1 /*---------------------------------------------------------------------\
3 | |__ / \ / / . \ . \ |
8 \---------------------------------------------------------------------*/
9 /** \file zypp/media/MediaMultiCurl.cc
14 #include <sys/types.h>
18 #include <arpa/inet.h>
25 #include "zypp/ZConfig.h"
26 #include "zypp/base/Logger.h"
27 #include "zypp/media/MediaMultiCurl.h"
28 #include "zypp/media/MetaLinkParser.h"
29 #include "zypp/ManagedFile.h"
32 using namespace zypp::base;
34 #undef CURLVERSION_AT_LEAST
35 #define CURLVERSION_AT_LEAST(M,N,O) LIBCURL_VERSION_NUM >= ((((M)<<8)+(N))<<8)+(O)
41 //////////////////////////////////////////////////////////////////////
44 class multifetchrequest;
46 // Hack: we derive from MediaCurl just to get the storage space for
47 // settings, url, curlerrors and the like
49 class multifetchworker : MediaCurl {
50 friend class multifetchrequest;
53 multifetchworker(int no, multifetchrequest &request, const Url &url);
58 bool recheckChecksum();
59 void disableCompetition();
62 void adddnsfd(fd_set &rset, int &maxfd);
63 void dnsevent(fd_set &rset);
87 size_t writefunction(void *ptr, size_t size);
88 static size_t _writefunction(void *ptr, size_t size, size_t nmemb, void *stream);
90 size_t headerfunction(char *ptr, size_t size);
91 static size_t _headerfunction(void *ptr, size_t size, size_t nmemb, void *stream);
93 multifetchrequest *_request;
104 #define WORKER_STARTING 0
105 #define WORKER_LOOKUP 1
106 #define WORKER_FETCH 2
107 #define WORKER_DISCARD 3
108 #define WORKER_DONE 4
109 #define WORKER_SLEEP 5
110 #define WORKER_BROKEN 6
114 class multifetchrequest {
116 multifetchrequest(const MediaMultiCurl *context, const Pathname &filename, const Url &baseurl, CURLM *multi, FILE *fp, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize);
117 ~multifetchrequest();
119 void run(std::vector<Url> &urllist);
122 friend class multifetchworker;
124 const MediaMultiCurl *_context;
125 const Pathname _filename;
129 callback::SendReport<DownloadProgressReport> *_report;
130 MediaBlockList *_blklist;
135 std::list<multifetchworker *> _workers;
141 size_t _activeworkers;
142 size_t _lookupworkers;
143 size_t _sleepworkers;
144 double _minsleepuntil;
148 off_t _fetchedgoodsize;
151 double _lastprogress;
153 double _lastperiodstart;
154 double _lastperiodfetched;
159 double _connect_timeout;
164 #define BLKSIZE 131072
168 //////////////////////////////////////////////////////////////////////
174 if (gettimeofday(&tv, NULL))
176 return tv.tv_sec + tv.tv_usec / 1000000.;
180 multifetchworker::writefunction(void *ptr, size_t size)
183 if (_state == WORKER_BROKEN)
186 double now = currentTime();
188 len = size > _size ? _size : size;
195 if (_blkstart && _off == _blkstart)
197 // make sure that the server replied with "partial content"
200 (void)curl_easy_getinfo(_curl, CURLINFO_EFFECTIVE_URL, &effurl);
201 if (effurl && !strncasecmp(effurl, "http", 4))
204 (void)curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &statuscode);
205 if (statuscode != 206)
213 _request->_lastprogress = now;
215 if (_state == WORKER_DISCARD || !_request->_fp)
217 // block is no longer needed
218 // still calculate the checksum so that we can throw out bad servers
219 if (_request->_blklist)
220 _dig.update((const char *)ptr, len);
225 if (fseeko(_request->_fp, _off, SEEK_SET))
227 cnt = fwrite(ptr, 1, len, _request->_fp);
230 _request->_fetchedsize += cnt;
231 if (_request->_blklist)
232 _dig.update((const char *)ptr, cnt);
242 multifetchworker::_writefunction(void *ptr, size_t size, size_t nmemb, void *stream)
244 multifetchworker *me = reinterpret_cast<multifetchworker *>(stream);
245 return me->writefunction(ptr, size * nmemb);
249 multifetchworker::headerfunction(char *p, size_t size)
252 if (l > 9 && !strncasecmp(p, "Location:", 9))
254 string line(p + 9, l - 9);
255 if (line[l - 10] == '\r')
256 line.erase(l - 10, 1);
257 XXX << "#" << _workerno << ": redirecting to" << line << endl;
260 if (l <= 14 || l >= 128 || strncasecmp(p, "Content-Range:", 14) != 0)
264 while (l && (*p == ' ' || *p == '\t'))
266 if (l < 6 || strncasecmp(p, "bytes", 5))
273 unsigned long long start, off, filesize;
274 if (sscanf(buf, "%llu-%llu/%llu", &start, &off, &filesize) != 3)
276 if (_request->_filesize == (off_t)-1)
278 WAR << "#" << _workerno << ": setting request filesize to " << filesize << endl;
