cpufunc.h revision 18265
1/*- 2 * Copyright (c) 1993 The Regents of the University of California. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * $Id: cpufunc.h,v 1.54 1996/08/01 20:29:28 wollman Exp $ 34 */ 35 36/* 37 * Functions to provide access to special i386 instructions. 38 */ 39 40#ifndef _MACHINE_CPUFUNC_H_ 41#define _MACHINE_CPUFUNC_H_ 42 43#include <sys/cdefs.h> 44#include <sys/types.h> 45 46#ifdef __GNUC__ 47 48static __inline void 49breakpoint(void) 50{ 51 __asm __volatile("int $3"); 52} 53 54static __inline void 55disable_intr(void) 56{ 57 __asm __volatile("cli" : : : "memory"); 58} 59 60static __inline void 61enable_intr(void) 62{ 63 __asm __volatile("sti"); 64} 65 66#define HAVE_INLINE_FFS 67 68static __inline int 69ffs(int mask) 70{ 71 int result; 72 /* 73 * bsfl turns out to be not all that slow on 486's. It can beaten 74 * using a binary search to reduce to 4 bits and then a table lookup, 75 * but only if the code is inlined and in the cache, and the code 76 * is quite large so inlining it probably busts the cache. 77 * 78 * Note that gcc-2's builtin ffs would be used if we didn't declare 79 * this inline or turn off the builtin. The builtin is faster but 80 * broken in gcc-2.4.5 and slower but working in gcc-2.5 and 2.6. 81 */ 82 __asm __volatile("testl %0,%0; je 1f; bsfl %0,%0; incl %0; 1:" 83 : "=r" (result) : "0" (mask)); 84 return (result); 85} 86 87#define HAVE_INLINE_FLS 88 89static __inline int 90fls(int mask) 91{ 92 int result; 93 __asm __volatile("testl %0,%0; je 1f; bsrl %0,%0; incl %0; 1:" 94 : "=r" (result) : "0" (mask)); 95 return (result); 96} 97 98#if __GNUC__ < 2 99 100#define inb(port) inbv(port) 101#define outb(port, data) outbv(port, data) 102 103#else /* __GNUC >= 2 */ 104 105/* 106 * The following complications are to get around gcc not having a 107 * constraint letter for the range 0..255. We still put "d" in the 108 * constraint because "i" isn't a valid constraint when the port 109 * isn't constant. This only matters for -O0 because otherwise 110 * the non-working version gets optimized away. 111 * 112 * Use an expression-statement instead of a conditional expression 113 * because gcc-2.6.0 would promote the operands of the conditional 114 * and produce poor code for "if ((inb(var) & const1) == const2)". 115 */ 116#define inb(port) ({ \ 117 u_char _data; \ 118 if (__builtin_constant_p((int) (port)) && (port) < 256ul) \ 119 _data = inbc(port); \ 120 else \ 121 _data = inbv(port); \ 122 _data; }) 123 124#define outb(port, data) \ 125 (__builtin_constant_p((int) (port)) && (port) < 256ul \ 126 ? outbc(port, data) : outbv(port, data)) 127 128static __inline u_char 129inbc(u_int port) 130{ 131 u_char data; 132 133 __asm __volatile("inb %1,%0" : "=a" (data) : "id" ((u_short)(port))); 134 return (data); 135} 136 137static __inline void 138outbc(u_int port, u_char data) 139{ 140 __asm __volatile("outb %0,%1" : : "a" (data), "id" ((u_short)(port))); 141} 142 143#endif /* __GNUC <= 2 */ 144 145static __inline u_char 146inbv(u_int port) 147{ 148 u_char data; 149 /* 150 * We use %%dx and not %1 here because i/o is done at %dx and not at 151 * %edx, while gcc generates inferior code (movw instead of movl) 152 * if we tell it to load (u_short) port. 153 */ 154 __asm __volatile("inb %%dx,%0" : "=a" (data) : "d" (port)); 155 return (data); 156} 157 158static __inline u_long 159inl(u_int port) 160{ 161 u_long data; 162 163 __asm __volatile("inl %%dx,%0" : "=a" (data) : "d" (port)); 164 return (data); 165} 166 167static __inline void 168insb(u_int port, void *addr, size_t cnt) 169{ 170 __asm __volatile("cld; rep; insb" 171 : : "d" (port), "D" (addr), "c" (cnt) 172 : "di", "cx", "memory"); 173} 174 175static __inline void 176insw(u_int port, void *addr, size_t cnt) 177{ 178 __asm __volatile("cld; rep; insw" 179 : : "d" (port), "D" (addr), "c" (cnt) 180 : "di", "cx", "memory"); 181} 182 183static __inline void 184insl(u_int port, void *addr, size_t cnt) 185{ 186 __asm __volatile("cld; rep; insl" 187 : : "d" (port), "D" (addr), "c" (cnt) 188 : "di", "cx", "memory"); 189} 190 191static __inline u_short 192inw(u_int port) 193{ 194 u_short data; 195 196 __asm __volatile("inw %%dx,%0" : "=a" (data) : "d" (port)); 197 return (data); 198} 199 200static __inline u_int 201loadandclear(u_int *addr) 202{ 203 u_int result; 204 205 __asm __volatile("xorl %0,%0; xchgl %1,%0" 206 : "=&r" (result) : "m" (*addr)); 207 return (result); 208} 209 210static __inline void 211outbv(u_int port, u_char data) 212{ 213 u_char al; 214 /* 215 * Use an unnecessary assignment to help gcc's register allocator. 