]> git.ipfire.org Git - people/pmueller/ipfire-2.x.git/blame - src/patches/suse-2.6.27.31/patches.xen/xen3-patch-2.6.23
projects / people / pmueller / ipfire-2.x.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Add a patch to fix Intel E100 wake-on-lan problems.
[people/pmueller/ipfire-2.x.git] / src / patches / suse-2.6.27.31 / patches.xen / xen3-patch-2.6.23
CommitLineData
2cb7cef9
BS		 1From: www.kernel.org
2Subject: Update to 2.6.23
3Patch-mainline: 2.6.23
4
5Automatically created from "patches.kernel.org/patch-2.6.23" by xen-port-patches.py
6
7Acked-by: jbeulich@novell.com
8
9--- sle11-2009-05-14.orig/arch/x86/Makefile 2008-12-01 11:11:08.000000000 +0100
10+++ sle11-2009-05-14/arch/x86/Makefile 2009-02-16 16:17:21.000000000 +0100
11@@ -148,7 +148,7 @@ libs-y += arch/x86/lib/
12 core-y += $(fcore-y)
13
14 # Xen paravirtualization support
15-core-$(CONFIG_XEN) += arch/x86/xen/
16+core-$(CONFIG_PARAVIRT_XEN) += arch/x86/xen/
17
18 # lguest paravirtualization support
19 core-$(CONFIG_LGUEST_GUEST) += arch/x86/lguest/
20--- sle11-2009-05-14.orig/arch/x86/kernel/acpi/sleep_32-xen.c 2009-05-14 10:56:29.000000000 +0200
21+++ sle11-2009-05-14/arch/x86/kernel/acpi/sleep_32-xen.c 2009-02-16 16:17:21.000000000 +0100
22@@ -15,7 +15,7 @@
23 #ifndef CONFIG_ACPI_PV_SLEEP
24 /* address in low memory of the wakeup routine. */
25 unsigned long acpi_wakeup_address = 0;
26-unsigned long acpi_video_flags;
27+unsigned long acpi_realmode_flags;
28 extern char wakeup_start, wakeup_end;
29
30 extern unsigned long FASTCALL(acpi_copy_wakeup_routine(unsigned long));
31@@ -74,9 +74,11 @@ static int __init acpi_sleep_setup(char
32 {
33 while ((str != NULL) && (*str != '\0')) {
34 if (strncmp(str, "s3_bios", 7) == 0)
35- acpi_video_flags = 1;
36+ acpi_realmode_flags |= 1;
37 if (strncmp(str, "s3_mode", 7) == 0)
38- acpi_video_flags |= 2;
39+ acpi_realmode_flags |= 2;
40+ if (strncmp(str, "s3_beep", 7) == 0)
41+ acpi_realmode_flags |= 4;
42 str = strchr(str, ',');
43 if (str != NULL)
44 str += strspn(str, ", \t");
45@@ -86,9 +88,11 @@ static int __init acpi_sleep_setup(char
46
47 __setup("acpi_sleep=", acpi_sleep_setup);
48
49+/* Ouch, we want to delete this. We already have better version in userspace, in
50+ s2ram from suspend.sf.net project */
51 static __init int reset_videomode_after_s3(struct dmi_system_id *d)
52 {
53- acpi_video_flags |= 2;
54+ acpi_realmode_flags |= 2;
55 return 0;
56 }
57
58--- sle11-2009-05-14.orig/arch/x86/kernel/asm-offsets_32.c 2008-12-15 11:27:22.000000000 +0100
59+++ sle11-2009-05-14/arch/x86/kernel/asm-offsets_32.c 2009-02-16 16:17:21.000000000 +0100
60@@ -19,7 +19,9 @@
61 #include <asm/bootparam.h>
62 #include <asm/elf.h>
63
64+#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN)
65 #include <xen/interface/xen.h>
66+#endif
67
68 #include <linux/lguest.h>
69 #include "../../../drivers/lguest/lg.h"
70@@ -121,7 +123,7 @@ void foo(void)
71 OFFSET(PV_CPU_read_cr0, pv_cpu_ops, read_cr0);
72 #endif
73
74-#ifdef CONFIG_XEN
75+#ifdef CONFIG_PARAVIRT_XEN
76 BLANK();
77 OFFSET(XEN_vcpu_info_mask, vcpu_info, evtchn_upcall_mask);
78 OFFSET(XEN_vcpu_info_pending, vcpu_info, evtchn_upcall_pending);
79--- sle11-2009-05-14.orig/arch/x86/kernel/cpu/common-xen.c 2008-12-15 11:27:22.000000000 +0100
80+++ sle11-2009-05-14/arch/x86/kernel/cpu/common-xen.c 2009-02-16 16:17:21.000000000 +0100
81@@ -360,6 +360,8 @@ static void __cpuinit generic_identify(s
82 if ( xlvl >= 0x80000004 )
83 get_model_name(c); /* Default name */
84 }
85+
86+ init_scattered_cpuid_features(c);
87 }
88
89 early_intel_workaround(c);
90@@ -611,7 +613,6 @@ extern int nsc_init_cpu(void);
91 extern int amd_init_cpu(void);
92 extern int centaur_init_cpu(void);
93 extern int transmeta_init_cpu(void);
94-extern int rise_init_cpu(void);
95 extern int nexgen_init_cpu(void);
96 extern int umc_init_cpu(void);
97
98@@ -623,7 +624,6 @@ void __init early_cpu_init(void)
99 amd_init_cpu();
100 centaur_init_cpu();
101 transmeta_init_cpu();
102- rise_init_cpu();
103 nexgen_init_cpu();
104 umc_init_cpu();
105 early_cpu_detect();
106--- sle11-2009-05-14.orig/arch/x86/kernel/cpu/mtrr/main-xen.c 2008-12-15 11:27:22.000000000 +0100
107+++ sle11-2009-05-14/arch/x86/kernel/cpu/mtrr/main-xen.c 2009-02-16 16:17:21.000000000 +0100
108@@ -167,7 +167,7 @@ mtrr_del(int reg, unsigned long base, un
109 EXPORT_SYMBOL(mtrr_add);
110 EXPORT_SYMBOL(mtrr_del);
111
112-__init void mtrr_bp_init(void)
113+void __init mtrr_bp_init(void)
114 {
115 }
116
117--- sle11-2009-05-14.orig/arch/x86/kernel/e820_32-xen.c 2008-12-15 11:27:22.000000000 +0100
118+++ sle11-2009-05-14/arch/x86/kernel/e820_32-xen.c 2009-02-16 16:17:21.000000000 +0100
119@@ -10,6 +10,7 @@
120 #include <linux/efi.h>
121 #include <linux/pfn.h>
122 #include <linux/uaccess.h>
123+#include <linux/suspend.h>
124
125 #include <asm/pgtable.h>
126 #include <asm/page.h>
127@@ -343,6 +344,37 @@ static int __init request_standard_resou
128
129 subsys_initcall(request_standard_resources);
130
131+#if defined(CONFIG_PM) && defined(CONFIG_HIBERNATION)
132+/**
133+ * e820_mark_nosave_regions - Find the ranges of physical addresses that do not
134+ * correspond to e820 RAM areas and mark the corresponding pages as nosave for
135+ * hibernation.
136+ *
137+ * This function requires the e820 map to be sorted and without any
138+ * overlapping entries and assumes the first e820 area to be RAM.
139+ */
140+void __init e820_mark_nosave_regions(void)
141+{
142+ int i;
143+ unsigned long pfn;
144+
145+ pfn = PFN_DOWN(e820.map[0].addr + e820.map[0].size);
146+ for (i = 1; i < e820.nr_map; i++) {
147+ struct e820entry *ei = &e820.map[i];
148+
149+ if (pfn < PFN_UP(ei->addr))
150+ register_nosave_region(pfn, PFN_UP(ei->addr));
151+
152+ pfn = PFN_DOWN(ei->addr + ei->size);
153+ if (ei->type != E820_RAM)
154+ register_nosave_region(PFN_UP(ei->addr), pfn);
155+
156+ if (pfn >= max_low_pfn)
157+ break;
158+ }
159+}
160+#endif
161+
162 void __init add_memory_region(unsigned long long start,
163 unsigned long long size, int type)
164 {
165@@ -804,7 +836,7 @@ void __init print_memory_map(char *who)
166 case E820_NVS:
167 printk("(ACPI NVS)\n");
168 break;
169- default: printk("type %lu\n", e820.map[i].type);
170+ default: printk("type %u\n", e820.map[i].type);
171 break;
172 }
173 }
174--- sle11-2009-05-14.orig/arch/x86/kernel/entry_32.S 2008-11-25 12:35:53.000000000 +0100
175+++ sle11-2009-05-14/arch/x86/kernel/entry_32.S 2009-02-16 16:17:21.000000000 +0100
176@@ -1112,7 +1112,7 @@ ENTRY(kernel_thread_helper)
177 CFI_ENDPROC
178 ENDPROC(kernel_thread_helper)
179
180-#ifdef CONFIG_XEN
181+#ifdef CONFIG_PARAVIRT_XEN
182 /* Xen doesn't set %esp to be precisely what the normal sysenter
183 entrypoint expects, so fix it up before using the normal path. */
184 ENTRY(xen_sysenter_target)
185@@ -1205,7 +1205,7 @@ ENTRY(xen_failsafe_callback)
186 .previous
187 ENDPROC(xen_failsafe_callback)
188
189-#endif /* CONFIG_XEN */
190+#endif /* CONFIG_PARAVIRT_XEN */
191
192 #ifdef CONFIG_FTRACE
193 #ifdef CONFIG_DYNAMIC_FTRACE
194--- sle11-2009-05-14.orig/arch/x86/kernel/entry_32-xen.S 2008-12-15 11:27:22.000000000 +0100
195+++ sle11-2009-05-14/arch/x86/kernel/entry_32-xen.S 2009-02-16 16:17:21.000000000 +0100
196@@ -452,9 +452,6 @@ restore_nocheck_notrace:
197 1: INTERRUPT_RETURN
198 .section .fixup,"ax"
199 iret_exc:
200-#ifndef CONFIG_XEN
201- ENABLE_INTERRUPTS(CLBR_NONE)
202-#endif
203 pushl $0 # no error code
204 pushl $do_iret_error
205 jmp error_code
206--- sle11-2009-05-14.orig/arch/x86/kernel/head_32-xen.S 2008-12-15 11:27:22.000000000 +0100
207+++ sle11-2009-05-14/arch/x86/kernel/head_32-xen.S 2009-02-16 16:17:21.000000000 +0100
208@@ -86,7 +86,10 @@ ENTRY(_stext)
209 /*
210 * BSS section
211 */
212-.section ".bss.page_aligned","w"
213+.section ".bss.page_aligned","wa"
214+ .align PAGE_SIZE_asm
215+ENTRY(swapper_pg_pmd)
216+ .fill 1024,4,0
217 ENTRY(empty_zero_page)
218 .fill 4096,1,0
219
220@@ -136,25 +139,25 @@ ENTRY(empty_zero_page)
221 #endif /* CONFIG_XEN_COMPAT <= 0x030002 */
222
223
224- ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS, .asciz, "linux")
225- ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION, .asciz, "2.6")
226- ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION, .asciz, "xen-3.0")
227- ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE, .long, __PAGE_OFFSET)
228+ ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS, .asciz "linux")
229+ ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION, .asciz "2.6")
230+ ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION, .asciz "xen-3.0")
231+ ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE, .long __PAGE_OFFSET)
232 #if CONFIG_XEN_COMPAT <= 0x030002
233- ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .long, __PAGE_OFFSET)
234+ ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .long __PAGE_OFFSET)
235 #else
236- ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .long, 0)
237+ ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .long 0)
238 #endif
239- ELFNOTE(Xen, XEN_ELFNOTE_ENTRY, .long, startup_32)
240- ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .long, hypercall_page)
241- ELFNOTE(Xen, XEN_ELFNOTE_HV_START_LOW, .long, HYPERVISOR_VIRT_START)
242- ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz, "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel")
243+ ELFNOTE(Xen, XEN_ELFNOTE_ENTRY, .long startup_32)
244+ ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .long hypercall_page)
245+ ELFNOTE(Xen, XEN_ELFNOTE_HV_START_LOW, .long HYPERVISOR_VIRT_START)
246+ ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel")
247 #ifdef CONFIG_X86_PAE
248- ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE, .asciz, "yes")
249- ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .quad, _PAGE_PRESENT,_PAGE_PRESENT)
250+ ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE, .asciz "yes")
251+ ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .quad _PAGE_PRESENT, _PAGE_PRESENT)
252 #else
253- ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE, .asciz, "no")
254- ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .long, _PAGE_PRESENT,_PAGE_PRESENT)
255+ ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE, .asciz "no")
256+ ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .long _PAGE_PRESENT, _PAGE_PRESENT)
257 #endif
258- ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz, "generic")
259- ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long, 1)
260+ ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz "generic")
261+ ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long 1)
262--- sle11-2009-05-14.orig/arch/x86/kernel/init_task-xen.c 2009-05-14 10:56:29.000000000 +0200
263+++ sle11-2009-05-14/arch/x86/kernel/init_task-xen.c 2009-02-16 16:17:21.000000000 +0100
264@@ -46,6 +46,6 @@ EXPORT_SYMBOL(init_task);
265 * per-CPU TSS segments. Threads are completely 'soft' on Linux,
266 * no more per-task TSS's.
267 */
268-DEFINE_PER_CPU(struct tss_struct, init_tss) ____cacheline_internodealigned_in_smp = INIT_TSS;
269+DEFINE_PER_CPU_SHARED_ALIGNED(struct tss_struct, init_tss) = INIT_TSS;
270 #endif
271
272--- sle11-2009-05-14.orig/arch/x86/kernel/io_apic_32-xen.c 2008-12-15 11:27:22.000000000 +0100
273+++ sle11-2009-05-14/arch/x86/kernel/io_apic_32-xen.c 2009-02-16 16:17:21.000000000 +0100
274@@ -402,14 +402,6 @@ static void set_ioapic_affinity_irq(unsi
275 # include <linux/slab.h> /* kmalloc() */
276 # include <linux/timer.h> /* time_after() */
277
278-#ifdef CONFIG_BALANCED_IRQ_DEBUG
279-# define TDprintk(x...) do { printk("<%ld:%s:%d>: ", jiffies, __FILE__, __LINE__); printk(x); } while (0)
280-# define Dprintk(x...) do { TDprintk(x); } while (0)
281-# else
282-# define TDprintk(x...)
283-# define Dprintk(x...)
284-# endif
285-
286 #define IRQBALANCE_CHECK_ARCH -999
287 #define MAX_BALANCED_IRQ_INTERVAL (5*HZ)
288 #define MIN_BALANCED_IRQ_INTERVAL (HZ/2)
289@@ -492,7 +484,7 @@ static inline void balance_irq(int cpu,
290 static inline void rotate_irqs_among_cpus(unsigned long useful_load_threshold)
291 {
292 int i, j;
293- Dprintk("Rotating IRQs among CPUs.\n");
294+
295 for_each_online_cpu(i) {
296 for (j = 0; j < NR_IRQS; j++) {
297 if (!irq_desc[j].action)
298@@ -609,19 +601,11 @@ tryanothercpu:
299 max_loaded = tmp_loaded; /* processor */
300 imbalance = (max_cpu_irq - min_cpu_irq) / 2;
301
302- Dprintk("max_loaded cpu = %d\n", max_loaded);
303- Dprintk("min_loaded cpu = %d\n", min_loaded);
304- Dprintk("max_cpu_irq load = %ld\n", max_cpu_irq);
305- Dprintk("min_cpu_irq load = %ld\n", min_cpu_irq);
306- Dprintk("load imbalance = %lu\n", imbalance);
307-
308 /* if imbalance is less than approx 10% of max load, then
309 * observe diminishing returns action. - quit
310 */
311- if (imbalance < (max_cpu_irq >> 3)) {
312- Dprintk("Imbalance too trivial\n");
313+ if (imbalance < (max_cpu_irq >> 3))
314 goto not_worth_the_effort;
315- }
316
317 tryanotherirq:
318 /* if we select an IRQ to move that can't go where we want, then
319@@ -678,9 +662,6 @@ tryanotherirq:
320 cpus_and(tmp, target_cpu_mask, allowed_mask);
321
322 if (!cpus_empty(tmp)) {
323-
324- Dprintk("irq = %d moved to cpu = %d\n",
325- selected_irq, min_loaded);
326 /* mark for change destination */
327 set_pending_irq(selected_irq, cpumask_of_cpu(min_loaded));
328
329@@ -700,7 +681,6 @@ not_worth_the_effort:
330 */
331 balanced_irq_interval = min((long)MAX_BALANCED_IRQ_INTERVAL,
332 balanced_irq_interval + BALANCED_IRQ_MORE_DELTA);
333- Dprintk("IRQ worth rotating not found\n");
334 return;
335 }
336
337@@ -716,6 +696,7 @@ static int balanced_irq(void *unused)
338 set_pending_irq(i, cpumask_of_cpu(0));
339 }
340
341+ set_freezable();
342 for ( ; ; ) {
343 time_remaining = schedule_timeout_interruptible(time_remaining);
344 try_to_freeze();
345@@ -825,14 +806,6 @@ static int pirq_entries [MAX_PIRQS];
346 static int pirqs_enabled;
347 int skip_ioapic_setup;
348
349-static int __init ioapic_setup(char *str)
350-{
351- skip_ioapic_setup = 1;
352- return 1;
353-}
354-
355-__setup("noapic", ioapic_setup);
356-
357 static int __init ioapic_pirq_setup(char *str)
358 {
359 int i, max;
360@@ -1323,12 +1296,15 @@ static struct irq_chip ioapic_chip;
361 static void ioapic_register_intr(int irq, int vector, unsigned long trigger)
362 {
363 if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) ||
364- trigger == IOAPIC_LEVEL)
365+ trigger == IOAPIC_LEVEL) {
366+ irq_desc[irq].status |= IRQ_LEVEL;
367 set_irq_chip_and_handler_name(irq, &ioapic_chip,
368 handle_fasteoi_irq, "fasteoi");
369- else
370+ } else {
371+ irq_desc[irq].status &= ~IRQ_LEVEL;
372 set_irq_chip_and_handler_name(irq, &ioapic_chip,
373 handle_edge_irq, "edge");
374+ }
375 set_intr_gate(vector, interrupt[irq]);
376 }
377 #else
378@@ -1957,7 +1933,7 @@ __setup("no_timer_check", notimercheck);
379 * - if this function detects that timer IRQs are defunct, then we fall
380 * back to ISA timer IRQs
381 */
382-int __init timer_irq_works(void)
383+static int __init timer_irq_works(void)
384 {
385 unsigned long t1 = jiffies;
386
387--- sle11-2009-05-14.orig/arch/x86/kernel/irq_32-xen.c 2008-12-15 11:27:22.000000000 +0100
388+++ sle11-2009-05-14/arch/x86/kernel/irq_32-xen.c 2009-02-16 16:17:21.000000000 +0100
389@@ -21,7 +21,7 @@
390 #include <asm/apic.h>
391 #include <asm/uaccess.h>
392
393-DEFINE_PER_CPU(irq_cpustat_t, irq_stat) ____cacheline_internodealigned_in_smp;
394+DEFINE_PER_CPU_SHARED_ALIGNED(irq_cpustat_t, irq_stat);
395 EXPORT_PER_CPU_SYMBOL(irq_stat);
396
397 DEFINE_PER_CPU(struct pt_regs *, irq_regs);
398@@ -149,15 +149,11 @@ fastcall unsigned int do_IRQ(struct pt_r
399
400 #ifdef CONFIG_4KSTACKS
401
402-/*
403- * These should really be __section__(".bss.page_aligned") as well, but
404- * gcc's 3.0 and earlier don't handle that correctly.
405- */
406 static char softirq_stack[NR_CPUS * THREAD_SIZE]
407- __attribute__((__aligned__(THREAD_SIZE)));
408+ __attribute__((__section__(".bss.page_aligned")));
409
410 static char hardirq_stack[NR_CPUS * THREAD_SIZE]
411- __attribute__((__aligned__(THREAD_SIZE)));
412+ __attribute__((__section__(".bss.page_aligned")));
413
414 /*
415 * allocate per-cpu stacks for hardirq and for softirq processing
416--- sle11-2009-05-14.orig/arch/x86/kernel/microcode-xen.c 2008-12-15 11:27:22.000000000 +0100
417+++ sle11-2009-05-14/arch/x86/kernel/microcode-xen.c 2009-02-16 16:17:21.000000000 +0100
418@@ -33,6 +33,7 @@
419 #include <linux/miscdevice.h>
420 #include <linux/spinlock.h>
421 #include <linux/mm.h>
422+#include <linux/fs.h>
423 #include <linux/mutex.h>
424 #include <linux/cpu.h>
425 #include <linux/firmware.h>
426--- sle11-2009-05-14.orig/arch/x86/kernel/pci-dma-xen.c 2008-12-15 11:27:22.000000000 +0100
427+++ sle11-2009-05-14/arch/x86/kernel/pci-dma-xen.c 2009-02-16 16:17:21.000000000 +0100
428@@ -24,7 +24,7 @@
429 #include <asm/bug.h>
430
431 #ifdef __x86_64__
432-#include <asm/proto.h>
433+#include <asm/iommu.h>
434
435 int iommu_merge __read_mostly = 0;
436 EXPORT_SYMBOL(iommu_merge);
437--- sle11-2009-05-14.orig/arch/x86/kernel/process_32-xen.c 2008-12-15 11:27:22.000000000 +0100
438+++ sle11-2009-05-14/arch/x86/kernel/process_32-xen.c 2009-02-16 16:17:21.000000000 +0100
439@@ -241,6 +241,7 @@ early_param("idle", idle_setup);
440 void show_regs(struct pt_regs * regs)
441 {
442 unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L;
443+ unsigned long d0, d1, d2, d3, d6, d7;
444
445 printk("\n");
446 printk("Pid: %d, comm: %20s\n", current->pid, current->comm);
447@@ -265,6 +266,17 @@ void show_regs(struct pt_regs * regs)
448 cr3 = read_cr3();
449 cr4 = read_cr4_safe();
450 printk("CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n", cr0, cr2, cr3, cr4);
451+
452+ get_debugreg(d0, 0);
453+ get_debugreg(d1, 1);
454+ get_debugreg(d2, 2);
455+ get_debugreg(d3, 3);
456+ printk("DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n",
457+ d0, d1, d2, d3);
458+ get_debugreg(d6, 6);
459+ get_debugreg(d7, 7);
460+ printk("DR6: %08lx DR7: %08lx\n", d6, d7);
461+
462 show_trace(NULL, regs, &regs->esp);
463 }
464
465@@ -473,7 +485,30 @@ int dump_task_regs(struct task_struct *t
466 return 1;
467 }
468
469-static noinline void __switch_to_xtra(struct task_struct *next_p)
470+#ifdef CONFIG_SECCOMP
471+void hard_disable_TSC(void)
472+{
473+ write_cr4(read_cr4() | X86_CR4_TSD);
474+}
475+void disable_TSC(void)
476+{
477+ preempt_disable();
478+ if (!test_and_set_thread_flag(TIF_NOTSC))
479+ /*
480+ * Must flip the CPU state synchronously with
481+ * TIF_NOTSC in the current running context.
482+ */
483+ hard_disable_TSC();
484+ preempt_enable();
485+}
486+void hard_enable_TSC(void)
487+{
488+ write_cr4(read_cr4() & ~X86_CR4_TSD);
489+}
490+#endif /* CONFIG_SECCOMP */
491+
492+static noinline void
493+__switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p)
494 {
495 struct thread_struct *next;
496
497@@ -488,33 +523,17 @@ static noinline void __switch_to_xtra(st
498 set_debugreg(next->debugreg[6], 6);
499 set_debugreg(next->debugreg[7], 7);
500 }
501-}
502
503-/*
504- * This function selects if the context switch from prev to next
505- * has to tweak the TSC disable bit in the cr4.
506- */
507-static inline void disable_tsc(struct task_struct *prev_p,
508- struct task_struct *next_p)
509-{
510- struct thread_info *prev, *next;
511-
512- /*
513- * gcc should eliminate the ->thread_info dereference if
514- * has_secure_computing returns 0 at compile time (SECCOMP=n).
515- */
516- prev = task_thread_info(prev_p);
517- next = task_thread_info(next_p);
518-
519- if (has_secure_computing(prev) || has_secure_computing(next)) {
520- /* slow path here */
521- if (has_secure_computing(prev) &&
522- !has_secure_computing(next)) {
523- write_cr4(read_cr4() & ~X86_CR4_TSD);
524- } else if (!has_secure_computing(prev) &&
525- has_secure_computing(next))
526- write_cr4(read_cr4() | X86_CR4_TSD);
527+#ifdef CONFIG_SECCOMP
528+ if (test_tsk_thread_flag(prev_p, TIF_NOTSC) ^
529+ test_tsk_thread_flag(next_p, TIF_NOTSC)) {
530+ /* prev and next are different */
531+ if (test_tsk_thread_flag(next_p, TIF_NOTSC))
532+ hard_disable_TSC();
533+ else
534+ hard_enable_TSC();
535 }
536+#endif
537 }
538
539 /*
540@@ -649,10 +668,9 @@ struct task_struct fastcall * __switch_t
541 /*
542 * Now maybe handle debug registers
543 */
544- if (unlikely(task_thread_info(next_p)->flags & _TIF_WORK_CTXSW))
545- __switch_to_xtra(next_p);
546-
547- disable_tsc(prev_p, next_p);
548+ if (unlikely(task_thread_info(prev_p)->flags & _TIF_WORK_CTXSW_PREV ||
549+ task_thread_info(next_p)->flags & _TIF_WORK_CTXSW_NEXT))
550+ __switch_to_xtra(prev_p, next_p);
551
552 /*
553 * Leave lazy mode, flushing any hypercalls made here.
554--- sle11-2009-05-14.orig/arch/x86/kernel/setup_32-xen.c 2009-03-04 11:25:55.000000000 +0100
555+++ sle11-2009-05-14/arch/x86/kernel/setup_32-xen.c 2009-02-16 16:17:21.000000000 +0100
556@@ -114,19 +114,10 @@ static unsigned int highmem_pages = -1;
557 /*
558 * Setup options
559 */
560-struct drive_info_struct { char dummy[32]; } drive_info;
561-#if defined(CONFIG_BLK_DEV_IDE) || defined(CONFIG_BLK_DEV_HD) || \
562- defined(CONFIG_BLK_DEV_IDE_MODULE) || defined(CONFIG_BLK_DEV_HD_MODULE)
563-EXPORT_SYMBOL(drive_info);
564-#endif
565 struct screen_info screen_info;
566 EXPORT_SYMBOL(screen_info);
567 struct apm_info apm_info;
568 EXPORT_SYMBOL(apm_info);
569-struct sys_desc_table_struct {
570- unsigned short length;
571- unsigned char table[0];
572-};
573 struct edid_info edid_info;
574 EXPORT_SYMBOL_GPL(edid_info);
575 #ifndef CONFIG_XEN
576@@ -149,7 +140,7 @@ unsigned long saved_videomode;
577
578 static char __initdata command_line[COMMAND_LINE_SIZE];
579
580-unsigned char __initdata boot_params[PARAM_SIZE];
581+struct boot_params __initdata boot_params;
582
583 /*
584 * Point at the empty zero page to start with. We map the real shared_info
585@@ -316,18 +307,18 @@ unsigned long __init find_max_low_pfn(vo
586 printk(KERN_WARNING "Warning only %ldMB will be used.\n",
587 MAXMEM>>20);
588 if (max_pfn > MAX_NONPAE_PFN)
589- printk(KERN_WARNING "Use a PAE enabled kernel.\n");
590+ printk(KERN_WARNING "Use a HIGHMEM64G enabled kernel.\n");
591 else
592 printk(KERN_WARNING "Use a HIGHMEM enabled kernel.\n");
593 max_pfn = MAXMEM_PFN;
594 #else /* !CONFIG_HIGHMEM */
595-#ifndef CONFIG_X86_PAE
596+#ifndef CONFIG_HIGHMEM64G
597 if (max_pfn > MAX_NONPAE_PFN) {
598 max_pfn = MAX_NONPAE_PFN;
599 printk(KERN_WARNING "Warning only 4GB will be used.\n");
600- printk(KERN_WARNING "Use a PAE enabled kernel.\n");
601+ printk(KERN_WARNING "Use a HIGHMEM64G enabled kernel.\n");
602 }
603-#endif /* !CONFIG_X86_PAE */
604+#endif /* !CONFIG_HIGHMEM64G */
605 #endif /* !CONFIG_HIGHMEM */
606 } else {
607 if (highmem_pages == -1)
608@@ -514,7 +505,7 @@ void __init setup_bootmem_allocator(void
609 *
610 * This should all compile down to nothing when NUMA is off.
611 */
612-void __init remapped_pgdat_init(void)
613+static void __init remapped_pgdat_init(void)
614 {
615 int nid;
616
617@@ -589,7 +580,6 @@ void __init setup_arch(char **cmdline_p)
618 properly. Setting ROOT_DEV to default to /dev/ram0 breaks initrd.
619 */
620 ROOT_DEV = MKDEV(UNNAMED_MAJOR,0);
621- drive_info = DRIVE_INFO;
622 screen_info = SCREEN_INFO;
623 copy_edid();
624 apm_info.bios = APM_BIOS_INFO;
625@@ -767,6 +757,8 @@ void __init setup_arch(char **cmdline_p)
626 * NOTE: at this point the bootmem allocator is fully available.
