Problems with RTEMS-4.7.1 pc386 BSP

[Eric Norum]

On May 23, 2007, at 10:04 AM, Joel Sherrill wrote:

> Eric Norum wrote:
>> I'm surprised to find problems with the pc386 BSP.   Surely this  
>> must  be one of the more heavily used architectures. so why am I  
>> the first  to run into this problem?
>>
>>
> I checked the build logs for the 4.7 branch and the pc386 BSP for  
> changes
> since 4.7.1 -- no build errors and no changes.  The last build did  
> include
> all tests and was on 21 May.
>
> Can you generate a .map and see what happened?

This is probably a link order issue.  If the BSP library is loaded  
before the librtemscpu.a library there's no problem -- or if no  
references to __assert have been made at the point where the  
librtemscpu.a library is linked theres no problem.  My application is  
being linked with a command from the EPICS build system which might  
be the reason that different code is being pulled from the libraries  
or that the libraries are linking in a different order.

>
> It does beg the question -- why does the pc386 BSP have its own  
> assert?

I think that this is so the BSP can put up the message on the console  
and wait for a key to be pressed.  Otherwise the message flashes onto  
the screen and disappears immediately.
For now I've just removed the assert in the BSP.

-- 
Eric Norum <norume at aps.anl.gov>
Advanced Photon Source
Argonne National Laboratory
(630) 252-4793