Duke4.net Forums: EDuke32 64bit crashlog creation broken [fixed] - Duke4.net Forums

Jump to content

Page 1 of 1
  • You cannot start a new topic
  • You cannot reply to this topic

EDuke32 64bit crashlog creation broken [fixed]

User is offline   LeoD 

  • Duke4.net topic/3513

#1

Reproduce by loading the corrupt unpatched version of the attached map.
Maybe this has never worked because even the very old 64bit executables provided by h266 (r3627, r3973) don't give a crashlog for that.
I remember however crashlogs produced by Mapster32 64bit.
I can't check under Linux, because:

Quote

Unable to connect to a repository at URL 'http://svn.eduke32.com/eduke32'
Unexpected HTTP status 500 'Internal Server Error' on '/eduke32'

Additional errors:
Could not open the requested SVN filesystem

Attached File(s)



This post has been edited by LeoD: 14 June 2014 - 10:41 AM

0

User is offline   TerminX 

  • el fundador

  #2

Thanks for reporting that SVN was broken, I upgraded to 1.8.5 earlier and apparently the package upgrade didn't restart Apache. It's working now after a simple restart of the service.
0

User is offline   Hendricks266 

  • Weaponized Autism

  #3

I can't seem to trigger the crash, 32- or 64- bit.
0

User is offline   LeoD 

  • Duke4.net topic/3513

#4

View PostHendricks266, on 02 December 2013 - 01:36 PM, said:

I can't seem to trigger the crash, 32- or 64- bit.
Oops, I forgot to mention that it's Polymer renderer related (HRP makes no difference).
Maybe it's machine dependent as well. I'm on AMD Bulldozer.
I have attached three more maps that crash at load time on my PC.
The not yet reproduced crashes which made me notice this issue did not occur while loading corrupt maps.

Attached File(s)


1

User is offline   LeoD 

  • Duke4.net topic/3513

#5

Bump. Problem persists for me (other recent crashes, not yet reproduced). Seen a crashlog created by EDuke32 (not Mapster32) 64bit on Windows, anyone?

This post has been edited by LeoD: 07 May 2014 - 04:26 PM

1

User is offline   Hendricks266 

  • Weaponized Autism

  #6

Found the problem. A fix and new builds will be in my next push.

EDIT: I fixed the map crashes too, so you may want to use an old EDuke32 build with the new DLLs if you want to confirm the 64-bit DLL.

This post has been edited by Hendricks266: 10 May 2014 - 07:40 PM

1

User is offline   LeoD 

  • Duke4.net topic/3513

#7

View PostHendricks266, on 10 May 2014 - 07:12 PM, said:

Found the problem. A fix and new builds will be in my next push.

EDIT: I fixed the map crashes too, so you may want to use an old EDuke32 build with the new DLLs if you want to confirm the 64-bit DLL.
Fixes confirmed for 1_bitch, dharga, TOXICWST, and Festpark. :)
1

User is offline   LeoD 

  • Duke4.net topic/3513

#8

Correct crashlog creation still fails if EDuke32 is built using RENDERTYPE=WIN :
Attached File  eduke32_or_mapster32.crash.log (27.46K)
Number of downloads: 162
1

User is offline   Hendricks266 

  • Weaponized Autism

  #9

Strange. I added some diagnostic information. What does this output?

http://hendricks266....ymInitialize.7z
0

User is offline   LeoD 

  • Duke4.net topic/3513

#10

View PostHendricks266, on 18 May 2014 - 12:03 PM, said:

Strange. I added some diagnostic information. What does this output?
There's quite some randomness involved, unfortunately. On some occasions the log seems somewhat useful but still has errors near the end. I have set up the compiler you're probably using:
gcc is /mingw/bin/gcc gcc.exe (rubenvb-4.8.0) 4.8.0
NASM version 2.11.04 compiled on May 18 2014
>make veryclean && make -j eduke32.exe RELEASE=0 LTO=0 RENDERTYPE=WIN

Two crashlogs using your DLL (error messages now in German - interesting): Attached File  crash-480rubenvb-h266DLL-1.log (48.99K)
Number of downloads: 133 Attached File  crash-480rubenvb-h266DLL-2.log (42.76K)
Number of downloads: 128

My setup:
gcc.exe (x86_64-posix-seh-rev1, Built by MinGW-W64 project) 4.9.0
NASM version 2.11.04 compiled on May 18 2014
>make veryclean && make -j eduke32.exe RELEASE=0 LTO=0 RENDERTYPE=WIN

Crashlogs (my DLL) : Attached File  crash-490plain-LeoDsDLL-1.log (51.86K)
Number of downloads: 143 Attached File  crash-490plain-LeoDsDLL-2.log (36.21K)
Number of downloads: 137 Attached File  crash-490LeoD-LeoDsDLL.log (35.12K)
Number of downloads: 136

