backport 4.5 fixes

example/main.c

@@ -23,6 +23,8 @@
 /*===========================================================================*
  * Defines:
  *===========================================================================*/
+//#define FAME_SOFT_FUP
+
 #define SB_BLS 1
 #define SB_ECDSA 2
 
@@ -34,13 +36,14 @@
 #define BLS_POINT_WSIZE (FP2_INT_WSIZE * 3)
 #define FAME_OW_SIZE (PH_CLUTILS_FAME_WORD_SIZE/2)
 
-#define FAME_WAIT(mask) do { dcw_dci(); } while(gphlibSfrUm_PKCC_STATUS.bVal & mask);
+#define FAME_WAIT(mask) while(gphlibSfrUm_PKCC_STATUS.bVal & mask) {}
+#define FAME_START(mask) dcw_dci(); gphlibSfrUm_PKCC_START.bVal = mask
 #define FAME_PTR(addr) (((uint16_t) addr) - SB_INFO_STRUCT_START)
 #define FAME_RPTR(addr) ((uint16_t*)(addr + SB_INFO_STRUCT_START))
 #define FAME_WALLOC(mem, size) &mem[size]; mem += size
 
-#define FUP_ENTRY(p0, p1, x, y, z, r) (p0|(p1 << 8)), (x|(y << 8)), (z|(r << 8))
-#define FMC_ENTRY(a, b) (a|(b << 8))
+#define FUP_ENTRY(p0, p1, x, y, z, r) (uint16_t)(p0|((uint16_t)p1 << 8)), (uint16_t)(x|(uint16_t)(y << 8)), (uint16_t)(z|(uint16_t)(r << 8))
+#define FMC_ENTRY(a, b) (uint16_t)(a|(uint16_t)(b << 8))
 #define FAME_USER_MACRO_ADDR 0x80
 #define FAME_USER_MACRO_WLEN 27
 #define FAME_MC_MODINVO FAME_USER_MACRO_ADDR
@@ -141,11 +144,25 @@ typedef struct LLVM_UNALIGNED {
   fp2_int_t z;
 } bls_point_t;
 
+typedef struct __attribute__((packed)) {
+  uint8_t op_type;
+  uint8_t op_code;
+  uint8_t x;
+  uint8_t y;
+  uint8_t z;
+  uint8_t r;
+} fup_entry_t;
+
 /*===========================================================================*
  * Globals:
  *===========================================================================*/
-uint16_t* FAME_MEMORY_TYPE fameMem = (uint16_t* FAME_MEMORY_TYPE) SB_WORKSPACE_START;
-DM32_MEMORY_TYPE uint16_t* fup = NULL;
+phNativeLibParameterArea_t parameterArea __attribute__((used, section(".section.infostruct")));
+uint16_t fameMem[0x3C0] __attribute__ ((section(".section.workspace")));
+uint8_t stack[0x120] __attribute__((used, section(".section.stack")));
+
+#ifndef FAME_SOFT_FUP
+volatile const uint16_t* fup;
+#endif
 
 const uint16_t SECP256K1_R2_NDASH_N[] = {
     // R^2
@@ -329,6 +346,9 @@ void fame_open() {
 
   gphlibSfrUm_PKCC_CTRL.bVal &= (uint8_t)~PHLIB_SFRUMPKCC_CTRL_RESET_MASK;
   while(gphlibSfrUm_PKCC_CTRL.bVal & PHLIB_SFRUMPKCC_CTRL_RESET_MASK);
+
+  gphlibSfrUm_PKCC_UPCTRL.bVal = PHLIB_SFRUMPKCC_UPCTRL_CACHE_EN_MASK;
+  gphlibSfrUm_PKCC_UPFIFO_CFG.bVal = (gphlibSfrUm_PKCC_VERSION0.wVal >> 8) & 0xf0;
 }
 
 void fame_close() {
@@ -339,6 +359,34 @@ void fame_close() {
   while((gphlibSfrUm_PKCC_CTRL.bVal & PHLIB_SFRUMPKCC_CTRL_STOP_MASK) != PHLIB_SFRUMPKCC_CTRL_STOP_MASK);
 }
 
