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"
31 using namespace zypp::base;
33 #undef CURLVERSION_AT_LEAST
34 #define CURLVERSION_AT_LEAST(M,N,O) LIBCURL_VERSION_NUM >= ((((M)<<8)+(N))<<8)+(O)
40 //////////////////////////////////////////////////////////////////////
43 class multifetchrequest;
45 // Hack: we derive from MediaCurl just to get the storage space for
46 // settings, url, curlerrors and the like
48 class multifetchworker : MediaCurl {
49 friend class multifetchrequest;
52 multifetchworker(int no, multifetchrequest &request, const Url &url);
57 bool recheckChecksum();
58 void disableCompetition();
61 void adddnsfd(fd_set &rset, int &maxfd);
62 void dnsevent(fd_set &rset);
86 size_t writefunction(void *ptr, size_t size);
87 static size_t _writefunction(void *ptr, size_t size, size_t nmemb, void *stream);
89 size_t headerfunction(char *ptr, size_t size);
90 static size_t _headerfunction(void *ptr, size_t size, size_t nmemb, void *stream);
92 multifetchrequest *_request;
103 #define WORKER_STARTING 0
104 #define WORKER_LOOKUP 1
105 #define WORKER_FETCH 2
106 #define WORKER_DISCARD 3
107 #define WORKER_DONE 4
108 #define WORKER_SLEEP 5
109 #define WORKER_BROKEN 6
113 class multifetchrequest {
115 multifetchrequest(const MediaMultiCurl *context, const Pathname &filename, const Url &baseurl, CURLM *multi, FILE *fp, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize);
116 ~multifetchrequest();
118 void run(std::vector<Url> &urllist);
121 friend class multifetchworker;
123 const MediaMultiCurl *_context;
124 const Pathname _filename;
128 callback::SendReport<DownloadProgressReport> *_report;
129 MediaBlockList *_blklist;
134 std::list<multifetchworker *> _workers;
140 size_t _activeworkers;
141 size_t _lookupworkers;
142 size_t _sleepworkers;
143 double _minsleepuntil;
147 off_t _fetchedgoodsize;
150 double _lastprogress;
152 double _lastperiodstart;
153 double _lastperiodfetched;
158 double _connect_timeout;
163 #define BLKSIZE 131072
167 //////////////////////////////////////////////////////////////////////
173 if (gettimeofday(&tv, NULL))
175 return tv.tv_sec + tv.tv_usec / 1000000.;
179 multifetchworker::writefunction(void *ptr, size_t size)
182 if (_state == WORKER_BROKEN)
185 double now = currentTime();
187 len = size > _size ? _size : size;
194 if (_blkstart && _off == _blkstart)
196 // make sure that the server replied with "partial content"
199 (void)curl_easy_getinfo(_curl, CURLINFO_EFFECTIVE_URL, &effurl);
200 if (effurl && !strncasecmp(effurl, "http", 4))
203 (void)curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &statuscode);
204 if (statuscode != 206)
212 _request->_lastprogress = now;
214 if (_state == WORKER_DISCARD || !_request->_fp)
216 // block is no longer needed
217 // still calculate the checksum so that we can throw out bad servers
218 if (_request->_blklist)
219 _dig.update((const char *)ptr, len);
224 if (fseeko(_request->_fp, _off, SEEK_SET))
226 cnt = fwrite(ptr, 1, len, _request->_fp);
229 _request->_fetchedsize += cnt;
230 if (_request->_blklist)
231 _dig.update((const char *)ptr, cnt);
241 multifetchworker::_writefunction(void *ptr, size_t size, size_t nmemb, void *stream)
243 multifetchworker *me = reinterpret_cast<multifetchworker *>(stream);
244 return me->writefunction(ptr, size * nmemb);
248 multifetchworker::headerfunction(char *p, size_t size)
251 if (l > 9 && !strncasecmp(p, "Location:", 9))
253 string line(p + 9, l - 9);
254 if (line[l - 10] == '\r')
255 line.erase(l - 10, 1);
256 XXX << "#" << _workerno << ": redirecting to" << line << endl;
259 if (l <= 14 || l >= 128 || strncasecmp(p, "Content-Range:", 14) != 0)
263 while (l && (*p == ' ' || *p == '\t'))
265 if (l < 6 || strncasecmp(p, "bytes", 5))
272 unsigned long long start, off, filesize;
273 if (sscanf(buf, "%llu-%llu/%llu", &start, &off, &filesize) != 3)
275 if (_request->_filesize == (off_t)-1)
277 WAR << "#" << _workerno << ": setting request filesize to " << filesize << endl;
278 _request->_filesize = filesize;
279 if (_request->_totalsize == 0 && !_request->_blklist)
280 _request->_totalsize = filesize;
282 if (_request->_filesize != (off_t)filesize)
284 XXX << "#" << _workerno << ": filesize mismatch" << endl;
285 _state = WORKER_BROKEN;
