Trace Generation

Methods available in the Shodan repo for generating traces from Spike, and Renode.

Trace Generation with Wrapper Script

The test_runner.py script currently supports both trace generation in both Spike and Renode.

Spike trace generation with `test_runner.py`

cd to the root shodan directory and run:

source build/setup.sh

(optionally) run m springbok to create potential elf targets
Run trace with the test_runner.py script and save trace to file:

TEST_RUNNER="${ROOTDIR}/sw/vec/scripts/test_runner.py"
ELF_FILE="${ROOTDIR}/out/springbok/rvv/softrvv/tests/softrvv_vxor_test.elf"
SPIKE_PATH="${ROOTDIR}/out/host/spike/bin/spike"
OUTPUT_FILE="spike_trace_output.txt"

"${TEST_RUNNER}" "spike" \
    "${ELF_FILE}" \
    "--spike-path" "${SPIKE_PATH}" \
    "--timeout=30" \
    "--trace-output" "${OUTPUT_FILE}"

Renode trace generation with `test_runner.py`

TEST_RUNNER="${ROOTDIR}/sw/vec/scripts/test_runner.py"
ELF_FILE="${ROOTDIR}/out/springbok/rvv/softrvv/tests/softrvv_vxor_test.elf"
RENODE_PATH="${ROOTDIR}/out/host/renode/renode"
OUTPUT_FILE="renode_trace_output.txt"

"${TEST_RUNNER}" "renode" \
    "${ELF_FILE}" \
    "--renode-path" "${RENODE_PATH}" \
    "--quick_test" \
    "--timeout=30" \
    "--trace-output" "${OUTPUT_FILE}"

Direct Trace Generation

This section provides examples for setting up trace generation, and command line options for Spike and Renode.

Direct Spike Trace Generation

Quickly Get Spike Trace With Springbok Defaults

To run a direct spike trace with the defaults for springbok, run:

ELF_FILE="${ROOTDIR}/out/springbok/rvv/softrvv/tests/softrvv_vxor_test.elf"

spike_sim_springbok "${ELF_FILE}"

Trace file will be saved to /tmp/spike_trace.txt

Granular Direct Spike Trace Generation

This section describes flags in depth for more granular trace generation.

The following example shows trace generation for a softrvv vxor test script, and how to adjust each of the flag values:

ELF_FILE="${ROOTDIR}/out/springbok/rvv/softrvv/tests/softrvv_vxor_test.elf"
LOG_FILE="spike_trace_output.txt"

"${OUT}/host/spike/bin/spike" \
    -m0x32000000:0x100000,0x34000000:0x1000000 \
    --pc=0x32000000 \
    --log="${LOG_FILE}" \
    -l \
    "${ELF_FILE}"

Sample output:

core   0: 0x00001000 (0x00000297) auipc   t0, 0x0
core   0: 0x00001004 (0x02028593) addi    a1, t0, 32
core   0: 0x00001008 (0xf1402573) csrr    a0, mhartid
core   0: 0x0000100c (0x0182a283) lw      t0, 24(t0)
core   0: 0x00001010 (0x00028067) jr      t0
core   0: >>>>  _boot_address
core   0: 0x32000000 (0x0080006f) j       pc + 0x8
core   0: >>>>  _start
core   0: 0x32000008 (0x02400117) auipc   sp, 0x2400
core   0: 0x3200000c (0xfb810113) addi    sp, sp, -72

Notes on Spike Flags:

The elf-file must be the last argument
Omitting the --log flag but keeping the -l flag sends trace to console's stderr instead of to a file.
add -d to to interactively run the trace
add --log-commits to add register value changes woven between the trace:

Sample output (recall that t0 is register x5, and a1 is x11):

core   0: 0x00001000 (0x00000297) auipc   t0, 0x0
core   0: 3 0x00001000 (0x00000297) x 5 0x00001000
core   0: 0x00001004 (0x02028593) addi    a1, t0, 32
core   0: 3 0x00001004 (0x02028593) x11 0x00001020

we use the -m flag as -m<IMEM>:<IMEM LENGTH>,<DMEM>:<DMEM LENGTH> or more generally from the spike documentation:

  -m<a:m,b:n,...>       Provide memory regions of size m and n bytes
                        at base addresses a and b (with 4 KiB alignment)

Trace Analysis

Number of Instructions

These commands assume the IMEM starts at 0x32000000, may need to s/0x32/0x80/ for 0x8000000 IMEM Starting address.

Spike

cat spike_trace_output.txt | awk '/_start/,/_finish/' | grep "0: 0x32" | wc -l

Renode

renode_trace_output.txt | wc -l

PC for Trace Analysis

SystemC trace output contains the 8 digit PC in hex (without the ‘0x’ prefix).

These commands reformat the raw output trace from Spike and Renode for comparison with the SystemC output trace, located at:

$ROOTDIR/out/springbok/systemc/sim/simulations/trace.dat

Spike

cat spike_trace.txt | awk '/_start/,/_finish/' | sed 's/.*: 0x\([0-9a-fA-F]\{8\}\).*/\1/g' | egrep "^32" > spike_trace_pc.txt

Renode

cat renode_trace.txt | awk -F":" '{print $1}' | tr "[:upper:]" "[:lower:]" > renode_trace_pc.txt

Trace Generation