blob: 67897152237cd4bcaa0cbcc80b127c0e47d8571b [file] [log] [blame]
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001/*
2 * Core of Xen paravirt_ops implementation.
3 *
4 * This file contains the xen_paravirt_ops structure itself, and the
5 * implementations for:
6 * - privileged instructions
7 * - interrupt flags
8 * - segment operations
9 * - booting and setup
10 *
11 * Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
12 */
13
Sheng Yang38e20b02010-05-14 12:40:51 +010014#include <linux/cpu.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070015#include <linux/kernel.h>
16#include <linux/init.h>
17#include <linux/smp.h>
18#include <linux/preempt.h>
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -070019#include <linux/hardirq.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070020#include <linux/percpu.h>
21#include <linux/delay.h>
22#include <linux/start_kernel.h>
23#include <linux/sched.h>
Jeremy Fitzhardinge6cac5a92009-03-29 19:56:29 -070024#include <linux/kprobes.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070025#include <linux/bootmem.h>
26#include <linux/module.h>
Jeremy Fitzhardingef4f97b32007-07-17 18:37:05 -070027#include <linux/mm.h>
28#include <linux/page-flags.h>
29#include <linux/highmem.h>
Markus Armbrusterb8c2d3d2008-02-27 14:56:35 +010030#include <linux/console.h>
Chris Wright5d990b62009-12-04 12:15:21 -080031#include <linux/pci.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090032#include <linux/gfp.h>
Jeremy Fitzhardinge236260b2010-10-06 15:52:29 -070033#include <linux/memblock.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070034
Jeremy Fitzhardinge1ccbf532009-10-06 15:11:14 -070035#include <xen/xen.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070036#include <xen/interface/xen.h>
Jeremy Fitzhardingeecbf29c2008-12-16 12:37:07 -080037#include <xen/interface/version.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070038#include <xen/interface/physdev.h>
39#include <xen/interface/vcpu.h>
Sheng Yangbee6ab52010-05-14 12:39:33 +010040#include <xen/interface/memory.h>
Liu, Jinsongcef12ee2012-06-07 19:56:51 +080041#include <xen/interface/xen-mca.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070042#include <xen/features.h>
43#include <xen/page.h>
Sheng Yang38e20b02010-05-14 12:40:51 +010044#include <xen/hvm.h>
Jeremy Fitzhardinge084a2a42008-07-08 15:06:50 -070045#include <xen/hvc-console.h>
Konrad Rzeszutek Wilk211063d2011-12-08 17:32:23 +080046#include <xen/acpi.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070047
48#include <asm/paravirt.h>
Ingo Molnar7b6aa332009-02-17 13:58:15 +010049#include <asm/apic.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070050#include <asm/page.h>
Alex Nixonb5401a92010-03-18 16:31:34 -040051#include <asm/xen/pci.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070052#include <asm/xen/hypercall.h>
53#include <asm/xen/hypervisor.h>
54#include <asm/fixmap.h>
55#include <asm/processor.h>
Jeremy Fitzhardinge707ebbc2009-03-27 11:29:02 -070056#include <asm/proto.h>
Jeremy Fitzhardinge11539682008-07-08 15:07:16 -070057#include <asm/msr-index.h>
Jeremy Fitzhardinge6cac5a92009-03-29 19:56:29 -070058#include <asm/traps.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070059#include <asm/setup.h>
60#include <asm/desc.h>
Ian Campbell817a8242010-02-26 17:16:00 +000061#include <asm/pgalloc.h>
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070062#include <asm/pgtable.h>
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -070063#include <asm/tlbflush.h>
Jeremy Fitzhardingefefa6292007-07-17 18:37:07 -070064#include <asm/reboot.h>
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -070065#include <asm/stackprotector.h>
Sheng Yangbee6ab52010-05-14 12:39:33 +010066#include <asm/hypervisor.h>
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -050067#include <asm/mwait.h>
David Vrabel76a8df72012-05-04 14:29:46 +010068#include <asm/pci_x86.h>
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -050069
70#ifdef CONFIG_ACPI
71#include <linux/acpi.h>
72#include <asm/acpi.h>
73#include <acpi/pdc_intel.h>
74#include <acpi/processor.h>
75#include <xen/interface/platform.h>
76#endif
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070077
78#include "xen-ops.h"
Jeremy Fitzhardinge3b827c12007-07-17 18:37:04 -070079#include "mmu.h"
Ben Guthrof447d562012-04-21 00:11:04 +080080#include "smp.h"
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070081#include "multicalls.h"
82
Konrad Rzeszutek Wilkb8b0f552012-08-21 14:49:34 -040083#include <xen/events.h>
84
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070085EXPORT_SYMBOL_GPL(hypercall_page);
86
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070087DEFINE_PER_CPU(struct vcpu_info *, xen_vcpu);
88DEFINE_PER_CPU(struct vcpu_info, xen_vcpu_info);
Jeremy Fitzhardinge9f799912007-10-16 11:51:30 -070089
Jeremy Fitzhardinge6e833582008-08-19 13:16:17 -070090enum xen_domain_type xen_domain_type = XEN_NATIVE;
91EXPORT_SYMBOL_GPL(xen_domain_type);
92
Ian Campbell7e775062010-09-30 12:37:26 +010093unsigned long *machine_to_phys_mapping = (void *)MACH2PHYS_VIRT_START;
94EXPORT_SYMBOL(machine_to_phys_mapping);
Jan Beulichccbcdf72011-08-16 15:07:41 +010095unsigned long machine_to_phys_nr;
96EXPORT_SYMBOL(machine_to_phys_nr);
Ian Campbell7e775062010-09-30 12:37:26 +010097
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -070098struct start_info *xen_start_info;
99EXPORT_SYMBOL_GPL(xen_start_info);
100
Jeremy Fitzhardingea0d695c2008-05-26 23:31:21 +0100101struct shared_info xen_dummy_shared_info;
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700102
Jeremy Fitzhardinge38341432009-02-02 13:55:31 -0800103void *xen_initial_gdt;
104
Sheng Yangbee6ab52010-05-14 12:39:33 +0100105RESERVE_BRK(shared_info_page_brk, PAGE_SIZE);
Sheng Yang38e20b02010-05-14 12:40:51 +0100106__read_mostly int xen_have_vector_callback;
107EXPORT_SYMBOL_GPL(xen_have_vector_callback);
Sheng Yangbee6ab52010-05-14 12:39:33 +0100108
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700109/*
110 * Point at some empty memory to start with. We map the real shared_info
111 * page as soon as fixmap is up and running.
112 */
Olaf Hering4648da72012-07-17 11:57:41 +0200113struct shared_info *HYPERVISOR_shared_info = &xen_dummy_shared_info;
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700114
115/*
116 * Flag to determine whether vcpu info placement is available on all
117 * VCPUs. We assume it is to start with, and then set it to zero on
118 * the first failure. This is because it can succeed on some VCPUs
119 * and not others, since it can involve hypervisor memory allocation,
120 * or because the guest failed to guarantee all the appropriate
121 * constraints on all VCPUs (ie buffer can't cross a page boundary).
122 *
123 * Note that any particular CPU may be using a placed vcpu structure,
124 * but we can only optimise if the all are.
125 *
126 * 0: not available, 1: available
127 */
Jeremy Fitzhardingee4d04072009-02-02 13:55:54 -0800128static int have_vcpu_info_placement = 1;
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700129
David Vrabel1c32cdc2012-07-09 11:39:08 +0100130struct tls_descs {
131 struct desc_struct desc[3];
132};
133
134/*
135 * Updating the 3 TLS descriptors in the GDT on every task switch is
136 * surprisingly expensive so we avoid updating them if they haven't
137 * changed. Since Xen writes different descriptors than the one
138 * passed in the update_descriptor hypercall we keep shadow copies to
139 * compare against.
140 */
141static DEFINE_PER_CPU(struct tls_descs, shadow_tls_desc);
142
Mukesh Rathorc06ee782010-07-19 10:25:08 -0700143static void clamp_max_cpus(void)
144{
145#ifdef CONFIG_SMP
146 if (setup_max_cpus > MAX_VIRT_CPUS)
147 setup_max_cpus = MAX_VIRT_CPUS;
148#endif
149}
150
Jeremy Fitzhardinge9c7a7942008-05-31 01:33:02 +0100151static void xen_vcpu_setup(int cpu)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700152{
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700153 struct vcpu_register_vcpu_info info;
154 int err;
155 struct vcpu_info *vcpup;
156
Jeremy Fitzhardingea0d695c2008-05-26 23:31:21 +0100157 BUG_ON(HYPERVISOR_shared_info == &xen_dummy_shared_info);
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700158
Mukesh Rathorc06ee782010-07-19 10:25:08 -0700159 if (cpu < MAX_VIRT_CPUS)
160 per_cpu(xen_vcpu,cpu) = &HYPERVISOR_shared_info->vcpu_info[cpu];
161
162 if (!have_vcpu_info_placement) {
163 if (cpu >= MAX_VIRT_CPUS)
164 clamp_max_cpus();
165 return;
166 }
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700167
168 vcpup = &per_cpu(xen_vcpu_info, cpu);
Jeremy Fitzhardinge9976b392009-02-27 09:19:26 -0800169 info.mfn = arbitrary_virt_to_mfn(vcpup);
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700170 info.offset = offset_in_page(vcpup);
171
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700172 /* Check to see if the hypervisor will put the vcpu_info
173 structure where we want it, which allows direct access via
174 a percpu-variable. */
175 err = HYPERVISOR_vcpu_op(VCPUOP_register_vcpu_info, cpu, &info);
176
177 if (err) {
178 printk(KERN_DEBUG "register_vcpu_info failed: err=%d\n", err);
179 have_vcpu_info_placement = 0;
Mukesh Rathorc06ee782010-07-19 10:25:08 -0700180 clamp_max_cpus();
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700181 } else {
182 /* This cpu is using the registered vcpu info, even if
183 later ones fail to. */
184 per_cpu(xen_vcpu, cpu) = vcpup;
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -0700185 }
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700186}
187
Jeremy Fitzhardinge9c7a7942008-05-31 01:33:02 +0100188/*
189 * On restore, set the vcpu placement up again.
190 * If it fails, then we're in a bad state, since
191 * we can't back out from using it...
