Also - another bug in the C++ code
BitSwarmClient::BitSwarmClient() should initialize cryptoKey = NULL. Otherwise, clients that aren't using encryption could get incorrect values and falsely determine that encryption is enabled. We've seen a couple crashes in this code.
Search found 38 matches
- 30 Nov 2015, 07:53
- Forum: SFS2X C++ API
- Topic: Recommended CryptoInitializer.cpp changes
- Replies: 2
- Views: 10796
- 30 Nov 2015, 07:31
- Forum: SFS2X C++ API
- Topic: Recommended CryptoInitializer.cpp changes
- Replies: 2
- Views: 10796
Recommended CryptoInitializer.cpp changes
Trying to turn on encryption for the C++ client. It works fine, but we ran into some blocks that we've resolved internally by making some changes to the way CryptoInitializer.cpp works. These are the issues we ran into: 1. It assumes the config-Host() is an IP address and doesn't account that it may...
- 06 Nov 2015, 11:33
- Forum: SFS2X C++ API
- Topic: C++ disconnect sequence incorrect?
- Replies: 2
- Views: 9896
C++ disconnect sequence incorrect?
Hi - We're still seeing some crashes (a lot less than before) on iOS devices. When we disconnect from SmartFox, occasionally we'll get some crashes because SmartFox still attempts to read or write something back to the closed socket. To try to alleviate this, we've used the same mechanism used elsew...
- 19 Oct 2015, 09:22
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
One last follow-up... We've had thousands of players on the new build that is compiled with C++ optimizer -O2 over the past week or so. Android devices are no longer crashing frequently from the problematic scenario we detailed out in previous posts. So this confirms that changing the NDK build sett...
- 13 Oct 2015, 06:58
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Ok and finally - this is what we've found works and the decision on which resolution to use in our next app update: boost::shared_ptr appears to be very compiler optimization dependent for about 1/3 of all Android devices. We changed Android NDK's NDK_DEBUG flag to 0 and this issue no longer occurs....
- 12 Oct 2015, 17:53
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Just to provide an update on things we've tried: 1. We've noticed on iOS that not actually setting ThreadManager::running to false when disconnecting while app is in background will cause 2 of the 3 crashes listed above. By actually forcing a stop on ThreadManager as detailed in the other posts in t...
- 12 Oct 2015, 06:42
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Also - from Android NDK 10-12 14:39:27.362: I/DEBUG(25044): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** 10-12 14:39:27.362: I/DEBUG(25044): Build fingerprint: 'google/occam/mako:4.2.2/JDQ39/573038:user/release-keys' 10-12 14:39:27.362: I/DEBUG(25044): Revision: '11' 10-12 14:39:2...
- 11 Oct 2015, 10:21
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
One other observation - this crash happens a lot more frequently when we use release builds. NDK_DEBUG = 0, which also sets the optimization flag to -O2.
- 10 Oct 2015, 07:02
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Hi Massimo - Tried the changes you posted. It still crashes and in the same spot on Android for iOS, we've also seen a similar crash but it happens at a different point. A few things to note, from what we've seen, there appears to be a couple ways that get to the same crash. Is it possible that this...
- 08 Oct 2015, 06:01
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Hi- Thanks for the thorough and insightful response. So a few points and even more questions. 1. We don't call SmartFox::Dispose() right after a Disconnect(). We call Dispose() if later on in the game flow, the player decides to reconnect to our game server. We don't call SmartFox::Dispose() right a...
- 07 Oct 2015, 12:58
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Ok - we think we may know what's wrong - but this is a bit hard to reproduce. In our game, there is a mechanism that disconnects the player if they repeatedly receive a slow ping to the SFS2X server (we set a threshold of continued slow or missed pings). This helps ensure that all players have a goo...
- 06 Oct 2015, 08:48
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Hi Lapo, Thanks for the response and follow-up. Just checking back again. The first crash listed is causing us the most problems (accounts for nearly 70% of all reported crashes): Thread : Crashed: SIGSEGV SEGV_ACCERR at 0x8000000000000000 0 ??? 0x9165d1b0 1 p.so 0x99b60908 Sfs2X::Util::DelegateOneA...
- 05 Oct 2015, 10:05
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Hi -
Any updates on these issues? For the first one we see on Android, it appears to happen when the player has been connected for a long time.
Any updates on these issues? For the first one we see on Android, it appears to happen when the player has been connected for a long time.
- 01 Oct 2015, 13:25
- Forum: The Gallery
- Topic: Texas Hold'em Poker +
- Replies: 1
- Views: 11169
Texas Hold'em Poker +
Our game using SmartFox Server 2X 2.10.0 on the back end and the SFS2X C++ API is live for iOS and Android (Google Play and Amazon). Please check it out! I'd be happy to answer any questions about how it was implemented and the hurdles we encountered using SFS2X and the C++ API's. Game Website: http...
- 01 Oct 2015, 13:06
- Forum: SFS2X C++ API
- Topic: Crash reports
- Replies: 18
- Views: 34816
Re: Crash reports
Ok we've finished upgrading to the latest 1.6.1 and added the deadlock fix noted in another thread. We are still seeing the same crashes. We get live crash reports for our live games. These crashes (smart fox C++ related ones) are the most frequent crashes that currently occur in all versions of our...