Recitation 8:Signals & ShellsAndrew Faulring15213 Section A28 October 2002Andrew Faulring• [email protected]• Office hours:– NSH 2504 (lab) / 2507 (conference room)– Thursday 5-6•Lab 5– due Thursday, 31 Oct @ 11:59pm•Halloween Night … happy reaping!Today’s Plan• Process IDs & Process Groups• Process Control•Signals• Preemptive Scheduler–Race hazards• Reaping Child ProcessesLab 5: Shell• tshref– Use as a guide for output– You shell should have same behaviorHow Programmers Play with Processes• Process: executing copy of program• Basic functions– fork() spawns new process– exit() terminates calling process– wait() and waitpid() wait for and reap terminated children– execl() and execve() run a new program in an existing processProcess IDs & Process Groups• Each process has its own, unique process ID– pid_t getpid();• Each process belongs to exactly one process group– pid_t getpgid();• To which process group does a new process initially belong?– Its parent’s process group• A process can make a process group for itself and its children– setpgid(0, 0);Fore-groundjobBack-groundjob #1Back-groundjob #2ShellChild Childpid=10pgid=10Foregroundprocess group 20Backgroundprocess group 32Backgroudprocess group 40pid=20pgid=20pid=32pgid=32pid=40pgid=40pid=21pgid=20pid=22pgid=20Signals•Section 8.5 in text– Read at least twice … really!• A signal tells our program that some event has occurred– For instance, a child process has terminated• Can we use signals to count events?–NoImportant Signals•SIGINT– Interrupt signal from keyboard (ctrl-c)• SIGTSTP– Stop signal from keyboard (ctrl-z)•SIGCHLD– A child process has stopped or terminatedLook at Figure 8.23 for a complete list of Linux signalsSending a Signal• Send a signal– Sent by either the kernel– Or another process• Why is a signal sent?– The kernel detects a system event.• Divide-by-zero (SIGFPE)• Termination of a child process (SIGCHLD)– Another process invokes a system call.• kill(pid_t pid, int SIGINT)– kill(1500, SIGINT)» Send SIGINT to process 1500– kill(-1500, SIGINT)» Send SIGINT to progress group 1500• alarm(unsigned int secs)Receiving a Signal• Default action– The process terminates [and dumps core]– The process stops until restarted by a SIGCONT signal– The process ignore the signal• Can modify the default action with the signal function– Additional action: “Handle the signal”• void sigint_handler(int sig);• signal(SIGINT, sigint_handler);– Cannot modify action for SIGSTOP and SIGKILLReceiving a Signal• pending: bit vector: bit k is set when signal type kis delivered, clear when signal received• blocked: bit vector of signals that should not be received• Only receive non-blocked, pending signals– pending & ~blockedSynchronizing Processes• Preemptive scheduler run multiple programs “concurrently” by time slicing– How does time slicing work? – The scheduler can stop a program at any point– Signal handler code can run at any point, too• Program behaviors depend on how the scheduler interleaves the execution of processes• Racing condition between parent and child!–Why? Race Hazard• Different behaviors of program depending upon how the schedule interleaves the execution of code.Parent & Child Race Hazardsigchld_handler() {pid = waitpid(…);deletejob(pid);}eval() {pid = fork();if(pid == 0){ /* child */execve(…);}/* parent *//* signal handler might run BEFORE addjob() */addjob(…);}ShellSignal Handler Childfork()addjob()execve()exit()sigchld_handler()deletejobs()timeAn Okay ScheduleShellSignal Handler Childfork()execve()exit()sigchld_handler()deletejobs()timeaddjob()Job added to job list afterthe signal handler tried to delete it!A Problematic Schedule Solution to Race Hazardsigchld_handler() {pid = waitpid(…);deletejob(pid);}eval() {sigprocmask(SIG_BLOCK, …)pid = fork();if(pid == 0){ /* child */sigprocmask(SIG_UNBLOCK, …)execve(…);}/* parent *//* signal handler might run BEFORE addjob() */addjob(…);sigprocmask(SIG_UNBLOCK, …)}More details 8.5.6 (page 633)Reaping Child Process• Child process becomes zombie when terminates– Still consume system resources– Parent performs reaping on terminated child• Using either wait or waitpid syscall• Where to wait children processes to terminate?–Two waits• sigchld_handler• eval: for foreground processes–One wait• sigchld_handler• But what about foreground processes?Busy Waitvoid eval() {…/* parent */addjob(…);while(fg process still alive){;}}sigchld_handler() {pid = waitpid(…);deletejob(pid);}Pausevoid eval() {…/* parent */addjob(…);while(fg process still alive){pause();}}sigchld_handler() {pid = waitpid(…);deletejob(pid);}If signal handled (SIGCHLD) before call to pause, then pause will not returnSleepvoid eval() {…/* parent */addjob(…);while(fg process still alive){sleep(1);}}sigchld_handler() {pid = waitpid(…);deletejob(pid);}waitpid• Used for reaping zombied child processes•pid_t waitpid(pid_t pid, int *status, intoptions)– pid: wait until child process with pid has terminated• -1: wait for any child process– status: tells why child terminated– options:• WNOHANG: return immediately if no children have exited (zombied)– waitpid returns -1• WUNTRACED: report status of stopped children toowaitpid’s status• int status;waitpid(pid,&status, NULL)• WIFEXITED(status): child exited normally– WEXITSTATUS(status): return code when child exits• WIFSIGNALED(status): child exited because a signal was not caught– WTERMSIG(status): gives the number of the terminating signal• WIFSTOPPED(status): child is stopped– WSTOPSIG(status): gives the number of the stop signalSummary• Process provides applications with the illusions of: – Exclusively use of the processor and the main memory• At the interface with OS, applications can:– Creating child processes– Run new programs– Catch signals from other processes•Use man if anything is not
View Full Document
Unlocking...