// Copyright (c) 2012 The Chromium OS Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

#include "update_engine/payload_generator/delta_diff_generator.h"

#include <errno.h>
#include <fcntl.h>
#include <inttypes.h>
#include <sys/stat.h>
#include <sys/types.h>

#include <algorithm>
#include <map>
#include <memory>
#include <set>
#include <string>
#include <utility>
#include <vector>

#include <base/files/file_path.h>
#include <base/files/file_util.h>
#include <base/logging.h>
#include <base/strings/stringprintf.h>
#include <base/strings/string_number_conversions.h>
#include <base/strings/string_util.h>
#include <bzlib.h>

#include "update_engine/bzip.h"
#include "update_engine/delta_performer.h"
#include "update_engine/extent_ranges.h"
#include "update_engine/file_writer.h"
#include "update_engine/omaha_hash_calculator.h"
#include "update_engine/payload_constants.h"
#include "update_engine/payload_generator/cycle_breaker.h"
#include "update_engine/payload_generator/extent_mapper.h"
#include "update_engine/payload_generator/filesystem_iterator.h"
#include "update_engine/payload_generator/full_update_generator.h"
#include "update_engine/payload_generator/graph_types.h"
#include "update_engine/payload_generator/graph_utils.h"
#include "update_engine/payload_generator/metadata.h"
#include "update_engine/payload_generator/payload_signer.h"
#include "update_engine/payload_generator/topological_sort.h"
#include "update_engine/payload_verifier.h"
#include "update_engine/subprocess.h"
#include "update_engine/update_metadata.pb.h"
#include "update_engine/utils.h"

using std::make_pair;
using std::map;
using std::max;
using std::min;
using std::pair;
using std::set;
using std::string;
using std::unique_ptr;
using std::vector;

namespace {

const uint64_t kVersionNumber = 1;
const uint64_t kFullUpdateChunkSize = 1024 * 1024;  // bytes

const size_t kBlockSize = 4096;  // bytes
const char kEmptyPath[] = "";

// The maximum destination size allowed for bsdiff. In general, bsdiff should
// work for arbitrary big files, but the payload generation and payload
// application requires a significant amount of RAM. We put a hard-limit of
// 200 MiB that should not affect any released board, but will limit the
// Chrome binary in ASan builders.
const off_t kMaxBsdiffDestinationSize = 200 * 1024 * 1024;  // bytes

static const char* kInstallOperationTypes[] = {
  "REPLACE",
  "REPLACE_BZ",
  "MOVE",
  "BSDIFF",
  "SOURCE_COPY",
  "SOURCE_BSDIFF"
};

}  // namespace

namespace chromeos_update_engine {

typedef DeltaDiffGenerator::Block Block;
typedef map<const DeltaArchiveManifest_InstallOperation*,
            string> OperationNameMap;

// bytes
const size_t kRootFSPartitionSize = static_cast<size_t>(2) * 1024 * 1024 * 1024;

// Needed for testing purposes, in case we can't use actual filesystem objects.
// TODO(garnold) (chromium:331965) Replace this hack with a properly injected
// parameter in form of a mockable abstract class.
bool (*get_extents_with_chunk_func)(const string&, off_t, off_t,
                                    vector<Extent>*) =
    extent_mapper::ExtentsForFileChunkFibmap;

namespace {

// Stores all the extents of |path| into |extents|. Returns true on success.
bool GatherExtents(const string& path,
                   off_t chunk_offset,
                   off_t chunk_size,
                   vector<Extent>* extents) {
  extents->clear();
  TEST_AND_RETURN_FALSE(
      get_extents_with_chunk_func(
          path, chunk_offset, chunk_size, extents));
  return true;
}

// For a given regular file which must exist at new_root + path, and
// may exist at old_root + path, creates a new InstallOperation and
// adds it to the graph. Also, populates the |blocks| array as
// necessary, if |blocks| is non-null.  Also, writes the data
// necessary to send the file down to the client into data_fd, which
// has length *data_file_size. *data_file_size is updated
// appropriately. If |existing_vertex| is no kInvalidIndex, use that
// rather than allocating a new vertex. Returns true on success.
bool DeltaReadFile(Graph* graph,
                   Vertex::Index existing_vertex,
                   vector<Block>* blocks,
                   const string& old_root,
                   const string& new_root,
                   const string& path,  // within new_root
                   off_t chunk_offset,
                   off_t chunk_size,
                   int data_fd,
                   off_t* data_file_size) {
  chromeos::Blob data;
  DeltaArchiveManifest_InstallOperation operation;

  string old_path = (old_root == kEmptyPath) ? kEmptyPath :
      old_root + path;

  // If bsdiff breaks again, blacklist the problem file by using:
  //   bsdiff_allowed = (path != "/foo/bar")
  //
  // TODO(dgarrett): chromium-os:15274 connect this test to the command line.
  bool bsdiff_allowed = true;

  if (utils::FileSize(new_root + path) > kMaxBsdiffDestinationSize)
    bsdiff_allowed = false;

  if (!bsdiff_allowed)
    LOG(INFO) << "bsdiff blacklisting: " << path;

  TEST_AND_RETURN_FALSE(DeltaDiffGenerator::ReadFileToDiff(old_path,
                                                           new_root + path,
                                                           chunk_offset,
                                                           chunk_size,
                                                           bsdiff_allowed,
                                                           &data,
                                                           &operation,
                                                           true));

  // Check if the operation writes nothing.
  if (operation.dst_extents_size() == 0) {
    if (operation.type() == DeltaArchiveManifest_InstallOperation_Type_MOVE) {
      LOG(INFO) << "Empty MOVE operation (" << new_root + path << "), skipping";
      return true;
    } else {
      LOG(ERROR) << "Empty non-MOVE operation";
      return false;
    }
  }

  // Write the data
  if (operation.type() != DeltaArchiveManifest_InstallOperation_Type_MOVE) {
    operation.set_data_offset(*data_file_size);
    operation.set_data_length(data.size());
  }

  TEST_AND_RETURN_FALSE(utils::WriteAll(data_fd, data.data(), data.size()));
  *data_file_size += data.size();

  // Now, insert into graph and blocks vector
  Vertex::Index vertex = existing_vertex;
  if (vertex == Vertex::kInvalidIndex) {
    graph->resize(graph->size() + 1);
    vertex = graph->size() - 1;
  }
  (*graph)[vertex].op = operation;
  CHECK((*graph)[vertex].op.has_type());
  (*graph)[vertex].file_name = path;
  (*graph)[vertex].chunk_offset = chunk_offset;
  (*graph)[vertex].chunk_size = chunk_size;

  if (blocks)
    TEST_AND_RETURN_FALSE(DeltaDiffGenerator::AddInstallOpToBlocksVector(
        (*graph)[vertex].op,
        *graph,
        vertex,
        blocks));
  return true;
}

// For each regular file within new_root, creates a node in the graph,
// determines the best way to compress it (REPLACE, REPLACE_BZ, COPY, BSDIFF),
// and writes any necessary data to the end of data_fd.
bool DeltaReadFiles(Graph* graph,
                    vector<Block>* blocks,
                    const string& old_root,
                    const string& new_root,
                    off_t chunk_size,
                    int data_fd,
                    off_t* data_file_size) {
  set<ino_t> visited_inodes;
  set<ino_t> visited_src_inodes;
  for (FilesystemIterator fs_iter(new_root,
                                  set<string>{"/lost+found"});
       !fs_iter.IsEnd(); fs_iter.Increment()) {
    // We never diff symlinks (here, we check that dst file is not a symlink).
    if (!S_ISREG(fs_iter.GetStat().st_mode))
      continue;

    // Make sure we visit each inode only once.
    if (utils::SetContainsKey(visited_inodes, fs_iter.GetStat().st_ino))
      continue;
    visited_inodes.insert(fs_iter.GetStat().st_ino);
    off_t dst_size = fs_iter.GetFileSize();
    if (dst_size == 0)
      continue;

    LOG(INFO) << "Encoding file " << fs_iter.GetPartialPath();

    // We can't visit each dst image inode more than once, as that would
    // duplicate work. Here, we avoid visiting each source image inode
    // more than once. Technically, we could have multiple operations
    // that read the same blocks from the source image for diffing, but
    // we choose not to avoid complexity. Eventually we will move away
    // from using a graph/cycle detection/etc to generate diffs, and at that
    // time, it will be easy (non-complex) to have many operations read
    // from the same source blocks. At that time, this code can die. -adlr
    bool should_diff_from_source = false;
    string src_path = old_root + fs_iter.GetPartialPath();
    struct stat src_stbuf;
    // We never diff symlinks (here, we check that src file is not a symlink).
    if (0 == lstat(src_path.c_str(), &src_stbuf) &&
        S_ISREG(src_stbuf.st_mode)) {
      should_diff_from_source = !utils::SetContainsKey(visited_src_inodes,
                                                       src_stbuf.st_ino);
      visited_src_inodes.insert(src_stbuf.st_ino);
    }

    off_t size = chunk_size == -1 ? dst_size : chunk_size;
    off_t step = size;
    for (off_t offset = 0; offset < dst_size; offset += step) {
      if (offset + size >= dst_size) {
        size = -1;  // Read through the end of the file.
      }
      TEST_AND_RETURN_FALSE(DeltaReadFile(graph,
                                          Vertex::kInvalidIndex,
                                          blocks,
                                          (should_diff_from_source ?
                                           old_root :
                                           kEmptyPath),
                                          new_root,
                                          fs_iter.GetPartialPath(),
                                          offset,
                                          size,
                                          data_fd,
                                          data_file_size));
    }
  }
  return true;
}

