Consider a C function (with external linkage) like the following one:
void f(void **p)
{
/* do something with *p */
}
Now assume that f is being called in a way such that p points to the return address of f on the stack, as in the following code (assuming the System V AMD64 ABI):
leaq -8(%rsp), %rdi
callq f
What may happen is that the code of f modifies the return address on the stack by assigning a value to *p. Thus the compiler will have to treat the return address on the stack as a volatile value. How can I tell the compiler, gcc in my case, that the return address is volatile?
Otherwise, the compiler could, at least in principle, generate the following code for f:
pushq %rbp
movq 8(%rsp), %r10
pushq %r10
## do something with (%rdi)
popq %r10
popq %rbp
addq 8,%rsp
jmpq *%r10
Admittedly, it is unlikely that a compiler would ever generate code like this but it does not seem to be forbidden without any further function attributes. And this code wouldn't notice if the return address on the stack is being modified in the middle of the function because the original return address is already retrieved at the beginning of the function.
P.S.: As has been suggested by Peter Cordes, I should better explain the purpose of my question: It is about garbage collecting dynamically generated machine code using a moving garbage collector: The function f stands for the garbage collector. The callee of f may be a function whose code is being moved around while f is running, so I came up with the idea of letting f know the return address so that f may modify it accordingly to whether the memory area the return address points to has been moved around or not.
Using the SysV ABI (Linux, FreeBSD, Solaris, Mac OS X / macOS) on AMD64/x86-64, you only need a trivial assembly function wrapped around the actual garbage collector function.
The following f.s defines void f(void *), and calls the real GC, real_f(void *, void **), with the added second parameter pointing to the return address.
.file "f.s"
.text
.p2align 4,,15
.globl f
.type f, #function
f:
movq %rsp, %rsi
call real_f
ret
.size f, .-f
If real_f() already has two other parameters, use %rdx (for the third) instead of %rsi. If three to five, use %rcx, %r8, or %r9, respectively. SysV ABI on AMD64/x86-64 only supports up to six non-floating-point parameters in registers.
Let's test the above with a small example.c:
#include <stdlib.h>
#include <stdio.h>
extern void f(void *);
void real_f(void *arg, void **retval)
{
printf("real_f(): Returning to %p instead of %p.\n", arg, *retval);
*retval = arg;
}
int main(void)
{
printf("Function and label addresses:\n");
printf("%p f()\n", f);
printf("%p real_f()\n", real_f);
printf("%p one_call:\n", &&one_call);
printf("%p one_fail:\n", &&one_fail);
printf("%p one_skip:\n", &&one_skip);
printf("\n");
printf("f(one_skip):\n");
fflush(stdout);
one_call:
f(&&one_skip);
one_fail:
printf("At one_fail.\n");
fflush(stdout);
one_skip:
printf("At one_skip.\n");
fflush(stdout);
return EXIT_SUCCESS;
}
Note that the above relies on both GCC behaviour (&& providing the address of a label) as well as GCC behaviour on AMD64/x86-64 architecture (object and function pointers being interchangeable), as well as the C compiler not making any of the myriad optimizations they are allowed to do to the code in main().
(It does not matter if real_f() is optimized; it's just that I was too lazy to work out a better example in main(). For example, one that creates a small function in an executable data segment that calls f(), with real_f() moving that data segment, and correspondingly adjusting the return address. That would match OP's scenario, and is just about the only practical use case for this kind of manipulation I can think of. Instead, I just hacked a crude example that might or might not work for others.)
Also, we might wish to declare f() as having two parameters (they would be passed in %rdi and %rsi) too, with the second being irrelevant, to make sure the compiler does not expect %rsi to stay unchanged. (If I recall correctly, the SysV ABI lets us clobber it, but I might remember wrong.)
