On 03/08/2012 03:27 AM, Yinghai Lu wrote: > 2012/3/7 Vivek Goyal<vgoyal at redhat.com>: >> On Wed, Mar 07, 2012 at 07:53:16PM +0900, Fernando Luis V?zquez Cao wrote: >>> We run into the same NMI problems and wrote some patches that tackle >>> the kernel boot side of things. They have been extensively tested using >>> qemu-kvm and things seem to be working as expected (after receiving an >>> early NMI the kernel continues without problem; after the iret there is no >>> stack corruption or register corruption). >> What happens if NMI happens while we are still in purgatory code? The system triple faults. As Eric said we need to fix both the reboot path to kdump and kdump kernel's boot code. The two patches I sent address the later. I sent these first because I had time to test them properly, fix a real issue and can be applied independently. > yes, you are right. that is too late to set that in arch/x86/kernel/head64.c Of course, the reboot path needs to be fixed too. Now that I solved some problems I was having with my qemu-kvm test environment (NMIs injected through the qemu monitor only get devivered to the BSP, etc), I will make sure that the code I wrote for it works as expected. In the meantime, I would appreciate it if you could review/ack yesterday's patches. Thanks, Fernando
