On 3/19/24 15:32, Peter Dyballa wrote:
My guess is that this old Mac OS X "rm" has corrupted internal memory somehow.
Let's not waste more time on it.
/bin/rm is not involved because the test programme is not using it but rmdir()
when cleaning up…
I was analyzing the trace you sent here:
https://lists.gnu.org/r/bug-gnulib/2024-03/msg00235.html
The third line of that trace says it's from /bin/rm, not from the test
program. Since the rest of the trace contains clearly incorrect
syscalls, it appears that /bin/rm is seriously buggy, in that it passes
arguments to rmdir that are so bogus that NAMI isn't logged (see, for
example, line 10049 of that trace). Possibly these are stray pointers.
Again, it's probably not worth wasting time going down this rabbit hole.
We already knew /bin/rm is buggy and the details don't matter.
If you can suggest a specific patch to gnulib/m4/getcwd-abort-bug.m4,
that might help. Preferably in "git format-patch" format. Thanks.
