]> git.ipfire.org Git - ipfire-2.x.git/blob - src/patches/suse-2.6.27.39/patches.xen/xen3-patch-2.6.24
projects / ipfire-2.x.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Add a patch to fix Intel E100 wake-on-lan problems.
[ipfire-2.x.git] / src / patches / suse-2.6.27.39 / 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 --- sle11-2009-10-16.orig/arch/x86/Kconfig 2009-02-05 10:22:38.000000000 +0100
10 +++ sle11-2009-10-16/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 --- sle11-2009-10-16.orig/arch/x86/Makefile 2009-02-16 16:17:21.000000000 +0100
160 +++ sle11-2009-10-16/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 --- sle11-2009-10-16.orig/arch/x86/ia32/ia32entry-xen.S 2009-02-16 16:17:21.000000000 +0100
182 +++ sle11-2009-10-16/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 $-ENOSYS,RAX(%rsp) /* really needed? */
216 movq %rsp,%rdi /* &pt_regs -> arg1 */
217 call syscall_trace_enter
218 LOAD_ARGS32 ARGOFFSET /* reload args from stack in case ptrace changed it */
219 RESTORE_REST
220 + xchgl %ebp,%r9d
221 movl RSP-ARGOFFSET(%rsp), %r8d
222 - /* no need to do an access_ok check here because r8 has been
223 - 32bit zero extended */
224 -1: movl (%r8),%r9d
225 - .section __ex_table,"a"
226 - .quad 1b,ia32_badarg
227 - .previous
228 jmp cstar_do_call
229 END(ia32_cstar_target)
230
231 --- sle11-2009-10-16.orig/arch/x86/kernel/Makefile 2008-12-15 11:27:22.000000000 +0100
232 +++ sle11-2009-10-16/arch/x86/kernel/Makefile 2009-02-16 16:18:36.000000000 +0100
233 @@ -127,4 +127,4 @@ endif
234 disabled-obj-$(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 --- sle11-2009-10-16.orig/arch/x86/kernel/acpi/sleep_32-xen.c 2009-02-16 16:17:21.000000000 +0100
240 +++ sle11-2009-10-16/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 --- sle11-2009-10-16.orig/arch/x86/kernel/acpi/sleep_64-xen.c 2009-02-16 16:17:21.000000000 +0100
251 +++ sle11-2009-10-16/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 --- sle11-2009-10-16.orig/arch/x86/kernel/apic_64-xen.c 2009-02-16 16:17:21.000000000 +0100
260 +++ sle11-2009-10-16/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 --- sle11-2009-10-16.orig/arch/x86/kernel/cpu/common-xen.c 2009-02-16 16:17:21.000000000 +0100
333 +++ sle11-2009-10-16/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 --- sle11-2009-10-16.orig/arch/x86/kernel/e820_32-xen.c 2009-02-16 16:17:21.000000000 +0100
344 +++ sle11-2009-10-16/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 --- sle11-2009-10-16.orig/arch/x86/kernel/e820_64-xen.c 2009-02-16 16:17:21.000000000 +0100
407 +++ sle11-2009-10-16/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 --- sle11-2009-10-16.orig/arch/x86/kernel/early_printk-xen.c 2009-02-16 16:17:21.000000000 +0100
489 +++ sle11-2009-10-16/arch/x86/kernel/early_printk-xen.c 2009-09-24 10:27:18.000000000 +0200
490 @@ -6,15 +6,10 @@
491 #include <asm/io.h>
492 #include <asm/processor.h>
493 #include <asm/fcntl.h>
494 -
495 -/* Simple VGA output */
496 -
497 -#ifdef __i386__
498 #include <asm/setup.h>
499 -#else
500 -#include <asm/bootsetup.h>
501 -#endif
502 +
503 #ifndef CONFIG_XEN
504 +/* Simple VGA output */
505 #define VGABASE (__ISA_IO_base + 0xb8000)
506
507 static int max_ypos = 25, max_xpos = 80;
508 @@ -264,10 +259,10 @@ static int __init setup_early_printk(cha
509 early_console = &early_serial_console;
510 } else if (!strncmp(buf, "vga", 3)) {
511 #ifndef CONFIG_XEN
512 - && SCREEN_INFO.orig_video_isVGA == 1) {
513 - max_xpos = SCREEN_INFO.orig_video_cols;
514 - max_ypos = SCREEN_INFO.orig_video_lines;
515 - current_ypos = SCREEN_INFO.orig_y;
516 + && boot_params.screen_info.orig_video_isVGA == 1) {
517 + max_xpos = boot_params.screen_info.orig_video_cols;
518 + max_ypos = boot_params.screen_info.orig_video_lines;
519 + current_ypos = boot_params.screen_info.orig_y;
520 #endif
521 early_console = &early_vga_console;
522 } else if (!strncmp(buf, "simnow", 6)) {
523 --- sle11-2009-10-16.orig/arch/x86/kernel/entry_32-xen.S 2009-02-16 16:17:21.000000000 +0100
524 +++ sle11-2009-10-16/arch/x86/kernel/entry_32-xen.S 2009-05-14 11:18:18.000000000 +0200
525 @@ -254,6 +254,7 @@ check_userspace:
526 jb resume_kernel # not returning to v8086 or userspace
527
528 ENTRY(resume_userspace)
529 + LOCKDEP_SYS_EXIT
530 DISABLE_INTERRUPTS(CLBR_ANY) # make sure we don't miss an interrupt
531 # setting need_resched or sigpending
532 # between sampling and the iret
533 @@ -341,6 +342,7 @@ sysenter_past_esp:
534 jae syscall_badsys
535 call *sys_call_table(,%eax,4)
536 movl %eax,PT_EAX(%esp)
537 + LOCKDEP_SYS_EXIT
538 DISABLE_INTERRUPTS(CLBR_ANY)
539 TRACE_IRQS_OFF
540 movl TI_flags(%ebp), %ecx
541 @@ -406,6 +408,7 @@ syscall_call:
542 call *sys_call_table(,%eax,4)
543 movl %eax,PT_EAX(%esp) # store the return value
544 syscall_exit:
545 + LOCKDEP_SYS_EXIT
546 DISABLE_INTERRUPTS(CLBR_ANY) # make sure we don't miss an interrupt
547 # setting need_resched or sigpending
548 # between sampling and the iret
549 @@ -478,7 +481,7 @@ ldt_ss:
550 * is still available to implement the setting of the high
551 * 16-bits in the INTERRUPT_RETURN paravirt-op.
552 */
553 - cmpl $0, paravirt_ops+PARAVIRT_enabled
554 + cmpl $0, pv_info+PARAVIRT_enabled
555 jne restore_nocheck
556 #endif
557
558 @@ -540,6 +543,7 @@ work_pending:
559 jz work_notifysig
560 work_resched:
561 call schedule
562 + LOCKDEP_SYS_EXIT
563 DISABLE_INTERRUPTS(CLBR_ANY) # make sure we don't miss an interrupt
564 # setting need_resched or sigpending
565 # between sampling and the iret
566 @@ -1268,6 +1272,6 @@ ENTRY(kernel_thread_helper)
567 ENDPROC(kernel_thread_helper)
568
569 .section .rodata,"a"
570 -#include "syscall_table.S"
571 +#include "syscall_table_32.S"
572
573 syscall_table_size=(.-sys_call_table)
574 --- sle11-2009-10-16.orig/arch/x86/kernel/entry_64-xen.S 2009-02-16 16:17:21.000000000 +0100
575 +++ sle11-2009-10-16/arch/x86/kernel/entry_64-xen.S 2009-02-16 16:18:36.000000000 +0100
576 @@ -57,7 +57,7 @@
577 #include <xen/interface/arch-x86_64.h>
578 #include <xen/interface/features.h>
579
580 -#include "xen_entry.S"
581 +#include "xen_entry_64.S"
582
583 .code64
584
585 @@ -275,6 +275,7 @@ ret_from_sys_call:
586 movl $_TIF_ALLWORK_MASK,%edi
587 /* edi: flagmask */
588 sysret_check:
589 + LOCKDEP_SYS_EXIT
590 GET_THREAD_INFO(%rcx)
591 XEN_BLOCK_EVENTS(%rsi)
592 TRACE_IRQS_OFF
593 @@ -365,6 +366,7 @@ int_ret_from_sys_call:
594 movl $_TIF_ALLWORK_MASK,%edi
595 /* edi: mask to check */
596 int_with_check:
597 + LOCKDEP_SYS_EXIT_IRQ
598 GET_THREAD_INFO(%rcx)
599 movl threadinfo_flags(%rcx),%edx
600 andl %edi,%edx
601 @@ -516,11 +518,12 @@ END(stub_rt_sigreturn)
602
603 retint_check:
604 CFI_DEFAULT_STACK adj=1
605 + LOCKDEP_SYS_EXIT_IRQ
606 movl threadinfo_flags(%rcx),%edx
607 andl %edi,%edx
608 CFI_REMEMBER_STATE
609 jnz retint_careful
610 -retint_restore_args:
611 +retint_restore_args: /* return to kernel space */
612 movl EFLAGS-REST_SKIP(%rsp), %eax
613 shr $9, %eax # EAX[0] == IRET_EFLAGS.IF
614 XEN_GET_VCPU_INFO(%rsi)
615 @@ -841,7 +844,7 @@ error_call_handler:
616 movq ORIG_RAX(%rsp),%rsi # get error code
617 movq $-1,ORIG_RAX(%rsp)
618 call *%rax
619 -error_exit:
620 +error_exit:
621 RESTORE_REST
622 /* cli */
623 XEN_BLOCK_EVENTS(%rsi)
624 @@ -849,14 +852,11 @@ error_exit:
625 GET_THREAD_INFO(%rcx)
626 testb $3,CS-ARGOFFSET(%rsp)
627 jz retint_kernel
628 + LOCKDEP_SYS_EXIT_IRQ
629 movl threadinfo_flags(%rcx),%edx
630 movl $_TIF_WORK_MASK,%edi
631 andl %edi,%edx
632 jnz retint_careful
633 - /*
634 - * The iret might restore flags:
635 - */
636 - TRACE_IRQS_IRETQ
637 jmp retint_restore_args
638
639 #if 0
640 @@ -1071,7 +1071,7 @@ child_rip:
641 movq %rsi, %rdi
642 call *%rax
643 # exit
644 - xorl %edi, %edi
645 + mov %eax, %edi
646 call do_exit
647 CFI_ENDPROC
648 ENDPROC(child_rip)
649 --- sle11-2009-10-16.orig/arch/x86/kernel/genapic_64-xen.c 2008-12-15 11:27:22.000000000 +0100
650 +++ sle11-2009-10-16/arch/x86/kernel/genapic_64-xen.c 2009-02-16 16:18:36.000000000 +0100
651 @@ -24,12 +24,21 @@
652 #include <acpi/acpi_bus.h>
653 #endif
654
655 -/* which logical CPU number maps to which CPU (physical APIC ID) */
656 -u8 x86_cpu_to_apicid[NR_CPUS] __read_mostly
657 +/*
658 + * which logical CPU number maps to which CPU (physical APIC ID)
659 + *
660 + * The following static array is used during kernel startup
661 + * and the x86_cpu_to_apicid_ptr contains the address of the
662 + * array during this time. Is it zeroed when the per_cpu
663 + * data area is removed.
664 + */
665 +#ifndef CONFIG_XEN
666 +u8 x86_cpu_to_apicid_init[NR_CPUS] __initdata
667 = { [0 ... NR_CPUS-1] = BAD_APICID };
668 -EXPORT_SYMBOL(x86_cpu_to_apicid);
669 -
670 -u8 x86_cpu_to_log_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
671 +void *x86_cpu_to_apicid_ptr;
672 +#endif
673 +DEFINE_PER_CPU(u8, x86_cpu_to_apicid) = BAD_APICID;
674 +EXPORT_PER_CPU_SYMBOL(x86_cpu_to_apicid);
675
676 #ifndef CONFIG_XEN
677 struct genapic __read_mostly *genapic = &apic_flat;
678 --- sle11-2009-10-16.orig/arch/x86/kernel/head64-xen.c 2009-02-16 16:17:21.000000000 +0100
679 +++ sle11-2009-10-16/arch/x86/kernel/head64-xen.c 2009-02-16 16:18:36.000000000 +0100
680 @@ -1,5 +1,5 @@
681 /*
682 - * linux/arch/x86_64/kernel/head64.c -- prepare to run common code
683 + * prepare to run common code
684 *
685 * Copyright (C) 2000 Andrea Arcangeli <andrea@suse.de> SuSE
686 *
687 @@ -21,7 +21,6 @@
688 #include <asm/processor.h>
689 #include <asm/proto.h>
690 #include <asm/smp.h>
691 -#include <asm/bootsetup.h>
692 #include <asm/setup.h>
693 #include <asm/desc.h>
694 #include <asm/pgtable.h>
695 @@ -47,27 +46,16 @@ static void __init clear_bss(void)
696 }
697 #endif
698
699 -#define NEW_CL_POINTER 0x228 /* Relative to real mode data */
700 -#define OLD_CL_MAGIC_ADDR 0x20
701 -#define OLD_CL_MAGIC 0xA33F
702 -#define OLD_CL_OFFSET 0x22
703 -
704 static void __init copy_bootdata(char *real_mode_data)
705 {
706 #ifndef CONFIG_XEN
707 - unsigned long new_data;
708 char * command_line;
709
710 - memcpy(x86_boot_params, real_mode_data, BOOT_PARAM_SIZE);
711 - new_data = *(u32 *) (x86_boot_params + NEW_CL_POINTER);
712 - if (!new_data) {
713 - if (OLD_CL_MAGIC != *(u16 *)(real_mode_data + OLD_CL_MAGIC_ADDR)) {
714 - return;
715 - }
716 - new_data = __pa(real_mode_data) + *(u16 *)(real_mode_data + OLD_CL_OFFSET);
717 + memcpy(&boot_params, real_mode_data, sizeof boot_params);
718 + if (boot_params.hdr.cmd_line_ptr) {
719 + command_line = __va(boot_params.hdr.cmd_line_ptr);
720 + memcpy(boot_command_line, command_line, COMMAND_LINE_SIZE);
721 }
722 - command_line = __va(new_data);
723 - memcpy(boot_command_line, command_line, COMMAND_LINE_SIZE);
724 #else
725 int max_cmdline;
726
727 @@ -117,7 +105,7 @@ void __init x86_64_start_kernel(char * r
728
729 for (i = 0; i < IDT_ENTRIES; i++)
730 set_intr_gate(i, early_idt_handler);
731 - asm volatile("lidt %0" :: "m" (idt_descr));
732 + load_idt((const struct desc_ptr *)&idt_descr);
733 #endif
734
735 early_printk("Kernel alive\n");
736 --- sle11-2009-10-16.orig/arch/x86/kernel/init_task-xen.c 2009-02-16 16:17:21.000000000 +0100
737 +++ sle11-2009-10-16/arch/x86/kernel/init_task-xen.c 2009-02-16 16:18:36.000000000 +0100
738 @@ -14,11 +14,11 @@ static struct fs_struct init_fs = INIT_F
739 static struct files_struct init_files = INIT_FILES;
740 static struct signal_struct init_signals = INIT_SIGNALS(init_signals);
741 static struct sighand_struct init_sighand = INIT_SIGHAND(init_sighand);
742 -
743 +#ifdef CONFIG_X86_XEN
744 #define swapper_pg_dir ((pgd_t *)NULL)
745 +#endif
746 struct mm_struct init_mm = INIT_MM(init_mm);
747 #undef swapper_pg_dir
748 -
749 EXPORT_SYMBOL(init_mm);
750
751 /*
752 @@ -28,7 +28,7 @@ EXPORT_SYMBOL(init_mm);
753 * way process stacks are handled. This is done by having a special
754 * "init_task" linker map entry..
755 */
756 -union thread_union init_thread_union
757 +union thread_union init_thread_union
758 __attribute__((__section__(".data.init_task"))) =
759 { INIT_THREAD_INFO(init_task) };
760
761 @@ -38,14 +38,15 @@ union thread_union init_thread_union
762 * All other task structs will be allocated on slabs in fork.c
763 */
764 struct task_struct init_task = INIT_TASK(init_task);
765 -
766 EXPORT_SYMBOL(init_task);
767
768 #ifndef CONFIG_X86_NO_TSS
769 /*
770 - * per-CPU TSS segments. Threads are completely 'soft' on Linux,
771 - * no more per-task TSS's.
772 - */
773 + * no more per-task TSS's. The TSS size is kept cacheline-aligned
774 + * so they are allowed to end up in the .data.cacheline_aligned
775 + * section. Since TSS's are completely CPU-local, we want them
776 + * on exact cacheline boundaries, to eliminate cacheline ping-pong.
777 + */
778 DEFINE_PER_CPU_SHARED_ALIGNED(struct tss_struct, init_tss) = INIT_TSS;
779 #endif
780
781 --- sle11-2009-10-16.orig/arch/x86/kernel/io_apic_32-xen.c 2009-02-16 16:17:21.000000000 +0100
782 +++ sle11-2009-10-16/arch/x86/kernel/io_apic_32-xen.c 2009-02-16 16:18:36.000000000 +0100
783 @@ -427,7 +427,7 @@ static struct irq_cpu_info {
784
785 #define IRQ_ALLOWED(cpu, allowed_mask) cpu_isset(cpu, allowed_mask)
786
787 -#define CPU_TO_PACKAGEINDEX(i) (first_cpu(cpu_sibling_map[i]))
788 +#define CPU_TO_PACKAGEINDEX(i) (first_cpu(per_cpu(cpu_sibling_map, i)))
789
790 static cpumask_t balance_irq_affinity[NR_IRQS] = {
791 [0 ... NR_IRQS-1] = CPU_MASK_ALL
792 @@ -633,7 +633,7 @@ tryanotherirq:
793
794 imbalance = move_this_load;
795
796 - /* For physical_balance case, we accumlated both load
797 + /* For physical_balance case, we accumulated both load
798 * values in the one of the siblings cpu_irq[],
799 * to use the same code for physical and logical processors
800 * as much as possible.
801 @@ -647,7 +647,7 @@ tryanotherirq:
802 * (A+B)/2 vs B
803 */
804 load = CPU_IRQ(min_loaded) >> 1;
805 - for_each_cpu_mask(j, cpu_sibling_map[min_loaded]) {
806 + for_each_cpu_mask(j, per_cpu(cpu_sibling_map, min_loaded)) {
807 if (load > CPU_IRQ(j)) {
808 /* This won't change cpu_sibling_map[min_loaded] */
809 load = CPU_IRQ(j);
810 @@ -1018,7 +1018,7 @@ static int EISA_ELCR(unsigned int irq)
811 #define default_MCA_trigger(idx) (1)
812 #define default_MCA_polarity(idx) (0)
813
814 -static int __init MPBIOS_polarity(int idx)
815 +static int MPBIOS_polarity(int idx)
816 {
817 int bus = mp_irqs[idx].mpc_srcbus;
818 int polarity;
819 @@ -1347,6 +1347,11 @@ static void __init setup_IO_APIC_irqs(vo
820 continue;
821 }
822
823 + if (!first_notcon) {
824 + apic_printk(APIC_VERBOSE, " not connected.\n");
825 + first_notcon = 1;
826 + }
827 +
828 entry.trigger = irq_trigger(idx);
829 entry.polarity = irq_polarity(idx);
830
831 @@ -1936,13 +1941,16 @@ __setup("no_timer_check", notimercheck);
832 static int __init timer_irq_works(void)
833 {
834 unsigned long t1 = jiffies;
835 + unsigned long flags;
836
837 if (no_timer_check)
838 return 1;
839
840 + local_save_flags(flags);
841 local_irq_enable();
842 /* Let ten ticks pass... */
843 mdelay((10 * 1000) / HZ);
844 + local_irq_restore(flags);
845
846 /*
847 * Expect a few ticks at least, to be sure some possible
848 @@ -2223,6 +2231,9 @@ static inline void __init check_timer(vo
849 {
850 int apic1, pin1, apic2, pin2;
851 int vector;
852 + unsigned long flags;
853 +
854 + local_irq_save(flags);
855
856 /*
857 * get/set the timer IRQ vector:
858 @@ -2268,7 +2279,7 @@ static inline void __init check_timer(vo
859 }
860 if (disable_timer_pin_1 > 0)
861 clear_IO_APIC_pin(0, pin1);
862 - return;
863 + goto out;
864 }
865 clear_IO_APIC_pin(apic1, pin1);
866 printk(KERN_ERR "..MP-BIOS bug: 8254 timer not connected to "
867 @@ -2291,7 +2302,7 @@ static inline void __init check_timer(vo
868 if (nmi_watchdog == NMI_IO_APIC) {
869 setup_nmi();
870 }
871 - return;
872 + goto out;
873 }
874 /*
875 * Cleanup, just in case ...
876 @@ -2315,7 +2326,7 @@ static inline void __init check_timer(vo
877
878 if (timer_irq_works()) {
879 printk(" works.\n");
880 - return;
881 + goto out;
882 }
883 apic_write_around(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | vector);
884 printk(" failed.\n");
885 @@ -2331,11 +2342,13 @@ static inline void __init check_timer(vo
886
887 if (timer_irq_works()) {
888 printk(" works.\n");
889 - return;
890 + goto out;
891 }
892 printk(" failed :(.\n");
893 panic("IO-APIC + timer doesn't work! Boot with apic=debug and send a "
894 "report. Then try booting with the 'noapic' option");
895 +out:
896 + local_irq_restore(flags);
897 }
898 #else
899 int timer_uses_ioapic_pin_0 = 0;
900 @@ -2353,6 +2366,14 @@ int timer_uses_ioapic_pin_0 = 0;
901
902 void __init setup_IO_APIC(void)
903 {
904 +#ifndef CONFIG_XEN
905 + int i;
906 +
907 + /* Reserve all the system vectors. */
908 + for (i = FIRST_SYSTEM_VECTOR; i < NR_VECTORS; i++)
909 + set_bit(i, used_vectors);
910 +#endif
911 +
912 enable_IO_APIC();
913
914 if (acpi_ioapic)
915 @@ -2542,7 +2563,7 @@ void destroy_irq(unsigned int irq)
916 #endif /* CONFIG_XEN */
917
918 /*
919 - * MSI mesage composition
920 + * MSI message composition
921 */
922 #if defined(CONFIG_PCI_MSI) && !defined(CONFIG_XEN)
923 static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg)
924 @@ -2899,6 +2920,25 @@ int io_apic_set_pci_routing (int ioapic,
925 return 0;
926 }
927
928 +int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
929 +{
930 + int i;
931 +
932 + if (skip_ioapic_setup)
933 + return -1;
934 +
935 + for (i = 0; i < mp_irq_entries; i++)
936 + if (mp_irqs[i].mpc_irqtype == mp_INT &&
937 + mp_irqs[i].mpc_srcbusirq == bus_irq)
938 + break;
939 + if (i >= mp_irq_entries)
940 + return -1;
941 +
942 + *trigger = irq_trigger(i);
943 + *polarity = irq_polarity(i);
944 + return 0;
945 +}
946 +
947 #endif /* CONFIG_ACPI */
948
949 static int __init parse_disable_timer_pin_1(char *arg)
950 --- sle11-2009-10-16.orig/arch/x86/kernel/io_apic_64-xen.c 2009-02-16 16:17:21.000000000 +0100
951 +++ sle11-2009-10-16/arch/x86/kernel/io_apic_64-xen.c 2009-02-16 16:18:36.000000000 +0100
952 @@ -31,6 +31,7 @@
953 #include <linux/sysdev.h>
954 #include <linux/msi.h>
955 #include <linux/htirq.h>
956 +#include <linux/dmar.h>
957 #ifdef CONFIG_ACPI
958 #include <acpi/acpi_bus.h>
959 #endif
960 @@ -584,7 +585,7 @@ int IO_APIC_get_PCI_irq_vector(int bus,
961 #define default_PCI_trigger(idx) (1)
962 #define default_PCI_polarity(idx) (1)
963
964 -static int __init MPBIOS_polarity(int idx)
965 +static int MPBIOS_polarity(int idx)
966 {
967 int bus = mp_irqs[idx].mpc_srcbus;
968 int polarity;
969 @@ -871,6 +872,10 @@ static void __init setup_IO_APIC_irqs(vo
970 apic_printk(APIC_VERBOSE, ", %d-%d", mp_ioapics[apic].mpc_apicid, pin);
971 continue;
972 }
973 + if (!first_notcon) {
974 + apic_printk(APIC_VERBOSE, " not connected.\n");
975 + first_notcon = 1;
976 + }
977
978 irq = pin_2_irq(idx, apic, pin);
979 add_pin_to_irq(irq, apic, pin);
980 @@ -881,7 +886,7 @@ static void __init setup_IO_APIC_irqs(vo
981 }
982
983 if (!first_notcon)
984 - apic_printk(APIC_VERBOSE," not connected.\n");
985 + apic_printk(APIC_VERBOSE, " not connected.\n");
986 }
987
988 #ifndef CONFIG_XEN
989 @@ -1277,10 +1282,13 @@ void disable_IO_APIC(void)
990 static int __init timer_irq_works(void)
991 {
992 unsigned long t1 = jiffies;
993 + unsigned long flags;
994
995 + local_save_flags(flags);
996 local_irq_enable();
997 /* Let ten ticks pass... */
998 mdelay((10 * 1000) / HZ);
999 + local_irq_restore(flags);
1000
1001 /*
1002 * Expect a few ticks at least, to be sure some possible
1003 @@ -1655,6 +1663,9 @@ static inline void check_timer(void)
1004 {
1005 struct irq_cfg *cfg = irq_cfg + 0;
1006 int apic1, pin1, apic2, pin2;
1007 + unsigned long flags;
1008 +
1009 + local_irq_save(flags);
1010
1011 /*
1012 * get/set the timer IRQ vector:
1013 @@ -1696,7 +1707,7 @@ static inline void check_timer(void)
1014 }
1015 if (disable_timer_pin_1 > 0)
1016 clear_IO_APIC_pin(0, pin1);
1017 - return;
1018 + goto out;
1019 }
1020 clear_IO_APIC_pin(apic1, pin1);
1021 apic_printk(APIC_QUIET,KERN_ERR "..MP-BIOS bug: 8254 timer not "
1022 @@ -1718,7 +1729,7 @@ static inline void check_timer(void)
1023 if (nmi_watchdog == NMI_IO_APIC) {
1024 setup_nmi();
1025 }
1026 - return;
1027 + goto out;
1028 }
1029 /*
1030 * Cleanup, just in case ...
1031 @@ -1741,7 +1752,7 @@ static inline void check_timer(void)
1032
1033 if (timer_irq_works()) {
1034 apic_printk(APIC_VERBOSE," works.\n");
1035 - return;
1036 + goto out;
1037 }
1038 apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | cfg->vector);
1039 apic_printk(APIC_VERBOSE," failed.\n");
1040 @@ -1756,10 +1767,12 @@ static inline void check_timer(void)
1041
1042 if (timer_irq_works()) {
1043 apic_printk(APIC_VERBOSE," works.\n");
1044 - return;
1045 + goto out;
1046 }
1047 apic_printk(APIC_VERBOSE," failed :(.\n");
1048 panic("IO-APIC + timer doesn't work! Try using the 'noapic' kernel parameter\n");
1049 +out:
1050 + local_irq_restore(flags);
1051 }
1052 #else
1053 #define check_timer() ((void)0)
1054 @@ -1775,7 +1788,7 @@ __setup("no_timer_check", notimercheck);
1055
1056 /*
1057 *
1058 - * IRQ's that are handled by the PIC in the MPS IOAPIC case.
1059 + * IRQs that are handled by the PIC in the MPS IOAPIC case.
1060 * - IRQ2 is the cascade IRQ, and cannot be a io-apic IRQ.
1061 * Linux doesn't really care, as it's not actually used
1062 * for any interrupt handling anyway.
1063 @@ -1858,7 +1871,7 @@ static struct sysdev_class ioapic_sysdev
1064 static int __init ioapic_init_sysfs(void)
1065 {
1066 struct sys_device * dev;
1067 - int i, size, error = 0;
1068 + int i, size, error;
1069
1070 error = sysdev_class_register(&ioapic_sysdev_class);
1071 if (error)
1072 @@ -1867,12 +1880,11 @@ static int __init ioapic_init_sysfs(void
1073 for (i = 0; i < nr_ioapics; i++ ) {
1074 size = sizeof(struct sys_device) + nr_ioapic_registers[i]
1075 * sizeof(struct IO_APIC_route_entry);
1076 - mp_ioapic_data[i] = kmalloc(size, GFP_KERNEL);
1077 + mp_ioapic_data[i] = kzalloc(size, GFP_KERNEL);
1078 if (!mp_ioapic_data[i]) {
1079 printk(KERN_ERR "Can't suspend/resume IOAPIC %d\n", i);
1080 continue;
1081 }
1082 - memset(mp_ioapic_data[i], 0, size);
1083 dev = &mp_ioapic_data[i]->dev;
1084 dev->id = i;
1085 dev->cls = &ioapic_sysdev_class;
1086 @@ -1933,7 +1945,7 @@ void destroy_irq(unsigned int irq)
1087 #endif /* CONFIG_XEN */
1088
1089 /*
1090 - * MSI mesage composition
1091 + * MSI message composition
1092 */
1093 #if defined(CONFIG_PCI_MSI) && !defined(CONFIG_XEN)
1094 static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_msg *msg)
1095 @@ -2043,8 +2055,64 @@ void arch_teardown_msi_irq(unsigned int
1096 destroy_irq(irq);
1097 }
1098
1099 -#endif /* CONFIG_PCI_MSI */
1100 +#ifdef CONFIG_DMAR
1101 +#ifdef CONFIG_SMP
1102 +static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask)
1103 +{
1104 + struct irq_cfg *cfg = irq_cfg + irq;
1105 + struct msi_msg msg;
1106 + unsigned int dest;
1107 + cpumask_t tmp;
1108 +
1109 + cpus_and(tmp, mask, cpu_online_map);
1110 + if (cpus_empty(tmp))
1111 + return;
1112 +
1113 + if (assign_irq_vector(irq, mask))
1114 + return;
1115 +
1116 + cpus_and(tmp, cfg->domain, mask);
1117 + dest = cpu_mask_to_apicid(tmp);
1118 +
1119 + dmar_msi_read(irq, &msg);
1120 +
1121 + msg.data &= ~MSI_DATA_VECTOR_MASK;
1122 + msg.data |= MSI_DATA_VECTOR(cfg->vector);
1123 + msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK;
1124 + msg.address_lo |= MSI_ADDR_DEST_ID(dest);
1125 +
1126 + dmar_msi_write(irq, &msg);
1127 + irq_desc[irq].affinity = mask;
1128 +}
1129 +#endif /* CONFIG_SMP */
1130 +
1131 +struct irq_chip dmar_msi_type = {
1132 + .name = "DMAR_MSI",
1133 + .unmask = dmar_msi_unmask,
1134 + .mask = dmar_msi_mask,
1135 + .ack = ack_apic_edge,
1136 +#ifdef CONFIG_SMP
1137 + .set_affinity = dmar_msi_set_affinity,
1138 +#endif
1139 + .retrigger = ioapic_retrigger_irq,
1140 +};
1141 +
1142 +int arch_setup_dmar_msi(unsigned int irq)
1143 +{
1144 + int ret;
1145 + struct msi_msg msg;
1146 +
1147 + ret = msi_compose_msg(NULL, irq, &msg);
1148 + if (ret < 0)
1149 + return ret;
1150 + dmar_msi_write(irq, &msg);
1151 + set_irq_chip_and_handler_name(irq, &dmar_msi_type, handle_edge_irq,
1152 + "edge");
1153 + return 0;
1154 +}
1155 +#endif
1156
1157 +#endif /* CONFIG_PCI_MSI */
1158 /*
1159 * Hypertransport interrupt support
1160 */
1161 @@ -2177,8 +2245,27 @@ int io_apic_set_pci_routing (int ioapic,
1162 return 0;
1163 }
1164
1165 -#endif /* CONFIG_ACPI */
1166
1167 +int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
1168 +{
1169 + int i;
1170 +
1171 + if (skip_ioapic_setup)
1172 + return -1;
1173 +
1174 + for (i = 0; i < mp_irq_entries; i++)
1175 + if (mp_irqs[i].mpc_irqtype == mp_INT &&
1176 + mp_irqs[i].mpc_srcbusirq == bus_irq)
1177 + break;
1178 + if (i >= mp_irq_entries)
1179 + return -1;
1180 +
1181 + *trigger = irq_trigger(i);
1182 + *polarity = irq_polarity(i);
1183 + return 0;
1184 +}
1185 +
1186 +#endif /* CONFIG_ACPI */
1187
1188 #ifndef CONFIG_XEN
1189 /*
1190 @@ -2217,3 +2304,4 @@ void __init setup_ioapic_dest(void)
1191 }
1192 #endif
1193 #endif /* !CONFIG_XEN */
1194 +
1195 --- sle11-2009-10-16.orig/arch/x86/kernel/ioport_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1196 +++ sle11-2009-10-16/arch/x86/kernel/ioport_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1197 @@ -1,6 +1,4 @@
1198 /*
1199 - * linux/arch/i386/kernel/ioport.c
1200 - *
1201 * This contains the io-permission bitmap code - written by obz, with changes
1202 * by Linus.
1203 */
1204 --- sle11-2009-10-16.orig/arch/x86/kernel/ioport_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1205 +++ sle11-2009-10-16/arch/x86/kernel/ioport_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1206 @@ -1,6 +1,4 @@
1207 /*
1208 - * linux/arch/x86_64/kernel/ioport.c
1209 - *
1210 * This contains the io-permission bitmap code - written by obz, with changes
1211 * by Linus.
1212 */
1213 --- sle11-2009-10-16.orig/arch/x86/kernel/irq_32-xen.c 2009-02-16 16:17:21.000000000 +0100
1214 +++ sle11-2009-10-16/arch/x86/kernel/irq_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1215 @@ -1,6 +1,4 @@
1216 /*
1217 - * linux/arch/i386/kernel/irq.c
1218 - *
1219 * Copyright (C) 1992, 1998 Linus Torvalds, Ingo Molnar
1220 *
1221 * This file contains the lowest level x86-specific interrupt
1222 @@ -231,8 +229,6 @@ asmlinkage void do_softirq(void)
1223
1224 local_irq_restore(flags);
1225 }
1226 -
1227 -EXPORT_SYMBOL(do_softirq);
1228 #endif
1229
1230 /*
1231 @@ -259,9 +255,17 @@ int show_interrupts(struct seq_file *p,
1232 }
1233
1234 if (i < NR_IRQS) {
1235 + unsigned any_count = 0;
1236 +
1237 spin_lock_irqsave(&irq_desc[i].lock, flags);
1238 +#ifndef CONFIG_SMP
1239 + any_count = kstat_irqs(i);
1240 +#else
1241 + for_each_online_cpu(j)
1242 + any_count |= kstat_cpu(j).irqs[i];
1243 +#endif
1244 action = irq_desc[i].action;
1245 - if (!action)
1246 + if (!action && !any_count)
1247 goto skip;
1248 seq_printf(p, "%3d: ",i);
1249 #ifndef CONFIG_SMP
1250 @@ -272,10 +276,12 @@ int show_interrupts(struct seq_file *p,
1251 #endif
1252 seq_printf(p, " %8s", irq_desc[i].chip->name);
1253 seq_printf(p, "-%-8s", irq_desc[i].name);
1254 - seq_printf(p, " %s", action->name);
1255
1256 - for (action=action->next; action; action = action->next)
1257 - seq_printf(p, ", %s", action->name);
1258 + if (action) {
1259 + seq_printf(p, " %s", action->name);
1260 + while ((action = action->next) != NULL)
1261 + seq_printf(p, ", %s", action->name);
1262 + }
1263
1264 seq_putc(p, '\n');
1265 skip:
1266 @@ -284,13 +290,46 @@ skip:
1267 seq_printf(p, "NMI: ");
1268 for_each_online_cpu(j)
1269 seq_printf(p, "%10u ", nmi_count(j));
1270 - seq_putc(p, '\n');
1271 + seq_printf(p, " Non-maskable interrupts\n");
1272 #ifdef CONFIG_X86_LOCAL_APIC
1273 seq_printf(p, "LOC: ");
1274 for_each_online_cpu(j)
1275 seq_printf(p, "%10u ",
1276 per_cpu(irq_stat,j).apic_timer_irqs);
1277 - seq_putc(p, '\n');
1278 + seq_printf(p, " Local timer interrupts\n");
1279 +#endif
1280 +#ifdef CONFIG_SMP
1281 + seq_printf(p, "RES: ");
1282 + for_each_online_cpu(j)
1283 + seq_printf(p, "%10u ",
1284 + per_cpu(irq_stat,j).irq_resched_count);
1285 + seq_printf(p, " Rescheduling interrupts\n");
1286 + seq_printf(p, "CAL: ");
1287 + for_each_online_cpu(j)
1288 + seq_printf(p, "%10u ",
1289 + per_cpu(irq_stat,j).irq_call_count);
1290 + seq_printf(p, " function call interrupts\n");
1291 +#ifndef CONFIG_XEN
1292 + seq_printf(p, "TLB: ");
1293 + for_each_online_cpu(j)
1294 + seq_printf(p, "%10u ",
1295 + per_cpu(irq_stat,j).irq_tlb_count);
1296 + seq_printf(p, " TLB shootdowns\n");
1297 +#endif
1298 +#endif
1299 +#ifdef CONFIG_X86_MCE
1300 + seq_printf(p, "TRM: ");
1301 + for_each_online_cpu(j)
1302 + seq_printf(p, "%10u ",
1303 + per_cpu(irq_stat,j).irq_thermal_count);
1304 + seq_printf(p, " Thermal event interrupts\n");
1305 +#endif
1306 +#ifdef CONFIG_X86_LOCAL_APIC
1307 + seq_printf(p, "SPU: ");
1308 + for_each_online_cpu(j)
1309 + seq_printf(p, "%10u ",
1310 + per_cpu(irq_stat,j).irq_spurious_count);
1311 + seq_printf(p, " Spurious interrupts\n");
1312 #endif
1313 seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count));
1314 #if defined(CONFIG_X86_IO_APIC)
1315 --- sle11-2009-10-16.orig/arch/x86/kernel/irq_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1316 +++ sle11-2009-10-16/arch/x86/kernel/irq_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1317 @@ -1,6 +1,4 @@
1318 /*
1319 - * linux/arch/x86_64/kernel/irq.c
1320 - *
1321 * Copyright (C) 1992, 1998 Linus Torvalds, Ingo Molnar
1322 *
1323 * This file contains the lowest level x86_64-specific interrupt
1324 @@ -64,9 +62,17 @@ int show_interrupts(struct seq_file *p,
1325 }
1326
1327 if (i < NR_IRQS) {
1328 + unsigned any_count = 0;
1329 +
1330 spin_lock_irqsave(&irq_desc[i].lock, flags);
1331 +#ifndef CONFIG_SMP
1332 + any_count = kstat_irqs(i);
1333 +#else
1334 + for_each_online_cpu(j)
1335 + any_count |= kstat_cpu(j).irqs[i];
1336 +#endif
1337 action = irq_desc[i].action;
1338 - if (!action)
1339 + if (!action && !any_count)
1340 goto skip;
1341 seq_printf(p, "%3d: ",i);
1342 #ifndef CONFIG_SMP
1343 @@ -78,9 +84,11 @@ int show_interrupts(struct seq_file *p,
1344 seq_printf(p, " %8s", irq_desc[i].chip->name);
1345 seq_printf(p, "-%-8s", irq_desc[i].name);
1346
1347 - seq_printf(p, " %s", action->name);
1348 - for (action=action->next; action; action = action->next)
1349 - seq_printf(p, ", %s", action->name);
1350 + if (action) {
1351 + seq_printf(p, " %s", action->name);
1352 + while ((action = action->next) != NULL)
1353 + seq_printf(p, ", %s", action->name);
1354 + }
1355 seq_putc(p, '\n');
1356 skip:
1357 spin_unlock_irqrestore(&irq_desc[i].lock, flags);
1358 @@ -88,12 +96,44 @@ skip:
1359 seq_printf(p, "NMI: ");
1360 for_each_online_cpu(j)
1361 seq_printf(p, "%10u ", cpu_pda(j)->__nmi_count);
1362 - seq_putc(p, '\n');
1363 + seq_printf(p, " Non-maskable interrupts\n");
1364 #ifdef CONFIG_X86_LOCAL_APIC
1365 seq_printf(p, "LOC: ");
1366 for_each_online_cpu(j)
1367 seq_printf(p, "%10u ", cpu_pda(j)->apic_timer_irqs);
1368 - seq_putc(p, '\n');
1369 + seq_printf(p, " Local timer interrupts\n");
1370 +#endif
1371 +#ifdef CONFIG_SMP
1372 + seq_printf(p, "RES: ");
1373 + for_each_online_cpu(j)
1374 + seq_printf(p, "%10u ", cpu_pda(j)->irq_resched_count);
1375 + seq_printf(p, " Rescheduling interrupts\n");
1376 + seq_printf(p, "CAL: ");
1377 + for_each_online_cpu(j)
1378 + seq_printf(p, "%10u ", cpu_pda(j)->irq_call_count);
1379 + seq_printf(p, " function call interrupts\n");
1380 +#ifndef CONFIG_XEN
1381 + seq_printf(p, "TLB: ");
1382 + for_each_online_cpu(j)
1383 + seq_printf(p, "%10u ", cpu_pda(j)->irq_tlb_count);
1384 + seq_printf(p, " TLB shootdowns\n");
1385 +#endif
1386 +#endif
1387 +#ifdef CONFIG_X86_MCE
1388 + seq_printf(p, "TRM: ");
1389 + for_each_online_cpu(j)
1390 + seq_printf(p, "%10u ", cpu_pda(j)->irq_thermal_count);
1391 + seq_printf(p, " Thermal event interrupts\n");
1392 + seq_printf(p, "THR: ");
1393 + for_each_online_cpu(j)
1394 + seq_printf(p, "%10u ", cpu_pda(j)->irq_threshold_count);
1395 + seq_printf(p, " Threshold APIC interrupts\n");
1396 +#endif
1397 +#ifdef CONFIG_X86_LOCAL_APIC
1398 + seq_printf(p, "SPU: ");
1399 + for_each_online_cpu(j)
1400 + seq_printf(p, "%10u ", cpu_pda(j)->irq_spurious_count);
1401 + seq_printf(p, " Spurious interrupts\n");
1402 #endif
1403 seq_printf(p, "ERR: %10u\n", atomic_read(&irq_err_count));
1404 }
1405 @@ -211,7 +251,6 @@ asmlinkage void do_softirq(void)
1406 }
1407 local_irq_restore(flags);
1408 }
1409 -EXPORT_SYMBOL(do_softirq);
1410
1411 #ifndef CONFIG_X86_LOCAL_APIC
1412 /*
1413 --- sle11-2009-10-16.orig/arch/x86/kernel/ldt_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1414 +++ sle11-2009-10-16/arch/x86/kernel/ldt_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1415 @@ -1,6 +1,4 @@
1416 /*
1417 - * linux/arch/i386/kernel/ldt.c
1418 - *
1419 * Copyright (C) 1992 Krishna Balasubramanian and Linus Torvalds
1420 * Copyright (C) 1999 Ingo Molnar <mingo@redhat.com>
1421 */
1422 @@ -106,14 +104,14 @@ int init_new_context(struct task_struct
1423 struct mm_struct * old_mm;
1424 int retval = 0;
1425
1426 - init_MUTEX(&mm->context.sem);
1427 + mutex_init(&mm->context.lock);
1428 mm->context.size = 0;
1429 mm->context.has_foreign_mappings = 0;
1430 old_mm = current->mm;
1431 if (old_mm && old_mm->context.size > 0) {
1432 - down(&old_mm->context.sem);
1433 + mutex_lock(&old_mm->context.lock);
1434 retval = copy_ldt(&mm->context, &old_mm->context);
1435 - up(&old_mm->context.sem);
1436 + mutex_unlock(&old_mm->context.lock);
1437 }
1438 return retval;
1439 }
1440 @@ -149,7 +147,7 @@ static int read_ldt(void __user * ptr, u
1441 if (bytecount > LDT_ENTRY_SIZE*LDT_ENTRIES)
1442 bytecount = LDT_ENTRY_SIZE*LDT_ENTRIES;
1443
1444 - down(&mm->context.sem);
1445 + mutex_lock(&mm->context.lock);
1446 size = mm->context.size*LDT_ENTRY_SIZE;
1447 if (size > bytecount)
1448 size = bytecount;
1449 @@ -157,7 +155,7 @@ static int read_ldt(void __user * ptr, u
1450 err = 0;
1451 if (copy_to_user(ptr, mm->context.ldt, size))
1452 err = -EFAULT;
1453 - up(&mm->context.sem);
1454 + mutex_unlock(&mm->context.lock);
1455 if (err < 0)
1456 goto error_return;
1457 if (size != bytecount) {
1458 @@ -213,7 +211,7 @@ static int write_ldt(void __user * ptr,
1459 goto out;
1460 }
1461
1462 - down(&mm->context.sem);
1463 + mutex_lock(&mm->context.lock);
1464 if (ldt_info.entry_number >= mm->context.size) {
1465 error = alloc_ldt(&current->mm->context, ldt_info.entry_number+1, 1);
1466 if (error < 0)
1467 @@ -240,7 +238,7 @@ install:
1468 entry_1, entry_2);
1469
1470 out_unlock:
1471 - up(&mm->context.sem);
1472 + mutex_unlock(&mm->context.lock);
1473 out:
1474 return error;
1475 }
1476 --- sle11-2009-10-16.orig/arch/x86/kernel/ldt_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1477 +++ sle11-2009-10-16/arch/x86/kernel/ldt_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1478 @@ -1,6 +1,4 @@
1479 /*
1480 - * linux/arch/x86_64/kernel/ldt.c
1481 - *
1482 * Copyright (C) 1992 Krishna Balasubramanian and Linus Torvalds
1483 * Copyright (C) 1999 Ingo Molnar <mingo@redhat.com>
1484 * Copyright (C) 2002 Andi Kleen
1485 @@ -112,19 +110,14 @@ int init_new_context(struct task_struct
1486 int retval = 0;
1487
1488 memset(&mm->context, 0, sizeof(mm->context));
1489 - init_MUTEX(&mm->context.sem);
1490 + mutex_init(&mm->context.lock);
1491 old_mm = current->mm;
1492 if (old_mm)
1493 mm->context.vdso = old_mm->context.vdso;
1494 if (old_mm && old_mm->context.size > 0) {
1495 - down(&old_mm->context.sem);
1496 + mutex_lock(&old_mm->context.lock);
1497 retval = copy_ldt(&mm->context, &old_mm->context);
1498 - up(&old_mm->context.sem);
1499 - }
1500 - if (retval == 0) {
1501 - spin_lock(&mm_unpinned_lock);
1502 - list_add(&mm->context.unpinned, &mm_unpinned);
1503 - spin_unlock(&mm_unpinned_lock);
1504 + mutex_unlock(&old_mm->context.lock);
1505 }
1506 return retval;
1507 }
1508 @@ -148,11 +141,6 @@ void destroy_context(struct mm_struct *m
1509 kfree(mm->context.ldt);
1510 mm->context.size = 0;
1511 }
1512 - if (!PagePinned(virt_to_page(mm->pgd))) {
1513 - spin_lock(&mm_unpinned_lock);
1514 - list_del(&mm->context.unpinned);
1515 - spin_unlock(&mm_unpinned_lock);
1516 - }
1517 }
1518
1519 static int read_ldt(void __user * ptr, unsigned long bytecount)
1520 @@ -166,7 +154,7 @@ static int read_ldt(void __user * ptr, u
1521 if (bytecount > LDT_ENTRY_SIZE*LDT_ENTRIES)
1522 bytecount = LDT_ENTRY_SIZE*LDT_ENTRIES;
1523
1524 - down(&mm->context.sem);
1525 + mutex_lock(&mm->context.lock);
1526 size = mm->context.size*LDT_ENTRY_SIZE;
1527 if (size > bytecount)
1528 size = bytecount;
1529 @@ -174,7 +162,7 @@ static int read_ldt(void __user * ptr, u
1530 err = 0;
1531 if (copy_to_user(ptr, mm->context.ldt, size))
1532 err = -EFAULT;
1533 - up(&mm->context.sem);
1534 + mutex_unlock(&mm->context.lock);
1535 if (err < 0)
1536 goto error_return;
1537 if (size != bytecount) {
1538 @@ -227,7 +215,7 @@ static int write_ldt(void __user * ptr,
1539 goto out;
1540 }
1541
1542 - down(&mm->context.sem);
1543 + mutex_lock(&mm->context.lock);
1544 if (ldt_info.entry_number >= (unsigned)mm->context.size) {
1545 error = alloc_ldt(&current->mm->context, ldt_info.entry_number+1, 1);
1546 if (error < 0)
1547 @@ -256,7 +244,7 @@ install:
1548 error = HYPERVISOR_update_descriptor(mach_lp, (unsigned long)((entry_1 | (unsigned long) entry_2 << 32)));
1549
1550 out_unlock:
1551 - up(&mm->context.sem);
1552 + mutex_unlock(&mm->context.lock);
1553 out:
1554 return error;
1555 }
1556 --- sle11-2009-10-16.orig/arch/x86/kernel/mpparse_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1557 +++ sle11-2009-10-16/arch/x86/kernel/mpparse_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1558 @@ -1023,7 +1023,7 @@ void __init mp_config_acpi_legacy_irqs (
1559
1560 /*
1561 * Use the default configuration for the IRQs 0-15. Unless
1562 - * overriden by (MADT) interrupt source override entries.