279 _request->_filesize = filesize;
280 if (_request->_totalsize == 0 && !_request->_blklist)
281 _request->_totalsize = filesize;
283 if (_request->_filesize != (off_t)filesize)
285 XXX << "#" << _workerno << ": filesize mismatch" << endl;
286 _state = WORKER_BROKEN;
287 strncpy(_curlError, "filesize mismatch", CURL_ERROR_SIZE);
293 multifetchworker::_headerfunction(void *ptr, size_t size, size_t nmemb, void *stream)
295 multifetchworker *me = reinterpret_cast<multifetchworker *>(stream);
296 return me->headerfunction((char *)ptr, size * nmemb);
299 multifetchworker::multifetchworker(int no, multifetchrequest &request, const Url &url)
300 : MediaCurl(url, Pathname())
304 _state = WORKER_STARTING;
306 _off = _blkstart = 0;
307 _size = _blksize = 0;
317 _maxspeed = _request->_maxspeed;
320 Url curlUrl( clearQueryString(url) );
321 _urlbuf = curlUrl.asString();
322 _curl = _request->_context->fromEasyPool(_url.getHost());
324 XXX << "reused worker from pool" << endl;
325 if (!_curl && !(_curl = curl_easy_init()))
327 _state = WORKER_BROKEN;
328 strncpy(_curlError, "curl_easy_init failed", CURL_ERROR_SIZE);
335 catch (Exception &ex)
337 curl_easy_cleanup(_curl);
339 _state = WORKER_BROKEN;
340 strncpy(_curlError, "curl_easy_setopt failed", CURL_ERROR_SIZE);
343 curl_easy_setopt(_curl, CURLOPT_PRIVATE, this);
344 curl_easy_setopt(_curl, CURLOPT_URL, _urlbuf.c_str());
345 curl_easy_setopt(_curl, CURLOPT_WRITEFUNCTION, &_writefunction);
346 curl_easy_setopt(_curl, CURLOPT_WRITEDATA, this);
347 if (_request->_filesize == off_t(-1) || !_request->_blklist || !_request->_blklist->haveChecksum(0))
349 curl_easy_setopt(_curl, CURLOPT_HEADERFUNCTION, &_headerfunction);
350 curl_easy_setopt(_curl, CURLOPT_HEADERDATA, this);
352 // if this is the same host copy authorization
353 // (the host check is also what curl does when doing a redirect)
354 // (note also that unauthorized exceptions are thrown with the request host)
355 if (url.getHost() == _request->_context->_url.getHost())
357 _settings.setUsername(_request->_context->_settings.username());
358 _settings.setPassword(_request->_context->_settings.password());
359 _settings.setAuthType(_request->_context->_settings.authType());
360 if ( _settings.userPassword().size() )
362 curl_easy_setopt(_curl, CURLOPT_USERPWD, _settings.userPassword().c_str());
363 string use_auth = _settings.authType();
364 if (use_auth.empty())
365 use_auth = "digest,basic"; // our default
366 long auth = CurlAuthData::auth_type_str2long(use_auth);
367 if( auth != CURLAUTH_NONE)
369 XXX << "#" << _workerno << ": Enabling HTTP authentication methods: " << use_auth
370 << " (CURLOPT_HTTPAUTH=" << auth << ")" << std::endl;
371 curl_easy_setopt(_curl, CURLOPT_HTTPAUTH, auth);
378 multifetchworker::~multifetchworker()
382 if (_state == WORKER_FETCH || _state == WORKER_DISCARD)
383 curl_multi_remove_handle(_request->_multi, _curl);
384 if (_state == WORKER_DONE || _state == WORKER_SLEEP)
386 #if CURLVERSION_AT_LEAST(7,15,5)
387 curl_easy_setopt(_curl, CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t)0);
389 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (void *)0);
390 curl_easy_setopt(_curl, CURLOPT_WRITEFUNCTION, (void *)0);
391 curl_easy_setopt(_curl, CURLOPT_WRITEDATA, (void *)0);
392 curl_easy_setopt(_curl, CURLOPT_HEADERFUNCTION, (void *)0);
393 curl_easy_setopt(_curl, CURLOPT_HEADERDATA, (void *)0);
394 _request->_context->toEasyPool(_url.getHost(), _curl);
397 curl_easy_cleanup(_curl);
404 while (waitpid(_pid, &status, 0) == -1)
414 // the destructor in MediaCurl doesn't call disconnect() if
415 // the media is not attached, so we do it here manually
419 static inline bool env_isset(string name)
421 const char *s = getenv(name.c_str());
422 return s && *s ? true : false;
426 multifetchworker::checkdns()
428 string host = _url.getHost();
433 if (_request->_context->isDNSok(host))
436 // no need to do dns checking for numeric hosts
438 if (inet_pton(AF_INET, host.c_str(), addrbuf) == 1)
440 if (inet_pton(AF_INET6, host.c_str(), addrbuf) == 1)
443 // no need to do dns checking if we use a proxy
444 if (!_settings.proxy().empty())
446 if (env_isset("all_proxy") || env_isset("ALL_PROXY"))
448 string schemeproxy = _url.getScheme() + "_proxy";
449 if (env_isset(schemeproxy))
451 if (schemeproxy != "http_proxy")
453 std::transform(schemeproxy.begin(), schemeproxy.end(), schemeproxy.begin(), ::toupper);
