src/kpx_cpu/sheepshaver_glue.cpp

/*
 *  sheepshaver_glue.cpp - Glue Kheperix CPU to SheepShaver CPU engine interface
 *
 *  SheepShaver (C) 1997-2004 Christian Bauer and Marc Hellwig
 *
 *  This program is free software; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License as published by
 *  the Free Software Foundation; either version 2 of the License, or
 *  (at your option) any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; if not, write to the Free Software
 *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */

#include "sysdeps.h"
#include "cpu_emulation.h"
#include "main.h"
#include "prefs.h"
#include "xlowmem.h"
#include "emul_op.h"
#include "rom_patches.h"
#include "macos_util.h"
#include "block-alloc.hpp"
#include "sigsegv.h"
#include "cpu/ppc/ppc-cpu.hpp"
#include "cpu/ppc/ppc-operations.hpp"
#include "cpu/ppc/ppc-instructions.hpp"
#include "thunks.h"

// Used for NativeOp trampolines
#include "video.h"
#include "name_registry.h"
#include "serial.h"
#include "ether.h"
#include "timer.h"

#include <stdio.h>
#include <stdlib.h>

#if ENABLE_MON
#include "mon.h"
#include "mon_disass.h"
#endif

#define DEBUG 0
#include "debug.h"

// Emulation time statistics
#define EMUL_TIME_STATS 1

#if EMUL_TIME_STATS
static clock_t emul_start_time;
static uint32 interrupt_count = 0;
static clock_t interrupt_time = 0;
static uint32 exec68k_count = 0;
static clock_t exec68k_time = 0;
static uint32 native_exec_count = 0;
static clock_t native_exec_time = 0;
static uint32 macos_exec_count = 0;
static clock_t macos_exec_time = 0;
#endif

static void enter_mon(void)
{
        // Start up mon in real-mode
#if ENABLE_MON
        char *arg[4] = {"mon", "-m", "-r", NULL};
        mon(3, arg);
#endif
}

// From main_*.cpp
extern uintptr SignalStackBase();

// From rsrc_patches.cpp
extern "C" void check_load_invoc(uint32 type, int16 id, uint32 h);

// PowerPC EmulOp to exit from emulation looop
const uint32 POWERPC_EXEC_RETURN = POWERPC_EMUL_OP | 1;

// Enable multicore (main/interrupts) cpu emulation?
#define MULTICORE_CPU (ASYNC_IRQ ? 1 : 0)

// Enable interrupt routine safety checks?
#define SAFE_INTERRUPT_PPC 1

// Enable Execute68k() safety checks?
#define SAFE_EXEC_68K 1

// Save FP state in Execute68k()?
#define SAVE_FP_EXEC_68K 1

// Interrupts in EMUL_OP mode?
#define INTERRUPTS_IN_EMUL_OP_MODE 1

// Interrupts in native mode?
#define INTERRUPTS_IN_NATIVE_MODE 1

// Enable native EMUL_OPs to be run without a mode switch
#define ENABLE_NATIVE_EMUL_OP 1

// Pointer to Kernel Data
static KernelData * const kernel_data = (KernelData *)KERNEL_DATA_BASE;

// SIGSEGV handler
static sigsegv_return_t sigsegv_handler(sigsegv_address_t, sigsegv_address_t);

#if PPC_ENABLE_JIT && PPC_REENTRANT_JIT
// Special trampolines for EmulOp and NativeOp
static uint8 *emul_op_trampoline;
static uint8 *native_op_trampoline;
#endif

// JIT Compiler enabled?
static inline bool enable_jit_p()
{
        return PrefsFindBool("jit");
}


/**
 *              PowerPC emulator glue with special 'sheep' opcodes
 **/

enum {
        PPC_I(SHEEP) = PPC_I(MAX),
        PPC_I(SHEEP_MAX)
};

class sheepshaver_cpu
        : public powerpc_cpu
{
        void init_decoder();
        void execute_sheep(uint32 opcode);

        // Filter out EMUL_OP routines that only call native code
        bool filter_execute_emul_op(uint32 emul_op);

        // "Native" EMUL_OP routines
        void execute_emul_op_microseconds();
        void execute_emul_op_idle_time_1();
        void execute_emul_op_idle_time_2();

public:

        // Constructor
        sheepshaver_cpu();

        // CR & XER accessors
        uint32 get_cr() const           { return cr().get(); }
        void set_cr(uint32 v)           { cr().set(v); }
        uint32 get_xer() const          { return xer().get(); }
        void set_xer(uint32 v)          { xer().set(v); }

        // Execute NATIVE_OP routine
        void execute_native_op(uint32 native_op);

        // Execute EMUL_OP routine
        void execute_emul_op(uint32 emul_op);

        // Execute 68k routine
        void execute_68k(uint32 entry, M68kRegisters *r);

        // Execute ppc routine
        void execute_ppc(uint32 entry);

        // Execute MacOS/PPC code
        uint32 execute_macos_code(uint32 tvect, int nargs, uint32 const *args);

        // Compile one instruction
        virtual int compile1(codegen_context_t & cg_context);

        // Resource manager thunk
        void get_resource(uint32 old_get_resource);

        // Handle MacOS interrupt
        void interrupt(uint32 entry);
        void handle_interrupt();

        // Make sure the SIGSEGV handler can access CPU registers
        friend sigsegv_return_t sigsegv_handler(sigsegv_address_t, sigsegv_address_t);
};

// Memory allocator returning areas aligned on 16-byte boundaries
void *operator new(size_t size)
{
        void *p;

