...uses hardware FP, whereas u-boot uses software FP...
Hi there,
I just tried
CROSS_COMPILE=arm-softfloat-linux-gnu- ./MAKEALL ARM9
on the CVS HEAD but got
arm-softfloat-linux-gnu-ld: ERROR:
/opt/crosstool/arm-softfloat-linux-gnu/gcc-3.3.3-glibc-2.3.2-linux-2.4.27/lib/gcc-lib/arm-softfloat-linux-gnu/3.3.3/libgcc.a(_divsi3.oS) uses hardware FP, whereas u-boot uses software FP
arm-softfloat-linux-gnu-ld: failed to merge target specific data of file /opt/crosstool/arm-softfloat-linux-gnu/gcc-3.3.3-glibc-2.3.2-linux-2.4.27/lib/gcc-lib/arm-softfloat-linux-gnu/3.3.3/libgcc.a(_divsi3.oS)
make: *** [u-boot] Fehler 1
for at91rm9200dk and cmc_pu2.
Removing " -msoft-float" from /cpu/at91rm9200/config.mk solves this problem.
But why?
My toolchain gcc-3.3.3-glibc-2.3.2-linux-2.4.27 was built with crosstool using
export TARGET=arm-softfloat-linux-gnu
export TARGET_CFLAGS="-O"
export GCC_EXTRA_CONFIG="--with-float=soft"
export GLIBC_EXTRA_CONFIG="--without-fp"
export BINUTILS_DIR=binutils-2.15
export GCC_DIR=gcc-3.3.3
export GLIBC_DIR=glibc-2.3.2
export LINUX_DIR=linux-2.4.27
so I am pretty sure that I am using softfloat by default. But then why is it a problem if I pass -msoft-float to the compiler?
Thanks!
Signature · 1 line
Steven
Quoted 8 lines · In message <[email protected]> you wrote:
CROSS_COMPILE=arm-softfloat-linux-gnu- ./MAKEALL ARM9
on the CVS HEAD but got
arm-softfloat-linux-gnu-ld: ERROR:
/opt/crosstool/arm-softfloat-linux-gnu/gcc-3.3.3-glibc-2.3.2-linux-2.4.27/lib/gcc-lib/arm-softfloat-linux-gnu/3.3.3/libgcc.a(_divsi3.oS) uses hardware FP, whereas u-boot uses software FP
This message is clear, isn't it? Your compiler libraries are built using hardware FP, whereas u-boot uses software FP
Quoted 1 line
Removing " -msoft-float" from /cpu/at91rm9200/config.mk solves this problem.
Not really.
Quoted 1 line
My toolchain gcc-3.3.3-glibc-2.3.2-linux-2.4.27 was built with crosstool using
Maybe you should ask this on a crosstools related mailing list (probably linux-arm-toolchain).
Quoted 2 lines
so I am pretty sure that I am using softfloat by default. But then why is it a problem if I pass -msoft-float to the compiler?
Maybe you are using softfloat, but your compiler libraries aren't.
Best regards,
Wolfgang Denk
Signature · 4 lines
Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de When you die, the first thing you lose is your life. The next thing is the illusions. - Terry Pratchett, _Pyramids_
Dear Wolfgang,
Quoted 10 lines
Quoted 3 lines
arm-softfloat-linux-gnu-ld: ERROR:
/opt/crosstool/arm-softfloat-linux-gnu/gcc-3.3.3-glibc-2.3.2-linux-2.4.27/lib/gcc-lib/arm-softfloat-linux-gnu/3.3.3/libgcc.a(_divsi3.oS) uses hardware FP, whereas u-boot uses software FPThis message is clear, isn't it? Your compiler libraries are built using hardware FP, whereas u-boot uses software FP
...Quoted 1 line
My toolchain gcc-3.3.3-glibc-2.3.2-linux-2.4.27 was built with crosstool using
Maybe you are using softfloat, but your compiler libraries aren't.
So your suggesting that my glibc is built with softfloat support but not the gcc itself?
IIUC then U-Boot does not need glibc?
Signature · 1 line
Steven
Dear Steven,
Quoted 3 lines · in message <[email protected]> you wrote:
Quoted 3 lines
Quoted 3 lines
/opt/crosstool/arm-softfloat-linux-gnu/gcc-3.3.3-glibc-2.3.2-linux-2.4.27/lib/gcc-lib/arm-softfloat-linux-gnu/3.3.3/libgcc.a(_divsi3.oS) uses hardware FP, whereas u-boot uses software FP
Please read this message again, and carefully.
Quoted 7 lines
Quoted 2 lines
This message is clear, isn't it? Your compiler libraries are built using hardware FP, whereas u-boot uses software FP
So your suggesting that my glibc is built with softfloat support but not the gcc itself?
IIUC then U-Boot does not need glibc?
I never mentioned glibc. Nor is it mentioned in the error message. The library in question is "libgcc.a", where the compiler itself (GCC) provides utility functions (like here divsi3 to perferm a signed division operation).
I repeat: Your compiler libraries are built using hardware FP, whereas u-boot uses software FP.
In other words: your toolchain is not configured as you expected.
