TWI327711B - Boot-switching apparatus and method for multiprocessor and multi-memory system - Google Patents

Boot-switching apparatus and method for multiprocessor and multi-memory system Download PDF

Info

Publication number
TWI327711B
TWI327711B TW96108929A TW96108929A TWI327711B TW I327711 B TWI327711 B TW I327711B TW 96108929 A TW96108929 A TW 96108929A TW 96108929 A TW96108929 A TW 96108929A TW I327711 B TWI327711 B TW I327711B
Authority
TW
Taiwan
Prior art keywords
memory
processor
signal
enable
switching
Prior art date
Application number
TW96108929A
Other languages
Chinese (zh)
Other versions
TW200837632A (en
Inventor
Ying Chih Lu
Original Assignee
Inventec Corp
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 Inventec Corp filed Critical Inventec Corp
Priority to TW96108929A priority Critical patent/TWI327711B/en
Publication of TW200837632A publication Critical patent/TW200837632A/en
Application granted granted Critical
Publication of TWI327711B publication Critical patent/TWI327711B/en

Links

Landscapes

  • Hardware Redundancy (AREA)
  • Stored Programmes (AREA)
  • Techniques For Improving Reliability Of Storages (AREA)

Description

1327711 IPD070011TW 23277twf.doc/n 九、發明說明: 【發明所屬之技術領域】 本發明是有關於一種開機裝置及方法,且特別是有關 於一種使用單一計時器執行多處理器及多記憶體之切換的 開機裝置及方法。 【先前技術】 過去幾年來,隨著資訊科技的進步,處理器的速度也 壬倍數成長,為了取得領導的地位,處理器大廠英代爾 (Intel)及超微(AMD)莫不致力開發更高頻率的處理器。 時至今日,處理器的時脈週期已經從10〇百萬赫茲(MHz) 進化到了 1千兆赫茲(GHz)以上。然而,隨著處理器時 脈週期的提高,硬體的複雜性也隨之增加,而處理器廠商 在追求速度之餘,也意識到了處理器的速度仍有其極限, 不可能如此無限制地發展下去,此時便需要採用其他的 法。 據此,多處理器(Multiprocessor)架構則發展成為最 新一代的處理器技術,利用將多個處理器串接在一起,而 達到多工處理的理想境界。此多處理κ统在開機時,必 須選擇其中—個處理ϋ做為啟動(Bootstrap)處理器,用 以執打系統開機的程序,而其餘的處理器顺視為應用 (Application)處理器,以做為啟動處理器的輔助之用。 其中,當系統開機時,各個處理器均會執行一個内建的自 我測試功能,假如有任何一個處理器因為某種因素發生開 機錯誤時,則會送出-個狀態旗標(Statusflag)以^知 5 1327711 IPD070011TW 23277twf.doc/n 、 統此狀況,而由系統做後續處理。然而,若是原本的啟動 處理器發生開機錯誤時,則必須改用其它的應用處理器取 代其任務,才能正常地執行開機功能。 ; 在此情況下,習知的作法是採用一種錯誤回復開機 . (Fault Resilient Booting, FRB )技術,藉由基本輸入輸出 系統(Basic Input/Output System, BIOS )、基板管理控制 态(Baseboard Management Controller,BMC)或是其他硬 φ 體執行開機回復的動作。以下即以包括兩個處理器的系統 • 為例,介紹習知的FRB技術。圖1所繪示為習知雙處理器 .°系統使用BMC進行FRB的架構圖,而圖2所繪示為習知 雙處理态系統使用BMC進行FRB的方法流程圖。請先參 照圖1,在此雙處理器系統1〇〇中,處理器11〇及處理器 120为別與BMC130的致能針腳131及致能針腳132相連 接,其中,處理器11〇係做為開機用的處理器。此外, BMC130中則内建—個馳計時器133,並與m〇s系統 的個通用輸入輪出(General purp〇se j叩ut/〇utput,) •針腳140相連接。 ’ 請同時參照圖1及圖2,在系統1〇〇開機之前,即先 • 在腕⑶0中设定處理器執行FRB的功能(步驟S210), -包括設置一個暫停時間(Time-out),並將BMC130的致 能針,131及致能針腳I32設置為卜使得處理器110及 處理裔120在開機後均會被致能(Enable)。每當系統100 開機(P〇Wer〇n)或是重新啟動(Reset)(步驟S220)時, BMC130的FRB計時器133即會被啟動而開始倒數暫停時 6 1327711 IPD070011TW 23277twf.doc/n 間(步驟S230) ’同時也會判斷是否接收到計時器禁能訊 號(步驟S240)。其中,當BIOS系統成功執行開^自我 測試(Power On Self Test,POST)時,即可藉由通用輸入 輸出針腳140發出禁能訊號通知BMC130將frb計時器 133禁能(Disable),此時系統1〇〇的處理器n〇及處理 器120則可繼續進行正常的開機程序(步驟S25〇)。然而, 若是開機用的處理器110在開機時即發生錯誤,則代表系 統100無法正常開機,此時FRB計時器133即不會被BI〇s 系統禁能’因此當判斷FRB計時器133的暫停時間倒數結 束(步驟S260 )時,BMC 130則會藉由將致能針腳131設 置為0而關閉處理器110的功能(步驟S270),並將系統 100重新啟動’而改由處理器120取代處理器11〇進行開 機(步驟S280)。 上述原理也同樣運用在雙BIOS唯讀記憶體(Read Only Memory,ROM)的技術上,雙BIOS ROM技術顧名 思義即是在主機板上配置兩個BIOS ROM以存放BIOS系 統的資料,採用雙BIOS ROM的系統即可避免在升級BIOS 時’因為發生BIOS文件與主機板不匹配、BIOS文件被修 改或是升級過程中斷電等情況,而造成BIOS升級失敗進 而導致系統無法開機的窘境。 圖3所繪示為習知雙BIOS ROM系統的架構圖,而圖 4則繪示為習知雙BIOS ROM系統的運作流程圖。請先參 照圖3,此BIOS ROM系統300包括一個輸入輸出控制 (Super 1/0, SIO)晶片310,此SIO晶片310即透過位址/ 7 1327711 IPD070011TW 23277twf.doc/n \ 資料(Address/Data)線與唯讀記憶體320及唯讀記憶體 330相連接’並藉由一個選擇針腳311分別與唯讀記恃體 320及唯讀記憶體330相連接’其中,在此假設唯讀記憶 體320係做為主要開機之用,而唯讀記憶體33〇則為備用, 此時選擇針腳311送出的訊號為1,而將唯讀記憶體32〇 致能,此訊號經由反向器340轉換後,則送到唯讀記憶體 330而將其禁能。此外,SIO晶片31〇中則配置有一個^^ 計時器312,用以做為選擇由唯讀記憶體32〇及唯讀記憶 體330讀取BIOS之用。 請同時參照圖3及圖4 ’在將系統300開機或重新啟 動(步驟S410)後,即啟動ROM計時器312開始倒數暫 停時間(步驟S420),並判斷是否接收到計時器禁能訊號 (步驟S430),若BIOS系統可正常執行P〇ST,則可同1 f藉由一個通用輸入輸出針腳350發送禁能訊號通知§1〇 晶片31〇將R〇M計時器312禁能。此時即不會影響到選 擇針腳311輸出的訊號’仍然沿用唯讀記憶體32〇進行開 機(步騍S440)。然而,若唯讀記憶體320在開機時發生1327711 IPD070011TW 23277twf.doc/n IX. Description of the Invention: [Technical Field] The present invention relates to a booting device and method, and more particularly to a multi-processor and multi-memory switching using a single timer Boot device and method. [Prior Art] Over the past few years, with the advancement of information technology, the speed of processors has also multiplied. In order to achieve leadership status, processor manufacturers Intel and AMD are not working hard to develop more. High frequency processor. Today, the processor's clock cycle has evolved from 10 megahertz (MHz) to more than 1 gigahertz (GHz). However, as the processor clock cycle increases, so does the complexity of the hardware. Processor manufacturers, in pursuit of speed, realize that the speed of the processor still has its limits. To develop, other laws are needed at this time. As a result, the multiprocessor architecture has evolved into the latest generation of processor technology, using the ideal combination of multiple processors in tandem to achieve multiplex processing. When the multi-processing system is powered on, it must select one of the processing ports as a bootstrap processor to execute the program for booting the