#if defined(HAVE_POSIX_MEMALIGN)
        if (posix_memalign(&p, 16, size) != 0)
                throw std::bad_alloc();
#elif defined(HAVE_MEMALIGN)
        p = memalign(16, size);
#elif defined(HAVE_VALLOC)
        p = valloc(size); // page-aligned!
#else
        /* XXX: handle padding ourselves */
        p = malloc(size);
#endif

        return p;
}

void operator delete(void *p)
{
#if defined(HAVE_MEMALIGN) || defined(HAVE_VALLOC)
#if defined(__GLIBC__)
        // this is known to work only with GNU libc
        free(p);
#endif
#else
        free(p);
#endif
}

sheepshaver_cpu::sheepshaver_cpu()
        : powerpc_cpu(enable_jit_p())
{
        init_decoder();
}

void sheepshaver_cpu::init_decoder()
{
        static const instr_info_t sheep_ii_table[] = {
                { "sheep",
                  (execute_pmf)&sheepshaver_cpu::execute_sheep,
                  NULL,
                  PPC_I(SHEEP),
                  D_form, 6, 0, CFLOW_JUMP | CFLOW_TRAP
                }
        };

        const int ii_count = sizeof(sheep_ii_table)/sizeof(sheep_ii_table[0]);
        D(bug("SheepShaver extra decode table has %d entries\n", ii_count));

        for (int i = 0; i < ii_count; i++) {
                const instr_info_t * ii = &sheep_ii_table[i];
                init_decoder_entry(ii);
        }
}

/*              NativeOp instruction format:
                +------------+-------------------------+--+-----------+------------+
                |      6     |                         |FN|    OP     |      2     |
                +------------+-------------------------+--+-----------+------------+
                 0         5 |6                      18 19 20      25 26        31
*/

typedef bit_field< 19, 19 > FN_field;
typedef bit_field< 20, 25 > NATIVE_OP_field;
typedef bit_field< 26, 31 > EMUL_OP_field;

// "Native" EMUL_OP routines
#define GPR_A(REG) gpr(16 + (REG))
#define GPR_D(REG) gpr( 8 + (REG))

void sheepshaver_cpu::execute_emul_op_microseconds()
{
        Microseconds(GPR_A(0), GPR_D(0));
}

void sheepshaver_cpu::execute_emul_op_idle_time_1()
{
        // Sleep if no events pending
        if (ReadMacInt32(0x14c) == 0)
                Delay_usec(16667);
        GPR_A(0) = ReadMacInt32(0x2b6);
}

void sheepshaver_cpu::execute_emul_op_idle_time_2()
{
        // Sleep if no events pending
        if (ReadMacInt32(0x14c) == 0)
                Delay_usec(16667);
        GPR_D(0) = (uint32)-2;
}

// Filter out EMUL_OP routines that only call native code
bool sheepshaver_cpu::filter_execute_emul_op(uint32 emul_op)
{
        switch (emul_op) {
        case OP_MICROSECONDS:
                execute_emul_op_microseconds();
                return true;
        case OP_IDLE_TIME:
                execute_emul_op_idle_time_1();
                return true;
        case OP_IDLE_TIME_2:
                execute_emul_op_idle_time_2();
                return true;
        }
        return false;
}

// Execute EMUL_OP routine
void sheepshaver_cpu::execute_emul_op(uint32 emul_op)
{
#if ENABLE_NATIVE_EMUL_OP
        // First, filter out EMUL_OPs that can be executed without a mode switch
        if (filter_execute_emul_op(emul_op))
                return;
#endif

        M68kRegisters r68;
        WriteMacInt32(XLM_68K_R25, gpr(25));
        WriteMacInt32(XLM_RUN_MODE, MODE_EMUL_OP);
        for (int i = 0; i < 8; i++)
                r68.d[i] = gpr(8 + i);
        for (int i = 0; i < 7; i++)
                r68.a[i] = gpr(16 + i);
        r68.a[7] = gpr(1);
        uint32 saved_cr = get_cr() & CR_field<2>::mask();
        uint32 saved_xer = get_xer();
        EmulOp(&r68, gpr(24), emul_op);
        set_cr(saved_cr);
        set_xer(saved_xer);
        for (int i = 0; i < 8; i++)
                gpr(8 + i) = r68.d[i];
        for (int i = 0; i < 7; i++)
                gpr(16 + i) = r68.a[i];
        gpr(1) = r68.a[7];
        WriteMacInt32(XLM_RUN_MODE, MODE_68K);
}

// Execute SheepShaver instruction
void sheepshaver_cpu::execute_sheep(uint32 opcode)
{
//      D(bug("Extended opcode %08x at %08x (68k pc %08x)\n", opcode, pc(), gpr(24)));
        assert((((opcode >> 26) & 0x3f) == 6) && OP_MAX <= 64 + 3);

        switch (opcode & 0x3f) {
        case 0:         // EMUL_RETURN
                QuitEmulator();
                break;

        case 1:         // EXEC_RETURN
                spcflags().set(SPCFLAG_CPU_EXEC_RETURN);
                break;

        case 2:         // EXEC_NATIVE
                execute_native_op(NATIVE_OP_field::extract(opcode));
                if (FN_field::test(opcode))
                        pc() = lr();
                else
                        pc() += 4;
                break;

        default:        // EMUL_OP
                execute_emul_op(EMUL_OP_field::extract(opcode) - 3);
                pc() += 4;
                break;
        }
}

// Compile one instruction
int sheepshaver_cpu::compile1(codegen_context_t & cg_context)
{
#if PPC_ENABLE_JIT
        const instr_info_t *ii = cg_context.instr_info;
        if (ii->mnemo != PPC_I(SHEEP))
                return COMPILE_FAILURE;

        int status = COMPILE_FAILURE;
        powerpc_dyngen & dg = cg_context.codegen;
        uint32 opcode = cg_context.opcode;

        switch (opcode & 0x3f) {
        case 0:         // EMUL_RETURN
                dg.gen_invoke(QuitEmulator);
                status = COMPILE_CODE_OK;
                break;

        case 1:         // EXEC_RETURN
                dg.gen_spcflags_set(SPCFLAG_CPU_EXEC_RETURN);
                // Don't check for pending interrupts, we do know we have to
                // get out of this block ASAP
                dg.gen_exec_return();
                status = COMPILE_EPILOGUE_OK;
                break;

