Blame view
arch/x86/kernel/xsave.c
11.1 KB
dc1e35c6e
|
1 2 3 4 5 6 7 8 |
/* * xsave/xrstor support. * * Author: Suresh Siddha <suresh.b.siddha@intel.com> */ #include <linux/bootmem.h> #include <linux/compat.h> #include <asm/i387.h> |
c37b5efea
|
9 10 11 |
#ifdef CONFIG_IA32_EMULATION #include <asm/sigcontext32.h> #endif |
6152e4b1c
|
12 |
#include <asm/xcr.h> |
dc1e35c6e
|
13 14 15 16 |
/* * Supported feature mask by the CPU and the kernel. */ |
6152e4b1c
|
17 |
u64 pcntxt_mask; |
dc1e35c6e
|
18 |
|
45c2d7f46
|
19 20 21 22 |
/* * Represents init state for the supported extended state. */ static struct xsave_struct *init_xstate_buf; |
c37b5efea
|
23 24 25 26 |
struct _fpx_sw_bytes fx_sw_reserved; #ifdef CONFIG_IA32_EMULATION struct _fpx_sw_bytes fx_sw_reserved_ia32; #endif |
a1488f8bf
|
27 |
static unsigned int *xstate_offsets, *xstate_sizes, xstate_features; |
c37b5efea
|
28 |
/* |
29104e101
|
29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 |
* If a processor implementation discern that a processor state component is * in its initialized state it may modify the corresponding bit in the * xsave_hdr.xstate_bv as '0', with out modifying the corresponding memory * layout in the case of xsaveopt. While presenting the xstate information to * the user, we always ensure that the memory layout of a feature will be in * the init state if the corresponding header bit is zero. This is to ensure * that the user doesn't see some stale state in the memory layout during * signal handling, debugging etc. */ void __sanitize_i387_state(struct task_struct *tsk) { u64 xstate_bv; int feature_bit = 0x2; struct i387_fxsave_struct *fx = &tsk->thread.fpu.state->fxsave; if (!fx) return; BUG_ON(task_thread_info(tsk)->status & TS_USEDFPU); xstate_bv = tsk->thread.fpu.state->xsave.xsave_hdr.xstate_bv; /* * None of the feature bits are in init state. So nothing else |
0d2eb44f6
|
53 |
* to do for us, as the memory layout is up to date. |
29104e101
|
54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 |
*/ if ((xstate_bv & pcntxt_mask) == pcntxt_mask) return; /* * FP is in init state */ if (!(xstate_bv & XSTATE_FP)) { fx->cwd = 0x37f; fx->swd = 0; fx->twd = 0; fx->fop = 0; fx->rip = 0; fx->rdp = 0; memset(&fx->st_space[0], 0, 128); } /* * SSE is in init state */ if (!(xstate_bv & XSTATE_SSE)) memset(&fx->xmm_space[0], 0, 256); xstate_bv = (pcntxt_mask & ~xstate_bv) >> 2; /* * Update all the other memory layouts for which the corresponding * header bit is in the init state. */ while (xstate_bv) { if (xstate_bv & 0x1) { int offset = xstate_offsets[feature_bit]; int size = xstate_sizes[feature_bit]; memcpy(((void *) fx) + offset, ((void *) init_xstate_buf) + offset, size); } xstate_bv >>= 1; feature_bit++; } } |
c37b5efea
|
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 |
/* * Check for the presence of extended state information in the * user fpstate pointer in the sigcontext. */ int check_for_xstate(struct i387_fxsave_struct __user *buf, void __user *fpstate, struct _fpx_sw_bytes *fx_sw_user) { int min_xstate_size = sizeof(struct i387_fxsave_struct) + sizeof(struct xsave_hdr_struct); unsigned int magic2; int err; err = __copy_from_user(fx_sw_user, &buf->sw_reserved[0], sizeof(struct _fpx_sw_bytes)); |
c37b5efea
|
112 |
if (err) |
d6d4d4205
|
113 |
return -EFAULT; |
c37b5efea
|
114 115 116 117 118 |
/* * First Magic check failed. */ if (fx_sw_user->magic1 != FP_XSTATE_MAGIC1) |
d6d4d4205
|
119 |
return -EINVAL; |
c37b5efea
|
120 121 122 123 124 125 126 |
