This is the mail archive of the
mailing list for the glibc project.
Re: Another GLIBC build error with GCC6
- From: Jeff Law <law at redhat dot com>
- To: Marek Polacek <polacek at redhat dot com>, Steve Ellcey <sellcey at imgtec dot com>
- Cc: Paul Eggert <eggert at cs dot ucla dot edu>, Joseph Myers <joseph at codesourcery dot com>, Roland McGrath <roland at hack dot frob dot com>, GNU C Library <libc-alpha at sourceware dot org>
- Date: Tue, 18 Aug 2015 11:25:49 -0600
- Subject: Re: Another GLIBC build error with GCC6
- Authentication-results: sourceware.org; auth=none
- References: <1437509796 dot 19674 dot 138 dot camel at ubuntu-sellcey> <55AEB7FF dot 1000405 at redhat dot com> <1437518197 dot 19674 dot 152 dot camel at ubuntu-sellcey> <20150721224637 dot 1E0102C3B32 at topped-with-meat dot com> <55AF30A1 dot 2030402 at cs dot ucla dot edu> <1439914341 dot 16261 dot 29 dot camel at ubuntu-sellcey> <alpine dot DEB dot 2 dot 10 dot 1508181618240 dot 17603 at digraph dot polyomino dot org dot uk> <1439915115 dot 16261 dot 32 dot camel at ubuntu-sellcey> <55D365B9 dot 6070407 at cs dot ucla dot edu> <1439917980 dot 16261 dot 36 dot camel at ubuntu-sellcey> <20150818171857 dot GA2729 at redhat dot com>
On 08/18/2015 11:18 AM, Marek Polacek wrote:
And note that left shifting 1 into the sign bit may be downgraded from
undefined to implementation defined behaviour for C (it's already
changed for C++14).
On Tue, Aug 18, 2015 at 10:13:00AM -0700, Steve Ellcey wrote:
On Tue, 2015-08-18 at 10:04 -0700, Paul Eggert wrote:
Steve Ellcey wrote:
I put the old and new
code segments in a small test program and compiled it on MIPS (32 and 64
bit modes) and verified that the generated code was identical for both
versions in that setting.
That's good enough for me. I installed the patch.
Wasn't there a similar problem with DT_EXTRATAGIDX? See, for example:
For some reason, I am no longer seeing this problem. I don't see any
changes in GLIBC or GCC that look like they would have made this problem
go away so I am not sure what is going on.
If this is about shift overflow warnings:
I've changed the -Wshift-overflow warning to not warn when left-shifting 1
into the sign bit. It caused way too much noise. If you still want to see
even those, use -Wshift-overflow=2.