From c18e02387b0628a9ecfc41a65af4802e8b95357e Mon Sep 17 00:00:00 2001 From: Drasko DRASKOVIC Date: Thu, 7 Jul 2011 17:59:13 +0200 Subject: [PATCH] mips32: Sync Caches to Make Instr Writes Effective Pprogram that loads another program into memory is actually writing the D- side cache. The instructions it has loaded can't be executed until they reach the I-cache. After the instructions have been written, the loader should arrange to write back any containing D-cache line and invalidate any locations already in the I-cache. For the MIPS Architecture Release2 cores, we can use synci command that does this job. For Release1 we must use "cache" instruction. --- src/target/mips32.h | 112 ++++++++++++++---- src/target/mips32_pracc.c | 239 +++++++++++++++++++++++++++++++++++++- 2 files changed, 322 insertions(+), 29 deletions(-) diff --git a/src/target/mips32.h b/src/target/mips32.h index 8b21b0ac3..5d5aa4650 100644 --- a/src/target/mips32.h +++ b/src/target/mips32.h @@ -31,6 +31,38 @@ #define MIPS32_COMMON_MAGIC 0xB320B320 +/** + * Memory segments (32bit kernel mode addresses) + * These are the traditional names used in the 32-bit universe. + */ +#define KUSEG 0x00000000 +#define KSEG0 0x80000000 +#define KSEG1 0xa0000000 +#define KSEG2 0xc0000000 +#define KSEG3 0xe0000000 + +/** Returns the kernel segment base of a given address */ +#define KSEGX(a) ((a) & 0xe0000000) + +/** CP0 CONFIG regites fields */ +#define MIPS32_CONFIG0_KU_SHIFT 25 +#define MIPS32_CONFIG0_KU_MASK (0x7 << MIPS32_CONFIG0_KU_SHIFT) + +#define MIPS32_CONFIG0_K0_SHIFT 0 +#define MIPS32_CONFIG0_K0_MASK (0x7 << MIPS32_CONFIG0_K0_SHIFT) + +#define MIPS32_CONFIG0_K23_SHIFT 28 +#define MIPS32_CONFIG0_K23_MASK (0x7 << MIPS32_CONFIG0_K23_SHIFT) + +#define MIPS32_CONFIG0_AR_SHIFT 10 +#define MIPS32_CONFIG0_AR_MASK (0x7 << MIPS32_CONFIG0_AR_SHIFT) + +#define MIPS32_CONFIG1_DL_SHIFT 10 +#define MIPS32_CONFIG1_DL_MASK (0x7 << MIPS32_CONFIG1_DL_SHIFT) + +#define MIPS32_ARCH_REL1 0x0 +#define MIPS32_ARCH_REL2 0x1 + /* offsets into mips32 core register cache */ enum { @@ -95,10 +127,14 @@ struct mips32_algorithm enum mips32_isa_mode isa_mode; }; +#define MIPS32_OP_ADDIU 0x21 +#define MIPS32_OP_ANDI 0x0C #define MIPS32_OP_BEQ 0x04 +#define MIPS32_OP_BGTZ 0x07 #define MIPS32_OP_BNE 0x05 #define MIPS32_OP_ADDI 0x08 #define MIPS32_OP_AND 0x24 +#define MIPS32_OP_CACHE 0x2F #define MIPS32_OP_COP0 0x10 #define MIPS32_OP_JR 0x08 #define MIPS32_OP_LUI 0x0F @@ -109,12 +145,21 @@ struct mips32_algorithm #define MIPS32_OP_MTHI 0x11 #define MIPS32_OP_MFLO 0x12 #define MIPS32_OP_MTLO 0x13 +#define MIPS32_OP_RDHWR 0x3B #define MIPS32_OP_SB 0x28 #define MIPS32_OP_SH 0x29 #define MIPS32_OP_SW 0x2B #define MIPS32_OP_ORI 0x0D #define MIPS32_OP_XOR 0x26 +#define MIPS32_OP_SLTU 0x2B #define MIPS32_OP_SRL 0x03 +#define