Cygwin: stat(): Fix "Bad address" error on stat() for /dev/tty.
As reported in
https://cygwin.com/pipermail/cygwin/2023-June/253888.html,
"Bad address" error occurs when stat() is called after the commit 3721a756b0d8 ("Cygwin: console: Make the console accessible from
other terminals.").
There are two problems in the current code. One is fhandler_console::
fstat() calls get_ttyp()->getsid(). However, fh_alloc() in dtable.cc
omits to initialize the fhandler_console instance when stat() is
called. Due to this, get_ttyp() returns NULL and access violation
occurs. The other problem is fh_alloc() assigns fhandler_console
even if the CTTY is not a console. So the first problem above occurs
even if the CTTY is a pty.
This patch fixes the issue by:
1) Call set_unit() to initialize _tc if the get_ttyp() returns NULL.
2) Assign fhandler_pty_slave for /dev/tty if CTTY is a pty in fh_alloc().
Fixes: 3721a756b0d8 ("Cygwin: console: Make the console accessible
from other terminals."). Fixes: 23771fa1f7028 ("dtable.cc (fh_alloc): Make different decisions
when generating fhandler for not-opened devices. Add kludge to deal
with opening /dev/tty.") Reported-by: Bruce Jerrick <bmj001@gmail.com> Reviewed-by: Corinna Vinschen <corinna@vinschen.de> Signed-off-by: Takashi Yano <takashi.yano@nifty.ne.jp>
Mark Geisert [Tue, 4 Jul 2023 00:51:41 +0000 (17:51 -0700)]
Cygwin: Make <sys/cpuset.h> safe for c89 compilations
Four modifications to include/sys/cpuset.h:
* Change C++-style comments to C-style also supported by C++
* Change "inline" to "__inline" on code lines
* Add "#include <sys/cdefs.h>" to make sure __inline is defined
* Don't declare loop variables on for-loop init clauses
Tested by first reproducing the reported issue with home-grown test
programs by compiling with gcc option "-std=c89", then compiling again
using the modified <sys/cpuset.h>. Other "-std=" options tested too.
Addresses: https://cygwin.com/pipermail/cygwin-patches/2023q3/012308.html Fixes: 315e5fbd99ec ("Cygwin: Fix type mismatch on sys/cpuset.h") Signed-off-by: Mark Geisert <mark@maxrnd.com>
Takashi Yano [Thu, 22 Jun 2023 15:07:26 +0000 (00:07 +0900)]
Cygwin: thread: Reset _my_tls.tid if it's pthread_null in init_mainthread().
Currently, _my_tls.tid is set to pthread_null if pthread::self()
is called before pthread::init_mainthread(). As a result, pthread::
init_mainthread() does not set _my_tls.tid appropriately. Due to
this, pthread_join() fails in LDAP environment if the program is
the first program which loads cygwin1.dll.
Thomas Schwinge [Fri, 14 Apr 2023 19:26:32 +0000 (21:26 +0200)]
For GCC, newlib combined tree, newlib build-tree testing, use standard search paths
For example, for GCC/GCN target (AMD GPUs), target libraries are built
individually per supported hardware ISA ('-march=[...]'). Testing such a
toolchain via, for example:
$ make RUNTESTFLAGS='--target_board=[...]/-march=gfx90a' check[...]
... does work fine for all 'check-gcc-[...]' as well as GCC-provided target
libraries, 'check-target-[...]'. Just for 'check-target-newlib', for the
example above, not the '-march=gfx90a' newlib libraries are linked in, but
instead always the default ones, which results in link FAILure. This is cured
simply by skipping use of 'newlib/testsuite/lib/flags.exp', so that the
standard search paths as determined by GCC, DejaGnu are used for newlib, too.
Brian Inglis [Mon, 19 Jun 2023 18:15:19 +0000 (12:15 -0600)]
fhandler/proc.cc: use wincap.has_user_shstk
In test for AMD/Intel Control flow Enforcement Technology user mode
shadow stack support replace Windows version tests with test of wincap
member addition has_user_shstk with Windows version dependent value
Fixes: 41fdb869f998 ("fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo") Signed-off-by: Brian Inglis <Brian.Inglis@Shaw.ca>
Do not rely on `getenv ("HOME")`'s path conversion
In the very early code path where `dll_crt0_1 ()` calls
`user_shared->initialize ()`, the Cygwin runtime calls `internal_pwsid ()`
to initialize the user name in preparation for reading the `fstab` file.
