On Thu, Aug 22, 2013 at 06:05:58PM -0700, Jordan Justen wrote:
> On Mon, Aug 19, 2013 at 3:00 AM, Michael Chang <[email protected]> wrote:
> > On Sat, Aug 17, 2013 at 11:16:16PM -0700, Jordan Justen wrote:
> >> The volatile 'NvVars' variable indicates that the variables do
> >> not need to be loaded from the file again. After we write the
> >> variables out to the file, there is clearly no need to load
> >> them back from the file.
> >>
> >> Contributed-under: TianoCore Contribution Agreement 1.0
> >> Signed-off-by: Jordan Justen <[email protected]>
> >> Cc: Michael Chang <[email protected]>
> >> Cc: Laszlo Ersek <[email protected]>
> >> ---
> >> Michael,
> >>
> >> Does this patch fix the issue you highlighted (way back) on June
> >> 21st in your patch:
> >> * OvmfPkg/NvVarsFileLib: handle the inital file not found
> >
> > It seems not fix the problem if I use regular openSUSE 12.3
> > installation to a new virtual disk or chooses to recreate the ESP
> > if it already exists.
> >
> > However the manual test step works, if without your patch it fails. I
> > list the steps for reference.
> >
> > Fistly to simulate the initial (disk) condition.
> >
> > Delete NvVars File
> >  $ rm /boot/efi/NvVars
> > Delete openSUSE boot entry
> >  $ efibootmgr -b 0005 -B
> > Delete NvVars variables
> >  $ cd /sys/firmware/efi/vars
> >  $ cat NvVars-<GUID ..>/raw_var > del_var
> > Power off
> >  $ poweroff
> >
> > Now we can start vm to install the bootloader, use efi shell to
> > launch the installed grub2. In the booted system, use
> >  $ grub2-install
> >
> > Check if boot variable are written correctly
> >  $ efibootmgr -v
> >
> > Reboot and check if opensuse shows in the EFI boot manager ..
> >
> > So there could be some other corner cases, but I cannot tell it now.
> > I can continue to investigate it or do you have any other better
> > idea?
> 
> Do these failing scenarios work with your original patch?

My patch fails as well.

I think the problem is I'm using virt-intall, and there might be some
tricks on how libvirt restarts domain after installation finished that
may actually restart from power-down ? Laszlo did you aware any issue
reported for libvirt before ?

If I swich to qemu-kvm, your patch works to fix the problem.

The inconsistent result of virt-install wrt last time may blame to the
version downgraded a while ago after unsuccessful upgrade. Somehow I
may downgrad too much.

Here listed the version I'm testing.

$ virt-install --version
0.600.4

$ virsh --version
1.1.1

$ qemu-kvm --version
QEMU emulator version 1.3.0 (kvm-1.3.0-3.3.2) ...

$ uname -r
3.10.6-1.gbd99dce-desktop

After all I'm quite convinced your patch actually works.

> 
> One thing to note about our current hack runtime non-volatile variable
> support: If you power-down then any changes that were made at runtime
> under the OS will be lost. If you reboot then there is a chance that
> we can see the changes that were made to non-volatile variables within
> the VM RAM.

Thanks for your remind.

regards,
Michael
> 
> This all should be fixed by the 'flash' support of QEMU, if I ever
> take some time to get that support into OVMF.
> 
> -Jordan

------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and 
AppDynamics. Performance Central is your source for news, insights, 
analysis and resources for efficient Application Performance Management. 
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
edk2-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/edk2-devel

Reply via email to