[U-Boot] [PATCH 3/3] Consolidate arch-specific mem_malloc_init() implementations

[Peter Tyser]

Jean-Christophe PLAGNIOL-VILLARD wrote:
> On 23:05 Fri 21 Aug     , Peter Tyser wrote:
>> Signed-off-by: Peter Tyser <ptyser at xes-inc.com>
>> ---
>>  common/dlmalloc.c      |   15 +++++++++++++++
>>  include/malloc.h       |    2 ++
>>  lib_arm/board.c        |   10 ----------
>>  lib_avr32/board.c      |   15 ++-------------
>>  lib_blackfin/board.c   |    9 ---------
>>  lib_m68k/board.c       |   14 +-------------
>>  lib_microblaze/board.c |   20 +++++---------------
>>  lib_mips/board.c       |   14 +-------------
>>  lib_nios/board.c       |   14 ++------------
>>  lib_nios2/board.c      |   14 ++------------
>>  lib_ppc/board.c        |   14 +-------------
>>  lib_sh/board.c         |    9 ---------
>>  lib_sparc/board.c      |   21 +--------------------
>>  13 files changed, 32 insertions(+), 139 deletions(-)
>>
>> diff --git a/common/dlmalloc.c b/common/dlmalloc.c
>> index f3bced4..241db8c 100644
>> --- a/common/dlmalloc.c
>> +++ b/common/dlmalloc.c
>> @@ -1520,6 +1520,21 @@ void *sbrk(ptrdiff_t increment)
>>  	return (void *)old;
>>  }
>>  
>> +#ifndef CONFIG_X86
> could we have a generic name (not arch)
> CONFIG_GENERIC_MALLOC_INIT
> 
> and then define it in the configs/arch.h

I thought the cleanest fix would be to update the x86 call to 
mem_malloc_init() to match all the other architectures', but didn't want 
to delve into the x86 arch init code.

I think adding a new CONFIG_* option to handle 1 non-conforming 
architecture is dirtier than syncing up the x86 init code in any case.

Best,
Peter