| Line |
Branch |
Exec |
Source |
| 1 |
|
|
/** |
| 2 |
|
|
* This file is part of the CernVM File System. |
| 3 |
|
|
*/ |
| 4 |
|
|
|
| 5 |
|
|
#define __STDC_FORMAT_MACROS |
| 6 |
|
|
|
| 7 |
|
|
#include "swissknife_scrub.h" |
| 8 |
|
|
#include "cvmfs_config.h" |
| 9 |
|
|
|
| 10 |
|
|
#include "util/fs_traversal.h" |
| 11 |
|
|
#include "util/logging.h" |
| 12 |
|
|
#include "util/posix.h" |
| 13 |
|
|
#include "util/smalloc.h" |
| 14 |
|
|
#include "util/string.h" |
| 15 |
|
|
|
| 16 |
|
|
using namespace std; // NOLINT |
| 17 |
|
|
|
| 18 |
|
|
namespace swissknife { |
| 19 |
|
|
|
| 20 |
|
|
const size_t kHashSubtreeLength = 2; |
| 21 |
|
|
const std::string kTxnDirectoryName = "txn"; |
| 22 |
|
|
|
| 23 |
|
✗ |
CommandScrub::CommandScrub() |
| 24 |
|
✗ |
: machine_readable_output_(false) |
| 25 |
|
✗ |
, alerts_(0) |
| 26 |
|
|
{ |
| 27 |
|
✗ |
int retval = pthread_mutex_init(&alerts_mutex_, NULL); |
| 28 |
|
✗ |
assert(retval == 0); |
| 29 |
|
|
} |
| 30 |
|
|
|
| 31 |
|
|
|
| 32 |
|
✗ |
CommandScrub::~CommandScrub() { |
| 33 |
|
✗ |
pthread_mutex_destroy(&alerts_mutex_); |
| 34 |
|
|
} |
| 35 |
|
|
|
| 36 |
|
|
|
| 37 |
|
✗ |
swissknife::ParameterList CommandScrub::GetParams() const { |
| 38 |
|
✗ |
swissknife::ParameterList r; |
| 39 |
|
✗ |
r.push_back(Parameter::Mandatory('r', "repository directory")); |
| 40 |
|
✗ |
r.push_back(Parameter::Switch('m', "machine readable output")); |
| 41 |
|
✗ |
return r; |
| 42 |
|
|
} |
| 43 |
|
|
|
| 44 |
|
✗ |
const char *CommandScrub::Alerts::ToString(const CommandScrub::Alerts::Type t) { |
| 45 |
|
✗ |
switch (t) { |
| 46 |
|
✗ |
case Alerts::kUnexpectedFile: |
| 47 |
|
✗ |
return "unexpected regular file"; |
| 48 |
|
✗ |
case Alerts::kUnexpectedSymlink: |
| 49 |
|
✗ |
return "unexpected symlink"; |
| 50 |
|
✗ |
case Alerts::kUnexpectedSubdir: |
| 51 |
|
✗ |
return "unexpected subdir in CAS subdir"; |
| 52 |
|
✗ |
case Alerts::kUnexpectedModifier: |
| 53 |
|
✗ |
return "unknown object modifier"; |
| 54 |
|
✗ |
case Alerts::kMalformedHash: |
| 55 |
|
✗ |
return "malformed content hash"; |
| 56 |
|
✗ |
case Alerts::kMalformedCasSubdir: |
| 57 |
|
✗ |
return "malformed CAS subdir length"; |
| 58 |
|
✗ |
case Alerts::kContentHashMismatch: |
| 59 |
|
✗ |
return "mismatch of file name and content hash"; |
| 60 |
|
✗ |
default: |
| 61 |
|
✗ |
return "unknown alert"; |
| 62 |
|
|
} |
| 63 |
|
|
} |
| 64 |
|
|
|
| 65 |
|
✗ |
void CommandScrub::FileCallback( |
| 66 |
|
|
const std::string &relative_path, |
| 67 |
|
|
const std::string &file_name) |
| 68 |
|
|
{ |
| 69 |
|
✗ |
assert(!file_name.empty()); |
| 70 |
|
|
|
| 71 |
|
✗ |
if (relative_path.empty()) { |
| 72 |
|
✗ |
PrintAlert(Alerts::kUnexpectedFile, repo_path_ + "/" + file_name); |
| 73 |
|
✗ |
return; |
| 74 |
|
|
} |
| 75 |
|
✗ |
if (relative_path == kTxnDirectoryName) { |
| 76 |
|
|
// transaction directory should be ignored |
| 77 |
|
✗ |
return; |
| 78 |
|
|
} |
| 79 |
|
|
|
| 80 |
|
✗ |
const string full_path = MakeFullPath(relative_path, file_name); |
| 81 |
|
|
const std::string hash_string = |
| 82 |
|
✗ |
CheckPathAndExtractHash(relative_path, file_name, full_path); |
| 83 |
|
✗ |
if (hash_string.empty()) { |
| 84 |
|
✗ |
return; |
| 85 |
|
|
} |
| 86 |
|
|
|
| 87 |
|
✗ |
if (!shash::HexPtr(hash_string).IsValid()) { |
| 88 |
|
✗ |
PrintAlert(Alerts::kMalformedHash, full_path, hash_string); |
| 89 |
|
✗ |
return; |
| 90 |
|
|
} |
| 91 |
|
|
|
| 92 |
|
|
shash::Any hash_from_name = |
| 93 |
|
✗ |
shash::MkFromSuffixedHexPtr(shash::HexPtr(hash_string)); |
| 94 |
|
✗ |
IngestionSource* full_path_source = new FileIngestionSource(full_path); |
| 95 |
|
✗ |
pipeline_scrubbing_.Process( |
| 96 |
|
|
full_path_source, |
| 97 |
|
|
hash_from_name.algorithm, |
| 98 |
|
✗ |
hash_from_name.suffix); |
| 99 |
|
|
} |
| 100 |
|
|
|
| 101 |
|
|
|
| 102 |
|
✗ |
void CommandScrub::DirCallback( |
| 103 |
|
|
const std::string &relative_path, |
| 104 |
|
|
const std::string &dir_name) |
| 105 |
|
|
{ |
| 106 |
|
✗ |
const string full_path = MakeFullPath(relative_path, dir_name); |
| 107 |
|
|
|
| 108 |
|
|
// The directory "/srv/cvmfs/<REPO_NAME>/data/txn/receiver" is whitelisted |
| 109 |
|
✗ |
if (HasSuffix(full_path, "data/txn/receiver", false)) { |
| 110 |
|
✗ |
return; |
| 111 |
|
|
} |
| 112 |
|
|
|
| 113 |
|
|
// Check for nested subdirs |
| 114 |
|
✗ |
if (relative_path.size() > 0) { |
| 115 |
|
✗ |
PrintAlert(Alerts::kUnexpectedSubdir, full_path); |
| 116 |
|
✗ |
return; |
| 117 |
|
|
} |
| 118 |
|
|
|
| 119 |
|
|
// Check CAS hash subdirectory name length |
| 120 |
|
✗ |
if (!dir_name.empty() && dir_name.size() != kHashSubtreeLength && |
| 121 |
|
✗ |
dir_name != kTxnDirectoryName) { |
| 122 |
|
✗ |
PrintAlert(Alerts::kMalformedCasSubdir, full_path); |
| 123 |
|
|
} |
| 124 |
|
|
} |
| 125 |
|
|
|
| 126 |
|
✗ |
void CommandScrub::SymlinkCallback(const std::string &relative_path, |
| 127 |
|
|
const std::string &symlink_name) { |
| 128 |
|
✗ |
const string full_path = MakeFullPath(relative_path, symlink_name); |
| 129 |
|
✗ |
PrintAlert(Alerts::kUnexpectedSymlink, full_path); |
| 130 |
|
|
} |
| 131 |
|
|
|
| 132 |
|
✗ |
void CommandScrub::OnFileHashed(const ScrubbingResult &scrubbing_result) { |
| 133 |
|
✗ |
const string full_path = scrubbing_result.path; |
| 134 |
|
✗ |
const string file_name = GetFileName(full_path); |
| 135 |
|
✗ |
const string parent_path = GetParentPath(full_path); |
| 136 |
|
✗ |
const string relative_path = MakeRelativePath(parent_path); |
| 137 |
|
✗ |
assert(!file_name.empty()); |
| 138 |
|
|
|
| 139 |
|
|
const std::string hash_string = |
| 140 |
|
✗ |
CheckPathAndExtractHash(relative_path, file_name, full_path); |
| 141 |
|
✗ |
assert(!hash_string.empty()); |
| 142 |
|
✗ |
assert(shash::HexPtr(hash_string).IsValid()); |
| 143 |
|
|
|
| 144 |
|
|
|
| 145 |
|
✗ |
if (scrubbing_result.hash != |
| 146 |
|
✗ |
shash::MkFromSuffixedHexPtr(shash::HexPtr(hash_string))) |
| 147 |
|
|
{ |
| 148 |
|
✗ |
PrintAlert(Alerts::kContentHashMismatch, full_path, |
| 149 |
|
✗ |
scrubbing_result.hash.ToString()); |
| 150 |
|
|
} |
| 151 |
|
|
} |
| 152 |
|
|
|
| 153 |
|
✗ |
std::string CommandScrub::CheckPathAndExtractHash( |
| 154 |
|
|
const std::string &relative_path, |
| 155 |
|
|
const std::string &file_name, |
| 156 |
|
|
const std::string &full_path) const |
| 157 |
|
|
{ |
| 158 |
|
|
// check for a valid object modifier on the end of the file name |
| 159 |
|
✗ |
const char last_character = *(file_name.end() - 1); |
| 160 |
|
✗ |
bool has_object_modifier = false; |
| 161 |
|
✗ |
if (std::isupper(last_character)) { |
| 162 |
|
✗ |
has_object_modifier = true; |
| 163 |
|
|
} |
| 164 |
|
✗ |
