From owner-freebsd-smp Sun Jul 20 09:04:36 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA21611 for smp-outgoing; Sun, 20 Jul 1997 09:04:36 -0700 (PDT) Received: from genghis.eng.demon.net (genghis.eng.demon.net [193.195.45.10]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id JAA21606 for ; Sun, 20 Jul 1997 09:04:31 -0700 (PDT) Received: from genghis.eng.demon.net [193.195.45.10] by genghis.eng.demon.net with esmtp (Exim 1.62 #1) id 0wpyT7-00009N-00; Sun, 20 Jul 1997 17:04:21 +0100 To: Steve Passe cc: smp@freebsd.org Subject: Re: HEADS UP: major timer/APIC code changes. Organization: Demon Internet Ltd. Reply-To: ade@demon.net In-reply-to: Your message of "Sat, 19 Jul 1997 13:45:35 MDT." <199707191945.NAA26940@Ilsa.StevesCafe.com> Date: Sun, 20 Jul 1997 17:04:21 +0100 From: Ade Lovett Message-Id: Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Steve Passe writes: > >looks like I need to fall back and punt... For now, anyone seeing a panic >message like: > >8254 redirect via APIC pin0 impossible! > >should comment out i386/include/smptests.h: > >#define APIC_PIN0_TIMER > >Please don't do this unless you encounter this panic, then try it and report >results. Include your mptable (-dmesg not necessary). I've been running some tests with recent and week-or-so-old 3.0-current SMP kernels (ie: before and after the major changes) Guess I must be lucky, but I haven't had a single kernel panic, even under extreme IO conditions (simulated high-intensity news reading machines), with lots of local disk, network, and NFS traffic. The newer kernels also feel somewhat more responsive under load, though as yet I don't have any quantitive data to back this assertion up. Machine #1: dual-P133 on a Micronics motherboard (identified as 430NX EISA/PCI), 128Mb RAM, 3COM 3C579 EISA interface, mixture of EIDE and SCSI disk (Adaptec 2940) Machine #2: dual-PPro200 on an Intel PR440FX motherboard 512Mb RAM, DEC DE500XA 100baseT and DEC DEFPA FDDI interfaces, fast-wide Seagate disks on an Adaptec 3940. So, for me at least, it's a big thumbs-up to the new code :) -aDe -- Ade Lovett, Demon Internet Ltd. From owner-freebsd-smp Sun Jul 20 09:06:35 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA21711 for smp-outgoing; Sun, 20 Jul 1997 09:06:35 -0700 (PDT) Received: from ady.warp.starnets.ro (ady.warp.starnets.ro [193.226.124.33]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id JAA21672; Sun, 20 Jul 1997 09:06:17 -0700 (PDT) Received: from localhost (ady@localhost) by ady.warp.starnets.ro (8.8.5/8.8.5) with SMTP id TAA09139; Sun, 20 Jul 1997 19:04:59 +0300 (EEST) Date: Sun, 20 Jul 1997 19:04:59 +0300 (EEST) From: Penisoara Adrian Reply-To: Penisoara Adrian To: freebsd-smp@FreeBSD.ORG, freebsd-current@FreeBSD.ORG Subject: Need advice for SMP production server Message-ID: MIME-Version: 1.0 Content-Type: MULTIPART/MIXED; BOUNDARY="0-65234016-869414262=:8584" Content-ID: Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk This message is in MIME format. The first part should be readable text, while the remaining parts are likely unreadable without MIME-aware tools. Send mail to mime@docserver.cac.washington.edu for more info. --0-65234016-869414262=:8584 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII Content-ID: Hi ! I need some advice regarding a full reinstall for an 2xP133 server (Tyan Tomcat III S1563D with AHA2940AU). Currently I have 3.0-9702090-SNAP with the pre-Lite2 kernel (and its aic7xxx SCSI problem); I plan to install a newer 3.0-current version and I have to choose between the 3.0-970618-SNAP on the official FreeBSD server and the daily snapshots from current.FreeBSD.org. The problem is that this is a production server (which sustains all our ISP services) so I need it to be as much stable as it's possible. What are your opinions ? From what I read on the -current and -smp lists I understood that there are some minor problems with the 06/18 snapshot and also that Steve is started the long awaited move from the 'giant lock' to the 'finer grained lock' kernel architecture, which means current kernels are pretty untrustable. For the situation to be clear I attached an 'mptable -verbose -dmesg' output. All your comments are welcomed and appreciated. Thank you very much; apologises for any trouble caused. Ady (@warp.starnets.ro) --0-65234016-869414262=:8584 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII; NAME="mptable.out" Content-Transfer-Encoding: BASE64 Content-ID: Content-Description: MP table output DQo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09DQoNCk1QVGFi bGUsIHZlcnNpb24gMi4wLjEyDQoNCi0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0NCg0KTVAgRmxvYXRpbmcgUG9pbnRlciBTdHJ1Y3R1cmU6 DQoNCiAgbG9jYXRpb246CQkJQklPUw0KICBwaHlzaWNhbCBhZGRyZXNzOgkJ MHgwMDBmMGM4MA0KICBzaWduYXR1cmU6CQkJJ19NUF8nDQogIGxlbmd0aDoJ CQkxNiBieXRlcw0KICB2ZXJzaW9uOgkJCTEuMQ0KICBjaGVja3N1bToJCQkw eGY0DQogIG1vZGU6CQkJCVZpcnR1YWwgV2lyZQ0KDQotLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tDQoNCk1QIENvbmZpZyBUYWJsZSBIZWFk ZXI6DQoNCiAgcGh5c2ljYWwgYWRkcmVzczoJCTB4MDAwZjBjOTQNCiAgc2ln bmF0dXJlOgkJCSdQQ01QJw0KICBiYXNlIHRhYmxlIGxlbmd0aDoJCTI5Mg0K ICB2ZXJzaW9uOgkJCTEuMQ0KICBjaGVja3N1bToJCQkweGE1DQogIE9FTSBJ RDoJCQknT0VNMDAwMDAnDQogIFByb2R1Y3QgSUQ6CQkJJ1BST0QwMDAwMDAw MCcNCiAgT0VNIHRhYmxlIHBvaW50ZXI6CQkweDAwMDAwMDAwDQogIE9FTSB0 YWJsZSBzaXplOgkJMA0KICBlbnRyeSBjb3VudDoJCQkyOA0KICBsb2NhbCBB UElDIGFkZHJlc3M6CQkweGZlZTAwMDAwDQogIGV4dGVuZGVkIHRhYmxlIGxl bmd0aDoJMA0KICBleHRlbmRlZCB0YWJsZSBjaGVja3N1bToJMA0KDQotLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tDQoNCk1QIENvbmZpZyBC YXNlIFRhYmxlIEVudHJpZXM6DQoNCi0tDQpQcm9jZXNzb3JzOglBUElDIElE CVZlcnNpb24JU3RhdGUJCUZhbWlseQlNb2RlbAlTdGVwCUZsYWdzDQoJCSAw CSAweDExCSBCU1AsIHVzYWJsZQkgNQkgMgkgMQkgMHgwN2JmDQoJCSAxCSAw eDExCSBBUCwgdXNhYmxlCSA1CSAyCSAxCSAweDA3YmYNCi0tDQpCdXM6CQlC dXMgSUQJVHlwZQ0KCQkgMAkgSVNBICAgDQoJCSAxCSBQQ0kgICANCi0tDQpJ L08gQVBJQ3M6CUFQSUMgSUQJVmVyc2lvbglTdGF0ZQkJQWRkcmVzcw0KCQkg MgkgMHgxMQkgdXNhYmxlCQkgMHhmZWMwMDAwMA0KLS0NCkkvTyBJbnRzOglU eXBlCVBvbGFyaXR5ICAgIFRyaWdnZXIJQnVzIElECSBJUlEJQVBJQyBJRAlJ TlQjDQoJCUV4dElOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAg IDAJICAgICAgMgkgICAwDQoJCUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJ ICAgICAwCSAgIDEJICAgICAgMgkgICAxDQoJCUlOVAkgY29uZm9ybXMgICAg Y29uZm9ybXMJICAgICAwCSAgIDAJICAgICAgMgkgICAyDQoJCUlOVAkgY29u Zm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAgIDMJICAgICAgMgkgICAzDQoJ CUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAgIDQJICAgICAg MgkgICA0DQoJCUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAg IDUJICAgICAgMgkgICA1DQoJCUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJ ICAgICAwCSAgIDYJICAgICAgMgkgICA2DQoJCUlOVAkgY29uZm9ybXMgICAg Y29uZm9ybXMJICAgICAwCSAgIDcJICAgICAgMgkgICA3DQoJCUlOVAkgY29u Zm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAgIDgJICAgICAgMgkgICA4DQoJ CUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAgIDkJICAgICAg MgkgICA5DQoJCUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAg MTAJICAgICAgMgkgIDEwDQoJCUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJ ICAgICAwCSAgMTEJICAgICAgMgkgIDExDQoJCUlOVAkgY29uZm9ybXMgICAg Y29uZm9ybXMJICAgICAwCSAgMTIJICAgICAgMgkgIDEyDQoJCUlOVAkgY29u Zm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAgMTMJICAgICAgMgkgIDEzDQoJ CUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAgMTQJICAgICAg MgkgIDE0DQoJCUlOVAkgY29uZm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAg MTUJICAgICAgMgkgIDE1DQoJCUlOVAlhY3RpdmUtbG8gICAgICAgbGV2ZWwJ ICAgICAxCTIwOkEJICAgICAgMgkgIDE2DQoJCUlOVAlhY3RpdmUtbG8gICAg ICAgbGV2ZWwJICAgICAxCTE5OkEJICAgICAgMgkgIDE3DQoJCUlOVAlhY3Rp dmUtbG8gICAgICAgbGV2ZWwJICAgICAxCTE4OkEJICAgICAgMgkgIDE4DQoJ CUlOVAlhY3RpdmUtbG8gICAgICAgbGV2ZWwJICAgICAxCTE3OkEJICAgICAg MgkgIDE5DQoJCVNNSQkgY29uZm9ybXMgICAgY29uZm9ybXMJICAgICAwCSAg IDAJICAgICAgMgkgIDIzDQotLQ0KTG9jYWwgSW50czoJVHlwZQlQb2xhcml0 eSAgICBUcmlnZ2VyCUJ1cyBJRAkgSVJRCUFQSUMgSUQJSU5UIw0KCQlFeHRJ TlQJYWN0aXZlLWhpICAgICAgICBlZGdlCSAgICAgMAkgICAwCSAgICAyNTUJ ICAgMA0KCQlOTUkJYWN0aXZlLWhpICAgICAgICBlZGdlCSAgICAgMAkgICAw CSAgICAyNTUJICAgMQ0KDQotLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tDQoNCiMgU01QIGtlcm5lbCBjb25maWcgZmlsZSBvcHRpb25zOg0K DQoNCiMgUmVxdWlyZWQ6DQpvcHRpb25zCQlTTVAJCQkjIFN5bW1ldHJpYyBN dWx0aVByb2Nlc3NvciBLZXJuZWwNCm9wdGlvbnMJCUFQSUNfSU8JCQkjIFN5 bW1ldHJpYyAoQVBJQykgSS9PDQoNCiMgVXNlZnVsOg0KI29wdGlvbnMJCVNN UF9BVVRPU1RBUlQJCSMgc3RhcnQgdGhlIGFkZGl0aW9uYWwgQ1BVcyBkdXJp bmcgYm9vdA0KDQojIE9wdGlvbmFsIChidWlsdC1pbiBkZWZhdWx0cyB3aWxs IHdvcmsgaW4gbW9zdCBjYXNlcyk6DQojb3B0aW9ucwkJTkNQVT0yCQkJIyBu dW1iZXIgb2YgQ1BVcw0KI29wdGlvbnMJCU5CVVM9MgkJCSMgbnVtYmVyIG9m IGJ1c3Nlcw0KI29wdGlvbnMJCU5BUElDPTEJCQkjIG51bWJlciBvZiBJTyBB UElDcw0KI29wdGlvbnMJCU5JTlRSPTI0CQkjIG51bWJlciBvZiBJTlRzDQoN CiMgUm9ndWUgaGFyZHdhcmU6DQojDQojICBUeWFuIFRvbWNhdCBJSToNCiNv cHRpb25zCQlTTVBfVElNRVJfTkMJCSMgDQojDQojICBTdXBlck1pY3JvIFA2 RE5FOg0KI29wdGlvbnMJCVNNUF9USU1FUl9OQwkJIyANCg0KPT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PQ0KDQo= --0-65234016-869414262=:8584-- From owner-freebsd-smp Sun Jul 20 09:19:31 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA22328 for smp-outgoing; Sun, 20 Jul 1997 09:19:31 -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 JAA22323 for ; Sun, 20 Jul 1997 09:19:27 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id KAA00693; Sun, 20 Jul 1997 10:19:15 -0600 (MDT) Message-Id: <199707201619.KAA00693@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: ade@demon.net cc: smp@FreeBSD.ORG Subject: Re: HEADS UP: major timer/APIC code changes. In-reply-to: Your message of "Sun, 20 Jul 1997 17:04:21 BST." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 20 Jul 1997 10:19:15 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > I've been running some tests with recent and week-or-so-old > 3.0-current SMP kernels (ie: before and after the major changes) > > Guess I must be lucky, but I haven't had a single kernel panic, > even under extreme IO conditions (simulated high-intensity news > reading machines), with lots of local disk, network, and NFS > traffic. this is good news. I also have both machines running nicely. The only expected panics are of a static nature during boot when the kernel discovers its missing a necessary APIC connection. No one has reported random panics (knock on silicon) yet. --- > The newer kernels also feel somewhat more responsive under load, > though as yet I don't have any quantitive data to back this > assertion up. I had the same feeling of improvement but figured it was just wishful thinking on my part. Thanx for the report. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sun Jul 20 09:38:10 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA23197 for smp-outgoing; Sun, 20 Jul 1997 09:38:10 -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 JAA23174; Sun, 20 Jul 1997 09:38:05 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id KAA00807; Sun, 20 Jul 1997 10:35:50 -0600 (MDT) Message-Id: <199707201635.KAA00807@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Penisoara Adrian cc: freebsd-smp@FreeBSD.ORG, freebsd-current@FreeBSD.ORG Subject: Re: Need advice for SMP production server In-reply-to: Your message of "Sun, 20 Jul 1997 19:04:59 +0300." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 20 Jul 1997 10:35:50 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > I need some advice regarding a full reinstall for an 2xP133 server (Tyan > Tomcat III S1563D with AHA2940AU). > ... > From what I read on the -current and -smp lists I understood that there > are some minor problems with the 06/18 snapshot and also that Steve is > started the long awaited move from the 'giant lock' to the 'finer grained > lock' kernel architecture, which means current kernels are pretty > untrustable. > For the situation to be clear I attached an 'mptable -verbose -dmesg' > output. your mptable shows the proper connections for the new code. I would tend to suggest the latest SNAP: 970718. Most reports are of good success. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sun Jul 20 09:52:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA23988 for smp-outgoing; Sun, 20 Jul 1997 09:52:49 -0700 (PDT) Received: from ady.warp.starnets.ro (ady.warp.starnets.ro [193.226.124.33]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id JAA23971 for ; Sun, 20 Jul 1997 09:52:37 -0700 (PDT) Received: from localhost (ady@localhost) by ady.warp.starnets.ro (8.8.5/8.8.5) with SMTP id TAA09251; Sun, 20 Jul 1997 19:50:58 +0300 (EEST) Date: Sun, 20 Jul 1997 19:50:57 +0300 (EEST) From: Penisoara Adrian To: Steve Passe cc: freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server In-Reply-To: <199707201635.KAA00807@Ilsa.StevesCafe.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi! On Sun, 20 Jul 1997, Steve Passe wrote: > Hi, > [...] > your mptable shows the proper connections for the new code. I would tend > to suggest the latest SNAP: 970718. Most reports are of good success. Great ! That means I can taste the new kernel architecture and also that I have the chance to remain -current in time... What about stability ? Did anybody had problems with configurations similar to mine (specially the SCSI subsystem)? > > -- > Steve Passe | powered by > smp@csn.net | Symmetric MultiProcessor FreeBSD Ady (@warp.starnets.ro) From owner-freebsd-smp Sun Jul 20 10:06:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA24526 for smp-outgoing; Sun, 20 Jul 1997 10:06:26 -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 KAA24492; Sun, 20 Jul 1997 10:06:10 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.6/8.8.5) id MAA03874; Sun, 20 Jul 1997 12:05:12 -0500 (EST) From: "John S. Dyson" Message-Id: <199707201705.MAA03874@dyson.iquest.net> Subject: Re: Need advice for SMP production server In-Reply-To: <199707201635.KAA00807@Ilsa.StevesCafe.com> from Steve Passe at "Jul 20, 97 10:35:50 am" To: smp@csn.net (Steve Passe) Date: Sun, 20 Jul 1997 12:05:12 -0500 (EST) Cc: ady@warp.starnets.ro, freebsd-smp@FreeBSD.ORG, freebsd-current@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-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Hi, > > > I need some advice regarding a full reinstall for an 2xP133 server (Tyan > > Tomcat III S1563D with AHA2940AU). > > ... > > From what I read on the -current and -smp lists I understood that there > > are some minor problems with the 06/18 snapshot and also that Steve is > > started the long awaited move from the 'giant lock' to the 'finer grained > > lock' kernel architecture, which means current kernels are pretty > > untrustable. > > For the situation to be clear I attached an 'mptable -verbose -dmesg' > > output. > > your mptable shows the proper connections for the new code. I would tend > to suggest the latest SNAP: 970718. Most reports are of good success. > One problem -- there is an issue with my latest VM changes. You'll want to remove pmap_addr_hint call from vm_mmap.c. It (indirectly) breaks X. John From owner-freebsd-smp Sun Jul 20 10:23:44 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA25093 for smp-outgoing; Sun, 20 Jul 1997 10:23:44 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA25083; Sun, 20 Jul 1997 10:23:31 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id TAA01514; Sun, 20 Jul 1997 19:15:15 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.6/8.8.6) id TAA09070; Sun, 20 Jul 1997 19:01:17 +0200 (CEST) Message-ID: <19970720190117.09549@gtn.com> Date: Sun, 20 Jul 1997 19:01:17 +0200 From: Andreas Klemm To: Penisoara Adrian Cc: freebsd-smp@FreeBSD.ORG, freebsd-current@FreeBSD.ORG Subject: Re: Need advice for SMP production server References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.79 In-Reply-To: ; from Penisoara Adrian on Sun, Jul 20, 1997 at 07:04:59PM +0300 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT SMP Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Sun, Jul 20, 1997 at 07:04:59PM +0300, Penisoara Adrian wrote: > > Hi ! > > I need some advice regarding a full reinstall for an 2xP133 server (Tyan > Tomcat III S1563D with AHA2940AU). > Currently I have 3.0-9702090-SNAP with the pre-Lite2 kernel (and its > aic7xxx SCSI problem); I plan to install a newer 3.0-current version and I > have to choose between the 3.0-970618-SNAP on the official FreeBSD server > and the daily snapshots from current.FreeBSD.org. Generally I'd want to warn you, that -current is not the typical environment for a production machine. You should use FreeBSD-2.2.2 and cvsup the 2.2-STABLE version. On the other hand -current is since a long time in a very good shape with only minor trouble from time to time. And it's the only release, that has SMP capabilities. Since I use better hardware and have a new power supply, I had no big trouble and am running a very fast 2 x 200 MHz PPro machine. Andreas /// -- Andreas Klemm | klemm.gtn.com - powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html http://www.freebsd.org/~fsmp/SMP/benches.html From owner-freebsd-smp Sun Jul 20 10:37:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA25703 for smp-outgoing; Sun, 20 Jul 1997 10:37:34 -0700 (PDT) Received: from ady.warp.starnets.ro (ady.warp.starnets.ro [193.226.124.33]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA25684; Sun, 20 Jul 1997 10:37:22 -0700 (PDT) Received: from localhost (ady@localhost) by ady.warp.starnets.ro (8.8.5/8.8.5) with SMTP id UAA09352; Sun, 20 Jul 1997 20:35:37 +0300 (EEST) Date: Sun, 20 Jul 1997 20:35:37 +0300 (EEST) From: Penisoara Adrian To: Andreas Klemm cc: freebsd-smp@FreeBSD.ORG, freebsd-current@FreeBSD.ORG Subject: Re: Need advice for SMP production server In-Reply-To: <19970720190117.09549@gtn.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, On Sun, 20 Jul 1997, Andreas Klemm wrote: [...] > Generally I'd want to warn you, that -current is not the typical > environment for a production machine. You should use FreeBSD-2.2.2 > and cvsup the 2.2-STABLE version. Hmmm, I NEED to run SMP, so it's obvious I'm talking 3.0-current... > > On the other hand -current is since a long time in a very good > shape with only minor trouble from time to time. And it's the > only release, that has SMP capabilities. Personally I had no trouble since installation, except for the buggy SCSI driver (aic7xx in 970209-SNAP) which forced me to assist to about 20-25 reboots since then :) I know, I could have tried an cvsupped kernel, but that happened right in the middle of the 4.4BSD Lite2 'upgrade', and lazy/scary me I haven't the gods to do a make world ;) > > Since I use better hardware and have a new power supply, I had no > big trouble and am running a very fast 2 x 200 MHz PPro machine. Happy SMPfying ! BTW, overclocked ? (Mine not, and not gonna happen...) > > Andreas /// > > -- > Andreas Klemm | klemm.gtn.com - powered by > Symmetric MultiProcessor FreeBSD > http://www.freebsd.org/~fsmp/SMP/SMP.html > http://www.freebsd.org/~fsmp/SMP/benches.html > Ady (@warp.starnets.ro) From owner-freebsd-smp Sun Jul 20 13:00:53 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA00942 for smp-outgoing; Sun, 20 Jul 1997 13:00:53 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [194.77.0.15]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA00920 for ; Sun, 20 Jul 1997 13:00:43 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id VAA23714; Sun, 20 Jul 1997 21:45:19 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.6/8.8.6) id VAA17183; Sun, 20 Jul 1997 21:17:27 +0200 (CEST) Message-ID: <19970720211727.40182@gtn.com> Date: Sun, 20 Jul 1997 21:17:27 +0200 From: Andreas Klemm To: Penisoara Adrian Cc: Andreas Klemm , freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server References: <19970720190117.09549@gtn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.79 In-Reply-To: ; from Penisoara Adrian on Sun, Jul 20, 1997 at 08:35:37PM +0300 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT SMP Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Sun, Jul 20, 1997 at 08:35:37PM +0300, Penisoara Adrian wrote: > Personally I had no trouble since installation, except for the buggy SCSI > driver (aic7xx in 970209-SNAP) which forced me to assist to about 20-25 > reboots since then :) Those times are gone, it runs really fine. > Happy SMPfying ! BTW, overclocked ? (Mine not, and not gonna happen...) I think my cooling isn't well enough. Running at 233 MHz the machine doesn't boot. -- Andreas Klemm | klemm.gtn.com - powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html http://www.freebsd.org/~fsmp/SMP/benches.html From owner-freebsd-smp Sun Jul 20 13:04:31 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA01065 for smp-outgoing; Sun, 20 Jul 1997 13:04:31 -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 NAA01060 for ; Sun, 20 Jul 1997 13:04:27 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id OAA01705 for ; Sun, 20 Jul 1997 14:03:54 -0600 (MDT) Message-Id: <199707202003.OAA01705@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: even newer code Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 20 Jul 1997 14:03:54 -0600 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, I have reworked the recent changes to the APIC cocde. Hopefully this will boot for everyone without needing to customize the compile. Try it first without any changes to smptests.h, thats the purpose to this round of changes! -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jul 21 01:04:48 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id BAA29175 for smp-outgoing; Mon, 21 Jul 1997 01:04:48 -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 BAA29166 for ; Mon, 21 Jul 1997 01:04:44 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id CAA04033 for ; Mon, 21 Jul 1997 02:04:44 -0600 (MDT) Message-Id: <199707210804.CAA04033@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: HEADS UP: new npx.c code. Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 21 Jul 1997 02:04:44 -0600 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, I just committed a new version of i386/isa/npx.c that eliminates all the INT13 interface code for the SMP case. This allows me to remove two APIC inlines that would otherwise need to be made MP-safe. It also reduces the overhead of npxsave() by a *bunch*! It should be invisible, but be on the lookout for any math related funkiness. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jul 21 02:04:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA02014 for smp-outgoing; Mon, 21 Jul 1997 02:04:26 -0700 (PDT) Received: from ady.warp.starnets.ro (ady.warp.starnets.ro [193.226.124.33]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id CAA01985 for ; Mon, 21 Jul 1997 02:04:15 -0700 (PDT) Received: from localhost (ady@localhost) by ady.warp.starnets.ro (8.8.5/8.8.5) with SMTP id MAA10637; Mon, 21 Jul 1997 12:02:31 +0300 (EEST) Date: Mon, 21 Jul 1997 12:02:31 +0300 (EEST) From: Penisoara Adrian Reply-To: Penisoara Adrian To: Steve Passe cc: freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, On Sun, 20 Jul 1997, Penisoara Adrian wrote: > Hi! [...] > Great ! That means I can taste the new kernel architecture and also that > I have the chance to remain -current in time... One more thing: if I get 3.0-970618-SNAP and cvsup to latest -current kernel do I need (for now) to do a make world ? And another one: except those mmap and vm_mmap.c things, is there anything else broken or not functioning at optimum stability/speed ? (I don't need the system to be too fast but rather functional/stable). > > > > > -- > > Steve Passe | powered by > > smp@csn.net | Symmetric MultiProcessor FreeBSD > > Ady (@warp.starnets.ro) As I said, Ady (@warp.starnets.ro) From owner-freebsd-smp Mon Jul 21 02:12:21 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA02305 for smp-outgoing; Mon, 21 Jul 1997 02:12:21 -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 CAA02300 for ; Mon, 21 Jul 1997 02:12:18 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id DAA04415; Mon, 21 Jul 1997 03:11:58 -0600 (MDT) Message-Id: <199707210911.DAA04415@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Penisoara Adrian cc: freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server In-reply-to: Your message of "Mon, 21 Jul 1997 12:02:31 +0300." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 21 Jul 1997 03:11:57 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > One more thing: if I get 3.0-970618-SNAP and cvsup to latest -current > ^ > kernel do I need (for now) to do a make world ? I recommended the 7-18 SNAP, if thats what you meant, no "world" will be new enough. but to use a 6-18 SNAP with the latest SMP sys you would probably want to cvsup & make world. --- > And another one: except those mmap and vm_mmap.c things, is there anything > else broken or not functioning at optimum stability/speed ? (I don't need > the system to be too fast but rather functional/stable). NFS is giving people some trouble, but thats all I'm aware of... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jul 21 02:29:13 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA02725 for smp-outgoing; Mon, 21 Jul 1997 02:29:13 -0700 (PDT) Received: from sendero-ppp.i-connect.net (sendero-ppp.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id CAA02717 for ; Mon, 21 Jul 1997 02:29:10 -0700 (PDT) Received: (qmail 8510 invoked by uid 1000); 21 Jul 1997 09:29:24 -0000 Message-ID: X-Mailer: XFMail 1.2-alpha [p0] on FreeBSD Content-Type: text/plain; charset=iso-8859-8 Content-Transfer-Encoding: 8bit MIME-Version: 1.0 Date: Mon, 21 Jul 1997 02:29:23 -0700 (PDT) Organization: Atlas Telecom From: Simon Shapiro To: freebsd-smp@freebsd.org Subject: P6DNH Question Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi Y'all, -current of tonight, built an SMP kernel, using the SAME configuration as an UP kernel, except: options SMP options APIC_IO Used to also have options SMP_TIMER_NC but this one appears very dead. Kernel used to boot and run well up to a week ago or so. Now we are getting this behavior: Boots normally up to the point just before mounting files systems. At which points it takes a vacation: .... changing root device to sd0a DEVFS: ready to run APIC_IO: routing 8254 via 8259 on pin 0 ccd0-63: concatenated disk drivers [ Nothing further happens, so we hit ctl-alt-ESC ] CPU0 stopping CPUs: 0x00000002 stopped [ddb trace command yields (approximately)] Debugger... scgetc scintr Xresume1() at Xresume1 _ 0x2b ---interrupt eip=0xf01ed729 esp=f022dff4 ebp=0 default_halt() at default_halt + 0x01 [ DDB continue command ] syncing disks... trap 12 CPU0: Stopping.... stopped at tsleep+0x1a movl b0(%ebx),%eax tsleep(f13fb000, 110, f01da601,2710,f01da5ff) The bit about tsleep comes from the DPT driver. It does NOT crash like this on normal boot, only when triggered from a DDB. I think it is worth fixing but feel it may be unrelated to the not-booting-but-halting problem. Any suggestion is welcome. Simon ent of tonight, built an SMP kernel, using the SAME configuration as an UP kernel, except: options SMP options APIC_IO Used to also have options SMP_TIMER_NC but this one appears very dead. Kernel used to boot and run we From owner-freebsd-smp Mon Jul 21 04:53:53 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id EAA09206 for smp-outgoing; Mon, 21 Jul 1997 04:53:53 -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 EAA09201 for ; Mon, 21 Jul 1997 04:53:45 -0700 (PDT) Received: from spinner.dialix.com.au (localhost.dialix.com.au [127.0.0.1]) by spinner.dialix.com.au with ESMTP id TAA25108; Mon, 21 Jul 1997 19:51:34 +0800 (WST) Message-Id: <199707211151.TAA25108@spinner.dialix.com.au> X-Mailer: exmh version 2.0delta 6/3/97 To: Andreas Klemm cc: Penisoara Adrian , freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server In-reply-to: Your message of "Sun, 20 Jul 1997 21:17:27 +0200." <19970720211727.40182@gtn.com> Date: Mon, 21 Jul 1997 19:51:33 +0800 From: Peter Wemm Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Andreas Klemm wrote: > On Sun, Jul 20, 1997 at 08:35:37PM +0300, Penisoara Adrian wrote: [..] > > Happy SMPfying ! BTW, overclocked ? (Mine not, and not gonna happen...) > > I think my cooling isn't well enough. Running at 233 MHz the machine > doesn't boot. I'm running an Intel PR440FX with 2xPPro200's at 233MHz each. It just screams along! :-) Steve has the same motherboard but it doesn't work reliably for him at 233.. He has 512K cache cpus though, I've got 256K. Cheers, -Peter From owner-freebsd-smp Mon Jul 21 06:58:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id GAA14376 for smp-outgoing; Mon, 21 Jul 1997 06:58:33 -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 GAA14371 for ; Mon, 21 Jul 1997 06:58:31 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id HAA01064; Mon, 21 Jul 1997 07:57:49 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id HAA04664; Mon, 21 Jul 1997 07:57:46 -0600 Date: Mon, 21 Jul 1997 07:57:46 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199707211357.HAA04664@fast.cs.utah.edu> To: andreas@klemm.gtn.com, peter@spinner.dialix.com.au Subject: Re: Need advice for SMP production server Cc: ady@warp.starnets.ro, freebsd-smp@FreeBSD.ORG Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > I'm running an Intel PR440FX with 2xPPro200's at 233MHz each. It just > screams along! :-) Steve has the same motherboard but it doesn't work > reliably for him at 233.. He has 512K cache cpus though, I've got 256K. But you started with 200's. Steve's are 166's already overclocked to 200. I'm not surprised his don't run at 233 ;) From owner-freebsd-smp Mon Jul 21 09:38:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA22495 for smp-outgoing; Mon, 21 Jul 1997 09:38:09 -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 JAA22487 for ; Mon, 21 Jul 1997 09:38:00 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id KAA05819; Mon, 21 Jul 1997 10:37:50 -0600 (MDT) Message-Id: <199707211637.KAA05819@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Simon Shapiro cc: freebsd-smp@FreeBSD.ORG Subject: Re: P6DNH Question In-reply-to: Your message of "Mon, 21 Jul 1997 02:29:23 PDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 21 Jul 1997 10:37:50 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > -current of tonight, built an SMP kernel, using the SAME configuration as > an UP kernel, except: > ... > Used to also have > > options SMP_TIMER_NC > > but this one appears very dead. > > Kernel used to boot and run well up to a week ago or so. Now we are > getting this behavior: > ... in your smptests.h file: comment out: NEW_STRATEGY APIC_PIN0_TIMER TEST_ALTTIMER TIMER_ALL then go to the i386/conf dir and restore the "options SMP_TIMER_NC". run config, go to the build dir, make depend && make && make install. let me know if this works (or not). -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jul 21 09:43:08 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA22921 for smp-outgoing; Mon, 21 Jul 1997 09:43:08 -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 JAA22914 for ; Mon, 21 Jul 1997 09:43:01 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id KAA05833; Mon, 21 Jul 1997 10:39:59 -0600 (MDT) Message-Id: <199707211639.KAA05833@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: vanmaren@fast.cs.utah.edu (Kevin Van Maren) cc: andreas@klemm.gtn.com, peter@spinner.dialix.com.au, ady@warp.starnets.ro, freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server In-reply-to: Your message of "Mon, 21 Jul 1997 07:57:46 MDT." <199707211357.HAA04664@fast.cs.utah.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 21 Jul 1997 10:39:59 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > I'm running an Intel PR440FX with 2xPPro200's at 233MHz each. It just > > screams along! :-) Steve has the same motherboard but it doesn't work > > reliably for him at 233.. He has 512K cache cpus though, I've got 256K. > > But you started with 200's. Steve's are 166's already overclocked to 200. > I'm not surprised his don't run at 233 ;) actually I'm running a gigabyte with 166's clocked at 200, and a pr440fx with 200's clocked at 200. the pr440fx used to run at 233, but after 1 set of code changes a couple months back I had to go back to 200 to maintain reliability. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jul 21 10:25:38 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA26290 for smp-outgoing; Mon, 21 Jul 1997 10:25:38 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA26275 for ; Mon, 21 Jul 1997 10:25:29 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id TAA06915; Mon, 21 Jul 1997 19:15:47 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.6/8.8.6) id TAA02824; Mon, 21 Jul 1997 19:06:59 +0200 (CEST) Message-ID: <19970721190659.28219@gtn.com> Date: Mon, 21 Jul 1997 19:06:59 +0200 From: Andreas Klemm To: Peter Wemm Cc: Andreas Klemm , Penisoara Adrian , freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server References: <19970720211727.40182@gtn.com> <199707211151.TAA25108@spinner.dialix.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.79 In-Reply-To: <199707211151.TAA25108@spinner.dialix.com.au>; from Peter Wemm on Mon, Jul 21, 1997 at 07:51:33PM +0800 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT SMP Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Mon, Jul 21, 1997 at 07:51:33PM +0800, Peter Wemm wrote: > I'm running an Intel PR440FX with 2xPPro200's at 233MHz each. It just > screams along! :-) Steve has the same motherboard but it doesn't work > reliably for him at 233.. He has 512K cache cpus though, I've got 256K. My heat sinks aren't large enough I think. A friend of a colleague of mine knows distributors for huge heat sinks. Will try such a beast and try to get 233 MHz ... Might 266 be possible ???? ;-)))))) -- Andreas Klemm | klemm.gtn.com - powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html http://www.freebsd.org/~fsmp/SMP/benches.html From owner-freebsd-smp Mon Jul 21 10:40:07 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA27358 for smp-outgoing; Mon, 21 Jul 1997 10:40:07 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA27287 for ; Mon, 21 Jul 1997 10:39:55 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id TAA09914; Mon, 21 Jul 1997 19:30:26 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.6/8.8.6) id TAA03332; Mon, 21 Jul 1997 19:21:30 +0200 (CEST) Message-ID: <19970721192130.08801@gtn.com> Date: Mon, 21 Jul 1997 19:21:30 +0200 From: Andreas Klemm To: Steve Passe Cc: Kevin Van Maren , andreas@klemm.gtn.com, peter@spinner.dialix.com.au, ady@warp.starnets.ro, freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server References: <199707211357.HAA04664@fast.cs.utah.edu> <199707211639.KAA05833@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.79 In-Reply-To: <199707211639.KAA05833@Ilsa.StevesCafe.com>; from Steve Passe on Mon, Jul 21, 1997 at 10:39:59AM -0600 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT SMP Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Mon, Jul 21, 1997 at 10:39:59AM -0600, Steve Passe wrote: > Hi, > > > > I'm running an Intel PR440FX with 2xPPro200's at 233MHz each. It just > > > screams along! :-) Steve has the same motherboard but it doesn't work > > > reliably for him at 233.. He has 512K cache cpus though, I've got 256K. > > > > But you started with 200's. Steve's are 166's already overclocked to 200. > > I'm not surprised his don't run at 233 ;) > > actually I'm running a gigabyte with 166's clocked at 200, and a pr440fx > with 200's clocked at 200. the pr440fx used to run at 233, but after > 1 set of code changes a couple months back I had to go back to 200 to > maintain reliability. What dimensions do your heatsinks have ??? As I said, I was unable to boot at 200 MHz. I have a fan on top of the heatsink and a fan is blowing over the CPU's (located in front of the ATX power supply). +---------------------------... | | | Power supply | | +--------------------------.... | FFFFFFFFFFFFFFFFFan | FFFFFFFFFFFFFFFFFan Blows to the CPU but not direktly onto it | | | | | CPU2_with_Heatsink | CPU FANFANFAN sink | CPU FANFANFAN sink | CPU FANFANFAN sink | CPU FANFANFAN sink | CPU2_with_Heatsink | | CPU1_with_Heatsink | CPU FANFANFAN sink | CPU FANFANFAN sink | CPU FANFANFAN sink | CPU FANFANFAN sink | CPU1_with_Heatsink | -- Andreas Klemm | klemm.gtn.com - powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html http://www.freebsd.org/~fsmp/SMP/benches.html From owner-freebsd-smp Mon Jul 21 10:47:48 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA28183 for smp-outgoing; Mon, 21 Jul 1997 10:47:48 -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 KAA28178 for ; Mon, 21 Jul 1997 10:47:46 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id LAA07569; Mon, 21 Jul 1997 11:47:43 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id LAA14873; Mon, 21 Jul 1997 11:47:42 -0600 Date: Mon, 21 Jul 1997 11:47:42 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199707211747.LAA14873@fast.cs.utah.edu> To: freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk If you think cooling is the only problem, get the dual Pro fans from PC Power and Cooling (www.pcpowercooling.com) and be sure to correctly apply the thermal grease. Using thermal grease properly dramatically reduces the thermal resistance, which enables lower operating temperatures. Kevin (ps, the dual fan retails for $39 each) From owner-freebsd-smp Mon Jul 21 10:55:02 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA28920 for smp-outgoing; Mon, 21 Jul 1997 10:55:02 -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 KAA28898 for ; Mon, 21 Jul 1997 10:54:54 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id LAA06243; Mon, 21 Jul 1997 11:52:31 -0600 (MDT) Message-Id: <199707211752.LAA06243@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Andreas Klemm cc: Kevin Van Maren , peter@spinner.dialix.com.au, ady@warp.starnets.ro, freebsd-smp@FreeBSD.ORG Subject: Re: Need advice for SMP production server In-reply-to: Your message of "Mon, 21 Jul 1997 19:21:30 +0200." <19970721192130.08801@gtn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 21 Jul 1997 11:52:31 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > actually I'm running a gigabyte with 166's clocked at 200, and a pr440fx > > with 200's clocked at 200. the pr440fx used to run at 233, but after > > 1 set of code changes a couple months back I had to go back to 200 to > > maintain reliability. > > What dimensions do your heatsinks have ??? As I said, I was unable > to boot at 200 MHz. I have a fan on top of the heatsink and a fan > is blowing over the CPU's (located in front of the ATX power supply). about same size as the chip. On the 166 -> 200 I put a nice layer of heat compound. the 200 -> 200 came assembled, I doubt they used compound... Looks like same brand on both. Next time I powerdown I'll try to remember to read the label. I have both in an enclosed rack, with a dryer hose running from window air-conditioner to the bottom of the rack. Temp sensors in rack typically show (farenheight): top: 80 middle: 78 bottom: 72 I've never seen the top get higher than 84. This rack has 5 systems running 7/24, external disks, tapes, modems, etc. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jul 21 11:41:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA02175 for smp-outgoing; Mon, 21 Jul 1997 11:41:49 -0700 (PDT) Received: from adsdevelop.autodebit.com (adsdevelop.autodebit.com [204.50.245.10]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA02169 for ; Mon, 21 Jul 1997 11:41:39 -0700 (PDT) Received: by adsdevelop.autodebit.com with Microsoft Exchange (IMC 4.0.837.3) id <01BC95CA.C4A26240@adsdevelop.autodebit.com>; Mon, 21 Jul 1997 11:39:37 -0700 Message-ID: From: David Green-Seed To: "'freebsd-smp@FreeBSD.ORG'" Subject: RE: Need advice for SMP production server Date: Mon, 21 Jul 1997 11:39:34 -0700 X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.837.3 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk You may also want to check out a Peltier effect cooler. Check out http://www.computernerd.com/coolspec.htm#PAX56B Dave. >-----Original Message----- >From: vanmaren@fast.cs.utah.edu [SMTP:vanmaren@fast.cs.utah.edu] >Sent: Monday, July 21, 1997 10:48 AM >To: freebsd-smp@FreeBSD.ORG >Subject: Re: Need advice for SMP production server > > >If you think cooling is the only problem, get the dual Pro fans >from PC Power and Cooling (www.pcpowercooling.com) and be sure to >correctly apply the thermal grease. Using thermal grease properly >dramatically reduces the thermal resistance, which enables lower >operating temperatures. > >Kevin > >(ps, the dual fan retails for $39 each) From owner-freebsd-smp Mon Jul 21 11:50:20 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA02682 for smp-outgoing; Mon, 21 Jul 1997 11:50:20 -0700 (PDT) Received: from bob.tri-lakes.net ([207.3.81.6]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA02673 for ; Mon, 21 Jul 1997 11:50:14 -0700 (PDT) Received: from [207.3.81.149] by bob.tri-lakes.net (NTMail 3.02.13) with ESMTP id sa233186 for ; Mon, 21 Jul 1997 13:50:32 -0500 Message-ID: <33D3691F.2781E494@tri-lakes.net> Date: Mon, 21 Jul 1997 13:50:23 +0000 From: Chris Dillon X-Mailer: Mozilla 3.01Gold (X11; I; FreeBSD 2.2-STABLE i386) MIME-Version: 1.0 To: Andreas Klemm CC: freebsd-smp@freebsd.org Subject: Re: Need advice for SMP production server References: <19970720211727.40182@gtn.com> <199707211151.TAA25108@spinner.dialix.com.au> <19970721190659.28219@gtn.