unpacker/cpp/volume_reader_javascript_stream.cc

// Copyright 2014 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 "volume_reader_javascript_stream.h"

#include <algorithm>
#include <limits>

#include "archive.h"
#include "ppapi/cpp/logging.h"

VolumeReaderJavaScriptStream::VolumeReaderJavaScriptStream(
    int64_t archive_size,
    JavaScriptRequestorInterface* requestor)
    : archive_size_(archive_size),
      requestor_(requestor),
      available_data_(false),
      read_error_(false),
      passphrase_error_(false),
      offset_(0),
      last_read_chunk_offset_(-1) /* For first call -1 will force a chunk
                                     request from JavaScript as offset
                                     parameter is 0. */,
      read_ahead_array_buffer_ptr_(&first_array_buffer_) {
  pthread_mutex_init(&shared_state_lock_, NULL);
  pthread_cond_init(&available_data_cond_, NULL);
  pthread_cond_init(&available_passphrase_cond_, NULL);

  // Dummy Map the second buffer as first buffer is used for read ahead by
  // read_ahead_array_buffer_ptr_. This operation is required in order for Unmap
  // to correctly work in the destructor and VolumeReaderJavaScriptStream::Read.
  second_array_buffer_.Map();
}

VolumeReaderJavaScriptStream::~VolumeReaderJavaScriptStream() {
  pthread_mutex_destroy(&shared_state_lock_);
  pthread_cond_destroy(&available_data_cond_);
  pthread_cond_destroy(&available_passphrase_cond_);

  // Unmap last mapped buffer. This is the other buffer to
  // read_ahead_array_buffer_ptr_ as read_ahead_array_buffer_ptr_ must be
  // available for SetBufferAndSignal to overwrite.
  if (read_ahead_array_buffer_ptr_ != &first_array_buffer_)
    first_array_buffer_.Unmap();
  else
    second_array_buffer_.Unmap();
};

void VolumeReaderJavaScriptStream::SetBufferAndSignal(
    const pp::VarArrayBuffer& array_buffer,
    int64_t read_offset) {
  PP_DCHECK(read_offset >= 0);

  // Ignore read ahead in case offset was changed using Skip or Seek and in case
  // we already have available data. This can happen in case of 2+ RequestChunk
  // calls done in parallel as a result of calling Read, Skip and Seek one after
  // another really fast. The usage of the buffer is not guarded so in case we
  // overwrite *read_ahead_array_buffer_ptr_ we will end up with memory
  // corruption.
  // In case read_offset and offset_ are different, then the read ahead data is
  // not valid anymore, but in case they are equal and available_data_ is set to
  // true then the second read ahead data is the same as the first read ahead
  // data so we can just ignore it.

  // TODO(mtomasz): We don't need to discard everything. Sometimes part of the
  // buffer can still be used. In such case we should use it. That can greatly
  // improve traversing headers for archives with small files!

  pthread_mutex_lock(&shared_state_lock_);
  if (read_offset == offset_ && !available_data_ && !read_error_) {
    // Signal VolumeReaderJavaScriptStream::Read to continue execution. Copies
    // buffer locally so libarchive has the buffer in memory when working with
    // it. Though we acquire a lock here this call is blocking only for a few
    // moments as VolumeReaderJavaScriptStream::Read will release the lock with
    // pthread_cond_wait. So we cannot arrive at a deadlock that will block the
    // main thread.

    *read_ahead_array_buffer_ptr_ = array_buffer;  // Copy operation.
    available_data_ = true;

    pthread_cond_signal(&available_data_cond_);
  }
  pthread_mutex_unlock(&shared_state_lock_);
}

void VolumeReaderJavaScriptStream::ReadErrorSignal() {
  pthread_mutex_lock(&shared_state_lock_);
  read_error_ = true;  // Read error from JavaScript.
  pthread_cond_signal(&available_data_cond_);
  pthread_mutex_unlock(&shared_state_lock_);
}

void VolumeReaderJavaScriptStream::SetPassphraseAndSignal(
    const std::string& passphrase) {
  pthread_mutex_lock(&shared_state_lock_);
  // Signal VolumeReaderJavaScriptStream::Passphrase to continue execution.
  available_passphrase_ = passphrase;
  pthread_cond_signal(&available_passphrase_cond_);
  pthread_mutex_unlock(&shared_state_lock_);
}

void VolumeReaderJavaScriptStream::PassphraseErrorSignal() {
  pthread_mutex_lock(&shared_state_lock_);
  passphrase_error_ = true;  // Passphrase error from JavaScript.
  pthread_cond_signal(&available_passphrase_cond_);
  pthread_mutex_unlock(&shared_state_lock_);
}

int64_t VolumeReaderJavaScriptStream::Read(int64_t bytes_to_read,
                                           const void** destination_buffer) {
  PP_DCHECK(bytes_to_read > 0);

  pthread_mutex_lock(&shared_state_lock_);

  // No more data, so signal end of reading.
  if (offset_ >= archive_size_) {
    pthread_mutex_unlock(&shared_state_lock_);
    return 0;
  }

  // Call in case of first read or read after Seek and Skip.
  if (last_read_chunk_offset_ != offset_)
    RequestChunk(bytes_to_read);

  if (!available_data_) {
    // Wait for data from JavaScript.
    while (!available_data_) {  // Check again available data as first call
                                // was done outside guarded zone.
      if (read_error_) {
        pthread_mutex_unlock(&shared_state_lock_);
        return ARCHIVE_FATAL;
      }
      pthread_cond_wait(&available_data_cond_, &shared_state_lock_);
    }
  }

  if (read_error_) {  // Read ahead failed.
    pthread_mutex_unlock(&shared_state_lock_);
    return ARCHIVE_FATAL;
  }

  // Make data available for libarchive custom read. No need to lock this part.
  // The reason is that VolumeReaderJavaScriptStream::RequestChunk is the only
  // function that can set available_data_ back to false and let
  // VolumeReaderJavaScriptStream::SetBufferAndSignal overwrite the buffer. But
  // reading ahead is done only at the end of this function after the buffers
  // are switched.
  *destination_buffer = read_ahead_array_buffer_ptr_->Map();
  int64_t bytes_read =
      std::min(static_cast<int64_t>(read_ahead_array_buffer_ptr_->ByteLength()),
               bytes_to_read);

  offset_ += bytes_read;
  last_read_chunk_offset_ = offset_;

  // Ask for more data from JavaScript in the other buffer. This is the only
  // time when we switch buffers. The reason is that libarchive must
  // always work on valid data and that data must be available until next
  // VolumeReaderJavaScriptStream::Read call, and as the data can be received
  // at any time from JavaScript, we need a buffer to store it in case of
  // reading ahead.
  read_ahead_array_buffer_ptr_ =
      read_ahead_array_buffer_ptr_ != &first_array_buffer_
          ? &first_array_buffer_
          : &second_array_buffer_;

  // Unmap old buffer. Only Read and constructor can Map the buffers so Read and
  // destructor should be the one to Unmap them. This will work because it is
  // called before RequestChunk which is the only method that overwrites the
  // buffer. The constructor should also Map a default pp::VarArrayBuffer and
  // destructor Unmap the last used array buffer (which is the other buffer than
  // read_ahead_array_buffer_ptr_). Unfortunately it's not clear from the
  // API description if this call is done automatically on pp::VarArrayBuffer
  // destructor.
  read_ahead_array_buffer_ptr_->Unmap();

  // Read ahead next chunk with a length similar to current read.
  RequestChunk(bytes_to_read);
  pthread_mutex_unlock(&shared_state_lock_);

  return bytes_read;
}

int64_t VolumeReaderJavaScriptStream::Seek(int64_t offset, int whence) {
  pthread_mutex_lock(&shared_state_lock_);

  int64_t new_offset = offset_;
  switch (whence) {
    case SEEK_SET:
      new_offset = offset;
      break;
    case SEEK_CUR:
      new_offset += offset;
      break;
    case SEEK_END:
      new_offset = archive_size_ + offset;
      break;
    default:
      PP_NOTREACHED();
      pthread_mutex_unlock(&shared_state_lock_);
      return ARCHIVE_FATAL;
  }

  if (new_offset < 0 || new_offset > archive_size_) {
    pthread_mutex_unlock(&shared_state_lock_);
    return ARCHIVE_FATAL;
  }

  offset_ = new_offset;
  pthread_mutex_unlock(&shared_state_lock_);

  return new_offset;
}

int64_t VolumeReaderJavaScriptStream::Skip(int64_t bytes_to_skip) {
  pthread_mutex_lock(&shared_state_lock_);
  // Invalid bytes_to_skip. This "if" can be triggered for corrupted archives.
  // We return 0 instead of ARCHIVE_FATAL in order for libarchive to use normal
  // Read and return the correct error. In case we return ARCHIVE_FATAL here
  // then libarchive just stops without telling us why it wasn't able to
  // process the archive.
  if (archive_size_ - offset_ < bytes_to_skip || bytes_to_skip < 0) {
    pthread_mutex_unlock(&shared_state_lock_);
    return 0;
  }

  offset_ += bytes_to_skip;
  pthread_mutex_unlock(&shared_state_lock_);

  return bytes_to_skip;
}

void VolumeReaderJavaScriptStream::SetRequestId(const std::string& request_id) {
  // No lock necessary, as request_id is used by one thread only.
  request_id_ = request_id;
}

const char* VolumeReaderJavaScriptStream::Passphrase() {
  // The error is not recoverable. Once passphrase fails to be provided, it is
  // never asked again. Note, that still users are able to retry entering the
  // password, unless they click Cancel.
  pthread_mutex_lock(&shared_state_lock_);
  if (passphrase_error_) {
    pthread_mutex_unlock(&shared_state_lock_);
    return NULL;
  }
  pthread_mutex_unlock(&shared_state_lock_);

  // Request the passphrase outside of the lock.
  requestor_->RequestPassphrase(request_id_);

  pthread_mutex_lock(&shared_state_lock_);
  // Wait for the passphrase from JavaScript.
  pthread_cond_wait(&available_passphrase_cond_, &shared_state_lock_);
  const char* result = NULL;
  if (!passphrase_error_)
    result = strdup(available_passphrase_.c_str());
  pthread_mutex_unlock(&shared_state_lock_);

  return result;
}

void VolumeReaderJavaScriptStream::RequestChunk(int64_t length) {
  // Read next chunk only if not at the end of archive.
  if (archive_size_ <= offset_)
    return;

  int64_t bytes_to_read =
      std::min(length, archive_size_ - offset_ /* Positive check above. */);
  available_data_ = false;

  requestor_->RequestFileChunk(request_id_, offset_, bytes_to_read);
}