mirror of https://gitlab.com/camelot/kickc.git
714 lines
25 KiB
Plaintext
714 lines
25 KiB
Plaintext
Inlined call vicSelectGfxBank::$0 = call toDd00(vicSelectGfxBank::gfx)
|
|
Eliminating unused variable with no statement main::$0
|
|
Eliminating unused variable with no statement main::$1
|
|
|
|
CONTROL FLOW GRAPH SSA
|
|
|
|
unsigned long mul16u(unsigned int a , unsigned int b)
|
|
mul16u: scope:[mul16u] from main
|
|
mul16u::a#5 = phi( main/mul16u::a#1 )
|
|
mul16u::b#1 = phi( main/mul16u::b#0 )
|
|
mul16u::res#0 = 0
|
|
mul16u::mb#0 = (unsigned long)mul16u::b#1
|
|
to:mul16u::@1
|
|
mul16u::@1: scope:[mul16u] from mul16u mul16u::@4
|
|
mul16u::mb#5 = phi( mul16u/mul16u::mb#0, mul16u::@4/mul16u::mb#1 )
|
|
mul16u::res#4 = phi( mul16u/mul16u::res#0, mul16u::@4/mul16u::res#6 )
|
|
mul16u::a#2 = phi( mul16u/mul16u::a#5, mul16u::@4/mul16u::a#0 )
|
|
mul16u::$0 = mul16u::a#2 != 0
|
|
if(mul16u::$0) goto mul16u::@2
|
|
to:mul16u::@3
|
|
mul16u::@2: scope:[mul16u] from mul16u::@1
|
|
mul16u::res#5 = phi( mul16u::@1/mul16u::res#4 )
|
|
mul16u::mb#4 = phi( mul16u::@1/mul16u::mb#5 )
|
|
mul16u::a#3 = phi( mul16u::@1/mul16u::a#2 )
|
|
mul16u::$1 = mul16u::a#3 & 1
|
|
mul16u::$2 = mul16u::$1 != 0
|
|
mul16u::$3 = ! mul16u::$2
|
|
if(mul16u::$3) goto mul16u::@4
|
|
to:mul16u::@5
|
|
mul16u::@3: scope:[mul16u] from mul16u::@1
|
|
mul16u::res#2 = phi( mul16u::@1/mul16u::res#4 )
|
|
mul16u::return#0 = mul16u::res#2
|
|
to:mul16u::@return
|
|
mul16u::@4: scope:[mul16u] from mul16u::@2 mul16u::@5
|
|
mul16u::res#6 = phi( mul16u::@2/mul16u::res#5, mul16u::@5/mul16u::res#1 )
|
|
mul16u::mb#2 = phi( mul16u::@2/mul16u::mb#4, mul16u::@5/mul16u::mb#3 )
|
|
mul16u::a#4 = phi( mul16u::@2/mul16u::a#3, mul16u::@5/mul16u::a#6 )
|
|
mul16u::$5 = mul16u::a#4 >> 1
|
|
mul16u::a#0 = mul16u::$5
|
|
mul16u::$6 = mul16u::mb#2 << 1
|
|
mul16u::mb#1 = mul16u::$6
|
|
to:mul16u::@1
|
|
mul16u::@5: scope:[mul16u] from mul16u::@2
|
|
mul16u::a#6 = phi( mul16u::@2/mul16u::a#3 )
|
|
mul16u::mb#3 = phi( mul16u::@2/mul16u::mb#4 )
|
|
mul16u::res#3 = phi( mul16u::@2/mul16u::res#5 )
|
|
mul16u::$4 = mul16u::res#3 + mul16u::mb#3
|
|
mul16u::res#1 = mul16u::$4
|
|
to:mul16u::@4
|
|
mul16u::@return: scope:[mul16u] from mul16u::@3
|
|
mul16u::return#3 = phi( mul16u::@3/mul16u::return#0 )
|
|
mul16u::return#1 = mul16u::return#3
|
|
return
|
|
to:@return
|
|
|
|
void main()
|
|
main: scope:[main] from __start
|
|
mul16u::a#1 = 4
|
|
mul16u::b#0 = $7b
|
|
call mul16u
|
|
mul16u::return#2 = mul16u::return#1
|
|
to:main::@1
|
|
main::@1: scope:[main] from main
|
|
mul16u::return#4 = phi( main/mul16u::return#2 )
|
|
main::result#0 = mul16u::return#4
|
|
main::kaputt#0 = word0 main::result#0
|
|
main::$2 = byte0 main::kaputt#0
|
|
*BORDER_COLOR = main::$2
|
|
main::$3 = byte1 main::kaputt#0
|
|
*BG_COLOR = main::$3
|
|
to:main::@return
|
|
main::@return: scope:[main] from main::@1
|
|
return
|
|
to:@return
|
|
|
|
void __start()
|
|
__start: scope:[__start] from
|
|
call main
|
|
to:__start::@1
|
|
__start::@1: scope:[__start] from __start
|
|
to:__start::@return
|
|
__start::@return: scope:[__start] from __start::@1
|
|
return
|
|
to:@return
|
|
|
|
SYMBOL TABLE SSA
|
|
__constant char * const BG_COLOR = (char *)$d021
|
|
__constant char * const BORDER_COLOR = (char *)$d020
|
|
void __start()
|
|
void main()
|
|
char main::$2
|
|
char main::$3
|
|
unsigned int main::kaputt
|
|
unsigned int main::kaputt#0
|
|
unsigned long main::result
|
|
unsigned long main::result#0
|
|
unsigned long mul16u(unsigned int a , unsigned int b)
|
|
bool mul16u::$0
|
|
number mul16u::$1
|
|
bool mul16u::$2
|
|
bool mul16u::$3
|
|
unsigned long mul16u::$4
|
|
unsigned int mul16u::$5
|
|
unsigned long mul16u::$6
|
|
unsigned int mul16u::a
|
|
unsigned int mul16u::a#0
|
|
unsigned int mul16u::a#1
|
|
unsigned int mul16u::a#2
|
|
unsigned int mul16u::a#3
|
|
unsigned int mul16u::a#4
|
|
unsigned int mul16u::a#5
|
|
unsigned int mul16u::a#6
|
|
unsigned int mul16u::b
|
|
unsigned int mul16u::b#0
|
|
unsigned int mul16u::b#1
|
|
unsigned long mul16u::mb
|
|
unsigned long mul16u::mb#0
|
|
unsigned long mul16u::mb#1
|
|
unsigned long mul16u::mb#2
|
|
unsigned long mul16u::mb#3
|
|
unsigned long mul16u::mb#4
|
|
unsigned long mul16u::mb#5
|
|
unsigned long mul16u::res
|
|
unsigned long mul16u::res#0
|
|
unsigned long mul16u::res#1
|
|
unsigned long mul16u::res#2
|
|
unsigned long mul16u::res#3
|
|
unsigned long mul16u::res#4
|
|
unsigned long mul16u::res#5
|
|
unsigned long mul16u::res#6
|
|
unsigned long mul16u::return
|
|
unsigned long mul16u::return#0
|
|
unsigned long mul16u::return#1
|
|
unsigned long mul16u::return#2
|
|
unsigned long mul16u::return#3
|
|
unsigned long mul16u::return#4
|
|
|
|
Adding number conversion cast (unumber) 0 in mul16u::$0 = mul16u::a#2 != 0
|
|
Adding number conversion cast (unumber) 1 in mul16u::$1 = mul16u::a#3 & 1
|
|
Adding number conversion cast (unumber) mul16u::$1 in mul16u::$1 = mul16u::a#3 & (unumber)1
|
|
Adding number conversion cast (unumber) 0 in mul16u::$2 = mul16u::$1 != 0
|
|
Adding number conversion cast (unumber) 1 in mul16u::$5 = mul16u::a#4 >> 1
|
|
Adding number conversion cast (unumber) 1 in mul16u::$6 = mul16u::mb#2 << 1
|
|
Adding number conversion cast (unumber) 4 in mul16u::a#1 = 4
|
|
Adding number conversion cast (unumber) $7b in mul16u::b#0 = $7b
|
|
Successful SSA optimization PassNAddNumberTypeConversions
|
|
Inlining cast mul16u::a#1 = (unumber)4
|
|
Inlining cast mul16u::b#0 = (unumber)$7b
|
|
Successful SSA optimization Pass2InlineCast
|
|
Simplifying constant pointer cast (char *) 53280
|
|
Simplifying constant pointer cast (char *) 53281
|
|
Simplifying constant integer cast 0
|
|
Simplifying constant integer cast 1
|
|
Simplifying constant integer cast 0
|
|
Simplifying constant integer cast 1
|
|
Simplifying constant integer cast 1
|
|
Simplifying constant integer cast 4
|
|
Simplifying constant integer cast $7b
|
|
Successful SSA optimization PassNCastSimplification
|
|
Finalized unsigned number type (char) 0
|
|
Finalized unsigned number type (char) 1
|
|
Finalized unsigned number type (char) 0
|
|
Finalized unsigned number type (char) 1
|
|
Finalized unsigned number type (char) 1
|
|
Finalized unsigned number type (char) 4
|
|
Finalized unsigned number type (char) $7b
|
|
Successful SSA optimization PassNFinalizeNumberTypeConversions
|
|
Inferred type updated to char in mul16u::$1 = mul16u::a#3 & 1
|
|
Inversing boolean not [9] mul16u::$3 = mul16u::$1 == 0 from [8] mul16u::$2 = mul16u::$1 != 0
|
|
Successful SSA optimization Pass2UnaryNotSimplification
|
|
Alias mul16u::a#2 = mul16u::a#3 mul16u::a#6
|
|
Alias mul16u::mb#3 = mul16u::mb#4 mul16u::mb#5
|
|
Alias mul16u::res#2 = mul16u::res#5 mul16u::res#4 mul16u::return#0 mul16u::res#3 mul16u::return#3 mul16u::return#1