/* * Check for error scenarios. */ if (fx_sw_user->xstate_size < min_xstate_size || fx_sw_user->xstate_size > xstate_size || fx_sw_user->xstate_size > fx_sw_user->extended_size) |
d6d4d4205
|
127 |
return -EINVAL; |
c37b5efea
|
128 129 130 131 |
err = __get_user(magic2, (__u32 *) (((void *)fpstate) + fx_sw_user->extended_size - FP_XSTATE_MAGIC2_SIZE)); |
d6d4d4205
|
132 133 |
if (err) return err; |
c37b5efea
|
134 135 136 137 138 139 |
/* * Check for the presence of second magic word at the end of memory * layout. This detects the case where the user just copied the legacy * fpstate layout with out copying the extended state information * in the memory layout. */ |
d6d4d4205
|
140 141 |
if (magic2 != FP_XSTATE_MAGIC2) return -EFAULT; |
c37b5efea
|
142 143 144 |
return 0; } |
ab5137015
|
145 146 147 148 149 150 151 152 153 154 155 156 |
#ifdef CONFIG_X86_64 /* * Signal frame handlers. */ int save_i387_xstate(void __user *buf) { struct task_struct *tsk = current; int err = 0; if (!access_ok(VERIFY_WRITE, buf, sig_xstate_size)) return -EACCES; |
f65bc214e
|
157 |
BUG_ON(sig_xstate_size < xstate_size); |
ab5137015
|
158 |
|
c37b5efea
|
159 |
if ((unsigned long)buf % 64) |
ab5137015
|
160 161 162 163 164 |
printk("save_i387_xstate: bad fpstate %p ", buf); if (!used_math()) return 0; |
06c38d5e3
|
165 |
|
ab5137015
|
166 |
if (task_thread_info(tsk)->status & TS_USEDFPU) { |
c9ad48828
|
167 |
if (use_xsave()) |
c37b5efea
|
168 169 170 |
err = xsave_user(buf); else err = fxsave_user(buf); |
ab5137015
|
171 172 173 174 175 |
if (err) return err; task_thread_info(tsk)->status &= ~TS_USEDFPU; stts(); } else { |
29104e101
|
176 |
sanitize_i387_state(tsk); |
866032833
|
177 |
if (__copy_to_user(buf, &tsk->thread.fpu.state->fxsave, |
ab5137015
|
178 179 180 |
xstate_size)) return -1; } |
c37b5efea
|
181 |
|
06c38d5e3
|
182 |
clear_used_math(); /* trigger finit */ |
c9ad48828
|
183 |
if (use_xsave()) { |
c37b5efea
|
184 |
struct _fpstate __user *fx = buf; |
04944b793
|
185 186 |
struct _xstate __user *x = buf; u64 xstate_bv; |
c37b5efea
|
187 188 189 190 191 192 193 |
err = __copy_to_user(&fx->sw_reserved, &fx_sw_reserved, sizeof(struct _fpx_sw_bytes)); err |= __put_user(FP_XSTATE_MAGIC2, (__u32 __user *) (buf + sig_xstate_size - FP_XSTATE_MAGIC2_SIZE)); |
04944b793
|
194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 |
/* * Read the xstate_bv which we copied (directly from the cpu or * from the state in task struct) to the user buffers and * set the FP/SSE bits. */ err |= __get_user(xstate_bv, &x->xstate_hdr.xstate_bv); /* * For legacy compatible, we always set FP/SSE bits in the bit * vector while saving the state to the user context. This will * enable us capturing any changes(during sigreturn) to * the FP/SSE bits by the legacy applications which don't touch * xstate_bv in the xsave header. * * xsave aware apps can change the xstate_bv in the xsave * header as well as change any contents in the memory layout. * xrestore as part of sigreturn will capture all the changes. */ xstate_bv |= XSTATE_FPSSE; err |= __put_user(xstate_bv, &x->xstate_hdr.xstate_bv); |
f364eadab
|
216 217 |
if (err) return err; |
c37b5efea
|
218 |
} |
ab5137015
|
219 220 221 222 |
return 1; } /* |
c37b5efea
|
223 224 225 |
* Restore the extended state if present. Otherwise, restore the FP/SSE * state. */ |
7820b7564
|
226 |
static int restore_user_xstate(void __user *buf) |
c37b5efea
|
227 228 |
{ struct _fpx_sw_bytes fx_sw_user; |
6152e4b1c
|
229 |
u64 mask; |
c37b5efea
|
230 231 232 233 234 |
int err; if (((unsigned long)buf % 64) || check_for_xstate(buf, buf, &fx_sw_user)) goto fx_only; |
6152e4b1c
|
235 |
mask = fx_sw_user.xstate_bv; |
