| 1 | /* SPDX-License-Identifier: GPL-2.0-or-later */ |
| 2 | /* |
| 3 | * This file contains miscellaneous low-level functions. |
| 4 | * Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org) |
| 5 | * |
| 6 | * Largely rewritten by Cort Dougan (cort@cs.nmt.edu) |
| 7 | * and Paul Mackerras. |
| 8 | * |
| 9 | */ |
| 10 | |
| 11 | #include <linux/export.h> |
| 12 | #include <linux/sys.h> |
| 13 | #include <asm/unistd.h> |
| 14 | #include <asm/errno.h> |
| 15 | #include <asm/reg.h> |
| 16 | #include <asm/page.h> |
| 17 | #include <asm/cache.h> |
| 18 | #include <asm/cputable.h> |
| 19 | #include <asm/mmu.h> |
| 20 | #include <asm/ppc_asm.h> |
| 21 | #include <asm/thread_info.h> |
| 22 | #include <asm/asm-offsets.h> |
| 23 | #include <asm/processor.h> |
| 24 | #include <asm/bug.h> |
| 25 | #include <asm/ptrace.h> |
| 26 | #include <asm/feature-fixups.h> |
| 27 | |
| 28 | .text |
| 29 | |
| 30 | /* |
| 31 | * reloc_got2 runs through the .got2 section adding an offset |
| 32 | * to each entry. |
| 33 | */ |
| 34 | _GLOBAL(reloc_got2) |
| 35 | mflr r11 |
| 36 | lis r7,__got2_start@ha |
| 37 | addi r7,r7,__got2_start@l |
| 38 | lis r8,__got2_end@ha |
| 39 | addi r8,r8,__got2_end@l |
| 40 | subf r8,r7,r8 |
| 41 | srwi. r8,r8,2 |
| 42 | beqlr |
| 43 | mtctr r8 |
| 44 | bcl 20,31,$+4 |
| 45 | 1: mflr r0 |
| 46 | lis r4,1b@ha |
| 47 | addi r4,r4,1b@l |
| 48 | subf r0,r4,r0 |
| 49 | add r7,r0,r7 |
| 50 | 2: lwz r0,0(r7) |
| 51 | add r0,r0,r3 |
| 52 | stw r0,0(r7) |
| 53 | addi r7,r7,4 |
| 54 | bdnz 2b |
| 55 | mtlr r11 |
| 56 | blr |
| 57 | |
| 58 | /* |
| 59 | * call_setup_cpu - call the setup_cpu function for this cpu |
| 60 | * r3 = data offset, r24 = cpu number |
| 61 | * |
| 62 | * Setup function is called with: |
| 63 | * r3 = data offset |
| 64 | * r4 = ptr to CPU spec (relocated) |
| 65 | */ |
| 66 | _GLOBAL(call_setup_cpu) |
| 67 | addis r4,r3,cur_cpu_spec@ha |
| 68 | addi r4,r4,cur_cpu_spec@l |
| 69 | lwz r4,0(r4) |
| 70 | add r4,r4,r3 |
| 71 | lwz r5,CPU_SPEC_SETUP(r4) |
| 72 | cmpwi 0,r5,0 |
| 73 | add r5,r5,r3 |
| 74 | beqlr |
| 75 | mtctr r5 |
| 76 | bctr |
| 77 | |
| 78 | #if defined(CONFIG_CPU_FREQ_PMAC) && defined(CONFIG_PPC_BOOK3S_32) |
| 79 | |
| 80 | /* This gets called by via-pmu.c to switch the PLL selection |
| 81 | * on 750fx CPU. This function should really be moved to some |
