summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Replace bogus resize-window-to-clear-screen logic.cygnal-3-1-6-98-branchKaz Kylheku2020-10-112-56/+23
| | | | | | | | | | | | | | | | | | | | | | | | | | | | This removes (ab)uses of SetConsoleScreenBufferSize which sometimes cause cmd.exe to badly misbehave. It probably doesn't like the closely spaced timing of shrinking the window down to one line followed by a restore of the size. Instead we just output newlines to clear the window. * winsup/cygwin/fhandler.h (dev_console::scroll_window): Member function declaration removed. (dev_console::clear_should_scroll): New member function declared. * winsup/cygwin/fhandler_console.cc (dev_console::scroll_window): Member function removed. (dev_console::clear_should_scroll): New member function. Performs only the test that was performed by scroll_window, not the actual scrolling. The scrolling is now done in the caller in the fhandler_console class. (fhandler_console::clear_screen): Call con.clear_should_scroll instead of con.scroll_window. If this returns false, act as before. Otherwise, clear the screen by scrolling the window. This is done not by making SetConsoleScreenBufferSize calls to shrink and restore the window, but by earnestly emitting a number of carriage returns equal to the vertical screen size and then restoring the cursor position.
* Revert "Fix console clear screen if buffer is full"Kaz Kylheku2020-10-111-25/+16
| | | | | | | | | | | | | | | This reverts commit 99a3f266c15af5bbb9a8cacda63ce11b094c10d1. Unfortunately Corinna's August 3, 2016 is unsatisfactory, and interferes with our patch for the issue cherry picked from cygnal-2.5.2 into cygnal-2.9.0-branch. When the screen-clearing escape sequence ESC[2J is issued material below the cursor is not cleared properly. For instance, in the TXR Lisp listener, the Ctrl-L command fails to clear the screen at all, because it first homes the cursor to the top line of the screen and then issues the clear screen.
* Use Windows values in passwd and HOME env var.Kaz Kylheku2020-10-114-18/+25
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | In this patch, the path to the Windows command interpreter is reported in the shell field of struct passwd by getpwent by default rather than "/bin/bash". The value of USERPROFILE is used for the home field rather than "/home/<user>". Also, the HOME environment variable is stuffed with a copy of USERPROFILE. The HOME issue solves the following problem: some OSS programs on Windows, such as Vim, respond to a HOME variable. If it has garbage contents that make no sense like "/home/bob", they don't behave well. * winsup/cygwin/grp.cc (pwdgrp::init_grp): Initialize new pwd_sep member. * winsup/cygwin/passwd.cc (pwdgrp::parse_passwd): Use pwd_sep rather than hard-coded colon. (pwdgrp::init_pwd): Initialize pwd_sep. * winsup/cygwin/pwdgrp.h (class pwdgrp): New member, pwd_sep. * winsup/cygwin/uinfo.cc (cygheap_user::ontherange): Copy value of USERPROFILE into HOME. (pwdgrp::next_num, pwdgrp::fetch_account_from_line): Use pwd_sep rather than ':'. (pwdgrp::next_num, pwdgrp::fetch_account_from_windows): Get real Windows shell as default shell field. Get USERPROFILE as home directory. Use '|' as the field separator because these fields contain colons. Set pwd_sep to '|'.
* Remove env var translation; PATH semicolon separated.Kaz Kylheku2020-10-112-99/+9
| | | | | | | | | | | | | | | * winsup/cygwin/environ.cc (conv_envvars): Static array removed. (conv_start_chars): Likewise. (getwinenv): Function gutted to just return NULL. No environment vars require conversion. (match_first_char): Static function removed. (build_env): Removed logic for eliminating those variables that require donversion, since there are no such variables and the needed functions and arrays are gone. * winsup/cygwin/environ.cc (find_exec): Recognize semicolon as PATH separator.
* Arrange for text mode on stdin, stdout and stderr.Kaz Kylheku2020-10-111-0/+20
| | | | | | * newlib/libc/stdio/findfp.c (__sinit): Add the Cygwin-specific __SCLE (stream convert line endings) flag when calling the std function to initializing the reent structure's _stdin, _stdout and _stderr members.
