TW436795B - Memory system - Google Patents

Memory system Download PDF

Info

Publication number
TW436795B
TW436795B TW88100946A TW88100946A TW436795B TW 436795 B TW436795 B TW 436795B TW 88100946 A TW88100946 A TW 88100946A TW 88100946 A TW88100946 A TW 88100946A TW 436795 B TW436795 B TW 436795B
Authority
TW
Taiwan
Prior art keywords
cell
memory
address
controller
erasable
Prior art date
Application number
TW88100946A
Other languages
Chinese (zh)
Inventor
Alan Welsh Sinclair
Sergey Anatolivich Gorobets
Original Assignee
Memory Corp Plc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Memory Corp Plc filed Critical Memory Corp Plc
Priority to TW88100946A priority Critical patent/TW436795B/en
Application granted granted Critical
Publication of TW436795B publication Critical patent/TW436795B/en

Links

Landscapes

  • Read Only Memory (AREA)
  • Techniques For Improving Reliability Of Storages (AREA)

Abstract

A memory system (10) comprising a non-volatile memory (18) having memory locations (38), and a controller (16) for writing data structures to and reading data structures from the memory. The system (10) is architecturally configured so that the locations (38) can be written to individually but are erasable only in blocks. The controller (16) forms one or more erasable units (39) which are each subdivided into cells (50) each consisting of a group of locations (38). The controller (16) writes data structures to and reads data structures from each cell (50) on a per cell basis. The system (10) may comprise a controller (16) may be embedded in, or implemented in, a host system such as a personal computer (PC).

