);
6668 -}
6669 -
6670 -static inline int __must_check
6671 -HYPERVISOR_mmuext_op(
6672 - struct mmuext_op *op, unsigned int count, unsigned int *success_count,
6673 - domid_t domid)
6674 -{
6675 - if (arch_use_lazy_mmu_mode())
6676 - return xen_multi_mmuext_op(op, count, success_count, domid);
6677 - return _hypercall4(int, mmuext_op, op, count, success_count, domid);
6678 -}
6679 -
6680 -static inline int __must_check
6681 -HYPERVISOR_set_gdt(
6682 - unsigned long *frame_list, unsigned int entries)
6683 -{
6684 - return _hypercall2(int, set_gdt, frame_list, entries);
6685 -}
6686 -
6687 -static inline int __must_check
6688 -HYPERVISOR_stack_switch(
6689 - unsigned long ss, unsigned long esp)
6690 -{
6691 - return _hypercall2(int, stack_switch, ss, esp);
6692 -}
6693 +#define HYPERCALL_arg1 "ebx"
6694 +#define HYPERCALL_arg2 "ecx"
6695 +#define HYPERCALL_arg3 "edx"
6696 +#define HYPERCALL_arg4 "esi"
6697 +#define HYPERCALL_arg5 "edi"
6698
6699 +#if CONFIG_XEN_COMPAT <= 0x030002
6700 static inline int __must_check
6701 HYPERVISOR_set_callbacks(
6702 unsigned long event_selector, unsigned long event_address,
6703 @@ -195,80 +14,24 @@ HYPERVISOR_set_callbacks(
6704 event_selector, event_address,
6705 failsafe_selector, failsafe_address);
6706 }
6707 -
6708 -static inline int
6709 -HYPERVISOR_fpu_taskswitch(
6710 - int set)
6711 -{
6712 - return _hypercall1(int, fpu_taskswitch, set);
6713 -}
6714 -
6715 -static inline int __must_check
6716 -HYPERVISOR_sched_op_compat(
6717 - int cmd, unsigned long arg)
6718 -{
6719 - return _hypercall2(int, sched_op_compat, cmd, arg);
6720 -}
6721 -
6722 -static inline int __must_check
6723 -HYPERVISOR_sched_op(
6724 - int cmd, void *arg)
6725 -{
6726 - return _hypercall2(int, sched_op, cmd, arg);
6727 -}
6728 +#endif
6729
6730 static inline long __must_check
6731 HYPERVISOR_set_timer_op(
6732 u64 timeout)
6733 {
6734 - unsigned long timeout_hi = (unsigned long)(timeout>>32);
6735 - unsigned long timeout_lo = (unsigned long)timeout;
6736 - return _hypercall2(long, set_timer_op, timeout_lo, timeout_hi);
6737 -}
6738 -
6739 -static inline int __must_check
6740 -HYPERVISOR_platform_op(
6741 - struct xen_platform_op *platform_op)
6742 -{
6743 - platform_op->interface_version = XENPF_INTERFACE_VERSION;
6744 - return _hypercall1(int, platform_op, platform_op);
6745 -}
6746 -
6747 -static inline int __must_check
6748 -HYPERVISOR_set_debugreg(
6749 - unsigned int reg, unsigned long value)
6750 -{
6751 - return _hypercall2(int, set_debugreg, reg, value);
6752 -}
6753 -
6754 -static inline unsigned long __must_check
6755 -HYPERVISOR_get_debugreg(
6756 - unsigned int reg)
6757 -{
6758 - return _hypercall1(unsigned long, get_debugreg, reg);
6759 + return _hypercall2(long, set_timer_op,
6760 + (unsigned long)timeout,
6761 + (unsigned long)(timeout>>32));
6762 }
6763
6764 static inline int __must_check
6765 HYPERVISOR_update_descriptor(
6766 u64 ma, u64 desc)
6767 {
6768 - return _hypercall4(int, update_descriptor, ma, ma>>32, desc, desc>>32);
6769 -}
6770 -
6771 -static inline int __must_check
6772 -HYPERVISOR_memory_op(
6773 - unsigned int cmd, void *arg)
6774 -{
6775 - if (arch_use_lazy_mmu_mode())
6776 - xen_multicall_flush(false);
6777 - return _hypercall2(int, memory_op, cmd, arg);
6778 -}
6779 -
6780 -static inline int __must_check
6781 -HYPERVISOR_multicall(
6782 - multicall_entry_t *call_list, unsigned int nr_calls)
6783 -{
6784 - return _hypercall2(int, multicall, call_list, nr_calls);
6785 + return _hypercall4(int, update_descriptor,
6786 + (unsigned long)ma, (unsigned long)(ma>>32),
6787 + (unsigned long)desc, (unsigned long)(desc>>32));
6788 }
6789
6790 static inline int __must_check
6791 @@ -287,67 +50,6 @@ HYPERVISOR_update_va_mapping(
6792 }
6793
6794 static inline int __must_check
6795 -HYPERVISOR_event_channel_op(
6796 - int cmd, void *arg)
6797 -{
6798 - int rc = _hypercall2(int, event_channel_op, cmd, arg);
6799 -
6800 -#if CONFIG_XEN_COMPAT <= 0x030002
6801 - if (unlikely(rc == -ENOSYS)) {
6802 - struct evtchn_op op;
6803 - op.cmd = cmd;
6804 - memcpy(&op.u, arg, sizeof(op.u));
6805 - rc = _hypercall1(int, event_channel_op_compat, &op);
6806 - memcpy(arg, &op.u, sizeof(op.u));
6807 - }
6808 -#endif
6809 -
6810 - return rc;
6811 -}
6812 -
6813 -static inline int __must_check
6814 -HYPERVISOR_xen_version(
6815 - int cmd, void *arg)
6816 -{
6817 - return _hypercall2(int, xen_version, cmd, arg);
6818 -}
6819 -
6820 -static inline int __must_check
6821 -HYPERVISOR_console_io(
6822 - int cmd, unsigned int count, char *str)
6823 -{
6824 - return _hypercall3(int, console_io, cmd, count, str);
6825 -}
6826 -
6827 -static inline int __must_check
6828 -HYPERVISOR_physdev_op(
6829 - int cmd, void *arg)
6830 -{
6831 - int rc = _hypercall2(int, physdev_op, cmd, arg);
6832 -
6833 -#if CONFIG_XEN_COMPAT <= 0x030002
6834 - if (unlikely(rc == -ENOSYS)) {
6835 - struct physdev_op op;
6836 - op.cmd = cmd;
6837 - memcpy(&op.u, arg, sizeof(op.u));
6838 - rc = _hypercall1(int, physdev_op_compat, &op);
6839 - memcpy(arg, &op.u, sizeof(op.u));
6840 - }
6841 -#endif
6842 -
6843 - return rc;
6844 -}
6845 -
6846 -static inline int __must_check
6847 -HYPERVISOR_grant_table_op(
6848 - unsigned int cmd, void *uop, unsigned int count)
6849 -{
6850 - if (arch_use_lazy_mmu_mode())
6851 - xen_multicall_flush(false);
6852 - return _hypercall3(int, grant_table_op, cmd, uop, count);
6853 -}
6854 -
6855 -static inline int __must_check
6856 HYPERVISOR_update_va_mapping_otherdomain(
6857 unsigned long va, pte_t new_val, unsigned long flags, domid_t domid)
6858 {
6859 @@ -358,80 +60,3 @@ HYPERVISOR_update_va_mapping_otherdomain
6860 return _hypercall5(int, update_va_mapping_otherdomain, va,
6861 new_val.pte_low, pte_hi, flags, domid);
6862 }
6863 -
6864 -static inline int __must_check
6865 -HYPERVISOR_vm_assist(
6866 - unsigned int cmd, unsigned int type)
6867 -{
6868 - return _hypercall2(int, vm_assist, cmd, type);
6869 -}
6870 -
6871 -static inline int __must_check
6872 -HYPERVISOR_vcpu_op(
6873 - int cmd, unsigned int vcpuid, void *extra_args)
6874 -{
6875 - return _hypercall3(int, vcpu_op, cmd, vcpuid, extra_args);
6876 -}
6877 -
6878 -static inline int __must_check
6879 -HYPERVISOR_suspend(
6880 - unsigned long srec)
6881 -{
6882 - struct sched_shutdown sched_shutdown = {
6883 - .reason = SHUTDOWN_suspend
6884 - };
6885 -
6886 - int rc = _hypercall3(int, sched_op, SCHEDOP_shutdown,
6887 - &sched_shutdown, srec);
6888 -
6889 -#if CONFIG_XEN_COMPAT <= 0x030002
6890 - if (rc == -ENOSYS)
6891 - rc = _hypercall3(int, sched_op_compat, SCHEDOP_shutdown,
6892 - SHUTDOWN_suspend, srec);
6893 -#endif
6894 -
6895 - return rc;
6896 -}
6897 -
6898 -#if CONFIG_XEN_COMPAT <= 0x030002
6899 -static inline int
6900 -HYPERVISOR_nmi_op(
6901 - unsigned long op, void *arg)
6902 -{
6903 - return _hypercall2(int, nmi_op, op, arg);
6904 -}
6905 -#endif
6906 -
6907 -#ifndef CONFIG_XEN
6908 -static inline unsigned long __must_check
6909 -HYPERVISOR_hvm_op(
6910 - int op, void *arg)
6911 -{
6912 - return _hypercall2(unsigned long, hvm_op, op, arg);
6913 -}
6914 -#endif
6915 -
6916 -static inline int __must_check
6917 -HYPERVISOR_callback_op(
6918 - int cmd, const void *arg)
6919 -{
6920 - return _hypercall2(int, callback_op, cmd, arg);
6921 -}
6922 -
6923 -static inline int __must_check
6924 -HYPERVISOR_xenoprof_op(
6925 - int op, void *arg)
6926 -{
6927 - return _hypercall2(int, xenoprof_op, op, arg);
6928 -}
6929 -
6930 -static inline int __must_check
6931 -HYPERVISOR_kexec_op(
6932 - unsigned long op, void *args)
6933 -{
6934 - return _hypercall2(int, kexec_op, op, args);
6935 -}
6936 -
6937 -
6938 -
6939 -#endif /* __HYPERCALL_H__ */
6940 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/hypercall_64.h 2009-03-04 11:28:34.000000000 +0100
6941 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/hypercall_64.h 2009-02-16 16:18:36.000000000 +0100
6942 @@ -1,197 +1,10 @@
6943 -/******************************************************************************
6944 - * hypercall.h
6945 - *
6946 - * Linux-specific hypervisor handling.
