blob: 1248f9ceabc168caf620be8aa06935f7d137a054 [file] [log] [blame]
Paul Gortmaker69c60c82011-05-26 12:22:53 -04001#include <linux/export.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07002#include <linux/init.h>
3#include <linux/bitops.h>
Stephen Rothwell5cdd1742011-08-10 11:49:56 +10004#include <linux/elf.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07005#include <linux/mm.h>
Yinghai Lu8d71a2e2008-09-07 17:58:53 -07006
Alan Cox8bdbd962009-07-04 00:35:45 +01007#include <linux/io.h>
Borislav Petkovc98fdea2012-02-07 13:08:52 +01008#include <linux/sched.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07009#include <asm/processor.h>
Andi Kleend3f7eae2007-08-10 22:31:07 +020010#include <asm/apic.h>
Yinghai Lu1f442d72009-03-07 23:46:26 -080011#include <asm/cpu.h>
Andreas Herrmann42937e82009-06-08 15:55:09 +020012#include <asm/pci-direct.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070013
Yinghai Lu8d71a2e2008-09-07 17:58:53 -070014#ifdef CONFIG_X86_64
15# include <asm/numa_64.h>
16# include <asm/mmconfig.h>
17# include <asm/cacheflush.h>
18#endif
19
Linus Torvalds1da177e2005-04-16 15:20:36 -070020#include "cpu.h"
21
Yinghai Lu6c62aa42008-09-07 17:58:54 -070022#ifdef CONFIG_X86_32
Linus Torvalds1da177e2005-04-16 15:20:36 -070023/*
24 * B step AMD K6 before B 9730xxxx have hardware bugs that can cause
25 * misexecution of code under Linux. Owners of such processors should
26 * contact AMD for precise details and a CPU swap.
27 *
28 * See http://www.multimania.com/poulot/k6bug.html
Andreas Herrmannd7de8642012-04-11 17:12:38 +020029 * and section 2.6.2 of "AMD-K6 Processor Revision Guide - Model 6"
30 * (Publication # 21266 Issue Date: August 1998)
Linus Torvalds1da177e2005-04-16 15:20:36 -070031 *
32 * The following test is erm.. interesting. AMD neglected to up
33 * the chip setting when fixing the bug but they also tweaked some
34 * performance at the same time..
35 */
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +010036
Linus Torvalds1da177e2005-04-16 15:20:36 -070037extern void vide(void);
38__asm__(".align 4\nvide: ret");
39
Yinghai Lu11fdd252008-09-07 17:58:50 -070040static void __cpuinit init_amd_k5(struct cpuinfo_x86 *c)
41{
42/*
43 * General Systems BIOSen alias the cpu frequency registers
44 * of the Elan at 0x000df000. Unfortuantly, one of the Linux
45 * drivers subsequently pokes it, and changes the CPU speed.
46 * Workaround : Remove the unneeded alias.
47 */
48#define CBAR (0xfffc) /* Configuration Base Address (32-bit) */
49#define CBAR_ENB (0x80000000)
50#define CBAR_KEY (0X000000CB)
51 if (c->x86_model == 9 || c->x86_model == 10) {
Alan Cox8bdbd962009-07-04 00:35:45 +010052 if (inl(CBAR) & CBAR_ENB)
53 outl(0 | CBAR_KEY, CBAR);
Yinghai Lu11fdd252008-09-07 17:58:50 -070054 }
55}
56
57
58static void __cpuinit init_amd_k6(struct cpuinfo_x86 *c)
59{
60 u32 l, h;
61 int mbytes = num_physpages >> (20-PAGE_SHIFT);
62
63 if (c->x86_model < 6) {
64 /* Based on AMD doc 20734R - June 2000 */
65 if (c->x86_model == 0) {
66 clear_cpu_cap(c, X86_FEATURE_APIC);
67 set_cpu_cap(c, X86_FEATURE_PGE);
68 }
69 return;
70 }
71
72 if (c->x86_model == 6 && c->x86_mask == 1) {
73 const int K6_BUG_LOOP = 1000000;
74 int n;
75 void (*f_vide)(void);
76 unsigned long d, d2;
77
78 printk(KERN_INFO "AMD K6 stepping B detected - ");
79
80 /*
81 * It looks like AMD fixed the 2.6.2 bug and improved indirect
82 * calls at the same time.
