Bug 2187644
| Summary: | SEV-es guest failed to find the boot disk when reseting nvram | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 9 | Reporter: | zixchen |
| Component: | edk2 | Assignee: | Gerd Hoffmann <kraxel> |
| Status: | CLOSED NOTABUG | QA Contact: | zixchen |
| Severity: | medium | Docs Contact: | |
| Priority: | medium | ||
| Version: | 9.3 | CC: | berrange, jinzhao, kraxel, pbonzini, vgoyal, virt-maint, xuwei, zhguo |
| Target Milestone: | rc | Keywords: | CustomerScenariosInitiative, Triaged |
| Target Release: | --- | ||
| Hardware: | x86_64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | If docs needed, set a value | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2023-06-27 03:56:58 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
|
Description
zixchen
2023-04-18 09:58:43 UTC
Relevant upstream discussion: https://edk2.groups.io/g/devel/message/103023 Seems the root cause is memory encryption attributes for variable flash can be wrong. (In reply to Gerd Hoffmann from comment #2) > Relevant upstream discussion: Oops, wrong thread, here is the real one: https://edk2.groups.io/g/devel/message/102098 Status update: edk2-stable202305 released without upstream fix. Increased risk this will miss the 9.3 boat. Testbuild: https://kojihub.stream.centos.org/koji/taskinfo?taskID=2275934 Any change in behavior? (In reply to Gerd Hoffmann from comment #6) > Testbuild: > https://kojihub.stream.centos.org/koji/taskinfo?taskID=2275934 > Any change in behavior? No change for this, virsh start $name --reset-nvram still failed with boot option restoration and then VM quits. Version: kernel-5.14.0-319.el9.x86_64 qemu-kvm-8.0.0-4.el9.x86_64 edk2-ovmf-20230301gitf80f052277c8-4.el9.bz2187644.20230601.1109.noarch Upload a firmware log: nvram_reset.log Is this a regression? What happens if you try boot the VM twice, first time with --reset-nvram and second time without? (In reply to Gerd Hoffmann from comment #9) > Is this a regression? It is not a regression, I test edk2-ovmf-20210527gite1999b264f1f-2.el9.noarch, and the issue reproduces. > What happens if you try boot the VM twice, first time with --reset-nvram and > second time without? The second time boot with #virsh start $guest_name, the VM can boot. After reset nvram of sev-es VM, the guests boot to "Boot Option Restoration" screen, if you select 'system reset' or wait for timeout to reset system, the VM will quit. If you select continue boot, then the guest van boots to the system. Sev-es doesn't support reboot, is that the reason why VM quit after resting nvram? Does this mean sev-es VM doesn't support reset nvram? (In reply to zixchen from comment #10) > (In reply to Gerd Hoffmann from comment #9) > > Is this a regression? > It is not a regression, I test > edk2-ovmf-20210527gite1999b264f1f-2.el9.noarch, and the issue reproduces. > > > What happens if you try boot the VM twice, first time with --reset-nvram and > > second time without? > The second time boot with #virsh start $guest_name, the VM can boot. > > > After reset nvram of sev-es VM, the guests boot to "Boot Option Restoration" > screen, if you select 'system reset' or wait for timeout to reset system, > the VM will quit. If you select continue boot, then the guest van boots to > the system. Sev-es doesn't support reboot, is that the reason why VM quit > after resting nvram? I think this is exactly what is happening. guest restores boot options. guest wants reboot, but because sev-es doesn't support that the vm is powered off instead. When starting it again it boots fine (using the boot options restored on first start). > Does this mean sev-es VM doesn't support reset nvram? resetting vmram works. But you need to start the vm twice because sev-es does not support reboots. I think you should update the test case accordingly. Thanks Gerd for confirming, I update the reset nvram case, and close this as not a bug. |