In case `db_home: env` is defined in `/etc/nsswitch.conf`, we need to
look at the environment variable `HOME` and use it, if set.
When all of this happens, though, the `pinfo_init ()` function has had no
chance to run yet (and therefore, `environ_init ()`). At this stage,
therefore, `getenv ()`'s `findenv_func ()` call still finds `getearly ()`
and we get the _verbatim_ value of `HOME`. That is, the Windows form.
But we need the "POSIX" form.
To add insult to injury, later calls to `getpwuid (getuid ())` will
receive a cached version of the home directory via
`cygheap->pg.pwd_cache.win.find_user ()` thanks to the first
`internal_pwsid ()` call caching the result via
`add_user_from_cygserver ()`, read: we will never receive the converted
`HOME` but always the Windows variant.
So, contrary to the assumptions made in 27376c60a9 (Allow deriving the
current user's home directory via the HOME variable, 2023-03-28), we
cannot assume that `getenv ("HOME")` returned a "POSIX" path.
This is a real problem. Even setting aside that common callers of
`getpwuid ()` (such as OpenSSH) are unable to handle Windows paths in the
`pw_dir` attribute, the Windows path never makes it back to the caller
unscathed. The value returned from `fetch_home_env ()` is not actually
used as-is. Instead, the `fetch_account_from_windows ()` method uses it
to write a pseudo `/etc/passwd`-formatted line that is _then_ parsed via
the `pwdgrp::parse_passwd ()` method which sees no problem with
misinterpreting the colon after the drive letter as a field separator of
that `/etc/passwd`-formatted line, and instead of a Windows path, we now
have a mere drive letter.
Let's detect when the `HOME` value is still in Windows format in
`fetch_home_env ()`, and convert it in that case.
For good measure, interpret this "Windows format" not only to include
absolute paths with drive prefixes, but also UNC paths.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The account under which Azure Web Apps run is an IIS APPOOL account that
is generated on the fly.
These are special because the virtual machines on which thes Apps run
are not domain-joined, yet the accounts are domain accounts.
To support the use case where such a Web App needs to call `ssh` (e.g.
to deploy from a Git repository that is accessible only via SSH), we do
need OpenSSH's `getpwuid (getuid ())` invocation to work.
But currently it does not. Concretely, `getuid ()` returns -1 for these
accounts, and OpenSSH fails to find the correct home directory
(_especially_ when that home directory was overridden via a `db_home:
env` line in `/etc/nsswitch.conf`).
This can be verified e.g. in a Kudu console (for details about Kudu
consoles, see https://github.com/projectkudu/kudu/wiki/Kudu-console):
the domain is `IIS APPPOOL`, the account name is the name of the Azure
Web App, the SID starts with 'S-1-5-82-`, and
`pwdgrp::fetch_account_from_windows()` runs into the code path where
"[...] the domain returned by LookupAccountSid is not our machine name,
and if our machine is no domain member, we lose. We have nobody to ask
for the POSIX offset."
Since these IIS APPPOOL accounts are relatively similar to AzureAD
accounts in this scenario, let's imitate the latter to support also the
former.
Reported-by: David Ebbo <david.ebbo@gmail.com> Helped-by: Corinna Vinschen <corinna@vinschen.de> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Takashi Yano [Fri, 2 Jun 2023 01:19:18 +0000 (10:19 +0900)]
Cygwin: pty: Additional fix for transferring input at exit.
The commit 9fc746d17dc3 does not fix transferring input at exit
appropriately. If the more than one non-cygwin apps are executed
simultaneously and one of them is terminated, the pty master failed
to send input to the other non-cygwin apps. This patch fixes that.
Takashi Yano [Thu, 1 Jun 2023 10:44:41 +0000 (19:44 +0900)]
Cygwin: pty: Fix transferring type-ahead input between input pipes.
After the commit e5fcc5837c95, transferring type-ahead input between
the pipe for cygwin app and the pipe for non-cygwin app will not be
done appropriately when the stdin of the non-cygwin app is not pty.
Due to this issue, sometimes the keyboard input might be lost which
should be sent to cygwin app. This patch fixes the issue.
Fixes: e5fcc5837c95 ("Cygwin: pty: Fix reading CONIN$ when stdin is not a pty.") Signed-off-by: Takashi Yano <takashi.yano@nifty.ne.jp>
Cygwin: Adjust CWD magic to accommodate for the latest Windows previews
Reportedly Windows 11 build 25*** from Insider changed the current
working directory logic a bit, and Cygwin's "magic" (or:
"technologically sufficiently advanced") code needs to be adjusted
accordingly.
This fixes https://github.com/git-for-windows/git/issues/4429
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
newlib: Add non LDBL_EQ_DBL math support for aarch64, i386, and x86_64
Rename s_nearbyint.c, s_fdim.c and s_scalbln.c to remove conflicts
Remove functions that are not needed from above files
Modify include paths
Add includes missing in cygwin build
Add missing types
Create Makefiles
Create header files to resolve dependencies between directories
Modify some instances of unsigned long to uint64_t for 32 bit platforms
Add HAVE_FPMATH_H
Brian Inglis [Mon, 8 May 2023 03:12:10 +0000 (21:12 -0600)]
fhandler/proc.cc(format_proc_cpuinfo): Add Linux 6.3 cpuinfo
cpuid 0x00000007:0 ecx:7 shstk Shadow Stack support & Windows [20]20H1/[20]2004+
=> user_shstk User mode program Shadow Stack support
AMD SVM 0x8000000a:0 edx:25 vnmi virtual Non-Maskable Interrrupts
Sync AMD 0x80000008:0 ebx flags across two output locations
Sebastian Huber [Wed, 26 Apr 2023 06:23:04 +0000 (08:23 +0200)]
Fix _REENT_EMERGENCY() if TLS is enabled
If the thread-local storage (TLS) support was enabled, the _REENT_EMERGENCY()
object had the wrong size. It must be a buffer of length _REENT_EMERGENCY_SIZE
and not just a single character.
Cygwin: locales: ignore @cjkwide and @cjksingle just like @cjknarrow
When the @cjkwide and @cjksingle modifiers have been added, the
patches missed to add checks for the new modifiers in the Cygwin
locale code. Along the same lines, commit c3e7f7609e46 forgot to
add a test for @cjksingle.
Merge check for cjk* modifiers into a macro set andf use that
throughout. Fix comments.
Fixes: f92f048528e6f ("Locale modifier @cjkwide to adjust ambiguous-width in non-CJK locales") Fixes: c8204b106988f ("Locale modifier "@cjksingle" to enforce single-width CJK width.") Fixes: c3e7f7609e46 ("Cygwin: locales: fix behaviour for @cjk* and @euro locales") Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Per the discussion starting with
https://cygwin.com/pipermail/cygwin/2023-April/253495.html
stop falling back to sh if the file given to posix_spawnp
is no executable.
This is not necessarily the last word on it, given
https://www.austingroupbugs.net/view.php?id=1674, but for now,
opt for following the proposal in the Austin Group bug entry,
as well as PASSing the GNULIB test-posix_spawnp-script test.
Fixes: c7c1a1ca1b33 ("Add support for new posix_spawn function.") Fixes: 3fbfcd11fb09 ("Cygwin: posix_spawn: add Cygwin-specific code fixing process synchronisation") Reported-by: Bruno Haible <bruno@clisp.org> Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
posix_spawn_file_actions_addfchdir_np: return EBADF on negative fd
FreeBSD and Musl implement posix_spawn_file_actions_addfchdir_np
so that it checks the incoming descriptor for being negative, and,
if so, return with EBADF. The POSIX proposal defining
posix_spawn_file_actions_addfchdir follows this behaviour, see
https://www.austingroupbugs.net/view.php?id=1208
In contrast to rename default behaviour, Linux' renameat2 returns -1
with errno set to EEXIST, if oldfile and newfile refer to the same
file, and the RENAME_NOREPLACE flag is set.
