From owner-freebsd-smp Sun Jun 22 19:22:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA05315 for smp-outgoing; Sun, 22 Jun 1997 19:22:03 -0700 (PDT) Received: from fredriks-1.pr.mcs.net (fredriks-1.pr.mcs.net [205.164.50.241]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA05310 for ; Sun, 22 Jun 1997 19:22:00 -0700 (PDT) Received: (from lars@localhost) by fredriks-1.pr.mcs.net (8.8.5/8.6.6) id VAA00305; Sun, 22 Jun 1997 21:20:58 -0500 (CDT) From: Lars Fredriksen Message-Id: <199706230220.VAA00305@fredriks-1.pr.mcs.net> Subject: SMP kernel hung...more info To: dyson@dyson.iquest.net Date: Sun, 22 Jun 1997 21:20:58 -0500 (CDT) Cc: root@dyson.iquest.net, freebsd-smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi John, I just booted the SMP kernel and broke into the debugger when it got hung (after saying that all 2 cpus are online). Well it isn't actually the kernel per se that is hung. It is in the smp_idleloop. The problem is that one of the rc scripts is running swapon and it is hung - on wait channel 0xf0b60a00 (i'll try to find out what this is) I have 2 scsi controllers in my system; a 2940 and a 2944. Here is the config file: Anything here ring a bell?? Lars # # GENERICAH -- Generic machine with WD/AHx/NCR family disks # # $Id: GENERIC,v 1.47 1995/08/18 11:26:16 jkh Exp $ # machine "i386" #cpu "I386_CPU" #cpu "I486_CPU" cpu "I586_CPU" cpu "I686_CPU" ident "LUDVIG_30_PRO_SMP" maxusers 30 options INET #InterNETworking options FFS #Berkeley Fast File System options NFS #Network File System options MSDOSFS #Msdos file system options "CD9660" #CD ISO9660 File System options PROCFS #Process File System options DEVFS #Device File System options "COMPAT_43" #Compatible with BSD 4.3 options PERFMON #Pentium Pro performance counters options MROUTING # Multicast routing options "SCSI_DELAY=15" #Be pessimistic about Joe SCSI device options SCSIDEBUG #Turn on debugging ability for SCSI options UCONSOLE #X Console support options "FAT_CURSOR" #block cursor in syscons or pccons options DDB #Kernel debugger options USER_LDT #allow user-level control of i386 ldt options "IBCS2" #Run SCO stuff options "MAXMEM=131072" #128Mb options "AUTO_EOI_1" options COMPAT_LINUX options KTRACE options LINUX options USERCONFIG #boot -c editor options VISUAL_USERCONFIG #visual boot -c editor options SYSVSHM options SYSVSEM options SYSVMSG options IPFIREWALL options AHC_TAGENABLE options AHC_SCBPAGING_ENABLE options AHC_ALLOW_MEMIO # Mandatory: options SMP # Symmetric MultiProcessor Kernel options APIC_IO # Symmetric (APIC) I/O # Useful: options SMP_AUTOSTART # start the additional CPUs during boot # Optional, these are the defaults: #options NCPU=2 # number of CPUs #options NBUS=4 # number of busses #options NAPIC=1 # number of IO APICs #options NINTR=24 # number of INTs # Currently unusable: #options SMP_PRIVPAGES # BROKEN: architecture problem # # Rogue SMP hardware: # # Tyan Tomcat II: #options SMP_TIMER_NC # 8254 NOT connected to APIC # SuperMicro P6DNE: options SMP_TIMER_NC # 8254 NOT connected to APIC config kernel root on sd0 controller isa0 controller pci0 controller fdc0 at isa? port "IO_FD1" bio irq 6 drq 2 vector fdintr disk fd0 at fdc0 drive 0 disk fd1 at fdc0 drive 1 controller ahc0 controller ahc1 controller scbus0 at ahc0 controller scbus1 at ahc1 disk sd0 at scbus1 target 0 unit 0 #disk sd1 at scbus0 target 1 tape st0 at scbus1 target 2 tape st1 at scbus1 target 5 device sd0 #Only need one of these, the code dynamically grows device st0 device cd0 device de0 device sc0 at isa? port "IO_KBD" tty irq 1 vector scintr #device mse0 at isa? port 0x23c tty irq 9 vector mseintr device npx0 at isa? port "IO_NPX" irq 13 vector npxintr device sio0 at isa? port "IO_COM1" tty irq 4 vector siointr device sio1 at isa? port "IO_COM2" tty irq 3 vector siointr device lpt0 at isa? port? tty irq 5 vector lptintr # Other standard PC hardware: `lpt', `mse', `psm', `sio', etc. # psm: PS/2 mouse port [note: conflicts with sc0/vt0, thus "conflicts" keywd] device psm0 at isa? port "IO_KBD" conflicts tty irq 12 vector psmintr # Options for psm: # options PSM_NO_RESET for the `psm' driver device ed0 at isa? port 0x300 net irq 10 iomem 0xd8000 vector edintr # Sound device drivers for Gus PnP Pro controller snd0 device gus0 at isa? vector gusintr # dgb: Digiboard PC/Xi and PC/Xe series driver (ALPHA QUALITY!) #device dgb0 at isa? port 0x110 flags 2 irq 15 iomem 0xfc0000 iosiz 16385 tty pseudo-device loop pseudo-device ether pseudo-device log pseudo-device sl 2 pseudo-device ppp 2 pseudo-device tun 2 pseudo-device vn pseudo-device bpfilter 1 pseudo-device pty 32 pseudo-device snp 2 pseudo-device speaker -- ------------------------------------------------------------------- Lars Fredriksen fredriks@mcs.com (home) lars@fredriks-1.pr.mcs.net (home-home) From owner-freebsd-smp Mon Jun 23 10:41:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA17113 for smp-outgoing; Mon, 23 Jun 1997 10:41:26 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA17107 for ; Mon, 23 Jun 1997 10:41:20 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id LAA06459; Mon, 23 Jun 1997 11:40:46 -0600 (MDT) Message-Id: <199706231740.LAA06459@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: david.myers@Corp.Sun.COM cc: freebsd-smp@FreeBSD.ORG Subject: Re: SMP current status? In-reply-to: Your message of "Fri, 20 Jun 1997 10:38:07 PDT." <199706201738.KAA25205@concord.Corp.Sun.COM> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 11:40:46 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > I'm about to install a second CPU in my Tyan Pentium Pro motherboard > and go for the FreeBSD 3.0 upgrade. The web pages on freebsd.org are > looking a bit dated, this is primarily because SMP is now part of 3.0-current and thus doesn't require as much special information to be able to use it. --- > however, so I thought I'd ask: how usable is > FreeBSD-SMP, for a person who's not necessarily a kernel hacker but > does like to tinker? It builds cleanly right out of the box, in most cases no special skills are needed. --- > How stable? hard to quantify, but it builds world and runs for weeks on most systems without problems. --- > Does all FreeBSD 2.2.2 code run? probably not, but this is true for the 3.0 UP kernel as well. the bigger issue is 2.2 vs. 3.0, not SMP --- > I'm wondering about Open Sound System, Linux compatibility, and other > dynamically-loaded libraries, in particular. I typically don't use any of these, except perhaps a few lkms that I'mn unaware of... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jun 23 10:50:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA17946 for smp-outgoing; Mon, 23 Jun 1997 10:50:49 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA17939 for ; Mon, 23 Jun 1997 10:50:46 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id LAA06493; Mon, 23 Jun 1997 11:50:00 -0600 (MDT) Message-Id: <199706231750.LAA06493@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Lars Fredriksen cc: freebsd-smp@FreeBSD.ORG Subject: Re: Booting SMP on SuperMicro P6DNE fails In-reply-to: Your message of "Sat, 21 Jun 1997 00:01:40 CDT." <199706210501.AAA12670@Mercury.mcs.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 11:49:59 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > Building a current kernel (as of last night) with the following > options: > SMP > APIC_IO > SMP_AUTOSTART > SMP_TIMER_NC > > and without I386_CPU and I486_CPU failed to boot. I booted to the point > where you get the TADA message saying that the second processor was > enabled and added to the scheduling queue and the issued a message about > it having found all the cpus. Then it just sat there. I will get a break > into the debugger tomorrow at that point to get a stack backtrace. really need to see the complete printout of the boot. best way to get this is to setup the machine to boot via a serial port, then direct that port thru another machine inside of a tip session under X. you can then cutNpaste the output into a file for examination. having said that this sounds like a missing INTerrupt problem. If I had to guess I would say the disk controller. is it SCSI or IDE? --- > Are there drivers in the 3.0 tree that are not SMP safe? Like the sound > drivers etc?? theoretically they should all be SMP safe, since we use a global "giant lock". but there undoubtedly are a few that do something 'unusual' that could get them in trouble... The general rule in your type problem is to first get a minimal kernel working that has ONLY the drivers you need for booting. then add the others (such as sound, etc.) -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jun 23 11:00:54 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA18484 for smp-outgoing; Mon, 23 Jun 1997 11:00:54 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA18474 for ; Mon, 23 Jun 1997 11:00:49 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id MAA06541; Mon, 23 Jun 1997 12:00:30 -0600 (MDT) Message-Id: <199706231800.MAA06541@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Lars Fredriksen cc: dyson@dyson.iquest.net, freebsd-smp@FreeBSD.ORG Subject: Re: SMP kernel hung...more info In-reply-to: Your message of "Sun, 22 Jun 1997 21:20:58 CDT." <199706230220.VAA00305@fredriks-1.pr.mcs.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 12:00:30 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > Hi John, > I just booted the SMP kernel and broke into the debugger when > it got hung (after saying that all 2 cpus are online). Well it isn't > actually the kernel per se that is hung. It is in the smp_idleloop. > > The problem is that one of the rc scripts is running swapon and it > is hung - on wait channel 0xf0b60a00 (i'll try to find out what this > is) > > I have 2 scsi controllers in my system; a 2940 and a 2944. > > Here is the config file: > > Anything here ring a bell?? first step is to get a kernel minimal running. I've marked the things to NUKE for this goal. Once you get something that boots, start adding them back one at a time til you identify the problem driver/feature: > options PERFMON #Pentium Pro performance counters > options MROUTING # Multicast routing > options USER_LDT #allow user-level control of i386 ldt > options "IBCS2" #Run SCO stuff > options "AUTO_EOI_1" > options COMPAT_LINUX > options LINUX > options AHC_TAGENABLE > options AHC_SCBPAGING_ENABLE > options AHC_ALLOW_MEMIO delete these for test kernel. --- > # SuperMicro P6DNE: > options SMP_TIMER_NC # 8254 NOT connected to APIC I assume you are using a supermicro? which specific model? --- > # psm: PS/2 mouse port [note: conflicts with sc0/vt0, thus "conflicts" keywd] > device psm0 at isa? port "IO_KBD" conflicts tty irq 12 vector psmintr delete this for test kernel. it has caused problems for some users in the past. --- > # Sound device drivers for Gus PnP Pro > controller snd0 > > device gus0 at isa? vector gusintr delete this for test kernel. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jun 23 12:39:54 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA25274 for smp-outgoing; Mon, 23 Jun 1997 12:39:54 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA25263 for ; Mon, 23 Jun 1997 12:39:49 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id NAA07051; Mon, 23 Jun 1997 13:39:06 -0600 (MDT) Message-Id: <199706231939.NAA07051@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Lars Fredriksen cc: smp@freebsd.org Subject: Re: SMP kernel hung...more info In-reply-to: Your message of "Mon, 23 Jun 1997 13:39:31 CDT." <199706231839.NAA00727@fredriks-1.pr.mcs.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 13:39:05 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, Lars has been struggling with his SMP system and just sent me his mptable output. It has something bogus in it, am wondering if anyone else with a supermicro has a similar table. specifically: I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# INT active-lo level 0 19:A 2 16 INT active-lo level 0 18:A 2 16 INT active-lo level 0 17:A 2 16 INT active-lo level 0 19:A 2 17 INT active-lo level 0 18:A 2 17 INT active-lo level 0 17:A 2 17 INT active-lo level 0 19:A 2 18 INT active-lo level 0 18:A 2 18 INT active-lo level 0 17:A 2 18 INT active-lo level 0 20:A 2 19 note that 19:A is declared to be attached to apic#2, pins 16, 17 and 18. similarily, 18:A is attached to pins 16,17,18. similarily, 17:A is attached to pins 16,17,18. This is totally BOGUS. more than one slot can be on an APIC pin (shared INTs) but each slot should NOT be attached to more than one APIC pin! if this were possible you would trigger three different INT vectors each time a device generated an INT. does anyone have any theories to explain this? does anyone have an mptable with similar entries, ie a slot that is claimed to be attached to more than one APIC pin? --- Lars, the current code just isn't going to work with this mptable. I suggest you don't waste further time trying. We need to talk to someone at supermicro tech support about this to find out whats up. do you have any contacts there? -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jun 23 13:18:46 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA27684 for smp-outgoing; Mon, 23 Jun 1997 13:18:46 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA27672 for ; Mon, 23 Jun 1997 13:18:33 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id OAA07315; Mon, 23 Jun 1997 14:18:19 -0600 (MDT) Message-Id: <199706232018.OAA07315@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org cc: mef@cs.washington.edu Subject: SMP_PRIVPAGES Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 14:18:18 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, Peter has committed the code for enabling SMP_PRIVPAGES. Not that it is automatic, ie you don't need to add the option to your config file. I successfully used it to "make world" with -current as of earlier this morning. Thanx, Peter! -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jun 23 14:55:42 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA03253 for smp-outgoing; Mon, 23 Jun 1997 14:55:42 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA03248 for ; Mon, 23 Jun 1997 14:55:39 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id PAA07810; Mon, 23 Jun 1997 15:55:04 -0600 (MDT) Message-Id: <199706232155.PAA07810@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: mef@cs.washington.edu cc: smp@freebsd.org Subject: Re: SMP_PRIVPAGES In-reply-to: Your message of "Mon, 23 Jun 1997 14:27:59 PDT." <199706232127.OAA09034@wile-e-coyote.cs.washington.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 15:55:04 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > I couldn't parse the following sentence in your email message: > > "Not that it is automatic, ie you don't need to add the option to your ^^^ > config file." > > Does one have to add the option or not? What's not automatic about that should have been: Note that it is automatic, ie you don't need to add the option to your config file. --- > it? How does it affect the system structure? I.e., does the system > still make a call to some assembly code to figure out its CPU ID? no, thats in a CPU private page, both as a number, and as the 24<< mask. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jun 23 15:20:53 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA04410 for smp-outgoing; Mon, 23 Jun 1997 15:20:53 -0700 (PDT) Received: from drawbridge.ascend.com (drawbridge.ascend.com [198.4.92.1]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id PAA04405 for ; Mon, 23 Jun 1997 15:20:51 -0700 (PDT) Received: from spud.ascend.com (fw-ext.ascend.com [198.4.92.5]) by drawbridge.ascend.com (8.6.12/8.6.12) with ESMTP id PAA13278 for ; Mon, 23 Jun 1997 15:20:50 -0700 Received: from ascend.com by ascend.com with ESMTP id PAA00338 for ; Mon, 23 Jun 1997 15:20:42 -0700 Received: from ascend.com by ascend.com id PAA07654; Mon, 23 Jun 1997 15:19:11 -0700 Message-Id: <2.2.32.19970623222250.00c036a8@sales_srv.ascend.com> X-Sender: jzeitlin@sales_srv.ascend.com X-Mailer: Windows Eudora Pro Version 2.2 (32) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 23 Jun 1997 15:22:50 -0700 To: smp@FreeBSD.ORG From: Jim Zeitlin Subject: Newbie query Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Greetings, Pardon the use of ( possibly ) excessive list bandwidth but I have some newbie questions. I am exploring some communications server development work and was looking into using the Intel PR440FX dual Pentium Pro motherboard. Do systems built around this board generally run both the FreeBSD uni-processor release and the current FreeBSD SMP release ? ( Not at the same time ! ) Are there working drivers for this boards integrated SCSI and 10/100BaseT ethernet controllers for these kernels ? The marketing literature describes the SCSI controller "Adaptec 2940UW (7880) SCSI Controller" and the ethernet as an "Ether Express". Thanks in advance. - Jim Zeitlin jzeitlin@ascend.com From owner-freebsd-smp Mon Jun 23 15:30:44 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA04648 for smp-outgoing; Mon, 23 Jun 1997 15:30:44 -0700 (PDT) Received: from drawbridge.ascend.com (drawbridge.ascend.com [198.4.92.1]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id PAA04641 for ; Mon, 23 Jun 1997 15:30:41 -0700 (PDT) Received: from spud.ascend.com (fw-ext.ascend.com [198.4.92.5]) by drawbridge.ascend.com (8.6.12/8.6.12) with ESMTP id PAA13727 for ; Mon, 23 Jun 1997 15:30:40 -0700 Received: from ascend.com by ascend.com with ESMTP id PAA00886 for ; Mon, 23 Jun 1997 15:30:37 -0700 Received: from ascend.com by ascend.com id PAA07695; Mon, 23 Jun 1997 15:29:07 -0700 Message-Id: <2.2.32.19970623223245.00c3c850@sales_srv.ascend.com> X-Sender: jzeitlin@sales_srv.ascend.com X-Mailer: Windows Eudora Pro Version 2.2 (32) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 23 Jun 1997 15:32:45 -0700 To: freebsd-smp@FreeBSD.ORG From: Jim Zeitlin Subject: Newbie query Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Greetings, Pardon the use of ( possibly ) excessive list bandwidth but I have some newbie questions. I am exploring some communications server development work and was looking into using the Intel PR440FX dual Pentium Pro motherboard. Do systems built around this board generally run both the FreeBSD uni-processor release and the current FreeBSD SMP release ? ( Not at the same time ! ) Are there working drivers for this boards integrated SCSI and 10/100BaseT ethernet controllers for these kernels ? The marketing literature describes the SCSI controller "Adaptec 2940UW (7880) SCSI Controller" and the ethernet as an "Ether Express". Thanks in advance. - Jim Zeitlin jzeitlin@ascend.com From owner-freebsd-smp Mon Jun 23 15:41:47 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA05397 for smp-outgoing; Mon, 23 Jun 1997 15:41:47 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id PAA05387 for ; Mon, 23 Jun 1997 15:41:34 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.5/8.8.5) id RAA01295; Mon, 23 Jun 1997 17:41:14 -0500 (EST) From: "John S. Dyson" Message-Id: <199706232241.RAA01295@dyson.iquest.net> Subject: Re: Newbie query In-Reply-To: <2.2.32.19970623222250.00c036a8@sales_srv.ascend.com> from Jim Zeitlin at "Jun 23, 97 03:22:50 pm" To: jzeitlin@ascend.com (Jim Zeitlin) Date: Mon, 23 Jun 1997 17:41:14 -0500 (EST) Cc: smp@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Greetings, > > Pardon the use of ( possibly ) excessive list bandwidth but I have > some newbie questions. > > I am exploring some communications server development work and was > looking into using the Intel PR440FX dual Pentium Pro motherboard. Do > systems built around this board generally run both the FreeBSD uni-processor > release and the current FreeBSD SMP release ? ( Not at the same time ! ) Are > there working drivers for this boards integrated SCSI and 10/100BaseT > ethernet controllers for these kernels ? The marketing literature describes > the SCSI controller "Adaptec 2940UW (7880) SCSI Controller" and the ethernet > as an "Ether Express". > I haven't checked the latest -current on that machine, but it does/has worked. use 'ahc0' for the SCSI, and fxp0 for the ethernet. I am currently running that machine with SMP-current as of a week or so ago. There is a bit of a trick that you need to select APIC interrupts or somesuch in the bios. John From owner-freebsd-smp Mon Jun 23 18:59:21 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id SAA13948 for smp-outgoing; Mon, 23 Jun 1997 18:59:21 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id SAA13938 for ; Mon, 23 Jun 1997 18:59:09 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id TAA08592 for ; Mon, 23 Jun 1997 19:59:05 -0600 (MDT) Message-Id: <199706240159.TAA08592@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@FreeBSD.ORG Subject: Re: SMP_PRIVPAGES In-reply-to: Your message of "Mon, 23 Jun 1997 14:18:18 MDT." <199706232018.OAA07315@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 19:59:05 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, I said: > Peter has committed the code for enabling SMP_PRIVPAGES. Not that it ^^^ > is automatic, ie you don't need to add the option to your config file. this should have been: Note that it is automatic, ie you don't need to add the option to your ^^^^ config file. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jun 23 19:38:05 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA15786 for smp-outgoing; Mon, 23 Jun 1997 19:38:05 -0700 (PDT) Received: from luke.pmr.com (luke.pmr.com [207.170.114.132]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA15759 for ; Mon, 23 Jun 1997 19:37:56 -0700 (PDT) Received: (from bob@localhost) by luke.pmr.com (8.8.5/8.7.3) id VAA01043; Mon, 23 Jun 1997 21:37:46 -0500 (CDT) Message-ID: <19970623213745.30803@luke.pmr.com> Date: Mon, 23 Jun 1997 21:37:45 -0500 From: Bob Willcox To: SMP list Subject: Recent (last two days) smp kernel is hanging for me Reply-To: Bob Willcox Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.76e Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk For the couple of days kernels built on my ASUS P/I-P65UP5 w/C-P6ND have been hanging (a kernel built on 6/19 boots fine, nothing changed in the config file). With recently cvsup'd -current sources my kernels will no longer boot all the way. They hang right after printing out the following messages: ncr0 scanning for targets 0..6 (V2 pl24 96/12/14) ncr0: waiting for scsi devices to settle scbus0 at ncr0 bus 0 --- HANGS HERE --- Again, my old 6/19 kernel runs right on past this point, printing out the scsi targets found. Anybody else seeing anything like this? Is there any additional info that I can provide to assist in diagnosing this? Thanks, -- Bob Willcox Deliberation, n.: The act of examining one's bread to determine bob@luke.pmr.com which side it is buttered on. Austin, TX -- Ambrose Bierce, "The Devil's Dictionary" From owner-freebsd-smp Mon Jun 23 20:10:53 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA17610 for smp-outgoing; Mon, 23 Jun 1997 20:10:53 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA17604 for ; Mon, 23 Jun 1997 20:10:49 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.5/8.8.5) id WAA01893; Mon, 23 Jun 1997 22:10:35 -0500 (EST) From: "John S. Dyson" Message-Id: <199706240310.WAA01893@dyson.iquest.net> Subject: Re: Recent (last two days) smp kernel is hanging for me In-Reply-To: <19970623213745.30803@luke.pmr.com> from Bob Willcox at "Jun 23, 97 09:37:45 pm" To: bob@luke.pmr.com Date: Mon, 23 Jun 1997 22:10:35 -0500 (EST) Cc: freebsd-smp@FreeBSD.ORG Reply-To: dyson@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > Again, my old 6/19 kernel runs right on past this point, printing > out the scsi targets found. > > Anybody else seeing anything like this? Is there any additional > info that I can provide to assist in diagnosing this? > I am having the same problem on my P6DNF. The P6 Intel SMP motherboard seems to work okay though. Must have something to do with the way the chipsets are config'ed. I am trying to help find out what is wrong now, but haven't made any progress yet. John From owner-freebsd-smp Mon Jun 23 20:18:48 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA17853 for smp-outgoing; Mon, 23 Jun 1997 20:18:48 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA17847 for ; Mon, 23 Jun 1997 20:18:43 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.5/8.8.5) id WAA01939; Mon, 23 Jun 1997 22:18:26 -0500 (EST) From: "John S. Dyson" Message-Id: <199706240318.WAA01939@dyson.iquest.net> Subject: Re: SMP kernel hung...more info In-Reply-To: <199706231939.NAA07051@Ilsa.StevesCafe.com> from Steve Passe at "Jun 23, 97 01:39:05 pm" To: smp@csn.net (Steve Passe) Date: Mon, 23 Jun 1997 22:18:26 -0500 (EST) Cc: lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG Reply-To: dyson@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Hi, > > Lars has been struggling with his SMP system and just sent me his mptable > output. It has something bogus in it, am wondering if anyone else with a > supermicro has a similar table. specifically: > > I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# > INT active-lo level 0 19:A 2 16 > INT active-lo level 0 18:A 2 16 > INT active-lo level 0 17:A 2 16 > INT active-lo level 0 19:A 2 17 > INT active-lo level 0 18:A 2 17 > INT active-lo level 0 17:A 2 17 > INT active-lo level 0 19:A 2 18 > INT active-lo level 0 18:A 2 18 > INT active-lo level 0 17:A 2 18 > INT active-lo level 0 20:A 2 19 > > note that 19:A is declared to be attached to apic#2, pins 16, 17 and 18. > similarily, 18:A is attached to pins 16,17,18. > similarily, 17:A is attached to pins 16,17,18. > Here is a portion of my mptable output. It appears that my system also has an "interesting" mptable. It is a Supermicro P6DNF. Does this mean that we might have to write some interrupt routing code??????? I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# INT conforms conforms 1 15 2 20 INT active-lo level 0 18:A 2 17 INT active-lo level 0 20:A 2 18 INT active-lo level 0 19:A 2 18 INT active-lo level 0 20:A 2 19 INT active-lo level 0 19:A 2 19 SMI conforms conforms 1 0 2 23 John From owner-freebsd-smp Mon Jun 23 20:28:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA18419 for smp-outgoing; Mon, 23 Jun 1997 20:28:33 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA18395; Mon, 23 Jun 1997 20:28:16 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id VAA08888; Mon, 23 Jun 1997 21:28:03 -0600 (MDT) Message-Id: <199706240328.VAA08888@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: dyson@FreeBSD.ORG cc: lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG Subject: Re: SMP kernel hung...more info In-reply-to: Your message of "Mon, 23 Jun 1997 22:18:26 CDT." <199706240318.WAA01939@dyson.iquest.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 21:28:02 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > Here is a portion of my mptable output. It appears that my system also has > an "interesting" mptable. It is a Supermicro P6DNF. Does this mean that > we might have to write some interrupt routing code??????? > > > I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# > INT conforms conforms 1 15 2 20 > INT active-lo level 0 18:A 2 17 > INT active-lo level 0 20:A 2 18 > INT active-lo level 0 19:A 2 18 > INT active-lo level 0 20:A 2 19 > INT active-lo level 0 19:A 2 19 > SMI conforms conforms 1 0 2 23 but doesn't your system work OK? your last mptable was: INT conforms conforms 1 15 2 20 INT active-lo level 0 18:A 2 17 INT active-lo level 0 20:A 2 19 SMI conforms conforms 1 0 2 23 what changed since you sent this version (sent 5-2-97)? -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jun 23 20:35:47 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA18766 for smp-outgoing; Mon, 23 Jun 1997 20:35:47 -0700 (PDT) Received: from spinner.dialix.com.au (spinner.dialix.com.au [192.203.228.67]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA18755; Mon, 23 Jun 1997 20:35:29 -0700 (PDT) Received: from spinner.dialix.com.au (localhost.dialix.com.au [127.0.0.1]) by spinner.dialix.com.au with ESMTP id LAA05567; Tue, 24 Jun 1997 11:35:06 +0800 (WST) Message-Id: <199706240335.LAA05567@spinner.dialix.com.au> X-Mailer: exmh version 2.0gamma 1/27/96 To: dyson@FreeBSD.ORG cc: smp@csn.net (Steve Passe), lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG Subject: Re: SMP kernel hung...more info In-reply-to: Your message of "Mon, 23 Jun 1997 22:18:26 EST." <199706240318.WAA01939@dyson.iquest.net> Date: Tue, 24 Jun 1997 11:35:05 +0800 From: Peter Wemm Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk "John S. Dyson" wrote: > > Hi, > > > > Lars has been struggling with his SMP system and just sent me his mptable > > output. It has something bogus in it, am wondering if anyone else with a > > supermicro has a similar table. specifically: > > > > I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# > > INT active-lo level 0 19:A 2 16 > > INT active-lo level 0 18:A 2 16 > > INT active-lo level 0 17:A 2 16 > > INT active-lo level 0 19:A 2 17 > > INT active-lo level 0 18:A 2 17 > > INT active-lo level 0 17:A 2 17 > > INT active-lo level 0 19:A 2 18 > > INT active-lo level 0 18:A 2 18 > > INT active-lo level 0 17:A 2 18 > > INT active-lo level 0 20:A 2 19 > > > > note that 19:A is declared to be attached to apic#2, pins 16, 17 and 18. > > similarily, 18:A is attached to pins 16,17,18. > > similarily, 17:A is attached to pins 16,17,18. > > > > Here is a portion of my mptable output. It appears that my system also has > an "interesting" mptable. It is a Supermicro P6DNF. Does this mean that > we might have to write some interrupt routing code??????? > > > I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# > INT conforms conforms 1 15 2 20 > INT active-lo level 0 18:A 2 17 > INT active-lo level 0 20:A 2 18 > INT active-lo level 0 19:A 2 18 > INT active-lo level 0 20:A 2 19 > INT active-lo level 0 19:A 2 19 > SMI conforms conforms 1 0 2 23 > > > John Umm, I don't see why this should suddenly be a problem _now_.. Nothing significant has changed in that part of the code over the last few days, except for brute-force substitutions of code constructs.. There should be no real difference in the handling of the routing etc, unless I've made a typo.. :-( Hmm.. The only other thing that has changed that I can think of that might affect this is the ordering of mptable parsing phases is different.. Some routines are called earlier, some later. Hmm, and then there's the PCI shared interrupt code too, but that is a few weeks old now. Cheers, -Peter From owner-freebsd-smp Mon Jun 23 20:37:46 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA18891 for smp-outgoing; Mon, 23 Jun 1997 20:37:46 -0700 (PDT) Received: from spinner.dialix.com.au (spinner.dialix.com.au [192.203.228.67]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA18883 for ; Mon, 23 Jun 1997 20:37:33 -0700 (PDT) Received: from spinner.dialix.com.au (localhost.dialix.com.au [127.0.0.1]) by spinner.dialix.com.au with ESMTP id LAA05596; Tue, 24 Jun 1997 11:36:47 +0800 (WST) Message-Id: <199706240336.LAA05596@spinner.dialix.com.au> X-Mailer: exmh version 2.0gamma 1/27/96 To: Bob Willcox cc: SMP list Subject: Re: Recent (last two days) smp kernel is hanging for me In-reply-to: Your message of "Mon, 23 Jun 1997 21:37:45 EST." <19970623213745.30803@luke.pmr.com> Date: Tue, 24 Jun 1997 11:36:46 +0800 From: Peter Wemm Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Bob Willcox wrote: > For the couple of days kernels built on my ASUS P/I-P65UP5 w/C-P6ND > have been hanging (a kernel built on 6/19 boots fine, nothing > changed in the config file). With recently cvsup'd -current sources > my kernels will no longer boot all the way. They hang right after > printing out the following messages: > > ncr0 scanning for targets 0..6 (V2 pl24 96/12/14) > ncr0: waiting for scsi devices to settle > scbus0 at ncr0 bus 0 > --- HANGS HERE --- > > Again, my old 6/19 kernel runs right on past this point, printing > out the scsi targets found. > > Anybody else seeing anything like this? Is there any additional > info that I can provide to assist in diagnosing this? > > Thanks, Hmm.. This sounds suspiciously like an interrupt problem. I presume you have APIC_IO enabled? What does your mptable output look like? What was the pci probe message for the ncr0 controller (ie: what interrupt?) Cheers, -Peter From owner-freebsd-smp Mon Jun 23 20:47:07 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA19387 for smp-outgoing; Mon, 23 Jun 1997 20:47:07 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA19381; Mon, 23 Jun 1997 20:46:57 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.5/8.8.5) id WAA03219; Mon, 23 Jun 1997 22:46:24 -0500 (EST) From: "John S. Dyson" Message-Id: <199706240346.WAA03219@dyson.iquest.net> Subject: Re: SMP kernel hung...more info In-Reply-To: <199706240335.LAA05567@spinner.dialix.com.au> from Peter Wemm at "Jun 24, 97 11:35:05 am" To: peter@spinner.dialix.com.au (Peter Wemm) Date: Mon, 23 Jun 1997 22:46:24 -0500 (EST) Cc: dyson@FreeBSD.ORG, smp@csn.net, lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG Reply-To: dyson@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > > > Here is a portion of my mptable output. It appears that my system also has > > an "interesting" mptable. It is a Supermicro P6DNF. Does this mean that > > we might have to write some interrupt routing code??????? > > > > > > I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# > > INT conforms conforms 1 15 2 20 > > INT active-lo level 0 18:A 2 17 > > INT active-lo level 0 20:A 2 18 > > INT active-lo level 0 19:A 2 18 > > INT active-lo level 0 20:A 2 19 > > INT active-lo level 0 19:A 2 19 > > SMI conforms conforms 1 0 2 23 > > > > > > John > > Umm, I don't see why this should suddenly be a problem _now_.. Nothing > significant has changed in that part of the code over the last few days, > except for brute-force substitutions of code constructs.. There should be > no real difference in the handling of the routing etc, unless I've made a > typo.. :-( > > Hmm.. The only other thing that has changed that I can think of that might > affect this is the ordering of mptable parsing phases is different.. Some > routines are called earlier, some later. Hmm, and then there's the PCI > shared interrupt code too, but that is a few weeks old now. > The code as of 21-JUN works fine, but after the latest bunch of SMP commits, the system hangs upon startup right during the NCR SCSI probes, very very early on. I am trying to figure things out, if you don't first -- I want to get this code working so that I can make some upper level VM system changes. :-). John From owner-freebsd-smp Mon Jun 23 20:48:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA19477 for smp-outgoing; Mon, 23 Jun 1997 20:48:03 -0700 (PDT) Received: from luke.pmr.com (luke.pmr.com [207.170.114.132]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA19408 for ; Mon, 23 Jun 1997 20:47:57 -0700 (PDT) Received: (from bob@localhost) by luke.pmr.com (8.8.5/8.7.3) id WAA01735; Mon, 23 Jun 1997 22:46:13 -0500 (CDT) Message-ID: <19970623224613.04995@luke.pmr.com> Date: Mon, 23 Jun 1997 22:46:13 -0500 From: Bob Willcox To: Peter Wemm Cc: SMP list Subject: Re: Recent (last two days) smp kernel is hanging for me Reply-To: Bob Willcox References: <19970623213745.30803@luke.pmr.com> <199706240336.LAA05596@spinner.dialix.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.76e In-Reply-To: <199706240336.LAA05596@spinner.dialix.com.au>; from Peter Wemm on Tue, Jun 24, 1997 at 11:36:46AM +0800 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Tue, Jun 24, 1997 at 11:36:46AM +0800, Peter Wemm wrote: > Bob Willcox wrote: > > For the couple of days kernels built on my ASUS P/I-P65UP5 w/C-P6ND > > have been hanging (a kernel built on 6/19 boots fine, nothing > > changed in the config file). With recently cvsup'd -current sources > > my kernels will no longer boot all the way. They hang right after > > printing out the following messages: > > > > ncr0 scanning for targets 0..6 (V2 pl24 96/12/14) > > ncr0: waiting for scsi devices to settle > > scbus0 at ncr0 bus 0 > > --- HANGS HERE --- > > > > Again, my old 6/19 kernel runs right on past this point, printing > > out the scsi targets found. > > > > Anybody else seeing anything like this? Is there any additional > > info that I can provide to assist in diagnosing this? > > > > Thanks, > > Hmm.. This sounds suspiciously like an interrupt problem. I presume you > have APIC_IO enabled? What does your mptable output look like? What was > the pci probe message for the ncr0 controller (ie: what interrupt?) Below is the mptable output (while running on the old working kermel). The NCR SCSI controller is listed as being on IRQ 17 (for both the old working and new not working kernels). Here is the dmesg line for it: ncr0: rev 0x02 int a irq 17 on pci0.11.0 Thanks, Bob =============================================================================== MPTable, version 2.0.11 ------------------------------------------------------------------------------- MP Floating Pointer Structure: location: BIOS physical address: 0x000f61b0 signature: '_MP_' length: 16 bytes version: 1.4 checksum: 0x8a mode: Virtual Wire ------------------------------------------------------------------------------- MP Config Table Header: physical address: 0x000f5daa signature: 'PCMP' base table length: 244 version: 1.4 checksum: 0x9f OEM ID: 'OEM00000' Product ID: 'PROD00000000' OEM table pointer: 0x00000000 OEM table size: 0 entry count: 22 local APIC address: 0xfee00000 extended table length: 0 extended table checksum: 0 ------------------------------------------------------------------------------- MP Config Base Table Entries: -- Processors: APIC ID Version State Family Model Step Flags 1 0x11 BSP, usable 6 1 2 0xfbff 0 0x11 AP, usable 6 1 2 0xfbff -- Bus: Bus ID Type 0 PCI 1 ISA -- I/O APICs: APIC ID Version State Address 2 0x11 usable 0xfec00000 -- I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# ExtINT conforms conforms 1 0 2 0 INT conforms conforms 1 1 2 1 INT conforms conforms 1 0 2 2 INT conforms conforms 1 3 2 3 INT conforms conforms 1 4 2 4 INT conforms conforms 1 5 2 5 INT conforms conforms 1 6 2 6 INT conforms conforms 1 7 2 7 INT conforms conforms 1 8 2 8 INT conforms conforms 1 12 2 12 INT conforms conforms 1 14 2 14 INT active-lo level 0 10:A 2 18 INT active-lo level 0 11:A 2 17 INT active-lo level 0 12:A 2 16 INT active-lo level 0 13:A 2 19 -- Local Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# ExtINT active-hi edge 1 0 255 0 NMI active-hi edge 1 0 255 1 ------------------------------------------------------------------------------- # SMP kernel config file options: # Required: options SMP # Symmetric MultiProcessor Kernel options APIC_IO # Symmetric (APIC) I/O # Useful: #options SMP_AUTOSTART # start the additional CPUs during boot # Optional (built-in defaults will work in most cases): #options NCPU=2 # number of CPUs #options NBUS=2 # number of busses #options NAPIC=1 # number of IO APICs #options NINTR=24 # number of INTs # Currently broken: #options SMP_PRIVPAGES # BROKEN, DO NOT use! # Rogue hardware: # # Tyan Tomcat II: #options SMP_TIMER_NC # # # SuperMicro P6DNE: #options SMP_TIMER_NC # =============================================================================== > > Cheers, > -Peter > -- Bob Willcox Deliberation, n.: The act of examining one's bread to determine bob@luke.pmr.com which side it is buttered on. Austin, TX -- Ambrose Bierce, "The Devil's Dictionary" From owner-freebsd-smp Mon Jun 23 20:48:19 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA19504 for smp-outgoing; Mon, 23 Jun 1997 20:48:19 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA19495; Mon, 23 Jun 1997 20:48:12 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.5/8.8.5) id WAA03241; Mon, 23 Jun 1997 22:48:08 -0500 (EST) From: "John S. Dyson" Message-Id: <199706240348.WAA03241@dyson.iquest.net> Subject: Re: SMP kernel hung...more info In-Reply-To: <199706240328.VAA08888@Ilsa.StevesCafe.com> from Steve Passe at "Jun 23, 97 09:28:02 pm" To: smp@csn.net (Steve Passe) Date: Mon, 23 Jun 1997 22:48:08 -0500 (EST) Cc: dyson@FreeBSD.ORG, lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG Reply-To: dyson@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > your last mptable was: > > INT conforms conforms 1 15 2 20 > INT active-lo level 0 18:A 2 17 > INT active-lo level 0 20:A 2 19 > SMI conforms conforms 1 0 2 23 > > what changed since you sent this version (sent 5-2-97)? > I added a GUSPNP, upgraded my Matrox Millenium to 8MB, and that is about it. I'll reset my bios from scratch, and see what I can do to nudge it into being more "standard." John From owner-freebsd-smp Mon Jun 23 20:54:10 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA19877 for smp-outgoing; Mon, 23 Jun 1997 20:54:10 -0700 (PDT) Received: from spinner.dialix.com.au (spinner.dialix.com.au [192.203.228.67]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA19855; Mon, 23 Jun 1997 20:53:45 -0700 (PDT) Received: from spinner.dialix.com.au (localhost.dialix.com.au [127.0.0.1]) by spinner.dialix.com.au with ESMTP id LAA05745; Tue, 24 Jun 1997 11:52:23 +0800 (WST) Message-Id: <199706240352.LAA05745@spinner.dialix.com.au> X-Mailer: exmh version 2.0gamma 1/27/96 To: Steve Passe cc: dyson@FreeBSD.ORG, lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG Subject: Re: SMP kernel hung...more info In-reply-to: Your message of "Mon, 23 Jun 1997 21:28:02 CST." <199706240328.VAA08888@Ilsa.StevesCafe.com> Date: Tue, 24 Jun 1997 11:52:22 +0800 From: Peter Wemm Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Steve Passe wrote: > Hi, > > > Here is a portion of my mptable output. It appears that my system also has > > an "interesting" mptable. It is a Supermicro P6DNF. Does this mean that > > we might have to write some interrupt routing code??????? > > > > > > I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# > > INT conforms conforms 1 15 2 20 > > INT active-lo level 0 18:A 2 17 > > INT active-lo level 0 20:A 2 18 > > INT active-lo level 0 19:A 2 18 > > INT active-lo level 0 20:A 2 19 > > INT active-lo level 0 19:A 2 19 > > SMI conforms conforms 1 0 2 23 > > but doesn't your system work OK? > > your last mptable was: > > INT conforms conforms 1 15 2 20 > INT active-lo level 0 18:A 2 17 > INT active-lo level 0 20:A 2 19 > SMI conforms conforms 1 0 2 23 > > what changed since you sent this version (sent 5-2-97)? Hmm.. PCI bus remapped to bus 0 sort routine perhaps? Nah, that only runs on the collected data, doesn't it? Or perhaps something has gone wrong with mptable.c and it's leading us on a wild goose chase? Cheers, -Peter From owner-freebsd-smp Mon Jun 23 21:28:37 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA21294 for smp-outgoing; Mon, 23 Jun 1997 21:28:37 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA21278; Mon, 23 Jun 1997 21:28:20 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id WAA09251; Mon, 23 Jun 1997 22:27:27 -0600 (MDT) Message-Id: <199706240427.WAA09251@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Peter Wemm cc: dyson@FreeBSD.ORG, bob@luke.pmr.com, lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG Subject: Re: SMP kernel hung...more info In-reply-to: Your message of "Tue, 24 Jun 1997 11:52:22 +0800." <199706240352.LAA05745@spinner.dialix.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 23 Jun 1997 22:27:26 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > what changed since you sent this version (sent 5-2-97)? > > Hmm.. PCI bus remapped to bus 0 sort routine perhaps? Nah, that only > runs on the collected data, doesn't it? > > Or perhaps something has gone wrong with mptable.c and it's leading us on > a wild goose chase? these listings are the output of mptable, so PCI remap no, mptable.c possibly yes. but I can't imagine what. It hasn't had any logic changes in ages. but reguardless 2 or 3 people are reporting hangs within the last week. Summary: --- From: Lars Fredriksen Subject: Booting SMP on SuperMicro P6DNE fails To: freebsd-smp@FreeBSD.ORG Date: Sat, 21 Jun 1997 00:01:40 -0500 (CDT) Hi, Building a current kernel (as of last night) with the following options: SMP APIC_IO SMP_AUTOSTART SMP_TIMER_NC and without I386_CPU and I486_CPU failed to boot. I booted to the point ... I just booted the SMP kernel and broke into the debugger when it got hung (after saying that all 2 cpus are online). Well it isn't actually the kernel per se that is hung. It is in the smp_idleloop. The problem is that one of the rc scripts is running swapon and it is hung - on wait channel 0xf0b60a00 (i'll try to find out what this is) --- Date: Mon, 23 Jun 1997 21:37:45 -0500 From: Bob Willcox To: SMP list Subject: Recent (last two days) smp kernel is hanging for me For the couple of days kernels built on my ASUS P/I-P65UP5 w/C-P6ND have been hanging (a kernel built on 6/19 boots fine, nothing changed in the config file). With recently cvsup'd -current sources my kernels will no longer boot all the way. They hang right after printing out the following messages: ncr0 scanning for targets 0..6 (V2 pl24 96/12/14) ncr0: waiting for scsi devices to settle scbus0 at ncr0 bus 0 --- HANGS HERE --- --- peter committed the new code: Date: Sun, 22 Jun 1997 09:04:24 -0700 (PDT) so it would appear to be unrelated to this latest SMP change (thanxs for little favors!) --- Lars, you mention it hangs in swapon called by rc, do you have an explicit "swaps on sdx" line in your config file? this is known to be a problem. you mention having a 2944, what is that, a 2940 with floppy? --- I'm thinking its something related to recent disk I/O changes perhaps? I cvsupped everything early today and successfully rebuilt the world with this source. No problems noticed, so I don't have any clues to offer. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Jun 24 00:32:25 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA00324 for smp-outgoing; Tue, 24 Jun 1997 00:32:25 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA00303; Tue, 24 Jun 1997 00:32:13 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id BAA10251; Tue, 24 Jun 1997 01:32:10 -0600 (MDT) Message-Id: <199706240732.BAA10251@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: dyson@FreeBSD.ORG cc: bob@luke.pmr.com, peter@spinner.dialix.com.au, freebsd-smp@FreeBSD.ORG Subject: Re: Recent (last two days) smp kernel is hanging for me In-reply-to: Your message of "Mon, 23 Jun 1997 22:10:35 CDT." <199706240310.WAA01893@dyson.iquest.