Re: Address space limit?

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

christos@ wrote:

> 1. I think this might break existing binaries that use jemalloc because:
> 
> bool
> pages_boot(void) {
>         os_page = os_page_detect();
>         if (os_page > PAGE) {
> 		boom.

Hmm. If the different PGSHIFT causes any problem,
it has already been broken because several m68k ports
have used 8kB page:

---
sys/arch/amiga/include/param.h:#define	PGSHIFT		13		/* LOG2(NBPG) */
sys/arch/atari/include/param.h:#define	PGSHIFT		13		/* LOG2(NBPG) */
sys/arch/cesfic/include/param.h:#define	PGSHIFT		12		/* LOG2(NBPG) */
sys/arch/hp300/include/param.h:#define	PGSHIFT		12		/* LOG2(NBPG) */
sys/arch/luna68k/include/param.h:#define PGSHIFT		12		/* LOG2(NBPG) */
sys/arch/mac68k/include/param.h:#define	PGSHIFT		13		/* LOG2(NBPG) */
sys/arch/mvme68k/include/param.h:#define	PGSHIFT		12		/* LOG2(NBPG) */
sys/arch/news68k/include/param.h:#define	PGSHIFT		13		/* LOG2(NBPG) */
sys/arch/next68k/include/param.h:#define	PGSHIFT		12		/* LOG2(NBPG) */
sys/arch/sun3/include/param.h:#define	PGSHIFT		13		/* LOG2(NBPG) */
sys/arch/x68k/include/param.h:#define	PGSHIFT		12		/* LOG2(NBPG) */
---

Maybe newer jemalloc should be fixed like sparc port,
which has both 4k (sun4c/sun4m) and 8k (sun4)?

> 2. There is a comment in vmparam.h about 4K pages..

Ah, it's leftover of hp300 copies.
It should be "4K or 8K page, chosen by PGSHIFT in param.h" or so..

---
Izumi Tsutsui