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Andreas Klemm wrote: > > On Mon, Jul 21, 1997 at 07:51:33PM +0800, Peter Wemm wrote: > > I'm running an Intel PR440FX with 2xPPro200's at 233MHz each. It just > > screams along! :-) Steve has the same motherboard but it doesn't work > > reliably for him at 233.. He has 512K cache cpus though, I've got 256K. > > My heat sinks aren't large enough I think. A friend of a colleague > of mine knows distributors for huge heat sinks. Will try such a beast > and try to get 233 MHz ... Might 266 be possible ???? ;-)))))) You may also want to try some coolers which use Peltier devices. Only drawback is they cost more ($40 to $70 each) and eat another 1.5 to 2.5 amps (at 12 volts) from your power supply. Most of the devices say they can keep the CPU below 60C (depends on CPU and cooling device combination, of course). From owner-freebsd-smp Mon Jul 21 12:45:54 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA06540 for smp-outgoing; Mon, 21 Jul 1997 12:45:54 -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 MAA06499 for ; Mon, 21 Jul 1997 12:45:15 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id NAA06718; Mon, 21 Jul 1997 13:44:43 -0600 (MDT) Message-Id: <199707211944.NAA06718@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: David Green-Seed cc: "'freebsd-smp@FreeBSD.ORG'" Subject: Re: Need advice for SMP production server In-reply-to: Your message of "Mon, 21 Jul 1997 11:39:34 PDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 21 Jul 1997 13:44:42 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > You may also want to check out a Peltier effect cooler. Check out > http://www.computernerd.com/coolspec.htm#PAX56B They don't seem to offer a PPRo version!! -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jul 21 14:53:51 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA14130 for smp-outgoing; Mon, 21 Jul 1997 14:53:51 -0700 (PDT) Received: from sendero-ppp.i-connect.net (sendero-ppp.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id OAA14123 for ; Mon, 21 Jul 1997 14:53:47 -0700 (PDT) Received: (qmail 19795 invoked by uid 1000); 21 Jul 1997 21:53:59 -0000 Message-ID: X-Mailer: XFMail 1.1 [p0] on FreeBSD Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <199707211637.KAA05819@Ilsa.StevesCafe.com> Date: Mon, 21 Jul 1997 14:22:19 -0700 (PDT) Organization: Atlas Telecom From: Simon Shapiro To: Steve Passe Subject: Re: P6DNH Question Cc: freebsd-smp@FreeBSD.ORG Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi Steve Passe; On 21-Jul-97 you wrote: > Hi, > > > -current of tonight, built an SMP kernel, using the SAME configuration > as > > an UP kernel, except: > > ... > > Used to also have > > > > options SMP_TIMER_NC > > > > but this one appears very dead. > > > > Kernel used to boot and run well up to a week ago or so. Now we are > > getting this behavior: > > ... > > in your smptests.h file: > comment out: > NEW_STRATEGY > APIC_PIN0_TIMER > TEST_ALTTIMER > TIMER_ALL > > then go to the i386/conf dir and restore the "options SMP_TIMER_NC". > run config, go to the build dir, make depend && make && make install. > let me know if this works (or not). Not :-( Same symptoms; stuck in default_halt just before execing init. Simon From owner-freebsd-smp Mon Jul 21 15:04:07 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA14777 for smp-outgoing; Mon, 21 Jul 1997 15:04:07 -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 PAA14728 for ; Mon, 21 Jul 1997 15:03:51 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id QAA07227; Mon, 21 Jul 1997 16:03:34 -0600 (MDT) Message-Id: <199707212203.QAA07227@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Simon Shapiro cc: freebsd-smp@FreeBSD.ORG Subject: Re: P6DNH Question In-reply-to: Your message of "Mon, 21 Jul 1997 14:22:19 PDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 21 Jul 1997 16:03:34 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > in your smptests.h file: > > comment out: > > NEW_STRATEGY > > APIC_PIN0_TIMER > > TEST_ALTTIMER > > TIMER_ALL > > > > then go to the i386/conf dir and restore the "options SMP_TIMER_NC". > > run config, go to the build dir, make depend && make && make install. > > let me know if this works (or not). > > Not :-( > > Same symptoms; stuck in default_halt just before execing init. I'm going to need detailed info to be able to be of help. The ideal situation is to setup your box to use a serial console, and use an xterm on another machine to connect to the target box. Then you can capture the complete output thru the point of failure. See the freebsd handbook for details. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Jul 21 19:01:12 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA27147 for smp-outgoing; Mon, 21 Jul 1997 19:01:12 -0700 (PDT) Received: from sendero-ppp.i-connect.net (sendero-ppp.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id TAA27136 for ; Mon, 21 Jul 1997 19:01:05 -0700 (PDT) Received: (qmail 15010 invoked by uid 1000); 22 Jul 1997 02:01:18 -0000 Message-ID: X-Mailer: XFMail 1.1 [p0] on FreeBSD Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <199707212203.QAA07227@Ilsa.StevesCafe.com> Date: Mon, 21 Jul 1997 15:21:22 -0700 (PDT) Organization: Atlas Telecom From: Simon Shapiro To: Steve Passe Subject: Re: P6DNH Question Cc: freebsd-smp@FreeBSD.ORG Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi Steve Passe; On 21-Jul-97 you wrote: > Hi, > > > > in your smptests.h file: > > > comment out: > > > NEW_STRATEGY > > > APIC_PIN0_TIMER > > > TEST_ALTTIMER > > > TIMER_ALL > > > > > > then go to the i386/conf dir and restore the "options SMP_TIMER_NC". > > > run config, go to the build dir, make depend && make && make install. > > > let me know if this works (or not). > > > > Not :-( > > > > Same symptoms; stuck in default_halt just before execing init. > > I'm going to need detailed info to be able to be of help. The ideal > situation is to setup your box to use a serial console, and use an > xterm on another machine to connect to the target box. Then you can > capture > the complete output thru the point of failure. See the freebsd handbook > for details. I already posted all the details (OK, minus some prompts). There is nothing much to tell; Boot proceeds totally normally until the point at which /etc/rc starts (on a UP). At this point nothing happens. At all. If I ctl-alt-esc into the debugger, I am told the processor is in default_halt + 0x01. If I tell the debugger to continue, it actually tries to call boot(). Boot gets confused because I have a function registered with at_shutdown which tsleep()s. This last part has nothing to do with SMP as this confusion exists in the UP kernel as well. I will be very happy to supply whatever information is necessary, including access to the errant machine. BTW, the handbook is out of date on serial console, last I checked. I wanted to add that for a long time... Simon From owner-freebsd-smp Mon Jul 21 21:45:56 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA05753 for smp-outgoing; Mon, 21 Jul 1997 21:45:56 -0700 (PDT) Received: from nirvana.genesyslab.com (nirvana.genesyslab.com [204.94.142.10]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA05748 for ; Mon, 21 Jul 1997 21:45:53 -0700 (PDT) Received: from dao.genesyslab.com (Dao.genesyslab.com [204.94.142.146]) by nirvana.genesyslab.com (Guinness/Extra Stout) with ESMTP id VAA21632 for ; Mon, 21 Jul 1997 21:50:37 -0700 (PDT) Received: (from dk@localhost) by dao.genesyslab.com (8.8.5/8.7.3) id VAA00927; Mon, 21 Jul 1997 21:45:50 -0700 (PDT) Message-ID: <19970721214550.09752@genesyslab.com> Date: Mon, 21 Jul 1997 21:45:50 -0700 From: Dmitry Kohmanyuk To: freebsd-smp@freebsd.org Subject: smp kernel for 2.2-stable? Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69e X-OS-of-the-Day: FreeBSD 2.2-STABLE X-Test: (browser detected) X-Home-URL: http://zen.genesyslab.com/~dk/ Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk is there any SMP kernel version which works with 2.2-branch? Or do I have to jump to 3.0-current? I have SMP kernel dated around last December (the smp.todo file is 1.3, dated 1996/11/26) used here in production on NIS/DNS server. It is running some old snapshot (about the same vintage). I also have SMP machine running 2.2.1-RELEASE (standard UP kernel). I am thinking about putting some SMP kernel on them, which would be more stable; what would you recommend?? p.s. please Cc: replies to me as I am not yet subscribed (I am doing this now). Thanks! From owner-freebsd-smp Tue Jul 22 04:58:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id EAA24011 for smp-outgoing; Tue, 22 Jul 1997 04:58:23 -0700 (PDT) Received: from weenix.guru.org (weenix.guru.org [198.82.200.65]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id EAA24006 for ; Tue, 22 Jul 1997 04:58:19 -0700 (PDT) Received: (from kmitch@localhost) by weenix.guru.org (8.8.6/8.8.5) id HAA03374; Tue, 22 Jul 1997 07:58:15 -0400 (EDT) Date: Tue, 22 Jul 1997 07:58:15 -0400 (EDT) From: Keith Mitchell Message-Id: <199707221158.HAA03374@weenix.guru.org> To: smp@csn.net (Steve Passe) CC: smp@freebsd.org, kmitch@guru.org Subject: Re: HEADS UP: major timer/APIC code changes. In-Reply-To: X-Newsreader: TIN [UNIX 1.3 unoff BETA release 970124] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 8bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > Please send me reports for ANY new kernel builds, working or not. I need > to know ASAP whether I've broken anything. This is one of those changes > that may behave differently on various hardware. I keep getting system freezes with this (or some other) new code. My -current from about two weeks ago works fine, but under the this and the later stuff my kernel seems to just freeze. At first, I thought it might be the the new VM changes, but the fix didn't seem to do anything. I am using an ASUS P65UP5 with Dual P5s. The freezes appear pretty random too. Sometimes as soon as I log on; sometimes after its been up a while. It still could be the new VM code (I guess), but I thought I would check to see if it could be SMP first. -- Keith Mitchell Head Administrator: acm.vt.edu Email: kmitch@weenix.guru.org PGP key available upon request http://weenix.guru.org/~kmitch From owner-freebsd-smp Tue Jul 22 07:05:07 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA29915 for smp-outgoing; Tue, 22 Jul 1997 07:05:07 -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 HAA29905 for ; Tue, 22 Jul 1997 07:05:01 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.6/8.8.5) id JAA00447; Tue, 22 Jul 1997 09:04:43 -0500 (EST) From: "John S. Dyson" Message-Id: <199707221404.JAA00447@dyson.iquest.net> Subject: Re: HEADS UP: major timer/APIC code changes. In-Reply-To: <199707221158.HAA03374@weenix.guru.org> from Keith Mitchell at "Jul 22, 97 07:58:15 am" To: kmitch@weenix.guru.org (Keith Mitchell) Date: Tue, 22 Jul 1997 09:04:43 -0500 (EST) Cc: smp@csn.net, smp@FreeBSD.ORG, kmitch@guru.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-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > Please send me reports for ANY new kernel builds, working or not. I need > > to know ASAP whether I've broken anything. This is one of those changes > > that may behave differently on various hardware. > > I keep getting system freezes with this (or some other) new code. My -current > from about two weeks ago works fine, but under the this and the later stuff > my kernel seems to just freeze. At first, I thought it might be the the new > VM changes, but the fix didn't seem to do anything. I am using an ASUS P65UP5 > with Dual P5s. The freezes appear pretty random too. Sometimes as soon > as I log on; sometimes after its been up a while. It still could be the new > VM code (I guess), but I thought I would check to see if it could be > SMP first. > It could still be the VM code, but it would be good to understand how your system is failing. For example, when your system "freezes" are you running X windows? Can you still ping or telnet into it? John From owner-freebsd-smp Tue Jul 22 07:22:59 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA00972 for smp-outgoing; Tue, 22 Jul 1997 07:22:59 -0700 (PDT) Received: from weenix.guru.org (weenix.guru.org [198.82.200.65]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id HAA00967 for ; Tue, 22 Jul 1997 07:22:53 -0700 (PDT) Received: (from kmitch@localhost) by weenix.guru.org (8.8.6/8.8.5) id KAA07745; Tue, 22 Jul 1997 10:22:42 -0400 (EDT) From: Keith Mitchell Message-Id: <199707221422.KAA07745@weenix.guru.org> Subject: Re: HEADS UP: major timer/APIC code changes. In-Reply-To: <199707221404.JAA00447@dyson.iquest.net> from "John S. Dyson" at "Jul 22, 97 09:04:43 am" To: toor@dyson.iquest.net (John S. Dyson) Date: Tue, 22 Jul 1997 10:22:42 -0400 (EDT) Cc: smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL30 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > It could still be the VM code, but it would be good to understand how your > system is failing. For example, when your system "freezes" are you running > X windows? Can you still ping or telnet into it? Yes, I am running X-windows. I can't ping telnet or anything to the machine. I can't even log in via the serial console attached to the machine. The activity light on the network cards are still active though. Sometimes it occurs immediately after I log in and start typing in the xterm window. Other times it locks up while I am out and not using the computer. Twice I was using nfs at the time of the lockups. -- Keith Mitchell Head Administrator: acm.vt.edu Email: kmitch@weenix.guru.org PGP key available upon request http://weenix.guru.org/~kmitch From owner-freebsd-smp Tue Jul 22 09:08:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA07083 for smp-outgoing; Tue, 22 Jul 1997 09:08:09 -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 JAA07065 for ; Tue, 22 Jul 1997 09:08:03 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id KAA10262; Tue, 22 Jul 1997 10:07:49 -0600 (MDT) Message-Id: <199707221607.KAA10262@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Keith Mitchell cc: smp@freebsd.org, kmitch@guru.org Subject: Re: HEADS UP: major timer/APIC code changes. In-reply-to: Your message of "Tue, 22 Jul 1997 07:58:15 EDT." <199707221158.HAA03374@weenix.guru.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 22 Jul 1997 10:07:49 -0600 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > I keep getting system freezes with this (or some other) new code. My -current > from about two weeks ago works fine, but under the this and the later stuff > my kernel seems to just freeze. > ... I have received several reports from people stating that the cure to their problems with the recent code has been fixed by a cvsup & new "make world". My own experience was that a dual P5 system with a world as new as 7-13 would panic with the 7-16 thru 7-18 code. If you have any problems with the new SMP code, you must cvsup the entire source tree (ie NOT just sys) and "make world". After making world, rebuild a new SMP kernel with it b4 rebooting. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Jul 22 10:31:21 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA13103 for smp-outgoing; Tue, 22 Jul 1997 10:31:21 -0700 (PDT) Received: from weenix.guru.org (weenix.guru.org [198.82.200.65]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA13092 for ; Tue, 22 Jul 1997 10:31:16 -0700 (PDT) Received: (from kmitch@localhost) by weenix.guru.org (8.8.6/8.8.5) id NAA13782; Tue, 22 Jul 1997 13:31:12 -0400 (EDT) From: Keith Mitchell Message-Id: <199707221731.NAA13782@weenix.guru.org> Subject: Re: HEADS UP: major timer/APIC code changes. In-Reply-To: <199707221607.KAA10262@Ilsa.StevesCafe.com> from Steve Passe at "Jul 22, 97 10:07:49 am" To: smp@csn.net (Steve Passe) Date: Tue, 22 Jul 1997 13:31:12 -0400 (EDT) Cc: smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL30 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > I have received several reports from people stating that the cure > to their problems with the recent code has been fixed by a cvsup & > new "make world". My own experience was that a dual P5 system with > a world as new as 7-13 would panic with the 7-16 thru 7-18 code. > If you have any problems with the new SMP code, you must cvsup > the entire source tree (ie NOT just sys) and "make world". After > making world, rebuild a new SMP kernel with it b4 rebooting. I did that when I first upgraded on the 19th. I then went to 7/21 -sys only. I will try going with a 7/22 world and kernel and see if that helps. -- Keith Mitchell Head Administrator: acm.vt.edu Email: kmitch@weenix.guru.org PGP key available upon request http://weenix.guru.org/~kmitch Address and URL (c) 1997 Keith Mitchell - All Rights Reserved Unauthorized use or duplication prohibited From owner-freebsd-smp Tue Jul 22 12:25:11 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA21267 for smp-outgoing; Tue, 22 Jul 1997 12:25:11 -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 MAA21249 for ; Tue, 22 Jul 1997 12:24:55 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id NAA11030 for ; Tue, 22 Jul 1997 13:24:29 -0600 (MDT) Message-Id: <199707221924.NAA11030@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: Please make my life easier! Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 22 Jul 1997 13:24:29 -0600 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, I'm doing way too much hand-holding and not near enough coding these days. I get about 5 SMP related email problems in my private e-mail for every one that goes to the official freeBSD SMP list. - Send ALL problem reports to smp@freebsd.org. DON'T cc me, I follow the smp list closely and don't need all the dups. After the initial report, I will take it offline if appropriate. - People need to jump in and start helping to diagnose failing systems. Its not rocket science, I know there are others out there capable of taking some of this load off my shoulders. - We need a method to organize information about the user machine base. It would be nice if people could be assigned a "contact number" by which all the info about their machines could be stored. This would include their mptable and a hardware description file(s). People send me an mptable, then 3 months later encounter a problem, but its heck to go back and find the info easily. The best solution would place this info on the web page so that others could access it when necessary. Any suggests/help on this item would be appreciated! -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Jul 22 14:28:08 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA29719 for smp-outgoing; Tue, 22 Jul 1997 14:28:08 -0700 (PDT) Received: from sci.brooklyn.cuny.edu (dayton@louis.sci.brooklyn.cuny.edu [146.245.1.7]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA29674; Tue, 22 Jul 1997 14:27:59 -0700 (PDT) Received: (from dayton@localhost) by sci.brooklyn.cuny.edu (8.8.5/8.8.4) id RAA02030; Tue, 22 Jul 1997 17:27:52 -0400 (EDT) Date: Tue, 22 Jul 1997 17:27:52 -0400 (EDT) Message-Id: <199707222127.RAA02030@sci.brooklyn.cuny.edu> From: Dayton Clark To: questions@freebsd.org, smp@freebsd.org Subject: Goliath and 2.2.2 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Folks, I'm sending this to questions and smp because I suspect that the smp folks may be in a better position to answer the question. To those who get the message twice, I apologize. My system: Vendor: SAG Motherboard: AMI Goliath Processors: 4 PPro 200 Video: Number9 Image 128 2e SCSI: Adaptec 2940 RAM: 128M Solaris and WindowsNT run fine on this machine, in multi-processor mode. FreeBSD 2.2.1 ran OK except for X Windows which made the machine hang. I've installed FreeBSD 2.2.2 and I get many many mysterious bus errors and segmentation faults. The system as a whole survives, but many programs dump core. Anyone know what's happening? Cache inconsistency seems like a likely culprit. I disabled the L1 cache from the setup, and got the same result (but much slower 8-). I hope to try the SMP FreeBSD. Perhaps I should jump to some variant of Fbsd 3 and see if the problem disappears. Which variant would you suggest? Any other suggestions? thanks dayton Dayton Clark CIS Department dayton@brooklyn.cuny.edu Brooklyn College/CUNY 1-718-951-4811 Brooklyn, New York 11210 1-718-951-4842 (fax) From owner-freebsd-smp Wed Jul 23 07:50:08 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA21668 for smp-outgoing; Wed, 23 Jul 1997 07:50:08 -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 HAA21658 for ; Wed, 23 Jul 1997 07:50:02 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id IAA28518; Wed, 23 Jul 1997 08:49:44 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id IAA28335; Wed, 23 Jul 1997 08:49:36 -0600 Date: Wed, 23 Jul 1997 08:49:36 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199707231449.IAA28335@fast.cs.utah.edu> To: freebsd-smp@FreeBSD.org Subject: Lots 'o PCI slots Sender: owner-freebsd-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Anyone can find 5 PCI slots. Here is my list of dual motherboards with 7+ (6 or more plus integrated PCI too) PCI slots. Note that I'm not intending to promote www.interpromicro.com; prices are just given as a convenience for comparison. http://www.supermicro.com/products/motherboards/pentium%20pro/p6dnh.htm Dual Pentium Pro, I2O 8 PCI $669 from interpromicro http://www.micronics.com/products/m6dpi.html Dual Pentium Pro, I2O 7 PCI with Adaptec 7880 $719 from interpromicro http://www.ami.com/motherboards/s758spec.html Dual Pentium II 6 PCI with dual NCR875 controllers, 2MB ATI video (Not yet shipping) There are also numerous dual SBC computers that will work with a 7+ slot passive backplane. Additionally, many of the higher-end servers have 6 or more PCI slots. The first two are I2O. The dedicated i960 processor is intended to improve PCI performance by offloading tasks from the main CPU. But I doubt FreeBSD will support I2O within a year or two. The AMI board looks nice, except (forgive me if I'm wrong) that the Symbios875's aren't supported in FreeBSD. They will work, but just like an 825. (Local script memory isn't used, and they won't negotiate ultra speeds). It is also likely to be very $$$ because it is AMI. Kevin From owner-freebsd-smp Wed Jul 23 08:37:35 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id IAA25051 for smp-outgoing; Wed, 23 Jul 1997 08:37:35 -0700 (PDT) Received: from time.cdrom.com (root@time.cdrom.com [204.216.27.226]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id IAA25045 for ; Wed, 23 Jul 1997 08:37:33 -0700 (PDT) Received: from time.cdrom.com (jkh@localhost.cdrom.com [127.0.0.1]) by time.cdrom.com (8.8.6/8.6.9) with ESMTP id IAA04682; Wed, 23 Jul 1997 08:36:08 -0700 (PDT) To: vanmaren@fast.cs.utah.edu (Kevin Van Maren) cc: freebsd-smp@FreeBSD.ORG Subject: Re: Lots 'o PCI slots In-reply-to: Your message of "Wed, 23 Jul 1997 08:49:36 MDT." <199707231449.IAA28335@fast.cs.utah.edu> Date: Wed, 23 Jul 1997 08:36:08 -0700 Message-ID: <4679.869672168@time.cdrom.com> From: "Jordan K. Hubbard" Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > The AMI board looks nice, except (forgive me if I'm wrong) that the > Symbios875's aren't supported in FreeBSD. They will work, but just That'd be news to freefall.cdrom.com, currently a user of a Symbios 875 based board. :-) Jordan From owner-freebsd-smp Wed Jul 23 08:50:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id IAA25908 for smp-outgoing; Wed, 23 Jul 1997 08:50:49 -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 IAA25897 for ; Wed, 23 Jul 1997 08:50:46 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id JAA01289; Wed, 23 Jul 1997 09:50:35 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id JAA09697; Wed, 23 Jul 1997 09:50:35 -0600 Date: Wed, 23 Jul 1997 09:50:35 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199707231550.JAA09697@fast.cs.utah.edu> To: jkh@time.cdrom.com Subject: Re: Lots 'o PCI slots Cc: freebsd-smp@FreeBSD.ORG Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > The AMI board looks nice, except (forgive me if I'm wrong) that the > > Symbios875's aren't supported in FreeBSD. They will work, but just > > That'd be news to freefall.cdrom.com, currently a user of a Symbios 875 > based board. :-) You cut out the part of my message where I explained `not supported': ] Symbios875's aren't supported in FreeBSD. They will work, but just ] like an 825. (Local script memory isn't used, and they won't negotiate ] ultra speeds). It is also likely to be very $$$ because it is AMI. I have a 875 at home, and under 2.2.x it won't negotaite ultra with my ultra scsi drive. If 2.2.x now supports this (or the local script memory), I'll gladly cvsup the latest version. If it doesn't, then "just like an 825" is correct :( Kevin From owner-freebsd-smp Wed Jul 23 11:26:41 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA05201 for smp-outgoing; Wed, 23 Jul 1997 11:26:41 -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 LAA05192 for ; Wed, 23 Jul 1997 11:26:36 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id LAA15994; Wed, 23 Jul 1997 11:22:46 -0700 From: Terry Lambert Message-Id: <199707231822.LAA15994@phaeton.artisoft.com> Subject: Re: Lots 'o PCI slots To: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Date: Wed, 23 Jul 1997 11:22:46 -0700 (MST) Cc: freebsd-smp@FreeBSD.ORG In-Reply-To: <199707231449.IAA28335@fast.cs.utah.edu> from "Kevin Van Maren" at Jul 23, 97 08:49:36 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-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > The first two are I2O. The dedicated i960 processor is intended to > improve PCI performance by offloading tasks from the main CPU. But > I doubt FreeBSD will support I2O within a year or two. If you have been following the -current list, you will have noticed a lot of PCI driver messages. Vinay Bannai has got the i960 probed and attached, and was last seen asking after gcc for the i960 (GCC does support the i960, in case you were wondering). So I kind of doubt it will take "a year or two". ;-). What I think is more interesting is if you could run the i960 as the default processor -- an i960 port of FreeBSD. 8-). 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 Wed Jul 23 11:59:16 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA07080 for smp-outgoing; Wed, 23 Jul 1997 11:59:16 -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 LAA07013 for ; Wed, 23 Jul 1997 11:57:53 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id MAA16991; Wed, 23 Jul 1997 12:56:12 -0600 (MDT) Message-Id: <199707231856.MAA16991@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Terry Lambert cc: vanmaren@fast.cs.utah.edu (Kevin Van Maren), freebsd-smp@FreeBSD.ORG Subject: Re: Lots 'o PCI slots In-reply-to: Your message of "Wed, 23 Jul 1997 11:22:46 PDT." <199707231822.LAA15994@phaeton.artisoft.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 23 Jul 1997 12:56:11 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > Vinay Bannai has got the i960 probed and > attached, and was last seen asking after gcc for the i960 (GCC > does support the i960, in case you were wondering). > > So I kind of doubt it will take "a year or two". ;-). > > > What I think is more interesting is if you could run the i960 > as the default processor -- an i960 port of FreeBSD. 8-). I'm ignorant of the I2O/i960 hardware setup, but I believe the i960 has an onboard APIC interrupt controller. Do such motherboards require the use of the APICs to handle the concept of driver offloading? If so, as the person writing most of the SMP APIC code, it might be a good idea for me to find out more so that I can consider this issue when making arch. decisions... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Jul 23 14:02:22 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA14250 for smp-outgoing; Wed, 23 Jul 1997 14:02:22 -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 OAA14228 for ; Wed, 23 Jul 1997 14:02:05 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id PAA17477 for ; Wed, 23 Jul 1997 15:02:00 -0600 (MDT) Message-Id: <199707232102.PAA17477@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: HEADS UP: new simple_lock code Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 23 Jul 1997 15:02:00 -0600 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, I just committed code to impliment simple locks for SMP. I created a simple lock (imen_lock) to lock the critical region of "IO APIC & apic_imen" accesses. I moved the code that accesses these critical regions *outside* of the GL (mp_lock). It seems to work!!! YMMV, be aware that I will be mucking with other sections of the ISRs next, so you may want to avoid updates for any "important" systems for awhile... (ie its time to break things). But otherwise, please test this code if you can. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Jul 23 14:10:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA14724 for smp-outgoing; Wed, 23 Jul 1997 14:10:34 -0700 (PDT) Received: from hutcs.cs.hut.fi (apl@hutcs.cs.hut.fi [130.233.192.6]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA14714 for ; Wed, 23 Jul 1997 14:10:30 -0700 (PDT) Received: (from apl@localhost) by hutcs.cs.hut.fi (8.8.5/8.7.3) id AAA16750; Thu, 24 Jul 1997 00:10:26 +0300 (EET DST) Message-ID: <19970724001025.34097@hutcs.cs.hut.fi> Date: Thu, 24 Jul 1997 00:10:25 +0300 From: Antti-Pekka Liedes To: freebsd-smp@freebsd.org Subject: AHA-3940U on Tomcat IIID, asking for trouble? Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.79 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Are there any experiences of how well AHA-3940U (bridged PCI card) works on Tomcat IIID running freebsd-smp? I read the FBSD-SMP pages concerning bridged cards and it doesn't look promising. -- Antti-Pekka Liedes * apl@IRC * In two hells there's JMT 6 B 406 * apl@iki.fi * one hell too many 02150 ESPOO * apl@apocalypse.tky.hut.fi * - Lucifer +358 - 9 - 468 3121 * +358 - 40 - 5873 593 * (in God's Army) From owner-freebsd-smp Wed Jul 23 14:24:41 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA15516 for smp-outgoing; Wed, 23 Jul 1997 14:24:41 -0700 (PDT) Received: from mx.serv.net (mx.serv.net [205.153.153.234]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA15509 for ; Wed, 23 Jul 1997 14:24:38 -0700 (PDT) Received: from MindBender.serv.net (root@mindbender.serv.net [205.153.153.98]) by mx.serv.net (8.8.5/8.8.5) with ESMTP id OAA04261; Wed, 23 Jul 1997 14:24:41 -0700 (PDT) Received: from localhost.HeadCandy.com (michaelv@localhost.HeadCandy.com [127.0.0.1]) by MindBender.serv.net (8.7.5/8.7.3) with SMTP id OAA08649; Wed, 23 Jul 1997 14:23:06 -0700 (PDT) Message-Id: <199707232123.OAA08649@MindBender.serv.net> X-Authentication-Warning: MindBender.serv.net: Host michaelv@localhost.HeadCandy.com [127.0.0.1] didn't use HELO protocol To: Antti-Pekka Liedes cc: freebsd-smp@freebsd.org Subject: Re: AHA-3940U on Tomcat IIID, asking for trouble? In-reply-to: Your message of Thu, 24 Jul 97 00:10:25 +0300. <19970724001025.34097@hutcs.cs.hut.fi> Date: Wed, 23 Jul 1997 14:22:55 -0700 From: "Michael L. VanLoon -- HeadCandy.com" Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >Are there any experiences of how well AHA-3940U (bridged PCI card) works >on Tomcat IIID running freebsd-smp? I read the FBSD-SMP pages concerning >bridged cards and it doesn't look promising. I have a 2940UW and a 3940UW (three SCSI channels total) running in a NetBSD box (Asus P55TP4N with Cyrix 6x86 P166+, if it matters). It works extremely well. Since the NetBSD and FreeBSD ahc drivers are derived from the same code base, there's no reason to believe it shouldn't work (unless, of course, your motherboard doesn't support bridges). ----------------------------------------------------------------------------- Michael L. VanLoon michaelv@MindBender.serv.net --< Free your mind and your machine -- NetBSD free un*x >-- NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... NetBSD ports in progress: PICA, others... ----------------------------------------------------------------------------- From owner-freebsd-smp Wed Jul 23 14:31:46 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA16112 for smp-outgoing; Wed, 23 Jul 1997 14:31:46 -0700 (PDT) Received: from daffy-duck.cs.washington.edu (daffy-duck.cs.washington.edu [128.95.2.11]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA16106 for ; Wed, 23 Jul 1997 14:31:44 -0700 (PDT) From: mef@cs.washington.edu Received: (mef@localhost) by daffy-duck.cs.washington.edu (8.8.5+CS/7.2ws+) id OAA26614; Wed, 23 Jul 1997 14:30:24 -0700 (PDT) Date: Wed, 23 Jul 1997 14:30:24 -0700 (PDT) Message-Id: <199707232130.OAA26614@daffy-duck.cs.washington.edu> To: terry@lambert.org CC: vanmaren@fast.cs.utah.edu, freebsd-smp@FreeBSD.ORG In-reply-to: <199707231822.LAA15994@phaeton.artisoft.com> (message from Terry Lambert on Wed, 23 Jul 1997 11:22:46 -0700 (MST)) Subject: Re: Lots 'o PCI slots Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk From: Terry Lambert Date: Wed, 23 Jul 1997 11:22:46 -0700 (MST) Vinay Bannai has got the i960 probed and attached, and was last seen asking after gcc for the i960 (GCC does support the i960, in case you were wondering). So I kind of doubt it will take "a year or two". ;-). It all depends on what Vinay plans to do with the i960 processor. If he intends to communicate to it via the I2O message protocol, then I doubt that FreeBSD will be able to publically distribute that source code for at least another year. That is, for some stupid reason the I2O stuff is not in the public domain and there are (ahem) legal restrictions that are supposed to prevent one from reverse engineering it. To become an I2O member you have to play a pretty penny ($2000). For $250 you can get an evaluation copy of the I2O specification, which runs out after 90days. However, you are not supposed to be developing I2O software without being a member. This makes it kinda hard to distribute any of that work to none I2O members. :( Marc From owner-freebsd-smp Wed Jul 23 14:42:07 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA16739 for smp-outgoing; Wed, 23 Jul 1997 14:42:07 -0700 (PDT) Received: from hutcs.cs.hut.fi (apl@hutcs.cs.hut.fi [130.233.192.6]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA16725 for ; Wed, 23 Jul 1997 14:42:00 -0700 (PDT) Received: (from apl@localhost) by hutcs.cs.hut.fi (8.8.5/8.7.3) id AAA16941; Thu, 24 Jul 1997 00:40:06 +0300 (EET DST) Message-ID: <19970724004005.12787@hutcs.cs.hut.fi> Date: Thu, 24 Jul 1997 00:40:05 +0300 From: Antti-Pekka Liedes To: "Michael L. VanLoon -- HeadCandy.com" Cc: freebsd-smp@freebsd.org Subject: Re: AHA-3940U on Tomcat IIID, asking for trouble? References: <19970724001025.34097@hutcs.cs.hut.fi> <199707232123.OAA08649@MindBender.serv.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.79 In-Reply-To: <199707232123.OAA08649@MindBender.serv.net>; from Michael L. VanLoon -- HeadCandy.com on Wed, Jul 23, 1997 at 02:22:55PM -0700 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Wed, Jul 23, 1997 at 02:22:55PM -0700, Michael L. VanLoon -- HeadCandy.com wrote: > > >Are there any experiences of how well AHA-3940U (bridged PCI card) works > >on Tomcat IIID running freebsd-smp? I read the FBSD-SMP pages concerning > >bridged cards and it doesn't look promising. > > I have a 2940UW and a 3940UW (three SCSI channels total) running in a > NetBSD box (Asus P55TP4N with Cyrix 6x86 P166+, if it matters). It > works extremely well. > > Since the NetBSD and FreeBSD ahc drivers are derived from the same > code base, there's no reason to believe it shouldn't work (unless, of > course, your motherboard doesn't support bridges). > This exactly is my concern, how will Tomcat IIID motherboard react to 3940? Being a generally good quality board I think it should work well with bridging, but does anyone have any real experience? Too bad I still don't have two multiprocessing capable CPUs, I have to wait until the end of this month to get some. > ----------------------------------------------------------------------------- > Michael L. VanLoon michaelv@MindBender.serv.net > --< Free your mind and your machine -- NetBSD free un*x >-- > NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, > Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... > NetBSD ports in progress: PICA, others... > ----------------------------------------------------------------------------- -- Antti-Pekka Liedes * apl@IRC * In two hells there's JMT 6 B 406 * apl@iki.fi * one hell too many 02150 ESPOO * apl@apocalypse.tky.hut.fi * - Lucifer +358 - 9 - 468 3121 * +358 - 40 - 5873 593 * (in God's Army) From owner-freebsd-smp Wed Jul 23 15:16:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA18905 for smp-outgoing; Wed, 23 Jul 1997 15:16:23 -0700 (PDT) Received: from cenotaph.snafu.de (gw-deadnet.snafu.de [194.121.229.33]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id PAA18898 for ; Wed, 23 Jul 1997 15:16:16 -0700 (PDT) Received: by cenotaph.snafu.de from deadline.snafu.de using smtp id m0wr9hK-000KE4C; Thu, 24 Jul 1997 00:15:54 +0200 (CEST) (Smail-3.2 1996-Jul-4 #1) Received: by deadline.snafu.de id m0wr9hJ-00049fC; Thu, 24 Jul 1997 00:15:53 +0200 (CEST) (Smail-3.2 1996-Jul-4 #1) Message-Id: From: root@deadline.snafu.de (Andreas S. Wetzel) Subject: 3.0-SMP + COMPAT_LINUX ? To: smp@freebsd.org Date: Thu, 24 Jul 1997 00:15:53 +0200 (CEST) Organization: A world stranger than you have ever imagined. X-Mailer: ELM [version 2.4ME+ PL13] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi! --- I recently tried to run some Linux binaries on top of 3.0-current SMP, but when I start them the machine hangs forever. Is this "supposed" to be so at the current stage of development or doesn't anybody notice until now? Regards, Mickey -- (__) (@@) Andreas S. Wetzel Mail: mickey@deadline.snafu.de /-------\/ Utrechter Strasse 41 Web: http://cenotaph.snafu.de/ / | || 13347 Berlin Fon: <+4930> 456 066 90 * ||----|| Germany Fax: <+4930> 456 066 91/92 ~~ ~~ From owner-freebsd-smp Wed Jul 23 15:24:43 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA19362 for smp-outgoing; Wed, 23 Jul 1997 15:24:43 -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 PAA19343 for ; Wed, 23 Jul 1997 15:24:29 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id QAA17793 for ; Wed, 23 Jul 1997 16:24:17 -0600 (MDT) Message-Id: <199707232224.QAA17793@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: user questions Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 23 Jul 1997 16:24:16 -0600 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk ------- Forwarded Message Date: Wed, 23 Jul 97 14:49:09 PDT Subject: SMP, state of -current? From: Troy Curtiss Message-Id: In-Reply-To: <199707232125.OAA19082@freefall.freebsd.org> Hi Steve, I see you've been working hot & heavy on the SMP FreeBSD stuff. I may be in a position to buy an MP motherboard here at work, which would you recommend? Secondly, how far along is the SMP in the way of process sharing/threading/etc... If I write a program that parallel tasks (ie... brute forcing password keyspaces), can I put a CPU on each task/thread/process to effectively double my speed (almost)? Just curious, - -Troy - -- /-----------------------------------------------------------\ | Troy Curtiss, HW/SW Engineer | Email: troyc@merix.com | | Merix Corporation, CL-302 | Phone: (970) 203-6643 | | Loveland, CO 80537 | Fax : (970) 203-6610 | \-----------------------------------------------------------/ ------- End of Forwarded Message From owner-freebsd-smp Wed Jul 23 15:24:41 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA19361 for smp-outgoing; Wed, 23 Jul 1997 15:24:41 -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 PAA19347 for ; Wed, 23 Jul 1997 15:24:33 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id PAA16378; Wed, 23 Jul 1997 15:20:37 -0700 From: Terry Lambert Message-Id: <199707232220.PAA16378@phaeton.artisoft.com> Subject: Re: Lots 'o PCI slots To: mef@cs.washington.edu Date: Wed, 23 Jul 1997 15:20:36 -0700 (MST) Cc: terry@lambert.org, vanmaren@fast.cs.utah.edu, freebsd-smp@FreeBSD.ORG In-Reply-To: <199707232130.OAA26614@daffy-duck.cs.washington.edu> from "mef@cs.washington.edu" at Jul 23, 97 02:30:24 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > It all depends on what Vinay plans to do with the i960 processor. If > he intends to communicate to it via the I2O message protocol, then I > doubt that FreeBSD will be able to publically distribute that source > code for at least another year. That is, for some stupid reason the > I2O stuff is not in the public domain and there are (ahem) legal > restrictions that are supposed to prevent one from reverse engineering > it. > > To become an I2O member you have to play a pretty penny ($2000). For > $250 you can get an evaluation copy of the I2O specification, which > runs out after 90days. However, you are not supposed to be developing > I2O software without being a member. This makes it kinda hard to > distribute any of that work to none I2O members. :( I would think that would be relatively easy to prosecute as racketeering (under RICO), restraint of trade, and as a monopoly (under Sherman). I would also think that if they started making noises about it, you would quickly see the EFF or a similar organization filing the suit. I don't think it's possible for it to be illegal to write code for hardware you own. 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 Wed Jul 23 15:46:01 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA20497 for smp-outgoing; Wed, 23 Jul 1997 15:46:01 -0700 (PDT) Received: from george.arc.nasa.gov (george.arc.nasa.gov [128.102.194.142]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id PAA20491 for ; Wed, 23 Jul 1997 15:45:58 -0700 (PDT) From: lamaster@george.arc.nasa.gov Received: by george.arc.nasa.gov (8.8.6/1.35) id PAA29632; Wed, 23 Jul 1997 15:44:11 -0700 (PDT) Date: Wed, 23 Jul 1997 15:44:11 -0700 (PDT) Message-Id: <199707232244.PAA29632@george.arc.nasa.gov> To: freebsd-smp@FreeBSD.ORG Subject: Re: Lots 'o PCI slots Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Terry Lambert wrote: > I don't think it's possible for it to be illegal to write code > for hardware you own. I always thought that reverse-engineering was generally protected in the U.S. (but not in all countries). Increasingly, however, I am noticing shrink-wrapped licenses that say something to the effect that by opening the package, I am agreeing not to reverse- engineer anything inside the package. Is this legally binding in the U.S. and/or other countries? -Hugh LaMaster Hugh LaMaster, M/S 258-5, ASCII Email: hlamaster@mail.arc.nasa.gov NASA Ames Research Center Or: lamaster@nas.nasa.gov Moffett Field, CA 94035-1000 No Junkmail: USC 18 section 2701 Phone: 415/604-1056 Disclaimer: Unofficial, personal *opinion*. From owner-freebsd-smp Wed Jul 23 16:36:37 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id QAA23054 for smp-outgoing; Wed, 23 Jul 1997 16:36:37 -0700 (PDT) Received: from weenix.guru.org (weenix.guru.org [198.82.200.65]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id QAA23045 for ; Wed, 23 Jul 1997 16:36:33 -0700 (PDT) Received: (from kmitch@localhost) by weenix.guru.org (8.8.6/8.8.5) id TAA15727; Wed, 23 Jul 1997 19:36:17 -0400 (EDT) Date: Wed, 23 Jul 1997 19:36:17 -0400 (EDT) From: Keith Mitchell Message-Id: <199707232336.TAA15727@weenix.guru.org> To: apl@hutcs.cs.hut.fi (Antti-Pekka Liedes) CC: smp@freebsd.org Subject: Re: AHA-3940U on Tomcat IIID, asking for trouble? X-Newsreader: TIN [UNIX 1.3 unoff BETA release 970124] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 8bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > Are there any experiences of how well AHA-3940U (bridged PCI card) works > on Tomcat IIID running freebsd-smp? I read the FBSD-SMP pages concerning > bridged cards and it doesn't look promising. It will work, but it will require kernel patches. I can't speak for the 4.0 BIOS, but earlier BIOSes did not build the MP table to include cards with PCI-PCI bridges on them. If you run the MPtable program and it lists more than one PCI bus you are probably fine though. -- Keith Mitchell Head Administrator: acm.vt.edu Email: kmitch@weenix.guru.org PGP key available upon request http://weenix.guru.org/~kmitch Address and URL (c) 1997 Keith Mitchell - All Rights Reserved Unauthorized use or duplication prohibited From owner-freebsd-smp Wed Jul 23 17:23:00 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id RAA25907 for smp-outgoing; Wed, 23 Jul 1997 17:23:00 -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 RAA25901 for ; Wed, 23 Jul 1997 17:22:57 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id RAA16606; Wed, 23 Jul 1997 17:18:58 -0700 From: Terry Lambert Message-Id: <199707240018.RAA16606@phaeton.artisoft.com> Subject: Re: Lots 'o PCI slots To: lamaster@george.arc.nasa.gov Date: Wed, 23 Jul 1997 17:18:58 -0700 (MST) Cc: freebsd-smp@FreeBSD.ORG In-Reply-To: <199707232244.PAA29632@george.arc.nasa.gov> from "lamaster@george.arc.nasa.gov" at Jul 23, 97 03:44:11 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Sean claims I'm giving legal advice; just in case someone thought I was, I'm not, so this statement is my disclaimer. > > I don't think it's possible for it to be illegal to write code > > for hardware you own. > > I always thought that reverse-engineering was generally protected > in the U.S. (but not in all countries). Increasingly, however, > I am noticing shrink-wrapped licenses that say something to the > effect that by opening the package, I am agreeing not to reverse- > engineer anything inside the package. Is this legally binding > in the U.S. and/or other countries? Well, I can tell you legal history without it being advice. The Berne convention specifically excepts reverse engineering for the purposes of documenting an interface. This is sometimes called "the piracy clause". However, it also states that local copyright laws take precedence, so it's possible that this is only legal in Germany, which has local laws allowing reverse engineering an interface. So in Germany, at least, I don't believe it is binding. Historically in the US, there have been judgements against companies for reverse engineering. Specifically, Microsoft sued and won against Stacker for "deep reverse engineering". Even after reading the briefs and the judgement for that case, I still have no idea how "deep reverse engineeering" differs from "plain old reverse engineering". Lotus Developement sued several companies for "infringing their look and feel". Several settled, but one fought (and won). If Lotus had won, I would have expected VisiCalc to sue them for the same thing, and win, citing the Lotus win. This was an apellate court decision, which means it's binding case law for the district served by the court. Apple sued Digital Research over a similar "look and feel" issue, citing the GEM interface's trash can icon as being an infringement on Apple's look and feel. DR backed down (settled out of court, part of which required them to change their icon), but interestingly enough, Atari, which had licensed GEM for the ST, refused to change the icon, and was apparently not harrassed about it. Perhaps one of the most interesting cases is COMPAQ, and then later Phoenix Technologies. They reverse engineered the IBM PC BIOS, and led to the existance of clones. To accomplish this, they used a technique called "clean-rooming". Basically, one group disassembled and documented the BIOS in sufficient detail that it could be rewritten. Then they threw the document over the wall to another group, which wrote the BIOS from the documentation produced by the first group. This was deemed to be legal by a court (obviously, IBM was a bit upset), since the people who wrote the BIOS never saw IBM's code. Now PC BIOS code is a commodity, and you can purchase one from several places. You can even get one in source form from an ad in Dr. Dobb's journal. So clean room reverse engineering is apparently OK, so long as as it isn't "deep"? ...I guess. As far as shrink wrap licenses, I don't believe any have been tested to the apellate level. I know of two non-apellate tests, one in Missouri, and one in Wisconsin. Both found the licenses invalid. Note that finding the license invalid does not necessarily mean that the people they found against were allowed to keep the software. I believe the Missouri suit was against a company which rented software to people, and the Wisconsin suit was against a company which sold used software. I also know that there are at least 5 stores in my area, mostly book stores which sell used books, selling used software with apparent impugnity. >From this, and from apellate level decisions regarding other types of implied contract, I don't think shrink wrap licenses are very enforcible. As implied contracts go, your entry into the contract is conditional upon an act, rather than on your consent to and acknowledgement of the agreement. On a more personal note, "Yes, I own a copy of V Communications 'Sourcer' product, and I use it to disassemble code and document hardware interfaces. Regards, 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 Wed Jul 23 19:18:04 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA00930 for smp-outgoing; Wed, 23 Jul 1997 19:18:04 -0700 (PDT) Received: from mpress.com (mpress.com [208.138.29.130]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id TAA00923 for ; Wed, 23 Jul 1997 19:18:01 -0700 (PDT) Received: (qmail 7224 invoked by uid 100); 24 Jul 1997 02:17:53 -0000 Message-ID: <19970723191751.30899@mpress.com> Date: Wed, 23 Jul 1997 19:17:51 -0700 From: Brian Litzinger To: "Andreas S. Wetzel" Cc: smp@FreeBSD.ORG Subject: Re: 3.0-SMP + COMPAT_LINUX ? References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.76 In-Reply-To: ; from Andreas S. Wetzel on Thu, Jul 24, 1997 at 12:15:53AM +0200 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Jul 24, "Andreas S. Wetzel" wrote: > I recently tried to run some Linux binaries on top of 3.0-current SMP, > but when I start them the machine hangs forever. Is this "supposed" to > be so at the current stage of development or doesn't anybody notice > until now? I run ApplixWare, and the linux version of netscape on an SMP -current machine with linux compatibility compiled into the kernel quite extensively. -- Brian Litzinger brian@mpress.com From owner-freebsd-smp Wed Jul 23 22:26:35 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA09553 for smp-outgoing; Wed, 23 Jul 1997 22:26:35 -0700 (PDT) Received: from scorpion.aracnet.com (root@[205.159.88.23]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA09548 for ; Wed, 23 Jul 1997 22:26:33 -0700 (PDT) Received: from speedy (ppp-a55.aracnet.com [205.238.13.182]) by scorpion.aracnet.com (8.8.5/8.8.5) with ESMTP id WAA22342 for ; Wed, 23 Jul 1997 22:26:16 -0700 Message-ID: <33D6F4BF.3FF44FA1@earthling.net> Date: Wed, 23 Jul 1997 22:22:55 -0800 From: Chris Browning Reply-To: cbrown@aracnet.com X-Mailer: Mozilla 4.01 [en] (WinNT; I) MIME-Version: 1.0 To: freebsd-smp@FreeBSD.ORG Subject: Re: Lots 'o PCI slots X-Priority: 3 (Normal) References: <199707231856.MAA16991@Ilsa.StevesCafe.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk All, I think this is a very interesting topic for FreeBSD to look at. I have been thinking about getting some of the free Un*xes to support I2O. There are many advantages to using it. 1) OS & Adapter drivers are separated. Now, you only need to write an OS driver the interfaces to an abstract devices, like a "SCSI controller" or a "NIC". The device specific stuff is all handled by a OS independent adaptor driver. Therefore, you would only need to have a couple of drivers to support all adapter cards that support I20 can then be used with the OS. 2) The IOP (I/O Platform) handles interrupts & servicing the devices it supports. This reduces the overhead that the OS & host platform must spend servicing I/O request and leaves them more time to do computations, etc. 3) Peer, Device-to-device transfers are possible. A SCSI adapter can copy a file out to the network without the main CPUs even being interrupted. It is all handled by the IOP. Well, that is a short plop about I2O > I'm ignorant of the I2O/i960 hardware setup, but I believe the i960 > has an > onboard APIC interrupt controller. Do such motherboards require the > use > of the APICs to handle the concept of driver offloading? If so, as > the person > writing most of the SMP APIC code, it might be a good idea for me to > find > out more so that I can consider this issue when making arch. > decisions... Well, I just thumbed through the I2O spec, and it made no mention of APICs or even much about how interrupts with the host system are handled. I am guessing that this is implementation specific. If I find out anything more that I can tell you, I'll let you know. Hmm, I wonder if I were to write a I20 driver for Linux or FreeBSD would anyone get upset (I think all Intel employees are I20 SIG members :-) Chris Not speaking for anyone, especially Intel. From owner-freebsd-smp Wed Jul 23 22:48:28 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA10571 for smp-outgoing; Wed, 23 Jul 1997 22:48:28 -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 WAA10564 for ; Wed, 23 Jul 1997 22:48:20 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id XAA19610; Wed, 23 Jul 1997 23:47:58 -0600 (MDT) Message-Id: <199707240547.XAA19610@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: cbrown@aracnet.com cc: freebsd-smp@FreeBSD.ORG Subject: Re: Lots 'o PCI slots In-reply-to: Your message of "Wed, 23 Jul 1997 22:22:55 -0800." <33D6F4BF.3FF44FA1@earthling.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 23 Jul 1997 23:47:58 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, >Well, I just thumbed through the I2O spec, and it made no mention of >APICs or even much about how interrupts with the host system are >handled. I am guessing that this is implementation specific. If I >find out anything more that I can tell you, I'll let you know. Indeed it is implementation specific. However the I2O boards I've noticed so far use the i960 for that "specific hardware". I can picture such boards using the APIC system for interprocessor communications. I'll try to find time to go look at the i960 datasheet... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Jul 23 23:10:57 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id XAA11327 for smp-outgoing; Wed, 23 Jul 1997 23:10:57 -0700 (PDT) Received: from silver.sms.fi (silver.sms.fi [194.111.122.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA11320 for ; Wed, 23 Jul 1997 23:10:53 -0700 (PDT) Received: (from pete@localhost) by silver.sms.fi (8.8.6/8.7.3) id JAA02186; Thu, 24 Jul 1997 09:10:29 +0300 (EEST) Date: Thu, 24 Jul 1997 09:10:29 +0300 (EEST) Message-Id: <199707240610.JAA02186@silver.sms.fi> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit From: Petri Helenius To: lamaster@george.arc.nasa.gov Cc: freebsd-smp@FreeBSD.ORG Subject: Re: Lots 'o PCI slots In-Reply-To: <199707232244.PAA29632@george.arc.nasa.gov> References: <199707232244.PAA29632@george.arc.nasa.gov> X-Mailer: VM 6.22 under 19.15p7 XEmacs Lucid Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk lamaster@george.arc.nasa.gov writes: > > Terry Lambert wrote: > > > I don't think it's possible for it to be illegal to write code > > for hardware you own. > > I always thought that reverse-engineering was generally protected > in the U.S. (but not in all countries). Increasingly, however, > I am noticing shrink-wrapped licenses that say something to the > effect that by opening the package, I am agreeing not to reverse- > engineer anything inside the package. Is this legally binding > in the U.S. and/or other countries? > I only know for sure for Finland but I think it's non-binding in most European countries. Pete From owner-freebsd-smp Wed Jul 23 23:18:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id XAA11560 for smp-outgoing; Wed, 23 Jul 1997 23:18:26 -0700 (PDT) Received: from tweetie-bird.cs.washington.edu (tweetie-bird.cs.washington.edu [128.95.2.46]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA11555 for ; Wed, 23 Jul 1997 23:18:24 -0700 (PDT) From: mef@cs.washington.edu Received: (mef@localhost) by tweetie-bird.cs.washington.edu (8.8.5+CS/7.2ws+) id XAA21456; Wed, 23 Jul 1997 23:18:23 -0700 (PDT) Date: Wed, 23 Jul 1997 23:18:23 -0700 (PDT) Message-Id: <199707240618.XAA21456@tweetie-bird.cs.washington.edu> To: cbrown@aracnet.com CC: freebsd-smp@FreeBSD.ORG In-reply-to: <33D6F4BF.3FF44FA1@earthling.net> (message from Chris Browning on Wed, 23 Jul 1997 22:22:55 -0800) Subject: Re: Lots 'o PCI slots Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Date: Wed, 23 Jul 1997 22:22:55 -0800 From: Chris Browning Hi Chris, All, I think this is a very interesting topic for FreeBSD to look at. I have been thinking about getting some of the free Un*xes to support I2O. There are many advantages to using it. I think it would be great if one could develop a LAN and mass storage I2O OSM for the free un*xes. Right now you can't even get the OSM binaries for SCO or NT without being an I2O member. Consequently, given the license restrictions, I doubt that you will be able to easily just give a loadable kernel module away to non I2O members (i.e., the rest of the world). My guess is that the I2O SIG will keep the I2O specification under lock and key until they can come up with a reasonable standard. Hmm, I wonder if I were to write a I20 driver for Linux or FreeBSD would anyone get upset (I think all Intel employees are I20 SIG members :-) I think that you would find yourself suid by the lawyers of the I2O SIG. Hope you got good lawyers. :) Marc From owner-freebsd-smp Wed Jul 23 23:20:29 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id XAA11669 for smp-outgoing; Wed, 23 Jul 1997 23:20:29 -0700 (PDT) Received: from tweetie-bird.cs.washington.edu (tweetie-bird.cs.washington.edu [128.95.2.46]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA11664 for ; Wed, 23 Jul 1997 23:20:26 -0700 (PDT) From: mef@cs.washington.edu Received: (mef@localhost) by tweetie-bird.cs.washington.edu (8.8.5+CS/7.2ws+) id XAA21455; Wed, 23 Jul 1997 23:20:25 -0700 (PDT) Date: Wed, 23 Jul 1997 23:20:25 -0700 (PDT) Message-Id: <199707240620.XAA21455@tweetie-bird.cs.washington.edu> To: smp@csn.net CC: cbrown@aracnet.com, freebsd-smp@FreeBSD.ORG In-reply-to: <199707240547.XAA19610@Ilsa.StevesCafe.com> (message from Steve Passe on Wed, 23 Jul 1997 23:47:58 -0600) Subject: Re: Lots 'o PCI slots Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk From: Steve Passe Date: Wed, 23 Jul 1997 23:47:58 -0600 >Well, I just thumbed through the I2O spec, and it made no mention of >APICs or even much about how interrupts with the host system are >handled. I am guessing that this is implementation specific. If I >find out anything more that I can tell you, I'll let you know. Indeed it is implementation specific. However the I2O boards I've noticed so far use the i960 for that "specific hardware". I can picture such boards using the APIC system for interprocessor communications. I'll try to find time to go look at the i960 datasheet... It is exactly as you suspect it to be. As far as I know, the i960 has an APIC controller. Marc From owner-freebsd-smp Wed Jul 23 23:33:28 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id XAA12213 for smp-outgoing; Wed, 23 Jul 1997 23:33:28 -0700 (PDT) Received: from critter.dk.tfs.com (critter.phk.freebsd.dk [195.8.133.1]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA12208 for ; Wed, 23 Jul 1997 23:33:24 -0700 (PDT) Received: from critter.dk.tfs.com (localhost [127.0.0.1]) by critter.dk.tfs.com (8.8.6/8.8.5) with ESMTP id IAA01170; Thu, 24 Jul 1997 08:31:56 +0200 (CEST) To: lamaster@george.arc.nasa.gov cc: freebsd-smp@FreeBSD.ORG From: Poul-Henning Kamp Subject: Re: Lots 'o PCI slots In-reply-to: Your message of "Wed, 23 Jul 1997 15:44:11 PDT." <199707232244.PAA29632@george.arc.nasa.gov> Date: Thu, 24 Jul 1997 08:31:55 +0200 Message-ID: <1168.869725915@critter.dk.tfs.com> Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk In message <199707232244.PAA29632@george.arc.nasa.gov>, lamaster@george.arc.nas a.gov writes: > >Terry Lambert wrote: > >> I don't think it's possible for it to be illegal to write code >> for hardware you own. > >I always thought that reverse-engineering was generally protected >in the U.S. (but not in all countries). Increasingly, however, >I am noticing shrink-wrapped licenses that say something to the >effect that by opening the package, I am agreeing not to reverse- >engineer anything inside the package. Is this legally binding >in the U.S. and/or other countries? Not in the EU. We are explicitly allowed to reverse-engineer if the "interface needed for usage" isn't well documented. -- Poul-Henning Kamp | phk@FreeBSD.ORG FreeBSD Core-team. http://www.freebsd.org/~phk | phk@login.dknet.dk Private mailbox. whois: [PHK] | phk@tfs.com TRW Financial Systems, Inc. Power and ignorance is a disgusting cocktail. From owner-freebsd-smp Thu Jul 24 01:04:51 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id BAA16478 for smp-outgoing; Thu, 24 Jul 1997 01:04:51 -0700 (PDT) Received: from hutcs.cs.hut.fi (apl@hutcs.cs.hut.fi [130.233.192.6]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id BAA16473 for ; Thu, 24 Jul 1997 01:04:47 -0700 (PDT) Received: (from apl@localhost) by hutcs.cs.hut.fi (8.8.5/8.7.3) id LAA19922; Thu, 24 Jul 1997 11:04:41 +0300 (EET DST) Message-ID: <19970724110440.48213@hutcs.cs.hut.fi> Date: Thu, 24 Jul 1997 11:04:40 +0300 From: Antti-Pekka Liedes To: Keith Mitchell Cc: freebsd-smp@freebsd.org Subject: Re: AHA-3940U on Tomcat IIID, asking for trouble? References: <199707232336.TAA15727@weenix.guru.