Related
When asking about common undefined behavior in C, people sometimes refer to the strict aliasing rule.
What are they talking about?
A typical situation where you encounter strict aliasing problems is when overlaying a struct (like a device/network msg) onto a buffer of the word size of your system (like a pointer to uint32_ts or uint16_ts). When you overlay a struct onto such a buffer, or a buffer onto such a struct through pointer casting you can easily violate strict aliasing rules.
So in this kind of setup, if I want to send a message to something I'd have to have two incompatible pointers pointing to the same chunk of memory. I might then naively code something like this:
typedef struct Msg
{
unsigned int a;
unsigned int b;
} Msg;
void SendWord(uint32_t);
int main(void)
{
// Get a 32-bit buffer from the system
uint32_t* buff = malloc(sizeof(Msg));
// Alias that buffer through message
Msg* msg = (Msg*)(buff);
// Send a bunch of messages
for (int i = 0; i < 10; ++i)
{
msg->a = i;
msg->b = i+1;
SendWord(buff[0]);
SendWord(buff[1]);
}
}
The strict aliasing rule makes this setup illegal: dereferencing a pointer that aliases an object that is not of a compatible type or one of the other types allowed by C 2011 6.5 paragraph 71 is undefined behavior. Unfortunately, you can still code this way, maybe get some warnings, have it compile fine, only to have weird unexpected behavior when you run the code.
(GCC appears somewhat inconsistent in its ability to give aliasing warnings, sometimes giving us a friendly warning and sometimes not.)
To see why this behavior is undefined, we have to think about what the strict aliasing rule buys the compiler. Basically, with this rule, it doesn't have to think about inserting instructions to refresh the contents of buff every run of the loop. Instead, when optimizing, with some annoyingly unenforced assumptions about aliasing, it can omit those instructions, load buff[0] and buff[1] into CPU registers once before the loop is run, and speed up the body of the loop. Before strict aliasing was introduced, the compiler had to live in a state of paranoia that the contents of buff could change by any preceding memory stores. So to get an extra performance edge, and assuming most people don't type-pun pointers, the strict aliasing rule was introduced.
Keep in mind, if you think the example is contrived, this might even happen if you're passing a buffer to another function doing the sending for you, if instead you have.
void SendMessage(uint32_t* buff, size_t size32)
{
for (int i = 0; i < size32; ++i)
{
SendWord(buff[i]);
}
}
And rewrote our earlier loop to take advantage of this convenient function
for (int i = 0; i < 10; ++i)
{
msg->a = i;
msg->b = i+1;
SendMessage(buff, 2);
}
The compiler may or may not be able to or smart enough to try to inline SendMessage and it may or may not decide to load or not load buff again. If SendMessage is part of another API that's compiled separately, it probably has instructions to load buff's contents. Then again, maybe you're in C++ and this is some templated header only implementation that the compiler thinks it can inline. Or maybe it's just something you wrote in your .c file for your own convenience. Anyway undefined behavior might still ensue. Even when we know some of what's happening under the hood, it's still a violation of the rule so no well defined behavior is guaranteed. So just by wrapping in a function that takes our word delimited buffer doesn't necessarily help.
So how do I get around this?
Use a union. Most compilers support this without complaining about strict aliasing. This is allowed in C99 and explicitly allowed in C11.
union {
Msg msg;
unsigned int asBuffer[sizeof(Msg)/sizeof(unsigned int)];
};
You can disable strict aliasing in your compiler (f[no-]strict-aliasing in gcc))
You can use char* for aliasing instead of your system's word. The rules allow an exception for char* (including signed char and unsigned char). It's always assumed that char* aliases other types. However this won't work the other way: there's no assumption that your struct aliases a buffer of chars.
Beginner beware
This is only one potential minefield when overlaying two types onto each other. You should also learn about endianness, word alignment, and how to deal with alignment issues through packing structs correctly.
Footnote
1 The types that C 2011 6.5 7 allows an lvalue to access are:
a type compatible with the effective type of the object,
a qualified version of a type compatible with the effective type of the object,
a type that is the signed or unsigned type corresponding to the effective type of the object,
a type that is the signed or unsigned type corresponding to a qualified version of the effective type of the object,
an aggregate or union type that includes one of the aforementioned types among its members (including, recursively, a member of a subaggregate or contained union), or
a character type.
The best explanation I have found is by Mike Acton, Understanding Strict Aliasing. It's focused a little on PS3 development, but that's basically just GCC.
From the article:
"Strict aliasing is an assumption, made by the C (or C++) compiler, that dereferencing pointers to objects of different types will never refer to the same memory location (i.e. alias each other.)"
So basically if you have an int* pointing to some memory containing an int and then you point a float* to that memory and use it as a float you break the rule. If your code does not respect this, then the compiler's optimizer will most likely break your code.
The exception to the rule is a char*, which is allowed to point to any type.
Note
This is excerpted from my "What is the Strict Aliasing Rule and Why do we care?" write-up.
What is strict aliasing?
In C and C++ aliasing has to do with what expression types we are allowed to access stored values through. In both C and C++ the standard specifies which expression types are allowed to alias which types. The compiler and optimizer are allowed to assume we follow the aliasing rules strictly, hence the term strict aliasing rule. If we attempt to access a value using a type not allowed it is classified as undefined behavior (UB). Once we have undefined behavior all bets are off, the results of our program are no longer reliable.
Unfortunately with strict aliasing violations, we will often obtain the results we expect, leaving the possibility the a future version of a compiler with a new optimization will break code we thought was valid. This is undesirable and it is a worthwhile goal to understand the strict aliasing rules and how to avoid violating them.
To understand more about why we care, we will discuss issues that come up when violating strict aliasing rules, type punning since common techniques used in type punning often violate strict aliasing rules and how to type pun correctly.
Preliminary examples
Let's look at some examples, then we can talk about exactly what the standard(s) say, examine some further examples and then see how to avoid strict aliasing and catch violations we missed. Here is an example that should not be surprising (live example):
int x = 10;
int *ip = &x;
std::cout << *ip << "\n";
*ip = 12;
std::cout << x << "\n";
We have a int* pointing to memory occupied by an int and this is a valid aliasing. The optimizer must assume that assignments through ip could update the value occupied by x.
The next example shows aliasing that leads to undefined behavior (live example):
int foo( float *f, int *i ) {
*i = 1;
*f = 0.f;
return *i;
}
int main() {
int x = 0;
std::cout << x << "\n"; // Expect 0
x = foo(reinterpret_cast<float*>(&x), &x);
std::cout << x << "\n"; // Expect 0?
}
In the function foo we take an int* and a float*, in this example we call foo and set both parameters to point to the same memory location which in this example contains an int. Note, the reinterpret_cast is telling the compiler to treat the expression as if it had the type specified by its template parameter. In this case we are telling it to treat the expression &x as if it had type float*. We may naively expect the result of the second cout to be 0 but with optimization enabled using -O2 both gcc and clang produce the following result:
0
1
Which may not be expected but is perfectly valid since we have invoked undefined behavior. A float can not validly alias an int object. Therefore the optimizer can assume the constant 1 stored when dereferencing i will be the return value since a store through f could not validly affect an int object. Plugging the code in Compiler Explorer shows this is exactly what is happening(live example):
foo(float*, int*): # #foo(float*, int*)
mov dword ptr [rsi], 1
mov dword ptr [rdi], 0
mov eax, 1
ret
The optimizer using Type-Based Alias Analysis (TBAA) assumes 1 will be returned and directly moves the constant value into register eax which carries the return value. TBAA uses the languages rules about what types are allowed to alias to optimize loads and stores. In this case TBAA knows that a float can not alias an int and optimizes away the load of i.
Now, to the Rule-Book
What exactly does the standard say we are allowed and not allowed to do? The standard language is not straightforward, so for each item I will try to provide code examples that demonstrates the meaning.