system, and the remaining processors are regarded as application processors. As an aid to booting the processor. Among them, when the system is powered on, each processor will perform a built-in self-test function. If any processor has a boot error due to some factor, it will send a status flag to know. 5 1327711 IPD070011TW 23277twf.doc/n, this situation is the same, and the system does the follow-up. However, if the original boot processor has a boot error, it must use another application processor to replace its task in order to perform the boot function normally. In this case, the conventional practice is to use a Fault Resilient Booting (FRB) technology, through the Basic Input/Output System (BIOS), and the Baseboard Management Controller (Baseboard Management Controller). , BMC) or other hard φ body to perform the boot response action. The following is an introduction to the conventional FRB technology with a system including two processors. 1 is a schematic diagram of a conventional dual processor system using a BMC for FRB, and FIG. 2 is a flow chart of a conventional dual processing system using BMC for FRB. Referring to FIG. 1 , in the dual processor system, the processor 11 and the processor 120 are connected to the enabling pin 131 and the enabling pin 132 of the BMC 130. The processor for booting. In addition, the BMC 130 has a built-in timer 133 and is connected to a universal input wheel (General purp〇se j叩ut/〇utput,) of the m〇s system. 'Please refer to FIG. 1 and FIG. 2 at the same time, before the system 1 is turned on, that is, firstly, the function of executing the FRB by the processor in the wrist (3) 0 (step S210), including setting a pause time (Time-out), The enabler pin 131 and the enable pin I32 of the BMC 130 are set such that the processor 110 and the processor 120 are enabled after being turned on. Whenever the system 100 is powered on (P〇Wer〇n) or restarted (step S220), the BMC 130's FRB timer 133 will be activated to start the countdown pause 6 1327711 IPD070011TW 23277twf.doc/n ( Step S230) 'It is also determined whether a timer disable signal is received (step S240). When the BIOS system successfully performs the Power On Self Test (POST), the general-purpose input/output pin 140 can be used to issue a disable signal to notify the BMC 130 to disable the frb timer 133 (Disable). The processor 〇 and the processor 120 can continue the normal boot process (step S25 〇). However, if the processor 110 for booting fails when booting, the system 100 cannot be turned on normally, and the FRB timer 133 is not disabled by the BI〇s system. Therefore, when the FRB timer 133 is suspended, it is judged. When the time countdown ends (step S260), the BMC 130 turns off the function of the processor 110 by setting the enable pin 131 to 0 (step S270), and restarts the system 100 instead of being processed by the processor 120. The device 11 is turned on (step S280). The above principle is also applied to the technology of dual BIOS read only memory (ROM). The dual BIOS ROM technology, as the name implies, is to configure two BIOS ROMs on the motherboard to store the BIOS system data, using dual BIOS ROM. The system can avoid the dilemma when the BIOS is upgraded because the BIOS file does not match the motherboard, the BIOS file is modified, or the upgrade process is interrupted. FIG. 3 is a block diagram of a conventional dual BIOS ROM system, and FIG. 4 is a flow chart showing the operation of a conventional dual BIOS ROM system. Referring first to FIG. 3, the BIOS ROM system 300 includes an input/output control (Super 1/0, SIO) chip 310, which is transmitted through the address / 7 1327711 IPD070011TW 23277twf.doc/n \ data (Address/Data The line is connected to the read-only memory 320 and the read-only memory 330 and is connected to the read-only memory 320 and the read-only memory 330 by a selection pin 311, respectively, where a read-only memory is assumed. The 320 series is used as the main power-on, and the read-only memory 33 is reserved. At this time, the signal sent by the selection pin 311 is 1, and the read-only memory 32 is enabled. This signal is converted via the inverter 340. After that, it is sent to the read-only memory 330 to disable it. In addition, a timer 312 is disposed in the SIO chip 31 for selecting the BIOS to be read by the read only memory 32 and the read only memory 330. Referring to FIG. 3 and FIG. 4 simultaneously, after the system 300 is powered on or restarted (step S410), the ROM timer 312 is started to start the countdown pause time (step S420), and it is determined whether the timer disable signal is received (step S430), if the BIOS system can execute P〇ST normally, the R can be disabled by sending a disable signal to the DDR1 by a general-purpose input/output pin 350. At this time, the signal that does not affect the output of the selection pin 311' is still started with the read-only memory 32 (step S440). However, if read-only memory 320 occurs at boot time

錯誤’此時B丨〇 S系統即無法正常開機,也就無法禁能R〇M 計時器312,而當判斷ROM計時器312的暫停時間倒數完 畢(步驟S450)時,即會進行唯讀記憶體的切換(步驟 S460),將選擇針腳311送出的訊號設為〇,此時唯讀記 隐體320即會被禁能’並改為致能唯讀記憶體33〇。最後 則將系統重新啟動,而由唯讀記憶體330進行開機動作(+ 驟 S470) 。 乂 8 1327711 IPD070011TW 23277twf.doc/n 由上述内容可知,多處理 均各自採用-個計時器做為切it及雙刪讀系统 又發生錯誤時,财可能會,而系統開機 的計時器先倒數完成,在沒有確處理器) 如是BIOS R〇M)所造成蚌,p涘疋否為另一方(例 系統誤判的可能性。__ 而造成 作,但是預設的器均能正常卫 器的計時器先倒數完畢,系统:機錯騎,則由於處理 而重新開機。在此情況下,最^理f發生錯誤 機。 …、去成功地啟動系統進行開 【發明内容】 本發明提供一種多處理器及多 =置僅_與處理器連接的_針腳;測 =:=r個計時器來判斷發生錯誤的處理= 本發明提供一種多心==開機: 開機切換裂置判斷系統無法開'=表::-個 誤或無法正常運作的處理器或記憶Γ而錯 開機動作。 而月匕夠正吊的執行 換方==在供—種7理^多記憶體系統之開機切 、 -在糸統開機之初倒數—暫停時間,並在倒數 9 1327711 IPD070011TW 23277twf.doc/n 完畢時’先判斷處理p 是否發生錯誤,而找“ 後再判斷記憶體 本發明提出-種多卢二 吊開機的問題。 換裝置,此裝置係配置^^ $及多記憶體系統之開機切 系統,其^=^=_;*叫個記憶趙之 多個記憶體選擇針腳及—個::。夕:處理器偵測針腳、 腳係分別輕接至處理器個:^ ’處理器致能針 有-致能訊號,以將處理器致:=,針:上均設置 亦分別輕接處理器,而這此處哭:处理益偵/則針腳 應之處理器的工作狀態,I卿適用於侦測對 擇針腳則耦接至上述之記憔S付二測訊號。記憶體選 以在這些記憶财娜。置-個選擇訊號, 間Γτ. 、 a十時态中包括設定有一個暫停時 倒數,二二士^在系統_時’從此暫停時間開始 腳上A-間倒數完畢時,判斷各個處理器致能針 符,it 财媽侧_的_訊號是否相 :、中右致能訊號與摘測訊號不符,則透過對應之處理 針腳將處理器禁能;若所有的致能訊號與積測訊號 :付’則透過錢、體選擇針腳切換所使用的記憶體。 在本發明之一實施例中,上述之多處理器及多記憶體 =统^機切換裝置’更包括一個警示裝置,其適於在所 5己憶體均被切換過而系統仍無法正常開機時,發出一 警示訊息。 、 在本發明之一實施例中,上述之記憶體為儲存基本輸 入輪出系統(Basic Input/Output System,BIOS)之唯讀記 1327711Error 'At this time, the B丨〇S system cannot be turned on normally, and the R〇M timer 312 cannot be disabled. When it is judged that the pause time of the ROM timer 312 is completed (step S450), the read-only memory is performed. The switching of the body (step S460) sets the signal sent by the selection pin 311 to 〇, and at this time, only the hidden object 320 is disabled and the read-only memory 33 is enabled. Finally, the system is restarted, and the read-only memory 330 is turned on (+ step S470).乂8 1327711 IPD070011TW 23277twf.doc/n It can be seen from the above that when multiple processors use one timer as the cut and double read system, when the error occurs, the money may be generated, and the timer of the system startup is counted down first. , in the absence of a processor (such as BIOS R〇M) caused by 蚌, p涘疋 is the other party (such as the possibility of system misjudgment. __ caused by, but the default device can be a normal timer After the countdown is completed, the system: the machine is wrongly riding, and then restarted due to the processing. In this case, the fault occurs in the most f. ..., to successfully start the system to open [invention] The present invention provides a multiprocessor And more = set only _ pin connected to the processor; test =: = r timers to determine the processing of the error = The present invention provides a multi-heart == boot: boot switching crack determination system can not open '= table ::- A faulty or inoperable processor or memory is wrong to start the action. And the moon is enough to hang the execution of the converter == in the supply - 7 kinds of ^ multi-memory system boot, - in the 糸Countdown to the beginning of the boot - pause time, and When the countdown 9 1327711 IPD070011TW 23277twf.doc/n is completed, 'first judge whether the processing p has an error, and then find the "memory of the memory proposed by the present invention - the problem of the multi-lulu two-lift boot. Replace the device, this device is configured ^ ^ $ and multi-memory system boot cutting system, its ^ = ^ = _; * called a memory of Zhao's multiple memory selection pins and a::: eve: processor detection pin, foot system respectively To the processor: ^ 'The processor enable pin has a - enable signal to the processor to: =, pin: the upper setting is also lightly connected to the processor, and here cry: processing benefit detection / then pin In response to the working state of the processor, I Qing is suitable for detecting the selected pin and is coupled to the above-mentioned record S. The memory is selected in these memories. Set the signal, Γτ. , a ten-time state includes setting a reciprocal when paused, and two or two ^ in the system _ when 'the pause time from the start of the foot on the A-between count, the judgment of each processor enable pin, it wealth side _ Whether the signal of the _ signal is correct: if the middle right signal does not match the signal, the corresponding The processing pin disables the processor; if all the enabling signals and the integrated signal: pay', the memory used is switched by the money and the body selection stitch. In one embodiment of the present invention, the multiprocessor described above The multi-memory=system switching device further includes a warning device adapted to issue a warning message when the 5th memory is switched and the system is still unable to be normally turned on. In an embodiment of the present invention In the above memory, the memory is stored in the Basic Input/Output System (BIOS), only reading 1327711

