Date: Mon, 30 Apr 2001 22:10:30 +0000 (GMT) From: Terry Lambert <tlambert@primenet.com> To: ohartman@klima.physik.uni-mainz.de (Hartmann, O.) Cc: freebsd-stable@FreeBSD.ORG, freebsd-smp@FreeBSD.ORG Subject: Re: Spontanous reboot of SMP system and FBSD 4.3 Message-ID: <200104302210.PAA12613@usr01.primenet.com> In-Reply-To: <Pine.BSF.4.33.0104292327270.83323-100000@klima.physik.uni-mainz.de> from "Hartmann, O." at Apr 29, 2001 11:42:13 PM
next in thread | previous in thread | raw e-mail | index | archive | help
> One of our server systems, a TYAN Thunder 2500 ServerWorks Set III HE > based dual PIII based machine with AMI Enterprise 1600 RAID and 2GB > RAM seems to reboot spontanously without any reason. This is the first > time I realise this without having APM enabled into the kernel, which > has been detected to be a reason for spontanous reboots on boards > without APM to us. > > Well, in the time this machine rebooted itself there was no load, no > extraordinary activity or heavy load. I realised this this morning > checking the logs. > > Several days ago someone asked in the list for stability issues concerning > SMP on several architectures and I responded that our TYAN Thunder 2500 > based boards have run for a long time. I think I have to revise this > statement. I hope it is a simple problem with the kernel config because > the selected hardware, power supplies and UPS system we use is first > choice for this system and showed no problems before. > > Does anyone know, whether the kernel options AUTO_EOI1 and AUTO_EOI2 > do have serious impacts on stability? On this machine we configured kernel > with both options "ON" and have had no problems within the last 9 > months. I know that many system boards do not work with both options > set, but the TYAN Thunder 2500 seems to do - but could this be the reason > for spontanous reboots? I have seen a number of issues with Tyan motherboards, most recently, the Tyan Tiger. I suggest you rebuild your kernel with DDB and BREAK_TO_DEBUGGER options, and do a traceback next time there is a problem. If this is a true "spontaneous reboot", you may want to tell us the load you are placing on the system, and the exact version of FreeBSD you are using. In particular, bad things can happen if you use more than 32,767 sockets on 4.3-RELEASE (among other load related issues, such as running out of mbuf clusters). On some of the Tyan Tiger boards I've fought withm the system becomes extremely unstable because the invltlb() call after switching to 4M pages is not correctly communicated to the I/O APIC, and unless you have enough memory allocations to force all 8 4M entries out, it can lose its mind, relative to the TLB of one or more of the main CPUs. If this is the case, using the DISABLE_PSE option might help you as well (this seems to be a bug in the BIOS programming of the I/O APIC; I haven't tracked it past there). Changing the code to cause an "unnecessary" reload of control register 3 could help things for you, assuining you are mapping large (4M) device memory or something like that... Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200104302210.PAA12613>