Re: Use a pcd to control PlatformRecovery

["Gao, Zhichao" <zhichao.gao@intel.com>]

> -----Original Message-----
> From: Gao, Zhichao
> Sent: Wednesday, October 16, 2019 10:09 AM
> To: 'Wang, Sunny (HPS SW)' <sunnywang@hpe.com>; devel@edk2.groups.io;
> Ni, Ray <ray.ni@intel.com>
> Cc: Wang, Jian J <jian.j.wang@intel.com>; Wu, Hao A <hao.a.wu@intel.com>;
> Zeng, Star <star.zeng@intel.com>; Gao, Liming <liming.gao@intel.com>;
> Sean Brogan <sean.brogan@microsoft.com>; Michael Turner
> <Michael.Turner@microsoft.com>; Bret Barkelew
> <Bret.Barkelew@microsoft.com>; Li, Walon <walon.li@hpe.com>; Wei, Kent
> (HPS SW) <kent.wei@hpe.com>
> Subject: RE: Use a pcd to control PlatformRecovery
> 
> First of all, the patch didn't aim to change the other part of the boot flow
> except PlatformRecovery.
> 
> Local variable PlatformRecovery is controlled by OsIndications variable. When
> the EFI_OS_INDICATIONS_START_PLATFORM_RECOVERY is set, the
> firmware should try to platform specific recovery. But that doesn't mean the
> platform must support the specific recovery. i.e. local PlatformRecovery is
> controlling the boot flow and the Pcd just indicated whether the platform
> support recovery or not.
> So, on my opinion, I don't agree with your change.

After discuss with Sunny and Ray, and refer to the spec section 3.4.1 and 3.4.2, the OsRecovery and PlatformRecovery should always be operated regardless of the value of OsIndication variable if fail to boot the BootOrder. I am wrong. We should change to use the PcdPlatformRecoverySupport to control the PlatformRecovery. Please help to send a patch to fix it. Thanks a lot.

> 
> Default Platform Recovery refer to the short file path to boot the OS. If the
> firmware supports platform recovery, then *short file path* option would be
> one part of the PlatformRecovery#### in case there are no other boot
> options. If the firmware doesn't support platform recovery, we still need this
> default boot thru a short file path and it should not depend on the
> PlatformRecovery#### variable for the compatibility thinking.
> 
> Thanks,
> Zhichao
> 
> > -----Original Message-----
> > From: Wang, Sunny (HPS SW) [mailto:sunnywang@hpe.com]
> > Sent: Tuesday, October 15, 2019 4:53 PM
> > To: devel@edk2.groups.io; Gao, Zhichao <zhichao.gao@intel.com>; Ni,
> > Ray <ray.ni@intel.com>
> > Cc: Wang, Jian J <jian.j.wang@intel.com>; Wu, Hao A
> > <hao.a.wu@intel.com>; Zeng, Star <star.zeng@intel.com>; Gao, Liming
> > <liming.gao@intel.com>; Sean Brogan <sean.brogan@microsoft.com>;
> > Michael Turner <Michael.Turner@microsoft.com>; Bret Barkelew
> > <Bret.Barkelew@microsoft.com>; Li, Walon <walon.li@hpe.com>; Wei,
> Kent
> > (HPS SW) <kent.wei@hpe.com>; Wang, Sunny (HPS SW)
> <sunnywang@hpe.com>
> > Subject: Use a pcd to control PlatformRecovery
> >
> > Hi Zhichao and Ray,
> >
> > I have some questions about this code change. Sorry for being late to
> > bring my questions here.
> >
> > For now, the code block for iterating the PlatformRecovery####
> > variables is controlled by OsIndications variable. However, it looks
> > to me like that the PlatformRecovery#### should still be attempted for
> > the case of that processing of BootOrder does NOT result in success
> > (according to section 3.4 in UEFI 2.8). In other words, I think we
> > should check PCD "PcdPlatformRecoverySupport" instead of Local variable
> "PlatformRecovery"
> > (from OsIndications variable) like the code below. What do you guys
> > think? If you need a meeting or short talk to discuss this, feel free to let me
> know.
> >
> >   if (!BootSuccess) {
> > -   if (PlatformRecovery) {
> > +  if (PcdGetBool (PcdPlatformRecoverySupport)) {
> >       LoadOptions = EfiBootManagerGetLoadOptions (&LoadOptionCount,
> > LoadOptionTypePlatformRecovery);
> >       ProcessLoadOptions (LoadOptions, LoadOptionCount);
> >       EfiBootManagerFreeLoadOptions (LoadOptions, LoadOptionCount);
> >     } else {
> >       //
> >       // When platform recovery is not enabled, still boot to platform
> > default file path.
> >       //
> >       EfiBootManagerProcessLoadOption (&PlatformDefaultBootOption);
> >     }
> >
> >
> > In addition, it looks like EDK2 don't have code to process
> > OsRecovery#### variables. Do we need to create a Bug on TianoCore
> Bugzilla system?

OsRecovery#### doesn't have an implement yet, we should co-work with the OS vendor to define the operation. For now, there is no requirement.

> >
> > Moreover, I saw that both of you had a discussion about "Default
> > PlatformRecovery", but I can't figure out the connection between the
> > discussion and the final code change. Isn't the "Default PlatformRecovery"
> > part of the Platform Recovery feature? At this moment, we don't have
> > OS recovery support, so I think that NO platform recovery support can
> > be identified as NO boot option recovery support. For this case,
> > shouldn't we use PCD "PcdPlatformRecoverySupport" to control "Default
> > PlatformRecovery" as well? For the next step, I think we need to get
> > further clarification from USWG to either not tie "Default Boot
> > Behavior" with PlatformRecovery or update the description to the
> following:
> >     - If system firmware supports Platform recovery as described in
> > Section 3.4.2, system firmware must include a PlatformRecovery####
> > variable specifying a short-form File Path Media Device Path....

As I said above the 'Default Platform Recovery' refer to the short-form file path boot. When the PcdPlatformRecoverySupport is TRUE, it is part of PlatformRecovery. Ohterwise, it is not part of platform recovery and we still need to support it because of the capability issue. And that isn't conflict with the spec.

Thanks,
Zhichao

> >
> > Regards,
> > Sunny Wang
> >