bitcoin-qt crashes with message : QMetaMethod::invoke: Unable to handle unregistered datatype 'bool*' #2402

miniupnp commented at 5:02 PM on March 22, 2013: none

It happens during the wallet synchronisation. On my debian linux 6.0.7 amd64 $ 64/bitcoin-qt QMetaMethod::invoke: Unable to handle unregistered datatype 'bool*'

I executed the precompiled binary from bitcoin-0.8.1-linux.tar.gz

miniupnp commented at 7:08 AM on March 23, 2013: none

I have exactly the same problem with the program freshly compiled from the gihub sources : $ bitcoin-qt QMetaMethod::invoke: Unable to handle unregistered datatype 'bool*' v0.8.0-57-gdfd71bb-beta

miniupnp commented at 7:11 AM on March 23, 2013: none

it looks like it happens when the free space on disk is getting low...

Diapolo commented at 10:08 AM on March 23, 2013: none

So is this the case? Can you try it with more free space available please.

miniupnp commented at 2:37 PM on March 23, 2013: none

the synchronisation is working well if there is plenty of free space. When there is only around 50 or 60MB free left, it crashes with the Unable to handle unregistered datatype 'bool*' and that's all. Difficult to understand what is the cause of the problem...

Diapolo commented at 3:10 PM on March 23, 2013: none

I have some feeling, that perhaps there could be a problem with the signal-handlers from core to GUI, when the disk-space check fails. @laanwj Can you take a look if this could be related to ThreadSafeMessageBox? I know @sipa added a bool, which we use to handle dialog-boxes. Perhaps there is something wrong with it?

sipa commented at 2:56 PM on April 13, 2013: member

@laanwj Can you have a look?

laanwj commented at 3:35 PM on April 13, 2013: member

I'll take a look

laanwj commented at 3:52 PM on April 13, 2013: member

Is there another disk space check apart from CheckDiskSpace? I rigged that function to always call AbortNode, but that works fine. I get the error in a popup box and the client exits. As expected.

I suppose it could be a Qt version issue? What version (About Qt) are you using? Mine is 4.8.2.

zyv commented at 5:25 PM on April 13, 2013: none

In terms of diskspace, how much is "low"? I'm having these crashes when ~5G are still free. I will try with 10G now.

zyv commented at 10:31 PM on April 13, 2013: none

Tried with ~10G free space, but sadly it didn't make a difference, it still crashes with the same error message. However, I've got something interesting in my debug.log:

received block 000000000000012eaf65592dadb1ffa27f72f1e65f74d588ebced08e7d19fb3d
LevelDB read failure: Corruption: block checksum mismatch
*** System error: Database corrupted
Flush(false)
DBFlush(false) ended               1ms
ThreadSocketHandler exited
ERROR: AcceptBlock() : AddToBlockIndex failed
ERROR: ProcessBlock() : AcceptBlock FAILED
StopNode()
ThreadOpenConnections exited
ThreadMessageHandler exited
ipcThread exited
Flushed 14747 addresses to peers.dat  154ms
Committing 14044 changed transactions to coin database...
Flush(true)
DBFlush(true) ended               1ms
Bitcoin exited

So, apparently, this message is just a misleading facade of a database corruption. Is there anything I can do about it? I tried re-starting, but it always crashes in the end... My database is inherited from a very old 0.3.x client.

Diapolo commented at 1:52 PM on April 14, 2013: none

@laanwj @sipa Should this error induce a question to rebuild via the known GUI dialog-box? That is still something where I think we also could have a problem with this users OS and ThreadSafeMessageBox.

https://github.com/bitcoin/bitcoin/blob/master/src/init.cpp#L833

laanwj referenced this in commit e83a90f127 on Apr 14, 2013

laanwj commented at 4:56 PM on April 14, 2013: member

I'm pretty sure #2525 solves the issue. Can anyone please test?

laanwj commented at 4:01 PM on April 18, 2013: member

Closing -- I'm assuming this is solved by #2525. Let me know if this is not the case.

laanwj closed this on Apr 18, 2013

zyv commented at 4:06 PM on April 18, 2013: none

Hi Wladimir, I think you nailed it. Considering my crash logs your explanation sounds very sensible. I can test is as soon as the new binary is released. Thanks!

namecoin-qt referenced this in commit bcf8fe927f on May 31, 2013

kyledrake referenced this in commit 5fed4dfd5c on Jul 25, 2013

luke-jr referenced this in commit 04b075e579 on Aug 16, 2013

davispuh referenced this in commit 5a0314bf29 on Feb 18, 2014

laudney referenced this in commit 6bea021fce on Mar 19, 2014

DrahtBot locked this on Sep 8, 2021