[PATCH] fix (again) MAX_USER_RT_PRIO and MAX_RT_PRIO (was: MAX_USER_RT_PRIO and MAX_RT_PRIO are wrong!)

[PATCH] fix (again) MAX_USER_RT_PRIO and MAX_RT_PRIO (was: MAX_USER_RT_PRIO and MAX_RT_PRIO are wrong!)

[Steven Rostedt]

Andrew, 

As we discussed in OLS, here's the patch again to fix the code to handle
if the values between MAX_USER_RT_PRIO and MAX_RT_PRIO are different.
Without this patch, an SMP system will crash if the values are
different. I've already submitted this patch and it made it to -mm, but
never made it to mainline, so you told me to send it again and here it
is.

Also, shouldn't these values (MAX_RT_PRIO, MAX_USER_RT_PRIO) be user
configurable via a config parameter?  If so, I can send a separate patch
to do just that.

Ingo,

I've CC you just because you are the schedule maintainer.  You already
accepted this patch into your RT tree.

Dean,

I've CC you since it also has the change to
linux-2.6.13-rc3/arch/ia64/sn/kernel/xpc_main.c in it. But I don't see
this in -mm.  I don't have a ia64 so I can't test it. You tested this
for me before, so it should still work. This part should be at least
acknowledged by you.

-- Steve

PS. I'm currently running this patched kernel with MAX_RT_USER set to 95
and MAX_RT_PRIO set to 100 on an SMP machine.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

--- linux-2.6.13-rc3/kernel/sched.c.orig	2005-07-25 10:16:31.000000000 -0400
+++ linux-2.6.13-rc3/kernel/sched.c	2005-07-25 10:23:35.000000000 -0400
@@ -3486,7 +3486,7 @@ static void __setscheduler(struct task_s
 	p->policy = policy;
 	p->rt_priority = prio;
 	if (policy != SCHED_NORMAL)
-		p->prio = MAX_USER_RT_PRIO-1 - p->rt_priority;
+		p->prio = MAX_RT_PRIO-1 - p->rt_priority;
 	else
 		p->prio = p->static_prio;
 }
@@ -3518,7 +3518,8 @@ recheck:
 	 * 1..MAX_USER_RT_PRIO-1, valid priority for SCHED_NORMAL is 0.
 	 */
 	if (param->sched_priority < 0 ||
-	    param->sched_priority > MAX_USER_RT_PRIO-1)
+	    (p->mm &&  param->sched_priority > MAX_USER_RT_PRIO-1) ||
+	    (!p->mm && param->sched_priority > MAX_RT_PRIO-1))
 		return -EINVAL;
 	if ((policy == SCHED_NORMAL) != (param->sched_priority == 0))
 		return -EINVAL;
--- linux-2.6.13-rc3/arch/ia64/sn/kernel/xpc_main.c.orig	2005-07-25 10:23:22.000000000 -0400
+++ linux-2.6.13-rc3/arch/ia64/sn/kernel/xpc_main.c	2005-07-25 10:23:35.000000000 -0400
@@ -420,7 +420,7 @@ xpc_activating(void *__partid)
 	partid_t partid = (u64) __partid;
 	struct xpc_partition *part = &xpc_partitions[partid];
 	unsigned long irq_flags;
-	struct sched_param param = { sched_priority: MAX_USER_RT_PRIO - 1 };
+	struct sched_param param = { sched_priority: MAX_RT_PRIO - 1 };
 	int ret;
 

Re: [PATCH] fix (again) MAX_USER_RT_PRIO and MAX_RT_PRIO (was: MAX_USER_RT_PRIO and MAX_RT_PRIO are wrong!)

[Lee Revell]

On Mon, 2005-07-25 at 16:28 -0400, Steven Rostedt wrote:
> Ingo,
> 
> I've CC you just because you are the schedule maintainer.  You already
> accepted this patch into your RT tree.
> 

Please make sure to pick up this patch from Andreas Steinmetz too,
otherwise the rlimits are broken:

--- linux.orig/kernel/sched.c   2005-07-22 19:45:05.000000000 +0200
+++ linux/kernel/sched.c        2005-07-22 19:45:42.000000000 +0200
@@ -3528,7 +3528,8 @@
         */
        if (!capable(CAP_SYS_NICE)) {
                /* can't change policy */
-               if (policy != p->policy)
+               if (policy != p->policy &&
+                       !p->signal->rlim[RLIMIT_RTPRIO].rlim_cur)
                        return -EPERM;
                /* can't increase priority */
                if (policy != SCHED_NORMAL &&

Lee

Re: [PATCH] fix (again) MAX_USER_RT_PRIO and MAX_RT_PRIO (was: MAX_USER_RT_PRIO and MAX_RT_PRIO are wrong!)

[Dean Nelson]

Steve, your change to XPC looks good. Thanks, Dean

Signed-off-by: Dean Nelson <dcn@sgi.com>


On Mon, Jul 25, 2005 at 04:28:39PM -0400, Steven Rostedt wrote:
> Dean,
> 
> I've CC you since it also has the change to
> linux-2.6.13-rc3/arch/ia64/sn/kernel/xpc_main.c in it. But I don't see
> this in -mm.  I don't have a ia64 so I can't test it. You tested this
> for me before, so it should still work. This part should be at least
> acknowledged by you.
> 
> -- Steve
> 
> PS. I'm currently running this patched kernel with MAX_RT_USER set to 95
> and MAX_RT_PRIO set to 100 on an SMP machine.
> 
> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
> 
> --- linux-2.6.13-rc3/kernel/sched.c.orig	2005-07-25 10:16:31.000000000 -0400
> +++ linux-2.6.13-rc3/kernel/sched.c	2005-07-25 10:23:35.000000000 -0400
> @@ -3486,7 +3486,7 @@ static void __setscheduler(struct task_s
>  	p->policy = policy;
>  	p->rt_priority = prio;
>  	if (policy != SCHED_NORMAL)
> -		p->prio = MAX_USER_RT_PRIO-1 - p->rt_priority;
> +		p->prio = MAX_RT_PRIO-1 - p->rt_priority;
>  	else
>  		p->prio = p->static_prio;
>  }
> @@ -3518,7 +3518,8 @@ recheck:
>  	 * 1..MAX_USER_RT_PRIO-1, valid priority for SCHED_NORMAL is 0.
>  	 */
>  	if (param->sched_priority < 0 ||
> -	    param->sched_priority > MAX_USER_RT_PRIO-1)
> +	    (p->mm &&  param->sched_priority > MAX_USER_RT_PRIO-1) ||
> +	    (!p->mm && param->sched_priority > MAX_RT_PRIO-1))
>  		return -EINVAL;
>  	if ((policy == SCHED_NORMAL) != (param->sched_priority == 0))
>  		return -EINVAL;
> --- linux-2.6.13-rc3/arch/ia64/sn/kernel/xpc_main.c.orig	2005-07-25 10:23:22.000000000 -0400
> +++ linux-2.6.13-rc3/arch/ia64/sn/kernel/xpc_main.c	2005-07-25 10:23:35.000000000 -0400
> @@ -420,7 +420,7 @@ xpc_activating(void *__partid)
>  	partid_t partid = (u64) __partid;
>  	struct xpc_partition *part = &xpc_partitions[partid];
>  	unsigned long irq_flags;
> -	struct sched_param param = { sched_priority: MAX_USER_RT_PRIO - 1 };
> +	struct sched_param param = { sched_priority: MAX_RT_PRIO - 1 };
>  	int ret;
>  
> 
>