SynchronizationDecember 6, 2006Topics Shared variables The need for synchronization Synchronizing with semaphores Thread safety and reentrancy Races and deadlocksclass26.ppt15-213“The course that gives CMU its Zip!”– 2 –15-213, F’06Shared Variables in Threaded C ProgramsQuestion: Which variables in a threaded C program are shared variables? The answer is not as simple as “global variables are shared”and “stack variables are private”.Requires answers to the following questions: What is the memory model for threads? How are variables mapped to memory instances? How many threads reference each of these instances?– 3 –15-213, F’06Threads Memory ModelConceptual model: Multiple threads run within the context of a single process. Each thread has its own separate thread contextz Thread ID, stack, stack pointer, program counter, condition codes, and general purpose registers. All threads share the remaining process context.z Code, data, heap, and shared library segments of the process virtual address spacez Open files and installed handlersOperationally, this model is not strictly enforced: While register values are truly separate and protected.... Any thread can read and write the stack of any other thread. Mismatch between the conceptual and operation model is a source of confusion and errors.– 4 –15-213, F’06Example of Threads Accessing Another Thread’s Stackchar **ptr; /* global */int main(){int i;pthread_t tid;char *msgs[N] = {"Hello from foo","Hello from bar"};ptr = msgs;for (i = 0; i < 2; i++)Pthread_create(&tid, NULL, thread, (void *)i);Pthread_exit(NULL);}/* thread routine */void *thread(void *vargp){int myid = (int) vargp;static int svar = 0;printf("[%d]: %s (svar=%d)\n", myid, ptr[myid], ++svar);}Peer threads access main thread’s stackindirectly through global ptr variable– 5 –15-213, F’06Mapping Variables to Mem. Instanceschar **ptr; /* global */int main(){int i;pthread_t tid;char *msgs[N] = {"Hello from foo","Hello from bar"};ptr = msgs;for (i = 0; i < 2; i++)Pthread_create(&tid, NULL, thread, (void *)i);Pthread_exit(NULL);}/* thread routine */void *thread(void *vargp){int myid = (int)vargp;static int svar = 0;printf("[%d]: %s (svar=%d)\n", myid, ptr[myid], ++svar);}Global var: 1 instance (ptr [data])Local static var: 1 instance (svar [data])Local automatic vars: 1 instance (i.m, msgs.m )Local automatic var: 2 instances (myid.p0[peer thread 0’s stack],myid.p1[peer thread 1’s stack])– 6 –15-213, F’06Shared Variable AnalysisWhich variables are shared?Variable Referenced by Referenced by Referenced byinstance main thread? peer thread 0? peer thread 1?ptr yes yes yessvar no yes yesi.m yes no nomsgs.m yes yes yesmyid.p0 no yes nomyid.p1 no no yesAnswer: A variable x is shared iff multiple threads reference at least one instance of x. Thus: ptr, svar, and msgs are shared. i and myid are NOT shared.– 7 –15-213, F’06badcnt.c: An Improperly Synchronized Threaded Program/* shared */volatile unsigned int cnt = 0;#define NITERS 100000000 int main() {pthread_t tid1, tid2;Pthread_create(&tid1, NULL, count, NULL);Pthread_create(&tid2, NULL, count, NULL);Pthread_join(tid1, NULL);Pthread_join(tid2, NULL);if (cnt != (unsigned)NITERS*2)printf("BOOM! cnt=%d\n", cnt);elseprintf("OK cnt=%d\n", cnt);}/* thread routine */void *count(void *arg) {int i;for (i=0; i<NITERS; i++)cnt++;return NULL;}linux> ./badcntBOOM! cnt=198841183linux> ./badcntBOOM! cnt=198261801linux> ./badcntBOOM! cnt=198269672cnt should beequal to 200,000,000. What went wrong?!– 8 –15-213, F’06Assembly Code for Counter Loop.L9:movl -4(%ebp),%eaxcmpl $99999999,%eaxjle .L12jmp .L10.L12:movl cnt,%eax # Loadleal 1(%eax),%edx # Updatemovl %edx,cnt # Store.L11:movl -4(%ebp),%eaxleal 1(%eax),%edxmovl %edx,-4(%ebp)jmp .L9.L10:Corresponding asm code for (i=0; i<NITERS; i++)cnt++;C code for counter loopHead (Hi)Tail (Ti)Load cnt (Li)Update cnt (Ui)Store cnt (Si)– 9 –15-213, F’06Concurrent ExecutionKey idea: In general, any sequentially consistent interleaving is possible, but some are incorrect! Iidenotes that thread i executes instruction I %eaxiis the contents of %eax in thread i’s contextH1L1U1S1H2L2U2S2T2T11111222221-011-----10001111222i (thread)instricnt%eax1OK-----1222-%eax2– 10 –15-213, F’06Concurrent Execution (cont)Incorrect ordering: two threads increment the counter, but the result is 1 instead of 2.H1L1U1H2L2S1T1U2S2T21112211222-01--11---0000011111i (thread)instricnt%eax1----0--111%eax2Oops!– 11 –15-213, F’06Concurrent Execution (cont)How about this ordering?H1L1H2L2U2S2U1S1T1T21122221112i (thread)instricnt%eax1%eax2We can clarify our understanding of concurrentexecution with the help of the progress graph– 12 –15-213, F’06Beware of Optimizing Compilers! Global variable cnt shared between threads Multiple threads could be trying to update within their iterations Compiler moved access to cnt out of loop Only shared accesses to cntoccur before loop (read) or after (write) What are possible program outcomes?#define NITERS 100000000/* shared counter variable */unsigned int cnt = 0;/* thread routine */void *count(void *arg) {int i;for (i = 0; i < NITERS; i++)cnt++;return NULL;}Code From Bookmovl cnt, %ecxmovl $99999999, %eax.L6:leal 1(%ecx), %edxdecl %eaxmovl %edx, %ecxjns .L6movl %edx, cntGenerated Code– 13 –15-213, F’06Controlling Optimizing Compilers! Declaring variable as volatile forces it to be kept in memory Shared variable read and written each iteration#define NITERS 100000000/* shared counter variable */volatile unsigned int cnt = 0;/* thread routine */void *count(void *arg) {int i;for (i = 0; i < NITERS; i++)cnt++;return NULL;}Revised Book Codemovl $99999999, %edx.L15:movl cnt, %eaxincl %eaxdecl %edxmovl %eax, cntjns .L15Generated Code– 14 –15-213, F’06Progress GraphsA progress graph depictsthe discrete execution state space of concurrentthreads.Each axis corresponds tothe sequential order ofinstructions in a thread.Each point corresponds toa possible execution state(Inst1, Inst2).E.g., (L1, S2) denotes statewhere thread 1 hascompleted L1and thread2 has completed S2.H1L1U1S1T1H2L2U2S2T2Thread 1Thread 2(L1, S2)– 15 –15-213, F’06Trajectories in Progress GraphsA trajectory is a sequence of legal state transitions that describes one possible concurrent execution ofthe threads.Example:H1, L1, U1, H2, L2, S1, T1, U2, S2,
View Full Document