MIPS32_OP_SYNCI 0x1F + +#define MIPS32_OP_REGIMM 0x01 +#define MIPS32_OP_SDBBP 0x3F +#define MIPS32_OP_SPECIAL 0x00 +#define MIPS32_OP_SPECIAL2 0x07 +#define MIPS32_OP_SPECIAL3 0x1F #define MIPS32_COP0_MF 0x00 #define MIPS32_COP0_MT 0x04 @@ -123,33 +168,52 @@ struct mips32_algorithm #define MIPS32_I_INST(opcode, rs, rt, immd) (((opcode) << 26) |((rs) << 21) | ((rt) << 16) | (immd)) #define MIPS32_J_INST(opcode, addr) (((opcode) << 26) |(addr)) -#define MIPS32_NOP 0 -#define MIPS32_ADDI(tar, src, val) MIPS32_I_INST(MIPS32_OP_ADDI, src, tar, val) -#define MIPS32_AND(reg, off, val) MIPS32_R_INST(0, off, val, reg, 0, MIPS32_OP_AND) -#define MIPS32_B(off) MIPS32_BEQ(0, 0, off) -#define MIPS32_BEQ(src,tar,off) MIPS32_I_INST(MIPS32_OP_BEQ, src, tar, off) -#define MIPS32_BNE(src,tar,off) MIPS32_I_INST(MIPS32_OP_BNE, src, tar, off) -#define MIPS32_JR(reg) MIPS32_R_INST(0, reg, 0, 0, 0, MIPS32_OP_JR) -#define MIPS32_MFC0(gpr, cpr, sel) MIPS32_R_INST(MIPS32_OP_COP0, MIPS32_COP0_MF, gpr, cpr, 0, sel) -#define MIPS32_MTC0(gpr,cpr, sel) MIPS32_R_INST(MIPS32_OP_COP0, MIPS32_COP0_MT, gpr, cpr, 0, sel) -#define MIPS32_LBU(reg, off, base) MIPS32_I_INST(MIPS32_OP_LBU, base, reg, off) -#define MIPS32_LHU(reg, off, base) MIPS32_I_INST(MIPS32_OP_LHU, base, reg, off) -#define MIPS32_LUI(reg, val) MIPS32_I_INST(MIPS32_OP_LUI, 0, reg, val) -#define MIPS32_LW(reg, off, base) MIPS32_I_INST(MIPS32_OP_LW, base, reg, off) -#define MIPS32_MFLO(reg) MIPS32_R_INST(0, 0, 0, reg, 0, MIPS32_OP_MFLO) -#define MIPS32_MFHI(reg) MIPS32_R_INST(0, 0, 0, reg, 0, MIPS32_OP_MFHI) -#define MIPS32_MTLO(reg) MIPS32_R_INST(0, reg, 0, 0, 0, MIPS32_OP_MTLO) -#define MIPS32_MTHI(reg) MIPS32_R_INST(0, reg, 0, 0, 0, MIPS32_OP_MTHI) -#define MIPS32_ORI(tar, src, val) MIPS32_I_INST(MIPS32_OP_ORI, src, tar, val) -#define MIPS32_SB(reg, off, base) MIPS32_I_INST(MIPS32_OP_SB, base, reg, off) -#define MIPS32_SH(reg, off, base) MIPS32_I_INST(MIPS32_OP_SH, base, reg, off) -#define MIPS32_SW(reg, off, base) MIPS32_I_INST(MIPS32_OP_SW, base, reg, off) -#define MIPS32_XOR(reg, val1, val2) MIPS32_R_INST(0, val1, val2, reg, 0, MIPS32_OP_XOR) -#define MIPS32_SRL(reg, src, off) MIPS32_R_INST(0, 0, src, reg, off, MIPS32_OP_SRL) +#define MIPS32_NOP 0 +#define MIPS32_ADDI(tar, src, val) MIPS32_I_INST(MIPS32_OP_ADDI, src, tar, val) +#define MIPS32_ADDU(dst, src, tar) MIPS32_R_INST(MIPS32_OP_SPECIAL, src, tar, dst, 0, MIPS32_OP_ADDIU) +#define MIPS32_AND(reg, off, val) MIPS32_R_INST(0, off, val, reg, 0, MIPS32_OP_AND) +#define MIPS32_ANDI(tar, src, val) MIPS32_I_INST(MIPS32_OP_ANDI, src, tar, val) +#define MIPS32_B(off) MIPS32_BEQ(0, 0, off) +#define MIPS32_BEQ(src, tar, off) MIPS32_I_INST(MIPS32_OP_BEQ, src, tar, off) +#define MIPS32_BGTZ(reg, off) MIPS32_I_INST(MIPS32_OP_BGTZ, reg, 0, off) +#define