Description

經濟部智慧財產局員工消費合作社印製 p Λ36795 at _____B7___ 五、發明説明(I ) 本發明有關一種用於儲存資料和取回資料的非揮發記 憶體系統,其記憶體系統包含一個具有記憶位置之記憶體 ,而其記憶位置能夠個別地寫入,但卻只能以整個區塊的 位置來淸除,其系統並且包含一個用來控制這些記憶位置 存取的控制器;本發明也有關一種非揮發記憶體,用於如 此的一個非揮發記憶體系統中,並且有關一種用來控制非 揮發性記憶體的控制器。特別的是,本發明有關具有失效 (defective)記憶位置的快閃記憶體(FLASH memory)系統, 以及有關用於快閃記憶體的控制器。 快閃EPROM(可淸除之可程式唯讀記憶體)裝置通常用 於電子工業非揮發性資料之儲存。快閃記憶體裝置的架構 具有可個別寫入但卻只能集體淸除的位置,稱之爲可淸除 的區塊(erasable block)。由於快閃記憶體中的電晶體集群 .乃是藉由一條共用的淸除線路所連結,因此而產生如此的 架構。所以,可淸除的區塊之大小(儲存位置的數量)乃是 由裝置的結構所決定,且在設計與製造時便已制定,並不 j 能爲使用者所修改。 資料儲存的一種應用則是儲存由,例如,個人電腦 (PC)所產生的資料結構。如果具有失效記憶位置的快閃記 憶體用來儲存資料結構,則由於這些失效記億位置並不能 確實地用於資料的儲存,因此而產生一個問題。 •解決記憶體中具有失效位置的問題之一個方法,乃是 快閃控制器避免使用任何包含失效位置的可淸除區塊。然 而,如果絕不使用包含失效位置的可淸除區塊(由控制器標 __-_ 4___ 本紙張尺度適用中國國家標準(CNS ) A4規格(2I0X297公釐) -----------^------ΐτ------ii (請先閲讀背面之注意事項再vik本頁) 經濟部钇恶財產局員工消費合作社印製 ί ' 4367 9 5 at _______Β7 五、發明説明(> ) 記爲不能使用的),則可能會大大地浪費了可用的記憶體儲 存空間(視可淸除的區塊之大小以及其中可使用的記憶位置 之數量而定),而導致低的記憶體使用率(可使用的記憶體 位置對總記憶體位置之低比率)。 本發明的一個目的乃是提供一種去除或減輕以上缺點 的非揮發性記憶體系統。 本發明的一個目的乃是提供一種非揮發性記憶體系統 ,其包含具有不能用來儲存資料的位置之記憶體,但其記 憶體能夠用於資料結構有效率的儲存和取回。 根據本發明第一個的觀點,設置一種連接到主系統的 記憶體系統,該記憶體系統包含: 一個具有記憶位置的非揮發性記憶體, 以及一個控制器,用來將資料結構寫入記憶體和將資 料由記憶體中取出,建構其系統以使得位置能夠被個別地 寫入,.但只能以位置的集體區塊淸除; 其進步之處爲:控制器設置至少一個的可淸除單元, 每一個可淸除單元至少包含一個可淸除區塊,以及控制器 細分每一個可淸除單元爲位置群(每一個組在此則稱爲一個 儲存格,cell),以及基於每一個儲存格,控制器將資料結 構寫入每一個儲存格並且從每一個儲存格將資料結構讀出 〇 .藉由本發明的優點,重新建構記憶體爲儲存格’以使 得控制器可以避免使用具有失效的個別儲存格’而不必避 免使用具有失效的可淸除區塊β如此則有效地增加了記憶 -----------S _ 本紙張尺度適用中國國家標準(CNS > Α4規格(210Χ:297公釐) ---------^---t.------ΐτ------0 (請先閱讀背面之注意事項再f..本頁)·\ A7 B7 Π367 9 5 一發明説明(今) 麟的使用率。 一個可淸除單元可能只包含一個可淸除的區塊。可替 代的是,一個可淸除單元可能包含複數個的可淸除區塊; 合宣的是,二進位倍數的可淸除區塊。 將會了解到的是,一個可淸除區塊典型地較一個儲存 格大許多。 在連接到主系統之前或期間,測試每一個儲存格中的 記億位置,並且如果失效即使只出現於儲存格恰恰一個位 置中,則控制器視整個的儲存格爲不能用來儲存資料結構 的,除此之外,整個的儲存格則視爲可用來儲存資料結構 的。 將會了解到的是,專有名詞記憶位置可以指記億體儲 存的單一位元而言;然而,專有名詞儲存格則指記憶體儲 存的多數位元而言,一個典型的儲存格可以儲存256個位 元組或者512個位元組。 將會了解到的是,能夠個別寫入的位置可以由多數個 的實體記憶位置所構成,例如實體記憶體的一列,以使得 在如此的範例中,一整個列爲能夠個別寫入的記憶體之最 小單元。 非揮發性記憶體可以包含多數個的記憶體裝置,或只 有單一個的記憶體裝置。較佳的是,控制器以一種具有附 屬控制器的主控制器之形式,其附屬控制器則倂入其記憶 體裝置或者每一個記憶體裝置之中。可替代地,控制器以 一種單一控制器的形式,其單一控制器控制著其記憶體裝 --;-------- 本紙浪尺度適用中國國家標準(CNS )八4規格(21〇><297公釐) (請先閲讀背面之注意事項再ί本頁) 裝· 訂- .線 經濟部智慧財J.局員工消费合作社印製 r 436795 五、發明説明(+) 置或者每一個記憶體裝置。 . 較佳地,控制器指定保留每一個可淸除單元中至少一 個的可用儲存格,用以具有包含失效的不可用儲存格、保. 留來儲存控制資訊的儲存格、以及可用的儲存格,用來儲 存從主系統所接收到的資料。 較佳地,所保留的儲存格用來儲存位址轉換資訊,位 址轉換資訊則是用來換算來自主系統的位址,成爲適合存 取記憶體的位址。. 較佳地,由於具有多數個連結在一起的保留儲存格, 而實現其位址的轉換。 較佳地,每一個所保留的儲存格用來儲存指標資訊 (pointer information),指標資訊則是用來指向下一個所保 留的儲存格,直到抵達最後的保留儲存格,其指標資訊乃 是指向記憶體中的一個位址,而其記憶體則是儲存了適合 存取記憶體的位址。 較佳地,建構所保留的儲存格以使之具有複數個的登 錄元(entry),其中的登錄元儲存多數個的欄位(field),其中 儲存於一個登錄元中的欄位則是用來指向儲存在另一個登 錄元中的欄位,並且只有視儲存於一個登錄元中最近才寫 入的欄位爲有效的。 根據本發明的第二個觀點,設置一個控制器,用於一 個具有位置的非揮發性記憶體,其能夠被個別地寫入,但 只能以位置區塊淸除,在使用中,控制器設置至少一個的 可淸除單元,其中每一個可淸除單元則包含至少一個的可 _________2______ 參紙張尺度適用中國國家榡準(CNS ) A4规格(2丨0X297公釐) (請先閲讀背面之注意事項再J>i本頁) -裝. 訂 經濟部智慧財產局員工消骨合作社印製 經濟部智蒹財產局員工消費合作社印製 4367 9 5 五、發明説明(ζ:) 淸除區塊,再者控制器細分每一個可淸除單元爲儲存格’ 且基於每一個儲存格’從儲存格中讀取資料結構並將資料 結構寫入儲存格中。 專有名詞資料結構(data struture)乃是用來涵蓋所有儲 存在記憶體中的資料。因此,專有名詞資料結構涵蓋讀取/ 寫入區塊(read/write biock)(即是於PC和記憶體之間所傳輸 的資料區塊)以及控制資訊(control information)(即是,例如 由記憶體控制器所產生的資訊)。 稂據本發明的第三個觀點,設置一種用於控制器的非 .揮發性記憶體,具有位置的記憶體能夠被個別地寫入,但 只能以位置區塊淸除,建構非揮發性記憶體以設置至少一· 個的可淸除單元,其中每一個的可淸除單元則包含至少一 個的可淸除區塊,並且再細分每一個可淸除單元成稱爲儲 存格的記憶體位置群,以使得在每一個基本儲存格上,藉 由控制器來讀取儲存格中的資料結構’以及將資料結構寫 入儲存格之中。 由於具有所测試的每一個可淸除單元中的每—個儲存 格,而建構了非揮發性記憶體,以及將每—個單元的測試 結果寫入一組位置中,而形成在其單元中的一個標頭 (header),使得每一個單元皆具有一個包含標頭資訊的標頭 〇 .可替代地,在一個包含多數個區塊的可淸除單元中, 在每個可淸除區塊中的每個儲存格可以被測試,並且將每 個可淸除區塊的測試結果寫入其區塊中的—個標頭內。控 ____« —___ 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) ' — I----.---„---參-------、玎------Μ (請先閱讀背面之注意事項再炉ί、本頁} 經濟部智慧財產局員工消費合作社印製 4367 95 at B7 五、發明説明(];) 制器可以從多數個可淸除區塊之中的每一個內讀取標頭資 訊,並且連結標頭資訊’以及將所連結的標頭資訊儲存於 一個可淸除單元中的單一標頭內,而其可淸除單元則包含 所讀取的多數個區塊。 較佳地,在每個可淸除單兀中指定保留至少一個的儲 存格,用來儲存控制資訊。 較佳地,將多數個所保留(控制)儲存格連結在一起’ 用以構成一個儲存格體系’藉以實現位址的轉.換’並且藉 由來自主系統所洪應的邏輯位址中個別的位元’來指示儲 存格體系中的各個層級,使得儲存格體系中最低的儲存格 不是提供所需的實際具體位址,便是提供用來指向另一個 儲存格體系的指標資訊。 經由所給定範例的以下明確之說明’以及參照附圖’ 本發明這些和其它的觀點將會是顯而易見的’在附圖中: 圖1根據本發明的一個實施例,顯示一個具有快閃記 憶體的記憶體系統之'方塊圖; 圖2顯示各種的失效結構,其可能發生於圖1的快閃 記憶體之可淸除區塊其中一個內; 圖3A顯示圖2的可淸除區塊細分爲儲存格’藉以構 成一個可淸除單元; 圖3B顯示一個可淸除單元,包含八個圖2的可淸除 區塊; 圖4顯示圖3A的可淸除單元,在其中的資料結構則 具有五個儲存; 本紙張尺度適用中國國家標準(CNS ) A4規格(2S0X297公釐) n Hi H - - I- --1 I t— - - i lil I 111 -- _______ - n I—____ _ i: I (請先閲讀背面之注意事項再.丨'-,本頁) Γ 4367 9 5 Α7 Β7 五、發明説明(1) 圖5闡述一個儲存格圖案的一部分,如何建構爲圖3A 的可淸除單元; 圖6A顯示一個典型儲存格圖案的格式; 圖6B闡述起因於圖5的圖6A中一個儲存格圖案之一 部分; 圖7顯示一個用來存取圖1記憶體的實體位址之格式 圖8更爲詳紐地顯示圖7的一個部分; 圖9顯示定址體系,用來將圖1系統的邏輯位址轉換 爲實體位址; 圖10更爲詳細地顯示圖9的一部分; 圖11顯示圖9的定址體系之鏈鎖結構,其以第二區塊 位址表的形式; 圖12闡述當儲存於一個控制儲存格時之圖11的表格 , 圖13顯示在可淸除單元刪除之後,其單元中的控制儲 存格之配置; 圖14顯示一個可淸除單元實體位址的格式; 圖15顯示一個具有外加記億裝置的可淸除單元;以及 圖16顯示一個主系統中的層級體系,爲本發明的一個 實施例,其中的控制器履行一個軟體層級。 .圖1顯示一個藉由標準PC介面Μ而連接到個人電腦 12的記憶體系統10。系統10具有一個連接到快閃記憶體 18的控制器16,其快閃記憶體18則是用來儲存資料結構 — — 10 ____ ___ 本紙張尺度適用中國國家榇準(CNS ) Α4規格(210X297公釐) I-----------餐------1T------^ (請先閲讀背面之注意事項再'yi?7本頁) 經濟部智慧財產局員工消费合作社印製 經濟部智慧財產局員工消費合作社印製 436795 at ____137 五、發明説明(沒) 。可以藉由PC I2(讀取/寫入區塊)或藉由控制器16(控制資 訊)來產生這些資料結構。控制器16管理pc 12和快閃記 憶體1δ之間的資料結構之傳輸。 控制器16具有連接到PC介面14的PC介面硬體20 。PC介面硬體20則是經由資料匯流排24,以及也可經由 控制匯流排26而連接到記憶體介面硬體22。一個緩衝記 憶體28也連接到資料匯流排24上,藉以提供將要寫入快 閃記憶體的資料之暫時儲存。一個微處理器29設置於控制 器16之中。微處理器29經由控制匯流排26並且使用揮發 性記憶體33,以聯繫著在非揮發性記憶體30中的PC介面 韌體以及也是在非揮發性記憶體32中的控制器管理韌體。 記憶體介面硬體22經由第二條資料匯流排34聯繫著快閃 記憶體18。 快閃記憶體18典型地爲一個16個百萬位元的記憶體 ,建構其具有三十二個的可淸除區塊(藉由範例區塊36顯 示其中的一個於遍2之中),每個可淸除區塊36則具有用 來儲存64千位元組的記憶位置。 在此的或每個的可淸除區塊36乃是位兀組定址的’並 且具有65535位元組寬的記憶位置38 ’其中的一些典型爲 失效的。經由範例,圖2顯不在區塊36的左上角接近十個 的一群非失效位置38,以及顯示失效位置不同的結構,例 如點4〇(即是一個簡單的單一失效位置)、磁叢42、重複的 磁叢44 '列46和行48的失效。淸楚地,相對於可淸除區 塊36 ’已加大了位置Μ的大小。 (請先閲讀背面之注意事項#/k本頁) 裝. 訂 線 逋用中國國( CNS )八视格_( 21。X 291?1公瘦] Α7 Β7 Γ 436795 五、發明説明(f ) 總是需要一個實體位址來存取記憶體18(其爲一個實 體記憶位置定址);因此,一個邏輯位址(例如藉由控制器 從PC 12所接收到的)必須轉換爲一個實體位址,以放置資 料結構,如同儲存於實體裝置(記憶體18)之中。 所有的讀取/寫入區塊皆塡寫於及讀取於記憶體18, 如同具有一個位元組(八個位元)的字組大小之一串區塊轉 換演算 根據本發明,.控制器16設置一個可淸除單元39,其 包含一個或多個的可淸除區塊36。在一個實施例中,如圖 3A所示的,可淸除單元39由一個可淸除區塊36所構成。 在另一個實施例中,一個可淸除單元39可以包含多數個的 可淸除區塊36,例如,二進位倍數個的可淸除區塊36,其 最終放置於記憶體18中,如圖3B所示,其中的可淸除單 元39包含八個的可淸除區塊36。控制器16,藉由確認儲 存於一個可淸除單元39的資料結構並沒有和另一個可淸除 單元39重複,而確定每個可淸除單元39與所有其它的可 淸除單元39乃是無關的。 在快閃記憶體18尙未倂入記憶體系統1〇之前’測試 記憶位置38並且將所有失效位置的實體位址儲存於一個靠 近每個可淸除單元39開端(最低的實體位址)的標頭中。標 頭爲一連串m個的記憶位置,其第一個位置則是靠近一個 可淸除單元的開端。因此,每個可淸除單元39包含所有失 效位置的實體位址,例如在其可淸除單元39中的40、44 、46_、佔。如此則意謂著,對儲存pc 12所傳送的資料而 ----- - 12 本紙張尺度適用中國國家操準(CNS ) A4規格(210 X 297公釐) ---------.---裝------'玎------0 (請先聞讀背面之注意事項再Γ'.,'本頁) "; 經濟部智慧財產局員工消費合作社印製 Μ 4367 9 5 五、發明説明() 言,m個(非失效)記憶位置爲無效用的’而由於能乡t_以行 來排列這些位置,在一個可淸除區塊中,並不需要這些m 個位置的第一個m(或甚至一連串的)之實體位址。 參照圖3A以及圖2,控制器16藉由將每個可淸除區 .塊36細分爲鄰近的記憶體位置38之群集,而構成一個可 淸除單元39 :這些鄰近的記憶體位置38之群集在此則稱 爲儲存格50。記憶體的實體劃分’實際上並沒有發生。控 制器16全然將每個可淸除單元39視爲某些儲存格的組合 ,其中每個的儲存格則爲一群鄰接的記憶位置。 爲了最小化所需的儲存空間,儲存格的大小較佳地爲 相同的。可以使用不定大小的儲存格’但是其將更爲複雜 〇 每個儲存格50包含512個可定址的記憶位置38 ’而 每個記憶位置38則爲一個位元組寬。其位置依照十六行及 三十二列,排列於一個儲存格之中。 ' 控制器16設置一個用於每個可淸除單元39的儲存格 圖表(cell map),其中的儲存格圖表則儲存單元39中的失 效儲存格(即是具有至少一個失效位置的儲存格)之位置(實 體位址)。設計用於一種特別的記憶體結構(可淸除區塊的 大小以即可淸除單元的大小)之控制器.16 ’以使得其儲存 格圖表相應於可淸除單元39中的位置之格狀陣列。控制器 16則儲存其儲存格圖表,藉以充當在所相應的可淸除單元 39標頭52中的標頭資訊。 •儲存格圖表乃是藉由一個測試單元(並無顯示)所產生 '——._____________1_3 .._______ 本紙張尺度適用中國國家橾準(CNS ) A4规格(210X297公釐) I------Γ--种衣------tT------^ (請先閲讀背面之注意事項再好1本頁) 經濟部智慧財產局員K消費合作社印^ rv Λ367 9 5 五、發明説明(\ () (請先閲讀背面之注意事項再"文本頁) 的,而且在快閃記憶體18的製造測試處理期間’寫入標頭 52中,但若是在裝置操作的有效期間中發生錯誤,則可以 藉由控制器16來更新儲存格圖表。 控制器16控制可淸除單元39的淸除,並且在淸除之 前,將所相應的儲存格圖表寫入可淸除區塊緩衝記憶體28 中,而且在淸除完成之後’則將所相應的儲存格圖表寫回 可淸除單元39中。爲了防備的目的,一個儲存格圖表的備 份副本’同樣地寫.入下一個司*淸除單兀39之中。 如果一個儲存格50包含一個或者多個的失效記憶位置 ,則儲存格被記爲失效的(在儲存格圖表中)’如同以下更 爲詳細的說明。 經濟部智慧財產局員工消費合作社印製 每個可淸除單元39能夠將資料儲存於,對儲存資料有 所效用的儲存格50中,而不儲存於失效的儲存格54中’ 或者不儲存於儲存控制資訊所配置(所保留)的儲存格(控制 儲存格)之中,例如標頭52。因此,每個可淸除單元39的 邏輯容量(對資料儲存有效用的可定址位置之數量)將會小 於每個可淸除單元39的實體容量(可定址記憶位置38的總 數)。 由於失效儲存格54以及控制儲存格56(當讀取或寫入 其讀取/寫入區塊時,此兩者皆略過)的關係’具有相同的 邏輯長度(即是,相同的資料位元數)之讀取/寫入區塊’可 能佔,用不同數量的實體記憶空間(不同數量的記憶位置38) 。此闡述於圖4中。 在圖4中,五個讀取/寫入區塊60a、b、c、d和q,儲 本紙張尺度適用中國國家操举(CNS ) A4规格(21〇X 297公釐) A7 4 3 6 7 9 5 五、發明説明(p"·) 存於可淸除單元39之中。讀取/寫入區塊通常鄰接地儲存 於可淸除單元39的一個或多個行中,例如,第一個的讀取 /寫入區塊可能終止於可用的儲存格51之第1列第4行, 而下一個的讀取/寫入區塊則通常開始於同一個可用儲存格 51的第1列第5行。在下一個列使用之前,讀取/寫入區塊 塡滿於可淸除單元39中的一整個列(辑常延展許多的可用 儲存格51)上。讀取/寫入區塊因此而連接整個可淸除單元 39的位置列,並旦與儲存格50無關。這便是60a、60b、 6c和60d儲存的方式。60b起始位置的位址即是緊接在 6〇a最後位置的位址之後,而60c起始位置的位址則是緊 接在60b最後位置的位址之後,諸如此類。 無論如何,爲了闡述具有許多相鄰近的儲存格之效用 ,所示的讀取/寫入區塊60q位於靠近可淸除單元39的底 部(由於圖4中靠近單元39底部,具有更爲失效的儲存格) 〇 如果沒有失效儲存格54和控制儲存格56時’在一個 單一可淸除單元39之中’將會以鄰接連續的位址’並且以 一列接著一列塡滿爲基礎’來配置讀取/寫入區塊。讀取/ 寫入區塊6〇b、60d和60q具有相同的邏輯長度。如同圖4 中所看到的,讀取/寫入區塊60能夠起始於儲存格中任 何—列的任何一行之位置;砬且’同樣地’可以結束於儲 存格50中任何一列的任何一行之位置。並不需要起始於特 別的一列或一行,或者結束於特別的一列或一行。 每個讀取/寫入區塊60具有一個標頭部份62和〜個資 度適用中國國家標準(CNS )A4規格(210X 297公ϋ " ——" 裝 、STii (請先閲讀背面之注意事項再姥^^本頁) 經濟部智憨財產局員工消費合作社印製 436795 A7 B7___ 五、發明説明() 料部份64。標頭部份62典型地佔用記憶體18十二個的位 元組,並且包含:一個旗標’用來指示在資料部份的資料 是否爲有效的;一個用於讀取/寫入區塊的邏輯區塊位址; 錯誤校正碼(ECC) ’用來保護標頭部份62中的資料;一個 用來指示所儲存的資料之型態的數碼’其型態例如壓縮、 非壓縮、保護和此類的(相類似的);以及下一個讀取/寫入 區塊60的實體位址。 資料部份64則包含所儲存的資料。資料部份64的長 度乃是由系統1〇所決定。一個錯誤校正碼藉由控制器16 可以附加於其資料上。讀從PC I2所接收到的取/寫入區塊 60具有一個固定的大小,而其大小乃是由PC介面韌體30 .所決定。, 將會體會到的是,每個可淸除單元39的邏輯容量乃是 不固定的,其乃使視失效儲存格54的數量和控制儲存格的 數量來決定。 當讀取/寫入區塊60將要寫入快閃記憶體18時,控制 器16則決定區塊60 —個合適的起始位址。起始位址的選 擇視控制器所履行的特定演算法決定。例如,在增加到下 一列之前,控制器16可以塡滿一整個列:同樣地,在增加 到下一行之前,控制器16可以塡滿一整個行。選擇所使用 的演算法則視記億體裝置18的結構和特性決定。 •在此實施例中,整個的列首先被塡滿。即是,最低的 可使用列位址用來當作起始位址,並且在列位址增加之前 ,塡滿一整個列(藉由增加行位址)。因此,控制器16搜尋 張尺度逍用中國國家標準(CNS ) Λ4規格(210X29】?公釐) ---------„---;^------訂------.^ (讀先閱讀背面之注$項再Γί·.本頁) 經濟部智慧財產局員工消費合作社印¾. 經濟部智慧財產局員工消費合作社印製Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs p Λ36795 at _____B7___ V. Description of the Invention (I) The present invention relates to a non-volatile memory system for storing and retrieving data. The memory system includes a memory with a memory location. Memory, and its memory locations can be written individually, but can only be erased by the location of the entire block, and its system also includes a controller for controlling access to these memory locations; the invention also relates to a non- Volatile memory is used in such a non-volatile memory system and relates to a controller for controlling non-volatile memory. In particular, the present invention relates to a flash memory system having a defective memory location, and to a controller for a flash memory. Flash EPROM (programmable read-only memory) can be used to store non-volatile data in the electronics industry. The architecture of flash memory devices has locations that can be individually written but can only be erased collectively, which is called erasable blocks. Because the transistor clusters in flash memory are connected by a common erasure circuit, such a structure is produced. Therefore, the size of the erasable block (the number of storage locations) is determined by the structure of the device, and it has been formulated at the time of design and manufacture, and cannot be modified by the user. One application of data storage is to store data structures generated by, for example, a personal computer (PC). If flash memory with invalid memory locations is used to store data structures, a problem arises because these invalid memory locations cannot be reliably used for data storage. • One way to solve the problem of having invalid locations in memory is to avoid the use of any erasable blocks that contain invalid locations by the flash controller. However, if the erasable block containing the location of failure is never used (marked by the controller __-_ 4___, this paper size applies the Chinese National Standard (CNS) A4 specification (2I0X297 mm) -------- --- ^ ------ ΐτ ------ ii (Please read the precautions on the back before vik this page) Printed by the Consumer Cooperatives of the Yttrium Evil Property Bureau of the Ministry of Economic Affairs ′ '4367 9 5 at _______ Β7 5 , The description of the invention (>) is marked as unusable), it may greatly waste the available memory storage space (depending on the size of the erasable block and the number of usable memory locations), This results in low memory usage (low ratio of available memory locations to total memory location). It is an object of the present invention to provide a non-volatile memory system which removes or alleviates the above disadvantages. It is an object of the present invention to provide a non-volatile memory system including a memory having a location where data cannot be stored, but the memory can be used for efficient storage and retrieval of data structures. According to a first aspect of the present invention, a memory system connected to a host system is provided. The memory system includes: a non-volatile memory having a memory location, and a controller for writing a data structure into the memory. And remove the data from the memory, and construct its system so that the location can be written individually, but can only be eliminated by the collective block of the location; its progress is: the controller sets at least one Each erasable unit includes at least one erasable block, and the controller subdivides each erasable unit into a location group (each group is referred to herein as a cell), and based on each A cell, the controller writes the data structure into each cell and reads the data structure from each cell. With the advantages of the present invention, the memory is reconstructed as a cell, so that the controller can avoid the use of Failure of individual cells' without having to avoid the use of erasable erasable blocks β This effectively increases memory ----------- S _ Use Chinese National Standard (CNS > Α4 Specification (210 ×: 297 mm) --------- ^ --- t .------ ΐτ ------ 0 (Please read first Note on the back f .. this page) · \ A7 B7 Π367 9 5 An invention description (today) The usage rate of Lin. An erasable unit may only contain one erasable block. Instead, An erasable unit may contain a plurality of erasable blocks; it is jointly declared that the erasable block is a multiple of a binary. It will be understood that one erasable block is typically more than one cell. Before connecting to the main system, test the hundreds of millions of positions in each cell, and if the failure occurs even in only one cell, the controller treats the entire cell as unusable In addition to the data structure, the entire cell is considered to be used to store the data structure. It will be understood that the proper noun memory location can refer to a single bit stored in the billion body; however, Proper noun cells refer to most bits stored in memory. A typical cell can Stores 256 bytes or 512 bytes. It will be understood that the locations that can be individually written can be composed of a plurality of physical memory locations, such as a row of physical memory, so that in such an example An entire row is the smallest unit of memory that can be individually written. Non-volatile memory can include multiple memory devices or only a single memory device. Preferably, the controller In the form of the main controller of the auxiliary controller, the auxiliary controller is incorporated into its memory device or each memory device. Alternatively, the controller is in the form of a single controller, and the single controller controls Its memory pack ----------- This paper wave standard is applicable to China National Standard (CNS) 8-4 specifications (21〇 > < 297mm) (Please read the precautions on the back before you copy this Page) Binding and ordering-Printed by the Ministry of Economic Affairs, Smart Money, J. Bureau Employee Cooperative Cooperative, printed r 436795 5. Invention Description (+) or each memory device. Preferably, the controller specifies that at least one of the available cells in each of the erasable units is reserved to have invalid cells that are invalidated, and the cells are reserved for the control information, and the available cells. To store data received from the main system. Preferably, the reserved cells are used to store the address translation information, and the address translation information is used to convert the address from the host system into an address suitable for storing memory. Preferably, since there are a plurality of reserved cells linked together, the address conversion is realized. Preferably, each reserved cell is used to store pointer information, and the pointer information is used to point to the next reserved cell until the last reserved cell is reached, and the pointer information is to point to An address in memory, which stores an address suitable for accessing memory. Preferably, the reserved cell is constructed so as to have a plurality of entries, wherein the entry stores a plurality of fields, and the fields stored in one entry are used To point to a field stored in another registry element, and only treat fields that were recently written in one registry field as valid. According to a second aspect of the present invention, a controller is provided for a non-volatile memory with a position, which can be individually written, but can only be erased by a position block. In use, the controller Set at least one erasable unit, each of which contains at least one erasable unit _________2______ Reference paper size Applies to China National Standard (CNS) A4 (2 丨 0X297 mm) (Please read the back Note again J > i this page)-Pack. Ordered by the Intellectual Property Bureau of the Ministry of Economic Affairs, printed by the bone-eliminating cooperative of the Ministry of Economic Affairs, printed by the Employees ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs, printed by the Consumer Cooperative of the Ministry of Economic Affairs, 4367 9 5 5. Description of invention (ζ :) Furthermore, the controller subdivides each erasable cell into a cell ', and reads the data structure from the cell and writes the data structure into the cell based on each cell'. The proper data structure (data struture) is used to cover all the data stored in memory. Therefore, the proper noun data structure covers read / write biock (that is, a block of data transmitted between a PC and memory) and control information (that is, for example, Information generated by the memory controller). According to the third aspect of the present invention, a non-volatile memory for the controller is provided. The memory with a position can be individually written, but it can only be erased by the position block to construct a non-volatile memory. The memory is provided with at least one erasable unit, and each erasable unit contains at least one erasable block, and each erasable unit is subdivided into a memory called a cell. Groups of locations so that on each basic cell, the controller reads the data structure in the cell and writes the data structure into the cell. Because each cell in each of the erasable units tested has been constructed, non-volatile memory is constructed, and the test results of each unit are written into a set of locations to form in its unit A header in each unit, so that each unit has a header containing header information. Alternatively, in a deletable unit containing a plurality of blocks, in each deletable area Each cell in the block can be tested, and the test results of each erasable block are written into a header in its block. Control ____ «—___ This paper size applies to the Chinese National Standard (CNS) A4 specification (210X297 mm) '— I ----.