627 */
628
629+ paravirt_post_allocator_init();
630+
631 if (is_initial_xendomain())
632 dmi_scan_machine();
633
634@@ -814,6 +806,7 @@ void __init setup_arch(char **cmdline_p)
635 #endif
636
637 e820_register_memory();
638+ e820_mark_nosave_regions();
639
640 if (is_initial_xendomain()) {
641 #ifdef CONFIG_VT
642--- sle11-2009-05-14.orig/arch/x86/kernel/smp_32-xen.c 2008-12-15 11:27:22.000000000 +0100
643+++ sle11-2009-05-14/arch/x86/kernel/smp_32-xen.c 2009-02-16 16:17:21.000000000 +0100
644@@ -22,6 +22,7 @@
645
646 #include <asm/mtrr.h>
647 #include <asm/tlbflush.h>
648+#include <asm/mmu_context.h>
649 #if 0
650 #include <mach_apic.h>
651 #endif
652@@ -217,13 +218,13 @@ static unsigned long flush_va;
653 static DEFINE_SPINLOCK(tlbstate_lock);
654
655 /*
656- * We cannot call mmdrop() because we are in interrupt context,
657+ * We cannot call mmdrop() because we are in interrupt context,
658 * instead update mm->cpu_vm_mask.
659 *
660 * We need to reload %cr3 since the page tables may be going
661 * away from under us..
662 */
663-static inline void leave_mm (unsigned long cpu)
664+void leave_mm(unsigned long cpu)
665 {
666 if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK)
667 BUG();
668--- sle11-2009-05-14.orig/arch/x86/kernel/time_32-xen.c 2009-03-24 10:11:31.000000000 +0100
669+++ sle11-2009-05-14/arch/x86/kernel/time_32-xen.c 2009-03-24 10:12:09.000000000 +0100
670@@ -76,11 +76,12 @@
671 #include <xen/evtchn.h>
672 #include <xen/interface/vcpu.h>
673
674-#ifdef CONFIG_X86_32
675 #include <asm/i8253.h>
676 DEFINE_SPINLOCK(i8253_lock);
677 EXPORT_SYMBOL(i8253_lock);
678-#else
679+
680+#ifdef CONFIG_X86_64
681+#include <asm/vsyscall.h>
682 volatile unsigned long __jiffies __section_jiffies = INITIAL_JIFFIES;
683 #endif
684
685@@ -756,56 +757,10 @@ unsigned long read_persistent_clock(void
686 return retval;
687 }
688
689-static void sync_cmos_clock(unsigned long dummy);
690-
691-static DEFINE_TIMER(sync_cmos_timer, sync_cmos_clock, 0, 0);
692-int no_sync_cmos_clock;
693-
694-static void sync_cmos_clock(unsigned long dummy)
695-{
696- struct timeval now, next;
697- int fail = 1;
698-
699- /*
700- * If we have an externally synchronized Linux clock, then update
701- * CMOS clock accordingly every ~11 minutes. Set_rtc_mmss() has to be
702- * called as close as possible to 500 ms before the new second starts.
703- * This code is run on a timer. If the clock is set, that timer
704- * may not expire at the correct time. Thus, we adjust...
705- */
706- if (!ntp_synced())
707- /*
708- * Not synced, exit, do not restart a timer (if one is
709- * running, let it run out).
710- */
711- return;
712-
713- do_gettimeofday(&now);
714- if (now.tv_usec >= USEC_AFTER - ((unsigned) TICK_SIZE) / 2 &&
715- now.tv_usec <= USEC_BEFORE + ((unsigned) TICK_SIZE) / 2)
716- fail = set_rtc_mmss(now.tv_sec);
717-
718- next.tv_usec = USEC_AFTER - now.tv_usec;
719- if (next.tv_usec <= 0)
720- next.tv_usec += USEC_PER_SEC;
721-
722- if (!fail)
723- next.tv_sec = 659;
724- else
725- next.tv_sec = 0;
726-
727- if (next.tv_usec >= USEC_PER_SEC) {
728- next.tv_sec++;
729- next.tv_usec -= USEC_PER_SEC;
730- }
731- mod_timer(&sync_cmos_timer, jiffies + timeval_to_jiffies(&next));
732-}
733-
734-void notify_arch_cmos_timer(void)
735+int update_persistent_clock(struct timespec now)
736 {
737- if (!no_sync_cmos_clock)
738- mod_timer(&sync_cmos_timer, jiffies + 1);
739 mod_timer(&sync_xen_wallclock_timer, jiffies + 1);
740+ return set_rtc_mmss(now.tv_sec);
741 }
742
743 extern void (*late_time_init)(void);
744--- sle11-2009-05-14.orig/arch/x86/kernel/traps_32-xen.c 2008-12-15 11:27:22.000000000 +0100
745+++ sle11-2009-05-14/arch/x86/kernel/traps_32-xen.c 2009-02-16 16:17:21.000000000 +0100
746@@ -41,6 +41,10 @@
747 #include <linux/mca.h>
748 #endif
749
750+#if defined(CONFIG_EDAC)
751+#include <linux/edac.h>
752+#endif
753+
754 #include <asm/processor.h>
755 #include <asm/system.h>
756 #include <asm/io.h>
757@@ -102,36 +106,45 @@ asmlinkage void machine_check(void);
758 int kstack_depth_to_print = 24;
759 static unsigned int code_bytes = 64;
760
761-static inline int valid_stack_ptr(struct thread_info *tinfo, void *p)
762+static inline int valid_stack_ptr(struct thread_info *tinfo, void *p, unsigned size)
763 {
764 return p > (void *)tinfo &&
765- p < (void *)tinfo + THREAD_SIZE - 3;
766+ p <= (void *)tinfo + THREAD_SIZE - size;
767 }
768
769+/* The form of the top of the frame on the stack */
770+struct stack_frame {
771+ struct stack_frame *next_frame;
772+ unsigned long return_address;
773+};
774+
775 static inline unsigned long print_context_stack(struct thread_info *tinfo,
776 unsigned long *stack, unsigned long ebp,
777 struct stacktrace_ops *ops, void *data)
778 {
779- unsigned long addr;
780-
781 #ifdef CONFIG_FRAME_POINTER
782- while (valid_stack_ptr(tinfo, (void *)ebp)) {
783- unsigned long new_ebp;
784- addr = *(unsigned long *)(ebp + 4);
785+ struct stack_frame *frame = (struct stack_frame *)ebp;
786+ while (valid_stack_ptr(tinfo, frame, sizeof(*frame))) {
787+ struct stack_frame *next;
788+ unsigned long addr;
789+
790+ addr = frame->return_address;
791 ops->address(data, addr);
792 /*
793 * break out of recursive entries (such as
794 * end_of_stack_stop_unwind_function). Also,
795 * we can never allow a frame pointer to
796 * move downwards!
797- */
798- new_ebp = *(unsigned long *)ebp;
799- if (new_ebp <= ebp)
800+ */
801+ next = frame->next_frame;
802+ if (next <= frame)
803 break;
804- ebp = new_ebp;
805+ frame = next;
806 }
807 #else
808- while (valid_stack_ptr(tinfo, stack)) {
809+ while (valid_stack_ptr(tinfo, stack, sizeof(*stack))) {
810+ unsigned long addr;
811+
812 addr = *stack++;
813 if (__kernel_text_address(addr))
814 ops->address(data, addr);
815@@ -154,7 +167,7 @@ void dump_trace(struct task_struct *task
816 if (!stack) {
817 unsigned long dummy;
818 stack = &dummy;
819- if (task && task != current)
820+ if (task != current)
821 stack = (unsigned long *)task->thread.esp;
822 }
823
824@@ -213,6 +226,7 @@ static void print_trace_address(void *da
825 {
826 printk("%s [<%08lx>] ", (char *)data, addr);
827 print_symbol("%s\n", addr);
828+ touch_nmi_watchdog();
829 }
830
831 static struct stacktrace_ops print_trace_ops = {
832@@ -396,7 +410,7 @@ void die(const char * str, struct pt_reg
833 unsigned long esp;
834 unsigned short ss;
835
836- report_bug(regs->eip);
837+ report_bug(regs->eip, regs);
838
839 printk(KERN_EMERG "%s: %04lx [#%d]\n", str, err & 0xffff, ++die_counter);
840 #ifdef CONFIG_PREEMPT
841@@ -439,6 +453,7 @@ void die(const char * str, struct pt_reg
842
843 bust_spinlocks(0);
844 die.lock_owner = -1;
845+ add_taint(TAINT_DIE);
846 spin_unlock_irqrestore(&die.lock, flags);
847
848 if (!regs)
849@@ -523,10 +538,12 @@ fastcall void do_##name(struct pt_regs *
850 do_trap(trapnr, signr, str, 0, regs, error_code, NULL); \
851 }
852
853-#define DO_ERROR_INFO(trapnr, signr, str, name, sicode, siaddr) \
854+#define DO_ERROR_INFO(trapnr, signr, str, name, sicode, siaddr, irq) \
855 fastcall void do_##name(struct pt_regs * regs, long error_code) \
856 { \
857 siginfo_t info; \
858+ if (irq) \
859+ local_irq_enable(); \
860 info.si_signo = signr; \
861 info.si_errno = 0; \
862 info.si_code = sicode; \
863@@ -566,13 +583,13 @@ DO_VM86_ERROR( 3, SIGTRAP, "int3", int3)
864 #endif
865 DO_VM86_ERROR( 4, SIGSEGV, "overflow", overflow)
866 DO_VM86_ERROR( 5, SIGSEGV, "bounds", bounds)
867-DO_ERROR_INFO( 6, SIGILL, "invalid opcode", invalid_op, ILL_ILLOPN, regs->eip)
868+DO_ERROR_INFO( 6, SIGILL, "invalid opcode", invalid_op, ILL_ILLOPN, regs->eip, 0)
869 DO_ERROR( 9, SIGFPE, "coprocessor segment overrun", coprocessor_segment_overrun)
870 DO_ERROR(10, SIGSEGV, "invalid TSS", invalid_TSS)
871 DO_ERROR(11, SIGBUS, "segment not present", segment_not_present)
872 DO_ERROR(12, SIGBUS, "stack segment", stack_segment)
873-DO_ERROR_INFO(17, SIGBUS, "alignment check", alignment_check, BUS_ADRALN, 0)
874-DO_ERROR_INFO(32, SIGSEGV, "iret exception", iret_error, ILL_BADSTK, 0)
875+DO_ERROR_INFO(17, SIGBUS, "alignment check", alignment_check, BUS_ADRALN, 0, 0)
876+DO_ERROR_INFO(32, SIGSEGV, "iret exception", iret_error, ILL_BADSTK, 0, 1)
877
878 fastcall void __kprobes do_general_protection(struct pt_regs * regs,
879 long error_code)
880@@ -585,6 +602,13 @@ fastcall void __kprobes do_general_prote
881
882 current->thread.error_code = error_code;
883 current->thread.trap_no = 13;
884+ if (show_unhandled_signals && unhandled_signal(current, SIGSEGV) &&
885+ printk_ratelimit())
886+ printk(KERN_INFO
887+ "%s[%d] general protection eip:%lx esp:%lx error:%lx\n",
888+ current->comm, current->pid,
889+ regs->eip, regs->esp, error_code);
890+
891 force_sig(SIGSEGV, current);
892 return;
893
894@@ -610,6 +634,14 @@ mem_parity_error(unsigned char reason, s
895 printk(KERN_EMERG "Uhhuh. NMI received for unknown reason %02x on "
896 "CPU %d.\n", reason, smp_processor_id());
897 printk(KERN_EMERG "You have some hardware problem, likely on the PCI bus.\n");
898+
899+#if defined(CONFIG_EDAC)
900+ if(edac_handler_set()) {
901+ edac_atomic_assert_error();
902+ return;
903+ }
904+#endif
905+
906 if (panic_on_unrecovered_nmi)
907 panic("NMI: Not continuing");
908
909@@ -720,6 +752,8 @@ static __kprobes void default_do_nmi(str
910 reassert_nmi();
911 }
912
913+static int ignore_nmis;
914+
915 fastcall __kprobes void do_nmi(struct pt_regs * regs, long error_code)
916 {
917 int cpu;
918@@ -730,11 +764,24 @@ fastcall __kprobes void do_nmi(struct pt
919
920 ++nmi_count(cpu);
921
922- default_do_nmi(regs);
923+ if (!ignore_nmis)
924+ default_do_nmi(regs);
925
926 nmi_exit();
927 }
928
929+void stop_nmi(void)
930+{
931+ acpi_nmi_disable();
932+ ignore_nmis++;
933+}
934+
935+void restart_nmi(void)
936+{
937+ ignore_nmis--;
938+ acpi_nmi_enable();
939+}
940+
941 #ifdef CONFIG_KPROBES
942 fastcall void __kprobes do_int3(struct pt_regs *regs, long error_code)
943 {
944@@ -1023,6 +1070,7 @@ asmlinkage void math_state_restore(void)
945 thread->status |= TS_USEDFPU; /* So we fnsave on switch_to() */
946 tsk->fpu_counter++;
947 }
948+EXPORT_SYMBOL_GPL(math_state_restore);
949
950 #ifndef CONFIG_MATH_EMULATION
951
952--- sle11-2009-05-14.orig/arch/x86/mach-xen/setup.c 2009-03-04 11:28:34.000000000 +0100
953+++ sle11-2009-05-14/arch/x86/mach-xen/setup.c 2009-02-16 16:17:21.000000000 +0100
954@@ -12,6 +12,7 @@
955 #include <asm/e820.h>
956 #include <asm/setup.h>
957 #include <asm/fixmap.h>
958+#include <asm/pgtable.h>
959
960 #include <xen/interface/callback.h>
961 #include <xen/interface/memory.h>
962@@ -101,7 +102,7 @@ void __init pre_setup_arch_hook(void)
963
964 init_mm.pgd = swapper_pg_dir = (pgd_t *)xen_start_info->pt_base;
965
966- setup_xen_features();
967+ xen_setup_features();
968
969 if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0) {
970 hypervisor_virt_start = pp.virt_start;
971@@ -157,4 +158,18 @@ void __init machine_specific_arch_setup(
972 HYPERVISOR_nmi_op(XENNMI_register_callback, &cb);
973 }
974 #endif
975+
976+ /* Do an early initialization of the fixmap area */
977+ {
978+ extern pte_t swapper_pg_pmd[PTRS_PER_PTE];
979+ unsigned long addr = __fix_to_virt(FIX_EARLYCON_MEM_BASE);
980+ pgd_t *pgd = (pgd_t *)xen_start_info->pt_base;
981+ pud_t *pud = pud_offset(pgd + pgd_index(addr), addr);
982+ pmd_t *pmd = pmd_offset(pud, addr);
983+
984+ swapper_pg_dir = pgd;
985+ init_mm.pgd = pgd;
986+ make_lowmem_page_readonly(swapper_pg_pmd, XENFEAT_writable_page_tables);
987+ set_pmd(pmd, __pmd(__pa_symbol(swapper_pg_pmd) | _PAGE_TABLE));
988+ }
989 }
990--- sle11-2009-05-14.orig/arch/x86/mm/fault_32-xen.c 2008-12-15 11:27:22.000000000 +0100
991+++ sle11-2009-05-14/arch/x86/mm/fault_32-xen.c 2009-02-16 16:17:21.000000000 +0100
992@@ -346,7 +346,10 @@ static inline pmd_t *vmalloc_sync_one(pg
993 pmd_k = pmd_offset(pud_k, address);
994 if (!pmd_present(*pmd_k))
995 return NULL;
996- if (!pmd_present(*pmd))
997+ if (!pmd_present(*pmd)) {
998+ bool lazy = x86_read_percpu(xen_lazy_mmu);
999+
1000+ x86_write_percpu(xen_lazy_mmu, false);
1001 #if CONFIG_XEN_COMPAT > 0x030002
1002 set_pmd(pmd, *pmd_k);
1003 #else
1004@@ -356,7 +359,8 @@ static inline pmd_t *vmalloc_sync_one(pg
1005 */
1006 set_pmd(pmd, __pmd(pmd_val(*pmd_k)));
1007 #endif
1008- else
1009+ x86_write_percpu(xen_lazy_mmu, lazy);
1010+ } else
1011 BUG_ON(pmd_page(*pmd) != pmd_page(*pmd_k));
1012 return pmd_k;
1013 }
1014@@ -388,6 +392,8 @@ static inline int vmalloc_fault(unsigned
1015 return 0;
1016 }
1017
1018+int show_unhandled_signals = 1;
1019+
1020 /*
1021 * This routine handles page faults. It determines the address,
1022 * and the problem, and then passes it off to one of the appropriate
1023@@ -408,6 +414,7 @@ fastcall void __kprobes do_page_fault(st
1024 struct vm_area_struct * vma;
1025 unsigned long address;
1026 int write, si_code;
1027+ int fault;
1028
1029 /* get the address */
1030 address = read_cr2();
1031@@ -541,20 +548,18 @@ good_area:
1032 * make sure we exit gracefully rather than endlessly redo
1033 * the fault.
1034 */
1035- switch (handle_mm_fault(mm, vma, address, write)) {
1036- case VM_FAULT_MINOR:
1037- tsk->min_flt++;
1038- break;
1039- case VM_FAULT_MAJOR:
1040- tsk->maj_flt++;
1041- break;
1042- case VM_FAULT_SIGBUS:
1043- goto do_sigbus;
1044- case VM_FAULT_OOM:
1045+ fault = handle_mm_fault(mm, vma, address, write);
1046+ if (unlikely(fault & VM_FAULT_ERROR)) {
1047+ if (fault & VM_FAULT_OOM)
1048 goto out_of_memory;
1049- default:
1050- BUG();
1051+ else if (fault & VM_FAULT_SIGBUS)
1052+ goto do_sigbus;
1053+ BUG();
1054 }
1055+ if (fault & VM_FAULT_MAJOR)
1056+ tsk->maj_flt++;
1057+ else
1058+ tsk->min_flt++;
1059
1060 /*
1061 * Did it hit the DOS screen memory VA from vm86 mode?
1062@@ -589,6 +594,14 @@ bad_area_nosemaphore:
1063 if (is_prefetch(regs, address, error_code))
1064 return;
1065
1066+ if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
1067+ printk_ratelimit()) {
1068+ printk("%s%s[%d]: segfault at %08lx eip %08lx "
1069+ "esp %08lx error %lx\n",
1070+ tsk->pid > 1 ? KERN_INFO : KERN_EMERG,
1071+ tsk->comm, tsk->pid, address, regs->eip,
1072+ regs->esp, error_code);
1073+ }
1074 tsk->thread.cr2 = address;
1075 /* Kernel addresses are always protection faults */
1076 tsk->thread.error_code = error_code | (address >= TASK_SIZE);
1077--- sle11-2009-05-14.orig/arch/x86/mm/highmem_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1078+++ sle11-2009-05-14/arch/x86/mm/highmem_32-xen.c 2009-02-16 16:17:21.000000000 +0100
1079@@ -34,17 +34,16 @@ void *kmap_atomic_prot(struct page *page
1080 /* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */
1081 pagefault_disable();
1082
1083- idx = type + KM_TYPE_NR*smp_processor_id();
1084- BUG_ON(!pte_none(*(kmap_pte-idx)));
1085-
1086 if (!PageHighMem(page))
1087 return page_address(page);
1088
1089+ idx = type + KM_TYPE_NR*smp_processor_id();
1090 vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
1091+ BUG_ON(!pte_none(*(kmap_pte-idx)));
1092 set_pte_at(&init_mm, vaddr, kmap_pte-idx, mk_pte(page, prot));
1093 /*arch_flush_lazy_mmu_mode();*/
1094
1095- return (void*) vaddr;
1096+ return (void *)vaddr;
1097 }
1098
1099 void *kmap_atomic(struct page *page, enum km_type type)
1100--- sle11-2009-05-14.orig/arch/x86/mm/init_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1101+++ sle11-2009-05-14/arch/x86/mm/init_32-xen.c 2009-02-16 16:17:21.000000000 +0100
1102@@ -96,7 +96,7 @@ static pte_t * __init one_page_table_ini
1103 #endif
1104 pte_t *page_table = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE);
1105
1106- paravirt_alloc_pt(__pa(page_table) >> PAGE_SHIFT);
1107+ paravirt_alloc_pt(&init_mm, __pa(page_table) >> PAGE_SHIFT);
1108 make_lowmem_page_readonly(page_table,
1109 XENFEAT_writable_page_tables);
1110 set_pmd(pmd, __pmd(__pa(page_table) | _PAGE_TABLE));
1111@@ -446,7 +446,7 @@ static void __init pagetable_init (void)
1112 xen_pagetable_setup_done(pgd_base);
1113 }
1114
1115-#if defined(CONFIG_SOFTWARE_SUSPEND) || defined(CONFIG_ACPI_SLEEP)
1116+#if defined(CONFIG_HIBERNATION) || defined(CONFIG_ACPI)
1117 /*
1118 * Swap suspend & friends need this for resume because things like the intel-agp
1119 * driver might have split up a kernel 4MB mapping.
1120@@ -485,9 +485,13 @@ void zap_low_mappings (void)
1121 flush_tlb_all();
1122 }
1123
1124+int nx_enabled = 0;
1125+
1126+#ifdef CONFIG_X86_PAE
1127+
1128 static int disable_nx __initdata = 0;
1129 u64 __supported_pte_mask __read_mostly = ~_PAGE_NX;
1130-EXPORT_SYMBOL(__supported_pte_mask);
1131+EXPORT_SYMBOL_GPL(__supported_pte_mask);
1132
1133 /*
1134 * noexec = on|off
1135@@ -514,9 +518,6 @@ static int __init noexec_setup(char *str
1136 }
1137 early_param("noexec", noexec_setup);
1138
1139-int nx_enabled = 0;
1140-#ifdef CONFIG_X86_PAE
1141-
1142 static void __init set_nx(void)
1143 {
1144 unsigned int v[4], l, h;
1145@@ -764,7 +765,7 @@ void __init mem_init(void)
1146 zap_low_mappings();
1147 #endif
1148
1149- set_bit(PG_pinned, &virt_to_page(init_mm.pgd)->flags);
1150+ SetPagePinned(virt_to_page(init_mm.pgd));
1151 }
1152
1153 #ifdef CONFIG_MEMORY_HOTPLUG
1154@@ -796,8 +797,7 @@ void __init pgtable_cache_init(void)
1155 PTRS_PER_PMD*sizeof(pmd_t),
1156 PTRS_PER_PMD*sizeof(pmd_t),
1157 SLAB_PANIC,
1158- pmd_ctor,
1159- NULL);
1160+ pmd_ctor);
1161 if (!SHARED_KERNEL_PMD) {
1162 /* If we're in PAE mode and have a non-shared
1163 kernel pmd, then the pgd size must be a
1164--- sle11-2009-05-14.orig/arch/x86/mm/ioremap_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1165+++ sle11-2009-05-14/arch/x86/mm/ioremap_32-xen.c 2009-02-16 16:17:21.000000000 +0100
1166@@ -363,9 +363,8 @@ void iounmap(volatile void __iomem *addr
1167
1168 /* Reset the direct mapping. Can block */
1169 if ((p->flags >> 20) && is_local_lowmem(p->phys_addr)) {
1170- /* p->size includes the guard page, but cpa doesn't like that */
1171 change_page_attr(virt_to_page(bus_to_virt(p->phys_addr)),
1172- (p->size - PAGE_SIZE) >> PAGE_SHIFT,
1173+ get_vm_area_size(p) >> PAGE_SHIFT,
1174 PAGE_KERNEL);
1175 global_flush_tlb();
1176 }
1177--- sle11-2009-05-14.orig/arch/x86/mm/pgtable_32-xen.c 2008-12-15 11:27:22.000000000 +0100
1178+++ sle11-2009-05-14/arch/x86/mm/pgtable_32-xen.c 2009-02-16 16:17:21.000000000 +0100
1179@@ -198,7 +198,7 @@ void pte_free(struct page *pte)
1180 va, pfn_pte(pfn, PAGE_KERNEL), 0))
1181 BUG();
1182 } else
1183- clear_bit(PG_pinned, &pte->flags);
1184+ ClearPagePinned(pte);
1185
1186 ClearPageForeign(pte);
1187 init_page_count(pte);
1188@@ -248,7 +248,7 @@ static inline void pgd_list_del(pgd_t *p
1189
1190 #if (PTRS_PER_PMD == 1)
1191 /* Non-PAE pgd constructor */
1192-void pgd_ctor(void *pgd)
1193+static void pgd_ctor(void *pgd)
1194 {
1195 unsigned long flags;
1196
1197@@ -271,7 +271,7 @@ void pgd_ctor(void *pgd)
1198 }
1199 #else /* PTRS_PER_PMD > 1 */
1200 /* PAE pgd constructor */
1201-void pgd_ctor(void *pgd)
1202+static void pgd_ctor(void *pgd)
1203 {
1204 /* PAE, kernel PMD may be shared */
1205
1206@@ -285,7 +285,7 @@ void pgd_ctor(void *pgd)
1207 }
1208 #endif /* PTRS_PER_PMD */
1209
1210-void pgd_dtor(void *pgd)
1211+static void pgd_dtor(void *pgd)
1212 {
1213 unsigned long flags; /* can be called from interrupt context */
1214
1215@@ -637,9 +637,9 @@ static inline unsigned int pgd_walk_set_
1216
1217 if (PageHighMem(page)) {
1218 if (pgprot_val(flags) & _PAGE_RW)
1219- clear_bit(PG_pinned, &page->flags);
1220+ ClearPagePinned(page);
1221 else
1222- set_bit(PG_pinned, &page->flags);
1223+ SetPagePinned(page);
1224 } else {
1225 MULTI_update_va_mapping(per_cpu(pb_mcl, cpu) + seq,
1226 (unsigned long)__va(pfn << PAGE_SHIFT),
1227@@ -709,19 +709,19 @@ static void __pgd_pin(pgd_t *pgd)
1228 pgd_walk(pgd, PAGE_KERNEL_RO);
1229 kmap_flush_unused();
1230 xen_pgd_pin(__pa(pgd));
1231- set_bit(PG_pinned, &virt_to_page(pgd)->flags);
1232+ SetPagePinned(virt_to_page(pgd));
1233 }
1234
1235 static void __pgd_unpin(pgd_t *pgd)
1236 {
1237 xen_pgd_unpin(__pa(pgd));
1238 pgd_walk(pgd, PAGE_KERNEL);
1239- clear_bit(PG_pinned, &virt_to_page(pgd)->flags);
1240+ ClearPagePinned(virt_to_page(pgd));
1241 }
1242
1243 static void pgd_test_and_unpin(pgd_t *pgd)
1244 {
1245- if (test_bit(PG_pinned, &virt_to_page(pgd)->flags))
1246+ if (PagePinned(virt_to_page(pgd)))
1247 __pgd_unpin(pgd);
1248 }
1249
1250@@ -759,7 +759,7 @@ void mm_pin_all(void)
1251 */
1252 spin_lock_irqsave(&pgd_lock, flags);
1253 for (page = pgd_list; page; page = (struct page *)page->index) {
1254- if (!test_bit(PG_pinned, &page->flags))
1255+ if (!PagePinned(page))
1256 __pgd_pin((pgd_t *)page_address(page));
1257 }
1258 spin_unlock_irqrestore(&pgd_lock, flags);
1259@@ -767,7 +767,7 @@ void mm_pin_all(void)
1260
1261 void arch_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
1262 {
1263- if (!test_bit(PG_pinned, &virt_to_page(mm->pgd)->flags))
1264+ if (!PagePinned(virt_to_page(mm->pgd)))
1265 mm_pin(mm);
1266 }
1267
1268@@ -793,7 +793,7 @@ void arch_exit_mmap(struct mm_struct *mm
1269
1270 task_unlock(tsk);
1271
1272- if (test_bit(PG_pinned, &virt_to_page(mm->pgd)->flags) &&
1273+ if (PagePinned(virt_to_page(mm->pgd)) &&
1274 (atomic_read(&mm->mm_count) == 1) &&
1275 !mm->context.has_foreign_mappings)
1276 mm_unpin(mm);
1277--- sle11-2009-05-14.orig/arch/x86/pci/irq-xen.c 2008-12-15 11:26:44.000000000 +0100
1278+++ sle11-2009-05-14/arch/x86/pci/irq-xen.c 2009-02-16 16:17:21.000000000 +0100
1279@@ -142,8 +142,9 @@ static void __init pirq_peer_trick(void)
1280 for(i = 1; i < 256; i++) {
1281 if (!busmap[i] || pci_find_bus(0, i))
1282 continue;
1283- if (pci_scan_bus(i, &pci_root_ops, NULL))
1284- printk(KERN_INFO "PCI: Discovered primary peer bus %02x [IRQ]\n", i);
1285+ if (pci_scan_bus_with_sysdata(i))
1286+ printk(KERN_INFO "PCI: Discovered primary peer "
1287+ "bus %02x [IRQ]\n", i);
1288 }
1289 pcibios_last_bus = -1;
1290 }
1291@@ -553,6 +554,7 @@ static __init int intel_router_probe(str
1292 case PCI_DEVICE_ID_INTEL_ICH9_3:
1293 case PCI_DEVICE_ID_INTEL_ICH9_4:
1294 case PCI_DEVICE_ID_INTEL_ICH9_5:
1295+ case PCI_DEVICE_ID_INTEL_TOLAPAI_0:
1296 r->name = "PIIX/ICH";
1297 r->get = pirq_piix_get;
1298 r->set = pirq_piix_set;
1299--- sle11-2009-05-14.orig/arch/x86/xen/Kconfig 2009-05-14 10:56:29.000000000 +0200
1300+++ sle11-2009-05-14/arch/x86/xen/Kconfig 2009-02-16 16:17:21.000000000 +0100
1301@@ -2,7 +2,7 @@
1302 # This Kconfig describes xen options
1303 #
1304
1305-config XEN
1306+config PARAVIRT_XEN
1307 bool "Xen guest support"
1308 select PARAVIRT
1309 select PARAVIRT_CLOCK
1310--- sle11-2009-05-14.orig/arch/x86/xen/xen-head.S 2009-05-14 10:56:29.000000000 +0200
1311+++ sle11-2009-05-14/arch/x86/xen/xen-head.S 2009-02-16 16:17:21.000000000 +0100
1312@@ -1,7 +1,7 @@
1313 /* Xen-specific pieces of head.S, intended to be included in the right
1314 place in head.S */
1315
1316-#ifdef CONFIG_XEN
1317+#ifdef CONFIG_PARAVIRT_XEN
1318
1319 #include <linux/elfnote.h>
1320 #include <linux/init.h>
1321@@ -52,4 +52,4 @@ ENTRY(hypercall_page)
1322 ELFNOTE(Xen, XEN_ELFNOTE_HV_START_LOW, _ASM_PTR __HYPERVISOR_VIRT_START)
1323 ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, _ASM_PTR 0)
1324
1325-#endif /*CONFIG_XEN */
1326+#endif /* CONFIG_PARAVIRT_XEN */
1327--- sle11-2009-05-14.orig/arch/x86/ia32/ia32entry-xen.S 2008-12-15 11:27:22.000000000 +0100
1328+++ sle11-2009-05-14/arch/x86/ia32/ia32entry-xen.S 2009-02-16 16:17:21.000000000 +0100
1329@@ -105,7 +105,7 @@ ENTRY(ia32_sysenter_target)
1330 movl $VSYSCALL32_SYSEXIT,8(%rsp)
1331 movq %rax,(%rsp)
1332 cld
1333- SAVE_ARGS 0,0,0
1334+ SAVE_ARGS 0,0,1
1335 /* no need to do an access_ok check here because rbp has been
1336 32bit zero extended */
1337 1: movl (%rbp),%r9d
1338@@ -244,7 +244,7 @@ ia32_badarg:
1339 */
1340
1341 ENTRY(ia32_syscall)
1342- CFI_STARTPROC simple
1343+ CFI_STARTPROC32 simple
1344 CFI_SIGNAL_FRAME
1345 CFI_DEF_CFA rsp,SS+8-RIP+16
1346 /*CFI_REL_OFFSET ss,SS-RIP+16*/
1347@@ -280,6 +280,7 @@ ia32_sysret:
1348
1349 ia32_tracesys:
1350 SAVE_REST
1351+ CLEAR_RREGS
1352 movq $-ENOSYS,RAX(%rsp) /* really needed? */
1353 movq %rsp,%rdi /* &pt_regs -> arg1 */
1354 call syscall_trace_enter
1355@@ -476,7 +477,7 @@ ia32_sys_call_table:
1356 .quad sys_init_module
1357 .quad sys_delete_module
1358 .quad quiet_ni_syscall /* 130 get_kernel_syms */
1359- .quad sys_quotactl
1360+ .quad sys32_quotactl
1361 .quad sys_getpgid
1362 .quad sys_fchdir
1363 .quad quiet_ni_syscall /* bdflush */
1364@@ -669,4 +670,5 @@ ia32_sys_call_table:
1365 .quad compat_sys_signalfd
1366 .quad compat_sys_timerfd
1367 .quad sys_eventfd
1368+ .quad sys32_fallocate
1369 ia32_syscall_end:
1370--- sle11-2009-05-14.orig/arch/x86/kernel/acpi/sleep_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1371+++ sle11-2009-05-14/arch/x86/kernel/acpi/sleep_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1372@@ -51,12 +51,10 @@
1373 Low-Level Sleep Support
1374 -------------------------------------------------------------------------- */
1375
1376-#ifdef CONFIG_ACPI_SLEEP
1377-
1378 #ifndef CONFIG_ACPI_PV_SLEEP
1379 /* address in low memory of the wakeup routine. */
1380 unsigned long acpi_wakeup_address = 0;
1381-unsigned long acpi_video_flags;
1382+unsigned long acpi_realmode_flags;
1383 extern char wakeup_start, wakeup_end;
1384
1385 extern unsigned long acpi_copy_wakeup_routine(unsigned long);
1386@@ -109,9 +107,11 @@ static int __init acpi_sleep_setup(char
1387 {
1388 while ((str != NULL) && (*str != '\0')) {
1389 if (strncmp(str, "s3_bios", 7) == 0)
1390- acpi_video_flags = 1;
1391+ acpi_realmode_flags |= 1;
1392 if (strncmp(str, "s3_mode", 7) == 0)
1393- acpi_video_flags |= 2;
1394+ acpi_realmode_flags |= 2;
1395+ if (strncmp(str, "s3_beep", 7) == 0)
1396+ acpi_realmode_flags |= 4;
1397 str = strchr(str, ',');
1398 if (str != NULL)
1399 str += strspn(str, ", \t");
1400@@ -123,8 +123,6 @@ static int __init acpi_sleep_setup(char
1401 __setup("acpi_sleep=", acpi_sleep_setup);
1402 #endif /* CONFIG_ACPI_PV_SLEEP */
1403
1404-#endif /*CONFIG_ACPI_SLEEP */
1405-
1406 void acpi_pci_link_exit(void)
1407 {
1408 }
1409--- sle11-2009-05-14.orig/arch/x86/kernel/apic_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1410+++ sle11-2009-05-14/arch/x86/kernel/apic_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1411@@ -50,7 +50,7 @@ void ack_bad_irq(unsigned int irq)
1412 * holds up an irq slot - in excessive cases (when multiple
1413 * unexpected vectors occur) that might lock up the APIC
1414 * completely.
