200807273 九、發明說明： 【發明所屬之技術領域】 本發明是有關於一種系統呼叫之方法，且特別是有關 於使用一密文以限制核心之進入點之一種系統呼叫之方 法0 【先前技術】 使用者模式程式（user mode program)從使用者模式 (user mode )轉換到核心模式（kernel mode )，通常需要透 過硬體之協助，以取得完整的硬體控制權。作業系統可區 分為執行於一個虛擬機器上的作業系統或直接執行於硬體 上的作業系統。執行於虛擬機器上之作業系統，舉例來說 JavaOS，係完全依賴軟體的動態檢查以限制指標（p〇inter) 以及硬體資源（hardware resources)的使用方式。但其主 要之缺點在於執行效率不佳。 另，使用直接執行於硬體上之作業系統，其係大幅度 的依賴硬體所提供之保護。因此，某些特別的指令 (instruction )及動作只能由系統中某些元件所執行（通常 是 kernel) 〇 現今，大部分之中央處理系統（central processing unit) 以及作業系統（operating system )，採用以軟體中斷 (software interrupt)為基礎之系統呼叫（system call)。其係藉 由使用軟體的分配（dispatch)機制以呼叫正確之核心服務常 式（kernel service routine)。舉例來說，Linux 執行於 Intel 5 200807273 x86系列處理器上所使用的指令int 0x80系列將弓丨起軟體中 斷，隨後核心（kernel)將依照使用者所要求的服務並執行 一相對的核心服務常式。 弟1圖是Linux之系統呼叫之步 六货'顯示運作200807273 IX. Description of the Invention: [Technical Field] The present invention relates to a method for system call, and in particular to a method for using a ciphertext to limit a system call of a core entry point. [Prior Art] The user mode program is switched from user mode to kernel mode, usually with hardware assistance to obtain complete hardware control. The operating system can be divided into operating systems that are executed on a virtual machine or operating systems that are directly executed on hardware. Operating systems that execute on virtual machines, for example, JavaOS, rely entirely on dynamic checking of software to limit the use of metrics and hardware resources. But its main drawback is the poor implementation efficiency. In addition, the use of operating systems that are directly implemented on hardware is heavily dependent on the protection provided by the hardware. Therefore, some special instructions and actions can only be performed by certain components in the system (usually kernel). Today, most central processing units and operating systems use A system call based on software interrupt. It uses the software's dispatch mechanism to call the correct kernel service routine. For example, the instruction int 0x80 series that Linux uses on the Intel 5 200807273 x86 series processor will break the software interrupt, and then the kernel will follow the service requested by the user and execute a relative core service. formula. Brother 1 is the Linux system call step six goods 'display operation
於奔騰處理(Pentium)上之Linux之系統哞叫之步驟。^主 參照第1圖’完成Linux之系統呼叫包含五個步驟（步驟 至步驟118)。首先，於步驟11〇時，使用者模式=式 產生系統呼叫（舉例來說’ int 〇χ8〇) ^接著，於步驟^2 时，奔騰處理器將中斷描述表（interrupt descript〇r 中之第0x80號位址的内容載入到核心空間中之程式計數哭 (program Counter )。在載入此新的位址於程式計數器後: 作”統於步驟114將依照使用者的指定（通常“ _ 在堆疊中（stack)中的AX暫存器的值）的不同而乎叫相 對應之存在於核心空間中之核心服務常式。接著，於步驟 116中，從核心服務常式返回核心空間之暫存器。最後，於 步驟118中’從核心模式切換回使用者模式。 ^第！圖之以軟體令斷為基礎之系統呼叫，因為必 门 的軟體的刀配機制以呼叫正確的核心 服矛力韦式。因此對時間女台 核心騎的程式必須轉移到 核1中執仃（例如：以職的ke酬module)。及，因 為大部分之核心為單一入口（ 敏卞々/ ( try P〇1nt)。因此，需要_ 繁複糸統呼叫之步驟（步㈣〇至步驟118)。 效率因此’需要-簡化之系統呼叫之步驟，進而提高執行 6 200807273 【發明内容】 因此本發明的目的就是在提供—種系統啤叫之方法， 用以簡化系統呼叫之步驟。 因此本發㈣ϋ㈣是在提供—種加密於核心服 務常式以限制核心的進入點，進而簡化作業系統於處理系 統呼叫時之流程。 根據本發明之上述目的，提出-種快速系統呼叫之方 法°首先’❹-邏輯運算計算—核錢料式以產生_ !輯：算結果。然後’比較此邏輯運算之結果與金鑰暫存 器之在文。其中，此邏輯運算方法之輸入為至少一與此核 心服務之相關數據。例如，核心服務f式之起始位址或核 心，務常式之起始位址所存放的内容或其組合。如果邏輯 運算之結果相等於金鑰暫存器之密文，則允許從使用者模 式切換到核心模式以讀取核心服務常式。反之，中央處理 系統將會執行相對應之例外處理常式。且，於例外處理常 气中作業系統會巾止模式轉換之請求，並回報錯誤至作 業系統。 【實施方式】 請參照第2圖，係繪式本發明所提出之加密於一核心 服務常式之示意圖。2〇2係系統核心所加入之一密文，由於 二通的使用者程式（user mGde prQgram)無法得知產生此 贫文所需的金錄（key)(妥善保護此金鑰需於系統中），因 7 200807273 此唯有核心（kernel )才可產生此密文。 本發明能保證系統核心的強固性，舉例來說，惡意糕 式（malicious program )可以利用系統呼叫，將程式偽裝成 資料以進入核心中之資料缓衝器（data buffer)。如果，惡 意程式得知資料緩衝器之位址（假設資料缓衝器之位址為 0xCF00044)，惡意程式可以使用jmp 0xCF00044 (即，跳 躍至此資料缓衝器之位址）之指令以執行這段非法之程式 碼。本發明利用位於所有核心服務常式前的密文202，以過 濾限制核心之進入點（entry point)。 傳統上，核心須將所有之頁區分為資料頁（datapage) 以及碼頁（codepage)，以避免安全性上之漏洞。參考第2 圖，本發明使用加入密文之方法以限制核心之進入點，因 而增加安全之保護，容許核心服務常式200之資料區段 (data segment)以及碼段（code segment)直接存放於核 心 204之同一頁中以減少内部片段（internal fragmentation )，進而提高核心記憶體（kernel memory )之 使用率。及，由於碼段以及資料區段存在於相鄰之位址， 編譯器（compiler )可以大量使用程式計算器之相對位址 (PC relative addressing )，進而提高核心服務常式2〇〇之 效率。 請參照第3A圖，係繪式本發明所提出之快速系統呼叫 (fast system call)之流程圖。第3 A圖之步驟3〇〇所示， 使用者模式程式需要做模式切換，以取得完整的硬體控制 權。首先，邏輯運算欲使用之核心服務之相關數據。在一 8 200807273 範例裡，步驟302將系統呼叫之位址與存放於此位址之指 令做邏輯運算【例如，及（AND)、或（OR)、非（NOT)、 或與非（AND-NOT )、互斥或（exclusive OR，Xo〇、反互 斥或（NOTXor)運算或其它更複雜的加密運算如資料加密 演算法（Data Encryption Standard，DES )等】。此邏輯運算 方法之輸入為至少一與使用者所欲使用之核心服務之相關 數據。例如，此邏輯運算方法可使用核心服務常式之起始 位址或核心服務常式之起始位址所存放的内容或其組合。 然後，比較此邏輯運算之結果與金鑰暫存器之密文 (ciphertext)。金錄暫存器之密文係由核心隨機指定或於核 心編譯時所指定，指定此密文的方法及時機可以隨系統架 構之不同而有所不同（例如，密文係由一核心所產生，且 用以產生該密文之演算法及產生時機並無限制）。如果邏輯 運算之結果係等於金鑰暫存器之密文，則步驟304會允許 從使用者模式切換到核心模式以取得硬體控制權。反之， 於步驟306中，中央處理系統將會執行相對應之例外處理 常式（exception handler routine )。且，於例外處理常式中， 作業系統會中止模式轉換（mode change )之請求，並回報 錯誤至作業系統。 請參照第3B圖，其繪示依照本發明所提出之快速系統 呼叫之一較佳實施例之一示意圖。步驟310中，使用者模 式程式執行存放於位址0X00FE0000之指令，其内容為jmp 0xC00FF004 (即，跳躍到位址0xC00FF004 )。然，位址 0xC00FF004係位於核心空間。因此，中央處理器（CPU) 9 200807273 使用本發明所提出之快速系統呼叫之方法以處理此要求。 第3B圖之步驟312,邏輯運算欲使用之核心服務之相 關數據。在一範例裡，將存放於位址〇x〇〇FE〇〇〇4之指令内 容0XFF0788FF和位址0X00FE0004做邏輯運算【例如:及 (AND)、或（OR)、非（NOT)、或與非（AND_N〇T)、互 斥或（exclusive OR，Χ〇Γ)、反互斥或（ΝσΓχ〇〇運算或 其它更複雜的加密運算如資料加密演算法（DataEm^ypti^ Standard，DES)等】。此邏輯運算方法之輸入為至少一與使 用者所欲使用之核心服務之相關數據。例如，此邏輯運算 方法可使㈣錢務常狀起純址或核錢務常式之: 始位址所存放的内容或其組合。經由邏輯運算所產生之結 果再與金錄暫存器之密文做比較。金输暫存器之密文係^ 核心隨機指定或於核心編譯時所指^ m密文的方法 =時機可以隨系統架構之不同而有所不同。如果，邏輯運 异所產生之結果等於金鑰暫存器之密文，射央處理系統 可切換到核心模式（步驟314)。或，邏輯運算所產生之結 果不相等於金輸暫存器之密文（步驟316)，則中央處理系 統將會產生一例外（excepti〇n)，並開始執行例外處理器常 式。 口口 "月多舨第4圖，本發明所提出之快速系統呼叫方法之 步驟圖。因’邏輯運算欲使用之核心服務之相關數據，並 比對金繪暫存器中之密文以及此運算之結果，以限制核心 之入口點。如果，此運算所產生之結果等於金鑰暫存器之 密文，則中央處理线可切換_心、模式1而摒除安全 200807273 上之顧慮並直接執行相對應之核心服務常式（步驟4㈧）。 及，執行核心服務常式後，直接返回使用者模式（步驟 4〇2)。所以，其管理負擔（〇verhead)小於第i圖所述2奔 騰處理器上之Linux之系統呼叫。 由上述本發明之較佳實施例可知，應用本發明具 列優點： 1·本發明之快速系統呼叫，係採用加密之方法以控制 核心之進入點。進而簡化作業系統於處理系統呼叫 時之流程。 2·因採用加密之方法以控制核心之進入點。在不破壞 系統安全性之前提下，允許系統核心在同一頁中同 知存放碼（code)以及資料（data)。因此，内部片 段（internal fragmentation)變得較不嚴重，核心可 以變小。 雖然本發明已以一較佳實施例揭露如上，然其並非用 乂限义本發明，任何熟習此技藝者，在不脫離本發明之精 神和範圍内，當可作各種之更動與潤飾，因此本發明之保 濩範圍當視後附之申請專利範圍所界定者為準。 【圖式簡單說明】 為讓本發明之上述和其他目的、特徵、優點與實施例 月匕更明顯易懂，所附圖式之詳細說明如下： 第1圖是Linux之系統呼叫之一步驟圖； 200807273 密於 核 第2圖係繪示依照本發明一實施例之—種加 心服務常式之一示意圖； 第3A圖係繪示依照本發明另一實施例之快逮系纟一 叫之一流程圖； 、’、統呼 第3B圖係繪示依照本發明另一實施例之快速系統呼 叫之一示意圖；以及 第4圖係繪示依照本發明另一實施例之快速系統呼叫 之一步驟圖。 主要元件符號說明 110、112、114、116、118、300、302、304、306、310、 312、314、316、400、402 步驟 200 :核心服務常式 202 :密文The steps of the Linux system on the Pentium. ^Main Referring to Figure 1 'Complete the Linux system call, there are five steps (step to step 118). First, in step 11〇, the user mode = formula generates a system call (for example, ' int 〇χ 8〇) ^ Then, in step ^2, the Pentium processor will interrupt the description table (the first in the interrupt descript〇r) The content of address 0x80 is loaded into the core space and the program counter is frozen. After loading this new address in the program counter: "Continue in step 114 will be specified by the user (usually " _ The value of the AX register in the stack is different from the core service routine that exists in the core space. Then, in step 116, the core space is returned from the core service routine. Finally, in step 118, 'switch from core mode back to user mode. ^第! Figure is a software-based system call, because the software of the required software is called to call the correct core service. Spear force type. Therefore, the program for the time female platform core ride must be transferred to the core 1 (for example: the job ke reward module). And, because most of the core is a single entry (min/ / try P 〇1nt). Therefore, Need _ complicated 呼叫 call step (step (4) 〇 to step 118). Efficiency therefore 'needs-simplified system call steps, and then improve execution 6 200807273 [Inventive content] Therefore, the object of the present invention is to provide a system of beer The method is used to simplify the steps of the system call. Therefore, the present invention (4) (4) is to provide a process of encrypting the core service routine to limit the entry point of the core, thereby simplifying the flow of the operating system in handling system calls. For the above purpose, a method for fast system call is proposed. Firstly, the '❹-logical operation calculation—the nuclear money type is used to generate the _! series: the result. Then the result of comparing the logical operation with the key register is The input of the logical operation method is at least one data related to the core service, for example, the starting address or core of the core service f, the content of the starting address of the routine, or a combination thereof. If the result of the logical operation is equal to the ciphertext of the key register, then it is allowed to switch from the user mode to the core mode to read the core service routine. The central processing system will execute the corresponding exception handling routine. Moreover, in the exception processing, the operating system will reject the request for mode conversion and report the error to the operating system. [Embodiment] Please refer to Figure 2, A schematic diagram of the encryption of a core service routine proposed by the present invention. 2 〇 2 system core is added to a ciphertext, because the user program of the two-way user (user mGde prQgram) cannot know that the poor text is generated. The required key (the protection of this key is required in the system), because 7 200807273 This core (kernel) can only produce this ciphertext. The invention can guarantee the robustness of the system core, for example The malicious program can use the system call to disguise the program as data to enter the data buffer in the core. If the malware knows the address of the data buffer (assuming the address of the data buffer is 0xCF00044), the malware can use the instruction jmp 0xCF00044 (ie, jump to the address of this data buffer) to execute this section. Illegal code. The present invention utilizes ciphertext 202 located before all core service routines to filter the entry point of the core. Traditionally, the core has to distinguish all pages into data pages and codepages to avoid security holes. Referring to FIG. 2, the present invention uses the method of adding ciphertext to limit the entry point of the core, thereby increasing the security protection, allowing the data segment and the code segment of the core service routine 200 to be directly stored in The same page of the core 204 is used to reduce internal fragmentation, thereby increasing the usage of the core memory. And, since the code segment and the data segment exist in adjacent addresses, the compiler can use the PC relative addressing in a large amount, thereby improving the efficiency of the core service routine. Referring to Figure 3A, a flow chart of the fast system call proposed by the present invention is shown. As shown in step 3 of Figure 3A, the user mode program needs to do mode switching to obtain complete hardware control. First, the logic calculates the relevant data for the core service to be used. In an example of 200807273, step 302 logically operates the address of the system call with the instruction stored in the address [eg, AND, OR, NOT, OR (AND-) NOT ), mutual exclusion or (exclusive OR, Xo〇, anti-mutual or (NOTXor) operation or other more complex encryption operations such as Data Encryption Standard (DES), etc.] The input of this logical operation method is At least one data related to the core service that the user wants to use. For example, the logical operation method may use the content of the starting address of the core service routine or the starting address of the core service routine or a combination thereof. Then, compare the result of the logical operation with the ciphertext of the key register. The ciphertext of the golden record register is randomly specified by the core or specified at the time of core compilation, and the method for specifying the ciphertext is timely. It can vary from system to architecture (for example, ciphertext is generated by a core, and there is no limit to the algorithm used to generate the ciphertext and the timing of the generation). If the result of the logical operation is equal to the key The ciphertext of the device, step 304 will allow switching from the user mode to the core mode to obtain hardware control. Otherwise, in step 306, the central processing system will execute the corresponding exception handler routine (exception handler routine) Moreover, in the exception handling routine, the operating system will abort the request for mode change and report the error to the operating system. Please refer to FIG. 3B, which illustrates one of the fast system calls proposed in accordance with the present invention. A schematic diagram of a preferred embodiment. In step 310, the user mode program executes an instruction stored at address 0X00FE0000, the content of which is jmp 0xC00FF004 (ie, jumps to address 0xC00FF004). However, the address 0xC00FF004 is located in the core space. Central Processing Unit (CPU) 9 200807273 The method of the fast system call proposed by the present invention is used to process this requirement. Step 312 of Figure 3B, logically computing the relevant data of the core service to be used. In one example, it will be stored. Perform logical operations on the instruction contents 0XFF0788FF and address 0X00FE0004 of the address 〇x〇〇FE〇〇〇4 [eg: AND (AND), (OR), NOT (NOT), or NAND (AND_N〇T), mutually exclusive or (exclusive OR, Χ〇Γ), anti-mutex or (ΝσΓχ〇〇 operation or other more complex cryptographic operations such as data encryption calculus Method (DataEm^ypti^ Standard, DES), etc. The input of the logic operation method is at least one data related to the core service that the user wants to use. For example, the logic operation method can make the (4) money routines pure. Address or nuclear payment routine: The content of the original address or a combination