"Bo Borgerson" <[EMAIL PROTECTED]> wrote: > Regarding rev. 94a2bd5b, `tests: improve perl-based tempdir handling', > I wonder if it might be worthwhile to add a PID check in the cleanup > handler for race avoidance. > > I think File::Temp does this internally as well, but it looks like > chmod_tree will just warn about the failed chdir and procede to > recursively chmod whatever directory it was in at the time if $dir is > yanked out from under it.
Yes. Good catch. It should obviously skip the find in that case. Want to write the patch? > I've included a patch that sets up the signal handler as a closure > around a copy of the PID at the time of installation and only cleans > up in the original process. Protecting against clean-up races in multi-threaded tests is definitely worthwhile. Thanks! Luckily for my ego, I think none of the tests in coreutils are susceptible. I prefer the closure, too. Applied. > From dd8e78633f60a4a266b870326ac87d9844dab02b Mon Sep 17 00:00:00 2001 > From: Bo Borgerson <[EMAIL PROTECTED]> > Date: Mon, 28 Apr 2008 10:30:22 -0400 > Subject: [PATCH] Only cleanup test dirs from the process that created them. > > * tests/CuTmpdir.pm (import): Use closure around current PID to avoid cleanup > races. _______________________________________________ Bug-coreutils mailing list Bug-coreutils@gnu.org http://lists.gnu.org/mailman/listinfo/bug-coreutils
