XRootD
XrdHttpTpcTPC.cc
Go to the documentation of this file.
2 #include "XrdNet/XrdNetAddr.hh"
3 #include "XrdNet/XrdNetUtils.hh"
4 #include "XrdOuc/XrdOucEnv.hh"
5 #include "XrdSec/XrdSecEntity.hh"
8 #include "XrdSys/XrdSysFD.hh"
9 #include "XrdVersion.hh"
10 
14 #include "XrdOuc/XrdOucTUtils.hh"
16 #include "XrdHttp/XrdHttpUtils.hh"
17 
18 #include <curl/curl.h>
19 
20 #include <dlfcn.h>
21 #include <fcntl.h>
22 
23 #include <algorithm>
24 #include <memory>
25 #include <sstream>
26 #include <stdexcept>
27 #include <thread>
28 
29 #include "XrdHttpTpcState.hh"
30 #include "XrdHttpTpcStream.hh"
31 #include "XrdHttpTpcTPC.hh"
32 #include <fstream>
33 
34 using namespace TPC;
35 
36 XrdXrootdTpcMon* TPCHandler::TPCLogRecord::tpcMonitor = 0;
37 
38 uint64_t TPCHandler::m_monid{0};
39 int TPCHandler::m_marker_period = 5;
40 size_t TPCHandler::m_block_size = 16*1024*1024;
41 size_t TPCHandler::m_small_block_size = 1*1024*1024;
42 XrdSysMutex TPCHandler::m_monid_mutex;
43 bool TPCHandler::allowMissingCRL = false;
44 
46 
47 /******************************************************************************/
48 /* T P C H a n d l e r : : T P C L o g R e c o r d D e s t r u c t o r */
49 /******************************************************************************/
50 
51 TPCHandler::TPCLogRecord::~TPCLogRecord()
52 {
53 // Record monitoring data is enabled
54 //
55  if (tpcMonitor)
56  {XrdXrootdTpcMon::TpcInfo monInfo;
57 
58  monInfo.clID = clID.c_str();
59  monInfo.begT = begT;
60  gettimeofday(&monInfo.endT, 0);
61 
62  if (mTpcType == TpcType::Pull)
63  {monInfo.dstURL = local.c_str();
64  monInfo.srcURL = remote.c_str();
65  } else {
66  monInfo.dstURL = remote.c_str();
67  monInfo.srcURL = local.c_str();
69  }
70 
71  if (!status) monInfo.endRC = 0;
72  else if (tpc_status > 0) monInfo.endRC = tpc_status;
73  else monInfo.endRC = 1;
74  monInfo.strm = static_cast<unsigned char>(streams);
75  monInfo.fSize = (bytes_transferred < 0 ? 0 : bytes_transferred);
76  if (!isIPv6) monInfo.opts |= XrdXrootdTpcMon::TpcInfo::isIPv4;
77 
78  tpcMonitor->Report(monInfo);
79  }
80 }
81 
82 /******************************************************************************/
83 /* C u r l D e l e t e r : : o p e r a t o r ( ) */
84 /******************************************************************************/
85 
87 {
88  if (curl) curl_easy_cleanup(curl);
89 }
90 
91 /******************************************************************************/
92 /* s o c k o p t _ s e t c l o e x e c _ c a l l b a c k */
93 /******************************************************************************/
94 
103 int TPCHandler::sockopt_callback(void *clientp, curl_socket_t curlfd, curlsocktype purpose) {
104  TPCLogRecord * rec = (TPCLogRecord *)clientp;
105  if (purpose == CURLSOCKTYPE_IPCXN && rec && rec->pmarkManager.isEnabled()) {
106  // We will not reach this callback if the corresponding socket could not have been connected
107  // the socket is already connected only if the packet marking is enabled
108  return CURL_SOCKOPT_ALREADY_CONNECTED;
109  }
110  return CURL_SOCKOPT_OK;
111 }
112 
113 /******************************************************************************/
114 /* o p e n s o c k e t _ c a l l b a c k */
115 /******************************************************************************/
116 
117 
122 int TPCHandler::opensocket_callback(void *clientp,
123  curlsocktype purpose,
124  struct curl_sockaddr *aInfo)
125 {
126  /* CURLSOCKTYPE_IPCXN (for IP based connections) is the only type currently known by curl,
127  * so let's make sure to reject other types if they appear in the furure */
128  if (purpose != CURLSOCKTYPE_IPCXN)
129  return CURL_SOCKET_BAD;
130 
131  if (!aInfo)
132  return CURL_SOCKET_BAD;
133 
134  // Create the socket (note that O_CLOEXEC flag will be set)
135  int fd = XrdSysFD_Socket(aInfo->family, aInfo->socktype, aInfo->protocol);
136 
137  if (fd < 0) {
138  return CURL_SOCKET_BAD;
139  }
140 
141  if (!clientp)
142  return fd;
143 
144  XrdNetAddr thePeer(&(aInfo->addr));
145  TPCLogRecord *rec = static_cast<TPCLogRecord*>(clientp);
146 
147  /* Reject attempts to connect to local/private addresses unless allowed by configuration */
148  if ((!rec->allow_private && thePeer.isPrivate()) || (!rec->allow_local && thePeer.isLocal())) {
149  rec->tpc_status = 403; // Forbidden
150  rec->m_log->Emsg(rec->log_prefix.c_str(),
151  "Connection to local/private address is forbidden");
152  close(fd);
153  return CURL_SOCKET_BAD;
154  }
155 
156  rec->isIPv6 = (thePeer.isIPType(XrdNetAddrInfo::IPv6) && !thePeer.isMapped());
157 
158  std::stringstream connectErrMsg;
159  if(!rec->pmarkManager.connect(fd, &(aInfo->addr), aInfo->addrlen, CONNECT_TIMEOUT, connectErrMsg)) {
160  // at this point fd has already been closed
161  rec->m_log->Emsg(rec->log_prefix.c_str(), "Unable to connect socket: ", connectErrMsg.str().c_str());
162  return CURL_SOCKET_BAD;
163  }
164 
165  return fd;
166 }
167 
168 int TPCHandler::closesocket_callback(void *clientp, curl_socket_t fd) {
169  TPCLogRecord * rec = (TPCLogRecord *)clientp;
170 
171  // Destroy the PMark handle associated to the file descriptor before closing it.
172  // Otherwise, we would lose the socket usage information if the socket is closed before
173  // the PMark handle is closed.
174  rec->pmarkManager.endPmark(fd);
175 
176  return close(fd);
177 }
178 
179 /******************************************************************************/
180 /* s s l _ c t x _ c a l l b a c k */
181 /******************************************************************************/
182 
189 int TPCHandler::ssl_ctx_callback(CURL *curl, void *ssl_ctx, void *clientp) {
190  TPCLogRecord * rec = (TPCLogRecord *)clientp;
191  SSL_CTX* ctx = static_cast<SSL_CTX*>(ssl_ctx);
192 
193  if (rec && rec->ca_store) {
194  // Bumps the store's reference count instead of re-parsing the CA and CRL
195  // bundles for this connection. libcurl runs this callback after it has
196  // applied its own TLS options, so this replaces whatever store it built.
197  SSL_CTX_set1_cert_store(ctx, rec->ca_store.get());
198  }
199  if (allowMissingCRL) {
200  // verify_callback only excuses X509_V_ERR_UNABLE_TO_GET_CRL, i.e. a CA in
201  // the chain for which no CRL could be found. Every other verification rule
202  // still applies, including revocation itself whenever a CRL is present.