// This class allocates non-existent temp blocks, starting from
// kTempBlockStart. Other code is responsible for converting these
// temp blocks into real blocks, as the client can't read or write to
// these blocks.
class DummyExtentAllocator {
 public:
  DummyExtentAllocator() : next_block_(kTempBlockStart) {}
  vector<Extent> Allocate(const uint64_t block_count) {
    vector<Extent> ret(1);
    ret[0].set_start_block(next_block_);
    ret[0].set_num_blocks(block_count);
    next_block_ += block_count;
    return ret;
  }
 private:
  uint64_t next_block_;
};

// Reads blocks from image_path that are not yet marked as being written
// in the blocks array. These blocks that remain are non-file-data blocks.
// In the future we might consider intelligent diffing between this data
// and data in the previous image, but for now we just bzip2 compress it
// and include it in the update.
// Creates a new node in the graph to write these blocks and writes the
// appropriate blob to blobs_fd. Reads and updates blobs_length;
bool ReadUnwrittenBlocks(const vector<Block>& blocks,
                         int blobs_fd,
                         off_t* blobs_length,
                         const string& image_path,
                         Vertex* vertex) {
  vertex->file_name = "<rootfs-non-file-data>";

  DeltaArchiveManifest_InstallOperation* out_op = &vertex->op;
  int image_fd = open(image_path.c_str(), O_RDONLY, 000);
  TEST_AND_RETURN_FALSE_ERRNO(image_fd >= 0);
  ScopedFdCloser image_fd_closer(&image_fd);

  string temp_file_path;
  TEST_AND_RETURN_FALSE(utils::MakeTempFile("CrAU_temp_data.XXXXXX",
                                            &temp_file_path,
                                            nullptr));

  FILE* file = fopen(temp_file_path.c_str(), "w");
  TEST_AND_RETURN_FALSE(file);
  int err = BZ_OK;

  BZFILE* bz_file = BZ2_bzWriteOpen(&err,
                                    file,
                                    9,  // max compression
                                    0,  // verbosity
                                    0);  // default work factor
  TEST_AND_RETURN_FALSE(err == BZ_OK);

  vector<Extent> extents;
  vector<Block>::size_type block_count = 0;

  LOG(INFO) << "Appending left over blocks to extents";
  for (vector<Block>::size_type i = 0; i < blocks.size(); i++) {
    if (blocks[i].writer != Vertex::kInvalidIndex)
      continue;
    if (blocks[i].reader != Vertex::kInvalidIndex) {
      graph_utils::AddReadBeforeDep(vertex, blocks[i].reader, i);
    }
    graph_utils::AppendBlockToExtents(&extents, i);
    block_count++;
  }

  // Code will handle 'buf' at any size that's a multiple of kBlockSize,
  // so we arbitrarily set it to 1024 * kBlockSize.
  chromeos::Blob buf(1024 * kBlockSize);

  LOG(INFO) << "Reading left over blocks";
  vector<Block>::size_type blocks_copied_count = 0;

  // For each extent in extents, write the data into BZ2_bzWrite which
  // sends it to an output file.
  // We use the temporary buffer 'buf' to hold the data, which may be
  // smaller than the extent, so in that case we have to loop to get
  // the extent's data (that's the inner while loop).
  for (const Extent& extent : extents) {
    vector<Block>::size_type blocks_read = 0;
    float printed_progress = -1;
    while (blocks_read < extent.num_blocks()) {
      const int copy_block_cnt =
          min(buf.size() / kBlockSize,
              static_cast<chromeos::Blob::size_type>(
                  extent.num_blocks() - blocks_read));
      ssize_t rc = pread(image_fd,
                         buf.data(),
                         copy_block_cnt * kBlockSize,
                         (extent.start_block() + blocks_read) * kBlockSize);
      TEST_AND_RETURN_FALSE_ERRNO(rc >= 0);
      TEST_AND_RETURN_FALSE(static_cast<size_t>(rc) ==
                            copy_block_cnt * kBlockSize);
      BZ2_bzWrite(&err, bz_file, buf.data(), copy_block_cnt * kBlockSize);
      TEST_AND_RETURN_FALSE(err == BZ_OK);
      blocks_read += copy_block_cnt;
      blocks_copied_count += copy_block_cnt;
      float current_progress =
          static_cast<float>(blocks_copied_count) / block_count;
      if (printed_progress + 0.1 < current_progress ||
          blocks_copied_count == block_count) {
        LOG(INFO) << "progress: " << current_progress;
        printed_progress = current_progress;
      }
    }
  }
  BZ2_bzWriteClose(&err, bz_file, 0, nullptr, nullptr);
  TEST_AND_RETURN_FALSE(err == BZ_OK);
  bz_file = nullptr;
  TEST_AND_RETURN_FALSE_ERRNO(0 == fclose(file));
  file = nullptr;

  chromeos::Blob compressed_data;
  LOG(INFO) << "Reading compressed data off disk";
  TEST_AND_RETURN_FALSE(utils::ReadFile(temp_file_path, &compressed_data));
  TEST_AND_RETURN_FALSE(unlink(temp_file_path.c_str()) == 0);

  // Add node to graph to write these blocks
  out_op->set_type(DeltaArchiveManifest_InstallOperation_Type_REPLACE_BZ);
  out_op->set_data_offset(*blobs_length);
  out_op->set_data_length(compressed_data.size());
  LOG(INFO) << "Rootfs non-data blocks compressed take up "
            << compressed_data.size();
  *blobs_length += compressed_data.size();
  out_op->set_dst_length(kBlockSize * block_count);
  DeltaDiffGenerator::StoreExtents(extents, out_op->mutable_dst_extents());

  TEST_AND_RETURN_FALSE(utils::WriteAll(blobs_fd,
                                        compressed_data.data(),
                                        compressed_data.size()));
  LOG(INFO) << "done with extra blocks";
  return true;
}

// Writes the uint64_t passed in in host-endian to the file as big-endian.
// Returns true on success.
bool WriteUint64AsBigEndian(FileWriter* writer, const uint64_t value) {
  uint64_t value_be = htobe64(value);
  TEST_AND_RETURN_FALSE(writer->Write(&value_be, sizeof(value_be)));
  return true;
}

// Adds each operation from |graph| to |out_manifest| in the order specified by
// |order| while building |out_op_name_map| with operation to name
// mappings. Adds all |kernel_ops| to |out_manifest|. Filters out no-op
// operations.
void InstallOperationsToManifest(
    const Graph& graph,
    const vector<Vertex::Index>& order,
    const vector<DeltaArchiveManifest_InstallOperation>& kernel_ops,
    DeltaArchiveManifest* out_manifest,
    OperationNameMap* out_op_name_map) {
  for (Vertex::Index vertex_index : order) {
    const Vertex& vertex = graph[vertex_index];
    const DeltaArchiveManifest_InstallOperation& add_op = vertex.op;
    if (DeltaDiffGenerator::IsNoopOperation(add_op)) {
      continue;
    }
    DeltaArchiveManifest_InstallOperation* op =
        out_manifest->add_install_operations();
    *op = add_op;
    string name = vertex.file_name;
    if (vertex.chunk_offset || vertex.chunk_size != -1) {
      string offset = base::Int64ToString(vertex.chunk_offset);
      if (vertex.chunk_size != -1) {
        name += " [" + offset + ", " +
            base::Int64ToString(vertex.chunk_offset + vertex.chunk_size - 1) +
            "]";
      } else {
        name += " [" + offset + ", end]";
      }
    }
    (*out_op_name_map)[op] = name;
  }
  for (vector<DeltaArchiveManifest_InstallOperation>::const_iterator it =
           kernel_ops.begin(); it != kernel_ops.end(); ++it) {
    const DeltaArchiveManifest_InstallOperation& add_op = *it;
    if (DeltaDiffGenerator::IsNoopOperation(add_op)) {
      continue;
    }
    DeltaArchiveManifest_InstallOperation* op =
        out_manifest->add_kernel_install_operations();
    *op = add_op;
  }
}

void CheckGraph(const Graph& graph) {
  for (const Vertex& v : graph) {
    CHECK(v.op.has_type());
  }
}

// Delta compresses a kernel partition |new_kernel_part| with knowledge of the
// old kernel partition |old_kernel_part|. If |old_kernel_part| is an empty
// string, generates a full update of the partition.
bool DeltaCompressKernelPartition(
    const string& old_kernel_part,
    const string& new_kernel_part,
    vector<DeltaArchiveManifest_InstallOperation>* ops,
    int blobs_fd,
    off_t* blobs_length) {
  LOG(INFO) << "Delta compressing kernel partition...";
  LOG_IF(INFO, old_kernel_part.empty()) << "Generating full kernel update...";

  DeltaArchiveManifest_InstallOperation op;
  chromeos::Blob data;
  TEST_AND_RETURN_FALSE(
      DeltaDiffGenerator::ReadFileToDiff(old_kernel_part,
                                         new_kernel_part,
                                         0,  // chunk_offset
                                         -1,  // chunk_size
                                         true,  // bsdiff_allowed
                                         &data,
                                         &op,
                                         false));

  // Check if the operation writes nothing.
  if (op.dst_extents_size() == 0) {
    if (op.type() == DeltaArchiveManifest_InstallOperation_Type_MOVE) {
      LOG(INFO) << "Empty MOVE operation, nothing to do.";
      return true;
    } else {
      LOG(ERROR) << "Empty non-MOVE operation";
      return false;
    }
  }

  // Write the data.
  if (op.type() != DeltaArchiveManifest_InstallOperation_Type_MOVE) {
    op.set_data_offset(*blobs_length);
    op.set_data_length(data.size());
  }

  // Add the new install operation.
  ops->clear();
  ops->push_back(op);

  TEST_AND_RETURN_FALSE(utils::WriteAll(blobs_fd, data.data(), data.size()));
  *blobs_length += data.size();

  LOG(INFO) << "Done delta compressing kernel partition: "
            << kInstallOperationTypes[op.type()];
  return true;
}

struct DeltaObject {
  DeltaObject(const string& in_name, const int in_type, const off_t in_size)
      : name(in_name),
        type(in_type),
        size(in_size) {}
  bool operator <(const DeltaObject& object) const {
    return (size != object.size) ? (size < object.size) : (name < object.name);
  }
  string name;
  int type;
  off_t size;
};

void ReportPayloadUsage(const DeltaArchiveManifest& manifest,
                        const int64_t manifest_metadata_size,
                        const OperationNameMap& op_name_map) {
  vector<DeltaObject> objects;
  off_t total_size = 0;

  // Rootfs install operations.
  for (int i = 0; i < manifest.install_operations_size(); ++i) {
    const DeltaArchiveManifest_InstallOperation& op =
        manifest.install_operations(i);
    objects.push_back(DeltaObject(op_name_map.find(&op)->second,
                                  op.type(),
                                  op.data_length()));
    total_size += op.data_length();
  }

  // Kernel install operations.
  for (int i = 0; i < manifest.kernel_install_operations_size(); ++i) {
    const DeltaArchiveManifest_InstallOperation& op =
        manifest.kernel_install_operations(i);
    objects.push_back(DeltaObject(base::StringPrintf("<kernel-operation-%d>",
                                                     i),
                                  op.type(),
                                  op.data_length()));
    total_size += op.data_length();
  }

  objects.push_back(DeltaObject("<manifest-metadata>",
                                -1,
                                manifest_metadata_size));
  total_size += manifest_metadata_size;

  std::sort(objects.begin(), objects.end());

  static const char kFormatString[] = "%6.2f%% %10jd %-10s %s\n";
  for (const DeltaObject& object : objects) {
    fprintf(stderr, kFormatString,
            object.size * 100.0 / total_size,
            static_cast<intmax_t>(object.size),
            object.type >= 0 ? kInstallOperationTypes[object.type] : "-",
            object.name.c_str());
  }
  fprintf(stderr, kFormatString,
          100.0, static_cast<intmax_t>(total_size), "", "<total>");
}

// Process a range of blocks from |range_start| to |range_end| in the extent at
// position |*idx_p| of |extents|. If |do_remove| is true, this range will be
// removed, which may cause the extent to be trimmed, split or removed entirely.
// The value of |*idx_p| is updated to point to the next extent to be processed.
// Returns true iff the next extent to process is a new or updated one.
bool ProcessExtentBlockRange(vector<Extent>* extents, size_t* idx_p,
                             const bool do_remove, uint64_t range_start,
                             uint64_t range_end) {
  size_t idx = *idx_p;
  uint64_t start_block = (*extents)[idx].start_block();
  uint64_t num_blocks = (*extents)[idx].num_blocks();
  uint64_t range_size = range_end - range_start;

  if (do_remove) {
    if (range_size == num_blocks) {
      // Remove the entire extent.
      extents->erase(extents->begin() + idx);
    } else if (range_end == num_blocks) {
      // Trim the end of the extent.
      (*extents)[idx].set_num_blocks(num_blocks - range_size);
      idx++;
    } else if (range_start == 0) {
      // Trim the head of the extent.
      (*extents)[idx].set_start_block(start_block + range_size);
      (*extents)[idx].set_num_blocks(num_blocks - range_size);
    } else {
      // Trim the middle, splitting the remainder into two parts.
      (*extents)[idx].set_num_blocks(range_start);
      Extent e;
      e.set_start_block(start_block + range_end);
      e.set_num_blocks(num_blocks - range_end);
      idx++;
      extents->insert(extents->begin() + idx, e);
    }
  } else if (range_end == num_blocks) {
    // Done with this extent.
    idx++;
  } else {
    return false;
  }

  *idx_p = idx;
  return true;
}

// Remove identical corresponding block ranges in |src_extents| and
// |dst_extents|. Used for preventing moving of blocks onto themselves during
// MOVE operations. The value of |total_bytes| indicates the actual length of
// content; this may be slightly less than the total size of blocks, in which
// case the last block is only partly occupied with data. Returns the total
// number of bytes removed.
size_t RemoveIdenticalBlockRanges(vector<Extent>* src_extents,
                                  vector<Extent>* dst_extents,
                                  const size_t total_bytes) {
  size_t src_idx = 0;
  size_t dst_idx = 0;
  uint64_t src_offset = 0, dst_offset = 0;
  bool new_src = true, new_dst = true;
  size_t removed_bytes = 0, nonfull_block_bytes;
  bool do_remove = false;
  while (src_idx < src_extents->size() && dst_idx < dst_extents->size()) {
    if (new_src) {
      src_offset = 0;
      new_src = false;
    }
    if (new_dst) {
      dst_offset = 0;
      new_dst = false;
    }

    do_remove = ((*src_extents)[src_idx].start_block() + src_offset ==
                 (*dst_extents)[dst_idx].start_block() + dst_offset);

    uint64_t src_num_blocks = (*src_extents)[src_idx].num_blocks();
    uint64_t dst_num_blocks = (*dst_extents)[dst_idx].num_blocks();
    uint64_t min_num_blocks = min(src_num_blocks - src_offset,
                                  dst_num_blocks - dst_offset);
    uint64_t prev_src_offset = src_offset;
    uint64_t prev_dst_offset = dst_offset;
    src_offset += min_num_blocks;
    dst_offset += min_num_blocks;

    new_src = ProcessExtentBlockRange(src_extents, &src_idx, do_remove,
                                      prev_src_offset, src_offset);
    new_dst = ProcessExtentBlockRange(dst_extents, &dst_idx, do_remove,
                                      prev_dst_offset, dst_offset);
    if (do_remove)
      removed_bytes += min_num_blocks * kBlockSize;
  }

  // If we removed the last block and this block is only partly used by file
  // content, deduct the unused portion from the total removed byte count.
  if (do_remove && (nonfull_block_bytes = total_bytes % kBlockSize))
    removed_bytes -= kBlockSize - nonfull_block_bytes;

  return removed_bytes;
}

}  // namespace

bool DeltaDiffGenerator::ReadFileToDiff(
    const string& old_filename,
    const string& new_filename,
    off_t chunk_offset,
    off_t chunk_size,
    bool bsdiff_allowed,
    chromeos::Blob* out_data,
    DeltaArchiveManifest_InstallOperation* out_op,
    bool gather_extents) {
  // Read new data in
  chromeos::Blob new_data;
  TEST_AND_RETURN_FALSE(
      utils::ReadFileChunk(new_filename, chunk_offset, chunk_size, &new_data));

  TEST_AND_RETURN_FALSE(!new_data.empty());
  TEST_AND_RETURN_FALSE(chunk_size == -1 ||
                        static_cast<off_t>(new_data.size()) <= chunk_size);

  chromeos::Blob new_data_bz;
  TEST_AND_RETURN_FALSE(BzipCompress(new_data, &new_data_bz));
  CHECK(!new_data_bz.empty());

  chromeos::Blob data;  // Data blob that will be written to delta file.

  DeltaArchiveManifest_InstallOperation operation;
  size_t current_best_size = 0;
  if (new_data.size() <= new_data_bz.size()) {
    operation.set_type(DeltaArchiveManifest_InstallOperation_Type_REPLACE);
    current_best_size = new_data.size();
    data = new_data;
  } else {
    operation.set_type(DeltaArchiveManifest_InstallOperation_Type_REPLACE_BZ);
    current_best_size = new_data_bz.size();
    data = new_data_bz;
  }

  // Do we have an original file to consider?
  off_t old_size = 0;
  bool original = !old_filename.empty();
  if (original && (old_size = utils::FileSize(old_filename)) < 0) {
    // If stat-ing the old file fails, it should be because it doesn't exist.
    TEST_AND_RETURN_FALSE(!utils::FileExists(old_filename.c_str()));
    original = false;
  }

  chromeos::Blob old_data;
  if (original) {
    // Read old data
    TEST_AND_RETURN_FALSE(
        utils::ReadFileChunk(
            old_filename, chunk_offset, chunk_size, &old_data));
    if (old_data == new_data) {
      // No change in data.
      operation.set_type(DeltaArchiveManifest_InstallOperation_Type_MOVE);
      current_best_size = 0;
      data.clear();
    } else if (!old_data.empty() && bsdiff_allowed) {
      // If the source file is considered bsdiff safe (no bsdiff bugs
      // triggered), see if BSDIFF encoding is smaller.
      base::FilePath old_chunk;
      TEST_AND_RETURN_FALSE(base::CreateTemporaryFile(&old_chunk));
      ScopedPathUnlinker old_unlinker(old_chunk.value());
      TEST_AND_RETURN_FALSE(
          utils::WriteFile(old_chunk.value().c_str(),
                           old_data.data(), old_data.size()));
      base::FilePath new_chunk;
      TEST_AND_RETURN_FALSE(base::CreateTemporaryFile(&new_chunk));
      ScopedPathUnlinker new_unlinker(new_chunk.value());
      TEST_AND_RETURN_FALSE(
          utils::WriteFile(new_chunk.value().c_str(),
                           new_data.data(), new_data.size()));

      chromeos::Blob bsdiff_delta;
      TEST_AND_RETURN_FALSE(
          BsdiffFiles(old_chunk.value(), new_chunk.value(), &bsdiff_delta));
      CHECK_GT(bsdiff_delta.size(), static_cast<chromeos::Blob::size_type>(0));
      if (bsdiff_delta.size() < current_best_size) {
        operation.set_type(DeltaArchiveManifest_InstallOperation_Type_BSDIFF);
        current_best_size = bsdiff_delta.size();
        data = bsdiff_delta;
      }
    }
  }

