1 /*---------------------------------------------------------------------\
3 | |__ / \ / / . \ . \ |
8 \---------------------------------------------------------------------*/
9 /** \file zypp/media/MediaMultiCurl.cc
14 #include <sys/types.h>
18 #include <arpa/inet.h>
25 #include "zypp/base/Logger.h"
26 #include "zypp/media/MediaMultiCurl.h"
27 #include "zypp/media/MetaLinkParser.h"
30 using namespace zypp::base;
36 //////////////////////////////////////////////////////////////////////
39 class multifetchrequest;
41 // Hack: we derive from MediaCurl just to get the storage space for
42 // settings, url, curlerrors and the like
44 class multifetchworker : MediaCurl {
45 friend class multifetchrequest;
48 multifetchworker(int no, multifetchrequest &request, const Url &url);
53 bool recheckChecksum();
54 void disableCompetition();
57 void adddnsfd(fd_set &rset, int &maxfd);
58 void dnsevent(fd_set &rset);
82 size_t writefunction(void *ptr, size_t size);
83 static size_t _writefunction(void *ptr, size_t size, size_t nmemb, void *stream);
85 size_t headerfunction(char *ptr, size_t size);
86 static size_t _headerfunction(void *ptr, size_t size, size_t nmemb, void *stream);
88 multifetchrequest *_request;
99 #define WORKER_STARTING 0
100 #define WORKER_LOOKUP 1
101 #define WORKER_FETCH 2
102 #define WORKER_DISCARD 3
103 #define WORKER_DONE 4
104 #define WORKER_SLEEP 5
105 #define WORKER_BROKEN 6
109 class multifetchrequest {
111 multifetchrequest(const MediaMultiCurl *context, const Pathname &filename, const Url &baseurl, CURLM *multi, FILE *fp, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize);
112 ~multifetchrequest();
114 void run(std::vector<Url> &urllist);
117 friend class multifetchworker;
119 const MediaMultiCurl *_context;
120 const Pathname _filename;
124 callback::SendReport<DownloadProgressReport> *_report;
125 MediaBlockList *_blklist;
130 std::list<multifetchworker *> _workers;
136 size_t _activeworkers;
137 size_t _lookupworkers;
138 size_t _sleepworkers;
139 double _minsleepuntil;
143 off_t _fetchedgoodsize;
146 double _lastprogress;
148 double _lastperiodstart;
149 double _lastperiodfetched;
154 double _connect_timeout;
158 #define BLKSIZE 131072
163 //////////////////////////////////////////////////////////////////////
169 if (gettimeofday(&tv, NULL))
171 return tv.tv_sec + tv.tv_usec / 1000000.;
175 multifetchworker::writefunction(void *ptr, size_t size)
178 if (_state == WORKER_BROKEN)
181 double now = currentTime();
183 len = size > _size ? _size : size;
193 _request->_lastprogress = now;
195 if (_state == WORKER_DISCARD || !_request->_fp)
197 // block is no longer needed
198 // still calculate the checksum so that we can throw out bad servers
199 if (_request->_blklist)
200 _dig.update((const char *)ptr, len);
205 if (fseeko(_request->_fp, _off, SEEK_SET))
207 cnt = fwrite(ptr, 1, len, _request->_fp);
210 _request->_fetchedsize += cnt;
211 if (_request->_blklist)
212 _dig.update((const char *)ptr, cnt);
222 multifetchworker::_writefunction(void *ptr, size_t size, size_t nmemb, void *stream)
224 multifetchworker *me = reinterpret_cast<multifetchworker *>(stream);
225 return me->writefunction(ptr, size * nmemb);
229 multifetchworker::headerfunction(char *p, size_t size)
232 if (l > 9 && !strncasecmp(p, "Location:", 9))
234 string line(p + 9, l - 9);
235 if (line[l - 10] == '\r')
236 line.erase(l - 10, 1);
237 DBG << "#" << _workerno << ": redirecting to" << line << endl;
240 if (l <= 14 || l >= 128 || strncasecmp(p, "Content-Range:", 14) != 0)
244 while (l && (*p == ' ' || *p == '\t'))
246 if (l < 6 || strncasecmp(p, "bytes", 5))