454 if (env_isset(schemeproxy))
458 XXX << "checking DNS lookup of " << host << endl;
462 _state = WORKER_BROKEN;
463 strncpy(_curlError, "DNS pipe creation failed", CURL_ERROR_SIZE);
467 if (_pid == pid_t(-1))
472 _state = WORKER_BROKEN;
473 strncpy(_curlError, "DNS checker fork failed", CURL_ERROR_SIZE);
479 // XXX: close all other file descriptors
480 struct addrinfo *ai, aihints;
481 memset(&aihints, 0, sizeof(aihints));
482 aihints.ai_family = PF_UNSPEC;
483 int tstsock = socket(PF_INET6, SOCK_DGRAM | SOCK_CLOEXEC, 0);
485 aihints.ai_family = PF_INET;
488 aihints.ai_socktype = SOCK_STREAM;
489 aihints.ai_flags = AI_CANONNAME;
490 unsigned int connecttimeout = _request->_connect_timeout;
492 alarm(connecttimeout);
493 signal(SIGALRM, SIG_DFL);
494 if (getaddrinfo(host.c_str(), NULL, &aihints, &ai))
499 _dnspipe = pipefds[0];
500 _state = WORKER_LOOKUP;
504 multifetchworker::adddnsfd(fd_set &rset, int &maxfd)
506 if (_state != WORKER_LOOKUP)
508 FD_SET(_dnspipe, &rset);
509 if (maxfd < _dnspipe)
514 multifetchworker::dnsevent(fd_set &rset)
517 if (_state != WORKER_LOOKUP || !FD_ISSET(_dnspipe, &rset))
520 while (waitpid(_pid, &status, 0) == -1)
531 if (!WIFEXITED(status))
533 _state = WORKER_BROKEN;
534 strncpy(_curlError, "DNS lookup failed", CURL_ERROR_SIZE);
535 _request->_activeworkers--;
538 int exitcode = WEXITSTATUS(status);
539 XXX << "#" << _workerno << ": DNS lookup returned " << exitcode << endl;
542 _state = WORKER_BROKEN;
543 strncpy(_curlError, "DNS lookup failed", CURL_ERROR_SIZE);
544 _request->_activeworkers--;
547 _request->_context->setDNSok(_url.getHost());
552 multifetchworker::checkChecksum()
554 // XXX << "checkChecksum block " << _blkno << endl;
555 if (!_blksize || !_request->_blklist)
557 return _request->_blklist->verifyDigest(_blkno, _dig);
561 multifetchworker::recheckChecksum()
563 // XXX << "recheckChecksum block " << _blkno << endl;
564 if (!_request->_fp || !_blksize || !_request->_blklist)
566 if (fseeko(_request->_fp, _blkstart, SEEK_SET))
570 _request->_blklist->createDigest(_dig); // resets digest
573 size_t cnt = l > sizeof(buf) ? sizeof(buf) : l;
574 if (fread(buf, cnt, 1, _request->_fp) != 1)
576 _dig.update(buf, cnt);
579 return _request->_blklist->verifyDigest(_blkno, _dig);
584 multifetchworker::stealjob()
586 if (!_request->_stealing)
588 XXX << "start stealing!" << endl;
589 _request->_stealing = true;
591 multifetchworker *best = 0;
592 std::list<multifetchworker *>::iterator workeriter = _request->_workers.begin();
594 for (; workeriter != _request->_workers.end(); ++workeriter)
596 multifetchworker *worker = *workeriter;
599 if (worker->_pass == -1)
600 continue; // do not steal!
601 if (worker->_state == WORKER_DISCARD || worker->_state == WORKER_DONE || worker->_state == WORKER_SLEEP || !worker->_blksize)
602 continue; // do not steal finished jobs
603 if (!worker->_avgspeed && worker->_blkreceived)
607 if (now > worker->_blkstarttime)
608 worker->_avgspeed = worker->_blkreceived / (now - worker->_blkstarttime);
610 if (!best || best->_pass > worker->_pass)
615 if (best->_pass < worker->_pass)
617 // if it is the same block, we want to know the best worker, otherwise the worst
618 if (worker->_blkstart == best->_blkstart)
620 if ((worker->_blksize - worker->_blkreceived) * best->_avgspeed < (best->_blksize - best->_blkreceived) * worker->_avgspeed)
625 if ((worker->_blksize - worker->_blkreceived) * best->_avgspeed > (best->_blksize - best->_blkreceived) * worker->_avgspeed)
631 _state = WORKER_DONE;
632 _request->_activeworkers--;
633 _request->_finished = true;
636 // do not sleep twice
637 if (_state != WORKER_SLEEP)
639 if (!_avgspeed && _blkreceived)
643 if (now > _blkstarttime)
644 _avgspeed = _blkreceived / (now - _blkstarttime);
647 // lets see if we should sleep a bit
648 XXX << "me #" << _workerno << ": " << _avgspeed << ", size " << best->_blksize << endl;
649 XXX << "best #" << best->_workerno << ": " << best->_avgspeed << ", size " << (best->_blksize - best->_blkreceived) << endl;
650 if (_avgspeed && best->_avgspeed && best->_blksize - best->_blkreceived > 0 &&
651 (best->_blksize - best->_blkreceived) * _avgspeed < best->_blksize * best->_avgspeed)
655 double sl = (best->_blksize - best->_blkreceived) / best->_avgspeed * 2;
658 XXX << "#" << _workerno << ": going to sleep for " << sl * 1000 << " ms" << endl;
