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;
37 //////////////////////////////////////////////////////////////////////
40 class multifetchrequest;
42 // Hack: we derive from MediaCurl just to get the storage space for
43 // settings, url, curlerrors and the like
45 class multifetchworker : MediaCurl {
46 friend class multifetchrequest;
49 multifetchworker(int no, multifetchrequest &request, const Url &url);
54 bool recheckChecksum();
55 void disableCompetition();
58 void adddnsfd(fd_set &rset, int &maxfd);
59 void dnsevent(fd_set &rset);
83 size_t writefunction(void *ptr, size_t size);
84 static size_t _writefunction(void *ptr, size_t size, size_t nmemb, void *stream);
86 size_t headerfunction(char *ptr, size_t size);
87 static size_t _headerfunction(void *ptr, size_t size, size_t nmemb, void *stream);
89 multifetchrequest *_request;
100 #define WORKER_STARTING 0
101 #define WORKER_LOOKUP 1
102 #define WORKER_FETCH 2
103 #define WORKER_DISCARD 3
104 #define WORKER_DONE 4
105 #define WORKER_SLEEP 5
106 #define WORKER_BROKEN 6
110 class multifetchrequest {
112 multifetchrequest(const MediaMultiCurl *context, const Pathname &filename, const Url &baseurl, CURLM *multi, FILE *fp, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize);
113 ~multifetchrequest();
115 void run(std::vector<Url> &urllist);
118 friend class multifetchworker;
120 const MediaMultiCurl *_context;
121 const Pathname _filename;
125 callback::SendReport<DownloadProgressReport> *_report;
126 MediaBlockList *_blklist;
131 std::list<multifetchworker *> _workers;
137 size_t _activeworkers;
138 size_t _lookupworkers;
139 size_t _sleepworkers;
140 double _minsleepuntil;
144 off_t _fetchedgoodsize;
147 double _lastprogress;
149 double _lastperiodstart;
150 double _lastperiodfetched;
155 double _connect_timeout;
160 #define BLKSIZE 131072
164 //////////////////////////////////////////////////////////////////////
170 if (gettimeofday(&tv, NULL))
172 return tv.tv_sec + tv.tv_usec / 1000000.;
176 multifetchworker::writefunction(void *ptr, size_t size)
179 if (_state == WORKER_BROKEN)
182 double now = currentTime();
184 len = size > _size ? _size : size;
191 if (_blkstart && _off == _blkstart)
193 // make sure that the server replied with "partial content"
196 (void)curl_easy_getinfo(_curl, CURLINFO_EFFECTIVE_URL, &effurl);
197 if (effurl && !strncasecmp(effurl, "http", 4))
200 (void)curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &statuscode);
201 if (statuscode != 206)
209 _request->_lastprogress = now;
211 if (_state == WORKER_DISCARD || !_request->_fp)
213 // block is no longer needed
214 // still calculate the checksum so that we can throw out bad servers
215 if (_request->_blklist)
216 _dig.update((const char *)ptr, len);
221 if (fseeko(_request->_fp, _off, SEEK_SET))
223 cnt = fwrite(ptr, 1, len, _request->_fp);
226 _request->_fetchedsize += cnt;
227 if (_request->_blklist)
228 _dig.update((const char *)ptr, cnt);
238 multifetchworker::_writefunction(void *ptr, size_t size, size_t nmemb, void *stream)
240 multifetchworker *me = reinterpret_cast<multifetchworker *>(stream);
241 return me->writefunction(ptr, size * nmemb);
245 multifetchworker::headerfunction(char *p, size_t size)
248 if (l > 9 && !strncasecmp(p, "Location:", 9))
250 string line(p + 9, l - 9);
251 if (line[l - 10] == '\r')
252 line.erase(l - 10, 1);
253 DBG << "#" << _workerno << ": redirecting to" << line << endl;
256 if (l <= 14 || l >= 128 || strncasecmp(p, "Content-Range:", 14) != 0)
260 while (l && (*p == ' ' || *p == '\t'))
262 if (l < 6 || strncasecmp(p, "bytes", 5))
269 unsigned long long start, off, filesize;