6947 - *
6948 - * Copyright (c) 2002-2004, K A Fraser
6949 - *
6950 - * 64-bit updates:
6951 - * Benjamin Liu <benjamin.liu@intel.com>
6952 - * Jun Nakajima <jun.nakajima@intel.com>
6953 - *
6954 - * This program is free software; you can redistribute it and/or
6955 - * modify it under the terms of the GNU General Public License version 2
6956 - * as published by the Free Software Foundation; or, when distributed
6957 - * separately from the Linux kernel or incorporated into other
6958 - * software packages, subject to the following license:
6959 - *
6960 - * Permission is hereby granted, free of charge, to any person obtaining a copy
6961 - * of this source file (the "Software"), to deal in the Software without
6962 - * restriction, including without limitation the rights to use, copy, modify,
6963 - * merge, publish, distribute, sublicense, and/or sell copies of the Software,
6964 - * and to permit persons to whom the Software is furnished to do so, subject to
6965 - * the following conditions:
6966 - *
6967 - * The above copyright notice and this permission notice shall be included in
6968 - * all copies or substantial portions of the Software.
6969 - *
6970 - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
6971 - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
6972 - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
6973 - * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
6974 - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
6975 - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
6976 - * IN THE SOFTWARE.
6977 - */
6978 -
6979 -#ifndef __HYPERCALL_H__
6980 -#define __HYPERCALL_H__
6981 -
6982 -#include <linux/string.h> /* memcpy() */
6983 -#include <linux/stringify.h>
6984 -
6985 -#ifndef __HYPERVISOR_H__
6986 -# error "please don't include this file directly"
6987 -#endif
6988 -
6989 -#ifdef CONFIG_XEN
6990 -#define HYPERCALL_STR(name) \
6991 - "call hypercall_page + ("__stringify(__HYPERVISOR_##name)" * 32)"
6992 -#else
6993 -#define HYPERCALL_STR(name) \
6994 - "mov $("__stringify(__HYPERVISOR_##name)" * 32),%%eax; "\
6995 - "add hypercall_stubs(%%rip),%%rax; " \
6996 - "call *%%rax"
6997 -#endif
6998 -
6999 -#define _hypercall0(type, name) \
7000 -({ \
7001 - type __res; \
7002 - asm volatile ( \
7003 - HYPERCALL_STR(name) \
7004 - : "=a" (__res) \
7005 - : \
7006 - : "memory" ); \
7007 - __res; \
7008 -})
7009 -
7010 -#define _hypercall1(type, name, a1) \
7011 -({ \
7012 - type __res; \
7013 - long __ign1; \
7014 - asm volatile ( \
7015 - HYPERCALL_STR(name) \
7016 - : "=a" (__res), "=D" (__ign1) \
7017 - : "1" ((long)(a1)) \
7018 - : "memory" ); \
7019 - __res; \
7020 -})
7021 -
7022 -#define _hypercall2(type, name, a1, a2) \
7023 -({ \
7024 - type __res; \
7025 - long __ign1, __ign2; \
7026 - asm volatile ( \
7027 - HYPERCALL_STR(name) \
7028 - : "=a" (__res), "=D" (__ign1), "=S" (__ign2) \
7029 - : "1" ((long)(a1)), "2" ((long)(a2)) \
7030 - : "memory" ); \
7031 - __res; \
7032 -})
7033 -
7034 -#define _hypercall3(type, name, a1, a2, a3) \
7035 -({ \
7036 - type __res; \
7037 - long __ign1, __ign2, __ign3; \
7038 - asm volatile ( \
7039 - HYPERCALL_STR(name) \
7040 - : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
7041 - "=d" (__ign3) \
7042 - : "1" ((long)(a1)), "2" ((long)(a2)), \
7043 - "3" ((long)(a3)) \
7044 - : "memory" ); \
7045 - __res; \
7046 -})
7047 -
7048 -#define _hypercall4(type, name, a1, a2, a3, a4) \
7049 -({ \
7050 - type __res; \
7051 - long __ign1, __ign2, __ign3; \
7052 - register long __arg4 asm("r10") = (long)(a4); \
7053 - asm volatile ( \
7054 - HYPERCALL_STR(name) \
7055 - : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
7056 - "=d" (__ign3), "+r" (__arg4) \
7057 - : "1" ((long)(a1)), "2" ((long)(a2)), \
7058 - "3" ((long)(a3)) \
7059 - : "memory" ); \
7060 - __res; \
7061 -})
7062 -
7063 -#define _hypercall5(type, name, a1, a2, a3, a4, a5) \
7064 -({ \
7065 - type __res; \
7066 - long __ign1, __ign2, __ign3; \
7067 - register long __arg4 asm("r10") = (long)(a4); \
7068 - register long __arg5 asm("r8") = (long)(a5); \
7069 - asm volatile ( \
7070 - HYPERCALL_STR(name) \
7071 - : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
7072 - "=d" (__ign3), "+r" (__arg4), "+r" (__arg5) \
7073 - : "1" ((long)(a1)), "2" ((long)(a2)), \
7074 - "3" ((long)(a3)) \
7075 - : "memory" ); \
7076 - __res; \
7077 -})
7078 -
7079 -#define _hypercall(type, op, a1, a2, a3, a4, a5) \
7080 -({ \
7081 - type __res; \
7082 - register typeof((a1)+0) __arg1 asm("rdi") = (a1); \
7083 - register typeof((a2)+0) __arg2 asm("rsi") = (a2); \
7084 - register typeof((a3)+0) __arg3 asm("rdx") = (a3); \
7085 - register typeof((a4)+0) __arg4 asm("r10") = (a4); \
7086 - register typeof((a5)+0) __arg5 asm("r8") = (a5); \
7087 - asm volatile ( \
7088 - "call *%6" \
7089 - : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
7090 - "+r" (__arg3), "+r" (__arg4), "+r" (__arg5) \
7091 - : "0" (hypercall_page + (op) * 32) \
7092 - : "memory" ); \
7093 - __res; \
7094 -})
7095 -
7096 -static inline int __must_check
7097 -HYPERVISOR_set_trap_table(
7098 - const trap_info_t *table)
7099 -{
7100 - return _hypercall1(int, set_trap_table, table);
7101 -}
7102 -
7103 -static inline int __must_check
7104 -HYPERVISOR_mmu_update(
7105 - mmu_update_t *req, unsigned int count, unsigned int *success_count,
7106 - domid_t domid)
7107 -{
7108 - if (arch_use_lazy_mmu_mode())
7109 - return xen_multi_mmu_update(req, count, success_count, domid);
7110 - return _hypercall4(int, mmu_update, req, count, success_count, domid);
7111 -}
7112 -
7113 -static inline int __must_check
7114 -HYPERVISOR_mmuext_op(
7115 - struct mmuext_op *op, unsigned int count, unsigned int *success_count,
7116 - domid_t domid)
7117 -{
7118 - if (arch_use_lazy_mmu_mode())
7119 - return xen_multi_mmuext_op(op, count, success_count, domid);
7120 - return _hypercall4(int, mmuext_op, op, count, success_count, domid);
7121 -}
7122 -
7123 -static inline int __must_check
7124 -HYPERVISOR_set_gdt(
7125 - unsigned long *frame_list, unsigned int entries)
7126 -{
7127 - return _hypercall2(int, set_gdt, frame_list, entries);
7128 -}
7129 -
7130 -static inline int __must_check
7131 -HYPERVISOR_stack_switch(
7132 - unsigned long ss, unsigned long esp)
7133 -{
7134 - return _hypercall2(int, stack_switch, ss, esp);
7135 -}
7136 +#define HYPERCALL_arg1 "rdi"
7137 +#define HYPERCALL_arg2 "rsi"
7138 +#define HYPERCALL_arg3 "rdx"
7139 +#define HYPERCALL_arg4 "r10"
7140 +#define HYPERCALL_arg5 "r8"
7141
7142 +#if CONFIG_XEN_COMPAT <= 0x030002
7143 static inline int __must_check
7144 HYPERVISOR_set_callbacks(
7145 unsigned long event_address, unsigned long failsafe_address,
7146 @@ -200,27 +13,7 @@ HYPERVISOR_set_callbacks(
7147 return _hypercall3(int, set_callbacks,
7148 event_address, failsafe_address, syscall_address);
7149 }
7150 -
7151 -static inline int
7152 -HYPERVISOR_fpu_taskswitch(
7153 - int set)
7154 -{
7155 - return _hypercall1(int, fpu_taskswitch, set);
7156 -}
7157 -
7158 -static inline int __must_check
7159 -HYPERVISOR_sched_op_compat(
7160 - int cmd, unsigned long arg)
7161 -{
7162 - return _hypercall2(int, sched_op_compat, cmd, arg);
7163 -}
7164 -
7165 -static inline int __must_check
7166 -HYPERVISOR_sched_op(
7167 - int cmd, void *arg)
7168 -{
7169 - return _hypercall2(int, sched_op, cmd, arg);
7170 -}
7171 +#endif
7172
7173 static inline long __must_check
7174 HYPERVISOR_set_timer_op(
7175 @@ -230,28 +23,6 @@ HYPERVISOR_set_timer_op(
7176 }
7177
7178 static inline int __must_check
7179 -HYPERVISOR_platform_op(
7180 - struct xen_platform_op *platform_op)
7181 -{
7182 - platform_op->interface_version = XENPF_INTERFACE_VERSION;
7183 - return _hypercall1(int, platform_op, platform_op);
7184 -}
7185 -
7186 -static inline int __must_check
7187 -HYPERVISOR_set_debugreg(
7188 - unsigned int reg, unsigned long value)
7189 -{
7190 - return _hypercall2(int, set_debugreg, reg, value);
7191 -}
7192 -