192 */
193void xen_vcpu_restore(void)
194{
Jeremy Fitzhardinge3905bb2aa2009-11-21 08:46:29 +0800195 int cpu;
Jeremy Fitzhardinge9c7a7942008-05-31 01:33:02 +0100196
Jeremy Fitzhardinge3905bb2aa2009-11-21 08:46:29 +0800197 for_each_online_cpu(cpu) {
198 bool other_cpu = (cpu != smp_processor_id());
Jeremy Fitzhardinge9c7a7942008-05-31 01:33:02 +0100199
Jeremy Fitzhardinge3905bb2aa2009-11-21 08:46:29 +0800200 if (other_cpu &&
201 HYPERVISOR_vcpu_op(VCPUOP_down, cpu, NULL))
202 BUG();
Jeremy Fitzhardinge9c7a7942008-05-31 01:33:02 +0100203
Jeremy Fitzhardinge3905bb2aa2009-11-21 08:46:29 +0800204 xen_setup_runstate_info(cpu);
Ian Campbellbe012922009-11-21 08:35:55 +0800205
Jeremy Fitzhardinge3905bb2aa2009-11-21 08:46:29 +0800206 if (have_vcpu_info_placement)
Jeremy Fitzhardinge9c7a7942008-05-31 01:33:02 +0100207 xen_vcpu_setup(cpu);
208
Jeremy Fitzhardinge3905bb2aa2009-11-21 08:46:29 +0800209 if (other_cpu &&
210 HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL))
211 BUG();
Jeremy Fitzhardinge9c7a7942008-05-31 01:33:02 +0100212 }
213}
214
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700215static void __init xen_banner(void)
216{
Jeremy Fitzhardinge95c7c232008-07-15 13:42:34 -0700217 unsigned version = HYPERVISOR_xen_version(XENVER_version, NULL);
218 struct xen_extraversion extra;
219 HYPERVISOR_xen_version(XENVER_extraversion, &extra);
220
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700221 printk(KERN_INFO "Booting paravirtualized kernel on %s\n",
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -0700222 pv_info.name);
Jeremy Fitzhardinge95c7c232008-07-15 13:42:34 -0700223 printk(KERN_INFO "Xen version: %d.%d%s%s\n",
224 version >> 16, version & 0xffff, extra.extraversion,
Jeremy Fitzhardingee57778a2008-06-16 04:30:02 -0700225 xen_feature(XENFEAT_mmu_pt_update_preserve_ad) ? " (preserve-AD)" : "");
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700226}
227
Andre Przywara5e626252012-05-29 13:07:31 +0200228#define CPUID_THERM_POWER_LEAF 6
229#define APERFMPERF_PRESENT 0
230
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800231static __read_mostly unsigned int cpuid_leaf1_edx_mask = ~0;
232static __read_mostly unsigned int cpuid_leaf1_ecx_mask = ~0;
233
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500234static __read_mostly unsigned int cpuid_leaf1_ecx_set_mask;
235static __read_mostly unsigned int cpuid_leaf5_ecx_val;
236static __read_mostly unsigned int cpuid_leaf5_edx_val;
237
H. Peter Anvin65ea5b02008-01-30 13:30:56 +0100238static void xen_cpuid(unsigned int *ax, unsigned int *bx,
239 unsigned int *cx, unsigned int *dx)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700240{
Jeremy Fitzhardinge82d64692009-10-22 16:41:15 -0700241 unsigned maskebx = ~0;
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800242 unsigned maskecx = ~0;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700243 unsigned maskedx = ~0;
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500244 unsigned setecx = 0;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700245 /*
246 * Mask out inconvenient features, to try and disable as many
247 * unsupported kernel subsystems as possible.
248 */
Jeremy Fitzhardinge82d64692009-10-22 16:41:15 -0700249 switch (*ax) {
250 case 1:
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800251 maskecx = cpuid_leaf1_ecx_mask;
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500252 setecx = cpuid_leaf1_ecx_set_mask;
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800253 maskedx = cpuid_leaf1_edx_mask;
Jeremy Fitzhardinge82d64692009-10-22 16:41:15 -0700254 break;
255
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500256 case CPUID_MWAIT_LEAF:
257 /* Synthesize the values.. */
258 *ax = 0;
259 *bx = 0;
260 *cx = cpuid_leaf5_ecx_val;
261 *dx = cpuid_leaf5_edx_val;
262 return;
263
Andre Przywara5e626252012-05-29 13:07:31 +0200264 case CPUID_THERM_POWER_LEAF:
265 /* Disabling APERFMPERF for kernel usage */
266 maskecx = ~(1 << APERFMPERF_PRESENT);
267 break;
268
Jeremy Fitzhardinge82d64692009-10-22 16:41:15 -0700269 case 0xb:
270 /* Suppress extended topology stuff */
271 maskebx = 0;
272 break;
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800273 }
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700274
275 asm(XEN_EMULATE_PREFIX "cpuid"
H. Peter Anvin65ea5b02008-01-30 13:30:56 +0100276 : "=a" (*ax),
277 "=b" (*bx),
278 "=c" (*cx),
279 "=d" (*dx)
280 : "0" (*ax), "2" (*cx));
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800281
Jeremy Fitzhardinge82d64692009-10-22 16:41:15 -0700282 *bx &= maskebx;
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800283 *cx &= maskecx;
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500284 *cx |= setecx;
H. Peter Anvin65ea5b02008-01-30 13:30:56 +0100285 *dx &= maskedx;
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500286
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700287}
288
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500289static bool __init xen_check_mwait(void)
290{
Konrad Rzeszutek Wilkdf88b2d2012-04-26 13:13:21 -0400291#if defined(CONFIG_ACPI) && !defined(CONFIG_ACPI_PROCESSOR_AGGREGATOR) && \
292 !defined(CONFIG_ACPI_PROCESSOR_AGGREGATOR_MODULE)
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500293 struct xen_platform_op op = {
294 .cmd = XENPF_set_processor_pminfo,
295 .u.set_pminfo.id = -1,
296 .u.set_pminfo.type = XEN_PM_PDC,
297 };
298 uint32_t buf[3];
299 unsigned int ax, bx, cx, dx;
300 unsigned int mwait_mask;
301
302 /* We need to determine whether it is OK to expose the MWAIT
303 * capability to the kernel to harvest deeper than C3 states from ACPI
304 * _CST using the processor_harvest_xen.c module. For this to work, we
305 * need to gather the MWAIT_LEAF values (which the cstate.c code
306 * checks against). The hypervisor won't expose the MWAIT flag because
307 * it would break backwards compatibility; so we will find out directly
308 * from the hardware and hypercall.
309 */
310 if (!xen_initial_domain())
311 return false;
312
313 ax = 1;
314 cx = 0;
315
316 native_cpuid(&ax, &bx, &cx, &dx);
317
318 mwait_mask = (1 << (X86_FEATURE_EST % 32)) |
319 (1 << (X86_FEATURE_MWAIT % 32));
320
321 if ((cx & mwait_mask) != mwait_mask)
322 return false;
323
324 /* We need to emulate the MWAIT_LEAF and for that we need both
325 * ecx and edx. The hypercall provides only partial information.
326 */
327
328 ax = CPUID_MWAIT_LEAF;
329 bx = 0;
330 cx = 0;
331 dx = 0;
332
333 native_cpuid(&ax, &bx, &cx, &dx);
334
335 /* Ask the Hypervisor whether to clear ACPI_PDC_C_C2C3_FFH. If so,
336 * don't expose MWAIT_LEAF and let ACPI pick the IOPORT version of C3.
337 */
338 buf[0] = ACPI_PDC_REVISION_ID;
339 buf[1] = 1;
340 buf[2] = (ACPI_PDC_C_CAPABILITY_SMP | ACPI_PDC_EST_CAPABILITY_SWSMP);
341
342 set_xen_guest_handle(op.u.set_pminfo.pdc, buf);
343
344 if ((HYPERVISOR_dom0_op(&op) == 0) &&
345 (buf[2] & (ACPI_PDC_C_C1_FFH | ACPI_PDC_C_C2C3_FFH))) {
346 cpuid_leaf5_ecx_val = cx;
347 cpuid_leaf5_edx_val = dx;
348 }
349 return true;
350#else
351 return false;
352#endif
353}
Daniel Kiperad3062a2011-05-04 20:15:18 +0200354static void __init xen_init_cpuid_mask(void)
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800355{
356 unsigned int ax, bx, cx, dx;
Shan Haitao947ccf92010-11-09 11:43:36 -0800357 unsigned int xsave_mask;
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800358
359 cpuid_leaf1_edx_mask =
Liu, Jinsongcef12ee2012-06-07 19:56:51 +0800360 ~((1 << X86_FEATURE_MTRR) | /* disable MTRR */
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800361 (1 << X86_FEATURE_ACC)); /* thermal monitoring */
362
363 if (!xen_initial_domain())
364 cpuid_leaf1_edx_mask &=
365 ~((1 << X86_FEATURE_APIC) | /* disable local APIC */
366 (1 << X86_FEATURE_ACPI)); /* disable ACPI */
Shan Haitao947ccf92010-11-09 11:43:36 -0800367 ax = 1;
Konrad Rzeszutek Wilk5e287832011-09-29 13:06:42 -0400368 cx = 0;
Shan Haitao947ccf92010-11-09 11:43:36 -0800369 xen_cpuid(&ax, &bx, &cx, &dx);
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800370
Shan Haitao947ccf92010-11-09 11:43:36 -0800371 xsave_mask =
372 (1 << (X86_FEATURE_XSAVE % 32)) |
373 (1 << (X86_FEATURE_OSXSAVE % 32));
374
375 /* Xen will set CR4.OSXSAVE if supported and not disabled by force */
376 if ((cx & xsave_mask) != xsave_mask)
377 cpuid_leaf1_ecx_mask &= ~xsave_mask; /* disable XSAVE & OSXSAVE */
Konrad Rzeszutek Wilk73c154c2012-02-13 22:26:32 -0500378 if (xen_check_mwait())
379 cpuid_leaf1_ecx_set_mask = (1 << (X86_FEATURE_MWAIT % 32));
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -0800380}
381
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700382static void xen_set_debugreg(int reg, unsigned long val)
383{
384 HYPERVISOR_set_debugreg(reg, val);
385}
386
387static unsigned long xen_get_debugreg(int reg)
388{
389 return HYPERVISOR_get_debugreg(reg);
390}
391
Jeremy Fitzhardinge224101e2009-02-18 11:18:57 -0800392static void xen_end_context_switch(struct task_struct *next)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700393{
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700394 xen_mc_flush();
Jeremy Fitzhardinge224101e2009-02-18 11:18:57 -0800395 paravirt_end_context_switch(next);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700396}
397
398static unsigned long xen_store_tr(void)
399{
400 return 0;
401}
402
Jeremy Fitzhardingea05d2eb2008-07-27 08:45:02 -0700403/*
Jeremy Fitzhardingecef43bf2008-07-28 13:33:44 -0700404 * Set the page permissions for a particular virtual address. If the
405 * address is a vmalloc mapping (or other non-linear mapping), then
406 * find the linear mapping of the page and also set its protections to
407 * match.
Jeremy Fitzhardingea05d2eb2008-07-27 08:45:02 -0700408 */
409static void set_aliased_prot(void *v, pgprot_t prot)
410{
411 int level;
412 pte_t *ptep;
413 pte_t pte;
414 unsigned long pfn;
415 struct page *page;
416
417 ptep = lookup_address((unsigned long)v, &level);
418 BUG_ON(ptep == NULL);
419
420 pfn = pte_pfn(*ptep);
421 page = pfn_to_page(pfn);
422
423 pte = pfn_pte(pfn, prot);
424
425 if (HYPERVISOR_update_va_mapping((unsigned long)v, pte, 0))
426 BUG();
427
428 if (!PageHighMem(page)) {
429 void *av = __va(PFN_PHYS(pfn));
430
431 if (av != v)
432 if (HYPERVISOR_update_va_mapping((unsigned long)av, pte, 0))
433 BUG();
434 } else
435 kmap_flush_unused();
436}
437
Jeremy Fitzhardinge38ffbe62008-07-23 14:21:18 -0700438static void xen_alloc_ldt(struct desc_struct *ldt, unsigned entries)
439{
Jeremy Fitzhardingea05d2eb2008-07-27 08:45:02 -0700440 const unsigned entries_per_page = PAGE_SIZE / LDT_ENTRY_SIZE;
Jeremy Fitzhardinge38ffbe62008-07-23 14:21:18 -0700441 int i;
442
Jeremy Fitzhardingea05d2eb2008-07-27 08:45:02 -0700443 for(i = 0; i < entries; i += entries_per_page)
444 set_aliased_prot(ldt + i, PAGE_KERNEL_RO);
Jeremy Fitzhardinge38ffbe62008-07-23 14:21:18 -0700445}
446
447static void xen_free_ldt(struct desc_struct *ldt, unsigned entries)
448{
Jeremy Fitzhardingea05d2eb2008-07-27 08:45:02 -0700449 const unsigned entries_per_page = PAGE_SIZE / LDT_ENTRY_SIZE;
Jeremy Fitzhardinge38ffbe62008-07-23 14:21:18 -0700450 int i;
451
Jeremy Fitzhardingea05d2eb2008-07-27 08:45:02 -0700452 for(i = 0; i < entries; i += entries_per_page)
453 set_aliased_prot(ldt + i, PAGE_KERNEL);
Jeremy Fitzhardinge38ffbe62008-07-23 14:21:18 -0700454}
455
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700456static void xen_set_ldt(const void *addr, unsigned entries)
457{
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700458 struct mmuext_op *op;
459 struct multicall_space mcs = xen_mc_entry(sizeof(*op));
460
Jeremy Fitzhardingeab78f7a2010-12-17 15:54:28 -0800461 trace_xen_cpu_set_ldt(addr, entries);
462
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700463 op = mcs.args;
464 op->cmd = MMUEXT_SET_LDT;
Jan Beulich4dbf7af2008-01-30 13:33:14 +0100465 op->arg1.linear_addr = (unsigned long)addr;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700466 op->arg2.nr_ents = entries;
467
468 MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
469
470 xen_mc_issue(PARAVIRT_LAZY_CPU);
471}
472
Glauber de Oliveira Costa6b68f012008-01-30 13:31:12 +0100473static void xen_load_gdt(const struct desc_ptr *dtr)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700474{
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700475 unsigned long va = dtr->address;
476 unsigned int size = dtr->size + 1;
477 unsigned pages = (size + PAGE_SIZE - 1) / PAGE_SIZE;
Jeremy Fitzhardinge3ce5fa72009-03-04 15:26:00 -0800478 unsigned long frames[pages];
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700479 int f;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700480
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -0700481 /*
482 * A GDT can be up to 64k in size, which corresponds to 8192
483 * 8-byte entries, or 16 4k pages..
484 */
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700485
486 BUG_ON(size > 65536);
487 BUG_ON(va & ~PAGE_MASK);
488
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700489 for (f = 0; va < dtr->address + size; va += PAGE_SIZE, f++) {
Jeremy Fitzhardinge6ed6bf42009-03-04 13:02:18 -0800490 int level;
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -0700491 pte_t *ptep;
Jeremy Fitzhardinge6ed6bf42009-03-04 13:02:18 -0800492 unsigned long pfn, mfn;
493 void *virt;
494
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -0700495 /*
496 * The GDT is per-cpu and is in the percpu data area.
497 * That can be virtually mapped, so we need to do a
498 * page-walk to get the underlying MFN for the
499 * hypercall. The page can also be in the kernel's
500 * linear range, so we need to RO that mapping too.
501 */
502 ptep = lookup_address(va, &level);
Jeremy Fitzhardinge6ed6bf42009-03-04 13:02:18 -0800503 BUG_ON(ptep == NULL);
504
505 pfn = pte_pfn(*ptep);
506 mfn = pfn_to_mfn(pfn);
507 virt = __va(PFN_PHYS(pfn));
508
509 frames[f] = mfn;
Jeremy Fitzhardinge9976b392009-02-27 09:19:26 -0800510
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700511 make_lowmem_page_readonly((void *)va);
Jeremy Fitzhardinge6ed6bf42009-03-04 13:02:18 -0800512 make_lowmem_page_readonly(virt);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700513 }
514
Jeremy Fitzhardinge3ce5fa72009-03-04 15:26:00 -0800515 if (HYPERVISOR_set_gdt(frames, size / sizeof(struct desc_struct)))
516 BUG();
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700517}
518
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -0700519/*
520 * load_gdt for early boot, when the gdt is only mapped once
521 */
Daniel Kiperad3062a2011-05-04 20:15:18 +0200522static void __init xen_load_gdt_boot(const struct desc_ptr *dtr)
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -0700523{
524 unsigned long va = dtr->address;
525 unsigned int size = dtr->size + 1;
526 unsigned pages = (size + PAGE_SIZE - 1) / PAGE_SIZE;
527 unsigned long frames[pages];
528 int f;
529
530 /*
531 * A GDT can be up to 64k in size, which corresponds to 8192
532 * 8-byte entries, or 16 4k pages..
533 */
534
535 BUG_ON(size > 65536);
536 BUG_ON(va & ~PAGE_MASK);
537
538 for (f = 0; va < dtr->address + size; va += PAGE_SIZE, f++) {
539 pte_t pte;
540 unsigned long pfn, mfn;
541
542 pfn = virt_to_pfn(va);
543 mfn = pfn_to_mfn(pfn);
544
545 pte = pfn_pte(pfn, PAGE_KERNEL_RO);
546
547 if (HYPERVISOR_update_va_mapping((unsigned long)va, pte, 0))
548 BUG();
549
550 frames[f] = mfn;
551 }
552
553 if (HYPERVISOR_set_gdt(frames, size / sizeof(struct desc_struct)))
554 BUG();
555}
556
David Vrabel59290362012-07-09 11:39:07 +0100557static inline bool desc_equal(const struct desc_struct *d1,
558 const struct desc_struct *d2)
559{
560 return d1->a == d2->a && d1->b == d2->b;
561}
562
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700563static void load_TLS_descriptor(struct thread_struct *t,
564 unsigned int cpu, unsigned int i)
565{
David Vrabel1c32cdc2012-07-09 11:39:08 +0100566 struct desc_struct *shadow = &per_cpu(shadow_tls_desc, cpu).desc[i];
567 struct desc_struct *gdt;
568 xmaddr_t maddr;
569 struct multicall_space mc;
570
571 if (desc_equal(shadow, &t->tls_array[i]))
572 return;
573
574 *shadow = t->tls_array[i];
575
576 gdt = get_cpu_gdt_table(cpu);
577 maddr = arbitrary_virt_to_machine(&gdt[GDT_ENTRY_TLS_MIN+i]);
578 mc = __xen_mc_entry(0);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700579
580 MULTI_update_descriptor(mc.mc, maddr.maddr, t->tls_array[i]);
581}
582
583static void xen_load_tls(struct thread_struct *t, unsigned int cpu)
584{
Jeremy Fitzhardinge8b84ad92007-07-17 18:37:06 -0700585 /*
Tejun Heoccbeed32009-02-09 22:17:40 +0900586 * XXX sleazy hack: If we're being called in a lazy-cpu zone
587 * and lazy gs handling is enabled, it means we're in a
588 * context switch, and %gs has just been saved. This means we
589 * can zero it out to prevent faults on exit from the
590 * hypervisor if the next process has no %gs. Either way, it
591 * has been saved, and the new value will get loaded properly.
592 * This will go away as soon as Xen has been modified to not
593 * save/restore %gs for normal hypercalls.
Eduardo Habkost8a954082008-07-08 15:07:10 -0700594 *
595 * On x86_64, this hack is not used for %gs, because gs points
596 * to KERNEL_GS_BASE (and uses it for PDA references), so we
597 * must not zero %gs on x86_64
598 *
599 * For x86_64, we need to zero %fs, otherwise we may get an
600 * exception between the new %fs descriptor being loaded and
601 * %fs being effectively cleared at __switch_to().
