Comment 2 for bug 109994

The coredump is not likely to be meaningful, since debug symbols aren't available for coreutils yet.

The combination of the value of the PATH, combined with the fact that cwd is /, leads me to believe it was probably fired off by hal. Need more clues as to what file tr was processing, though (on standard input). Had you recently plugged/unplugged something into your PC when this crash happened?

I have my doubts as to whether we'll be able to reproduce this; SIGILL is "illegal instruction". Possible explanations would include a corrupted binary for tr (probably not in the package itself; more likely a bad extraction); somehow running an instruction that wasn't meant for that architecture (trying to run i686 code on a lesser CPU); a subtle problem with the CPU, such as overheating.

Regarding the commandline: I doubt very much that apport is reporting it accurately. There is only one space after the first backslash, meaning there was only one argument to the command: ' \n" (space backslash n), which is not a legal invocation of tr. I suspect it was more likely two arguments: ' ' '\n', or arg1: space, arg2: backslash n (representing newline in tr). I'll try to investigate to see whether apport is handling such things correctly.

Setting to "Needs Info", for the question asked in the second paragraph, above.