Re: FreeBSD core dumps when wrapping big files!

Jean-Claude Wippler <[email protected]>
Newsgroups gmane.comp.lang.tcl.starkit
Message-ID <[email protected]>
Joey Mukherjee wrote:

>> I have no idea if this is coincidence, but 2^25 == 67108864.
>
> Winner!
>
> As guessed, using "dd if=/dev/random of=filename bs=1  
> count=67108864" to create a file, it crashes the creation.  If I  
> use 67108863, it works fine.
[...]
>> Out of curiosity, what are the resultant sizes of the starpack?
>
> This seems to be random everytime.  After this latest crash, the  
> file size of setup is 1913111.  I seen as small as 380 or so bytes  
> to huge.  When it succeeds at the 67108863, the file size is  
> 73079792.  This has my other junk in it as well, though.

The input size seems to trigger this.

> To answer some other questions:
>
> 1) Adding -nocomp makes no difference.  Either way it crashes.
> 2) If I wrap only the big file, it crashes.
> 3) If I wrap the big file on Linux, I *CAN* unwrap it on FreeBSD.
>
> The only thing I did not do that was asked, was if it was text or  
> not.  I am pretty convinced it's size oriented so I didn't bother.
>
> Finally, my last attachment is the long traceback with a symbols  
> enabled tclkit-cli.

That's interesting - so sdx wrap breaks without any Mk4tcl/mk4vfs in  
there.  I'm continuing to suspect tclvfs... if the stat buffer has a  
64-bit item while the rest of the system expects 32, or simply if  
wrong struct sizes put the wrong data in the wrong memory bytes, then  
yes a 2^25 bit change could well trigger it (say that particular byte  
is always supposed to be 0 or 1).

I'll go through the stack trace a bit more.  Nothing jumps out yet.

-jcw
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.