Re: Part II: Running SMP kernel but only one cpu
On Mon, May 4, 2009 at 11:49 PM, Duane du...@cheekymonkey.us wrote: On 5/4/09, Tim Judd taj...@gmail.com wrote: IRQ15 is typically your secondary IDE controller; but due to PCI (or E-ISA) plugplay, including the PnP the BIOS may setup, lots of others can be on that bus too. This box has one SCSI card running two SCSI drives. The IDE's are disabled in the BIOS. But the SCSI card does feature in the problem: # grep -i irq15 /var/run/dmesg.boot No 'irq15' is found. One can see that the interrupt storm begins when the SCSI drives begin to spin up, IF the machine is booting with two cpus initialized. dmesg.boot is attached for everyone's edification and amusement! Another interesting datapoint is that if the machine is booted in Safe Mode the interrupt storm disappears, but so does the second cpu. Best regards, -- Duane I've seen the IRQ storms when an IDE channel is disabled too. And to silence it I had to enable the channel (using an IRQ), with no devices on it. it's gonna be a chatterbox, and the question is if the system can work with a shortage of up to 2 IRQs (irq 14 = primary IDE, irq 15 = secondary IDE) Especially early 586 and 686 classes I saw that often. Would you try enabling the IDE channels and see if the IRQ storms stop? --TJ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Part II: Running SMP kernel but only one cpu
On 5/5/09, Tim Judd taj...@gmail.com wrote: Would you try enabling the IDE channels and see if the IRQ storms stop? There is joy in Mudville. g I couldn't really change the IDE settings that much since there are no IDE drives in this thing. The channels are all on 'auto' and that yields a NONE for each of them (iirc, it's a kinda late). But I did turn on some stuff that was off in the BIOS: serial port A, the sound card, and the parallell port. I know it's not good engineering practice to change a bunch of stuff at once, but there it is. Whatever I did along the lines of turning stuff *on* seems to have done the trick! Thanks all! -- Duane ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Part II: Running SMP kernel but only one cpu
The bios in this old Micron dual PPro-180 full tower antique only initializes the second CPU if the machine is cold-booted. A simple 'reboot' results in a single processor machine regardless of the kernel that is launched. This fact -- unknown to me before last night -- was the source of a great deal of lost time! My 6.4 SMP kernel (now customized) runs just fine, with both cpus active, *except* for this message streaming constantly up the boot console. (from /var/log/messages:) May 4 20:20:33 poobah kernel: interrupt storm detected on irq15:; throttling interrupt source May 4 20:21:02 poobah last message repeated 42 times May 4 20:21:03 poobah login: ROOT LOGIN (root) ON ttyv1 May 4 20:21:03 poobah kernel: interrupt storm detected on irq15:; throttling interrupt source May 4 20:21:33 poobah last message repeated 30 times May 4 20:23:33 poobah last message repeated 120 times May 4 20:33:33 poobah last message repeated 599 times May 4 20:40:01 poobah last message repeated 387 times etc etc ad repetitum infinitum Question1: Is this something I should go to some lengths to eliminate? Question2: What the heck is it? Best regards, -- Duane ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Part II: Running SMP kernel but only one cpu
Duane wrote: The bios in this old Micron dual PPro-180 full tower antique only initializes the second CPU if the machine is cold-booted. A simple 'reboot' results in a single processor machine regardless of the kernel that is launched. This fact -- unknown to me before last night -- was the source of a great deal of lost time! Good you found a work around. You might try ensuring your running latest avail bios as well. My 6.4 SMP kernel (now customized) runs just fine, with both cpus active, *except* for this message streaming constantly up the boot console. (from /var/log/messages:) May 4 20:20:33 poobah kernel: interrupt storm detected on irq15:; throttling interrupt source May 4 20:21:02 poobah last message repeated 42 times May 4 20:21:03 poobah login: ROOT LOGIN (root) ON ttyv1 May 4 20:21:03 poobah kernel: interrupt storm detected on irq15:; throttling interrupt source May 4 20:21:33 poobah last message repeated 30 times May 4 20:23:33 poobah last message repeated 120 times May 4 20:33:33 poobah last message repeated 599 times May 4 20:40:01 poobah last message repeated 387 times etc etc ad repetitum infinitum Question1: Is this something I should go to some lengths to eliminate? Yes, it's probably something you should eliminate. Question2: What the heck is it? A poor explanation is the devices are fighting over an IRQ. Generally, simplest fix is to find what devs are on that IRQ, and manually reassign one dev to a different IRQ. Best regards, ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Part II: Running SMP kernel but only one cpu
On Mon, May 4, 2009 at 6:46 PM, Duane du...@cheekymonkey.us wrote: The bios in this old Micron dual PPro-180 full tower antique only initializes the second CPU if the machine is cold-booted. A simple 'reboot' results in a single processor machine regardless of the kernel that is launched. This fact -- unknown to me before last night -- was the source of a great deal of lost time! My 6.4 SMP kernel (now customized) runs just fine, with both cpus active, *except* for this message streaming constantly up the boot console. (from /var/log/messages:) May 4 20:20:33 poobah kernel: interrupt storm detected on irq15:; throttling interrupt source May 4 20:21:02 poobah last message repeated 42 times May 4 20:21:03 poobah login: ROOT LOGIN (root) ON ttyv1 May 4 20:21:03 poobah kernel: interrupt storm detected on irq15:; throttling interrupt source May 4 20:21:33 poobah last message repeated 30 times May 4 20:23:33 poobah last message repeated 120 times May 4 20:33:33 poobah last message repeated 599 times May 4 20:40:01 poobah last message repeated 387 times etc etc ad repetitum infinitum Question1: Is this something I should go to some lengths to eliminate? Question2: What the heck is it? Best regards, -- Duane IRQ15 is typically your secondary IDE controller; but due to PCI (or E-ISA) plugplay, including the PnP the BIOS may setup, lots of others can be on that bus too. Likely candidates are PCI devices, such as modems, NICs, sound cards, etc I think you'd be able to find what's on IRQ15 by a simple: # grep -i irq15 /var/run/dmesg.boot You will probably not be able to pull your secondary IDE controller off 15. The possible other device that's been configured for irq15 might stop if you disable PnP OS in the BIOS (if it exists), or setting irq15 to the equivelant of 'reserved' in the BIOS might aleviate the problem. Good luck. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Part II: Running SMP kernel but only one cpu
On 5/4/09, Tim Judd taj...@gmail.com wrote: IRQ15 is typically your secondary IDE controller; but due to PCI (or E-ISA) plugplay, including the PnP the BIOS may setup, lots of others can be on that bus too. This box has one SCSI card running two SCSI drives. The IDE's are disabled in the BIOS. But the SCSI card does feature in the problem: # grep -i irq15 /var/run/dmesg.boot No 'irq15' is found. One can see that the interrupt storm begins when the SCSI drives begin to spin up, IF the machine is booting with two cpus initialized. dmesg.boot is attached for everyone's edification and amusement! Another interesting datapoint is that if the machine is booted in Safe Mode the interrupt storm disappears, but so does the second cpu. Best regards, -- Duane dmesg.boot Description: Binary data ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org