linux-mips
[Top] [All Lists]

Re: compiling kernel 2.6.13

To: ths@networkno.de
Subject: Re: compiling kernel 2.6.13
From: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Date: Thu, 01 Sep 2005 01:22:47 +0900 (JST)
Cc: ralf@linux-mips.org, geoman@gentoo.org, linux-mips@linux-mips.org
In-reply-to: <20050831155526.GW21717@hattusa.textio>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <4315CD1C.80203@gentoo.org> <20050831153509.GF3377@linux-mips.org> <20050831155526.GW21717@hattusa.textio>
Sender: linux-mips-bounce@linux-mips.org
>>>>> On Wed, 31 Aug 2005 17:55:26 +0200, Thiemo Seufer <ths@networkno.de> said:
>> But that seems an IP22-specific problem.

ths> I _think_ it hits every 64bit kernel which uses mappings in
ths> CKSEG0.  Do you know a system where this works?

Though I do not have IP22, I think this line in mach-ip22/space.h is
inappropriate.

#define MAP_BASE                0xffffffffc0000000

It will make VMALLOC_END in pgtabe-64.h overflow.

#define VMALLOC_START           MAP_BASE
#define VMALLOC_END     \
        (VMALLOC_START + PTRS_PER_PGD * PTRS_PER_PMD * PTRS_PER_PTE * PAGE_SIZE)

Shoule we use 0xc000000000000000 as MAP_BASE for IP22 ?

---
Atsushi Nemoto

<Prev in Thread] Current Thread [Next in Thread>