From owner-freebsd-smp Mon Oct 28 16:42:16 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA21273 for smp-outgoing; Mon, 28 Oct 1996 16:42:16 -0800 (PST) Received: from soda.CSUA.Berkeley.EDU (soda.CSUA.Berkeley.EDU [128.32.43.52]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id QAA21265 for ; Mon, 28 Oct 1996 16:42:13 -0800 (PST) Received: from localhost (richardc@localhost) by soda.CSUA.Berkeley.EDU (8.6.12/8.6.12) with SMTP id QAA12573 for ; Mon, 28 Oct 1996 16:43:18 -0800 Date: Mon, 28 Oct 1996 16:43:16 -0800 (PST) From: Veggy Vinny To: smp@FreeBSD.ORG Subject: Earth.GaiaNet.NET Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi everyone, After doing a make world on earth.GAIANET.NET last Wednesday night and then rebuilt a new SMP kernel and this is what happens: When earth comes up and detects the hardware, it pauses at: WARNING: mapping PCI INT #10 If I hit enter, it continues until it gets to sio1: type 16550A and freezes. Anyone have any clues? Could this be the new commit for the COMPAT_LINUX to make the SMP kernel compile? Vince GaiaNet Corporation - Unix Networking Operations - GUS Mailing Lists Admin From owner-freebsd-smp Mon Oct 28 16:49:49 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA21776 for smp-outgoing; Mon, 28 Oct 1996 16:49:49 -0800 (PST) Received: from ormail.intel.com (ormail.intel.com [134.134.248.3]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id QAA21767; Mon, 28 Oct 1996 16:49:44 -0800 (PST) Received: from ichips.intel.com (ichips.intel.com [134.134.50.200]) by ormail.intel.com (8.7.6/8.7.3) with ESMTP id QAA14731; Mon, 28 Oct 1996 16:49:39 -0800 (PST) Received: from pdxlx008.intel.com by ichips.intel.com (8.7.4/jIII) id QAA12123; Mon, 28 Oct 1996 16:48:24 -0800 (PST) Received: from pdxlx008.intel.com (localhost.jf.intel.com [127.0.0.1]) by pdxlx008.intel.com (8.7.5/8.7.3) with ESMTP id QAA05186; Mon, 28 Oct 1996 16:49:37 -0800 (PST) Message-Id: <199610290049.QAA05186@pdxlx008.intel.com> To: asami@freebsd.org (Satoshi Asami) Cc: rgrimes@GndRsh.aac.dev.com, ejs@bfd.com, michaelv@MindBender.serv.net, scrappy@ki.net, smp@freebsd.org Subject: Re: Recommendations... In-reply-to: Your message of "Fri, 25 Oct 1996 15:55:08 PDT." <199610252255.PAA29409@sunrise.cs.berkeley.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Date: Mon, 28 Oct 1996 16:49:37 -0800 From: Wayne Scott Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > move 70-80MB/s (depending on EDO or non-EDO), but I can't get more > than 45MB/s out of a P6-200 with Natoma/server (at least that's what > Intel told us). Be careful. Intel told you nothing. I told you what my machine was able to get. I have never run your benchmarks on a Natoma system. My numbers showed a 70+ MB/s using REP MOVs on an Orion motherboard with 8 SIMMs. -Wayne From owner-freebsd-smp Mon Oct 28 17:06:41 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id RAA28602 for smp-outgoing; Mon, 28 Oct 1996 17:06:41 -0800 (PST) Received: from horst.bfd.com (horst.bfd.com [204.160.242.10]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id RAA28558 for ; Mon, 28 Oct 1996 17:06:39 -0800 (PST) Received: from harlie (bastion.bfd.com [204.160.242.2]) by horst.bfd.com (8.7.6/8.7.3) with SMTP id RAA10132; Mon, 28 Oct 1996 17:05:08 -0800 (PST) Date: Mon, 28 Oct 1996 17:05:08 -0800 (PST) From: "Eric J. Schwertfeger" X-Sender: ejs@harlie To: Wayne Scott cc: smp@freebsd.org Subject: Re: Recommendations... In-Reply-To: <199610290049.QAA05186@pdxlx008.intel.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 Mon, 28 Oct 1996, Wayne Scott wrote: > Be careful. > > Intel told you nothing. I told you what my machine was able to get. > I have never run your benchmarks on a Natoma system. > > My numbers showed a 70+ MB/s using REP MOVs on an Orion motherboard > with 8 SIMMs. The server version of orion will interleave 4 banks of memory, and benchmarks in the same ballpark as a good 430HX chipset with a single bank of EDO RAM. (for memory speed only). It should have a strong advantage, but doesn't show it. It's the Natoma, and Orion with fewer banks that are in question. Why do the P6 chipsets have lower memory throughput when on equal footing with a good P5 chipset? From owner-freebsd-smp Wed Oct 30 01:28:25 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id BAA21188 for smp-outgoing; Wed, 30 Oct 1996 01:28:25 -0800 (PST) Received: from soda.CSUA.Berkeley.EDU (soda.CSUA.Berkeley.EDU [128.32.43.52]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id