View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001813 | Spring engine | General | public | 2010-02-01 01:20 | 2010-02-10 00:27 |
| Reporter | Auswaschbar | Assigned To | tvo | ||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | resolved | Resolution | fixed | ||
| Product Version | 0.81.1.3 | ||||
| Fixed in Version | 0.81.2 | ||||
| Summary | 0001813: Reproducable desync | ||||
| Description | [01:02:25] <[ZXC]t0> Linux jebuntu 2.6.28-17-generic 0000058-Ubuntu SMP Tue Dec 1 21:27:25 UTC 2009 x86_64 GNU/Linux [01:02:39] <[ZXC]t0> gcc version 4.3.3 (Ubuntu 4.3.3-5ubuntu4) constantly desynced with me after about 10000 frames. | ||||
| Tags | No tags attached. | ||||
| Attached Files | |||||
| Checked infolog.txt for Errors | |||||
|
|
Me and [SIN]zCram were desyncing too with gcc-4.3.4. gcc-4.4 works fine. g++ (Debian 4.3.4-6) 4.3.4 -> desync gcc-4.4 (Debian 4.4.2-9) 4.4.3 20100108 (prerelease) -> no desync |
|
|
I got a desync with the new version too: http://replays.adune.nl/?2081 with my spring version, demo.zip is my local recorded file, i'm using gentoo amd64. gcc (Gentoo 4.3.4 p1.0, pie-10.1.5) 4.3.4 i hope this helps! |
|
|
i've recompiled with gcc (Gentoo 4.4.2 p1.0) 4.4.2 and the sync errors seems to be gone... |
|
|
recompiled with gcc 4.4(20090418) on ubuntu sync errors gone |
|
|
on Gentoo with gcc 4.3.4 p1.0 using DEBUG2 build of spring no desyncs in online matches yet (in about 4-5 BA matches) extracted from springs CMake build file: DEBUG -O1 (the default when compiling with cmake) DEBUG2 -O0 DEBUG3 -O0 RELWITHDEBINFO -02 (release build uses this) is there a reference for changes between 4.3 and 4.4 with -O2? if so, could that probably help in finding the problem? |
|
|
Solved in syncdebug-gcc-4.3 branch. It was introduced by adding __attribute__((const)) to (in particular) GetHeadingFromVector. (Commits a1c92dfe9d2aafc1e9f44439ad530f758faeac98 and 65a8593768a2e35801403091bb217daa167b5971) |
|
|
Thanks to Kloot for his help with testing btw :-) I'll see how to fix definitely in master another day. |
|
|
Btw this is how the desync could be reproduced with 100% reliability: - Start game using XTA 9.6. - Make rotated airlab (not sure if the rotation actually mattered) - Make peeper - Desync when peeper is at 94% |
|
|
IMO desyncs like these (and those caused by "ordinary" uninitialized memory-access bugs in Spring, etc.) form the best argument why any attempt at a resync system would be misguided :) |
|
|
Should be solved by the fix to 0001818. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2010-02-01 01:20 | Auswaschbar | New Issue | |
| 2010-02-01 01:20 | Auswaschbar | File Added: syncdebug-server.log | |
| 2010-02-01 01:25 | Auswaschbar | File Added: syncdebug-server2.log | |
| 2010-02-01 01:49 | SirMaverick | Note Added: 0004581 | |
| 2010-02-01 23:26 | abma | File Added: demo.zip | |
| 2010-02-01 23:27 | abma | Note Added: 0004584 | |
| 2010-02-01 23:28 | abma | Note Edited: 0004584 | |
| 2010-02-02 18:47 | abma | Note Added: 0004585 | |
| 2010-02-03 15:05 | t0 | Note Added: 0004592 | |
| 2010-02-04 10:48 | imbaczek | Status | new => acknowledged |
| 2010-02-06 17:00 | hoijui | Note Added: 0004608 | |
| 2010-02-07 00:58 | tvo | Note Added: 0004610 | |
| 2010-02-07 01:00 | tvo | File Added: syncdebug-server3.log | |
| 2010-02-07 01:12 | tvo | Note Added: 0004611 | |
| 2010-02-07 12:01 | tvo | Note Added: 0004615 | |
| 2010-02-07 12:02 | tvo | Note Edited: 0004615 | |
| 2010-02-07 12:28 | Kloot | Note Added: 0004616 | |
| 2010-02-07 12:28 | Kloot | Note Edited: 0004616 | |
| 2010-02-10 00:27 | tvo | Note Added: 0004632 | |
| 2010-02-10 00:27 | tvo | Status | acknowledged => resolved |
| 2010-02-10 00:27 | tvo | Fixed in Version | => 0.81.2 |
| 2010-02-10 00:27 | tvo | Resolution | open => fixed |
| 2010-02-10 00:27 | tvo | Assigned To | => tvo |
| 2010-02-10 00:27 | tvo | Note Edited: 0004632 |
