Linux进程之间的通信-内存共享（System V）

https://github.com/carloscn/blog/issues/16

Linux进程之间的通信-内存共享（System V）

• 信号量 (sem) ： 管理资源的访问

• 共享内存 (shm)： 高效的数据分享

• 消息队列 (msg)：在进程之间简易的传数据的方法

• 互斥算法（Dekker, Peterson, Filter, Szymanski, Lamport面包店算法)

IPC（Inter-Process Communication，进程间通讯）包含三种通信方式，信号量、共享内存和消息队列。在linux编程里面可以有两个不同的标准，一个是SYSTEM-V标准，一个是POSIX标准。以下是两个标准之间的区别。简单的说，POSIX更轻量，常面向于线程；SYSTEM-V更重一些，需要深陷Linux内核之中，面向于进程。

1. 内存共享

在大数据量的交换领域，使用memcpy是一个十分占据CPU和内存带宽的方法，共享内存作为Linux进程之间的通信是一个比较高效和节约资源的方法。现代多核心处理器，还有一个处理方法，比如对于视频的处理VPU，设定专门VDMA的RTL来处理。

1.1 原理

内存共享是随内核持续的，在两个毫无关系的进程中在userspace区域创建一个一块共享的内存，使两个进程都有读写的权限。既然是共享，必然涉及资源的竞争问题，故还要引入随内核持续的进程同步机制，来辅助访问共享的内存空间。如图所示，为进程的LMA在共享同一块share memory。

image-20220401121808422

既然是内存，这里就不得不提到，shm机制到底从哪里分配的内存，是heap还是stack。这里借用一个这个人的问题：

Understand-shmat-and-attachment-to-the-process-memory?

It is recommended/conventional that the second argument to shmat(int id , void * addr,int flg) should be NULL.

But if i want to give it a specific address (void* addr), should that address be from the stack or the heap?

I mean do i have to malloc() and then pass that address to shmat or i can just declare void * adrr(or char * addr) and pass it to shmat.

9ZHwD

If addr is NULL, the system selects the first available address without corrupting the BSS segment. Most probably that will be in the heap. So you don't need to allocate.

If addr is from the stack segment of your application, calling shmat will corrupt the stack. Most probably that will result in a segmentation fault in your program. shmat will overwrite the variables on the stack, located at addresses lower than the one you gave as parameter.

• 如果addr是NULL，系统紧凑选择不会覆盖BSS段展开的地址空间，最可能的是heap。

• 如果addr是一个栈地址，shmat将会覆盖栈空间，可能会造成段错误。

因此，shmat会绕开BSS展开的空间，紧凑地在HEAP上分配空间。

1.2 APIs

1.2.1 shmat, shmdt

shmat() attaches the System V shared memory segment identified by shmid to the address space of the calling process.

#include <sys/shm.h>
void *shmat(int shmid, const void *shmaddr, int shmflg);
int shmdt(const void *shmaddr);

Parameters:

Params	I/O	Details
int shmid	Input	由shmget返回的id值
const void *shmaddr	Input	指定共享内存连接到当前进程中的地址的位置。下面引用手册中的，可以看到shmaddr是NULL或者是其他情况。
int shmflg	Input	SHM_RND[与addr联合使用，用于控制共享内存的连接地址] 或者 SHM_RDONLY[连接的内存只读]

• If shmaddr is NULL, the system chooses a suitable (unused) page-aligned address to attach the segment.

• If shmaddr isn't NULL and SHM_RND is specified in shmflg, the attach occurs at the address equal to shmaddr rounded down to the nearest multiple of SHMLBA.

• Otherwise, shmaddr must be a page-aligned address at which the attach occurs.

Return:

On success, shmat() returns the address of the attached shared memory segment; on error, (void *) -1 is returned, and errno is set to indicate the error.

On success, shmdt() returns 0; on error -1 is returned, and errno is set to indicate the error.

Note:

• After a fork(2), the child inherits the attached shared memory segments.
• After an execve(2), all attached shared memory segments are detached from the process.
• Upon _exit(2), all attached shared memory segments are detached from the process.

1.2.2

shmget() returns the identifier of the System V shared memory segment associated with the value of the argument key. It may be used either to obtain the identifier of a previously created shared memory segment (when shmflg is zero and key does not have the value IPC_PRIVATE), or to create a new set.

A new shared memory segment, with size equal to the value of size rounded up to a multiple of PAGE_SIZE, is created if key has the value IPC_PRIVATE or key isn't IPC_PRIVATE, no shared memory segment corresponding to key exists, and IPC_CREAT is specified in shmflg.

If shmflg specifies both IPC_CREAT and IPC_EXCL and a shared memory segment already exists for key, then shmget() fails with errno set to EEXIST. (This is analogous to the effect of the combination O_CREAT | O_EXCL for open(2).)

#include <sys/shm.h>
int shmget(key_t key, size_t size, int shmflg);

Parameters:

Params	I/O	Details
key_t key	Input	提供一个key
size_t size	Input	需要共享内存的容量
int shmflg	Input	**IPC_CREAT **/IPC_EXCL /**SHM_HUGETLB **/SHM_HUGE_2MB /SHM_HUGE_1GB/ SHM_NORESERVE

• If shmaddr is NULL, the system chooses a suitable (unused) page-aligned address to attach the segment.

• If shmaddr isn't NULL and SHM_RND is specified in shmflg, the attach occurs at the address equal to shmaddr rounded down to the nearest multiple of SHMLBA.

• Otherwise, shmaddr must be a page-aligned address at which the attach occurs.

Return:

On success, a valid shared memory identifier is returned. On error, -1 is returned, and errno is set to indicate the error.