1563 + * overridden by (MADT) interrupt source override entries.
1564 */
1565 for (i = 0; i < 16; i++) {
1566 int idx;
1567 --- sle11-2009-10-16.orig/arch/x86/kernel/mpparse_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1568 +++ sle11-2009-10-16/arch/x86/kernel/mpparse_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1569 @@ -57,6 +57,8 @@ unsigned long mp_lapic_addr = 0;
1570
1571 /* Processor that is doing the boot up */
1572 unsigned int boot_cpu_id = -1U;
1573 +EXPORT_SYMBOL(boot_cpu_id);
1574 +
1575 /* Internal processor count */
1576 unsigned int num_processors __cpuinitdata = 0;
1577
1578 @@ -87,7 +89,7 @@ static int __init mpf_checksum(unsigned
1579 }
1580
1581 #ifndef CONFIG_XEN
1582 -static void __cpuinit MP_processor_info (struct mpc_config_processor *m)
1583 +static void __cpuinit MP_processor_info(struct mpc_config_processor *m)
1584 {
1585 int cpu;
1586 cpumask_t tmp_map;
1587 @@ -124,13 +126,24 @@ static void __cpuinit MP_processor_info
1588 cpu = 0;
1589 }
1590 bios_cpu_apicid[cpu] = m->mpc_apicid;
1591 - x86_cpu_to_apicid[cpu] = m->mpc_apicid;
1592 + /*
1593 + * We get called early in the the start_kernel initialization
1594 + * process when the per_cpu data area is not yet setup, so we
1595 + * use a static array that is removed after the per_cpu data
1596 + * area is created.
1597 + */
1598 + if (x86_cpu_to_apicid_ptr) {
1599 + u8 *x86_cpu_to_apicid = (u8 *)x86_cpu_to_apicid_ptr;
1600 + x86_cpu_to_apicid[cpu] = m->mpc_apicid;
1601 + } else {
1602 + per_cpu(x86_cpu_to_apicid, cpu) = m->mpc_apicid;
1603 + }
1604
1605 cpu_set(cpu, cpu_possible_map);
1606 cpu_set(cpu, cpu_present_map);
1607 }
1608 #else
1609 -static void __cpuinit MP_processor_info (struct mpc_config_processor *m)
1610 +static void __cpuinit MP_processor_info(struct mpc_config_processor *m)
1611 {
1612 num_processors++;
1613 }
1614 --- sle11-2009-10-16.orig/arch/x86/kernel/pci-dma-xen.c 2009-02-16 16:17:21.000000000 +0100
1615 +++ sle11-2009-10-16/arch/x86/kernel/pci-dma-xen.c 2009-02-16 16:18:36.000000000 +0100
1616 @@ -13,14 +13,13 @@
1617 #include <linux/pci.h>
1618 #include <linux/module.h>
1619 #include <linux/version.h>
1620 -#include <linux/pci.h>
1621 #include <asm/io.h>
1622 #include <xen/balloon.h>
1623 #include <xen/gnttab.h>
1624 #include <asm/swiotlb.h>
1625 #include <asm/tlbflush.h>
1626 -#include <asm-i386/mach-xen/asm/swiotlb.h>
1627 -#include <asm-i386/mach-xen/asm/gnttab_dma.h>
1628 +#include <asm/swiotlb_32.h>
1629 +#include <asm/gnttab_dma.h>
1630 #include <asm/bug.h>
1631
1632 #ifdef __x86_64__
1633 @@ -106,27 +105,29 @@ int range_straddles_page_boundary(paddr_
1634 }
1635
1636 int
1637 -dma_map_sg(struct device *hwdev, struct scatterlist *sg, int nents,
1638 +dma_map_sg(struct device *hwdev, struct scatterlist *sgl, int nents,
1639 enum dma_data_direction direction)
1640 {
1641 int i, rc;
1642
1643 BUG_ON(!valid_dma_direction(direction));
1644 - WARN_ON(nents == 0 || sg[0].length == 0);
1645 + WARN_ON(nents == 0 || sgl->length == 0);
1646
1647 if (swiotlb) {
1648 - rc = swiotlb_map_sg(hwdev, sg, nents, direction);
1649 + rc = swiotlb_map_sg(hwdev, sgl, nents, direction);
1650 } else {
1651 - for (i = 0; i < nents; i++ ) {
1652 - BUG_ON(!sg[i].page);
1653 - sg[i].dma_address =
1654 - gnttab_dma_map_page(sg[i].page) + sg[i].offset;
1655 - sg[i].dma_length = sg[i].length;
1656 + struct scatterlist *sg;
1657 +
1658 + for_each_sg(sgl, sg, nents, i) {
1659 + BUG_ON(!sg_page(sg));
1660 + sg->dma_address =
1661 + gnttab_dma_map_page(sg_page(sg)) + sg->offset;
1662 + sg->dma_length = sg->length;
1663 IOMMU_BUG_ON(address_needs_mapping(
1664 - hwdev, sg[i].dma_address));
1665 + hwdev, sg->dma_address));
1666 IOMMU_BUG_ON(range_straddles_page_boundary(
1667 - page_to_pseudophys(sg[i].page) + sg[i].offset,
1668 - sg[i].length));
1669 + page_to_pseudophys(sg_page(sg)) + sg->offset,
1670 + sg->length));
1671 }
1672 rc = nents;
1673 }
1674 @@ -137,17 +138,19 @@ dma_map_sg(struct device *hwdev, struct
1675 EXPORT_SYMBOL(dma_map_sg);
1676
1677 void
1678 -dma_unmap_sg(struct device *hwdev, struct scatterlist *sg, int nents,
1679 +dma_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nents,
1680 enum dma_data_direction direction)
1681 {
1682 int i;
1683
1684 BUG_ON(!valid_dma_direction(direction));
1685 if (swiotlb)
1686 - swiotlb_unmap_sg(hwdev, sg, nents, direction);
1687 + swiotlb_unmap_sg(hwdev, sgl, nents, direction);
1688 else {
1689 - for (i = 0; i < nents; i++ )
1690 - gnttab_dma_unmap_page(sg[i].dma_address);
1691 + struct scatterlist *sg;
1692 +
1693 + for_each_sg(sgl, sg, nents, i)
1694 + gnttab_dma_unmap_page(sg->dma_address);
1695 }
1696 }
1697 EXPORT_SYMBOL(dma_unmap_sg);
1698 @@ -261,7 +264,8 @@ void dma_free_coherent(struct device *de
1699 {
1700 struct dma_coherent_mem *mem = dev ? dev->dma_mem : NULL;
1701 int order = get_order(size);
1702 -
1703 +
1704 + WARN_ON(irqs_disabled()); /* for portability */
1705 if (mem && vaddr >= mem->virt_base && vaddr < (mem->virt_base + (mem->size << PAGE_SHIFT))) {
1706 int page = (vaddr - mem->virt_base) >> PAGE_SHIFT;
1707
1708 --- sle11-2009-10-16.orig/arch/x86/kernel/process_32-xen.c 2009-02-16 16:17:21.000000000 +0100
1709 +++ sle11-2009-10-16/arch/x86/kernel/process_32-xen.c 2009-02-16 16:18:36.000000000 +0100
1710 @@ -1,6 +1,4 @@
1711 /*
1712 - * linux/arch/i386/kernel/process.c
1713 - *
1714 * Copyright (C) 1995 Linus Torvalds
1715 *
1716 * Pentium III FXSR, SSE support
1717 @@ -190,6 +188,10 @@ void cpu_idle(void)
1718 }
1719 }
1720
1721 +static void do_nothing(void *unused)
1722 +{
1723 +}
1724 +
1725 void cpu_idle_wait(void)
1726 {
1727 unsigned int cpu, this_cpu = get_cpu();
1728 @@ -214,13 +216,20 @@ void cpu_idle_wait(void)
1729 cpu_clear(cpu, map);
1730 }
1731 cpus_and(map, map, cpu_online_map);
1732 + /*
1733 + * We waited 1 sec, if a CPU still did not call idle
1734 + * it may be because it is in idle and not waking up
1735 + * because it has nothing to do.
1736 + * Give all the remaining CPUS a kick.
1737 + */
1738 + smp_call_function_mask(map, do_nothing, 0, 0);
1739 } while (!cpus_empty(map));
1740
1741 set_cpus_allowed(current, tmp);
1742 }
1743 EXPORT_SYMBOL_GPL(cpu_idle_wait);
1744
1745 -void __devinit select_idle_routine(const struct cpuinfo_x86 *c)
1746 +void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
1747 {
1748 }
1749
1750 @@ -238,34 +247,52 @@ static int __init idle_setup(char *str)
1751 }
1752 early_param("idle", idle_setup);
1753
1754 -void show_regs(struct pt_regs * regs)
1755 +void __show_registers(struct pt_regs *regs, int all)
1756 {
1757 unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L;
1758 unsigned long d0, d1, d2, d3, d6, d7;
1759 + unsigned long esp;
1760 + unsigned short ss, gs;
1761 +
1762 + if (user_mode_vm(regs)) {
1763 + esp = regs->esp;
1764 + ss = regs->xss & 0xffff;
1765 + savesegment(gs, gs);
1766 + } else {
1767 + esp = (unsigned long) (&regs->esp);
1768 + savesegment(ss, ss);
1769 + savesegment(gs, gs);
1770 + }
1771
1772 printk("\n");
1773 - printk("Pid: %d, comm: %20s\n", current->pid, current->comm);
1774 - printk("EIP: %04x:[<%08lx>] CPU: %d\n",0xffff & regs->xcs,regs->eip, smp_processor_id());
1775 + printk("Pid: %d, comm: %s %s (%s %.*s)\n",
1776 + task_pid_nr(current), current->comm,
1777 + print_tainted(), init_utsname()->release,
1778 + (int)strcspn(init_utsname()->version, " "),
1779 + init_utsname()->version);
1780 +
1781 + printk("EIP: %04x:[<%08lx>] EFLAGS: %08lx CPU: %d\n",
1782 + 0xffff & regs->xcs, regs->eip, regs->eflags,
1783 + smp_processor_id());
1784 print_symbol("EIP is at %s\n", regs->eip);
1785
1786 - if (user_mode_vm(regs))
1787 - printk(" ESP: %04x:%08lx",0xffff & regs->xss,regs->esp);
1788 - printk(" EFLAGS: %08lx %s (%s %.*s)\n",
1789 - regs->eflags, print_tainted(), init_utsname()->release,
1790 - (int)strcspn(init_utsname()->version, " "),
1791 - init_utsname()->version);
1792 printk("EAX: %08lx EBX: %08lx ECX: %08lx EDX: %08lx\n",
1793 - regs->eax,regs->ebx,regs->ecx,regs->edx);
1794 - printk("ESI: %08lx EDI: %08lx EBP: %08lx",
1795 - regs->esi, regs->edi, regs->ebp);
1796 - printk(" DS: %04x ES: %04x FS: %04x\n",
1797 - 0xffff & regs->xds,0xffff & regs->xes, 0xffff & regs->xfs);
1798 + regs->eax, regs->ebx, regs->ecx, regs->edx);
1799 + printk("ESI: %08lx EDI: %08lx EBP: %08lx ESP: %08lx\n",
1800 + regs->esi, regs->edi, regs->ebp, esp);
1801 + printk(" DS: %04x ES: %04x FS: %04x GS: %04x SS: %04x\n",
1802 + regs->xds & 0xffff, regs->xes & 0xffff,
1803 + regs->xfs & 0xffff, gs, ss);
1804 +
1805 + if (!all)
1806 + return;
1807
1808 cr0 = read_cr0();
1809 cr2 = read_cr2();
1810 cr3 = read_cr3();
1811 cr4 = read_cr4_safe();
1812 - printk("CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n", cr0, cr2, cr3, cr4);
1813 + printk("CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n",
1814 + cr0, cr2, cr3, cr4);
1815
1816 get_debugreg(d0, 0);
1817 get_debugreg(d1, 1);
1818 @@ -273,10 +300,16 @@ void show_regs(struct pt_regs * regs)
1819 get_debugreg(d3, 3);
1820 printk("DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n",
1821 d0, d1, d2, d3);
1822 +
1823 get_debugreg(d6, 6);
1824 get_debugreg(d7, 7);
1825 - printk("DR6: %08lx DR7: %08lx\n", d6, d7);
1826 + printk("DR6: %08lx DR7: %08lx\n",
1827 + d6, d7);
1828 +}
1829
1830 +void show_regs(struct pt_regs *regs)
1831 +{
1832 + __show_registers(regs, 1);
1833 show_trace(NULL, regs, &regs->esp);
1834 }
1835
1836 --- sle11-2009-10-16.orig/arch/x86/kernel/process_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1837 +++ sle11-2009-10-16/arch/x86/kernel/process_64-xen.c 2009-02-16 16:18:36.000000000 +0100
1838 @@ -1,6 +1,4 @@
1839 /*
1840 - * linux/arch/x86-64/kernel/process.c
1841 - *
1842 * Copyright (C) 1995 Linus Torvalds
1843 *
1844 * Pentium III FXSR, SSE support
1845 @@ -41,6 +39,7 @@
1846 #include <linux/notifier.h>
1847 #include <linux/kprobes.h>
1848 #include <linux/kdebug.h>
1849 +#include <linux/tick.h>
1850
1851 #include <asm/uaccess.h>
1852 #include <asm/pgtable.h>
1853 @@ -172,6 +171,9 @@ void cpu_idle (void)
1854
1855 if (__get_cpu_var(cpu_idle_state))
1856 __get_cpu_var(cpu_idle_state) = 0;
1857 +
1858 + tick_nohz_stop_sched_tick();
1859 +
1860 rmb();
1861 idle = xen_idle; /* no alternatives */
1862 if (cpu_is_offline(smp_processor_id()))
1863 @@ -190,12 +192,17 @@ void cpu_idle (void)
1864 __exit_idle();
1865 }
1866
1867 + tick_nohz_restart_sched_tick();
1868 preempt_enable_no_resched();
1869 schedule();
1870 preempt_disable();
1871 }
1872 }
1873
1874 +static void do_nothing(void *unused)
1875 +{
1876 +}
1877 +
1878 void cpu_idle_wait(void)
1879 {
1880 unsigned int cpu, this_cpu = get_cpu();
1881 @@ -221,6 +228,13 @@ void cpu_idle_wait(void)
1882 cpu_clear(cpu, map);
1883 }
1884 cpus_and(map, map, cpu_online_map);
1885 + /*
1886 + * We waited 1 sec, if a CPU still did not call idle
1887 + * it may be because it is in idle and not waking up
1888 + * because it has nothing to do.
1889 + * Give all the remaining CPUS a kick.
1890 + */
1891 + smp_call_function_mask(map, do_nothing, 0, 0);
1892 } while (!cpus_empty(map));
1893
1894 set_cpus_allowed(current, tmp);
1895 @@ -528,7 +542,7 @@ static inline void __switch_to_xtra(stru
1896 *
1897 * Kprobes not supported here. Set the probe on schedule instead.
1898 */
1899 -__kprobes struct task_struct *
1900 +struct task_struct *
1901 __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
1902 {
1903 struct thread_struct *prev = &prev_p->thread,
1904 --- sle11-2009-10-16.orig/arch/x86/kernel/quirks-xen.c 2008-12-15 11:27:22.000000000 +0100
1905 +++ sle11-2009-10-16/arch/x86/kernel/quirks-xen.c 2009-02-16 16:18:36.000000000 +0100
1906 @@ -41,7 +41,353 @@ static void __devinit quirk_intel_irqbal
1907 if (!(config & 0x2))
1908 pci_write_config_byte(dev, 0xf4, config);
1909 }
1910 -DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7320_MCH, quirk_intel_irqbalance);
1911 -DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7525_MCH, quirk_intel_irqbalance);
1912 -DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7520_MCH, quirk_intel_irqbalance);
1913 +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7320_MCH,
1914 + quirk_intel_irqbalance);
1915 +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7525_MCH,
1916 + quirk_intel_irqbalance);
1917 +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_E7520_MCH,
1918 + quirk_intel_irqbalance);
1919 +#endif
1920 +
1921 +#if defined(CONFIG_HPET_TIMER)
1922 +#include <asm/hpet.h>
1923 +
1924 +unsigned long force_hpet_address;
1925 +
1926 +static enum {
1927 + NONE_FORCE_HPET_RESUME,
1928 + OLD_ICH_FORCE_HPET_RESUME,
1929 + ICH_FORCE_HPET_RESUME,
1930 + VT8237_FORCE_HPET_RESUME,
1931 + NVIDIA_FORCE_HPET_RESUME,
1932 +} force_hpet_resume_type;
1933 +
1934 +static void __iomem *rcba_base;
1935 +
1936 +static void ich_force_hpet_resume(void)
1937 +{
1938 + u32 val;
1939 +
1940 + if (!force_hpet_address)
1941 + return;
1942 +
1943 + if (rcba_base == NULL)
1944 + BUG();
1945 +
1946 + /* read the Function Disable register, dword mode only */
1947 + val = readl(rcba_base + 0x3404);
1948 + if (!(val & 0x80)) {
1949 + /* HPET disabled in HPTC. Trying to enable */
1950 + writel(val | 0x80, rcba_base + 0x3404);
1951 + }
1952 +
1953 + val = readl(rcba_base + 0x3404);
1954 + if (!(val & 0x80))
1955 + BUG();
1956 + else
1957 + printk(KERN_DEBUG "Force enabled HPET at resume\n");
1958 +
1959 + return;
1960 +}
1961 +
1962 +static void ich_force_enable_hpet(struct pci_dev *dev)
1963 +{
1964 + u32 val;
1965 + u32 uninitialized_var(rcba);
1966 + int err = 0;
1967 +
1968 + if (hpet_address || force_hpet_address)
1969 + return;
1970 +
1971 + pci_read_config_dword(dev, 0xF0, &rcba);
1972 + rcba &= 0xFFFFC000;
1973 + if (rcba == 0) {
1974 + printk(KERN_DEBUG "RCBA disabled. Cannot force enable HPET\n");
1975 + return;
1976 + }
1977 +
1978 + /* use bits 31:14, 16 kB aligned */
1979 + rcba_base = ioremap_nocache(rcba, 0x4000);
1980 + if (rcba_base == NULL) {
1981 + printk(KERN_DEBUG "ioremap failed. Cannot force enable HPET\n");
1982 + return;
1983 + }
1984 +
1985 + /* read the Function Disable register, dword mode only */
1986 + val = readl(rcba_base + 0x3404);
1987 +
1988 + if (val & 0x80) {
1989 + /* HPET is enabled in HPTC. Just not reported by BIOS */
1990 + val = val & 0x3;
1991 + force_hpet_address = 0xFED00000 | (val << 12);
1992 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
1993 + force_hpet_address);
1994 + iounmap(rcba_base);
1995 + return;
1996 + }
1997 +
1998 + /* HPET disabled in HPTC. Trying to enable */
1999 + writel(val | 0x80, rcba_base + 0x3404);
2000 +
2001 + val = readl(rcba_base + 0x3404);
2002 + if (!(val & 0x80)) {
2003 + err = 1;
2004 + } else {
2005 + val = val & 0x3;
2006 + force_hpet_address = 0xFED00000 | (val << 12);
2007 + }
2008 +
2009 + if (err) {
2010 + force_hpet_address = 0;
2011 + iounmap(rcba_base);
2012 + printk(KERN_DEBUG "Failed to force enable HPET\n");
2013 + } else {
2014 + force_hpet_resume_type = ICH_FORCE_HPET_RESUME;
2015 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
2016 + force_hpet_address);
2017 + }
2018 +}
2019 +
2020 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ESB2_0,
2021 + ich_force_enable_hpet);
2022 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH6_1,
2023 + ich_force_enable_hpet);
2024 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_0,
2025 + ich_force_enable_hpet);
2026 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_1,
2027 + ich_force_enable_hpet);
2028 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_31,
2029 + ich_force_enable_hpet);
2030 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH8_1,
2031 + ich_force_enable_hpet);
2032 +
2033 +
2034 +static struct pci_dev *cached_dev;
2035 +
2036 +static void old_ich_force_hpet_resume(void)
2037 +{
2038 + u32 val;
2039 + u32 uninitialized_var(gen_cntl);
2040 +
2041 + if (!force_hpet_address || !cached_dev)
2042 + return;
2043 +
2044 + pci_read_config_dword(cached_dev, 0xD0, &gen_cntl);
2045 + gen_cntl &= (~(0x7 << 15));
2046 + gen_cntl |= (0x4 << 15);
2047 +
2048 + pci_write_config_dword(cached_dev, 0xD0, gen_cntl);
2049 + pci_read_config_dword(cached_dev, 0xD0, &gen_cntl);
2050 + val = gen_cntl >> 15;
2051 + val &= 0x7;
2052 + if (val == 0x4)
2053 + printk(KERN_DEBUG "Force enabled HPET at resume\n");
2054 + else
2055 + BUG();
2056 +}
2057 +
2058 +static void old_ich_force_enable_hpet(struct pci_dev *dev)
2059 +{
2060 + u32 val;
2061 + u32 uninitialized_var(gen_cntl);
2062 +
2063 + if (hpet_address || force_hpet_address)
2064 + return;
2065 +
2066 + pci_read_config_dword(dev, 0xD0, &gen_cntl);
2067 + /*
2068 + * Bit 17 is HPET enable bit.
2069 + * Bit 16:15 control the HPET base address.
2070 + */
2071 + val = gen_cntl >> 15;
2072 + val &= 0x7;
2073 + if (val & 0x4) {
2074 + val &= 0x3;
2075 + force_hpet_address = 0xFED00000 | (val << 12);
2076 + printk(KERN_DEBUG "HPET at base address 0x%lx\n",
2077 + force_hpet_address);
2078 + return;
2079 + }
2080 +
2081 + /*
2082 + * HPET is disabled. Trying enabling at FED00000 and check
2083 + * whether it sticks
2084 + */
2085 + gen_cntl &= (~(0x7 << 15));
2086 + gen_cntl |= (0x4 << 15);
2087 + pci_write_config_dword(dev, 0xD0, gen_cntl);
2088 +
2089 + pci_read_config_dword(dev, 0xD0, &gen_cntl);
2090 +
2091 + val = gen_cntl >> 15;
2092 + val &= 0x7;
2093 + if (val & 0x4) {
2094 + /* HPET is enabled in HPTC. Just not reported by BIOS */
2095 + val &= 0x3;
2096 + force_hpet_address = 0xFED00000 | (val << 12);
2097 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
2098 + force_hpet_address);
2099 + cached_dev = dev;
2100 + force_hpet_resume_type = OLD_ICH_FORCE_HPET_RESUME;
2101 + return;
2102 + }
2103 +
2104 + printk(KERN_DEBUG "Failed to force enable HPET\n");
2105 +}
2106 +
2107 +/*
2108 + * Undocumented chipset features. Make sure that the user enforced
2109 + * this.
2110 + */
2111 +static void old_ich_force_enable_hpet_user(struct pci_dev *dev)
2112 +{
2113 + if (hpet_force_user)
2114 + old_ich_force_enable_hpet(dev);
2115 +}
2116 +
2117 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801CA_0,
2118 + old_ich_force_enable_hpet_user);
2119 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801CA_12,
2120 + old_ich_force_enable_hpet_user);
2121 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801DB_0,
2122 + old_ich_force_enable_hpet_user);
2123 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801DB_12,
2124 + old_ich_force_enable_hpet_user);
2125 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801EB_0,
2126 + old_ich_force_enable_hpet);
2127 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82801EB_12,
2128 + old_ich_force_enable_hpet);
2129 +
2130 +
2131 +static void vt8237_force_hpet_resume(void)
2132 +{
2133 + u32 val;
2134 +
2135 + if (!force_hpet_address || !cached_dev)
2136 + return;
2137 +
2138 + val = 0xfed00000 | 0x80;
2139 + pci_write_config_dword(cached_dev, 0x68, val);
2140 +
2141 + pci_read_config_dword(cached_dev, 0x68, &val);
2142 + if (val & 0x80)
2143 + printk(KERN_DEBUG "Force enabled HPET at resume\n");
2144 + else
2145 + BUG();
2146 +}
2147 +
2148 +static void vt8237_force_enable_hpet(struct pci_dev *dev)
2149 +{
2150 + u32 uninitialized_var(val);
2151 +
2152 + if (!hpet_force_user || hpet_address || force_hpet_address)
2153 + return;
2154 +
2155 + pci_read_config_dword(dev, 0x68, &val);
2156 + /*
2157 + * Bit 7 is HPET enable bit.
2158 + * Bit 31:10 is HPET base address (contrary to what datasheet claims)
2159 + */
2160 + if (val & 0x80) {
2161 + force_hpet_address = (val & ~0x3ff);
2162 + printk(KERN_DEBUG "HPET at base address 0x%lx\n",
2163 + force_hpet_address);
2164 + return;
2165 + }
2166 +
2167 + /*
2168 + * HPET is disabled. Trying enabling at FED00000 and check
2169 + * whether it sticks
2170 + */
2171 + val = 0xfed00000 | 0x80;
2172 + pci_write_config_dword(dev, 0x68, val);
2173 +
2174 + pci_read_config_dword(dev, 0x68, &val);
2175 + if (val & 0x80) {
2176 + force_hpet_address = (val & ~0x3ff);
2177 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
2178 + force_hpet_address);
2179 + cached_dev = dev;
2180 + force_hpet_resume_type = VT8237_FORCE_HPET_RESUME;
2181 + return;
2182 + }
2183 +
2184 + printk(KERN_DEBUG "Failed to force enable HPET\n");
2185 +}
2186 +
2187 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_VIA, PCI_DEVICE_ID_VIA_8235,
2188 + vt8237_force_enable_hpet);
2189 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_VIA, PCI_DEVICE_ID_VIA_8237,
2190 + vt8237_force_enable_hpet);
2191 +
2192 +/*
2193 + * Undocumented chipset feature taken from LinuxBIOS.
2194 + */
2195 +static void nvidia_force_hpet_resume(void)
2196 +{
2197 + pci_write_config_dword(cached_dev, 0x44, 0xfed00001);
2198 + printk(KERN_DEBUG "Force enabled HPET at resume\n");
2199 +}
2200 +
2201 +static void nvidia_force_enable_hpet(struct pci_dev *dev)
2202 +{
2203 + u32 uninitialized_var(val);
2204 +
2205 + if (!hpet_force_user || hpet_address || force_hpet_address)
2206 + return;
2207 +
2208 + pci_write_config_dword(dev, 0x44, 0xfed00001);
2209 + pci_read_config_dword(dev, 0x44, &val);
2210 + force_hpet_address = val & 0xfffffffe;
2211 + force_hpet_resume_type = NVIDIA_FORCE_HPET_RESUME;
2212 + printk(KERN_DEBUG "Force enabled HPET at base address 0x%lx\n",
2213 + force_hpet_address);
2214 + cached_dev = dev;
2215 + return;
2216 +}
2217 +
2218 +/* ISA Bridges */
2219 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0050,
2220 + nvidia_force_enable_hpet);
2221 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0051,
2222 + nvidia_force_enable_hpet);
2223 +
2224 +/* LPC bridges */
2225 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0360,
2226 + nvidia_force_enable_hpet);
2227 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0361,
2228 + nvidia_force_enable_hpet);
2229 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0362,
2230 + nvidia_force_enable_hpet);
2231 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0363,
2232 + nvidia_force_enable_hpet);
2233 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0364,
2234 + nvidia_force_enable_hpet);
2235 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0365,
2236 + nvidia_force_enable_hpet);
2237 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0366,
2238 + nvidia_force_enable_hpet);
2239 +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NVIDIA, 0x0367,
2240 + nvidia_force_enable_hpet);
2241 +
2242 +void force_hpet_resume(void)
2243 +{
2244 + switch (force_hpet_resume_type) {
2245 + case ICH_FORCE_HPET_RESUME:
2246 + return ich_force_hpet_resume();
2247 +
2248 + case OLD_ICH_FORCE_HPET_RESUME:
2249 + return old_ich_force_hpet_resume();
2250 +
2251 + case VT8237_FORCE_HPET_RESUME:
2252 + return vt8237_force_hpet_resume();
2253 +
2254 + case NVIDIA_FORCE_HPET_RESUME:
2255 + return nvidia_force_hpet_resume();
2256 +
2257 + default:
2258 + break;
2259 + }
2260 +}
2261 +
2262 #endif
2263 --- sle11-2009-10-16.orig/arch/x86/kernel/setup64-xen.c 2009-02-16 16:17:21.000000000 +0100
2264 +++ sle11-2009-10-16/arch/x86/kernel/setup64-xen.c 2009-02-16 16:18:36.000000000 +0100
2265 @@ -15,7 +15,6 @@
2266 #include <linux/bootmem.h>
2267 #include <linux/bitops.h>
2268 #include <linux/module.h>
2269 -#include <asm/bootsetup.h>
2270 #include <asm/pda.h>
2271 #include <asm/pgtable.h>
2272 #include <asm/processor.h>
2273 @@ -27,11 +26,12 @@
2274 #include <asm/percpu.h>
2275 #include <asm/proto.h>
2276 #include <asm/sections.h>
2277 +#include <asm/setup.h>
2278 #ifdef CONFIG_XEN
2279 #include <asm/hypervisor.h>
2280 #endif
2281
2282 -char x86_boot_params[BOOT_PARAM_SIZE] __initdata;
2283 +struct boot_params __initdata boot_params;
2284
2285 cpumask_t cpu_initialized __cpuinitdata = CPU_MASK_NONE;
2286
2287 @@ -159,8 +159,8 @@ static void switch_pt(void)
2288
2289 static void __cpuinit cpu_gdt_init(const struct desc_ptr *gdt_descr)
2290 {
2291 - asm volatile("lgdt %0" :: "m" (*gdt_descr));
2292 - asm volatile("lidt %0" :: "m" (idt_descr));
2293 + load_gdt(gdt_descr);
2294 + load_idt(idt_descr);
2295 }
2296 #endif
2297
2298 @@ -252,6 +252,14 @@ void __cpuinit check_efer(void)
2299
2300 unsigned long kernel_eflags;
2301
2302 +#ifndef CONFIG_X86_NO_TSS
2303 +/*
2304 + * Copies of the original ist values from the tss are only accessed during
2305 + * debugging, no special alignment required.
2306 + */
2307 +DEFINE_PER_CPU(struct orig_ist, orig_ist);
2308 +#endif
2309 +
2310 /*
2311 * cpu_init() initializes state that is per-CPU. Some data is already
2312 * initialized (naturally) in the bootstrap process, such as the GDT
2313 --- sle11-2009-10-16.orig/arch/x86/kernel/setup_32-xen.c 2009-02-16 16:17:21.000000000 +0100
2314 +++ sle11-2009-10-16/arch/x86/kernel/setup_32-xen.c 2009-02-16 16:18:36.000000000 +0100
2315 @@ -1,6 +1,4 @@
2316 /*
2317 - * linux/arch/i386/kernel/setup.c
2318 - *
2319 * Copyright (C) 1995 Linus Torvalds
2320 *
2321 * Support of BIGMEM added by Gerhard Wichert, Siemens AG, July 1999
2322 @@ -70,6 +68,7 @@
2323 #include <xen/xencons.h>
2324 #include <setup_arch.h>
2325 #include <bios_ebda.h>
2326 +#include <asm/cacheflush.h>
2327
2328 #ifdef CONFIG_XEN
2329 #include <xen/interface/kexec.h>
2330 @@ -80,13 +79,14 @@ static struct notifier_block xen_panic_b
2331 xen_panic_event, NULL, 0 /* try to go last */
2332 };
2333
2334 -int disable_pse __devinitdata = 0;
2335 +int disable_pse __cpuinitdata = 0;
2336
2337 /*
2338 * Machine setup..
2339 */
2340 extern struct resource code_resource;
2341 extern struct resource data_resource;
2342 +extern struct resource bss_resource;
2343
2344 /* cpu data as detected by the assembly code in head.S */
2345 struct cpuinfo_x86 new_cpu_data __cpuinitdata = { 0, 0, 0, 0, -1, 1, 0, 0, -1 };
2346 @@ -98,9 +98,6 @@ unsigned long mmu_cr4_features;
2347
2348 /* for MCA, but anyone else can use it if they want */
2349 unsigned int machine_id;
2350 -#ifdef CONFIG_MCA
2351 -EXPORT_SYMBOL(machine_id);
2352 -#endif
2353 unsigned int machine_submodel_id;
2354 unsigned int BIOS_revision;
2355 unsigned int mca_pentium_flag;
2356 @@ -121,7 +118,7 @@ EXPORT_SYMBOL(apm_info);
2357 struct edid_info edid_info;
2358 EXPORT_SYMBOL_GPL(edid_info);
2359 #ifndef CONFIG_XEN
2360 -#define copy_edid() (edid_info = EDID_INFO)
2361 +#define copy_edid() (edid_info = boot_params.edid_info)
2362 #endif
2363 struct ist_info ist_info;
2364 #if defined(CONFIG_X86_SPEEDSTEP_SMI) || \
2365 @@ -170,10 +167,11 @@ EXPORT_SYMBOL(edd);
2366 */
2367 static inline void copy_edd(void)
2368 {
2369 - memcpy(edd.mbr_signature, EDD_MBR_SIGNATURE, sizeof(edd.mbr_signature));
2370 - memcpy(edd.edd_info, EDD_BUF, sizeof(edd.edd_info));
2371 - edd.mbr_signature_nr = EDD_MBR_SIG_NR;
2372 - edd.edd_info_nr = EDD_NR;
2373 + memcpy(edd.mbr_signature, boot_params.edd_mbr_sig_buffer,
2374 + sizeof(edd.mbr_signature));
2375 + memcpy(edd.edd_info, boot_params.eddbuf, sizeof(edd.edd_info));
2376 + edd.mbr_signature_nr = boot_params.edd_mbr_sig_buf_entries;
2377 + edd.edd_info_nr = boot_params.eddbuf_entries;
2378 }
2379 #endif
2380 #else
2381 @@ -416,6 +414,53 @@ extern unsigned long __init setup_memory
2382 extern void zone_sizes_init(void);
2383 #endif /* !CONFIG_NEED_MULTIPLE_NODES */
2384
2385 +static inline unsigned long long get_total_mem(void)
2386 +{
2387 + unsigned long long total;
2388 +
2389 + total = max_low_pfn - min_low_pfn;
2390 +#ifdef CONFIG_HIGHMEM
2391 + total += highend_pfn - highstart_pfn;
2392 +#endif
2393 +
2394 + return total << PAGE_SHIFT;
2395 +}
2396 +
2397 +#ifdef CONFIG_KEXEC
2398 +#ifndef CONFIG_XEN
2399 +static void __init reserve_crashkernel(void)
2400 +{
2401 + unsigned long long total_mem;
2402 + unsigned long long crash_size, crash_base;
2403 + int ret;
2404 +
2405 + total_mem = get_total_mem();
2406 +
2407 + ret = parse_crashkernel(boot_command_line, total_mem,
2408 + &crash_size, &crash_base);
2409 + if (ret == 0 && crash_size > 0) {
2410 + if (crash_base > 0) {
2411 + printk(KERN_INFO "Reserving %ldMB of memory at %ldMB "
2412 + "for crashkernel (System RAM: %ldMB)\n",
2413 + (unsigned long)(crash_size >> 20),
2414 + (unsigned long)(crash_base >> 20),
2415 + (unsigned long)(total_mem >> 20));
2416 + crashk_res.start = crash_base;
2417 + crashk_res.end = crash_base + crash_size - 1;
2418 + reserve_bootmem(crash_base, crash_size);
2419 + } else
2420 + printk(KERN_INFO "crashkernel reservation failed - "
2421 + "you have to specify a base address\n");
2422 + }
2423 +}
2424 +#else
2425 +#define reserve_crashkernel xen_machine_kexec_setup_resources
2426 +#endif
2427 +#else
2428 +static inline void __init reserve_crashkernel(void)
2429 +{}
2430 +#endif
2431 +
2432 void __init setup_bootmem_allocator(void)
2433 {
2434 unsigned long bootmap_size;
2435 @@ -471,30 +516,25 @@ void __init setup_bootmem_allocator(void
2436
2437 #ifdef CONFIG_BLK_DEV_INITRD
2438 if (xen_start_info->mod_start) {
2439 - if (INITRD_START + INITRD_SIZE <= (max_low_pfn << PAGE_SHIFT)) {
2440 - /*reserve_bootmem(INITRD_START, INITRD_SIZE);*/
2441 - initrd_start = INITRD_START + PAGE_OFFSET;
2442 - initrd_end = initrd_start+INITRD_SIZE;
2443 + unsigned long ramdisk_image = __pa(xen_start_info->mod_start);
2444 + unsigned long ramdisk_size = xen_start_info->mod_len;
2445 + unsigned long ramdisk_end = ramdisk_image + ramdisk_size;
2446 + unsigned long end_of_lowmem = max_low_pfn << PAGE_SHIFT;
2447 +
2448 + if (ramdisk_end <= end_of_lowmem) {
2449 + /*reserve_bootmem(ramdisk_image, ramdisk_size);*/
2450 + initrd_start = ramdisk_image + PAGE_OFFSET;
2451 + initrd_end = initrd_start+ramdisk_size;
2452 initrd_below_start_ok = 1;
2453 - }
2454 - else {
2455 + } else {
2456 printk(KERN_ERR "initrd extends beyond end of memory "
2457 - "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2458 - INITRD_START + INITRD_SIZE,
2459 - max_low_pfn << PAGE_SHIFT);
2460 + "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2461 + ramdisk_end, end_of_lowmem);
2462 initrd_start = 0;
2463 }
2464 }
2465 #endif
2466 -#ifdef CONFIG_KEXEC
2467 -#ifdef CONFIG_XEN
2468 - xen_machine_kexec_setup_resources();
2469 -#else
2470 - if (crashk_res.start != crashk_res.end)
2471 - reserve_bootmem(crashk_res.start,
2472 - crashk_res.end - crashk_res.start + 1);
2473 -#endif
2474 -#endif
2475 + reserve_crashkernel();
2476 }
2477
2478 /*
2479 @@ -572,7 +612,8 @@ void __init setup_arch(char **cmdline_p)
2480 * the system table is valid. If not, then initialize normally.
2481 */
2482 #ifdef CONFIG_EFI
2483 - if ((LOADER_TYPE == 0x50) && EFI_SYSTAB)
2484 + if ((boot_params.hdr.type_of_loader == 0x50) &&
2485 + boot_params.efi_info.efi_systab)
2486 efi_enabled = 1;
2487 #endif
2488
2489 @@ -580,18 +621,18 @@ void __init setup_arch(char **cmdline_p)
2490 properly. Setting ROOT_DEV to default to /dev/ram0 breaks initrd.
2491 */
2492 ROOT_DEV = MKDEV(UNNAMED_MAJOR,0);
2493 - screen_info = SCREEN_INFO;
2494 + screen_info = boot_params.screen_info;
2495 copy_edid();
2496 - apm_info.bios = APM_BIOS_INFO;
2497 - ist_info = IST_INFO;
2498 - saved_videomode = VIDEO_MODE;
2499 - if( SYS_DESC_TABLE.length != 0 ) {
2500 - set_mca_bus(SYS_DESC_TABLE.table[3] & 0x2);
2501 - machine_id = SYS_DESC_TABLE.table[0];
2502 - machine_submodel_id = SYS_DESC_TABLE.table[1];
2503 - BIOS_revision = SYS_DESC_TABLE.table[2];
2504 + apm_info.bios = boot_params.apm_bios_info;
2505 + ist_info = boot_params.ist_info;
2506 + saved_videomode = boot_params.hdr.vid_mode;
2507 + if( boot_params.sys_desc_table.length != 0 ) {
2508 + set_mca_bus(boot_params.sys_desc_table.table[3] & 0x2);
2509 + machine_id = boot_params.sys_desc_table.table[0];
2510 + machine_submodel_id = boot_params.sys_desc_table.table[1];
2511 + BIOS_revision = boot_params.sys_desc_table.table[2];
2512 }
2513 - bootloader_type = LOADER_TYPE;
2514 + bootloader_type = boot_params.hdr.type_of_loader;
2515
2516 if (is_initial_xendomain()) {
2517 const struct dom0_vga_console_info *info =
2518 @@ -606,9 +647,9 @@ void __init setup_arch(char **cmdline_p)
2519 screen_info.orig_video_isVGA = 0;
2520
2521 #ifdef CONFIG_BLK_DEV_RAM
2522 - rd_image_start = RAMDISK_FLAGS & RAMDISK_IMAGE_START_MASK;
2523 - rd_prompt = ((RAMDISK_FLAGS & RAMDISK_PROMPT_FLAG) != 0);
2524 - rd_doload = ((RAMDISK_FLAGS & RAMDISK_LOAD_FLAG) != 0);
2525 + rd_image_start = boot_params.hdr.ram_size & RAMDISK_IMAGE_START_MASK;
2526 + rd_prompt = ((boot_params.hdr.ram_size & RAMDISK_PROMPT_FLAG) != 0);
2527 + rd_doload = ((boot_params.hdr.ram_size & RAMDISK_LOAD_FLAG) != 0);
2528 #endif
2529
2530 ARCH_SETUP
2531 @@ -621,7 +662,7 @@ void __init setup_arch(char **cmdline_p)
2532
2533 copy_edd();
2534
2535 - if (!MOUNT_ROOT_RDONLY)
2536 + if (!boot_params.hdr.root_flags)
2537 root_mountflags &= ~MS_RDONLY;
2538 init_mm.start_code = (unsigned long) _text;
2539 init_mm.end_code = (unsigned long) _etext;
2540 @@ -633,6 +674,8 @@ void __init setup_arch(char **cmdline_p)
2541 code_resource.end = virt_to_phys(_etext)-1;
2542 data_resource.start = virt_to_phys(_etext);
2543 data_resource.end = virt_to_phys(_edata)-1;
2544 + bss_resource.start = virt_to_phys(&__bss_start);
2545 + bss_resource.end = virt_to_phys(&__bss_stop)-1;
2546
2547 if ((i = MAX_GUEST_CMDLINE) > COMMAND_LINE_SIZE)
2548 i = COMMAND_LINE_SIZE;
2549 @@ -661,7 +704,7 @@ void __init setup_arch(char **cmdline_p)
2550 /*
2551 * NOTE: before this point _nobody_ is allowed to allocate
2552 * any memory using the bootmem allocator. Although the
2553 - * alloctor is now initialised only the first 8Mb of the kernel
2554 + * allocator is now initialised only the first 8Mb of the kernel
2555 * virtual address space has been mapped. All allocations before
2556 * paging_init() has completed must use the alloc_bootmem_low_pages()
2557 * variant (which allocates DMA'able memory) and care must be taken
2558 @@ -784,10 +827,8 @@ void __init setup_arch(char **cmdline_p)
2559 acpi_boot_table_init();
2560 #endif
2561
2562 -#ifdef CONFIG_PCI
2563 -#ifdef CONFIG_X86_IO_APIC
2564 - check_acpi_pci(); /* Checks more than just ACPI actually */
2565 -#endif
2566 +#if defined(CONFIG_PCI) && !defined(CONFIG_XEN)
2567 + early_quirks();
2568 #endif
2569
2570 #ifdef CONFIG_ACPI
2571 --- sle11-2009-10-16.orig/arch/x86/kernel/setup_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2572 +++ sle11-2009-10-16/arch/x86/kernel/setup_64-xen.c 2009-02-16 16:18:36.000000000 +0100
2573 @@ -1,10 +1,5 @@
2574 /*
2575 - * linux/arch/x86-64/kernel/setup.c
2576 - *
2577 * Copyright (C) 1995 Linus Torvalds
2578 - *
2579 - * Nov 2001 Dave Jones <davej@suse.de>
2580 - * Forked from i386 setup code.
