This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: PING: [PATCH] Add --enable-static-pie to build static PIE
- From: Joseph Myers <joseph at codesourcery dot com>
- To: "H.J. Lu" <hjl dot tools at gmail dot com>
- Cc: GNU C Library <libc-alpha at sourceware dot org>
- Date: Tue, 26 Sep 2017 23:11:07 +0000
- Subject: Re: PING: [PATCH] Add --enable-static-pie to build static PIE
- Authentication-results: sourceware.org; auth=none
- References: <CAMe9rOqdAZkw6rAj16hnbURQVswbOAbseuK+cRfm7yzC6_ktOA@mail.gmail.com> <CAMe9rOp2jnA19-xU0PE=2410yHnyC9Hu_QMkD6wUaYi4rOBVTQ@mail.gmail.com>
On Tue, 26 Sep 2017, H.J. Lu wrote:
> Any objections?
This is a complicated feature; it clearly needs someone to do a detailed
technical review before it can go in, however long that takes. It's only
borderline-obvious patches one might consider committing after a while in
the absence of objections but without review or relevant maintainership.
Perhaps you could expand more on the design and the rationale for the
choices made? Why are x86_64 configure changes needed, for example?
What changes might other architectures need to support this feature (when
would an architecture need or not need architecture-specific configure
changes, etc.), and have you tried a build-many-glibcs.py run with this
option passed to all the builds to see how other architectures do? Why
build libc.a with -fPIE rather than building a separate PIE copy of libc.a
(i.e., making this a bit more like --enable-profile as an option to build
an additional library variant) and making GCC use that separate copy in
the static-PIE case?
The patch is missing documentation in install.texi of the new configure
option and associated regeneration of INSTALL.
--
Joseph S. Myers
joseph@codesourcery.com