7193 -static inline unsigned long __must_check
7194 -HYPERVISOR_get_debugreg(
7195 - unsigned int reg)
7196 -{
7197 - return _hypercall1(unsigned long, get_debugreg, reg);
7198 -}
7199 -
7200 -static inline int __must_check
7201 HYPERVISOR_update_descriptor(
7202 unsigned long ma, unsigned long word)
7203 {
7204 @@ -259,22 +30,6 @@ HYPERVISOR_update_descriptor(
7205 }
7206
7207 static inline int __must_check
7208 -HYPERVISOR_memory_op(
7209 - unsigned int cmd, void *arg)
7210 -{
7211 - if (arch_use_lazy_mmu_mode())
7212 - xen_multicall_flush(false);
7213 - return _hypercall2(int, memory_op, cmd, arg);
7214 -}
7215 -
7216 -static inline int __must_check
7217 -HYPERVISOR_multicall(
7218 - multicall_entry_t *call_list, unsigned int nr_calls)
7219 -{
7220 - return _hypercall2(int, multicall, call_list, nr_calls);
7221 -}
7222 -
7223 -static inline int __must_check
7224 HYPERVISOR_update_va_mapping(
7225 unsigned long va, pte_t new_val, unsigned long flags)
7226 {
7227 @@ -284,67 +39,6 @@ HYPERVISOR_update_va_mapping(
7228 }
7229
7230 static inline int __must_check
7231 -HYPERVISOR_event_channel_op(
7232 - int cmd, void *arg)
7233 -{
7234 - int rc = _hypercall2(int, event_channel_op, cmd, arg);
7235 -
7236 -#if CONFIG_XEN_COMPAT <= 0x030002
7237 - if (unlikely(rc == -ENOSYS)) {
7238 - struct evtchn_op op;
7239 - op.cmd = cmd;
7240 - memcpy(&op.u, arg, sizeof(op.u));
7241 - rc = _hypercall1(int, event_channel_op_compat, &op);
7242 - memcpy(arg, &op.u, sizeof(op.u));
7243 - }
7244 -#endif
7245 -
7246 - return rc;
7247 -}
7248 -
7249 -static inline int __must_check
7250 -HYPERVISOR_xen_version(
7251 - int cmd, void *arg)
7252 -{
7253 - return _hypercall2(int, xen_version, cmd, arg);
7254 -}
7255 -
7256 -static inline int __must_check
7257 -HYPERVISOR_console_io(
7258 - int cmd, unsigned int count, char *str)
7259 -{
7260 - return _hypercall3(int, console_io, cmd, count, str);
7261 -}
7262 -
7263 -static inline int __must_check
7264 -HYPERVISOR_physdev_op(
7265 - int cmd, void *arg)
7266 -{
7267 - int rc = _hypercall2(int, physdev_op, cmd, arg);
7268 -
7269 -#if CONFIG_XEN_COMPAT <= 0x030002
7270 - if (unlikely(rc == -ENOSYS)) {
7271 - struct physdev_op op;
7272 - op.cmd = cmd;
7273 - memcpy(&op.u, arg, sizeof(op.u));
7274 - rc = _hypercall1(int, physdev_op_compat, &op);
7275 - memcpy(arg, &op.u, sizeof(op.u));
7276 - }
7277 -#endif
7278 -
7279 - return rc;
7280 -}
7281 -
7282 -static inline int __must_check
7283 -HYPERVISOR_grant_table_op(
7284 - unsigned int cmd, void *uop, unsigned int count)
7285 -{
7286 - if (arch_use_lazy_mmu_mode())
7287 - xen_multicall_flush(false);
7288 - return _hypercall3(int, grant_table_op, cmd, uop, count);
7289 -}
7290 -
7291 -static inline int __must_check
7292 HYPERVISOR_update_va_mapping_otherdomain(
7293 unsigned long va, pte_t new_val, unsigned long flags, domid_t domid)
7294 {
7295 @@ -353,83 +47,8 @@ HYPERVISOR_update_va_mapping_otherdomain
7296 }
7297
7298 static inline int __must_check
7299 -HYPERVISOR_vm_assist(
7300 - unsigned int cmd, unsigned int type)
7301 -{
7302 - return _hypercall2(int, vm_assist, cmd, type);
7303 -}
7304 -
7305 -static inline int __must_check
7306 -HYPERVISOR_vcpu_op(
7307 - int cmd, unsigned int vcpuid, void *extra_args)
7308 -{
7309 - return _hypercall3(int, vcpu_op, cmd, vcpuid, extra_args);
7310 -}
7311 -
7312 -static inline int __must_check
7313 HYPERVISOR_set_segment_base(
7314 int reg, unsigned long value)
7315 {
7316 return _hypercall2(int, set_segment_base, reg, value);
7317 }
7318 -
7319 -static inline int __must_check
7320 -HYPERVISOR_suspend(
7321 - unsigned long srec)
7322 -{
7323 - struct sched_shutdown sched_shutdown = {
7324 - .reason = SHUTDOWN_suspend
7325 - };
7326 -
7327 - int rc = _hypercall3(int, sched_op, SCHEDOP_shutdown,
7328 - &sched_shutdown, srec);
7329 -
7330 -#if CONFIG_XEN_COMPAT <= 0x030002
7331 - if (rc == -ENOSYS)
7332 - rc = _hypercall3(int, sched_op_compat, SCHEDOP_shutdown,
7333 - SHUTDOWN_suspend, srec);
7334 -#endif
7335 -
7336 - return rc;
7337 -}
7338 -
7339 -#if CONFIG_XEN_COMPAT <= 0x030002
7340 -static inline int
7341 -HYPERVISOR_nmi_op(
7342 - unsigned long op, void *arg)
7343 -{
7344 - return _hypercall2(int, nmi_op, op, arg);
7345 -}
7346 -#endif
7347 -
7348 -#ifndef CONFIG_XEN
7349 -static inline unsigned long __must_check
7350 -HYPERVISOR_hvm_op(
7351 - int op, void *arg)
7352 -{
7353 - return _hypercall2(unsigned long, hvm_op, op, arg);
7354 -}
7355 -#endif
7356 -
7357 -static inline int __must_check
7358 -HYPERVISOR_callback_op(
7359 - int cmd, const void *arg)
7360 -{
7361 - return _hypercall2(int, callback_op, cmd, arg);
7362 -}
7363 -
7364 -static inline int __must_check
7365 -HYPERVISOR_xenoprof_op(
7366 - int op, void *arg)
7367 -{
7368 - return _hypercall2(int, xenoprof_op, op, arg);
7369 -}
7370 -
7371 -static inline int __must_check
7372 -HYPERVISOR_kexec_op(
7373 - unsigned long op, void *args)
7374 -{
7375 - return _hypercall2(int, kexec_op, op, args);
7376 -}
7377 -
7378 -#endif /* __HYPERCALL_H__ */
7379 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/hypervisor.h 2009-03-04 11:28:11.000000000 +0100
7380 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/hypervisor.h 2009-02-16 16:18:36.000000000 +0100
7381 @@ -194,7 +194,6 @@ static inline void xen_multicall_flush(b
7382 extern char hypercall_page[PAGE_SIZE];
7383 #else
7384 extern char *hypercall_stubs;
7385 -#define hypercall_page hypercall_stubs
7386 #define is_running_on_xen() (!!hypercall_stubs)
7387 #endif
7388
7389 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7390 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/io.h 2009-02-16 16:18:36.000000000 +0100
7391 @@ -0,0 +1,5 @@
7392 +#ifdef CONFIG_X86_32
7393 +# include "io_32.h"
7394 +#else
7395 +# include "io_64.h"
7396 +#endif
7397 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7398 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/irqflags.h 2009-02-16 16:18:36.000000000 +0100
7399 @@ -0,0 +1,5 @@
7400 +#ifdef CONFIG_X86_32
7401 +# include "irqflags_32.h"
7402 +#else
7403 +# include "irqflags_64.h"
7404 +#endif
7405 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/irqflags_32.h 2008-12-15 11:27:22.000000000 +0100
7406 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/irqflags_32.h 2009-02-16 16:18:36.000000000 +0100
7407 @@ -148,6 +148,23 @@ static inline int raw_irqs_disabled_flag
7408 \
7409 raw_irqs_disabled_flags(flags); \
7410 })
7411 +
7412 +/*
7413 + * makes the traced hardirq state match with the machine state
7414 + *
7415 + * should be a rarely used function, only in places where its
7416 + * otherwise impossible to know the irq state, like in traps.
7417 + */
7418 +static inline void trace_hardirqs_fixup_flags(unsigned long flags)
7419 +{
7420 + if (raw_irqs_disabled_flags(flags))
7421 + trace_hardirqs_off();
7422 + else
7423 + trace_hardirqs_on();
7424 +}
7425 +
7426 +#define trace_hardirqs_fixup() \
7427 + trace_hardirqs_fixup_flags(__raw_local_save_flags())
7428 #endif /* __ASSEMBLY__ */
7429
7430 /*
7431 @@ -179,4 +196,17 @@ static inline int raw_irqs_disabled_flag
7432 # define TRACE_IRQS_OFF
7433 #endif
7434
7435 +#ifdef CONFIG_DEBUG_LOCK_ALLOC
7436 +# define LOCKDEP_SYS_EXIT \
7437 + pushl %eax; \
7438 + pushl %ecx; \
7439 + pushl %edx; \
7440 + call lockdep_sys_exit; \
7441 + popl %edx; \
7442 + popl %ecx; \
7443 + popl %eax;
7444 +#else
7445 +# define LOCKDEP_SYS_EXIT
7446 +#endif
7447 +
7448 #endif
7449 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/irqflags_64.h 2008-12-15 11:27:22.000000000 +0100
7450 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/irqflags_64.h 2009-02-16 16:18:36.000000000 +0100
7451 @@ -116,6 +116,22 @@ static inline int raw_irqs_disabled_flag
7452 })
7453
7454 /*
7455 + * makes the traced hardirq state match with the machine state
7456 + *
7457 + * should be a rarely used function, only in places where its
7458 + * otherwise impossible to know the irq state, like in traps.
7459 + */
7460 +static inline void trace_hardirqs_fixup_flags(unsigned long flags)
7461 +{
7462 + if (raw_irqs_disabled_flags(flags))
7463 + trace_hardirqs_off();
7464 + else
7465 + trace_hardirqs_on();
7466 +}
7467 +
7468 +#define trace_hardirqs_fixup() \
7469 + trace_hardirqs_fixup_flags(__raw_local_save_flags())
7470 +/*
7471 * Used in the idle loop; sti takes one instruction cycle
7472 * to complete:
7473 */
7474 @@ -143,6 +159,20 @@ static inline void halt(void)
7475 # define TRACE_IRQS_ON
7476 # define TRACE_IRQS_OFF
7477 # endif
7478 +# ifdef CONFIG_DEBUG_LOCK_ALLOC
7479 +# define LOCKDEP_SYS_EXIT call lockdep_sys_exit_thunk
7480 +# define LOCKDEP_SYS_EXIT_IRQ \
7481 + TRACE_IRQS_ON; \
7482 + sti; \
7483 + SAVE_REST; \
7484 + LOCKDEP_SYS_EXIT; \
7485 + RESTORE_REST; \
7486 + cli; \
7487 + TRACE_IRQS_OFF;
7488 +# else
7489 +# define LOCKDEP_SYS_EXIT
7490 +# define LOCKDEP_SYS_EXIT_IRQ
7491 +# endif
7492 #endif
7493
7494 #endif
7495 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7496 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/maddr.h 2009-02-16 16:18:36.000000000 +0100
7497 @@ -0,0 +1,5 @@
7498 +#ifdef CONFIG_X86_32
7499 +# include "maddr_32.h"
7500 +#else
7501 +# include "maddr_64.h"
7502 +#endif
7503 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7504 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/mmu_context.h 2009-02-16 16:18:36.000000000 +0100
7505 @@ -0,0 +1,5 @@
7506 +#ifdef CONFIG_X86_32
7507 +# include "mmu_context_32.h"
7508 +#else
7509 +# include "mmu_context_64.h"
7510 +#endif
7511 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7512 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/page.h 2009-02-16 16:18:36.000000000 +0100
7513 @@ -0,0 +1,13 @@
7514 +#ifdef __KERNEL__
7515 +# ifdef CONFIG_X86_32
7516 +# include "page_32.h"
7517 +# else
7518 +# include "page_64.h"
7519 +# endif
7520 +#else
7521 +# ifdef __i386__
7522 +# include "page_32.h"
7523 +# else
7524 +# include "page_64.h"
7525 +# endif
7526 +#endif
7527 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/page_64.h 2009-02-16 16:17:21.000000000 +0100
7528 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/page_64.h 2009-02-16 16:18:36.000000000 +0100
7529 @@ -207,6 +207,7 @@ static inline unsigned long __phys_addr(
7530 VM_READ | VM_WRITE | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
7531
7532 #define __HAVE_ARCH_GATE_AREA 1
7533 +#define vmemmap ((struct page *)VMEMMAP_START)
7534
7535 #include <asm-generic/memory_model.h>
7536 #include <asm-generic/page.h>
7537 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7538 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pci.h 2009-02-16 16:18:36.000000000 +0100
7539 @@ -0,0 +1,100 @@
7540 +#ifndef __x86_PCI_H
7541 +#define __x86_PCI_H
7542 +
7543 +#include <linux/mm.h> /* for struct page */
7544 +#include <linux/types.h>
7545 +#include <linux/slab.h>
7546 +#include <linux/string.h>
7547 +#include <asm/scatterlist.h>
7548 +#include <asm/io.h>
7549 +
7550 +
7551 +#ifdef __KERNEL__
7552 +
7553 +struct pci_sysdata {
7554 + int domain; /* PCI domain */
7555 + int node; /* NUMA node */
7556 +#ifdef CONFIG_X86_64
7557 + void* iommu; /* IOMMU private data */
7558 +#endif
7559 +#ifdef CONFIG_XEN_PCIDEV_FRONTEND
7560 + struct pcifront_device *pdev;
7561 +#endif
7562 +};
7563 +
7564 +/* scan a bus after allocating a pci_sysdata for it */
7565 +extern struct pci_bus *pci_scan_bus_with_sysdata(int busno);
7566 +
7567 +static inline int pci_domain_nr(struct pci_bus *bus)
7568 +{
7569 + struct pci_sysdata *sd = bus->sysdata;
7570 + return sd->domain;
7571 +}
7572 +
7573 +static inline int pci_proc_domain(struct pci_bus *bus)
7574 +{
7575 + return pci_domain_nr(bus);
7576 +}
7577 +
7578 +
7579 +/* Can be used to override the logic in pci_scan_bus for skipping
7580 + already-configured bus numbers - to be used for buggy BIOSes
7581 + or architectures with incomplete PCI setup by the loader */
7582 +
7583 +#ifdef CONFIG_PCI
7584 +extern unsigned int pcibios_assign_all_busses(void);
7585 +#else
7586 +#define pcibios_assign_all_busses() 0
7587 +#endif
7588 +
7589 +#include <asm/hypervisor.h>
7590 +#define pcibios_scan_all_fns(a, b) (!is_initial_xendomain())
7591 +
7592 +extern unsigned long pci_mem_start;
7593 +#define PCIBIOS_MIN_IO 0x1000
7594 +#define PCIBIOS_MIN_MEM (pci_mem_start)
7595 +
7596 +#define PCIBIOS_MIN_CARDBUS_IO 0x4000
7597 +
7598 +void pcibios_config_init(void);
7599 +struct pci_bus * pcibios_scan_root(int bus);
7600 +
7601 +void pcibios_set_master(struct pci_dev *dev);
7602 +void pcibios_penalize_isa_irq(int irq, int active);
7603 +struct irq_routing_table *pcibios_get_irq_routing_table(void);
7604 +int pcibios_set_irq_routing(struct pci_dev *dev, int pin, int irq);
7605 +
7606 +
7607 +#define HAVE_PCI_MMAP
7608 +extern int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
7609 + enum pci_mmap_state mmap_state, int write_combine);
7610 +
7611 +
7612 +#ifdef CONFIG_PCI
7613 +static inline void pci_dma_burst_advice(struct pci_dev *pdev,
7614 + enum pci_dma_burst_strategy *strat,
7615 + unsigned long *strategy_parameter)
7616 +{
7617 + *strat = PCI_DMA_BURST_INFINITY;
7618 + *strategy_parameter = ~0UL;
7619 +}
7620 +#endif
7621 +
7622 +
7623 +#endif /* __KERNEL__ */
7624 +
7625 +#ifdef CONFIG_X86_32
7626 +# include "pci_32.h"
7627 +#else
7628 +# include "pci_64.h"
7629 +#endif
7630 +
7631 +/* implement the pci_ DMA API in terms of the generic device dma_ one */
7632 +#include <asm-generic/pci-dma-compat.h>
7633 +
7634 +/* generic pci stuff */
7635 +#include <asm-generic/pci.h>
7636 +
7637 +
7638 +
7639 +#endif
7640 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pci_32.h 2009-02-16 16:17:21.000000000 +0100
7641 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pci_32.h 2009-02-16 16:18:36.000000000 +0100
7642 @@ -4,52 +4,10 @@
7643
7644 #ifdef __KERNEL__
7645
7646 -struct pci_sysdata {
7647 - int node; /* NUMA node */
7648 -};
7649 -
7650 -/* scan a bus after allocating a pci_sysdata for it */
7651 -extern struct pci_bus *pci_scan_bus_with_sysdata(int busno);
7652 -
7653 -#include <linux/mm.h> /* for struct page */
7654 -
7655 -/* Can be used to override the logic in pci_scan_bus for skipping
7656 - already-configured bus numbers - to be used for buggy BIOSes
7657 - or architectures with incomplete PCI setup by the loader */
7658 -
7659 -#ifdef CONFIG_PCI
7660 -extern unsigned int pcibios_assign_all_busses(void);
7661 -#else
7662 -#define pcibios_assign_all_busses() 0
7663 -#endif
7664 -
7665 -#include <asm/hypervisor.h>
7666 -#define pcibios_scan_all_fns(a, b) (!is_initial_xendomain())
7667 -
7668 -extern unsigned long pci_mem_start;
7669 -#define PCIBIOS_MIN_IO 0x1000
7670 -#define PCIBIOS_MIN_MEM (pci_mem_start)
7671 -
7672 -#define PCIBIOS_MIN_CARDBUS_IO 0x4000
7673 -
7674 -void pcibios_config_init(void);
7675 -struct pci_bus * pcibios_scan_root(int bus);
7676 -
7677 -void pcibios_set_master(struct pci_dev *dev);
7678 -void pcibios_penalize_isa_irq(int irq, int active);
7679 -struct irq_routing_table *pcibios_get_irq_routing_table(void);
7680 -int pcibios_set_irq_routing(struct pci_dev *dev, int pin, int irq);
7681 -
7682 /* Dynamic DMA mapping stuff.
7683 * i386 has everything mapped statically.
7684 */
7685
7686 -#include <linux/types.h>
7687 -#include <linux/slab.h>
7688 -#include <asm/scatterlist.h>
7689 -#include <linux/string.h>
7690 -#include <asm/io.h>
7691 -
7692 struct pci_dev;
7693
7694 #ifdef CONFIG_SWIOTLB
7695 @@ -89,31 +47,8 @@ struct pci_dev;
7696
7697 #endif
7698
7699 -#define HAVE_PCI_MMAP
7700 -extern int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
7701 - enum pci_mmap_state mmap_state, int write_combine);
7702 -
7703 -
7704 -#ifdef CONFIG_PCI
7705 -static inline void pci_dma_burst_advice(struct pci_dev *pdev,
7706 - enum pci_dma_burst_strategy *strat,
7707 - unsigned long *strategy_parameter)
7708 -{
7709 - *strat = PCI_DMA_BURST_INFINITY;
7710 - *strategy_parameter = ~0UL;
7711 -}
7712 -#endif
7713
7714 #endif /* __KERNEL__ */
7715
7716 -#ifdef CONFIG_XEN_PCIDEV_FRONTEND
7717 -#include <xen/pcifront.h>
7718 -#endif /* CONFIG_XEN_PCIDEV_FRONTEND */
7719 -
7720 -/* implement the pci_ DMA API in terms of the generic device dma_ one */
7721 -#include <asm-generic/pci-dma-compat.h>
7722 -
7723 -/* generic pci stuff */
7724 -#include <asm-generic/pci.h>
7725
7726 #endif /* __i386_PCI_H */
7727 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7728 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgalloc.h 2009-02-16 16:18:36.000000000 +0100
7729 @@ -0,0 +1,5 @@
7730 +#ifdef CONFIG_X86_32
7731 +# include "pgalloc_32.h"
7732 +#else
7733 +# include "pgalloc_64.h"
7734 +#endif
7735 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pgalloc_64.h 2009-02-16 16:17:21.000000000 +0100
7736 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgalloc_64.h 2009-02-16 16:18:36.000000000 +0100
7737 @@ -112,6 +112,8 @@ static inline void pgd_list_del(pgd_t *p
7738 spin_unlock(&pgd_lock);
7739 }
7740
7741 +extern void pgd_test_and_unpin(pgd_t *);
7742 +
7743 static inline pgd_t *pgd_alloc(struct mm_struct *mm)
7744 {
7745 /*
7746 @@ -122,6 +124,7 @@ static inline pgd_t *pgd_alloc(struct mm
7747 if (!pgd)
7748 return NULL;
7749 pgd_list_add(pgd);
7750 + pgd_test_and_unpin(pgd);
7751 /*
7752 * Copy kernel pointers in from init.
7753 * Could keep a freelist or slab cache of those because the kernel
7754 @@ -144,27 +147,7 @@ static inline pgd_t *pgd_alloc(struct mm
7755
7756 static inline void pgd_free(pgd_t *pgd)
7757 {
7758 - pte_t *ptep = virt_to_ptep(pgd);
7759 -
7760 - if (!pte_write(*ptep)) {
7761 - xen_pgd_unpin(__pa(pgd));
7762 - BUG_ON(HYPERVISOR_update_va_mapping(
7763 - (unsigned long)pgd,
7764 - pfn_pte(virt_to_phys(pgd)>>PAGE_SHIFT, PAGE_KERNEL),
7765 - 0));
7766 - }
7767 -
7768 - ptep = virt_to_ptep(__user_pgd(pgd));
7769 -
7770 - if (!pte_write(*ptep)) {
7771 - xen_pgd_unpin(__pa(__user_pgd(pgd)));
7772 - BUG_ON(HYPERVISOR_update_va_mapping(
7773 - (unsigned long)__user_pgd(pgd),
7774 - pfn_pte(virt_to_phys(__user_pgd(pgd))>>PAGE_SHIFT,
7775 - PAGE_KERNEL),
7776 - 0));
7777 - }
7778 -
7779 + pgd_test_and_unpin(pgd);
7780 pgd_list_del(pgd);
7781 free_pages((unsigned long)pgd, 1);
7782 }
7783 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7784 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgtable.h 2009-02-16 16:18:36.000000000 +0100
7785 @@ -0,0 +1,5 @@
7786 +#ifdef CONFIG_X86_32
7787 +# include "pgtable_32.h"
7788 +#else
7789 +# include "pgtable_64.h"
7790 +#endif
7791 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pgtable_32.h 2009-02-16 16:17:21.000000000 +0100
7792 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgtable_32.h 2009-02-16 16:18:36.000000000 +0100
7793 @@ -17,10 +17,7 @@
7794 #include <asm/fixmap.h>
7795 #include <linux/threads.h>
7796
7797 -#ifndef _I386_BITOPS_H
7798 -#include <asm/bitops.h>
7799 -#endif
7800 -
7801 +#include <linux/bitops.h>
7802 #include <linux/slab.h>
7803 #include <linux/list.h>
7804 #include <linux/spinlock.h>
7805 @@ -40,7 +37,7 @@ extern spinlock_t pgd_lock;
7806 extern struct page *pgd_list;
7807 void check_pgt_cache(void);
7808
7809 -void pmd_ctor(void *, struct kmem_cache *, unsigned long);
7810 +void pmd_ctor(struct kmem_cache *, void *);
7811 void pgtable_cache_init(void);
7812 void paging_init(void);
7813
7814 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pgtable_64.h 2009-02-16 16:17:21.000000000 +0100
7815 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgtable_64.h 2009-02-16 16:18:36.000000000 +0100
7816 @@ -9,7 +9,7 @@
7817 * the x86-64 page table tree.
