Commit v8.23-63-g111a2b9 removed the expensive tag on this test,
as it runs quickly on systems with inotify. However without that
it would take about 8 minutes for the test to complete all iterations.
* tests/tail-2/inotify-rotate.sh: Tag as expensive without inotify.
Also adjust the polling parameters used on systems without inotify
so that the test completes within about 15 seconds.
Included in this are gnulib changes 3ea43e02 2768ceb7
which make the device IDs from /proc/self/mountinfo
available to df. This can be leveraged by a subsequent
change to df to present a more accurate list of file systems.
* bootstrap: Merge from gnulib.
* src/ls.c (dev_ino_pop): s/obstack_blank/obstack_blank_fast/
as this API/ABI has changed, giving memory exhausted errors
if negative (large positive) numbers are passed to obstack_blank().
* tests/df/skip-duplicates.sh: Adjust as the new gnulib code
requires a non NULL mnt_opts even when mnt_type is not "none".
http://hydra.nixos.org/build/18129583 identified (on OS X)
an incorrect test assumption in the previous commit.
* gl/lib/xdectoint.c (__xnumtoint): Suppress the EINVAL
error message as it's redundant in this context.
* tests/misc/tail.pl: Suppress _optionally_ appended
strerror messages.
* tests/fmt/base.pl: Likewise.
* tests/pr/pr-tests.pl: Likewise.
* tests/split/l-chunk.sh: Likewise.
Following on from commit v8.23-82-gaddae94, consistently diagnose
numbers that are too large, so as to distinguish from other errors,
and make the limits obvious.
* gl/modules/xdectoint: A new module implementing xdecto[iu]max(),
which handles the common case of parsing a bounded integer and
exiting with a diagnostic on error.
* gl/lib/xdectoimax.c: The signed variant.
* gl/lib/xdectoint.c: The parameterized implementation.
* gl/lib/xdectoint.h: The interface.
* gl/lib/xdectoumax.c: The unsigned variant.
* bootstrap.conf: Reference the new module.
* cfg.mk (exclude_file_name_regexp--sc_require_config_h_first):
Exclude the parameterized templates.
* src/csplit.c: Output EOVERFLOW or ERANGE errors if appropriate.
* src/fmt.c: Likewise.
* src/fold.c: Likewise.
* src/head.c: Likewise.
* src/ls.c: Likewise.
* src/nl.c: Likewise.
* src/nproc.c: Likewise.
* src/shred.c: Likewise.
* src/shuf.c: Likewise.
* src/stdbuf.c: Likewise.
* src/stty.c: Likewise.
* src/tail.c: Likewise.
* src/truncate.c: Likewise.
* src/split.c: Likewise.
* src/pr.c: Likewise.
* tests/pr/pr-tests.pl: Adjust to avoid matching errno diagnostic.
* tests/fmt/base.pl: Likewise.
* tests/split/l-chunk.sh: Likewise.
* tests/misc/shred-negative.sh: Likewise.
* tests/misc/tail.pl: Likewise. Also remove the redundant
existing ERR_SUBST from test err-6.
* tests/ls/hex-option.sh: Check HEX/OCT options.
* tests/misc/shred-size.sh: Likewise.
* tests/misc/stty-row-col.sh: Likewise.
OS/2 traditional shells(cmd) do not expand a response file(@file)
or a wildcard. Expand them in each utility itself.
* src/system.h (initialize_main): Define on OS/2. Expand a response
file and a wildcard.
* THANKS.in: Change the comment at the top to send change requests
regarding this file to the main mailing list rather than referring
to cp's --help output for the mailing list's address - which does
not include that information anymore.
glibc <= 2.5 would crash when passed invalid long double values,
therefore internal gnulib routines were used, essentially only by od,
to output such invalid values. Later glibc versions don't crash,
as per https://sourceware.org/bugzilla/show_bug.cgi?id=4586
and subsequently od was adjusted to use the system printf routines
through the use of the ftoastr module with commit v8.7-22-ga71c22f.
Consequently our testing of this feature was moot, and use of
the gnulib printf replacement for printf(1), od(1) and error(3) etc.
was redundant.
* configure.ac (gl_printf_safe): Unset so that we don't check that
"nan" is output for these long double values.
