GCC-ARM Directory Reference

Directories
	Header
	Source

Detailed Description

This example is for Cortex M series, but can be adopted for other architectures.

Design

• All library options are defined in Header/user_settings.h.
• The memory map is located in the linker file in linker.ld.
• Entry point function is reset_handler in armtarget.c.
• The RTC and RNG hardware interface needs implemented for real production applications in armtarget.c

Building

1. Make sure you have gcc-arm-none-eabi installed.
2. Modify the Makefile.common:
   • Use correct toolchain path TOOLCHAIN.
   • Use correct architecture 'ARCHFLAGS'. See GCC ARM Options -mcpu=name.
   • Confirm memory map in linker.ld matches your flash/ram or comment out SRC_LD = -T./linker.ld in Makefile.common.
3. Use make to build the static library (libwolfssl.a), wolfCrypt test/benchmark and wolfSSL TLS client targets as .elf and .hex in /Build.

Building for Raspberry Pi

Example Makefile.common changes for Raspberry Pi with Cortex-A53:

1. In Makefile.common change ARCHFLAGS to -mcpu=cortex-a53 -mthumb.
2. Comment out SRC_LD, since custom memory map is not applicable.
3. Clear TOOLCHAIN, so it will use default gcc. Set TOOLCHAIN =
4. Comment out LDFLAGS += --specs=nano.specs and LDFLAGS += --specs=nosys.specs to nosys and nano.

Note: To comment out a line in a Makefile use place # in front of line.

Building for FIPS

1. Request evaluation from wolfSSL by emailing fips@.nosp@m.wolf.nosp@m.ss.co.nosp@m.m.
2. Modify user_settings.h so section for HAVE_FIPS is enabled.
3. Use make.
4. Run the wolfCrypt test ./Build/WolfCryptTest.elf to generate the FIPS boundary HASH

Example:

$ Crypt Test
error    test passed!
base64   test passed!
base16   test passed!
asn      test passed!
in my Fips callback, ok = 0, err = -203
message = In Core Integrity check FIPS error
hash = F607C7B983D1D283590448A56381DE460F1E83CB02584F4D77B7F2C583A8F5CD
In core integrity hash check failure, copy above hash
into verifyCore[] in fips_test.c and rebuild
SHA      test failed!
 error = -1802
Crypt Test: Return code -1

1. Update the ../../wolfcrypt/src/fips_test.c array static const char verifyCore[] = {} with the correct core hash check.
2. Build again using make.
3. Run the wolfCrypt test.

Building with configure

The configure script in the main project directory can perform a cross-compile build with the the gcc-arm-none-eabi tools. Assuming the tools are installed in your executable path:

$ ./configure \
    --host=arm-non-eabi \
    CC=arm-none-eabi-gcc \
    AR=arm-none-eabi-ar \
    STRIP=arm-none-eabi-strip \
    RANLIB=arm-none-eabi-ranlib \
    --prefix=/path/to/build/wolfssl-arm \
    CFLAGS="-march=armv8-a --specs=nosys.specs \
        -DHAVE_PK_CALLBACKS -DWOLFSSL_USER_IO -DNO_WRITEV" \
    --disable-filesystem --enable-fastmath \
    --disable-shared
$ make
$ make install

If you are building for a 32-bit architecture, add -DTIME_T_NOT_64BIT to the list of CFLAGS.

Example Build Output

make clean && make
 
   text    data     bss     dec     hex   filename
  50076    2508      44   52628    cd94   ./Build/WolfCryptTest.elf
 
   text    data     bss     dec     hex   filename
  39155    2508      60   41723    a2fb   ./Build/WolfCryptBench.elf
 
   text    data     bss     dec     hex filename
  70368     464      36   70868   114d4 ./Build/WolfSSLClient.elf

Performance Tuning Options

These settings are located in Header/user_settings.h.

• DEBUG_WOLFSSL: Undefine this to disable debug logging.
• NO_ERROR_STRINGS: Disables error strings to save code space.
• NO_INLINE: Disabling inline function saves about 1KB, but is slower.
• WOLFSSL_SMALL_STACK: Enables stack reduction techniques to allocate stack sections over 100 bytes from heap.
• USE_FAST_MATH: Uses stack based math, which is faster than the heap based math.
• ALT_ECC_SIZE: If using fast math and RSA/DH you can define this to reduce your ECC memory consumption.
• FP_MAX_BITS: Is the maximum math size (key size * 2). Used only with USE_FAST_MATH.
• ECC_TIMING_RESISTANT: Enables timing resistance for ECC and uses slightly less memory.
• ECC_SHAMIR: Doubles heap usage, but slightly faster
• RSA_LOW_MEM: Half as much memory but twice as slow. Uses Non-CRT method for private key.
• AES GCM: GCM_SMALL, GCM_WORD32 or GCM_TABLE: Tunes performance and flash/memory usage.
• CURVED25519_SMALL: Enables small versions of Ed/Curve (FE/GE math).
• USE_SLOW_SHA: Enables smaller/slower version of SHA.
• USE_SLOW_SHA256: About 2k smaller and about 25% slower
• USE_SLOW_SHA512: Over twice as small, but 50% slower
• USE_CERT_BUFFERS_1024 or USE_CERT_BUFFERS_2048: Size of RSA certs / keys to test with.
• BENCH_EMBEDDED: Define this if using the wolfCrypt test/benchmark and using a low memory target.
• ECC_USER_CURVES: Allows user to define curve sizes to enable. Default is 256-bit on. To enable others use HAVE_ECC192, HAVE_ECC224, etc....
• TFM_ARM, TFM_SSE2, TFM_AVR32, TFM_PPC32, TFM_MIPS, TFM_X86 or TFM_X86_64: These are assembly optimizations available with USE_FAST_MATH.
• Single Precision Math for ARM: See WOLFSSL_SP. Optimized math for ARM performance of specific RSA, DH and ECC algorithms.