What does the C11 standard say?
The C11 standard says the following in section 6.5 Expressions paragraph 7:
An object shall have its stored value accessed only by an lvalue expression that has one of the following types:88)
— a type compatible with the effective type of the object,
int x = 1;
int *p = &x;
printf("%d\n", *p); // *p gives us an lvalue expression of type int which is compatible with int
— a qualified version of a type compatible with the effective type of the object,
int x = 1;
const int *p = &x;
printf("%d\n", *p); // *p gives us an lvalue expression of type const int which is compatible with int
— a type that is the signed or unsigned type corresponding to the effective type of the object,
int x = 1;
unsigned int *p = (unsigned int*)&x;
printf("%u\n", *p ); // *p gives us an lvalue expression of type unsigned int which corresponds to
// the effective type of the object
gcc/clang has an extension and also that allows assigning unsigned int* to int* even though they are not compatible types.
— a type that is the signed or unsigned type corresponding to a qualified version of the effective type of the object,
int x = 1;
const unsigned int *p = (const unsigned int*)&x;
printf("%u\n", *p ); // *p gives us an lvalue expression of type const unsigned int which is a unsigned type
// that corresponds with to a qualified version of the effective type of the object
— an aggregate or union type that includes one of the aforementioned types among its members (including, recursively, a member of a subaggregate or contained union), or
struct foo {
int x;
};
void foobar( struct foo *fp, int *ip ); // struct foo is an aggregate that includes int among its members so it
// can alias with *ip
foo f;
foobar( &f, &f.x );
— a character type.
int x = 65;
char *p = (char *)&x;
printf("%c\n", *p ); // *p gives us an lvalue expression of type char which is a character type.
// The results are not portable due to endianness issues.
What the C++17 Draft Standard says
The C++17 draft standard in section [basic.lval] paragraph 11 says:
If a program attempts to access the stored value of an object through a glvalue of other than one of the following types the behavior is undefined:63
(11.1) — the dynamic type of the object,
void *p = malloc( sizeof(int) ); // We have allocated storage but not started the lifetime of an object
int *ip = new (p) int{0}; // Placement new changes the dynamic type of the object to int
std::cout << *ip << "\n"; // *ip gives us a glvalue expression of type int which matches the dynamic type
// of the allocated object
(11.2) — a cv-qualified version of the dynamic type of the object,
int x = 1;
const int *cip = &x;
std::cout << *cip << "\n"; // *cip gives us a glvalue expression of type const int which is a cv-qualified
// version of the dynamic type of x
(11.3) — a type similar (as defined in 7.5) to the dynamic type of the object,
(11.4) — a type that is the signed or unsigned type corresponding to the dynamic type of the object,
// Both si and ui are signed or unsigned types corresponding to each others dynamic types
// We can see from this godbolt(https://godbolt.org/g/KowGXB) the optimizer assumes aliasing.
signed int foo( signed int &si, unsigned int &ui ) {
si = 1;
ui = 2;
return si;
}
(11.5) — a type that is the signed or unsigned type corresponding to a cv-qualified version of the dynamic type of the object,
signed int foo( const signed int &si1, int &si2); // Hard to show this one assumes aliasing
(11.6) — an aggregate or union type that includes one of the aforementioned types among its elements or nonstatic data members (including, recursively, an element or non-static data member of a subaggregate or contained union),
struct foo {
int x;
};
// Compiler Explorer example(https://godbolt.org/g/z2wJTC) shows aliasing assumption
int foobar( foo &fp, int &ip ) {
fp.x = 1;
ip = 2;
return fp.x;
}
foo f;
foobar( f, f.x );
(11.7) — a type that is a (possibly cv-qualified) base class type of the dynamic type of the object,
struct foo { int x; };
struct bar : public foo {};
int foobar( foo &f, bar &b ) {
f.x = 1;
b.x = 2;
return f.x;
}
(11.8) — a char, unsigned char, or std::byte type.
int foo( std::byte &b, uint32_t &ui ) {
b = static_cast<std::byte>('a');
ui = 0xFFFFFFFF;
return std::to_integer<int>( b ); // b gives us a glvalue expression of type std::byte which can alias
// an object of type uint32_t
}
Worth noting signed char is not included in the list above, this is a notable difference from C which says a character type.
What is Type Punning
We have gotten to this point and we may be wondering, why would we want to alias for? The answer typically is to type pun, often the methods used violate strict aliasing rules.
Sometimes we want to circumvent the type system and interpret an object as a different type. This is called type punning, to reinterpret a segment of memory as another type. Type punning is useful for tasks that want access to the underlying representation of an object to view, transport or manipulate. Typical areas we find type punning being used are compilers, serialization, networking code, etc…
Traditionally this has been accomplished by taking the address of the object, casting it to a pointer of the type we want to reinterpret it as and then accessing the value, or in other words by aliasing. For example:
int x = 1;
// In C
float *fp = (float*)&x; // Not a valid aliasing
// In C++
float *fp = reinterpret_cast<float*>(&x); // Not a valid aliasing
printf( "%f\n", *fp );
As we have seen earlier this is not a valid aliasing, so we are invoking undefined behavior. But traditionally compilers did not take advantage of strict aliasing rules and this type of code usually just worked, developers have unfortunately gotten used to doing things this way. A common alternate method for type punning is through unions, which is valid in C but undefined behavior in C++ (see live example):
union u1
{
int n;
float f;
};
union u1 u;
u.f = 1.0f;
printf( "%d\n", u.n ); // UB in C++ n is not the active member
This is not valid in C++ and some consider the purpose of unions to be solely for implementing variant types and feel using unions for type punning is an abuse.
How do we Type Pun correctly?
The standard method for type punning in both C and C++ is memcpy. This may seem a little heavy handed but the optimizer should recognize the use of memcpy for type punning and optimize it away and generate a register to register move. For example if we know int64_t is the same size as double:
static_assert( sizeof( double ) == sizeof( int64_t ) ); // C++17 does not require a message
we can use memcpy:
void func1( double d ) {
std::int64_t n;
std::memcpy(&n, &d, sizeof d);
//...
At a sufficient optimization level any decent modern compiler generates identical code to the previously mentioned reinterpret_cast method or union method for type punning. Examining the generated code we see it uses just register mov (live Compiler Explorer Example).
C++20 and bit_cast
In C++20 we may gain bit_cast (implementation available in link from proposal) which gives a simple and safe way to type-pun as well as being usable in a constexpr context.
The following is an example of how to use bit_cast to type pun a unsigned int to float, (see it live):
std::cout << bit_cast<float>(0x447a0000) << "\n"; //assuming sizeof(float) == sizeof(unsigned int)
In the case where To and From types don't have the same size, it requires us to use an intermediate struct15. We will use a struct containing a sizeof( unsigned int ) character array (assumes 4 byte unsigned int) to be the From type and unsigned int as the To type.:
struct uint_chars {
unsigned char arr[sizeof( unsigned int )] = {}; // Assume sizeof( unsigned int ) == 4
};
// Assume len is a multiple of 4
int bar( unsigned char *p, size_t len ) {
int result = 0;
for( size_t index = 0; index < len; index += sizeof(unsigned int) ) {
uint_chars f;
std::memcpy( f.arr, &p[index], sizeof(unsigned int));
unsigned int result = bit_cast<unsigned int>(f);
result += foo( result );
}
return result;
}
It is unfortunate that we need this intermediate type but that is the current constraint of bit_cast.
Catching Strict Aliasing Violations
We don't have a lot of good tools for catching strict aliasing in C++, the tools we have will catch some cases of strict aliasing violations and some cases of misaligned loads and stores.
gcc using the flag -fstrict-aliasing and -Wstrict-aliasing can catch some cases although not without false positives/negatives. For example the following cases will generate a warning in gcc (see it live):
int a = 1;
short j;
float f = 1.f; // Originally not initialized but tis-kernel caught
// it was being accessed w/ an indeterminate value below
printf("%i\n", j = *(reinterpret_cast<short*>(&a)));
printf("%i\n", j = *(reinterpret_cast<int*>(&f)));
although it will not catch this additional case (see it live):
int *p;
p = &a;
printf("%i\n", j = *(reinterpret_cast<short*>(p)));
Although clang allows these flags it apparently does not actually implement the warnings.
Another tool we have available to us is ASan which can catch misaligned loads and stores. Although these are not directly strict aliasing violations they are a common result of strict aliasing violations. For example the following cases will generate runtime errors when built with clang using -fsanitize=address
int *x = new int[2]; // 8 bytes: [0,7].
int *u = (int*)((char*)x + 6); // regardless of alignment of x this will not be an aligned address
*u = 1; // Access to range [6-9]
printf( "%d\n", *u ); // Access to range [6-9]
The last tool I will recommend is C++ specific and not strictly a tool but a coding practice, don't allow C-style casts. Both gcc and clang will produce a diagnostic for C-style casts using -Wold-style-cast. This will force any undefined type puns to use reinterpret_cast, in general reinterpret_cast should be a flag for closer code review. It is also easier to search your code base for reinterpret_cast to perform an audit.
For C we have all the tools already covered and we also have tis-interpreter, a static analyzer that exhaustively analyzes a program for a large subset of the C language. Given a C version of the earlier example where using -fstrict-aliasing misses one case (see it live)
int a = 1;
short j;
float f = 1.0;
printf("%i\n", j = *((short*)&a));
printf("%i\n", j = *((int*)&f));
int *p;
p = &a;
printf("%i\n", j = *((short*)p));
tis-interpeter is able to catch all three, the following example invokes tis-kernel as tis-interpreter (output is edited for brevity):
./bin/tis-kernel -sa example1.c
...
example1.c:9:[sa] warning: The pointer (short *)(& a) has type short *. It violates strict aliasing
rules by accessing a cell with effective type int.
...
example1.c:10:[sa] warning: The pointer (int *)(& f) has type int *. It violates strict aliasing rules by
accessing a cell with effective type float.
Callstack: main
...
example1.c:15:[sa] warning: The pointer (short *)p has type short *. It violates strict aliasing rules by
accessing a cell with effective type int.
Finally there is TySan which is currently in development. This sanitizer adds type checking information in a shadow memory segment and checks accesses to see if they violate aliasing rules. The tool potentially should be able to catch all aliasing violations but may have a large run-time overhead.
This is the strict aliasing rule, found in section 3.10 of the C++03 standard (other answers provide good explanation, but none provided the rule itself):
If a program attempts to access the stored value of an object through an lvalue of other than one of the following types the behavior is undefined:
the dynamic type of the object,
a cv-qualified version of the dynamic type of the object,
a type that is the signed or unsigned type corresponding to the dynamic type of the object,
a type that is the signed or unsigned type corresponding to a cv-qualified version of the dynamic type of the object,
an aggregate or union type that includes one of the aforementioned types among its members (including, recursively, a member of a subaggregate or contained union),
a type that is a (possibly cv-qualified) base class type of the dynamic type of the object,
a char or unsigned char type.
C++11 and C++14 wording (changes emphasized):
If a program attempts to access the stored value of an object through a glvalue of other than one of the following types the behavior is undefined:
the dynamic type of the object,
a cv-qualified version of the dynamic type of the object,
a type similar (as defined in 4.4) to the dynamic type of the object,
a type that is the signed or unsigned type corresponding to the dynamic type of the object,
a type that is the signed or unsigned type corresponding to a cv-qualified version of the dynamic type of the object,
an aggregate or union type that includes one of the aforementioned types among its elements or non-static data members (including, recursively, an element or non-static data member of a subaggregate or contained union),
a type that is a (possibly cv-qualified) base class type of the dynamic type of the object,
a char or unsigned char type.
Two changes were small: glvalue instead of lvalue, and clarification of the aggregate/union case.
The third change makes a stronger guarantee (relaxes the strong aliasing rule): The new concept of similar types that are now safe to alias.
Also the C wording (C99; ISO/IEC 9899:1999 6.5/7; the exact same wording is used in ISO/IEC 9899:2011 §6.5 ¶7):
An object shall have its stored value accessed only by an lvalue
expression that has one of the following types 73) or 88):
a type compatible with the effective type of the object,
a qualified version of a type compatible with the effective type of
the object,
a type that is the signed or unsigned type corresponding to the
effective type of the object,
a type that is the signed or unsigned type corresponding to a
qualified version of the effective type of the object,
an aggregate or union type that includes one of the aforementioned
types among its members (including, recursively, a member of a
subaggregate or contained union), or
a character type.
73) or 88) The intent of this list is to specify those circumstances in which an object may or may not be aliased.
Strict aliasing doesn't refer only to pointers, it affects references as well, I wrote a paper about it for the boost developer wiki and it was so well received that I turned it into a page on my consulting web site. It explains completely what it is, why it confuses people so much and what to do about it. Strict Aliasing White Paper. In particular it explains why unions are risky behavior for C++, and why using memcpy is the only fix portable across both C and C++. Hope this is helpful.
As addendum to what Doug T. already wrote, here
is a simple test case which probably triggers it with gcc :
check.c
#include <stdio.h>
void check(short *h,long *k)
{
*h=5;
*k=6;
if (*h == 5)
printf("strict aliasing problem\n");
}
int main(void)
{
long k[1];
check((short *)k,k);
return 0;
}
Compile with gcc -O2 -o check check.c .
Usually (with most gcc versions I tried) this outputs "strict aliasing problem", because the compiler assumes that "h" cannot be the same address as "k" in the "check" function. Because of that the compiler optimizes the if (*h == 5) away and always calls the printf.
For those who are interested here is the x64 assembler code, produced by gcc 4.6.3, running on ubuntu 12.04.2 for x64:
movw $5, (%rdi)
movq $6, (%rsi)
movl $.LC0, %edi
jmp puts
So the if condition is completely gone from the assembler code.
According to the C89 rationale, the authors of the Standard did not want to require that compilers given code like:
int x;
int test(double *p)
{
x=5;
*p = 1.0;
return x;
}
should be required to reload the value of x between the assignment and return statement so as to allow for the possibility that p might point to x, and the assignment to *p might consequently alter the value of x. The notion that a compiler should be entitled to presume that there won't be aliasing in situations like the above was non-controversial.
Unfortunately, the authors of the C89 wrote their rule in a way that, if read literally, would make even the following function invoke Undefined Behavior:
void test(void)
{
struct S {int x;} s;
s.x = 1;
}
because it uses an lvalue of type int to access an object of type struct S, and int is not among the types that may be used accessing a struct S. Because it would be absurd to treat all use of non-character-type members of structs and unions as Undefined Behavior, almost everyone recognizes that there are at least some circumstances where an lvalue of one type may be used to access an object of another type. Unfortunately, the C Standards Committee has failed to define what those circumstances are.
Much of the problem is a result of Defect Report #028, which asked about the behavior of a program like:
int test(int *ip, double *dp)
{
*ip = 1;
*dp = 1.23;
return *ip;
}
int test2(void)
{
union U { int i; double d; } u;
return test(&u.i, &u.d);
}
Defect Report #28 states that the program invokes Undefined Behavior because the action of writing a union member of type "double" and reading one of type "int" invokes Implementation-Defined behavior. Such reasoning is nonsensical, but forms the basis for the Effective Type rules which needlessly complicate the language while doing nothing to address the original problem.