2581 */
2582
2583 /*
2584 @@ -57,13 +52,13 @@
2585 #include <asm/dma.h>
2586 #include <asm/mpspec.h>
2587 #include <asm/mmu_context.h>
2588 -#include <asm/bootsetup.h>
2589 #include <asm/proto.h>
2590 #include <asm/setup.h>
2591 #include <asm/mach_apic.h>
2592 #include <asm/numa.h>
2593 #include <asm/sections.h>
2594 #include <asm/dmi.h>
2595 +#include <asm/cacheflush.h>
2596 #ifdef CONFIG_XEN
2597 #include <linux/percpu.h>
2598 #include <xen/interface/physdev.h>
2599 @@ -180,6 +175,12 @@ struct resource code_resource = {
2600 .end = 0,
2601 .flags = IORESOURCE_RAM,
2602 };
2603 +struct resource bss_resource = {
2604 + .name = "Kernel bss",
2605 + .start = 0,
2606 + .end = 0,
2607 + .flags = IORESOURCE_RAM,
2608 +};
2609
2610 #ifdef CONFIG_PROC_VMCORE
2611 /* elfcorehdr= specifies the location of elf core header
2612 @@ -231,10 +232,11 @@ EXPORT_SYMBOL(edd);
2613 */
2614 static inline void copy_edd(void)
2615 {
2616 - memcpy(edd.mbr_signature, EDD_MBR_SIGNATURE, sizeof(edd.mbr_signature));
2617 - memcpy(edd.edd_info, EDD_BUF, sizeof(edd.edd_info));
2618 - edd.mbr_signature_nr = EDD_MBR_SIG_NR;
2619 - edd.edd_info_nr = EDD_NR;
2620 + memcpy(edd.mbr_signature, boot_params.edd_mbr_sig_buffer,
2621 + sizeof(edd.mbr_signature));
2622 + memcpy(edd.edd_info, boot_params.eddbuf, sizeof(edd.edd_info));
2623 + edd.mbr_signature_nr = boot_params.edd_mbr_sig_buf_entries;
2624 + edd.edd_info_nr = boot_params.eddbuf_entries;
2625 }
2626 #endif
2627 #else
2628 @@ -243,6 +245,41 @@ static inline void copy_edd(void)
2629 }
2630 #endif
2631
2632 +#ifdef CONFIG_KEXEC
2633 +#ifndef CONFIG_XEN
2634 +static void __init reserve_crashkernel(void)
2635 +{
2636 + unsigned long long free_mem;
2637 + unsigned long long crash_size, crash_base;
2638 + int ret;
2639 +
2640 + free_mem = ((unsigned long long)max_low_pfn - min_low_pfn) << PAGE_SHIFT;
2641 +
2642 + ret = parse_crashkernel(boot_command_line, free_mem,
2643 + &crash_size, &crash_base);
2644 + if (ret == 0 && crash_size) {
2645 + if (crash_base > 0) {
2646 + printk(KERN_INFO "Reserving %ldMB of memory at %ldMB "
2647 + "for crashkernel (System RAM: %ldMB)\n",
2648 + (unsigned long)(crash_size >> 20),
2649 + (unsigned long)(crash_base >> 20),
2650 + (unsigned long)(free_mem >> 20));
2651 + crashk_res.start = crash_base;
2652 + crashk_res.end = crash_base + crash_size - 1;
2653 + reserve_bootmem(crash_base, crash_size);
2654 + } else
2655 + printk(KERN_INFO "crashkernel reservation failed - "
2656 + "you have to specify a base address\n");
2657 + }
2658 +}
2659 +#else
2660 +#define reserve_crashkernel xen_machine_kexec_setup_resources
2661 +#endif
2662 +#else
2663 +static inline void __init reserve_crashkernel(void)
2664 +{}
2665 +#endif
2666 +
2667 #ifndef CONFIG_XEN
2668 #define EBDA_ADDR_POINTER 0x40E
2669
2670 @@ -283,7 +320,7 @@ void __init setup_arch(char **cmdline_p)
2671 atomic_notifier_chain_register(&panic_notifier_list, &xen_panic_block);
2672
2673 ROOT_DEV = MKDEV(RAMDISK_MAJOR,0);
2674 - screen_info = SCREEN_INFO;
2675 + screen_info = boot_params.screen_info;
2676
2677 if (is_initial_xendomain()) {
2678 const struct dom0_vga_console_info *info =
2679 @@ -306,22 +343,22 @@ void __init setup_arch(char **cmdline_p)
2680 #else
2681 printk(KERN_INFO "Command line: %s\n", boot_command_line);
2682
2683 - ROOT_DEV = old_decode_dev(ORIG_ROOT_DEV);
2684 - screen_info = SCREEN_INFO;
2685 - edid_info = EDID_INFO;
2686 + ROOT_DEV = old_decode_dev(boot_params.hdr.root_dev);
2687 + screen_info = boot_params.screen_info;
2688 + edid_info = boot_params.edid_info;
2689 #endif /* !CONFIG_XEN */
2690 - saved_video_mode = SAVED_VIDEO_MODE;
2691 - bootloader_type = LOADER_TYPE;
2692 + saved_video_mode = boot_params.hdr.vid_mode;
2693 + bootloader_type = boot_params.hdr.type_of_loader;
2694
2695 #ifdef CONFIG_BLK_DEV_RAM
2696 - rd_image_start = RAMDISK_FLAGS & RAMDISK_IMAGE_START_MASK;
2697 - rd_prompt = ((RAMDISK_FLAGS & RAMDISK_PROMPT_FLAG) != 0);
2698 - rd_doload = ((RAMDISK_FLAGS & RAMDISK_LOAD_FLAG) != 0);
2699 + rd_image_start = boot_params.hdr.ram_size & RAMDISK_IMAGE_START_MASK;
2700 + rd_prompt = ((boot_params.hdr.ram_size & RAMDISK_PROMPT_FLAG) != 0);
2701 + rd_doload = ((boot_params.hdr.ram_size & RAMDISK_LOAD_FLAG) != 0);
2702 #endif
2703 setup_memory_region();
2704 copy_edd();
2705
2706 - if (!MOUNT_ROOT_RDONLY)
2707 + if (!boot_params.hdr.root_flags)
2708 root_mountflags &= ~MS_RDONLY;
2709 init_mm.start_code = (unsigned long) &_text;
2710 init_mm.end_code = (unsigned long) &_etext;
2711 @@ -332,6 +369,8 @@ void __init setup_arch(char **cmdline_p)
2712 code_resource.end = virt_to_phys(&_etext)-1;
2713 data_resource.start = virt_to_phys(&_etext);
2714 data_resource.end = virt_to_phys(&_edata)-1;
2715 + bss_resource.start = virt_to_phys(&__bss_start);
2716 + bss_resource.end = virt_to_phys(&__bss_stop)-1;
2717
2718 early_identify_cpu(&boot_cpu_data);
2719
2720 @@ -359,6 +398,11 @@ void __init setup_arch(char **cmdline_p)
2721 if (is_initial_xendomain())
2722 dmi_scan_machine();
2723
2724 +#if defined(CONFIG_SMP) && !defined(CONFIG_XEN)
2725 + /* setup to use the static apicid table during kernel startup */
2726 + x86_cpu_to_apicid_ptr = (void *)&x86_cpu_to_apicid_init;
2727 +#endif
2728 +
2729 /* How many end-of-memory variables you have, grandma! */
2730 max_low_pfn = end_pfn;
2731 max_pfn = end_pfn;
2732 @@ -423,52 +467,37 @@ void __init setup_arch(char **cmdline_p)
2733 */
2734 acpi_reserve_bootmem();
2735 #endif
2736 -#ifdef CONFIG_XEN
2737 #ifdef CONFIG_BLK_DEV_INITRD
2738 +#ifdef CONFIG_XEN
2739 if (xen_start_info->mod_start) {
2740 - if (INITRD_START + INITRD_SIZE <= (end_pfn << PAGE_SHIFT)) {
2741 - /*reserve_bootmem_generic(INITRD_START, INITRD_SIZE);*/
2742 - initrd_start = INITRD_START + PAGE_OFFSET;
2743 - initrd_end = initrd_start+INITRD_SIZE;
2744 + unsigned long ramdisk_image = __pa(xen_start_info->mod_start);
2745 + unsigned long ramdisk_size = xen_start_info->mod_len;
2746 +#else
2747 + if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) {
2748 + unsigned long ramdisk_image = boot_params.hdr.ramdisk_image;
2749 + unsigned long ramdisk_size = boot_params.hdr.ramdisk_size;
2750 +#endif
2751 + unsigned long ramdisk_end = ramdisk_image + ramdisk_size;
2752 + unsigned long end_of_mem = end_pfn << PAGE_SHIFT;
2753 +
2754 + if (ramdisk_end <= end_of_mem) {
2755 +#ifndef CONFIG_XEN
2756 + reserve_bootmem_generic(ramdisk_image, ramdisk_size);
2757 +#endif
2758 + initrd_start = ramdisk_image + PAGE_OFFSET;
2759 + initrd_end = initrd_start+ramdisk_size;
2760 +#ifdef CONFIG_XEN
2761 initrd_below_start_ok = 1;
2762 - } else {
2763 - printk(KERN_ERR "initrd extends beyond end of memory "
2764 - "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2765 - (unsigned long)(INITRD_START + INITRD_SIZE),
2766 - (unsigned long)(end_pfn << PAGE_SHIFT));
2767 - initrd_start = 0;
2768 - }
2769 - }
2770 #endif
2771 -#else /* CONFIG_XEN */
2772 -#ifdef CONFIG_BLK_DEV_INITRD
2773 - if (LOADER_TYPE && INITRD_START) {
2774 - if (INITRD_START + INITRD_SIZE <= (end_pfn << PAGE_SHIFT)) {
2775 - reserve_bootmem_generic(INITRD_START, INITRD_SIZE);
2776 - initrd_start = INITRD_START + PAGE_OFFSET;
2777 - initrd_end = initrd_start+INITRD_SIZE;
2778 - }
2779 - else {
2780 + } else {
2781 printk(KERN_ERR "initrd extends beyond end of memory "
2782 - "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2783 - (unsigned long)(INITRD_START + INITRD_SIZE),
2784 - (unsigned long)(end_pfn << PAGE_SHIFT));
2785 + "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
2786 + ramdisk_end, end_of_mem);
2787 initrd_start = 0;
2788 }
2789 }
2790 #endif
2791 -#endif /* !CONFIG_XEN */
2792 -#ifdef CONFIG_KEXEC
2793 -#ifdef CONFIG_XEN
2794 - xen_machine_kexec_setup_resources();
2795 -#else
2796 - if (crashk_res.start != crashk_res.end) {
2797 - reserve_bootmem_generic(crashk_res.start,
2798 - crashk_res.end - crashk_res.start + 1);
2799 - }
2800 -#endif
2801 -#endif
2802 -
2803 + reserve_crashkernel();
2804 paging_init();
2805 #ifdef CONFIG_X86_LOCAL_APIC
2806 /*
2807 @@ -783,7 +812,7 @@ static void __init amd_detect_cmp(struct
2808 but in the same order as the HT nodeids.
2809 If that doesn't result in a usable node fall back to the
2810 path for the previous case. */
2811 - int ht_nodeid = apicid - (cpu_data[0].phys_proc_id << bits);
2812 + int ht_nodeid = apicid - (cpu_data(0).phys_proc_id << bits);
2813 if (ht_nodeid >= 0 &&
2814 apicid_to_node[ht_nodeid] != NUMA_NO_NODE)
2815 node = apicid_to_node[ht_nodeid];
2816 @@ -798,6 +827,39 @@ static void __init amd_detect_cmp(struct
2817 #endif
2818 }
2819
2820 +#define ENABLE_C1E_MASK 0x18000000
2821 +#define CPUID_PROCESSOR_SIGNATURE 1
2822 +#define CPUID_XFAM 0x0ff00000
2823 +#define CPUID_XFAM_K8 0x00000000
2824 +#define CPUID_XFAM_10H 0x00100000
2825 +#define CPUID_XFAM_11H 0x00200000
2826 +#define CPUID_XMOD 0x000f0000
2827 +#define CPUID_XMOD_REV_F 0x00040000
2828 +
2829 +#ifndef CONFIG_XEN
2830 +/* AMD systems with C1E don't have a working lAPIC timer. Check for that. */
2831 +static __cpuinit int amd_apic_timer_broken(void)
2832 +{
2833 + u32 lo, hi;
2834 + u32 eax = cpuid_eax(CPUID_PROCESSOR_SIGNATURE);
2835 + switch (eax & CPUID_XFAM) {
2836 + case CPUID_XFAM_K8:
2837 + if ((eax & CPUID_XMOD) < CPUID_XMOD_REV_F)
2838 + break;
2839 + case CPUID_XFAM_10H:
2840 + case CPUID_XFAM_11H:
2841 + rdmsr(MSR_K8_ENABLE_C1E, lo, hi);
2842 + if (lo & ENABLE_C1E_MASK)
2843 + return 1;
2844 + break;
2845 + default:
2846 + /* err on the side of caution */
2847 + return 1;
2848 + }
2849 + return 0;
2850 +}
2851 +#endif
2852 +
2853 static void __cpuinit init_amd(struct cpuinfo_x86 *c)
2854 {
2855 unsigned level;
2856 @@ -827,7 +889,7 @@ static void __cpuinit init_amd(struct cp
2857 level = cpuid_eax(1);
2858 if (c->x86 == 15 && ((level >= 0x0f48 && level < 0x0f50) || level >= 0x0f58))
2859 set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability);
2860 - if (c->x86 == 0x10)
2861 + if (c->x86 == 0x10 || c->x86 == 0x11)
2862 set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability);
2863
2864 /* Enable workaround for FXSAVE leak */
2865 @@ -869,6 +931,11 @@ static void __cpuinit init_amd(struct cp
2866 /* Family 10 doesn't support C states in MWAIT so don't use it */
2867 if (c->x86 == 0x10 && !force_mwait)
2868 clear_bit(X86_FEATURE_MWAIT, &c->x86_capability);
2869 +
2870 +#ifndef CONFIG_XEN
2871 + if (amd_apic_timer_broken())
2872 + disable_apic_timer = 1;
2873 +#endif
2874 }
2875
2876 static void __cpuinit detect_ht(struct cpuinfo_x86 *c)
2877 @@ -1179,6 +1246,7 @@ void __cpuinit print_cpu_info(struct cpu
2878 static int show_cpuinfo(struct seq_file *m, void *v)
2879 {
2880 struct cpuinfo_x86 *c = v;
2881 + int cpu = 0;
2882
2883 /*
2884 * These flag bits must match the definitions in <asm/cpufeature.h>.
2885 @@ -1188,7 +1256,7 @@ static int show_cpuinfo(struct seq_file
2886 * applications want to get the raw CPUID data, they should access
2887 * /dev/cpu/<cpu_nr>/cpuid instead.
2888 */
2889 - static char *x86_cap_flags[] = {
2890 + static const char *const x86_cap_flags[] = {
2891 /* Intel-defined */
2892 "fpu", "vme", "de", "pse", "tsc", "msr", "pae", "mce",
2893 "cx8", "apic", NULL, "sep", "mtrr", "pge", "mca", "cmov",
2894 @@ -1219,7 +1287,7 @@ static int show_cpuinfo(struct seq_file
2895 /* Intel-defined (#2) */
2896 "pni", NULL, NULL, "monitor", "ds_cpl", "vmx", "smx", "est",
2897 "tm2", "ssse3", "cid", NULL, NULL, "cx16", "xtpr", NULL,
2898 - NULL, NULL, "dca", NULL, NULL, NULL, NULL, "popcnt",
2899 + NULL, NULL, "dca", "sse4_1", "sse4_2", NULL, NULL, "popcnt",
2900 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2901
2902 /* VIA/Cyrix/Centaur-defined */
2903 @@ -1229,10 +1297,10 @@ static int show_cpuinfo(struct seq_file
2904 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2905
2906 /* AMD-defined (#2) */
2907 - "lahf_lm", "cmp_legacy", "svm", "extapic", "cr8_legacy",
2908 - "altmovcr8", "abm", "sse4a",
2909 - "misalignsse", "3dnowprefetch",
2910 - "osvw", "ibs", NULL, NULL, NULL, NULL,
2911 + "lahf_lm", "cmp_legacy", "svm", "extapic",
2912 + "cr8_legacy", "abm", "sse4a", "misalignsse",
2913 + "3dnowprefetch", "osvw", "ibs", "sse5",
2914 + "skinit", "wdt", NULL, NULL,
2915 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2916 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2917
2918 @@ -1242,7 +1310,7 @@ static int show_cpuinfo(struct seq_file
2919 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2920 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2921 };
2922 - static char *x86_power_flags[] = {
2923 + static const char *const x86_power_flags[] = {
2924 "ts", /* temperature sensor */
2925 "fid", /* frequency id control */
2926 "vid", /* voltage id control */
2927 @@ -1257,8 +1325,7 @@ static int show_cpuinfo(struct seq_file
2928
2929
2930 #ifdef CONFIG_SMP
2931 - if (!cpu_online(c-cpu_data))
2932 - return 0;
2933 + cpu = c->cpu_index;
2934 #endif
2935
2936 seq_printf(m,"processor\t: %u\n"
2937 @@ -1266,7 +1333,7 @@ static int show_cpuinfo(struct seq_file
2938 "cpu family\t: %d\n"
2939 "model\t\t: %d\n"
2940 "model name\t: %s\n",
2941 - (unsigned)(c-cpu_data),
2942 + (unsigned)cpu,
2943 c->x86_vendor_id[0] ? c->x86_vendor_id : "unknown",
2944 c->x86,
2945 (int)c->x86_model,
2946 @@ -1278,7 +1345,7 @@ static int show_cpuinfo(struct seq_file
2947 seq_printf(m, "stepping\t: unknown\n");
2948
2949 if (cpu_has(c,X86_FEATURE_TSC)) {
2950 - unsigned int freq = cpufreq_quick_get((unsigned)(c-cpu_data));
2951 + unsigned int freq = cpufreq_quick_get((unsigned)cpu);
2952 if (!freq)
2953 freq = cpu_khz;
2954 seq_printf(m, "cpu MHz\t\t: %u.%03u\n",
2955 @@ -1291,9 +1358,9 @@ static int show_cpuinfo(struct seq_file
2956
2957 #ifdef CONFIG_SMP
2958 if (smp_num_siblings * c->x86_max_cores > 1) {
2959 - int cpu = c - cpu_data;
2960 seq_printf(m, "physical id\t: %d\n", c->phys_proc_id);
2961 - seq_printf(m, "siblings\t: %d\n", cpus_weight(cpu_core_map[cpu]));
2962 + seq_printf(m, "siblings\t: %d\n",
2963 + cpus_weight(per_cpu(cpu_core_map, cpu)));
2964 seq_printf(m, "core id\t\t: %d\n", c->cpu_core_id);
2965 seq_printf(m, "cpu cores\t: %d\n", c->booted_cores);
2966 }
2967 @@ -1348,12 +1415,16 @@ static int show_cpuinfo(struct seq_file
2968
2969 static void *c_start(struct seq_file *m, loff_t *pos)
2970 {
2971 - return *pos < NR_CPUS ? cpu_data + *pos : NULL;
2972 + if (*pos == 0) /* just in case, cpu 0 is not the first */
2973 + *pos = first_cpu(cpu_online_map);
2974 + if ((*pos) < NR_CPUS && cpu_online(*pos))
2975 + return &cpu_data(*pos);
2976 + return NULL;
2977 }
2978
2979 static void *c_next(struct seq_file *m, void *v, loff_t *pos)
2980 {
2981 - ++*pos;
2982 + *pos = next_cpu(*pos, cpu_online_map);
2983 return c_start(m, pos);
2984 }
2985
2986 --- sle11-2009-10-16.orig/arch/x86/kernel/smp_32-xen.c 2009-02-16 16:17:21.000000000 +0100
2987 +++ sle11-2009-10-16/arch/x86/kernel/smp_32-xen.c 2009-02-16 16:18:36.000000000 +0100
2988 @@ -72,7 +72,7 @@
2989 *
2990 * B stepping CPUs may hang. There are hardware work arounds
2991 * for this. We warn about it in case your board doesn't have the work
2992 - * arounds. Basically thats so I can tell anyone with a B stepping
2993 + * arounds. Basically that's so I can tell anyone with a B stepping
2994 * CPU and SMP problems "tough".
2995 *
2996 * Specific items [From Pentium Processor Specification Update]
2997 @@ -241,7 +241,7 @@ void leave_mm(unsigned long cpu)
2998 * 1a1) cpu_clear(cpu, old_mm->cpu_vm_mask);
2999 * Stop ipi delivery for the old mm. This is not synchronized with
3000 * the other cpus, but smp_invalidate_interrupt ignore flush ipis
3001 - * for the wrong mm, and in the worst case we perform a superflous
3002 + * for the wrong mm, and in the worst case we perform a superfluous
3003 * tlb flush.
3004 * 1a2) set cpu_tlbstate to TLBSTATE_OK
3005 * Now the smp_invalidate_interrupt won't call leave_mm if cpu0
3006 @@ -309,6 +309,7 @@ irqreturn_t smp_invalidate_interrupt(int
3007 smp_mb__after_clear_bit();
3008 out:
3009 put_cpu_no_resched();
3010 + __get_cpu_var(irq_stat).irq_tlb_count++;
3011
3012 return IRQ_HANDLED;
3013 }
3014 @@ -580,7 +581,7 @@ static void stop_this_cpu (void * dummy)
3015 */
3016 cpu_clear(smp_processor_id(), cpu_online_map);
3017 disable_all_local_evtchn();
3018 - if (cpu_data[smp_processor_id()].hlt_works_ok)
3019 + if (cpu_data(smp_processor_id()).hlt_works_ok)
3020 for(;;) halt();
3021 for (;;);
3022 }
3023 @@ -610,6 +611,7 @@ void xen_smp_send_stop(void)
3024 */
3025 irqreturn_t smp_reschedule_interrupt(int irq, void *dev_id)
3026 {
3027 + __get_cpu_var(irq_stat).irq_resched_count++;
3028
3029 return IRQ_HANDLED;
3030 }
3031 @@ -632,6 +634,7 @@ irqreturn_t smp_call_function_interrupt(
3032 */
3033 irq_enter();
3034 (*func)(info);
3035 + __get_cpu_var(irq_stat).irq_call_count++;
3036 irq_exit();
3037
3038 if (wait) {
3039 --- sle11-2009-10-16.orig/arch/x86/kernel/smp_64-xen.c 2009-02-16 16:17:21.000000000 +0100
3040 +++ sle11-2009-10-16/arch/x86/kernel/smp_64-xen.c 2009-02-16 16:18:36.000000000 +0100
3041 @@ -167,6 +167,7 @@ asmlinkage void smp_invalidate_interrupt
3042 out:
3043 ack_APIC_irq();
3044 cpu_clear(cpu, f->flush_cpumask);
3045 + add_pda(irq_tlb_count, 1);
3046 }
3047
3048 static void flush_tlb_others(cpumask_t cpumask, struct mm_struct *mm,
3049 @@ -326,17 +327,27 @@ void unlock_ipi_call_lock(void)
3050 }
3051
3052 /*
3053 - * this function sends a 'generic call function' IPI to one other CPU
3054 - * in the system.
3055 - *
3056 - * cpu is a standard Linux logical CPU number.
3057 + * this function sends a 'generic call function' IPI to all other CPU
3058 + * of the system defined in the mask.
3059 */
3060 -static void
3061 -__smp_call_function_single(int cpu, void (*func) (void *info), void *info,
3062 - int nonatomic, int wait)
3063 +
3064 +static int
3065 +__smp_call_function_mask(cpumask_t mask,
3066 + void (*func)(void *), void *info,
3067 + int wait)
3068 {
3069 struct call_data_struct data;
3070 - int cpus = 1;
3071 + cpumask_t allbutself;
3072 + int cpus;
3073 +
3074 + allbutself = cpu_online_map;
3075 + cpu_clear(smp_processor_id(), allbutself);
3076 +
3077 + cpus_and(mask, mask, allbutself);
3078 + cpus = cpus_weight(mask);
3079 +
3080 + if (!cpus)
3081 + return 0;
3082
3083 data.func = func;
3084 data.info = info;
3085 @@ -347,19 +358,55 @@ __smp_call_function_single(int cpu, void
3086
3087 call_data = &data;
3088 wmb();
3089 - /* Send a message to all other CPUs and wait for them to respond */
3090 - send_IPI_mask(cpumask_of_cpu(cpu), CALL_FUNCTION_VECTOR);
3091 +
3092 + /* Send a message to other CPUs */
3093 + if (cpus_equal(mask, allbutself))
3094 + send_IPI_allbutself(CALL_FUNCTION_VECTOR);
3095 + else
3096 + send_IPI_mask(mask, CALL_FUNCTION_VECTOR);
3097
3098 /* Wait for response */
3099 while (atomic_read(&data.started) != cpus)
3100 cpu_relax();
3101
3102 if (!wait)
3103 - return;
3104 + return 0;
3105
3106 while (atomic_read(&data.finished) != cpus)
3107 cpu_relax();
3108 +
3109 + return 0;
3110 +}
3111 +/**
3112 + * smp_call_function_mask(): Run a function on a set of other CPUs.
3113 + * @mask: The set of cpus to run on. Must not include the current cpu.
3114 + * @func: The function to run. This must be fast and non-blocking.
3115 + * @info: An arbitrary pointer to pass to the function.
3116 + * @wait: If true, wait (atomically) until function has completed on other CPUs.
3117 + *
3118 + * Returns 0 on success, else a negative status code.
3119 + *
3120 + * If @wait is true, then returns once @func has returned; otherwise
3121 + * it returns just before the target cpu calls @func.
3122 + *
3123 + * You must not call this function with disabled interrupts or from a
3124 + * hardware interrupt handler or from a bottom half handler.
3125 + */
3126 +int smp_call_function_mask(cpumask_t mask,
3127 + void (*func)(void *), void *info,
3128 + int wait)
3129 +{
3130 + int ret;
3131 +
3132 + /* Can deadlock when called with interrupts disabled */
3133 + WARN_ON(irqs_disabled());
3134 +
3135 + spin_lock(&call_lock);
3136 + ret = __smp_call_function_mask(mask, func, info, wait);
3137 + spin_unlock(&call_lock);
3138 + return ret;
3139 }
3140 +EXPORT_SYMBOL(smp_call_function_mask);
3141
3142 /*
3143 * smp_call_function_single - Run a function on a specific CPU
3144 @@ -378,6 +425,7 @@ int smp_call_function_single (int cpu, v
3145 int nonatomic, int wait)
3146 {
3147 /* prevent preemption and reschedule on another processor */
3148 + int ret;
3149 int me = get_cpu();
3150
3151 /* Can deadlock when called with interrupts disabled */
3152 @@ -391,51 +439,14 @@ int smp_call_function_single (int cpu, v
3153 return 0;
3154 }
3155
3156 - spin_lock(&call_lock);
3157 - __smp_call_function_single(cpu, func, info, nonatomic, wait);
3158 - spin_unlock(&call_lock);
3159 + ret = smp_call_function_mask(cpumask_of_cpu(cpu), func, info, wait);
3160 +
3161 put_cpu();
3162 - return 0;
3163 + return ret;
3164 }
3165 EXPORT_SYMBOL(smp_call_function_single);
3166
3167 /*
3168 - * this function sends a 'generic call function' IPI to all other CPUs
3169 - * in the system.
3170 - */
3171 -static void __smp_call_function (void (*func) (void *info), void *info,
3172 - int nonatomic, int wait)
3173 -{
3174 - struct call_data_struct data;
3175 - int cpus = num_online_cpus()-1;
3176 -
3177 - if (!cpus)
3178 - return;
3179 -
3180 - data.func = func;
3181 - data.info = info;
3182 - atomic_set(&data.started, 0);
3183 - data.wait = wait;
3184 - if (wait)
3185 - atomic_set(&data.finished, 0);
3186 -
3187 - call_data = &data;
3188 - wmb();
3189 - /* Send a message to all other CPUs and wait for them to respond */
3190 - send_IPI_allbutself(CALL_FUNCTION_VECTOR);
3191 -
3192 - /* Wait for response */
3193 - while (atomic_read(&data.started) != cpus)
3194 - cpu_relax();
3195 -
3196 - if (!wait)
3197 - return;
3198 -
3199 - while (atomic_read(&data.finished) != cpus)
3200 - cpu_relax();
3201 -}
3202 -
3203 -/*
3204 * smp_call_function - run a function on all other CPUs.
3205 * @func: The function to run. This must be fast and non-blocking.
3206 * @info: An arbitrary pointer to pass to the function.
3207 @@ -453,10 +464,7 @@ static void __smp_call_function (void (*
3208 int smp_call_function (void (*func) (void *info), void *info, int nonatomic,
3209 int wait)
3210 {
3211 - spin_lock(&call_lock);
3212 - __smp_call_function(func,info,nonatomic,wait);
3213 - spin_unlock(&call_lock);
3214 - return 0;
3215 + return smp_call_function_mask(cpu_online_map, func, info, wait);
3216 }
3217 EXPORT_SYMBOL(smp_call_function);
3218
3219 @@ -485,7 +493,7 @@ void smp_send_stop(void)
3220 /* Don't deadlock on the call lock in panic */
3221 nolock = !spin_trylock(&call_lock);
3222 local_irq_save(flags);
3223 - __smp_call_function(stop_this_cpu, NULL, 0, 0);
3224 + __smp_call_function_mask(cpu_online_map, stop_this_cpu, NULL, 0);
3225 if (!nolock)
3226 spin_unlock(&call_lock);
3227 disable_all_local_evtchn();
3228 @@ -505,7 +513,9 @@ asmlinkage irqreturn_t smp_reschedule_in
3229 {
3230 #ifndef CONFIG_XEN
3231 ack_APIC_irq();
3232 -#else
3233 +#endif
3234 + add_pda(irq_resched_count, 1);
3235 +#ifdef CONFIG_XEN
3236 return IRQ_HANDLED;
3237 #endif
3238 }
3239 @@ -535,6 +545,7 @@ asmlinkage irqreturn_t smp_call_function
3240 exit_idle();
3241 irq_enter();
3242 (*func)(info);
3243 + add_pda(irq_call_count, 1);
3244 irq_exit();
3245 if (wait) {
3246 mb();
3247 --- sle11-2009-10-16.orig/arch/x86/kernel/time_32-xen.c 2009-10-28 14:57:53.000000000 +0100
3248 +++ sle11-2009-10-16/arch/x86/kernel/time_32-xen.c 2009-10-28 14:57:59.000000000 +0100
3249 @@ -1,6 +1,4 @@
3250 /*
3251 - * linux/arch/i386/kernel/time.c
3252 - *
3253 * Copyright (C) 1991, 1992, 1995 Linus Torvalds
3254 *
3255 * This file contains the PC-specific time handling details:
3256 @@ -74,6 +72,7 @@
3257 #include <asm/arch_hooks.h>
3258
3259 #include <xen/evtchn.h>
3260 +#include <xen/sysctl.h>
3261 #include <xen/interface/vcpu.h>
3262
3263 #include <asm/i8253.h>
3264 @@ -544,6 +543,13 @@ irqreturn_t timer_interrupt(int irq, voi
3265 struct shadow_time_info *shadow = &per_cpu(shadow_time, cpu);
3266 struct vcpu_runstate_info runstate;
3267
3268 + /* Keep nmi watchdog up to date */
3269 +#ifdef __i386__
3270 + per_cpu(irq_stat, smp_processor_id()).irq0_irqs++;
3271 +#else
3272 + add_pda(irq0_irqs, 1);
3273 +#endif
3274 +
3275 /*
3276 * Here we are in the timer irq handler. We just have irqs locally
3277 * disabled but we don't know if the timer_bh is running on the other
3278 @@ -995,7 +1001,7 @@ static int time_cpufreq_notifier(struct
3279 struct cpufreq_freqs *freq = data;
3280 struct xen_platform_op op;
3281
3282 - if (cpu_has(&cpu_data[freq->cpu], X86_FEATURE_CONSTANT_TSC))
3283 + if (cpu_has(&cpu_data(freq->cpu), X86_FEATURE_CONSTANT_TSC))
3284 return 0;
3285
3286 if (val == CPUFREQ_PRECHANGE)
3287 @@ -1033,30 +1039,33 @@ core_initcall(cpufreq_time_setup);
3288 */
3289 static ctl_table xen_subtable[] = {
3290 {
3291 - .ctl_name = 1,
3292 + .ctl_name = CTL_XEN_INDEPENDENT_WALLCLOCK,
3293 .procname = "independent_wallclock",
3294 .data = &independent_wallclock,
3295 .maxlen = sizeof(independent_wallclock),
3296 .mode = 0644,
3297 + .strategy = sysctl_data,
3298 .proc_handler = proc_dointvec
3299 },
3300 {
3301 - .ctl_name = 2,
3302 + .ctl_name = CTL_XEN_PERMITTED_CLOCK_JITTER,
3303 .procname = "permitted_clock_jitter",
3304 .data = &permitted_clock_jitter,
3305 .maxlen = sizeof(permitted_clock_jitter),
3306 .mode = 0644,
3307 + .strategy = sysctl_data,
3308 .proc_handler = proc_doulongvec_minmax
3309 },
3310 - { 0 }
3311 + { }
3312 };
3313 static ctl_table xen_table[] = {
3314 {
3315 - .ctl_name = 123,
3316 + .ctl_name = CTL_XEN,
3317 .procname = "xen",
3318 .mode = 0555,
3319 - .child = xen_subtable},
3320 - { 0 }
3321 + .child = xen_subtable
3322 + },
3323 + { }
3324 };
3325 static int __init xen_sysctl_init(void)
3326 {
3327 --- sle11-2009-10-16.orig/arch/x86/kernel/traps_32-xen.c 2009-02-16 16:17:21.000000000 +0100
3328 +++ sle11-2009-10-16/arch/x86/kernel/traps_32-xen.c 2009-02-16 16:18:36.000000000 +0100
3329 @@ -1,6 +1,4 @@
3330 /*
3331 - * linux/arch/i386/traps.c
3332 - *
3333 * Copyright (C) 1991, 1992 Linus Torvalds
3334 *
3335 * Pentium III FXSR, SSE support
3336 @@ -65,6 +63,11 @@
3337
3338 int panic_on_unrecovered_nmi;
3339
3340 +#ifndef CONFIG_XEN
3341 +DECLARE_BITMAP(used_vectors, NR_VECTORS);
3342 +EXPORT_SYMBOL_GPL(used_vectors);
3343 +#endif
3344 +
3345 asmlinkage int system_call(void);
3346
3347 /* Do we ignore FPU interrupts ? */
3348 @@ -120,7 +123,7 @@ struct stack_frame {
3349
3350 static inline unsigned long print_context_stack(struct thread_info *tinfo,
3351 unsigned long *stack, unsigned long ebp,
3352 - struct stacktrace_ops *ops, void *data)
3353 + const struct stacktrace_ops *ops, void *data)
3354 {
3355 #ifdef CONFIG_FRAME_POINTER
3356 struct stack_frame *frame = (struct stack_frame *)ebp;
3357 @@ -157,7 +160,7 @@ static inline unsigned long print_contex
3358
3359 void dump_trace(struct task_struct *task, struct pt_regs *regs,
3360 unsigned long *stack,
3361 - struct stacktrace_ops *ops, void *data)
3362 + const struct stacktrace_ops *ops, void *data)
3363 {
3364 unsigned long ebp = 0;
3365
3366 @@ -229,7 +232,7 @@ static void print_trace_address(void *da
3367 touch_nmi_watchdog();
3368 }
3369
3370 -static struct stacktrace_ops print_trace_ops = {
3371 +static const struct stacktrace_ops print_trace_ops = {
3372 .warning = print_trace_warning,
3373 .warning_symbol = print_trace_warning_symbol,
3374 .stack = print_trace_stack,
3375 @@ -288,6 +291,11 @@ void dump_stack(void)
3376 {
3377 unsigned long stack;
3378
3379 + printk("Pid: %d, comm: %.20s %s %s %.*s\n",
3380 + current->pid, current->comm, print_tainted(),
3381 + init_utsname()->release,
3382 + (int)strcspn(init_utsname()->version, " "),
3383 + init_utsname()->version);
3384 show_trace(current, NULL, &stack);
3385 }
3386
3387 @@ -296,48 +304,24 @@ EXPORT_SYMBOL(dump_stack);
3388 void show_registers(struct pt_regs *regs)
3389 {
3390 int i;
3391 - int in_kernel = 1;
3392 - unsigned long esp;
3393 - unsigned short ss, gs;
3394 -
3395 - esp = (unsigned long) (&regs->esp);
3396 - savesegment(ss, ss);
3397 - savesegment(gs, gs);
3398 - if (user_mode_vm(regs)) {
3399 - in_kernel = 0;
3400 - esp = regs->esp;
3401 - ss = regs->xss & 0xffff;
3402 - }
3403 +
3404 print_modules();
3405 - printk(KERN_EMERG "CPU: %d\n"
3406 - KERN_EMERG "EIP: %04x:[<%08lx>] %s VLI\n"
3407 - KERN_EMERG "EFLAGS: %08lx (%s %.*s)\n",
3408 - smp_processor_id(), 0xffff & regs->xcs, regs->eip,
3409 - print_tainted(), regs->eflags, init_utsname()->release,
3410 - (int)strcspn(init_utsname()->version, " "),
3411 - init_utsname()->version);
3412 - print_symbol(KERN_EMERG "EIP is at %s\n", regs->eip);
3413 - printk(KERN_EMERG "eax: %08lx ebx: %08lx ecx: %08lx edx: %08lx\n",
3414 - regs->eax, regs->ebx, regs->ecx, regs->edx);
3415 - printk(KERN_EMERG "esi: %08lx edi: %08lx ebp: %08lx esp: %08lx\n",
3416 - regs->esi, regs->edi, regs->ebp, esp);
3417 - printk(KERN_EMERG "ds: %04x es: %04x fs: %04x gs: %04x ss: %04x\n",
3418 - regs->xds & 0xffff, regs->xes & 0xffff, regs->xfs & 0xffff, gs, ss);
3419 + __show_registers(regs, 0);
3420 printk(KERN_EMERG "Process %.*s (pid: %d, ti=%p task=%p task.ti=%p)",
3421 - TASK_COMM_LEN, current->comm, current->pid,
3422 + TASK_COMM_LEN, current->comm, task_pid_nr(current),
3423 current_thread_info(), current, task_thread_info(current));
3424 /*
3425 * When in-kernel, we also print out the stack and code at the
3426 * time of the fault..
3427 */
3428 - if (in_kernel) {
3429 + if (!user_mode_vm(regs)) {
3430 u8 *eip;
3431 unsigned int code_prologue = code_bytes * 43 / 64;
3432 unsigned int code_len = code_bytes;
3433 unsigned char c;
3434
3435 printk("\n" KERN_EMERG "Stack: ");
3436 - show_stack_log_lvl(NULL, regs, (unsigned long *)esp, KERN_EMERG);
3437 + show_stack_log_lvl(NULL, regs, &regs->esp, KERN_EMERG);
3438
3439 printk(KERN_EMERG "Code: ");
3440
3441 @@ -382,11 +366,11 @@ int is_valid_bugaddr(unsigned long eip)
3442 void die(const char * str, struct pt_regs * regs, long err)
3443 {
3444 static struct {
3445 - spinlock_t lock;
3446 + raw_spinlock_t lock;
3447 u32 lock_owner;
3448 int lock_owner_depth;
3449 } die = {
3450 - .lock = __SPIN_LOCK_UNLOCKED(die.lock),
3451 + .lock = __RAW_SPIN_LOCK_UNLOCKED,
3452 .lock_owner = -1,
3453 .lock_owner_depth = 0
3454 };
3455 @@ -397,40 +381,33 @@ void die(const char * str, struct pt_reg
3456
3457 if (die.lock_owner != raw_smp_processor_id()) {
3458 console_verbose();
3459 - spin_lock_irqsave(&die.lock, flags);
3460 + raw_local_irq_save(flags);
3461 + __raw_spin_lock(&die.lock);
3462 die.lock_owner = smp_processor_id();
3463 die.lock_owner_depth = 0;
3464 bust_spinlocks(1);
3465 - }
3466 - else
3467 - local_save_flags(flags);
3468 + } else
3469 + raw_local_irq_save(flags);
3470
3471 if (++die.lock_owner_depth < 3) {
3472 - int nl = 0;
3473 unsigned long esp;
3474 unsigned short ss;
3475
3476 report_bug(regs->eip, regs);
3477
3478 - printk(KERN_EMERG "%s: %04lx [#%d]\n", str, err & 0xffff, ++die_counter);
3479 + printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff,
3480 + ++die_counter);
3481 #ifdef CONFIG_PREEMPT
3482 - printk(KERN_EMERG "PREEMPT ");
3483 - nl = 1;
3484 + printk("PREEMPT ");
3485 #endif
3486 #ifdef CONFIG_SMP
3487 - if (!nl)
3488 - printk(KERN_EMERG);
3489 printk("SMP ");
3490 - nl = 1;
3491 #endif
3492 #ifdef CONFIG_DEBUG_PAGEALLOC
3493 - if (!nl)
3494 - printk(KERN_EMERG);
3495 printk("DEBUG_PAGEALLOC");
3496 - nl = 1;
3497 #endif
3498 - if (nl)
3499 - printk("\n");
3500 + printk("\n");
3501 +
3502 if (notify_die(DIE_OOPS, str, regs, err,
3503 current->thread.trap_no, SIGSEGV) !=
3504 NOTIFY_STOP) {
3505 @@ -454,7 +431,8 @@ void die(const char * str, struct pt_reg
3506 bust_spinlocks(0);
3507 die.lock_owner = -1;
3508 add_taint(TAINT_DIE);
3509 - spin_unlock_irqrestore(&die.lock, flags);
3510 + __raw_spin_unlock(&die.lock);
3511 + raw_local_irq_restore(flags);
3512
3513 if (!regs)
3514 return;
3515 @@ -571,6 +549,7 @@ fastcall void do_##name(struct pt_regs *
3516 info.si_errno = 0; \
3517 info.si_code = sicode; \
3518 info.si_addr = (void __user *)siaddr; \
3519 + trace_hardirqs_fixup(); \
3520 if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr) \
3521 == NOTIFY_STOP) \
3522 return; \
3523 @@ -606,7 +585,7 @@ fastcall void __kprobes do_general_prote
3524 printk_ratelimit())
3525 printk(KERN_INFO
3526 "%s[%d] general protection eip:%lx esp:%lx error:%lx\n",
3527 - current->comm, current->pid,
3528 + current->comm, task_pid_nr(current),
3529 regs->eip, regs->esp, error_code);
3530
3531 force_sig(SIGSEGV, current);
3532 @@ -785,6 +764,8 @@ void restart_nmi(void)
3533 #ifdef CONFIG_KPROBES
3534 fastcall void __kprobes do_int3(struct pt_regs *regs, long error_code)
3535 {
3536 + trace_hardirqs_fixup();
3537 +
3538 if (notify_die(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP)
3539 == NOTIFY_STOP)
3540 return;
3541 @@ -822,6 +803,8 @@ fastcall void __kprobes do_debug(struct
3542 unsigned int condition;
3543 struct task_struct *tsk = current;
3544
3545 + trace_hardirqs_fixup();
3546 +
3547 get_debugreg(condition, 6);
3548
3549 if (notify_die(DIE_DEBUG, "debug", regs, condition, error_code,
3550 @@ -1084,20 +1067,6 @@ asmlinkage void math_emulate(long arg)
3551
3552 #endif /* CONFIG_MATH_EMULATION */
3553
3554 -#ifdef CONFIG_X86_F00F_BUG
3555 -void __init trap_init_f00f_bug(void)
3556 -{
3557 - __set_fixmap(FIX_F00F_IDT, __pa(&idt_table), PAGE_KERNEL_RO);
3558 -
3559 - /*
3560 - * Update the IDT descriptor and reload the IDT so that
3561 - * it uses the read-only mapped virtual address.