270 if (sscanf(buf, "%llu-%llu/%llu", &start, &off, &filesize) != 3)
272 if (_request->_filesize == (off_t)-1)
274 WAR << "#" << _workerno << ": setting request filesize to " << filesize << endl;
275 _request->_filesize = filesize;
276 if (_request->_totalsize == 0 && !_request->_blklist)
277 _request->_totalsize = filesize;
279 if (_request->_filesize != (off_t)filesize)
281 DBG << "#" << _workerno << ": filesize mismatch" << endl;
282 _state = WORKER_BROKEN;
283 strncpy(_curlError, "filesize mismatch", CURL_ERROR_SIZE);
289 multifetchworker::_headerfunction(void *ptr, size_t size, size_t nmemb, void *stream)
291 multifetchworker *me = reinterpret_cast<multifetchworker *>(stream);
292 return me->headerfunction((char *)ptr, size * nmemb);
295 multifetchworker::multifetchworker(int no, multifetchrequest &request, const Url &url)
296 : MediaCurl(url, Pathname())
300 _state = WORKER_STARTING;
302 _off = _blkstart = 0;
303 _size = _blksize = 0;
313 _maxspeed = _request->_maxspeed;
316 Url curlUrl( clearQueryString(url) );
317 _urlbuf = curlUrl.asString();
318 _curl = _request->_context->fromEasyPool(_url.getHost());
320 DBG << "reused worker from pool" << endl;
321 if (!_curl && !(_curl = curl_easy_init()))
323 _state = WORKER_BROKEN;
324 strncpy(_curlError, "curl_easy_init failed", CURL_ERROR_SIZE);
331 catch (Exception &ex)
333 curl_easy_cleanup(_curl);
335 _state = WORKER_BROKEN;
336 strncpy(_curlError, "curl_easy_setopt failed", CURL_ERROR_SIZE);
339 curl_easy_setopt(_curl, CURLOPT_PRIVATE, this);
340 curl_easy_setopt(_curl, CURLOPT_URL, _urlbuf.c_str());
341 curl_easy_setopt(_curl, CURLOPT_WRITEFUNCTION, &_writefunction);
342 curl_easy_setopt(_curl, CURLOPT_WRITEDATA, this);
343 if (_request->_filesize == off_t(-1) || !_request->_blklist || !_request->_blklist->haveChecksum(0))
345 curl_easy_setopt(_curl, CURLOPT_HEADERFUNCTION, &_headerfunction);
346 curl_easy_setopt(_curl, CURLOPT_HEADERDATA, this);
348 // if this is the same host copy authorization
349 // (the host check is also what curl does when doing a redirect)
350 // (note also that unauthorized exceptions are thrown with the request host)
351 if (url.getHost() == _request->_context->_url.getHost())
353 _settings.setUsername(_request->_context->_settings.username());
354 _settings.setPassword(_request->_context->_settings.password());
355 _settings.setAuthType(_request->_context->_settings.authType());
356 if ( _settings.userPassword().size() )
358 curl_easy_setopt(_curl, CURLOPT_USERPWD, _settings.userPassword().c_str());
359 string use_auth = _settings.authType();
360 if (use_auth.empty())
361 use_auth = "digest,basic"; // our default
362 long auth = CurlAuthData::auth_type_str2long(use_auth);
363 if( auth != CURLAUTH_NONE)
365 DBG << "#" << _workerno << ": Enabling HTTP authentication methods: " << use_auth
366 << " (CURLOPT_HTTPAUTH=" << auth << ")" << std::endl;
367 curl_easy_setopt(_curl, CURLOPT_HTTPAUTH, auth);
374 multifetchworker::~multifetchworker()
378 if (_state == WORKER_FETCH || _state == WORKER_DISCARD)
379 curl_multi_remove_handle(_request->_multi, _curl);
380 if (_state == WORKER_DONE || _state == WORKER_SLEEP)
382 #if LIBCURL_VERSION_NUMBER >= 0x071505
383 curl_easy_setopt(_curl, CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t)0);
385 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (void *)0);
386 curl_easy_setopt(_curl, CURLOPT_WRITEFUNCTION, (void *)0);
387 curl_easy_setopt(_curl, CURLOPT_WRITEDATA, (void *)0);
388 curl_easy_setopt(_curl, CURLOPT_HEADERFUNCTION, (void *)0);
389 curl_easy_setopt(_curl, CURLOPT_HEADERDATA, (void *)0);
390 _request->_context->toEasyPool(_url.getHost(), _curl);
393 curl_easy_cleanup(_curl);
400 while (waitpid(_pid, &status, 0) == -1)
410 // the destructor in MediaCurl doesn't call disconnect() if
411 // the media is not attached, so we do it here manually
415 static inline bool env_isset(string name)
417 const char *s = getenv(name.c_str());
418 return s && *s ? true : false;
422 multifetchworker::checkdns()
424 string host = _url.getHost();
429 if (_request->_context->isDNSok(host))
432 // no need to do dns checking for numeric hosts
434 if (inet_pton(AF_INET, host.c_str(), addrbuf) == 1)
436 if (inet_pton(AF_INET6, host.c_str(), addrbuf) == 1)
439 // no need to do dns checking if we use a proxy
440 if (!_settings.proxy().empty())
442 if (env_isset("all_proxy") || env_isset("ALL_PROXY"))
444 string schemeproxy = _url.getScheme() + "_proxy";
445 if (env_isset(schemeproxy))
447 if (schemeproxy != "http_proxy")
449 std::transform(schemeproxy.begin(), schemeproxy.end(), schemeproxy.begin(), ::toupper);
450 if (env_isset(schemeproxy))
454 DBG << "checking DNS lookup of " << host << endl;
458 _state = WORKER_BROKEN;
459 strncpy(_curlError, "DNS pipe creation failed", CURL_ERROR_SIZE);
463 if (_pid == pid_t(-1))
468 _state = WORKER_BROKEN;
469 strncpy(_curlError, "DNS checker fork failed", CURL_ERROR_SIZE);
475 // XXX: close all other file descriptors
476 struct addrinfo *ai, aihints;
477 memset(&aihints, 0, sizeof(aihints));
478 aihints.ai_family = PF_UNSPEC;
479 int tstsock = socket(PF_INET6, SOCK_DGRAM, 0);
481 aihints.ai_family = PF_INET;
484 aihints.ai_socktype = SOCK_STREAM;
485 aihints.ai_flags = AI_CANONNAME;
486 unsigned int connecttimeout = _request->_connect_timeout;
488 alarm(connecttimeout);
489 signal(SIGALRM, SIG_DFL);
490 if (getaddrinfo(host.c_str(), NULL, &aihints, &ai))
495 _dnspipe = pipefds[0];
496 _state = WORKER_LOOKUP;
500 multifetchworker::adddnsfd(fd_set &rset, int &maxfd)
502 if (_state != WORKER_LOOKUP)
504 FD_SET(_dnspipe, &rset);
505 if (maxfd < _dnspipe)
510 multifetchworker::dnsevent(fd_set &rset)
513 if (_state != WORKER_LOOKUP || !FD_ISSET(_dnspipe, &rset))
516 while (waitpid(_pid, &status, 0) == -1)
527 if (!WIFEXITED(status))
529 _state = WORKER_BROKEN;
530 strncpy(_curlError, "DNS lookup failed", CURL_ERROR_SIZE);
531 _request->_activeworkers--;
534 int exitcode = WEXITSTATUS(status);
535 DBG << "#" << _workerno << ": DNS lookup returned " << exitcode << endl;
538 _state = WORKER_BROKEN;
539 strncpy(_curlError, "DNS lookup failed", CURL_ERROR_SIZE);
540 _request->_activeworkers--;
543 _request->_context->setDNSok(_url.getHost());
548 multifetchworker::checkChecksum()
550 // DBG << "checkChecksum block " << _blkno << endl;
551 if (!_blksize || !_request->_blklist)
553 return _request->_blklist->verifyDigest(_blkno, _dig);
557 multifetchworker::recheckChecksum()
559 // DBG << "recheckChecksum block " << _blkno << endl;
560 if (!_request->_fp || !_blksize || !_request->_blklist)
562 if (fseeko(_request->_fp, _blkstart, SEEK_SET))
566 _request->_blklist->createDigest(_dig); // resets digest
569 size_t cnt = l > sizeof(buf) ? sizeof(buf) : l;
570 if (fread(buf, cnt, 1, _request->_fp) != 1)
572 _dig.update(buf, cnt);
575 return _request->_blklist->verifyDigest(_blkno, _dig);
580 multifetchworker::stealjob()
582 if (!_request->_stealing)
584 DBG << "start stealing!" << endl;
585 _request->_stealing = true;
587 multifetchworker *best = 0;
588 std::list<multifetchworker *>::iterator workeriter = _request->_workers.begin();
590 for (; workeriter != _request->_workers.end(); ++workeriter)
592 multifetchworker *worker = *workeriter;
595 if (worker->_pass == -1)
596 continue; // do not steal!
