(Sorry for the ridiculously large check-in, wants to fix lots of stuff.)
Other Changes:
* Added SSL Test Harness. (authssltest)
* Added GPG Test Harness. (authssltest)
* Reworked SSL/GPG startup to enable test harnesses
* pqiperson: added notify of FAIL if connection method unavailable.
* added p3tunnel to pqissltunnel init.
* pqimonitor: added ipHistory to connect callback.
p3ConnectMgr Changes:
* removed STUN system - wasn't operating correctly anyway without DHT.
* switched to new IpAddress history data types. (removed lots of code)
* Added Addr history for both Local and Ext addresses.
* removed p3tunnel references in p3connmgr
* fixed up mUseTunnelConnection flags (now used!)
* fixed up mUseExtFinder flags (now used!)
* added improved Net Status.
* corrected UPNP / EXT / UDP startup. (was always UPNP).
* fixed netReset() and netStartup().
* removed unnecessary DEBUG code.
* added UPNP timeout code (600 secs - why does it take so long?)
* added improved netExtCheck().
* removed wierd netConsistencyCheck() ... to rework.
* corrected connect / reconnect code.
* removed DHT notify code.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3247 b45a01b8-16f6-495d-af2f-9b41ad6348cc
* Removed nasty list of ip addresses from rsPeerNetItem
* create TLV type to handle ip address history instead.
* switched to new data type. (should be seemless?)
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3246 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- stores an iterator rather than searching a map twice
- deletes a message when it can't be routed in routeGenericTunnelItem()
- deletes a message when its contents allocation failed in sendFileData()
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3244 b45a01b8-16f6-495d-af2f-9b41ad6348cc
Every call to RsInit::LoadCertificates() now creates a file:
~/.retroshare/xxxxxxxxxxxxxxxxxxxx/lock
which is then bound to a system lock (fcntl F_SETLK).
If the lock request fails, it means another instance is already
running with the same profile.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3241 b45a01b8-16f6-495d-af2f-9b41ad6348cc
Rework of AuthSSL: cleanup and split up.
* Split out independent SSL functions into sslfns.h / sslfns.cc
* Reworked SSL certificate storage.
* Reworked SignDataBin / VerifyDataBin (fixed memory leaks).
* Removed funny code: /* cert->cert_info->key->pkey is NULL */
- just use X509_get_pubkey() instead.
* Removed lots of old code.
* Fixed up Mutex usage in AuthSSL - which was random.
* Removed certificates from tlvSignature serialiser obj.
* removed certificates from p3distrib messages.
* Starting removing "unused parameter" compiler warnings.
* Various related changes to make libretroshare compile.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3222 b45a01b8-16f6-495d-af2f-9b41ad6348cc
* restored AuthSSL::Failed/CheckCertificate() calls.
* removed mConnMgr->isConnected() checks. These should be elsewhere.
* enabled exit(1) call if cannot open listening socket.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3218 b45a01b8-16f6-495d-af2f-9b41ad6348cc
p3connmgr: This needs major help... and will be revisited.
Only changes to get code compiling for now.
* revert peerStatus() function signatures (to match pqimonitor.h)
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3217 b45a01b8-16f6-495d-af2f-9b41ad6348cc
This functionality might need to be restored later,
but should be done with list<> for both Local and Remote Addrs.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3215 b45a01b8-16f6-495d-af2f-9b41ad6348cc
AuthSSL / AuthGPG: Bits and pieces still need fixing up.
AuthSSL will be rechecked later ;)
* Mutex bug in authGPG (not very serious one).
* removed _locked() suffix from functions (as they do the locking now).
* added dummy FailedCertificate / CheckCertificate fns to AuthSSL
* removed last bits of awful SSL_ex_data hack.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3213 b45a01b8-16f6-495d-af2f-9b41ad6348cc
so I can properly document what is being changed!
Revert / Merge : pqissl / pqissludp code.
* restored original timeout system.
* removed weird testing code - should never have been here!
* switched extra checks to generate ALERT msgs - we want to know if they happen.
* removed quietShutdown (breaks TLS protocol - why was this added?)
* switched listening on all the time. (helps restore disconnects).
* restored common connect parameters (removed duplicates in ssludp).
* restored AuthSSL->Failed/CheckCertificate() calls ... useful.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3212 b45a01b8-16f6-495d-af2f-9b41ad6348cc
This used an unprotected singleton -> bad... Probably killed a lot of connections.
* removed mConnMgr tests in authssl
AuthSSL shouldn't use ConnectMgr. what happened to abstraction / seperation of function???
This code prevents quick reconnections.
* Disabled AuthSSL from adding new SSL peers.
This functionality is important... but should be handled elsewhere (p3connmgr or pqissl)
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3200 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- also added support src for serialiser test - add functions here that could be generally used throughout serialiser tests
- fixed empty grp id set of grp keys in creategrp function
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3196 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- added a destructor to RsTunnelDataItem, to remove a memory leak on all data packets.
- apparently the code would allow data items of zero size, hence calling malloc(0), which
has undetermined behavior.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3191 b45a01b8-16f6-495d-af2f-9b41ad6348cc
NOTE: This tests packets up to size of 1Mb. This causes the test to FAIL
due to getRsPktMaxSize() returning 262142.... What to do? This needs to be fixed!
see comments at serialiser/rsserial.cc line 448
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3180 b45a01b8-16f6-495d-af2f-9b41ad6348cc
* Added existing ft tests to tests directory (don't work yet!)
* Added dbase tests to tests directory (don't work yet!)
* moved net_setup test to tests/general directory (don't work yet!)
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3176 b45a01b8-16f6-495d-af2f-9b41ad6348cc
use for handling the packet: RsItem::DATA_QUEUE for data and RsItem::CONTROL_QUEUE for
control packets, to be used in pqistreamer.
Up to now, File data chunks where the only packets that were going into the --less prioritized--
data queue. Now, the default in RsItem::queueType() being to return CONTROL_QUEUE, developpers can
choose which queue t use by simply overloading this method so that it returns a different value.
The first use of this was to make generic turtle tunnel packets use the data queue instead of
the control queuem, which restores a correct balance between turtle downloads and friends downloads.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3153 b45a01b8-16f6-495d-af2f-9b41ad6348cc
3148: modified the update system for address lists, so that the list is totally
replaced (and not merged) when it comes from the peer itself. This helps
cleaning wrong addresses
libretroshare/src/pqi/authssl.cc
libretroshare/src/pqi/p3connmgr.cc
libretroshare/src/pqi/p3connmgr.h
libretroshare/src/services/p3disc.cc
3149: removed uninitialized memory read
libretroshare/src/pqi/authgpg.cc
3151: corrected missed update of peer info when connected. Disabled setting
peer connexion IP from p3disc info (does not make sense, and leads to
errors)
libretroshare/src/services/p3disc.cc
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3152 b45a01b8-16f6-495d-af2f-9b41ad6348cc
Corrects wrong logic when using Win32 API for handling the filesystem, Win32 functions return 0 for an error while POSIX functions return 0 for a success.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3146 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- uses joss's/ open ssl sealinit implementation. public key encryption
- invoked by rs_group_encrypted group flag
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3126 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- memory leak -> "RsItem* item" was not freed, when "dynamic_cast<RsStatusItem*>" failed
- optimized return of std::list as parameter and not as return. return will copy the list and its not necessary
p3StatusService::getStatus
- memory leak -> items in "std::list<RsStatusItem* > status_items" was not freed
- potential crash, when receiving a status for an unknown peer
p3Peers::getPeerDetails
- optimized - call to "AuthSSL::getAuthSSL()->OwnId()" only once
- optimized - add ip addresses directly to "d.ipAddressList"
PeersDialog::insertPeers
- fixed possible crash, when ssl child has disappeared and was removed from tree, there was a missing continue
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3125 b45a01b8-16f6-495d-af2f-9b41ad6348cc
3099: libretroshare/src/tcponudp/udplayer.cc
libretroshare/src/tcponudp/udplayer.h
libretroshare/src/util/rsnet.cc
libretroshare/src/util/rsnet.h
moved operator<<(ostream&,sockaddr_in) to rsnet.h, as it is used at
different places
3100: libretroshare/src/util/rsdiscspace.cc
suppressed debug mode
3101: libretroshare/src/serialiser/rsdiscitems.cc
put correct (i.e. non conservative) size for RsDiscReply packets
3102: libretroshare/src/services/p3disc.cc
Made a pass on the code in p3disc. Corrected some bugs:
- suppressed an unwanted return in packet treatment
- prevented sending info to a peer about itself
- changed askInfoToAllPeers() such as not to discard info
from peers with NODISC flag (because we especially need
info for these)
- enabled receiving p3disc info even if p3discovery is
disabled. Indeed, disabeling p3disc is a measure of
protection, so it should limit the export of p3disc info,
not the import.
