This change is performance neutral as tested with:
$ yes $(yes eeeaae | head -n10K | paste -s -d,) | head -n10K > ll.in
$ hyperfine src/cut{,.orig}' -f10 -w ll.in >/dev/null'
* gl/lib/mbbuf.h (mbbuf_fill): Repeat the mbbuf_topup(), i.e. the read()
if the previous run didn't read enough. Note we add the explicit
common case short-circuit to avoid a 20% regression
for the test case above, with GCC 16.2 with -O3 -flto.
* tests/cut/mb-short-read.sh: A new test file.
* tests/local.mk: Reference the new test.
* NEWS: Mention the bug fix.
Discovered by Anthropic's OSS Scanner, as bug ANT-2026-JSCMJCN2.
This was a mistake in my commit 16a673638 (fold: move multi-byte
character reading to a module, 2025-09-28), which added the file. The
correct license is in the module description and the header file, which
contains the module's implementation.
* gl/lib/mbbuf.c: Change license header to LGPLv2+, matching the
associated header and module description.
configure accepted (and needed) both --without-libsmack
and --disable-libsmack specified, to fully disable support.
* gl/modules/smack: Require gl_FUNC_ACL_ARG so we
use a consistent and single --without-libsmack option.
This also better aligns with --without-selinux.
* NEWS: Mention the Build-related change.
* bootstrap.conf: Reference the new "utf8" module.
* cfg.mk: Adjust checks as per mbbuf.
* gl/lib/utf8.c: Include the implementation.
* gl/lib/utf8.h: Add is_utf8_charset() and u8_count().
Note we don't give is_utf8_charset external-linkage
due to its static cache.
* gl/local.mk: Reference the new module files.
* gl/modules/utf8: Add the module definition and dependencies.
* src/cut.c: Use is_utf8_charset() and u8_count().
* src/numfmt.c: Use is_utf8_charset().
* src/system.h: Remove is_utf8_charset() definition.
Change some uses of unreachable() to affirm(false),
so that we would get an abort in normal builds.
This is safer than unreachable() which results in undefined behavior.
See commit v9.8-27-ge661c7a52 for a real case of this.
Uses of unreachable() should be provably unreachable, and trivially so.
Trivial cases: e.g. after error(NON_ZERO, ...) etc,
or a fully handled switch on enum (proved with -Wswitch-enum).
Note unreachable() can be defined to exit with -funreachable-traps,
but it's better to use standard build options and distinguish the
different classes of "unreachable" in code. Also clang (22) doesn't
seem to support the -funreachable-traps option.
Note we prefer affirm to abort, since it has better diagnostics in the
normal case and better performance with -DNDEBUG.
and prefer affirm to assert, as it allows for better static
checking when compiling with -DNDEBUG.
Neither of these apply to affirm(false),
but we avoid assert for consistency.
* gl/lib/mbbuf.h: s/unreachable/affirm(false)/.
* src/copy-file-data.c: Likewise.
* src/copy.c: Likewise.
* src/cut.c: Likewise.
* src/ls.c: Likewise.
* src/numfmt.c: Likewise.
* src/od.c: Likewise.
* src/stdbuf.c: Likewise.
* src/sync.c: Likewise.
* src/tr.c: Likewise.
* src/tsort.c: Likewise.
* src/wc.c: Likewise.
* gl/lib/mbbuf.h: Adjust mbbuf_fill() to process full characters
in the slop at the end of a read(). Previously valid characters
in the last MCEL_LEN_MAX bytes were ignored until the next read().
* src/cut.c (cut_fields_bytesearch): Adjust to the new naming.
* NEWS: Mention the fold(1) responsiveness fix, which was
improved with the change from fread() to read(),
and completed with this patch.
* gl/lib/mbbuf.h (fill_buf): Switch from fread() to read()
as the former retries read() internally to fill the buffer.
* src/cut.c: Adjust accordingly, and avoid getc() interface entirely.
* bootstrap.h: Depend explicitly on fseterr. This is already depended
on transitively, so should not introduce new build portability issues.
We can only byte search with uni-byte or utf-8.
utf-8 implicitly can't false match a delimiter at buffer boundary.