597 if (worker->_state == WORKER_DISCARD || worker->_state == WORKER_DONE || worker->_state == WORKER_SLEEP || !worker->_blksize)
598 continue; // do not steal finished jobs
599 if (!worker->_avgspeed && worker->_blkreceived)
603 if (now > worker->_blkstarttime)
604 worker->_avgspeed = worker->_blkreceived / (now - worker->_blkstarttime);
606 if (!best || best->_pass > worker->_pass)
611 if (best->_pass < worker->_pass)
613 // if it is the same block, we want to know the best worker, otherwise the worst
614 if (worker->_blkstart == best->_blkstart)
616 if ((worker->_blksize - worker->_blkreceived) * best->_avgspeed < (best->_blksize - best->_blkreceived) * worker->_avgspeed)
621 if ((worker->_blksize - worker->_blkreceived) * best->_avgspeed > (best->_blksize - best->_blkreceived) * worker->_avgspeed)
627 _state = WORKER_DONE;
628 _request->_activeworkers--;
629 _request->_finished = true;
632 // do not sleep twice
633 if (_state != WORKER_SLEEP)
635 if (!_avgspeed && _blkreceived)
639 if (now > _blkstarttime)
640 _avgspeed = _blkreceived / (now - _blkstarttime);
643 // lets see if we should sleep a bit
644 DBG << "me #" << _workerno << ": " << _avgspeed << ", size " << best->_blksize << endl;
645 DBG << "best #" << best->_workerno << ": " << best->_avgspeed << ", size " << (best->_blksize - best->_blkreceived) << endl;
646 if (_avgspeed && best->_avgspeed && best->_blksize - best->_blkreceived > 0 &&
647 (best->_blksize - best->_blkreceived) * _avgspeed < best->_blksize * best->_avgspeed)
651 double sl = (best->_blksize - best->_blkreceived) / best->_avgspeed * 2;
654 DBG << "#" << _workerno << ": going to sleep for " << sl * 1000 << " ms" << endl;
655 _sleepuntil = now + sl;
656 _state = WORKER_SLEEP;
657 _request->_sleepworkers++;
663 best->_competing = true;
664 _blkstart = best->_blkstart;
665 _blksize = best->_blksize;
668 _blkno = best->_blkno;
673 multifetchworker::disableCompetition()
675 std::list<multifetchworker *>::iterator workeriter = _request->_workers.begin();
676 for (; workeriter != _request->_workers.end(); ++workeriter)
678 multifetchworker *worker = *workeriter;
681 if (worker->_blkstart == _blkstart)
683 if (worker->_state == WORKER_FETCH)
684 worker->_state = WORKER_DISCARD;
685 worker->_pass = -1; /* do not steal this one, we already have it */
692 multifetchworker::nextjob()
695 if (_request->_stealing)
701 MediaBlockList *blklist = _request->_blklist;
705 if (_request->_filesize != off_t(-1))
707 if (_request->_blkoff >= _request->_filesize)
712 _blksize = _request->_filesize - _request->_blkoff;
713 if (_blksize > BLKSIZE)
719 MediaBlock blk = blklist->getBlock(_request->_blkno);
720 while (_request->_blkoff >= blk.off + blk.size)
722 if (++_request->_blkno == blklist->numBlocks())
727 blk = blklist->getBlock(_request->_blkno);
728 _request->_blkoff = blk.off;
730 _blksize = blk.off + blk.size - _request->_blkoff;
731 if (_blksize > BLKSIZE && !blklist->haveChecksum(_request->_blkno))
734 _blkno = _request->_blkno;
735 _blkstart = _request->_blkoff;
736 _request->_blkoff += _blksize;
741 multifetchworker::run()
745 if (_state == WORKER_BROKEN || _state == WORKER_DONE)
746 return; // just in case...
748 sprintf(rangebuf, "%llu-", (unsigned long long)_blkstart);
750 sprintf(rangebuf, "%llu-%llu", (unsigned long long)_blkstart, (unsigned long long)_blkstart + _blksize - 1);
751 DBG << "#" << _workerno << ": BLK " << _blkno << ":" << rangebuf << " " << _url << endl;
752 if (curl_easy_setopt(_curl, CURLOPT_RANGE, !_noendrange || _blkstart != 0 ? rangebuf : (char *)0) != CURLE_OK)
754 _request->_activeworkers--;
755 _state = WORKER_BROKEN;
756 strncpy(_curlError, "curl_easy_setopt range failed", CURL_ERROR_SIZE);
759 if (curl_multi_add_handle(_request->_multi, _curl) != CURLM_OK)
761 _request->_activeworkers--;
762 _state = WORKER_BROKEN;
763 strncpy(_curlError, "curl_multi_add_handle failed", CURL_ERROR_SIZE);
766 _request->_havenewjob = true;
769 if (_request->_blklist)
770 _request->_blklist->createDigest(_dig); // resets digest
771 _state = WORKER_FETCH;
773 double now = currentTime();
779 //////////////////////////////////////////////////////////////////////
782 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)
787 _filesize = filesize;
793 _blkoff = _blklist->getBlock(0).off;
802 _fetchedgoodsize = 0;
804 _lastperiodstart = _lastprogress = _starttime = currentTime();
805 _lastperiodfetched = 0;
808 _connect_timeout = 0;
813 for (size_t blkno = 0; blkno < blklist->numBlocks(); blkno++)
815 MediaBlock blk = blklist->getBlock(blkno);
816 _totalsize += blk.size;
819 else if (filesize != off_t(-1))
820 _totalsize = filesize;
823 multifetchrequest::~multifetchrequest()
825 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
827 multifetchworker *worker = *workeriter;
835 multifetchrequest::run(std::vector<Url> &urllist)
838 std::vector<Url>::iterator urliter = urllist.begin();
841 fd_set rset, wset, xset;
846 DBG << "finished!" << endl;
850 if (_activeworkers < _maxworkers && urliter != urllist.end() && _workers.size() < MAXURLS)
852 // spawn another worker!
