From 54c5f7f607484ca7692dd217eeb22bf97f3a8e93 Mon Sep 17 00:00:00 2001
From: bbracker <bbracker@hmc.edu>
Date: Thu, 7 Apr 2022 19:43:22 -0700
Subject: [PATCH] deprecate remove_dup.awk in favor of expanding
 parseGDBtoTrace.py to internally remove duplicates; this way the instruction
 counts in traps.txt are hopefully now in sync with the line numbers of
 all.txt

---
 linux/testvector-generation/genTrace.sh       |  2 +-
 .../testvector-generation/parseGDBtoTrace.py  | 44 +++++++++++--------
 linux/testvector-generation/remove_dup.awk    | 20 ---------
 3 files changed, 26 insertions(+), 40 deletions(-)
 delete mode 100755 linux/testvector-generation/remove_dup.awk

diff --git a/linux/testvector-generation/genTrace.sh b/linux/testvector-generation/genTrace.sh
index df540c53..76a6a10c 100755
--- a/linux/testvector-generation/genTrace.sh
+++ b/linux/testvector-generation/genTrace.sh
@@ -41,7 +41,7 @@ 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 \
     -d nochain,cpu,in_asm,int \
-    2>&1 >./qemu-serial | ./parseQEMUtoGDB.py | ./parseGDBtoTrace.py $trapsFile | ./remove_dup.awk > $traceFile)
+    2>&1 >./qemu-serial | ./parseQEMUtoGDB.py | ./parseGDBtoTrace.py $trapsFile > $traceFile)
 
     ./filterTrapsToInterrupts.py $tvDir
 
diff --git a/linux/testvector-generation/parseGDBtoTrace.py b/linux/testvector-generation/parseGDBtoTrace.py
index 97145a35..9a467e3e 100755
--- a/linux/testvector-generation/parseGDBtoTrace.py
+++ b/linux/testvector-generation/parseGDBtoTrace.py
@@ -81,7 +81,7 @@ def WhatAddrSC(text, Regs):
     Src = Src.strip(')').strip()
     return Regs[Src]
 
-def PrintInstr(instr, fp):
+def PrintInstr(instr):
     if instr[2] == None:
         return
     ChangedRegisters = instr[4]
@@ -107,28 +107,27 @@ def PrintInstr(instr, fp):
     #print(instr)
 
     if (HUMAN_READABLE == True):
-        fp.write('{:016x} {:08x} {:25s}'.format(instr[0], instr[1], instr[2]))
+        outString='{:016x} {:08x} {:25s}'.format(instr[0], instr[1], instr[2])
         if(len(GPR) != 0):
-            fp.write(' GPR {}'.format(GPR))
+            outString+=' GPR {}'.format(GPR)
         if(instr[3] == 'load' or instr[3] == 'lr'):
-            fp.write(' MemR {:016x} {:016x} {:016x}'.format(instr[5], 0, instr[7]))
+            outString+=' MemR {:016x} {:016x} {:016x}'.format(instr[5], 0, instr[7])
         if(instr[3] == 'store'):
-            fp.write('\t\t\t    MemW {:016x} {:016x} {:016x}'.format(instr[5], instr[6], 0))
-
+            outString+='\t\t\t    MemW {:016x} {:016x} {:016x}'.format(instr[5], instr[6], 0)
         if(len(CSR) != 0):
-            fp.write(' CSR {}'.format(CSRStr))
+            outString+=' CSR {}'.format(CSRStr)
     else:
-        fp.write('{:x} {:x} {:s}'.format(instr[0], instr[1], instr[2].replace(' ', '_')))
+        outString='{:x} {:x} {:s}'.format(instr[0], instr[1], instr[2].replace(' ', '_'))
         if(len(GPR) != 0):
-            fp.write(' GPR {}'.format(GPR))
+            outString+=' GPR {}'.format(GPR)
         if(instr[3] == 'load' or instr[3] == 'lr'):
-            fp.write(' MemR {:x} {:x} {:x}'.format(instr[5], 0, instr[7]))
+            outString+=' MemR {:x} {:x} {:x}'.format(instr[5], 0, instr[7])
         if(instr[3] == 'store'):
-            fp.write(' MemW {:x} {:x} {:x}'.format(instr[5], instr[6], 0))
-
+            outString+=' MemW {:x} {:x} {:x}'.format(instr[5], instr[6], 0)
         if(len(CSR) != 0):
-            fp.write(' CSR {}'.format(CSRStr))
-    fp.write('\n')
+            outString+=' CSR {}'.format(CSRStr)
+    outString+='\n'
+    return outString
 
 # =========
 # Main Code
@@ -154,6 +153,8 @@ lines = []
 interrupts=open(interruptFname,'w')
 interrupts.close()
 
+prevInstrOutString=''
+currInstrOutString=''
 for line in fileinput.input('-'):
     if line.startswith('riscv_cpu_do_interrupt'):
         with open(interruptFname,'a') as interrupts:
@@ -219,11 +220,16 @@ for line in fileinput.input('-'):
 
         lines.clear()
         #instructions.append(MoveInstrToRegWriteLst)
-        PrintInstr(MoveInstrToRegWriteLst, sys.stdout)
-        numInstrs +=1
-        if (numInstrs % 1e5 == 0):
-            sys.stderr.write('GDB trace parser reached '+str(numInstrs/1.0e6)+' million instrs.\n')
-            sys.stderr.flush()
+
+        prevInstrOutString = currInstrOutString
+        currInstrOutString = PrintInstr(MoveInstrToRegWriteLst)
+        # Remove duplicates
+        if (PreviousInstr[0] != CurrentInstr[0]) and (currInstrOutString != None):
+            sys.stdout.write(currInstrOutString)
+            numInstrs += 1
+            if (numInstrs % 1e5 == 0):
+                sys.stderr.write('GDB trace parser reached '+str(numInstrs/1.0e6)+' million instrs.\n')
+                sys.stderr.flush()
     lineNum += 1
 
 
diff --git a/linux/testvector-generation/remove_dup.awk b/linux/testvector-generation/remove_dup.awk
deleted file mode 100755
index 7963d76a..00000000
--- a/linux/testvector-generation/remove_dup.awk
+++ /dev/null
@@ -1,20 +0,0 @@
-#!/usr/bin/awk -f
-
-BEGIN{
-    old = "first"
-}
-
-{
-    if($1 != old){
-	if(old != "first"){
-	    print oldAll
-	}
-    }
-    old=$1
-    oldAll=$0
-}
-
-END{
-    print oldAll
-}
-