286 strncpy(_curlError, "filesize mismatch", CURL_ERROR_SIZE);
292 multifetchworker::_headerfunction(void *ptr, size_t size, size_t nmemb, void *stream)
294 multifetchworker *me = reinterpret_cast<multifetchworker *>(stream);
295 return me->headerfunction((char *)ptr, size * nmemb);
298 multifetchworker::multifetchworker(int no, multifetchrequest &request, const Url &url)
299 : MediaCurl(url, Pathname())
303 _state = WORKER_STARTING;
305 _off = _blkstart = 0;
306 _size = _blksize = 0;
316 _maxspeed = _request->_maxspeed;
319 Url curlUrl( clearQueryString(url) );
320 _urlbuf = curlUrl.asString();
321 _curl = _request->_context->fromEasyPool(_url.getHost());
323 XXX << "reused worker from pool" << endl;
324 if (!_curl && !(_curl = curl_easy_init()))
326 _state = WORKER_BROKEN;
327 strncpy(_curlError, "curl_easy_init failed", CURL_ERROR_SIZE);
334 catch (Exception &ex)
336 curl_easy_cleanup(_curl);
338 _state = WORKER_BROKEN;
339 strncpy(_curlError, "curl_easy_setopt failed", CURL_ERROR_SIZE);
342 curl_easy_setopt(_curl, CURLOPT_PRIVATE, this);
343 curl_easy_setopt(_curl, CURLOPT_URL, _urlbuf.c_str());
344 curl_easy_setopt(_curl, CURLOPT_WRITEFUNCTION, &_writefunction);
345 curl_easy_setopt(_curl, CURLOPT_WRITEDATA, this);
346 if (_request->_filesize == off_t(-1) || !_request->_blklist || !_request->_blklist->haveChecksum(0))
348 curl_easy_setopt(_curl, CURLOPT_HEADERFUNCTION, &_headerfunction);
349 curl_easy_setopt(_curl, CURLOPT_HEADERDATA, this);
351 // if this is the same host copy authorization
352 // (the host check is also what curl does when doing a redirect)
353 // (note also that unauthorized exceptions are thrown with the request host)
354 if (url.getHost() == _request->_context->_url.getHost())
356 _settings.setUsername(_request->_context->_settings.username());
357 _settings.setPassword(_request->_context->_settings.password());
358 _settings.setAuthType(_request->_context->_settings.authType());
359 if ( _settings.userPassword().size() )
361 curl_easy_setopt(_curl, CURLOPT_USERPWD, _settings.userPassword().c_str());
362 string use_auth = _settings.authType();
363 if (use_auth.empty())
364 use_auth = "digest,basic"; // our default
365 long auth = CurlAuthData::auth_type_str2long(use_auth);
366 if( auth != CURLAUTH_NONE)
368 XXX << "#" << _workerno << ": Enabling HTTP authentication methods: " << use_auth
369 << " (CURLOPT_HTTPAUTH=" << auth << ")" << std::endl;
370 curl_easy_setopt(_curl, CURLOPT_HTTPAUTH, auth);
377 multifetchworker::~multifetchworker()
381 if (_state == WORKER_FETCH || _state == WORKER_DISCARD)
382 curl_multi_remove_handle(_request->_multi, _curl);
383 if (_state == WORKER_DONE || _state == WORKER_SLEEP)
385 #if CURLVERSION_AT_LEAST(7,15,5)
386 curl_easy_setopt(_curl, CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t)0);
388 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (void *)0);
389 curl_easy_setopt(_curl, CURLOPT_WRITEFUNCTION, (void *)0);
390 curl_easy_setopt(_curl, CURLOPT_WRITEDATA, (void *)0);
391 curl_easy_setopt(_curl, CURLOPT_HEADERFUNCTION, (void *)0);
392 curl_easy_setopt(_curl, CURLOPT_HEADERDATA, (void *)0);
393 _request->_context->toEasyPool(_url.getHost(), _curl);
396 curl_easy_cleanup(_curl);
403 while (waitpid(_pid, &status, 0) == -1)
413 // the destructor in MediaCurl doesn't call disconnect() if
414 // the media is not attached, so we do it here manually
418 static inline bool env_isset(string name)
420 const char *s = getenv(name.c_str());
421 return s && *s ? true : false;
425 multifetchworker::checkdns()
427 string host = _url.getHost();
432 if (_request->_context->isDNSok(host))
435 // no need to do dns checking for numeric hosts
437 if (inet_pton(AF_INET, host.c_str(), addrbuf) == 1)
439 if (inet_pton(AF_INET6, host.c_str(), addrbuf) == 1)
442 // no need to do dns checking if we use a proxy
443 if (!_settings.proxy().empty())
445 if (env_isset("all_proxy") || env_isset("ALL_PROXY"))
447 string schemeproxy = _url.getScheme() + "_proxy";
448 if (env_isset(schemeproxy))
450 if (schemeproxy != "http_proxy")
452 std::transform(schemeproxy.begin(), schemeproxy.end(), schemeproxy.begin(), ::toupper);
453 if (env_isset(schemeproxy))
457 XXX << "checking DNS lookup of " << host << endl;