* tests/misc/od-float.sh: Adjust all existing checks to fail if od
exits with failure status (like crashing for example). Add a new case
for one of the problematic invalid long double values for x86_64.
We only check that od exits successfully at present, which may change
if https://sourceware.org/bugzilla/show_bug.cgi?id=17661 is resolved.
* gl/lib/rand-isaac.c (isaac_refill): readisaac() purposefully passes
unaligned pointers to avoid memory copies. This is only done on
platforms where this is defined, so avoid the associated
runtime warning generated with -fsanitize=undefined, which is:
lib/rand-isaac.c:125:182: runtime error: store to misaligned address
0x63100003d7fd for type 'isaac_word', which requires 8 byte alignment
0x63100003d7fd: note: pointer points here
47 ce ed a4 be be be 00 00 00 00 00 00 00 00 ...
^
Prompted by the implicit -O3 added by american-fuzzy-lop,
seen with GCC 4.9.2 on x86_64.
src/pr.c: In function 'print_files.part.5':
src/pr.c:1781:6: error: assuming signed overflow does not occur
when simplifying conditional to constant [-Werror=strict-overflow]
if (cols_ready_to_print () == 0)
This happens because cols_ready_to_print() is inlined
thus reducing the comparison to the N variable in print_page().
Now this can't overflow due to the protection when parsing the
specified column, but use an unsigned type to avoid the
apparent signed overflow.
* src/pr.c (cols_ready_to_print): Increment an unsigned type to
avoid the subsequent signed overflow warning.
This patch fixes the handling of sub-bind-mount cycles which are
incorrectly detected as the file system errors. If you bind mount the
directory 'a' to its subdirectory 'a/b/c' and then run 'du a/b' you
will get the circular dependency warning even though nothing is wrong
with the file system. This happens because the first directory that is
traversed twice in this case is not a bind mount but a child of bind
mount. The solution is to traverse all the directories in the cycle
that fts detected and check whether they are not a (bind) mount.
* src/du.c (mount_point_in_fts_cycle): New function that checks whether
any of the directories in the cycle that fts detected is a mount point.
* src/du.c (process_file): Update the function to use the new function
that looks up all the directories in the fts cycle instead of only the
last one.
* tests/du/bind-mount-dir-cycle-v2.sh: New test case that exhibits the
described behavior.
* tests/local.mk: Reference the new root test.
* NEWS: Mention the bug fix.
These checks weren't correctly avoided in commit v8.23-66-g222d7ac
* tests/cp/same-file.sh: Avoid all hardlink to symlink tests
on platforms where that's not supported.
Identified by http://hydra.nixos.org/build/17636446
"zu" was output on solaris 8 for example rather than the number,
since coreutils-8.22.
* cfg.mk: Disallow %z, since we don't currently use the gnulib
fprintf module, so any usage with it is non portable. Also
our usage with error() currently works only through an ancillary
dependency on the vfprintf gnulib module.
* src/rm.c (main): Use %PRIuMAX rather than %zu for portability.
* src/dd.c (alloc_[io]buf): Likewise for consistency.
* src/od.c (main): Likewise.
* src/split.c (set_suffix_length): Likewise.
* NEWS: Mention the rm bug fix.
Reported in http://bugs.gnu.org/19184
Solaris 8 was seen to issue this error:
"printf: `&': illegal format character"
* test/dd/ascii.sh: Use the coreutils printf in this test
rather than the system one, to avoid portability issues.
If '\n' was present at the size_t boundary of a file,
then that and subsequent data would be discarded.
* src/paste.c (paste_parallel): Avoid the overflow issue
by changing the flag to a boolean rather than a count.
* NEWS: Mention the bug fix.
* src/df.c (filter_mount_list): Separate remote locations are
generally explicitly mounted, so list each even if they share
the same remote device and thus storage. However with --total
keep the suppression to give a more accurate value for the
total storage available.
(usage): Expand on the new implications of --total and move
it in the options list according to alphabetic order.
doc/coreutils.texi (df invocation): Mention that --total impacts
on deduplication of remote file systems and also move location
according to alphabetic order.
* tests/df/skip-duplicates.sh: Add remote test cases.
* NEWS: Mention the change in behavior.
Reported in http://bugs.debian.org/737399
Reported in http://bugzilla.redhat.com/920806
Reported in http://bugzilla.opensuse.org/866010
Reported in http://bugzilla.opensuse.org/901905
commit v8.22-125-g9d736f8 printed placeholder "-" values
for device names that didn't match the preferred device name
for a particular mount point. However that was seen to erroneously
suppress values for aliased host names or exports, common with
remote file systems.
* src/df.c (me_for_dev): Rename from devname_for_dev() so that
we can determine the remoteness as well as the name for the
preferred mount entry.
(get_dev): Don't output place holder values when both
current and preferred mount entries are remote.
Reported in http://bugs.debian.org/737399
* NEWS: Update the recent entry to also mention the avoidance
of incorrectly unlinking a multi-hardlinked "source" file when
presented with source and dest that only differ in case.
* src/copy.c (same_file_ok): Mention the case issue with same_name().
* tests/mv/hardlink-case.sh: Test the issue on HFS+.
* tests/local.mk: Reference the new test case.
* tests/mv/vfat: Remove an old related but unused test case.
file_t is now mapped to unlabeled_t as per:
http://danwalsh.livejournal.com/68189.html
Therefore use the latter to ensure we match correctly.
This is needed on >= Fedora 21 for example,
while it also works on earlier releases.
We may run into a race condition if we treat hard links to the same file
as distinct files. If we do 'mv a b' and 'mv b a' in parallel, both a
and b can disappear from the file system. The reason is that in this
case the unlink on src is called and the system calls can end up being
run in the order where unlink(a) and unlink(b) are the last two system
calls. Therefore exit with an error code so that we avoid the potential
data loss.
* src/copy.c (same_file_ok): Don't set unlink_src that was used by mv,
and return false for two hardlinks to a file in move_mode.
*src/copy.c (copy_internal): No longer honor the unlink_src option,
used only by mv.
NEWS: Mention the change in behavior.
* tests/cp/same-file.sh: Augment to cover the `cp -a hlsl1 sl1` case.
* tests/mv/hard-verbose.sh: Remove no longer needed test.
* tests/local.mk: Remove the reference to hard-verbose.sh.
* tests/mv/hard-4.sh: Adjust so we fail in this case.
* tests/mv/i-4.sh: Likewise.
* tests/mv/symlink-onto-hardlink-to-self.sh: Likewise.
* tests/tail-2/inotify-rotate.sh: Use retry_delay_
to employ an exponential backoff with a total delay of
up to 25.5s. The 15s delay was seen to trigger a false
failure in http://hydra.nixos.org/build/16546517
Also remove the .1s sleep in each of the 50 iterations
to reduce the running time of the test and thus the
expensive_ tag on this test was removed.
Also ensure that we use the standard exit procedure
upon failure to avoid any erroneous diagnostics due
to persistent files on NFS.
* src/df.c (usage): Mention that duplicate file systems are shown
with this option, not just dummy file systems.
* doc/coreutils.texi (df invocation): For the --all option, expand
on the class of normally suppressed mount entries that it includes.
Reported in http://bugs.debian.org/737399
sc_long_lines was the slowest syntax check
before$ time make sc_long_lines
long_lines
real 0m2.740s
after $ time make sc_long_lines
long_lines
real 0m0.677s
* src/cfg.mk (sc_dd_max_sym_length): s/--max-line-length/-L/
for compat with BSDs.
(sc_long_lines): Prefilter with wc -L to only identify lines
in files that have lines longer than 80 characters.
* tests/d_type-check: The hardcoded name doesn't hold true for all
Linux/glibc platforms, let alone Linux/non-glibc.
Use ctypes.util.find_library() instead to search for the library.
At least the MHz number in /proc/cpuinfo may change, thus leading to
a false positive failure when comparing the expected against the
actual output file. Use an invariant file instead: /proc/version.
* tests/misc/head-c.sh: s/cpuinfo/version/
* src/chroot.c (is_root): Adjust to compare canonicalized paths
rather than inodes, to handle (return false in) the case where
we have a tree that is constructed by first bind mounting "/"
(thus having the same inode).
(main): Unconditionally call chroot() because it's safer
and of minimal performance benefit to avoid in this case.
This will cause inconsistency with some platforms
not allowing `chroot / true` for non root users.
* tests/misc/chroot-fail.sh: Adjust appropriately.