* Console: integrate end-of-line hack with cursor positioning.Kaz Kylheku2020-10-111-8/+23
| | | | | | | | | | | | | | | | | | | | | | | | | | The previous work for simulating the beyond-right-edge of screen cursor position is incomplete without integrating this into the cursor positioning routines. By making the cursor routines aware of the eat_newline flag, we allow cursor movements to work correctly with regard to this simulated position. * winsup/cygwin/fhandler_console.cc (fhandler_console::cursor_set): If the X position is beyond the right edge of the screen, then set the cursor to the start of the following line, rather than clipping to the right edge, set the eat_newline flag, indicating that the the true position is actually one character beyond the previous line. In all other cases, clear the eat_newline flag. (fhandler_console::cursor_rel): Do not apply the delta vector to the raw Win32 cursor position; call cursor_get and apply it to the virtual cursor position which takes into account the eat_newline flag. (fhandler_console::cursor_get): Take into account the eat_newline flag. If it is set, then report an adjusted position that is one column beyond the end of the previous line.
* Symlink creation defaults to Windows shortcuts.Kaz Kylheku2020-10-111-1/+1
| | | | | * winsup/cygwin/globals.cc (allow_winsymlinks): Change initial value from WSYM_sysfile to WSYM_lnk.
* Respond to CYGNAL environment variable, not CYGWIN.Kaz Kylheku2020-10-111-8/+8
| | | | | | | | | * winsup/cygwin/environ.cc (set_winsymlinks): Refer to CYGNAL variable in diagnostic message. (parse_options): Pass "CYGNAL" to setenv. (_addenv, environ_init): Call parse_options if variable name is "CYGNAL" rather than "CYGWIN". Comments updated. (spenvs): Change CYGWIN_DEBUG to CYGNAL_DEBUG.
* Implement sh -> cmd.exe translation hack.Kaz Kylheku2020-10-111-0/+15
| | | | | * winsup/cygwin/spawn.cc (spawnve): Rewrite /bin/sh -c cmd invocations to use cmd.exe /c cmd instead.
* Default stdio streams to text mode.Kaz Kylheku2020-10-111-0/+4
| | | | | * newlib/libc/stdio/flags.c (__sflags): If mode is not O_BINARY, then add O_TEXT.
* Implement drive-relative paths and per-drive working dir.Kaz Kylheku2020-10-111-1/+40
| | | | | | | | | | | | | | | | | | | | | | * winsup/cygwin/path.cc (normalize_win32_path): When a drive-relative path is normalized, look up the remembered working directory of that drive in the environment. A drive-relative path is, for example, "C:file.txt". Or just "C:", with no component. If there is no path for the drive in the environment, then the root directory is used, and the "C:" part thus becomes "C:\", causing the path to refer to "C:\file.txt". Otherwise the path is inserted, with a backslash, like "C:\users\bob\file.txt". The Windows convention for storing these per-drive paths in the environment is to use environment variables based on drive letters. For instance the path for the C drive is stored in the environment variable "!C:" (bang, letter, colon). The path includes the C:\ prefix. (cwdstuff::override_win32_cwd): Add the behavior of associating the current working directory with its drive (if it is a current working directory based on a drive). For instance if the overriding cwd is "C:\Users", then the "C:\Users" path is stored into the "!C:" environment variable.
* Remove conversion of native paths to POSIX mount points.Kaz Kylheku2020-10-111-74/+1
| | | | | | | | | | | | We don't want the behavior in Cygnal whereby a native path like C:\path\to\app is converted to /app in getcwd and other situations, or C:\random\path is converted to /cygdrive/c/random/path. * winsup/cygwin/mount.cc (mount_info::conv_to_posix_path): Remove entire section of code which scans mount points, mapping native paths to their mount points paths. All we do is "slashify" and exit.
