Re: [PATCH v19 070/130] KVM: TDX: TDP MMU TDX support

[Chen Yu]

On 2024-03-28 at 11:12:57 +0800, Chao Gao wrote:
> >+#if IS_ENABLED(CONFIG_HYPERV)
> >+static int vt_flush_remote_tlbs(struct kvm *kvm);
> >+#endif
> >+
> > static __init int vt_hardware_setup(void)
> > {
> > 	int ret;
> >@@ -49,11 +53,29 @@ static __init int vt_hardware_setup(void)
> > 		pr_warn_ratelimited("TDX requires mmio caching.  Please enable mmio caching for TDX.\n");
> > 	}
> > 
> >+#if IS_ENABLED(CONFIG_HYPERV)
> >+	/*
> >+	 * TDX KVM overrides flush_remote_tlbs method and assumes
> >+	 * flush_remote_tlbs_range = NULL that falls back to
> >+	 * flush_remote_tlbs.  Disable TDX if there are conflicts.
> >+	 */
> >+	if (vt_x86_ops.flush_remote_tlbs ||
> >+	    vt_x86_ops.flush_remote_tlbs_range) {
> >+		enable_tdx = false;
> >+		pr_warn_ratelimited("TDX requires baremetal. Not Supported on VMM guest.\n");
> >+	}
> >+#endif
> >+
> > 	enable_tdx = enable_tdx && !tdx_hardware_setup(&vt_x86_ops);
> > 	if (enable_tdx)
> > 		vt_x86_ops.vm_size = max_t(unsigned int, vt_x86_ops.vm_size,
> > 					   sizeof(struct kvm_tdx));
> > 
> >+#if IS_ENABLED(CONFIG_HYPERV)
> >+	if (enable_tdx)
> >+		vt_x86_ops.flush_remote_tlbs = vt_flush_remote_tlbs;
> 
> Is this hook necessary/beneficial to TDX?
>

I think so.

We happended to encounter the following error and breaks the boot up:
"SEAMCALL (0x000000000000000f) failed: 0xc0000b0800000001"
0xc0000b0800000001 indicates the TDX_TLB_TRACKING_NOT_DONE, and it is caused
by page demotion but not yet doing a tlb shotdown by tlb track.


It was found on my system the CONFIG_HYPERV is not set, and it makes
kvm_arch_flush_remote_tlbs() not invoking tdx_track() before the
tdh_mem_page_demote(), which caused the problem.

> if no, we can leave .flush_remote_tlbs as NULL. if yes, we should do:
> 
> struct kvm_x86_ops {
> ...
> #if IS_ENABLED(CONFIG_HYPERV) || IS_ENABLED(TDX...)
> 	int  (*flush_remote_tlbs)(struct kvm *kvm);
> 	int  (*flush_remote_tlbs_range)(struct kvm *kvm, gfn_t gfn,
> 					gfn_t nr_pages);
> #endif

If the flush_remote_tlbs implementation are both available in HYPERV and TDX,
does it make sense to remove the config checks? I thought when commit 0277022a77a5
was introduced, the only user of flush_remote_tlbs() is hyperv, and now
there is TDX.

thanks,
Chenyu