461 _state = WORKER_BROKEN;
462 strncpy(_curlError, "DNS pipe creation failed", CURL_ERROR_SIZE);
466 if (_pid == pid_t(-1))
471 _state = WORKER_BROKEN;
472 strncpy(_curlError, "DNS checker fork failed", CURL_ERROR_SIZE);
478 // XXX: close all other file descriptors
479 struct addrinfo *ai, aihints;
480 memset(&aihints, 0, sizeof(aihints));
481 aihints.ai_family = PF_UNSPEC;
482 int tstsock = socket(PF_INET6, SOCK_DGRAM | SOCK_CLOEXEC, 0);
484 aihints.ai_family = PF_INET;
487 aihints.ai_socktype = SOCK_STREAM;
488 aihints.ai_flags = AI_CANONNAME;
489 unsigned int connecttimeout = _request->_connect_timeout;
491 alarm(connecttimeout);
492 signal(SIGALRM, SIG_DFL);
493 if (getaddrinfo(host.c_str(), NULL, &aihints, &ai))
498 _dnspipe = pipefds[0];
499 _state = WORKER_LOOKUP;
503 multifetchworker::adddnsfd(fd_set &rset, int &maxfd)
505 if (_state != WORKER_LOOKUP)
507 FD_SET(_dnspipe, &rset);
508 if (maxfd < _dnspipe)
513 multifetchworker::dnsevent(fd_set &rset)
516 if (_state != WORKER_LOOKUP || !FD_ISSET(_dnspipe, &rset))
519 while (waitpid(_pid, &status, 0) == -1)
530 if (!WIFEXITED(status))
532 _state = WORKER_BROKEN;
533 strncpy(_curlError, "DNS lookup failed", CURL_ERROR_SIZE);
534 _request->_activeworkers--;
537 int exitcode = WEXITSTATUS(status);
538 XXX << "#" << _workerno << ": DNS lookup returned " << exitcode << endl;
541 _state = WORKER_BROKEN;
542 strncpy(_curlError, "DNS lookup failed", CURL_ERROR_SIZE);
543 _request->_activeworkers--;
546 _request->_context->setDNSok(_url.getHost());
551 multifetchworker::checkChecksum()
553 // XXX << "checkChecksum block " << _blkno << endl;
554 if (!_blksize || !_request->_blklist)
556 return _request->_blklist->verifyDigest(_blkno, _dig);
560 multifetchworker::recheckChecksum()
562 // XXX << "recheckChecksum block " << _blkno << endl;
563 if (!_request->_fp || !_blksize || !_request->_blklist)
565 if (fseeko(_request->_fp, _blkstart, SEEK_SET))
569 _request->_blklist->createDigest(_dig); // resets digest
572 size_t cnt = l > sizeof(buf) ? sizeof(buf) : l;
573 if (fread(buf, cnt, 1, _request->_fp) != 1)
575 _dig.update(buf, cnt);
578 return _request->_blklist->verifyDigest(_blkno, _dig);
583 multifetchworker::stealjob()
585 if (!_request->_stealing)
587 XXX << "start stealing!" << endl;
588 _request->_stealing = true;
590 multifetchworker *best = 0;
591 std::list<multifetchworker *>::iterator workeriter = _request->_workers.begin();
593 for (; workeriter != _request->_workers.end(); ++workeriter)
595 multifetchworker *worker = *workeriter;
598 if (worker->_pass == -1)
599 continue; // do not steal!
600 if (worker->_state == WORKER_DISCARD || worker->_state == WORKER_DONE || worker->_state == WORKER_SLEEP || !worker->_blksize)
601 continue; // do not steal finished jobs
602 if (!worker->_avgspeed && worker->_blkreceived)
606 if (now > worker->_blkstarttime)
607 worker->_avgspeed = worker->_blkreceived / (now - worker->_blkstarttime);
609 if (!best || best->_pass > worker->_pass)
614 if (best->_pass < worker->_pass)
616 // if it is the same block, we want to know the best worker, otherwise the worst
617 if (worker->_blkstart == best->_blkstart)
619 if ((worker->_blksize - worker->_blkreceived) * best->_avgspeed < (best->_blksize - best->_blkreceived) * worker->_avgspeed)
624 if ((worker->_blksize - worker->_blkreceived) * best->_avgspeed > (best->_blksize - best->_blkreceived) * worker->_avgspeed)
630 _state = WORKER_DONE;
631 _request->_activeworkers--;
632 _request->_finished = true;
635 // do not sleep twice
636 if (_state != WORKER_SLEEP)
638 if (!_avgspeed && _blkreceived)
642 if (now > _blkstarttime)
643 _avgspeed = _blkreceived / (now - _blkstarttime);
646 // lets see if we should sleep a bit
647 XXX << "me #" << _workerno << ": " << _avgspeed << ", size " << best->_blksize << endl;
648 XXX << "best #" << best->_workerno << ": " << best->_avgspeed << ", size " << (best->_blksize - best->_blkreceived) << endl;
649 if (_avgspeed && best->_avgspeed && best->_blksize - best->_blkreceived > 0 &&
650 (best->_blksize - best->_blkreceived) * _avgspeed < best->_blksize * best->_avgspeed)
654 double sl = (best->_blksize - best->_blkreceived) / best->_avgspeed * 2;
657 XXX << "#" << _workerno << ": going to sleep for " << sl * 1000 << " ms" << endl;