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.79 In-Reply-To: <199707232336.TAA15727@weenix.guru.org>; from Keith Mitchell on Wed, Jul 23, 1997 at 07:36:17PM -0400 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Wed, Jul 23, 1997 at 07:36:17PM -0400, Keith Mitchell wrote: > > Are there any experiences of how well AHA-3940U (bridged PCI card) works > > on Tomcat IIID running freebsd-smp? I read the FBSD-SMP pages concerning > > bridged cards and it doesn't look promising. > > It will work, but it will require kernel patches. I can't speak for the > 4.0 BIOS, but earlier BIOSes did not build the MP table to include > cards with PCI-PCI bridges on them. If you run the MPtable program and > it lists more than one PCI bus you are probably fine though. > Replying my own questions myself, this is what I found from www.tyan.com: "S56AW401.ZIP Ver 4.01 adds support for Intel MMX, AMD K5, and Cyrix CPUs; 512 MB memory; USB; LS120 bootable, APM1.2; SCO Unix SMP with SCSI; DMI & Year2000 ready. For MBs shipped with 3.xx / 4.xx BIOSs ONLY - Fixes 3IDE Spindown and second channel support for Adaptec 3940UW." I'll give it a try when I get the second multiprocessing capapble CPU. > -- > Keith Mitchell Head Administrator: acm.vt.edu > Email: kmitch@weenix.guru.org PGP key available upon request > http://weenix.guru.org/~kmitch > Address and URL (c) 1997 Keith Mitchell - All Rights Reserved > Unauthorized use or duplication prohibited -- Antti-Pekka Liedes * apl@IRC * In two hells there's JMT 6 B 406 * apl@iki.fi * one hell too many 02150 ESPOO * apl@apocalypse.tky.hut.fi * - Lucifer +358 - 9 - 468 3121 * +358 - 40 - 5873 593 * (in God's Army) From owner-freebsd-smp Thu Jul 24 03:15:18 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id DAA20893 for smp-outgoing; Thu, 24 Jul 1997 03:15:18 -0700 (PDT) Received: from cenotaph.snafu.de (gw-deadnet.snafu.de [194.121.229.33]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id DAA20888 for ; Thu, 24 Jul 1997 03:15:04 -0700 (PDT) Received: by cenotaph.snafu.de from deadline.snafu.de using smtp id m0wrKv4-000KBUC; Thu, 24 Jul 1997 12:14:50 +0200 (CEST) (Smail-3.2 1996-Jul-4 #1) Received: by deadline.snafu.de id m0wrKv4-00049oC; Thu, 24 Jul 1997 12:14:50 +0200 (CEST) (Smail-3.2 1996-Jul-4 #1) Message-Id: Date: Thu, 24 Jul 1997 12:14:50 +0200 (CEST) Mime-Version: 1.0 X-Newsreader: knews 0.9.8 References: In-Reply-To: From: mickey@deadline.snafu.de (Andreas S. Wetzel) Subject: Re: user questions X-Original-Newsgroups: lists.freebsd-smp To: Troy Curtiss Cc: smp@freebsd.org Content-Type: text/plain; charset=us-ascii Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi! --- In article , Steve Passe writes: > Hi Steve, > I see you've been working hot & heavy on the SMP FreeBSD stuff. > I may be in a position to buy an MP motherboard here at work, which > would you recommend? Secondly, how far along is the SMP in the way > of process sharing/threading/etc... If I write a program that parallel > tasks (ie... brute forcing password keyspaces), can I put a CPU on each > task/thread/process to effectively double my speed (almost)? I've been running FreeBSD SMP on a dual P5 box (ASUS P65Up5) since about March 97 without having any serious problems with it. I used it to render some animation sequences with Povray 3.0. I splitted the work onto two processes (one rendering all even image numbers and one rendering the odd image numbers). All I can say is that it runs like a breeze. Both povrays run parallel at nearly full single processor speed. Same thing is true for compiling things that support parallel make i.e. SMP kernel. My current kernel configuration takes about 4:59m with -j 8 enabled on a dual P133 and hits at about 180-190 % of single processor speed. Regards, Mickey -- (__) (@@) Andreas S. Wetzel Mail: mickey@deadline.snafu.de /-------\/ Utrechter Strasse 41 Web: http://cenotaph.snafu.de/ / | || 13347 Berlin Fon: <+4930> 456 066 90 * ||----|| Germany Fax: <+4930> 456 066 91/92 ~~ ~~ From owner-freebsd-smp Thu Jul 24 06:47:59 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id GAA28550 for smp-outgoing; Thu, 24 Jul 1997 06:47:59 -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 GAA28545 for ; Thu, 24 Jul 1997 06:47:57 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id HAA24835; Thu, 24 Jul 1997 07:47:51 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id HAA29271; Thu, 24 Jul 1997 07:47:50 -0600 Date: Thu, 24 Jul 1997 07:47:50 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199707241347.HAA29271@fast.cs.utah.edu> To: cbrown@aracnet.com, smp@csn.net Subject: Re: Lots 'o PCI slots Cc: freebsd-smp@FreeBSD.ORG Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Hi, > > >Well, I just thumbed through the I2O spec, and it made no mention of > >APICs or even much about how interrupts with the host system are > >handled. I am guessing that this is implementation specific. If I > >find out anything more that I can tell you, I'll let you know. The I/O processor is off the PCI bus. In fact, Intel's i960 RP/RD has two bridges integrated, which provides for a secondary PCI bus with just one chip. Since the i960 is on the `other side' of the PCI bus, I think it would be pretty stupid to have an x86-specific multiprocessor support bus on the controller. However, it appears that is exactly what they do: http://134.134.214.1/design/iio/des.htm which is annother reason Digital has their own I2O processor based on the StrongARM. http://www.digital.com/info/semiconductor/press0303hm.htm > Indeed it is implementation specific. However the I2O boards I've noticed > so far use the i960 for that "specific hardware". I can picture such boards > using the APIC system for interprocessor communications. I'll try to find > time to go look at the i960 datasheet... > -- The i960 is the INTEL embedded processor for the "Intelligent I/O" system. DEC (of course) has their own processor; the I2O spec doesn't specify which I/O procesor should be used; most people (WinTel anyway) are using the i960 from Intel. So using a i960 IS implementation-specific. Try looking from Intel's site: http://www1.intel.com/design/iio/ The I2O Sig is at: http://www.i2osig.org/ I don't think releasing technical info about the I2O spec is really any different than not releasing info about a RAID controller or a proprietary system bus. Many companies still want NDAs for their hardware specs; this is just a `big company' that has lots of little companies (like IBM and Intel :) This doesn't mean that I like it, nor do I even remotely resemble a lawyer. It would be much better if the specs were $30 and available to everyone (like the PCI spec). Hopefully in a few years it will be that way. http://134.134.214.1/design/i960/prodbref/27274001.htm Shows the logical layout of the i960 RP. Note that cycles are NOT allowed in the PCI spec; however, intel `fudged' this one by having the i960 connected to both PCI busses: there are now two ways for a message (data) to flow, which invalidates the consistency guarantees. Of course, it is also broken with PCI 2.1's split transactions across a bridge chip, as the requests aren't tagged. (A reads C, told to `retry', bridge gets value from C, B reads C and bridge gives it to B instead of C). HP did formal verification work to prove these problems exist, although they have not publically released `fixes'. Kevin From owner-freebsd-smp Thu Jul 24 07:43:17 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA02036 for smp-outgoing; Thu, 24 Jul 1997 07:43:17 -0700 (PDT) Received: from star.cirrus.com (star.cirrus.com [141.131.7.10]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id HAA02027 for ; Thu, 24 Jul 1997 07:43:15 -0700 (PDT) Received: from ss563.corp.cirrus.com (ss563.corp.cirrus.com [141.131.8.55]) by star.cirrus.com (8.6.12/8.6.12) with ESMTP id HAA00593 for ; Thu, 24 Jul 1997 07:43:14 -0700 Received: from elbert.colorado.cirrus.com (elbert.colorado.cirrus.com [198.90.142.11]) by ss563.corp.cirrus.com with ESMTP id HAA29907 (8.7.5/IDA-1.6 for ); Thu, 24 Jul 1997 07:43:12 -0700 (PDT) Received: from longs.colorado.cirrus.com (longs.colorado.cirrus.com [198.90.142.55]) by elbert.colorado.cirrus.com with SMTP id IAA29566 (8.6.12/IDA-1.6 for ); Thu, 24 Jul 1997 08:43:10 -0600 Received: by longs.colorado.cirrus.com (4.1-Colorado/2.00) id AA02450; Thu, 24 Jul 97 08:43:08 MDT Date: Thu, 24 Jul 97 08:43:08 MDT From: clintw@colorado.cirrus.com (Clint Wolff) Message-Id: <9707241443.AA02450@longs.colorado.cirrus.com> To: smp@FreeBSD.ORG Subject: Re: Lots 'o PCI slots Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > The I2O Sig is at: > http://www.i2osig.org/ > > I don't think releasing technical info about the I2O spec is really > any different than not releasing info about a RAID controller or a > proprietary system bus. Many companies still want NDAs for their > hardware specs; this is just a `big company' that has lots of little > companies (like IBM and Intel :) This doesn't mean that I like it, > nor do I even remotely resemble a lawyer. It would be much better if > the specs were $30 and available to everyone (like the PCI spec). > Hopefully in a few years it will be that way. > <<>> Just to poke my two cents worth in here... I believe the I2O Sig isn't trying to get rich selling specs and memberships, just trying to pay their bills... Charging Intel or MicroSoft $2000 to join provides the companies a method to formally write off the costs without the bookeeping required for two hours of this engineer's time or four hours of that secretaries... I suspect a the FreeBSD group could work out an equitable arrangement with the I2O sig to allow a core team to develop the system and distribute the source code... I'd even be willing to kick in some cash to cover the membership fee (but not the whole thing... Sorry :)... With all the discussions about legal/illegal to do it, has anyone asked I2O? Well... Back to lurking... clint From owner-freebsd-smp Thu Jul 24 10:20:02 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA12683 for smp-outgoing; Thu, 24 Jul 1997 10:20:02 -0700 (PDT) Received: from icicle.winternet.com (root@icicle.winternet.com [198.174.169.13]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA12639 for ; Thu, 24 Jul 1997 10:19:58 -0700 (PDT) Received: (from adm@localhost) by icicle.winternet.com (8.7.5/8.7.5) id MAA27455; Thu, 24 Jul 1997 12:13:35 -0500 (CDT) Posted-Date: Thu, 24 Jul 1997 12:13:35 -0500 (CDT) Received: from tundra.winternet.com(198.174.169.11) by icicle.winternet.com via smap (V2.0) id xma027414; Thu, 24 Jul 97 12:13:10 -0500 Received: from localhost (mestery@localhost) by tundra.winternet.com (8.8.4/8.8.4) with SMTP id MAA04854; Thu, 24 Jul 1997 12:13:10 -0500 (CDT) X-Authentication-Warning: tundra.winternet.com: mestery owned process doing -bs Date: Thu, 24 Jul 1997 12:13:10 -0500 (CDT) From: Kyle Mestery To: "Andreas S. Wetzel" cc: smp@FreeBSD.ORG Subject: Re: 3.0-SMP + COMPAT_LINUX ? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Thu, 24 Jul 1997, Andreas S. Wetzel wrote: > Hi! > --- > > I recently tried to run some Linux binaries on top of 3.0-current SMP, > but when I start them the machine hangs forever. Is this "supposed" to > be so at the current stage of development or doesn't anybody notice > until now? > I have the same problem, only my machine hangs for about 2 seconds and then reboots. Not sure if it is related to SMP or just compat_linux though. I only noticed this with quake btw, and it is this Linux qyake binary: i386-ELF-xquake-1.06 hope.winternet.com$ file /usr/local/quakeserver/i386-ELF-xquake-1.06 /usr/local/quakeserver/i386-ELF-xquake-1.06: ELF 32-bit LSB executable, Intel 80386, version 1, dynamically linked, stripped Kyle Mestery StorageTek's Network Systems Group 7600 Boone Ave. N., Brooklyn Park, MN 55428 mesteka@anubis.network.com, mestery@winternet.com From owner-freebsd-smp Thu Jul 24 11:47:36 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA17571 for smp-outgoing; Thu, 24 Jul 1997 11:47:36 -0700 (PDT) Received: from cenotaph.snafu.de (gw-deadnet.snafu.de [194.121.229.33]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA17566 for ; Thu, 24 Jul 1997 11:47:24 -0700 (PDT) Received: by cenotaph.snafu.de from deadline.snafu.de using smtp id m0wrSv1-000KEpC; Thu, 24 Jul 1997 20:47:19 +0200 (CEST) (Smail-3.2 1996-Jul-4 #1) Received: by deadline.snafu.de id m0wrSv0-00049oC; Thu, 24 Jul 1997 20:47:18 +0200 (CEST) (Smail-3.2 1996-Jul-4 #1) Message-Id: From: root@deadline.snafu.de (Andreas S. Wetzel) Subject: Re: 3.0-SMP + COMPAT_LINUX ? To: mestery@winternet.com (Kyle Mestery) Date: Thu, 24 Jul 1997 20:47:18 +0200 (CEST) Cc: smp@FreeBSD.ORG In-Reply-To: from Kyle Mestery at "Jul 24, 97 12:13:10 pm" Organization: A world stranger than you have ever imagined. X-Mailer: ELM [version 2.4ME+ PL13] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi! --- Kyle Mestery writes: ] On Thu, 24 Jul 1997, Andreas S. Wetzel wrote: ] > I recently tried to run some Linux binaries on top of 3.0-current SMP, ] > but when I start them the machine hangs forever. Is this "supposed" to ] > be so at the current stage of development or doesn't anybody notice ] > until now? ] > ] I have the same problem, only my machine hangs for about 2 seconds and ] then reboots. Not sure if it is related to SMP or just compat_linux ] though. I only noticed this with quake btw, and it is this Linux qyake ] binary: ] ] i386-ELF-xquake-1.06 ] hope.winternet.com$ file /usr/local/quakeserver/i386-ELF-xquake-1.06 ] /usr/local/quakeserver/i386-ELF-xquake-1.06: ELF 32-bit LSB executable, ] Intel 80386, version 1, dynamically linked, stripped It doesn't reboot but it hangs forever. The binary I was using is somewhat different: xf86quake: ELF 32-bit LSB executable, Intel 80386, version 1 (Linux), dynamically linked, stripped The QW Client was version 1.64 and hanged itself upon the same way as xf86quake does. I also tried to stop the second processor and then run the linux emulator but that did not make any difference. Regards, Mickey -- (__) (@@) Andreas S. Wetzel Mail: mickey@deadline.snafu.de /-------\/ Utrechter Strasse 41 Web: http://cenotaph.snafu.de/ / | || 13347 Berlin Fon: <+4930> 456 066 90 * ||----|| Germany Fax: <+4930> 456 066 91/92 ~~ ~~ From owner-freebsd-smp Thu Jul 24 12:07:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA18209 for smp-outgoing; Thu, 24 Jul 1997 12:07:09 -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 MAA18167 for ; Thu, 24 Jul 1997 12:06:06 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id NAA22355; Thu, 24 Jul 1997 13:04:55 -0600 (MDT) Message-Id: <199707241904.NAA22355@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: root@deadline.snafu.de (Andreas S. Wetzel) cc: mestery@winternet.com (Kyle Mestery), smp@FreeBSD.ORG Subject: Re: 3.0-SMP + COMPAT_LINUX ? In-reply-to: Your message of "Thu, 24 Jul 1997 20:47:18 +0200." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 24 Jul 1997 13:04:55 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, add "options DDB" to the kernel, run it and hit "CNTL PRINT_SCREEN" to jump to the debugger when it hangs. then do: ddb> trace and send us the results. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu Jul 24 13:04:11 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA20994 for smp-outgoing; Thu, 24 Jul 1997 13:04:11 -0700 (PDT) Received: from george.arc.nasa.gov (george.arc.nasa.gov [128.102.194.142]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id NAA20989 for ; Thu, 24 Jul 1997 13:04:08 -0700 (PDT) From: lamaster@george.arc.nasa.gov Received: by george.arc.nasa.gov (SMI-8.6/1.35) id NAA24606; Thu, 24 Jul 1997 13:02:19 -0700 Date: Thu, 24 Jul 1997 13:02:19 -0700 Message-Id: <199707242002.NAA24606@george.arc.nasa.gov> To: freebsd-smp@FreeBSD.ORG Subject: Re: Lots 'o PCI slots Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk vanmaren@fast.cs.utah.edu (Kevin Van Maren) writes: > The I/O processor is off the PCI bus. In fact, Intel's i960 RP/RD > has two bridges integrated, which provides for a secondary PCI bus > with just one chip. Since the i960 is on the `other side' of the PCI > bus, I think it would be pretty stupid to have an x86-specific > multiprocessor support bus on the controller. However, it appears > that is exactly what they do: > http://134.134.214.1/design/iio/des.htm > which is annother reason Digital has their own I2O processor based on > the StrongARM. http://www.digital.com/info/semiconductor/press0303hm.htm When I first heard about the I2O processor, I thought maybe that it would be something like a Cray IOP, or at least an IBM channel. After I read through the online documentation, it sounds less interesting. As I read the docs, it sounds like it could help performance by fielding PCI interrupts for interrupt-crazy boards. That could be welcome news in some cases, but, since Unix/BSD folks tend to shy away from such boards anyway whenever possible, it seems that the only real "benefit" for well-behaved boards is going to be increased latency. For example, the most performance-critical path is usually the SCSI controller. If the controller already has a minimal-interrupt message-passing flavor, does gather-scatter, etc., what's the difference? Except for the added PCI bus traffic and the added latency due to the I/O request having to get processed in the I2O processor. What am I missing? > I don't think releasing technical info about the I2O spec is really > any different than not releasing info about a RAID controller or a > proprietary system bus. Many companies still want NDAs for their > hardware specs; this is just a `big company' that has lots of little If the reverse-engineering to document the interface takes place in Europe, and someone writes drivers based on the documented interface, it would appear that distribution of the resulting code would be legal, as long as no one violates an NDA. But, I'm not a lawyer and not giving legal advice. It could also be that there are some sneaky landmines hidden in the interface to make reverse-engineering more difficult, and to facilitate detection of NDA violations. > Shows the logical layout of the i960 RP. Note that cycles are NOT > allowed in the PCI spec; however, intel `fudged' this one by having > the i960 connected to both PCI busses: there are now two ways for > a message (data) to flow, which invalidates the consistency guarantees. > Of course, it is also broken with PCI 2.1's split transactions across > a bridge chip, as the requests aren't tagged. (A reads C, told to > `retry', bridge gets value from C, B reads C and bridge gives it to B > instead of C). HP did formal verification work to prove these > problems exist, although they have not publically released `fixes'. Presumably, the software will fix it so that either the CPU would own a particular device, or, the I2O processor would own the device, but, not both. In that case, is it still a problem? I'm not clear on how the I2O will prevent the CPU from seeing the PCI interrupts for the devices it owns, though, if they are sharing access to the same the PCI bus. Presumably, if the architecture ever migrates to the point where the I2O has direct access to system memory and has its own, multiple, PCI buses, then it would have much more of a benefit. Hugh LaMaster, M/S 258-5, ASCII Email: hlamaster@mail.arc.nasa.gov NASA Ames Research Center Or: lamaster@nas.nasa.gov Moffett Field, CA 94035-1000 No Junkmail: USC 18 section 2701 Phone: 415/604-1056 Disclaimer: Unofficial, personal *opinion*. From owner-freebsd-smp Thu Jul 24 13:23:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA21844 for smp-outgoing; Thu, 24 Jul 1997 13:23:23 -0700 (PDT) Received: from sendero-ppp.i-connect.net (sendero-ppp.