853 multifetchworker *worker = new multifetchworker(workerno++, *this, *urliter);
854 _workers.push_back(worker);
855 if (worker->_state != WORKER_BROKEN)
858 if (worker->_state != WORKER_LOOKUP)
870 WAR << "No more active workers!" << endl;
871 // show the first worker error we find
872 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
874 if ((*workeriter)->_state != WORKER_BROKEN)
876 ZYPP_THROW(MediaCurlException(_baseurl, "Server error", (*workeriter)->_curlError));
885 curl_multi_fdset(_multi, &rset, &wset, &xset, &maxfd);
888 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
889 (*workeriter)->adddnsfd(rset, maxfd);
892 // if we added a new job we have to call multi_perform once
893 // to make it show up in the fd set. do not sleep in this case.
895 tv.tv_usec = _havenewjob ? 0 : 200000;
896 if (_sleepworkers && !_havenewjob)
898 if (_minsleepuntil == 0)
900 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
902 multifetchworker *worker = *workeriter;
903 if (worker->_state != WORKER_SLEEP)
905 if (!_minsleepuntil || _minsleepuntil > worker->_sleepuntil)
906 _minsleepuntil = worker->_sleepuntil;
909 double sl = _minsleepuntil - currentTime();
916 tv.tv_usec = sl * 1000000;
918 int r = select(maxfd + 1, &rset, &wset, &xset, &tv);
919 if (r == -1 && errno != EINTR)
920 ZYPP_THROW(MediaCurlException(_baseurl, "select() failed", "unknown error"));
921 if (r != 0 && _lookupworkers)
922 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
924 multifetchworker *worker = *workeriter;
925 if (worker->_state != WORKER_LOOKUP)
927 (*workeriter)->dnsevent(rset);
928 if (worker->_state != WORKER_LOOKUP)
938 mcode = curl_multi_perform(_multi, &tasks);
939 if (mcode == CURLM_CALL_MULTI_PERFORM)
941 if (mcode != CURLM_OK)
942 ZYPP_THROW(MediaCurlException(_baseurl, "curl_multi_perform", "unknown error"));
946 double now = currentTime();
949 if (now > _lastperiodstart + .5)
952 _periodavg = (_fetchedsize - _lastperiodfetched) / (now - _lastperiodstart);
954 _periodavg = (_periodavg + (_fetchedsize - _lastperiodfetched) / (now - _lastperiodstart)) / 2;
955 _lastperiodfetched = _fetchedsize;
956 _lastperiodstart = now;
962 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
964 multifetchworker *worker = *workeriter;
965 if (worker->_state != WORKER_SLEEP)
967 if (worker->_sleepuntil > now)
969 if (_minsleepuntil == worker->_sleepuntil)
971 DBG << "#" << worker->_workerno << ": sleep done, wake up" << endl;
973 // nextjob chnages the state
978 // collect all curl results, reschedule new jobs
980 while ((msg = curl_multi_info_read(_multi, &nqueue)) != 0)
982 if (msg->msg != CURLMSG_DONE)
984 CURL *easy = msg->easy_handle;
985 CURLcode cc = msg->data.result;
986 multifetchworker *worker;
987 if (curl_easy_getinfo(easy, CURLINFO_PRIVATE, &worker) != CURLE_OK)
988 ZYPP_THROW(MediaCurlException(_baseurl, "curl_easy_getinfo", "unknown error"));
989 if (worker->_blkreceived && now > worker->_blkstarttime)
991 if (worker->_avgspeed)
992 worker->_avgspeed = (worker->_avgspeed + worker->_blkreceived / (now - worker->_blkstarttime)) / 2;
994 worker->_avgspeed = worker->_blkreceived / (now - worker->_blkstarttime);
996 DBG << "#" << worker->_workerno << ": BLK " << worker->_blkno << " done code " << cc << " speed " << worker->_avgspeed << endl;
997 curl_multi_remove_handle(_multi, easy);
998 if (cc == CURLE_HTTP_RETURNED_ERROR)
1000 long statuscode = 0;
1001 (void)curl_easy_getinfo(easy, CURLINFO_RESPONSE_CODE, &statuscode);
1002 DBG << "HTTP status " << statuscode << endl;
1003 if (statuscode == 416 && !_blklist) /* Range error */
1005 if (_filesize == off_t(-1))
1007 if (!worker->_noendrange)
1009 DBG << "#" << worker->_workerno << ": retrying with no end range" << endl;
1010 worker->_noendrange = true;
1014 worker->_noendrange = false;
1018 if (worker->_blkstart >= _filesize)
1027 if (!worker->checkChecksum())
1029 WAR << "#" << worker->_workerno << ": checksum error, disable worker" << endl;
1030 worker->_state = WORKER_BROKEN;
1031 strncpy(worker->_curlError, "checksum error", CURL_ERROR_SIZE);
1035 if (worker->_state == WORKER_FETCH)
1037 if (worker->_competing)
1039 worker->disableCompetition();
1040 // multiple workers wrote into this block. We already know that our
1041 // data was correct, but maybe some other worker overwrote our data
1042 // with something broken. Thus we have to re-check the block.
