fold/tools/pcode_dump.cpp

// Phase-1 smoke tool for the vendored Ghidra decompiler.
//
// Lifts a raw machine-code buffer to Ghidra P-Code using a compiled .sla and
// prints disassembly + P-Code per instruction. This validates the vendored
// engine builds, the .sla loads, and lifting works end-to-end — before any
// fold-side IR/encoder glue (Phase 2+). Bytes are served via a custom
// LoadImage (no BFD): fold's loader contract.
//
//   pcode_dump <path/to.sla> <base_vaddr_hex> <raw_bytes_file>
//
#include <fstream>
#include <iostream>
#include <sstream>
#include <vector>

#include "loadimage.hh"
#include "sleigh.hh"

using namespace ghidra;

namespace {

// Serves a flat byte buffer at a fixed base address; zero-fills outside it.
class BufferLoadImage : public LoadImage {
    uintb base_;
    std::vector<uint1> data_;

  public:
    BufferLoadImage(uintb base, std::vector<uint1> data)
        : LoadImage("nofile"), base_(base), data_(std::move(data)) {}
    void loadFill(uint1* ptr, int4 size, const Address& addr) override {
        uintb start = addr.getOffset();
        for (int4 i = 0; i < size; ++i) {
            uintb off = start + i;
            ptr[i] = (off >= base_ && off < base_ + data_.size()) ? data_[off - base_] : 0;
        }
    }
    std::string getArchType() const override { return "buffer"; }
    void adjustVma(long) override {}
    uintb end() const { return base_ + data_.size(); }
};

class AsmPrinter : public AssemblyEmit {
  public:
    void dump(const Address& addr, const std::string& mnem, const std::string& body) override {
        addr.printRaw(std::cout);
        std::cout << ":  " << mnem << ' ' << body << '\n';
    }
};

void printVar(std::ostream& s, const VarnodeData& v) {
    s << '(' << v.space->getName() << ',';
    v.space->printOffset(s, v.offset);
    s << ',' << std::dec << v.size << ')';
}

class PcodePrinter : public PcodeEmit {
  public:
    void dump(const Address&, OpCode opc, VarnodeData* outvar, VarnodeData* vars,
              int4 isize) override {
        std::cout << "      ";
        if (outvar != nullptr) {
            printVar(std::cout, *outvar);
            std::cout << " = ";
        }
        std::cout << get_opname(opc);
        for (int4 i = 0; i < isize; ++i) {
            std::cout << ' ';
            printVar(std::cout, vars[i]);
        }
        std::cout << '\n';
    }
};

} // namespace

int main(int argc, char** argv) {
    if (argc != 4) {
        std::cerr << "usage: " << argv[0] << " <path.sla> <base_vaddr_hex> <raw_bytes_file>\n";
        return 2;
    }
    const std::string sla_path = argv[1];
    const uintb base = std::strtoull(argv[2], nullptr, 16);

    std::ifstream bf(argv[3], std::ios::binary);
    if (!bf) {
        std::cerr << "cannot open bytes file: " << argv[3] << '\n';
        return 2;
    }
    std::vector<uint1> bytes((std::istreambuf_iterator<char>(bf)),
                             std::istreambuf_iterator<char>());
    if (bytes.empty()) {
        std::cerr << "empty bytes file\n";
        return 2;
    }

    AttributeId::initialize();
    ElementId::initialize();

    BufferLoadImage loader(base, bytes);
    ContextInternal context;
    Sleigh trans(&loader, &context);

    std::istringstream slatag("<sleigh>" + sla_path + "</sleigh>");
    DocumentStorage docstorage;
    Element* root = docstorage.parseDocument(slatag)->getRoot();
    docstorage.registerTag(root);
    try {
        trans.initialize(docstorage);
    } catch (const LowlevelError& e) {
        std::cerr << "sleigh init failed: " << e.explain << '\n';
        return 1;
    }

    AsmPrinter asmp;
    PcodePrinter pcodep;
    Address addr(trans.getDefaultCodeSpace(), base);
    Address last(trans.getDefaultCodeSpace(), loader.end());
    int instrs = 0;
    while (addr < last) {
        try {
            trans.printAssembly(asmp, addr);
            int4 len = trans.oneInstruction(pcodep, addr);
            if (len <= 0)
                break;
            addr = addr + len;
            ++instrs;
        } catch (const BadDataError& e) {
            std::cerr << "bad data at ";
            addr.printRaw(std::cerr);
            std::cerr << ": " << e.explain << '\n';
            break;
        } catch (const LowlevelError& e) {
            std::cerr << "lift error: " << e.explain << '\n';
            break;
        }
    }
    std::cerr << "[pcode_dump] lifted " << instrs << " instruction(s)\n";
    return 0;
}