| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Create mingw_strptime() to return timezone offset as a separate
argument (since Microsoft's struct tm doesn't have the required
member).
Import timegm() from musl.
Update parse_datestr() to use mingw_strptime().
Enable FEATURE_TIMEZONE in the default configuration.
GitHub issue #230.
|
|
|
|
| |
Saves 88 bytes.
|
|
|
|
|
|
|
|
|
| |
The condition to detect the end of the environment string was wrong.
Don't bother calculating the MD5SUM of the environment, just XOR
the bytes into the data.
This reduces bloat by 320 bytes but only in the non-default case.
|
|
|
|
|
|
|
|
| |
Turn off the 'if-conversion' optimisation for err_win_to_posix().
My tests actually have this being slightly faster than with the
optimisation enabled, though within the variation of the measurement.
Saves 288 bytes.
|
|
|
|
|
|
|
| |
If GetVersionEx() fails just assume the OS version numbers remain
set to zero and print them as-is.
Saves 48 bytes.
|
|
|
|
| |
Saves 176 bytes.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add a new function, getsysdir(), to fetch and cache the system
directory. This avoids the non-intuitive use of getpwuid() in
get_system_drive().
The call to GetSystemDirectory() in get_proc_addr() can't be
replaced because getsysdir() calls realpath() which requires a
call to get_proc_addr().
No change in the size of the binary.
|
|
|
|
|
|
|
|
|
| |
There doesn't seem to be any need to call OpenThreadToken() in
file_owner(): OpenProcessToken() should suffice.
Also, tidy up gethomedir() without any change in functionality.
Saves 56 bytes.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 15fcbd19c (win32: special case for devices files in stat(2))
added special treatment in stat(2) for device files.
As a result of this change device files appeared to be regular files
which broke the use of /dev/null with noclobber in the shell.
Device files now appear as character special files (as they do on
Unix).
GitHub issue #225.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Our realpath(3) implementation uses xmalloc_follow_symlinks() to
expand symlinks. This detects when symlinks are too deeply nested
but didn't set errno, so anything calling realpath(3) was unable to
say what had gone wrong. (For example, 'ls -L' or 'stat -L'.)
Set errno to ELOOP.
This then leads to the problem that Windows doesn't know about
ELOOP so reports 'Unknown error'. Add a replacement for strerror(3)
which returns a sensible message.
Costs 96 bytes.
|
|
|
|
|
|
|
| |
Add a routine to detect the names of character classes. Use it
in fnmatch(3) and regcomp(3), replacing local code in the former.
Saves 216 bytes.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The glob pattern '[[:xdigit::]]*' didn't return the matches
expected. It turns out the implementation (from glibc) fails
to detect 'xdigit' as a valid character class. Changing the
definition of CHAR_CLASS_MAX_LENGTH to 7 fixes the problem.
This was never an issue in glibc because it uses a different
definition. More modern versions of fnmatch(3) in glibc and
gnulib also make CHAR_CLASS_MAX_LENGTH long enough.
The code for fnmatch(3) was taken from glibc at commit
7814856974388a856a575fa45f88d502c8a1ab29. This was the last
version before the code was rearraged to better support multibyte
characters.
|
|
|
|
|
|
|
| |
Remove filetime_to_time_t(): it's no longer used.
Align style of time{spec,val}_to_filetime() to make it easier to
compare what they do.
|
|
|
|
|
|
|
|
|
| |
Save a few bytes:
- When collecting entropy prefer functions we call elsewhere.
- In uname(2) set 32-bit processor type to i686 and tweak it for
i386.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In read_keys():
- Identify all keys on the numeric pad used to enter character
codes. Otherwise Alt-Left/Alt-Right on the numeric pad are
treated as word movements in command line editing, preventing
the entry of character codes containing 4 or 6.
- Add modifier flag bits to the virtual keycodes we care about.
This means, for example, that only the unmodified Up/Down arrow
keys move through shell history, not Up/Down plus arbitrary
modifiers.
|
|
|
|
| |
No change in functionality. Saves 16 bytes.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 7874ca73b (win32: allow characters to be entered as ALTNNN)
made it possible to enter characters using Alt and a character code
on the numeric pad. This required detecting the release of the
Alt key.
Sometimes this caused a problem when using Alt+Tab to cycle between
applications. If Alt was released before Tab the code passed a tab
character to the application.