  // Set parameters of the operations
  CHECK_EQ(data.size(), current_best_size);

  vector<Extent> src_extents, dst_extents;
  if (operation.type() == DeltaArchiveManifest_InstallOperation_Type_MOVE ||
      operation.type() == DeltaArchiveManifest_InstallOperation_Type_BSDIFF) {
    if (gather_extents) {
      TEST_AND_RETURN_FALSE(
          GatherExtents(old_filename,
                        chunk_offset,
                        chunk_size,
                        &src_extents));
    } else {
      Extent* src_extent = operation.add_src_extents();
      src_extent->set_start_block(0);
      src_extent->set_num_blocks((old_size + kBlockSize - 1) / kBlockSize);
    }
    operation.set_src_length(old_data.size());
  }

  if (gather_extents) {
    TEST_AND_RETURN_FALSE(
        GatherExtents(new_filename,
                      chunk_offset,
                      chunk_size,
                      &dst_extents));
  } else {
    Extent* dst_extent = operation.add_dst_extents();
    dst_extent->set_start_block(0);
    dst_extent->set_num_blocks((new_data.size() + kBlockSize - 1) / kBlockSize);
  }
  operation.set_dst_length(new_data.size());

  if (gather_extents) {
    // Remove identical src/dst block ranges in MOVE operations.
    if (operation.type() == DeltaArchiveManifest_InstallOperation_Type_MOVE) {
      size_t removed_bytes = RemoveIdenticalBlockRanges(
          &src_extents, &dst_extents, new_data.size());

      // Adjust the file length field accordingly.
      if (removed_bytes) {
        operation.set_src_length(old_data.size() - removed_bytes);
        operation.set_dst_length(new_data.size() - removed_bytes);
      }
    }

    // Embed extents in the operation.
    DeltaDiffGenerator::StoreExtents(src_extents,
                                     operation.mutable_src_extents());
    DeltaDiffGenerator::StoreExtents(dst_extents,
                                     operation.mutable_dst_extents());
  }

  out_data->swap(data);
  *out_op = operation;

  return true;
}

bool DeltaDiffGenerator::InitializePartitionInfo(bool is_kernel,
                                                 const string& partition,
                                                 PartitionInfo* info) {
  int64_t size = 0;
  if (is_kernel) {
    size = utils::FileSize(partition);
  } else {
    int block_count = 0, block_size = 0;
    TEST_AND_RETURN_FALSE(utils::GetFilesystemSize(partition,
                                                   &block_count,
                                                   &block_size));
    size = static_cast<int64_t>(block_count) * block_size;
  }
  TEST_AND_RETURN_FALSE(size > 0);
  info->set_size(size);
  OmahaHashCalculator hasher;
  TEST_AND_RETURN_FALSE(hasher.UpdateFile(partition, size) == size);
  TEST_AND_RETURN_FALSE(hasher.Finalize());
  const chromeos::Blob& hash = hasher.raw_hash();
  info->set_hash(hash.data(), hash.size());
  LOG(INFO) << partition << ": size=" << size << " hash=" << hasher.hash();
  return true;
}

bool InitializePartitionInfos(const string& old_kernel,
                              const string& new_kernel,
                              const string& old_rootfs,
                              const string& new_rootfs,
                              DeltaArchiveManifest* manifest) {
  if (!old_kernel.empty()) {
    TEST_AND_RETURN_FALSE(DeltaDiffGenerator::InitializePartitionInfo(
        true,
        old_kernel,
        manifest->mutable_old_kernel_info()));
  }
  TEST_AND_RETURN_FALSE(DeltaDiffGenerator::InitializePartitionInfo(
      true,
      new_kernel,
      manifest->mutable_new_kernel_info()));
  if (!old_rootfs.empty()) {
    TEST_AND_RETURN_FALSE(DeltaDiffGenerator::InitializePartitionInfo(
        false,
        old_rootfs,
        manifest->mutable_old_rootfs_info()));
  }
  TEST_AND_RETURN_FALSE(DeltaDiffGenerator::InitializePartitionInfo(
      false,
      new_rootfs,
      manifest->mutable_new_rootfs_info()));
  return true;
}

namespace {

// Takes a collection (vector or RepeatedPtrField) of Extent and
// returns a vector of the blocks referenced, in order.
template<typename T>
vector<uint64_t> ExpandExtents(const T& extents) {
  vector<uint64_t> ret;
  for (size_t i = 0, e = static_cast<size_t>(extents.size()); i != e; ++i) {
    const Extent extent = graph_utils::GetElement(extents, i);
    if (extent.start_block() == kSparseHole) {
      ret.resize(ret.size() + extent.num_blocks(), kSparseHole);
    } else {
      for (uint64_t block = extent.start_block();
           block < (extent.start_block() + extent.num_blocks()); block++) {
        ret.push_back(block);
      }
    }
  }
  return ret;
}

// Takes a vector of blocks and returns an equivalent vector of Extent
// objects.
vector<Extent> CompressExtents(const vector<uint64_t>& blocks) {
  vector<Extent> new_extents;
  for (uint64_t block : blocks) {
    graph_utils::AppendBlockToExtents(&new_extents, block);
  }
  return new_extents;
}

}  // namespace

void DeltaDiffGenerator::SubstituteBlocks(
    Vertex* vertex,
    const vector<Extent>& remove_extents,
    const vector<Extent>& replace_extents) {
  // First, expand out the blocks that op reads from
  vector<uint64_t> read_blocks = ExpandExtents(vertex->op.src_extents());
  {
    // Expand remove_extents and replace_extents
    vector<uint64_t> remove_extents_expanded =
        ExpandExtents(remove_extents);
    vector<uint64_t> replace_extents_expanded =
        ExpandExtents(replace_extents);
    CHECK_EQ(remove_extents_expanded.size(), replace_extents_expanded.size());
    map<uint64_t, uint64_t> conversion;
    for (vector<uint64_t>::size_type i = 0;
         i < replace_extents_expanded.size(); i++) {
      conversion[remove_extents_expanded[i]] = replace_extents_expanded[i];
    }
    utils::ApplyMap(&read_blocks, conversion);
    for (auto& edge_prop_pair : vertex->out_edges) {
      vector<uint64_t> write_before_deps_expanded =
          ExpandExtents(edge_prop_pair.second.write_extents);
      utils::ApplyMap(&write_before_deps_expanded, conversion);
      edge_prop_pair.second.write_extents =
          CompressExtents(write_before_deps_expanded);
    }
  }
  // Convert read_blocks back to extents
  vertex->op.clear_src_extents();
  vector<Extent> new_extents = CompressExtents(read_blocks);
  DeltaDiffGenerator::StoreExtents(new_extents,
                                   vertex->op.mutable_src_extents());
}

bool DeltaDiffGenerator::CutEdges(Graph* graph,
                                  const set<Edge>& edges,
                                  vector<CutEdgeVertexes>* out_cuts) {
  DummyExtentAllocator scratch_allocator;
  vector<CutEdgeVertexes> cuts;
  cuts.reserve(edges.size());

  uint64_t scratch_blocks_used = 0;
  for (const Edge& edge : edges) {
    cuts.resize(cuts.size() + 1);
    vector<Extent> old_extents =
        (*graph)[edge.first].out_edges[edge.second].extents;
    // Choose some scratch space
    scratch_blocks_used += graph_utils::EdgeWeight(*graph, edge);
    cuts.back().tmp_extents =
        scratch_allocator.Allocate(graph_utils::EdgeWeight(*graph, edge));
    // create vertex to copy original->scratch
    cuts.back().new_vertex = graph->size();
    graph->resize(graph->size() + 1);
    cuts.back().old_src = edge.first;
    cuts.back().old_dst = edge.second;

    EdgeProperties& cut_edge_properties =
        (*graph)[edge.first].out_edges.find(edge.second)->second;

    // This should never happen, as we should only be cutting edges between
    // real file nodes, and write-before relationships are created from
    // a real file node to a temp copy node:
    CHECK(cut_edge_properties.write_extents.empty())
        << "Can't cut edge that has write-before relationship.";

    // make node depend on the copy operation
    (*graph)[edge.first].out_edges.insert(make_pair(graph->size() - 1,
                                                   cut_edge_properties));

    // Set src/dst extents and other proto variables for copy operation
    graph->back().op.set_type(DeltaArchiveManifest_InstallOperation_Type_MOVE);
    DeltaDiffGenerator::StoreExtents(
        cut_edge_properties.extents,
        graph->back().op.mutable_src_extents());
    DeltaDiffGenerator::StoreExtents(cuts.back().tmp_extents,
                                     graph->back().op.mutable_dst_extents());
    graph->back().op.set_src_length(
        graph_utils::EdgeWeight(*graph, edge) * kBlockSize);
    graph->back().op.set_dst_length(graph->back().op.src_length());

