Age | Commit message (Collapse) | Author |
|
Running a command like this on a multi-core system
sort < big-file | less
would peg all processors at near 100% utilization.
* src/sort.c: (struct merge_node) Change member lock to mutex.
All uses changed.
* tests/Makefile.am (XFAIL_TESTS): Remove definition, now that
this test passes once again. I.e., the sort-spinlock-abuse test
no longer fails.
* NEWS (Bug reports): Mention this.
Reported by DJ Lucas in http://debbugs.gnu.org/7489.
|
|
* src/sort.c (struct thread_args, sortlines_thread, sortlines, sort):
Rename "merge_queue" to "queue", for consistency with other functions
that just use the name "queue" for these things.
|
|
* src/sort.c (queue_check_insert): Clarify body a bit, and remove
no-longer-needed comment.
|
|
* src/sort.c (mergelines_node): Return void, not size_t. All
callers changed. Change *node->dest here, not in caller.
Do not change node->dest: it's not needed and could cause problems
on (mostly theoretical) hosts that do not allow adding integers to
null pointers.
(queue_check_insert_parent): Omit MERGED parameter; no longer needed.
All callers changed.
|
|
* src/sort.c (write_unique): Simplify slightly so that there is
just one call to write_line, not two.
|
|
* src/sort.c (queue_check_insert, queue_check_insert_parent): Make
the queue arg first, for consistency with other functions such as
queue_insert that put the queue arg first. Rename from
check_insert and update_parent, respectively. All callers
changed.
|
|
* src/sort.c (struct merge_node): 'lock' is now the actual lock,
not a pointer to the lock; there's no need for indirection here.
Make 'level' unsigned int instead of size_t, since it is a
bit-shift count; also, move it next to a bool so that it's more
likely to take less space. All uses changed.
(sortlines, sort): Spell out initialization instead of using an
initializer. This makes the initializer a bit easier to understand,
and avoids unnecessary stores into the spin lock.
|
|
* src/sort.c: Improve comments a bit.
|
|
* src/sort.c (MAX_MERGE): Avoid integer overflow when on a machine
with (say) 32-bit int and 64-bit size_t and when level == 15.
Without this fix, on such a machine with 32768 or more processors,
the level computation could overflow on large input, and this
would result in division by zero.
|
|
* src/sort.c (write_unique): Save the entire "struct line", not
just a pointer to one. Otherwise, with a multi-thread run,
sometimes, with some inputs, fillbuf would would win a race
and clobber a "saved->text" pointer in one thread just before
it was dereferenced in a comparison in another thread.
* NEWS (Bug fixes): Mention it.
|
|
obsolete keys
* src/sort.c (key_warnings): Local buffer should be of size
INT_BUFSIZE_BOUND (uintmax_t), not INT_BUFSIZE_BOUND (sword).
This bug was discovered by running 'make check' on a 32-bit
Solaris 8 sparc host, using Sun cc. I saw several other instances
of invoking umaxtostr on a buffer declared to be of size
INT_BUFSIZE_BOUND (VAR), and these instances should at some point
be replaced by INT_BUFSIZE_BOUND (uintmax_t) too, as that's a
less error-prone style.
|
|
* src/sort.c (fold_toupper): Change this back from char to
unsigned char, fixing a portability issue introduced in commit
59e2e55d0f154a388adc9bac37d2b45f2ba971f8 dated February 26, as the
C Standard doesn't let you convert from unsigned char * to char *
without a cast, and the (in theory more portable) style here is to
convert char values, not pointer values.
(getmonth): Convert char to unsigned char when needed for
comparison.
|
|
* src/sort.c (sortlines, sort): Use pthread_spin_destroy when a
spin lock is no longer used. This isn't needed on GNU/Linux or
Solaris, but POSIX says it may free up resources on some platforms.
|
|
This change was prompted by the previous one: I audited the code
looking for similar examples. Too bad valgrind doesn't catch this.
* src/sort.c (check, mergefps): xrealloc -> free + xmalloc
* src/who.c (print_user): Likewise.
|
|
* src/sort.c (compare_random): Use free/xmalloc rather than
xrealloc, since the old buffer contents need not be preserved.
