On 13/12/16 10:11, Christophe Lyon wrote:
On 13 December 2016 at 10:54, Thomas Preudhomme
<thomas.preudho...@foss.arm.com> wrote:
On 12/12/16 21:17, Christophe Lyon wrote:
Hi Thomas,
Thanks for working on this,
On 12 December 2016 at 18:52, Thomas Preudhomme
<thomas.preudho...@foss.arm.com> wrote:
Hi,
The logic to make -mthumb optional for Thumb-only devices is only
executed
when no -marm or -mthumb is given on the command-line. This includes
configuring GCC with --with-mode= because this makes the option to be
passed
before any others. The optional_mthumb-* testcases are skipped when -marm
or
-mthumb is passed on the command line but not when GCC was configured
with
--with-mode. Not only are the tests meaningless in these configurations,
they also spuriously FAIL if --with-mode=arm was used since the test are
built for Thumb-only devices, as reported by Christophe Lyon in [1].
[1] https://gcc.gnu.org/ml/gcc-patches/2016-11/msg02082.html
This patch adds logic to target-support.exp to check how was GCC
configured
and changes the optional_mthumb testcases to be skipped if there is a
default mode (--with-mode=). It also fixes a couple of typo on the
selectors.
How hard would it be to skip these tests only if --with-mode=arm,
such that they would still pass in configurations --with-mode=thumb?
It seems easy to extend what you propose here, doesn't it?
It is but IMO it gives a false sense of quality since it does not test the
optional -mthumb logic.
What is the motivation to make it run with --with-mode=thumb?
With your patch, they will appear as unsupported, right?
It's not a big deal.
yes indeed.
Do you know how most people configure their toolchains?
Linaro and Ubuntu (I think) use --with-mode=thumb.
Do most people use no --with-mode configure flag?
I cannot answer for people in general but in our case we use
--with-multilib-list which will build the toolchain for several -march, -mfpu
and -mfloat combinations. But it's good to have a mix, it finds more issues in
the testsuite as this very examples shows.
In the validations I run against trunk, I always have --with-mode,
except for one arm-none-eabi configuration where I also use
default cpu/fpu.
For good reason because without those it will compile target libraries without
any option and so the default would be used (ie ARMv4T in default mode). Either
you need to specify some --with-* or you built with multilib.
Maybe I should remove --with-mode=thumb from my config
--with-cpu=cortex-m3? (but I don't remember if the new logic
was backported to gcc-5/gcc-6 ?)
No it wasn't backported.
Best regards,
Thomas
