Contents of /trunk/kernel26-xen/patches-2.6.25-r1/1024-2.6.25-xen-patch-2.6.23.patch
Parent Directory | Revision Log
Revision 609 -
(show annotations)
(download)
Fri May 23 17:35:37 2008 UTC (16 years, 4 months ago) by niro
File size: 171857 byte(s)
Fri May 23 17:35:37 2008 UTC (16 years, 4 months ago) by niro
File size: 171857 byte(s)
-using opensuse xen patchset, updated kernel configs
1 | diff -Naur linux-2.6.25/arch/x86/ia32/ia32entry-xen.S linux-2.6.25-xen/arch/x86/ia32/ia32entry-xen.S |
2 | --- linux-2.6.25/arch/x86/ia32/ia32entry-xen.S 2008-05-23 18:33:27.000000000 +0200 |
3 | +++ linux-2.6.25-xen/arch/x86/ia32/ia32entry-xen.S 2008-05-23 18:27:40.000000000 +0200 |
4 | @@ -105,7 +105,7 @@ |
5 | movl $VSYSCALL32_SYSEXIT,8(%rsp) |
6 | movq %rax,(%rsp) |
7 | cld |
8 | - SAVE_ARGS 0,0,0 |
9 | + SAVE_ARGS 0,0,1 |
10 | /* no need to do an access_ok check here because rbp has been |
11 | 32bit zero extended */ |
12 | 1: movl (%rbp),%r9d |
13 | @@ -244,7 +244,7 @@ |
14 | */ |
15 | |
16 | ENTRY(ia32_syscall) |
17 | - CFI_STARTPROC simple |
18 | + CFI_STARTPROC32 simple |
19 | CFI_SIGNAL_FRAME |
20 | CFI_DEF_CFA rsp,SS+8-RIP+16 |
21 | /*CFI_REL_OFFSET ss,SS-RIP+16*/ |
22 | @@ -280,6 +280,7 @@ |
23 | |
24 | ia32_tracesys: |
25 | SAVE_REST |
26 | + CLEAR_RREGS |
27 | movq $-ENOSYS,RAX(%rsp) /* really needed? */ |
28 | movq %rsp,%rdi /* &pt_regs -> arg1 */ |
29 | call syscall_trace_enter |
30 | @@ -476,7 +477,7 @@ |
31 | .quad sys_init_module |
32 | .quad sys_delete_module |
33 | .quad quiet_ni_syscall /* 130 get_kernel_syms */ |
34 | - .quad sys_quotactl |
35 | + .quad sys32_quotactl |
36 | .quad sys_getpgid |
37 | .quad sys_fchdir |
38 | .quad quiet_ni_syscall /* bdflush */ |
39 | @@ -669,4 +670,5 @@ |
40 | .quad compat_sys_signalfd |
41 | .quad compat_sys_timerfd |
42 | .quad sys_eventfd |
43 | + .quad sys32_fallocate |
44 | ia32_syscall_end: |
45 | diff -Naur linux-2.6.25/arch/x86/kernel/acpi/sleep_32-xen.c linux-2.6.25-xen/arch/x86/kernel/acpi/sleep_32-xen.c |
46 | --- linux-2.6.25/arch/x86/kernel/acpi/sleep_32-xen.c 2008-05-23 18:32:29.000000000 +0200 |
47 | +++ linux-2.6.25-xen/arch/x86/kernel/acpi/sleep_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
48 | @@ -15,7 +15,7 @@ |
49 | #ifndef CONFIG_ACPI_PV_SLEEP |
50 | /* address in low memory of the wakeup routine. */ |
51 | unsigned long acpi_wakeup_address = 0; |
52 | -unsigned long acpi_video_flags; |
53 | +unsigned long acpi_realmode_flags; |
54 | extern char wakeup_start, wakeup_end; |
55 | |
56 | extern unsigned long FASTCALL(acpi_copy_wakeup_routine(unsigned long)); |
57 | @@ -74,9 +74,11 @@ |
58 | { |
59 | while ((str != NULL) && (*str != '\0')) { |
60 | if (strncmp(str, "s3_bios", 7) == 0) |
61 | - acpi_video_flags = 1; |
62 | + acpi_realmode_flags |= 1; |
63 | if (strncmp(str, "s3_mode", 7) == 0) |
64 | - acpi_video_flags |= 2; |
65 | + acpi_realmode_flags |= 2; |
66 | + if (strncmp(str, "s3_beep", 7) == 0) |
67 | + acpi_realmode_flags |= 4; |
68 | str = strchr(str, ','); |
69 | if (str != NULL) |
70 | str += strspn(str, ", \t"); |
71 | @@ -86,9 +88,11 @@ |
72 | |
73 | __setup("acpi_sleep=", acpi_sleep_setup); |
74 | |
75 | +/* Ouch, we want to delete this. We already have better version in userspace, in |
76 | + s2ram from suspend.sf.net project */ |
77 | static __init int reset_videomode_after_s3(struct dmi_system_id *d) |
78 | { |
79 | - acpi_video_flags |= 2; |
80 | + acpi_realmode_flags |= 2; |
81 | return 0; |
82 | } |
83 | |
84 | diff -Naur linux-2.6.25/arch/x86/kernel/acpi/sleep_64-xen.c linux-2.6.25-xen/arch/x86/kernel/acpi/sleep_64-xen.c |
85 | --- linux-2.6.25/arch/x86/kernel/acpi/sleep_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
86 | +++ linux-2.6.25-xen/arch/x86/kernel/acpi/sleep_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
87 | @@ -51,12 +51,10 @@ |
88 | Low-Level Sleep Support |
89 | -------------------------------------------------------------------------- */ |
90 | |
91 | -#ifdef CONFIG_ACPI_SLEEP |
92 | - |
93 | #ifndef CONFIG_ACPI_PV_SLEEP |
94 | /* address in low memory of the wakeup routine. */ |
95 | unsigned long acpi_wakeup_address = 0; |
96 | -unsigned long acpi_video_flags; |
97 | +unsigned long acpi_realmode_flags; |
98 | extern char wakeup_start, wakeup_end; |
99 | |
100 | extern unsigned long acpi_copy_wakeup_routine(unsigned long); |
101 | @@ -109,9 +107,11 @@ |
102 | { |
103 | while ((str != NULL) && (*str != '\0')) { |
104 | if (strncmp(str, "s3_bios", 7) == 0) |
105 | - acpi_video_flags = 1; |
106 | + acpi_realmode_flags |= 1; |
107 | if (strncmp(str, "s3_mode", 7) == 0) |
108 | - acpi_video_flags |= 2; |
109 | + acpi_realmode_flags |= 2; |
110 | + if (strncmp(str, "s3_beep", 7) == 0) |
111 | + acpi_realmode_flags |= 4; |
112 | str = strchr(str, ','); |
113 | if (str != NULL) |
114 | str += strspn(str, ", \t"); |
115 | @@ -123,8 +123,6 @@ |
116 | __setup("acpi_sleep=", acpi_sleep_setup); |
117 | #endif /* CONFIG_ACPI_PV_SLEEP */ |
118 | |
119 | -#endif /*CONFIG_ACPI_SLEEP */ |
120 | - |
121 | void acpi_pci_link_exit(void) |
122 | { |
123 | } |
124 | diff -Naur linux-2.6.25/arch/x86/kernel/apic_64-xen.c linux-2.6.25-xen/arch/x86/kernel/apic_64-xen.c |
125 | --- linux-2.6.25/arch/x86/kernel/apic_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
126 | +++ linux-2.6.25-xen/arch/x86/kernel/apic_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
127 | @@ -50,7 +50,7 @@ |
128 | * holds up an irq slot - in excessive cases (when multiple |
129 | * unexpected vectors occur) that might lock up the APIC |
130 | * completely. |
131 | - * But don't ack when the APIC is disabled. -AK |
132 | + * But don't ack when the APIC is disabled. -AK |
133 | */ |
134 | if (!disable_apic) |
135 | ack_APIC_irq(); |
136 | @@ -132,20 +132,6 @@ |
137 | if (v & (1 << (SPURIOUS_APIC_VECTOR & 0x1f))) |
138 | ack_APIC_irq(); |
139 | |
140 | -#if 0 |
141 | - static unsigned long last_warning; |
142 | - static unsigned long skipped; |
143 | - |
144 | - /* see sw-dev-man vol 3, chapter 7.4.13.5 */ |
145 | - if (time_before(last_warning+30*HZ,jiffies)) { |
146 | - printk(KERN_INFO "spurious APIC interrupt on CPU#%d, %ld skipped.\n", |
147 | - smp_processor_id(), skipped); |
148 | - last_warning = jiffies; |
149 | - skipped = 0; |
150 | - } else { |
151 | - skipped++; |
152 | - } |
153 | -#endif |
154 | irq_exit(); |
155 | } |
156 | |
157 | @@ -177,7 +163,7 @@ |
158 | 7: Illegal register address |
159 | */ |
160 | printk (KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n", |
161 | - smp_processor_id(), v , v1); |
162 | + smp_processor_id(), v , v1); |
163 | irq_exit(); |
164 | } |
165 | |
166 | diff -Naur linux-2.6.25/arch/x86/kernel/asm-offsets_32.c linux-2.6.25-xen/arch/x86/kernel/asm-offsets_32.c |
167 | --- linux-2.6.25/arch/x86/kernel/asm-offsets_32.c 2008-05-23 18:33:26.000000000 +0200 |
168 | +++ linux-2.6.25-xen/arch/x86/kernel/asm-offsets_32.c 2008-05-23 18:27:40.000000000 +0200 |
169 | @@ -18,7 +18,9 @@ |
170 | #include <asm/bootparam.h> |
171 | #include <asm/elf.h> |
172 | |
173 | +#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN) |
174 | #include <xen/interface/xen.h> |
175 | +#endif |
176 | |
177 | #include <linux/lguest.h> |
178 | #include "../../../drivers/lguest/lg.h" |
179 | @@ -133,7 +135,7 @@ |
180 | OFFSET(PV_CPU_read_cr0, pv_cpu_ops, read_cr0); |
181 | #endif |
182 | |
183 | -#ifdef CONFIG_XEN |
184 | +#ifdef CONFIG_PARAVIRT_XEN |
185 | BLANK(); |
186 | OFFSET(XEN_vcpu_info_mask, vcpu_info, evtchn_upcall_mask); |
187 | OFFSET(XEN_vcpu_info_pending, vcpu_info, evtchn_upcall_pending); |
188 | diff -Naur linux-2.6.25/arch/x86/kernel/cpu/common-xen.c linux-2.6.25-xen/arch/x86/kernel/cpu/common-xen.c |
189 | --- linux-2.6.25/arch/x86/kernel/cpu/common-xen.c 2008-05-23 18:33:27.000000000 +0200 |
190 | +++ linux-2.6.25-xen/arch/x86/kernel/cpu/common-xen.c 2008-05-23 18:27:40.000000000 +0200 |
191 | @@ -360,6 +360,8 @@ |
192 | if ( xlvl >= 0x80000004 ) |
193 | get_model_name(c); /* Default name */ |
194 | } |
195 | + |
196 | + init_scattered_cpuid_features(c); |
197 | } |
198 | |
199 | early_intel_workaround(c); |
200 | @@ -611,7 +613,6 @@ |
201 | extern int amd_init_cpu(void); |
202 | extern int centaur_init_cpu(void); |
203 | extern int transmeta_init_cpu(void); |
204 | -extern int rise_init_cpu(void); |
205 | extern int nexgen_init_cpu(void); |
206 | extern int umc_init_cpu(void); |
207 | |
208 | @@ -623,7 +624,6 @@ |
209 | amd_init_cpu(); |
210 | centaur_init_cpu(); |
211 | transmeta_init_cpu(); |
212 | - rise_init_cpu(); |
213 | nexgen_init_cpu(); |
214 | umc_init_cpu(); |
215 | early_cpu_detect(); |
216 | diff -Naur linux-2.6.25/arch/x86/kernel/cpu/mtrr/main-xen.c linux-2.6.25-xen/arch/x86/kernel/cpu/mtrr/main-xen.c |
217 | --- linux-2.6.25/arch/x86/kernel/cpu/mtrr/main-xen.c 2008-05-23 18:33:27.000000000 +0200 |
218 | +++ linux-2.6.25-xen/arch/x86/kernel/cpu/mtrr/main-xen.c 2008-05-23 18:27:40.000000000 +0200 |
219 | @@ -167,7 +167,7 @@ |
220 | EXPORT_SYMBOL(mtrr_add); |
221 | EXPORT_SYMBOL(mtrr_del); |
222 | |
223 | -__init void mtrr_bp_init(void) |
224 | +void __init mtrr_bp_init(void) |
225 | { |
226 | } |
227 | |
228 | diff -Naur linux-2.6.25/arch/x86/kernel/e820_32-xen.c linux-2.6.25-xen/arch/x86/kernel/e820_32-xen.c |
229 | --- linux-2.6.25/arch/x86/kernel/e820_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
230 | +++ linux-2.6.25-xen/arch/x86/kernel/e820_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
231 | @@ -10,6 +10,7 @@ |
232 | #include <linux/efi.h> |
233 | #include <linux/pfn.h> |
234 | #include <linux/uaccess.h> |
235 | +#include <linux/suspend.h> |
236 | |
237 | #include <asm/pgtable.h> |
238 | #include <asm/page.h> |
239 | @@ -343,6 +344,37 @@ |
240 | |
241 | subsys_initcall(request_standard_resources); |
242 | |
243 | +#if defined(CONFIG_PM) && defined(CONFIG_HIBERNATION) |
244 | +/** |
245 | + * e820_mark_nosave_regions - Find the ranges of physical addresses that do not |
246 | + * correspond to e820 RAM areas and mark the corresponding pages as nosave for |
247 | + * hibernation. |
248 | + * |
249 | + * This function requires the e820 map to be sorted and without any |
250 | + * overlapping entries and assumes the first e820 area to be RAM. |
251 | + */ |
252 | +void __init e820_mark_nosave_regions(void) |
253 | +{ |
254 | + int i; |
255 | + unsigned long pfn; |
256 | + |
257 | + pfn = PFN_DOWN(e820.map[0].addr + e820.map[0].size); |
258 | + for (i = 1; i < e820.nr_map; i++) { |
259 | + struct e820entry *ei = &e820.map[i]; |
260 | + |
261 | + if (pfn < PFN_UP(ei->addr)) |
262 | + register_nosave_region(pfn, PFN_UP(ei->addr)); |
263 | + |
264 | + pfn = PFN_DOWN(ei->addr + ei->size); |
265 | + if (ei->type != E820_RAM) |
266 | + register_nosave_region(PFN_UP(ei->addr), pfn); |
267 | + |
268 | + if (pfn >= max_low_pfn) |
269 | + break; |
270 | + } |
271 | +} |
272 | +#endif |
273 | + |
274 | void __init add_memory_region(unsigned long long start, |
275 | unsigned long long size, int type) |
276 | { |
277 | @@ -789,7 +821,7 @@ |
278 | case E820_NVS: |
279 | printk("(ACPI NVS)\n"); |
280 | break; |
281 | - default: printk("type %lu\n", e820.map[i].type); |
282 | + default: printk("type %u\n", e820.map[i].type); |
283 | break; |
284 | } |
285 | } |
286 | diff -Naur linux-2.6.25/arch/x86/kernel/e820_64-xen.c linux-2.6.25-xen/arch/x86/kernel/e820_64-xen.c |
287 | --- linux-2.6.25/arch/x86/kernel/e820_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
288 | +++ linux-2.6.25-xen/arch/x86/kernel/e820_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
289 | @@ -224,37 +224,6 @@ |
290 | } |
291 | |
292 | /* |
293 | - * Find the hole size in the range. |
294 | - */ |
295 | -unsigned long __init e820_hole_size(unsigned long start, unsigned long end) |
296 | -{ |
297 | - unsigned long ram = 0; |
298 | - int i; |
299 | - |
300 | - for (i = 0; i < e820.nr_map; i++) { |
301 | - struct e820entry *ei = &e820.map[i]; |
302 | - unsigned long last, addr; |
303 | - |
304 | - if (ei->type != E820_RAM || |
305 | - ei->addr+ei->size <= start || |
306 | - ei->addr >= end) |
307 | - continue; |
308 | - |
309 | - addr = round_up(ei->addr, PAGE_SIZE); |
310 | - if (addr < start) |
311 | - addr = start; |
312 | - |
313 | - last = round_down(ei->addr + ei->size, PAGE_SIZE); |
314 | - if (last >= end) |
315 | - last = end; |
316 | - |
317 | - if (last > addr) |
318 | - ram += last - addr; |
319 | - } |
320 | - return ((end - start) - ram); |
321 | -} |
322 | - |
323 | -/* |
324 | * Mark e820 reserved areas as busy for the resource manager. |
325 | */ |
326 | void __init e820_reserve_resources(struct e820entry *e820, int nr_map) |
327 | @@ -327,47 +296,61 @@ |
328 | } |
329 | #endif |
330 | |
331 | +/* |
332 | + * Finds an active region in the address range from start_pfn to end_pfn and |
333 | + * returns its range in ei_startpfn and ei_endpfn for the e820 entry. |
334 | + */ |
335 | +static int __init e820_find_active_region(const struct e820entry *ei, |
336 | + unsigned long start_pfn, |
337 | + unsigned long end_pfn, |
338 | + unsigned long *ei_startpfn, |
339 | + unsigned long *ei_endpfn) |
340 | +{ |
341 | + *ei_startpfn = round_up(ei->addr, PAGE_SIZE) >> PAGE_SHIFT; |
342 | + *ei_endpfn = round_down(ei->addr + ei->size, PAGE_SIZE) >> PAGE_SHIFT; |
343 | + |
344 | + /* Skip map entries smaller than a page */ |
345 | + if (*ei_startpfn >= *ei_endpfn) |
346 | + return 0; |
347 | + |
348 | + /* Check if end_pfn_map should be updated */ |
349 | + if (ei->type != E820_RAM && *ei_endpfn > end_pfn_map) |
350 | + end_pfn_map = *ei_endpfn; |
351 | + |
352 | + /* Skip if map is outside the node */ |
353 | + if (ei->type != E820_RAM || *ei_endpfn <= start_pfn || |
354 | + *ei_startpfn >= end_pfn) |
355 | + return 0; |
356 | + |
357 | + /* Check for overlaps */ |
358 | + if (*ei_startpfn < start_pfn) |
359 | + *ei_startpfn = start_pfn; |
360 | + if (*ei_endpfn > end_pfn) |
361 | + *ei_endpfn = end_pfn; |
362 | + |
363 | + /* Obey end_user_pfn to save on memmap */ |
364 | + if (*ei_startpfn >= end_user_pfn) |
365 | + return 0; |
366 | + if (*ei_endpfn > end_user_pfn) |
367 | + *ei_endpfn = end_user_pfn; |
368 | + |
369 | + return 1; |
370 | +} |
371 | + |
372 | /* Walk the e820 map and register active regions within a node */ |
373 | void __init |
374 | e820_register_active_regions(int nid, unsigned long start_pfn, |
375 | unsigned long end_pfn) |
376 | { |
377 | + unsigned long ei_startpfn; |
378 | + unsigned long ei_endpfn; |
379 | int i; |
380 | - unsigned long ei_startpfn, ei_endpfn; |
381 | - for (i = 0; i < e820.nr_map; i++) { |
382 | - struct e820entry *ei = &e820.map[i]; |
383 | - ei_startpfn = round_up(ei->addr, PAGE_SIZE) >> PAGE_SHIFT; |
384 | - ei_endpfn = round_down(ei->addr + ei->size, PAGE_SIZE) |
385 | - >> PAGE_SHIFT; |
386 | - |
387 | - /* Skip map entries smaller than a page */ |
388 | - if (ei_startpfn >= ei_endpfn) |
389 | - continue; |
390 | - |
391 | - /* Check if end_pfn_map should be updated */ |
392 | - if (ei->type != E820_RAM && ei_endpfn > end_pfn_map) |
393 | - end_pfn_map = ei_endpfn; |
394 | - |
395 | - /* Skip if map is outside the node */ |
396 | - if (ei->type != E820_RAM || |
397 | - ei_endpfn <= start_pfn || |
398 | - ei_startpfn >= end_pfn) |
399 | - continue; |
400 | - |
401 | - /* Check for overlaps */ |
402 | - if (ei_startpfn < start_pfn) |
403 | - ei_startpfn = start_pfn; |
404 | - if (ei_endpfn > end_pfn) |
405 | - ei_endpfn = end_pfn; |
406 | - |
407 | - /* Obey end_user_pfn to save on memmap */ |
408 | - if (ei_startpfn >= end_user_pfn) |
409 | - continue; |
410 | - if (ei_endpfn > end_user_pfn) |
411 | - ei_endpfn = end_user_pfn; |
412 | |
413 | - add_active_range(nid, ei_startpfn, ei_endpfn); |
414 | - } |
415 | + for (i = 0; i < e820.nr_map; i++) |
416 | + if (e820_find_active_region(&e820.map[i], |
417 | + start_pfn, end_pfn, |
418 | + &ei_startpfn, &ei_endpfn)) |
419 | + add_active_range(nid, ei_startpfn, ei_endpfn); |
420 | } |
421 | |
422 | /* |
423 | @@ -388,12 +371,35 @@ |
424 | e820.nr_map++; |
425 | } |
426 | |
427 | +/* |
428 | + * Find the hole size (in bytes) in the memory range. |
429 | + * @start: starting address of the memory range to scan |
430 | + * @end: ending address of the memory range to scan |
431 | + */ |
432 | +unsigned long __init e820_hole_size(unsigned long start, unsigned long end) |
433 | +{ |
434 | + unsigned long start_pfn = start >> PAGE_SHIFT; |
435 | + unsigned long end_pfn = end >> PAGE_SHIFT; |
436 | + unsigned long ei_startpfn; |
437 | + unsigned long ei_endpfn; |
438 | + unsigned long ram = 0; |
439 | + int i; |
440 | + |
441 | + for (i = 0; i < e820.nr_map; i++) { |
442 | + if (e820_find_active_region(&e820.map[i], |
443 | + start_pfn, end_pfn, |
444 | + &ei_startpfn, &ei_endpfn)) |
445 | + ram += ei_endpfn - ei_startpfn; |
446 | + } |
447 | + return end - start - (ram << PAGE_SHIFT); |
448 | +} |
449 | + |
450 | void __init e820_print_map(char *who) |
451 | { |
452 | int i; |
453 | |
454 | for (i = 0; i < e820.nr_map; i++) { |
455 | - printk(" %s: %016Lx - %016Lx ", who, |
456 | + printk(KERN_INFO " %s: %016Lx - %016Lx ", who, |
457 | (unsigned long long) e820.map[i].addr, |
458 | (unsigned long long) (e820.map[i].addr + e820.map[i].size)); |
459 | switch (e820.map[i].type) { |
460 | diff -Naur linux-2.6.25/arch/x86/kernel/early_printk-xen.c linux-2.6.25-xen/arch/x86/kernel/early_printk-xen.c |
461 | --- linux-2.6.25/arch/x86/kernel/early_printk-xen.c 2008-05-23 18:33:27.000000000 +0200 |
462 | +++ linux-2.6.25-xen/arch/x86/kernel/early_printk-xen.c 2008-05-23 18:27:40.000000000 +0200 |
463 | @@ -174,6 +174,7 @@ |
464 | * mappings. Someone should fix this for domain 0. For now, use fake serial. |
465 | */ |
466 | #define early_vga_console early_serial_console |
467 | +#define xenboot_console early_serial_console |
468 | |
469 | #endif |
470 | |
471 | @@ -261,20 +262,22 @@ |
472 | } else if (!strncmp(buf, "ttyS", 4)) { |
473 | early_serial_init(buf); |
474 | early_console = &early_serial_console; |
475 | - } else if (!strncmp(buf, "vga", 3) |
476 | + } else if (!strncmp(buf, "vga", 3)) { |
477 | #ifndef CONFIG_XEN |
478 | && SCREEN_INFO.orig_video_isVGA == 1) { |
479 | max_xpos = SCREEN_INFO.orig_video_cols; |
480 | max_ypos = SCREEN_INFO.orig_video_lines; |
481 | current_ypos = SCREEN_INFO.orig_y; |
482 | -#else |
483 | - || !strncmp(buf, "xen", 3)) { |
484 | #endif |
485 | early_console = &early_vga_console; |
486 | } else if (!strncmp(buf, "simnow", 6)) { |
487 | simnow_init(buf + 6); |
488 | early_console = &simnow_console; |
489 | keep_early = 1; |
490 | +#ifdef CONFIG_XEN |
491 | + } else if (!strncmp(buf, "xen", 3)) { |
492 | + early_console = &xenboot_console; |
493 | +#endif |
494 | } |
495 | |
496 | if (keep_early) |
497 | diff -Naur linux-2.6.25/arch/x86/kernel/entry_32.S linux-2.6.25-xen/arch/x86/kernel/entry_32.S |
498 | --- linux-2.6.25/arch/x86/kernel/entry_32.S 2008-05-23 18:32:54.000000000 +0200 |
499 | +++ linux-2.6.25-xen/arch/x86/kernel/entry_32.S 2008-05-23 18:27:40.000000000 +0200 |
500 | @@ -1022,7 +1022,7 @@ |
501 | CFI_ENDPROC |
502 | ENDPROC(kernel_thread_helper) |
503 | |
504 | -#ifdef CONFIG_XEN |
505 | +#ifdef CONFIG_PARAVIRT_XEN |
506 | ENTRY(xen_hypervisor_callback) |
507 | CFI_STARTPROC |
508 | pushl $0 |
509 | @@ -1105,7 +1105,7 @@ |
510 | .previous |
511 | ENDPROC(xen_failsafe_callback) |
512 | |
513 | -#endif /* CONFIG_XEN */ |
514 | +#endif /* CONFIG_PARAVIRT_XEN */ |
515 | |
516 | .section .rodata,"a" |
517 | #include "syscall_table_32.S" |
518 | diff -Naur linux-2.6.25/arch/x86/kernel/entry_32-xen.S linux-2.6.25-xen/arch/x86/kernel/entry_32-xen.S |
519 | --- linux-2.6.25/arch/x86/kernel/entry_32-xen.S 2008-05-23 18:33:27.000000000 +0200 |
520 | +++ linux-2.6.25-xen/arch/x86/kernel/entry_32-xen.S 2008-05-23 18:27:40.000000000 +0200 |
521 | @@ -452,9 +452,6 @@ |
522 | 1: INTERRUPT_RETURN |
523 | .section .fixup,"ax" |
524 | iret_exc: |
525 | -#ifndef CONFIG_XEN |
526 | - ENABLE_INTERRUPTS(CLBR_NONE) |
527 | -#endif |
528 | pushl $0 # no error code |
529 | pushl $do_iret_error |
530 | jmp error_code |
531 | diff -Naur linux-2.6.25/arch/x86/kernel/entry_64-xen.S linux-2.6.25-xen/arch/x86/kernel/entry_64-xen.S |
532 | --- linux-2.6.25/arch/x86/kernel/entry_64-xen.S 2008-05-23 18:33:27.000000000 +0200 |
533 | +++ linux-2.6.25-xen/arch/x86/kernel/entry_64-xen.S 2008-05-23 18:27:40.000000000 +0200 |
534 | @@ -310,7 +310,7 @@ |
535 | TRACE_IRQS_ON |
536 | /* sti */ |
537 | XEN_UNBLOCK_EVENTS(%rsi) |
538 | - testl $(_TIF_SIGPENDING|_TIF_NOTIFY_RESUME|_TIF_SINGLESTEP),%edx |
539 | + testl $(_TIF_SIGPENDING|_TIF_SINGLESTEP|_TIF_MCE_NOTIFY),%edx |
540 | jz 1f |
541 | |
542 | /* Really a signal */ |
543 | @@ -409,7 +409,7 @@ |
544 | jmp int_restore_rest |
545 | |
546 | int_signal: |
547 | - testl $(_TIF_NOTIFY_RESUME|_TIF_SIGPENDING|_TIF_SINGLESTEP),%edx |
548 | + testl $(_TIF_SIGPENDING|_TIF_SINGLESTEP|_TIF_MCE_NOTIFY),%edx |
549 | jz 1f |
550 | movq %rsp,%rdi # &ptregs -> arg1 |
551 | xorl %esi,%esi # oldset -> arg2 |
552 | @@ -552,7 +552,7 @@ |
553 | jmp retint_check |
554 | |
555 | retint_signal: |
556 | - testl $(_TIF_SIGPENDING|_TIF_NOTIFY_RESUME|_TIF_SINGLESTEP),%edx |
557 | + testl $(_TIF_SIGPENDING|_TIF_SINGLESTEP|_TIF_MCE_NOTIFY),%edx |
558 | jz retint_restore_args |
559 | TRACE_IRQS_ON |
560 | XEN_UNBLOCK_EVENTS(%rsi) |
561 | diff -Naur linux-2.6.25/arch/x86/kernel/head_32-xen.S linux-2.6.25-xen/arch/x86/kernel/head_32-xen.S |
562 | --- linux-2.6.25/arch/x86/kernel/head_32-xen.S 2008-05-23 18:33:27.000000000 +0200 |
563 | +++ linux-2.6.25-xen/arch/x86/kernel/head_32-xen.S 2008-05-23 18:27:40.000000000 +0200 |
564 | @@ -86,7 +86,10 @@ |
565 | /* |
566 | * BSS section |
567 | */ |
568 | -.section ".bss.page_aligned","w" |
569 | +.section ".bss.page_aligned","wa" |
570 | + .align PAGE_SIZE_asm |
571 | +ENTRY(swapper_pg_pmd) |
572 | + .fill 1024,4,0 |
573 | ENTRY(empty_zero_page) |
574 | .fill 4096,1,0 |
575 | |
576 | @@ -136,25 +139,25 @@ |
577 | #endif /* CONFIG_XEN_COMPAT <= 0x030002 */ |
578 | |
579 | |
580 | - ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS, .asciz, "linux") |
581 | - ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION, .asciz, "2.6") |
582 | - ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION, .asciz, "xen-3.0") |
583 | - ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE, .long, __PAGE_OFFSET) |
584 | + ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS, .asciz "linux") |
585 | + ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION, .asciz "2.6") |
586 | + ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION, .asciz "xen-3.0") |
587 | + ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE, .long __PAGE_OFFSET) |
588 | #if CONFIG_XEN_COMPAT <= 0x030002 |
589 | - ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .long, __PAGE_OFFSET) |
590 | + ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .long __PAGE_OFFSET) |
591 | #else |
592 | - ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .long, 0) |
593 | + ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .long 0) |
594 | #endif |
595 | - ELFNOTE(Xen, XEN_ELFNOTE_ENTRY, .long, startup_32) |
596 | - ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .long, hypercall_page) |
597 | - ELFNOTE(Xen, XEN_ELFNOTE_HV_START_LOW, .long, HYPERVISOR_VIRT_START) |
598 | - ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz, "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel") |
599 | + ELFNOTE(Xen, XEN_ELFNOTE_ENTRY, .long startup_32) |
600 | + ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .long hypercall_page) |
601 | + ELFNOTE(Xen, XEN_ELFNOTE_HV_START_LOW, .long HYPERVISOR_VIRT_START) |
602 | + ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel") |
603 | #ifdef CONFIG_X86_PAE |
604 | - ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE, .asciz, "yes") |
605 | - ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .quad, _PAGE_PRESENT,_PAGE_PRESENT) |
606 | + ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE, .asciz "yes") |
607 | + ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .quad _PAGE_PRESENT, _PAGE_PRESENT) |
608 | #else |
609 | - ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE, .asciz, "no") |
610 | - ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .long, _PAGE_PRESENT,_PAGE_PRESENT) |
611 | + ELFNOTE(Xen, XEN_ELFNOTE_PAE_MODE, .asciz "no") |
612 | + ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .long _PAGE_PRESENT, _PAGE_PRESENT) |
613 | #endif |
614 | - ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz, "generic") |
615 | - ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long, 1) |
616 | + ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz "generic") |
617 | + ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long 1) |
618 | diff -Naur linux-2.6.25/arch/x86/kernel/head64-xen.c linux-2.6.25-xen/arch/x86/kernel/head64-xen.c |
619 | --- linux-2.6.25/arch/x86/kernel/head64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
620 | +++ linux-2.6.25-xen/arch/x86/kernel/head64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
621 | @@ -87,7 +87,7 @@ |
622 | unsigned long machine_to_phys_nr_ents; |
623 | int i; |
624 | |
625 | - setup_xen_features(); |
626 | + xen_setup_features(); |
627 | |
628 | xen_start_info = (struct start_info *)real_mode_data; |
629 | if (!xen_feature(XENFEAT_auto_translated_physmap)) |
630 | diff -Naur linux-2.6.25/arch/x86/kernel/head_64-xen.S linux-2.6.25-xen/arch/x86/kernel/head_64-xen.S |
631 | --- linux-2.6.25/arch/x86/kernel/head_64-xen.S 2008-05-23 18:33:27.000000000 +0200 |
632 | +++ linux-2.6.25-xen/arch/x86/kernel/head_64-xen.S 2008-05-23 18:27:40.000000000 +0200 |
633 | @@ -23,7 +23,7 @@ |
634 | #include <asm/dwarf2.h> |
635 | #include <xen/interface/elfnote.h> |
636 | |
637 | - .section .bootstrap.text, "ax", @progbits |
638 | + .section .text.head, "ax", @progbits |
639 | .code64 |
640 | .globl startup_64 |
641 | startup_64: |
642 | @@ -47,7 +47,7 @@ |
643 | |
644 | #define NEXT_PAGE(name) \ |
645 | .balign PAGE_SIZE; \ |
646 | - phys_##name = . - .bootstrap.text; \ |
647 | + phys_##name = . - .text.head; \ |
648 | ENTRY(name) |
649 | |
650 | NEXT_PAGE(init_level4_pgt) |
651 | @@ -75,6 +75,12 @@ |
652 | NEXT_PAGE(level2_kernel_pgt) |
653 | .fill 512,8,0 |
654 | |
655 | +NEXT_PAGE(level2_fixmap_pgt) |
656 | + .fill 512,8,0 |
657 | + |
658 | +NEXT_PAGE(level1_fixmap_pgt) |
659 | + .fill 512,8,0 |
660 | + |
661 | NEXT_PAGE(hypercall_page) |
662 | CFI_STARTPROC |
663 | .rept 0x1000 / 0x20 |
664 | @@ -189,18 +195,18 @@ |
665 | .byte 0 |
666 | #endif /* CONFIG_XEN_COMPAT <= 0x030002 */ |
667 | |
668 | - ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS, .asciz, "linux") |
669 | - ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION, .asciz, "2.6") |
670 | - ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION, .asciz, "xen-3.0") |
