From owner-freebsd-hackers Thu Mar 11 9:59:18 1999 Delivered-To: freebsd-hackers@freebsd.org Received: from apollo.backplane.com (apollo.backplane.com [209.157.86.2]) by hub.freebsd.org (Postfix) with ESMTP id 1BD9015160 for ; Thu, 11 Mar 1999 09:59:15 -0800 (PST) (envelope-from dillon@apollo.backplane.com) Received: (from dillon@localhost) by apollo.backplane.com (8.9.3/8.9.1) id JAA66670; Thu, 11 Mar 1999 09:58:55 -0800 (PST) (envelope-from dillon) Date: Thu, 11 Mar 1999 09:58:55 -0800 (PST) From: Matthew Dillon Message-Id: <199903111758.JAA66670@apollo.backplane.com> To: Greg Rowe Cc: freebsd-hackers@FreeBSD.ORG, David Greenman Subject: Re: SMP Woes References: Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG : :Bingo !!! The system is a 1 gig of memory, 4 cpu's. Maxusers down to 64 solved :the Fatal Trap problem. I'll try moving the number up in stages and see where :it breaks. I had been using 256 and a couple times 512 in testing. Thanks. : :Greg Uh huh! David, hackers, Gentlemen! This is the third person to hit this problem. It is especially nasty because the failure condition is not always something easily traceable. I, for one, would personally like to see the problem fixed and damn the BSDI compatibility. At the very least, we have to panic when the kernel's page table is overrun! Can we commit Tor.Egge's patch or something similar? I've re-included it below for reference. -Matt :To: jfieber@indiana.edu :Cc: dillon@apollo.backplane.com :Subject: Re: Problems in VM structure ? :From: Tor.Egge@fast.no :In-Reply-To: Your message of "Tue, 16 Feb 1999 09:28:32 -0500 (EST)" :References: :X-Mailer: Mew version 1.70 on Emacs 19.34.1 :Mime-Version: 1.0 :Content-Type: Text/Plain; charset=us-ascii :Content-Transfer-Encoding: 7bit :Date: Tue, 16 Feb 1999 23:11:37 +0100 :Sender: tegge@fast.no : :... : :If your system has 512 MB or more memory, you might need to increase :the kernel KVA range. ------------ Index: sys/i386/include/pmap.h =================================================================== RCS file: /home/ncvs/src/sys/i386/include/pmap.h,v retrieving revision 1.57 diff -u -r1.57 pmap.h --- pmap.h 1998/11/24 20:25:52 1.57 +++ pmap.h 1999/02/07 00:18:50 @@ -92,9 +94,9 @@ #endif #ifndef NKPDE #ifdef SMP -#define NKPDE 62 /* addressable number of page tables/pde's */ +#define NKPDE 126 /* addressable number of page tables/pde's */ #else -#define NKPDE 63 /* addressable number of page tables/pde's */ +#define NKPDE 127 /* addressable number of page tables/pde's */ #endif /* SMP */ #endif Index: sys/i386/conf/Makefile.i386 =================================================================== RCS file: /home/ncvs/src/sys/i386/conf/Makefile.i386,v retrieving revision 1.139 diff -u -r1.139 Makefile.i386 --- Makefile.i386 1999/02/14 13:56:15 1.139 +++ Makefile.i386 1999/02/15 00:24:21 @@ -56,7 +56,7 @@ CFLAGS+= -aout .endif -LOAD_ADDRESS?= F0100000 +LOAD_ADDRESS?= E0100000 DEFINED_PROF= ${PROF} .if defined(PROF) CFLAGS+= -malign-functions=4 Index: sys/i386/conf/kernel.script =================================================================== RCS file: /home/ncvs/src/sys/i386/conf/kernel.script,v retrieving revision 1.1 diff -u -r1.1 kernel.script --- kernel.script 1998/09/30 12:14:39 1.1 +++ kernel.script 1999/01/16 00:49:59 @@ -5,7 +5,7 @@ SECTIONS { /* Read-only sections, merged into text segment: */ - . = 0xf0100000 + SIZEOF_HEADERS; + . = 0xe0100000 + SIZEOF_HEADERS; .interp : { *(.interp) } .hash : { *(.hash) } .dynsym : { *(.dynsym) } -------------------- Combined with some extra kernel options options "VM_KMEM_SIZE=(128*1024*1024)" options "VM_KMEM_SIZE_MAX=(128*1024*1024)" You'll need a fairly recent version of /boot/loader in order to boot ELF kernels with larger KVA range. - Tor Egge To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message