This is the mail archive of the
mailing list for the glibc project.
Re: [PATCH] dlfcn: do not mark dlopen/dlclose as leaf functions
- From: Roland McGrath <roland at hack dot frob dot com>
- To: Mike Frysinger <vapier at gentoo dot org>
- Cc: libc-alpha at sourceware dot org
- Date: Mon, 26 Aug 2013 15:25:11 -0700 (PDT)
- Subject: Re: [PATCH] dlfcn: do not mark dlopen/dlclose as leaf functions
- Authentication-results: sourceware.org; auth=none
- References: <1377493203-1498-1-git-send-email-vapier at gentoo dot org>
> Since the dlopen funcs might invoke a constructor that is in the same
> compilation unit as the caller, we cannot mark them as leaf funcs.
> Similarly, dlclose might invoke a destructor that is in the same
> compilation unit as the caller.
This description doesn't track. You meant to say "indirectly invoke".
The comments in the test case are clear enough, but this introduction
was misleading enough that I thought you might be wrong about something.
The comments at the top of the test sources are not in the canonical style.
Nowadays I think we are doing copyright notices on new test cases too.
Every new test case should use test-skeleton.c unless there's a big reason
that it cannot.
And, needs a BZ# filed as was already mentioned.