On this particular machine, compiling the above with
gcc -Wall -O0 f.s example.c -o example
running it
./example
produces
Function and label addresses:
0x400650 f()
0x400659 real_f()
0x400729 one_call:
0x400733 one_fail:
0x40074c one_skip:
f(one_skip):
real_f(): Returning to 0x40074c instead of 0x400733.
At one_skip.
Note that if you tell GCC to optimize the code (say, -O2), it will make assumptions about the code in main() it is perfectly allowed to do by the C standard, but which may lead to all three labels having the exact same address. This happens on my particular machine and GCC-5.4.0, and of course causes an endless loop. It does not reflect on the implementation of f() or real_f() at all, only that my example in main() is quite poor. I'm lazy.
Related
In a C99 program, under the (theoretical) assumption that I'm not using variable-length arrays, and each of my automatic variables can only exist once at a time in the whole stack (by forbidding circular function calls and explicit recursion), if I sum up all the space they are consuming, could I declare that this is the maximal stack size that can ever happen?
A bit of context here: I told a friend that I wrote a program not using dynamic memory allocation ("malloc") and allocate all memory static (by modeling all my state variables in a struct, which I then declared global). He then told me that if I'm using automatic variables, I still make use of dynamic memory. I argued that my automatic variables are not state variables but control variables, so my program is still to be considered static. We then discussed that there has to be a way to make a statement about the absolute worst-case behaviour about my program, so I came up with the above question.
Bonus question: If the assumptions above hold, I could simply declare all automatic variables static and would end up with a "truly" static program?
Even if array sizes are constant a C implementation could allocate arrays and even structures dynamically. I'm not aware of any that do (anyone) and it would appear quite unhelpful. But the C Standard doesn't make such guarantees.
There is also (almost certainly) some further overhead in the stack frame (the data added to the stack on call and released on return).
You would need to declare all your functions as taking no parameters and returning void to ensure no program variables in the stack. Finally the 'return address' of where execution of a function is to continue after return is pushed onto the stack (at least logically).
So having removed all parameters, automatic variables and return values to you 'state' struct there will still be something going on to the stack - probably.
I say probably because I'm aware of a (non-standard) embedded C compiler that forbids recursion that can determine the maximum size of the stack by examining the call tree of the whole program and identify the call chain that reaches the peek size of the stack.
You could achieve this a monstrous pile of goto statements (some conditional where a functon is logically called from two places or by duplicating code.
It's often important in embedded code on devices with tiny memory to avoid any dynamic memory allocation and know that any 'stack-space' will never overflow.
I'm happy this is a theoretical discussion. What you suggest is a mad way to write code and would throw away most of (ultimately limited) services C provides to infrastructure of procedural coding (pretty much the call stack)
Footnote: See the comment below about the 8-bit PIC architecture.
Bonus question: If the assumptions above hold, I could simply declare
all automatic variables static and would end up with a "truly" static
program?
No. This would change the function of the program. static variables are initialized only once.
Compare this 2 functions:
int canReturn0Or1(void)
{
static unsigned a=0;
a++;
if(a>1)
{
return 1;
}
return 0;
}
int willAlwaysReturn0(void)
{
unsigned a=0;
a++;
if(a>1)
{
return 1;
}
return 0;
}
In a C99 program, under the (theoretical) assumption that I'm not using variable-length arrays, and each of my automatic variables can only exist once at a time in the whole stack (by forbidding circular function calls and explicit recursion), if I sum up all the space they are consuming, could I declare that this is the maximal stack size that can ever happen?
No, because of function pointers..... Read n1570.
Consider the following code, where rand(3) is some pseudo random number generator (it could also be some input from a sensor) :
typedef int foosig(int);
int foo(int x) {
foosig* fptr = (x>rand())?&foo:NULL;
if (fptr)
return (*fptr)(x);
else
return x+rand();
}
An optimizing compiler (such as some recent GCC suitably invoked with enough optimizations) would make a tail-recursive call for (*fptr)(x). Some other compiler won't.