1043 if (!worker->recheckChecksum())
1045 DBG << "#" << worker->_workerno << ": recheck checksum error, refetch block" << endl;
1046 // re-fetch! No need to worry about the bad workers,
1047 // they will now be set to DISCARD. At the end of their block
1048 // they will notice that they wrote bad data and go into BROKEN.
1053 _fetchedgoodsize += worker->_blksize;
1056 // make bad workers sleep a little
1058 int maxworkerno = 0;
1060 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
1062 multifetchworker *oworker = *workeriter;
1063 if (oworker->_state == WORKER_BROKEN)
1065 if (oworker->_avgspeed > maxavg)
1067 maxavg = oworker->_avgspeed;
1068 maxworkerno = oworker->_workerno;
1070 if (oworker->_avgspeed > worker->_avgspeed)
1073 if (maxavg && !_stealing)
1075 double ratio = worker->_avgspeed / maxavg;
1077 if (numbetter < 3) // don't sleep that much if we're in the top two
1078 ratio = ratio * ratio;
1081 DBG << "#" << worker->_workerno << ": too slow ("<< ratio << ", " << worker->_avgspeed << ", #" << maxworkerno << ": " << maxavg << "), going to sleep for " << ratio * 1000 << " ms" << endl;
1082 worker->_sleepuntil = now + ratio;
1083 worker->_state = WORKER_SLEEP;
1089 // do rate control (if requested)
1090 // should use periodavg, but that's not what libcurl does
1091 if (_maxspeed && now > _starttime)
1093 double avg = _fetchedsize / (now - _starttime);
1094 avg = worker->_maxspeed * _maxspeed / avg;
1095 if (avg < _maxspeed / _maxworkers)
1096 avg = _maxspeed / _maxworkers;
1097 if (avg > _maxspeed)
1101 worker->_maxspeed = avg;
1102 #if LIBCURL_VERSION_NUMBER >= 0x071505
1103 curl_easy_setopt(worker->_curl, CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t)(avg));
1111 worker->_state = WORKER_BROKEN;
1113 if (!_activeworkers && !(urliter != urllist.end() && _workers.size() < MAXURLS))
1115 // end of workers reached! goodbye!
1116 worker->evaluateCurlCode(Pathname(), cc, false);
1124 int percent = _totalsize ? (100 * (_fetchedgoodsize + _fetchedsize)) / (_totalsize + _fetchedsize) : 0;
1126 if (now > _starttime)
1127 avg = _fetchedsize / (now - _starttime);
1128 if (!(*(_report))->progress(percent, _baseurl, avg, _lastperiodstart == _starttime ? avg : _periodavg))
1129 ZYPP_THROW(MediaCurlException(_baseurl, "User abort", "cancelled"));
1132 if (_timeout && now - _lastprogress > _timeout)
1137 ZYPP_THROW(MediaTimeoutException(_baseurl));
1139 // print some download stats
1140 WAR << "overall result" << endl;
1141 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
1143 multifetchworker *worker = *workeriter;
1144 WAR << "#" << worker->_workerno << ": state: " << worker->_state << " received: " << worker->_received << " url: " << worker->_url << endl;
1149 //////////////////////////////////////////////////////////////////////
1152 MediaMultiCurl::MediaMultiCurl(const Url &url_r, const Pathname & attach_point_hint_r)
1153 : MediaCurl(url_r, attach_point_hint_r)
1155 MIL << "MediaMultiCurl::MediaMultiCurl(" << url_r << ", " << attach_point_hint_r << ")" << endl;
1157 _customHeadersMetalink = 0;
1160 MediaMultiCurl::~MediaMultiCurl()
1162 if (_customHeadersMetalink)
1164 curl_slist_free_all(_customHeadersMetalink);
1165 _customHeadersMetalink = 0;
1169 curl_multi_cleanup(_multi);
1172 std::map<std::string, CURL *>::iterator it;
1173 for (it = _easypool.begin(); it != _easypool.end(); it++)
1175 CURL *easy = it->second;
1178 curl_easy_cleanup(easy);
1184 void MediaMultiCurl::setupEasy()
1186 MediaCurl::setupEasy();
1188 if (_customHeadersMetalink)
1190 curl_slist_free_all(_customHeadersMetalink);
1191 _customHeadersMetalink = 0;
1193 struct curl_slist *sl = _customHeaders;
1194 for (; sl; sl = sl->next)
1195 _customHeadersMetalink = curl_slist_append(_customHeadersMetalink, sl->data);
1196 _customHeadersMetalink = curl_slist_append(_customHeadersMetalink, "Accept: */*, application/metalink+xml, application/metalink4+xml");