BAA21173 for ; Wed, 30 Oct 1996 01:28:20 -0800 (PST) Received: from localhost (richardc@localhost) by soda.CSUA.Berkeley.EDU (8.6.12/8.6.12) with SMTP id BAA01838 for ; Wed, 30 Oct 1996 01:29:22 -0800 Date: Wed, 30 Oct 1996 01:29:21 -0800 (PST) From: Veggy Vinny To: smp@FreeBSD.ORG Subject: SMP Kernel problems Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk It seems as of October 24, 1996's -current , compiling a SMP kernel with CONFIG_LINUX now works but there seems to be a problem which is experience with the SMP kernel and not the -current kernel: When the machine comes up and detects the hardware, it pauses at: WARNING: mapping PCI INT #10 If I hit enter, it continues until it gets to sio1: type 16550A and freezes. I rebooted 3 times and the same thing happens... Any ideas? Vince GaiaNet Corporation - Unix Networking Operations - GUS Mailing Lists Admin From owner-freebsd-smp Wed Oct 30 09:55:52 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id JAA16510 for smp-outgoing; Wed, 30 Oct 1996 09:55:52 -0800 (PST) Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id JAA16499 for ; Wed, 30 Oct 1996 09:55:46 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.12/8.6.12) with SMTP id KAA18233; Wed, 30 Oct 1996 10:55:36 -0700 Message-Id: <199610301755.KAA18233@clem.systemsix.com> X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.5 12/11/95 From: Steve Passe To: Veggy Vinny cc: smp@FreeBSD.ORG Subject: Re: SMP Kernel problems In-reply-to: Your message of "Wed, 30 Oct 1996 01:29:21 PST." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 30 Oct 1996 10:55:36 -0700 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > It seems as of October 24, 1996's -current , compiling a SMP > kernel with CONFIG_LINUX now works but there seems to be a problem which > is experience with the SMP kernel and not the -current kernel: > > When the machine comes up and detects the hardware, it pauses at: > > WARNING: mapping PCI INT #10 > > If I hit enter, it continues until it gets to > > sio1: type 16550A > > and freezes. My machine is dead righrt now so I can't check the sources to see where this message is coming from. It might be from my hacks to pci.c, or possibly Stefan has started to add the PCI support I requested (although I haven't seen a merge of -current to the SMP tree so this is doubtful) I should have my machine running by end of day. What hardware is accessed immediately prior to this message?; By any chance, did you enable SMP_SYMIOXXX (this is not ready yet)? -- Steve Passe | powered by smp@csn.net | FreeBSD From owner-freebsd-smp Wed Oct 30 12:47:42 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id MAA28038 for smp-outgoing; Wed, 30 Oct 1996 12:47:42 -0800 (PST) Received: from soda.CSUA.Berkeley.EDU (soda.CSUA.Berkeley.EDU [128.32.43.52]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id MAA28033 for ; Wed, 30 Oct 1996 12:47:40 -0800 (PST) Received: from localhost (richardc@localhost) by soda.CSUA.Berkeley.EDU (8.6.12/8.6.12) with SMTP id MAA29644; Wed, 30 Oct 1996 12:48:43 -0800 Date: Wed, 30 Oct 1996 12:48:41 -0800 (PST) From: Veggy Vinny To: Steve Passe cc: smp@FreeBSD.ORG Subject: Re: SMP Kernel problems In-Reply-To: <199610301755.KAA18233@clem.systemsix.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, 30 Oct 1996, Steve Passe wrote: Hi Steve and everyone, > > It seems as of October 24, 1996's -current , compiling a SMP > > kernel with CONFIG_LINUX now works but there seems to be a problem which > > is experience with the SMP kernel and not the -current kernel: > > > > When the machine comes up and detects the hardware, it pauses at: > > > > WARNING: mapping PCI INT #10 > > > > If I hit enter, it continues until it gets to > > > > sio1: type 16550A > > > > and freezes. > > My machine is dead righrt now so I can't check the sources to see where > this message is coming from. It might be from my hacks to pci.c, or > possibly Stefan has started to add the PCI support I requested (although > I haven't seen a merge of -current to the SMP tree so this is doubtful) > I should have my machine running by end of day. > > What hardware is accessed immediately prior to this message?; Don't remember since my friend is the one who is in front of the machines 600 miles away =) and that's the only info he gave... I didn't see anything committed to the list except that patch for COMPAT_LINUX though. > By any chance, did you enable SMP_SYMIOXXX (this is not ready yet)? Nope.... If it was