EmbDev.net

Forum: ARM programming with GCC/GNU tools WinARM newlib float


Author: kasuppke (Guest)
Posted on:

Rate this post
0 useful
not useful
Hi,

I just asked this in the uC forum but it seems to be erased?
When building my project with WinARM I get linker errors saying
somthing like: newlib is compiled with hardware point, whereas
project is compiled with software floating point.
(I have the original text not here ;-)
So it seems that newlib is compiled with hardware floating point
and reentrant? I don't think Philips and Atmel ARM7 have a hardware fpu?
Am I something missing in the makefile?
Greets, Joerg

Author: Martin Thomas (Guest)
Posted on:

Rate this post
0 useful
not useful
kasuppke wrote:

> When building my project with WinARM I get linker errors saying
> somthing like: newlib is compiled with hardware point, whereas
> project is compiled with software floating point.

"Something like" is not a very usefull error-report. Create a minimal
example with source-code, linker-script and makefile, place it on a
server and send the URL or send it as attachment to the e-mail adress
mentioned in the WinARM readme-file.

> So it seems that newlib is compiled with hardware floating point
> and reentrant? I don't think Philips and Atmel ARM7 have a hardware fpu?

The newlib is build for reentrant syscalls, yes. I do not remember any
configuration option for a default floating-point setting.

Martin Thomas

Author: Joerg (Guest)
Posted on:

Rate this post
0 useful
not useful
Hello Martin,

> "Something like" is not a very usefull error-report. Create a minimal
> example with source-code, linker-script and makefile, place it on a
> server and send the URL or send it as attachment to the e-mail adress
> mentioned in the WinARM readme-file.

d:\programme\winarm\bin\..\lib\gcc\arm-elf\4.0.2\..\..\..\..\arm-elf\bin 
\ld.exe:
ERROR:
d:/programme/winarm/bin/../lib/gcc/arm-elf/4.0.2/../../../../arm-elf/lib 
\libc.a(errno.o)
uses hardware FP, whereas tidnreadarm.elf uses software FP


> The newlib is build for reentrant syscalls, yes. I do not remember any
> configuration option for a default floating-point setting.

I do compile with -msoft-float for my sources.
I think you need this if you use floating point operations?
Maybe I can send you some of my project.
Regards,
Joerg

Author: Joerg (Guest)
Posted on:

Rate this post
0 useful
not useful
Hello again,

I started to use my own version off newlib which worked until
the same problem with the libgcc came up.

d:\programme\winarm\bin\..\lib\gcc\arm-elf\4.0.2\..\..\..\..\arm-elf\bin 
\ld.exe:
ERROR:
d:/programme/winarm/bin/../lib/gcc/arm-elf/4.0.2\libgcc.a(_fixsfsi.o)
uses hardware FP, whereas tidnreadarm.elf uses software FP

Any ideas yet?
Regards,
Joerg

Author: Martin Thomas (Guest)
Posted on:

Rate this post
0 useful
not useful
Joerg wrote:
> Hello again,
>
> I started to use my own version off newlib which worked until
> the same problem with the libgcc came up.
>
> d:\programme\winarm\bin\..\lib\gcc\arm-elf\4.0.2\..\..\..\..\arm-elf\bin 
\ld.exe:
> ERROR:
> d:/programme/winarm/bin/../lib/gcc/arm-elf/4.0.2\libgcc.a(_fixsfsi.o)
> uses hardware FP, whereas tidnreadarm.elf uses software FP

Did you try the -EB option for the linker?

Author: Mh Tay (lpc2103)
Posted on:

Rate this post
0 useful
not useful
When I installed WinARM to a 2nd computer today,
I messed up the %PATH% and got some files compiled
with GCC3.2 and some with GCC4.0.  And I had same
linker error messages as yours.
Could it be you are running 2 versions of compilers
accidentally??

Author: Joerg (Guest)
Posted on:

Rate this post
0 useful
not useful
Hi,

after several clean- and re- builds I found
the -msoft-float option as reason for the problem.
(maybe some confusion with GNUARM but I don't think so)
Floating point treatment of newlib (and gcc-lib?)
still seems to work without this option.
Still wonder why my own newlib compilation
didn't came up with this problem....
I think ARM with Hardware-FP needs the -mhard-float
to use it then?
Thanks for your support (so far :-)

Joerg

Author: Gerard Sequeira (bluehash)
Posted on:

Rate this post
0 useful
not useful
> the -msoft-float option as reason for the problem.