659 _sleepuntil = now + sl;
660 _state = WORKER_SLEEP;
661 _request->_sleepworkers++;
667 best->_competing = true;
668 _blkstart = best->_blkstart;
669 _blksize = best->_blksize;
672 _blkno = best->_blkno;
677 multifetchworker::disableCompetition()
679 std::list<multifetchworker *>::iterator workeriter = _request->_workers.begin();
680 for (; workeriter != _request->_workers.end(); ++workeriter)
682 multifetchworker *worker = *workeriter;
685 if (worker->_blkstart == _blkstart)
687 if (worker->_state == WORKER_FETCH)
688 worker->_state = WORKER_DISCARD;
689 worker->_pass = -1; /* do not steal this one, we already have it */
696 multifetchworker::nextjob()
699 if (_request->_stealing)
705 MediaBlockList *blklist = _request->_blklist;
709 if (_request->_filesize != off_t(-1))
711 if (_request->_blkoff >= _request->_filesize)
716 _blksize = _request->_filesize - _request->_blkoff;
717 if (_blksize > BLKSIZE)
723 MediaBlock blk = blklist->getBlock(_request->_blkno);
724 while (_request->_blkoff >= (off_t)(blk.off + blk.size))
726 if (++_request->_blkno == blklist->numBlocks())
731 blk = blklist->getBlock(_request->_blkno);
732 _request->_blkoff = blk.off;
734 _blksize = blk.off + blk.size - _request->_blkoff;
735 if (_blksize > BLKSIZE && !blklist->haveChecksum(_request->_blkno))
738 _blkno = _request->_blkno;
739 _blkstart = _request->_blkoff;
740 _request->_blkoff += _blksize;
745 multifetchworker::run()
749 if (_state == WORKER_BROKEN || _state == WORKER_DONE)
750 return; // just in case...
752 sprintf(rangebuf, "%llu-", (unsigned long long)_blkstart);
754 sprintf(rangebuf, "%llu-%llu", (unsigned long long)_blkstart, (unsigned long long)_blkstart + _blksize - 1);
755 XXX << "#" << _workerno << ": BLK " << _blkno << ":" << rangebuf << " " << _url << endl;
756 if (curl_easy_setopt(_curl, CURLOPT_RANGE, !_noendrange || _blkstart != 0 ? rangebuf : (char *)0) != CURLE_OK)
758 _request->_activeworkers--;
759 _state = WORKER_BROKEN;
760 strncpy(_curlError, "curl_easy_setopt range failed", CURL_ERROR_SIZE);
763 if (curl_multi_add_handle(_request->_multi, _curl) != CURLM_OK)
765 _request->_activeworkers--;
766 _state = WORKER_BROKEN;
767 strncpy(_curlError, "curl_multi_add_handle failed", CURL_ERROR_SIZE);
770 _request->_havenewjob = true;
773 if (_request->_blklist)
774 _request->_blklist->createDigest(_dig); // resets digest
775 _state = WORKER_FETCH;
777 double now = currentTime();
783 //////////////////////////////////////////////////////////////////////
786 multifetchrequest::multifetchrequest(const MediaMultiCurl *context, const Pathname &filename, const Url &baseurl, CURLM *multi, FILE *fp, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize) : _context(context), _filename(filename), _baseurl(baseurl)
791 _filesize = filesize;
797 _blkoff = _blklist->getBlock(0).off;
806 _fetchedgoodsize = 0;
808 _lastperiodstart = _lastprogress = _starttime = currentTime();
809 _lastperiodfetched = 0;
812 _connect_timeout = 0;
817 for (size_t blkno = 0; blkno < blklist->numBlocks(); blkno++)
819 MediaBlock blk = blklist->getBlock(blkno);
820 _totalsize += blk.size;
823 else if (filesize != off_t(-1))
824 _totalsize = filesize;
827 multifetchrequest::~multifetchrequest()
829 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
831 multifetchworker *worker = *workeriter;
839 multifetchrequest::run(std::vector<Url> &urllist)
842 std::vector<Url>::iterator urliter = urllist.begin();
845 fd_set rset, wset, xset;
850 XXX << "finished!" << endl;
854 if ((int)_activeworkers < _maxworkers && urliter != urllist.end() && _workers.size() < MAXURLS)
856 // spawn another worker!
857 multifetchworker *worker = new multifetchworker(workerno++, *this, *urliter);
858 _workers.push_back(worker);
859 if (worker->_state != WORKER_BROKEN)
862 if (worker->_state != WORKER_LOOKUP)
874 WAR << "No more active workers!" << endl;
875 // show the first worker error we find
876 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
878 if ((*workeriter)->_state != WORKER_BROKEN)
880 ZYPP_THROW(MediaCurlException(_baseurl, "Server error", (*workeriter)->_curlError));
889 curl_multi_fdset(_multi, &rset, &wset, &xset, &maxfd);
892 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
893 (*workeriter)->adddnsfd(rset, maxfd);
896 // if we added a new job we have to call multi_perform once
897 // to make it show up in the fd set. do not sleep in this case.