* First steps toward native path handling.Kaz Kylheku2020-10-114-31/+73
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - /proc and /dev are are still available, accessed as proc:/ and dev:/ - All other paths are native, and do not "see" virtual Cygwin items; /cygdrive is just C:\cygdrive (if the current drive is C). - chdir() to a virtual directory like dev:/ or proc:/ results in errno EOPNOTSUPP. * winsup/cygwin/mount.cc (mount_info::conv_to_win32_path): Takes new bool argument, hide_cygwin_virtuals. If true all that the function does is "backslashify" the path. (mount_info::add_item): Special hack inserted here so that we can create the virtual root directory which holds dev and proc and whatnot, and is passed here as "/". We cannot let this go through normalize_posix_path any more because it will turn to C:\. * winsup/cygwin/mount.h (mount_info::conv_to_win32_path): Declaration updated. * winsup/cygwin/path.cc (is_posix_space): New static function: tests for paths in special spaces, currently "dev:/" and "proc:/". Used by normalize_posix_path. (normalize_posix_path): Any path that doesn't satisfy the is_posix_space test is treated as Win32. Since the bulk of the code is now only used for these spaces, the relative path handling is not required and a the corresponding block of code is removed. Paths satisfying is_posix_space are transformed. I.e. the underlying path resolution machine in the path_conv class still recognizes /proc and /dev. It's just that these will not occur, because normalize_posix_path will convert them to references with drive names. (path_conv::check): Pass the is_msdos flag down to mount_info::conv_to_win32_path as the new argument. Thus if normalize_posix_path indicates a native path, this function will hide the virtual spaces. Also, we add MOUNT_NOPOSIX and MOUNT_NOACL to the object's mount_flags. This is used in chdir. (normalize_win32_path): A small piece of logic works against our plan here: it checks for the leading forward slash on the path, and prevents such paths from being converted to Win32 paths with a drive reference. We eliminate this test, and treat paths unconditionally. (chdir): Here, if the path is not native, we return EOPNOTSUPP. Thus it is impossible to chdir into Cygwin virtual directories like /dev (now referenced as dev:/). They can be listed but not turned into the current directory. Eventually we want chdir to actually set the Win32 current directory of the process; that can't work for virtual dirs. * winsup/cygwin/path.h (path_conv::is_native): New inline accessor which tests for the MOUNT_NOPOSIX flag.
* Small fixes in get_cmd_exe_path.Kaz Kylheku2020-10-111-11/+13
| | | | | | | | | | * winsup/cygwin/spawn.cc (init_cmd_exe_path): Restructure code to initialize rather than assign nchars. Include backslash in cmd.exe name; then it can be omitted from the size calculation and sprintf. Do not allocate an excess byte for the string. Thanks to user forsvarir of the code review stackechange. Also reformatted to the GNU style used inside Cygwin. (init_cmd_exe_path): Remove spurious whitespace.
* More secure way of obtaining command interpreter.Kaz Kylheku2020-10-113-4/+30
| | | | | | | | | | | | | | | | | | Instead of relying on the COMSPEC environment variable, what we can do is assume that the program is called "cmd.exe", and then look for it in the directory reported by the GetSystemDirectoryA Win32 function in kernel32.dll. * winsup/cygwin/path.h (get_cmd_exe_path): New function declared. * winsup/cygwin/spawn.cc (av::setup): Use get_cmd_exe_path instead of getenv("COMSPEC"). (cmd_exe_path): New static variable. (init_cmd_exe_path): New static function. (get_cmd_exe_path): New function. * winsup/cygwin/syscalls.cc (system, getusershell, popen): Use get_cmd_exe_path instead of getenv("COMSPEC").
* Use COMSPEC env var, not hard-coded CMD.EXE path.Kaz Kylheku2020-10-113-19/+25
| | | | | | | | | | | | | | | | | | | | It is with some reluctance I make this change, due to the security implications of relying on environment variables. But we can't have a hard-coded path. * winsup/cygwin/include/paths.h (_PATH_CMDEXE): Macro removed. * winsup/cygwin/spawn.cc (av::setup): Use COMSPEC environment variable instead of hard-coded path. If missing, bail with errno set to EOPNOTSUPP. * winsup/cygwin/syscalls.cc (system): Use COMSPEC environment variable. If missing, return -1. (getusershell): Eliminate static array of shell names. If shell_index is zero, return value of COMSPEC env var, if it exists, and increment shell_index to 1. (popen): Use COMSPEC and if that is missing, set errno to EOPNOTSUPP and return NULL.