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 24 Jun 1997 01:32:10 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, I found something that would explain the new code working on the intels, but failing on both the ASUS and the P6DNF. Specifically we now create the mplock with the LOGICAL cpu id, BUT the initial lock is still built with the PHYSICAL cpu id. since the intel #s the BSP as 0 it works, but both the ASUS and the P6DNF # the BSP as 1, so it gets hung on the first attempt to grab the lock (this is all theory). try changing mp_machdep.c:start_all_aps(), line 1376: mp_lock = (lapic.id & APIC_ID_MASK) + 1; to: mp_lock = 1; -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Jun 24 02:10:27 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA04920 for smp-outgoing; Tue, 24 Jun 1997 02:10:27 -0700 (PDT) Received: from gw.itfs.nsk.su (ns.nsk.ru [193.124.36.33]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id CAA04912 for ; Tue, 24 Jun 1997 02:10:22 -0700 (PDT) Received: from itfs.UUCP (uucp@localhost) by gw.itfs.nsk.su (8.6.12/8.6.12) with UUCP id QAA26400 for smp@freebsd.org; Tue, 24 Jun 1997 16:10:19 +0700 Received: by itfs.nsk.su; Tue, 24 Jun 97 16:20:49 +0700 (NST) Received: (from daemon@localhost) by news.itfs.nsk.su (8.7.5/8.6.12) id PAA23468; Tue, 24 Jun 1997 15:52:38 +0700 (NSD) From: nnd@itfs.nsk.su To: smp@freebsd.org Subject: Re: SMP_PRIVPAGES Date: 24 Jun 1997 08:52:37 GMT Message-ID: <5oo1sl$mdv@news.itfs.nsk.su> References: <199706232018.OAA07315@Ilsa.StevesCafe.com> Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Steve Passe wrote: > Peter has committed the code for enabling SMP_PRIVPAGES. Not that it > is automatic, ie you don't need to add the option to your config file. > I successfully used it to "make world" with -current as of earlier this > morning. > Thanx, Peter! I also make world and make kernel with this code and after that discover that famous 'dd if=/dev/zero of=/dev/null...' bencmark shows even smaller results than previous SMP-kernel. After that I've decide to try the next patch on /sys/i386/isa/npx.c: --- npx.c Tue Jun 24 14:42:03 1997 +++ npx.c.orig Tue Jun 24 09:02:56 1997 @@ -413,7 +413,7 @@ } npxinit(__INITIAL_NPXCW__); -#if defined(I586_CPU) +#if defined(I586_CPU) && !defined(SMP) /* FPU not working under SMP yet */ if (cpu_class == CPUCLASS_586 && npx_ex16) { if (!(dvp->id_flags & NPX_DISABLE_I586_OPTIMIZED_BCOPY)) { and ... this is a result: (before my patch) dd if=/dev/zero of=/dev/null bs=1m count=1000 1000+0 records in 1000+0 records out 1048576000 bytes transferred in 21.219967 secs (49414591 bytes/sec) (and after it) dd if=/dev/zero of=/dev/null bs=1m count=1000 1000+0 records in 1000+0 records out 1048576000 bytes transferred in 11.985571 secs (87486529 bytes/sec) The last number is still lose to NON-SMP case, but it seems to me that there is another place to gain speed for bzero/bcopy - in file /sys/i386/i386/support.s there is a "label" kernel_fpu_lock: As I can (NOT?) understand it can be duplicated for each CPU and may be this can give us some more speedup of bzero/bcopy ? N.Dudorov From owner-freebsd-smp Tue Jun 24 03:55:36 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id DAA08458 for smp-outgoing; Tue, 24 Jun 1997 03:55:36 -0700 (PDT) Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id DAA08445 for ; Tue, 24 Jun 1997 03:55:15 -0700 (PDT) Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.5/8.6.9) id UAA16458; Tue, 24 Jun 1997 20:37:15 +1000 Date: Tue, 24 Jun 1997 20:37:15 +1000 From: Bruce Evans Message-Id: <199706241037.UAA16458@godzilla.zeta.org.au> To: nnd@itfs.nsk.su, smp@freebsd.org Subject: Re: SMP_PRIVPAGES Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > After that I've decide to try the next patch on >/sys/i386/isa/npx.c: > >--- npx.c Tue Jun 24 14:42:03 1997 >+++ npx.c.orig Tue Jun 24 09:02:56 1997 >@@ -413,7 +413,7 @@ > } > npxinit(__INITIAL_NPXCW__); > >-#if defined(I586_CPU) >+#if defined(I586_CPU) && !defined(SMP) > /* FPU not working under SMP yet */ > if (cpu_class == CPUCLASS_586 && npx_ex16) { > if (!(dvp->id_flags & NPX_DISABLE_I586_OPTIMIZED_BCOPY)) { I thought that copying through the FPU (not the FPU itself) still doesn't work yet, but it seems that the private page changes have automagically fixed it - there is now at least a chance that direct accesses to _npxproc and _curpcb work right for the same reasons that direct accesses to C variables npxproc and curpcb work right. >and ... this is a result: > >dd if=/dev/zero of=/dev/null bs=1m count=1000 >1000+0 records in >1000+0 records out >1048576000 bytes transferred in 11.985571 secs (87486529 bytes/sec) > > The last number is still lose to NON-SMP case, but it seems Did you get > 110MB/sec for non-SMP? >to me that there is another place to gain speed for bzero/bcopy - > >in file /sys/i386/i386/support.s there is a "label" kernel_fpu_lock: > >As I can (NOT?) understand it can be duplicated for each CPU >and may be this can give us some more speedup of bzero/bcopy ? kernel_fpu_lock may already work under SMP too, except for the obvious problem that the bus is not locked when it is accessed. It should be an array instead of a per-process variable, since it prevents reentry by interrupt handlers and interrupt handling is unrelated to processes. I don't think there large speedups to be gained here, since multiple large concurrent bcopy/bzero's are probably rare, and anyway, a single fast bzero/bcopy or a couple of slow ones will saturate the memory bus. Bruce From owner-freebsd-smp Tue Jun 24 05:19:38 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id FAA11479 for smp-outgoing; Tue, 24 Jun 1997 05:19:38 -0700 (PDT) Received: from luke.pmr.com (luke.pmr.com [207.170.114.132]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id FAA11465; Tue, 24 Jun 1997 05:19:25 -0700 (PDT) Received: (from bob@localhost) by luke.pmr.com (8.8.5/8.7.3) id HAA06089; Tue, 24 Jun 1997 07:19:04 -0500 (CDT) Message-ID: <19970624071903.56017@luke.pmr.com> Date: Tue, 24 Jun 1997 07:19:03 -0500 From: Bob Willcox To: Steve Passe Cc: dyson@FreeBSD.ORG, peter@spinner.dialix.com.au, freebsd-smp@FreeBSD.ORG Subject: Re: Recent (last two days) smp kernel is hanging for me Reply-To: Bob Willcox References: <199706240310.WAA01893@dyson.iquest.net> <199706240732.BAA10251@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.76e In-Reply-To: <199706240732.BAA10251@Ilsa.StevesCafe.com>; from Steve Passe on Tue, Jun 24, 1997 at 01:32:10AM -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Tue, Jun 24, 1997 at 01:32:10AM -0600, Steve Passe wrote: > Hi, > > I found something that would explain the new code working on the intels, but > failing on both the ASUS and the P6DNF. Specifically we now create the mplock > with the LOGICAL cpu id, BUT the initial lock is still built with the > PHYSICAL cpu id. since the intel #s the BSP as 0 it works, but both the ASUS > and the P6DNF # the BSP as 1, so it gets hung on the first attempt to grab the > lock (this is all theory). try changing mp_machdep.c:start_all_aps(), line > 1376: > > mp_lock = (lapic.id & APIC_ID_MASK) + 1; > > to: > > mp_lock = 1; Well, that seemed to do the trick on my system!! :-) > > -- > Steve Passe | powered by > smp@csn.net | Symmetric MultiProcessor FreeBSD > -- Bob Willcox Deliberation, n.: The act of examining one's bread to determine bob@luke.pmr.com which side it is buttered on. Austin, TX -- Ambrose Bierce, "The Devil's Dictionary" From owner-freebsd-smp Tue Jun 24 05:55:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id FAA12812 for smp-outgoing; Tue, 24 Jun 1997 05:55:23 -0700 (PDT) Received: from agora.rdrop.com (root@agora.rdrop.com [199.2.210.241]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id FAA12804; Tue, 24 Jun 1997 05:55:17 -0700 (PDT) Received: from spinner.dialix.com.au (spinner.dialix.com.au [192.203.228.67]) by agora.rdrop.com (8.8.5/8.8.5) with ESMTP id FAA09907; Tue, 24 Jun 1997 05:55:11 -0700 (PDT) Received: from spinner.dialix.com.au (localhost.dialix.com.au [127.0.0.1]) by spinner.dialix.com.au with ESMTP id UAA10065; Tue, 24 Jun 1997 20:53:01 +0800 (WST) Message-Id: <199706241253.UAA10065@spinner.dialix.com.au> X-Mailer: exmh version 2.0gamma 1/27/96 To: Bob Willcox cc: Steve Passe , dyson@FreeBSD.ORG, freebsd-smp@FreeBSD.ORG Subject: Re: Recent (last two days) smp kernel is hanging for me In-reply-to: Your message of "Tue, 24 Jun 1997 07:19:03 EST." <19970624071903.56017@luke.pmr.com> Date: Tue, 24 Jun 1997 20:53:00 +0800 From: Peter Wemm Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Bob Willcox wrote: > On Tue, Jun 24, 1997 at 01:32:10AM -0600, Steve Passe wrote: > > Hi, > > > > I found something that would explain the new code working on the intels, bu t > > failing on both the ASUS and the P6DNF. Specifically we now create the mpl ock > > with the LOGICAL cpu id, BUT the initial lock is still built with the > > PHYSICAL cpu id. since the intel #s the BSP as 0 it works, but both the AS US > > and the P6DNF # the BSP as 1, so it gets hung on the first attempt to grab the > > lock (this is all theory). try changing mp_machdep.c:start_all_aps(), line > > 1376: > > > > mp_lock = (lapic.id & APIC_ID_MASK) + 1; > > > > to: > > > > mp_lock = 1; > > > Well, that seemed to do the trick on my system!! :-) Well, pass me the pointy hat! *-<8-) Cheers, -Peter From owner-freebsd-smp Tue Jun 24 11:09:38 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA28251 for smp-outgoing; Tue, 24 Jun 1997 11:09:38 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA28232; Tue, 24 Jun 1997 11:09:15 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id MAA11971; Tue, 24 Jun 1997 12:07:01 -0600 (MDT) Message-Id: <199706241807.MAA11971@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Peter Wemm cc: Bob Willcox , lars@fredriks-1.pr.mcs.net, dyson@FreeBSD.ORG, freebsd-smp@FreeBSD.ORG Subject: Re: Recent (last two days) smp kernel is hanging for me In-reply-to: Your message of "Tue, 24 Jun 1997 20:53:00 +0800." <199706241253.UAA10065@spinner.dialix.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 24 Jun 1997 12:07:01 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > Bob Willcox wrote: > > On Tue, Jun 24, 1997 at 01:32:10AM -0600, Steve Passe wrote: > > > > > > I found something that would explain the new code working on the intels, > > > but failing on both the ASUS and the P6DNF. > > > ... > > Well, that seemed to do the trick on my system!! :-) > > Well, pass me the pointy hat! *-<8-) > > Cheers, > -Peter sorry, Peter, but if thats the only thing you missed in that massive set of changes you don't get to wear the hat! John Dyson also reports that SMP is working again for him. Lars reports that this fix gets him further along, but is still hanging in swapon. --- So to summarize the failure on Lar's system: I just booted the SMP kernel and broke into the debugger when it got hung (after saying that all 2 cpus are online). Well it isn't actually the kernel per se that is hung. It is in the smp_idleloop. The problem is that one of the rc scripts is running swapon and it is hung - on wait channel 0xf0b60a00 (i'll try to find out what this is) I have 2 scsi controllers in my system; a 2940 and a 2944. - config: # ahc1 is the controller for sd0, swap devices on ahc0 controller ahc0 controller ahc1 controller scbus0 at ahc0 controller scbus1 at ahc1 disk sd0 at scbus1 target 0 unit 0 #disk sd1 at scbus0 target 1 tape st0 at scbus1 target 2 tape st1 at scbus1 target 5 device sd0 #Only need one of these, the code dynamically grows device st0 device cd0 - dmesg: ahc0: rev 0x03 int a irq 11 on pci0.19.0 using shared irq11. ahc0: Reading SEEPROM...done. low byte termination enabled, high byte termination enabled ahc0: aic7870 Wide Channel, SCSI Id=7, 16/255 SCBs ahc0: Resetting Channel A ahc0: Downloading Sequencer Program...ahc0: 411 instructions downloaded Done ahc0: Probing channel A Choosing drivers for scbus configured at 0 ahc0: waiting for scsi devices to settle scbus0 at ahc0 bus 0 ahc0: target 0 synchronous at 8.0MHz, offset = 0xf ahc0: target 0 Tagged Queuing Device scbus0 target 0 lun 0: type 0 fixed SCSI 2 sd1 at scbus0 target 0 lun 0 sd1: Direct-Access 2030MB (4159462 512 byte sectors) sd1: with 2626 cyls, 19 heads, and an average 83 sectors/track ahc0: target 1 synchronous at 4.4MHz, offset = 0xf scbus0 target 1 lun 0: type 0 fixed SCSI 1 sd2 at scbus0 target 1 lun 0 sd2: Direct-Access 955MB (1956864 512 byte sectors) sd2: with 1931 cyls, 15 heads, and an average 67 sectors/track ahc0: target 2 synchronous at 10.0MHz, offset = 0x8 ahc0: target 2 Tagged Queuing Device scbus0 target 2 lun 0: type 0 fixed SCSI 2 sd3 at scbus0 target 2 lun 0 sd3: Direct-Access 1003MB (2054864 512 byte sectors) sd3: with 2051 cyls, 13 heads, and an average 77 sectors/track ahc0: target 3 using asynchronous transfers ahc0:A:3: Warning - unknown message received from target (0x1). SEQ_FLAGS == 0x6. Rejecting scbus0 target 3 lun 0: type 1 removable SCSI 2 st2 at scbus0 target 3 lun 0 st2: Sequential-Access density code 0x0, drive empty ahc0: target 4 synchronous at 4.4MHz, offset = 0xf scbus0 target 4 lun 0: type 0 fixed SCSI 1 sd4 at scbus0 target 4 lun 0 sd4: Direct-Access 955MB (1956864 512 byte sectors) sd4: with 1931 cyls, 15 heads, and an average 67 sectors/track ahc0: target 5 synchronous at 4.4MHz, offset = 0xf scbus0 target 5 lun 0: type 0 fixed SCSI 1 sd5 at scbus0 target 5 lun 0 sd5: Direct-Access 955MB (1956864 512 byte sectors) sd5: with 1931 cyls, 15 heads, and an average 67 sectors/track ahc0: target 6 synchronous at 4.4MHz, offset = 0xf scbus0 target 6 lun 0: type 0 fixed SCSI 1 sd6 at scbus0 target 6 lun 0 sd6: Direct-Access 955MB (1956864 512 byte sectors) sd6: with 1931 cyls, 15 heads, and an average 67 sectors/track probe0(ahc0:9:0): scsi_cmd probe0(ahc0:9:0): scsi_done scbus0 target 9 lun 0: command: 0,0,0,0,0,0-[0 bytes] probe0(ahc0:9:0): scsi_cmd probe0(ahc0:9:0): scsi_done scbus0 target 9 lun 0: command: 12,0,0,0,2c,0-[44 bytes] ------------------------------ 000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 016: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 032: 00 00 00 00 00 00 00 00 00 00 00 00 ------------------------------ found-> vendor=0x9004, dev=0x8178, revid=0x00 class=01-00-00, hdrtype=0x00, mfdev=0 intpin=a, irq=14 map[0]: type 4, range 32, base 0000e800, size 8 map[1]: type 1, range 32, base fe3e7000, size 12 ahc1: rev 0x00 int a irq 14 on pci0.20.0 ahc1: Reading SEEPROM...done. internal50 cable is present internal68 cable not present brdctl == 0xac external cable is present eprom is present brdctl == 0xac low byte termination disabled, high byte termination enabled ahc1: aic7880 Wide Channel, SCSI Id=7, 16/255 SCBs ahc1: Resetting Channel A ahc1: Downloading Sequencer Program...ahc1: 418 instructions downloaded Done ahc1: Probing channel A Choosing drivers for scbus configured at 1 ahc1: waiting for scsi devices to settle scbus1 at ahc1 bus 0 ahc1: target 0 synchronous at 4.4MHz, offset = 0xf scbus1 target 0 lun 0: type 0 fixed SCSI 1 sd is configured at 0 sd0 at scbus1 target 0 lun 0 sd0: Direct-Access 955MB (1956864 512 byte sectors) sd0: with 1931 cyls, 15 heads, and an average 67 sectors/track ahc1:A:3: refuses synchronous negotiation. Using asynchronous transfers scbus1 target 3 lun 0: type 5 removable SCSI 2 cd0 at scbus1 target 3 lun 0 cd0: CD-ROM cd present [78823 x 2048 byte records] scbus1 target 3 lun 1: type 5 removable SCSI 2 cd1 at scbus1 target 3 lun 1 cd1: CD-ROM can't get the size scbus1 target 3 lun 2: type 5 removable SCSI 2 cd2 at scbus1 target 3 lun 2 cd2: CD-ROM cd present [196535 x 2048 byte records] scbus1 target 3 lun 3: type 5 removable SCSI 2 cd3 at scbus1 target 3 lun 3 cd3: CD-ROM cd present [227432 x 2048 byte records] scbus1 target 3 lun 4: type 5 removable SCSI 2 cd4 at scbus1 target 3 lun 4 cd4: CD-ROM cd present [167490 x 2048 byte records] scbus1 target 3 lun 5: type 5 removable SCSI 2 cd5 at scbus1 target 3 lun 5 cd5: CD-ROM cd present [87066 x 2048 byte records] scbus1 target 3 lun 6: type 5 removable SCSI 2 cd6 at scbus1 target 3 lun 6 cd6: CD-ROM cd present [303991 x 2048 byte records] ahc1: target 5 synchronous at 5.0MHz, offset = 0x8 scbus1 target 5 lun 0: type 1 removable SCSI 2 st is configured at 1 st1 at scbus1 target 5 lun 0 st1: Sequential-Access density code 0x13, drive empty probe0(ahc1:9:0): scsi_cmd probe0(ahc1:9:0): scsi_done scbus1 target 9 lun 0: command: 0,0,0,0,0,0-[0 bytes] probe0(ahc1:9:0): scsi_cmd probe0(ahc1:9:0): scsi_done scbus1 target 9 lun 0: command: 12,0,0,0,2c,0-[44 bytes] ------------------------------ 000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 016: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 032: 00 00 00 00 00 00 00 00 00 00 00 00 ------------------------------ ------------------------------------------------------------------------------- I noticed long ago that a line in the config file like: config kernel root on sd0 swaps on sd0 ^^^^^^^^^^^^ would hang SMP. So since these might be related, I am going to try adding that here and hopefully attack the problem locally. If anyone has any clues/theories plese speak up... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Jun 24 12:17:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA01115 for smp-outgoing; Tue, 24 Jun 1997 12:17:26 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA01107; Tue, 24 Jun 1997 12:17:06 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id NAA12231; Tue, 24 Jun 1997 13:15:07 -0600 (MDT) Message-Id: <199706241915.NAA12231@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Steve Passe cc: Peter Wemm , Bob Willcox , lars@fredriks-1.pr.mcs.net, dyson@FreeBSD.ORG, freebsd-smp@FreeBSD.ORG Subject: Re: Recent (last two days) smp kernel is hanging for me In-reply-to: Your message of "Tue, 24 Jun 1997 12:07:01 MDT." <199706241807.MAA11971@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 24 Jun 1997 13:15:06 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > I noticed long ago that a line in the config file like: > > config kernel root on sd0 swaps on sd0 > ^^^^^^^^^^^^ > would hang SMP. please excuse my last brain-fart, it was "dumps on" that caused the problem... guess I'll go look at swapon source. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Jun 24 12:18:39 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA01172 for smp-outgoing; Tue, 24 Jun 1997 12:18:39 -0700 (PDT) Received: from Kitten.mcs.com (Kitten.mcs.com [192.160.127.90]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA01136; Tue, 24 Jun 1997 12:18:20 -0700 (PDT) Received: from Mercury.mcs.net (fredriks@Mercury.mcs.net [192.160.127.80]) by Kitten.mcs.com (8.8.5/8.8.2) with ESMTP id OAA18884; Tue, 24 Jun 1997 14:18:18 -0500 (CDT) Received: (from fredriks@localhost) by Mercury.mcs.net (8.8.5/8.8.2) id OAA19076; Tue, 24 Jun 1997 14:18:17 -0500 (CDT) From: Lars Fredriksen Message-Id: <199706241918.OAA19076@Mercury.mcs.net> Subject: Re: Recent (last two days) smp kernel is hanging for me To: smp@csn.net (Steve Passe) Date: Tue, 24 Jun 1997 14:18:16 -0500 (CDT) Cc: peter@spinner.dialix.com.au, bob@luke.pmr.com, lars@fredriks-1.pr.mcs.net, dyson@FreeBSD.ORG, freebsd-smp@FreeBSD.ORG In-Reply-To: <199706241807.MAA11971@Ilsa.StevesCafe.com> from "Steve Passe" at Jun 24, 97 12:07:01 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Steve Passe writes: > > --- > So to summarize the failure on Lar's system: > > I just booted the SMP kernel and broke into the debugger when > it got hung (after saying that all 2 cpus are online). Well it isn't > actually the kernel per se that is hung. It is in the smp_idleloop. > > The problem is that one of the rc scripts is running swapon and it > is hung - on wait channel 0xf0b60a00 (i'll try to find out what this > is) > > I have 2 scsi controllers in my system; a 2940 and a 2944. > This morning when I booted the SMP kernel with the lock fixes I let it sit for about 10 minutes or so after it got hung and I did get error messages from the scsi driver where it said that it timed out "Timeout SCB handled by another timeout" on the controller that has the swapdevices on it which to me indicates that the kernel does not see the interrupt back from the controller board. Now I am making one hell of an assumption here, and that is that the firmware download and initialization of the board succeeded. The dmesg did indicate that. Now I'll do one more test and that is to comment out the swapon in /etc/rc. I expect that if I do that, the machine will just fail to mount any of the file systems that is connected through the second (ahc0) controller. Now from what I can see, ahc0 is sharing the interrupt with vga and ethernet controllers, even though they seem to be steered to different pins(?) by means of pci0.19.0, pci.18.0, and so forth. With respect to the request of getting the kernel messages through the serial port, that is going to take some time. I first need to get the other box up and running :-) Lars -- ------------------------------------------------------------------- Lars Fredriksen fredriks@mcs.com (home) lars@fredriks-2.pr.mcs.net (home-home) From owner-freebsd-smp Tue Jun 24 12:42:54 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA02145 for smp-outgoing; Tue, 24 Jun 1997 12:42:54 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA02134; Tue, 24 Jun 1997 12:42:40 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id NAA12340; Tue, 24 Jun 1997 13:38:54 -0600 (MDT) Message-Id: <199706241938.NAA12340@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Lars Fredriksen cc: peter@spinner.dialix.com.au, lars@fredriks-1.pr.mcs.net, dyson@FreeBSD.ORG, freebsd-smp@FreeBSD.ORG Subject: Re: Recent (last two days) smp kernel is hanging for me In-reply-to: Your message of "Tue, 24 Jun 1997 14:18:16 CDT." <199706241918.OAA19076@Mercury.mcs.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 24 Jun 1997 13:38:54 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Lars, --- which of the above addresses is preferred for contacting you: fredriks@Mcs.Net or lars@fredriks-1.pr.mcs.net --- > This morning when I booted the SMP kernel with the lock fixes I let > it sit for about 10 minutes or so after it got hung and I did get > error messages from the scsi driver where it said that it timed out > "Timeout SCB handled by another timeout" > on the controller that has the swapdevices on it which to me indicates > that the kernel does not see the interrupt back from the controller board. > Now I am making one hell of an assumption here, and that is that the > firmware download and initialization of the board succeeded. The dmesg > did indicate that. Now I'll do one more test and that is to comment > out the swapon in /etc/rc. I expect that if I do that, the machine will > just fail to mount any of the file systems that is connected through > the second (ahc0) controller. Now from what I can see, ahc0 is sharing > the interrupt with vga and ethernet controllers, even though they seem > to be steered to different pins(?) by means of pci0.19.0, pci.18.0, and > so forth. So it seems we are back to the odd mptable contents. This would certainly explain a missing INT. The supermicro site lists the P6DNE as curently using BIOS version 1.1, which version are you using? -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Jun 24 18:40:25 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id SAA21737 for smp-outgoing; Tue, 24 Jun 1997 18:40:25 -0700 (PDT) Received: from squid.pdc.kth.se (squid.pdc.kth.se [130.237.221.65]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id SAA21712 for ; Tue, 24 Jun 1997 18:40:14 -0700 (PDT) Received: from localhost (tege@localhost) by squid.pdc.kth.se (8.8.5/8.6.9) with SMTP id DAA01747 for ; Wed, 25 Jun 1997 03:40:10 +0200 (MET DST) Message-Id: <199706250140.DAA01747@squid.pdc.kth.se> X-Authentication-Warning: squid.pdc.kth.se: tege@localhost didn't use HELO protocol To: freebsd-smp@freebsd.org Subject: New monster server Date: Wed, 25 Jun 1997 03:40:10 +0200 From: Torbjorn Granlund Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk My company intends to purchase a new compute and disk server early this fall. We originally intended to get an Alpha system, but it seems hard to get a reasonable OS without paying an outrageous amount of money. (NetBSD might be good enough for the alpha platform; but it doesn't seem to get a lot of usage.) The FreeBSD SMP work is very good news, since it may enable us to stay with FreeBSD while getting the machine power we need. (We currently run FreeBSD 2.2.1 or 2.2.2 on seven of our eight machines. The only non-FreeBSD machine is a Sun...) We would like the machine to have on the order of 25 SPECint95 (summing the SPECint95 on the individual CPUs in case we go for an SMP solution). So a single CPU 21164 won't do. A single cpu 21264 would do. A 4 CPU P6/200MHz/512KB would do. (It seems difficult to get 4 CPU P6 motherboards, since the Orion isn't made, and the 440BX is not yet available.) I'd be interessted in hearing your comments. Realistically, will I be able to get a 4 CPU FreeBSD system to work well enough this fall? If not, when do you think that could happen? I am willing to spend about $20000 for the hardware. It is highly unlikely that Digital will want to sell me a dual CPU 21164 or a single CPU 21264 for that amount money. Will a 4 CPU P6 be cheap enough and will it be stable enough to use for heavy load? Torbjorn From owner-freebsd-smp Tue Jun 24 21:33:07 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA29513 for smp-outgoing; Tue, 24 Jun 1997 21:33:07 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA29508 for ; Tue, 24 Jun 1997 21:33:03 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id WAA13791; Tue, 24 Jun 1997 22:32:52 -0600 (MDT) Message-Id: <199706250432.WAA13791@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Torbjorn Granlund cc: freebsd-smp@FreeBSD.ORG Subject: Re: New monster server In-reply-to: Your message of "Wed, 25 Jun 1997 03:40:10 +0200." <199706250140.DAA01747@squid.pdc.kth.se> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 24 Jun 1997 22:32:52 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > I'd be interessted in hearing your comments. Realistically, will I be able > to get a 4 CPU FreeBSD system to work well enough this fall? If not, when > do you think that could happen? > > I am willing to spend about $20000 for the hardware. It is highly unlikely > that Digital will want to sell me a dual CPU 21164 or a single CPU 21264 for > that amount money. Will a 4 CPU P6 be cheap enough and will it be stable > enough to use for heavy load? we expect to have made a fair amount of progress by late fall, but there are no guarantees... It is reasonably stable now. check out the AMI goliath, a 4 P6 MB. I have seen the bare boards + CPU boards for around $4500: http://www.megatrends.com/Motherboards/ This specific machine has not been tested under FreeBSD SMP yet that I know of, but AMI usually makes quality hardware. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Jun 25 09:50:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA02433 for smp-outgoing; Wed, 25 Jun 1997 09:50:23 -0700 (PDT) Received: from sanjuan.cs.washington.edu (sanjuan.cs.washington.edu [128.95.8.118]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id JAA02427 for ; Wed, 25 Jun 1997 09:50:20 -0700 (PDT) Received: from localhost (ulbright@localhost) by sanjuan.cs.washington.edu (8.8.5+CS/7.2ws+) with SMTP id JAA10430 for ; Wed, 25 Jun 1997 09:50:19 -0700 (PDT) Date: Wed, 25 Jun 1997 09:50:19 -0700 (PDT) From: Christopher Ulbright To: freebsd-smp@freebsd.org Subject: Debugger Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Has anyone done debugger development for the SMP version? I'm working on the issue of what to do with the other processors when one cpu receives a breakpoint interrupt. My initial thought is to stop all processors via an IPI but I'm not sure what type of INT to send. Any ideas would be much appreciated. -chris ulbright From owner-freebsd-smp Wed Jun 25 10:13:27 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA03619 for smp-outgoing; Wed, 25 Jun 1997 10:13:27 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA03614 for ; Wed, 25 Jun 1997 10:13:24 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id LAA16124; Wed, 25 Jun 1997 11:12:52 -0600 (MDT) Message-Id: <199706251712.LAA16124@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Lars Fredriksen cc: smp@freebsd.org Subject: Re: Recent (last two days) smp kernel is hanging for me In-reply-to: Your message of "Wed, 25 Jun 1997 10:15:21 CDT." <199706251515.KAA00540@fredriks-1.pr.mcs.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 25 Jun 1997 11:12:51 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Lars, --- Would all current users of supermicro motherboards & SMP please send in a fresh mptable output? read further for details. --- > > so just for grins modify sys/i386/i386/mp_machdep.c: > > > > get_pci_apic_irq(int pciBus, int pciDevice, int pciInt) > > { > > int intr; > > > > --pciInt; /* zero based */ > > > > + if ( pciDevice == 17 ) return 16; > > + else if ( pciDevice == 18 ) return 17; > > + else if ( pciDevice == 19 ) return 18; > > > > for (intr = 0; intr < nintrs; ++intr) /* search each record */ > > > > Hi Steve, > Well it worked (for the most part)!!! The only thing that doesn't work > right now is X. The videocard seems to get its dma or something similar screwed > up. The cursor is there but the image is not (just a couple of random(seemingly > so) vertical lines). The output of mptable didn't change but that is probably > not surprising. I assumme that X works OK in same setup but with UP kernel. Does this card actually use INTs? So as long as grinning seems to work, can you rearrange the cards so that the VGA tries to use pin 20? In other words get it next to the ahc1 card so that it tries to share an INT with it. This will probably reguire minor changes to the patch I sent, but they should be obvious. Having said this I CANNOT think of any reason X should break relative to the APIC code... I've never seen a modern PCI VGA card that actually uses INTs. If anyone has theories for this please shout. So at least we seem to have confirmed that the mptable is the main problem. Now we have to figure out why yours is so obviously wrong, and no-one else with supermicro MBs has encountered it. Are there any options in the BIOS involving MP spec version? it should be 1.4 if settable. does anyone have an electronic version of the BIOS manual for the P6DNxxx boards, I couldn't find any at their web site. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Jun 25 11:51:38 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA08464 for smp-outgoing; Wed, 25 Jun 1997 11:51:38 -0700 (PDT) Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA08445 for ; Wed, 25 Jun 1997 11:50:50 -0700 (PDT) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.8.4/8.8.4) with SMTP id WAA19837; Wed, 25 Jun 1997 22:18:53 +0300 (EEST) Date: Wed, 25 Jun 1997 22:18:53 +0300 (EEST) From: Narvi To: Steve Passe cc: Lars Fredriksen , smp@FreeBSD.ORG Subject: Re: Recent (last two days) smp kernel is hanging for me In-Reply-To: <199706251712.LAA16124@Ilsa.StevesCafe.