blob: 3a674ecc10af8dc9a8cde579a3b39e29b7645e03 [file] [log] [blame]
// Copyright 2014 The Chromium 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 "extensions/browser/content_hash_fetcher.h"
#include <stddef.h>
#include <algorithm>
#include <memory>
#include <vector>
#include "base/files/file_util.h"
#include "base/json/json_reader.h"
#include "base/macros.h"
#include "base/synchronization/lock.h"
#include "base/task_scheduler/post_task.h"
#include "content/public/browser/browser_thread.h"
#include "extensions/browser/content_verifier/content_hash.h"
#include "extensions/browser/content_verifier_delegate.h"
#include "extensions/browser/extension_file_task_runner.h"
#include "extensions/browser/verified_contents.h"
#include "extensions/common/extension.h"
#include "extensions/common/file_util.h"
#include "net/base/load_flags.h"
#include "net/traffic_annotation/network_traffic_annotation.h"
#include "net/url_request/url_fetcher.h"
#include "net/url_request/url_fetcher_delegate.h"
#include "net/url_request/url_request_status.h"
namespace extensions {
namespace {
std::unique_ptr<ContentHash> ReadContentHashAfterVerifiedContentsWritten(
const ContentHash::ExtensionKey& extension_key,
bool force,
const ContentHash::IsCancelledCallback& is_cancelled) {
// This function is called after writing verified_contents.json, so we'd
// expect a valid file at this point. Even in the case where the file turns
// out to be invalid right after writing it (malicous behavior?) don't delete
// it.
int mode = 0;
if (force)
mode |= ContentHash::Mode::kForceRecreateExistingComputedHashesFile;
return ContentHash::Create(extension_key, mode, is_cancelled);
}
} // namespace
// This class takes care of doing the disk and network I/O work to ensure we
// have both verified_contents.json files from the webstore and
// computed_hashes.json files computed over the files in an extension's
// directory.
class ContentHashFetcherJob
: public base::RefCountedThreadSafe<ContentHashFetcherJob>,
public net::URLFetcherDelegate {
public:
using CompletionCallback =
base::Callback<void(scoped_refptr<ContentHashFetcherJob>)>;
ContentHashFetcherJob(net::URLRequestContextGetter* request_context,
const ContentHash::ExtensionKey& extension_key,
const GURL& fetch_url,
bool force,
const CompletionCallback& callback);
void Start();
// Cancels this job, which will attempt to stop I/O operations sooner than
// just waiting for the entire job to complete. Safe to call from any thread.
void Cancel();
// Checks whether this job has been cancelled. Safe to call from any thread.
bool IsCancelled();
// Returns whether this job was successful (we have both verified contents
// and computed hashes). Even if the job was a success, there might have been
// files that were found to have contents not matching expectations; these
// are available by calling hash_mismatch_unix_paths().
bool success() { return success_; }
bool force() const { return force_; }
const std::string& extension_id() { return extension_key_.extension_id; }
// Returns the set of paths (with unix style '/' separators) that had a hash
// mismatch.
const std::set<base::FilePath>& hash_mismatch_unix_paths() {
return hash_mismatch_unix_paths_;
}
private:
friend class base::RefCountedThreadSafe<ContentHashFetcherJob>;
~ContentHashFetcherJob() override;
// Methods called when job completes.
void Succeeded(const std::set<base::FilePath>& hash_mismatch_unix_paths);
void Failed();
// Returns content hashes after trying to load them.
// Invalid verified_contents.json will be removed from disk.
std::unique_ptr<ContentHash> LoadVerifiedContents();
// Callback for when we're done doing file I/O to see if we already have
// content hashes. If we don't have verified_contents.json, this will kick off
// a network request to get one.
void DoneCheckingForVerifiedContents(
std::unique_ptr<ContentHash> content_hash);
// URLFetcherDelegate interface
void OnURLFetchComplete(const net::URLFetcher* source) override;
// Callback for when we've read verified contents after fetching it from
// network and writing it to disk.
void ReadCompleteAfterWrite(std::unique_ptr<ContentHash> content_hash);
// Callback for the job to write the verified contents to the filesystem.
void OnVerifiedContentsWritten(size_t expected_size, int write_result);
// Will call the callback, if we haven't been cancelled.
void DispatchCallback();
net::URLRequestContextGetter* request_context_;
// Note: Read from multiple threads/sequences.
const ContentHash::ExtensionKey extension_key_;
// The url we'll need to use to fetch a verified_contents.json file.
GURL fetch_url_;
const bool force_;
CompletionCallback callback_;
content::BrowserThread::ID creation_thread_;
// Used for fetching content signatures.
// Created and destroyed on |creation_thread_|.
std::unique_ptr<net::URLFetcher> url_fetcher_;
// The parsed contents of the verified_contents.json file, either read from
// disk or fetched from the network and then written to disk.
std::unique_ptr<VerifiedContents> verified_contents_;
// Whether this job succeeded.
bool success_;
// Paths that were found to have a mismatching hash.
std::set<base::FilePath> hash_mismatch_unix_paths_;
// Note: this may be accessed from multiple threads, so all access should
// be protected by |cancelled_lock_|.
bool cancelled_;
// A lock for synchronizing access to |cancelled_|.
base::Lock cancelled_lock_;
DISALLOW_COPY_AND_ASSIGN(ContentHashFetcherJob);
};
ContentHashFetcherJob::ContentHashFetcherJob(
net::URLRequestContextGetter* request_context,
const ContentHash::ExtensionKey& extension_key,
const GURL& fetch_url,
bool force,
const CompletionCallback& callback)
: request_context_(request_context),
extension_key_(extension_key),
fetch_url_(fetch_url),
force_(force),
callback_(callback),
success_(false),
cancelled_(false) {
bool got_id =
content::BrowserThread::GetCurrentThreadIdentifier(&creation_thread_);
DCHECK(got_id);
}
void ContentHashFetcherJob::Start() {
base::PostTaskAndReplyWithResult(
GetExtensionFileTaskRunner().get(), FROM_HERE,
base::BindOnce(&ContentHashFetcherJob::LoadVerifiedContents, this),
base::BindOnce(&ContentHashFetcherJob::DoneCheckingForVerifiedContents,
this));
}
void ContentHashFetcherJob::Cancel() {
base::AutoLock autolock(cancelled_lock_);
cancelled_ = true;
}
bool ContentHashFetcherJob::IsCancelled() {
base::AutoLock autolock(cancelled_lock_);
bool result = cancelled_;
return result;
}
ContentHashFetcherJob::~ContentHashFetcherJob() {
// Destroy |url_fetcher_| on correct thread.
// It was possible for it to be deleted on blocking pool. See
// https://crbug.com/702300 for details.
// TODO(lazyboy): Make ContentHashFetcherJob non ref-counted.
if (url_fetcher_ && !content::BrowserThread::CurrentlyOn(creation_thread_)) {
content::BrowserThread::DeleteSoon(creation_thread_, FROM_HERE,
url_fetcher_.release());
}
}
std::unique_ptr<ContentHash> ContentHashFetcherJob::LoadVerifiedContents() {
// Will delete invalid verified contents file.
int mode = ContentHash::Mode::kDeleteInvalidVerifiedContents;
if (force_)
mode |= ContentHash::Mode::kForceRecreateExistingComputedHashesFile;
return ContentHash::Create(
extension_key_, mode,
base::BindRepeating(&ContentHashFetcherJob::IsCancelled, this));
}
void ContentHashFetcherJob::DoneCheckingForVerifiedContents(
std::unique_ptr<ContentHash> content_hash) {
if (IsCancelled())
return;
if (content_hash->has_verified_contents()) {
VLOG(1) << "Found verified contents for " << extension_key_.extension_id;
if (content_hash->succeeded()) {
// We've found both verified_contents.json and computed_hashes.json, we're
// done.
Succeeded(content_hash->hash_mismatch_unix_paths());
} else {
// Even though we attempted to write computed_hashes.json after reading
// verified_contents.json, something went wrong.