253 unsigned long long start, off, filesize;
254 if (sscanf(buf, "%llu-%llu/%llu", &start, &off, &filesize) != 3)
256 if (_request->_filesize == (off_t)-1)
258 WAR << "#" << _workerno << ": setting request filesize to " << filesize << endl;
259 _request->_filesize = filesize;
261 if (_request->_filesize != (off_t)filesize)
263 DBG << "#" << _workerno << ": filesize mismatch" << endl;
264 _state = WORKER_BROKEN;
265 strncpy(_curlError, "filesize mismatch", CURL_ERROR_SIZE);
271 multifetchworker::_headerfunction(void *ptr, size_t size, size_t nmemb, void *stream)
273 multifetchworker *me = reinterpret_cast<multifetchworker *>(stream);
274 return me->headerfunction((char *)ptr, size * nmemb);
277 multifetchworker::multifetchworker(int no, multifetchrequest &request, const Url &url)
278 : MediaCurl(url, Pathname())
282 _state = WORKER_STARTING;
284 _off = _blkstart = 0;
285 _size = _blksize = 0;
295 _maxspeed = _request->_maxspeed;
299 curlUrl.setUsername( "" );
300 curlUrl.setPassword( "" );
301 curlUrl.setPathParams( "" );
302 curlUrl.setQueryString( "" );
303 curlUrl.setFragment( "" );
304 _urlbuf = curlUrl.asString();
305 _curl = _request->_context->fromEasyPool(_url.getHost());
307 DBG << "reused worker from pool" << endl;
308 if (!_curl && !(_curl = curl_easy_init()))
310 _state = WORKER_BROKEN;
311 strncpy(_curlError, "curl_easy_init failed", CURL_ERROR_SIZE);
318 catch (Exception &ex)
320 curl_easy_cleanup(_curl);
322 _state = WORKER_BROKEN;
323 strncpy(_curlError, "curl_easy_setopt failed", CURL_ERROR_SIZE);
326 curl_easy_setopt(_curl, CURLOPT_PRIVATE, this);
327 curl_easy_setopt(_curl, CURLOPT_URL, _urlbuf.c_str());
328 curl_easy_setopt(_curl, CURLOPT_WRITEFUNCTION, &_writefunction);
329 curl_easy_setopt(_curl, CURLOPT_WRITEDATA, this);
330 if (_request->_filesize == off_t(-1) || !_request->_blklist || !_request->_blklist->haveChecksum(0))
332 curl_easy_setopt(_curl, CURLOPT_HEADERFUNCTION, &_headerfunction);
333 curl_easy_setopt(_curl, CURLOPT_HEADERDATA, this);
335 // if this is the same host copy authorization
336 // (the host check is also what curl does when doing a redirect)
337 // (note also that unauthorized exceptions are thrown with the request host)
338 if (url.getHost() == _request->_context->_url.getHost())
340 _settings.setUsername(_request->_context->_settings.username());
341 _settings.setPassword(_request->_context->_settings.password());
342 _settings.setAuthType(_request->_context->_settings.authType());
343 if ( _settings.userPassword().size() )
345 curl_easy_setopt(_curl, CURLOPT_USERPWD, _settings.userPassword().c_str());
346 string use_auth = _settings.authType();
347 if (use_auth.empty())
348 use_auth = "digest,basic"; // our default
349 long auth = CurlAuthData::auth_type_str2long(use_auth);
350 if( auth != CURLAUTH_NONE)
352 DBG << "#" << _workerno << ": Enabling HTTP authentication methods: " << use_auth
353 << " (CURLOPT_HTTPAUTH=" << auth << ")" << std::endl;
354 curl_easy_setopt(_curl, CURLOPT_HTTPAUTH, auth);
361 multifetchworker::~multifetchworker()
365 if (_state == WORKER_FETCH || _state == WORKER_DISCARD)
366 curl_multi_remove_handle(_request->_multi, _curl);
367 if (_state == WORKER_DONE || _state == WORKER_SLEEP)
369 curl_easy_setopt(_curl, CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t)0);
370 curl_easy_setopt(_curl, CURLOPT_PRIVATE, (void *)0);
371 curl_easy_setopt(_curl, CURLOPT_WRITEFUNCTION, (void *)0);
372 curl_easy_setopt(_curl, CURLOPT_WRITEDATA, (void *)0);
373 curl_easy_setopt(_curl, CURLOPT_HEADERFUNCTION, (void *)0);
374 curl_easy_setopt(_curl, CURLOPT_HEADERDATA, (void *)0);
375 _request->_context->toEasyPool(_url.getHost(), _curl);