3562 - */
3563 - idt_descr.address = fix_to_virt(FIX_F00F_IDT);
3564 - load_idt(&idt_descr);
3565 -}
3566 -#endif
3567 -
3568
3569 /*
3570 * NB. All these are "trap gates" (i.e. events_mask isn't set) except
3571 --- sle11-2009-10-16.orig/arch/x86/kernel/traps_64-xen.c 2009-02-16 16:17:21.000000000 +0100
3572 +++ sle11-2009-10-16/arch/x86/kernel/traps_64-xen.c 2009-02-16 16:18:36.000000000 +0100
3573 @@ -1,6 +1,4 @@
3574 /*
3575 - * linux/arch/x86-64/traps.c
3576 - *
3577 * Copyright (C) 1991, 1992 Linus Torvalds
3578 * Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
3579 *
3580 @@ -33,6 +31,7 @@
3581 #include <linux/uaccess.h>
3582 #include <linux/bug.h>
3583 #include <linux/kdebug.h>
3584 +#include <linux/utsname.h>
3585
3586 #if defined(CONFIG_EDAC)
3587 #include <linux/edac.h>
3588 @@ -205,7 +204,7 @@ static unsigned long *in_exception_stack
3589 #define MSG(txt) ops->warning(data, txt)
3590
3591 /*
3592 - * x86-64 can have upto three kernel stacks:
3593 + * x86-64 can have up to three kernel stacks:
3594 * process stack
3595 * interrupt stack
3596 * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
3597 @@ -219,7 +218,7 @@ static inline int valid_stack_ptr(struct
3598
3599 void dump_trace(struct task_struct *tsk, struct pt_regs *regs,
3600 unsigned long *stack,
3601 - struct stacktrace_ops *ops, void *data)
3602 + const struct stacktrace_ops *ops, void *data)
3603 {
3604 const unsigned cpu = get_cpu();
3605 unsigned long *irqstack_end = (unsigned long*)cpu_pda(cpu)->irqstackptr;
3606 @@ -340,7 +339,7 @@ static void print_trace_address(void *da
3607 printk_address(addr);
3608 }
3609
3610 -static struct stacktrace_ops print_trace_ops = {
3611 +static const struct stacktrace_ops print_trace_ops = {
3612 .warning = print_trace_warning,
3613 .warning_symbol = print_trace_warning_symbol,
3614 .stack = print_trace_stack,
3615 @@ -404,6 +403,12 @@ void show_stack(struct task_struct *tsk,
3616 void dump_stack(void)
3617 {
3618 unsigned long dummy;
3619 +
3620 + printk("Pid: %d, comm: %.20s %s %s %.*s\n",
3621 + current->pid, current->comm, print_tainted(),
3622 + init_utsname()->release,
3623 + (int)strcspn(init_utsname()->version, " "),
3624 + init_utsname()->version);
3625 show_trace(NULL, NULL, &dummy);
3626 }
3627
3628 @@ -466,7 +471,7 @@ void out_of_line_bug(void)
3629 EXPORT_SYMBOL(out_of_line_bug);
3630 #endif
3631
3632 -static DEFINE_SPINLOCK(die_lock);
3633 +static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED;
3634 static int die_owner = -1;
3635 static unsigned int die_nest_count;
3636
3637 @@ -478,13 +483,13 @@ unsigned __kprobes long oops_begin(void)
3638 oops_enter();
3639
3640 /* racy, but better than risking deadlock. */
3641 - local_irq_save(flags);
3642 + raw_local_irq_save(flags);
3643 cpu = smp_processor_id();
3644 - if (!spin_trylock(&die_lock)) {
3645 + if (!__raw_spin_trylock(&die_lock)) {
3646 if (cpu == die_owner)
3647 /* nested oops. should stop eventually */;
3648 else
3649 - spin_lock(&die_lock);
3650 + __raw_spin_lock(&die_lock);
3651 }
3652 die_nest_count++;
3653 die_owner = cpu;
3654 @@ -498,12 +503,10 @@ void __kprobes oops_end(unsigned long fl
3655 die_owner = -1;
3656 bust_spinlocks(0);
3657 die_nest_count--;
3658 - if (die_nest_count)
3659 - /* We still own the lock */
3660 - local_irq_restore(flags);
3661 - else
3662 + if (!die_nest_count)
3663 /* Nest count reaches zero, release the lock. */
3664 - spin_unlock_irqrestore(&die_lock, flags);
3665 + __raw_spin_unlock(&die_lock);
3666 + raw_local_irq_restore(flags);
3667 if (panic_on_oops)
3668 panic("Fatal exception");
3669 oops_exit();
3670 @@ -636,6 +639,7 @@ asmlinkage void do_##name(struct pt_regs
3671 info.si_errno = 0; \
3672 info.si_code = sicode; \
3673 info.si_addr = (void __user *)siaddr; \
3674 + trace_hardirqs_fixup(); \
3675 if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr) \
3676 == NOTIFY_STOP) \
3677 return; \
3678 @@ -741,11 +745,8 @@ mem_parity_error(unsigned char reason, s
3679
3680 printk(KERN_EMERG "Dazed and confused, but trying to continue\n");
3681
3682 -#if 0 /* XEN */
3683 /* Clear and disable the memory parity error line. */
3684 - reason = (reason & 0xf) | 4;
3685 - outb(reason, 0x61);
3686 -#endif /* XEN */
3687 + clear_mem_error(reason);
3688 }
3689
3690 static __kprobes void
3691 @@ -754,14 +755,8 @@ io_check_error(unsigned char reason, str
3692 printk("NMI: IOCK error (debug interrupt?)\n");
3693 show_registers(regs);
3694
3695 -#if 0 /* XEN */
3696 /* Re-enable the IOCK line, wait for a few seconds */
3697 - reason = (reason & 0xf) | 8;
3698 - outb(reason, 0x61);
3699 - mdelay(2000);
3700 - reason &= ~8;
3701 - outb(reason, 0x61);
3702 -#endif /* XEN */
3703 + clear_io_check_error(reason);
3704 }
3705
3706 static __kprobes void
3707 @@ -821,6 +816,8 @@ asmlinkage __kprobes void default_do_nmi
3708 /* runs on IST stack. */
3709 asmlinkage void __kprobes do_int3(struct pt_regs * regs, long error_code)
3710 {
3711 + trace_hardirqs_fixup();
3712 +
3713 if (notify_die(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP) == NOTIFY_STOP) {
3714 return;
3715 }
3716 @@ -858,6 +855,8 @@ asmlinkage void __kprobes do_debug(struc
3717 struct task_struct *tsk = current;
3718 siginfo_t info;
3719
3720 + trace_hardirqs_fixup();
3721 +
3722 get_debugreg(condition, 6);
3723
3724 if (notify_die(DIE_DEBUG, "debug", regs, condition, error_code,
3725 --- sle11-2009-10-16.orig/arch/x86/kernel/vsyscall_64-xen.c 2009-02-16 16:17:21.000000000 +0100
3726 +++ sle11-2009-10-16/arch/x86/kernel/vsyscall_64-xen.c 2009-02-16 16:18:36.000000000 +0100
3727 @@ -1,6 +1,4 @@
3728 /*
3729 - * linux/arch/x86_64/kernel/vsyscall.c
3730 - *
3731 * Copyright (C) 2001 Andrea Arcangeli <andrea@suse.de> SuSE
3732 * Copyright 2003 Andi Kleen, SuSE Labs.
3733 *
3734 @@ -50,12 +48,12 @@
3735 ({unsigned long v; \
3736 extern char __vsyscall_0; \
3737 asm("" : "=r" (v) : "0" (x)); \
3738 - ((v - VSYSCALL_FIRST_PAGE) + __pa_symbol(&__vsyscall_0)); })
3739 + ((v - VSYSCALL_START) + __pa_symbol(&__vsyscall_0)); })
3740
3741 /*
3742 * vsyscall_gtod_data contains data that is :
3743 * - readonly from vsyscalls
3744 - * - writen by timer interrupt or systcl (/proc/sys/kernel/vsyscall64)
3745 + * - written by timer interrupt or systcl (/proc/sys/kernel/vsyscall64)
3746 * Try to keep this structure as small as possible to avoid cache line ping pongs
3747 */
3748 int __vgetcpu_mode __section_vgetcpu_mode;
3749 @@ -66,6 +64,16 @@ struct vsyscall_gtod_data __vsyscall_gto
3750 .sysctl_enabled = 1,
3751 };
3752
3753 +void update_vsyscall_tz(void)
3754 +{
3755 + unsigned long flags;
3756 +
3757 + write_seqlock_irqsave(&vsyscall_gtod_data.lock, flags);
3758 + /* sys_tz has changed */
3759 + vsyscall_gtod_data.sys_tz = sys_tz;
3760 + write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
3761 +}
3762 +
3763 void update_vsyscall(struct timespec *wall_time, struct clocksource *clock)
3764 {
3765 unsigned long flags;
3766 @@ -79,8 +87,6 @@ void update_vsyscall(struct timespec *wa
3767 vsyscall_gtod_data.clock.shift = clock->shift;
3768 vsyscall_gtod_data.wall_time_sec = wall_time->tv_sec;
3769 vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
3770 - vsyscall_gtod_data.sys_tz = sys_tz;
3771 - vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
3772 vsyscall_gtod_data.wall_to_monotonic = wall_to_monotonic;
3773 write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
3774 }
3775 @@ -166,7 +172,7 @@ time_t __vsyscall(1) vtime(time_t *t)
3776 if (unlikely(!__vsyscall_gtod_data.sysctl_enabled))
3777 return time_syscall(t);
3778
3779 - vgettimeofday(&tv, 0);
3780 + vgettimeofday(&tv, NULL);
3781 result = tv.tv_sec;
3782 if (t)
3783 *t = result;
3784 @@ -260,18 +266,10 @@ out:
3785 return ret;
3786 }
3787
3788 -static int vsyscall_sysctl_nostrat(ctl_table *t, int __user *name, int nlen,
3789 - void __user *oldval, size_t __user *oldlenp,
3790 - void __user *newval, size_t newlen)
3791 -{
3792 - return -ENOSYS;
3793 -}
3794 -
3795 static ctl_table kernel_table2[] = {
3796 - { .ctl_name = 99, .procname = "vsyscall64",
3797 + { .procname = "vsyscall64",
3798 .data = &vsyscall_gtod_data.sysctl_enabled, .maxlen = sizeof(int),
3799 .mode = 0644,
3800 - .strategy = vsyscall_sysctl_nostrat,
3801 .proc_handler = vsyscall_sysctl_change },
3802 {}
3803 };
3804 @@ -291,9 +289,9 @@ static void __cpuinit vsyscall_set_cpu(i
3805 unsigned long d;
3806 unsigned long node = 0;
3807 #ifdef CONFIG_NUMA
3808 - node = cpu_to_node[cpu];
3809 + node = cpu_to_node(cpu);
3810 #endif
3811 - if (cpu_has(&cpu_data[cpu], X86_FEATURE_RDTSCP))
3812 + if (cpu_has(&cpu_data(cpu), X86_FEATURE_RDTSCP))
3813 write_rdtscp_aux((node << 12) | cpu);
3814
3815 /* Store cpu number in limit so that it can be loaded quickly
3816 --- sle11-2009-10-16.orig/arch/x86/mm/fault_32-xen.c 2009-02-16 16:17:21.000000000 +0100
3817 +++ sle11-2009-10-16/arch/x86/mm/fault_32-xen.c 2009-02-16 16:18:36.000000000 +0100
3818 @@ -25,6 +25,7 @@
3819 #include <linux/kprobes.h>
3820 #include <linux/uaccess.h>
3821 #include <linux/kdebug.h>
3822 +#include <linux/kprobes.h>
3823
3824 #include <asm/system.h>
3825 #include <asm/desc.h>
3826 @@ -32,33 +33,27 @@
3827
3828 extern void die(const char *,struct pt_regs *,long);
3829
3830 -static ATOMIC_NOTIFIER_HEAD(notify_page_fault_chain);
3831 -
3832 -int register_page_fault_notifier(struct notifier_block *nb)
3833 +#ifdef CONFIG_KPROBES
3834 +static inline int notify_page_fault(struct pt_regs *regs)
3835 {
3836 - vmalloc_sync_all();
3837 - return atomic_notifier_chain_register(&notify_page_fault_chain, nb);
3838 -}
3839 -EXPORT_SYMBOL_GPL(register_page_fault_notifier);
3840 + int ret = 0;
3841
3842 -int unregister_page_fault_notifier(struct notifier_block *nb)
3843 -{
3844 - return atomic_notifier_chain_unregister(&notify_page_fault_chain, nb);
3845 -}
3846 -EXPORT_SYMBOL_GPL(unregister_page_fault_notifier);
3847 + /* kprobe_running() needs smp_processor_id() */
3848 + if (!user_mode_vm(regs)) {
3849 + preempt_disable();
3850 + if (kprobe_running() && kprobe_fault_handler(regs, 14))
3851 + ret = 1;
3852 + preempt_enable();
3853 + }
3854
3855 -static inline int notify_page_fault(struct pt_regs *regs, long err)
3856 + return ret;
3857 +}
3858 +#else
3859 +static inline int notify_page_fault(struct pt_regs *regs)
3860 {
3861 - struct die_args args = {
3862 - .regs = regs,
3863 - .str = "page fault",
3864 - .err = err,
3865 - .trapnr = 14,
3866 - .signr = SIGSEGV
3867 - };
3868 - return atomic_notifier_call_chain(&notify_page_fault_chain,
3869 - DIE_PAGE_FAULT, &args);
3870 + return 0;
3871 }
3872 +#endif
3873
3874 /*
3875 * Return EIP plus the CS segment base. The segment limit is also
3876 @@ -110,7 +105,7 @@ static inline unsigned long get_segment_
3877 LDT and other horrors are only used in user space. */
3878 if (seg & (1<<2)) {
3879 /* Must lock the LDT while reading it. */
3880 - down(&current->mm->context.sem);
3881 + mutex_lock(&current->mm->context.lock);
3882 desc = current->mm->context.ldt;
3883 desc = (void *)desc + (seg & ~7);
3884 } else {
3885 @@ -123,7 +118,7 @@ static inline unsigned long get_segment_
3886 base = get_desc_base((unsigned long *)desc);
3887
3888 if (seg & (1<<2)) {
3889 - up(&current->mm->context.sem);
3890 + mutex_unlock(&current->mm->context.lock);
3891 } else
3892 put_cpu();
3893
3894 @@ -244,7 +239,7 @@ static void dump_fault_path(unsigned lon
3895 if (mfn_to_pfn(mfn) >= highstart_pfn)
3896 return;
3897 #endif
3898 - if (p[0] & _PAGE_PRESENT) {
3899 + if ((p[0] & _PAGE_PRESENT) && !(p[0] & _PAGE_PSE)) {
3900 page = mfn_to_pfn(mfn) << PAGE_SHIFT;
3901 p = (unsigned long *) __va(page);
3902 address &= 0x001fffff;
3903 @@ -270,7 +265,8 @@ static void dump_fault_path(unsigned lon
3904 * it's allocated already.
3905 */
3906 if ((machine_to_phys(page) >> PAGE_SHIFT) < max_low_pfn
3907 - && (page & _PAGE_PRESENT)) {
3908 + && (page & _PAGE_PRESENT)
3909 + && !(page & _PAGE_PSE)) {
3910 page = machine_to_phys(page & PAGE_MASK);
3911 page = ((unsigned long *) __va(page))[(address >> PAGE_SHIFT)
3912 & (PTRS_PER_PTE - 1)];
3913 @@ -416,6 +412,11 @@ fastcall void __kprobes do_page_fault(st
3914 int write, si_code;
3915 int fault;
3916
3917 + /*
3918 + * We can fault from pretty much anywhere, with unknown IRQ state.
3919 + */
3920 + trace_hardirqs_fixup();
3921 +
3922 /* get the address */
3923 address = read_cr2();
3924
3925 @@ -453,7 +454,7 @@ fastcall void __kprobes do_page_fault(st
3926 /* Can take a spurious fault if mapping changes R/O -> R/W. */
3927 if (spurious_fault(regs, address, error_code))
3928 return;
3929 - if (notify_page_fault(regs, error_code) == NOTIFY_STOP)
3930 + if (notify_page_fault(regs))
3931 return;
3932 /*
3933 * Don't take the mm semaphore here. If we fixup a prefetch
3934 @@ -462,7 +463,7 @@ fastcall void __kprobes do_page_fault(st
3935 goto bad_area_nosemaphore;
3936 }
3937
3938 - if (notify_page_fault(regs, error_code) == NOTIFY_STOP)
3939 + if (notify_page_fault(regs))
3940 return;
3941
3942 /* It's safe to allow irq's after cr2 has been saved and the vmalloc
3943 @@ -481,7 +482,7 @@ fastcall void __kprobes do_page_fault(st
3944
3945 /* When running in the kernel we expect faults to occur only to
3946 * addresses in user space. All other faults represent errors in the
3947 - * kernel and should generate an OOPS. Unfortunatly, in the case of an
3948 + * kernel and should generate an OOPS. Unfortunately, in the case of an
3949 * erroneous fault occurring in a code path which already holds mmap_sem
3950 * we will deadlock attempting to validate the fault against the
3951 * address space. Luckily the kernel only validly references user
3952 @@ -489,7 +490,7 @@ fastcall void __kprobes do_page_fault(st
3953 * exceptions table.
3954 *
3955 * As the vast majority of faults will be valid we will only perform
3956 - * the source reference check when there is a possibilty of a deadlock.
3957 + * the source reference check when there is a possibility of a deadlock.
3958 * Attempt to lock the address space, if we cannot we then validate the
3959 * source. If this is invalid we can skip the address space check,
3960 * thus avoiding the deadlock.
3961 @@ -598,8 +599,8 @@ bad_area_nosemaphore:
3962 printk_ratelimit()) {
3963 printk("%s%s[%d]: segfault at %08lx eip %08lx "
3964 "esp %08lx error %lx\n",
3965 - tsk->pid > 1 ? KERN_INFO : KERN_EMERG,
3966 - tsk->comm, tsk->pid, address, regs->eip,
3967 + task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
3968 + tsk->comm, task_pid_nr(tsk), address, regs->eip,
3969 regs->esp, error_code);
3970 }
3971 tsk->thread.cr2 = address;
3972 @@ -664,8 +665,7 @@ no_context:
3973 printk(KERN_ALERT "BUG: unable to handle kernel paging"
3974 " request");
3975 printk(" at virtual address %08lx\n",address);
3976 - printk(KERN_ALERT " printing eip:\n");
3977 - printk("%08lx\n", regs->eip);
3978 + printk(KERN_ALERT "printing eip: %08lx\n", regs->eip);
3979 dump_fault_path(address);
3980 }
3981 tsk->thread.cr2 = address;
3982 @@ -681,14 +681,14 @@ no_context:
3983 */
3984 out_of_memory:
3985 up_read(&mm->mmap_sem);
3986 - if (is_init(tsk)) {
3987 + if (is_global_init(tsk)) {
3988 yield();
3989 down_read(&mm->mmap_sem);
3990 goto survive;
3991 }
3992 printk("VM: killing process %s\n", tsk->comm);
3993 if (error_code & 4)
3994 - do_exit(SIGKILL);
3995 + do_group_exit(SIGKILL);
3996 goto no_context;
3997
3998 do_sigbus:
3999 --- sle11-2009-10-16.orig/arch/x86/mm/fault_64-xen.c 2009-02-16 16:17:21.000000000 +0100
4000 +++ sle11-2009-10-16/arch/x86/mm/fault_64-xen.c 2009-02-16 16:18:36.000000000 +0100
4001 @@ -25,6 +25,7 @@
4002 #include <linux/kprobes.h>
4003 #include <linux/uaccess.h>
4004 #include <linux/kdebug.h>
4005 +#include <linux/kprobes.h>
4006
4007 #include <asm/system.h>
4008 #include <asm/pgalloc.h>
4009 @@ -40,34 +41,27 @@
4010 #define PF_RSVD (1<<3)
4011 #define PF_INSTR (1<<4)
4012
4013 -static ATOMIC_NOTIFIER_HEAD(notify_page_fault_chain);
4014 -
4015 -/* Hook to register for page fault notifications */
4016 -int register_page_fault_notifier(struct notifier_block *nb)
4017 +#ifdef CONFIG_KPROBES
4018 +static inline int notify_page_fault(struct pt_regs *regs)
4019 {
4020 - vmalloc_sync_all();
4021 - return atomic_notifier_chain_register(&notify_page_fault_chain, nb);
4022 -}
4023 -EXPORT_SYMBOL_GPL(register_page_fault_notifier);
4024 + int ret = 0;
4025
4026 -int unregister_page_fault_notifier(struct notifier_block *nb)
4027 -{
4028 - return atomic_notifier_chain_unregister(&notify_page_fault_chain, nb);
4029 -}
4030 -EXPORT_SYMBOL_GPL(unregister_page_fault_notifier);
4031 + /* kprobe_running() needs smp_processor_id() */
4032 + if (!user_mode(regs)) {
4033 + preempt_disable();
4034 + if (kprobe_running() && kprobe_fault_handler(regs, 14))
4035 + ret = 1;
4036 + preempt_enable();
4037 + }
4038
4039 -static inline int notify_page_fault(struct pt_regs *regs, long err)
4040 + return ret;
4041 +}
4042 +#else
4043 +static inline int notify_page_fault(struct pt_regs *regs)
4044 {
4045 - struct die_args args = {
4046 - .regs = regs,
4047 - .str = "page fault",
4048 - .err = err,
4049 - .trapnr = 14,
4050 - .signr = SIGSEGV
4051 - };
4052 - return atomic_notifier_call_chain(&notify_page_fault_chain,
4053 - DIE_PAGE_FAULT, &args);
4054 + return 0;
4055 }
4056 +#endif
4057
4058 /* Sometimes the CPU reports invalid exceptions on prefetch.
4059 Check that here and ignore.
4060 @@ -175,7 +169,7 @@ void dump_pagetable(unsigned long addres
4061 pmd = pmd_offset(pud, address);
4062 if (bad_address(pmd)) goto bad;
4063 printk("PMD %lx ", pmd_val(*pmd));
4064 - if (!pmd_present(*pmd)) goto ret;
4065 + if (!pmd_present(*pmd) || pmd_large(*pmd)) goto ret;
4066
4067 pte = pte_offset_kernel(pmd, address);
4068 if (bad_address(pte)) goto bad;
4069 @@ -294,7 +288,6 @@ static int vmalloc_fault(unsigned long a
4070 return 0;
4071 }
4072
4073 -static int page_fault_trace;
4074 int show_unhandled_signals = 1;
4075
4076
4077 @@ -371,6 +364,11 @@ asmlinkage void __kprobes do_page_fault(
4078 if (!user_mode(regs))
4079 error_code &= ~PF_USER; /* means kernel */
4080
4081 + /*
4082 + * We can fault from pretty much anywhere, with unknown IRQ state.
4083 + */
4084 + trace_hardirqs_fixup();
4085 +
4086 tsk = current;
4087 mm = tsk->mm;
4088 prefetchw(&mm->mmap_sem);
4089 @@ -408,7 +406,7 @@ asmlinkage void __kprobes do_page_fault(
4090 /* Can take a spurious fault if mapping changes R/O -> R/W. */
4091 if (spurious_fault(regs, address, error_code))
4092 return;
4093 - if (notify_page_fault(regs, error_code) == NOTIFY_STOP)
4094 + if (notify_page_fault(regs))
4095 return;
4096 /*
4097 * Don't take the mm semaphore here. If we fixup a prefetch
4098 @@ -417,16 +415,12 @@ asmlinkage void __kprobes do_page_fault(
4099 goto bad_area_nosemaphore;
4100 }
4101
4102 - if (notify_page_fault(regs, error_code) == NOTIFY_STOP)
4103 + if (notify_page_fault(regs))
4104 return;
4105
4106 if (likely(regs->eflags & X86_EFLAGS_IF))
4107 local_irq_enable();
4108
4109 - if (unlikely(page_fault_trace))
4110 - printk("pagefault rip:%lx rsp:%lx cs:%lu ss:%lu address %lx error %lx\n",
4111 - regs->rip,regs->rsp,regs->cs,regs->ss,address,error_code);
4112 -
4113 if (unlikely(error_code & PF_RSVD))
4114 pgtable_bad(address, regs, error_code);
4115
4116 @@ -447,7 +441,7 @@ asmlinkage void __kprobes do_page_fault(
4117 again:
4118 /* When running in the kernel we expect faults to occur only to
4119 * addresses in user space. All other faults represent errors in the
4120 - * kernel and should generate an OOPS. Unfortunatly, in the case of an
4121 + * kernel and should generate an OOPS. Unfortunately, in the case of an
4122 * erroneous fault occurring in a code path which already holds mmap_sem
4123 * we will deadlock attempting to validate the fault against the
4124 * address space. Luckily the kernel only validly references user
4125 @@ -455,7 +449,7 @@ asmlinkage void __kprobes do_page_fault(
4126 * exceptions table.
4127 *
4128 * As the vast majority of faults will be valid we will only perform
4129 - * the source reference check when there is a possibilty of a deadlock.
4130 + * the source reference check when there is a possibility of a deadlock.
4131 * Attempt to lock the address space, if we cannot we then validate the
4132 * source. If this is invalid we can skip the address space check,
4133 * thus avoiding the deadlock.
4134 @@ -557,7 +551,7 @@ bad_area_nosemaphore:
4135 if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
4136 printk_ratelimit()) {
4137 printk(
4138 - "%s%s[%d]: segfault at %016lx rip %016lx rsp %016lx error %lx\n",
4139 + "%s%s[%d]: segfault at %lx rip %lx rsp %lx error %lx\n",
4140 tsk->pid > 1 ? KERN_INFO : KERN_EMERG,
4141 tsk->comm, tsk->pid, address, regs->rip,
4142 regs->rsp, error_code);
4143 @@ -623,7 +617,7 @@ no_context:
4144 */
4145 out_of_memory:
4146 up_read(&mm->mmap_sem);
4147 - if (is_init(current)) {
4148 + if (is_global_init(current)) {
4149 yield();
4150 goto again;
4151 }
4152 @@ -690,10 +684,3 @@ void vmalloc_sync_all(void)
4153 BUILD_BUG_ON(!(((MODULES_END - 1) & PGDIR_MASK) ==
4154 (__START_KERNEL & PGDIR_MASK)));
4155 }
4156 -
4157 -static int __init enable_pagefaulttrace(char *str)
4158 -{
4159 - page_fault_trace = 1;
4160 - return 1;
4161 -}
4162 -__setup("pagefaulttrace", enable_pagefaulttrace);
4163 --- sle11-2009-10-16.orig/arch/x86/mm/hypervisor.c 2009-03-04 11:28:34.000000000 +0100
4164 +++ sle11-2009-10-16/arch/x86/mm/hypervisor.c 2009-05-06 10:23:43.000000000 +0200
4165 @@ -496,6 +496,9 @@ int xen_create_contiguous_region(
4166 unsigned long frame, flags;
4167 unsigned int i;
4168 int rc, success;
4169 +#ifdef CONFIG_64BIT
4170 + pte_t *ptep = NULL;
4171 +#endif
4172 struct xen_memory_exchange exchange = {
4173 .in = {
4174 .nr_extents = 1UL << order,
4175 @@ -521,6 +524,27 @@ int xen_create_contiguous_region(
4176 if (unlikely(order > MAX_CONTIG_ORDER))
4177 return -ENOMEM;
4178
4179 +#ifdef CONFIG_64BIT
4180 + if (unlikely(vstart > PAGE_OFFSET + MAXMEM)) {
4181 + unsigned int level;
4182 +
4183 + if (vstart < __START_KERNEL_map
4184 + || vstart + (PAGE_SIZE << order) > (unsigned long)_end)
4185 + return -EINVAL;
4186 + ptep = lookup_address((unsigned long)__va(__pa(vstart)),
4187 + &level);
4188 + if (ptep && pte_none(*ptep))
4189 + ptep = NULL;
4190 + if (vstart < __START_KERNEL && ptep)
4191 + return -EINVAL;
4192 + if (order > MAX_CONTIG_ORDER - 1)
4193 + return -ENOMEM;
4194 + }
4195 +#else
4196 + if (unlikely(vstart + (PAGE_SIZE << order) > (unsigned long)high_memory))
4197 + return -EINVAL;
4198 +#endif
4199 +
4200 set_xen_guest_handle(exchange.in.extent_start, in_frames);
4201 set_xen_guest_handle(exchange.out.extent_start, &out_frame);
4202
4203 @@ -533,9 +557,19 @@ int xen_create_contiguous_region(
4204 in_frames[i] = pfn_to_mfn((__pa(vstart) >> PAGE_SHIFT) + i);
4205 MULTI_update_va_mapping(cr_mcl + i, vstart + (i*PAGE_SIZE),
4206 __pte_ma(0), 0);
4207 +#ifdef CONFIG_64BIT
4208 + if (ptep)
4209 + MULTI_update_va_mapping(cr_mcl + i + (1U << order),
4210 + (unsigned long)__va(__pa(vstart)) + (i*PAGE_SIZE),
4211 + __pte_ma(0), 0);
4212 +#endif
4213 set_phys_to_machine((__pa(vstart)>>PAGE_SHIFT)+i,
4214 INVALID_P2M_ENTRY);
4215 }
4216 +#ifdef CONFIG_64BIT
4217 + if (ptep)
4218 + i += i;
4219 +#endif
4220 if (HYPERVISOR_multicall_check(cr_mcl, i, NULL))
4221 BUG();
4222
4223 @@ -569,9 +603,18 @@ int xen_create_contiguous_region(
4224 frame = success ? (out_frame + i) : in_frames[i];
4225 MULTI_update_va_mapping(cr_mcl + i, vstart + (i*PAGE_SIZE),
4226 pfn_pte_ma(frame, PAGE_KERNEL), 0);
4227 +#ifdef CONFIG_64BIT
4228 + if (ptep)
4229 + MULTI_update_va_mapping(cr_mcl + i + (1U << order),
4230 + (unsigned long)__va(__pa(vstart)) + (i*PAGE_SIZE),
4231 + pfn_pte_ma(frame, PAGE_KERNEL_RO), 0);
4232 +#endif
4233 set_phys_to_machine((__pa(vstart)>>PAGE_SHIFT)+i, frame);
4234 }
4235 -
4236 +#ifdef CONFIG_64BIT
4237 + if (ptep)
4238 + i += i;
4239 +#endif
4240 cr_mcl[i - 1].args[MULTI_UVMFLAGS_INDEX] = order
4241 ? UVMF_TLB_FLUSH|UVMF_ALL
4242 : UVMF_INVLPG|UVMF_ALL;
4243 --- sle11-2009-10-16.orig/arch/x86/mm/init_32-xen.c 2009-02-16 16:17:21.000000000 +0100
4244 +++ sle11-2009-10-16/arch/x86/mm/init_32-xen.c 2009-02-16 16:18:36.000000000 +0100
4245 @@ -94,7 +94,14 @@ static pte_t * __init one_page_table_ini
4246 #else
4247 if (!(__pmd_val(*pmd) & _PAGE_PRESENT)) {
4248 #endif
4249 - pte_t *page_table = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE);
4250 + pte_t *page_table = NULL;
4251 +
4252 +#ifdef CONFIG_DEBUG_PAGEALLOC
4253 + page_table = (pte_t *) alloc_bootmem_pages(PAGE_SIZE);
4254 +#endif
4255 + if (!page_table)
4256 + page_table =
4257 + (pte_t *)alloc_bootmem_low_pages(PAGE_SIZE);
4258
4259 paravirt_alloc_pt(&init_mm, __pa(page_table) >> PAGE_SHIFT);
4260 make_lowmem_page_readonly(page_table,
4261 @@ -102,7 +109,7 @@ static pte_t * __init one_page_table_ini
4262 set_pmd(pmd, __pmd(__pa(page_table) | _PAGE_TABLE));
4263 BUG_ON(page_table != pte_offset_kernel(pmd, 0));
4264 }
4265 -
4266 +
4267 return pte_offset_kernel(pmd, 0);
4268 }
4269
4270 @@ -360,8 +367,13 @@ extern void set_highmem_pages_init(int);
4271 static void __init set_highmem_pages_init(int bad_ppro)
4272 {
4273 int pfn;
4274 - for (pfn = highstart_pfn; pfn < highend_pfn; pfn++)
4275 - add_one_highpage_init(pfn_to_page(pfn), pfn, bad_ppro);
4276 + for (pfn = highstart_pfn; pfn < highend_pfn; pfn++) {
4277 + /*
4278 + * Holes under sparsemem might not have no mem_map[]:
4279 + */
4280 + if (pfn_valid(pfn))
4281 + add_one_highpage_init(pfn_to_page(pfn), pfn, bad_ppro);
4282 + }
4283 totalram_pages += totalhigh_pages;
4284 }
4285 #endif /* CONFIG_FLATMEM */
4286 @@ -779,35 +791,18 @@ int arch_add_memory(int nid, u64 start,
4287 return __add_pages(zone, start_pfn, nr_pages);
4288 }
4289
4290 -int remove_memory(u64 start, u64 size)
4291 -{
4292 - return -EINVAL;
4293 -}
4294 -EXPORT_SYMBOL_GPL(remove_memory);
4295 #endif
4296
4297 struct kmem_cache *pmd_cache;
4298
4299 void __init pgtable_cache_init(void)
4300 {
4301 - size_t pgd_size = PTRS_PER_PGD*sizeof(pgd_t);
4302 -
4303 - if (PTRS_PER_PMD > 1) {
4304 + if (PTRS_PER_PMD > 1)
4305 pmd_cache = kmem_cache_create("pmd",
4306 - PTRS_PER_PMD*sizeof(pmd_t),
4307 - PTRS_PER_PMD*sizeof(pmd_t),
4308 - SLAB_PANIC,
4309 - pmd_ctor);
4310 - if (!SHARED_KERNEL_PMD) {
4311 - /* If we're in PAE mode and have a non-shared
4312 - kernel pmd, then the pgd size must be a
4313 - page size. This is because the pgd_list
4314 - links through the page structure, so there
4315 - can only be one pgd per page for this to
4316 - work. */
4317 - pgd_size = PAGE_SIZE;
4318 - }
4319 - }
4320 + PTRS_PER_PMD*sizeof(pmd_t),
4321 + PTRS_PER_PMD*sizeof(pmd_t),
4322 + SLAB_PANIC,
4323 + pmd_ctor);
4324 }
4325
4326 /*
4327 --- sle11-2009-10-16.orig/arch/x86/mm/init_64-xen.c 2009-02-16 16:17:21.000000000 +0100
4328 +++ sle11-2009-10-16/arch/x86/mm/init_64-xen.c 2009-02-16 16:18:36.000000000 +0100
4329 @@ -767,7 +767,7 @@ static void xen_finish_init_mapping(void
4330 /* Setup the direct mapping of the physical memory at PAGE_OFFSET.
4331 This runs before bootmem is initialized and gets pages directly from the
4332 physical memory. To access them they are temporarily mapped. */
4333 -void __meminit init_memory_mapping(unsigned long start, unsigned long end)
4334 +void __init_refok init_memory_mapping(unsigned long start, unsigned long end)
4335 {
4336 unsigned long next;
4337
4338 @@ -901,12 +901,6 @@ error:
4339 }
4340 EXPORT_SYMBOL_GPL(arch_add_memory);
4341
4342 -int remove_memory(u64 start, u64 size)
4343 -{
4344 - return -EINVAL;
4345 -}
4346 -EXPORT_SYMBOL_GPL(remove_memory);
4347 -
4348 #if !defined(CONFIG_ACPI_NUMA) && defined(CONFIG_NUMA)
4349 int memory_add_physaddr_to_nid(u64 start)
4350 {
4351 @@ -1173,14 +1167,6 @@ int in_gate_area_no_task(unsigned long a
4352 return (addr >= VSYSCALL_START) && (addr < VSYSCALL_END);
4353 }
4354
4355 -#ifndef CONFIG_XEN
4356 -void * __init alloc_bootmem_high_node(pg_data_t *pgdat, unsigned long size)
4357 -{
4358 - return __alloc_bootmem_core(pgdat->bdata, size,
4359 - SMP_CACHE_BYTES, (4UL*1024*1024*1024), 0);
4360 -}
4361 -#endif
4362 -
4363 const char *arch_vma_name(struct vm_area_struct *vma)
4364 {
4365 if (vma->vm_mm && vma->vm_start == (long)vma->vm_mm->context.vdso)
4366 @@ -1189,3 +1175,48 @@ const char *arch_vma_name(struct vm_area
4367 return "[vsyscall]";
4368 return NULL;
4369 }
4370 +
4371 +#ifdef CONFIG_SPARSEMEM_VMEMMAP
4372 +/*
4373 + * Initialise the sparsemem vmemmap using huge-pages at the PMD level.
4374 + */
4375 +int __meminit vmemmap_populate(struct page *start_page,
4376 + unsigned long size, int node)
4377 +{
4378 + unsigned long addr = (unsigned long)start_page;
4379 + unsigned long end = (unsigned long)(start_page + size);
4380 + unsigned long next;
4381 + pgd_t *pgd;
4382 + pud_t *pud;
4383 + pmd_t *pmd;
4384 +
4385 + for (; addr < end; addr = next) {
4386 + next = pmd_addr_end(addr, end);
4387 +
4388 + pgd = vmemmap_pgd_populate(addr, node);
4389 + if (!pgd)
4390 + return -ENOMEM;
4391 + pud = vmemmap_pud_populate(pgd, addr, node);
4392 + if (!pud)
4393 + return -ENOMEM;
4394 +
4395 + pmd = pmd_offset(pud, addr);
4396 + if (pmd_none(*pmd)) {
4397 + pte_t entry;
4398 + void *p = vmemmap_alloc_block(PMD_SIZE, node);
4399 + if (!p)
4400 + return -ENOMEM;
4401 +
4402 + entry = pfn_pte(__pa(p) >> PAGE_SHIFT, PAGE_KERNEL);
4403 + mk_pte_huge(entry);
4404 + set_pmd(pmd, __pmd(pte_val(entry)));
4405 +
4406 + printk(KERN_DEBUG " [%lx-%lx] PMD ->%p on node %d\n",
4407 + addr, addr + PMD_SIZE - 1, p, node);
4408 + } else
4409 + vmemmap_verify((pte_t *)pmd, node, addr, next);
4410 + }
4411 +
4412 + return 0;
4413 +}
4414 +#endif
4415 --- sle11-2009-10-16.orig/arch/x86/mm/pageattr_64-xen.c 2009-02-16 16:17:21.000000000 +0100
4416 +++ sle11-2009-10-16/arch/x86/mm/pageattr_64-xen.c 2009-02-16 16:18:36.000000000 +0100
4417 @@ -17,9 +17,6 @@
4418 #include <asm/pgalloc.h>
4419 #include <asm/mmu_context.h>
4420
4421 -LIST_HEAD(mm_unpinned);
4422 -DEFINE_SPINLOCK(mm_unpinned_lock);
4423 -
4424 static void _pin_lock(struct mm_struct *mm, int lock) {
4425 if (lock)
4426 spin_lock(&mm->page_table_lock);
4427 @@ -81,8 +78,8 @@ static void _pin_lock(struct mm_struct *
4428 #define PIN_BATCH 8
4429 static DEFINE_PER_CPU(multicall_entry_t[PIN_BATCH], pb_mcl);
4430
4431 -static inline unsigned int mm_walk_set_prot(void *pt, pgprot_t flags,
4432 - unsigned int cpu, unsigned int seq)
4433 +static inline unsigned int pgd_walk_set_prot(void *pt, pgprot_t flags,
4434 + unsigned int cpu, unsigned int seq)
4435 {
4436 struct page *page = virt_to_page(pt);
4437 unsigned long pfn = page_to_pfn(page);
4438 @@ -100,9 +97,9 @@ static inline unsigned int mm_walk_set_p
4439 return seq;
4440 }
4441
4442 -static void mm_walk(struct mm_struct *mm, pgprot_t flags)
4443 +static void pgd_walk(pgd_t *pgd_base, pgprot_t flags)
4444 {
4445 - pgd_t *pgd;
4446 + pgd_t *pgd = pgd_base;
4447 pud_t *pud;
4448 pmd_t *pmd;
4449 pte_t *pte;
4450 @@ -110,7 +107,6 @@ static void mm_walk(struct mm_struct *mm
4451 unsigned int cpu, seq;
4452 multicall_entry_t *mcl;
4453
4454 - pgd = mm->pgd;
4455 cpu = get_cpu();
4456
4457 /*
4458 @@ -125,18 +121,18 @@ static void mm_walk(struct mm_struct *mm
4459 continue;
4460 pud = pud_offset(pgd, 0);
4461 if (PTRS_PER_PUD > 1) /* not folded */
4462 - seq = mm_walk_set_prot(pud,flags,cpu,seq);
4463 + seq = pgd_walk_set_prot(pud,flags,cpu,seq);
4464 for (u = 0; u < PTRS_PER_PUD; u++, pud++) {
4465 if (pud_none(*pud))
4466 continue;
4467 pmd = pmd_offset(pud, 0);
4468 if (PTRS_PER_PMD > 1) /* not folded */
4469 - seq = mm_walk_set_prot(pmd,flags,cpu,seq);
4470 + seq = pgd_walk_set_prot(pmd,flags,cpu,seq);
4471 for (m = 0; m < PTRS_PER_PMD; m++, pmd++) {
4472 if (pmd_none(*pmd))
4473 continue;
4474 pte = pte_offset_kernel(pmd,0);
4475 - seq = mm_walk_set_prot(pte,flags,cpu,seq);
4476 + seq = pgd_walk_set_prot(pte,flags,cpu,seq);
4477 }
4478 }
4479 }
4480 @@ -148,12 +144,12 @@ static void mm_walk(struct mm_struct *mm
4481 seq = 0;
4482 }
4483 MULTI_update_va_mapping(mcl + seq,
4484 - (unsigned long)__user_pgd(mm->pgd),
4485 - pfn_pte(virt_to_phys(__user_pgd(mm->pgd))>>PAGE_SHIFT, flags),
4486 + (unsigned long)__user_pgd(pgd_base),
4487 + pfn_pte(virt_to_phys(__user_pgd(pgd_base))>>PAGE_SHIFT, flags),
4488 0);
4489 MULTI_update_va_mapping(mcl + seq + 1,
4490 - (unsigned long)mm->pgd,
4491 - pfn_pte(virt_to_phys(mm->pgd)>>PAGE_SHIFT, flags),
4492 + (unsigned long)pgd_base,
4493 + pfn_pte(virt_to_phys(pgd_base)>>PAGE_SHIFT, flags),
4494 UVMF_TLB_FLUSH);
4495 if (unlikely(HYPERVISOR_multicall_check(mcl, seq + 2, NULL)))
4496 BUG();
4497 @@ -161,21 +157,35 @@ static void mm_walk(struct mm_struct *mm
4498 put_cpu();
4499 }
4500
4501 +static void __pgd_pin(pgd_t *pgd)
4502 +{
4503 + pgd_walk(pgd, PAGE_KERNEL_RO);
4504 + xen_pgd_pin(__pa(pgd)); /* kernel */
4505 + xen_pgd_pin(__pa(__user_pgd(pgd))); /* user */
4506 + SetPagePinned(virt_to_page(pgd));
4507 +}
4508 +
4509 +static void __pgd_unpin(pgd_t *pgd)
4510 +{
4511 + xen_pgd_unpin(__pa(pgd));
4512 + xen_pgd_unpin(__pa(__user_pgd(pgd)));
4513 + pgd_walk(pgd, PAGE_KERNEL);
4514 + ClearPagePinned(virt_to_page(pgd));
4515 +}
4516 +
4517 +void pgd_test_and_unpin(pgd_t *pgd)
4518 +{
4519 + if (PagePinned(virt_to_page(pgd)))
4520 + __pgd_unpin(pgd);
4521 +}
4522 +
4523 void mm_pin(struct mm_struct *mm)
4524 {
4525 if (xen_feature(XENFEAT_writable_page_tables))
4526 return;
4527
4528 pin_lock(mm);
4529 -
4530 - mm_walk(mm, PAGE_KERNEL_RO);
4531 - xen_pgd_pin(__pa(mm->pgd)); /* kernel */
4532 - xen_pgd_pin(__pa(__user_pgd(mm->pgd))); /* user */
4533 - SetPagePinned(virt_to_page(mm->pgd));
4534 - spin_lock(&mm_unpinned_lock);
4535 - list_del(&mm->context.unpinned);
4536 - spin_unlock(&mm_unpinned_lock);
4537 -
4538 + __pgd_pin(mm->pgd);
4539 pin_unlock(mm);
4540 }
4541
4542 @@ -185,34 +195,30 @@ void mm_unpin(struct mm_struct *mm)
4543 return;
4544
4545 pin_lock(mm);
4546 -
4547 - xen_pgd_unpin(__pa(mm->pgd));
4548 - xen_pgd_unpin(__pa(__user_pgd(mm->pgd)));
4549 - mm_walk(mm, PAGE_KERNEL);
4550 - ClearPagePinned(virt_to_page(mm->pgd));
4551 - spin_lock(&mm_unpinned_lock);
4552 - list_add(&mm->context.unpinned, &mm_unpinned);
4553 - spin_unlock(&mm_unpinned_lock);
4554 -
4555 + __pgd_unpin(mm->pgd);
4556 pin_unlock(mm);
4557 }
4558
4559 void mm_pin_all(void)
4560 {
4561 + struct page *page;
4562 + unsigned long flags;
4563 +
4564 if (xen_feature(XENFEAT_writable_page_tables))
4565 return;
4566
4567 /*
4568 - * Allow uninterrupted access to the mm_unpinned list. We don't
4569 - * actually take the mm_unpinned_lock as it is taken inside mm_pin().
