Re: [bug-gzip] gzip-1.5 test failures on FreeBSD 9.0 and NetBSD 5.1

[Paul Eggert]

On 06/19/2012 06:40 AM, Nelson H. F. Beebe wrote:

>          PID USERNAME THR PRI NICE  SIZE   RES STATE    TIME    CPU COMMAND
>        80442 beebe      1 201    0 3732K 2092K RUN      0:27 99.89% bash-4.2

Thanks for reporting this.

The symptoms are of a bug in the test suite, not in gzip itself,
so I don't think we need to worry about gzip going bad on your
platform.  Still, it'd be nice to fix the bug.

Can you show us the process tree when bash is looping?
That is, there's a 'make', and underneath that there's
a 'bash'; but are there other 'bash'es underneath that,
and which one is looping?  That would help us figure out
whether the bug is in the test suite files (tests/init.sh,
tests/help-version) or in the Automake machinery that
invokes the tests.

Is it possible to run 'lsof' or equivalent, to find out
which files the looping process has open, and the offsets
into those files?