This is the mail archive of the
newlib@sourceware.org
mailing list for the newlib project.
RE: FW: checking whether <arm-none-eabi-gcc accepts> -g... <no>
- From: "Kleiber Tamas (BCDS/ENG2)" <Tamas dot Kleiber at bosch-connectivity dot com>
- To: Sebastian Huber <sebastian dot huber at embedded-brains dot de>, "newlib at sourceware dot org" <newlib at sourceware dot org>
- Date: Thu, 4 Dec 2014 10:17:03 +0000
- Subject: RE: FW: checking whether <arm-none-eabi-gcc accepts> -g... <no>
- Authentication-results: sourceware.org; auth=none
- References: <fe7ec752437a424eb680254b08c247c4 at FE-MBX1020 dot de dot bosch dot com> <547EE679 dot 70403 at op dot pl> <547EE901 dot 7050004 at embedded-brains dot de>
Hello Sebastian and Freddie,
thanks for your suggestions. I have moved to linux (ubuntu 10.x) and compiled newlib 2.1.0 successfully. My cross compiler is again the same as before as I am recompiling newlib for that particular compiler revision... The linux version of the compiler is this one: https://launchpad.net/gcc-arm-embedded/4.7/4.7-2014-q2-update/+download/gcc-arm-none-eabi-4_7-2014q2-20140408-linux.tar.bz2 .
Configure throws exactly the same error on Linux as it threw before on Windows and all the compiled libg.a files are 0 bytes in size for all target combinations. So now that Windows and MSYS are out of the picture, do you have another idea on what could be wrong? Based on what I have seen, again the compiler invoked in the test actually compiles the conftest.c file successfully and returns without errors.
Failing configure tests from the linux build session:
configure:3449: checking whether arm-none-eabi-gcc -mthumb -march=armv7-m accepts -g
configure:3469: arm-none-eabi-gcc -mthumb -march=armv7-m -c -g conftest.c >&5
configure:3469: $? = 0
configure: failed program was:
| /* confdefs.h */
| #define PACKAGE_NAME "newlib"
| #define PACKAGE_TARNAME "newlib"
| #define PACKAGE_VERSION "2.1.0"
| #define PACKAGE_STRING "newlib 2.1.0"
| #define PACKAGE_BUGREPORT ""
| #define PACKAGE_URL ""
| /* end confdefs.h. */
|
| int
| main ()
| {
|
| ;
| return 0;
| }
configure:3484: arm-none-eabi-gcc -mthumb -march=armv7-m -c conftest.c >&5
configure:3484: $? = 0
configure: failed program was:
| /* confdefs.h */
| #define PACKAGE_NAME "newlib"
| #define PACKAGE_TARNAME "newlib"
| #define PACKAGE_VERSION "2.1.0"
| #define PACKAGE_STRING "newlib 2.1.0"
| #define PACKAGE_BUGREPORT ""
| #define PACKAGE_URL ""
| /* end confdefs.h. */
|
| int
| main ()
| {
|
| ;
| return 0;
| }
configure:3500: arm-none-eabi-gcc -mthumb -march=armv7-m -c -g conftest.c >&5
configure:3500: $? = 0
configure: failed program was:
| /* confdefs.h */
| #define PACKAGE_NAME "newlib"
| #define PACKAGE_TARNAME "newlib"
| #define PACKAGE_VERSION "2.1.0"
| #define PACKAGE_STRING "newlib 2.1.0"
| #define PACKAGE_BUGREPORT ""
| #define PACKAGE_URL ""
| /* end confdefs.h. */
|
| int
| main ()
| {
|
| ;
| return 0;
| }
configure:3510: result: no
Mit freundlichen Grüßen / Best regards
Tamas Kleiber
BeQIK
-----Original Message-----
From: Sebastian Huber [mailto:sebastian.huber@embedded-brains.de]
Sent: Wednesday, December 03, 2014 11:42 AM
To: Kleiber Tamas (BCDS/ENG2); newlib@sourceware.org
Subject: Re: FW: checking whether <arm-none-eabi-gcc accepts> -g... <no>
On 03/12/14 11:31, Freddie Chopin wrote:
> On 12/03/2014 11:02 AM, Kleiber Tamas (BCDS/ENG2) wrote:
>> Could you please help me understand what could be the problem with my
>> setup?
>
> I didn't dig deep into that issue, but I would assume that the problem
> is Windows and MSYS. If you configure/build in the same way as
> launchpad toolchain is built (check their scripts in the "...-src"
> package), then there's no other option.
>
> I also compile newlib (lates master) myself in a similar way as
> launchpad guys do (my compilation is here -
> https://sourceforge.net/projects/bleeding-edge/ , scripts are also
> included), but I'm doing that on Linux and it "just works"... Trying
> to build any "typical" project on Windows is just too much trouble -
> it's easier and faster to install any Linux distro directly on your
> machine or using VirtualBox.
Yes, this is also my experience. Building GCC on Windows is a waste of
time. Its much faster and reliable to Canadian Cross compile on Linux.
Most distributions provide MinGW cross-compiler packages, e.g. for
openSUSE https://build.opensuse.org/project/show/windows:mingw:win32.
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail : sebastian.huber@embedded-brains.de
PGP : Public key available on request.
Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.