| 1 | /*
|
|---|
| 2 | * Copyright (c) 2012 Jan Vesely
|
|---|
| 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 | *
|
|---|
| 9 | * - Redistributions of source code must retain the above copyright
|
|---|
| 10 | * notice, this list of conditions and the following disclaimer.
|
|---|
| 11 | * - Redistributions in binary form must reproduce the above copyright
|
|---|
| 12 | * notice, this list of conditions and the following disclaimer in the
|
|---|
| 13 | * documentation and/or other materials provided with the distribution.
|
|---|
| 14 | * - The name of the author may not be used to endorse or promote products
|
|---|
| 15 | * derived from this software without specific prior written permission.
|
|---|
| 16 | *
|
|---|
| 17 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
|
|---|
| 18 | * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
|
|---|
| 19 | * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
|
|---|
| 20 | * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
|
|---|
| 21 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
|
|---|
| 22 | * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|---|
| 23 | * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|---|
| 24 | * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|---|
| 25 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
|
|---|
| 26 | * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|---|
| 27 | */
|
|---|
| 28 |
|
|---|
| 29 | /** @addtogroup arm32
|
|---|
| 30 | * @{
|
|---|
| 31 | */
|
|---|
| 32 | /** @file
|
|---|
| 33 | * @brief arm32 FPU context
|
|---|
| 34 | */
|
|---|
| 35 |
|
|---|
| 36 | #include <fpu_context.h>
|
|---|
| 37 | #include <arch.h>
|
|---|
| 38 | #include <arch/types.h>
|
|---|
| 39 | #include <arch/security_ext.h>
|
|---|
| 40 | #include <arch/cp15.h>
|
|---|
| 41 | #include <cpu.h>
|
|---|
| 42 |
|
|---|
| 43 | #define FPSID_IMPLEMENTER(r) ((r) >> 24)
|
|---|
| 44 | #define FPSID_SW_ONLY_FLAG (1 << 23)
|
|---|
| 45 | #define FPSID_SUBACHITECTURE(r) (((r) >> 16) & 0x7f)
|
|---|
| 46 | #define FPSID_PART_NUMBER(r) (((r) >> 8) & 0xff)
|
|---|
| 47 | #define FPSID_VARIANT(r) (((r) >> 4) 0xf)
|
|---|
| 48 | #define FPSID_REVISION(r) (((r) >> 0) 0xf)
|
|---|
| 49 |
|
|---|
| 50 |
|
|---|
| 51 | enum {
|
|---|
| 52 | FPU_VFPv1 = 0x00,
|
|---|
| 53 | FPU_VFPv2_COMMONv1 = 0x01,
|
|---|
| 54 | FPU_VFPv3_COMMONv2 = 0x02,
|
|---|
| 55 | FPU_VFPv3_NO_COMMON = 0x3, /* Does not support fpu exc. traps */
|
|---|
| 56 | FPU_VFPv3_COMMONv3 = 0x4,
|
|---|
| 57 | };
|
|---|
| 58 |
|
|---|
| 59 | extern uint32_t fpsid_read(void);
|
|---|
| 60 | extern uint32_t mvfr0_read(void);
|
|---|
| 61 |
|
|---|
| 62 | enum {
|
|---|
| 63 | FPEXC_EX_FLAG = (1 << 31),
|
|---|
| 64 | FPEXC_ENABLED_FLAG = (1 << 30),
|
|---|
| 65 | };
|
|---|
| 66 | extern uint32_t fpexc_read(void);
|
|---|
| 67 | extern void fpexc_write(uint32_t);
|
|---|
| 68 |
|
|---|
| 69 | /** ARM Architecture Reference Manual ch. B4.1.58, p. B$-1551 */
|
|---|
| 70 | enum {
|
|---|
| 71 | FPSCR_N_FLAG = (1 << 31),
|
|---|
| 72 | FPSCR_Z_FLAG = (1 << 30),
|
|---|
| 73 | FPSCR_C_FLAG = (1 << 29),
|
|---|
| 74 | FPSCR_V_FLAG = (1 << 28),
|
|---|
| 75 | FPSCR_QC_FLAG = (1 << 27),
|
|---|
| 76 | FPSCR_AHP_FLAG = (1 << 26),
|
|---|
| 77 | FPSCR_DN_FLAG = (1 << 25),
|
|---|
| 78 | FPSCR_FZ_FLAG = (1 << 24),
|
|---|
| 79 | FPSCR_ROUND_MODE_MASK = (0x3 << 22),
|
|---|
| 80 | FPSCR_ROUND_TO_NEAREST = (0x0 << 22),
|
|---|
| 81 | FPSCR_ROUND_TO_POS_INF = (0x1 << 22),
|
|---|
| 82 | FPSCR_ROUND_TO_NEG_INF = (0x2 << 22),
|
|---|
| 83 | FPSCR_ROUND_TO_ZERO = (0x3 << 22),
|
|---|
| 84 | FPSCR_STRIDE_MASK = (0x3 << 20),
|
|---|
| 85 | FPSCR_STRIDE_SHIFT = 20,
|
|---|
| 86 | FPSCR_LEN_MASK = (0x7 << 16),
|
|---|
| 87 | FPSCR_LEN_SHIFT = 16,
|
|---|
| 88 | FPSCR_DENORMAL_EN_FLAG = (1 << 15),
|
|---|
| 89 | FPSCR_INEXACT_EN_FLAG = (1 << 12),
|
|---|
| 90 | FPSCR_UNDERFLOW_EN_FLAG = (1 << 11),
|
|---|
| 91 | FPSCR_OVERFLOW_EN_FLAG = (1 << 10),
|
|---|
| 92 | FPSCR_ZERO_DIV_EN_FLAG = (1 << 9),
|
|---|
| 93 | FPSCR_INVALID_OP_EN_FLAG = (1 << 8),
|
|---|
| 94 | FPSCR_DENORMAL_FLAG = (1 << 7),
|
|---|
| 95 | FPSCR_INEXACT_FLAG = (1 << 4),
|
|---|
| 96 | FPSCR_UNDERFLOW_FLAG = (1 << 3),
|
|---|
| 97 | FPSCR_OVERLOW_FLAG = (1 << 2),
|
|---|
| 98 | FPSCR_DIV_ZERO_FLAG = (1 << 1),
|
|---|
| 99 | FPSCR_INVALID_OP_FLAG = (1 << 0),
|
|---|
| 100 |
|
|---|
| 101 | FPSCR_EN_ALL = FPSCR_DENORMAL_EN_FLAG | FPSCR_INEXACT_EN_FLAG | FPSCR_UNDERFLOW_EN_FLAG | FPSCR_OVERFLOW_EN_FLAG | FPSCR_ZERO_DIV_EN_FLAG | FPSCR_INVALID_OP_EN_FLAG,
|
|---|
| 102 | };
|
|---|
| 103 |
|
|---|
| 104 | extern uint32_t fpscr_read(void);
|
|---|
| 105 | extern void fpscr_write(uint32_t);
|
|---|
| 106 |
|
|---|
| 107 | extern void fpu_context_save_s32(fpu_context_t *);
|
|---|
| 108 | extern void fpu_context_restore_s32(fpu_context_t *);
|
|---|
| 109 | extern void fpu_context_save_d16(fpu_context_t *);
|
|---|
| 110 | extern void fpu_context_restore_d16(fpu_context_t *);
|
|---|
| 111 | extern void fpu_context_save_d32(fpu_context_t *);
|
|---|
| 112 | extern void fpu_context_restore_d32(fpu_context_t *);
|
|---|
| 113 |
|
|---|
| 114 | static void (*save_context)(fpu_context_t *ctx);
|
|---|
| 115 | static void (*restore_context)(fpu_context_t *ctx);
|
|---|
| 116 |
|
|---|
| 117 | static int fpu_have_coprocessor_access(void)
|
|---|
| 118 | {
|
|---|
| 119 | /*
|
|---|
| 120 | * The register containing the information (CPACR) is not available on armv6-
|
|---|
| 121 | * rely on user decision to use CONFIG_FPU.
|
|---|
| 122 | */
|
|---|
| 123 | #ifdef PROCESSOR_ARCH_armv7_a
|
|---|
| 124 | const uint32_t cpacr = CPACR_read();
|
|---|
| 125 | /* FPU needs access to coprocessor 10 and 11.
|
|---|
| 126 | * Moreover they need to have same access enabled */
|
|---|
| 127 | if (((cpacr & CPACR_CP_MASK(10)) != CPACR_CP_FULL_ACCESS(10)) &&
|
|---|
| 128 | ((cpacr & CPACR_CP_MASK(11)) != CPACR_CP_FULL_ACCESS(11))) {
|
|---|
| 129 | printf("No access to CP10 and CP11: %" PRIx32 "\n", cpacr);
|
|---|
| 130 | return 0;
|
|---|
| 131 | }
|
|---|
| 132 | #endif
|
|---|
| 133 | return 1;
|
|---|
| 134 | }
|
|---|
| 135 |
|
|---|
| 136 | /** Enable coprocessor access. Turn both non-secure mode bit and generic access.
|
|---|
| 137 | * Cortex A8 Manual says:
|
|---|
| 138 | * "You must execute an Instruction Memory Barrier (IMB) sequence immediately
|
|---|
| 139 | * after an update of the Coprocessor Access Control Register, see Memory
|
|---|
| 140 | * Barriers in the ARM Architecture Reference Manual. You must not attempt to
|
|---|
| 141 | * execute any instructions that are affected by the change of access rights
|
|---|
| 142 | * between the IMB sequence and the register update."
|
|---|
| 143 | * Cortex a8 TRM ch. 3.2.27. c1, Coprocessor Access Control Register
|
|---|
| 144 | *
|
|---|
| 145 | * @note do we need to call secure monitor here?
|
|---|
| 146 | */
|
|---|
| 147 | static void fpu_enable_coprocessor_access(void)
|
|---|
| 148 | {
|
|---|
| 149 | /*
|
|---|
| 150 | * The register containing the information (CPACR) is not available on armv6-
|
|---|
| 151 | * rely on user decision to use CONFIG_FPU.
|
|---|
| 152 | */
|
|---|
| 153 | #ifdef PROCESSOR_ARCH_armv7_a
|
|---|
| 154 | /* Allow coprocessor access */
|
|---|
| 155 | uint32_t cpacr = CPACR_read();
|
|---|
| 156 | /* FPU needs access to coprocessor 10 and 11.
|
|---|
| 157 | * Moreover, they need to have same access enabled */
|
|---|
| 158 | cpacr &= ~(CPACR_CP_MASK(10) | CPACR_CP_MASK(11));
|
|---|
| 159 | cpacr |= (CPACR_CP_FULL_ACCESS(10) | CPACR_CP_FULL_ACCESS(11));
|
|---|
| 160 | CPACR_write(cpacr);
|
|---|
| 161 | #endif
|
|---|
| 162 | }
|
|---|
| 163 |
|
|---|
| 164 |
|
|---|
| 165 | void fpu_init(void)
|
|---|
| 166 | {
|
|---|
| 167 | /* Check if we have access */
|
|---|
| 168 | if (!fpu_have_coprocessor_access())
|
|---|
| 169 | return;
|
|---|
| 170 |
|
|---|
| 171 | /* Clear all fpu flags */
|
|---|
| 172 | fpexc_write(0);
|
|---|
| 173 | fpu_enable();
|
|---|
| 174 | /* Mask all exception traps,
|
|---|
| 175 | * The bits are RAZ/WI on archs that don't support fpu exc traps.
|
|---|
| 176 | */
|
|---|
| 177 | fpscr_write(fpscr_read() & ~FPSCR_EN_ALL);
|
|---|
| 178 | }
|
|---|
| 179 |
|
|---|
| 180 | void fpu_setup(void)
|
|---|
| 181 | {
|
|---|
| 182 | /* Enable coprocessor access*/
|
|---|
| 183 | fpu_enable_coprocessor_access();
|
|---|
| 184 |
|
|---|
| 185 | /* Check if we succeeded */
|
|---|
| 186 | if (!fpu_have_coprocessor_access())
|
|---|
| 187 | return;
|
|---|
| 188 |
|
|---|
| 189 | const uint32_t fpsid = fpsid_read();
|
|---|
| 190 | if (fpsid & FPSID_SW_ONLY_FLAG) {
|
|---|
| 191 | printf("No FPU avaiable\n");
|
|---|
| 192 | return;
|
|---|
| 193 | }
|
|---|
| 194 | switch (FPSID_SUBACHITECTURE(fpsid)) {
|
|---|
| 195 | case FPU_VFPv1:
|
|---|
| 196 | printf("Detected VFPv1\n");
|
|---|
| 197 | save_context = fpu_context_save_s32;
|
|---|
| 198 | restore_context = fpu_context_restore_s32;
|
|---|
| 199 | break;
|
|---|
| 200 | case FPU_VFPv2_COMMONv1:
|
|---|
| 201 | printf("Detected VFPv2\n");
|
|---|
| 202 | save_context = fpu_context_save_d16;
|
|---|
| 203 | restore_context = fpu_context_restore_d16;
|
|---|
| 204 | break;
|
|---|
| 205 | case FPU_VFPv3_COMMONv2:
|
|---|
| 206 | case FPU_VFPv3_NO_COMMON:
|
|---|
| 207 | case FPU_VFPv3_COMMONv3: {
|
|---|
| 208 | const uint32_t mvfr0 = mvfr0_read();
|
|---|
| 209 | /* See page B4-1637 */
|
|---|
| 210 | if ((mvfr0 & 0xf) == 0x1) {
|
|---|
| 211 | printf("Detected VFPv3+ with 16 regs\n");
|
|---|
| 212 | save_context = fpu_context_save_d16;
|
|---|
| 213 | restore_context = fpu_context_restore_d16;
|
|---|
| 214 | } else {
|
|---|
| 215 | printf("Detected VFPv3+ with 32 regs\n");
|
|---|
| 216 | save_context = fpu_context_save_d32;
|
|---|
| 217 | restore_context = fpu_context_restore_d32;
|
|---|
| 218 | }
|
|---|
| 219 | break;
|
|---|
| 220 | }
|
|---|
| 221 |
|
|---|
| 222 | }
|
|---|
| 223 | }
|
|---|
| 224 |
|
|---|
| 225 | bool handle_if_fpu_exception(void)
|
|---|
| 226 | {
|
|---|
| 227 | /* Check if we have access */
|
|---|
| 228 | if (!fpu_have_coprocessor_access())
|
|---|
| 229 | return false;
|
|---|
| 230 |
|
|---|
| 231 | const uint32_t fpexc = fpexc_read();
|
|---|
| 232 | if (fpexc & FPEXC_ENABLED_FLAG) {
|
|---|
| 233 | const uint32_t fpscr = fpscr_read();
|
|---|
| 234 | printf("FPU exception\n"
|
|---|
| 235 | "\tFPEXC: %" PRIx32 " FPSCR: %" PRIx32 "\n", fpexc, fpscr);
|
|---|
| 236 | return false;
|
|---|
| 237 | }
|
|---|
| 238 | #ifdef CONFIG_FPU_LAZY
|
|---|
| 239 | scheduler_fpu_lazy_request();
|
|---|
| 240 | return true;
|
|---|
| 241 | #else
|
|---|
| 242 | return false;
|
|---|
| 243 | #endif
|
|---|
| 244 | }
|
|---|
| 245 |
|
|---|
| 246 | void fpu_enable(void)
|
|---|
| 247 | {
|
|---|
| 248 | /* Check if we have access */
|
|---|
| 249 | if (!fpu_have_coprocessor_access())
|
|---|
| 250 | return;
|
|---|
| 251 | /* Enable FPU instructions */
|
|---|
| 252 | fpexc_write(fpexc_read() | FPEXC_ENABLED_FLAG);
|
|---|
| 253 | }
|
|---|
| 254 |
|
|---|
| 255 | void fpu_disable(void)
|
|---|
| 256 | {
|
|---|
| 257 | /* Check if we have access */
|
|---|
| 258 | if (!fpu_have_coprocessor_access())
|
|---|
| 259 | return;
|
|---|
| 260 | /* Disable FPU instructions */
|
|---|
| 261 | fpexc_write(fpexc_read() & ~FPEXC_ENABLED_FLAG);
|
|---|
| 262 | }
|
|---|
| 263 |
|
|---|
| 264 | void fpu_context_save(fpu_context_t *ctx)
|
|---|
| 265 | {
|
|---|
| 266 | /* This is only necessary if we enable fpu exceptions. */
|
|---|
| 267 | #if 0
|
|---|
| 268 | const uint32_t fpexc = fpexc_read();
|
|---|
| 269 |
|
|---|
| 270 | if (fpexc & FPEXC_EX_FLAG) {
|
|---|
| 271 | printf("EX FPU flag is on, things will fail\n");
|
|---|
| 272 | //TODO implement common subarch context saving
|
|---|
| 273 | }
|
|---|
| 274 | #endif
|
|---|
| 275 | if (save_context)
|
|---|
| 276 | save_context(ctx);
|
|---|
| 277 | }
|
|---|
| 278 |
|
|---|
| 279 | void fpu_context_restore(fpu_context_t *ctx)
|
|---|
| 280 | {
|
|---|
| 281 | if (restore_context)
|
|---|
| 282 | restore_context(ctx);
|
|---|
| 283 | }
|
|---|