Is Parallel Programming Hard, And, If So, What Can You Do About It?

More documents

Recommendations

Info

$TeX op Mac OS X, met teTeX en TeXShop - Nluug$

204 APPENDIX D. READ-COPY UPDATE IMPLEMENTATIONSthe size of the level array is thus specified byNUM_RCU_LVLS, which is computed as describedin Section D.3.1.5. The ->level field is oftenused in combination with ->node to scana level of the rcu_node hierarchy, for example,all of the leaf nodes. The elements of ->levelare filled in by the boot-time rcu_init_one()function.levelcnt: Thisfieldisanarraycontainingthenumberof rcu_node structures in each level of thehierarchy, including the number of rcu_datastructures referencing the leaf rcu_node structures,so that this array has one more elementthan does the ->level array. Note that->levelcnt[0] will always contain a value ofone, corresponding to the single root rcu_nodeat the top of the hierarchy. This array is initializedwith the values NUM_RCU_LVL_0, NUM_RCU_LVL_1, NUM_RCU_LVL_2, and NUM_RCU_LVL_3,which are C-preprocessor macros computed asdescribed in Section D.3.1.5. The ->levelcntfield is used to initialize other parts of the hierarchyand for debugging purposes.levelspread: Each element of this field containsthe desired number of children for thecorresponding level of the rcu_node hierarchy.This array’s element’s values are computedat runtime by one of the two rcu_init_levelspread() functions, selected by theCONFIG_RCU_FANOUT_EXACT kernel parameter.rda: Each element of this field contains a pointer tothe corresponding CPU’s rcu_data structure.ThisarrayisinitializedatboottimebytheRCU_DATA_PTR_INIT() macro.signaled: This field is used to maintain state usedby the force_quiescent_state() function, asdescribed in Section D.3.8. This field takes onvalues as follows:RCU_GP_INIT: This value indicates that thecurrent grace period is still in the processof being initialized, so that force_quiescent_state() should take no action.Of course, grace-period initializationwould need to stretch out for three jiffiesbefore this race could arise, but if you havea very large number of CPUs, this racecould in fact occur. Once grace-period initializationis complete, this value is set toeither RCU_SAVE_DYNTICK (if CONFIG_NO_HZ) or RCU_FORCE_QS otherwise.RCU_SAVE_DYNTICK: This value indicates thatforce_quiescent_state() should checkthe dynticks state of any CPUs that havenot yet reported quiescent states for thecurrent grace period. Quiescent states willbe reported on behalf of any CPUs thatare in dyntick-idle mode.RCU_FORCE_QS: This value indicates thatforce_quiescent_state() shouldrecheck dynticks state along with theonline/offline state of any CPUs that havenot yet reported quiescent states for thecurrent grace period. The rechecking ofdynticks states allows the implementationto handle cases where a given CPU mightbe in dynticks-idle state, but have been inan irq or NMI handler both times it waschecked. If all else fails, a reschedule IPIwill be sent to the laggart CPU.This field is guarded by the root rcu_nodestructure’s lock.Quick Quiz D.26: So what guards the earlierfields in this structure?gpnum: Thisfieldcontains thenumberofthecurrentgrace period, or that of the last grace periodif no grace period is currently in effect. Thisfield is guarded by the root rcu_node