if (has_object_modifier && last_character != shash::kSuffixHistory && |
| 165 |
|
✗ |
last_character != shash::kSuffixCatalog && |
| 166 |
|
✗ |
last_character != shash::kSuffixPartial && |
| 167 |
|
✗ |
last_character != shash::kSuffixCertificate && |
| 168 |
|
✗ |
last_character != shash::kSuffixMicroCatalog && |
| 169 |
|
|
last_character != shash::kSuffixMetainfo) { |
| 170 |
|
✗ |
PrintAlert(Alerts::kUnexpectedModifier, full_path); |
| 171 |
|
✗ |
return ""; |
| 172 |
|
|
} |
| 173 |
|
|
|
| 174 |
|
|
const string hash_string = |
| 175 |
|
✗ |
GetFileName(GetParentPath(full_path)) + |
| 176 |
|
✗ |
(has_object_modifier ? file_name.substr(0, file_name.length() - 1) |
| 177 |
|
✗ |
: file_name); |
| 178 |
|
✗ |
return hash_string; |
| 179 |
|
|
} |
| 180 |
|
|
|
| 181 |
|
|
|
| 182 |
|
✗ |
int CommandScrub::Main(const swissknife::ArgumentList &args) { |
| 183 |
|
✗ |
repo_path_ = MakeCanonicalPath(*args.find('r')->second); |
| 184 |
|
✗ |
machine_readable_output_ = (args.find('m') != args.end()); |
| 185 |
|
|
|
| 186 |
|
✗ |
pipeline_scrubbing_.RegisterListener(&CommandScrub::OnFileHashed, this); |
| 187 |
|
✗ |
pipeline_scrubbing_.Spawn(); |
| 188 |
|
|
|
| 189 |
|
|
// initialize file system recursion engine |
| 190 |
|
✗ |
FileSystemTraversal<CommandScrub> traverser(this, repo_path_, true); |
| 191 |
|
✗ |
traverser.fn_new_file = &CommandScrub::FileCallback; |
| 192 |
|
✗ |
traverser.fn_enter_dir = &CommandScrub::DirCallback; |
| 193 |
|
✗ |
traverser.fn_new_symlink = &CommandScrub::SymlinkCallback; |
| 194 |
|
✗ |
traverser.Recurse(repo_path_); |
| 195 |
|
|
|
| 196 |
|
|
// wait for reader to finish all jobs |
| 197 |
|
✗ |
pipeline_scrubbing_.WaitFor(); |
| 198 |
|
|
|
| 199 |
|
✗ |
return (alerts_ == 0) ? 0 : 1; |
| 200 |
|
|
} |
| 201 |
|
|
|
| 202 |
|
✗ |
void CommandScrub::PrintAlert( |
| 203 |
|
|
const Alerts::Type type, |
| 204 |
|
|
const std::string &path, |
| 205 |
|
|
const std::string &affected_hash) const |
| 206 |
|
|
{ |
| 207 |
|
✗ |
MutexLockGuard l(alerts_mutex_); |
| 208 |
|
|
|
| 209 |
|
✗ |
const char *msg = Alerts::ToString(type); |
| 210 |
|
✗ |
if (machine_readable_output_) { |
| 211 |
|
✗ |
LogCvmfs(kLogUtility, kLogStderr, "%d %s %s", type, |
| 212 |
|
|
((affected_hash.empty()) ? "-" : affected_hash.c_str()), |
| 213 |
|
|
path.c_str()); |
| 214 |
|
|
} else { |
| 215 |
|
✗ |
LogCvmfs(kLogUtility, kLogStderr, "%s | at: %s", msg, path.c_str()); |
| 216 |
|
|
} |
| 217 |
|
|
|
| 218 |
|
✗ |
++alerts_; |
| 219 |
|
|
} |
| 220 |
|
|
|
| 221 |
|
✗ |
std::string CommandScrub::MakeFullPath(const std::string &relative_path, |
| 222 |
|
|
const std::string &file_name) const { |
| 223 |
|
✗ |
return (relative_path.empty()) |
| 224 |
|
✗ |
? repo_path_ + "/" + file_name |
| 225 |
|
✗ |
: repo_path_ + "/" + relative_path + "/" + file_name; |
| 226 |
|
|
} |
| 227 |
|
|
|
| 228 |
|
✗ |
std::string CommandScrub::MakeRelativePath(const std::string &full_path) { |
| 229 |
|
✗ |
assert(HasPrefix(full_path, repo_path_ + "/", false)); |
| 230 |
|
✗ |
return full_path.substr(repo_path_.length() + 1); |
| 231 |
|
|
} |
| 232 |
|
|
|
| 233 |
|
✗ |
void CommandScrub::ShowAlertsHelpMessage() const { |
| 234 |
|
✗ |
LogCvmfs(kLogUtility, kLogStdout, "to come..."); |
| 235 |
|
|
} |
| 236 |
|
|
|
| 237 |
|
|
} // namespace swissknife |
| 238 |
|
|
|