Re: ia64 GNU/Linux pwd-long test failure in CVS coreutils

[Jim Meyering]

Paul Eggert <address@hidden> wrote:
> Thanks, Andreas.  I have filed this as a Debian bug (against glibc) here:
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=373930
>
> A possible workaround for coreutils is to use a path of more than
> 16384 bytes in length for the test, rather than just more than 4096 --
> this is according to Aurelian Jarno in
> <http://sourceware.org/bugzilla/show_bug.cgi?id=2418>.
> I dunno, though, that sounds like just masking the error.

I read that, too, but am not inclined to make the test program
create such a deep hierarchy, just for ia64 (or whatever systems
have PATH_MAX << pagesize).  Already, creating a hierarchy deep
enough to trigger getcwd misbehavior is a little onerous.
Quadrupling the depth/length seems excessive.

However, if more systems end up being affected, or
if someone can come up with a run-test that's not too unreasonable,
then maybe it'd be worthwhile.

FWIW, I think the proposed change would involve creating deeper
and deeper directories, in spite of partially-failing getcwd
calls, until the real failure (bug) when the length passes 16384.