671 | - ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE, .quad, __START_KERNEL_map) |
672 | + ELFNOTE(Xen, XEN_ELFNOTE_GUEST_OS, .asciz "linux") |
673 | + ELFNOTE(Xen, XEN_ELFNOTE_GUEST_VERSION, .asciz "2.6") |
674 | + ELFNOTE(Xen, XEN_ELFNOTE_XEN_VERSION, .asciz "xen-3.0") |
675 | + ELFNOTE(Xen, XEN_ELFNOTE_VIRT_BASE, .quad __START_KERNEL_map) |
676 | #if CONFIG_XEN_COMPAT <= 0x030002 |
677 | - ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .quad, __START_KERNEL_map) |
678 | + ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .quad __START_KERNEL_map) |
679 | #else |
680 | - ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .quad, 0) |
681 | + ELFNOTE(Xen, XEN_ELFNOTE_PADDR_OFFSET, .quad 0) |
682 | #endif |
683 | - ELFNOTE(Xen, XEN_ELFNOTE_ENTRY, .quad, startup_64) |
684 | - ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .quad, hypercall_page) |
685 | - ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .quad, _PAGE_PRESENT,_PAGE_PRESENT) |
686 | - ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz, "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel") |
687 | - ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz, "generic") |
688 | - ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long, 1) |
689 | + ELFNOTE(Xen, XEN_ELFNOTE_ENTRY, .quad startup_64) |
690 | + ELFNOTE(Xen, XEN_ELFNOTE_HYPERCALL_PAGE, .quad hypercall_page) |
691 | + ELFNOTE(Xen, XEN_ELFNOTE_L1_MFN_VALID, .quad _PAGE_PRESENT, _PAGE_PRESENT) |
692 | + ELFNOTE(Xen, XEN_ELFNOTE_FEATURES, .asciz "writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel") |
693 | + ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz "generic") |
694 | + ELFNOTE(Xen, XEN_ELFNOTE_SUSPEND_CANCEL, .long 1) |
695 | diff -Naur linux-2.6.25/arch/x86/kernel/init_task-xen.c linux-2.6.25-xen/arch/x86/kernel/init_task-xen.c |
696 | --- linux-2.6.25/arch/x86/kernel/init_task-xen.c 2008-05-23 18:32:29.000000000 +0200 |
697 | +++ linux-2.6.25-xen/arch/x86/kernel/init_task-xen.c 2008-05-23 18:27:40.000000000 +0200 |
698 | @@ -46,6 +46,6 @@ |
699 | * per-CPU TSS segments. Threads are completely 'soft' on Linux, |
700 | * no more per-task TSS's. |
701 | */ |
702 | -DEFINE_PER_CPU(struct tss_struct, init_tss) ____cacheline_internodealigned_in_smp = INIT_TSS; |
703 | +DEFINE_PER_CPU_SHARED_ALIGNED(struct tss_struct, init_tss) = INIT_TSS; |
704 | #endif |
705 | |
706 | diff -Naur linux-2.6.25/arch/x86/kernel/io_apic_32-xen.c linux-2.6.25-xen/arch/x86/kernel/io_apic_32-xen.c |
707 | --- linux-2.6.25/arch/x86/kernel/io_apic_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
708 | +++ linux-2.6.25-xen/arch/x86/kernel/io_apic_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
709 | @@ -397,14 +397,6 @@ |
710 | # include <linux/slab.h> /* kmalloc() */ |
711 | # include <linux/timer.h> /* time_after() */ |
712 | |
713 | -#ifdef CONFIG_BALANCED_IRQ_DEBUG |
714 | -# define TDprintk(x...) do { printk("<%ld:%s:%d>: ", jiffies, __FILE__, __LINE__); printk(x); } while (0) |
715 | -# define Dprintk(x...) do { TDprintk(x); } while (0) |
716 | -# else |
717 | -# define TDprintk(x...) |
718 | -# define Dprintk(x...) |
719 | -# endif |
720 | - |
721 | #define IRQBALANCE_CHECK_ARCH -999 |
722 | #define MAX_BALANCED_IRQ_INTERVAL (5*HZ) |
723 | #define MIN_BALANCED_IRQ_INTERVAL (HZ/2) |
724 | @@ -487,7 +479,7 @@ |
725 | static inline void rotate_irqs_among_cpus(unsigned long useful_load_threshold) |
726 | { |
727 | int i, j; |
728 | - Dprintk("Rotating IRQs among CPUs.\n"); |
729 | + |
730 | for_each_online_cpu(i) { |
731 | for (j = 0; j < NR_IRQS; j++) { |
732 | if (!irq_desc[j].action) |
733 | @@ -604,19 +596,11 @@ |
734 | max_loaded = tmp_loaded; /* processor */ |
735 | imbalance = (max_cpu_irq - min_cpu_irq) / 2; |
736 | |
737 | - Dprintk("max_loaded cpu = %d\n", max_loaded); |
738 | - Dprintk("min_loaded cpu = %d\n", min_loaded); |
739 | - Dprintk("max_cpu_irq load = %ld\n", max_cpu_irq); |
740 | - Dprintk("min_cpu_irq load = %ld\n", min_cpu_irq); |
741 | - Dprintk("load imbalance = %lu\n", imbalance); |
742 | - |
743 | /* if imbalance is less than approx 10% of max load, then |
744 | * observe diminishing returns action. - quit |
745 | */ |
746 | - if (imbalance < (max_cpu_irq >> 3)) { |
747 | - Dprintk("Imbalance too trivial\n"); |
748 | + if (imbalance < (max_cpu_irq >> 3)) |
749 | goto not_worth_the_effort; |
750 | - } |
751 | |
752 | tryanotherirq: |
753 | /* if we select an IRQ to move that can't go where we want, then |
754 | @@ -673,9 +657,6 @@ |
755 | cpus_and(tmp, target_cpu_mask, allowed_mask); |
756 | |
757 | if (!cpus_empty(tmp)) { |
758 | - |
759 | - Dprintk("irq = %d moved to cpu = %d\n", |
760 | - selected_irq, min_loaded); |
761 | /* mark for change destination */ |
762 | set_pending_irq(selected_irq, cpumask_of_cpu(min_loaded)); |
763 | |
764 | @@ -695,7 +676,6 @@ |
765 | */ |
766 | balanced_irq_interval = min((long)MAX_BALANCED_IRQ_INTERVAL, |
767 | balanced_irq_interval + BALANCED_IRQ_MORE_DELTA); |
768 | - Dprintk("IRQ worth rotating not found\n"); |
769 | return; |
770 | } |
771 | |
772 | @@ -711,6 +691,7 @@ |
773 | set_pending_irq(i, cpumask_of_cpu(0)); |
774 | } |
775 | |
776 | + set_freezable(); |
777 | for ( ; ; ) { |
778 | time_remaining = schedule_timeout_interruptible(time_remaining); |
779 | try_to_freeze(); |
780 | @@ -820,14 +801,6 @@ |
781 | static int pirqs_enabled; |
782 | int skip_ioapic_setup; |
783 | |
784 | -static int __init ioapic_setup(char *str) |
785 | -{ |
786 | - skip_ioapic_setup = 1; |
787 | - return 1; |
788 | -} |
789 | - |
790 | -__setup("noapic", ioapic_setup); |
791 | - |
792 | static int __init ioapic_pirq_setup(char *str) |
793 | { |
794 | int i, max; |
795 | @@ -1313,12 +1286,15 @@ |
796 | static void ioapic_register_intr(int irq, int vector, unsigned long trigger) |
797 | { |
798 | if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || |
799 | - trigger == IOAPIC_LEVEL) |
800 | + trigger == IOAPIC_LEVEL) { |
801 | + irq_desc[irq].status |= IRQ_LEVEL; |
802 | set_irq_chip_and_handler_name(irq, &ioapic_chip, |
803 | handle_fasteoi_irq, "fasteoi"); |
804 | - else |
805 | + } else { |
806 | + irq_desc[irq].status &= ~IRQ_LEVEL; |
807 | set_irq_chip_and_handler_name(irq, &ioapic_chip, |
808 | handle_edge_irq, "edge"); |
809 | + } |
810 | set_intr_gate(vector, interrupt[irq]); |
811 | } |
812 | #else |
813 | @@ -1943,7 +1919,7 @@ |
814 | * - if this function detects that timer IRQs are defunct, then we fall |
815 | * back to ISA timer IRQs |
816 | */ |
817 | -int __init timer_irq_works(void) |
818 | +static int __init timer_irq_works(void) |
819 | { |
820 | unsigned long t1 = jiffies; |
821 | |
822 | diff -Naur linux-2.6.25/arch/x86/kernel/io_apic_64-xen.c linux-2.6.25-xen/arch/x86/kernel/io_apic_64-xen.c |
823 | --- linux-2.6.25/arch/x86/kernel/io_apic_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
824 | +++ linux-2.6.25-xen/arch/x86/kernel/io_apic_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
825 | @@ -164,7 +164,9 @@ |
826 | #endif |
827 | } |
828 | |
829 | -#ifndef CONFIG_XEN |
830 | +#ifdef CONFIG_XEN |
831 | +#define io_apic_modify io_apic_write |
832 | +#else |
833 | /* |
834 | * Re-write a value: to be used for read-modify-write |
835 | * cycles where the read already set up the index register. |
836 | @@ -174,8 +176,32 @@ |
837 | struct io_apic __iomem *io_apic = io_apic_base(apic); |
838 | writel(value, &io_apic->data); |
839 | } |
840 | -#else |
841 | -#define io_apic_modify io_apic_write |
842 | + |
843 | +static int io_apic_level_ack_pending(unsigned int irq) |
844 | +{ |
845 | + struct irq_pin_list *entry; |
846 | + unsigned long flags; |
847 | + int pending = 0; |
848 | + |
849 | + spin_lock_irqsave(&ioapic_lock, flags); |
850 | + entry = irq_2_pin + irq; |
851 | + for (;;) { |
852 | + unsigned int reg; |
853 | + int pin; |
854 | + |
855 | + pin = entry->pin; |
856 | + if (pin == -1) |
857 | + break; |
858 | + reg = io_apic_read(entry->apic, 0x10 + pin*2); |
859 | + /* Is the remote IRR bit set? */ |
860 | + pending |= (reg >> 14) & 1; |
861 | + if (!entry->next) |
862 | + break; |
863 | + entry = irq_2_pin + entry->next; |
864 | + } |
865 | + spin_unlock_irqrestore(&ioapic_lock, flags); |
866 | + return pending; |
867 | +} |
868 | #endif |
869 | |
870 | /* |
871 | @@ -403,14 +429,12 @@ |
872 | int skip_ioapic_setup; |
873 | int ioapic_force; |
874 | |
875 | -/* dummy parsing: see setup.c */ |
876 | - |
877 | -static int __init disable_ioapic_setup(char *str) |
878 | +static int __init parse_noapic(char *str) |
879 | { |
880 | - skip_ioapic_setup = 1; |
881 | + disable_ioapic_setup(); |
882 | return 0; |
883 | } |
884 | -early_param("noapic", disable_ioapic_setup); |
885 | +early_param("noapic", parse_noapic); |
886 | |
887 | /* Actually the next is obsolete, but keep it for paranoid reasons -AK */ |
888 | static int __init disable_timer_pin_setup(char *arg) |
889 | @@ -758,12 +782,15 @@ |
890 | |
891 | static void ioapic_register_intr(int irq, unsigned long trigger) |
892 | { |
893 | - if (trigger) |
894 | + if (trigger) { |
895 | + irq_desc[irq].status |= IRQ_LEVEL; |
896 | set_irq_chip_and_handler_name(irq, &ioapic_chip, |
897 | handle_fasteoi_irq, "fasteoi"); |
898 | - else |
899 | + } else { |
900 | + irq_desc[irq].status &= ~IRQ_LEVEL; |
901 | set_irq_chip_and_handler_name(irq, &ioapic_chip, |
902 | handle_edge_irq, "edge"); |
903 | + } |
904 | } |
905 | #else |
906 | #define ioapic_register_intr(irq,trigger) ((void)0) |
907 | @@ -1412,9 +1439,37 @@ |
908 | ack_APIC_irq(); |
909 | |
910 | /* Now we can move and renable the irq */ |
911 | - move_masked_irq(irq); |
912 | - if (unlikely(do_unmask_irq)) |
913 | + if (unlikely(do_unmask_irq)) { |
914 | + /* Only migrate the irq if the ack has been received. |
915 | + * |
916 | + * On rare occasions the broadcast level triggered ack gets |
917 | + * delayed going to ioapics, and if we reprogram the |
918 | + * vector while Remote IRR is still set the irq will never |
919 | + * fire again. |
920 | + * |
921 | + * To prevent this scenario we read the Remote IRR bit |
922 | + * of the ioapic. This has two effects. |
923 | + * - On any sane system the read of the ioapic will |
924 | + * flush writes (and acks) going to the ioapic from |
925 | + * this cpu. |
926 | + * - We get to see if the ACK has actually been delivered. |
927 | + * |
928 | + * Based on failed experiments of reprogramming the |
929 | + * ioapic entry from outside of irq context starting |
930 | + * with masking the ioapic entry and then polling until |
931 | + * Remote IRR was clear before reprogramming the |
932 | + * ioapic I don't trust the Remote IRR bit to be |
933 | + * completey accurate. |
934 | + * |
935 | + * However there appears to be no other way to plug |
936 | + * this race, so if the Remote IRR bit is not |
937 | + * accurate and is causing problems then it is a hardware bug |
938 | + * and you can go talk to the chipset vendor about it. |
939 | + */ |
940 | + if (!io_apic_level_ack_pending(irq)) |
941 | + move_masked_irq(irq); |
942 | unmask_IO_APIC_irq(irq); |
943 | + } |
944 | } |
945 | |
946 | static struct irq_chip ioapic_chip __read_mostly = { |
947 | diff -Naur linux-2.6.25/arch/x86/kernel/irq_32-xen.c linux-2.6.25-xen/arch/x86/kernel/irq_32-xen.c |
948 | --- linux-2.6.25/arch/x86/kernel/irq_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
949 | +++ linux-2.6.25-xen/arch/x86/kernel/irq_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
950 | @@ -21,7 +21,7 @@ |
951 | #include <asm/apic.h> |
952 | #include <asm/uaccess.h> |
953 | |
954 | -DEFINE_PER_CPU(irq_cpustat_t, irq_stat) ____cacheline_internodealigned_in_smp; |
955 | +DEFINE_PER_CPU_SHARED_ALIGNED(irq_cpustat_t, irq_stat); |
956 | EXPORT_PER_CPU_SYMBOL(irq_stat); |
957 | |
958 | DEFINE_PER_CPU(struct pt_regs *, irq_regs); |
959 | @@ -149,15 +149,11 @@ |
960 | |
961 | #ifdef CONFIG_4KSTACKS |
962 | |
963 | -/* |
964 | - * These should really be __section__(".bss.page_aligned") as well, but |
965 | - * gcc's 3.0 and earlier don't handle that correctly. |
966 | - */ |
967 | static char softirq_stack[NR_CPUS * THREAD_SIZE] |
968 | - __attribute__((__aligned__(THREAD_SIZE))); |
969 | + __attribute__((__section__(".bss.page_aligned"))); |
970 | |
971 | static char hardirq_stack[NR_CPUS * THREAD_SIZE] |
972 | - __attribute__((__aligned__(THREAD_SIZE))); |
973 | + __attribute__((__section__(".bss.page_aligned"))); |
974 | |
975 | /* |
976 | * allocate per-cpu stacks for hardirq and for softirq processing |
977 | diff -Naur linux-2.6.25/arch/x86/kernel/ldt_64-xen.c linux-2.6.25-xen/arch/x86/kernel/ldt_64-xen.c |
978 | --- linux-2.6.25/arch/x86/kernel/ldt_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
979 | +++ linux-2.6.25-xen/arch/x86/kernel/ldt_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
980 | @@ -114,6 +114,8 @@ |
981 | memset(&mm->context, 0, sizeof(mm->context)); |
982 | init_MUTEX(&mm->context.sem); |
983 | old_mm = current->mm; |
984 | + if (old_mm) |
985 | + mm->context.vdso = old_mm->context.vdso; |
986 | if (old_mm && old_mm->context.size > 0) { |
987 | down(&old_mm->context.sem); |
988 | retval = copy_ldt(&mm->context, &old_mm->context); |
989 | diff -Naur linux-2.6.25/arch/x86/kernel/microcode-xen.c linux-2.6.25-xen/arch/x86/kernel/microcode-xen.c |
990 | --- linux-2.6.25/arch/x86/kernel/microcode-xen.c 2008-05-23 18:33:27.000000000 +0200 |
991 | +++ linux-2.6.25-xen/arch/x86/kernel/microcode-xen.c 2008-05-23 18:27:40.000000000 +0200 |
992 | @@ -33,6 +33,7 @@ |
993 | #include <linux/miscdevice.h> |
994 | #include <linux/spinlock.h> |
995 | #include <linux/mm.h> |
996 | +#include <linux/fs.h> |
997 | #include <linux/mutex.h> |
998 | #include <linux/cpu.h> |
999 | #include <linux/firmware.h> |
1000 | diff -Naur linux-2.6.25/arch/x86/kernel/mpparse_64-xen.c linux-2.6.25-xen/arch/x86/kernel/mpparse_64-xen.c |
1001 | --- linux-2.6.25/arch/x86/kernel/mpparse_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1002 | +++ linux-2.6.25-xen/arch/x86/kernel/mpparse_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1003 | @@ -32,7 +32,6 @@ |
1004 | |
1005 | /* Have we found an MP table */ |
1006 | int smp_found_config; |
1007 | -unsigned int __initdata maxcpus = NR_CPUS; |
1008 | |
1009 | /* |
1010 | * Various Linux-internal data structures created from the |
1011 | @@ -657,6 +656,20 @@ |
1012 | return -1; |
1013 | } |
1014 | |
1015 | +static u8 uniq_ioapic_id(u8 id) |
1016 | +{ |
1017 | + int i; |
1018 | + DECLARE_BITMAP(used, 256); |
1019 | + bitmap_zero(used, 256); |
1020 | + for (i = 0; i < nr_ioapics; i++) { |
1021 | + struct mpc_config_ioapic *ia = &mp_ioapics[i]; |
1022 | + __set_bit(ia->mpc_apicid, used); |
1023 | + } |
1024 | + if (!test_bit(id, used)) |
1025 | + return id; |
1026 | + return find_first_zero_bit(used, 256); |
1027 | +} |
1028 | + |
1029 | void __init mp_register_ioapic(u8 id, u32 address, u32 gsi_base) |
1030 | { |
1031 | int idx = 0; |
1032 | @@ -664,7 +677,7 @@ |
1033 | if (bad_ioapic(address)) |
1034 | return; |
1035 | |
1036 | - idx = nr_ioapics++; |
1037 | + idx = nr_ioapics; |
1038 | |
1039 | mp_ioapics[idx].mpc_type = MP_IOAPIC; |
1040 | mp_ioapics[idx].mpc_flags = MPC_APIC_USABLE; |
1041 | @@ -673,7 +686,7 @@ |
1042 | #ifndef CONFIG_XEN |
1043 | set_fixmap_nocache(FIX_IO_APIC_BASE_0 + idx, address); |
1044 | #endif |
1045 | - mp_ioapics[idx].mpc_apicid = id; |
1046 | + mp_ioapics[idx].mpc_apicid = uniq_ioapic_id(id); |
1047 | mp_ioapics[idx].mpc_apicver = 0; |
1048 | |
1049 | /* |
1050 | @@ -690,6 +703,8 @@ |
1051 | mp_ioapics[idx].mpc_apicaddr, |
1052 | mp_ioapic_routing[idx].gsi_start, |
1053 | mp_ioapic_routing[idx].gsi_end); |
1054 | + |
1055 | + nr_ioapics++; |
1056 | } |
1057 | |
1058 | void __init |
1059 | diff -Naur linux-2.6.25/arch/x86/kernel/pci-dma_32-xen.c linux-2.6.25-xen/arch/x86/kernel/pci-dma_32-xen.c |
1060 | --- linux-2.6.25/arch/x86/kernel/pci-dma_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1061 | +++ linux-2.6.25-xen/arch/x86/kernel/pci-dma_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1062 | @@ -24,7 +24,7 @@ |
1063 | #include <asm/bug.h> |
1064 | |
1065 | #ifdef __x86_64__ |
1066 | -#include <asm/proto.h> |
1067 | +#include <asm/iommu.h> |
1068 | |
1069 | int iommu_merge __read_mostly = 0; |
1070 | EXPORT_SYMBOL(iommu_merge); |
1071 | diff -Naur linux-2.6.25/arch/x86/kernel/pci-swiotlb_64-xen.c linux-2.6.25-xen/arch/x86/kernel/pci-swiotlb_64-xen.c |
1072 | --- linux-2.6.25/arch/x86/kernel/pci-swiotlb_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1073 | +++ linux-2.6.25-xen/arch/x86/kernel/pci-swiotlb_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1074 | @@ -5,7 +5,7 @@ |
1075 | #include <linux/module.h> |
1076 | #include <linux/dma-mapping.h> |
1077 | |
1078 | -#include <asm/proto.h> |
1079 | +#include <asm/iommu.h> |
1080 | #include <asm/swiotlb.h> |
1081 | #include <asm/dma.h> |
1082 | |
1083 | diff -Naur linux-2.6.25/arch/x86/kernel/process_32-xen.c linux-2.6.25-xen/arch/x86/kernel/process_32-xen.c |
1084 | --- linux-2.6.25/arch/x86/kernel/process_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1085 | +++ linux-2.6.25-xen/arch/x86/kernel/process_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1086 | @@ -241,6 +241,7 @@ |
1087 | void show_regs(struct pt_regs * regs) |
1088 | { |
1089 | unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L; |
1090 | + unsigned long d0, d1, d2, d3, d6, d7; |
1091 | |
1092 | printk("\n"); |
1093 | printk("Pid: %d, comm: %20s\n", current->pid, current->comm); |
1094 | @@ -265,6 +266,17 @@ |
1095 | cr3 = read_cr3(); |
1096 | cr4 = read_cr4_safe(); |
1097 | printk("CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n", cr0, cr2, cr3, cr4); |
1098 | + |
1099 | + get_debugreg(d0, 0); |
1100 | + get_debugreg(d1, 1); |
1101 | + get_debugreg(d2, 2); |
1102 | + get_debugreg(d3, 3); |
1103 | + printk("DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n", |
1104 | + d0, d1, d2, d3); |
1105 | + get_debugreg(d6, 6); |
1106 | + get_debugreg(d7, 7); |
1107 | + printk("DR6: %08lx DR7: %08lx\n", d6, d7); |
1108 | + |
1109 | show_trace(NULL, regs, ®s->esp); |
1110 | } |
1111 | |
1112 | @@ -473,7 +485,30 @@ |
1113 | return 1; |
1114 | } |
1115 | |
1116 | -static noinline void __switch_to_xtra(struct task_struct *next_p) |
1117 | +#ifdef CONFIG_SECCOMP |
1118 | +void hard_disable_TSC(void) |
1119 | +{ |
1120 | + write_cr4(read_cr4() | X86_CR4_TSD); |
1121 | +} |
1122 | +void disable_TSC(void) |
1123 | +{ |
1124 | + preempt_disable(); |
1125 | + if (!test_and_set_thread_flag(TIF_NOTSC)) |
1126 | + /* |
1127 | + * Must flip the CPU state synchronously with |
1128 | + * TIF_NOTSC in the current running context. |
1129 | + */ |
1130 | + hard_disable_TSC(); |
1131 | + preempt_enable(); |
1132 | +} |
1133 | +void hard_enable_TSC(void) |
1134 | +{ |
1135 | + write_cr4(read_cr4() & ~X86_CR4_TSD); |
1136 | +} |
1137 | +#endif /* CONFIG_SECCOMP */ |
1138 | + |
1139 | +static noinline void |
1140 | +__switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p) |
1141 | { |
1142 | struct thread_struct *next; |
1143 | |
1144 | @@ -488,33 +523,17 @@ |
1145 | set_debugreg(next->debugreg[6], 6); |
1146 | set_debugreg(next->debugreg[7], 7); |
1147 | } |
1148 | -} |
1149 | |
1150 | -/* |
1151 | - * This function selects if the context switch from prev to next |
1152 | - * has to tweak the TSC disable bit in the cr4. |
1153 | - */ |
1154 | -static inline void disable_tsc(struct task_struct *prev_p, |
1155 | - struct task_struct *next_p) |
1156 | -{ |
1157 | - struct thread_info *prev, *next; |
1158 | - |
1159 | - /* |
1160 | - * gcc should eliminate the ->thread_info dereference if |
1161 | - * has_secure_computing returns 0 at compile time (SECCOMP=n). |
1162 | - */ |
1163 | - prev = task_thread_info(prev_p); |
1164 | - next = task_thread_info(next_p); |
1165 | - |
1166 | - if (has_secure_computing(prev) || has_secure_computing(next)) { |
1167 | - /* slow path here */ |
1168 | - if (has_secure_computing(prev) && |
1169 | - !has_secure_computing(next)) { |
1170 | - write_cr4(read_cr4() & ~X86_CR4_TSD); |
1171 | - } else if (!has_secure_computing(prev) && |
1172 | - has_secure_computing(next)) |
1173 | - write_cr4(read_cr4() | X86_CR4_TSD); |
1174 | +#ifdef CONFIG_SECCOMP |
1175 | + if (test_tsk_thread_flag(prev_p, TIF_NOTSC) ^ |
1176 | + test_tsk_thread_flag(next_p, TIF_NOTSC)) { |
1177 | + /* prev and next are different */ |
1178 | + if (test_tsk_thread_flag(next_p, TIF_NOTSC)) |
1179 | + hard_disable_TSC(); |
1180 | + else |
1181 | + hard_enable_TSC(); |
1182 | } |
1183 | +#endif |
1184 | } |
1185 | |
1186 | /* |
1187 | @@ -628,10 +647,9 @@ |
1188 | /* |
1189 | * Now maybe handle debug registers |
1190 | */ |
1191 | - if (unlikely(task_thread_info(next_p)->flags & _TIF_WORK_CTXSW)) |
1192 | - __switch_to_xtra(next_p); |
1193 | - |
1194 | - disable_tsc(prev_p, next_p); |
1195 | + if (unlikely(task_thread_info(prev_p)->flags & _TIF_WORK_CTXSW_PREV || |
1196 | + task_thread_info(next_p)->flags & _TIF_WORK_CTXSW_NEXT)) |
1197 | + __switch_to_xtra(prev_p, next_p); |
1198 | |
1199 | /* |
1200 | * Leave lazy mode, flushing any hypercalls made here. |
1201 | diff -Naur linux-2.6.25/arch/x86/kernel/process_64-xen.c linux-2.6.25-xen/arch/x86/kernel/process_64-xen.c |
1202 | --- linux-2.6.25/arch/x86/kernel/process_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1203 | +++ linux-2.6.25-xen/arch/x86/kernel/process_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1204 | @@ -26,6 +26,7 @@ |
1205 | #include <linux/sched.h> |
1206 | #include <linux/kernel.h> |
1207 | #include <linux/mm.h> |
1208 | +#include <linux/fs.h> |
1209 | #include <linux/elfcore.h> |
1210 | #include <linux/smp.h> |
1211 | #include <linux/slab.h> |
1212 | @@ -249,6 +250,7 @@ |
1213 | void __show_regs(struct pt_regs * regs) |
1214 | { |
1215 | unsigned long fs, gs, shadowgs; |
1216 | + unsigned long d0, d1, d2, d3, d6, d7; |
1217 | unsigned int fsindex,gsindex; |
1218 | unsigned int ds,cs,es; |
1219 | |
1220 | @@ -288,6 +290,14 @@ |
1221 | fs,fsindex,gs,gsindex,shadowgs); |
1222 | printk("CS: %04x DS: %04x ES: %04x\n", cs, ds, es); |
1223 | |
1224 | + get_debugreg(d0, 0); |
1225 | + get_debugreg(d1, 1); |
1226 | + get_debugreg(d2, 2); |
1227 | + printk("DR0: %016lx DR1: %016lx DR2: %016lx\n", d0, d1, d2); |
1228 | + get_debugreg(d3, 3); |
1229 | + get_debugreg(d6, 6); |
1230 | + get_debugreg(d7, 7); |
1231 | + printk("DR3: %016lx DR6: %016lx DR7: %016lx\n", d3, d6, d7); |
1232 | } |
1233 | |
1234 | void show_regs(struct pt_regs *regs) |
1235 | diff -Naur linux-2.6.25/arch/x86/kernel/setup_32-xen.c linux-2.6.25-xen/arch/x86/kernel/setup_32-xen.c |
1236 | --- linux-2.6.25/arch/x86/kernel/setup_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1237 | +++ linux-2.6.25-xen/arch/x86/kernel/setup_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1238 | @@ -117,19 +117,10 @@ |
1239 | /* |
1240 | * Setup options |
1241 | */ |
1242 | -struct drive_info_struct { char dummy[32]; } drive_info; |
1243 | -#if defined(CONFIG_BLK_DEV_IDE) || defined(CONFIG_BLK_DEV_HD) || \ |
1244 | - defined(CONFIG_BLK_DEV_IDE_MODULE) || defined(CONFIG_BLK_DEV_HD_MODULE) |
1245 | -EXPORT_SYMBOL(drive_info); |
1246 | -#endif |
1247 | struct screen_info screen_info; |
1248 | EXPORT_SYMBOL(screen_info); |
1249 | struct apm_info apm_info; |
1250 | EXPORT_SYMBOL(apm_info); |
1251 | -struct sys_desc_table_struct { |
1252 | - unsigned short length; |
1253 | - unsigned char table[0]; |
1254 | -}; |
1255 | struct edid_info edid_info; |
1256 | EXPORT_SYMBOL_GPL(edid_info); |
1257 | #ifndef CONFIG_XEN |
1258 | @@ -152,7 +143,7 @@ |
1259 | |
1260 | static char __initdata command_line[COMMAND_LINE_SIZE]; |
1261 | |
1262 | -unsigned char __initdata boot_params[PARAM_SIZE]; |
1263 | +struct boot_params __initdata boot_params; |
1264 | |
1265 | /* |
1266 | * Point at the empty zero page to start with. We map the real shared_info |
1267 | @@ -319,18 +310,18 @@ |
1268 | printk(KERN_WARNING "Warning only %ldMB will be used.\n", |
1269 | MAXMEM>>20); |
1270 | if (max_pfn > MAX_NONPAE_PFN) |
1271 | - printk(KERN_WARNING "Use a PAE enabled kernel.\n"); |
1272 | + printk(KERN_WARNING "Use a HIGHMEM64G enabled kernel.\n"); |
1273 | else |
1274 | printk(KERN_WARNING "Use a HIGHMEM enabled kernel.\n"); |
1275 | max_pfn = MAXMEM_PFN; |
1276 | #else /* !CONFIG_HIGHMEM */ |
1277 | -#ifndef CONFIG_X86_PAE |
1278 | +#ifndef CONFIG_HIGHMEM64G |
1279 | if (max_pfn > MAX_NONPAE_PFN) { |
1280 | max_pfn = MAX_NONPAE_PFN; |
1281 | printk(KERN_WARNING "Warning only 4GB will be used.\n"); |
1282 | - printk(KERN_WARNING "Use a PAE enabled kernel.\n"); |
1283 | + printk(KERN_WARNING "Use a HIGHMEM64G enabled kernel.\n"); |
1284 | } |
1285 | -#endif /* !CONFIG_X86_PAE */ |
1286 | +#endif /* !CONFIG_HIGHMEM64G */ |
1287 | #endif /* !CONFIG_HIGHMEM */ |
1288 | } else { |
1289 | if (highmem_pages == -1) |
1290 | @@ -517,7 +508,7 @@ |
1291 | * |
1292 | * This should all compile down to nothing when NUMA is off. |
1293 | */ |
1294 | -void __init remapped_pgdat_init(void) |
1295 | +static void __init remapped_pgdat_init(void) |
1296 | { |
1297 | int nid; |
1298 | |
1299 | @@ -592,7 +583,6 @@ |
1300 | properly. Setting ROOT_DEV to default to /dev/ram0 breaks initrd. |
1301 | */ |
1302 | ROOT_DEV = MKDEV(UNNAMED_MAJOR,0); |
1303 | - drive_info = DRIVE_INFO; |
1304 | screen_info = SCREEN_INFO; |
1305 | copy_edid(); |
1306 | apm_info.bios = APM_BIOS_INFO; |
1307 | @@ -770,6 +760,8 @@ |
1308 | * NOTE: at this point the bootmem allocator is fully available. |
1309 | */ |
1310 | |
1311 | + paravirt_post_allocator_init(); |
1312 | + |
1313 | if (is_initial_xendomain()) |
1314 | dmi_scan_machine(); |
1315 | |
1316 | @@ -817,6 +809,7 @@ |
1317 | #endif |
1318 | |
1319 | e820_register_memory(); |
1320 | + e820_mark_nosave_regions(); |
1321 | |
1322 | if (is_initial_xendomain()) { |
1323 | #ifdef CONFIG_VT |
1324 | diff -Naur linux-2.6.25/arch/x86/kernel/setup64-xen.c linux-2.6.25-xen/arch/x86/kernel/setup64-xen.c |
1325 | --- linux-2.6.25/arch/x86/kernel/setup64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1326 | +++ linux-2.6.25-xen/arch/x86/kernel/setup64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1327 | @@ -125,11 +125,14 @@ |
1328 | } |
1329 | |
1330 | #ifdef CONFIG_XEN |
1331 | -static void switch_pt(void) |
1332 | +static void __init_refok switch_pt(int cpu) |
1333 | { |
1334 | + if (cpu == 0) |
1335 | + xen_init_pt(); |
1336 | xen_pt_switch(__pa_symbol(init_level4_pgt)); |
1337 | xen_new_user_pt(__pa_symbol(__user_pgd(init_level4_pgt))); |
1338 | } |
1339 | +#define switch_pt() switch_pt(cpu) |
1340 | |
1341 | static void __cpuinit cpu_gdt_init(const struct desc_ptr *gdt_descr) |
1342 | { |
1343 | @@ -185,9 +188,6 @@ |
1344 | pda->mmu_state = 0; |
1345 | |
1346 | if (cpu == 0) { |
1347 | -#ifdef CONFIG_XEN |
1348 | - xen_init_pt(); |
1349 | -#endif |
1350 | /* others are initialized in smpboot.c */ |
1351 | pda->pcurrent = &init_task; |
1352 | pda->irqstackptr = boot_cpu_stack; |
1353 | diff -Naur linux-2.6.25/arch/x86/kernel/setup_64-xen.c linux-2.6.25-xen/arch/x86/kernel/setup_64-xen.c |
1354 | --- linux-2.6.25/arch/x86/kernel/setup_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1355 | +++ linux-2.6.25-xen/arch/x86/kernel/setup_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1356 | @@ -840,6 +840,8 @@ |
1357 | level = cpuid_eax(1); |
1358 | if (c->x86 == 15 && ((level >= 0x0f48 && level < 0x0f50) || level >= 0x0f58)) |
1359 | set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability); |
1360 | + if (c->x86 == 0x10) |
1361 | + set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability); |
1362 | |
1363 | /* Enable workaround for FXSAVE leak */ |
1364 | if (c->x86 >= 6) |
1365 | @@ -865,8 +867,14 @@ |
1366 | if (c->extended_cpuid_level >= 0x80000008) |
1367 | amd_detect_cmp(c); |
1368 | |
1369 | - /* Fix cpuid4 emulation for more */ |
1370 | - num_cache_leaves = 3; |
1371 | + if (c->extended_cpuid_level >= 0x80000006 && |
1372 | + (cpuid_edx(0x80000006) & 0xf000)) |
1373 | + num_cache_leaves = 4; |
1374 | + else |
1375 | + num_cache_leaves = 3; |
1376 | + |
1377 | + if (c->x86 == 0xf || c->x86 == 0x10 || c->x86 == 0x11) |
1378 | + set_bit(X86_FEATURE_K8, &c->x86_capability); |
1379 | |
1380 | /* RDTSC can be speculated around */ |
1381 | clear_bit(X86_FEATURE_SYNC_RDTSC, &c->x86_capability); |
1382 | @@ -1111,6 +1119,8 @@ |
1383 | c->x86_capability[2] = cpuid_edx(0x80860001); |
1384 | } |
1385 | |
1386 | + init_scattered_cpuid_features(c); |
1387 | + |
1388 | c->apicid = phys_pkg_id(0); |
1389 | |
1390 | /* |
1391 | @@ -1196,7 +1206,7 @@ |
1392 | "fpu", "vme", "de", "pse", "tsc", "msr", "pae", "mce", |
1393 | "cx8", "apic", NULL, "sep", "mtrr", "pge", "mca", "cmov", |
1394 | "pat", "pse36", "pn", "clflush", NULL, "dts", "acpi", "mmx", |
1395 | - "fxsr", "sse", "sse2", "ss", "ht", "tm", "ia64", NULL, |
1396 | + "fxsr", "sse", "sse2", "ss", "ht", "tm", "ia64", "pbe", |
1397 | |
1398 | /* AMD-defined */ |
1399 | NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1400 | @@ -1212,10 +1222,11 @@ |
1401 | NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1402 | |
1403 | /* Other (Linux-defined) */ |
1404 | - "cxmmx", NULL, "cyrix_arr", "centaur_mcr", NULL, |
1405 | - "constant_tsc", NULL, NULL, |
1406 | - "up", NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1407 | - NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1408 | + "cxmmx", "k6_mtrr", "cyrix_arr", "centaur_mcr", |
1409 | + NULL, NULL, NULL, NULL, |
1410 | + "constant_tsc", "up", NULL, "arch_perfmon", |
1411 | + "pebs", "bts", NULL, "sync_rdtsc", |
1412 | + "rep_good", NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1413 | NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1414 | |
1415 | /* Intel-defined (#2) */ |
1416 | @@ -1226,7 +1237,7 @@ |
1417 | |
1418 | /* VIA/Cyrix/Centaur-defined */ |
1419 | NULL, NULL, "rng", "rng_en", NULL, NULL, "ace", "ace_en", |
1420 | - NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1421 | + "ace2", "ace2_en", "phe", "phe_en", "pmm", "pmm_en", NULL, NULL, |
1422 | NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1423 | NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1424 | |
1425 | @@ -1237,6 +1248,12 @@ |
1426 | "osvw", "ibs", NULL, NULL, NULL, NULL, |
1427 | NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1428 | NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1429 | + |
1430 | + /* Auxiliary (Linux-defined) */ |
1431 | + "ida", NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1432 | + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1433 | + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1434 | + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, |
1435 | }; |
1436 | static char *x86_power_flags[] = { |
1437 | "ts", /* temperature sensor */ |
1438 | diff -Naur linux-2.6.25/arch/x86/kernel/smp_32-xen.c linux-2.6.25-xen/arch/x86/kernel/smp_32-xen.c |
1439 | --- linux-2.6.25/arch/x86/kernel/smp_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1440 | +++ linux-2.6.25-xen/arch/x86/kernel/smp_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1441 | @@ -22,6 +22,7 @@ |
1442 | |
1443 | #include <asm/mtrr.h> |
1444 | #include <asm/tlbflush.h> |
1445 | +#include <asm/mmu_context.h> |
1446 | #if 0 |
1447 | #include <mach_apic.h> |
1448 | #endif |
1449 | @@ -217,13 +218,13 @@ |
1450 | static DEFINE_SPINLOCK(tlbstate_lock); |
1451 | |
1452 | /* |
1453 | - * We cannot call mmdrop() because we are in interrupt context, |
1454 | + * We cannot call mmdrop() because we are in interrupt context, |
1455 | * instead update mm->cpu_vm_mask. |
1456 | * |
1457 | * We need to reload %cr3 since the page tables may be going |
1458 | * away from under us.. |
1459 | */ |
1460 | -static inline void leave_mm (unsigned long cpu) |
1461 | +void leave_mm(unsigned long cpu) |
1462 | { |
1463 | if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK) |
1464 | BUG(); |
1465 | diff -Naur linux-2.6.25/arch/x86/kernel/smp_64-xen.c linux-2.6.25-xen/arch/x86/kernel/smp_64-xen.c |
1466 | --- linux-2.6.25/arch/x86/kernel/smp_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1467 | +++ linux-2.6.25-xen/arch/x86/kernel/smp_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1468 | @@ -362,7 +362,7 @@ |
1469 | } |
1470 | |
1471 | /* |
1472 | - * smp_call_function_single - Run a function on another CPU |
1473 | + * smp_call_function_single - Run a function on a specific CPU |
1474 | * @func: The function to run. This must be fast and non-blocking. |
1475 | * @info: An arbitrary pointer to pass to the function. |
1476 | * @nonatomic: Currently unused. |
1477 | @@ -379,17 +379,21 @@ |
1478 | { |
1479 | /* prevent preemption and reschedule on another processor */ |
1480 | int me = get_cpu(); |
1481 | + |
1482 | + /* Can deadlock when called with interrupts disabled */ |
1483 | + WARN_ON(irqs_disabled()); |
1484 | + |
1485 | if (cpu == me) { |
1486 | + local_irq_disable(); |
1487 | + func(info); |
1488 | + local_irq_enable(); |
1489 | put_cpu(); |
1490 | return 0; |
1491 | } |
1492 | |
1493 | - /* Can deadlock when called with interrupts disabled */ |
1494 | - WARN_ON(irqs_disabled()); |
1495 | - |
1496 | - spin_lock_bh(&call_lock); |
1497 | + spin_lock(&call_lock); |
1498 | __smp_call_function_single(cpu, func, info, nonatomic, wait); |
1499 | - spin_unlock_bh(&call_lock); |
1500 | + spin_unlock(&call_lock); |
1501 | put_cpu(); |
1502 | return 0; |
1503 | } |
1504 | diff -Naur linux-2.6.25/arch/x86/kernel/time_32-xen.c linux-2.6.25-xen/arch/x86/kernel/time_32-xen.c |
1505 | --- linux-2.6.25/arch/x86/kernel/time_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1506 | +++ linux-2.6.25-xen/arch/x86/kernel/time_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1507 | @@ -76,11 +76,12 @@ |
1508 | #include <xen/evtchn.h> |
1509 | #include <xen/interface/vcpu.h> |
1510 | |
1511 | -#ifdef CONFIG_X86_32 |
1512 | #include <asm/i8253.h> |
1513 | DEFINE_SPINLOCK(i8253_lock); |
1514 | EXPORT_SYMBOL(i8253_lock); |
1515 | -#else |
1516 | + |
1517 | +#ifdef CONFIG_X86_64 |
1518 | +#include <asm/vsyscall.h> |
1519 | volatile unsigned long __jiffies __section_jiffies = INITIAL_JIFFIES; |
1520 | #endif |
1521 | |
1522 | @@ -209,6 +210,26 @@ |
1523 | return product; |
1524 | } |
1525 | |
1526 | +static inline u64 get64(volatile u64 *ptr) |
1527 | +{ |
1528 | +#ifndef CONFIG_64BIT |
1529 | + return cmpxchg64(ptr, 0, 0); |
1530 | +#else |
1531 | + return *ptr; |
1532 | +#define cmpxchg64 cmpxchg |
1533 | +#endif |
1534 | +} |
1535 | + |
1536 | +static inline u64 get64_local(volatile u64 *ptr) |
1537 | +{ |
1538 | +#ifndef CONFIG_64BIT |
1539 | + return cmpxchg64_local(ptr, 0, 0); |
1540 | +#else |
1541 | + return *ptr; |
1542 | +#define cmpxchg64_local cmpxchg_local |
1543 | +#endif |
1544 | +} |
1545 | + |
1546 | #if 0 /* defined (__i386__) */ |
1547 | int read_current_timer(unsigned long *timer_val) |
1548 | { |
1549 | @@ -391,7 +412,7 @@ |
1550 | return retval; |
1551 | } |
1552 | |
1553 | -unsigned long long sched_clock(void) |
1554 | +static unsigned long long local_clock(void) |
1555 | { |
1556 | unsigned int cpu = get_cpu(); |
1557 | struct shadow_time_info *shadow = &per_cpu(shadow_time, cpu); |
1558 | @@ -412,6 +433,61 @@ |
1559 | return time; |
1560 | } |
1561 | |
1562 | +/* |
1563 | + * Runstate accounting |
1564 | + */ |
1565 | +static void get_runstate_snapshot(struct vcpu_runstate_info *res) |
1566 | +{ |
1567 | + u64 state_time; |
1568 | + struct vcpu_runstate_info *state; |
1569 | + |
1570 | + BUG_ON(preemptible()); |
1571 | + |
1572 | + state = &__get_cpu_var(runstate); |
1573 | + |
1574 | + do { |
1575 | + state_time = get64_local(&state->state_entry_time); |
1576 | + *res = *state; |
1577 | + } while (get64_local(&state->state_entry_time) != state_time); |
1578 | + |
1579 | + WARN_ON_ONCE(res->state != RUNSTATE_running); |
1580 | +} |
1581 | + |
1582 | +/* |
1583 | + * Xen sched_clock implementation. Returns the number of unstolen |
1584 | + * nanoseconds, which is nanoseconds the VCPU spent in RUNNING+BLOCKED |
1585 | + * states. |
1586 | + */ |
1587 | +unsigned long long sched_clock(void) |
1588 | +{ |
1589 | + struct vcpu_runstate_info runstate; |
1590 | + cycle_t now; |
1591 | + u64 ret; |
1592 | + s64 offset; |
1593 | + |
1594 | + /* |
1595 | + * Ideally sched_clock should be called on a per-cpu basis |
1596 | + * anyway, so preempt should already be disabled, but that's |
1597 | + * not current practice at the moment. |
1598 | + */ |
1599 | + preempt_disable(); |
1600 | + |
1601 | + now = local_clock(); |
1602 | + |
1603 | + get_runstate_snapshot(&runstate); |
1604 | + |
1605 | + offset = now - runstate.state_entry_time; |
1606 | + if (offset < 0) |
1607 | + offset = 0; |
1608 | + |
1609 | + ret = offset + runstate.time[RUNSTATE_running] |
1610 | + + runstate.time[RUNSTATE_blocked]; |
1611 | + |
1612 | + preempt_enable(); |
1613 | + |
1614 | + return ret; |
1615 | +} |
1616 | + |
1617 | unsigned long profile_pc(struct pt_regs *regs) |
1618 | { |
1619 | unsigned long pc = instruction_pointer(regs); |
1620 | @@ -459,10 +535,9 @@ |
1621 | irqreturn_t timer_interrupt(int irq, void *dev_id) |
1622 | { |
1623 | s64 delta, delta_cpu, stolen, blocked; |
1624 | - u64 sched_time; |
1625 | unsigned int i, cpu = smp_processor_id(); |
1626 | struct shadow_time_info *shadow = &per_cpu(shadow_time, cpu); |
1627 | - struct vcpu_runstate_info *runstate = &per_cpu(runstate, cpu); |
1628 | + struct vcpu_runstate_info runstate; |
1629 | |
1630 | /* |
1631 | * Here we are in the timer irq handler. We just have irqs locally |
1632 | @@ -482,20 +557,7 @@ |
1633 | delta -= processed_system_time; |
1634 | delta_cpu -= per_cpu(processed_system_time, cpu); |
1635 | |
1636 | - /* |
1637 | - * Obtain a consistent snapshot of stolen/blocked cycles. We |
1638 | - * can use state_entry_time to detect if we get preempted here. |
1639 | - */ |
1640 | - do { |
1641 | - sched_time = runstate->state_entry_time; |
1642 | - barrier(); |
1643 | - stolen = runstate->time[RUNSTATE_runnable] + |
1644 | - runstate->time[RUNSTATE_offline] - |
1645 | - per_cpu(processed_stolen_time, cpu); |
1646 | - blocked = runstate->time[RUNSTATE_blocked] - |
1647 | - per_cpu(processed_blocked_time, cpu); |
1648 | - barrier(); |
1649 | - } while (sched_time != runstate->state_entry_time); |
1650 | + get_runstate_snapshot(&runstate); |
1651 | } while (!time_values_up_to_date(cpu)); |
1652 | |
1653 | if ((unlikely(delta < -(s64)permitted_clock_jitter) || |
1654 | @@ -537,6 +599,9 @@ |
1655 | * HACK: Passing NULL to account_steal_time() |
1656 | * ensures that the ticks are accounted as stolen. |
1657 | */ |
1658 | + stolen = runstate.time[RUNSTATE_runnable] |
1659 | + + runstate.time[RUNSTATE_offline] |
1660 | + - per_cpu(processed_stolen_time, cpu); |
1661 | if ((stolen > 0) && (delta_cpu > 0)) { |
1662 | delta_cpu -= stolen; |
1663 | if (unlikely(delta_cpu < 0)) |
1664 | @@ -552,6 +617,8 @@ |
1665 | * HACK: Passing idle_task to account_steal_time() |
1666 | * ensures that the ticks are accounted as idle/wait. |
1667 | */ |
1668 | + blocked = runstate.time[RUNSTATE_blocked] |
1669 | + - per_cpu(processed_blocked_time, cpu); |
1670 | if ((blocked > 0) && (delta_cpu > 0)) { |
1671 | delta_cpu -= blocked; |
1672 | if (unlikely(delta_cpu < 0)) |
1673 | @@ -600,16 +667,11 @@ |
1674 | |
1675 | static cycle_t xen_clocksource_read(void) |
1676 | { |
1677 | - cycle_t ret = sched_clock(); |
1678 | + cycle_t ret = local_clock(); |
1679 | |
1680 | #ifdef CONFIG_SMP |
1681 | for (;;) { |
1682 | -#ifndef CONFIG_64BIT |
1683 | - cycle_t last = cmpxchg64(&cs_last, 0, 0); |
1684 | -#else |
1685 | - cycle_t last = cs_last; |
1686 | -#define cmpxchg64 cmpxchg |
1687 | -#endif |
1688 | + cycle_t last = get64(&cs_last); |
1689 | |
1690 | if ((s64)(ret - last) < 0) { |
1691 | if (last - ret > permitted_clock_jitter |
1692 | @@ -640,7 +702,7 @@ |
1693 | extern void time_resume(void); |
1694 | |
1695 | time_resume(); |
1696 | - cs_last = sched_clock(); |
1697 | + cs_last = local_clock(); |
1698 | } |
1699 | |
1700 | static struct clocksource clocksource_xen = { |
1701 | @@ -688,56 +750,10 @@ |
1702 | return retval; |
1703 | } |
1704 | |
1705 | -static void sync_cmos_clock(unsigned long dummy); |
1706 | - |
1707 | -static DEFINE_TIMER(sync_cmos_timer, sync_cmos_clock, 0, 0); |
1708 | -int no_sync_cmos_clock; |
1709 | - |
1710 | -static void sync_cmos_clock(unsigned long dummy) |
1711 | -{ |
1712 | - struct timeval now, next; |
1713 | - int fail = 1; |
1714 | - |
1715 | - /* |
1716 | - * If we have an externally synchronized Linux clock, then update |
1717 | - * CMOS clock accordingly every ~11 minutes. Set_rtc_mmss() has to be |
1718 | - * called as close as possible to 500 ms before the new second starts. |
1719 | - * This code is run on a timer. If the clock is set, that timer |
1720 | - * may not expire at the correct time. Thus, we adjust... |
1721 | - */ |
1722 | - if (!ntp_synced()) |
1723 | - /* |
1724 | - * Not synced, exit, do not restart a timer (if one is |
1725 | - * running, let it run out). |
1726 | - */ |
1727 | - return; |
1728 | - |
1729 | - do_gettimeofday(&now); |
1730 | - if (now.tv_usec >= USEC_AFTER - ((unsigned) TICK_SIZE) / 2 && |
1731 | - now.tv_usec <= USEC_BEFORE + ((unsigned) TICK_SIZE) / 2) |
1732 | - fail = set_rtc_mmss(now.tv_sec); |
1733 | - |
1734 | - next.tv_usec = USEC_AFTER - now.tv_usec; |
1735 | - if (next.tv_usec <= 0) |
1736 | - next.tv_usec += USEC_PER_SEC; |
1737 | - |
1738 | - if (!fail) |
1739 | - next.tv_sec = 659; |
1740 | - else |
1741 | - next.tv_sec = 0; |
1742 | - |
1743 | - if (next.tv_usec >= USEC_PER_SEC) { |
1744 | - next.tv_sec++; |
1745 | - next.tv_usec -= USEC_PER_SEC; |
1746 | - } |
1747 | - mod_timer(&sync_cmos_timer, jiffies + timeval_to_jiffies(&next)); |
1748 | -} |
1749 | - |
1750 | -void notify_arch_cmos_timer(void) |
1751 | +int update_persistent_clock(struct timespec now) |
1752 | { |
1753 | - if (!no_sync_cmos_clock) |
1754 | - mod_timer(&sync_cmos_timer, jiffies + 1); |
1755 | mod_timer(&sync_xen_wallclock_timer, jiffies + 1); |
1756 | + return set_rtc_mmss(now.tv_sec); |
1757 | } |
1758 | |
1759 | extern void (*late_time_init)(void); |
1760 | diff -Naur linux-2.6.25/arch/x86/kernel/traps_32-xen.c linux-2.6.25-xen/arch/x86/kernel/traps_32-xen.c |
1761 | --- linux-2.6.25/arch/x86/kernel/traps_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1762 | +++ linux-2.6.25-xen/arch/x86/kernel/traps_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1763 | @@ -41,6 +41,10 @@ |
1764 | #include <linux/mca.h> |
1765 | #endif |
1766 | |
1767 | +#if defined(CONFIG_EDAC) |
1768 | +#include <linux/edac.h> |
1769 | +#endif |
1770 | + |
1771 | #include <asm/processor.h> |
1772 | #include <asm/system.h> |
1773 | #include <asm/io.h> |
1774 | @@ -102,36 +106,45 @@ |
1775 | int kstack_depth_to_print = 24; |
1776 | static unsigned int code_bytes = 64; |
1777 | |
1778 | -static inline int valid_stack_ptr(struct thread_info *tinfo, void *p) |
1779 | +static inline int valid_stack_ptr(struct thread_info *tinfo, void *p, unsigned size) |
1780 | { |
1781 | return p > (void *)tinfo && |
1782 | - p < (void *)tinfo + THREAD_SIZE - 3; |
1783 | + p <= (void *)tinfo + THREAD_SIZE - size; |
1784 | } |
1785 | |
1786 | +/* The form of the top of the frame on the stack */ |
1787 | +struct stack_frame { |
1788 | + struct stack_frame *next_frame; |
1789 | + unsigned long return_address; |
1790 | +}; |
1791 | + |
1792 | static inline unsigned long print_context_stack(struct thread_info *tinfo, |
1793 | unsigned long *stack, unsigned long ebp, |
1794 | struct stacktrace_ops *ops, void *data) |
1795 | { |
1796 | - unsigned long addr; |
1797 | - |
1798 | #ifdef CONFIG_FRAME_POINTER |
1799 | - while (valid_stack_ptr(tinfo, (void *)ebp)) { |
1800 | - unsigned long new_ebp; |
1801 | - addr = *(unsigned long *)(ebp + 4); |
1802 | + struct stack_frame *frame = (struct stack_frame *)ebp; |
1803 | + while (valid_stack_ptr(tinfo, frame, sizeof(*frame))) { |
1804 | + struct stack_frame *next; |
1805 | + unsigned long addr; |
1806 | + |
1807 | + addr = frame->return_address; |
1808 | ops->address(data, addr); |
1809 | /* |
1810 | * break out of recursive entries (such as |
1811 | * end_of_stack_stop_unwind_function). Also, |
1812 | * we can never allow a frame pointer to |
1813 | * move downwards! |
1814 | - */ |
1815 | - new_ebp = *(unsigned long *)ebp; |
1816 | - if (new_ebp <= ebp) |
1817 | + */ |
1818 | + next = frame->next_frame; |
1819 | + if (next <= frame) |
1820 | break; |
1821 | - ebp = new_ebp; |
1822 | + frame = next; |
1823 | } |
1824 | #else |
1825 | - while (valid_stack_ptr(tinfo, stack)) { |
1826 | + while (valid_stack_ptr(tinfo, stack, sizeof(*stack))) { |
1827 | + unsigned long addr; |
1828 | + |
1829 | addr = *stack++; |
1830 | if (__kernel_text_address(addr)) |
1831 | ops->address(data, addr); |
1832 | @@ -154,7 +167,7 @@ |
1833 | if (!stack) { |
1834 | unsigned long dummy; |
1835 | stack = &dummy; |
1836 | - if (task && task != current) |
1837 | + if (task != current) |
1838 | stack = (unsigned long *)task->thread.esp; |
1839 | } |
1840 | |
1841 | @@ -213,6 +226,7 @@ |
1842 | { |
1843 | printk("%s [<%08lx>] ", (char *)data, addr); |
1844 | print_symbol("%s\n", addr); |
1845 | + touch_nmi_watchdog(); |
1846 | } |
1847 | |
1848 | static struct stacktrace_ops print_trace_ops = { |
1849 | @@ -396,7 +410,7 @@ |
1850 | unsigned long esp; |
1851 | unsigned short ss; |
1852 | |
1853 | - report_bug(regs->eip); |
1854 | + report_bug(regs->eip, regs); |
1855 | |
1856 | printk(KERN_EMERG "%s: %04lx [#%d]\n", str, err & 0xffff, ++die_counter); |
1857 | #ifdef CONFIG_PREEMPT |
1858 | @@ -439,6 +453,7 @@ |
1859 | |
1860 | bust_spinlocks(0); |
1861 | die.lock_owner = -1; |
1862 | + add_taint(TAINT_DIE); |
1863 | spin_unlock_irqrestore(&die.lock, flags); |
1864 | |
1865 | if (!regs) |
1866 | @@ -523,10 +538,12 @@ |
1867 | do_trap(trapnr, signr, str, 0, regs, error_code, NULL); \ |
1868 | } |
1869 | |
1870 | -#define DO_ERROR_INFO(trapnr, signr, str, name, sicode, siaddr) \ |
1871 | +#define DO_ERROR_INFO(trapnr, signr, str, name, sicode, siaddr, irq) \ |
1872 | fastcall void do_##name(struct pt_regs * regs, long error_code) \ |
1873 | { \ |
1874 | siginfo_t info; \ |
1875 | + if (irq) \ |
1876 | + local_irq_enable(); \ |
1877 | info.si_signo = signr; \ |
1878 | info.si_errno = 0; \ |
1879 | info.si_code = sicode; \ |
1880 | @@ -566,13 +583,13 @@ |
1881 | #endif |
1882 | DO_VM86_ERROR( 4, SIGSEGV, "overflow", overflow) |
1883 | DO_VM86_ERROR( 5, SIGSEGV, "bounds", bounds) |
1884 | -DO_ERROR_INFO( 6, SIGILL, "invalid opcode", invalid_op, ILL_ILLOPN, regs->eip) |
1885 | +DO_ERROR_INFO( 6, SIGILL, "invalid opcode", invalid_op, ILL_ILLOPN, regs->eip, 0) |
1886 | DO_ERROR( 9, SIGFPE, "coprocessor segment overrun", coprocessor_segment_overrun) |
1887 | DO_ERROR(10, SIGSEGV, "invalid TSS", invalid_TSS) |
1888 | DO_ERROR(11, SIGBUS, "segment not present", segment_not_present) |
1889 | DO_ERROR(12, SIGBUS, "stack segment", stack_segment) |
1890 | -DO_ERROR_INFO(17, SIGBUS, "alignment check", alignment_check, BUS_ADRALN, 0) |
1891 | -DO_ERROR_INFO(32, SIGSEGV, "iret exception", iret_error, ILL_BADSTK, 0) |
1892 | +DO_ERROR_INFO(17, SIGBUS, "alignment check", alignment_check, BUS_ADRALN, 0, 0) |
1893 | +DO_ERROR_INFO(32, SIGSEGV, "iret exception", iret_error, ILL_BADSTK, 0, 1) |
1894 | |
1895 | fastcall void __kprobes do_general_protection(struct pt_regs * regs, |
1896 | long error_code) |
1897 | @@ -585,6 +602,13 @@ |
1898 | |
1899 | current->thread.error_code = error_code; |
1900 | current->thread.trap_no = 13; |
1901 | + if (show_unhandled_signals && unhandled_signal(current, SIGSEGV) && |
1902 | + printk_ratelimit()) |
1903 | + printk(KERN_INFO |
1904 | + "%s[%d] general protection eip:%lx esp:%lx error:%lx\n", |
1905 | + current->comm, current->pid, |
1906 | + regs->eip, regs->esp, error_code); |
1907 | + |
1908 | force_sig(SIGSEGV, current); |
1909 | return; |
1910 | |
1911 | @@ -610,6 +634,14 @@ |
1912 | printk(KERN_EMERG "Uhhuh. NMI received for unknown reason %02x on " |
1913 | "CPU %d.\n", reason, smp_processor_id()); |
1914 | printk(KERN_EMERG "You have some hardware problem, likely on the PCI bus.\n"); |
1915 | + |
1916 | +#if defined(CONFIG_EDAC) |
1917 | + if(edac_handler_set()) { |
1918 | + edac_atomic_assert_error(); |
1919 | + return; |
1920 | + } |
1921 | +#endif |
1922 | + |
1923 | if (panic_on_unrecovered_nmi) |
1924 | panic("NMI: Not continuing"); |
1925 | |
1926 | @@ -720,6 +752,8 @@ |
1927 | reassert_nmi(); |
1928 | } |
1929 | |
1930 | +static int ignore_nmis; |
1931 | + |
1932 | fastcall __kprobes void do_nmi(struct pt_regs * regs, long error_code) |
1933 | { |
1934 | int cpu; |
1935 | @@ -730,11 +764,24 @@ |
1936 | |
1937 | ++nmi_count(cpu); |
1938 | |
1939 | - default_do_nmi(regs); |
1940 | + if (!ignore_nmis) |
1941 | + default_do_nmi(regs); |
1942 | |
1943 | nmi_exit(); |
1944 | } |
1945 | |
1946 | +void stop_nmi(void) |
1947 | +{ |
1948 | + acpi_nmi_disable(); |
1949 | + ignore_nmis++; |
1950 | +} |
1951 | + |
1952 | +void restart_nmi(void) |
1953 | +{ |
1954 | + ignore_nmis--; |
1955 | + acpi_nmi_enable(); |
1956 | +} |
1957 | + |
1958 | #ifdef CONFIG_KPROBES |
1959 | fastcall void __kprobes do_int3(struct pt_regs *regs, long error_code) |
1960 | { |
1961 | @@ -1023,6 +1070,7 @@ |
1962 | thread->status |= TS_USEDFPU; /* So we fnsave on switch_to() */ |
1963 | tsk->fpu_counter++; |
1964 | } |
1965 | +EXPORT_SYMBOL_GPL(math_state_restore); |
1966 | |
1967 | #ifndef CONFIG_MATH_EMULATION |
1968 | |
1969 | diff -Naur linux-2.6.25/arch/x86/kernel/traps_64-xen.c linux-2.6.25-xen/arch/x86/kernel/traps_64-xen.c |
1970 | --- linux-2.6.25/arch/x86/kernel/traps_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
1971 | +++ linux-2.6.25-xen/arch/x86/kernel/traps_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
1972 | @@ -34,6 +34,10 @@ |
1973 | #include <linux/bug.h> |
1974 | #include <linux/kdebug.h> |
1975 | |
1976 | +#if defined(CONFIG_EDAC) |
1977 | +#include <linux/edac.h> |
1978 | +#endif |
1979 | + |
1980 | #include <asm/system.h> |
1981 | #include <asm/io.h> |
1982 | #include <asm/atomic.h> |
1983 | @@ -332,6 +336,7 @@ |
1984 | |
1985 | static void print_trace_address(void *data, unsigned long addr) |
1986 | { |
1987 | + touch_nmi_watchdog(); |
1988 | printk_address(addr); |
1989 | } |
1990 | |
1991 | @@ -520,6 +525,7 @@ |
1992 | printk("\n"); |
1993 | notify_die(DIE_OOPS, str, regs, err, current->thread.trap_no, SIGSEGV); |
1994 | show_registers(regs); |
1995 | + add_taint(TAINT_DIE); |
1996 | /* Executive summary in case the oops scrolled away */ |
1997 | printk(KERN_ALERT "RIP "); |
1998 | printk_address(regs->rip); |
1999 | @@ -533,7 +539,7 @@ |
2000 | unsigned long flags = oops_begin(); |
2001 | |
2002 | if (!user_mode(regs)) |
2003 | - report_bug(regs->rip); |
2004 | + report_bug(regs->rip, regs); |
2005 | |
2006 | __die(str, regs, err); |
2007 | oops_end(flags); |
2008 | @@ -582,7 +588,8 @@ |
2009 | tsk->thread.error_code = error_code; |
2010 | tsk->thread.trap_no = trapnr; |
2011 | |
2012 | - if (exception_trace && unhandled_signal(tsk, signr)) |
2013 | + if (show_unhandled_signals && unhandled_signal(tsk, signr) && |
2014 | + printk_ratelimit()) |
2015 | printk(KERN_INFO |
2016 | "%s[%d] trap %s rip:%lx rsp:%lx error:%lx\n", |
2017 | tsk->comm, tsk->pid, str, |
2018 | @@ -686,7 +693,8 @@ |
2019 | tsk->thread.error_code = error_code; |
2020 | tsk->thread.trap_no = 13; |
2021 | |
2022 | - if (exception_trace && unhandled_signal(tsk, SIGSEGV)) |
2023 | + if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) && |
2024 | + printk_ratelimit()) |
2025 | printk(KERN_INFO |
2026 | "%s[%d] general protection rip:%lx rsp:%lx error:%lx\n", |
2027 | tsk->comm, tsk->pid, |
2028 | @@ -721,6 +729,13 @@ |
2029 | reason); |
2030 | printk(KERN_EMERG "You have some hardware problem, likely on the PCI bus.\n"); |
2031 | |
2032 | +#if defined(CONFIG_EDAC) |
2033 | + if(edac_handler_set()) { |
2034 | + edac_atomic_assert_error(); |
2035 | + return; |
2036 | + } |
2037 | +#endif |
2038 | + |
2039 | if (panic_on_unrecovered_nmi) |
2040 | panic("NMI: Not continuing"); |
2041 | |
2042 | diff -Naur linux-2.6.25/arch/x86/kernel/vsyscall_64-xen.c linux-2.6.25-xen/arch/x86/kernel/vsyscall_64-xen.c |
2043 | --- linux-2.6.25/arch/x86/kernel/vsyscall_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2044 | +++ linux-2.6.25-xen/arch/x86/kernel/vsyscall_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2045 | @@ -42,6 +42,7 @@ |
2046 | #include <asm/segment.h> |
2047 | #include <asm/desc.h> |
2048 | #include <asm/topology.h> |
2049 | +#include <asm/vgtod.h> |
2050 | |
2051 | #define __vsyscall(nr) __attribute__ ((unused,__section__(".vsyscall_" #nr))) |
2052 | #define __syscall_clobber "r11","rcx","memory" |
2053 | @@ -57,26 +58,9 @@ |
2054 | * - writen by timer interrupt or systcl (/proc/sys/kernel/vsyscall64) |
2055 | * Try to keep this structure as small as possible to avoid cache line ping pongs |
2056 | */ |
2057 | -struct vsyscall_gtod_data_t { |
2058 | - seqlock_t lock; |
2059 | - |
2060 | - /* open coded 'struct timespec' */ |
2061 | - time_t wall_time_sec; |
2062 | - u32 wall_time_nsec; |
2063 | - |
2064 | - int sysctl_enabled; |
2065 | - struct timezone sys_tz; |
2066 | - struct { /* extract of a clocksource struct */ |
2067 | - cycle_t (*vread)(void); |
2068 | - cycle_t cycle_last; |
2069 | - cycle_t mask; |
2070 | - u32 mult; |
2071 | - u32 shift; |
2072 | - } clock; |
2073 | -}; |
2074 | int __vgetcpu_mode __section_vgetcpu_mode; |
2075 | |
2076 | -struct vsyscall_gtod_data_t __vsyscall_gtod_data __section_vsyscall_gtod_data = |
2077 | +struct vsyscall_gtod_data __vsyscall_gtod_data __section_vsyscall_gtod_data = |
2078 | { |
2079 | .lock = SEQLOCK_UNLOCKED, |
2080 | .sysctl_enabled = 1, |
2081 | @@ -96,6 +80,8 @@ |
2082 | vsyscall_gtod_data.wall_time_sec = wall_time->tv_sec; |
2083 | vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec; |
2084 | vsyscall_gtod_data.sys_tz = sys_tz; |
2085 | + vsyscall_gtod_data.wall_time_nsec = wall_time->tv_nsec; |
2086 | + vsyscall_gtod_data.wall_to_monotonic = wall_to_monotonic; |
2087 | write_sequnlock_irqrestore(&vsyscall_gtod_data.lock, flags); |
2088 | } |
2089 | |
2090 | diff -Naur linux-2.6.25/arch/x86/mach-xen/setup.c linux-2.6.25-xen/arch/x86/mach-xen/setup.c |
2091 | --- linux-2.6.25/arch/x86/mach-xen/setup.c 2008-05-23 18:33:23.000000000 +0200 |
2092 | +++ linux-2.6.25-xen/arch/x86/mach-xen/setup.c 2008-05-23 18:27:40.000000000 +0200 |
2093 | @@ -12,6 +12,7 @@ |
2094 | #include <asm/e820.h> |
2095 | #include <asm/setup.h> |
2096 | #include <asm/fixmap.h> |
2097 | +#include <asm/pgtable.h> |
2098 | |
2099 | #include <xen/interface/callback.h> |
2100 | #include <xen/interface/memory.h> |
2101 | @@ -101,7 +102,7 @@ |
2102 | |
2103 | init_mm.pgd = swapper_pg_dir = (pgd_t *)xen_start_info->pt_base; |
2104 | |
2105 | - setup_xen_features(); |
2106 | + xen_setup_features(); |
2107 | |
2108 | if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0) { |
2109 | hypervisor_virt_start = pp.virt_start; |
2110 | @@ -157,4 +158,18 @@ |
2111 | HYPERVISOR_nmi_op(XENNMI_register_callback, &cb); |
2112 | } |
2113 | #endif |
2114 | + |
2115 | + /* Do an early initialization of the fixmap area */ |
2116 | + { |
2117 | + extern pte_t swapper_pg_pmd[PTRS_PER_PTE]; |
2118 | + unsigned long addr = __fix_to_virt(FIX_EARLYCON_MEM_BASE); |
2119 | + pgd_t *pgd = (pgd_t *)xen_start_info->pt_base; |
2120 | + pud_t *pud = pud_offset(pgd + pgd_index(addr), addr); |
2121 | + pmd_t *pmd = pmd_offset(pud, addr); |
2122 | + |
2123 | + swapper_pg_dir = pgd; |
2124 | + init_mm.pgd = pgd; |
2125 | + make_lowmem_page_readonly(swapper_pg_pmd, XENFEAT_writable_page_tables); |
2126 | + set_pmd(pmd, __pmd(__pa_symbol(swapper_pg_pmd) | _PAGE_TABLE)); |
2127 | + } |
2128 | } |
2129 | diff -Naur linux-2.6.25/arch/x86/Makefile linux-2.6.25-xen/arch/x86/Makefile |
2130 | --- linux-2.6.25/arch/x86/Makefile 2008-05-23 18:32:52.000000000 +0200 |
2131 | +++ linux-2.6.25-xen/arch/x86/Makefile 2008-05-23 18:27:40.000000000 +0200 |
2132 | @@ -173,7 +173,7 @@ |
2133 | core-y += $(fcore-y) |
2134 | |
2135 | # Xen paravirtualization support |
2136 | -core-$(CONFIG_XEN) += arch/x86/xen/ |
2137 | +core-$(CONFIG_PARAVIRT_XEN) += arch/x86/xen/ |
2138 | |
2139 | # lguest paravirtualization support |
2140 | core-$(CONFIG_LGUEST_GUEST) += arch/x86/lguest/ |
2141 | diff -Naur linux-2.6.25/arch/x86/mm/fault_32-xen.c linux-2.6.25-xen/arch/x86/mm/fault_32-xen.c |
2142 | --- linux-2.6.25/arch/x86/mm/fault_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2143 | +++ linux-2.6.25-xen/arch/x86/mm/fault_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2144 | @@ -346,7 +346,7 @@ |
2145 | pmd_k = pmd_offset(pud_k, address); |
2146 | if (!pmd_present(*pmd_k)) |
2147 | return NULL; |
2148 | - if (!pmd_present(*pmd)) |
2149 | + if (!pmd_present(*pmd)) { |
2150 | #if CONFIG_XEN_COMPAT > 0x030002 |
2151 | set_pmd(pmd, *pmd_k); |
2152 | #else |
2153 | @@ -356,7 +356,8 @@ |
2154 | */ |
2155 | set_pmd(pmd, __pmd(pmd_val(*pmd_k))); |
2156 | #endif |
2157 | - else |
2158 | + arch_flush_lazy_mmu_mode(); |
2159 | + } else |
2160 | BUG_ON(pmd_page(*pmd) != pmd_page(*pmd_k)); |
2161 | return pmd_k; |
2162 | } |
2163 | @@ -388,6 +389,8 @@ |
2164 | return 0; |
2165 | } |
2166 | |
2167 | +int show_unhandled_signals = 1; |
2168 | + |
2169 | /* |
2170 | * This routine handles page faults. It determines the address, |
2171 | * and the problem, and then passes it off to one of the appropriate |
2172 | @@ -408,6 +411,7 @@ |
2173 | struct vm_area_struct * vma; |
2174 | unsigned long address; |
2175 | int write, si_code; |
2176 | + int fault; |
2177 | |
2178 | /* get the address */ |
2179 | address = read_cr2(); |
2180 | @@ -541,20 +545,18 @@ |
2181 | * make sure we exit gracefully rather than endlessly redo |
2182 | * the fault. |
2183 | */ |
2184 | - switch (handle_mm_fault(mm, vma, address, write)) { |
2185 | - case VM_FAULT_MINOR: |
2186 | - tsk->min_flt++; |
2187 | - break; |
2188 | - case VM_FAULT_MAJOR: |
2189 | - tsk->maj_flt++; |
2190 | - break; |
2191 | - case VM_FAULT_SIGBUS: |
2192 | - goto do_sigbus; |
2193 | - case VM_FAULT_OOM: |
2194 | + fault = handle_mm_fault(mm, vma, address, write); |
2195 | + if (unlikely(fault & VM_FAULT_ERROR)) { |
2196 | + if (fault & VM_FAULT_OOM) |
2197 | goto out_of_memory; |
2198 | - default: |
2199 | - BUG(); |
2200 | + else if (fault & VM_FAULT_SIGBUS) |
2201 | + goto do_sigbus; |
2202 | + BUG(); |
2203 | } |
2204 | + if (fault & VM_FAULT_MAJOR) |
2205 | + tsk->maj_flt++; |
2206 | + else |
2207 | + tsk->min_flt++; |
2208 | |
2209 | /* |
2210 | * Did it hit the DOS screen memory VA from vm86 mode? |
2211 | @@ -589,6 +591,14 @@ |
2212 | if (is_prefetch(regs, address, error_code)) |
2213 | return; |
2214 | |
2215 | + if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) && |
2216 | + printk_ratelimit()) { |
2217 | + printk("%s%s[%d]: segfault at %08lx eip %08lx " |
2218 | + "esp %08lx error %lx\n", |
2219 | + tsk->pid > 1 ? KERN_INFO : KERN_EMERG, |
2220 | + tsk->comm, tsk->pid, address, regs->eip, |
2221 | + regs->esp, error_code); |
2222 | + } |
2223 | tsk->thread.cr2 = address; |
2224 | /* Kernel addresses are always protection faults */ |
2225 | tsk->thread.error_code = error_code | (address >= TASK_SIZE); |
2226 | diff -Naur linux-2.6.25/arch/x86/mm/fault_64-xen.c linux-2.6.25-xen/arch/x86/mm/fault_64-xen.c |
2227 | --- linux-2.6.25/arch/x86/mm/fault_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2228 | +++ linux-2.6.25-xen/arch/x86/mm/fault_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2229 | @@ -159,7 +159,9 @@ |
2230 | pmd_t *pmd; |
2231 | pte_t *pte; |
2232 | |
2233 | - pgd = __va(read_cr3() & PHYSICAL_PAGE_MASK); |
2234 | + pgd = (pgd_t *)read_cr3(); |
2235 | + |
2236 | + pgd = __va((unsigned long)pgd & PHYSICAL_PAGE_MASK); |
2237 | pgd += pgd_index(address); |
2238 | if (bad_address(pgd)) goto bad; |
2239 | printk("PGD %lx ", pgd_val(*pgd)); |
2240 | @@ -219,16 +221,6 @@ |
2241 | return 0; |
2242 | } |
2243 | |
2244 | -int unhandled_signal(struct task_struct *tsk, int sig) |
2245 | -{ |
2246 | - if (is_init(tsk)) |
2247 | - return 1; |
2248 | - if (tsk->ptrace & PT_PTRACED) |
2249 | - return 0; |
2250 | - return (tsk->sighand->action[sig-1].sa.sa_handler == SIG_IGN) || |
2251 | - (tsk->sighand->action[sig-1].sa.sa_handler == SIG_DFL); |
2252 | -} |
2253 | - |
2254 | static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs, |
2255 | unsigned long error_code) |
2256 | { |
2257 | @@ -302,8 +294,8 @@ |
2258 | return 0; |
2259 | } |
2260 | |
2261 | -int page_fault_trace = 0; |
2262 | -int exception_trace = 1; |
2263 | +static int page_fault_trace; |
2264 | +int show_unhandled_signals = 1; |
2265 | |
2266 | |
2267 | #define MEM_VERBOSE 1 |
2268 | @@ -372,7 +364,7 @@ |
2269 | struct vm_area_struct * vma; |
2270 | unsigned long address; |
2271 | const struct exception_table_entry *fixup; |
2272 | - int write; |
2273 | + int write, fault; |
2274 | unsigned long flags; |
2275 | siginfo_t info; |
2276 | |
2277 | @@ -384,7 +376,7 @@ |
2278 | prefetchw(&mm->mmap_sem); |
2279 | |
2280 | /* get the address */ |
2281 | - address = current_vcpu_info()->arch.cr2; |
2282 | + address = read_cr2(); |
2283 | |
2284 | info.si_code = SEGV_MAPERR; |
2285 | |
2286 | @@ -445,6 +437,13 @@ |
2287 | if (unlikely(in_atomic() || !mm)) |
2288 | goto bad_area_nosemaphore; |
2289 | |
2290 | + /* |
2291 | + * User-mode registers count as a user access even for any |
2292 | + * potential system fault or CPU buglet. |
2293 | + */ |
2294 | + if (user_mode_vm(regs)) |
2295 | + error_code |= PF_USER; |
2296 | + |
2297 | again: |
2298 | /* When running in the kernel we expect faults to occur only to |
2299 | * addresses in user space. All other faults represent errors in the |
2300 | @@ -511,19 +510,18 @@ |
2301 | * make sure we exit gracefully rather than endlessly redo |
2302 | * the fault. |
2303 | */ |
2304 | - switch (handle_mm_fault(mm, vma, address, write)) { |
2305 | - case VM_FAULT_MINOR: |
2306 | - tsk->min_flt++; |
2307 | - break; |
2308 | - case VM_FAULT_MAJOR: |
2309 | - tsk->maj_flt++; |
2310 | - break; |
2311 | - case VM_FAULT_SIGBUS: |
2312 | - goto do_sigbus; |
2313 | - default: |
2314 | - goto out_of_memory; |
2315 | + fault = handle_mm_fault(mm, vma, address, write); |
2316 | + if (unlikely(fault & VM_FAULT_ERROR)) { |
2317 | + if (fault & VM_FAULT_OOM) |
2318 | + goto out_of_memory; |
2319 | + else if (fault & VM_FAULT_SIGBUS) |
2320 | + goto do_sigbus; |
2321 | + BUG(); |
2322 | } |
2323 | - |
2324 | + if (fault & VM_FAULT_MAJOR) |
2325 | + tsk->maj_flt++; |
2326 | + else |
2327 | + tsk->min_flt++; |
2328 | up_read(&mm->mmap_sem); |
2329 | return; |
2330 | |
2331 | @@ -556,7 +554,8 @@ |
2332 | (address >> 32)) |
2333 | return; |
2334 | |
2335 | - if (exception_trace && unhandled_signal(tsk, SIGSEGV)) { |
2336 | + if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) && |
2337 | + printk_ratelimit()) { |
2338 | printk( |
2339 | "%s%s[%d]: segfault at %016lx rip %016lx rsp %016lx error %lx\n", |
2340 | tsk->pid > 1 ? KERN_INFO : KERN_EMERG, |
2341 | @@ -630,7 +629,7 @@ |
2342 | } |
2343 | printk("VM: killing process %s\n", tsk->comm); |
2344 | if (error_code & 4) |
2345 | - do_exit(SIGKILL); |
2346 | + do_group_exit(SIGKILL); |
2347 | goto no_context; |
2348 | |
2349 | do_sigbus: |
2350 | diff -Naur linux-2.6.25/arch/x86/mm/highmem_32-xen.c linux-2.6.25-xen/arch/x86/mm/highmem_32-xen.c |
2351 | --- linux-2.6.25/arch/x86/mm/highmem_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2352 | +++ linux-2.6.25-xen/arch/x86/mm/highmem_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2353 | @@ -34,17 +34,16 @@ |
2354 | /* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */ |
2355 | pagefault_disable(); |
2356 | |
2357 | - idx = type + KM_TYPE_NR*smp_processor_id(); |
2358 | - BUG_ON(!pte_none(*(kmap_pte-idx))); |
2359 | - |
2360 | if (!PageHighMem(page)) |
2361 | return page_address(page); |
2362 | |
2363 | + idx = type + KM_TYPE_NR*smp_processor_id(); |
2364 | vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx); |
2365 | + BUG_ON(!pte_none(*(kmap_pte-idx))); |
2366 | set_pte_at(&init_mm, vaddr, kmap_pte-idx, mk_pte(page, prot)); |
2367 | arch_flush_lazy_mmu_mode(); |
2368 | |
2369 | - return (void*) vaddr; |
2370 | + return (void *)vaddr; |
2371 | } |
2372 | |
2373 | void *kmap_atomic(struct page *page, enum km_type type) |
2374 | diff -Naur linux-2.6.25/arch/x86/mm/init_32-xen.c linux-2.6.25-xen/arch/x86/mm/init_32-xen.c |
2375 | --- linux-2.6.25/arch/x86/mm/init_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2376 | +++ linux-2.6.25-xen/arch/x86/mm/init_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2377 | @@ -98,7 +98,7 @@ |
2378 | #endif |
2379 | pte_t *page_table = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE); |
2380 | |
2381 | - paravirt_alloc_pt(__pa(page_table) >> PAGE_SHIFT); |
2382 | + paravirt_alloc_pt(&init_mm, __pa(page_table) >> PAGE_SHIFT); |
2383 | make_lowmem_page_readonly(page_table, |
2384 | XENFEAT_writable_page_tables); |
2385 | set_pmd(pmd, __pmd(__pa(page_table) | _PAGE_TABLE)); |
2386 | @@ -448,7 +448,7 @@ |
2387 | xen_pagetable_setup_done(pgd_base); |
2388 | } |
2389 | |
2390 | -#if defined(CONFIG_SOFTWARE_SUSPEND) || defined(CONFIG_ACPI_SLEEP) |
2391 | +#if defined(CONFIG_HIBERNATION) || defined(CONFIG_ACPI) |
2392 | /* |
2393 | * Swap suspend & friends need this for resume because things like the intel-agp |
2394 | * driver might have split up a kernel 4MB mapping. |
2395 | @@ -487,9 +487,13 @@ |
2396 | flush_tlb_all(); |
2397 | } |
2398 | |
2399 | +int nx_enabled = 0; |
2400 | + |
2401 | +#ifdef CONFIG_X86_PAE |
2402 | + |
2403 | static int disable_nx __initdata = 0; |
2404 | u64 __supported_pte_mask __read_mostly = ~_PAGE_NX; |
2405 | -EXPORT_SYMBOL(__supported_pte_mask); |
2406 | +EXPORT_SYMBOL_GPL(__supported_pte_mask); |
2407 | |
2408 | /* |
2409 | * noexec = on|off |
2410 | @@ -516,9 +520,6 @@ |
2411 | } |
2412 | early_param("noexec", noexec_setup); |
2413 | |
2414 | -int nx_enabled = 0; |
2415 | -#ifdef CONFIG_X86_PAE |
2416 | - |
2417 | static void __init set_nx(void) |
2418 | { |
2419 | unsigned int v[4], l, h; |
2420 | @@ -803,8 +804,7 @@ |
2421 | PTRS_PER_PMD*sizeof(pmd_t), |
2422 | PTRS_PER_PMD*sizeof(pmd_t), |
2423 | SLAB_PANIC, |
2424 | - pmd_ctor, |
2425 | - NULL); |
2426 | + pmd_ctor); |
2427 | if (!SHARED_KERNEL_PMD) { |
2428 | /* If we're in PAE mode and have a non-shared |
2429 | kernel pmd, then the pgd size must be a |
2430 | diff -Naur linux-2.6.25/arch/x86/mm/init_64-xen.c linux-2.6.25-xen/arch/x86/mm/init_64-xen.c |
2431 | --- linux-2.6.25/arch/x86/mm/init_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2432 | +++ linux-2.6.25-xen/arch/x86/mm/init_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2433 | @@ -66,6 +66,9 @@ |
2434 | DEFINE_PER_CPU(struct mmu_gather, mmu_gathers); |
2435 | extern unsigned long start_pfn; |
2436 | |
2437 | +extern pmd_t level2_fixmap_pgt[PTRS_PER_PMD]; |
2438 | +extern pte_t level1_fixmap_pgt[PTRS_PER_PTE]; |
2439 | + |
2440 | /* |
2441 | * Use this until direct mapping is established, i.e. before __va() is |
2442 | * available in init_memory_mapping(). |
2443 | @@ -353,6 +356,10 @@ |
2444 | set_pte_phys(address, phys, prot, 0); |
2445 | set_pte_phys(address, phys, prot, 1); |
2446 | break; |
2447 | + case FIX_EARLYCON_MEM_BASE: |
2448 | + xen_l1_entry_update(level1_fixmap_pgt + pte_index(address), |
2449 | + pfn_pte_ma(phys >> PAGE_SHIFT, prot)); |
2450 | + break; |
2451 | default: |
2452 | set_pte_phys_ma(address, phys, prot); |
2453 | break; |
2454 | @@ -588,6 +595,13 @@ |
2455 | __user_pgd(init_level4_pgt)[pgd_index(VSYSCALL_START)] = |
2456 | __pgd(__pa_symbol(level3_user_pgt) | _PAGE_TABLE); |
2457 | |
2458 | + /* Do an early initialization of the fixmap area. */ |
2459 | + addr = __fix_to_virt(FIX_EARLYCON_MEM_BASE); |
2460 | + level3_kernel_pgt[pud_index(addr)] = |
2461 | + __pud(__pa_symbol(level2_fixmap_pgt) | _PAGE_TABLE); |
2462 | + level2_fixmap_pgt[pmd_index(addr)] = |
2463 | + __pmd(__pa_symbol(level1_fixmap_pgt) | _PAGE_TABLE); |
2464 | + |
2465 | early_make_page_readonly(init_level4_pgt, |
2466 | XENFEAT_writable_page_tables); |
2467 | early_make_page_readonly(__user_pgd(init_level4_pgt), |
2468 | @@ -598,6 +612,10 @@ |
2469 | XENFEAT_writable_page_tables); |
2470 | early_make_page_readonly(level2_kernel_pgt, |
2471 | XENFEAT_writable_page_tables); |
2472 | + early_make_page_readonly(level2_fixmap_pgt, |
2473 | + XENFEAT_writable_page_tables); |
2474 | + early_make_page_readonly(level1_fixmap_pgt, |
2475 | + XENFEAT_writable_page_tables); |
2476 | |
2477 | if (!xen_feature(XENFEAT_writable_page_tables)) { |
2478 | xen_pgd_pin(__pa_symbol(init_level4_pgt)); |
2479 | @@ -1121,41 +1139,6 @@ |
2480 | return pfn_valid(pte_pfn(*pte)); |
2481 | } |
2482 | |
2483 | -#ifdef CONFIG_SYSCTL |
2484 | -#include <linux/sysctl.h> |
2485 | - |
2486 | -extern int exception_trace, page_fault_trace; |
2487 | - |
2488 | -static ctl_table debug_table2[] = { |
2489 | - { |
2490 | - .ctl_name = 99, |
2491 | - .procname = "exception-trace", |
2492 | - .data = &exception_trace, |
2493 | - .maxlen = sizeof(int), |
2494 | - .mode = 0644, |
2495 | - .proc_handler = proc_dointvec |
2496 | - }, |
2497 | - {} |
2498 | -}; |
2499 | - |
2500 | -static ctl_table debug_root_table2[] = { |
2501 | - { |
2502 | - .ctl_name = CTL_DEBUG, |
2503 | - .procname = "debug", |
2504 | - .mode = 0555, |
2505 | - .child = debug_table2 |
2506 | - }, |
2507 | - {} |
2508 | -}; |
2509 | - |
2510 | -static __init int x8664_sysctl_init(void) |
2511 | -{ |
2512 | - register_sysctl_table(debug_root_table2); |
2513 | - return 0; |
2514 | -} |
2515 | -__initcall(x8664_sysctl_init); |
2516 | -#endif |
2517 | - |
2518 | /* A pseudo VMA to allow ptrace access for the vsyscall page. This only |
2519 | covers the 64bit vsyscall page now. 32bit has a real VMA now and does |
2520 | not need special handling anymore. */ |
2521 | @@ -1194,9 +1177,18 @@ |
2522 | } |
2523 | |
2524 | #ifndef CONFIG_XEN |
2525 | -void *alloc_bootmem_high_node(pg_data_t *pgdat, unsigned long size) |
2526 | +void * __init alloc_bootmem_high_node(pg_data_t *pgdat, unsigned long size) |
2527 | { |
2528 | return __alloc_bootmem_core(pgdat->bdata, size, |
2529 | SMP_CACHE_BYTES, (4UL*1024*1024*1024), 0); |
2530 | } |
2531 | #endif |
2532 | + |
2533 | +const char *arch_vma_name(struct vm_area_struct *vma) |
2534 | +{ |
2535 | + if (vma->vm_mm && vma->vm_start == (long)vma->vm_mm->context.vdso) |
2536 | + return "[vdso]"; |
2537 | + if (vma == &gate_vma) |
2538 | + return "[vsyscall]"; |
2539 | + return NULL; |
2540 | +} |
2541 | diff -Naur linux-2.6.25/arch/x86/mm/ioremap_32-xen.c linux-2.6.25-xen/arch/x86/mm/ioremap_32-xen.c |
2542 | --- linux-2.6.25/arch/x86/mm/ioremap_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2543 | +++ linux-2.6.25-xen/arch/x86/mm/ioremap_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2544 | @@ -363,9 +363,8 @@ |
2545 | |
2546 | /* Reset the direct mapping. Can block */ |
2547 | if ((p->flags >> 20) && is_local_lowmem(p->phys_addr)) { |
2548 | - /* p->size includes the guard page, but cpa doesn't like that */ |
2549 | change_page_attr(virt_to_page(bus_to_virt(p->phys_addr)), |
2550 | - (p->size - PAGE_SIZE) >> PAGE_SHIFT, |
2551 | + get_vm_area_size(p) >> PAGE_SHIFT, |
2552 | PAGE_KERNEL); |
2553 | global_flush_tlb(); |
2554 | } |
2555 | diff -Naur linux-2.6.25/arch/x86/mm/pageattr_64-xen.c linux-2.6.25-xen/arch/x86/mm/pageattr_64-xen.c |
2556 | --- linux-2.6.25/arch/x86/mm/pageattr_64-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2557 | +++ linux-2.6.25-xen/arch/x86/mm/pageattr_64-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2558 | @@ -276,7 +276,7 @@ |
2559 | } |
2560 | #endif /* CONFIG_XEN */ |
2561 | |
2562 | -static inline pte_t *lookup_address(unsigned long address) |
2563 | +pte_t *lookup_address(unsigned long address) |
2564 | { |
2565 | pgd_t *pgd = pgd_offset_k(address); |
2566 | pud_t *pud; |
2567 | @@ -337,14 +337,13 @@ |
2568 | struct page *pg; |
2569 | |
2570 | /* When clflush is available always use it because it is |
2571 | - much cheaper than WBINVD. Disable clflush for now because |
2572 | - the high level code is not ready yet */ |
2573 | + much cheaper than WBINVD. */ |
2574 | + /* clflush is still broken. Disable for now. */ |
2575 | if (1 || !cpu_has_clflush) |
2576 | asm volatile("wbinvd" ::: "memory"); |
2577 | else list_for_each_entry(pg, l, lru) { |
2578 | void *adr = page_address(pg); |
2579 | - if (cpu_has_clflush) |
2580 | - cache_flush_page(adr); |
2581 | + cache_flush_page(adr); |
2582 | } |
2583 | __flush_tlb_all(); |
2584 | } |
2585 | @@ -358,7 +357,8 @@ |
2586 | |
2587 | static inline void save_page(struct page *fpage) |
2588 | { |
2589 | - list_add(&fpage->lru, &deferred_pages); |
2590 | + if (!test_and_set_bit(PG_arch_1, &fpage->flags)) |
2591 | + list_add(&fpage->lru, &deferred_pages); |
2592 | } |
2593 | |
2594 | /* |
2595 | @@ -392,9 +392,12 @@ |
2596 | pte_t *kpte; |
2597 | struct page *kpte_page; |
2598 | pgprot_t ref_prot2; |
2599 | + |
2600 | kpte = lookup_address(address); |
2601 | if (!kpte) return 0; |
2602 | kpte_page = virt_to_page(((unsigned long)kpte) & PAGE_MASK); |
2603 | + BUG_ON(PageLRU(kpte_page)); |
2604 | + BUG_ON(PageCompound(kpte_page)); |
2605 | if (pgprot_val(prot) != pgprot_val(ref_prot)) { |
2606 | if (!pte_huge(*kpte)) { |
2607 | set_pte(kpte, pfn_pte(pfn, prot)); |
2608 | @@ -433,10 +436,9 @@ |
2609 | return 0; |
2610 | #endif |
2611 | |
2612 | - if (page_private(kpte_page) == 0) { |
2613 | - save_page(kpte_page); |
2614 | + save_page(kpte_page); |
2615 | + if (page_private(kpte_page) == 0) |
2616 | revert_page(address, ref_prot); |
2617 | - } |
2618 | return 0; |
2619 | } |
2620 | |
2621 | @@ -508,6 +510,10 @@ |
2622 | flush_map(&l); |
2623 | |
2624 | list_for_each_entry_safe(pg, next, &l, lru) { |
2625 | + list_del(&pg->lru); |
2626 | + clear_bit(PG_arch_1, &pg->flags); |
2627 | + if (page_private(pg) != 0) |
2628 | + continue; |
2629 | ClearPagePrivate(pg); |
2630 | __free_page(pg); |
2631 | } |
2632 | diff -Naur linux-2.6.25/arch/x86/mm/pgtable_32-xen.c linux-2.6.25-xen/arch/x86/mm/pgtable_32-xen.c |
2633 | --- linux-2.6.25/arch/x86/mm/pgtable_32-xen.c 2008-05-23 18:33:27.000000000 +0200 |
2634 | +++ linux-2.6.25-xen/arch/x86/mm/pgtable_32-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2635 | @@ -242,7 +242,7 @@ |
2636 | |
2637 | #if (PTRS_PER_PMD == 1) |
2638 | /* Non-PAE pgd constructor */ |
2639 | -void pgd_ctor(void *pgd) |
2640 | +static void pgd_ctor(void *pgd) |
2641 | { |
2642 | unsigned long flags; |
2643 | |
2644 | @@ -265,7 +265,7 @@ |
2645 | } |
2646 | #else /* PTRS_PER_PMD > 1 */ |
2647 | /* PAE pgd constructor */ |
2648 | -void pgd_ctor(void *pgd) |
2649 | +static void pgd_ctor(void *pgd) |
2650 | { |
2651 | /* PAE, kernel PMD may be shared */ |
2652 | |
2653 | @@ -284,7 +284,7 @@ |
2654 | } |
2655 | #endif /* PTRS_PER_PMD */ |
2656 | |
2657 | -void pgd_dtor(void *pgd) |
2658 | +static void pgd_dtor(void *pgd) |
2659 | { |
2660 | unsigned long flags; /* can be called from interrupt context */ |
2661 | |
2662 | diff -Naur linux-2.6.25/arch/x86/pci/irq-xen.c linux-2.6.25-xen/arch/x86/pci/irq-xen.c |
2663 | --- linux-2.6.25/arch/x86/pci/irq-xen.c 2008-05-23 18:33:26.000000000 +0200 |
2664 | +++ linux-2.6.25-xen/arch/x86/pci/irq-xen.c 2008-05-23 18:27:40.000000000 +0200 |
2665 | @@ -142,8 +142,9 @@ |
2666 | for(i = 1; i < 256; i++) { |
2667 | if (!busmap[i] || pci_find_bus(0, i)) |
2668 | continue; |
2669 | - if (pci_scan_bus(i, &pci_root_ops, NULL)) |
2670 | - printk(KERN_INFO "PCI: Discovered primary peer bus %02x [IRQ]\n", i); |
2671 | + if (pci_scan_bus_with_sysdata(i)) |
2672 | + printk(KERN_INFO "PCI: Discovered primary peer " |
2673 | + "bus %02x [IRQ]\n", i); |
2674 | } |
2675 | pcibios_last_bus = -1; |
2676 | } |
2677 | @@ -553,6 +554,7 @@ |
2678 | case PCI_DEVICE_ID_INTEL_ICH9_3: |
2679 | case PCI_DEVICE_ID_INTEL_ICH9_4: |
2680 | case PCI_DEVICE_ID_INTEL_ICH9_5: |
2681 | + case PCI_DEVICE_ID_INTEL_TOLAPAI_0: |
2682 | r->name = "PIIX/ICH"; |
2683 | r->get = pirq_piix_get; |
2684 | r->set = pirq_piix_set; |
2685 | diff -Naur linux-2.6.25/arch/x86/vdso/vdso32/note.S linux-2.6.25-xen/arch/x86/vdso/vdso32/note.S |
2686 | --- linux-2.6.25/arch/x86/vdso/vdso32/note.S 2008-04-17 04:49:44.000000000 +0200 |
2687 | +++ linux-2.6.25-xen/arch/x86/vdso/vdso32/note.S 2008-05-23 18:27:40.000000000 +0200 |
2688 | @@ -13,7 +13,7 @@ |
2689 | .long LINUX_VERSION_CODE |
2690 | ELFNOTE_END |
2691 | |
2692 | -#ifdef CONFIG_XEN |
2693 | +#if defined(CONFIG_X86_XEN) || defined(CONFIG_PARAVIRT_XEN) |
2694 | /* |
2695 | * Add a special note telling glibc's dynamic linker a fake hardware |
2696 | * flavor that it will use to choose the search path for libraries in the |
2697 | @@ -37,8 +37,12 @@ |
2698 | |
2699 | ELFNOTE_START(GNU, 2, "a") |
2700 | .long 1 /* ncaps */ |
2701 | +#ifdef CONFIG_PARAVIRT_XEN |
2702 | VDSO32_NOTE_MASK: /* Symbol used by arch/x86/xen/setup.c */ |
2703 | .long 0 /* mask */ |
2704 | +#else |
2705 | + .long 1 << VDSO_NOTE_NONEGSEG_BIT /* mask */ |
2706 | +#endif |
2707 | .byte VDSO_NOTE_NONEGSEG_BIT; .asciz "nosegneg" /* bit, name */ |
2708 | ELFNOTE_END |
2709 | #endif |
2710 | diff -Naur linux-2.6.25/arch/x86/xen/Kconfig linux-2.6.25-xen/arch/x86/xen/Kconfig |
2711 | --- linux-2.6.25/arch/x86/xen/Kconfig 2008-04-17 04:49:44.000000000 +0200 |
2712 | +++ linux-2.6.25-xen/arch/x86/xen/Kconfig 2008-05-23 18:27:40.000000000 +0200 |
2713 | @@ -2,7 +2,7 @@ |
2714 | # This Kconfig describes xen options |
2715 | # |
2716 | |
2717 | -config XEN |
2718 | +config PARAVIRT_XEN |
2719 | bool "Xen guest support" |
2720 | select PARAVIRT |
2721 | depends on X86_32 |
2722 | diff -Naur linux-2.6.25/arch/x86/xen/xen-head.S linux-2.6.25-xen/arch/x86/xen/xen-head.S |
2723 | --- linux-2.6.25/arch/x86/xen/xen-head.S 2008-04-17 04:49:44.000000000 +0200 |
2724 | +++ linux-2.6.25-xen/arch/x86/xen/xen-head.S 2008-05-23 18:27:40.000000000 +0200 |
2725 | @@ -1,7 +1,7 @@ |
2726 | /* Xen-specific pieces of head.S, intended to be included in the right |
2727 | place in head.S */ |
2728 | |
2729 | -#ifdef CONFIG_XEN |
2730 | +#ifdef CONFIG_PARAVIRT_XEN |
2731 | |
2732 | #include <linux/elfnote.h> |
2733 | #include <linux/init.h> |
2734 | @@ -37,4 +37,4 @@ |
2735 | #endif |
2736 | ELFNOTE(Xen, XEN_ELFNOTE_LOADER, .asciz "generic") |
2737 | |
2738 | -#endif /*CONFIG_XEN */ |
2739 | +#endif /* CONFIG_PARAVIRT_XEN */ |
2740 | diff -Naur linux-2.6.25/drivers/block/Kconfig linux-2.6.25-xen/drivers/block/Kconfig |
2741 | --- linux-2.6.25/drivers/block/Kconfig 2008-04-17 04:49:44.000000000 +0200 |
2742 | +++ linux-2.6.25-xen/drivers/block/Kconfig 2008-05-23 18:27:40.000000000 +0200 |
2743 | @@ -417,9 +417,9 @@ |
2744 | help |
2745 | Include support for the Xilinx SystemACE CompactFlash interface |
2746 | |
2747 | -config XEN_BLKDEV_FRONTEND |
2748 | +config XEN_BLKFRONT |
2749 | tristate "Xen virtual block device support" |
2750 | - depends on XEN |
2751 | + depends on PARAVIRT_XEN |
2752 | default y |
2753 | help |
2754 | This driver implements the front-end of the Xen virtual |
2755 | diff -Naur linux-2.6.25/drivers/block/Makefile linux-2.6.25-xen/drivers/block/Makefile |
2756 | --- linux-2.6.25/drivers/block/Makefile 2008-04-17 04:49:44.000000000 +0200 |
2757 | +++ linux-2.6.25-xen/drivers/block/Makefile 2008-05-23 18:29:02.000000000 +0200 |
2758 | @@ -30,4 +30,4 @@ |
2759 | obj-$(CONFIG_BLK_DEV_SX8) += sx8.o |
2760 | obj-$(CONFIG_BLK_DEV_UB) += ub.o |
2761 | |
2762 | -obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += xen-blkfront.o |
2763 | +obj-$(CONFIG_XEN_BLKFRONT) += xen-blkfront.o |
2764 | diff -Naur linux-2.6.25/drivers/block/xen-blkfront.c linux-2.6.25-xen/drivers/block/xen-blkfront.c |
2765 | --- linux-2.6.25/drivers/block/xen-blkfront.c 2008-04-17 04:49:44.000000000 +0200 |
2766 | +++ linux-2.6.25-xen/drivers/block/xen-blkfront.c 2008-05-23 18:27:40.000000000 +0200 |
2767 | @@ -965,7 +965,6 @@ |
2768 | |
2769 | static struct xenbus_driver blkfront = { |
2770 | .name = "vbd", |
2771 | - .owner = THIS_MODULE, |
2772 | .ids = blkfront_ids, |
2773 | .probe = blkfront_probe, |
2774 | .remove = blkfront_remove, |
2775 | diff -Naur linux-2.6.25/drivers/char/Kconfig linux-2.6.25-xen/drivers/char/Kconfig |
2776 | --- linux-2.6.25/drivers/char/Kconfig 2008-04-17 04:49:44.000000000 +0200 |
2777 | +++ linux-2.6.25-xen/drivers/char/Kconfig 2008-05-23 18:27:40.000000000 +0200 |
2778 | @@ -584,7 +584,7 @@ |
2779 | |
2780 | config HVC_XEN |
2781 | bool "Xen Hypervisor Console support" |
2782 | - depends on XEN |
2783 | + depends on PARAVIRT_XEN |
2784 | select HVC_DRIVER |
2785 | default y |
2786 | help |
2787 | diff -Naur linux-2.6.25/drivers/Makefile linux-2.6.25-xen/drivers/Makefile |
2788 | --- linux-2.6.25/drivers/Makefile 2008-05-23 18:32:47.000000000 +0200 |
2789 | +++ linux-2.6.25-xen/drivers/Makefile 2008-05-23 18:27:40.000000000 +0200 |
2790 | @@ -19,7 +19,7 @@ |
2791 | obj-$(CONFIG_PNP) += pnp/ |
2792 | obj-$(CONFIG_ARM_AMBA) += amba/ |
2793 | |
2794 | -obj-$(CONFIG_XEN) += xen/ |
2795 | +obj-$(CONFIG_PARAVIRT_XEN) += xen/ |
2796 | |
2797 | obj-$(CONFIG_CONNECTOR) += connector/ |
2798 | |
2799 | diff -Naur linux-2.6.25/drivers/net/Kconfig linux-2.6.25-xen/drivers/net/Kconfig |
2800 | --- linux-2.6.25/drivers/net/Kconfig 2008-05-23 18:32:02.000000000 +0200 |
2801 | +++ linux-2.6.25-xen/drivers/net/Kconfig 2008-05-23 18:27:40.000000000 +0200 |
2802 | @@ -2690,9 +2690,9 @@ |
2803 | |
2804 | source "drivers/s390/net/Kconfig" |
2805 | |
2806 | -config XEN_NETDEV_FRONTEND |
2807 | +config XEN_NETFRONT |
2808 | tristate "Xen network device frontend driver" |
2809 | - depends on XEN |
2810 | + depends on PARAVIRT_XEN |
2811 | default y |
2812 | help |
2813 | The network device frontend driver allows the kernel to |
2814 | diff -Naur linux-2.6.25/drivers/net/Makefile linux-2.6.25-xen/drivers/net/Makefile |
2815 | --- linux-2.6.25/drivers/net/Makefile 2008-05-23 18:32:02.000000000 +0200 |
2816 | +++ linux-2.6.25-xen/drivers/net/Makefile 2008-05-23 18:27:40.000000000 +0200 |
2817 | @@ -140,7 +140,7 @@ |
2818 | obj-$(CONFIG_SLIP) += slip.o |
2819 | obj-$(CONFIG_SLHC) += slhc.o |
2820 | |
2821 | -obj-$(CONFIG_XEN_NETDEV_FRONTEND) += xen-netfront.o |
2822 | +obj-$(CONFIG_XEN_NETFRONT) += xen-netfront.o |
2823 | |
2824 | obj-$(CONFIG_DUMMY) += dummy.o |
2825 | obj-$(CONFIG_IFB) += ifb.o |
2826 | diff -Naur linux-2.6.25/drivers/net/xen-netfront.c linux-2.6.25-xen/drivers/net/xen-netfront.c |
2827 | --- linux-2.6.25/drivers/net/xen-netfront.c 2008-04-17 04:49:44.000000000 +0200 |
2828 | +++ linux-2.6.25-xen/drivers/net/xen-netfront.c 2008-05-23 18:27:40.000000000 +0200 |
2829 | @@ -36,8 +36,6 @@ |
2830 | #include <linux/skbuff.h> |
2831 | #include <linux/ethtool.h> |
2832 | #include <linux/if_ether.h> |
2833 | -#include <linux/tcp.h> |
2834 | -#include <linux/udp.h> |
2835 | #include <linux/moduleparam.h> |
2836 | #include <linux/mm.h> |
2837 | #include <net/ip.h> |
2838 | @@ -754,45 +752,6 @@ |
2839 | return cons; |
2840 | } |
2841 | |
2842 | -static int skb_checksum_setup(struct sk_buff *skb) |
2843 | -{ |
2844 | - struct iphdr *iph; |
2845 | - unsigned char *th; |
2846 | - int err = -EPROTO; |
2847 | - |
2848 | - if (skb->protocol != htons(ETH_P_IP)) |
2849 | - goto out; |
2850 | - |
2851 | - iph = (void *)skb->data; |
2852 | - th = skb->data + 4 * iph->ihl; |
2853 | - if (th >= skb_tail_pointer(skb)) |
2854 | - goto out; |
2855 | - |
2856 | - skb->csum_start = th - skb->head; |
2857 | - switch (iph->protocol) { |
2858 | - case IPPROTO_TCP: |
2859 | - skb->csum_offset = offsetof(struct tcphdr, check); |
2860 | - break; |
2861 | - case IPPROTO_UDP: |
2862 | - skb->csum_offset = offsetof(struct udphdr, check); |
2863 | - break; |
2864 | - default: |
2865 | - if (net_ratelimit()) |
2866 | - printk(KERN_ERR "Attempting to checksum a non-" |
2867 | - "TCP/UDP packet, dropping a protocol" |
2868 | - " %d packet", iph->protocol); |
2869 | - goto out; |
2870 | - } |
2871 | - |
2872 | - if ((th + skb->csum_offset + 2) > skb_tail_pointer(skb)) |
2873 | - goto out; |
2874 | - |
2875 | - err = 0; |
2876 | - |
2877 | -out: |
2878 | - return err; |
2879 | -} |
2880 | - |
2881 | static int handle_incoming_queue(struct net_device *dev, |
2882 | struct sk_buff_head *rxq) |
2883 | { |
2884 | @@ -1775,7 +1734,6 @@ |
2885 | |
2886 | static struct xenbus_driver netfront = { |
2887 | .name = "vif", |
2888 | - .owner = THIS_MODULE, |
2889 | .ids = netfront_ids, |
2890 | .probe = netfront_probe, |
2891 | .remove = __devexit_p(xennet_remove), |
2892 | diff -Naur linux-2.6.25/drivers/xen/blkback/blkback.c linux-2.6.25-xen/drivers/xen/blkback/blkback.c |
2893 | --- linux-2.6.25/drivers/xen/blkback/blkback.c 2008-05-23 18:33:26.000000000 +0200 |
2894 | +++ linux-2.6.25-xen/drivers/xen/blkback/blkback.c 2008-05-23 18:27:40.000000000 +0200 |
2895 | @@ -154,7 +154,7 @@ |
2896 | |
2897 | static void plug_queue(blkif_t *blkif, struct bio *bio) |
2898 | { |
2899 | - request_queue_t *q = bdev_get_queue(bio->bi_bdev); |
2900 | + struct request_queue *q = bdev_get_queue(bio->bi_bdev); |
2901 | |
2902 | if (q == blkif->plug) |
2903 | return; |
2904 | diff -Naur linux-2.6.25/drivers/xen/blkback/common.h linux-2.6.25-xen/drivers/xen/blkback/common.h |
2905 | --- linux-2.6.25/drivers/xen/blkback/common.h 2008-05-23 18:33:23.000000000 +0200 |
2906 | +++ linux-2.6.25-xen/drivers/xen/blkback/common.h 2008-05-23 18:27:40.000000000 +0200 |
2907 | @@ -79,7 +79,7 @@ |
2908 | wait_queue_head_t wq; |
2909 | struct task_struct *xenblkd; |
2910 | unsigned int waiting_reqs; |
2911 | - request_queue_t *plug; |
2912 | + struct request_queue *plug; |
2913 | |
2914 | /* statistics */ |
2915 | unsigned long st_print; |
2916 | diff -Naur linux-2.6.25/drivers/xen/blkback/interface.c linux-2.6.25-xen/drivers/xen/blkback/interface.c |
2917 | --- linux-2.6.25/drivers/xen/blkback/interface.c 2008-05-23 18:33:26.000000000 +0200 |
2918 | +++ linux-2.6.25-xen/drivers/xen/blkback/interface.c 2008-05-23 18:27:40.000000000 +0200 |
2919 | @@ -177,5 +177,5 @@ |
2920 | void __init blkif_interface_init(void) |
2921 | { |
2922 | blkif_cachep = kmem_cache_create("blkif_cache", sizeof(blkif_t), |
2923 | - 0, 0, NULL, NULL); |
2924 | + 0, 0, NULL); |
2925 | } |
2926 | diff -Naur linux-2.6.25/drivers/xen/blkfront/blkfront.c linux-2.6.25-xen/drivers/xen/blkfront/blkfront.c |
2927 | --- linux-2.6.25/drivers/xen/blkfront/blkfront.c 2008-05-23 18:33:27.000000000 +0200 |
2928 | +++ linux-2.6.25-xen/drivers/xen/blkfront/blkfront.c 2008-05-23 18:27:40.000000000 +0200 |
2929 | @@ -653,7 +653,7 @@ |
2930 | * do_blkif_request |
2931 | * read a block; request is in a request queue |
2932 | */ |
2933 | -void do_blkif_request(request_queue_t *rq) |
2934 | +void do_blkif_request(struct request_queue *rq) |
2935 | { |
2936 | struct blkfront_info *info = NULL; |
2937 | struct request *req; |
2938 | diff -Naur linux-2.6.25/drivers/xen/blkfront/block.h linux-2.6.25-xen/drivers/xen/blkfront/block.h |
2939 | --- linux-2.6.25/drivers/xen/blkfront/block.h 2008-05-23 18:32:32.000000000 +0200 |
2940 | +++ linux-2.6.25-xen/drivers/xen/blkfront/block.h 2008-05-23 18:27:40.000000000 +0200 |
2941 | @@ -105,7 +105,7 @@ |
2942 | blkif_front_ring_t ring; |
2943 | unsigned int irq; |
2944 | struct xlbd_major_info *mi; |
2945 | - request_queue_t *rq; |
2946 | + struct request_queue *rq; |
2947 | struct work_struct work; |
2948 | struct gnttab_free_callback callback; |
2949 | struct blk_shadow shadow[BLK_RING_SIZE]; |
2950 | @@ -129,7 +129,7 @@ |
2951 | extern int blkif_getgeo(struct block_device *, struct hd_geometry *); |
2952 | extern int blkif_check(dev_t dev); |
2953 | extern int blkif_revalidate(dev_t dev); |
2954 | -extern void do_blkif_request (request_queue_t *rq); |
2955 | +extern void do_blkif_request (struct request_queue *rq); |
2956 | |
2957 | /* Virtual block-device subsystem. */ |
2958 | /* Note that xlvbd_add doesn't call add_disk for you: you're expected |
2959 | diff -Naur linux-2.6.25/drivers/xen/blkfront/vbd.c linux-2.6.25-xen/drivers/xen/blkfront/vbd.c |
2960 | --- linux-2.6.25/drivers/xen/blkfront/vbd.c 2008-05-23 18:32:32.000000000 +0200 |
2961 | +++ linux-2.6.25-xen/drivers/xen/blkfront/vbd.c 2008-05-23 18:27:40.000000000 +0200 |
2962 | @@ -186,7 +186,7 @@ |
2963 | static int |
2964 | xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size) |
2965 | { |
2966 | - request_queue_t *rq; |
2967 | + struct request_queue *rq; |
2968 | |
2969 | rq = blk_init_queue(do_blkif_request, &blkif_io_lock); |
2970 | if (rq == NULL) |
2971 | diff -Naur linux-2.6.25/drivers/xen/blktap/common.h linux-2.6.25-xen/drivers/xen/blktap/common.h |
2972 | --- linux-2.6.25/drivers/xen/blktap/common.h 2008-05-23 18:33:23.000000000 +0200 |
2973 | +++ linux-2.6.25-xen/drivers/xen/blktap/common.h 2008-05-23 18:27:40.000000000 +0200 |
2974 | @@ -68,7 +68,7 @@ |
2975 | wait_queue_head_t wq; |
2976 | struct task_struct *xenblkd; |
2977 | unsigned int waiting_reqs; |
2978 | - request_queue_t *plug; |
2979 | + struct request_queue *plug; |
2980 | |
2981 | /* statistics */ |
2982 | unsigned long st_print; |
2983 | diff -Naur linux-2.6.25/drivers/xen/blktap/interface.c linux-2.6.25-xen/drivers/xen/blktap/interface.c |
2984 | --- linux-2.6.25/drivers/xen/blktap/interface.c 2008-05-23 18:33:26.000000000 +0200 |
2985 | +++ linux-2.6.25-xen/drivers/xen/blktap/interface.c 2008-05-23 18:27:40.000000000 +0200 |
2986 | @@ -170,5 +170,5 @@ |
2987 | void __init tap_blkif_interface_init(void) |
2988 | { |
2989 | blkif_cachep = kmem_cache_create("blktapif_cache", sizeof(blkif_t), |
2990 | - 0, 0, NULL, NULL); |
2991 | + 0, 0, NULL); |
2992 | } |
2993 | diff -Naur linux-2.6.25/drivers/xen/core/features.c linux-2.6.25-xen/drivers/xen/core/features.c |
2994 | --- linux-2.6.25/drivers/xen/core/features.c 2008-05-23 18:32:32.000000000 +0200 |
2995 | +++ linux-2.6.25-xen/drivers/xen/core/features.c 2008-05-23 18:27:40.000000000 +0200 |
2996 | @@ -19,7 +19,7 @@ |
2997 | /* Not a GPL symbol: used in ubiquitous macros, so too restrictive. */ |
2998 | EXPORT_SYMBOL(xen_features); |
2999 | |
3000 | -void setup_xen_features(void) |
3001 | +void xen_setup_features(void) |
3002 | { |
3003 | xen_feature_info_t fi; |
3004 | int i, j; |
3005 | diff -Naur linux-2.6.25/drivers/xen/core/reboot.c linux-2.6.25-xen/drivers/xen/core/reboot.c |
3006 | --- linux-2.6.25/drivers/xen/core/reboot.c 2008-05-23 18:33:26.000000000 +0200 |
3007 | +++ linux-2.6.25-xen/drivers/xen/core/reboot.c 2008-05-23 18:27:40.000000000 +0200 |
3008 | @@ -4,6 +4,7 @@ |
3009 | #include <linux/unistd.h> |
3010 | #include <linux/module.h> |
3011 | #include <linux/reboot.h> |
3012 | +#include <linux/sched.h> |
3013 | #include <linux/sysrq.h> |
3014 | #include <asm/hypervisor.h> |
3015 | #include <xen/xenbus.h> |
3016 | diff -Naur linux-2.6.25/drivers/xen/Makefile linux-2.6.25-xen/drivers/xen/Makefile |
3017 | --- linux-2.6.25/drivers/xen/Makefile 2008-05-23 18:33:03.000000000 +0200 |
3018 | +++ linux-2.6.25-xen/drivers/xen/Makefile 2008-05-23 18:27:40.000000000 +0200 |
3019 | @@ -1,10 +1,12 @@ |
3020 | -obj-y += core/ |
3021 | -obj-y += console/ |
3022 | -obj-y += evtchn/ |
3023 | -obj-y += xenbus/ |
3024 | -obj-y += char/ |
3025 | +obj-$(CONFIG_PARAVIRT_XEN) += grant-table.o |
3026 | |
3027 | -obj-y += util.o |
3028 | +obj-$(CONFIG_XEN) += core/ |
3029 | +obj-$(CONFIG_XEN) += console/ |
3030 | +obj-$(CONFIG_XEN) += evtchn/ |
3031 | +obj-y += xenbus/ |
3032 | +obj-$(CONFIG_XEN) += char/ |
3033 | + |
3034 | +obj-$(CONFIG_XEN) += util.o |
3035 | obj-$(CONFIG_XEN_BALLOON) += balloon/ |
3036 | obj-$(CONFIG_XEN_BLKDEV_BACKEND) += blkback/ |
3037 | obj-$(CONFIG_XEN_BLKDEV_TAP) += blktap/ |
3038 | diff -Naur linux-2.6.25/drivers/xen/tpmback/interface.c linux-2.6.25-xen/drivers/xen/tpmback/interface.c |
3039 | --- linux-2.6.25/drivers/xen/tpmback/interface.c 2008-05-23 18:33:27.000000000 +0200 |
3040 | +++ linux-2.6.25-xen/drivers/xen/tpmback/interface.c 2008-05-23 18:27:40.000000000 +0200 |
3041 | @@ -12,6 +12,7 @@ |
3042 | */ |
3043 | |
3044 | #include "common.h" |
3045 | +#include <linux/err.h> |
3046 | #include <xen/balloon.h> |
3047 | #include <xen/gnttab.h> |
3048 | |
3049 | @@ -159,7 +160,7 @@ |
3050 | int __init tpmif_interface_init(void) |
3051 | { |
3052 | tpmif_cachep = kmem_cache_create("tpmif_cache", sizeof (tpmif_t), |
3053 | - 0, 0, NULL, NULL); |
3054 | + 0, 0, NULL); |
3055 | return tpmif_cachep ? 0 : -ENOMEM; |
3056 | } |
3057 | |
3058 | diff -Naur linux-2.6.25/drivers/xen/util.c linux-2.6.25-xen/drivers/xen/util.c |
3059 | --- linux-2.6.25/drivers/xen/util.c 2008-05-23 18:32:32.000000000 +0200 |
3060 | +++ linux-2.6.25-xen/drivers/xen/util.c 2008-05-23 18:27:40.000000000 +0200 |
3061 | @@ -1,8 +1,5 @@ |
3062 | -#include <linux/mm.h> |
3063 | +#include <linux/err.h> |
3064 | #include <linux/module.h> |
3065 | -#include <linux/slab.h> |
3066 | -#include <linux/vmalloc.h> |
3067 | -#include <asm/uaccess.h> |
3068 | #include <xen/driver_util.h> |
3069 | |
3070 | struct class *get_xen_class(void) |
3071 | @@ -21,45 +18,3 @@ |
3072 | return xen_class; |
3073 | } |
3074 | EXPORT_SYMBOL_GPL(get_xen_class); |
3075 | - |
3076 | -#ifdef CONFIG_X86 |
3077 | -static int f(pte_t *pte, struct page *pmd_page, unsigned long addr, void *data) |
3078 | -{ |
3079 | - /* apply_to_page_range() does all the hard work. */ |
3080 | - return 0; |
3081 | -} |
3082 | - |
3083 | -struct vm_struct *alloc_vm_area(unsigned long size) |
3084 | -{ |
3085 | - struct vm_struct *area; |
3086 | - |
3087 | - area = get_vm_area(size, VM_IOREMAP); |
3088 | - if (area == NULL) |
3089 | - return NULL; |
3090 | - |
3091 | - /* |
3092 | - * This ensures that page tables are constructed for this region |
3093 | - * of kernel virtual address space and mapped into init_mm. |
3094 | - */ |
3095 | - if (apply_to_page_range(&init_mm, (unsigned long)area->addr, |
3096 | - area->size, f, NULL)) { |
3097 | - free_vm_area(area); |
3098 | - return NULL; |
3099 | - } |
3100 | - |
3101 | - /* Map page directories into every address space. */ |
3102 | - vmalloc_sync_all(); |
3103 | - |
3104 | - return area; |
3105 | -} |
3106 | -EXPORT_SYMBOL_GPL(alloc_vm_area); |
3107 | - |
3108 | -void free_vm_area(struct vm_struct *area) |
3109 | -{ |
3110 | - struct vm_struct *ret; |
3111 | - ret = remove_vm_area(area->addr); |
3112 | - BUG_ON(ret != area); |
3113 | - kfree(area); |
3114 | -} |
3115 | -EXPORT_SYMBOL_GPL(free_vm_area); |
3116 | -#endif /* CONFIG_X86 */ |
3117 | diff -Naur linux-2.6.25/drivers/xen/xenbus/xenbus_client.c linux-2.6.25-xen/drivers/xen/xenbus/xenbus_client.c |
3118 | --- linux-2.6.25/drivers/xen/xenbus/xenbus_client.c 2008-05-23 18:33:03.000000000 +0200 |
3119 | +++ linux-2.6.25-xen/drivers/xen/xenbus/xenbus_client.c 2008-05-23 18:27:40.000000000 +0200 |
3120 | @@ -30,19 +30,26 @@ |
3121 | * IN THE SOFTWARE. |
3122 | */ |
3123 | |
3124 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3125 | #include <linux/slab.h> |
3126 | #include <xen/evtchn.h> |
3127 | #include <xen/gnttab.h> |
3128 | -#include <xen/xenbus.h> |
3129 | #include <xen/driver_util.h> |
3130 | +#else |
3131 | +#include <linux/types.h> |
3132 | +#include <linux/vmalloc.h> |
3133 | +#include <asm/xen/hypervisor.h> |
3134 | +#include <xen/interface/xen.h> |
3135 | +#include <xen/interface/event_channel.h> |
3136 | +#include <xen/events.h> |
3137 | +#include <xen/grant_table.h> |
3138 | +#endif |
3139 | +#include <xen/xenbus.h> |
3140 | |
3141 | #ifdef HAVE_XEN_PLATFORM_COMPAT_H |
3142 | #include <xen/platform-compat.h> |
3143 | #endif |
3144 | |
3145 | -#define DPRINTK(fmt, args...) \ |
3146 | - pr_debug("xenbus_client (%s:%d) " fmt ".\n", __FUNCTION__, __LINE__, ##args) |
3147 | - |
3148 | const char *xenbus_strstate(enum xenbus_state state) |
3149 | { |
3150 | static const char *const name[] = { |
3151 | @@ -58,6 +65,20 @@ |
3152 | } |
3153 | EXPORT_SYMBOL_GPL(xenbus_strstate); |
3154 | |
3155 | +/** |
3156 | + * xenbus_watch_path - register a watch |
3157 | + * @dev: xenbus device |
3158 | + * @path: path to watch |
3159 | + * @watch: watch to register |
3160 | + * @callback: callback to register |
3161 | + * |
3162 | + * Register a @watch on the given path, using the given xenbus_watch structure |
3163 | + * for storage, and the given @callback function as the callback. Return 0 on |
3164 | + * success, or -errno on error. On success, the given @path will be saved as |
3165 | + * @watch->node, and remains the caller's to free. On error, @watch->node will |
3166 | + * be NULL, the device will switch to %XenbusStateClosing, and the error will |
3167 | + * be saved in the store. |
3168 | + */ |
3169 | int xenbus_watch_path(struct xenbus_device *dev, const char *path, |
3170 | struct xenbus_watch *watch, |
3171 | void (*callback)(struct xenbus_watch *, |
3172 | @@ -81,6 +102,7 @@ |
3173 | EXPORT_SYMBOL_GPL(xenbus_watch_path); |
3174 | |
3175 | |
3176 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3177 | int xenbus_watch_path2(struct xenbus_device *dev, const char *path, |
3178 | const char *path2, struct xenbus_watch *watch, |
3179 | void (*callback)(struct xenbus_watch *, |
3180 | @@ -99,8 +121,60 @@ |
3181 | return err; |
3182 | } |
3183 | EXPORT_SYMBOL_GPL(xenbus_watch_path2); |
3184 | +#else |
3185 | +/** |
3186 | + * xenbus_watch_pathfmt - register a watch on a sprintf-formatted path |
3187 | + * @dev: xenbus device |
3188 | + * @watch: watch to register |
3189 | + * @callback: callback to register |
3190 | + * @pathfmt: format of path to watch |
3191 | + * |
3192 | + * Register a watch on the given @path, using the given xenbus_watch |
3193 | + * structure for storage, and the given @callback function as the callback. |
3194 | + * Return 0 on success, or -errno on error. On success, the watched path |
3195 | + * (@path/@path2) will be saved as @watch->node, and becomes the caller's to |
3196 | + * kfree(). On error, watch->node will be NULL, so the caller has nothing to |
3197 | + * free, the device will switch to %XenbusStateClosing, and the error will be |
3198 | + * saved in the store. |
3199 | + */ |
3200 | +int xenbus_watch_pathfmt(struct xenbus_device *dev, |
3201 | + struct xenbus_watch *watch, |
3202 | + void (*callback)(struct xenbus_watch *, |
3203 | + const char **, unsigned int), |
3204 | + const char *pathfmt, ...) |
3205 | +{ |
3206 | + int err; |
3207 | + va_list ap; |
3208 | + char *path; |
3209 | |
3210 | + va_start(ap, pathfmt); |
3211 | + path = kvasprintf(GFP_KERNEL, pathfmt, ap); |
3212 | + va_end(ap); |
3213 | |
3214 | + if (!path) { |
3215 | + xenbus_dev_fatal(dev, -ENOMEM, "allocating path for watch"); |
3216 | + return -ENOMEM; |
3217 | + } |
3218 | + err = xenbus_watch_path(dev, path, watch, callback); |
3219 | + |
3220 | + if (err) |
3221 | + kfree(path); |
3222 | + return err; |
3223 | +} |
3224 | +EXPORT_SYMBOL_GPL(xenbus_watch_pathfmt); |
3225 | +#endif |
3226 | + |
3227 | + |
3228 | +/** |
3229 | + * xenbus_switch_state |
3230 | + * @dev: xenbus device |
3231 | + * @xbt: transaction handle |
3232 | + * @state: new state |
3233 | + * |
3234 | + * Advertise in the store a change of the given driver to the given new_state. |
3235 | + * Return 0 on success, or -errno on error. On error, the device will switch |
3236 | + * to XenbusStateClosing, and the error will be saved in the store. |
3237 | + */ |
3238 | int xenbus_switch_state(struct xenbus_device *dev, enum xenbus_state state) |
3239 | { |
3240 | /* We check whether the state is currently set to the given value, and |
3241 | @@ -159,8 +233,8 @@ |
3242 | } |
3243 | |
3244 | |
3245 | -void _dev_error(struct xenbus_device *dev, int err, const char *fmt, |
3246 | - va_list ap) |
3247 | +static void _dev_error(struct xenbus_device *dev, int err, |
3248 | + const char *fmt, va_list ap) |
3249 | { |
3250 | int ret; |
3251 | unsigned int len; |
3252 | @@ -181,14 +255,16 @@ |
3253 | path_buffer = error_path(dev); |
3254 | |
3255 | if (path_buffer == NULL) { |
3256 | - printk("xenbus: failed to write error node for %s (%s)\n", |
3257 | - dev->nodename, printf_buffer); |
3258 | + dev_err(&dev->dev, |
3259 | + "xenbus: failed to write error node for %s (%s)\n", |
3260 | + dev->nodename, printf_buffer); |
3261 | goto fail; |
3262 | } |
3263 | |
3264 | if (xenbus_write(XBT_NIL, path_buffer, "error", printf_buffer) != 0) { |
3265 | - printk("xenbus: failed to write error node for %s (%s)\n", |
3266 | - dev->nodename, printf_buffer); |
3267 | + dev_err(&dev->dev, |
3268 | + "xenbus: failed to write error node for %s (%s)\n", |
3269 | + dev->nodename, printf_buffer); |
3270 | goto fail; |
3271 | } |
3272 | |
3273 | @@ -200,6 +276,15 @@ |
3274 | } |
3275 | |
3276 | |
3277 | +/** |
3278 | + * xenbus_dev_error |
3279 | + * @dev: xenbus device |
3280 | + * @err: error to report |
3281 | + * @fmt: error message format |
3282 | + * |
3283 | + * Report the given negative errno into the store, along with the given |
3284 | + * formatted message. |
3285 | + */ |
3286 | void xenbus_dev_error(struct xenbus_device *dev, int err, const char *fmt, |
3287 | ...) |
3288 | { |
3289 | @@ -212,6 +297,16 @@ |
3290 | EXPORT_SYMBOL_GPL(xenbus_dev_error); |
3291 | |
3292 | |
3293 | +/** |
3294 | + * xenbus_dev_fatal |
3295 | + * @dev: xenbus device |
3296 | + * @err: error to report |
3297 | + * @fmt: error message format |
3298 | + * |
3299 | + * Equivalent to xenbus_dev_error(dev, err, fmt, args), followed by |
3300 | + * xenbus_switch_state(dev, NULL, XenbusStateClosing) to schedule an orderly |
3301 | + * closedown of this driver and its peer. |
3302 | + */ |
3303 | void xenbus_dev_fatal(struct xenbus_device *dev, int err, const char *fmt, |
3304 | ...) |
3305 | { |
3306 | @@ -226,6 +321,15 @@ |
3307 | EXPORT_SYMBOL_GPL(xenbus_dev_fatal); |
3308 | |
3309 | |
3310 | +/** |
3311 | + * xenbus_grant_ring |
3312 | + * @dev: xenbus device |
3313 | + * @ring_mfn: mfn of ring to grant |
3314 | + * |
3315 | + * Grant access to the given @ring_mfn to the peer of the given device. Return |
3316 | + * 0 on success, or -errno on error. On error, the device will switch to |
3317 | + * XenbusStateClosing, and the error will be saved in the store. |
3318 | + */ |
3319 | int xenbus_grant_ring(struct xenbus_device *dev, unsigned long ring_mfn) |
3320 | { |
3321 | int err = gnttab_grant_foreign_access(dev->otherend_id, ring_mfn, 0); |
3322 | @@ -236,6 +340,12 @@ |
3323 | EXPORT_SYMBOL_GPL(xenbus_grant_ring); |
3324 | |
3325 | |
3326 | +/** |
3327 | + * Allocate an event channel for the given xenbus_device, assigning the newly |
3328 | + * created local port to *port. Return 0 on success, or -errno on error. On |
3329 | + * error, the device will switch to XenbusStateClosing, and the error will be |
3330 | + * saved in the store. |
3331 | + */ |
3332 | int xenbus_alloc_evtchn(struct xenbus_device *dev, int *port) |
3333 | { |
3334 | struct evtchn_alloc_unbound alloc_unbound; |
3335 | @@ -256,6 +366,38 @@ |
3336 | EXPORT_SYMBOL_GPL(xenbus_alloc_evtchn); |
3337 | |
3338 | |
3339 | +#if 0 /* !defined(CONFIG_XEN) && !defined(MODULE) */ |
3340 | +/** |
3341 | + * Bind to an existing interdomain event channel in another domain. Returns 0 |
3342 | + * on success and stores the local port in *port. On error, returns -errno, |
3343 | + * switches the device to XenbusStateClosing, and saves the error in XenStore. |
3344 | + */ |
3345 | +int xenbus_bind_evtchn(struct xenbus_device *dev, int remote_port, int *port) |
3346 | +{ |
3347 | + struct evtchn_bind_interdomain bind_interdomain; |
3348 | + int err; |
3349 | + |
3350 | + bind_interdomain.remote_dom = dev->otherend_id; |
3351 | + bind_interdomain.remote_port = remote_port; |
3352 | + |
3353 | + err = HYPERVISOR_event_channel_op(EVTCHNOP_bind_interdomain, |
3354 | + &bind_interdomain); |
3355 | + if (err) |
3356 | + xenbus_dev_fatal(dev, err, |
3357 | + "binding to event channel %d from domain %d", |
3358 | + remote_port, dev->otherend_id); |
3359 | + else |
3360 | + *port = bind_interdomain.local_port; |
3361 | + |
3362 | + return err; |
3363 | +} |
3364 | +EXPORT_SYMBOL_GPL(xenbus_bind_evtchn); |
3365 | +#endif |
3366 | + |
3367 | + |
3368 | +/** |
3369 | + * Free an existing event channel. Returns 0 on success or -errno on error. |
3370 | + */ |
3371 | int xenbus_free_evtchn(struct xenbus_device *dev, int port) |
3372 | { |
3373 | struct evtchn_close close; |
3374 | @@ -272,6 +414,191 @@ |
3375 | EXPORT_SYMBOL_GPL(xenbus_free_evtchn); |
3376 | |
3377 | |
3378 | +#if 0 /* !defined(CONFIG_XEN) && !defined(MODULE) */ |
3379 | +/** |
3380 | + * xenbus_map_ring_valloc |
3381 | + * @dev: xenbus device |
3382 | + * @gnt_ref: grant reference |
3383 | + * @vaddr: pointer to address to be filled out by mapping |
3384 | + * |
3385 | + * Based on Rusty Russell's skeleton driver's map_page. |
3386 | + * Map a page of memory into this domain from another domain's grant table. |
3387 | + * xenbus_map_ring_valloc allocates a page of virtual address space, maps the |
3388 | + * page to that address, and sets *vaddr to that address. |
3389 | + * Returns 0 on success, and GNTST_* (see xen/include/interface/grant_table.h) |
3390 | + * or -ENOMEM on error. If an error is returned, device will switch to |
3391 | + * XenbusStateClosing and the error message will be saved in XenStore. |
3392 | + */ |
3393 | +int xenbus_map_ring_valloc(struct xenbus_device *dev, int gnt_ref, void **vaddr) |
3394 | +{ |
3395 | + struct gnttab_map_grant_ref op = { |
3396 | + .flags = GNTMAP_host_map, |
3397 | + .ref = gnt_ref, |
3398 | + .dom = dev->otherend_id, |
3399 | + }; |
3400 | + struct vm_struct *area; |
3401 | + |
3402 | + *vaddr = NULL; |
3403 | + |
3404 | + area = alloc_vm_area(PAGE_SIZE); |
3405 | + if (!area) |
3406 | + return -ENOMEM; |
3407 | + |
3408 | + op.host_addr = (unsigned long)area->addr; |
3409 | + |
3410 | + if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1)) |
3411 | + BUG(); |
3412 | + |
3413 | + if (op.status != GNTST_okay) { |
3414 | + free_vm_area(area); |
3415 | + xenbus_dev_fatal(dev, op.status, |
3416 | + "mapping in shared page %d from domain %d", |
3417 | + gnt_ref, dev->otherend_id); |
3418 | + return op.status; |
3419 | + } |
3420 | + |
3421 | + /* Stuff the handle in an unused field */ |
3422 | + area->phys_addr = (unsigned long)op.handle; |
3423 | + |
3424 | + *vaddr = area->addr; |
3425 | + return 0; |
3426 | +} |
3427 | +EXPORT_SYMBOL_GPL(xenbus_map_ring_valloc); |
3428 | + |
3429 | + |
3430 | +/** |
3431 | + * xenbus_map_ring |
3432 | + * @dev: xenbus device |
3433 | + * @gnt_ref: grant reference |
3434 | + * @handle: pointer to grant handle to be filled |
3435 | + * @vaddr: address to be mapped to |
3436 | + * |
3437 | + * Map a page of memory into this domain from another domain's grant table. |
3438 | + * xenbus_map_ring does not allocate the virtual address space (you must do |
3439 | + * this yourself!). It only maps in the page to the specified address. |
3440 | + * Returns 0 on success, and GNTST_* (see xen/include/interface/grant_table.h) |
3441 | + * or -ENOMEM on error. If an error is returned, device will switch to |
