Re: PING^N: [PATCH] Add --enable-static-pie to build static PIE [BZ #19574]

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Mon, Nov 20, 2017 at 6:17 AM, Joseph Myers <joseph@codesourcery.com> wrote:
> On Sun, 19 Nov 2017, H.J. Lu wrote:
>
>> > about the remaining tasks listed in it?
>>
>> Target maintainers must work with binutils maintainers to resolve them
>> since static PIE may require linker bug fixes.
>
> A binutils segfault / BFD internal error is always a bug, even on invalid
> input.  Those should all have binutils bugs filed (if present with current
> binutils master) and those bugs referenced in the commit message.

I have filed and listed 2 of linker bugs in my commit message.  There may be
more once target maintainers start working on static PIE.

> If this patch reaches consensus for commit, the status information for
> architectures that don't build, or that build but have not been verified
> to have execution tests work in this configuration, will need to go on a
> wiki page linked from https://sourceware.org/glibc/wiki/PortStatus (or on
> that page itself), describing how to verify if the feature works (given
> that it builds) and what's needed to fix things if there is a build
> problem.  Then that architecture information can be removed as and when
> someone gets things working, and verifies that they are working, on
> particular architectures.

I will do that once we have reached consensus for commit.  It is very unlikely
that non-x86 targets will have working static PIE before glibc 2.27 is branched.
I'd like to get my patch in and developers can work on it.  Are there
any objections
to enable static PIE in glibc 2.27 even if it is only verified to work on x86?

Thanks.

-- 
H.J.