Also, don't fail if the guessed-sized malloc fails. Suggested by
Bruno Haible.
|
|
* src/sort.c: Include "ignore-value.h".
(debug_key): Use ignore_value.
|
|
* src/sort.c (compare_random): Guess that the output will be
3X the input. This avoids the overhead of calling strxfrm
twice on typical implementations. Suggested by Bruno Haible.
|
|
* NEWS: Document this.
* src/sort.c (getmonth): Omit LEN arg, as MONTH is now null-terminated.
(compare_random): Don't null-terminate keys, as caller now does that.
(compare_version): Remove.
(debug_key): Null-terminate string for getmonth.
(keycompare): Support combining -R with any of -d, -f, -i, -V.
Also, support combining -V with any of -d, -i.
(check_ordering_compatibility): Allow newly-supported combinations.
* tests/misc/sort (02q, 02r, 02s): New tests, for new combinations.
(incompat2): Now test -nR, since -fR are now compatible.
|
|
Formerly, the 'compare' function and some of its subroutines had a
debugging flag, which caused them to output underlines. This
change refactors the code so that debugging output is
more-separated from the actual sorting. In the process, the
change fixes a minor error in the debugging output. The change
shortens the source code and executable size a tad, and improves
CPU performance by 2.4% on my platform with a simple benchmark (C
locale, line sorting, no debug).
* src/sort.c (long_double, strtold): Move back to prelude, since
they're now used by multiple functions again.
(unit_order): Move to file scope, since it's now used by two functions.
(find_unit_order, human_numcompare, numcompare, general_numcompare):
Remove endptr parameter. All callers changed.
(human_numcompare): Args are now const pointers.
(getmonth): Endptr is now non-const.
(key_numeric): Move up, since it's needed earlier.
(debug_key): Take a line and a key as argument, instead of having
the caller figure out where the field is.
(debug_line): New function.
(keycompare, compare): Omit debug parameter; debug output now done
elsewhere. All callers changed.
(write_line): Renamed from write_bytes; all callers changed.
Use debug_line (not 'compare') to output debug info.
Use a slightly faster check for whether output file is stdout.
(check): Don't do debugging output; it's not that useful here,
and it confuses the code.
(main): Check for incompatibility between -c and --debug.
Use standard diagnostic for incompatible options.
* tests/misc/sort-debug-keys: Fix test case: "--Mi-1" is not
a number, so its first character should not be underlined when
debugging a numeric sort.
|
|
* lib/Makefile.am (libcoreutils_a_SOURCES): Remove xmemxfrm.c,
xmemxfrm.h.
* lib/memxfrm.c, lib/memxfrm.h, lib/xmemxfrm.c, lib/xmemxfrm.h: Remove.
* m4/memxfrm.m4: Likewise.
* m4/prereq.m4 (gl_PREREQ): Remove gl_MEMXFRM.
* po/POTFILES.in: Remove lib/xmemxfrm.c.
* src/sort.c: Don't include xmemxfrm.h.
(cmp_hashes): Remove.
(xstrxfrm): New function.
(compare_random): If a line contains NULs, don't create a big
buffer that contains the strxfrm output of each string in the line.
Instead, accumulate checksums and differences as we go, so that
at any one time we have to store at most the output of a single
strxfrm call when processing the line. This removes the need for
an memxfrm function.
|
|
* src/sort.c (debug_width): New function, which does not stop
counting tabs at \0, and also invokes mbsnwidth. Stamp out strnlen!
(count_tabs): Remove.
(debug_key): Use debug_width instead of mbsnwidth and count_tabs.
* tests/misc/sort-debug-keys: Check that \0 and \t intermix.
|
|
* NEWS: Document changes to sort -h, which are now minor with
respect to the pre-July-30th version.
* doc/coreutils.texi (sort invocation): Likewise. The
documentation now describes how -h comparison is done rather than
being vague with border cases.
* src/sort.c (long_double, strtold): Move back to general_numcompare.
(LD, compute_human): Remove.