203  SSL_CTX_set_verify(ctx, SSL_VERIFY_PEER, verify_callback);
204  }
205  return CURLE_OK;
206 }
207 
208 int TPCHandler::verify_callback(int preverify_ok, X509_STORE_CTX* ctx) {
209  if (preverify_ok == 1) return 1;
210 
211  int err = X509_STORE_CTX_get_error(ctx);
212 
213  if (err == X509_V_ERR_UNABLE_TO_GET_CRL) {
214  X509_STORE_CTX_set_error(ctx, X509_V_OK);
215  return 1;
216  }
217 
218  return 0;
219 }
220 
221 /******************************************************************************/
222 /* p r e p a r e U R L */
223 /******************************************************************************/
224 
225 // See XrdHttpTpcUtils::prepareOpenURL() documentation
226 std::string TPCHandler::prepareURL(XrdHttpExtReq &req) {
227  XrdHttpTpcUtils::PrepareOpenURLParams parms {req.resource, req.headers, hdr2cgimap,req.mReprDigest};
228  return XrdHttpTpcUtils::prepareOpenURL(parms);
229 }
230 
231 bool TPCHandler::mismatchReprDigest(const std::map<std::string, std::string> & passiveSrvReprDigest, XrdHttpExtReq &req,
232  TPCLogRecord &rec) {
233  if(passiveSrvReprDigest.size()) {
234  for (const auto & [digestName, digestValue]: passiveSrvReprDigest) {
235  auto clientDigestMatch = req.mReprDigest.find(digestName);
236  if (clientDigestMatch != req.mReprDigest.end()) {
237  // We found a checksum type match between the client-provided one and the source server-provided one
238  if (clientDigestMatch->second != digestValue) {
239  // The checksum value does not match, return an error to the client 412 PRECONDITION_FAILED
240  std::stringstream errMsg;
241  errMsg << "Mismatch between client-provided and remote server checksums:"
242  << " client = (" << clientDigestMatch->first << "=" << clientDigestMatch->second << ")"
243  << " server = (" << digestName << "=" << digestValue << ")";
244  logTransferEvent(LogMask::Error, rec, "REPRDIGEST_VERIFY_FAIL", errMsg.str());
245  rec.status=412;
246  req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(errMsg, rec, CURLcode::CURLE_OK).c_str(), 0);
247  return true;
248  }
249  }
250  }
251  }
252  return false;
253 }
254 
255 /******************************************************************************/
256 /* e n c o d e _ x r o o t d _ o p a q u e _ t o _ u r i */
257 /******************************************************************************/
258 
259 // When processing a redirection from the filesystem layer, it is permitted to return
260 // some xrootd opaque data. The quoting rules for xrootd opaque data are significantly
261 // more permissive than a URI (basically, only '&' and '=' are disallowed while some
262 // URI parsers may dislike characters like '"'). This function takes an opaque string
263 // (e.g., foo=1&bar=2&baz=") and makes it safe for all URI parsers.
264 std::string encode_xrootd_opaque_to_uri(CURL *curl, const std::string &opaque)
265 {
266  std::stringstream parser(opaque);
267  std::string sequence;
268  std::stringstream output;
269  bool first = true;
270  while (getline(parser, sequence, '&')) {
271  if (sequence.empty()) {continue;}
272  size_t equal_pos = sequence.find('=');
273  char *val = NULL;
274  if (equal_pos != std::string::npos)
275  val = curl_easy_escape(curl, sequence.c_str() + equal_pos + 1, sequence.size() - equal_pos - 1);
276  // Do not emit parameter if value exists and escaping failed.
277  if (!val && equal_pos != std::string::npos) {continue;}
278 
279  if (!first) output << "&";
280  first = false;
281  output << sequence.substr(0, equal_pos);
282  if (val) {
283  output << "=" << val;
284  curl_free(val);
285  }
286  }
287  return output.str();
288 }
289 
290 /******************************************************************************/
291 /* T P C H a n d l e r : : C o n f i g u r e C u r l C A */
292 /******************************************************************************/
293 
294 bool
295 TPCHandler::ConfigureCurlCA(CURL *curl, TPCLogRecord &rec)
296 {
297  // Preferred path: hand libcurl the CA/CRL store that XrdTlsTempCA already
298  // parsed, rather than the bundle filenames. Passing filenames makes libcurl
299  // build a private X509_STORE per connection, which costs tens of MB for a grid
300  // CA directory and is held for the whole transfer; sharing one store makes that
301  // a reference count. See https://github.com/xrootd/xrootd/issues/2873
302  //
303  // Skipped when m_cafile is set, so that the http.cafile precedence established
304  // at the bottom of this function is preserved.
305  if (m_ca_file && m_sslctx_supported && m_cafile.empty()) {
306  rec.ca_store = m_ca_file->CAStore();
307  if (!rec.ca_store) {
308  m_log.Log(Error, "TpcHandler", "No CA store is available; refusing to "
309  "fall back to libcurl's default CA bundle");
310  return false;
311  }
312  // Stop libcurl loading its build-time default bundle, which the callback
313  // below would only discard; the callback supplies the trust anchors.
314  curl_easy_setopt(curl, CURLOPT_CAINFO, static_cast<char *>(nullptr));
315  curl_easy_setopt(curl, CURLOPT_CAPATH, static_cast<char *>(nullptr));
316  curl_easy_setopt(curl, CURLOPT_SSL_CTX_FUNCTION, ssl_ctx_callback);
317  curl_easy_setopt(curl, CURLOPT_SSL_CTX_DATA, &rec);
318  return true;
319  }
320 
321  auto ca_filename = m_ca_file ? m_ca_file->CAFilename() : "";
322  auto crl_filename = m_ca_file ? m_ca_file->CRLFilename() : "";
323  if (!ca_filename.empty() && !crl_filename.empty()) {
324  curl_easy_setopt(curl, CURLOPT_CAINFO, ca_filename.c_str());
325  //Check that the CRL file contains at least one entry before setting this option to curl
326  //Indeed, an empty CRL file will make curl unhappy and therefore will fail
327  //all HTTP TPC transfers (https://github.com/xrootd/xrootd/issues/1543)
328  std::ifstream in(crl_filename, std::ifstream::ate | std::ifstream::binary);
329  if(in.tellg() > 0 && m_ca_file->atLeastOneValidCRLFound()){
330  curl_easy_setopt(curl, CURLOPT_CRLFILE, crl_filename.c_str());
331  if (allowMissingCRL) {
332  // No need to set the callback if there is no need to do it
333  curl_easy_setopt(curl, CURLOPT_SSL_CTX_FUNCTION, ssl_ctx_callback);
334  }
335  } else {
336  std::ostringstream oss;
337  oss << "No valid CRL file has been found in the file " << crl_filename << ". Disabling CRL checking.";
338  m_log.Log(Warning,"TpcHandler",oss.str().c_str());
339  }
340  }
341  else if (!m_cadir.empty()) {
342  curl_easy_setopt(curl, CURLOPT_CAPATH, m_cadir.c_str());
343  }
344  if (!m_cafile.empty()) {
345  curl_easy_setopt(curl, CURLOPT_CAINFO, m_cafile.c_str());
346  }
347  return true;
348 }
349 
350 void
351 TPCHandler::ConfigureCurlLowSpeed(CURL *curl)
352 {
353  // Older versions have poor transfer performance when low-speed limits are
354  // enabled; this was corrected in curl commit cacdc27f for version 7.38.0.
355  curl_version_info_data *curl_ver = curl_version_info(CURLVERSION_NOW);
356  if (m_low_speed_limit > 0 && curl_ver && curl_ver->age > 0 &&
357  curl_ver->version_num >= 0x072600) {
358  curl_easy_setopt(curl, CURLOPT_LOW_SPEED_TIME, m_low_speed_time);
359  curl_easy_setopt(curl, CURLOPT_LOW_SPEED_LIMIT, m_low_speed_limit);
360  }
361 }
362 
363 
364 bool TPCHandler::MatchesPath(const char *verb, const char *path) {
365  return !strcmp(verb, "COPY") || !strcmp(verb, "OPTIONS");
366 }
367 
368 /******************************************************************************/
369 /* P r e p a r e U R L */
370 /******************************************************************************/
371 
372 static std::string PrepareURL(const std::string &url)
373 {
374  const std::string replace_schemes[] = { "davs://", "s3://", "s3s://" };
375 
376  for (const auto& s : replace_schemes)
377  if (url.compare(0, s.size(), s) == 0)
378  return "https://" + url.substr(s.size());
379 
380  return url;
381 }
382 
383 static bool IsAllowedScheme(const std::string& url)
384 {
385  const std::string allowed_schemes[] = { "https://", "http://" };
386 
387  for (const auto& s : allowed_schemes)
388  if (url.compare(0, s.size(), s) == 0)
389  return true;
390 
391  return false;
392 }
393 
394 /******************************************************************************/
395 /* T P C H a n d l e r : : P r o c e s s R e q */
396 /******************************************************************************/
397 
399  if (req.verb == "OPTIONS") {
400  return ProcessOptionsReq(req);
401  }
402  auto header = XrdOucTUtils::caseInsensitiveFind(req.headers,"credential");
403  if (header != req.headers.end()) {
404  if (header->second != "none") {
405  m_log.Emsg("ProcessReq", "COPY requested an unsupported credential type: ", header->second.c_str());
406  return req.SendSimpleResp(400, NULL, NULL, "COPY requestd an unsupported Credential type", 0);
407  }
408  }
409  header = XrdOucTUtils::caseInsensitiveFind(req.headers,"source");
410  if (header != req.headers.end()) {
411  std::string src = PrepareURL(header->second);
412  if (!IsAllowedScheme(src)) {
413  const char *error_src = "COPY rejected: disallowed scheme in source URL";
414  m_log.Emsg("ProcessReq", error_src, src.c_str());
415  return req.SendSimpleResp(400, NULL, NULL, error_src, 0);
416  }
417  return ProcessPullReq(src, req);
418  }
419  header = XrdOucTUtils::caseInsensitiveFind(req.headers,"destination");
420  if (header != req.headers.end()) {
421  const std::string& dst = header->second;
422  if (!IsAllowedScheme(dst)) {
423  const char *error_dst = "COPY rejected: disallowed scheme in destination URL";
424  m_log.Emsg("ProcessReq", error_dst, dst.c_str());
425  return req.SendSimpleResp(400, NULL, NULL, error_dst, 0);
426  }
427  return ProcessPushReq(header->second, req);
428  }
429  m_log.Emsg("ProcessReq", "COPY verb requested but no source or destination specified.");
430  return req.SendSimpleResp(400, NULL, NULL, "No Source or Destination specified", 0);
431 }
432 
433 /******************************************************************************/
434 /* T P C H a n d l e r D e s t r u c t o r */
435 /******************************************************************************/
436 
438  m_sfs = NULL;
439 }
440 
441 /******************************************************************************/
442 /* T P C H a n d l e r C o n s t r u c t o r */
443 /******************************************************************************/
444 
445 TPCHandler::TPCHandler(XrdSysError *log, const char *config, XrdOucEnv *myEnv) :
446  m_allow_local(false),
447  m_allow_private(true),
448  m_desthttps(false),
449  m_fixed_route(false),
450  m_low_speed_limit(10*1024),
451  m_low_speed_time(2*60),
452  m_timeout(60),
453  m_first_timeout(120),
454  m_log(log->logger(), "TPC_"),
455  m_sfs(NULL)
456 {
457  if (!Configure(config, myEnv)) {
458  throw std::runtime_error("Failed to configure the HTTP third-party-copy handler.");
459  }
460 
461 // Extract out the TPC monitoring object (we share it with xrootd).
462 //
463  XrdXrootdGStream *gs = (XrdXrootdGStream*)myEnv->GetPtr("Tpc.gStream*");
464  if (gs)
465  TPCLogRecord::tpcMonitor = new XrdXrootdTpcMon("http",log->logger(),*gs);
466 }
467 
468 /******************************************************************************/
469 /* T P C H a n d l e r : : P r o c e s s O p t i o n s R e q */
470 /******************************************************************************/
471 
475 int TPCHandler::ProcessOptionsReq(XrdHttpExtReq &req) {
476  return req.SendSimpleResp(200, NULL, (char *) "DAV: 1\r\nDAV: <http://apache.org/dav/propset/fs/1>\r\nAllow: HEAD,GET,PUT,PROPFIND,DELETE,OPTIONS,COPY", NULL, 0);
477 }
478 
479 /******************************************************************************/
480 /* T P C H a n d l e r : : G e t A u t h z */
481 /******************************************************************************/
482 
483 std::string TPCHandler::GetAuthz(XrdHttpExtReq &req) {
484  std::string authz;
485  auto authz_header = XrdOucTUtils::caseInsensitiveFind(req.headers,"authorization");
486  if (authz_header != req.headers.end()) {
487  std::stringstream ss;
488  ss << "authz=" << encode_str(authz_header->second);
489  authz += ss.str();
490  }
491  return authz;
492 }
493 
494 /******************************************************************************/
495 /* T P C H a n d l e r : : R e d i r e c t T r a n s f e r */
496 /******************************************************************************/
497 
498 int TPCHandler::RedirectTransfer(CURL *curl, const std::string &redirect_resource,
499  XrdHttpExtReq &req, XrdOucErrInfo &error, TPCLogRecord &rec)
500 {
501  int port;
502  const char *ptr = error.getErrText(port);
503  if ((ptr == NULL) || (*ptr == '\0') || (port == 0)) {
504  rec.status = 500;
505  std::stringstream ss;
506  ss << "Internal error: redirect without hostname";
507  logTransferEvent(LogMask::Error, rec, "REDIRECT_INTERNAL_ERROR", ss.str());
508  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
509  }
510 
511  // The XrdSfs layer hands back the redirect target in host[?cgi] form; the
512  // port arrived separately via getErrText() above. Default to that target
513  // and let the redirect plugin block below override it if it rewrites it.
514  std::string finalTarget = ptr;
515 
516  // When a redirect plugin (XrdXrootdRedirPI) is configured, run the COPY
517  // redirect target through it so the same plugin-driven routing applies
518  // as in the XRootD protocol's fsRedirPI(). The plugin may rewrite host,
519  // port, and CGI; on a fatal plugin error surface a 500 with the plugin's
520  // message rather than emit a redirect we know is wrong. See issue #2767.
521  if (XrdNetAddrInfo *clientAddr = req.GetSecEntity().addrInfo;
522  XrdXrootdRedirHelper::IsActive() && clientAddr) {
523  // Redirect() takes the host[?cgi] target as a single string and splits
524  // it itself; the non-negative port selects its host+port form.
525  int newPort = port;
526  std::string newTarget;
527  std::string errMsg;
528  auto outcome = XrdXrootdRedirHelper::Redirect(ptr, newPort, *clientAddr,
529  newTarget, errMsg);
531  finalTarget = std::move(newTarget);
532  port = newPort;
533  logTransferEvent(LogMask::Info, rec, "REDIRECT_PLUGIN_REWRITE",
534  finalTarget);
535  } else if (outcome == XrdXrootdRedirHelper::Outcome::Error) {
536  rec.status = 500;
537  std::stringstream ess;
538  ess << "Redirect plugin error: " << errMsg;
539  logTransferEvent(LogMask::Error, rec, "REDIRECT_PLUGIN_ERROR",
540  ess.str());
541  return req.SendSimpleResp(rec.status, nullptr, nullptr,
542  generateClientErr(ess, rec).c_str(), 0);
543  }
544  // Outcome::Unchanged: keep the original target.
545  }
546 
547  // Split the (possibly plugin-rewritten) host[?cgi] target: the host goes
548  // into the Location authority, the cgi into its query string. splitHostCgi
549  // keeps the leading '?' on cgi; the Location builder below wants the bare
550  // opaque body, so drop that '?' here.
551  std::string host;
552  std::string cgi;
553  splitHostCgi(finalTarget, host, cgi);
554  std::string opaque = cgi.empty() ? std::string() : cgi.substr(1);
555 
556  std::stringstream ss;
557  ss << "Location: http" << (m_desthttps ? "s" : "") << "://" << host << ":" << port << "/" << redirect_resource;
558 
559  if (!opaque.empty()) {
560  // redirect_resource (sourced from xrd-http-fullresource) may already
561  // carry the client's query string, so pick the separator accordingly
562  // to avoid emitting a malformed URL with two '?'.