MIPS32_BNE(src,tar,off) MIPS32_I_INST(MIPS32_OP_BNE, src, tar, off) +#define MIPS32_CACHE(op, off, base) MIPS32_I_INST(MIPS32_OP_CACHE, base, op, off) +#define MIPS32_JR(reg) MIPS32_R_INST(0, reg, 0, 0, 0, MIPS32_OP_JR) +#define MIPS32_MFC0(gpr, cpr, sel) MIPS32_R_INST(MIPS32_OP_COP0, MIPS32_COP0_MF, gpr, cpr, 0, sel) +#define MIPS32_MTC0(gpr, cpr, sel) MIPS32_R_INST(MIPS32_OP_COP0, MIPS32_COP0_MT, gpr, cpr, 0, sel) +#define MIPS32_LBU(reg, off, base) MIPS32_I_INST(MIPS32_OP_LBU, base, reg, off) +#define MIPS32_LHU(reg, off, base) MIPS32_I_INST(MIPS32_OP_LHU, base, reg, off) +#define MIPS32_LUI(reg, val) MIPS32_I_INST(MIPS32_OP_LUI, 0, reg, val) +#define MIPS32_LW(reg, off, base) MIPS32_I_INST(MIPS32_OP_LW, base, reg, off) +#define MIPS32_MFLO(reg) MIPS32_R_INST(0, 0, 0, reg, 0, MIPS32_OP_MFLO) +#define MIPS32_MFHI(reg) MIPS32_R_INST(0, 0, 0, reg, 0, MIPS32_OP_MFHI) +#define MIPS32_MTLO(reg) MIPS32_R_INST(0, reg, 0, 0, 0, MIPS32_OP_MTLO) +#define MIPS32_MTHI(reg) MIPS32_R_INST(0, reg, 0, 0, 0, MIPS32_OP_MTHI) +#define MIPS32_ORI(tar, src, val) MIPS32_I_INST(MIPS32_OP_ORI, src, tar, val) +#define MIPS32_RDHWR(tar, dst) MIPS32_R_INST(MIPS32_OP_SPECIAL3, 0, tar, dst, 0, MIPS32_OP_RDHWR) +#define MIPS32_SB(reg, off, base) MIPS32_I_INST(MIPS32_OP_SB, base, reg, off) +#define MIPS32_SH(reg, off, base) MIPS32_I_INST(MIPS32_OP_SH, base, reg, off) +#define MIPS32_SW(reg, off, base) MIPS32_I_INST(MIPS32_OP_SW, base, reg, off) +#define MIPS32_XOR(reg, val1, val2) MIPS32_R_INST(0, val1, val2, reg, 0, MIPS32_OP_XOR) +#define MIPS32_SRL(reg, src, off) MIPS32_R_INST(0, 0, src, reg, off, MIPS32_OP_SRL) +#define MIPS32_SLTU(dst, src, tar) MIPS32_R_INST(MIPS32_OP_SPECIAL, src, tar, dst, 0, MIPS32_OP_SLTU) +#define MIPS32_SYNCI(off, base) MIPS32_I_INST(MIPS32_OP_REGIMM, base, MIPS32_OP_SYNCI, off) + +#define MIPS32_SYNC 0xF +#define MIPS32_SYNCI_STEP 0x1 /* reg num od address step size to be used with synci instruction */ + +/** + * Cache operations definietions + * Operation field is 5 bits long : + * 1) bits 1..0 hold cache type + * 2) bits 4..2 hold operation code + */ +#define MIPS32_CACHE_D_HIT_WRITEBACK ((0x1 << 0) | (0x6 << 2)) +#define MIPS32_CACHE_I_HIT_INVALIDATE ((0x0 << 0) | (0x4 << 2)) /* ejtag specific instructions */ #define MIPS32_DRET 0x4200001F -#define MIPS32_SDBBP 0x7000003F +#define MIPS32_SDBBP 0x7000003F /* MIPS32_J_INST(MIPS32_OP_SPECIAL2, MIPS32_OP_SDBBP) */ #define MIPS16_SDBBP 0xE801 extern const struct command_registration mips32_command_handlers[]; diff --git a/src/target/mips32_pracc.c b/src/target/mips32_pracc.c index 1986e5ca2..6b43479fc 100644 --- a/src/target/mips32_pracc.c +++ b/src/target/mips32_pracc.c @@ -113,6 +113,11 @@ static int mips32_pracc_write_mem32(struct mips_ejtag *ejtag_info, static int mips32_pracc_write_u32(struct mips_ejtag *ejtag_info, uint32_t