Systems Programming and Computer Architecture ( )

Transcription

1 Systems Group Department of Computer Science ETH Zürich Systems Programming and Computer Architecture ( ) Timothy Roscoe Herbstsemester

2 3: Integers in C Computer Architecture and Systems Programming , Herbstsemester 2016 Timothy Roscoe 2

3 Representing integers int A = 15213; int B = ; long int C = 15213; Decimal: Binary: Hex: 3 B 6 D A on ia32, x86-64 A on SPARC C on ia32 C on x86-64 C on SPARC 6D 3B B on ia32, x C4 FF FF B 6D B on SPARC FF FF C4 93 6D 3B D 3B Two s complement representation Integers in C B 6D 3

4 Bit-level operations in C Operations &,, ~, ^ available in C Apply to any integral data type long, int, short, char, unsigned View arguments as bit vectors Arguments applied bit-wise &,, ~, ^ Examples (char data type): ~0x41 0xBE ~ ~0x00 0xFF ~ x69 & 0x55 0x & x69 0x55 0x7D

5 Contrast: Logic operations in C &&,,! View 0 as False Anything nonzero as True Always return 0 or 1 Early termination &&,,! Examples (char data type)!0x41 0x00!0x00 0x01!!0x41 0x01 0x69 && 0x55 0x01 0x69 0x55 0x01 p && *p (avoids null pointer access) 5

6 Representing & manipulating sets Width w bit vector represents subsets of {0,, w 1} a j = 1 if j A: { 0, 3, 5, 6 } { 0, 2, 4, 6 } Operations: Operator Operation Result Meaning & Intersection { 0, 6 } Union { 2, 3, 4, 5, 6 } ^ Symmetric difference { 2, 3, 4, 5 } ~ Complement { 1, 3, 5, 7 } 6

7 Shift operations Left shift: x << y Shift bit-vector x left y positions Throw away extra bits on left Fill with 0 s on right Right shift: x >> y Shift bit-vector x right y positions Throw away extra bits on right Logical shift Fill with 0 s on left Arithmetic shift Replicate most significant bit on right Undefined behavior Shift amount < 0 or word size Argument x << 3 Log. >> 2 Arith. >> 2 Argument x << 3 Log. >> 2 Arith. >> Java writes this >>>. 7

8 Summary Integer encoding Bit-level operators Set representation Logic operators Shift operators 8

9 3.1: Integer ranges Computer Architecture and Systems Programming , Herbstsemester 2016 Timothy Roscoe 9

10 Encoding integers Unsigned w 1 B2U X = x i 2 i i=0 Two s complement w 2 B2T X = x w 1 2 w 1 + x i 2 i i=0 short int x = 15213; short int y = ; Sign Bit A C short is 2 bytes long: Decimal Hex Binary x B 6D y C Sign bit For 2 s complement, most significant bit = 1 indicates negative 10

11 Numeric ranges Unsigned values UMin = UMax = 2 w Two s complement values TMin = 2 w TMax = 2 w Decimal Hex Binary Values for w=16 UMax FF FF TMax F FF TMin FF FF

12 Values for different word sizes UMax ,535 4,294,967,295 18,446,744,073,709,551,615 TMax ,767 2,147,483,647 9,223,372,036,854,775,807 TMin ,768-2,147,483,648-9,223,372,036,854,775,808 Observations TMin = TMax + 1 Asymmetric range UMax = 2 * TMax + 1 C Programming #include <limits.h> Declares constants, e.g., ULONG_MAX LONG_MAX LONG_MIN Values platform specific 12

13 Mapping signed unsigned = +16 Signed Unsigned Bits

14 Conversion visualized 2 s Comp. Unsigned Ordering inversion Negative big positive TMax UMax UMax 1 TMax + 1 TMax Unsigned range 2 s complement range TMin AS

15 Signed vs. unsigned in C Constants By default are considered to be signed integers Unsigned if have U as suffix: Casting Explicit casting between signed & unsigned same as U2T and T2U: 0U U int tx, ty; unsigned ux, uy; tx = (int) ux; uy = (unsigned) ty; Implicit casting also occurs via assignments and procedure calls tx = ux; uy = ty; 15

16 Casting surprises Expression evaluation If mix unsigned and signed in single expression, signed values implicitly cast to unsigned Including comparison operations <, >, ==, <=, >= Examples for W = 32: TMIN = -2,147,483,648, TMAX = 2,147,483,647 Constant 1 Constant 2 Relation Evaluation 0 0U == Unsigned -1 0 < Signed -1 0U > Unsigned > Signed U < Unsigned -1-2 > Signed (unsigned)-1-2 > Unsigned U < Unsigned (int) U > signed AS

17 Code security example /* Kernel memory region holding user-accessible data */ #define KSIZE 1024 char kbuf[ksize]; /* Copy at most maxlen bytes from kernel region to user buffer */ int copy_from_kernel(void *user_dest, int maxlen) { /* Byte count len is minimum of buffer size and maxlen */ int len = KSIZE < maxlen? KSIZE : maxlen; memcpy(user_dest, kbuf, len); return len; } Similar to code found in FreeBSD s implementation of getpeername There are legions of smart people trying to find vulnerabilities in programs 17

18 Typical usage /* Kernel memory region holding user-accessible data */ #define KSIZE 1024 char kbuf[ksize]; /* Copy at most maxlen bytes from kernel region to user buffer */ int copy_from_kernel(void *user_dest, int maxlen) { /* Byte count len is minimum of buffer size and maxlen */ int len = KSIZE < maxlen? KSIZE : maxlen; memcpy(user_dest, kbuf, len); return len; } #define MSIZE 528 void getstuff() { char mybuf[msize]; copy_from_kernel(mybuf, MSIZE); printf( %s\n, mybuf); } 18

19 Malicious usage /* Kernel memory region holding user-accessible data */ #define KSIZE 1024 char kbuf[ksize]; /* Copy at most maxlen bytes from kernel region to user buffer */ int copy_from_kernel(void *user_dest, int maxlen) { /* Byte count len is minimum of buffer size and maxlen */ int len = KSIZE < maxlen? KSIZE : maxlen; memcpy(user_dest, kbuf, len); return len; } #define MSIZE 528 void getstuff() { char mybuf[msize]; copy_from_kernel(mybuf, -MSIZE);... } /* Declaration of library function memcpy */ void *memcpy(void *dest, void *src, size_t n); 19

20 Sign extension Task: Given w-bit signed integer x Convert it to w+k-bit integer with same value Rule: Make k copies of sign bit: X = x w 1,, x w 1, x w 1, x w 2,, x 0 w k copies of MSB X X k w 20

21 Sign extension example short int x = 15213; int ix = (int) x; short int y = ; int iy = (int) y; Decimal Hex Binary x B 6D ix B 6D y C iy FF FF C Converting from smaller to larger integer data type C automatically performs sign extension for signed values 21

22 Summary Casting signed unsigned Bit pattern is maintained but reinterpreted Can have unexpected effects: adding or subtracting 2 w Expression containing signed and unsigned int int is cast to unsigned! Expanding (e.g., short int int) Unsigned: zeros added Signed: sign extension Truncating (e.g., unsigned unsigned short) Unsigned: modulus operation Signed: similar to modulus For small numbers yields expected behaviour 22

23 3.2: Integer addition and subtraction in C Computer Architecture and Systems Programming , Herbstsemester 2016 Timothy Roscoe 23

24 Negation: complement & increment Recall the following holds for 2 s complement: ~x + 1 == -x Complement Observation: ~x + x == == -1 + x ~x Complete proof? 24

25 Complement & increment examples x = x = 0 Decimal Hex Binary x B 6D ~x C ~x C y C Decimal Hex Binary ~0-1 FF FF ~

26 Unsigned addition Operands: w bits True sum: w+1 bits Discard carry: w bits u + v u + v UAdd w u, v Standard addition function Ignores carry output Implements modular arithmetic s = UAdd w u, v = u + v mod 2 w UAdd w u, v = u + v, u + v < 2w u + v 2 w, u + v 2 w 26

27 Visualizing (mathematical) integer addition Integer addition 4-bit integers u, v Add 4 (u, v) Integer Addition Compute true sum Add 4 (u, v) Values increase linearly with u and v Forms planar surface u v

28 Visualizing unsigned addition Wraps around If true sum 2 w At most once UAdd 4 (u, v) Overflow True Sum 2 w+1 Overflow w 0 Modular Sum u v 14 28

29 Mathematical properties Modular addition forms an Abelian group Closed under addition 0 UAdd w u, v 2 w 1 Commutative UAdd w u, v = UAdd w v, u Associative UAdd w t, UAdd w u, v = UAdd w UAdd w t, u, v 0 is additive identity UAdd w u, 0 = u Every element has additive inverse Let: UComp w u = 2 w u Then: UAdd w u, UComp w u = 0 29

30 Two s complement addition Operands: w bits True sum: w+1 bits Discard carry: w bits u + v u + v TAdd w u, v TAdd and UAdd have identical bit-level behavior Signed vs. unsigned addition in C: int s, t, u, v; s = (int) ((unsigned) u + (unsigned) v); t = u + v; Will give: s == t 30

31 TAdd overflow Functionality True sum requires w+1 bits Drop off MSB True Sum 2 w 1 2 w 1 0 PosOver TAdd Result Treat remaining bits as 2 s comp. integer w w NegOver

32 Visualizing 2 s compl. addition Values 4-bit two s comp. Range from -8 to +7 Wraps around If sum 2 w 1 Becomes negative At most once If sum < 2 w 1 Becomes positive At most once NegOver u 0 2 TAdd 4 (u, v) v PosOver 32

33 Characterizing TAdd Functionality True sum requires w+1 bits Drop off MSB Treat remaining bits as 2 s comp. integer TAdd(u, v) > 0 v < 0 < 0 > 0 u Negative Overflow Positive Overflow TAdd w u, v = u + v + 2 w u + v u + v 2 w u + v < TMin w TMin w u + v TMax w TMax w < u + v (Neg. overflow) (Pos. overflow) 33

34 Mathematical properties of TAdd Group isomorphic to unsigneds with UAdd Since both have identical bit patterns TAdd w u, v = U2T UAdd w T2U u, T2U v 2 s complement under TAdd forms a group Closed, commutative, associative, 0 is additive identity Every element has additive inverse TComp w u = u TMin w u TMin w u = TMin w 34

35 3.3: Integer multiplication in C Computer Architecture and Systems Programming , Herbstsemester 2016 Timothy Roscoe 35

36 Multiplication Computing exact product of w-bit numbers x, y Either signed or unsigned Ranges Unsigned (up to 2 w bits): 0 x y 2 w 1 2 = 2 2w 2 w Two s complement min (up to 2 w 1 bits): x y 2 w 1 2 w 1 1 = 2 2w w 1 Two s complement max (up to 2 w bits, but only for TMin 2 w ): x y 2 w 1 2 = 2 2w 2 Maintaining exact results Would need to keep expanding word size with each product computed Done in software by arbitrary precision arithmetic packages 36

37 Unsigned multiplication in C Operands: w bits u v True product: 2*w bits u v Discard w bits: w bits UMult w u, v Standard multiplication function Ignores high order w bits Implements modular arithmetic UMult w u, v = u v mod 2 w 37

38 Unsigned multiplication Unsigned multiplication with addition forms a commutative ring: Addition is a commutative group Closed under multiplication 0 UMult w u, v 2 w 1 Multiplication is commutative UMult w u, v = UMult w v, u Multiplication is associative UMult w t, UMult w u, v = UMult w UMult w t, u, v 1 is multiplicative identity UMult w u, 1 = u Multiplication distributes over addition UMult w t, UAdd w u, v = UAdd w UMult w t, u, UMult w t, v 38

39 Signed multiplication in C Operands: w bits u v True product: 2*w bits u v Discard w bits: w bits TMult w u, v Standard multiplication function Ignores high order w bits Some of which are different for signed vs. unsigned multiplication Lower bits are the same 39

40 Signed multiplication Isomorphic algebra to unsigned multiplication and addition Both isomorphic to ring of integers mod 2 w Comparison to (mathematical) integer arithmetic Both are rings True integers obey ordering properties, e.g., u > 0 u + v > v u > 0, v > 0 u v > 0 Not the case for two s complement arithmetic: TMax + 1 == TMin * == (e.g. for 16-bit words) 40

41 3.4: Integer multiplication and division using shifts Computer Architecture and Systems Programming , Herbstsemester 2016 Timothy Roscoe 41

42 Power-of-2 multiply with shift Operation u << k gives u * 2 k Both signed and unsigned Operands: w bits True product: w+k bits Discard k bits: w bits u 2 k Examples u << 3 == u * 8 u << 5 - u << 3 == u * 24 Most machines shift and add faster than multiply Compiler generates this code automatically u * 2 k UMult w u, 2 k TMult w u, 2 k k

43 Compiled multiplication code C Function int mul12(int x) { return x*12; } Compiled arithmetic operations leal (%eax,%eax,2), %eax sall $2, %eax Explanation t <- x+x*2 return t << 2; C compiler automatically generates shift/add code when multiplying by constant 43

44 Unsigned power-of-2 divide w/ shift Quotient of unsigned by power of 2 u >> k gives u / 2 k Uses logical shift k u Operands: / 2 k Binary Point Division: u / 2 k Result: u / 2 k Division Computed Hex Binary x B 6D x >> D B x >> B x >> B

45 Compiled unsigned division code C Function unsigned udiv8(unsigned x) { return x/8; } Compiled Arithmetic Operations shrl $3, %eax Explanation # Logical shift return x >> 3; Uses logical shift for unsigned For Java users: logical shift written as >>> 45

46 Signed power-of-2 divide w/ shift Quotient of Signed by Power of 2 x >> k gives x / 2 k Uses arithmetic shift Rounds wrong direction when u < 0 k x Operands: 2 k x Division: 2 k. Binary Point Result: RoundDown x 2 k Division Computed Hex Binary y C y >> E y >> FC y >> FF C

47 Correct power-of-2 divide Quotient of negative number by power of 2 We want x 2 k (round toward 0) We compute it as x + 2 k 1 2 k In C: (x + (1<<k)-1) >> k Biases the dividend toward 0 47

48 Correct power-of-2 divide Quotient of negative number by power of 2 We want x 2 k (round toward 0) We compute it as x + 2 k 1 2 k In C: (x + (1<<k)-1) >> k Biases the dividend toward 0 Case 1: No rounding Dividend: Divisor: u 2 k u 2 k k k Binary Point Biasing has no effect 48

49 Correct power-of-2 divide (Cont.) Case 2: Rounding: Dividend: u +2 k 1 k Incremented by 1 Binary Point Divisor: 2 k u 2 k Incremented by 1 Biasing adds 1 to final result 49

50 C function Compiled signed division code int idiv8(int x) { return x/8; } Compiled arithmetic operations testl %eax, %eax js L4 L3: sarl $3, %eax ret L4: addl $7, %eax jmp L3 Explanation if x < 0 x += 7; # Arithmetic shift return x >> 3; Uses arithmetic shift for int For Java users Arith. shift written as >> 50

51 Summary Signed/unsigned multiply: Unsigned divide: Signed divide: x * 2 k = x << k x / 2 k = x >> k Logical shift x / 2 k = x >> k for x > 0 x / 2 k = x + (2 k 1) >> k for x < 0 Arithmetic shift AS

52 3.5: C Integer puzzles Computer Architecture and Systems Programming , Herbstsemester 2016 Timothy Roscoe 52

53 Integer C Puzzles Assume 32-bit word size, two s complement integers For each of the following C expressions, either: Argue that it is true for all argument values Give an example where it is not true Initialization int x = foo(); int y = bar(); unsigned ux = x; unsigned uy = y; x < 0 ((x*2) < 0) ux >= 0 x & 7 == 7 (x<<30) < 0 ux > -1 x > y -x < -y x * x >= 0 x>0 && y>0 x + y > 0 x >= 0 -x <= 0 x <= 0 -x >= 0 (x -x)>>31 == -1 ux >> 3 == ux/8 x >> 3 == x/8 x & (x-1)!= 0 53