Jeremy Fitzhardinge8b84ad92007-07-17 18:37:06 -0700602 */
Eduardo Habkost8a954082008-07-08 15:07:10 -0700603 if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_CPU) {
604#ifdef CONFIG_X86_32
Tejun Heoccbeed32009-02-09 22:17:40 +0900605 lazy_load_gs(0);
Eduardo Habkost8a954082008-07-08 15:07:10 -0700606#else
607 loadsegment(fs, 0);
608#endif
609 }
610
611 xen_mc_batch();
612
613 load_TLS_descriptor(t, cpu, 0);
614 load_TLS_descriptor(t, cpu, 1);
615 load_TLS_descriptor(t, cpu, 2);
616
617 xen_mc_issue(PARAVIRT_LAZY_CPU);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700618}
619
Eduardo Habkosta8fc1082008-07-08 15:07:05 -0700620#ifdef CONFIG_X86_64
621static void xen_load_gs_index(unsigned int idx)
622{
623 if (HYPERVISOR_set_segment_base(SEGBASE_GS_USER_SEL, idx))
624 BUG();
625}
626#endif
627
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700628static void xen_write_ldt_entry(struct desc_struct *dt, int entrynum,
Glauber de Oliveira Costa75b8bb32008-01-30 13:31:13 +0100629 const void *ptr)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700630{
Jeremy Fitzhardingecef43bf2008-07-28 13:33:44 -0700631 xmaddr_t mach_lp = arbitrary_virt_to_machine(&dt[entrynum]);
Glauber de Oliveira Costa75b8bb32008-01-30 13:31:13 +0100632 u64 entry = *(u64 *)ptr;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700633
Jeremy Fitzhardingeab78f7a2010-12-17 15:54:28 -0800634 trace_xen_cpu_write_ldt_entry(dt, entrynum, entry);
635
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -0700636 preempt_disable();
637
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700638 xen_mc_flush();
639 if (HYPERVISOR_update_descriptor(mach_lp.maddr, entry))
640 BUG();
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -0700641
642 preempt_enable();
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700643}
644
Eduardo Habkoste176d362008-07-08 15:06:59 -0700645static int cvt_gate_to_trap(int vector, const gate_desc *val,
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700646 struct trap_info *info)
647{
Jeremy Fitzhardinge6cac5a92009-03-29 19:56:29 -0700648 unsigned long addr;
649
Jeremy Fitzhardinge6d02c422009-03-29 22:57:15 -0700650 if (val->type != GATE_TRAP && val->type != GATE_INTERRUPT)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700651 return 0;
652
653 info->vector = vector;
Jeremy Fitzhardinge6cac5a92009-03-29 19:56:29 -0700654
655 addr = gate_offset(*val);
656#ifdef CONFIG_X86_64
Jeremy Fitzhardingeb80119b2009-04-24 00:22:08 -0700657 /*
658 * Look for known traps using IST, and substitute them
659 * appropriately. The debugger ones are the only ones we care
Liu, Jinsong05e36002012-06-07 20:10:37 +0800660 * about. Xen will handle faults like double_fault,
661 * so we should never see them. Warn if
Jeremy Fitzhardingeb80119b2009-04-24 00:22:08 -0700662 * there's an unexpected IST-using fault handler.
663 */
Jeremy Fitzhardinge6cac5a92009-03-29 19:56:29 -0700664 if (addr == (unsigned long)debug)
665 addr = (unsigned long)xen_debug;
666 else if (addr == (unsigned long)int3)
667 addr = (unsigned long)xen_int3;
668 else if (addr == (unsigned long)stack_segment)
669 addr = (unsigned long)xen_stack_segment;
Jeremy Fitzhardingeb80119b2009-04-24 00:22:08 -0700670 else if (addr == (unsigned long)double_fault ||
671 addr == (unsigned long)nmi) {
672 /* Don't need to handle these */
673 return 0;
674#ifdef CONFIG_X86_MCE
675 } else if (addr == (unsigned long)machine_check) {
Liu, Jinsong05e36002012-06-07 20:10:37 +0800676 /*
677 * when xen hypervisor inject vMCE to guest,
678 * use native mce handler to handle it
679 */
680 ;
Jeremy Fitzhardingeb80119b2009-04-24 00:22:08 -0700681#endif
682 } else {
683 /* Some other trap using IST? */
684 if (WARN_ON(val->ist != 0))
685 return 0;
686 }
Jeremy Fitzhardinge6cac5a92009-03-29 19:56:29 -0700687#endif /* CONFIG_X86_64 */
688 info->address = addr;
689
Eduardo Habkoste176d362008-07-08 15:06:59 -0700690 info->cs = gate_segment(*val);
691 info->flags = val->dpl;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700692 /* interrupt gates clear IF */
Jeremy Fitzhardinge6d02c422009-03-29 22:57:15 -0700693 if (val->type == GATE_INTERRUPT)
694 info->flags |= 1 << 2;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700695
696 return 1;
697}
698
699/* Locations of each CPU's IDT */
Glauber de Oliveira Costa6b68f012008-01-30 13:31:12 +0100700static DEFINE_PER_CPU(struct desc_ptr, idt_desc);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700701
702/* Set an IDT entry. If the entry is part of the current IDT, then
703 also update Xen. */
Glauber de Oliveira Costa8d947342008-01-30 13:31:12 +0100704static void xen_write_idt_entry(gate_desc *dt, int entrynum, const gate_desc *g)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700705{
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700706 unsigned long p = (unsigned long)&dt[entrynum];
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -0700707 unsigned long start, end;
708
Jeremy Fitzhardingeab78f7a2010-12-17 15:54:28 -0800709 trace_xen_cpu_write_idt_entry(dt, entrynum, g);
710
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -0700711 preempt_disable();
712
Christoph Lameter780f36d2010-12-06 11:16:29 -0600713 start = __this_cpu_read(idt_desc.address);
714 end = start + __this_cpu_read(idt_desc.size) + 1;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700715
716 xen_mc_flush();
717
Glauber de Oliveira Costa8d947342008-01-30 13:31:12 +0100718 native_write_idt_entry(dt, entrynum, g);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700719
720 if (p >= start && (p + 8) <= end) {
721 struct trap_info info[2];
722
723 info[1].address = 0;
724
Eduardo Habkoste176d362008-07-08 15:06:59 -0700725 if (cvt_gate_to_trap(entrynum, g, &info[0]))
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700726 if (HYPERVISOR_set_trap_table(info))
727 BUG();
728 }
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -0700729
730 preempt_enable();
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700731}
732
Glauber de Oliveira Costa6b68f012008-01-30 13:31:12 +0100733static void xen_convert_trap_info(const struct desc_ptr *desc,
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700734 struct trap_info *traps)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700735{
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700736 unsigned in, out, count;
737
Eduardo Habkoste176d362008-07-08 15:06:59 -0700738 count = (desc->size+1) / sizeof(gate_desc);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700739 BUG_ON(count > 256);
740
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700741 for (in = out = 0; in < count; in++) {
Eduardo Habkoste176d362008-07-08 15:06:59 -0700742 gate_desc *entry = (gate_desc*)(desc->address) + in;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700743
Eduardo Habkoste176d362008-07-08 15:06:59 -0700744 if (cvt_gate_to_trap(in, entry, &traps[out]))
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700745 out++;
746 }
747 traps[out].address = 0;
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700748}
749
750void xen_copy_trap_info(struct trap_info *traps)
751{
Glauber de Oliveira Costa6b68f012008-01-30 13:31:12 +0100752 const struct desc_ptr *desc = &__get_cpu_var(idt_desc);
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700753
754 xen_convert_trap_info(desc, traps);
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700755}
756
757/* Load a new IDT into Xen. In principle this can be per-CPU, so we
758 hold a spinlock to protect the static traps[] array (static because
759 it avoids allocation, and saves stack space). */
Glauber de Oliveira Costa6b68f012008-01-30 13:31:12 +0100760static void xen_load_idt(const struct desc_ptr *desc)
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700761{
762 static DEFINE_SPINLOCK(lock);
763 static struct trap_info traps[257];
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700764
Jeremy Fitzhardingeab78f7a2010-12-17 15:54:28 -0800765 trace_xen_cpu_load_idt(desc);
766
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700767 spin_lock(&lock);
768
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -0700769 __get_cpu_var(idt_desc) = *desc;
770
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700771 xen_convert_trap_info(desc, traps);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700772
773 xen_mc_flush();
774 if (HYPERVISOR_set_trap_table(traps))
775 BUG();
776
777 spin_unlock(&lock);
778}
779
780/* Write a GDT descriptor entry. Ignore LDT descriptors, since
781 they're handled differently. */
782static void xen_write_gdt_entry(struct desc_struct *dt, int entry,
Glauber de Oliveira Costa014b15b2008-01-30 13:31:13 +0100783 const void *desc, int type)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700784{
Jeremy Fitzhardingeab78f7a2010-12-17 15:54:28 -0800785 trace_xen_cpu_write_gdt_entry(dt, entry, desc, type);
786
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -0700787 preempt_disable();
788
Glauber de Oliveira Costa014b15b2008-01-30 13:31:13 +0100789 switch (type) {
790 case DESC_LDT:
791 case DESC_TSS:
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700792 /* ignore */
793 break;
794
795 default: {
Jeremy Fitzhardinge9976b392009-02-27 09:19:26 -0800796 xmaddr_t maddr = arbitrary_virt_to_machine(&dt[entry]);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700797
798 xen_mc_flush();
Glauber de Oliveira Costa014b15b2008-01-30 13:31:13 +0100799 if (HYPERVISOR_update_descriptor(maddr.maddr, *(u64 *)desc))
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700800 BUG();
801 }
802
803 }
Jeremy Fitzhardingef120f132007-07-17 18:37:06 -0700804
805 preempt_enable();
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700806}
807
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -0700808/*
809 * Version of write_gdt_entry for use at early boot-time needed to
810 * update an entry as simply as possible.