658 _sleepuntil = now + sl;
659 _state = WORKER_SLEEP;
660 _request->_sleepworkers++;
666 best->_competing = true;
667 _blkstart = best->_blkstart;
668 _blksize = best->_blksize;
671 _blkno = best->_blkno;
676 multifetchworker::disableCompetition()
678 std::list<multifetchworker *>::iterator workeriter = _request->_workers.begin();
679 for (; workeriter != _request->_workers.end(); ++workeriter)
681 multifetchworker *worker = *workeriter;
684 if (worker->_blkstart == _blkstart)
686 if (worker->_state == WORKER_FETCH)
687 worker->_state = WORKER_DISCARD;
688 worker->_pass = -1; /* do not steal this one, we already have it */
695 multifetchworker::nextjob()
698 if (_request->_stealing)
704 MediaBlockList *blklist = _request->_blklist;
708 if (_request->_filesize != off_t(-1))
710 if (_request->_blkoff >= _request->_filesize)
715 _blksize = _request->_filesize - _request->_blkoff;
716 if (_blksize > BLKSIZE)
722 MediaBlock blk = blklist->getBlock(_request->_blkno);
723 while (_request->_blkoff >= (off_t)(blk.off + blk.size))
725 if (++_request->_blkno == blklist->numBlocks())
730 blk = blklist->getBlock(_request->_blkno);
731 _request->_blkoff = blk.off;
733 _blksize = blk.off + blk.size - _request->_blkoff;
734 if (_blksize > BLKSIZE && !blklist->haveChecksum(_request->_blkno))
737 _blkno = _request->_blkno;
738 _blkstart = _request->_blkoff;
739 _request->_blkoff += _blksize;
744 multifetchworker::run()
748 if (_state == WORKER_BROKEN || _state == WORKER_DONE)
749 return; // just in case...
751 sprintf(rangebuf, "%llu-", (unsigned long long)_blkstart);
753 sprintf(rangebuf, "%llu-%llu", (unsigned long long)_blkstart, (unsigned long long)_blkstart + _blksize - 1);
754 XXX << "#" << _workerno << ": BLK " << _blkno << ":" << rangebuf << " " << _url << endl;
755 if (curl_easy_setopt(_curl, CURLOPT_RANGE, !_noendrange || _blkstart != 0 ? rangebuf : (char *)0) != CURLE_OK)
757 _request->_activeworkers--;
758 _state = WORKER_BROKEN;
759 strncpy(_curlError, "curl_easy_setopt range failed", CURL_ERROR_SIZE);
762 if (curl_multi_add_handle(_request->_multi, _curl) != CURLM_OK)
764 _request->_activeworkers--;
765 _state = WORKER_BROKEN;
766 strncpy(_curlError, "curl_multi_add_handle failed", CURL_ERROR_SIZE);
769 _request->_havenewjob = true;
772 if (_request->_blklist)
773 _request->_blklist->createDigest(_dig); // resets digest
774 _state = WORKER_FETCH;
776 double now = currentTime();
782 //////////////////////////////////////////////////////////////////////
785 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)
790 _filesize = filesize;
796 _blkoff = _blklist->getBlock(0).off;
805 _fetchedgoodsize = 0;
807 _lastperiodstart = _lastprogress = _starttime = currentTime();
808 _lastperiodfetched = 0;
811 _connect_timeout = 0;
816 for (size_t blkno = 0; blkno < blklist->numBlocks(); blkno++)
818 MediaBlock blk = blklist->getBlock(blkno);
819 _totalsize += blk.size;
822 else if (filesize != off_t(-1))
823 _totalsize = filesize;
826 multifetchrequest::~multifetchrequest()
828 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
830 multifetchworker *worker = *workeriter;
838 multifetchrequest::run(std::vector<Url> &urllist)
841 std::vector<Url>::iterator urliter = urllist.begin();
844 fd_set rset, wset, xset;
849 XXX << "finished!" << endl;
853 if ((int)_activeworkers < _maxworkers && urliter != urllist.end() && _workers.size() < MAXURLS)
855 // spawn another worker!
856 multifetchworker *worker = new multifetchworker(workerno++, *this, *urliter);
857 _workers.push_back(worker);
858 if (worker->_state != WORKER_BROKEN)
861 if (worker->_state != WORKER_LOOKUP)
873 WAR << "No more active workers!" << endl;
874 // show the first worker error we find
875 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
877 if ((*workeriter)->_state != WORKER_BROKEN)
879 ZYPP_THROW(MediaCurlException(_baseurl, "Server error", (*workeriter)->_curlError));
888 curl_multi_fdset(_multi, &rset, &wset, &xset, &maxfd);
891 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
892 (*workeriter)->adddnsfd(rset, maxfd);
895 // if we added a new job we have to call multi_perform once
896 // to make it show up in the fd set. do not sleep in this case.
