Re: md5sum freezes system

[Daniel]

Hi,
this is one of the 'I really had a bug' problems.
I don't know what I have done, but md5sum works know. I swear that I had this 
problem and that it was reproducable. I will give you any feedback if I know 
more.

thanks for fast help
Daniel

Am Samstag, 15. Juli 2006 17:58 schrieb Philip Rowlands:
> On Sat, 15 Jul 2006, Daniel wrote:
> > If I run md5sum on big files (>700MB) it freezes the system. There are
> > no logs, no panic, nothing. It only shows the last screen.
>
> This will be difficult to diagnose if the problem occurs only on your
> system. Are you able to reproduce it on another computer?
>
> When you say freeze, is the computer entirely unresponsive? Can you
> interrupt (Ctrl-C) md5sum to get back to the shell prompt?
>
> 700MB is a reasonably large amount of data, which might longer than
> you'd expect to process on an older system, say, with slow disk.
>
> > I runned md5sum with strace and gdb because I want to know what
> > happens - but everything was fine and md5sum returns the hash.
>
> Although I don't expect this is an md5sum bug, I'll need to convince you
> of that. Is 700MB the smallest file that breaks? Is it any 700MB file,
> or a specific one on your disk? The following commands may help narrow
> down the cause:
>
> (I'm going to name the problematic 700MB file "badfile")
> $ cat badfile > /dev/null
>      (should return to the prompt with no output)
>
> $ dd bs=1M count=700 if=/dev/zero | md5sum
> 700+0 records in
> 700+0 records out
> 734003200 bytes (734 MB) copied, 3.92202 seconds, 187 MB/s
> 7fe5ca2a051d6dbb9ef191fbee0af98c  -
>      (tests md5sum handling 700MB of non-disk data)
>
> $ dd bs=1M count=700 if/dev/zero > /tmp/bigfile
> $ md5sum /tmp/bigfile badfile
>
> > So, can anyone of you explain this behavior? What happens there? Why
> > is md5sum running fine if it is under control? Are there ways to get
> > some more informations?
>
> The above tests should isolate whether it's md5sum or the specific file
> which triggers your problem.
>
> > PS: it is _no_ hardware defect ;)
>
> How can you be sure?
>
>
> Cheers,
> Phil
>
>
> _______________________________________________
> Bug-coreutils mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/bug-coreutils