Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 23 Apr 2017 12:37:57 +0200
From:      Tijl Coosemans <tijl@FreeBSD.org>
To:        Gerald Pfeifer <gerald@pfeifer.com>
Cc:        ports@FreeBSD.org, Martin Wilke <miwi@FreeBSD.org>
Subject:   Re: INSTALL_TARGET=install-strip runs into "permission denied"
Message-ID:  <20170423123757.5f111189@kalimero.tijl.coosemans.org>
In-Reply-To: <alpine.LSU.2.21.1704222137060.2928@anthias.pfeifer.com>
References:  <alpine.LSU.2.11.1501181210570.2527@tuna.site> <20150118130127.71b8cba9@kalimero.tijl.coosemans.org> <alpine.LSU.2.11.1501182140060.2527@tuna.site> <20150119092404.0a448f9f@kalimero.tijl.coosemans.org> <alpine.LSU.2.21.1704222137060.2928@anthias.pfeifer.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 22 Apr 2017 22:37:57 +0200 (CEST) Gerald Pfeifer <gerald@pfeifer.com> wrote:
> [ Old thread alert, but still relevant. ]
> 
> On Mon, 19 Jan 2015, Tijl Coosemans wrote:
>>>>> install   -m 555 mkheaders .../prefix/gcc5/libexec/gcc5/gcc/i386-portbld-freebsd10.1/5.0.0/install-tools/mkheaders
>>>>> test -z 'strip' || strip .../prefix/gcc5/libexec/gcc5/gcc/i386-portbld-freebsd10.1/5.0.0/install-tools/fixincl
>>>>> strip: unable to copy file '.../prefix/gcc5/libexec/gcc5/gcc/i386-portbld-freebsd10.1/5.0.0/install-tools/fixincl'; reason: Permission denied
>>>>> Makefile:191: recipe for target 'install-strip' failed
>>>>> gmake[3]: *** [install-strip] Error 1  
>>>> This strip command seems redundant.  Isn't fixincl already stripped by
>>>> the "install -s" command above?  
>>> Good point.
>>>   
>>>> What does this piece of the log look like outside the ports framework?  
>>> 
>>> /usr/bin/install -c fixinc.sh .../gcc-ref8-amd64/libexec/gcc/x86_64-unknown-freebsd8.4/5.0.0/install-tools/fixinc.sh
>>> /usr/bin/install -c fixincl .../gcc-ref8-amd64/libexec/gcc/x86_64-unknown-freebsd8.4/5.0.0/install-tools/fixincl
>>> /usr/bin/install -c mkheaders .../gcc-ref8-amd64/libexec/gcc/x86_64-unknown-freebsd8.4/5.0.0/install-tools/mkheaders
>>> test -z 'strip' || strip .../gcc-ref8-amd64/libexec/gcc/x86_64-unknown-freebsd8.4/5.0.0/install-tools/fixincl
>>> gmake[2]: Leaving directory '.../OBJ-0118-1528/fixincludes'
>>> 
>>> (I also tried setting STRIP_CMD to true, alas that is not used by GCC.)  
>> Try adding BINMODE=755 to the port Makefile or STRIP=true to CONFIGURE_ARGS  
> 
> Both of these allow the build to succeed as a regular user (non-root), 
> when INSTALL_TARGET=install-strip is set.
> 
> Alas with STRIP=true many files end up being not stripped, whereas with 
> the BINMODE setting the list is down to one file.
>   
> This appears to be the case since various aspects of GCC do not use our
> install-* tools, but a script install-sh (which you can find at the root 
> of the GCC source tree that uses cp, chmod, strip,... for compatibility 
> reasons with many systems).
> 
> So I guess setting BINMODE=755 is the best option if we want binaries
> and libraries in the gcc* ports stripped?
> 
> 
> (This is now also tracked in https://reviews.freebsd.org/D10357 where
> miwi proposed a patch originally.)

Yes, but in my opinion we should stop relying on upstream build systems
to get stripping right and let bsd.port.mk strip ELF files after staging.
It's less work for maintainers.  Then instead of stripping, bsd.port.mk
could also extract debug symbols into separate files and put them into a
debug subpackage.  Using something like this:

objcopy --only-keep-debug file file.debug
objcopy -S --add-gnu-debuglink=file.debug file



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20170423123757.5f111189>