83 */
84
85 n = K6_BUG_LOOP;
86 f_vide = vide;
87 rdtscl(d);
88 while (n--)
89 f_vide();
90 rdtscl(d2);
91 d = d2-d;
92
93 if (d > 20*K6_BUG_LOOP)
Alan Cox8bdbd962009-07-04 00:35:45 +010094 printk(KERN_CONT
95 "system stability may be impaired when more than 32 MB are used.\n");
Yinghai Lu11fdd252008-09-07 17:58:50 -070096 else
Alan Cox8bdbd962009-07-04 00:35:45 +010097 printk(KERN_CONT "probably OK (after B9730xxxx).\n");
Yinghai Lu11fdd252008-09-07 17:58:50 -070098 }
99
100 /* K6 with old style WHCR */
101 if (c->x86_model < 8 ||
102 (c->x86_model == 8 && c->x86_mask < 8)) {
103 /* We can only write allocate on the low 508Mb */
104 if (mbytes > 508)
105 mbytes = 508;
106
107 rdmsr(MSR_K6_WHCR, l, h);
108 if ((l&0x0000FFFF) == 0) {
109 unsigned long flags;
110 l = (1<<0)|((mbytes/4)<<1);
111 local_irq_save(flags);
112 wbinvd();
113 wrmsr(MSR_K6_WHCR, l, h);
114 local_irq_restore(flags);
115 printk(KERN_INFO "Enabling old style K6 write allocation for %d Mb\n",
116 mbytes);
117 }
118 return;
119 }
120
121 if ((c->x86_model == 8 && c->x86_mask > 7) ||
122 c->x86_model == 9 || c->x86_model == 13) {
123 /* The more serious chips .. */
124
125 if (mbytes > 4092)
126 mbytes = 4092;
127
128 rdmsr(MSR_K6_WHCR, l, h);
129 if ((l&0xFFFF0000) == 0) {
130 unsigned long flags;
131 l = ((mbytes>>2)<<22)|(1<<16);
132 local_irq_save(flags);
133 wbinvd();
134 wrmsr(MSR_K6_WHCR, l, h);
135 local_irq_restore(flags);
136 printk(KERN_INFO "Enabling new style K6 write allocation for %d Mb\n",
137 mbytes);
138 }
139
140 return;
141 }
142
143 if (c->x86_model == 10) {
144 /* AMD Geode LX is model 10 */
145 /* placeholder for any needed mods */
146 return;
147 }
148}
149
Yinghai Lu1f442d72009-03-07 23:46:26 -0800150static void __cpuinit amd_k7_smp_check(struct cpuinfo_x86 *c)
151{
Yinghai Lu1f442d72009-03-07 23:46:26 -0800152 /* calling is from identify_secondary_cpu() ? */
Robert Richterf6e9456c2010-07-21 19:03:58 +0200153 if (!c->cpu_index)
Yinghai Lu1f442d72009-03-07 23:46:26 -0800154 return;
155
156 /*
157 * Certain Athlons might work (for various values of 'work') in SMP
158 * but they are not certified as MP capable.
159 */
160 /* Athlon 660/661 is valid. */
161 if ((c->x86_model == 6) && ((c->x86_mask == 0) ||
162 (c->x86_mask == 1)))
163 goto valid_k7;
164
165 /* Duron 670 is valid */
166 if ((c->x86_model == 7) && (c->x86_mask == 0))
167 goto valid_k7;
168
169 /*
170 * Athlon 662, Duron 671, and Athlon >model 7 have capability
171 * bit. It's worth noting that the A5 stepping (662) of some
172 * Athlon XP's have the MP bit set.
173 * See http://www.heise.de/newsticker/data/jow-18.10.01-000 for
174 * more.
175 */
176 if (((c->x86_model == 6) && (c->x86_mask >= 2)) ||
177 ((c->x86_model == 7) && (c->x86_mask >= 1)) ||
178 (c->x86_model > 7))
179 if (cpu_has_mp)
180 goto valid_k7;
181
182 /* If we get here, not a certified SMP capable AMD system. */
183
184 /*
185 * Don't taint if we are running SMP kernel on a single non-MP
186 * approved Athlon
187 */
188 WARN_ONCE(1, "WARNING: This combination of AMD"
Michael Tokarev7da8b6d2009-07-22 17:50:23 +0400189 " processors is not suitable for SMP.\n");
Yinghai Lu1f442d72009-03-07 23:46:26 -0800190 if (!test_taint(TAINT_UNSAFE_SMP))
191 add_taint(TAINT_UNSAFE_SMP);
192
193valid_k7:
194 ;
Yinghai Lu1f442d72009-03-07 23:46:26 -0800195}
196
Yinghai Lu11fdd252008-09-07 17:58:50 -0700197static void __cpuinit init_amd_k7(struct cpuinfo_x86 *c)
198{
199 u32 l, h;
200
201 /*
202 * Bit 15 of Athlon specific MSR 15, needs to be 0
203 * to enable SSE on Palomino/Morgan/Barton CPU's.
204 * If the BIOS didn't enable it already, enable it here.
205 */
206 if (c->x86_model >= 6 && c->x86_model <= 10) {
207 if (!cpu_has(c, X86_FEATURE_XMM)) {
208 printk(KERN_INFO "Enabling disabled K7/SSE Support.\n");
209 rdmsr(MSR_K7_HWCR, l, h);
210 l &= ~0x00008000;
211 wrmsr(MSR_K7_HWCR, l, h);
212 set_cpu_cap(c, X86_FEATURE_XMM);
213 }
214 }
215
216 /*
217 * It's been determined by AMD that Athlons since model 8 stepping 1
218 * are more robust with CLK_CTL set to 200xxxxx instead of 600xxxxx
219 * As per AMD technical note 27212 0.2
220 */
221 if ((c->x86_model == 8 && c->x86_mask >= 1) || (c->x86_model > 8)) {
222 rdmsr(MSR_K7_CLK_CTL, l, h);
223 if ((l & 0xfff00000) != 0x20000000) {
Alan Cox8bdbd962009-07-04 00:35:45 +0100224 printk(KERN_INFO
225 "CPU: CLK_CTL MSR was %x. Reprogramming to %x\n",
226 l, ((l & 0x000fffff)|0x20000000));
Yinghai Lu11fdd252008-09-07 17:58:50 -0700227 wrmsr(MSR_K7_CLK_CTL, (l & 0x000fffff)|0x20000000, h);
228 }
229 }
230
231 set_cpu_cap(c, X86_FEATURE_K7);
Yinghai Lu1f442d72009-03-07 23:46:26 -0800232
233 amd_k7_smp_check(c);
Yinghai Lu11fdd252008-09-07 17:58:50 -0700234}
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700235#endif
236
Tejun Heo645a7912011-01-23 14:37:40 +0100237#ifdef CONFIG_NUMA
Tejun Heobbc9e2f2011-01-23 14:37:39 +0100238/*
239 * To workaround broken NUMA config. Read the comment in
240 * srat_detect_node().