structure’slock,butisfrequentlyaccessed(butnevermodified) without holding this lock.completed: This field contains the number of thelast completed grace period. As such, it is equalto ->gpnum when there is no grace period inprogress, or one less than ->gpnum when thereis a grace period in progress. In principle, onecould replace this pair of fields with a singleboolean, as is done in Classic RCU in some versionsof Linux, but in practice race resolution ismuch simpler given the pair of numbers. Thisfield is guarded by the root rcu_node structure’slock,butisfrequentlyaccessed(butnevermodified) without holding this lock.onofflock: This field prevents online/offline processingfrom running concurrently with graceperiodinitialization. There is one exception tothis: if the rcu_node hierarchy consists of buta single structure, then that single structure’s->lock field will instead take on this job.fqslock: This field prevents more than one taskfrom forcing quiescent states with force_quiescent_state().
D.3. HIERARCHICAL RCU CODE WALKTHROUGH 205jiffies_force_qs: This field contains the time, injiffies, when force_quiescent_state() shouldbeinvokedinordertoforceCPUsintoquiescentstates and/or report extended quiescent states.This field is guarded by the root rcu_nodestructure’s lock, but is frequently accessed (butnever modified) without holding this lock.n_force_qs: This field counts the number of callsto force_quiescent_state() that actually dowork, as opposed to leaving early due tothe grace period having already completed,some other CPU currently running force_quiescent_state(), or force_quiescent_state() having run too recently. This field isused for tracing and debugging, and is guardedby ->fqslock.n_force_qs_lh: This field holds an approximatecount of the number of times that force_quiescent_state() returned early due to the->fqslockbeingheldbysomeotherCPU.Thisfield is used for tracing and debugging, and isnot guarded by any lock, hence its approximatenature.n_force_qs_ngp: This field counts the numberof times that force_quiescent_state() thatsuccessfully acquire ->fqslock, but then findthat there is no grace period in progress. Thisfield is used for tracing and debugging, and isguarded by ->fqslock.gp_start: This field records the time at which themost recent grace period began, in jiffies. Thisis used to detect stalled CPUs, but only whenthe CONFIG_RCU_CPU_STALL_DETECTOR kernelparameter is selected. This field is guarded bythe root rcu_node’s ->lock, but is sometimesaccessed (but not modified) outside of this lock.jiffies_stall: This field holds the time, in jiffies,at which the current grace period will have extendedfor so long that it will be appropriate tocheck for CPU stalls. As with ->gp_start, thisfield exists only when the CONFIG_RCU_CPU_STALL_DETECTOR kernel parameter is selected.This field is guarded by the root rcu_node’s->lock, but is sometimes accessed (but notmodified) outside of this lock.dynticks_completed: This field records the valueof ->completed at the time when force_quiescent_state() snapshots dyntick state,but is also initialized to an earlier grace periodat the beginning of each grace period. This fieldis used to prevent dyntick-idle quiescent statesfrom a prior grace period from being appliedto the current grace period. As such, this fieldexists only when the CONFIG_NO_HZ kernel parameteris selected. This field is guarded by theroot rcu_node’s ->lock, but is sometimes accessed(but not modified) outside of this lock.D.3.1.5 Kernel ParametersThe following kernel parameters affect this variantof RCU:• NR_CPUS, the maximum number of CPUs in thesystem.