[PATCH v3 07/10] Drivers: hv: vmbus: Bring the post_msg_page back for TDX VMs with the paravisor

[Dexuan Cui <decui@microsoft.com>]

The post_msg_page was removed in
commit 9a6b1a170ca8 ("Drivers: hv: vmbus: Remove the per-CPU post_msg_page")

However, it turns out that we need to bring it back, but only for a TDX VM
with the paravisor: in such a VM, the hyperv_pcpu_input_arg is not decrypted,
but the HVCALL_POST_MESSAGE in such a VM needs a decrypted page as the
hypercall input page: see the comments in hyperv_init() for a detailed
explanation.

Except for HVCALL_POST_MESSAGE and HVCALL_SIGNAL_EVENT, the other hypercalls
in a TDX VM with the paravisor still use hv_hypercall_pg and must use the
hyperv_pcpu_input_arg (which is encrypted in such a VM), when a hypercall
input page is used.

Signed-off-by: Dexuan Cui <decui@microsoft.com>
---

Changes in v2: None

Changes in v3:
  hyperv_paravisor_present -> ms_hyperv.paravisor_present

 arch/x86/hyperv/hv_init.c | 20 +++++++++++--
 drivers/hv/hv.c           | 59 +++++++++++++++++++++++++++++++++++----
 drivers/hv/hyperv_vmbus.h | 11 ++++++++
 3 files changed, 82 insertions(+), 8 deletions(-)

diff --git a/arch/x86/hyperv/hv_init.c b/arch/x86/hyperv/hv_init.c
index eca5c4b7e3b5..3729eee21e47 100644
--- a/arch/x86/hyperv/hv_init.c
+++ b/arch/x86/hyperv/hv_init.c
@@ -480,6 +480,22 @@ void __init hyperv_init(void)
 	 * Setup the hypercall page and enable hypercalls.
 	 * 1. Register the guest ID
 	 * 2. Enable the hypercall and register the hypercall page
+	 *
+	 * A TDX VM with no paravisor only uses TDX GHCI rather than hv_hypercall_pg:
+	 * when the hypercall input is a page, such a VM must pass a decrypted
+	 * page to Hyper-V, e.g. hv_post_message() uses the per-CPU page
+	 * hyperv_pcpu_input_arg, which is decrypted if no paravisor is present.
+	 *
+	 * A TDX VM with the paravisor uses hv_hypercall_pg for most hypercalls,
+	 * which are handled by the paravisor and the VM must use an encrypted
+	 * input page: in such a VM, the hyperv_pcpu_input_arg is encrypted and
+	 * used in the hypercalls, e.g. see hv_mark_gpa_visibility() and
+	 * hv_arch_irq_unmask(). Such a VM uses TDX GHCI for two hypercalls:
+	 * 1. HVCALL_SIGNAL_EVENT: see vmbus_set_event() and _hv_do_fast_hypercall8().
+	 * 2. HVCALL_POST_MESSAGE: the input page must be a decrypted page, i.e.
+	 * hv_post_message() in such a VM can't use the encrypted hyperv_pcpu_input_arg;
+	 * instead, hv_post_message() uses the post_msg_page, which is decrypted
+	 * in such a VM and is only used in such a VM.
 	 */
 	guest_id = hv_generate_guest_id(LINUX_VERSION_CODE);
 	wrmsrl(HV_X64_MSR_GUEST_OS_ID, guest_id);
@@ -487,8 +503,8 @@ void __init hyperv_init(void)
 	/* Hyper-V requires to write guest os id via ghcb in SNP IVM. */
 	hv_ghcb_msr_write(HV_X64_MSR_GUEST_OS_ID, guest_id);
 
-	/* A TDX guest uses the GHCI call rather than hv_hypercall_pg. */
-	if (hv_isolation_type_tdx())
+	/* A TDX VM with no paravisor only uses TDX GHCI rather than hv_hypercall_pg */
+	if (hv_isolation_type_tdx() && !ms_hyperv.paravisor_present)
 		goto skip_hypercall_pg_init;
 
 	hv_hypercall_pg = __vmalloc_node_range(PAGE_SIZE, 1, VMALLOC_START,
diff --git a/drivers/hv/hv.c b/drivers/hv/hv.c
index 48b1623112f0..523c5d99f375 100644
--- a/drivers/hv/hv.c
+++ b/drivers/hv/hv.c
@@ -57,20 +57,37 @@ int hv_post_message(union hv_connection_id connection_id,
 
 	local_irq_save(flags);
 