in the kernel config file, I didn't... Vince GaiaNet Corporation - Unix Networking Operations - GUS Mailing Lists Admin From owner-freebsd-smp Wed Oct 30 13:25:41 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id NAA01117 for smp-outgoing; Wed, 30 Oct 1996 13:25:41 -0800 (PST) Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id NAA01106 for ; Wed, 30 Oct 1996 13:25:35 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.12/8.6.12) with SMTP id OAA00873; Wed, 30 Oct 1996 14:25:13 -0700 Message-Id: <199610302125.OAA00873@clem.systemsix.com> X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.5 12/11/95 From: Steve Passe To: Veggy Vinny cc: smp@FreeBSD.ORG Subject: Re: SMP Kernel problems In-reply-to: Your message of "Wed, 30 Oct 1996 12:48:41 PST." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 30 Oct 1996 14:25:13 -0700 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > > On Wed, 30 Oct 1996, Steve Passe wrote: > > Hi Steve and everyone, > > > > It seems as of October 24, 1996's -current , compiling a SMP > > > kernel with CONFIG_LINUX now works but there seems to be a problem which > > > is experience with the SMP kernel and not the -current kernel: > > > > > > When the machine comes up and detects the hardware, it pauses at: > > > > > > WARNING: mapping PCI INT #10 > > > > > > If I hit enter, it continues until it gets to > > > > > > sio1: type 16550A > > > > > > and freezes. > > > > My machine is dead righrt now so I can't check the sources to see where > > this message is coming from. It might be from my hacks to pci.c, or > > possibly Stefan has started to add the PCI support I requested (although > > I haven't seen a merge of -current to the SMP tree so this is doubtful) > > I should have my machine running by end of day. > > > > What hardware is accessed immediately prior to this message?; > > Don't remember since my friend is the one who is in front of the > machines 600 miles away =) and that's the only info he gave... I didn't > see anything committed to the list except that patch for COMPAT_LINUX > though. > > > By any chance, did you enable SMP_SYMIOXXX (this is not ready yet)? > > Nope.... If it was in the kernel config file, I didn't... here's the offending line in pci/pci.c: --- int pci_map_int (pcici_t tag, pci_inthand_t *func, void *arg, unsigned *maskptr) { ... #if defined( SMP ) && defined( SMP_SYMIOXXX ) if (irq!=19) {printf("WARNING: mapping PCI INT #%d\n",irq);cngetc();} #endif /* SMP && SMP_SYMIOXXX */ ... --- SMP_SYMIOXXX is clearly turned on, probably in i386/include/apic.h This won't work for anyone but me yet, turn it off and this problem should go away. I am about to commit new code where all the APIC/symmetric IO code is conditional on "APIC_IO", ie I am going to fold "SMP_SYMIO" & "SMP_SYMIOXXX" into 1 define and start to request that people turn it on for testing. This will happen as soon as I get 1 outstanding problem with PCI resolved. The warning you saw is associated with the bandaid that makes it work for my particualr hardware. -- Steve Passe | powered by smp@csn.net | FreeBSD From owner-freebsd-smp Wed Oct 30 13:37:28 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id NAA02269 for smp-outgoing; Wed, 30 Oct 1996 13:37:28 -0800 (PST) Received: from soda.CSUA.Berkeley.EDU (soda.CSUA.Berkeley.EDU [128.32.43.52]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id NAA02263 for ; Wed, 30 Oct 1996 13:37:24 -0800 (PST) Received: from localhost (richardc@localhost) by soda.CSUA.Berkeley.EDU (8.6.12/8.6.12) with SMTP id NAA03415; Wed, 30 Oct 1996 13:33:38 -0800 Date: Wed, 30 Oct 1996 13:33:36 -0800 (PST) From: Veggy Vinny To: Steve Passe cc: smp@FreeBSD.ORG Subject: Re: SMP Kernel problems In-Reply-To: <199610302125.OAA00873@clem.systemsix.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, 30 Oct 1996, Steve Passe wrote: Hi Steve and everyone, > > On Wed, 30 Oct 1996, Steve Passe wrote: > > > > > > It seems as of October 24, 1996's -current , compiling a SMP > > > > kernel with CONFIG_LINUX now works but there seems to be a problem which > > > > is experience with the SMP kernel and not the -current kernel: > > > > > > > > When the machine comes up and detects the hardware, it pauses at: > > > > > > > > WARNING: mapping PCI INT #10 > > > > > > > > If I hit enter, it continues until it gets to > > > > > > > > sio1: type 16550A > > > > > > > > and freezes. > > > > > > My machine is dead righrt now so I can't check the sources to see where > > > this message