    // make the dest node read from the scratch space
    DeltaDiffGenerator::SubstituteBlocks(
        &((*graph)[edge.second]),
        (*graph)[edge.first].out_edges[edge.second].extents,
        cuts.back().tmp_extents);

    // delete the old edge
    CHECK_EQ(static_cast<Graph::size_type>(1),
             (*graph)[edge.first].out_edges.erase(edge.second));

    // Add an edge from dst to copy operation
    EdgeProperties write_before_edge_properties;
    write_before_edge_properties.write_extents = cuts.back().tmp_extents;
    (*graph)[edge.second].out_edges.insert(
        make_pair(graph->size() - 1, write_before_edge_properties));
  }
  out_cuts->swap(cuts);
  return true;
}

// Stores all Extents in 'extents' into 'out'.
void DeltaDiffGenerator::StoreExtents(
    const vector<Extent>& extents,
    google::protobuf::RepeatedPtrField<Extent>* out) {
  for (const Extent& extent : extents) {
    Extent* new_extent = out->Add();
    *new_extent = extent;
  }
}

// Creates all the edges for the graph. Writers of a block point to
// readers of the same block. This is because for an edge A->B, B
// must complete before A executes.
void DeltaDiffGenerator::CreateEdges(Graph* graph,
                                     const vector<Block>& blocks) {
  for (vector<Block>::size_type i = 0; i < blocks.size(); i++) {
    // Blocks with both a reader and writer get an edge
    if (blocks[i].reader == Vertex::kInvalidIndex ||
        blocks[i].writer == Vertex::kInvalidIndex)
      continue;
    // Don't have a node depend on itself
    if (blocks[i].reader == blocks[i].writer)
      continue;
    // See if there's already an edge we can add onto
    Vertex::EdgeMap::iterator edge_it =
        (*graph)[blocks[i].writer].out_edges.find(blocks[i].reader);
    if (edge_it == (*graph)[blocks[i].writer].out_edges.end()) {
      // No existing edge. Create one
      (*graph)[blocks[i].writer].out_edges.insert(
          make_pair(blocks[i].reader, EdgeProperties()));
      edge_it = (*graph)[blocks[i].writer].out_edges.find(blocks[i].reader);
      CHECK(edge_it != (*graph)[blocks[i].writer].out_edges.end());
    }
    graph_utils::AppendBlockToExtents(&edge_it->second.extents, i);
  }
}

namespace {

class SortCutsByTopoOrderLess {
 public:
  explicit SortCutsByTopoOrderLess(
      const vector<vector<Vertex::Index>::size_type>& table)
      : table_(table) {}
  bool operator()(const CutEdgeVertexes& a, const CutEdgeVertexes& b) {
    return table_[a.old_dst] < table_[b.old_dst];
  }
 private:
  const vector<vector<Vertex::Index>::size_type>& table_;
};

}  // namespace

void DeltaDiffGenerator::GenerateReverseTopoOrderMap(
    const vector<Vertex::Index>& op_indexes,
    vector<vector<Vertex::Index>::size_type>* reverse_op_indexes) {
  vector<vector<Vertex::Index>::size_type> table(op_indexes.size());
  for (vector<Vertex::Index>::size_type i = 0, e = op_indexes.size();
       i != e; ++i) {
    Vertex::Index node = op_indexes[i];
    if (table.size() < (node + 1)) {
      table.resize(node + 1);
    }
    table[node] = i;
  }
  reverse_op_indexes->swap(table);
}

void DeltaDiffGenerator::SortCutsByTopoOrder(
    const vector<Vertex::Index>& op_indexes,
    vector<CutEdgeVertexes>* cuts) {
  // first, make a reverse lookup table.
  vector<vector<Vertex::Index>::size_type> table;
  GenerateReverseTopoOrderMap(op_indexes, &table);
  SortCutsByTopoOrderLess less(table);
  sort(cuts->begin(), cuts->end(), less);
}

void DeltaDiffGenerator::MoveFullOpsToBack(Graph* graph,
                                           vector<Vertex::Index>* op_indexes) {
  vector<Vertex::Index> ret;
  vector<Vertex::Index> full_ops;
  ret.reserve(op_indexes->size());
  for (vector<Vertex::Index>::size_type i = 0, e = op_indexes->size(); i != e;
       ++i) {
    DeltaArchiveManifest_InstallOperation_Type type =
        (*graph)[(*op_indexes)[i]].op.type();
    if (type == DeltaArchiveManifest_InstallOperation_Type_REPLACE ||
        type == DeltaArchiveManifest_InstallOperation_Type_REPLACE_BZ) {
      full_ops.push_back((*op_indexes)[i]);
    } else {
      ret.push_back((*op_indexes)[i]);
    }
  }
  LOG(INFO) << "Stats: " << full_ops.size() << " full ops out of "
            << (full_ops.size() + ret.size()) << " total ops.";
  ret.insert(ret.end(), full_ops.begin(), full_ops.end());
  op_indexes->swap(ret);
}

namespace {

template<typename T>
bool TempBlocksExistInExtents(const T& extents) {
  for (int i = 0, e = extents.size(); i < e; ++i) {
    Extent extent = graph_utils::GetElement(extents, i);
    uint64_t start = extent.start_block();
    uint64_t num = extent.num_blocks();
    if (start == kSparseHole)
      continue;
    if (start >= kTempBlockStart ||
        (start + num) >= kTempBlockStart) {
      LOG(ERROR) << "temp block!";
      LOG(ERROR) << "start: " << start << ", num: " << num;
      LOG(ERROR) << "kTempBlockStart: " << kTempBlockStart;
      LOG(ERROR) << "returning true";
      return true;
    }
    // check for wrap-around, which would be a bug:
    CHECK(start <= (start + num));
  }
  return false;
}

// Converts the cuts, which must all have the same |old_dst| member,
// to full. It does this by converting the |old_dst| to REPLACE or
// REPLACE_BZ, dropping all incoming edges to |old_dst|, and marking
// all temp nodes invalid.
bool ConvertCutsToFull(
    Graph* graph,
    const string& new_root,
    int data_fd,
    off_t* data_file_size,
    vector<Vertex::Index>* op_indexes,
    vector<vector<Vertex::Index>::size_type>* reverse_op_indexes,
    const vector<CutEdgeVertexes>& cuts) {
  CHECK(!cuts.empty());
  set<Vertex::Index> deleted_nodes;
  for (const CutEdgeVertexes& cut : cuts) {
    TEST_AND_RETURN_FALSE(DeltaDiffGenerator::ConvertCutToFullOp(
        graph,
        cut,
        new_root,
        data_fd,
        data_file_size));
    deleted_nodes.insert(cut.new_vertex);
  }
  deleted_nodes.insert(cuts[0].old_dst);

  vector<Vertex::Index> new_op_indexes;
  new_op_indexes.reserve(op_indexes->size());
  for (Vertex::Index vertex_index : *op_indexes) {
    if (utils::SetContainsKey(deleted_nodes, vertex_index))
      continue;
    new_op_indexes.push_back(vertex_index);
  }
  new_op_indexes.push_back(cuts[0].old_dst);
  op_indexes->swap(new_op_indexes);
  DeltaDiffGenerator::GenerateReverseTopoOrderMap(*op_indexes,
                                                  reverse_op_indexes);
  return true;
}

// Tries to assign temp blocks for a collection of cuts, all of which share
// the same old_dst member. If temp blocks can't be found, old_dst will be
// converted to a REPLACE or REPLACE_BZ operation. Returns true on success,
// which can happen even if blocks are converted to full. Returns false
// on exceptional error cases.
bool AssignBlockForAdjoiningCuts(
    Graph* graph,
    const string& new_root,
    int data_fd,
    off_t* data_file_size,
    vector<Vertex::Index>* op_indexes,
    vector<vector<Vertex::Index>::size_type>* reverse_op_indexes,
    const vector<CutEdgeVertexes>& cuts) {
  CHECK(!cuts.empty());
  const Vertex::Index old_dst = cuts[0].old_dst;
  // Calculate # of blocks needed
  uint64_t blocks_needed = 0;
  vector<uint64_t> cuts_blocks_needed(cuts.size());
  for (vector<CutEdgeVertexes>::size_type i = 0; i < cuts.size(); ++i) {
    uint64_t cut_blocks_needed = 0;
    for (const Extent& extent : cuts[i].tmp_extents) {
      cut_blocks_needed += extent.num_blocks();
    }
    blocks_needed += cut_blocks_needed;
    cuts_blocks_needed[i] = cut_blocks_needed;
  }