563  char sep = (redirect_resource.find('?') == std::string::npos) ? '?' : '&';
564  ss << sep << encode_xrootd_opaque_to_uri(curl, opaque);
565  }
566 
567  rec.status = 307;
568  logTransferEvent(LogMask::Info, rec, "REDIRECT", ss.str());
569  return req.SendSimpleResp(rec.status, NULL, const_cast<char *>(ss.str().c_str()),
570  NULL, 0);
571 }
572 
573 /******************************************************************************/
574 /* T P C H a n d l e r : : O p e n W a i t S t a l l */
575 /******************************************************************************/
576 
577 int TPCHandler::OpenWaitStall(XrdSfsFile &fh, const std::string &resource,
578  int mode, int openMode, const XrdSecEntity &sec,
579  const std::string &authz)
580 {
581  int open_result;
582  while (1) {
583  int orig_ucap = fh.error.getUCap();
584  fh.error.setUCap(orig_ucap | XrdOucEI::uIPv64);
585  std::string opaque;
586  size_t pos = resource.find('?');
587  // Extract the path and opaque info from the resource
588  std::string path = resource.substr(0, pos);
589 
590  if (pos != std::string::npos) {
591  opaque = resource.substr(pos + 1);
592  }
593 
594  // Append the authz information if there are some
595  if(!authz.empty()) {
596  opaque += (opaque.empty() ? "" : "&");
597  opaque += authz;
598  }
599  open_result = fh.open(path.c_str(), mode, openMode, &sec, opaque.c_str());
600 
601  if ((open_result == SFS_STALL) || (open_result == SFS_STARTED)) {
602  int secs_to_stall = fh.error.getErrInfo();
603  if (open_result == SFS_STARTED) {secs_to_stall = secs_to_stall/2 + 5;}
604  std::this_thread::sleep_for (std::chrono::seconds(secs_to_stall));
605  }
606  break;
607  }
608  return open_result;
609 }
610 
611 /******************************************************************************/
612 /* T P C H a n d l e r : : D e t e r m i n e X f e r S i z e */
613 /******************************************************************************/
614 
615 
616 
620 int TPCHandler::PerformHEADRequest(CURL *curl, XrdHttpExtReq &req, State &state,
621  bool &success, TPCLogRecord &rec, bool shouldReturnErrorToClient) {
622  success = false;
623  curl_easy_setopt(curl, CURLOPT_NOBODY, 1);
624  // Set a custom timeout of 60 seconds (= CONNECT_TIMEOUT for convenience) for the HEAD request
625  curl_easy_setopt(curl, CURLOPT_TIMEOUT, CONNECT_TIMEOUT);
626  CURLcode res;
627  res = curl_easy_perform(curl);
628  //Immediately set the CURLOPT_NOBODY flag to 0 as we anyway
629  //don't want the next curl call to do be a HEAD request
630  curl_easy_setopt(curl, CURLOPT_NOBODY, 0);
631  // Reset the CURLOPT_TIMEOUT to no timeout (default)
632  curl_easy_setopt(curl, CURLOPT_TIMEOUT, 0L);
633  curl_easy_setopt(curl, CURLOPT_FAILONERROR, true);
634 
635  std::stringstream ss;
636 
637  if (state.GetStatusCode() >= 400)
638  res = CURLE_HTTP_RETURNED_ERROR;
639 
640  if (res != CURLE_OK) { /* curl failed */
641  ss << curl_easy_strerror(res);
642  switch (res) {
643  case CURLE_HTTP_RETURNED_ERROR: /* remote side may have returned an error */
644  rec.tpc_status = state.GetStatusCode(); /* relay status received from remote side to the client */
645  ss << ": remote host returned '" << rec.tpc_status << " "
646  << httpStatusToString(rec.tpc_status) << "' while fetching file size";
647  break;
648  case CURLE_COULDNT_CONNECT: /* socket callback may have failed */
649  switch (rec.tpc_status) {
650  case 403:
651  ss << ": connection to local/private addresses is forbidden";
652  break;
653  default:
654  ss << ": internal server failure";
655  rec.tpc_status = 500;
656  }
657  break;
658  default:
659  rec.tpc_status = 500;
660  state.SetErrorCode(500);
661  }
662  }
663 
664  if (rec.tpc_status >= 400) {
665  logTransferEvent(LogMask::Error, rec, "HEAD_FAIL", ss.str());
666  return shouldReturnErrorToClient ? req.SendSimpleResp(rec.tpc_status, NULL, NULL, generateClientErr(ss, rec, res).c_str(), 0) : -1;
667  }
668 
669  success = true;
670  ss << "Successfully determined remote file information for pull request: "
671  << "size=" << state.GetContentLength();
672  if(state.GetReprDigest().size()) {
673  unsigned int cksumIndex = 1;
674  for(const auto & [cksumType,cksumValue]: state.GetReprDigest()) {
675  ss << " chksum" << cksumIndex << "=(" << cksumType << "," << cksumValue << ")";
676  cksumIndex++;
677  }
678  }
679  logTransferEvent(LogMask::Debug, rec, "HEAD_SUCCESS", ss.str());
680  return 0;
681 }
682 
683 int TPCHandler::GetRemoteFileInfoTPCPull(CURL *curl, XrdHttpExtReq &req, uint64_t &contentLength, std::map<std::string,std::string> & reprDigest, bool & success, TPCLogRecord &rec) {
684  State state(curl,req.tpcForwardCreds);
685  //Don't forget to copy the headers of the client's request before doing the HEAD call. Otherwise, if there is a need for authentication,
686  //it will fail
687  state.SetupHeadersForHEAD(req);
688  int result;
689  //In case we cannot get the file HEAD request, we return the error to the client
690  if ((result = PerformHEADRequest(curl, req, state, success, rec)) || !success) {
691  return result;
692  }
693  contentLength = state.GetContentLength();
694  reprDigest = state.GetReprDigest();
695  return result;
696 }
697 
698 /******************************************************************************/
699 /* T P C H a n d l e r : : S e n d P e r f M a r k e r */
700 /******************************************************************************/
701 
702 int TPCHandler::SendPerfMarker(XrdHttpExtReq &req, TPCLogRecord &rec, TPC::State &state) {
703  std::stringstream ss;
704  const std::string crlf = "\n";
705  ss << "Perf Marker" << crlf;
706  ss << "Timestamp: " << time(NULL) << crlf;
707  ss << "Stripe Index: 0" << crlf;
708  ss << "Stripe Bytes Transferred: " << state.BytesTransferred() << crlf;
709  ss << "Total Stripe Count: 1" << crlf;
710  // Include the TCP connection associated with this transfer; used by
711  // the TPC client for monitoring purposes.
712  std::string desc = state.GetConnectionDescription();
713  if (!desc.empty())
714  ss << "RemoteConnections: " << desc << crlf;
715  ss << "End" << crlf;
716  rec.bytes_transferred = state.BytesTransferred();
717  logTransferEvent(LogMask::Debug, rec, "PERF_MARKER");
718 
719  return req.ChunkResp(ss.str().c_str(), 0);
720 }
721 
722 /******************************************************************************/
723 /* T P C H a n d l e r : : S e n d P e r f M a r k e r */
724 /******************************************************************************/
725 
726 int TPCHandler::SendPerfMarker(XrdHttpExtReq &req, TPCLogRecord &rec, std::vector<State*> &state,
727  off_t bytes_transferred)
728 {
729  // The 'performance marker' format is largely derived from how GridFTP works
730  // (e.g., the concept of `Stripe` is not quite so relevant here). See:
731  // https://twiki.cern.ch/twiki/bin/view/LCG/HttpTpcTechnical
732  // Example marker:
733  // Perf Marker\n
734  // Timestamp: 1537788010\n
735  // Stripe Index: 0\n
736  // Stripe Bytes Transferred: 238745\n
737  // Total Stripe Count: 1\n
738  // RemoteConnections: tcp:129.93.3.4:1234,tcp:[2600:900:6:1301:268a:7ff:fef6:a590]:2345\n
739  // End\n
740  //
741  std::stringstream ss;
742  const std::string crlf = "\n";
743  ss << "Perf Marker" << crlf;
744  ss << "Timestamp: " << time(NULL) << crlf;
745  ss << "Stripe Index: 0" << crlf;
746  ss << "Stripe Bytes Transferred: " << bytes_transferred << crlf;
747  ss << "Total Stripe Count: 1" << crlf;
748  // Build a list of TCP connections associated with this transfer; used by
749  // the TPC client for monitoring purposes.
750  bool first = true;
751  std::stringstream ss2;
752  for (std::vector<State*>::const_iterator iter = state.begin();
753  iter != state.end(); iter++)
754  {
755  std::string desc = (*iter)->GetConnectionDescription();
756  if (!desc.empty()) {
757  ss2 << (first ? "" : ",") << desc;
758  first = false;
759  }
760  }
761  if (!first)
762  ss << "RemoteConnections: " << ss2.str() << crlf;
763  ss << "End" << crlf;
764  rec.bytes_transferred = bytes_transferred;
765  logTransferEvent(LogMask::Debug, rec, "PERF_MARKER");
766 
767  return req.ChunkResp(ss.str().c_str(), 0);
768 }
769 
770 /******************************************************************************/
771 /* T P C H a n d l e r : : R u n C u r l W i t h U p d a t e s */
772 /******************************************************************************/
773 
774 int TPCHandler::RunCurlWithUpdates(CURL *curl, XrdHttpExtReq &req, State &state,
775  TPCLogRecord &rec)
776 {
777  // Create the multi-handle and add in the current transfer to it.
778  CURLM *multi_handle = curl_multi_init();
779  if (!multi_handle) {
780  rec.status = 500;
781  logTransferEvent(LogMask::Error, rec, "CURL_INIT_FAIL",
782  "Failed to initialize a libcurl multi-handle");
783  std::stringstream ss;
784  ss << "Failed to initialize internal server memory";
785  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
786  }
787 
788  //curl_easy_setopt(curl, CURLOPT_BUFFERSIZE, 128*1024);
789 
790  CURLMcode mres;
791  mres = curl_multi_add_handle(multi_handle, curl);
792  if (mres) {
793  rec.status = 500;
794  std::stringstream ss;
795  ss << "Failed to add transfer to libcurl multi-handle: HTTP library failure=" << curl_multi_strerror(mres);
796  logTransferEvent(LogMask::Error, rec, "CURL_INIT_FAIL", ss.str());
797  curl_multi_cleanup(multi_handle);
798  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
799  }
800 
801  // Start response to client prior to the first call to curl_multi_perform
802  int retval = req.StartChunkedResp(202, NULL, "Content-Type: text/plain");
803  if (retval) {
804  curl_multi_cleanup(multi_handle);
805  logTransferEvent(LogMask::Error, rec, "RESPONSE_FAIL",
806  "Failed to send the initial response to the TPC client");
807  return retval;
808  } else {
809  logTransferEvent(LogMask::Debug, rec, "RESPONSE_START",
810  "Initial transfer response sent to the TPC client");
811  }
812 
813  // Transfer loop: use curl to actually run the transfer, but periodically
814  // interrupt things to send back performance updates to the client.
815  int running_handles = 1;
816  time_t last_marker = 0;
817  // Track how long it's been since the last time we recorded more bytes being transferred.
818  off_t last_advance_bytes = 0;
819  time_t last_advance_time = time(NULL);
820  time_t transfer_start = last_advance_time;
821  CURLcode res = static_cast<CURLcode>(-1);
822  do {
823  time_t now = time(NULL);
824  time_t next_marker = last_marker + m_marker_period;
825  if (now >= next_marker) {
826  off_t bytes_xfer = state.BytesTransferred();
827  if (bytes_xfer > last_advance_bytes) {
828  last_advance_bytes = bytes_xfer;
829  last_advance_time = now;
830  }
831  if (SendPerfMarker(req, rec, state)) {
832  curl_multi_remove_handle(multi_handle, curl);
833  curl_multi_cleanup(multi_handle);
834  logTransferEvent(LogMask::Error, rec, "PERFMARKER_FAIL",
835  "Failed to send a perf marker to the TPC client");
836  return -1;
837  }
838  int timeout = (transfer_start == last_advance_time) ? m_first_timeout : m_timeout;
839  if (now > last_advance_time + timeout) {
840  const char *log_prefix = rec.log_prefix.c_str();
841  bool tpc_pull = strncmp("Pull", log_prefix, 4) == 0;
842 
844  std::stringstream ss;
845  ss << "Transfer failed because no bytes have been "
846  << (tpc_pull ? "received from the source (pull mode) in "
847  : "transmitted to the destination (push mode) in ") << timeout << " seconds.";
848  state.SetErrorMessage(ss.str());
849  curl_multi_remove_handle(multi_handle, curl);
850  curl_multi_cleanup(multi_handle);
851  break;
852  }
853  last_marker = now;
854  }
855  // The transfer will start after this point, notify the packet marking manager
856  rec.pmarkManager.startTransfer();
857  mres = curl_multi_perform(multi_handle, &running_handles);
858  if (mres == CURLM_CALL_MULTI_PERFORM) {
859  // curl_multi_perform should be called again immediately. On newer
860  // versions of curl, this is no longer used.
861  continue;
862  } else if (mres != CURLM_OK) {
863  break;
864  } else if (running_handles == 0) {
865  break;
866  }
867 
868  rec.pmarkManager.beginPMarks();
869  //printf("There are %d running handles\n", running_handles);
870 
871  // Harvest any messages, looking for CURLMSG_DONE.
872  CURLMsg *msg;
873  do {
874  int msgq = 0;
875  msg = curl_multi_info_read(multi_handle, &msgq);
876  if (msg && (msg->msg == CURLMSG_DONE)) {
877  CURL *easy_handle = msg->easy_handle;
878  res = msg->data.result;
879  curl_multi_remove_handle(multi_handle, easy_handle);
880  }
881  } while (msg);
882 
883  int64_t max_sleep_time = next_marker - time(NULL);
884  if (max_sleep_time <= 0) {
885  continue;
886  }
887  int fd_count;
888  mres = curl_multi_wait(multi_handle, NULL, 0, max_sleep_time*1000, &fd_count);
889  if (mres != CURLM_OK) {
890  break;
891  }
892  } while (running_handles);
893 
894  if (mres != CURLM_OK) {
895  std::stringstream ss;
896  ss << "Internal libcurl multi-handle error: HTTP library failure=" << curl_multi_strerror(mres);
897  logTransferEvent(LogMask::Error, rec, "TRANSFER_CURL_ERROR", ss.str());
898 
899  curl_multi_remove_handle(multi_handle, curl);
900  curl_multi_cleanup(multi_handle);
901 
902  if ((retval = req.ChunkResp(generateClientErr(ss, rec).c_str(), 0))) {
903  logTransferEvent(LogMask::Error, rec, "RESPONSE_FAIL",
904  "Failed to send error message to the TPC client");
905  return retval;
906  }
907  return req.ChunkResp(NULL, 0);
908  }
909 
910  // Harvest any messages, looking for CURLMSG_DONE.
911  CURLMsg *msg;
912  do {
913  int msgq = 0;
914  msg = curl_multi_info_read(multi_handle, &msgq);
915  if (msg && (msg->msg == CURLMSG_DONE)) {
916  CURL *easy_handle = msg->easy_handle;
917  res = msg->data.result;
918  curl_multi_remove_handle(multi_handle, easy_handle);
919  }
920  } while (msg);
921 
922  if (!state.GetErrorCode() && res == static_cast<CURLcode>(-1)) { // No transfers returned?!?
923  curl_multi_remove_handle(multi_handle, curl);
924  curl_multi_cleanup(multi_handle);
925  std::stringstream ss;
926  ss << "Internal state error in libcurl";
927  logTransferEvent(LogMask::Error, rec, "TRANSFER_CURL_ERROR", ss.str());
928 
929  if ((retval = req.ChunkResp(generateClientErr(ss, rec).c_str(), 0))) {
930  logTransferEvent(LogMask::Error, rec, "RESPONSE_FAIL",
931  "Failed to send error message to the TPC client");
932  return retval;
933  }
934  return req.ChunkResp(NULL, 0);
935  }
936  curl_multi_cleanup(multi_handle);
937 
938  // The transfer is over at this point: any error recorded so far - a failed
939  // write to the local filesystem or the stall detector having fired - is the
940  // reason why the transfer failed. Flushing and closing the destination file
941  // below may fail as well but, as such a failure is usually a consequence of
942  // the transfer failure, it must not be reported instead of it.
943  const int transferErrorCode = state.GetErrorCode();
944  std::string transferErrorMsg = state.GetErrorMessage();
945 
946  state.Flush();
947 
948  rec.bytes_transferred = state.BytesTransferred();
949  rec.tpc_status = state.GetStatusCode();
950 
951  // Explicitly finalize the stream (which will close the underlying file
952  // handle) before the response is sent. In some cases, subsequent HTTP
953  // requests can occur before the filesystem is done closing the handle -
954  // and those requests may occur against partial data.
955  state.Finalize();
956 
957  // A failure to flush or to close the destination file is always logged and is
958  // appended to the error reported to the client, but it never replaces the
959  // transfer failure itself: it is usually a consequence of it.
960  std::string finalizeErrorMsg, finalizeErrorSuffix;
961  if (state.GetFinalizeErrorCode()) {
962  std::stringstream ss2;
963  ss2 << (state.GetFinalizeErrorCode() == State::errFlush
964  ? "Failed to flush the file to the local filesystem."
965  : "Failed to finalize and close file handle.");
966  std::string err = state.GetFinalizeErrorMessage();
967  if (!err.empty()) {
968  std::replace(err.begin(), err.end(), '\n', ' ');
969  ss2 << " " << err;
970  }
971  finalizeErrorMsg = ss2.str();
972  logTransferEvent(LogMask::Error, rec, "CLOSE_FAIL", finalizeErrorMsg);
973  finalizeErrorSuffix = "; " + finalizeErrorMsg;
974  }
975 
976  // Generate the final response back to the client.
977  std::stringstream ss;
978  bool success = false;
979  if (state.GetStatusCode() >= 400) {
980  std::string err = state.GetErrorMessage();
981  std::stringstream ss2;
982  ss2 << "Remote side failed with status code " << state.GetStatusCode();
983  if (!err.empty()) {
984  std::replace(err.begin(), err.end(), '\n', ' ');
985  ss2 << "; error message: \"" << err << "\"";
986  }
987  logTransferEvent(LogMask::Error, rec, "TRANSFER_FAIL", ss2.str());
988  ss2 << finalizeErrorSuffix;
989  ss << generateClientErr(ss2, rec);
990  } else if (transferErrorCode == State::errTimeout) {
991  // The stall detector fired; its message already describes precisely
992  // what happened, report it as-is.
993  std::stringstream ss2;
994  ss2 << transferErrorMsg;
995  logTransferEvent(LogMask::Error, rec, "TRANSFER_FAIL", ss2.str());
996  ss2 << finalizeErrorSuffix;
997  ss << generateClientErr(ss2, rec);
998  } else if (transferErrorCode) {
999  if (transferErrorMsg.empty()) {transferErrorMsg = "(no error message provided)";}
1000  else {std::replace(transferErrorMsg.begin(), transferErrorMsg.end(), '\n', ' ');}
1001  std::stringstream ss2;
1002  ss2 << "Error when interacting with local filesystem: " << transferErrorMsg;
1003  logTransferEvent(LogMask::Error, rec, "TRANSFER_FAIL", ss2.str());
1004  ss2 << finalizeErrorSuffix;
1005  ss << generateClientErr(ss2, rec);
1006  } else if (res != CURLE_OK) {
1007  std::stringstream ss2;
1008  ss2 << "Internal transfer failure";
1009  std::stringstream ss3;
1010  ss3 << ss2.str() << ": " << curl_easy_strerror(res);
1011  logTransferEvent(LogMask::Error, rec, "TRANSFER_FAIL", ss3.str());
1012  ss2 << finalizeErrorSuffix;
1013  ss << generateClientErr(ss2, rec, res);
1014  } else if (!finalizeErrorMsg.empty()) {
1015  // Nothing else went wrong: the flush/close failure is the reason of the failure.
1016  std::stringstream ss2;
1017  ss2 << finalizeErrorMsg;
1018  ss << generateClientErr(ss2, rec);
1019  } else {
1020  ss << "success: Created";
1021  success = true;
1022  }
1023 
1024  if ((retval = req.ChunkResp(ss.str().c_str(), 0))) {
1025  logTransferEvent(LogMask::Error, rec, "TRANSFER_ERROR",
1026  "Failed to send last update to remote client");
1027  return retval;
1028  } else if (success) {
1029  logTransferEvent(LogMask::Info, rec, "TRANSFER_SUCCESS");
1030  rec.status = 0;
1031  }
1032  return req.ChunkResp(NULL, 0);
1033 }
1034 
1035 /******************************************************************************/
1036 /* T P C H a n d l e r : : P r o c e s s P u s h R e q */
1037 /******************************************************************************/
1038 
1039 int TPCHandler::ProcessPushReq(const std::string & resource, XrdHttpExtReq &req) {
1040  TPCLogRecord rec(req, TpcType::Push);
1041  rec.allow_local = m_allow_local;
1042  rec.allow_private = m_allow_private;
1043  rec.log_prefix = "PushRequest";
1044  rec.local = req.resource;
1045  rec.remote = resource;
1046  rec.m_log = &m_log;
1047  char *name = req.GetSecEntity().name;
1048  req.GetClientID(rec.clID);
1049  if (name) rec.name = name;
1050  logTransferEvent(LogMask::Info, rec, "PUSH_START", "Starting a push request");
1051 
1052  ManagedCurlHandle curlPtr(curl_easy_init());
1053  auto curl = curlPtr.get();
1054  if (!curl) {
1055  std::stringstream ss;
1056  ss << "Failed to initialize internal transfer resources";
1057  rec.status = 500;
1058  logTransferEvent(LogMask::Error, rec, "PUSH_FAIL", ss.str());
1059  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
1060  }
1061  ConfigureCurlLowSpeed(curl);
1062  curl_easy_setopt(curl, CURLOPT_NOSIGNAL, 1);
1063  curl_easy_setopt(curl, CURLOPT_SSLVERSION, CURL_SSLVERSION_TLSv1_2);
1064  curl_easy_setopt(curl, CURLOPT_HTTP_VERSION, (long) CURL_HTTP_VERSION_1_1);
1065 #if CURL_AT_LEAST_VERSION(7, 85, 0)
1066  curl_easy_setopt(curl, CURLOPT_PROTOCOLS_STR, "https,http");
1067  curl_easy_setopt(curl, CURLOPT_REDIR_PROTOCOLS_STR, "https,http");
1068 #else
1069  long protocols = CURLPROTO_HTTP | CURLPROTO_HTTPS;
1070  curl_easy_setopt(curl, CURLOPT_PROTOCOLS, protocols);
1071  curl_easy_setopt(curl, CURLOPT_REDIR_PROTOCOLS, protocols);
1072 #endif
1073  curl_easy_setopt(curl, CURLOPT_OPENSOCKETFUNCTION, opensocket_callback);
1074  curl_easy_setopt(curl, CURLOPT_OPENSOCKETDATA, &rec);
1075  curl_easy_setopt(curl, CURLOPT_CLOSESOCKETFUNCTION, closesocket_callback);
1076  curl_easy_setopt(curl, CURLOPT_SOCKOPTFUNCTION, sockopt_callback);
1077  curl_easy_setopt(curl, CURLOPT_CLOSESOCKETDATA, &rec);
1078  curl_easy_setopt(curl, CURLOPT_CONNECTTIMEOUT, CONNECT_TIMEOUT);
1079 
1080  auto query_header = XrdOucTUtils::caseInsensitiveFind(req.headers,"xrd-http-fullresource");
1081  std::string redirect_resource = req.resource;
1082  if (query_header != req.headers.end()) {
1083  redirect_resource = query_header->second;
1084  }
1085 
1086  AtomicBeg(m_monid_mutex);
1087  uint64_t file_monid = AtomicInc(m_monid);
1088  AtomicEnd(m_monid_mutex);
1089  std::unique_ptr<XrdSfsFile> fh(m_sfs->newFile(name, file_monid));
1090  if (!fh.get()) {
1091  rec.status = 500;
1092  std::stringstream ss;
1093  ss << "Failed to initialize internal transfer file handle";
1094  logTransferEvent(LogMask::Error, rec, "OPEN_FAIL",
1095  ss.str());
1096  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
1097  }
1098  std::string full_url = prepareURL(req);
1099 
1100  std::string authz = GetAuthz(req);
1101 
1102  int open_results = OpenWaitStall(*fh, full_url, SFS_O_RDONLY, 0644,
1103  req.GetSecEntity(), authz);
1104  if (SFS_REDIRECT == open_results) {
1105  int result = RedirectTransfer(curl, redirect_resource, req, fh->error, rec);
1106  return result;
1107  } else if (SFS_OK != open_results) {
1108  int code;
1109  std::stringstream ss;
1110  const char *msg = fh->error.getErrText(code);
1111  if (msg == NULL) ss << "Failed to open local resource";
1112  else ss << msg;
1113  rec.status = mapErrNoToHttp(code);
1114  logTransferEvent(LogMask::Error, rec, "OPEN_FAIL", msg);
1115  int resp_result = req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
1116  fh->close();
1117  return resp_result;
1118  }
1119  if (!ConfigureCurlCA(curl, rec)) {
1120  std::stringstream ss;
1121  ss << "Failed to configure the certificate authorities for the transfer";
1122  rec.status = 500;
1123  logTransferEvent(LogMask::Error, rec, "PUSH_FAIL", ss.str());
1124  int resp_result = req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
1125  fh->close();
1126  return resp_result;
1127  }
1128  curl_easy_setopt(curl, CURLOPT_URL, resource.c_str());
1129 
1130  Stream stream(std::move(fh), 0, 0, m_log);
1131  State state(0, stream, curl, true, req.tpcForwardCreds);
1132  state.SetupHeaders(req);
1133 
1134  return RunCurlWithUpdates(curl, req, state, rec);
1135 }
1136 
1137 /******************************************************************************/
1138 /* T P C H a n d l e r : : P r o c e s s P u l l R e q */
1139 /******************************************************************************/
1140 
1141 int TPCHandler::ProcessPullReq(const std::string &resource, XrdHttpExtReq &req) {
1142  TPCLogRecord rec(req,TpcType::Pull);
1143  rec.allow_local = m_allow_local;
1144  rec.allow_private = m_allow_private;
1145  rec.log_prefix = "PullRequest";
1146  rec.local = req.resource;
1147  rec.remote = resource;
1148  rec.m_log = &m_log;
1149  char *name = req.GetSecEntity().name;
1150  req.GetClientID(rec.clID);
1151  if (name) rec.name = name;
1152  logTransferEvent(LogMask::Info, rec, "PULL_START", "Starting a pull request");
1153 
1154  ManagedCurlHandle curlPtr(curl_easy_init());
1155  auto curl = curlPtr.get();
1156  if (!curl) {
1157  std::stringstream ss;
1158  ss << "Failed to initialize internal transfer resources";
1159  rec.status = 500;
1160  logTransferEvent(LogMask::Error, rec, "PULL_FAIL", ss.str());
1161  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
1162  }
1163  ConfigureCurlLowSpeed(curl);
1164 
1165  // ddavila 2023-01-05:
1166  // The following change was required by the Rucio/SENSE project where
1167  // multiple IP addresses, each from a different subnet, are assigned to a
1168  // single server and routed differently by SENSE.
1169  // The above requires the server to utilize the same IP, that was used to
1170  // start the TPC, for the resolution of the given TPC instead of
1171  // using any of the IPs available.
1172  if (m_fixed_route) {
1173  char ip[64];
1174  char ipType = 0;
1175 
1176  XrdNetAddrInfo *addrInfo = req.GetSecEntity().addrInfo;
1177  int sockFD = addrInfo ? addrInfo->SockFD() : -1;
1178 
1179  if (sockFD < 0 || XrdNetUtils::GetSokInfo(-sockFD, ip, sizeof(ip), ipType) < 0) {
1180  // The socket information could not be fetched for some reason, treat this tpc.fixed_route as "best-effort" instead
1181  // of failing the transfer
1182  logTransferEvent(LogMask::Error, rec, "FIXED_ROUTE_ERR", "Failed to determine local address of incoming fixed route request");
1183  } else {
1184  logTransferEvent(LogMask::Info, rec, "LOCAL IP", ip);
1185  curl_easy_setopt(curl, CURLOPT_INTERFACE, ip);
1186  }
1187  }
1188  curl_easy_setopt(curl, CURLOPT_NOSIGNAL, 1);
1189  curl_easy_setopt(curl, CURLOPT_SSLVERSION, CURL_SSLVERSION_TLSv1_2);
1190  curl_easy_setopt(curl, CURLOPT_HTTP_VERSION, (long) CURL_HTTP_VERSION_1_1);
1191 #if CURL_AT_LEAST_VERSION(7, 85, 0)
1192  curl_easy_setopt(curl, CURLOPT_PROTOCOLS_STR, "https,http");
1193  curl_easy_setopt(curl, CURLOPT_REDIR_PROTOCOLS_STR, "https,http");
1194 #else
1195  long protocols = CURLPROTO_HTTP | CURLPROTO_HTTPS;
1196  curl_easy_setopt(curl, CURLOPT_PROTOCOLS, protocols);
1197  curl_easy_setopt(curl, CURLOPT_REDIR_PROTOCOLS, protocols);
1198 #endif
1199  curl_easy_setopt(curl, CURLOPT_OPENSOCKETFUNCTION, opensocket_callback);
1200  curl_easy_setopt(curl, CURLOPT_OPENSOCKETDATA, &rec);
1201  curl_easy_setopt(curl, CURLOPT_SOCKOPTFUNCTION, sockopt_callback);
1202  curl_easy_setopt(curl, CURLOPT_SOCKOPTDATA , &rec);
1203  curl_easy_setopt(curl, CURLOPT_CLOSESOCKETFUNCTION, closesocket_callback);
1204  curl_easy_setopt(curl, CURLOPT_CLOSESOCKETDATA, &rec);
1205  curl_easy_setopt(curl, CURLOPT_CONNECTTIMEOUT, CONNECT_TIMEOUT);
1206  std::unique_ptr<XrdSfsFile> fh(m_sfs->newFile(name, m_monid++));
1207  if (!fh.get()) {
1208  std::stringstream ss;
1209  ss << "Failed to initialize internal transfer file handle";
1210  rec.status = 500;
1211  logTransferEvent(LogMask::Error, rec, "PULL_FAIL", ss.str());
1212  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
1213  }
1214  auto query_header = XrdOucTUtils::caseInsensitiveFind(req.headers,"xrd-http-fullresource");
1215  std::string redirect_resource = req.resource;
1216  if (query_header != req.headers.end()) {
1217  redirect_resource = query_header->second;
1218  }
1220  auto overwrite_header = XrdOucTUtils::caseInsensitiveFind(req.headers,"overwrite");
1221  if ((overwrite_header == req.headers.end()) || (overwrite_header->second == "T")) {
1222  if (! usingEC) mode = SFS_O_TRUNC;
1223  }
1224  int streams = 1;
1225  {
1226  auto streams_header = XrdOucTUtils::caseInsensitiveFind(req.headers,"x-number-of-streams");
1227  if (streams_header != req.headers.end()) {
1228  int stream_req = -1;
1229  try {
1230  stream_req = std::stol(streams_header->second);
1231  } catch (...) { // Handled below
1232  }
1233  if (stream_req < 0 || stream_req > 100) {
1234  std::stringstream ss;
1235  ss << "Invalid request for number of streams";
1236  rec.status = 400;
1237  logTransferEvent(LogMask::Info, rec, "INVALID_REQUEST", ss.str());
1238  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
1239  }
1240  streams = stream_req == 0 ? 1 : stream_req;
1241  }
1242  }
1243  rec.streams = streams;
1244  std::string full_url = prepareURL(req);
1245  std::string authz = GetAuthz(req);
1246  curl_easy_setopt(curl, CURLOPT_URL, resource.c_str());
1247  if (!ConfigureCurlCA(curl, rec)) {
1248  std::stringstream ss;
1249  ss << "Failed to configure the certificate authorities for the transfer";
1250  rec.status = 500;
1251  logTransferEvent(LogMask::Error, rec, "PULL_FAIL", ss.str());
1252  return req.SendSimpleResp(rec.status, NULL, NULL, generateClientErr(ss, rec).c_str(), 0);
1253  }
1254  uint64_t sourceFileContentLength = 0;
1255  {
1256  //Get the content-length of the source file and pass it to the OSS layer
1257  //during the open
1258  bool success = false;
1259  bool mismatchDigests = false;
1260  std::map<std::string,std::string> sourceFileReprDigest;
1261  GetRemoteFileInfoTPCPull(curl, req, sourceFileContentLength, sourceFileReprDigest, success, rec);
1262  if(success) {
1263  //In the case we cannot get the information from the source server (offline or other error)
1264  //we just don't add the file information to the opaque of the local file to open
1265  full_url += "&oss.asize=" + std::to_string(sourceFileContentLength);
1266  mismatchDigests = mismatchReprDigest(sourceFileReprDigest,req,rec);
1267  }
1268  if(!success || mismatchDigests) {
1269  // We could not get remote file information, or the checksum provided by the client
1270  // does not match the source file one, we already sent the error to the client so we
1271  // just exit here
1272  return 0;
1273  }
1274  }
1275  int open_result = OpenWaitStall(*fh, full_url, mode|SFS_O_WRONLY,
1276  0644 | SFS_O_MKPTH,
1277  req.GetSecEntity(), authz);
1278  if (SFS_REDIRECT == open_result) {
1279  int result = RedirectTransfer(curl, redirect_resource, req, fh->error, rec);
1280  return result;
1281  } else if (SFS_OK != open_result) {
1282  int code;
1283  std::stringstream ss;
1284  const char *msg = fh->error.getErrText(code);
1285  if ((msg == NULL) || (*msg == '\0')) ss << "Failed to open local resource";
1286  else ss << msg;
1287  rec.status = mapErrNoToHttp(code);
1288  logTransferEvent(LogMask::Error, rec, "OPEN_FAIL", ss.str());
1289  int resp_result = req.SendSimpleResp(rec.status, NULL, NULL,
1290  generateClientErr(ss, rec).c_str(), 0);
1291  fh->close();
1292  return resp_result;
1293  }
1294  Stream stream(std::move(fh), streams * m_pipelining_multiplier, streams > 1 ? m_block_size : m_small_block_size, m_log);
1295  State state(0, stream, curl, false, req.tpcForwardCreds);
1296  state.SetupHeaders(req);
1297  state.SetContentLength(sourceFileContentLength);
1298 
1299  if (streams > 1) {
1300  return RunCurlWithStreams(req, state, streams, rec);
1301  } else {
1302  return RunCurlWithUpdates(curl, req, state, rec);
1303  }
1304 }
1305 
1306 /******************************************************************************/
1307 /* T P C H a n d l e r : : l o g T r a n s f e r E v e n t */
1308 /******************************************************************************/
