signal/kill calls

Smith, Gene gene.smith at siemens.com
Thu Jan 20 23:44:13 UTC 2005


Smith, Gene wrote, On 1/20/2005 6:20 PM:
> Joel Sherrill <joel at OARcorp.com> wrote, On 1/18/2005 11:32 AM:
> 
>>Smith, Gene wrote:
>>
>>
>>>Smith, Gene wrote, On 1/17/2005 3:17 PM:
>>>
>>>
>>>
>>>>The application which I am porting makes a posix style signal() and 
>>>>kill() call in the code. The application itself does not define these 
>>>>function. But when I link against rtems, kill() is resolved while 
>>>>signal() is not. I did not configure my rtems install with 
>>>>--enable-posix. Anyone know right off where the kill() could be coming 
>>>>from? Don't see it in libc.a with rtems-objdump.
>>>>
>>>>-gene
>>>
>>>
>>>I reconfigured rtems with --enable-posix added and now the call to 
>>>signal() resolves OK too. Is it OK to mix posix and standard RTEMS calls 
>>>in the same application?
>>
>>
>>Most definitely. :)  It was designed long ago to do this.
>>
> 
> 
> In init.c I did not define any CONFIGURE_MAXIMUM_POSIX_... items so it 
> seem it should use defaults for these (all zero). Yet when I run the 
> code it fails due to a large (uninitialized?) value for 
> api_configuration->maximum_queued_signals in posixapi.c because 
> api_configuration never gets set to &_POSIX_Default_configuration, I think.
> 
> So I guess my question is how do you configure a system that will use 
> 99.99% Classic API and just a little posix?
> 

Never mind on the first problem above. When I recompiled init.c (with no 
changes) and linked all posix params are now zero and I don't crash.

On the second question, pretty sure all zero for posix params not be good.






More information about the users mailing list