Re: Fix mcontext_t sigcontext namespace (bug 21457)

[Florian Weimer <fweimer@redhat.com>]

On 06/27/2017 05:56 PM, Joseph Myers wrote:
> On Tue, 27 Jun 2017, Florian Weimer wrote:
> 
>> On 06/27/2017 05:27 PM, Szabolcs Nagy wrote:
>>
>>>> For some previous patches in this series, changing stack_t and ucontext_t, 
>>>> Florian checked there were no affected mangled names in Fedora, as 
>>>> evidence for such problems being rare.  Such a check might be worthwhile 
>>>> here (remembering that what is being looked for is architecture-specific, 
>>>> but on x86_64 / x86 the mangled name is already mcontext_t and would not 
>>>> change because there aren't namespace issues there).
>>>>
>>>
>>> Florian, if you can check, then please do.
>>
>> What's the old mangled name I would have to look for?  sigcontext?
> 
> On x86_64 and x86, it's mcontext_t (and won't change).  On AArch64, it's 
> sigcontext.

Only one package is affected, the Dolphin emulator (we have version 5.0
in Fedora).  It uses SContext as typedef for mcontext_t.  All symbols
are definitions, and there are no references.  I assume they are all
accidental exports, so this program would not be affected by the
mangling change.

Singular (a “Computer Algebra System for polynomial computations”) uses
a sigcontext type as well, but again only in a definition.  But this
happens on i386 (where it even brings its own definition) and x86-64,
and mcontext_t is not used, so it's not related.

I don't know if the harmless mangling change within Dolphin should be
considered a blocker.  Probably not.  But in general, I agree with the
sentiment that POSIX conformance is usually not sufficient grounds for
ABI breakage (or deviation from kernel-defined data structure layouts).

Thanks,
Florian