---------- Forwarded message ----------
From: Jack Howarth <howarth.at.f...@gmail.com>
Date: Tue, Oct 28, 2014 at 10:08 PM
Subject: Re: [Fink-devel] validation test for Puma-linked shared libraries
on darwin14 and later
To: Alexander Hansen <alexanderk.han...@gmail.com>
Alternatively, the Apple linker developer pointed out, we can use...
# otool -hv libgdbm.3.0.0.dylib
libgdbm.3.0.0.dylib:
Mach header
magic cputype cpusubtype caps filetype ncmds sizeofcmds flags
MH_MAGIC_64 X86_64 ALL 0x00 DYLIB 14 1536
NOUNDEFS DYLDLINK TWOLEVEL NO_REEXPORTED_DYLIBS
vs
# otool -hv libgdbm.3.0.0.dylib
libgdbm.3.0.0.dylib:
Mach header
magic cputype cpusubtype caps filetype ncmds sizeofcmds flags
MH_MAGIC_64 X86_64 ALL 0x00 DYLIB 14 1536 DYLDLINK
NO_REEXPORTED_DYLIBS
and look for the presence of TWOLEVEL.
On Tue, Oct 28, 2014 at 9:38 PM, Jack Howarth <howarth.at.f...@gmail.com>
wrote:
> As I mentioned to Alexander on fink irc, it appears that we can use...
>
> # /Library/Developer/CommandLineTools/usr/bin/dyldinfo -bind
> .libs/libgdbm.3.0.0.dylib
>
> bind information:
>
> segment section address type addend dylib
> symbol
>
> __DATA __got 0x00004000 pointer 0 libSystem
> ___stack_chk_guard
>
> __DATA __nl_symbol_ptr 0x00004008 pointer 0 libSystem
> dyld_stub_binder
>
> vs
>
> # /Library/Developer/CommandLineTools/usr/bin/dyldinfo -bind
> .libs/libgdbm.3.0.0.dylib | grep flat
>
> __DATA __got 0x00004000 pointer 0 flat-namespace
> ___stack_chk_guard
>
> __DATA __got 0x00004008 pointer 0 flat-namespace
> _gdbm_errno
>
> __DATA __nl_symbol_ptr 0x00004010 pointer 0 flat-namespace
> dyld_stub_binder
>
> and just grep the output for flat-namespace.
> Jack
>
> On Tue, Oct 28, 2014 at 9:14 PM, Alexander Hansen <
> alexanderk.han...@gmail.com> wrote:
>
>> On 10/28/14, 6:05 PM, Jack Howarth wrote:
>>
>>> As I mentioned to Alexander on fink irc, I think we can add a
>>> fatal fink validation error for the dynamic shared libraries on darwin14
>>> and later that are erroneously linked with -flat_namespace rather than
>>> -undefined dynamic_lookup due to the buggy libtool. We can use the
>>> following approach...
>>>
>>> 1) test with the 'file' command to make sure the shared library isn't
>>> really a bundle
>>> 2) if not a bundle, use 'otool -H <filename> | grep Hint' to look for...
>>>
>>> Two-level namespace hints table (0 hints)
>>>
>>>
>>> which shared libraries created with -flat_namespace should be missing.
>>>
>>>
>>> If this test fails on darwin14 or later, this should be treated as a
>>> fatal validation error with the appropriate error message.
>>>
>>> Jack
>>>
>>>
>>>
>>>
>> The good news is that Validation.pm already has a loop over the installed
>> dylibs in a .deb, excluding bundles, so adding an extra test won't be too
>> difficult.
>>
>> The bad news is that my test case (libtcd-shlibs on 10.10 with and
>> without the PatchScript) didn't seem to give different "otool -H " output
>> whether or not -flat_namespace was used.
>> --
>> Alexander Hansen, Ph.D.
>> Fink User Liaison
>> My package updates: http://finkakh.wordpress.com/
>>
>
>
------------------------------------------------------------------------------
_______________________________________________
Fink-devel mailing list
Fink-devel@lists.sourceforge.net
List archive:
http://news.gmane.org/gmane.os.apple.fink.devel
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-devel