  // Find enough blocks
  ExtentRanges scratch_ranges;
  // Each block that's supplying temp blocks and the corresponding blocks:
  typedef vector<pair<Vertex::Index, ExtentRanges>> SupplierVector;
  SupplierVector block_suppliers;
  uint64_t scratch_blocks_found = 0;
  for (vector<Vertex::Index>::size_type i = (*reverse_op_indexes)[old_dst] + 1,
           e = op_indexes->size(); i < e; ++i) {
    Vertex::Index test_node = (*op_indexes)[i];
    if (!(*graph)[test_node].valid)
      continue;
    // See if this node has sufficient blocks
    ExtentRanges ranges;
    ranges.AddRepeatedExtents((*graph)[test_node].op.dst_extents());
    ranges.SubtractExtent(ExtentForRange(
        kTempBlockStart, kSparseHole - kTempBlockStart));
    ranges.SubtractRepeatedExtents((*graph)[test_node].op.src_extents());
    // For now, for simplicity, subtract out all blocks in read-before
    // dependencies.
    for (Vertex::EdgeMap::const_iterator edge_i =
             (*graph)[test_node].out_edges.begin(),
             edge_e = (*graph)[test_node].out_edges.end();
         edge_i != edge_e; ++edge_i) {
      ranges.SubtractExtents(edge_i->second.extents);
    }
    if (ranges.blocks() == 0)
      continue;

    if (ranges.blocks() + scratch_blocks_found > blocks_needed) {
      // trim down ranges
      vector<Extent> new_ranges = ranges.GetExtentsForBlockCount(
          blocks_needed - scratch_blocks_found);
      ranges = ExtentRanges();
      ranges.AddExtents(new_ranges);
    }
    scratch_ranges.AddRanges(ranges);
    block_suppliers.push_back(make_pair(test_node, ranges));
    scratch_blocks_found += ranges.blocks();
    if (scratch_ranges.blocks() >= blocks_needed)
      break;
  }
  if (scratch_ranges.blocks() < blocks_needed) {
    LOG(INFO) << "Unable to find sufficient scratch";
    TEST_AND_RETURN_FALSE(ConvertCutsToFull(graph,
                                            new_root,
                                            data_fd,
                                            data_file_size,
                                            op_indexes,
                                            reverse_op_indexes,
                                            cuts));
    return true;
  }
  // Use the scratch we found
  TEST_AND_RETURN_FALSE(scratch_ranges.blocks() == scratch_blocks_found);

  // Make all the suppliers depend on this node
  for (const auto& index_range_pair : block_suppliers) {
    graph_utils::AddReadBeforeDepExtents(
        &(*graph)[index_range_pair.first],
        old_dst,
        index_range_pair.second.GetExtentsForBlockCount(
            index_range_pair.second.blocks()));
  }

  // Replace temp blocks in each cut
  for (vector<CutEdgeVertexes>::size_type i = 0; i < cuts.size(); ++i) {
    const CutEdgeVertexes& cut = cuts[i];
    vector<Extent> real_extents =
        scratch_ranges.GetExtentsForBlockCount(cuts_blocks_needed[i]);
    scratch_ranges.SubtractExtents(real_extents);

    // Fix the old dest node w/ the real blocks
    DeltaDiffGenerator::SubstituteBlocks(&(*graph)[old_dst],
                                         cut.tmp_extents,
                                         real_extents);

    // Fix the new node w/ the real blocks. Since the new node is just a
    // copy operation, we can replace all the dest extents w/ the real
    // blocks.
    DeltaArchiveManifest_InstallOperation *op = &(*graph)[cut.new_vertex].op;
    op->clear_dst_extents();
    DeltaDiffGenerator::StoreExtents(real_extents, op->mutable_dst_extents());
  }
  return true;
}

}  // namespace

// Returns true if |op| is a no-op operation that doesn't do any useful work
// (e.g., a move operation that copies blocks onto themselves).
bool DeltaDiffGenerator::IsNoopOperation(
    const DeltaArchiveManifest_InstallOperation& op) {
  return (op.type() == DeltaArchiveManifest_InstallOperation_Type_MOVE &&
          ExpandExtents(op.src_extents()) == ExpandExtents(op.dst_extents()));
}

bool DeltaDiffGenerator::AssignTempBlocks(
    Graph* graph,
    const string& new_root,
    int data_fd,
    off_t* data_file_size,
    vector<Vertex::Index>* op_indexes,
    vector<vector<Vertex::Index>::size_type>* reverse_op_indexes,
    const vector<CutEdgeVertexes>& cuts) {
  CHECK(!cuts.empty());

  // group of cuts w/ the same old_dst:
  vector<CutEdgeVertexes> cuts_group;

  for (vector<CutEdgeVertexes>::size_type i = cuts.size() - 1, e = 0;
       true ; --i) {
    LOG(INFO) << "Fixing temp blocks in cut " << i
              << ": old dst: " << cuts[i].old_dst << " new vertex: "
              << cuts[i].new_vertex << " path: "
              << (*graph)[cuts[i].old_dst].file_name;

    if (cuts_group.empty() || (cuts_group[0].old_dst == cuts[i].old_dst)) {
      cuts_group.push_back(cuts[i]);
    } else {
      CHECK(!cuts_group.empty());
      TEST_AND_RETURN_FALSE(AssignBlockForAdjoiningCuts(graph,
                                                        new_root,
                                                        data_fd,
                                                        data_file_size,
                                                        op_indexes,
                                                        reverse_op_indexes,
                                                        cuts_group));
      cuts_group.clear();
      cuts_group.push_back(cuts[i]);
    }

    if (i == e) {
      // break out of for() loop
      break;
    }
  }
  CHECK(!cuts_group.empty());
  TEST_AND_RETURN_FALSE(AssignBlockForAdjoiningCuts(graph,
                                                    new_root,
                                                    data_fd,
                                                    data_file_size,
                                                    op_indexes,
                                                    reverse_op_indexes,
                                                    cuts_group));
  return true;
}

bool DeltaDiffGenerator::NoTempBlocksRemain(const Graph& graph) {
  size_t idx = 0;
  for (Graph::const_iterator it = graph.begin(), e = graph.end(); it != e;
       ++it, ++idx) {
    if (!it->valid)
      continue;
    const DeltaArchiveManifest_InstallOperation& op = it->op;
    if (TempBlocksExistInExtents(op.dst_extents()) ||
        TempBlocksExistInExtents(op.src_extents())) {
      LOG(INFO) << "bad extents in node " << idx;
      LOG(INFO) << "so yeah";
      return false;
    }

    // Check out-edges:
    for (const auto& edge_prop_pair : it->out_edges) {
      if (TempBlocksExistInExtents(edge_prop_pair.second.extents) ||
          TempBlocksExistInExtents(edge_prop_pair.second.write_extents)) {
        LOG(INFO) << "bad out edge in node " << idx;
        LOG(INFO) << "so yeah";
        return false;
      }
    }
  }
  return true;
}

bool DeltaDiffGenerator::ReorderDataBlobs(
    DeltaArchiveManifest* manifest,
    const string& data_blobs_path,
    const string& new_data_blobs_path) {
  int in_fd = open(data_blobs_path.c_str(), O_RDONLY, 0);
  TEST_AND_RETURN_FALSE_ERRNO(in_fd >= 0);
  ScopedFdCloser in_fd_closer(&in_fd);

  DirectFileWriter writer;
  TEST_AND_RETURN_FALSE(
      writer.Open(new_data_blobs_path.c_str(),
                  O_WRONLY | O_TRUNC | O_CREAT,
                  0644) == 0);
  ScopedFileWriterCloser writer_closer(&writer);
  uint64_t out_file_size = 0;

  for (int i = 0; i < (manifest->install_operations_size() +
                       manifest->kernel_install_operations_size()); i++) {
    DeltaArchiveManifest_InstallOperation* op = nullptr;
    if (i < manifest->install_operations_size()) {
      op = manifest->mutable_install_operations(i);
    } else {
      op = manifest->mutable_kernel_install_operations(
          i - manifest->install_operations_size());
    }
    if (!op->has_data_offset())
      continue;
    CHECK(op->has_data_length());
    chromeos::Blob buf(op->data_length());
    ssize_t rc = pread(in_fd, buf.data(), buf.size(), op->data_offset());
    TEST_AND_RETURN_FALSE(rc == static_cast<ssize_t>(buf.size()));

    // Add the hash of the data blobs for this operation
    TEST_AND_RETURN_FALSE(AddOperationHash(op, buf));

    op->set_data_offset(out_file_size);
    TEST_AND_RETURN_FALSE(writer.Write(buf.data(), buf.size()));
    out_file_size += buf.size();
  }
  return true;
}

bool DeltaDiffGenerator::AddOperationHash(
    DeltaArchiveManifest_InstallOperation* op,
    const chromeos::Blob& buf) {
  OmahaHashCalculator hasher;

  TEST_AND_RETURN_FALSE(hasher.Update(buf.data(), buf.size()));
  TEST_AND_RETURN_FALSE(hasher.Finalize());

  const chromeos::Blob& hash = hasher.raw_hash();
  op->set_data_sha256_hash(hash.data(), hash.size());
  return true;
}

bool DeltaDiffGenerator::ConvertCutToFullOp(Graph* graph,
                                            const CutEdgeVertexes& cut,
                                            const string& new_root,
                                            int data_fd,
                                            off_t* data_file_size) {
  // Drop all incoming edges, keep all outgoing edges

  // Keep all outgoing edges
  if ((*graph)[cut.old_dst].op.type() !=
      DeltaArchiveManifest_InstallOperation_Type_REPLACE_BZ &&
      (*graph)[cut.old_dst].op.type() !=
      DeltaArchiveManifest_InstallOperation_Type_REPLACE) {
    Vertex::EdgeMap out_edges = (*graph)[cut.old_dst].out_edges;
    graph_utils::DropWriteBeforeDeps(&out_edges);

    TEST_AND_RETURN_FALSE(DeltaReadFile(graph,
                                        cut.old_dst,
                                        nullptr,
                                        kEmptyPath,
                                        new_root,
                                        (*graph)[cut.old_dst].file_name,
                                        (*graph)[cut.old_dst].chunk_offset,
                                        (*graph)[cut.old_dst].chunk_size,
                                        data_fd,
                                        data_file_size));