thereof. The result of the logical operation is then compared with the ciphertext of the transcript register. The ciphertext system of the gold transfer register ^ The method of randomly specifying the core or the ciphertext indicated at the time of core compilation = the timing may vary depending on the system architecture. If the result of the logical assignment is equal to the ciphertext of the key register, the central processing system can switch to the core mode (step 314). Or, if the result of the logical operation is not equal to the ciphertext of the gold register (step 316), the central processing system will generate an exception (excepti〇n) and begin to execute the exception processor routine. Port "More than Figure 4, a step diagram of the fast system calling method proposed by the present invention. Because of the data related to the core services that the logical operation wants to use, and compare the ciphertext in the gold-plated register and the result of this operation to limit the entry point of the core. If the result of this operation is equal to the ciphertext of the key register, the central processing line can switch _ heart, mode 1 and remove the concerns on security 200807273 and directly execute the corresponding core service routine (step 4 (eight)) . And, after executing the core service routine, return directly to the user mode (step 4〇2). Therefore, its management burden (〇verhead) is smaller than the Linux system call on the 2 Pentium processors described in Figure i. It will be apparent from the above-described preferred embodiments of the present invention that the present invention has the advantages of: 1. The fast system call of the present invention employs an encryption method to control the entry point of the core. This simplifies the process of the operating system in handling system calls. 2. Use the encryption method to control the entry point of the core. Allowing the system core to share the code and data on the same page without prejudice to system security. Therefore, the internal fragmentation becomes less severe and the core can be made smaller. Although the present invention has been described above in terms of a preferred embodiment, it is not intended to limit the invention, and various modifications and changes may be made without departing from the spirit and scope of the invention. The scope of the present invention is defined by the scope of the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS In order to make the above and other objects, features, advantages and embodiments of the present invention more obvious and obvious, the detailed description of the drawings is as follows: FIG. 1 is a step diagram of a system call of Linux 200807273 密密核图 2 is a schematic diagram showing a kind of centering service routine according to an embodiment of the present invention; FIG. 3A is a diagram showing a quick catching system according to another embodiment of the present invention; FIG. 3B is a schematic diagram showing one of fast system calls according to another embodiment of the present invention; and FIG. 4 is a diagram showing one of fast system calls according to another embodiment of the present invention. Step chart. Main component symbol description 110, 112, 114, 116, 118, 300, 302, 304, 306, 310, 312, 314, 316, 400, 402 Step 200: Core service routine 202: ciphertext
204 ·•核心 12204 ·•Core 12