coreutils/coreutils-skip-gnulib-test-tls.patch
Bernhard Voelker e7f4664e8f Accepting request 1080971 from home:berny:branches:Base:System
- update to 9.3:
  Bug fixes:
  * cp --reflink=auto (the default), mv, and install
    will again fall back to a standard copy in more cases.
    Previously copies could fail with permission errors on
    more restricted systems like android or containers etc.
    [bug introduced in coreutils-9.2]
  * cp --recursive --backup will again operate correctly.
    Previousy it may have issued "File exists" errors when
    it failed to appropriately rename files being replaced.
    [bug introduced in coreutils-9.2]
  * date --file and dircolors will now diagnose a failure to read a file.
    Previously they would have silently ignored the failure.
    [This bug was present in "the beginning".]
  * md5sum --check again correctly prints the status of each file checked.
    Previously the status for files was printed as 'OK' once any file had passed.
    This also applies to cksum, sha*sum, and b2sum.
    [bug introduced in coreutils-9.2]
  * wc will now diagnose if any total counts have overflowed.
    [This bug was present in "the beginning".]
  * `wc -c` will again correctly update the read offset of inputs.
    Previously it deduced the size of inputs while leaving the offset unchanged.
    [bug introduced in coreutils-8.27]
  * Coreutils programs no longer fail for timestamps past the year 2038
    on obsolete configurations with 32-bit signed time_t, because the
    build procedure now rejects these configurations.
    [This bug was present in "the beginning".]
  Changes in behavior:
  * 'cp -n' and 'mv -n' now issue an error diagnostic if skipping a file,
    to correspond with -n inducing a nonzero exit status as of coreutils 9.2.
    Similarly 'cp -v' and 'mv -v' will output a message for each file skipped
    due to -n, -i, or -u.
  New features:
  * cp and mv now support --update=none to always skip existing files
    in the destination, while not affecting the exit status.
    This is equivalent to the --no-clobber behavior from before v9.2.
- drop fix-reflink-fallback.patch (upstream).
- add coreutils-tests-skip-cpuinfo-replaced.patch: avoid FP test failure.

OBS-URL: https://build.opensuse.org/request/show/1080971
OBS-URL: https://build.opensuse.org/package/show/Base:System/coreutils?expand=0&rev=350
2023-04-20 21:06:24 +00:00

38 lines
1.2 KiB
Diff

Subject: Skip the gnulib test 'test-tls' on some platforms
On i586, x86_64, ppc and ppc64, this test is known to sometimes fail
with a diagnostic like:
Starting test_tls ...*** Error in `./test-tls': free(): invalid pointer: 0x00007f21500008c0 ***
======= Backtrace: =========
/lib64/libc.so.6(+0x7406f)[0x7f215845006f]
/lib64/libc.so.6(+0x7989e)[0x7f215845589e]
/lib64/libpthread.so.0(+0x7ee2)[0x7f215878fee2]
/lib64/libpthread.so.0(+0x813e)[0x7f215879013e]
/lib64/libc.so.6(clone+0x6d)[0x7f21584c3d6d]
* gnulib-tests/gnulib.mk (test-tls): Comment to skip for now.
---
gnulib-tests/gnulib.mk | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
Index: gnulib-tests/gnulib.mk
===================================================================
--- gnulib-tests/gnulib.mk.orig
+++ gnulib-tests/gnulib.mk
@@ -2765,9 +2765,10 @@ EXTRA_DIST += test-timespec.c macros.h
## begin gnulib module tls-tests
-TESTS += test-tls
-check_PROGRAMS += test-tls
-test_tls_LDADD = $(LDADD) @LIBMULTITHREAD@ @YIELD_LIB@
+# Fails on i586 and x86_64.
+#TESTS += test-tls
+#check_PROGRAMS += test-tls
+#test_tls_LDADD = $(LDADD) @LIBMULTITHREAD@ @YIELD_LIB@
EXTRA_DIST += test-tls.c