    (*graph)[cut.old_dst].out_edges = out_edges;

    // Right now we don't have doubly-linked edges, so we have to scan
    // the whole graph.
    graph_utils::DropIncomingEdgesTo(graph, cut.old_dst);
  }

  // Delete temp node
  (*graph)[cut.old_src].out_edges.erase(cut.new_vertex);
  CHECK((*graph)[cut.old_dst].out_edges.find(cut.new_vertex) ==
        (*graph)[cut.old_dst].out_edges.end());
  (*graph)[cut.new_vertex].valid = false;
  LOG(INFO) << "marked node invalid: " << cut.new_vertex;
  return true;
}

bool DeltaDiffGenerator::ConvertGraphToDag(Graph* graph,
                                           const string& new_root,
                                           int fd,
                                           off_t* data_file_size,
                                           vector<Vertex::Index>* final_order,
                                           Vertex::Index scratch_vertex) {
  CycleBreaker cycle_breaker;
  LOG(INFO) << "Finding cycles...";
  set<Edge> cut_edges;
  cycle_breaker.BreakCycles(*graph, &cut_edges);
  LOG(INFO) << "done finding cycles";
  CheckGraph(*graph);

  // Calculate number of scratch blocks needed

  LOG(INFO) << "Cutting cycles...";
  vector<CutEdgeVertexes> cuts;
  TEST_AND_RETURN_FALSE(CutEdges(graph, cut_edges, &cuts));
  LOG(INFO) << "done cutting cycles";
  LOG(INFO) << "There are " << cuts.size() << " cuts.";
  CheckGraph(*graph);

  LOG(INFO) << "Creating initial topological order...";
  TopologicalSort(*graph, final_order);
  LOG(INFO) << "done with initial topo order";
  CheckGraph(*graph);

  LOG(INFO) << "Moving full ops to the back";
  MoveFullOpsToBack(graph, final_order);
  LOG(INFO) << "done moving full ops to back";

  vector<vector<Vertex::Index>::size_type> inverse_final_order;
  GenerateReverseTopoOrderMap(*final_order, &inverse_final_order);

  SortCutsByTopoOrder(*final_order, &cuts);

  if (!cuts.empty())
    TEST_AND_RETURN_FALSE(AssignTempBlocks(graph,
                                           new_root,
                                           fd,
                                           data_file_size,
                                           final_order,
                                           &inverse_final_order,
                                           cuts));
  LOG(INFO) << "Making sure all temp blocks have been allocated";

  // Remove the scratch node, if any
  if (scratch_vertex != Vertex::kInvalidIndex) {
    final_order->erase(final_order->begin() +
                       inverse_final_order[scratch_vertex]);
    (*graph)[scratch_vertex].valid = false;
    GenerateReverseTopoOrderMap(*final_order, &inverse_final_order);
  }

  graph_utils::DumpGraph(*graph);
  CHECK(NoTempBlocksRemain(*graph));
  LOG(INFO) << "done making sure all temp blocks are allocated";
  return true;
}

void DeltaDiffGenerator::CreateScratchNode(uint64_t start_block,
                                           uint64_t num_blocks,
                                           Vertex* vertex) {
  vertex->file_name = "<scratch>";
  vertex->op.set_type(DeltaArchiveManifest_InstallOperation_Type_REPLACE_BZ);
  vertex->op.set_data_offset(0);
  vertex->op.set_data_length(0);
  Extent* extent = vertex->op.add_dst_extents();
  extent->set_start_block(start_block);
  extent->set_num_blocks(num_blocks);
}

bool DeltaDiffGenerator::GenerateDeltaUpdateFile(
    const string& old_root,
    const string& old_image,
    const string& new_root,
    const string& new_image,
    const string& old_kernel_part,
    const string& new_kernel_part,
    const string& output_path,
    const string& private_key_path,
    off_t chunk_size,
    size_t rootfs_partition_size,
    uint32_t minor_version,
    const ImageInfo* old_image_info,
    const ImageInfo* new_image_info,
    uint64_t* metadata_size) {
  TEST_AND_RETURN_FALSE(chunk_size == -1 || chunk_size % kBlockSize == 0);
  int old_image_block_count = 0, old_image_block_size = 0;
  int new_image_block_count = 0, new_image_block_size = 0;
  TEST_AND_RETURN_FALSE(utils::GetFilesystemSize(new_image,
                                                 &new_image_block_count,
                                                 &new_image_block_size));
  if (!old_image.empty()) {
    TEST_AND_RETURN_FALSE(utils::GetFilesystemSize(old_image,
                                                   &old_image_block_count,
                                                   &old_image_block_size));
    TEST_AND_RETURN_FALSE(old_image_block_size == new_image_block_size);
    LOG_IF(WARNING, old_image_block_count != new_image_block_count)
        << "Old and new images have different block counts.";

    // If new_image_info is present, old_image_info must be present.
    TEST_AND_RETURN_FALSE(!old_image_info == !new_image_info);
  } else {
    // old_image_info must not be present for a full update.
    TEST_AND_RETURN_FALSE(!old_image_info);
  }

  // Sanity checks for the partition size.
  TEST_AND_RETURN_FALSE(rootfs_partition_size % kBlockSize == 0);
  size_t fs_size = static_cast<size_t>(new_image_block_size) *
                   new_image_block_count;
  LOG(INFO) << "Rootfs partition size: " << rootfs_partition_size;
  LOG(INFO) << "Actual filesystem size: " << fs_size;
  TEST_AND_RETURN_FALSE(rootfs_partition_size >= fs_size);

  // Sanity check kernel partition arg
  TEST_AND_RETURN_FALSE(utils::FileSize(new_kernel_part) >= 0);

  vector<Block> blocks(max(old_image_block_count, new_image_block_count));
  LOG(INFO) << "Invalid block index: " << Vertex::kInvalidIndex;
  LOG(INFO) << "Block count: " << blocks.size();
  for (vector<Block>::size_type i = 0; i < blocks.size(); i++) {
    CHECK(blocks[i].reader == Vertex::kInvalidIndex);
    CHECK(blocks[i].writer == Vertex::kInvalidIndex);
  }
  Graph graph;
  CheckGraph(graph);

  const string kTempFileTemplate("CrAU_temp_data.XXXXXX");
  string temp_file_path;
  unique_ptr<ScopedPathUnlinker> temp_file_unlinker;
  off_t data_file_size = 0;

  LOG(INFO) << "Reading files...";

  // Create empty protobuf Manifest object
  DeltaArchiveManifest manifest;

  vector<DeltaArchiveManifest_InstallOperation> kernel_ops;

  vector<Vertex::Index> final_order;
  Vertex::Index scratch_vertex = Vertex::kInvalidIndex;
  {
    int fd;
    TEST_AND_RETURN_FALSE(
        utils::MakeTempFile(kTempFileTemplate, &temp_file_path, &fd));
    temp_file_unlinker.reset(new ScopedPathUnlinker(temp_file_path));
    TEST_AND_RETURN_FALSE(fd >= 0);
    ScopedFdCloser fd_closer(&fd);
    if (!old_image.empty()) {
      // Delta update

      // Set the minor version for this payload.
      LOG(INFO) << "Adding Delta Minor Version.";
      manifest.set_minor_version(minor_version);

      TEST_AND_RETURN_FALSE(DeltaReadFiles(&graph,
                                           &blocks,
                                           old_root,
                                           new_root,
                                           chunk_size,
                                           fd,
                                           &data_file_size));
      LOG(INFO) << "done reading normal files";
      CheckGraph(graph);

      LOG(INFO) << "Starting metadata processing";
      TEST_AND_RETURN_FALSE(Metadata::DeltaReadMetadata(&graph,
                                                        &blocks,
                                                        old_image,
                                                        new_image,
                                                        fd,
                                                        &data_file_size));
      LOG(INFO) << "Done metadata processing";
      CheckGraph(graph);

      graph.resize(graph.size() + 1);
      TEST_AND_RETURN_FALSE(ReadUnwrittenBlocks(blocks,
                                                fd,
                                                &data_file_size,
                                                new_image,
                                                &graph.back()));

      // Final scratch block (if there's space)
      if (blocks.size() < (rootfs_partition_size / kBlockSize)) {
        scratch_vertex = graph.size();
        graph.resize(graph.size() + 1);
        CreateScratchNode(blocks.size(),
                          (rootfs_partition_size / kBlockSize) - blocks.size(),
                          &graph.back());
      }

      // Read kernel partition
      TEST_AND_RETURN_FALSE(DeltaCompressKernelPartition(old_kernel_part,
                                                         new_kernel_part,
                                                         &kernel_ops,
                                                         fd,
                                                         &data_file_size));

      LOG(INFO) << "done reading kernel";
      CheckGraph(graph);

      LOG(INFO) << "Creating edges...";
      CreateEdges(&graph, blocks);
      LOG(INFO) << "Done creating edges";
      CheckGraph(graph);

      TEST_AND_RETURN_FALSE(ConvertGraphToDag(&graph,
                                              new_root,
                                              fd,
                                              &data_file_size,
                                              &final_order,
                                              scratch_vertex));
    } else {
      // Full update
      off_t new_image_size =
          static_cast<off_t>(new_image_block_count) * new_image_block_size;
      TEST_AND_RETURN_FALSE(FullUpdateGenerator::Run(&graph,
                                                     new_kernel_part,
                                                     new_image,
                                                     new_image_size,
                                                     fd,
                                                     &data_file_size,
                                                     kFullUpdateChunkSize,
                                                     kBlockSize,
                                                     &kernel_ops,
                                                     &final_order));