|
|
Alias mul16u::a#0 = mul16u::$5
|
|
Alias mul16u::mb#1 = mul16u::$6
|
|
Alias mul16u::res#1 = mul16u::$4
|
|
Alias mul16u::return#2 = mul16u::return#4
|
|
Successful SSA optimization Pass2AliasElimination
|
|
Alias mul16u::a#2 = mul16u::a#4
|
|
Alias mul16u::mb#2 = mul16u::mb#3
|
|
Successful SSA optimization Pass2AliasElimination
|
|
Identical Phi Values mul16u::b#1 mul16u::b#0
|
|
Identical Phi Values mul16u::a#5 mul16u::a#1
|
|
Successful SSA optimization Pass2IdenticalPhiElimination
|
|
Simple Condition mul16u::$0 [5] if(mul16u::a#2!=0) goto mul16u::@2
|
|
Simple Condition mul16u::$3 [8] if(mul16u::$1==0) goto mul16u::@4
|
|
Successful SSA optimization Pass2ConditionalJumpSimplification
|
|
Constant mul16u::res#0 = 0
|
|
Constant mul16u::a#1 = 4
|
|
Constant mul16u::b#0 = $7b
|
|
Successful SSA optimization Pass2ConstantIdentification
|
|
Constant mul16u::mb#0 = (unsigned long)mul16u::b#0
|
|
Successful SSA optimization Pass2ConstantIdentification
|
|
Removing unused procedure __start
|
|
Removing unused procedure block __start
|
|
Removing unused procedure block __start::@1
|
|
Removing unused procedure block __start::@return
|
|
Successful SSA optimization PassNEliminateEmptyStart
|
|
Inlining constant with var siblings mul16u::res#0
|
|
Inlining constant with var siblings mul16u::a#1
|
|
Inlining constant with var siblings mul16u::mb#0
|
|
Constant inlined mul16u::res#0 = 0
|
|
Constant inlined mul16u::mb#0 = (unsigned long)mul16u::b#0
|
|
Constant inlined mul16u::a#1 = 4
|
|
Successful SSA optimization Pass2ConstantInlining
|
|
Added new block during phi lifting mul16u::@6(between mul16u::@2 and mul16u::@4)
|
|
Adding NOP phi() at start of main
|
|
Adding NOP phi() at start of mul16u
|
|
Adding NOP phi() at start of mul16u::@3
|
|
CALL GRAPH
|
|
Calls in [main] to mul16u:1
|
|
|
|
Created 4 initial phi equivalence classes
|
|
Coalesced [18] mul16u::res#9 = mul16u::res#1
|
|
Coalesced [22] mul16u::a#7 = mul16u::a#0
|
|
Coalesced [23] mul16u::res#7 = mul16u::res#6
|
|
Coalesced [24] mul16u::mb#6 = mul16u::mb#1
|
|
Coalesced (already) [25] mul16u::res#8 = mul16u::res#2
|
|
Coalesced down to 3 phi equivalence classes
|
|
Culled Empty Block label mul16u::@3
|
|
Culled Empty Block label mul16u::@6
|
|
Renumbering block mul16u::@4 to mul16u::@3
|
|
Renumbering block mul16u::@5 to mul16u::@4
|
|
Adding NOP phi() at start of main
|
|
Adding NOP phi() at start of mul16u
|
|
|
|
FINAL CONTROL FLOW GRAPH
|
|
|
|
void main()
|
|
main: scope:[main] from
|
|
[0] phi()
|
|
[1] call mul16u
|
|
[2] mul16u::return#2 = mul16u::res#2
|
|
to:main::@1
|
|
main::@1: scope:[main] from main
|
|
[3] main::result#0 = mul16u::return#2
|
|
[4] main::kaputt#0 = word0 main::result#0
|
|
[5] main::$2 = byte0 main::kaputt#0
|
|
[6] *BORDER_COLOR = main::$2
|
|
[7] main::$3 = byte1 main::kaputt#0
|
|
[8] *BG_COLOR = main::$3
|
|
to:main::@return
|
|
main::@return: scope:[main] from main::@1
|
|
[9] return
|
|
to:@return
|
|
|
|
unsigned long mul16u(unsigned int a , unsigned int b)
|
|
mul16u: scope:[mul16u] from main
|
|
[10] phi()
|
|
to:mul16u::@1
|
|
mul16u::@1: scope:[mul16u] from mul16u mul16u::@3
|
|
[11] mul16u::mb#2 = phi( mul16u/(unsigned long)mul16u::b#0, mul16u::@3/mul16u::mb#1 )
|
|
[11] mul16u::res#2 = phi( mul16u/0, mul16u::@3/mul16u::res#6 )
|
|
[11] mul16u::a#2 = phi( mul16u/4, mul16u::@3/mul16u::a#0 )
|
|
[12] if(mul16u::a#2!=0) goto mul16u::@2