The best way to resolve the original problem would probably be to treat the
footnote about the purpose of the rule as though it were normative, and made
the rule unenforceable except in cases which actually involve conflicting accesses using aliases. Given something like:
void inc_int(int *p) { *p = 3; }
int test(void)
{
int *p;
struct S { int x; } s;
s.x = 1;
p = &s.x;
inc_int(p);
return s.x;
}
There's no conflict within inc_int because all accesses to the storage accessed through *p are done with an lvalue of type int, and there's no conflict in test because p is visibly derived from a struct S, and by the next time s is used, all accesses to that storage that will ever be made through p will have already happened.
If the code were changed slightly...
void inc_int(int *p) { *p = 3; }
int test(void)
{
int *p;
struct S { int x; } s;
p = &s.x;
s.x = 1; // !!*!!
*p += 1;
return s.x;
}
Here, there is an aliasing conflict between p and the access to s.x on the marked line because at that point in execution another reference exists that will be used to access the same storage.
Had Defect Report 028 said the original example invoked UB because of the overlap between the creation and use of the two pointers, that would have made things a lot more clear without having to add "Effective Types" or other such complexity.
Type punning via pointer casts (as opposed to using a union) is a major example of breaking strict aliasing.
After reading many of the answers, I feel the need to add something:
Strict aliasing (which I'll describe in a bit) is important because:
Memory access can be expensive (performance wise), which is why data is manipulated in CPU registers before being written back to the physical memory.
If data in two different CPU registers will be written to the same memory space, we can't predict which data will "survive" when we code in C.
In assembly, where we code the loading and unloading of CPU registers manually, we will know which data remains intact. But C (thankfully) abstracts this detail away.
Since two pointers can point to the same location in the memory, this could result in complex code that handles possible collisions.
This extra code is slow and hurts performance since it performs extra memory read / write operations which are both slower and (possibly) unnecessary.
The Strict aliasing rule allows us to avoid redundant machine code in cases in which it should be safe to assume that two pointers don't point to the same memory block (see also the restrict keyword).
The Strict aliasing states it's safe to assume that pointers to different types point to different locations in the memory.
If a compiler notices that two pointers point to different types (for example, an int * and a float *), it will assume the memory address is different and it will not protect against memory address collisions, resulting in faster machine code.
For example:
Lets assume the following function:
void merge_two_ints(int *a, int *b) {
*b += *a;
*a += *b;
}
In order to handle the case in which a == b (both pointers point to the same memory), we need to order and test the way we load data from the memory to the CPU registers, so the code might end up like this:
load a and b from memory.
add a to b.
save b and reload a.
(save from CPU register to the memory and load from the memory to the CPU register).
add b to a.
save a (from the CPU register) to the memory.
Step 3 is very slow because it needs to access the physical memory. However, it's required to protect against instances where a and b point to the same memory address.
Strict aliasing would allow us to prevent this by telling the compiler that these memory addresses are distinctly different (which, in this case, will allow even further optimization which can't be performed if the pointers share a memory address).
This can be told to the compiler in two ways, by using different types to point to. i.e.:
void merge_two_numbers(int *a, long *b) {...}
Using the restrict keyword. i.e.:
void merge_two_ints(int * restrict a, int * restrict b) {...}
Now, by satisfying the Strict Aliasing rule, step 3 can be avoided and the code will run significantly faster.
In fact, by adding the restrict keyword, the whole function could be optimized to:
load a and b from memory.
add a to b.
save result both to a and to b.
This optimization couldn't have been done before, because of the possible collision (where a and b would be tripled instead of doubled).
Strict aliasing is not allowing different pointer types to the same data.
This article should help you understand the issue in full detail.
Technically in C++, the strict aliasing rule is probably never applicable.
Note the definition of indirection (* operator):
The unary * operator performs indirection: the expression to which it
is applied shall be a pointer to an object type, or a pointer to a
function type and the result is an lvalue referring to the object or
function to which the expression points.
Also from the definition of glvalue
A glvalue is an expression whose evaluation determines the identity of
an object, (...snip)
So in any well defined program trace, a glvalue refers to an object. So the so called strict aliasing rule doesn't apply, ever. This may not be what the designers wanted.
Is it legal to cast a pointer to an array of ints to an int pointer?
int arr[4];
int (*a)[4] = &arr;
int *p = (int*)a;
C 2018 6.3.2.3 7 says we can convert an int (*)[4] to an int *:
A pointer to an object type may be converted to a pointer to a different object type. If the resulting pointer is not correctly aligned for the referenced type, the behavior is undefined…
The alignment is necessarily correct since an array of int must have the alignment required for an int.
However, the only thing the C standard says about the value resulting from this conversion is:
… when converted back again, the result shall compare equal to the original pointer.
This means that an int * can temporarily hold the value of an int (*)[4]. If we execute:
int arr[4];
int (*x)[4] = &arr;
int *y = (int *) x;
int (*z)[4] = (int (*)[4]) y;
then we know x == z is true because the standard tells us that. But we do not know what y is. Because the standard permits different types of pointers to have different representations (use the bits that represent their values in different ways), it is possible that y has no useful meaning as an int *. The C standard does not say the converted pointer can be used to access objects.
Most C implementations either support this deliberately or as an artifact of how they are designed. However, in terms of what the C standard specifies, no guarantee is given.
If the original pointer's initialized to either NULL or a valid pointer to an int[4], then yes. Pointer casts must not violate alignment requirements lest you get UB. A cast such as what I've described won't violate such requirement ̶a̶n̶d̶ ̶f̶u̶r̶t̶h̶e̶r̶m̶o̶r̶e̶ ̶i̶t̶ ̶w̶i̶l̶l̶ ̶b̶e̶ ̶u̶s̶a̶b̶l̶e̶ ̶f̶o̶r̶ ̶d̶e̶r̶e̶f̶e̶r̶e̶n̶c̶i̶n̶g̶ ̶b̶e̶c̶a̶u̶s̶e̶ ̶i̶f̶ ̶t̶h̶e̶ ̶̶i̶n̶t̶(̶*̶a̶)̶[̶4̶]̶̶ ̶i̶s̶ ̶v̶a̶l̶i̶d̶ ̶a̶n̶d̶ ̶n̶o̶n̶n̶u̶l̶l̶ ̶t̶h̶e̶n̶ ̶t̶h̶e̶r̶e̶ ̶i̶n̶d̶e̶e̶d̶ ̶i̶s̶ ̶a̶n̶ ̶̶i̶n̶t̶̶ ̶a̶t̶ ̶̶(̶i̶n̶t̶*̶)̶a̶̶.
If you feel uneasy about pointer casts (as you should), you can effect the conversion in this case without casting by simply doing *a (will get int[4] which will decay to int*) or a[0] or &a[0][0] or &(*a)[0]. That way, you can also dereference the result while adhering to the letter of the standard.
When asking about common undefined behavior in C, people sometimes refer to the strict aliasing rule.
What are they talking about?
A typical situation where you encounter strict aliasing problems is when overlaying a struct (like a device/network msg) onto a buffer of the word size of your system (like a pointer to uint32_ts or uint16_ts). When you overlay a struct onto such a buffer, or a buffer onto such a struct through pointer casting you can easily violate strict aliasing rules.
So in this kind of setup, if I want to send a message to something I'd have to have two incompatible pointers pointing to the same chunk of memory. I might then naively code something like this:
typedef struct Msg
{
unsigned int a;
unsigned int b;
} Msg;
void SendWord(uint32_t);
int main(void)
{
// Get a 32-bit buffer from the system
uint32_t* buff = malloc(sizeof(Msg));
// Alias that buffer through message
Msg* msg = (Msg*)(buff);
// Send a bunch of messages
for (int i = 0; i < 10; ++i)
{
msg->a = i;
msg->b = i+1;
SendWord(buff[0]);
SendWord(buff[1]);
}
}
The strict aliasing rule makes this setup illegal: dereferencing a pointer that aliases an object that is not of a compatible type or one of the other types allowed by C 2011 6.5 paragraph 71 is undefined behavior. Unfortunately, you can still code this way, maybe get some warnings, have it compile fine, only to have weird unexpected behavior when you run the code.
(GCC appears somewhat inconsistent in its ability to give aliasing warnings, sometimes giving us a friendly warning and sometimes not.)
To see why this behavior is undefined, we have to think about what the strict aliasing rule buys the compiler. Basically, with this rule, it doesn't have to think about inserting instructions to refresh the contents of buff every run of the loop. Instead, when optimizing, with some annoyingly unenforced assumptions about aliasing, it can omit those instructions, load buff[0] and buff[1] into CPU registers once before the loop is run, and speed up the body of the loop. Before strict aliasing was introduced, the compiler had to live in a state of paranoia that the contents of buff could change by any preceding memory stores. So to get an extra performance edge, and assuming most people don't type-pun pointers, the strict aliasing rule was introduced.
Keep in mind, if you think the example is contrived, this might even happen if you're passing a buffer to another function doing the sending for you, if instead you have.
void SendMessage(uint32_t* buff, size_t size32)
{
for (int i = 0; i < size32; ++i)
{
SendWord(buff[i]);
}
}
And rewrote our earlier loop to take advantage of this convenient function
for (int i = 0; i < 10; ++i)
{
msg->a = i;
msg->b = i+1;
SendMessage(buff, 2);
}
The compiler may or may not be able to or smart enough to try to inline SendMessage and it may or may not decide to load or not load buff again. If SendMessage is part of another API that's compiled separately, it probably has instructions to load buff's contents. Then again, maybe you're in C++ and this is some templated header only implementation that the compiler thinks it can inline. Or maybe it's just something you wrote in your .c file for your own convenience. Anyway undefined behavior might still ensue. Even when we know some of what's happening under the hood, it's still a violation of the rule so no well defined behavior is guaranteed. So just by wrapping in a function that takes our word delimited buffer doesn't necessarily help.
So how do I get around this?
Use a union. Most compilers support this without complaining about strict aliasing. This is allowed in C99 and explicitly allowed in C11.
union {
Msg msg;
unsigned int asBuffer[sizeof(Msg)/sizeof(unsigned int)];
};
You can disable strict aliasing in your compiler (f[no-]strict-aliasing in gcc))
You can use char* for aliasing instead of your system's word. The rules allow an exception for char* (including signed char and unsigned char). It's always assumed that char* aliases other types. However this won't work the other way: there's no assumption that your struct aliases a buffer of chars.
Beginner beware
This is only one potential minefield when overlaying two types onto each other. You should also learn about endianness, word alignment, and how to deal with alignment issues through packing structs correctly.
Footnote
1 The types that C 2011 6.5 7 allows an lvalue to access are:
a type compatible with the effective type of the object,
a qualified version of a type compatible with the effective type of the object,
a type that is the signed or unsigned type corresponding to the effective type of the object,
a type that is the signed or unsigned type corresponding to a qualified version of the effective type of the object,
an aggregate or union type that includes one of the aforementioned types among its members (including, recursively, a member of a subaggregate or contained union), or
a character type.
The best explanation I have found is by Mike Acton, Understanding Strict Aliasing. It's focused a little on PS3 development, but that's basically just GCC.
From the article:
"Strict aliasing is an assumption, made by the C (or C++) compiler, that dereferencing pointers to objects of different types will never refer to the same memory location (i.e. alias each other.)"
So basically if you have an int* pointing to some memory containing an int and then you point a float* to that memory and use it as a float you break the rule. If your code does not respect this, then the compiler's optimizer will most likely break your code.
The exception to the rule is a char*, which is allowed to point to any type.
Note
This is excerpted from my "What is the Strict Aliasing Rule and Why do we care?" write-up.
What is strict aliasing?
In C and C++ aliasing has to do with what expression types we are allowed to access stored values through. In both C and C++ the standard specifies which expression types are allowed to alias which types. The compiler and optimizer are allowed to assume we follow the aliasing rules strictly, hence the term strict aliasing rule. If we attempt to access a value using a type not allowed it is classified as undefined behavior (UB). Once we have undefined behavior all bets are off, the results of our program are no longer reliable.
Unfortunately with strict aliasing violations, we will often obtain the results we expect, leaving the possibility the a future version of a compiler with a new optimization will break code we thought was valid. This is undesirable and it is a worthwhile goal to understand the strict aliasing rules and how to avoid violating them.
To understand more about why we care, we will discuss issues that come up when violating strict aliasing rules, type punning since common techniques used in type punning often violate strict aliasing rules and how to type pun correctly.
Preliminary examples
Let's look at some examples, then we can talk about exactly what the standard(s) say, examine some further examples and then see how to avoid strict aliasing and catch violations we missed. Here is an example that should not be surprising (live example):
int x = 10;
int *ip = &x;
std::cout << *ip << "\n";
*ip = 12;
std::cout << x << "\n";
We have a int* pointing to memory occupied by an int and this is a valid aliasing. The optimizer must assume that assignments through ip could update the value occupied by x.
The next example shows aliasing that leads to undefined behavior (live example):
int foo( float *f, int *i ) {
*i = 1;
*f = 0.f;
return *i;
}
int main() {
int x = 0;
std::cout << x << "\n"; // Expect 0
x = foo(reinterpret_cast<float*>(&x), &x);
std::cout << x << "\n"; // Expect 0?
}
In the function foo we take an int* and a float*, in this example we call foo and set both parameters to point to the same memory location which in this example contains an int. Note, the reinterpret_cast is telling the compiler to treat the expression as if it had the type specified by its template parameter. In this case we are telling it to treat the expression &x as if it had type float*. We may naively expect the result of the second cout to be 0 but with optimization enabled using -O2 both gcc and clang produce the following result:
0
1
Which may not be expected but is perfectly valid since we have invoked undefined behavior. A float can not validly alias an int object. Therefore the optimizer can assume the constant 1 stored when dereferencing i will be the return value since a store through f could not validly affect an int object. Plugging the code in Compiler Explorer shows this is exactly what is happening(live example):
foo(float*, int*): # #foo(float*, int*)
mov dword ptr [rsi], 1
mov dword ptr [rdi], 0
mov eax, 1
ret
The optimizer using Type-Based Alias Analysis (TBAA) assumes 1 will be returned and directly moves the constant value into register eax which carries the return value. TBAA uses the languages rules about what types are allowed to alias to optimize loads and stores. In this case TBAA knows that a float can not alias an int and optimizes away the load of i.
Now, to the Rule-Book
What exactly does the standard say we are allowed and not allowed to do? The standard language is not straightforward, so for each item I will try to provide code examples that demonstrates the meaning.
What does the C11 standard say?
The C11 standard says the following in section 6.5 Expressions paragraph 7:
An object shall have its stored value accessed only by an lvalue expression that has one of the following types:88)
— a type compatible with the effective type of the object,
int x = 1;
int *p = &x;
printf("%d\n", *p); // *p gives us an lvalue expression of type int which is compatible with int
— a qualified version of a type compatible with the effective type of the object,
int x = 1;
const int *p = &x;
printf("%d\n", *p); // *p gives us an lvalue expression of type const int which is compatible with int
— a type that is the signed or unsigned type corresponding to the effective type of the object,
int x = 1;
unsigned int *p = (unsigned int*)&x;
printf("%u\n", *p ); // *p gives us an lvalue expression of type unsigned int which corresponds to
// the effective type of the object
gcc/clang has an extension and also that allows assigning unsigned int* to int* even though they are not compatible types.