811 */
Daniel Kiperad3062a2011-05-04 20:15:18 +0200812static void __init xen_write_gdt_entry_boot(struct desc_struct *dt, int entry,
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -0700813 const void *desc, int type)
814{
Jeremy Fitzhardingeab78f7a2010-12-17 15:54:28 -0800815 trace_xen_cpu_write_gdt_entry(dt, entry, desc, type);
816
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -0700817 switch (type) {
818 case DESC_LDT:
819 case DESC_TSS:
820 /* ignore */
821 break;
822
823 default: {
824 xmaddr_t maddr = virt_to_machine(&dt[entry]);
825
826 if (HYPERVISOR_update_descriptor(maddr.maddr, *(u64 *)desc))
827 dt[entry] = *(struct desc_struct *)desc;
828 }
829
830 }
831}
832
H. Peter Anvinfaca6222008-01-30 13:31:02 +0100833static void xen_load_sp0(struct tss_struct *tss,
Jeremy Fitzhardingea05d2eb2008-07-27 08:45:02 -0700834 struct thread_struct *thread)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700835{
Jeremy Fitzhardingeab78f7a2010-12-17 15:54:28 -0800836 struct multicall_space mcs;
837
838 mcs = xen_mc_entry(0);
H. Peter Anvinfaca6222008-01-30 13:31:02 +0100839 MULTI_stack_switch(mcs.mc, __KERNEL_DS, thread->sp0);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700840 xen_mc_issue(PARAVIRT_LAZY_CPU);
841}
842
843static void xen_set_iopl_mask(unsigned mask)
844{
845 struct physdev_set_iopl set_iopl;
846
847 /* Force the change at ring 0. */
848 set_iopl.iopl = (mask == 0) ? 1 : (mask >> 12) & 3;
849 HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
850}
851
852static void xen_io_delay(void)
853{
854}
855
856#ifdef CONFIG_X86_LOCAL_APIC
Konrad Rzeszutek Wilk558daa22012-05-02 15:04:51 -0400857static unsigned long xen_set_apic_id(unsigned int x)
858{
859 WARN_ON(1);
860 return x;
861}
862static unsigned int xen_get_apic_id(unsigned long x)
863{
864 return ((x)>>24) & 0xFFu;
865}
Suresh Siddhaad66dd32008-07-11 13:11:56 -0700866static u32 xen_apic_read(u32 reg)
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700867{
Konrad Rzeszutek Wilk558daa22012-05-02 15:04:51 -0400868 struct xen_platform_op op = {
869 .cmd = XENPF_get_cpuinfo,
870 .interface_version = XENPF_INTERFACE_VERSION,
871 .u.pcpu_info.xen_cpuid = 0,
872 };
873 int ret = 0;
874
875 /* Shouldn't need this as APIC is turned off for PV, and we only
876 * get called on the bootup processor. But just in case. */
877 if (!xen_initial_domain() || smp_processor_id())
878 return 0;
879
880 if (reg == APIC_LVR)
881 return 0x10;
882
883 if (reg != APIC_ID)
884 return 0;
885
886 ret = HYPERVISOR_dom0_op(&op);
887 if (ret)
888 return 0;
889
890 return op.u.pcpu_info.apic_id << 24;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700891}
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700892
Suresh Siddhaad66dd32008-07-11 13:11:56 -0700893static void xen_apic_write(u32 reg, u32 val)
Jeremy Fitzhardingef87e4ca2007-07-17 18:37:06 -0700894{
895 /* Warn to see if there's any stray references */
896 WARN_ON(1);
897}
Suresh Siddhaad66dd32008-07-11 13:11:56 -0700898
Suresh Siddhaad66dd32008-07-11 13:11:56 -0700899static u64 xen_apic_icr_read(void)
900{
901 return 0;
902}
903
904static void xen_apic_icr_write(u32 low, u32 id)
905{
906 /* Warn to see if there's any stray references */
907 WARN_ON(1);
908}
909
910static void xen_apic_wait_icr_idle(void)
911{
912 return;
913}
914
Yinghai Lu94a8c3c2008-07-13 22:19:35 -0700915static u32 xen_safe_apic_wait_icr_idle(void)
916{
917 return 0;
918}
919
Yinghai Luc1eeb2d2009-02-16 23:02:14 -0800920static void set_xen_basic_apic_ops(void)
921{
922 apic->read = xen_apic_read;
923 apic->write = xen_apic_write;
924 apic->icr_read = xen_apic_icr_read;
925 apic->icr_write = xen_apic_icr_write;
926 apic->wait_icr_idle = xen_apic_wait_icr_idle;
927 apic->safe_wait_icr_idle = xen_safe_apic_wait_icr_idle;
Konrad Rzeszutek Wilk558daa22012-05-02 15:04:51 -0400928 apic->set_apic_id = xen_set_apic_id;
929 apic->get_apic_id = xen_get_apic_id;
Ben Guthrof447d562012-04-21 00:11:04 +0800930
931#ifdef CONFIG_SMP
932 apic->send_IPI_allbutself = xen_send_IPI_allbutself;
933 apic->send_IPI_mask_allbutself = xen_send_IPI_mask_allbutself;
934 apic->send_IPI_mask = xen_send_IPI_mask;
935 apic->send_IPI_all = xen_send_IPI_all;
936 apic->send_IPI_self = xen_send_IPI_self;
937#endif
Yinghai Luc1eeb2d2009-02-16 23:02:14 -0800938}
Suresh Siddhaad66dd32008-07-11 13:11:56 -0700939
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700940#endif
941
Jeremy Fitzhardinge7b1333a2008-05-26 23:31:01 +0100942static void xen_clts(void)
943{
944 struct multicall_space mcs;
945
946 mcs = xen_mc_entry(0);
947
948 MULTI_fpu_taskswitch(mcs.mc, 0);
949
950 xen_mc_issue(PARAVIRT_LAZY_CPU);
951}
952
Jeremy Fitzhardingea789ed52009-04-24 00:26:50 -0700953static DEFINE_PER_CPU(unsigned long, xen_cr0_value);
954
955static unsigned long xen_read_cr0(void)
956{
Alex Shi2113f462012-01-13 23:53:35 +0800957 unsigned long cr0 = this_cpu_read(xen_cr0_value);
Jeremy Fitzhardingea789ed52009-04-24 00:26:50 -0700958
959 if (unlikely(cr0 == 0)) {
960 cr0 = native_read_cr0();
Alex Shi2113f462012-01-13 23:53:35 +0800961 this_cpu_write(xen_cr0_value, cr0);
Jeremy Fitzhardingea789ed52009-04-24 00:26:50 -0700962 }
963
964 return cr0;
965}
966
Jeremy Fitzhardinge7b1333a2008-05-26 23:31:01 +0100967static void xen_write_cr0(unsigned long cr0)
968{
969 struct multicall_space mcs;
970
Alex Shi2113f462012-01-13 23:53:35 +0800971 this_cpu_write(xen_cr0_value, cr0);
Jeremy Fitzhardingea789ed52009-04-24 00:26:50 -0700972
Jeremy Fitzhardinge7b1333a2008-05-26 23:31:01 +0100973 /* Only pay attention to cr0.TS; everything else is
974 ignored. */
975 mcs = xen_mc_entry(0);
976
977 MULTI_fpu_taskswitch(mcs.mc, (cr0 & X86_CR0_TS) != 0);
978
979 xen_mc_issue(PARAVIRT_LAZY_CPU);
980}
981
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700982static void xen_write_cr4(unsigned long cr4)
983{
Jeremy Fitzhardinge2956a352008-05-26 23:31:04 +0100984 cr4 &= ~X86_CR4_PGE;
985 cr4 &= ~X86_CR4_PSE;
986
987 native_write_cr4(cr4);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -0700988}
989
Jeremy Fitzhardinge11539682008-07-08 15:07:16 -0700990static int xen_write_msr_safe(unsigned int msr, unsigned low, unsigned high)
991{
992 int ret;
993
994 ret = 0;
995
Tejf63c2f22008-12-16 11:56:06 -0800996 switch (msr) {
Jeremy Fitzhardinge11539682008-07-08 15:07:16 -0700997#ifdef CONFIG_X86_64
998 unsigned which;
999 u64 base;
1000
1001 case MSR_FS_BASE: which = SEGBASE_FS; goto set;
1002 case MSR_KERNEL_GS_BASE: which = SEGBASE_GS_USER; goto set;
1003 case MSR_GS_BASE: which = SEGBASE_GS_KERNEL; goto set;
1004
1005 set:
1006 base = ((u64)high << 32) | low;
1007 if (HYPERVISOR_set_segment_base(which, base) != 0)
H. Peter Anvin0cc02132009-08-31 14:23:29 -07001008 ret = -EIO;
Jeremy Fitzhardinge11539682008-07-08 15:07:16 -07001009 break;
1010#endif
Jeremy Fitzhardinged89961e2008-07-24 13:48:58 -07001011
1012 case MSR_STAR:
1013 case MSR_CSTAR:
1014 case MSR_LSTAR:
1015 case MSR_SYSCALL_MASK:
1016 case MSR_IA32_SYSENTER_CS:
1017 case MSR_IA32_SYSENTER_ESP:
1018 case MSR_IA32_SYSENTER_EIP:
1019 /* Fast syscall setup is all done in hypercalls, so
1020 these are all ignored. Stub them out here to stop
1021 Xen console noise. */
1022 break;
1023
Jeremy Fitzhardinge41f2e472010-03-30 11:47:40 -07001024 case MSR_IA32_CR_PAT:
1025 if (smp_processor_id() == 0)
1026 xen_set_pat(((u64)high << 32) | low);
1027 break;
1028
Jeremy Fitzhardinge11539682008-07-08 15:07:16 -07001029 default:
1030 ret = native_write_msr_safe(msr, low, high);
1031 }
1032
1033 return ret;
1034}
1035
Jeremy Fitzhardinge0e913982008-05-26 23:31:27 +01001036void xen_setup_shared_info(void)
Jeremy Fitzhardinge2e8fe712008-03-17 16:36:53 -07001037{
1038 if (!xen_feature(XENFEAT_auto_translated_physmap)) {
Jeremy Fitzhardinge15664f92008-07-08 15:06:47 -07001039 set_fixmap(FIX_PARAVIRT_BOOTMAP,
1040 xen_start_info->shared_info);
Jeremy Fitzhardinge2e8fe712008-03-17 16:36:53 -07001041
Jeremy Fitzhardinge15664f92008-07-08 15:06:47 -07001042 HYPERVISOR_shared_info =
1043 (struct shared_info *)fix_to_virt(FIX_PARAVIRT_BOOTMAP);
Jeremy Fitzhardinge2e8fe712008-03-17 16:36:53 -07001044 } else
1045 HYPERVISOR_shared_info =
1046 (struct shared_info *)__va(xen_start_info->shared_info);
1047
1048#ifndef CONFIG_SMP
1049 /* In UP this is as good a place as any to set up shared info */
1050 xen_setup_vcpu_info_placement();
1051#endif
Jeremy Fitzhardinged5edbc12008-05-26 23:31:22 +01001052
1053 xen_setup_mfn_list_list();
Jeremy Fitzhardinge2e8fe712008-03-17 16:36:53 -07001054}
1055
Rusty Russell5f054e32012-03-29 15:38:31 +10301056/* This is called once we have the cpu_possible_mask */
Jeremy Fitzhardinge0e913982008-05-26 23:31:27 +01001057void xen_setup_vcpu_info_placement(void)
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -07001058{
1059 int cpu;
1060
1061 for_each_possible_cpu(cpu)
1062 xen_vcpu_setup(cpu);
1063
1064 /* xen_vcpu_setup managed to place the vcpu_info within the
1065 percpu area for all cpus, so make use of it */
1066 if (have_vcpu_info_placement) {