899 tv.tv_usec = _havenewjob ? 0 : 200000;
900 if (_sleepworkers && !_havenewjob)
902 if (_minsleepuntil == 0)
904 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
906 multifetchworker *worker = *workeriter;
907 if (worker->_state != WORKER_SLEEP)
909 if (!_minsleepuntil || _minsleepuntil > worker->_sleepuntil)
910 _minsleepuntil = worker->_sleepuntil;
913 double sl = _minsleepuntil - currentTime();
920 tv.tv_usec = sl * 1000000;
922 int r = select(maxfd + 1, &rset, &wset, &xset, &tv);
923 if (r == -1 && errno != EINTR)
924 ZYPP_THROW(MediaCurlException(_baseurl, "select() failed", "unknown error"));
925 if (r != 0 && _lookupworkers)
926 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
928 multifetchworker *worker = *workeriter;
929 if (worker->_state != WORKER_LOOKUP)
931 (*workeriter)->dnsevent(rset);
932 if (worker->_state != WORKER_LOOKUP)
942 mcode = curl_multi_perform(_multi, &tasks);
943 if (mcode == CURLM_CALL_MULTI_PERFORM)
945 if (mcode != CURLM_OK)
946 ZYPP_THROW(MediaCurlException(_baseurl, "curl_multi_perform", "unknown error"));
950 double now = currentTime();
953 if (now > _lastperiodstart + .5)
956 _periodavg = (_fetchedsize - _lastperiodfetched) / (now - _lastperiodstart);
958 _periodavg = (_periodavg + (_fetchedsize - _lastperiodfetched) / (now - _lastperiodstart)) / 2;
959 _lastperiodfetched = _fetchedsize;
960 _lastperiodstart = now;
966 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
968 multifetchworker *worker = *workeriter;
969 if (worker->_state != WORKER_SLEEP)
971 if (worker->_sleepuntil > now)
973 if (_minsleepuntil == worker->_sleepuntil)
975 XXX << "#" << worker->_workerno << ": sleep done, wake up" << endl;
977 // nextjob chnages the state
982 // collect all curl results, reschedule new jobs
984 while ((msg = curl_multi_info_read(_multi, &nqueue)) != 0)
986 if (msg->msg != CURLMSG_DONE)
988 CURL *easy = msg->easy_handle;
989 CURLcode cc = msg->data.result;
990 multifetchworker *worker;
991 if (curl_easy_getinfo(easy, CURLINFO_PRIVATE, &worker) != CURLE_OK)
992 ZYPP_THROW(MediaCurlException(_baseurl, "curl_easy_getinfo", "unknown error"));
993 if (worker->_blkreceived && now > worker->_blkstarttime)
995 if (worker->_avgspeed)
996 worker->_avgspeed = (worker->_avgspeed + worker->_blkreceived / (now - worker->_blkstarttime)) / 2;
998 worker->_avgspeed = worker->_blkreceived / (now - worker->_blkstarttime);
1000 XXX << "#" << worker->_workerno << ": BLK " << worker->_blkno << " done code " << cc << " speed " << worker->_avgspeed << endl;
1001 curl_multi_remove_handle(_multi, easy);
1002 if (cc == CURLE_HTTP_RETURNED_ERROR)
1004 long statuscode = 0;
1005 (void)curl_easy_getinfo(easy, CURLINFO_RESPONSE_CODE, &statuscode);
1006 XXX << "HTTP status " << statuscode << endl;
1007 if (statuscode == 416 && !_blklist) /* Range error */
1009 if (_filesize == off_t(-1))
1011 if (!worker->_noendrange)
1013 XXX << "#" << worker->_workerno << ": retrying with no end range" << endl;
1014 worker->_noendrange = true;
1018 worker->_noendrange = false;
1022 if (worker->_blkstart >= _filesize)
1031 if (!worker->checkChecksum())
1033 WAR << "#" << worker->_workerno << ": checksum error, disable worker" << endl;
1034 worker->_state = WORKER_BROKEN;
1035 strncpy(worker->_curlError, "checksum error", CURL_ERROR_SIZE);
1039 if (worker->_state == WORKER_FETCH)
1041 if (worker->_competing)
1043 worker->disableCompetition();
1044 // multiple workers wrote into this block. We already know that our
1045 // data was correct, but maybe some other worker overwrote our data
1046 // with something broken. Thus we have to re-check the block.
1047 if (!worker->recheckChecksum())
1049 XXX << "#" << worker->_workerno << ": recheck checksum error, refetch block" << endl;
1050 // re-fetch! No need to worry about the bad workers,
1051 // they will now be set to DISCARD. At the end of their block
1052 // they will notice that they wrote bad data and go into BROKEN.
1057 _fetchedgoodsize += worker->_blksize;
1060 // make bad workers sleep a little
1062 int maxworkerno = 0;
1064 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
1066 multifetchworker *oworker = *workeriter;
1067 if (oworker->_state == WORKER_BROKEN)
1069 if (oworker->_avgspeed > maxavg)
1071 maxavg = oworker->_avgspeed;
1072 maxworkerno = oworker->_workerno;
1074 if (oworker->_avgspeed > worker->_avgspeed)
1077 if (maxavg && !_stealing)
1079 double ratio = worker->_avgspeed / maxavg;
1081 if (numbetter < 3) // don't sleep that much if we're in the top two
1082 ratio = ratio * ratio;
1085 XXX << "#" << worker->_workerno << ": too slow ("<< ratio << ", " << worker->_avgspeed << ", #" << maxworkerno << ": " << maxavg << "), going to sleep for " << ratio * 1000 << " ms" << endl;
1086 worker->_sleepuntil = now + ratio;
1087 worker->_state = WORKER_SLEEP;
1093 // do rate control (if requested)
1094 // should use periodavg, but that's not what libcurl does
1095 if (_maxspeed && now > _starttime)
1097 double avg = _fetchedsize / (now - _starttime);
1098 avg = worker->_maxspeed * _maxspeed / avg;
1099 if (avg < _maxspeed / _maxworkers)
1100 avg = _maxspeed / _maxworkers;
1101 if (avg > _maxspeed)
1105 worker->_maxspeed = avg;
1106 #if CURLVERSION_AT_LEAST(7,15,5)
1107 curl_easy_setopt(worker->_curl, CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t)(avg));
1115 worker->_state = WORKER_BROKEN;
1117 if (!_activeworkers && !(urliter != urllist.end() && _workers.size() < MAXURLS))
1119 // end of workers reached! goodbye!
