Unable to handle kernel paging request

Hi,

booting a FC6 on my intel box (xeon) with a kernel 2.6.22.5 and KVM git, I had the following error (not reproducible):

# kvm-userspace/qemu/x86_64-softmmu/qemu-system-x86_64 -hda fc6.qcow2 -net nic
-net tap -serial stdio -smp 4
...
INIT: version 2.86 booting
                Welcome to Fedora Core
                Press 'I' to enter interactive startup.
Setting clock  (utc): Thu Sep 27 18:06:27 CEST 2007 [  OK  ]
Starting udev: Unable to handle kernel paging request at ffffffff880e9000 RIP:
 [<ffffffff8104ebc1>] sys_init_module+0x985/0x1786
PGD 203067 PUD 205063 PMD 7fc4067 PTE 6b50163
Oops: 0002 [1] SMP
CPU 1
Modules linked in: dm_snapshot dm_zero dm_mirror dm_mod ata_piix libata sd_mod s
csi_mod ext3 jbd mbcache ehci_hcd ohci_hcd uhci_hcd
Pid: 641, comm: modprobe Not tainted 2.6.22.5 #1
RIP: 0010:[<ffffffff8104ebc1>]  [<ffffffff8104ebc1>] sys_init_module+0x985/0x178
6
RSP: 0018:ffff810006399e68  EFLAGS: 00010246
RAX: 0000000000000000 RBX: ffffc20000090a20 RCX: 0000000000000f84
RDX: ffffffff880e6000 RSI: 0000000000000163 RDI: ffffffff880e9000
RBP: 0000000000000026 R08: ffff810007d94254 R09: 00000000000050cf
R10: 0000000000000000 R11: 0000000000000001 R12: ffffc2000007c300
R13: 0000000000000004 R14: ffffc200000900e0 R15: 00002ab868ac2010
FS:  00002ab8690096e0(0000) GS:ffff810007d94280(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: ffffffff880e9000 CR3: 0000000007cfd000 CR4: 00000000000006e0
Process modprobe (pid: 641, threadinfo ffff810006398000, task ffff81000788e000)
Stack:  00000000000276d8 0000000000000000 0000000000000000 000000000608f340
 0000000000000000 0000000000000000 0000000000000000 0000000000000000
 0000000000000000 0000000000000000 ffffc200000a2328 ffff810006402000
Call Trace:
 [<ffffffff8108e67d>] vfs_read+0xcb/0x173
 [<ffffffff81009b9e>] system_call+0x7e/0x83

Code: f3 aa 49 89 94 24 88 01 00 00 49 8b bc 24 90 01 00 00 e8 32
RIP  [<ffffffff8104ebc1>] sys_init_module+0x985/0x1786
 RSP <ffff810006399e68>

CR2: ffffffff880e9000

The instruction at [<ffffffff8104ebc1>] sys_init_module+0x985/0x1786 is:

0xffffffff8104ebc1 <sys_init_module+2437>: rep stos %al,%es:(%rdi)

Any idea of what happened ?

Laurent

------------- Laurent.Vivier-6ktuUTfB/[email protected]  --------------
          "Software is hard" - Donald Knuth

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

_______________________________________________
kvm-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel

Correct me if I'm wrong, perhaps it's the bug Nitin is hunting.

There is always something wrong in the management in the REP prefix.

I think what is happen is:

- we have a REP prefix, we save ECX and EIP.
- we set c->dst to emulate a "stos"
- goto writeback
- writeback: we try a "write_emulated()" with c->dst
- write_emulated failed() AND WE DON'T RESTORE ECX AND EIP -> it's bad...
- exit to QEMU
- re-enter in x86_emulate_insn() with already modified ECX and EIP.

Any comment ?
(Yes, I know, it's again another bug I've introduced into KVM...)

Laurent

------------- Laurent.Vivier-6ktuUTfB/[email protected]  --------------
          "Software is hard" - Donald Knuth

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

_______________________________________________
kvm-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel

To avoid this, I suggest that Nitin and yourself review each other's patches. While I review every patch I commit, it works much better when someone who's involved daily with the code reviews the patch.

error compiling committee.c: too many arguments to function


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

I agree...

Laurent

------------- Laurent.Vivier-6ktuUTfB/[email protected]  --------------
          "Software is hard" - Donald Knuth

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

_______________________________________________
kvm-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel