[Coin-discuss] *** glibc detected *** double free or corruption (fasttop)

[Ted Ralphs]

After checking out Cgl from two weeks ago, this error goes away (see my 
previous post).

I observed this error only when compiling COIN on a 64-bit architecture, 
and I suspect it may be unique to that setting. It seems to be just what 
would normally be a segmentation fault. Compiling and running the exact 
same version of the code on a 32-bit architecture results in a 
segmentation fault. Unfortunately, valgrind cannot interpret 64-bit 
code. A quick google search on

*** glibc detected *** double free or corruption

turns up lots of hits that seem to support this. Spencer, did you happen 
to be on a 64-bit machine? At any rate, getting the Cgl version from two 
weeks ago should fix it for now until we find the cause of the problem.

Cheers,

Ted

Ted Ralphs wrote:
> I just did a CVS update and now am also getting this exact error. I'm 
> trying to track down the cause and will report back.
> 
> Ted
> 
> Matthew Saltzman wrote:
> 
>> On Mon, 28 Feb 2005, Spencer Fung wrote:
>>
>>> Hi,
>>>
>>> I have a problem of executing the OSI unitTest program. I have 
>>> installed the latest COIN package COIN_2005Feb26.tgz and compile them 
>>> with gcc-3.4.2-6.fc3 and glibc-2.3.4-2.fc3. The following error 
>>> message arised when executing the program,
>>>
>>> *** glibc detected *** double free or corruption (fasttop): 
>>> 0x08cf9f58 *** Abort
>>>
>>> How can I fix the problem?
>>
>>
>>
>> Help us reproduce it.  What solver are you linking with?
>>
>>>
>>> Thanks.
>>>
>>> Spencer
>>
>>
>>
> 
> 


-- 
Dr. Ted Ralphs
Assistant Professor
Industrial and Systems Engineering
Lehigh University
(610)758-4784
tkralphs at lehigh.edu
www.lehigh.edu/~tkr2