898 tv.tv_usec = _havenewjob ? 0 : 200000;
899 if (_sleepworkers && !_havenewjob)
901 if (_minsleepuntil == 0)
903 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
905 multifetchworker *worker = *workeriter;
906 if (worker->_state != WORKER_SLEEP)
908 if (!_minsleepuntil || _minsleepuntil > worker->_sleepuntil)
909 _minsleepuntil = worker->_sleepuntil;
912 double sl = _minsleepuntil - currentTime();
919 tv.tv_usec = sl * 1000000;
921 int r = select(maxfd + 1, &rset, &wset, &xset, &tv);
922 if (r == -1 && errno != EINTR)
923 ZYPP_THROW(MediaCurlException(_baseurl, "select() failed", "unknown error"));
924 if (r != 0 && _lookupworkers)
925 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
927 multifetchworker *worker = *workeriter;
928 if (worker->_state != WORKER_LOOKUP)
930 (*workeriter)->dnsevent(rset);
931 if (worker->_state != WORKER_LOOKUP)
941 mcode = curl_multi_perform(_multi, &tasks);
942 if (mcode == CURLM_CALL_MULTI_PERFORM)
944 if (mcode != CURLM_OK)
945 ZYPP_THROW(MediaCurlException(_baseurl, "curl_multi_perform", "unknown error"));
949 double now = currentTime();
952 if (now > _lastperiodstart + .5)
955 _periodavg = (_fetchedsize - _lastperiodfetched) / (now - _lastperiodstart);
957 _periodavg = (_periodavg + (_fetchedsize - _lastperiodfetched) / (now - _lastperiodstart)) / 2;
958 _lastperiodfetched = _fetchedsize;
959 _lastperiodstart = now;
965 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
967 multifetchworker *worker = *workeriter;
968 if (worker->_state != WORKER_SLEEP)
970 if (worker->_sleepuntil > now)
972 if (_minsleepuntil == worker->_sleepuntil)
974 XXX << "#" << worker->_workerno << ": sleep done, wake up" << endl;
976 // nextjob chnages the state
981 // collect all curl results, reschedule new jobs
983 while ((msg = curl_multi_info_read(_multi, &nqueue)) != 0)
985 if (msg->msg != CURLMSG_DONE)
987 CURL *easy = msg->easy_handle;
988 CURLcode cc = msg->data.result;
989 multifetchworker *worker;
990 if (curl_easy_getinfo(easy, CURLINFO_PRIVATE, &worker) != CURLE_OK)
991 ZYPP_THROW(MediaCurlException(_baseurl, "curl_easy_getinfo", "unknown error"));
992 if (worker->_blkreceived && now > worker->_blkstarttime)
994 if (worker->_avgspeed)
995 worker->_avgspeed = (worker->_avgspeed + worker->_blkreceived / (now - worker->_blkstarttime)) / 2;
997 worker->_avgspeed = worker->_blkreceived / (now - worker->_blkstarttime);
999 XXX << "#" << worker->_workerno << ": BLK " << worker->_blkno << " done code " << cc << " speed " << worker->_avgspeed << endl;
1000 curl_multi_remove_handle(_multi, easy);
1001 if (cc == CURLE_HTTP_RETURNED_ERROR)
1003 long statuscode = 0;
1004 (void)curl_easy_getinfo(easy, CURLINFO_RESPONSE_CODE, &statuscode);
1005 XXX << "HTTP status " << statuscode << endl;
1006 if (statuscode == 416 && !_blklist) /* Range error */
1008 if (_filesize == off_t(-1))
1010 if (!worker->_noendrange)
1012 XXX << "#" << worker->_workerno << ": retrying with no end range" << endl;
1013 worker->_noendrange = true;
1017 worker->_noendrange = false;
1021 if (worker->_blkstart >= _filesize)
1030 if (!worker->checkChecksum())
1032 WAR << "#" << worker->_workerno << ": checksum error, disable worker" << endl;
1033 worker->_state = WORKER_BROKEN;
1034 strncpy(worker->_curlError, "checksum error", CURL_ERROR_SIZE);
1038 if (worker->_state == WORKER_FETCH)
1040 if (worker->_competing)
1042 worker->disableCompetition();
1043 // multiple workers wrote into this block. We already know that our
1044 // data was correct, but maybe some other worker overwrote our data
1045 // with something broken. Thus we have to re-check the block.
1046 if (!worker->recheckChecksum())
1048 XXX << "#" << worker->_workerno << ": recheck checksum error, refetch block" << endl;
1049 // re-fetch! No need to worry about the bad workers,
1050 // they will now be set to DISCARD. At the end of their block
1051 // they will notice that they wrote bad data and go into BROKEN.
