From patchwork Thu Dec  3 12:48:58 2015
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
X-Patchwork-Submitter: Andrew Senkevich <andrew.n.senkevich@gmail.com>
X-Patchwork-Id: 9865
Received: (qmail 50660 invoked by alias); 3 Dec 2015 12:49:36 -0000
Mailing-List: contact libc-alpha-help@sourceware.org; run by ezmlm
Precedence: bulk
List-Id: <libc-alpha.sourceware.org>
List-Unsubscribe: <mailto:libc-alpha-unsubscribe-##L=##H@sourceware.org>
List-Subscribe: <mailto:libc-alpha-subscribe@sourceware.org>
List-Archive: <http://sourceware.org/ml/libc-alpha/>
List-Post: <mailto:libc-alpha@sourceware.org>
List-Help: <mailto:libc-alpha-help@sourceware.org>,
	<http://sourceware.org/ml/#faqs>
Sender: libc-alpha-owner@sourceware.org
Delivered-To: mailing list libc-alpha@sourceware.org
Received: (qmail 50559 invoked by uid 89); 3 Dec 2015 12:49:35 -0000
Authentication-Results: sourceware.org; auth=none
X-Virus-Found: No
X-Spam-SWARE-Status: No, score=-2.4 required=5.0 tests=AWL, BAYES_00,
	FREEMAIL_FROM, RCVD_IN_DNSWL_LOW,
	SPF_PASS autolearn=ham version=3.3.2
X-HELO: mail-lb0-f181.google.com
X-Received: by 10.112.137.132 with SMTP id qi4mr5571021lbb.120.1449146967880;
	Thu, 03 Dec 2015 04:49:27 -0800 (PST)
MIME-Version: 1.0
In-Reply-To: <20151202203736.GA6861@aurel32.net>
References: 
 <CAMXFM3uVghLzryBppyfCny2vxkhW4tORyFU=PM-O4=4+qp5yEw@mail.gmail.com>
	<20151202203736.GA6861@aurel32.net>
From: Andrew Senkevich <andrew.n.senkevich@gmail.com>
Date: Thu, 3 Dec 2015 15:48:58 +0300
Message-ID: 
 <CAMXFM3sazbbYWZ-eYqm8F00oWAFjWcTuo2j8AHs_Cv5AethnLQ@mail.gmail.com>
Subject: Re: [PATCH] [BZ #19058] Better workaround for aliases of *_finite
	symbols in vector math library
To: Andrew Senkevich <andrew.n.senkevich@gmail.com>,
	libc-alpha <libc-alpha@sourceware.org>

2015-12-02 23:37 GMT+03:00 Aurelien Jarno <aurelien@aurel32.net>:
> On 2015-11-26 19:17, Andrew Senkevich wrote:
>> Hi,
>>
>> BZ #19058 shows that current workaround based on assembly aliases can
>> lead to link fail with -fopenmp and -flto options.
>> This patch makes workaround in another way to avoid such linker fails.
>>
>> 2015-11-26  Andrew Senkevich  <andrew.senkevich@intel.com>
>>
>>        [BZ #19058]
>>        * math/Makefile ($(inst_libdir)/libm.so): Added libmvec_nonshared.a to
>>        AS_NEEDED.
>>        * sysdeps/x86/fpu/bits/math-vector.h: Removed code with old
>>        workaround.
>>        * sysdeps/x86_64/fpu/Makefile (libmvec-support,
>>        libmvec-static-only-routines): Added new file.
>>        * sysdeps/x86_64/fpu/svml_finite_alias.S: New file.
>>
>> diff --git a/math/Makefile b/math/Makefile
>> index 0ff3baa..6a926e9 100644
>> --- a/math/Makefile
>> +++ b/math/Makefile
>> @@ -99,7 +99,7 @@ $(inst_libdir)/libm.so: $(common-objpfx)format.lds \
>>         (echo '/* GNU ld script'; echo '*/';\
>>          cat $<; \
>>          echo 'GROUP ( $(slibdir)/libm.so$(libm.so-version) ' \
>> -             'AS_NEEDED ( $(slibdir)/libmvec.so$(libmvec.so-version) ) )' \
>> +             'AS_NEEDED ( $(slibdir)/libmvec_nonshared.a
>> $(slibdir)/libmvec.so$(libmvec.so-version) ) )' \
>>         ) > $@
>>  endif
>
> This is wrong. libmvec_nonshared.a is installed in $(libdir) and not in
> $(slibdir). Thus linking fails.

Thank you, unfortunately it was not found on x86-64 testing.

        * math/Makefile ($(inst_libdir)/libm.so): Correct path for

 endif


Is it ok for trunk and 2.22 branch?
---
WBR,
Andrew

diff --git a/math/Makefile b/math/Makefile
index 6a926e9..20f90d9 100644
--- a/math/Makefile
+++ b/math/Makefile
@@ -99,7 +99,7 @@ $(inst_libdir)/libm.so: $(common-objpfx)format.lds \
        (echo '/* GNU ld script'; echo '*/';\
         cat $<; \
         echo 'GROUP ( $(slibdir)/libm.so$(libm.so-version) ' \
-             'AS_NEEDED ( $(slibdir)/libmvec_nonshared.a
$(slibdir)/libmvec.so$(libmvec.so-version) ) )' \
+             'AS_NEEDED ( $(libdir)/libmvec_nonshared.a
$(slibdir)/libmvec.so$(libmvec.so-version) ) )' \
        ) > $@