So don't worry about finding the exact utf8 boundary at end of buffer,
rather just ensuring the buffer always starts with a valid character
(by ensuring MCEL_LEN_MAX-1 moved to start of buffer on each refill).
* gl/lib/mbbuf.h: To safely search a bounded buffer,
without needed to have unbounded memory with getndelim2.
TODO: rename mbbuf_utf8_safe_prefix to mbbuf_fill_utf8
The explicit_bzero function is a common extension, but memset_explicit
was standardized in C23. It will likely become more portable in the
future, and Gnulib provides an implementation if needed.
* bootstrap.conf (gnulib_modules): Add memset_explicit. Remove
explicit_bzero.
* gl/lib/randint.c (randint_free): Use memset_explicit instead of
explicit_bzero.
* gl/lib/randread.c (randread_free_body): Likewise.
On signed char platforms, 0xFF was converted to -1
which matches MBBUF_EOF, causing fold to stop processing.
* NEWS: Mention the bug fix.
* gl/lib/mbbuf.h: Avoid sign extension on signed char platforms.
* tests/fold/fold-characters.sh: Adjust test case.
Reported at https://src.fedoraproject.org/rpms/coreutils/pull-request/20
Make this change temporarily in coreutils.
Move to gnulib after v9.10 is released.
* top/maint.mk (web-manual): Propagate MAKEINFO explicitly,
rather than requiring it exported in the environment,
which can be awkward on Solaris make for example.
* gl/modules/mbbuf: New file.
* gl/lib/mbbuf.c: Likewise.
* gl/lib/mbbuf.h: Likewise.
* gl/local.mk (EXTRA_DIST): Add the new files.
* bootstrap.conf (gnulib_modules): Add mbbuf.
* src/fold.c: Include mbbuf.h.
(fold_file): Use the mbbuf functions instead of calling fread and
handling the input buffer ourselves.
* cfg.mk (exclude_file_name_regexp--sc_preprocessor_indentation)
(exclude_file_name_regexp--sc_GPL_version): Match gl/lib/mbbuf.c and
gl/lib/mbbuf.h.
Use new Gnulib streq function instead of rolling our own macro.
* bootstrap.conf (gnulib_modules): Add stringeq.
* src/rm.c (main): Don’t assume streq is a macro that expands to (...),
as it is now a function.
* src/system.h:
* tests/df/no-mtab-status.sh, tests/df/skip-duplicates.sh:
(STREQ): Remove. All uses replaced by streq.
Recent changes in Gnulib guarantee O_DIRECTORY != 0.
* gl/lib/targetdir.c (target_directory_operand):
* src/dd.c (usage):
* src/ln.c (main):
Simplify now that O_DIRECTORY must be nonzero.
Also, change sleep and tail to not sleep less than requested.
* bootstrap.conf (gnulib_modules): Add dtimespec-bound.
* gl/lib/dtimespec-bound.c, gl/lib/dtimespec-bound.h:
* gl/modules/dtimespec-bound: New files.
* src/sleep.c, src/tail.c, src/timeout.c: Include dtimespec-bound.h.
* src/sleep.c, src/tail.c: Don’t include xstrtod.h.
* src/sleep.c (apply_suffix, main):
* src/tail.c (parse_options):
* src/timeout.c (apply_time_suffix):
Don’t sleep less than the true number of seconds.
* src/timeout.c: Don’t include ctype.h.
(is_negative): Remove; no longer needed.
(parse_duration): Use a slightly looser bound on the timeout, one
that doesn’t need -lm on GNU/Linux. Clear errno before calling
cl_strtod.
Update to latest gnulib with new copyright year.
Run "make update-copyright" and then...
* gnulib: Update included in this commit as copyright years
are the only change from the previous gnulib commit.
* tests/init.sh: Sync with gnulib to pick up copyright year.
* bootstrap: Likewise.
* tests/sample-test: Adjust to use the single most recent year.
* gl/lib/strnumcmp-in.h (ISDIGIT):
* src/system.h (ISDIGIT): Remove. All uses replaced by c_isdigit,
with appropriate inclusions of c-ctype.h. This is more regular,
and is more portable to existing (but unlikely) platforms where
INT_MAX == UINT_MAX.
* gl/lib/heap.c (struct heap, heap_alloc, heap_insert)
(heapify_down, heapify_up): Prefer idx_t to size_t for sizes.
(heap_insert): Use xpalloc instead of x2nrealloc.
(heapify_down): Return void since no caller cares about value.
* gl/modules/heap: Depend on idx.
* cfg.mk (sc_gldist_missing): Add a new target to ensure we don't
forget to distribute any new gl/ files.
* gl/local.mk: Remove generation comment since it's
now encapsulated in the syntax-check, which outputs a consumable
diff to make any future adjustments.
Also adjust ordering to that of the C locale used in the syntax check.
This issue was noticed with -flto on GCC 14.2.1
* gl/lib/xdectoint.c (__xnumtoint): Only inspect the
returned value if LONGINT_INVALID is not set,
as the returned value is uninitialized in that case.
Fixes https://bugs.gnu.org/72842
Also, prepare for allowing some arguments to overflow
without that being an error.
* gl/lib/xdectoint.c: Do not include stddef.h,
since we no longer use ‘unreachable’.
(xnumtoimax, xnumtoumax, __xnumtoint):
New arg FLAGS. All callers changed.
Stop using __xdectoint_signed. All definers removed.
* gl/lib/xdectoint.h (XTOINT_MIN_QUIET, XTOINT_MAX_QUIET)
(XTOINT_MIN_RANGE, XTOINT_MAX_RANGE): New flag constants.
* src/fmt.c (main):
* src/fold.c (main):
* src/nl.c (main):
* src/pr.c (getoptnum):
* src/split.c (main):
Use XTOINT_MIN_RANGE and XTOINT_MAX_RANGE if appropriate.
* src/pr.c (getoptnum): Return int rather than returning void
and storing through int *.
* src/stty.c (apply_settings):
Use ckd_add to check for overflow instead of doing it by hand.
(integer_arg): Accept and return uintmax_t, not unsigned long.
* gl/lib/randperm.c: Include <stdckdint.h>.
(randperm_bound): Return SIZE_MAX if the multiplication overflows.
Do not overflow when converting bit count to byte count.
Problem reported by Daniel Carpenter <https://bugs.gnu.org/72445>.
* gl/lib/randread.c (randread_new): Fill the ISAAC buffer
instead of storing at most BYTES_BOUND bytes into it.
<stdbit.h> is in C23 and should be more portable in the long run,
now that Gnulib supports it.
* bootstrap.conf (gnulib_modules): Remove count-leading-zeros.
Add stdc_leading_zeros.
* gl/lib/randperm.c, src/ioblksize.h:
Include stdbit.h instead of count-leading-zeros.h.
* gl/lib/randperm.c (floor_lg): Remove; no longer needed.
(randperm_bound): Use stdc_bit_width instead of floor_lg;
* gl/modules/randperm (Depends-on): Remove count-leading-zeros.
Add stdc_bit_width.
* src/ioblksize.h (io_blksize): Use stdc_leading_zeros_ull
instead of count_leading_zeros_ll.
* gl/lib/fadvise.c: Remove pragma that works around GCC bug 83559
<https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83559>.
This bug was fixed in GCC 9, and we needn’t worry about
--enable-gcc-warnings for compilers that old.
* src/test.c: Likewise.
Update to latest gnulib with new copyright year.
Run "make update-copyright" and then...
* gnulib: Update included in this commit as copyright years
are the only change from the previous gnulib commit.
* tests/init.sh: Sync with gnulib to pick up copyright year.
* bootstrap: Manually update copyright year,
until we fully sync with gnulib at a later stage.
* tests/sample-test: Adjust to use the single most recent year.
* gl/lib/strnumcmp-in.h (numcompare): After commit v9.0-8-g6cafb122f,
we need to treat characters as signed to avoid invalid comparisons
between negative integers and unsigned characters.
* NEWS: Mention the bug fix.
* NEWS: Mention this.
* bootstrap.conf (gnulib_modules): Remove cu-ctype, as this module
is now more trouble than it’s worth. All uses removed.
Add skipchars.
* gl/lib/cu-ctype.c, gl/lib/cu-ctype.h, gl/modules/cu-ctype:
Remove.