1056 _fetchedgoodsize += worker->_blksize;
1059 // make bad workers sleep a little
1061 int maxworkerno = 0;
1063 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
1065 multifetchworker *oworker = *workeriter;
1066 if (oworker->_state == WORKER_BROKEN)
1068 if (oworker->_avgspeed > maxavg)
1070 maxavg = oworker->_avgspeed;
1071 maxworkerno = oworker->_workerno;
1073 if (oworker->_avgspeed > worker->_avgspeed)
1076 if (maxavg && !_stealing)
1078 double ratio = worker->_avgspeed / maxavg;
1080 if (numbetter < 3) // don't sleep that much if we're in the top two
1081 ratio = ratio * ratio;
1084 XXX << "#" << worker->_workerno << ": too slow ("<< ratio << ", " << worker->_avgspeed << ", #" << maxworkerno << ": " << maxavg << "), going to sleep for " << ratio * 1000 << " ms" << endl;
1085 worker->_sleepuntil = now + ratio;
1086 worker->_state = WORKER_SLEEP;
1092 // do rate control (if requested)
1093 // should use periodavg, but that's not what libcurl does
1094 if (_maxspeed && now > _starttime)
1096 double avg = _fetchedsize / (now - _starttime);
1097 avg = worker->_maxspeed * _maxspeed / avg;
1098 if (avg < _maxspeed / _maxworkers)
1099 avg = _maxspeed / _maxworkers;
1100 if (avg > _maxspeed)
1104 worker->_maxspeed = avg;
1105 #if CURLVERSION_AT_LEAST(7,15,5)
1106 curl_easy_setopt(worker->_curl, CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t)(avg));
1114 worker->_state = WORKER_BROKEN;
1116 if (!_activeworkers && !(urliter != urllist.end() && _workers.size() < MAXURLS))
1118 // end of workers reached! goodbye!
1119 worker->evaluateCurlCode(Pathname(), cc, false);
1127 int percent = _totalsize ? (100 * (_fetchedgoodsize + _fetchedsize)) / (_totalsize + _fetchedsize) : 0;
1129 if (now > _starttime)
1130 avg = _fetchedsize / (now - _starttime);
1131 if (!(*(_report))->progress(percent, _baseurl, avg, _lastperiodstart == _starttime ? avg : _periodavg))
1132 ZYPP_THROW(MediaCurlException(_baseurl, "User abort", "cancelled"));
1135 if (_timeout && now - _lastprogress > _timeout)
1140 ZYPP_THROW(MediaTimeoutException(_baseurl));
1142 // print some download stats
1143 WAR << "overall result" << endl;
1144 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
1146 multifetchworker *worker = *workeriter;
1147 WAR << "#" << worker->_workerno << ": state: " << worker->_state << " received: " << worker->_received << " url: " << worker->_url << endl;
1152 //////////////////////////////////////////////////////////////////////
1155 MediaMultiCurl::MediaMultiCurl(const Url &url_r, const Pathname & attach_point_hint_r)
1156 : MediaCurl(url_r, attach_point_hint_r)
1158 MIL << "MediaMultiCurl::MediaMultiCurl(" << url_r << ", " << attach_point_hint_r << ")" << endl;
1160 _customHeadersMetalink = 0;
1163 MediaMultiCurl::~MediaMultiCurl()
1165 if (_customHeadersMetalink)
1167 curl_slist_free_all(_customHeadersMetalink);
1168 _customHeadersMetalink = 0;
1172 curl_multi_cleanup(_multi);
1175 std::map<std::string, CURL *>::iterator it;
1176 for (it = _easypool.begin(); it != _easypool.end(); it++)
1178 CURL *easy = it->second;
1181 curl_easy_cleanup(easy);
1187 void MediaMultiCurl::setupEasy()
1189 MediaCurl::setupEasy();
1191 if (_customHeadersMetalink)
1193 curl_slist_free_all(_customHeadersMetalink);
1194 _customHeadersMetalink = 0;
1196 struct curl_slist *sl = _customHeaders;
1197 for (; sl; sl = sl->next)
1198 _customHeadersMetalink = curl_slist_append(_customHeadersMetalink, sl->data);
1199 _customHeadersMetalink = curl_slist_append(_customHeadersMetalink, "Accept: */*, application/metalink+xml, application/metalink4+xml");
1202 static bool looks_like_metalink_fd(int fd)
1206 while ((l = pread(fd, buf, sizeof(buf) - 1, (off_t)0)) == -1 && errno == EINTR)
1212 while (*p == ' ' || *p == '\t' || *p == '\r' || *p == '\n')
1214 if (!strncasecmp(p, "<?xml", 5))
1216 while (*p && *p != '>')
1220 while (*p == ' ' || *p == '\t' || *p == '\r' || *p == '\n')
1223 bool ret = !strncasecmp(p, "<metalink", 9) ? true : false;
1227 static bool looks_like_metalink(const Pathname & file)
1230 if ((fd = open(file.asString().c_str(), O_RDONLY|O_CLOEXEC)) == -1)
1232 bool ret = looks_like_metalink_fd(fd);
1234 DBG << "looks_like_metalink(" << file << "): " << ret << endl;
1238 // here we try to suppress all progress coming from a metalink download
1239 int MediaMultiCurl::progressCallback( void *clientp, double dltotal, double dlnow, double ultotal, double ulnow)
1241 CURL *_curl = MediaCurl::progressCallback_getcurl(clientp);
1245 // work around curl bug that gives us old data
1246 long httpReturnCode = 0;
1247 if (curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &httpReturnCode ) != CURLE_OK || httpReturnCode == 0)
1251 bool ismetalink = false;
1252 if (curl_easy_getinfo(_curl, CURLINFO_CONTENT_TYPE, &ptr) == CURLE_OK && ptr)
1254 string ct = string(ptr);
1255 if (ct.find("application/metalink+xml") == 0 || ct.find("application/metalink4+xml") == 0)
1258 if (!ismetalink && dlnow < 256)
1260 // can't tell yet, suppress callback
1266 if (curl_easy_getinfo(_curl, CURLINFO_PRIVATE, &fp) != CURLE_OK)
1271 ismetalink = looks_like_metalink_fd(fileno(fp));
1272 DBG << "looks_like_metalink_fd: " << ismetalink << endl;
1276 // we're downloading the metalink file. no progress please.
