In nptl/pthread_attr_setschedpolicy.c there are the lines

  /* Catch invalid values.  */
  if (policy != SCHED_OTHER && policy != SCHED_FIFO && policy != SCHED_RR)
    return EINVAL;

In pthread_setschedparam(), there is not an analogous check.

This means that one can set, for example, a SCHED_BATCH policy with 
pthread_setschedparam(), but one can't initialize a thread attributes object
that can be used to create a thread running under the SCHED_BATCH policy.  

This seems rather inconsistent.  Either pthread_setschedparam() should place the
same restriction as pthread_attr_setschedpolicy(), or the restriction should be
removed from pthread_attr_setschedpolicy().  Given that the POSIX.1
specification of pthread_attr_setschedpolicy() says:

    The  supported  values  of  policy  shall  include  SCHED_FIFO,
    SCHED_RR, and SCHED_OTHER, which are defined in  the  <sched.h>
    header.

removing the restriction from pthread_attr_setschedpolicy() seems the more
reasonable course, so that applications can take advantage of the additional
scheduling policies provided by Linux.