1120 worker->evaluateCurlCode(Pathname(), cc, false);
1124 if ( _filesize > 0 && _fetchedgoodsize > _filesize ) {
1125 ZYPP_THROW(MediaFileSizeExceededException(_baseurl, _filesize));
1132 int percent = _totalsize ? (100 * (_fetchedgoodsize + _fetchedsize)) / (_totalsize + _fetchedsize) : 0;
1135 if (now > _starttime)
1136 avg = _fetchedsize / (now - _starttime);
1137 if (!(*(_report))->progress(percent, _baseurl, avg, _lastperiodstart == _starttime ? avg : _periodavg))
1138 ZYPP_THROW(MediaCurlException(_baseurl, "User abort", "cancelled"));
1141 if (_timeout && now - _lastprogress > _timeout)
1146 ZYPP_THROW(MediaTimeoutException(_baseurl));
1148 // print some download stats
1149 WAR << "overall result" << endl;
1150 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
1152 multifetchworker *worker = *workeriter;
1153 WAR << "#" << worker->_workerno << ": state: " << worker->_state << " received: " << worker->_received << " url: " << worker->_url << endl;
1158 //////////////////////////////////////////////////////////////////////
1161 MediaMultiCurl::MediaMultiCurl(const Url &url_r, const Pathname & attach_point_hint_r)
1162 : MediaCurl(url_r, attach_point_hint_r)
1164 MIL << "MediaMultiCurl::MediaMultiCurl(" << url_r << ", " << attach_point_hint_r << ")" << endl;
1166 _customHeadersMetalink = 0;
1169 MediaMultiCurl::~MediaMultiCurl()
1171 if (_customHeadersMetalink)
1173 curl_slist_free_all(_customHeadersMetalink);
1174 _customHeadersMetalink = 0;
1178 curl_multi_cleanup(_multi);
1181 std::map<std::string, CURL *>::iterator it;
1182 for (it = _easypool.begin(); it != _easypool.end(); it++)
1184 CURL *easy = it->second;
1187 curl_easy_cleanup(easy);
1193 void MediaMultiCurl::setupEasy()
1195 MediaCurl::setupEasy();
1197 if (_customHeadersMetalink)
1199 curl_slist_free_all(_customHeadersMetalink);
1200 _customHeadersMetalink = 0;
1202 struct curl_slist *sl = _customHeaders;
1203 for (; sl; sl = sl->next)
1204 _customHeadersMetalink = curl_slist_append(_customHeadersMetalink, sl->data);
1205 _customHeadersMetalink = curl_slist_append(_customHeadersMetalink, "Accept: */*, application/metalink+xml, application/metalink4+xml");
1208 static bool looks_like_metalink_fd(int fd)
1212 while ((l = pread(fd, buf, sizeof(buf) - 1, (off_t)0)) == -1 && errno == EINTR)
1218 while (*p == ' ' || *p == '\t' || *p == '\r' || *p == '\n')
1220 if (!strncasecmp(p, "<?xml", 5))
1222 while (*p && *p != '>')
1226 while (*p == ' ' || *p == '\t' || *p == '\r' || *p == '\n')
1229 bool ret = !strncasecmp(p, "<metalink", 9) ? true : false;
1233 static bool looks_like_metalink(const Pathname & file)
1236 if ((fd = open(file.asString().c_str(), O_RDONLY|O_CLOEXEC)) == -1)
1238 bool ret = looks_like_metalink_fd(fd);
1240 DBG << "looks_like_metalink(" << file << "): " << ret << endl;
1244 // here we try to suppress all progress coming from a metalink download
1245 // bsc#1021291: Nevertheless send alive trigger (without stats), so UIs
1246 // are able to abort a hanging metalink download via callback response.
1247 int MediaMultiCurl::progressCallback( void *clientp, double dltotal, double dlnow, double ultotal, double ulnow)
1249 CURL *_curl = MediaCurl::progressCallback_getcurl(clientp);
1251 return MediaCurl::aliveCallback(clientp, dltotal, dlnow, ultotal, ulnow);
1253 // bsc#408814: Don't report any sizes before we don't have data on disk. Data reported
1254 // due to redirection etc. are not interesting, but may disturb filesize checks.
1256 if ( curl_easy_getinfo( _curl, CURLINFO_PRIVATE, &fp ) != CURLE_OK || !fp )
1257 return MediaCurl::aliveCallback( clientp, dltotal, dlnow, ultotal, ulnow );
1258 if ( ftell( fp ) == 0 )
1259 return MediaCurl::aliveCallback( clientp, dltotal, 0.0, ultotal, ulnow );
1261 // (no longer needed due to the filesize check above?)
1262 // work around curl bug that gives us old data
1263 long httpReturnCode = 0;
1264 if (curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &httpReturnCode ) != CURLE_OK || httpReturnCode == 0)
1265 return MediaCurl::aliveCallback(clientp, dltotal, dlnow, ultotal, ulnow);
1268 bool ismetalink = false;
1269 if (curl_easy_getinfo(_curl, CURLINFO_CONTENT_TYPE, &ptr) == CURLE_OK && ptr)
1271 string ct = string(ptr);
1272 if (ct.find("application/metalink+xml") == 0 || ct.find("application/metalink4+xml") == 0)
1275 if (!ismetalink && dlnow < 256)
1277 // can't tell yet, ...
1278 return MediaCurl::aliveCallback(clientp, dltotal, dlnow, ultotal, ulnow);
1283 ismetalink = looks_like_metalink_fd(fileno(fp));
1284 DBG << "looks_like_metalink_fd: " << ismetalink << endl;
1288 // this is a metalink file change the expected filesize
1289 MediaCurl::resetExpectedFileSize( clientp, ByteCount( 2, ByteCount::MB) );
1290 // we're downloading the metalink file. Just trigger aliveCallbacks
1291 curl_easy_setopt(_curl, CURLOPT_PROGRESSFUNCTION, &MediaCurl::aliveCallback);
1292 return MediaCurl::aliveCallback(clientp, dltotal, dlnow, ultotal, ulnow);
1294 curl_easy_setopt(_curl, CURLOPT_PROGRESSFUNCTION, &MediaCurl::progressCallback);
1295 return MediaCurl::progressCallback(clientp, dltotal, dlnow, ultotal, ulnow);
1298 void MediaMultiCurl::doGetFileCopy( const Pathname & filename , const Pathname & target, callback::SendReport<DownloadProgressReport> & report, const ByteCount &expectedFileSize_r, RequestOptions options ) const
1300 Pathname dest = target.absolutename();
1301 if( assert_dir( dest.dirname() ) )
1303 DBG << "assert_dir " << dest.dirname() << " failed" << endl;
1304 ZYPP_THROW( MediaSystemException(getFileUrl(filename), "System error on " + dest.dirname().asString()) );
1307 ManagedFile destNew { target.extend( ".new.zypp.XXXXXX" ) };
1310 AutoFREE<char> buf { ::strdup( (*destNew).c_str() ) };
1313 ERR << "out of memory for temp file name" << endl;
1314 ZYPP_THROW(MediaSystemException(getFileUrl(filename), "out of memory for temp file name"));
1317 AutoFD tmp_fd { ::mkostemp( buf, O_CLOEXEC ) };
1320 ERR << "mkstemp failed for file '" << destNew << "'" << endl;
1321 ZYPP_THROW(MediaWriteException(destNew));
1323 destNew = ManagedFile( (*buf), filesystem::unlink );
1325 file = ::fdopen( tmp_fd, "we" );
1328 ERR << "fopen failed for file '" << destNew << "'" << endl;
1329 ZYPP_THROW(MediaWriteException(destNew));
1331 tmp_fd.resetDispose(); // don't close it here! ::fdopen moved ownership to file
1334 DBG << "dest: " << dest << endl;
1335 DBG << "temp: " << destNew << endl;
1337 // set IFMODSINCE time condition (no download if not modified)
1338 if( PathInfo(target).isExist() && !(options & OPTION_NO_IFMODSINCE) )
1340 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_IFMODSINCE);
1341 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, (long)PathInfo(target).mtime());
1345 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1346 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1348 // change header to include Accept: metalink
1349 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeadersMetalink);
1350 // change to our own progress funcion
1351 curl_easy_setopt(_curl, CURLOPT_PROGRESSFUNCTION, &progressCallback);
1352 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (*file) ); // important to pass the FILE* explicitly (passing through varargs)
1355 MediaCurl::doGetFileCopyFile(filename, dest, file, report, expectedFileSize_r, options);
1357 catch (Exception &ex)
1359 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1360 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1361 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeaders);
1362 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (void *)0);
1365 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1366 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1367 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeaders);
1368 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (void *)0);
1369 long httpReturnCode = 0;
1370 CURLcode infoRet = curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &httpReturnCode);
1371 if (infoRet == CURLE_OK)
1373 DBG << "HTTP response: " + str::numstring(httpReturnCode) << endl;
1374 if ( httpReturnCode == 304
1375 || ( httpReturnCode == 213 && _url.getScheme() == "ftp" ) ) // not modified
1377 DBG << "not modified: " << PathInfo(dest) << endl;
1383 WAR << "Could not get the reponse code." << endl;
1386 bool ismetalink = false;
1389 if (curl_easy_getinfo(_curl, CURLINFO_CONTENT_TYPE, &ptr) == CURLE_OK && ptr)
1391 string ct = string(ptr);
1392 if (ct.find("application/metalink+xml") == 0 || ct.find("application/metalink4+xml") == 0)
1398 // some proxies do not store the content type, so also look at the file to find
1399 // out if we received a metalink (bnc#649925)
1401 if (looks_like_metalink(destNew))
1407 bool userabort = false;
1408 Pathname failedFile = ZConfig::instance().repoCachePath() / "MultiCurl.failed";
1409 file = nullptr; // explicitly close destNew before the parser reads it.