1277 curl_easy_setopt(_curl, CURLOPT_NOPROGRESS, 1L);
1280 curl_easy_setopt(_curl, CURLOPT_PROGRESSFUNCTION, &MediaCurl::progressCallback);
1281 return MediaCurl::progressCallback(clientp, dltotal, dlnow, ultotal, ulnow);
1284 void MediaMultiCurl::doGetFileCopy( const Pathname & filename , const Pathname & target, callback::SendReport<DownloadProgressReport> & report, RequestOptions options ) const
1286 Pathname dest = target.absolutename();
1287 if( assert_dir( dest.dirname() ) )
1289 DBG << "assert_dir " << dest.dirname() << " failed" << endl;
1290 Url url(getFileUrl(filename));
1291 ZYPP_THROW( MediaSystemException(url, "System error on " + dest.dirname().asString()) );
1293 string destNew = target.asString() + ".new.zypp.XXXXXX";
1294 char *buf = ::strdup( destNew.c_str());
1297 ERR << "out of memory for temp file name" << endl;
1298 Url url(getFileUrl(filename));
1299 ZYPP_THROW(MediaSystemException(url, "out of memory for temp file name"));
1302 int tmp_fd = ::mkostemp( buf, O_CLOEXEC );
1306 ERR << "mkstemp failed for file '" << destNew << "'" << endl;
1307 ZYPP_THROW(MediaWriteException(destNew));
1312 FILE *file = ::fdopen( tmp_fd, "we" );
1315 filesystem::unlink( destNew );
1316 ERR << "fopen failed for file '" << destNew << "'" << endl;
1317 ZYPP_THROW(MediaWriteException(destNew));
1319 DBG << "dest: " << dest << endl;
1320 DBG << "temp: " << destNew << endl;
1322 // set IFMODSINCE time condition (no download if not modified)
1323 if( PathInfo(target).isExist() && !(options & OPTION_NO_IFMODSINCE) )
1325 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_IFMODSINCE);
1326 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, (long)PathInfo(target).mtime());
1330 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1331 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1333 // change header to include Accept: metalink
1334 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeadersMetalink);
1335 // change to our own progress funcion
1336 curl_easy_setopt(_curl, CURLOPT_PROGRESSFUNCTION, &progressCallback);
1337 curl_easy_setopt(_curl, CURLOPT_PRIVATE, file);
1340 MediaCurl::doGetFileCopyFile(filename, dest, file, report, options);
1342 catch (Exception &ex)
1345 filesystem::unlink(destNew);
1346 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1347 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1348 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeaders);
1349 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (void *)0);
1352 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1353 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1354 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeaders);
1355 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (void *)0);
1356 long httpReturnCode = 0;
1357 CURLcode infoRet = curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &httpReturnCode);
1358 if (infoRet == CURLE_OK)
1360 DBG << "HTTP response: " + str::numstring(httpReturnCode) << endl;
1361 if ( httpReturnCode == 304
1362 || ( httpReturnCode == 213 && _url.getScheme() == "ftp" ) ) // not modified
1364 DBG << "not modified: " << PathInfo(dest) << endl;
1370 WAR << "Could not get the reponse code." << endl;
1373 bool ismetalink = false;
1376 if (curl_easy_getinfo(_curl, CURLINFO_CONTENT_TYPE, &ptr) == CURLE_OK && ptr)
1378 string ct = string(ptr);
1379 if (ct.find("application/metalink+xml") == 0 || ct.find("application/metalink4+xml") == 0)
1385 // some proxies do not store the content type, so also look at the file to find
1386 // out if we received a metalink (bnc#649925)
1388 if (looks_like_metalink(Pathname(destNew)))
1394 bool userabort = false;
1397 Pathname failedFile = ZConfig::instance().repoCachePath() / "MultiCurl.failed";
1401 mlp.parse(Pathname(destNew));
1402 MediaBlockList bl = mlp.getBlockList();
1403 vector<Url> urls = mlp.getUrls();
1405 file = fopen(destNew.c_str(), "w+e");
1407 ZYPP_THROW(MediaWriteException(destNew));
1408 if (PathInfo(target).isExist())
1410 XXX << "reusing blocks from file " << target << endl;
1411 bl.reuseBlocks(file, target.asString());
1414 if (bl.haveChecksum(1) && PathInfo(failedFile).isExist())
1416 XXX << "reusing blocks from file " << failedFile << endl;
1417 bl.reuseBlocks(file, failedFile.asString());
