blob: 77e8a220c85a472e44b09406969a091868f93dcb (
plain) (
blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
|
instruction: 0.607
runtime: 0.209
syscall: 0.185
linux-user: incorrect alignment of sigframe::pretcode & rt_sigframe::pretcode cause crash
Description of problem:
Corrent Print Result:
sp: cdd3b4e8
SUCCEEDED!
qemu-x86_64 Print Result:
sp: 2804170
qemu: uncaught target signal 11 (Segmentation fault) - core dumped
Segmentation fault
Reason of Bug:
sigframe::pretcode & rt_sigframe::pretcode must align of 16n-sizeof(void*) instead of 16n, Because rsp align of 16n before instruction "call" in caller, After "call", push address of "call" in caller. sp of begin in callee is 16n-sizeof(void*)
For example on x86_64:
reference to "qemu/linux-user/i386/signal.c"
```
# define TARGET_FPSTATE_FXSAVE_OFFSET 0
struct rt_sigframe {
abi_ulong pretcode;
struct target_ucontext uc;
struct target_siginfo info;
struct target_fpstate fpstate QEMU_ALIGNED(16);
};
#define TARGET_RT_SIGFRAME_FXSAVE_OFFSET ( \
offsetof(struct rt_sigframe, fpstate) + TARGET_FPSTATE_FXSAVE_OFFSET)
```
offsetof(struct rt_sigframe, fpstate) align of 16
TARGET_FPSTATE_FXSAVE_OFFSET is 0
TARGET_RT_SIGFRAME_FXSAVE_OFFSET is 16n, also alignment of fxsave is 64
so address of rt_sigframe::pretcode is 16n instead of 16n - sizeof(void*), It is incorect!
Fix the bug:
```
struct rt_sigframe {
abi_ulong pretcode;
struct target_ucontext uc;
struct target_siginfo info;
abi_ulong unused QEMU_ALIGNED(16);
struct target_fpstate fpstate;
};
```
offsetof(struct rt_sigframe, fpstate) is 16n+8, so address of rt_sigframe::pretcode is 16n-8 on x86_64.
|