http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50583
Bug #: 50583
Summary: Many __sync_XXX builtin functions are incorrect
Classification: Unclassified
Product: gcc
Version: 4.7.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
AssignedTo: unassig...@gcc.gnu.org
ReportedBy: hjl.to...@gmail.com
CC: kirill.yuk...@intel.com, ubiz...@gmail.com
We have
`TYPE __sync_fetch_and_add (TYPE *ptr, TYPE value, ...)'
`TYPE __sync_fetch_and_sub (TYPE *ptr, TYPE value, ...)'
`TYPE __sync_fetch_and_or (TYPE *ptr, TYPE value, ...)'
`TYPE __sync_fetch_and_and (TYPE *ptr, TYPE value, ...)'
`TYPE __sync_fetch_and_xor (TYPE *ptr, TYPE value, ...)'
`TYPE __sync_fetch_and_nand (TYPE *ptr, TYPE value, ...)'
These builtins perform the operation suggested by the name, and
returns the value that had previously been in memory. That is,
{ tmp = *ptr; *ptr OP= value; return tmp; }
{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; } // nand
On x86, they may be implemented as
[hjl@gnu-33 gcc]$ cat /tmp/x.c
int
foo (int *p)
{
return __sync_fetch_and_and(p, 7);
}
[hjl@gnu-33 gcc]$ gcc -S -O2 /tmp/x.c
[hjl@gnu-33 gcc]$ cat x.s
.file "x.c"
.text
.p2align 4,,15
.globl foo
.type foo, @function
foo:
.LFB0:
.cfi_startproc
movl (%rdi), %eax
.L2:
movl %eax, %edx
movl %eax, %ecx
andl $7, %edx
lock cmpxchgl %edx, (%rdi)
jne .L2
movl %ecx, %eax
ret
.cfi_endproc
.LFE0:
.size foo, .-foo
.ident "GCC: (GNU) 4.6.0 20110603 (Red Hat 4.6.0-10)"
.section .note.GNU-stack,"",@progbits
[hjl@gnu-33 gcc]$
This isn't equivalent to
{ tmp = *ptr; *ptr OP= value; return tmp; }
It is
{
tmp = *ptr;
tmp1 = tmp OP value;
return __sync_val_compare_and_swap (ptr, tmp, tmp1);
}
The only thing supported on x86 are
[hjl@gnu-33 gcc]$ cat /tmp/y.c
int
foo1 (int *p)
{
return __sync_fetch_and_add (p, 7);
}
int
foo2 (int *p)
{
return __sync_fetch_and_sub (p, 7);
}
[hjl@gnu-33 gcc]$ gcc -S -O2 /tmp/y.c
[hjl@gnu-33 gcc]$ cat y.s
.file "y.c"
.text
.p2align 4,,15
.globl foo1
.type foo1, @function
foo1:
.LFB0:
.cfi_startproc
movl $7, %eax
lock xaddl %eax, (%rdi)
ret
.cfi_endproc
.LFE0:
.size foo1, .-foo1
.p2align 4,,15
.globl foo2
.type foo2, @function
foo2:
.LFB1:
.cfi_startproc
movl $-7, %eax
lock xaddl %eax, (%rdi)
ret
.cfi_endproc
.LFE1:
.size foo2, .-foo2
.ident "GCC: (GNU) 4.6.0 20110603 (Red Hat 4.6.0-10)"
.section .note.GNU-stack,"",@progbits
[hjl@gnu-33 gcc]$
