Re: [PATCHv6 00/16] x86/tdx: Add kexec support

[Nikolay Borisov]

On 31.01.24 г. 14:47 ч., Baoquan He wrote:
> On 01/31/24 at 09:31am, Nikolay Borisov wrote:
> > On 30.01.24 г. 15:43 ч., Paolo Bonzini wrote:
> > > On 1/24/24 13:55, Kirill A. Shutemov wrote:
> > > > The patchset adds bits and pieces to get kexec (and crashkernel) work on
> > > > TDX guest.
> > > >
> > > > The last patch implements CPU offlining according to the approved ACPI
> > > > spec change poposal[1]. It unlocks kexec with all CPUs visible in
> > > > the target
> > > > kernel. It requires BIOS-side enabling. If it missing we fallback to
> > > > booting
> > > > 2nd kernel with single CPU.
> > > >
> > > > Please review. I would be glad for any feedback.
> > >
> > > Hi Kirill,
> > >
> > > I have a very basic question: is there a reason why this series does not
> > > revert commit cb8eb06d50fc, "x86/virt/tdx: Disable TDX host support when
> > > kexec is enabled"?
> >
> > While on the topic, Paolo do you think it's  better to have a runtime
> > disable of kexec rather than at compile time:
> >
> > [RFC PATCH] x86/virt/tdx: Disable KEXEC in the presence of TDX
> >
> > 20240118160118.1899299-1-nik.borisov@xxxxxxxx
>
> Runtime disabling kexec looks better than at cmpile time, esp for
> distros. While from above patch, making using of kexec_load_disabled to
> achive the runtime disabling may not be so good. Because we have a front
> door to enable it through:
>
> /proc/sys/kernel/kexec_load_disabled

AFAIU it can't be enabled via this sysctl because the handler for it 
expects only 1 to be written to it:

     2                 .proc_handler   = proc_dointvec_minmax, 

     1                 .extra1         = SYSCTL_ONE, 

  994                  .extra2         = SYSCTL_ONE,

> If there's a flag or status to check if TDX host is enabled, and does
> the checking in kexec_load_permitted(), that could be better. Anyway, I
> saw Huang, Kai has posted the tdx host support patchset.
>
> > I'm trying to get traction for this patch.
> >
> >
> > > Thanks,
> > >
> > > Paolo