+void fame_run_fup(const uint16_t* prg, uint8_t prg_len, uint16_t* FAME_MEMORY_TYPE uptable) {
+#ifndef FAME_SOFT_FUP
+  fup = prg;
+
+  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
+  gphlibSfrUm_PKCC_UPLEN.bVal = prg_len;
+  gphlibSfrUm_PKCC_UPTRT.wVal = FAME_PTR(uptable);
+
+  FAME_START(PHLIB_SFRUMPKCC_START_GOU_MASK);
+  gphlibSfrUm_PKCC_UPIRQ.bVal |= PHLIB_SFRUMPKCC_UPIRQ_LVL_MASK_VALUE_HIGH;
+
+  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
+#else
+  fup_entry_t* ops = (fup_entry_t*) prg;
+  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
+
+  for (int i = 0; i < prg_len; i++) {
+    gphlibSfrUm_PKCC_XPTR.wVal = uptable[ops[i].x];
+    gphlibSfrUm_PKCC_YPTR.wVal = uptable[ops[i].y];
+    gphlibSfrUm_PKCC_TOPP.wVal = uptable[ops[i].z];
+    gphlibSfrUm_PKCC_RPTR.wVal = uptable[ops[i].r];
+    gphlibSfrUm_PKCC_MODE.bVal = ops[i].op_code;
+    FAME_START(PHLIB_SFRUMPKCC_START_GOM1_MASK); // we always use macros in our FUP programs
+    FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
+  }
+#endif
+}
+
 uint8_t fame_cmp(uint16_t* FAME_MEMORY_TYPE a, uint16_t* FAME_MEMORY_TYPE b) {
   FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
 
@@ -346,7 +394,7 @@ uint8_t fame_cmp(uint16_t* FAME_MEMORY_TYPE a, uint16_t* FAME_MEMORY_TYPE b) {
   gphlibSfrUm_PKCC_TOPP.wVal = FAME_PTR(b);
   gphlibSfrUm_PKCC_MODE.bVal = FM_CMP;
 
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOD1_MASK;
+  FAME_START(PHLIB_SFRUMPKCC_START_GOD1_MASK);
   FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
 
   return gphlibSfrUm_PKCC_STATUS.bVal & 0x3;
@@ -360,28 +408,28 @@ void fame_rshift(uint16_t* FAME_MEMORY_TYPE a, uint16_t n) {
   gphlibSfrUm_PKCC_RPTR.wVal = FAME_PTR(a);
   gphlibSfrUm_PKCC_MODE.bVal = FM_SHR;
 
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOD1_MASK;
+  FAME_START(PHLIB_SFRUMPKCC_START_GOD1_MASK);
   FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
 }
 
 void fame_modInvEven(uint16_t* ab, uint16_t* uv, uint16_t* FAME_MEMORY_TYPE n) {
-  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
-
   n[-1] = 0x8000;
+
+  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
   gphlibSfrUm_PKCC_XPTR.wVal = FAME_PTR(n) - gphlibSfrUm_PKCC_OPLEN.wVal;
   gphlibSfrUm_PKCC_YPTR.wVal = FAME_PTR(n);
   gphlibSfrUm_PKCC_TOPP.wVal = FAME_PTR(ab);
   gphlibSfrUm_PKCC_RPTR.wVal = FAME_PTR(uv);
   gphlibSfrUm_PKCC_MODE.bVal = FAME_MC_MODINVE;
 
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOM1_MASK;
+  FAME_START(PHLIB_SFRUMPKCC_START_GOM1_MASK);
   FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
 }
 
 void fame_modInvOdd(uint16_t* FAME_MEMORY_TYPE ab, uint16_t* FAME_MEMORY_TYPE ba, uint16_t* FAME_MEMORY_TYPE uv, uint16_t* FAME_MEMORY_TYPE vu, uint16_t* FAME_MEMORY_TYPE n) {
-  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
-
   n[-1] = 0x8000;
+
+  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
   gphlibSfrUm_PKCC_XPTR.wVal = FAME_PTR(vu);
   gphlibSfrUm_PKCC_YPTR.wVal = FAME_PTR(uv);
   gphlibSfrUm_PKCC_TOPP.wVal = FAME_PTR(ab);
@@ -389,7 +437,7 @@ void fame_modInvOdd(uint16_t* FAME_MEMORY_TYPE ab, uint16_t* FAME_MEMORY_TYPE ba
   gphlibSfrUm_PKCC_MCLEN.wVal = FAME_PTR(n);
   gphlibSfrUm_PKCC_MODE.bVal = FAME_MC_MODINVO;
 
