Red Hat Bugzilla – Bug 78559
Can't boot SMP kernel on UP machine
Last modified: 2008-08-01 12:22:53 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; .NET CLR 1.0.3705) Description of problem: Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Install stock kernel-smp-2.4.18-18.7.x.i686 2. on IBM netfinity 5100 Uniprocessor (scsi) 3. Reboot.... Actual Results: The server hung during the boot sequence Expected Results: Should be able to boot an SMP kernel on a Uni Processor server. Additional info: I installed the non-smp of the same kernel, and that worked fine, I also went back to a 2.4.18-10 smp kernel, and booted fine. I have always been able to run SMP kernels on UP machines (it is a requirement to be able to run on smp servers if we have a cpu fail) I also tested a few other varieties of i686 servers, and they cannot boot this version of the smp kernel either.
Me too, the boot hangs after the 2048 pty line. PowerEdge 1300 Single PIII 450 Intel 440BX mainboard (smap capable) 256MB ECC SDRAM 40GB IDE bootdrive on onboard IDE controller. Latest BIOS 2.4.19-smp succesful bootlog attached.
Created attachment 86484 [details] 2.4.19-smp succesful boot log.
More helpful info- boot fails at this line: Real Time Clock Driver
This sounds very similar to the bug I'm seeing with kernel 2.4.18-26.7.x - the SMP kernel hangs at boot on a number of UP systems where 2.4.18-26.7.x UP works fine, as does 2.4.18-10 SMP. I believe the common thing with the systems that fail is that they have an "IO-APIC", e.g. systems with SMP support but only one CPU installed, but also others. E.g. here is the final output on a Dell-1550 with one CPU installed: ... PCI->APIC IRQ transform: (B2,I5,P1) -> 28 PCI->APIC IRQ transform: (B2,I6,P0) -> 29 Linux NET4.0 for Linux 2.4 Based upon Swansea University Computer Society NET3.039 Initializing RT netlink socket Starting kswapd pt The modification in the diff below seems to fix or at least work around the problem, but I'm unsure whether it is the "proper" one - any comments appreciated. Thanks --Per Hedeland --- arch/i386/kernel/io_apic.c.ORIG 5 Mar 2003 11:24:51 -0000 +++ arch/i386/kernel/io_apic.c 19 Mar 2003 23:04:28 -0000 @@ -249,6 +249,8 @@ irq_balance_t *entry = irq_balance + irq; unsigned long now = jiffies; + if (smp_num_cpus == 1) + return; if (unlikely(entry->timestamp != now)) { unsigned long allowed_mask; int random_number;
Thanks for the bug report. However, Red Hat no longer maintains this version of the product. Please upgrade to the latest version and open a new bug if the problem persists. The Fedora Legacy project (http://fedoralegacy.org/) maintains some older releases, and if you believe this bug is interesting to them, please report the problem in the bug tracker at: http://bugzilla.fedora.us/