(find_unit_order): Remove THOU_SEP parameter, since thousands
separators are now allowed by all callers. Revert to previous
behavior of sorting by suffix, and returning the order rather than
2 * order + binary, since we no longer care whether binary powers
are being used. However, treat all zeros the same, instead of
sorting 0M before 0G; this is more consistent with the desired
behavior of sorting -1G before -1M.
* tests/misc/sort (h1, h3, h6): Adjust to match mostly-reverted
behavior. However, check that all zeros sort together.
* tests/misc/sort-debug-keys: Omit a "_", since the trailing "i"
in "1234Gi" is no longer part of the key.
|
|
* NEWS: Document changes to sort -h.
* doc/coreutils.texi (sort invocation): Likewise.
* src/sort.c (long_double, strtold): Move to prelude, since they're
now used by multiple functions.
(LD): New macro.
(struct keyfield.iec_present): Remove this member. All uses removed.
(check_mixed_SI_IEC): Remove. This code was busted in the presence
of multiple threads, as it had a race condition.
(find_unit_order): Remove arg KEY; add arg THOU_SEP; arg ENDPTR is
now char ** rather than char const **. Return an integer that
distinguishes decimal from binary powers. Parse the number
consistently with the intersection of strtold and strnumcmp.
Set *ENDPTR unconditionally.
(compute_human): New static function.
(human_numcompare): Remove arg KEY. Remove 'const' from other args.
Use strnumcmp if possible, but fall back on floating point if not.
(numcompare, general_numcompare): Arg EA is now char ** rather
than char const **.
(numcompare): Adjust to new find_unit_order signature and behavior.
(keycompare): Adjus to new human_numcompare signature.
* tests/misc/sort (h1, h3, h4, h6): Adjust to new behavior.
* tests/misc/sort-debug-keys: Likewise.
|
|
* src/sort.c (mark_key): Don't assume offset <= INT_MAX.
Make the code a bit clearer when width != 0.
|
|
* src/sort.c (fillbuf): Don't append eol unless the line is nonempty.
This fixes a bug that was partly but not completely fixed by
the aadc67dfdb47f28bb8d1fa5e0fe0f52e2a8c51bf commit (dated July 15).
* tests/misc/sort (realloc-buf-2): New test, which catches this
bug on 64-bit hosts.
|
|
* src/sort.c (mergelines, queue_destroy, queue_init, queue_insert):
(queue_pop, write_unique, mergelines_node, check_insert):
(update_parent): No longer inline; these uses of "inline"
seemed unlikely to help performance much.
|
|
* src/sort.c (find_unit_order): Don't assume ASCII.
|
|
* src/sort.c (queue_pop): Omit unnecessary unlock+lock after
pthread_cond_wait returns. Don't access "count" member of the
heap; any efficiency gains should be quite minor, the access
complicates this code, and "count" should be private anyway.
|
|
* src/sort.c (lock_node, unlock_node, queue_destroy, queue_init):
(queue_pop):
Omit 'restrict'; it shouldn't help here, as these functions have just
one pointer parameter and don't access static storage.
(queue_insert, check_insert, update_parent): Omit 'restrict', as
the pointer types differ, and are not char * or unsigned char *,
and therefore can't alias.
(write_unique): Omit 'restrict', as the pointer types are all
read-only.
(merge_loop, sortlines): Omit 'restrict', as any performance
advantages are extremely unlikely and it's not worth cluttering
the code for that.
(struct thread_args): Omit 'restrict': this seems to be incorrect.
It's unlikely for 'restrict' to be correct inside a typedef.
|
|
* src/sort.c (inittables, general_numcompare, compare_nodes):
(queue_init, queue_pop): Omit casts that are not needed, typically
because they are between void * and some other pointer type.
|
|
* src/sort.c (proctab_hasher, proctab_comparator, stream_open, xfopen):
(open_temp, zaptemp, struct_month_cmp, begfield, limfield):
(find_unit_order, human_numcompare, numcompare, general_numcompare):
(count_tabs, keycompare, compare, compare_nodes, lock_node):
(unlock_node, queue_destroy, queue_init, queue_insert, queue_pop):
(write_unique, mergelines_node, check_insert, update_parent):
(merge_loop, sortlines, struct thread_args, set_ordering):
Prefer the style "T const" to "const T".
