Re: [PATCH RFC 0/7] revert RNG seed mess

[Dov Murik]

On 09/02/2023 8:03, Dov Murik wrote:
> Hi Michael,
> 
> On 08/02/2023 23:12, Michael S. Tsirkin wrote:
>> All attempts to fix up passing RNG seed via setup_data entry failed.
>> Let's just rip out all of it.  We'll start over.
>>
>>
>> Warning: all I did was git revert the relevant patches and resolve the
>> (trivial) conflicts. Not even compiled - it's almost midnight here.
>>
>> Jason this is the kind of approach I'd like to see, not yet another
>> pointer math rich patch I need to spend time reviewing. Just get us back
>> to where we started. We can redo "x86: use typedef for SetupData struct"
>> later if we want, it's benign.
>>
>> Could you do something like this pls?
>> Or test and ack if this patchset happens to work by luck.
>>
>> Michael S. Tsirkin (7):
>>   Revert "x86: don't let decompressed kernel image clobber setup_data"
>>   Revert "x86: do not re-randomize RNG seed on snapshot load"
>>   Revert "x86: re-initialize RNG seed when selecting kernel"
>>   Revert "x86: reinitialize RNG seed on system reboot"
>>   Revert "x86: use typedef for SetupData struct"
>>   Revert "x86: return modified setup_data only if read as memory, not as
>>     file"
>>   Revert "hw/i386: pass RNG seed via setup_data entry"
>>
>>  include/hw/i386/microvm.h |   5 +-
>>  include/hw/i386/pc.h      |   3 -
>>  include/hw/i386/x86.h     |   3 +-
>>  include/hw/nvram/fw_cfg.h |  31 ----------
>>  hw/i386/microvm.c         |  17 ++----
>>  hw/i386/pc.c              |   4 +-
>>  hw/i386/pc_piix.c         |   2 -
>>  hw/i386/pc_q35.c          |   2 -
>>  hw/i386/x86.c             | 122 ++++++++++----------------------------
>>  hw/nvram/fw_cfg.c         |  21 ++-----
>>  10 files changed, 49 insertions(+), 161 deletions(-)
>>
> 
> 
> I tested this series with SEV measured boot using AmdSev OVMF.  The boot
> succeeds, and the hashes computed for kernel/initrd/cmdline are
> identical to the ones computed by qemu 7.1.0, which are the hashes that
> the guest owner would expect.
> 

I also tested that backporting this series to 7.2.0 (skipping
PATCH 1/7 because it was added after the 7.2.0 release) works OK for
booting SEV guest with AmdSev measured boot (and retains the same
kernel/initrd/cmdline hashes as qemu 7.1.0).

-Dov


> As for non-EFI (non-SEV) guests, I did a very simple test of starting a
> non-EFI guest and it looks OK, but more testing is needed.
> 
> So:
> 
> Tested-by: Dov Murik <dovmurik@linux.ibm.com>
> 
> 
> Thank you!
> 
> -Dov