Depending on how you compile that code, it would use a bounded stack or could produce a stack overflow. With some ABI and calling conventions, both the argument and the result could go thru a processor register and won't consume any stack space.
Experiment with a recent GCC (e.g. on Linux/x86-64, some GCC 10 in 2020) invoked as gcc -O2 -fverbose-asm -S foo.c then look inside foo.s. Change the -O2 to a -O0.
Observe that the naive recursive factorial function could be compiled into some iterative machine code with a good enough C compiler and optimizer. In practice GCC 10 on Linux compiling the below code:
int fact(int n)
{
if (n<1) return 1;
else return n*fact(n-1);
}
as gcc -O3 -fverbose-asm tmp/fact.c -S -o tmp/fact.s produces the following assembler code:
.type fact, #function
fact:
.LFB0:
.cfi_startproc
endbr64
# tmp/fact.c:3: if (n<1) return 1;
movl $1, %eax #, <retval>
testl %edi, %edi # n
jle .L1 #,
.p2align 4,,10
.p2align 3
.L2:
imull %edi, %eax # n, <retval>
subl $1, %edi #, n
jne .L2 #,
.L1:
# tmp/fact.c:5: }
ret
.cfi_endproc
.LFE0:
.size fact, .-fact
.ident "GCC: (Ubuntu 10.2.0-5ubuntu1~20.04) 10.2.0"
And you can observe that the call stack is not increasing above.
If you have serious and documented arguments against GCC, please submit a bug report.
BTW, you could write your own GCC plugin which would choose to randomly apply or not such an optimization. I believe it stays conforming to the C standard.
The above optimization is essential for many compilers generating C code, such as Chicken/Scheme or Bigloo.
A related theorem is Rice's theorem. See also this draft report funded by the CHARIOT project.
See also the Compcert project.
Let's there is a function:
int caller()
{
int arg1 = 1;
int arg2 = 2
int a = test(&arg1, &arg2)
}
test(int *a, int *b)
{
...
}
so I don't understand why &arg1 and &arg2 have to be pushed on the stack too like this
I can understand that we can get address of arg1 and arg2 in the callee by using
movl 8(%ebp), %edx
movl 12(%ebp), %ecx
but if we don't push these two on the stack,
we can also can their address by using:
leal 8(%ebp), %edx
leal 12(%ebp), %ecx
so why bother pushing &arg1 and &arg2 on the stack?
In the general case, test has to work when you pass it arbitrary pointers, including to extern int global_var or whatever. Then main has to call it according to the ABI / calling convention.
So the asm definition of test can't assume anything about where int *a points, e.g. that it points into its caller's stack frame.
(Or you could look at that as optimizing away the addresses in a call-by-reference on locals, so the caller must place the pointed-to objects in the arg-passing slots, and on return those 2 dwords of stack memory hold the potentially-updated values of *a and *b.)
You compiled with optimization disabled. Especially for the special case where the caller is passing pointers to locals, the solution to this problem is to inline the whole function, which compilers will do when optimization is enabled.
Compilers are allowed to make a private clone of test that takes its args by value, or in registers, or with whatever custom calling convention the compiler wants to use. Most compilers don't actually do this, though, and rely on inlining instead of custom calling conventions for private functions to get rid of arg-passing overhead.
Or if it had been declared static test, then the compiler would already know it was private and could in theory use whatever custom calling convention it wanted without making a clone with a name like test.clone1234. gcc does sometimes actually do that for constant-propagation, e.g. if the caller passes a compile-time constant but gcc chooses not to inline. (Or can't because you used __attribute__((noinline)) static test() {})
And BTW, with a good register-args calling convention like x86-64 System V, the caller would do lea 12(%rsp), %rdi / lea 8(%rsp), %rsi / call test or something. The i386 System V calling convention is old and inefficient, passing everything on the stack forcing a store/reload.