1414 MediaBlockList bl = mlp.getBlockList();
1415 vector<Url> urls = mlp.getUrls();
1417 file = fopen((*destNew).c_str(), "w+e");
1419 ZYPP_THROW(MediaWriteException(destNew));
1420 if (PathInfo(target).isExist())
1422 XXX << "reusing blocks from file " << target << endl;
1423 bl.reuseBlocks(file, target.asString());
1426 if (bl.haveChecksum(1) && PathInfo(failedFile).isExist())
1428 XXX << "reusing blocks from file " << failedFile << endl;
1429 bl.reuseBlocks(file, failedFile.asString());
1431 filesystem::unlink(failedFile);
1433 Pathname df = deltafile();
1436 XXX << "reusing blocks from file " << df << endl;
1437 bl.reuseBlocks(file, df.asString());
1442 multifetch(filename, file, &urls, &report, &bl, expectedFileSize_r);
1444 catch (MediaCurlException &ex)
1446 userabort = ex.errstr() == "User abort";
1450 catch (MediaFileSizeExceededException &ex) {
1453 catch (Exception &ex)
1455 // something went wrong. fall back to normal download
1456 file = nullptr; // explicitly close destNew before moving it
1457 if (PathInfo(destNew).size() >= 63336)
1459 ::unlink(failedFile.asString().c_str());
1460 filesystem::hardlinkCopy(destNew, failedFile);
1466 file = fopen((*destNew).c_str(), "w+e");
1468 ZYPP_THROW(MediaWriteException(destNew));
1469 MediaCurl::doGetFileCopyFile(filename, dest, file, report, expectedFileSize_r, options | OPTION_NO_REPORT_START);
1473 if (::fchmod( ::fileno(file), filesystem::applyUmaskTo( 0644 )))
1475 ERR << "Failed to chmod file " << destNew << endl;
1478 file.resetDispose(); // we're going to close it manually here
1481 filesystem::unlink(destNew);
1482 ERR << "Fclose failed for file '" << destNew << "'" << endl;
1483 ZYPP_THROW(MediaWriteException(destNew));
1486 if ( rename( destNew, dest ) != 0 )
1488 ERR << "Rename failed" << endl;
1489 ZYPP_THROW(MediaWriteException(dest));
1491 destNew.resetDispose(); // no more need to unlink it
1493 DBG << "done: " << PathInfo(dest) << endl;
1496 void MediaMultiCurl::multifetch(const Pathname & filename, FILE *fp, std::vector<Url> *urllist, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize) const
1498 Url baseurl(getFileUrl(filename));
1499 if (blklist && filesize == off_t(-1) && blklist->haveFilesize())
1500 filesize = blklist->getFilesize();
1501 if (blklist && !blklist->haveBlocks() && filesize != 0)
1503 if (blklist && (filesize == 0 || !blklist->numBlocks()))
1505 checkFileDigest(baseurl, fp, blklist);
1512 _multi = curl_multi_init();
1514 ZYPP_THROW(MediaCurlInitException(baseurl));
1517 multifetchrequest req(this, filename, baseurl, _multi, fp, report, blklist, filesize);
1518 req._timeout = _settings.timeout();
1519 req._connect_timeout = _settings.connectTimeout();
1520 req._maxspeed = _settings.maxDownloadSpeed();
1521 req._maxworkers = _settings.maxConcurrentConnections();
1522 if (req._maxworkers > MAXURLS)
1523 req._maxworkers = MAXURLS;
1524 if (req._maxworkers <= 0)
1525 req._maxworkers = 1;
1526 std::vector<Url> myurllist;
1527 for (std::vector<Url>::iterator urliter = urllist->begin(); urliter != urllist->end(); ++urliter)
1531 string scheme = urliter->getScheme();
1532 if (scheme == "http" || scheme == "https" || scheme == "ftp" || scheme == "tftp")
1534 checkProtocol(*urliter);
1535 myurllist.push_back(*urliter);
1542 if (!myurllist.size())
1543 myurllist.push_back(baseurl);
1545 checkFileDigest(baseurl, fp, blklist);
1548 void MediaMultiCurl::checkFileDigest(Url &url, FILE *fp, MediaBlockList *blklist) const
1550 if (!blklist || !blklist->haveFileChecksum())
1552 if (fseeko(fp, off_t(0), SEEK_SET))
1553 ZYPP_THROW(MediaCurlException(url, "fseeko", "seek error"));
1555 blklist->createFileDigest(dig);
1558 while ((l = fread(buf, 1, sizeof(buf), fp)) > 0)
1560 if (!blklist->verifyFileDigest(dig))
1561 ZYPP_THROW(MediaCurlException(url, "file verification failed", "checksum error"));
1564 bool MediaMultiCurl::isDNSok(const string &host) const
1566 return _dnsok.find(host) == _dnsok.end() ? false : true;
1569 void MediaMultiCurl::setDNSok(const string &host) const
1571 _dnsok.insert(host);
1574 CURL *MediaMultiCurl::fromEasyPool(const string &host) const
1576 if (_easypool.find(host) == _easypool.end())
1578 CURL *ret = _easypool[host];
1579 _easypool.erase(host);
1583 void MediaMultiCurl::toEasyPool(const std::string &host, CURL *easy) const
1585 CURL *oldeasy = _easypool[host];
1586 _easypool[host] = easy;
1588 curl_easy_cleanup(oldeasy);
1591 } // namespace media