fix(fvp): enable cache state modelling
Cache state modelling is currently disabled in these models due to
performance cliffs observed in v11.18 and earlier. This caused certains
tests to run very slowly, and others to crash.
The latest version of the models (v11.19) address cache state
modelling performance. The following models are excluded from the
patch as they appear to still have the performance issues observed
previously:
* fvp-cortexa710x4
* fvp-cortexx2
* fvp-neoverse_n2x4
* fvp-neoversev1x4
Change-Id: Ied2d6cc42a02fab5d9765ccb9ecb2e449bd20a51
Signed-off-by: Harrison Mutai <harrison.mutai@arm.com>
diff --git a/model/fvp_common.sh b/model/fvp_common.sh
index 19a06a3..f31111c 100644
--- a/model/fvp_common.sh
+++ b/model/fvp_common.sh
@@ -47,6 +47,7 @@
${secure_memory+-C bp.secure_memory=$secure_memory}
${cache_state_modelled+-C cache_state_modelled=$cache_state_modelled}
+${use_pchannel_for_threads+-C pctl.use_pchannel_for_threads=$use_pchannel_for_threads}
${secure_ram_fill+-C bp.secureSRAM.fill1=0x00000000}
${secure_ram_fill+-C bp.secureSRAM.fill2=0x00000000}