138c67bf
From: grahn+nntp@snipabacken.se
On Sat, 2010-05-01, Prakash wrote:
> Hello,
>
> I am migrating the c application from Solaris 9 SPARC to RHEL 5.x
> x86_64 platform.
>
> Here are the platform details:
> -------------------------------------------
> 1. OS: Red Hat Linux 5.x, kernel 2.6.x
> 2. gcc compiler : 4.1.2.x V
>
> ioctl() call on Linux returning errno as EFAULT (Bad Address).
It's more precise than that. The Linux ioctl(2) man page says:
ERRORS
EBADF d is not a valid descriptor.
EFAULT argp references an inaccessible memory area.
EINVAL Request or argp is not valid.
...
here it means 'argp references an inaccessible memory area'.
> I have
> simulated the problem with below sample program. Can I know what is
> going wrong on Linux with this program.
>
> --------------
> /* Sample program to test the ioctl() call which sets the read mode to
> message-nondiscard mode.
> * For some reason this sample is not working on Linux x84_64, gcc :
> 4.1.x V and Red Hat 5.x.
> * However this code works on Solaris SPARC.
> */
>
> # include
> # include
> # include
> # include
> # include
> # include
> # include
> # include
> # include / macros I_SRDOPT and RMSGN are defined */
^^^^^^^
Hm -- apparently you haven't tried to compile this program. But that
seemed to be the only error.
>
>
> int
> main(){
>
> int fd[2];
>
> int fd_read;
> int fd_write;
>
> fd_read= fd_write = -1; /* Initila value */
>
> /* Create unnamed socket pairs */
> if ( socketpair(AF_UNIX, SOCK_STREAM,0,fd) == -1 ){
> fprintf(stdout, "socketpair call failed with errno: %d
> \n", errno);
> exit(1);
> }
>
> fd_read = fd[0];
> fd_write = fd[1];
>
> /* fcntl code comes here to set non blocking read */
>
> if ( ioctl(fd_read, I_SRDOPT, RMSGN ) == -1 ){ /* set read mode to
message -nondiscard mode */
> fprintf(stdout, "ioctl call failed with below error:\n" );
And as mr Magol wrote, there's perror() and strerror() for printing
errnos. (You don't even have to do that -- it's easier to run the
program under strace(1) and have it explain everything.)
> }else
> fprintf(stdout, "ioctl works!\n");
>
>
> return 0;
> }// main() ends
^^^^^^^^^^^^^^
Removing that to make it compile as C89.
>
> Thanks in Advance.
I_SRDOPT seems to be related to STREAMS. I cannot find it documented
in Stevens' "Unix network programming" books, and I have never used
STREAMS myself. But I know the Linux people generally dislike it, and
last time I heard it was /not/ part of Linux except as a kernel patch.
It is most likely /not/ in RedHat's kernel.
But the question is: if this is indeed STREAMS, and you don't have
that, then why are the ioctls defined in the header files? Perhaps
there are good reasons, but it's better when code that cannot possibly
work doesn't compile ...
Also, normally I expect 'gcc -std=c89 -Wall -W -pedantic ...' not to
compile unless I say "I want BSD APIs" or "I want POSIX this and
that". See features.h. But for these things, there was *no*
protection. That's surely a bug in the headers.
Someone else has to answer, but it seems to me you have a problem
here. The program you try to port uses Solaris-specific APIs. (OK,
some other Unices in the past may have had STREAMS too, but neither of
the healthy Unices -- Linux and the BSDs -- have it).
By the way, running the program under valgrind(1) is also enlightening.
Syscall param ioctl(CDROMREADTOCENTRY (cdte_format, char)) points to
unaddressable byte(s)
and so on.
/Jorgen
--
// Jorgen Grahn O o .
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