Best regards,
Wolfgang Denk
Signature · 3 lines
Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de "I like your game but we have to change the rules."
Dear Wolfgang,
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
This message is clear, isn't it? Your compiler libraries are built using hardware FP, whereas u-boot uses software FP
Sure it is.
Quoted 10 lines
Quoted 4 lines
So your suggesting that my glibc is built with softfloat support but not the gcc itself?
IIUC then U-Boot does not need glibc?
I never mentioned glibc. Nor is it mentioned in the error message. The library in question is "libgcc.a", where the compiler itself (GCC) provides utility functions (like here divsi3 to perferm a signed division operation).
Quoted 4 lines
I repeat: Your compiler libraries are built using hardware FP, whereas u-boot uses software FP.
In other words: your toolchain is not configured as you expected.
I just wanted to make sure that I understood corretcly.
Since my user applications seems to run fine under linux I assume that my glibc supports softfloat.
U-Boot does not use glibc but libgcc.a. Since the compiler complains while compiling U-Boot, it seems that my libgcc.a does not support softfloat. So the problems lies in how my compiler was built and not how the glibc was built. Did I get this right?
Thanks a million!
--
Steven
Quoted 3 lines · In message <[email protected]> you wrote:
Quoted 3 lines
I repeat: Your compiler libraries are built using hardware FP, whereas u-boot uses software FP.
...
Quoted 2 lines
Since my user applications seems to run fine under linux I assume that my glibc supports softfloat.
Maybe, maybe not.
Quoted 1 line
U-Boot does not use glibc but libgcc.a. Since the compiler complains while
No. U-Boot does NOT use libgcc.a.
It is your compiler which requires libgcc.a.
Quoted 3 lines
compiling U-Boot, it seems that my libgcc.a does not support softfloat. So the problems lies in how my compiler was built and not how the glibc was built. Did I get this right?
Right. It has nothing to do with glibc. It's the compiler libraries which use hardware FP, which does not mix with any other code that uses soft-float.
[This is why I have doubts if your Linux stuff really uses soft- float; you'd run in the same problem there, too. But this is a different issue.]
Best regards,
Wolfgang Denk
Signature · 4 lines
Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de The universe does not have laws - it has habits, and habits can be broken.
Quoted 5 lines · Wolfgang Denk
Quoted 2 lines · In message <[email protected]> you wrote:
Quoted 2 lines
I repeat: Your compiler libraries are built using hardware FP, whereas u-boot uses software FP.
Quoted 7 lines
...
Quoted 2 lines
Since my user applications seems to run fine under linux I assume that my glibc supports softfloat.
Maybe, maybe not.
Any hint how I could make sure and test it?
Quoted 5 lines
Quoted 1 line
U-Boot does not use glibc but libgcc.a. Since the compiler complains while
No. U-Boot does NOT use libgcc.a.
It is your compiler which requires libgcc.a.
Ok. Thanks.
Quoted 11 lines
Quoted 3 lines
compiling U-Boot, it seems that my libgcc.a does not support softfloat. So the problems lies in how my compiler was built and not how the glibc was built. Did I get this right?
Right. It has nothing to do with glibc. It's the compiler libraries which use hardware FP, which does not mix with any other code that uses soft-float.
[This is why I have doubts if your Linux stuff really uses soft- float; you'd run in the same problem there, too. But this is a different issue.]
Ok.
Thanks a million so far. I guess I try my luck on crossgcc at sources.redhat.com then.
--
Steven
Quoted 4 lines · In message <[email protected]> you wrote:
Quoted 2 lines
Maybe, maybe not.
Any hint how I could make sure and test it?
Disassemble and grep for FP instructions?
Quoted 1 line
Thanks a million so far. I guess I try my luck on crossgcc at sources.redhat.com then.
Or, as mentioned before, on the ARM toolchain mailing list.
Best regards,
Wolfgang Denk
Signature · 4 lines
Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de What nonsense people talk about happy marriages! A man can be happy with any woman so long as he doesn't love her. -- Oscar Wilde
Quoted 7 lines · Wolfgang Denk
Quoted 3 lines · In message <[email protected]> you wrote:
Quoted 1 line
Maybe, maybe not.
Any hint how I could make sure and test it?
Quoted 7 lines
Disassemble and grep for FP instructions?
Quoted 1 line
Thanks a million so far. I guess I try my luck on crossgcc at sources.redhat.com then.
Or, as mentioned before, on the ARM toolchain mailing list.
I did not know, there's an ARM specific toolchain mailing list. Do you mind giving me a hint where to find it? Thanks.
Have a nice weekend.
Signature · 1 line
Steven
Quoted 5 lines · In message <[email protected]> you wrote:
Quoted 2 lines
Or, as mentioned before, on the ARM toolchain mailing list.
I did not know, there's an ARM specific toolchain mailing list. Do you mind giving me a hint where to find it?
linux-arm-toolchain at lists.arm.linux.org.uk
Best regards,
Wolfgang Denk
Signature · 6 lines
Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de "The glory of creation is in its infinite diversity." "And in the way our differences combine to create meaning and beauty." -- Dr. Miranda Jones and Spock, "Is There in Truth No Beauty?", stardate 5630.8