Advanced Operating Systems: Lecture 37 - Mr. Farhan Zaidi

CS703 ­ Advanced 
Operating Systems
By Mr. Farhan Zaidi


Lecture No. 
37


Overview of today’s lecture







Interrupt handlers
Interrupts and exceptions
Linux interrupt handling
Top halfs, bottom halfs and tasklets
Timings and timer devices
Linux kernel timers and interval timers


Interrupt Handlers 


Interrupt handlers are best hidden



Interrupt procedure does its task


Interrupts vs Exceptions


Varying terminology but for Intel:


Interrupt (device generated)
 Maskable: device-generated, associated with IRQs (interrupt
request lines); may be temporarily disabled (still pending)
 Nonmaskable: some critical hardware failures



Exceptions (Processor-detected)
 Faults – correctable (restartable); e.g. page fault
 Traps – no reexecution needed; e.g. breakpoint
 Aborts – severe error; process usually terminated (by
signal)
 Programmed exceptions (software interrupts)
 int (system call), int3 (breakpoint)
 into (overflow), bounds (address check)


Interrupts and Exceptions


Hardware support for getting CPUs attention






Often transfers from user to kernel mode
Asynchronous: device or timer generated
Synchronous: immediate result of last instruction

Intel terminology and hardware


Irqs, vectors, IDT, gates, PIC, APIC


Interrupt Handling


More complex than exceptions




Some issues:




Requires registry, deferred processing, etc.
IRQs are often shared; all handlers (ISRs) are executed so they
must query device

Three types of actions:




Critical: Top-half (interrupts disabled – briefly!)
Non-critical: Top-half (interrupts enabled)
Non-critical deferrable: Do it “later” (interrupts enabled)


Interrupt Handling Components
IRQs

vector

Memory Bus

0

PIC

INTR

CPU
idtr

0

IDT

15

Mask points
255

handler


Timing and Timers


Accurate timing crucial for many aspects of OS








Intel timer hardware







Device-related timeouts
File timestamps (created, accessed, written)
Time-of-day (gettimeofday()
High-precision timers (code profiling, etc.)
Scheduling, cpu usage, etc.
RTC: Real Time Clock
PIT: Programmable Interrupt Timer
TSC: TimeStamp Counter (cycle counter)
Local APIC Timer: per-cpu alarms

Timer implementations



Kernel timers (dynamic timers)
User “interval” timers (alarm(), setitimer())




Dynamic timers may be dynamically created and destroyed. No limit is
placed on the number of currently active dynamic timers.



A dynamic timer is stored in the following timer_list structure:

struct timer_list

{
struct list_head list;
unsigned long expires;
unsigned long data;
void (*function)(unsigned long);
};


Software timers in Linux


Each timer contains a field that indicates how far in the future the timer
should expire. This field is initially calculated by adding the right number
of ticks to the current value of jiffies.



The field does not change. Every time the kernel checks a timer, it
compares the expiration field to the value of jiffies at the current
moment, and the timer expires when jiffies is greater or equal to the
stored value.


User Mode Interval Timers


setitimer(): 3 distinct user mode interval timers





three timers:






can set for one-time or periodic alarm
sends signals on timer expiry
ITIMER_REAL: elapsed time (SIGALRM)
ITIMER_VIRT: user (SIGVTALRM)
ITIMER_PROF: user + kernel (SIGPROF)

implementations:



VIRT, PROF – updates by PIT or APIC
REAL – requires kernel timer
 may need to deliver to blocked process
 current->real_timer
 shared by alarm() API so can’t use both