apu_valid_i not memorized causing deadlock condition when both alu and apu write into same register at the same time

Created by: dd-baoshan

Many tests are failing with fpu tests especially in pulp_fpu_zfinx_2cyclat.

The test hang issue was due to same regfile write by both alu and apu. In this case alu has priority than apu (see M1 marker below).
The apu_valid is suppressed during writeback from fsub (at marker M1) that due to deadlock and this keep apu_busy_i stuck at high thus hang the test.

During daily discussion, @dd-vaibhavjain stated that the issue is due to recent update at below rtl.
I tried by update to insert condition with data_rvalid_i (line 374) and rerun, the hang test is passing now.

Component

Component:RTL

Steps to Reproduce

cv32e40p git hash: 7df7d2df
core-v-verif git hash: 02ee75284e34acf908dda5344f642a7cb3b0ac0c
Users need to checkout below core-v-verif to replicate the issue. The reproduce steps as followings.

git clone --branch cv32e40p/dev_dd https://github.com/XavierAubert/core-v-verif.git sandbox

cd sandbox/cv32e40p/sim/uvmt

make gen_corev-dv test TEST=corev_rand_fp_instr_data_fwd_test CFG_PLUSARGS=+UVM_TIMEOUT=1000000 CHECK_SIM_RESULT=YES COMP=1 CV_CORE=cv32e40p COREV=1 CFG=pulp_fpu_zfinx_2cyclat TEST_CFG_FILE=floating_pt_instr_en,floating_pt_zfinx_instr_en SIMULATOR=vsim USE_ISS=yes COV=NO RUN_INDEX=764058775 GEN_START_INDEX=764058775 SEED=764058775 WAVES=1