        case 2: {       // EXEC_NATIVE
                uint32 selector = NATIVE_OP_field::extract(opcode);
                switch (selector) {
#if !PPC_REENTRANT_JIT
                // Filter out functions that may invoke Execute68k() or
                // CallMacOS(), this would break reentrancy as they could
                // invalidate the translation cache and even overwrite
                // continuation code when we are done with them.
                case NATIVE_PATCH_NAME_REGISTRY:
                        dg.gen_invoke(DoPatchNameRegistry);
                        status = COMPILE_CODE_OK;
                        break;
                case NATIVE_VIDEO_INSTALL_ACCEL:
                        dg.gen_invoke(VideoInstallAccel);
                        status = COMPILE_CODE_OK;
                        break;
                case NATIVE_VIDEO_VBL:
                        dg.gen_invoke(VideoVBL);
                        status = COMPILE_CODE_OK;
                        break;
                case NATIVE_GET_RESOURCE:
                case NATIVE_GET_1_RESOURCE:
                case NATIVE_GET_IND_RESOURCE:
                case NATIVE_GET_1_IND_RESOURCE:
                case NATIVE_R_GET_RESOURCE: {
                        static const uint32 get_resource_ptr[] = {
                                XLM_GET_RESOURCE,
                                XLM_GET_1_RESOURCE,
                                XLM_GET_IND_RESOURCE,
                                XLM_GET_1_IND_RESOURCE,
                                XLM_R_GET_RESOURCE
                        };
                        uint32 old_get_resource = ReadMacInt32(get_resource_ptr[selector - NATIVE_GET_RESOURCE]);
                        typedef void (*func_t)(dyngen_cpu_base, uint32);
                        func_t func = (func_t)nv_mem_fun(&sheepshaver_cpu::get_resource).ptr();
                        dg.gen_invoke_CPU_im(func, old_get_resource);
                        status = COMPILE_CODE_OK;
                        break;
                }
                case NATIVE_CHECK_LOAD_INVOC:
                        dg.gen_load_T0_GPR(3);
                        dg.gen_load_T1_GPR(4);
                        dg.gen_se_16_32_T1();
                        dg.gen_load_T2_GPR(5);
                        dg.gen_invoke_T0_T1_T2((void (*)(uint32, uint32, uint32))check_load_invoc);
                        status = COMPILE_CODE_OK;
                        break;
#endif
                case NATIVE_DISABLE_INTERRUPT:
                        dg.gen_invoke(DisableInterrupt);
                        status = COMPILE_CODE_OK;
                        break;
                case NATIVE_ENABLE_INTERRUPT:
                        dg.gen_invoke(EnableInterrupt);
                        status = COMPILE_CODE_OK;
                        break;
                case NATIVE_BITBLT:
                        dg.gen_load_T0_GPR(3);
                        dg.gen_invoke_T0((void (*)(uint32))NQD_bitblt);
                        status = COMPILE_CODE_OK;
                        break;
                case NATIVE_INVRECT:
                        dg.gen_load_T0_GPR(3);
                        dg.gen_invoke_T0((void (*)(uint32))NQD_invrect);
                        status = COMPILE_CODE_OK;
                        break;
                case NATIVE_FILLRECT:
                        dg.gen_load_T0_GPR(3);
                        dg.gen_invoke_T0((void (*)(uint32))NQD_fillrect);
                        status = COMPILE_CODE_OK;
                        break;
                }
                // Could we fully translate this NativeOp?
                if (FN_field::test(opcode)) {
                        if (status != COMPILE_FAILURE) {
                                dg.gen_load_A0_LR();
                                dg.gen_set_PC_A0();
                        }
                        cg_context.done_compile = true;
                        break;
                }
                else if (status != COMPILE_FAILURE) {
                        cg_context.done_compile = false;
                        break;
                }
#if PPC_REENTRANT_JIT
                // Try to execute NativeOp trampoline
                dg.gen_set_PC_im(cg_context.pc + 4);
                dg.gen_mov_32_T0_im(selector);
                dg.gen_jmp(native_op_trampoline);
                cg_context.done_compile = true;
                status = COMPILE_EPILOGUE_OK;
                break;
#endif
                // Invoke NativeOp handler
                typedef void (*func_t)(dyngen_cpu_base, uint32);
                func_t func = (func_t)nv_mem_fun(&sheepshaver_cpu::execute_native_op).ptr();
                dg.gen_invoke_CPU_im(func, selector);
                cg_context.done_compile = false;
                status = COMPILE_CODE_OK;
                break;
        }

        default: {      // EMUL_OP
                uint32 emul_op = EMUL_OP_field::extract(opcode) - 3;
#if ENABLE_NATIVE_EMUL_OP
                typedef void (*emul_op_func_t)(dyngen_cpu_base);
                emul_op_func_t emul_op_func = 0;
                switch (emul_op) {
                case OP_MICROSECONDS:
                        emul_op_func = (emul_op_func_t)nv_mem_fun(&sheepshaver_cpu::execute_emul_op_microseconds).ptr();
                        break;
                case OP_IDLE_TIME:
                        emul_op_func = (emul_op_func_t)nv_mem_fun(&sheepshaver_cpu::execute_emul_op_idle_time_1).ptr();
                        break;
                case OP_IDLE_TIME_2:
                        emul_op_func = (emul_op_func_t)nv_mem_fun(&sheepshaver_cpu::execute_emul_op_idle_time_2).ptr();
                        break;
                }
                if (emul_op_func) {
                        dg.gen_invoke_CPU(emul_op_func);
                        cg_context.done_compile = false;
                        status = COMPILE_CODE_OK;
                        break;
                }
#endif
#if PPC_REENTRANT_JIT
                // Try to execute EmulOp trampoline
                dg.gen_set_PC_im(cg_context.pc + 4);
                dg.gen_mov_32_T0_im(emul_op);
                dg.gen_jmp(emul_op_trampoline);
                cg_context.done_compile = true;
                status = COMPILE_EPILOGUE_OK;
                break;
#endif
                // Invoke EmulOp handler
                typedef void (*func_t)(dyngen_cpu_base, uint32);
                func_t func = (func_t)nv_mem_fun(&sheepshaver_cpu::execute_emul_op).ptr();
                dg.gen_invoke_CPU_im(func, emul_op);
                cg_context.done_compile = false;
                status = COMPILE_CODE_OK;
                break;
        }
        }
        return status;
#endif
        return COMPILE_FAILURE;
}

// Handle MacOS interrupt
void sheepshaver_cpu::interrupt(uint32 entry)
{
#if EMUL_TIME_STATS
        interrupt_count++;
        const clock_t interrupt_start = clock();
#endif

