David Garske 4b74a2f006 Updated STM32L4 (Cortex-M at 80MHz) benchmarks with v5.6.4 using the new ARM ASM speedups. | 1 yıl önce | |
---|---|---|
.. | ||
README.md | 1 yıl önce | |
STM32_Benchmarks.md | 11 ay önce | |
default_conf.ftl | 1 yıl önce | |
include.am | 3 yıl önce | |
main.c | 1 yıl önce | |
wolfssl_example.c | 1 yıl önce | |
wolfssl_example.h | 1 yıl önce |
This example includes:
These examples use the Cube HAL for STM32.
You need both the STM32 IDE and the STM32 initialization code generator (STM32CubeMX) tools. The STM32CubeMX tool is used to setup a project which is used by the IDE to make any required code level changes and program / debug the STM32.
error: r7 cannot be used in 'asm
add -fomit-frame-pointer
to the CFLAGS. This only happens in debug builds, because r7 is used for debug.main.c
changes below in the STM32 Printf section.If none of the examples fit your STM32 type then you can create your own in STM32CubeMX by doing the following:
When you get to the IDE make sure you edit wolfSSL.I-CUBE-wolfSSL_conf.h
to set the HAL_CONSOLE_UART
to the correct one for your configuration.
The settings for the wolfSSL CubeMX pack are in the generated wolfSSL.I-CUBE-wolfSSL_conf.h
file. An example of a generated file can be found at examples/configs/user_settings_stm32.h
.
The template used for generation is IDE/STM32Cube/default_conf.ftl
, which is stored in the pack here: STM32Cube/Repository/Packs/wolfSSL/wolfSSL/[Version]/CubeMX/templates/default_conf.ftl
.
If the default settings for the Cube GUI are insufficient you can customize the build using one of these methods to prevent the changes from being overwritten when generating the code:
wolfSSL.I-CUBE-wolfSSL_conf.h
to Core/Inc
and rename to user_settings.h
. Then add the preprocessor macro WOLFSSL_USER_SETTINGS
to your project. This will use the user_settings.h
instead of the generated configuration.OR
STM32Cube/Repository/Packs/wolfSSL/wolfSSL/[Version]/CubeMX/templates/default_conf.ftl
.The section for "Hardware platform" may need to be adjusted depending on your processor and board:
WOLFSSL_STM32F1
.WOLFSSL_STM32F2
.WOLFSSL_STM32F4
.WOLFSSL_STM32F7
.WOLFSSL_STM32L4
.WOLFSSL_STM32L5
.WOLFSSL_STM32H7
.WOLFSSL_STM32WB
.WOLFSSL_STM32WL
.WOLFSSL_STM32U5
.WOLFSSL_STM32H5
.To use the STM32 Cube HAL support make sure WOLFSSL_STM32_CUBEMX
is defined.
The PKA acceleration for ECC is available on some U5, L5 and WB55 chips.
This is enabled with WOLFSSL_STM32_PKA
. You can see some of the benchmarks here.
To disable hardware crypto acceleration you can define:
NO_STM32_HASH
NO_STM32_CRYPTO
NO_STM32_RNG
To enable the latest Cube HAL support please define STM32_HAL_V2
.
If you'd like to use the older Standard Peripheral library undefine WOLFSSL_STM32_CUBEMX
.
With STM32 Cube HAL v2 some AES GCM hardware has a limitation for the AAD header, which must be a multiple of 4 bytes. If your HAL does not support CRYP_HEADERWIDTHUNIT_BYTE
then consider adding STM32_AESGCM_PARTIAL
if you are getting AES GCM authentication failures. This bug existed in v1.16.0 or later.
The STM32F7 v1.17.0 pack has a bug in the AES GCM code for handling of additional authentication data when not a multiple of 4 bytes. To patch see stm32f7xx_hal_cryp.c
-> CRYP_GCMCCM_SetHeaderPhase
:
diff --git a/stm32f7xx_hal_cryp.c b/stm32f7xx_hal_cryp.c
index 2ae42d0..9666f26 100644
--- a/stm32f7xx_hal_cryp.c
+++ b/stm32f7xx_hal_cryp.c
@@ -5600,7 +5600,6 @@ static HAL_StatusTypeDef CRYP_GCMCCM_SetHeaderPhase(CRYP_HandleTypeDef *hcryp, u
uint32_t size_in_bytes;
uint32_t tmp;
uint32_t mask[12] = {0x0U, 0xFF000000U, 0xFFFF0000U, 0xFFFFFF00U, /* 32-bit data type */
- 0x0U, 0x0000FF00U, 0x0000FFFFU, 0xFF00FFFFU, /* 16-bit data type */
0x0U, 0x000000FFU, 0x0000FFFFU, 0x00FFFFFFU}; /* 8-bit data type */
/***************************** Header phase for GCM/GMAC or CCM *********************************/
@@ -5842,7 +5841,7 @@ static HAL_StatusTypeDef CRYP_GCMCCM_SetHeaderPhase(CRYP_HandleTypeDef *hcryp, u
{
/* Enter last bytes, padded with zeroes */
tmp = *(uint32_t *)(hcryp->Init.Header + hcryp->CrypHeaderCount);
- tmp &= mask[(hcryp->Init.DataType * 2U) + (size_in_bytes % 4U)];
+ tmp &= mask[(hcryp->Init.HeaderWidthUnit * 4U) + (size_in_bytes % 4U)];
hcryp->Instance->DINR = tmp;
loopcounter++;
/* Pad the data with zeros to have a complete block */
If you are using FreeRTOS make sure your FreeRTOSConfig.h
has its configTOTAL_HEAP_SIZE
increased.
The TLS client/server benchmark example requires about 76 KB for allocated tasks (with stack) and peak heap. This uses both a TLS client and server to test a TLS connection locally for each enabled TLS cipher suite.
IDE/STM32Cube/wolfssl_example.c
Output....MENU
.t. WolfCrypt Test
.b. WolfCrypt Benchmark
.l. WolfSSL TLS Bench
.e. Show Cipher List
.s. Run TLS 1.3 Server over UART
.c. Run TLS 1.3 Client over UART
Please select one of the above options:
A tutorial for setting this up can be found here: https://www.youtube.com/watch?v=OK6MKXYiVBY
The TLS v1.3 client/server examples over UART are paired with these host-side applications:
To use this example you will need to use the STM32Cube interface to enable an additional USART and enable DMA for the RX with defaults. Enabling DMA for the USART requires adding the USART RX DMA in the STM32Cube tool. Under Connectivity click on your TLS USART# and goto DMA Settings and "Add" one for USART#_RX with default options.
On some boards, such as U5, there is GPDMA support. In this case when you click on "DMA Settings" you will be given a button to take you to GPDMA1 configuration. Click it. You can then enable a channel (any of the ones from 0 to 11 should be fine.) as "Standard Request Mode" and set the "Request Configuration" section's "Request" to USART#_RX. In the "System Core" tab, find NVIC and click on it. Make sure that the GPDMA1 global interrupt for your channel is enabled as well as USARTx global interrupt.
Then set the TLS_UART macro to the correct huart#
instance. This USART will be used as a TLS transport.
#define TLS_UART huart2
To disable the TLS UART example you can define NO_TLS_UART_TEST
.
See STM32_Benchmarks.md.
Note: The Benchmark example uses float. To enable go to "Project Properties" -> "C/C++ Build" -> "Settings" -> "Tool Settings" -> "MCU Settings" -> Check "Use float with printf".
In main.c make the following changes:
This section needs to go below the UART_HandleTypeDef
line, otherwise wolfssl/wolfcrypt/settings.h
will error.
/* Retargets the C library printf function to the USART. */
#include <stdio.h>
#include <wolfssl/wolfcrypt/settings.h>
#ifdef __GNUC__
int __io_putchar(int ch)
#else
int fputc(int ch, FILE *f)
#endif
{
HAL_UART_Transmit(&HAL_CONSOLE_UART, (uint8_t *)&ch, 1, 0xFFFF);
return ch;
}
#ifdef __GNUC__
int _write(int file,char *ptr, int len)
{
int DataIdx;
for (DataIdx= 0; DataIdx< len; DataIdx++) {
__io_putchar(*ptr++);
}
return len;
}
#endif
In the main()
function make the follow setvbuf()
additions after HAL_Init()
.
int main(void)
{
/* Reset of all peripherals, Initializes the Flash interface and the Systick. */
HAL_Init();
/* Turn off buffers, so I/O occurs immediately */
setvbuf(stdin, NULL, _IONBF, 0);
setvbuf(stdout, NULL, _IONBF, 0);
setvbuf(stderr, NULL, _IONBF, 0);
For questions please email support@wolfssl.com