|
|
to:mul16u::@return
|
|
mul16u::@return: scope:[mul16u] from mul16u::@1
|
|
[13] return
|
|
to:@return
|
|
mul16u::@2: scope:[mul16u] from mul16u::@1
|
|
[14] mul16u::$1 = mul16u::a#2 & 1
|
|
[15] if(mul16u::$1==0) goto mul16u::@3
|
|
to:mul16u::@4
|
|
mul16u::@4: scope:[mul16u] from mul16u::@2
|
|
[16] mul16u::res#1 = mul16u::res#2 + mul16u::mb#2
|
|
to:mul16u::@3
|
|
mul16u::@3: scope:[mul16u] from mul16u::@2 mul16u::@4
|
|
[17] mul16u::res#6 = phi( mul16u::@2/mul16u::res#2, mul16u::@4/mul16u::res#1 )
|
|
[18] mul16u::a#0 = mul16u::a#2 >> 1
|
|
[19] mul16u::mb#1 = mul16u::mb#2 << 1
|
|
to:mul16u::@1
|
|
|
|
|
|
VARIABLE REGISTER WEIGHTS
|
|
void main()
|
|
char main::$2 // 4.0
|
|
char main::$3 // 4.0
|
|
unsigned int main::kaputt
|
|
unsigned int main::kaputt#0 // 2.0
|
|
unsigned long main::result
|
|
unsigned long main::result#0 // 4.0
|
|
unsigned long mul16u(unsigned int a , unsigned int b)
|
|
char mul16u::$1 // 202.0
|
|
unsigned int mul16u::a
|
|
unsigned int mul16u::a#0 // 101.0
|
|
unsigned int mul16u::a#2 // 67.33333333333333
|
|
unsigned int mul16u::b
|
|
unsigned long mul16u::mb
|
|
unsigned long mul16u::mb#1 // 202.0
|
|
unsigned long mul16u::mb#2 // 43.285714285714285
|
|
unsigned long mul16u::res
|
|
unsigned long mul16u::res#1 // 202.0
|
|
unsigned long mul16u::res#2 // 50.83333333333333
|
|
unsigned long mul16u::res#6 // 101.0
|
|
unsigned long mul16u::return
|
|
unsigned long mul16u::return#2 // 4.0
|
|
|
|
Initial phi equivalence classes
|
|
[ mul16u::a#2 mul16u::a#0 ]
|
|
[ mul16u::res#2 mul16u::res#6 mul16u::res#1 ]
|
|
[ mul16u::mb#2 mul16u::mb#1 ]
|
|
Added variable mul16u::return#2 to live range equivalence class [ mul16u::return#2 ]
|
|
Added variable main::result#0 to live range equivalence class [ main::result#0 ]
|
|
Added variable main::kaputt#0 to live range equivalence class [ main::kaputt#0 ]
|
|
Added variable main::$2 to live range equivalence class [ main::$2 ]
|
|
Added variable main::$3 to live range equivalence class [ main::$3 ]
|
|
Added variable mul16u::$1 to live range equivalence class [ mul16u::$1 ]
|
|
Complete equivalence classes
|
|
[ mul16u::a#2 mul16u::a#0 ]
|
|
[ mul16u::res#2 mul16u::res#6 mul16u::res#1 ]
|
|
[ mul16u::mb#2 mul16u::mb#1 ]
|
|
[ mul16u::return#2 ]
|
|
[ main::result#0 ]
|
|
[ main::kaputt#0 ]
|
|
[ main::$2 ]
|
|
[ main::$3 ]
|
|
[ mul16u::$1 ]
|
|
Allocated zp[4]:2 [ mul16u::res#2 mul16u::res#6 mul16u::res#1 ]
|
|
Allocated zp[4]:6 [ mul16u::mb#2 mul16u::mb#1 ]
|
|
Allocated zp[1]:10 [ mul16u::$1 ]
|
|
Allocated zp[2]:11 [ mul16u::a#2 mul16u::a#0 ]
|
|
Allocated zp[4]:13 [ mul16u::return#2 ]
|
|
Allocated zp[4]:17 [ main::result#0 ]
|
|
Allocated zp[1]:21 [ main::$2 ]
|
|
Allocated zp[1]:22 [ main::$3 ]
|
|
Allocated zp[2]:23 [ main::kaputt#0 ]
|
|
REGISTER UPLIFT POTENTIAL REGISTERS
|
|
Statement [2] mul16u::return#2 = mul16u::res#2 [ mul16u::return#2 ] ( [ mul16u::return#2 ] { { mul16u::return#2 = mul16u::res#2 } } ) always clobbers reg byte a
|
|
Statement [3] main::result#0 = mul16u::return#2 [ main::result#0 ] ( [ main::result#0 ] { } ) always clobbers reg byte a
|
|
Statement [4] main::kaputt#0 = word0 main::result#0 [ main::kaputt#0 ] ( [ main::kaputt#0 ] { } ) always clobbers reg byte a
|
|
Statement [12] if(mul16u::a#2!=0) goto mul16u::@2 [ mul16u::res#2 mul16u::a#2 mul16u::mb#2 ] ( mul16u:1 [ mul16u::res#2 mul16u::a#2 mul16u::mb#2 ] { { mul16u::return#2 = mul16u::res#2 } } ) always clobbers reg byte a
|
|
Statement [14] mul16u::$1 = mul16u::a#2 & 1 [ mul16u::res#2 mul16u::a#2 mul16u::mb#2 mul16u::$1 ] ( mul16u:1 [ mul16u::res#2 mul16u::a#2 mul16u::mb#2 mul16u::$1 ] { { mul16u::return#2 = mul16u::res#2 } } ) always clobbers reg byte a
|
|
Statement [16] mul16u::res#1 = mul16u::res#2 + mul16u::mb#2 [ mul16u::a#2 mul16u::mb#2 mul16u::res#1 ] ( mul16u:1 [ mul16u::a#2 mul16u::mb#2 mul16u::res#1 ] { { mul16u::return#2 = mul16u::res#2 } } ) always clobbers reg byte a
|
|
Potential registers zp[2]:11 [ mul16u::a#2 mul16u::a#0 ] : zp[2]:11 ,
|
|
Potential registers zp[4]:2 [ mul16u::res#2 mul16u::res#6 mul16u::res#1 ] : zp[4]:2 ,
|
|
Potential registers zp[4]:6 [ mul16u::mb#2 mul16u::mb#1 ] : zp[4]:6 ,
|
|
Potential registers zp[4]:13 [ mul16u::return#2 ] : zp[4]:13 ,
|
|
Potential registers zp[4]:17 [ main::result#0 ] : zp[4]:17 ,
|
|
Potential registers zp[2]:23 [ main::kaputt#0 ] : zp[2]:23 ,
|
|
Potential registers zp[1]:21 [ main::$2 ] : zp[1]:21 , reg byte a , reg byte x , reg byte y ,
|
|
Potential registers zp[1]:22 [ main::$3 ] : zp[1]:22 , reg byte a , reg byte x , reg byte y ,
|
|
Potential registers zp[1]:10 [ mul16u::$1 ] : zp[1]:10 , reg byte a , reg byte x , reg byte y ,
|
|
|
|
REGISTER UPLIFT SCOPES
|
|
Uplift Scope [mul16u] 353.83: zp[4]:2 [ mul16u::res#2 mul16u::res#6 mul16u::res#1 ] 245.29: zp[4]:6 [ mul16u::mb#2 mul16u::mb#1 ] 202: zp[1]:10 [ mul16u::$1 ] 168.33: zp[2]:11 [ mul16u::a#2 mul16u::a#0 ] 4: zp[4]:13 [ mul16u::return#2 ]
|
|
Uplift Scope [main] 4: zp[4]:17 [ main::result#0 ] 4: zp[1]:21 [ main::$2 ] 4: zp[1]:22 [ main::$3 ] 2: zp[2]:23 [ main::kaputt#0 ]
|
|
Uplift Scope [MOS6526_CIA]
|
|
Uplift Scope [MOS6569_VICII]
|
|
Uplift Scope [MOS6581_SID]
|
|
Uplift Scope []
|
|
|
|
Uplifting [mul16u] best 1620 combination zp[4]:2 [ mul16u::res#2 mul16u::res#6 mul16u::res#1 ] zp[4]:6 [ mul16u::mb#2 mul16u::mb#1 ] reg byte a [ mul16u::$1 ] zp[2]:11 [ mul16u::a#2 mul16u::a#0 ] zp[4]:13 [ mul16u::return#2 ]
|
|
Uplifting [main] best 1608 combination zp[4]:17 [ main::result#0 ] reg byte a [ main::$2 ] reg byte a [ main::$3 ] zp[2]:23 [ main::kaputt#0 ]
|
|
Uplifting [MOS6526_CIA] best 1608 combination
|
|
Uplifting [MOS6569_VICII] best 1608 combination
|
|
Uplifting [MOS6581_SID] best 1608 combination
|
|
Uplifting [] best 1608 combination
|
|
Coalescing zero page register [ zp[4]:2 [ mul16u::res#2 mul16u::res#6 mul16u::res#1 ] ] with [ zp[4]:13 [ mul16u::return#2 ] ] - score: 1
|
|
Coalescing zero page register [ zp[4]:2 [ mul16u::res#2 mul16u::res#6 mul16u::res#1 mul16u::return#2 ] ] with [ zp[4]:17 [ main::result#0 ] ] - score: 1
|
|
Coalescing zero page register [ zp[2]:23 [ main::kaputt#0 ] ] with [ zp[2]:11 [ mul16u::a#2 mul16u::a#0 ] ]
|
|
Allocated (was zp[2]:23) zp[2]:10 [ main::kaputt#0 mul16u::a#2 mul16u::a#0 ]
|
|
|
|
ASSEMBLER BEFORE OPTIMIZATION
|
|
// File Comments
|
|
// Program where loop-head optimization produces wrong return value
|
|
// Reported by Richard-William Loerakker
|
|
// Upstart
|
|
// Commodore 64 PRG executable file
|
|
.file [name="loophead-problem-3.prg", type="prg", segments="Program"]
|
|
.segmentdef Program [segments="Basic, Code, Data"]
|
|
.segmentdef Basic [start=$0801]
|
|
.segmentdef Code [start=$80d]
|
|
.segmentdef Data [startAfter="Code"]
|
|
.segment Basic
|
|
:BasicUpstart(main)
|
|
// Global Constants & labels