IPD070011TW 23277twf.d〇c/n 憶體(Read Only Memory,ROM)。上述之開機切換裝置 匕括配置於基板言理控制(Baseb〇ard Management Control, BMC)襄置中。此外’上述之計時器包括錯誤回復開機 (Fault Resilient Booting,FRB)計時器。 在本發明之-實施例中,上述之處理器致能針腳之致 能訊號包括均預先設置為致能狀態。$一方面,上述之記 ,體選擇針腳之選擇訊號包括將上述記憶體其中之一預先 叹置為致缺態’喊餘記紐則設置為禁紐態,而當 =要切換記憶料,則包括將另—個魏體設置為致能狀 恶’而其餘記憶體則設置為禁能狀態。 ^發明提出-種多處理器及多記憶體系統,其中包括 2處:器、多個記憶體,以及開機切換裝置。此開機換 L括多個處理1^致能針腳、多個處理器偵測針腳、 夕個記憶體選擇針腳及一個計時器。 :係分接至處理器,而各個處理器致能二= 理,或禁能。處理器偵測針腳 7接處理5,而追些處理⑸貞測針腳適用於谓測對 作狀態,並獲得—個_訊號。記憶體選 以在、_至上34之記憶體’適於設置—個選擇訊號, 間^記憶體中切換。計時器中包括設定有一個暫停時 甸數,乂適-於在系統開機時’從此暫停時間開始 :上=訊號及各個處理器_針腳的侧訊號是否相 -中右致能訊號與制訊號不符,則透過對應之處理 η 1327711 IPD0700HTW 23277twf.d〇c/n 器致能針腳將處理器禁能;若 均相兮,目右所有的致此讯唬與偵測訊號 物付’則透較憶體選擇針腳切換所使 在本發明之-實施例中,更包括一個多工器j传配 針腳及記憶體之間’適於將選擇訊號傳遞 在本發明之一實施例中,更包括組,⑽ 配置於處理㈣測針腳及處理器之間,適於偵測處理器是 否運作正常’並將對應之翻訊號發送給處理以貞測針腳。 本發明提出-種多處理!!及多記憶體系統之開機切 換方法’適驗包括?個處理肢多他憶體之系統,此 方法包括下列步驟'統職時,開始倒數一個 暫停時間’在倒數中判斷是否接收到計時器禁能訊號,若 未接收到计時為禁能訊號而倒數完畢時,則進一步偵測處 理器是否正常運作’並取得對應之多個制職。接著則 判斷各_測訊號是否與對應 < 致能m號相符,若積測訊 唬與對應之致能訊號不符,則將致能訊號設置為禁能狀 態,以將對應之處理器禁能;反之,若所有之偵測訊號均 與對應之致能訊號相符,則調整發送給記憶體之選擇訊 號’以切換用以開機的記憶體。 在本發明之一實施例t,在判斷是否接收到計時器禁 能訊號的步驟之後,若接收到計時器禁能訊號,則以目前 設定之處理器及記憶體執行開機動作。 在本發明之一實施例t ’上述取消暫停時間之倒數的 方式包括在基本輸入輸出系統正常執行開機動作時,透過 12 1327711 IPD070011TW 23277twf.doc/n 基本輸入輸出糸統取消暫停時間的倒數。 在本發明之一實施例中,在將對應之處理器禁能,以 及切換記憶體的步驟之後,更包括將系統重新啟動,並以 切換後的處理器及記憶體執行開機動作。IPD070011TW 23277twf.d〇c/n Read Only Memory (ROM). The power-on switching device described above is disposed in a BaseBardard Management Control (BMC) device. In addition, the above timer includes a Fault Resilient Booting (FRB) timer. In an embodiment of the invention, the enabling signals of the processor enable pins described above are each pre-set to an enabled state. On the one hand, the above-mentioned note, the selection signal of the body selection stitch includes pre-slipping one of the above-mentioned memories into a missing state, and the shouting is set to the forbidden state, and when = the memory is to be switched, This includes setting the other Wei body to enable the evil and the rest of the memory is set to the disabled state. The invention proposes a multiprocessor and multi-memory system, which includes two places: a device, a plurality of memories, and a power-on switching device. The power-on switch includes a plurality of processing 1^ enable pins, a plurality of processor detection pins, a memory selection pin, and a timer. : The system is tapped to the processor, and each processor is enabled or disabled. The processor detects pin 7 and processes 5, and pursues some processing (5). The pin is suitable for the status of the counter test and obtains a signal. The memory is selected by the memory of the _ to the upper 34, which is suitable for setting a selection signal, and switching between the memory. The timer includes setting a pause number, which is appropriate - when the system is turned on, 'from the pause time: the upper signal and the side signal of each processor_pin are phase-to-right. The enable signal does not match the signal. , the processor is disabled by the corresponding processing η 1327711 IPD0700HTW 23277twf.d〇c/n enable pin; if the phase is right, all the messages to the right and the detection signal are 'received' The body selection pin switching is such that, in the embodiment of the present invention, a multiplexer j is also provided between the pin and the memory, which is adapted to transfer the selection signal in an embodiment of the present invention, and further includes a group. (10) Configured between processing (4) between the stylus and the processor, it is suitable for detecting whether the processor is working properly and sending the corresponding rip signal to the processing to detect the stitch. The invention proposes a multi-processing! ! And the booting method of multi-memory systems. A system that handles multiple limbs, this method includes the following steps: when starting a job, start counting down a pause time. In the countdown, determine whether a timer disable signal is received. If the timer is not received, the timer is disabled. When the countdown is completed, it further detects whether the processor is functioning properly and obtains corresponding multiple jobs. Then, it is determined whether each _test signal is consistent with the corresponding < enable m number. If the product test signal does not match the corresponding enable signal, the enable signal is set to the disabled state to disable the corresponding processor. On the other hand, if all the detection signals match the corresponding enable signals, adjust the selection signal sent to the memory to switch the memory for booting. In an embodiment t of the present invention, after the step of determining whether the timer disable signal is received, if the timer disable signal is received, the power-on operation is performed by the currently set processor and the memory. In one embodiment of the present invention, the manner in which the reciprocal of the pause time is canceled includes the reversal of the pause time by the 12 1327711 IPD070011TW 23277twf.doc/n basic input/output system when the basic input/output system normally performs the power-on action. In an embodiment of the present invention, after the step of disabling the corresponding processor and switching the memory, the system further includes restarting the system and performing the booting operation with the switched processor and the memory.

在本發明之一實施例中’在將系統開機之前,更包括 將所有的致能訊號設置為致能狀態,以及選擇一個記憶體 做為開機記憶體,並將其對應之選擇訊號設置為致能狀 態,而其餘記憶體之選擇訊號則設置為禁能狀態。 在本發明之一實施例中,切換記憶體的步驟包括將原 本選擇之記憶體的選擇訊號設置為禁能狀態,以及在剩餘 之§己憶體中選擇一個做為開機記憶體,並將其對應之選擇 訊號設置為致能狀態。 在本發明之-實施例中’在所有記憶體均被選擇過, 而系統仍無法正常開機時,包括發出—則訊自。In an embodiment of the present invention, before the system is powered on, the method further includes setting all the enable signals to the enabled state, and selecting a memory as the boot memory, and setting the corresponding selection signal to The state can be set, and the selection signal of the remaining memory is set to the disabled state. In an embodiment of the invention, the step of switching the memory includes setting the selection signal of the originally selected memory to the disabled state, and selecting one of the remaining § memories as the boot memory, and The corresponding selection signal is set to the enabled state. In the embodiment of the invention - when all the memory has been selected, and the system still fails to boot normally, including issuing - then the message.

,刚採用單一個計時器判斷系統的處:器或記 憶體疋否在職發生錯誤,並藉㈣ = = 再決定是否切換記憶體,而能夠找出 、二取用其他的處理器或記憶體重新進行 ^缺決先前技術中,無法正確判斷錯誤發生來 13 1PD070011XW 23277twf.doc/i RDM為彡統在機發倾辦,自於FRB計時器及 置上上奴之暫停時糾同,而造絲選擇代用裝 可此存在的風險,較佳的做法即是將兩個計時器合而 触並在發現開機錯誤時,分別 題’以正碟的复清錯誤發生的原因,讓系 心地選擇替代的處理器或記紐來啟動。本發明即是基 於上述概念所發展^來的—套多處理器及多記憶體系統之 開機切換裝置及方法。為了使本發明之内容更為明瞭,以 下特舉實施例作為本發明確實能夠據以實施的範例。 圖5是依照本發明—實施例所繪示的多處理器及多記 憶體系統的方塊圖。請參照圖5,本實施例之多處理器及 多記憶體系統500包括基板管理控制器(baseboardJust use a single timer to judge the system: the device or the memory is wrong, and borrow (4) = = to decide whether to switch the memory, but to find out, and use another processor or memory to re- In the prior art, the error cannot be correctly judged. 13 1PD070011XW 23277twf.doc/i RDM is the machine in the machine, since the FRB timer and the suspension of the slave, the same, and the silk Choosing a substitute can be a risk. The better way is to combine the two timers and when you find the power-on error, ask the question of the occurrence of the correcting error of the original disc, and let the system choose the alternative. The processor or the counter is activated. The present invention is based on the above-described concept - a multi-processor and multi-memory system boot switching apparatus and method. In order to clarify the content of the present invention, the following specific examples are given as examples in which the present invention can be implemented. Figure 5 is a block diagram of a multiprocessor and multi-memory system in accordance with an embodiment of the present invention. Referring to FIG. 5, the multiprocessor and multi-memory system 500 of the present embodiment includes a baseboard management controller (baseboard).

Management Controller,BMC) 510、偵測模組 520、多工 斋 530、輸入輸出控制(Super Input/Output,SI0 )晶片 540、 η個處理器CPU0、CPU1、…、CPUn-1與m個唯讀記憶 體ROMO、R〇M1、…、R〇Mm-l。其巾,唯讀記憶體R〇M〇、 R0M1.....ROMm-1例如是用以儲存基本輸入輸出系統 (Basic Input/Output System,BIOS )資料之唯讀記憶體。 基板管理控制器510中則包括致能針腳512〜514、>(貞 測針腳515〜517、選擇針腳518及計時器511。致能針腳 512、513、514分別與處理器〇卩1;0、〇?1;卜〇?1;11-1耦接, 適於提供基板管理控制器510設置一個致能訊號,以將處 理器CPU0、CPU1.....CPUn-Ι致能或禁能。其中,這 些致能針腳之致能訊號一般均先設置為致能狀態,也就是 1327711 IPD070011TW 23277twf.doc/n 說,原先是假設每個處理器CPUO、CPU1.....CPUn l 均能正常工作,因此在系統開機時即先將這些處理器 CPU0、CPU1.....CPUn-Ι 致能。 @ 偵測針腳515、516、517則透過偵測模組520分別與 處理器CPU0、CPU1、CPUn-Ι相連接,而用以債測處^ 器CPU0、CPIH、CPUn-Ι的工作狀態,並獲得—個對應 的偵測訊號。其中’偵測模組520係用以偵測各個處理器 CPU0、CPU1.....CPUn-Ι的運作是否正常,並發送對二 之偵測訊號給偵測針腳515-517。 〜 此外’選擇針腳518則透過多工器530分別耦接至唯 讀記憶體ROMO、ROM1、…、R〇Mm-l,而適於提供其 板官理控制器510設置一個選擇訊號,以在唯讀記^體 ROMO、ROM1、…、R〇Mm-l中切換。其中,由於系統 500只需要從一個唯讀記憶體中讀取BI〇s資料即可進行 開機動作,因此在這些由選擇針腳傳送到唯讀記憶體 ROMO ^ ROM1.....ROMm-1的選擇訊號中,例如僅先 設置-個唯敎憶料致能錢,⑽其餘記憶體設置為 禁能狀態。然而,若在開機時發現此職的唯讀記憶體發 ^錯誤,而需切換_的唯讀記憶_,則可將此預設^ 讀記憶體禁能’而選擇將另—個唯讀記憶體設置為致能狀 遙,除此之外’其餘記憶體則仍保持原紋置的禁能狀態, 如此即能夠達到本發明切換記憶體的功效。 再者’基板管理控制器51〇中亦包括有一個計時器 511 ’此计時5 511例如是一個錯誤回復開機…也 15 1327711 IPD070011TW 23277twf.doc/nManagement Controller (BMC) 510, detection module 520, multi-work 530, input/output control (Super Input/Output, SI0) wafer 540, n processors CPU0, CPU1, ..., CPUn-1 and m read-only Memory ROMO, R〇M1, ..., R〇Mm-1. The towel, the read-only memory R〇M〇, R0M1.....ROMm-1 is, for example, a read-only memory for storing basic input/output system (BIOS) data. The substrate management controller 510 includes enable pins 512 to 514, > (measurement pins 515 to 517, selection pins 518, and timer 511. The enable pins 512, 513, and 514 are respectively coupled to the processor 〇卩1; 〇1; 〇? 1; 11-1 coupled, is adapted to provide a substrate management controller 510 to set an enable signal to enable or disable the processor CPU0, CPU1.....CPUn-Ι Among them, the enable signals of these enable pins are generally set to the enable state, that is, 1327711 IPD070011TW 23277twf.doc/n said that it is assumed that each processor CPUO, CPU1.....CPUn l can Normal operation, so the CPU0, CPU1.....CPUn-Ι are enabled first when the system is powered on. @Detection pins 515, 516, 517 pass through the detection module 520 and the processor CPU0, respectively. CPU1, CPUn-Ι are connected, and used to measure the working state of CPU0, CPIH, CPUn-Ι, and obtain a corresponding detection signal. The detection module 520 is used to detect each Whether the processor CPU0, CPU1.....CPUn-Ι is operating normally, and sends the detection signal of the second to the detection pins 515-517. The selection pins 518 are respectively coupled to the read-only memory ROMO, ROM1, ..., R〇Mm-1 through the multiplexer 530, and are adapted to provide their board controller 510 to set a selection signal for reading only. The body ROMO, ROM1, ..., R〇Mm-1 are switched. Among them, since the system 500 only needs to read the BI〇s data from a read-only memory, the boot operation can be performed, and therefore these are transmitted to the selected pin. In the read signal of the ROM-only memory ROMO ^ ROM1.....ROMm-1, for example, only the first set-only memory is enabled, and (10) the remaining memory is set to the disabled state. However, if it is turned on It is found that the read-only memory of this job is wrong, and the read-only memory _ needs to be switched _, then the preset read memory can be disabled and the other read-only memory is set to enable In addition, the rest of the memory remains in the disabled state of the original pattern, so that the function of the switching memory of the present invention can be achieved. Further, the substrate management controller 51 includes a timer 511. 'This timing 5 511 is for example an error recovery boot... also 15 1327711 IPD070011T W 23277twf.doc/n

Resilient Booting,FRB)計時器。由於基板管理控制器510 的電源為獨立供應,因此使用者可以在開啟系統500之 前,即開啟基板管理控制器510的電源,據以設定計時器 511的暫停時間(Time-out),而在系統500的電源開啟的 同時,計時器511則會開始倒數暫停時間。 .輸入輸出控制晶片540係透過位址/資料線分別與唯 5賣§己憶體ROMO、ROM1、…、R〇Mm-l相連接,而在系 统500實際開機時,則會根據這些唯讀記憶體、 R〇M1.....ROMm·1的致能狀態,從被致能的唯讀記憶 體上讀取BIOS系統資料。 值得一提的是’本實施例之多處理器及多記憶體系統 5〇〇還包括一個警示裝置(未繪示),此警示裝置會在唯 % 5己憶體R〇M0、ROM1、…、R〇Mm-l均切換過,而系 统500仍無法正常開機時,發出警示訊息,以告知使用者 窝要將唯讀記憶體ROMO、ROM1、…、R〇Mm-l更換。 本發明還包括基於上述多處理器及多記憶體系統5〇〇 的架構發展一套開機切換方法,以下則另舉一個實施例說 月上述夕處理益及多§己憶體系統中各個元件之間訊號與資 ,傳遞的詳細步驟。圖6是依照本發明—實施例所繪示的 夕處理态及多記憶體系統之開機切換方法的流程圖。請泉 照圖6,在本實施例中,假設所有的處理器均可正常運;/, 因此將所有致能針腳上的致能訊號均設置為致能狀態。此 外,亦可先選定一個唯讀記憶體做為開機記憶體,並將其 對應之選擇訊號設置為致能狀態,而其餘唯讀記憶體之^ 1327711 IPD070011TW 23277twf.doc/n 擇訊號則設置為禁能狀態。在上述的環境均設定完成後, 即可進行下列的步驟: 首先,由使用者將系統500開機,或是由系統5〇〇重 新啟動(步驟S610),而在系統500開機的同時,則會由 計時器511開始倒數一暫停時間(步驟S620)。 而在倒數期間,基板管理控制器510會偵測是否接收 到由BIOS系統之通用輸入輸出針腳(未繪示)傳送而來 的計時禁能訊號,以決定是否取消暫停時間的倒數(步驟 S630)。其中’若接收到計時禁能訊號,則代表系統5〇〇 可正常開機’此時即會將計時器511禁能,停止倒數暫停 時間,並以目前設定之處理器及唯讀記憶體執行開機動作 (步驟 S640)。 反之’右未接收到計時禁能訊號’而在暫停時間的倒 數完畢時,則會進一步比較並判斷各個偵測針腳(例如摘 測針腳515)所接收到的偵測訊號是否與相對應之各個致 能針腳(例如致能針腳512)的致能訊號相符(步驟§65〇)。 其中’若判斷出偵測訊號與對應之致能訊號不符,則 可藉由基板管理控制器510將致能針腳上的致能訊號設置 為禁能狀態,以將對應之處理器禁能(步驟S68〇);反之, 若判斷出偵測訊號與對應之致能訊號相符,則代表處理器 的運作狀態與預設相符,因此也無需對此處理器執行任何 動作。 值得注意的是,若在比較所有之偵測訊號與致能訊號 之後,發現兩者均相符,則可判定所有處理器的運作狀態^ 17 1327711 IPD070011TW 23277twf.doc/n 均與預設相符,此即代錢理器均可正常 論出造成系統無法開機的片因β & 運作,因此可推 即可調整發送給唯讀記憶===記=而來。此時 憶體(步驟S690)。其中,卜5儿刀換至另一個記 憶體的選擇訊號設置為禁能心2 ^將原本選擇之記 中選擇-個做為開機記悴體二:後再從剩餘之記憶體 為致能狀態。 其對紅選擇訊號設置Resilient Booting, FRB) timer. Since the power of the baseboard management controller 510 is independently supplied, the user can turn on the power of the baseboard management controller 510 before the system 500 is turned on, thereby setting the timeout of the timer 511, and in the system. At the same time that the power of the 500 is turned on, the timer 511 starts the countdown pause time. The input/output control chip 540 is respectively connected to the ROM/ROM1, ROM1, ..., R〇Mm-1 through the address/data line, and when the system 500 is actually powered on, it is based on these read-only The enable state of the memory, R〇M1.....ROMm·1, reads the BIOS system data from the enabled read-only memory. It is worth mentioning that the multi-processor and multi-memory system 5 of the embodiment further includes a warning device (not shown), and the warning device will be in the memory, R0, ROM1, ROM1, ... R〇Mm-l has been switched, and when the system 500 still fails to boot normally, a warning message is sent to inform the user to replace the read-only memory ROMO, ROM1, ..., R〇Mm-1. The present invention also includes developing a set of power-on switching methods based on the architecture of the multi-processor and multi-memory system 5, and the following is another embodiment of the present invention. Detailed steps between the signal and the capital. FIG. 6 is a flow chart showing a method for switching on and off of a multi-memory system according to an embodiment of the present invention. Please refer to FIG. 6. In this embodiment, it is assumed that all the processors can be normally operated; /, therefore, the enable signals on all enabled pins are set to the enabled state. In addition, a read-only memory can be selected as the boot memory, and the corresponding selection signal is set to the enabled state, and the remaining read-only memory is set to be the 1327711 IPD070011TW 23277twf.doc/n selection signal. Disabled state. After the above environment settings are completed, the following steps can be performed: First, the system 500 is powered on by the user, or restarted by the system 5 (step S610), and when the system 500 is powered on, The countdown time is started by the timer 511 (step S620). During the countdown period, the baseboard management controller 510 detects whether a timed disable signal transmitted by a general-purpose input/output pin (not shown) of the BIOS system is received to determine whether to cancel the reciprocal of the pause time (step S630). . If 'when the timed disable signal is received, it means the system can be turned on normally.' At this time, the timer 511 will be disabled, the countdown pause time will be stopped, and the current processor and read-only memory will be used to boot. Action (step S640). On the other hand, if the countdown disable signal is not received right, and the countdown of the pause time is completed, it will further compare and judge whether the detection signals received by the respective detection pins (for example, the measurement pin 515) are corresponding to each other. The enable signal for the enable pin (eg, enable pin 512) matches (step § 65〇). If it is determined that the detection signal does not match the corresponding enable signal, the substrate management controller 510 can set the enable signal on the enable pin to the disabled state to disable the corresponding processor (steps). S68〇); conversely, if it is determined that the detection signal matches the corresponding enable signal, the operating state of the processor is consistent with the preset, and thus no action is required on the processor. It is worth noting that if all the detection signals and enable signals are compared and it is found that the two match, then it can be determined that the operating states of all processors ^ 17 1327711 IPD070011TW 23277twf.doc/n are consistent with the preset, this That is, the money processor can normally say that the film that cannot be turned on by the system is operated by β & therefore, it can be adjusted and sent to the read-only memory =====. At this time, the body is restored (step S690). Among them, the selection signal of the switch to the other memory is set to the forbidden heart 2 ^ The selected one is selected as the start-up note 2: then the remaining memory is enabled . Its setting for red selection signal