241 */
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700242static int __cpuinit nearby_node(int apicid)
243{
244 int i, node;
245
246 for (i = apicid - 1; i >= 0; i--) {
Tejun Heobbc9e2f2011-01-23 14:37:39 +0100247 node = __apicid_to_node[i];
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700248 if (node != NUMA_NO_NODE && node_online(node))
249 return node;
250 }
251 for (i = apicid + 1; i < MAX_LOCAL_APIC; i++) {
Tejun Heobbc9e2f2011-01-23 14:37:39 +0100252 node = __apicid_to_node[i];
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700253 if (node != NUMA_NO_NODE && node_online(node))
254 return node;
255 }
256 return first_node(node_online_map); /* Shouldn't happen */
257}
258#endif
Yinghai Lu11fdd252008-09-07 17:58:50 -0700259
260/*
Andreas Herrmann23588c32010-09-30 14:36:28 +0200261 * Fixup core topology information for
262 * (1) AMD multi-node processors
263 * Assumption: Number of cores in each internal node is the same.
Andreas Herrmann6057b4d2010-09-30 14:38:57 +0200264 * (2) AMD processors supporting compute units
Andreas Herrmann4a376ec2009-09-03 09:40:21 +0200265 */
266#ifdef CONFIG_X86_HT
Andreas Herrmann23588c32010-09-30 14:36:28 +0200267static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
Andreas Herrmann4a376ec2009-09-03 09:40:21 +0200268{
Borislav Petkov9e815092011-02-14 18:14:51 +0100269 u32 nodes, cores_per_cu = 1;
Andreas Herrmann23588c32010-09-30 14:36:28 +0200270 u8 node_id;
Andreas Herrmann4a376ec2009-09-03 09:40:21 +0200271 int cpu = smp_processor_id();
272
Andreas Herrmann23588c32010-09-30 14:36:28 +0200273 /* get information required for multi-node processors */
274 if (cpu_has(c, X86_FEATURE_TOPOEXT)) {
Andreas Herrmann6057b4d2010-09-30 14:38:57 +0200275 u32 eax, ebx, ecx, edx;
276
277 cpuid(0x8000001e, &eax, &ebx, &ecx, &edx);
278 nodes = ((ecx >> 8) & 7) + 1;
279 node_id = ecx & 7;
280
281 /* get compute unit information */
282 smp_num_siblings = ((ebx >> 8) & 3) + 1;
283 c->compute_unit_id = ebx & 0xff;
Borislav Petkov9e815092011-02-14 18:14:51 +0100284 cores_per_cu += ((ebx >> 8) & 3);
Andreas Herrmann23588c32010-09-30 14:36:28 +0200285 } else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) {
Andreas Herrmann6057b4d2010-09-30 14:38:57 +0200286 u64 value;
287
Andreas Herrmann23588c32010-09-30 14:36:28 +0200288 rdmsrl(MSR_FAM10H_NODE_ID, value);
289 nodes = ((value >> 3) & 7) + 1;
290 node_id = value & 7;
291 } else
Andreas Herrmann9d260eb2009-12-16 15:43:55 +0100292 return;
293
Andreas Herrmann23588c32010-09-30 14:36:28 +0200294 /* fixup multi-node processor information */
295 if (nodes > 1) {
Andreas Herrmann6057b4d2010-09-30 14:38:57 +0200296 u32 cores_per_node;
Andreas Herrmannd5185732011-01-24 16:05:40 +0100297 u32 cus_per_node;
Andreas Herrmann6057b4d2010-09-30 14:38:57 +0200298
Andreas Herrmann23588c32010-09-30 14:36:28 +0200299 set_cpu_cap(c, X86_FEATURE_AMD_DCM);
300 cores_per_node = c->x86_max_cores / nodes;
Andreas Herrmannd5185732011-01-24 16:05:40 +0100301 cus_per_node = cores_per_node / cores_per_cu;
Andreas Herrmann4a376ec2009-09-03 09:40:21 +0200302
Andreas Herrmann23588c32010-09-30 14:36:28 +0200303 /* store NodeID, use llc_shared_map to store sibling info */
304 per_cpu(cpu_llc_id, cpu) = node_id;
Andreas Herrmann9d260eb2009-12-16 15:43:55 +0100305
Borislav Petkov9e815092011-02-14 18:14:51 +0100306 /* core id has to be in the [0 .. cores_per_node - 1] range */
Andreas Herrmannd5185732011-01-24 16:05:40 +0100307 c->cpu_core_id %= cores_per_node;
308 c->compute_unit_id %= cus_per_node;
Andreas Herrmann23588c32010-09-30 14:36:28 +0200309 }
Andreas Herrmann4a376ec2009-09-03 09:40:21 +0200310}
311#endif
312
313/*
Yinghai Lu11fdd252008-09-07 17:58:50 -0700314 * On a AMD dual core setup the lower bits of the APIC id distingush the cores.
315 * Assumes number of cores is a power of two.
316 */
317static void __cpuinit amd_detect_cmp(struct cpuinfo_x86 *c)
318{
319#ifdef CONFIG_X86_HT
320 unsigned bits;
Andreas Herrmann99bd0c02009-06-19 10:59:09 +0200321 int cpu = smp_processor_id();
Yinghai Lu11fdd252008-09-07 17:58:50 -0700322
323 bits = c->x86_coreid_bits;
Yinghai Lu11fdd252008-09-07 17:58:50 -0700324 /* Low order bits define the core id (index of core in socket) */
325 c->cpu_core_id = c->initial_apicid & ((1 << bits)-1);
326 /* Convert the initial APIC ID into the socket ID */
327 c->phys_proc_id = c->initial_apicid >> bits;
Andreas Herrmann99bd0c02009-06-19 10:59:09 +0200328 /* use socket ID also for last level cache */
329 per_cpu(cpu_llc_id, cpu) = c->phys_proc_id;
Andreas Herrmann23588c32010-09-30 14:36:28 +0200330 amd_get_topology(c);
Yinghai Lu11fdd252008-09-07 17:58:50 -0700331#endif
332}
333
Andreas Herrmann6a812692009-09-16 11:33:40 +0200334int amd_get_nb_id(int cpu)
335{
336 int id = 0;
337#ifdef CONFIG_SMP
338 id = per_cpu(cpu_llc_id, cpu);
339#endif
340 return id;
341}
342EXPORT_SYMBOL_GPL(amd_get_nb_id);
343
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700344static void __cpuinit srat_detect_node(struct cpuinfo_x86 *c)
345{
Tejun Heo645a7912011-01-23 14:37:40 +0100346#ifdef CONFIG_NUMA
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700347 int cpu = smp_processor_id();
348 int node;
Yinghai Lu0d96b9f2009-08-29 13:17:14 -0700349 unsigned apicid = c->apicid;
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700350
Tejun Heobbc9e2f2011-01-23 14:37:39 +0100351 node = numa_cpu_node(cpu);
352 if (node == NUMA_NO_NODE)
353 node = per_cpu(cpu_llc_id, cpu);
Andreas Herrmann4a376ec2009-09-03 09:40:21 +0200354
Daniel J Blueman64be4c12011-12-05 16:20:37 +0800355 /*
356 * If core numbers are inconsistent, it's likely a multi-fabric platform,
357 * so invoke platform-specific handler
358 */
359 if (c->phys_proc_id != node)
360 x86_cpuinit.fixup_cpu_id(c, node);
361
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700362 if (!node_online(node)) {
Tejun Heobbc9e2f2011-01-23 14:37:39 +0100363 /*
364 * Two possibilities here:
365 *
366 * - The CPU is missing memory and no node was created. In
367 * that case try picking one from a nearby CPU.
368 *
369 * - The APIC IDs differ from the HyperTransport node IDs
370 * which the K8 northbridge parsing fills in. Assume
371 * they are all increased by a constant offset, but in
372 * the same order as the HT nodeids. If that doesn't
373 * result in a usable node fall back to the path for the
374 * previous case.
375 *
376 * This workaround operates directly on the mapping between
377 * APIC ID and NUMA node, assuming certain relationship
378 * between APIC ID, HT node ID and NUMA topology. As going
379 * through CPU mapping may alter the outcome, directly
380 * access __apicid_to_node[].