1415- * But don't ack when the APIC is disabled. -AK
1416+ * But don't ack when the APIC is disabled. -AK
1417 */
1418 if (!disable_apic)
1419 ack_APIC_irq();
1420@@ -132,20 +132,6 @@ asmlinkage void smp_spurious_interrupt(v
1421 if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f)))
1422 ack_APIC_irq();
1423
1424-#if 0
1425- static unsigned long last_warning;
1426- static unsigned long skipped;
1427-
1428- /* see sw-dev-man vol 3, chapter 7.4.13.5 */
1429- if (time_before(last_warning+30*HZ,jiffies)) {
1430- printk(KERN_INFO "spurious APIC interrupt on CPU#%d, %ld skipped.\n",
1431- smp_processor_id(), skipped);
1432- last_warning = jiffies;
1433- skipped = 0;
1434- } else {
1435- skipped++;
1436- }
1437-#endif
1438 irq_exit();
1439 }
1440
1441@@ -177,7 +163,7 @@ asmlinkage void smp_error_interrupt(void
1442 7: Illegal register address
1443 */
1444 printk (KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n",
1445- smp_processor_id(), v , v1);
1446+ smp_processor_id(), v , v1);
1447 irq_exit();
1448 }
1449
1450--- sle11-2009-05-14.orig/arch/x86/kernel/e820_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1451+++ sle11-2009-05-14/arch/x86/kernel/e820_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1452@@ -222,37 +222,6 @@ unsigned long __init e820_end_of_ram(voi
1453 }
1454
1455 /*
1456- * Find the hole size in the range.
1457- */
1458-unsigned long __init e820_hole_size(unsigned long start, unsigned long end)
1459-{
1460- unsigned long ram = 0;
1461- int i;
1462-
1463- for (i = 0; i < e820.nr_map; i++) {
1464- struct e820entry *ei = &e820.map[i];
1465- unsigned long last, addr;
1466-
1467- if (ei->type != E820_RAM ||
1468- ei->addr+ei->size <= start ||
1469- ei->addr >= end)
1470- continue;
1471-
1472- addr = round_up(ei->addr, PAGE_SIZE);
1473- if (addr < start)
1474- addr = start;
1475-
1476- last = round_down(ei->addr + ei->size, PAGE_SIZE);
1477- if (last >= end)
1478- last = end;
1479-
1480- if (last > addr)
1481- ram += last - addr;
1482- }
1483- return ((end - start) - ram);
1484-}
1485-
1486-/*
1487 * Mark e820 reserved areas as busy for the resource manager.
1488 */
1489 void __init e820_reserve_resources(struct e820entry *e820, int nr_map)
1490@@ -325,47 +294,61 @@ void __init e820_mark_nosave_regions(voi
1491 }
1492 #endif
1493
1494+/*
1495+ * Finds an active region in the address range from start_pfn to end_pfn and
1496+ * returns its range in ei_startpfn and ei_endpfn for the e820 entry.
1497+ */
1498+static int __init e820_find_active_region(const struct e820entry *ei,
1499+ unsigned long start_pfn,
1500+ unsigned long end_pfn,
1501+ unsigned long *ei_startpfn,
1502+ unsigned long *ei_endpfn)
1503+{
1504+ *ei_startpfn = round_up(ei->addr, PAGE_SIZE) >> PAGE_SHIFT;
1505+ *ei_endpfn = round_down(ei->addr + ei->size, PAGE_SIZE) >> PAGE_SHIFT;
1506+
1507+ /* Skip map entries smaller than a page */
1508+ if (*ei_startpfn >= *ei_endpfn)
1509+ return 0;
1510+
1511+ /* Check if end_pfn_map should be updated */
1512+ if (ei->type != E820_RAM && *ei_endpfn > end_pfn_map)
1513+ end_pfn_map = *ei_endpfn;
1514+
1515+ /* Skip if map is outside the node */
1516+ if (ei->type != E820_RAM || *ei_endpfn <= start_pfn ||
1517+ *ei_startpfn >= end_pfn)
1518+ return 0;
1519+
1520+ /* Check for overlaps */
1521+ if (*ei_startpfn < start_pfn)
1522+ *ei_startpfn = start_pfn;
1523+ if (*ei_endpfn > end_pfn)
1524+ *ei_endpfn = end_pfn;
1525+
1526+ /* Obey end_user_pfn to save on memmap */
1527+ if (*ei_startpfn >= end_user_pfn)
1528+ return 0;
1529+ if (*ei_endpfn > end_user_pfn)
1530+ *ei_endpfn = end_user_pfn;
1531+
1532+ return 1;
1533+}
1534+
1535 /* Walk the e820 map and register active regions within a node */
1536 void __init
1537 e820_register_active_regions(int nid, unsigned long start_pfn,
1538 unsigned long end_pfn)
1539 {
1540+ unsigned long ei_startpfn;
1541+ unsigned long ei_endpfn;
1542 int i;
1543- unsigned long ei_startpfn, ei_endpfn;
1544- for (i = 0; i < e820.nr_map; i++) {
1545- struct e820entry *ei = &e820.map[i];
1546- ei_startpfn = round_up(ei->addr, PAGE_SIZE) >> PAGE_SHIFT;
1547- ei_endpfn = round_down(ei->addr + ei->size, PAGE_SIZE)
1548- >> PAGE_SHIFT;
1549-
1550- /* Skip map entries smaller than a page */
1551- if (ei_startpfn >= ei_endpfn)
1552- continue;
1553-
1554- /* Check if end_pfn_map should be updated */
1555- if (ei->type != E820_RAM && ei_endpfn > end_pfn_map)
1556- end_pfn_map = ei_endpfn;
1557-
1558- /* Skip if map is outside the node */
1559- if (ei->type != E820_RAM ||
1560- ei_endpfn <= start_pfn ||
1561- ei_startpfn >= end_pfn)
1562- continue;
1563-
1564- /* Check for overlaps */
1565- if (ei_startpfn < start_pfn)
1566- ei_startpfn = start_pfn;
1567- if (ei_endpfn > end_pfn)
1568- ei_endpfn = end_pfn;
1569-
1570- /* Obey end_user_pfn to save on memmap */
1571- if (ei_startpfn >= end_user_pfn)
1572- continue;
1573- if (ei_endpfn > end_user_pfn)
1574- ei_endpfn = end_user_pfn;
1575
1576- add_active_range(nid, ei_startpfn, ei_endpfn);
1577- }
1578+ for (i = 0; i < e820.nr_map; i++)
1579+ if (e820_find_active_region(&e820.map[i],
1580+ start_pfn, end_pfn,
1581+ &ei_startpfn, &ei_endpfn))
1582+ add_active_range(nid, ei_startpfn, ei_endpfn);
1583 }
1584
1585 /*
1586@@ -386,12 +369,35 @@ void __init add_memory_region(unsigned l
1587 e820.nr_map++;
1588 }
1589
1590+/*
1591+ * Find the hole size (in bytes) in the memory range.
1592+ * @start: starting address of the memory range to scan
1593+ * @end: ending address of the memory range to scan
1594+ */
1595+unsigned long __init e820_hole_size(unsigned long start, unsigned long end)
1596+{
1597+ unsigned long start_pfn = start >> PAGE_SHIFT;
1598+ unsigned long end_pfn = end >> PAGE_SHIFT;
1599+ unsigned long ei_startpfn;
1600+ unsigned long ei_endpfn;
1601+ unsigned long ram = 0;
1602+ int i;
1603+
1604+ for (i = 0; i < e820.nr_map; i++) {
1605+ if (e820_find_active_region(&e820.map[i],
1606+ start_pfn, end_pfn,
1607+ &ei_startpfn, &ei_endpfn))
1608+ ram += ei_endpfn - ei_startpfn;
1609+ }
1610+ return end - start - (ram << PAGE_SHIFT);
1611+}
1612+
1613 void __init e820_print_map(char *who)
1614 {
1615 int i;
1616
1617 for (i = 0; i < e820.nr_map; i++) {
1618- printk(" %s: %016Lx - %016Lx ", who,
1619+ printk(KERN_INFO " %s: %016Lx - %016Lx ", who,
1620 (unsigned long long) e820.map[i].addr,
1621 (unsigned long long) (e820.map[i].addr + e820.map[i].size));
1622 switch (e820.map[i].type) {
1623--- sle11-2009-05-14.orig/arch/x86/kernel/early_printk-xen.c 2008-12-15 11:27:22.000000000 +0100
1624+++ sle11-2009-05-14/arch/x86/kernel/early_printk-xen.c 2009-02-16 16:17:21.000000000 +0100
1625@@ -174,6 +174,7 @@ static __init void early_serial_init(cha
1626 * mappings. Someone should fix this for domain 0. For now, use fake serial.
1627 */
1628 #define early_vga_console early_serial_console
1629+#define xenboot_console early_serial_console
1630
1631 #endif
1632
1633@@ -261,20 +262,22 @@ static int __init setup_early_printk(cha
1634 } else if (!strncmp(buf, "ttyS", 4)) {
1635 early_serial_init(buf);
1636 early_console = &early_serial_console;
1637- } else if (!strncmp(buf, "vga", 3)
1638+ } else if (!strncmp(buf, "vga", 3)) {
1639 #ifndef CONFIG_XEN
1640 && SCREEN_INFO.orig_video_isVGA == 1) {
1641 max_xpos = SCREEN_INFO.orig_video_cols;
1642 max_ypos = SCREEN_INFO.orig_video_lines;
1643 current_ypos = SCREEN_INFO.orig_y;
1644-#else
1645- || !strncmp(buf, "xen", 3)) {
1646 #endif
1647 early_console = &early_vga_console;
1648 } else if (!strncmp(buf, "simnow", 6)) {
1649 simnow_init(buf + 6);
1650 early_console = &simnow_console;
1651 keep_early = 1;
1652+#ifdef CONFIG_XEN
1653+ } else if (!strncmp(buf, "xen", 3)) {
1654+ early_console = &xenboot_console;
1655+#endif
1656 }
1657
1658 if (keep_early)
1659--- sle11-2009-05-14.orig/arch/x86/kernel/entry_64-xen.S 2008-12-15 11:27:22.000000000 +0100
1660+++ sle11-2009-05-14/arch/x86/kernel/entry_64-xen.S 2009-02-16 16:17:21.000000000 +0100
1661@@ -310,7 +310,7 @@ sysret_signal:
1662 TRACE_IRQS_ON
1663 /* sti */
1664 XEN_UNBLOCK_EVENTS(%rsi)
1665- testl $(_TIF_SIGPENDING|_TIF_NOTIFY_RESUME|_TIF_SINGLESTEP),%edx
1666+ testl $(_TIF_SIGPENDING|_TIF_SINGLESTEP|_TIF_MCE_NOTIFY),%edx
1667 jz 1f
1668
1669 /* Really a signal */
1670@@ -409,7 +409,7 @@ int_very_careful:
1671 jmp int_restore_rest
1672
1673 int_signal:
1674- testl $(_TIF_NOTIFY_RESUME|_TIF_SIGPENDING|_TIF_SINGLESTEP),%edx
1675+ testl $(_TIF_SIGPENDING|_TIF_SINGLESTEP|_TIF_MCE_NOTIFY),%edx
1676 jz 1f
1677 movq %rsp,%rdi # &ptregs -> arg1
1678 xorl %esi,%esi # oldset -> arg2
1679@@ -552,7 +552,7 @@ retint_careful:
1680 jmp retint_check
1681
1682 retint_signal:
1683- testl $(_TIF_SIGPENDING|_TIF_NOTIFY_RESUME|_TIF_SINGLESTEP),%edx
1684+ testl $(_TIF_SIGPENDING|_TIF_SINGLESTEP|_TIF_MCE_NOTIFY),%edx
1685 jz retint_restore_args
1686 TRACE_IRQS_ON
1687 XEN_UNBLOCK_EVENTS(%rsi)
1688--- sle11-2009-05-14.orig/arch/x86/kernel/head_64-xen.S 2008-12-15 11:27:22.000000000 +0100
1689+++ sle11-2009-05-14/arch/x86/kernel/head_64-xen.S 2009-02-16 16:17:21.000000000 +0100
1690@@ -23,7 +23,7 @@
1691 #include <asm/dwarf2.h>
1692 #include <xen/interface/elfnote.h>
1693
1694- .section .bootstrap.text, "ax", @progbits
1695+ .section .text.head, "ax", @progbits
1696 .code64
1697 .globl startup_64
1698 startup_64:
1699@@ -39,7 +39,7 @@ startup_64:
1700
1701 #define NEXT_PAGE(name) \
1702 .balign PAGE_SIZE; \
1703- phys_##name = . - .bootstrap.text; \
1704+ phys_##name = . - .text.head; \
1705 ENTRY(name)
1706
1707 NEXT_PAGE(init_level4_pgt)
1708@@ -66,6 +66,12 @@ NEXT_PAGE(level3_user_pgt)
1709 NEXT_PAGE(level2_kernel_pgt)
1710 .fill 512,8,0
1711
1712+NEXT_PAGE(level2_fixmap_pgt)
1713+ .fill 512,8,0
1714+
1715+NEXT_PAGE(level1_fixmap_pgt)
1716+ .fill 512,8,0
1717+
1718 NEXT_PAGE(hypercall_page)
1719 CFI_STARTPROC
1720 .rept 0x1000 / 0x20
1721@@ -172,18 +178,18 @@ ENTRY(empty_zero_page)
1722 .byte 0
1723 #endif /* CONFIG_XEN_COMPAT <= 0x030002 */
1724
1725- ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS, .asciz, "linux")
1726- ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION, .asciz, "2.6")
1727- ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION, .asciz, "xen-3.0")
1728- ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE, .quad, __START_KERNEL_map)
1729+ ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS, .asciz "linux")
1730+ ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION, .asciz "2.6")
1731+ ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION, .asciz "xen-3.0")
1732+ ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE, .quad __START_KERNEL_map)
1733 #if CONFIG_XEN_COMPAT <= 0x030002
1734- ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .quad, __START_KERNEL_map)
1735+ ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .quad __START_KERNEL_map)
1736 #else
1737- ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .quad, 0)
1738+ ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .quad 0)
1739 #endif
1740- ELFNOTE(Xen, XEN_ELFNOTE_ENTRY, .quad, startup_64)
1741- ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .quad, hypercall_page)
1742- ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .quad, _PAGE_PRESENT,_PAGE_PRESENT)
1743- ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz, "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel")
1744- ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz, "generic")
1745- ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long, 1)
1746+ ELFNOTE(Xen, XEN_ELFNOTE_ENTRY, .quad startup_64)
1747+ ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .quad hypercall_page)
1748+ ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .quad _PAGE_PRESENT, _PAGE_PRESENT)
1749+ ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel")
1750+ ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz "generic")
1751+ ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long 1)
1752--- sle11-2009-05-14.orig/arch/x86/kernel/head64-xen.c 2008-12-15 11:27:22.000000000 +0100
1753+++ sle11-2009-05-14/arch/x86/kernel/head64-xen.c 2009-02-16 16:17:21.000000000 +0100
1754@@ -90,7 +90,7 @@ void __init x86_64_start_kernel(char * r
1755 unsigned long machine_to_phys_nr_ents;
1756 int i;
1757
1758- setup_xen_features();
1759+ xen_setup_features();
1760
1761 xen_start_info = (struct start_info *)real_mode_data;
1762 if (!xen_feature(XENFEAT_auto_translated_physmap))
1763--- sle11-2009-05-14.orig/arch/x86/kernel/io_apic_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1764+++ sle11-2009-05-14/arch/x86/kernel/io_apic_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1765@@ -165,7 +165,9 @@ static inline void io_apic_write(unsigne
1766 #endif
1767 }
1768
1769-#ifndef CONFIG_XEN
1770+#ifdef CONFIG_XEN
1771+#define io_apic_modify io_apic_write
1772+#else
1773 /*
1774 * Re-write a value: to be used for read-modify-write
1775 * cycles where the read already set up the index register.
1776@@ -175,8 +177,32 @@ static inline void io_apic_modify(unsign
1777 struct io_apic __iomem *io_apic = io_apic_base(apic);
1778 writel(value, &io_apic->data);
1779 }
1780-#else
1781-#define io_apic_modify io_apic_write
1782+
1783+static int io_apic_level_ack_pending(unsigned int irq)
1784+{
1785+ struct irq_pin_list *entry;
1786+ unsigned long flags;
1787+ int pending = 0;
1788+
1789+ spin_lock_irqsave(&ioapic_lock, flags);
1790+ entry = irq_2_pin + irq;
1791+ for (;;) {
1792+ unsigned int reg;
1793+ int pin;
1794+
1795+ pin = entry->pin;
1796+ if (pin == -1)
1797+ break;
1798+ reg = io_apic_read(entry->apic, 0x10 + pin*2);
1799+ /* Is the remote IRR bit set? */
1800+ pending |= (reg >> 14) & 1;
1801+ if (!entry->next)
1802+ break;
1803+ entry = irq_2_pin + entry->next;
1804+ }
1805+ spin_unlock_irqrestore(&ioapic_lock, flags);
1806+ return pending;
1807+}
1808 #endif
1809
1810 /*
1811@@ -406,14 +432,12 @@ static void clear_IO_APIC (void)
1812 int skip_ioapic_setup;
1813 int ioapic_force;
1814
1815-/* dummy parsing: see setup.c */
1816-
1817-static int __init disable_ioapic_setup(char *str)
1818+static int __init parse_noapic(char *str)
1819 {
1820- skip_ioapic_setup = 1;
1821+ disable_ioapic_setup();
1822 return 0;
1823 }
1824-early_param("noapic", disable_ioapic_setup);
1825+early_param("noapic", parse_noapic);
1826
1827 /* Actually the next is obsolete, but keep it for paranoid reasons -AK */
1828 static int __init disable_timer_pin_setup(char *arg)
1829@@ -765,12 +789,15 @@ static struct irq_chip ioapic_chip;
1830
1831 static void ioapic_register_intr(int irq, unsigned long trigger)
1832 {
1833- if (trigger)
1834+ if (trigger) {
1835+ irq_desc[irq].status |= IRQ_LEVEL;
1836 set_irq_chip_and_handler_name(irq, &ioapic_chip,
1837 handle_fasteoi_irq, "fasteoi");
1838- else
1839+ } else {
1840+ irq_desc[irq].status &= ~IRQ_LEVEL;
1841 set_irq_chip_and_handler_name(irq, &ioapic_chip,
1842 handle_edge_irq, "edge");
1843+ }
1844 }
1845 #else
1846 #define ioapic_register_intr(irq, trigger) evtchn_register_pirq(irq)
1847@@ -1419,9 +1446,37 @@ static void ack_apic_level(unsigned int
1848 ack_APIC_irq();
1849
1850 /* Now we can move and renable the irq */
1851- move_masked_irq(irq);
1852- if (unlikely(do_unmask_irq))
1853+ if (unlikely(do_unmask_irq)) {
1854+ /* Only migrate the irq if the ack has been received.
1855+ *
1856+ * On rare occasions the broadcast level triggered ack gets
1857+ * delayed going to ioapics, and if we reprogram the
1858+ * vector while Remote IRR is still set the irq will never
1859+ * fire again.
1860+ *
1861+ * To prevent this scenario we read the Remote IRR bit
1862+ * of the ioapic. This has two effects.
1863+ * - On any sane system the read of the ioapic will
1864+ * flush writes (and acks) going to the ioapic from
1865+ * this cpu.
1866+ * - We get to see if the ACK has actually been delivered.
1867+ *
1868+ * Based on failed experiments of reprogramming the
1869+ * ioapic entry from outside of irq context starting
1870+ * with masking the ioapic entry and then polling until
1871+ * Remote IRR was clear before reprogramming the
1872+ * ioapic I don't trust the Remote IRR bit to be
1873+ * completey accurate.
1874+ *
1875+ * However there appears to be no other way to plug
1876+ * this race, so if the Remote IRR bit is not
1877+ * accurate and is causing problems then it is a hardware bug
1878+ * and you can go talk to the chipset vendor about it.
1879+ */
1880+ if (!io_apic_level_ack_pending(irq))
1881+ move_masked_irq(irq);
1882 unmask_IO_APIC_irq(irq);
1883+ }
1884 }
1885
1886 static struct irq_chip ioapic_chip __read_mostly = {
1887--- sle11-2009-05-14.orig/arch/x86/kernel/ldt_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1888+++ sle11-2009-05-14/arch/x86/kernel/ldt_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1889@@ -114,6 +114,8 @@ int init_new_context(struct task_struct
1890 memset(&mm->context, 0, sizeof(mm->context));
1891 init_MUTEX(&mm->context.sem);
1892 old_mm = current->mm;
1893+ if (old_mm)
1894+ mm->context.vdso = old_mm->context.vdso;
1895 if (old_mm && old_mm->context.size > 0) {
1896 down(&old_mm->context.sem);
1897 retval = copy_ldt(&mm->context, &old_mm->context);
1898@@ -146,7 +148,7 @@ void destroy_context(struct mm_struct *m
1899 kfree(mm->context.ldt);
1900 mm->context.size = 0;
1901 }
1902- if (!mm->context.pinned) {
1903+ if (!PagePinned(virt_to_page(mm->pgd))) {
1904 spin_lock(&mm_unpinned_lock);
1905 list_del(&mm->context.unpinned);
1906 spin_unlock(&mm_unpinned_lock);
1907--- sle11-2009-05-14.orig/arch/x86/kernel/mpparse_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1908+++ sle11-2009-05-14/arch/x86/kernel/mpparse_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1909@@ -32,7 +32,6 @@
1910
1911 /* Have we found an MP table */
1912 int smp_found_config;
1913-unsigned int __initdata maxcpus = NR_CPUS;
1914
1915 /*
1916 * Various Linux-internal data structures created from the
1917@@ -657,6 +656,20 @@ static int mp_find_ioapic(int gsi)
1918 return -1;
1919 }
1920
1921+static u8 uniq_ioapic_id(u8 id)
1922+{
1923+ int i;
1924+ DECLARE_BITMAP(used, 256);
1925+ bitmap_zero(used, 256);
1926+ for (i = 0; i < nr_ioapics; i++) {
1927+ struct mpc_config_ioapic *ia = &mp_ioapics[i];
1928+ __set_bit(ia->mpc_apicid, used);
1929+ }
1930+ if (!test_bit(id, used))
1931+ return id;
1932+ return find_first_zero_bit(used, 256);
1933+}
1934+
1935 void __init mp_register_ioapic(u8 id, u32 address, u32 gsi_base)
1936 {
1937 int idx = 0;
1938@@ -664,7 +677,7 @@ void __init mp_register_ioapic(u8 id, u3
1939 if (bad_ioapic(address))
1940 return;
1941
1942- idx = nr_ioapics++;
1943+ idx = nr_ioapics;
1944
1945 mp_ioapics[idx].mpc_type = MP_IOAPIC;
1946 mp_ioapics[idx].mpc_flags = MPC_APIC_USABLE;
1947@@ -673,7 +686,7 @@ void __init mp_register_ioapic(u8 id, u3
1948 #ifndef CONFIG_XEN
1949 set_fixmap_nocache(FIX_IO_APIC_BASE_0 + idx, address);
1950 #endif
1951- mp_ioapics[idx].mpc_apicid = id;
1952+ mp_ioapics[idx].mpc_apicid = uniq_ioapic_id(id);
1953 mp_ioapics[idx].mpc_apicver = 0;
1954
1955 /*
1956@@ -690,6 +703,8 @@ void __init mp_register_ioapic(u8 id, u3
1957 mp_ioapics[idx].mpc_apicaddr,
1958 mp_ioapic_routing[idx].gsi_start,
1959 mp_ioapic_routing[idx].gsi_end);
1960+
1961+ nr_ioapics++;
1962 }
1963
1964 void __init
1965--- sle11-2009-05-14.orig/arch/x86/kernel/process_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1966+++ sle11-2009-05-14/arch/x86/kernel/process_64-xen.c 2009-02-16 16:17:21.000000000 +0100
1967@@ -26,6 +26,7 @@
1968 #include <linux/sched.h>
1969 #include <linux/kernel.h>
1970 #include <linux/mm.h>
1971+#include <linux/fs.h>
1972 #include <linux/elfcore.h>
1973 #include <linux/smp.h>
1974 #include <linux/slab.h>
1975@@ -249,6 +250,7 @@ early_param("idle", idle_setup);
1976 void __show_regs(struct pt_regs * regs)
1977 {
1978 unsigned long fs, gs, shadowgs;
1979+ unsigned long d0, d1, d2, d3, d6, d7;
1980 unsigned int fsindex,gsindex;
1981 unsigned int ds,cs,es;
1982
1983@@ -288,6 +290,14 @@ void __show_regs(struct pt_regs * regs)
1984 fs,fsindex,gs,gsindex,shadowgs);
1985 printk("CS: %04x DS: %04x ES: %04x\n", cs, ds, es);
1986
1987+ get_debugreg(d0, 0);
1988+ get_debugreg(d1, 1);
1989+ get_debugreg(d2, 2);
1990+ printk("DR0: %016lx DR1: %016lx DR2: %016lx\n", d0, d1, d2);
1991+ get_debugreg(d3, 3);
1992+ get_debugreg(d6, 6);
1993+ get_debugreg(d7, 7);
1994+ printk("DR3: %016lx DR6: %016lx DR7: %016lx\n", d3, d6, d7);
1995 }
1996
1997 void show_regs(struct pt_regs *regs)
1998--- sle11-2009-05-14.orig/arch/x86/kernel/setup_64-xen.c 2008-12-15 11:27:22.000000000 +0100
1999+++ sle11-2009-05-14/arch/x86/kernel/setup_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2000@@ -827,6 +827,8 @@ static void __cpuinit init_amd(struct cp
2001 level = cpuid_eax(1);
2002 if (c->x86 == 15 && ((level >= 0x0f48 && level < 0x0f50) || level >= 0x0f58))
2003 set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability);
2004+ if (c->x86 == 0x10)
2005+ set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability);
2006
2007 /* Enable workaround for FXSAVE leak */
2008 if (c->x86 >= 6)
2009@@ -852,8 +854,14 @@ static void __cpuinit init_amd(struct cp
2010 if (c->extended_cpuid_level >= 0x80000008)
2011 amd_detect_cmp(c);
2012
2013- /* Fix cpuid4 emulation for more */
2014- num_cache_leaves = 3;
2015+ if (c->extended_cpuid_level >= 0x80000006 &&
2016+ (cpuid_edx(0x80000006) & 0xf000))
2017+ num_cache_leaves = 4;
2018+ else
2019+ num_cache_leaves = 3;
2020+
2021+ if (c->x86 == 0xf || c->x86 == 0x10 || c->x86 == 0x11)
2022+ set_bit(X86_FEATURE_K8, &c->x86_capability);
2023
2024 /* RDTSC can be speculated around */
2025 clear_bit(X86_FEATURE_SYNC_RDTSC, &c->x86_capability);
2026@@ -1098,6 +1106,8 @@ void __cpuinit identify_cpu(struct cpuin
2027 c->x86_capability[2] = cpuid_edx(0x80860001);
2028 }
2029
2030+ init_scattered_cpuid_features(c);
2031+
2032 c->apicid = phys_pkg_id(0);
2033
2034 /*
2035@@ -1183,7 +1193,7 @@ static int show_cpuinfo(struct seq_file
2036 "fpu", "vme", "de", "pse", "tsc", "msr", "pae", "mce",
2037 "cx8", "apic", NULL, "sep", "mtrr", "pge", "mca", "cmov",
2038 "pat", "pse36", "pn", "clflush", NULL, "dts", "acpi", "mmx",
2039- "fxsr", "sse", "sse2", "ss", "ht", "tm", "ia64", NULL,
2040+ "fxsr", "sse", "sse2", "ss", "ht", "tm", "ia64", "pbe",
2041
2042 /* AMD-defined */
2043 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2044@@ -1199,10 +1209,11 @@ static int show_cpuinfo(struct seq_file
2045 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2046
2047 /* Other (Linux-defined) */
2048- "cxmmx", NULL, "cyrix_arr", "centaur_mcr", NULL,
2049- "constant_tsc", NULL, NULL,
2050- "up", NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2051- NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2052+ "cxmmx", "k6_mtrr", "cyrix_arr", "centaur_mcr",
2053+ NULL, NULL, NULL, NULL,
2054+ "constant_tsc", "up", NULL, "arch_perfmon",
2055+ "pebs", "bts", NULL, "sync_rdtsc",
2056+ "rep_good", NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2057 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2058
2059 /* Intel-defined (#2) */
2060@@ -1213,7 +1224,7 @@ static int show_cpuinfo(struct seq_file
2061
2062 /* VIA/Cyrix/Centaur-defined */
2063 NULL, NULL, "rng", "rng_en", NULL, NULL, "ace", "ace_en",
2064- NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2065+ "ace2", "ace2_en", "phe", "phe_en", "pmm", "pmm_en", NULL, NULL,
2066 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2067 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2068
2069@@ -1224,6 +1235,12 @@ static int show_cpuinfo(struct seq_file
2070 "osvw", "ibs", NULL, NULL, NULL, NULL,
2071 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2072 NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2073+
2074+ /* Auxiliary (Linux-defined) */
2075+ "ida", NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2076+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2077+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2078+ NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
2079 };
2080 static char *x86_power_flags[] = {
2081 "ts", /* temperature sensor */
2082--- sle11-2009-05-14.orig/arch/x86/kernel/setup64-xen.c 2008-12-15 11:27:22.000000000 +0100
2083+++ sle11-2009-05-14/arch/x86/kernel/setup64-xen.c 2009-02-16 16:17:21.000000000 +0100
2084@@ -125,11 +125,14 @@ void __init setup_per_cpu_areas(void)
2085 }
2086
2087 #ifdef CONFIG_XEN
2088-static void switch_pt(void)
2089+static void __init_refok switch_pt(int cpu)
2090 {
2091+ if (cpu == 0)
2092+ xen_init_pt();
2093 xen_pt_switch(__pa_symbol(init_level4_pgt));
2094 xen_new_user_pt(__pa_symbol(__user_pgd(init_level4_pgt)));
2095 }
2096+#define switch_pt() switch_pt(cpu)
2097
2098 static void __cpuinit cpu_gdt_init(const struct desc_ptr *gdt_descr)
2099 {
2100@@ -185,9 +188,6 @@ void pda_init(int cpu)
2101 pda->mmu_state = 0;
2102
2103 if (cpu == 0) {
2104-#ifdef CONFIG_XEN
2105- xen_init_pt();
2106-#endif
2107 /* others are initialized in smpboot.c */
2108 pda->pcurrent = &init_task;
2109 pda->irqstackptr = boot_cpu_stack;
2110--- sle11-2009-05-14.orig/arch/x86/kernel/smp_64-xen.c 2008-12-15 11:27:22.000000000 +0100
2111+++ sle11-2009-05-14/arch/x86/kernel/smp_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2112@@ -362,7 +362,7 @@ __smp_call_function_single(int cpu, void
2113 }
2114
2115 /*
2116- * smp_call_function_single - Run a function on another CPU
2117+ * smp_call_function_single - Run a function on a specific CPU
2118 * @func: The function to run. This must be fast and non-blocking.
2119 * @info: An arbitrary pointer to pass to the function.
2120 * @nonatomic: Currently unused.
2121@@ -379,17 +379,21 @@ int smp_call_function_single (int cpu, v
2122 {
2123 /* prevent preemption and reschedule on another processor */
2124 int me = get_cpu();
2125+
2126+ /* Can deadlock when called with interrupts disabled */
2127+ WARN_ON(irqs_disabled());
2128+
2129 if (cpu == me) {
2130+ local_irq_disable();
2131+ func(info);
2132+ local_irq_enable();
2133 put_cpu();
2134 return 0;
2135 }
2136
2137- /* Can deadlock when called with interrupts disabled */
2138- WARN_ON(irqs_disabled());
2139-
2140- spin_lock_bh(&call_lock);
2141+ spin_lock(&call_lock);
2142 __smp_call_function_single(cpu, func, info, nonatomic, wait);
2143- spin_unlock_bh(&call_lock);
2144+ spin_unlock(&call_lock);
2145 put_cpu();
2146 return 0;
2147 }
2148--- sle11-2009-05-14.orig/arch/x86/kernel/traps_64-xen.c 2008-12-15 11:27:22.000000000 +0100
2149+++ sle11-2009-05-14/arch/x86/kernel/traps_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2150@@ -34,6 +34,10 @@
2151 #include <linux/bug.h>
2152 #include <linux/kdebug.h>
2153
2154+#if defined(CONFIG_EDAC)
2155+#include <linux/edac.h>
2156+#endif
2157+
2158 #include <asm/system.h>
2159 #include <asm/io.h>
2160 #include <asm/atomic.h>
2161@@ -332,6 +336,7 @@ static int print_trace_stack(void *data,
2162
2163 static void print_trace_address(void *data, unsigned long addr)
2164 {
2165+ touch_nmi_watchdog();
2166 printk_address(addr);
2167 }
2168
2169@@ -520,6 +525,7 @@ void __kprobes __die(const char * str, s
2170 printk("\n");
2171 notify_die(DIE_OOPS, str, regs, err, current->thread.trap_no, SIGSEGV);
2172 show_registers(regs);
2173+ add_taint(TAINT_DIE);
2174 /* Executive summary in case the oops scrolled away */
2175 printk(KERN_ALERT "RIP ");
2176 printk_address(regs->rip);
2177@@ -533,7 +539,7 @@ void die(const char * str, struct pt_reg
2178 unsigned long flags = oops_begin();
2179
2180 if (!user_mode(regs))
2181- report_bug(regs->rip);
2182+ report_bug(regs->rip, regs);
2183
2184 __die(str, regs, err);
2185 oops_end(flags);
2186@@ -582,7 +588,8 @@ static void __kprobes do_trap(int trapnr
2187 tsk->thread.error_code = error_code;
2188 tsk->thread.trap_no = trapnr;
2189
2190- if (exception_trace && unhandled_signal(tsk, signr))
2191+ if (show_unhandled_signals && unhandled_signal(tsk, signr) &&
2192+ printk_ratelimit())
2193 printk(KERN_INFO
2194 "%s[%d] trap %s rip:%lx rsp:%lx error:%lx\n",
2195 tsk->comm, tsk->pid, str,
2196@@ -686,7 +693,8 @@ asmlinkage void __kprobes do_general_pro
2197 tsk->thread.error_code = error_code;
2198 tsk->thread.trap_no = 13;
2199
2200- if (exception_trace && unhandled_signal(tsk, SIGSEGV))
2201+ if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
2202+ printk_ratelimit())
2203 printk(KERN_INFO
2204 "%s[%d] general protection rip:%lx rsp:%lx error:%lx\n",
2205 tsk->comm, tsk->pid,
2206@@ -721,6 +729,13 @@ mem_parity_error(unsigned char reason, s
2207 reason);
2208 printk(KERN_EMERG "You have some hardware problem, likely on the PCI bus.\n");
2209
2210+#if defined(CONFIG_EDAC)
2211+ if(edac_handler_set()) {
2212+ edac_atomic_assert_error();
2213+ return;
2214+ }
2215+#endif
2216+
2217 if (panic_on_unrecovered_nmi)
2218 panic("NMI: Not continuing");
2219
2220--- sle11-2009-05-14.orig/arch/x86/kernel/vsyscall_64-xen.c 2008-12-15 11:27:22.000000000 +0100
2221+++ sle11-2009-05-14/arch/x86/kernel/vsyscall_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2222@@ -42,6 +42,7 @@
2223 #include <asm/segment.h>
2224 #include <asm/desc.h>
2225 #include <asm/topology.h>
2226+#include <asm/vgtod.h>
2227
2228 #define __vsyscall(nr) __attribute__ ((unused,__section__(".vsyscall_" #nr)))
2229 #define __syscall_clobber "r11","rcx","memory"
2230@@ -57,26 +58,9 @@
2231 * - writen by timer interrupt or systcl (/proc/sys/kernel/vsyscall64)
2232 * Try to keep this structure as small as possible to avoid cache line ping pongs
2233 */
2234-struct vsyscall_gtod_data_t {
2235- seqlock_t lock;
2236-
2237- /* open coded 'struct timespec' */
2238- time_t wall_time_sec;
2239- u32 wall_time_nsec;
2240-
2241- int sysctl_enabled;
2242- struct timezone sys_tz;
2243- struct { /* extract of a clocksource struct */
2244- cycle_t (*vread)(void);
2245- cycle_t cycle_last;
2246- cycle_t mask;
2247- u32 mult;
2248- u32 shift;
2249- } clock;
2250-};
2251 int __vgetcpu_mode __section_vgetcpu_mode;
2252
2253-struct vsyscall_gtod_data_t __vsyscall_gtod_data __section_vsyscall_gtod_data =
2254+struct vsyscall_gtod_data __vsyscall_gtod_data __section_vsyscall_gtod_data =
2255 {
2256 .lock = SEQLOCK_UNLOCKED,
2257 .sysctl_enabled = 1,
2258@@ -96,6 +80,8 @@ void update_vsyscall(struct timespec *wa
2259 vsyscall_gtod_data.wall_time_sec = wall_time->tv_sec;
2260 vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
2261 vsyscall_gtod_data.sys_tz = sys_tz;
2262+ vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec;
2263+ vsyscall_gtod_data.wall_to_monotonic = wall_to_monotonic;
2264 write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags);
2265 }
2266
2267--- sle11-2009-05-14.orig/arch/x86/mm/fault_64-xen.c 2008-12-15 11:27:22.000000000 +0100
2268+++ sle11-2009-05-14/arch/x86/mm/fault_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2269@@ -159,7 +159,9 @@ void dump_pagetable(unsigned long addres
2270 pmd_t *pmd;
2271 pte_t *pte;
2272
2273- pgd = __va(read_cr3() & PHYSICAL_PAGE_MASK);
2274+ pgd = (pgd_t *)read_cr3();
2275+
2276+ pgd = __va((unsigned long)pgd & PHYSICAL_PAGE_MASK);
2277 pgd += pgd_index(address);
2278 if (bad_address(pgd)) goto bad;
2279 printk("PGD %lx ", pgd_val(*pgd));
2280@@ -219,16 +221,6 @@ static int is_errata93(struct pt_regs *r
2281 return 0;
2282 }
2283
2284-int unhandled_signal(struct task_struct *tsk, int sig)
2285-{
2286- if (is_init(tsk))
2287- return 1;
2288- if (tsk->ptrace & PT_PTRACED)
2289- return 0;
2290- return (tsk->sighand->action[sig-1].sa.sa_handler == SIG_IGN) ||
2291- (tsk->sighand->action[sig-1].sa.sa_handler == SIG_DFL);
2292-}
2293-
2294 static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs,
2295 unsigned long error_code)
2296 {
2297@@ -302,8 +294,8 @@ static int vmalloc_fault(unsigned long a
2298 return 0;
2299 }
2300
2301-int page_fault_trace = 0;
2302-int exception_trace = 1;
2303+static int page_fault_trace;
2304+int show_unhandled_signals = 1;
2305
2306
2307 #define MEM_VERBOSE 1
2308@@ -372,7 +364,7 @@ asmlinkage void __kprobes do_page_fault(
2309 struct vm_area_struct * vma;
2310 unsigned long address;
2311 const struct exception_table_entry *fixup;
2312- int write;
2313+ int write, fault;
2314 unsigned long flags;
2315 siginfo_t info;
2316
2317@@ -384,7 +376,7 @@ asmlinkage void __kprobes do_page_fault(
2318 prefetchw(&mm->mmap_sem);
2319
2320 /* get the address */
2321- address = current_vcpu_info()->arch.cr2;
2322+ address = read_cr2();
2323
2324 info.si_code = SEGV_MAPERR;
2325
2326@@ -445,6 +437,13 @@ asmlinkage void __kprobes do_page_fault(
2327 if (unlikely(in_atomic() || !mm))
2328 goto bad_area_nosemaphore;
2329
2330+ /*
2331+ * User-mode registers count as a user access even for any
2332+ * potential system fault or CPU buglet.
2333+ */
2334+ if (user_mode_vm(regs))
2335+ error_code |= PF_USER;
2336+
2337 again:
2338 /* When running in the kernel we expect faults to occur only to
2339 * addresses in user space. All other faults represent errors in the
2340@@ -511,19 +510,18 @@ good_area:
2341 * make sure we exit gracefully rather than endlessly redo
2342 * the fault.
2343 */
2344- switch (handle_mm_fault(mm, vma, address, write)) {
2345- case VM_FAULT_MINOR:
2346- tsk->min_flt++;
2347- break;
2348- case VM_FAULT_MAJOR:
2349- tsk->maj_flt++;
2350- break;
2351- case VM_FAULT_SIGBUS:
2352- goto do_sigbus;
2353- default:
2354- goto out_of_memory;
2355+ fault = handle_mm_fault(mm, vma, address, write);
2356+ if (unlikely(fault & VM_FAULT_ERROR)) {
2357+ if (fault & VM_FAULT_OOM)
2358+ goto out_of_memory;
2359+ else if (fault & VM_FAULT_SIGBUS)
2360+ goto do_sigbus;
2361+ BUG();
2362 }
2363-
2364+ if (fault & VM_FAULT_MAJOR)
2365+ tsk->maj_flt++;
2366+ else
2367+ tsk->min_flt++;
2368 up_read(&mm->mmap_sem);
2369 return;
2370
2371@@ -556,7 +554,8 @@ bad_area_nosemaphore:
2372 (address >> 32))
2373 return;
2374
2375- if (exception_trace && unhandled_signal(tsk, SIGSEGV)) {
2376+ if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
2377+ printk_ratelimit()) {
2378 printk(
2379 "%s%s[%d]: segfault at %016lx rip %016lx rsp %016lx error %lx\n",
2380 tsk->pid > 1 ? KERN_INFO : KERN_EMERG,
2381@@ -630,7 +629,7 @@ out_of_memory:
2382 }
2383 printk("VM: killing process %s\n", tsk->comm);
2384 if (error_code & 4)
2385- do_exit(SIGKILL);
2386+ do_group_exit(SIGKILL);
2387 goto no_context;
2388
2389 do_sigbus:
2390--- sle11-2009-05-14.orig/arch/x86/mm/init_64-xen.c 2008-12-15 11:27:22.000000000 +0100
2391+++ sle11-2009-05-14/arch/x86/mm/init_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2392@@ -66,6 +66,9 @@ int after_bootmem;
2393 DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
2394 extern unsigned long start_pfn;
2395
2396+extern pmd_t level2_fixmap_pgt[PTRS_PER_PMD];
2397+extern pte_t level1_fixmap_pgt[PTRS_PER_PTE];
2398+
2399 /*
2400 * Use this until direct mapping is established, i.e. before __va() is
2401 * available in init_memory_mapping().
2402@@ -362,6 +365,10 @@ __set_fixmap (enum fixed_addresses idx,
2403 set_pte_phys(address, phys, prot, 0);
2404 set_pte_phys(address, phys, prot, 1);
2405 break;
2406+ case FIX_EARLYCON_MEM_BASE:
2407+ xen_l1_entry_update(level1_fixmap_pgt + pte_index(address),
2408+ pfn_pte_ma(phys >> PAGE_SHIFT, prot));
2409+ break;
2410 default:
2411 set_pte_phys_ma(address, phys, prot);
2412 break;
2413@@ -594,6 +601,13 @@ void __init xen_init_pt(void)
2414 __user_pgd(init_level4_pgt)[pgd_index(VSYSCALL_START)] =
2415 __pgd(__pa_symbol(level3_user_pgt) | _PAGE_TABLE);
2416
2417+ /* Do an early initialization of the fixmap area. */
2418+ addr = __fix_to_virt(FIX_EARLYCON_MEM_BASE);
2419+ level3_kernel_pgt[pud_index(addr)] =
2420+ __pud(__pa_symbol(level2_fixmap_pgt) | _PAGE_TABLE);
2421+ level2_fixmap_pgt[pmd_index(addr)] =
2422+ __pmd(__pa_symbol(level1_fixmap_pgt) | _PAGE_TABLE);
2423+
2424 early_make_page_readonly(init_level4_pgt,
2425 XENFEAT_writable_page_tables);
2426 early_make_page_readonly(__user_pgd(init_level4_pgt),
2427@@ -604,6 +618,10 @@ void __init xen_init_pt(void)
2428 XENFEAT_writable_page_tables);
2429 early_make_page_readonly(level2_kernel_pgt,
2430 XENFEAT_writable_page_tables);
2431+ early_make_page_readonly(level2_fixmap_pgt,
2432+ XENFEAT_writable_page_tables);
2433+ early_make_page_readonly(level1_fixmap_pgt,
2434+ XENFEAT_writable_page_tables);
2435
2436 if (!xen_feature(XENFEAT_writable_page_tables)) {
2437 xen_pgd_pin(__pa_symbol(init_level4_pgt));
2438@@ -807,7 +825,7 @@ void __init paging_init(void)
2439 sparse_init();
2440 free_area_init_nodes(max_zone_pfns);
2441
2442- init_mm.context.pinned = 1;
2443+ SetPagePinned(virt_to_page(init_mm.pgd));
2444 }
2445 #endif
2446
2447@@ -1118,41 +1136,6 @@ int kern_addr_valid(unsigned long addr)
2448 return pfn_valid(pte_pfn(*pte));
2449 }
2450
2451-#ifdef CONFIG_SYSCTL
2452-#include <linux/sysctl.h>
2453-
2454-extern int exception_trace, page_fault_trace;
2455-
2456-static ctl_table debug_table2[] = {
2457- {
2458- .ctl_name = 99,
2459- .procname = "exception-trace",
2460- .data = &exception_trace,
2461- .maxlen = sizeof(int),
2462- .mode = 0644,
2463- .proc_handler = proc_dointvec
2464- },
2465- {}
2466-};
2467-
2468-static ctl_table debug_root_table2[] = {
2469- {
2470- .ctl_name = CTL_DEBUG,
2471- .procname = "debug",
2472- .mode = 0555,
2473- .child = debug_table2
2474- },
2475- {}
2476-};
2477-
2478-static __init int x8664_sysctl_init(void)
2479-{
2480- register_sysctl_table(debug_root_table2);
2481- return 0;
2482-}
2483-__initcall(x8664_sysctl_init);
2484-#endif
2485-
2486 /* A pseudo VMA to allow ptrace access for the vsyscall page. This only
2487 covers the 64bit vsyscall page now. 32bit has a real VMA now and does
2488 not need special handling anymore. */
2489@@ -1191,9 +1174,18 @@ int in_gate_area_no_task(unsigned long a
2490 }
2491
2492 #ifndef CONFIG_XEN
2493-void *alloc_bootmem_high_node(pg_data_t *pgdat, unsigned long size)
2494+void * __init alloc_bootmem_high_node(pg_data_t *pgdat, unsigned long size)
2495 {
2496 return __alloc_bootmem_core(pgdat->bdata, size,
2497 SMP_CACHE_BYTES, (4UL*1024*1024*1024), 0);
2498 }
2499 #endif
2500+
2501+const char *arch_vma_name(struct vm_area_struct *vma)
2502+{
2503+ if (vma->vm_mm && vma->vm_start == (long)vma->vm_mm->context.vdso)
2504+ return "[vdso]";
2505+ if (vma == &gate_vma)
2506+ return "[vsyscall]";
2507+ return NULL;
2508+}
2509--- sle11-2009-05-14.orig/arch/x86/mm/pageattr_64-xen.c 2008-12-15 11:27:22.000000000 +0100
2510+++ sle11-2009-05-14/arch/x86/mm/pageattr_64-xen.c 2009-02-16 16:17:21.000000000 +0100
2511@@ -171,7 +171,7 @@ void mm_pin(struct mm_struct *mm)
2512 mm_walk(mm, PAGE_KERNEL_RO);
2513 xen_pgd_pin(__pa(mm->pgd)); /* kernel */
2514 xen_pgd_pin(__pa(__user_pgd(mm->pgd))); /* user */
2515- mm->context.pinned = 1;
2516+ SetPagePinned(virt_to_page(mm->pgd));
2517 spin_lock(&mm_unpinned_lock);
2518 list_del(&mm->context.unpinned);
2519 spin_unlock(&mm_unpinned_lock);
2520@@ -189,7 +189,7 @@ void mm_unpin(struct mm_struct *mm)
2521 xen_pgd_unpin(__pa(mm->pgd));
2522 xen_pgd_unpin(__pa(__user_pgd(mm->pgd)));
2523 mm_walk(mm, PAGE_KERNEL);
2524- mm->context.pinned = 0;
2525+ ClearPagePinned(virt_to_page(mm->pgd));
2526 spin_lock(&mm_unpinned_lock);
2527 list_add(&mm->context.unpinned, &mm_unpinned);
2528 spin_unlock(&mm_unpinned_lock);
2529@@ -217,7 +217,7 @@ void mm_pin_all(void)
2530
2531 void arch_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
2532 {
2533- if (!mm->context.pinned)
2534+ if (!PagePinned(virt_to_page(mm->pgd)))
2535 mm_pin(mm);
2536 }
2537
2538@@ -243,8 +243,9 @@ void arch_exit_mmap(struct mm_struct *mm
2539
2540 task_unlock(tsk);
2541
2542- if ( mm->context.pinned && (atomic_read(&mm->mm_count) == 1) &&
2543- !mm->context.has_foreign_mappings )
2544+ if (PagePinned(virt_to_page(mm->pgd))
2545+ && (atomic_read(&mm->mm_count) == 1)
2546+ && !mm->context.has_foreign_mappings)
2547 mm_unpin(mm);
2548 }
2549
2550@@ -343,14 +344,13 @@ static void flush_kernel_map(void *arg)
2551 struct page *pg;
2552
2553 /* When clflush is available always use it because it is
2554- much cheaper than WBINVD. Disable clflush for now because
2555- the high level code is not ready yet */
2556+ much cheaper than WBINVD. */
2557+ /* clflush is still broken. Disable for now. */
2558 if (1 || !cpu_has_clflush)
2559 asm volatile("wbinvd" ::: "memory");
2560 else list_for_each_entry(pg, l, lru) {
2561 void *adr = page_address(pg);
2562- if (cpu_has_clflush)
2563- cache_flush_page(adr);
2564+ cache_flush_page(adr);
2565 }
2566 __flush_tlb_all();
2567 }
2568@@ -364,7 +364,8 @@ static LIST_HEAD(deferred_pages); /* pro
2569
2570 static inline void save_page(struct page *fpage)
2571 {
2572- list_add(&fpage->lru, &deferred_pages);
2573+ if (!test_and_set_bit(PG_arch_1, &fpage->flags))
2574+ list_add(&fpage->lru, &deferred_pages);
2575 }
2576
2577 /*
2578@@ -398,9 +399,12 @@ __change_page_attr(unsigned long address
2579 pte_t *kpte;
2580 struct page *kpte_page;
2581 pgprot_t ref_prot2;
2582+
2583 kpte = lookup_address(address);
2584 if (!kpte) return 0;
2585 kpte_page = virt_to_page(((unsigned long)kpte) & PAGE_MASK);
2586+ BUG_ON(PageLRU(kpte_page));
2587+ BUG_ON(PageCompound(kpte_page));
2588 if (pgprot_val(prot) != pgprot_val(ref_prot)) {
2589 if (!pte_huge(*kpte)) {
2590 set_pte(kpte, pfn_pte(pfn, prot));
2591@@ -439,10 +443,9 @@ __change_page_attr(unsigned long address
2592 return 0;
2593 #endif
2594
2595- if (page_private(kpte_page) == 0) {
2596- save_page(kpte_page);
2597+ save_page(kpte_page);
2598+ if (page_private(kpte_page) == 0)
2599 revert_page(address, ref_prot);
2600- }
2601 return 0;
2602 }
2603
2604@@ -514,6 +517,10 @@ void global_flush_tlb(void)
2605 flush_map(&l);
2606
2607 list_for_each_entry_safe(pg, next, &l, lru) {
2608+ list_del(&pg->lru);
2609+ clear_bit(PG_arch_1, &pg->flags);
2610+ if (page_private(pg) != 0)
2611+ continue;
2612 ClearPagePrivate(pg);
2613 __free_page(pg);
2614 }
2615--- sle11-2009-05-14.orig/arch/x86/vdso/vdso32/note.S 2009-05-14 10:56:29.000000000 +0200
2616+++ sle11-2009-05-14/arch/x86/vdso/vdso32/note.S 2009-02-16 16:17:21.000000000 +0100
2617@@ -13,7 +13,7 @@ ELFNOTE_START(Linux, 0, "a")
2618 .long LINUX_VERSION_CODE
2619 ELFNOTE_END
2620
2621-#ifdef CONFIG_XEN
2622+#if defined(CONFIG_X86_XEN) || defined(CONFIG_PARAVIRT_XEN)
2623 /*
2624 * Add a special note telling glibc's dynamic linker a fake hardware
2625 * flavor that it will use to choose the search path for libraries in the
2626@@ -37,8 +37,12 @@ ELFNOTE_END
2627
2628 ELFNOTE_START(GNU, 2, "a")
2629 .long 1 /* ncaps */
2630+#ifdef CONFIG_PARAVIRT_XEN
2631 VDSO32_NOTE_MASK: /* Symbol used by arch/x86/xen/setup.c */
2632 .long 0 /* mask */
2633+#else
2634+ .long 1 << VDSO_NOTE_NONEGSEG_BIT /* mask */
2635+#endif
2636 .byte VDSO_NOTE_NONEGSEG_BIT; .asciz "nosegneg" /* bit, name */
2637 ELFNOTE_END
2638 #endif
2639--- sle11-2009-05-14.orig/drivers/Makefile 2009-02-16 15:58:14.000000000 +0100
2640+++ sle11-2009-05-14/drivers/Makefile 2009-02-16 16:17:21.000000000 +0100
2641@@ -16,7 +16,7 @@ obj-$(CONFIG_ACPI) += acpi/
2642 obj-$(CONFIG_PNP) += pnp/
2643 obj-$(CONFIG_ARM_AMBA) += amba/
2644
2645-obj-$(CONFIG_XEN) += xen/
2646+obj-$(CONFIG_PARAVIRT_XEN) += xen/
2647
2648 # char/ comes before serial/ etc so that the VT console is the boot-time
2649 # default.
2650--- sle11-2009-05-14.orig/drivers/block/Kconfig 2009-05-14 10:56:29.000000000 +0200
2651+++ sle11-2009-05-14/drivers/block/Kconfig 2009-02-16 16:17:21.000000000 +0100
2652@@ -423,9 +423,9 @@ config XILINX_SYSACE
2653 help
2654 Include support for the Xilinx SystemACE CompactFlash interface
2655
2656-config XEN_BLKDEV_FRONTEND
2657+config XEN_BLKFRONT
2658 tristate "Xen virtual block device support"
2659- depends on XEN
2660+ depends on PARAVIRT_XEN
2661 default y
2662 help
2663 This driver implements the front-end of the Xen virtual
2664--- sle11-2009-05-14.orig/drivers/block/Makefile 2009-05-14 10:56:29.000000000 +0200
2665+++ sle11-2009-05-14/drivers/block/Makefile 2009-02-16 16:17:21.000000000 +0100
2666@@ -31,5 +31,5 @@ obj-$(CONFIG_BLK_DEV_SX8) += sx8.o
2667 obj-$(CONFIG_BLK_DEV_UB) += ub.o
2668 obj-$(CONFIG_BLK_DEV_HD) += hd.o
2669
2670-obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += xen-blkfront.o
2671+obj-$(CONFIG_XEN_BLKFRONT) += xen-blkfront.o
2672 obj-$(CONFIG_CIPHER_TWOFISH) += loop_fish2.o
2673--- sle11-2009-05-14.orig/drivers/block/xen-blkfront.c 2009-05-14 10:56:29.000000000 +0200
2674+++ sle11-2009-05-14/drivers/block/xen-blkfront.c 2009-02-16 16:17:21.000000000 +0100
2675@@ -1021,7 +1021,6 @@ static struct xenbus_device_id blkfront_
2676
2677 static struct xenbus_driver blkfront = {
2678 .name = "vbd",
2679- .owner = THIS_MODULE,
2680 .ids = blkfront_ids,
2681 .probe = blkfront_probe,
2682 .remove = blkfront_remove,
2683--- sle11-2009-05-14.orig/drivers/char/Kconfig 2009-02-16 15:58:02.000000000 +0100
2684+++ sle11-2009-05-14/drivers/char/Kconfig 2009-02-16 16:17:21.000000000 +0100
2685@@ -624,7 +624,7 @@ config HVC_BEAT
2686
2687 config HVC_XEN
2688 bool "Xen Hypervisor Console support"
2689- depends on XEN
2690+ depends on PARAVIRT_XEN
2691 select HVC_DRIVER
2692 select HVC_IRQ
2693 default y
2694--- sle11-2009-05-14.orig/drivers/net/Kconfig 2009-05-14 10:56:29.000000000 +0200
2695+++ sle11-2009-05-14/drivers/net/Kconfig 2009-02-16 16:17:21.000000000 +0100
2696@@ -2545,9 +2545,9 @@ source "drivers/atm/Kconfig"
2697
2698 source "drivers/s390/net/Kconfig"
2699
2700-config XEN_NETDEV_FRONTEND
2701+config XEN_NETFRONT
2702 tristate "Xen network device frontend driver"
2703- depends on XEN
2704+ depends on PARAVIRT_XEN
2705 default y
2706 help
2707 The network device frontend driver allows the kernel to
2708--- sle11-2009-05-14.orig/drivers/net/Makefile 2009-05-14 10:56:29.000000000 +0200
2709+++ sle11-2009-05-14/drivers/net/Makefile 2009-02-16 16:17:21.000000000 +0100
2710@@ -142,7 +142,7 @@ obj-$(CONFIG_PPPOL2TP) += pppox.o pppol2
2711 obj-$(CONFIG_SLIP) += slip.o
2712 obj-$(CONFIG_SLHC) += slhc.o
2713
2714-obj-$(CONFIG_XEN_NETDEV_FRONTEND) += xen-netfront.o
2715+obj-$(CONFIG_XEN_NETFRONT) += xen-netfront.o
2716
2717 obj-$(CONFIG_DUMMY) += dummy.o
2718 obj-$(CONFIG_IFB) += ifb.o
2719--- sle11-2009-05-14.orig/drivers/net/xen-netfront.c 2009-05-14 10:56:29.000000000 +0200
2720+++ sle11-2009-05-14/drivers/net/xen-netfront.c 2009-02-16 16:17:21.000000000 +0100
2721@@ -36,8 +36,6 @@
2722 #include <linux/skbuff.h>
2723 #include <linux/ethtool.h>
2724 #include <linux/if_ether.h>
2725-#include <linux/tcp.h>
2726-#include <linux/udp.h>
2727 #include <linux/moduleparam.h>
2728 #include <linux/mm.h>
2729 #include <net/ip.h>
2730@@ -765,45 +763,6 @@ static RING_IDX xennet_fill_frags(struct
2731 return cons;
2732 }
2733
2734-static int skb_checksum_setup(struct sk_buff *skb)
2735-{
2736- struct iphdr *iph;
2737- unsigned char *th;
2738- int err = -EPROTO;
2739-
2740- if (skb->protocol != htons(ETH_P_IP))
2741- goto out;
2742-
2743- iph = (void *)skb->data;
2744- th = skb->data + 4 * iph->ihl;
2745- if (th >= skb_tail_pointer(skb))
2746- goto out;
2747-
2748- skb->csum_start = th - skb->head;
2749- switch (iph->protocol) {
2750- case IPPROTO_TCP:
2751- skb->csum_offset = offsetof(struct tcphdr, check);
2752- break;
2753- case IPPROTO_UDP:
2754- skb->csum_offset = offsetof(struct udphdr, check);
2755- break;
2756- default:
2757- if (net_ratelimit())
2758- printk(KERN_ERR "Attempting to checksum a non-"
2759- "TCP/UDP packet, dropping a protocol"
2760- " %d packet", iph->protocol);
2761- goto out;
2762- }
2763-
2764- if ((th + skb->csum_offset + 2) > skb_tail_pointer(skb))
2765- goto out;
2766-
2767- err = 0;
2768-
2769-out:
2770- return err;
2771-}
2772-
2773 static int handle_incoming_queue(struct net_device *dev,
2774 struct sk_buff_head *rxq)
2775 {
2776@@ -1784,7 +1743,6 @@ static int __devexit xennet_remove(struc
2777
2778 static struct xenbus_driver netfront = {
2779 .name = "vif",
2780- .owner = THIS_MODULE,
2781 .ids = netfront_ids,
2782 .probe = netfront_probe,
2783 .remove = __devexit_p(xennet_remove),
2784--- sle11-2009-05-14.orig/drivers/xen/Makefile 2008-11-25 12:35:56.000000000 +0100
2785+++ sle11-2009-05-14/drivers/xen/Makefile 2009-02-16 16:17:21.000000000 +0100
2786@@ -1,10 +1,12 @@
2787-obj-y += core/
2788-obj-y += console/
2789-obj-y += evtchn/
2790-obj-y += xenbus/
2791-obj-y += char/
2792+obj-$(CONFIG_PARAVIRT_XEN) += grant-table.o
2793
2794-obj-y += util.o
2795+obj-$(CONFIG_XEN) += core/
2796+obj-$(CONFIG_XEN) += console/
2797+obj-$(CONFIG_XEN) += evtchn/
2798+obj-y += xenbus/
2799+obj-$(CONFIG_XEN) += char/
2800+
2801+obj-$(CONFIG_XEN) += util.o
2802 obj-$(CONFIG_XEN_BALLOON) += balloon/
2803 obj-$(CONFIG_XEN_BLKDEV_BACKEND) += blkback/
2804 obj-$(CONFIG_XEN_BLKDEV_TAP) += blktap/
2805--- sle11-2009-05-14.orig/drivers/xen/blkback/blkback.c 2008-12-15 11:26:44.000000000 +0100
2806+++ sle11-2009-05-14/drivers/xen/blkback/blkback.c 2009-02-16 16:17:21.000000000 +0100
2807@@ -154,7 +154,7 @@ static void unplug_queue(blkif_t *blkif)
2808
2809 static void plug_queue(blkif_t *blkif, struct block_device *bdev)
2810 {
2811- request_queue_t *q = bdev_get_queue(bdev);
2812+ struct request_queue *q = bdev_get_queue(bdev);
2813
2814 if (q == blkif->plug)
2815 return;
2816--- sle11-2009-05-14.orig/drivers/xen/blkback/common.h 2009-03-04 11:28:34.000000000 +0100
2817+++ sle11-2009-05-14/drivers/xen/blkback/common.h 2009-02-16 16:17:21.000000000 +0100
2818@@ -79,7 +79,7 @@ typedef struct blkif_st {
2819 wait_queue_head_t wq;
2820 struct task_struct *xenblkd;
2821 unsigned int waiting_reqs;
2822- request_queue_t *plug;
2823+ struct request_queue *plug;
2824
2825 /* statistics */
2826 unsigned long st_print;
2827--- sle11-2009-05-14.orig/drivers/xen/blkback/interface.c 2008-12-15 11:26:44.000000000 +0100
2828+++ sle11-2009-05-14/drivers/xen/blkback/interface.c 2009-02-16 16:17:21.000000000 +0100
2829@@ -177,5 +177,5 @@ void blkif_free(blkif_t *blkif)
2830 void __init blkif_interface_init(void)
2831 {
2832 blkif_cachep = kmem_cache_create("blkif_cache", sizeof(blkif_t),
2833- 0, 0, NULL, NULL);
2834+ 0, 0, NULL);
2835 }
2836--- sle11-2009-05-14.orig/drivers/xen/blkfront/blkfront.c 2009-03-24 10:11:24.000000000 +0100
2837+++ sle11-2009-05-14/drivers/xen/blkfront/blkfront.c 2009-03-24 10:12:03.000000000 +0100
2838@@ -662,7 +662,7 @@ static int blkif_queue_request(struct re
2839 * do_blkif_request
2840 * read a block; request is in a request queue
2841 */
2842-void do_blkif_request(request_queue_t *rq)
2843+void do_blkif_request(struct request_queue *rq)
2844 {
2845 struct blkfront_info *info = NULL;
2846 struct request *req;
2847--- sle11-2009-05-14.orig/drivers/xen/blkfront/block.h 2009-03-05 15:42:00.000000000 +0100
2848+++ sle11-2009-05-14/drivers/xen/blkfront/block.h 2009-03-24 10:11:58.000000000 +0100
2849@@ -106,7 +106,7 @@ struct blkfront_info
2850 struct scatterlist sg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
2851 unsigned int irq;
2852 struct xlbd_major_info *mi;
2853- request_queue_t *rq;
2854+ struct request_queue *rq;
2855 struct work_struct work;
2856 struct gnttab_free_callback callback;
2857 struct blk_shadow shadow[BLK_RING_SIZE];
2858@@ -130,7 +130,7 @@ extern int blkif_ioctl(struct inode *ino
2859 extern int blkif_getgeo(struct block_device *, struct hd_geometry *);
2860 extern int blkif_check(dev_t dev);
2861 extern int blkif_revalidate(dev_t dev);
2862-extern void do_blkif_request (request_queue_t *rq);
2863+extern void do_blkif_request (struct request_queue *rq);
2864
2865 /* Virtual block-device subsystem. */
2866 /* Note that xlvbd_add doesn't call add_disk for you: you're expected
2867--- sle11-2009-05-14.orig/drivers/xen/blkfront/vbd.c 2009-05-14 10:56:29.000000000 +0200
2868+++ sle11-2009-05-14/drivers/xen/blkfront/vbd.c 2009-02-16 16:17:21.000000000 +0100
2869@@ -211,7 +211,7 @@ xlbd_put_major_info(struct xlbd_major_in
2870 static int
2871 xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
2872 {
2873- request_queue_t *rq;
2874+ struct request_queue *rq;
2875
2876 rq = blk_init_queue(do_blkif_request, &blkif_io_lock);
2877 if (rq == NULL)
2878--- sle11-2009-05-14.orig/drivers/xen/blktap/common.h 2009-03-04 11:28:34.000000000 +0100
2879+++ sle11-2009-05-14/drivers/xen/blktap/common.h 2009-02-16 16:17:21.000000000 +0100
2880@@ -68,7 +68,7 @@ typedef struct blkif_st {
2881 wait_queue_head_t wq;
2882 struct task_struct *xenblkd;
2883 unsigned int waiting_reqs;
2884- request_queue_t *plug;
2885+ struct request_queue *plug;
2886
2887 /* statistics */
2888 unsigned long st_print;
2889--- sle11-2009-05-14.orig/drivers/xen/blktap/interface.c 2008-12-15 11:26:44.000000000 +0100
2890+++ sle11-2009-05-14/drivers/xen/blktap/interface.c 2009-02-16 16:17:21.000000000 +0100
2891@@ -177,5 +177,5 @@ void tap_blkif_kmem_cache_free(blkif_t *
2892 void __init tap_blkif_interface_init(void)
2893 {
2894 blkif_cachep = kmem_cache_create("blktapif_cache", sizeof(blkif_t),
2895- 0, 0, NULL, NULL);
2896+ 0, 0, NULL);
2897 }
2898--- sle11-2009-05-14.orig/drivers/xen/core/features.c 2009-05-14 10:56:29.000000000 +0200
2899+++ sle11-2009-05-14/drivers/xen/core/features.c 2009-02-16 16:17:21.000000000 +0100
2900@@ -19,7 +19,7 @@ u8 xen_features[XENFEAT_NR_SUBMAPS * 32]
2901 /* Not a GPL symbol: used in ubiquitous macros, so too restrictive. */
2902 EXPORT_SYMBOL(xen_features);
2903
2904-void setup_xen_features(void)
2905+void xen_setup_features(void)
2906 {
2907 xen_feature_info_t fi;
2908 int i, j;
2909--- sle11-2009-05-14.orig/drivers/xen/core/reboot.c 2008-12-15 11:26:44.000000000 +0100
2910+++ sle11-2009-05-14/drivers/xen/core/reboot.c 2009-02-16 16:17:21.000000000 +0100
2911@@ -4,6 +4,7 @@
2912 #include <linux/unistd.h>
2913 #include <linux/module.h>
2914 #include <linux/reboot.h>
2915+#include <linux/sched.h>
2916 #include <linux/sysrq.h>
2917 #include <asm/hypervisor.h>
2918 #include <xen/xenbus.h>
2919--- sle11-2009-05-14.orig/drivers/xen/scsiback/interface.c 2008-12-15 11:26:44.000000000 +0100
2920+++ sle11-2009-05-14/drivers/xen/scsiback/interface.c 2009-02-16 16:17:21.000000000 +0100
2921@@ -167,7 +167,7 @@ void scsiback_free(struct vscsibk_info *
2922 int __init scsiback_interface_init(void)
2923 {
2924 scsiback_cachep = kmem_cache_create("vscsiif_cache",
2925- sizeof(struct vscsibk_info), 0, 0, NULL, NULL);
2926+ sizeof(struct vscsibk_info), 0, 0, NULL);
2927 if (!scsiback_cachep) {
2928 printk(KERN_ERR "scsiback: can't init scsi cache\n");
2929 return -ENOMEM;
2930--- sle11-2009-05-14.orig/drivers/xen/scsifront/scsifront.c 2009-03-04 11:28:34.000000000 +0100
2931+++ sle11-2009-05-14/drivers/xen/scsifront/scsifront.c 2009-02-16 16:17:21.000000000 +0100
2932@@ -147,7 +147,7 @@ static void scsifront_cdb_cmd_done(struc
2933 add_id_to_freelist(info, id);
2934
2935 sc->result = ring_res->rslt;
2936- sc->resid = ring_res->residual_len;
2937+ scsi_set_resid(sc, ring_res->residual_len);
2938
2939 if (ring_res->sense_len > VSCSIIF_SENSE_BUFFERSIZE)
2940 sense_len = VSCSIIF_SENSE_BUFFERSIZE;
2941--- sle11-2009-05-14.orig/drivers/xen/tpmback/interface.c 2009-03-04 11:25:55.000000000 +0100
2942+++ sle11-2009-05-14/drivers/xen/tpmback/interface.c 2009-02-16 16:17:21.000000000 +0100
2943@@ -12,6 +12,7 @@
2944 */
2945
2946 #include "common.h"
2947+#include <linux/err.h>
2948 #include <xen/balloon.h>
2949 #include <xen/gnttab.h>
2950
2951@@ -159,7 +160,7 @@ void tpmif_disconnect_complete(tpmif_t *
2952 int __init tpmif_interface_init(void)
2953 {
2954 tpmif_cachep = kmem_cache_create("tpmif_cache", sizeof (tpmif_t),
2955- 0, 0, NULL, NULL);
2956+ 0, 0, NULL);
2957 return tpmif_cachep ? 0 : -ENOMEM;
2958 }
2959
2960--- sle11-2009-05-14.orig/drivers/xen/util.c 2009-05-14 10:56:29.000000000 +0200
2961+++ sle11-2009-05-14/drivers/xen/util.c 2009-02-16 16:17:21.000000000 +0100
2962@@ -1,8 +1,5 @@
2963-#include <linux/mm.h>
2964+#include <linux/err.h>
2965 #include <linux/module.h>
2966-#include <linux/slab.h>
2967-#include <linux/vmalloc.h>
2968-#include <asm/uaccess.h>
2969 #include <xen/driver_util.h>
2970
2971 struct class *get_xen_class(void)
2972@@ -21,45 +18,3 @@ struct class *get_xen_class(void)
2973 return xen_class;
2974 }
2975 EXPORT_SYMBOL_GPL(get_xen_class);
2976-
2977-#ifdef CONFIG_X86
2978-static int f(pte_t *pte, struct page *pmd_page, unsigned long addr, void *data)
2979-{
2980- /* apply_to_page_range() does all the hard work. */
2981- return 0;
2982-}
2983-
2984-struct vm_struct *alloc_vm_area(unsigned long size)
2985-{
2986- struct vm_struct *area;
2987-
2988- area = get_vm_area(size, VM_IOREMAP);
2989- if (area == NULL)
2990- return NULL;
2991-
2992- /*
2993- * This ensures that page tables are constructed for this region
2994- * of kernel virtual address space and mapped into init_mm.
2995- */
2996- if (apply_to_page_range(&init_mm, (unsigned long)area->addr,
2997- area->size, f, NULL)) {
2998- free_vm_area(area);
2999- return NULL;
3000- }
3001-
3002- /* Map page directories into every address space. */
3003- vmalloc_sync_all();
3004-
3005- return area;
3006-}
3007-EXPORT_SYMBOL_GPL(alloc_vm_area);
3008-
3009-void free_vm_area(struct vm_struct *area)
3010-{
3011- struct vm_struct *ret;
3012- ret = remove_vm_area(area->addr);
3013- BUG_ON(ret != area);
3014- kfree(area);
3015-}
3016-EXPORT_SYMBOL_GPL(free_vm_area);
3017-#endif /* CONFIG_X86 */
3018--- sle11-2009-05-14.orig/drivers/xen/xenbus/xenbus_client.c 2009-03-17 15:27:31.000000000 +0100
3019+++ sle11-2009-05-14/drivers/xen/xenbus/xenbus_client.c 2009-03-24 10:12:22.000000000 +0100
3020@@ -30,19 +30,26 @@
3021 * IN THE SOFTWARE.
3022 */
3023
3024+#if defined(CONFIG_XEN) || defined(MODULE)
3025 #include <linux/slab.h>
3026 #include <xen/evtchn.h>
3027 #include <xen/gnttab.h>
3028-#include <xen/xenbus.h>
3029 #include <xen/driver_util.h>
3030+#else
3031+#include <linux/types.h>
3032+#include <linux/vmalloc.h>
3033+#include <asm/xen/hypervisor.h>
3034+#include <xen/interface/xen.h>
3035+#include <xen/interface/event_channel.h>
3036+#include <xen/events.h>
3037+#include <xen/grant_table.h>
3038+#endif
3039+#include <xen/xenbus.h>
3040
3041 #ifdef HAVE_XEN_PLATFORM_COMPAT_H
3042 #include <xen/platform-compat.h>
3043 #endif
3044
3045-#define DPRINTK(fmt, args...) \
3046- pr_debug("xenbus_client (%s:%d) " fmt ".\n", __FUNCTION__, __LINE__, ##args)
3047-
3048 const char *xenbus_strstate(enum xenbus_state state)
3049 {
3050 static const char *const name[] = {
3051@@ -60,6 +67,20 @@ const char *xenbus_strstate(enum xenbus_
3052 }
3053 EXPORT_SYMBOL_GPL(xenbus_strstate);
3054
3055+/**
3056+ * xenbus_watch_path - register a watch
3057+ * @dev: xenbus device
3058+ * @path: path to watch
3059+ * @watch: watch to register
3060+ * @callback: callback to register
3061+ *
3062+ * Register a @watch on the given path, using the given xenbus_watch structure
3063+ * for storage, and the given @callback function as the callback. Return 0 on
3064+ * success, or -errno on error. On success, the given @path will be saved as
3065+ * @watch->node, and remains the caller's to free. On error, @watch->node will
3066+ * be NULL, the device will switch to %XenbusStateClosing, and the error will
3067+ * be saved in the store.
3068+ */
3069 int xenbus_watch_path(struct xenbus_device *dev, const char *path,
3070 struct xenbus_watch *watch,
3071 void (*callback)(struct xenbus_watch *,
3072@@ -83,6 +104,7 @@ int xenbus_watch_path(struct xenbus_devi
3073 EXPORT_SYMBOL_GPL(xenbus_watch_path);
3074
3075
3076+#if defined(CONFIG_XEN) || defined(MODULE)
3077 int xenbus_watch_path2(struct xenbus_device *dev, const char *path,
3078 const char *path2, struct xenbus_watch *watch,
3079 void (*callback)(struct xenbus_watch *,
3080@@ -101,8 +123,60 @@ int xenbus_watch_path2(struct xenbus_dev
3081 return err;
3082 }
3083 EXPORT_SYMBOL_GPL(xenbus_watch_path2);
3084+#else
3085+/**
3086+ * xenbus_watch_pathfmt - register a watch on a sprintf-formatted path
3087+ * @dev: xenbus device
3088+ * @watch: watch to register
3089+ * @callback: callback to register
3090+ * @pathfmt: format of path to watch
3091+ *
3092+ * Register a watch on the given @path, using the given xenbus_watch
3093+ * structure for storage, and the given @callback function as the callback.
3094+ * Return 0 on success, or -errno on error. On success, the watched path
3095+ * (@path/@path2) will be saved as @watch->node, and becomes the caller's to
3096+ * kfree(). On error, watch->node will be NULL, so the caller has nothing to
3097+ * free, the device will switch to %XenbusStateClosing, and the error will be
3098+ * saved in the store.
3099+ */
3100+int xenbus_watch_pathfmt(struct xenbus_device *dev,
3101+ struct xenbus_watch *watch,
3102+ void (*callback)(struct xenbus_watch *,
3103+ const char **, unsigned int),
3104+ const char *pathfmt, ...)
3105+{
3106+ int err;
3107+ va_list ap;
3108+ char *path;
3109
3110+ va_start(ap, pathfmt);
3111+ path = kvasprintf(GFP_KERNEL, pathfmt, ap);
3112+ va_end(ap);
3113
3114+ if (!path) {
3115+ xenbus_dev_fatal(dev, -ENOMEM, "allocating path for watch");
3116+ return -ENOMEM;
3117+ }
3118+ err = xenbus_watch_path(dev, path, watch, callback);
3119+
3120+ if (err)
3121+ kfree(path);
3122+ return err;
3123+}
3124+EXPORT_SYMBOL_GPL(xenbus_watch_pathfmt);
3125+#endif
3126+
3127+
3128+/**
3129+ * xenbus_switch_state
3130+ * @dev: xenbus device
3131+ * @xbt: transaction handle
3132+ * @state: new state
3133+ *
3134+ * Advertise in the store a change of the given driver to the given new_state.
3135+ * Return 0 on success, or -errno on error. On error, the device will switch
3136+ * to XenbusStateClosing, and the error will be saved in the store.
3137+ */
3138 int xenbus_switch_state(struct xenbus_device *dev, enum xenbus_state state)
3139 {
3140 /* We check whether the state is currently set to the given value, and
3141@@ -161,8 +235,8 @@ static char *error_path(struct xenbus_de
3142 }
3143
3144
3145-void _dev_error(struct xenbus_device *dev, int err, const char *fmt,
3146- va_list ap)
3147+static void _dev_error(struct xenbus_device *dev, int err,
3148+ const char *fmt, va_list ap)
3149 {
3150 int ret;
3151 unsigned int len;
3152@@ -183,14 +257,16 @@ void _dev_error(struct xenbus_device *de
3153 path_buffer = error_path(dev);
3154
3155 if (path_buffer == NULL) {
3156- printk("xenbus: failed to write error node for %s (%s)\n",
3157- dev->nodename, printf_buffer);
3158+ dev_err(&dev->dev,
3159+ "xenbus: failed to write error node for %s (%s)\n",
3160+ dev->nodename, printf_buffer);
3161 goto fail;
3162 }
3163
3164 if (xenbus_write(XBT_NIL, path_buffer, "error", printf_buffer) != 0) {
3165- printk("xenbus: failed to write error node for %s (%s)\n",
3166- dev->nodename, printf_buffer);
3167+ dev_err(&dev->dev,
3168+ "xenbus: failed to write error node for %s (%s)\n",
3169+ dev->nodename, printf_buffer);
3170 goto fail;
3171 }
3172
3173@@ -202,6 +278,15 @@ fail:
3174 }
3175
3176
3177+/**
3178+ * xenbus_dev_error
3179+ * @dev: xenbus device
3180+ * @err: error to report
3181+ * @fmt: error message format
3182+ *
3183+ * Report the given negative errno into the store, along with the given
3184+ * formatted message.
3185+ */
3186 void xenbus_dev_error(struct xenbus_device *dev, int err, const char *fmt,
3187 ...)
3188 {
3189@@ -214,6 +299,16 @@ void xenbus_dev_error(struct xenbus_devi
3190 EXPORT_SYMBOL_GPL(xenbus_dev_error);
3191
3192
3193+/**
3194+ * xenbus_dev_fatal
3195+ * @dev: xenbus device
3196+ * @err: error to report
3197+ * @fmt: error message format
3198+ *
3199+ * Equivalent to xenbus_dev_error(dev, err, fmt, args), followed by
3200+ * xenbus_switch_state(dev, NULL, XenbusStateClosing) to schedule an orderly
3201+ * closedown of this driver and its peer.
3202+ */
3203 void xenbus_dev_fatal(struct xenbus_device *dev, int err, const char *fmt,
3204 ...)
3205 {
3206@@ -228,6 +323,15 @@ void xenbus_dev_fatal(struct xenbus_devi
3207 EXPORT_SYMBOL_GPL(xenbus_dev_fatal);
3208
3209
3210+/**
3211+ * xenbus_grant_ring
3212+ * @dev: xenbus device
3213+ * @ring_mfn: mfn of ring to grant
3214+ *
3215+ * Grant access to the given @ring_mfn to the peer of the given device. Return
3216+ * 0 on success, or -errno on error. On error, the device will switch to
3217+ * XenbusStateClosing, and the error will be saved in the store.
3218+ */
3219 int xenbus_grant_ring(struct xenbus_device *dev, unsigned long ring_mfn)
3220 {
3221 int err = gnttab_grant_foreign_access(dev->otherend_id, ring_mfn, 0);
3222@@ -238,6 +342,12 @@ int xenbus_grant_ring(struct xenbus_devi
3223 EXPORT_SYMBOL_GPL(xenbus_grant_ring);
3224
3225
3226+/**
3227+ * Allocate an event channel for the given xenbus_device, assigning the newly
3228+ * created local port to *port. Return 0 on success, or -errno on error. On
3229+ * error, the device will switch to XenbusStateClosing, and the error will be
3230+ * saved in the store.
3231+ */
3232 int xenbus_alloc_evtchn(struct xenbus_device *dev, int *port)
3233 {
3234 struct evtchn_alloc_unbound alloc_unbound;
3235@@ -258,6 +368,38 @@ int xenbus_alloc_evtchn(struct xenbus_de
3236 EXPORT_SYMBOL_GPL(xenbus_alloc_evtchn);
3237
3238
3239+#if 0 /* !defined(CONFIG_XEN) && !defined(MODULE) */
3240+/**
3241+ * Bind to an existing interdomain event channel in another domain. Returns 0
3242+ * on success and stores the local port in *port. On error, returns -errno,
3243+ * switches the device to XenbusStateClosing, and saves the error in XenStore.
