fix(intel): handle cold reset via physical reset switch
On the Agilex5 platform when cold reset is issued via CLI application
in the OS, it is received in the BL31 via a SMC call and handled
accordingly like flush/invalidate the caches. However, when the cold
reset is issued via an external switch these handlings are missed.
This patch addresses those missed cache operations.
Also, this patch is to restoring SCR_EL3 NS bit to its previous value
in order to avoid unintended behavior especially if subsequent code
expects the SCR_EL3 register to be in its original configuration.
Change-Id: I9737f2db649e483ba61fffa6eeb0b56a9d15074a
Signed-off-by: Girisha Dengi <girisha.dengi@intel.com>
Signed-off-by: Sieu Mun Tang <sieu.mun.tang@intel.com>
diff --git a/plat/intel/soc/agilex5/bl31_plat_setup.c b/plat/intel/soc/agilex5/bl31_plat_setup.c
index c090117..ab03928 100644
--- a/plat/intel/soc/agilex5/bl31_plat_setup.c
+++ b/plat/intel/soc/agilex5/bl31_plat_setup.c
@@ -196,6 +196,9 @@
NOTICE("BL31: CPU ID = %x\n", cpuid);
INFO("BL31: Invalidate Data cache\n");
invalidate_dcache_all();
+
+ /* Invalidate for NS EL2 and EL1 */
+ invalidate_cache_low_el();
}
/* Get non-secure image entrypoint for BL33. Zephyr and Linux */