| 82 | * other place (as most of the cpufreq code in via-pmu |
| 83 | */ |
| 84 | _GLOBAL(low_choose_750fx_pll) |
| 85 | /* Clear MSR:EE */ |
| 86 | mfmsr r7 |
| 87 | rlwinm r0,r7,0,17,15 |
| 88 | mtmsr r0 |
| 89 | |
| 90 | /* If switching to PLL1, disable HID0:BTIC */ |
| 91 | cmplwi cr0,r3,0 |
| 92 | beq 1f |
| 93 | mfspr r5,SPRN_HID0 |
| 94 | rlwinm r5,r5,0,27,25 |
| 95 | sync |
| 96 | mtspr SPRN_HID0,r5 |
| 97 | isync |
| 98 | sync |
| 99 | |
| 100 | 1: |
| 101 | /* Calc new HID1 value */ |
| 102 | mfspr r4,SPRN_HID1 /* Build a HID1:PS bit from parameter */ |
| 103 | rlwinm r5,r3,16,15,15 /* Clear out HID1:PS from value read */ |
| 104 | rlwinm r4,r4,0,16,14 /* Could have I used rlwimi here ? */ |
| 105 | or r4,r4,r5 |
| 106 | mtspr SPRN_HID1,r4 |
| 107 | |
| 108 | #ifdef CONFIG_SMP |
| 109 | /* Store new HID1 image */ |
| 110 | lwz r6,TASK_CPU(r2) |
| 111 | slwi r6,r6,2 |
| 112 | #else |
| 113 | li r6, 0 |
| 114 | #endif |
| 115 | addis r6,r6,nap_save_hid1@ha |
| 116 | stw r4,nap_save_hid1@l(r6) |
| 117 | |
| 118 | /* If switching to PLL0, enable HID0:BTIC */ |
| 119 | cmplwi cr0,r3,0 |
| 120 | bne 1f |
| 121 | mfspr r5,SPRN_HID0 |
| 122 | ori r5,r5,HID0_BTIC |
| 123 | sync |
| 124 | mtspr SPRN_HID0,r5 |
| 125 | isync |
| 126 | sync |
| 127 | |
| 128 | 1: |
| 129 | /* Return */ |
| 130 | mtmsr r7 |
| 131 | blr |
| 132 | |
| 133 | _GLOBAL(low_choose_7447a_dfs) |
| 134 | /* Clear MSR:EE */ |
| 135 | mfmsr r7 |
| 136 | rlwinm r0,r7,0,17,15 |
| 137 | mtmsr r0 |
| 138 | |
| 139 | /* Calc new HID1 value */ |
| 140 | mfspr r4,SPRN_HID1 |
| 141 | insrwi r4,r3,1,9 /* insert parameter into bit 9 */ |
| 142 | sync |
| 143 | mtspr SPRN_HID1,r4 |
| 144 | sync |
| 145 | isync |
| 146 | |
| 147 | /* Return */ |
| 148 | mtmsr r7 |
| 149 | blr |
| 150 | |
| 151 | #endif /* CONFIG_CPU_FREQ_PMAC && CONFIG_PPC_BOOK3S_32 */ |
| 152 | |
| 153 | /* |
| 154 | * Copy a whole page. We use the dcbz instruction on the destination |
| 155 | * to reduce memory traffic (it eliminates the unnecessary reads of |
| 156 | * the destination into cache). This requires that the destination |
| 157 | * is cacheable. |
| 158 | */ |
| 159 | #define COPY_16_BYTES \ |
| 160 | lwz r6,4(r4); \ |
| 161 | lwz r7,8(r4); \ |
| 162 | lwz r8,12(r4); \ |
| 163 | lwzu r9,16(r4); \ |
| 164 | stw r6,4(r3); \ |
| 165 | stw r7,8(r3); \ |
| 166 | stw r8,12(r3); \ |
| 167 | stwu r9,16(r3) |