3244+ */
3245+int xenbus_bind_evtchn(struct xenbus_device *dev, int remote_port, int *port)
3246+{
3247+ struct evtchn_bind_interdomain bind_interdomain;
3248+ int err;
3249+
3250+ bind_interdomain.remote_dom = dev->otherend_id;
3251+ bind_interdomain.remote_port = remote_port;
3252+
3253+ err = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain,
3254+ &bind_interdomain);
3255+ if (err)
3256+ xenbus_dev_fatal(dev, err,
3257+ "binding to event channel %d from domain %d",
3258+ remote_port, dev->otherend_id);
3259+ else
3260+ *port = bind_interdomain.local_port;
3261+
3262+ return err;
3263+}
3264+EXPORT_SYMBOL_GPL(xenbus_bind_evtchn);
3265+#endif
3266+
3267+
3268+/**
3269+ * Free an existing event channel. Returns 0 on success or -errno on error.
3270+ */
3271 int xenbus_free_evtchn(struct xenbus_device *dev, int port)
3272 {
3273 struct evtchn_close close;
3274@@ -274,6 +416,191 @@ int xenbus_free_evtchn(struct xenbus_dev
3275 EXPORT_SYMBOL_GPL(xenbus_free_evtchn);
3276
3277
3278+#if 0 /* !defined(CONFIG_XEN) && !defined(MODULE) */
3279+/**
3280+ * xenbus_map_ring_valloc
3281+ * @dev: xenbus device
3282+ * @gnt_ref: grant reference
3283+ * @vaddr: pointer to address to be filled out by mapping
3284+ *
3285+ * Based on Rusty Russell's skeleton driver's map_page.
3286+ * Map a page of memory into this domain from another domain's grant table.
3287+ * xenbus_map_ring_valloc allocates a page of virtual address space, maps the
3288+ * page to that address, and sets *vaddr to that address.
3289+ * Returns 0 on success, and GNTST_* (see xen/include/interface/grant_table.h)
3290+ * or -ENOMEM on error. If an error is returned, device will switch to
3291+ * XenbusStateClosing and the error message will be saved in XenStore.
3292+ */
3293+int xenbus_map_ring_valloc(struct xenbus_device *dev, int gnt_ref, void **vaddr)
3294+{
3295+ struct gnttab_map_grant_ref op = {
3296+ .flags = GNTMAP_host_map,
3297+ .ref = gnt_ref,
3298+ .dom = dev->otherend_id,
3299+ };
3300+ struct vm_struct *area;
3301+
3302+ *vaddr = NULL;
3303+
3304+ area = alloc_vm_area(PAGE_SIZE);
3305+ if (!area)
3306+ return -ENOMEM;
3307+
3308+ op.host_addr = (unsigned long)area->addr;
3309+
3310+ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
3311+ BUG();
3312+
3313+ if (op.status != GNTST_okay) {
3314+ free_vm_area(area);
3315+ xenbus_dev_fatal(dev, op.status,
3316+ "mapping in shared page %d from domain %d",
3317+ gnt_ref, dev->otherend_id);
3318+ return op.status;
3319+ }
3320+
3321+ /* Stuff the handle in an unused field */
3322+ area->phys_addr = (unsigned long)op.handle;
3323+
3324+ *vaddr = area->addr;
3325+ return 0;
3326+}
3327+EXPORT_SYMBOL_GPL(xenbus_map_ring_valloc);
3328+
3329+
3330+/**
3331+ * xenbus_map_ring
3332+ * @dev: xenbus device
3333+ * @gnt_ref: grant reference
3334+ * @handle: pointer to grant handle to be filled
3335+ * @vaddr: address to be mapped to
3336+ *
3337+ * Map a page of memory into this domain from another domain's grant table.
3338+ * xenbus_map_ring does not allocate the virtual address space (you must do
3339+ * this yourself!). It only maps in the page to the specified address.
3340+ * Returns 0 on success, and GNTST_* (see xen/include/interface/grant_table.h)
3341+ * or -ENOMEM on error. If an error is returned, device will switch to
3342+ * XenbusStateClosing and the error message will be saved in XenStore.
3343+ */
3344+int xenbus_map_ring(struct xenbus_device *dev, int gnt_ref,
3345+ grant_handle_t *handle, void *vaddr)
3346+{
3347+ struct gnttab_map_grant_ref op = {
3348+ .host_addr = (unsigned long)vaddr,
3349+ .flags = GNTMAP_host_map,
3350+ .ref = gnt_ref,
3351+ .dom = dev->otherend_id,
3352+ };
3353+
3354+ if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1))
3355+ BUG();
3356+
3357+ if (op.status != GNTST_okay) {
3358+ xenbus_dev_fatal(dev, op.status,
3359+ "mapping in shared page %d from domain %d",
3360+ gnt_ref, dev->otherend_id);
3361+ } else
3362+ *handle = op.handle;
3363+
3364+ return op.status;
3365+}
3366+EXPORT_SYMBOL_GPL(xenbus_map_ring);
3367+
3368+
3369+/**
3370+ * xenbus_unmap_ring_vfree
3371+ * @dev: xenbus device
3372+ * @vaddr: addr to unmap
3373+ *
3374+ * Based on Rusty Russell's skeleton driver's unmap_page.
3375+ * Unmap a page of memory in this domain that was imported from another domain.
3376+ * Use xenbus_unmap_ring_vfree if you mapped in your memory with
3377+ * xenbus_map_ring_valloc (it will free the virtual address space).
3378+ * Returns 0 on success and returns GNTST_* on error
3379+ * (see xen/include/interface/grant_table.h).
3380+ */
3381+int xenbus_unmap_ring_vfree(struct xenbus_device *dev, void *vaddr)
3382+{
3383+ struct vm_struct *area;
3384+ struct gnttab_unmap_grant_ref op = {
3385+ .host_addr = (unsigned long)vaddr,
3386+ };
3387+
3388+ /* It'd be nice if linux/vmalloc.h provided a find_vm_area(void *addr)
3389+ * method so that we don't have to muck with vmalloc internals here.
3390+ * We could force the user to hang on to their struct vm_struct from
3391+ * xenbus_map_ring_valloc, but these 6 lines considerably simplify
3392+ * this API.
3393+ */
3394+ read_lock(&vmlist_lock);
3395+ for (area = vmlist; area != NULL; area = area->next) {
3396+ if (area->addr == vaddr)
3397+ break;
3398+ }
3399+ read_unlock(&vmlist_lock);
3400+
3401+ if (!area) {
3402+ xenbus_dev_error(dev, -ENOENT,
3403+ "can't find mapped virtual address %p", vaddr);
3404+ return GNTST_bad_virt_addr;
3405+ }
3406+
3407+ op.handle = (grant_handle_t)area->phys_addr;
3408+
3409+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
3410+ BUG();
3411+
3412+ if (op.status == GNTST_okay)
3413+ free_vm_area(area);
3414+ else
3415+ xenbus_dev_error(dev, op.status,
3416+ "unmapping page at handle %d error %d",
3417+ (int16_t)area->phys_addr, op.status);
3418+
3419+ return op.status;
3420+}
3421+EXPORT_SYMBOL_GPL(xenbus_unmap_ring_vfree);
3422+
3423+
3424+/**
3425+ * xenbus_unmap_ring
3426+ * @dev: xenbus device
3427+ * @handle: grant handle
3428+ * @vaddr: addr to unmap
3429+ *
3430+ * Unmap a page of memory in this domain that was imported from another domain.
3431+ * Returns 0 on success and returns GNTST_* on error
3432+ * (see xen/include/interface/grant_table.h).
3433+ */
3434+int xenbus_unmap_ring(struct xenbus_device *dev,
3435+ grant_handle_t handle, void *vaddr)
3436+{
3437+ struct gnttab_unmap_grant_ref op = {
3438+ .host_addr = (unsigned long)vaddr,
3439+ .handle = handle,
3440+ };
3441+
3442+ if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1))
3443+ BUG();
3444+
3445+ if (op.status != GNTST_okay)
3446+ xenbus_dev_error(dev, op.status,
3447+ "unmapping page at handle %d error %d",
3448+ handle, op.status);
3449+
3450+ return op.status;
3451+}
3452+EXPORT_SYMBOL_GPL(xenbus_unmap_ring);
3453+#endif
3454+
3455+
3456+/**
3457+ * xenbus_read_driver_state
3458+ * @path: path for driver
3459+ *
3460+ * Return the state of the driver rooted at the given store path, or
3461+ * XenbusStateUnknown if no state can be read.
3462+ */
3463 enum xenbus_state xenbus_read_driver_state(const char *path)
3464 {
3465 enum xenbus_state result;
3466--- sle11-2009-05-14.orig/drivers/xen/xenbus/xenbus_comms.c 2008-12-15 11:26:44.000000000 +0100
3467+++ sle11-2009-05-14/drivers/xen/xenbus/xenbus_comms.c 2009-02-16 16:17:21.000000000 +0100
3468@@ -34,12 +34,15 @@
3469 #include <linux/interrupt.h>
3470 #include <linux/sched.h>
3471 #include <linux/err.h>
3472-#include <linux/ptrace.h>
3473-#include <linux/workqueue.h>
3474-#include <xen/evtchn.h>
3475 #include <xen/xenbus.h>
3476-
3477+#if defined(CONFIG_XEN) || defined(MODULE)
3478+#include <xen/evtchn.h>
3479 #include <asm/hypervisor.h>
3480+#else
3481+#include <asm/xen/hypervisor.h>
3482+#include <xen/events.h>
3483+#include <xen/page.h>
3484+#endif
3485
3486 #include "xenbus_comms.h"
3487
3488@@ -91,6 +94,13 @@ static const void *get_input_chunk(XENST
3489 return buf + MASK_XENSTORE_IDX(cons);
3490 }
3491
3492+/**
3493+ * xb_write - low level write
3494+ * @data: buffer to send
3495+ * @len: length of buffer
3496+ *
3497+ * Returns 0 on success, error otherwise.
3498+ */
3499 int xb_write(const void *data, unsigned len)
3500 {
3501 struct xenstore_domain_interface *intf = xen_store_interface;
3502@@ -199,7 +209,9 @@ int xb_read(void *data, unsigned len)
3503 return 0;
3504 }
3505
3506-/* Set up interrupt handler off store event channel. */
3507+/**
3508+ * xb_init_comms - Set up interrupt handler off store event channel.
3509+ */
3510 int xb_init_comms(void)
3511 {
3512 struct xenstore_domain_interface *intf = xen_store_interface;
3513@@ -219,7 +231,11 @@ int xb_init_comms(void)
3514 if (xenbus_irq)
3515 unbind_from_irqhandler(xenbus_irq, &xb_waitq);
3516
3517+#if defined(CONFIG_XEN) || defined(MODULE)
3518 err = bind_caller_port_to_irqhandler(
3519+#else
3520+ err = bind_evtchn_to_irqhandler(
3521+#endif
3522 xen_store_evtchn, wake_waiting,
3523 0, "xenbus", &xb_waitq);
3524 if (err <= 0) {
3525--- sle11-2009-05-14.orig/drivers/xen/xenbus/xenbus_probe.c 2009-03-04 11:25:55.000000000 +0100
3526+++ sle11-2009-05-14/drivers/xen/xenbus/xenbus_probe.c 2009-02-16 16:17:21.000000000 +0100
3527@@ -43,12 +43,11 @@
3528 #include <linux/mm.h>
3529 #include <linux/notifier.h>
3530 #include <linux/mutex.h>
3531-#include <linux/module.h>
3532+#include <linux/io.h>
3533
3534-#include <asm/io.h>
3535 #include <asm/page.h>
3536-#include <asm/maddr.h>
3537 #include <asm/pgtable.h>
3538+#if defined(CONFIG_XEN) || defined(MODULE)
3539 #include <asm/hypervisor.h>
3540 #include <xen/xenbus.h>
3541 #include <xen/xen_proc.h>
3542@@ -57,6 +56,12 @@
3543 #ifdef MODULE
3544 #include <xen/hvm.h>
3545 #endif
3546+#else
3547+#include <asm/xen/hypervisor.h>
3548+#include <xen/xenbus.h>
3549+#include <xen/events.h>
3550+#include <xen/page.h>
3551+#endif
3552
3553 #include "xenbus_comms.h"
3554 #include "xenbus_probe.h"
3555@@ -168,7 +173,7 @@ static int read_backend_details(struct x
3556 return read_otherend_details(xendev, "backend-id", "backend");
3557 }
3558
3559-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16)
3560+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) && (defined(CONFIG_XEN) || defined(MODULE))
3561 static int xenbus_uevent_frontend(struct device *dev, char **envp,
3562 int num_envp, char *buffer, int buffer_size)
3563 {
3564@@ -207,12 +212,16 @@ static struct xen_bus_type xenbus_fronte
3565 .probe = xenbus_dev_probe,
3566 .remove = xenbus_dev_remove,
3567 .shutdown = xenbus_dev_shutdown,
3568+#if defined(CONFIG_XEN) || defined(MODULE)
3569 .uevent = xenbus_uevent_frontend,
3570 #endif
3571+#endif
3572 },
3573+#if defined(CONFIG_XEN) || defined(MODULE)
3574 .dev = {
3575 .bus_id = "xen",
3576 },
3577+#endif
3578 };
3579
3580 static void otherend_changed(struct xenbus_watch *watch,
3581@@ -228,14 +237,15 @@ static void otherend_changed(struct xenb
3582 if (!dev->otherend ||
3583 strncmp(dev->otherend, vec[XS_WATCH_PATH],
3584 strlen(dev->otherend))) {
3585- DPRINTK("Ignoring watch at %s", vec[XS_WATCH_PATH]);
3586+ dev_dbg(&dev->dev, "Ignoring watch at %s", vec[XS_WATCH_PATH]);
3587 return;
3588 }
3589
3590 state = xenbus_read_driver_state(dev->otherend);
3591
3592- DPRINTK("state is %d (%s), %s, %s", state, xenbus_strstate(state),
3593- dev->otherend_watch.node, vec[XS_WATCH_PATH]);
3594+ dev_dbg(&dev->dev, "state is %d (%s), %s, %s",
3595+ state, xenbus_strstate(state), dev->otherend_watch.node,
3596+ vec[XS_WATCH_PATH]);
3597
3598 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16)
3599 /*
3600@@ -271,8 +281,13 @@ static int talk_to_otherend(struct xenbu
3601
3602 static int watch_otherend(struct xenbus_device *dev)
3603 {
3604+#if defined(CONFIG_XEN) || defined(MODULE)
3605 return xenbus_watch_path2(dev, dev->otherend, "state",
3606 &dev->otherend_watch, otherend_changed);
3607+#else
3608+ return xenbus_watch_pathfmt(dev, &dev->otherend_watch, otherend_changed,
3609+ "%s/%s", dev->otherend, "state");
3610+#endif
3611 }
3612
3613
3614@@ -298,9 +313,9 @@ int xenbus_dev_probe(struct device *_dev
3615
3616 err = talk_to_otherend(dev);
3617 if (err) {
3618- printk(KERN_WARNING
3619- "xenbus_probe: talk_to_otherend on %s failed.\n",
3620- dev->nodename);
3621+ dev_warn(&dev->dev,
3622+ "xenbus_probe: talk_to_otherend on %s failed.\n",
3623+ dev->nodename);
3624 return err;
3625 }
3626
3627@@ -310,9 +325,9 @@ int xenbus_dev_probe(struct device *_dev
3628
3629 err = watch_otherend(dev);
3630 if (err) {
3631- printk(KERN_WARNING
3632- "xenbus_probe: watch_otherend on %s failed.\n",
3633- dev->nodename);
3634+ dev_warn(&dev->dev,
3635+ "xenbus_probe: watch_otherend on %s failed.\n",
3636+ dev->nodename);
3637 return err;
3638 }
3639
3640@@ -352,14 +367,15 @@ static void xenbus_dev_shutdown(struct d
3641
3642 get_device(&dev->dev);
3643 if (dev->state != XenbusStateConnected) {
3644- printk("%s: %s: %s != Connected, skipping\n", __FUNCTION__,
3645- dev->nodename, xenbus_strstate(dev->state));
3646+ dev_info(&dev->dev, "%s: %s: %s != Connected, skipping\n", __FUNCTION__,
3647+ dev->nodename, xenbus_strstate(dev->state));
3648 goto out;
3649 }
3650 xenbus_switch_state(dev, XenbusStateClosing);
3651 timeout = wait_for_completion_timeout(&dev->down, timeout);
3652 if (!timeout)
3653- printk("%s: %s timeout closing device\n", __FUNCTION__, dev->nodename);
3654+ dev_info(&dev->dev, "%s: %s timeout closing device\n",
3655+ __FUNCTION__, dev->nodename);
3656 out:
3657 put_device(&dev->dev);
3658 }
3659@@ -547,7 +563,9 @@ int xenbus_probe_node(struct xen_bus_typ
3660 xendev->devicetype = tmpstring;
3661 init_completion(&xendev->down);
3662
3663+#if defined(CONFIG_XEN) || defined(MODULE)
3664 xendev->dev.parent = &bus->dev;
3665+#endif
3666 xendev->dev.bus = &bus->bus;
3667 xendev->dev.release = xenbus_dev_release;
3668
3669@@ -562,15 +580,16 @@ int xenbus_probe_node(struct xen_bus_typ
3670
3671 err = device_create_file(&xendev->dev, &dev_attr_nodename);
3672 if (err)
3673- goto unregister;
3674+ goto fail_unregister;
3675+
3676 err = device_create_file(&xendev->dev, &dev_attr_devtype);
3677 if (err)
3678- goto unregister;
3679+ goto fail_remove_file;
3680
3681 return 0;
3682-unregister:
3683+fail_remove_file:
3684 device_remove_file(&xendev->dev, &dev_attr_nodename);
3685- device_remove_file(&xendev->dev, &dev_attr_devtype);
3686+fail_unregister:
3687 device_unregister(&xendev->dev);
3688 fail:
3689 kfree(xendev);
3690@@ -583,7 +602,8 @@ static int xenbus_probe_frontend(const c
3691 char *nodename;
3692 int err;
3693
3694- nodename = kasprintf(GFP_KERNEL, "%s/%s/%s", xenbus_frontend.root, type, name);
3695+ nodename = kasprintf(GFP_KERNEL, "%s/%s/%s",
3696+ xenbus_frontend.root, type, name);
3697 if (!nodename)
3698 return -ENOMEM;
3699
3700@@ -659,7 +679,7 @@ static int strsep_len(const char *str, c
3701 return (len == 0) ? i : -ERANGE;
3702 }
3703
3704-void dev_changed(const char *node, struct xen_bus_type *bus)
3705+void xenbus_dev_changed(const char *node, struct xen_bus_type *bus)
3706 {
3707 int exists, rootlen;
3708 struct xenbus_device *dev;
3709@@ -667,7 +687,7 @@ void dev_changed(const char *node, struc
3710 const char *p, *root;
3711
3712 if (bus->error || char_count(node, '/') < 2)
3713- return;
3714+ return;
3715
3716 exists = xenbus_exists(XBT_NIL, node, "");
3717 if (!exists) {
3718@@ -701,7 +721,7 @@ static void frontend_changed(struct xenb
3719 {
3720 DPRINTK("");
3721
3722- dev_changed(vec[XS_WATCH_PATH], &xenbus_frontend);
3723+ xenbus_dev_changed(vec[XS_WATCH_PATH], &xenbus_frontend);
3724 }
3725
3726 /* We watch for devices appearing and vanishing. */
3727@@ -904,10 +924,16 @@ static int xsd_port_read(char *page, cha
3728 }
3729 #endif
3730
3731-static int xenbus_probe_init(void)
3732+#ifndef MODULE
3733+static int __init xenbus_probe_init(void)
3734+#else
3735+static int __devinit xenbus_probe_init(void)
3736+#endif
3737 {
3738 int err = 0;
3739+#if defined(CONFIG_XEN) || defined(MODULE)
3740 unsigned long page = 0;
3741+#endif
3742
3743 DPRINTK("");
3744
3745@@ -926,6 +952,7 @@ static int xenbus_probe_init(void)
3746 * Domain0 doesn't have a store_evtchn or store_mfn yet.
3747 */
3748 if (is_initial_xendomain()) {
3749+#if defined(CONFIG_XEN) || defined(MODULE)
3750 struct evtchn_alloc_unbound alloc_unbound;
3751
3752 /* Allocate page. */
3753@@ -963,10 +990,13 @@ static int xenbus_probe_init(void)
3754 if (xsd_port_intf)
3755 xsd_port_intf->read_proc = xsd_port_read;
3756 #endif
3757+#else
3758+ /* dom0 not yet supported */
3759+#endif
3760 xen_store_interface = mfn_to_virt(xen_store_mfn);
3761 } else {
3762 xenstored_ready = 1;
3763-#ifdef CONFIG_XEN
3764+#ifndef MODULE
3765 xen_store_evtchn = xen_start_info->store_evtchn;
3766 xen_store_mfn = xen_start_info->store_mfn;
3767 xen_store_interface = mfn_to_virt(xen_store_mfn);
3768@@ -979,7 +1009,9 @@ static int xenbus_probe_init(void)
3769 }
3770
3771
3772+#if defined(CONFIG_XEN) || defined(MODULE)
3773 xenbus_dev_init();
3774+#endif
3775
3776 /* Initialize the interface to xenstore. */
3777 err = xs_init();
3778@@ -989,6 +1021,7 @@ static int xenbus_probe_init(void)
3779 goto err;
3780 }
3781
3782+#if defined(CONFIG_XEN) || defined(MODULE)
3783 /* Register ourselves with the kernel device subsystem */
3784 if (!xenbus_frontend.error) {
3785 xenbus_frontend.error = device_register(&xenbus_frontend.dev);
3786@@ -999,6 +1032,7 @@ static int xenbus_probe_init(void)
3787 xenbus_frontend.error);
3788 }
3789 }
3790+#endif
3791 xenbus_backend_device_register();
3792
3793 if (!is_initial_xendomain())
3794@@ -1007,8 +1041,10 @@ static int xenbus_probe_init(void)
3795 return 0;
3796
3797 err:
3798+#if defined(CONFIG_XEN) || defined(MODULE)
3799 if (page)
3800 free_page(page);
3801+#endif
3802
3803 /*
3804 * Do not unregister the xenbus front/backend buses here. The buses
3805@@ -1019,11 +1055,15 @@ static int xenbus_probe_init(void)
3806 return err;
3807 }
3808
3809-#ifdef CONFIG_XEN
3810+#ifndef MODULE
3811 postcore_initcall(xenbus_probe_init);
3812+#ifdef CONFIG_XEN
3813 MODULE_LICENSE("Dual BSD/GPL");
3814 #else
3815-int xenbus_init(void)
3816+MODULE_LICENSE("GPL");
3817+#endif
3818+#else
3819+int __devinit xenbus_init(void)
3820 {
3821 return xenbus_probe_init();
3822 }
3823--- sle11-2009-05-14.orig/drivers/xen/xenbus/xenbus_probe.h 2009-03-04 11:25:55.000000000 +0100
3824+++ sle11-2009-05-14/drivers/xen/xenbus/xenbus_probe.h 2009-02-16 16:17:21.000000000 +0100
3825@@ -56,7 +56,9 @@ struct xen_bus_type
3826 int (*get_bus_id)(char bus_id[BUS_ID_SIZE], const char *nodename);
3827 int (*probe)(const char *type, const char *dir);
3828 struct bus_type bus;
3829+#if defined(CONFIG_XEN) || defined(MODULE)
3830 struct device dev;
3831+#endif
3832 };
3833
3834 extern int xenbus_match(struct device *_dev, struct device_driver *_drv);
3835@@ -71,7 +73,7 @@ extern int xenbus_probe_node(struct xen_
3836 const char *nodename);
3837 extern int xenbus_probe_devices(struct xen_bus_type *bus);
3838
3839-extern void dev_changed(const char *node, struct xen_bus_type *bus);
3840+extern void xenbus_dev_changed(const char *node, struct xen_bus_type *bus);
3841
3842 #endif
3843
3844--- sle11-2009-05-14.orig/drivers/xen/xenbus/xenbus_probe_backend.c 2009-03-04 11:25:55.000000000 +0100
3845+++ sle11-2009-05-14/drivers/xen/xenbus/xenbus_probe_backend.c 2009-02-16 16:17:21.000000000 +0100
3846@@ -236,7 +236,7 @@ static void backend_changed(struct xenbu
3847 {
3848 DPRINTK("");
3849
3850- dev_changed(vec[XS_WATCH_PATH], &xenbus_backend);
3851+ xenbus_dev_changed(vec[XS_WATCH_PATH], &xenbus_backend);
3852 }
3853
3854 static struct xenbus_watch be_watch = {
3855--- sle11-2009-05-14.orig/drivers/xen/xenbus/xenbus_xs.c 2008-11-25 12:35:56.000000000 +0100
3856+++ sle11-2009-05-14/drivers/xen/xenbus/xenbus_xs.c 2009-02-16 16:17:21.000000000 +0100
3857@@ -221,7 +221,7 @@ static void *xs_talkv(struct xenbus_tran
3858 }
3859
3860 for (i = 0; i < num_vecs; i++) {
3861- err = xb_write(iovec[i].iov_base, iovec[i].iov_len);;
3862+ err = xb_write(iovec[i].iov_base, iovec[i].iov_len);
3863 if (err) {
3864 mutex_unlock(&xs_state.request_mutex);
3865 return ERR_PTR(err);
3866@@ -630,7 +630,9 @@ void unregister_xenbus_watch(struct xenb
3867 char token[sizeof(watch) * 2 + 1];
3868 int err;
3869
3870+#if defined(CONFIG_XEN) || defined(MODULE)
3871 BUG_ON(watch->flags & XBWF_new_thread);
3872+#endif
3873
3874 sprintf(token, "%lX", (long)watch);
3875
3876@@ -649,6 +651,11 @@ void unregister_xenbus_watch(struct xenb
3877
3878 up_read(&xs_state.watch_mutex);
3879
3880+ /* Make sure there are no callbacks running currently (unless
3881+ its us) */
3882+ if (current->pid != xenwatch_pid)
3883+ mutex_lock(&xenwatch_mutex);
3884+
3885 /* Cancel pending watch events. */
3886 spin_lock(&watch_events_lock);
3887 list_for_each_entry_safe(msg, tmp, &watch_events, list) {
3888@@ -660,11 +667,8 @@ void unregister_xenbus_watch(struct xenb
3889 }
3890 spin_unlock(&watch_events_lock);
3891
3892- /* Flush any currently-executing callback, unless we are it. :-) */
3893- if (current->pid != xenwatch_pid) {
3894- mutex_lock(&xenwatch_mutex);
3895+ if (current->pid != xenwatch_pid)
3896 mutex_unlock(&xenwatch_mutex);
3897- }
3898 }
3899 EXPORT_SYMBOL_GPL(unregister_xenbus_watch);
3900
3901@@ -702,6 +706,7 @@ void xs_suspend_cancel(void)
3902 up_write(&xs_state.transaction_mutex);
3903 }
3904
3905+#if defined(CONFIG_XEN) || defined(MODULE)
3906 static int xenwatch_handle_callback(void *data)
3907 {
3908 struct xs_stored_msg *msg = data;
3909@@ -719,6 +724,7 @@ static int xenwatch_handle_callback(void
3910
3911 return 0;
3912 }
3913+#endif
3914
3915 static int xenwatch_thread(void *unused)
3916 {
3917@@ -748,6 +754,7 @@ static int xenwatch_thread(void *unused)
3918
3919 msg = list_entry(ent, struct xs_stored_msg, list);
3920
3921+#if defined(CONFIG_XEN) || defined(MODULE)
3922 /*
3923 * Unlock the mutex before running an XBWF_new_thread
3924 * handler. kthread_run can block which can deadlock
3925@@ -764,6 +771,15 @@ static int xenwatch_thread(void *unused)
3926 xenwatch_handle_callback(msg);
3927 mutex_unlock(&xenwatch_mutex);
3928 }
3929+#else
3930+ msg->u.watch.handle->callback(
3931+ msg->u.watch.handle,
3932+ (const char **)msg->u.watch.vec,
3933+ msg->u.watch.vec_size);
3934+ mutex_unlock(&xenwatch_mutex);
3935+ kfree(msg->u.watch.vec);
3936+ kfree(msg);
3937+#endif
3938 }
3939
3940 return 0;
3941--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/fixmap_32.h 2008-12-15 11:27:22.000000000 +0100
3942+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/fixmap_32.h 2009-02-16 16:17:21.000000000 +0100
3943@@ -53,6 +53,8 @@ extern unsigned long __FIXADDR_TOP;
3944 enum fixed_addresses {
3945 FIX_HOLE,
3946 FIX_VDSO,
3947+ FIX_DBGP_BASE,
3948+ FIX_EARLYCON_MEM_BASE,
3949 #ifdef CONFIG_X86_LOCAL_APIC
3950 FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */
3951 #endif
3952--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/highmem.h 2008-12-15 11:27:22.000000000 +0100
3953+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/highmem.h 2009-02-16 16:17:21.000000000 +0100
3954@@ -76,8 +76,7 @@ struct page *kmap_atomic_to_page(void *p
3955
3956 #define kmap_atomic_pte(page, type) \
3957 kmap_atomic_prot(page, type, \
3958- test_bit(PG_pinned, &(page)->flags) \
3959- ? PAGE_KERNEL_RO : kmap_prot)
3960+ PagePinned(page) ? PAGE_KERNEL_RO : kmap_prot)
3961
3962 #define flush_cache_kmaps() do { } while (0)
3963
3964--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/maddr_32.h 2009-05-14 10:56:29.000000000 +0200
3965+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/maddr_32.h 2009-02-16 16:17:21.000000000 +0100
3966@@ -155,6 +155,7 @@ static inline paddr_t pte_machine_to_phy
3967
3968 #ifdef CONFIG_X86_PAE
3969 #define __pte_ma(x) ((pte_t) { (x), (maddr_t)(x) >> 32 } )
3970+extern unsigned long long __supported_pte_mask;
3971 static inline pte_t pfn_pte_ma(unsigned long page_nr, pgprot_t pgprot)
3972 {
3973 pte_t pte;
3974--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/mmu_context_32.h 2008-12-15 11:27:22.000000000 +0100
3975+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/mmu_context_32.h 2009-02-16 16:17:21.000000000 +0100
3976@@ -16,7 +16,7 @@ void mm_pin_all(void);
3977 static inline void xen_activate_mm(struct mm_struct *prev,
3978 struct mm_struct *next)
3979 {
3980- if (!test_bit(PG_pinned, &virt_to_page(next->pgd)->flags))
3981+ if (!PagePinned(virt_to_page(next->pgd)))
3982 mm_pin(next);
3983 }
3984
3985@@ -51,6 +51,8 @@ static inline void __prepare_arch_switch
3986 : : "r" (0) );
3987 }
3988
3989+void leave_mm(unsigned long cpu);
3990+
3991 static inline void switch_mm(struct mm_struct *prev,
3992 struct mm_struct *next,
3993 struct task_struct *tsk)
3994@@ -60,7 +62,7 @@ static inline void switch_mm(struct mm_s
3995
3996 if (likely(prev != next)) {
3997 BUG_ON(!xen_feature(XENFEAT_writable_page_tables) &&
3998- !test_bit(PG_pinned, &virt_to_page(next->pgd)->flags));
3999+ !PagePinned(virt_to_page(next->pgd)));
4000
4001 /* stop flush ipis for the previous mm */
4002 cpu_clear(cpu, prev->cpu_vm_mask);
4003--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pci_32.h 2009-05-14 10:56:29.000000000 +0200
4004+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pci_32.h 2009-02-16 16:17:21.000000000 +0100
4005@@ -3,6 +3,14 @@
4006
4007
4008 #ifdef __KERNEL__
4009+
4010+struct pci_sysdata {
4011+ int node; /* NUMA node */
4012+};
4013+
4014+/* scan a bus after allocating a pci_sysdata for it */
4015+extern struct pci_bus *pci_scan_bus_with_sysdata(int busno);
4016+
4017 #include <linux/mm.h> /* for struct page */
4018
4019 /* Can be used to override the logic in pci_scan_bus for skipping
4020@@ -81,48 +89,11 @@ struct pci_dev;
4021
4022 #endif
4023
4024-/* This is always fine. */
4025-#define pci_dac_dma_supported(pci_dev, mask) (1)
4026-
4027-static inline dma64_addr_t
4028-pci_dac_page_to_dma(struct pci_dev *pdev, struct page *page, unsigned long offset, int direction)
4029-{
4030- return ((dma64_addr_t) page_to_phys(page) +
4031- (dma64_addr_t) offset);
4032-}
4033-
4034-static inline struct page *
4035-pci_dac_dma_to_page(struct pci_dev *pdev, dma64_addr_t dma_addr)
4036-{
4037- return pfn_to_page(dma_addr >> PAGE_SHIFT);
4038-}
4039-
4040-static inline unsigned long
4041-pci_dac_dma_to_offset(struct pci_dev *pdev, dma64_addr_t dma_addr)
4042-{
4043- return (dma_addr & ~PAGE_MASK);
4044-}
4045-
4046-static inline void
4047-pci_dac_dma_sync_single_for_cpu(struct pci_dev *pdev, dma64_addr_t dma_addr, size_t len, int direction)
4048-{
4049-}
4050-
4051-static inline void
4052-pci_dac_dma_sync_single_for_device(struct pci_dev *pdev, dma64_addr_t dma_addr, size_t len, int direction)
4053-{
4054- flush_write_buffers();
4055-}
4056-
4057 #define HAVE_PCI_MMAP
4058 extern int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
4059 enum pci_mmap_state mmap_state, int write_combine);
4060
4061
4062-static inline void pcibios_add_platform_entries(struct pci_dev *dev)
4063-{
4064-}
4065-
4066 #ifdef CONFIG_PCI
4067 static inline void pci_dma_burst_advice(struct pci_dev *pdev,
4068 enum pci_dma_burst_strategy *strat,
4069--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pgalloc_32.h 2008-12-15 11:27:22.000000000 +0100
4070+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgalloc_32.h 2009-02-16 16:17:21.000000000 +0100
4071@@ -5,7 +5,7 @@
4072 #include <linux/mm.h> /* for struct page */
4073 #include <asm/io.h> /* for phys_to_virt and page_to_pseudophys */
4074
4075-#define paravirt_alloc_pt(pfn) do { } while (0)
4076+#define paravirt_alloc_pt(mm, pfn) do { } while (0)
4077 #define paravirt_alloc_pd(pfn) do { } while (0)
4078 #define paravirt_alloc_pd(pfn) do { } while (0)
4079 #define paravirt_alloc_pd_clone(pfn, clonepfn, start, count) do { } while (0)
4080@@ -14,15 +14,15 @@
4081
4082 #define pmd_populate_kernel(mm, pmd, pte) \
4083 do { \
4084- paravirt_alloc_pt(__pa(pte) >> PAGE_SHIFT); \
4085+ paravirt_alloc_pt(mm, __pa(pte) >> PAGE_SHIFT); \
4086 set_pmd(pmd, __pmd(_PAGE_TABLE + __pa(pte))); \
4087 } while (0)
4088
4089 #define pmd_populate(mm, pmd, pte) \
4090 do { \
4091 unsigned long pfn = page_to_pfn(pte); \
4092- paravirt_alloc_pt(pfn); \
4093- if (test_bit(PG_pinned, &virt_to_page((mm)->pgd)->flags)) { \
4094+ paravirt_alloc_pt(mm, pfn); \
4095+ if (PagePinned(virt_to_page((mm)->pgd))) { \
4096 if (!PageHighMem(pte)) \
4097 BUG_ON(HYPERVISOR_update_va_mapping( \
4098 (unsigned long)__va(pfn << PAGE_SHIFT), \
4099--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pgtable_32.h 2008-12-15 11:27:22.000000000 +0100
4100+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgtable_32.h 2009-02-16 16:17:21.000000000 +0100
4101@@ -26,9 +26,6 @@
4102 #include <linux/spinlock.h>
4103 #include <linux/sched.h>
4104
4105-/* Is this pagetable pinned? */
4106-#define PG_pinned PG_arch_1
4107-
4108 struct vm_area_struct;
4109
4110 /*
4111@@ -82,7 +79,7 @@ void paging_init(void);
4112 * area for the same reason. ;)
4113 */
4114 #define VMALLOC_OFFSET (8*1024*1024)
4115-#define VMALLOC_START (((unsigned long) high_memory + vmalloc_earlyreserve + \
4116+#define VMALLOC_START (((unsigned long) high_memory + \
4117 2*VMALLOC_OFFSET-1) & ~(VMALLOC_OFFSET-1))
4118 #ifdef CONFIG_HIGHMEM
4119 # define VMALLOC_END (PKMAP_BASE-2*PAGE_SIZE)
4120@@ -231,8 +228,6 @@ extern unsigned long pg0[];
4121 * The following only work if pte_present() is true.