|
|
/// $D020 Border Color
|
|
.label BORDER_COLOR = $d020
|
|
/// $D021 Background Color 0
|
|
.label BG_COLOR = $d021
|
|
.segment Code
|
|
// main
|
|
main: {
|
|
.label result = 2
|
|
.label kaputt = $a
|
|
// [1] call mul16u
|
|
// [10] phi from main to mul16u [phi:main->mul16u]
|
|
mul16u_from_main:
|
|
jsr mul16u
|
|
// [2] mul16u::return#2 = mul16u::res#2
|
|
jmp __b1
|
|
// main::@1
|
|
__b1:
|
|
// [3] main::result#0 = mul16u::return#2
|
|
// [4] main::kaputt#0 = word0 main::result#0 -- vwuz1=_word0_vduz2
|
|
lda.z result
|
|
sta.z kaputt
|
|
lda.z result+1
|
|
sta.z kaputt+1
|
|
// [5] main::$2 = byte0 main::kaputt#0 -- vbuaa=_byte0_vwuz1
|
|
lda.z kaputt
|
|
// [6] *BORDER_COLOR = main::$2 -- _deref_pbuc1=vbuaa
|
|
sta BORDER_COLOR
|
|
// [7] main::$3 = byte1 main::kaputt#0 -- vbuaa=_byte1_vwuz1
|
|
lda.z kaputt+1
|
|
// [8] *BG_COLOR = main::$3 -- _deref_pbuc1=vbuaa
|
|
sta BG_COLOR
|
|
jmp __breturn
|
|
// main::@return
|
|
__breturn:
|
|
// [9] return
|
|
rts
|
|
}
|
|
// mul16u
|
|
// Perform binary multiplication of two unsigned 16-bit unsigned ints into a 32-bit unsigned long
|
|
// __zp(2) unsigned long mul16u(__zp($a) unsigned int a, unsigned int b)
|
|
mul16u: {
|
|
.const b = $7b
|
|
.label a = $a
|
|
.label mb = 6
|
|
.label res = 2
|
|
.label return = 2
|
|
// [11] phi from mul16u to mul16u::@1 [phi:mul16u->mul16u::@1]
|
|
__b1_from_mul16u:
|
|
// [11] phi mul16u::mb#2 = (unsigned long)mul16u::b#0 [phi:mul16u->mul16u::@1#0] -- vduz1=vduc1
|
|
lda #<b
|
|
sta.z mb
|
|
lda #>b
|
|
sta.z mb+1
|
|
lda #<b>>$10
|
|
sta.z mb+2
|
|
lda #>b>>$10
|
|
sta.z mb+3
|
|
// [11] phi mul16u::res#2 = 0 [phi:mul16u->mul16u::@1#1] -- vduz1=vduc1
|
|
lda #<0
|
|
sta.z res
|
|
lda #>0
|
|
sta.z res+1
|
|
lda #<0>>$10
|
|
sta.z res+2
|
|
lda #>0>>$10
|
|
sta.z res+3
|
|
// [11] phi mul16u::a#2 = 4 [phi:mul16u->mul16u::@1#2] -- vwuz1=vbuc1
|
|
lda #<4
|
|
sta.z a
|
|
lda #>4
|
|
sta.z a+1
|
|
jmp __b1
|
|
// mul16u::@1
|
|
__b1:
|
|
// [12] if(mul16u::a#2!=0) goto mul16u::@2 -- vwuz1_neq_0_then_la1
|
|
lda.z a
|
|
ora.z a+1
|
|
bne __b2
|
|
jmp __breturn
|
|
// mul16u::@return
|
|
__breturn:
|
|
// [13] return
|
|
rts
|
|
// mul16u::@2
|
|
__b2:
|
|
// [14] mul16u::$1 = mul16u::a#2 & 1 -- vbuaa=vwuz1_band_vbuc1
|
|
lda #1
|
|
and.z a
|
|
// [15] if(mul16u::$1==0) goto mul16u::@3 -- vbuaa_eq_0_then_la1
|
|
cmp #0
|
|
beq __b3_from___b2
|
|
jmp __b4
|
|
// mul16u::@4
|
|
__b4:
|
|
// [16] mul16u::res#1 = mul16u::res#2 + mul16u::mb#2 -- vduz1=vduz1_plus_vduz2
|
|
clc
|
|
lda.z res
|
|
adc.z mb
|
|
sta.z res
|
|
lda.z res+1
|
|
adc.z mb+1
|
|
sta.z res+1
|
|
lda.z res+2
|
|
adc.z mb+2
|
|
sta.z res+2
|
|
lda.z res+3
|
|
adc.z mb+3
|
|
sta.z res+3
|
|
// [17] phi from mul16u::@2 mul16u::@4 to mul16u::@3 [phi:mul16u::@2/mul16u::@4->mul16u::@3]
|
|
__b3_from___b2:
|
|
__b3_from___b4:
|
|
// [17] phi mul16u::res#6 = mul16u::res#2 [phi:mul16u::@2/mul16u::@4->mul16u::@3#0] -- register_copy
|
|
jmp __b3
|
|
// mul16u::@3
|
|
__b3:
|
|
// [18] mul16u::a#0 = mul16u::a#2 >> 1 -- vwuz1=vwuz1_ror_1
|
|
lsr.z a+1
|
|
ror.z a
|
|
// [19] mul16u::mb#1 = mul16u::mb#2 << 1 -- vduz1=vduz1_rol_1
|
|
asl.z mb
|
|
rol.z mb+1
|
|
rol.z mb+2
|
|
rol.z mb+3
|
|
// [11] phi from mul16u::@3 to mul16u::@1 [phi:mul16u::@3->mul16u::@1]
|
|
__b1_from___b3:
|
|
// [11] phi mul16u::mb#2 = mul16u::mb#1 [phi:mul16u::@3->mul16u::@1#0] -- register_copy
|
|