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Wed, 25 Jun 1997, Steve Passe wrote: > Lars, > > --- > Would all current users of supermicro motherboards & SMP please send in > a fresh mptable output? read further for details. > > --- > > > so just for grins modify sys/i386/i386/mp_machdep.c: > > > > > > get_pci_apic_irq(int pciBus, int pciDevice, int pciInt) > > > { > > > int intr; > > > > > > --pciInt; /* zero based */ > > > > > > + if ( pciDevice == 17 ) return 16; > > > + else if ( pciDevice == 18 ) return 17; > > > + else if ( pciDevice == 19 ) return 18; > > > > > > for (intr = 0; intr < nintrs; ++intr) /* search each record */ > > > > > > > Hi Steve, > > Well it worked (for the most part)!!! The only thing that doesn't work > > right now is X. The videocard seems to get its dma or something similar screwed > > up. The cursor is there but the image is not (just a couple of random(seemingly > > so) vertical lines). The output of mptable didn't change but that is probably > > not surprising. > > I assumme that X works OK in same setup but with UP kernel. Does this card > actually use INTs? > > So as long as grinning seems to work, can you rearrange the cards so that > the VGA tries to use pin 20? In other words get it next to the ahc1 card > so that it tries to share an INT with it. This will probably reguire minor > changes to the patch I sent, but they should be obvious. > > Having said this I CANNOT think of any reason X should break relative to > the APIC code... I've never seen a modern PCI VGA card that actually uses > INTs. If anyone has theories for this please shout. Well, actually they do. Or at least seem to be. If I remember right, the docu that came with my Diamont Video Vram 64 card claim that with the optional harhware MPEG daughtercard it uses interrupts (for sunc with sound or something like that). Also - the ATI cards seem to be using interrupts - maybe in certain situations, but they are. For what? I don't know. But you will lose an INT to it under Windows, and in the case of sharing berformance (no, it was not my machine and not a FreeBSD machine - SawPlus runs exclusively on that machine and the ATI has been replaced for a S3 since). Sander There is no love, no good, no happiness and no future - all these are just illusions. > > So at least we seem to have confirmed that the mptable is the main problem. > Now we have to figure out why yours is so obviously wrong, and no-one else > with supermicro MBs has encountered it. Are there any options in the BIOS > involving MP spec version? it should be 1.4 if settable. does anyone have > an electronic version of the BIOS manual for the P6DNxxx boards, I couldn't > find any at their web site. > -- > Steve Passe | powered by > smp@csn.net | Symmetric MultiProcessor FreeBSD > > > From owner-freebsd-smp Wed Jun 25 17:04:40 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id RAA21514 for smp-outgoing; Wed, 25 Jun 1997 17:04:40 -0700 (PDT) Received: from Kitten.mcs.com (Kitten.mcs.com [192.160.127.90]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id RAA21504 for ; Wed, 25 Jun 1997 17:04:28 -0700 (PDT) Received: from Mercury.mcs.net (fredriks@Mercury.mcs.net [192.160.127.80]) by Kitten.mcs.com (8.8.5/8.8.2) with ESMTP id TAA02756; Wed, 25 Jun 1997 19:04:25 -0500 (CDT) Received: (from fredriks@localhost) by Mercury.mcs.net (8.8.5/8.8.2) id TAA02493; Wed, 25 Jun 1997 19:04:25 -0500 (CDT) From: Lars Fredriksen Message-Id: <199706260004.TAA02493@Mercury.mcs.net> Subject: Re: Recent (last two days) smp kernel is hanging for me To: smp@csn.net (Steve Passe) Date: Wed, 25 Jun 1997 19:04:24 -0500 (CDT) Cc: lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG In-Reply-To: <199706251712.LAA16124@Ilsa.StevesCafe.com> from "Steve Passe" at Jun 25, 97 11:12:51 am X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Steve Passe writes: > > > I assumme that X works OK in same setup but with UP kernel. Does this card > actually use INTs? > > So as long as grinning seems to work, can you rearrange the cards so that > the VGA tries to use pin 20? In other words get it next to the ahc1 card > so that it tries to share an INT with it. This will probably reguire minor > changes to the patch I sent, but they should be obvious. > > Having said this I CANNOT think of any reason X should break relative to > the APIC code... I've never seen a modern PCI VGA card that actually uses > INTs. If anyone has theories for this please shout. > > So at least we seem to have confirmed that the mptable is the main problem. > Now we have to figure out why yours is so obviously wrong, and no-one else > with supermicro MBs has encountered it. Are there any options in the BIOS > involving MP spec version? it should be 1.4 if settable. does anyone have > an electronic version of the BIOS manual for the P6DNxxx boards, I couldn't > find any at their web site. Hi Steve, Well I guess I didn't try hard enough. A complete poweroff seemed to have fixed the video controller. The only thing I can think off here is that the Imagine-128-2 doesn't alway initialize correctly. There are some issues with the X server for that cards in other areas. So it appears that The only problem right now is my mptable. I'll send you a hard copy of the BIOS manual first thing in the morning. Thanks for all the help folks! Lars -- ------------------------------------------------------------------- Lars Fredriksen fredriks@mcs.com (home) lars@fredriks-2.pr.mcs.net (home-home) From owner-freebsd-smp Wed Jun 25 17:12:35 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id RAA21955 for smp-outgoing; Wed, 25 Jun 1997 17:12:35 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id RAA21949 for ; Wed, 25 Jun 1997 17:12:30 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id SAA17440; Wed, 25 Jun 1997 18:12:04 -0600 (MDT) Message-Id: <199706260012.SAA17440@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Lars Fredriksen cc: lars@fredriks-1.pr.mcs.net, smp@FreeBSD.ORG Subject: Re: Recent (last two days) smp kernel is hanging for me In-reply-to: Your message of "Wed, 25 Jun 1997 19:04:24 CDT." <199706260004.TAA02493@Mercury.mcs.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 25 Jun 1997 18:12:04 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Lars, > Well I guess I didn't try hard enough. A complete poweroff seemed >to have fixed the video controller. The only thing I can think off here >is that the Imagine-128-2 doesn't alway initialize correctly. There are some >issues with the X server for that cards in other areas. So it appears that >The only problem right now is my mptable. I'll send you a hard copy of the >BIOS manual first thing in the morning. this is good news, I was at a complete loss for a way to associate the APIC code and failure of a VGA card to operate (short of a special case like the MPEG hardware mentioned by an earlier poster). HINT-HINT: I still need mptable outputs for supermicro MBs from others running SMP on them! -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Jun 25 19:33:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA28698 for smp-outgoing; Wed, 25 Jun 1997 19:33:15 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA28686 for ; Wed, 25 Jun 1997 19:32:49 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id UAA17883; Wed, 25 Jun 1997 20:32:35 -0600 (MDT) Message-Id: <199706260232.UAA17883@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Christopher Ulbright cc: freebsd-smp@FreeBSD.ORG Subject: Re: Debugger In-reply-to: Your message of "Wed, 25 Jun 1997 09:50:19 PDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 25 Jun 1997 20:32:35 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, some suggested pseudo-code follows. please comment. ------------------------------------- cut ------------------------------------- --- /* * example usage: */ breakpoint() { ... stop_other_cpus(cpu_map); do_stuff(); restart_other_cpus(stopped_cpus); ... } --- i386/include/smp.h: /* global data in i386/isa/apic_vector.s: */ extern volatile u_int stopped_cpus; extern volatile u_int started_cpus; extern u_int cpu_map; --- i386/isa/intr_machdep.h: #define XCPUSTOP_OFFSET 33 inthand_t Xcpustop; --- i386/i386/mp_machdep.c: /* * existing functions: */ mp_enable(u_int boot_addr) { ... #if defined( APIC_IO ) ... + /* install an inter-CPU IPI for CPU stop/restart */ + setidt(ICU_OFFSET + XCPUSTOP_OFFSET, Xcpustop, + SDT_SYS386IGT, SEL_KPL, GSEL(GCODE_SEL, SEL_KPL)); #endif /* APIC_IO */ ... } start_all_aps(u_int boot_addr) { ... /* start each AP */ for (x = 1; x <= mp_naps; ++x) { ... + cpu_map |= (1 << x); /* record AP in global map */ } ... + cpu_map |= 1; /* record BSP in global map */ } /* * new functions: */ /* * When called the executing CPU will send an IPI to all other CPUs * requesting that they halt execution. * * Usually (but not necessarily) called with 'cpu_map' as its arg. * * - Signals all CPUs in map to stop. * - Waits for each to stop. * * Returns: * -1: error * 0: NA * 1: ok * * XXX FIXME: this is not MP-safe, needs a lock to prevent multiple CPUs * from executing at same time. */ int stop_other_cpus( u_int map ) { if (!smp_active) return 0; stopped_cpus = 0; map &= ~(1 << cpuid); /* clear our bit */ /* send IPI to all CPUs in map */ selected_apic_ipi(map, ICU_OFFSET + XSTOP_CPUS_OFFSET, APIC_DELMODE_FIXED); while (stopped_cpus != map) /* spin */ ; return 1; } /* * Called by a CPU to restart stopped CPUs. * * Usually (but not necessarily) called with 'stopped_cpus' as its arg. * * - Signals all CPUs in map to restart. * - Waits for each to restart. * * Returns: * -1: error * 0: NA * 1: ok */ int restart_other_cpus( u_int map ) { if (!smp_active) return 0; map &= ~(1 << cpuid); /* XXX paranoia */ started_cpus = map; /* signal other cpus to restart */ while (started_cpus) /* wait for each to clear its bit */ /* spin */ ; return 1; } --- i386/isa/apic_vector.s: .globl _stopped_cpus _stopped_cpus: .long 0 .globl _started_cpus _started_cpus: .long 0 .globl _cpu_map _cpu_map: .long 0 /* * Executed by a CPU when it receives an Xcpustop IPI from another CPU, * * - Signals its receipt. * - Waits for permission to restart. * - Signals its restart. */ .text SUPERALIGN_TEXT .globl _Xcpustop _Xcpustop: pushl %eax pushl %ds /* save current data segment */ movl $KDSEL, %eax movl %ax, %ds /* use KERNEL data segment */ movl _cpuid, %eax /* id */ lock btsl %eax, _stopped_cpus /* stopped_cpus |= (1< Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA02445 for smp-outgoing; Wed, 25 Jun 1997 21:20:08 -0700 (PDT) Received: from Kitten.mcs.com (Kitten.mcs.com [192.160.127.90]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA02437 for ; Wed, 25 Jun 1997 21:20:06 -0700 (PDT) Received: from Mercury.mcs.net (fredriks@Mercury.mcs.net [192.160.127.80]) by Kitten.mcs.com (8.8.5/8.8.2) with ESMTP id XAA10653 for ; Wed, 25 Jun 1997 23:19:58 -0500 (CDT) Received: (from fredriks@localhost) by Mercury.mcs.net (8.8.5/8.8.2) id XAA17826 for smp@freebsd.org.; Wed, 25 Jun 1997 23:19:58 -0500 (CDT) From: Lars Fredriksen Message-Id: <199706260419.XAA17826@Mercury.mcs.net> Subject: Silo overflows with SMP kernel To: smp@freebsd.org Date: Wed, 25 Jun 1997 23:19:58 -0500 (CDT) X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, Just a quick comment that I have noticed. This might have something to do with the granularity of locking that we are currently running with(ie giant lock), but I have seen a fair amount of silo overflows while running the SMP kernel. I have a 64Kbs ppp link that from time to time have a medium load on it.(ie big ftp transfers) Is this to be expected until the locking gets more granular or is this related to something else? Lars -- ------------------------------------------------------------------- Lars Fredriksen fredriks@mcs.com (home) lars@fredriks-2.pr.mcs.net (home-home) From owner-freebsd-smp Wed Jun 25 22:12:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA03940 for smp-outgoing; Wed, 25 Jun 1997 22:12:23 -0700 (PDT) Received: from jumping-spider.aracnet.com (root@jumping-spider.aracnet.com [204.188.47.14]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA03935 for ; Wed, 25 Jun 1997 22:12:20 -0700 (PDT) Received: from speedy (ppp-a31.aracnet.com [205.238.13.158]) by jumping-spider.aracnet.com (8.8.5/8.8.5) with ESMTP id WAA30681; Wed, 25 Jun 1997 22:15:30 -0700 Message-ID: <33B2073E.DEBDE29F@aracnet.com> Date: Wed, 25 Jun 1997 22:07:58 -0800 From: Chris Browning Reply-To: cbrown@aracnet.com X-Mailer: Mozilla 4.01 [en] (WinNT; I) MIME-Version: 1.0 To: Torbjorn Granlund CC: freebsd-smp@FreeBSD.ORG Subject: Re: New monster server X-Priority: 3 (Normal) References: <199706250140.DAA01747@squid.pdc.kth.se> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > My company intends to purchase a new compute and disk server early > this > fall. We originally intended to get an Alpha system, but it seems > hard to > get a reasonable OS without paying an outrageous amount of money. > (NetBSD > might be good enough for the alpha platform; but it doesn't seem to > get a > lot of usage.) Well, it sounds like you are looking for one of the free Un*xes. The price is certainly right :-) > We would like the machine to have on the order of 25 SPECint95 > (summing the > SPECint95 on the individual CPUs in case we go for an SMP solution). > So a > single CPU 21164 won't do. A single cpu 21264 would do. A 4 CPU > P6/200MHz/512KB would do. Hmm, I guess a 4-way Pentium(R) Pro Processor system would do. I think that the benchmarks say 8.5 SPECint95 and 6.5 SPECfp95. I'm not to sure that you can "add" them together to come up with a 26 SPECint95 machine :-). > (It seems difficult to get 4 CPU P6 > motherboards, > since the Orion isn't made, and the 440BX is not yet available.) Now, you TOTAL confused me here. The 82450GX (Orion) isn't made? Please tell our factory! We build boards every day with this chipset. Believe me, it is alive and well. Also, you should have no trouble getting a 4-way PPP box. I can name atleast 6 vendors off the top of my head. > I'd be interessted in hearing your comments. Realistically, will I be > able > to get a 4 CPU FreeBSD system to work well enough this fall? If not, > when > do you think that could happen? Hmm, I am pretty new at FreeBSD, but 4way Linux does work right now. > I am willing to spend about $20000 for the hardware. It is highly > unlikely > that Digital will want to sell me a dual CPU 21164 or a single CPU > 21264 for > that amount money. Will a 4 CPU P6 be cheap enough and will it be > stable > enough to use for heavy load? You can certainly get a 4way PPP system for $20000. Of course, this depends on what HW you want to through on it. You might want to check http://www.onsale.com. They have 4way capable boxes for sale almost every day on there. As for the "heavy load". I am not sure what type of load you will be dealing with, but the 4 way PPP system have set many price/performance records on the TPC-C benchmarks. (Yikes, that marketing side of me is coming out :-). Chris Disclaimer: I don't speak for anyone, especially Intel. From owner-freebsd-smp Wed Jun 25 22:24:37 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA04297 for smp-outgoing; Wed, 25 Jun 1997 22:24:37 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA04288 for ; Wed, 25 Jun 1997 22:24:29 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.5/8.8.5) id AAA00225; Thu, 26 Jun 1997 00:24:22 -0500 (EST) From: "John S. Dyson" Message-Id: <199706260524.AAA00225@dyson.iquest.net> Subject: Re: Silo overflows with SMP kernel In-Reply-To: <199706260419.XAA17826@Mercury.mcs.net> from Lars Fredriksen at "Jun 25, 97 11:19:58 pm" To: fredriks@Mcs.Net (Lars Fredriksen) Date: Thu, 26 Jun 1997 00:24:21 -0500 (EST) Cc: smp@FreeBSD.ORG Reply-To: dyson@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Hi, > Just a quick comment that I have noticed. This might have > something to do with the granularity of locking that we are currently > running with(ie giant lock), but I have seen a fair amount of silo > overflows while running the SMP kernel. I have a 64Kbs ppp link that > from time to time have a medium load on it.(ie big ftp transfers) > > Is this to be expected until the locking gets more granular or is this > related to something else? > I have been seeing that also... I have a modified SIO driver that fully uses the new 32 byte fifo chips with automatic hardware RTS/CTS flow control. Haven't committed it yet, because it covers up the problem. The good news is that it is likely due to the "big lock" SMP kernel model. If you want a copy of the modified SIO driver, let me know. I am not comfortable enough with it yet to commit it though. Right now, I am working on finer grained VM locking which should be a first step on the VFS/VM high level side to make things better. (I am not one of the "main" SMP people, but am doing what I can do to support them.) By the weekend, my copy of the VM code should be SMP safe, all the way down to the VM object/page level. There are still many many issues to make it "correct", but it is getting closer. Essentially, I should be able to handle a page fault (with no disk I/O) without doing the "big lock" thing. In fact, the VM code should be able to handle simultaneous user page faults on different processors, and manage the locking when data structures are in common. This stuff is less trivial than I had thought... (Translation: it is DA*N difficult.) I am working from 12:00 Noon to 3AM trying to understand and work all of the issues. It is not likely that the stuff that I'll have this weekend will be the final version, but it will be a first step on making the higher VM code real-SMP capable. (Note that I am building on the hard SMP work that has been done by the SMP guys...) John From owner-freebsd-smp Thu Jun 26 08:06:35 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id IAA25102 for smp-outgoing; Thu, 26 Jun 1997 08:06:35 -0700 (PDT) Received: from agora.rdrop.com (root@agora.rdrop.com [199.2.210.241]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id IAA25089 for ; Thu, 26 Jun 1997 08:06:31 -0700 (PDT) Received: from amaretto.csv.warwick.ac.uk (csubl@amaretto.csv.warwick.ac.uk [137.205.148.138]) by agora.rdrop.com (8.8.5/8.8.5) with ESMTP id IAA23427 for ; Thu, 26 Jun 1997 08:06:04 -0700 (PDT) From: Mr M P Searle Message-Id: <6632.199706261459@amaretto.csv.warwick.ac.uk> Received: by amaretto.csv.warwick.ac.uk id PAA06632; Thu, 26 Jun 1997 15:59:32 +0100 (BST) Subject: Re: New monster server In-Reply-To: <33B2073E.DEBDE29F@aracnet.com> from Chris Browning at "Jun 25, 97 10:07:58 pm" To: cbrown@aracnet.com Date: Thu, 26 Jun 1997 15:59:26 +0100 (BST) Cc: tege@pdc.kth.se, freebsd-smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > My company intends to purchase a new compute and disk server early > > this > > fall. We originally intended to get an Alpha system, but it seems > > hard to > > get a reasonable OS without paying an outrageous amount of money. > > (NetBSD > > might be good enough for the alpha platform; but it doesn't seem to > > get a > > lot of usage.) > > Well, it sounds like you are looking for one of the free Un*xes. The > price is certainly right :-) > > > We would like the machine to have on the order of 25 SPECint95 > > (summing the > > SPECint95 on the individual CPUs in case we go for an SMP solution). > > So a > > single CPU 21164 won't do. A single cpu 21264 would do. A 4 CPU > > P6/200MHz/512KB would do. > > Hmm, I guess a 4-way Pentium(R) Pro Processor system would do. I think > that > the benchmarks say 8.5 SPECint95 and 6.5 SPECfp95. I'm not to sure that > you can > "add" them together to come up with a 26 SPECint95 machine :-). If this does mean that a 4 CPU machine isn't enough, I have heard of a 6 CPU P6. I'm not sure that you can get it for $20000 though, and I don't have the URL. From owner-freebsd-smp Thu Jun 26 08:20:55 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id IAA25740 for smp-outgoing; Thu, 26 Jun 1997 08:20:55 -0700 (PDT) Received: from cs.utah.edu (cs.utah.edu [128.110.4.21]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id IAA25734 for ; Thu, 26 Jun 1997 08:20:52 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id JAA15097; Thu, 26 Jun 1997 09:20:34 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id JAA09740; Thu, 26 Jun 1997 09:20:28 -0600 Date: Thu, 26 Jun 1997 09:20:28 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199706261520.JAA09740@fast.cs.utah.edu> To: cbrown@aracnet.com, csubl@csv.warwick.ac.uk Subject: Re: New monster server Cc: freebsd-smp@freebsd.org, tege@pdc.kth.se Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Also, for $100k you can get a 10-way Unisys server. 8-way Pros should be out in serious force this summer. From owner-freebsd-smp Thu Jun 26 09:34:00 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA28954 for smp-outgoing; Thu, 26 Jun 1997 09:34:00 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id JAA28949; Thu, 26 Jun 1997 09:33:47 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id JAA07792; Thu, 26 Jun 1997 09:21:47 -0700 From: Terry Lambert Message-Id: <199706261621.JAA07792@phaeton.artisoft.com> Subject: Re: Silo overflows with SMP kernel To: dyson@FreeBSD.ORG Date: Thu, 26 Jun 1997 09:21:47 -0700 (MST) Cc: fredriks@Mcs.Net, smp@FreeBSD.ORG In-Reply-To: <199706260524.AAA00225@dyson.iquest.net> from "John S. Dyson" at Jun 26, 97 00:24:21 am X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Right now, I am working on finer grained VM locking which should be > a first step on the VFS/VM high level side to make things better. (I am > not one of the "main" SMP people, but am doing what I can do to support > them.) Is this locking for which deadlock detection should be done, or is it purely opportunistic? I ask because the locks probably need to be checked agains non-VM locks for deadly embrace deadlocks, if it's not. > By the weekend, my copy of the VM code should be SMP safe, all the > way down to the VM object/page level. There are still many many > issues to make it "correct", but it is getting closer. Essentially, > I should be able to handle a page fault (with no disk I/O) without > doing the "big lock" thing. In fact, the VM code should be able > to handle simultaneous user page faults on different processors, and > manage the locking when data structures are in common. JOHN DYSON IS A STUD! Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Thu Jun 26 11:16:04 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA03844 for smp-outgoing; Thu, 26 Jun 1997 11:16:04 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA03780; Thu, 26 Jun 1997 11:15:23 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.5/8.8.5) id NAA12413; Thu, 26 Jun 1997 13:13:57 -0500 (EST) From: "John S. Dyson" Message-Id: <199706261813.NAA12413@dyson.iquest.net> Subject: Re: Silo overflows with SMP kernel In-Reply-To: <199706261621.JAA07792@phaeton.artisoft.com> from Terry Lambert at "Jun 26, 97 09:21:47 am" To: terry@lambert.org (Terry Lambert) Date: Thu, 26 Jun 1997 13:13:57 -0500 (EST) Cc: dyson@FreeBSD.ORG, fredriks@Mcs.Net, smp@FreeBSD.ORG Reply-To: dyson@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > Right now, I am working on finer grained VM locking which should be > > a first step on the VFS/VM high level side to make things better. (I am > > not one of the "main" SMP people, but am doing what I can do to support > > them.) > > Is this locking for which deadlock detection should be done, or is > it purely opportunistic? I ask because the locks probably need to > be checked agains non-VM locks for deadly embrace deadlocks, if it's > not. > Yep... That is where alot of the complexity is coming in. > > JOHN DYSON IS A STUD! > Not really, lets see how things work first :-). Also, if I was really a stud, I wouldn't be working on FreeBSD all night, but would have other interests :-). Now, I am working on FreeBSD for both my full time job and after hours... The stuff is going to be preliminary, and there is a lot of coordination that is needed between me and the "real" SMP crew. What I am doing is mostly going to be a "proposal" to both the SMP people, and my co-workers at work for review and criticism. As I know that you know, this has to be done correctly -- I really don't want to create another "interesting" thing like our VFS layering. :-). John From owner-freebsd-smp Thu Jun 26 18:44:51 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id SAA26220 for smp-outgoing; Thu, 26 Jun 1997 18:44:51 -0700 (PDT) Received: from chaos.amber.org (root@chaos.amber.org [205.231.232.12]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id SAA26214 for ; Thu, 26 Jun 1997 18:44:45 -0700 (PDT) Received: from [205.231.232.197] (ab2-005.dial.nova.org [205.231.232.197]) by chaos.amber.org (8.7.5/8.6.12) with SMTP id VAA00608; Thu, 26 Jun 1997 21:44:11 -0400 (EDT) Message-Id: <199706270144.VAA00608@chaos.amber.org> Subject: Re: New monster server Date: Thu, 26 Jun 97 21:44:11 -0400 x-sender: petrilli@mx.amber.org x-mailer: Claris Emailer 2.0, March 15, 1997 From: Christopher Petrilli To: "Mr M P Searle" , cc: , Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk I will only remind people involved in these discussions that processor performance is RARELY the bottleneck in a system, I/O almost always is. If it isn't I/O, you're not measuring correctly, and if you are, then it's memory. THrowing processors at a disk/IO intensive application will not help one bit, so until you're sure what the problem is, measure it :-) Christopher Petrilli -- | Christopher Petrilli "That's right you're | petrilli@amber.org not from Texas." From owner-freebsd-smp Fri Jun 27 10:12:06 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA02426 for smp-outgoing; Fri, 27 Jun 1997 10:12:06 -0700 (PDT) Received: from bagpuss.visint.co.uk (bagpuss.visint.co.uk [194.207.134.1]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA02403 for ; Fri, 27 Jun 1997 10:11:48 -0700 (PDT) Received: from dylan.visint.co.uk (dylan.visint.co.uk [194.207.134.180]) by bagpuss.visint.co.uk (8.7.5/8.7.3) with SMTP id SAA08477 for ; Fri, 27 Jun 1997 18:10:54 +0100 (BST) Date: Fri, 27 Jun 1997 18:10:54 +0100 (BST) From: Stephen Roome To: smp@freebsd.org Subject: Weird segfaults etc.. Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I've just upgraded (?) to 3.0-970618-SNAP with SMP, and I'm getting weird errors this time (last time I supped it down on top of current about 2 months ago it was rock solid). I tend to be getting stuff like, cc1: died with internal signal 10 or 11 quite often. I saw a similar problem on a standard 2.2.2 machine, but only once, this happens more regularly while trying to build kernels and the like. I can't find anything about this in the bugs list at the website and thought someone might have a clue? Anyway, here's the mptable output, I think mptable might be screwing up, because I sure as hell put SMP_AUTOSTART in the kernel, and it did it as well. (I booted with -v so it should give enough info below to explain the hardware which is a GA586-DX(Rev 3B) + 2*P133, fxp0, graphics and memory (32MB) -- minimalist enough ?) mptable -dmesg -verbose ... Ta.. =============================================================================== MPTable, version 2.0.11 looking for EBDA pointer @ 0x040e, NOT found searching CMOS 'top of mem' @ 0x0009fc00 (639K) searching BIOS @ 0x000f0000 MP FPS found in BIOS @ physical addr: 0x000f0dd0 ------------------------------------------------------------------------------- MP Floating Pointer Structure: location: BIOS physical address: 0x000f0dd0 signature: '_MP_' length: 16 bytes version: 1.1 checksum: 0xa3 mode: Virtual Wire ------------------------------------------------------------------------------- MP Config Table Header: physical address: 0x000f0de4 signature: 'PCMP' base table length: 292 version: 1.1 checksum: 0x2e OEM ID: 'OEM00000' Product ID: 'PROD00000000' OEM table pointer: 0x00000000 OEM table size: 0 entry count: 28 local APIC address: 0xfee00000 extended table length: 0 extended table checksum: 0 ------------------------------------------------------------------------------- MP Config Base Table Entries: -- Processors: APIC ID Version State Family Model Step Flags 0 0x11 BSP, usable 5 2 1 0x07bf 1 0x11 AP, usable 5 2 1 0x07bf -- Bus: Bus ID Type 0 PCI 1 ISA -- I/O APICs: APIC ID Version State Address 2 0x11 usable 0xfec00000 -- I/O Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# ExtINT conforms conforms 1 0 2 0 INT conforms conforms 1 1 2 1 INT conforms conforms 1 0 2 2 INT conforms conforms 1 3 2 3 INT conforms conforms 1 4 2 4 INT conforms conforms 1 5 2 5 INT conforms conforms 1 6 2 6 INT conforms conforms 1 7 2 7 INT conforms conforms 1 8 2 8 INT conforms conforms 1 9 2 9 INT conforms conforms 1 10 2 10 INT conforms conforms 1 11 2 11 INT conforms conforms 1 12 2 12 INT conforms conforms 1 13 2 13 INT conforms conforms 1 14 2 14 INT conforms conforms 1 15 2 15 INT active-lo level 0 8:A 2 16 INT active-lo level 0 9:A 2 17 INT active-lo level 0 10:A 2 18 INT active-lo level 0 12:A 2 19 SMI conforms conforms 1 0 2 23 -- Local Ints: Type Polarity Trigger Bus ID IRQ APIC ID INT# ExtINT conforms conforms 0 0:A 255 0 NMI conforms conforms 0 0:A 255 1 ------------------------------------------------------------------------------- # SMP kernel config file options: # Required: options SMP # Symmetric MultiProcessor Kernel options APIC_IO # Symmetric (APIC) I/O # Useful: #options SMP_AUTOSTART # start the additional CPUs during boot # Optional (built-in defaults will work in most cases): #options NCPU=2 # number of CPUs #options NBUS=2 # number of busses #options NAPIC=1 # number of IO APICs #options NINTR=24 # number of INTs # Currently broken: #options SMP_PRIVPAGES # BROKEN, DO NOT use! # Rogue hardware: # # Tyan Tomcat II: #options SMP_TIMER_NC # # # SuperMicro P6DNE: #options SMP_TIMER_NC # ------------------------------------------------------------------------------- dmesg output: Copyright (c) 1992-1997 FreeBSD Inc. Copyright (c) 1982, 1986, 1989, 1991, 1993 The Regents of the University of California. All rights reserved. FreeBSD 3.0-970618-SNAP #0: Thu Jun 26 18:29:33 BST 1997 root@yakko.visint.co.uk:/usr/src/sys/compile/YAKKO FreeBSD/SMP: Multiprocessor motherboard cpu0 (BSP): apic id: 0, version: 0x00030010 cpu1 (AP): apic id: 1, version: 0x00030010 io0 (APIC): apic id: 2, version: 0x00170011 Calibrating clock(s) ... i8254 clock: 1193581 Hz CLK_USE_I8254_CALIBRATION not specified - using default frequency CPU: Pentium (586-class CPU) Origin = "GenuineIntel" Id = 0x52c Stepping=12 Features=0x3bf real memory = 33554432 (32768K bytes) avail memory = 30830592 (30108K bytes) pci_open(1): mode 1 addr port (0x0cf8) is 0x8000005c pci_open(1a): mode1res=0x80000000 (0x80000000) pci_cfgcheck: device 0 [class=060000] [hdr=00] is there (id=12508086) Probing for devices on PCI bus 0: found-> vendor=0x8086, dev=0x1250, revid=0x03 class=06-00-00, hdrtype=0x00, mfdev=0 chip0: rev 0x03 on pci0.0.0 