Re: [Gc] FreeBSD/ia64 build error: ./include/private/gcconfig.h:2436:3: error: #error --> undefined ALIGNMENT

>From [email protected] Fri Apr 25 12:20:52 2014
>
>On 2014-04-24 15:41, Anton Shterenlikht wrote:
>> ===>  Building for boehm-gc-7.2e
>> /bin/sh ./libtool  --tag=CC    --mode=compile cc -DHAVE_CONFIG_H    -I./include -I./include  -I./libatomic_ops/src -I./libatomic_ops/src   -fexceptions -O2 -pipe  -fno-strict-aliasing -fno-strict-aliasing -MT allchblk.lo -MD -MP -MF .deps/allchblk.Tpo -c -o allchblk.lo allchblk.c
>> libtool: compile:  cc -DHAVE_CONFIG_H -I./include -I./include -I./libatomic_ops/src -I./libatomic_ops/src -fexceptions -O2 -pipe -fno-strict-aliasing -fno-strict-aliasing -MT allchblk.lo -MD -MP -MF .deps/allchblk.Tpo -c allchblk.c  -fPIC -DPIC -o .libs/allchblk.o
>> In file included from ./include/private/gc_priv.h:94,
>>                  from allchblk.c:17:
>> ./include/private/gcconfig.h:2436:3: error: #error --> undefined ALIGNMENT
>> *** Error code 1
>> 
>> [...]
>> 
>> I'm not sure if this is a regression, or that boehm-gc never
>> built on ia64, and some recent changes in the FreeBSD ports
>> infrastructure made lots of ports to depend on it.
>> 
>> I'm happy to try patches, or provide other log files.
>
>It never built on FreeBSD/ia64 (which is why the port is marked with
>NOT_FOR_ARCHS=ia64). See gcconfig.h [1]: the "ifdef IA64" section
>doesn't have "ifdef FREEBSD" clause in it. Perhaps you could add
>something like this to that section:
>
>  # ifdef FREEBSD
>  #  define OS_TYPE "FREEBSD"
>  #  define CPP_WORDSZ 64
>  #  define ALIGNMENT 8
>  #  define SEARCH_FOR_DATA_START
>  #  define FREEBSD_STACKBOTTOM
>  #  ifdef __ELF__
>  #   define DYNAMIC_LOADING
>  #  endif
>  # endif
>
>... and see if it builds and passes regression tests.
>
>(This is a wild guess though; even if it will compile, applications may
>still crash or leak memory; unfortunately more knowledge of the process
>memory layout in FreeBSD/ia64 is needed than I have).
>
>[1] https://github.com/ivmai/bdwgc/blob/master/include/private/gcconfig.h#L1814
>[2] https://github.com/ivmai/bdwgc/blob/master/doc/porting.html

Thank you. With your patch I get to this failure:

/bin/sh ./libtool  --tag=CC    --mode=compile cc -DHAVE_CONFIG_H    -I./include -I./include  -I./libatomic_ops/src -I./libatomic_ops/src   -fexceptions -O2 -pipe  -fno-strict-aliasing -fno-strict-aliasing -MT mark_rts.lo -MD -MP -MF .deps/mark_rts.Tpo -c -o mark_rts.lo mark_rts.c
libtool: compile:  cc -DHAVE_CONFIG_H -I./include -I./include -I./libatomic_ops/src -I./libatomic_ops/src -fexceptions -O2 -pipe -fno-strict-aliasing -fno-strict-aliasing -MT mark_rts.lo -MD -MP -MF .deps/mark_rts.Tpo -c mark_rts.c  -fPIC -DPIC -o .libs/mark_rts.o
mark_rts.c: In function 'GC_push_current_stack':
mark_rts.c:652: error: 'BACKING_STORE_BASE' undeclared (first use in this function)
mark_rts.c:652: error: (Each undeclared identifier is reported only once
mark_rts.c:652: error: for each function it appears in.)
*** Error code 1

Many thanks

Anton