Using float with sprintf generates hardware fault

Question

nRF51 SDK_v5.2.0.39364
GNU Compiler version 4.7 2013q1
Eclipse Juno SR2
nRF51822 Development Kit 
 uint8_t str[128]; //Declared globally 
 float tempval;
tempval = measure_temperature();
memset(&str[0], 0, 128);
sprintf((char*)&str[0], "%3.2f", tempval); <<- Hardware Fault here 
 I have attached the snapshot of the callstack.
All source files compiled with "-mfloat-abi=soft" and linked with "-u _printf_float" 
 Can someone please help? 
 Best regards,
Karthik 
 Capture.PNG

FormerMember · Accepted Answer

Hi Karthik,
Its working for me. I've added this code in the timer handler called from the scheduler. 
 
 float tempval;
 uint32_t n;
 tempval = 128.54;
 memset(&str[0], 0, 128);
 n = sprintf((char*)&str[0], "%3.2f", tempval);
 printf ("[%s] is a string %d chars long
",str,n);

 
 where str is a global uint8_t variable of 128 octet length. The string > [128.54] is a string 6 chars long
is printed every second, from the timer handler.
The linker flags used are 
 LDFLAGS = -Xlinker
LDFLAGS += -Map=$(BIN_DIR)/$(OUTPUT_NAME).map
LDFLAGS += --specs=nano.specs
LDFLAGS += -u _printf_float
LDFLAGS += -mcpu=$(CPU) 
LDFLAGS += -mthumb 
LDFLAGS += -mabi=aapcs 
LDFLAGS += -T$(LD_SCRIPT)
LDFLAGS	+= -D$(DEVICE)
LDFLAGS	+= -D$(BOARD)
LDFLAGS += -D$(BLE_REQD)
LDFLAGS	+= $(patsubst %,-L%, $(LDDIRS))
 
 The difference that I see are that i am not specifying the mfloat-abi option, while I am using the nano library for printf and sprintf. See if this works.
Cheers,
Prithvi

Using float with sprintf generates hardware fault

Top Replies