7818 */
7819 #include <asm/processor.h>
7820 -#include <asm/bitops.h>
7821 +#include <linux/bitops.h>
7822 #include <linux/threads.h>
7823 #include <linux/sched.h>
7824 #include <asm/pda.h>
7825 @@ -139,6 +139,7 @@ static inline void pgd_clear (pgd_t * pg
7826 #define MAXMEM _AC(0x3fffffffffff, UL)
7827 #define VMALLOC_START _AC(0xffffc20000000000, UL)
7828 #define VMALLOC_END _AC(0xffffe1ffffffffff, UL)
7829 +#define VMEMMAP_START _AC(0xffffe20000000000, UL)
7830 #define MODULES_VADDR _AC(0xffffffff88000000, UL)
7831 #define MODULES_END _AC(0xfffffffffff00000, UL)
7832 #define MODULES_LEN (MODULES_END - MODULES_VADDR)
7833 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7834 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/processor.h 2009-02-16 16:18:36.000000000 +0100
7835 @@ -0,0 +1,5 @@
7836 +#ifdef CONFIG_X86_32
7837 +# include "processor_32.h"
7838 +#else
7839 +# include "processor_64.h"
7840 +#endif
7841 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/processor_32.h 2009-02-16 16:17:21.000000000 +0100
7842 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/processor_32.h 2009-02-16 16:18:36.000000000 +0100
7843 @@ -80,6 +80,7 @@ struct cpuinfo_x86 {
7844 unsigned char booted_cores; /* number of cores as seen by OS */
7845 __u8 phys_proc_id; /* Physical processor id. */
7846 __u8 cpu_core_id; /* Core id */
7847 + __u8 cpu_index; /* index into per_cpu list */
7848 #endif
7849 } __attribute__((__aligned__(SMP_CACHE_BYTES)));
7850
7851 @@ -106,14 +107,19 @@ DECLARE_PER_CPU(struct tss_struct, init_
7852 #endif
7853
7854 #ifdef CONFIG_SMP
7855 -extern struct cpuinfo_x86 cpu_data[];
7856 -#define current_cpu_data cpu_data[smp_processor_id()]
7857 +DECLARE_PER_CPU(struct cpuinfo_x86, cpu_info);
7858 +#define cpu_data(cpu) per_cpu(cpu_info, cpu)
7859 +#define current_cpu_data cpu_data(smp_processor_id())
7860 #else
7861 -#define cpu_data (&boot_cpu_data)
7862 -#define current_cpu_data boot_cpu_data
7863 +#define cpu_data(cpu) boot_cpu_data
7864 +#define current_cpu_data boot_cpu_data
7865 #endif
7866
7867 -extern int cpu_llc_id[NR_CPUS];
7868 +/*
7869 + * the following now lives in the per cpu area:
7870 + * extern int cpu_llc_id[NR_CPUS];
7871 + */
7872 +DECLARE_PER_CPU(u8, cpu_llc_id);
7873 extern char ignore_fpu_irq;
7874
7875 void __init cpu_detect(struct cpuinfo_x86 *c);
7876 @@ -560,7 +566,9 @@ static inline void xen_set_iopl_mask(uns
7877 * clear %ecx since some cpus (Cyrix MII) do not set or clear %ecx
7878 * resulting in stale register contents being returned.
7879 */
7880 -static inline void cpuid(unsigned int op, unsigned int *eax, unsigned int *ebx, unsigned int *ecx, unsigned int *edx)
7881 +static inline void cpuid(unsigned int op,
7882 + unsigned int *eax, unsigned int *ebx,
7883 + unsigned int *ecx, unsigned int *edx)
7884 {
7885 *eax = op;
7886 *ecx = 0;
7887 @@ -568,8 +576,9 @@ static inline void cpuid(unsigned int op
7888 }
7889
7890 /* Some CPUID calls want 'count' to be placed in ecx */
7891 -static inline void cpuid_count(int op, int count, int *eax, int *ebx, int *ecx,
7892 - int *edx)
7893 +static inline void cpuid_count(unsigned int op, int count,
7894 + unsigned int *eax, unsigned int *ebx,
7895 + unsigned int *ecx, unsigned int *edx)
7896 {
7897 *eax = op;
7898 *ecx = count;
7899 @@ -639,6 +648,17 @@ static inline unsigned int cpuid_edx(uns
7900 #define K7_NOP7 ".byte 0x8D,0x04,0x05,0,0,0,0\n"
7901 #define K7_NOP8 K7_NOP7 ASM_NOP1
7902
7903 +/* P6 nops */
7904 +/* uses eax dependencies (Intel-recommended choice) */
7905 +#define P6_NOP1 GENERIC_NOP1
7906 +#define P6_NOP2 ".byte 0x66,0x90\n"
7907 +#define P6_NOP3 ".byte 0x0f,0x1f,0x00\n"
7908 +#define P6_NOP4 ".byte 0x0f,0x1f,0x40,0\n"
7909 +#define P6_NOP5 ".byte 0x0f,0x1f,0x44,0x00,0\n"
7910 +#define P6_NOP6 ".byte 0x66,0x0f,0x1f,0x44,0x00,0\n"
7911 +#define P6_NOP7 ".byte 0x0f,0x1f,0x80,0,0,0,0\n"
7912 +#define P6_NOP8 ".byte 0x0f,0x1f,0x84,0x00,0,0,0,0\n"
7913 +
7914 #ifdef CONFIG_MK8
7915 #define ASM_NOP1 K8_NOP1
7916 #define ASM_NOP2 K8_NOP2
7917 @@ -657,6 +677,17 @@ static inline unsigned int cpuid_edx(uns
7918 #define ASM_NOP6 K7_NOP6
7919 #define ASM_NOP7 K7_NOP7
7920 #define ASM_NOP8 K7_NOP8
7921 +#elif defined(CONFIG_M686) || defined(CONFIG_MPENTIUMII) || \
7922 + defined(CONFIG_MPENTIUMIII) || defined(CONFIG_MPENTIUMM) || \
7923 + defined(CONFIG_MCORE2) || defined(CONFIG_PENTIUM4)
7924 +#define ASM_NOP1 P6_NOP1
7925 +#define ASM_NOP2 P6_NOP2
7926 +#define ASM_NOP3 P6_NOP3
7927 +#define ASM_NOP4 P6_NOP4
7928 +#define ASM_NOP5 P6_NOP5
7929 +#define ASM_NOP6 P6_NOP6
7930 +#define ASM_NOP7 P6_NOP7
7931 +#define ASM_NOP8 P6_NOP8
7932 #else
7933 #define ASM_NOP1 GENERIC_NOP1
7934 #define ASM_NOP2 GENERIC_NOP2
7935 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/processor_64.h 2009-02-16 16:17:21.000000000 +0100
7936 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/processor_64.h 2009-02-16 16:18:36.000000000 +0100
7937 @@ -74,6 +74,7 @@ struct cpuinfo_x86 {
7938 __u8 booted_cores; /* number of cores as seen by OS */
7939 __u8 phys_proc_id; /* Physical Processor id. */
7940 __u8 cpu_core_id; /* Core id. */
7941 + __u8 cpu_index; /* index into per_cpu list */
7942 #endif
7943 } ____cacheline_aligned;
7944
7945 @@ -88,11 +89,12 @@ struct cpuinfo_x86 {
7946 #define X86_VENDOR_UNKNOWN 0xff
7947
7948 #ifdef CONFIG_SMP
7949 -extern struct cpuinfo_x86 cpu_data[];
7950 -#define current_cpu_data cpu_data[smp_processor_id()]
7951 +DECLARE_PER_CPU(struct cpuinfo_x86, cpu_info);
7952 +#define cpu_data(cpu) per_cpu(cpu_info, cpu)
7953 +#define current_cpu_data cpu_data(smp_processor_id())
7954 #else
7955 -#define cpu_data (&boot_cpu_data)
7956 -#define current_cpu_data boot_cpu_data
7957 +#define cpu_data(cpu) boot_cpu_data
7958 +#define current_cpu_data boot_cpu_data
7959 #endif
7960
7961 extern char ignore_irq13;
7962 @@ -343,6 +345,16 @@ struct extended_sigtable {
7963 };
7964
7965
7966 +#if defined(CONFIG_MPSC) || defined(CONFIG_MCORE2)
7967 +#define ASM_NOP1 P6_NOP1
7968 +#define ASM_NOP2 P6_NOP2
7969 +#define ASM_NOP3 P6_NOP3
7970 +#define ASM_NOP4 P6_NOP4
7971 +#define ASM_NOP5 P6_NOP5
7972 +#define ASM_NOP6 P6_NOP6
7973 +#define ASM_NOP7 P6_NOP7
7974 +#define ASM_NOP8 P6_NOP8
7975 +#else
7976 #define ASM_NOP1 K8_NOP1
7977 #define ASM_NOP2 K8_NOP2
7978 #define ASM_NOP3 K8_NOP3
7979 @@ -351,6 +363,7 @@ struct extended_sigtable {
7980 #define ASM_NOP6 K8_NOP6
7981 #define ASM_NOP7 K8_NOP7
7982 #define ASM_NOP8 K8_NOP8
7983 +#endif
7984
7985 /* Opteron nops */
7986 #define K8_NOP1 ".byte 0x90\n"
7987 @@ -362,6 +375,17 @@ struct extended_sigtable {
7988 #define K8_NOP7 K8_NOP4 K8_NOP3
7989 #define K8_NOP8 K8_NOP4 K8_NOP4
7990
7991 +/* P6 nops */
7992 +/* uses eax dependencies (Intel-recommended choice) */
7993 +#define P6_NOP1 ".byte 0x90\n"
7994 +#define P6_NOP2 ".byte 0x66,0x90\n"
7995 +#define P6_NOP3 ".byte 0x0f,0x1f,0x00\n"
7996 +#define P6_NOP4 ".byte 0x0f,0x1f,0x40,0\n"
7997 +#define P6_NOP5 ".byte 0x0f,0x1f,0x44,0x00,0\n"
7998 +#define P6_NOP6 ".byte 0x66,0x0f,0x1f,0x44,0x00,0\n"
7999 +#define P6_NOP7 ".byte 0x0f,0x1f,0x80,0,0,0,0\n"
8000 +#define P6_NOP8 ".byte 0x0f,0x1f,0x84,0x00,0,0,0,0\n"
8001 +
8002 #define ASM_NOP_MAX 8
8003
8004 /* REP NOP (PAUSE) is a good thing to insert into busy-wait loops. */