4570 + * Allow uninterrupted access to the pgd_list. Also protects
4571 + * __pgd_pin() by disabling preemption.
4572 * All other CPUs must be at a safe point (e.g., in stop_machine
4573 * or offlined entirely).
4574 */
4575 - preempt_disable();
4576 - while (!list_empty(&mm_unpinned))
4577 - mm_pin(list_entry(mm_unpinned.next, struct mm_struct,
4578 - context.unpinned));
4579 - preempt_enable();
4580 + spin_lock_irqsave(&pgd_lock, flags);
4581 + list_for_each_entry(page, &pgd_list, lru) {
4582 + if (!PagePinned(page))
4583 + __pgd_pin((pgd_t *)page_address(page));
4584 + }
4585 + spin_unlock_irqrestore(&pgd_lock, flags);
4586 }
4587
4588 void arch_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
4589 @@ -331,11 +337,11 @@ static struct page *split_large_page(uns
4590 return base;
4591 }
4592
4593 -static void cache_flush_page(void *adr)
4594 +void clflush_cache_range(void *adr, int size)
4595 {
4596 int i;
4597 - for (i = 0; i < PAGE_SIZE; i += boot_cpu_data.x86_clflush_size)
4598 - asm volatile("clflush (%0)" :: "r" (adr + i));
4599 + for (i = 0; i < size; i += boot_cpu_data.x86_clflush_size)
4600 + clflush(adr+i);
4601 }
4602
4603 static void flush_kernel_map(void *arg)
4604 @@ -350,7 +356,7 @@ static void flush_kernel_map(void *arg)
4605 asm volatile("wbinvd" ::: "memory");
4606 else list_for_each_entry(pg, l, lru) {
4607 void *adr = page_address(pg);
4608 - cache_flush_page(adr);
4609 + clflush_cache_range(adr, PAGE_SIZE);
4610 }
4611 __flush_tlb_all();
4612 }
4613 @@ -418,6 +424,7 @@ __change_page_attr(unsigned long address
4614 split = split_large_page(address, prot, ref_prot2);
4615 if (!split)
4616 return -ENOMEM;
4617 + pgprot_val(ref_prot2) &= ~_PAGE_NX;
4618 set_pte(kpte, mk_pte(split, ref_prot2));
4619 kpte_page = split;
4620 }
4621 @@ -510,9 +517,14 @@ void global_flush_tlb(void)
4622 struct page *pg, *next;
4623 struct list_head l;
4624
4625 - down_read(&init_mm.mmap_sem);
4626 + /*
4627 + * Write-protect the semaphore, to exclude two contexts
4628 + * doing a list_replace_init() call in parallel and to
4629 + * exclude new additions to the deferred_pages list:
4630 + */
4631 + down_write(&init_mm.mmap_sem);
4632 list_replace_init(&deferred_pages, &l);
4633 - up_read(&init_mm.mmap_sem);
4634 + up_write(&init_mm.mmap_sem);
4635
4636 flush_map(&l);
4637
4638 --- sle11-2009-10-16.orig/arch/x86/mm/pgtable_32-xen.c 2009-02-16 16:17:21.000000000 +0100
4639 +++ sle11-2009-10-16/arch/x86/mm/pgtable_32-xen.c 2009-02-16 16:18:36.000000000 +0100
4640 @@ -6,6 +6,7 @@
4641 #include <linux/kernel.h>
4642 #include <linux/errno.h>
4643 #include <linux/mm.h>
4644 +#include <linux/nmi.h>
4645 #include <linux/swap.h>
4646 #include <linux/smp.h>
4647 #include <linux/highmem.h>
4648 @@ -46,6 +47,8 @@ void show_mem(void)
4649 for_each_online_pgdat(pgdat) {
4650 pgdat_resize_lock(pgdat, &flags);
4651 for (i = 0; i < pgdat->node_spanned_pages; ++i) {
4652 + if (unlikely(i % MAX_ORDER_NR_PAGES == 0))
4653 + touch_nmi_watchdog();
4654 page = pgdat_page_nr(pgdat, i);
4655 total++;
4656 if (PageHighMem(page))
4657 @@ -206,7 +209,7 @@ void pte_free(struct page *pte)
4658 __free_page(pte);
4659 }
4660
4661 -void pmd_ctor(void *pmd, struct kmem_cache *cache, unsigned long flags)
4662 +void pmd_ctor(struct kmem_cache *cache, void *pmd)
4663 {
4664 memset(pmd, 0, PTRS_PER_PMD*sizeof(pmd_t));
4665 }
4666 --- sle11-2009-10-16.orig/arch/x86/pci/irq-xen.c 2009-02-16 16:17:21.000000000 +0100
4667 +++ sle11-2009-10-16/arch/x86/pci/irq-xen.c 2009-02-16 16:18:36.000000000 +0100
4668 @@ -173,7 +173,7 @@ void eisa_set_level_irq(unsigned int irq
4669 }
4670
4671 /*
4672 - * Common IRQ routing practice: nybbles in config space,
4673 + * Common IRQ routing practice: nibbles in config space,
4674 * offset by some magic constant.
4675 */
4676 static unsigned int read_config_nybble(struct pci_dev *router, unsigned offset, unsigned nr)
4677 @@ -496,6 +496,26 @@ static int pirq_amd756_set(struct pci_de
4678 return 1;
4679 }
4680
4681 +/*
4682 + * PicoPower PT86C523
4683 + */
4684 +static int pirq_pico_get(struct pci_dev *router, struct pci_dev *dev, int pirq)
4685 +{
4686 + outb(0x10 + ((pirq - 1) >> 1), 0x24);
4687 + return ((pirq - 1) & 1) ? (inb(0x26) >> 4) : (inb(0x26) & 0xf);
4688 +}
4689 +
4690 +static int pirq_pico_set(struct pci_dev *router, struct pci_dev *dev, int pirq,
4691 + int irq)
4692 +{
4693 + unsigned int x;
4694 + outb(0x10 + ((pirq - 1) >> 1), 0x24);
4695 + x = inb(0x26);
4696 + x = ((pirq - 1) & 1) ? ((x & 0x0f) | (irq << 4)) : ((x & 0xf0) | (irq));
4697 + outb(x, 0x26);
4698 + return 1;
4699 +}
4700 +
4701 #ifdef CONFIG_PCI_BIOS
4702
4703 static int pirq_bios_set(struct pci_dev *router, struct pci_dev *dev, int pirq, int irq)
4704 @@ -569,7 +589,7 @@ static __init int via_router_probe(struc
4705 /* FIXME: We should move some of the quirk fixup stuff here */
4706
4707 /*
4708 - * work arounds for some buggy BIOSes
4709 + * workarounds for some buggy BIOSes
4710 */
4711 if (device == PCI_DEVICE_ID_VIA_82C586_0) {
4712 switch(router->device) {
4713 @@ -725,6 +745,24 @@ static __init int amd_router_probe(struc
4714 return 1;
4715 }
4716
4717 +static __init int pico_router_probe(struct irq_router *r, struct pci_dev *router, u16 device)
4718 +{
4719 + switch (device) {
4720 + case PCI_DEVICE_ID_PICOPOWER_PT86C523:
4721 + r->name = "PicoPower PT86C523";
4722 + r->get = pirq_pico_get;
4723 + r->set = pirq_pico_set;
4724 + return 1;
4725 +
4726 + case PCI_DEVICE_ID_PICOPOWER_PT86C523BBP:
4727 + r->name = "PicoPower PT86C523 rev. BB+";
4728 + r->get = pirq_pico_get;
4729 + r->set = pirq_pico_set;
4730 + return 1;
4731 + }
4732 + return 0;
4733 +}
4734 +
4735 static __initdata struct irq_router_handler pirq_routers[] = {
4736 { PCI_VENDOR_ID_INTEL, intel_router_probe },
4737 { PCI_VENDOR_ID_AL, ali_router_probe },
4738 @@ -736,6 +774,7 @@ static __initdata struct irq_router_hand
4739 { PCI_VENDOR_ID_VLSI, vlsi_router_probe },
4740 { PCI_VENDOR_ID_SERVERWORKS, serverworks_router_probe },
4741 { PCI_VENDOR_ID_AMD, amd_router_probe },
4742 + { PCI_VENDOR_ID_PICOPOWER, pico_router_probe },
4743 /* Someone with docs needs to add the ATI Radeon IGP */
4744 { 0, NULL }
4745 };
4746 @@ -1014,7 +1053,7 @@ static void __init pcibios_fixup_irqs(vo
4747 * Work around broken HP Pavilion Notebooks which assign USB to
4748 * IRQ 9 even though it is actually wired to IRQ 11
4749 */
4750 -static int __init fix_broken_hp_bios_irq9(struct dmi_system_id *d)
4751 +static int __init fix_broken_hp_bios_irq9(const struct dmi_system_id *d)
4752 {
4753 if (!broken_hp_bios_irq9) {
4754 broken_hp_bios_irq9 = 1;
4755 @@ -1027,7 +1066,7 @@ static int __init fix_broken_hp_bios_irq
4756 * Work around broken Acer TravelMate 360 Notebooks which assign
4757 * Cardbus to IRQ 11 even though it is actually wired to IRQ 10
4758 */
4759 -static int __init fix_acer_tm360_irqrouting(struct dmi_system_id *d)
4760 +static int __init fix_acer_tm360_irqrouting(const struct dmi_system_id *d)
4761 {
4762 if (!acer_tm360_irqrouting) {
4763 acer_tm360_irqrouting = 1;
4764 --- sle11-2009-10-16.orig/drivers/acpi/processor_idle.c 2009-08-26 11:52:33.000000000 +0200
4765 +++ sle11-2009-10-16/drivers/acpi/processor_idle.c 2009-06-29 15:29:06.000000000 +0200
4766 @@ -1749,6 +1749,13 @@ int acpi_processor_cst_has_changed(struc
4767 if (!pr->flags.power_setup_done)
4768 return -ENODEV;
4769
4770 + if (processor_pm_external()) {
4771 + acpi_processor_get_power_info(pr);
4772 + processor_notify_external(pr,
4773 + PROCESSOR_PM_CHANGE, PM_TYPE_IDLE);
4774 + return ret;
4775 + }
4776 +
4777 cpuidle_pause_and_lock();
4778 cpuidle_disable_device(&pr->power.dev);
4779 acpi_processor_get_power_info(pr);
4780 --- sle11-2009-10-16.orig/drivers/cpuidle/Kconfig 2009-10-28 14:55:06.000000000 +0100
4781 +++ sle11-2009-10-16/drivers/cpuidle/Kconfig 2009-02-16 16:18:36.000000000 +0100
4782 @@ -1,6 +1,7 @@
4783
4784 config CPU_IDLE
4785 bool "CPU idle PM support"
4786 + depends on !PROCESSOR_EXTERNAL_CONTROL
4787 default ACPI
4788 help
4789 CPU idle is a generic framework for supporting software-controlled
4790 --- sle11-2009-10-16.orig/drivers/oprofile/cpu_buffer.c 2009-02-16 16:01:39.000000000 +0100
4791 +++ sle11-2009-10-16/drivers/oprofile/cpu_buffer.c 2009-03-12 16:15:32.000000000 +0100
4792 @@ -308,6 +308,37 @@ void oprofile_add_trace(unsigned long pc
4793 }
4794
4795 #ifdef CONFIG_XEN
4796 +/*
4797 + * This is basically log_sample(b, ESCAPE_CODE, cpu_mode, CPU_TRACE_BEGIN),
4798 + * as was previously accessible through oprofile_add_pc().
4799 + */
4800 +void oprofile_add_mode(int cpu_mode)
4801 +{
4802 + struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(cpu_buffer);
4803 + struct task_struct *task;
4804 +
4805 + if (nr_available_slots(cpu_buf) < 3) {
4806 + cpu_buf->sample_lost_overflow++;
4807 + return;
4808 + }
4809 +
4810 + task = current;
4811 +
4812 + /* notice a switch from user->kernel or vice versa */
4813 + if (cpu_buf->last_cpu_mode != cpu_mode) {
4814 + cpu_buf->last_cpu_mode = cpu_mode;
4815 + add_code(cpu_buf, cpu_mode);
4816 + }
4817 +
4818 + /* notice a task switch */
4819 + if (cpu_buf->last_task != task) {
4820 + cpu_buf->last_task = task;
4821 + add_code(cpu_buf, (unsigned long)task);
4822 + }
4823 +
4824 + add_code(cpu_buf, CPU_TRACE_BEGIN);
4825 +}
4826 +
4827 int oprofile_add_domain_switch(int32_t domain_id)
4828 {
4829 struct oprofile_cpu_buffer * cpu_buf = &cpu_buffer[smp_processor_id()];
4830 --- sle11-2009-10-16.orig/drivers/pci/msi-xen.c 2008-12-15 11:27:22.000000000 +0100
4831 +++ sle11-2009-10-16/drivers/pci/msi-xen.c 2009-02-16 16:18:36.000000000 +0100
4832 @@ -264,6 +264,12 @@ static int msi_map_vector(struct pci_dev
4833 return msi_map_pirq_to_vector(dev, -1, entry_nr, table_base);
4834 }
4835
4836 +static void pci_intx_for_msi(struct pci_dev *dev, int enable)
4837 +{
4838 + if (!(dev->dev_flags & PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG))
4839 + pci_intx(dev, enable);
4840 +}
4841 +
4842 #ifdef CONFIG_PM
4843 static void __pci_restore_msi_state(struct pci_dev *dev)
4844 {
4845 @@ -272,7 +278,7 @@ static void __pci_restore_msi_state(stru
4846 if (!dev->msi_enabled)
4847 return;
4848
4849 - pci_intx(dev, 0); /* disable intx */
4850 + pci_intx_for_msi(dev, 0);
4851 msi_set_enable(dev, 0);
4852
4853 pirq = msi_map_pirq_to_vector(dev, dev->irq, 0, 0);
4854 @@ -295,7 +301,7 @@ static void __pci_restore_msix_state(str
4855 if (!dev->msix_enabled)
4856 return;
4857
4858 - pci_intx(dev, 0); /* disable intx */
4859 + pci_intx_for_msi(dev, 0);
4860 msix_set_enable(dev, 0);
4861
4862 msi_dev_entry = get_msi_dev_pirq_list(dev);
4863 @@ -348,7 +354,7 @@ static int msi_capability_init(struct pc
4864 return -EBUSY;
4865
4866 /* Set MSI enabled bits */
4867 - pci_intx(dev, 0); /* disable intx */
4868 + pci_intx_for_msi(dev, 0);
4869 msi_set_enable(dev, 1);
4870 dev->msi_enabled = 1;
4871
4872 @@ -422,7 +428,7 @@ static int msix_capability_init(struct p
4873 return avail;
4874 }
4875
4876 - pci_intx(dev, 0); /* disable intx */
4877 + pci_intx_for_msi(dev, 0);
4878 msix_set_enable(dev, 1);
4879 dev->msix_enabled = 1;
4880
4881 @@ -557,7 +563,7 @@ void pci_disable_msi(struct pci_dev* dev
4882
4883 /* Disable MSI mode */
4884 msi_set_enable(dev, 0);
4885 - pci_intx(dev, 1); /* enable intx */
4886 + pci_intx_for_msi(dev, 1);
4887 dev->msi_enabled = 0;
4888 }
4889 EXPORT_SYMBOL(pci_disable_msi);
4890 @@ -696,7 +702,7 @@ void pci_disable_msix(struct pci_dev* de
4891
4892 /* Disable MSI mode */
4893 msix_set_enable(dev, 0);
4894 - pci_intx(dev, 1); /* enable intx */
4895 + pci_intx_for_msi(dev, 1);
4896 dev->msix_enabled = 0;
4897 }
4898 EXPORT_SYMBOL(pci_disable_msix);
4899 --- sle11-2009-10-16.orig/drivers/xen/blkback/blkback.c 2009-02-16 16:17:21.000000000 +0100
4900 +++ sle11-2009-10-16/drivers/xen/blkback/blkback.c 2009-02-16 16:18:36.000000000 +0100
4901 @@ -269,13 +269,10 @@ static void __end_block_io_op(pending_re
4902 }
4903 }
4904
4905 -static int end_block_io_op(struct bio *bio, unsigned int done, int error)
4906 +static void end_block_io_op(struct bio *bio, int error)
4907 {
4908 - if (bio->bi_size != 0)
4909 - return 1;
4910 __end_block_io_op(bio->bi_private, error);
4911 bio_put(bio);
4912 - return error;
4913 }
4914
4915
4916 --- sle11-2009-10-16.orig/drivers/xen/blkfront/blkfront.c 2009-03-24 10:12:03.000000000 +0100
4917 +++ sle11-2009-10-16/drivers/xen/blkfront/blkfront.c 2009-02-16 16:18:36.000000000 +0100
4918 @@ -233,7 +233,7 @@ static int setup_blkring(struct xenbus_d
4919 SHARED_RING_INIT(sring);
4920 FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
4921
4922 - memset(info->sg, 0, sizeof(info->sg));
4923 + sg_init_table(info->sg, BLKIF_MAX_SEGMENTS_PER_REQUEST);
4924
4925 err = xenbus_grant_ring(dev, virt_to_mfn(info->ring.sring));
4926 if (err < 0) {
4927 @@ -625,9 +625,8 @@ static int blkif_queue_request(struct re
4928
4929 ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg);
4930 BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);
4931 - for (i = 0; i < ring_req->nr_segments; ++i) {
4932 - sg = info->sg + i;
4933 - buffer_mfn = page_to_phys(sg->page) >> PAGE_SHIFT;
4934 + for_each_sg(info->sg, sg, ring_req->nr_segments, i) {
4935 + buffer_mfn = page_to_phys(sg_page(sg)) >> PAGE_SHIFT;
4936 fsect = sg->offset >> 9;
4937 lsect = fsect + (sg->length >> 9) - 1;
4938 /* install a grant reference. */
4939 --- sle11-2009-10-16.orig/drivers/xen/core/firmware.c 2009-10-28 14:55:06.000000000 +0100
4940 +++ sle11-2009-10-16/drivers/xen/core/firmware.c 2009-03-25 18:10:23.000000000 +0100
4941 @@ -1,4 +1,5 @@
4942 #include <linux/kernel.h>
4943 +#include <linux/string.h>
4944 #include <linux/errno.h>
4945 #include <linux/init.h>
4946 #include <linux/edd.h>
4947 --- sle11-2009-10-16.orig/drivers/xen/core/machine_kexec.c 2009-10-28 14:55:06.000000000 +0100
4948 +++ sle11-2009-10-16/drivers/xen/core/machine_kexec.c 2009-02-17 11:46:41.000000000 +0100
4949 @@ -29,6 +29,10 @@ void __init xen_machine_kexec_setup_reso
4950 int k = 0;
4951 int rc;
4952
4953 + if (strstr(boot_command_line, "crashkernel="))
4954 + printk(KERN_WARNING "Ignoring crashkernel command line, "
4955 + "parameter will be supplied by xen\n");
4956 +
4957 if (!is_initial_xendomain())
4958 return;
4959
4960 @@ -130,6 +134,13 @@ void __init xen_machine_kexec_setup_reso
4961 xen_max_nr_phys_cpus))
4962 goto err;
4963
4964 +#ifdef CONFIG_X86
4965 + if (xen_create_contiguous_region((unsigned long)&vmcoreinfo_note,
4966 + get_order(sizeof(vmcoreinfo_note)),
4967 + BITS_PER_LONG))
4968 + goto err;
4969 +#endif
4970 +
4971 return;
4972
4973 err:
4974 @@ -205,6 +216,13 @@ NORET_TYPE void machine_kexec(struct kim
4975 panic("KEXEC_CMD_kexec hypercall should not return\n");
4976 }
4977
4978 +#ifdef CONFIG_X86
4979 +unsigned long paddr_vmcoreinfo_note(void)
4980 +{
4981 + return virt_to_machine(&vmcoreinfo_note);
4982 +}
4983 +#endif
4984 +
4985 void machine_shutdown(void)
4986 {
4987 /* do nothing */
4988 --- sle11-2009-10-16.orig/drivers/xen/core/smpboot.c 2008-12-15 11:27:22.000000000 +0100
4989 +++ sle11-2009-10-16/drivers/xen/core/smpboot.c 2009-02-16 16:18:36.000000000 +0100
4990 @@ -45,8 +45,8 @@ cpumask_t cpu_possible_map;
4991 EXPORT_SYMBOL(cpu_possible_map);
4992 cpumask_t cpu_initialized_map;
4993
4994 -struct cpuinfo_x86 cpu_data[NR_CPUS] __cacheline_aligned;
4995 -EXPORT_SYMBOL(cpu_data);
4996 +DEFINE_PER_CPU(struct cpuinfo_x86, cpu_info);
4997 +EXPORT_PER_CPU_SYMBOL(cpu_info);
4998
4999 static DEFINE_PER_CPU(int, resched_irq);
5000 static DEFINE_PER_CPU(int, callfunc_irq);
5001 @@ -55,13 +55,13 @@ static char callfunc_name[NR_CPUS][15];
5002
5003 u8 cpu_2_logical_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
5004
5005 -cpumask_t cpu_sibling_map[NR_CPUS] __cacheline_aligned;
5006 -cpumask_t cpu_core_map[NR_CPUS] __cacheline_aligned;
5007 -EXPORT_SYMBOL(cpu_core_map);
5008 +DEFINE_PER_CPU(cpumask_t, cpu_sibling_map);
5009 +DEFINE_PER_CPU(cpumask_t, cpu_core_map);
5010 +EXPORT_PER_CPU_SYMBOL(cpu_core_map);
5011
5012 #if defined(__i386__)
5013 -u8 x86_cpu_to_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = 0xff };
5014 -EXPORT_SYMBOL(x86_cpu_to_apicid);
5015 +DEFINE_PER_CPU(u8, x86_cpu_to_apicid) = BAD_APICID;
5016 +EXPORT_PER_CPU_SYMBOL(x86_cpu_to_apicid);
5017 #endif
5018
5019 void __init prefill_possible_map(void)
5020 @@ -86,25 +86,25 @@ void __init smp_alloc_memory(void)
5021 static inline void
5022 set_cpu_sibling_map(unsigned int cpu)
5023 {
5024 - cpu_data[cpu].phys_proc_id = cpu;
5025 - cpu_data[cpu].cpu_core_id = 0;
5026 + cpu_data(cpu).phys_proc_id = cpu;
5027 + cpu_data(cpu).cpu_core_id = 0;
5028
5029 - cpu_sibling_map[cpu] = cpumask_of_cpu(cpu);
5030 - cpu_core_map[cpu] = cpumask_of_cpu(cpu);
5031 + per_cpu(cpu_sibling_map, cpu) = cpumask_of_cpu(cpu);
5032 + per_cpu(cpu_core_map, cpu) = cpumask_of_cpu(cpu);
5033
5034 - cpu_data[cpu].booted_cores = 1;
5035 + cpu_data(cpu).booted_cores = 1;
5036 }
5037
5038 static void
5039 remove_siblinginfo(unsigned int cpu)
5040 {
5041 - cpu_data[cpu].phys_proc_id = BAD_APICID;
5042 - cpu_data[cpu].cpu_core_id = BAD_APICID;
5043 + cpu_data(cpu).phys_proc_id = BAD_APICID;
5044 + cpu_data(cpu).cpu_core_id = BAD_APICID;
5045
5046 - cpus_clear(cpu_sibling_map[cpu]);
5047 - cpus_clear(cpu_core_map[cpu]);
5048 + cpus_clear(per_cpu(cpu_sibling_map, cpu));
5049 + cpus_clear(per_cpu(cpu_core_map, cpu));
5050
5051 - cpu_data[cpu].booted_cores = 0;
5052 + cpu_data(cpu).booted_cores = 0;
5053 }
5054
5055 static int __cpuinit xen_smp_intr_init(unsigned int cpu)
5056 @@ -163,9 +163,9 @@ void __cpuinit cpu_bringup(void)
5057 {
5058 cpu_init();
5059 #ifdef __i386__
5060 - identify_secondary_cpu(cpu_data + smp_processor_id());
5061 + identify_secondary_cpu(&current_cpu_data);
5062 #else
5063 - identify_cpu(cpu_data + smp_processor_id());
5064 + identify_cpu(&current_cpu_data);
5065 #endif
5066 touch_softlockup_watchdog();
5067 preempt_disable();
5068 @@ -266,16 +266,16 @@ void __init smp_prepare_cpus(unsigned in
5069 if (HYPERVISOR_vcpu_op(VCPUOP_get_physid, 0, &cpu_id) == 0)
5070 apicid = xen_vcpu_physid_to_x86_apicid(cpu_id.phys_id);
5071 boot_cpu_data.apicid = apicid;
5072 - cpu_data[0] = boot_cpu_data;
5073 + cpu_data(0) = boot_cpu_data;
5074
5075 cpu_2_logical_apicid[0] = apicid;
5076 - x86_cpu_to_apicid[0] = apicid;
5077 + per_cpu(x86_cpu_to_apicid, 0) = apicid;
5078
5079 current_thread_info()->cpu = 0;
5080
5081 for (cpu = 0; cpu < NR_CPUS; cpu++) {
5082 - cpus_clear(cpu_sibling_map[cpu]);
5083 - cpus_clear(cpu_core_map[cpu]);
5084 + cpus_clear(per_cpu(cpu_sibling_map, cpu));
5085 + cpus_clear(per_cpu(cpu_core_map, cpu));
5086 }
5087
5088 set_cpu_sibling_map(0);
5089 @@ -320,11 +320,12 @@ void __init smp_prepare_cpus(unsigned in
5090 apicid = cpu;
5091 if (HYPERVISOR_vcpu_op(VCPUOP_get_physid, cpu, &cpu_id) == 0)
5092 apicid = xen_vcpu_physid_to_x86_apicid(cpu_id.phys_id);
5093 - cpu_data[cpu] = boot_cpu_data;
5094 - cpu_data[cpu].apicid = apicid;
5095 + cpu_data(cpu) = boot_cpu_data;
5096 + cpu_data(cpu).cpu_index = cpu;
5097 + cpu_data(cpu).apicid = apicid;
5098
5099 cpu_2_logical_apicid[cpu] = apicid;
5100 - x86_cpu_to_apicid[cpu] = apicid;
5101 + per_cpu(x86_cpu_to_apicid, cpu) = apicid;
5102
5103 #ifdef __x86_64__
5104 cpu_pda(cpu)->pcurrent = idle;
5105 --- sle11-2009-10-16.orig/drivers/xen/netback/loopback.c 2008-12-15 11:26:44.000000000 +0100
5106 +++ sle11-2009-10-16/drivers/xen/netback/loopback.c 2009-02-16 16:18:36.000000000 +0100
5107 @@ -285,9 +285,9 @@ static void __exit clean_loopback(int i)
5108 char dev_name[IFNAMSIZ];
5109
5110 sprintf(dev_name, "vif0.%d", i);
5111 - dev1 = dev_get_by_name(dev_name);
5112 + dev1 = dev_get_by_name(&init_net, dev_name);
5113 sprintf(dev_name, "veth%d", i);
5114 - dev2 = dev_get_by_name(dev_name);
5115 + dev2 = dev_get_by_name(&init_net, dev_name);
5116 if (dev1 && dev2) {
5117 unregister_netdev(dev2);
5118 unregister_netdev(dev1);
5119 --- sle11-2009-10-16.orig/drivers/xen/netback/netback.c 2008-12-23 09:33:22.000000000 +0100
5120 +++ sle11-2009-10-16/drivers/xen/netback/netback.c 2009-02-16 16:18:36.000000000 +0100
5121 @@ -350,8 +350,8 @@ static void xen_network_done_notify(void
5122 {
5123 static struct net_device *eth0_dev = NULL;
5124 if (unlikely(eth0_dev == NULL))
5125 - eth0_dev = __dev_get_by_name("eth0");
5126 - netif_rx_schedule(eth0_dev);
5127 + eth0_dev = __dev_get_by_name(&init_net, "eth0");
5128 + netif_rx_schedule(eth0_dev, ???);
5129 }
5130 /*
5131 * Add following to poll() function in NAPI driver (Tigon3 is example):
5132 --- sle11-2009-10-16.orig/drivers/xen/netback/xenbus.c 2009-03-04 11:25:55.000000000 +0100
5133 +++ sle11-2009-10-16/drivers/xen/netback/xenbus.c 2009-02-16 16:18:36.000000000 +0100
5134 @@ -149,12 +149,10 @@ fail:
5135 * and vif variables to the environment, for the benefit of the vif-* hotplug
5136 * scripts.
5137 */
5138 -static int netback_uevent(struct xenbus_device *xdev, char **envp,
5139 - int num_envp, char *buffer, int buffer_size)
5140 +static int netback_uevent(struct xenbus_device *xdev, struct kobj_uevent_env *env)
5141 {
5142 struct backend_info *be = xdev->dev.driver_data;
5143 netif_t *netif = be->netif;
5144 - int i = 0, length = 0;
5145 char *val;
5146
5147 DPRINTK("netback_uevent");
5148 @@ -166,15 +164,11 @@ static int netback_uevent(struct xenbus_
5149 return err;
5150 }
5151 else {
5152 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size,
5153 - &length, "script=%s", val);
5154 + add_uevent_var(env, "script=%s", val);
5155 kfree(val);
5156 }
5157
5158 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5159 - "vif=%s", netif->dev->name);
5160 -
5161 - envp[i] = NULL;
5162 + add_uevent_var(env, "vif=%s", netif->dev->name);
5163
5164 return 0;
5165 }
5166 --- sle11-2009-10-16.orig/drivers/xen/netfront/accel.c 2009-04-09 14:43:45.000000000 +0200
5167 +++ sle11-2009-10-16/drivers/xen/netfront/accel.c 2009-03-30 16:39:19.000000000 +0200
5168 @@ -313,7 +313,7 @@ accelerator_set_vif_state_hooks(struct n
5169 DPRINTK("%p\n",vif_state);
5170
5171 /* Make sure there are no data path operations going on */
5172 - netif_poll_disable(vif_state->np->netdev);
5173 + napi_disable(&vif_state->np->napi);
5174 netif_tx_lock_bh(vif_state->np->netdev);
5175
5176 accelerator = vif_state->np->accelerator;
5177 @@ -322,7 +322,7 @@ accelerator_set_vif_state_hooks(struct n
5178 spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
5179
5180 netif_tx_unlock_bh(vif_state->np->netdev);
5181 - netif_poll_enable(vif_state->np->netdev);
5182 + napi_enable(&vif_state->np->napi);
5183 }
5184
5185
5186 @@ -496,7 +496,7 @@ accelerator_remove_single_hook(struct ne
5187 unsigned long flags;
5188
5189 /* Make sure there are no data path operations going on */
5190 - netif_poll_disable(vif_state->np->netdev);
5191 + napi_disable(&vif_state->np->napi);
5192 netif_tx_lock_bh(vif_state->np->netdev);
5193
5194 spin_lock_irqsave(&accelerator->vif_states_lock, flags);
5195 @@ -512,7 +512,7 @@ accelerator_remove_single_hook(struct ne
5196 spin_unlock_irqrestore(&accelerator->vif_states_lock, flags);
5197
5198 netif_tx_unlock_bh(vif_state->np->netdev);
5199 - netif_poll_enable(vif_state->np->netdev);
5200 + napi_enable(&vif_state->np->napi);
5201 }
5202
5203
5204 --- sle11-2009-10-16.orig/drivers/xen/netfront/netfront.c 2009-03-30 16:36:30.000000000 +0200
5205 +++ sle11-2009-10-16/drivers/xen/netfront/netfront.c 2009-03-30 16:39:44.000000000 +0200
5206 @@ -626,6 +626,7 @@ static int network_open(struct net_devic
5207 struct netfront_info *np = netdev_priv(dev);
5208
5209 memset(&np->stats, 0, sizeof(np->stats));
5210 + napi_enable(&np->napi);
5211
5212 spin_lock_bh(&np->rx_lock);
5213 if (netfront_carrier_ok(np)) {
5214 @@ -634,7 +635,7 @@ static int network_open(struct net_devic
5215 if (RING_HAS_UNCONSUMED_RESPONSES(&np->rx)){
5216 netfront_accelerator_call_stop_napi_irq(np, dev);
5217
5218 - netif_rx_schedule(dev);
5219 + netif_rx_schedule(dev, &np->napi);
5220 }
5221 }
5222 spin_unlock_bh(&np->rx_lock);
5223 @@ -706,7 +707,7 @@ static void rx_refill_timeout(unsigned l
5224
5225 netfront_accelerator_call_stop_napi_irq(np, dev);
5226
5227 - netif_rx_schedule(dev);
5228 + netif_rx_schedule(dev, &np->napi);
5229 }
5230
5231 static void network_alloc_rx_buffers(struct net_device *dev)
5232 @@ -1063,7 +1064,7 @@ static irqreturn_t netif_int(int irq, vo
5233 if (RING_HAS_UNCONSUMED_RESPONSES(&np->rx)) {
5234 netfront_accelerator_call_stop_napi_irq(np, dev);
5235
5236 - netif_rx_schedule(dev);
5237 + netif_rx_schedule(dev, &np->napi);
5238 dev->last_rx = jiffies;
5239 }
5240 }
5241 @@ -1316,16 +1317,17 @@ static int xennet_set_skb_gso(struct sk_
5242 #endif
5243 }
5244
5245 -static int netif_poll(struct net_device *dev, int *pbudget)
5246 +static int netif_poll(struct napi_struct *napi, int budget)
5247 {
5248 - struct netfront_info *np = netdev_priv(dev);
5249 + struct netfront_info *np = container_of(napi, struct netfront_info, napi);
5250 + struct net_device *dev = np->netdev;
5251 struct sk_buff *skb;
5252 struct netfront_rx_info rinfo;
5253 struct netif_rx_response *rx = &rinfo.rx;
5254 struct netif_extra_info *extras = rinfo.extras;
5255 RING_IDX i, rp;
5256 struct multicall_entry *mcl;
5257 - int work_done, budget, more_to_do = 1, accel_more_to_do = 1;
5258 + int work_done, more_to_do = 1, accel_more_to_do = 1;
5259 struct sk_buff_head rxq;
5260 struct sk_buff_head errq;
5261 struct sk_buff_head tmpq;
5262 @@ -1345,8 +1347,6 @@ static int netif_poll(struct net_device
5263 skb_queue_head_init(&errq);
5264 skb_queue_head_init(&tmpq);
5265
5266 - if ((budget = *pbudget) > dev->quota)
5267 - budget = dev->quota;
5268 rp = np->rx.sring->rsp_prod;
5269 rmb(); /* Ensure we see queued responses up to 'rp'. */
5270
5271 @@ -1508,9 +1508,6 @@ err:
5272 accel_more_to_do = 0;
5273 }
5274
5275 - *pbudget -= work_done;
5276 - dev->quota -= work_done;
5277 -
5278 if (work_done < budget) {
5279 local_irq_save(flags);
5280
5281 @@ -1527,14 +1524,14 @@ err:
5282 }
5283
5284 if (!more_to_do && !accel_more_to_do)
5285 - __netif_rx_complete(dev);
5286 + __netif_rx_complete(dev, napi);
5287
5288 local_irq_restore(flags);
5289 }
5290
5291 spin_unlock(&np->rx_lock);
5292
5293 - return more_to_do | accel_more_to_do;
5294 + return work_done;
5295 }
5296
5297 static void netif_release_tx_bufs(struct netfront_info *np)
5298 @@ -1681,6 +1678,7 @@ static int network_close(struct net_devi
5299 {
5300 struct netfront_info *np = netdev_priv(dev);
5301 netif_stop_queue(np->netdev);
5302 + napi_disable(&np->napi);
5303 return 0;
5304 }
5305
5306 @@ -2088,16 +2086,14 @@ static struct net_device * __devinit cre
5307 netdev->hard_start_xmit = network_start_xmit;
5308 netdev->stop = network_close;
5309 netdev->get_stats = network_get_stats;
5310 - netdev->poll = netif_poll;
5311 + netif_napi_add(netdev, &np->napi, netif_poll, 64);
5312 netdev->set_multicast_list = network_set_multicast_list;
5313 netdev->uninit = netif_uninit;
5314 netdev->set_mac_address = xennet_set_mac_address;
5315 netdev->change_mtu = xennet_change_mtu;
5316 - netdev->weight = 64;
5317 netdev->features = NETIF_F_IP_CSUM;
5318
5319 SET_ETHTOOL_OPS(netdev, &network_ethtool_ops);
5320 - SET_MODULE_OWNER(netdev);
5321 SET_NETDEV_DEV(netdev, &dev->dev);
5322
5323 np->netdev = netdev;
5324 --- sle11-2009-10-16.orig/drivers/xen/netfront/netfront.h 2009-10-28 14:55:06.000000000 +0100
5325 +++ sle11-2009-10-16/drivers/xen/netfront/netfront.h 2009-02-16 16:18:36.000000000 +0100
5326 @@ -157,6 +157,8 @@ struct netfront_info {
5327 spinlock_t tx_lock;
5328 spinlock_t rx_lock;
5329
5330 + struct napi_struct napi;
5331 +
5332 unsigned int irq;
5333 unsigned int copying_receiver;
5334 unsigned int carrier;
5335 --- sle11-2009-10-16.orig/drivers/xen/pciback/Makefile 2009-10-28 14:55:06.000000000 +0100
5336 +++ sle11-2009-10-16/drivers/xen/pciback/Makefile 2009-02-16 16:18:36.000000000 +0100
5337 @@ -12,6 +12,4 @@ pciback-$(CONFIG_XEN_PCIDEV_BACKEND_SLOT
5338 pciback-$(CONFIG_XEN_PCIDEV_BACKEND_PASS) += passthrough.o
5339 pciback-$(CONFIG_XEN_PCIDEV_BACKEND_CONTROLLER) += controller.o
5340
5341 -ifeq ($(CONFIG_XEN_PCIDEV_BE_DEBUG),y)
5342 -EXTRA_CFLAGS += -DDEBUG
5343 -endif
5344 +ccflags-$(CONFIG_XEN_PCIDEV_BE_DEBUG) += -DDEBUG
5345 --- sle11-2009-10-16.orig/drivers/xen/pcifront/Makefile 2009-10-28 14:55:06.000000000 +0100
5346 +++ sle11-2009-10-16/drivers/xen/pcifront/Makefile 2009-02-16 16:18:36.000000000 +0100
5347 @@ -2,6 +2,4 @@ obj-y += pcifront.o
5348
5349 pcifront-y := pci_op.o xenbus.o pci.o
5350
5351 -ifeq ($(CONFIG_XEN_PCIDEV_FE_DEBUG),y)
5352 -EXTRA_CFLAGS += -DDEBUG
5353 -endif
5354 +ccflags-$(CONFIG_XEN_PCIDEV_FE_DEBUG) += -DDEBUG
5355 --- sle11-2009-10-16.orig/drivers/xen/scsiback/emulate.c 2009-10-28 14:55:06.000000000 +0100
5356 +++ sle11-2009-10-16/drivers/xen/scsiback/emulate.c 2009-02-16 16:18:36.000000000 +0100
5357 @@ -104,9 +104,10 @@ static void resp_not_supported_cmd(pendi
5358 }
5359
5360
5361 -static int __copy_to_sg(struct scatterlist *sg, unsigned int nr_sg,
5362 +static int __copy_to_sg(struct scatterlist *sgl, unsigned int nr_sg,
5363 void *buf, unsigned int buflen)
5364 {
5365 + struct scatterlist *sg;
5366 void *from = buf;
5367 void *to;
5368 unsigned int from_rest = buflen;
5369 @@ -115,8 +116,8 @@ static int __copy_to_sg(struct scatterli
5370 unsigned int i;
5371 unsigned long pfn;
5372
5373 - for (i = 0; i < nr_sg; i++) {
5374 - if (sg->page == NULL) {
5375 + for_each_sg (sgl, sg, nr_sg, i) {
5376 + if (sg_page(sg) == NULL) {
5377 printk(KERN_WARNING "%s: inconsistent length field in "
5378 "scatterlist\n", __FUNCTION__);
5379 return -ENOMEM;
5380 @@ -125,7 +126,7 @@ static int __copy_to_sg(struct scatterli
5381 to_capa = sg->length;
5382 copy_size = min_t(unsigned int, to_capa, from_rest);
5383
5384 - pfn = page_to_pfn(sg->page);
5385 + pfn = page_to_pfn(sg_page(sg));
5386 to = pfn_to_kaddr(pfn) + (sg->offset);
5387 memcpy(to, from, copy_size);
5388
5389 @@ -134,7 +135,6 @@ static int __copy_to_sg(struct scatterli
5390 return 0;
5391 }
5392
5393 - sg++;
5394 from += copy_size;
5395 }
5396
5397 @@ -143,9 +143,10 @@ static int __copy_to_sg(struct scatterli
5398 return -ENOMEM;
5399 }
5400
5401 -static int __copy_from_sg(struct scatterlist *sg, unsigned int nr_sg,
5402 +static int __copy_from_sg(struct scatterlist *sgl, unsigned int nr_sg,
5403 void *buf, unsigned int buflen)
5404 {
5405 + struct scatterlist *sg;
5406 void *from;
5407 void *to = buf;
5408 unsigned int from_rest;
5409 @@ -154,8 +155,8 @@ static int __copy_from_sg(struct scatter
5410 unsigned int i;
5411 unsigned long pfn;
5412
5413 - for (i = 0; i < nr_sg; i++) {
5414 - if (sg->page == NULL) {
5415 + for_each_sg (sgl, sg, nr_sg, i) {
5416 + if (sg_page(sg) == NULL) {
5417 printk(KERN_WARNING "%s: inconsistent length field in "
5418 "scatterlist\n", __FUNCTION__);
5419 return -ENOMEM;
5420 @@ -170,13 +171,11 @@ static int __copy_from_sg(struct scatter
5421 }
5422 copy_size = from_rest;
5423
5424 - pfn = page_to_pfn(sg->page);
5425 + pfn = page_to_pfn(sg_page(sg));
5426 from = pfn_to_kaddr(pfn) + (sg->offset);
5427 memcpy(to, from, copy_size);
5428
5429 to_capa -= copy_size;
5430 -
5431 - sg++;
5432 to += copy_size;
5433 }
5434
5435 --- sle11-2009-10-16.orig/drivers/xen/scsiback/scsiback.c 2008-12-15 11:26:44.000000000 +0100
5436 +++ sle11-2009-10-16/drivers/xen/scsiback/scsiback.c 2009-02-16 16:18:36.000000000 +0100
5437 @@ -247,6 +247,8 @@ static int scsiback_gnttab_data_map(vscs
5438 write = (data_dir == DMA_TO_DEVICE);
5439
5440 if (nr_segments) {
5441 + struct scatterlist *sg;
5442 +
5443 /* free of (sgl) in fast_flush_area()*/
5444 pending_req->sgl = kmalloc(sizeof(struct scatterlist) * nr_segments,
5445 GFP_KERNEL);
5446 @@ -255,6 +257,8 @@ static int scsiback_gnttab_data_map(vscs
5447 return -ENOMEM;
5448 }
5449
5450 + sg_init_table(pending_req->sgl, nr_segments);
5451 +
5452 for (i = 0; i < nr_segments; i++) {
5453 flags = GNTMAP_host_map;
5454 if (write)
5455 @@ -267,7 +271,7 @@ static int scsiback_gnttab_data_map(vscs
5456 err = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nr_segments);
5457 BUG_ON(err);
5458
5459 - for (i = 0; i < nr_segments; i++) {
5460 + for_each_sg (pending_req->sgl, sg, nr_segments, i) {
5461 if (unlikely(map[i].status != 0)) {
5462 printk(KERN_ERR "scsiback: invalid buffer -- could not remap it\n");
5463 map[i].handle = SCSIBACK_INVALID_HANDLE;
5464 @@ -283,15 +287,15 @@ static int scsiback_gnttab_data_map(vscs
5465 pending_req, i)) >> PAGE_SHIFT,
5466 FOREIGN_FRAME(map[i].dev_bus_addr >> PAGE_SHIFT));
5467
5468 - pending_req->sgl[i].page = virt_to_page(vaddr(pending_req, i));
5469 - pending_req->sgl[i].offset = ring_req->seg[i].offset;
5470 - pending_req->sgl[i].length = ring_req->seg[i].length;
5471 - data_len += pending_req->sgl[i].length;
5472 + sg_set_page(sg, virt_to_page(vaddr(pending_req, i)),
5473 + ring_req->seg[i].length,
5474 + ring_req->seg[i].offset);
5475 + data_len += sg->length;
5476
5477 barrier();
5478 - if (pending_req->sgl[i].offset >= PAGE_SIZE ||
5479 - pending_req->sgl[i].length > PAGE_SIZE ||
5480 - pending_req->sgl[i].offset + pending_req->sgl[i].length > PAGE_SIZE)
5481 + if (sg->offset >= PAGE_SIZE ||
5482 + sg->length > PAGE_SIZE ||
5483 + sg->offset + sg->length > PAGE_SIZE)
5484 err |= 1;
5485
5486 }
5487 @@ -320,27 +324,14 @@ static int scsiback_merge_bio(struct req
5488
5489 blk_queue_bounce(q, &bio);
5490
5491 - if (!rq->bio)
5492 - blk_rq_bio_prep(q, rq, bio);
5493 - else if (!ll_back_merge_fn(q, rq, bio))
5494 - return -EINVAL;
5495 - else {
5496 - rq->biotail->bi_next = bio;
5497 - rq->biotail = bio;
5498 - }
5499 -
5500 - return 0;
5501 + return blk_rq_append_bio(q, rq, bio);
5502 }
5503
5504
5505 /* quoted scsi_lib.c/scsi_bi_endio */
5506 -static int scsiback_bi_endio(struct bio *bio, unsigned int bytes_done, int error)
5507 +static void scsiback_bi_endio(struct bio *bio, int error)
5508 {
5509 - if (bio->bi_size)
5510 - return 1;
5511 -
5512 bio_put(bio);
5513 - return 0;
5514 }
5515
5516
5517 @@ -351,16 +342,16 @@ static int request_map_sg(struct request
5518 struct request_queue *q = rq->q;
5519 int nr_pages;
5520 unsigned int nsegs = count;
5521 -
5522 unsigned int data_len = 0, len, bytes, off;
5523 + struct scatterlist *sg;
5524 struct page *page;
5525 struct bio *bio = NULL;
5526 int i, err, nr_vecs = 0;
5527
5528 - for (i = 0; i < nsegs; i++) {
5529 - page = pending_req->sgl[i].page;
5530 - off = (unsigned int)pending_req->sgl[i].offset;
5531 - len = (unsigned int)pending_req->sgl[i].length;
5532 + for_each_sg (pending_req->sgl, sg, nsegs, i) {
5533 + page = sg_page(sg);
5534 + off = sg->offset;
5535 + len = sg->length;
5536 data_len += len;
5537
5538 nr_pages = (len + off + PAGE_SIZE - 1) >> PAGE_SHIFT;
5539 @@ -388,7 +379,7 @@ static int request_map_sg(struct request
5540 if (bio->bi_vcnt >= nr_vecs) {
5541 err = scsiback_merge_bio(rq, bio);
5542 if (err) {
5543 - bio_endio(bio, bio->bi_size, 0);
5544 + bio_endio(bio, 0);
5545 goto free_bios;
5546 }
5547 bio = NULL;
5548 @@ -411,7 +402,7 @@ free_bios:
5549 /*
5550 * call endio instead of bio_put incase it was bounced
5551 */
5552 - bio_endio(bio, bio->bi_size, 0);
