Im Prinzip ja, aber bei deinem Code bestimmt nicht. Du hast 112 System-Takte pro IO-Takt. Ich bezweifle, dass jeder Pfad durch deinen ISR-Code innerhalb dieser Laufzeitgrenze bleibt, zumal du ja sogar 2 Interrupts pro IO-Takt hast.Zitat von Trabukh
Hallo zusammen,
ich versuche gerade, mittels eines ATmega8 eine SLE4442-Emulation zu entwerfen. Gleich vorweg: Nein, ich versuche nicht, irgendwelche Sonnenstudios, TV-Receiver, Waschautomaten oder sonstiges zu hacken.Die Emulation ist Teil einer freiwilligen Projektarbeit an der Uni (Thema "Sicherheit und Datenschutz").
Jedenfalls bin ich soweit eigentlich fertig, die Programmierung anhand des Datenblattes zur Karte war rein von der funktionalen Seite betrachtet kein großes Drama, allerdings befürchte ich, dass mein Code schlicht zu langsam ist. Mit dem USB-Kartenleser funktioniert meine "Karte" nämlich wunderbar, im Zielsystem (welches etwa 3.5 mal so schnell arbeitet) geht nichts mehr.
Ein mit 8 MHz getakteter ATmega8 sollte mit den 71.4 KHz Taktfrequenz des Systems meiner Meinung nach ohne jegliche Probleme zurechtkommen.Eventuell hab' ich auch einfach irgendwo etwas übersehen oder Mist gebaut, das will ich natürlich nicht ausschließen...
Der Code ist zugegebenermaßen nicht überufernd kommentiert, aber alle wichtigen Funktionen des Datenblattes sollten sofort ersichtlich sein.
Falls jemand eine Idee oder allgemeine Kritik zum Code hat, bin ich für jeden Beitrag dankbar.
Gruß und gute Nacht!
Trabukh
Code:// main.c // // cpu: ATMega8 // speed: 8 MHz #include "sle4442.h" #include "memory.h" void edgeFalling() { switch (mode) { case MODE_ATR: // Answer-to-Reset if (pointerByte <= 3) { setOutput(); setIO(memoryMain[pointerByte] & (1 << pointerBit)); } else { // ATR finished setInput(); setMode(MODE_IDLE); } break; case MODE_DATA: // Outgoing Data Mode switch (command[0]) { case 0x30: // Read Main Memory if (pointerByte <= 255) { setIO(memoryMain[pointerByte] & (1 << pointerBit)); setOutput(); } else { // Reading finished setInput(); setMode(MODE_IDLE); } break; case 0x34: // Read Protection Memory if (pointerByte <= 3) { setIO(memoryProtected[pointerByte * 8 + pointerBit]); setOutput(); } else { // Reading finished setInput(); setMode(MODE_IDLE); } break; case 0x31: // Read Security Memory if (pointerByte <= 3) { if ((unlocked == 3) | (pointerByte == 0x00)) setIO(memorySecurity[pointerByte] & (1 << pointerBit)); else setIO(0); setOutput(); } else { // Reading finished setInput(); setMode(MODE_IDLE); } break; } break; case MODE_PROC: // Processing Mode switch (command[0]) { case 0x39: // Update Security Memory setIO(0); setOutput(); if (unlocked == 3) { memorySecurity[command[1]] = command[2]; memorySecurity[0x00] &= 0x07; } else if ((command[1] == 0x00) & (command[2] < memorySecurity[0x00])) memorySecurity[0x00] = command[2]; waitCycles(124); setInput(); setMode(MODE_IDLE); break; case 0x33: // Compare Verification Data setIO(0); setOutput(); if ((memorySecurity[command[1]] == command[2]) & (memorySecurity[0x00] != 0x00)) { unlocked++; if (unlocked >= 3) { unlocked = 3; memorySecurity[0x00] = 