Re: [RFC][PATCH] sched: SCHED_FIFO watchdog timer
On 10/14/07, Peter Zijlstra <[EMAIL PROTECTED]> wrote: > The below patch is an idea proposed by tglx and depends on sched-devel + > the hrtick patch previously posted. > > The current watchdog action is to demote the task to SCHED_NORMAL, > however it might be wanted to deliver a signal instead (or have more per > task configuration state). Which is why I added Lennart to the CC list > as I gathered he would like something like this for PulseAudio. Great, this looks very promising. Thanks for doing this. Kay - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC][PATCH] sched: SCHED_FIFO watchdog timer
On Sun, 14.10.07 00:51, Peter Zijlstra ([EMAIL PROTECTED]) wrote: > The below patch is an idea proposed by tglx and depends on sched-devel + > the hrtick patch previously posted. > > The current watchdog action is to demote the task to SCHED_NORMAL, > however it might be wanted to deliver a signal instead (or have more per > task configuration state). Which is why I added Lennart to the CC list > as I gathered he would like something like this for PulseAudio. Indeed! Having this in the kernel would allow us to enable RT scheduling for PulseAudio by default without bad effects. I was thinking about adding some kind of babysitting process to userspace -- but doing this as an RLIMIT in the kernel strikes me a much better idea! I think it would make a lot of sense to make the API very similar to RLIMIT_CPU, i.e. also send out SIGXCPU and SIGKILL, with the single difference that RLIMIT_CPU sends out a signal depending on the total CPU time used for the process and the new RLIMIT based on the time the process spent without sleeping. That would be a very reasonable extension to the current RLIMIT_CPU model. Thank you very much for doing this patch! Lennart -- Lennart PoetteringRed Hat, Inc. lennart [at] poettering [dot] net ICQ# 11060553 http://0pointer.net/lennart/ GnuPG 0x1A015CC4 - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC][PATCH] sched: SCHED_FIFO watchdog timer
On Mon, 2007-10-15 at 15:26 +0200, Dmitry Adamushko wrote: > On 14/10/2007, Peter Zijlstra <[EMAIL PROTECTED]> wrote: > > The below patch is an idea proposed by tglx and depends on sched-devel + > > the hrtick patch previously posted. > > > > The current watchdog action is to demote the task to SCHED_NORMAL, > > however it might be wanted to deliver a signal instead (or have more per > > task configuration state). Which is why I added Lennart to the CC list > > as I gathered he would like something like this for PulseAudio. > > > > --- > > Subject: sched: SCHED_FIFO watchdog timer > > Why only SHCED_FIFO and not SCHED_RR? > Their (mis)behavior is similar wrt SCHED_NORMAL tasks. Because SCHED_FIFO is easier, _RR is for later. It was mostly an RFC to request behavioural wishes from the users. > > > +#ifdef CONFIG_SCHED_HRT_TICK > > +static int fifo_watchdog(struct rq *rq, struct task_struct *p, int queued) > > +{ > > + if (likely(!queued || p->policy != SCHED_FIFO)) > > + return 0; > > + > > + /* > > +* task has been naughty, turn into SCHED_NORMAL > > +*/ > > + printk(KERN_INFO "SCHED_FIFO task %s/%d exceeded his runtime quota," > > + " demoting to regular task\n", p->comm, > > task_pid_nr(p)); > > + deactivate_task(rq, p, 0); > > + __setscheduler(rq, p, SCHED_NORMAL, 0); > > + activate_task(rq, p, 0); > > + resched_task(p); > > I guess, put_prev_task() / set_curr_task() should be called (for the > case of task_running(p)) to make it group-scheduler-friendly (as it's > done e.g. in sched_setscheduler()). > > (normilize_task() should probably do the same) Right, that is where I copied from, I'll pull the functionality into a single function and make this and the sysrq stuff use it. Thanks! signature.asc Description: This is a digitally signed message part
Re: [RFC][PATCH] sched: SCHED_FIFO watchdog timer
On 14/10/2007, Peter Zijlstra <[EMAIL PROTECTED]> wrote: > The below patch is an idea proposed by tglx and depends on sched-devel + > the hrtick patch previously posted. > > The current watchdog action is to demote the task to SCHED_NORMAL, > however it might be wanted to deliver a signal instead (or have more per > task configuration state). Which is why I added Lennart to the CC list > as I gathered he would like something like this for PulseAudio. > > --- > Subject: sched: SCHED_FIFO watchdog timer Why only SHCED_FIFO and not SCHED_RR? Their (mis)behavior is similar wrt SCHED_NORMAL tasks. > +#ifdef CONFIG_SCHED_HRT_TICK > +static int fifo_watchdog(struct rq *rq, struct task_struct *p, int queued) > +{ > + if (likely(!queued || p->policy != SCHED_FIFO)) > + return 0; > + > + /* > +* task has been naughty, turn into SCHED_NORMAL > +*/ > + printk(KERN_INFO "SCHED_FIFO task %s/%d exceeded his runtime quota," > + " demoting to regular task\n", p->comm, > task_pid_nr(p)); > + deactivate_task(rq, p, 0); > + __setscheduler(rq, p, SCHED_NORMAL, 0); > + activate_task(rq, p, 0); > + resched_task(p); I guess, put_prev_task() / set_curr_task() should be called (for the case of task_running(p)) to make it group-scheduler-friendly (as it's done e.g. in sched_setscheduler()). (normilize_task() should probably do the same) -- Best regards, Dmitry Adamushko - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC][PATCH] sched: SCHED_FIFO watchdog timer
On 14/10/2007, Peter Zijlstra [EMAIL PROTECTED] wrote: The below patch is an idea proposed by tglx and depends on sched-devel + the hrtick patch previously posted. The current watchdog action is to demote the task to SCHED_NORMAL, however it might be wanted to deliver a signal instead (or have more per task configuration state). Which is why I added Lennart to the CC list as I gathered he would like something like this for PulseAudio. --- Subject: sched: SCHED_FIFO watchdog timer Why only SHCED_FIFO and not SCHED_RR? Their (mis)behavior is similar wrt SCHED_NORMAL tasks. +#ifdef CONFIG_SCHED_HRT_TICK +static int fifo_watchdog(struct rq *rq, struct task_struct *p, int queued) +{ + if (likely(!queued || p-policy != SCHED_FIFO)) + return 0; + + /* +* task has been naughty, turn into SCHED_NORMAL +*/ + printk(KERN_INFO SCHED_FIFO task %s/%d exceeded his runtime quota, +demoting to regular task\n, p-comm, task_pid_nr(p)); + deactivate_task(rq, p, 0); + __setscheduler(rq, p, SCHED_NORMAL, 0); + activate_task(rq, p, 0); + resched_task(p); I guess, put_prev_task() / set_curr_task() should be called (for the case of task_running(p)) to make it group-scheduler-friendly (as it's done e.g. in sched_setscheduler()). (normilize_task() should probably do the same) -- Best regards, Dmitry Adamushko - To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC][PATCH] sched: SCHED_FIFO watchdog timer
On Mon, 2007-10-15 at 15:26 +0200, Dmitry Adamushko wrote: On 14/10/2007, Peter Zijlstra [EMAIL PROTECTED] wrote: The below patch is an idea proposed by tglx and depends on sched-devel + the hrtick patch previously posted. The current watchdog action is to demote the task to SCHED_NORMAL, however it might be wanted to deliver a signal instead (or have more per task configuration state). Which is why I added Lennart to the CC list as I gathered he would like something like this for PulseAudio. --- Subject: sched: SCHED_FIFO watchdog timer Why only SHCED_FIFO and not SCHED_RR? Their (mis)behavior is similar wrt SCHED_NORMAL tasks. Because SCHED_FIFO is easier, _RR is for later. It was mostly an RFC to request behavioural wishes from the users. +#ifdef CONFIG_SCHED_HRT_TICK +static int fifo_watchdog(struct rq *rq, struct task_struct *p, int queued) +{ + if (likely(!queued || p-policy != SCHED_FIFO)) + return 0; + + /* +* task has been naughty, turn into SCHED_NORMAL +*/ + printk(KERN_INFO SCHED_FIFO task %s/%d exceeded his runtime quota, +demoting to regular task\n, p-comm, task_pid_nr(p)); + deactivate_task(rq, p, 0); + __setscheduler(rq, p, SCHED_NORMAL, 0); + activate_task(rq, p, 0); + resched_task(p); I guess, put_prev_task() / set_curr_task() should be called (for the case of task_running(p)) to make it group-scheduler-friendly (as it's done e.g. in sched_setscheduler()). (normilize_task() should probably do the same) Right, that is where I copied from, I'll pull the functionality into a single function and make this and the sysrq stuff use it. Thanks! signature.asc Description: This is a digitally signed message part
Re: [RFC][PATCH] sched: SCHED_FIFO watchdog timer
On Sun, 14.10.07 00:51, Peter Zijlstra ([EMAIL PROTECTED]) wrote: The below patch is an idea proposed by tglx and depends on sched-devel + the hrtick patch previously posted. The current watchdog action is to demote the task to SCHED_NORMAL, however it might be wanted to deliver a signal instead (or have more per task configuration state). Which is why I added Lennart to the CC list as I gathered he would like something like this for PulseAudio. Indeed! Having this in the kernel would allow us to enable RT scheduling for PulseAudio by default without bad effects. I was thinking about adding some kind of babysitting process to userspace -- but doing this as an RLIMIT in the kernel strikes me a much better idea! I think it would make a lot of sense to make the API very similar to RLIMIT_CPU, i.e. also send out SIGXCPU and SIGKILL, with the single difference that RLIMIT_CPU sends out a signal depending on the total CPU time used for the process and the new RLIMIT based on the time the process spent without sleeping. That would be a very reasonable extension to the current RLIMIT_CPU model. Thank you very much for doing this patch! Lennart -- Lennart PoetteringRed Hat, Inc. lennart [at] poettering [dot] net ICQ# 11060553 http://0pointer.net/lennart/ GnuPG 0x1A015CC4 - To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [RFC][PATCH] sched: SCHED_FIFO watchdog timer
On 10/14/07, Peter Zijlstra [EMAIL PROTECTED] wrote: The below patch is an idea proposed by tglx and depends on sched-devel + the hrtick patch previously posted. The current watchdog action is to demote the task to SCHED_NORMAL, however it might be wanted to deliver a signal instead (or have more per task configuration state). Which is why I added Lennart to the CC list as I gathered he would like something like this for PulseAudio. Great, this looks very promising. Thanks for doing this. Kay - To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[RFC][PATCH] sched: SCHED_FIFO watchdog timer
The below patch is an idea proposed by tglx and depends on sched-devel + the hrtick patch previously posted. The current watchdog action is to demote the task to SCHED_NORMAL, however it might be wanted to deliver a signal instead (or have more per task configuration state). Which is why I added Lennart to the CC list as I gathered he would like something like this for PulseAudio. --- Subject: sched: SCHED_FIFO watchdog timer Set a per task (rlimit based) limit on SCHED_FIFO runtime. When the limit is exceeded the task is demoted back to SCHED_NORMAL. Signed-off-by: Peter Zijlstra <[EMAIL PROTECTED]> --- include/asm-generic/resource.h |5 +++-- kernel/sched.c |5 + kernel/sched_rt.c | 36 3 files changed, 44 insertions(+), 2 deletions(-) Index: linux-2.6/include/asm-generic/resource.h === --- linux-2.6.orig/include/asm-generic/resource.h +++ linux-2.6/include/asm-generic/resource.h @@ -44,8 +44,8 @@ #define RLIMIT_NICE13 /* max nice prio allowed to raise to 0-39 for nice level 19 .. -20 */ #define RLIMIT_RTPRIO 14 /* maximum realtime priority */ - -#define RLIM_NLIMITS 15 +#define RLIMIT_FIFOTIME15 /* timeout for fifo slices in us */ +#define RLIM_NLIMITS 16 /* * SuS says limits have to be unsigned. @@ -86,6 +86,7 @@ [RLIMIT_MSGQUEUE] = { MQ_BYTES_MAX, MQ_BYTES_MAX }, \ [RLIMIT_NICE] = { 0, 0 }, \ [RLIMIT_RTPRIO] = { 0, 0 }, \ + [RLIMIT_FIFOTIME] = { RLIM_INFINITY, RLIM_INFINITY }, \ } #endif /* __KERNEL__ */ Index: linux-2.6/kernel/sched_rt.c === --- linux-2.6.orig/kernel/sched_rt.c +++ linux-2.6/kernel/sched_rt.c @@ -89,6 +89,14 @@ static struct task_struct *pick_next_tas next->se.exec_start = rq->clock; + if (next->policy == SCHED_FIFO) { + unsigned long fifotime; + + fifotime = rq->curr->signal->rlim[RLIMIT_FIFOTIME].rlim_cur; + if (fifotime != RLIM_INFINITY) + hrtick_start(rq, (u64)fifotime * 1000, 0); + } + return next; } @@ -194,8 +202,36 @@ load_balance_rt(struct rq *this_rq, int return load_moved; } +#ifdef CONFIG_SCHED_HRT_TICK +static int fifo_watchdog(struct rq *rq, struct task_struct *p, int queued) +{ + if (likely(!queued || p->policy != SCHED_FIFO)) + return 0; + + /* +* task has been naughty, turn into SCHED_NORMAL +*/ + printk(KERN_INFO "SCHED_FIFO task %s/%d exceeded his runtime quota," + " demoting to regular task\n", p->comm, task_pid_nr(p)); + deactivate_task(rq, p, 0); + __setscheduler(rq, p, SCHED_NORMAL, 0); + activate_task(rq, p, 0); + resched_task(p); + + return 