#if SAFE_INTERRUPT_PPC
        static int depth = 0;
        if (depth != 0)
                printf("FATAL: sheepshaver_cpu::interrupt() called more than once: %d\n", depth);
        depth++;
#endif
#if SAFE_INTERRUPT_PPC >= 2
        uint32 saved_regs[32];
        memcpy(&saved_regs[0], &gpr(0), sizeof(saved_regs));
#endif

#if !MULTICORE_CPU
        // Save program counters and branch registers
        uint32 saved_pc = pc();
        uint32 saved_lr = lr();
        uint32 saved_ctr= ctr();
        uint32 saved_sp = gpr(1);
#endif

        // Initialize stack pointer to SheepShaver alternate stack base
        gpr(1) = SignalStackBase() - 64;

        // Build trampoline to return from interrupt
        SheepVar32 trampoline = POWERPC_EXEC_RETURN;

        // Prepare registers for nanokernel interrupt routine
        kernel_data->v[0x004 >> 2] = htonl(gpr(1));
        kernel_data->v[0x018 >> 2] = htonl(gpr(6));

        gpr(6) = ntohl(kernel_data->v[0x65c >> 2]);
        assert(gpr(6) != 0);
        WriteMacInt32(gpr(6) + 0x13c, gpr(7));
        WriteMacInt32(gpr(6) + 0x144, gpr(8));
        WriteMacInt32(gpr(6) + 0x14c, gpr(9));
        WriteMacInt32(gpr(6) + 0x154, gpr(10));
        WriteMacInt32(gpr(6) + 0x15c, gpr(11));
        WriteMacInt32(gpr(6) + 0x164, gpr(12));
        WriteMacInt32(gpr(6) + 0x16c, gpr(13));

        gpr(1)  = KernelDataAddr;
        gpr(7)  = ntohl(kernel_data->v[0x660 >> 2]);
        gpr(8)  = 0;
        gpr(10) = trampoline.addr();
        gpr(12) = trampoline.addr();
        gpr(13) = get_cr();

        // rlwimi. r7,r7,8,0,0
        uint32 result = op_ppc_rlwimi::apply(gpr(7), 8, 0x80000000, gpr(7));
        record_cr0(result);
        gpr(7) = result;

        gpr(11) = 0xf072; // MSR (SRR1)
        cr().set((gpr(11) & 0x0fff0000) | (get_cr() & ~0x0fff0000));

        // Enter nanokernel
        execute(entry);

#if !MULTICORE_CPU
        // Restore program counters and branch registers
        pc() = saved_pc;
        lr() = saved_lr;
        ctr()= saved_ctr;
        gpr(1) = saved_sp;
#endif

#if EMUL_TIME_STATS
        interrupt_time += (clock() - interrupt_start);
#endif

#if SAFE_INTERRUPT_PPC >= 2
        if (memcmp(&saved_regs[0], &gpr(0), sizeof(saved_regs)) != 0)
                printf("FATAL: dirty PowerPC registers\n");
#endif
#if SAFE_INTERRUPT_PPC
        depth--;
#endif
}

// Execute 68k routine
void sheepshaver_cpu::execute_68k(uint32 entry, M68kRegisters *r)
{
#if EMUL_TIME_STATS
        exec68k_count++;
        const clock_t exec68k_start = clock();
#endif

#if SAFE_EXEC_68K
        if (ReadMacInt32(XLM_RUN_MODE) != MODE_EMUL_OP)
                printf("FATAL: Execute68k() not called from EMUL_OP mode\n");
#endif

        // Save program counters and branch registers
        uint32 saved_pc = pc();
        uint32 saved_lr = lr();
        uint32 saved_ctr= ctr();
        uint32 saved_cr = get_cr();

        // Create MacOS stack frame
        // FIXME: make sure MacOS doesn't expect PPC registers to live on top
        uint32 sp = gpr(1);
        gpr(1) -= 56;
        WriteMacInt32(gpr(1), sp);

        // Save PowerPC registers
        uint32 saved_GPRs[19];
        memcpy(&saved_GPRs[0], &gpr(13), sizeof(uint32)*(32-13));
#if SAVE_FP_EXEC_68K
        double saved_FPRs[18];
        memcpy(&saved_FPRs[0], &fpr(14), sizeof(double)*(32-14));
#endif

        // Setup registers for 68k emulator
        cr().set(CR_SO_field<2>::mask());                       // Supervisor mode
        for (int i = 0; i < 8; i++)                                     // d[0]..d[7]
          gpr(8 + i) = r->d[i];
        for (int i = 0; i < 7; i++)                                     // a[0]..a[6]
          gpr(16 + i) = r->a[i];
        gpr(23) = 0;
        gpr(24) = entry;
        gpr(25) = ReadMacInt32(XLM_68K_R25);            // MSB of SR
        gpr(26) = 0;
        gpr(28) = 0;                                                            // VBR
        gpr(29) = ntohl(kernel_data->ed.v[0x74 >> 2]);          // Pointer to opcode table
        gpr(30) = ntohl(kernel_data->ed.v[0x78 >> 2]);          // Address of emulator
        gpr(31) = KernelDataAddr + 0x1000;

        // Push return address (points to EXEC_RETURN opcode) on stack
        gpr(1) -= 4;
        WriteMacInt32(gpr(1), XLM_EXEC_RETURN_OPCODE);
        
        // Rentering 68k emulator
        WriteMacInt32(XLM_RUN_MODE, MODE_68K);

        // Set r0 to 0 for 68k emulator
        gpr(0) = 0;

        // Execute 68k opcode
        uint32 opcode = ReadMacInt16(gpr(24));
        gpr(27) = (int32)(int16)ReadMacInt16(gpr(24) += 2);
        gpr(29) += opcode * 8;
        execute(gpr(29));

        // Save r25 (contains current 68k interrupt level)
        WriteMacInt32(XLM_68K_R25, gpr(25));

        // Reentering EMUL_OP mode
        WriteMacInt32(XLM_RUN_MODE, MODE_EMUL_OP);

