Changing compilation options for BSP building

[Jiri Gaisler]

I have now uploaded a new patch to:

ftp://gaisler.com/gaisler.com/rcc/src/rtems-4.6.5-12.diff

It restores the shared memory driver and set the CPU_MODEL_NAME
for all sparc bsp to "generic".

Please let me know if it can be accepted into the next release.

Jiri.




Ralf Corsepius wrote:
> On Sat, 2006-03-25 at 00:32 +0100, Jiri Gaisler wrote:
> 
>>I have created a patch agains rtems-4.6.5 that includes
>>the all the remaining FPU modification for the sparc port,
>>and some new drivers for leon2/3. It can be downloaded at:
>>
>>ftp://gaisler.com/gaisler.com/rcc/src/rtems-4.6.5.diff
>>
>>The patch also includes multi-processor (shared memory)
>>support for leon3. We have run rtems on a 4-cpu leon3
>>system quite successfully, all the (relevant) mp tests
>>ran OK.
> 
> 
>>Can somebody (Joel?) merge the patch into the 4.6.6 branch?
> 
> 
> I am doubts on some parts of your patches:
> 
> 1. Here, you have commented out code being shared between all BSPs and
> thereby are breaking the API:
> 
> --- rtems-4.6.5/c/src/lib/libbsp/shmdr/shm_driver.h     2003-09-04 20:53:36.000000000 +0200
> +++ rtems-4.6.5.12/c/src/lib/libbsp/shmdr/shm_driver.h  2006-02-03 18:31:26.000000000 +0100
> @@ -144,6 +144,9 @@
>  #elif defined(_AM29K)
>  #define SHM_LOCK_VALUE    0
>  #define SHM_UNLOCK_VALUE  1
> +#elif defined(__sparc__)
> +#define SHM_LOCK_VALUE    1
> +#define SHM_UNLOCK_VALUE  0
>  #elif defined(no_cpu)               /* for this values are irrelevant */
>  #define SHM_LOCK_VALUE    1
>  #define SHM_UNLOCK_VALUE  0
> @@ -499,6 +502,7 @@
>  void           Shm_Unlock( Shm_Locked_queue_Control * );
> 
>  /* MPCI entry points */
> +/*
>  rtems_mpci_entry Shm_Get_packet(
>    rtems_packet_prefix **
>  );
> @@ -517,6 +521,7 @@
>    rtems_unsigned32,
>    rtems_packet_prefix *
>  );
> +*/
> 
>  extern rtems_mpci_table MPCI_table;
> 
> @@ -524,15 +529,15 @@
> 
>  /* multiprocessor communications interface (MPCI) table */
> 
> -rtems_mpci_table MPCI_table  = {
> -  100000,                     /* default timeout value in ticks */
> -  MAX_PACKET_SIZE,            /* maximum packet size */
> -  Shm_Initialization,         /* initialization procedure   */
> -  Shm_Get_packet,             /* get packet procedure       */
> -  Shm_Return_packet,          /* return packet procedure    */
> -  Shm_Send_packet,            /* packet send procedure      */
> -  Shm_Receive_packet          /* packet receive procedure   */
> -};
> +/* rtems_mpci_table MPCI_table  = { */
> +/*   100000,                     /\* default timeout value in ticks *\/ */
> +/*   MAX_PACKET_SIZE,            /\* maximum packet size *\/ */
> +/*   Shm_Initialization,         /\* initialization procedure   *\/ */
> +/*   Shm_Get_packet,             /\* get packet procedure       *\/ */
> +/*   Shm_Return_packet,          /\* return packet procedure    *\/ */
> +/*   Shm_Send_packet,            /\* packet send procedure      *\/ */
> +/*   Shm_Receive_packet          /\* packet receive procedure   *\/ */
> +/* }; */
> 
>  #endif
> 
> 
> 2. Can you explain this patch below:
> 
> --- rtems-4.6.5/cpukit/score/cpu/sparc/rtems/score/sparc.h      2003-09-04 20:47:40.000000000 +0200
> +++ rtems-4.6.5.12/cpukit/score/cpu/sparc/rtems/score/sparc.h   2005-10-24 15:34:57.000000000 +0200
> @@ -66,11 +66,7 @@
>   */
> 
> 
> -#if defined(_SOFT_FLOAT)
> -#define SPARC_HAS_FPU 0
> -#else
>  #define SPARC_HAS_FPU 1
> -#endif
> 
>  #if SPARC_HAS_FPU
>  #define CPU_MODEL_NAME "w/FPU"
> 
> This doesn't seem right to me.
> 
> At least these code fragment can not be merged into CVS. The other parts
> of your patch only affect your BSPs and therefore shouldn't be critical 
> 
> Ralf
> 
> 
> .
>