216 * This make a large difference for gcc-1.40 and a tiny difference 217 * for gcc-2.6.0. For gcc-1.40, al had to be ``asm("ax")'' for 218 * best results. gcc-2.6.0 can't handle this. 219 */ 220 al = data; 221 __asm __volatile("outb %0,%%dx" : : "a" (al), "d" (port)); 222} 223 224static __inline void 225outl(u_int port, u_long data) 226{ 227 /* 228 * outl() and outw() aren't used much so we haven't looked at 229 * possible micro-optimizations such as the unnecessary 230 * assignment for them. 231 */ 232 __asm __volatile("outl %0,%%dx" : : "a" (data), "d" (port)); 233} 234 235static __inline void 236outsb(u_int port, void *addr, size_t cnt) 237{ 238 __asm __volatile("cld; rep; outsb" 239 : : "d" (port), "S" (addr), "c" (cnt) 240 : "si", "cx"); 241} 242 243static __inline void 244outsw(u_int port, void *addr, size_t cnt) 245{ 246 __asm __volatile("cld; rep; outsw" 247 : : "d" (port), "S" (addr), "c" (cnt) 248 : "si", "cx"); 249} 250 251static __inline void 252outsl(u_int port, void *addr, size_t cnt) 253{ 254 __asm __volatile("cld; rep; outsl" 255 : : "d" (port), "S" (addr), "c" (cnt) 256 : "si", "cx"); 257} 258 259static __inline void 260outw(u_int port, u_short data) 261{ 262 __asm __volatile("outw %0,%%dx" : : "a" (data), "d" (port)); 263} 264 265 266static __inline void 267pmap_update(void) 268{ 269 u_long temp; 270 /* 271 * This should be implemented as load_cr3(rcr3()) when load_cr3() 272 * is inlined. 273 */ 274 __asm __volatile("movl %%cr3, %0; movl %0, %%cr3" : "=r" (temp) 275 : : "memory"); 276} 277 278static __inline u_long 279rcr2(void) 280{ 281 u_long data; 282 283 __asm __volatile("movl %%cr2,%0" : "=r" (data)); 284 return (data); 285} 286 287static __inline u_long 288read_eflags(void) 289{ 290 u_long ef; 291 292 __asm __volatile("pushfl; popl %0" : "=r" (ef)); 293 return (ef); 294} 295 296static __inline quad_t 297rdmsr(u_int msr) 298{ 299 quad_t rv; 300 301 __asm __volatile(".byte 0x0f, 0x32" : "=A" (rv) : "c" (msr)); 302 return (rv); 303} 304 305static __inline quad_t 306rdpmc(u_int pmc) 307{ 308 quad_t rv; 309 310 __asm __volatile(".byte 0x0f, 0x33" : "=A" (rv) : "c" (pmc)); 311 return (rv); 312} 313 314static __inline quad_t 315rdtsc(void) 316{ 317 quad_t rv; 318 319 __asm __volatile(".byte 0x0f, 0x31" : "=A" (rv)); 320 return (rv); 321} 322 323static __inline void 324setbits(volatile unsigned *addr, u_int bits) 325{ 326 __asm __volatile("orl %1,%0" : "=m" (*addr) : "ir" (bits)); 327} 328 329static __inline void 330write_eflags(u_long ef) 331{ 332 __asm __volatile("pushl %0; popfl" : : "r" (ef)); 333} 334 335static __inline void 336wrmsr(u_int msr, quad_t newval) 337{ 338 __asm __volatile(".byte 0x0f, 0x30" : : "A" (newval), "c" (msr)); 339} 340 341#else /* !__GNUC__ */ 342 343int breakpoint __P((void)); 344void disable_intr __P((void)); 345void enable_intr __P((void)); 346u_char inb __P((u_int port)); 347u_long inl __P((u_int port)); 348void insb __P((u_int port, void *addr, size_t cnt)); 349void insl __P((u_int port, void *addr, size_t cnt)); 350void insw __P((u_int port, void *addr, size_t cnt)); 351u_short inw __P((u_int port)); 352u_int loadandclear __P((u_int *addr)); 353void outb __P((u_int port, u_char data)); 354void outl __P((u_int port, u_long data)); 355void outsb __P((u_int port, void *addr, size_t cnt)); 356void outsl __P((u_int port, void *addr, size_t cnt)); 357void outsw __P((u_int port, void *addr, size_t cnt)); 358void outw __P((u_int port, u_short data)); 359void pmap_update __P((void)); 360u_long rcr2 __P((void)); 361quad_t rdmsr __P((u_int msr)); 362quad_t rdpmc __P((u_int pmc)); 363quad_t rdtsc __P((void)); 364u_long read_eflags __P((void)); 365void setbits __P((volatile unsigned *addr, u_int bits)); 366void write_eflags __P((u_long ef)); 367void wrmsr __P((u_int msr, quad_t newval)); 368 369#endif /* __GNUC__ */ 370 371void load_cr0 __P((u_long cr0)); 372void load_cr3 __P((u_long cr3)); 373void ltr __P((u_short sel)); 374u_int rcr0 __P((void)); 375u_long rcr3 __P((void)); 376 377#include <machine/spl.h> /* XXX belongs elsewhere */ 378 379#endif /* !_MACHINE_CPUFUNC_H_ */ 380