[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20160216201324.GX9349@brightrain.aerifal.cx>
Date: Tue, 16 Feb 2016 15:13:24 -0500
From: Rich Felker <dalias@...c.org>
To: musl@...ts.openwall.com
Subject: Re: test results from latest git
On Tue, Feb 16, 2016 at 07:07:34PM +0100, Waldemar Brodkorb wrote:
> Hi,
> Szabolcs Nagy wrote,
>
> > * Szabolcs Nagy <nsz@...t70.net> [2016-02-15 22:04:52 +0100]:
> > > * Rich Felker <dalias@...c.org> [2016-02-15 15:11:22 -0500]:
> > > > On Mon, Feb 15, 2016 at 09:05:40PM +0100, Waldemar Brodkorb wrote:
> > > > > Hi,
> > > > > Szabolcs Nagy wrote,
> > > > >
> > > > > > * Waldemar Brodkorb <wbx@...nadk.org> [2016-02-15 17:45:11 +0100]:
> > > > > > >
> > > > > > > I uploaded latest test results from embedded-test here:
> > > > > > > http://tests.embedded-test.org/musl/git-14022016/
> > > > > > >
> > > > > >
> > > > > > hm, almost all armv7 float/math tests failed
> > > > > >
> > ...
> > > there is some fundamental problem with the toolchain/qemu/..
> > > and i dont yet see what.. e.g. wrong float.h was used or
> > > similarly bad thing.
> > >
> >
> > can you send
> >
> > objdump -rdw src/math/scalbnf.lo
> > objdump -rdw src/math/rintf.lo
> > gcc -dM -E - </dev/null |sort
>
> Find attached.
>
> best regards
> Waldemar
> toolchain_build_qemu-arm-vexpress-a9_musl_arm_hard_eabihf/w-musl-git-1/musl-git/obj/src/math/rintf.lo: file format elf32-littlearm
>
>
> Disassembly of section .text.rintf:
>
> 00000000 <rintf>:
> 0: e12fff1e bx lr
This was the result of massive CSE optimization which could not have
taken place without assuming that ±0.0 are equal and that addition is
associative. Based on that, I tracked down the bug to here:
http://cgit.openadk.org/cgi/cgit/openadk.git/tree/mk/vars.mk?id=479d256fcd479d3237970a6b41daf3c364d97910#n267
Rich
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
