| /trunk/kernel/arch/amd64/include/atomic.h |
|---|
| 108,13 → 108,13 |
| #endif |
| "mov %0, %1\n" |
| "testq %1, %1\n" |
| "jnz 0b\n" /* Lightweight looping on locked spinlock */ |
| "jnz 0b\n" /* lightweight looping on locked spinlock */ |
| "incq %1\n" /* now use the atomic operation */ |
| "xchgq %0, %1\n" |
| "testq %1, %1\n" |
| "jnz 0b\n" |
| : "+m" (val->count), "=r"(tmp) |
| : "+m" (val->count), "=&r" (tmp) |
| ); |
| /* |
| * Prevent critical section code from bleeding out this way up. |