6.50RC2 Build 3088 has been crashing a lot on me...

[SnakeByte]

Access violation at 1402a76b9.

Here's the call stack:

Code: Select all

Address          Function                                                   
-------          --------                                                   
00000001401DA39B ??0CMFCRibbonPanelMenuBar@@IEAA@XZ_5+0xF40B               
00000001401DA747 ??0CMFCRibbonPanelMenuBar@@IEAA@XZ_5+0xF7B7               
000000014021DA69 ?AddControlBar@CFrameWnd@@QEAAXPEAVCControlBar@@@Z_2+0x27A9
000000014021F250 ?AddControlBar@CFrameWnd@@QEAAXPEAVCControlBar@@@Z_2+0x3F90
000000014021F284 ?AddControlBar@CFrameWnd@@QEAAXPEAVCControlBar@@@Z_2+0x3FC4
000000014021F5F8 ?AddControlBar@CFrameWnd@@QEAAXPEAVCControlBar@@@Z_2+0x4338
00000001402220A3 nullsub_19+0x2A43                                         
00000001402AABAE DeleteCriticalSection+0x17E                               
0000000140474057 _callthreadstartex+0x17                                   
00000001404740EB _threadstartex+0x7F                                       
000007FE250F1832 KERNEL32:kernel32_BaseThreadInitThunk+1A                   
000007FE27D5D609 ntdll:ntdll_RtlUserThreadStart+21                         


The program was downloading updates from the various newsgroups at the time.

Give me the pdb file if you want more details.

[SnakeByte]

Crashed again. Same place, and again, while the program was downloading headers from all the newsgroups.

Code: Select all

1402A76B9: The instruction at 0x1402A76B9 referenced memory at 0xFFFFFFFFFFFFFFFF. The memory could not be read (exc.code c0000005, tid 11756)

Stack trace:

Code: Select all

Address          Function                                                   
-------          --------                                                   
00000001401DA39B ??0CMFCRibbonPanelMenuBar@@IEAA@XZ_5+0xF40B               
00000001401DA747 ??0CMFCRibbonPanelMenuBar@@IEAA@XZ_5+0xF7B7               
000000014021DA69 ?AddControlBar@CFrameWnd@@QEAAXPEAVCControlBar@@@Z_2+0x27A9
000000014021F250 ?AddControlBar@CFrameWnd@@QEAAXPEAVCControlBar@@@Z_2+0x3F90
000000014021F5F8 ?AddControlBar@CFrameWnd@@QEAAXPEAVCControlBar@@@Z_2+0x4338
00000001402220A3 nullsub_19+0x2A43                                         
00000001402AABAE DeleteCriticalSection+0x17E                               
0000000140474057 _callthreadstartex+0x17                                   
00000001404740EB _threadstartex+0x7F                                       
000007FE250F1832 KERNEL32:kernel32_BaseThreadInitThunk+1A                   
000007FE27D5D609 ntdll:ntdll_RtlUserThreadStart+21                         


[Quade]

I'd guess is specific to some data it's importing in some group. Might be worth the effort to run procmon and watch fileio to see if it's alway happening when a specific group is getting imported.

In the registry, if you add the value

HKEY_CURRENT_USER\Software\DJI Interprises\Newsbin50\Settings\Logfile

As a string Then set it to a filename then restart Newsbin. Newsbin should log to disk. You might not catch the actual crash log but, you might.

[Moondawgie]

I believe you're correct, Quade.

I experienced the same problem. After 17 crashes and 17 restarts I finally managed to finish the downloading of the latest message headers for all of my subscribed newsgroups. The 18th downloading of the latest of message headers for all of my subscribed newsgroups completed without any crashes. Woo hoo!

[SnakeByte]

I'd guess is specific to some data it's importing in some group.

I've set the key and have restarted. Note that the crash doesn't happen each time it downloads the headers, and it doesn't happen during import of the cache. It only seems to happen the moment it starts to quickly populate the "Downloading Files" tab. The stack trace also seems to point to a gui population issue. (RibbonPanelMenuBar)

If I had the pdb file I could tell you the exact function responsible, and maybe even the parameters passed to it, but let's see what the log file reveals.

[Quade]

Newsbin doesn't actually use a ribbonpanel so, I'm a bit skeptical about the stack trace's accuracy. Could be something deep in the GUI lib uses it. I'd be more inclined to think this is some heap corruption issue.

I've had zero crashes with it but, then again I don't have anywhere near your group count.

[Moondawgie]

I believe you're correct, Quade.

I experienced the same problem. After 17 crashes and 17 restarts I finally managed to finish the downloading of the latest message headers for all of my subscribed newsgroups. The 18th downloading of the latest of message headers for all of my subscribed newsgroups completed without any crashes. Woo hoo!

Starting with the 19th downloading of the latest message headers for all 57 of my subscribed newsgroups, and continuing with each and every of the subsequent 24 attempts, Newsbin 6.50RC2 crashes, but never while retrieving headers from the same newsgroup. I've tried downloading headers for one group at a time, five groups at a time, and ten groups at a time. Still crashes.

I utilize both Astraweb and Easynews nntp servers.

I delete any existing AutoPAR2.db3, Downloads.db3, and Downloads.db3-journal files each time before restarting Newsbin.

I've scanned and checked both of the drive partitions used with Newsbin, and no errors have ever been found.

I've uninstalled/reinstalled Newsbin 6.50RC2 three times since the problem started, the third time after wiping my system partition and reinstalling Windows 7 Ultimate 64-bit.

It's frustrating.

Note: I'm hoping that adding to this thread started by someone else but dealing with the same problem is the acceptable for the reporting of this problem. But please let me know if I should start a separate message thread of my own.

[Quade]

I have a fix for crash when doing to simul-local searches. I just don't know if it'll help you King. Your symptoms are different than the OP's. Dex might be contacting you to let you try it before too long.

[Quade]

If this doesn't fix the problem. I'd try renaming the spool_v6 folder then set the download ago to 10 and download headers and see what happens. I'm thinking corrupt spool file.