Re: [PATCH 6/6] ipc/sem.c: make semctl(,,{GETNCNT,GETZCNT}) standard compliant

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 14 May 2014 07:52:38 -0700 Davidlohr Bueso <davidlohr@hp.com> wrote:

> > --- a/ipc/sem.c
> > +++ b/ipc/sem.c
> > @@ -993,38 +993,33 @@ static void do_smart_update(struct sem_array *sma, struct sembuf *sops, int nsop
> >  }
> >  
> >  /*
> > - * check_qop: Test how often a queued operation sleeps on the semaphore semnum
> > + * check_qop: Test if a queued operation sleeps on the semaphore semnum
> >   */
> >  static int check_qop(struct sem_array *sma, int semnum, struct sem_queue *q,
> >  			bool count_zero)
> >  {
> > -	struct sembuf *sops = q->sops;
> > -	int nsops = q->nsops;
> > -	int i, semcnt;
> > +	struct sembuf *sop = q->blocking;
> >  
> > -	semcnt = 0;
> > +	BUG_ON(sop->sem_flg & IPC_NOWAIT);
> > +	BUG_ON(sop->sem_op > 0);
> 
> Hmm in light of Linus' recent criticism about randomly sprinkling
> BUG_ONs in the kernel I'm not sure we want this. Yes, all those calls
> are correct from a logical pov and should never occur, however, would
> WARN be more suitable instead? I don't know. 

Well, this BUG_ON is so old that a decent approach would be to just
delete the thing, if only Manfred wasn't changing stuff.

Yes, if we can reasonably warn-then-recover then I guess that's worth
doing.