On Wed, Nov 29, 2017 at 6:54 PM, Dennis Zhou <dennisszhou@xxxxxxxxx> wrote:
> Hi everyone,
>
> I spent a bit of time learning more about this problem as Fengguang was
> able to determine the root commit f7dd2507893cc3. I reproduced the bug
> in userspace to make life a bit easier and below the assignment occurs
> before the unpoison. This is fine if we're sequentially proceeding, but
> as in the case in percpu, it's calling the function in a for loop
> causing the assignment to happen after it has been poisoned in the prior
> iteration.
>
> <bb 3> [0.00%]:
> _1 = (long unsigned int) i_4;
> _2 = _1 * 16;
> _3 = p_8 + _2;
> list_14 = _3;
> __u = {};
> ASAN_MARK (UNPOISON, &__u, 8);
> __u.__val = list_14;
>
> <bb 9> [0.00%]:
> _24 = __u.__val;
> ASAN_MARK (POISON, &__u, 8);
> list_14->prev = list_14;
> i_13 = i_4 + 1;
>
> <bb 10> [0.00%]:
> # i_4 = PHI <i_9(2), i_13(9)>
> if (i_4 <= 9)
> goto <bb 3>; [0.00%]
> else
> goto <bb 11>; [0.00%]
>
> I don't know how to go about fixing this though. The reproducing code is
> below and was compiled with gcc-7 and the structleak_plugin.
Are we sure that structleak plugin is not at fault? If yes, then we
need to report this to https://gcc.gnu.org/bugzilla/ with instructions
on how to build/use the plugin.
> I hope this helps.
>
> Thanks,
> Dennis
>
> ----
> #include <stdint.h>
> #include <stdlib.h>
>
> #define barrier()
>
> #define WRITE_ONCE(x, val) \
> ({ \
> union { typeof(x) __val; char __c[1]; } __u = \
> { .__val = (typeof(x)) (val) }; \
> __write_once_size(&(x), __u.__c, sizeof(x)); \
> __u.__val; \
> })
>
> typedef uint8_t __u8;
> typedef uint16_t __u16;
> typedef uint32_t __u32;
> typedef uint64_t __u64;
>
> static inline __attribute__((always_inline)) void __write_once_size(volatile void *p, void *res, int size)
> {
> switch (size) {
> case 1: *(volatile __u8 *)p = *(__u8 *)res; break;
> case 2: *(volatile __u16 *)p = *(__u16 *)res; break;
> case 4: *(volatile __u32 *)p = *(__u32 *)res; break;
> case 8: *(volatile __u64 *)p = *(__u64 *)res; break;
> default:
> barrier();
> __builtin_memcpy((void *)p, (const void *)res, size);
> barrier();
> }
> }
>
> struct list_head {
> struct list_head *next, *prev;
> };
>
> static inline __attribute__((always_inline)) void INIT_LIST_HEAD(struct list_head *list)
> {
> WRITE_ONCE(list->next, list);
> list->prev = list;
> }
>
> int main(int argc, char *argv[])
> {
> struct list_head *p = malloc(10 * sizeof(struct list_head));
> int i;
>
> for (i = 0; i < 10; i++) {
> INIT_LIST_HEAD(&p[i]);
> }
>
> free(p);
>
> return 0;
> }
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