378 curl_easy_cleanup(_curl);
385 while (waitpid(_pid, &status, 0) == -1)
395 // the destructor in MediaCurl doesn't call disconnect() if
396 // the media is not attached, so we do it here manually
400 static inline bool env_isset(string name)
402 const char *s = getenv(name.c_str());
403 return s && *s ? true : false;
407 multifetchworker::checkdns()
409 string host = _url.getHost();
414 if (_request->_context->isDNSok(host))
417 // no need to do dns checking for numeric hosts
419 if (inet_pton(AF_INET, host.c_str(), addrbuf) == 1)
421 if (inet_pton(AF_INET6, host.c_str(), addrbuf) == 1)
424 // no need to do dns checking if we use a proxy
425 if (!_settings.proxy().empty())
427 if (env_isset("all_proxy") || env_isset("ALL_PROXY"))
429 string schemeproxy = _url.getScheme() + "_proxy";
430 if (env_isset(schemeproxy))
432 if (schemeproxy != "http_proxy")
434 std::transform(schemeproxy.begin(), schemeproxy.end(), schemeproxy.begin(), ::toupper);
435 if (env_isset(schemeproxy))
439 DBG << "checking DNS lookup of " << host << endl;
443 _state = WORKER_BROKEN;
444 strncpy(_curlError, "DNS pipe creation failed", CURL_ERROR_SIZE);
448 if (_pid == pid_t(-1))
453 _state = WORKER_BROKEN;
454 strncpy(_curlError, "DNS checker fork failed", CURL_ERROR_SIZE);
460 // XXX: close all other file descriptors
461 struct addrinfo *ai, aihints;
462 memset(&aihints, 0, sizeof(aihints));
463 aihints.ai_family = PF_UNSPEC;
464 int tstsock = socket(PF_INET6, SOCK_DGRAM, 0);
466 aihints.ai_family = PF_INET;
469 aihints.ai_socktype = SOCK_STREAM;
470 aihints.ai_flags = AI_CANONNAME;
471 unsigned int connecttimeout = _request->_connect_timeout;
473 alarm(connecttimeout);
474 signal(SIGALRM, SIG_DFL);
475 if (getaddrinfo(host.c_str(), NULL, &aihints, &ai))
480 _dnspipe = pipefds[0];
481 _state = WORKER_LOOKUP;
485 multifetchworker::adddnsfd(fd_set &rset, int &maxfd)
487 if (_state != WORKER_LOOKUP)
489 FD_SET(_dnspipe, &rset);
490 if (maxfd < _dnspipe)
495 multifetchworker::dnsevent(fd_set &rset)
498 if (_state != WORKER_LOOKUP || !FD_ISSET(_dnspipe, &rset))
501 while (waitpid(_pid, &status, 0) == -1)
512 if (!WIFEXITED(status))
514 _state = WORKER_BROKEN;
515 strncpy(_curlError, "DNS lookup failed", CURL_ERROR_SIZE);
516 _request->_activeworkers--;
519 int exitcode = WEXITSTATUS(status);
520 DBG << "#" << _workerno << ": DNS lookup returned " << exitcode << endl;
523 _state = WORKER_BROKEN;
524 strncpy(_curlError, "DNS lookup failed", CURL_ERROR_SIZE);
525 _request->_activeworkers--;
528 _request->_context->setDNSok(_url.getHost());
533 multifetchworker::checkChecksum()
535 // DBG << "checkChecksum block " << _blkno << endl;
536 if (!_blksize || !_request->_blklist)
538 return _request->_blklist->verifyDigest(_blkno, _dig);
542 multifetchworker::recheckChecksum()
544 // DBG << "recheckChecksum block " << _blkno << endl;
545 if (!_request->_fp || !_blksize || !_request->_blklist)
547 if (fseeko(_request->_fp, _blkstart, SEEK_SET))
551 _request->_blklist->createDigest(_dig); // resets digest
554 size_t cnt = l > sizeof(buf) ? sizeof(buf) : l;
555 if (fread(buf, cnt, 1, _request->_fp) != 1)
557 _dig.update(buf, cnt);
560 return _request->_blklist->verifyDigest(_blkno, _dig);
565 multifetchworker::stealjob()
567 if (!_request->_stealing)
569 DBG << "start stealing!" << endl;
570 _request->_stealing = true;
572 multifetchworker *best = 0;
573 std::list<multifetchworker *>::iterator workeriter = _request->_workers.begin();
575 for (; workeriter != _request->_workers.end(); ++workeriter)
577 multifetchworker *worker = *workeriter;
580 if (worker->_pass == -1)
581 continue; // do not steal!