Note:

The following limits on shared memory segment resources affect the shmget() call:

• SHMALL，系统宽度限制总共的资源系统页大小统计。

• SHMMAX，共享最大段空间

• SHMMIN，共享最小段空间大小，1byte

具体参考文献

1.2.3 shmctl

shmctl() performs the control operation specified by cmd on the System V shared memory segment whose identifier is given in shmid.

#include <sys/shm.h>

struct shmid_ds {
    struct ipc_perm shm_perm;    /* Ownership and permissions */
    size_t          shm_segsz;   /* Size of segment (bytes) */
    time_t          shm_atime;   /* Last attach time */
    time_t          shm_dtime;   /* Last detach time */
    time_t          shm_ctime;   /* Creation time/time of last
                                    modification via shmctl() */
    pid_t           shm_cpid;    /* PID of creator */
    pid_t           shm_lpid;    /* PID of last shmat(2)/shmdt(2) */
    shmatt_t        shm_nattch;  /* No. of current attaches */
    ...
};
int shmctl(int shmid, int cmd, struct shmid_ds *buf);

Parameters:

Params	I/O	Details
int shmid	Input	get返回的shmid
int cmd	Input	采取的命令行动。 IPC_STAT，把shmid_ds结构中的数据设置为共享内存的当前关联值； IPC_SET，如果进程有足够的权限，就把共享内存的当前关联值设定为shmid_ds结构中给出的值；IPC_RMID，删除共享内存段。
struct shmid_ds *buf	Input	指向包含共享内存模式和访问权限的结构，具体参考

• If shmaddr is NULL, the system chooses a suitable (unused) page-aligned address to attach the segment.

• If shmaddr isn't NULL and SHM_RND is specified in shmflg, the attach occurs at the address equal to shmaddr rounded down to the nearest multiple of SHMLBA.

• Otherwise, shmaddr must be a page-aligned address at which the attach occurs.

Return:

0 成功， -1失败。

1.3 Example

test_shm_1.c: server,消耗数据client的数据

#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#include <sys/shm.h>
#include "test_shm.h"

#define debug_log printf("%s:%d--", __FUNCTION__, __LINE__);printf

int main(int argc, char *argv[])
{
    int i, ret;
    char op_chars[20];
    int count = 0;
    int shm_id = 0;
    int running = 1;

    void *share_memory = NULL;
    struct shared_use_st *share_stuff = NULL;

    debug_log("call the shmget function\n");
    shm_id = shmget((key_t)1234, sizeof(struct shared_use_st), 0666|IPC_CREAT);
    if (shm_id < 0) {
        debug_log("failed on semget\n");
        goto finish2;
    }

    debug_log("call the shmat\n");
    share_memory = shmat(shm_id, NULL, 0);
    if (share_memory == NULL - 1) {
        debug_log("failed on shmat\n");
        goto finish2;
    }
    debug_log("allocate memory attached at %x and main entry %x\n", (int)share_memory, (int)&main);
    share_stuff = (struct shared_use_st *)share_memory;
    share_stuff->written_by_you = 0;
    while(running) {
        if (share_stuff->written_by_you) {
            debug_log("You wrote: %s", share_stuff->some_text);
            sleep( rand() % 4);
            share_stuff->written_by_you = 0;
            if (strncmp(share_stuff->some_text, "end", 3) == 0) {
                running = 0;
            }
        }
    }
    debug_log("finish.....\n");
finish2:
    if (shmdt(share_memory) == -1) {
        debug_log("failed on share_memory\n");
        goto finish1;
    }
finish1:
    if (shmctl(shm_id, IPC_RMID, 0) == -1) {
        debug_log("failed on shmctl\n");
    }
    debug_log("finish test...\n");
    return ret;
}

test_shm_2.c: client,从stdin读入数据传送给server

#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#include <sys/shm.h>
#include "test_shm.h"

#define debug_log printf("%s:%d--", __FUNCTION__, __LINE__);printf

int main(int argc, char *argv[])
{
    int i, ret;
    char op_chars[20];
    int count = 0;
    int shm_id = 0;
    int running = 1;
    char buffer[BUFSIZ];

    void *share_memory = NULL;
    struct shared_use_st *share_stuff = NULL;

    debug_log("call the shmget function\n");
    shm_id = shmget((key_t)1234, sizeof(struct shared_use_st), 0666|IPC_CREAT);
    if (shm_id < 0) {
        debug_log("failed on semget\n");
        goto finish2;
    }

    debug_log("call the shmat\n");
    share_memory = shmat(shm_id, NULL, 0);
    if (share_memory == NULL - 1) {
        debug_log("failed on shmat\n");
        goto finish2;
    }
    debug_log("allocate memory attached at %x and main entry %x\n", (int*)share_memory, (int)&main);
    share_stuff = (struct shared_use_st *)share_memory;
    share_stuff->written_by_you = 0;
    while (running) {
        while (share_stuff->written_by_you == 1) {
            sleep(1);
            debug_log("waiting for client\n");
        }
        debug_log("Enter some test: ");
        fgets(buffer, BUFSIZ, stdin);
        strncpy(share_stuff->some_text, buffer, TEXT_SZ);
        share_stuff->written_by_you = 1;
        if (strncmp(buffer, "end", 3) == 0) {
            running = 0;
        }
    }
    debug_log("finish.....\n");
finish2:
    if (shmdt(share_memory) == -1) {
        debug_log("failed on share_memory\n");
    }
    debug_log("finish test...\n");
    return ret;
}

$ ./test_shm_1.elf

$ ./test_shm_2.elf

Ref

1. Linux Programmer's Manual - shmat, shmdt - System V shared memory operations
2. Linux Programmer's Manual - shmctl - System V shared memory control