Skip to contentSkip to conversation

builtin/gc: correct physical memory detection for OpenBSD / NetBSD

PATCH Collin Funk 2 participants 2 messages Jun 2, 2025

lore
Share & tools
lore

Collin Funk

author Reviewed-by
#0 lore raw

Hi Brad,

Quoted 8 lines · You

OpenBSD / NetBSD use HW_PHYSMEM64 to detect the amount of physical memory in a system. HW_PHYSMEM will not provide the correct amount on a system with >=4GB of memory.

---
builtin/gc.c | 4 +++-

1 file changed, 3 insertions(+), 1 deletion(-)

I confirm this patch is correct.

I also used the following test program:

------------------------------------------------------------------------
#include <inttypes.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/sysctl.h>
#define ARRAY_SIZE(array) (sizeof array / sizeof *array) int
main (void)
{

  {
    unsigned int physmem;
    size_t len = sizeof physmem;
    static int mib[2] = { CTL_HW, HW_PHYSMEM };
    if (!(sysctl (mib, ARRAY_SIZE (mib), &physmem, &len, NULL, 0) == 0
          && len == sizeof (physmem)))
      abort ();
    printf ("HW_PHYSMEM: %jd\n", (intmax_t) physmem);
  }
  {
    int64_t physmem;
    size_t len = sizeof physmem;
    static int mib[2] = { CTL_HW, HW_PHYSMEM64 };
    if (!(sysctl (mib, ARRAY_SIZE (mib), &physmem, &len, NULL, 0) == 0
          && len == sizeof (physmem)))
      abort ();
    printf ("HW_PHYSMEM64: %jd\n", (intmax_t) physmem);

}
return 0;
}
------------------------------------------------------------------------

On NetBSD 10.0:

    $ ./a.out
    HW_PHYSMEM: 4294967295
    HW_PHYSMEM64: 17153662976

OpenBSD 7.6:

    $ ./a.out
    HW_PHYSMEM: 4286128128
    HW_PHYSMEM64: 17171030016

Thanks for the fix.

Collin

#1 lore raw
Quoted 8 lines · Collin Funk

Hi Brad,

Quoted 4 lines · You

OpenBSD / NetBSD use HW_PHYSMEM64 to detect the amount of physical memory in a system. HW_PHYSMEM will not provide the correct amount on a system with >=4GB of memory.

Quoted 8 lines
Quoted 4 lines

---
builtin/gc.c | 4 +++-

1 file changed, 3 insertions(+), 1 deletion(-)

I confirm this patch is correct.

Thanks (also thanks for the online_cpus() one).