8005 @@ -377,12 +401,6 @@ static inline void sync_core(void)
8006 asm volatile("cpuid" : "=a" (tmp) : "0" (1) : "ebx","ecx","edx","memory");
8007 }
8008
8009 -#define ARCH_HAS_PREFETCH
8010 -static inline void prefetch(void *x)
8011 -{
8012 - asm volatile("prefetcht0 (%0)" :: "r" (x));
8013 -}
8014 -
8015 #define ARCH_HAS_PREFETCHW 1
8016 static inline void prefetchw(void *x)
8017 {
8018 @@ -398,11 +416,6 @@ static inline void prefetchw(void *x)
8019
8020 #define cpu_relax() rep_nop()
8021
8022 -static inline void serialize_cpu(void)
8023 -{
8024 - __asm__ __volatile__ ("cpuid" : : : "ax", "bx", "cx", "dx");
8025 -}
8026 -
8027 static inline void __monitor(const void *eax, unsigned long ecx,
8028 unsigned long edx)
8029 {
8030 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8031 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/segment.h 2009-02-16 16:18:36.000000000 +0100
8032 @@ -0,0 +1,5 @@
8033 +#ifdef CONFIG_X86_32
8034 +# include "segment_32.h"
8035 +#else
8036 +# include "../../segment_64.h"
8037 +#endif
8038 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8039 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/smp.h 2009-02-16 16:18:36.000000000 +0100
8040 @@ -0,0 +1,5 @@
8041 +#ifdef CONFIG_X86_32
8042 +# include "smp_32.h"
8043 +#else
8044 +# include "smp_64.h"
8045 +#endif
8046 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/smp_32.h 2008-12-15 11:27:22.000000000 +0100
8047 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/smp_32.h 2009-02-16 16:18:36.000000000 +0100
8048 @@ -11,7 +11,7 @@
8049 #endif
8050
8051 #if defined(CONFIG_X86_LOCAL_APIC) && !defined(__ASSEMBLY__)
8052 -#include <asm/bitops.h>
8053 +#include <linux/bitops.h>
8054 #include <asm/mpspec.h>
8055 #include <asm/apic.h>
8056 #ifdef CONFIG_X86_IO_APIC
8057 @@ -30,8 +30,8 @@
8058 extern void smp_alloc_memory(void);
8059 extern int pic_mode;
8060 extern int smp_num_siblings;
8061 -extern cpumask_t cpu_sibling_map[];
8062 -extern cpumask_t cpu_core_map[];
8063 +DECLARE_PER_CPU(cpumask_t, cpu_sibling_map);
8064 +DECLARE_PER_CPU(cpumask_t, cpu_core_map);
8065
8066 extern void (*mtrr_hook) (void);
8067 extern void zap_low_mappings (void);
8068 @@ -39,9 +39,11 @@ extern void lock_ipi_call_lock(void);
8069 extern void unlock_ipi_call_lock(void);
8070
8071 #define MAX_APICID 256
8072 -extern u8 x86_cpu_to_apicid[];
8073 +extern u8 __initdata x86_cpu_to_apicid_init[];
8074 +extern void *x86_cpu_to_apicid_ptr;
8075 +DECLARE_PER_CPU(u8, x86_cpu_to_apicid);
8076
8077 -#define cpu_physical_id(cpu) x86_cpu_to_apicid[cpu]
8078 +#define cpu_physical_id(cpu) per_cpu(x86_cpu_to_apicid, cpu)
8079
8080 #ifdef CONFIG_HOTPLUG_CPU
8081 extern void cpu_exit_clear(void);
8082 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/smp_64.h 2008-12-15 11:27:22.000000000 +0100
8083 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/smp_64.h 2009-02-16 16:18:36.000000000 +0100
8084 @@ -40,10 +40,19 @@ extern void lock_ipi_call_lock(void);
8085 extern void unlock_ipi_call_lock(void);
8086 extern int smp_num_siblings;
8087 extern void smp_send_reschedule(int cpu);
8088 +extern int smp_call_function_mask(cpumask_t mask, void (*func)(void *),
8089 + void *info, int wait);
8090
8091 -extern cpumask_t cpu_sibling_map[NR_CPUS];
8092 -extern cpumask_t cpu_core_map[NR_CPUS];
8093 -extern u8 cpu_llc_id[NR_CPUS];
8094 +/*
8095 + * cpu_sibling_map and cpu_core_map now live
8096 + * in the per cpu area
8097 + *
8098 + * extern cpumask_t cpu_sibling_map[NR_CPUS];
8099 + * extern cpumask_t cpu_core_map[NR_CPUS];
8100 + */
8101 +DECLARE_PER_CPU(cpumask_t, cpu_sibling_map);
8102 +DECLARE_PER_CPU(cpumask_t, cpu_core_map);
8103 +DECLARE_PER_CPU(u8, cpu_llc_id);
8104
8105 #define SMP_TRAMPOLINE_BASE 0x6000
8106
8107 @@ -70,6 +79,8 @@ extern unsigned __cpuinitdata disabled_c
8108
8109 #endif /* CONFIG_SMP */
8110
8111 +#define safe_smp_processor_id() smp_processor_id()
8112 +
8113 #ifdef CONFIG_X86_LOCAL_APIC
8114 static inline int hard_smp_processor_id(void)
8115 {
8116 @@ -82,8 +93,9 @@ static inline int hard_smp_processor_id(
8117 * Some lowlevel functions might want to know about
8118 * the real APIC ID <-> CPU # mapping.
8119 */
8120 -extern u8 x86_cpu_to_apicid[NR_CPUS]; /* physical ID */
8121 -extern u8 x86_cpu_to_log_apicid[NR_CPUS];
8122 +extern u8 __initdata x86_cpu_to_apicid_init[];
8123 +extern void *x86_cpu_to_apicid_ptr;
8124 +DECLARE_PER_CPU(u8, x86_cpu_to_apicid); /* physical ID */
8125 extern u8 bios_cpu_apicid[];
8126
8127 #ifdef CONFIG_X86_LOCAL_APIC
8128 @@ -118,8 +130,9 @@ static __inline int logical_smp_processo
8129 #endif
8130
8131 #ifdef CONFIG_SMP
8132 -#define cpu_physical_id(cpu) x86_cpu_to_apicid[cpu]
8133 +#define cpu_physical_id(cpu) per_cpu(x86_cpu_to_apicid, cpu)
8134 #else
8135 +extern unsigned int boot_cpu_id;
8136 #define cpu_physical_id(cpu) boot_cpu_id
8137 #endif /* !CONFIG_SMP */
8138 #endif
8139 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8140 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/swiotlb.h 2009-02-16 16:18:36.000000000 +0100
8141 @@ -0,0 +1,5 @@
8142 +#ifdef CONFIG_X86_32
8143 +# include "swiotlb_32.h"
8144 +#else
8145 +# include "../../swiotlb.h"
8146 +#endif
8147 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8148 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/system.h 2009-02-16 16:18:36.000000000 +0100
8149 @@ -0,0 +1,5 @@
8150 +#ifdef CONFIG_X86_32
8151 +# include "system_32.h"
8152 +#else
8153 +# include "system_64.h"
8154 +#endif
8155 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/system_32.h 2009-02-16 16:17:21.000000000 +0100
8156 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/system_32.h 2009-02-16 16:18:36.000000000 +0100
8157 @@ -9,6 +9,7 @@
8158 #include <asm/hypervisor.h>
8159
8160 #ifdef __KERNEL__
8161 +#define AT_VECTOR_SIZE_ARCH 2 /* entries in ARCH_DLINFO */
8162
8163 struct task_struct; /* one of the stranger aspects of C forward declarations.. */
8164 extern struct task_struct * FASTCALL(__switch_to(struct task_struct *prev, struct task_struct *next));
8165 @@ -138,7 +139,7 @@ static inline unsigned long xen_read_cr4
8166 {
8167 unsigned long val;
8168 /* This could fault if %cr4 does not exist */
8169 - asm("1: movl %%cr4, %0 \n"
8170 + asm volatile("1: movl %%cr4, %0 \n"
8171 "2: \n"
8172 ".section __ex_table,\"a\" \n"
8173 ".long 1b,2b \n"
8174 @@ -157,6 +158,11 @@ static inline void xen_wbinvd(void)
8175 asm volatile("wbinvd": : :"memory");
8176 }
8177
8178 +static inline void clflush(volatile void *__p)
8179 +{
8180 + asm volatile("clflush %0" : "+m" (*(char __force *)__p));
8181 +}
8182 +
8183 #define read_cr0() (xen_read_cr0())
8184 #define write_cr0(x) (xen_write_cr0(x))
8185 #define read_cr2() (xen_read_cr2())
8186 @@ -207,6 +213,7 @@ static inline unsigned long get_limit(un
8187
8188 #define mb() alternative("lock; addl $0,0(%%esp)", "mfence", X86_FEATURE_XMM2)
8189 #define rmb() alternative("lock; addl $0,0(%%esp)", "lfence", X86_FEATURE_XMM2)
8190 +#define wmb() alternative("lock; addl $0,0(%%esp)", "sfence", X86_FEATURE_XMM)
8191
8192 /**
8193 * read_barrier_depends - Flush all pending reads that subsequents reads
8194 @@ -262,18 +269,18 @@ static inline unsigned long get_limit(un
8195
8196 #define read_barrier_depends() do { } while(0)
8197
8198 +#ifdef CONFIG_SMP
