Re: [edk2-devel] [PATCH v2 0/5] Fix OvmfXen HVM Direct kernel boot failure

[Jim Fehlig <jfehlig@suse.com>]

On 8/15/21 7:34 PM, Gary Lin (HPS OE-Linux TDC) wrote:
> On Fri, Aug 13, 2021 at 09:55:48AM +0000, Yao, Jiewen wrote:
>> HI Gary
> Hi Jiewen,
> 
>> Several comment:
>>
>> 1) According to our process, we need a Bugzilla. Would you please file it?
>>
> Okay, will create a bugzilla entry to track the issue.
> 
>> 2) I do not understand how HVM direct kernel boot is related to S3 enabling.
>> It seems 1/3/4/5 are for S3, while 2 is for missing driver.
>> Should we split them to 2 patch set?
>>
> Actually the inconsistency between QemuFwCfgS3Enabled() and
> PcdAcpiS3Enable casued the boot failure of HVM direct kernel boot.
> S3SaveStateDxe checked PcdAcpiS3Enable(=FALSE) and skipped S3BootScript.
> On the other hand, PlatformBootManagerBeforeConsole() invoked
> QemuFwCfgS3Enabled()(=TRUE) and tried to locate S3BootScript. Since
> S3BootScript wasn't installed, it failed with EFI_NOT_FOUND and stopped
> the booting. So Patch 1 and 2 are mainly for the booting issues in
> OvmfXen. For Patch 3~5, I just try to eliminate the inconsistency to
> avoid the potential problems.
> 
> I can split them into 2 or 3 patch sets if necessary.
> 
>> 3) Does the S3 issue only happen in direct kernel boot? Or is it a generic issue.
>>
>> 4) Have you validated non direct kernel boot to ensure it still works?
>>
> Jim,
> 
> Did you encounter any problem with non direct kernel boot VM?

No problems encountered, works fine in my testing. Although I did not test these 
patches specifically, only the ones in the openSUSE Factory ovmf package.

Regards,
Jim