1; +} +#else +static inline int fifo_watchdog(struct rq *rq, struct task_struct *p, int queued) +{ + return 0; +} +#endif + static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued) { + if (fifo_watchdog(rq, p, queued)) + return; + /* * RR tasks need a special form of timeslice management. * FIFO tasks have no timeslices. Index: linux-2.6/kernel/sched.c === --- linux-2.6.orig/kernel/sched.c +++ linux-2.6/kernel/sched.c @@ -132,6 +132,11 @@ static inline void sg_inc_cpu_power(stru } #endif +static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep); +static void activate_task(struct rq *rq, struct task_struct *p, int wakeup); +static void +__setscheduler(struct rq *rq, struct task_struct *p, int policy, int prio); + static inline int rt_policy(int policy) { if (unlikely(policy == SCHED_FIFO) || unlikely(policy == SCHED_RR)) - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[RFC][PATCH] sched: SCHED_FIFO watchdog timer
The below patch is an idea proposed by tglx and depends on sched-devel + the hrtick patch previously posted. The current watchdog action is to demote the task to SCHED_NORMAL, however it might be wanted to deliver a signal instead (or have more per task configuration state). Which is why I added Lennart to the CC list as I gathered he would like something like this for PulseAudio. --- Subject: sched: SCHED_FIFO watchdog timer Set a per task (rlimit based) limit on SCHED_FIFO runtime. When the limit is exceeded the task is demoted back to SCHED_NORMAL. Signed-off-by: Peter Zijlstra [EMAIL PROTECTED] --- include/asm-generic/resource.h |5 +++-- kernel/sched.c |5 + kernel/sched_rt.c | 36 3 files changed, 44 insertions(+), 2 deletions(-) Index: linux-2.6/include/asm-generic/resource.h === --- linux-2.6.orig/include/asm-generic/resource.h +++ linux-2.6/include/asm-generic/resource.h @@ -44,8 +44,8 @@ #define RLIMIT_NICE13 /* max nice prio allowed to raise to 0-39 for nice level 19 .. -20 */ #define RLIMIT_RTPRIO 14 /* maximum realtime priority */ - -#define RLIM_NLIMITS 15 +#define RLIMIT_FIFOTIME15 /* timeout for fifo slices in us */ +#define RLIM_NLIMITS 16 /* * SuS says limits have to be unsigned. @@ -86,6 +86,7 @@ [RLIMIT_MSGQUEUE] = { MQ_BYTES_MAX, MQ_BYTES_MAX }, \ [RLIMIT_NICE] = { 0, 0 }, \ [RLIMIT_RTPRIO] = { 0, 0 }, \ + [RLIMIT_FIFOTIME] = { RLIM_INFINITY, RLIM_INFINITY }, \ } #endif /* __KERNEL__ */ Index: linux-2.6/kernel/sched_rt.c === --- linux-2.6.orig/kernel/sched_rt.c +++ linux-2.6/kernel/sched_rt.c @@ -89,6 +89,14 @@ static struct task_struct *pick_next_tas next-se.exec_start = rq-clock; + if (next-policy == SCHED_FIFO) { + unsigned long fifotime; + + fifotime = rq-curr-signal-rlim[RLIMIT_FIFOTIME].rlim_cur; + if (fifotime != RLIM_INFINITY) + hrtick_start(rq, (u64)fifotime * 1000, 0); + } + return next; } @@ -194,8 +202,36 @@ load_balance_rt(struct rq *this_rq, int return load_moved; } +#ifdef CONFIG_SCHED_HRT_TICK +static int fifo_watchdog(struct rq *rq, struct task_struct *p, int queued) +{ + if (likely(!queued || p-policy != SCHED_FIFO)) + return 0; + + /* +* task has been naughty, turn into SCHED_NORMAL +*/ + printk(KERN_INFO SCHED_FIFO task %s/%d exceeded his runtime quota, +demoting to regular task\n, p-comm, task_pid_nr(p)); + deactivate_task(rq, p, 0); + __setscheduler(rq, p, SCHED_NORMAL, 0); + activate_task(rq, p, 0); + resched_task(p); + + return 1; +} +#else +static inline int fifo_watchdog(struct rq *rq, struct task_struct *p, int queued) +{ + return 0; +} +#endif + static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued) { + if (fifo_watchdog(rq, p, queued)) + return; + /* * RR tasks need a special form of timeslice management. * FIFO tasks have no timeslices. Index: linux-2.6/kernel/sched.c === --- linux-2.6.orig/kernel/sched.c +++ linux-2.6/kernel/sched.c @@ -132,6 +132,11 @@ static inline void sg_inc_cpu_power(stru } #endif +static void deactivate_task(struct rq *rq, struct task_struct *p, int sleep); +static void activate_task(struct rq *rq, struct task_struct *p, int wakeup); +static void +__setscheduler(struct rq *rq, struct task_struct *p, int policy, int prio); + static inline int rt_policy(int policy) { if (unlikely(policy == SCHED_FIFO) || unlikely(policy == SCHED_RR)) - To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/