8199 +#define smp_mb() mb()
8200 +#ifdef CONFIG_X86_PPRO_FENCE
8201 +# define smp_rmb() rmb()
8202 +#else
8203 +# define smp_rmb() barrier()
8204 +#endif
8205 #ifdef CONFIG_X86_OOSTORE
8206 -/* Actually there are no OOO store capable CPUs for now that do SSE,
8207 - but make it already an possibility. */
8208 -#define wmb() alternative("lock; addl $0,0(%%esp)", "sfence", X86_FEATURE_XMM)
8209 +# define smp_wmb() wmb()
8210 #else
8211 -#define wmb() __asm__ __volatile__ ("": : :"memory")
8212 +# define smp_wmb() barrier()
8213 #endif
8214 -
8215 -#ifdef CONFIG_SMP
8216 -#define smp_mb() mb()
8217 -#define smp_rmb() rmb()
8218 -#define smp_wmb() wmb()
8219 #define smp_read_barrier_depends() read_barrier_depends()
8220 #define set_mb(var, value) do { (void) xchg(&var, value); } while (0)
8221 #else
8222 @@ -300,5 +307,6 @@ extern unsigned long arch_align_stack(un
8223 extern void free_init_pages(char *what, unsigned long begin, unsigned long end);
8224
8225 void default_idle(void);
8226 +void __show_registers(struct pt_regs *, int all);
8227
8228 #endif
8229 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/system_64.h 2009-02-16 16:17:21.000000000 +0100
8230 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/system_64.h 2009-02-16 16:18:36.000000000 +0100
8231 @@ -11,8 +11,12 @@
8232
8233 #ifdef __KERNEL__
8234
8235 -#define __STR(x) #x
8236 -#define STR(x) __STR(x)
8237 +/* entries in ARCH_DLINFO: */
8238 +#ifdef CONFIG_IA32_EMULATION
8239 +# define AT_VECTOR_SIZE_ARCH 2
8240 +#else
8241 +# define AT_VECTOR_SIZE_ARCH 1
8242 +#endif
8243
8244 #define __SAVE(reg,offset) "movq %%" #reg ",(14-" #offset ")*8(%%rsp)\n\t"
8245 #define __RESTORE(reg,offset) "movq (14-" #offset ")*8(%%rsp),%%" #reg "\n\t"
8246 @@ -92,7 +96,7 @@ static inline void write_cr0(unsigned lo
8247
8248 #define read_cr3() ({ \
8249 unsigned long __dummy; \
8250 - asm("movq %%cr3,%0" : "=r" (__dummy)); \
8251 + asm volatile("movq %%cr3,%0" : "=r" (__dummy)); \
8252 machine_to_phys(__dummy); \
8253 })
8254
8255 @@ -105,7 +109,7 @@ static inline void write_cr3(unsigned lo
8256 static inline unsigned long read_cr4(void)
8257 {
8258 unsigned long cr4;
8259 - asm("movq %%cr4,%0" : "=r" (cr4));
8260 + asm volatile("movq %%cr4,%0" : "=r" (cr4));
8261 return cr4;
8262 }
8263
8264 @@ -131,12 +135,17 @@ static inline void write_cr8(unsigned lo
8265
8266 #endif /* __KERNEL__ */
8267
8268 +static inline void clflush(volatile void *__p)
8269 +{
8270 + asm volatile("clflush %0" : "+m" (*(char __force *)__p));
8271 +}
8272 +
8273 #define nop() __asm__ __volatile__ ("nop")
8274
8275 #ifdef CONFIG_SMP
8276 #define smp_mb() mb()
8277 -#define smp_rmb() rmb()
8278 -#define smp_wmb() wmb()
8279 +#define smp_rmb() barrier()
8280 +#define smp_wmb() barrier()
8281 #define smp_read_barrier_depends() do {} while(0)
8282 #else
8283 #define smp_mb() barrier()
8284 @@ -153,12 +162,8 @@ static inline void write_cr8(unsigned lo
8285 */
8286 #define mb() asm volatile("mfence":::"memory")
8287 #define rmb() asm volatile("lfence":::"memory")
8288 -
8289 -#ifdef CONFIG_UNORDERED_IO
8290 #define wmb() asm volatile("sfence" ::: "memory")
8291 -#else
8292 -#define wmb() asm volatile("" ::: "memory")
8293 -#endif
8294 +
8295 #define read_barrier_depends() do {} while(0)
8296 #define set_mb(var, value) do { (void) xchg(&var, value); } while (0)
8297
8298 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8299 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/tlbflush.h 2009-02-16 16:18:36.000000000 +0100
8300 @@ -0,0 +1,5 @@
8301 +#ifdef CONFIG_X86_32
8302 +# include "tlbflush_32.h"
8303 +#else
8304 +# include "tlbflush_64.h"
8305 +#endif
8306 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/tlbflush_32.h 2009-02-16 16:17:21.000000000 +0100
8307 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/tlbflush_32.h 2009-02-16 16:18:36.000000000 +0100
8308 @@ -23,7 +23,6 @@
8309 * - flush_tlb_page(vma, vmaddr) flushes one page
8310 * - flush_tlb_range(vma, start, end) flushes a range of pages
8311 * - flush_tlb_kernel_range(start, end) flushes a range of kernel pages
8312 - * - flush_tlb_pgtables(mm, start, end) flushes a range of page tables
8313 *
8314 * ..but the i386 has somewhat limited tlb flushing capabilities,
8315 * and page-granular flushes are available only on i486 and up.
8316 @@ -97,10 +96,4 @@ static inline void flush_tlb_kernel_rang
8317 flush_tlb_all();
8318 }
8319
8320 -static inline void flush_tlb_pgtables(struct mm_struct *mm,
8321 - unsigned long start, unsigned long end)
8322 -{
8323 - /* i386 does not keep any page table caches in TLB */
8324 -}
8325 -
8326 #endif /* _I386_TLBFLUSH_H */
8327 --- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/tlbflush_64.h 2009-02-16 16:17:21.000000000 +0100
8328 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/tlbflush_64.h 2009-02-16 16:18:36.000000000 +0100
8329 @@ -28,7 +28,6 @@
8330 * - flush_tlb_page(vma, vmaddr) flushes one page
8331 * - flush_tlb_range(vma, start, end) flushes a range of pages
8332 * - flush_tlb_kernel_range(start, end) flushes a range of kernel pages
8333 - * - flush_tlb_pgtables(mm, start, end) flushes a range of page tables
8334 *
8335 * x86-64 can only flush individual pages or full VMs. For a range flush
8336 * we always do the full VM. Might be worth trying if for a small
8337 @@ -95,12 +94,4 @@ static inline void flush_tlb_kernel_rang
8338 flush_tlb_all();
8339 }
8340
8341 -static inline void flush_tlb_pgtables(struct mm_struct *mm,
8342 - unsigned long start, unsigned long end)
8343 -{
8344 - /* x86_64 does not keep any page table caches in a software TLB.
8345 - The CPUs do in their hardware TLBs, but they are handled
8346 - by the normal TLB flushing algorithms. */
8347 -}
8348 -
8349 #endif /* _X8664_TLBFLUSH_H */
8350 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8351 +++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/xor.h 2009-02-16 16:18:36.000000000 +0100
8352 @@ -0,0 +1,5 @@
8353 +#ifdef CONFIG_X86_32
8354 +# include "../../xor_32.h"
8355 +#else
8356 +# include "xor_64.h"
8357 +#endif
8358 --- sle11-2009-05-14.orig/include/asm-x86/mmu.h 2009-05-14 10:56:29.000000000 +0200
8359 +++ sle11-2009-05-14/include/asm-x86/mmu.h 2009-02-16 16:18:36.000000000 +0100
8360 @@ -16,6 +16,9 @@ typedef struct {
8361 rwlock_t ldtlock;
8362 #endif
8363 int size;
8364 +#ifdef CONFIG_XEN
8365 + unsigned has_foreign_mappings:1;
8366 +#endif
8367 struct mutex lock;
8368 void *vdso;
8369 } mm_context_t;
8370 --- sle11-2009-05-14.orig/include/linux/kexec.h 2009-02-16 15:58:14.000000000 +0100
8371 +++ sle11-2009-05-14/include/linux/kexec.h 2009-02-17 12:43:57.000000000 +0100
8372 @@ -202,8 +202,15 @@ extern int dump_after_notifier;
8373 #define VMCOREINFO_BYTES (4096)
8374 #define VMCOREINFO_NOTE_NAME "VMCOREINFO"
8375 #define VMCOREINFO_NOTE_NAME_BYTES ALIGN(sizeof(VMCOREINFO_NOTE_NAME), 4)
8376 +#if !defined(CONFIG_XEN) || !defined(CONFIG_X86)
8377 #define VMCOREINFO_NOTE_SIZE (KEXEC_NOTE_HEAD_BYTES*2 + VMCOREINFO_BYTES \
8378 + VMCOREINFO_NOTE_NAME_BYTES)
8379 +#else
8380 +#define VMCOREINFO_NOTE_SIZE ALIGN(KEXEC_NOTE_HEAD_BYTES*2 \
8381 + + VMCOREINFO_BYTES \
8382 + + VMCOREINFO_NOTE_NAME_BYTES, \
8383 + PAGE_SIZE)
8384 +#endif
8385
8386 /* Location of a reserved region to hold the crash kernel.