| 168 | |
| 169 | _GLOBAL(copy_page) |
| 170 | rlwinm r5, r3, 0, L1_CACHE_BYTES - 1 |
| 171 | addi r3,r3,-4 |
| 172 | |
| 173 | 0: twnei r5, 0 /* WARN if r3 is not cache aligned */ |
| 174 | EMIT_WARN_ENTRY 0b,__FILE__,__LINE__, BUGFLAG_WARNING |
| 175 | |
| 176 | addi r4,r4,-4 |
| 177 | |
| 178 | li r5,4 |
| 179 | |
| 180 | #if MAX_COPY_PREFETCH > 1 |
| 181 | li r0,MAX_COPY_PREFETCH |
| 182 | li r11,4 |
| 183 | mtctr r0 |
| 184 | 11: dcbt r11,r4 |
| 185 | addi r11,r11,L1_CACHE_BYTES |
| 186 | bdnz 11b |
| 187 | #else /* MAX_COPY_PREFETCH == 1 */ |
| 188 | dcbt r5,r4 |
| 189 | li r11,L1_CACHE_BYTES+4 |
| 190 | #endif /* MAX_COPY_PREFETCH */ |
| 191 | li r0,PAGE_SIZE/L1_CACHE_BYTES - MAX_COPY_PREFETCH |
| 192 | crclr 4*cr0+eq |
| 193 | 2: |
| 194 | mtctr r0 |
| 195 | 1: |
| 196 | dcbt r11,r4 |
| 197 | dcbz r5,r3 |
| 198 | COPY_16_BYTES |
| 199 | #if L1_CACHE_BYTES >= 32 |
| 200 | COPY_16_BYTES |
| 201 | #if L1_CACHE_BYTES >= 64 |
| 202 | COPY_16_BYTES |
| 203 | COPY_16_BYTES |
| 204 | #if L1_CACHE_BYTES >= 128 |
| 205 | COPY_16_BYTES |
| 206 | COPY_16_BYTES |
| 207 | COPY_16_BYTES |
| 208 | COPY_16_BYTES |
| 209 | #endif |
| 210 | #endif |
| 211 | #endif |
| 212 | bdnz 1b |
| 213 | beqlr |
| 214 | crnot 4*cr0+eq,4*cr0+eq |
| 215 | li r0,MAX_COPY_PREFETCH |
| 216 | li r11,4 |
| 217 | b 2b |
| 218 | EXPORT_SYMBOL(copy_page) |
| 219 | |
| 220 | /* |
| 221 | * Extended precision shifts. |
| 222 | * |
| 223 | * Updated to be valid for shift counts from 0 to 63 inclusive. |
| 224 | * -- Gabriel |
| 225 | * |
| 226 | * R3/R4 has 64 bit value |
| 227 | * R5 has shift count |
| 228 | * result in R3/R4 |
| 229 | * |
| 230 | * ashrdi3: arithmetic right shift (sign propagation) |
| 231 | * lshrdi3: logical right shift |
| 232 | * ashldi3: left shift |
| 233 | */ |
| 234 | _GLOBAL(__ashrdi3) |
| 235 | subfic r6,r5,32 |
| 236 | srw r4,r4,r5 # LSW = count > 31 ? 0 : LSW >> count |
| 237 | addi r7,r5,32 # could be xori, or addi with -32 |
| 238 | slw r6,r3,r6 # t1 = count > 31 ? 0 : MSW << (32-count) |
| 239 | rlwinm r8,r7,0,32 # t3 = (count < 32) ? 32 : 0 |
| 240 | sraw r7,r3,r7 # t2 = MSW >> (count-32) |
| 241 | or r4,r4,r6 # LSW |= t1 |
| 242 | slw r7,r7,r8 # t2 = (count < 32) ? 0 : t2 |
| 243 | sraw r3,r3,r5 # MSW = MSW >> count |
| 244 | or r4,r4,r7 # LSW |= t2 |
| 245 | blr |
| 246 | EXPORT_SYMBOL(__ashrdi3) |
| 247 | |