You have basically identified one of the reasons that stack-args calling conventions have higher overhead and generally suck.
if you access arg1 and arg2 directly, it means you are accessing a portion of stack that does not belong to this function. This is somehow what happens when someone uses a buffer overflow attack to access additional data from calling stack.
When your call has arguments, arguments are pushed into stack(in your case &arg1 and &arg2) and function can use them as valid list of arguments for this function.
How Get arguments value using inline assembly in C without Glibc?
i require this code for Linux archecture x86_64 and i386.
if you know about MAC OS X or Windows , also submit and please guide.
void exit(int code)
{
//This function not important!
//...
}
void _start()
{
//How Get arguments value using inline assembly
//in C without Glibc?
//argc
//argv
exit(0);
}
New Update
https://gist.github.com/apsun/deccca33244471c1849d29cc6bb5c78e
and
#define ReadRdi(To) asm("movq %%rdi,%0" : "=r"(To));
#define ReadRsi(To) asm("movq %%rsi,%0" : "=r"(To));
long argcL;
long argvL;
ReadRdi(argcL);
ReadRsi(argvL);
int argc = (int) argcL;
//char **argv = (char **) argvL;
exit(argc);
But it still returns 0.
So this code is wrong!
please help.
As specified in the comment, argc and argv are provided on the stack, so you cannot use a regular C function to get them, even with inline assembly, as the compiler will touch the stack pointer to allocate the local variables, setup the stack frame & co.; hence, _start must be written in assembly, as it's done in glibc (x86; x86_64). A small stub can be written to just grab the stuff and forward it to your "real" C entrypoint according to the regular calling convention.
Here a minimal example of a program (both for x86 and x86_64) that reads argc and argv, prints all the values in argv on stdout (separated by newline) and exits using argc as status code; it can be compiled with the usual gcc -nostdlib (and -static to make sure ld.so isn't involved; not that it does any harm here).
#ifdef __x86_64__
asm(
".global _start\n"
"_start:\n"
" xorl %ebp,%ebp\n" // mark outermost stack frame
" movq 0(%rsp),%rdi\n" // get argc
" lea 8(%rsp),%rsi\n" // the arguments are pushed just below, so argv = %rbp + 8
" call bare_main\n" // call our bare_main
" movq %rax,%rdi\n" // take the main return code and use it as first argument for...
" movl $60,%eax\n" // ... the exit syscall
" syscall\n"
" int3\n"); // just in case
asm(
"bare_write:\n" // write syscall wrapper; the calling convention is pretty much ok as is
" movq $1,%rax\n" // 1 = write syscall on x86_64
" syscall\n"
" ret\n");
#endif
#ifdef __i386__
asm(
".global _start\n"
"_start:\n"
" xorl %ebp,%ebp\n" // mark outermost stack frame
" movl 0(%esp),%edi\n" // argc is on the top of the stack
" lea 4(%esp),%esi\n" // as above, but with 4-byte pointers
" sub $8,%esp\n" // the start starts 16-byte aligned, we have to push 2*4 bytes; "waste" 8 bytes
" pushl %esi\n" // to keep it aligned after pushing our arguments
" pushl %edi\n"
" call bare_main\n" // call our bare_main
" add $8,%esp\n" // fix the stack after call (actually useless here)
" movl %eax,%ebx\n" // take the main return code and use it as first argument for...
" movl $1,%eax\n" // ... the exit syscall
" int $0x80\n"
" int3\n"); // just in case
asm(
"bare_write:\n" // write syscall wrapper; convert the user-mode calling convention to the syscall convention
" pushl %ebx\n" // ebx is callee-preserved
" movl 8(%esp),%ebx\n" // just move stuff from the stack to the correct registers
" movl 12(%esp),%ecx\n"
" movl 16(%esp),%edx\n"
" mov $4,%eax\n" // 4 = write syscall on i386
" int $0x80\n"
" popl %ebx\n" // restore ebx
" ret\n"); // notice: the return value is already ok in %eax
#endif
int bare_write(int fd, const void *buf, unsigned count);
unsigned my_strlen(const char *ch) {
const char *ptr;
for(ptr = ch; *ptr; ++ptr);
return ptr-ch;
}
int bare_main(int argc, char *argv[]) {
for(int i = 0; i < argc; ++i) {
int len = my_strlen(argv[i]);
bare_write(1, argv[i], len);
bare_write(1, "\n", 1);
}
return argc;
}
Notice that here several subtleties are ignored - in particular, the atexit bit. All the documentation about the machine-specific startup state has been extracted from the comments in the two glibc files linked above.
This answer is for x86-64, 64-bit Linux ABI, only. All the other OSes and ABIs mentioned will be broadly similar, but different enough in the fine details that you will need to write your custom _start once for each.
You are looking for the specification of the initial process state in the "x86-64 psABI", or, to give it its full title, "System V Application Binary Interface, AMD64 Architecture Processor Supplement (With LP64 and ILP32 Programming Models)". I will reproduce figure 3.9, "Initial Process Stack", here:
Purpose Start Address Length
------------------------------------------------------------------------
Information block, including varies
argument strings, environment
strings, auxiliary information
...
------------------------------------------------------------------------
Null auxiliary vector entry 1 eightbyte
Auxiliary vector entries... 2 eightbytes each
0 eightbyte
Environment pointers... 1 eightbyte each
0 8+8*argc+%rsp eightbyte
Argument pointers... 8+%rsp argc eightbytes
Argument count %rsp eightbyte
It goes on to say that the initial registers are unspecified except
for %rsp, which is of course the stack pointer, and %rdx, which may contain "a function pointer to register with atexit".
So all the information you are looking for is already present in memory, but it hasn't been laid out according to the normal calling convention, which means you must write _start in assembly language. It is _start's responsibility to set everything up to call main with, based on the above. A minimal _start would look something like this:
_start:
xorl %ebp, %ebp # mark the deepest stack frame
# Current Linux doesn't pass an atexit function,
# so you could leave out this part of what the ABI doc says you should do
# You can't just keep the function pointer in a call-preserved register
# and call it manually, even if you know the program won't call exit
# directly, because atexit functions must be called in reverse order
# of registration; this one, if it exists, is meant to be called last.
testq %rdx, %rdx # is there "a function pointer to
je skip_atexit # register with atexit"?
movq %rdx, %rdi # if so, do it
call atexit
skip_atexit:
movq (%rsp), %rdi # load argc
leaq 8(%rsp), %rsi # calc argv (pointer to the array on the stack)
leaq 8(%rsp,%rdi,8), %rdx # calc envp (starts after the NULL terminator for argv[])
call main
movl %eax, %edi # pass return value of main to exit
call exit
hlt # should never get here
(Completely untested.)
(In case you're wondering why there's no adjustment to maintain stack pointer alignment, this is because upon a normal procedure call, 8(%rsp) is 16-byte aligned, but when _start is called, %rsp itself is 16-byte aligned. Each call instruction displaces %rsp down by eight, producing the alignment situation expected by normal compiled functions.)
A more thorough _start would do more things, such as clearing all the other registers, arranging for greater stack pointer alignment than the default if desired, calling into the C library's own initialization functions, setting up environ, initializing the state used by thread-local storage, doing something constructive with the auxiliary vector, etc.
You should also be aware that if there is a dynamic linker (PT_INTERP section in the executable), it receives control before _start does. Glibc's ld.so cannot be used with any C library other than glibc itself; if you are writing your own C library, and you want to support dynamic linkage, you will also need to write your own ld.so. (Yes, this is unfortunate; ideally, the dynamic linker would be a separate development project and its complete interface would be specified.)
As a quick and dirty hack, you can make an executable with a compiled C function as the ELF entry point. Just make sure you use exit or _exit instead of returning.
