[英]Declaring array in pthreads start_routine function causing segmentation fault
#include <stdio.h>
#include <pthread.h>
void* function(void* arg){
int picture[4096][4096];
}
int main(){
int N=10, S=10;
pthread_t pids[10];
pthread_create(&pids[0], NULL, function, NULL);
pthread_join(pids[0], NULL);
return 0;
}
I compiled the above code with: gcc test.c -pthread
.我编译了上面的代码:
gcc test.c -pthread
。
On running the executable, it crashes, displaying: Segmentation fault
.在运行可执行文件时,它会崩溃,并显示:
Segmentation fault
。
But, if I remove the int picture[4096][4096];
但是,如果我删除
int picture[4096][4096];
definition, it doesn't crash.定义,它不会崩溃。
What could be the reason for this?这可能是什么原因?
The crashing program is:崩溃的程序是:
#include <stdio.h>
#include <pthread.h>
void *function(void *arg)
{
int picture[4096][4096]; // 4096*4096*sizeof(int) = 67108864 bytes = 64 MB
}
int main()
{
pthread_t pids[10];
pthread_create(&pids[0],NULL, function, NULL);
pthread_join(pids[0],NULL);
return 0;
}
The program crashes at execution time:程序在执行时崩溃:
$ gcc p.c -lpthread
$ ./a.out
Segmentation fault (core dumped)
The default stack size for a thread in the GLIBC/pthread is 8 MB . GLIBC/pthread中线程的默认堆栈大小为8 MB 。 At thread creation time, the Thread descriptor also called Task Control Block (TCB), is stored at the bottom of the stack and a red zone (guard page of 4 KB without read/write permission is set at the top of the stack).
在线程创建时,线程描述符也称为任务控制块(TCB),存储在堆栈底部和一个红色区域(在堆栈顶部设置了 4 KB 的没有读/写权限的保护页)。 The stack grows from the high to low addresses.
堆栈从高地址向低地址增长。
Result of the program under the control of strace
: strace
控制下的程序结果:
$ strace -f ./a.out
[...]
prlimit64(0, RLIMIT_STACK, NULL, {rlim_cur=8192*1024, rlim_max=RLIM64_INFINITY}) = 0
mmap(NULL, 8392704, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0) = 0x7fee8d4dc000
mprotect(0x7fee8d4dd000, 8388608, PROT_READ|PROT_WRITE) = 0
brk(NULL) = 0x556cf1b72000
brk(0x556cf1b93000) = 0x556cf1b93000
clone(child_stack=0x7fee8dcdbfb0, flags=CLONE_VM|CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_THREAD|CLONE_SYSVSEM|CLONE_SETTLS|CLONE_PARENT_SETTID|CLONE_CHILD_CLEARTIDstrace: Process 3338 attached
, parent_tid=[3338], tls=0x7fee8dcdc700, child_tidptr=0x7fee8dcdc9d0) = 3338
[pid 3338] set_robust_list(0x7fee8dcdc9e0, 24 <unfinished ...>
[pid 3337] futex(0x7fee8dcdc9d0, FUTEX_WAIT, 3338, NULL <unfinished ...>
[pid 3338] <... set_robust_list resumed>) = 0
[pid 3338] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fee8d4dcef0} ---
[pid 3337] <... futex resumed>) = ?
[pid 3338] +++ killed by SIGSEGV (core dumped) +++
+++ killed by SIGSEGV (core dumped) +++
Segmentation fault (core dumped)
In the preceding:在前面:
prlimit64(0, RLIMIT_STACK, NULL, {rlim_cur=8192*1024, rlim_max=RLIM64_INFINITY}) = 0
prlimit64(0, RLIMIT_STACK, NULL, {rlim_cur=8192*1024, rlim_max=RLIM64_INFINITY}) = 0
mmap(NULL, 8392704, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0) = 0x7fee8d4dc000
mmap(NULL, 8392704, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0) = 0x7fee8d4dc000
mprotect(0x7fee8d4dd000, 8388608, PROT_READ|PROT_WRITE) = 0
mprotect(0x7fee8d4dd000, 8388608, PROT_READ|PROT_WRITE) = 0
clone(child_stack=0x7fee8dcdbfb0, flags=CLONE_VM|CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_THREAD|CLONE_SYSVSEM|CLONE_SETTLS|CLONE_PARENT_SETTID|CLONE_CHILD_CLEARTID, parent_tid=[3338], tls=0x7fee8dcdc700, child_tidptr=0x7fee8dcdc9d0) = 3338
clone(child_stack=0x7fee8dcdbfb0, flags=CLONE_VM|CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_THREAD|CLONE_SYSVSEM|CLONE_SETTLS|CLONE_PARENT_SETTID|CLONE_CHILD_CLEARTID, parent_tid=[3338], tls=0x7fee8dcdc700, child_tidptr=0x7fee8dcdc9d0) = 3338
Hence the following memory space layout:因此以下 memory 空间布局:
+ +--------------------+ 0x7fee8d4dc000
| | |
4 KB | | RED ZONE |
(PROT_NONE)| | (guard page) |
+ +--------------------+ 0x7fee8d4dd000
| | |
| | |
| | ^ |
8192 KB | | | |
(PROT_READ/WRITE) | Stack |
| | | |
| | | |
| +--------------------+ 0x7fee8dcdbfb0
| | |
| | TCB + TLS |
| | |
+ +--------------------+ 0x7fee8dcdd000
The thread entry point defines a table of 4096x4096x4
bytes which is equal to 64 MB.线程入口点定义了一个
4096x4096x4
字节的表,等于 64 MB。 This is too much for the 8 MB long stack area.这对于 8 MB 长堆栈区域来说太多了。 However, we could expect no crash at all as the function defines a huge local table but there is no read/write access into it.
但是,我们可以预期根本不会崩溃,因为 function 定义了一个巨大的本地表,但没有对其进行读/写访问。 So, no crash should occur .
因此,不应发生崩溃。
The strace
logs show that the crash occurs upon access to address 0x7fee8d4dcef0 which is above the stack area in the allocated memory zone: [pid 3338] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fee8d4dcef0} ---
strace
日志显示,访问地址0x7fee8d4dcef0时发生崩溃,该地址位于分配的 memory 区域中的堆栈区域上方: [pid 3338] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x7fee8d4dcef0} ---
It is actually in the guard page :它实际上在保护页面中:
+ +--------------------+ 0x7fee8d4dc000
| | |
4 KB | | RED ZONE <--------- Trap @ si_addr=0x7fee8d4dcef0
(PROT_NONE)| | | si_code=SEGV_ACCERR
+ +--------------------+ 0x7fee8d4dd000
| | |
| | |
| | ^ |
8192 KB | | | |
(PROT_READ/WRITE) | Stack |
| | | |
| | | |
| +--------------------+ 0x7fee8dcdbfb0
| | |
| | TCB + TLS |
| | |
+ +--------------------+ 0x7fee8dcdd000
The core dump analysis under gdb
provides the following location for the crash: gdb
下的核心转储分析提供了以下崩溃位置:
$ gdb a.out core
[...]