Failed();
}
} else {
VLOG(1) << "Missing verified contents for " << extension_key_.extension_id
<< ", fetching...";
net::NetworkTrafficAnnotationTag traffic_annotation =
net::DefineNetworkTrafficAnnotation("content_hash_verification_job", R"(
semantics {
sender: "Web Store Content Verification"
description:
"The request sent to retrieve the file required for content "
"verification for an extension from the Web Store."
trigger:
"An extension from the Web Store is missing the "
"verified_contents.json file required for extension content "
"verification."
data: "The extension id and extension version."
destination: GOOGLE_OWNED_SERVICE
}
policy {
cookies_allowed: NO
setting:
"This feature cannot be directly disabled; it is enabled if any "
"extension from the webstore is installed in the browser."
policy_exception_justification:
"Not implemented, not required. If the user has extensions from "
"the Web Store, this feature is required to ensure the "
"extensions match what is distributed by the store."
})");
url_fetcher_ = net::URLFetcher::Create(fetch_url_, net::URLFetcher::GET,
this, traffic_annotation);
url_fetcher_->SetRequestContext(request_context_);
url_fetcher_->SetLoadFlags(net::LOAD_DO_NOT_SEND_COOKIES |
net::LOAD_DO_NOT_SAVE_COOKIES |
net::LOAD_DISABLE_CACHE);
url_fetcher_->SetAutomaticallyRetryOnNetworkChanges(3);
url_fetcher_->Start();
}
}
// Helper function to let us pass ownership of a string via base::Bind with the
// contents to be written into a file. Also ensures that the directory for
// |path| exists, creating it if needed.
static int WriteFileHelper(const base::FilePath& path,
std::unique_ptr<std::string> content) {
base::FilePath dir = path.DirName();
if (!base::CreateDirectoryAndGetError(dir, nullptr))
return -1;
return base::WriteFile(path, content->data(), content->size());
}
void ContentHashFetcherJob::OnURLFetchComplete(const net::URLFetcher* source) {
VLOG(1) << "URLFetchComplete for " << extension_key_.extension_id
<< " is_success:" << url_fetcher_->GetStatus().is_success() << " "
<< fetch_url_.possibly_invalid_spec();
// Delete |url_fetcher_| once we no longer need it.
std::unique_ptr<net::URLFetcher> url_fetcher = std::move(url_fetcher_);
if (IsCancelled())
return;
std::unique_ptr<std::string> response(new std::string);
if (!url_fetcher->GetStatus().is_success() ||
!url_fetcher->GetResponseAsString(response.get())) {
Failed();
return;
}
// Parse the response to make sure it is valid json (on staging sometimes it
// can be a login redirect html, xml file, etc. if you aren't logged in with
// the right cookies). TODO(asargent) - It would be a nice enhancement to
// move to parsing this in a sandboxed helper (crbug.com/372878).
std::unique_ptr<base::Value> parsed(base::JSONReader::Read(*response));
if (!parsed) {
Failed();
return;
}
VLOG(1) << "JSON parsed ok for " << extension_key_.extension_id;
parsed.reset(); // no longer needed
base::FilePath destination =
file_util::GetVerifiedContentsPath(extension_key_.extension_root);
size_t size = response->size();
base::PostTaskWithTraitsAndReplyWithResult(
FROM_HERE, {base::MayBlock(), base::TaskPriority::USER_VISIBLE},
// TODO(lazyboy): Consider creating VerifiedContents directly from
// |response| instead of reading the file again.
base::BindOnce(&WriteFileHelper, destination, base::Passed(&response)),
base::BindOnce(&ContentHashFetcherJob::OnVerifiedContentsWritten, this,
size));
}
void ContentHashFetcherJob::OnVerifiedContentsWritten(size_t expected_size,
int write_result) {
bool success =
write_result >= 0 && static_cast<size_t>(write_result) == expected_size;
if (!success) {
Failed();
return;
}
base::PostTaskAndReplyWithResult(
GetExtensionFileTaskRunner().get(), FROM_HERE,
// TODO(lazyboy): Consider creating VerifiedContents directly from
// |response| instead of reading the file again.
