...
Wiki Markup |
---|
However, as demonstrated in \[[Eide and Regehr|AA. C References#Eide and Regehr]\], all tested compilers generated some percentage of incorrect compiled code with regards to {{volatile}} accesses. Therefore, it is necessary to know how your compiler behaves when the standard {{volatile}} behavior is required. There is also a workaround that eliminates some or all of these errors \[[Eide and Regehr|AA. C References#Eide and Regehr]\]. |
...
Noncompliant Code Example
Wiki Markup |
---|
As demonstrated in \[[Eide and Regehr|AA. C References#Eide and Regehr]\], the following code example compiles incorrectly using GCC version 4.3.0 for IA32 and the {{\-Os}} optimization flag: |
Code Block |
---|
|
const volatile int x;
volatile int y;
void foo(void) {
for(y = 0; y < 10; y++) {
int z = x;
}
}
|
Wiki Markup |
---|
Because the variable {{x}} is {{volatile}}\-qualified, it should be accessed ten times in this program. However, as shown in the compiled object code, it is only accessed once due to a loop-hoisting optimization \[[Eide and Regehr|AA. C References#Eide and Regehr]\]: |
Code Block |
---|
|
foo:
movl $0, y
movl x, %eax
jmp .L2
.L3:
movl y, %eax
incl %eax
movl %eax, y
.L2:
movl y, %eax
cmpl $10, %eax
jg .L3
ret
|
Should x
represent a hardware register or some other memory-mapped device that has side effects when accessed, the previous miscompiled code example may produce unexpected behavior.
...
Compliant Solution
Wiki Markup |
---|
Eide and Regehr tested a workaround by wrapping {{volatile}} accesses with function calls. They describe it with the intuition that "we can replace an action that compilers empirically get wrong by a different action â--- a function call â--- that compilers can get right" \[[Eide and Regehr|AA. C References#Eide and Regehr]\]. For example, the workaround considerfor the followingnoncompliant code example thatwould could be miscompiled: |
Code Block |
---|
volatile x;
x = x;
|
Wiki Markup |
---|
To work around some of the {{volatile}}-access compiler bugs, wrap both accesses to {{x}} in function calls \[[Eide and Regehr]\]: |
|
Code Block |
---|
int vol_read_int(volatile int *vp) {
return *vp;
}
volatile int *vol_id_int(volatile int *vp) {
return vp;
}
const volatile int x;
xvolatile =int xy;
void foo(void) {
for(*vol_id_int(&xy) = 0; vol_read_int(&y) < 10; *vol_id_int(&y) = vol_read_int(&y) + 1) {
int z = vol_read_int(&x);
}
}
|
Wiki Markup |
---|
The workarounds proposed in \[[Eide and Regehr|AA. C References#Eide and Regehr]\] fix many of the {{volatile}}\-access bugs in the tested compilers. However, compilers are always changing so critical sections of code should be compiled as if for deployment and the compiled object code should be inspected for the correct behavior. |
...
Wiki Markup |
---|
\[[Eide and Regehr|AA. C References#Eide and Regehr]\] "Volatiles Areare Miscompiledmiscompiled, and Whatwhat to Dodo about Itit"
\[[ISO/IEC 9899:1999|AA. C References#ISO/IEC 9899-1999]\] Section 6.7.3, "Type qualifiers" |