8387 */
8388 --- sle11-2009-05-14.orig/include/linux/oprofile.h 2009-02-16 16:01:39.000000000 +0100
8389 +++ sle11-2009-05-14/include/linux/oprofile.h 2009-03-12 16:16:07.000000000 +0100
8390 @@ -116,6 +116,8 @@ void oprofile_add_pc(unsigned long pc, i
8391 /* add a backtrace entry, to be called from the ->backtrace callback */
8392 void oprofile_add_trace(unsigned long eip);
8393
8394 +void oprofile_add_mode(int cpu_mode);
8395 +
8396 /* add a domain switch entry */
8397 int oprofile_add_domain_switch(int32_t domain_id);
8398
8399 --- sle11-2009-05-14.orig/include/linux/sysctl.h 2009-05-14 10:56:29.000000000 +0200
8400 +++ sle11-2009-05-14/include/linux/sysctl.h 2009-02-16 16:18:36.000000000 +0100
8401 @@ -69,6 +69,7 @@ enum
8402 CTL_BUS=8, /* Busses */
8403 CTL_ABI=9, /* Binary emulation */
8404 CTL_CPU=10, /* CPU stuff (speed scaling, etc) */
8405 + CTL_XEN=123, /* Xen info and control */
8406 CTL_ARLAN=254, /* arlan wireless driver */
8407 CTL_S390DBF=5677, /* s390 debug */
8408 CTL_SUNRPC=7249, /* sunrpc debug */
8409 --- sle11-2009-05-14.orig/include/xen/pcifront.h 2009-05-14 10:56:29.000000000 +0200
8410 +++ sle11-2009-05-14/include/xen/pcifront.h 2009-02-16 16:18:36.000000000 +0100
8411 @@ -12,13 +12,11 @@
8412
8413 #ifndef __ia64__
8414
8415 +#include <asm/pci.h>
8416 +
8417 struct pcifront_device;
8418 struct pci_bus;
8419 -
8420 -struct pcifront_sd {
8421 - int domain;
8422 - struct pcifront_device *pdev;
8423 -};
8424 +#define pcifront_sd pci_sysdata
8425
8426 static inline struct pcifront_device *
8427 pcifront_get_pdev(struct pcifront_sd *sd)
8428 @@ -34,18 +32,6 @@ static inline void pcifront_init_sd(stru
8429 sd->pdev = pdev;
8430 }
8431
8432 -#if defined(CONFIG_PCI_DOMAINS)
8433 -static inline int pci_domain_nr(struct pci_bus *bus)
8434 -{
8435 - struct pcifront_sd *sd = bus->sysdata;
8436 - return sd->domain;
8437 -}
8438 -static inline int pci_proc_domain(struct pci_bus *bus)
8439 -{
8440 - return pci_domain_nr(bus);
8441 -}
8442 -#endif /* CONFIG_PCI_DOMAINS */
8443 -
8444 static inline void pcifront_setup_root_resources(struct pci_bus *bus,
8445 struct pcifront_sd *sd)
8446 {
8447 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8448 +++ sle11-2009-05-14/include/xen/sysctl.h 2009-02-16 16:18:36.000000000 +0100
8449 @@ -0,0 +1,11 @@
8450 +#ifndef _XEN_SYSCTL_H
8451 +#define _XEN_SYSCTL_H
8452 +
8453 +/* CTL_XEN names: */
8454 +enum
8455 +{
8456 + CTL_XEN_INDEPENDENT_WALLCLOCK=1,
8457 + CTL_XEN_PERMITTED_CLOCK_JITTER=2,
8458 +};
8459 +
8460 +#endif /* _XEN_SYSCTL_H */
8461 --- sle11-2009-05-14.orig/include/xen/xenbus.h 2009-02-16 16:17:21.000000000 +0100
8462 +++ sle11-2009-05-14/include/xen/xenbus.h 2009-02-16 16:18:36.000000000 +0100
8463 @@ -107,7 +107,7 @@ struct xenbus_driver {
8464 int (*suspend)(struct xenbus_device *dev);
8465 int (*suspend_cancel)(struct xenbus_device *dev);
8466 int (*resume)(struct xenbus_device *dev);
8467 - int (*uevent)(struct xenbus_device *, char **, int, char *, int);
8468 + int (*uevent)(struct xenbus_device *, struct kobj_uevent_env *);
8469 struct device_driver driver;
8470 int (*read_otherend_details)(struct xenbus_device *dev);
8471 int (*is_ready)(struct xenbus_device *dev);
8472 --- sle11-2009-05-14.orig/kernel/kexec.c 2009-02-17 11:34:22.000000000 +0100
8473 +++ sle11-2009-05-14/kernel/kexec.c 2009-02-17 12:38:20.000000000 +0100
8474 @@ -52,7 +52,11 @@ int dump_after_notifier;
8475
8476 /* vmcoreinfo stuff */
8477 unsigned char vmcoreinfo_data[VMCOREINFO_BYTES];
8478 -u32 vmcoreinfo_note[VMCOREINFO_NOTE_SIZE/4];
8479 +u32
8480 +#if defined(CONFIG_XEN) && defined(CONFIG_X86)
8481 +__attribute__((__section__(".bss.page_aligned"), __aligned__(PAGE_SIZE)))
8482 +#endif
8483 +vmcoreinfo_note[VMCOREINFO_NOTE_SIZE/4];
8484 size_t vmcoreinfo_size;
8485 size_t vmcoreinfo_max_size = sizeof(vmcoreinfo_data);
8486
8487 @@ -1240,6 +1244,7 @@ static int __init crash_notes_memory_ini
8488 module_init(crash_notes_memory_init)
8489
8490
8491 +#ifndef CONFIG_XEN
8492 /*
8493 * parsing the "crashkernel" commandline
8494 *
8495 @@ -1402,7 +1407,7 @@ int __init parse_crashkernel(char *cm
8496
8497 return 0;
8498 }
8499 -
8500 +#endif
8501
8502
8503 void crash_save_vmcoreinfo(void)
8504 @@ -1459,7 +1464,18 @@ static int __init crash_save_vmcoreinfo_
8505
8506 VMCOREINFO_SYMBOL(init_uts_ns);
8507 VMCOREINFO_SYMBOL(node_online_map);
8508 +#ifndef CONFIG_X86_XEN
8509 VMCOREINFO_SYMBOL(swapper_pg_dir);
8510 +#else
8511 +/*
8512 + * Since for x86-32 Xen swapper_pg_dir is a pointer rather than an array,
8513 + * make the value stored consistent with native (i.e. the base address of
8514 + * the page directory).
8515 + */
8516 +# define swapper_pg_dir *swapper_pg_dir
8517 + VMCOREINFO_SYMBOL(swapper_pg_dir);
8518 +# undef swapper_pg_dir
8519 +#endif
8520 VMCOREINFO_SYMBOL(_stext);
8521
8522 #ifndef CONFIG_NEED_MULTIPLE_NODES
8523 --- sle11-2009-05-14.orig/kernel/sysctl_check.c 2009-05-14 10:56:29.000000000 +0200
8524 +++ sle11-2009-05-14/kernel/sysctl_check.c 2009-02-16 16:18:36.000000000 +0100
8525 @@ -4,6 +4,7 @@
8526 #include <linux/sunrpc/debug.h>
8527 #include <linux/string.h>
8528 #include <net/ip_vs.h>
8529 +#include <xen/sysctl.h>
8530
8531 struct trans_ctl_table {
8532 int ctl_name;
8533 @@ -897,6 +898,14 @@ static const struct trans_ctl_table tran
8534 {}
8535 };
8536
8537 +#ifdef CONFIG_XEN
8538 +static struct trans_ctl_table trans_xen_table[] = {
8539 + { CTL_XEN_INDEPENDENT_WALLCLOCK, "independent_wallclock" },
8540 + { CTL_XEN_PERMITTED_CLOCK_JITTER, "permitted_clock_jitter" },
8541 + {}
8542 +};
8543 +#endif
8544 +
8545 static const struct trans_ctl_table trans_arlan_conf_table0[] = {
8546 { 1, "spreadingCode" },
8547 { 2, "channelNumber" },
8548 @@ -1232,6 +1241,9 @@ static const struct trans_ctl_table tran
8549 { CTL_BUS, "bus", trans_bus_table },
8550 { CTL_ABI, "abi" },
8551 /* CTL_CPU not used */
8552 +#ifdef CONFIG_XEN
8553 + { CTL_XEN, "xen", trans_xen_table },
8554 +#endif
8555 { CTL_ARLAN, "arlan", trans_arlan_table },
8556 { CTL_S390DBF, "s390dbf", trans_s390dbf_table },
8557 { CTL_SUNRPC, "sunrpc", trans_sunrpc_table },
8558 --- sle11-2009-05-14.orig/lib/swiotlb-xen.c 2009-02-05 11:16:51.000000000 +0100
8559 +++ sle11-2009-05-14/lib/swiotlb-xen.c 2009-02-16 16:18:36.000000000 +0100
8560 @@ -27,7 +27,7 @@
8561 #include <asm/uaccess.h>
8562 #include <xen/gnttab.h>
8563 #include <xen/interface/memory.h>
8564 -#include <asm-i386/mach-xen/asm/gnttab_dma.h>
8565 +#include <asm/gnttab_dma.h>
8566
8567 int swiotlb;
8568 EXPORT_SYMBOL(swiotlb);
8569 @@ -574,9 +574,10 @@ swiotlb_sync_single_for_device(struct de
8570 * same here.
8571 */
8572 int
8573 -swiotlb_map_sg(struct device *hwdev, struct scatterlist *sg, int nelems,
8574 +swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
8575 int dir)
8576 {
8577 + struct scatterlist *sg;
8578 struct phys_addr buffer;
8579 dma_addr_t dev_addr;
8580 char *map;
8581 @@ -584,22 +585,22 @@ swiotlb_map_sg(struct device *hwdev, str
8582
8583 BUG_ON(dir == DMA_NONE);
8584
8585 - for (i = 0; i < nelems; i++, sg++) {
8586 - dev_addr = gnttab_dma_map_page(sg->page) + sg->offset;
8587 + for_each_sg(sgl, sg, nelems, i) {
8588 + dev_addr = gnttab_dma_map_page(sg_page(sg)) + sg->offset;
8589
8590 - if (range_straddles_page_boundary(page_to_pseudophys(sg->page)
8591 + if (range_straddles_page_boundary(page_to_pseudophys(sg_page(sg))
8592 + sg->offset, sg->length)
8593 || address_needs_mapping(hwdev, dev_addr)) {
8594 gnttab_dma_unmap_page(dev_addr);
8595 - buffer.page = sg->page;
8596 + buffer.page = sg_page(sg);
8597 buffer.offset = sg->offset;
8598 map = map_single(hwdev, buffer, sg->length, dir);
8599 if (!map) {
8600 /* Don't panic here, we expect map_sg users
8601 to do proper error handling. */
8602 swiotlb_full(hwdev, sg->length, dir, 0);
8603 - swiotlb_unmap_sg(hwdev, sg - i, i, dir);
8604 - sg[0].dma_length = 0;
8605 + swiotlb_unmap_sg(hwdev, sgl, i, dir);
8606 + sgl[0].dma_length = 0;
8607 return 0;
8608 }
8609 sg->dma_address = virt_to_bus(map);
8610 @@ -615,19 +616,21 @@ swiotlb_map_sg(struct device *hwdev, str
8611 * concerning calls here are the same as for swiotlb_unmap_single() above.
8612 */
8613 void
8614 -swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sg, int nelems,
8615 +swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
8616 int dir)
8617 {
8618 + struct scatterlist *sg;
8619 int i;
8620
8621 BUG_ON(dir == DMA_NONE);
8622
8623 - for (i = 0; i < nelems; i++, sg++)
8624 + for_each_sg(sgl, sg, nelems, i) {
8625 if (in_swiotlb_aperture(sg->dma_address))
8626 unmap_single(hwdev, bus_to_virt(sg->dma_address),
8627 sg->dma_length, dir);
8628 else
8629 gnttab_dma_unmap_page(sg->dma_address);
8630 + }
8631 }
8632
8633 /*
8634 @@ -638,31 +641,35 @@ swiotlb_unmap_sg(struct device *hwdev, s
8635 * and usage.
8636 */
8637 void
8638 -swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sg,
8639 +swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sgl,
8640 int nelems, int dir)
8641 {
8642 + struct scatterlist *sg;
8643 int i;
8644
8645 BUG_ON(dir == DMA_NONE);
8646
8647 - for (i = 0; i < nelems; i++, sg++)
8648 + for_each_sg(sgl, sg, nelems, i) {
8649 if (in_swiotlb_aperture(sg->dma_address))
8650 sync_single(hwdev, bus_to_virt(sg->dma_address),
8651 sg->dma_length, dir);
8652 + }
8653 }
8654
8655 void
8656 -swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg,
8657 +swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sgl,
8658 int nelems, int dir)
8659 {
8660 + struct scatterlist *sg;
8661 int i;
8662
8663 BUG_ON(dir == DMA_NONE);
8664
8665 - for (i = 0; i < nelems; i++, sg++)
8666 + for_each_sg(sgl, sg, nelems, i) {
8667 if (in_swiotlb_aperture(sg->dma_address))
8668 sync_single(hwdev, bus_to_virt(sg->dma_address),
8669 sg->dma_length, dir);
8670 + }
8671 }
8672
8673 #ifdef CONFIG_HIGHMEM
Timo's tree of IPFire 2.x