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOM1_MASK;
+  FAME_START(PHLIB_SFRUMPKCC_START_GOM1_MASK);
   FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
 }
 
@@ -432,7 +480,7 @@ void fame_monpro(uint16_t* FAME_MEMORY_TYPE a, uint16_t* FAME_MEMORY_TYPE b, uin
   gphlibSfrUm_PKCC_MCLEN.wVal = ECDSA_OP_SIZE;
   gphlibSfrUm_PKCC_MODE.bVal = FMC_MONPRO;
 
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOM1_MASK;
+  FAME_START(PHLIB_SFRUMPKCC_START_GOM1_MASK);
   FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
 }
 
@@ -447,19 +495,9 @@ static inline void fame_unblind(uint16_t* FAME_MEMORY_TYPE a, uint16_t* FAME_MEM
 
 void fame_ecdsaSign(uint16_t* FAME_MEMORY_TYPE uptable) {
   FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
-
   gphlibSfrUm_PKCC_MCLEN.wVal = ECDSA_OP_SIZE;
-  gphlibSfrUm_PKCC_UPCTRL.bVal = PHLIB_SFRUMPKCC_UPCTRL_CACHE_EN_MASK;
-  gphlibSfrUm_PKCC_UPLEN.bVal = ECDSA_SIGN_LEN;
-  gphlibSfrUm_PKCC_UPTRT.wVal = FAME_PTR(uptable);
 
-  fup = ECDSA_SIGN;
-
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOU_MASK;
-
-  gphlibSfrUm_PKCC_UPIRQ.bVal |= PHLIB_SFRUMPKCC_UPIRQ_LVL_MASK_VALUE_HIGH;
-
-  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
+  fame_run_fup(ECDSA_SIGN, ECDSA_SIGN_LEN, uptable);
 }
 
 void flipmpi(uint8_t* mpi, uint16_t size) {
@@ -473,21 +511,13 @@ void flipmpi(uint8_t* mpi, uint16_t size) {
 
 }
 
-void fame_fp2_do(uint16_t* FAME_MEMORY_TYPE uptable, fp2_int_t* a, fp2_int_t* b, fp2_int_t* r, const uint16_t* op, uint16_t oplen) {
-  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
-
+void fame_fp2_do(uint16_t* FAME_MEMORY_TYPE uptable, fp2_int_t* a, fp2_int_t* b, fp2_int_t* r, const uint16_t* op, uint8_t oplen) {
   uptable[0] = FAME_PTR(a->re);
   uptable[1] = FAME_PTR(a->im);
   uptable[2] = FAME_PTR(b->re);
   uptable[3] = FAME_PTR(b->im);
 
-  gphlibSfrUm_PKCC_UPLEN.bVal = oplen;
-  fup = op;
-
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOU_MASK;
-  gphlibSfrUm_PKCC_UPIRQ.bVal |= PHLIB_SFRUMPKCC_UPIRQ_LVL_MASK_VALUE_HIGH;
-
-  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
+  fame_run_fup(op, oplen, uptable);
 
   uint16_t* rre = FAME_RPTR(uptable[6]);
   uint16_t* rim = FAME_RPTR(uptable[7]);
@@ -525,19 +555,11 @@ void fame_fp2_mult_fp(uint16_t* FAME_MEMORY_TYPE uptable, fp2_int_t* a, const ui
 }
 
 void fame_fp2_invert(uint16_t* FAME_MEMORY_TYPE uptable, fp2_int_t* a, fp2_int_t* r) {
-  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_GOANY_MASK);
-
   uptable[0] = FAME_PTR(a->re);
   uptable[1] = FAME_PTR(a->im);
   uptable[2] = uptable[9];
 
