Re: [Gc] patch for gc to work on newer NetBSD releases

[Ivan Maidanski]

Hi Tsugutomo,

Committed your patch (with some modifications) to master branch (7.3alpha3). Please check.
I'll commit to release branch (gc7.2) as well after testing.

Q: Which is newer NetBSD? I guess rel 9.0, right?

Regards,
Ivan


Tue, 29 May 2012 12:08:26 +0900 tsugutomo.enami at jp.sony.com:
> Hi.
> 
> Attached patch is necessary for boehm-gc to work on newer NetBSD
> release.  Without it, for example, make check fails as attached log.
> 
> The change itself is distributed for a while as a part of pkgsrc, which
> is 3rd party software management system.
> 
> I hope this is integrated in the future release of boehm-gc.
> 
> enami.
> 
> 
> $NetBSD: patch-ac,v 1.15 2012/05/29 03:01:15 enami Exp $
> 
> --- dyn_load.c.orig	2007-12-19 02:37:12.000000000 +0300
> +++ dyn_load.c	2009-11-11 01:54:26.000000000 +0300
> @@ -78,6 +78,8 @@
> #endif
> 
> #if defined(NETBSD)
> +#   include <sys/param.h>
> +#   include <dlfcn.h>
> #   include <machine/elf_machdep.h>
> #   define ELFSIZE ARCH_ELFSIZE
> #endif
> @@ -499,6 +501,15 @@
> return(0);
> }
> if( cachedResult == 0 ) {
> +#if defined(NETBSD) && defined(RTLD_DI_LINKMAP)
> +        struct link_map *lm = NULL;
> +        int rv = dlinfo(RTLD_SELF, RTLD_DI_LINKMAP, &lm); 
> +        if (rv != 0)
> +            return (0);
> +        if (lm == NULL)
> +            return (0);
> +        cachedResult = lm;
> +#else  /* !(defined(NETBSD) && defined(RTLD_DI_LINKMAP)) */
> int tag;
> for( dp = _DYNAMIC; (tag = dp->d_tag) != 0; dp++ ) {
> if( tag == DT_DEBUG ) {
> @@ -508,6 +519,7 @@
> break;
> }
> }
> +#endif /* !(defined(NETBSD) && defined(RTLD_DI_LINKMAP)) */
> }
> return cachedResult;
> }
> 
> make  check-TESTS
> [1]   Segmentation fault (core dumped) ${dir}${tst}
> FAIL: gctest
> Leaked composite object at 0x631eb0 (tests/leak_test.c:19, sz=8, NORMAL)
> Caller at allocation:
> ##PC##= 0x4009d2
> 
> PASS: leaktest
> Final heap size is 65536
> PASS: middletest
> GC_check_heap_block: found smashed heap objects:
> 0x63ffb8 in or near object at 0x63ff90(tests/smash_test.c:21, sz=40)
> Caller at allocation:
> ##PC##= 0x400942
> GC_check_heap_block: found smashed heap objects:
> 0x63ffb8 in or near object at 0x63ff90(tests/smash_test.c:21, sz=40)
> Caller at allocation:
> ##PC##= 0x400942
> GC_check_heap_block: found smashed heap objects:
> 0x63ffb8 in or near object at 0x63ff90(tests/smash_test.c:21, sz=40)
> Caller at allocation:
> ##PC##= 0x400942
> GC_check_heap_block: found smashed heap objects:
> 0x6c6cb8 in or near object at 0x6c6c90(tests/smash_test.c:21, sz=40)
> Caller at allocation:
> ##PC##= 0x400942
> 0x63ffb8 in or near object at 0x63ff90(tests/smash_test.c:21, sz=40)
> Caller at allocation:
> ##PC##= 0x400942
> PASS: smashtest
> GC Warning: Out of Memory!  Returning NIL!
> GC Warning: Out of Memory!  Returning NIL!
> GC Warning: Out of Memory!  Returning NIL!
> PASS: hugetest
> usage: test_cpp number-of-iterations
> Assuming 10 iters
> Starting iteration 1
> [1]   Segmentation fault (core dumped) ${dir}${tst}
> FAIL: test_cpp
> ==================================
> 2 of 6 tests failed
> Please report to Hans.Boehm at hp.com
> ==================================
> *** Error code 1
> 
> Stop.
> 
> _______________________________________________
> Gc mailing list
> Gc at linux.hpl.hp.com
> http://www.hpl.hp.com/hosted/linux/mail-archives/gc/