Follow suit, given this is a Linux-only function anyway.
readlinkat(fd, "", ...) is supposed to return ENOENT per POSIX,
but Cygwin returns EBADF.
At the same time, we have to maintain the special feature of
glibc that readlinkat(fd, "", ...) operates on fd, if fd is pointing
at a symlink opened with O_PATH|O_NOFOLLOW.
And, while fixing that, readlinkat(fd, path, ...) *still* has to set errno
to EBADF, if fd is an invalid descriptor *and* path is a relative path.
This required to change the evaluation order in the helper function
gen_full_path_at.
Last but not least, in case of the aforementioned glibc-like special
handling for symlink descriptors, we have to make sure that errors from
gen_full_path_at are not spilled into that special handling.
Fixes: 6cc05784e16a ("Cygwin: readlinkat: allow pathname to be empty") Reported-by: Bruno Haible <bruno@clisp.org> Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
When cygheap_max was > CYGHEAP_STORAGE_INITIAL, commit_size would be set to
allocsize(cygheap_max), which is an address, not a size. VirtualAlloc would be
called to commit commit_size bytes, which would fail, and then child_copy would
be called with zero as the base address.
Fixes: 2f9b8ff00cce ("Cygwin: decouple cygheap from Cygwin DLL") Signed-off-by: David McFarland <corngood@gmail.com>
Cygwin: pty: Fix reading CONIN$ when stdin is not a pty.
Previously, the pty master sends inputs to the pipe for cygwin app
even when pseudo console is activated if stdin is not the pty.
This causes the problem that key input is not sent to non cygwin
app even if the app opens CONIN$. This patch sets switch_to_nat_pipe
to true regardless whether stdin is the pty or not to allow that case.
E.g. known errors:
x = 0x1.000002c5e2e99p+0, y = 0x1.c9eee35374af6p+31 had an error of 639
ULP and is now correctly rounded.
x = 0x1.fffffd2e3e669p-1, y = 0x1.344c9823eb66cp+32 had an error of 428
ULP and is now correctly rounded.
Cygwin: chattr: fix description of requirements for casesensitive directories
Preconditions of WSL or empty directories dependent on Windows
versions was totally screwed up. Drop the description from
--help, describe the preconditions for case-sensitive dirs in the
man page instead.
Respect `db_home` setting even for SYSTEM/Microsoft accounts
We should not blindly set the home directory of the SYSTEM account (or
of Microsoft accounts) to `/home/<name>`, especially
`/etc/nsswitch.conf` defines `db_home: env`, in which case we want to
respect the `HOME` variable.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Corinna Vinschen [Wed, 29 Mar 2023 08:18:23 +0000 (10:18 +0200)]
Cygwin: dirname: fix handling of leading slashes
Per https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/xbd_chap04.html:
"A pathname that begins with two successive slashes may be interpreted
in an implementation-defined manner, although more than two leading
slashes shall be treated as a single slash."
So more than 2 leading slashes are supposed to be folded into one,
which our dirname neglected. Fix that.
Allow deriving the current user's home directory via the HOME variable
This patch hails from Git for Windows (where the Cygwin runtime is used
in the form of a slightly modified MSYS2 runtime), where it is a
well-established technique to let the `$HOME` variable define where the
current user's home directory is, falling back to `$HOMEDRIVE$HOMEPATH`
and `$USERPROFILE`.
The idea is that we want to share user-specific settings between
programs, whether they be Cygwin, MSYS2 or not. Unfortunately, we
cannot blindly activate the "db_home: windows" setting because in some
setups, the user's home directory is set to a hidden directory via an
UNC path (\\share\some\hidden\folder$) -- something many programs
cannot handle correctly, e.g. `cmd.exe` and other native Windows
applications that users want to employ as Git helpers.
The established technique is to allow setting the user's home directory
via the environment variables mentioned above: `$HOMEDRIVE$HOMEPATH` or
`$USERPROFILE`. This has the additional advantage that it is much
faster than querying the Windows user database.