addr, uint32_t *buf); +static int mips32_pracc_sync_cache(struct mips_ejtag *ejtag_info, + uint32_t start_addr, uint32_t end_addr); +static int mips32_pracc_clean_invalidate_cache(struct mips_ejtag *ejtag_info, + uint32_t start_addr, uint32_t end_addr); + static int wait_for_pracc_rw(struct mips_ejtag *ejtag_info, uint32_t *ctrl) { uint32_t ejtag_ctrl; @@ -663,22 +668,246 @@ int mips32_cp0_write(struct mips_ejtag *ejtag_info, return retval; } +/** + * \b mips32_pracc_sync_cache + * + * Synchronize Caches to Make Instruction Writes Effective + * (ref. doc. MIPS32 Architecture For Programmers Volume II: The MIPS32 Instruction Set, + * Document Number: MD00086, Revision 2.00, June 9, 2003) + * + * When the instruction stream is written, the SYNCI instruction should be used + * in conjunction with other instructions to make the newly-written instructions effective. + * + * Explanation : + * A program that loads another program into memory is actually writing the D- side cache. + * The instructions it has loaded can't be executed until they reach the I-cache. + * + * After the instructions have been written, the loader should arrange + * to write back any containing D-cache line and invalidate any locations + * already in the I-cache. + * + * You can do that with cache instructions, but those instructions are only available in kernel mode, + * and a loader writing instructions for the use of its own process need not be privileged software. + * + * In the latest MIPS32/64 CPUs, MIPS provides the synci instruction, + * which does the whole job for a cache-line-sized chunk of the memory you just loaded: + * That is, it arranges a D-cache write-back and an I-cache invalidate. + * + * To employ synci at user level, you need to know the size of a cache line, + * and that can be obtained with a rdhwr SYNCI_Step + * from one of the standard “hardware registers”. + */ +static int mips32_pracc_sync_cache(struct mips_ejtag *ejtag_info, + uint32_t start_addr, uint32_t end_addr) +{ + static const uint32_t code[] = { + /* start: */ + MIPS32_MTC0(15,31,0), /* move $15 to COP0 DeSave */ + MIPS32_LUI(15,UPPER16(MIPS32_PRACC_STACK)), /* $15 = MIPS32_PRACC_STACK */ + MIPS32_ORI(15,15,LOWER16(MIPS32_PRACC_STACK)), + MIPS32_SW(8,0,15), /* sw $8,($15) */ + MIPS32_SW(9,0,15), /* sw $9,($15) */ + MIPS32_SW(10,0,15), /* sw $10,($15) */ + MIPS32_SW(11,0,15), /* sw $11,($15) */ + + MIPS32_LUI(8,UPPER16(MIPS32_PRACC_PARAM_IN)), /* $8 = MIPS32_PRACC_PARAM_IN */ + MIPS32_ORI(8,8,LOWER16(MIPS32_PRACC_PARAM_IN)), + MIPS32_LW(9,0,8), /* Load write start_addr to $9 */ + MIPS32_LW(10,4,8), /* Load write end_addr to $10 */ + + MIPS32_RDHWR(11, MIPS32_SYNCI_STEP), /* $11 = MIPS32_SYNCI_STEP */ + MIPS32_BEQ(11,0,6), /* beq $11, $0, end */ + MIPS32_NOP, + /* synci_loop : */ + MIPS32_SYNCI(0,9), /* synci 0($9) */ + MIPS32_SLTU(8,10,9), /* sltu $8, $10, $9 # $8 = $10 < $9 ? 