Avoid such issues by looking specifically for the release of the Alt
key.
|
|
|
|
|
|
|
| |
Use is_prefixed_with() rather than strncmp() in a few places,
and the case-insensitive analogues.
Saves 96 bytes in 64-bit build, 192 bytes in 32-bit.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the code to detect binaries:
- use '|' rather than '+' to combine bytes;
- fix the test that the PE header is within the buffer;
- once we have the offset to the PE header make a pointer to it;
- cosmetic changes.
Saves 96 bytes.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The diff applet calls stat(2) on the files it's asked to process.
This includes /dev/null when the -N flag is used and /dev/fd/*
files when process substitution is used.
Treat device files as a special case in get_file_attr(), returning
a fake set of attributes with FILE_ATTRIBUTE_DEVICE set. This
value is unused elsewhere in busybox-w32. Ensure it's unset in
other cases.
When FILE_ATTRIBUTE_DEVICE is set:
- adjust some permissions;
- avoid calling has_exec_format() as this opens/closes the file
which breaks process substitution.
These changes improve the behaviour of diff but they also have other
effects. For example, the stat and ls applets now report details of
device files. There may be unintended consequences.
|
|
|
|
|
|
|
|
|
| |
The lzma and xz applets don't support compression. Any attempt
to generate a tar file using these compressors should try to use
an external program. (Which probably won't work on a typical
Windows system, but hey, at least we made the effort.)
See GitHub issue #222.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Make the structure returned by getmntent(3) static so it persists
after endmntent(3) closes the stream. No current caller in the
WIN32 port needs this functionality but it's good to match the
documented behaviour.
Populate more fields of the mntent structure in find_mount_point().
This is required to support the df -t and -T flags recently added
upstream.
The static structures used here are allocated on demand. Separate
static structures are used in each case because df iterates through
mounts calling statfs(2) on each and the WIN32 implementation of
statfs(2) calls find_mount_point().
|
|
|
|
|
|
|
|
| |
The implementations of execve(2) and execvp(3) were unable to
distinguish between failure and a program returning a status of -1.
Check the error return code to discriminate between these cases.
See GitHub issue #218.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add a cut down version of the dirent implementation from git.
The git developers said:
The mingw-runtime implemenation of opendir, readdir and closedir
sets errno to 0 on success, something that POSIX explicitly
forbids.
This also avoids having to link against libssp.a (commit 13eb34205)
and reduces the size of the binary by 2KB.
|
|
|
|
|
|
|
|
|
|
| |
Provide a partial implementation of sync(2), so sync(1) can actually
do something in some circumstances:
- Only logical drives are handled.
- Flushing buffers requires administrative privileges. If run as
a normal user nothing will happen.
|
|
|
|
| |
Add support for virtual /dev/fd files to represent file descriptors.
|
|
|
|
| |
Avoid confusion between special devices and /dev/fd.
|
|
|
|
|
| |
The touch applet has been changed to use futimens(2)/utimensat(2).
Provide implementations of these for WIN32.
|
|
|
|
|
|
|
|
| |
The 64-bit build of busybox-w32 failed to create symbolic links
on Windows 7 but claimed to have succeeded. The declaration of
CreateSymbolicLinkA had the wrong return value.
See GitHub issue #217.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The following commands (reported in GitHub issue #201):
printf "\033[38;2;255;0;0mX\033[m\n"
printf "\033[38;2;255;0;0m;\033[m\n"
produce different results. The first correctly displays a red 'X'
while the second incorrectly displays a white ';'.
The problem is that process_24bit() overruns the extent of the
escape sequence. As a result the loop in process_escape() which
handles 'ESC[...m' sequences sees the ';' in the text as a
continuation of the escape sequence.
Fix this by:
- reworking process_24bit() so that the overrun is avoided;
- changing the test in the loop in process_escape() so that even
if an overrun happens it stops processing at the end of the escape
sequence.