---„ -------------, 玎- ---- Μ (Please read the precautions on the back before making this page again.} Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 4367 95 at B7 V. Description of the invention (); Read header information in each of the blocks, and link header information 'and store the linked header information in a single header in an erasable unit, and its erasable unit is Contains most of the read blocks. Preferably, at least one cell is designated to be reserved in each erasable unit for storing control information. Preferably, a plurality of reserved (control) cells are stored Linked together 'to form a cell system' to achieve address translation. And 'and to indicate the various levels in the cell system by the individual bits in the logical address that the main system responds to, The lowest cell in the cell system either provides the actual specific address required, or Provides indicator information to point to another cell system. These and other aspects of the invention will be apparent from the following explicit description of the given example 'and with reference to the drawings' in the drawings: Figure 1 according to An embodiment of the present invention shows a block diagram of a memory system with flash memory. FIG. 2 shows various failure structures that may occur in one of the erasable blocks of the flash memory of FIG. 1. Fig. 3A shows the erasable block of Fig. 2 subdivided into cells to form an erasable unit; Fig. 3B shows an erasable unit including eight erasable blocks of Fig. 2; Fig. 4 shows The erasable unit in FIG. 3A has five storages in the data structure; The paper size is applicable to the Chinese National Standard (CNS) A4 specification (2S0X297 mm) n Hi H--I- --1 I t-- -i lil I 111-_______-n I —____ _ i: I (please read the precautions on the back .'-, this page) Γ 4367 9 5 Α7 Β7 5. Description of the invention (1) Figure 5 illustrates How a part of a cell pattern is structured as Deletion unit; FIG. 6A shows the format of a typical cell pattern; FIG. 6B illustrates a part of a cell pattern in FIG. 6A resulting from FIG. 5; FIG. 7 shows a physical address used to access the memory of FIG. Figure 8 shows a part of Figure 7 in more detail; Figure 9 shows the addressing system used to convert the logical address of the system of Figure 1 into a physical address; Figure 10 shows a part of Figure 9 in more detail; 11 shows the chain structure of the addressing system of FIG. 9 in the form of an address table of the second block; FIG. 12 illustrates the table of FIG. 11 when stored in a control cell, and FIG. After that, the configuration of the control cell in its unit; Figure 14 shows the format of a unit address that can be removed; Figure 15 shows a unit that can be added with a billion-dollar device; and Figure 16 shows the The hierarchical system is an embodiment of the present invention, in which the controller implements a software hierarchy. FIG. 1 shows a memory system 10 connected to a personal computer 12 through a standard PC interface M. The system 10 has a controller 16 connected to a flash memory 18, and the flash memory 18 is used to store data structures-10 ____ ___ This paper size is applicable to China National Standards (CNS) A4 specifications (210X297) Li) I ----------- Meal ------ 1T ------ ^ (Please read the notes on the back before 'yi? 7 pages) Intellectual Property Bureau of the Ministry of Economic Affairs Printed by the Employee Consumption Cooperative. Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs. Printed by the Consumer Consumption Cooperative at 436795 at ____137. 5. Description of the invention (none). These data structures can be generated by PC I2 (read / write block) or by controller 16 (control information). The controller 16 manages the transmission of the data structure between the pc 12 and the flash memory 1δ. The controller 16 has a PC interface hardware 20 connected to the PC interface 14. The PC interface hardware 20 is connected to the memory interface hardware 22 via a data bus 24 and also via a control bus 26. A buffer memory 28 is also connected to the data bus 24 to provide temporary storage of data to be written to the flash memory. A microprocessor 29 is provided in the controller 16. The microprocessor 29 uses the volatile memory 33 via the control bus 26 to communicate with the PC interface firmware in the non-volatile memory 30 and the controller management firmware also in the non-volatile memory 32. The memory interface hardware 22 is connected to the flash memory 18 via a second data bus 34. The flash memory 18 is typically a 16-megabit memory, constructed with thirty-two erasable blocks (one of which is shown in Pass 2 by the example block 36), Each erasable block 36 has a memory location for storing 64 kilobytes. Some or all of the erasable blocks 36 here are bit-addressed ' and have a memory location 38 ' of 65535 bytes wide, some of which are typically invalid. By way of example, FIG. 2 shows a group of non-failure locations 38 that are not close to ten in the upper left corner of block 36, and shows structures with different failure locations, such as point 40 (that is, a simple single failure location), magnetic cluster 42, Repeated magnetic clusters 44 'column 46 and row 48 fail. Specifically, the size of the position M has been increased relative to the erasable block 36 '. (Please read the precautions on the back # / k this page first). Binding. The Chinese national (CNS) eight-view grid _ (21. X 291? 1 thin) Α7 Β7 Γ 436795 V. Description of the invention (f) A physical address is always needed to access memory 18 (which addresses an actual memory location); therefore, a logical address (such as that received by the controller from PC 12) must be converted to a physical address To store the data structure, as if stored in a physical device (memory 18). All read / write blocks are transcribed and read in memory 18, as if having a byte (eight bits A block conversion algorithm for a string size of a block size according to the present invention. The controller 16 is provided with an erasable unit 39 which includes one or more erasable blocks 36. In one embodiment, such as As shown in Fig. 3A, the erasable unit 39 is composed of one erasable block 36. In another embodiment, one erasable unit 39 may include a plurality of erasable blocks 36, for example, two Carryable multiple erasable blocks 36 are finally placed in the memory 18, as shown in FIG. 3B The erasable unit 39 includes eight erasable blocks 36. The controller 16 confirms that the data structure stored in one erasable unit 39 does not overlap with the other erasable unit 39, and Make sure that each erasable unit 39 is irrelevant to all other erasable units 39. Before the flash memory 18 is not inserted into the memory system 10 ', test the memory location 38 and replace all entities in the invalid location The address is stored in a header near the beginning (lowest physical address) of each erasable unit 39. The header is a series of m memory locations, and the first location is near a erasable unit Beginning. Therefore, each erasable unit 39 contains the physical addresses of all invalid locations, such as 40, 44, 46_, and occupies in its erasable unit 39. This means that the storage pc 12 is transmitted ------12 This paper size applies to China National Standards (CNS) A4 specification (210 X 297 mm) ---------.------------------ '玎 ------ 0 (please read the precautions on the back before Γ'., 'This page) " Cooperative printed M 4367 9 5 V. Description of the invention () The m (non-invalid) memory locations are invalid, and because the energy t_ is used to arrange these locations in rows, in a deletable block, The physical address of the first m (or even a series) of these m locations is not required. Referring to Figure 3A and Figure 2, the controller 16 subdivides each erasable area. Block 36 is divided into adjacent memories The clusters of body positions 38 constitute an eradicable unit 39: these clusters of adjacent memory positions 38 are referred to herein as cell 50. The physical division of memory ’has not actually happened. The controller 16 considers each erasable unit 39 as a combination of certain cells, where each cell is a group of adjacent memory locations. To minimize the required storage space, the cell sizes are preferably the same. Cells of variable size can be used but they will be more complex. Each cell 50 contains 512 addressable memory locations 38 'and each memory location 38 is one byte wide. Its position is arranged in a cell according to 16 rows and 32 columns. 'The controller 16 sets a cell map for each erasable unit 39, wherein the cell map stores the failed cells in the unit 39 (that is, cells with at least one failed position) Location (physical address). Controller designed for a special memory structure (the size of the block can be deleted to eliminate the size of the unit). 16 'so that its cell chart corresponds to the position of the position in the erasable unit 39状 Array. The controller 16 stores its cell chart to serve as header information in the corresponding erasable unit 39 header 52. • Cell chart is generated by a test unit (not shown) '——._____________ 1_3 .._______ This paper size applies to China National Standard (CNS) A4 (210X297 mm) I ----- -Γ--Seed coat ------ tT ------ ^ (Please read the precautions on the back first and then the first page) Printed by K Consumer Cooperative, Member of Intellectual Property Bureau, Ministry of Economic Affairs ^ rv Λ367 9 5 V. The description of the invention (\ () (please read the precautions on the back and then the "text page") and write it into the header 52 during the manufacturing test process of the flash memory 18, but if it is during the valid period of the device operation If an error occurs, the cell diagram can be updated by the controller 16. The controller 16 controls the erasure of the erasable unit 39, and writes the corresponding cell diagram to the erasable block before erasing. In the buffer memory 28, and after erasing is completed, 'the corresponding cell chart is written back to the erasable unit 39. For precautionary purposes, a backup copy of the cell chart is likewise written. Enter the next Division * Divide the unit 39. If a cell 50 contains one or If there are multiple invalid memory locations, the cell is marked as invalid (in the cell chart) as described in more detail below. Each consumer unit of the Intellectual Property Bureau of the Ministry of Economic Affairs ’s consumer cooperative can print each deletable unit 39. Data is stored in cell 50, which is useful for storing data, but not in invalid cell 54 'or in the cell (control cell) where the control information is configured (retained) , Such as header 52. Therefore, the logical capacity of each erasable unit 39 (the number of addressable locations effective for data storage) will be less than the physical capacity of each erasable unit 39 (the addressable memory location 38 Total). Since the relationship between the stale cell 54 and the control cell 56 (which are both ignored when reading or writing its read / write block) has the same logical length (ie, Read / write blocks of the same number of data bits) may occupy different amounts of physical memory space (different number of memory locations 38). This is illustrated in Figure 4. In Figure 4, five reads Fetch / write block 60a, b, c, d, and q, the paper size of this paper is applicable to China National Standards (CNS) A4 specification (21〇X 297 mm) A7 4 3 6 7 9 5 V. Description of invention (p " ·) is stored in Erasable unit 39. Read / write blocks are usually stored contiguously in one or more rows of erasable unit 39, for example, the first read / write block may terminate when available Column 51, row 4 of cell 51, and the next read / write block usually starts at column 1, row 5 of the same available cell 51. Before the next column is used, read / The write block is full on an entire column in the erasable unit 39 (the number of available cells 51 is often extended). The read / write block thus connects the entire column of the erasable unit 39 and is independent of the cell 50. This is how 60a, 60b, 6c and 60d are stored. The address of the start position of 60b is immediately after the address of the last position of 60a, and the address of the start position of 60c is immediately after the address of the last position of 60b, and so on. However, in order to illustrate the utility of having many adjacent cells, the read / write block 60q shown is located near the bottom of the erasable unit 39 (because it is near the bottom of the unit 39 in FIG. 4, it has a more invalid (Cells) 〇 If there is no invalid cell 54 and control cell 56, 'in a single erasable unit 39' will be based on adjacent consecutive addresses 'and based on the row after row full' configuration reading Fetch / write block. The read / write blocks 60b, 60d, and 60q have the same logical length. As can be seen in Figure 4, the read / write block 60 can start at any position of any row in the cell; and 'samely' can end any of any row in the cell 50 The position of a line. It does not need to start in a particular column or row, or end in a particular column or row. Each read / write block 60 has a header portion 62 and ~ qualifications applicable to China National Standard (CNS) A4 specifications (210X 297 public ϋ " —— " 装 、 STii (Please read the back first Please note again ^^ this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 436795 A7 B7___ 5. Description of the invention () Material part 64. The header part 62 typically occupies 18 twelve of the memory Bytes, and contains: a flag 'used to indicate whether the data in the data part is valid; a logical block address to read / write the block; an error correction code (ECC)' used To protect the data in the header section 62; a digital 'indicating' the type of the stored data; its type such as compressed, uncompressed, protected, and the like (similar); and the next read / Write the physical address of block 60. The data part 64 contains the stored data. The length of the data part 64 is determined by the system 10. An error correction code can be appended to it by the controller 16. On the data. Read the fetch / write block 60 received from PC I2 has It is a fixed size, and its size is determined by the PC interface firmware 30. It will be appreciated that the logical capacity of each erasable unit 39 is not fixed, which is caused by the failure storage The number of cells 54 and the number of control cells are determined. When the read / write block 60 is to be written to the flash memory 18, the controller 16 determines the block 60-a suitable starting address. The selection of the starting address is determined by the specific algorithm performed by the controller. For example, the controller 16 may fill a whole column before adding to the next column: Similarly, the controller 16 may add before adding to the next column. A full row. The algorithm used depends on the structure and characteristics of the billion-body device 18. • In this embodiment, the entire column is filled first. That is, the lowest available column address is used to As the starting address, and before the column address is increased, fill the entire column (by adding the row address). Therefore, the controller 16 searches for the Zhang scale and applies the Chinese National Standard (CNS) Λ4 specification (210X29) ? Mm) --------- „---; ^ ------ Order ------. ^ ( Please read the note on the back of the page before reading. This page) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs.