-	aligned_msg = *this_cpu_ptr(hyperv_pcpu_input_arg);
+	/*
+	 * A TDX VM with the paravisor must use the decrypted post_msg_page: see
+	 * the comment in struct hv_per_cpu_context. A SNP VM with the paravisor
+	 * can use the encrypted hyperv_pcpu_input_arg because it copies the
+	 * input into the GHCB page, which has been decrypted by the paravisor.
+	 */
+	if (hv_isolation_type_tdx() && ms_hyperv.paravisor_present)
+		aligned_msg = this_cpu_ptr(hv_context.cpu_context)->post_msg_page;
+	else
+		aligned_msg = *this_cpu_ptr(hyperv_pcpu_input_arg);
+
 	aligned_msg->connectionid = connection_id;
 	aligned_msg->reserved = 0;
 	aligned_msg->message_type = message_type;
 	aligned_msg->payload_size = payload_size;
 	memcpy((void *)aligned_msg->payload, payload, payload_size);
 
-	if (hv_isolation_type_snp())
-		status = hv_ghcb_hypercall(HVCALL_POST_MESSAGE,
-				(void *)aligned_msg, NULL,
-				sizeof(*aligned_msg));
-	else
+	if (ms_hyperv.paravisor_present) {
+		if (hv_isolation_type_tdx())
+			status = hv_tdx_hypercall(HVCALL_POST_MESSAGE,
+						  virt_to_phys(aligned_msg), 0);
+		else if (hv_isolation_type_snp())
+			status = hv_ghcb_hypercall(HVCALL_POST_MESSAGE,
+						   aligned_msg, NULL,
+						   sizeof(*aligned_msg));
+		else
+			status = HV_STATUS_INVALID_PARAMETER;
+	} else {
 		status = hv_do_hypercall(HVCALL_POST_MESSAGE,
 				aligned_msg, NULL);
+	}
 
 	local_irq_restore(flags);
 
@@ -105,6 +122,24 @@ int hv_synic_alloc(void)
 		tasklet_init(&hv_cpu->msg_dpc,
 			     vmbus_on_msg_dpc, (unsigned long) hv_cpu);
 
+		if (ms_hyperv.paravisor_present && hv_isolation_type_tdx()) {
+			hv_cpu->post_msg_page = (void *)get_zeroed_page(GFP_ATOMIC);
+			if (hv_cpu->post_msg_page == NULL) {
+				pr_err("Unable to allocate post msg page\n");
+				goto err;
+			}
+
+			ret = set_memory_decrypted((unsigned long)hv_cpu->post_msg_page, 1);
+			if (ret) {
+				pr_err("Failed to decrypt post msg page: %d\n", ret);
+				/* Just leak the page, as it's unsafe to free the page. */
+				hv_cpu->post_msg_page = NULL;
+				goto err;
+			}
+
+			memset(hv_cpu->post_msg_page, 0, PAGE_SIZE);
+		}
+
 		/*
 		 * Synic message and event pages are allocated by paravisor.
 		 * Skip these pages allocation here.
@@ -178,6 +213,17 @@ void hv_synic_free(void)
 			= per_cpu_ptr(hv_context.cpu_context, cpu);
 
 		/* It's better to leak the page if the encryption fails. */
+		if (ms_hyperv.paravisor_present && hv_isolation_type_tdx()) {
+			if (hv_cpu->post_msg_page) {
+				ret = set_memory_encrypted((unsigned long)
+					hv_cpu->post_msg_page, 1);
+				if (ret) {
+					pr_err("Failed to encrypt post msg page: %d\n", ret);
+					hv_cpu->post_msg_page = NULL;
+				}
+			}
+		}
+
 		if (!ms_hyperv.paravisor_present &&
 		    (hv_isolation_type_en_snp() || hv_isolation_type_tdx())) {
 			if (hv_cpu->synic_message_page) {
@@ -199,6 +245,7 @@ void hv_synic_free(void)
 			}
 		}
 
+		free_page((unsigned long)hv_cpu->post_msg_page);
 		free_page((unsigned long)hv_cpu->synic_event_page);
 		free_page((unsigned long)hv_cpu->synic_message_page);
 	}
diff --git a/drivers/hv/hyperv_vmbus.h b/drivers/hv/hyperv_vmbus.h
index 55f2086841ae..f6b1e710f805 100644
--- a/drivers/hv/hyperv_vmbus.h
+++ b/drivers/hv/hyperv_vmbus.h
@@ -123,6 +123,17 @@ struct hv_per_cpu_context {
 	void *synic_message_page;
 	void *synic_event_page;
 
+	/*
+	 * The page is only used in hv_post_message() for a TDX VM (with the
+	 * paravisor) to post a messages to Hyper-V: when such a VM calls
+	 * HVCALL_POST_MESSAGE, it can't use the hyperv_pcpu_input_arg (which
+	 * is encrypted in such a VM) as the hypercall input page, because
+	 * the input page for HVCALL_POST_MESSAGE must be decrypted in such a
+	 * VM, so post_msg_page (which is decrypted in hv_synic_alloc()) is
+	 * introduced for this purpose. See hyperv_init() for more comments.
+	 */
+	void *post_msg_page;
+
 	/*
 	 * Starting with win8, we can take channel interrupts on any CPU;
 	 * we will manage the tasklet that handles events messages on a per CPU
-- 
2.25.1