This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: patch applied to sh-elf-4.1-branch sh.h
Joern RENNECKE <joern.rennecke@st.com> wrote:
> It turns out that I didn't see the IEEE problems because the
> configuration I tested defaulted to SH1.
> Hence, the test (TARGET_SH2E && !TARGET_IEEE) in OPTIMIZATION_OPTIONS was
> always false.
With this patch, almost regressions on sh4-unknown-linux-gnu went
away. Here is the current list of tests that newly FAIL, comparing
with the vanilla 4.1.0 (20050407):
=== libstdc++ tests ===
18_support/numeric_limits.cc execution test
=== libjava tests ===
CompareNaN -O3 output - source compiled test
CompareNaN output - source compiled test
=== gcc tests ===
gcc.dg/pr15784-3.c scan-tree-dump-times ABS_EXPR 1
=== gfortran tests ===
gfortran.dg/pr20755.f -O0 (test for excess errors)
gfortran.dg/pr20755.f -O1 (test for excess errors)
gfortran.dg/pr20755.f -O2 (test for excess errors)
gfortran.dg/pr20755.f -O3 -fomit-frame-pointer (test for excess errors)
gfortran.dg/pr20755.f -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions (test for excess errors)
gfortran.dg/pr20755.f -O3 -fomit-frame-pointer -funroll-loops (test for excess errors)
gfortran.dg/pr20755.f -O3 -g (test for excess errors)
gfortran.dg/pr20755.f -Os (test for excess errors)
It seems that they aren't showstoppers, though. Anyway, I'll take
a look at them more closely.
Regards,
kaz