[PATCH] fix #19444 - build failures with -O1 due to -Wmaybe-uninitialized

[H.J. Lu]

On Thu, Jan 14, 2016 at 10:06 AM, Joseph Myers <joseph@codesourcery.com> wrote:
> On Wed, 13 Jan 2016, H.J. Lu wrote:
>
>> > It would be good to have an updated list of issues seen.  I'd expect most
>> > such issues to be easy to fix (adding missing *_hidden_* for functions
>> > where PLT avoidance currently relies on inlining, for example).
>>
>> This is a regression:
>>
>> https://sourceware.org/bugzilla/show_bug.cgi?id=19462
>
> I think the right way to fix those _STRING_ARCH_unaligned issues is: get
> Wilco's patch <https://sourceware.org/ml/libc-alpha/2016-01/msg00170.html>
> reviewed (which we need to do for 2.23 anyway as an ABI issue for
> AArch64), then move _STRING_ARCH_unaligned to a separate non-installed
> header, not in bits/, which is included directly by the files that need
> it.  (It's possible that Wilco's patch depends on other patches to
> eliminate _STRING_ARCH_unaligned tests from installed headers except where
> they are replaced by _STRING_INLINE_unaligned.)
>

This is the right direction, but may not work with -Os.  I am working
on a patch to add bits/string-1.h to define _STRING_ARCH_unaligned,
which is included unconditionally.

-- 
H.J.