(Link with gcc -nostartfiles to omit CRT but still link other libraries, and write a _start() in C. Beware of ABI violations like stack alignment, e.g. use -mincoming-stack-boundary=2 or an __attribte__ on _start, as in Compiling without libc)
If it's dynamically linked, you can still use glibc functions on Linux (because the dynamic linker runs glibc's init functions). Not all systems are like this, e.g. on cygwin you definitely can't call libc functions if you (or the CRT start code) hasn't called the libc init functions in the correct order. I'm not sure it's even guaranteed that this works on Linux, so don't depend on it except for experimentation on your own system.
I have used a C _start(void){ ... } + calling _exit() for making a static executable to microbenchmark some compiler-generated code with less startup overhead for perf stat ./a.out.
Glibc's _exit() works even if glibc wasn't initialized (gcc -O3 -static), or use inline asm to run xor %edi,%edi / mov $60, %eax / syscall (sys_exit(0) on Linux) so you don't have to even statically link libc. (gcc -O3 -nostdlib)
With even more dirty hacking and UB, you can access argc and argv by knowing the x86-64 System V ABI that you're compiling for (see #zwol's answer for a quote from ABI doc), and how the process startup state differers from the function calling convention:
argc is where the return address would be for a normal function (pointed to by RSP). GNU C has a builtin for accessing the return address of the current function (or for walking up the stack.)
argv[0] is where the 7th integer/pointer arg should be (the first stack arg, just above the return address). It happens to / seems to work to take its address and use that as an array!
// Works only for the x86-64 SystemV ABI; only tested on Linux.
// DO NOT USE THIS EXCEPT FOR EXPERIMENTS ON YOUR OWN COMPUTER.
#include <stdio.h>
#include <stdlib.h>
// tell gcc *this* function is called with a misaligned RSP
__attribute__((force_align_arg_pointer))
void _start(int dummy1, int dummy2, int dummy3, int dummy4, int dummy5, int dummy6, // register args
char *argv0) {
int argc = (int)(long)__builtin_return_address(0); // load (%rsp), casts to silence gcc warnings.
char **argv = &argv0;
printf("argc = %d, argv[argc-1] = %s\n", argc, argv[argc-1]);
printf("%f\n", 1.234); // segfaults if RSP is misaligned
exit(0);
//_exit(0); // without flushing stdio buffers!
}
# with a version without the FP printf
peter#volta:~/src/SO$ gcc -nostartfiles _start.c -o bare_start
peter#volta:~/src/SO$ ./bare_start
argc = 1, argv[argc-1] = ./bare_start
peter#volta:~/src/SO$ ./bare_start abc def hij
argc = 4, argv[argc-1] = hij
peter#volta:~/src/SO$ file bare_start
bare_start: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, BuildID[sha1]=af27c8416b31bb74628ef9eec51a8fc84e49550c, not stripped
# I could have used -fno-pie -no-pie to make a non-PIE executable
This works with or without optimization, with gcc7.3. I was worried that without optimization, the address of argv0 would be below rbp where it copies the arg, rather than its original location. But apparently it works.
gcc -nostartfiles links glibc but not the CRT start files.
gcc -nostdlib omits both libraries and CRT startup files.
Very little of this is guaranteed to work, but it does in practice work with current gcc on current x86-64 Linux, and has worked in the past for years. If it breaks, you get to keep both pieces. IDK what C features are broken by omitting the CRT startup code and just relying on the dynamic linker to run glibc init functions. Also, taking the address of an arg and accessing pointers above it is UB, so you could maybe get broken code-gen. gcc7.3 happens to do what you'd expect in this case.