1309 
1310 void TPCHandler::logTransferEvent(LogMask mask, const TPCLogRecord &rec,
1311  const std::string &event, const std::string &message)
1312 {
1313  if (!(m_log.getMsgMask() & mask)) {return;}
1314 
1315  std::stringstream ss;
1316  ss << "event=" << event << ", local=" << rec.local << ", remote=" << rec.remote;
1317  if (rec.name.empty())
1318  ss << ", user=(anonymous)";
1319  else
1320  ss << ", user=" << rec.name;
1321  if (rec.streams != 1)
1322  ss << ", streams=" << rec.streams;
1323  if (rec.bytes_transferred >= 0)
1324  ss << ", bytes_transferred=" << rec.bytes_transferred;
1325  if (rec.status >= 0)
1326  ss << ", status=" << rec.status;
1327  if (rec.tpc_status >= 0)
1328  ss << ", tpc_status=" << rec.tpc_status;
1329  if (!message.empty())
1330  ss << "; " << message;
1331  m_log.Log(mask, rec.log_prefix.c_str(), ss.str().c_str());
1332 }
1333 
1334 std::string TPCHandler::generateClientErr(std::stringstream &err_ss, const TPCLogRecord &rec, CURLcode cCode) {
1335  std::stringstream ssret;
1336  ssret << "failure: " << err_ss.str() << ", local=" << rec.local <<", remote=" << rec.remote;
1337  if(cCode != CURLcode::CURLE_OK) {
1338  ssret << ", HTTP library failure=" << curl_easy_strerror(cCode);
1339  }
1340  return ssret.str();
1341 }
1342 /******************************************************************************/
1343 /* X r d H t t p G e t E x t H a n d l e r */
1344 /******************************************************************************/
1345 
1346 extern "C" {
1347 
1348 XrdHttpExtHandler *XrdHttpGetExtHandler(XrdSysError *log, const char * config, const char * /*parms*/, XrdOucEnv *myEnv) {
1349  if (curl_global_init(CURL_GLOBAL_DEFAULT)) {
1350  log->Emsg("TPCInitialize", "libcurl failed to initialize");
1351  return NULL;
1352  }
1353 
1354  TPCHandler *retval{NULL};
1355  if (!config) {
1356  log->Emsg("TPCInitialize", "TPC handler requires a config filename in order to load");
1357  return NULL;
1358  }
1359  try {
1360  log->Emsg("TPCInitialize", "Will load configuration for the TPC handler from", config);
1361  retval = new TPCHandler(log, config, myEnv);
1362  } catch (std::runtime_error &re) {
1363  log->Emsg("TPCInitialize", "Encountered a runtime failure when loading ", re.what());
1364  //printf("Provided env vars: %p, XrdInet*: %p\n", myEnv, myEnv->GetPtr("XrdInet*"));
1365  }
1366  return retval;
1367 }
1368 
1369 }
void CURL
XrdVERSIONINFO(XrdHttpGetExtHandler, HttpTPC)
XrdHttpExtHandler * XrdHttpGetExtHandler(XrdSysError *log, const char *config, const char *, XrdOucEnv *myEnv)
static std::string PrepareURL(const std::string &url)
std::string encode_xrootd_opaque_to_uri(CURL *curl, const std::string &opaque)
static bool IsAllowedScheme(const std::string &url)
int mapErrNoToHttp(int errNo)
std::string httpStatusToString(int status)
Utility functions for XrdHTTP.
std::string encode_str(const std::string &str)
void splitHostCgi(std::string_view target, std::string &host, std::string &cgi)
#define close(a)
Definition: XrdPosix.hh:48
bool Debug
void getline(uchar *buff, int blen)
#define SFS_REDIRECT
#define SFS_O_MKPTH
#define SFS_STALL
#define SFS_O_RDONLY
#define SFS_STARTED
#define SFS_O_WRONLY
#define SFS_O_CREAT
int XrdSfsFileOpenMode
#define SFS_OK
#define SFS_O_TRUNC
#define AtomicInc(x)
#define AtomicBeg(Mtx)
#define AtomicEnd(Mtx)
@ Error
int GetFinalizeErrorCode() const
int GetStatusCode() const
off_t BytesTransferred() const
void SetErrorMessage(const std::string &error_msg)
int GetErrorCode() const
std::string GetFinalizeErrorMessage() const
std::string GetErrorMessage() const
std::string GetConnectionDescription()
void SetupHeaders(XrdHttpExtReq &req)
void SetContentLength(const off_t content_length)
off_t GetContentLength() const
void SetErrorCode(int error_code)
const std::map< std::string, std::string > & GetReprDigest() const
void SetupHeadersForHEAD(XrdHttpExtReq &req)
TPCHandler(XrdSysError *log, const char *config, XrdOucEnv *myEnv)
virtual int ProcessReq(XrdHttpExtReq &req)
virtual ~TPCHandler()
virtual bool MatchesPath(const char *verb, const char *path)
Tells if the incoming path is recognized as one of the paths that have to be processed.
int ChunkResp(const char *body, long long bodylen)
Send a (potentially partial) body in a chunked response; invoking with NULL body.
void GetClientID(std::string &clid)
std::map< std::string, std::string > & headers
std::string resource
std::string verb
std::map< std::string, std::string > mReprDigest
Repr-Digest map where the key is the digest name and the value is the base64 encoded digest value.
int StartChunkedResp(int code, const char *desc, const char *header_to_add)
Starts a chunked response; body of request is sent over multiple parts using the SendChunkResp.
const XrdSecEntity & GetSecEntity() const
int SendSimpleResp(int code, const char *desc, const char *header_to_add, const char *body, long long bodylen)
Sends a basic response. If the length is < 0 then it is calculated internally.
static std::string prepareOpenURL(PrepareOpenURLParams &params)
static int GetSokInfo(int fd, char *theAddr, int theALen, char &theType)
Definition: XrdNetUtils.cc:533
void * GetPtr(const char *varname)
Definition: XrdOucEnv.cc:281
const char * getErrText()
void setUCap(int ucval)
Set user capabilties.
static std::map< std::string, T >::const_iterator caseInsensitiveFind(const std::map< std::string, T > &m, const std::string &lowerCaseSearchKey)
Definition: XrdOucTUtils.hh:79
XrdNetAddrInfo * addrInfo
Entity's connection details.
Definition: XrdSecEntity.hh:80
char * name
Entity's name.
Definition: XrdSecEntity.hh:69
virtual XrdSfsFile * newFile(char *user=0, int MonID=0)=0
XrdOucErrInfo & error
virtual int open(const char *fileName, XrdSfsFileOpenMode openMode, mode_t createMode, const XrdSecEntity *client=0, const char *opaque=0)=0
virtual int close()=0
int Emsg(const char *esfx, int ecode, const char *text1, const char *text2=0)
Definition: XrdSysError.cc:116
XrdSysLogger * logger(XrdSysLogger *lp=0)
Definition: XrdSysError.hh:175
int getMsgMask()
Definition: XrdSysError.hh:190
void Log(int mask, const char *esfx, const char *text1, const char *text2=0, const char *text3=0)
Definition: XrdSysError.hh:167
static Outcome Redirect(const char *trg, int &port, XrdNetAddrInfo &clientAddr, std::string &outTarget, std::string &errMsg)
std::unique_ptr< CURL, CurlDeleter > ManagedCurlHandle
@ Warning
void operator()(CURL *curl)
static const int uIPv64
ucap: Supports only IPv4 info
static const int isaPush