l··: 4367 9 5 A A7 B7 -丨 I ' 1 ' " 五、發明説明(\屮) 最低的可使用列位址之儲存格圖表。 控制器則在到達第一個失效儲存格之前’判斷用來儲 存讀取/寫入區塊60的可用儲存格之數量。由可用的儲存 格51之數量及每個儲存格51具有十六行的位置’能容易 地判斷出有效記憶位置的數量’所以有效記憶位置、的數量 等於第一個失效儲存格之前的可用儲存格51之數量乘以十 六。 如果能夠儲存整個區塊60 ’而不與失效儲存格54(或 控制儲存格56)有所衝突’則儲存區塊60 °如此的範例爲 圖4中的區塊60a和6c。 如果,無論如何,在到達失效儲存格54(或控制儲存 格56)之前,只有區塊60的一部分能夠儲存’則此部份能 夠傳輸至記憶體並且儲存’之後’當控制器16判斷下一個 可用的儲存格51之位置時’則暫停其傳輸。控制器16然 後判斷有多少資料能夠儲存於下一個可用的儲存格51以及 緊接著下一個可用儲存格51的失效儲存格54(或控制儲存 格56)之間。控制器16然後增加將寫入的位址,以使得將 寫入的位址等於第一個可用的位址,並且開始資料的傳輸 ,諸如此類。 可替代地,如果快閃記憶體裝置的特性並不輕易地允 許設定一個用於資料傳輸的新位址,例如,由於建構其裝 置乃是用於連續的資料傳輸,則可以藉由控制器16產生一 個資料傳輸時脈,並且供給記憶體裝置,而並無實際地傳 輸資料。當記憶體裝置(由時脈信號所驅動)中的內部存取 一__17__一一 ' 本紙度ϋ中囤囤家標準(CNS〉A4規格(2HTX297公釐) 一"^~ ---iw---^---押衣------1T------0 (請先閲讀背面之注$項再妒九本頁) 4367 95 五、發明説明(丨<) 指標達到下一個可用的位址時,其資料傳輸則藉由控制器 16而重新開始。 因此,分裂了 PC 12和快閃記憶體18之間的資料傳輸 :每一個節段(segment)儲存於一組連續的記憶位置中。每 個節段的長度等於連續可用的記憶位置之數目° 一個節段 在到達失效儲存格54(或控制儲存格56)之前,終止於最後 一個位址,並且在達到下一個可用儲存格51之,時重新開 始(於儲存格中最低的可用位址)。因此’控制器16決定將 要傳輸的節段之大小。當許多的失效儲存格出現於記憶體 18時,這種節段傳輸的效用則藉由讀取/寫入區塊60q來 闡述。 由於在讀取記憶體之時,必須跳過失效儲存格(和控制 儲存格),則當資料將從快閃記憶體中讀取時,將進行相類 似的程序。控制器16以下列的方式略過儲存格。當一個節 段中的最後一個位址被讀取時,控制器16則在緊接於失效 儲存格54或控制儲存格56之後的第一個可用儲存格51中 ( ,增加其位址爲第一個可用的列位址。 當讀取/寫入區塊60被讀取,以及任何一個相應其讀 取/寫入區塊60的ECC所指示的資料讀取之中有所錯誤時 +,控制器16則識別出特定的失效位元,並且將包含這些位 元的可用儲存格或儲存格51標記爲失效的,以及更新相應 的儲存格圖表。可替代地,控制器16可以將所有包含讀取 /寫人區塊60的可用儲存格或儲存格SI標記爲失效的。以 如此的方式,控制器16在每一個時間資料中,確認每個整 Γ-Τ---――__1£_ 表張尺度適用中國國家榣準(CMS〉Μ規格(2〖0Χ297公釐) ^ ' 436 7 9 5 A7 B7 五、發明説明(α) 體的可用儲存格51。因此,控制器16能夠識別出現於可 用的儲存格51中之新的失效。在另一個實施例中,可以使 用一種自我測試程序。在自我測試的起始’控制器16將資 料寫入並且之後讀出每一個可用儲存格51,用以判斷任何 一個可用儲存格是否具有失效。 控制器16使用一個寫入指標(如GB 2 291 911所說明 的)來指示下一個將要寫入的記億體位置,並且使用一個淸 除指標來指示下一.個將要淸除的可淸除單兀39。總是藉由 其寫入指標的指示,而將讀出/寫入區塊寫入所指示的位置 ,控制器16並且確認在寫入指標和淸除指標之間至少有一 個已淸除的可淸除單元39。控制器也使用一個單獨的控制 儲存格篇入指標(control cell write pointer),來指示控制資 訊將要寫入的記憶位置。其控制儲存格寫入指標僅僅指示 控制儲存格56中的位置。 圖5顯示一個具有標記(mark)的可淸除單元,用來闡 述如何建構一個儲存格圖表的一部份,其中儲存格的行編 號爲〇至15,而列編號則爲0至7。 圖6A顯示一個典型的儲存格圖表70之格式。每個儲 存格圖表70包含··與圖表70有關的可淸除單元39之位址 h ;在可淸除單元39中有效用的可控制儲存格56之數目 72 ;在可淸除單元39中可用的儲存格51之數目73 ;以及 —組的位元74,其表示在區塊36中的可用儲存格51。 圖6B闡述一組位元74如何從圖5產生。在圖6B中 ’以一種相應於圖5的儲存格型式之格狀型式來顯示一組 狀财咖家標準(CNS ) A4· (21Qx297公疫 (請先聞讀背面之注意事項再Ϊ.本頁) --5 經濟部智慧財產局員工消費合作社印製 鎚 越 財 產 局 X 消 費 合 作 社 印 製 436795 A7 B7 五、發明説明(j ) 的位元74。在圖6B中’爲了淸楚的目的’而顯示一個外 加的第一行75和第一列76 ’其在實際上並不儲存。包含 —和零(一組的位元74)的八乘以十六陣列之儲存格圖表 ,儲存於每個可淸除單元39的標頭52中。在一個包含多 數個可淸除區塊36的可淸除單元39之實施例中,一個儲 存格圖表可以只儲存於其可淸除單元的第一個可淸除區塊 中。可替代地,一個儲存格圖表重疊到數個的可淸除區塊 ,其儲存格圖則是用於具有多數個可淸除區塊的可淸除單 元,特別是在每一個可淸除區塊如果包含一列或少數列的 _記憶位置時。第一彳了(Drow)75指不儲存格的列編威’而弟 一列76則指示儲存格的行編號(如同圖5中的每個標記方 式)。行75和列76並不儲存於標頭52中。如果一個儲存 格爲可用的,則將數字〇插入相關的列和行之中;無論如 何,如果一個儲存格爲不可用的(例如,由於其爲有失效的 儲存格,或爲控制儲存格),則將數字1插入相關的列和行 之中。就外加的防護措施而言’ 一個用於可清除單元39的 儲存格圖表70之備份’可以儲存於下一個可淸除單元39 的標頭中。如此乃是爲了擔保:如果可淸除單元39標頭 52中的一個位置失去作用時,則可以藉由下一個標頭52 而將儲存於標頭52中的儲存格圖表7〇恢復原狀。儲存格 圖表70中的資訊由一個ECC所防護,而其ECC則是倂入 標頭52所儲存的資訊中。 標頭52包含一個用於相關可淸除單元39的儲存格圖 表70、一個用於前一個的可淸除單元39之儲存格圖表70 (請先閲讀背面之注意事項再ί'本頁) -裝· 訂 線 本紙l · ·: 4367 9 5 A A7 B7-丨 I '1' " V. Description of the invention (\ 屮) The cell chart with the lowest available column address. The controller judges the number of available cells used to store the read / write block 60 before reaching the first invalid cell. From the number of available cell 51 and the position of each cell 51 with sixteen rows 'the number of valid memory positions can be easily determined' so the number of valid memory positions equals the available memory before the first invalid cell Multiply the number of cells by sixteen. If it is possible to store the entire block 60 'without conflicting with the stale cell 54 (or the control cell 56)', the example of storing the block 60 ° is block 60a and 6c in FIG. If, in any case, before reaching the invalid cell 54 (or control cell 56), only a part of the block 60 can be stored 'then this part can be transferred to memory and stored' after 'when the controller 16 judges the next When the position of available cell 51 is', its transmission is suspended. The controller 16 then determines how much data can be stored between the next available cell 51 and the stale cell 54 (or control cell 56) immediately after the next available cell 51. The controller 16 then increases the address to be written so that the address to be written is equal to the first available address, and starts the transmission of data, and so on. Alternatively, if the characteristics of the flash memory device do not easily allow setting a new address for data transmission, for example, because the device is constructed for continuous data transmission, the controller 16 can be used A data transmission clock is generated and supplied to the memory device without actually transmitting the data. When the internal access in the memory device (driven by the clock signal)-__17__ one-one "in the paper store standard (CNS> A4 specification (2HTX297 mm)-" ^ ~ --- iw --- ^ --- Lily clothes ------ 1T ------ 0 (Please read the note on the back and then jealous nine pages) 4367 95 V. Description of the invention (丨 <) When the indicator reaches the next available address, its data transmission is restarted by the controller 16. Therefore, the data transmission between the PC 12 and the flash memory 18 is split: each segment is stored in A set of consecutive memory locations. The length of each segment is equal to the number of consecutively available memory locations. A segment terminates at the last address before reaching the invalid cell 54 (or control cell 56), and When the next available cell 51 is reached, it restarts (at the lowest available address in the cell). Therefore, the 'controller 16 determines the size of the segment to be transmitted. When many invalid cells appear in memory 18 The utility of this segment transfer is explained by reading / writing the block 60q. , You must skip the invalid cell (and control cell), when the data will be read from the flash memory, a similar process will be performed. The controller 16 skips the cell in the following way. When a section When the last address in the segment is read, the controller 16 is in the first available cell 51 immediately after the failed cell 54 or the control cell 56 (, increasing its address to be the first available When the read / write block 60 is read, and there is an error in any of the data reads indicated by the ECC corresponding to its read / write block 60+, the controller 16 Then identify specific invalid bits, and mark the available cells or cells 51 containing those bits as invalid, and update the corresponding cell chart. Alternatively, the controller 16 may read all The available cell or cell SI of the writer block 60 is marked as invalid. In this way, the controller 16 confirms in each time data each integer Γ-Τ -------__ 1 £ _ sheet Standards apply to China National Standards (CMS> M specifications (2 〖0 × 297mm ) '436 7 9 5 A7 B7 V. Description of the invention (available in the available cell 51 of the volume (α). Therefore, the controller 16 can identify a new failure that appears in the available cell 51. In another embodiment, A self-test procedure can be used. At the beginning of the self-test, the controller 16 writes data and then reads out each available cell 51 to determine whether any available cell has a failure. The controller 16 uses a write Input indicator (as explained in GB 2 291 911) to indicate the next memory location to be written, and use an erasure indicator to indicate the next erasable unit 39 to be erased. The controller 16 also confirms that at least one of the erased and written indicators has been deleted between the write indicator and the erase indicator by the instruction of the write indicator and the read / write block is written to the indicated position. Annihilation unit 39. The controller also uses a separate control cell write pointer to indicate the memory location where control information will be written. Its control cell write indicator only indicates the position in the control cell 56. Figure 5 shows a eliminable unit with a mark to illustrate how to construct a part of a cell chart, where the row numbers of the cells are 0 to 15 and the column numbers are 0 to 7. FIG. 6A shows the format of a typical cell chart 70. FIG. Each cell chart 70 contains the address h of the erasable unit 39 related to the chart 70; the number of controllable cells 56 that are effective in the erasable unit 39 72; in the erasable unit 39 The number 73 of available cell 51; and-a set of bits 74, which represents the available cell 51 in block 36. FIG. 6B illustrates how a set of bits 74 are generated from FIG. 5. In FIG. 6B, a set of state-like financial standards (CNS) A4 (21Qx297) is displayed in a grid pattern corresponding to the cell pattern of FIG. 5 (please read the precautions on the back before reading this.) Page) --5 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, printed by Hammer Property Agency X Printed by the Consumer Cooperatives, 436795 A7 B7 5. Bit 74 of the invention description (j). In Figure 6B 'for the sake of clarity' And an extra first row 75 and first column 76 'are not actually stored. A cell chart of eight by sixteen arrays containing-and zero (set of bit 74) is stored in each In the header 52 of each erasable unit 39. In an embodiment of the erasable unit 39 including a plurality of erasable blocks 36, a cell chart may be stored only in the first erasable unit of the erasable unit 39. There are several erasable blocks. Alternatively, a cell chart is overlapped with several erasable blocks, and the cell map is used for erasable units with a large number of erasable blocks, especially Is the memory location in each erasable block if it contains one or a few rows The first Drow 75 refers to the column editor of the cell, and the younger column 76 indicates the row number of the cell (as each labeling method in Figure 5). Row 75 and column 76 are not stored in In header 52. If a cell is available, insert the number 0 into the relevant column and row; however, if a cell is unavailable (for example, because it is a stale cell, or To control the cell), insert the number 1 into the relevant column and row. For added protection, 'a backup of cell chart 70 for clearable unit 39' can be stored in the next erasable In the header of cell 39. This is to guarantee that if one of the positions in header 52 of cell 39 can be deleted, the next header 52 can be used to store the cell in header 52. Figure 70 is restored. The information in cell chart 70 is protected by an ECC, and the ECC is inserted into the information stored in the header 52. The header 52 contains a storage for the relevant erasable unit 39 Grid chart 70, one divisible for the previous Cell chart 70 in unit 39 (please read the precautions on the reverse side before going to this page)-binding and binding