- removed test discarding info about dummy friends, because
it is useless
- don't discard info about peers that have the NODISC flag
(meaning that we especially need info for them)
- added safety check about received GPG keys. Before we
relied on this test being performed by
p3ConnMgr::addFriend()
- added some debug info
3103: libretroshare/src/ft/ftcontroller.cc
removed debug info
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3104 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- memory leaks in pqi/cleanupxpgp.cc , 2 new char[] => 2 delete[]
- reordering some constructors' initializing order
- char* myString = "foo" constructs must be prefixed by "const"
- using size_t for variable types when dealing with STL
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3079 b45a01b8-16f6-495d-af2f-9b41ad6348cc
3068: added verification for file size computation (bug correction)
3069: set heartbeat values to intermediate values
3070: suppressed uninitialised memory read
3071: automatic removal of file lists from deleted peers, at restart (After
double check that this does not alter exchange of file lists in any way)
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3072 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- save msgId in configuration file
- enable previous improvements read/unread state and tags
new function for creating backups of a file
bool createBackup (std::string sFilename, unsigned int nCount = 5);
currently its only available for windows compile. maybe there is a linux developer who change it for linux.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3064 b45a01b8-16f6-495d-af2f-9b41ad6348cc
This avoids that an attacker that captures a broadcast hash request can also ask for the file.
This needed to add back explicit sources in SearchDialog and private chat transfer. I dynamically search
for sources in file lists just before calling FileRequest.
When RS links anchorClick() will work properly with Qt, we can remove the search flag "EXTRA" from the
tunnel digging in p3turtle.cc, otherwise, people having this bug can't transfer files in private chat by
copy+paste of a RS link.
Also:
- added missing mutexes in search requests into fimonitor.cc
- suppressed some debug info in connect manager
- added check for write fails when dumping config files
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3027 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- added failure tests for fwrite. Some where missing.
- added a test against the IP 1.0.0.0, which pops up on MacOS (when the peer is not connected to the internet?)
- put some additional debug messages.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@3018 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- enabled draft box
bugfix MessagesDialog:
- correct from and to in list and message header
changes MessageComposer:
- cleaned channel part
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2994 b45a01b8-16f6-495d-af2f-9b41ad6348cc
2951: cleaned up some desastrous code in rsdiscitems.cc
2952: put back an exit when the local address+port are already in use.
Otherwise this corrupts file lists, config files etc.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2953 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- it was previously corrupting group keys
added dynamics size to grps created
- previously limited to 16000 bytes
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2928 b45a01b8-16f6-495d-af2f-9b41ad6348cc
I got confused somehow between the different versions of my modifications and Linux/Windows, anyway it should work now.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2925 b45a01b8-16f6-495d-af2f-9b41ad6348cc
The problem was that libretroshare handles files in UTF-8 but Windows's ANSI/POSIX C functions automatically assume that the char* parameters are encoded with the system's code page. There is no way to set that code page as UTF-8. So now under Windows the code translates the file name to UTF-16 before feeding it to one of the Unicode functions (they are usually prefixed or suffixed by 'w'). Please note that it is not very efficient.
Furthermore, Windows does not provide a Unicode version of opendir/readdir/closedir, so it was necessary to use FindFirstFileW/FindNextFileW/FindClose which has a different behaviour as well as different structures. The FolderIterator class was created in order to mimic the Unix way of traversing folders contents. Hence the algorithm is unchanged and the systems differences masked.
As it was necessary to use some functions from the Windows API, <windows.h> had to be included in a few files were it didn't appear before, creating macros and #define conflicts. In order to solve them, util/rswin.h must be included first in a file (if necessary). Otherwise the preprocessor will fail on purpose to avoid the code being compiled with different _WIN32_WINNT values.