> I think ARM with Hardware-FP needs the -mhard-float


Where could I find or put in those switches?

Author: Gerard Sequeira (bluehash)
Posted on:

Rate this post
0 useful
not useful
These are the messages i get

"c:\program
files\yagarto\bin\..\lib\gcc\arm-elf\4.1.1\..\..\..\..\arm-elf\bin\ld.ex 
e:
ERROR: startup/libea_startup_thumb.a(framework.o) uses software FP,
whereas srv1.elf uses hardware FP"

Im using gcc 4.1.1 on the LPC2106 board from olimex.

My makefile is as follows:

"
##########################################################
#
# General makefile for building executable programs and
# libraries for Embedded Artists' QuickStart Boards.
# (C) 2001-2005 Embedded Artists AB
#
##########################################################

# Name of target (executable program or library)
NAME      = srv1

# Link program to RAM or ROM (possible values for LD_RAMROM is RAM or
ROM,
# if not specified = ROM)
LD_RAMROM = ROM

# Name if specific CPU used (used by linker scripts to define correct
memory map)
# Valid CPUs are: LPC2101, LPC2102, LPC2103, LPC2104, LPC2105, LPC2106
#                 LPC2114, LPC2119
#                 LPC2124, LPC2129
#                 LPC2131, LPC2132, LPC2134, LPC2136, LPC2138
#                 LPC2141, LPC2142, LPC2144, LPC2146, LPC2148
#                 LPC2194
#                 LPC2210, LPC2220, LPC2212, LPC2214,
#                 LPC2290, LPC2292, LPC2294
# If you have a new version not specified above, just select one of the
old
# versions with the same memory map.
CPU_VARIANT = LPC2106

# It is possible to override the automatic linker file selection with
the variable below.
# No not use this opion unless you have very specific needs.
#LD_SCRIPT      = build_files/myOwnLinkScript_rom.ld
#LD_SCRIPT_PATH =

# ELF-file contains debug information, or not
# (possible values for DEBUG are 0 or 1)
# Extra debug flags can be specified in DBFLAGS
DEBUG   = 1
#DBFLAGS =

# Optimization setting
# (-Os for small code size, -O2 for speed)
OFLAGS  = -Os

# Extra general flags
# For example, compile for ARM / THUMB interworking (EFLAGS =
-mthumb-interwork)
EFLAGS  =

# Program code run in ARM or THUMB mode
# Can be [ARM | THUMB]
CODE    = THUMB

# List C source files here.
CSRCS   = main.c \
          uart0.c \
          uart1.c \
          jpeg.c \
          camera.c \
          iap.c \
          utils.c

# List assembler source files here
ASRCS   =

# List subdirectories to recursively invoke make in
SUBDIRS = startup

# List additional libraries to link with
LIBS    = startup/libea_startup_thumb.a

# Add include search path for startup files, and other include
directories
INC     = -I./startup

# Select if an executable program or a library shall be created
PROGRAM_MK  = true
#LIBRARY_MK  = true

# Output format on hex file (if making a program); can be [srec | ihex]
HEX_FORMAT  = ihex

# Program to download executable program file into microcontroller's
FLASH
DOWNLOAD    = lpc21isp.exe

# Configurations for download program
# Which com-pot that is used, which download speed and what crystal
frequency on the board.
DL_COMPORT  = com1
DL_BAUDRATE = 115200
DL_CRYSTAL  = 14746

#######################################################################
include build_files/general.mk
#######################################################################
"

Author: Clifford Slocombe (clifford)
Posted on:

Rate this post
0 useful
not useful
Joerg wrote:
> Hi,
>
> after several clean- and re- builds I found
> the -msoft-float option as reason for the problem.
> (maybe some confusion with GNUARM but I don't think so)
> Floating point treatment of newlib (and gcc-lib?)
> still seems to work without this option.
> Still wonder why my own newlib compilation
> didn't came up with this problem....
> I think ARM with Hardware-FP needs the -mhard-float
> to use it then?
> Thanks for your support (so far :-)
>
> Joerg

The default is to use -mhard-float, but this does not mean that
you have to have a hardware FPU. An FPU instruction without an FPU
present causes an invalid instruction exception, the exception handler
emulates the hardware floating point operation in software. (see
http://gcc.gnu.org/onlinedocs/gcc-4.0.4/gcc/ARM-Op...
for details). The idea is that the same code will work with or without
an FPU, with -msoft-float, you would never see any performance increase
when an FPU is present because it would not be used.

