Added support for running test-app and benchmarks. Added example trace32 scripts.

Author: dgarske

config/examples/nxp-t2080.config

@@ -38,26 +38,26 @@ OPTIMIZATION_LEVEL?=1
 # NOR Base Address
 ARCH_FLASH_OFFSET?=0xEFFE0000
 
-# Flash Sector Size (must match physical NOR erase block: 128 KB)
-WOLFBOOT_SECTOR_SIZE?=0x20000
+# Flash Sector Size
+WOLFBOOT_SECTOR_SIZE?=0x10000
 
 # wolfBoot start address
 WOLFBOOT_ORIGIN?=0xEFFE0000
 # wolfBoot partition size (custom)
 BOOTLOADER_PARTITION_SIZE=0x20000
 
 # Application Partition Size
-WOLFBOOT_PARTITION_SIZE?=0x20000
+WOLFBOOT_PARTITION_SIZE?=0x100000
 # Location in Flash for Application Partition
-WOLFBOOT_PARTITION_BOOT_ADDRESS?=0xEFFC0000
+WOLFBOOT_PARTITION_BOOT_ADDRESS?=0xEFEE0000
 # Load Partition to RAM Address
 WOLFBOOT_LOAD_ADDRESS?=0x19000
 
 # Location in Flash for Update Partition
-WOLFBOOT_PARTITION_UPDATE_ADDRESS?=0xEFFA0000
+WOLFBOOT_PARTITION_UPDATE_ADDRESS?=0xEFDE0000
 
 # Location of temporary sector used during updates
-WOLFBOOT_PARTITION_SWAP_ADDRESS?=0xEFF90000
+WOLFBOOT_PARTITION_SWAP_ADDRESS?=0xEFDD0000
 
 # DTS (Device Tree)
 WOLFBOOT_DTS_BOOT_ADDRESS?=0xE8040000
@@ -67,3 +67,7 @@ WOLFBOOT_LOAD_DTS_ADDRESS?=0x200000
 
 # Flash erase/write/read test at update partition address
 #TEST_FLASH?=1
+
+# wolfCrypt Test and Benchmark (requires larger partition size)
+#WOLFCRYPT_TEST?=1
+#WOLFCRYPT_BENCHMARK?=1

docs/Targets.md

@@ -3315,10 +3315,13 @@ placed in `.ramcode`, copied to DDR, and remapped via TLB9. Key rules:
 
 - **No calls to flash-resident code.** The linker generates trampolines that
   jump back to flash addresses. Any helper called from RAMFUNCTION code must
-  itself be RAMFUNCTION or fully inlined. Delay ticks are pre-computed in
-  `hal_flash_init()` to avoid calling `hal_get_plat_clk()` from flash `.text`.
+  itself be RAMFUNCTION or fully inlined. Delay/clock helpers (for example,
+  `udelay` and associated clock accessors) are provided by `nxp_ppc.c` and
+  are marked `RAMFUNCTION` so they can be safely invoked without executing
+  from flash `.text`.
 - **Inline TLB/cache ops.** `hal_flash_cache_disable/enable` use
-  `ram_write_tlb()` (inline mtspr) and direct L1CSR0/L1CSR1 manipulation.
+  `set_tlb()` / `write_tlb()` (inline `mtspr` helpers) and direct
+  L1CSR0/L1CSR1 manipulation.
 - **WBP timing.** The write-buffer-program sequence (unlock → 0x25 → count →
   data → 0x29) must execute without bus-stalling delays. UART output between
   steps (~87us per character at 115200) triggers DQ1 abort.

hal/nxp_t2080.c

@@ -400,6 +400,16 @@ void hal_init(void)
 #ifndef WOLFBOOT_REPRODUCIBLE_BUILD
     wolfBoot_printf("Build: %s %s\n", __DATE__, __TIME__);
 #endif
+    wolfBoot_printf("System Clock: %lu MHz\n",
+        (unsigned long)(SYS_CLK / 1000000));
+    wolfBoot_printf("Platform Clock: %lu MHz\n",
+        (unsigned long)(hal_get_plat_clk() / 1000000));
+    wolfBoot_printf("Core Clock: %lu MHz\n",
+        (unsigned long)(hal_get_core_clk() / 1000000));
+    wolfBoot_printf("Bus Clock: %lu MHz\n",
+        (unsigned long)(hal_get_bus_clk() / 1000000));
+    wolfBoot_printf("Timebase: %lu MHz\n",
+        (unsigned long)(TIMEBASE_HZ / 1000000));
 #endif
 
     hal_flash_init();

include/user_settings.h

@@ -478,11 +478,11 @@ extern int tolower(int c);
 