582 if (worker->_state == WORKER_DISCARD || worker->_state == WORKER_DONE || worker->_state == WORKER_SLEEP || !worker->_blksize)
583 continue; // do not steal finished jobs
584 if (!worker->_avgspeed && worker->_blkreceived)
588 if (now > worker->_blkstarttime)
589 worker->_avgspeed = worker->_blkreceived / (now - worker->_blkstarttime);
591 if (!best || best->_pass > worker->_pass)
596 if (best->_pass < worker->_pass)
598 // if it is the same block, we want to know the best worker, otherwise the worst
599 if (worker->_blkstart == best->_blkstart)
601 if ((worker->_blksize - worker->_blkreceived) * best->_avgspeed < (best->_blksize - best->_blkreceived) * worker->_avgspeed)
606 if ((worker->_blksize - worker->_blkreceived) * best->_avgspeed > (best->_blksize - best->_blkreceived) * worker->_avgspeed)
612 _state = WORKER_DONE;
613 _request->_activeworkers--;
614 _request->_finished = true;
617 // do not sleep twice
618 if (_state != WORKER_SLEEP)
620 if (!_avgspeed && _blkreceived)
624 if (now > _blkstarttime)
625 _avgspeed = _blkreceived / (now - _blkstarttime);
628 // lets see if we should sleep a bit
629 DBG << "me #" << _workerno << ": " << _avgspeed << ", size " << best->_blksize << endl;
630 DBG << "best #" << best->_workerno << ": " << best->_avgspeed << ", size " << (best->_blksize - best->_blkreceived) << endl;
631 if (_avgspeed && best->_avgspeed && (best->_blksize - best->_blkreceived) * _avgspeed < best->_blksize * best->_avgspeed)
635 double sl = (best->_blksize - best->_blkreceived) / best->_avgspeed * 2;
638 DBG << "#" << _workerno << ": going to sleep for " << sl * 1000 << " ms" << endl;
639 _sleepuntil = now + sl;
640 _state = WORKER_SLEEP;
641 _request->_sleepworkers++;
647 best->_competing = true;
648 _blkstart = best->_blkstart;
649 _blksize = best->_blksize;
652 _blkno = best->_blkno;
657 multifetchworker::disableCompetition()
659 std::list<multifetchworker *>::iterator workeriter = _request->_workers.begin();
660 for (; workeriter != _request->_workers.end(); ++workeriter)
662 multifetchworker *worker = *workeriter;
665 if (worker->_blkstart == _blkstart)
667 if (worker->_state == WORKER_FETCH)
668 worker->_state = WORKER_DISCARD;
669 worker->_pass = -1; /* do not steal this one, we already have it */
676 multifetchworker::nextjob()
679 if (_request->_stealing)
685 MediaBlockList *blklist = _request->_blklist;
689 if (_request->_filesize != off_t(-1))
691 if (_request->_blkoff >= _request->_filesize)
696 _blksize = _request->_filesize - _request->_blkoff;
697 if (_blksize > BLKSIZE)
703 MediaBlock blk = blklist->getBlock(_request->_blkno);
704 while (_request->_blkoff >= blk.off + blk.size)
706 if (++_request->_blkno == blklist->numBlocks())
711 blk = blklist->getBlock(_request->_blkno);
712 _request->_blkoff = blk.off;
714 _blksize = blk.off + blk.size - _request->_blkoff;
715 if (_blksize > BLKSIZE && !blklist->haveChecksum(_request->_blkno))
718 _blkno = _request->_blkno;
719 _blkstart = _request->_blkoff;
720 _request->_blkoff += _blksize;
725 multifetchworker::run()
729 if (_state == WORKER_BROKEN || _state == WORKER_DONE)
730 return; // just in case...
732 sprintf(rangebuf, "%llu-", (unsigned long long)_blkstart);
734 sprintf(rangebuf, "%llu-%llu", (unsigned long long)_blkstart, (unsigned long long)_blkstart + _blksize - 1);
735 DBG << "#" << _workerno << ": BLK " << _blkno << ":" << rangebuf << " " << _url << endl;
736 if (curl_easy_setopt(_curl, CURLOPT_RANGE, !_noendrange || _blkstart != 0 ? rangebuf : (char *)0) != CURLE_OK)
738 _request->_activeworkers--;
739 _state = WORKER_BROKEN;
740 strncpy(_curlError, "curl_easy_setopt range failed", CURL_ERROR_SIZE);
743 if (curl_multi_add_handle(_request->_multi, _curl) != CURLM_OK)
745 _request->_activeworkers--;
746 _state = WORKER_BROKEN;
747 strncpy(_curlError, "curl_multi_add_handle failed", CURL_ERROR_SIZE);
750 _request->_havenewjob = true;
753 if (_request->_blklist)
754 _request->_blklist->createDigest(_dig); // resets digest
755 _state = WORKER_FETCH;
757 double now = currentTime();
763 //////////////////////////////////////////////////////////////////////
766 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)
771 _filesize = filesize;
777 _blkoff = _blklist->getBlock(0).off;
786 _fetchedgoodsize = 0;
788 _lastperiodstart = _lastprogress = _starttime = currentTime();
789 _lastperiodfetched = 0;
792 _connect_timeout = 0;
796 for (size_t blkno = 0; blkno < blklist->numBlocks(); blkno++)
798 MediaBlock blk = blklist->getBlock(blkno);
799 _totalsize += blk.size;
802 else if (filesize != off_t(-1))
803 _totalsize = filesize;
806 multifetchrequest::~multifetchrequest()
808 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
810 multifetchworker *worker = *workeriter;
818 multifetchrequest::run(std::vector<Url> &urllist)
821 std::vector<Url>::iterator urliter = urllist.begin();
824 fd_set rset, wset, xset;
829 DBG << "finished!" << endl;
833 if (_activeworkers < MAXWORKERS && urliter != urllist.end() && _workers.size() < MAXURLS)
835 // spawn another worker!
836 multifetchworker *worker = new multifetchworker(workerno++, *this, *urliter);
837 _workers.push_back(worker);
838 if (worker->_state != WORKER_BROKEN)
841 if (worker->_state != WORKER_LOOKUP)
853 WAR << "No more active workers!" << endl;
854 // show the first worker error we find
855 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
857 if ((*workeriter)->_state != WORKER_BROKEN)
859 ZYPP_THROW(MediaCurlException(_baseurl, "Server error", (*workeriter)->_curlError));
868 curl_multi_fdset(_multi, &rset, &wset, &xset, &maxfd);
871 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
872 (*workeriter)->adddnsfd(rset, maxfd);
875 // if we added a new job we have to call multi_perform once
876 // to make it show up in the fd set. do not sleep in this case.
878 tv.tv_usec = _havenewjob ? 0 : 200000;
879 if (_sleepworkers && !_havenewjob)
881 if (_minsleepuntil == 0)
883 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
885 multifetchworker *worker = *workeriter;
886 if (worker->_state != WORKER_SLEEP)
888 if (!_minsleepuntil || _minsleepuntil > worker->_sleepuntil)
889 _minsleepuntil = worker->_sleepuntil;
892 double sl = _minsleepuntil - currentTime();
899 tv.tv_usec = sl * 1000000;
901 int r = select(maxfd + 1, &rset, &wset, &xset, &tv);
902 if (r == -1 && errno != EINTR)
903 ZYPP_THROW(MediaCurlException(_baseurl, "select() failed", "unknown error"));
904 if (r != 0 && _lookupworkers)
905 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
907 multifetchworker *worker = *workeriter;
908 if (worker->_state != WORKER_LOOKUP)
910 (*workeriter)->dnsevent(rset);
911 if (worker->_state != WORKER_LOOKUP)
921 mcode = curl_multi_perform(_multi, &tasks);
922 if (mcode == CURLM_CALL_MULTI_PERFORM)
924 if (mcode != CURLM_OK)
925 ZYPP_THROW(MediaCurlException(_baseurl, "curl_multi_perform", "unknown error"));
929 double now = currentTime();
932 if (now > _lastperiodstart + .5)
935 _periodavg = (_fetchedsize - _lastperiodfetched) / (now - _lastperiodstart);
937 _periodavg = (_periodavg + (_fetchedsize - _lastperiodfetched) / (now - _lastperiodstart)) / 2;
938 _lastperiodfetched = _fetchedsize;
939 _lastperiodstart = now;
945 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
947 multifetchworker *worker = *workeriter;
948 if (worker->_state != WORKER_SLEEP)
950 if (worker->_sleepuntil > now)
952 if (_minsleepuntil == worker->_sleepuntil)
954 DBG << "#" << worker->_workerno << ": sleep done, wake up" << endl;
956 // nextjob chnages the state
961 // collect all curl results, reschedule new jobs
963 while ((msg = curl_multi_info_read(_multi, &nqueue)) != 0)
965 if (msg->msg != CURLMSG_DONE)
967 multifetchworker *worker;
968 if (curl_easy_getinfo(msg->easy_handle, CURLINFO_PRIVATE, &worker) != CURLE_OK)
969 ZYPP_THROW(MediaCurlException(_baseurl, "curl_easy_getinfo", "unknown error"));
970 CURLcode cc = msg->data.result;
971 if (worker->_blkreceived && now > worker->_blkstarttime)
973 if (worker->_avgspeed)
974 worker->_avgspeed = (worker->_avgspeed + worker->_blkreceived / (now - worker->_blkstarttime)) / 2;