found-> vendor=0x8086, dev=0x7000, revid=0x01 class=06-01-00, hdrtype=0x00, mfdev=1 chip1: rev 0x01 on pci0.7.0 found-> vendor=0x8086, dev=0x7010, revid=0x00 class=01-01-80, hdrtype=0x00, mfdev=0 chip2: rev 0x00 on pci0.7.1 I/O Recovery Timing: 8-bit 3.5 clocks, 16-bit 3.5 clocks Extended BIOS: disabled Lower BIOS: disabled Coprocessor IRQ13: disabled Mouse IRQ12: disabled Interrupt Routing: A: , B: , C: , D: MB0: , MB1: Freeing (NOT implemented) redirected PCI irq 15. found-> vendor=0x8086, dev=0x1229, revid=0x01 class=02-00-00, hdrtype=0x00, mfdev=0 intpin=a, irq=18 map[0]: type 3, range 32, base e1100000, size 12 map[1]: type 4, range 32, base 00006400, size 5 map[2]: type 1, range 32, base e1000000, size 20 fxp0: rev 0x01 int a irq 18 on pci0.10.0 fxp0: Ethernet address 00:a0:c9:2f:85:41 bpf: fxp0 attached found-> vendor=0x1002, dev=0x5654, revid=0x48 class=03-00-00, hdrtype=0x00, mfdev=0 map[0]: type 1, range 32, base e0000000, size 24 map[1]: type 4, range 32, base 00006500, size 8 vga0: rev 0x48 on pci0.11.0 Freeing (NOT implemented) redirected PCI irq 11. found-> vendor=0x9004, dev=0x8078, revid=0x00 class=01-00-00, hdrtype=0x00, mfdev=0 intpin=a, irq=19 map[0]: type 4, range 32, base 00006000, size 8 map[1]: type 1, range 32, base e1101000, size 12 ahc0: rev 0x00 int a irq 19 on pci0.12.0 ahc0: Reading SEEPROM...checksum error ahc0: No SEEPROM available ahc0: Using left over BIOS settings ahc0: aic7880 Wide Channel, SCSI Id=7, 16 SCBs ahc0: Resetting Channel A ahc0: Downloading Sequencer Program...ahc0: 369 instructions downloaded Done ahc0: Probing channel A ahc0: waiting for scsi devices to settle scbus0 at ahc0 bus 0 ahc0: target 0 synchronous at 10.0MHz, offset = 0xf sd0 at scbus0 target 0 lun 0 sd0: type 0 fixed SCSI 2 sd0: Direct-Access 3067MB (6281856 512 byte sectors) sd0: with 6810 cyls, 5 heads, and an average 184 sectors/track ahc0: target 1 synchronous at 10.0MHz, offset = 0xf sd1 at scbus0 target 1 lun 0 sd1: type 0 fixed SCSI 2 sd1: Direct-Access 3067MB (6281856 512 byte sectors) sd1: with 6810 cyls, 5 heads, and an average 184 sectors/track Probing for devices on the ISA bus: sc0: the current keyboard controller command byte 0047 kbdio: RESET_KBD return code:00fa kbdio: RESET_KBD status:00aa sc0 at 0x60-0x6f irq 1 on motherboard sc0: BIOS video mode:3 sc0: VGA registers upon power-up 50 18 10 00 10 00 03 00 02 67 5f 4f 50 82 55 81 bf 1f 00 4f 0e 0f 00 00 ff ff 9c 8e 8f 28 1f 96 b9 a3 ff 00 01 02 03 04 05 14 07 38 39 3a 3b 3c 3d 3e 3f 0c 00 0f 08 00 00 00 00 00 10 0e 00 ff sc0: video mode:24 sc0: VGA registers for mode:24 50 18 10 00 10 00 03 00 02 67 5f 4f 50 82 55 81 bf 1f 00 4f 0d 0e 00 00 00 00 9c 8e 8f 28 1f 96 b9 a3 ff 00 01 02 03 04 05 14 07 38 39 3a 3b 3c 3d 3e 3f 0c 00 0f 08 00 00 00 00 00 10 0e 00 ff sc0: VGA color <16 virtual consoles, flags=0x0> sio0 at 0x3f8-0x3ff irq 4 on isa sio0: type 16550A sio1 at 0x2f8-0x2ff irq 3 on isa sio1: type 16550A lpt0 at 0x378-0x37f irq 7 on isa lpt0: Interrupt-driven port lp0: TCP/IP capable interface bpf: lp0 attached lpt1 not found fdc0 at 0x3f0-0x3f7 irq 6 drq 2 on isa fdc0: NEC 72065B fd0: 1.44MB 3.5in npx0 on motherboard npx0: INT 16 interface imasks: bio c0080040, tty c304009a, net c304009a BIOS Geometries: 0:03fe3f20 0..1022=1023 cylinders, 0..63=64 heads, 1..32=32 sectors 1:03fe3f20 0..1022=1023 cylinders, 0..63=64 heads, 1..32=32 sectors 0 accounted for Device configuration finished. Considering FFS root f/s. configure() finished. Enabled INTs: 1, 2, 3, 4, 6, 7, 8, 18, 19, imen: 0x00f3fe21 bpf: tun0 attached bpf: sl0 attached bpf: lo0 attached sd0s1: type 0xa5, start 0, end = 6281855, size 6281856 sd0s1: C/H/S end 391/6/63 (172871) != end 6281855: invalid SMP: All idle procs online. SMP: *** AUTO *** starting 1st AP! SMP: AP CPU #1 LAUNCHED!! Starting Scheduling... SMP: TADA! CPU #1 made it into the scheduler!. SMP: All 2 CPU's are online! =============================================================================== Steve Roome - Vision Interactive Ltd. Tel:+44(0)117 9730597 Home:+44(0)976 241342 WWW: http://dylan.visint.co.uk/ From owner-freebsd-smp Fri Jun 27 11:23:25 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA05697 for smp-outgoing; Fri, 27 Jun 1997 11:23:25 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA05681 for ; Fri, 27 Jun 1997 11:23:02 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id MAA02798; Fri, 27 Jun 1997 12:21:08 -0600 (MDT) Message-Id: <199706271821.MAA02798@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Stephen Roome cc: smp@freebsd.org Subject: Re: Weird segfaults etc.. In-reply-to: Your message of "Fri, 27 Jun 1997 18:10:54 BST." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 27 Jun 1997 12:21:08 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > I've just upgraded (?) to 3.0-970618-SNAP with SMP, and I'm getting weird > errors this time (last time I supped it down on top of current about 2 > months ago it was rock solid). > > I tend to be getting stuff like, cc1: died with internal signal 10 or 11 > quite often. I saw a similar problem on a standard 2.2.2 machine, but only > once, this happens more regularly while trying to build kernels and the > like. there were minor changes on 5/30, but they are of the nature of "should work or fail". the next major changes were on 6/22, the SMP_PRIVPAGES code. I strongly suggest cvsupping -current (complete world, not just sys) and remaking both kernel and world. --- > Anyway, here's the mptable output, I think mptable might be screwing up, > because I sure as hell put SMP_AUTOSTART in the kernel, and it did it as > well. if you are refering to the lines: # Useful: #options SMP_AUTOSTART # start the additional CPUs you are misunderstanding how the mptable program works. mptable merely reads the MP spec data from the hardware as the BIOS sets it up. It will read the same reguardless of whether a UP or an SMP kernel is currently booted. "option" lines do NOT come from MP spec data, but instead are compiled into the mptable program. they reflect your current choices (as of the time mptable was last modified) for configuring your SMP kernel. the presense or absense of a leading '#' char on a line reflects whether that option would be used in a 'GENERIC' kernel, and is a judgement call in most cases. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Fri Jun 27 11:25:53 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA05859 for smp-outgoing; Fri, 27 Jun 1997 11:25:53 -0700 (PDT) Received: from Symbion.srrc.usda.gov ([199.78.118.118]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA05837 for ; Fri, 27 Jun 1997 11:25:15 -0700 (PDT) Received: from Symbion (localhost [127.0.0.1]) by Symbion.srrc.usda.gov (8.8.5/8.8.5) with ESMTP id NAA00284 for ; Fri, 27 Jun 1997 13:22:58 -0500 (CDT) Message-Id: <199706271822.NAA00284@Symbion.srrc.usda.gov> X-Mailer: exmh version 2.0gamma 1/27/96 To: freebsd-smp@freebsd.org From: Glenn Johnson Subject: kernel panic on dual processor Tyan Titan Pro AT Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 27 Jun 1997 13:22:57 -0500 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Greetings, I have two MP Pentium Pro machines. One is a Micron Millenia Pro, not sure of motherboard brand, the other is a MaxVision with a Tyan Titan Pro AT motherboard. Each has two 200 MHz Pentium Pro chips, and both were running the 3.0-970615-SNAP with both processors in each machine working fine. I just updated the source yesterday via cvsup on both machines, did a 'make world' on each, rebuilt the kernels, and rebooted. The Micron has no problems but the Tyan Titan Pro AT system gives me a kernel panic upon boot, and the machine does an automatic reboot. I don't have the exact message in front of me, but can get it. I backed out my kernel source to 6-15-97 and rebuilt the kernel again but this did not help. Does any one have any ideas? Do I need to go back to a prior date of 3.0, not just kernel source? Or did I lose a CPU? Thanks. -- Glenn Johnson Phone: (504) 286-4252 USDA-ARS-SRRC; CFQ FAX: (504) 286-4217 1100 Robert E. Lee email: gjohnson@nola.srrc.usda.gov New Orleans, LA 70124 --------------------------------------------------------------------------- From owner-freebsd-smp Fri Jun 27 14:52:27 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA17328 for smp-outgoing; Fri, 27 Jun 1997 14:52:27 -0700 (PDT) Received: from Symbion.srrc.usda.gov ([199.78.118.118]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA17319 for ; Fri, 27 Jun 1997 14:52:19 -0700 (PDT) Received: from Symbion (localhost [127.0.0.1]) by Symbion.srrc.usda.gov (8.8.5/8.8.5) with ESMTP id QAA00243 for ; Fri, 27 Jun 1997 16:50:35 -0500 (CDT) Message-Id: <199706272150.QAA00243@Symbion.srrc.usda.gov> X-Mailer: exmh version 2.0gamma 1/27/96 To: freebsd-smp@freebsd.org From: Glenn Johnson Subject: Follow up to problem with Tyan Titan Pro AT Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 27 Jun 1997 16:50:35 -0500 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I posted a question earlier about my system with a Tyan Titan Pro AT motherboard not booting, giving an error about the CPU after a recent 'make world'. It turns out that if I boot from the floppy I had of 3.0-06-15-SNAPSHOT and specify sd(0,a)kernel at the boot propmt, then both processors then come on line and the system boots and runs fine. What exactly does this mean? I can provide details of the error message if anyone is interested. Thanks. -- Glenn Johnson Phone: (504) 286-4252 USDA-ARS-SRRC; CFQ FAX: (504) 286-4217 1100 Robert E. Lee email: gjohnson@nola.srrc.usda.gov New Orleans, LA 70124 --------------------------------------------------------------------------- From owner-freebsd-smp Fri Jun 27 15:43:50 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA19481 for smp-outgoing; Fri, 27 Jun 1997 15:43:50 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id PAA19473 for ; Fri, 27 Jun 1997 15:43:44 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id QAA03755 for ; Fri, 27 Jun 1997 16:43:42 -0600 (MDT) Message-Id: <199706272243.QAA03755@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: NMI and debugging in general Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 27 Jun 1997 16:43:42 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, I just committed code to the SMP kernel that allows the BSP to handle NMIs. Till now they were completely ignored. Aps still ignore it, the plan is for the BSP to send an IPI to each of them informing them to stop. I just recently discovered that we were ignoring them. I bought one of those POST code boards from JDR for debugging purposes. It has a 2 digit LED display that shows the BIOS POST codes as the machine boots. kernel code can write an 8 bit value to address 0x80 to display it. This allows a simpler method for displaying the progress of code thru "problem areas". it will be particularily nice for those bugs that run for days before hanging (no need to collect a 100MB of printf output). the design of the card includes a db-15 connector on the bracket for remote mounting the LEDs. I plan to build such a cable, along with a pair of switches for RESET and NMI (finally I get to the point). this will hopefully prove useful for breaking out of those "hung" conditions into the debugger with useful information. with my machine kept in a machine closet this extension box will also allow me to view the display and RESET/NMI the machine without leaving my seat. now if I could just figure out why my stop_cpus()/restart_cpus() IPIs don't work reliably... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Fri Jun 27 15:46:28 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA19588 for smp-outgoing; Fri, 27 Jun 1997 15:46:28 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id PAA19583 for ; Fri, 27 Jun 1997 15:46:23 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id QAA03773; Fri, 27 Jun 1997 16:46:14 -0600 (MDT) Message-Id: <199706272246.QAA03773@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Glenn Johnson cc: freebsd-smp@FreeBSD.ORG Subject: Re: Follow up to problem with Tyan Titan Pro AT In-reply-to: Your message of "Fri, 27 Jun 1997 16:50:35 CDT." <199706272150.QAA00243@Symbion.srrc.usda.gov> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 27 Jun 1997 16:46:14 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > I posted a question earlier about my system with a Tyan Titan Pro AT > motherboard not booting, giving an error about the CPU after a recent 'make > world'. It turns out that if I boot from the floppy I had of > 3.0-06-15-SNAPSHOT and specify sd(0,a)kernel at the boot propmt, then both > processors then come on line and the system boots and runs fine. What exactly > does this mean? sounds like corrupted book blocks to me, but this is not within my area of knowledge... --- > I can provide details of the error message if anyone is > interested. Thanks. always send such info, we can't help without it! -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sat Jun 28 05:51:16 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id FAA18909 for smp-outgoing; Sat, 28 Jun 1997 05:51:16 -0700 (PDT) Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id FAA18904 for ; Sat, 28 Jun 1997 05:51:10 -0700 (PDT) Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.5/8.6.9) id WAA28620; Sat, 28 Jun 1997 22:46:29 +1000 Date: Sat, 28 Jun 1997 22:46:29 +1000 From: Bruce Evans Message-Id: <199706281246.WAA28620@godzilla.zeta.org.au> To: smp@csn.net, smp@FreeBSD.ORG Subject: Re: NMI and debugging in general Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk >the design of the card includes a db-15 connector on the bracket for remote >mounting the LEDs. I plan to build such a cable, along with a pair of >switches for RESET and NMI (finally I get to the point). this will hopefully >prove useful for breaking out of those "hung" conditions into the debugger >with useful information. with my machine kept in a machine closet this This will hopefully prove useful for fixing the uniprocessor handing of NMI :-). Accepting an NMI masks NMIs until the next IRET. Intel docs recommend using an interrupt gate for NMI's in order to disable nested maskable interrupts, since an IRET instruction from the maskable-interrupt handler would re-enable NMI. We are sloppy and use a trap gate for NMI's. trap() isn't designed to work when interrupts are masked (especially at the cpu level). I have no problems using interrupt gates for debugger interrupts, except for the obvious one that it stops everything including clocks while the debugger is active, but this is the point of using interrupt gates. If the system is about to panic, then stopping everything is OK, but reenabling interrupts later for sync() may cause reentrancy problems (this problem is not restricted to NMIs). In all other cases, stopping everything is unacceptable. There are also reentrancy problems, even for calling printf() (printf() is reentrant, but some console drivers aren't). Completely hung systems should rarely occur. sio interrupts rarely hang, so the breakpoint in siointr1() is usually reachable if the BREAK_TO_DEBUGGER option is configured. A similar result can be obtained by removing the keyboard IRQ bit from tty_imask. Then the breakpoint in scintr() is usually reachable. This is fairly safe (perhaps 100% safe) because ISA keyboard interrupts are edge triggered. The SMP lazy masking code makes it safer. In any case, it is safe if you don't touch the keyboard between corrupting tty_imask and attempting to get into the debugger when the system hangs. Bruce