 #if !defined(WOLFCRYPT_SECURE_MODE) && !defined(WOLFBOOT_TPM_PARMENC) && \
     !defined(WOLFCRYPT_TEST) && !defined(WOLFCRYPT_BENCHMARK)
-#if !(defined(WOLFBOOT_ENABLE_WOLFHSM_CLIENT) && \
-      defined(WOLFBOOT_SIGN_ML_DSA)) && \
-    !defined(WOLFBOOT_ENABLE_WOLFHSM_SERVER)
-#define WC_NO_RNG
-#endif
+    #if !(defined(WOLFBOOT_ENABLE_WOLFHSM_CLIENT) && \
+        defined(WOLFBOOT_SIGN_ML_DSA)) && \
+        !defined(WOLFBOOT_ENABLE_WOLFHSM_SERVER)
+        #define WC_NO_RNG
+    #endif
     #define WC_NO_HASHDRBG
     #define NO_AES_CBC
 #else
@@ -494,6 +494,9 @@ extern int tolower(int c);
         #define CUSTOM_RAND_GENERATE_SEED my_rng_seed_gen
         #define CUSTOM_RAND_GENERATE_BLOCK my_rng_seed_gen
         extern int my_rng_seed_gen(unsigned char* output, unsigned int sz);
+        
+        #define HAVE_AESGCM
+        #define GCM_TABLE
     #else
         #define HAVE_HASHDRBG
         #define WOLFSSL_AES_CFB
@@ -591,6 +594,9 @@ extern int tolower(int c);
 #if defined(WOLFCRYPT_TEST) || defined(WOLFCRYPT_BENCHMARK)
     #define NO_WRITE_TEMP_FILES
 
+    /* Use printf for wolfSSL logging (redirected to UART via syscalls.c) */
+    #define WOLFSSL_LOG_PRINTF
+
     /* Use static memory pool to avoid system malloc dependency.
      * benchmark.c provides gBenchMemory static buffer.
      * Default is 50KB with BENCH_EMBEDDED, override for smaller targets */

src/string.c

@@ -346,7 +346,7 @@ void uart_writenum(int num, int base, int zeropad, int maxdigits)
 void uart_vprintf(const char* fmt, va_list argp)
 {
     char* fmtp = (char*)fmt;
-    int zeropad, maxdigits;
+    int zeropad, maxdigits, precision, leftjust;
     while (fmtp != NULL && *fmtp != '\0') {
         /* print non formatting characters */
         if (*fmtp != '%') {
@@ -356,17 +356,42 @@ void uart_vprintf(const char* fmt, va_list argp)
         fmtp++; /* skip % */
 
