This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Compiler for testing a bootstrapped glibc?
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: Brooks Moses <bmoses at google dot com>, libc-alpha <libc-alpha at sourceware dot org>
- Date: Thu, 27 Feb 2014 16:05:17 -0500
- Subject: Re: Compiler for testing a bootstrapped glibc?
- Authentication-results: sourceware.org; auth=none
- References: <CAOxa4KrctdEQWfXbfd7Xk-=cFcskzUSMASCFaa2Lgf3USPPmKg at mail dot gmail dot com>
On 02/27/2014 02:49 PM, Brooks Moses wrote:
> Glibc folks -
>
> I've run into a conundrum with testing our bootstrapped-from-scratch
> glibc builds, and I'm curious how other people's processes work.
>
> Recently, we updated to a version of glibc that can do a single-step
> bootstrap -- and so we now build a static-only GCC with no glibc, and
> then build glibc using that GCC. This works great -- but then we want
> to run "make check". And, unfortunately, glibc's testsuite requires
> more than just a static-only GCC.
>
> There are a couple of solution directions that I've thought of:
>
> 1) Build a dynamic GCC using the just-built glibc, and install it on
> top of the static-only GCC we used for building. This is arguably
> simpler, but means rebuilding anything in glibc after testing will get
> different results than before testing. It also increases the overall
> build-and-test time.
>
> 2) Test the glibc with a different (pre-existing) compiler instead of
> with the compiler we used to build it. This is what I'd prefer, but I
> couldn't figure out how to do that without doing evil sed things to
> the generated makefile fragments.
>
> Is there a good way to test against a different compiler than the
> build compiler? Or is there a good alternative here that I'm missing?
> How do other people handle this?
My opinion is that everyone does #1.
Cheers,
Carlos.