Things that definitely break
atexit() cleanup, e.g. flushing stdio buffers.
static destructors for static objects in dynamically-linked libraries. (On entry to _start, RDX is a function pointer you should register with atexit for this reason. In a dynamically linked executable, the dynamic linker runs before your _start and sets RDX before jumping to your _start. Statically linked executables have RDX=0 under Linux.)
gcc -mincoming-stack-boundary=3 (i.e. 2^3 = 8 bytes) is another way to get gcc to realign the stack, because the -mpreferred-stack-boundary=4 default of 2^4 = 16 is still in place. But that makes gcc assume under-aligned RSP for all functions, not just for _start, which is why I looked in the docs and found an attribute that was intended for 32-bit when the ABI transitioned from only requiring 4-byte stack alignment to the current requirement of 16-byte alignment for ESP in 32-bit mode.
The SysV ABI requirement for 64-bit mode has always been 16-byte alignment, but gcc options let you make code that doesn't follow the ABI.
// test call to a function the compiler can't inline
// to see if gcc emits extra code to re-align the stack
// like it would if we'd used -mincoming-stack-boundary=3 to assume *all* functions
// have only 8-byte (2^3) aligned RSP on entry, with the default -mpreferred-stack-boundary=4
void foo() {
int i = 0;
atoi(NULL);
}
With -mincoming-stack-boundary=3, we get stack-realignment code there, where we don't need it. gcc's stack-realignment code is pretty clunky, so we'd like to avoid that. (Not that you'd really ever use this to compile a significant program where you care about efficiency, please only use this stupid computer trick as a learning experiment.)
But anyway, see the code on the Godbolt compiler explorer with and without -mpreferred-stack-boundary=3.
Consider:
#include <stdio.h>
char toUpper(char);
int main(void)
{
char ch, ch2;
printf("lowercase input: ");
ch = getchar();
ch2 = toUpper(ch);
printf("%c ==> %c\n", ch, ch2);
return 0;
}
char toUpper(char c)
{
if(c>='a' && c<='z')
c = c - 32;
}
In the toUpper function, the return type is char, but there isn't any "return" in toUpper(). And compile the source code with gcc (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4), Fedora 14.
Of course, a warning is issued: "warning: control reaches end of non-void function", but, working well.
What has happened in that code during compile with gcc?
When the C program was compiled into assembly language, your toUpper function ended up like this, perhaps:
_toUpper:
LFB4:
pushq %rbp
LCFI3:
movq %rsp, %rbp
LCFI4:
movb %dil, -4(%rbp)
cmpb $96, -4(%rbp)
jle L8
cmpb $122, -4(%rbp)
jg L8
movzbl -4(%rbp), %eax
subl $32, %eax
movb %al, -4(%rbp)
L8:
leave
ret
The subtraction of 32 was carried out in the %eax register. And in the x86 calling convention, that is the register in which the return value is expected to be! So... you got lucky.
But please pay attention to the warnings. They are there for a reason!
It depends on the Application Binary Interface and which registers are used for the computation.
E.g. on x86, the first function parameter and the return value is stored in EAX and so gcc is most likely using this to store the result of the calculation as well.
Essentially, c is pushed into the spot that should later be filled with the return value; since it's not overwritten by use of return, it ends up as the value returned.
Note that relying on this (in C, or any other language where this isn't an explicit language feature, like Perl), is a Bad Idea™. In the extreme.
One missing thing that's important to understand is that it's rarely a diagnosable error to omit a return statement. Consider this function:
int f(int x)
{
if (x!=42) return x*x;
}
As long as you never call it with an argument of 42, a program containing this function is perfectly valid C and does not invoke any undefined behavior, despite the fact that it would invoke UB if you called f(42) and subsequently attempted to use the return value.
As such, while it's possible for a compiler to provide warning heuristics for missing return statements, it's impossible to do so without false positives or false negatives. This is a consequence of the impossibility of solving the halting problem.
I can't tell you the specifics of your platform as I don't know it, but there is a general answer to the behaviour you see.
When some function that has a return is compiled, the compiler will use a convention on how to return that data. It could be a machine register, or a defined memory location such as via a stack or whatever (though generally machine registers are used). The compiled code may also use that location (register or otherwise) while doing the work of the function.
If the function doesn't return anything, then the compiler will not generate code that explicitly fills that location with a return value. However, like I said above, it may use that location during the function. When you write code that reads the return value (ch2 = toUpper(ch);), the compiler will write code that uses its convention on how retrieve that return from the conventional location. As far as the caller code is concerned, it will just read that value from the location, even if nothing was written explicitly there. Hence you get a value.
Now look at Ray's example. The compiler used the EAX register to store the results of the upper casing operation. It just so happens, this is probably the location that return values are written to. On the calling side, ch2 is loaded with the value that's in EAX - hence a phantom return. This is only true of the x86 range of processors, as on other architectures the compiler may use a completely different scheme in deciding how the convention should be organised.
However, good compilers will try optimise according to a set of local conditions, knowledge of code, rules, and heuristics. So an important thing to note is that this is just luck that it works. The compiler could optimise and not do this or whatever - you should not reply on the behaviour.
You should keep in mind that such code may crash depending on the compiler. For example, Clang generates a ud2 instruction at the end of such function and your app will crash at run time.
There are no local variables, so the value on the top of the stack at the end of the function will be the parameter c. The value at the top of the stack upon exiting, is the return value. So whatever c holds, that's the return value.
I have tried a small program:
#include <stdio.h>
int f1() {
}
int main() {
printf("TEST: <%d>\n", f1());
printf("TEST: <%d>\n", f1());
printf("TEST: <%d>\n", f1());
printf("TEST: <%d>\n", f1());
printf("TEST: <%d>\n", f1());
}
Result:
TEST: <1>
TEST: <10>
TEST: <11>
TEST: <11>
TEST: <11>
I have used the MinGW-GCC compiler, so there might be differences.
You could just play around and try, e.g., a char function.
As long you don't use the result value, it will still work fine.
#include <stdio.h>
char f1() {
}
int main() {
f1();
}
But I still would recommend to set either void function or give some return value.
Your function seems to need a return:
char toUpper(char c)
{
if(c>='a'&&c<='z')
c = c - 32;
return c;
}
Sometimes a function doesn't use an argument (perhaps because another "flags" argument doesn't enable a specific feature).
However, you have to specify something, so usually you just put 0. But if you do that, and the function is external, gcc will emit code to "really make sure" that parameter gets set to 0.
Is there a way to tell gcc that a particular argument to a function doesn't matter and it can leave alone whatever value it is that happens to be in the argument register right then?
Update: Someone asked about the XY problem. The context behind this question is I want to implement a varargs function in x86_64 without using the compiler varargs support. This is simplest when the parameters are on the stack, so I declare my functions to take 5 or 6 dummy parameters first, so that the last non-vararg parameter and all of the vararg parameters end up on the stack. This works fine, except it's clearly not optimal - when looking at the assembly code it's clear that gcc is initializing all those argument registers to zero in the caller.
Please don't take below answer seriously. The question asks for a hack so there you go.
GCC will effectively treat value of uninitialized variable as "don't care" so we can try exploiting this:
int foo(int x, int y);
int bar_1(int y) {
int tmp = tmp; // Suppress uninitialized warnings
return foo(tmp, y);
}
Unfortunately my version of GCC still cowardly initializes tmp to zero but yours may be more aggressive:
bar_1:
.LFB0:
.cfi_startproc
movl %edi, %esi
xorl %edi, %edi
jmp foo
.cfi_endproc
Another option is (ab)using inline assembly to fake GCC into thinking that tmp is defined (when in fact it isn't):
int bar_2(int y) {
int tmp;
asm("" : "=r"(tmp));
return foo(tmp, y);
}
With this GCC managed to get rid of parameter initializations:
bar_2:
.LFB1:
.cfi_startproc
movl %edi, %esi
jmp foo
.cfi_endproc
Note that inline asm must be immediately before the function call, otherwise GCC will think it has to preserve output values which would harm register allocation.