* gl/lib/heap.h (struct heap, heap_alloc): Likewise.
* gl/lib/heap.c (heap_default_compare, heapify_down, heapify_up):
(heap_alloc): Likewise.
|
|
* bootstrap.conf: Include the new module
* gl/lib/fadvise.c: Provide a simpler interface to posix_fadvise.
(fadvise): Provide hint to the whole file associated with a stream.
(fdadvise): Provide hint to the specific portion of a file
associated with a file descriptor.
* gl/lib/fadvise.h: Redefine POSIX_FADV_* to FADVISE_* enums.
* gl/modules/fadvise: New file.
* m4/jm-macros.m4: Remove the no longer needed posix_fadvise check.
* .x-sc_program_name: Exclude test-fadvise.c from this check.
* gl/tests/test-fadvise (main): New test program.
* gl/modules/fadvise-testss: A new index to reference the tests.
* src/sort.c (stream_open): Use the new interface.
* src/dd.c (iwrite): Likewise.
|
|
* src/sort.c (random_md5_state): New static var.
(random_md5_state_init): New function, to initialize random_md5_state.
(random_state, randread_source): Remove.
(cmp_hashes): Use random_md5_state rather than random_state.
Break ties using memcmp, not by getting more randomness.
If MD5 collisions turn into a problem in practice, we should
simply use a better checksum.
(main): If -R is given, call random_md5_state_init rather than
going single-threaded.
|
|
* src/sort.c (keycompare): Use xmemcoll0, as it avoids
a couple of stores.
(write_bytes): Leave the buffer the way we found it,
as it might be used again for a later comparison,
if -u is used.
|
|
* src/sort.c (fillbuf): The previous commit incorrectly
terminated the buffer when the last line of input
didn't contain a terminating character.
|
|
Don't write NUL after the comparison buffers on each compare,
which increases performance by about 3% for short lines
on a pentium-m with gcc-4.4.1
* src/sort.c: (fillbuf): Delimit input items with NUL.
(write_bytes): Restore the item delimiter char which was
replaced with NUL in fillbuf().
|
|
This patch is by Gene Auyeung, Chris Dickens, Chen Guo, and Mike
Nichols, based off of a patch by Paul Eggert, Glen Lenker, et. al.,
with a basic heap implementation based off of the GDSL heap,
originally by Nicolas Darnis.
The number of sorts done in parallel is limited to the number
of available processors by default, or can be further restricted
with the --parallel option.
On a dual-die, 8 core Intel Xeon, results show sorting with
8 threads is almost 4 times faster than using a single thread.
Timings when sorting a 96MB file:
THREADS TIME (s)
1 5.10
2 2.87
4 1.75
8 1.31
Single threaded sorting has also been improved,
especially for cheaper comparison operations:
COMMAND BEFORE (s) AFTER (s)
sort 8.822 8.716
sort -g 10.336 10.222
sort -n 3.077 2.961
LANG=C sort 2.169 2.066
* bootstrap.conf: Add heap, pthread.
* coreutils.texi (sort): Describe the new --parallel option.
* gl/lib/heap.c: New file. Very basic heap implementation.
* gl/lib/heap.h: New file.
* gl/modules/heap: New file.
* src/Makefile.am: Add LIB_PTHREAD.
* src/sort.c: Include heap.h, nproc.h, pthread.h.
(MAX_MERGE): New macro.
(SUBTHREAD_LINES_HEURISTIC, PARALLEL_OPTION): New constants.
(MERGE_END, MERGE_ROOT): New constants.
(struct merge_node): New struct.
(struct merge_node_queue): New struct.
(sortlines temp): Remove declaration.
(usage, long_options, main): New option, --parallel.
(specify_nthreads): New function.
(mergelines): New signature, to emphasize the fact that the HI area
must be part of the destination. All callers changed.
(sequential_sort): New function, renamed from sortlines. Merge in
the functionality of sortlines_temp.
(compare_nodes): New function.
(lock_node, unlock_node): New functions.
(queue_destroy): New function.
(queue_init): New function.