1419 filesystem::unlink(failedFile);
1421 Pathname df = deltafile();
1424 XXX << "reusing blocks from file " << df << endl;
1425 bl.reuseBlocks(file, df.asString());
1430 multifetch(filename, file, &urls, &report, &bl);
1432 catch (MediaCurlException &ex)
1434 userabort = ex.errstr() == "User abort";
1438 catch (Exception &ex)
1440 // something went wrong. fall back to normal download
1444 if (PathInfo(destNew).size() >= 63336)
1446 ::unlink(failedFile.asString().c_str());
1447 filesystem::hardlinkCopy(destNew, failedFile);
1451 filesystem::unlink(destNew);
1454 file = fopen(destNew.c_str(), "w+e");
1456 ZYPP_THROW(MediaWriteException(destNew));
1457 MediaCurl::doGetFileCopyFile(filename, dest, file, report, options | OPTION_NO_REPORT_START);
1461 if (::fchmod( ::fileno(file), filesystem::applyUmaskTo( 0644 )))
1463 ERR << "Failed to chmod file " << destNew << endl;
1467 filesystem::unlink(destNew);
1468 ERR << "Fclose failed for file '" << destNew << "'" << endl;
1469 ZYPP_THROW(MediaWriteException(destNew));
1471 if ( rename( destNew, dest ) != 0 )
1473 ERR << "Rename failed" << endl;
1474 ZYPP_THROW(MediaWriteException(dest));
1476 DBG << "done: " << PathInfo(dest) << endl;
1479 ///////////////////////////////////////////////////////////////////
1481 // bsc#933839: propagate proxy settings passed in the repo URL
1482 inline Url propagateQueryParams( Url url_r, const Url & template_r )
1484 for ( std::string param : { "proxy", "proxyport", "proxyuser", "proxypass"} )
1486 const std::string & value( template_r.getQueryParam( param ) );
1487 if ( ! value.empty() )
1488 url_r.setQueryParam( param, value );
1493 ///////////////////////////////////////////////////////////////////
1495 void MediaMultiCurl::multifetch(const Pathname & filename, FILE *fp, std::vector<Url> *urllist, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize) const
1497 Url baseurl(getFileUrl(filename));
1498 if (blklist && filesize == off_t(-1) && blklist->haveFilesize())
1499 filesize = blklist->getFilesize();
1500 if (blklist && !blklist->haveBlocks() && filesize != 0)
1502 if (blklist && (filesize == 0 || !blklist->numBlocks()))
1504 checkFileDigest(baseurl, fp, blklist);
1511 _multi = curl_multi_init();
1513 ZYPP_THROW(MediaCurlInitException(baseurl));
1515 multifetchrequest req(this, filename, baseurl, _multi, fp, report, blklist, filesize);
1516 req._timeout = _settings.timeout();
1517 req._connect_timeout = _settings.connectTimeout();
1518 req._maxspeed = _settings.maxDownloadSpeed();
1519 req._maxworkers = _settings.maxConcurrentConnections();
1520 if (req._maxworkers > MAXURLS)
1521 req._maxworkers = MAXURLS;
1522 if (req._maxworkers <= 0)
1523 req._maxworkers = 1;
1524 std::vector<Url> myurllist;
1525 for (std::vector<Url>::iterator urliter = urllist->begin(); urliter != urllist->end(); ++urliter)
1529 string scheme = urliter->getScheme();
1530 if (scheme == "http" || scheme == "https" || scheme == "ftp" || scheme == "tftp")
1532 checkProtocol(*urliter);
1533 myurllist.push_back(propagateQueryParams(*urliter, _url));
1540 if (!myurllist.size())
1541 myurllist.push_back(baseurl);
1543 checkFileDigest(baseurl, fp, blklist);
1546 void MediaMultiCurl::checkFileDigest(Url &url, FILE *fp, MediaBlockList *blklist) const
1548 if (!blklist || !blklist->haveFileChecksum())
1550 if (fseeko(fp, off_t(0), SEEK_SET))
1551 ZYPP_THROW(MediaCurlException(url, "fseeko", "seek error"));
1553 blklist->createFileDigest(dig);
1556 while ((l = fread(buf, 1, sizeof(buf), fp)) > 0)
1558 if (!blklist->verifyFileDigest(dig))
1559 ZYPP_THROW(MediaCurlException(url, "file verification failed", "checksum error"));
1562 bool MediaMultiCurl::isDNSok(const string &host) const
1564 return _dnsok.find(host) == _dnsok.end() ? false : true;
1567 void MediaMultiCurl::setDNSok(const string &host) const
1569 _dnsok.insert(host);
1572 CURL *MediaMultiCurl::fromEasyPool(const string &host) const
1574 if (_easypool.find(host) == _easypool.end())
1576 CURL *ret = _easypool[host];
1577 _easypool.erase(host);
1581 void MediaMultiCurl::toEasyPool(const std::string &host, CURL *easy) const
1583 CURL *oldeasy = _easypool[host];
1584 _easypool[host] = easy;
1586 curl_easy_cleanup(oldeasy);
1589 } // namespace media