is coming from. It might be from my hacks to pci.c, or > > > possibly Stefan has started to add the PCI support I requested (although > > > I haven't seen a merge of -current to the SMP tree so this is doubtful) > > > I should have my machine running by end of day. > > > > > > What hardware is accessed immediately prior to this message?; > > > > Don't remember since my friend is the one who is in front of the > > machines 600 miles away =) and that's the only info he gave... I didn't > > see anything committed to the list except that patch for COMPAT_LINUX > > though. > > > > > By any chance, did you enable SMP_SYMIOXXX (this is not ready yet)? > > > > Nope.... If it was in the kernel config file, I didn't... > > here's the offending line in pci/pci.c: > > --- > int pci_map_int (pcici_t tag, pci_inthand_t *func, void *arg, unsigned > *maskptr) > { > ... > #if defined( SMP ) && defined( SMP_SYMIOXXX ) > if (irq!=19) {printf("WARNING: mapping PCI INT #%d\n",irq);cngetc();} > #endif /* SMP && SMP_SYMIOXXX */ > ... Oh okay... > --- > > SMP_SYMIOXXX is clearly turned on, probably in i386/include/apic.h This won't > work for anyone but me yet, turn it off and this problem should go away. > I am about to commit new code where all the APIC/symmetric IO code is > conditional on "APIC_IO", ie I am going to fold "SMP_SYMIO" & "SMP_SYMIOXXX" > into 1 define and start to request that people turn it on for testing. This > will happen as soon as I get 1 outstanding problem with PCI resolved. > The warning you saw is associated with the bandaid that makes it work for > my particualr hardware. Oh okay so after your commit, would it be turned off? I guess I can just wait until the next merge of -current =) Vince GaiaNet Corporation - Unix Networking Operations - GUS Mailing Lists Admin From owner-freebsd-smp Wed Oct 30 13:46:44 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id NAA02916 for smp-outgoing; Wed, 30 Oct 1996 13:46:44 -0800 (PST) Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id NAA02909 for ; Wed, 30 Oct 1996 13:46:38 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.12/8.6.12) with SMTP id OAA01009; Wed, 30 Oct 1996 14:45:55 -0700 Message-Id: <199610302145.OAA01009@clem.systemsix.com> X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.5 12/11/95 From: Steve Passe To: Veggy Vinny cc: smp@FreeBSD.ORG Subject: Re: SMP Kernel problems In-reply-to: Your message of "Wed, 30 Oct 1996 13:33:36 PST." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 30 Oct 1996 14:45:55 -0700 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > Oh okay so after your commit, would it be turned off? I guess I >can just wait until the next merge of -current =) SMP_SYMIOXXX is turned OFF in the current version of apic.h, ie: #define SMP_SYMIOXXX_NOT If this is not true let me know and I will go to freefall and fix it. It shouldn't have anything to do with -current. I'm guessing your friend edited apic.h to turn it on. I suspect the new APIC_IO will also be OFF by default for some time, there are A LOT of issues to be ironed out b4 it will work on some of the more 'unique' hardware out there... -- Steve Passe | powered by smp@csn.net | FreeBSD -----BEGIN PGP PUBLIC KEY BLOCK----- Version: 2.6.2 mQCNAzHe7tEAAAEEAM274wAEEdP+grIrV6UtBt54FB5ufifFRA5ujzflrvlF8aoE 04it5BsUPFi3jJLfvOQeydbegexspPXL6kUejYt2OeptHuroIVW5+y2M2naTwqtX WVGeBP6s2q/fPPAS+g+sNZCpVBTbuinKa/C4Q6HJ++M9AyzIq5EuvO0a8Rr9AAUR tBlTdGV2ZSBQYXNzZSA8c21wQGNzbi5uZXQ+ =ds99 -----END PGP PUBLIC KEY BLOCK----- From owner-freebsd-smp Wed Oct 30 14:00:47 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id OAA03684 for smp-outgoing; Wed, 30 Oct 1996 14:00:47 -0800 (PST) Received: (from peter@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id OAA03676 for freebsd-smp; Wed, 30 Oct 1996 14:00:44 -0800 (PST) Date: Wed, 30 Oct 1996 14:00:44 -0800 (PST) From: Peter Wemm Message-Id: <199610302200.OAA03676@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/i386/include apic.h Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk peter 96/10/30 14:00:42 Modified: i386/include apic.h Log: aargh! oops! It seems I accidently turned on Steve's experimental code here last time. Revision Changes Path 1.11 +1 -1 sys/i386/include/apic.h From owner-freebsd-smp Wed Oct 30 14:14:01 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id OAA04415 for smp-outgoing; Wed, 30 Oct 1996 14:14:01 -0800 (PST) Received: from soda.CSUA.Berkeley.EDU (soda.CSUA.Berkeley.EDU [128.32.43.52]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id OAA04409 for ; Wed, 30 Oct 1996 14:13:57 -0800 (PST) Received: from localhost (richardc@localhost) by soda.CSUA.Berkeley.EDU (8.6.12/8.6.12) with SMTP id OAA06806; Wed, 30 Oct 1996 14:15:03 -0800 Date: Wed, 30 Oct 1996 14:15:01 -0800 (PST) From: Veggy Vinny To: Steve Passe cc: smp@FreeBSD.ORG Subject: Re: SMP Kernel problems In-Reply-To: <199610302145.OAA01009@clem.systemsix.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, 30 Oct 1996, Steve Passe wrote: Hi Steve and everyone, > > Oh okay so after your commit, would it be turned off? I guess I > >can just wait until the next merge of -current =) > > SMP_SYMIOXXX is turned OFF in the current version of apic.h, ie: > > #define SMP_SYMIOXXX_NOT > > If this is not true let me know and I will go to freefall and fix it. It > shouldn't have anything to do with -current. I'm guessing your friend > edited apic.h to turn it on. I suspect the new APIC_IO will also be OFF by > default for some time, there are A LOT of issues to be ironed out b4 it > will work on some of the more 'unique' hardware out there... Mines said SMP_SYMIOXXX so it's turned on.... It's not my friend since all he does is turn the machine on and off ;) I'm the one who runs the machine remotely and keep the machine working in a happy form... I just saw Peter commit the fix since he accidentally turned it on... Vince GaiaNet Corporation - Unix Networking Operations - GUS Mailing Lists Admin From owner-freebsd-smp Thu Oct 31 16:52:48 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA20381 for smp-outgoing; Thu, 31 Oct 1996 16:52:48 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA20374 for freebsd-smp; Thu, 31 Oct 1996 16:52:46 -0800 (PST) Date: Thu, 31 Oct 1996 16:52:46 -0800 (PST) From: Steve Passe Message-Id: <199611010052.QAA20374@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/kern init_main.c Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/10/31 16:52:45 Modified: kern init_main.c Log: official ALPHA support for IO APICs 'SMP_SYMIO' & 'SMP_SYMIOXXX' merged into APIC_IO 'APIC_IO' OFF by default, enable in i386/include/apic.h ISA works. PCI works. EISA does NOT work. Revision Changes Path 1.28 +20 -1 sys/kern/init_main.c From owner-freebsd-smp Thu Oct 31 16:54:50 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA20447 for smp-outgoing; Thu, 31 Oct 1996 16:54:50 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA20440 for freebsd-smp; Thu, 31 Oct 1996 16:54:48 -0800 (PST) Date: Thu, 31 Oct 1996 16:54:48 -0800 (PST) From: Steve Passe Message-Id: <199611010054.QAA20440@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/pci pci.c pcibus.h pcireg.h Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/10/31 16:54:47 Modified: pci pci.c pcibus.h pcireg.h Log: official ALPHA support for IO APICs 'SMP_SYMIO' & 'SMP_SYMIOXXX' merged into APIC_IO 'APIC_IO' OFF by default, enable in i386/include/apic.h ISA works. PCI works. EISA does NOT work. Revision Changes Path 1.3 +12 -49 sys/pci/pci.c 1.3 +3 -3 sys/pci/pcibus.h 1.2 +2 -2 sys/pci/pcireg.h From owner-freebsd-smp Thu Oct 31 16:57:48 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA20809 for smp-outgoing; Thu, 31 Oct 1996 16:57:48 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA20800 for freebsd-smp; Thu, 31 Oct 1996 16:57:47 -0800 (PST) Date: Thu, 31 Oct 1996 16:57:47 -0800 (PST) From: Steve Passe Message-Id: <199611010057.QAA20800@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/i386/isa clock.c icu.h icu.s if_ze.c isa.c isa_device.h npx.c pcibus.c sio.c vector.s sys/i386/i386 autoconf.c microtime.s mp_machdep.c mpapic.c mplock.s pmap.c sys/i386/include apic.h ipl.h smp.h Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/10/31 16:57:46 Modified: i386/i386 autoconf.c microtime.s mp_machdep.c mpapic.c mplock.s pmap.c i386/include apic.h ipl.h smp.h i386/isa clock.c icu.h icu.s if_ze.c isa.c isa_device.h npx.c pcibus.c sio.c vector.s Log: official ALPHA support for IO APICs 'SMP_SYMIO' & 'SMP_SYMIOXXX' merged into APIC_IO 'APIC_IO' OFF by default, enable in i386/include/apic.h ISA works. PCI works. EISA does NOT work. Revision Changes Path 1.4 +7 -12 sys/i386/i386/autoconf.c 1.12 +12 -12 sys/i386/i386/microtime.s 1.8 +41 -14 sys/i386/i386/mp_machdep.c 1.9 +67 -20 sys/i386/i386/mpapic.c 1.13 +10 -1 sys/i386/i386/mplock.s 1.24 +14 -11 sys/i386/i386/pmap.c 1.12 +40 -6 sys/i386/include/apic.h 1.3 +3 -3 sys/i386/include/ipl.h 1.14 +52 -54 sys/i386/include/smp.h 1.7 +11 -11 