(gdb) where
#0 0x00005594eb9461a0 in function (arg=<error reading variable: Cannot access memory at address 0x7fe95459ded8>) at p.c:56
#1 0x00007fe95879d609 in start_thread (arg=<optimized out>) at pthread_create.c:477
#2 0x00007fe9586c4293 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
(gdb) disas /m
Dump of assembler code for function function:
56 void* function(void* arg){
0x00005594eb946189 <+0>: endbr64
0x00005594eb94618d <+4>: push %rbp
0x00005594eb94618e <+5>: mov %rsp,%rbp
0x00005594eb946191 <+8>: lea -0x4000000(%rsp),%r11
0x00005594eb946199 <+16>: sub $0x1000,%rsp
=> 0x00005594eb9461a0 <+23>: orq $0x0,(%rsp)
0x00005594eb9461a5 <+28>: cmp %r11,%rsp
0x00005594eb9461a8 <+31>: jne 0x5594eb946199 <function+16>
0x00005594eb9461aa <+33>: sub $0x20,%rsp
0x00005594eb9461ae <+37>: mov %rdi,-0x4000018(%rbp)
0x00005594eb9461b5 <+44>: mov %fs:0x28,%rax
0x00005594eb9461be <+53>: mov %rax,-0x8(%rbp)
0x00005594eb9461c2 <+57>: xor %eax,%eax
57 int picture[4096][4096];
58 }
The above disassembly code of the thread entry point shows that gcc
generates stack accesses every 4 KB (memory page size).上述线程入口点的反汇编代码显示,
gcc
每 4 KB(内存页大小)生成一次堆栈访问。 It first sets R11
register with the address of the beginning of the local table ( 0x4000000 is 4096x4096xsizeof(int) = 67108864 bytes ):它首先将
R11
寄存器设置为本地表开头的地址( 0x4000000是4096x4096xsizeof(int) = 67108864 字节):
0x00005594eb946191 <+8>: lea -0x4000000(%rsp),%r11
Then, it loops "oring" the content of the stack with 0 every 4096 bytes (0x1000):然后,它每 4096 个字节(0x1000)循环“oring”堆栈的内容为 0:
0x00005594eb946199 <+16>: sub $0x1000,%rsp
=> 0x00005594eb9461a0 <+23>: orq $0x0,(%rsp)
0x00005594eb9461a5 <+28>: cmp %r11,%rsp
0x00005594eb9461a8 <+31>: jne 0x5594eb946199 <function+16>
Hence, the crash because at some point, the orq
instruction occurs in the guard page of the stack!因此,崩溃是因为在某些时候,
orq
指令出现在堆栈的保护页中!
NB :注意:
$ gcc p.c -lpthread -O2
$ ./a.out
The optimized disassembly code of function() is a simple "return": function()的优化反汇编代码是一个简单的“返回”:
$ objdump -S a.out
[...]
00000000000011f0 <function>:
11f0: f3 0f 1e fa endbr64
11f4: c3 retq
11f5: 66 2e 0f 1f 84 00 00 nopw %cs:0x0(%rax,%rax,1)
11fc: 00 00 00
11ff: 90 nop
As seen above, by default, the GLIBC/pthread library allocates a default stack of 8 MB.如上所示,默认情况下, GLIBC/pthread库分配 8 MB 的默认堆栈。 But it also provides the ability to set a stack allocated by the user or simply define the stack size with the following steps:
但它也提供了设置用户分配的堆栈的能力,或者通过以下步骤简单地定义堆栈大小:
Here is an enhanced version of the program which defines a stack of 65 MB for the thread:这是该程序的增强版本,它为线程定义了 65 MB 的堆栈:
#include <stdio.h>
#include <pthread.h>
void* function(void* arg)
{
int picture[4096][4096]; // 4096*4096*sizeof(int) = 67108864 bytes = 64 MB
}
int main(void)
{
pthread_t pids[10];
pthread_attr_t attr;
pthread_attr_init(&attr);
pthread_attr_setstacksize(&attr, 65*1024*1024);
pthread_create(&pids[0], &attr, function, NULL);
pthread_join(pids[0], NULL);
pthread_attr_destroy(&attr);
return 0;
}
Build and execution:构建和执行:
$ gcc p2.c -lpthread
$ ./a.out
There is no crash.没有崩溃。 With
strace
, we can verify the behavior:使用
strace
,我们可以验证行为:
$ strace ./a.out
[...]
