Re: Segmentation fault, dd or egrep

[Jim Meyering]

Zev Weiss <address@hidden> wrote:
> I'm trying to recover some data from a disk that was unexpectedly
> reformatted, and am running into a segmentation fault, though it's
> not clear from the output whether it's coming from dd or egrep (hence
> this email being sent to both grep and coreutils).  I'm running the
> following command in bash:
>
> dd if=/dev/sdc | egrep -A 50 -B 50 -a '^[\\\:]\ [^\ \t\n]{1,15}\ \(\
> [A-Za-z0-9\ ]{0,15}\ \-\-\ [A-Za-z0-9\ ]{0,15}\ \)$' > ./outputfile
>
> (Looking for telltale signs of Forth source code.)
>
> It runs for about about two minutes and then fails with a segfault,
> though the console output doesn't tell which process segfaulted.  I'm
> running Fedora 8 x86_64 with Linux kernel version 2.6.23.1-42.fc8 (in
> runlevel 3, for what it's worth).
>
> dd is version 6.9
> egrep is version 2.5.1
> bash is version 3.2.25(1)-release (x86_64-redhat-linux-gnu) [if that's
> of any relevance]

Thanks for taking the time to report it.
Rerun the pipeline, using strace on one command at a time or both.
I'd log egrep's syscalls first:

dd if=/dev/sdc | strace -o egrep-log \
  egrep -A 50 -B 50 -a '^[\\\:]\ [^\ \t\n]{1,15}\ \(\
[A-Za-z0-9\ ]{0,15}\ \-\-\ [A-Za-z0-9\ ]{0,15}\ \)$' > ./outputfile

Then when it segfaults, look at egrep-log.
If that's not helpful, do it for dd instead.

Or easier still, just eliminate dd and the pipeline:

egrep -A 50 -B 50 -a '^[\\\:]\ [^\ \t\n]{1,15}\ \(\
[A-Za-z0-9\ ]{0,15}\ \-\-\ [A-Za-z0-9\ ]{0,15}\ \)$' < /dev/sdc > ./outputfile

if that fails, it's egrep's fault.