0x07; } } else unlocked = 0; waitCycles(2); setInput(); setMode(MODE_IDLE); break; case 0x3C: // Write Protection Memory setIO(0); setOutput(); if ((unlocked == 3) & (memoryMain[command[1]] == command[2])) { memoryProtected[command[1]] = 1; } waitCycles(124); setInput(); setMode(MODE_IDLE); break; case 0x38: // Update Main Memory setIO(0); setOutput(); if (unlocked == 3) memoryMain[command[1]] = command[2]; waitCycles(124); setInput(); setMode(MODE_IDLE); break; } break; } } void edgeRising(bool bit) { // Increase pointers pointerBit++; if (pointerBit > 7) { pointerByte++; pointerBit = 0; } switch (mode) { case MODE_IDLE: // Idle Mode - Waiting for command loop_until_bit_is_clear(PIN,PIN_IO); pointerByte = 0; pointerBit = -1; setMode(MODE_CMD); break; case MODE_CMD: // Command Mode if (pointerByte <= 2) { if (bit) command[pointerByte] |= (1 << pointerBit); else command[pointerByte] &= ~(1 << pointerBit); } else { loop_until_bit_is_set(PIN,PIN_IO); pointerByte = 0; pointerBit = 0; switch (command[0]) { case 0x30: // Read Main Memory pointerByte = command[1]; case 0x34: // Read Protection Memory case 0x31: // Read Security Memory setMode(MODE_DATA); break; case 0x39: // Update Security Memory case 0x33: // Compare Verification Data case 0x3C: // Write Protection Memory case 0x38: // Update Main Memory setMode(MODE_PROC); break; } } break; } } // Clock interrupt ISR(INT0_vect) { if (!(PIN & (1 << PIN_CLK))) edgeFalling(); else edgeRising(PIN & (1 << PIN_IO)); } // Reset interrupt ISR(INT1_vect) { setMode(MODE_ATR); pointerByte = 0; pointerBit = 0; unlocked = 0; edgeFalling(); } int main(void) { // Activate interrupts MCUCR |= (1 << ISC00) | (1 << ISC11); GICR |= (1 << INT0) | (1 << INT1); sei(); // Initialize port PORT = 0x00; DDR = 0x00; PIN = 0x00; setMode(MODE_ATR); for (;;) { ; } }Code:// sle4442.h #include <avr/io.h> #include <avr/interrupt.h> #include <stdbool.h> // Global variables volatile unsigned char mode; volatile unsigned char unlocked = 0; // Pin/Port configuration #define PIN_CLK PD2 // INT0 #define PIN_RST PD3 // INT1 #define PIN_IO PD4 #define PORT PORTD #define PIN PIND #define DDR DDRD // Some helpful functions bool getReset() { return PIN & (1 << PIN_RST); } void setOutput() { DDR |= (1 << PIN_IO); } void setInput() { DDR &= ~(1 << PIN_IO); } void setIO(bool b) { if (b) PORT |= (1 << PIN_IO); else PORT &= ~(1 << PIN_IO); } void setMode(unsigned char m) { mode = m; } void waitCycles(unsigned char c) { for (unsigned char i = 0; i < c; i++) { loop_until_bit_is_set(PIN,PIN_CLK); loop_until_bit_is_clear(PIN,PIN_CLK); } } // Mode constants #define MODE_ATR 1 #define MODE_CMD 2 #define MODE_DATA 3 #define MODE_PROC 4 #define MODE_IDLE 5 // Memory pointers volatile unsigned int pointerByte = 0; volatile signed char pointerBit = 0; // Command bytes unsigned char command[3];Code:// memory.h unsigned char memoryMain[256] = { 0xA2, 0x13, 0x10, 0x91, 0x46, 0x0B, 0x81, 0x15, 0x42, 0x45, 0x00, 0x30, 0x03, 0x09, 0xED, 0xEA, 0x40, 0x12, 0xFF, 0xFF, 0xFF, 0xD2, 0x76, 0x00, 0x00, 0x63, 0x01, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x30, 0x30, 0x32, 0x37, 0x33, 0x38, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x00, 0x0A, 0x00, 0xBC, 0x64, 0x64, 0x00, 0x01, 0x00, 0x35, 0xF8, 0xD6, 0x19, 0x00, 0x00, 0x02, 0xFF, 0xFF, 0xFF, 0xFF, 0x11, 0x1A, 0x7C, 0x02, 0x05, 0x10, 0x08, 0x34, 0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x02, 0x66, 0x09, 0x01, 0x14, 0x10, 0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x10, 0xDA, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF }; bool memoryProtected[32] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1 }; unsigned char memorySecurity[] = { 0b00000111, 0xCD, 0x95, 0xE4 }; // Last 3 bytes = PSC
Im Prinzip ja, aber bei deinem Code bestimmt nicht. Du hast 112 System-Takte pro IO-Takt. Ich bezweifle, dass jeder Pfad durch deinen ISR-Code innerhalb dieser Laufzeitgrenze bleibt, zumal du ja sogar 2 Interrupts pro IO-Takt hast.Zitat von Trabukh
MfG
Stefan
Hmm, das ist natürlich ein Argument. Also nochmal neu aufziehen ohne Interrupts? Dafür müsste ich das meiste quasi neuschreiben, aber so lernt man wohl am besten...Du hast 112 System-Takte pro IO-Takt.![]()
Jedenfalls schonmal Danke, ich hätte nicht gedacht, dass soviel Zeit in meinem Code verloren geht.
EDIT: Hättest du eventuell auf den ersten Blick direkt Ideen, was man verbessern könnte? Ich bin leider kein C-Experte.
Nö, du musst aber den Interrupt deutlich verschlanken. Erster Schritt wäre z.B. im Interrupt keine Funktionen aufzurufen, sondern den kompletten Code direkt in die ISR zu packen (oder alternativ dafür sorgen, dass die Funktionen auf jeden Fall ge-inlined werden). Zweiter Schritt wäre Konstrukte der Art "(1 << pointerBit)" rauszuschmeißen. Die sind auf einem AVR sehr ineffektiv, weil der AVR keinen Assembler-Befehl hat für das Verschieben um mehrere Stellen, also wird daraus eine Schleife. Besser ist es direkt mit Masken zu arbeiten und immer nur um ein Bit zu verschieben.Zitat von Trabukh
(Das nur so auf die Schnelle. Da gibt es mit Sicherheit noch mehr "Verschlankungs-Potential")Code:Also statt pointerBit++; ... |= (1 << pointerBit); eher sowas pointerMask <<= 1; ... |= pointerMask;
MfG
Stefan
Selbst mit inline-Funktionen und Bitmasken ist das Programm noch zu langsam. Ich fürchte, mit einem Mega8 und den internen 8 MHz werd' ich wohl nicht an den benötigten Speed rankommen, zumal ich tatsächlich nur grob geschätzte 56 Takte für die Reaktion zur Verfügung habe... Wirklich schade![]()
So, Studium und Arbeit lassen im Augenblick wieder ein wenig Luft für Freizeit. Auch wenn das Uni-Projekt inzwischen abgeschlossen ist (als Ersatzprojekt habe ich eine AES-Implementierung mit der CUDA-Plattform für NVIDIA-Grafikkarten entwickelt), so würde ich doch gerne weiterhin versuchen, einen funktionierenden Code hinzubekommen.
Der allgemeine Stand ist jetzt mit Bitmasken und "ge-inlineten" Funktionen, aber auch damit erreiche ich keine ausreichende Reaktionszeit für die geforderten knapp 72 KHz... Eventuell hat jemand noch Anmerkungen, Kritik, Vorschläge etc.?
Gruß
Trabukh
Geändert von Trabukh (12.12.2011 um 19:08 Uhr) Grund: Typo
Lesezeichen