        // Save 68k registers
        for (int i = 0; i < 8; i++)                                     // d[0]..d[7]
          r->d[i] = gpr(8 + i);
        for (int i = 0; i < 7; i++)                                     // a[0]..a[6]
          r->a[i] = gpr(16 + i);

        // Restore PowerPC registers
        memcpy(&gpr(13), &saved_GPRs[0], sizeof(uint32)*(32-13));
#if SAVE_FP_EXEC_68K
        memcpy(&fpr(14), &saved_FPRs[0], sizeof(double)*(32-14));
#endif

        // Cleanup stack
        gpr(1) += 56;

        // Restore program counters and branch registers
        pc() = saved_pc;
        lr() = saved_lr;
        ctr()= saved_ctr;
        set_cr(saved_cr);

#if EMUL_TIME_STATS
        exec68k_time += (clock() - exec68k_start);
#endif
}

// Call MacOS PPC code
uint32 sheepshaver_cpu::execute_macos_code(uint32 tvect, int nargs, uint32 const *args)
{
#if EMUL_TIME_STATS
        macos_exec_count++;
        const clock_t macos_exec_start = clock();
#endif

        // Save program counters and branch registers
        uint32 saved_pc = pc();
        uint32 saved_lr = lr();
        uint32 saved_ctr= ctr();

        // Build trampoline with EXEC_RETURN
        SheepVar32 trampoline = POWERPC_EXEC_RETURN;
        lr() = trampoline.addr();

        gpr(1) -= 64;                                                           // Create stack frame
        uint32 proc = ReadMacInt32(tvect);                      // Get routine address
        uint32 toc = ReadMacInt32(tvect + 4);           // Get TOC pointer

        // Save PowerPC registers
        uint32 regs[8];
        regs[0] = gpr(2);
        for (int i = 0; i < nargs; i++)
                regs[i + 1] = gpr(i + 3);

        // Prepare and call MacOS routine
        gpr(2) = toc;
        for (int i = 0; i < nargs; i++)
                gpr(i + 3) = args[i];
        execute(proc);
        uint32 retval = gpr(3);

        // Restore PowerPC registers
        for (int i = 0; i <= nargs; i++)
                gpr(i + 2) = regs[i];

        // Cleanup stack
        gpr(1) += 64;

        // Restore program counters and branch registers
        pc() = saved_pc;
        lr() = saved_lr;
        ctr()= saved_ctr;

#if EMUL_TIME_STATS
        macos_exec_time += (clock() - macos_exec_start);
#endif

        return retval;
}

// Execute ppc routine
inline void sheepshaver_cpu::execute_ppc(uint32 entry)
{
        // Save branch registers
        uint32 saved_lr = lr();

        SheepVar32 trampoline = POWERPC_EXEC_RETURN;
        WriteMacInt32(trampoline.addr(), POWERPC_EXEC_RETURN);
        lr() = trampoline.addr();

        execute(entry);

        // Restore branch registers
        lr() = saved_lr;
}

// Resource Manager thunk
inline void sheepshaver_cpu::get_resource(uint32 old_get_resource)
{
        uint32 type = gpr(3);
        int16 id = gpr(4);

        // Create stack frame
        gpr(1) -= 56;

        // Call old routine
        execute_ppc(old_get_resource);

        // Call CheckLoad()
        uint32 handle = gpr(3);
        check_load_invoc(type, id, handle);
        gpr(3) = handle;

        // Cleanup stack
        gpr(1) += 56;
}


/**
 *              SheepShaver CPU engine interface
 **/

static sheepshaver_cpu *main_cpu = NULL;                // CPU emulator to handle usual control flow
static sheepshaver_cpu *interrupt_cpu = NULL;   // CPU emulator to handle interrupts
static sheepshaver_cpu *current_cpu = NULL;             // Current CPU emulator context

void FlushCodeCache(uintptr start, uintptr end)
{
        D(bug("FlushCodeCache(%08x, %08x)\n", start, end));
        main_cpu->invalidate_cache_range(start, end);
#if MULTICORE_CPU
        interrupt_cpu->invalidate_cache_range(start, end);
#endif
}

static inline void cpu_push(sheepshaver_cpu *new_cpu)
{
#if MULTICORE_CPU
        current_cpu = new_cpu;
#endif
}

static inline void cpu_pop()
{
#if MULTICORE_CPU
        current_cpu = main_cpu;
#endif
}

// Dump PPC registers
static void dump_registers(void)
{
        current_cpu->dump_registers();
}

// Dump log
static void dump_log(void)
{
        current_cpu->dump_log();
}

/*
 *  Initialize CPU emulation
 */

static sigsegv_return_t sigsegv_handler(sigsegv_address_t fault_address, sigsegv_address_t fault_instruction)
{
#if ENABLE_VOSF
        // Handle screen fault
        extern bool Screen_fault_handler(sigsegv_address_t, sigsegv_address_t);
        if (Screen_fault_handler(fault_address, fault_instruction))
                return SIGSEGV_RETURN_SUCCESS;
#endif

        const uintptr addr = (uintptr)fault_address;
#if HAVE_SIGSEGV_SKIP_INSTRUCTION
        // Ignore writes to ROM
        if ((addr - ROM_BASE) < ROM_SIZE)
                return SIGSEGV_RETURN_SKIP_INSTRUCTION;

        // Get program counter of target CPU
        sheepshaver_cpu * const cpu = current_cpu;
        const uint32 pc = cpu->pc();
        
        // Fault in Mac ROM or RAM?
        bool mac_fault = (pc >= ROM_BASE) && (pc < (ROM_BASE + ROM_AREA_SIZE)) || (pc >= RAMBase) && (pc < (RAMBase + RAMSize));
        if (mac_fault) {

                // "VM settings" during MacOS 8 installation
                if (pc == ROM_BASE + 0x488160 && cpu->gpr(20) == 0xf8000000)
                        return SIGSEGV_RETURN_SKIP_INSTRUCTION;
        
                // MacOS 8.5 installation
                else if (pc == ROM_BASE + 0x488140 && cpu->gpr(16) == 0xf8000000)
                        return SIGSEGV_RETURN_SKIP_INSTRUCTION;
        
