re: Some pkgsrc observations

John Klos writes:
> If anyone wants to figure out a VAX lockup that's still happening even 
> with NetBSD 10, try to compile pkgsrc/parallel/openpa-1.0.4. On both real 
> VAX hardware and simh, the system eventually locks up.

i'll see what happens.

> Also, does anyone know why net/bind918 fails on VAX when it compiles fine 
> on other CPUs?
>
> In file included from mem.c:60:
> jemalloc_shim.h:81:1: error: conflicting types for 'sallocx'
>     81 | sallocx(void *ptr, int flags) {
>        | ^~~~~~~
> In file included from jemalloc_shim.h:77,
>                   from mem.c:60:
> /usr/include/malloc.h:39:8: note: previous declaration of 'sallocx' was 
> here
>     39 | size_t sallocx(const void *, int);
>        |        ^~~~~~~

odd.  vax does use old jemalloc, but this is <malloc.h>, which
is the same, and both old/new have the same sallocx().

looks like the jemalloc_shim.h is just wrong.  does it have
some conditionals in it that would make it otherwise for other
platforms?


.mrg.