| Commit message (Collapse) | Author | Files | Lines |
|---|
|
function old new delta
get_malloc_cpu_affinity - 76 +76
nproc_main 216 206 -10
process_pid_str 250 206 -44
------------------------------------------------------------------------------
(add/remove: 2/0 grow/shrink: 0/2 up/down: 76/-54) Total: 22 bytes
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
|
|
Commit d6b764116 (make: fix detection of target rules) checked
for target rules before macro assignments. This failed for some
Makefiles generated by autotools because partially defined macros
were expanded while testing for a target rule.
Revert to checking for macro assignments first, but try to detect
if the proposed left hand side of the assignment might form part
of a target rule with an inline command.
Also handle the case where the ';' separator of the inline command
has been obfuscated by putting it in a macro.
Saves 128-160 bytes.
(GitHub pdpmake issues 31, 44)
|
|
GNU make and bmake have different implementations for := macro
assignment. In POSIX 202X these are supported by the forms ::=
and :::= respectively. Only the former was supported on the
pdpmake command line.
Add the required support in process_macros() and update the usage
message.
Adds 48-64 bytes.
|
|
The test 'Return error if command fails' needed an exit code of 2.
POSIX only requires an exit code greater than 0.
|
|
The -q option returns an exit status to indicate if targets are
up-to-date (0) or in need of updating (1) but without updating
them. As an exception (imported from GNU make) build commands
with a '+' prefix are executed.
pdpmake didn't implement the exception. Doing so required moving
handling of the -q option down into docmds().
Saves 48 bytes.
|
|
The -t option (in general) causes targets to be touched instead of
having build commands run to create them. There were two problems.
The flag variable 'ssilent' in docmds was too small (uint8_t) to
contain the value of 'dotouch' (uint32_t). Truncation of the
value resulted in build commands being echoed when they shouldn't
have been.
The POSIX specification is unclear as to how build commands with
a '+' prefix interact with touch. The rationale indicates that
this feature was imported from GNU make, so the behaviour has been
made to match what it does: if a '+' build command is run the
target is not touched.
The code has been rearranged to move the call to touch() up into
docmds().
Adds 48 bytes.
|
|
The default POSIX level to be enforced in strict mode is now a
configuration option.
Print details of supported POSIX levels in the usage message.
Adds 56-64 bytes.
|
|
The pragma 'posix_202x' causes the 202X POSIX standard to be
enforced in POSIX mode. Add an equivalent 'posix_2017' for the
2017 standard.
There's now a DEFAULT_POSIX_LEVEL preprocessor symbol to configure
the default standard. This is hardcoded to the 2017 standard but
it can also be set to 202X.
Adds 48 bytes.
|
|
A couple of tests require backslashes to be escaped.
The test for CURDIR should use 'pwd -P' to resolve symlinks.
|
|
Add support for the conditional directives ifeq and ifneq.
These follow GNU make in allowing things like:
ifeq (arg1,arg2)
ifeq 'arg1' 'arg2'
In the second case single or double quotes may be used.
Macros are expanded in the two arguments and the resulting strings
are compared.
Adds 240-248 bytes.
|
|
POSIX only requires $< and $* to be set for inference rules, not
target rules.
As an extension allow them to be set for target rules, as in
GNU make. This may or may not be useful.
In POSIX mode, when $< and $* are only set for inference rules,
they're set to an empty string for target rules. This avoids
the possibility of stale values being used.
Adds 64-80 bytes.
(GitHub issue #407)
|
|
Austin Group defect report 1626 introduced support for the CURDIR
macro:
https://www.austingroupbugs.net/view.php?id=1626
Implement this as a POSIX 202X feature.
Adds 160-176 bytes.
|
|
|
|
The Cygwin terminal program mintty sets the HOME environment
variable. Attempt to detect this and unset HOME so the usual
busybox-w32 initialisation of HOME is used instead.
Adds 80 bytes.
(GitHub issue #420)
|
|
Merge the kill() and kill_pids() functions.
Allocate an array for the PIDs rather than use a hardcoded one.
Adds 32 bytes to the 32-bit build, none to 64-bit.
|
|
busybox-w32 had a dummy implementation of getppid(2) which always
returned 1. Provide a more realistic version.
The effect is limited:
- The PPID shell variable should report a sensible value.
- The special value to omit the parent PID 'pidof -o %PPID'
should work.
Costs 48 bytes.
|
|
Recent changes to allow orphaned processes to report a parent PID
of 1 rely on the assumption that Process32First/Process32Next
return parents before children. This isn't guaranteed by the API.
Obtain all known PIDs on the first call to procps_scan() so that
dead parents can be detected reliably.
Costs 48 bytes.
|
|
Saves 16-32 bytes
|
|
If the parent PID doesn't appear in the process table, report it
as 1. This more closely matches how orphaned children are handled
on UNIX.
Adds 96-128 bytes.
(GitHub issue #416)
|
|
A process which has exited may still have its process handle
held open by its children. Such a process doesn't appear in
the process table. It is thus similar to a zombie process in
UNIX. Using kill(1) to interact with such a process was seen
to succeed, contrary to expectation.
The code for "ordinary" signals in kill(2) did check if the
process was still active but didn't treat an attempt to kill
an inactive process as an error. Furthermore, sending SIGKILL
or the fake signal 0 to a process didn't even check if the
process was still active.
Rearrange the implementation of kill(2) so that an attempt to
signal an inactive process is treated as an error. This also
consolidates handling of SIGKILL and signal 0 with "ordinary"
signals.
Saves 96 bytes.
(GitHub issue #416)
|
|
When a shell was started with the -X option, environment variables
had forward slashes changed to backslashes. This is unnecessary
and counterproductive.
Adjust how the state of winxp is handled to avoid this.
(GitHub issue #415)
|
|
It's possible that files in remote storage may not be available
locally. Avoid downloading such files just to obtain file
attributes.
(GitHub issue #414)
|
|
|
|
An alias expansion immediately followed by '&' is parsed
incorrectly:
~ $ alias x='sleep 2'
~ $ x&
~ $
sh: syntax error: unexpected "&"
The sleep happens in the foreground and the '&' is left in the
input buffer. The same problem occurs in upstream BusyBox but
not dash.
The difference between BusyBox and dash is that BusyBox supports
bash-style output redirection (BASH_REDIR_OUTPUT in the code).
This requires checking for '&>' in readtoken1().
When the end of the alias is found, the '&' and the following
newline are both read to check for '&>'. Since there's no match
both characters are pushed back.
The alias is then expanded and __pgetc() is called to fetch the
next character. Since there are none left in the alias string
__pgetc() calls preadbuffer() which pops the string, reverts to
the previous input and recursively calls __pgetc(). This request
is satisified from the pungetc buffer. But the first __pgetc()
doesn't know this: it sees the character has come from
preadbuffer() so it (incorrectly) updates the pungetc buffer.
Resolve the issue by moving the code to pop the string and fetch
the next character up from preadbuffer() into __pgetc().
Saves 32-48 bytes.
(GitHub issue #413)
|
|
The 'noiconify' option controls how the console window is concealed
when the 'noconsole' option is used. The default is to iconify
the console. When 'noiconify' is 'on' the console is hidden.
Adds 8-16 bytes.
(GitHub issue #325)
|
|
Commit 67ed7484be (ash: detect console state on shell start up)
synchronised the noconsole option with the actual state of the
window on shell start up.
This is insufficient. The user can change the state of the window
independently of the noconsole option, leading to confusion and
unwanted iconification of the window when unrelated 'set' commands
are issued.
Detect the current console state on every call to options().
Saves 16-32 bytes.
(GitHub issue #325)
|
|
Recent versions of gcc fail to build the binary to test for
ncurses because main() is lacking a return type.
|
|
Allow the default history size (used if HISTFILESIZE isn't set)
to be configured at build time. This may be less than or equal
to the standard history size.
(GitHub issue #411)
|
|
Keep the maximum history size at 1023 but make the default 383.
This gives a modest increase over the previous default of 255
while allowing users to increase or decrease the history size
using the HISTFILESIZE environment variable.
(GitHub issue #411)
|
|
The upstream default history size is 255. Increase it to 1023
for default busybox-w32 builds.
(GitHub issue #411)
|
|
The 'quick fix' in the previous commit unnecessarily checks the
fractional timestamp value to determine if a target exists. This
isn't how it's done elsewhere in the code.
|
|
POSIX says:
A target shall be considered up-to-date... if it has already
been made up-to-date by the current invocation of make
(regardless of the target's existence or age).
If the target does not exist after the target has been
successfully made up-to-date, the target shall be treated as
being newer than any target for which it is a prerequisite.
Previously 'make' assumed that if a rule had succeeded the
modification time of the target would be the current time. This
isn't necessarily the case. Instead:
- If the file exists use the modification time of the file as the
the time of the target.
- If it doesn't exist use the current time, which should be more
recent than the time of any file for which it's a prerequisite.
Adds 16 bytes.
(GitHub issue #410)
|
|
Mixing Windows and Unix-style filename extensions was causing
problems. Tweak how extensions are handled to try and improve
matters:
- Consistently check whether the unaltered filename is an
executable before trying adding extensions.
- Check .exe and .com before .sh.
Saves up to 16 bytes.
(GitHub issue #405)
|
|
POSIX requires macro definitions to appear before targets on the
command line. Allow mixed macros and targets as an extension.
All macros on the command line are read first, then the targets
are processed.
Costs 64-80 bytes.
(GitHub issue #406)
|
|
Originally there was no validation of macro names imported from
the environment. However, process_macros() failed to account for
the additional flag bit at this level. Thus, for example, SHELL
was imported when it shouldn't have been.
The problem the lack of validation was supposed to address was
that of environment variables with invalid names, which the user
may not have control over, causing a fatal error.
As an alternative, silently ignore variables with invalid names
on import from the environment.
|
|
POSIX says, regarding execution of commands:
The execution line shall then be executed by a shell as if it
were passed as the argument to the system() interface, except
that if errors are not being ignored then the shell -e option
shall also be in effect.
As a non-POSIX extension, skip the use of the -e option. This is
how GNU make and BSD make behave.
(GitHub issue #409)
|
|
/dev/zero and /dev/urandom are only available internally and as
arguments to 'dd'. Since users can't otherwise access them they
shouldn't be treated as existing by stat(2).
With this change stat(1) and test(1) will deny their existence.
(GitHub issue #282)
|
|
Now that putchar() is implemented using winansi_fputc() it isn't
necessary to perform an explicit substitution in 'yes'.
This was previously done so 'yes' could check for a broken pipe.
Saves 16-32 bytes.
|
|
Allow the compiler to inline parse_meminfo() so it can spot some
optimisations.
Saves 208-240 bytes.
|
|
For shells, this is mandated by standards
function old new delta
input_backward 215 231 +16
read_line_input 3015 3028 +13
draw_custom 66 78 +12
put_cur_glyph_and_inc_cursor 149 159 +10
put_prompt_custom 47 56 +9
show_history 40 46 +6
input_tab 927 933 +6
input_delete 136 142 +6
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 8/0 up/down: 78/0) Total: 78 bytes
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
|
|
In the function find_export_symbols, since the fopen file does not
exit when it fails, there is a dereference problem in fclose(fp),
which will cause a segmentation fault.
Signed-off-by: Yan Zhu <zhuyan2015@foxmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
|
|
The undocumented '-d' shell option is used to set the current
directory in shells started by the 'su' applet of busybox-w32.
In this case, the shell isn't a login shell.