(queue_insert): New function.
(queue_pop): New function.
(write_unique): New function.
(mergelines_node): New function.
(check_insert): New function.
(update_parent): New function.
(merge_loop): New function.
(sortlines): Rewrite to support and use parallelism, with a new
signature. All callers changed.
(struct thread_args): New struct.
(sortlines_thread): New function.
(sortlines_temp): Remove.
(sort): New argument NTHREADS. All uses changed. Output moved to
mergelines_node.
(main): disable threading if we are sorting at random.
* tests/Makefile.am (TESTS): Add misc/sort-benchmark-random.
* tests/misc/sort-benchmark-random: New file.
Signed-off-by: Pádraig Brady <P@draigBrady.com>
|
|
* src/sort.c (usage): Reference the additional description
of the POS syntax.
|
|
sprintf is relatively heavy-weight.
* src/sort.c (key_warnings): Use umaxtostr and stpcpy rather
than sprintf.
Also, replace each INT_BUFSIZE_BOUND "type_name" argument
with the equivalent variable name. More maintainable that way.
|
|
E.g.,
- size_t desired_width IF_LINT (= 0);
+ size_t desired_width IF_LINT ( = 0);
Use this command:
g grep -l IF_LINT | grep '\.[ch]$' \
| xargs perl -pi -e 's/(IF_LINT \()= /$1 = /'
|
|
Run this command:
git ls-files | grep '\.[ch]$' \
| xargs perl -pi -e 's/for \(;;\)/while (true)/g'
...except for randint.c, which does not include stdbool.h.
In that case, use "while (1)".
* gl/lib/randint.c (randint_genmax): Use "while (1)" for infloops.
* src/cat.c (simple_cat, cat): Use "while (true)" for infloops.
* gl/lib/randread.c (readsource, readisaac): Likewise.
* src/copy.c (copy_reg): Likewise.
* src/csplit.c (record_line_starts, process_regexp): Likewise.
* src/cut.c (set_fields): Likewise.
* src/dd.c (iread, parse_symbols): Likewise.
* src/df.c (find_mount_point, main): Likewise.
* src/du.c (main): Likewise.
* src/expand.c (expand): Likewise.
* src/factor.c (factor_using_division, do_stdin): Likewise.
* src/fmt.c (get_space): Likewise.
* src/ls.c (decode_switches): Likewise.
* src/od.c (main): Likewise.
* src/pr.c (main, read_line): Likewise.
* src/shred.c (dopass, genpattern): Likewise.
* src/sort.c (initbuf, fillbuf, getmonth, keycompare): Likewise.
* src/split.c (bytes_split, lines_split): Likewise.
* src/tac.c (tac_seekable): Likewise.
* src/test.c (and, or): Likewise.
* src/tr.c (squeeze_filter, main): Likewise.
* src/tsort.c (search_item): Likewise.
* src/unexpand.c (unexpand): Likewise.
* src/uniq.c (main): Likewise.
* src/yes.c (main): Likewise.
|
|
* src/sort.c (key_warnings): Always warn about significant leading
blanks when character offsets are specified, unless they key is
possibly a line offset, i.e. of the form -k1.x,1.y. Also suppress
this warning if the user could be sorting right aligned indexes.
|
|
* src/sort.c (mark_key): Add a cast-to-int of a printf field width,
to placate -Wformat.
|
|
* src/sort.c (usage): Mention --debug can output warnings to stderr.
Also split the translatable string to aid translation.
(default_key_compare): A new function refactored from main(),
and now also called from the new key_warnings() function.
(key_to_opts): A new function refactored from incompatible_options(),
and now also called from the new key_warnings() function.
(key_numeric): A new function refactored to test if key is numeric.
(key_warnings): A new function to output warnings to stderr,
about questionable use of various options. Currently it warns
about zero length keys and ineffective global options.
(incompatible_options): Refactor out key_to_opts()
(main): Use key_init() to initialize gkey. Refactor out
default_key_compare(). Call key_warnings() in debug mode.
* doc/coreutils.texi (sort invocation): Mention that warnings
are output by --debug.