5553 + bio_endio(bio, 0);
5554 }
5555
5556 return err;
5557 --- sle11-2009-10-16.orig/drivers/xen/scsifront/scsifront.c 2009-02-16 16:17:21.000000000 +0100
5558 +++ sle11-2009-10-16/drivers/xen/scsifront/scsifront.c 2009-02-16 16:18:36.000000000 +0100
5559 @@ -246,11 +246,10 @@ static int map_data_for_request(struct v
5560 {
5561 grant_ref_t gref_head;
5562 struct page *page;
5563 - int err, i, ref, ref_cnt = 0;
5564 + int err, ref, ref_cnt = 0;
5565 int write = (sc->sc_data_direction == DMA_TO_DEVICE);
5566 - int nr_pages, off, len, bytes;
5567 + unsigned int i, nr_pages, off, len, bytes;
5568 unsigned long buffer_pfn;
5569 - unsigned int data_len = 0;
5570
5571 if (sc->sc_data_direction == DMA_NONE)
5572 return 0;
5573 @@ -263,25 +262,31 @@ static int map_data_for_request(struct v
5574
5575 if (sc->use_sg) {
5576 /* quoted scsi_lib.c/scsi_req_map_sg . */
5577 - struct scatterlist *sg = (struct scatterlist *)sc->request_buffer;
5578 - nr_pages = (sc->request_bufflen + sg[0].offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
5579 + struct scatterlist *sg, *sgl = (struct scatterlist *)sc->request_buffer;
5580 + unsigned int data_len = sc->request_bufflen;
5581
5582 + nr_pages = (sc->request_bufflen + sgl->offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
5583 if (nr_pages > VSCSIIF_SG_TABLESIZE) {
5584 printk(KERN_ERR "scsifront: Unable to map request_buffer for command!\n");
5585 ref_cnt = (-E2BIG);
5586 goto big_to_sg;
5587 }
5588
5589 - for (i = 0; i < sc->use_sg; i++) {
5590 - page = sg[i].page;
5591 - off = sg[i].offset;
5592 - len = sg[i].length;
5593 - data_len += len;
5594 + for_each_sg (sgl, sg, sc->use_sg, i) {
5595 + page = sg_page(sg);
5596 + off = sg->offset;
5597 + len = sg->length;
5598
5599 buffer_pfn = page_to_phys(page) >> PAGE_SHIFT;
5600
5601 - while (len > 0) {
5602 + while (len > 0 && data_len > 0) {
5603 + /*
5604 + * sg sends a scatterlist that is larger than
5605 + * the data_len it wants transferred for certain
5606 + * IO sizes
5607 + */
5608 bytes = min_t(unsigned int, len, PAGE_SIZE - off);
5609 + bytes = min(bytes, data_len);
5610
5611 ref = gnttab_claim_grant_reference(&gref_head);
5612 BUG_ON(ref == -ENOSPC);
5613 @@ -296,6 +301,7 @@ static int map_data_for_request(struct v
5614
5615 buffer_pfn++;
5616 len -= bytes;
5617 + data_len -= bytes;
5618 off = 0;
5619 ref_cnt++;
5620 }
5621 --- sle11-2009-10-16.orig/drivers/xen/sfc_netback/accel_fwd.c 2008-12-15 11:27:22.000000000 +0100
5622 +++ sle11-2009-10-16/drivers/xen/sfc_netback/accel_fwd.c 2009-02-16 16:18:36.000000000 +0100
5623 @@ -181,10 +181,11 @@ int netback_accel_fwd_add(const __u8 *ma
5624 unsigned long flags;
5625 cuckoo_hash_mac_key key = cuckoo_mac_to_key(mac);
5626 struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
5627 + DECLARE_MAC_BUF(buf);
5628
5629 BUG_ON(fwd_priv == NULL);
5630
5631 - DPRINTK("Adding mac " MAC_FMT "\n", MAC_ARG(mac));
5632 + DPRINTK("Adding mac %s\n", print_mac(buf, mac));
5633
5634 spin_lock_irqsave(&fwd_set->fwd_lock, flags);
5635
5636 @@ -199,8 +200,8 @@ int netback_accel_fwd_add(const __u8 *ma
5637 if (cuckoo_hash_lookup(&fwd_set->fwd_hash_table,
5638 (cuckoo_hash_key *)(&key), &rc) != 0) {
5639 spin_unlock_irqrestore(&fwd_set->fwd_lock, flags);
5640 - EPRINTK("MAC address " MAC_FMT " already accelerated.\n",
5641 - MAC_ARG(mac));
5642 + EPRINTK("MAC address %s already accelerated.\n",
5643 + print_mac(buf, mac));
5644 return -EEXIST;
5645 }
5646
5647 @@ -235,8 +236,9 @@ void netback_accel_fwd_remove(const __u8
5648 unsigned long flags;
5649 cuckoo_hash_mac_key key = cuckoo_mac_to_key(mac);
5650 struct port_fwd *fwd_set = (struct port_fwd *)fwd_priv;
5651 + DECLARE_MAC_BUF(buf);
5652
5653 - DPRINTK("Removing mac " MAC_FMT "\n", MAC_ARG(mac));
5654 + DPRINTK("Removing mac %s\n", print_mac(buf, mac));
5655
5656 BUG_ON(fwd_priv == NULL);
5657
5658 @@ -394,14 +396,16 @@ void netback_accel_tx_packet(struct sk_b
5659
5660 if (is_broadcast_ether_addr(skb_mac_header(skb))
5661 && packet_is_arp_reply(skb)) {
5662 + DECLARE_MAC_BUF(buf);
5663 +
5664 /*
5665 * update our fast path forwarding to reflect this
5666 * gratuitous ARP
5667 */
5668 mac = skb_mac_header(skb)+ETH_ALEN;
5669
5670 - DPRINTK("%s: found gratuitous ARP for " MAC_FMT "\n",
5671 - __FUNCTION__, MAC_ARG(mac));
5672 + DPRINTK("%s: found gratuitous ARP for %s\n",
5673 + __FUNCTION__, print_mac(buf, mac));
5674
5675 spin_lock_irqsave(&fwd_set->fwd_lock, flags);
5676 /*
5677 --- sle11-2009-10-16.orig/drivers/xen/sfc_netback/accel_msg.c 2009-10-28 14:55:06.000000000 +0100
5678 +++ sle11-2009-10-16/drivers/xen/sfc_netback/accel_msg.c 2009-02-16 16:18:36.000000000 +0100
5679 @@ -57,11 +57,11 @@ static void netback_accel_msg_tx_localma
5680 {
5681 unsigned long lock_state;
5682 struct net_accel_msg *msg;
5683 + DECLARE_MAC_BUF(buf);
5684
5685 BUG_ON(bend == NULL || mac == NULL);
5686
5687 - VPRINTK("Sending local mac message: " MAC_FMT "\n",
5688 - MAC_ARG((const char *)mac));
5689 + VPRINTK("Sending local mac message: %s\n", print_mac(buf, mac));
5690
5691 msg = net_accel_msg_start_send(bend->shared_page, &bend->to_domU,
5692 &lock_state);
5693 --- sle11-2009-10-16.orig/drivers/xen/sfc_netfront/accel_msg.c 2009-03-04 11:28:34.000000000 +0100
5694 +++ sle11-2009-10-16/drivers/xen/sfc_netfront/accel_msg.c 2009-02-16 16:18:36.000000000 +0100
5695 @@ -41,11 +41,13 @@ static void vnic_start_interrupts(netfro
5696 /* Prime our interrupt */
5697 spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
5698 if (!netfront_accel_vi_enable_interrupts(vnic)) {
5699 + struct netfront_info *np = netdev_priv(vnic->net_dev);
5700 +
5701 /* Cripes, that was quick, better pass it up */
5702 netfront_accel_disable_net_interrupts(vnic);
5703 vnic->irq_enabled = 0;
5704 NETFRONT_ACCEL_STATS_OP(vnic->stats.poll_schedule_count++);
5705 - netif_rx_schedule(vnic->net_dev);
5706 + netif_rx_schedule(vnic->net_dev, &np->napi);
5707 } else {
5708 /*
5709 * Nothing yet, make sure we get interrupts through
5710 @@ -72,6 +74,7 @@ static void vnic_stop_interrupts(netfron
5711 static void vnic_start_fastpath(netfront_accel_vnic *vnic)
5712 {
5713 struct net_device *net_dev = vnic->net_dev;
5714 + struct netfront_info *np = netdev_priv(net_dev);
5715 unsigned long flags;
5716
5717 DPRINTK("%s\n", __FUNCTION__);
5718 @@ -80,9 +83,9 @@ static void vnic_start_fastpath(netfront
5719 vnic->tx_enabled = 1;
5720 spin_unlock_irqrestore(&vnic->tx_lock, flags);
5721
5722 - netif_poll_disable(net_dev);
5723 + napi_disable(&np->napi);
5724 vnic->poll_enabled = 1;
5725 - netif_poll_enable(net_dev);
5726 + napi_enable(&np->napi);
5727
5728 vnic_start_interrupts(vnic);
5729 }
5730 @@ -114,11 +117,11 @@ void vnic_stop_fastpath(netfront_accel_v
5731 spin_unlock_irqrestore(&vnic->tx_lock, flags1);
5732
5733 /* Must prevent polls and hold lock to modify poll_enabled */
5734 - netif_poll_disable(net_dev);
5735 + napi_disable(&np->napi);
5736 spin_lock_irqsave(&vnic->irq_enabled_lock, flags1);
5737 vnic->poll_enabled = 0;
5738 spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags1);
5739 - netif_poll_enable(net_dev);
5740 + napi_enable(&np->napi);
5741 }
5742
5743
5744 @@ -326,8 +329,10 @@ static int vnic_process_localmac_msg(net
5745 cuckoo_hash_mac_key key;
5746
5747 if (msg->u.localmac.flags & NET_ACCEL_MSG_ADD) {
5748 - DPRINTK("MAC has moved, could be local: " MAC_FMT "\n",
5749 - MAC_ARG(msg->u.localmac.mac));
5750 + DECLARE_MAC_BUF(buf);
5751 +
5752 + DPRINTK("MAC has moved, could be local: %s\n",
5753 + print_mac(buf, msg->u.localmac.mac));
5754 key = cuckoo_mac_to_key(msg->u.localmac.mac);
5755 spin_lock_irqsave(&vnic->table_lock, flags);
5756 /* Try to remove it, not a big deal if not there */
5757 @@ -515,6 +520,8 @@ irqreturn_t netfront_accel_net_channel_i
5758
5759 spin_lock_irqsave(&vnic->irq_enabled_lock, flags);
5760 if (vnic->irq_enabled) {
5761 + struct netfront_info *np = netdev_priv(net_dev);
5762 +
5763 netfront_accel_disable_net_interrupts(vnic);
5764 vnic->irq_enabled = 0;
5765 spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
5766 @@ -527,7 +534,7 @@ irqreturn_t netfront_accel_net_channel_i
5767 vnic->stats.event_count_since_irq;
5768 vnic->stats.event_count_since_irq = 0;
5769 #endif
5770 - netif_rx_schedule(net_dev);
5771 + netif_rx_schedule(net_dev, &np->napi);
5772 }
5773 else {
5774 spin_unlock_irqrestore(&vnic->irq_enabled_lock, flags);
5775 --- sle11-2009-10-16.orig/drivers/xen/sfc_netfront/accel_vi.c 2009-03-30 16:36:26.000000000 +0200
5776 +++ sle11-2009-10-16/drivers/xen/sfc_netfront/accel_vi.c 2009-03-30 16:39:38.000000000 +0200
5777 @@ -641,8 +641,10 @@ netfront_accel_vi_tx_post(netfront_accel
5778 (cuckoo_hash_key *)(&key), &value);
5779
5780 if (!try_fastpath) {
5781 - VPRINTK("try fast path false for mac: " MAC_FMT "\n",
5782 - MAC_ARG(skb->data));
5783 + DECLARE_MAC_BUF(buf);
5784 +
5785 + VPRINTK("try fast path false for mac: %s\n",
5786 + print_mac(buf, skb->data));
5787
5788 return NETFRONT_ACCEL_STATUS_CANT;
5789 }
5790 @@ -768,9 +770,10 @@ static void netfront_accel_vi_rx_comple
5791 if (compare_ether_addr(skb->data, vnic->mac)) {
5792 struct iphdr *ip = (struct iphdr *)(skb->data + ETH_HLEN);
5793 u16 port;
5794 + DECLARE_MAC_BUF(buf);
5795
5796 - DPRINTK("%s: saw wrong MAC address " MAC_FMT "\n",
5797 - __FUNCTION__, MAC_ARG(skb->data));
5798 + DPRINTK("%s: saw wrong MAC address %s\n",
5799 + __FUNCTION__, print_mac(buf, skb->data));
5800
5801 if (ip->protocol == IPPROTO_TCP) {
5802 struct tcphdr *tcp = (struct tcphdr *)
5803 --- sle11-2009-10-16.orig/drivers/xen/sfc_netutil/accel_util.h 2009-10-28 14:55:06.000000000 +0100
5804 +++ sle11-2009-10-16/drivers/xen/sfc_netutil/accel_util.h 2009-02-16 16:18:36.000000000 +0100
5805 @@ -63,9 +63,6 @@
5806 DPRINTK("%s at %s:%d\n", #exp, __FILE__, __LINE__); \
5807 } while(0)
5808
5809 -#define MAC_FMT "%.2x:%.2x:%.2x:%.2x:%.2x:%.2x"
5810 -#define MAC_ARG(_mac) (_mac)[0], (_mac)[1], (_mac)[2], (_mac)[3], (_mac)[4], (_mac)[5]
5811 -
5812 #include <xen/xenbus.h>
5813
5814 /*! Map a set of pages from another domain
5815 --- sle11-2009-10-16.orig/drivers/xen/xenbus/xenbus_probe.c 2009-02-16 16:17:21.000000000 +0100
5816 +++ sle11-2009-10-16/drivers/xen/xenbus/xenbus_probe.c 2009-02-16 16:18:36.000000000 +0100
5817 @@ -174,11 +174,9 @@ static int read_backend_details(struct x
5818 }
5819
5820 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) && (defined(CONFIG_XEN) || defined(MODULE))
5821 -static int xenbus_uevent_frontend(struct device *dev, char **envp,
5822 - int num_envp, char *buffer, int buffer_size)
5823 +static int xenbus_uevent_frontend(struct device *dev, struct kobj_uevent_env *env)
5824 {
5825 struct xenbus_device *xdev;
5826 - int length = 0, i = 0;
5827
5828 if (dev == NULL)
5829 return -ENODEV;
5830 @@ -187,12 +185,9 @@ static int xenbus_uevent_frontend(struct
5831 return -ENODEV;
5832
5833 /* stuff we want to pass to /sbin/hotplug */
5834 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5835 - "XENBUS_TYPE=%s", xdev->devicetype);
5836 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5837 - "XENBUS_PATH=%s", xdev->nodename);
5838 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5839 - "MODALIAS=xen:%s", xdev->devicetype);
5840 + add_uevent_var(env, "XENBUS_TYPE=%s", xdev->devicetype);
5841 + add_uevent_var(env, "XENBUS_PATH=%s", xdev->nodename);
5842 + add_uevent_var(env, "MODALIAS=xen:%s", xdev->devicetype);
5843
5844 return 0;
5845 }
5846 --- sle11-2009-10-16.orig/drivers/xen/xenbus/xenbus_probe_backend.c 2009-02-16 16:17:21.000000000 +0100
5847 +++ sle11-2009-10-16/drivers/xen/xenbus/xenbus_probe_backend.c 2009-02-16 16:18:36.000000000 +0100
5848 @@ -60,8 +60,7 @@
5849 #include <xen/platform-compat.h>
5850 #endif
5851
5852 -static int xenbus_uevent_backend(struct device *dev, char **envp,
5853 - int num_envp, char *buffer, int buffer_size);
5854 +static int xenbus_uevent_backend(struct device *dev, struct kobj_uevent_env *env);
5855 static int xenbus_probe_backend(const char *type, const char *domid);
5856
5857 extern int read_otherend_details(struct xenbus_device *xendev,
5858 @@ -128,13 +127,10 @@ static struct xen_bus_type xenbus_backen
5859 },
5860 };
5861
5862 -static int xenbus_uevent_backend(struct device *dev, char **envp,
5863 - int num_envp, char *buffer, int buffer_size)
5864 +static int xenbus_uevent_backend(struct device *dev, struct kobj_uevent_env *env)
5865 {
5866 struct xenbus_device *xdev;
5867 struct xenbus_driver *drv;
5868 - int i = 0;
5869 - int length = 0;
5870
5871 DPRINTK("");
5872
5873 @@ -146,27 +142,16 @@ static int xenbus_uevent_backend(struct
5874 return -ENODEV;
5875
5876 /* stuff we want to pass to /sbin/hotplug */
5877 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5878 - "XENBUS_TYPE=%s", xdev->devicetype);
5879 + add_uevent_var(env, "XENBUS_TYPE=%s", xdev->devicetype);
5880
5881 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5882 - "XENBUS_PATH=%s", xdev->nodename);
5883 + add_uevent_var(env, "XENBUS_PATH=%s", xdev->nodename);
5884
5885 - add_uevent_var(envp, num_envp, &i, buffer, buffer_size, &length,
5886 - "XENBUS_BASE_PATH=%s", xenbus_backend.root);
5887 -
5888 - /* terminate, set to next free slot, shrink available space */
5889 - envp[i] = NULL;
5890 - envp = &envp[i];
5891 - num_envp -= i;
5892 - buffer = &buffer[length];
5893 - buffer_size -= length;
5894 + add_uevent_var(env, "XENBUS_BASE_PATH=%s", xenbus_backend.root);
5895
5896 if (dev->driver) {
5897 drv = to_xenbus_driver(dev->driver);
5898 if (drv && drv->uevent)
5899 - return drv->uevent(xdev, envp, num_envp, buffer,
5900 - buffer_size);
5901 + return drv->uevent(xdev, env);
5902 }
5903
5904 return 0;
5905 --- sle11-2009-10-16.orig/drivers/xen/xenoprof/xenoprofile.c 2009-03-04 11:25:55.000000000 +0100
5906 +++ sle11-2009-10-16/drivers/xen/xenoprof/xenoprofile.c 2009-03-11 15:39:38.000000000 +0100
5907 @@ -29,7 +29,6 @@
5908 #include <xen/driver_util.h>
5909 #include <xen/interface/xen.h>
5910 #include <xen/interface/xenoprof.h>
5911 -#include "../../../drivers/oprofile/cpu_buffer.h"
5912 #include "../../../drivers/oprofile/event_buffer.h"
5913
5914 #define MAX_XENOPROF_SAMPLES 16
5915 @@ -142,8 +141,7 @@ static void xenoprof_add_pc(xenoprof_buf
5916 if (xenoprof_is_escape(buf, tail) &&
5917 xenoprof_get_event(buf, tail) == XENOPROF_TRACE_BEGIN) {
5918 tracing=1;
5919 - oprofile_add_pc(ESCAPE_CODE, buf->event_log[tail].mode,
5920 - CPU_TRACE_BEGIN);
5921 + oprofile_add_mode(buf->event_log[tail].mode);
5922 if (!is_passive)
5923 oprofile_samples++;
5924 else
5925 --- sle11-2009-10-16.orig/fs/xfs/linux-2.6/xfs_buf.c 2009-10-28 14:55:06.000000000 +0100
5926 +++ sle11-2009-10-16/fs/xfs/linux-2.6/xfs_buf.c 2009-02-16 16:18:36.000000000 +0100
5927 @@ -187,7 +187,7 @@ free_address(
5928 {
5929 a_list_t *aentry;
5930
5931 -#ifdef CONFIG_XEN
5932 +#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN)
5933 /*
5934 * Xen needs to be able to make sure it can get an exclusive
5935 * RO mapping of pages it wants to turn into a pagetable. If
5936 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/agp.h 2009-10-28 14:55:06.000000000 +0100
5937 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/agp.h 2009-02-16 16:18:36.000000000 +0100
5938 @@ -1,20 +1,22 @@
5939 -#ifndef AGP_H
5940 -#define AGP_H 1
5941 +#ifndef _ASM_X86_AGP_H
5942 +#define _ASM_X86_AGP_H
5943
5944 #include <asm/pgtable.h>
5945 #include <asm/cacheflush.h>
5946 #include <asm/system.h>
5947
5948 -/*
5949 - * Functions to keep the agpgart mappings coherent with the MMU.
5950 - * The GART gives the CPU a physical alias of pages in memory. The alias region is
5951 - * mapped uncacheable. Make sure there are no conflicting mappings
5952 - * with different cachability attributes for the same page. This avoids
5953 - * data corruption on some CPUs.
5954 +/*
5955 + * Functions to keep the agpgart mappings coherent with the MMU. The
5956 + * GART gives the CPU a physical alias of pages in memory. The alias
5957 + * region is mapped uncacheable. Make sure there are no conflicting
5958 + * mappings with different cachability attributes for the same
5959 + * page. This avoids data corruption on some CPUs.
5960 */
5961
5962 -/* Caller's responsibility to call global_flush_tlb() for
5963 - * performance reasons */
5964 +/*
5965 + * Caller's responsibility to call global_flush_tlb() for performance
5966 + * reasons
5967 + */
5968 #define map_page_into_agp(page) ( \
5969 xen_create_contiguous_region((unsigned long)page_address(page), 0, 32) \
5970 ?: change_page_attr(page, 1, PAGE_KERNEL_NOCACHE))
5971 @@ -24,9 +26,11 @@
5972 change_page_attr(page, 1, PAGE_KERNEL))
5973 #define flush_agp_mappings() global_flush_tlb()
5974
5975 -/* Could use CLFLUSH here if the cpu supports it. But then it would
5976 - need to be called for each cacheline of the whole page so it may not be
5977 - worth it. Would need a page for it. */
5978 +/*
5979 + * Could use CLFLUSH here if the cpu supports it. But then it would
5980 + * need to be called for each cacheline of the whole page so it may
5981 + * not be worth it. Would need a page for it.
5982 + */
5983 #define flush_agp_cache() wbinvd()
5984
5985 /* Convert a physical address to an address suitable for the GART. */
5986 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
5987 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/desc.h 2009-02-16 16:18:36.000000000 +0100
5988 @@ -0,0 +1,5 @@
5989 +#ifdef CONFIG_X86_32
5990 +# include "desc_32.h"
5991 +#else
5992 +# include "desc_64.h"
5993 +#endif
5994 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/desc_64.h 2008-12-15 11:27:22.000000000 +0100
5995 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/desc_64.h 2009-02-16 16:18:36.000000000 +0100
5996 @@ -34,6 +34,18 @@ static inline void clear_LDT(void)
5997 put_cpu();
5998 }
5999
6000 +#ifndef CONFIG_X86_NO_TSS
6001 +static inline unsigned long __store_tr(void)
6002 +{
6003 + unsigned long tr;
6004 +
6005 + asm volatile ("str %w0":"=r" (tr));
6006 + return tr;
6007 +}
6008 +
6009 +#define store_tr(tr) (tr) = __store_tr()
6010 +#endif
6011 +
6012 /*
6013 * This is the ldt that every process will get unless we need
6014 * something other than this.
6015 @@ -47,6 +59,18 @@ extern struct desc_ptr cpu_gdt_descr[];
6016 /* the cpu gdt accessor */
6017 #define cpu_gdt(_cpu) ((struct desc_struct *)cpu_gdt_descr[_cpu].address)
6018
6019 +#ifndef CONFIG_XEN
6020 +static inline void load_gdt(const struct desc_ptr *ptr)
6021 +{
6022 + asm volatile("lgdt %w0"::"m" (*ptr));
6023 +}
6024 +
6025 +static inline void store_gdt(struct desc_ptr *ptr)
6026 +{
6027 + asm("sgdt %w0":"=m" (*ptr));
6028 +}
6029 +#endif
6030 +
6031 static inline void _set_gate(void *adr, unsigned type, unsigned long func, unsigned dpl, unsigned ist)
6032 {
6033 struct gate_struct s;
6034 @@ -87,6 +111,16 @@ static inline void set_system_gate_ist(i
6035 {
6036 _set_gate(&idt_table[nr], GATE_INTERRUPT, (unsigned long) func, 3, ist);
6037 }
6038 +
6039 +static inline void load_idt(const struct desc_ptr *ptr)
6040 +{
6041 + asm volatile("lidt %w0"::"m" (*ptr));
6042 +}
6043 +
6044 +static inline void store_idt(struct desc_ptr *dtr)
6045 +{
6046 + asm("sidt %w0":"=m" (*dtr));
6047 +}
6048 #endif
6049
6050 static inline void set_tssldt_descriptor(void *ptr, unsigned long tss, unsigned type,
6051 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
6052 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/dma-mapping.h 2009-02-16 16:18:36.000000000 +0100
6053 @@ -0,0 +1,5 @@
6054 +#ifdef CONFIG_X86_32
6055 +# include "dma-mapping_32.h"
6056 +#else
6057 +# include "dma-mapping_64.h"
6058 +#endif
6059 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/dma-mapping_32.h 2008-12-15 11:26:44.000000000 +0100
6060 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/dma-mapping_32.h 2009-02-16 16:18:36.000000000 +0100
6061 @@ -7,9 +7,9 @@
6062 */
6063
6064 #include <linux/mm.h>
6065 +#include <linux/scatterlist.h>
6066 #include <asm/cache.h>
6067 #include <asm/io.h>
6068 -#include <asm/scatterlist.h>
6069 #include <asm/swiotlb.h>
6070
6071 static inline int
6072 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/dma-mapping_64.h 2008-12-15 11:27:22.000000000 +0100
6073 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/dma-mapping_64.h 2009-02-16 16:18:36.000000000 +0100
6074 @@ -6,8 +6,7 @@
6075 * documentation.
6076 */
6077
6078 -
6079 -#include <asm/scatterlist.h>
6080 +#include <linux/scatterlist.h>
6081 #include <asm/io.h>
6082
6083 struct dma_mapping_ops {
6084 @@ -203,4 +202,4 @@ extern int panic_on_overflow;
6085
6086 #endif /* _X8664_DMA_MAPPING_H */
6087
6088 -#include <asm-i386/mach-xen/asm/dma-mapping.h>
6089 +#include "dma-mapping_32.h"
6090 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
6091 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/fixmap.h 2009-02-16 16:18:36.000000000 +0100
6092 @@ -0,0 +1,5 @@
6093 +#ifdef CONFIG_X86_32
6094 +# include "fixmap_32.h"
6095 +#else
6096 +# include "fixmap_64.h"
6097 +#endif
6098 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
6099 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/hypercall.h 2009-02-16 16:18:36.000000000 +0100
6100 @@ -0,0 +1,404 @@
6101 +/******************************************************************************
6102 + * hypercall.h
6103 + *
6104 + * Linux-specific hypervisor handling.
6105 + *
6106 + * Copyright (c) 2002-2004, K A Fraser
6107 + *
6108 + * 64-bit updates:
6109 + * Benjamin Liu <benjamin.liu@intel.com>
6110 + * Jun Nakajima <jun.nakajima@intel.com>
6111 + *
6112 + * This program is free software; you can redistribute it and/or
6113 + * modify it under the terms of the GNU General Public License version 2
6114 + * as published by the Free Software Foundation; or, when distributed
6115 + * separately from the Linux kernel or incorporated into other
6116 + * software packages, subject to the following license:
6117 + *
6118 + * Permission is hereby granted, free of charge, to any person obtaining a copy
6119 + * of this source file (the "Software"), to deal in the Software without
6120 + * restriction, including without limitation the rights to use, copy, modify,
6121 + * merge, publish, distribute, sublicense, and/or sell copies of the Software,
6122 + * and to permit persons to whom the Software is furnished to do so, subject to
6123 + * the following conditions:
6124 + *
6125 + * The above copyright notice and this permission notice shall be included in
6126 + * all copies or substantial portions of the Software.
6127 + *
6128 + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
6129 + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
6130 + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
6131 + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
6132 + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
6133 + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
6134 + * IN THE SOFTWARE.
6135 + */
6136 +
6137 +#ifndef __HYPERCALL_H__
6138 +#define __HYPERCALL_H__
6139 +
6140 +#ifndef __HYPERVISOR_H__
6141 +# error "please don't include this file directly"
6142 +#endif
6143 +
6144 +#if CONFIG_XEN_COMPAT <= 0x030002
6145 +# include <linux/string.h> /* memcpy() */
6146 +#endif
6147 +
6148 +#ifdef CONFIG_XEN
6149 +#define HYPERCALL_ASM_OPERAND "%c"
6150 +#define HYPERCALL_LOCATION(op) (hypercall_page + (op) * 32)
6151 +#define HYPERCALL_C_OPERAND(name) "i" (HYPERCALL_LOCATION(__HYPERVISOR_##name))
6152 +#else
6153 +#define HYPERCALL_ASM_OPERAND "*%"
6154 +#define HYPERCALL_LOCATION(op) (hypercall_stubs + (op) * 32)
6155 +#define HYPERCALL_C_OPERAND(name) "g" (HYPERCALL_LOCATION(__HYPERVISOR_##name))
6156 +#endif
6157 +
6158 +#define HYPERCALL_ARG(arg, n) \
6159 + register typeof((arg)+0) __arg##n asm(HYPERCALL_arg##n) = (arg)
6160 +
6161 +#define _hypercall0(type, name) \
6162 +({ \
6163 + type __res; \
6164 + asm volatile ( \
6165 + "call " HYPERCALL_ASM_OPERAND "1" \
6166 + : "=a" (__res) \
6167 + : HYPERCALL_C_OPERAND(name) \
6168 + : "memory" ); \
6169 + __res; \
6170 +})
6171 +
6172 +#define _hypercall1(type, name, arg) \
6173 +({ \
6174 + type __res; \
6175 + HYPERCALL_ARG(arg, 1); \
6176 + asm volatile ( \
6177 + "call " HYPERCALL_ASM_OPERAND "2" \
6178 + : "=a" (__res), "+r" (__arg1) \
6179 + : HYPERCALL_C_OPERAND(name) \
6180 + : "memory" ); \
6181 + __res; \
6182 +})
6183 +
6184 +#define _hypercall2(type, name, a1, a2) \
6185 +({ \
6186 + type __res; \
6187 + HYPERCALL_ARG(a1, 1); \
6188 + HYPERCALL_ARG(a2, 2); \
6189 + asm volatile ( \
6190 + "call " HYPERCALL_ASM_OPERAND "3" \
6191 + : "=a" (__res), "+r" (__arg1), "+r" (__arg2) \
6192 + : HYPERCALL_C_OPERAND(name) \
6193 + : "memory" ); \
6194 + __res; \
6195 +})
6196 +
6197 +#define _hypercall3(type, name, a1, a2, a3) \
6198 +({ \
6199 + type __res; \
6200 + HYPERCALL_ARG(a1, 1); \
6201 + HYPERCALL_ARG(a2, 2); \
6202 + HYPERCALL_ARG(a3, 3); \
6203 + asm volatile ( \
6204 + "call " HYPERCALL_ASM_OPERAND "4" \
6205 + : "=a" (__res), "+r" (__arg1), \
6206 + "+r" (__arg2), "+r" (__arg3) \
6207 + : HYPERCALL_C_OPERAND(name) \
6208 + : "memory" ); \
6209 + __res; \
6210 +})
6211 +
6212 +#define _hypercall4(type, name, a1, a2, a3, a4) \
6213 +({ \
6214 + type __res; \
6215 + HYPERCALL_ARG(a1, 1); \
6216 + HYPERCALL_ARG(a2, 2); \
6217 + HYPERCALL_ARG(a3, 3); \
6218 + HYPERCALL_ARG(a4, 4); \
6219 + asm volatile ( \
6220 + "call " HYPERCALL_ASM_OPERAND "5" \
6221 + : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
6222 + "+r" (__arg3), "+r" (__arg4) \
6223 + : HYPERCALL_C_OPERAND(name) \
6224 + : "memory" ); \
6225 + __res; \
6226 +})
6227 +
6228 +#define _hypercall5(type, name, a1, a2, a3, a4, a5) \
6229 +({ \
6230 + type __res; \
6231 + HYPERCALL_ARG(a1, 1); \
6232 + HYPERCALL_ARG(a2, 2); \
6233 + HYPERCALL_ARG(a3, 3); \
6234 + HYPERCALL_ARG(a4, 4); \
6235 + HYPERCALL_ARG(a5, 5); \
6236 + asm volatile ( \
6237 + "call " HYPERCALL_ASM_OPERAND "6" \
6238 + : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
6239 + "+r" (__arg3), "+r" (__arg4), "+r" (__arg5) \
6240 + : HYPERCALL_C_OPERAND(name) \
6241 + : "memory" ); \
6242 + __res; \
6243 +})
6244 +
6245 +#define _hypercall(type, op, a1, a2, a3, a4, a5) \
6246 +({ \
6247 + type __res; \
6248 + HYPERCALL_ARG(a1, 1); \
6249 + HYPERCALL_ARG(a2, 2); \
6250 + HYPERCALL_ARG(a3, 3); \
6251 + HYPERCALL_ARG(a4, 4); \
6252 + HYPERCALL_ARG(a5, 5); \
6253 + asm volatile ( \
6254 + "call *%6" \
6255 + : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
6256 + "+r" (__arg3), "+r" (__arg4), "+r" (__arg5) \
6257 + : "g" (HYPERCALL_LOCATION(op)) \
6258 + : "memory" ); \
6259 + __res; \
6260 +})
6261 +
6262 +#ifdef CONFIG_X86_32
6263 +# include "hypercall_32.h"
6264 +#else
6265 +# include "hypercall_64.h"
6266 +#endif
6267 +
6268 +static inline int __must_check
6269 +HYPERVISOR_set_trap_table(
6270 + const trap_info_t *table)
6271 +{
6272 + return _hypercall1(int, set_trap_table, table);
6273 +}
6274 +
6275 +static inline int __must_check
6276 +HYPERVISOR_mmu_update(
6277 + mmu_update_t *req, unsigned int count, unsigned int *success_count,
6278 + domid_t domid)
6279 +{
6280 + if (arch_use_lazy_mmu_mode())
6281 + return xen_multi_mmu_update(req, count, success_count, domid);
6282 + return _hypercall4(int, mmu_update, req, count, success_count, domid);
6283 +}
6284 +
6285 +static inline int __must_check
6286 +HYPERVISOR_mmuext_op(
6287 + struct mmuext_op *op, unsigned int count, unsigned int *success_count,
6288 + domid_t domid)
6289 +{
6290 + if (arch_use_lazy_mmu_mode())
6291 + return xen_multi_mmuext_op(op, count, success_count, domid);
6292 + return _hypercall4(int, mmuext_op, op, count, success_count, domid);
6293 +}
6294 +
6295 +static inline int __must_check
6296 +HYPERVISOR_set_gdt(
6297 + unsigned long *frame_list, unsigned int entries)
6298 +{
6299 + return _hypercall2(int, set_gdt, frame_list, entries);
6300 +}
6301 +
6302 +static inline int __must_check
6303 +HYPERVISOR_stack_switch(
6304 + unsigned long ss, unsigned long esp)
6305 +{
6306 + return _hypercall2(int, stack_switch, ss, esp);
6307 +}
6308 +
6309 +static inline int
6310 +HYPERVISOR_fpu_taskswitch(
6311 + int set)
6312 +{
6313 + return _hypercall1(int, fpu_taskswitch, set);
6314 +}
6315 +
6316 +#if CONFIG_XEN_COMPAT <= 0x030002
6317 +static inline int __must_check
6318 +HYPERVISOR_sched_op_compat(
6319 + int cmd, unsigned long arg)
6320 +{
6321 + return _hypercall2(int, sched_op_compat, cmd, arg);
6322 +}
6323 +#endif
6324 +
6325 +static inline int __must_check
6326 +HYPERVISOR_sched_op(
6327 + int cmd, void *arg)
6328 +{
6329 + return _hypercall2(int, sched_op, cmd, arg);
6330 +}
6331 +
6332 +static inline int __must_check
6333 +HYPERVISOR_platform_op(
6334 + struct xen_platform_op *platform_op)
6335 +{
6336 + platform_op->interface_version = XENPF_INTERFACE_VERSION;
6337 + return _hypercall1(int, platform_op, platform_op);
6338 +}
6339 +
6340 +static inline int __must_check
6341 +HYPERVISOR_set_debugreg(
6342 + unsigned int reg, unsigned long value)
6343 +{
6344 + return _hypercall2(int, set_debugreg, reg, value);
6345 +}
6346 +
6347 +static inline unsigned long __must_check
6348 +HYPERVISOR_get_debugreg(
6349 + unsigned int reg)
6350 +{
6351 + return _hypercall1(unsigned long, get_debugreg, reg);
6352 +}
6353 +
6354 +static inline int __must_check
6355 +HYPERVISOR_memory_op(
6356 + unsigned int cmd, void *arg)
6357 +{
6358 + if (arch_use_lazy_mmu_mode())
6359 + xen_multicall_flush(false);
6360 + return _hypercall2(int, memory_op, cmd, arg);
6361 +}
6362 +
6363 +static inline int __must_check
6364 +HYPERVISOR_multicall(
6365 + multicall_entry_t *call_list, unsigned int nr_calls)
6366 +{
6367 + return _hypercall2(int, multicall, call_list, nr_calls);
6368 +}
6369 +
6370 +static inline int __must_check
6371 +HYPERVISOR_event_channel_op(
6372 + int cmd, void *arg)
6373 +{
6374 + int rc = _hypercall2(int, event_channel_op, cmd, arg);
6375 +
6376 +#if CONFIG_XEN_COMPAT <= 0x030002
6377 + if (unlikely(rc == -ENOSYS)) {
6378 + struct evtchn_op op;
6379 + op.cmd = cmd;
6380 + memcpy(&op.u, arg, sizeof(op.u));
6381 + rc = _hypercall1(int, event_channel_op_compat, &op);
6382 + memcpy(arg, &op.u, sizeof(op.u));
6383 + }
6384 +#endif
6385 +
6386 + return rc;
6387 +}
6388 +
6389 +static inline int __must_check
6390 +HYPERVISOR_xen_version(
6391 + int cmd, void *arg)
6392 +{
6393 + return _hypercall2(int, xen_version, cmd, arg);
6394 +}
6395 +
6396 +static inline int __must_check
6397 +HYPERVISOR_console_io(
6398 + int cmd, unsigned int count, char *str)
6399 +{
6400 + return _hypercall3(int, console_io, cmd, count, str);
6401 +}
6402 +
6403 +static inline int __must_check
6404 +HYPERVISOR_physdev_op(
6405 + int cmd, void *arg)
6406 +{
6407 + int rc = _hypercall2(int, physdev_op, cmd, arg);
6408 +
6409 +#if CONFIG_XEN_COMPAT <= 0x030002
6410 + if (unlikely(rc == -ENOSYS)) {
6411 + struct physdev_op op;
6412 + op.cmd = cmd;
6413 + memcpy(&op.u, arg, sizeof(op.u));
6414 + rc = _hypercall1(int, physdev_op_compat, &op);
6415 + memcpy(arg, &op.u, sizeof(op.u));
6416 + }
6417 +#endif
6418 +
6419 + return rc;
6420 +}
6421 +
6422 +static inline int __must_check
6423 +HYPERVISOR_grant_table_op(
6424 + unsigned int cmd, void *uop, unsigned int count)
6425 +{
6426 + if (arch_use_lazy_mmu_mode())
6427 + xen_multicall_flush(false);
6428 + return _hypercall3(int, grant_table_op, cmd, uop, count);
6429 +}
6430 +
6431 +static inline int __must_check
6432 +HYPERVISOR_vm_assist(
6433 + unsigned int cmd, unsigned int type)
6434 +{
6435 + return _hypercall2(int, vm_assist, cmd, type);
6436 +}
6437 +
6438 +static inline int __must_check
6439 +HYPERVISOR_vcpu_op(
6440 + int cmd, unsigned int vcpuid, void *extra_args)
6441 +{
6442 + return _hypercall3(int, vcpu_op, cmd, vcpuid, extra_args);
6443 +}
6444 +
6445 +static inline int __must_check
6446 +HYPERVISOR_suspend(
6447 + unsigned long srec)
6448 +{
6449 + struct sched_shutdown sched_shutdown = {
6450 + .reason = SHUTDOWN_suspend
6451 + };
6452 +
6453 + int rc = _hypercall3(int, sched_op, SCHEDOP_shutdown,
6454 + &sched_shutdown, srec);
6455 +
6456 +#if CONFIG_XEN_COMPAT <= 0x030002
6457 + if (rc == -ENOSYS)
6458 + rc = _hypercall3(int, sched_op_compat, SCHEDOP_shutdown,
6459 + SHUTDOWN_suspend, srec);
6460 +#endif
6461 +
6462 + return rc;
6463 +}
6464 +
6465 +#if CONFIG_XEN_COMPAT <= 0x030002
6466 +static inline int
6467 +HYPERVISOR_nmi_op(
6468 + unsigned long op, void *arg)
6469 +{
6470 + return _hypercall2(int, nmi_op, op, arg);
6471 +}
6472 +#endif
6473 +
6474 +#ifndef CONFIG_XEN
6475 +static inline unsigned long __must_check
6476 +HYPERVISOR_hvm_op(
6477 + int op, void *arg)
6478 +{
6479 + return _hypercall2(unsigned long, hvm_op, op, arg);
6480 +}
6481 +#endif
6482 +
6483 +static inline int __must_check
6484 +HYPERVISOR_callback_op(
6485 + int cmd, const void *arg)
6486 +{
6487 + return _hypercall2(int, callback_op, cmd, arg);
6488 +}
6489 +
6490 +static inline int __must_check
6491 +HYPERVISOR_xenoprof_op(
6492 + int op, void *arg)
6493 +{
6494 + return _hypercall2(int, xenoprof_op, op, arg);
6495 +}
6496 +
6497 +static inline int __must_check
6498 +HYPERVISOR_kexec_op(
6499 + unsigned long op, void *args)
6500 +{
6501 + return _hypercall2(int, kexec_op, op, args);
6502 +}
6503 +
6504 +#endif /* __HYPERCALL_H__ */
6505 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/hypercall_32.h 2009-03-04 11:28:34.000000000 +0100
6506 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/hypercall_32.h 2009-02-16 16:18:36.000000000 +0100
6507 @@ -1,191 +1,10 @@
6508 -/******************************************************************************
6509 - * hypercall.h
6510 - *
6511 - * Linux-specific hypervisor handling.
6512 - *
6513 - * Copyright (c) 2002-2004, K A Fraser
6514 - *
6515 - * This program is free software; you can redistribute it and/or
6516 - * modify it under the terms of the GNU General Public License version 2
6517 - * as published by the Free Software Foundation; or, when distributed
6518 - * separately from the Linux kernel or incorporated into other
6519 - * software packages, subject to the following license:
6520 - *
6521 - * Permission is hereby granted, free of charge, to any person obtaining a copy
6522 - * of this source file (the "Software"), to deal in the Software without
6523 - * restriction, including without limitation the rights to use, copy, modify,
6524 - * merge, publish, distribute, sublicense, and/or sell copies of the Software,
6525 - * and to permit persons to whom the Software is furnished to do so, subject to
6526 - * the following conditions:
6527 - *
6528 - * The above copyright notice and this permission notice shall be included in
6529 - * all copies or substantial portions of the Software.
6530 - *
6531 - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
6532 - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
6533 - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
6534 - * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
6535 - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
6536 - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
6537 - * IN THE SOFTWARE.