If a login shell sets the current working directory in /etc/profile
it's possible the user may wish to override this with '-d'. This
didn't work, though, because the directory is changed before
/etc/profile is processed.
Move the changing of the directory to that specified by '-d' so it
happens after the processing of /etc/profile and ~/.profile. This
won't affect the intended use of '-d'.
(GitHub issue #403)
|
|
Commit 54dbf0fa5 (time: mitigation for interleaved output) added
buffering to stderr in the 'time' applet.
@avih pointed out that it isn't necessary to provide an explicit
buffer.
Saves 16 bytes in the 64-bit build.
(GitHub issue #396)
|
|
Implement a 'title' built-in for ash. It's very simple-minded,
performs almost no error checking and is completely non-portable.
- With no arguments it prints the current console title.
- If arguments are provided the *first only* is set as the console
title.
Costs 88-116 bytes.
(GitHub issue #401)
|
|
The 'timeout' applet, by default, accepted fractional seconds but
ignored the fraction. Moreover, in busybox-w32, even the integer
part was incorrectly handled.
Rewrite the (Windows) code to fix both problems.
A patch has also been submitted to fix upstream. If that is
accepted the Windows port will need to be updated to match.
Saves 8-16 bytes.
(GitHub issue #400)
|
|
Other winansi IO wrappers, like winansi_fputs, optimize by calling
the libc API directly if no special handling is needed, e.g. if the
input is fully ASCII and without escape sequences - without converting
the output codepage or interpreting escapes.
Now the fputc and putchar wrappers do that as well.
And as a simplification, putchar is now also a wrapper of fputc.
Other than possibly minor speedup, this can also help buffered streams
remain buffered, because the codepage conversion using writeCon_utf8
is unbuffered and first flushes the stream, so by avoiding the
conversion and calling the libc API directly, we also avoid
premature flush of a buffered stream.
This did happen, as "time" is buffered since commit 54dbf0fa5, so
previously it was flushed early when using putchar, while now it
remains buffered by default (but can still be flushed early if the -f
format string contains non-ASCII chars).
|
|
writeCon_utf8 is unbuffered - it writes directly using WriteConsoleW,
but some winansi libc IO wrappers, like fputs, use the libc API
directly if the content doesn't need any special handling (e.g. all
ASCII and no escape sequences), and so if the stream is buffered, and
if only some parts of it go through writeCon_utf8, then we can get
wrong output order due to some parts being buffered and some not.
Case in point, the recent commit 54dbf0fa5 made the output of "time"
buffered, and so if only parts of it go through writeCon_utf8, then we
get bad output order.
This did actually happen, because not all the winasi wrappers have
this ASCII-only optimization (e.g. winansi_putchar), and "time" did
end up with wrong output order. Even if all the winansi wrappers were
ASCII-optimized, "time" could still have unicode output, e.g. with -f.
Fix it by flushing the stream before converting using writeCon_utf8.
|
|
When the 'time' applet is run in a pipeline, like so:
time seq 1 10 | tail -2
stdout from 'tail' and stderr from 'time' can be interleaved.
This is particularly the case with the ConEmu terminal emulator.
The interleaving can be reduced, though not eliminated, by
buffering the output of 'time'.
Adds 40-44 bytes.
(GitHub issue #396)
|
|
The coreutils versions of md5sum and the like accept uppercase hex
strings from checksum files specified with the '-c' option.
Use a case-insensitive comparison so BusyBox does the same.
(GitHub issue #394)
|
|
interval, in_interval_table, and in_uint16_table were previously
not compiled when using the new wcwidth (commit c188a345a) because
they're used by the old wcwidth but not by the new one.
But they're also used by the BIDI routines. mingw64u_defconfig doesn't
enable bidi (rightly - it's not working well), but it'd still be nice
to allow enabling bidi while the new wcwidth is in effect.
Enable the tables lookup code if BIDI is enabled.
|