Of course this scheme needs to be opt-in. For that reason, it needs
to be activated explicitly via `db_home: env` in `/etc/nsswitch.conf`.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Corinna Vinschen [Sun, 26 Mar 2023 11:06:38 +0000 (13:06 +0200)]
Cygwin: locales: fix behaviour for @euro locales
Latest Windows supports more EU locales than GLibc, so some of the
@euro locales are not covered by checking the GLibc locale defaults.
Those locales have no long history, they are all UTF-8. So just
check for @euro in the UTF-8 case and set them to ISO-8859-15.
Fixes: 2483e54be852e ("Cygwin: locale: Set default charset from Linux locale -> codeset mapping") Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Corinna Vinschen [Sun, 26 Mar 2023 10:50:57 +0000 (12:50 +0200)]
Cygwin: /proc/locales: use modifier when checking default codeset
Drop usage of newlocale/nl_langinfo_l/freelocale.
Call __set_charset_from_locale instead, and make sure to call it
with modifier, if any, otherwise suffer wrong results.
Corinna Vinschen [Sat, 25 Mar 2023 21:48:00 +0000 (22:48 +0100)]
Cygwin: locales: fix return value check of ResolveLocaleName
ResolveLocaleName does not simply return an error value if it
can't resolve a locale. Rather, it returns an empty string and
the length of this string: 1.
The idea behind this patch was wrong. Systems are supposed to
support iso639-only strings as settings for the locale environment
variables, and they are not necessarily available in the
/usr/share/locale/locale.alias file.
Corinna Vinschen [Fri, 24 Mar 2023 11:43:47 +0000 (12:43 +0100)]
Cygwin: locales: implement own method to check locale validity
The Windows function ResolveLocaleName is next to useless to
convert a partial locale identifier into a full, supported
locale identifier. It converts anything which vaguely resembles
a locale into some other locale it supports.
Bad examples are:
"en-XY" gets converted to "en-US", and worse,
"ff-BF" gets converted to "ff-Latn-SN", even though "ff-Adlm-BF"
exists!
To check if a locale is supported, we have to enumerate all valid
Windows locales, and return the match, even if the locale in Windows
requires a script. Implement resolve_locale_name() as replacement
function for ResolveLocaleName.
Fixes: e95a7a795522 ("Cygwin: convert Windows locale handling from LCID to ISO5646 strings") Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Corinna Vinschen [Fri, 24 Mar 2023 10:47:23 +0000 (11:47 +0100)]
Cygwin: /proc/locales: fix sd_IN locale
Due to the way locales are evaluated in Windows, we can't ask for
the script of the "sd-IN" locale, because Windows only knows the
"sd-Deva-IN" locale. So asking for the script of the "sd" locale
returns "Arab;", because "sd" is converted to "sd-Arab-PK".
Deleting files returns STATUS_INVALID_PARAMETE on a bind mounted file system
in hyper-v container with FILE_DISPOSITION_POSIX_SEMANTICS.
Therefore fall back to default method.
This code is suggested by Johannes Schindelin on github
and I change it more simple.
Corinna Vinschen [Tue, 21 Mar 2023 17:29:59 +0000 (18:29 +0100)]
Cygwin: don't use unlink/rename POSIX semantics on certain NTFS
If a host NTFS is mapped into a Hyper-V isolated container, the
OPEN_BY_FILE_ID filesystem flag is missing, just as if that NTFS
is a remote drive. However, NtQueryVolumeInformationFile claims
the drive is a local drive.
We can use this fact to learn that the process is running under
Hyper-V, and that the Hyper-V isolated process can't use rename/unlink
with POSIX semantics. Strange enough, the POSIX_UNLINK_RENAME filesystem
flag is still set...
These flags are used to check a remote filesystem. Not all
flags supported by a local NTFS are available when checking
a remote NTFS. Fix the flag set accordingly, otherwise
the remote NTFS will ba handled as CIFS.
Fixes: fcccdc4021ff ("Cygwin: fs_info: update filesystem flags and check Windows 7 flags") Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Corinna Vinschen [Thu, 16 Mar 2023 20:09:53 +0000 (21:09 +0100)]
Cygwin: regex: fix faulty check for valid range expression
Except for the "C" or "POSIX" locale, checking for start <= finish
is always wrong. Range start must be <= range finish in terms of the
locale's collating order. So make sure to call always wcscoll(), even
in the "C"/"POSIX" locale, which makes wcscoll equivalent to wcscmp
anyway.
