[PATCHv10 10/18] x86/tdx: Convert shared memory back to private on kexec

[Borislav Petkov]

On Mon, May 06, 2024 at 06:37:19PM +0300, Kirill A. Shutemov wrote:
> "second kernel" is nomenclature kexec folks are using, but okay.

And the "third kernel" is the one which got kexec-ed the second time?

You can make it: "The second, kexec-ed kernel" and then it is perfectly
clear.

> > > +	/*
> > > +	 * Crash kernel reaches here with interrupts disabled: can't wait for
> > > +	 * conversions to finish.
> > > +	 *
> > > +	 * If race happened, just report and proceed.
> > > +	 */
> > > +	bool wait_for_lock = !crash;
> > 
> > You don't need that bool - use crash.
> 
> Dave suggested the variable for documentation purposes.
> 
> https://lore.kernel.org/all/0b70ee1e-4bb5-4867-9378-f5723ca091d5@intel.com
> 
> I'm fine either way.

But you have the comment above it which already explains what's going
on...

> > Why are we printing something here if we're not really acting up on it?
> > 
> > Who should care here?
> 
> The only thing we can do at this point on failure is panic. It think
> it is reasonable to proceed, especially for crash case.
> 
> The print leaves a trace in the log to give a clue for debug.

Sure but you'll leave a trace if you panic right then and there, on the
first failure. Why noodle through the pages if the first failure is
already fatal?

> One possible reason for the failure is if kdump raced with memory
> conversion. In this case shared bit in page table got set (or not cleared
> form shared->private conversion), but the page is actually private. So
> this failure is not going to affect the kexec'ed kernel.

Lemme make sure I understand what you're saying here:

1. This is a fatal failure and we should panic

However,

2. the kexec-ed kernel is using a different page table so there won't be
   a mismatch between shared/private marking of the page so it doesn't
   matter

Close?

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette