change checkpoint generation to integrate GDB scripting more cleanly and save UART and PLIC state

linux/testvector-generation/createGenCheckpointScript.py

@@ -1,63 +0,0 @@
-#! /usr/bin/python3
-import sys
-
-if len(sys.argv) != 8:
-    sys.exit("""Error createGenCheckpointScript.py expects 7 args:
-    <TCP port number>
-    <path to vmlinux>
-    <checkpoint instruction count>
-    <path to GDB checkpoint state dump>
-    <path to GDB ram dump>
-    <checkpoint pc address>
-    <number of times pc has already been hit before checkpoint>""")
-
-tcpPort=sys.argv[1]
-vmlinux=sys.argv[2]
-instrCount=sys.argv[3]
-statePath=sys.argv[4]
-ramPath=sys.argv[5]
-checkPC=sys.argv[6]
-checkPCoccurences=sys.argv[7]
-
-GDBscript = f"""
-# GDB config
-set pagination off
-set logging overwrite on
-set logging redirect on
-set confirm off
-
-# Connect to QEMU session
-target extended-remote :{tcpPort}
-
-# QEMU Config
-maintenance packet Qqemu.PhyMemMode:1
-
-# Symbol file
-file {vmlinux}
-
-# Step over reset vector into actual code
-stepi 100
-# Proceed to checkpoint 
-print "GDB proceeding to checkpoint at {instrCount} instrs, pc {checkPC}\\n"
-b *0x{checkPC}
-ignore 1 {checkPCoccurences}
-c
-print "Reached checkpoint at {instrCount} instrs\\n"
-
-# Log all registers to a file
-printf "GDB storing state to {statePath}\\n"
-set logging file {statePath}
-set logging on
-info all-registers
-set logging off
-
-# Log main memory to a file
-print "GDB storing RAM to {ramPath}\\n"
-dump binary memory {ramPath} 0x80000000 0xffffffff
-
-kill
-q
-"""
-GDBscriptFile = open("genCheckpoint.gdb",'w')
-GDBscriptFile.write(GDBscript)
-GDBscriptFile.close()

linux/testvector-generation/genCheckpoint.sh

@@ -19,8 +19,11 @@ fi
 
 checkPtDir="$tvDir/checkpoint$instrs"
 outTraceFile="$checkPtDir/all.txt"
-interruptsFile="$checkPtDir/interrupts.txt"
 rawStateFile="$checkPtDir/stateGDB.txt"
+rawUartStateFile="$checkPtDir/uartStateGDB.txt"
+uartStateFile="$checkPtDir/checkpoint-UART"
+rawPlicStateFile="$checkPtDir/plicStateGDB.txt"
+plicStateFile="$checkPtDir/checkpoint-PLIC"
 rawRamFile="$checkPtDir/ramGDB.bin"
 ramFile="$checkPtDir/ram.bin"
 
@@ -55,7 +58,48 @@ then
     echo "It occurs ${occurences} times before the ${instrs}th instr." 
 
     # Create GDB script because GDB is terrible at handling arguments / variables
-    ./createGenCheckpointScript.py $tcpPort $imageDir/vmlinux $instrs $rawStateFile $rawRamFile $pc $occurences
+    cat > genCheckpoint.gdb <<- end_of_script 
+    set pagination off
+    set logging overwrite on
+    set logging redirect on
+    set confirm off
+    target extended-remote :$tcpPort
+    maintenance packet Qqemu.PhyMemMode:1
+    file $imageDir/vmlinux
+    # Step over reset vector into actual code
+    stepi 100
+    shell echo \"GDB proceeding to checkpoint at $instrs instrs, pc $pc\"
+    b *0x$pc
+    ignore 1 $occurences
+    c
+    shell echo \"Reached checkpoint at $instrs instrs\"
+    shell echo \"GDB storing CPU state to $rawStateFile\"
+    set logging file $rawStateFile
+    set logging on
+    info all-registers
+    set logging off
+    shell echo \"GDB storing UART state to $rawUartStateFile\"
+    set logging file $rawUartStateFile
+    set logging on
+    x/8xb 0x10000000
+    set logging off
+    shell echo \"GDB storing PLIC state to $rawPlicStateFile\"
+    shell echo \"Note: this dumping assumes a maximum of 63 PLIC sources\"
+    set logging file $rawPlicStateFile
+    set logging on
+    # Priority Levels for sources 1 thru 63
+    x/63xw 0x0C000004
+    # Interrupt Enables
+    x/2xw 0x0C020000
+    # Global Priority Threshold
+    x/1xw 0x0C200000
+    set logging off
+    shell echo \"GDB storing RAM to $rawRamFile\"
+    dump binary memory $rawRamFile 0x80000000 0xffffffff
+    kill
+    q
+end_of_script
+
     # GDB+QEMU
     echo "Starting QEMU in replay mode with attached GDB script at $(date +%H:%M:%S)"
     (qemu-system-riscv64 \
@@ -64,12 +108,15 @@ then
     -bios $imageDir/fw_jump.elf -kernel $imageDir/Image -append "root=/dev/vda ro" -initrd $imageDir/rootfs.cpio \
     -singlestep -rtc clock=vm -icount shift=0,align=off,sleep=on,rr=replay,rrfile=$recordFile \
     -gdb tcp::$tcpPort -S \
-    2>&1 1>./qemu-serial | ./parseQEMUtoGDB.py | ./parseGDBtoTrace.py $interruptsFile | ./remove_dup.awk > $outTraceFile) \
+     1>./qemu-serial) \
-    & riscv64-unknown-elf-gdb --quiet -ex "source genCheckpoint.gdb"
+    & riscv64-unknown-elf-gdb --quiet -x genCheckpoint.gdb
+
     echo "Completed GDB script at $(date +%H:%M:%S)"
 
     # Post-Process GDB outputs
     ./parseState.py "$checkPtDir"
+    ./parseUartState.py "$rawUartStateFile" "$uartStateFile"
+    ./parsePlicState.py "$rawPlicStateFile" "$plicStateFile"
     echo "Changing Endianness at $(date +%H:%M:%S)"
     make fixBinMem
     ./fixBinMem "$rawRamFile" "$ramFile"

linux/testvector-generation/parsePlicState.py

@@ -0,0 +1,87 @@
+#! /usr/bin/python3
+import sys, os
+
+################
+# Helper Funcs #
+################
+
+def tokenize(string):
+    tokens = []
+    token = ''
+    whitespace = 0
+    prevWhitespace = 0
+    for char in string:
+        prevWhitespace = whitespace
+        whitespace = char in ' \t\n'
+        if (whitespace):
+            if ((not prevWhitespace) and (token != '')): 
+                tokens.append(token)
+            token = ''
+        else:
+            token = token + char
+    return tokens
+
+#############
+# Main Code #
+#############
+print("Begin parsing PLIC state.")
+
+# Parse Args
+if len(sys.argv) != 3:
+    sys.exit('Error parsePlicState.py expects 2 args: <raw GDB state dump> <output state file>')
+rawPlicStateFile=sys.argv[1]
+outPlicStateFile=sys.argv[2]
+if not os.path.exists(rawPlicStateFile):
+    sys.exit('Error input file '+rawPlicStateFile+'not found')
+
+# Main Loop
+with open(rawPlicStateFile, 'r') as rawPlicStateFile:
+    plicIntPriorityArray=[]
+    # 0x0C000004 thru 0x0C000010
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000014 thru 0x0C000020
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000024 thru 0x0C000030
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000034 thru 0x0C000040
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000044 thru 0x0C000050
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000054 thru 0x0C000060
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000064 thru 0x0C000070
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000074 thru 0x0C000080
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000084 thru 0x0C000090
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C000094 thru 0x0C0000a0
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C0000a4 thru 0x0C0000b0
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C0000b4 thru 0x0C0000c0
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C0000c4 thru 0x0C0000d0
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C0000d4 thru 0x0C0000e0
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C0000e4 thru 0x0C0000f0
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+    # 0x0C0000f4 thru 0x0C0000fc
+    plicIntPriorityArray += tokenize(rawPlicStateFile.readline())[1:]
+
+    # 0x0C020000 thru 0x0C020004
+    plicIntEnable = tokenize(rawPlicStateFile.readline())[1:]
+
+    # 0x0C200000
+    plicIntPriorityThreshold = tokenize(rawPlicStateFile.readline())[1:]
+
+with open(outPlicStateFile, 'w') as outPlicStateFile:
+    for word in plicIntPriorityArray:
+        outPlicStateFile.write(word[2:]+'\n')
+    for word in plicIntEnable:
+        outPlicStateFile.write(word[2:]+'\n')
+    for word in plicIntPriorityThreshold:
+        outPlicStateFile.write(word[2:]+'\n')
+
+print("Finished parsing PLIC state!")

linux/testvector-generation/parseState.py

@@ -24,7 +24,7 @@ def tokenize(string):
 #############
 # Main Code #
 #############
-print("Begin parsing state.")
+print("Begin parsing CPU state.")
 
 # Parse Args
 if len(sys.argv) != 2:
@@ -96,4 +96,4 @@ with open(stateGDBpath, 'r') as stateGDB:
                     outFile.write(hex(byte)[2:]+'\n')
                 outFile.close()
 
-print("Finished parsing state!")
+print("Finished parsing CPU state!")

linux/testvector-generation/parseUartState.py

@@ -0,0 +1,53 @@
+#! /usr/bin/python3
+import sys, os
+
+################
+# Helper Funcs #
+################
+
+def tokenize(string):
+    tokens = []
+    token = ''
+    whitespace = 0
+    prevWhitespace = 0
+    for char in string:
+        prevWhitespace = whitespace
+        whitespace = char in ' \t\n'
+        if (whitespace):
+            if ((not prevWhitespace) and (token != '')): 
+                tokens.append(token)
+            token = ''
+        else:
+            token = token + char
+    return tokens
+
+#############
+# Main Code #
+#############
+print("Begin parsing UART state.")
+
+# Parse Args
+if len(sys.argv) != 3:
+    sys.exit('Error parseUartState.py expects 2 args: <raw GDB state dump> <output state file>')
+rawUartStateFile=sys.argv[1]
+outUartStateFile=sys.argv[2]
+if not os.path.exists(rawUartStateFile):
+    sys.exit('Error input file '+rawUartStateFile+'not found')
+
+# Main Loop
+with open(rawUartStateFile, 'r') as rawUartStateFile:
+    with open(outUartStateFile, 'w') as outUartStateFile:
+        uartBytes = tokenize(rawUartStateFile.readline())[1:]
+        # Stores
+        # 0: RBR / Divisor Latch Low
+        # 1: IER / Divisor Latch High
+        # 2: IIR
+        # 3: LCR
+        # 4: MCR
+        # 5: LSR
+        # 6: MSR
+        # 7: SCR
+        for uartByte in uartBytes:
+            outUartStateFile.write(uartByte[2:]+'\n')
+
+print("Finished parsing UART state!")