Signals

Khaj.pandey · 06-18-2010, 06:42 AM

Code:

 1  static volatile sig_atomic_t sigreceived = 0;
 2
 3  sigset_t maskall, maskmost, maskold;
 4  int signum = SIGUSR1;
 5
 6  sigfillset(&maskall);
 7  sigfillset(&maskmost);
 8  sigdelset(&maskmost, signum);
 9  sigprocmask(SIG_SETMASK, &maskall, &maskold);
10  if (sigreceived == 0)
11     sigsuspend(&maskmost);
12  sigprocmask(SIG_SETMASK, &maskold, NULL);

Consider the above code. Text i have says that this is the correct way to wait for SIGUSR1 signal , but i am unable to wrap my head around it.

so maskall -> all signals , maskmost-> all except sigusr1

Can some one explain what does this do?

sigprocmask(SIG_SETMASK, &maskall, &maskold);

and why is this :

sigsuspend(&maskmost);

Shouldn't the mask of sigsuspend contain only signal SIGUSR1?

Cheers!

JohnGraham · 06-18-2010, 07:20 AM

Code:

sigprocmask(SIG_SETMASK, &maskall, &maskold);

This will set the current mask to maskall, and store the old masks in maskold so they can be restored later.

Have you tried reading the man pages? You might want to try 'apt-get install manpages manpages-dev' on Debian/Ubuntu, not sure about other distros but then there will be something similar. Then 'man sigprocmask' should give you all the info you need on that particular function (and the others in your example.

Khaj.pandey · 06-18-2010, 08:52 AM

Thanks jhon, that does clear up a couple of things. I got a little confused bw sigsuspend & sigwait.

I have another doubt though , if sigsuspend restores the mask, why is this required ?

sigprocmask(SIG_SETMASK, &maskold, NULL);

ForzaItalia2006 · 06-18-2010, 12:10 PM

Quote:

Originally Posted by Khaj.pandey

I have another doubt though , if sigsuspend restores the mask, why is this required ?

sigprocmask(SIG_SETMASK, &maskold, NULL);

First of all, the signal mask holds al those signals which should be blocked by the application. Though, by using maskmost, you would block all signals except SIGUSR1. For more details, please check the man pages as suggested by John or the POSIX rationale/spec which contains very useful information about how signals are handled, when signals are blocked, when they are just suspended, etc.

Regarding your second question, why this second sigprocmask(2) is required: please verify your code that maskmost != maskall != maskold. Though the last statement just restores the maskold state while sigsuspend restores from maskmost to maskall.

Andi

Khaj.pandey · 06-18-2010, 01:18 PM

Thanks forza , i think i got what is happening now... Early on all signals are blocked , then all signals except sigusr1 is blocked, on receipt the state is restored to the last configuration.

This is a way to prevent a signal being missed from the looks of it.

I am now going to hijack this in a new direction -> Sockets!

When speaking of file descriptors, as i understand it is nothing more than an integer. This is an excerpt from Beej's guide to n/w programming. Stripped for clarity.

Code:

select(maxfd+1, &readfd,NULL,NULL,NULL)

for(int i =0;;i<maxfd ; i++)
 if(FD_ISSET(i,&readfd))

first up , there is no ordering between the fd's returned by the OS if i understand correctly. So read fd could have fd's like 10,24,28,32. Does this make sense to iterate all they way from 0 to maxfd??

PS: My unix box is dead atm , hence i am asking some silly stuff which would be clearer to program .

Cheers!

ForzaItalia2006 · 06-18-2010, 04:24 PM

You didn't really ask any explicit question, though I'm just commenting some of your statements :-)

Quote:

Originally Posted by Khaj.pandey

When speaking of file descriptors, as i understand it is nothing more than an integer. This is an excerpt from Beej's guide to n/w programming. Stripped for clarity.

Yes, file descriptors are integer values

Quote:

Originally Posted by Khaj.pandey

first up , there is no ordering between the fd's returned by the OS if i understand correctly. So read fd could have fd's like 10,24,28,32. Does this make sense to iterate all they way from 0 to maxfd??

Well, it depends on what you understand by ordering. The file/socket descriptors returned by the OS are in a specific order. The OS returns the lowest available socket descriptor, starting at 3 (due to stdin, stdout, stderr) for a newly created program. For example, if you would close stdin before creating a new socket, you'll get a socket id of 0.

Andi

paulsm4 · 06-18-2010, 07:40 PM

Quote:

Q: So read fd could have fd's like 10,24,28,32. Does this make sense to iterate all they way from 0 to maxfd??

Code:

// WRONG
  select(maxfd+1, &readfd,NULL,NULL,NULL)

  for(int i =0;;i<maxfd ; i++)
    if(FD_ISSET(i,&readfd))

Code:

// CORRECT
int sock_ready(int sock) 
{ 
    int             res; 
    fd_set          sready; 
    struct timeval  nowait; 

    FD_ZERO(&sready); 
    FD_SET((unsigned int)sock,&sready); 
    /*bzero((char *)&nowait,sizeof(nowait));*/ 
    memset((char *)&nowait,0,sizeof(nowait)); 

    res = select(sock+1,&sready,NULL,NULL,&nowait); 
    if( FD_ISSET(sock,&sready) ) 
        res = 1; 
    else 
        res = 0; 

    return(res); 
}

So in answer to your question:
1. Yes, it can make sense to iterate 0 .. maxfd

2. It usually makes MORE sense to check a specific list of descriptors

3. *Whatever* you do, you should *always* declare and zero out "sready" each and every time you call "select()"

'Hope that helps .. PSM

Khaj.pandey · 06-19-2010, 07:54 AM

Thanks guys!

As i understand, the complete data might not be sent in one go by the send command... i.e. which pretty much means, the complete data is not recovered at the other end.... Any mechanisms to handle this ? i.e. if one was to write a structure and send it (only half is sent), and only half is received then how is the same handled?

From the top of my head , i would say , allocating memory using malloc , copying as much as you can in it.... doing a memcpy of the address returned via malloc + data read and the new offset. Obviously all this with another send by te sender.

I theory this sounds nice, not sure if it actually works.