sys/i386/isa/clock.c 1.5 +5 -5 sys/i386/isa/icu.h 1.12 +16 -11 sys/i386/isa/icu.s 1.4 +13 -8 sys/i386/isa/if_ze.c 1.6 +32 -14 sys/i386/isa/isa.c 1.3 +7 -7 sys/i386/isa/isa_device.h 1.9 +19 -19 sys/i386/isa/npx.c 1.3 +1 -5 sys/i386/isa/pcibus.c 1.8 +9 -15 sys/i386/isa/sio.c 1.16 +22 -22 sys/i386/isa/vector.s From owner-freebsd-smp Thu Oct 31 17:59:35 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id RAA28238 for smp-outgoing; Thu, 31 Oct 1996 17:59:35 -0800 (PST) Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id RAA28216 for ; Thu, 31 Oct 1996 17:59:29 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.12/8.6.12) with SMTP id SAA08923 for ; Thu, 31 Oct 1996 18:59:26 -0700 Message-Id: <199611010159.SAA08923@clem.systemsix.com> X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.5 12/11/95 From: Steve Passe To: freebsd-smp@freefall.freebsd.org Subject: IO APIC code ready for testers. Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 31 Oct 1996 18:59:26 -0700 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, The ALPHA level version of the I/O APIC code is ready for testers. By default it is turned OFF. I would appreciate it if you could enable it and test on your hardware. The first problems to be reported will get the first attention. For details see: http://www.freebsd.org/~fsmp/SMP/apicio.html -- Steve Passe | powered by smp@csn.net | FreeBSD From owner-freebsd-smp Fri Nov 1 00:27:56 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id AAA07646 for smp-outgoing; Fri, 1 Nov 1996 00:27:56 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id AAA07635 for freebsd-smp; Fri, 1 Nov 1996 00:27:54 -0800 (PST) Date: Fri, 1 Nov 1996 00:27:54 -0800 (PST) From: Steve Passe Message-Id: <199611010827.AAA07635@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/i386/i386 mpapic.c mpboot.s Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/11/01 00:27:54 Modified: i386/i386 mpapic.c mpboot.s Log: fixed bug preventing APs from handling INTs in LOW PRIORITY mode. Revision Changes Path 1.10 +4 -4 sys/i386/i386/mpapic.c 1.8 +7 -0 sys/i386/i386/mpboot.s From owner-freebsd-smp Fri Nov 1 21:50:57 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id VAA02949 for smp-outgoing; Fri, 1 Nov 1996 21:50:57 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id VAA02930 for freebsd-smp; Fri, 1 Nov 1996 21:50:51 -0800 (PST) Date: Fri, 1 Nov 1996 21:50:51 -0800 (PST) From: Steve Passe Message-Id: <199611020550.VAA02930@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/kern init_main.c Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/11/01 21:50:47 Modified: kern init_main.c Log: general cleanup. previously merged SMP_SYMIO/SMP_SYMIOXXX into APIC_IO, now moving APIC_IO from i386/include/apic.h to i386/conf/options.i386, ie, it is now a config file option. removed "include from files only using it for SMP_SYMIO/SMP_SYMIOXXX/APIC_IO defines. created i386/include/smptests.h as placeholder for test defines, etc. moved all test defines from apic.h to smptests.h. removed "include from files only using it for test defines. further logical separation of SMP and APIC_IO. final (for now) cleanup of PCI/APIC hooks. Revision Changes Path 1.29 +5 -1 sys/kern/init_main.c From owner-freebsd-smp Fri Nov 1 21:49:13 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id VAA02717 for smp-outgoing; Fri, 1 Nov 1996 21:49:13 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id VAA02702 for freebsd-smp; Fri, 1 Nov 1996 21:49:08 -0800 (PST) Date: Fri, 1 Nov 1996 21:49:08 -0800 (PST) From: Steve Passe Message-Id: <199611020549.VAA02702@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/i386/isa clock.c icu.h icu.s if_ze.c isa.c isa_device.h npx.c pcibus.c sio.c vector.s sys/i386/conf options.i386 sys/i386/i386 autoconf.c microtime.s mp_machdep.c mpapic.c mpboot.s mplock.s pmap.c sys/i386/include smptests.h apic.h ipl.h smp.h smpasm.h Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/11/01 21:49:06 Modified: i386/conf options.i386 i386/i386 autoconf.c microtime.s mp_machdep.c mpapic.c mpboot.s mplock.s pmap.c i386/include apic.h ipl.h smp.h smpasm.h i386/isa clock.c icu.h icu.s if_ze.c isa.c isa_device.h npx.c pcibus.c sio.c vector.s Added: i386/include smptests.h Log: general cleanup. previously merged SMP_SYMIO/SMP_SYMIOXXX into APIC_IO, now moving APIC_IO from i386/include/apic.h to i386/conf/options.i386, ie, it is now a config file option. removed "include from files only using it for SMP_SYMIO/SMP_SYMIOXXX/APIC_IO