diff options
author | Hans-Peter Nilsson <hp@axis.com> | 2023-02-28 18:37:32 +0100 |
---|---|---|
committer | Hans-Peter Nilsson <hp@bitrange.com> | 2023-03-01 02:39:15 +0100 |
commit | 6622f7e8b896071cc554774171353cc82a4a3a5f (patch) | |
tree | 7c915bf6835a650c04192c7e3b878c103e38e5ef /gcc/fortran/trans-openmp.cc | |
parent | 6f9e2f144ec5b76c139073bae2a3c1cc65aed309 (diff) | |
download | gcc-6622f7e8b896071cc554774171353cc82a4a3a5f.zip gcc-6622f7e8b896071cc554774171353cc82a4a3a5f.tar.gz gcc-6622f7e8b896071cc554774171353cc82a4a3a5f.tar.bz2 |
testsuite: Fix analyzer errors for newlib-fd
Investigating analyzer testsuite errors for cris-elf. The same are
seen for pru-elf according to posts to gcc-testresults@.
The test fd-access-mode-target-headers.c uses the analyzer
"sm-fd" which for this use requires (e.g.) that constants
O_ACCMODE, O_RDONLY and O_WRONLY are defined as literal
constants. While for glibc, O_ACCMODE is defined as:
#define O_ACCMODE 0003
in newlib, it's defined as:
#define O_ACCMODE (O_RDONLY|O_WRONLY|O_RDWR)
and the analyzer is not able to make use of an expression
like this (even though O_RDONLY, O_WRONLY and O_RDWR are
defined as literal constants and the whole evaluates to 3).
Better do as for AIX and skip this test.
testsuite:
* gcc.dg/analyzer/fd-access-mode-target-headers.c: Skip for
newlib targets too.
Diffstat (limited to 'gcc/fortran/trans-openmp.cc')
0 files changed, 0 insertions, 0 deletions