6538 - */
6539 -
6540 -#ifndef __HYPERCALL_H__
6541 -#define __HYPERCALL_H__
6542 -
6543 -#include <linux/string.h> /* memcpy() */
6544 -#include <linux/stringify.h>
6545 -
6546 -#ifndef __HYPERVISOR_H__
6547 -# error "please don't include this file directly"
6548 -#endif
6549 -
6550 -#ifdef CONFIG_XEN
6551 -#define HYPERCALL_STR(name) \
6552 - "call hypercall_page + ("__stringify(__HYPERVISOR_##name)" * 32)"
6553 -#else
6554 -#define HYPERCALL_STR(name) \
6555 - "mov hypercall_stubs,%%eax; " \
6556 - "add $("__stringify(__HYPERVISOR_##name)" * 32),%%eax; "\
6557 - "call *%%eax"
6558 -#endif
6559 -
6560 -#define _hypercall0(type, name) \
6561 -({ \
6562 - type __res; \
6563 - asm volatile ( \
6564 - HYPERCALL_STR(name) \
6565 - : "=a" (__res) \
6566 - : \
6567 - : "memory" ); \
6568 - __res; \
6569 -})
6570 -
6571 -#define _hypercall1(type, name, a1) \
6572 -({ \
6573 - type __res; \
6574 - long __ign1; \
6575 - asm volatile ( \
6576 - HYPERCALL_STR(name) \
6577 - : "=a" (__res), "=b" (__ign1) \
6578 - : "1" ((long)(a1)) \
6579 - : "memory" ); \
6580 - __res; \
6581 -})
6582 -
6583 -#define _hypercall2(type, name, a1, a2) \
6584 -({ \
6585 - type __res; \
6586 - long __ign1, __ign2; \
6587 - asm volatile ( \
6588 - HYPERCALL_STR(name) \
6589 - : "=a" (__res), "=b" (__ign1), "=c" (__ign2) \
6590 - : "1" ((long)(a1)), "2" ((long)(a2)) \
6591 - : "memory" ); \
6592 - __res; \
6593 -})
6594 -
6595 -#define _hypercall3(type, name, a1, a2, a3) \
6596 -({ \
6597 - type __res; \
6598 - long __ign1, __ign2, __ign3; \
6599 - asm volatile ( \
6600 - HYPERCALL_STR(name) \
6601 - : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
6602 - "=d" (__ign3) \
6603 - : "1" ((long)(a1)), "2" ((long)(a2)), \
6604 - "3" ((long)(a3)) \
6605 - : "memory" ); \
6606 - __res; \
6607 -})
6608 -
6609 -#define _hypercall4(type, name, a1, a2, a3, a4) \
6610 -({ \
6611 - type __res; \
6612 - long __ign1, __ign2, __ign3, __ign4; \
6613 - asm volatile ( \
6614 - HYPERCALL_STR(name) \
6615 - : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
6616 - "=d" (__ign3), "=S" (__ign4) \
6617 - : "1" ((long)(a1)), "2" ((long)(a2)), \
6618 - "3" ((long)(a3)), "4" ((long)(a4)) \
6619 - : "memory" ); \
6620 - __res; \
6621 -})
6622 -
6623 -#define _hypercall5(type, name, a1, a2, a3, a4, a5) \
6624 -({ \
6625 - type __res; \
6626 - long __ign1, __ign2, __ign3, __ign4, __ign5; \
6627 - asm volatile ( \
6628 - HYPERCALL_STR(name) \
6629 - : "=a" (__res), "=b" (__ign1), "=c" (__ign2), \
6630 - "=d" (__ign3), "=S" (__ign4), "=D" (__ign5) \
6631 - : "1" ((long)(a1)), "2" ((long)(a2)), \
6632 - "3" ((long)(a3)), "4" ((long)(a4)), \
6633 - "5" ((long)(a5)) \
6634 - : "memory" ); \
6635 - __res; \
6636 -})
6637 -
6638 -#define _hypercall(type, op, a1, a2, a3, a4, a5) \
6639 -({ \
6640 - type __res; \
6641 - register typeof((a1)+0) __arg1 asm("ebx") = (a1); \
6642 - register typeof((a2)+0) __arg2 asm("ecx") = (a2); \
6643 - register typeof((a3)+0) __arg3 asm("edx") = (a3); \
6644 - register typeof((a4)+0) __arg4 asm("esi") = (a4); \
6645 - register typeof((a5)+0) __arg5 asm("edi") = (a5); \
6646 - asm volatile ( \
6647 - "call *%6" \
6648 - : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
6649 - "+r" (__arg3), "+r" (__arg4), "+r" (__arg5) \
6650 - : "0" (hypercall_page + (op) * 32) \
6651 - : "memory" ); \
6652 - __res; \
6653 -})
6654 -
6655 -static inline int __must_check
6656 -HYPERVISOR_set_trap_table(
6657 - const trap_info_t *table)
6658 -{
6659 - return _hypercall1(int, set_trap_table, table);
6660 -}
6661 -
6662 -static inline int __must_check
6663 -HYPERVISOR_mmu_update(
6664 - mmu_update_t *req, unsigned int count, unsigned int *success_count,
6665 - domid_t domid)
6666 -{
6667 - if (arch_use_lazy_mmu_mode())
6668 - return xen_multi_mmu_update(req, count, success_count, domid);
6669 - return _hypercall4(int, mmu_update, req, count, success_count, domid);
6670 -}
6671 -
6672 -static inline int __must_check
6673 -HYPERVISOR_mmuext_op(
6674 - struct mmuext_op *op, unsigned int count, unsigned int *success_count,
6675 - domid_t domid)
6676 -{
6677 - if (arch_use_lazy_mmu_mode())
6678 - return xen_multi_mmuext_op(op, count, success_count, domid);
6679 - return _hypercall4(int, mmuext_op, op, count, success_count, domid);
6680 -}
6681 -
6682 -static inline int __must_check
6683 -HYPERVISOR_set_gdt(
6684 - unsigned long *frame_list, unsigned int entries)
6685 -{
6686 - return _hypercall2(int, set_gdt, frame_list, entries);
6687 -}
6688 -
6689 -static inline int __must_check
6690 -HYPERVISOR_stack_switch(
6691 - unsigned long ss, unsigned long esp)
6692 -{
6693 - return _hypercall2(int, stack_switch, ss, esp);
6694 -}
6695 +#define HYPERCALL_arg1 "ebx"
6696 +#define HYPERCALL_arg2 "ecx"
6697 +#define HYPERCALL_arg3 "edx"
6698 +#define HYPERCALL_arg4 "esi"
6699 +#define HYPERCALL_arg5 "edi"
6700
6701 +#if CONFIG_XEN_COMPAT <= 0x030002
6702 static inline int __must_check
6703 HYPERVISOR_set_callbacks(
6704 unsigned long event_selector, unsigned long event_address,
6705 @@ -195,80 +14,24 @@ HYPERVISOR_set_callbacks(
6706 event_selector, event_address,
6707 failsafe_selector, failsafe_address);
6708 }
6709 -
6710 -static inline int
6711 -HYPERVISOR_fpu_taskswitch(
6712 - int set)
6713 -{
6714 - return _hypercall1(int, fpu_taskswitch, set);
6715 -}
6716 -
6717 -static inline int __must_check
6718 -HYPERVISOR_sched_op_compat(
6719 - int cmd, unsigned long arg)
6720 -{
6721 - return _hypercall2(int, sched_op_compat, cmd, arg);
6722 -}
6723 -
6724 -static inline int __must_check
6725 -HYPERVISOR_sched_op(
6726 - int cmd, void *arg)
6727 -{
6728 - return _hypercall2(int, sched_op, cmd, arg);
6729 -}
6730 +#endif
6731
6732 static inline long __must_check
6733 HYPERVISOR_set_timer_op(
6734 u64 timeout)
6735 {
6736 - unsigned long timeout_hi = (unsigned long)(timeout>>32);
6737 - unsigned long timeout_lo = (unsigned long)timeout;
6738 - return _hypercall2(long, set_timer_op, timeout_lo, timeout_hi);
6739 -}
6740 -
6741 -static inline int __must_check
6742 -HYPERVISOR_platform_op(
6743 - struct xen_platform_op *platform_op)
6744 -{
6745 - platform_op->interface_version = XENPF_INTERFACE_VERSION;
6746 - return _hypercall1(int, platform_op, platform_op);
6747 -}
6748 -
6749 -static inline int __must_check
6750 -HYPERVISOR_set_debugreg(
6751 - unsigned int reg, unsigned long value)
6752 -{
6753 - return _hypercall2(int, set_debugreg, reg, value);
6754 -}
6755 -
6756 -static inline unsigned long __must_check
6757 -HYPERVISOR_get_debugreg(
6758 - unsigned int reg)
6759 -{
6760 - return _hypercall1(unsigned long, get_debugreg, reg);
6761 + return _hypercall2(long, set_timer_op,
6762 + (unsigned long)timeout,
6763 + (unsigned long)(timeout>>32));
6764 }
6765
6766 static inline int __must_check
6767 HYPERVISOR_update_descriptor(
6768 u64 ma, u64 desc)
6769 {
6770 - return _hypercall4(int, update_descriptor, ma, ma>>32, desc, desc>>32);
6771 -}
6772 -
6773 -static inline int __must_check
6774 -HYPERVISOR_memory_op(
6775 - unsigned int cmd, void *arg)
6776 -{
6777 - if (arch_use_lazy_mmu_mode())
6778 - xen_multicall_flush(false);
6779 - return _hypercall2(int, memory_op, cmd, arg);
6780 -}
6781 -
6782 -static inline int __must_check
6783 -HYPERVISOR_multicall(
6784 - multicall_entry_t *call_list, unsigned int nr_calls)
6785 -{
6786 - return _hypercall2(int, multicall, call_list, nr_calls);
6787 + return _hypercall4(int, update_descriptor,
6788 + (unsigned long)ma, (unsigned long)(ma>>32),
6789 + (unsigned long)desc, (unsigned long)(desc>>32));
6790 }
6791
6792 static inline int __must_check
6793 @@ -287,67 +50,6 @@ HYPERVISOR_update_va_mapping(
6794 }
6795
6796 static inline int __must_check
6797 -HYPERVISOR_event_channel_op(
6798 - int cmd, void *arg)
6799 -{
6800 - int rc = _hypercall2(int, event_channel_op, cmd, arg);
6801 -
6802 -#if CONFIG_XEN_COMPAT <= 0x030002
6803 - if (unlikely(rc == -ENOSYS)) {
6804 - struct evtchn_op op;
6805 - op.cmd = cmd;
6806 - memcpy(&op.u, arg, sizeof(op.u));
6807 - rc = _hypercall1(int, event_channel_op_compat, &op);
6808 - memcpy(arg, &op.u, sizeof(op.u));
6809 - }
6810 -#endif
6811 -
6812 - return rc;
6813 -}
6814 -
6815 -static inline int __must_check
6816 -HYPERVISOR_xen_version(
6817 - int cmd, void *arg)
6818 -{
6819 - return _hypercall2(int, xen_version, cmd, arg);
6820 -}
6821 -
6822 -static inline int __must_check
6823 -HYPERVISOR_console_io(
6824 - int cmd, unsigned int count, char *str)
6825 -{
6826 - return _hypercall3(int, console_io, cmd, count, str);
6827 -}
6828 -
6829 -static inline int __must_check
6830 -HYPERVISOR_physdev_op(
6831 - int cmd, void *arg)
6832 -{
6833 - int rc = _hypercall2(int, physdev_op, cmd, arg);
6834 -
6835 -#if CONFIG_XEN_COMPAT <= 0x030002
6836 - if (unlikely(rc == -ENOSYS)) {
6837 - struct physdev_op op;
6838 - op.cmd = cmd;
6839 - memcpy(&op.u, arg, sizeof(op.u));
6840 - rc = _hypercall1(int, physdev_op_compat, &op);
6841 - memcpy(arg, &op.u, sizeof(op.u));
6842 - }
6843 -#endif
6844 -
6845 - return rc;
6846 -}
6847 -
6848 -static inline int __must_check
6849 -HYPERVISOR_grant_table_op(
6850 - unsigned int cmd, void *uop, unsigned int count)
6851 -{
6852 - if (arch_use_lazy_mmu_mode())
6853 - xen_multicall_flush(false);
6854 - return _hypercall3(int, grant_table_op, cmd, uop, count);
6855 -}
6856 -
6857 -static inline int __must_check
6858 HYPERVISOR_update_va_mapping_otherdomain(
6859 unsigned long va, pte_t new_val, unsigned long flags, domid_t domid)
6860 {
6861 @@ -358,80 +60,3 @@ HYPERVISOR_update_va_mapping_otherdomain
6862 return _hypercall5(int, update_va_mapping_otherdomain, va,
6863 new_val.pte_low, pte_hi, flags, domid);
6864 }
6865 -
6866 -static inline int __must_check
6867 -HYPERVISOR_vm_assist(
6868 - unsigned int cmd, unsigned int type)
6869 -{
6870 - return _hypercall2(int, vm_assist, cmd, type);
6871 -}
6872 -
6873 -static inline int __must_check
6874 -HYPERVISOR_vcpu_op(
6875 - int cmd, unsigned int vcpuid, void *extra_args)
6876 -{
6877 - return _hypercall3(int, vcpu_op, cmd, vcpuid, extra_args);
6878 -}
6879 -
6880 -static inline int __must_check
6881 -HYPERVISOR_suspend(
6882 - unsigned long srec)
6883 -{
6884 - struct sched_shutdown sched_shutdown = {
6885 - .reason = SHUTDOWN_suspend
6886 - };
6887 -
6888 - int rc = _hypercall3(int, sched_op, SCHEDOP_shutdown,
6889 - &sched_shutdown, srec);
6890 -
6891 -#if CONFIG_XEN_COMPAT <= 0x030002
6892 - if (rc == -ENOSYS)
6893 - rc = _hypercall3(int, sched_op_compat, SCHEDOP_shutdown,
6894 - SHUTDOWN_suspend, srec);
6895 -#endif
6896 -
6897 - return rc;
6898 -}
6899 -
6900 -#if CONFIG_XEN_COMPAT <= 0x030002
6901 -static inline int
6902 -HYPERVISOR_nmi_op(
6903 - unsigned long op, void *arg)
6904 -{
6905 - return _hypercall2(int, nmi_op, op, arg);
6906 -}
6907 -#endif
6908 -
6909 -#ifndef CONFIG_XEN
6910 -static inline unsigned long __must_check
6911 -HYPERVISOR_hvm_op(
6912 - int op, void *arg)
6913 -{
6914 - return _hypercall2(unsigned long, hvm_op, op, arg);
6915 -}
6916 -#endif
6917 -
6918 -static inline int __must_check
6919 -HYPERVISOR_callback_op(
6920 - int cmd, const void *arg)
6921 -{
6922 - return _hypercall2(int, callback_op, cmd, arg);
6923 -}
6924 -
6925 -static inline int __must_check
6926 -HYPERVISOR_xenoprof_op(
6927 - int op, void *arg)
6928 -{
6929 - return _hypercall2(int, xenoprof_op, op, arg);
6930 -}
6931 -
6932 -static inline int __must_check
6933 -HYPERVISOR_kexec_op(
6934 - unsigned long op, void *args)
6935 -{
6936 - return _hypercall2(int, kexec_op, op, args);
6937 -}
6938 -
6939 -
6940 -
6941 -#endif /* __HYPERCALL_H__ */
6942 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/hypercall_64.h 2009-03-04 11:28:34.000000000 +0100
6943 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/hypercall_64.h 2009-02-16 16:18:36.000000000 +0100
6944 @@ -1,197 +1,10 @@
6945 -/******************************************************************************
6946 - * hypercall.h
6947 - *
6948 - * Linux-specific hypervisor handling.
6949 - *
6950 - * Copyright (c) 2002-2004, K A Fraser
6951 - *
6952 - * 64-bit updates:
6953 - * Benjamin Liu <benjamin.liu@intel.com>
6954 - * Jun Nakajima <jun.nakajima@intel.com>
6955 - *
6956 - * This program is free software; you can redistribute it and/or
6957 - * modify it under the terms of the GNU General Public License version 2
6958 - * as published by the Free Software Foundation; or, when distributed
6959 - * separately from the Linux kernel or incorporated into other
6960 - * software packages, subject to the following license:
6961 - *
6962 - * Permission is hereby granted, free of charge, to any person obtaining a copy
6963 - * of this source file (the "Software"), to deal in the Software without
6964 - * restriction, including without limitation the rights to use, copy, modify,
6965 - * merge, publish, distribute, sublicense, and/or sell copies of the Software,
6966 - * and to permit persons to whom the Software is furnished to do so, subject to
6967 - * the following conditions:
6968 - *
6969 - * The above copyright notice and this permission notice shall be included in
6970 - * all copies or substantial portions of the Software.
6971 - *
6972 - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
6973 - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
6974 - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
6975 - * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
6976 - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
6977 - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
6978 - * IN THE SOFTWARE.
6979 - */
6980 -
6981 -#ifndef __HYPERCALL_H__
6982 -#define __HYPERCALL_H__
6983 -
6984 -#include <linux/string.h> /* memcpy() */
6985 -#include <linux/stringify.h>
6986 -
6987 -#ifndef __HYPERVISOR_H__
6988 -# error "please don't include this file directly"
6989 -#endif
6990 -
6991 -#ifdef CONFIG_XEN
6992 -#define HYPERCALL_STR(name) \
6993 - "call hypercall_page + ("__stringify(__HYPERVISOR_##name)" * 32)"
6994 -#else
6995 -#define HYPERCALL_STR(name) \
6996 - "mov $("__stringify(__HYPERVISOR_##name)" * 32),%%eax; "\
6997 - "add hypercall_stubs(%%rip),%%rax; " \
6998 - "call *%%rax"
6999 -#endif
7000 -
7001 -#define _hypercall0(type, name) \
7002 -({ \
7003 - type __res; \
7004 - asm volatile ( \
7005 - HYPERCALL_STR(name) \
7006 - : "=a" (__res) \
7007 - : \
7008 - : "memory" ); \
7009 - __res; \
7010 -})
7011 -
7012 -#define _hypercall1(type, name, a1) \
7013 -({ \
7014 - type __res; \
7015 - long __ign1; \
7016 - asm volatile ( \
7017 - HYPERCALL_STR(name) \
7018 - : "=a" (__res), "=D" (__ign1) \
7019 - : "1" ((long)(a1)) \
7020 - : "memory" ); \
7021 - __res; \
7022 -})
7023 -
7024 -#define _hypercall2(type, name, a1, a2) \
7025 -({ \
7026 - type __res; \
7027 - long __ign1, __ign2; \
7028 - asm volatile ( \
7029 - HYPERCALL_STR(name) \
7030 - : "=a" (__res), "=D" (__ign1), "=S" (__ign2) \
7031 - : "1" ((long)(a1)), "2" ((long)(a2)) \
7032 - : "memory" ); \
7033 - __res; \
7034 -})
7035 -
7036 -#define _hypercall3(type, name, a1, a2, a3) \
7037 -({ \
7038 - type __res; \
7039 - long __ign1, __ign2, __ign3; \
7040 - asm volatile ( \
7041 - HYPERCALL_STR(name) \
7042 - : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
7043 - "=d" (__ign3) \
7044 - : "1" ((long)(a1)), "2" ((long)(a2)), \
7045 - "3" ((long)(a3)) \
7046 - : "memory" ); \
7047 - __res; \
7048 -})
7049 -
7050 -#define _hypercall4(type, name, a1, a2, a3, a4) \
7051 -({ \
7052 - type __res; \
7053 - long __ign1, __ign2, __ign3; \
7054 - register long __arg4 asm("r10") = (long)(a4); \
7055 - asm volatile ( \
7056 - HYPERCALL_STR(name) \
7057 - : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
7058 - "=d" (__ign3), "+r" (__arg4) \
7059 - : "1" ((long)(a1)), "2" ((long)(a2)), \
7060 - "3" ((long)(a3)) \
7061 - : "memory" ); \
7062 - __res; \
7063 -})
7064 -
7065 -#define _hypercall5(type, name, a1, a2, a3, a4, a5) \
7066 -({ \
7067 - type __res; \
7068 - long __ign1, __ign2, __ign3; \
7069 - register long __arg4 asm("r10") = (long)(a4); \
7070 - register long __arg5 asm("r8") = (long)(a5); \
7071 - asm volatile ( \
7072 - HYPERCALL_STR(name) \
7073 - : "=a" (__res), "=D" (__ign1), "=S" (__ign2), \
7074 - "=d" (__ign3), "+r" (__arg4), "+r" (__arg5) \
7075 - : "1" ((long)(a1)), "2" ((long)(a2)), \
7076 - "3" ((long)(a3)) \
7077 - : "memory" ); \
7078 - __res; \
7079 -})
7080 -
7081 -#define _hypercall(type, op, a1, a2, a3, a4, a5) \
7082 -({ \
7083 - type __res; \
7084 - register typeof((a1)+0) __arg1 asm("rdi") = (a1); \
7085 - register typeof((a2)+0) __arg2 asm("rsi") = (a2); \
7086 - register typeof((a3)+0) __arg3 asm("rdx") = (a3); \
7087 - register typeof((a4)+0) __arg4 asm("r10") = (a4); \
7088 - register typeof((a5)+0) __arg5 asm("r8") = (a5); \
7089 - asm volatile ( \
7090 - "call *%6" \
7091 - : "=a" (__res), "+r" (__arg1), "+r" (__arg2), \
7092 - "+r" (__arg3), "+r" (__arg4), "+r" (__arg5) \
7093 - : "0" (hypercall_page + (op) * 32) \
7094 - : "memory" ); \
7095 - __res; \
7096 -})
7097 -
7098 -static inline int __must_check
7099 -HYPERVISOR_set_trap_table(
7100 - const trap_info_t *table)
7101 -{
7102 - return _hypercall1(int, set_trap_table, table);
7103 -}
7104 -
7105 -static inline int __must_check
7106 -HYPERVISOR_mmu_update(
7107 - mmu_update_t *req, unsigned int count, unsigned int *success_count,
7108 - domid_t domid)
7109 -{
7110 - if (arch_use_lazy_mmu_mode())
7111 - return xen_multi_mmu_update(req, count, success_count, domid);
7112 - return _hypercall4(int, mmu_update, req, count, success_count, domid);
7113 -}
7114 -
7115 -static inline int __must_check
7116 -HYPERVISOR_mmuext_op(
7117 - struct mmuext_op *op, unsigned int count, unsigned int *success_count,
7118 - domid_t domid)
7119 -{
7120 - if (arch_use_lazy_mmu_mode())
7121 - return xen_multi_mmuext_op(op, count, success_count, domid);
7122 - return _hypercall4(int, mmuext_op, op, count, success_count, domid);
7123 -}
7124 -
7125 -static inline int __must_check
7126 -HYPERVISOR_set_gdt(
7127 - unsigned long *frame_list, unsigned int entries)
7128 -{
7129 - return _hypercall2(int, set_gdt, frame_list, entries);
7130 -}
7131 -
7132 -static inline int __must_check
7133 -HYPERVISOR_stack_switch(
7134 - unsigned long ss, unsigned long esp)
7135 -{
7136 - return _hypercall2(int, stack_switch, ss, esp);
7137 -}
7138 +#define HYPERCALL_arg1 "rdi"
7139 +#define HYPERCALL_arg2 "rsi"
7140 +#define HYPERCALL_arg3 "rdx"
7141 +#define HYPERCALL_arg4 "r10"
7142 +#define HYPERCALL_arg5 "r8"
7143
7144 +#if CONFIG_XEN_COMPAT <= 0x030002
7145 static inline int __must_check
7146 HYPERVISOR_set_callbacks(
7147 unsigned long event_address, unsigned long failsafe_address,
7148 @@ -200,27 +13,7 @@ HYPERVISOR_set_callbacks(
7149 return _hypercall3(int, set_callbacks,
7150 event_address, failsafe_address, syscall_address);
7151 }
7152 -
7153 -static inline int
7154 -HYPERVISOR_fpu_taskswitch(
7155 - int set)
7156 -{
7157 - return _hypercall1(int, fpu_taskswitch, set);
7158 -}
7159 -
7160 -static inline int __must_check
7161 -HYPERVISOR_sched_op_compat(
7162 - int cmd, unsigned long arg)
7163 -{
7164 - return _hypercall2(int, sched_op_compat, cmd, arg);
7165 -}
7166 -
7167 -static inline int __must_check
7168 -HYPERVISOR_sched_op(
7169 - int cmd, void *arg)
7170 -{
7171 - return _hypercall2(int, sched_op, cmd, arg);
7172 -}
7173 +#endif
7174
7175 static inline long __must_check
7176 HYPERVISOR_set_timer_op(
7177 @@ -230,28 +23,6 @@ HYPERVISOR_set_timer_op(
7178 }
7179
7180 static inline int __must_check
7181 -HYPERVISOR_platform_op(
7182 - struct xen_platform_op *platform_op)
7183 -{
7184 - platform_op->interface_version = XENPF_INTERFACE_VERSION;
7185 - return _hypercall1(int, platform_op, platform_op);
7186 -}
7187 -
7188 -static inline int __must_check
7189 -HYPERVISOR_set_debugreg(
7190 - unsigned int reg, unsigned long value)
7191 -{
7192 - return _hypercall2(int, set_debugreg, reg, value);
7193 -}
7194 -
7195 -static inline unsigned long __must_check
7196 -HYPERVISOR_get_debugreg(
7197 - unsigned int reg)
7198 -{
7199 - return _hypercall1(unsigned long, get_debugreg, reg);
7200 -}
7201 -
7202 -static inline int __must_check
7203 HYPERVISOR_update_descriptor(
7204 unsigned long ma, unsigned long word)
7205 {
7206 @@ -259,22 +30,6 @@ HYPERVISOR_update_descriptor(
7207 }
7208
7209 static inline int __must_check
7210 -HYPERVISOR_memory_op(
7211 - unsigned int cmd, void *arg)
7212 -{
7213 - if (arch_use_lazy_mmu_mode())
7214 - xen_multicall_flush(false);
7215 - return _hypercall2(int, memory_op, cmd, arg);
7216 -}
7217 -
7218 -static inline int __must_check
7219 -HYPERVISOR_multicall(
7220 - multicall_entry_t *call_list, unsigned int nr_calls)
7221 -{
7222 - return _hypercall2(int, multicall, call_list, nr_calls);
7223 -}
7224 -
7225 -static inline int __must_check
7226 HYPERVISOR_update_va_mapping(
7227 unsigned long va, pte_t new_val, unsigned long flags)
7228 {
7229 @@ -284,67 +39,6 @@ HYPERVISOR_update_va_mapping(
7230 }
7231
7232 static inline int __must_check
7233 -HYPERVISOR_event_channel_op(
7234 - int cmd, void *arg)
7235 -{
7236 - int rc = _hypercall2(int, event_channel_op, cmd, arg);
7237 -
7238 -#if CONFIG_XEN_COMPAT <= 0x030002
7239 - if (unlikely(rc == -ENOSYS)) {
7240 - struct evtchn_op op;
7241 - op.cmd = cmd;
7242 - memcpy(&op.u, arg, sizeof(op.u));
7243 - rc = _hypercall1(int, event_channel_op_compat, &op);
7244 - memcpy(arg, &op.u, sizeof(op.u));
7245 - }
7246 -#endif
7247 -
7248 - return rc;
7249 -}
7250 -
7251 -static inline int __must_check
7252 -HYPERVISOR_xen_version(
7253 - int cmd, void *arg)
7254 -{
7255 - return _hypercall2(int, xen_version, cmd, arg);
7256 -}
7257 -
7258 -static inline int __must_check
7259 -HYPERVISOR_console_io(
7260 - int cmd, unsigned int count, char *str)
7261 -{
7262 - return _hypercall3(int, console_io, cmd, count, str);
7263 -}
7264 -
7265 -static inline int __must_check
7266 -HYPERVISOR_physdev_op(
7267 - int cmd, void *arg)
7268 -{
7269 - int rc = _hypercall2(int, physdev_op, cmd, arg);
7270 -
7271 -#if CONFIG_XEN_COMPAT <= 0x030002
7272 - if (unlikely(rc == -ENOSYS)) {
7273 - struct physdev_op op;
7274 - op.cmd = cmd;
7275 - memcpy(&op.u, arg, sizeof(op.u));
7276 - rc = _hypercall1(int, physdev_op_compat, &op);
7277 - memcpy(arg, &op.u, sizeof(op.u));
7278 - }
7279 -#endif
7280 -
7281 - return rc;
7282 -}
7283 -
7284 -static inline int __must_check
7285 -HYPERVISOR_grant_table_op(
7286 - unsigned int cmd, void *uop, unsigned int count)
7287 -{
7288 - if (arch_use_lazy_mmu_mode())
7289 - xen_multicall_flush(false);
7290 - return _hypercall3(int, grant_table_op, cmd, uop, count);
7291 -}
7292 -
7293 -static inline int __must_check
7294 HYPERVISOR_update_va_mapping_otherdomain(
7295 unsigned long va, pte_t new_val, unsigned long flags, domid_t domid)
7296 {
7297 @@ -353,83 +47,8 @@ HYPERVISOR_update_va_mapping_otherdomain
7298 }
7299
7300 static inline int __must_check
7301 -HYPERVISOR_vm_assist(
7302 - unsigned int cmd, unsigned int type)
7303 -{
7304 - return _hypercall2(int, vm_assist, cmd, type);
7305 -}
7306 -
7307 -static inline int __must_check
7308 -HYPERVISOR_vcpu_op(
7309 - int cmd, unsigned int vcpuid, void *extra_args)
7310 -{
7311 - return _hypercall3(int, vcpu_op, cmd, vcpuid, extra_args);
7312 -}
7313 -
7314 -static inline int __must_check
7315 HYPERVISOR_set_segment_base(
7316 int reg, unsigned long value)
7317 {
7318 return _hypercall2(int, set_segment_base, reg, value);
7319 }
7320 -
7321 -static inline int __must_check
7322 -HYPERVISOR_suspend(
7323 - unsigned long srec)
7324 -{
7325 - struct sched_shutdown sched_shutdown = {
7326 - .reason = SHUTDOWN_suspend
7327 - };
7328 -
7329 - int rc = _hypercall3(int, sched_op, SCHEDOP_shutdown,
7330 - &sched_shutdown, srec);
7331 -
7332 -#if CONFIG_XEN_COMPAT <= 0x030002
7333 - if (rc == -ENOSYS)
7334 - rc = _hypercall3(int, sched_op_compat, SCHEDOP_shutdown,
7335 - SHUTDOWN_suspend, srec);
7336 -#endif
7337 -
7338 - return rc;
7339 -}
7340 -
7341 -#if CONFIG_XEN_COMPAT <= 0x030002
7342 -static inline int
7343 -HYPERVISOR_nmi_op(
7344 - unsigned long op, void *arg)
7345 -{
7346 - return _hypercall2(int, nmi_op, op, arg);
7347 -}
7348 -#endif
7349 -
7350 -#ifndef CONFIG_XEN
7351 -static inline unsigned long __must_check
7352 -HYPERVISOR_hvm_op(
7353 - int op, void *arg)
7354 -{
7355 - return _hypercall2(unsigned long, hvm_op, op, arg);
7356 -}
7357 -#endif
7358 -
7359 -static inline int __must_check
7360 -HYPERVISOR_callback_op(
7361 - int cmd, const void *arg)
7362 -{
7363 - return _hypercall2(int, callback_op, cmd, arg);
7364 -}
7365 -
7366 -static inline int __must_check
7367 -HYPERVISOR_xenoprof_op(
7368 - int op, void *arg)
7369 -{
7370 - return _hypercall2(int, xenoprof_op, op, arg);
7371 -}
7372 -
7373 -static inline int __must_check
7374 -HYPERVISOR_kexec_op(
7375 - unsigned long op, void *args)
7376 -{
7377 - return _hypercall2(int, kexec_op, op, args);
7378 -}
7379 -
7380 -#endif /* __HYPERCALL_H__ */
7381 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/hypervisor.h 2009-03-04 11:28:11.000000000 +0100
7382 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/hypervisor.h 2009-02-16 16:18:36.000000000 +0100
7383 @@ -194,7 +194,6 @@ static inline void xen_multicall_flush(b
7384 extern char hypercall_page[PAGE_SIZE];
7385 #else
7386 extern char *hypercall_stubs;
7387 -#define hypercall_page hypercall_stubs
7388 #define is_running_on_xen() (!!hypercall_stubs)
7389 #endif
7390
7391 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7392 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/io.h 2009-02-16 16:18:36.000000000 +0100
7393 @@ -0,0 +1,5 @@
7394 +#ifdef CONFIG_X86_32
7395 +# include "io_32.h"
7396 +#else
7397 +# include "io_64.h"
7398 +#endif
7399 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7400 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/irqflags.h 2009-02-16 16:18:36.000000000 +0100
7401 @@ -0,0 +1,5 @@
7402 +#ifdef CONFIG_X86_32
7403 +# include "irqflags_32.h"
7404 +#else
7405 +# include "irqflags_64.h"
7406 +#endif
7407 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/irqflags_32.h 2008-12-15 11:27:22.000000000 +0100
7408 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/irqflags_32.h 2009-02-16 16:18:36.000000000 +0100
7409 @@ -148,6 +148,23 @@ static inline int raw_irqs_disabled_flag
7410 \
7411 raw_irqs_disabled_flags(flags); \
7412 })
7413 +
7414 +/*
7415 + * makes the traced hardirq state match with the machine state
7416 + *
7417 + * should be a rarely used function, only in places where its
7418 + * otherwise impossible to know the irq state, like in traps.
7419 + */
7420 +static inline void trace_hardirqs_fixup_flags(unsigned long flags)
7421 +{
7422 + if (raw_irqs_disabled_flags(flags))
7423 + trace_hardirqs_off();
7424 + else
7425 + trace_hardirqs_on();
7426 +}
7427 +
7428 +#define trace_hardirqs_fixup() \
7429 + trace_hardirqs_fixup_flags(__raw_local_save_flags())
7430 #endif /* __ASSEMBLY__ */
7431
7432 /*
7433 @@ -179,4 +196,17 @@ static inline int raw_irqs_disabled_flag
7434 # define TRACE_IRQS_OFF
7435 #endif
7436
7437 +#ifdef CONFIG_DEBUG_LOCK_ALLOC
7438 +# define LOCKDEP_SYS_EXIT \
7439 + pushl %eax; \
7440 + pushl %ecx; \
7441 + pushl %edx; \
7442 + call lockdep_sys_exit; \
7443 + popl %edx; \
7444 + popl %ecx; \
7445 + popl %eax;
7446 +#else
7447 +# define LOCKDEP_SYS_EXIT
7448 +#endif
7449 +
7450 #endif
7451 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/irqflags_64.h 2008-12-15 11:27:22.000000000 +0100
7452 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/irqflags_64.h 2009-02-16 16:18:36.000000000 +0100
7453 @@ -116,6 +116,22 @@ static inline int raw_irqs_disabled_flag
7454 })
7455
7456 /*
7457 + * makes the traced hardirq state match with the machine state
7458 + *
7459 + * should be a rarely used function, only in places where its
7460 + * otherwise impossible to know the irq state, like in traps.
7461 + */
7462 +static inline void trace_hardirqs_fixup_flags(unsigned long flags)
7463 +{
7464 + if (raw_irqs_disabled_flags(flags))
7465 + trace_hardirqs_off();
7466 + else
7467 + trace_hardirqs_on();
7468 +}
7469 +
7470 +#define trace_hardirqs_fixup() \
7471 + trace_hardirqs_fixup_flags(__raw_local_save_flags())
7472 +/*
7473 * Used in the idle loop; sti takes one instruction cycle
7474 * to complete:
7475 */
7476 @@ -143,6 +159,20 @@ static inline void halt(void)
7477 # define TRACE_IRQS_ON
7478 # define TRACE_IRQS_OFF
7479 # endif
7480 +# ifdef CONFIG_DEBUG_LOCK_ALLOC
7481 +# define LOCKDEP_SYS_EXIT call lockdep_sys_exit_thunk
7482 +# define LOCKDEP_SYS_EXIT_IRQ \
7483 + TRACE_IRQS_ON; \
7484 + sti; \
7485 + SAVE_REST; \
7486 + LOCKDEP_SYS_EXIT; \
7487 + RESTORE_REST; \
7488 + cli; \
7489 + TRACE_IRQS_OFF;
7490 +# else
7491 +# define LOCKDEP_SYS_EXIT
7492 +# define LOCKDEP_SYS_EXIT_IRQ
7493 +# endif
7494 #endif
7495
7496 #endif
7497 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7498 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/maddr.h 2009-02-16 16:18:36.000000000 +0100
7499 @@ -0,0 +1,5 @@
7500 +#ifdef CONFIG_X86_32
7501 +# include "maddr_32.h"
7502 +#else
7503 +# include "maddr_64.h"
7504 +#endif
7505 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7506 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/mmu_context.h 2009-02-16 16:18:36.000000000 +0100
7507 @@ -0,0 +1,5 @@
7508 +#ifdef CONFIG_X86_32
7509 +# include "mmu_context_32.h"
7510 +#else
7511 +# include "mmu_context_64.h"
7512 +#endif
7513 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7514 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/page.h 2009-02-16 16:18:36.000000000 +0100
7515 @@ -0,0 +1,13 @@
7516 +#ifdef __KERNEL__
7517 +# ifdef CONFIG_X86_32
7518 +# include "page_32.h"
7519 +# else
7520 +# include "page_64.h"
7521 +# endif
7522 +#else
7523 +# ifdef __i386__
7524 +# include "page_32.h"
7525 +# else
7526 +# include "page_64.h"
7527 +# endif
7528 +#endif
7529 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/page_64.h 2009-02-16 16:17:21.000000000 +0100
7530 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/page_64.h 2009-02-16 16:18:36.000000000 +0100
7531 @@ -207,6 +207,7 @@ static inline unsigned long __phys_addr(
7532 VM_READ | VM_WRITE | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
7533
7534 #define __HAVE_ARCH_GATE_AREA 1
7535 +#define vmemmap ((struct page *)VMEMMAP_START)
7536
7537 #include <asm-generic/memory_model.h>
7538 #include <asm-generic/page.h>
7539 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7540 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/pci.h 2009-02-16 16:18:36.000000000 +0100
7541 @@ -0,0 +1,100 @@
7542 +#ifndef __x86_PCI_H
7543 +#define __x86_PCI_H
7544 +
7545 +#include <linux/mm.h> /* for struct page */
7546 +#include <linux/types.h>
7547 +#include <linux/slab.h>
7548 +#include <linux/string.h>
7549 +#include <asm/scatterlist.h>
7550 +#include <asm/io.h>
7551 +
7552 +
7553 +#ifdef __KERNEL__
7554 +
7555 +struct pci_sysdata {
7556 + int domain; /* PCI domain */
7557 + int node; /* NUMA node */
7558 +#ifdef CONFIG_X86_64
7559 + void* iommu; /* IOMMU private data */
7560 +#endif
7561 +#ifdef CONFIG_XEN_PCIDEV_FRONTEND
7562 + struct pcifront_device *pdev;
7563 +#endif
7564 +};
7565 +
7566 +/* scan a bus after allocating a pci_sysdata for it */
7567 +extern struct pci_bus *pci_scan_bus_with_sysdata(int busno);
7568 +
7569 +static inline int pci_domain_nr(struct pci_bus *bus)
7570 +{
7571 + struct pci_sysdata *sd = bus->sysdata;
7572 + return sd->domain;
7573 +}
7574 +
7575 +static inline int pci_proc_domain(struct pci_bus *bus)
7576 +{
7577 + return pci_domain_nr(bus);
7578 +}
7579 +
7580 +
7581 +/* Can be used to override the logic in pci_scan_bus for skipping
7582 + already-configured bus numbers - to be used for buggy BIOSes
7583 + or architectures with incomplete PCI setup by the loader */
7584 +
7585 +#ifdef CONFIG_PCI
7586 +extern unsigned int pcibios_assign_all_busses(void);
7587 +#else
7588 +#define pcibios_assign_all_busses() 0
7589 +#endif
7590 +
7591 +#include <asm/hypervisor.h>
7592 +#define pcibios_scan_all_fns(a, b) (!is_initial_xendomain())
7593 +
7594 +extern unsigned long pci_mem_start;
7595 +#define PCIBIOS_MIN_IO 0x1000
7596 +#define PCIBIOS_MIN_MEM (pci_mem_start)
7597 +
7598 +#define PCIBIOS_MIN_CARDBUS_IO 0x4000
7599 +
7600 +void pcibios_config_init(void);
7601 +struct pci_bus * pcibios_scan_root(int bus);
7602 +
7603 +void pcibios_set_master(struct pci_dev *dev);
7604 +void pcibios_penalize_isa_irq(int irq, int active);
7605 +struct irq_routing_table *pcibios_get_irq_routing_table(void);
7606 +int pcibios_set_irq_routing(struct pci_dev *dev, int pin, int irq);
7607 +
7608 +
7609 +#define HAVE_PCI_MMAP
7610 +extern int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
7611 + enum pci_mmap_state mmap_state, int write_combine);
7612 +
7613 +
7614 +#ifdef CONFIG_PCI
7615 +static inline void pci_dma_burst_advice(struct pci_dev *pdev,
7616 + enum pci_dma_burst_strategy *strat,
7617 + unsigned long *strategy_parameter)
7618 +{
7619 + *strat = PCI_DMA_BURST_INFINITY;
7620 + *strategy_parameter = ~0UL;
7621 +}
7622 +#endif
7623 +
7624 +
7625 +#endif /* __KERNEL__ */
7626 +
7627 +#ifdef CONFIG_X86_32
7628 +# include "pci_32.h"
7629 +#else
7630 +# include "pci_64.h"
7631 +#endif
7632 +
7633 +/* implement the pci_ DMA API in terms of the generic device dma_ one */
7634 +#include <asm-generic/pci-dma-compat.h>
7635 +
7636 +/* generic pci stuff */
7637 +#include <asm-generic/pci.h>
7638 +
7639 +
7640 +
7641 +#endif
7642 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/pci_32.h 2009-02-16 16:17:21.000000000 +0100
7643 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/pci_32.h 2009-02-16 16:18:36.000000000 +0100
7644 @@ -4,52 +4,10 @@
7645
7646 #ifdef __KERNEL__
7647
7648 -struct pci_sysdata {
7649 - int node; /* NUMA node */
7650 -};
7651 -
7652 -/* scan a bus after allocating a pci_sysdata for it */
7653 -extern struct pci_bus *pci_scan_bus_with_sysdata(int busno);
7654 -
7655 -#include <linux/mm.h> /* for struct page */
7656 -
7657 -/* Can be used to override the logic in pci_scan_bus for skipping
7658 - already-configured bus numbers - to be used for buggy BIOSes
7659 - or architectures with incomplete PCI setup by the loader */
7660 -
7661 -#ifdef CONFIG_PCI
7662 -extern unsigned int pcibios_assign_all_busses(void);
7663 -#else
7664 -#define pcibios_assign_all_busses() 0
7665 -#endif
7666 -
7667 -#include <asm/hypervisor.h>
7668 -#define pcibios_scan_all_fns(a, b) (!is_initial_xendomain())
7669 -
7670 -extern unsigned long pci_mem_start;
7671 -#define PCIBIOS_MIN_IO 0x1000
7672 -#define PCIBIOS_MIN_MEM (pci_mem_start)
7673 -
7674 -#define PCIBIOS_MIN_CARDBUS_IO 0x4000
7675 -
7676 -void pcibios_config_init(void);
7677 -struct pci_bus * pcibios_scan_root(int bus);
7678 -
7679 -void pcibios_set_master(struct pci_dev *dev);
7680 -void pcibios_penalize_isa_irq(int irq, int active);
7681 -struct irq_routing_table *pcibios_get_irq_routing_table(void);
7682 -int pcibios_set_irq_routing(struct pci_dev *dev, int pin, int irq);
7683 -
7684 /* Dynamic DMA mapping stuff.
7685 * i386 has everything mapped statically.
7686 */
7687
7688 -#include <linux/types.h>
7689 -#include <linux/slab.h>
7690 -#include <asm/scatterlist.h>
7691 -#include <linux/string.h>
7692 -#include <asm/io.h>
7693 -
7694 struct pci_dev;
7695
7696 #ifdef CONFIG_SWIOTLB
7697 @@ -89,31 +47,8 @@ struct pci_dev;
7698
7699 #endif
7700
7701 -#define HAVE_PCI_MMAP
7702 -extern int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
7703 - enum pci_mmap_state mmap_state, int write_combine);
7704 -
7705 -
7706 -#ifdef CONFIG_PCI
7707 -static inline void pci_dma_burst_advice(struct pci_dev *pdev,
7708 - enum pci_dma_burst_strategy *strat,
7709 - unsigned long *strategy_parameter)
7710 -{
7711 - *strat = PCI_DMA_BURST_INFINITY;
7712 - *strategy_parameter = ~0UL;
7713 -}
7714 -#endif
7715
7716 #endif /* __KERNEL__ */
7717
7718 -#ifdef CONFIG_XEN_PCIDEV_FRONTEND
7719 -#include <xen/pcifront.h>
7720 -#endif /* CONFIG_XEN_PCIDEV_FRONTEND */
7721 -
7722 -/* implement the pci_ DMA API in terms of the generic device dma_ one */
7723 -#include <asm-generic/pci-dma-compat.h>
7724 -
7725 -/* generic pci stuff */
7726 -#include <asm-generic/pci.h>
7727
7728 #endif /* __i386_PCI_H */
7729 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7730 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/pgalloc.h 2009-02-16 16:18:36.000000000 +0100
7731 @@ -0,0 +1,5 @@
7732 +#ifdef CONFIG_X86_32
7733 +# include "pgalloc_32.h"
7734 +#else
7735 +# include "pgalloc_64.h"
7736 +#endif
7737 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/pgalloc_64.h 2009-02-16 16:17:21.000000000 +0100
7738 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/pgalloc_64.h 2009-02-16 16:18:36.000000000 +0100
7739 @@ -112,6 +112,8 @@ static inline void pgd_list_del(pgd_t *p
7740 spin_unlock(&pgd_lock);
7741 }
7742
7743 +extern void pgd_test_and_unpin(pgd_t *);
7744 +
7745 static inline pgd_t *pgd_alloc(struct mm_struct *mm)
7746 {
7747 /*
7748 @@ -122,6 +124,7 @@ static inline pgd_t *pgd_alloc(struct mm
7749 if (!pgd)
7750 return NULL;
7751 pgd_list_add(pgd);
7752 + pgd_test_and_unpin(pgd);
7753 /*
7754 * Copy kernel pointers in from init.
