diff options
| author | Karl Palsson <karlp@remake.is> | 2015-10-15 15:28:54 +0300 |
|---|---|---|
| committer | Karl Palsson <karlp@remake.is> | 2015-10-15 15:28:54 +0300 |
| commit | caeeed2b0f3e231a8667deb57f0d75f5f63d0ec3 (patch) | |
| tree | bd66ee013767c9500f23044198bc1d7f70460475 | |
| parent | 5ece67ad5dbe7ad9b5c378d41d77e2337caffe07 (diff) | |
stm32f7: FPU only single precision on f7 :(
Boo ST.
| -rw-r--r-- | README.md | 5 | ||||
| -rw-r--r-- | lib/stm32/f7/Makefile | 4 |
2 files changed, 5 insertions, 4 deletions
@@ -99,8 +99,9 @@ them as environment variables, for example: behavior can be specified by setting `FP_FLAGS`. Currently, M4F cores default to `-mfloat-abi=hard -mfpu=fpv4-sp-d16`, - M7 core defaults to double precision `-mfloat-abi=hard -mfpu=fpv5-d16` - and other architectures to no FP flags forcing to use software computation. + M7 cores defaults to double precision `-mfloat-abi=hard -mfpu=fpv5-d16` if available, + and single precision `-mfloat-abi=hard -mfpu=fpv5-sp-d16` otherwise. + Other architectures use no FP flags, in otherwords, traditional softfp. You may find which FP_FLAGS you can use in particular architecture in readme.txt shipped with gcc-arm-embedded package. diff --git a/lib/stm32/f7/Makefile b/lib/stm32/f7/Makefile index d7f314f7..a543d3e7 100644 --- a/lib/stm32/f7/Makefile +++ b/lib/stm32/f7/Makefile @@ -26,8 +26,8 @@ PREFIX ?= arm-none-eabi CC = $(PREFIX)-gcc AR = $(PREFIX)-ar -# By default, use double precision FPU -FP_FLAGS ?= -mfloat-abi=hard -mfpu=fpv5-d16 +# STM32F7 only supports single precision FPU +FP_FLAGS ?= -mfloat-abi=hard -mfpu=fpv5-sp-d16 CFLAGS = -Os -g \ -Wall -Wextra -Wimplicit-function-declaration \ |