-  gphlibSfrUm_PKCC_UPLEN.bVal = FP2_PRE_INV_LEN;
-  fup = FP2_PRE_INV;
-
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOU_MASK;
-  gphlibSfrUm_PKCC_UPIRQ.bVal |= PHLIB_SFRUMPKCC_UPIRQ_LVL_MASK_VALUE_HIGH;
-
-  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
+  fame_run_fup(FP2_PRE_INV, FP2_PRE_INV_LEN, uptable);
 
   uint16_t* factor = FAME_RPTR(uptable[9]);
   uint16_t* b = FAME_RPTR(uptable[6]);
@@ -560,13 +582,7 @@ void fame_fp2_invert(uint16_t* FAME_MEMORY_TYPE uptable, fp2_int_t* a, fp2_int_t
   }
 
   gphlibSfrUm_PKCC_MCLEN.wVal = FP2_OP_SIZE;
-  gphlibSfrUm_PKCC_UPLEN.bVal = FP2_POST_INV_LEN;
-  fup = FP2_POST_INV;
-
-  gphlibSfrUm_PKCC_START.bVal = PHLIB_SFRUMPKCC_START_GOU_MASK;
-  gphlibSfrUm_PKCC_UPIRQ.bVal |= PHLIB_SFRUMPKCC_UPIRQ_LVL_MASK_VALUE_HIGH;
-
-  FAME_WAIT(PHLIB_SFRUMPKCC_STATUS_ACTV_MASK);
+  fame_run_fup(FP2_POST_INV, FP2_POST_INV_LEN, uptable);
 
   for (uint8_t i = 0; i < FP2_OP_WSIZE; i++) {
     r->re[i] = b[i];
@@ -673,17 +689,15 @@ uint16_t ecdsaSign() {
   res[off++] = recIdInv;
 
 _ecdsa_finish:;
-  phNativeLibParameterArea_t* params = (phNativeLibParameterArea_t*)SB_INFO_STRUCT_START;
-
-  params->siCopyBackAddress = (puint8_uni_t) res;
-  params->siCopyBackSize = off;
-  params->scCopyBackAtomic = 0xff;
+  parameterArea.siCopyBackAddress = (puint8_uni_t) res;
+  parameterArea.siCopyBackSize = off;
+  parameterArea.scCopyBackAtomic = 0xff;
 
   return off;
 }
 
 uint8_t bls_point_is_zero(bls_point_t* FAME_MEMORY_TYPE p) {
-  uint8_t res = 0;
+  uint16_t res = 0;
 
   for (uint16_t i = 0; i < FP2_OP_WSIZE; i++) {
     res |= p->z.re[i];
@@ -892,8 +906,6 @@ uint16_t blsSign() {
 
   gphlibSfrUm_PKCC_OPLEN.wVal = FP2_OP_SIZE;
   gphlibSfrUm_PKCC_MCLEN.wVal = FP2_OP_SIZE;
-  gphlibSfrUm_PKCC_UPCTRL.bVal = PHLIB_SFRUMPKCC_UPCTRL_CACHE_EN_MASK;
-  gphlibSfrUm_PKCC_UPTRT.wVal = FAME_PTR(uptable);
 
   fame_load_macros();
 
@@ -908,39 +920,53 @@ uint16_t blsSign() {
 
   bls_toggle_montgomeryxy(uptable, s);
 
-  phNativeLibParameterArea_t* params = (phNativeLibParameterArea_t*)SB_INFO_STRUCT_START;
-
   flipmpi((uint8_t*) s->x.re, FP2_OP_SIZE);
   flipmpi((uint8_t*) s->x.im, FP2_OP_SIZE);
   flipmpi((uint8_t*) s->y.re, FP2_OP_SIZE);
   flipmpi((uint8_t*) s->y.im, FP2_OP_SIZE);
 
-  params->siCopyBackAddress = (puint8_uni_t) s;
-  params->siCopyBackSize = (FP2_OP_SIZE * 4);
-  params->scCopyBackAtomic = 0xff;
+  parameterArea.siCopyBackAddress = (puint8_uni_t) s;
+  parameterArea.siCopyBackSize = (FP2_OP_SIZE * 4);
+  parameterArea.scCopyBackAtomic = 0xff;
 
   return (FP2_OP_SIZE * 4);
 }
 
 /* Main entry to test native library -> USR_0 */
 LLVM_ISR_WITH_ARGS uint16_t mainEntry(uint16_t funcId) {
-  sil(0);
+#ifndef FAME_SOFT_FUP
+  sil(6);
+#endif
+
   OpenResources_SYS(NATIVELIB_OPEN_FAME_MASK|NATIVELIB_OPEN_RNG_MASK);
+  //SetClock_SYS(NATIVELIB_CLOCK_ID_PKCC, NATIVELIB_CLOCK_MODE_ENABLE);
+
   fame_open();
 