1 : 0 */ + MIPS32_BNE(8,0,NEG16(3)), /* bne $8, $0, synci_loop */ + MIPS32_ADDU(9, 9, 11), /* $9 += MIPS32_SYNCI_STEP */ + MIPS32_SYNC, + /* end: */ + MIPS32_LW(11,0,15), /* lw $11,($15) */ + MIPS32_LW(10,0,15), /* lw $10,($15) */ + MIPS32_LW(9,0,15), /* lw $9,($15) */ + MIPS32_LW(8,0,15), /* lw $8,($15) */ + MIPS32_B(NEG16(24)), /* b start */ + MIPS32_MFC0(15,31,0), /* move COP0 DeSave to $15 */ + }; + + /* TODO remove array */ + uint32_t *param_in = malloc(2 * sizeof(uint32_t)); + int retval; + param_in[0] = start_addr; + param_in[1] = end_addr; + + retval = mips32_pracc_exec(ejtag_info, ARRAY_SIZE(code), code, 2, param_in, 0, NULL, 1); + + free(param_in); + + return retval; +} + +/** + * \b mips32_pracc_clean_invalidate_cache + * + * Writeback D$ and Invalidate I$ + * so that the instructions written can be visible to CPU + */ +static int mips32_pracc_clean_invalidate_cache(struct mips_ejtag *ejtag_info, + uint32_t start_addr, uint32_t end_addr) +{ + static const uint32_t code[] = { + /* start: */ + MIPS32_MTC0(15,31,0), /* move $15 to COP0 DeSave */ + MIPS32_LUI(15,UPPER16(MIPS32_PRACC_STACK)), /* $15 = MIPS32_PRACC_STACK */ + MIPS32_ORI(15,15,LOWER16(MIPS32_PRACC_STACK)), + MIPS32_SW(8,0,15), /* sw $8,($15) */ + MIPS32_SW(9,0,15), /* sw $9,($15) */ + MIPS32_SW(10,0,15), /* sw $10,($15) */ + MIPS32_SW(11,0,15), /* sw $11,($15) */ + + MIPS32_LUI(8,UPPER16(MIPS32_PRACC_PARAM_IN)), /* $8 = MIPS32_PRACC_PARAM_IN */ + MIPS32_ORI(8,8,LOWER16(MIPS32_PRACC_PARAM_IN)), + MIPS32_LW(9,0,8), /* Load write start_addr to $9 */ + MIPS32_LW(10,4,8), /* Load write end_addr to $10 */ + MIPS32_LW(11,8,8), /* Load write clsiz to $11 */ + + /* cache_loop: */ + MIPS32_SLTU(8,10,9), /* sltu $8, $10, $9 : $8 <- $10 < $9 ? */ + MIPS32_BGTZ(8,6), /* bgtz $8, end */ + MIPS32_NOP, + + MIPS32_CACHE(MIPS32_CACHE_D_HIT_WRITEBACK,0,9), /* cache Hit_Writeback_D, 0($9) */ + MIPS32_CACHE(MIPS32_CACHE_I_HIT_INVALIDATE,0,9), /* cache Hit_Invalidate_I, 0($9) */ + + MIPS32_ADDU(9,9,11), /* $9 += $11 */ + + MIPS32_B(NEG16(7)), /* b cache_loop */ + MIPS32_NOP, + /* end: */ + MIPS32_LW(11,0,15), /* lw $11,($15) */ + MIPS32_LW(10,0,15), /* lw $10,($15) */ + MIPS32_LW(9,0,15), /* lw $9,($15) */ + MIPS32_LW(8,0,15), /* lw $8,($15) */ + MIPS32_B(NEG16(25)), /* b start */ + MIPS32_MFC0(15,31,0), /* move COP0 DeSave to $15 */ + }; + + /** + * Find cache line size in bytes + */ + uint32_t conf; + uint32_t dl, clsiz; + + mips32_cp0_read(ejtag_info, &conf, 16, 1); + dl = (conf & MIPS32_CONFIG1_DL_MASK) >> MIPS32_CONFIG1_DL_SHIFT; + + /* dl encoding : dl=1 => 4 bytes, dl=2 => 8 bytes, etc... */ + clsiz = 0x2 << dl; + + /* TODO remove array */ + uint32_t *param_in = malloc(3 * sizeof(uint32_t)); + int