base::BindOnce(
&ReadContentHashAfterVerifiedContentsWritten, extension_key_, force_,
base::BindRepeating(&ContentHashFetcherJob::IsCancelled, this)),
base::BindOnce(&ContentHashFetcherJob::ReadCompleteAfterWrite, this));
}
void ContentHashFetcherJob::ReadCompleteAfterWrite(
std::unique_ptr<ContentHash> content_hash) {
if (IsCancelled())
return;
if (content_hash->succeeded()) {
Succeeded(content_hash->hash_mismatch_unix_paths());
} else {
Failed();
}
}
void ContentHashFetcherJob::Failed() {
success_ = false;
DispatchCallback();
}
void ContentHashFetcherJob::Succeeded(
const std::set<base::FilePath>& hash_mismatch_unix_paths) {
success_ = true;
// TODO(lazyboy): Avoid this copy.
hash_mismatch_unix_paths_ = hash_mismatch_unix_paths;
DispatchCallback();
}
void ContentHashFetcherJob::DispatchCallback() {
{
base::AutoLock autolock(cancelled_lock_);
if (cancelled_)
return;
}
callback_.Run(base::WrapRefCounted(this));
}
// ----
ContentHashFetcher::ContentHashFetcher(
net::URLRequestContextGetter* context_getter,
ContentVerifierDelegate* delegate,
const FetchCallback& callback)
: context_getter_(context_getter),
delegate_(delegate),
fetch_callback_(callback),
weak_ptr_factory_(this) {}
ContentHashFetcher::~ContentHashFetcher() {
for (JobMap::iterator i = jobs_.begin(); i != jobs_.end(); ++i) {
i->second->Cancel();
}
}
void ContentHashFetcher::DoFetch(const Extension* extension, bool force) {
DCHECK(extension);
IdAndVersion key(extension->id(), extension->version().GetString());
JobMap::iterator found = jobs_.find(key);
if (found != jobs_.end()) {
if (!force || found->second->force()) {
// Just let the existing job keep running.
return;
} else {
// Kill the existing non-force job, so we can start a new one below.
found->second->Cancel();
jobs_.erase(found);
}
}
// TODO(asargent) - we should do something here to remember recent attempts
// to fetch signatures by extension id, and use exponential backoff to avoid
// hammering the server when we aren't successful in getting them.
// crbug.com/373397
GURL url =
delegate_->GetSignatureFetchUrl(extension->id(), extension->version());
ContentHashFetcherJob* job = new ContentHashFetcherJob(
context_getter_,
ContentHash::ExtensionKey(extension->id(), extension->path(),
extension->version(),
delegate_->GetPublicKey()),
url, force,
// TODO(lazyboy): Use BindOnce, ContentHashFetcherJob should respond at
// most once.
base::BindRepeating(&ContentHashFetcher::JobFinished,
weak_ptr_factory_.GetWeakPtr()));
jobs_.insert(std::make_pair(key, job));
job->Start();
}
void ContentHashFetcher::ExtensionLoaded(const Extension* extension) {
CHECK(extension);
DoFetch(extension, false);
}
void ContentHashFetcher::ExtensionUnloaded(const Extension* extension) {
CHECK(extension);
IdAndVersion key(extension->id(), extension->version().GetString());
JobMap::iterator found = jobs_.find(key);
if (found != jobs_.end()) {
found->second->Cancel();
jobs_.erase(found);
}
}
void ContentHashFetcher::JobFinished(scoped_refptr<ContentHashFetcherJob> job) {
if (!job->IsCancelled()) {
// Note: Run can result in ContentHashFetcher::ExtensionUnloaded.
//
// TODO(lazyboy): Add a unit test to cover the case where Run can result in
// ContentHashFetcher::ExtensionUnloaded, once https://crbug.com/702300 is
// fixed.
fetch_callback_.Run(job->extension_id(), job->success(), job->force(),
job->hash_mismatch_unix_paths());
}
for (JobMap::iterator i = jobs_.begin(); i != jobs_.end(); ++i) {
if (i->second.get() == job.get()) {
jobs_.erase(i);
break;
}
}
}
} // namespace extensions