• CONFIG_RCU_FANOUT, the desired number ofchildren for each node in the rcu_node hierarchy.• CONFIG_RCU_FANOUT_EXACT, abooleanpreventingrebalancing of the rcu_node hierarchy.• CONFIG_HOTPLUG_CPU, permitting CPUs tocome online and go offline.• CONFIG_NO_HZ, indicating that dynticks-idlemode is supported.• CONFIG_SMP, indicating that multiple CPUsmay be present.• CONFIG_RCU_CPU_STALL_DETECTOR, indicatingthat RCU should check for stalled CPUs whenRCU grace periods extend too long.• CONFIG_RCU_TRACE, indicating that RCUshould provide tracing information in debugfs.The CONFIG_RCU_FANOUT and NR_CPUS parametersareusedtodeterminetheshapeofthercu_nodehierarchy at compile time, as shown in Figure D.20.Line 1 defines the maximum depth of the rcu_nodehierarchy, currently three. Note that increasing themaximum permitted depth requires changes elsewhere,for example, adding another leg to the #ifstatement running from lines 6-26. Lines 2-4 computethe fanout, the square of the fanout, and thecube of the fanout, respectively.Then these values are compared to NR_CPUS todetermine the required depth of the rcu_node hierarchy,which is placed into NUM_RCU_LVLS, whichis used to size a number of arrays in the rcu_statestructure. Thereisalwaysonenodeattherootlevel,and there are always NUM_CPUS number of rcu_datastructures below the leaf level. If there is morethan just the root level, the number of nodes at
Page 1 and 2:
Is Parallel Programming Hard, And,
Page 3 and 4:
Contents1 Introduction 11.1 Histori
Page 5 and 6:
CONTENTSv6 Locking 676.1 Staying Al
Page 7 and 8:
CONTENTSviiB Synchronization Primit
Page 9 and 10:
CONTENTSixE.7.1 Introduction to Pre
Page 11 and 12:
PrefaceThe purpose of this book is
Page 13 and 14:
Chapter 1IntroductionParallel progr
Page 15 and 16:
1.2. PARALLEL PROGRAMMING GOALS 3CP
Page 17 and 18:
1.3. ALTERNATIVES TO PARALLEL PROGR
Page 19 and 20:
1.4. WHAT MAKES PARALLEL PROGRAMMIN
Page 21 and 22:
1.5. GUIDE TO THIS BOOK 9other hand
Page 23 and 24:
Chapter 2Hardware and its HabitsMos
Page 25:
2.1. OVERVIEW 13Therefore, as shown
Page 28 and 29:
16 CHAPTER 2. HARDWARE AND ITS HABI
Page 30 and 31:
18 CHAPTER 2. HARDWARE AND ITS HABI
Page 32 and 33:
20 CHAPTER 3. TOOLS OF THE TRADE1 p
Page 34 and 35:
22 CHAPTER 3. TOOLS OF THE TRADE1 p
Page 36 and 37:
24 CHAPTER 3. TOOLS OF THE TRADE1.1
Page 38 and 39:
26 CHAPTER 3. TOOLS OF THE TRADEQui
Page 40 and 41:
28 CHAPTER 3. TOOLS OF THE TRADE
Page 42 and 43:
30 CHAPTER 4. COUNTING1 atomic_t co
Page 44 and 45:
32 CHAPTER 4. COUNTING4.2.3 Eventua
Page 46 and 47:
34 CHAPTER 4. COUNTINGvanish when t
Page 48 and 49:
36 CHAPTER 4. COUNTINGper-thread va
Page 50 and 51:
38 CHAPTER 4. COUNTING1 unsigned lo
Page 52 and 53:
Page 54 and 55:
42 CHAPTER 4. COUNTING1 #define THE
Page 56 and 57:
Page 58 and 59:
46 CHAPTER 4. COUNTINGReadsAlgorith
Page 60 and 61:
48 CHAPTER 5. PARTITIONING AND SYNC
Page 62 and 63:
Page 64 and 65:
Page 66 and 67:
Page 68 and 69:
Page 70 and 71:
Page 72 and 73:
Page 74 and 75:
Page 76 and 77:
Page 78 and 79:
Page 80 and 81:
68 CHAPTER 6. LOCKING1 int delete(i
Page 82 and 83:
70 CHAPTER 7. DATA OWNERSHIP
Page 84 and 85:
72 CHAPTER 8. DEFERRED PROCESSINGfo
Page 86 and 87:
74 CHAPTER 8. DEFERRED PROCESSINGth
Page 88 and 89:
76 CHAPTER 8. DEFERRED PROCESSING
Page 90 and 91:
78 CHAPTER 8. DEFERRED PROCESSINGfi
Page 92 and 93:
80 CHAPTER 8. DEFERRED PROCESSINGti
Page 94 and 95:
82 CHAPTER 8. DEFERRED PROCESSINGNo
Page 96 and 97:
84 CHAPTER 8. DEFERRED PROCESSING12
Page 98 and 99:
Page 100 and 101:
88 CHAPTER 8. DEFERRED PROCESSINGvo
Page 102 and 103:
90 CHAPTER 8. DEFERRED PROCESSINGLi
Page 104 and 105:
92 CHAPTER 8. DEFERRED PROCESSINGpe
Page 106 and 107:
94 CHAPTER 8. DEFERRED PROCESSINGCa
Page 108 and 109:
96 CHAPTER 8. DEFERRED PROCESSINGTh
Page 110 and 111:
Page 112 and 113:
Page 114 and 115:
Page 116 and 117:
104 CHAPTER 8. DEFERRED PROCESSINGs
Page 118 and 119:
106 CHAPTER 8. DEFERRED PROCESSINGo
Page 120 and 121:
108 CHAPTER 8. DEFERRED PROCESSING
Page 122 and 123:
110 CHAPTER 9. APPLYING RCU1 struct
Page 124 and 125:
112 CHAPTER 9. APPLYING RCU
Page 126 and 127:
114 CHAPTER 10. VALIDATION: DEBUGGI
Page 128 and 129:
116 CHAPTER 11. DATA STRUCTURES
Page 130 and 131:
118 CHAPTER 12. ADVANCED SYNCHRONIZ
Page 132 and 133:
Page 134 and 135:
Page 136 and 137:
Page 138 and 139:
Page 140 and 141:
Page 142 and 143:
Page 144 and 145:
Page 146 and 147:
Page 148 and 149:
Page 150 and 151:
138 CHAPTER 13. EASE OF USEFigure 1
Page 152 and 153:
140 CHAPTER 13. EASE OF USE
Page 154 and 155:
142 CHAPTER 14. TIME MANAGEMENT
Page 156 and 157:
144 CHAPTER 15. CONFLICTING VISIONS
Page 158 and 159:
Page 160 and 161:
Page 162 and 163:
Page 164 and 165:
Page 166 and 167: 154 APPENDIX A. IMPORTANT QUESTIONS
Page 168 and 169: 156 APPENDIX A. IMPORTANT QUESTIONS
Page 170 and 171: 158 APPENDIX B. SYNCHRONIZATION PRI
Page 172 and 173: 160 APPENDIX B. SYNCHRONIZATION PRI
Page 174 and 175: 162 APPENDIX C. WHY MEMORY BARRIERS
Page 196 and 197: 184 APPENDIX D. READ-COPY UPDATE IM
Page 256 and 257: 244 APPENDIX E. FORMAL VERIFICATION
Page 266 and 267:
254 APPENDIX E. FORMAL VERIFICATION
Page 268 and 269:
Page 270 and 271:
Page 272 and 273:
Page 274 and 275:
Page 276 and 277:
Page 278 and 279:
Page 280 and 281:
Page 282 and 283:
Page 284 and 285:
272 APPENDIX F. ANSWERS TO QUICK QU
Page 286 and 287:
Page 288 and 289:
Page 290 and 291:
Page 292 and 293:
Page 294 and 295:
Page 296 and 297:
Page 298 and 299:
Page 300 and 301:
Page 302 and 303:
Page 304 and 305:
Page 306 and 307:
Page 308 and 309:
Page 310 and 311:
Page 312 and 313:
Page 314 and 315:
Page 316 and 317:
Page 318 and 319:
Page 320 and 321:
Page 322 and 323:
Page 324 and 325:
Page 326 and 327:
Page 328 and 329:
Page 330 and 331:
Page 332 and 333:
Page 334 and 335:
Page 336 and 337:
Page 338 and 339:
Page 340 and 341:
Page 342 and 343:
330 APPENDIX G. GLOSSARY(2) A physi
Page 344 and 345:
332 APPENDIX G. GLOSSARYnear by. Th
Page 346 and 347:
334 APPENDIX G. GLOSSARY
Page 348 and 349:
336 BIBLIOGRAPHY[But97]USA, March 2
Page 350 and 351:
338 BIBLIOGRAPHY[HMB06][Hol03][HP95
Page 352 and 353:
340 BIBLIOGRAPHY[McK06] Paul E. McK
Page 354 and 355:
342 BIBLIOGRAPHYtor. Software - Pra
Page 356 and 357:
344 BIBLIOGRAPHY[UoC08][VGS08]Berke
Page 358:
346 APPENDIX H. CREDITSH.4 Original
show all

Is Parallel Programming Hard, And, If So, What Can You Do About It?

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?