How to use IPC msg Q's, msgget etc. ??

[Kevin Swift]

I am trying to use IPC message queues. On HPLX but they do not seem to work.

msgget seems to return a handle but if I try to use this in any way I get EACCES or EPERM.

ipcs(8) also reports nothing.

Is there some configuration I need to make to set permissions for message queues?

[Ron Vladick]

What compartment are you in when you are trying to use msgget, msgrcv?

Are you trying to talk from Compartment to Compartment? if so what are your rules?

[Kevin Swift]

Here's a bit more detail.

I have a new compartment

IPC is between processes in the same compartment.

I wrote a little test program which works fine on other linux but not on HP-LX. (attached)

Is there anything I need to configure for IPC? I found no mention in the man pages.

On other linux the test program reports:-

mid is 196613 and errno is 0
msgctl/STAT return is 0 and errno is 0
msgctl/SET return is 0 and errno is 0
msgsnd return is 0 and errno is 0


and ipcs -q reports

key msqid owner perms used-bytes messages
0x00000000 229382 omu104 660 10 1

on HP-LX I just get.
mid is 163845 and errno is 0
msgctl/STAT return is -1 and errno is 13
msgctl/SET return is -1 and errno is 1
msgsnd return is -1 and errno is 13

and ipcs never reports anything.


Thanks,


Kevin

[Ron Vladick · Accepted solution]

Kevin,

Thanks for the attacthment.

I was able to run this program successfully, here were my steps.

1) create a new compartment
/sbin/tlcompadd

2) change into this compartment
/sbin/tlsetcomp

3) no rules will be required if you stay in the same comparment.

BTW I did get the errors if I tried to run this program from the "system" compartment.

Here are my results and an strace of the success:

[tlinuxadm@viper tlinuxadm]$ /sbin/tlsetcomp msg1
sh-2.04$ ls
a.out message.c
sh-2.04$ ./a.out
mid is 294919 and errno is 0
msgctl/STAT return is 0 and errno is 0
msgctl/SET return is 0 and errno is 0
msgsnd return is 0 and errno is 0

sh-2.04$ ipcs -q
------ Message Queues --------
key msqid owner perms used-bytes messages
0x0000000c 131074 tlinuxadm 660 0 0
0x00000000 294919 104 660 10 1

sh-2.04$ strace ./a.out

execve("./a.out", ["./a.out"], [/* 24 vars */]) = 0

uname({sys="Linux",

node="viper.idev.atl.hp.com", ...}) = 0

brk(0) = 0x8049954

old_mmap(NULL, 4096, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x40017000

open("/etc/ld.so.preload", O_RDONLY) = -1 ENOENT (No such file or directory)

open("/etc/ld.so.cache", O_RDONLY) = 3

fstat64(3, {st_mode=S_IFREG|0644, st_size=32402, ...}) = 0

old_mmap(NULL, 32402, PROT_READ, MAP_PRIVATE, 3, 0) = 0x40018000

close(3) = 0

open("/lib/i686/libc.so.6", O_RDONLY) = 3

read(3, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\200\302"..., 1024) = 1024

fstat64(3, {st_mode=S_IFREG|0755,
st_size=5634864, ...}) = 0

old_mmap(NULL, 1242920, PROT_READ|PROT_EXEC,

MAP_PRIVATE, 3, 0) = 0x40020000

mprotect(0x40146000, 38696, PROT_NONE) = 0

old_mmap(0x40146000, 24576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 3, 0x125000) = 0x40146000

old_mmap(0x4014c000, 14120, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x4014c000

close(3) = 0

munmap(0x40018000, 32402) = 0

getpid() = 1556

msgget(IPC_PRIVATE, 0x1b0|0660) = 327688

fstat64(1, {st_mode=S_IFCHR|0620,

st_rdev=makedev(136, 1), ...}) = 0

mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x40018000

ioctl(1, TCGETS, {B38400 opost isig icanon echo ...}) = 0

write(1, "mid is 327688 and errno is 0\n", 29mid is 327688 and errno is 0
) = 29

msgctl(327688, MSG_???, 0xbffff200) = 0

write(1, "msgctl/STAT return is 0 and errn"..., 39msgctl/STAT return is 0 and errno is 0
) = 39

msgctl(327688, MSG_???, 0xbffff200) = 0

write(1, "msgctl/SET return is 0 and errno"..., 38msgctl/SET return is 0 and errno is 0
) = 38

msgsnd(327688, {1, "123456789\0"}, 10, 0) = 0

write(1, "msgsnd return is 0 and errno is "..., 34msgsnd return is 0 and errno is 0
) = 34

munmap(0x40018000, 4096) = 0
_exit(34) = ?
sh-2.04$

[Kevin Swift]

Thanks for that, It seems that all my testing happend in the system compartment, clearly not a good idea!


Kevin