         /* find formatters */
-        zeropad = maxdigits = 0;
+        zeropad = maxdigits = leftjust = 0;
+        precision = -1; /* -1 = not specified */
+        /* check for left-justify flag */
+        if (*fmtp == '-') {
+            leftjust = 1;
+            fmtp++;
+        }
         while (*fmtp != '\0') {
-            if (*fmtp >= '0' && *fmtp <= '9') {
+            if (*fmtp == '*') {
+                /* width from argument */
+                maxdigits = va_arg(argp, int);
+                fmtp++;
+            }
+            else if (*fmtp >= '0' && *fmtp <= '9') {
                 /* length formatter */
-                if (*fmtp == '0') {
+                if (*fmtp == '0' && maxdigits == 0) {
                     zeropad = 1;
                 }
                 maxdigits <<= 8;
                 maxdigits += (*fmtp - '0');
                 fmtp++;
             }
+            else if (*fmtp == '.') {
+                /* precision */
+                fmtp++;
+                if (*fmtp == '*') {
+                    precision = va_arg(argp, int);
+                    fmtp++;
+                } else {
+                    precision = 0;
+                    while (*fmtp >= '0' && *fmtp <= '9') {
+                        precision = precision * 10 + (*fmtp - '0');
+                        fmtp++;
+                    }
+                }
+            }
             else if (*fmtp == 'l') {
                 /* long - skip */
                 fmtp++;
@@ -405,7 +430,20 @@ void uart_vprintf(const char* fmt, va_list argp)
             case 's':
             {
                 char* str = (char*)va_arg(argp, char*);
-                uart_write(str, (uint32_t)strlen(str));
+                int slen = (int)strlen(str);
+                if (leftjust) {
+                    uart_write(str, slen);
+                    while (slen < maxdigits) {
+                        uart_write(" ", 1);
+                        slen++;
+                    }
+                } else {
+                    while (slen < maxdigits) {
+                        uart_write(" ", 1);
+                        slen++;
+                    }
+                    uart_write(str, (uint32_t)strlen(str));
+                }
                 break;
             }
             case 'c':
@@ -414,6 +452,44 @@ void uart_vprintf(const char* fmt, va_list argp)
                 uart_write(&c, 1);
                 break;
             }
+#ifdef UART_PRINTF_FLOAT
+            case 'f':
+            case 'e':
+            case 'g':
+            {
+                double val = va_arg(argp, double);
+                int prec = (precision >= 0) ? precision : 3;
+                int digit;
+                unsigned int ipart;
+
+                /* handle negative */
+                if (val < 0.0) {
+                    uart_write("-", 1);
+                    val = -val;
+                }
+
+                /* integer part */
+                ipart = (unsigned int)val;
+                uart_writenum((int)ipart, 10, 0, 0);
+
+                /* fractional part */
+                if (prec > 0) {
+                    double frac = val - (double)ipart;
+                    char c;
+                    int i;
+                    uart_write(".", 1);
+                    for (i = 0; i < prec; i++) {
+                        frac *= 10.0;
+                        digit = (int)frac;
+                        if (digit > 9) digit = 9;
+                        c = '0' + digit;
+                        uart_write(&c, 1);
+                        frac -= (double)digit;
+                    }
+                }
+                break;
+            }
+#endif /* UART_PRINTF_FLOAT */
             default:
                 break;
         }

test-app/Makefile

@@ -107,9 +107,14 @@ endif
 ifeq ($(WOLFCRYPT_BENCHMARK),1)
   CFLAGS+=-DWOLFCRYPT_BENCHMARK
   WOLFCRYPT_SUPPORT=1
+  UART_PRINTF_FLOAT?=1
   APP_OBJS+=$(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/benchmark/benchmark.o
 endif
 
+ifeq ($(UART_PRINTF_FLOAT),1)
+  CFLAGS+=-DUART_PRINTF_FLOAT
+endif
+
 ifeq ($(WOLFCRYPT_SUPPORT),1)
   # Add support infrastructure
   APP_OBJS+=wolfcrypt_support.o
@@ -141,25 +146,60 @@ ifeq ($(WOLFCRYPT_SUPPORT),1)
   APP_OBJS+=$(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/ecc.o
   APP_OBJS+=$(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/sp_int.o
 
-  # Add SP math implementations for ARM Cortex-M
-  APP_OBJS+=$(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/sp_cortexm.o
+  # Add SP C math (all architectures)
   APP_OBJS+=$(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/sp_c32.o
 
