refactor: always eret FFA_RUN to the caller
The FFA_RUN interface was not running an SP if it didn't
have any pending interrupts, notifications or messages.
This was inherited from the hypervisor implementation.
With the present change, if the vCPU is in VCPU_STATE_WAITING
always eret FFA_RUN to the target SP.
This behaviour aligns better with what is defined by the
runtime model for FFA_RUN interface in the FF-A specification
(FF-A v1.1 onwards).
BREAKING: this change the expected behaviour of the FFA_RUN
ABI. This breaks Hafnium tests, which are fixed in other
patches.
Signed-off-by: J-Alves <joao.alves@arm.com>
Change-Id: Ib37a430ef357720e7368fb4654b26f610986756e
diff --git a/src/vcpu.c b/src/vcpu.c
index b54fb90..b5604ef 100644
--- a/src/vcpu.c
+++ b/src/vcpu.c
@@ -301,7 +301,8 @@
* However, this action is restricted to WAITING and BLOCKED states,
* as such, assert accordingly.
*/
-void vcpu_set_running(struct vcpu_locked target_locked, struct ffa_value *args)
+void vcpu_set_running(struct vcpu_locked target_locked,
+ const struct ffa_value *args)
{
struct vcpu *target_vcpu = target_locked.vcpu;