If I boot a KVM instance configured with 4 CPU's but pass the "nosmp" flag
Host kernel is 2.6.30-rc1, guest kernel is 2.6.29.1
It dies really early in boot, on a module I never use..
This system is open-source software. The exact distribution terms for
Press ESC to enter the Grub menu... 0 modprobe: FATAL: Could not load /lib/modules/2.6.29-1-amd64-vyatta/modules.dep: No such file or directory
Loading, please wait...
BUG: unable to handle kernel NULL pointer dereference at (null)
IP: [<ffffffff803d6b6b>] dma_channel_rebalance+0x46/0x1d1
PGD 1d8db067 PUD 179bc067 PMD 0
Oops: 0002 [#1] SMP
last sysfs file: /sys/kernel/uevent_seqnum
CPU 0
Modules linked in: async_tx(+) xor raid1 raid0 multipath linear md_mod virtio_blk sg sr_mod cdrom pata_acpi virtio_pci virtio_ring virtio floppy uhci_hcd usbcore ata_piix ata_generic thermal processor fan thermal_sys
Pid: 1789, comm: modprobe Not tainted 2.6.29-1-amd64-vyatta #1
RIP: 0010:[<ffffffff803d6b6b>] [<ffffffff803d6b6b>] dma_channel_rebalance+0x46/0x1d1
RSP: 0018:ffff880017a6de78 EFLAGS: 00010283
RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000001
RDX: 0000000000000001 RSI: 0000000000000001 RDI: 0000000000000000
RBP: 0000000000000000 R08: 0000000000000020 R09: ffffffff8063ab00
R10: ffffffffa0106d80 R11: ffff880001010300 R12: 000000000000000c
R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
FS: 00007fe9c51536e0(0000) GS:ffffffff806ac080(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 0000000000000000 CR3: 000000001d903000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process modprobe (pid: 1789, threadinfo ffff880017a6c000, task ffff88001fb14380)
Stack:
0000000000000000 ffffffff805fc4e8 0000000000001ee9 ffffffff805fc4e8
0000000000000000 0000000000000000 0000000000000000 ffffffff803d730a
00000000fffffffe ffffffffa0106d80 0000000000001ee9 ffffffffa0109000
Call Trace:
[<ffffffff803d730a>] dmaengine_get+0xca/0xdd
[<ffffffffa0109000>] async_tx_init+0x0/0x1b [async_tx]
[<ffffffffa0109009>] async_tx_init+0x9/0x1b [async_tx]
[<ffffffff80209051>] _stext+0x51/0x120
[<ffffffff8032bfde>] __up_read+0x13/0x8d
[<ffffffff8024aa4d>] __blocking_notifier_call_chain+0x51/0x5f
[<ffffffff80257c75>] sys_init_module+0xa0/0x1ba
[<ffffffff8020be5b>] system_call_fastpath+0x16/0x1b
Code: 53 48 83 ec 08 e8 ae 15 f5 ff 89 c3 83 f8 0c b8 0c 00 00 00 0f 4f d8 eb 5a 48 8b 04 ed e0 12 73 80 48 63 d1 48 f7 d0 48 8b 04 d0 <48> c7 00 00 00 00 00 8d 51 01 48 8b 3d b4 d3 09 00 be 20 00 00
RIP [<ffffffff803d6b6b>] dma_channel_rebalance+0x46/0x1d1
RSP <ffff880017a6de78>
CR2: 0000000000000000
---[ end trace af2ed99f087c36ac ]---
Killed
On Tue, Apr 14, 2009 at 9:18 PM, Stephen Hemminger
<[email protected]> wrote:
> If I boot a KVM instance configured with 4 CPU's but pass the "nosmp" flag
> Host kernel is 2.6.30-rc1, guest kernel is 2.6.29.1
> It dies really early in boot, on a module I never use..
>
Hi Stephen,
I'll try to reproduce this. In the meantime you can set
CONFIG_DMA_ENGINE=n as a workaround. The output of
scripts/markup_oops.pl may help...
Thanks for the report,
Dan