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id NAA21838 for ; Thu, 24 Jul 1997 13:23:16 -0700 (PDT) Received: (qmail 7450 invoked by uid 1000); 24 Jul 1997 20:23:28 -0000 Message-ID: X-Mailer: XFMail 1.2-alpha [p0] on FreeBSD X-PRIORITY: 2 (High) Priority: urgent Content-Type: text/plain; charset=iso-8859-8 Content-Transfer-Encoding: 8bit MIME-Version: 1.0 Date: Thu, 24 Jul 1997 13:23:28 -0700 (PDT) Organization: Atlas Telecom From: Simon Shapiro To: freebsd-smp@freebsd.org Subject: SMP & P6DNH - Still the same (problem) Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi Y'all, SMP kernel as of last night, still gives me the same exact problem: It boot normally and then, instead of starting init, it goes into default_halt(). I only found reference to default_halt in i386/i386/dwtch.s. Could not find any calls made to it, except, indirectly from idle_loop (in the same file). I do not exactly understand (last assembly code I saw was PDP-11, sorry) what idle loop does, but it appears as if vm_page_zero_idle returns 0, and this is why we halt. This can only happen (?) (in vm_machdep.c) if WRONG is defined, or the nimbr of non-zeroed free pages is smaller than half the number of reserved pages. Now, none of this makes sense to me, which most likely means it is correct. The fact is that I could boot and run SMP on that motherboard and now I cannot. Does anyone have a clue? does anyone have a /usr/src/sys that compiles and builds and boots on a P6DNH? Simon From owner-freebsd-smp Thu Jul 24 13:54:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA23495 for smp-outgoing; Thu, 24 Jul 1997 13:54:33 -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 NAA23490 for ; Thu, 24 Jul 1997 13:54:31 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id NAA18027; Thu, 24 Jul 1997 13:49:06 -0700 From: Terry Lambert Message-Id: <199707242049.NAA18027@phaeton.artisoft.com> Subject: Re: 3.0-SMP + COMPAT_LINUX ? To: brian@mpress.com (Brian Litzinger) Date: Thu, 24 Jul 1997 13:49:06 -0700 (MST) Cc: root@deadline.snafu.de, smp@FreeBSD.ORG In-Reply-To: <19970723191751.30899@mpress.com> from "Brian Litzinger" at Jul 23, 97 07:17:51 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > I recently tried to run some Linux binaries on top of 3.0-current SMP, > > but when I start them the machine hangs forever. Is this "supposed" to > > be so at the current stage of development or doesn't anybody notice > > until now? > > I run ApplixWare, and the linux version of netscape on an > SMP -current machine with linux compatibility compiled into the > kernel quite extensively. This issue was discussed on the list in the past. The upshot was "if it's an LKM, it will crash with an SMP kernel, but if it's compiled in, it will work". I believe the problem is that there is not a dual mapping established for kernel memory allocations of a particular class. It may be as simple as stale cache data, since the data load is a two stage process, or it may be as complex as the particular kernel reoutine used to allocate the logically contiguous pages. My personaly suspicion is that if the LKM load occurs entirely on one processor, then code using the LKM will run, assuming it always runs on that same processor (hence the intermittent behaviour). One issue is that the LKM code required the kernel address space to be mapped in its entirety. I believe the SMP code *requires* seperate regions (per processor data areas and kernel stacks), *but* (and this is a biggie) the memory map for the rest of the address space was not adjusted when the change went in. Regards, 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 Jul 24 13:56:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA23594 for smp-outgoing; Thu, 24 Jul 1997 13:56:49 -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 NAA23589 for ; Thu, 24 Jul 1997 13:56:46 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id NAA18047; Thu, 24 Jul 1997 13:52:29 -0700 From: Terry Lambert Message-Id: <199707242052.NAA18047@phaeton.artisoft.com> Subject: Re: Lots 'o PCI slots To: phk@dk.tfs.com (Poul-Henning Kamp) Date: Thu, 24 Jul 1997 13:52:28 -0700 (MST) Cc: lamaster@george.arc.nasa.gov, freebsd-smp@FreeBSD.ORG In-Reply-To: <1168.869725915@critter.dk.tfs.com> from "Poul-Henning Kamp" at Jul 24, 97 08:31:55 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-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Not in the EU. We are explicitly allowed to reverse-engineer if > the "interface needed for usage" isn't well documented. Is this every EU country? I know the "priacy clause" was insisted upon by Germany, specifically. Has the EU gotten rid of contradictory local laws? How Unified are you guys these days? 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 Jul 24 14:01:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA23832 for smp-outgoing; Thu, 24 Jul 1997 14:01:34 -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 OAA23822 for ; Thu, 24 Jul 1997 14:01:28 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id NAA18060; Thu, 24 Jul 1997 13:57:17 -0700 From: Terry Lambert Message-Id: <199707242057.NAA18060@phaeton.artisoft.com> Subject: Re: Lots 'o PCI slots To: lamaster@george.arc.nasa.gov Date: Thu, 24 Jul 1997 13:57:17 -0700 (MST) Cc: freebsd-smp@FreeBSD.ORG In-Reply-To: <199707242002.NAA24606@george.arc.nasa.gov> from "lamaster@george.arc.nasa.gov" at Jul 24, 97 01:02:19 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > That could be welcome news in some cases, but, since Unix/BSD folks > tend to shy away from such boards anyway whenever possible, it seems > that the only real "benefit" for well-behaved boards is going to be > increased latency. For example, the most performance-critical > path is usually the SCSI controller. If the controller already > has a minimal-interrupt message-passing flavor, does gather-scatter, > etc., what's the difference? Except for the added PCI bus traffic > and the added latency due to the I/O request having to get processed > in the I2O processor. > > What am I missing? PCI-PCI bridging, I would guess, where the i960 is a cluster communications processor more than simply a method of offloading interrupts. It would be a relatively cheap method of obtaining background distributed cache coherency processing (among other things). One might even use the i960 for the scheduler and for forcible process migration. Only a theory... 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 Jul 24 14:49:27 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA26347 for smp-outgoing; Thu, 24 Jul 1997 14:49:27 -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 OAA26330 for ; Thu, 24 Jul 1997 14:49:21 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id PAA23073; Thu, 24 Jul 1997 15:48:32 -0600 (MDT) Message-Id: <199707242148.PAA23073@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Simon Shapiro cc: freebsd-smp@FreeBSD.ORG Subject: Re: SMP & P6DNH - Still the same (problem) In-reply-to: Your message of "Thu, 24 Jul 1997 13:23:28 PDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 24 Jul 1997 15:48:32 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > SMP kernel as of last night, still gives me the same exact problem: > > It boot normally and then, instead of starting init, it goes into > default_halt(). > > I only found reference to default_halt in i386/i386/dwtch.s. > Could not find any calls made to it, except, indirectly from idle_loop (in > the same file). > > I do not exactly understand (last assembly code I saw was PDP-11, sorry) > what idle loop does, but it appears as if vm_page_zero_idle returns 0, > and this is why we halt. This can only happen (?) (in vm_machdep.c) if > WRONG is defined, or the nimbr of non-zeroed free pages is smaller than > half the number of reserved pages. what idle_loop does: check each queue for a process to run see if there are any vm pages that can be zeroed while waiting for a proc. If a page was zeroed repeat the loop. else no pages left to zero, go into a halt until an INT wakes us up. Since you get to ddb from the keyboard, kb INTs are being seen, it sounds like your not getting timer INTerrupts. mail me both your config file and i386/include/smptests.h file. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu Jul 24 15:42:54 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA29185 for smp-outgoing; Thu, 24 Jul 1997 15:42:54 -0700 (PDT) Received: from sendero-ppp.i-connect.net (sendero-ppp.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id PAA29180 for ; Thu, 24 Jul 1997 15:42:48 -0700 (PDT) Received: (qmail 10248 invoked by uid 1000); 24 Jul 1997 22:43:01 -0000 Message-ID: X-Mailer: XFMail 1.2-alpha [p0] on FreeBSD MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="_=XFMail.1.2-alpha.p0.FreeBSD:970724153834:27423=_" In-Reply-To: <199707242148.PAA23073@Ilsa.StevesCafe.com> Date: Thu, 24 Jul 1997 15:43:01 -0700 (PDT) Organization: Atlas Telecom From: Simon Shapiro To: Steve Passe Subject: Re: SMP & P6DNH - Still the same (problem) Cc: freebsd-smp@FreeBSD.ORG Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk This message is in MIME format --_=XFMail.1.2-alpha.p0.FreeBSD:970724153834:27423=_ Content-Type: text/plain; charset=iso-8859-8 Hi Steve Passe; On 24-Jul-97 you wrote: > Hi, > > > SMP kernel as of last night, still gives me the same exact problem: > > > > It boot normally and then, instead of starting init, it goes into > > default_halt(). > > > > I only found reference to default_halt in i386/i386/dwtch.s. > > Could not find any calls made to it, except, indirectly from idle_loop > (in > > the same file). > > > > I do not exactly understand (last assembly code I saw was PDP-11, > sorry) > > what idle loop does, but it appears as if vm_page_zero_idle returns 0, > > and this is why we halt. This can only happen (?) (in vm_machdep.c) if > > WRONG is defined, or the nimbr of non-zeroed free pages is smaller than > > half the number of reserved pages. > > what idle_loop does: > > check each queue for a process to run > see if there are any vm pages that can be zeroed while waiting for a > proc. > If a page was zeroed repeat the loop. > else no pages left to zero, go into a halt until an INT wakes us up. > > Since you get to ddb from the keyboard, kb INTs are being seen, > it sounds like your not getting timer INTerrupts. mail me both your > config > file and i386/include/smptests.h file. Here you go... Thanx! Simon --_=XFMail.1.2-alpha.p0.FreeBSD:970724153834:27423=_ Content-Disposition: attachment; filename="smptests.h" Content-Transfer-Encoding: none Content-Description: smptests.h Content-Type: application/octet-stream; name=smptests.h; SizeOnDisk=5009 /* * Copyright (c) 1996, by Steve Passe * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. The name of the developer may NOT be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $Id: smptests.h,v 1.15 1997/07/22 18:46:41 fsmp Exp $ */ #ifndef _MACHINE_SMPTESTS_H_ #define _MACHINE_SMPTESTS_H_ /* * various 'tests in progress' */ /* * Address of POST hardware port. * Defining this enables POSTCODE macros. * #define POST_ADDR 0x80 */ /* * 1st attempt to use ExtInt connected 8259 to attach 8254 timer. * failing that, attempt to attach 8254 timer via direct APIC pin. * failing that, panic! * This overrides both APIC_PIN0_TIMER & TEST_ALTTIMER */ #undef NEW_STRATEGY /* * Use 'regular Int' method to connect external 8254 timer via IO APIC pin 0. * See "Intel I486 Microprocessors and Related Products", page 4-292: * 82489DX/8259A DUAL MODE CONNECTION * */ #undef APIC_PIN0_TIMER /* * Use non 'ExtInt' method of external (non-conected) 8254 timer * See "Intel I486 Microprocessors and Related Products", page 4-292: * 82489DX/8259A DUAL MODE CONNECTION * */ #undef TEST_ALTTIMER /* * Send 8254 timer INTs to all CPUs in LOPRIO mode. * */ #undef TIMER_ALL /* * Send CPUSTOP IPI for stop/restart of other CPUs on DDB break. * */ #define CPUSTOP_ON_DDBBREAK #define VERBOSE_CPUSTOP_ON_DDBBREAK /* * Bracket code/comments relevant to the current 'giant lock' model. * Everything is now the 'giant lock' model, but we will use this as * we start to "push down" the lock. */ #define GIANT_LOCK /* * use 'lowest priority' for sending IRQs to CPUs * * i386/i386/mplock.s, i386/i386/mpapic.c, kern/init_main.c * */ #define TEST_LOPRIO /* * deal with broken smp_idleloop() */ #define IGNORE_IDLEPROCS /* * misc. counters * #define COUNT_XINVLTLB_HITS #define COUNT_SPURIOUS_INTS #define COUNT_CSHITS */ /** * hack to "fake-out" kernel into thinking it is running on a 'default config' * * value == default type #define TEST_DEFAULT_CONFIG 6 */ /* * simple test code for IPI interaction, save for future... * #define TEST_TEST1 #define IPI_TARGET_TEST1 1 */ /* * POST hardware macros. */ #ifdef POST_ADDR #define ASMPOSTCODE_INC \ pushl %eax ; \ movl _current_postcode, %eax ; \ incl %eax ; \ andl $0xff, %eax ; \ movl %eax, _current_postcode ; \ outb %al, $POST_ADDR ; \ popl %eax /* * Overwrite the current_postcode value. */ #define ASMPOSTCODE(X) \ pushl %eax ; \ movl $X, %eax ; \ movl %eax, _current_postcode ; \ outb %al, $POST_ADDR ; \ popl %eax /* * Overwrite the current_postcode low nibble. */ #define ASMPOSTCODE_LO(X) \ pushl %eax ; \ movl _current_postcode, %eax ; \ andl $0xf0, %eax ; \ orl $X, %eax ; \ movl %eax, _current_postcode ; \ outb %al, $POST_ADDR ; \ popl %eax /* * Overwrite the current_postcode high nibble. */ #define ASMPOSTCODE_HI(X) \ pushl %eax ; \ movl _current_postcode, %eax ; \ andl $0x0f, %eax ; \ orl $(X<<4), %eax ; \ movl %eax, _current_postcode ; \ outb %al, $POST_ADDR ; \ popl %eax #else #define ASMPOSTCODE_INC #define ASMPOSTCODE(X) #define ASMPOSTCODE_LO(X) #define ASMPOSTCODE_HI(X) #endif /* POST_ADDR */ /* * these are all temps for debugging... * #define GUARD_INTS */ /* * This macro traps unexpected INTs to a specific CPU, eg. GUARD_CPU. */ #ifdef GUARD_INTS #define GUARD_CPU 1 #define MAYBE_PANIC(irq_num) \ cmpl $GUARD_CPU, _cpuid ; \ jne 9f ; \ cmpl $1, _ok_test1 ; \ jne 9f ; \ pushl lapic_isr3 ; \ pushl lapic_isr2 ; \ pushl lapic_isr1 ; \ pushl lapic_isr0 ; \ pushl lapic_irr3 ; \ pushl lapic_irr2 ; \ pushl lapic_irr1 ; \ pushl lapic_irr0 ; \ pushl $irq_num ; \ pushl _cpuid ; \ pushl $panic_msg ; \ call _printf ; \ addl $44, %esp ; \ 9: #else #define MAYBE_PANIC(irq_num) #endif /* GUARD_INTS */ #endif /* _MACHINE_SMPTESTS_H_ */ --_=XFMail.1.2-alpha.p0.FreeBSD:970724153834:27423=_ Content-Type: application/octet-stream; name=DPT-smp; SizeOnDisk=11342 Content-Description: DPT-smp Content-Transfer-Encoding: none Content-Disposition: attachment; filename="DPT-smp" # # SENDERO -- Sendero configuration # # $Id: SENDERO,v 1.11 1997/07/06 20:10:56 ShimonR Exp $ machine "i386" cpu "I686_CPU" cpu "I586_CPU" ident SENDERO maxusers 64 options SMP # Symmetric MultiProcessor Kernel options APIC_IO # Symmetric (APIC) I/O #options SMP_AUTOSTART # 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 # SuperMicro P6DNxxx: options SMP_TIMER_NC # 8254 NOT connected to APIC options CHILD_MAX=128 options OPEN_MAX=128 options "MAXDSIZ=(64*1024*1024)" options "DFLDSIZ=(16*1024*1024)" options EXTRAVNODES=2048 config kernel root on wd0a # ... dumps on sd0s2b options "COMPAT_43" #Compatible with BSD 4.3 [KEEP THIS!] options USER_LDT #allow user-level control of i386 ldt options SYSVSHM options SYSVSEM options SYSVMSG options DDB options DDB_UNATTENDED options KTRACE #kernel tracing options PERFMON options UCONSOLE #Allow users to grab the console options USERCONFIG #boot -c editor options USERCONFIG_BOOT #imply -c and parse info area #options VISUAL_USERCONFIG #visual boot -c editor options INET #InterNETworking pseudo-device ether pseudo-device fddi pseudo-device sppp pseudo-device loop pseudo-device ppp 4 pseudo-device bpfilter 16 pseudo-device disc pseudo-device tun 4 options MROUTING # Multicast routing options IPDIVERT #divert sockets options TCPDEBUG options FFS #Berkeley Fast Filesystem options NFS #Network Filesystem options "CD9660" #ISO 9660 Filesystem options PROCFS #Process filesystem options DEVFS #devices filesystem options QUOTA #enable disk quotas options NULLFS_DIAGNOSTIC options KERNFS_DIAGNOSTIC options UMAPFS_DIAGNOSTIC options SAFETY controller pci0 device ch0 #SCSI media changers device sd0 #SCSI disks device st0 #SCSI tapes device cd0 #SCSI CD-ROMs device od0 #SCSI optical disk options SCSI_DELAY=5 # Don't be too pessimistic about SCSI device # Disk Addressing Guidelines: # We assume wide SCSI, not FCAL, with this release. # There is never a target 7, nor target 15; They are reserved for initiators! # Each ADAPTEr takes 4 busses. # Each bus consumes 16 addresses # Non-zero LUNs have the same ID as LUN 0 but bit 7 is set on (64 + target) controller dpt0 controller dpt1 controller scbus0 at dpt0 bus 0 controller scbus1 at dpt0 bus 1 controller scbus2 at dpt0 bus 2 controller scbus4 at dpt1 bus 0 controller scbus5 at dpt1 bus 1 controller scbus6 at dpt1 bus 2 # First DPT starts here options DPT_USE_SINTR=1 # Use software interrupts (crucial) options DPT_MEASURE_PERFORMANCE # Lose ~0.1% performance but measure it options DPT_TRACK_CCB_STATES # Diagnostic value only. @ Simon's direction options DPT_HANDLE_TIMEOUTS # Leave on if things hang options DPT_COMMAND_SPLHIGH # Experimental, briefly locks eata ports options DPT_INTR_CHECK_SOFTC # Experimental; checks soft structs is valid device worm0 at scbus0 target 4 unit 0 tape st0 at scbus0 target 5 unit 0 device cd0 at scbus0 target 6 unit 0 disk sd0 at scbus0 target 0 unit 0 disk sd1 at scbus0 target 1 unit 0 disk sd2 at scbus0 target 2 unit 0 disk sd3 at scbus0 target 3 unit 0 disk sd4 at scbus0 target 4 unit 0 disk sd5 at scbus0 target 5 unit 0 disk sd6 at scbus0 target 6 unit 0 disk sd8 at scbus0 target 8 unit 0 disk sd9 at scbus0 target 9 unit 0 disk sd10 at scbus0 target 10 unit 0 disk sd11 at scbus0 target 11 unit 0 disk sd12 at scbus0 target 12 unit 0 disk sd13 at scbus0 target 13 unit 0 disk sd14 at scbus0 target 14 unit 0 device worm1 at scbus1 target 4 unit 0 tape st1 at scbus1 target 5 unit 0 device cd1 at scbus1 target 6 unit 0 disk sd16 at scbus1 target 0 unit 0 disk sd17 at scbus1 target 1 unit 0 disk sd18 at scbus1 target 2 unit 0 disk sd19 at scbus1 target 3 unit 0 disk sd20 at scbus1 target 4 unit 0 disk sd21 at scbus1 target 5 unit 0 disk sd22 at scbus1 target 6 unit 0 disk sd24 at scbus1 target 8 unit 0 disk sd25 at scbus1 target 9 unit 0 disk sd26 at scbus1 target 10 unit 0 disk sd27 at scbus1 target 11 unit 0 disk sd28 at scbus1 target 12 unit 0 disk sd29 at scbus1 target 13 unit 0 disk sd30 at scbus1 target 14 unit 0 device worm2 at scbus2 target 4 unit 0 tape st2 at scbus2 target 5 unit 0 device cd2 at scbus2 target 6 unit 0 disk sd32 at scbus2 target 0 unit 0 disk sd33 at scbus2 target 1 unit 0 disk sd34 at scbus2 target 2 unit 0 disk sd35 at scbus2 target 3 unit 0 disk sd36 at scbus2 target 4 unit 0 disk sd37 at scbus2 target 5 unit 0 disk sd38 at scbus2 target 6 unit 0 disk sd41 at scbus2 target 9 unit 0 disk sd42 at scbus2 target 10 unit 0 disk sd43 at scbus2 target 11 unit 0 disk sd44 at scbus2 target 12 unit 0 disk sd45 at scbus2 target 13 unit 0 disk sd46 at scbus2 target 14 unit 0 # Second DPT starts here device worm3 at scbus4 target 4 unit 0 tape st3 at scbus4 target 5 unit 0 device cd3 at scbus4 target 6 unit 0 disk sd64 at scbus4 target 0 unit 0 disk sd65 at scbus4 target 1 unit 0 disk sd66 at scbus4 target 2 unit 0 disk sd67 at scbus4 target 3 unit 0 disk sd68 at scbus4 target 4 unit 0 disk sd69 at scbus4 target 5 unit 0 disk sd70 at scbus4 target 6 unit 0 disk sd72 at scbus4 target 8 unit 0 disk sd73 at scbus4 target 9 unit 0 disk sd74 at scbus4 target 10 unit 0 disk sd75 at scbus4 target 11 unit 0 disk sd76 at scbus4 target 12 unit 0 disk sd77 at scbus4 target 13 unit 0 disk sd78 at scbus4 target 14 unit 0 disk sd79 at scbus4 target 15 unit 0 device worm4 at scbus5 target 4 unit 0 tape st4 at scbus5 target 5 unit 0 device cd4 at scbus5 target 6 unit 0 disk sd80 at scbus5 target 0 unit 0 disk sd81 at scbus5 target 1 unit 0 disk sd82 at scbus5 target 2 unit 0 disk sd83 at scbus5 target 3 unit 0 disk sd84 at scbus5 target 4 unit 0 disk sd85 at scbus5 target 5 unit 0 disk sd86 at scbus5 target 6 unit 0 disk sd88 at scbus5 target 8 unit 0 disk sd89 at scbus5 target 9 unit 0 disk sd90 at scbus5 target 10 unit 0 disk sd91 at scbus5 target 11 unit 0 disk sd92 at scbus5 target 12 unit 0 disk sd93 at scbus5 target 13 unit 0 disk sd94 at scbus5 target 14 unit 0 disk sd95 at scbus5 target 15 unit 0 device worm5 at scbus6 target 4 unit 0 tape st5 at scbus6 target 5 unit 0 device cd5 at scbus6 target 6 unit 0 disk sd96 at scbus6 target 0 unit 0 disk sd97 at scbus6 target 1 unit 0 disk sd98 at scbus6 target 2 unit 0 disk sd99 at scbus6 target 