That said, the only off-the-shelf ARM micro-controller with floating
point hardware currently available is the NXP LPC3180, but it uses the
VFP co-processor unit rather than the older FPA co-processor. The
default Newlib builds are for the FPA only, you would have to re-build
it yourself to get VFP support. The consequence of this is that without
a rebuild library FP functions (libm.a) will be software emulated even
when VFP hardware is available.

The solution is simple: don't use -msoft-float. You don't need it, and
it is not always supported. There may be a marginal performance
improvement from not having to do the mode switch on exception, but it
is not significant - it is dog slow either way! Consider avoiding
floating point altogether. And as a general rule, do not use any
non-default compiler options unless you are sure exactly what they do!

Clifford

Author: Gerard Sequeira (bluehash)
Posted on:

Rate this post
0 useful
not useful
> The solution is simple: don't use -msoft-float. You don't need it, and
> it is not always supported.

Yes..but I havent specified those switches anywhere.
I still get those errors.

Thanks.

Author: Clifford Slocombe (clifford)
Posted on:

Rate this post
0 useful
not useful
gerard sequeira wrote:
>
> These are the messages i get
>
> "c:\program
> files\yagarto\bin\..\lib\gcc\arm-elf\4.1.1\..\..\..\..\arm-elf\bin\ld.ex e:
> ERROR: startup/libea_startup_thumb.a(framework.o) uses software FP,
> whereas srv1.elf uses hardware FP"
>
> Im using gcc 4.1.1 on the LPC2106 board from olimex.

Your problem is different (you probably should not have chipped in on
Joerg's thread on that basis) srv1.elf is built with the default
hard-float, but framework.o was built with soft-float.

You have to have consistency; either add -msoft-float to the EFLAGS
macro in the make file, or if you are able, rebuild
libea_startup_thumb.a without -msoft-float (I'd recommend the latter if
you have the source and build files).

Clifford

Author: Clifford Slocombe (clifford)
Posted on:

Rate this post
0 useful
not useful
gerard sequeira wrote:
>> The solution is simple: don't use -msoft-float. You don't need it, and
>> it is not always supported.
>
> Yes..but I havent specified those switches anywhere.
> I still get those errors.
>
> Thanks.

This is not your thread, and I was not referring to your problem, which
is different, which is why the solution does not apply! I have however
just answered your problem - hope it helps, but take care not to hijack
other peoples threads.

Author: Gerard Sequeira (bluehash)
Posted on:

Rate this post
0 useful
not useful
Clifford Slocombe wrote:
> gerard sequeira wrote:
>>> The solution is simple: don't use -msoft-float. You don't need it, and
>>> it is not always supported.
>>
>> Yes..but I havent specified those switches anywhere.
>> I still get those errors.
>>
>> Thanks.
>
> This is not your thread, and I was not referring to your problem, which
> is different, which is why the solution does not apply! I have however
> just answered your problem - hope it helps, but take care not to hijack
> other peoples threads.

Sorry about that. I didnt mean to.
I'll start a new one.
Thank you for your help

Reply

Entering an e-mail address is optional. If you want to receive reply notifications by e-mail, please log in.

Rules — please read before posting

  • Post long source code as attachment, not in the text
  • Posting advertisements is forbidden.

Formatting options

  • [c]C code[/c]
  • [avrasm]AVR assembler code[/avrasm]
  • [code]code in other languages, ASCII drawings[/code]
  • [math]formula (LaTeX syntax)[/math]




Bild automatisch verkleinern, falls nötig
Note: the original post is older than 6 months. Please don't ask any new questions in this thread, but start a new one.