Description of problem: Cannot boot any 2.6 kernel on compaq proliant 8500R's. First i upgraded a fc1 system to fc2 via yum/apt and then when I rebooted it gets to the point where it says: mice: ps2 mouse drive common for all devices ( or something ) and then it reboots... It doesn't get far enough to boog to single or find my scsi drives or free unused kernel memory or anything Version-Release number of selected component (if applicable): How reproducible: I've tried it on two machines and done the following. Steps to Reproduce: 1. Tried with fc2 boot.iso, fc2-disk1, fc2-rescue cd and a gentoo 2.6.5-livecd all fail in the exact same place, except that when it fails with gentoo it doesn't reboot, it just permanently hangs the machine. 2. I've tried passing the following options nousb, noprobe, noapic, nofb, text, skipddc, acpi=off, pci=off, pci=biosirq, ide=nodma 3. Actual results: No changes... Expected results: I expected it to boot. Additional info:
Has this been checked on or reproduced? I can reproduce in-house on three different Compaq 8500R machines. These are fairly common servers so I expect that this could be verified pretty easily.
I can reproduce on our 8500s, as well. I used a four processor (pent III, 550), 4 gig machine currently running FC1. I tried a normal upgrade using the CDs and I also copied the initrd.img and vmlinuz files from /isolinux and configured grub so I could boot locally. Both attempts hang at the same place. With the CD I see a normal boot to: NET: Registered protocol family 2 IP: routing cache hash table of 32768 buckets, 256Kbytes Unable to handle kernel paging request at virtual address c16f0018 From grub I see the hang in the same place with the same message, but continues: printing eip: c01301ff *pde = 014001e3 Oops: 0009 [#1] CPU: 0 EIP: 0060:[<c01301ff>] Not tainted followed by a register dump and stack trace.
can you get the full dump & trace please ?
NET: Registered protocol family 2 IP: routing cache hash table of 32768 buckets, 256Kbytes Unable to handle kernel paging request at virtual address c16f0018 printing eip: c013201ff *pde = 014001e3 Oops: 0009 [#1] CPU: 0 EIP: 0060:[<c01301ff>] Not tainted EFLAGS: 00010002 (2.6.5-1.358) EIP is at __rmqueue+0x38/0x102 eax: 00000078 ebx: 0000000a ecx: c16f0018 edx: c02d1d44 esi: 00000000 edi: c02d1dbc ebp: 0000000a esp: f7d90f54 ds: 007b es: 007b ss: 0068 Process swapper (pid: 1, threadinfo=f7d90000 task=c304f630) Stack: 0000003c c02d1e08 0000000a c02d1d44 c02d1d44 00000000 00000206 0000000a c0130579 00000000 00000220 c02d1d44 c02d1d44 00000000 c02d1f8c 00000220 c01306b9 00000001 c304f630 00000000 0000000a 00000000 0000000a 00000000 Call Trace: [<c0130579>] buffered_rmqueue+0xa2/0x147
Adding i8042.nomux to the boot prompt gets around this problem.
A patch for this problem has been has been merged in the upstream kernel a few months ago. It should be fixed in the latest kernels from Arjan's page: http://people.redhat.com/arjanv/2.5/ (IIRC)
should be fixed in the current tree. For FC2, use the workaround in comment #5