prlimit64(0, RLIMIT_STACK, NULL, {rlim_cur=8192*1024, rlim_max=RLIM64_INFINITY}) = 0
mmap(NULL, 68161536, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0) = 0x7fe55afd3000
mprotect(0x7fe55afd4000, 68157440, PROT_READ|PROT_WRITE) = 0
brk(NULL) = 0x55b9d7ade000
brk(0x55b9d7aff000) = 0x55b9d7aff000
clone(child_stack=0x7fe55f0d2fb0, flags=CLONE_VM|CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_THREAD|CLONE_SYSVSEM|CLONE_SETTLS|CLONE_PARENT_SETTID|CLONE_CHILD_CLEARTID, parent_tid=[5199], tls=0x7fe55f0d3700, child_tidptr=0x7fe55f0d39d0) = 5199
futex(0x7fe55f0d39d0, FUTEX_WAIT, 5199, NULL) = 0
munmap(0x7fe55afd3000, 68161536) = 0
exit_group(0) = ?
+++ exited with 0 +++
We can see in the above traces:我们可以在上面的痕迹中看到:
mmap(NULL, 68161536, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0) = 0x7fe55afd3000
mprotect(0x7fe55afd4000, 68157440, PROT_READ|PROT_WRITE) = 0
Hence the new memory space layout:因此,新的 memory 空间布局:
+ +--------------------+ 0x7fe55afd3000
| | |
4 KB | | RED ZONE |
(PROT_NONE)| | |
+ +--------------------+ 0x7fe55afd4000
| | |
| | |
| | ^ |
66560 KB | | | |
(PROT_READ/WRITE) | Stack |
| | | |
| | | |
| +--------------------+ 0x7fe55f0d2fb0
| | |
| | TCB + TLS |
| | |
+ +--------------------+ 0x7FE55F0D4000
From a simple program ending into a strange crash, we took the opportunity to study the thread's stack layout in the GLIBC/pthread library as well as the protection mechanism against the stack overflows and the stack size configuration.从一个简单的程序结束到一个奇怪的崩溃,我们借此机会研究了GLIBC/pthread库中线程的堆栈布局以及堆栈溢出的保护机制和堆栈大小配置。
However, from a program design point of view , we should never allocate so huge variables in the stack.但是,从程序设计的角度来看,我们不应该在堆栈中分配这么大的变量。 In the current program, the table should be dynamically allocated or defined as a global variable (in Thread Local Storage) for examples.
在当前程序中,表应该是动态分配或定义为全局变量(在线程本地存储中)的例子。 But it is another story...
但这是另一个故事……
I generated core dump file.我生成了核心转储文件。 I ran the core dump file.
我运行了核心转储文件。 It gave me the following:
它给了我以下信息:
#0 0x00005643352ba745 in function (arg=<error reading variable: Cannot access memory at address 0x7fe80b054ed8>) at Pthred_kk.c:5
picture = <error reading variable picture (value requires 67108864 bytes, which is more than max-value-size)>
#1 0x00007fe80f6526db in start_thread (arg=0x7fe80f055700) at pthread_create.c:463
pd = 0x7fe80f055700
now = <optimized out>
unwind_buf = {cancel_jmp_buf = {{jmp_buf = {140634661148416, 8554578219241222147, 140634661146560, 0, 0, 140724934020640,
-8545604918547140605, -8545605192128745469}, mask_was_saved = 0}}, priv = {pad = {0x0, 0x0, 0x0, 0x0}, data = {prev = 0x0,
cleanup = 0x0, canceltype = 0}}}
not_first_call = <optimized out>
#2 0x00007fe80f37b88f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
picture = <error reading variable picture (value requires 67108864 bytes, which is more than max-value-size)
图片=<错误读取变量图片(值需要67108864字节,大于max-value-size)
In linux, the maxmum stack size for a thread is about 8MB.在 linux 中,线程的最大堆栈大小约为 8MB。
As you can see the size (67108864 bytes) of picture
is more than the maximum size (8MB = 8 * 1024 *1024 = 8388608).如您所见,
picture
的大小(67108864 字节)大于最大大小(8MB = 8 * 1024 *1024 = 8388608)。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.