> On 3 Apr 2020, at 17:31, Andreas Fritiofson <andreas.fritiof...@gmail.com>
> wrote:
>
>
>
> On Fri, Apr 3, 2020 at 1:09 PM Liviu Ionescu <i...@livius.net> wrote:
>
>
> For example, I test the arm-none-eabi-gcc xPack by compiling two simple hello
> world, one with c and one with c++, with and without -flto; the purpose is to
> make the compiler use as many execution paths as possible, so if there are
> any missing plug-ins or libraries, to identify this. I'm definitelly not
> doing functional tests, to see if the compiler generates the correct code, I
> expect these test to be done by those who maintain the compiler.
>
>
> Reinventing the wheel, GCC already has an extensive testsuite which is
> recommended to run to make sure *your* build generates correct binaries:
> https://gcc.gnu.org/install/test.html
>
> Although the full suite will take quite a while to run.
Exactly!
Those tests are way, way, beyond my scope; in my toolchain distributions I use
exactly the same sources as Arm does in their cross toolchain, and Arm claims
to have tested it intensively, so I gamble and assume that if I can compile a
simple hello world, I can compile the same programs as they do.
And it is not reinventing the wheel, it is a different purpose, using a
compromise that seems to have an acceptable risk.
Regards,
Liviu
_______________________________________________
OpenOCD-devel mailing list
OpenOCD-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/openocd-devel