7755 * Could keep a freelist or slab cache of those because the kernel
7756 @@ -144,27 +147,7 @@ static inline pgd_t *pgd_alloc(struct mm
7757
7758 static inline void pgd_free(pgd_t *pgd)
7759 {
7760 - pte_t *ptep = virt_to_ptep(pgd);
7761 -
7762 - if (!pte_write(*ptep)) {
7763 - xen_pgd_unpin(__pa(pgd));
7764 - BUG_ON(HYPERVISOR_update_va_mapping(
7765 - (unsigned long)pgd,
7766 - pfn_pte(virt_to_phys(pgd)>>PAGE_SHIFT, PAGE_KERNEL),
7767 - 0));
7768 - }
7769 -
7770 - ptep = virt_to_ptep(__user_pgd(pgd));
7771 -
7772 - if (!pte_write(*ptep)) {
7773 - xen_pgd_unpin(__pa(__user_pgd(pgd)));
7774 - BUG_ON(HYPERVISOR_update_va_mapping(
7775 - (unsigned long)__user_pgd(pgd),
7776 - pfn_pte(virt_to_phys(__user_pgd(pgd))>>PAGE_SHIFT,
7777 - PAGE_KERNEL),
7778 - 0));
7779 - }
7780 -
7781 + pgd_test_and_unpin(pgd);
7782 pgd_list_del(pgd);
7783 free_pages((unsigned long)pgd, 1);
7784 }
7785 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7786 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/pgtable.h 2009-02-16 16:18:36.000000000 +0100
7787 @@ -0,0 +1,5 @@
7788 +#ifdef CONFIG_X86_32
7789 +# include "pgtable_32.h"
7790 +#else
7791 +# include "pgtable_64.h"
7792 +#endif
7793 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/pgtable_32.h 2009-02-16 16:17:21.000000000 +0100
7794 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/pgtable_32.h 2009-02-16 16:18:36.000000000 +0100
7795 @@ -17,10 +17,7 @@
7796 #include <asm/fixmap.h>
7797 #include <linux/threads.h>
7798
7799 -#ifndef _I386_BITOPS_H
7800 -#include <asm/bitops.h>
7801 -#endif
7802 -
7803 +#include <linux/bitops.h>
7804 #include <linux/slab.h>
7805 #include <linux/list.h>
7806 #include <linux/spinlock.h>
7807 @@ -40,7 +37,7 @@ extern spinlock_t pgd_lock;
7808 extern struct page *pgd_list;
7809 void check_pgt_cache(void);
7810
7811 -void pmd_ctor(void *, struct kmem_cache *, unsigned long);
7812 +void pmd_ctor(struct kmem_cache *, void *);
7813 void pgtable_cache_init(void);
7814 void paging_init(void);
7815
7816 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/pgtable_64.h 2009-02-16 16:17:21.000000000 +0100
7817 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/pgtable_64.h 2009-02-16 16:18:36.000000000 +0100
7818 @@ -9,7 +9,7 @@
7819 * the x86-64 page table tree.
7820 */
7821 #include <asm/processor.h>
7822 -#include <asm/bitops.h>
7823 +#include <linux/bitops.h>
7824 #include <linux/threads.h>
7825 #include <linux/sched.h>
7826 #include <asm/pda.h>
7827 @@ -139,6 +139,7 @@ static inline void pgd_clear (pgd_t * pg
7828 #define MAXMEM _AC(0x3fffffffffff, UL)
7829 #define VMALLOC_START _AC(0xffffc20000000000, UL)
7830 #define VMALLOC_END _AC(0xffffe1ffffffffff, UL)
7831 +#define VMEMMAP_START _AC(0xffffe20000000000, UL)
7832 #define MODULES_VADDR _AC(0xffffffff88000000, UL)
7833 #define MODULES_END _AC(0xfffffffffff00000, UL)
7834 #define MODULES_LEN (MODULES_END - MODULES_VADDR)
7835 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
7836 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/processor.h 2009-02-16 16:18:36.000000000 +0100
7837 @@ -0,0 +1,5 @@
7838 +#ifdef CONFIG_X86_32
7839 +# include "processor_32.h"
7840 +#else
7841 +# include "processor_64.h"
7842 +#endif
7843 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/processor_32.h 2009-02-16 16:17:21.000000000 +0100
7844 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/processor_32.h 2009-02-16 16:18:36.000000000 +0100
7845 @@ -80,6 +80,7 @@ struct cpuinfo_x86 {
7846 unsigned char booted_cores; /* number of cores as seen by OS */
7847 __u8 phys_proc_id; /* Physical processor id. */
7848 __u8 cpu_core_id; /* Core id */
7849 + __u8 cpu_index; /* index into per_cpu list */
7850 #endif
7851 } __attribute__((__aligned__(SMP_CACHE_BYTES)));
7852
7853 @@ -106,14 +107,19 @@ DECLARE_PER_CPU(struct tss_struct, init_
7854 #endif
7855
7856 #ifdef CONFIG_SMP
7857 -extern struct cpuinfo_x86 cpu_data[];
7858 -#define current_cpu_data cpu_data[smp_processor_id()]
7859 +DECLARE_PER_CPU(struct cpuinfo_x86, cpu_info);
7860 +#define cpu_data(cpu) per_cpu(cpu_info, cpu)
7861 +#define current_cpu_data cpu_data(smp_processor_id())
7862 #else
7863 -#define cpu_data (&boot_cpu_data)
7864 -#define current_cpu_data boot_cpu_data
7865 +#define cpu_data(cpu) boot_cpu_data
7866 +#define current_cpu_data boot_cpu_data
7867 #endif
7868
7869 -extern int cpu_llc_id[NR_CPUS];
7870 +/*
7871 + * the following now lives in the per cpu area:
7872 + * extern int cpu_llc_id[NR_CPUS];
7873 + */
7874 +DECLARE_PER_CPU(u8, cpu_llc_id);
7875 extern char ignore_fpu_irq;
7876
7877 void __init cpu_detect(struct cpuinfo_x86 *c);
7878 @@ -560,7 +566,9 @@ static inline void xen_set_iopl_mask(uns
7879 * clear %ecx since some cpus (Cyrix MII) do not set or clear %ecx
7880 * resulting in stale register contents being returned.
7881 */
7882 -static inline void cpuid(unsigned int op, unsigned int *eax, unsigned int *ebx, unsigned int *ecx, unsigned int *edx)
7883 +static inline void cpuid(unsigned int op,
7884 + unsigned int *eax, unsigned int *ebx,
7885 + unsigned int *ecx, unsigned int *edx)
7886 {
7887 *eax = op;
7888 *ecx = 0;
7889 @@ -568,8 +576,9 @@ static inline void cpuid(unsigned int op
7890 }
7891
7892 /* Some CPUID calls want 'count' to be placed in ecx */
7893 -static inline void cpuid_count(int op, int count, int *eax, int *ebx, int *ecx,
7894 - int *edx)
7895 +static inline void cpuid_count(unsigned int op, int count,
7896 + unsigned int *eax, unsigned int *ebx,
7897 + unsigned int *ecx, unsigned int *edx)
7898 {
7899 *eax = op;
7900 *ecx = count;
7901 @@ -639,6 +648,17 @@ static inline unsigned int cpuid_edx(uns
7902 #define K7_NOP7 ".byte 0x8D,0x04,0x05,0,0,0,0\n"
7903 #define K7_NOP8 K7_NOP7 ASM_NOP1
7904
7905 +/* P6 nops */
7906 +/* uses eax dependencies (Intel-recommended choice) */
7907 +#define P6_NOP1 GENERIC_NOP1
7908 +#define P6_NOP2 ".byte 0x66,0x90\n"
7909 +#define P6_NOP3 ".byte 0x0f,0x1f,0x00\n"
7910 +#define P6_NOP4 ".byte 0x0f,0x1f,0x40,0\n"
7911 +#define P6_NOP5 ".byte 0x0f,0x1f,0x44,0x00,0\n"
7912 +#define P6_NOP6 ".byte 0x66,0x0f,0x1f,0x44,0x00,0\n"
7913 +#define P6_NOP7 ".byte 0x0f,0x1f,0x80,0,0,0,0\n"
7914 +#define P6_NOP8 ".byte 0x0f,0x1f,0x84,0x00,0,0,0,0\n"
7915 +
7916 #ifdef CONFIG_MK8
7917 #define ASM_NOP1 K8_NOP1
7918 #define ASM_NOP2 K8_NOP2
7919 @@ -657,6 +677,17 @@ static inline unsigned int cpuid_edx(uns
7920 #define ASM_NOP6 K7_NOP6
7921 #define ASM_NOP7 K7_NOP7
7922 #define ASM_NOP8 K7_NOP8
7923 +#elif defined(CONFIG_M686) || defined(CONFIG_MPENTIUMII) || \
7924 + defined(CONFIG_MPENTIUMIII) || defined(CONFIG_MPENTIUMM) || \
7925 + defined(CONFIG_MCORE2) || defined(CONFIG_PENTIUM4)
7926 +#define ASM_NOP1 P6_NOP1
7927 +#define ASM_NOP2 P6_NOP2
7928 +#define ASM_NOP3 P6_NOP3
7929 +#define ASM_NOP4 P6_NOP4
7930 +#define ASM_NOP5 P6_NOP5
7931 +#define ASM_NOP6 P6_NOP6
7932 +#define ASM_NOP7 P6_NOP7
7933 +#define ASM_NOP8 P6_NOP8
7934 #else
7935 #define ASM_NOP1 GENERIC_NOP1
7936 #define ASM_NOP2 GENERIC_NOP2
7937 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/processor_64.h 2009-02-16 16:17:21.000000000 +0100
7938 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/processor_64.h 2009-02-16 16:18:36.000000000 +0100
7939 @@ -74,6 +74,7 @@ struct cpuinfo_x86 {
7940 __u8 booted_cores; /* number of cores as seen by OS */
7941 __u8 phys_proc_id; /* Physical Processor id. */
7942 __u8 cpu_core_id; /* Core id. */
7943 + __u8 cpu_index; /* index into per_cpu list */
7944 #endif
7945 } ____cacheline_aligned;
7946
7947 @@ -88,11 +89,12 @@ struct cpuinfo_x86 {
7948 #define X86_VENDOR_UNKNOWN 0xff
7949
7950 #ifdef CONFIG_SMP
7951 -extern struct cpuinfo_x86 cpu_data[];
7952 -#define current_cpu_data cpu_data[smp_processor_id()]
7953 +DECLARE_PER_CPU(struct cpuinfo_x86, cpu_info);
7954 +#define cpu_data(cpu) per_cpu(cpu_info, cpu)
7955 +#define current_cpu_data cpu_data(smp_processor_id())
7956 #else
7957 -#define cpu_data (&boot_cpu_data)
7958 -#define current_cpu_data boot_cpu_data
7959 +#define cpu_data(cpu) boot_cpu_data
7960 +#define current_cpu_data boot_cpu_data
7961 #endif
7962
7963 extern char ignore_irq13;
7964 @@ -343,6 +345,16 @@ struct extended_sigtable {
7965 };
7966
7967
7968 +#if defined(CONFIG_MPSC) || defined(CONFIG_MCORE2)
7969 +#define ASM_NOP1 P6_NOP1
7970 +#define ASM_NOP2 P6_NOP2
7971 +#define ASM_NOP3 P6_NOP3
7972 +#define ASM_NOP4 P6_NOP4
7973 +#define ASM_NOP5 P6_NOP5
7974 +#define ASM_NOP6 P6_NOP6
7975 +#define ASM_NOP7 P6_NOP7
7976 +#define ASM_NOP8 P6_NOP8
7977 +#else
7978 #define ASM_NOP1 K8_NOP1
7979 #define ASM_NOP2 K8_NOP2
7980 #define ASM_NOP3 K8_NOP3
7981 @@ -351,6 +363,7 @@ struct extended_sigtable {
7982 #define ASM_NOP6 K8_NOP6
7983 #define ASM_NOP7 K8_NOP7
7984 #define ASM_NOP8 K8_NOP8
7985 +#endif
7986
7987 /* Opteron nops */
7988 #define K8_NOP1 ".byte 0x90\n"
7989 @@ -362,6 +375,17 @@ struct extended_sigtable {
7990 #define K8_NOP7 K8_NOP4 K8_NOP3
7991 #define K8_NOP8 K8_NOP4 K8_NOP4
7992
7993 +/* P6 nops */
7994 +/* uses eax dependencies (Intel-recommended choice) */
7995 +#define P6_NOP1 ".byte 0x90\n"
7996 +#define P6_NOP2 ".byte 0x66,0x90\n"
7997 +#define P6_NOP3 ".byte 0x0f,0x1f,0x00\n"
7998 +#define P6_NOP4 ".byte 0x0f,0x1f,0x40,0\n"
7999 +#define P6_NOP5 ".byte 0x0f,0x1f,0x44,0x00,0\n"
8000 +#define P6_NOP6 ".byte 0x66,0x0f,0x1f,0x44,0x00,0\n"
8001 +#define P6_NOP7 ".byte 0x0f,0x1f,0x80,0,0,0,0\n"
8002 +#define P6_NOP8 ".byte 0x0f,0x1f,0x84,0x00,0,0,0,0\n"
8003 +
8004 #define ASM_NOP_MAX 8
8005
8006 /* REP NOP (PAUSE) is a good thing to insert into busy-wait loops. */
8007 @@ -377,12 +401,6 @@ static inline void sync_core(void)
8008 asm volatile("cpuid" : "=a" (tmp) : "0" (1) : "ebx","ecx","edx","memory");
8009 }
8010
8011 -#define ARCH_HAS_PREFETCH
8012 -static inline void prefetch(void *x)
8013 -{
8014 - asm volatile("prefetcht0 (%0)" :: "r" (x));
8015 -}
8016 -
8017 #define ARCH_HAS_PREFETCHW 1
8018 static inline void prefetchw(void *x)
8019 {
8020 @@ -398,11 +416,6 @@ static inline void prefetchw(void *x)
8021
8022 #define cpu_relax() rep_nop()
8023
8024 -static inline void serialize_cpu(void)
8025 -{
8026 - __asm__ __volatile__ ("cpuid" : : : "ax", "bx", "cx", "dx");
8027 -}
8028 -
8029 static inline void __monitor(const void *eax, unsigned long ecx,
8030 unsigned long edx)
8031 {
8032 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8033 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/segment.h 2009-02-16 16:18:36.000000000 +0100
8034 @@ -0,0 +1,5 @@
8035 +#ifdef CONFIG_X86_32
8036 +# include "segment_32.h"
8037 +#else
8038 +# include "../../segment_64.h"
8039 +#endif
8040 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8041 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/smp.h 2009-02-16 16:18:36.000000000 +0100
8042 @@ -0,0 +1,5 @@
8043 +#ifdef CONFIG_X86_32
8044 +# include "smp_32.h"
8045 +#else
8046 +# include "smp_64.h"
8047 +#endif
8048 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/smp_32.h 2008-12-15 11:27:22.000000000 +0100
8049 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/smp_32.h 2009-02-16 16:18:36.000000000 +0100
8050 @@ -11,7 +11,7 @@
8051 #endif
8052
8053 #if defined(CONFIG_X86_LOCAL_APIC) && !defined(__ASSEMBLY__)
8054 -#include <asm/bitops.h>
8055 +#include <linux/bitops.h>
8056 #include <asm/mpspec.h>
8057 #include <asm/apic.h>
8058 #ifdef CONFIG_X86_IO_APIC
8059 @@ -30,8 +30,8 @@
8060 extern void smp_alloc_memory(void);
8061 extern int pic_mode;
8062 extern int smp_num_siblings;
8063 -extern cpumask_t cpu_sibling_map[];
8064 -extern cpumask_t cpu_core_map[];
8065 +DECLARE_PER_CPU(cpumask_t, cpu_sibling_map);
8066 +DECLARE_PER_CPU(cpumask_t, cpu_core_map);
8067
8068 extern void (*mtrr_hook) (void);
8069 extern void zap_low_mappings (void);
8070 @@ -39,9 +39,11 @@ extern void lock_ipi_call_lock(void);
8071 extern void unlock_ipi_call_lock(void);
8072
8073 #define MAX_APICID 256
8074 -extern u8 x86_cpu_to_apicid[];
8075 +extern u8 __initdata x86_cpu_to_apicid_init[];
8076 +extern void *x86_cpu_to_apicid_ptr;
8077 +DECLARE_PER_CPU(u8, x86_cpu_to_apicid);
8078
8079 -#define cpu_physical_id(cpu) x86_cpu_to_apicid[cpu]
8080 +#define cpu_physical_id(cpu) per_cpu(x86_cpu_to_apicid, cpu)
8081
8082 #ifdef CONFIG_HOTPLUG_CPU
8083 extern void cpu_exit_clear(void);
8084 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/smp_64.h 2008-12-15 11:27:22.000000000 +0100
8085 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/smp_64.h 2009-02-16 16:18:36.000000000 +0100
8086 @@ -40,10 +40,19 @@ extern void lock_ipi_call_lock(void);
8087 extern void unlock_ipi_call_lock(void);
8088 extern int smp_num_siblings;
8089 extern void smp_send_reschedule(int cpu);
8090 +extern int smp_call_function_mask(cpumask_t mask, void (*func)(void *),
8091 + void *info, int wait);
8092
8093 -extern cpumask_t cpu_sibling_map[NR_CPUS];
8094 -extern cpumask_t cpu_core_map[NR_CPUS];
8095 -extern u8 cpu_llc_id[NR_CPUS];
8096 +/*
8097 + * cpu_sibling_map and cpu_core_map now live
8098 + * in the per cpu area
8099 + *
8100 + * extern cpumask_t cpu_sibling_map[NR_CPUS];
8101 + * extern cpumask_t cpu_core_map[NR_CPUS];
8102 + */
8103 +DECLARE_PER_CPU(cpumask_t, cpu_sibling_map);
8104 +DECLARE_PER_CPU(cpumask_t, cpu_core_map);
8105 +DECLARE_PER_CPU(u8, cpu_llc_id);
8106
8107 #define SMP_TRAMPOLINE_BASE 0x6000
8108
8109 @@ -70,6 +79,8 @@ extern unsigned __cpuinitdata disabled_c
8110
8111 #endif /* CONFIG_SMP */
8112
8113 +#define safe_smp_processor_id() smp_processor_id()
8114 +
8115 #ifdef CONFIG_X86_LOCAL_APIC
8116 static inline int hard_smp_processor_id(void)
8117 {
8118 @@ -82,8 +93,9 @@ static inline int hard_smp_processor_id(
8119 * Some lowlevel functions might want to know about
8120 * the real APIC ID <-> CPU # mapping.
8121 */
8122 -extern u8 x86_cpu_to_apicid[NR_CPUS]; /* physical ID */
8123 -extern u8 x86_cpu_to_log_apicid[NR_CPUS];
8124 +extern u8 __initdata x86_cpu_to_apicid_init[];
8125 +extern void *x86_cpu_to_apicid_ptr;
8126 +DECLARE_PER_CPU(u8, x86_cpu_to_apicid); /* physical ID */
8127 extern u8 bios_cpu_apicid[];
8128
8129 #ifdef CONFIG_X86_LOCAL_APIC
8130 @@ -118,8 +130,9 @@ static __inline int logical_smp_processo
8131 #endif
8132
8133 #ifdef CONFIG_SMP
8134 -#define cpu_physical_id(cpu) x86_cpu_to_apicid[cpu]
8135 +#define cpu_physical_id(cpu) per_cpu(x86_cpu_to_apicid, cpu)
8136 #else
8137 +extern unsigned int boot_cpu_id;
8138 #define cpu_physical_id(cpu) boot_cpu_id
8139 #endif /* !CONFIG_SMP */
8140 #endif
8141 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8142 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/swiotlb.h 2009-02-16 16:18:36.000000000 +0100
8143 @@ -0,0 +1,5 @@
8144 +#ifdef CONFIG_X86_32
8145 +# include "swiotlb_32.h"
8146 +#else
8147 +# include "../../swiotlb.h"
8148 +#endif
8149 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8150 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/system.h 2009-02-16 16:18:36.000000000 +0100
8151 @@ -0,0 +1,5 @@
8152 +#ifdef CONFIG_X86_32
8153 +# include "system_32.h"
8154 +#else
8155 +# include "system_64.h"
8156 +#endif
8157 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/system_32.h 2009-02-16 16:17:21.000000000 +0100
8158 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/system_32.h 2009-02-16 16:18:36.000000000 +0100
8159 @@ -9,6 +9,7 @@
8160 #include <asm/hypervisor.h>
8161
8162 #ifdef __KERNEL__
8163 +#define AT_VECTOR_SIZE_ARCH 2 /* entries in ARCH_DLINFO */
8164
8165 struct task_struct; /* one of the stranger aspects of C forward declarations.. */
8166 extern struct task_struct * FASTCALL(__switch_to(struct task_struct *prev, struct task_struct *next));
8167 @@ -138,7 +139,7 @@ static inline unsigned long xen_read_cr4
8168 {
8169 unsigned long val;
8170 /* This could fault if %cr4 does not exist */
8171 - asm("1: movl %%cr4, %0 \n"
8172 + asm volatile("1: movl %%cr4, %0 \n"
8173 "2: \n"
8174 ".section __ex_table,\"a\" \n"
8175 ".long 1b,2b \n"
8176 @@ -157,6 +158,11 @@ static inline void xen_wbinvd(void)
8177 asm volatile("wbinvd": : :"memory");
8178 }
8179
8180 +static inline void clflush(volatile void *__p)
8181 +{
8182 + asm volatile("clflush %0" : "+m" (*(char __force *)__p));
8183 +}
8184 +
8185 #define read_cr0() (xen_read_cr0())
8186 #define write_cr0(x) (xen_write_cr0(x))
8187 #define read_cr2() (xen_read_cr2())
8188 @@ -207,6 +213,7 @@ static inline unsigned long get_limit(un
8189
8190 #define mb() alternative("lock; addl $0,0(%%esp)", "mfence", X86_FEATURE_XMM2)
8191 #define rmb() alternative("lock; addl $0,0(%%esp)", "lfence", X86_FEATURE_XMM2)
8192 +#define wmb() alternative("lock; addl $0,0(%%esp)", "sfence", X86_FEATURE_XMM)
8193
8194 /**
8195 * read_barrier_depends - Flush all pending reads that subsequents reads
8196 @@ -262,18 +269,18 @@ static inline unsigned long get_limit(un
8197
8198 #define read_barrier_depends() do { } while(0)
8199
8200 +#ifdef CONFIG_SMP
8201 +#define smp_mb() mb()
8202 +#ifdef CONFIG_X86_PPRO_FENCE
8203 +# define smp_rmb() rmb()
8204 +#else
8205 +# define smp_rmb() barrier()
8206 +#endif
8207 #ifdef CONFIG_X86_OOSTORE
8208 -/* Actually there are no OOO store capable CPUs for now that do SSE,
8209 - but make it already an possibility. */
8210 -#define wmb() alternative("lock; addl $0,0(%%esp)", "sfence", X86_FEATURE_XMM)
8211 +# define smp_wmb() wmb()
8212 #else
8213 -#define wmb() __asm__ __volatile__ ("": : :"memory")
8214 +# define smp_wmb() barrier()
8215 #endif
8216 -
8217 -#ifdef CONFIG_SMP
8218 -#define smp_mb() mb()
8219 -#define smp_rmb() rmb()
8220 -#define smp_wmb() wmb()
8221 #define smp_read_barrier_depends() read_barrier_depends()
8222 #define set_mb(var, value) do { (void) xchg(&var, value); } while (0)
8223 #else
8224 @@ -300,5 +307,6 @@ extern unsigned long arch_align_stack(un
8225 extern void free_init_pages(char *what, unsigned long begin, unsigned long end);
8226
8227 void default_idle(void);
8228 +void __show_registers(struct pt_regs *, int all);
8229
8230 #endif
8231 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/system_64.h 2009-02-16 16:17:21.000000000 +0100
8232 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/system_64.h 2009-02-16 16:18:36.000000000 +0100
8233 @@ -11,8 +11,12 @@
8234
8235 #ifdef __KERNEL__
8236
8237 -#define __STR(x) #x
8238 -#define STR(x) __STR(x)
8239 +/* entries in ARCH_DLINFO: */
8240 +#ifdef CONFIG_IA32_EMULATION
8241 +# define AT_VECTOR_SIZE_ARCH 2
8242 +#else
8243 +# define AT_VECTOR_SIZE_ARCH 1
8244 +#endif
8245
8246 #define __SAVE(reg,offset) "movq %%" #reg ",(14-" #offset ")*8(%%rsp)\n\t"
8247 #define __RESTORE(reg,offset) "movq (14-" #offset ")*8(%%rsp),%%" #reg "\n\t"
8248 @@ -92,7 +96,7 @@ static inline void write_cr0(unsigned lo
8249
8250 #define read_cr3() ({ \
8251 unsigned long __dummy; \
8252 - asm("movq %%cr3,%0" : "=r" (__dummy)); \
8253 + asm volatile("movq %%cr3,%0" : "=r" (__dummy)); \
8254 machine_to_phys(__dummy); \
8255 })
8256
8257 @@ -105,7 +109,7 @@ static inline void write_cr3(unsigned lo
8258 static inline unsigned long read_cr4(void)
8259 {
8260 unsigned long cr4;
8261 - asm("movq %%cr4,%0" : "=r" (cr4));
8262 + asm volatile("movq %%cr4,%0" : "=r" (cr4));
8263 return cr4;
8264 }
8265
8266 @@ -131,12 +135,17 @@ static inline void write_cr8(unsigned lo
8267
8268 #endif /* __KERNEL__ */
8269
8270 +static inline void clflush(volatile void *__p)
8271 +{
8272 + asm volatile("clflush %0" : "+m" (*(char __force *)__p));
8273 +}
8274 +
8275 #define nop() __asm__ __volatile__ ("nop")
8276
8277 #ifdef CONFIG_SMP
8278 #define smp_mb() mb()
8279 -#define smp_rmb() rmb()
8280 -#define smp_wmb() wmb()
8281 +#define smp_rmb() barrier()
8282 +#define smp_wmb() barrier()
8283 #define smp_read_barrier_depends() do {} while(0)
8284 #else
8285 #define smp_mb() barrier()
8286 @@ -153,12 +162,8 @@ static inline void write_cr8(unsigned lo
8287 */
8288 #define mb() asm volatile("mfence":::"memory")
8289 #define rmb() asm volatile("lfence":::"memory")
8290 -
8291 -#ifdef CONFIG_UNORDERED_IO
8292 #define wmb() asm volatile("sfence" ::: "memory")
8293 -#else
8294 -#define wmb() asm volatile("" ::: "memory")
8295 -#endif
8296 +
8297 #define read_barrier_depends() do {} while(0)
8298 #define set_mb(var, value) do { (void) xchg(&var, value); } while (0)
8299
8300 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8301 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/tlbflush.h 2009-02-16 16:18:36.000000000 +0100
8302 @@ -0,0 +1,5 @@
8303 +#ifdef CONFIG_X86_32
8304 +# include "tlbflush_32.h"
8305 +#else
8306 +# include "tlbflush_64.h"
8307 +#endif
8308 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/tlbflush_32.h 2009-02-16 16:17:21.000000000 +0100
8309 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/tlbflush_32.h 2009-02-16 16:18:36.000000000 +0100
8310 @@ -23,7 +23,6 @@
8311 * - flush_tlb_page(vma, vmaddr) flushes one page
8312 * - flush_tlb_range(vma, start, end) flushes a range of pages
8313 * - flush_tlb_kernel_range(start, end) flushes a range of kernel pages
8314 - * - flush_tlb_pgtables(mm, start, end) flushes a range of page tables
8315 *
8316 * ..but the i386 has somewhat limited tlb flushing capabilities,
8317 * and page-granular flushes are available only on i486 and up.
8318 @@ -97,10 +96,4 @@ static inline void flush_tlb_kernel_rang
8319 flush_tlb_all();
8320 }
8321
8322 -static inline void flush_tlb_pgtables(struct mm_struct *mm,
8323 - unsigned long start, unsigned long end)
8324 -{
8325 - /* i386 does not keep any page table caches in TLB */
8326 -}
8327 -
8328 #endif /* _I386_TLBFLUSH_H */
8329 --- sle11-2009-10-16.orig/include/asm-x86/mach-xen/asm/tlbflush_64.h 2009-02-16 16:17:21.000000000 +0100
8330 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/tlbflush_64.h 2009-02-16 16:18:36.000000000 +0100
8331 @@ -28,7 +28,6 @@
8332 * - flush_tlb_page(vma, vmaddr) flushes one page
8333 * - flush_tlb_range(vma, start, end) flushes a range of pages
8334 * - flush_tlb_kernel_range(start, end) flushes a range of kernel pages
8335 - * - flush_tlb_pgtables(mm, start, end) flushes a range of page tables
8336 *
8337 * x86-64 can only flush individual pages or full VMs. For a range flush
8338 * we always do the full VM. Might be worth trying if for a small
8339 @@ -95,12 +94,4 @@ static inline void flush_tlb_kernel_rang
8340 flush_tlb_all();
8341 }
8342
8343 -static inline void flush_tlb_pgtables(struct mm_struct *mm,
8344 - unsigned long start, unsigned long end)
8345 -{
8346 - /* x86_64 does not keep any page table caches in a software TLB.
8347 - The CPUs do in their hardware TLBs, but they are handled
8348 - by the normal TLB flushing algorithms. */
8349 -}
8350 -
8351 #endif /* _X8664_TLBFLUSH_H */
8352 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8353 +++ sle11-2009-10-16/include/asm-x86/mach-xen/asm/xor.h 2009-02-16 16:18:36.000000000 +0100
8354 @@ -0,0 +1,5 @@
8355 +#ifdef CONFIG_X86_32
8356 +# include "../../xor_32.h"
8357 +#else
8358 +# include "xor_64.h"
8359 +#endif
8360 --- sle11-2009-10-16.orig/include/asm-x86/mmu.h 2009-10-28 14:55:06.000000000 +0100
8361 +++ sle11-2009-10-16/include/asm-x86/mmu.h 2009-02-16 16:18:36.000000000 +0100
8362 @@ -16,6 +16,9 @@ typedef struct {
8363 rwlock_t ldtlock;
8364 #endif
8365 int size;
8366 +#ifdef CONFIG_XEN
8367 + unsigned has_foreign_mappings:1;
8368 +#endif
8369 struct mutex lock;
8370 void *vdso;
8371 } mm_context_t;
8372 --- sle11-2009-10-16.orig/include/linux/kexec.h 2009-08-26 11:52:33.000000000 +0200
8373 +++ sle11-2009-10-16/include/linux/kexec.h 2009-02-17 12:43:57.000000000 +0100
8374 @@ -202,8 +202,15 @@ extern int dump_after_notifier;
8375 #define VMCOREINFO_BYTES (4096)
8376 #define VMCOREINFO_NOTE_NAME "VMCOREINFO"
8377 #define VMCOREINFO_NOTE_NAME_BYTES ALIGN(sizeof(VMCOREINFO_NOTE_NAME), 4)
8378 +#if !defined(CONFIG_XEN) || !defined(CONFIG_X86)
8379 #define VMCOREINFO_NOTE_SIZE (KEXEC_NOTE_HEAD_BYTES*2 + VMCOREINFO_BYTES \
8380 + VMCOREINFO_NOTE_NAME_BYTES)
8381 +#else
8382 +#define VMCOREINFO_NOTE_SIZE ALIGN(KEXEC_NOTE_HEAD_BYTES*2 \
8383 + + VMCOREINFO_BYTES \
8384 + + VMCOREINFO_NOTE_NAME_BYTES, \
8385 + PAGE_SIZE)
8386 +#endif
8387
8388 /* Location of a reserved region to hold the crash kernel.
8389 */
8390 --- sle11-2009-10-16.orig/include/linux/oprofile.h 2009-06-29 15:28:01.000000000 +0200
8391 +++ sle11-2009-10-16/include/linux/oprofile.h 2009-06-29 15:28:57.000000000 +0200
8392 @@ -119,6 +119,8 @@ void oprofile_add_pc(unsigned long pc, i
8393 /* add a backtrace entry, to be called from the ->backtrace callback */
8394 void oprofile_add_trace(unsigned long eip);
8395
8396 +void oprofile_add_mode(int cpu_mode);
8397 +
8398 /* add a domain switch entry */
8399 int oprofile_add_domain_switch(int32_t domain_id);
8400
8401 --- sle11-2009-10-16.orig/include/linux/sysctl.h 2009-10-28 14:55:06.000000000 +0100
8402 +++ sle11-2009-10-16/include/linux/sysctl.h 2009-02-16 16:18:36.000000000 +0100
8403 @@ -69,6 +69,7 @@ enum
8404 CTL_BUS=8, /* Busses */
8405 CTL_ABI=9, /* Binary emulation */
8406 CTL_CPU=10, /* CPU stuff (speed scaling, etc) */
8407 + CTL_XEN=123, /* Xen info and control */
8408 CTL_ARLAN=254, /* arlan wireless driver */
8409 CTL_S390DBF=5677, /* s390 debug */
8410 CTL_SUNRPC=7249, /* sunrpc debug */
8411 --- sle11-2009-10-16.orig/include/xen/pcifront.h 2009-10-28 14:55:06.000000000 +0100
8412 +++ sle11-2009-10-16/include/xen/pcifront.h 2009-02-16 16:18:36.000000000 +0100
8413 @@ -12,13 +12,11 @@
8414
8415 #ifndef __ia64__
8416
8417 +#include <asm/pci.h>
8418 +
8419 struct pcifront_device;
8420 struct pci_bus;
8421 -
8422 -struct pcifront_sd {
8423 - int domain;
8424 - struct pcifront_device *pdev;
8425 -};
8426 +#define pcifront_sd pci_sysdata
8427
8428 static inline struct pcifront_device *
8429 pcifront_get_pdev(struct pcifront_sd *sd)
8430 @@ -34,18 +32,6 @@ static inline void pcifront_init_sd(stru
8431 sd->pdev = pdev;
8432 }
8433
8434 -#if defined(CONFIG_PCI_DOMAINS)
8435 -static inline int pci_domain_nr(struct pci_bus *bus)
8436 -{
8437 - struct pcifront_sd *sd = bus->sysdata;
8438 - return sd->domain;
8439 -}
8440 -static inline int pci_proc_domain(struct pci_bus *bus)
8441 -{
8442 - return pci_domain_nr(bus);
8443 -}
8444 -#endif /* CONFIG_PCI_DOMAINS */
8445 -
8446 static inline void pcifront_setup_root_resources(struct pci_bus *bus,
8447 struct pcifront_sd *sd)
8448 {
8449 --- /dev/null 1970-01-01 00:00:00.000000000 +0000
8450 +++ sle11-2009-10-16/include/xen/sysctl.h 2009-02-16 16:18:36.000000000 +0100
8451 @@ -0,0 +1,11 @@
8452 +#ifndef _XEN_SYSCTL_H
8453 +#define _XEN_SYSCTL_H
8454 +
8455 +/* CTL_XEN names: */
8456 +enum
8457 +{
8458 + CTL_XEN_INDEPENDENT_WALLCLOCK=1,
8459 + CTL_XEN_PERMITTED_CLOCK_JITTER=2,
8460 +};
8461 +
8462 +#endif /* _XEN_SYSCTL_H */
8463 --- sle11-2009-10-16.orig/include/xen/xenbus.h 2009-02-16 16:17:21.000000000 +0100
8464 +++ sle11-2009-10-16/include/xen/xenbus.h 2009-02-16 16:18:36.000000000 +0100
8465 @@ -107,7 +107,7 @@ struct xenbus_driver {
8466 int (*suspend)(struct xenbus_device *dev);
8467 int (*suspend_cancel)(struct xenbus_device *dev);
8468 int (*resume)(struct xenbus_device *dev);
8469 - int (*uevent)(struct xenbus_device *, char **, int, char *, int);
8470 + int (*uevent)(struct xenbus_device *, struct kobj_uevent_env *);
8471 struct device_driver driver;
8472 int (*read_otherend_details)(struct xenbus_device *dev);
8473 int (*is_ready)(struct xenbus_device *dev);
8474 --- sle11-2009-10-16.orig/kernel/kexec.c 2009-02-17 11:34:22.000000000 +0100
8475 +++ sle11-2009-10-16/kernel/kexec.c 2009-02-17 12:38:20.000000000 +0100
8476 @@ -52,7 +52,11 @@ int dump_after_notifier;
8477
8478 /* vmcoreinfo stuff */
8479 unsigned char vmcoreinfo_data[VMCOREINFO_BYTES];
8480 -u32 vmcoreinfo_note[VMCOREINFO_NOTE_SIZE/4];
8481 +u32
8482 +#if defined(CONFIG_XEN) && defined(CONFIG_X86)
8483 +__attribute__((__section__(".bss.page_aligned"), __aligned__(PAGE_SIZE)))
8484 +#endif
8485 +vmcoreinfo_note[VMCOREINFO_NOTE_SIZE/4];
8486 size_t vmcoreinfo_size;
8487 size_t vmcoreinfo_max_size = sizeof(vmcoreinfo_data);
8488
8489 @@ -1240,6 +1244,7 @@ static int __init crash_notes_memory_ini
8490 module_init(crash_notes_memory_init)
8491
8492
8493 +#ifndef CONFIG_XEN
8494 /*
8495 * parsing the "crashkernel" commandline
8496 *
8497 @@ -1402,7 +1407,7 @@ int __init parse_crashkernel(char *cm
8498
8499 return 0;
8500 }
8501 -
8502 +#endif
8503
8504
8505 void crash_save_vmcoreinfo(void)
8506 @@ -1459,7 +1464,18 @@ static int __init crash_save_vmcoreinfo_
8507
8508 VMCOREINFO_SYMBOL(init_uts_ns);
8509 VMCOREINFO_SYMBOL(node_online_map);
8510 +#ifndef CONFIG_X86_XEN
8511 VMCOREINFO_SYMBOL(swapper_pg_dir);
8512 +#else
8513 +/*
8514 + * Since for x86-32 Xen swapper_pg_dir is a pointer rather than an array,
8515 + * make the value stored consistent with native (i.e. the base address of
8516 + * the page directory).
8517 + */
8518 +# define swapper_pg_dir *swapper_pg_dir
8519 + VMCOREINFO_SYMBOL(swapper_pg_dir);
8520 +# undef swapper_pg_dir
8521 +#endif
8522 VMCOREINFO_SYMBOL(_stext);
8523
8524 #ifndef CONFIG_NEED_MULTIPLE_NODES
8525 --- sle11-2009-10-16.orig/kernel/sysctl_check.c 2009-10-28 14:55:06.000000000 +0100
8526 +++ sle11-2009-10-16/kernel/sysctl_check.c 2009-02-16 16:18:36.000000000 +0100
8527 @@ -4,6 +4,7 @@
8528 #include <linux/sunrpc/debug.h>
8529 #include <linux/string.h>
8530 #include <net/ip_vs.h>
8531 +#include <xen/sysctl.h>
8532
8533 struct trans_ctl_table {
8534 int ctl_name;
8535 @@ -897,6 +898,14 @@ static const struct trans_ctl_table tran
8536 {}
8537 };
8538
8539 +#ifdef CONFIG_XEN
8540 +static struct trans_ctl_table trans_xen_table[] = {
8541 + { CTL_XEN_INDEPENDENT_WALLCLOCK, "independent_wallclock" },
8542 + { CTL_XEN_PERMITTED_CLOCK_JITTER, "permitted_clock_jitter" },
8543 + {}
8544 +};
8545 +#endif
8546 +
8547 static const struct trans_ctl_table trans_arlan_conf_table0[] = {
8548 { 1, "spreadingCode" },
8549 { 2, "channelNumber" },
8550 @@ -1232,6 +1241,9 @@ static const struct trans_ctl_table tran
8551 { CTL_BUS, "bus", trans_bus_table },
8552 { CTL_ABI, "abi" },
8553 /* CTL_CPU not used */
8554 +#ifdef CONFIG_XEN
8555 + { CTL_XEN, "xen", trans_xen_table },
8556 +#endif
8557 { CTL_ARLAN, "arlan", trans_arlan_table },
8558 { CTL_S390DBF, "s390dbf", trans_s390dbf_table },
8559 { CTL_SUNRPC, "sunrpc", trans_sunrpc_table },
8560 --- sle11-2009-10-16.orig/lib/swiotlb-xen.c 2009-02-05 11:16:51.000000000 +0100
8561 +++ sle11-2009-10-16/lib/swiotlb-xen.c 2009-02-16 16:18:36.000000000 +0100
8562 @@ -27,7 +27,7 @@
8563 #include <asm/uaccess.h>
8564 #include <xen/gnttab.h>
8565 #include <xen/interface/memory.h>
8566 -#include <asm-i386/mach-xen/asm/gnttab_dma.h>
8567 +#include <asm/gnttab_dma.h>
8568
8569 int swiotlb;
8570 EXPORT_SYMBOL(swiotlb);
8571 @@ -574,9 +574,10 @@ swiotlb_sync_single_for_device(struct de
8572 * same here.
8573 */
8574 int
8575 -swiotlb_map_sg(struct device *hwdev, struct scatterlist *sg, int nelems,
8576 +swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
8577 int dir)
8578 {
8579 + struct scatterlist *sg;
8580 struct phys_addr buffer;
8581 dma_addr_t dev_addr;
8582 char *map;
8583 @@ -584,22 +585,22 @@ swiotlb_map_sg(struct device *hwdev, str
8584
8585 BUG_ON(dir == DMA_NONE);
8586
8587 - for (i = 0; i < nelems; i++, sg++) {
8588 - dev_addr = gnttab_dma_map_page(sg->page) + sg->offset;
8589 + for_each_sg(sgl, sg, nelems, i) {
8590 + dev_addr = gnttab_dma_map_page(sg_page(sg)) + sg->offset;
8591
8592 - if (range_straddles_page_boundary(page_to_pseudophys(sg->page)
8593 + if (range_straddles_page_boundary(page_to_pseudophys(sg_page(sg))
8594 + sg->offset, sg->length)
8595 || address_needs_mapping(hwdev, dev_addr)) {
8596 gnttab_dma_unmap_page(dev_addr);
8597 - buffer.page = sg->page;
8598 + buffer.page = sg_page(sg);
8599 buffer.offset = sg->offset;
8600 map = map_single(hwdev, buffer, sg->length, dir);
8601 if (!map) {
8602 /* Don't panic here, we expect map_sg users
8603 to do proper error handling. */
8604 swiotlb_full(hwdev, sg->length, dir, 0);
8605 - swiotlb_unmap_sg(hwdev, sg - i, i, dir);
8606 - sg[0].dma_length = 0;
8607 + swiotlb_unmap_sg(hwdev, sgl, i, dir);
8608 + sgl[0].dma_length = 0;
8609 return 0;
8610 }
8611 sg->dma_address = virt_to_bus(map);
8612 @@ -615,19 +616,21 @@ swiotlb_map_sg(struct device *hwdev, str
8613 * concerning calls here are the same as for swiotlb_unmap_single() above.
8614 */
8615 void
8616 -swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sg, int nelems,
8617 +swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
8618 int dir)
8619 {
8620 + struct scatterlist *sg;
8621 int i;
8622
8623 BUG_ON(dir == DMA_NONE);
8624
8625 - for (i = 0; i < nelems; i++, sg++)
8626 + for_each_sg(sgl, sg, nelems, i) {
8627 if (in_swiotlb_aperture(sg->dma_address))
8628 unmap_single(hwdev, bus_to_virt(sg->dma_address),
8629 sg->dma_length, dir);
8630 else
8631 gnttab_dma_unmap_page(sg->dma_address);
8632 + }
8633 }
8634
8635 /*
8636 @@ -638,31 +641,35 @@ swiotlb_unmap_sg(struct device *hwdev, s
8637 * and usage.
8638 */
8639 void
8640 -swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sg,
8641 +swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sgl,
8642 int nelems, int dir)
8643 {
8644 + struct scatterlist *sg;
8645 int i;
8646
8647 BUG_ON(dir == DMA_NONE);
8648
8649 - for (i = 0; i < nelems; i++, sg++)
8650 + for_each_sg(sgl, sg, nelems, i) {
8651 if (in_swiotlb_aperture(sg->dma_address))
8652 sync_single(hwdev, bus_to_virt(sg->dma_address),
8653 sg->dma_length, dir);
8654 + }
8655 }
8656
8657 void
8658 -swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg,
8659 +swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sgl,
8660 int nelems, int dir)
8661 {
8662 + struct scatterlist *sg;
8663 int i;
8664
8665 BUG_ON(dir == DMA_NONE);
8666
8667 - for (i = 0; i < nelems; i++, sg++)
8668 + for_each_sg(sgl, sg, nelems, i) {
8669 if (in_swiotlb_aperture(sg->dma_address))
8670 sync_single(hwdev, bus_to_virt(sg->dma_address),
8671 sg->dma_length, dir);
8672 + }
8673 }
8674
8675 #ifdef CONFIG_HIGHMEM
IPFire 2.x development tree