— a type that is the signed or unsigned type corresponding to a qualified version of the effective type of the object,
int x = 1;
const unsigned int *p = (const unsigned int*)&x;
printf("%u\n", *p ); // *p gives us an lvalue expression of type const unsigned int which is a unsigned type
// that corresponds with to a qualified version of the effective type of the object
— an aggregate or union type that includes one of the aforementioned types among its members (including, recursively, a member of a subaggregate or contained union), or
struct foo {
int x;
};
void foobar( struct foo *fp, int *ip ); // struct foo is an aggregate that includes int among its members so it
// can alias with *ip
foo f;
foobar( &f, &f.x );
— a character type.
int x = 65;
char *p = (char *)&x;
printf("%c\n", *p ); // *p gives us an lvalue expression of type char which is a character type.
// The results are not portable due to endianness issues.
What the C++17 Draft Standard says
The C++17 draft standard in section [basic.lval] paragraph 11 says:
If a program attempts to access the stored value of an object through a glvalue of other than one of the following types the behavior is undefined:63
(11.1) — the dynamic type of the object,
void *p = malloc( sizeof(int) ); // We have allocated storage but not started the lifetime of an object
int *ip = new (p) int{0}; // Placement new changes the dynamic type of the object to int
std::cout << *ip << "\n"; // *ip gives us a glvalue expression of type int which matches the dynamic type
// of the allocated object
(11.2) — a cv-qualified version of the dynamic type of the object,
int x = 1;
const int *cip = &x;
std::cout << *cip << "\n"; // *cip gives us a glvalue expression of type const int which is a cv-qualified
// version of the dynamic type of x
(11.3) — a type similar (as defined in 7.5) to the dynamic type of the object,
(11.4) — a type that is the signed or unsigned type corresponding to the dynamic type of the object,
// Both si and ui are signed or unsigned types corresponding to each others dynamic types
// We can see from this godbolt(https://godbolt.org/g/KowGXB) the optimizer assumes aliasing.
signed int foo( signed int &si, unsigned int &ui ) {
si = 1;
ui = 2;
return si;
}
(11.5) — a type that is the signed or unsigned type corresponding to a cv-qualified version of the dynamic type of the object,
signed int foo( const signed int &si1, int &si2); // Hard to show this one assumes aliasing
(11.6) — an aggregate or union type that includes one of the aforementioned types among its elements or nonstatic data members (including, recursively, an element or non-static data member of a subaggregate or contained union),
struct foo {
int x;
};
// Compiler Explorer example(https://godbolt.org/g/z2wJTC) shows aliasing assumption
int foobar( foo &fp, int &ip ) {
fp.x = 1;
ip = 2;
return fp.x;
}
foo f;
foobar( f, f.x );
(11.7) — a type that is a (possibly cv-qualified) base class type of the dynamic type of the object,
struct foo { int x; };
struct bar : public foo {};
int foobar( foo &f, bar &b ) {
f.x = 1;
b.x = 2;
return f.x;
}
(11.8) — a char, unsigned char, or std::byte type.
int foo( std::byte &b, uint32_t &ui ) {
b = static_cast<std::byte>('a');
ui = 0xFFFFFFFF;
return std::to_integer<int>( b ); // b gives us a glvalue expression of type std::byte which can alias
// an object of type uint32_t
}
Worth noting signed char is not included in the list above, this is a notable difference from C which says a character type.
What is Type Punning
We have gotten to this point and we may be wondering, why would we want to alias for? The answer typically is to type pun, often the methods used violate strict aliasing rules.
Sometimes we want to circumvent the type system and interpret an object as a different type. This is called type punning, to reinterpret a segment of memory as another type. Type punning is useful for tasks that want access to the underlying representation of an object to view, transport or manipulate. Typical areas we find type punning being used are compilers, serialization, networking code, etc…
Traditionally this has been accomplished by taking the address of the object, casting it to a pointer of the type we want to reinterpret it as and then accessing the value, or in other words by aliasing. For example:
int x = 1;
// In C
float *fp = (float*)&x; // Not a valid aliasing
// In C++
float *fp = reinterpret_cast<float*>(&x); // Not a valid aliasing
printf( "%f\n", *fp );
As we have seen earlier this is not a valid aliasing, so we are invoking undefined behavior. But traditionally compilers did not take advantage of strict aliasing rules and this type of code usually just worked, developers have unfortunately gotten used to doing things this way. A common alternate method for type punning is through unions, which is valid in C but undefined behavior in C++ (see live example):
union u1
{
int n;
float f;
};
union u1 u;
u.f = 1.0f;
printf( "%d\n", u.n ); // UB in C++ n is not the active member
This is not valid in C++ and some consider the purpose of unions to be solely for implementing variant types and feel using unions for type punning is an abuse.
How do we Type Pun correctly?
The standard method for type punning in both C and C++ is memcpy. This may seem a little heavy handed but the optimizer should recognize the use of memcpy for type punning and optimize it away and generate a register to register move. For example if we know int64_t is the same size as double:
static_assert( sizeof( double ) == sizeof( int64_t ) ); // C++17 does not require a message
we can use memcpy:
void func1( double d ) {
std::int64_t n;
std::memcpy(&n, &d, sizeof d);
//...
At a sufficient optimization level any decent modern compiler generates identical code to the previously mentioned reinterpret_cast method or union method for type punning. Examining the generated code we see it uses just register mov (live Compiler Explorer Example).
C++20 and bit_cast
In C++20 we may gain bit_cast (implementation available in link from proposal) which gives a simple and safe way to type-pun as well as being usable in a constexpr context.
The following is an example of how to use bit_cast to type pun a unsigned int to float, (see it live):
std::cout << bit_cast<float>(0x447a0000) << "\n"; //assuming sizeof(float) == sizeof(unsigned int)
In the case where To and From types don't have the same size, it requires us to use an intermediate struct15. We will use a struct containing a sizeof( unsigned int ) character array (assumes 4 byte unsigned int) to be the From type and unsigned int as the To type.:
struct uint_chars {
unsigned char arr[sizeof( unsigned int )] = {}; // Assume sizeof( unsigned int ) == 4
};
// Assume len is a multiple of 4
int bar( unsigned char *p, size_t len ) {
int result = 0;
for( size_t index = 0; index < len; index += sizeof(unsigned int) ) {
uint_chars f;
std::memcpy( f.arr, &p[index], sizeof(unsigned int));
unsigned int result = bit_cast<unsigned int>(f);
result += foo( result );
}
return result;
}
It is unfortunate that we need this intermediate type but that is the current constraint of bit_cast.
Catching Strict Aliasing Violations
We don't have a lot of good tools for catching strict aliasing in C++, the tools we have will catch some cases of strict aliasing violations and some cases of misaligned loads and stores.
gcc using the flag -fstrict-aliasing and -Wstrict-aliasing can catch some cases although not without false positives/negatives. For example the following cases will generate a warning in gcc (see it live):
int a = 1;
short j;
float f = 1.f; // Originally not initialized but tis-kernel caught
// it was being accessed w/ an indeterminate value below
printf("%i\n", j = *(reinterpret_cast<short*>(&a)));
printf("%i\n", j = *(reinterpret_cast<int*>(&f)));
although it will not catch this additional case (see it live):
int *p;
p = &a;
printf("%i\n", j = *(reinterpret_cast<short*>(p)));
Although clang allows these flags it apparently does not actually implement the warnings.
Another tool we have available to us is ASan which can catch misaligned loads and stores. Although these are not directly strict aliasing violations they are a common result of strict aliasing violations. For example the following cases will generate runtime errors when built with clang using -fsanitize=address
int *x = new int[2]; // 8 bytes: [0,7].
int *u = (int*)((char*)x + 6); // regardless of alignment of x this will not be an aligned address
*u = 1; // Access to range [6-9]
printf( "%d\n", *u ); // Access to range [6-9]
The last tool I will recommend is C++ specific and not strictly a tool but a coding practice, don't allow C-style casts. Both gcc and clang will produce a diagnostic for C-style casts using -Wold-style-cast. This will force any undefined type puns to use reinterpret_cast, in general reinterpret_cast should be a flag for closer code review. It is also easier to search your code base for reinterpret_cast to perform an audit.
For C we have all the tools already covered and we also have tis-interpreter, a static analyzer that exhaustively analyzes a program for a large subset of the C language. Given a C version of the earlier example where using -fstrict-aliasing misses one case (see it live)
int a = 1;
short j;
float f = 1.0;
printf("%i\n", j = *((short*)&a));
printf("%i\n", j = *((int*)&f));
int *p;
p = &a;
printf("%i\n", j = *((short*)p));
tis-interpeter is able to catch all three, the following example invokes tis-kernel as tis-interpreter (output is edited for brevity):
./bin/tis-kernel -sa example1.c
...
example1.c:9:[sa] warning: The pointer (short *)(& a) has type short *. It violates strict aliasing
rules by accessing a cell with effective type int.
...
example1.c:10:[sa] warning: The pointer (int *)(& f) has type int *. It violates strict aliasing rules by
accessing a cell with effective type float.
Callstack: main
...
example1.c:15:[sa] warning: The pointer (short *)p has type short *. It violates strict aliasing rules by
accessing a cell with effective type int.
Finally there is TySan which is currently in development. This sanitizer adds type checking information in a shadow memory segment and checks accesses to see if they violate aliasing rules. The tool potentially should be able to catch all aliasing violations but may have a large run-time overhead.
This is the strict aliasing rule, found in section 3.10 of the C++03 standard (other answers provide good explanation, but none provided the rule itself):
If a program attempts to access the stored value of an object through an lvalue of other than one of the following types the behavior is undefined:
the dynamic type of the object,
a cv-qualified version of the dynamic type of the object,
a type that is the signed or unsigned type corresponding to the dynamic type of the object,
a type that is the signed or unsigned type corresponding to a cv-qualified version of the dynamic type of the object,
an aggregate or union type that includes one of the aforementioned types among its members (including, recursively, a member of a subaggregate or contained union),
a type that is a (possibly cv-qualified) base class type of the dynamic type of the object,
a char or unsigned char type.
C++11 and C++14 wording (changes emphasized):
If a program attempts to access the stored value of an object through a glvalue of other than one of the following types the behavior is undefined:
the dynamic type of the object,
a cv-qualified version of the dynamic type of the object,
a type similar (as defined in 4.4) to the dynamic type of the object,
a type that is the signed or unsigned type corresponding to the dynamic type of the object,
a type that is the signed or unsigned type corresponding to a cv-qualified version of the dynamic type of the object,
an aggregate or union type that includes one of the aforementioned types among its elements or non-static data members (including, recursively, an element or non-static data member of a subaggregate or contained union),
a type that is a (possibly cv-qualified) base class type of the dynamic type of the object,
a char or unsigned char type.
Two changes were small: glvalue instead of lvalue, and clarification of the aggregate/union case.
The third change makes a stronger guarantee (relaxes the strong aliasing rule): The new concept of similar types that are now safe to alias.
Also the C wording (C99; ISO/IEC 9899:1999 6.5/7; the exact same wording is used in ISO/IEC 9899:2011 §6.5 ¶7):
An object shall have its stored value accessed only by an lvalue
expression that has one of the following types 73) or 88):
a type compatible with the effective type of the object,
a qualified version of a type compatible with the effective type of
the object,
a type that is the signed or unsigned type corresponding to the
effective type of the object,
a type that is the signed or unsigned type corresponding to a
qualified version of the effective type of the object,
an aggregate or union type that includes one of the aforementioned
types among its members (including, recursively, a member of a
subaggregate or contained union), or
a character type.
73) or 88) The intent of this list is to specify those circumstances in which an object may or may not be aliased.
Strict aliasing doesn't refer only to pointers, it affects references as well, I wrote a paper about it for the boost developer wiki and it was so well received that I turned it into a page on my consulting web site. It explains completely what it is, why it confuses people so much and what to do about it. Strict Aliasing White Paper. In particular it explains why unions are risky behavior for C++, and why using memcpy is the only fix portable across both C and C++. Hope this is helpful.
As addendum to what Doug T. already wrote, here
is a simple test case which probably triggers it with gcc :
check.c
#include <stdio.h>
void check(short *h,long *k)
{
*h=5;
*k=6;
if (*h == 5)
printf("strict aliasing problem\n");
}
int main(void)
{
long k[1];
check((short *)k,k);
return 0;
}
Compile with gcc -O2 -o check check.c .
Usually (with most gcc versions I tried) this outputs "strict aliasing problem", because the compiler assumes that "h" cannot be the same address as "k" in the "check" function. Because of that the compiler optimizes the if (*h == 5) away and always calls the printf.
For those who are interested here is the x64 assembler code, produced by gcc 4.6.3, running on ubuntu 12.04.2 for x64:
movw $5, (%rdi)
movq $6, (%rsi)
movl $.LC0, %edi
jmp puts
So the if condition is completely gone from the assembler code.
According to the C89 rationale, the authors of the Standard did not want to require that compilers given code like:
int x;
int test(double *p)
{
x=5;
*p = 1.0;
return x;
}
should be required to reload the value of x between the assignment and return statement so as to allow for the possibility that p might point to x, and the assignment to *p might consequently alter the value of x. The notion that a compiler should be entitled to presume that there won't be aliasing in situations like the above was non-controversial.
Unfortunately, the authors of the C89 wrote their rule in a way that, if read literally, would make even the following function invoke Undefined Behavior:
void test(void)
{
struct S {int x;} s;
s.x = 1;
}
because it uses an lvalue of type int to access an object of type struct S, and int is not among the types that may be used accessing a struct S. Because it would be absurd to treat all use of non-character-type members of structs and unions as Undefined Behavior, almost everyone recognizes that there are at least some circumstances where an lvalue of one type may be used to access an object of another type. Unfortunately, the C Standards Committee has failed to define what those circumstances are.
Much of the problem is a result of Defect Report #028, which asked about the behavior of a program like:
int test(int *ip, double *dp)
{
*ip = 1;
*dp = 1.23;
return *ip;
}
int test2(void)
{
union U { int i; double d; } u;
return test(&u.i, &u.d);
}
Defect Report #28 states that the program invokes Undefined Behavior because the action of writing a union member of type "double" and reading one of type "int" invokes Implementation-Defined behavior. Such reasoning is nonsensical, but forms the basis for the Effective Type rules which needlessly complicate the language while doing nothing to address the original problem.
The best way to resolve the original problem would probably be to treat the
footnote about the purpose of the rule as though it were normative, and made
the rule unenforceable except in cases which actually involve conflicting accesses using aliases. Given something like:
void inc_int(int *p) { *p = 3; }
int test(void)
{
int *p;
struct S { int x; } s;
s.x = 1;
p = &s.x;
inc_int(p);
return s.x;
}
There's no conflict within inc_int because all accesses to the storage accessed through *p are done with an lvalue of type int, and there's no conflict in test because p is visibly derived from a struct S, and by the next time s is used, all accesses to that storage that will ever be made through p will have already happened.
If the code were changed slightly...
void inc_int(int *p) { *p = 3; }
int test(void)
{
int *p;
struct S { int x; } s;
p = &s.x;
s.x = 1; // !!*!!
*p += 1;
return s.x;
}
Here, there is an aliasing conflict between p and the access to s.x on the marked line because at that point in execution another reference exists that will be used to access the same storage.
Had Defect Report 028 said the original example invoked UB because of the overlap between the creation and use of the two pointers, that would have made things a lot more clear without having to add "Effective Types" or other such complexity.
Type punning via pointer casts (as opposed to using a union) is a major example of breaking strict aliasing.
After reading many of the answers, I feel the need to add something:
Strict aliasing (which I'll describe in a bit) is important because:
Memory access can be expensive (performance wise), which is why data is manipulated in CPU registers before being written back to the physical memory.
If data in two different CPU registers will be written to the same memory space, we can't predict which data will "survive" when we code in C.
In assembly, where we code the loading and unloading of CPU registers manually, we will know which data remains intact. But C (thankfully) abstracts this detail away.
Since two pointers can point to the same location in the memory, this could result in complex code that handles possible collisions.
This extra code is slow and hurts performance since it performs extra memory read / write operations which are both slower and (possibly) unnecessary.
The Strict aliasing rule allows us to avoid redundant machine code in cases in which it should be safe to assume that two pointers don't point to the same memory block (see also the restrict keyword).
The Strict aliasing states it's safe to assume that pointers to different types point to different locations in the memory.
If a compiler notices that two pointers point to different types (for example, an int * and a float *), it will assume the memory address is different and it will not protect against memory address collisions, resulting in faster machine code.
For example:
Lets assume the following function:
void merge_two_ints(int *a, int *b) {
*b += *a;
*a += *b;
}
In order to handle the case in which a == b (both pointers point to the same memory), we need to order and test the way we load data from the memory to the CPU registers, so the code might end up like this:
load a and b from memory.
add a to b.
save b and reload a.
(save from CPU register to the memory and load from the memory to the CPU register).
add b to a.
save a (from the CPU register) to the memory.
Step 3 is very slow because it needs to access the physical memory. However, it's required to protect against instances where a and b point to the same memory address.
Strict aliasing would allow us to prevent this by telling the compiler that these memory addresses are distinctly different (which, in this case, will allow even further optimization which can't be performed if the pointers share a memory address).
This can be told to the compiler in two ways, by using different types to point to. i.e.:
void merge_two_numbers(int *a, long *b) {...}
Using the restrict keyword. i.e.:
void merge_two_ints(int * restrict a, int * restrict b) {...}
Now, by satisfying the Strict Aliasing rule, step 3 can be avoided and the code will run significantly faster.
In fact, by adding the restrict keyword, the whole function could be optimized to:
load a and b from memory.
add a to b.
save result both to a and to b.
This optimization couldn't have been done before, because of the possible collision (where a and b would be tripled instead of doubled).
Strict aliasing is not allowing different pointer types to the same data.
This article should help you understand the issue in full detail.
Technically in C++, the strict aliasing rule is probably never applicable.
Note the definition of indirection (* operator):
The unary * operator performs indirection: the expression to which it
is applied shall be a pointer to an object type, or a pointer to a
function type and the result is an lvalue referring to the object or
function to which the expression points.
Also from the definition of glvalue
A glvalue is an expression whose evaluation determines the identity of
an object, (...snip)
So in any well defined program trace, a glvalue refers to an object. So the so called strict aliasing rule doesn't apply, ever. This may not be what the designers wanted.
Let's say that I have some code that looks like this:
void foo(void* obj, int type)
{
TypeA* a = (TypeA*)obj;
TypeB* b = (TypeB*)obj;
TypeC* c = (TypeC*)obj;
if( type == 1 ) {
// code that uses "a"
}
else if( type == 2 ) {
// code that uses "b"
}
else if( type == 3 ) {
// code that uses "c"
}
}
I'm curious to know whether or not this function breaks strict aliasing a
priori, or if the validity of the function depends on how it's actually used.
A certain part of me wants to hope that things will work out fine if I only
ever used the function as foo(&objectOfTypeA, 1), foo(&objectOfTypeB, 2),
or foo(&objectOfTypeC, 3). Yet I've only know about aliasing errors for a
few days, and so I worry that the function foo() is broken right from the
start simply because it casts the incoming void* to various different types to
begin with.
This does not break strict aliasing, because you only access *obj through the correct type. Casting obj does not result in access to *obj.
The strict aliasing rule
Strict aliasing only applies to accessing an object, not to pointer manipulations.
You can manipulate the pointer all you want.
From n1570 §6.5 ¶7
An object shall have its stored value accessed only by an lvalue expression that has one of the following types...
Since you're not accessing *obj but just obj, the strict aliasing rule is not violated.
Conversions
The second question here is whether the conversions are permissible. The answer is yes, because you can convert a void * to any other object pointer type. The void type is special, it is the only type with this property.
int *i;
void *ptr = &i;
float *f = ptr; // Permissible
The standard is silent on what the value of f is. In other words, while the conversion is permissible, there is nothing useful you can do with f. You can't even convert it back to void *.
This is explained in n1570 §6.3.2.3 ¶1:
A pointer to void may be converted to or from a pointer to any object type. A pointer to any object type may be converted to a pointer to void and back again, the result shall compare equal to the original pointer.
So, to repeat, the conversion is always permissible but not useful in any way.
First, to clarify, I am not talking about dereferencing invalid pointers!
Consider the following two examples.
Example 1
typedef struct { int *p; } T;
T a = { malloc(sizeof(int) };
free(a.p); // a.p is now indeterminate?
T b = a; // Access through a non-character type?
Example 2
void foo(int *p) {}
int *p = malloc(sizeof(int));
free(p); // p is now indeterminate?
foo(p); // Access through a non-character type?
Question
Do either of the above examples invoke undefined behaviour?
Context
This question is posed in response to this discussion. The suggestion was that, for example, pointer arguments may be passed to a function via x86 segment registers, which could cause a hardware exception.
From the C99 standard, we learn the following (emphasis mine):
[3.17] indeterminate value - either an unspecified value or a trap representation
and then:
[6.2.4 p2] The value of a pointer becomes indeterminate when
the object it points to reaches the end of its lifetime.
and then:
[6.2.6.1 p5] Certain object representations need not represent a value of the object type. If the stored value of an object has such a representation and is read by an lvalue expression that does not have character type, the behavior is undefined. If such a representation is produced by a side effect that modifies all or any part of the object by an lvalue expression that does not have character type, the behavior is undefined. Such a representation is called a trap representation.
Taking all of this together, what restrictions do we have on accessing pointers to "dead" objects?
Addendum
Whilst I've quoted the C99 standard above, I'd be interested to know if the behaviour differs in any of the C++ standards.
Example 2 is invalid. The analysis in your question is correct.
Example 1 is valid. A structure type never holds a trap representation, even if one of its members does. This means that structure assignment, on a system where trap representations would cause problems, must be implemented as a bytewise copy, rather than a member-by-member copy.
6.2.6 Representations of types
6.2.6.1 General
6 [...] The value of a structure or union object is never a t rap
representation, even though the value of a member of the structure or union object may be
a trap representation.
My interpretation is that while only non-character types can have trap representations, any type can have indeterminate value, and that accessing an object with indeterminate value in any way invokes undefined behavior. The most infamous example might be OpenSSL's invalid use of uninitialized objects as a random seed.
So, the answer to your question would be: never.
By the way, an interesting consequence of not just the pointed-to object but the pointer itself being indeterminate after free or realloc is that this idiom invokes undefined behavior:
void *tmp = realloc(ptr, newsize);
if (tmp != ptr) {
/* ... */
}
C++ discussion
Short answer: In C++, there is no such thing as accessing "reading" a class instance; you can only "read" non-class object, and this is done by a lvalue-to-rvalue conversion.
Detailed answer:
typedef struct { int *p; } T;
T designates an unnamed class. For the sake of the discussion let's name this class T:
struct T {
int *p;
};
Because you did not declare a copy constructor, the compiler implicitly declares one, so the class definition reads:
struct T {
int *p;
T (const T&);
};
So we have:
T a;
T b = a; // Access through a non-character type?
Yes, indeed; this is initialization by copy constructor, so the copy constructor definition will be generated by the compiler; the definition is equivalent with
inline T::T (const T& rhs)
: p(rhs.p) {
}
So you are accessing the value as a pointer, not a bunch of bytes.
If the pointer value is invalid (not initialized, freed), the behavior is not defined.