Re: pam_limits and Linux 2.4

[Pavel Kankovsky <peak argo troja mff cuni cz>]

On Sun, 13 May 2001, Andrew Morgan wrote:

> Off hand, it looks like you've hit this:
> 
>     for(i = 0; i < RLIM_NLIMITS; i++) 
>         retval |= getrlimit(i, &pl->limits[i].limit);
> 
> Now, if you compiled this module with the headers for 2.4, but used it
> with a 2.2 kernel where RLIM_NLIMITS might not be the same, I could see
> that this might fail. Could you check that this is so?

Yes, RLIM_NLIMITS is 11 in 2.4 but only 10 in 2.2.
Nevertheless, getrlimit() should return EINVAL iff its first argument is
incorrect (i.e. has a value not supported by the running kernel). This
means the following code could solve the problem:

  char supported[RLIM_NLIMITS];
  ...
  for (i = 0; i < RLIM_NLIMITS; ++i) {
    int r = getrlimit(i, ...);
    supported[i] = 1;
    if (r == -1 && errno == EINVAL) {
      supported[i] = 0;
      r = 0;
    }
    retval |= r;
  }
  ...
  for (i = 0; i < RLIM_NLIMITS; ++i) {
    if (!supported[i])
      continue;
    ...
    ... = setrlimit(i, ...);
    ...
  }

--Pavel Kankovsky aka Peak  [ Boycott Microsoft--http://www.vcnet.com/bms ]
"Resistance is futile. Open your source code and prepare for assimilation."