Also, save a few bytes by replacing '++str' with 'str + 1' in a
few places.
|
|
|
|
|
|
| |
In xmalloc_follow_symlinks() the code to detect relative paths
needs to be altered for WIN32. We don't want C:/path to be
treated as a relative path.
|
|
|
|
| |
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
|
|
|
| |
That function would rename the _target_, not the link.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
|
|
|
|
|
|
| |
When one tries to call `CreateFile()` with a reparse point, it will
trigger an `ERROR_INVALID_NAME` unless `FILE_FLAG_OPEN_REPARSE_POINT` is
passed. However, _when_ that flag is passed, it does not open a handle
to the symlink _target_, but to the symlink itself.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
|
|
|
| |
Otherwise we're provoking buffer overruns.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On Windows, there are file symlinks and directory symlinks. When trying
to `opendir()` a symlink marked as `file`, it will fail. Even if the
target is a directory. Because it's the wrong symlink type.
To address this, our `symlink()` function calls `stat(target, ...)` to
see whether the target exists and is a directory.
The problem is that this `target` can be a relative path, and the link
path can _also_ be a relative path. Example: `symlink("dir", "uh/oh")`.
In this example, the target might say `dir`, but it is relative to
`uh/oh`, i.e. we need to `stat("uh/dir", ...)`.
This is necessary to pass the `cp` tests because they first create such
a directory symlink and then try to copy it while dereferencing
symlinks, i.e. calling `opendir()` on the symlink.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
|
|
| |
Make is_absolute_path() a function rather than a macro and move it
from ash.c into mingw.c.
|
|
|
|
|
|
|
|
| |
Investigating why free(1) wasn't working on Windows 7 I found it's
possible for LoadLibraryEx to fail with exactly the flags we're
using. Work around this.
This probably also explains GitHub issue #204.
|
|
|
|
|
|
|
|
|
|
| |
Use another API call: EnumPageFiles(). This seems to provide more
reliable information about page file usage than the previous ad hoc
method. It also allows the call to GlobalMemoryStatusEx() to be
removed.
With these changes free(1) works sensibly on Windows XP, though not
ReactOS.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There seems to be a bug in the Windows/MSYS2/mingw-w64 toolchain
which results in the generated win32/resources/built-in.o lacking
relocation data.
Forcing an actual C program to be linked along with the resources
appears to be an effective workaround.
When the toolchain bug is resolved this commit can be reverted.
See GitHub issue #200.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
To investigate GitHub issue #200 it was necessary to perform
build on Window using the MSYS2/mingw-w64 toolchain. This
threw up some issues:
- The settings for _WIN32_WINNT and __USE_MINGW_ANSI_STDIO differ
from those in Fedora resulting in compiler errors and warnings.
Force the defaults I'm used to.
- The workaround to allow native compilation of mconf.c was broken
by a subsequent upstream change. Make it work again.
|
|
|
|
|
|
|
|
|
|
| |
Add a call to GetPerformanceInfo. Treat the SystemCache member of
the PERFORMANCE_INFORMATION structure as buffer RAM. Deduct it
from available RAM.
The numbers reported by 'free' move about in vaguely sensible
ways when I start and stop programs, though I still don't know
if they're in any way accurate.
|
|
|
|
|
|
|
|
|
|
| |
Fill the uptime member of the sysinfo structure. With this change
we can:
- use sysinfo(2) in the 'ps' applet;
- enable the 'uptime' applet (though without useful support for
load averages).
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is an experimental implementation of sysinfo(2)/free(1).
It uses the WIN32 API GlobalMemoryStatusEx() to obtain information
about memory.
It seems that the 'total pagefile' value includes total RAM as well
as pagefile and 'available pagefile' includes available RAM. So the
RAM values are deducted.
I've no idea what corresponds to Linux buffers and cache.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit b0b7ab792 (winansi: code shrink) noted that combining reverse
vidoe escape sequences (e.g. ESC[7;27m) didn't work properly.
Make further changes to improve the situation:
- revert to keeping the current attributes in a static variable;
- when reverse video is enabled switch the intensity as well as the
colour components;
- move the code from set_console_attr() into its only caller,
process_escape();
- use 0xffff instead of 0 as a flag to indicate the attributes
haven't been initialised.
Saves 44 bytes and seems to work better.
|
| |
|
|
|
|
|
|
|
|
| |
The alternative console screen buffer (used by less and vi) doesn't
work in Wine.
Setting the environment variable BB_ALT_BUFFER to 0 causes a screen
reset instead.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Running the test suite on Wine failed because in
seq 4 0 8 | head -n 10
'seq' didn't detect the broken pipe when 'head' terminated and
carried on forever. Fix this by adding a call to ferror(3) in
winansi_vfprintf().
Also, use xstrdup() and xmalloc() in a couple of places.
|