Corinna Vinschen [Thu, 16 Mar 2023 20:52:48 +0000 (21:52 +0100)]
Cygwin: regex: fix freeing g->charjump in low memory condition
computejumps() moves g->charjump to a position relativ to the value of
CHAR_MIN. As such, g->charjump doesn't necessarily point to the address
actually allocated. While regfree() takes that into account, the low
memory handling in regcomp_internal() doesn't. Fix that by free'ing
the actually allocated address, as in regfree().
Corinna Vinschen [Mon, 20 Mar 2023 11:22:07 +0000 (12:22 +0100)]
Cygwin: fs_info: update filesystem flags and check Windows 7 flags
Update the list of filesystem flags to the flags supported since
Windows 7. Make sure to use the new flags only with Windows
filesystems, not with 3rd party filesystems.
Corinna Vinschen [Thu, 16 Mar 2023 12:47:07 +0000 (13:47 +0100)]
Cygwin: regex: NONCHAR: re-add cast to int
wint_t is unsigned int and the test checks for a negative value. Thus,
it's optimized out by gcc. Add the cast from commit 44caccfca2433 to
avoid this.
Make kill -V and kill -l exit immediately, thus stopping to
print "not enough arguments" accidentally.
Fixes: ef48a2cad3704 ("* kill.cc (prog_name) New global variable.") Fixes: c49fa762631f9 ("* Makefile.in (kill.exe): Add as a specific target.") Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Take the opportunity to follow FreeBSD's and Linux's lead in recasting
macro inline code as calls to static inline functions. This allows the
macros to be type-safe. In addition, added a lower bound check to the
functions that use a cpu number to avoid a potential buffer underrun on
a bad argument. h/t to Corinna for the advice on recasting.
Takashi Yano [Sun, 5 Mar 2023 09:17:39 +0000 (18:17 +0900)]
Cygwin: ctty: Remove old 'kludge' code.
Remove old 'kludge' code which does not seem necessary anymore. The
comment of the 'kludge' is as follows.
* syscalls.cc (setsid): On second thought, in the spirit of keeping
things kludgy, set ctty to -2 here as a special flag, and...
(open): ...only eschew setting O_NOCTTY when that case is detected.
Takashi Yano [Sun, 5 Mar 2023 10:02:13 +0000 (19:02 +0900)]
Cygwin: ctty: Replace ctty constant with more descriptive macros.
This patch replaces ctty constants with more descriptive macros
(CTTY_UNINITIALIZED and CTTY_RELEASED) rather than -1 and -2 as
well as checking sign with CTTY_IS_VALID().
Fixes: 3b7df69aaa57 (Cygwin: ctty: Add comments for the special values: -1 and -2.) Suggested-by: Corinna Vinschen <corinna@vinschen.de> Signed-off-by: Takashi Yano <takashi.yano@nifty.ne.jp>
Cygwin: glob: fix conversion from UTF-32 to multibyte
The conversion function g_Ctoc missed to drop the flag
values from the wint_t value. That wasn't noticable with
the original version because it used a 64 bit Char type
and the flags were in the upper 32 bit region. So the
flag values were silently dropped when wcrtomb was called.
After converting Char to wint_t, we have to do drop the
flags explicitely.
arm: Restrict processor mode change when in hypervisor mode
If a CPU implements EL2 as its highest exception level then programs
using newlib may start in hypervisor mode. In that state it is not
trivial to switch into the various EL1 modes to configure the
individual exception stacks, so do not try.
Cygwin: fnmatch: support collating symbols in [. .] brackets
This requires quite a few changes in how fnmatch operates.
It always operates on wint_t strings now, just like regex and glob,
and it always keeps a pointer on the character inside the string,
rather than operating on a single character.
As a result, just drop the ifdef's for Cygwin. The code is
non-portable now anyway...
git://sourceware.org / newlib-cygwin.git / log