c37b5efea
|
236 237 238 239 |
/* * restore the state passed by the user. */ |
6152e4b1c
|
240 |
err = xrestore_user(buf, mask); |
c37b5efea
|
241 242 243 244 245 246 |
if (err) return err; /* * init the state skipped by the user. */ |
6152e4b1c
|
247 |
mask = pcntxt_mask & ~mask; |
8e221b6db
|
248 249 |
if (unlikely(mask)) xrstor_state(init_xstate_buf, mask); |
c37b5efea
|
250 251 252 253 254 255 256 257 258 |
return 0; fx_only: /* * couldn't find the extended state information in the * memory layout. Restore just the FP/SSE and init all * the other extended state. */ |
6152e4b1c
|
259 |
xrstor_state(init_xstate_buf, pcntxt_mask & ~XSTATE_FPSSE); |
c37b5efea
|
260 261 262 263 |
return fxrstor_checking((__force struct i387_fxsave_struct *)buf); } /* |
ab5137015
|
264 265 266 267 268 |
* This restores directly out of user space. Exceptions are handled. */ int restore_i387_xstate(void __user *buf) { struct task_struct *tsk = current; |
c37b5efea
|
269 |
int err = 0; |
ab5137015
|
270 271 |
if (!buf) { |
c37b5efea
|
272 273 |
if (used_math()) goto clear; |
ab5137015
|
274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 |
return 0; } else if (!access_ok(VERIFY_READ, buf, sig_xstate_size)) return -EACCES; if (!used_math()) { err = init_fpu(tsk); if (err) return err; } if (!(task_thread_info(current)->status & TS_USEDFPU)) { clts(); task_thread_info(current)->status |= TS_USEDFPU; } |
c9ad48828
|
289 |
if (use_xsave()) |
c37b5efea
|
290 291 292 293 |
err = restore_user_xstate(buf); else err = fxrstor_checking((__force struct i387_fxsave_struct *) buf); |
ab5137015
|
294 295 296 297 298 |
if (unlikely(err)) { /* * Encountered an error while doing the restore from the * user buffer, clear the fpu state. */ |
c37b5efea
|
299 |
clear: |
ab5137015
|
300 301 302 303 304 305 |
clear_fpu(tsk); clear_used_math(); } return err; } #endif |
dc1e35c6e
|
306 |
/* |
c37b5efea
|
307 308 309 310 311 312 |
* Prepare the SW reserved portion of the fxsave memory layout, indicating * the presence of the extended state information in the memory layout * pointed by the fpstate pointer in the sigcontext. * This will be saved when ever the FP and extended state context is * saved on the user stack during the signal handler delivery to the user. */ |
8bcad30f2
|
313 |
static void prepare_fx_sw_frame(void) |
c37b5efea
|
314 315 316 317 318 319 320 321 322 323 324 325 326 327 |
{ int size_extended = (xstate_size - sizeof(struct i387_fxsave_struct)) + FP_XSTATE_MAGIC2_SIZE; sig_xstate_size = sizeof(struct _fpstate) + size_extended; #ifdef CONFIG_IA32_EMULATION sig_xstate_ia32_size = sizeof(struct _fpstate_ia32) + size_extended; #endif memset(&fx_sw_reserved, 0, sizeof(fx_sw_reserved)); fx_sw_reserved.magic1 = FP_XSTATE_MAGIC1; fx_sw_reserved.extended_size = sig_xstate_size; |
6152e4b1c
|
328 |
fx_sw_reserved.xstate_bv = pcntxt_mask; |
c37b5efea
|
329 330 331 332 333 334 335 |
fx_sw_reserved.xstate_size = xstate_size; #ifdef CONFIG_IA32_EMULATION memcpy(&fx_sw_reserved_ia32, &fx_sw_reserved, sizeof(struct _fpx_sw_bytes)); fx_sw_reserved_ia32.extended_size = sig_xstate_ia32_size; #endif } |
3c1c7f101
|
336 337 338 |
#ifdef CONFIG_X86_64 unsigned int sig_xstate_size = sizeof(struct _fpstate); #endif |
dc1e35c6e
|
339 340 341 |
/* * Enable the extended processor state save/restore feature */ |
1cff92d8f
|
342 |
static inline void xstate_enable(void) |
dc1e35c6e
|
343 |
{ |
dc1e35c6e
|
344 |
set_in_cr4(X86_CR4_OSXSAVE); |
6152e4b1c
|
345 |
xsetbv(XCR_XFEATURE_ENABLED_MASK, pcntxt_mask); |
dc1e35c6e
|
346 347 348 |
} /* |
a1488f8bf
|
349 350 351 |
* Record the offsets and sizes of different state managed by the xsave * memory layout. */ |
4995b9dba
|
352 |
static void __init setup_xstate_features(void) |
a1488f8bf
|
353 354 355 356 357 358 359 360 |
{ int eax, ebx, ecx, edx, leaf = 0x2; xstate_features = fls64(pcntxt_mask); xstate_offsets = alloc_bootmem(xstate_features * sizeof(int)); xstate_sizes = alloc_bootmem(xstate_features * sizeof(int)); do { |
ee813d53a
|
361 |
cpuid_count(XSTATE_CPUID, leaf, &eax, &ebx, &ecx, &edx); |
a1488f8bf
|
362 363 364 365 366 367 368 369 370 371 372 373 |
if (eax == 0) break; xstate_offsets[leaf] = ebx; xstate_sizes[leaf] = eax; leaf++; } while (1); } /* |
dc1e35c6e
|
374 375 |
* setup the xstate image representing the init state */ |
a19aac854
|
376 |
static void __init setup_xstate_init(void) |
dc1e35c6e
|
377 |
{ |
29104e101
|
378 379 380 381 382 383 |
setup_xstate_features(); /* * Setup init_xstate_buf to represent the init state of * all the features managed by the xsave */ |
10340ae13
|
384 385 |
init_xstate_buf = alloc_bootmem_align(xstate_size, __alignof__(struct xsave_struct)); |
dc1e35c6e
|
386 |
init_xstate_buf->i387.mxcsr = MXCSR_DEFAULT; |
a1488f8bf
|
387 |
|
29104e101
|
388 389 390 391 392 393 394 395 396 397 398 |
clts(); /* * Init all the features state with header_bv being 0x0 */ xrstor_state(init_xstate_buf, -1); /* * Dump the init state again. This is to identify the init state * of any feature which is not represented by all zero's. */ xsave_state(init_xstate_buf, -1); stts(); |
dc1e35c6e
|
399 400 401 402 403 |
} /* * Enable and initialize the xsave feature. */ |
1cff92d8f
|
404 |
static void __init xstate_enable_boot_cpu(void) |
dc1e35c6e
|
405 406 |
{ unsigned int eax, ebx, ecx, edx; |
ee813d53a
|
407 408 409 410 411 412 413 |
if (boot_cpu_data.cpuid_level < XSTATE_CPUID) { WARN(1, KERN_ERR "XSTATE_CPUID missing "); return; } cpuid_count(XSTATE_CPUID, 0, &eax, &ebx, &ecx, &edx); |
6152e4b1c
|
414 |
pcntxt_mask = eax + ((u64)edx << 32); |
dc1e35c6e
|
415 |
|
6152e4b1c
|
416 417 418 419 |
if ((pcntxt_mask & XSTATE_FPSSE) != XSTATE_FPSSE) { printk(KERN_ERR "FP/SSE not shown under xsave features 0x%llx ", pcntxt_mask); |
dc1e35c6e
|
420 421 422 423 |
BUG(); } /* |
a30469e79
|
424 |
* Support only the state known to OS. |
dc1e35c6e
|
425 |
*/ |
6152e4b1c
|
426 |
pcntxt_mask = pcntxt_mask & XCNTXT_MASK; |
97e80a70d
|
427 |
|
1cff92d8f
|
428 |
xstate_enable(); |
dc1e35c6e
|
429 430 431 432 |
/* * Recompute the context size for enabled features */ |
ee813d53a
|
433 |
cpuid_count(XSTATE_CPUID, 0, &eax, &ebx, &ecx, &edx); |
dc1e35c6e
|
434 |
xstate_size = ebx; |
5b3efd500
|
435 |
update_regset_xstate_info(xstate_size, pcntxt_mask); |
c37b5efea
|
436 |
prepare_fx_sw_frame(); |
dc1e35c6e
|
437 |
setup_xstate_init(); |
6152e4b1c
|
438 |
printk(KERN_INFO "xsave/xrstor: enabled xstate_bv 0x%llx, " |
dc1e35c6e
|
439 440 |
"cntxt size 0x%x ", |
6152e4b1c
|
441 |
pcntxt_mask, xstate_size); |
dc1e35c6e
|
442 |
} |
82d4150ce
|
443 |
|
1cff92d8f
|
444 445 446 447 448 449 450 |
/* * For the very first instance, this calls xstate_enable_boot_cpu(); * for all subsequent instances, this calls xstate_enable(). * * This is somewhat obfuscated due to the lack of powerful enough * overrides for the section checks. */ |
82d4150ce
|
451 452 |
void __cpuinit xsave_init(void) { |
1cff92d8f
|
453 454 |
static __refdata void (*next_func)(void) = xstate_enable_boot_cpu; void (*this_func)(void); |
0e49bf66d
|
455 456 |
if (!cpu_has_xsave) return; |
1cff92d8f
|
457 458 459 |
this_func = next_func; next_func = xstate_enable; this_func(); |
82d4150ce
|
460 |
} |