J2£L 張尺度適用中國國家標準(CNS〉A4規格(2丨0 X 297公釐) Α7 Γ' 4367 9 5 五、發明説明(丨名) 、一個唯一的識別方式、以使得控制器能夠識別標頭52、 一個寫入起始旗標'一個寫入終止旗標、一個控制儲存格 起始旗標、一個控制儲存格終止旗標'以及不是控制儲存 格標頭資訊便是控制儲存格的指標資訊。 在可淸除單元39清除之後的初始時間’當一個讀取/ 寫入區塊寫入可淸除單元39時,則聲明寫入起始旗標(設 定爲非淸除狀態)。而當讀取/寫入區塊塡滿可淸除單元39 時,則聲明寫入終止旗標。 在可淸除單元39清除之後,當首次寫入控制儲存格( 除了包含標頭的控制儲存格),則聲明控制儲存格起始旗標 。而當所有的控制儲存格皆被寫入時’則聲明控制儲存格 終止旗標。 如果在用於整個標頭資訊之單一個控制儲存格中的空 間不足,則只需要控制其#存格標頭指標’在範例中’控 制儲存格標頭資訊乃是儲存於另一個由控制儲存格指標所 指示的控制儲存格之中。 圖7顯示用於存取(即是,從其讀取以及寫入其中)快 閃記憶體18的一個實體位址80之格式。實體位址8〇具有 第一個欄位82,表示一個可淸除單元39。在此實施例中, —個可淸除單元39的大小等於一個可淸除區塊36。實體 位址80也具有第二個欄位84,其包含可淸除單元39中的 一個·位址。在第二個欄位84中的位址,有關於可淸除單元 39的總實體位址空間。第二個欄位84乃是用來定義可淸 除單元39中任何一個資料結構的位置。 本紙張尺度適用中國國家標準(CNS ) Α4规格(210 X 297公釐) -----^-----襄------IT------^ (請先閱讀背面之注意事項再^^:本頁) ί 經濟部智慧財產局員工消費合作社印製 經濟部智慧財產局員工消費合作社印製 4 3 6 7 9 5 at ____B7____ 五、發明説明(l f ) 圖8顯示更爲詳細的第二個欄位84。第二個欄位84 包含一儲存格列92、一列94、一儲存格f了 96、以及一 fT 98。儲存格列92乃是將要存取的可淸除單元39中儲存格 一個特別的列。列94則是將要存取的位置(儲存格列92中 )特定列的實體位址。此列的位址則是有關儲存格中的第一 列。同樣地,儲存格行96乃是將要存取的司'淸除單元39 中儲存格一特定的行,而行98則是將要存取的位置(儲存 格行96中)特定行的實體位址。此行的位址即是有關儲存 格中的第一行。 儲存於實體位址空間中的資料結構之次序’完全無關 於每個資料結構的邏輯區塊位址(LBA)。相應於任何一個 特定邏輯區塊的資料結構,可以配置於任何一個實體位址 的空間中。指定每一個LBA的資料結構實體位址(PBA)設 置於一組表格中,如所知的區塊位址表(BAT),而能夠經 由一種指標表(PT)的樹狀結構來存取其BAT,如圖9和10 中所闡述的。 一種靴帶式結構1〇〇儲存於記憶體18第一個可淸除 單元39中的一個控制儲存格56之中。就防護的目的而言 ,靴帶式結構100的備份也儲存於記憶體18第二個和最後 一個可淸除單元39之中。用來儲存靴帶式結構1〇〇的控制 儲存格56,乃是藉由控制器16來選擇其位置,而其位置 則是·緊接在包含標頭52的控制儲存格之後的第一個控制儲 存格之位置。 靴帶式結構100乃是第一個的資料結構,爲控制器16 _______ _22_____ 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) I--------^------ΐτ------^ (请先閲讀背Ι&之注意事項再r'-私本頁) 經濟部智慧財產局員工消費合作社印製 f ' 4 3 6 7 9 5 A7 B7 五、發明説明( 在每次系統初始化(即是,每次在關掉之後重新啓動系統) 所必須配置的。首先必須配置靴帶式結構100的理由,則 是其包含兩個十分重要的指標:一個標頭指標52,以及一 個區塊位址指標104。標頭指標52用來指向一個標頭位址 表(HAT),以判斷每個可淸除單元39中的每個標頭52之 位址。HAT所指的乃是最高層級(或根段)的HAT。區塊位 址指標104用來指向一個區塊位址表(BAT),藉以轉換從 PC I2所接收到的邏輯位址,成爲適合存取快閃記憶體18 的實體位址。BAT所指的乃是最高層級(或根段)的BAT。 靴帶式結構100儲存以下的資訊: 1. 一個唯一的標記,用來輔助控制器16的識別; 2. 一個指標,指向配置有結構1〇的可淸除單兀之標 頭; 3. 記憶體18中最高可清除單元39的數目; 4. —個儲存格5〇的總行數; 5. 一個儲存格50的總列數; 6. 一個可淸除單元中儲存格的總行數(例如在圖3中 16 個); 7. —個可淸除單元中儲存格的總列數(例如在圖3中 8個); 8. 一個控制區塊的邏輯位址(將在以下說明); 9. 一般用於系統10中的讀取/寫入區塊之長度; 10. 用於讀取/寫入區塊的任何一個ECC之長度; 11. 區塊位址指標104 ;以及 裝 訂 線 {請先聞讀背面之注意事項再f本頁) 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) 經濟部智M?財產局員工消費合作杜印製 |% 4367 9 5 Λ7 Β7 五、發明説明(>/ ) 12.標頭位址指標102。 區塊位址指標104定義控制儲存格的實體位址,其控 制儲存格中則配置有最高層級(也如同所知的根段)的區塊 位址表。同樣地,標頭位址指標102定義控制儲存格的實 體位址,其控制儲存格中則配置有最高層級(也如同所知的 根段)的標頭位址表。 圖1〇顯示讀取/寫入區塊60的一個邏輯位址11〇 ’其 位址包含n+1個欄位(表示層級),第η個欄位(表示最高的 層級,或樹的根段)112爲位址110最重要的位元,而第零 個欄位(表示最低的層級,或樹的葉部)114則是位址110最 不重要的位元。 對每一個標頭位址指標102和區塊位址指標104而言 ,靴帶式結構’ 100包含一個位址,指向單一第η個層級位 址表的指標表116之(實體)起始位址。控制器16使用第n 個欄位Π2的數値,來指示一個在第n個指標表116中的 單一登錄元。指向第η個指標表116的單一登錄元,乃是 用來指示位址表第n-Ι個層級的指標表118其中一個。在 所指向的第n-Ι個層級表118中特定的登錄元乃是由邏輯 位址的第n-Ι個欄位所決定。持續這種多重層級的指標處 理’直到第零個欄位114指向位址表U0中的一個登錄元 之時’其中的位址表120則包含將要存取的實際之實體位 址。‘所需的層級數目乃是由每個表116、US、120的登錄 元數目以及記憶體18的總邏輯容量所決定。 將會察知的是,相同形式的多重層級樹狀結構能夠用 本紙I尺度適用中國國碎(CNS ) ------:„„^r--辦本-------ΪΤ------0 (請先閲讀背面之注意事項再f-本頁) . 436795 五、發明説明(>^) - 來判斷標頭52的實體位址,以致判斷讀取/寫入區塊6〇的 實體位址。 圖9闡述一種用於標頭52和讀取/寫入區塊6〇的樹狀 結構配置。一個標頭位址樹130用來判斷標頭52的實體位 址,而一個區塊位址樹132則用來判斷讀取/寫入區塊6〇 的實體位址。 由區塊位址樹Π2所指向的第零個欄位表(相應於圖 10中的12〇)乃是區塊位址表(BAT)136。同樣地,標頭位址 樹130指向標頭位址表138。 區塊位址表136有兩種形式:主BATs(PBATs)以及次 BATs(SBATs)。一個PBAT可以搭配一個特定的SBAT。 PBATs和SBATs可以配置於實體位址空間內的任何一個位 置中,無論如何,它們總是儲存其控制儲存格56。 每個可淸除單元39具有至少一個的標頭控制儲存格, 用來維持可淸除單元39中PBATs和SBATs的位址(以控 制儲存格標頭的格式)。視將要使用的控制儲存格56之大 小而定,這種標頭控制儲存格也可以包含標頭52。每個 PBAT包含用來預測讀取/寫入區塊60數目的登錄元。 在PBAT中的一個登錄元可以包含將要存取的記憶體 之實體位址,或者可以包含一個指向所相應的SBAT 一個 登錄元的指標。同樣地,SBAT可以包含將要存取的記憶 體之實體位址’或者可以包含一個指向同一個SBAT另― 登錄元的指標。因此,可以建構一個間接的定址鏈,藉以 追蹤一個讀取/寫入區塊的動向。如此則提供了一種間接的 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) (請先閲讀背面之注意事項再ί本頁) 裝. 訂 經濟部智蒽財產局員工消費合作杜印製 ^ 436795 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明説明 定址機構,當一個讀取/寫入區塊重新寫入記憶體18不同 的位置時,允許更新BAT 136。 圖11顯示一個次區塊位址表(SBAT)139的鏈鎖結構。 每個SBAT登錄元H0具有三個SBAT欄位H2。能夠連 續並且獨立地寫入每個欄位142。在圖11和12所示的範 例中,於SBAT需要淸除之前,可以寫入(重新配置)一個 讀取/寫入區塊60三-(次。圖11中的最大鍵鎖長度(鍵 鎖的每個連接爲一.個SBAT欄位142)則爲四個的欄位H2 〇 第一次需要一個SBAT(區段一),新的讀取/寫入區塊 位址寫入欄位142a。下一次寫入(重新配置)其讀取/寫入區 塊60,其相應於區段二,並將新的位址寫入欄位142b之 中,且藉由欄位142c指向其新的位址;因此,此時必須寫 入兩個欄位(142b和142c)。下一次寫入(重新配置)其讀取/ 寫入區塊60時,將新的位址寫入欄位142d之中,並且藉 由欄位142e指向其新的位址;再次地,必須寫入兩個欄位 (142d和142e)。因此,對每一個登錄元140而言,只有認 爲最近所寫入的欄位才是有效的。重複如此的程序,直到 所有的攔位皆被寫入爲止,在圖Η中,其乃是發生在區段 三^*—被寫入之時。另一個範例則是當寫入其讀取/寫入區 塊十七次(在圖11中的區段17)時,則寫入欄位142f和 142g。 當所有的欄位皆被寫入時,在能夠使用之前,則需要 淸除整個SBAT 139。將會察知的是,對如此的架構而言, _______2fi_____ 本紙張尺度適用中國國家標準(CNS ) Μ規格(2H)X297公釐) ----^--„---裝------訂—^----線 (請先閲讀背面之注意事項再ylh本頁) 4367 9 5 a? B7 五、發明説明(斗) 在重新配置讀取/寫入區塊60時,每次至多必須更新兩個 欄位。一個SBAT 139的鏈鎖結構所具有的優點爲:其允 許簡易地重新建構其鏈鎖,如同控制器寫入和重新寫入資 料區塊,而不用處理會阻礙記憶體系統1〇操作速度的冗長 鏈鎖之連結。圖13顯示如圖3至5所示的一個具有相同失 效儲存格54之可淸除單元39,其增加一個失效儲存格於 第二列的第一行中。每一個控制儲存格56配置於每一列的 第一個可用儲存格51之中。每次可淸除單元39清除時, 根據一些預定演算法,必須藉由重新覆蓋可用儲存格51來 初始化,以使得可用儲存格51充當控制儲存格使用。 可用來充當每個可淸除單元39中控制儲存格56的可 用儲存格51之數目,乃是藉由儲存格圖表70中所儲存的 一個參數所決'定。此參數的數値不能夠小於用來配置靴帶 式結構100和標頭52所需的控制儲存格56之數目。在此 一實施例中,控制儲存格56配置於可淸除單元39每個不 完全失效列中第一個可用的儲存格51。 由於一些快閃記憶體不能免除記憶體裝置相同列中多 重個別寫入操作所引起的擾動效應,而產生可用的控制儲 存格56最大數目之限制。爲了消除如此的結果,而限制可 能配置於相同列的控制儲存格56之數目。 在此一實施例中,由於每個可淸除單元%具有十六行 儲存格,並且限制在每個列只能有一個儲存格,其參數的 最大値則等於可用儲存格的總數除以十六。 當寫入控制儲存格56時,則只有寫入一個列,然後增 本紙張尺度適用中國國家標準(CNS ) A4規格(2! 0 X 297公釐〉 (請先閱讀背面之注意事項再續爲本頁) : -'5 經濟部智慧財產局員工消費合作社印製 A7 ^ 436795 五、發明説明(><) (請先閱讀背面之注意事項再#減本頁) 加其列,再寫入控制儲存格中的下一個列’諸如此類。相 對地,當塡寫可用儲存格51時,則在增加其列之前’塡寫 可淸除單元39的一整個列。控制資料主要包含各次獨立塡 寫的資料之簡短欄位。因此,使用—列中用來儲存控制資 料所限制的控制儲存格之數目,來提供限制記憶體裝置相 同列中個別塡寫操作的數目,而不使用類似可用儲存格中 一個讀取/寫入區塊的資料結構,來提供限制記憶體裝置相 同列中分離塡寫操作的數目。 這些控制儲存格56乃是用來儲存區塊位址表 (BAT)136、指標表(PT)116 和 118、標頭位址表(HAT)138 、標頭52、以及靴帶式結構1〇〇。 每個控制儲存格56具有一個標頭’其帶有一個唯一的 編碼,用來辨識其是否儲存與BAT 136、PT 116、HAT 138、標頭52或者靴帶式結構100有關的資訊。控制儲存 格標頭也具有一個無效旗標(obsolete flag),用來指示儲存 於控制儲存格標頭中的資訊,是否爲有效或者無效的。 經濟部智慧財產局員工消費合作社印製 如同以上所說明的,用於一個可淸除單元39中所有控 制儲存格56的標頭,皆共同儲存於單元39的一個單一標 頭控制儲存格內,其中的標頭也可以涵蓋用於單元39的標 頭52。 每個PBAT 136具有多數個的登錄元,每個登錄元有 關於一個特定的讀取/寫入區塊之邏輯位址。每個PBAT 136則有關一組連續的邏輯位址。離析獨立的PBAT之存 在,以提供裝置中所有邏輯讀取/寫入區塊充足的容量。例 __2S--— - _________ 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) 436795 A7 B7 五、發明説明 如,每個PBAT登錄元具有三個位址欄位。於淸除之後, 當初次塡寫一個PBAT登錄元時,讀取/寫入區塊的實體位 址,則塡寫於第一個欄位中。在之後某次,當重新塡寫相 同的邏輯區塊時,新的實體位址輸入第二個位址欄位中, 而第一個欄位則不再是有效用的。於其後某次,當再一次 重新塡寫相同的邏輯區塊時,則第三個欄位用來供給一個 間接位址(一個SBAT登錄元140的位址)。控制器16自動 地認爲三個位址欄位的最高欄位(最近所塡寫的)爲有效的 ,並且不理會所有在較低欄位中的位址。 某些控制資訊必須保留於記憶體18中,藉以在PC 12 開始檢測系統10之時,提供系統1〇〇正確的初始化。爲了 能夠簡易地配置,此一資訊必須.儲存於一個預定的位置中 。無論如何,在記憶體系統1〇正常操作期間中,週期性地 更新此一資訊,因此,記憶體的耐用性乃是起因於其預定 位置的重複淸除。爲了避免此一問題,儲存其控制資訊如 同一個具有預定邏輯位址的控制區塊,而不像一個具有預 定實體位址的控制區塊。因此,每次更新時,控制區塊重 新塡寫於記憶體不同的區域中,但總是藉由相同的邏輯位 址來存取。控制區塊的邏輯位址接近於記憶體18邏輯位址 空間的頂端,且不能藉由PC 12來存取。可用於PC I2的 最高邏輯位址則緊接在控制區塊位址之下。使用區塊位址 表來判斷在實體記憶體中控制區塊的位置。 標頭位址表138使用以下的格式儲存於控制儲存格56 之中。HATs 138具有記億體18每個可淸除單元39的一個 _;__ _ ――29 — _ _ 本紙張尺度適用中國國家標準(CNS ) A4規格(210 X 297公釐) 0¾ (請先^讀背面之注意事項异本莧) "' 經濟部智慧財產局員工消費合作社印製 A7 436795 五、發明説明( 登錄元。只有一個HAT表配置於控制儲存格56中。每一 個HAT登錄元的長度爲16個位元組,所以連續的HAT登 錄元以16個位元組爲界限’配置於一個控制儲存格%之 中。因此,一個控制儲存格56能夠容納32個的登錄元。 控制器16以下列的方式將一個邏輯位址轉換成爲一個 實體位址(其中將要讀取/寫入的讀取/寫入區塊之邏輯區塊 位址格式,爲邏輯位址Π0的格式)。 控制器16使用區塊位址指標1〇4來配置最高(第η個) 層級區塊位址指標表116,並且藉由使用讀取/寫入邏輯位 •址第η個欄位112的數値,存取在表116中的正確登錄元 〇 此一登錄元指向下一個(第η-1個層級)的區塊位址指 標表118。之後則存取由讀取/寫入位址第n-Ι個欄位所指 向的登錄元。表118所有的η個層級皆重複如此的程序’ 直到指向在區塊位址表中的一個登錄元爲止。 如果此一區塊位址表登錄元爲一個直接指向記憶體18 一個位址的指標,則控制器16讀取可淸除單元39的可淸 除單元標頭52,而其中的讀取/寫入區塊則是配置在其可淸 除單元之中。控制器16之後讀取儲存格圖表70。再者, 藉由控制器16讀取其讀取/寫入區塊標頭部份62,用以確 認讀取/寫入區塊標頭部份62的邏輯區塊位址欄位,與讀 取/寫入區塊60的邏輯位址11〇相匹配。 在PC 12或緩衝記憶體28和記憶體18間的資料傳輸 操作初始化之前,控制器16從所要存取的可淸除單元39 _____________3J3________ .紙張尺度適用中國國家標卒(CNS ) A4規格(210Χ29?公釐)J2 £ L scale is applicable to Chinese national standard (CNS> A4 specification (2 丨 0 X 297mm) Α7 Γ '4367 9 5 V. Description of invention (丨 name), a unique identification method, so that the controller can identify Header 52, a write start flag 'a write end flag, a control cell start flag, a control cell end flag', and either the control cell header information or the control cell Index information. Initial time after erasable unit 39 is cleared 'When a read / write block is written into erasable unit 39, the write start flag is declared (set to a non-erasable state). When the read / write block is full of erasable unit 39, the write termination flag is declared. After erasable unit 39 is cleared, when the control cell (except the control memory containing the header) is written for the first time. Cell), the control cell start flag is declared. When all control cells are written, 'the control cell end flag is declared. If in a single control cell for the entire header information Out of space, you just need to control # 存 格 Header indicator 'in the example' The control cell header information is stored in another control cell indicated by the control cell indicator. Figure 7 shows for access (ie, from its Read and write to) the format of a physical address 80 of the flash memory 18. The physical address 80 has a first field 82, which indicates a removable unit 39. In this embodiment, one The size of the erasable unit 39 is equal to one erasable block 36. The physical address 80 also has a second field 84, which contains an address of the erasable unit 39. In the second field 84 The address in is about the total physical address space of the erasable unit 39. The second field 84 is used to define the position of any data structure in the erasable unit 39. This paper scale applies Chinese national standards (CNS) Α4 specification (210 X 297 mm) ----- ^ ----- Xiang ------ IT ------ ^ (Please read the precautions on the back before ^^: (This page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 4 3 6 7 9 5 at ____B7____ Description of the Invention (lf) Figure 8 shows a more detailed second field 84. The second field 84 contains a cell row 92, a row 94, a cell f96, and an fT 98. The cell Column 92 is a special column of the cells in the erasable unit 39 to be accessed. Column 94 is the physical address of the specific column of the location (in cell column 92) to be accessed. The address of this column is Is the first column in the cell. Similarly, cell row 96 is a specific row of cell in division unit 39 to be accessed, and row 98 is the location to be accessed (cell (Line 96) The physical address of a particular line. The address of this row is the first row in the relevant cell. The order of the data structures stored in the physical address space is completely independent of the logical block address (LBA) of each data structure. The data structure corresponding to any specific logical block can be allocated in the space of any physical address. The data structure entity address (PBA) of each LBA is specified in a set of tables, such as the known block address table (BAT), which can be accessed through a tree structure of an index table (PT) BAT, as illustrated in Figures 9 and 10. A shoelace structure 100 is stored in a control cell 56 in the first erasable unit 39 of the memory 18. For the purpose of protection, a backup of the boot strap structure 100 is also stored in the second and last erasable unit 39 of the memory 18. The control cell 56 for storing the shoelace structure 100 is selected by the controller 16 and its position is the first one immediately after the control cell including the header 52 Controls the location of the cell. The shoelace structure 100 is the first data structure, which is the controller 16 _______ _22_____ This paper size is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) I -------- ^ --- --- ΐτ ------ ^ (Please read the notes of Ι & before r'-private page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs f '4 3 6 7 9 5 A7 B7 5 Description of the invention (necessary for each system initialization (that is, each time the system is restarted after being shut down). The reason why the bootstrap structure 100 must be configured first is that it contains two very important indicators: A header index 52 and a block address index 104. The header index 52 is used to point to a header address table (HAT) to determine the position of each header 52 in each erasable unit 39 HAT refers to the HAT at the highest level (or root segment). The block address indicator 104 is used to point to a block address table (BAT) to convert the logical address received from PC I2. Become a physical address suitable for accessing flash memory 18. BAT refers to the highest-level (or root) BAT. Bootstrap structure 100 The following information is stored: 1. A unique mark to assist the identification of the controller 16; 2. An indicator pointing to the removable header with the structure 10 configured; 3. The highest available memory in the memory 18 Clear the number of cells 39; 4. The total number of rows in cell 50; 5. The total number of rows in one cell 50; 6. The total number of rows in a removable cell (e.g. 16 in Figure 3) ); 7. The total number of columns of cells in a divisible cell (for example, 8 in Figure 3); 8. The logical address of a control block (to be described below); 9. Generally used in the system The length of the read / write block in 10; 10. the length of any ECC used to read / write the block; 11. the block address index 104; and the gutter {please read the back (Notes on this page again) This paper size applies Chinese National Standard (CNS) A4 (210X297 mm) Printed by the Ministry of Economic Affairs, M? Property Bureau, Consumer Consumption Cooperation |% 4367 9 5 Λ7 Β7 V. Description of the invention (> /) 12. Header address indicator 102. The block address index 104 defines the physical address of the control cell, and the control cell is configured with a block address table of the highest level (also known as the root segment). Similarly, the header address indicator 102 defines the physical address of the control cell, and the control cell is configured with the header address table of the highest level (also known as the root segment). FIG. 10 shows a logical address 11 ′ of the read / write block 60, whose address contains n + 1 fields (representing the hierarchy), and the nth field (representing the highest hierarchy, or the root of the tree) Segment) 112 is the most significant bit of address 110, and the zeroth field (indicating the lowest level, or the leaf of the tree) 114 is the least significant bit of address 110. For each header address index 102 and block address index 104, the shoelaced structure '100 contains an address that points to the (physical) starting position of the index table 116 of a single n-th level address table site. The controller 16 uses the number of the n-th field Π2 to indicate a single entry element in the n-th index table 116. The single sign-on element pointing to the n-th index table 116 is one of the index tables 118 for indicating the n-1th level of the address table. The specific entry element in the n-1th level table 118 pointed to is determined by the n-1th field of the logical address. This multi-level index processing is continued until the zeroth field 114 points to a registration element in the address table U0, where the address table 120 contains the actual physical address to be accessed. ‘The number of levels required is determined by the number of entries in each table 116, US, 120, and the total logical capacity of memory 18. It will be known that the same form of multi-level tree structure can be applied to Chinese National Fragmentation (CNS) using the I scale of this paper ------: „^ r-- 办 本 ------ ΪΤ- ----- 0 (Please read the precautions on the back before f-this page). 436795 V. Description of the invention (> ^)-to determine the physical address of the header 52, so as to determine the read / write area The physical address of block 60. FIG. 9 illustrates a tree structure configuration for the header 52 and the read / write block 60. A header address tree 130 is used to determine the physical address of the header 52, and a block address tree 132 is used to determine the physical address of the read / write block 60. The zero field table (corresponding to 12 in FIG. 10) pointed by the block address tree Π2 is the block address table (BAT) 136. Similarly, the header address tree 130 points to the header address table 138. The block address table 136 has two forms: primary BATs (PBATs) and secondary BATs (SBATs). A PBAT can be paired with a specific SBAT. PBATs and SBATs can be placed in any location in the physical address space, and they always store their control cell 56 no matter what. Each erasable unit 39 has at least one header control cell for maintaining the addresses of PBATs and SBATs in the erasable unit 39 (to control the format of the cell header). Depending on the size of the control cell 56 to be used, this header control cell may also include a header 52. Each PBAT contains a registry element used to predict the number of read / write blocks 60. A registry element in the PBAT may contain the physical address of the memory to be accessed, or it may contain an indicator pointing to a registry element of the corresponding SBAT. Similarly, the SBAT can contain the physical address of the memory to be accessed 'or it can contain an indicator that points to the same SBAT as another registrar. Therefore, an indirect addressing chain can be constructed to track the movement of a read / write block. In this way, it provides an indirect standard for this paper, which is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) (please read the precautions on the back first and then this page). Printed ^ 435 795 Printed by the Consumer Property Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the invention The addressing mechanism allows updating of BAT 136 when a read / write block is rewritten to a different location in memory 18. FIG. 11 shows a chain structure of a sub-block address table (SBAT) 139. Each SBAT login element H0 has three SBAT fields H2. Each field 142 can be written continuously and independently. In the examples shown in Figures 11 and 12, before the SBAT needs to be erased, a read / write block 60 can be written (reconfigured) three times (the maximum key lock length (key lock in Figure 11) Each connection is one. One SBAT field 142) Four fields H2 〇 One SBAT (Sector 1) is required for the first time, and the new read / write block address is written to field 142a The next write (reconfiguration) of its read / write block 60, which corresponds to sector two, and writes a new address into field 142b, and points to its new one through field 142c Address; therefore, two fields (142b and 142c) must be written at this time. The next write (reconfiguration) of its read / write block 60, a new address is written to field 142d , And point to its new address by field 142e; again, two fields (142d and 142e) must be written. Therefore, for each entry element 140, only the most recently written The field is only valid. Repeat this process until all the blocks are written. In the figure, it happens when the section three ^ *-is written. Another example is When its read / write block is written seventeen times (in section 17 in Figure 11), fields 142f and 142g are written. When all fields are written, before they can be used , You need to eliminate the entire SBAT 139. It will be known that, for such an architecture, _______2fi_____ This paper size applies to the Chinese National Standard (CNS) M specification (2H) X297 mm) ---- ^-„ --- install ------ order-^ ---- line (please read the precautions on the back before ylh this page) 4367 9 5 a? B7 V. Description of the invention When writing to block 60, up to two fields must be updated each time. The advantage of a SBAT 139 chain structure is that it allows easy reconstruction of its chain, as the controller writes and rewrites data Block without processing the lengthy chain link that would hinder the operating speed of the memory system 10. Figure 13 shows a removable unit 39 with the same invalid cell 54 as shown in Figures 3 to 5, which adds one Invalid cells are in the first row of the second column. Each control cell 56 is placed in the first available cell in each column Each time the erasable unit 39 is cleared, according to some predetermined algorithms, it must be initialized by rewriting the available cell 51 so that the available cell 51 can be used as the control cell. It can be used as each The number of available cell 51 in control cell 56 in erasable unit 39 is determined by a parameter stored in cell chart 70. The number of this parameter cannot be less than that used to configure the shoelace The number of control cells 56 required by the formula structure 100 and the header 52. In this embodiment, the control cell 56 is arranged in the first available cell 51 in each incomplete failure column of the erasable unit 39 . Since some flash memories cannot eliminate the perturbation effect caused by multiple individual write operations in the same column of the memory device, there is a limit on the maximum number of usable control bank 56. To eliminate such a result, the number of control cells 56 that may be arranged in the same column is limited. In this embodiment, since each erasable unit% has sixteen rows of cells and is limited to only one cell in each column, its maximum parameter is equal to the total number of available cells divided by ten. six. When writing to control cell 56, only one column is written, and then the paper size is increased to apply the Chinese National Standard (CNS) A4 specification (2! 0 X 297 mm> (Please read the precautions on the back before continuing) (This page): -'5 Printed A7 by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs ^ 436795 V. Invention Description (> <) (Please read the notes on the back before #minus this page) Add it and write Enter the next column in the control cell 'and so on. In contrast, when writing available cell 51, before writing its entire row,' write an entire column of unit 39. Control data mainly includes each independent Short field for transcribing data. Therefore, the number of control cells limited by the row used to store the control data is used to provide a limit on the number of individual transcribing operations in the same row of the memory device, instead of using similarly available A data structure of a read / write block in a cell to limit the number of separate write operations in the same row of the memory device. These control cells 56 are used to store the block address table (BAT) 136 Index table PT) 116 and 118, Header Address Table (HAT) 138, Header 52, and Bootstrap Structure 100. Each control cell 56 has a header 'with a unique code for Identifies whether it stores information related to BAT 136, PT 116, HAT 138, header 52, or bootstrap structure 100. The control cell header also has an obsolete flag to indicate storage in the control storage Whether the information in the header of the grid is valid or invalid. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs, as described above, the header for all control cells 56 in one erasable unit 39 is common. Stored in a single header control cell in unit 39, the header of which can also cover header 52 for unit 39. Each PBAT 136 has a plurality of login elements, and each login element has a specific The logical address of the read / write block. Each PBAT 136 is related to a set of consecutive logical addresses. The existence of independent PBAT is isolated to provide sufficient capacity for all logical read / write blocks in the device. Example __2S -----__ _______ This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) 436795 A7 B7 V. Description of the invention For example, each PBAT login element has three address fields. After the deletion, when one is written for the first time When the PBAT is registered, the physical address of the read / write block is written in the first field. At a later time, when the same logical block is rewritten, a new physical address is entered. In the second address field, the first field is no longer valid. At a later time, when the same logical block is rewritten again, the third field is used Provide an indirect address (the address of a SBAT login element 140). The controller 16 automatically considers the highest field (most recently transcribed) of the three address fields to be valid and ignores all addresses in the lower field. Certain control information must be retained in the memory 18 so as to provide the system 100 with proper initialization when the PC 12 begins to detect the system 10. For easy configuration, this information must be stored in a predetermined location. In any case, this information is periodically updated during the normal operation of the memory system 10. Therefore, the durability of the memory is caused by the repeated erasure of its predetermined position. To avoid this problem, its control information is stored as a control block with a predetermined logical address, rather than a control block with a predetermined physical address. Therefore, with each update, the control block is rewritten in a different area of the memory, but is always accessed through the same logical address. The logical address of the control block is close to the top of the logical address space of the memory 18 and cannot be accessed by the PC 12. The highest logical address available for PC I2 is immediately below the control block address. Use the block address table to determine the location of the control block in physical memory. The header address table 138 is stored in the control cell 56 using the following format. HATs 138 has one _; __ _ _ 29 — _ _ for each deletable unit 39 of the record body 18. This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) 0¾ (please first ^ Notes on the back of the book are different.) &Quot; 'Printed by A7 436795 of the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Description of the invention (registration element. There is only one HAT table configured in the control cell 56. For each HAT registration element The length is 16 bytes, so consecutive HAT login elements are arranged in a control cell% with a limit of 16 bytes. Therefore, one control cell 56 can accommodate 32 login elements. Controller 16 Convert a logical address into a physical address in the following manner (where the logical block address format of the read / write block to be read / written is the format of the logical address Π0). Control The device 16 uses the block address index 104 to configure the highest (nth) level block address index table 116, and by using the read / write logic bit address of the nth field 112, , Access the correct login element in table 116 The entry points to the next (n-1th level) block address index table 118. After that, it accesses the entry point pointed to by the read / write address n-1 field. Table 118 all This process is repeated for all η levels of the block until it points to a registry element in the block address table. If the registry element of this block address table is an index directly pointing to an address in the memory 18, then control The processor 16 reads the erasable unit header 52 of the erasable unit 39, and the read / write block therein is configured in its erasable unit. The controller 16 then reads the cell chart 70 In addition, the controller 16 reads its read / write block header portion 62 to confirm the logical block address field of the read / write block header portion 62, and The logical address 11 of the read / write block 60 matches. Before the data transfer operation between the PC 12 or the buffer memory 28 and the memory 18 is initialized, the controller 16 removes the erasable unit 39 to be accessed. _____________3J3________. The paper size is applicable to China National Standards (CNS) A4 specification (210 × 29? Mm)