* NEWS: Mention the bug fixes.
Fixes http://bugs.gnu.org/18736
With --sparse=always use fallocate(...PUNCH_HOLE...) to
avoid any permanent allocation due to speculative
preallocation employed by file systems such as XFS.
* m4/jm-macros.m4: Check for <linux/falloc.h> and fallocate().
* src/copy.c (punch_hole): A new function to try and punch
a hole at the specified offset if supported.
(create_hole): Call punch_hole() after requesting a hole.
(extent_copy): Likewise.
* NEWS: Mention the improvement.
Previously cp would not detect runs of NULs that were
smaller than the buffer size used for I/O (currently 128KiB).
* src/copy.c (copy_reg): Use an independent hole_size, set to
st_blksize, to increase the chances of detecting a representable hole,
in a run of NULs read from the input.
(create_hole): A new function refactored from sparse_copy() and
extent_copy() so we have a single place to handle holes.
(sparse_copy): Adjust to loop over the larger input buffer
in chunks of the passed hole size. Also adjust to only call
lseek once per hole, rather than at least once per input buffer.
* tests/cp/sparse.sh: Add test cases for various sparse chunk sizes.
* NEWS: Mention the improvement.
Prompted by http://hydra.nixos.org/build/15682577
with GCC 4.8.3 on i686
src/tac.c:557:6: error: assuming signed overflow does not occur
when simplifying conditional to constant [-Werror=strict-overflow]
if (bytes_copied < 0)
This happens because copy_to_temp() is inlined in tac_nonseekable(),
thus reducing the comparison to the bytes_copied variable in
copy_to_temp. Now this can't overflow on either 32 or 64 bit
due to the protection of the preceding fwrite(). We could use a
guard like "if (bytes_copied <= OFF_T_MAX - bytes_read)" to avoid
the warning, but rather than a runtime branch, just use an unsigned
type to avoid apparent signed overflow on systems where the accumulation
is not promoted to unsigned (32 bit size_t, 64 bit off_t).
* src/tac.c (copy_to_temp): Increment an unsigned type to
avoid the subsequent signed overflow warning.
Fix similar problems in head, od, split, tac, and tail.
Reported by George Shuklin in: http://bugs.gnu.org/18621
* NEWS: Document this.
* src/head.c (elseek): Move up.
(elide_tail_bytes_pipe, elide_tail_lines_pipe): New arg
CURRENT_POS. All uses changed.
(elide_tail_bytes_file, elide_tail_lines_file):
New arg ST and remove arg SIZE. All uses changed.
* src/head.c (elide_tail_bytes_file):
* src/od.c (skip): Avoid optimization for /sys files, where
st_size is bogus and st_size == st_blksize.
Don't report error at EOF when not optimizing.
* src/head.c, src/od.c, src/tail.c: Include "stat-size.h".
* src/split.c (input_file_size): New function.
(bytes_split, lines_chunk_split, bytes_chunk_extract): New arg
INITIAL_READ. All uses changed. Use it to double-check st_size.
* src/tac.c (tac_seekable): New arg FILE_POS. All uses changed.
(copy_to_temp): Return size of temp file. All uses changed.
* src/tac.c (tac_seekable):
* src/tail.c (tail_bytes):
* src/wc.c (wc):
Don't trust st_size; double-check by reading.
* src/wc.c (wc): New arg CURRENT_POS. All uses changed.
* tests/local.mk (all_tests): Add tests/misc/wc-proc.sh,
tests/misc/od-j.sh, tests/tail-2/tail-c.sh.
* tests/misc/head-c.sh:
* tests/misc/tac-2-nonseekable.sh:
* tests/split/b-chunk.sh:
Add tests for problems with /proc and /sys files.
* tests/misc/od-j.sh, tests/misc/wc-proc.sh, tests/tail-2/tail-c.sh:
New files.
* doc/coreutils.texi (stat invocation): Add a paragraph documenting
stat's output format when the --terse option is specified, both in
normal and in --file-system mode.
Reported by Dan Jacobson <jidanni@jidanni.org>
in http://bugs.gnu.org/18624
* init.cfg (gcc_shared_): -ldl has to be positioned after the object
files that may rely upon it. This fixes tests/cp/nfs-removal-race.sh
which references dlsym() from libdl.