976 worker->_avgspeed = worker->_blkreceived / (now - worker->_blkstarttime);
978 DBG << "#" << worker->_workerno << ": BLK " << worker->_blkno << " done code " << cc << " speed " << worker->_avgspeed << endl;
979 curl_multi_remove_handle(_multi, msg->easy_handle);
980 if (cc == CURLE_HTTP_RETURNED_ERROR)
983 (void)curl_easy_getinfo(msg->easy_handle, CURLINFO_RESPONSE_CODE, &statuscode);
984 DBG << "HTTP status " << statuscode << endl;
985 if (statuscode == 416 && !_blklist) /* Range error */
987 if (_filesize == off_t(-1))
989 if (!worker->_noendrange)
991 DBG << "#" << worker->_workerno << ": retrying with no end range" << endl;
992 worker->_noendrange = true;
996 worker->_noendrange = false;
1000 if (worker->_blkstart >= _filesize)
1009 if (!worker->checkChecksum())
1011 WAR << "#" << worker->_workerno << ": checksum error, disable worker" << endl;
1012 worker->_state = WORKER_BROKEN;
1013 strncpy(worker->_curlError, "checksum error", CURL_ERROR_SIZE);
1017 if (worker->_state == WORKER_FETCH)
1019 if (worker->_competing)
1021 worker->disableCompetition();
1022 // multiple workers wrote into this block. We already know that our
1023 // data was correct, but maybe some other worker overwrote our data
1024 // with something broken. Thus we have to re-check the block.
1025 if (!worker->recheckChecksum())
1027 DBG << "#" << worker->_workerno << ": recheck checksum error, refetch block" << endl;
1028 // re-fetch! No need to worry about the bad workers,
1029 // they will now be set to DISCARD. At the end of their block
1030 // they will notice that they wrote bad data and go into BROKEN.
1035 _fetchedgoodsize += worker->_blksize;
1038 // make bad workers sleep a little
1040 int maxworkerno = 0;
1042 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
1044 multifetchworker *oworker = *workeriter;
1045 if (oworker->_state == WORKER_BROKEN)
1047 if (oworker->_avgspeed > maxavg)
1049 maxavg = oworker->_avgspeed;
1050 maxworkerno = oworker->_workerno;
1052 if (oworker->_avgspeed > worker->_avgspeed)
1055 if (maxavg && !_stealing)
1057 double ratio = worker->_avgspeed / maxavg;
1059 if (numbetter < 3) // don't sleep that much if we're in the top two
1060 ratio = ratio * ratio;
1063 DBG << "#" << worker->_workerno << ": too slow ("<< ratio << ", " << worker->_avgspeed << ", #" << maxworkerno << ": " << maxavg << "), going to sleep for " << ratio * 1000 << " ms" << endl;
1064 worker->_sleepuntil = now + ratio;
1065 worker->_state = WORKER_SLEEP;
1071 // do rate control (if requested)
1072 // should use periodavg, but that's not what libcurl does
1073 if (_maxspeed && now > _starttime)
1075 double avg = _fetchedsize / (now - _starttime);
1076 avg = worker->_maxspeed * _maxspeed / avg;
1077 if (avg < _maxspeed / MAXWORKERS)
1078 avg = _maxspeed / MAXWORKERS;
1079 if (avg > _maxspeed)
1083 worker->_maxspeed = avg;
1084 curl_easy_setopt(worker->_curl, CURLOPT_MAX_RECV_SPEED_LARGE, (curl_off_t)(avg));
1091 worker->_state = WORKER_BROKEN;
1093 if (!_activeworkers && !(urliter != urllist.end() && _workers.size() < MAXURLS))
1095 // end of workers reached! goodbye!