                // MacOS 8 serial drivers on startup
                else if (pc == ROM_BASE + 0x48e080 && (cpu->gpr(8) == 0xf3012002 || cpu->gpr(8) == 0xf3012000))
                        return SIGSEGV_RETURN_SKIP_INSTRUCTION;
        
                // MacOS 8.1 serial drivers on startup
                else if (pc == ROM_BASE + 0x48c5e0 && (cpu->gpr(20) == 0xf3012002 || cpu->gpr(20) == 0xf3012000))
                        return SIGSEGV_RETURN_SKIP_INSTRUCTION;
                else if (pc == ROM_BASE + 0x4a10a0 && (cpu->gpr(20) == 0xf3012002 || cpu->gpr(20) == 0xf3012000))
                        return SIGSEGV_RETURN_SKIP_INSTRUCTION;

                // Ignore writes to the zero page
                else if ((uint32)(addr - SheepMem::ZeroPage()) < (uint32)SheepMem::PageSize())
                        return SIGSEGV_RETURN_SKIP_INSTRUCTION;

                // Ignore all other faults, if requested
                if (PrefsFindBool("ignoresegv"))
                        return SIGSEGV_RETURN_SKIP_INSTRUCTION;
        }
#else
#error "FIXME: You don't have the capability to skip instruction within signal handlers"
#endif

        printf("SIGSEGV\n");
        printf("  pc %p\n", fault_instruction);
        printf("  ea %p\n", fault_address);
        printf(" cpu %s\n", current_cpu == main_cpu ? "main" : "interrupts");
        dump_registers();
        current_cpu->dump_log();
        enter_mon();
        QuitEmulator();

        return SIGSEGV_RETURN_FAILURE;
}

void init_emul_ppc(void)
{
        // Initialize main CPU emulator
        main_cpu = new sheepshaver_cpu();
        main_cpu->set_register(powerpc_registers::GPR(3), any_register((uint32)ROM_BASE + 0x30d000));
        main_cpu->set_register(powerpc_registers::GPR(4), any_register(KernelDataAddr + 0x1000));
        WriteMacInt32(XLM_RUN_MODE, MODE_68K);

#if MULTICORE_CPU
        // Initialize alternate CPU emulator to handle interrupts
        interrupt_cpu = new sheepshaver_cpu();
#endif

        // Install the handler for SIGSEGV
        sigsegv_install_handler(sigsegv_handler);

#if ENABLE_MON
        // Install "regs" command in cxmon
        mon_add_command("regs", dump_registers, "regs                     Dump PowerPC registers\n");
        mon_add_command("log", dump_log, "log                      Dump PowerPC emulation log\n");
#endif

#if EMUL_TIME_STATS
        emul_start_time = clock();
#endif
}

/*
 *  Deinitialize emulation
 */

void exit_emul_ppc(void)
{
#if EMUL_TIME_STATS
        clock_t emul_end_time = clock();

        printf("### Statistics for SheepShaver emulation parts\n");
        const clock_t emul_time = emul_end_time - emul_start_time;
        printf("Total emulation time : %.1f sec\n", double(emul_time) / double(CLOCKS_PER_SEC));
        printf("Total interrupt count: %d (%2.1f Hz)\n", interrupt_count,
                   (double(interrupt_count) * CLOCKS_PER_SEC) / double(emul_time));

#define PRINT_STATS(LABEL, VAR_PREFIX) do {                                                             \
                printf("Total " LABEL " count : %d\n", VAR_PREFIX##_count);             \
                printf("Total " LABEL " time  : %.1f sec (%.1f%%)\n",                   \
                           double(VAR_PREFIX##_time) / double(CLOCKS_PER_SEC),          \
                           100.0 * double(VAR_PREFIX##_time) / double(emul_time));      \
        } while (0)

        PRINT_STATS("Execute68k[Trap] execution", exec68k);
        PRINT_STATS("NativeOp execution", native_exec);
        PRINT_STATS("MacOS routine execution", macos_exec);

#undef PRINT_STATS
        printf("\n");
#endif

        delete main_cpu;
#if MULTICORE_CPU
        delete interrupt_cpu;
#endif
}

#if PPC_ENABLE_JIT && PPC_REENTRANT_JIT
// Initialize EmulOp trampolines
void init_emul_op_trampolines(basic_dyngen & dg)
{
        typedef void (*func_t)(dyngen_cpu_base, uint32);
        func_t func;

        // EmulOp
        emul_op_trampoline = dg.gen_start();
        func = (func_t)nv_mem_fun(&sheepshaver_cpu::execute_emul_op).ptr();
        dg.gen_invoke_CPU_T0(func);
        dg.gen_exec_return();
        dg.gen_end();

        // NativeOp
        native_op_trampoline = dg.gen_start();
        func = (func_t)nv_mem_fun(&sheepshaver_cpu::execute_native_op).ptr();
        dg.gen_invoke_CPU_T0(func);     
        dg.gen_exec_return();
        dg.gen_end();

        D(bug("EmulOp trampoline:   %p\n", emul_op_trampoline));
        D(bug("NativeOp trampoline: %p\n", native_op_trampoline));
}
#endif

/*
 *  Emulation loop
 */

void emul_ppc(uint32 entry)
{
        current_cpu = main_cpu;
#if 0
        current_cpu->start_log();
#endif
        // start emulation loop and enable code translation or caching
        current_cpu->execute(entry);
}

/*
 *  Handle PowerPC interrupt
 */

#if ASYNC_IRQ
void HandleInterrupt(void)
{
        main_cpu->handle_interrupt();
}
#else
void TriggerInterrupt(void)
{
#if 0
  WriteMacInt32(0x16a, ReadMacInt32(0x16a) + 1);
#else
  // Trigger interrupt to main cpu only
  if (main_cpu)
          main_cpu->trigger_interrupt();
#endif
}
#endif

void sheepshaver_cpu::handle_interrupt(void)
{
        // Do nothing if interrupts are disabled
        if (*(int32 *)XLM_IRQ_NEST > 0)
                return;

        // Do nothing if there is no interrupt pending
        if (InterruptFlags == 0)
                return;

        // Disable MacOS stack sniffer
        WriteMacInt32(0x110, 0);

