Re: [PATCH v3 1/2] KVM: arm64: Support FFA_MSG_SEND_DIRECT_REQ in host handler

On Thu, Jan 08, 2026 at 03:26:21PM +0000, Will Deacon wrote:

Hi Will,

On Wed, Nov 19, 2025 at 02:07:53AM +0000, Per Larsen via B4 Relay wrote:

quoted

From: Sebastian Ene <redacted>

Allow direct messages to be forwarded from the host. The host should
not be sending framework messages so they are filtered out.

Signed-off-by: Sebastian Ene <redacted>
Reviewed-by: Yeoreum Yun <redacted>
Signed-off-by: Per Larsen <redacted>
---
 arch/arm64/kvm/hyp/nvhe/ffa.c | 22 ++++++++++++++++++++++
 include/linux/arm_ffa.h       |  3 +++
 2 files changed, 25 insertions(+)

diff --git a/arch/arm64/kvm/hyp/nvhe/ffa.c b/arch/arm64/kvm/hyp/nvhe/ffa.c
index 58b7d0c477d7fce235fc70d089d175c7879861b5..a38a3ab497e5eac11777109684a33f02d88d09a1 100644
--- a/arch/arm64/kvm/hyp/nvhe/ffa.c
+++ b/arch/arm64/kvm/hyp/nvhe/ffa.c

@@ -862,6 +862,23 @@ static void do_ffa_part_get(struct arm_smccc_1_2_regs *res,
 	hyp_spin_unlock(&host_buffers.lock);
 }
 
+static void do_ffa_direct_msg(struct arm_smccc_1_2_regs *res,
+			      struct kvm_cpu_context *ctxt,
+			      u64 vm_handle)
+{
+	DECLARE_REG(u32, flags, ctxt, 2);
+
+	struct arm_smccc_1_2_regs *args = (void *)&ctxt->regs.regs[0];
+
+	/* filter out framework messages */
+	if (FIELD_GET(FFA_MSG_FLAGS_MSG_TYPE, flags)) {

Wouldn't we be better off just checking that flags is 0? The rest of it
is SBZ or MBZ in the current spec.

Yes, we can simplify it in this way.

quoted

+		ffa_to_smccc_error(res, FFA_RET_INVALID_PARAMETERS);
+		return;
+	}
+
+	arm_smccc_1_2_smc(args, res);
+}
+
 bool kvm_host_ffa_handler(struct kvm_cpu_context *host_ctxt, u32 func_id)
 {
 	struct arm_smccc_1_2_regs res;

@@ -920,6 +937,11 @@ bool kvm_host_ffa_handler(struct kvm_cpu_context *host_ctxt, u32 func_id)
 	case FFA_PARTITION_INFO_GET:
 		do_ffa_part_get(&res, host_ctxt);
 		goto out_handled;
+	case FFA_MSG_SEND_DIRECT_REQ:
+	case FFA_FN64_MSG_SEND_DIRECT_REQ:
+

Weird whitespace addition ^^

Let me clear this space out.

quoted

+		do_ffa_direct_msg(&res, host_ctxt, HOST_FFA_ID);

What's the point of passing HOST_FFA_ID here? Is that supposed to end up
in the Sender ID bits of W1?

I can remove it, this doesn't bring too much for upstream but on the
android kernel with guest-ffa it makes sense because we need to validate
the sender to prevent impersonation.

Will

Thanks,
Sebastian