retval; + param_in[0] = start_addr; + param_in[1] = end_addr; + param_in[2] = clsiz; + + retval = mips32_pracc_exec(ejtag_info, ARRAY_SIZE(code), code, 3, param_in, 0, NULL, 1); + + free(param_in); + + return retval; +} + + int mips32_pracc_write_mem(struct mips_ejtag *ejtag_info, uint32_t addr, int size, int count, void *buf) { + int retval; + switch (size) { case 1: - return mips32_pracc_write_mem8(ejtag_info, addr, count, (uint8_t*)buf); + retval = mips32_pracc_write_mem8(ejtag_info, addr, count, (uint8_t*)buf); + break; case 2: - return mips32_pracc_write_mem16(ejtag_info, addr, count,(uint16_t*)buf); + retval = mips32_pracc_write_mem16(ejtag_info, addr, count,(uint16_t*)buf); + break; case 4: if (count == 1) - return mips32_pracc_write_u32(ejtag_info, addr, (uint32_t*)buf); + { + retval = mips32_pracc_write_u32(ejtag_info, addr, (uint32_t*)buf); + } else - return mips32_pracc_write_mem32(ejtag_info, addr, count, (uint32_t*)buf); + { + retval = mips32_pracc_write_mem32(ejtag_info, addr, count, (uint32_t*)buf); + } + break; + default: + retval = ERROR_FAIL; } - return ERROR_OK; + /** + * If we are in the cachable regoion and cache is activated, + * we must clean D$ + invalidate I$ after we did the write, + * so that changes do not continue to live only in D$, but to be + * replicated in I$ also (maybe we wrote the istructions) + */ + uint32_t conf = 0; + int cached = 0; + + mips32_cp0_read(ejtag_info, &conf, 16, 0); + + switch (KSEGX(addr)) + { + case KUSEG: + cached = (conf & MIPS32_CONFIG0_KU_MASK) >> MIPS32_CONFIG0_KU_SHIFT; + break; + case KSEG0 : + cached = (conf & MIPS32_CONFIG0_K0_MASK) >> MIPS32_CONFIG0_K0_SHIFT; + break; + case KSEG1: + /* uncachable segment - nothing to do */ + break; + case KSEG2: + case KSEG3: + cached = (conf & MIPS32_CONFIG0_K23_MASK) >> MIPS32_CONFIG0_K23_SHIFT; + break; + default: + /* what ? */ + break; + } + + /** + * Check cachablitiy bits coherency algorithm - + * is the region cacheable or uncached. + * If cacheable we have to synchronize the cache + */ + if (cached == 0x3) + { + uint32_t start_addr, end_addr; + uint32_t rel; + + start_addr = addr; + end_addr = addr + count * size; + + /** select cache synchronisation mechanism based on Architecture Release */ + rel = (conf & MIPS32_CONFIG0_AR_MASK) >> MIPS32_CONFIG0_AR_SHIFT; + switch (rel) + { + case MIPS32_ARCH_REL1 : + /* MIPS32/64 Release 1 - we must use cache instruction */ + mips32_pracc_clean_invalidate_cache(ejtag_info, start_addr, end_addr); + break; + case MIPS32_ARCH_REL2 : + /* MIPS32/64 Release 2 - we can use synci instruction */ + mips32_pracc_sync_cache(ejtag_info, start_addr, end_addr); + break; + default : + /* what ? */ + break; + } + } + + return retval; } static int mips32_pracc_write_mem32(struct mips_ejtag *ejtag_info, uint32_t addr, int count, uint32_t *buf)