// [11] phi mul16u::res#2 = mul16u::res#6 [phi:mul16u::@3->mul16u::@1#1] -- register_copy
|
|
// [11] phi mul16u::a#2 = mul16u::a#0 [phi:mul16u::@3->mul16u::@1#2] -- register_copy
|
|
jmp __b1
|
|
}
|
|
// File Data
|
|
|
|
ASSEMBLER OPTIMIZATIONS
|
|
Removing instruction jmp __b1
|
|
Removing instruction jmp __breturn
|
|
Removing instruction jmp __b1
|
|
Removing instruction jmp __breturn
|
|
Removing instruction jmp __b4
|
|
Removing instruction jmp __b3
|
|
Succesful ASM optimization Pass5NextJumpElimination
|
|
Removing instruction lda #>0
|
|
Succesful ASM optimization Pass5UnnecesaryLoadElimination
|
|
Replacing label __b3_from___b2 with __b3
|
|
Removing instruction __b3_from___b2:
|
|
Removing instruction __b3_from___b4:
|
|
Succesful ASM optimization Pass5RedundantLabelElimination
|
|
Removing instruction mul16u_from_main:
|
|
Removing instruction __b1:
|
|
Removing instruction __breturn:
|
|
Removing instruction __b1_from_mul16u:
|
|
Removing instruction __breturn:
|
|
Removing instruction __b4:
|
|
Removing instruction __b1_from___b3:
|
|
Succesful ASM optimization Pass5UnusedLabelElimination
|
|
|
|
FINAL SYMBOL TABLE
|
|
__constant char * const BG_COLOR = (char *) 53281
|
|
__constant char * const BORDER_COLOR = (char *) 53280
|
|
void main()
|
|
char main::$2 // reg byte a 4.0
|
|
char main::$3 // reg byte a 4.0
|
|
unsigned int main::kaputt
|
|
unsigned int main::kaputt#0 // kaputt zp[2]:10 2.0
|
|
unsigned long main::result
|
|
unsigned long main::result#0 // result zp[4]:2 4.0
|
|
unsigned long mul16u(unsigned int a , unsigned int b)
|
|
char mul16u::$1 // reg byte a 202.0
|
|
unsigned int mul16u::a
|
|
unsigned int mul16u::a#0 // a zp[2]:10 101.0
|
|
unsigned int mul16u::a#2 // a zp[2]:10 67.33333333333333
|
|
unsigned int mul16u::b
|
|
__constant unsigned int mul16u::b#0 = $7b // b
|
|
unsigned long mul16u::mb
|
|
unsigned long mul16u::mb#1 // mb zp[4]:6 202.0
|
|
unsigned long mul16u::mb#2 // mb zp[4]:6 43.285714285714285
|
|
unsigned long mul16u::res
|
|
unsigned long mul16u::res#1 // res zp[4]:2 202.0
|
|
unsigned long mul16u::res#2 // res zp[4]:2 50.83333333333333
|
|
unsigned long mul16u::res#6 // res zp[4]:2 101.0
|
|
unsigned long mul16u::return
|
|
unsigned long mul16u::return#2 // return zp[4]:2 4.0
|
|
|
|
zp[4]:2 [ mul16u::res#2 mul16u::res#6 mul16u::res#1 mul16u::return#2 main::result#0 ]
|
|
zp[4]:6 [ mul16u::mb#2 mul16u::mb#1 ]
|
|
zp[2]:10 [ main::kaputt#0 mul16u::a#2 mul16u::a#0 ]
|
|
reg byte a [ main::$2 ]
|
|
reg byte a [ main::$3 ]
|
|
reg byte a [ mul16u::$1 ]
|
|
|
|
|
|
FINAL ASSEMBLER
|
|
Score: 1414
|
|
|
|
// File Comments
|
|
// Program where loop-head optimization produces wrong return value
|
|
// Reported by Richard-William Loerakker
|
|
// Upstart
|
|
// Commodore 64 PRG executable file
|
|
.file [name="loophead-problem-3.prg", type="prg", segments="Program"]
|
|
.segmentdef Program [segments="Basic, Code, Data"]
|
|
.segmentdef Basic [start=$0801]
|
|
.segmentdef Code [start=$80d]
|
|
.segmentdef Data [startAfter="Code"]
|
|
.segment Basic
|
|
:BasicUpstart(main)
|
|
// Global Constants & labels
|
|
/// $D020 Border Color
|
|
.label BORDER_COLOR = $d020
|
|
/// $D021 Background Color 0
|
|
.label BG_COLOR = $d021
|
|
.segment Code
|
|
// main
|
|
main: {
|
|
.label result = 2
|
|
.label kaputt = $a
|
|
// dword result = mul16u(4,123)
|
|
// [1] call mul16u
|
|