1199 static bool looks_like_metalink(const Pathname & file)
1203 if ((fd = open(file.asString().c_str(), O_RDONLY)) == -1)
1205 while ((l = read(fd, buf, sizeof(buf) - 1)) == -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;
1224 DBG << "looks_like_metalink(" << file << "): " << ret << endl;
1228 void MediaMultiCurl::doGetFileCopy( const Pathname & filename , const Pathname & target, callback::SendReport<DownloadProgressReport> & report, RequestOptions options ) const
1230 Pathname dest = target.absolutename();
1231 if( assert_dir( dest.dirname() ) )
1233 DBG << "assert_dir " << dest.dirname() << " failed" << endl;
1234 Url url(getFileUrl(filename));
1235 ZYPP_THROW( MediaSystemException(url, "System error on " + dest.dirname().asString()) );
1237 string destNew = target.asString() + ".new.zypp.XXXXXX";
1238 char *buf = ::strdup( destNew.c_str());
1241 ERR << "out of memory for temp file name" << endl;
1242 Url url(getFileUrl(filename));
1243 ZYPP_THROW(MediaSystemException(url, "out of memory for temp file name"));
1246 int tmp_fd = ::mkstemp( buf );
1250 ERR << "mkstemp failed for file '" << destNew << "'" << endl;
1251 ZYPP_THROW(MediaWriteException(destNew));
1256 FILE *file = ::fdopen( tmp_fd, "w" );
1259 filesystem::unlink( destNew );
1260 ERR << "fopen failed for file '" << destNew << "'" << endl;
1261 ZYPP_THROW(MediaWriteException(destNew));
1263 DBG << "dest: " << dest << endl;
1264 DBG << "temp: " << destNew << endl;
1266 // set IFMODSINCE time condition (no download if not modified)
1267 if( PathInfo(target).isExist() && !(options & OPTION_NO_IFMODSINCE) )
1269 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_IFMODSINCE);
1270 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, (long)PathInfo(target).mtime());
1274 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1275 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1277 // change header to include Accept: metalink
1278 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeadersMetalink);
1281 MediaCurl::doGetFileCopyFile(filename, dest, file, report, options);
1283 catch (Exception &ex)
1286 filesystem::unlink(destNew);
1287 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1288 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1289 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeaders);
1292 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1293 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1294 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeaders);
1295 long httpReturnCode = 0;
1296 CURLcode infoRet = curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &httpReturnCode);
1297 if (infoRet == CURLE_OK)
1299 DBG << "HTTP response: " + str::numstring(httpReturnCode) << endl;
1300 if ( httpReturnCode == 304
1301 || ( httpReturnCode == 213 && _url.getScheme() == "ftp" ) ) // not modified
1303 DBG << "not modified: " << PathInfo(dest) << endl;
1309 WAR << "Could not get the reponse code." << endl;
1312 bool ismetalink = false;
1315 if (curl_easy_getinfo(_curl, CURLINFO_CONTENT_TYPE, &ptr) == CURLE_OK && ptr)
1317 string ct = string(ptr);
1318 if (ct.find("application/metalink+xml") == 0 || ct.find("application/metalink4+xml") == 0)
1324 // some proxies do not store the content type, so also look at the file to find
1325 // out if we received a metalink (bnc#649925)
1327 if (looks_like_metalink(Pathname(destNew)))
1333 bool userabort = false;
1336 Pathname failedFile = ZConfig::instance().repoCachePath() / "MultiCurl.failed";
1340 mlp.parse(Pathname(destNew));
1341 MediaBlockList bl = mlp.getBlockList();
1342 vector<Url> urls = mlp.getUrls();
1344 file = fopen(destNew.c_str(), "w+");
1346 ZYPP_THROW(MediaWriteException(destNew));
1347 if (PathInfo(target).isExist())
1349 DBG << "reusing blocks from file " << target << endl;