在執行完步驟S680的處理哭林 憶體切換後,_到步驟S61Q,°°^新驟S_的記 後的處理器及記憶體執行開機新欠動:統’並以切換 理哭及々愔·I#廿户,機動作,如此重複執行檢查處 後究;:選到二組運換記憶體’最 開機動作,,萬—在執==== 的處理μ是所㈣域截均紐正常工作,貞丨代表做任After the execution of the process of step S680, the crying memory is switched, _ to step S61Q, and the processor and memory of the new step S_ are executed to start a new undershoot: "and switch to cry and 々愔·I# Seto, machine action, so repeat the execution of the inspection office;: select the two groups of memory memory 'the most power-on action, 10,000-the processing of the ==== μ is the (four) domain intercept New Zealand is working normally, and the representative is doing the job.

何切換也沒有用,此時即可發出—則警示訊息,以告知使 用者替換損壞的處理器或是記憶體。 為了更清楚地描述上述方法實際運作的情形,以下再 舉另貝把例說明。本實施例假設致能針腳512原先設置 的致能訊號為1,❼若積測模组52〇㈣測處理器cpu〇 ,工作狀態後,骑其運作正t,财發送的伽彳訊號也 是1。因此,兩相比較之下,則可確定處理器CPU()的運 作狀態與預設相符,而無需將處理器cpu〇禁能;然而, 若偵測模組520在偵測處理器CPU〇的工作狀態後,判斷 其無法正常運作,因此發送的偵測訊號變為0。此即代表 18 1327711 EPD070011TW 23277twf.doc/n 處理器cpuos造成系統無法開機的原因,據此即將致能 訊號改為G,而將處理H CPUG帛能。此外,若依昭上^ 比對處理器CPU〇、_、…、CPUnq後發現, ,、運作白為正常。此即代表系統開機錯誤的問題是在於記 ,體,此時即可進行記憶體的切換動作。其中,若原先設 定的開機記紐為RQM0,則可城成由RQM1開機。由 上述可知,藉由本發明之開機切換方法,即可找出問題發 生的原因’並對應做出補救的動作,最終能使系統正常開 機。 最後,可參照圖7及圖8所繪示之本發明與習知技術 的比車乂表,此比較表是基於在一個採用雙處理器(CpU〇 及CPU1)及雙唯讀記憶體(R〇M〇及R〇M1)的系統上’ 分別進行本發明與習知技術所獲得的結果。其中,圖7中 習知技術的部分是假設R0M暫停時間大於FRB暫停時間 的情況;而圖8中習知技術的部分則是假設R〇M暫停時 間丨、於FRB暫停時間的情況。此外,圖式中打勾的部分代 表運作正常,打X的部分則代表運作異常或無法運作。比 較兩邊的結果可發現,採用習知的技術仍會造成許多判斷 失敗的情況,此即會造成系統永遠無法正常開機;相對地, 採用本案的結果均可正確判斷出錯誤發生的原因,並在發 現兩個處理器或兩個記憶體均為異常的狀況下,提示警示 訊號,以告知使用者替換,明顯地比習知技術更為可靠。 綜上所述,本發明之多處理器及多記憶體系統之開機 切換裝置及方法至少具有下列優點: 19 1327711 IPD070011TW 23277twf.doc/n 避免4二判斷系統是否發生開機錯誤,可 避免刀別使用兩料時器所造成之錯 夠導引系統選擇運作正當—理^^^^兄’並此 作。 R巾的處理錢雜體執行開機動 2/刀別透過勤〗針腳及致能針 夠在發生開機錯誤的忤.p〇n_ °。連接,而忐 哭,並將1林障況下❿發現發生錯誤的處理 I統用其他的處理㈣機,達到正常啟動 是由運彳’即可判斷錯誤 代的記in而利㈣賴的記憶體切換到另-個替 資料。“ 讓核在下次⑽時能夠讀収確的開機 限定㈣腦如上,然其並非用以 脫離本發明之精=屬,術領,有通常知識者,在不 因此本获明夕位▲和乾圍内,當可作些許之更動與潤飾, 為準。X呆護範圍當視後附之申請專利範圍所界定者 【圖式簡單說明】 的架^圖^ s7F4f知雙處理11祕使用删^進行㈣ 的方^流程S r為f知雙處理11系統使用BMC進行聽 圖4戶二二為習知雙BIOS ROM系統的架構圖。 Θ斤曰示為習知雙BIOS ROM系統的運作流程圖。 20 1327711 IPD070011TW 23277twf.doc/n 圖5是依照本發明一實施例所繪示的多處理器及多記 憶體系統的方塊圖。 圖6是依照本發明一實施例所繪示的多處理器及多記 - 憶體系統之開機切換方法的流程圖。 圖7所繪示為本發明與習知技術的比較表。 圖8所繪示為本發明與習知技術的比較表。 【主要元件符號說明】 ^ 100:雙處理器系統 110、120 :處理器 130 :基板管理控制器 131、132 :致能針腳 133 :錯誤回復開機計時器 140 :通用輸入輸出針腳 300 : BIOS ROM 系統 310 .輸入輸出控制晶片 311 :選擇針腳 • 312 :唯讀記憶體計時器 320、330 :唯讀記憶體 • 340 :反向器 350:通用輸入輸出針腳 500:多處理器及多記憶體系統 510 :基板管理控制器 511 :計時器 512、513、514 :致能針腳 21 1327711 IPD070011TW 23277twf.doc/n 515、516、517 :偵測針腳 518 :選擇針腳 520 :偵測模組 530 :多工器 540 :輸入輸出控制晶片 CPU0 ' CPU1.....CPUn-1 :處理器 ROMO、ROM1、...、R〇Mm-l :唯讀記憶體 S210〜S280 :習知雙處理器系統使用BMC進行 的方法之各步驟 S410〜S470 :習知雙BIOS ROM系統運作之各步驟 S610〜S690 :本發明—實施例之多處理器及多記 系統的開機切換方法之各步輝^ 22The switch is also useless. At this point, a warning message can be sent to inform the user to replace the damaged processor or memory. In order to more clearly describe the actual operation of the above method, the following is an example. In this embodiment, it is assumed that the enable signal of the enable pin 512 is set to 1, and if the test module 52 〇 (four) measures the processor cpu 〇, after the working state, the rider operates positively, and the gamma signal sent by the money is also 1 . Therefore, in the case of two-phase comparison, it can be determined that the operating state of the processor CPU() is consistent with the preset without the CPU cpu being disabled; however, if the detection module 520 is detecting the processor CPU After the working state, it is judged that it cannot operate normally, so the detection signal sent becomes 0. This means that 18 1327711 EPD070011TW 23277twf.doc/n The reason why the processor cpuos caused the system to fail to boot up is that the enable signal will be changed to G, and the H CPUG will be processed. In addition, if the processor CPU 〇, _, ..., CPUnq is found after the comparison, the operation white is normal. This means that the problem of the system boot error is that it is recorded, and the memory can be switched at this time. Among them, if the originally set power-on counter is RQM0, then Chengcheng can be powered on by RQM1. From the above, it can be seen that by the power-on switching method of the present invention, the cause of the problem can be found and the remedial action is performed, and the system can be normally started. Finally, referring to FIG. 7 and FIG. 8 , the comparison table of the present invention and the prior art is based on a dual processor (CpU〇 and CPU1) and dual read-only memory (R). The results obtained by the present invention and the conventional techniques are respectively performed on the systems of 〇M〇 and R〇M1). Among them, the part of the prior art in Fig. 7 assumes that the ROM pause time is greater than the FRB pause time; and the part of the prior art in Fig. 8 assumes that the R〇M pause time 丨 and the FRB pause time. In addition, some of the parts marked in the figure are working properly, and the part marked with X means that the operation is abnormal or inoperable. Comparing the results on both sides, it can be found that the use of conventional techniques will still cause many failures in judgment, which will cause the system to never start up normally; relatively, the results of this case can correctly determine the cause of the error, and When two processors or two memories are found to be abnormal, a warning signal is displayed to inform the user to replace, which is obviously more reliable than the prior art. In summary, the multi-processor and multi-memory system boot switching apparatus and method of the present invention have at least the following advantages: 19 1327711 IPD070011TW 23277twf.doc/n Avoiding the 4th judgment system whether a boot error occurs, and avoiding the use of the cutter The error caused by the two timing devices is enough to guide the system to choose to operate properly - rational ^^^^ brother' and do this. R towel handling money miscellaneous execution of the motor 2 / knife through the hard feet and enable the needle enough to occur in the boot error 忤.p〇n_ °. Connected, and crying, and I will find that the error occurred in a forest disorder, I use other processing (four) machine, to achieve normal start-up is to be able to judge the wrong generation of the memory of the inferior (four) Lai Switch to another one for the data. "Let the nuclear can read the correct start-up limit at the next (10) (4) brain as above, but it is not used to get rid of the essence of the invention, the genus, the general knowledge, the 408 and the dry Within the circumference, when a little change and retouching can be made, the scope of the X-protected area is defined by the scope of the patent application attached to the attached [simplified description of the figure] ^ s7F4f know the double processing 11 secret use delete ^ The process of performing (4) is the architecture diagram of the dual BIOS ROM system. The system is shown as the operation flow diagram of the conventional dual BIOS ROM system. 20 1327711 IPD070011TW 23277twf.doc/n Figure 5 is a block diagram of a multiprocessor and multi-memory system according to an embodiment of the invention. Figure 6 is a multiprocessor according to an embodiment of the invention. And a multi-recording-memory system switching method of the booting system. Figure 7 is a comparison table of the present invention and the prior art. Figure 8 is a comparison table of the present invention and the prior art. DESCRIPTION OF SYMBOLS ^ 100: Dual processor system 110, 120: Processor 130: Board Management Controller 131, 132: Enable Pin 133: Error Recovery Power On Timer 140: General Purpose Output Pin 300: BIOS ROM System 310. Input Output Control Chip 311: Select Pins • 312: Read Only Memory Timer 320, 330: Read-only memory • 340: Inverter 350: General-purpose input/output pin 500: Multiprocessor and multi-memory system 510: Baseboard management controller 511: Timers 512, 513, 514: Enable pin 21 1327711 IPD070011TW 23277twf.doc/n 515, 516, 517: detection pin 518: selection pin 520: detection module 530: multiplexer 540: input and output control chip CPU0 'CPU1.....CPUn-1: processor ROMO , ROM1, ..., R〇Mm-1: read-only memory S210 to S280: steps S410 to S470 of the conventional dual processor system using BMC: steps S610 of the conventional dual BIOS ROM system operation ~S690: Each step of the power-on switching method of the multiprocessor and multi-record system of the present invention-embodiment ^ 22