-----:-----^------’訂------^ (請先閱讀背面之注意事項再f.本頁) C 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明説明(5及) 中讀取儲存格圖表70,來判斷所有儲存格50之可用/不可 用狀態,而其中的儲存格50乃是由讀取/寫入區塊60所展 延的。在資料傳輸期間’控制器16監視資料的傳輸,用以 檢測出可用儲存格51和控制儲存格56,或失效儲存格54 之間的邊界。當遇到一個邊界之時,則暫停其傳輸,以允 許配置新的(繼續)位址。 在可能的情況下,可能由於電源的失效而失去寫入指 標位址,則藉由掃.描記憶體18的資料結構直到發現部份已 寫入的一個可淸除單元39,而能夠再獲得寫入指標的正確 位址,其中部份已寫入的可淸除單元39,即是具有已聲明 的寫入起始旗標,並且具有未聲明的寫入終止旗標。此則 是藉由讀取繼續的可淸除單元標頭52之旗標狀態來執行的 〇 當找到正確的可淸除單元39時,則掃描單元39,直 到發現已淸除記億體的第一個位置(在可用儲存格中)爲止 。此即是寫入指標的位置。 根據所實現的特定之淸除表格化演算法’藉由從寫入 指標的位置開始掃描可淸除單元36,直到發現應該淸除的 第一個可淸除單元爲止,來找到淸除指標。藉由讀取緊接 在可淸除單元標頭52之後的資料結構標頭’來確認每一個 可淸除單元39。 ,圖14顯示一種修改過的可淸除單元實體位址84b之 格式,其使用於具有外加記憶裝置的快閃記憶體架構。圖 15顯示一個快閃記憶體的可淸除單兀39,其快閃記憶體則 ^___^__—u_ 本紙張尺度適用中國國家標準(CNS ) A4規格(21 〇 X 297公釐) (請先閲讀背面之注意事項再ί.本頁) .裝 線 經濟部智¾財產局員工消費合作社印製 經濟部智慧財產局員工消費合作社印製 Γ' 4367 9 5 Α7 Β7 五、發明説明(>f) 具有主要的儲存裝置150和額外之單行位元組形式的儲存 裝置U2。可淸除單元實體位址84b(圖H)具有一個外加的 位址位元(xcell位元)154,用來指示其位址是表示主要的( 位元154設定爲零)記憶體空間,或者是表示外加的(位元 154設定爲一)記憶體空間。 當xcell位元設定爲零之時,則以如同上述的位址欄 位84之相同方法,來使用位元84b。當xcell位元設定爲1 之時,無論如何,.則存取具有相應於行92和94的列位址 之xcell。當使用外加的儲存裝置152時,則儲存格圖表7〇 涵蓋一個外加行,用來指示每一列的xcell是否爲失效的或 有效的。如果使用xcell的多數行,則行96和98可以用來 存取外加儲存裝置的正確列。 在使用中,藉由提供資料起始位置之一個實體位址, 以及藉由提供將要塡寫的資料段之長度,控制器16將所有 的資料結構塡寫於記憶體18。在設有一個或多個失效或控 制儲存格的讀取/寫入區塊之一個範例中,或者從一個實體 記憶列重疊到另外一列的讀取/寫入區塊之一個範例中,依 照一些個別的資料段來塡寫其資料結構。在控制儲存格56 內控制資料結構的一個範例中,依照一個或多個的資料段 來塡寫其資料,其每一個皆佔攄控制儲存格56的一列或一 列的部份。一個控制資料結構的實體位址,不是藉由控制 器16來計算,便是藉由讀取另一個控制資料結構來得到。 以上的實施例乃是有關於控制器16 ’其控制器16貝IJ 具有一個設置於快閃記憶體卡中的處理器以及所需的韌體 __ _12_______ 本紙張尺度適用中國園家橾準(CNS ) A4规格(210X297公釐) i I 訂 線 (請先閲讀背面之注意事項再^Γ本頁) A7 B7 :436795 五、發明説明(今0 ) 。無論如何,控制器16可以設置於一個主系統中,例如 PC I2,而可替代地,依照使用線性記億卡(linear memory cards)的主系統中之一個軟體層級(使用控制演算法),可以 實現控制器16。圖16闡述主系統中的層級160之體系, 以及主系統所連接的一個線性快閃記憶體162。在圖16的 系統中,能夠藉由一個主要軟體驅動器來產生儲存格圖表 70 ° 其將會察知的是,本發明所具有P優點爲:藉由使用 儲存格來儲存不是控制資訊便是讀取/寫入區塊,一個可淸 除區塊則可以儲存控制資訊以及讀取/寫入區塊兩者。另外 ,藉由使用儲存格中的指標來指向其它的儲存格,則容易 儲存需要經常更新的控制資訊,並且容易持續地追蹤其需 要經常更新的控制資訊。由於能夠簡易地確認和重新配置 任何一個有效的控制儲存格,因此可淸除區塊之淸除乃是 簡易的。 在本發明的觀點之中,可以建構種種的變體成爲以上 所說明的實施例。例如,一個可淸除區塊36可以大於或者 小於64個千位元組;同樣地,一個儲存格可以大於或者小 於512個位元組,就範例而言,256個位元組則可能是較 佳的。選用特定數値的可淸除單元之大小,將視製造者的 選擇而定,但一般則是與記憶體裝置18的架構、技術和儲 存格形式有關。選用特定數値的儲存格之大小,將視所需 的記憶容量而定:如果使用較小的儲存格尺寸,則可用的 容量爲較大的,但是隨著使用更多的儲存格,卻使得每個 ___ -- __—_ U____ 本紙浪尺度適用中國國家標準(CNS ) A4規格(21〇X297公釐) 丨 裝 、^1線 (請先閲讀背面之注意事項再^_本頁) 經濟部智慧財產局員4消費合作社印製 經濟部智慧財產局員Χ-消費合作社印製 436 7 9 5 1_ml i_ 五、發明說明() 可淸除單元需要較大的儲存格圖表。錯誤校正碼可以實現 於以上的資料結構中,用以最小化遺失重要資料的可能性 。在其它的實施例中將會察知,資料結構的大小能夠不同 於所說明的,例如,如果使用一個較大或較小的記憶體。 在其它的實施例中,快閃記憶體18可以由包含多數量的可 淸除區塊之記憶體裝置所組成,每一個可淸除區塊可以是 記憶體裝置的一列(區段),或典型爲十六的少數連續列。 在此範例中,控制器16所管理的可淸除單元39可以由多 數個的可淸除區塊所組成。在其它的實施例中,字組的大 小可以較大於或較小於八個位元,例如,十六個位元或六 十四個位元。在其它的實施例中,可以使用大於四的鏈鎖 長度。在如此的實施例中,由於隨著一個讀取/寫入區塊60 多數的重新配置,而可以展開SBAT登錄元的長鏈鎖。在 如此的範例中,可以藉由更新指向SBAT鏈鎖起始端的 PBAT登錄元來打斷其鏈鎖,以使得其指向新的SBAT。在 其它的實施例中,能以不同的方式來配置控制儲存格56 ’ 例如’控制儲存格可以配置於可淸除單元39每一列最後的 可用儲存格51中、配置於每個第二列的第一個儲存格中、 或者每列可以配置多於一個的控制儲存格56。在其它的實 施例中,可以藉由控制器丨6、而不藉由一個個別的裝置( 測試單元)來建構儲存格圖表70。 〔元件符號說明〕 10 記憶體系統 12 個人電腦 ___ 34 本紙張尺度適用中國國家標準(CNS)A4規格(210 x 297公釐) -------------裝--------訂·一-----1 * 線 (請先閱讀背面之注意事項再填寫本育) 89.12. 0 年月曰 補无 436795 A7 B7 五、發明說明( 經濟部智慧財產局員工消費合作社印製 14 PC介面 16 控制器 18 快閃記憶體 20 PC介面硬體 22 記憶體介面硬體 24 資料匯流排 26 控制匯流排 28 緩衝記憶體 29 微處理器 30 非揮發性記憶體 32 非揮發性記憶體 33 非揮發性記憶體 34 資料匯流排 36 區塊 38 記憶位置 39 可淸除單元 40 1占 42 磁叢 44 磁叢 46 列 48 行 50 儲存格 51 儲存格 52 標頭 35 (請先閱讀背面之注意事項再填寫本頁) 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐)----- : ----- ^ ------ 'Order ------ ^ (Please read the notes on the back before f. This page) C Employees' Cooperatives, Intellectual Property Bureau, Ministry of Economic Affairs Print A7 B7 5. In the description of the invention (5 and), read the cell chart 70 to determine the available / unavailable status of all cells 50, and the cell 50 is read / written to block 60. Stretched out. During the data transmission ', the controller 16 monitors the transmission of data to detect the boundary between the available cell 51 and the control cell 56, or the failed cell 54. When a boundary is encountered, its transmission is suspended to allow the allocation of a new (continued) address. Where possible, the write index address may be lost due to power failure. By scanning the data structure of the memory 18 until it is found that a partially erasable unit 39 has been written, it can be obtained again. The correct address of the write index, some of which can be written into the erasable unit 39, have a write start flag that has been declared, and a write end flag that has not been declared. This is performed by reading the flag state of the continued erasable unit header 52. When the correct erasable unit 39 is found, the unit 39 is scanned until the first erasable unit is found. Up to one location (in the available cells). This is where the indicator is written. According to the specific erasing table algorithm implemented ', the erasing index is found by scanning the erasing unit 36 from the position where the index is written until the first erasing unit that should be erasing is found. Each erasable unit 39 is confirmed by reading the data structure header 'immediately after the erasable unit header 52. FIG. 14 shows a modified format of the erasable unit physical address 84b, which is used in a flash memory architecture with an external memory device. Figure 15 shows the removable unit 39 of a flash memory, and the flash memory is ^ ___ ^ ___ u_ This paper size is applicable to the Chinese National Standard (CNS) A4 specification (21 × 297 mm) (Please Read the precautions on the back first, and then go to this page). Printed by the Ministry of Economic Affairs, Intellectual Property Bureau, the Consumer Cooperative of the Property Bureau, and printed by the Ministry of Economic Affairs, the Intellectual Property Bureau, the Employee Consumer Cooperative, and printed by the employee ’s cooperative. Γ '4367 9 5 Α7 Β7 5. Description of the invention f) It has a main storage device 150 and an additional single-row byte storage device U2. The eliminable unit physical address 84b (Figure H) has an additional address bit (xcell bit) 154, which is used to indicate that its address represents the primary (bit 154 is set to zero) memory space, or Yes means extra (bit 154 is set to one) memory space. When the xcell bit is set to zero, bit 84b is used in the same way as address field 84 described above. When the xcell bit is set to 1, the xcell having a column address corresponding to rows 92 and 94 is accessed anyway. When the additional storage device 152 is used, the cell chart 70 covers an additional row to indicate whether the xcell of each column is invalid or valid. If most rows of xcell are used, rows 96 and 98 can be used to access the correct row of the external storage device. In use, the controller 16 writes all data structures into the memory 18 by providing a physical address of the starting position of the data and by providing the length of the data segment to be written. In an example of a read / write block with one or more invalid or controlled cells, or an example of a read / write block that overlaps from one physical memory row to another, follow some Individual data segments to describe their data structure. In an example of the control data structure in the control cell 56, its data is written according to one or more data segments, each of which occupies part of a row or a row of the control cell 56. The physical address of a control data structure is either calculated by the controller 16 or obtained by reading another control data structure. The above embodiment is related to the controller 16 ', the controller 16 of which has a processor set in a flash memory card and the required firmware __ _12_______ This paper standard is applicable to China Garden Home Standard ( CNS) A4 specification (210X297mm) i I line (please read the precautions on the back before ^ Γ this page) A7 B7: 436795 V. Description of the invention (now 0). In any case, the controller 16 may be provided in a host system, such as PC I2, and alternatively, in accordance with a software level (using a control algorithm) in the host system using linear memory cards, Implement the controller 16. FIG. 16 illustrates the hierarchy 160 system in the host system and a linear flash memory 162 connected to the host system. In the system of FIG. 16, a cell chart can be generated by a main software driver. 70 ° It will be seen that the present invention has a P advantage: by using cells to store either control information or reading / Write block, a erasable block can store both control information and read / write block. In addition, by using the pointer in the cell to point to other cells, it is easy to store control information that needs to be updated frequently, and it is easy to continuously track the control information that needs to be updated frequently. Since any valid control cell can be easily identified and reconfigured, erasing blocks is easy. From the viewpoint of the present invention, various modifications can be constructed into the embodiments described above. For example, an erasable block 36 may be larger or smaller than 64 kilobytes; similarly, a cell may be larger or smaller than 512 bytes. As an example, 256 bytes may be more than Good. The size of the selectable unit of the particular number will depend on the manufacturer's choice, but is generally related to the architecture, technology, and form of the memory cells of the memory device 18. The size of the selected number of cells will depend on the required memory capacity: if you use a smaller cell size, the available capacity will be larger, but as more cells are used, it will make Each ___-____ U____ This paper wave scale is applicable to the Chinese National Standard (CNS) A4 specification (21 × 297 mm) 丨 installed, ^ 1 line (please read the precautions on the back before ^ _ this page) Economy Member of the Ministry of Intellectual Property Bureau 4 Printed by the Consumer Cooperative Cooperative Member of the Ministry of Economic Affairs Intellectual Property Bureau X-Consumer Cooperative Printed 436 7 9 5 1_ml i_ V. Description of the Invention () Cancellable units require larger cell charts. Error correction codes can be implemented in the above data structure to minimize the possibility of losing important data. It will be appreciated in other embodiments that the size of the data structure can be different from what is illustrated, for example, if a larger or smaller memory is used. In other embodiments, the flash memory 18 may be composed of a memory device including a large number of erasable blocks, and each erasable block may be a row (section) of the memory device, or It is typically a small number of consecutive columns of sixteen. In this example, the erasable unit 39 managed by the controller 16 may be composed of a plurality of erasable blocks. In other embodiments, the size of the block may be larger or smaller than eight bits, for example, sixteen bits or sixty-four bits. In other embodiments, a chain length greater than four may be used. In such an embodiment, the long chain lock of the SBAT login element can be expanded due to the reconfiguration of a majority of a read / write block 60. In such an example, the chain lock can be interrupted by updating the PBAT registry element pointing to the beginning of the SBAT chain lock so that it points to the new SBAT. In other embodiments, the control cell 56 can be configured in different ways. For example, the control cell can be configured in the last available cell 51 in each column of the erasable unit 39 and in each of the second columns. More than one control cell 56 can be arranged in the first cell or each row. In other embodiments, the cell chart 70 can be constructed by the controller 6 without using a separate device (test unit). [Explanation of Component Symbols] 10 Memory System 12 Personal Computer ___ 34 This paper size applies to China National Standard (CNS) A4 (210 x 297 mm) ------------- installation- ------ Order · 一 ----- 1 * line (please read the notes on the back before filling in this education) 89.12. 0 month and month supplement 436795 A7 B7 V. Description of Invention (Intellectual Property of the Ministry of Economic Affairs Printed by Bureau Consumer Cooperatives 14 PC interface 16 Controller 18 Flash memory 20 PC interface hardware 22 Memory interface hardware 24 Data bus 26 Control bus 28 Buffer memory 29 Microprocessor 30 Non-volatile memory 32 non-volatile memory 33 non-volatile memory 34 data bus 36 block 38 memory location 39 erasable unit 40 1 occupant 42 magnetic plexus 44 magnetic plexus 46 rows 48 rows 50 cell 51 cell 52 header 35 (Please read the precautions on the back before filling this page) This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm)

