On 07/09/16 08:52, Pavel Pisa wrote:
The smpfatal08 test is quite a hack since it re-defines some global SMP >support functions locally. In order to support this, the ARM BSPs must >place the _CPU_SMP_Start_processor() function as the only global >function into the bspsmp.c file.I expected that CFLAGS_OPTIMIZE_V += -ffunction-sections -fdata-sections in BSP and linker -Wl,--gc-section could/sould resolve that. So should I divide bspsmp.c into two files? For example, bspsmp-startcpu.c, bspsmp-init.c. What about uint32_t _CPU_SMP_Initialize(void) { uint32_t cpu_count = (uint32_t)bsp_processor_count; if ( cpu_count > 4 ) cpu_count = 4; return cpu_count; } void _CPU_SMP_Finalize_initialization( uint32_t cpu_count ) { /* Do nothing */ } void _CPU_SMP_Prepare_start_multitasking( void ) { /* Do nothing */ } void _CPU_SMP_Send_interrupt( uint32_t target_cpu_index ) { /* Generates IPI */ BCM2835_REG(BCM2836_MAILBOX_3_WRITE_SET_BASE + 0x10 * target_cpu_index) = 0x1; } should they go to the same compilation unit as _CPU_SMP_Start_processor or bspsmp-init.c separate one. Or I should not care about smpfatal08 or add required symbol to it.
It would be nice to get the smpfatal08 to work with this BSP. There is already a special case for the the qoriq BSPs in this test, so maybe we need something similar for the RPI.
-- Sebastian Huber, embedded brains GmbH Address : Dornierstr. 4, D-82178 Puchheim, Germany Phone : +49 89 189 47 41-16 Fax : +49 89 189 47 41-09 E-Mail : sebastian.hu...@embedded-brains.de PGP : Public key available on request. Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG. _______________________________________________ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel
