while going through linux kernel code inside __scheduler() function I saw hrtick_clear(rq).
Can anyone explain what is this and why it is used?
it seems something related to timer, but unable to proceed further.
Classic OS design involves system timer - an entity that ticks at fixed intervals. During each tick, scheduler is called and if process/thread should be switched. But system timer frequency is pretty low (i.e. 1000 HZ, which means once in 1 ms), and if process have only 100us of its timeslice left, it will get extra time (under certain circumstances), while other processes are starve.
However, modern CPUs provide more precision hardware timers like HPET on Intel, which are provided by hrtimers subsystem. They can be enabled for be used in scheduler by CONFIG_SCHED_HRTICK option.
But if you already called __schedule() (i.e. on path of system call), you do not need to call it second time from hrtimer, because you already scheduling, so before doing so, hrtick_clear disables that hrtimer.
Related
I have this code:
let k = time::Instant::now();
thread::sleep(time::Duration::from_micros(10));
let elapsed = k.elapsed().as_micros();
println!("{}", elapsed);
My output is always somewhere between 70 and 90. I expect it to be 10, why is this number 7x higher?
This actually doesn't really have anything to do with Rust.
On a typical multi-processing, user-interactive operating system (i.e., every consumer OS you've used), your thread isn't special. It's one among many, and the CPUs need to be shared.
You operating system has a component called a scheduler, whose job it is to share the hardware resources. It will boot off your thread off the CPU quite often. This typically happens:
On every system call
Every time an interrupt hits the CPU
When the scheduler kicks you off to give other processes/threads a chance (this is called preemption, and typically happens 10s of times a second)
Thus, your userland process can't possibly do anything timing-related with such fine precision.
There's several solution paths you can explore:
Increase the amount of CPU your operating system gives you. Some ideas:
Increase the process' priortiy
Pin the thread to a particular CPU core, to give it exclusive use (this means you lose throughput, because if your thread is idle, no other thread's work can borrow that CPU)
Switch to a real-time operating system which makes guarantees about latency and timing.
Offload the work to some hardware that's specialized to do with, without the involvement of your process.
E.g. offload sine wave generation to a hardware sound-card, WiFi radio processing to a radio controller, etc.
Use your own micro controller to do the real-time stuff, and communicate to it over something like I2C or SPI.
In your case of running some simple code on a userland process, I think your easiest bet is to just pin your process. Your existing code will work as-is, you'll just lose the throughput of one of your cores (but luckily, you haven multiple).
Is it possible to execute tasks on a Linux host with microsecond precision? I.e., I'd like to execute a task at a specific instant of time. I know, Linux is no real-time system but I'm searching for the best solution on Linux.
So far, I've created a kernel module, setup hrtimer and measured the jitter when the callback function is entered (I don't really care too much about the actual delay, it's jitter that counts) - it's about 20-50us. That's not significantly better than using timerfd in userspace (also tried using real-time priority for the process but that did not really change anything).
I'm running Linux 3.5.0 (just an example, tried different kernels from 2.6.35 to 3.7), /proc/timer_list shows hrtimer_interrupt, I'm not running in failsafe mode which disables hrtimer functionality. Tried on different CPUs (Intel Atom to Core i7).
My best idea so far would be using hrtimer in combination with ndelay/udelay. Is this really the best way to do it? I can't believe it's not possible to trigger a task with microsecond precision. Running the code in kernel space as module is acceptable, would be great if the code was not interrupted by other tasks though. I dont' really care too much about the rest of the system, the task will be executed only very few times a second so using mdelay/ndelay for burning the CPU for some microseconds every time the task should be executed would not really matter. Altough, I'd prefer a more elegent solution.
I hope the question is clear, found a lot of topics concerning timer precision but no real answer to that problem.
You can do what you want from user space
use clock_gettime() with CLOCK_REALTIME to get the time-of-day with nano-second resolution
use nanosleep() to yield the CPU until you are close to the time you need to execute your task (it is at least milli-second resolution).
use a spin loop with clock_gettime() until you reach the desired time
execute your task
The clock_gettime() function is implemented as a VDSO in recent kernels and modern x86 processors - it takes 20-30 nanoseconds to get the time-of-day with nano-second resolution - you should be able to call clock_gettime() over 30 times per micro-second. Using this method your task should dispatch within 1/30th of a micro-second of the intended time.
The default Linux kernel timer ticks each millisecond. Microseconds is way beyond anything current user hardware is capable of.
The jitter you see is due to a host of factors, like interrupt handling and servicing higher priority tasks. You can cut that down somewhat by selecting hardware carefully, only enabling what is really needed. The real-time patchseries to the kernel (see the HOWTO) might be an option to reduce it a bit further.