436195 A7 B7 經濟部智慧財產局員工消費合作社印製 (請先閱讀背面之注意事項再填寫本頁) 五、發明說明( 54 56 60a 60b 60c 60d 60q 62 64 70 71 72 73 74 75 76 80 82 84 92 94 96 98 100 儲存格 儲存格 讀取/寫入區塊 讚取/寫入區塊 讀取/寫入區塊 讀取/寫入區塊 讀取/寫入區塊 標頭部份 資料部份 圖表 位址 數目 數目 一組位元 行 列 . 實體位址 攔位 欄位 儲存格列 列 儲存格行 行 靴帶式結構 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 年月曰4367 9 5 補无 A7 ------ ------裝 - ----- ---訂 --------線 (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 五、發明說明( 102 104 110 112 114 116 118 120 130 132 136 138 139 140 142 142a 142b 142c 142d 142e 142f 142g 150 152 ) 標頭 區塊位址指標 位址 欄位’ 欄位 指標表 指標表 位址表 標頭位址樹 區塊位址樹 區塊位址表 標頭位址表 區塊位址表 登錄元 欄位 攔位 欄位 欄位 欄位 欄位 欄位 欄位 儲存位置 儲存裝置 本紙張尺度適用中國國家標準(CNS>A4規格(210 X 297公釐)436195 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs (please read the precautions on the back before filling this page) V. Invention Description (54 56 60a 60b 60c 60d 60q 62 64 70 71 72 73 74 75 76 80 82 84 92 94 96 98 100 cell read / write block like / write block read / write block read / write block read / write block header part data department The number of copies of the chart address is a group of bits and rows. The physical address block field, the storage cell, the storage cell, the row, and the boot strap structure. The paper dimensions are applicable to China National Standard (CNS) A4 (210 X 297 mm). Year, month and year 4367 9 5 Supplement without A7 ------ ------ installation------ --- order -------- line (please read the precautions on the back before (Fill in this page) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs V. Invention Description (102 104 110 112 114 116 118 120 130 132 136 138 139 140 142 142a 142b 142c 142d 142e 142f 142g 150 152) Header Block Address Index Address Field 'Field Index Table Index Table Address Table Header Address Tree Block Address Tree Block Address Table Icon Block address table entry table address field blocking element bit Fields field Fields field Fields field storage location the storage device of the present paper is suitable China National Standard Scale (CNS > A4 size (210 X 297 mm)

