Michal Mazurek wrote:
> Depending on the NOLOCK flag, the KERNEL_LOCK will, or will not be acquired
> before calling pledge_syscall(). I therefore conclude that KERNEL_LOCK
> is not needed to call pledge_syscall().
>
> Also remove the goto. The code is simple enough to avoid it easily.
>
> I think this complexity was caused by some reshuffling during tame/pledge
> deployment.
I think some of the code in pledge_syscall was also more complex. Now it's
only a lookup into a static table, so I think this is ok.
>
> Index: sys/sys/syscall_mi.h
> ===================================================================
> RCS file: /cvs/src/sys/sys/syscall_mi.h,v
> retrieving revision 1.15
> diff -u -p -r1.15 syscall_mi.h
> --- sys/sys/syscall_mi.h 3 Nov 2015 16:14:14 -0000 1.15
> +++ sys/sys/syscall_mi.h 7 Apr 2016 11:01:08 -0000
> @@ -69,29 +69,24 @@ mi_syscall(struct proc *p, register_t co
> }
> #endif
>
> - if (lock)
> - KERNEL_LOCK();
> pledged = (p->p_p->ps_flags & PS_PLEDGE);
> if (pledged && (error = pledge_syscall(p, code, &tval))) {
> - if (!lock)
> - KERNEL_LOCK();
> + KERNEL_LOCK();
> error = pledge_fail(p, error, tval);
> KERNEL_UNLOCK();
> return (error);
> }
> #if NSYSTRACE > 0
> if (ISSET(p->p_flag, P_SYSTRACE)) {
> - if (!lock)
> - KERNEL_LOCK();
> + KERNEL_LOCK();
> error = systrace_redirect(code, p, argp, retval);
> - lock = 1;
> - goto done;
> + KERNEL_UNLOCK();
> + return (error);
> }
> #endif
> + if (lock)
> + KERNEL_LOCK();
> error = (*callp->sy_call)(p, argp, retval);
> -#if NSYSTRACE > 0
> -done:
> -#endif
> if (lock)
> KERNEL_UNLOCK();
>
>
> --
> Michal Mazurek
>