        // Interrupt action depends on current run mode
        switch (ReadMacInt32(XLM_RUN_MODE)) {
        case MODE_68K:
                // 68k emulator active, trigger 68k interrupt level 1
                assert(current_cpu == main_cpu);
                WriteMacInt16(tswap32(kernel_data->v[0x67c >> 2]), 1);
                set_cr(get_cr() | tswap32(kernel_data->v[0x674 >> 2]));
                break;
    
#if INTERRUPTS_IN_NATIVE_MODE
        case MODE_NATIVE:
                // 68k emulator inactive, in nanokernel?
                assert(current_cpu == main_cpu);
                if (gpr(1) != KernelDataAddr) {
                        // Prepare for 68k interrupt level 1
                        WriteMacInt16(tswap32(kernel_data->v[0x67c >> 2]), 1);
                        WriteMacInt32(tswap32(kernel_data->v[0x658 >> 2]) + 0xdc,
                                                  ReadMacInt32(tswap32(kernel_data->v[0x658 >> 2]) + 0xdc)
                                                  | tswap32(kernel_data->v[0x674 >> 2]));
      
                        // Execute nanokernel interrupt routine (this will activate the 68k emulator)
                        DisableInterrupt();
                        cpu_push(interrupt_cpu);
                        if (ROMType == ROMTYPE_NEWWORLD)
                                current_cpu->interrupt(ROM_BASE + 0x312b1c);
                        else
                                current_cpu->interrupt(ROM_BASE + 0x312a3c);
                        cpu_pop();
                }
                break;
#endif
    
#if INTERRUPTS_IN_EMUL_OP_MODE
        case MODE_EMUL_OP:
                // 68k emulator active, within EMUL_OP routine, execute 68k interrupt routine directly when interrupt level is 0
                if ((ReadMacInt32(XLM_68K_R25) & 7) == 0) {
#if 1
                        // Execute full 68k interrupt routine
                        M68kRegisters r;
                        uint32 old_r25 = ReadMacInt32(XLM_68K_R25);     // Save interrupt level
                        WriteMacInt32(XLM_68K_R25, 0x21);                       // Execute with interrupt level 1
                        static const uint8 proc[] = {
                                0x3f, 0x3c, 0x00, 0x00,                 // move.w       #$0000,-(sp)    (fake format word)
                                0x48, 0x7a, 0x00, 0x0a,                 // pea          @1(pc)                  (return address)
                                0x40, 0xe7,                                             // move         sr,-(sp)                (saved SR)
                                0x20, 0x78, 0x00, 0x064,                // move.l       $64,a0
                                0x4e, 0xd0,                                             // jmp          (a0)
                                M68K_RTS >> 8, M68K_RTS & 0xff  // @1
                        };
                        Execute68k((uint32)proc, &r);
                        WriteMacInt32(XLM_68K_R25, old_r25);            // Restore interrupt level
#else
                        // Only update cursor
                        if (HasMacStarted()) {
                                if (InterruptFlags & INTFLAG_VIA) {
                                        ClearInterruptFlag(INTFLAG_VIA);
                                        ADBInterrupt();
                                        ExecuteNative(NATIVE_VIDEO_VBL);
                                }
                        }
#endif
                }
                break;
#endif
        }
}

static void get_resource(void);
static void get_1_resource(void);
static void get_ind_resource(void);
static void get_1_ind_resource(void);
static void r_get_resource(void);

// Execute NATIVE_OP routine
void sheepshaver_cpu::execute_native_op(uint32 selector)
{
#if EMUL_TIME_STATS
        native_exec_count++;
        const clock_t native_exec_start = clock();
#endif

        switch (selector) {
        case NATIVE_PATCH_NAME_REGISTRY:
                DoPatchNameRegistry();
                break;
        case NATIVE_VIDEO_INSTALL_ACCEL:
                VideoInstallAccel();
                break;
        case NATIVE_VIDEO_VBL:
                VideoVBL();
                break;
        case NATIVE_VIDEO_DO_DRIVER_IO:
                gpr(3) = (int32)(int16)VideoDoDriverIO((void *)gpr(3), (void *)gpr(4),
                                                                                           (void *)gpr(5), gpr(6), gpr(7));
                break;
#ifdef WORDS_BIGENDIAN
        case NATIVE_ETHER_IRQ:
                EtherIRQ();
                break;
        case NATIVE_ETHER_INIT:
                gpr(3) = InitStreamModule((void *)gpr(3));
                break;
        case NATIVE_ETHER_TERM:
                TerminateStreamModule();
                break;
        case NATIVE_ETHER_OPEN:
                gpr(3) = ether_open((queue_t *)gpr(3), (void *)gpr(4), gpr(5), gpr(6), (void*)gpr(7));
                break;
        case NATIVE_ETHER_CLOSE:
                gpr(3) = ether_close((queue_t *)gpr(3), gpr(4), (void *)gpr(5));
                break;
        case NATIVE_ETHER_WPUT:
                gpr(3) = ether_wput((queue_t *)gpr(3), (mblk_t *)gpr(4));
                break;
        case NATIVE_ETHER_RSRV:
                gpr(3) = ether_rsrv((queue_t *)gpr(3));
                break;
#else
        case NATIVE_ETHER_INIT:
                // FIXME: needs more complicated thunks
                gpr(3) = false;
                break;
#endif
        case NATIVE_SYNC_HOOK:
                gpr(3) = NQD_sync_hook(gpr(3));
                break;
        case NATIVE_BITBLT_HOOK:
                gpr(3) = NQD_bitblt_hook(gpr(3));
                break;
        case NATIVE_BITBLT:
                NQD_bitblt(gpr(3));
                break;
        case NATIVE_FILLRECT_HOOK:
                gpr(3) = NQD_fillrect_hook(gpr(3));
                break;
        case NATIVE_INVRECT:
                NQD_invrect(gpr(3));
                break;
        case NATIVE_FILLRECT:
                NQD_fillrect(gpr(3));
                break;
        case NATIVE_SERIAL_NOTHING:
        case NATIVE_SERIAL_OPEN:
        case NATIVE_SERIAL_PRIME_IN:
        case NATIVE_SERIAL_PRIME_OUT:
        case NATIVE_SERIAL_CONTROL:
        case NATIVE_SERIAL_STATUS:
        case NATIVE_SERIAL_CLOSE: {
                typedef int16 (*SerialCallback)(uint32, uint32);
                static const SerialCallback serial_callbacks[] = {
                        SerialNothing,
                        SerialOpen,
                        SerialPrimeIn,
                        SerialPrimeOut,
                        SerialControl,
                        SerialStatus,
                        SerialClose
                };
                gpr(3) = serial_callbacks[selector - NATIVE_SERIAL_NOTHING](gpr(3), gpr(4));
                break;
        }
        case NATIVE_GET_RESOURCE:
        case NATIVE_GET_1_RESOURCE:
        case NATIVE_GET_IND_RESOURCE:
        case NATIVE_GET_1_IND_RESOURCE:
        case NATIVE_R_GET_RESOURCE: {
                typedef void (*GetResourceCallback)(void);
                static const GetResourceCallback get_resource_callbacks[] = {
                        ::get_resource,
                        ::get_1_resource,
                        ::get_ind_resource,
                        ::get_1_ind_resource,
                        ::r_get_resource
                };
                get_resource_callbacks[selector - NATIVE_GET_RESOURCE]();
                break;
        }
        case NATIVE_DISABLE_INTERRUPT:
                DisableInterrupt();
                break;
        case NATIVE_ENABLE_INTERRUPT:
                EnableInterrupt();
                break;
        case NATIVE_MAKE_EXECUTABLE:
                MakeExecutable(0, (void *)gpr(4), gpr(5));
                break;
        case NATIVE_CHECK_LOAD_INVOC:
                check_load_invoc(gpr(3), gpr(4), gpr(5));
                break;
        default:
                printf("FATAL: NATIVE_OP called with bogus selector %d\n", selector);
                QuitEmulator();
                break;
        }