+  uint16_t res;
+
   switch (funcId) {
-    case SB_BLS: return blsSign();
-    case SB_ECDSA: return ecdsaSign();
+    case SB_BLS:
+      res = blsSign();
+      break;
+    case SB_ECDSA:
+      res = ecdsaSign();
+      break;
+    default:
+      res = 0xffff;
+      break;
   }
 
   fame_close();
+  //SetClock_SYS(NATIVELIB_CLOCK_ID_PKCC, NATIVELIB_CLOCK_MODE_DISABLE);
   CloseResources_SYS(NATIVELIB_OPEN_FAME_MASK|NATIVELIB_OPEN_RNG_MASK);
 
-  return 0x6d00;
+  return res;
 }
 
 /* FAME UPFIFO Empty ISR */
 void phIntHandler_FAME3_UPFIFO_Empty_Isr() LLVM_ISR {
+#ifndef FAME_SOFT_FUP
   while(!(gphlibSfrUm_PKCC_UPIRQ.bVal & PHLIB_SFRUMPKCC_UPIRQ_STOP_READING_MASK)) {
     gphlibSfrUm_PKCC_UPFIFO.wVal = *(fup++);
     gphlibSfrUm_PKCC_UPFIFO.wVal = *(fup++);
@@ -956,6 +982,7 @@ void phIntHandler_FAME3_UPFIFO_Empty_Isr() LLVM_ISR {
   }
 
   gphlibSfrUm_PKCC_UPIRQ.bVal = (gphlibSfrUm_PKCC_UPIRQ.bVal & ~mask) | PHLIB_SFRUMPKCC_UPIRQ_CLR_MASK;
+#endif
 }
 
 /* Timer ISR */
@@ -971,4 +998,3 @@ LLVM_ISR_WITH_ARGS void phappSysCall_Reset_POR() {
 LLVM_ISR_WITH_ARGS void phSecureBox_Exception_UsrCall(uint16_t liExceptionType, uint16_t liExceptionReason) {
   Exception_SYS(liExceptionType, liExceptionReason);
 } 
-

example/secureBoxNatLibMemSettings.ldi

@@ -9,11 +9,10 @@
  secureBoxWorkSpaceSize = 0x780;
  
  /* SecureBox Stack size (located in FRAM) */
- secureBoxStackSize = 0x100;
+ secureBoxStackSize = 0x120;
  
  /* SecureBox Transient State size (located in T-Heap)*/
- secureBoxTransienStateSize = 0x100;
+ secureBoxTransienStateSize = 0x20;
  
  /* SecureBox Persistent State size (located in P-Heap)*/
- secureBoxPersistentStateSize = 0x00;
- 
+ secureBoxPersistentStateSize = 0x00;

example/securebox_local.ldi

@@ -106,9 +106,8 @@ SECTIONS
  .nxp.jcop.securebox.bss.SECUREBOX_PRODUCT_NAME (NOLOAD) :
  {
   FILL(0);
-
-  *(SORT_BY_USAGE(.bss*)
-          SORT_BY_USAGE(COMMON))
+  KEEP (*(SORT_BY_NAME(.bss*) 
+  SORT_BY_USAGE(COMMON)))
   . = .;
 
   . = ALIGN(__MMU_RAM_GRANULARITY_BYTES);
@@ -123,17 +122,14 @@ SECTIONS
 
 
  .nxp.jcop.securebox.workspace.SECUREBOX_PRODUCT_NAME (ORIGIN(DATA) + (0x4000)) :
- {
-  FILL(0);
-
-  KEEP (*(.section.workspace))
-  . = .;
-
-  . = ALIGN(__MMU_RAM_GRANULARITY_BYTES);
-
- } > DATA
-
-
+{
+	FILL(0);
+	KEEP (*(.section.infostruct))
+	ASSERT(. == 0x30, "InfoStruct allocation does not have the correct length");
+	KEEP (*(.section.workspace))
+	. = .;
+	. = ALIGN(__MMU_RAM_GRANULARITY_BYTES);
+} > DATA
 
  .nxp.jcop.securebox.stack.SECUREBOX_PRODUCT_NAME :
  {