I i'm trying to install CentOS 5.3 in VMWare Server 2.x. The host system is running Windows 2000 SP4 with an AMD Athelon XP 1800 CPU, 1G RAM and an NVidia FX5200. Every time i try to install CentOS 5.3 the result is always the same, a kernel panic/kernel oops. I have other VM's running in the same host hardware without problem, but those VM's are running FC4 and previous RedHat versions.
As far i can diagnose the problem doesn't seams to be CentOS related. I believe this is true because i have tried to install CentOS 5.3 directly in the host system and i can reach the step where the partitions are created, stage that i can't reach when trying to install CentOS 5.3 in my VMWare Server 2.0 virtual machine.
I have tried to raise the RAM in use by the guest OS, but the problem persists. I have tried to disable acpi (setting the acpi=no as a boot option), doing the install using text mode, but the same problem, kernel panic.
Acording to the kernel panic diagnostic info the problem happens in function xor_sse_2.
The call trace shows the following:
-
apic_timer_interrupt
xor_sse_2
do_xor_speed
calibrate_xor_block
sys_init_module
syscall_call
Code: Bad EIP value
-
I have upploaded the kernel panic message as it has been printed in the VMWare Console.
I have tried to disable all kinds of acceleration in the VM configuration, changed the SCSI vitual controller type from BusLogic to LSI, but the same issue. I have searched for the functions reported in the call trace, and it seams that calibrate_xor_block is used inside module/driver md, so this must be a virtual hardware issue (disk, cdrom, etc). I have changed the virtual CDrom hardware from IDE to SCSI and now the guest OS can't find the CDRom nor the hard disk, but there is also no kernel panic!. Is there any LSI driver disk that can be used so the virtual SCSI cdrom and hardisk can be accessed?
Does any one has any hint?
Thanks
Mike