#include <windows.h>
#include "debug.h"
#include "stringutils.h"
#include "args.h"
#include  "kernelmisc.h"
#include  "arm_cp.h"

// tstbxj 
// r[i] = 0xe0000000+0x1110*i
//   -> prefetch abort @ pc=0xe000eee0  bva=0x27ffc9e4 pexi=0xe000dda4{SP-0x2c} r5 -= 0x140
//   -> data abort @ pc=0x8c10a368, bva=0xe0005810{r5+0x2c0}
//
// tstbxj -5
// r[i] = 0xe0000000+0x1110*i,  r[5]={handler, ..., handler}
//   -> prefetch abort @ pc=0xe000eee0 bva=0x2dffc9e4 pexi=0xe000dda4 r5 = 0x1a90730-0x300
//
// tstbxj -5
//   -> prefetch abort @ pc=0xe000eee0 bva=0xe0007770{r7} pexi=0xe000dda4 r5 = 0x1a90730-0x300
//   -> data abort @ pc=0x8c10a368, bva=0xe0007770
//
// tstbxj -5,  r5={0xd0000000+i*0x1000}
//   -> prefetchabort @ pc=0xd0024000, ra=0x1a902cc, bva=0xd0024000
//
#define structofs(s,x)  ((int)(&(((s*)0)->x)))
struct regbank {
    uint32_t cpsr;
    uint32_t r[16];
};

class bxjcode {
private:
typedef void (*voidfn)();
    DwordVector _code;
    voidfn _fn;
    regbank *_regs;
public:
    bxjcode() {
        genbxj(_code, &_regs, &_fn);
        debug("%08lx: bxjcall: %s\n", &_code[0], hexdump((BYTE*)(&_code[0]), _code.size(), 4).c_str());
    }
    regbank call(const regbank& regs)
    {
        memcpy(_regs, &regs, sizeof(regbank));
        _fn();
        regbank out;
        memcpy(&out, _regs+1, sizeof(regbank));
        return out;
    }
private:
    void genbxj(DwordVector& code, regbank** regs, voidfn *bxjfn)
    {
        typedef std::map<int,int> fixupmap_t;
        fixupmap_t fixups;
        code.push_back(0xe92d7ff0);  // STMFD SP!, {R4..R14}
        fixups[code.size()]= structofs(regbank, cpsr);
        code.push_back(0xe59f0000);  // LDR R0, [PC,#...]
        // cond 0001 0R10 field_mask : SBO SBZ 0000 Rm  - msr
        code.push_back(0xe128f000);  // MSR R0, CPSR
        fixups[code.size()]= structofs(regbank, r[0]);
        code.push_back(0xe28f0000);  // ADD R0, PC, #...

        // don't touch LR and SP
        code.push_back(0xe8901fff);  // LDMIA R0, {R0..R12}
        code.push_back(0xe1a0e00f);  // MOV LR, PC
        code.push_back(0xe12fff2c);  // BXJ R12
        // cond 01IP U0W0 Rn : Rd addr_mode
        fixups[code.size()]= sizeof(regbank)+structofs(regbank, r[0]);
        code.push_back(0xe58f0000);  // STR R0, [PC,#...]
        // cond 0001 0R00 SBO : Rd SBZ  - mrs
        code.push_back(0xe10f0000);  // MRS R0, CPSR
        fixups[code.size()]= sizeof(regbank)+structofs(regbank, cpsr);
        code.push_back(0xe58f0000);  // STR R0, [PC,#...]
        fixups[code.size()]= sizeof(regbank)+structofs(regbank, r[1]);
        code.push_back(0xe28f0000);  // ADD R0, PC, #...
        code.push_back(0xe8807ffe);  // STMIA R0, {R1..R14}
        code.push_back(0xe8bd7ff0);  // LDMFD SP!, {R4..R14}
        code.push_back(0xe12fff1e);  // BX LR
        int ofstoregs=code.size();
        code.resize(code.size()+2*(sizeof(regbank)/sizeof(uint32_t)));    // twice r0..r14 + twice cpsr + once spsr

        for (fixupmap_t::iterator i= fixups.begin() ; i!=fixups.end() ; i++)
        {
            code[(*i).first] += sizeof(uint32_t)*(ofstoregs-(*i).first-2)+(*i).second;
        }
        CacheRangeFlush(0, 0, CACHE_SYNC_ALL);

        *regs= reinterpret_cast<regbank*>(&code[ofstoregs]);
        *bxjfn= reinterpret_cast<voidfn>(&code[0]);
    }
};
void unhandled(int r0, int r1, int r2, int r3)
{
    *(DWORD*)0xeeee0000;
}
int WINAPI WinMain( HINSTANCE hInstance,
                   HINSTANCE hPrevInstance,
                   LPTSTR    lpCmdLine,
                   int       nCmdShow)
{
    DebugSetLogfile("tstbxj.log");
    StringList args;
    if (!SplitString(ToString(lpCmdLine), args, false))
    {
        error("Error in commandline");
        return false;
    }
    bool bPointR5= false;
    bool bPointR14=false;
    for (StringList::iterator i= args.begin() ; i!=args.end() ; ++i)
    {
        std::string& arg= *i;
        if (arg[0]=='-') switch(arg[1])
        {
            case '5': bPointR5=true; break;

            case 'e':
            case 'E': bPointR14=true; break;
        }
    }

    KernelMode _km;

    bxjcode bxj;

    armcoprocessor cp;

    debug("jzid=       %08lx [AIISSiii]\n", cp.read(14,7,0,0,0));
    debug("jzoscontrol=%08lx [0000000C]  cv + cd bits\n", cp.read(14,7,1,0,0));
    debug("jzconfig=   %08lx [0000000J]  JE bit\n", cp.read(14,7,2,0,0));

    cp.write(14,7,2,0,0, 1);

    regbank r;
    memset(&r, 0, sizeof(r));
    for (int i=0 ; i<16 ; i++)
        r.r[i]= 0x1110*i+0xe0000000;

    const uint8_t undefbc[]={0xcb,0xfe,0xff};
    const uint8_t load42bc[]={0x10,0x00,0x00,0x00,0x2a,0x3b,0x1a,0xff,0xfe};

    if (bPointR14)
        r.r[14]= reinterpret_cast<uint32_t>(undefbc);

    // todo: generate different code for each handler
    DwordVector r5(1024);
    for (unsigned i=0 ; i<1024 ; i++) {
        r5[i]= 0xd0000000+i*0x1000; // reinterpret_cast<uint32_t>(&unhandled);
    }
    if (bPointR5)
        r.r[5]= reinterpret_cast<uint32_t>(&r5[0]);

    debug("unh=%08lx,  r5=%08lx r14=%08lx\n", &unhandled, r.r[5], r.r[14]);

    DWORD heapblock= ((DWORD)&r5[0])&~0xfff;
//  debug(".... heap block (ph=%08lx)\n%s\n", GetProcessHeap(),
//          hexdump(heapblock, (BYTE*)heapblock, 0x1000).c_str());

    regbank rout= bxj.call(r);

    debug("cpsr=%08lx\n", r.cpsr);
    debug("0-3 =%08lx %08lx %08lx %08lx\n", rout.r[0], rout.r[1], rout.r[2], rout.r[3]);
    debug("4-7 =%08lx %08lx %08lx %08lx\n", rout.r[4], rout.r[5], rout.r[6], rout.r[7]);
    debug("8-b =%08lx %08lx %08lx %08lx\n", rout.r[8], rout.r[9], rout.r[10], rout.r[11]);
    debug("c-f =%08lx %08lx %08lx %08lx\n", rout.r[12], rout.r[13], rout.r[14], rout.r[15]);

    debug("done\n\n");
}