execvpe and glibc 2.10

[Joseph D. Darcy]

Martin Buchholz wrote:
> Joe, thanks for the review.
>   

Thanks for the fixes :-)

-Joe

> I've pushed fixes for this problem to openjdk6 and openjdk7/tl
>
> Martin
>
> On Wed, Jul 29, 2009 at 21:22, Joseph D. Darcy<Joe.Darcy at sun.com> wrote:
>   
>> Martin Buchholz wrote:
>>     
>>> Of course, it's all my fault.
>>> First, for having used a symbol that libc implementers are likely to add.
>>> Second, for actually asking glibc implementers to add that very symbol.
>>> Third, for forgetting that this is an issue in openjdk6 as well.
>>>
>>> Anyways, I intend to commit these patches to their respective forests:
>>>
>>> http://cr.openjdk.java.net/~martin/webrevs/openjdk6/rename-execvpe/
>>> http://cr.openjdk.java.net/~martin/webrevs/openjdk7/rename-execvpe/
>>>
>>> As ever, we need a Sun bug (could Michael or Joe file it?):
>>>
>>> Synopsis: Rename execvpe to avoid symbol clash with glibc 2.10.
>>> Description:
>>> glibc 2.10 added the long-awaited "missing link" function execvpe
>>> (thank you!  No really!)
>>> But the JDK already has a function of that name, which needs renaming,
>>> to avoid a compile time failure in UNIXProcess_md.c
>>> Evaluation: Yup
>>>
>>> Thanks,
>>>
>>>       
>> I've filed bug 6866719 and I approve the fix going back into 7 and OpenJDK
>> 6.
>>
>> Cheers,
>>
>> -Joe
>>
>>     
>>> Martin
>>>
>>> On Wed, Jul 29, 2009 at 08:24, Andrew John
>>> Hughes<gnu_andrew at member.fsf.org> wrote:
>>>
>>>       
>>>> 2009/7/28 Martin Buchholz <martinrb at google.com>:
>>>>
>>>>         
>>>>> On Tue, Jul 28, 2009 at 06:08, Michael McMahon<Michael.McMahon at sun.com>
>>>>> wrote:
>>>>>
>>>>>           
>>>>>> Andrew John Hughes wrote:
>>>>>>
>>>>>>             
>>>>>>> 2009/7/28 Michael McMahon <Michael.McMahon at sun.com>:
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>>>> Andrew John Hughes wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>>>> 2009/7/9 Martin Buchholz <martinrb at google.com>:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                   
>>>>>>> rename-execvpe is the one I'm particularly concerned about.  It's a
>>>>>>> trivial patch, but without it, OpenJDK builds are going to start
>>>>>>> failing as distros move to the new glibc (e.g. Fedora 12).  It's
>>>>>>> already an issue for users of Fedora rawhide.
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>> Ok. Maybe we can push rename-execvpe and RESTARTABLE first.
>>>>>>
>>>>>>             
>>>>> RESTARTABLE depends on vfork-exec, and I would not like to do
>>>>> the work to reorder them.
>>>>>
>>>>> Changes to fork-exec are always high-risk,
>>>>> and I would like to see some more testing on these
>>>>> before committing to openjdk proper.
>>>>> That could be done by having icedtea7 import them,
>>>>> and by having Michael or another Sun person run them
>>>>> through Sun testing.
>>>>>
>>>>> There are more changes to fork-exec to come,
>>>>> although they will probably not affect the average Linux user's
>>>>> experience.
>>>>>
>>>>> Michael and I have been doing other things the past few weeks.
>>>>>
>>>>> Martin
>>>>>
>>>>>
>>>>>           
>>>> Sure, we can do some testing with IcedTea7 after the release for M4,
>>>> which should be sometime in the next week or so.
>>>>
>>>> In the meantime, can you or me push the execvpe cleanup to the 6 and 7
>>>> forests (tl presumably for 7)?  Or is there a further issue there?
>>>> You've not mentioned it in either reply.
>>>>
>>>> Joe, I assume this is okay for 6?  Without it the build is broken on
>>>> newer distributions.
>>>>
>>>> Thanks,
>>>> --
>>>> Andrew :-)
>>>>
>>>> Free Java Software Engineer
>>>> Red Hat, Inc. (http://www.redhat.com)
>>>>
>>>> Support Free Java!
>>>> Contribute to GNU Classpath and the OpenJDK
>>>> http://www.gnu.org/software/classpath
>>>> http://openjdk.java.net
>>>>
>>>> PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
>>>> Fingerprint: F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8
>>>>
>>>>
>>>>         
>>