fix(spmd): relax use of EHF with SPMC at S-EL2
Follow up to [1] and [2], for systems implementing the SPMC at S-EL2,
it is necessary to leave the option for handling Group0 interrupts
(while the normal world runs) through the EHF by the use of the
EL3_EXCEPTION_HANDLING option.
Specifically for RAS, the handling through EHF is still required because
the platform function provided by the SPMD doesn't provide the facility
to link back to the RAS handling framework.
[1] https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/16047
[2] https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/19897
Signed-off-by: Olivier Deprez <olivier.deprez@arm.com>
Change-Id: Idf8741887904a286fb3f5ab2d754afd2fc78d3b0
diff --git a/services/std_svc/spmd/spmd_main.c b/services/std_svc/spmd/spmd_main.c
index 80b506b..6b16373 100644
--- a/services/std_svc/spmd/spmd_main.c
+++ b/services/std_svc/spmd/spmd_main.c
@@ -249,6 +249,7 @@
SMC_RET0(&ctx->cpu_ctx);
}
+#if (EL3_EXCEPTION_HANDLING == 0)
/*******************************************************************************
* spmd_group0_interrupt_handler_nwd
* Group0 secure interrupt in the normal world are trapped to EL3. Delegate the
@@ -281,6 +282,7 @@
return 0U;
}
+#endif
/*******************************************************************************
* spmd_handle_group0_intr_swd
@@ -561,6 +563,18 @@
}
/*
+ * Permit configurations where the SPM resides at S-EL1/2 and upon a
+ * Group0 interrupt triggering while the normal world runs, the
+ * interrupt is routed either through the EHF or directly to the SPMD:
+ *
+ * EL3_EXCEPTION_HANDLING=0: the Group0 interrupt is routed to the SPMD
+ * for handling by spmd_group0_interrupt_handler_nwd.
+ *
+ * EL3_EXCEPTION_HANDLING=1: the Group0 interrupt is routed to the EHF.
+ *
+ */
+#if (EL3_EXCEPTION_HANDLING == 0)
+ /*
* Register an interrupt handler routing Group0 interrupts to SPMD
* while the NWd is running.
*/
@@ -570,6 +584,8 @@
if (rc != 0) {
panic();
}
+#endif
+
return 0;
}