3442 | + * XenbusStateClosing and the error message will be saved in XenStore. |
3443 | + */ |
3444 | +int xenbus_map_ring(struct xenbus_device *dev, int gnt_ref, |
3445 | + grant_handle_t *handle, void *vaddr) |
3446 | +{ |
3447 | + struct gnttab_map_grant_ref op = { |
3448 | + .host_addr = (unsigned long)vaddr, |
3449 | + .flags = GNTMAP_host_map, |
3450 | + .ref = gnt_ref, |
3451 | + .dom = dev->otherend_id, |
3452 | + }; |
3453 | + |
3454 | + if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1)) |
3455 | + BUG(); |
3456 | + |
3457 | + if (op.status != GNTST_okay) { |
3458 | + xenbus_dev_fatal(dev, op.status, |
3459 | + "mapping in shared page %d from domain %d", |
3460 | + gnt_ref, dev->otherend_id); |
3461 | + } else |
3462 | + *handle = op.handle; |
3463 | + |
3464 | + return op.status; |
3465 | +} |
3466 | +EXPORT_SYMBOL_GPL(xenbus_map_ring); |
3467 | + |
3468 | + |
3469 | +/** |
3470 | + * xenbus_unmap_ring_vfree |
3471 | + * @dev: xenbus device |
3472 | + * @vaddr: addr to unmap |
3473 | + * |
3474 | + * Based on Rusty Russell's skeleton driver's unmap_page. |
3475 | + * Unmap a page of memory in this domain that was imported from another domain. |
3476 | + * Use xenbus_unmap_ring_vfree if you mapped in your memory with |
3477 | + * xenbus_map_ring_valloc (it will free the virtual address space). |
3478 | + * Returns 0 on success and returns GNTST_* on error |
3479 | + * (see xen/include/interface/grant_table.h). |
3480 | + */ |
3481 | +int xenbus_unmap_ring_vfree(struct xenbus_device *dev, void *vaddr) |
3482 | +{ |
3483 | + struct vm_struct *area; |
3484 | + struct gnttab_unmap_grant_ref op = { |
3485 | + .host_addr = (unsigned long)vaddr, |
3486 | + }; |
3487 | + |
3488 | + /* It'd be nice if linux/vmalloc.h provided a find_vm_area(void *addr) |
3489 | + * method so that we don't have to muck with vmalloc internals here. |
3490 | + * We could force the user to hang on to their struct vm_struct from |
3491 | + * xenbus_map_ring_valloc, but these 6 lines considerably simplify |
3492 | + * this API. |
3493 | + */ |
3494 | + read_lock(&vmlist_lock); |
3495 | + for (area = vmlist; area != NULL; area = area->next) { |
3496 | + if (area->addr == vaddr) |
3497 | + break; |
3498 | + } |
3499 | + read_unlock(&vmlist_lock); |
3500 | + |
3501 | + if (!area) { |
3502 | + xenbus_dev_error(dev, -ENOENT, |
3503 | + "can't find mapped virtual address %p", vaddr); |
3504 | + return GNTST_bad_virt_addr; |
3505 | + } |
3506 | + |
3507 | + op.handle = (grant_handle_t)area->phys_addr; |
3508 | + |
3509 | + if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1)) |
3510 | + BUG(); |
3511 | + |
3512 | + if (op.status == GNTST_okay) |
3513 | + free_vm_area(area); |
3514 | + else |
3515 | + xenbus_dev_error(dev, op.status, |
3516 | + "unmapping page at handle %d error %d", |
3517 | + (int16_t)area->phys_addr, op.status); |
3518 | + |
3519 | + return op.status; |
3520 | +} |
3521 | +EXPORT_SYMBOL_GPL(xenbus_unmap_ring_vfree); |
3522 | + |
3523 | + |
3524 | +/** |
3525 | + * xenbus_unmap_ring |
3526 | + * @dev: xenbus device |
3527 | + * @handle: grant handle |
3528 | + * @vaddr: addr to unmap |
3529 | + * |
3530 | + * Unmap a page of memory in this domain that was imported from another domain. |
3531 | + * Returns 0 on success and returns GNTST_* on error |
3532 | + * (see xen/include/interface/grant_table.h). |
3533 | + */ |
3534 | +int xenbus_unmap_ring(struct xenbus_device *dev, |
3535 | + grant_handle_t handle, void *vaddr) |
3536 | +{ |
3537 | + struct gnttab_unmap_grant_ref op = { |
3538 | + .host_addr = (unsigned long)vaddr, |
3539 | + .handle = handle, |
3540 | + }; |
3541 | + |
3542 | + if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1)) |
3543 | + BUG(); |
3544 | + |
3545 | + if (op.status != GNTST_okay) |
3546 | + xenbus_dev_error(dev, op.status, |
3547 | + "unmapping page at handle %d error %d", |
3548 | + handle, op.status); |
3549 | + |
3550 | + return op.status; |
3551 | +} |
3552 | +EXPORT_SYMBOL_GPL(xenbus_unmap_ring); |
3553 | +#endif |
3554 | + |
3555 | + |
3556 | +/** |
3557 | + * xenbus_read_driver_state |
3558 | + * @path: path for driver |
3559 | + * |
3560 | + * Return the state of the driver rooted at the given store path, or |
3561 | + * XenbusStateUnknown if no state can be read. |
3562 | + */ |
3563 | enum xenbus_state xenbus_read_driver_state(const char *path) |
3564 | { |
3565 | enum xenbus_state result; |
3566 | diff -Naur linux-2.6.25/drivers/xen/xenbus/xenbus_comms.c linux-2.6.25-xen/drivers/xen/xenbus/xenbus_comms.c |
3567 | --- linux-2.6.25/drivers/xen/xenbus/xenbus_comms.c 2008-05-23 18:33:26.000000000 +0200 |
3568 | +++ linux-2.6.25-xen/drivers/xen/xenbus/xenbus_comms.c 2008-05-23 18:27:40.000000000 +0200 |
3569 | @@ -34,12 +34,15 @@ |
3570 | #include <linux/interrupt.h> |
3571 | #include <linux/sched.h> |
3572 | #include <linux/err.h> |
3573 | -#include <linux/ptrace.h> |
3574 | -#include <linux/workqueue.h> |
3575 | -#include <xen/evtchn.h> |
3576 | #include <xen/xenbus.h> |
3577 | - |
3578 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3579 | +#include <xen/evtchn.h> |
3580 | #include <asm/hypervisor.h> |
3581 | +#else |
3582 | +#include <asm/xen/hypervisor.h> |
3583 | +#include <xen/events.h> |
3584 | +#include <xen/page.h> |
3585 | +#endif |
3586 | |
3587 | #include "xenbus_comms.h" |
3588 | |
3589 | @@ -91,6 +94,13 @@ |
3590 | return buf + MASK_XENSTORE_IDX(cons); |
3591 | } |
3592 | |
3593 | +/** |
3594 | + * xb_write - low level write |
3595 | + * @data: buffer to send |
3596 | + * @len: length of buffer |
3597 | + * |
3598 | + * Returns 0 on success, error otherwise. |
3599 | + */ |
3600 | int xb_write(const void *data, unsigned len) |
3601 | { |
3602 | struct xenstore_domain_interface *intf = xen_store_interface; |
3603 | @@ -199,7 +209,9 @@ |
3604 | return 0; |
3605 | } |
3606 | |
3607 | -/* Set up interrupt handler off store event channel. */ |
3608 | +/** |
3609 | + * xb_init_comms - Set up interrupt handler off store event channel. |
3610 | + */ |
3611 | int xb_init_comms(void) |
3612 | { |
3613 | struct xenstore_domain_interface *intf = xen_store_interface; |
3614 | @@ -219,7 +231,11 @@ |
3615 | if (xenbus_irq) |
3616 | unbind_from_irqhandler(xenbus_irq, &xb_waitq); |
3617 | |
3618 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3619 | err = bind_caller_port_to_irqhandler( |
3620 | +#else |
3621 | + err = bind_evtchn_to_irqhandler( |
3622 | +#endif |
3623 | xen_store_evtchn, wake_waiting, |
3624 | 0, "xenbus", &xb_waitq); |
3625 | if (err <= 0) { |
3626 | diff -Naur linux-2.6.25/drivers/xen/xenbus/xenbus_probe_backend.c linux-2.6.25-xen/drivers/xen/xenbus/xenbus_probe_backend.c |
3627 | --- linux-2.6.25/drivers/xen/xenbus/xenbus_probe_backend.c 2008-05-23 18:33:27.000000000 +0200 |
3628 | +++ linux-2.6.25-xen/drivers/xen/xenbus/xenbus_probe_backend.c 2008-05-23 18:27:40.000000000 +0200 |
3629 | @@ -236,7 +236,7 @@ |
3630 | { |
3631 | DPRINTK(""); |
3632 | |
3633 | - dev_changed(vec[XS_WATCH_PATH], &xenbus_backend); |
3634 | + xenbus_dev_changed(vec[XS_WATCH_PATH], &xenbus_backend); |
3635 | } |
3636 | |
3637 | static struct xenbus_watch be_watch = { |
3638 | diff -Naur linux-2.6.25/drivers/xen/xenbus/xenbus_probe.c linux-2.6.25-xen/drivers/xen/xenbus/xenbus_probe.c |
3639 | --- linux-2.6.25/drivers/xen/xenbus/xenbus_probe.c 2008-05-23 18:33:27.000000000 +0200 |
3640 | +++ linux-2.6.25-xen/drivers/xen/xenbus/xenbus_probe.c 2008-05-23 18:27:40.000000000 +0200 |
3641 | @@ -43,12 +43,11 @@ |
3642 | #include <linux/mm.h> |
3643 | #include <linux/notifier.h> |
3644 | #include <linux/mutex.h> |
3645 | -#include <linux/module.h> |
3646 | +#include <linux/io.h> |
3647 | |
3648 | -#include <asm/io.h> |
3649 | #include <asm/page.h> |
3650 | -#include <asm/maddr.h> |
3651 | #include <asm/pgtable.h> |
3652 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3653 | #include <asm/hypervisor.h> |
3654 | #include <xen/xenbus.h> |
3655 | #include <xen/xen_proc.h> |
3656 | @@ -57,6 +56,12 @@ |
3657 | #ifdef MODULE |
3658 | #include <xen/hvm.h> |
3659 | #endif |
3660 | +#else |
3661 | +#include <asm/xen/hypervisor.h> |
3662 | +#include <xen/xenbus.h> |
3663 | +#include <xen/events.h> |
3664 | +#include <xen/page.h> |
3665 | +#endif |
3666 | |
3667 | #include "xenbus_comms.h" |
3668 | #include "xenbus_probe.h" |
3669 | @@ -168,7 +173,7 @@ |
3670 | return read_otherend_details(xendev, "backend-id", "backend"); |
3671 | } |
3672 | |
3673 | -#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) |
3674 | +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) && (defined(CONFIG_XEN) || defined(MODULE)) |
3675 | static int xenbus_uevent_frontend(struct device *dev, char **envp, |
3676 | int num_envp, char *buffer, int buffer_size) |
3677 | { |
3678 | @@ -207,12 +212,16 @@ |
3679 | .probe = xenbus_dev_probe, |
3680 | .remove = xenbus_dev_remove, |
3681 | .shutdown = xenbus_dev_shutdown, |
3682 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3683 | .uevent = xenbus_uevent_frontend, |
3684 | #endif |
3685 | +#endif |
3686 | }, |
3687 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3688 | .dev = { |
3689 | .bus_id = "xen", |
3690 | }, |
3691 | +#endif |
3692 | }; |
3693 | |
3694 | static void otherend_changed(struct xenbus_watch *watch, |
3695 | @@ -228,14 +237,15 @@ |
3696 | if (!dev->otherend || |
3697 | strncmp(dev->otherend, vec[XS_WATCH_PATH], |
3698 | strlen(dev->otherend))) { |
3699 | - DPRINTK("Ignoring watch at %s", vec[XS_WATCH_PATH]); |
3700 | + dev_dbg(&dev->dev, "Ignoring watch at %s", vec[XS_WATCH_PATH]); |
3701 | return; |
3702 | } |
3703 | |
3704 | state = xenbus_read_driver_state(dev->otherend); |
3705 | |
3706 | - DPRINTK("state is %d (%s), %s, %s", state, xenbus_strstate(state), |
3707 | - dev->otherend_watch.node, vec[XS_WATCH_PATH]); |
3708 | + dev_dbg(&dev->dev, "state is %d (%s), %s, %s", |
3709 | + state, xenbus_strstate(state), dev->otherend_watch.node, |
3710 | + vec[XS_WATCH_PATH]); |
3711 | |
3712 | #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) |
3713 | /* |
3714 | @@ -271,8 +281,13 @@ |
3715 | |
3716 | static int watch_otherend(struct xenbus_device *dev) |
3717 | { |
3718 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3719 | return xenbus_watch_path2(dev, dev->otherend, "state", |
3720 | &dev->otherend_watch, otherend_changed); |
3721 | +#else |
3722 | + return xenbus_watch_pathfmt(dev, &dev->otherend_watch, otherend_changed, |
3723 | + "%s/%s", dev->otherend, "state"); |
3724 | +#endif |
3725 | } |
3726 | |
3727 | |
3728 | @@ -298,9 +313,9 @@ |
3729 | |
3730 | err = talk_to_otherend(dev); |
3731 | if (err) { |
3732 | - printk(KERN_WARNING |
3733 | - "xenbus_probe: talk_to_otherend on %s failed.\n", |
3734 | - dev->nodename); |
3735 | + dev_warn(&dev->dev, |
3736 | + "xenbus_probe: talk_to_otherend on %s failed.\n", |
3737 | + dev->nodename); |
3738 | return err; |
3739 | } |
3740 | |
3741 | @@ -310,9 +325,9 @@ |
3742 | |
3743 | err = watch_otherend(dev); |
3744 | if (err) { |
3745 | - printk(KERN_WARNING |
3746 | - "xenbus_probe: watch_otherend on %s failed.\n", |
3747 | - dev->nodename); |
3748 | + dev_warn(&dev->dev, |
3749 | + "xenbus_probe: watch_otherend on %s failed.\n", |
3750 | + dev->nodename); |
3751 | return err; |
3752 | } |
3753 | |
3754 | @@ -349,14 +364,15 @@ |
3755 | |
3756 | get_device(&dev->dev); |
3757 | if (dev->state != XenbusStateConnected) { |
3758 | - printk("%s: %s: %s != Connected, skipping\n", __FUNCTION__, |
3759 | - dev->nodename, xenbus_strstate(dev->state)); |
3760 | + dev_info(&dev->dev, "%s: %s: %s != Connected, skipping\n", __FUNCTION__, |
3761 | + dev->nodename, xenbus_strstate(dev->state)); |
3762 | goto out; |
3763 | } |
3764 | xenbus_switch_state(dev, XenbusStateClosing); |
3765 | timeout = wait_for_completion_timeout(&dev->down, timeout); |
3766 | if (!timeout) |
3767 | - printk("%s: %s timeout closing device\n", __FUNCTION__, dev->nodename); |
3768 | + dev_info(&dev->dev, "%s: %s timeout closing device\n", |
3769 | + __FUNCTION__, dev->nodename); |
3770 | out: |
3771 | put_device(&dev->dev); |
3772 | } |
3773 | @@ -544,7 +560,9 @@ |
3774 | xendev->devicetype = tmpstring; |
3775 | init_completion(&xendev->down); |
3776 | |
3777 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3778 | xendev->dev.parent = &bus->dev; |
3779 | +#endif |
3780 | xendev->dev.bus = &bus->bus; |
3781 | xendev->dev.release = xenbus_dev_release; |
3782 | |
3783 | @@ -559,15 +577,16 @@ |
3784 | |
3785 | err = device_create_file(&xendev->dev, &dev_attr_nodename); |
3786 | if (err) |
3787 | - goto unregister; |
3788 | + goto fail_unregister; |
3789 | + |
3790 | err = device_create_file(&xendev->dev, &dev_attr_devtype); |
3791 | if (err) |
3792 | - goto unregister; |
3793 | + goto fail_remove_file; |
3794 | |
3795 | return 0; |
3796 | -unregister: |
3797 | +fail_remove_file: |
3798 | device_remove_file(&xendev->dev, &dev_attr_nodename); |
3799 | - device_remove_file(&xendev->dev, &dev_attr_devtype); |
3800 | +fail_unregister: |
3801 | device_unregister(&xendev->dev); |
3802 | fail: |
3803 | kfree(xendev); |
3804 | @@ -580,7 +599,8 @@ |
3805 | char *nodename; |
3806 | int err; |
3807 | |
3808 | - nodename = kasprintf(GFP_KERNEL, "%s/%s/%s", xenbus_frontend.root, type, name); |
3809 | + nodename = kasprintf(GFP_KERNEL, "%s/%s/%s", |
3810 | + xenbus_frontend.root, type, name); |
3811 | if (!nodename) |
3812 | return -ENOMEM; |
3813 | |
3814 | @@ -656,7 +676,7 @@ |
3815 | return (len == 0) ? i : -ERANGE; |
3816 | } |
3817 | |
3818 | -void dev_changed(const char *node, struct xen_bus_type *bus) |
3819 | +void xenbus_dev_changed(const char *node, struct xen_bus_type *bus) |
3820 | { |
3821 | int exists, rootlen; |
3822 | struct xenbus_device *dev; |
3823 | @@ -664,7 +684,7 @@ |
3824 | const char *p, *root; |
3825 | |
3826 | if (bus->error || char_count(node, '/') < 2) |
3827 | - return; |
3828 | + return; |
3829 | |
3830 | exists = xenbus_exists(XBT_NIL, node, ""); |
3831 | if (!exists) { |
3832 | @@ -698,7 +718,7 @@ |
3833 | { |
3834 | DPRINTK(""); |
3835 | |
3836 | - dev_changed(vec[XS_WATCH_PATH], &xenbus_frontend); |
3837 | + xenbus_dev_changed(vec[XS_WATCH_PATH], &xenbus_frontend); |
3838 | } |
3839 | |
3840 | /* We watch for devices appearing and vanishing. */ |
3841 | @@ -774,9 +794,9 @@ |
3842 | |
3843 | if (drv->resume) { |
3844 | err = drv->resume(xdev); |
3845 | - if (err) { |
3846 | + if (err) { |
3847 | printk(KERN_WARNING |
3848 | - "xenbus: resume %s failed: %i\n", |
3849 | + "xenbus: resume %s failed: %i\n", |
3850 | dev->bus_id, err); |
3851 | return err; |
3852 | } |
3853 | @@ -901,10 +921,16 @@ |
3854 | } |
3855 | #endif |
3856 | |
3857 | -static int xenbus_probe_init(void) |
3858 | +#ifndef MODULE |
3859 | +static int __init xenbus_probe_init(void) |
3860 | +#else |
3861 | +static int __devinit xenbus_probe_init(void) |
3862 | +#endif |
3863 | { |
3864 | int err = 0; |
3865 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3866 | unsigned long page = 0; |
3867 | +#endif |
3868 | |
3869 | DPRINTK(""); |
3870 | |
3871 | @@ -923,6 +949,7 @@ |
3872 | * Domain0 doesn't have a store_evtchn or store_mfn yet. |
3873 | */ |
3874 | if (is_initial_xendomain()) { |
3875 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3876 | struct evtchn_alloc_unbound alloc_unbound; |
3877 | |
3878 | /* Allocate page. */ |
3879 | @@ -960,10 +987,13 @@ |
3880 | if (xsd_port_intf) |
3881 | xsd_port_intf->read_proc = xsd_port_read; |
3882 | #endif |
3883 | +#else |
3884 | + /* dom0 not yet supported */ |
3885 | +#endif |
3886 | xen_store_interface = mfn_to_virt(xen_store_mfn); |
3887 | } else { |
3888 | xenstored_ready = 1; |
3889 | -#ifdef CONFIG_XEN |
3890 | +#ifndef MODULE |
3891 | xen_store_evtchn = xen_start_info->store_evtchn; |
3892 | xen_store_mfn = xen_start_info->store_mfn; |
3893 | xen_store_interface = mfn_to_virt(xen_store_mfn); |
3894 | @@ -976,7 +1006,9 @@ |
3895 | } |
3896 | |
3897 | |
3898 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3899 | xenbus_dev_init(); |
3900 | +#endif |
3901 | |
3902 | /* Initialize the interface to xenstore. */ |
3903 | err = xs_init(); |
3904 | @@ -986,6 +1018,7 @@ |
3905 | goto err; |
3906 | } |
3907 | |
3908 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3909 | /* Register ourselves with the kernel device subsystem */ |
3910 | if (!xenbus_frontend.error) { |
3911 | xenbus_frontend.error = device_register(&xenbus_frontend.dev); |
3912 | @@ -996,6 +1029,7 @@ |
3913 | xenbus_frontend.error); |
3914 | } |
3915 | } |
3916 | +#endif |
3917 | xenbus_backend_device_register(); |
3918 | |
3919 | if (!is_initial_xendomain()) |
3920 | @@ -1004,8 +1038,10 @@ |
3921 | return 0; |
3922 | |
3923 | err: |
3924 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3925 | if (page) |
3926 | free_page(page); |
3927 | +#endif |
3928 | |
3929 | /* |
3930 | * Do not unregister the xenbus front/backend buses here. The buses |
3931 | @@ -1016,11 +1052,15 @@ |
3932 | return err; |
3933 | } |
3934 | |
3935 | -#ifdef CONFIG_XEN |
3936 | +#ifndef MODULE |
3937 | postcore_initcall(xenbus_probe_init); |
3938 | +#ifdef CONFIG_XEN |
3939 | MODULE_LICENSE("Dual BSD/GPL"); |
3940 | #else |
3941 | -int xenbus_init(void) |
3942 | +MODULE_LICENSE("GPL"); |
3943 | +#endif |
3944 | +#else |
3945 | +int __devinit xenbus_init(void) |
3946 | { |
3947 | return xenbus_probe_init(); |
3948 | } |
3949 | diff -Naur linux-2.6.25/drivers/xen/xenbus/xenbus_probe.h linux-2.6.25-xen/drivers/xen/xenbus/xenbus_probe.h |
3950 | --- linux-2.6.25/drivers/xen/xenbus/xenbus_probe.h 2008-05-23 18:33:27.000000000 +0200 |
3951 | +++ linux-2.6.25-xen/drivers/xen/xenbus/xenbus_probe.h 2008-05-23 18:27:40.000000000 +0200 |
3952 | @@ -56,7 +56,9 @@ |
3953 | int (*get_bus_id)(char bus_id[BUS_ID_SIZE], const char *nodename); |
3954 | int (*probe)(const char *type, const char *dir); |
3955 | struct bus_type bus; |
3956 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3957 | struct device dev; |
3958 | +#endif |
3959 | }; |
3960 | |
3961 | extern int xenbus_match(struct device *_dev, struct device_driver *_drv); |
3962 | @@ -71,7 +73,7 @@ |
3963 | const char *nodename); |
3964 | extern int xenbus_probe_devices(struct xen_bus_type *bus); |
3965 | |
3966 | -extern void dev_changed(const char *node, struct xen_bus_type *bus); |
3967 | +extern void xenbus_dev_changed(const char *node, struct xen_bus_type *bus); |
3968 | |
3969 | #endif |
3970 | |
3971 | diff -Naur linux-2.6.25/drivers/xen/xenbus/xenbus_xs.c linux-2.6.25-xen/drivers/xen/xenbus/xenbus_xs.c |
3972 | --- linux-2.6.25/drivers/xen/xenbus/xenbus_xs.c 2008-05-23 18:33:03.000000000 +0200 |
3973 | +++ linux-2.6.25-xen/drivers/xen/xenbus/xenbus_xs.c 2008-05-23 18:27:40.000000000 +0200 |
3974 | @@ -221,7 +221,7 @@ |
3975 | } |
3976 | |
3977 | for (i = 0; i < num_vecs; i++) { |
3978 | - err = xb_write(iovec[i].iov_base, iovec[i].iov_len);; |
3979 | + err = xb_write(iovec[i].iov_base, iovec[i].iov_len); |
3980 | if (err) { |
3981 | mutex_unlock(&xs_state.request_mutex); |
3982 | return ERR_PTR(err); |
3983 | @@ -629,7 +629,9 @@ |
3984 | char token[sizeof(watch) * 2 + 1]; |
3985 | int err; |
3986 | |
3987 | +#if defined(CONFIG_XEN) || defined(MODULE) |
3988 | BUG_ON(watch->flags & XBWF_new_thread); |
3989 | +#endif |
3990 | |
3991 | sprintf(token, "%lX", (long)watch); |
3992 | |
3993 | @@ -648,6 +650,11 @@ |
3994 | |
3995 | up_read(&xs_state.watch_mutex); |
3996 | |
3997 | + /* Make sure there are no callbacks running currently (unless |
3998 | + its us) */ |
3999 | + if (current->pid != xenwatch_pid) |
4000 | + mutex_lock(&xenwatch_mutex); |
4001 | + |
4002 | /* Cancel pending watch events. */ |
4003 | spin_lock(&watch_events_lock); |
4004 | list_for_each_entry_safe(msg, tmp, &watch_events, list) { |
4005 | @@ -659,11 +666,8 @@ |
4006 | } |
4007 | spin_unlock(&watch_events_lock); |
4008 | |
4009 | - /* Flush any currently-executing callback, unless we are it. :-) */ |
4010 | - if (current->pid != xenwatch_pid) { |
4011 | - mutex_lock(&xenwatch_mutex); |
4012 | + if (current->pid != xenwatch_pid) |
4013 | mutex_unlock(&xenwatch_mutex); |
4014 | - } |
4015 | } |
4016 | EXPORT_SYMBOL_GPL(unregister_xenbus_watch); |
4017 | |
4018 | @@ -701,6 +705,7 @@ |
4019 | up_write(&xs_state.transaction_mutex); |
4020 | } |
4021 | |
4022 | +#if defined(CONFIG_XEN) || defined(MODULE) |
4023 | static int xenwatch_handle_callback(void *data) |
4024 | { |
4025 | struct xs_stored_msg *msg = data; |
4026 | @@ -718,6 +723,7 @@ |
4027 | |
4028 | return 0; |
4029 | } |
4030 | +#endif |
4031 | |
4032 | static int xenwatch_thread(void *unused) |
4033 | { |
4034 | @@ -747,6 +753,7 @@ |
4035 | |
4036 | msg = list_entry(ent, struct xs_stored_msg, list); |
4037 | |
4038 | +#if defined(CONFIG_XEN) || defined(MODULE) |
4039 | /* |
4040 | * Unlock the mutex before running an XBWF_new_thread |
4041 | * handler. kthread_run can block which can deadlock |
4042 | @@ -763,6 +770,15 @@ |
4043 | xenwatch_handle_callback(msg); |
4044 | mutex_unlock(&xenwatch_mutex); |
4045 | } |
4046 | +#else |
4047 | + msg->u.watch.handle->callback( |
4048 | + msg->u.watch.handle, |
4049 | + (const char **)msg->u.watch.vec, |
4050 | + msg->u.watch.vec_size); |
4051 | + mutex_unlock(&xenwatch_mutex); |
4052 | + kfree(msg->u.watch.vec); |
4053 | + kfree(msg); |
4054 | +#endif |
4055 | } |
4056 | |
4057 | return 0; |
4058 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/e820_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/e820_64.h |
4059 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/e820_64.h 2008-05-23 18:33:27.000000000 +0200 |
4060 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/e820_64.h 2008-05-23 18:27:40.000000000 +0200 |
4061 | @@ -11,8 +11,6 @@ |
4062 | #ifndef __E820_HEADER |
4063 | #define __E820_HEADER |
4064 | |
4065 | -#include <linux/mmzone.h> |
4066 | - |
4067 | #define E820MAP 0x2d0 /* our map */ |
4068 | #define E820MAX 128 /* number of entries in E820MAP */ |
4069 | #define E820NR 0x1e8 /* # entries in E820MAP */ |
4070 | @@ -30,7 +28,7 @@ |
4071 | } __attribute__((packed)); |
4072 | |
4073 | struct e820map { |
4074 | - int nr_map; |
4075 | + u32 nr_map; |
4076 | struct e820entry map[E820MAX]; |
4077 | }; |
4078 | |
4079 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/fixmap_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/fixmap_32.h |
4080 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/fixmap_32.h 2008-05-23 18:33:27.000000000 +0200 |
4081 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/fixmap_32.h 2008-05-23 18:27:40.000000000 +0200 |
4082 | @@ -53,6 +53,8 @@ |
4083 | enum fixed_addresses { |
4084 | FIX_HOLE, |
4085 | FIX_VDSO, |
4086 | + FIX_DBGP_BASE, |
4087 | + FIX_EARLYCON_MEM_BASE, |
4088 | #ifdef CONFIG_X86_LOCAL_APIC |
4089 | FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */ |
4090 | #endif |
4091 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/fixmap_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/fixmap_64.h |
4092 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/fixmap_64.h 2008-05-23 18:33:27.000000000 +0200 |
4093 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/fixmap_64.h 2008-05-23 18:27:40.000000000 +0200 |
4094 | @@ -23,9 +23,9 @@ |
4095 | * compile time, but to set the physical address only |
4096 | * in the boot process. |
4097 | * |
4098 | - * these 'compile-time allocated' memory buffers are |
4099 | - * fixed-size 4k pages. (or larger if used with an increment |
4100 | - * highger than 1) use fixmap_set(idx,phys) to associate |
4101 | + * These 'compile-time allocated' memory buffers are |
4102 | + * fixed-size 4k pages (or larger if used with an increment |
4103 | + * higher than 1). Use set_fixmap(idx,phys) to associate |
4104 | * physical memory with fixmap indices. |
4105 | * |
4106 | * TLB entries of such buffers will not be flushed across |
4107 | @@ -36,6 +36,8 @@ |
4108 | VSYSCALL_LAST_PAGE, |
4109 | VSYSCALL_FIRST_PAGE = VSYSCALL_LAST_PAGE + ((VSYSCALL_END-VSYSCALL_START) >> PAGE_SHIFT) - 1, |
4110 | VSYSCALL_HPET, |
4111 | + FIX_DBGP_BASE, |
4112 | + FIX_EARLYCON_MEM_BASE, |
4113 | FIX_HPET_BASE, |
4114 | #ifdef CONFIG_X86_LOCAL_APIC |
4115 | FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */ |
4116 | @@ -105,7 +107,7 @@ |
4117 | if (idx >= __end_of_fixed_addresses) |
4118 | __this_fixmap_does_not_exist(); |
4119 | |
4120 | - return __fix_to_virt(idx); |
4121 | + return __fix_to_virt(idx); |
4122 | } |
4123 | |
4124 | #endif |
4125 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/hw_irq_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/hw_irq_64.h |
4126 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/hw_irq_64.h 2008-05-23 18:33:27.000000000 +0200 |
4127 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/hw_irq_64.h 2008-05-23 18:27:40.000000000 +0200 |
4128 | @@ -97,6 +97,26 @@ |
4129 | #endif |
4130 | |
4131 | #ifndef __ASSEMBLY__ |
4132 | + |
4133 | +/* Interrupt handlers registered during init_IRQ */ |
4134 | +void apic_timer_interrupt(void); |
4135 | +void spurious_interrupt(void); |
4136 | +void error_interrupt(void); |
4137 | +void reschedule_interrupt(void); |
4138 | +void call_function_interrupt(void); |
4139 | +void irq_move_cleanup_interrupt(void); |
4140 | +void invalidate_interrupt0(void); |
4141 | +void invalidate_interrupt1(void); |
4142 | +void invalidate_interrupt2(void); |
4143 | +void invalidate_interrupt3(void); |
4144 | +void invalidate_interrupt4(void); |
4145 | +void invalidate_interrupt5(void); |
4146 | +void invalidate_interrupt6(void); |
4147 | +void invalidate_interrupt7(void); |
4148 | +void thermal_interrupt(void); |
4149 | +void threshold_interrupt(void); |
4150 | +void i8254_timer_resume(void); |
4151 | + |
4152 | typedef int vector_irq_t[NR_VECTORS]; |
4153 | DECLARE_PER_CPU(vector_irq_t, vector_irq); |
4154 | extern void __setup_vector_irq(int cpu); |
4155 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/io_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/io_32.h |
4156 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/io_32.h 2008-05-23 18:33:27.000000000 +0200 |
4157 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/io_32.h 2008-05-23 18:27:40.000000000 +0200 |
4158 | @@ -125,6 +125,9 @@ |
4159 | * writew/writel functions and the other mmio helpers. The returned |
4160 | * address is not guaranteed to be usable directly as a virtual |
4161 | * address. |
4162 | + * |
4163 | + * If the area you are trying to map is a PCI BAR you should have a |
4164 | + * look at pci_iomap(). |
4165 | */ |
4166 | |
4167 | static inline void __iomem * ioremap(unsigned long offset, unsigned long size) |
4168 | @@ -142,6 +145,7 @@ |
4169 | */ |
4170 | extern void *bt_ioremap(unsigned long offset, unsigned long size); |
4171 | extern void bt_iounmap(void *addr, unsigned long size); |
4172 | +extern void __iomem *fix_ioremap(unsigned idx, unsigned long phys); |
4173 | |
4174 | /* Use early IO mappings for DMI because it's initialized early */ |
4175 | #define dmi_ioremap bt_ioremap |
4176 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/io_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/io_64.h |
4177 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/io_64.h 2008-05-23 18:33:27.000000000 +0200 |
4178 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/io_64.h 2008-05-23 18:27:40.000000000 +0200 |
4179 | @@ -162,6 +162,7 @@ |
4180 | */ |
4181 | extern void __iomem * ioremap_nocache (unsigned long offset, unsigned long size); |
4182 | extern void iounmap(volatile void __iomem *addr); |
4183 | +extern void __iomem *fix_ioremap(unsigned idx, unsigned long phys); |
4184 | |
4185 | /* |
4186 | * ISA I/O bus memory addresses are 1:1 with the physical address. |
4187 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/maddr_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/maddr_32.h |
4188 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/maddr_32.h 2008-05-23 18:32:29.000000000 +0200 |
4189 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/maddr_32.h 2008-05-23 18:27:40.000000000 +0200 |
4190 | @@ -155,6 +155,7 @@ |
4191 | |
4192 | #ifdef CONFIG_X86_PAE |
4193 | #define __pte_ma(x) ((pte_t) { (x), (maddr_t)(x) >> 32 } ) |
4194 | +extern unsigned long long __supported_pte_mask; |
4195 | static inline pte_t pfn_pte_ma(unsigned long page_nr, pgprot_t pgprot) |
4196 | { |
4197 | pte_t pte; |
4198 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/mmu_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/mmu_64.h |
4199 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/mmu_64.h 2008-05-23 18:33:27.000000000 +0200 |
4200 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/mmu_64.h 2008-05-23 18:27:40.000000000 +0200 |
4201 | @@ -15,6 +15,7 @@ |
4202 | rwlock_t ldtlock; |
4203 | int size; |
4204 | struct semaphore sem; |
4205 | + void *vdso; |
4206 | #ifdef CONFIG_XEN |
4207 | unsigned pinned:1; |
4208 | unsigned has_foreign_mappings:1; |
4209 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/mmu_context_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/mmu_context_32.h |
4210 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/mmu_context_32.h 2008-05-23 18:33:27.000000000 +0200 |
4211 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/mmu_context_32.h 2008-05-23 18:27:40.000000000 +0200 |
4212 | @@ -51,6 +51,8 @@ |
4213 | : : "r" (0) ); |
4214 | } |
4215 | |
4216 | +void leave_mm(unsigned long cpu); |
4217 | + |
4218 | static inline void switch_mm(struct mm_struct *prev, |
4219 | struct mm_struct *next, |
4220 | struct task_struct *tsk) |
4221 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/page_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/page_64.h |
4222 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/page_64.h 2008-05-23 18:33:27.000000000 +0200 |
4223 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/page_64.h 2008-05-23 18:27:40.000000000 +0200 |
4224 | @@ -72,7 +72,8 @@ |
4225 | #define clear_user_page(page, vaddr, pg) clear_page(page) |
4226 | #define copy_user_page(to, from, vaddr, pg) copy_page(to, from) |
4227 | |
4228 | -#define alloc_zeroed_user_highpage(vma, vaddr) alloc_page_vma(GFP_HIGHUSER | __GFP_ZERO, vma, vaddr) |
4229 | +#define __alloc_zeroed_user_highpage(movableflags, vma, vaddr) \ |
4230 | + alloc_page_vma(GFP_HIGHUSER | __GFP_ZERO | movableflags, vma, vaddr) |
4231 | #define __HAVE_ARCH_ALLOC_ZEROED_USER_HIGHPAGE |
4232 | |
4233 | /* |
4234 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/pci_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pci_32.h |
4235 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/pci_32.h 2008-05-23 18:32:29.000000000 +0200 |
4236 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pci_32.h 2008-05-23 18:27:40.000000000 +0200 |
4237 | @@ -3,6 +3,14 @@ |
4238 | |
4239 | |
4240 | #ifdef __KERNEL__ |
4241 | + |
4242 | +struct pci_sysdata { |
4243 | + int node; /* NUMA node */ |
4244 | +}; |
4245 | + |
4246 | +/* scan a bus after allocating a pci_sysdata for it */ |
4247 | +extern struct pci_bus *pci_scan_bus_with_sysdata(int busno); |
4248 | + |
4249 | #include <linux/mm.h> /* for struct page */ |
4250 | |
4251 | /* Can be used to override the logic in pci_scan_bus for skipping |
4252 | @@ -81,48 +89,11 @@ |
4253 | |
4254 | #endif |
4255 | |
4256 | -/* This is always fine. */ |
4257 | -#define pci_dac_dma_supported(pci_dev, mask) (1) |
4258 | - |
4259 | -static inline dma64_addr_t |
4260 | -pci_dac_page_to_dma(struct pci_dev *pdev, struct page *page, unsigned long offset, int direction) |
4261 | -{ |
4262 | - return ((dma64_addr_t) page_to_phys(page) + |
4263 | - (dma64_addr_t) offset); |
4264 | -} |
4265 | - |
4266 | -static inline struct page * |
4267 | -pci_dac_dma_to_page(struct pci_dev *pdev, dma64_addr_t dma_addr) |
4268 | -{ |
4269 | - return pfn_to_page(dma_addr >> PAGE_SHIFT); |
4270 | -} |
4271 | - |
4272 | -static inline unsigned long |
4273 | -pci_dac_dma_to_offset(struct pci_dev *pdev, dma64_addr_t dma_addr) |
4274 | -{ |
4275 | - return (dma_addr & ~PAGE_MASK); |
4276 | -} |
4277 | - |
4278 | -static inline void |
4279 | -pci_dac_dma_sync_single_for_cpu(struct pci_dev *pdev, dma64_addr_t dma_addr, size_t len, int direction) |
4280 | -{ |
4281 | -} |
4282 | - |
4283 | -static inline void |
4284 | -pci_dac_dma_sync_single_for_device(struct pci_dev *pdev, dma64_addr_t dma_addr, size_t len, int direction) |
4285 | -{ |
4286 | - flush_write_buffers(); |
4287 | -} |
4288 | - |
4289 | #define HAVE_PCI_MMAP |
4290 | extern int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma, |
4291 | enum pci_mmap_state mmap_state, int write_combine); |
4292 | |
4293 | |
4294 | -static inline void pcibios_add_platform_entries(struct pci_dev *dev) |
4295 | -{ |
4296 | -} |
4297 | - |
4298 | #ifdef CONFIG_PCI |
4299 | static inline void pci_dma_burst_advice(struct pci_dev *pdev, |
4300 | enum pci_dma_burst_strategy *strat, |
4301 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/pci_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pci_64.h |
4302 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/pci_64.h 2008-05-23 18:32:29.000000000 +0200 |
4303 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pci_64.h 2008-05-23 18:27:40.000000000 +0200 |
4304 | @@ -5,6 +5,27 @@ |
4305 | |
4306 | #ifdef __KERNEL__ |
4307 | |
4308 | +struct pci_sysdata { |
4309 | + int node; /* NUMA node */ |
4310 | + void* iommu; /* IOMMU private data */ |
4311 | +}; |
4312 | + |
4313 | +extern struct pci_bus *pci_scan_bus_with_sysdata(int busno); |
4314 | + |
4315 | +#ifdef CONFIG_CALGARY_IOMMU |
4316 | +static inline void* pci_iommu(struct pci_bus *bus) |
4317 | +{ |
4318 | + struct pci_sysdata *sd = bus->sysdata; |
4319 | + return sd->iommu; |
4320 | +} |
4321 | + |
4322 | +static inline void set_pci_iommu(struct pci_bus *bus, void *val) |
4323 | +{ |
4324 | + struct pci_sysdata *sd = bus->sysdata; |
4325 | + sd->iommu = val; |
4326 | +} |
4327 | +#endif /* CONFIG_CALGARY_IOMMU */ |
4328 | + |
4329 | #include <linux/mm.h> /* for struct page */ |
4330 | |
4331 | /* Can be used to override the logic in pci_scan_bus for skipping |
4332 | @@ -56,14 +77,6 @@ |
4333 | |
4334 | #if defined(CONFIG_IOMMU) || defined(CONFIG_CALGARY_IOMMU) |
4335 | |
4336 | -/* |
4337 | - * x86-64 always supports DAC, but sometimes it is useful to force |
4338 | - * devices through the IOMMU to get automatic sg list merging. |
4339 | - * Optional right now. |
4340 | - */ |
4341 | -extern int iommu_sac_force; |
4342 | -#define pci_dac_dma_supported(pci_dev, mask) (!iommu_sac_force) |
4343 | - |
4344 | #define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME) \ |
4345 | dma_addr_t ADDR_NAME; |
4346 | #define DECLARE_PCI_UNMAP_LEN(LEN_NAME) \ |
4347 | @@ -97,8 +110,6 @@ |
4348 | #else |
4349 | /* No IOMMU */ |
4350 | |
4351 | -#define pci_dac_dma_supported(pci_dev, mask) 1 |
4352 | - |
4353 | #define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME) |
4354 | #define DECLARE_PCI_UNMAP_LEN(LEN_NAME) |
4355 | #define pci_unmap_addr(PTR, ADDR_NAME) (0) |
4356 | @@ -110,36 +121,6 @@ |
4357 | |
4358 | #include <asm-generic/pci-dma-compat.h> |
4359 | |
4360 | -static inline dma64_addr_t |
4361 | -pci_dac_page_to_dma(struct pci_dev *pdev, struct page *page, unsigned long offset, int direction) |
4362 | -{ |
4363 | - return ((dma64_addr_t) page_to_phys(page) + |
4364 | - (dma64_addr_t) offset); |
4365 | -} |
4366 | - |
4367 | -static inline struct page * |
4368 | -pci_dac_dma_to_page(struct pci_dev *pdev, dma64_addr_t dma_addr) |
4369 | -{ |
4370 | - return virt_to_page(__va(dma_addr)); |
4371 | -} |
4372 | - |
4373 | -static inline unsigned long |
4374 | -pci_dac_dma_to_offset(struct pci_dev *pdev, dma64_addr_t dma_addr) |
4375 | -{ |
4376 | - return (dma_addr & ~PAGE_MASK); |
4377 | -} |
4378 | - |
4379 | -static inline void |
4380 | -pci_dac_dma_sync_single_for_cpu(struct pci_dev *pdev, dma64_addr_t dma_addr, size_t len, int direction) |
4381 | -{ |
4382 | -} |
4383 | - |
4384 | -static inline void |
4385 | -pci_dac_dma_sync_single_for_device(struct pci_dev *pdev, dma64_addr_t dma_addr, size_t len, int direction) |
4386 | -{ |
4387 | - flush_write_buffers(); |
4388 | -} |
4389 | - |
4390 | #ifdef CONFIG_PCI |
4391 | static inline void pci_dma_burst_advice(struct pci_dev *pdev, |
4392 | enum pci_dma_burst_strategy *strat, |
4393 | @@ -154,10 +135,6 @@ |
4394 | extern int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma, |
4395 | enum pci_mmap_state mmap_state, int write_combine); |
4396 | |
4397 | -static inline void pcibios_add_platform_entries(struct pci_dev *dev) |
4398 | -{ |
4399 | -} |
4400 | - |
4401 | #endif /* __KERNEL__ */ |
4402 | |
4403 | /* generic pci stuff */ |
4404 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/pgalloc_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgalloc_32.h |
4405 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/pgalloc_32.h 2008-05-23 18:33:27.000000000 +0200 |
4406 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgalloc_32.h 2008-05-23 18:27:40.000000000 +0200 |
4407 | @@ -5,7 +5,7 @@ |
4408 | #include <linux/mm.h> /* for struct page */ |
4409 | #include <asm/io.h> /* for phys_to_virt and page_to_pseudophys */ |
4410 | |
4411 | -#define paravirt_alloc_pt(pfn) do { } while (0) |
4412 | +#define paravirt_alloc_pt(mm, pfn) do { } while (0) |
4413 | #define paravirt_alloc_pd(pfn) do { } while (0) |
4414 | #define paravirt_alloc_pd(pfn) do { } while (0) |
4415 | #define paravirt_alloc_pd_clone(pfn, clonepfn, start, count) do { } while (0) |
4416 | @@ -14,14 +14,14 @@ |
4417 | |
4418 | #define pmd_populate_kernel(mm, pmd, pte) \ |
4419 | do { \ |
4420 | - paravirt_alloc_pt(__pa(pte) >> PAGE_SHIFT); \ |
4421 | + paravirt_alloc_pt(mm, __pa(pte) >> PAGE_SHIFT); \ |
4422 | set_pmd(pmd, __pmd(_PAGE_TABLE + __pa(pte))); \ |
4423 | } while (0) |
4424 | |
4425 | #define pmd_populate(mm, pmd, pte) \ |
4426 | do { \ |
4427 | unsigned long pfn = page_to_pfn(pte); \ |
4428 | - paravirt_alloc_pt(pfn); \ |
4429 | + paravirt_alloc_pt(mm, pfn); \ |
4430 | if (test_bit(PG_pinned, &virt_to_page((mm)->pgd)->flags)) { \ |
4431 | if (!PageHighMem(pte)) \ |
4432 | BUG_ON(HYPERVISOR_update_va_mapping( \ |
4433 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/pgtable-2level.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgtable-2level.h |
4434 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/pgtable-2level.h 2008-05-23 18:33:27.000000000 +0200 |
4435 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgtable-2level.h 2008-05-23 18:27:40.000000000 +0200 |
4436 | @@ -82,14 +82,6 @@ |
4437 | #define pfn_pmd(pfn, prot) __pmd(((pfn) << PAGE_SHIFT) | pgprot_val(prot)) |
4438 | |
4439 | /* |
4440 | - * All present user pages are user-executable: |
4441 | - */ |
4442 | -static inline int pte_exec(pte_t pte) |
4443 | -{ |
4444 | - return pte_user(pte); |
4445 | -} |
4446 | - |
4447 | -/* |
4448 | * All present pages are kernel-executable: |
4449 | */ |
4450 | static inline int pte_exec_kernel(pte_t pte) |
4451 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/pgtable_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgtable_32.h |
4452 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/pgtable_32.h 2008-05-23 18:33:27.000000000 +0200 |
4453 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgtable_32.h 2008-05-23 18:27:40.000000000 +0200 |
4454 | @@ -26,9 +26,6 @@ |
4455 | #include <linux/spinlock.h> |
4456 | #include <linux/sched.h> |
4457 | |
4458 | -/* Is this pagetable pinned? */ |
4459 | -#define PG_pinned PG_arch_1 |
4460 | - |
4461 | struct vm_area_struct; |
4462 | |
4463 | /* |
4464 | @@ -82,7 +79,7 @@ |
4465 | * area for the same reason. ;) |
4466 | */ |
4467 | #define VMALLOC_OFFSET (8*1024*1024) |
4468 | -#define VMALLOC_START (((unsigned long) high_memory + vmalloc_earlyreserve + \ |
4469 | +#define VMALLOC_START (((unsigned long) high_memory + \ |
4470 | 2*VMALLOC_OFFSET-1) & ~(VMALLOC_OFFSET-1)) |
4471 | #ifdef CONFIG_HIGHMEM |
4472 | # define VMALLOC_END (PKMAP_BASE-2*PAGE_SIZE) |
4473 | @@ -231,8 +228,6 @@ |
4474 | * The following only work if pte_present() is true. |
4475 | * Undefined behaviour if not.. |
4476 | */ |
4477 | -static inline int pte_user(pte_t pte) { return (pte).pte_low & _PAGE_USER; } |
4478 | -static inline int pte_read(pte_t pte) { return (pte).pte_low & _PAGE_USER; } |
4479 | static inline int pte_dirty(pte_t pte) { return (pte).pte_low & _PAGE_DIRTY; } |
4480 | static inline int pte_young(pte_t pte) { return (pte).pte_low & _PAGE_ACCESSED; } |
4481 | static inline int pte_write(pte_t pte) { return (pte).pte_low & _PAGE_RW; } |
4482 | @@ -243,13 +238,9 @@ |
4483 | */ |
4484 | static inline int pte_file(pte_t pte) { return (pte).pte_low & _PAGE_FILE; } |
4485 | |
4486 | -static inline pte_t pte_rdprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_USER; return pte; } |
4487 | -static inline pte_t pte_exprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_USER; return pte; } |
4488 | static inline pte_t pte_mkclean(pte_t pte) { (pte).pte_low &= ~_PAGE_DIRTY; return pte; } |
4489 | static inline pte_t pte_mkold(pte_t pte) { (pte).pte_low &= ~_PAGE_ACCESSED; return pte; } |
4490 | static inline pte_t pte_wrprotect(pte_t pte) { (pte).pte_low &= ~_PAGE_RW; return pte; } |
4491 | -static inline pte_t pte_mkread(pte_t pte) { (pte).pte_low |= _PAGE_USER; return pte; } |
4492 | -static inline pte_t pte_mkexec(pte_t pte) { (pte).pte_low |= _PAGE_USER; return pte; } |
4493 | static inline pte_t pte_mkdirty(pte_t pte) { (pte).pte_low |= _PAGE_DIRTY; return pte; } |
4494 | static inline pte_t pte_mkyoung(pte_t pte) { (pte).pte_low |= _PAGE_ACCESSED; return pte; } |
4495 | static inline pte_t pte_mkwrite(pte_t pte) { (pte).pte_low |= _PAGE_RW; return pte; } |
4496 | @@ -295,22 +286,20 @@ |
4497 | #define ptep_set_access_flags(vma, address, ptep, entry, dirty) \ |
4498 | ({ \ |
4499 | int __changed = !pte_same(*(ptep), entry); \ |
4500 | - if (__changed && (dirty)) \ |
4501 | - ptep_establish(vma, address, ptep, entry); \ |
4502 | + if (__changed && (dirty)) { \ |
4503 | + if ( likely((vma)->vm_mm == current->mm) ) { \ |
4504 | + BUG_ON(HYPERVISOR_update_va_mapping(address, \ |
4505 | + entry, \ |
4506 | + (unsigned long)(vma)->vm_mm->cpu_vm_mask.bits| \ |
4507 | + UVMF_INVLPG|UVMF_MULTI)); \ |
4508 | + } else { \ |
4509 | + xen_l1_entry_update(ptep, entry); \ |
4510 | + flush_tlb_page(vma, address); \ |
4511 | + } \ |
4512 | + } \ |
4513 | __changed; \ |
4514 | }) |
4515 | |
4516 | -#define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_DIRTY |
4517 | -#define ptep_test_and_clear_dirty(vma, addr, ptep) ({ \ |
4518 | - int __ret = 0; \ |
4519 | - if (pte_dirty(*(ptep))) \ |
4520 | - __ret = test_and_clear_bit(_PAGE_BIT_DIRTY, \ |
4521 | - &(ptep)->pte_low); \ |
4522 | - if (__ret) \ |
4523 | - pte_update((vma)->vm_mm, addr, ptep); \ |
4524 | - __ret; \ |
4525 | -}) |
4526 | - |
4527 | #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG |
4528 | #define ptep_test_and_clear_young(vma, addr, ptep) ({ \ |
4529 | int __ret = 0; \ |
4530 | @@ -322,37 +311,6 @@ |
4531 | __ret; \ |
4532 | }) |
4533 | |
4534 | -/* |
4535 | - * Rules for using ptep_establish: the pte MUST be a user pte, and |
4536 | - * must be a present->present transition. |
4537 | - */ |
4538 | -#define __HAVE_ARCH_PTEP_ESTABLISH |
4539 | -#define ptep_establish(vma, address, ptep, pteval) \ |
4540 | -do { \ |
4541 | - if ( likely((vma)->vm_mm == current->mm) ) { \ |
4542 | - BUG_ON(HYPERVISOR_update_va_mapping(address, \ |
4543 | - pteval, \ |
4544 | - (unsigned long)(vma)->vm_mm->cpu_vm_mask.bits| \ |
4545 | - UVMF_INVLPG|UVMF_MULTI)); \ |
4546 | - } else { \ |
4547 | - xen_l1_entry_update(ptep, pteval); \ |
4548 | - flush_tlb_page(vma, address); \ |
4549 | - } \ |
4550 | -} while (0) |
4551 | - |
4552 | -#define __HAVE_ARCH_PTEP_CLEAR_DIRTY_FLUSH |
4553 | -#define ptep_clear_flush_dirty(vma, address, ptep) \ |
4554 | -({ \ |
4555 | - pte_t __pte = *(ptep); \ |
4556 | - int __dirty = pte_dirty(__pte); \ |
4557 | - __pte = pte_mkclean(__pte); \ |
4558 | - if (test_bit(PG_pinned, &virt_to_page((vma)->vm_mm->pgd)->flags)) \ |
4559 | - (void)ptep_set_access_flags(vma, address, ptep, __pte, __dirty); \ |
4560 | - else if (__dirty) \ |
4561 | - (ptep)->pte_low = __pte.pte_low; \ |
4562 | - __dirty; \ |
4563 | -}) |
4564 | - |
4565 | #define __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH |
4566 | #define ptep_clear_flush_young(vma, address, ptep) \ |
4567 | ({ \ |
4568 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/pgtable-3level.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgtable-3level.h |
4569 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/pgtable-3level.h 2008-05-23 18:33:27.000000000 +0200 |
4570 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgtable-3level.h 2008-05-23 18:27:40.000000000 +0200 |
4571 | @@ -23,26 +23,11 @@ |
4572 | #define pud_present(pud) 1 |
4573 | |
4574 | /* |
4575 | - * Is the pte executable? |
4576 | - */ |
4577 | -static inline int pte_x(pte_t pte) |
4578 | -{ |
4579 | - return !(__pte_val(pte) & _PAGE_NX); |
4580 | -} |
4581 | - |
4582 | -/* |
4583 | - * All present user-pages with !NX bit are user-executable: |
4584 | - */ |
4585 | -static inline int pte_exec(pte_t pte) |
4586 | -{ |
4587 | - return pte_user(pte) && pte_x(pte); |
4588 | -} |
4589 | -/* |
4590 | * All present pages with !NX bit are kernel-executable: |
4591 | */ |
4592 | static inline int pte_exec_kernel(pte_t pte) |
4593 | { |
4594 | - return pte_x(pte); |
4595 | + return !(__pte_val(pte) & _PAGE_NX); |
4596 | } |
4597 | |
4598 | /* Rules for using set_pte: the pte being assigned *must* be |
4599 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/pgtable_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgtable_64.h |
4600 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/pgtable_64.h 2008-05-23 18:33:27.000000000 +0200 |
4601 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/pgtable_64.h 2008-05-23 18:27:40.000000000 +0200 |
4602 | @@ -332,21 +332,15 @@ |
4603 | * Undefined behaviour if not.. |
4604 | */ |
4605 | #define __LARGE_PTE (_PAGE_PSE|_PAGE_PRESENT) |
4606 | -static inline int pte_user(pte_t pte) { return __pte_val(pte) & _PAGE_USER; } |
4607 | -static inline int pte_read(pte_t pte) { return __pte_val(pte) & _PAGE_USER; } |
4608 | -static inline int pte_exec(pte_t pte) { return !(__pte_val(pte) & _PAGE_NX); } |
4609 | static inline int pte_dirty(pte_t pte) { return __pte_val(pte) & _PAGE_DIRTY; } |
4610 | static inline int pte_young(pte_t pte) { return __pte_val(pte) & _PAGE_ACCESSED; } |
4611 | static inline int pte_write(pte_t pte) { return __pte_val(pte) & _PAGE_RW; } |
4612 | static inline int pte_file(pte_t pte) { return __pte_val(pte) & _PAGE_FILE; } |
4613 | static inline int pte_huge(pte_t pte) { return __pte_val(pte) & _PAGE_PSE; } |
4614 | |
4615 | -static inline pte_t pte_rdprotect(pte_t pte) { __pte_val(pte) &= ~_PAGE_USER; return pte; } |
4616 | -static inline pte_t pte_exprotect(pte_t pte) { __pte_val(pte) &= ~_PAGE_USER; return pte; } |
4617 | static inline pte_t pte_mkclean(pte_t pte) { __pte_val(pte) &= ~_PAGE_DIRTY; return pte; } |
4618 | static inline pte_t pte_mkold(pte_t pte) { __pte_val(pte) &= ~_PAGE_ACCESSED; return pte; } |
4619 | static inline pte_t pte_wrprotect(pte_t pte) { __pte_val(pte) &= ~_PAGE_RW; return pte; } |
4620 | -static inline pte_t pte_mkread(pte_t pte) { __pte_val(pte) |= _PAGE_USER; return pte; } |
4621 | static inline pte_t pte_mkexec(pte_t pte) { __pte_val(pte) &= ~_PAGE_NX; return pte; } |
4622 | static inline pte_t pte_mkdirty(pte_t pte) { __pte_val(pte) |= _PAGE_DIRTY; return pte; } |
4623 | static inline pte_t pte_mkyoung(pte_t pte) { __pte_val(pte) |= _PAGE_ACCESSED; return pte; } |
4624 | @@ -354,13 +348,6 @@ |
4625 | static inline pte_t pte_mkhuge(pte_t pte) { __pte_val(pte) |= _PAGE_PSE; return pte; } |
4626 | static inline pte_t pte_clrhuge(pte_t pte) { __pte_val(pte) &= ~_PAGE_PSE; return pte; } |
4627 | |
4628 | -static inline int ptep_test_and_clear_dirty(struct vm_area_struct *vma, unsigned long addr, pte_t *ptep) |
4629 | -{ |
4630 | - if (!pte_dirty(*ptep)) |
4631 | - return 0; |
4632 | - return test_and_clear_bit(_PAGE_BIT_DIRTY, &ptep->pte); |
4633 | -} |
4634 | - |
4635 | static inline int ptep_test_and_clear_young(struct vm_area_struct *vma, unsigned long addr, pte_t *ptep) |
4636 | { |
4637 | if (!pte_young(*ptep)) |
4638 | @@ -500,19 +487,6 @@ |
4639 | __changed; \ |
4640 | }) |
4641 | |
4642 | -#define __HAVE_ARCH_PTEP_CLEAR_DIRTY_FLUSH |
4643 | -#define ptep_clear_flush_dirty(vma, address, ptep) \ |
4644 | -({ \ |
4645 | - pte_t __pte = *(ptep); \ |
4646 | - int __dirty = pte_dirty(__pte); \ |
4647 | - __pte = pte_mkclean(__pte); \ |
4648 | - if ((vma)->vm_mm->context.pinned) \ |
4649 | - (void)ptep_set_access_flags(vma, address, ptep, __pte, __dirty); \ |
4650 | - else if (__dirty) \ |
4651 | - set_pte(ptep, __pte); \ |
4652 | - __dirty; \ |
4653 | -}) |
4654 | - |
4655 | #define __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH |
4656 | #define ptep_clear_flush_young(vma, address, ptep) \ |
4657 | ({ \ |
4658 | @@ -570,6 +544,8 @@ |
4659 | #define arch_change_pte_range(mm, pmd, addr, end, newprot, dirty_accountable) \ |
4660 | xen_change_pte_range(mm, pmd, addr, end, newprot, dirty_accountable) |
4661 | |
4662 | +pte_t *lookup_address(unsigned long addr); |
4663 | + |
4664 | #define io_remap_pfn_range(vma, vaddr, pfn, size, prot) \ |
4665 | direct_remap_pfn_range(vma,vaddr,pfn,size,prot,DOMID_IO) |
4666 | |
4667 | @@ -587,7 +563,6 @@ |
4668 | (((o) & (1UL << (__VIRTUAL_MASK_SHIFT-1))) ? ((o) | (~__VIRTUAL_MASK)) : (o)) |
4669 | |
4670 | #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG |
4671 | -#define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_DIRTY |
4672 | #define __HAVE_ARCH_PTEP_GET_AND_CLEAR |
4673 | #define __HAVE_ARCH_PTEP_GET_AND_CLEAR_FULL |
4674 | #define __HAVE_ARCH_PTEP_CLEAR_FLUSH |
4675 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/processor_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/processor_32.h |
4676 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/processor_32.h 2008-05-23 18:33:27.000000000 +0200 |
4677 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/processor_32.h 2008-05-23 18:27:40.000000000 +0200 |
4678 | @@ -89,7 +89,6 @@ |
4679 | #define X86_VENDOR_UMC 3 |
4680 | #define X86_VENDOR_NEXGEN 4 |
4681 | #define X86_VENDOR_CENTAUR 5 |
4682 | -#define X86_VENDOR_RISE 6 |
4683 | #define X86_VENDOR_TRANSMETA 7 |
4684 | #define X86_VENDOR_NSC 8 |
4685 | #define X86_VENDOR_NUM 9 |
4686 | @@ -122,6 +121,7 @@ |
4687 | extern void identify_boot_cpu(void); |
4688 | extern void identify_secondary_cpu(struct cpuinfo_x86 *); |
4689 | extern void print_cpu_info(struct cpuinfo_x86 *); |
4690 | +extern void init_scattered_cpuid_features(struct cpuinfo_x86 *c); |
4691 | extern unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c); |
4692 | extern unsigned short num_cache_leaves; |
4693 | |
4694 | @@ -171,17 +171,6 @@ |
4695 | write_cr4(cr4); |
4696 | } |
4697 | |
4698 | -/* |
4699 | - * NSC/Cyrix CPU indexed register access macros |
4700 | - */ |
4701 | - |
4702 | -#define getCx86(reg) ({ outb((reg), 0x22); inb(0x23); }) |
4703 | - |
4704 | -#define setCx86(reg, data) do { \ |
4705 | - outb((reg), 0x22); \ |
4706 | - outb((data), 0x23); \ |
4707 | -} while (0) |
4708 | - |
4709 | /* Stop speculative execution */ |
4710 | static inline void sync_core(void) |
4711 | { |
4712 | @@ -230,6 +219,10 @@ |
4713 | |
4714 | #define HAVE_ARCH_PICK_MMAP_LAYOUT |
4715 | |
4716 | +extern void hard_disable_TSC(void); |
4717 | +extern void disable_TSC(void); |
4718 | +extern void hard_enable_TSC(void); |
4719 | + |
4720 | /* |
4721 | * Size of io_bitmap. |
4722 | */ |
4723 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/processor_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/processor_64.h |
4724 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/processor_64.h 2008-05-23 18:33:27.000000000 +0200 |
4725 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/processor_64.h 2008-05-23 18:27:40.000000000 +0200 |
4726 | @@ -83,7 +83,6 @@ |
4727 | #define X86_VENDOR_UMC 3 |
4728 | #define X86_VENDOR_NEXGEN 4 |
4729 | #define X86_VENDOR_CENTAUR 5 |
4730 | -#define X86_VENDOR_RISE 6 |
4731 | #define X86_VENDOR_TRANSMETA 7 |
4732 | #define X86_VENDOR_NUM 8 |
4733 | #define X86_VENDOR_UNKNOWN 0xff |
4734 | @@ -100,6 +99,7 @@ |
4735 | |
4736 | extern void identify_cpu(struct cpuinfo_x86 *); |
4737 | extern void print_cpu_info(struct cpuinfo_x86 *); |
4738 | +extern void init_scattered_cpuid_features(struct cpuinfo_x86 *c); |
4739 | extern unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c); |
4740 | extern unsigned short num_cache_leaves; |
4741 | |
4742 | @@ -377,12 +377,10 @@ |
4743 | asm volatile("cpuid" : "=a" (tmp) : "0" (1) : "ebx","ecx","edx","memory"); |
4744 | } |
4745 | |
4746 | -#define cpu_has_fpu 1 |
4747 | - |
4748 | #define ARCH_HAS_PREFETCH |
4749 | static inline void prefetch(void *x) |
4750 | { |
4751 | - asm volatile("prefetcht0 %0" :: "m" (*(unsigned long *)x)); |
4752 | + asm volatile("prefetcht0 (%0)" :: "r" (x)); |
4753 | } |
4754 | |
4755 | #define ARCH_HAS_PREFETCHW 1 |
4756 | @@ -400,17 +398,6 @@ |
4757 | |
4758 | #define cpu_relax() rep_nop() |
4759 | |
4760 | -/* |
4761 | - * NSC/Cyrix CPU indexed register access macros |
4762 | - */ |
4763 | - |
4764 | -#define getCx86(reg) ({ outb((reg), 0x22); inb(0x23); }) |
4765 | - |
4766 | -#define setCx86(reg, data) do { \ |
4767 | - outb((reg), 0x22); \ |
4768 | - outb((data), 0x23); \ |
4769 | -} while (0) |
4770 | - |