Jeremy Fitzhardingeecb93d12009-01-28 14:35:05 -08001067 pv_irq_ops.save_fl = __PV_IS_CALLEE_SAVE(xen_save_fl_direct);
1068 pv_irq_ops.restore_fl = __PV_IS_CALLEE_SAVE(xen_restore_fl_direct);
1069 pv_irq_ops.irq_disable = __PV_IS_CALLEE_SAVE(xen_irq_disable_direct);
1070 pv_irq_ops.irq_enable = __PV_IS_CALLEE_SAVE(xen_irq_enable_direct);
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001071 pv_mmu_ops.read_cr2 = xen_read_cr2_direct;
Jeremy Fitzhardinge60223a32007-07-17 18:37:07 -07001072 }
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001073}
1074
Andi Kleenab144f52007-08-10 22:31:03 +02001075static unsigned xen_patch(u8 type, u16 clobbers, void *insnbuf,
1076 unsigned long addr, unsigned len)
Jeremy Fitzhardinge64876732007-07-17 18:37:07 -07001077{
1078 char *start, *end, *reloc;
1079 unsigned ret;
1080
1081 start = end = reloc = NULL;
1082
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001083#define SITE(op, x) \
1084 case PARAVIRT_PATCH(op.x): \
Jeremy Fitzhardinge64876732007-07-17 18:37:07 -07001085 if (have_vcpu_info_placement) { \
1086 start = (char *)xen_##x##_direct; \
1087 end = xen_##x##_direct_end; \
1088 reloc = xen_##x##_direct_reloc; \
1089 } \
1090 goto patch_site
1091
1092 switch (type) {
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001093 SITE(pv_irq_ops, irq_enable);
1094 SITE(pv_irq_ops, irq_disable);
1095 SITE(pv_irq_ops, save_fl);
1096 SITE(pv_irq_ops, restore_fl);
Jeremy Fitzhardinge64876732007-07-17 18:37:07 -07001097#undef SITE
1098
1099 patch_site:
1100 if (start == NULL || (end-start) > len)
1101 goto default_patch;
1102
Andi Kleenab144f52007-08-10 22:31:03 +02001103 ret = paravirt_patch_insns(insnbuf, len, start, end);
Jeremy Fitzhardinge64876732007-07-17 18:37:07 -07001104
1105 /* Note: because reloc is assigned from something that
1106 appears to be an array, gcc assumes it's non-null,
1107 but doesn't know its relationship with start and
1108 end. */
1109 if (reloc > start && reloc < end) {
1110 int reloc_off = reloc - start;
Andi Kleenab144f52007-08-10 22:31:03 +02001111 long *relocp = (long *)(insnbuf + reloc_off);
1112 long delta = start - (char *)addr;
Jeremy Fitzhardinge64876732007-07-17 18:37:07 -07001113
1114 *relocp += delta;
1115 }
1116 break;
1117
1118 default_patch:
1119 default:
Andi Kleenab144f52007-08-10 22:31:03 +02001120 ret = paravirt_patch_default(type, clobbers, insnbuf,
1121 addr, len);
Jeremy Fitzhardinge64876732007-07-17 18:37:07 -07001122 break;
1123 }
1124
1125 return ret;
1126}
1127
Daniel Kiperad3062a2011-05-04 20:15:18 +02001128static const struct pv_info xen_info __initconst = {
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001129 .paravirt_enabled = 1,
1130 .shared_kernel_pmd = 0,
1131
Andy Lutomirski318f5a22011-08-03 09:31:53 -04001132#ifdef CONFIG_X86_64
1133 .extra_user_64bit_cs = FLAT_USER_CS64,
1134#endif
1135
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001136 .name = "Xen",
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001137};
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001138
Daniel Kiperad3062a2011-05-04 20:15:18 +02001139static const struct pv_init_ops xen_init_ops __initconst = {
Jeremy Fitzhardinge64876732007-07-17 18:37:07 -07001140 .patch = xen_patch,
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001141};
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001142
Daniel Kiperad3062a2011-05-04 20:15:18 +02001143static const struct pv_cpu_ops xen_cpu_ops __initconst = {
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001144 .cpuid = xen_cpuid,
1145
1146 .set_debugreg = xen_set_debugreg,
1147 .get_debugreg = xen_get_debugreg,
1148
Jeremy Fitzhardinge7b1333a2008-05-26 23:31:01 +01001149 .clts = xen_clts,
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001150
Jeremy Fitzhardingea789ed52009-04-24 00:26:50 -07001151 .read_cr0 = xen_read_cr0,
Jeremy Fitzhardinge7b1333a2008-05-26 23:31:01 +01001152 .write_cr0 = xen_write_cr0,
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001153
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001154 .read_cr4 = native_read_cr4,
1155 .read_cr4_safe = native_read_cr4_safe,
1156 .write_cr4 = xen_write_cr4,
1157
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001158 .wbinvd = native_wbinvd,
1159
1160 .read_msr = native_read_msr_safe,
Jeremy Fitzhardinge11539682008-07-08 15:07:16 -07001161 .write_msr = xen_write_msr_safe,
Konrad Rzeszutek Wilk1ab46fd2012-05-30 18:23:56 -04001162
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001163 .read_tsc = native_read_tsc,
1164 .read_pmc = native_read_pmc,
1165
Jeremy Fitzhardinge81e103f2008-04-17 17:40:51 +02001166 .iret = xen_iret,
Jeremy Fitzhardinged75cd222008-06-25 00:19:26 -04001167 .irq_enable_sysexit = xen_sysexit,
Jeremy Fitzhardinge6fcac6d2008-07-08 15:07:14 -07001168#ifdef CONFIG_X86_64
1169 .usergs_sysret32 = xen_sysret32,
1170 .usergs_sysret64 = xen_sysret64,
1171#endif
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001172
1173 .load_tr_desc = paravirt_nop,
1174 .set_ldt = xen_set_ldt,
1175 .load_gdt = xen_load_gdt,
1176 .load_idt = xen_load_idt,
1177 .load_tls = xen_load_tls,
Eduardo Habkosta8fc1082008-07-08 15:07:05 -07001178#ifdef CONFIG_X86_64
1179 .load_gs_index = xen_load_gs_index,
1180#endif
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001181
Jeremy Fitzhardinge38ffbe62008-07-23 14:21:18 -07001182 .alloc_ldt = xen_alloc_ldt,
1183 .free_ldt = xen_free_ldt,
1184
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001185 .store_gdt = native_store_gdt,
1186 .store_idt = native_store_idt,
1187 .store_tr = xen_store_tr,
1188
1189 .write_ldt_entry = xen_write_ldt_entry,
1190 .write_gdt_entry = xen_write_gdt_entry,
1191 .write_idt_entry = xen_write_idt_entry,
H. Peter Anvinfaca6222008-01-30 13:31:02 +01001192 .load_sp0 = xen_load_sp0,
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001193
1194 .set_iopl_mask = xen_set_iopl_mask,
1195 .io_delay = xen_io_delay,
1196
Jeremy Fitzhardinge952d1d72008-07-08 15:07:01 -07001197 /* Xen takes care of %gs when switching to usermode for us */
1198 .swapgs = paravirt_nop,
1199
Jeremy Fitzhardinge224101e2009-02-18 11:18:57 -08001200 .start_context_switch = paravirt_start_context_switch,
1201 .end_context_switch = xen_end_context_switch,
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001202};
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001203
Daniel Kiperad3062a2011-05-04 20:15:18 +02001204static const struct pv_apic_ops xen_apic_ops __initconst = {
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001205#ifdef CONFIG_X86_LOCAL_APIC
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001206 .startup_ipi_hook = paravirt_nop,
1207#endif
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001208};
1209
Jeremy Fitzhardingefefa6292007-07-17 18:37:07 -07001210static void xen_reboot(int reason)
1211{
Jeremy Fitzhardinge349c7092008-05-26 23:31:02 +01001212 struct sched_shutdown r = { .reason = reason };
1213
Jeremy Fitzhardinge349c7092008-05-26 23:31:02 +01001214 if (HYPERVISOR_sched_op(SCHEDOP_shutdown, &r))
Jeremy Fitzhardingefefa6292007-07-17 18:37:07 -07001215 BUG();
1216}
1217
1218static void xen_restart(char *msg)
1219{
1220 xen_reboot(SHUTDOWN_reboot);
1221}
1222
1223static void xen_emergency_restart(void)
1224{
1225 xen_reboot(SHUTDOWN_reboot);
1226}
1227
1228static void xen_machine_halt(void)
1229{
1230 xen_reboot(SHUTDOWN_poweroff);
1231}
1232
Tom Goetzb2abe502011-05-16 15:06:26 -04001233static void xen_machine_power_off(void)
1234{
1235 if (pm_power_off)
1236 pm_power_off();
1237 xen_reboot(SHUTDOWN_poweroff);
1238}
1239
Jeremy Fitzhardingefefa6292007-07-17 18:37:07 -07001240static void xen_crash_shutdown(struct pt_regs *regs)
1241{
1242 xen_reboot(SHUTDOWN_crash);
1243}
1244
Donald Dutilef09f6d12010-07-15 14:56:49 -04001245static int
1246xen_panic_event(struct notifier_block *this, unsigned long event, void *ptr)
1247{
Ian Campbell086748e2010-08-03 14:55:14 -07001248 xen_reboot(SHUTDOWN_crash);
Donald Dutilef09f6d12010-07-15 14:56:49 -04001249 return NOTIFY_DONE;
1250}
1251
1252static struct notifier_block xen_panic_block = {
1253 .notifier_call= xen_panic_event,
1254};
1255
1256int xen_panic_handler_init(void)
1257{
1258 atomic_notifier_chain_register(&panic_notifier_list, &xen_panic_block);
1259 return 0;
1260}
1261
Daniel Kiperad3062a2011-05-04 20:15:18 +02001262static const struct machine_ops xen_machine_ops __initconst = {
Jeremy Fitzhardingefefa6292007-07-17 18:37:07 -07001263 .restart = xen_restart,
1264 .halt = xen_machine_halt,
Tom Goetzb2abe502011-05-16 15:06:26 -04001265 .power_off = xen_machine_power_off,
Jeremy Fitzhardingefefa6292007-07-17 18:37:07 -07001266 .shutdown = xen_machine_halt,
1267 .crash_shutdown = xen_crash_shutdown,
1268 .emergency_restart = xen_emergency_restart,
1269};
1270
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -07001271/*
1272 * Set up the GDT and segment registers for -fstack-protector. Until
1273 * we do this, we have to be careful not to call any stack-protected
1274 * function, which is most of the kernel.
1275 */
1276static void __init xen_setup_stackprotector(void)
1277{
1278 pv_cpu_ops.write_gdt_entry = xen_write_gdt_entry_boot;
1279 pv_cpu_ops.load_gdt = xen_load_gdt_boot;
1280
1281 setup_stack_canary_segment(0);
1282 switch_to_new_gdt(0);
1283
1284 pv_cpu_ops.write_gdt_entry = xen_write_gdt_entry;
1285 pv_cpu_ops.load_gdt = xen_load_gdt;
1286}
1287
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001288/* First C function to be called on Xen boot */
1289asmlinkage void __init xen_start_kernel(void)
1290{
Konrad Rzeszutek Wilkec35a692010-11-16 12:09:59 -05001291 struct physdev_set_iopl set_iopl;
1292 int rc;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001293
1294 if (!xen_start_info)
1295 return;
1296
Jeremy Fitzhardinge6e833582008-08-19 13:16:17 -07001297 xen_domain_type = XEN_PV_DOMAIN;
1298
Ian Campbell7e775062010-09-30 12:37:26 +01001299 xen_setup_machphys_mapping();
1300
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001301 /* Install Xen paravirt ops */
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001302 pv_info = xen_info;
1303 pv_init_ops = xen_init_ops;
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001304 pv_cpu_ops = xen_cpu_ops;
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001305 pv_apic_ops = xen_apic_ops;
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001306
Thomas Gleixner6b18ae32009-08-20 10:19:54 +02001307 x86_init.resources.memory_setup = xen_memory_setup;
Thomas Gleixner42bbdb42009-08-20 13:04:10 +02001308 x86_init.oem.arch_setup = xen_arch_setup;
Thomas Gleixner6f30c1a2009-08-20 13:19:57 +02001309 x86_init.oem.banner = xen_banner;
Thomas Gleixner845b3942009-08-19 15:37:03 +02001310
Stefano Stabellini409771d2010-05-14 12:48:19 +01001311 xen_init_time_ops();
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001312
Jeremy Fitzhardingece2eef32009-08-17 12:26:53 -07001313 /*
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -07001314 * Set up some pagetable state before starting to set any ptes.
Jeremy Fitzhardingece2eef32009-08-17 12:26:53 -07001315 */
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -07001316
Jeremy Fitzhardinge973df352009-10-27 16:54:19 -07001317 xen_init_mmu_ops();
1318
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -07001319 /* Prevent unwanted bits from being set in PTEs. */
1320 __supported_pte_mask &= ~_PAGE_GLOBAL;
Konrad Rzeszutek Wilk8eaffa62012-02-10 09:16:27 -05001321#if 0
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -07001322 if (!xen_initial_domain())
Konrad Rzeszutek Wilk8eaffa62012-02-10 09:16:27 -05001323#endif
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -07001324 __supported_pte_mask &= ~(_PAGE_PWT | _PAGE_PCD);
1325
1326 __supported_pte_mask |= _PAGE_IOMAP;
1327
Ian Campbell817a8242010-02-26 17:16:00 +00001328 /*
1329 * Prevent page tables from being allocated in highmem, even
1330 * if CONFIG_HIGHPTE is enabled.
1331 */
1332 __userpte_alloc_gfp &= ~__GFP_HIGHMEM;
1333
Jeremy Fitzhardingeb75fe4e2009-09-21 13:34:06 -07001334 /* Work out if we support NX */
H. Peter Anvin4763ed42009-11-13 15:28:16 -08001335 x86_configure_nx();
Jeremy Fitzhardingeb75fe4e2009-09-21 13:34:06 -07001336
Jeremy Fitzhardinge577eebe2009-08-27 12:46:35 -07001337 xen_setup_features();
1338
1339 /* Get mfn list */
1340 if (!xen_feature(XENFEAT_auto_translated_physmap))
1341 xen_build_dynamic_phys_to_machine();
1342
1343 /*
1344 * Set up kernel GDT and segment registers, mainly so that
1345 * -fstack-protector code can be executed.
1346 */
1347 xen_setup_stackprotector();
Jeremy Fitzhardinge0d1edf42008-07-28 11:53:57 -07001348
Jeremy Fitzhardingece2eef32009-08-17 12:26:53 -07001349 xen_init_irq_ops();
Jeremy Fitzhardingee826fe12009-03-07 17:09:27 -08001350 xen_init_cpuid_mask();
1351
Yinghai Lu94a8c3c2008-07-13 22:19:35 -07001352#ifdef CONFIG_X86_LOCAL_APIC
Suresh Siddhaad66dd32008-07-11 13:11:56 -07001353 /*
Yinghai Lu94a8c3c2008-07-13 22:19:35 -07001354 * set up the basic apic ops.
Suresh Siddhaad66dd32008-07-11 13:11:56 -07001355 */
Yinghai Luc1eeb2d2009-02-16 23:02:14 -08001356 set_xen_basic_apic_ops();
Suresh Siddhaad66dd32008-07-11 13:11:56 -07001357#endif
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001358
Jeremy Fitzhardingee57778a2008-06-16 04:30:02 -07001359 if (xen_feature(XENFEAT_mmu_pt_update_preserve_ad)) {
1360 pv_mmu_ops.ptep_modify_prot_start = xen_ptep_modify_prot_start;
1361 pv_mmu_ops.ptep_modify_prot_commit = xen_ptep_modify_prot_commit;
1362 }
1363
Jeremy Fitzhardingefefa6292007-07-17 18:37:07 -07001364 machine_ops = xen_machine_ops;
1365
Jeremy Fitzhardinge38341432009-02-02 13:55:31 -08001366 /*
1367 * The only reliable way to retain the initial address of the
1368 * percpu gdt_page is to remember it here, so we can go and
1369 * mark it RW later, when the initial percpu area is freed.
1370 */
1371 xen_initial_gdt = &per_cpu(gdt_page, 0);
Jeremy Fitzhardinge795f99b2009-01-30 17:47:54 +09001372
Jeremy Fitzhardingea9e70622008-07-08 15:06:41 -07001373 xen_smp_init();
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001374
Ian Campbellc1f5db12010-12-03 09:54:03 +00001375#ifdef CONFIG_ACPI_NUMA
1376 /*
1377 * The pages we from Xen are not related to machine pages, so
1378 * any NUMA information the kernel tries to get from ACPI will
1379 * be meaningless. Prevent it from trying.
1380 */
1381 acpi_numa = -1;
1382#endif
1383
Jeremy Fitzhardinge084a2a42008-07-08 15:06:50 -07001384 /* Don't do the full vcpu_info placement stuff until we have a
1385 possible map and a non-dummy shared_info. */
1386 per_cpu(xen_vcpu, 0) = &HYPERVISOR_shared_info->vcpu_info[0];
1387
Jeremy Fitzhardinge55d80852009-02-25 09:42:25 -08001388 local_irq_disable();
Tejun Heo2ce802f2011-01-20 12:06:35 +01001389 early_boot_irqs_disabled = true;
Jeremy Fitzhardinge55d80852009-02-25 09:42:25 -08001390
Jeremy Fitzhardinge084a2a42008-07-08 15:06:50 -07001391 xen_raw_console_write("mapping kernel into physical memory\n");
Konrad Rzeszutek Wilk3699aad2012-06-28 22:47:35 -04001392 xen_setup_kernel_pagetable((pgd_t *)xen_start_info->pt_base, xen_start_info->nr_pages);
Jeremy Fitzhardinge084a2a42008-07-08 15:06:50 -07001393
Jeremy Fitzhardinge33a84752010-08-27 15:18:19 -07001394 /* Allocate and initialize top and mid mfn levels for p2m structure */
1395 xen_build_mfn_list_list();
1396
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001397 /* keep using Xen gdt for now; no urgent need to change it */
1398
Ian Campbelle68266b2010-01-13 10:16:08 +00001399#ifdef CONFIG_X86_32
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001400 pv_info.kernel_rpl = 1;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001401 if (xen_feature(XENFEAT_supervisor_mode_kernel))
Jeremy Fitzhardinge93b1eab2007-10-16 11:51:29 -07001402 pv_info.kernel_rpl = 0;
Ian Campbelle68266b2010-01-13 10:16:08 +00001403#else
1404 pv_info.kernel_rpl = 0;
1405#endif
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001406 /* set the limit of our address space */
Jeremy Fitzhardingefb1d8402007-10-16 11:51:31 -07001407 xen_reserve_top();
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001408
Konrad Rzeszutek Wilkec35a692010-11-16 12:09:59 -05001409 /* We used to do this in xen_arch_setup, but that is too late on AMD
1410 * were early_cpu_init (run before ->arch_setup()) calls early_amd_init
1411 * which pokes 0xcf8 port.
