[testsuite][arm] Fix cmse-15.c expected output
authorChristophe Lyon <christophe.lyon@linaro.org>
Thu, 9 Apr 2020 14:01:22 +0000 (14:01 +0000)
committerChristophe Lyon <christophe.lyon@linaro.org>
Thu, 9 Apr 2020 14:17:48 +0000 (14:17 +0000)
commit8b5bc7d12de344707e89c5111649dcc1ae4cd1cd
tree3bfa41e5463aaac262f6f3e37a97eb1b7b3a6fdd
parent93674a72309f986c9ace2d6060916053a00da2a1
[testsuite][arm] Fix cmse-15.c expected output

The cmse-15.c testcase fails at -Os because ICF means that we
generate
nonsecure2:
        b       nonsecure0

which is OK, but does not match the currently expected
nonsecure2:
...
        bl      __gnu_cmse_nonsecure_call

(see https://gcc.gnu.org/pipermail/gcc-patches/2020-April/543190.html)

The test has already different expectations for v8-M and v8.1-M.

This patch uses check-function-bodies to account for the
different possibilities:
- v8-M vs v8.1-M via different target selectors where needed
- code generation variants (-0?) via multiple regexps

I've tested that the test now passes with --target-board=-march=armv8-m.main
and --target-board=-march=armv8.1-m.main.
gcc/testsuite/ChangeLog
gcc/testsuite/gcc.target/arm/cmse/cmse-15.c