defines. created i386/include/smptests.h as placeholder for test defines, etc. moved all test defines from apic.h to smptests.h. removed "include from files only using it for test defines. further logical separation of SMP and APIC_IO. final (for now) cleanup of PCI/APIC hooks. Revision Changes Path 1.8 +3 -1 sys/i386/conf/options.i386 1.5 +3 -3 sys/i386/i386/autoconf.c 1.13 +4 -4 sys/i386/i386/microtime.s 1.9 +5 -2 sys/i386/i386/mp_machdep.c 1.11 +5 -5 sys/i386/i386/mpapic.c 1.9 +2 -4 sys/i386/i386/mpboot.s 1.14 +2 -1 sys/i386/i386/mplock.s 1.25 +7 -9 sys/i386/i386/pmap.c 1.13 +84 -97 sys/i386/include/apic.h 1.4 +1 -4 sys/i386/include/ipl.h 1.15 +4 -4 sys/i386/include/smp.h 1.7 +6 -5 sys/i386/include/smpasm.h 1.8 +1 -4 sys/i386/isa/clock.c 1.6 +13 -16 sys/i386/isa/icu.h 1.13 +5 -3 sys/i386/isa/icu.s 1.5 +4 -5 sys/i386/isa/if_ze.c 1.7 +3 -3 sys/i386/isa/isa.c 1.4 +1 -4 sys/i386/isa/isa_device.h 1.10 +4 -4 sys/i386/isa/npx.c 1.4 +1 -7 sys/i386/isa/pcibus.c 1.9 +1 -4 sys/i386/isa/sio.c 1.17 +18 -1 sys/i386/isa/vector.s From owner-freebsd-smp Fri Nov 1 21:51:42 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id VAA03080 for smp-outgoing; Fri, 1 Nov 1996 21:51:42 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id VAA03064 for freebsd-smp; Fri, 1 Nov 1996 21:51:37 -0800 (PST) Date: Fri, 1 Nov 1996 21:51:37 -0800 (PST) From: Steve Passe Message-Id: <199611020551.VAA03064@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/pci pci.c pcibus.h pcireg.h Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/11/01 21:51:36 Modified: pci pci.c pcibus.h pcireg.h Log: general cleanup. previously merged SMP_SYMIO/SMP_SYMIOXXX into APIC_IO, now moving APIC_IO from i386/include/apic.h to i386/conf/options.i386, ie, it is now a config file option. removed "include from files only using it for SMP_SYMIO/SMP_SYMIOXXX/APIC_IO defines. created i386/include/smptests.h as placeholder for test defines, etc. moved all test defines from apic.h to smptests.h. removed "include from files only using it for test defines. further logical separation of SMP and APIC_IO. final (for now) cleanup of PCI/APIC hooks. Revision Changes Path 1.4 +25 -18 sys/pci/pci.c 1.4 +1 -5 sys/pci/pcibus.h 1.3 +2 -2 sys/pci/pcireg.h From owner-freebsd-smp Fri Nov 1 22:19:03 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id WAA07272 for smp-outgoing; Fri, 1 Nov 1996 22:19:03 -0800 (PST) Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id WAA07252 for ; Fri, 1 Nov 1996 22:18:57 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.12/8.6.12) with SMTP id XAA16485 for ; Fri, 1 Nov 1996 23:18:50 -0700 Message-Id: <199611020618.XAA16485@clem.systemsix.com> X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.5 12/11/95 From: Steve Passe To: freebsd-smp@freefall.freebsd.org Subject: APIC testers Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 01 Nov 1996 23:18:49 -0700 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, I just commited new APIC code with cleanup and minor bug fixes. This code now uses "options APIC_IO" in the kernel config file instead of a define in i386/include/apic.h. For details see: http://www.freebsd.org/~fsmp/SMP/apicio.html So far I have received NO reports of anyone testing this code. Bite the bullet gang, this has to be done sooner or later, sooner if we expect SMP to get into the main tree... -- Steve Passe | powered by smp@csn.net | FreeBSD From owner-freebsd-smp Sat Nov 2 13:33:27 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id NAA25392 for smp-outgoing; Sat, 2 Nov 1996 13:33:27 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id NAA25383 for freebsd-smp; Sat, 2 Nov 1996 13:33:23 -0800 (PST) Date: Sat, 2 Nov 1996 13:33:23 -0800 (PST) From: Steve Passe Message-Id: <199611022133.NAA25383@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/i386/i386 autoconf.c mp_machdep.c sys/i386/include smp.h Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/11/02 13:33:22 Modified: i386/i386 autoconf.c mp_machdep.c i386/include smp.h Log: added support for default MP configurations. Note: I have no way to test this code. added support for motherboards that start in "PIC" mode. Note: I have no way to test this code. Revision Changes Path 1.6 +6 -4 sys/i386/i386/autoconf.c 1.10 +62 -14 sys/i386/i386/mp_machdep.c 1.16 +3 -2 sys/i386/include/smp.h From owner-freebsd-smp Sat Nov 2 15:05:58 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id PAA01964 for smp-outgoing; Sat, 2 Nov 1996 15:05:58 -0800 (PST) Received: (from fsmp@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id PAA01957 for freebsd-smp; Sat, 2 Nov 1996 15:05:56 -0800 (PST) Date: Sat, 2 Nov 1996 15:05:56 -0800 (PST) From: Steve Passe Message-Id: <199611022305.PAA01957@freefall.freebsd.org> To: freebsd-smp Subject: cvs commit: sys/i386/i386 mp_machdep.c mpapic.c Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk fsmp 96/11/02 15:05:55 Modified: i386/i386 mp_machdep.c mpapic.c Log: broke all panic() calls into printf()/panic() calls. panic() evidently freezes some machines when called early in boot process. preceeding printf() hopefully gets the reason to the screen b4 the freeze. Revision Changes Path 1.11 +23 -11 sys/i386/i386/mp_machdep.c 1.12 +9 -5 sys/i386/i386/mpapic.c From owner-freebsd-smp Sat Nov 2 15:20:06 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id PAA02672 for smp-outgoing; Sat, 2 Nov 1996 15:20:06 -0800 (PST) Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id PAA02667 for ; Sat, 2 Nov 1996 15:20:03 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.12/8.6.12) with SMTP id QAA23170 for ; Sat, 2 Nov 1996 16:19:59 -0700 Message-Id: <199611022319.QAA23170@clem.systemsix.com> X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.5 12/11/95 From: Steve Passe To: freebsd-smp@freefall.freebsd.org Subject: options.i386 Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sat, 02 Nov 1996 16:19:59 -0700 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, I just helped a user get past the problem of having 3 busses on their SMP motherboard (needed to bump "NBUS" from 2 to 3). I propose placing the following into options.i386: NCPU opt_smp.h # number of CPUs NBUS opt_smp.h # number of busses NAPIC opt_smp.h # number of IO APICs Comments? -- Steve Passe | powered by smp@csn.net | FreeBSD From owner-freebsd-smp Sat Nov 2 22:44:12 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id WAA00284 for smp-outgoing; Sat, 2 Nov 1996 22:44:12 -0800 (PST) Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id WAA00273; Sat, 2 Nov 1996 22:44:05 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.12/8.6.12) with SMTP id XAA25052; Sat, 2 Nov 1996 23:43:56 -0700 Message-Id: <199611030643.XAA25052@clem.systemsix.com> X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.5 12/11/95 From: Steve Passe To: dg@Root.COM cc: hackers@freefall.freebsd.org, smp@freefall.freebsd.org Subject: Re: ed0 timeouts In-reply-to: Your message of "Fri, 25 Oct 1996 14:10:45 PDT." <199610252110.OAA08906@root.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sat, 02 Nov 1996 23:43:56 -0700 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, >>I currently am writing code for the FreeBSD SMP kernel that will allow >> ... >>I am occasionally loosing the INTs expected to be generated by ed_xmit(). >>The ed_watchdog() routine usually recovers gracefully (network actually locked >>twice since running this code: approx 2/3 weeks). Could someone explain >>why the ed0 cards might loose this INT? Any clues/insight appreciated! > ... It sounds much more >like a problem acking ISA interrupts, or perhaps a bug in the handling of >the interrupt masks. ... Question: Is it possible for a new INT to be asserted by the if_ed driver WHILE it is currently being serviced by the edintr() routine? What I have discovered is that unlike the 8259, the IO APIC will ignore (ie NOT delivered or held pending) an edge level INT if it currently is masked. The routine in vector.s masks the INT, calls edintr(), then after edintr() returns it unmasks the INT. If another INT fired as a result of ed_start() being called in edintr() BEFORE the INT was unmasked it would be LOST. What about a scenerio where an async receive packet occured while servicing a TXMIT INT: The received packet causes a bus level edge INT to be asserted, which is lost, as it is masked by edintr() (for the XMIT INT). Now edintr() finishes, unmasking the INT, and eventually another XMIT INT occurs. BUT the board doesn't assert another edge because it already has the RCV INT "active", which was lost, but the hardware assumes it is pending. Does this sound feasable? -- Steve Passe | powered by smp@csn.net | FreeBSD -----BEGIN PGP PUBLIC KEY BLOCK----- Version: 2.6.2 mQCNAzHe7tEAAAEEAM274wAEEdP+grIrV6UtBt54FB5ufifFRA5ujzflrvlF8aoE 04it5BsUPFi3jJLfvOQeydbegexspPXL6kUejYt2OeptHuroIVW5+y2M2naTwqtX WVGeBP6s2q/fPPAS+g+sNZCpVBTbuinKa/C4Q6HJ++M9AyzIq5EuvO0a8Rr9AAUR tBlTdGV2ZSBQYXNzZSA8c21wQGNzbi5uZXQ+ =ds99 -----END PGP PUBLIC KEY BLOCK-----