#if EMUL_TIME_STATS
        native_exec_time += (clock() - native_exec_start);
#endif
}

/*
 *  Execute 68k subroutine (must be ended with EXEC_RETURN)
 *  This must only be called by the emul_thread when in EMUL_OP mode
 *  r->a[7] is unused, the routine runs on the caller's stack
 */

void Execute68k(uint32 pc, M68kRegisters *r)
{
        current_cpu->execute_68k(pc, r);
}

/*
 *  Execute 68k A-Trap from EMUL_OP routine
 *  r->a[7] is unused, the routine runs on the caller's stack
 */

void Execute68kTrap(uint16 trap, M68kRegisters *r)
{
        SheepVar proc_var(4);
        uint32 proc = proc_var.addr();
        WriteMacInt16(proc, trap);
        WriteMacInt16(proc + 2, M68K_RTS);
        Execute68k(proc, r);
}

/*
 *  Call MacOS PPC code
 */

uint32 call_macos(uint32 tvect)
{
        return current_cpu->execute_macos_code(tvect, 0, NULL);
}

uint32 call_macos1(uint32 tvect, uint32 arg1)
{
        const uint32 args[] = { arg1 };
        return current_cpu->execute_macos_code(tvect, sizeof(args)/sizeof(args[0]), args);
}

uint32 call_macos2(uint32 tvect, uint32 arg1, uint32 arg2)
{
        const uint32 args[] = { arg1, arg2 };
        return current_cpu->execute_macos_code(tvect, sizeof(args)/sizeof(args[0]), args);
}

uint32 call_macos3(uint32 tvect, uint32 arg1, uint32 arg2, uint32 arg3)
{
        const uint32 args[] = { arg1, arg2, arg3 };
        return current_cpu->execute_macos_code(tvect, sizeof(args)/sizeof(args[0]), args);
}

uint32 call_macos4(uint32 tvect, uint32 arg1, uint32 arg2, uint32 arg3, uint32 arg4)
{
        const uint32 args[] = { arg1, arg2, arg3, arg4 };
        return current_cpu->execute_macos_code(tvect, sizeof(args)/sizeof(args[0]), args);
}

uint32 call_macos5(uint32 tvect, uint32 arg1, uint32 arg2, uint32 arg3, uint32 arg4, uint32 arg5)
{
        const uint32 args[] = { arg1, arg2, arg3, arg4, arg5 };
        return current_cpu->execute_macos_code(tvect, sizeof(args)/sizeof(args[0]), args);
}

uint32 call_macos6(uint32 tvect, uint32 arg1, uint32 arg2, uint32 arg3, uint32 arg4, uint32 arg5, uint32 arg6)
{
        const uint32 args[] = { arg1, arg2, arg3, arg4, arg5, arg6 };
        return current_cpu->execute_macos_code(tvect, sizeof(args)/sizeof(args[0]), args);
}

uint32 call_macos7(uint32 tvect, uint32 arg1, uint32 arg2, uint32 arg3, uint32 arg4, uint32 arg5, uint32 arg6, uint32 arg7)
{
        const uint32 args[] = { arg1, arg2, arg3, arg4, arg5, arg6, arg7 };
        return current_cpu->execute_macos_code(tvect, sizeof(args)/sizeof(args[0]), args);
}

/*
 *  Resource Manager thunks
 */

void get_resource(void)
{
        current_cpu->get_resource(ReadMacInt32(XLM_GET_RESOURCE));
}

void get_1_resource(void)
{
        current_cpu->get_resource(ReadMacInt32(XLM_GET_1_RESOURCE));
}

void get_ind_resource(void)
{
        current_cpu->get_resource(ReadMacInt32(XLM_GET_IND_RESOURCE));
}

void get_1_ind_resource(void)
{
        current_cpu->get_resource(ReadMacInt32(XLM_GET_1_IND_RESOURCE));
}

void r_get_resource(void)
{
        current_cpu->get_resource(ReadMacInt32(XLM_R_GET_RESOURCE));
}
Revision:	1.38
Committed:	2004-05-19T21:23:16Z (20 years, 4 months ago) by gbeauche
Branch:	MAIN
Changes since 1.37:	+126 -57 lines
Log Message:	Make NativeOp() handler a sheepshaver_cpu handler, thus getting rid of ugly GPR macro definition. Make the JIT engine somewhat reentrant. This brings a massive performance boost for applications that cause many Execute68k(). e.g. audio in PlayerPRO.