On Tue, 4 Feb 2020 14:51:39 +1100 Gavin Shan <gs...@redhat.com> wrote:
[...] > The mechanism I figured out is to inject SError to guest, as below snippet > shows. It > helps to get a panic and guest rebooting, which looks similar to what x86/p > pc have. I think being able to inject a SError is a reasonable thing to have for QEMU. After all, the RAS spec[1] says that "An SError interrupt can also be generated for IMPLEMENTATION DEFINED causes." What I'm more concerned about is how this is going to coexist with the RAS extension itself (which is mandated with ARMv8.2), and whether we should classify this with a particular syndrome (by populating the ESR field in the event injection structure), if only to be able to distinguish the SW-injected SError from a RAS error propagated by KVM. It looks to me that this SError could be described as a "Recoverable Error", but I'd like someone who knows more about the RAS extension to comment on this. Thanks, M. [1] https://static.docs.arm.com/ddi0587/cb/2019_07_05_DD_0587_C_b.pdf -- Jazz is not dead. It just smells funny...