381 */
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700382 int ht_nodeid = c->initial_apicid;
383
384 if (ht_nodeid >= 0 &&
Tejun Heobbc9e2f2011-01-23 14:37:39 +0100385 __apicid_to_node[ht_nodeid] != NUMA_NO_NODE)
386 node = __apicid_to_node[ht_nodeid];
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700387 /* Pick a nearby node */
388 if (!node_online(node))
389 node = nearby_node(apicid);
390 }
391 numa_set_node(cpu, node);
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700392#endif
393}
394
Yinghai Lu11fdd252008-09-07 17:58:50 -0700395static void __cpuinit early_init_amd_mc(struct cpuinfo_x86 *c)
396{
397#ifdef CONFIG_X86_HT
398 unsigned bits, ecx;
399
400 /* Multi core CPU? */
401 if (c->extended_cpuid_level < 0x80000008)
402 return;
403
404 ecx = cpuid_ecx(0x80000008);
405
406 c->x86_max_cores = (ecx & 0xff) + 1;
407
408 /* CPU telling us the core id bits shift? */
409 bits = (ecx >> 12) & 0xF;
410
411 /* Otherwise recompute */
412 if (bits == 0) {
413 while ((1 << bits) < c->x86_max_cores)
414 bits++;
415 }
416
417 c->x86_coreid_bits = bits;
418#endif
419}
420
Borislav Petkov8fa8b032011-08-05 20:04:09 +0200421static void __cpuinit bsp_init_amd(struct cpuinfo_x86 *c)
422{
423 if (cpu_has(c, X86_FEATURE_CONSTANT_TSC)) {
424
425 if (c->x86 > 0x10 ||
426 (c->x86 == 0x10 && c->x86_model >= 0x2)) {
427 u64 val;
428
429 rdmsrl(MSR_K7_HWCR, val);
430 if (!(val & BIT(24)))
431 printk(KERN_WARNING FW_BUG "TSC doesn't count "
432 "with P0 frequency!\n");
433 }
434 }
435
436 if (c->x86 == 0x15) {
437 unsigned long upperbit;
438 u32 cpuid, assoc;
439
440 cpuid = cpuid_edx(0x80000005);
441 assoc = cpuid >> 16 & 0xff;
442 upperbit = ((cpuid >> 24) << 10) / assoc;
443
444 va_align.mask = (upperbit - 1) & PAGE_MASK;
445 va_align.flags = ALIGN_VA_32 | ALIGN_VA_64;
446 }
447}
448
Thomas Petazzoni03ae5762008-02-15 12:00:23 +0100449static void __cpuinit early_init_amd(struct cpuinfo_x86 *c)
Andi Kleen2b16a232008-01-30 13:32:40 +0100450{
Yinghai Lu11fdd252008-09-07 17:58:50 -0700451 early_init_amd_mc(c);
452
Venki Pallipadi40fb1712008-11-17 16:11:37 -0800453 /*
454 * c->x86_power is 8000_0007 edx. Bit 8 is TSC runs at constant rate
455 * with P/T states and does not stop in deep C-states
456 */
457 if (c->x86_power & (1 << 8)) {
Yinghai Lue3224232008-09-06 01:52:28 -0700458 set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC);
Venki Pallipadi40fb1712008-11-17 16:11:37 -0800459 set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC);
Borislav Petkovc98fdea2012-02-07 13:08:52 +0100460 if (!check_tsc_unstable())
461 sched_clock_stable = 1;
Venki Pallipadi40fb1712008-11-17 16:11:37 -0800462 }
Yinghai Lu5fef55f2008-09-04 21:09:43 +0200463
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700464#ifdef CONFIG_X86_64
465 set_cpu_cap(c, X86_FEATURE_SYSCALL32);
466#else
Yinghai Lu5fef55f2008-09-04 21:09:43 +0200467 /* Set MTRR capability flag if appropriate */
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700468 if (c->x86 == 5)
469 if (c->x86_model == 13 || c->x86_model == 9 ||
470 (c->x86_model == 8 && c->x86_mask >= 8))
471 set_cpu_cap(c, X86_FEATURE_K6_MTRR);
472#endif
Andreas Herrmann42937e82009-06-08 15:55:09 +0200473#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_PCI)
474 /* check CPU config space for extended APIC ID */
Jeremy Fitzhardinge2cb07862009-07-22 09:59:35 -0700475 if (cpu_has_apic && c->x86 >= 0xf) {
Andreas Herrmann42937e82009-06-08 15:55:09 +0200476 unsigned int val;
477 val = read_pci_config(0, 24, 0, 0x68);
478 if ((val & ((1 << 17) | (1 << 18))) == ((1 << 17) | (1 << 18)))
479 set_cpu_cap(c, X86_FEATURE_EXTD_APICID);
480 }
481#endif
Andi Kleen2b16a232008-01-30 13:32:40 +0100482}
483
Magnus Dammb4af3f72006-09-26 10:52:36 +0200484static void __cpuinit init_amd(struct cpuinfo_x86 *c)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700485{
Linus Torvalds8e8da022011-12-04 11:57:09 -0800486 u32 dummy;
487
Andi Kleen7d318d72005-09-29 22:05:55 +0200488#ifdef CONFIG_SMP
Andi Kleen3c92c2b2005-10-11 01:28:33 +0200489 unsigned long long value;
Andi Kleen7d318d72005-09-29 22:05:55 +0200490
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100491 /*
492 * Disable TLB flush filter by setting HWCR.FFDIS on K8
Andi Kleen7d318d72005-09-29 22:05:55 +0200493 * bit 6 of msr C001_0015
494 *
495 * Errata 63 for SH-B3 steppings
496 * Errata 122 for all steppings (F+ have it disabled by default)
497 */
Yinghai Lu11fdd252008-09-07 17:58:50 -0700498 if (c->x86 == 0xf) {
Andi Kleen7d318d72005-09-29 22:05:55 +0200499 rdmsrl(MSR_K7_HWCR, value);
500 value |= 1 << 6;
501 wrmsrl(MSR_K7_HWCR, value);
502 }
503#endif
504
Andi Kleen2b16a232008-01-30 13:32:40 +0100505 early_init_amd(c);
506
Linus Torvalds1da177e2005-04-16 15:20:36 -0700507 /*
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100508 * Bit 31 in normal CPUID used for nonstandard 3DNow ID;
Ingo Molnar16282a82008-02-26 08:49:57 +0100509 * 3DNow is IDd by bit 31 in extended CPUID (1*32+31) anyway
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100510 */
Ingo Molnar16282a82008-02-26 08:49:57 +0100511 clear_cpu_cap(c, 0*32+31);
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100512
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700513#ifdef CONFIG_X86_64
514 /* On C+ stepping K8 rep microcode works well for copy/memset */
515 if (c->x86 == 0xf) {
516 u32 level;
517
518 level = cpuid_eax(1);
Alan Cox8bdbd962009-07-04 00:35:45 +0100519 if ((level >= 0x0f48 && level < 0x0f50) || level >= 0x0f58)
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700520 set_cpu_cap(c, X86_FEATURE_REP_GOOD);
Kevin Winchesterfbd8b182009-08-10 19:56:45 -0300521
522 /*
523 * Some BIOSes incorrectly force this feature, but only K8
524 * revision D (model = 0x14) and later actually support it.
Borislav Petkov6b0f43d2009-08-31 09:50:11 +0200525 * (AMD Erratum #110, docId: 25759).
Kevin Winchesterfbd8b182009-08-10 19:56:45 -0300526 */
Borislav Petkov6b0f43d2009-08-31 09:50:11 +0200527 if (c->x86_model < 0x14 && cpu_has(c, X86_FEATURE_LAHF_LM)) {
528 u64 val;
529
Kevin Winchesterfbd8b182009-08-10 19:56:45 -0300530 clear_cpu_cap(c, X86_FEATURE_LAHF_LM);
Borislav Petkov6b0f43d2009-08-31 09:50:11 +0200531 if (!rdmsrl_amd_safe(0xc001100d, &val)) {
532 val &= ~(1ULL << 32);
533 wrmsrl_amd_safe(0xc001100d, val);
534 }
535 }
536
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700537 }
Borislav Petkov12d8a962010-06-02 20:29:21 +0200538 if (c->x86 >= 0x10)
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700539 set_cpu_cap(c, X86_FEATURE_REP_GOOD);
Yinghai Lu0d96b9f2009-08-29 13:17:14 -0700540
541 /* get apicid instead of initial apic id from cpuid */
542 c->apicid = hard_smp_processor_id();
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700543#else
544
545 /*
546 * FIXME: We should handle the K5 here. Set up the write
547 * range and also turn on MSR 83 bits 4 and 31 (write alloc,
548 * no bus pipeline)
549 */
550
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100551 switch (c->x86) {
552 case 4:
Yinghai Lu11fdd252008-09-07 17:58:50 -0700553 init_amd_k5(c);
554 break;
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100555 case 5:
Yinghai Lu11fdd252008-09-07 17:58:50 -0700556 init_amd_k6(c);
557 break;
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100558 case 6: /* An Athlon/Duron */
Yinghai Lu11fdd252008-09-07 17:58:50 -0700559 init_amd_k7(c);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700560 break;
Andi Kleen67cddd92007-07-21 17:10:03 +0200561 }
Andi Kleen3556ddf2007-04-02 12:14:12 +0200562
Andi Kleenc12ceb72007-05-21 14:31:47 +0200563 /* K6s reports MCEs but don't actually have all the MSRs */
564 if (c->x86 < 6)
Ingo Molnar16282a82008-02-26 08:49:57 +0100565 clear_cpu_cap(c, X86_FEATURE_MCE);
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700566#endif
Andi Kleende421862008-01-30 13:32:37 +0100567
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700568 /* Enable workaround for FXSAVE leak */
Yinghai Lu11fdd252008-09-07 17:58:50 -0700569 if (c->x86 >= 6)
570 set_cpu_cap(c, X86_FEATURE_FXSAVE_LEAK);
571
572 if (!c->x86_model_id[0]) {
573 switch (c->x86) {
574 case 0xf:
575 /* Should distinguish Models here, but this is only
576 a fallback anyways. */
577 strcpy(c->x86_model_id, "Hammer");
578 break;
579 }
580 }
581
Borislav Petkov27c13ec2009-11-21 14:01:45 +0100582 cpu_detect_cache_sizes(c);
Yinghai Lu11fdd252008-09-07 17:58:50 -0700583
584 /* Multi core CPU? */
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700585 if (c->extended_cpuid_level >= 0x80000008) {
Yinghai Lu11fdd252008-09-07 17:58:50 -0700586 amd_detect_cmp(c);
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700587 srat_detect_node(c);
588 }
Yinghai Lu11fdd252008-09-07 17:58:50 -0700589
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700590#ifdef CONFIG_X86_32
Yinghai Lu11fdd252008-09-07 17:58:50 -0700591 detect_ht(c);
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700592#endif
Yinghai Lu11fdd252008-09-07 17:58:50 -0700593
594 if (c->extended_cpuid_level >= 0x80000006) {
Andreas Herrmannd9fadd72010-09-02 15:37:10 +0200595 if (cpuid_edx(0x80000006) & 0xf000)
Yinghai Lu11fdd252008-09-07 17:58:50 -0700596 num_cache_leaves = 4;
597 else
598 num_cache_leaves = 3;
599 }
600
Borislav Petkov12d8a962010-06-02 20:29:21 +0200601 if (c->x86 >= 0xf)
Yinghai Lu11fdd252008-09-07 17:58:50 -0700602 set_cpu_cap(c, X86_FEATURE_K8);
603
604 if (cpu_has_xmm2) {
605 /* MFENCE stops RDTSC speculation */
Ingo Molnar16282a82008-02-26 08:49:57 +0100606 set_cpu_cap(c, X86_FEATURE_MFENCE_RDTSC);
Yinghai Lu11fdd252008-09-07 17:58:50 -0700607 }
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700608
609#ifdef CONFIG_X86_64
610 if (c->x86 == 0x10) {
611 /* do this for boot cpu */
612 if (c == &boot_cpu_data)
613 check_enable_amd_mmconf_dmi();
614
615 fam10h_check_enable_mmcfg();
616 }
617
Borislav Petkov12d8a962010-06-02 20:29:21 +0200618 if (c == &boot_cpu_data && c->x86 >= 0xf) {
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700619 unsigned long long tseg;
620
621 /*
622 * Split up direct mapping around the TSEG SMM area.
623 * Don't do it for gbpages because there seems very little
624 * benefit in doing so.
625 */
626 if (!rdmsrl_safe(MSR_K8_TSEG_ADDR, &tseg)) {
Alan Cox8bdbd962009-07-04 00:35:45 +0100627 printk(KERN_DEBUG "tseg: %010llx\n", tseg);
628 if ((tseg>>PMD_SHIFT) <
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700629 (max_low_pfn_mapped>>(PMD_SHIFT-PAGE_SHIFT)) ||
Alan Cox8bdbd962009-07-04 00:35:45 +0100630 ((tseg>>PMD_SHIFT) <
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700631 (max_pfn_mapped>>(PMD_SHIFT-PAGE_SHIFT)) &&
Alan Cox8bdbd962009-07-04 00:35:45 +0100632 (tseg>>PMD_SHIFT) >= (1ULL<<(32 - PMD_SHIFT))))
633 set_memory_4k((unsigned long)__va(tseg), 1);
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700634 }
635 }
636#endif
Boris Ostrovskyb87cf802011-03-15 12:13:44 -0400637
Boris Ostrovskye9cdd342011-05-26 11:19:52 -0400638 /*
639 * Family 0x12 and above processors have APIC timer
640 * running in deep C states.
641 */
642 if (c->x86 > 0x11)
Boris Ostrovskyb87cf802011-03-15 12:13:44 -0400643 set_cpu_cap(c, X86_FEATURE_ARAT);
Joerg Roedel5bbc0972011-04-15 14:47:40 +0200644
645 /*
646 * Disable GART TLB Walk Errors on Fam10h. We do this here
647 * because this is always needed when GART is enabled, even in a
648 * kernel which has no MCE support built in.
649 */
650 if (c->x86 == 0x10) {
651 /*
652 * BIOS should disable GartTlbWlk Errors themself. If
653 * it doesn't do it here as suggested by the BKDG.
654 *
655 * Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=33012
656 */
657 u64 mask;
Roedel, Joergd47cc0d2011-05-19 11:13:39 +0200658 int err;
Joerg Roedel5bbc0972011-04-15 14:47:40 +0200659
Roedel, Joergd47cc0d2011-05-19 11:13:39 +0200660 err = rdmsrl_safe(MSR_AMD64_MCx_MASK(4), &mask);
661 if (err == 0) {
662 mask |= (1 << 10);
663 checking_wrmsrl(MSR_AMD64_MCx_MASK(4), mask);
664 }
Joerg Roedel5bbc0972011-04-15 14:47:40 +0200665 }
Linus Torvalds8e8da022011-12-04 11:57:09 -0800666
667 rdmsr_safe(MSR_AMD64_PATCH_LEVEL, &c->microcode, &dummy);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700668}
669
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700670#ifdef CONFIG_X86_32
Alan Cox8bdbd962009-07-04 00:35:45 +0100671static unsigned int __cpuinit amd_size_cache(struct cpuinfo_x86 *c,
672 unsigned int size)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700673{
674 /* AMD errata T13 (order #21922) */
675 if ((c->x86 == 6)) {
Alan Cox8bdbd962009-07-04 00:35:45 +0100676 /* Duron Rev A0 */
677 if (c->x86_model == 3 && c->x86_mask == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700678 size = 64;
Alan Cox8bdbd962009-07-04 00:35:45 +0100679 /* Tbird rev A1/A2 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700680 if (c->x86_model == 4 &&
Alan Cox8bdbd962009-07-04 00:35:45 +0100681 (c->x86_mask == 0 || c->x86_mask == 1))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700682 size = 256;
683 }
684 return size;
685}
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700686#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700687
Jan Beulich02dde8b2009-03-12 12:08:49 +0000688static const struct cpu_dev __cpuinitconst amd_cpu_dev = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700689 .c_vendor = "AMD",
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100690 .c_ident = { "AuthenticAMD" },
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700691#ifdef CONFIG_X86_32
Linus Torvalds1da177e2005-04-16 15:20:36 -0700692 .c_models = {
693 { .vendor = X86_VENDOR_AMD, .family = 4, .model_names =
694 {
695 [3] = "486 DX/2",
696 [7] = "486 DX/2-WB",
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100697 [8] = "486 DX/4",
698 [9] = "486 DX/4-WB",
Linus Torvalds1da177e2005-04-16 15:20:36 -0700699 [14] = "Am5x86-WT",
Paolo Ciarrocchifb87a292008-02-22 23:10:33 +0100700 [15] = "Am5x86-WB"
Linus Torvalds1da177e2005-04-16 15:20:36 -0700701 }
702 },
703 },
Yinghai Lu6c62aa42008-09-07 17:58:54 -0700704 .c_size_cache = amd_size_cache,
705#endif
Thomas Petazzoni03ae5762008-02-15 12:00:23 +0100706 .c_early_init = early_init_amd,
Borislav Petkov8fa8b032011-08-05 20:04:09 +0200707 .c_bsp_init = bsp_init_amd,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700708 .c_init = init_amd,
Yinghai Lu10a434f2008-09-04 21:09:45 +0200709 .c_x86_vendor = X86_VENDOR_AMD,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700710};
711
Yinghai Lu10a434f2008-09-04 21:09:45 +0200712cpu_dev_register(amd_cpu_dev);
Hans Rosenfeldd78d6712010-07-28 19:09:30 +0200713
714/*
715 * AMD errata checking
716 *
717 * Errata are defined as arrays of ints using the AMD_LEGACY_ERRATUM() or
718 * AMD_OSVW_ERRATUM() macros. The latter is intended for newer errata that
719 * have an OSVW id assigned, which it takes as first argument. Both take a
720 * variable number of family-specific model-stepping ranges created by
721 * AMD_MODEL_RANGE(). Each erratum also has to be declared as extern const
722 * int[] in arch/x86/include/asm/processor.h.
723 *
724 * Example:
725 *
726 * const int amd_erratum_319[] =
727 * AMD_LEGACY_ERRATUM(AMD_MODEL_RANGE(0x10, 0x2, 0x1, 0x4, 0x2),
728 * AMD_MODEL_RANGE(0x10, 0x8, 0x0, 0x8, 0x0),
729 * AMD_MODEL_RANGE(0x10, 0x9, 0x0, 0x9, 0x0));
730 */
731
Hans Rosenfeld9d8888c2010-07-28 19:09:31 +0200732const int amd_erratum_400[] =
Borislav Petkov328935e2011-05-17 14:55:18 +0200733 AMD_OSVW_ERRATUM(1, AMD_MODEL_RANGE(0xf, 0x41, 0x2, 0xff, 0xf),
Hans Rosenfeld9d8888c2010-07-28 19:09:31 +0200734 AMD_MODEL_RANGE(0x10, 0x2, 0x1, 0xff, 0xf));
H. Peter Anvina5b91602010-07-28 16:23:20 -0700735EXPORT_SYMBOL_GPL(amd_erratum_400);
Hans Rosenfeld9d8888c2010-07-28 19:09:31 +0200736
Hans Rosenfeld1be85a62010-07-28 19:09:32 +0200737const int amd_erratum_383[] =
738 AMD_OSVW_ERRATUM(3, AMD_MODEL_RANGE(0x10, 0, 0, 0xff, 0xf));
H. Peter Anvina5b91602010-07-28 16:23:20 -0700739EXPORT_SYMBOL_GPL(amd_erratum_383);
Hans Rosenfeld9d8888c2010-07-28 19:09:31 +0200740
Hans Rosenfeldd78d6712010-07-28 19:09:30 +0200741bool cpu_has_amd_erratum(const int *erratum)
742{
Tejun Heo7b543a52010-12-18 16:30:05 +0100743 struct cpuinfo_x86 *cpu = __this_cpu_ptr(&cpu_info);
Hans Rosenfeldd78d6712010-07-28 19:09:30 +0200744 int osvw_id = *erratum++;
745 u32 range;
746 u32 ms;
747
748 /*
749 * If called early enough that current_cpu_data hasn't been initialized
750 * yet, fall back to boot_cpu_data.
751 */
752 if (cpu->x86 == 0)
753 cpu = &boot_cpu_data;
754
755 if (cpu->x86_vendor != X86_VENDOR_AMD)
756 return false;
757
758 if (osvw_id >= 0 && osvw_id < 65536 &&
759 cpu_has(cpu, X86_FEATURE_OSVW)) {
760 u64 osvw_len;
761
762 rdmsrl(MSR_AMD64_OSVW_ID_LENGTH, osvw_len);
763 if (osvw_id < osvw_len) {
764 u64 osvw_bits;
765
766 rdmsrl(MSR_AMD64_OSVW_STATUS + (osvw_id >> 6),
767 osvw_bits);
768 return osvw_bits & (1ULL << (osvw_id & 0x3f));
769 }
770 }
771
772 /* OSVW unavailable or ID unknown, match family-model-stepping range */
Hans Rosenfeld07a77952010-08-18 16:19:50 +0200773 ms = (cpu->x86_model << 4) | cpu->x86_mask;
Hans Rosenfeldd78d6712010-07-28 19:09:30 +0200774 while ((range = *erratum++))
775 if ((cpu->x86 == AMD_MODEL_RANGE_FAMILY(range)) &&
776 (ms >= AMD_MODEL_RANGE_START(range)) &&
777 (ms <= AMD_MODEL_RANGE_END(range)))
778 return true;
779
780 return false;
781}
H. Peter Anvina5b91602010-07-28 16:23:20 -0700782
783EXPORT_SYMBOL_GPL(cpu_has_amd_erratum);