| 248 | _GLOBAL(__ashldi3) |
| 249 | subfic r6,r5,32 |
| 250 | slw r3,r3,r5 # MSW = count > 31 ? 0 : MSW << count |
| 251 | addi r7,r5,32 # could be xori, or addi with -32 |
| 252 | srw r6,r4,r6 # t1 = count > 31 ? 0 : LSW >> (32-count) |
| 253 | slw r7,r4,r7 # t2 = count < 32 ? 0 : LSW << (count-32) |
| 254 | or r3,r3,r6 # MSW |= t1 |
| 255 | slw r4,r4,r5 # LSW = LSW << count |
| 256 | or r3,r3,r7 # MSW |= t2 |
| 257 | blr |
| 258 | EXPORT_SYMBOL(__ashldi3) |
| 259 | |
| 260 | _GLOBAL(__lshrdi3) |
| 261 | subfic r6,r5,32 |
| 262 | srw r4,r4,r5 # LSW = count > 31 ? 0 : LSW >> count |
| 263 | addi r7,r5,32 # could be xori, or addi with -32 |
| 264 | slw r6,r3,r6 # t1 = count > 31 ? 0 : MSW << (32-count) |
| 265 | srw r7,r3,r7 # t2 = count < 32 ? 0 : MSW >> (count-32) |
| 266 | or r4,r4,r6 # LSW |= t1 |
| 267 | srw r3,r3,r5 # MSW = MSW >> count |
| 268 | or r4,r4,r7 # LSW |= t2 |
| 269 | blr |
| 270 | EXPORT_SYMBOL(__lshrdi3) |
| 271 | |
| 272 | /* |
| 273 | * 64-bit comparison: __cmpdi2(s64 a, s64 b) |
| 274 | * Returns 0 if a < b, 1 if a == b, 2 if a > b. |
| 275 | */ |
| 276 | _GLOBAL(__cmpdi2) |
| 277 | cmpw r3,r5 |
| 278 | li r3,1 |
| 279 | bne 1f |
| 280 | cmplw r4,r6 |
| 281 | beqlr |
| 282 | 1: li r3,0 |
| 283 | bltlr |
| 284 | li r3,2 |
| 285 | blr |
| 286 | EXPORT_SYMBOL(__cmpdi2) |
| 287 | /* |
| 288 | * 64-bit comparison: __ucmpdi2(u64 a, u64 b) |
| 289 | * Returns 0 if a < b, 1 if a == b, 2 if a > b. |
| 290 | */ |
| 291 | _GLOBAL(__ucmpdi2) |
| 292 | cmplw r3,r5 |
| 293 | li r3,1 |
| 294 | bne 1f |
| 295 | cmplw r4,r6 |
| 296 | beqlr |
| 297 | 1: li r3,0 |
| 298 | bltlr |
| 299 | li r3,2 |
| 300 | blr |
| 301 | EXPORT_SYMBOL(__ucmpdi2) |
| 302 | |
| 303 | _GLOBAL(__bswapdi2) |
| 304 | rotlwi r9,r4,8 |
| 305 | rotlwi r10,r3,8 |
| 306 | rlwimi r9,r4,24,0,7 |
| 307 | rlwimi r10,r3,24,0,7 |
| 308 | rlwimi r9,r4,24,16,23 |
| 309 | rlwimi r10,r3,24,16,23 |
| 310 | mr r3,r9 |
| 311 | mr r4,r10 |
| 312 | blr |
| 313 | EXPORT_SYMBOL(__bswapdi2) |
| 314 | |
| 315 | #ifdef CONFIG_SMP |
| 316 | _GLOBAL(start_secondary_resume) |
| 317 | /* Reset stack */ |
| 318 | rlwinm r1, r1, 0, 0, 31 - THREAD_SHIFT |
| 319 | addi r1,r1,THREAD_SIZE-STACK_FRAME_MIN_SIZE |
| 320 | li r3,0 |
| 321 | stw r3,0(r1) /* Zero the stack frame pointer */ |
| 322 | bl start_secondary |
| 323 | b . |
| 324 | #endif /* CONFIG_SMP */ |
| 325 | |