1412 */
1413 set_iopl.iopl = 1;
1414 rc = HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
1415 if (rc != 0)
1416 xen_raw_printk("physdev_op failed %d\n", rc);
1417
Jeremy Fitzhardinge7d087b62008-07-08 15:06:48 -07001418#ifdef CONFIG_X86_32
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001419 /* set up basic CPUID stuff */
1420 cpu_detect(&new_cpu_data);
1421 new_cpu_data.hard_math = 1;
Jeremy Fitzhardinged560bc62009-08-25 12:53:02 -07001422 new_cpu_data.wp_works_ok = 1;
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001423 new_cpu_data.x86_capability[0] = cpuid_edx(1);
Jeremy Fitzhardinge7d087b62008-07-08 15:06:48 -07001424#endif
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001425
1426 /* Poke various useful things into boot_params */
H. Peter Anvin30c82642007-10-15 17:13:22 -07001427 boot_params.hdr.type_of_loader = (9 << 4) | 0;
1428 boot_params.hdr.ramdisk_image = xen_start_info->mod_start
1429 ? __pa(xen_start_info->mod_start) : 0;
1430 boot_params.hdr.ramdisk_size = xen_start_info->mod_len;
Jeremy Fitzhardingeb7c3c5c2008-07-08 15:07:07 -07001431 boot_params.hdr.cmd_line_ptr = __pa(xen_start_info->cmd_line);
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001432
Jeremy Fitzhardinge6e833582008-08-19 13:16:17 -07001433 if (!xen_initial_domain()) {
Jeremy Fitzhardinge83abc702008-05-26 23:31:12 +01001434 add_preferred_console("xenboot", 0, NULL);
Markus Armbruster9e124fe2008-05-26 23:31:07 +01001435 add_preferred_console("tty", 0, NULL);
Markus Armbrusterb8c2d3d2008-02-27 14:56:35 +01001436 add_preferred_console("hvc", 0, NULL);
Alex Nixonb5401a92010-03-18 16:31:34 -04001437 if (pci_xen)
1438 x86_init.pci.arch_init = pci_xen_init;
Chris Wright5d990b62009-12-04 12:15:21 -08001439 } else {
Jeremy Fitzhardingec2419b42011-05-31 10:50:10 -04001440 const struct dom0_vga_console_info *info =
1441 (void *)((char *)xen_start_info +
1442 xen_start_info->console.dom0.info_off);
Konrad Rzeszutek Wilkffb8b232012-09-21 12:30:35 -04001443 struct xen_platform_op op = {
1444 .cmd = XENPF_firmware_info,
1445 .interface_version = XENPF_INTERFACE_VERSION,
1446 .u.firmware_info.type = XEN_FW_KBD_SHIFT_FLAGS,
1447 };
Jeremy Fitzhardingec2419b42011-05-31 10:50:10 -04001448
1449 xen_init_vga(info, xen_start_info->console.dom0.info_size);
1450 xen_start_info->console.domU.mfn = 0;
1451 xen_start_info->console.domU.evtchn = 0;
1452
Konrad Rzeszutek Wilkffb8b232012-09-21 12:30:35 -04001453 if (HYPERVISOR_dom0_op(&op) == 0)
1454 boot_params.kbd_status = op.u.firmware_info.u.kbd_shift_flags;
1455
Konrad Rzeszutek Wilk31b3c9d2012-03-20 18:53:10 -04001456 xen_init_apic();
1457
Chris Wright5d990b62009-12-04 12:15:21 -08001458 /* Make sure ACS will be enabled */
1459 pci_request_acs();
Konrad Rzeszutek Wilk211063d2011-12-08 17:32:23 +08001460
1461 xen_acpi_sleep_register();
Markus Armbruster9e124fe2008-05-26 23:31:07 +01001462 }
David Vrabel76a8df72012-05-04 14:29:46 +01001463#ifdef CONFIG_PCI
1464 /* PCI BIOS service won't work from a PV guest. */
1465 pci_probe &= ~PCI_PROBE_BIOS;
1466#endif
Jeremy Fitzhardinge084a2a42008-07-08 15:06:50 -07001467 xen_raw_console_write("about to get started...\n");
1468
Jeremy Fitzhardinge499d19b2009-11-24 09:38:25 -08001469 xen_setup_runstate_info(0);
1470
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001471 /* Start the world */
Jeremy Fitzhardingef5d36de2008-07-08 15:06:39 -07001472#ifdef CONFIG_X86_32
Yinghai Luf0d43102008-05-29 12:56:36 -07001473 i386_start_kernel();
Jeremy Fitzhardingef5d36de2008-07-08 15:06:39 -07001474#else
Jeremy Fitzhardinge084a2a42008-07-08 15:06:50 -07001475 x86_64_start_reservations((char *)__pa_symbol(&boot_params));
Jeremy Fitzhardingef5d36de2008-07-08 15:06:39 -07001476#endif
Jeremy Fitzhardinge5ead97c2007-07-17 18:37:04 -07001477}
Sheng Yangbee6ab52010-05-14 12:39:33 +01001478
Ian Campbell44b46c32011-02-11 16:37:41 +00001479void __ref xen_hvm_init_shared_info(void)
Sheng Yangbee6ab52010-05-14 12:39:33 +01001480{
Stefano Stabellini016b6f52010-05-14 12:45:07 +01001481 int cpu;
Sheng Yangbee6ab52010-05-14 12:39:33 +01001482 struct xen_add_to_physmap xatp;
Stefano Stabellini016b6f52010-05-14 12:45:07 +01001483 static struct shared_info *shared_info_page = 0;
Sheng Yangbee6ab52010-05-14 12:39:33 +01001484
Stefano Stabellini016b6f52010-05-14 12:45:07 +01001485 if (!shared_info_page)
1486 shared_info_page = (struct shared_info *)
1487 extend_brk(PAGE_SIZE, PAGE_SIZE);
Sheng Yangbee6ab52010-05-14 12:39:33 +01001488 xatp.domid = DOMID_SELF;
1489 xatp.idx = 0;
1490 xatp.space = XENMAPSPACE_shared_info;
1491 xatp.gpfn = __pa(shared_info_page) >> PAGE_SHIFT;
1492 if (HYPERVISOR_memory_op(XENMEM_add_to_physmap, &xatp))
1493 BUG();
1494
1495 HYPERVISOR_shared_info = (struct shared_info *)shared_info_page;
1496
Stefano Stabellini016b6f52010-05-14 12:45:07 +01001497 /* xen_vcpu is a pointer to the vcpu_info struct in the shared_info
1498 * page, we use it in the event channel upcall and in some pvclock
1499 * related functions. We don't need the vcpu_info placement
1500 * optimizations because we don't use any pv_mmu or pv_irq op on
1501 * HVM.
1502 * When xen_hvm_init_shared_info is run at boot time only vcpu 0 is
1503 * online but xen_hvm_init_shared_info is run at resume time too and
1504 * in that case multiple vcpus might be online. */
1505 for_each_online_cpu(cpu) {
1506 per_cpu(xen_vcpu, cpu) = &HYPERVISOR_shared_info->vcpu_info[cpu];
1507 }
Sheng Yangbee6ab52010-05-14 12:39:33 +01001508}
1509
Stefano Stabellinica65f9f2010-07-29 14:37:48 +01001510#ifdef CONFIG_XEN_PVHVM
Olaf Hering4ff2d062012-07-17 11:59:15 +02001511static void __init init_hvm_pv_info(void)
1512{
1513 int major, minor;
Sheng Yangbee6ab52010-05-14 12:39:33 +01001514 uint32_t eax, ebx, ecx, edx, pages, msr, base;
1515 u64 pfn;
1516
1517 base = xen_cpuid_base();
1518 cpuid(base + 1, &eax, &ebx, &ecx, &edx);
1519
Olaf Hering4ff2d062012-07-17 11:59:15 +02001520 major = eax >> 16;
1521 minor = eax & 0xffff;
1522 printk(KERN_INFO "Xen version %d.%d.\n", major, minor);
Sheng Yangbee6ab52010-05-14 12:39:33 +01001523
1524 cpuid(base + 2, &pages, &msr, &ecx, &edx);
1525
1526 pfn = __pa(hypercall_page);
1527 wrmsr_safe(msr, (u32)pfn, (u32)(pfn >> 32));
1528
1529 xen_setup_features();
1530
1531 pv_info.name = "Xen HVM";
1532
1533 xen_domain_type = XEN_HVM_DOMAIN;
Sheng Yangbee6ab52010-05-14 12:39:33 +01001534}
1535
Sheng Yang38e20b02010-05-14 12:40:51 +01001536static int __cpuinit xen_hvm_cpu_notify(struct notifier_block *self,
1537 unsigned long action, void *hcpu)
1538{
1539 int cpu = (long)hcpu;
1540 switch (action) {
1541 case CPU_UP_PREPARE:
Zhenzhong Duan90d4f552011-10-27 22:28:59 -07001542 xen_vcpu_setup(cpu);
Stefano Stabellini99bbb3a2010-12-02 17:55:10 +00001543 if (xen_have_vector_callback)
1544 xen_init_lock_cpu(cpu);
Sheng Yang38e20b02010-05-14 12:40:51 +01001545 break;
1546 default:
1547 break;
1548 }
1549 return NOTIFY_OK;
1550}
1551
Daniel Kiperad3062a2011-05-04 20:15:18 +02001552static struct notifier_block xen_hvm_cpu_notifier __cpuinitdata = {
Sheng Yang38e20b02010-05-14 12:40:51 +01001553 .notifier_call = xen_hvm_cpu_notify,
1554};
1555
Sheng Yangbee6ab52010-05-14 12:39:33 +01001556static void __init xen_hvm_guest_init(void)
1557{
Olaf Hering4ff2d062012-07-17 11:59:15 +02001558 init_hvm_pv_info();
Sheng Yangbee6ab52010-05-14 12:39:33 +01001559
Stefano Stabellini016b6f52010-05-14 12:45:07 +01001560 xen_hvm_init_shared_info();
Sheng Yang38e20b02010-05-14 12:40:51 +01001561
1562 if (xen_feature(XENFEAT_hvm_callback_vector))
1563 xen_have_vector_callback = 1;
Stefano Stabellini99bbb3a2010-12-02 17:55:10 +00001564 xen_hvm_smp_init();
Sheng Yang38e20b02010-05-14 12:40:51 +01001565 register_cpu_notifier(&xen_hvm_cpu_notifier);
Stefano Stabellinic1c54132010-05-14 12:44:30 +01001566 xen_unplug_emulated_devices();
Sheng Yang38e20b02010-05-14 12:40:51 +01001567 x86_init.irqs.intr_init = xen_init_IRQ;
Stefano Stabellini409771d2010-05-14 12:48:19 +01001568 xen_hvm_init_time_ops();
Stefano Stabellini59151002010-06-17 14:22:52 +01001569 xen_hvm_init_mmu_ops();
Sheng Yangbee6ab52010-05-14 12:39:33 +01001570}
1571
1572static bool __init xen_hvm_platform(void)
1573{
1574 if (xen_pv_domain())
1575 return false;
1576
1577 if (!xen_cpuid_base())
1578 return false;
1579
1580 return true;
1581}
1582
Sheng Yangd9b8ca82010-12-21 14:18:49 +08001583bool xen_hvm_need_lapic(void)
1584{
1585 if (xen_pv_domain())
1586 return false;
1587 if (!xen_hvm_domain())
1588 return false;
1589 if (xen_feature(XENFEAT_hvm_pirqs) && xen_have_vector_callback)
1590 return false;
1591 return true;
1592}
1593EXPORT_SYMBOL_GPL(xen_hvm_need_lapic);
1594
Daniel Kiperad3062a2011-05-04 20:15:18 +02001595const struct hypervisor_x86 x86_hyper_xen_hvm __refconst = {
Sheng Yangbee6ab52010-05-14 12:39:33 +01001596 .name = "Xen HVM",
1597 .detect = xen_hvm_platform,
1598 .init_platform = xen_hvm_guest_init,
1599};
1600EXPORT_SYMBOL(x86_hyper_xen_hvm);
Stefano Stabellinica65f9f2010-07-29 14:37:48 +01001601#endif