* gl/lib/skipchars.c, gl/lib/skipchars.h, gl/modules/skipchars:
* tests/misc/join-utf8.sh:
New files.
* src/join.c: Include skipchars.h and mcel.h instead of cu-ctype.h.
(tab): Now mcel_t, not int. All uses changed.
(output_separator, output_seplen): New static vars.
(eq_tab, newline_or_blank, comma_or_blank): New functions.
(xfields, prfields, prjoin, add_field_list, main):
Support multi-byte characters.
* src/numfmt.c: Include ctype.h, skipchars.h.
Do not include cu-ctype.h.
(newline_or_blank): New function.
(next_field): Support multi-byte characters.
* src/sort.c: Include ctype.h instead of cu-ctype.h.
(inittables): Open-code field_sep since it no longer exists.
‘sort’ is not multi-byte safe yet, but when it is this code
will need revamping anyway.
* src/uniq.c: Include mcel.h and skipchars.h instead of cu-ctype.h.
(newline_or_blank): New function.
(find_field): Support multi-byte characters.
* tests/local.mk (all_tests): Add tests/misc/join-utf8.sh
This is so that we don’t need to have every source file
include ctype.h.
* bootstrap.conf (gnulib_modules): Add cu-ctype.
* gl/lib/cu-ctype.c, gl/lib/cu-ctype.h, gl/modules/cu-ctype:
New files.
* src/join.c, src/numfmt.c, src/sort.c, src/uniq.c:
Include cu-ctype.h, for field_sep.
* src/system.h (field_sep): Remove; now supplied by cu-ctype.
This simplifies memory allocation a bit, and removes an arbitrary
limitation from numfmt, which formerly limited cell output to 127
bytes.
* bootstrap.conf (gnulib_modules): Remove mbsalign, strncat.
Add strnlen (the code already used strnlen directly, and we were
saved only because Gnulib used the module indirectly)
* gl/lib/mbsalign.c, gl/lib/mbsalign.h, gl/modules/mbsalign:
* gl/modules/mbsalign-tests, gl/tests/test-mbsalign.c: Remove.
* src/df.c, src/ls.c: Do not include mbsalign.h.
(MBSWIDTH_FLAGS): New constant, now used for all
mbswidth calls. All callers changed to check for -1 return.
* src/df.c (struct field_data_t): ‘width’ is now int not size_t,
since mbswidth can’t do widths greater than INT_MAX anyway.
Replace ‘align’ with ‘align_right’. All uses changed.
(print_table): Redo to avoid the need for ambsalign.
(get_header, get_dev): mbswidth returns int, not size_t.
* src/ls.c (MAX_MON_WIDTH): Remove; no longer used.
(abmon_init): Use strnlen to cheaply discard too-long month names.
Align by hand instead of using mbsalign.
* src/numfmt.c: Include stdckdint.h, mbswidth.h.
Do not include mbsalign.h.
(padding_buffer_size): Now idx_t. All uses changed.
(padding_width): Now intmax_t, since it’s no longer an object
size. Its sign now records alignment. All uses changed.
(zero_padding_width): Now int, since it’s given to sprintf.
All uses changed.
(padding_alignment): Remove; it’s now taken from padding_width’s sign.
(double_to_human): Return string length. BUF_SIZE arg is now idx_t.
Include suffix in output. All callers changed. Simplify by not
calling strncat or stpcpy. Calculate fmt size bound more carefully.
(setup_padding_buffer): Remove. All uses removed.
(parse_format_string): Use intmax_t, not long, for pad.
On overflow, set widths to large values that cause later code
to do the right thing, rather than separately checking for
overflow here.
(prepare_padded_number): Return bool, not int 0/1. New arg
PADDING. All uses changed. Do not limit padded output to 127
bytes; instead, use xpalloc to expand the output buffer.
(print_padded_number): New arg PADDING. All uses changed.
(process_suffixed_number): Simplify.
(main): Take extremum if xstrtoimax overflows, as this does
the right thing.
* tests/misc/numfmt.pl: New test suf-20 to test for truncation bug.
Remove tests pad-3.2, fmt-err-7, as they’re no longer invalid but
are quite expensive.