As another side-effect, rstlvutil.h and rstlvutil.cc have been removed from libretroshare.pro file. They are only used by testing units and include util/utest.h which defines a macro FAILED that already exists in <windows.h>. I don't know if unit tests are still in use and I don't plan on coding often on Windows, so I'll leave that as an exercise (hot potato?) to a motivated fellow Windows programmer.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2924 b45a01b8-16f6-495d-af2f-9b41ad6348cc
pqi/authgpg.cc:
- remove French comment (with nasty non-unicode characters)
- use std::cerr instead of fprintf
tcponudp/extaddrfinder.cc:
- initialize correctly the sockaddr_in, former code wouldn't pad sin_zero
(which is required by the standard)
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2909 b45a01b8-16f6-495d-af2f-9b41ad6348cc
Now used in MainWindow.cpp, MessagesDialog.cpp and peerstatus.cpp
void p3Peers::getPeerCount (unsigned int *pnFriendCount, unsigned int *pnOnlineCount);
void p3Msgs::getMessageCount(unsigned int *pnInbox, unsigned int *pnInboxNew, unsigned int *pnOutbox, unsigned int *pnDraftbox, unsigned int *pnSentbox);
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2898 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- ftController::copyFile -> buffer and file pointer leaks in case of error
- PeersDialog::insertSendList -> new QTreeWidgetItem for all online peers, this function is not ready yet
- PeersDialog::insertPeers -> gpg_item = new QTreeWidgetItem(0); in case of no more accept connection
- pqissl::reset -> missing SSL_free of ssl_connection
Missing return:
- IntroPage::nextId
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2891 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- clean up of channels old (> storeperiod) cache file downloads called at rsglobal shutdown
- added extrafilehash and extra-remove to rschannels, so it chooses appropriate way to add files (correct book-keeping)
- added extra files not copied to chan dir if over 100mb (high overhead)
- added limit to channels auto-download (1gig)
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2866 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- corrected calculation of message counts
- p3MsgService::checkOutgoingMessages -> notify when message was sent
- refill of message list without clear, selected messages and scroll position are not changed
- corrected sorting of date with only time for today
- changed context menu in message list
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2863 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- since libretroshare doesn't use Qt at all, we remove it from the CONFIG
directive,
- adding files in HEADERS doesn't include them for compilation but only for
installation. Instead we must use gpg-error-config and gpgme-config to fill
correctly INCLUDEPATH.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2856 b45a01b8-16f6-495d-af2f-9b41ad6348cc
RetroShare checks for locale gpg.exe and gpgme-w32spawn.exe and if exists, it starts as portable version.
Modified version of libgpgme-11.dll needed.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2815 b45a01b8-16f6-495d-af2f-9b41ad6348cc
Currently the only way to induce a bug is start rs for an ssl_id which
which does not have a password saved, then try use an ssl id at start
dialog which does or does not have a pword (help_pword:bool enters an
illegal state)
I have fixed it by checking if rsinit::passwd == "" then trying to load for
that ssl_id, if that fails then i set help_pword to false; does not have
password and pass flow to gpg to ask for passwd. (problem was rs thought
user had help/pass file and they didn't).
i have checked it in and tested for
1. starting with ssl_id which has saved pword, and switching ids at start
dialog
2. secondary id may or may not have saved pword.
- added check for the result of fscanf
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2802 b45a01b8-16f6-495d-af2f-9b41ad6348cc
I have fixed it by checking if rsinit::passwd == "" then trying to load for that ssl_id,
if that fails then i set help_pword to false; does not have password and pass flow to gpg to ask for passwd. (problem was rs thought user had help/pass file and they didn't).
i have checked it in and tested for
1. starting with ssl_id which has saved pword, and switching ids at start dialog
2. secondary id may or may not have saved pword.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2801 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- Bug fix: Checkbox for automatic share of the incoming directory doesn't correctly set with setDown, use setChecked
- ShareManager and RSettingsWin doesn't need to be created all the time
- RSettingsWin: Save the last active page for the current runtime
- ShareManager and DirectoryPage: Show changed shared directories direct after the change
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2777 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- getcustom status now generate a requests to that peer.
- setowncustom status now generates an 'status available' item sent to all peers
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2769 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- call back function did not pass prev_was_bad as parameter
to gui/qtnotify function which requests password
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2738 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- in the situation where autologin is enabled
- BUG found (or rather autologin induced it), if one asks to make friends to a peer (in ConfCertDialg) 'within' a session that uses autologin it freezes
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2703 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- Using rc4 to store ssl pword on file for linux, for windows already using cryptprotect
- yes its not safe, but its the user's choice to keep pword or not
- consider using method closely hooked to linux kernel (another layer of obfuscation)
- windows i believe is stronger (because its proprietary/ heavily obfuscated)
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2690 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- suppressed old DwlQueue class
- turned mDownloads into a list of pointers to allow easy cross-info update between queue and std::map of downloads
- added queue functionality for moving files top/bottom/up/down
- added the necessary functions in rsFiles
Put back display of exact sizes in FileTransferInfoWidget
Suppressed some warnings.
**Warning**: this commit requires a make clean.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2493 b45a01b8-16f6-495d-af2f-9b41ad6348cc
- essentially if any part of the file saving procedure fails, there should be config files whose signatures/hash match up
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2476 b45a01b8-16f6-495d-af2f-9b41ad6348cc
1) b64.c used c++ functions -> b64.cc
2) OSX will use miniupnpc for now -> modified #defines.
3) change in openssl fn signature?
4) updated build file for OSX.
git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@2344 b45a01b8-16f6-495d-af2f-9b41ad6348cc