1350 bl.reuseBlocks(file, target.asString());
1353 if (bl.haveChecksum(1) && PathInfo(failedFile).isExist())
1355 DBG << "reusing blocks from file " << failedFile << endl;
1356 bl.reuseBlocks(file, failedFile.asString());
1358 filesystem::unlink(failedFile);
1360 Pathname df = deltafile();
1363 DBG << "reusing blocks from file " << df << endl;
1364 bl.reuseBlocks(file, df.asString());
1369 multifetch(filename, file, &urls, &report, &bl);
1371 catch (MediaCurlException &ex)
1373 userabort = ex.errstr() == "User abort";
1377 catch (Exception &ex)
1379 // something went wrong. fall back to normal download
1383 if (PathInfo(destNew).size() >= 63336)
1385 ::unlink(failedFile.asString().c_str());
1386 filesystem::hardlinkCopy(destNew, failedFile);
1390 filesystem::unlink(destNew);
1393 file = fopen(destNew.c_str(), "w+");
1395 ZYPP_THROW(MediaWriteException(destNew));
1396 MediaCurl::doGetFileCopyFile(filename, dest, file, report, options | OPTION_NO_REPORT_START);
1400 if (::fchmod( ::fileno(file), filesystem::applyUmaskTo( 0644 )))
1402 ERR << "Failed to chmod file " << destNew << endl;
1406 filesystem::unlink(destNew);
1407 ERR << "Fclose failed for file '" << destNew << "'" << endl;
1408 ZYPP_THROW(MediaWriteException(destNew));
1410 if ( rename( destNew, dest ) != 0 )
1412 ERR << "Rename failed" << endl;
1413 ZYPP_THROW(MediaWriteException(dest));
1415 DBG << "done: " << PathInfo(dest) << endl;
1418 void MediaMultiCurl::multifetch(const Pathname & filename, FILE *fp, std::vector<Url> *urllist, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize) const
1420 Url baseurl(getFileUrl(filename));
1421 if (blklist && filesize == off_t(-1) && blklist->haveFilesize())
1422 filesize = blklist->getFilesize();
1423 if (blklist && !blklist->haveBlocks() && filesize != 0)
1425 if (blklist && (filesize == 0 || !blklist->numBlocks()))
1427 checkFileDigest(baseurl, fp, blklist);
1434 _multi = curl_multi_init();
1436 ZYPP_THROW(MediaCurlInitException(baseurl));
1438 multifetchrequest req(this, filename, baseurl, _multi, fp, report, blklist, filesize);
1439 req._timeout = _settings.timeout();
1440 req._connect_timeout = _settings.connectTimeout();
1441 req._maxspeed = _settings.maxDownloadSpeed();
1442 req._maxworkers = _settings.maxConcurrentConnections();
1443 if (req._maxworkers > MAXURLS)
1444 req._maxworkers = MAXURLS;
1445 if (req._maxworkers <= 0)
1446 req._maxworkers = 1;
1447 std::vector<Url> myurllist;
1448 for (std::vector<Url>::iterator urliter = urllist->begin(); urliter != urllist->end(); ++urliter)
1452 string scheme = urliter->getScheme();
1453 if (scheme == "http" || scheme == "https" || scheme == "ftp")
1455 checkProtocol(*urliter);
1456 myurllist.push_back(*urliter);
1463 if (!myurllist.size())
1464 myurllist.push_back(baseurl);
1466 checkFileDigest(baseurl, fp, blklist);
1469 void MediaMultiCurl::checkFileDigest(Url &url, FILE *fp, MediaBlockList *blklist) const
1471 if (!blklist || !blklist->haveFileChecksum())
1473 if (fseeko(fp, off_t(0), SEEK_SET))
1474 ZYPP_THROW(MediaCurlException(url, "fseeko", "seek error"));
1476 blklist->createFileDigest(dig);
1479 while ((l = fread(buf, 1, sizeof(buf), fp)) > 0)
1481 if (!blklist->verifyFileDigest(dig))
1482 ZYPP_THROW(MediaCurlException(url, "file verification failed", "checksum error"));
1485 bool MediaMultiCurl::isDNSok(const string &host) const
1487 return _dnsok.find(host) == _dnsok.end() ? false : true;
1490 void MediaMultiCurl::setDNSok(const string &host) const
1492 _dnsok.insert(host);
1495 CURL *MediaMultiCurl::fromEasyPool(const string &host) const
1497 if (_easypool.find(host) == _easypool.end())
1499 CURL *ret = _easypool[host];
1500 _easypool.erase(host);
1504 void MediaMultiCurl::toEasyPool(const std::string &host, CURL *easy) const
1506 CURL *oldeasy = _easypool[host];
1507 _easypool[host] = easy;
1509 curl_easy_cleanup(oldeasy);
1512 } // namespace media