-  ifneq ($(NO_ARM_ASM),1)
-    APP_OBJS+= \
-          $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-aes-asm_c.o \
-          $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-sha256-asm_c.o \
-          $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-sha512-asm_c.o \
-          $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-sha3-asm_c.o \
-          $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-chacha-asm_c.o
+  ifneq ($(NO_ASM),1)
+    # SP Cortex M
+    ifneq ($(filter ARM ARM_BE,$(ARCH)),)
+      ifeq ($(SPMATH),1)
+        APP_OBJS+=$(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/sp_cortexm.o
+      endif
+
+      ifneq ($(NO_ARM_ASM),1)
+        APP_OBJS+= \
+              $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-aes-asm_c.o \
+              $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-sha256-asm_c.o \
+              $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-sha512-asm_c.o \
+              $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-sha3-asm_c.o \
+              $(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/arm/thumb2-chacha-asm_c.o
+
+        CFLAGS+=-DWOLFSSL_ARMASM -DWOLFSSL_ARMASM_NO_HW_CRYPTO -DWOLFSSL_ARMASM_INLINE \
+                -DWOLFSSL_ARMASM_NO_NEON -DWOLFSSL_ARMASM_THUMB2
+      endif
+    endif
+
+    # Power PC
+    ifeq ($(ARCH),PPC)
+      # SP mp int PowerPC ASM
+      ifeq ($(PPC64),1)
+        CFLAGS+=-DWOLFSSL_SP_PPC64
+      else
+        CFLAGS+=-DWOLFSSL_SP_PPC
+      endif
+
+      # SHA256
+      APP_OBJS+=$(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/ppc32/ppc32-sha256-asm_c.o
+      CFLAGS+=-DWOLFSSL_PPC32_ASM
+      CFLAGS+=-DWOLFSSL_PPC32_ASM_INLINE
+      CFLAGS+=-DWOLFSSL_PPC32_ASM_SMALL
+
+      ifeq ($(PPC64),1) # requires wolfssl PR 9852
+        # AES
+        APP_OBJS+=$(WOLFBOOT_LIB_WOLFSSL)/wolfcrypt/src/port/ppc64/ppc64-aes-asm_c.o
+        CFLAGS+=-DWOLFSSL_PPC64_ASM 
+        CFLAGS+=-DWOLFSSL_PPC64_ASM_INLINE
+        CFLAGS+=-DWOLFSSL_PPC64_ASM_SMALL
+        CFLAGS+=-DWOLFSSL_PPC64_ASM_AES_NO_HARDEN
+      endif
 
-    CFLAGS+=-DWOLFSSL_ARMASM -DWOLFSSL_ARMASM_NO_HW_CRYPTO -DWOLFSSL_ARMASM_INLINE \
-                              -DWOLFSSL_ARMASM_NO_NEON -DWOLFSSL_ARMASM_THUMB2
+      CFLAGS+=-fomit-frame-pointer
+    endif
   endif
 
   CFLAGS+=-DWOLFSSL_USER_SETTINGS
   CFLAGS+=-I"$(WOLFBOOT_LIB_WOLFSSL)"
-endif
+endif # WOLFCRYPT_SUPPORT
 
 ifeq ($(TZEN),1)
     CFLAGS+=-DNONSECURE_APP
@@ -414,6 +454,22 @@ ifeq ($(TARGET),va416x0)
   CFLAGS+=-ffunction-sections -fdata-sections
 endif
 
+PPC64=0
+ifeq ($(TARGET),nxp_t2080)
+  ifneq ($(SIGN),NONE)
+    APP_OBJS+=../src/keystore.o
+  endif
+  CFLAGS+=-ffunction-sections -fdata-sections
+
+  # PowerPC e6500
+  PPC64=1
+endif
+
+ifeq ($(TARGET),nxp_t1024)
+  # PowerPC e5500
+  PPC64=1
+endif
+
 ifeq ($(TARGET),sim)
   # LD on MacOS does not support "-Map="
   LDMAPSUPPORTED=$(shell $(CC) -Wl,-Map=image.map 2>&1 | grep 'unknown option')

test-app/PPC.ld

@@ -1,6 +1,6 @@
 OUTPUT_ARCH( "powerpc" )
 
-ENTRY( main )
+ENTRY( _app_entry )
 
 PHDRS
 {
@@ -14,6 +14,7 @@ SECTIONS
 
     .text :
     {
+       *(.text._app_entry)
        *(.text.main)
         *(.text*)
         *(.rodata*)
@@ -23,6 +24,7 @@ SECTIONS
     {
         _start_data = .;
         KEEP(*(.data*))
+        *(.sdata*)       /* PPC small initialized data */
         . = ALIGN(4);
         KEEP(*(.ramcode))
 
@@ -54,6 +56,7 @@ SECTIONS
         _start_bss = .;
         __bss_start__ = .;
         *(.bss*)
+        *(.sbss*)        /* PPC small uninitialized data */
         *(COMMON)
         . = ALIGN(4);
         _end_bss = .;
@@ -62,4 +65,12 @@ SECTIONS
     }
     . = ALIGN(4);
 
+    /* Heap for _sbrk (used by syscalls.c malloc) */
+    _Min_Heap_Size = 0x10000; /* 64KB */
+
+    /* Stack: 64KB, grows downward from _stack_top.
+     * wolfCrypt ECC384 operations need deep stack (~11KB per frame). */
+    _Min_Stack_Size = 0x10000; /* 64KB */
+    _stack_end = _end + _Min_Heap_Size;
+    _stack_top = _stack_end + _Min_Stack_Size;
 }