* tests/misc/sort-debug-warn: A new test for debug warnings.
* tests/Makefile.am: Reference the new test.
* NEWS: Mention the new feature
|
|
* src/sort (usage): Add description for --debug.
(write_bytes): Pass a line structure so it can subsequently
be passed to compare to highlight the keys when in debug mode.
Also transform TAB and NUL characters written to stdout so
that the highlighting in debug mode aligns correctly.
(human_numcompare): Pass an "endptr" so we can record the extent
of the number matched.
(general_numcompare): Likewise.
(find_unit_order): Likewise.
(getmonth): Likewise.
(numcompare): Likewise. Note we reuse find_unit_order() for this,
which is a good enough approximation, and means we don't need to
change the strnumcmp() interface.
(check_mixed_SI_IEC): Return whether iec_present, so that can be
used to set the "endptr" in find_unit_order. Also make the key
parameter optional, which will be the case from numcompare().
(count_tabs): A new function to determine how much to adjust
the mbswidth() values by (TABs don't have a width).
(mark_key): A new function to output the key highlighting to stdout.
(debug_key): A new function to determine the offset and width
of the key highlighting.
(key_compare): Pass the show_debug parameter so the key highlighting
is only displayed when explicitly called. For each key type, set
the length (lena) and whether leading blanks are auto skipped (skipb)
which are then used by debug_key() to highlight the portion of the
key used in the comparison.
(compare): Pass the show_debug parameter so the key highlighting
is only displayed when explicitly called. Call debug_key() to
highlight the last resort comparison.
(check): Output highlighting for disorder line to stdout.
(main): Process the --debug option and make it mutually exlusive
with the -o option as I don't see it useful there, even potentially
harmful if someone left a --debug in by mistake when updating a file.
Also restricting debug output to stdout, simplifies the logic
for dealing with temporary files.
* doc/coreutils.texi (sort invocation): Describe the --debug option,
and reference it from the --key description.
* tests/misc/sort-debug-keys: A new test for highlighting keys.
* tests/Makefile.am: Reference the new test.
* NEWS: Mention the new feature.
|
|
* src/sort.c (general_numcompare): Don't use long double if strtold
is not available, as it may introduce needless overhead.
|
|
* src/sort.c (general_numcompare): Use long doubles unconditionally,
and strtold when available, to convert numbers with greater range and
precision. Performance was seen to be on par with standard doubles.
* doc/coreutils.texi (sort invocation): Amend the -g description to
mention long double rather than double, and strtold rather than strtod.
* src/getlimits.c (main): Output floating point limits for use in tests.
* tests/misc/sort-float: A new test to ensure sort is using long
doubles when possible, and that locale specific floats are handled.
* tests/Makefile.am: Reference the new test.
* tests/test-lib.sh (getlimits_): Normalize indenting.
* NEWS: Mention the new behaviour.
Reported by Nelson Beebe.
|
|
This regression was introduced in commit 224a69b5, 2009-02-24,
"sort: Fix two bugs with determining the end of field".
The specific regression being that we include 1 field too many when
an end field is specified using obsolescent key syntax (+POS -POS).
* src/sort.c (struct keyfield): Clarify the description of the eword
member, as suggested by Alan Curry.
(main): When processing obsolescent format key specifications,
normalize eword to a zero based count when no specific end char is given
for an end field. This matches what's done when keys are specified with -k.
* tests/misc/sort: Add a few more tests for the obsolescent key formats,
with test 07i being the particular failure addressed by this change.
* THANKS: Add Alan Curry who precisely identified the issue.
* NEWS: Mention the fix.
Reported by Santiago Rodríguez
|
|
* src/copy.c (copy_reg): Likewise.
* src/cut.c (main): Likewise.
* src/dd.c (main): Likewise.
* src/getlimits.c (print_int): Likewise.
* src/join.c (join): Likewise.
* src/pwd.c (logical_getcwd): Likewise.
* src/sort.c (specify_nmerge, mergefps, avoid_trashing_input): Likewise.
(merge): Likewise.
* src/uptime.c (usage): Likewise.
|
|
* src/sort.c: The units containing 'i' are actually IEC not SI.
|