Re: [SSSD] [PATCH] sss_cache: Remove fastcache even if sssd is not running.

[Michal Židek]

On 11/06/2012 09:45 PM, Sumit Bose wrote:
> Hi,
> > +errno_t sss_br_lock_file(int fd, size_t start, size_t len,
> >                     int retries, useconds_t wait)
> >
> > +{
> > int ret;
> > struct flock lock;
> > int retries_left;
> > lock.l_type = F_WRLCK;
> > lock.l_whence = SEEK_SET;
> > lock.l_start = start;
> > lock.l_len = len;
> > lock.l_pid = 0;
> > for (retries_left = retries; retries_left > 0; retries_left--) {
> >    ret = fcntl(fd, F_SETLK, &lock);
> >    if (ret == -1) {
> >        ret = errno;
> >        if (ret == EACCES || ret == EAGAIN || ret == EINTR) {
> >            DEBUG(SSSDBG_TRACE_FUNC,
> >                  ("Failed to lock file. Retries left: %d\n",
> >                   retries_left - 1));
> >            if ((ret == EACCES || ret == EAGAIN) && (retries_left <= 1)) {
> >                /* File is locked by someone else. Return EACCESS
> >                 * if this is the last try. */
> >                return EACCES;
> >            }
> >            if (retries_left - 1 > 0) {
> >                ret = usleep(wait);
> >                if (ret == -1) {
> >                    DEBUG(SSSDBG_MINOR_FAILURE,
> >                          ("usleep() failed -> ignoring\n"));
> >                }
> >            }
> >        } else {
> >            /* Error occurred */
> >            DEBUG(SSSDBG_CRIT_FAILURE,
> >                  ("Unable to lock file.\n"));
> >            return ret;
> >        }
> >    } else if (ret == 0) {
> >        /* File successfuly locked */
> >        break;
> >    }
> > }
> > if (retries_left == 0) {
> >    DEBUG(SSSDBG_CRIT_FAILURE, ("Unable to lock file.\n"));
> >    return ret;
>
> if retries is 0, ret is uninitialized. Can you fix this? All credits
> goes to gcc, I've just seen the compiler warning.
>
> bye,
> Sumit
> > }
> > return EOK;
> >
> > +}
>
> sssd-devel mailing list
> [e-mail]
> https://lists.fedorahosted.org/mailman/listinfo/sssd-devel

Good catch. Having a value of 0 or <= 0 makes no sense here. I'll send a 
patch that returns EINVAL in such situation in a new thread.

Thanks
Michal