nptl: Add missing err declaration __pthread_initialize_minimal_internal

  Fixes build problems on mips
nptl-init.c: In function '__pthread_initialize_minimal_internal':
nptl-init.c:333:48: error: 'err' undeclared (first use in this function)

Signed-off-by: Khem Raj <raj.khem@gmail.com>
---
 ChangeLog        | 5 +++++
 nptl/nptl-init.c | 1 +
 2 files changed, 6 insertions(+)
  

On 05/22/2015 08:23 AM, Khem Raj wrote:
> Fixes build problems on mips
> nptl-init.c: In function '__pthread_initialize_minimal_internal':
> nptl-init.c:333:48: error: 'err' undeclared (first use in this function)

I think you should move the err definition closer to its users,
introducing a new scope if necessary.  The use sites are conditional on
preprocessor symbols, and a function-global definition might result in
unused variable warnings on other architectures (which are treated as
errors).
  

On 05/22/2015 06:02 AM, Florian Weimer wrote:
> On 05/22/2015 08:23 AM, Khem Raj wrote:
>> Fixes build problems on mips
>> nptl-init.c: In function '__pthread_initialize_minimal_internal':
>> nptl-init.c:333:48: error: 'err' undeclared (first use in this function)
> 
> I think you should move the err definition closer to its users,
> introducing a new scope if necessary.  The use sites are conditional on
> preprocessor symbols, and a function-global definition might result in
> unused variable warnings on other architectures (which are treated as
> errors).
> 

Agreed. The err declaration should be right beside the syscall and in
a new scope.

Cheers,
Carlos.
  

On 05/23/2015 10:25 PM, Carlos O'Donell wrote:
> On 05/22/2015 06:02 AM, Florian Weimer wrote:
>> On 05/22/2015 08:23 AM, Khem Raj wrote:
>>> Fixes build problems on mips
>>> nptl-init.c: In function '__pthread_initialize_minimal_internal':
>>> nptl-init.c:333:48: error: 'err' undeclared (first use in this function)
>>
>> I think you should move the err definition closer to its users,
>> introducing a new scope if necessary.  The use sites are conditional on
>> preprocessor symbols, and a function-global definition might result in
>> unused variable warnings on other architectures (which are treated as
>> errors).
>>
> 
> Agreed. The err declaration should be right beside the syscall and in
> a new scope.

I see you're fixing the same thing Steve is fixing.

See his patch.

Cheers,
Carlos.