4771 | static inline void serialize_cpu(void) |
4772 | { |
4773 | __asm__ __volatile__ ("cpuid" : : : "ax", "bx", "cx", "dx"); |
4774 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/system_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/system_32.h |
4775 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/system_32.h 2008-05-23 18:33:27.000000000 +0200 |
4776 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/system_32.h 2008-05-23 18:27:40.000000000 +0200 |
4777 | @@ -205,11 +205,6 @@ |
4778 | */ |
4779 | |
4780 | |
4781 | -/* |
4782 | - * Actually only lfence would be needed for mb() because all stores done |
4783 | - * by the kernel should be already ordered. But keep a full barrier for now. |
4784 | - */ |
4785 | - |
4786 | #define mb() alternative("lock; addl $0,0(%%esp)", "mfence", X86_FEATURE_XMM2) |
4787 | #define rmb() alternative("lock; addl $0,0(%%esp)", "lfence", X86_FEATURE_XMM2) |
4788 | |
4789 | @@ -301,15 +296,6 @@ |
4790 | extern int es7000_plat; |
4791 | void cpu_idle_wait(void); |
4792 | |
4793 | -/* |
4794 | - * On SMP systems, when the scheduler does migration-cost autodetection, |
4795 | - * it needs a way to flush as much of the CPU's caches as possible: |
4796 | - */ |
4797 | -static inline void sched_cacheflush(void) |
4798 | -{ |
4799 | - wbinvd(); |
4800 | -} |
4801 | - |
4802 | extern unsigned long arch_align_stack(unsigned long sp); |
4803 | extern void free_init_pages(char *what, unsigned long begin, unsigned long end); |
4804 | |
4805 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/system_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/system_64.h |
4806 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/system_64.h 2008-05-23 18:33:27.000000000 +0200 |
4807 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/system_64.h 2008-05-23 18:27:40.000000000 +0200 |
4808 | @@ -79,12 +79,16 @@ |
4809 | unsigned long cr0; |
4810 | asm volatile("movq %%cr0,%0" : "=r" (cr0)); |
4811 | return cr0; |
4812 | -} |
4813 | +} |
4814 | |
4815 | static inline void write_cr0(unsigned long val) |
4816 | { |
4817 | asm volatile("movq %0,%%cr0" :: "r" (val)); |
4818 | -} |
4819 | +} |
4820 | + |
4821 | +#define read_cr2() current_vcpu_info()->arch.cr2 |
4822 | + |
4823 | +#define write_cr2(val) ((void)(current_vcpu_info()->arch.cr2 = (val))) |
4824 | |
4825 | #define read_cr3() ({ \ |
4826 | unsigned long __dummy; \ |
4827 | @@ -103,27 +107,28 @@ |
4828 | unsigned long cr4; |
4829 | asm("movq %%cr4,%0" : "=r" (cr4)); |
4830 | return cr4; |
4831 | -} |
4832 | +} |
4833 | |
4834 | static inline void write_cr4(unsigned long val) |
4835 | { |
4836 | asm volatile("movq %0,%%cr4" :: "r" (val) : "memory"); |
4837 | -} |
4838 | - |
4839 | -#define stts() (HYPERVISOR_fpu_taskswitch(1)) |
4840 | +} |
4841 | |
4842 | -#define wbinvd() \ |
4843 | - __asm__ __volatile__ ("wbinvd": : :"memory"); |
4844 | +static inline unsigned long read_cr8(void) |
4845 | +{ |
4846 | + return 0; |
4847 | +} |
4848 | |
4849 | -/* |
4850 | - * On SMP systems, when the scheduler does migration-cost autodetection, |
4851 | - * it needs a way to flush as much of the CPU's caches as possible. |
4852 | - */ |
4853 | -static inline void sched_cacheflush(void) |
4854 | +static inline void write_cr8(unsigned long val) |
4855 | { |
4856 | - wbinvd(); |
4857 | + BUG_ON(val); |
4858 | } |
4859 | |
4860 | +#define stts() (HYPERVISOR_fpu_taskswitch(1)) |
4861 | + |
4862 | +#define wbinvd() \ |
4863 | + __asm__ __volatile__ ("wbinvd": : :"memory") |
4864 | + |
4865 | #endif /* __KERNEL__ */ |
4866 | |
4867 | #define nop() __asm__ __volatile__ ("nop") |
4868 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/tlbflush_32.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/tlbflush_32.h |
4869 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/tlbflush_32.h 2008-05-23 18:33:27.000000000 +0200 |
4870 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/tlbflush_32.h 2008-05-23 18:27:40.000000000 +0200 |
4871 | @@ -91,7 +91,11 @@ |
4872 | DECLARE_PER_CPU(struct tlb_state, cpu_tlbstate); |
4873 | #endif /* SMP */ |
4874 | |
4875 | -#define flush_tlb_kernel_range(start, end) flush_tlb_all() |
4876 | +static inline void flush_tlb_kernel_range(unsigned long start, |
4877 | + unsigned long end) |
4878 | +{ |
4879 | + flush_tlb_all(); |
4880 | +} |
4881 | |
4882 | static inline void flush_tlb_pgtables(struct mm_struct *mm, |
4883 | unsigned long start, unsigned long end) |
4884 | diff -Naur linux-2.6.25/include/asm-x86/mach-xen/asm/tlbflush_64.h linux-2.6.25-xen/include/asm-x86/mach-xen/asm/tlbflush_64.h |
4885 | --- linux-2.6.25/include/asm-x86/mach-xen/asm/tlbflush_64.h 2008-05-23 18:33:27.000000000 +0200 |
4886 | +++ linux-2.6.25-xen/include/asm-x86/mach-xen/asm/tlbflush_64.h 2008-05-23 18:27:40.000000000 +0200 |
4887 | @@ -89,7 +89,11 @@ |
4888 | |
4889 | #endif |
4890 | |
4891 | -#define flush_tlb_kernel_range(start, end) flush_tlb_all() |
4892 | +static inline void flush_tlb_kernel_range(unsigned long start, |
4893 | + unsigned long end) |
4894 | +{ |
4895 | + flush_tlb_all(); |
4896 | +} |
4897 | |
4898 | static inline void flush_tlb_pgtables(struct mm_struct *mm, |
4899 | unsigned long start, unsigned long end) |
4900 | diff -Naur linux-2.6.25/include/asm-x86/thread_info_32.h linux-2.6.25-xen/include/asm-x86/thread_info_32.h |
4901 | --- linux-2.6.25/include/asm-x86/thread_info_32.h 2008-05-23 18:33:22.000000000 +0200 |
4902 | +++ linux-2.6.25-xen/include/asm-x86/thread_info_32.h 2008-05-23 18:27:40.000000000 +0200 |
4903 | @@ -177,7 +177,8 @@ |
4904 | #define _TIF_WORK_CTXSW_PREV _TIF_WORK_CTXSW |
4905 | #define _TIF_WORK_CTXSW_NEXT (_TIF_WORK_CTXSW | _TIF_DEBUG) |
4906 | #else |
4907 | -#define _TIF_WORK_CTXSW _TIF_DEBUG |
4908 | +#define _TIF_WORK_CTXSW_NEXT (_TIF_NOTSC | _TIF_DEBUG) |
4909 | +#define _TIF_WORK_CTXSW_PREV (_TIF_NOTSC) |
4910 | #endif |
4911 | |
4912 | |
4913 | diff -Naur linux-2.6.25/include/asm-x86/xen/interface.h linux-2.6.25-xen/include/asm-x86/xen/interface.h |
4914 | --- linux-2.6.25/include/asm-x86/xen/interface.h 2008-04-17 04:49:44.000000000 +0200 |
4915 | +++ linux-2.6.25-xen/include/asm-x86/xen/interface.h 2008-05-23 18:27:40.000000000 +0200 |
4916 | @@ -10,27 +10,21 @@ |
4917 | #define __XEN_PUBLIC_ARCH_X86_32_H__ |
4918 | |
4919 | #ifdef __XEN__ |
4920 | -#define __DEFINE_GUEST_HANDLE(name, type) \ |
4921 | +#define __DEFINE_XEN_GUEST_HANDLE(name, type) \ |
4922 | typedef struct { type *p; } __guest_handle_ ## name |
4923 | #else |
4924 | -#define __DEFINE_GUEST_HANDLE(name, type) \ |
4925 | +#define __DEFINE_XEN_GUEST_HANDLE(name, type) \ |
4926 | typedef type * __guest_handle_ ## name |
4927 | #endif |
4928 | |
4929 | -#define DEFINE_GUEST_HANDLE_STRUCT(name) \ |
4930 | - __DEFINE_GUEST_HANDLE(name, struct name) |
4931 | -#define DEFINE_GUEST_HANDLE(name) __DEFINE_GUEST_HANDLE(name, name) |
4932 | -#define GUEST_HANDLE(name) __guest_handle_ ## name |
4933 | +#define DEFINE_XEN_GUEST_HANDLE_STRUCT(name) \ |
4934 | + __DEFINE_XEN_GUEST_HANDLE(name, struct name) |
4935 | +#define DEFINE_XEN_GUEST_HANDLE(name) __DEFINE_XEN_GUEST_HANDLE(name, name) |
4936 | +#define XEN_GUEST_HANDLE(name) __guest_handle_ ## name |
4937 | |
4938 | #ifndef __ASSEMBLY__ |
4939 | -/* Guest handles for primitive C types. */ |
4940 | -__DEFINE_GUEST_HANDLE(uchar, unsigned char); |
4941 | -__DEFINE_GUEST_HANDLE(uint, unsigned int); |
4942 | -__DEFINE_GUEST_HANDLE(ulong, unsigned long); |
4943 | -DEFINE_GUEST_HANDLE(char); |
4944 | -DEFINE_GUEST_HANDLE(int); |
4945 | -DEFINE_GUEST_HANDLE(long); |
4946 | -DEFINE_GUEST_HANDLE(void); |
4947 | +typedef unsigned long xen_pfn_t; |
4948 | +typedef unsigned long xen_ulong_t; |
4949 | #endif |
4950 | |
4951 | /* |
4952 | @@ -105,7 +99,7 @@ |
4953 | uint16_t cs; /* code selector */ |
4954 | unsigned long address; /* code offset */ |
4955 | }; |
4956 | -DEFINE_GUEST_HANDLE_STRUCT(trap_info); |
4957 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(trap_info); |
4958 | |
4959 | struct cpu_user_regs { |
4960 | uint32_t ebx; |
4961 | @@ -129,7 +123,7 @@ |
4962 | uint16_t fs, _pad4; |
4963 | uint16_t gs, _pad5; |
4964 | }; |
4965 | -DEFINE_GUEST_HANDLE_STRUCT(cpu_user_regs); |
4966 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(cpu_user_regs); |
4967 | |
4968 | typedef uint64_t tsc_timestamp_t; /* RDTSC timestamp */ |
4969 | |
4970 | @@ -157,7 +151,7 @@ |
4971 | unsigned long failsafe_callback_eip; |
4972 | unsigned long vm_assist; /* VMASST_TYPE_* bitmap */ |
4973 | }; |
4974 | -DEFINE_GUEST_HANDLE_STRUCT(vcpu_guest_context); |
4975 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(vcpu_guest_context); |
4976 | |
4977 | struct arch_shared_info { |
4978 | unsigned long max_pfn; /* max pfn that appears in table */ |
4979 | diff -Naur linux-2.6.25/include/linux/elfnote.h linux-2.6.25-xen/include/linux/elfnote.h |
4980 | --- linux-2.6.25/include/linux/elfnote.h 2008-04-17 04:49:44.000000000 +0200 |
4981 | +++ linux-2.6.25-xen/include/linux/elfnote.h 2008-05-23 18:27:40.000000000 +0200 |
4982 | @@ -52,7 +52,7 @@ |
4983 | 4484:.balign 4 ; \ |
4984 | .popsection ; |
4985 | |
4986 | -#define ELFNOTE(name, type, desc) \ |
4987 | +#define ELFNOTE(name, type, desc...) \ |
4988 | ELFNOTE_START(name, type, "") \ |
4989 | desc ; \ |
4990 | ELFNOTE_END |
4991 | diff -Naur linux-2.6.25/include/linux/page-flags.h linux-2.6.25-xen/include/linux/page-flags.h |
4992 | --- linux-2.6.25/include/linux/page-flags.h 2008-05-23 18:33:11.000000000 +0200 |
4993 | +++ linux-2.6.25-xen/include/linux/page-flags.h 2008-05-23 18:27:40.000000000 +0200 |
4994 | @@ -83,6 +83,11 @@ |
4995 | #define PG_private 11 /* If pagecache, has fs-private data */ |
4996 | |
4997 | #define PG_writeback 12 /* Page is under writeback */ |
4998 | +#ifdef CONFIG_XEN |
4999 | +/* Cannot alias with PG_owner_priv_1 since bag_page() checks include this bit. |
5000 | + * Also cannot use PG_arch_1 since that now has a different purpose on x86. */ |
5001 | +#define PG_pinned 13 |
5002 | +#endif |
5003 | #define PG_compound 14 /* Part of a compound page */ |
5004 | #define PG_swapcache 15 /* Swap page: swp_entry_t in private */ |
5005 | |
5006 | @@ -96,7 +101,9 @@ |
5007 | |
5008 | /* PG_owner_priv_1 users should have descriptive aliases */ |
5009 | #define PG_checked PG_owner_priv_1 /* Used by some filesystems */ |
5010 | +#ifdef CONFIG_PARAVIRT_XEN |
5011 | #define PG_pinned PG_owner_priv_1 /* Xen pinned pagetable */ |
5012 | +#endif |
5013 | |
5014 | #if (BITS_PER_LONG > 32) |
5015 | /* |
5016 | diff -Naur linux-2.6.25/include/linux/skbuff.h linux-2.6.25-xen/include/linux/skbuff.h |
5017 | --- linux-2.6.25/include/linux/skbuff.h 2008-05-23 18:33:24.000000000 +0200 |
5018 | +++ linux-2.6.25-xen/include/linux/skbuff.h 2008-05-23 18:27:40.000000000 +0200 |
5019 | @@ -1822,7 +1822,7 @@ |
5020 | |
5021 | bool skb_partial_csum_set(struct sk_buff *skb, u16 start, u16 off); |
5022 | |
5023 | -#ifdef CONFIG_XEN |
5024 | +#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN) |
5025 | int skb_checksum_setup(struct sk_buff *skb); |
5026 | #else |
5027 | static inline int skb_checksum_setup(struct sk_buff *skb) { return 0; } |
5028 | diff -Naur linux-2.6.25/include/xen/driver_util.h linux-2.6.25-xen/include/xen/driver_util.h |
5029 | --- linux-2.6.25/include/xen/driver_util.h 2008-05-23 18:32:29.000000000 +0200 |
5030 | +++ linux-2.6.25-xen/include/xen/driver_util.h 2008-05-23 18:27:40.000000000 +0200 |
5031 | @@ -5,10 +5,6 @@ |
5032 | #include <linux/vmalloc.h> |
5033 | #include <linux/device.h> |
5034 | |
5035 | -/* Allocate/destroy a 'vmalloc' VM area. */ |
5036 | -extern struct vm_struct *alloc_vm_area(unsigned long size); |
5037 | -extern void free_vm_area(struct vm_struct *area); |
5038 | - |
5039 | extern struct class *get_xen_class(void); |
5040 | |
5041 | #endif /* __ASM_XEN_DRIVER_UTIL_H__ */ |
5042 | diff -Naur linux-2.6.25/include/xen/features.h linux-2.6.25-xen/include/xen/features.h |
5043 | --- linux-2.6.25/include/xen/features.h 2008-04-17 04:49:44.000000000 +0200 |
5044 | +++ linux-2.6.25-xen/include/xen/features.h 2008-05-23 18:27:40.000000000 +0200 |
5045 | @@ -10,6 +10,7 @@ |
5046 | #define __XEN_FEATURES_H__ |
5047 | |
5048 | #include <xen/interface/features.h> |
5049 | +#include <xen/interface/version.h> |
5050 | |
5051 | void xen_setup_features(void); |
5052 | |
5053 | @@ -20,4 +21,4 @@ |
5054 | return xen_features[flag]; |
5055 | } |
5056 | |
5057 | -#endif /* __ASM_XEN_FEATURES_H__ */ |
5058 | +#endif /* __XEN_FEATURES_H__ */ |
5059 | diff -Naur linux-2.6.25/include/xen/interface/arch-x86/xen.h linux-2.6.25-xen/include/xen/interface/arch-x86/xen.h |
5060 | --- linux-2.6.25/include/xen/interface/arch-x86/xen.h 2008-05-23 18:32:37.000000000 +0200 |
5061 | +++ linux-2.6.25-xen/include/xen/interface/arch-x86/xen.h 2008-05-23 18:27:40.000000000 +0200 |
5062 | @@ -49,6 +49,9 @@ |
5063 | #define get_xen_guest_handle(val, hnd) do { val = (hnd).p; } while (0) |
5064 | #endif |
5065 | |
5066 | +/* Allow co-existing Linux 2.6.23+ Xen interface definitions. */ |
5067 | +#define DEFINE_XEN_GUEST_HANDLE_STRUCT(name) struct name |
5068 | + |
5069 | #if defined(__i386__) |
5070 | #include "xen-x86_32.h" |
5071 | #elif defined(__x86_64__) |
5072 | diff -Naur linux-2.6.25/include/xen/interface/event_channel.h linux-2.6.25-xen/include/xen/interface/event_channel.h |
5073 | --- linux-2.6.25/include/xen/interface/event_channel.h 2008-05-23 18:33:03.000000000 +0200 |
5074 | +++ linux-2.6.25-xen/include/xen/interface/event_channel.h 2008-05-23 18:27:40.000000000 +0200 |
5075 | @@ -248,6 +248,7 @@ |
5076 | struct evtchn_unmask unmask; |
5077 | } u; |
5078 | }; |
5079 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(evtchn_op); |
5080 | typedef struct evtchn_op evtchn_op_t; |
5081 | DEFINE_XEN_GUEST_HANDLE(evtchn_op_t); |
5082 | |
5083 | diff -Naur linux-2.6.25/include/xen/interface/io/netif.h linux-2.6.25-xen/include/xen/interface/io/netif.h |
5084 | --- linux-2.6.25/include/xen/interface/io/netif.h 2008-05-23 18:33:03.000000000 +0200 |
5085 | +++ linux-2.6.25-xen/include/xen/interface/io/netif.h 2008-05-23 18:27:40.000000000 +0200 |
5086 | @@ -183,8 +183,22 @@ |
5087 | * Generate netif ring structures and types. |
5088 | */ |
5089 | |
5090 | +#if defined(CONFIG_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H) |
5091 | DEFINE_RING_TYPES(netif_tx, struct netif_tx_request, struct netif_tx_response); |
5092 | DEFINE_RING_TYPES(netif_rx, struct netif_rx_request, struct netif_rx_response); |
5093 | +#else |
5094 | +#define xen_netif_tx_request netif_tx_request |
5095 | +#define xen_netif_rx_request netif_rx_request |
5096 | +#define xen_netif_tx_response netif_tx_response |
5097 | +#define xen_netif_rx_response netif_rx_response |
5098 | +DEFINE_RING_TYPES(xen_netif_tx, |
5099 | + struct xen_netif_tx_request, |
5100 | + struct xen_netif_tx_response); |
5101 | +DEFINE_RING_TYPES(xen_netif_rx, |
5102 | + struct xen_netif_rx_request, |
5103 | + struct xen_netif_rx_response); |
5104 | +#define xen_netif_extra_info netif_extra_info |
5105 | +#endif |
5106 | |
5107 | #define NETIF_RSP_DROPPED -2 |
5108 | #define NETIF_RSP_ERROR -1 |
5109 | diff -Naur linux-2.6.25/include/xen/interface/memory.h linux-2.6.25-xen/include/xen/interface/memory.h |
5110 | --- linux-2.6.25/include/xen/interface/memory.h 2008-05-23 18:33:03.000000000 +0200 |
5111 | +++ linux-2.6.25-xen/include/xen/interface/memory.h 2008-05-23 18:27:40.000000000 +0200 |
5112 | @@ -67,6 +67,7 @@ |
5113 | */ |
5114 | domid_t domid; |
5115 | }; |
5116 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(xen_memory_reservation); |
5117 | typedef struct xen_memory_reservation xen_memory_reservation_t; |
5118 | DEFINE_XEN_GUEST_HANDLE(xen_memory_reservation_t); |
5119 | |
5120 | @@ -160,6 +161,7 @@ |
5121 | */ |
5122 | unsigned int nr_extents; |
5123 | }; |
5124 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(xen_machphys_mfn_list); |
5125 | typedef struct xen_machphys_mfn_list xen_machphys_mfn_list_t; |
5126 | DEFINE_XEN_GUEST_HANDLE(xen_machphys_mfn_list_t); |
5127 | |
5128 | @@ -198,6 +200,7 @@ |
5129 | /* GPFN where the source mapping page should appear. */ |
5130 | xen_pfn_t gpfn; |
5131 | }; |
5132 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(xen_add_to_physmap); |
5133 | typedef struct xen_add_to_physmap xen_add_to_physmap_t; |
5134 | DEFINE_XEN_GUEST_HANDLE(xen_add_to_physmap_t); |
5135 | |
5136 | @@ -222,6 +225,7 @@ |
5137 | */ |
5138 | XEN_GUEST_HANDLE(xen_pfn_t) mfn_list; |
5139 | }; |
5140 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(xen_translate_gpfn_list); |
5141 | typedef struct xen_translate_gpfn_list xen_translate_gpfn_list_t; |
5142 | DEFINE_XEN_GUEST_HANDLE(xen_translate_gpfn_list_t); |
5143 | |
5144 | diff -Naur linux-2.6.25/include/xen/interface/sched.h linux-2.6.25-xen/include/xen/interface/sched.h |
5145 | --- linux-2.6.25/include/xen/interface/sched.h 2008-05-23 18:33:03.000000000 +0200 |
5146 | +++ linux-2.6.25-xen/include/xen/interface/sched.h 2008-05-23 18:27:40.000000000 +0200 |
5147 | @@ -67,6 +67,7 @@ |
5148 | struct sched_shutdown { |
5149 | unsigned int reason; /* SHUTDOWN_* */ |
5150 | }; |
5151 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(sched_shutdown); |
5152 | typedef struct sched_shutdown sched_shutdown_t; |
5153 | DEFINE_XEN_GUEST_HANDLE(sched_shutdown_t); |
5154 | |
5155 | @@ -81,6 +82,7 @@ |
5156 | unsigned int nr_ports; |
5157 | uint64_t timeout; |
5158 | }; |
5159 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(sched_poll); |
5160 | typedef struct sched_poll sched_poll_t; |
5161 | DEFINE_XEN_GUEST_HANDLE(sched_poll_t); |
5162 | |
5163 | diff -Naur linux-2.6.25/include/xen/interface/version.h linux-2.6.25-xen/include/xen/interface/version.h |
5164 | --- linux-2.6.25/include/xen/interface/version.h 2008-05-23 18:33:03.000000000 +0200 |
5165 | +++ linux-2.6.25-xen/include/xen/interface/version.h 2008-05-23 18:27:40.000000000 +0200 |
5166 | @@ -36,6 +36,9 @@ |
5167 | /* arg == xen_extraversion_t. */ |
5168 | #define XENVER_extraversion 1 |
5169 | typedef char xen_extraversion_t[16]; |
5170 | +struct xen_extraversion { |
5171 | + xen_extraversion_t extraversion; |
5172 | +}; |
5173 | #define XEN_EXTRAVERSION_LEN (sizeof(xen_extraversion_t)) |
5174 | |
5175 | /* arg == xen_compile_info_t. */ |
5176 | @@ -50,10 +53,16 @@ |
5177 | |
5178 | #define XENVER_capabilities 3 |
5179 | typedef char xen_capabilities_info_t[1024]; |
5180 | +struct xen_capabilities_info { |
5181 | + xen_capabilities_info_t info; |
5182 | +}; |
5183 | #define XEN_CAPABILITIES_INFO_LEN (sizeof(xen_capabilities_info_t)) |
5184 | |
5185 | #define XENVER_changeset 4 |
5186 | typedef char xen_changeset_info_t[64]; |
5187 | +struct xen_changeset_info { |
5188 | + xen_changeset_info_t info; |
5189 | +}; |
5190 | #define XEN_CHANGESET_INFO_LEN (sizeof(xen_changeset_info_t)) |
5191 | |
5192 | #define XENVER_platform_parameters 5 |
5193 | diff -Naur linux-2.6.25/include/xen/interface/xen.h linux-2.6.25-xen/include/xen/interface/xen.h |
5194 | --- linux-2.6.25/include/xen/interface/xen.h 2008-05-23 18:33:03.000000000 +0200 |
5195 | +++ linux-2.6.25-xen/include/xen/interface/xen.h 2008-05-23 18:27:40.000000000 +0200 |
5196 | @@ -29,7 +29,9 @@ |
5197 | |
5198 | #include "xen-compat.h" |
5199 | |
5200 | -#if defined(__i386__) || defined(__x86_64__) |
5201 | +#if defined(CONFIG_PARAVIRT_XEN) && !defined(HAVE_XEN_PLATFORM_COMPAT_H) |
5202 | +#include <asm/xen/interface.h> |
5203 | +#elif defined(__i386__) || defined(__x86_64__) |
5204 | #include "arch-x86/xen.h" |
5205 | #elif defined(__ia64__) |
5206 | #include "arch-ia64.h" |
5207 | @@ -109,7 +111,7 @@ |
5208 | */ |
5209 | |
5210 | /* New sched_op hypercall introduced in 0x00030101. */ |
5211 | -#if __XEN_INTERFACE_VERSION__ < 0x00030101 |
5212 | +#if __XEN_INTERFACE_VERSION__ < 0x00030101 || (defined(CONFIG_PARAVIRT_XEN) && !defined(HAVE_XEN_PLATFORM_COMPAT_H)) |
5213 | #undef __HYPERVISOR_sched_op |
5214 | #define __HYPERVISOR_sched_op __HYPERVISOR_sched_op_compat |
5215 | #endif |
5216 | @@ -123,7 +125,7 @@ |
5217 | #endif |
5218 | |
5219 | /* New platform_op hypercall introduced in 0x00030204. */ |
5220 | -#if __XEN_INTERFACE_VERSION__ < 0x00030204 |
5221 | +#if __XEN_INTERFACE_VERSION__ < 0x00030204 || (defined(CONFIG_PARAVIRT_XEN) && !defined(HAVE_XEN_PLATFORM_COMPAT_H)) |
5222 | #define __HYPERVISOR_dom0_op __HYPERVISOR_platform_op |
5223 | #endif |
5224 | |
5225 | @@ -270,6 +272,7 @@ |
5226 | #endif |
5227 | } arg2; |
5228 | }; |
5229 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(mmuext_op); |
5230 | typedef struct mmuext_op mmuext_op_t; |
5231 | DEFINE_XEN_GUEST_HANDLE(mmuext_op_t); |
5232 | #endif |
5233 | @@ -352,6 +355,7 @@ |
5234 | uint64_t ptr; /* Machine address of PTE. */ |
5235 | uint64_t val; /* New contents of PTE. */ |
5236 | }; |
5237 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(mmu_update); |
5238 | typedef struct mmu_update mmu_update_t; |
5239 | DEFINE_XEN_GUEST_HANDLE(mmu_update_t); |
5240 | |
5241 | @@ -360,9 +364,15 @@ |
5242 | * NB. The fields are natural register size for this architecture. |
5243 | */ |
5244 | struct multicall_entry { |
5245 | - unsigned long op, result; |
5246 | + unsigned long op; |
5247 | +#if !defined(CONFIG_PARAVIRT_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H) |
5248 | + unsigned long result; |
5249 | +#else |
5250 | + long result; |
5251 | +#endif |
5252 | unsigned long args[6]; |
5253 | }; |
5254 | +DEFINE_XEN_GUEST_HANDLE_STRUCT(multicall_entry); |
5255 | typedef struct multicall_entry multicall_entry_t; |
5256 | DEFINE_XEN_GUEST_HANDLE(multicall_entry_t); |
5257 | |
5258 | diff -Naur linux-2.6.25/include/xen/xenbus.h linux-2.6.25-xen/include/xen/xenbus.h |
5259 | --- linux-2.6.25/include/xen/xenbus.h 2008-05-23 18:33:27.000000000 +0200 |
5260 | +++ linux-2.6.25-xen/include/xen/xenbus.h 2008-05-23 18:27:40.000000000 +0200 |
5261 | @@ -57,16 +57,20 @@ |
5262 | void (*callback)(struct xenbus_watch *, |
5263 | const char **vec, unsigned int len); |
5264 | |
5265 | +#if defined(CONFIG_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H) |
5266 | /* See XBWF_ definitions below. */ |
5267 | unsigned long flags; |
5268 | +#endif |
5269 | }; |
5270 | |
5271 | +#if defined(CONFIG_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H) |
5272 | /* |
5273 | * Execute callback in its own kthread. Useful if the callback is long |
5274 | * running or heavily serialised, to avoid taking out the main xenwatch thread |
5275 | * for a long period of time (or even unwittingly causing a deadlock). |
5276 | */ |
5277 | #define XBWF_new_thread 1 |
5278 | +#endif |
5279 | |
5280 | /* A xenbus device. */ |
5281 | struct xenbus_device { |
5282 | @@ -214,6 +218,7 @@ |
5283 | const char **, unsigned int)); |
5284 | |
5285 | |
5286 | +#if defined(CONFIG_XEN) || defined(HAVE_XEN_PLATFORM_COMPAT_H) |
5287 | /** |
5288 | * Register a watch on the given path/path2, using the given xenbus_watch |
5289 | * structure for storage, and the given callback function as the callback. |
5290 | @@ -227,7 +232,13 @@ |
5291 | const char *path2, struct xenbus_watch *watch, |
5292 | void (*callback)(struct xenbus_watch *, |
5293 | const char **, unsigned int)); |
5294 | - |
5295 | +#else |
5296 | +int xenbus_watch_pathfmt(struct xenbus_device *dev, struct xenbus_watch *watch, |
5297 | + void (*callback)(struct xenbus_watch *, |
5298 | + const char **, unsigned int), |
5299 | + const char *pathfmt, ...) |
5300 | + __attribute__ ((format (printf, 4, 5))); |
5301 | +#endif |
5302 | |
5303 | /** |
5304 | * Advertise in the store a change of the given driver to the given new_state. |
5305 | diff -Naur linux-2.6.25/net/core/dev.c linux-2.6.25-xen/net/core/dev.c |
5306 | --- linux-2.6.25/net/core/dev.c 2008-05-23 18:33:27.000000000 +0200 |
5307 | +++ linux-2.6.25-xen/net/core/dev.c 2008-05-23 18:27:40.000000000 +0200 |
5308 | @@ -122,7 +122,7 @@ |
5309 | |
5310 | #include "net-sysfs.h" |
5311 | |
5312 | -#ifdef CONFIG_XEN |
5313 | +#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN) |
5314 | #include <net/ip.h> |
5315 | #include <linux/tcp.h> |
5316 | #include <linux/udp.h> |
5317 | @@ -1582,42 +1582,54 @@ |
5318 | return 0; |
5319 | } |
5320 | |
5321 | -#ifdef CONFIG_XEN |
5322 | +#if defined(CONFIG_XEN) || defined(CONFIG_PARAVIRT_XEN) |
5323 | inline int skb_checksum_setup(struct sk_buff *skb) |
5324 | { |
5325 | - if (skb->proto_csum_blank) { |
5326 | - struct iphdr *iph; |
5327 | - unsigned char *th; |
5328 | + struct iphdr *iph; |
5329 | + unsigned char *th; |
5330 | + int err = -EPROTO; |
5331 | |
5332 | - if (skb->protocol != htons(ETH_P_IP)) |
5333 | - goto out; |
5334 | - iph = ip_hdr(skb); |
5335 | - th = skb_network_header(skb) + 4 * iph->ihl; |
5336 | - if (th >= skb_tail_pointer(skb)) |
5337 | - goto out; |
5338 | - skb->csum_start = th - skb->head; |
5339 | - switch (iph->protocol) { |
5340 | - case IPPROTO_TCP: |
5341 | - skb->csum_offset = offsetof(struct tcphdr, check); |
5342 | - break; |
5343 | - case IPPROTO_UDP: |
5344 | - skb->csum_offset = offsetof(struct udphdr, check); |
5345 | - break; |
5346 | - default: |
5347 | - if (net_ratelimit()) |
5348 | - printk(KERN_ERR "Attempting to checksum a non-" |
5349 | - "TCP/UDP packet, dropping a protocol" |
5350 | - " %d packet", iph->protocol); |
5351 | - goto out; |
5352 | - } |
5353 | - if ((th + skb->csum_offset + 2) > skb_tail_pointer(skb)) |
5354 | - goto out; |
5355 | - skb->ip_summed = CHECKSUM_PARTIAL; |
5356 | - skb->proto_csum_blank = 0; |
5357 | +#ifdef CONFIG_XEN |
5358 | + if (!skb->proto_csum_blank) |
5359 | + return 0; |
5360 | +#endif |
5361 | + |
5362 | + if (skb->protocol != htons(ETH_P_IP)) |
5363 | + goto out; |
5364 | + |
5365 | + iph = ip_hdr(skb); |
5366 | + th = skb_network_header(skb) + 4 * iph->ihl; |
5367 | + if (th >= skb_tail_pointer(skb)) |
5368 | + goto out; |
5369 | + |
5370 | + skb->csum_start = th - skb->head; |
5371 | + switch (iph->protocol) { |
5372 | + case IPPROTO_TCP: |
5373 | + skb->csum_offset = offsetof(struct tcphdr, check); |
5374 | + break; |
5375 | + case IPPROTO_UDP: |
5376 | + skb->csum_offset = offsetof(struct udphdr, check); |
5377 | + break; |
5378 | + default: |
5379 | + if (net_ratelimit()) |
5380 | + printk(KERN_ERR "Attempting to checksum a non-" |
5381 | + "TCP/UDP packet, dropping a protocol" |
5382 | + " %d packet", iph->protocol); |
5383 | + goto out; |
5384 | } |
5385 | - return 0; |
5386 | + |
5387 | + if ((th + skb->csum_offset + 2) > skb_tail_pointer(skb)) |
5388 | + goto out; |
5389 | + |
5390 | +#ifdef CONFIG_XEN |
5391 | + skb->ip_summed = CHECKSUM_PARTIAL; |
5392 | + skb->proto_csum_blank = 0; |
5393 | +#endif |
5394 | + |
5395 | + err = 0; |
5396 | + |
5397 | out: |
5398 | - return -EPROTO; |
5399 | + return err; |
5400 | } |
5401 | EXPORT_SYMBOL(skb_checksum_setup); |
5402 | #endif |