4122 * Undefined behaviour if not..
4123 */
4124-static inline int pte_user(pte_t pte) { return (pte).pte_low & _PAGE_USER; }
4125-static inline int pte_read(pte_t pte) { return (pte).pte_low & _PAGE_USER; }
4126 static inline int pte_dirty(pte_t pte) { return (pte).pte_low & _PAGE_DIRTY; }
4127 static inline int pte_young(pte_t pte) { return (pte).pte_low & _PAGE_ACCESSED; }
4128 static inline int pte_write(pte_t pte) { return (pte).pte_low & _PAGE_RW; }
4129@@ -243,13 +238,9 @@ static inline int pte_huge(pte_t pte) {
4130 */
4131 static inline int pte_file(pte_t pte) { return (pte).pte_low & _PAGE_FILE; }
4132
4133-static inline pte_t pte_rdprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_USER; return pte; }
4134-static inline pte_t pte_exprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_USER; return pte; }
4135 static inline pte_t pte_mkclean(pte_t pte) { (pte).pte_low &= ~_PAGE_DIRTY; return pte; }
4136 static inline pte_t pte_mkold(pte_t pte) { (pte).pte_low &= ~_PAGE_ACCESSED; return pte; }
4137 static inline pte_t pte_wrprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_RW; return pte; }
4138-static inline pte_t pte_mkread(pte_t pte) { (pte).pte_low |= _PAGE_USER; return pte; }
4139-static inline pte_t pte_mkexec(pte_t pte) { (pte).pte_low |= _PAGE_USER; return pte; }
4140 static inline pte_t pte_mkdirty(pte_t pte) { (pte).pte_low |= _PAGE_DIRTY; return pte; }
4141 static inline pte_t pte_mkyoung(pte_t pte) { (pte).pte_low |= _PAGE_ACCESSED; return pte; }
4142 static inline pte_t pte_mkwrite(pte_t pte) { (pte).pte_low |= _PAGE_RW; return pte; }
4143@@ -295,22 +286,20 @@ static inline pte_t xen_local_ptep_get_a
4144 #define ptep_set_access_flags(vma, address, ptep, entry, dirty) \
4145 ({ \
4146 int __changed = !pte_same(*(ptep), entry); \
4147- if (__changed && (dirty)) \
4148- ptep_establish(vma, address, ptep, entry); \
4149+ if (__changed && (dirty)) { \
4150+ if ( likely((vma)->vm_mm == current->mm) ) { \
4151+ BUG_ON(HYPERVISOR_update_va_mapping(address, \
4152+ entry, \
4153+ (unsigned long)(vma)->vm_mm->cpu_vm_mask.bits| \
4154+ UVMF_INVLPG|UVMF_MULTI)); \
4155+ } else { \
4156+ xen_l1_entry_update(ptep, entry); \
4157+ flush_tlb_page(vma, address); \
4158+ } \
4159+ } \
4160 __changed; \
4161 })
4162
4163-#define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_DIRTY
4164-#define ptep_test_and_clear_dirty(vma, addr, ptep) ({ \
4165- int __ret = 0; \
4166- if (pte_dirty(*(ptep))) \
4167- __ret = test_and_clear_bit(_PAGE_BIT_DIRTY, \
4168- &(ptep)->pte_low); \
4169- if (__ret) \
4170- pte_update((vma)->vm_mm, addr, ptep); \
4171- __ret; \
4172-})
4173-
4174 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
4175 #define ptep_test_and_clear_young(vma, addr, ptep) ({ \
4176 int __ret = 0; \
4177@@ -322,44 +311,13 @@ static inline pte_t xen_local_ptep_get_a
4178 __ret; \
4179 })
4180
4181-/*
4182- * Rules for using ptep_establish: the pte MUST be a user pte, and
4183- * must be a present->present transition.
4184- */
4185-#define __HAVE_ARCH_PTEP_ESTABLISH
4186-#define ptep_establish(vma, address, ptep, pteval) \
4187-do { \
4188- if ( likely((vma)->vm_mm == current->mm) ) { \
4189- BUG_ON(HYPERVISOR_update_va_mapping(address, \
4190- pteval, \
4191- (unsigned long)(vma)->vm_mm->cpu_vm_mask.bits| \
4192- UVMF_INVLPG|UVMF_MULTI)); \
4193- } else { \
4194- xen_l1_entry_update(ptep, pteval); \
4195- flush_tlb_page(vma, address); \
4196- } \
4197-} while (0)
4198-
4199-#define __HAVE_ARCH_PTEP_CLEAR_DIRTY_FLUSH
4200-#define ptep_clear_flush_dirty(vma, address, ptep) \
4201-({ \
4202- pte_t __pte = *(ptep); \
4203- int __dirty = pte_dirty(__pte); \
4204- __pte = pte_mkclean(__pte); \
4205- if (test_bit(PG_pinned, &virt_to_page((vma)->vm_mm->pgd)->flags)) \
4206- (void)ptep_set_access_flags(vma, address, ptep, __pte, __dirty); \
4207- else if (__dirty) \
4208- (ptep)->pte_low = __pte.pte_low; \
4209- __dirty; \
4210-})
4211-
4212 #define __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH
4213 #define ptep_clear_flush_young(vma, address, ptep) \
4214 ({ \
4215 pte_t __pte = *(ptep); \
4216 int __young = pte_young(__pte); \
4217 __pte = pte_mkold(__pte); \
4218- if (test_bit(PG_pinned, &virt_to_page((vma)->vm_mm->pgd)->flags)) \
4219+ if (PagePinned(virt_to_page((vma)->vm_mm->pgd))) \
4220 (void)ptep_set_access_flags(vma, address, ptep, __pte, __young); \
4221 else if (__young) \
4222 (ptep)->pte_low = __pte.pte_low; \
4223@@ -383,7 +341,7 @@ static inline pte_t ptep_get_and_clear(s
4224 #define ptep_get_and_clear_full(mm, addr, ptep, full) \
4225 ((full) ? ({ \
4226 pte_t __res = *(ptep); \
4227- if (test_bit(PG_pinned, &virt_to_page((mm)->pgd)->flags)) \
4228+ if (PagePinned(virt_to_page((mm)->pgd))) \
4229 xen_l1_entry_update(ptep, __pte(0)); \
4230 else \
4231 *(ptep) = __pte(0); \
4232--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pgtable-3level.h 2008-12-15 11:27:22.000000000 +0100
4233+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgtable-3level.h 2009-02-16 16:17:21.000000000 +0100
4234@@ -23,26 +23,11 @@
4235 #define pud_present(pud) 1
4236
4237 /*
4238- * Is the pte executable?
4239- */
4240-static inline int pte_x(pte_t pte)
4241-{
4242- return !(__pte_val(pte) & _PAGE_NX);
4243-}
4244-
4245-/*
4246- * All present user-pages with !NX bit are user-executable:
4247- */
4248-static inline int pte_exec(pte_t pte)
4249-{
4250- return pte_user(pte) && pte_x(pte);
4251-}
4252-/*
4253 * All present pages with !NX bit are kernel-executable:
4254 */
4255 static inline int pte_exec_kernel(pte_t pte)
4256 {
4257- return pte_x(pte);
4258+ return !(__pte_val(pte) & _PAGE_NX);
4259 }
4260
4261 /* Rules for using set_pte: the pte being assigned *must* be
4262--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/processor_32.h 2008-12-15 11:27:22.000000000 +0100
4263+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/processor_32.h 2009-02-16 16:17:21.000000000 +0100
4264@@ -89,7 +89,6 @@ struct cpuinfo_x86 {
4265 #define X86_VENDOR_UMC 3
4266 #define X86_VENDOR_NEXGEN 4
4267 #define X86_VENDOR_CENTAUR 5
4268-#define X86_VENDOR_RISE 6
4269 #define X86_VENDOR_TRANSMETA 7
4270 #define X86_VENDOR_NSC 8
4271 #define X86_VENDOR_NUM 9
4272@@ -122,6 +121,7 @@ void __init cpu_detect(struct cpuinfo_x8
4273 extern void identify_boot_cpu(void);
4274 extern void identify_secondary_cpu(struct cpuinfo_x86 *);
4275 extern void print_cpu_info(struct cpuinfo_x86 *);
4276+extern void init_scattered_cpuid_features(struct cpuinfo_x86 *c);
4277 extern unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c);
4278 extern unsigned short num_cache_leaves;
4279
4280@@ -171,17 +171,6 @@ static inline void clear_in_cr4 (unsigne
4281 write_cr4(cr4);
4282 }
4283
4284-/*
4285- * NSC/Cyrix CPU indexed register access macros
4286- */
4287-
4288-#define getCx86(reg) ({ outb((reg), 0x22); inb(0x23); })
4289-
4290-#define setCx86(reg, data) do { \
4291- outb((reg), 0x22); \
4292- outb((data), 0x23); \
4293-} while (0)
4294-
4295 /* Stop speculative execution */
4296 static inline void sync_core(void)
4297 {
4298@@ -230,6 +219,10 @@ extern int bootloader_type;
4299
4300 #define HAVE_ARCH_PICK_MMAP_LAYOUT
4301
4302+extern void hard_disable_TSC(void);
4303+extern void disable_TSC(void);
4304+extern void hard_enable_TSC(void);
4305+
4306 /*
4307 * Size of io_bitmap.
4308 */
4309--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/system_32.h 2008-12-15 11:27:22.000000000 +0100
4310+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/system_32.h 2009-02-16 16:17:21.000000000 +0100
4311@@ -205,11 +205,6 @@ static inline unsigned long get_limit(un
4312 */
4313
4314
4315-/*
4316- * Actually only lfence would be needed for mb() because all stores done
4317- * by the kernel should be already ordered. But keep a full barrier for now.
4318- */
4319-
4320 #define mb() alternative("lock; addl $0,0(%%esp)", "mfence", X86_FEATURE_XMM2)
4321 #define rmb() alternative("lock; addl $0,0(%%esp)", "lfence", X86_FEATURE_XMM2)
4322
4323@@ -301,15 +296,6 @@ void enable_hlt(void);
4324 extern int es7000_plat;
4325 void cpu_idle_wait(void);
4326
4327-/*
4328- * On SMP systems, when the scheduler does migration-cost autodetection,
4329- * it needs a way to flush as much of the CPU's caches as possible:
4330- */
4331-static inline void sched_cacheflush(void)
4332-{
4333- wbinvd();
4334-}
4335-
4336 extern unsigned long arch_align_stack(unsigned long sp);
4337 extern void free_init_pages(char *what, unsigned long begin, unsigned long end);
4338
4339--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/tlbflush_32.h 2008-12-15 11:27:22.000000000 +0100
4340+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/tlbflush_32.h 2009-02-16 16:17:21.000000000 +0100
4341@@ -91,7 +91,11 @@ struct tlb_state
4342 DECLARE_PER_CPU(struct tlb_state, cpu_tlbstate);
4343 #endif /* SMP */
4344
4345-#define flush_tlb_kernel_range(start, end) flush_tlb_all()
4346+static inline void flush_tlb_kernel_range(unsigned long start,
4347+ unsigned long end)
4348+{
4349+ flush_tlb_all();
4350+}
4351
4352 static inline void flush_tlb_pgtables(struct mm_struct *mm,
4353 unsigned long start, unsigned long end)
4354--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/fixmap_64.h 2008-12-15 11:27:22.000000000 +0100
4355+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/fixmap_64.h 2009-02-16 16:17:21.000000000 +0100
4356@@ -23,9 +23,9 @@
4357 * compile time, but to set the physical address only
4358 * in the boot process.
4359 *
4360- * these 'compile-time allocated' memory buffers are
4361- * fixed-size 4k pages. (or larger if used with an increment
4362- * highger than 1) use fixmap_set(idx,phys) to associate
4363+ * These 'compile-time allocated' memory buffers are
4364+ * fixed-size 4k pages (or larger if used with an increment
4365+ * higher than 1). Use set_fixmap(idx,phys) to associate
4366 * physical memory with fixmap indices.
4367 *
4368 * TLB entries of such buffers will not be flushed across
4369@@ -36,6 +36,8 @@ enum fixed_addresses {
4370 VSYSCALL_LAST_PAGE,
4371 VSYSCALL_FIRST_PAGE = VSYSCALL_LAST_PAGE + ((VSYSCALL_END-VSYSCALL_START) >> PAGE_SHIFT) - 1,
4372 VSYSCALL_HPET,
4373+ FIX_DBGP_BASE,
4374+ FIX_EARLYCON_MEM_BASE,
4375 FIX_HPET_BASE,
4376 #ifdef CONFIG_X86_LOCAL_APIC
4377 FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */
4378@@ -105,7 +107,7 @@ static __always_inline unsigned long fix
4379 if (idx >= __end_of_fixed_addresses)
4380 __this_fixmap_does_not_exist();
4381
4382- return __fix_to_virt(idx);
4383+ return __fix_to_virt(idx);
4384 }
4385
4386 #endif
4387--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/mmu_context_64.h 2008-12-15 11:27:22.000000000 +0100
4388+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/mmu_context_64.h 2009-02-16 16:17:21.000000000 +0100
4389@@ -76,7 +76,7 @@ static inline void switch_mm(struct mm_s
4390
4391 if (likely(prev != next)) {
4392 BUG_ON(!xen_feature(XENFEAT_writable_page_tables) &&
4393- !next->context.pinned);
4394+ !PagePinned(virt_to_page(next->pgd)));
4395
4396 /* stop flush ipis for the previous mm */
4397 cpu_clear(cpu, prev->cpu_vm_mask);
4398@@ -131,7 +131,7 @@ static inline void switch_mm(struct mm_s
4399
4400 static inline void activate_mm(struct mm_struct *prev, struct mm_struct *next)
4401 {
4402- if (!next->context.pinned)
4403+ if (!PagePinned(virt_to_page(next->pgd)))
4404 mm_pin(next);
4405 switch_mm(prev, next, NULL);
4406 }
4407--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/page_64.h 2008-12-15 11:27:22.000000000 +0100
4408+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/page_64.h 2009-02-16 16:17:21.000000000 +0100
4409@@ -72,7 +72,8 @@ void copy_page(void *, void *);
4410 #define clear_user_page(page, vaddr, pg) clear_page(page)
4411 #define copy_user_page(to, from, vaddr, pg) copy_page(to, from)
4412
4413-#define alloc_zeroed_user_highpage(vma, vaddr) alloc_page_vma(GFP_HIGHUSER | __GFP_ZERO, vma, vaddr)
4414+#define __alloc_zeroed_user_highpage(movableflags, vma, vaddr) \
4415+ alloc_page_vma(GFP_HIGHUSER | __GFP_ZERO | movableflags, vma, vaddr)
4416 #define __HAVE_ARCH_ALLOC_ZEROED_USER_HIGHPAGE
4417
4418 /*
4419--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pgalloc_64.h 2008-12-15 11:27:22.000000000 +0100
4420+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgalloc_64.h 2009-02-16 16:17:21.000000000 +0100
4421@@ -21,7 +21,7 @@ static inline void pmd_populate_kernel(s
4422
4423 static inline void pmd_populate(struct mm_struct *mm, pmd_t *pmd, struct page *pte)
4424 {
4425- if (unlikely((mm)->context.pinned)) {
4426+ if (unlikely(PagePinned(virt_to_page((mm)->pgd)))) {
4427 BUG_ON(HYPERVISOR_update_va_mapping(
4428 (unsigned long)__va(page_to_pfn(pte) << PAGE_SHIFT),
4429 pfn_pte(page_to_pfn(pte), PAGE_KERNEL_RO), 0));
4430@@ -33,7 +33,7 @@ static inline void pmd_populate(struct m
4431
4432 static inline void pud_populate(struct mm_struct *mm, pud_t *pud, pmd_t *pmd)
4433 {
4434- if (unlikely((mm)->context.pinned)) {
4435+ if (unlikely(PagePinned(virt_to_page((mm)->pgd)))) {
4436 BUG_ON(HYPERVISOR_update_va_mapping(
4437 (unsigned long)pmd,
4438 pfn_pte(virt_to_phys(pmd)>>PAGE_SHIFT,
4439@@ -50,7 +50,7 @@ static inline void pud_populate(struct m
4440 */
4441 static inline void pgd_populate(struct mm_struct *mm, pgd_t *pgd, pud_t *pud)
4442 {
4443- if (unlikely((mm)->context.pinned)) {
4444+ if (unlikely(PagePinned(virt_to_page((mm)->pgd)))) {
4445 BUG_ON(HYPERVISOR_update_va_mapping(
4446 (unsigned long)pud,
4447 pfn_pte(virt_to_phys(pud)>>PAGE_SHIFT,
4448--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/pgtable_64.h 2008-12-15 11:27:22.000000000 +0100
4449+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/pgtable_64.h 2009-02-16 16:17:21.000000000 +0100
4450@@ -304,7 +304,7 @@ static inline pte_t ptep_get_and_clear_f
4451 {
4452 if (full) {
4453 pte_t pte = *ptep;
4454- if (mm->context.pinned)
4455+ if (PagePinned(virt_to_page(mm->pgd)))
4456 xen_l1_entry_update(ptep, __pte(0));
4457 else
4458 *ptep = __pte(0);
4459@@ -333,21 +333,15 @@ static inline pte_t ptep_get_and_clear_f
4460 * Undefined behaviour if not..
4461 */
4462 #define __LARGE_PTE (_PAGE_PSE|_PAGE_PRESENT)
4463-static inline int pte_user(pte_t pte) { return __pte_val(pte) & _PAGE_USER; }
4464-static inline int pte_read(pte_t pte) { return __pte_val(pte) & _PAGE_USER; }
4465-static inline int pte_exec(pte_t pte) { return !(__pte_val(pte) & _PAGE_NX); }
4466 static inline int pte_dirty(pte_t pte) { return __pte_val(pte) & _PAGE_DIRTY; }
4467 static inline int pte_young(pte_t pte) { return __pte_val(pte) & _PAGE_ACCESSED; }
4468 static inline int pte_write(pte_t pte) { return __pte_val(pte) & _PAGE_RW; }
4469 static inline int pte_file(pte_t pte) { return __pte_val(pte) & _PAGE_FILE; }
4470 static inline int pte_huge(pte_t pte) { return __pte_val(pte) & _PAGE_PSE; }
4471
4472-static inline pte_t pte_rdprotect(pte_t pte) { __pte_val(pte) &= ~_PAGE_USER; return pte; }
4473-static inline pte_t pte_exprotect(pte_t pte) { __pte_val(pte) &= ~_PAGE_USER; return pte; }
4474 static inline pte_t pte_mkclean(pte_t pte) { __pte_val(pte) &= ~_PAGE_DIRTY; return pte; }
4475 static inline pte_t pte_mkold(pte_t pte) { __pte_val(pte) &= ~_PAGE_ACCESSED; return pte; }
4476 static inline pte_t pte_wrprotect(pte_t pte) { __pte_val(pte) &= ~_PAGE_RW; return pte; }
4477-static inline pte_t pte_mkread(pte_t pte) { __pte_val(pte) |= _PAGE_USER; return pte; }
4478 static inline pte_t pte_mkexec(pte_t pte) { __pte_val(pte) &= ~_PAGE_NX; return pte; }
4479 static inline pte_t pte_mkdirty(pte_t pte) { __pte_val(pte) |= _PAGE_DIRTY; return pte; }
4480 static inline pte_t pte_mkyoung(pte_t pte) { __pte_val(pte) |= _PAGE_ACCESSED; return pte; }
4481@@ -355,13 +349,6 @@ static inline pte_t pte_mkwrite(pte_t pt
4482 static inline pte_t pte_mkhuge(pte_t pte) { __pte_val(pte) |= _PAGE_PSE; return pte; }
4483 static inline pte_t pte_clrhuge(pte_t pte) { __pte_val(pte) &= ~_PAGE_PSE; return pte; }
4484
4485-static inline int ptep_test_and_clear_dirty(struct vm_area_struct *vma, unsigned long addr, pte_t *ptep)
4486-{
4487- if (!pte_dirty(*ptep))
4488- return 0;
4489- return test_and_clear_bit(_PAGE_BIT_DIRTY, &ptep->pte);
4490-}
4491-
4492 static inline int ptep_test_and_clear_young(struct vm_area_struct *vma, unsigned long addr, pte_t *ptep)
4493 {
4494 if (!pte_young(*ptep))
4495@@ -500,26 +487,13 @@ static inline pte_t pte_modify(pte_t pte
4496 __changed; \
4497 })
4498
4499-#define __HAVE_ARCH_PTEP_CLEAR_DIRTY_FLUSH
4500-#define ptep_clear_flush_dirty(vma, address, ptep) \
4501-({ \
4502- pte_t __pte = *(ptep); \
4503- int __dirty = pte_dirty(__pte); \
4504- __pte = pte_mkclean(__pte); \
4505- if ((vma)->vm_mm->context.pinned) \
4506- (void)ptep_set_access_flags(vma, address, ptep, __pte, __dirty); \
4507- else if (__dirty) \
4508- set_pte(ptep, __pte); \
4509- __dirty; \
4510-})
4511-
4512 #define __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH
4513 #define ptep_clear_flush_young(vma, address, ptep) \
4514 ({ \
4515 pte_t __pte = *(ptep); \
4516 int __young = pte_young(__pte); \
4517 __pte = pte_mkold(__pte); \
4518- if ((vma)->vm_mm->context.pinned) \
4519+ if (PagePinned(virt_to_page((vma)->vm_mm->pgd))) \
4520 (void)ptep_set_access_flags(vma, address, ptep, __pte, __young); \
4521 else if (__young) \
4522 set_pte(ptep, __pte); \
4523@@ -570,6 +544,8 @@ int xen_change_pte_range(struct mm_struc
4524 #define arch_change_pte_range(mm, pmd, addr, end, newprot, dirty_accountable) \
4525 xen_change_pte_range(mm, pmd, addr, end, newprot, dirty_accountable)
4526
4527+pte_t *lookup_address(unsigned long addr);
4528+
4529 #define io_remap_pfn_range(vma, vaddr, pfn, size, prot) \
4530 direct_remap_pfn_range(vma,vaddr,pfn,size,prot,DOMID_IO)
4531
4532@@ -587,7 +563,6 @@ int xen_change_pte_range(struct mm_struc
4533 (((o) & (1UL << (__VIRTUAL_MASK_SHIFT-1))) ? ((o) | (~__VIRTUAL_MASK)) : (o))
4534
4535 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
4536-#define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_DIRTY
4537 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR
4538 #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL
4539 #define __HAVE_ARCH_PTEP_CLEAR_FLUSH
4540--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/processor_64.h 2008-12-15 11:27:22.000000000 +0100
4541+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/processor_64.h 2009-02-16 16:17:21.000000000 +0100
4542@@ -83,7 +83,6 @@ struct cpuinfo_x86 {
4543 #define X86_VENDOR_UMC 3
4544 #define X86_VENDOR_NEXGEN 4
4545 #define X86_VENDOR_CENTAUR 5
4546-#define X86_VENDOR_RISE 6
4547 #define X86_VENDOR_TRANSMETA 7
4548 #define X86_VENDOR_NUM 8
4549 #define X86_VENDOR_UNKNOWN 0xff
4550@@ -100,6 +99,7 @@ extern char ignore_irq13;
4551
4552 extern void identify_cpu(struct cpuinfo_x86 *);
4553 extern void print_cpu_info(struct cpuinfo_x86 *);
4554+extern void init_scattered_cpuid_features(struct cpuinfo_x86 *c);
4555 extern unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c);
4556 extern unsigned short num_cache_leaves;
4557
4558@@ -377,12 +377,10 @@ static inline void sync_core(void)
4559 asm volatile("cpuid" : "=a" (tmp) : "0" (1) : "ebx","ecx","edx","memory");
4560 }
4561
4562-#define cpu_has_fpu 1
4563-
4564 #define ARCH_HAS_PREFETCH
4565 static inline void prefetch(void *x)
4566 {
4567- asm volatile("prefetcht0 %0" :: "m" (*(unsigned long *)x));
4568+ asm volatile("prefetcht0 (%0)" :: "r" (x));
4569 }
4570
4571 #define ARCH_HAS_PREFETCHW 1
4572@@ -400,17 +398,6 @@ static inline void prefetchw(void *x)
4573
4574 #define cpu_relax() rep_nop()
4575
4576-/*
4577- * NSC/Cyrix CPU indexed register access macros
4578- */
4579-
4580-#define getCx86(reg) ({ outb((reg), 0x22); inb(0x23); })
4581-
4582-#define setCx86(reg, data) do { \
4583- outb((reg), 0x22); \
4584- outb((data), 0x23); \
4585-} while (0)
4586-
4587 static inline void serialize_cpu(void)
4588 {
4589 __asm__ __volatile__ ("cpuid" : : : "ax", "bx", "cx", "dx");
4590--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/system_64.h 2008-12-15 11:27:22.000000000 +0100
4591+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/system_64.h 2009-02-16 16:17:21.000000000 +0100
4592@@ -79,12 +79,16 @@ static inline unsigned long read_cr0(voi
4593 unsigned long cr0;
4594 asm volatile("movq %%cr0,%0" : "=r" (cr0));
4595 return cr0;
4596-}
4597+}
4598
4599 static inline void write_cr0(unsigned long val)
4600 {
4601 asm volatile("movq %0,%%cr0" :: "r" (val));
4602-}
4603+}
4604+
4605+#define read_cr2() current_vcpu_info()->arch.cr2
4606+
4607+#define write_cr2(val) ((void)(current_vcpu_info()->arch.cr2 = (val)))
4608
4609 #define read_cr3() ({ \
4610 unsigned long __dummy; \
4611@@ -103,27 +107,28 @@ static inline unsigned long read_cr4(voi
4612 unsigned long cr4;
4613 asm("movq %%cr4,%0" : "=r" (cr4));
4614 return cr4;
4615-}
4616+}
4617
4618 static inline void write_cr4(unsigned long val)
4619 {
4620 asm volatile("movq %0,%%cr4" :: "r" (val) : "memory");
4621-}
4622-
4623-#define stts() (HYPERVISOR_fpu_taskswitch(1))
4624+}
4625
4626-#define wbinvd() \
4627- __asm__ __volatile__ ("wbinvd": : :"memory");
4628+static inline unsigned long read_cr8(void)
4629+{
4630+ return 0;
4631+}
4632
4633-/*
4634- * On SMP systems, when the scheduler does migration-cost autodetection,
4635- * it needs a way to flush as much of the CPU's caches as possible.