1096 worker->evaluateCurlCode(Pathname(), cc, false);
1104 int percent = _totalsize ? (100 * (_fetchedgoodsize + _fetchedsize)) / (_totalsize + _fetchedsize) : 0;
1106 if (now > _starttime)
1107 avg = _fetchedsize / (now - _starttime);
1108 if (!(*(_report))->progress(percent, _baseurl, avg, _lastperiodstart == _starttime ? avg : _periodavg))
1109 ZYPP_THROW(MediaCurlException(_baseurl, "User abort", "cancelled"));
1112 if (_timeout && now - _lastprogress > _timeout)
1117 ZYPP_THROW(MediaTimeoutException(_baseurl));
1119 // print some download stats
1120 WAR << "overall result" << endl;
1121 for (std::list<multifetchworker *>::iterator workeriter = _workers.begin(); workeriter != _workers.end(); ++workeriter)
1123 multifetchworker *worker = *workeriter;
1124 WAR << "#" << worker->_workerno << ": state: " << worker->_state << " received: " << worker->_received << " url: " << worker->_url << endl;
1129 //////////////////////////////////////////////////////////////////////
1132 MediaMultiCurl::MediaMultiCurl(const Url &url_r, const Pathname & attach_point_hint_r)
1133 : MediaCurl(url_r, attach_point_hint_r)
1135 MIL << "MediaMultiCurl::MediaMultiCurl(" << url_r << ", " << attach_point_hint_r << ")" << endl;
1137 _customHeadersMetalink = 0;
1140 MediaMultiCurl::~MediaMultiCurl()
1142 if (_customHeadersMetalink)
1144 curl_slist_free_all(_customHeadersMetalink);
1145 _customHeadersMetalink = 0;
1149 curl_multi_cleanup(_multi);
1152 std::map<std::string, CURL *>::iterator it;
1153 for (it = _easypool.begin(); it != _easypool.end(); it++)
1155 CURL *easy = it->second;
1158 curl_easy_cleanup(easy);
1164 void MediaMultiCurl::setupEasy()
1166 MediaCurl::setupEasy();
1168 if (_customHeadersMetalink)
1170 curl_slist_free_all(_customHeadersMetalink);
1171 _customHeadersMetalink = 0;
1173 struct curl_slist *sl = _customHeaders;
1174 for (; sl; sl = sl->next)
1175 _customHeadersMetalink = curl_slist_append(_customHeadersMetalink, sl->data);
1176 _customHeadersMetalink = curl_slist_append(_customHeadersMetalink, "Accept: */*, application/metalink+xml, application/metalink4+xml");
1180 void MediaMultiCurl::doGetFileCopy( const Pathname & filename , const Pathname & target, callback::SendReport<DownloadProgressReport> & report, RequestOptions options ) const
1182 Pathname dest = target.absolutename();
1183 if( assert_dir( dest.dirname() ) )
1185 DBG << "assert_dir " << dest.dirname() << " failed" << endl;
1186 Url url(getFileUrl(filename));
1187 ZYPP_THROW( MediaSystemException(url, "System error on " + dest.dirname().asString()) );
1189 string destNew = target.asString() + ".new.zypp.XXXXXX";
1190 char *buf = ::strdup( destNew.c_str());
1193 ERR << "out of memory for temp file name" << endl;
1194 Url url(getFileUrl(filename));
1195 ZYPP_THROW(MediaSystemException(url, "out of memory for temp file name"));
1198 int tmp_fd = ::mkstemp( buf );
1202 ERR << "mkstemp failed for file '" << destNew << "'" << endl;
1203 ZYPP_THROW(MediaWriteException(destNew));
1208 FILE *file = ::fdopen( tmp_fd, "w" );
1211 filesystem::unlink( destNew );
1212 ERR << "fopen failed for file '" << destNew << "'" << endl;
1213 ZYPP_THROW(MediaWriteException(destNew));
1215 DBG << "dest: " << dest << endl;
1216 DBG << "temp: " << destNew << endl;
1218 // set IFMODSINCE time condition (no download if not modified)
1219 if( PathInfo(target).isExist() && !(options & OPTION_NO_IFMODSINCE) )
1221 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_IFMODSINCE);
1222 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, (long)PathInfo(target).mtime());
1226 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1227 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1229 // change header to include Accept: metalink
1230 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeadersMetalink);
1233 MediaCurl::doGetFileCopyFile(filename, dest, file, report, options);
1235 catch (Exception &ex)
1238 filesystem::unlink(destNew);
1239 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1240 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1241 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeaders);
1244 curl_easy_setopt(_curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_NONE);
1245 curl_easy_setopt(_curl, CURLOPT_TIMEVALUE, 0L);
1246 curl_easy_setopt(_curl, CURLOPT_HTTPHEADER, _customHeaders);
1247 long httpReturnCode = 0;
1248 CURLcode infoRet = curl_easy_getinfo(_curl, CURLINFO_RESPONSE_CODE, &httpReturnCode);
1249 if (infoRet == CURLE_OK)
1251 DBG << "HTTP response: " + str::numstring(httpReturnCode) << endl;
1252 if ( httpReturnCode == 304
1253 || ( httpReturnCode == 213 && _url.getScheme() == "ftp" ) ) // not modified
1255 DBG << "not modified: " << PathInfo(dest) << endl;
1261 WAR << "Could not get the reponse code." << endl;
1264 if (curl_easy_getinfo(_curl, CURLINFO_CONTENT_TYPE, &ptr) == CURLE_OK && ptr)
1266 string ct = string(ptr);
1267 if (ct.find("application/metalink+xml") == 0 || ct.find("application/metalink4+xml") == 0)
1269 bool userabort = false;
1276 MediaBlockList bl = mlp.getBlockList();
1277 vector<Url> urls = mlp.getUrls();
1279 file = fopen(destNew.c_str(), "w+");
1281 ZYPP_THROW(MediaWriteException(destNew));
1282 if (PathInfo(target).isExist())
1284 DBG << "reusing blocks from file " << target << endl;
1285 bl.reuseBlocks(file, target.asString());
1288 Pathname df = deltafile();
1291 DBG << "reusing blocks from file " << df << endl;
1292 bl.reuseBlocks(file, df.asString());
1297 multifetch(filename, file, &urls, &report, &bl);
1299 catch (MediaCurlException &ex)
1302 userabort = ex.errstr() == "User abort";
1306 catch (Exception &ex)
1308 // something went wrong. fall back to normal download
1311 filesystem::unlink(destNew);
1314 MediaCurl::doGetFileCopyFile(filename, dest, file, report, options | OPTION_NO_REPORT_START);
1318 if (::fchmod( ::fileno(file), filesystem::applyUmaskTo( 0644 )))
1320 ERR << "Failed to chmod file " << destNew << endl;
1324 filesystem::unlink(destNew);
1325 ERR << "Fclose failed for file '" << destNew << "'" << endl;
1326 ZYPP_THROW(MediaWriteException(destNew));
1328 if ( rename( destNew, dest ) != 0 )
1330 ERR << "Rename failed" << endl;
1331 ZYPP_THROW(MediaWriteException(dest));
1333 DBG << "done: " << PathInfo(dest) << endl;
1336 void MediaMultiCurl::multifetch(const Pathname & filename, FILE *fp, std::vector<Url> *urllist, callback::SendReport<DownloadProgressReport> *report, MediaBlockList *blklist, off_t filesize) const
1338 Url baseurl(getFileUrl(filename));
1339 if (blklist && filesize == off_t(-1) && blklist->haveFilesize())
1340 filesize = blklist->getFilesize();
1341 if (blklist && !blklist->haveBlocks() && filesize != 0)
1343 if (blklist && (filesize == 0 || !blklist->numBlocks()))
1345 checkFileDigest(baseurl, fp, blklist);
1352 _multi = curl_multi_init();
1354 ZYPP_THROW(MediaCurlInitException(baseurl));
1356 multifetchrequest req(this, filename, baseurl, _multi, fp, report, blklist, filesize);
1357 req._timeout = _settings.timeout();
1358 req._connect_timeout = _settings.connectTimeout();
1359 req._maxspeed = _settings.maxDownloadSpeed();
1360 std::vector<Url> myurllist;
1361 for (std::vector<Url>::iterator urliter = urllist->begin(); urliter != urllist->end(); ++urliter)
1365 string scheme = urliter->getScheme();
1366 if (scheme == "http" || scheme == "https" || scheme == "ftp")
1368 checkProtocol(*urliter);
1369 myurllist.push_back(*urliter);
1376 if (!myurllist.size())
1377 myurllist.push_back(baseurl);
1379 checkFileDigest(baseurl, fp, blklist);
1382 void MediaMultiCurl::checkFileDigest(Url &url, FILE *fp, MediaBlockList *blklist) const
1384 if (!blklist || !blklist->haveFileChecksum())
1386 if (fseeko(fp, off_t(0), SEEK_SET))
1387 ZYPP_THROW(MediaCurlException(url, "fseeko", "seek error"));
1389 blklist->createFileDigest(dig);
1392 while ((l = fread(buf, 1, sizeof(buf), fp)) > 0)
1394 if (!blklist->verifyFileDigest(dig))
1395 ZYPP_THROW(MediaCurlException(url, "file verification failed", "checksum error"));
1398 bool MediaMultiCurl::isDNSok(const string &host) const
1400 return _dnsok.find(host) == _dnsok.end() ? false : true;
1403 void MediaMultiCurl::setDNSok(const string &host) const
1405 _dnsok.insert(host);
1408 CURL *MediaMultiCurl::fromEasyPool(const string &host) const
1410 if (_easypool.find(host) == _easypool.end())
1412 CURL *ret = _easypool[host];
1413 _easypool.erase(host);
1417 void MediaMultiCurl::toEasyPool(const std::string &host, CURL *easy) const
1419 CURL *oldeasy = _easypool[host];
1420 _easypool[host] = easy;
1422 curl_easy_cleanup(oldeasy);
1425 } // namespace media