Always keep in mind that any gain has a definite cost in terms of interactiveness, stability, and (last, but by far not least) your time in building, tuning, troubleshooting, and keeping the house of cards from falling apart.
I need a thread in my process to wakeup every 5ms(precise) and do some work.
I have used posix timers, they seems to be accurate 90% and accuracy further decreases when cpu is somewhat loaded.
I believe that is because posix timer have to fork new thread on every expiry.
Is there some other reliable way to implement high resolution timer in linux and will increasing priority of thread help?
I am on CentOS 5.6.
The POSIX timers (created with timer_create()) are already high-resolution. Your problem is in the delivery method - if you want very precise timing then SIGEV_THREAD is not a good idea.
You could instead use SIGEV_SIGNAL so that timer expiry is notified via a signal, then use sigwaitinfo() to wait for it to expire. Alternately, you could use a timerfd instead of a POSIX timer (created with timerfd_create()).
Additionally, if you want your thread to preempt other running threads when the timer expires, you'll need to give it a real-time scheduling policy (SCHED_FIFO or SCHED_RR) with sched_setscheduler().
You will also want to ensure that your kernel is compiled with the CONFIG_PREEMPT option, which allows most kernel code to be preemptable. There will still be some level of jitter, caused by non-preemptible kernel work like hardware interrupts and softirqs. To reduce this further, you can try using the CONFIG_PREEMPT_RT kernel patchset.
I recently started to learn how the CPU and the operating system works, and I am a bit confused about the operation of a single-CPU machine with an operating system that provides multitasking.
Supposing my machine has a single CPU, this would mean that, at any given time, only one process could be running.
Now, I can only assume that the scheduler used by the operating system to control the access to the precious CPU time is also a process.
Thus, in this machine, either the user process or the scheduling system process is running at any given point in time, but not both.
So here's a question:
Once the scheduler gives up control of the CPU to another process, how can it regain CPU time to run itself again to do its scheduling work? I mean, if any given process currently running does not yield the CPU, how could the scheduler itself ever run again and ensure proper multitasking?
So far, I had been thinking, well, if the user process requests an I/O operation through a system call, then in the system call we could ensure the scheduler is allocated some CPU time again. But I am not even sure if this works in this way.
On the other hand, if the user process in question were inherently CPU-bound, then, from this point of view, it could run forever, never letting other processes, not even the scheduler run again.
Supposing time-sliced scheduling, I have no idea how the scheduler could slice the time for the execution of another process when it is not even running?
I would really appreciate any insight or references that you can provide in this regard.
The OS sets up a hardware timer (Programmable interval timer or PIT) that generates an interrupt every N milliseconds. That interrupt is delivered to the kernel and user-code is interrupted.
It works like any other hardware interrupt. For example your disk will force a switch to the kernel when it has completed an IO.
Google "interrupts". Interrupts are at the centre of multithreading, preemptive kernels like Linux/Windows. With no interrupts, the OS will never do anything.
While investigating/learning, try to ignore any explanations that mention "timer interrupt", "round-robin" and "time-slice", or "quantum" in the first paragraph – they are dangerously misleading, if not actually wrong.
Interrupts, in OS terms, come in two flavours:
Hardware interrupts – those initiated by an actual hardware signal from a peripheral device. These can happen at (nearly) any time and switch execution from whatever thread might be running to code in a driver.
Software interrupts – those initiated by OS calls from currently running threads.
Either interrupt may request the scheduler to make threads that were waiting ready/running or cause threads that were waiting/running to be preempted.
The most important interrupts are those hardware interrupts from peripheral drivers – those that make threads ready that were waiting on IO from disks, NIC cards, mice, keyboards, USB etc. The overriding reason for using preemptive kernels, and all the problems of locking, synchronization, signaling etc., is that such systems have very good IO performance because hardware peripherals can rapidly make threads ready/running that were waiting for data from that hardware, without any latency resulting from threads that do not yield, or waiting for a periodic timer reschedule.
The hardware timer interrupt that causes periodic scheduling runs is important because many system calls have timeouts in case, say, a response from a peripheral takes longer than it should.
On multicore systems the OS has an interprocessor driver that can cause a hardware interrupt on other cores, allowing the OS to interrupt/schedule/dispatch threads onto multiple cores.
On seriously overloaded boxes, or those running CPU-intensive apps (a small minority), the OS can use the periodic timer interrupts, and the resulting scheduling, to cycle through a set of ready threads that is larger than the number of available cores, and allow each a share of available CPU resources. On most systems this happens rarely and is of little importance.
Every time I see "quantum", "give up the remainder of their time-slice", "round-robin" and similar, I just cringe...
To complement #usr's answer, quoting from Understanding the Linux Kernel:
The schedule( ) Function
schedule( ) implements the scheduler. Its objective is to find a
process in the runqueue list and then assign the CPU to it. It is
invoked, directly or in a lazy way, by several kernel routines.
[...]
Lazy invocation
The scheduler can also be invoked in a lazy way by setting the
need_resched field of current [process] to 1. Since a check on the value of this
field is always made before resuming the execution of a User Mode
process (see the section "Returning from Interrupts and Exceptions" in
Chapter 4), schedule( ) will definitely be invoked at some close
future time.
In a simple experiment I set NOHZ=OFF and used printk() to print how often the do_timer() function gets called. It gets called every 10 ms on my machine.
However if NOHZ=ON then there is a lot of jitter in the way do_timer() gets called. Most of the times it does get called every 10 ms but there are times when it completely misses the deadlines.
I have researched about both do_timer() and NOHZ. do_timer() is the function responsible for updating jiffies value and is also responsible for the round robin scheduling of the processes.
NOHZ feature switches off the hi-res timers on the system.
What I am unable to understand is how can hi-res timers affect the do_timer()? Even if hi-res hardware is in sleep state the persistent clock is more than capable to execute do_timer() every 10 ms. Secondly if do_timer() is not executing when it should, that means some processes are not getting their timeshare when they should ideally be getting it. A lot of googling does show that for many people many applications start working much better when NOHZ=OFF.
To make long story short, how does NOHZ=ON affect do_timer()?
Why does do_timer() miss its deadlines?
First lets understand what is a tickless kernel ( NOHZ=On or CONFIG_NO_HZ set ) and what was the motivation of introducing it into the Linux Kernel from 2.6.17
From http://www.lesswatts.org/projects/tickless/index.php,
Traditionally, the Linux kernel used a periodic timer for each CPU.
This timer did a variety of things, such as process accounting,
scheduler load balancing, and maintaining per-CPU timer events. Older
Linux kernels used a timer with a frequency of 100Hz (100 timer events
per second or one event every 10ms), while newer kernels use 250Hz
(250 events per second or one event every 4ms) or 1000Hz (1000 events
per second or one event every 1ms).
This periodic timer event is often called "the timer tick". The timer
tick is simple in its design, but has a significant drawback: the
timer tick happens periodically, irrespective of the processor state,
whether it's idle or busy. If the processor is idle, it has to wake up
from its power saving sleep state every 1, 4, or 10 milliseconds. This
costs quite a bit of energy, consuming battery life in laptops and
causing unnecessary power consumption in servers.
With "tickless idle", the Linux kernel has eliminated this periodic
timer tick when the CPU is idle. This allows the CPU to remain in
power saving states for a longer period of time, reducing the overall
system power consumption.
So reducing power consumption was one of the main motivations of the tickless kernel. But as it goes, most of the times, Performance takes a hit with decreased power consumption. For desktop computers, performance is of utmost concern and hence you see that for most of them NOHZ=OFF works pretty well.
In Ingo Molnar's own words
The tickless kernel feature (CONFIG_NO_HZ) enables 'on-demand' timer
interrupts: if there is no timer to be expired for say 1.5 seconds
when the system goes idle, then the system will stay totally idle for
1.5 seconds. This should bring cooler CPUs and power savings: on our (x86) testboxes we have measured the effective IRQ rate to go from HZ
to 1-2 timer interrupts per second.
Now, lets try to answer your queries-
What I am unable to understand is how can hi-res timers affect the
do_timer ?
If a system supports high-res timers, timer interrupts can occur more frequently than the usual 10ms on most systems. i.e these timers try to make the system more responsive by leveraging the system capabilities and by firing timer interrupts even faster, say every 100us. So with NOHZ option, these timers are cooled down and hence the lower execution of do_timer
Even if hi-res hardware is in sleep state the persistent clock is more
than capable to execute do_timer every 10ms
Yes it is capable. But the intention of NOHZ is exactly the opposite. To prevent frequent timer interrupts!
Secondly if do_timer is not executing when it should that means some
processes are not getting their timeshare when they should ideally be
getting it
As caf noted in the comments, NOHZ does not cause processes to get scheduled less often, because it only kicks in when the CPU is idle - in other words, when no processes are schedulable. Only the process accounting stuff will be done at a delayed time.
Why does do_timer miss it's deadlines ?
As elaborated, it is the intended design of NOHZ
I suggest you go through the tick-sched.c kernel sources as a starting point. Search for CONFIG_NO_HZ and try understanding the new functionality added for the NOHZ feature
Here is one test performed to measure the Impact of a Tickless Kernel