// [10] phi from main to mul16u [phi:main->mul16u]
|
|
jsr mul16u
|
|
// dword result = mul16u(4,123)
|
|
// [2] mul16u::return#2 = mul16u::res#2
|
|
// main::@1
|
|
// [3] main::result#0 = mul16u::return#2
|
|
// word kaputt = WORD0(result)
|
|
// [4] main::kaputt#0 = word0 main::result#0 -- vwuz1=_word0_vduz2
|
|
lda.z result
|
|
sta.z kaputt
|
|
lda.z result+1
|
|
sta.z kaputt+1
|
|
// BYTE0(kaputt)
|
|
// [5] main::$2 = byte0 main::kaputt#0 -- vbuaa=_byte0_vwuz1
|
|
lda.z kaputt
|
|
// *BORDER_COLOR = BYTE0(kaputt)
|
|
// [6] *BORDER_COLOR = main::$2 -- _deref_pbuc1=vbuaa
|
|
sta BORDER_COLOR
|
|
// BYTE1(kaputt)
|
|
// [7] main::$3 = byte1 main::kaputt#0 -- vbuaa=_byte1_vwuz1
|
|
lda.z kaputt+1
|
|
// *BG_COLOR = BYTE1(kaputt)
|
|
// [8] *BG_COLOR = main::$3 -- _deref_pbuc1=vbuaa
|
|
sta BG_COLOR
|
|
// main::@return
|
|
// }
|
|
// [9] return
|
|
rts
|
|
}
|
|
// mul16u
|
|
// Perform binary multiplication of two unsigned 16-bit unsigned ints into a 32-bit unsigned long
|
|
// __zp(2) unsigned long mul16u(__zp($a) unsigned int a, unsigned int b)
|
|
mul16u: {
|
|
.const b = $7b
|
|
.label a = $a
|
|
.label mb = 6
|
|
.label res = 2
|
|
.label return = 2
|
|
// [11] phi from mul16u to mul16u::@1 [phi:mul16u->mul16u::@1]
|
|
// [11] phi mul16u::mb#2 = (unsigned long)mul16u::b#0 [phi:mul16u->mul16u::@1#0] -- vduz1=vduc1
|
|
lda #<b
|
|
sta.z mb
|
|
lda #>b
|
|
sta.z mb+1
|
|
lda #<b>>$10
|
|
sta.z mb+2
|
|
lda #>b>>$10
|
|
sta.z mb+3
|
|
// [11] phi mul16u::res#2 = 0 [phi:mul16u->mul16u::@1#1] -- vduz1=vduc1
|
|
lda #<0
|
|
sta.z res
|
|
sta.z res+1
|
|
lda #<0>>$10
|
|
sta.z res+2
|
|
lda #>0>>$10
|
|
sta.z res+3
|
|
// [11] phi mul16u::a#2 = 4 [phi:mul16u->mul16u::@1#2] -- vwuz1=vbuc1
|
|
lda #<4
|
|
sta.z a
|
|
lda #>4
|
|
sta.z a+1
|
|
// mul16u::@1
|
|
__b1:
|
|
// while(a!=0)
|
|
// [12] if(mul16u::a#2!=0) goto mul16u::@2 -- vwuz1_neq_0_then_la1
|
|
lda.z a
|
|
ora.z a+1
|
|
bne __b2
|
|
// mul16u::@return
|
|
// }
|
|
// [13] return
|
|
rts
|
|
// mul16u::@2
|
|
__b2:
|
|
// a&1
|
|
// [14] mul16u::$1 = mul16u::a#2 & 1 -- vbuaa=vwuz1_band_vbuc1
|
|
lda #1
|
|
and.z a
|
|
// if( (a&1) != 0)
|
|
// [15] if(mul16u::$1==0) goto mul16u::@3 -- vbuaa_eq_0_then_la1
|
|
cmp #0
|
|
beq __b3
|
|
// mul16u::@4
|
|
// res = res + mb
|
|
// [16] mul16u::res#1 = mul16u::res#2 + mul16u::mb#2 -- vduz1=vduz1_plus_vduz2
|
|
clc
|
|
lda.z res
|
|
adc.z mb
|
|
sta.z res
|
|
lda.z res+1
|
|
adc.z mb+1
|
|
sta.z res+1
|
|
lda.z res+2
|
|
adc.z mb+2
|
|
sta.z res+2
|
|
lda.z res+3
|
|
adc.z mb+3
|
|
sta.z res+3
|
|
// [17] phi from mul16u::@2 mul16u::@4 to mul16u::@3 [phi:mul16u::@2/mul16u::@4->mul16u::@3]
|
|
// [17] phi mul16u::res#6 = mul16u::res#2 [phi:mul16u::@2/mul16u::@4->mul16u::@3#0] -- register_copy
|
|
// mul16u::@3
|
|
__b3:
|
|
// a = a>>1
|
|
// [18] mul16u::a#0 = mul16u::a#2 >> 1 -- vwuz1=vwuz1_ror_1
|
|
lsr.z a+1
|
|
ror.z a
|
|
// mb = mb<<1
|
|
// [19] mul16u::mb#1 = mul16u::mb#2 << 1 -- vduz1=vduz1_rol_1
|
|
asl.z mb
|
|
rol.z mb+1
|
|
rol.z mb+2
|
|
rol.z mb+3
|
|
// [11] phi from mul16u::@3 to mul16u::@1 [phi:mul16u::@3->mul16u::@1]
|
|
// [11] phi mul16u::mb#2 = mul16u::mb#1 [phi:mul16u::@3->mul16u::@1#0] -- register_copy
|
|
// [11] phi mul16u::res#2 = mul16u::res#6 [phi:mul16u::@3->mul16u::@1#1] -- register_copy
|
|
// [11] phi mul16u::a#2 = mul16u::a#0 [phi:mul16u::@3->mul16u::@1#2] -- register_copy
|
|
jmp __b1
|
|
}
|
|
// File Data
|
|
|