436795 A7 B7 五、發明說明( 154 位址位元 160 層級 162 快閃記憶體 — — — — — — — —— — — It ' — — — III— ·11{1*1111 (請先閱讀背面之沈意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 38 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐)436795 A7 B7 V. Description of the invention (154 address bit 160 level 162 flash memory — — — — — — — — — — It '— — — III — · 11 {1 * 1111 (Please read the back Please fill in this page again if you are concerned about the matter) Printed by the Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 38 This paper size applies to China National Standard (CNS) A4 (210 X 297 mm)

Claims (1)

正充修補 GJ ο I 5 9 7 6 3 Λα U0 8 8 8 ABCD 經濟部智袅財沌局:^工冶^合作社印製 六、申請專利範圍 1.一種連接到主系統的記憶體系統,該系統包含:一 個具有記憶位置的非揮發性記憶體,以及一個控制器’用 來將資料結構寫入記憶體及用來從記憶體中讀取資料結構 ,且建構該系統以使得該位置能夠個別地寫入,但卻只能 以整個區塊的位置來淸除; 其改良爲:控制器構成至少一個可淸除單元,其中’ 每一個可淸除單元包含至少一個可淸除區塊,以及控制器 將每個可淸除單元細分爲位置群(其中每群稱爲一個儲存格 ),以及控制器基於每個儲存格,而將資料結構寫入每個儲 存格且從每個儲存格中讀取資料結構。 根據申請專利範圍第1項之記憶體系統,其中測試 每個儲存格中的記憶位置,並且如果一個失效出現於儲存 格中,則認爲整個儲存格皆不能用來儲存資料結構,否則 ,則認爲該儲存格可用於儲存資料結構》 3. 根據申請專利範圍第1或第2項之記憶體系統,其 中該非揮發性記憶體包含多數個記憶體裝置。 4. 根據申請專利範圍第1或第2項之一種記憶體系統 ,其中的非揮發性記憶體僅爲單一的記憶體裝置。 5-根據申請專利範圍第3項之記憶體系統,其中的控 制器控制著合倂於其每一個記憶體裝置中的附屬控制器。 6.根據申請專利範圍第3項之記憶體系統,其中的控 制器乃是以一種單一控制器的形式控制著其每一個記憶體 裝置。 7·根據申請專利範圍第1項之記憶體系統,其中指定 尽紙浪尺度適用中國國家標準{ CMS ) Α4規格(210X Μ7公釐) ---------1¾衣--------ΪΤ------- ^ [ (請先閱讀背面之注意事項再填寫本頁)Positive charge repair GJ ο I 5 9 7 6 3 Λα U0 8 8 8 ABCD Ministry of Economic Affairs, Ministry of Economic Affairs, Finance and Economic Chaos Bureau: ^ Industry and Metallurgy ^ Printed by the cooperative 6. Application scope 1. A memory system connected to the main system, the system Includes: a non-volatile memory with a memory location, and a controller 'for writing and reading data structures to and from the memory, and constructing the system so that the location can be individually Write, but can only be erased by the position of the entire block; its improvement is: the controller constitutes at least one erasable unit, where 'each erasable unit contains at least one erasable block, and the control The device subdivides each eliminable unit into groups of locations (where each group is called a cell), and the controller writes a data structure to and reads from each cell based on each cell Take the data structure. The memory system according to item 1 of the scope of the patent application, in which the memory location in each cell is tested, and if a failure occurs in the cell, the entire cell cannot be considered to store the data structure, otherwise, It is considered that the cell can be used for storing data structures. 3. According to the memory system of the first or second patent application scope, the non-volatile memory includes a plurality of memory devices. 4. A memory system according to item 1 or 2 of the scope of patent application, wherein the non-volatile memory is only a single memory device. 5- The memory system according to item 3 of the scope of patent application, wherein the controller controls the attached controller integrated in each of its memory devices. 6. The memory system according to item 3 of the scope of patent application, wherein the controller controls each memory device in the form of a single controller. 7. Memory system according to item 1 of the scope of patent application, in which the specified paper scale is applicable to the Chinese national standard {CMS Α4 specification (210X Μ7 mm) --------- 1¾ clothing ---- ---- ΪΤ ------- ^ [(Please read the notes on the back before filling this page) 4367 9 D8 六、申請專利範園 每個可淸除單元中至少一個的儲存格’保留來儲存控制資 訊。 8. 根據申請專利範圍第2項之記憶體系統,其中控制 器指定保留每個可淸除單元中至少一個的可用儲存格’以 使得其具有包含失效的不可用儲存格、儲存控制資訊所保 留.的儲存格 '以及用來儲存從主系統所收到的資料之可用 儲存格。 9. 根據申請專利範圍第7項之記憶體系統,其中一些 所保留的儲存格用來儲存位址轉換資訊,而其位址轉換資 訊則是用來轉換一個來自主系統的位址,成爲一個適合存 取記憶體的位址。 10. 根據申請專利範圍第7項之記憶體系統,其中每一 個所保留的儲存格用來儲存指向下一個保留儲存格的指標 資訊,直到抵及最後一個保留儲存格,其指向儲存適合存 取記憶體的位址之記憶體中的位址。 11. 根據申請專利範圍第1項之記憶體系統,其中由於 具有多數個連接在一起而形成儲存格體系的儲存格,而實 現位址的轉換,其中藉由來自主系統所供應的一個邏輯位 址之不同位元,來爲儲存格體系的不同層級定址’以使得 最低儲存格不是提供所需的實際實體位址,便是提供指向 另一個儲存格體系的指標資訊。 I2·根據申請專利範圍第11項之記憶體系統,其中的 儲存格體系則是由保留來儲存控制資訊的儲存格所組成。 Π.根據申請專利範圍第7項之記憶體系統,其中建構 2 -δ 句-' (請先閲讀背面之注意事項再填寫本f)4367 9 D8 VI. Patent Application Park Each cell of at least one of the erasable units is reserved to store control information. 8. The memory system according to item 2 of the scope of the patent application, wherein the controller designates to retain at least one of the available cells in each of the erasable units so that it has reserved unusable cells and storage control information. 'Cells' and available cells for storing data received from the host system. 9. According to the memory system of the 7th patent application, some of the reserved cells are used to store the address translation information, and the address translation information is used to convert an address from the main system into a An address suitable for accessing memory. 10. The memory system according to item 7 of the scope of the patent application, wherein each reserved cell is used to store index information pointing to the next reserved cell until the last reserved cell is reached, which points to storage suitable for access The address in memory. 11. The memory system according to item 1 of the scope of the patent application, in which a plurality of cells connected together to form a cell system realize address conversion, in which a logical address supplied from the main system is used Different bits to address different levels of the cell system 'so that the lowest cell provides either the actual physical address required or the index information pointing to another cell system. I2. The memory system according to item 11 of the patent application, where the cell system is composed of cells reserved to store control information. Π. The memory system according to item 7 of the scope of the patent application, which constructs a 2 -δ sentence-'(Please read the precautions on the back before filling in this f) 夂地*乂度適用中國國家標準(CNS ) A4坭格(210X297公着) 436795 A8 B8 C8 D8 1%这料修正 L 補充 、申請專利範圍 所保留的儲存格具有多數個的登錄元,並且其中的每個登 錄元儲存多數個的欄位’因此—個儲存於登錄元中的欄位 用來指向儲存於另一個登錄元的欄位’並且認爲只有儲存 於一個登錄元中最近所塡寫的欄位是有效的。 14. 一種使用於具有位置的非揮發性記憶體之控制器’ 其中的位置能夠個別地寫入’但卻只能以整個區塊的位置 來淸除,藉以在使用中,控制器設置至少一個的可淸除單 元,其中每一個可淸除單元包含至少一個的可淸除區塊, 控制器將每一個可淸除單元細分成稱爲儲存格的記憶位置 群,以及控制器基於每個儲存格而從儲存格中讀取資料結 構,並且將資料結構寫入儲存格中。 15. —種使用於控制器的非揮發性記憶體,該記億體具 有能夠個別地寫入,但卻只能以整個區塊位置來淸除的位 置,建構此非揮發性記憶體以使得至少設置一個可淸除單 元,其中每一個可淸除單元包含至少一個可淸除區塊’並 且每一個可淸除區塊細分成稱爲儲存格的記憶位置群’以 使得控制器基於每個儲存格,而從儲存格中讀取資料結構 ,並且將資料結構寫入儲存格中。 16. 根據申請專利範圍第15項之非揮發性記憶體’其 中指定每個可淸除單元中至少一個的儲存格’保留來儲存 控制資訊。 17. 根據申請專利範圍第16項之非揮發性記憶體’其 中多數個的保留儲存格連接在一起,以形成一個儲存格體 系,用來實現位址的轉換,並且其中藉由來自主系統所供 ,—裝 訂 τ線-^ (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧时產AM工消t合作社印% 本紙張义度適用中國國家標準(CNS > A4現格(2! Ο X 297公釐)夂 地 * 乂 Applicable to China National Standard (CNS) A4 grid (210X297) 436795 A8 B8 C8 D8 1% This material is modified L supplement, and the cell reserved by the scope of patent application has a large number of registered elements, and among them Each registry element stores multiple fields 'hence-a field stored in a registry element is used to point to a field stored in another registry element' and it is considered that only the most recent script stored in one registry element is written The fields are valid. 14. A controller for non-volatile memory with a position, where the position can be individually written, but can only be erased by the position of the entire block, so that in use, the controller is provided with at least one Erasable units, where each erasable unit contains at least one erasable block, the controller subdivides each erasable unit into a memory location group called a cell, and the controller is based on each storage Read the data structure from the cell and write the data structure to the cell. 15. A type of non-volatile memory used in the controller. The memory can be individually written, but can only be erased by the entire block position. This non-volatile memory is constructed so that At least one erasable unit is provided, wherein each erasable unit contains at least one erasable block 'and each erasable block is subdivided into a memory location group called a cell' so that the controller is based on each Cell, and read the data structure from the cell and write the data structure to the cell. 16. According to the non-volatile memory of item 15 of the patent application, at least one cell in each erasable unit is designated to be reserved for storing control information. 17. According to the non-volatile memory of item 16 of the patent application, most of the reserved cells are connected together to form a cell system, which is used for address conversion, and provided by the host system. , —Binding τ 线-^ (Please read the precautions on the back before filling out this page) Printed by AM Ministry of Economic Affairs AM Industrial Consumer Cooperative Co., Ltd.% This paper is applicable to Chinese national standards (CNS > A4 now (2! Ο X 297 mm) 43679¾ C8 D8 々、申請專利範圍 應的一個邏輯位址之不同位元,來爲儲存格體系的不同層 級定址,以使得最低儲存格不是提供所需的實際實體位址 ,便是提供指向另一個儲存格體系的指標資訊。 ----------^_ 裝--^-----訂 1-----:線 (請先閱讀背面之注意事項再填寫本頁) 經濟部智总財-^诗工消^合作社卬製 卷紙長义度边用中國國家標準(CNS ) A4現格(210X 297公釐)43679¾ C8 D8 々 Different bits of a logical address should be applied for patent application to address different levels of the cell system, so that the lowest cell provides either the actual physical address required or a pointer to another Metric information for the cell system. ---------- ^ _ Install-^ ----- Order 1 -----: line (please read the precautions on the back before filling out this page) Poetry Consumers Cooperative Co., Ltd. produced rolled paper with long-term use and Chinese National Standard (CNS) A4 (210X 297 mm)
TW88100946A 1999-01-21 1999-01-21 Memory system TW436795B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW88100946A TW436795B (en) 1999-01-21 1999-01-21 Memory system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW88100946A TW436795B (en) 1999-01-21 1999-01-21 Memory system

Publications (1)

Publication Number Publication Date
TW436795B true TW436795B (en) 2001-05-28

Family

ID=21639482

Family Applications (1)

Application Number Title Priority Date Filing Date
TW88100946A TW436795B (en) 1999-01-21 1999-01-21 Memory system

Country Status (1)

Country Link
TW (1) TW436795B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9665478B2 (en) 2002-12-09 2017-05-30 Innovative Memory Systems, Inc. Zone boundary adjustments for defects in non-volatile memories

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9665478B2 (en) 2002-12-09 2017-05-30 Innovative Memory Systems, Inc. Zone boundary adjustments for defects in non-volatile memories

Similar Documents

Publication Publication Date Title
JP4787266B2 (en) Scratch pad block
US8792277B2 (en) Split data error correction code circuits
KR100684887B1 (en) Data storing device including flash memory and merge method of thereof
US6725321B1 (en) Memory system
EP0712067B1 (en) Flash disk card
KR0128218B1 (en) Nonvolatile memory with flash erase capability
TWI417717B (en) Non-volatile memory and method with multi-stream updating
KR101076830B1 (en) Partial block data programming and reading operations in a non-volatile memory
US7752412B2 (en) Methods of managing file allocation table information
TWI221553B (en) Semiconductor memory device with block alignment function
US7681008B2 (en) Systems for managing file allocation table information
US8296503B2 (en) Data updating and recovering methods for a non-volatile memory array
WO1999038066A1 (en) Flash memory system
WO2002054247A1 (en) Data recorder and method for recording data in flash memory
JP2007094764A (en) Memory system
JPWO2007000862A1 (en) MEMORY CONTROLLER, NONVOLATILE MEMORY DEVICE, NONVOLATILE MEMORY SYSTEM, AND DATA WRITE METHOD
WO2006009322A2 (en) Memory card, nonvolatile semiconductor memory, and method of controlling semiconductor memory
CN109521944A (en) data storage device and data storage method
US20100325524A1 (en) Control circuit capable of identifying error data in flash memory and storage system and method thereof
EP1139210A1 (en) Method of logic partitioning of a nonvolatile memory array
JP4661497B2 (en) Memory controller, flash memory system, and flash memory control method
TW436795B (en) Memory system
US20060059296A1 (en) Emulating small block size of flash memory
TW200832132A (en) Managing file allocation table information
JP4177301B2 (en) Memory controller, flash memory system, and flash memory control method

Legal Events

Date Code Title Description
GD4A Issue of patent certificate for granted invention patent
MK4A Expiration of patent term of an invention patent