My method to crash the game: Save a DukePlus game and load it under normal Duke setup: Attached File  dukesav1.zip (736.96K)
Number of downloads: 102
Crashes while loading, or when exiting the game right after load.
1

User is offline   Hendricks266 

  • Weaponized Autism

  #11

I was hoping it wouldn't be error 87. :) It seems that SymInitialize can return that error, yet the symbols still work. http://emptyhammock....initialize.html

Try this. The backtrace will be attempted even if SymInitialize returns an error, and error messages will be in en-US. I included the backtrace.c this time.

http://hendricks266....Initialize-2.7z

If a backtrace always works under these conditions (test the backtrace.c with your 4.9.0 setup too), then I'll look into the cause of error 87. Very interesting that RENDERTYPE=WIN triggers this.
1

User is offline   LeoD 

  • Duke4.net topic/3513

#12

View PostHendricks266, on 18 May 2014 - 03:10 PM, said:

Try this. The backtrace will be attempted even if SymInitialize returns an error, and error messages will be in en-US. I included the backtrace.c this time. [...]
If a backtrace always works under these conditions (test the backtrace.c with your 4.9.0 setup too), then I'll look into the cause of error 87. Very interesting that RENDERTYPE=WIN triggers this.
I have created a few dozen crashlogs and attached a couple of representative ones:
Attached File  crashlogs.zip (17.24K)
Number of downloads: 103
It seems indeed that backtrace always works now, either with or without error 87 messages. I've tried three r4476 executables, both your DLL and my compiled ebacktrace.cl, with and without HRP, and two different savegames. Plus an older r4465 build / your/my DLL / the aforementioned corrupt maps.
Btw., why isn't backtrace.c part of the repository?

This post has been edited by LeoD: 19 May 2014 - 09:59 AM

1

User is offline   Hendricks266 

  • Weaponized Autism

  #13

Interesting.

View PostLeoD, on 19 May 2014 - 09:58 AM, said:

Btw., why isn't backtrace.c part of the repository?

eduke32/polymer/eduke32/platform/Windows/src/backtrace.c
0

User is offline   LeoD 

  • Duke4.net topic/3513

#14

View PostHendricks266, on 19 May 2014 - 10:01 AM, said:

eduke32/polymer/eduke32/platform/Windows/src/backtrace.c
:) I can only guess that I searched for ebacktrace... :)
0

User is offline   Hendricks266 

  • Weaponized Autism

  #15

r4503 should be producing clean backtraces now. I can only guess that the bug in question is some sort of insidious memory corruption, and that is why the crashlogs vary so widely.
1

User is offline   LeoD 

  • Duke4.net topic/3513

#16

View PostHendricks266, on 13 June 2014 - 01:23 AM, said:

r4503 should be producing clean backtraces now. I can only guess that the bug in question is some sort of insidious memory corruption, and that is why the crashlogs vary so widely.
Looks better indeed. ^_^
I get less crashes overall, but that might be due to some randomness. Most crashlogs look like this:
Attached File  eduke32_or_mapster32.crash.log (1019bytes)
Number of downloads: 163
One exception is this:
Attached File  eduke32_or_mapster32.crash2.log (3.3K)
Number of downloads: 161
My local build has produced something which might be interesting (involving error 87 and backtrace.c):
Attached File  eduke32_or_mapster32.crash3.log (13.09K)
Number of downloads: 244

Now back to football ( Chile 2:1 Australia ) :)
0

User is offline   The Commander 

  • I used to be a Brown Fuzzy Fruit, but I've changed bro...

#17

Fuck Fagball, go play Rugby.
0

User is offline   LeoD 

  • Duke4.net topic/3513

#18

View PostThe Commander, on 13 June 2014 - 05:15 PM, said:

Fuck Fagball, go play Rugby.


Get back to Minecraft you slacker.
0

User is offline   The Commander 

  • I used to be a Brown Fuzzy Fruit, but I've changed bro...

#19

Minecraft?
0

User is offline   Hendricks266 

  • Weaponized Autism

  #20

View PostLeoD, on 13 June 2014 - 03:29 PM, said:

My local build has produced something which might be interesting (involving error 87 and backtrace.c):
Attachment eduke32_or_mapster32.crash3.log

When it repeats like that it means something in ebacktrace crashed and it loops until the log buffer fills up. I thought I fixed cases of that. That backtrace itself doesn't make much sense though... it doesn't match up to my backtrace.c even though you clearly are using my most recent commit.
0

Share this topic:


Page 1 of 1
  • You cannot start a new topic
  • You cannot reply to this topic


All copyrights and trademarks not owned by Voidpoint, LLC are the sole property of their respective owners. Play Ion Fury! ;) © Voidpoint, LLC

Enter your sign in name and password


Sign in options