3 unit 0 disk sd100 at scbus6 target 4 unit 0 disk sd101 at scbus6 target 5 unit 0 disk sd102 at scbus6 target 6 unit 0 disk sd104 at scbus6 target 8 unit 0 disk sd105 at scbus6 target 9 unit 0 disk sd106 at scbus6 target 10 unit 0 disk sd107 at scbus6 target 11 unit 0 disk sd108 at scbus6 target 12 unit 0 disk sd109 at scbus6 target 13 unit 0 disk sd110 at scbus6 target 14 unit 0 disk sd111 at scbus6 target 15 unit 0 controller ahc0 options AHC_TAGENABLE options AHC_SCBPAGING_ENABLE options AHC_ALLOW_MEMIO controller scbus8 at ahc0 disk sd128 at scbus8 target 0 unit 0 disk sd129 at scbus8 target 1 unit 0 disk sd130 at scbus8 target 2 unit 0 disk sd131 at scbus8 target 3 unit 0 disk sd132 at scbus8 target 4 unit 0 disk sd133 at scbus8 target 5 unit 0 disk sd134 at scbus8 target 6 unit 0 disk sd136 at scbus8 target 8 unit 0 disk sd137 at scbus8 target 9 unit 0 disk sd138 at scbus8 target 10 unit 0 disk sd139 at scbus8 target 11 unit 0 disk sd140 at scbus8 target 12 unit 0 disk sd141 at scbus8 target 13 unit 0 disk sd142 at scbus8 target 14 unit 0 disk sd143 at scbus8 target 15 unit 0 pseudo-device pty 128 pseudo-device speaker #Play IBM BASIC-style noises out your speaker pseudo-device log pseudo-device gzip # Exec gzipped a.out's pseudo-device vn #Vnode driver (turns a file into a device) pseudo-device snp 3 #Snoop device - to look at pty/vty/etc.. pseudo-device ccd 64 #Concatenated disk driver options FAILSAFE #Be conservative controller isa0 options "AUTO_EOI_1" options "AUTO_EOI_2" options BOUNCE_BUFFERS #include support for DMA bounce buffers options "MAXMEM=(256*1024)" # 128M, no autodetect over 64M :-( #device vt0 at isa? port "IO_KBD" tty irq 1 vector pcrint #options PCVT_FREEBSD=210 # pcvt running on FreeBSD >= 2.0.5 options XSERVER # include code for XFree86 #options FAT_CURSOR # start with block cursor device sc0 at isa? port "IO_KBD" tty irq 1 vector scintr options MAXCONS=12 # number of virtual consoles device npx0 at isa? port "IO_NPX" irq 13 vector npxintr controller wdc0 at isa? port "IO_WD1" bio irq 14 vector wdintr disk wd0 at wdc0 drive 0 disk wd1 at wdc0 drive 1 controller wdc1 at isa? port "IO_WD2" bio irq 15 vector wdintr disk wd2 at wdc1 drive 0 disk wd3 at wdc1 drive 1 options ATAPI #Enable ATAPI support for IDE bus options ATAPI_STATIC #Don't do it as an LKM device wcd0 #IDE CD-ROM 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 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 sio2 at isa? port "IO_COM3" tty irq 5 vector siointr device sio3 at isa? port "IO_COM4" tty irq 7 vector siointr device ed0 at isa? port 0x300 net irq 10 iomem 0xd8000 vector edintr controller snd0 device sb0 at isa? port 0x220 irq 7 drq 1 vector sbintr device sbxvi0 at isa? drq 5 device sbmidi0 at isa? port 0x300 device opl0 at isa? port 0x388 device de0 device fxp0 device fpa0 options COMPAT_LINUX options "EXT2FS" options "IBCS2" options SHOW_BUSYBUFS # List buffers that prevent root unmount --_=XFMail.1.2-alpha.p0.FreeBSD:970724153834:27423=_-- End of MIME message From owner-freebsd-smp Thu Jul 24 16:02:17 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id QAA29919 for smp-outgoing; Thu, 24 Jul 1997 16:02:17 -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 QAA29913 for ; Thu, 24 Jul 1997 16:02:13 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id RAA23420 for ; Thu, 24 Jul 1997 17:01:59 -0600 (MDT) Message-Id: <199707242301.RAA23420@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@FreeBSD.ORG Subject: Re: 3.0-SMP + COMPAT_LINUX ? Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 24 Jul 1997 17:01:59 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > I recently tried to run some Linux binaries on top of 3.0-current SMP, > > but when I start them the machine hangs forever. Is this "supposed" to > > be so at the current stage of development or doesn't anybody notice > > until now? > > > I have the same problem, only my machine hangs for about 2 seconds and > then reboots. Not sure if it is related to SMP or just compat_linux > though. I only noticed this with quake btw, and it is this Linux qyake > binary: both Kyle and Andreas have confirmed that this problem exists with a current UP kernel, so evidently something in 3.0-current itself is recently broken for COMPAT_LINUX. I'm removing this from my SMP problem list. on to the next one... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu Jul 24 16:08:08 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id QAA00373 for smp-outgoing; Thu, 24 Jul 1997 16:08:08 -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 QAA00327 for ; Thu, 24 Jul 1997 16:07:46 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id RAA04229; Thu, 24 Jul 1997 17:07:36 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id RAA14924; Thu, 24 Jul 1997 17:07:36 -0600 Date: Thu, 24 Jul 1997 17:07:36 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199707242307.RAA14924@fast.cs.utah.edu> To: freebsd-smp@FreeBSD.ORG, lamaster@george.arc.nasa.gov Subject: Re: Lots 'o PCI slots Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > When I first heard about the I2O processor, I thought maybe that > it would be something like a Cray IOP, or at least an IBM channel. > After I read through the online documentation, it sounds less > interesting. As I read the docs, it sounds like it could help > performance by fielding PCI interrupts for interrupt-crazy boards. > That could be welcome news in some cases, but, since Unix/BSD folks > tend to shy away from such boards anyway whenever possible, it seems > that the only real "benefit" for well-behaved boards is going to be > increased latency. For example, the most performance-critical > path is usually the SCSI controller. If the controller already > has a minimal-interrupt message-passing flavor, does gather-scatter, > etc., what's the difference? Except for the added PCI bus traffic > and the added latency due to the I/O request having to get processed > in the I2O processor. > > What am I missing? Additional processing that is done by the host CPU. Ie, routing between multiple ethernet devices could be done completly by the i960, without any `interference' by the host CPU. `software' RAID (as under NT) could be done by the dedicated `RAID processor' the i960, rather than by the host CPU. If this ever takes off, there will be `high end' motherboards with very fast I2O processors, and then the `low end' with the 30 MIPS ones. > > Shows the logical layout of the i960 RP. Note that cycles are NOT > > allowed in the PCI spec; however, intel `fudged' this one by having > > the i960 connected to both PCI busses: there are now two ways for > > a message (data) to flow, which invalidates the consistency guarantees. > > Of course, it is also broken with PCI 2.1's split transactions across > > a bridge chip, as the requests aren't tagged. (A reads C, told to > > `retry', bridge gets value from C, B reads C and bridge gives it to B > > instead of C). HP did formal verification work to prove these > > problems exist, although they have not publically released `fixes'. > > Presumably, the software will fix it so that either the CPU > would own a particular device, or, the I2O processor would own > the device, but, not both. In that case, is it still a problem? The example that I was shown looks like this (with the i960) secondary bus ---------- Primary PCI ==============|PCI/PCI |================= x86 processor(s) | | Bridge | | | ---------- | | | ----------------------------- | i960 processor/memory | ----------------------------- Device on secondary bus writes to the i960's memory (as it should); main processor reads that location. Note that NEITHER WENT THROUGH the PCI bridge! However, if the processor reads off the secondary bus, then it does go through the bridge. There are therefore cycles; the use of `mailboxes' to signal completion has thus been invalidated, as it is now possible for a read to pass a write! (FOr more information about the PCI ordering requirements, look in the PCI specs). The problem with PCI is that no formal (verification) methods were used until after it was aproved. That and Intel `bent' (or broke) the rules with this i960 controller. It makes sense, from a performance standpoint, to not have to go through the bridge -- just add annother PCI interface, right? > I'm not clear on how the I2O will prevent the CPU from seeing > the PCI interrupts for the devices it owns, though, if they Neither am I :) It could be that they are `disabled' and only come in through the APIC connected to the i960? (on x86) Kevin From owner-freebsd-smp Thu Jul 24 16:10:29 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id QAA00494 for smp-outgoing; Thu, 24 Jul 1997 16:10:29 -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 QAA00486 for ; Thu, 24 Jul 1997 16:10:24 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id RAA04253; Thu, 24 Jul 1997 17:09:02 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id RAA14938; Thu, 24 Jul 1997 17:09:02 -0600 Date: Thu, 24 Jul 1997 17:09:02 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199707242309.RAA14938@fast.cs.utah.edu> To: lamaster@george.arc.nasa.gov, terry@lambert.org Subject: Re: Lots 'o PCI slots Cc: freebsd-smp@FreeBSD.ORG Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > It would be a relatively cheap method of obtaining background > distributed cache coherency processing (among other things). One > might even use the i960 for the scheduler and for forcible process > migration. Actually, the PCI cache-coherence isn't exactly `normal'. Also, as I mentioned, it is broken with PCI 2.1 bridges as well as with the i960 I2O controller. Kevin From owner-freebsd-smp Thu Jul 24 18:25:21 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id SAA06805 for smp-outgoing; Thu, 24 Jul 1997 18:25:21 -0700 (PDT) Received: from mail.cdsnet.net (mail.cdsnet.net [204.118.244.5]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id SAA06793 for ; Thu, 24 Jul 1997 18:25:14 -0700 (PDT) Received: from mail.cdsnet.net (mail.cdsnet.net [204.118.244.5]) by mail.cdsnet.net (8.8.6/8.8.6) with SMTP id SAA25182 for ; Thu, 24 Jul 1997 18:25:13 -0700 (PDT) Date: Thu, 24 Jul 1997 18:25:09 -0700 (PDT) From: Jaye Mathisen To: smp@freebsd.org Subject: Anybody know kernel tweak stuff for a Digital HX or ZX P6? Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I have a 4 processor P6 box I was thinking of converting from Solaris to FreeBSD-SMP, but am not intimately familiar with the hardware enough to know what the various kernel options (APIC related I guess) should be set to. Anybody have any ideas? From owner-freebsd-smp Thu Jul 24 19:07:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA08196 for smp-outgoing; Thu, 24 Jul 1997 19:07:26 -0700 (PDT) Received: from icicle.winternet.com (adm@icicle.winternet.com [198.174.169.13]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA08190 for ; Thu, 24 Jul 1997 19:07:19 -0700 (PDT) Received: (from adm@localhost) by icicle.winternet.com (8.7.5/8.7.5) id VAA29480; Thu, 24 Jul 1997 21:06:57 -0500 (CDT) Posted-Date: Thu, 24 Jul 1997 21:06:57 -0500 (CDT) Received: from tundra.winternet.com(198.174.169.11) by icicle.winternet.com via smap (V2.0) id xma029454; Thu, 24 Jul 97 21:06:40 -0500 Received: from localhost (mestery@localhost) by tundra.winternet.com (8.8.4/8.8.4) with SMTP id VAA12639; Thu, 24 Jul 1997 21:06:40 -0500 (CDT) X-Authentication-Warning: tundra.winternet.com: mestery owned process doing -bs Date: Thu, 24 Jul 1997 21:06:40 -0500 (CDT) From: Kyle Mestery To: Jaye Mathisen cc: smp@FreeBSD.ORG Subject: Re: Anybody know kernel tweak stuff for a Digital HX or ZX P6? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Thu, 24 Jul 1997, Jaye Mathisen wrote: > > I have a 4 processor P6 box I was thinking of converting from Solaris to > FreeBSD-SMP, but am not intimately familiar with the hardware enough to > know what the various kernel options (APIC related I guess) should be > set to. > Here are the options for SMP: options SMP options APIC_IO options SMP_AUTOSTART Those should get you going. Keep in mind that SMP support is only in 3.0-current. It is stable as of yesterday, but I know Steve is working on doing some changes on the locks, so it has the potential to become unstable in the coming weeks. Kyle Mestery StorageTek's Network Systems Group 7600 Boone Ave. N., Brooklyn Park, MN 55428 mesteka@anubis.network.com, mestery@winternet.com From owner-freebsd-smp Fri Jul 25 12:06:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA18732 for smp-outgoing; Fri, 25 Jul 1997 12:06:03 -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 MAA18720 for ; Fri, 25 Jul 1997 12:05:55 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id NAA27306 for ; Fri, 25 Jul 1997 13:05:47 -0600 (MDT) Message-Id: <199707251905.NAA27306@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: HEADS UP: cleanup coming. Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 25 Jul 1997 13:05:46 -0600 Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, Just got another of the "broken machines" on its feet. I think that except for a possible slowdown for isa drives, all the other concerns have been meet. I want to do a cleanup/purge of alot of legacy stuff in the tree. I need to hear ASAP from anyone who still has problems with the code of the last several weeks. Once I purge these old algorithms it will be much harder to fall back to them for comparison testing. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sat Jul 26 08:46:17 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id IAA04766 for smp-outgoing; Sat, 26 Jul 1997 08:46:17 -0700 (PDT) Received: from mhub1.tc.umn.edu (0@mhub1.tc.umn.edu [128.101.131.51]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id IAA04757 for ; Sat, 26 Jul 1997 08:46:12 -0700 (PDT) Received: from gold.tc.umn.edu by mhub1.tc.umn.edu; Sat, 26 Jul 97 10:46:08 -0500 Received: from pub-14-b-30.dialup.umn.edu by gold.tc.umn.edu; Sat, 26 Jul 97 10:46:07 -0500 Date: Sat, 26 Jul 1997 10:43:57 -0500 (CDT) From: dave adkins To: smp@FreeBSD.ORG cc: David A Adkins Subject: Re: ide and SMP current (07/26/97) In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, Looks like some minor problems with wd when used with current as of 07/26/97 (apic_vector.s v1.11). I'm configured with / on an IDE drive and all other partions on a NCR53c810. When the IDE controller is configured 0x80008000 I see the following: Jul 26 11:18:50 samthedog /kernel.test: wd2: interrupt timeout: Jul 26 11:18:55 samthedog /kernel.test: wd2: status 58 error 0 Jul 26 11:19:04 samthedog /kernel.test: wd2: interrupt timeout: Jul 26 11:19:04 samthedog /kernel.test: wd2: status 58 error 1 The system continues to run but stalls periodically on wd2 accesses. During the stalls systat shows the interrupt load go to nearly 50%. I assume thats because the system is stalled in the wd driver. If I reconfigure to use multi-sector transfers (0x80ff80ff) then I can run normally with very infrequent timeout messages. The problem still exists with multi-sector transfers enabled, its just less frequent. Possibly a race in the driver? I've got drives on wcd0 and wcd1 but I'm only mounting wd2. When I mount the drives on wcd0 they have the same problem in single-sector transfer mode. I originally switched to single-sector in an attempt to minimize silo overflows and never changed it back. The motherboard is a tyan S1563D. There were no timeouts with apic_vector.s V1.10. dave From owner-freebsd-smp Sat Jul 26 09:04:22 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA05460 for smp-outgoing; Sat, 26 Jul 1997 09:04:22 -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 JAA05454 for ; Sat, 26 Jul 1997 09:04:18 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id KAA04757; Sat, 26 Jul 1997 10:04:07 -0600 (MDT) Message-Id: <199707261604.KAA04757@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: dave adkins cc: smp@FreeBSD.ORG Subject: Re: ide and SMP current (07/26/97) In-reply-to: Your message of "Sat, 26 Jul 1997 10:43:57 CDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sat, 26 Jul 1997 10:04:07 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, edit i386/include/smptests.h, undef'ing PEND_INTS, then rebuild kernel. let me know if that helps. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sat Jul 26 09:55:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA07195 for smp-outgoing; Sat, 26 Jul 1997 09:55:03 -0700 (PDT) Received: from mhub2.tc.umn.edu (0@mhub2.tc.umn.edu [128.101.131.52]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id JAA07188 for ; Sat, 26 Jul 1997 09:55:00 -0700 (PDT) Received: from gold.tc.umn.edu by mhub2.tc.umn.edu; Sat, 26 Jul 97 11:54:24 -0500 Received: from pub-15-c-177.dialup.umn.edu by gold.tc.umn.edu; Sat, 26 Jul 97 11:54:23 -0500 Date: Sat, 26 Jul 1997 11:52:13 -0500 (CDT) From: dave adkins Reply-To: dave adkins To: smp@csn.net cc: smp@FreeBSD.ORG Subject: ide and PEND_INTS Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk steve, With PEND_INTS disabled in smptest.h the wd driver works (no timeouts) without multi-sector transfer enabled. dave From owner-freebsd-smp Sat Jul 26 10:14:07 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA07958 for smp-outgoing; Sat, 26 Jul 1997 10:14:07 -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 KAA07929 for ; Sat, 26 Jul 1997 10:13:42 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id LAA05033; Sat, 26 Jul 1997 11:13:30 -0600 (MDT) Message-Id: <199707261713.LAA05033@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: dave adkins cc: smp@FreeBSD.ORG Subject: Re: ide and PEND_INTS In-reply-to: Your message of "Sat, 26 Jul 1997 11:52:13 CDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sat, 26 Jul 1997 11:13:29 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > With PEND_INTS disabled in smptest.h the wd driver works (no timeouts) > without multi-sector transfer enabled. I understand the problem and am working on a fix. For now anyone using ISA cards (including the onboard ide controller) that generate INTs will have to #undef PEND_INTS in i386/isa/smptests.h. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sat Jul 26 11:00:55 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA10533 for smp-outgoing; Sat, 26 Jul 1997 11:00:55 -0700 (PDT) Received: from uhf.wdc.net (uhf.wdc.net [198.147.74.44]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA10525 for ; Sat, 26 Jul 1997 11:00:50 -0700 (PDT) Received: from localhost (bad@localhost) by uhf.wdc.net (8.8.5/8.6.12) with SMTP id OAA01123; Sat, 26 Jul 1997 14:03:33 -0400 (EDT) Date: Sat, 26 Jul 1997 14:03:31 -0400 (EDT) From: Bernie Doehner X-Sender: bad@uhf.wdc.net To: dave adkins cc: smp@FreeBSD.ORG Subject: Re: ide and SMP current (07/26/97) In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Looks like some minor problems with wd when used with current as of > 07/26/97 (apic_vector.s v1.11). I'm configured with / on an IDE drive and > all other partions on a NCR53c810. When the IDE controller is configured > 0x80008000 I see the following: > > Jul 26 11:18:50 samthedog /kernel.test: wd2: interrupt timeout: > Jul 26 11:18:55 samthedog /kernel.test: wd2: status 58 > error 0 > Jul 26 11:19:04 samthedog /kernel.test: wd2: interrupt timeout: > Jul 26 11:19:04 samthedog /kernel.test: wd2: status 58 > error 1 I get that ocassionaly with my 486 Acer AP43 board when running fxtv with a Wincast/TV board. fxtv uses DMA to talk to the video card. Are you running something that hits the PCI bus heavily? Bernie From owner-freebsd-smp Sat Jul 26 11:24:52 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA11498 for smp-outgoing; Sat, 26 Jul 1997 11:24:52 -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 LAA11493 for ; Sat, 26 Jul 1997 11:24:48 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.6/8.8.5) with ESMTP id MAA05394; Sat, 26 Jul 1997 12:24:33 -0600 (MDT) Message-Id: <199707261824.MAA05394@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Bernie Doehner cc: dave adkins , smp@FreeBSD.ORG Subject: Re: ide and SMP current (07/26/97) In-reply-to: Your message of "Sat, 26 Jul 1997 14:03:31 EDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sat, 26 Jul 1997 12:24:33 -0600 Sender: owner-freebsd-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, this is definately a problem with the newest SMP code. I know what the problem is, but don't have time to fix it today. Removing PEND_INTS from smptests.h will correct it for now. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD