984 lines · plain
1.. SPDX-License-Identifier: GPL-2.0 OR GFDL-1.2-no-invariants-only2 3===========================4Lockless Ring Buffer Design5===========================6 7Copyright 2009 Red Hat Inc.8 9:Author: Steven Rostedt <srostedt@redhat.com>10:License: The GNU Free Documentation License, Version 1.211 (dual licensed under the GPL v2)12:Reviewers: Mathieu Desnoyers, Huang Ying, Hidetoshi Seto,13 and Frederic Weisbecker.14 15 16Written for: 2.6.3117 18Terminology used in this Document19---------------------------------20 21tail22 - where new writes happen in the ring buffer.23 24head25 - where new reads happen in the ring buffer.26 27producer28 - the task that writes into the ring buffer (same as writer)29 30writer31 - same as producer32 33consumer34 - the task that reads from the buffer (same as reader)35 36reader37 - same as consumer.38 39reader_page40 - A page outside the ring buffer used solely (for the most part)41 by the reader.42 43head_page44 - a pointer to the page that the reader will use next45 46tail_page47 - a pointer to the page that will be written to next48 49commit_page50 - a pointer to the page with the last finished non-nested write.51 52cmpxchg53 - hardware-assisted atomic transaction that performs the following::54 55 A = B if previous A == C56 57 R = cmpxchg(A, C, B) is saying that we replace A with B if and only58 if current A is equal to C, and we put the old (current)59 A into R60 61 R gets the previous A regardless if A is updated with B or not.62 63 To see if the update was successful a compare of ``R == C``64 may be used.65 66The Generic Ring Buffer67-----------------------68 69The ring buffer can be used in either an overwrite mode or in70producer/consumer mode.71 72Producer/consumer mode is where if the producer were to fill up the73buffer before the consumer could free up anything, the producer74will stop writing to the buffer. This will lose most recent events.75 76Overwrite mode is where if the producer were to fill up the buffer77before the consumer could free up anything, the producer will78overwrite the older data. This will lose the oldest events.79 80No two writers can write at the same time (on the same per-cpu buffer),81but a writer may interrupt another writer, but it must finish writing82before the previous writer may continue. This is very important to the83algorithm. The writers act like a "stack". The way interrupts works84enforces this behavior::85 86 87 writer1 start88 <preempted> writer2 start89 <preempted> writer3 start90 writer3 finishes91 writer2 finishes92 writer1 finishes93 94This is very much like a writer being preempted by an interrupt and95the interrupt doing a write as well.96 97Readers can happen at any time. But no two readers may run at the98same time, nor can a reader preempt/interrupt another reader. A reader99cannot preempt/interrupt a writer, but it may read/consume from the100buffer at the same time as a writer is writing, but the reader must be101on another processor to do so. A reader may read on its own processor102and can be preempted by a writer.103 104A writer can preempt a reader, but a reader cannot preempt a writer.105But a reader can read the buffer at the same time (on another processor)106as a writer.107 108The ring buffer is made up of a list of pages held together by a linked list.109 110At initialization a reader page is allocated for the reader that is not111part of the ring buffer.112 113The head_page, tail_page and commit_page are all initialized to point114to the same page.115 116The reader page is initialized to have its next pointer pointing to117the head page, and its previous pointer pointing to a page before118the head page.119 120The reader has its own page to use. At start up time, this page is121allocated but is not attached to the list. When the reader wants122to read from the buffer, if its page is empty (like it is on start-up),123it will swap its page with the head_page. The old reader page will124become part of the ring buffer and the head_page will be removed.125The page after the inserted page (old reader_page) will become the126new head page.127 128Once the new page is given to the reader, the reader could do what129it wants with it, as long as a writer has left that page.130 131A sample of how the reader page is swapped: Note this does not132show the head page in the buffer, it is for demonstrating a swap133only.134 135::136 137 +------+138 |reader| RING BUFFER139 |page |140 +------+141 +---+ +---+ +---+142 | |-->| |-->| |143 | |<--| |<--| |144 +---+ +---+ +---+145 ^ | ^ |146 | +-------------+ |147 +-----------------+148 149 150 +------+151 |reader| RING BUFFER152 |page |-------------------+153 +------+ v154 | +---+ +---+ +---+155 | | |-->| |-->| |156 | | |<--| |<--| |<-+157 | +---+ +---+ +---+ |158 | ^ | ^ | |159 | | +-------------+ | |160 | +-----------------+ |161 +------------------------------------+162 163 +------+164 |reader| RING BUFFER165 |page |-------------------+166 +------+ <---------------+ v167 | ^ +---+ +---+ +---+168 | | | |-->| |-->| |169 | | | | | |<--| |<-+170 | | +---+ +---+ +---+ |171 | | | ^ | |172 | | +-------------+ | |173 | +-----------------------------+ |174 +------------------------------------+175 176 +------+177 |buffer| RING BUFFER178 |page |-------------------+179 +------+ <---------------+ v180 | ^ +---+ +---+ +---+181 | | | | | |-->| |182 | | New | | | |<--| |<-+183 | | Reader +---+ +---+ +---+ |184 | | page ----^ | |185 | | | |186 | +-----------------------------+ |187 +------------------------------------+188 189 190 191It is possible that the page swapped is the commit page and the tail page,192if what is in the ring buffer is less than what is held in a buffer page.193 194::195 196 reader page commit page tail page197 | | |198 v | |199 +---+ | |200 | |<----------+ |201 | |<------------------------+202 | |------+203 +---+ |204 |205 v206 +---+ +---+ +---+ +---+207 <---| |--->| |--->| |--->| |--->208 --->| |<---| |<---| |<---| |<---209 +---+ +---+ +---+ +---+210 211This case is still valid for this algorithm.212When the writer leaves the page, it simply goes into the ring buffer213since the reader page still points to the next location in the ring214buffer.215 216 217The main pointers:218 219 reader page220 - The page used solely by the reader and is not part221 of the ring buffer (may be swapped in)222 223 head page224 - the next page in the ring buffer that will be swapped225 with the reader page.226 227 tail page228 - the page where the next write will take place.229 230 commit page231 - the page that last finished a write.232 233The commit page only is updated by the outermost writer in the234writer stack. A writer that preempts another writer will not move the235commit page.236 237When data is written into the ring buffer, a position is reserved238in the ring buffer and passed back to the writer. When the writer239is finished writing data into that position, it commits the write.240 241Another write (or a read) may take place at anytime during this242transaction. If another write happens it must finish before continuing243with the previous write.244 245 246 Write reserve::247 248 Buffer page249 +---------+250 |written |251 +---------+ <--- given back to writer (current commit)252 |reserved |253 +---------+ <--- tail pointer254 | empty |255 +---------+256 257 Write commit::258 259 Buffer page260 +---------+261 |written |262 +---------+263 |written |264 +---------+ <--- next position for write (current commit)265 | empty |266 +---------+267 268 269 If a write happens after the first reserve::270 271 Buffer page272 +---------+273 |written |274 +---------+ <-- current commit275 |reserved |276 +---------+ <--- given back to second writer277 |reserved |278 +---------+ <--- tail pointer279 280 After second writer commits::281 282 283 Buffer page284 +---------+285 |written |286 +---------+ <--(last full commit)287 |reserved |288 +---------+289 |pending |290 |commit |291 +---------+ <--- tail pointer292 293 When the first writer commits::294 295 Buffer page296 +---------+297 |written |298 +---------+299 |written |300 +---------+301 |written |302 +---------+ <--(last full commit and tail pointer)303 304 305The commit pointer points to the last write location that was306committed without preempting another write. When a write that307preempted another write is committed, it only becomes a pending commit308and will not be a full commit until all writes have been committed.309 310The commit page points to the page that has the last full commit.311The tail page points to the page with the last write (before312committing).313 314The tail page is always equal to or after the commit page. It may315be several pages ahead. If the tail page catches up to the commit316page then no more writes may take place (regardless of the mode317of the ring buffer: overwrite and produce/consumer).318 319The order of pages is::320 321 head page322 commit page323 tail page324 325Possible scenario::326 327 tail page328 head page commit page |329 | | |330 v v v331 +---+ +---+ +---+ +---+332 <---| |--->| |--->| |--->| |--->333 --->| |<---| |<---| |<---| |<---334 +---+ +---+ +---+ +---+335 336There is a special case that the head page is after either the commit page337and possibly the tail page. That is when the commit (and tail) page has been338swapped with the reader page. This is because the head page is always339part of the ring buffer, but the reader page is not. Whenever there340has been less than a full page that has been committed inside the ring buffer,341and a reader swaps out a page, it will be swapping out the commit page.342 343::344 345 reader page commit page tail page346 | | |347 v | |348 +---+ | |349 | |<----------+ |350 | |<------------------------+351 | |------+352 +---+ |353 |354 v355 +---+ +---+ +---+ +---+356 <---| |--->| |--->| |--->| |--->357 --->| |<---| |<---| |<---| |<---358 +---+ +---+ +---+ +---+359 ^360 |361 head page362 363 364In this case, the head page will not move when the tail and commit365move back into the ring buffer.366 367The reader cannot swap a page into the ring buffer if the commit page368is still on that page. If the read meets the last commit (real commit369not pending or reserved), then there is nothing more to read.370The buffer is considered empty until another full commit finishes.371 372When the tail meets the head page, if the buffer is in overwrite mode,373the head page will be pushed ahead one. If the buffer is in producer/consumer374mode, the write will fail.375 376Overwrite mode::377 378 tail page379 |380 v381 +---+ +---+ +---+ +---+382 <---| |--->| |--->| |--->| |--->383 --->| |<---| |<---| |<---| |<---384 +---+ +---+ +---+ +---+385 ^386 |387 head page388 389 390 tail page391 |392 v393 +---+ +---+ +---+ +---+394 <---| |--->| |--->| |--->| |--->395 --->| |<---| |<---| |<---| |<---396 +---+ +---+ +---+ +---+397 ^398 |399 head page400 401 402 tail page403 |404 v405 +---+ +---+ +---+ +---+406 <---| |--->| |--->| |--->| |--->407 --->| |<---| |<---| |<---| |<---408 +---+ +---+ +---+ +---+409 ^410 |411 head page412 413Note, the reader page will still point to the previous head page.414But when a swap takes place, it will use the most recent head page.415 416 417Making the Ring Buffer Lockless:418--------------------------------419 420The main idea behind the lockless algorithm is to combine the moving421of the head_page pointer with the swapping of pages with the reader.422State flags are placed inside the pointer to the page. To do this,423each page must be aligned in memory by 4 bytes. This will allow the 2424least significant bits of the address to be used as flags, since425they will always be zero for the address. To get the address,426simply mask out the flags::427 428 MASK = ~3429 430 address & MASK431 432Two flags will be kept by these two bits:433 434 HEADER435 - the page being pointed to is a head page436 437 UPDATE438 - the page being pointed to is being updated by a writer439 and was or is about to be a head page.440 441::442 443 reader page444 |445 v446 +---+447 | |------+448 +---+ |449 |450 v451 +---+ +---+ +---+ +---+452 <---| |--->| |-H->| |--->| |--->453 --->| |<---| |<---| |<---| |<---454 +---+ +---+ +---+ +---+455 456 457The above pointer "-H->" would have the HEADER flag set. That is458the next page is the next page to be swapped out by the reader.459This pointer means the next page is the head page.460 461When the tail page meets the head pointer, it will use cmpxchg to462change the pointer to the UPDATE state::463 464 465 tail page466 |467 v468 +---+ +---+ +---+ +---+469 <---| |--->| |-H->| |--->| |--->470 --->| |<---| |<---| |<---| |<---471 +---+ +---+ +---+ +---+472 473 tail page474 |475 v476 +---+ +---+ +---+ +---+477 <---| |--->| |-U->| |--->| |--->478 --->| |<---| |<---| |<---| |<---479 +---+ +---+ +---+ +---+480 481"-U->" represents a pointer in the UPDATE state.482 483Any access to the reader will need to take some sort of lock to serialize484the readers. But the writers will never take a lock to write to the485ring buffer. This means we only need to worry about a single reader,486and writes only preempt in "stack" formation.487 488When the reader tries to swap the page with the ring buffer, it489will also use cmpxchg. If the flag bit in the pointer to the490head page does not have the HEADER flag set, the compare will fail491and the reader will need to look for the new head page and try again.492Note, the flags UPDATE and HEADER are never set at the same time.493 494The reader swaps the reader page as follows::495 496 +------+497 |reader| RING BUFFER498 |page |499 +------+500 +---+ +---+ +---+501 | |--->| |--->| |502 | |<---| |<---| |503 +---+ +---+ +---+504 ^ | ^ |505 | +---------------+ |506 +-----H-------------+507 508The reader sets the reader page next pointer as HEADER to the page after509the head page::510 511 512 +------+513 |reader| RING BUFFER514 |page |-------H-----------+515 +------+ v516 | +---+ +---+ +---+517 | | |--->| |--->| |518 | | |<---| |<---| |<-+519 | +---+ +---+ +---+ |520 | ^ | ^ | |521 | | +---------------+ | |522 | +-----H-------------+ |523 +--------------------------------------+524 525It does a cmpxchg with the pointer to the previous head page to make it526point to the reader page. Note that the new pointer does not have the HEADER527flag set. This action atomically moves the head page forward::528 529 +------+530 |reader| RING BUFFER531 |page |-------H-----------+532 +------+ v533 | ^ +---+ +---+ +---+534 | | | |-->| |-->| |535 | | | |<--| |<--| |<-+536 | | +---+ +---+ +---+ |537 | | | ^ | |538 | | +-------------+ | |539 | +-----------------------------+ |540 +------------------------------------+541 542After the new head page is set, the previous pointer of the head page is543updated to the reader page::544 545 +------+546 |reader| RING BUFFER547 |page |-------H-----------+548 +------+ <---------------+ v549 | ^ +---+ +---+ +---+550 | | | |-->| |-->| |551 | | | | | |<--| |<-+552 | | +---+ +---+ +---+ |553 | | | ^ | |554 | | +-------------+ | |555 | +-----------------------------+ |556 +------------------------------------+557 558 +------+559 |buffer| RING BUFFER560 |page |-------H-----------+ <--- New head page561 +------+ <---------------+ v562 | ^ +---+ +---+ +---+563 | | | | | |-->| |564 | | New | | | |<--| |<-+565 | | Reader +---+ +---+ +---+ |566 | | page ----^ | |567 | | | |568 | +-----------------------------+ |569 +------------------------------------+570 571Another important point: The page that the reader page points back to572by its previous pointer (the one that now points to the new head page)573never points back to the reader page. That is because the reader page is574not part of the ring buffer. Traversing the ring buffer via the next pointers575will always stay in the ring buffer. Traversing the ring buffer via the576prev pointers may not.577 578Note, the way to determine a reader page is simply by examining the previous579pointer of the page. If the next pointer of the previous page does not580point back to the original page, then the original page is a reader page::581 582 583 +--------+584 | reader | next +----+585 | page |-------->| |<====== (buffer page)586 +--------+ +----+587 | | ^588 | v | next589 prev | +----+590 +------------->| |591 +----+592 593The way the head page moves forward:594 595When the tail page meets the head page and the buffer is in overwrite mode596and more writes take place, the head page must be moved forward before the597writer may move the tail page. The way this is done is that the writer598performs a cmpxchg to convert the pointer to the head page from the HEADER599flag to have the UPDATE flag set. Once this is done, the reader will600not be able to swap the head page from the buffer, nor will it be able to601move the head page, until the writer is finished with the move.602 603This eliminates any races that the reader can have on the writer. The reader604must spin, and this is why the reader cannot preempt the writer::605 606 tail page607 |608 v609 +---+ +---+ +---+ +---+610 <---| |--->| |-H->| |--->| |--->611 --->| |<---| |<---| |<---| |<---612 +---+ +---+ +---+ +---+613 614 tail page615 |616 v617 +---+ +---+ +---+ +---+618 <---| |--->| |-U->| |--->| |--->619 --->| |<---| |<---| |<---| |<---620 +---+ +---+ +---+ +---+621 622The following page will be made into the new head page::623 624 tail page625 |626 v627 +---+ +---+ +---+ +---+628 <---| |--->| |-U->| |-H->| |--->629 --->| |<---| |<---| |<---| |<---630 +---+ +---+ +---+ +---+631 632After the new head page has been set, we can set the old head page633pointer back to NORMAL::634 635 tail page636 |637 v638 +---+ +---+ +---+ +---+639 <---| |--->| |--->| |-H->| |--->640 --->| |<---| |<---| |<---| |<---641 +---+ +---+ +---+ +---+642 643After the head page has been moved, the tail page may now move forward::644 645 tail page646 |647 v648 +---+ +---+ +---+ +---+649 <---| |--->| |--->| |-H->| |--->650 --->| |<---| |<---| |<---| |<---651 +---+ +---+ +---+ +---+652 653 654The above are the trivial updates. Now for the more complex scenarios.655 656 657As stated before, if enough writes preempt the first write, the658tail page may make it all the way around the buffer and meet the commit659page. At this time, we must start dropping writes (usually with some kind660of warning to the user). But what happens if the commit was still on the661reader page? The commit page is not part of the ring buffer. The tail page662must account for this::663 664 665 reader page commit page666 | |667 v |668 +---+ |669 | |<----------+670 | |671 | |------+672 +---+ |673 |674 v675 +---+ +---+ +---+ +---+676 <---| |--->| |-H->| |--->| |--->677 --->| |<---| |<---| |<---| |<---678 +---+ +---+ +---+ +---+679 ^680 |681 tail page682 683If the tail page were to simply push the head page forward, the commit when684leaving the reader page would not be pointing to the correct page.685 686The solution to this is to test if the commit page is on the reader page687before pushing the head page. If it is, then it can be assumed that the688tail page wrapped the buffer, and we must drop new writes.689 690This is not a race condition, because the commit page can only be moved691by the outermost writer (the writer that was preempted).692This means that the commit will not move while a writer is moving the693tail page. The reader cannot swap the reader page if it is also being694used as the commit page. The reader can simply check that the commit695is off the reader page. Once the commit page leaves the reader page696it will never go back on it unless a reader does another swap with the697buffer page that is also the commit page.698 699 700Nested writes701-------------702 703In the pushing forward of the tail page we must first push forward704the head page if the head page is the next page. If the head page705is not the next page, the tail page is simply updated with a cmpxchg.706 707Only writers move the tail page. This must be done atomically to protect708against nested writers::709 710 temp_page = tail_page711 next_page = temp_page->next712 cmpxchg(tail_page, temp_page, next_page)713 714The above will update the tail page if it is still pointing to the expected715page. If this fails, a nested write pushed it forward, the current write716does not need to push it::717 718 719 temp page720 |721 v722 tail page723 |724 v725 +---+ +---+ +---+ +---+726 <---| |--->| |--->| |--->| |--->727 --->| |<---| |<---| |<---| |<---728 +---+ +---+ +---+ +---+729 730Nested write comes in and moves the tail page forward::731 732 tail page (moved by nested writer)733 temp page |734 | |735 v v736 +---+ +---+ +---+ +---+737 <---| |--->| |--->| |--->| |--->738 --->| |<---| |<---| |<---| |<---739 +---+ +---+ +---+ +---+740 741The above would fail the cmpxchg, but since the tail page has already742been moved forward, the writer will just try again to reserve storage743on the new tail page.744 745But the moving of the head page is a bit more complex::746 747 tail page748 |749 v750 +---+ +---+ +---+ +---+751 <---| |--->| |-H->| |--->| |--->752 --->| |<---| |<---| |<---| |<---753 +---+ +---+ +---+ +---+754 755The write converts the head page pointer to UPDATE::756 757 tail page758 |759 v760 +---+ +---+ +---+ +---+761 <---| |--->| |-U->| |--->| |--->762 --->| |<---| |<---| |<---| |<---763 +---+ +---+ +---+ +---+764 765But if a nested writer preempts here, it will see that the next766page is a head page, but it is also nested. It will detect that767it is nested and will save that information. The detection is the768fact that it sees the UPDATE flag instead of a HEADER or NORMAL769pointer.770 771The nested writer will set the new head page pointer::772 773 tail page774 |775 v776 +---+ +---+ +---+ +---+777 <---| |--->| |-U->| |-H->| |--->778 --->| |<---| |<---| |<---| |<---779 +---+ +---+ +---+ +---+780 781But it will not reset the update back to normal. Only the writer782that converted a pointer from HEAD to UPDATE will convert it back783to NORMAL::784 785 tail page786 |787 v788 +---+ +---+ +---+ +---+789 <---| |--->| |-U->| |-H->| |--->790 --->| |<---| |<---| |<---| |<---791 +---+ +---+ +---+ +---+792 793After the nested writer finishes, the outermost writer will convert794the UPDATE pointer to NORMAL::795 796 797 tail page798 |799 v800 +---+ +---+ +---+ +---+801 <---| |--->| |--->| |-H->| |--->802 --->| |<---| |<---| |<---| |<---803 +---+ +---+ +---+ +---+804 805 806It can be even more complex if several nested writes came in and moved807the tail page ahead several pages::808 809 810 (first writer)811 812 tail page813 |814 v815 +---+ +---+ +---+ +---+816 <---| |--->| |-H->| |--->| |--->817 --->| |<---| |<---| |<---| |<---818 +---+ +---+ +---+ +---+819 820The write converts the head page pointer to UPDATE::821 822 tail page823 |824 v825 +---+ +---+ +---+ +---+826 <---| |--->| |-U->| |--->| |--->827 --->| |<---| |<---| |<---| |<---828 +---+ +---+ +---+ +---+829 830Next writer comes in, and sees the update and sets up the new831head page::832 833 (second writer)834 835 tail page836 |837 v838 +---+ +---+ +---+ +---+839 <---| |--->| |-U->| |-H->| |--->840 --->| |<---| |<---| |<---| |<---841 +---+ +---+ +---+ +---+842 843The nested writer moves the tail page forward. But does not set the old844update page to NORMAL because it is not the outermost writer::845 846 tail page847 |848 v849 +---+ +---+ +---+ +---+850 <---| |--->| |-U->| |-H->| |--->851 --->| |<---| |<---| |<---| |<---852 +---+ +---+ +---+ +---+853 854Another writer preempts and sees the page after the tail page is a head page.855It changes it from HEAD to UPDATE::856 857 (third writer)858 859 tail page860 |861 v862 +---+ +---+ +---+ +---+863 <---| |--->| |-U->| |-U->| |--->864 --->| |<---| |<---| |<---| |<---865 +---+ +---+ +---+ +---+866 867The writer will move the head page forward::868 869 870 (third writer)871 872 tail page873 |874 v875 +---+ +---+ +---+ +---+876 <---| |--->| |-U->| |-U->| |-H->877 --->| |<---| |<---| |<---| |<---878 +---+ +---+ +---+ +---+879 880But now that the third writer did change the HEAD flag to UPDATE it881will convert it to normal::882 883 884 (third writer)885 886 tail page887 |888 v889 +---+ +---+ +---+ +---+890 <---| |--->| |-U->| |--->| |-H->891 --->| |<---| |<---| |<---| |<---892 +---+ +---+ +---+ +---+893 894 895Then it will move the tail page, and return back to the second writer::896 897 898 (second writer)899 900 tail page901 |902 v903 +---+ +---+ +---+ +---+904 <---| |--->| |-U->| |--->| |-H->905 --->| |<---| |<---| |<---| |<---906 +---+ +---+ +---+ +---+907 908 909The second writer will fail to move the tail page because it was already910moved, so it will try again and add its data to the new tail page.911It will return to the first writer::912 913 914 (first writer)915 916 tail page917 |918 v919 +---+ +---+ +---+ +---+920 <---| |--->| |-U->| |--->| |-H->921 --->| |<---| |<---| |<---| |<---922 +---+ +---+ +---+ +---+923 924The first writer cannot know atomically if the tail page moved925while it updates the HEAD page. It will then update the head page to926what it thinks is the new head page::927 928 929 (first writer)930 931 tail page932 |933 v934 +---+ +---+ +---+ +---+935 <---| |--->| |-U->| |-H->| |-H->936 --->| |<---| |<---| |<---| |<---937 +---+ +---+ +---+ +---+938 939Since the cmpxchg returns the old value of the pointer the first writer940will see it succeeded in updating the pointer from NORMAL to HEAD.941But as we can see, this is not good enough. It must also check to see942if the tail page is either where it use to be or on the next page::943 944 945 (first writer)946 947 A B tail page948 | | |949 v v v950 +---+ +---+ +---+ +---+951 <---| |--->| |-U->| |-H->| |-H->952 --->| |<---| |<---| |<---| |<---953 +---+ +---+ +---+ +---+954 955If tail page != A and tail page != B, then it must reset the pointer956back to NORMAL. The fact that it only needs to worry about nested957writers means that it only needs to check this after setting the HEAD page::958 959 960 (first writer)961 962 A B tail page963 | | |964 v v v965 +---+ +---+ +---+ +---+966 <---| |--->| |-U->| |--->| |-H->967 --->| |<---| |<---| |<---| |<---968 +---+ +---+ +---+ +---+969 970Now the writer can update the head page. This is also why the head page must971remain in UPDATE and only reset by the outermost writer. This prevents972the reader from seeing the incorrect head page::973 974 975 (first writer)976 977 A B tail page978 | | |979 v v v980 +---+ +---+ +---+ +---+981 <---| |--->| |--->| |--->| |-H->982 --->| |<---| |<---| |<---| |<---983 +---+ +---+ +---+ +---+984