      // Set the minor version for this payload.
      LOG(INFO) << "Adding Full Minor Version.";
      manifest.set_minor_version(DeltaPerformer::kFullPayloadMinorVersion);
    }
  }

  if (old_image_info)
    *(manifest.mutable_old_image_info()) = *old_image_info;

  if (new_image_info)
    *(manifest.mutable_new_image_info()) = *new_image_info;

  OperationNameMap op_name_map;
  CheckGraph(graph);
  InstallOperationsToManifest(graph,
                              final_order,
                              kernel_ops,
                              &manifest,
                              &op_name_map);
  CheckGraph(graph);
  manifest.set_block_size(kBlockSize);

  // Reorder the data blobs with the newly ordered manifest
  string ordered_blobs_path;
  TEST_AND_RETURN_FALSE(utils::MakeTempFile(
      "CrAU_temp_data.ordered.XXXXXX",
      &ordered_blobs_path,
      nullptr));
  ScopedPathUnlinker ordered_blobs_unlinker(ordered_blobs_path);
  TEST_AND_RETURN_FALSE(ReorderDataBlobs(&manifest,
                                         temp_file_path,
                                         ordered_blobs_path));
  temp_file_unlinker.reset();

  // Check that install op blobs are in order.
  uint64_t next_blob_offset = 0;
  {
    for (int i = 0; i < (manifest.install_operations_size() +
                         manifest.kernel_install_operations_size()); i++) {
      DeltaArchiveManifest_InstallOperation* op =
          i < manifest.install_operations_size() ?
          manifest.mutable_install_operations(i) :
          manifest.mutable_kernel_install_operations(
              i - manifest.install_operations_size());
      if (op->has_data_offset()) {
        if (op->data_offset() != next_blob_offset) {
          LOG(FATAL) << "bad blob offset! " << op->data_offset() << " != "
                     << next_blob_offset;
        }
        next_blob_offset += op->data_length();
      }
    }
  }

  // Signatures appear at the end of the blobs. Note the offset in the
  // manifest
  if (!private_key_path.empty()) {
    uint64_t signature_blob_length = 0;
    TEST_AND_RETURN_FALSE(
        PayloadSigner::SignatureBlobLength(vector<string>(1, private_key_path),
                                           &signature_blob_length));
    AddSignatureOp(next_blob_offset, signature_blob_length, &manifest);
  }

  TEST_AND_RETURN_FALSE(InitializePartitionInfos(old_kernel_part,
                                                 new_kernel_part,
                                                 old_image,
                                                 new_image,
                                                 &manifest));

  // Serialize protobuf
  string serialized_manifest;

  CheckGraph(graph);
  TEST_AND_RETURN_FALSE(manifest.AppendToString(&serialized_manifest));
  CheckGraph(graph);

  LOG(INFO) << "Writing final delta file header...";
  DirectFileWriter writer;
  TEST_AND_RETURN_FALSE_ERRNO(writer.Open(output_path.c_str(),
                                          O_WRONLY | O_CREAT | O_TRUNC,
                                          0644) == 0);
  ScopedFileWriterCloser writer_closer(&writer);

  // Write header
  TEST_AND_RETURN_FALSE(writer.Write(kDeltaMagic, strlen(kDeltaMagic)));

  // Write version number
  TEST_AND_RETURN_FALSE(WriteUint64AsBigEndian(&writer, kVersionNumber));

  // Write protobuf length
  TEST_AND_RETURN_FALSE(WriteUint64AsBigEndian(&writer,
                                               serialized_manifest.size()));

  // Write protobuf
  LOG(INFO) << "Writing final delta file protobuf... "
            << serialized_manifest.size();
  TEST_AND_RETURN_FALSE(writer.Write(serialized_manifest.data(),
                                     serialized_manifest.size()));

  // Append the data blobs
  LOG(INFO) << "Writing final delta file data blobs...";
  int blobs_fd = open(ordered_blobs_path.c_str(), O_RDONLY, 0);
  ScopedFdCloser blobs_fd_closer(&blobs_fd);
  TEST_AND_RETURN_FALSE(blobs_fd >= 0);
  for (;;) {
    char buf[kBlockSize];
    ssize_t rc = read(blobs_fd, buf, sizeof(buf));
    if (0 == rc) {
      // EOF
      break;
    }
    TEST_AND_RETURN_FALSE_ERRNO(rc > 0);
    TEST_AND_RETURN_FALSE(writer.Write(buf, rc));
  }

  // Write signature blob.
  if (!private_key_path.empty()) {
    LOG(INFO) << "Signing the update...";
    chromeos::Blob signature_blob;
    TEST_AND_RETURN_FALSE(PayloadSigner::SignPayload(
        output_path,
        vector<string>(1, private_key_path),
        &signature_blob));
    TEST_AND_RETURN_FALSE(writer.Write(signature_blob.data(),
                                       signature_blob.size()));
  }

  *metadata_size =
      strlen(kDeltaMagic) + 2 * sizeof(uint64_t) + serialized_manifest.size();
  ReportPayloadUsage(manifest, *metadata_size, op_name_map);

  LOG(INFO) << "All done. Successfully created delta file with "
            << "metadata size = " << *metadata_size;
  return true;
}

// Runs the bsdiff tool on two files and returns the resulting delta in
// 'out'. Returns true on success.
bool DeltaDiffGenerator::BsdiffFiles(const string& old_file,
                                     const string& new_file,
                                     chromeos::Blob* out) {
  const string kPatchFile = "delta.patchXXXXXX";
  string patch_file_path;

  TEST_AND_RETURN_FALSE(
      utils::MakeTempFile(kPatchFile, &patch_file_path, nullptr));

  vector<string> cmd;
  cmd.push_back(kBsdiffPath);
  cmd.push_back(old_file);
  cmd.push_back(new_file);
  cmd.push_back(patch_file_path);

  int rc = 1;
  chromeos::Blob patch_file;
  TEST_AND_RETURN_FALSE(Subprocess::SynchronousExec(cmd, &rc, nullptr));
  TEST_AND_RETURN_FALSE(rc == 0);
  TEST_AND_RETURN_FALSE(utils::ReadFile(patch_file_path, out));
  unlink(patch_file_path.c_str());
  return true;
}

// The |blocks| vector contains a reader and writer for each block on the
// filesystem that's being in-place updated. We populate the reader/writer
// fields of |blocks| by calling this function.
// For each block in |operation| that is read or written, find that block
// in |blocks| and set the reader/writer field to the vertex passed.
// |graph| is not strictly necessary, but useful for printing out
// error messages.
bool DeltaDiffGenerator::AddInstallOpToBlocksVector(
    const DeltaArchiveManifest_InstallOperation& operation,
    const Graph& graph,
    Vertex::Index vertex,
    vector<Block>* blocks) {
  // See if this is already present.
  TEST_AND_RETURN_FALSE(operation.dst_extents_size() > 0);

  enum BlockField { READER = 0, WRITER, BLOCK_FIELD_COUNT };
  for (int field = READER; field < BLOCK_FIELD_COUNT; field++) {
    const int extents_size =
        (field == READER) ? operation.src_extents_size() :
        operation.dst_extents_size();
    const char* past_participle = (field == READER) ? "read" : "written";
    const google::protobuf::RepeatedPtrField<Extent>& extents =
        (field == READER) ? operation.src_extents() : operation.dst_extents();
    Vertex::Index Block::*access_type =
        (field == READER) ? &Block::reader : &Block::writer;

    for (int i = 0; i < extents_size; i++) {
      const Extent& extent = extents.Get(i);
      if (extent.start_block() == kSparseHole) {
        // Hole in sparse file. skip
        continue;
      }
      for (uint64_t block = extent.start_block();
           block < (extent.start_block() + extent.num_blocks()); block++) {
        if ((*blocks)[block].*access_type != Vertex::kInvalidIndex) {
          LOG(FATAL) << "Block " << block << " is already "
                     << past_participle << " by "
                     << (*blocks)[block].*access_type << "("
                     << graph[(*blocks)[block].*access_type].file_name
                     << ") and also " << vertex << "("
                     << graph[vertex].file_name << ")";
        }
        (*blocks)[block].*access_type = vertex;
      }
    }
  }
  return true;
}

void DeltaDiffGenerator::AddSignatureOp(uint64_t signature_blob_offset,
                                        uint64_t signature_blob_length,
                                        DeltaArchiveManifest* manifest) {
  LOG(INFO) << "Making room for signature in file";
  manifest->set_signatures_offset(signature_blob_offset);
  LOG(INFO) << "set? " << manifest->has_signatures_offset();
  // Add a dummy op at the end to appease older clients
  DeltaArchiveManifest_InstallOperation* dummy_op =
      manifest->add_kernel_install_operations();
  dummy_op->set_type(DeltaArchiveManifest_InstallOperation_Type_REPLACE);
  dummy_op->set_data_offset(signature_blob_offset);
  manifest->set_signatures_offset(signature_blob_offset);
  dummy_op->set_data_length(signature_blob_length);
  manifest->set_signatures_size(signature_blob_length);
  Extent* dummy_extent = dummy_op->add_dst_extents();
  // Tell the dummy op to write this data to a big sparse hole
  dummy_extent->set_start_block(kSparseHole);
  dummy_extent->set_num_blocks((signature_blob_length + kBlockSize - 1) /
                               kBlockSize);
}

const char* const kBsdiffPath = "bsdiff";

};  // namespace chromeos_update_engine