4636- */
4637-static inline void sched_cacheflush(void)
4638+static inline void write_cr8(unsigned long val)
4639 {
4640- wbinvd();
4641+ BUG_ON(val);
4642 }
4643
4644+#define stts() (HYPERVISOR_fpu_taskswitch(1))
4645+
4646+#define wbinvd() \
4647+ __asm__ __volatile__ ("wbinvd": : :"memory")
4648+
4649 #endif /* __KERNEL__ */
4650
4651 #define nop() __asm__ __volatile__ ("nop")
4652--- sle11-2009-05-14.orig/include/asm-x86/mach-xen/asm/tlbflush_64.h 2008-12-15 11:27:22.000000000 +0100
4653+++ sle11-2009-05-14/include/asm-x86/mach-xen/asm/tlbflush_64.h 2009-02-16 16:17:21.000000000 +0100
4654@@ -89,7 +89,11 @@ static inline void flush_tlb_range(struc
4655
4656 #endif
4657
4658-#define flush_tlb_kernel_range(start, end) flush_tlb_all()
4659+static inline void flush_tlb_kernel_range(unsigned long start,
4660+ unsigned long end)
4661+{
4662+ flush_tlb_all();
4663+}
4664
4665 static inline void flush_tlb_pgtables(struct mm_struct *mm,
4666 unsigned long start, unsigned long end)
4667--- sle11-2009-05-14.orig/include/asm-x86/thread_info.h 2009-04-20 11:36:10.000000000 +0200
4668+++ sle11-2009-05-14/include/asm-x86/thread_info.h 2009-02-16 16:17:21.000000000 +0100
4669@@ -150,7 +150,8 @@ struct thread_info {
4670 #define _TIF_WORK_CTXSW_PREV _TIF_WORK_CTXSW
4671 #define _TIF_WORK_CTXSW_NEXT (_TIF_WORK_CTXSW|_TIF_DEBUG)
4672 #else
4673-#define _TIF_WORK_CTXSW _TIF_DEBUG
4674+#define _TIF_WORK_CTXSW_NEXT (_TIF_NOTSC | _TIF_DEBUG)
4675+#define _TIF_WORK_CTXSW_PREV (_TIF_NOTSC)
4676 #endif
4677
4678 #define PREEMPT_ACTIVE 0x10000000
4679--- sle11-2009-05-14.orig/include/asm-x86/xen/interface.h 2009-05-14 10:56:29.000000000 +0200
4680+++ sle11-2009-05-14/include/asm-x86/xen/interface.h 2009-02-16 16:17:21.000000000 +0100
4681@@ -10,17 +10,17 @@
4682 #define __ASM_X86_XEN_INTERFACE_H
4683
4684 #ifdef __XEN__
4685-#define __DEFINE_GUEST_HANDLE(name, type) \
4686+#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
4687 typedef struct { type *p; } __guest_handle_ ## name
4688 #else
4689-#define __DEFINE_GUEST_HANDLE(name, type) \
4690+#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
4691 typedef type * __guest_handle_ ## name
4692 #endif
4693
4694-#define DEFINE_GUEST_HANDLE_STRUCT(name) \
4695- __DEFINE_GUEST_HANDLE(name, struct name)
4696-#define DEFINE_GUEST_HANDLE(name) __DEFINE_GUEST_HANDLE(name, name)
4697-#define GUEST_HANDLE(name) __guest_handle_ ## name
4698+#define DEFINE_XEN_GUEST_HANDLE_STRUCT(name) \
4699+ __DEFINE_XEN_GUEST_HANDLE(name, struct name)
4700+#define DEFINE_XEN_GUEST_HANDLE(name) __DEFINE_XEN_GUEST_HANDLE(name, name)
4701+#define XEN_GUEST_HANDLE(name) __guest_handle_ ## name
4702
4703 #ifdef __XEN__
4704 #if defined(__i386__)
4705@@ -47,14 +47,8 @@
4706 #endif
4707
4708 #ifndef __ASSEMBLY__
4709-/* Guest handles for primitive C types. */
4710-__DEFINE_GUEST_HANDLE(uchar, unsigned char);
4711-__DEFINE_GUEST_HANDLE(uint, unsigned int);
4712-__DEFINE_GUEST_HANDLE(ulong, unsigned long);
4713-DEFINE_GUEST_HANDLE(char);
4714-DEFINE_GUEST_HANDLE(int);
4715-DEFINE_GUEST_HANDLE(long);
4716-DEFINE_GUEST_HANDLE(void);
4717+typedef unsigned long xen_pfn_t;
4718+typedef unsigned long xen_ulong_t;
4719 #endif
4720
4721 #ifndef HYPERVISOR_VIRT_START
4722@@ -103,7 +97,7 @@ struct trap_info {
4723 uint16_t cs; /* code selector */
4724 unsigned long address; /* code offset */
4725 };
4726-DEFINE_GUEST_HANDLE_STRUCT(trap_info);
4727+DEFINE_XEN_GUEST_HANDLE_STRUCT(trap_info);
4728
4729 struct arch_shared_info {
4730 unsigned long max_pfn; /* max pfn that appears in table */
4731@@ -157,7 +151,7 @@ struct vcpu_guest_context {
4732 uint64_t gs_base_user;
4733 #endif
4734 };
4735-DEFINE_GUEST_HANDLE_STRUCT(vcpu_guest_context);
4736+DEFINE_XEN_GUEST_HANDLE_STRUCT(vcpu_guest_context);
4737 #endif /* !__ASSEMBLY__ */
4738
4739 /*
4740--- sle11-2009-05-14.orig/include/asm-x86/xen/interface_32.h 2009-05-14 10:56:29.000000000 +0200
4741+++ sle11-2009-05-14/include/asm-x86/xen/interface_32.h 2009-02-16 16:17:21.000000000 +0100
4742@@ -62,7 +62,7 @@ struct cpu_user_regs {
4743 uint16_t fs, _pad4;
4744 uint16_t gs, _pad5;
4745 };
4746-DEFINE_GUEST_HANDLE_STRUCT(cpu_user_regs);
4747+DEFINE_XEN_GUEST_HANDLE_STRUCT(cpu_user_regs);
4748
4749 typedef uint64_t tsc_timestamp_t; /* RDTSC timestamp */
4750
4751--- sle11-2009-05-14.orig/include/linux/elfnote.h 2009-05-14 10:56:29.000000000 +0200
4752+++ sle11-2009-05-14/include/linux/elfnote.h 2009-02-16 16:17:21.000000000 +0100
4753@@ -52,7 +52,7 @@
4754 4484:.balign 4 ; \
4755 .popsection ;
4756
4757-#define ELFNOTE(name, type, desc) \
4758+#define ELFNOTE(name, type, desc...) \
4759 ELFNOTE_START(name, type, "") \
4760 desc ; \
4761 ELFNOTE_END
4762--- sle11-2009-05-14.orig/include/linux/page-flags.h 2009-01-16 10:20:18.000000000 +0100
4763+++ sle11-2009-05-14/include/linux/page-flags.h 2009-02-16 16:17:21.000000000 +0100
4764@@ -100,6 +100,10 @@ enum pageflags {
4765 #endif
4766 #ifdef CONFIG_XEN
4767 PG_foreign, /* Page is owned by foreign allocator. */
4768+ PG_pinned, /* Cannot alias with PG_owner_priv_1 since
4769+ * bad_page() checks include this bit.
4770+ * Also cannot use PG_arch_1 since that now
4771+ * has a different purpose on x86. */
4772 #endif
4773 __NR_PAGEFLAGS,
4774
4775--- sle11-2009-05-14.orig/include/linux/skbuff.h 2009-03-04 11:28:34.000000000 +0100
4776+++ sle11-2009-05-14/include/linux/skbuff.h 2009-02-16 16:17:21.000000000 +0100
4777@@ -1772,7 +1772,7 @@ static inline void skb_forward_csum(stru
4778
4779 bool skb_partial_csum_set(struct sk_buff *skb, u16 start, u16 off);
4780
4781-#ifdef CONFIG_XEN
4782+#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN)
4783 int skb_checksum_setup(struct sk_buff *skb);
4784 #else
4785 static inline int skb_checksum_setup(struct sk_buff *skb) { return 0; }
4786--- sle11-2009-05-14.orig/include/xen/driver_util.h 2009-05-14 10:56:29.000000000 +0200
4787+++ sle11-2009-05-14/include/xen/driver_util.h 2009-02-16 16:17:21.000000000 +0100
4788@@ -5,10 +5,6 @@
4789 #include <linux/vmalloc.h>
4790 #include <linux/device.h>
4791
4792-/* Allocate/destroy a 'vmalloc' VM area. */
4793-extern struct vm_struct *alloc_vm_area(unsigned long size);
4794-extern void free_vm_area(struct vm_struct *area);
4795-
4796 extern struct class *get_xen_class(void);
4797
4798 #endif /* __ASM_XEN_DRIVER_UTIL_H__ */
4799--- sle11-2009-05-14.orig/include/xen/features.h 2009-05-14 10:56:29.000000000 +0200
4800+++ sle11-2009-05-14/include/xen/features.h 2009-02-16 16:17:21.000000000 +0100
4801@@ -10,6 +10,7 @@
4802 #define __XEN_FEATURES_H__
4803
4804 #include <xen/interface/features.h>
4805+#include <xen/interface/version.h>
4806
4807 void xen_setup_features(void);
4808
4809@@ -20,4 +21,4 @@ static inline int xen_feature(int flag)
4810 return xen_features[flag];
4811 }
4812
4813-#endif /* __ASM_XEN_FEATURES_H__ */
4814+#endif /* __XEN_FEATURES_H__ */
4815--- sle11-2009-05-14.orig/include/xen/interface/arch-x86/xen.h 2009-05-14 10:56:29.000000000 +0200
4816+++ sle11-2009-05-14/include/xen/interface/arch-x86/xen.h 2009-02-16 16:17:21.000000000 +0100
4817@@ -49,6 +49,9 @@
4818 #define get_xen_guest_handle(val, hnd) do { val = (hnd).p; } while (0)
4819 #endif
4820
4821+/* Allow co-existing Linux 2.6.23+ Xen interface definitions. */
4822+#define DEFINE_XEN_GUEST_HANDLE_STRUCT(name) struct name
4823+
4824 #if defined(__i386__)
4825 #include "xen-x86_32.h"
4826 #elif defined(__x86_64__)
4827--- sle11-2009-05-14.orig/include/xen/interface/event_channel.h 2008-11-25 12:35:56.000000000 +0100
4828+++ sle11-2009-05-14/include/xen/interface/event_channel.h 2009-02-16 16:17:21.000000000 +0100
4829@@ -248,6 +248,7 @@ struct evtchn_op {
4830 struct evtchn_unmask unmask;
4831 } u;
4832 };
4833+DEFINE_XEN_GUEST_HANDLE_STRUCT(evtchn_op);
4834 typedef struct evtchn_op evtchn_op_t;
4835 DEFINE_XEN_GUEST_HANDLE(evtchn_op_t);
4836
4837--- sle11-2009-05-14.orig/include/xen/interface/io/netif.h 2008-11-25 12:35:56.000000000 +0100
4838+++ sle11-2009-05-14/include/xen/interface/io/netif.h 2009-02-16 16:17:21.000000000 +0100
4839@@ -183,8 +183,22 @@ typedef struct netif_rx_response netif_r
4840 * Generate netif ring structures and types.
4841 */
4842
4843+#if defined(CONFIG_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H)
4844 DEFINE_RING_TYPES(netif_tx, struct netif_tx_request, struct netif_tx_response);
4845 DEFINE_RING_TYPES(netif_rx, struct netif_rx_request, struct netif_rx_response);
4846+#else
4847+#define xen_netif_tx_request netif_tx_request
4848+#define xen_netif_rx_request netif_rx_request
4849+#define xen_netif_tx_response netif_tx_response
4850+#define xen_netif_rx_response netif_rx_response
4851+DEFINE_RING_TYPES(xen_netif_tx,
4852+ struct xen_netif_tx_request,
4853+ struct xen_netif_tx_response);
4854+DEFINE_RING_TYPES(xen_netif_rx,
4855+ struct xen_netif_rx_request,
4856+ struct xen_netif_rx_response);
4857+#define xen_netif_extra_info netif_extra_info
4858+#endif
4859
4860 #define NETIF_RSP_DROPPED -2
4861 #define NETIF_RSP_ERROR -1
4862--- sle11-2009-05-14.orig/include/xen/interface/memory.h 2008-11-25 12:35:56.000000000 +0100
4863+++ sle11-2009-05-14/include/xen/interface/memory.h 2009-02-16 16:17:21.000000000 +0100
4864@@ -82,6 +82,7 @@ struct xen_memory_reservation {
4865 domid_t domid;
4866
4867 };
4868+DEFINE_XEN_GUEST_HANDLE_STRUCT(xen_memory_reservation);
4869 typedef struct xen_memory_reservation xen_memory_reservation_t;
4870 DEFINE_XEN_GUEST_HANDLE(xen_memory_reservation_t);
4871
4872@@ -175,6 +176,7 @@ struct xen_machphys_mfn_list {
4873 */
4874 unsigned int nr_extents;
4875 };
4876+DEFINE_XEN_GUEST_HANDLE_STRUCT(xen_machphys_mfn_list);
4877 typedef struct xen_machphys_mfn_list xen_machphys_mfn_list_t;
4878 DEFINE_XEN_GUEST_HANDLE(xen_machphys_mfn_list_t);
4879
4880@@ -214,6 +216,7 @@ struct xen_add_to_physmap {
4881 /* GPFN where the source mapping page should appear. */
4882 xen_pfn_t gpfn;
4883 };
4884+DEFINE_XEN_GUEST_HANDLE_STRUCT(xen_add_to_physmap);
4885 typedef struct xen_add_to_physmap xen_add_to_physmap_t;
4886 DEFINE_XEN_GUEST_HANDLE(xen_add_to_physmap_t);
4887
4888@@ -254,6 +257,7 @@ struct xen_translate_gpfn_list {
4889 */
4890 XEN_GUEST_HANDLE(xen_pfn_t) mfn_list;
4891 };
4892+DEFINE_XEN_GUEST_HANDLE_STRUCT(xen_translate_gpfn_list);
4893 typedef struct xen_translate_gpfn_list xen_translate_gpfn_list_t;
4894 DEFINE_XEN_GUEST_HANDLE(xen_translate_gpfn_list_t);
4895
4896--- sle11-2009-05-14.orig/include/xen/interface/sched.h 2008-11-25 12:35:56.000000000 +0100
4897+++ sle11-2009-05-14/include/xen/interface/sched.h 2009-02-16 16:17:21.000000000 +0100
4898@@ -67,6 +67,7 @@
4899 struct sched_shutdown {
4900 unsigned int reason; /* SHUTDOWN_* */
4901 };
4902+DEFINE_XEN_GUEST_HANDLE_STRUCT(sched_shutdown);
4903 typedef struct sched_shutdown sched_shutdown_t;
4904 DEFINE_XEN_GUEST_HANDLE(sched_shutdown_t);
4905
4906@@ -81,6 +82,7 @@ struct sched_poll {
4907 unsigned int nr_ports;
4908 uint64_t timeout;
4909 };
4910+DEFINE_XEN_GUEST_HANDLE_STRUCT(sched_poll);
4911 typedef struct sched_poll sched_poll_t;
4912 DEFINE_XEN_GUEST_HANDLE(sched_poll_t);
4913
4914--- sle11-2009-05-14.orig/include/xen/interface/version.h 2008-11-25 12:35:56.000000000 +0100
4915+++ sle11-2009-05-14/include/xen/interface/version.h 2009-02-16 16:17:21.000000000 +0100
4916@@ -36,6 +36,9 @@
4917 /* arg == xen_extraversion_t. */
4918 #define XENVER_extraversion 1
4919 typedef char xen_extraversion_t[16];
4920+struct xen_extraversion {
4921+ xen_extraversion_t extraversion;
4922+};
4923 #define XEN_EXTRAVERSION_LEN (sizeof(xen_extraversion_t))
4924
4925 /* arg == xen_compile_info_t. */
4926@@ -50,10 +53,16 @@ typedef struct xen_compile_info xen_comp
4927
4928 #define XENVER_capabilities 3
4929 typedef char xen_capabilities_info_t[1024];
4930+struct xen_capabilities_info {
4931+ xen_capabilities_info_t info;
4932+};
4933 #define XEN_CAPABILITIES_INFO_LEN (sizeof(xen_capabilities_info_t))
4934
4935 #define XENVER_changeset 4
4936 typedef char xen_changeset_info_t[64];
4937+struct xen_changeset_info {
4938+ xen_changeset_info_t info;
4939+};
4940 #define XEN_CHANGESET_INFO_LEN (sizeof(xen_changeset_info_t))
4941
4942 #define XENVER_platform_parameters 5
4943--- sle11-2009-05-14.orig/include/xen/interface/xen.h 2008-11-25 12:35:56.000000000 +0100
4944+++ sle11-2009-05-14/include/xen/interface/xen.h 2009-05-14 11:17:48.000000000 +0200
4945@@ -32,7 +32,9 @@
4946 #include <asm/pvclock-abi.h>
4947 #endif
4948
4949-#if defined(__i386__) || defined(__x86_64__)
4950+#if defined(CONFIG_PARAVIRT_XEN) && !defined(HAVE_XEN_PLATFORM_COMPAT_H)
4951+#include <asm/xen/interface.h>
4952+#elif defined(__i386__) || defined(__x86_64__)
4953 #include "arch-x86/xen.h"
4954 #elif defined(__ia64__)
4955 #include "arch-ia64.h"
4956@@ -110,7 +112,7 @@ DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
4957 */
4958
4959 /* New sched_op hypercall introduced in 0x00030101. */
4960-#if __XEN_INTERFACE_VERSION__ < 0x00030101
4961+#if __XEN_INTERFACE_VERSION__ < 0x00030101 || (defined(CONFIG_PARAVIRT_XEN) && !defined(HAVE_XEN_PLATFORM_COMPAT_H))
4962 #undef __HYPERVISOR_sched_op
4963 #define __HYPERVISOR_sched_op __HYPERVISOR_sched_op_compat
4964 #else
4965@@ -126,7 +128,7 @@ DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
4966 #endif
4967
4968 /* New platform_op hypercall introduced in 0x00030204. */
4969-#if __XEN_INTERFACE_VERSION__ < 0x00030204
4970+#if __XEN_INTERFACE_VERSION__ < 0x00030204 || (defined(CONFIG_PARAVIRT_XEN) && !defined(HAVE_XEN_PLATFORM_COMPAT_H))
4971 #define __HYPERVISOR_dom0_op __HYPERVISOR_platform_op
4972 #endif
4973
4974@@ -285,6 +287,7 @@ struct mmuext_op {
4975 xen_pfn_t src_mfn;
4976 } arg2;
4977 };
4978+DEFINE_XEN_GUEST_HANDLE_STRUCT(mmuext_op);
4979 typedef struct mmuext_op mmuext_op_t;
4980 DEFINE_XEN_GUEST_HANDLE(mmuext_op_t);
4981 #endif
4982@@ -367,6 +370,7 @@ struct mmu_update {
4983 uint64_t ptr; /* Machine address of PTE. */
4984 uint64_t val; /* New contents of PTE. */
4985 };
4986+DEFINE_XEN_GUEST_HANDLE_STRUCT(mmu_update);
4987 typedef struct mmu_update mmu_update_t;
4988 DEFINE_XEN_GUEST_HANDLE(mmu_update_t);
4989
4990@@ -375,9 +379,15 @@ DEFINE_XEN_GUEST_HANDLE(mmu_update_t);
4991 * NB. The fields are natural register size for this architecture.
4992 */
4993 struct multicall_entry {
4994- unsigned long op, result;
4995+ unsigned long op;
4996+#if !defined(CONFIG_PARAVIRT_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H)
4997+ unsigned long result;
4998+#else
4999+ long result;
5000+#endif
5001 unsigned long args[6];
5002 };
5003+DEFINE_XEN_GUEST_HANDLE_STRUCT(multicall_entry);
5004 typedef struct multicall_entry multicall_entry_t;
5005 DEFINE_XEN_GUEST_HANDLE(multicall_entry_t);
5006
5007--- sle11-2009-05-14.orig/include/xen/xenbus.h 2009-03-04 11:25:55.000000000 +0100
5008+++ sle11-2009-05-14/include/xen/xenbus.h 2009-02-16 16:17:21.000000000 +0100
5009@@ -57,16 +57,20 @@ struct xenbus_watch
5010 void (*callback)(struct xenbus_watch *,
5011 const char **vec, unsigned int len);
5012
5013+#if defined(CONFIG_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H)
5014 /* See XBWF_ definitions below. */
5015 unsigned long flags;
5016+#endif
5017 };
5018
5019+#if defined(CONFIG_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H)
5020 /*
5021 * Execute callback in its own kthread. Useful if the callback is long
5022 * running or heavily serialised, to avoid taking out the main xenwatch thread
5023 * for a long period of time (or even unwittingly causing a deadlock).
5024 */
5025 #define XBWF_new_thread 1
5026+#endif
5027
5028 /* A xenbus device. */
5029 struct xenbus_device {
5030@@ -214,6 +218,7 @@ int xenbus_watch_path(struct xenbus_devi
5031 const char **, unsigned int));
5032
5033
5034+#if defined(CONFIG_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H)
5035 /**
5036 * Register a watch on the given path/path2, using the given xenbus_watch
5037 * structure for storage, and the given callback function as the callback.
5038@@ -227,7 +232,13 @@ int xenbus_watch_path2(struct xenbus_dev
5039 const char *path2, struct xenbus_watch *watch,
5040 void (*callback)(struct xenbus_watch *,
5041 const char **, unsigned int));
5042-
5043+#else
5044+int xenbus_watch_pathfmt(struct xenbus_device *dev, struct xenbus_watch *watch,
5045+ void (*callback)(struct xenbus_watch *,
5046+ const char **, unsigned int),
5047+ const char *pathfmt, ...)
5048+ __attribute__ ((format (printf, 4, 5)));
5049+#endif
5050
5051 /**
5052 * Advertise in the store a change of the given driver to the given new_state.
5053--- sle11-2009-05-14.orig/net/core/dev.c 2008-12-15 11:27:22.000000000 +0100
5054+++ sle11-2009-05-14/net/core/dev.c 2009-02-16 16:17:21.000000000 +0100
5055@@ -131,7 +131,7 @@
5056
5057 #include "net-sysfs.h"
5058
5059-#ifdef CONFIG_XEN
5060+#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN)
5061 #include <net/ip.h>
5062 #include <linux/tcp.h>
5063 #include <linux/udp.h>
5064@@ -1740,42 +1740,54 @@ static struct netdev_queue *dev_pick_tx(
5065 return netdev_get_tx_queue(dev, queue_index);
5066 }
5067
5068-#ifdef CONFIG_XEN
5069+#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN)
5070 inline int skb_checksum_setup(struct sk_buff *skb)
5071 {
5072- if (skb->proto_csum_blank) {
5073- struct iphdr *iph;
5074- unsigned char *th;
5075+ struct iphdr *iph;
5076+ unsigned char *th;
5077+ int err = -EPROTO;
5078
5079- if (skb->protocol != htons(ETH_P_IP))
5080- goto out;
5081- iph = ip_hdr(skb);
5082- th = skb_network_header(skb) + 4 * iph->ihl;
5083- if (th >= skb_tail_pointer(skb))
5084- goto out;
5085- skb->csum_start = th - skb->head;
5086- switch (iph->protocol) {
5087- case IPPROTO_TCP:
5088- skb->csum_offset = offsetof(struct tcphdr, check);
5089- break;
5090- case IPPROTO_UDP:
5091- skb->csum_offset = offsetof(struct udphdr, check);
5092- break;
5093- default:
5094- if (net_ratelimit())
5095- printk(KERN_ERR "Attempting to checksum a non-"
5096- "TCP/UDP packet, dropping a protocol"
5097- " %d packet", iph->protocol);
5098- goto out;
5099- }
5100- if ((th + skb->csum_offset + 2) > skb_tail_pointer(skb))
5101- goto out;
5102- skb->ip_summed = CHECKSUM_PARTIAL;
5103- skb->proto_csum_blank = 0;
5104+#ifdef CONFIG_XEN
5105+ if (!skb->proto_csum_blank)
5106+ return 0;
5107+#endif
5108+
5109+ if (skb->protocol != htons(ETH_P_IP))
5110+ goto out;
5111+
5112+ iph = ip_hdr(skb);
5113+ th = skb_network_header(skb) + 4 * iph->ihl;
5114+ if (th >= skb_tail_pointer(skb))
5115+ goto out;
5116+
5117+ skb->csum_start = th - skb->head;
5118+ switch (iph->protocol) {
5119+ case IPPROTO_TCP:
5120+ skb->csum_offset = offsetof(struct tcphdr, check);
5121+ break;
5122+ case IPPROTO_UDP:
5123+ skb->csum_offset = offsetof(struct udphdr, check);
5124+ break;
5125+ default:
5126+ if (net_ratelimit())
5127+ printk(KERN_ERR "Attempting to checksum a non-"
5128+ "TCP/UDP packet, dropping a protocol"
5129+ " %d packet", iph->protocol);
5130+ goto out;
5131 }
5132- return 0;
5133+
5134+ if ((th + skb->csum_offset + 2) > skb_tail_pointer(skb))
5135+ goto out;
5136+
5137+#ifdef CONFIG_XEN
5138+ skb->ip_summed = CHECKSUM_PARTIAL;
5139+ skb->proto_csum_blank = 0;
5140+#endif
5141+
5142+ err = 0;
5143+
5144 out:
5145- return -EPROTO;
5146+ return err;
5147 }
5148 EXPORT_SYMBOL(skb_checksum_setup);
5149 #endif
Peter's tree of IPFire 2.x