Claims (1)

IPD070011TW 23277twf.d〇c/i 十、申請專利範面: 1·種多處理器及多記憶體系統之開機切換褒置,配 置於包料個處理ϋ及多個記憶體之—系統,包括: 夕,處理ϋ致能針腳,分別_接至該些處理器 ,各該 =5致此針腳適於設置—致能訊號,以致能或禁能該 些處理器;IPD070011TW 23277twf.d〇c/i X. Application for patents: 1. Multi-processor and multi-memory system boot switching device, configured in the processing of multiple materials and multiple memory systems, including: In the evening, the processing pins are respectively connected to the processors, and each of the pins is adapted to set a enable signal to enable or disable the processors; “夕驗⑻貞崎腳,分別祕至該些處理11,各該 一 制針腳適於彳貞測對應之該處理器料作狀態, 而獲得一偵測訊號; 一^個記紐選擇針腳,缺至該些記㈣,適於設置 一达擇訊號,以切換該些記憶體;以及 一 6十時益,包括設定有—暫停時間(Time_〇ut),適 於在該系統開機時’開始倒數該暫停時間,其中 在該暫停時間倒數完畢時,判斷各該些處理器致能針"Xizheng (8) Sakizaki's foot, respectively, to the processing 11, each of the one-pin is suitable for detecting the state of the corresponding processor material, and obtaining a detection signal; a selection of stitches, Missing these records (4), suitable for setting a selection signal to switch the memory; and a 60-hour benefit, including setting the timeout (Time_〇ut), suitable for when the system is powered on' Starting to count down the pause time, wherein when the countdown is completed, it is determined that each of the processor enable pins ^該致能訊號及各該些處理以貞測針腳的該偵測訊 否相符, 若該致能訊號與該偵測訊號不符時 理器致能針腳將該處理器禁能, 了應之處 若所有致能訊號與偵測訊號均相符時,則透過 憶體選擇針腳切換該些記憶體。 二心 ^ = 圍第1項所述之多處理器及多記憶體 矛'既之開機切換裝置,更包括: 警示震置’適於在該些記憶體如換後仍無法正常 23 1327711 IPD070011TW 23277twf. doc/n 開機時,發出一警示訊息。 3. 如申請專利範圍第1項所述之多處理器及多記憶體 糸統之開機切換裝置’其中該些記憶體為儲存·基本輸入 輸出系統(Basic Input/Output System,BIOS)之唯讀記憶 體(Read Only Memory,ROM ) 〇 4. 如申請專利範圍第1項所述之多處理器及多記憶體 系統之開機切換裝置,其中該開機切換裝置包括配置於一^ The enable signal and each of the processes are consistent with the detection of the detected pin. If the enable signal does not match the detection signal, the processor enables the pin to disable the processor. If all the enable signals and the detection signals match, the memory is switched through the memory selection pins. Two hearts ^ = The multi-processor and multi-memory spears mentioned in the first item are both switched on and off, and include: the alarm is set to be suitable for the memory, such as after the change, still not normal 23 1327711 IPD070011TW 23277twf . doc/n Sends a warning message when you turn it on. 3. The multi-processor and multi-memory switching device as described in claim 1 wherein the memories are read only for the Basic Input/Output System (BIOS). (Read Only Memory, ROM) 〇4. The multi-processor and multi-memory system boot switching device of claim 1, wherein the power-on switching device comprises: 基板管理控制(Baseboard Management Control,BMC)裝 置中。 5.如申請專利範圍第1項所述之多處理器及多記憶體 系統之開機切換裝置,其中該計時器包括錯誤回復開機 (Fault Resilient Booting,FRB )計時器。 6·如申請專利範圍第1項所述之多處理器及多記憶體 系統之開機切換裝置,其中該些處理器致能針腳之該些致 能訊號包括均預先設置為致能狀態。In the Baseboard Management Control (BMC) device. 5. The power-on switching device of the multiprocessor and multi-memory system of claim 1, wherein the timer comprises a Fault Resilient Booting (FRB) timer. 6. The power-on switching device of the multi-processor and multi-memory system of claim 1, wherein the enabling signals of the processor-enabled pins are all preset to an enabled state. 7.如申請專鄉圍第丨項所述之多處理器及多記憶體 糸統之開機_裝置,其中該些記憶體選擇針 談 些選擇訊號包括預先設置該些記憶體宜 能而甘从u租,、T之—為致能狀 L ’而兵餘§己憶體則設置為禁能狀態。 8.如申請專利範圍第7項所述之多處理器及 糸統之^機切換裝置,其中當需要切換該些記憶體時Γ包 括將該'體其中之另-設置為致能狀綠,:你… 體則設置為禁能狀態。 〜、’而八餘冗憶 9_一種多處理器及多記憶體系統,包括: 24 1327711 IPD0700HTW 23277twf.doc/n 體系統之開機切換方法,其中在判斷 禁能訊號的步驟之後,更包括: 接收到該計時器 若接收到該計時器禁能訊號,則以目吁 理器及該些記憶體執行開機動作。 則叹定之該些處 22.如申請專利範圍第2〇項所述之多声^7. If you apply for the multi-processor and multi-memory system boot device as described in the section headed by the hometown section, the memory selection pins need to talk about some selection signals, including pre-setting the memory. u rent, T - for the enablement of L 'and the military § ** recall is set to disable state. 8. The multi-processor and switching device of claim 7, wherein when the memory needs to be switched, the setting of the body is set to enable green. : You... The body is set to the disabled state. ~, 'And eight more memory 9_ a multi-processor and multi-memory system, including: 24 1327711 IPD0700HTW 23277twf.doc / n system switching method, in the judgment of the disable signal step, including: Receiving the timer, if the timer disable signal is received, the booting operation is performed by the target processor and the memory. Then sigh somewhere. 22. As stated in the second paragraph of the patent application scope ^ 體糸統之開機切換方法,其中取消該暫作睡=及夕a己’("思 式包括在-基本輸人輸出系統正常執^^倒數的方 該基本輸人輸出系統取消該暫停時間的^動作時,透過 / 23.如申請專利範圍第20項所述之多處理 體系統之開機切換方法,其巾在將對應之該處理了 以及切換該些記憶體的步驟之後,更包括· 时不此, 重新啟動該系統,並以切換後的該些處理器及該些圮 憶體執行開機動作。 24. 如申請專利範圍第20項所述之多處理器及多記憶 體系統之開機切換方法,其中在該系統開機之前,更包括: 將所有該些致能訊號設置為致能狀態。The power-on switching method of the system, in which the temporary sleep = and eve a self is removed (" thinking is included in the basic input output system normal ^ ^ countdown party, the basic input output system cancels the pause time The operation of the multi-processor system according to claim 20, wherein the towel is processed and the steps of switching the memory are further included. If this is not the case, restart the system, and perform the booting operation with the switched processors and the memory. 24. Booting the multiprocessor and multi-memory system as described in claim 20 The switching method, wherein before the system is powered on, the method further comprises: setting all of the enabling signals to an enabled state. 25. 如申請專利範圍第20項所述之多處理器及多記憶 體系統之開機切換方法,其中在該系統開機之前,更包括: 選擇該些記憶體其中之一做為一開機記憶體,並將其 對應之該選擇訊號設置為致能狀態,而其餘記憶體之該些 選擇訊號則設置為禁能狀態。 26. 如申請專利範圍第25項所述之多處理器及多記憶 體系統之開機切換方法,其中切換該些記憶體的步驟包括: 將原本選擇之該記憶體的該選擇訊號設置為禁能狀 28 1327711 IPD070011TW 23277twf.doc/n 態;以及 選擇剩餘之該些記憶體其中之一做為該開機記憶 體,並將其對應之該選擇訊號設置為致能狀態。 27.如申請專利範圍第26項所述之多處理器及多記憶 體系統之開機切換方法,其中在該些記憶體均被選擇過, 而該系統仍無法正常開機時,包括發出一警示訊息。25. The power-on switching method of the multi-processor and multi-memory system according to claim 20, wherein before the system is powered on, the method further comprises: selecting one of the memories as a boot memory, And the corresponding selection signal is set to the enabled state, and the selection signals of the remaining memories are set to the disabled state. 26. The method of switching between the plurality of processors and the multi-memory system of claim 25, wherein the step of switching the memory comprises: setting the selected signal of the memory that is originally selected to be disabled 28 2827711 IPD070011TW 23277twf.doc/n state; and select one of the remaining memories as the boot memory, and set the corresponding selection signal to the enable state. 27. The method for switching the multiprocessor and multi-memory system according to claim 26, wherein the memory is selected, and the system still fails to boot normally, including issuing a warning message . 2929
TW96108929A 2007-03-15 2007-03-15 Boot-switching apparatus and method for multiprocessor and multi-memory system TWI327711B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW96108929A TWI327711B (en) 2007-03-15 2007-03-15 Boot-switching apparatus and method for multiprocessor and multi-memory system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW96108929A TWI327711B (en) 2007-03-15 2007-03-15 Boot-switching apparatus and method for multiprocessor and multi-memory system

Publications (2)

Publication Number Publication Date
TW200837632A TW200837632A (en) 2008-09-16
TWI327711B true TWI327711B (en) 2010-07-21

Family

ID=44820275

Family Applications (1)

Application Number Title Priority Date Filing Date
TW96108929A TWI327711B (en) 2007-03-15 2007-03-15 Boot-switching apparatus and method for multiprocessor and multi-memory system

Country Status (1)

Country Link
TW (1) TWI327711B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI489287B (en) * 2013-06-26 2015-06-21 Inventec Corp Multiplexer device and switching method therefor

Also Published As

Publication number Publication date
TW200837632A (en) 2008-09-16

Similar Documents

Publication Publication Date Title
US20190278677A1 (en) Runtime Software-Based Self-Test with Mutual Inter-Core Checking
US7783877B2 (en) Boot-switching apparatus and method for multiprocessor and multi-memory system
TW201237753A (en) Expedited computer boot system and method
US8392761B2 (en) Memory checkpointing using a co-located processor and service processor
EP2175371B1 (en) Synchronization control apparatuses, information processing apparatuses, and synchronization management methods
RU2405192C2 (en) Method and device for adjustment of multi-nuclear system
TW200944996A (en) Overclocking module and computer system and method thereof
US10379931B2 (en) Computer system
JP4393954B2 (en) Microcomputer
JP2003248598A (en) Micro-controller and method for detecting micro- controller trouble
CN116775141A (en) Abnormality detection method, abnormality detection device, computer device, and storage medium
TWI327711B (en) Boot-switching apparatus and method for multiprocessor and multi-memory system
CN100568189C (en) The power-on switching mechanism of multiprocessor and multi-memory system and method
CN113076222A (en) Ampere platform-based double-EEPROM system and double-EEPROM switching method
JP5788611B2 (en) Method and apparatus for saving state prior to reset for evaluation after reset
JP2004302731A (en) Information processor and method for trouble diagnosis
JPH0693229B2 (en) Data processing device
US7334167B2 (en) Circuit for detection of internal microprocessor watchdog device execution and method for resetting microprocessor system
JP2005190038A (en) Diagnostic processing method and diagnostic processing program for processor
JPH1153225A (en) Fault processor
TW200532433A (en) Device and method for automatically detecting and announcing error on booting a motherboard
JPWO2010032291A1 (en) Information processing apparatus and control method thereof
JP2011170691A (en) Cpu board with online test function of peripheral device part, and online test method thereof
JPH09212470A (en) Multiprocessor system
WO2016145774A1 (en) Electronic equipment start-up method and device

Legal Events

Date Code Title Description
MM4A Annulment or lapse of patent due to non-payment of fees