* Fix previous VT100 fix.Kaz Kylheku2020-10-111-11/+11
| | | | | | | | | We must only set or clear the eat_newline flag if we output something. * winsup/cygwin/fhandler_console.cc (fhandler_console::write_normal): Move the flag setting code inside the if statement which performs output.
* Fix spawned process window not foregrounding.Kaz Kylheku2020-10-112-1/+7
| | | | | | | | | | | | | | | | | | | | | | | | This patch addresses an issue whereby the window of a process created with CreateProcess fails to come to the foreground. This occurs when the calling process itself hasn't run any Windows event processing loop. A repro test case is to make a program with a main, and and call CreateProcess to spawn calc.exe or notepad.exe before doing anything else. It turns out that a dummy call to TranslateMessage makes this issue goes away. If such a call is made before CreateProcess, then the spawned process' window comes up in the foreground as expected. * winsup/cygwin/Makefile.in (DLL_IMPORTS): We need to link in user32.dll to call TranslateMessage. Condense the multiple ${shell ...} call repetition with a foreach. * winsup/cygwin/spawn.cc (child_info_spawn::worker): Do the dummy TranslateMessage call before the section of code that calls CreateProcess or CreateProcessAsUser.
* Console: provide VT100-like end-of-line print behavior.Kaz Kylheku2020-10-112-0/+24
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | On the majority of VT100-like terminals, when you print a character in the last column, the cursor sits in a kind of "limbo", as if the cursor position were one column past the edge of the screen. Thus when a line feed is output next, the cursor is then at the start of the next line. The Win32 console write doesn't do this; the cursor position wraps to the start of the next line. If a newline is put out, it translates to an extra newline going to the next-next line. This spoils the behavior of programs which depend on the VT100 behavior. In this patch, the VT100 behavior is simulated as follows. A new flag in the dev_console structure is set when a character is output to the last column. This flag is then observed in order to discard a newline (or rather any character which is mapped mapped to the DWN action). * winsup/cygwin/fhandler.h (class dev_console): New boolean data member, eat_newline. This indicates that if a character is output which moves down to the next line, it should be discarded rather than sent to the console. * winsup/cygwin/fhandler_console.cc (fhandler_console::write_normal): Set the eat_newline flag if the text output ends up at column zero. Honor the eat_newline flag when processing a DWN character, and clear it when processing certain other control characters.
* Report as CYGNAL in uname syscall.Kaz Kylheku2020-10-111-2/+2
| | | | | | | | This allows applications to distinguish whether they are running on the stock Cygwin or Cygnal. * winsup/cygwin/uname.cc (uname, uname_x): sysname is now filled in with CYGNAL_ prefix rather than CYGWIN_.
* Cygnal apps use cmd.exe, not /bin/sh.Kaz Kylheku2020-10-103-47/+13
| | | | | | | | | | | | | | | | | | | * winsup/cygwin/include/paths.h (_PATH_CMDEXE): New preprocessor symbol. * winsup/cygwin/spawn.cc (av_setup): Use _PATH_CMDEXE rather than "/bin/sh". * winsup/cygwin/syscalls.cc (system): Spawn _PATH_CMDEXE with /c option rather than /bin/sh. (ETC_SHELLS): Preprocessor symbol removed. (shell_fp): Global variable removed. (getusershell): Don't open ETC_SHELLS, just march through static array of shell names. That array contains only one entry: _PATH_CMDEXE. (setusershell, endusershell): Remove references to shell_fp. (popen): Exec _PATH_CMDEXE rather than "/bin/sh", and the option is /c.
* When spawning, don't try to make invisible window.Kaz Kylheku2020-10-101-2/+0
| | | | | | | | | * winsup/cygwin/spawn.cc (child_info_spawn::worker): Do not call fhandler_console::need_invisible. It's not working properly. In an application which has no console because it was compiled -mwindows, calling this funcion causes a visible console window to appear. We don't need this in Cygnal; the Microsoft spawn functions don't pop up such windows.
* Use wShowWindow when calling CreateProcess.Kaz Kylheku2020-10-101-1/+2
| | | | | | * winsup/cygwin/spawn.cc (child_info_spawn::worker): Add STARTF_USESHOWWINDOW to dwFlags of the STARTUPINFOW structure, and set wShowWindow to SW_SHOWNORMAL.
* Cygwin: avoid GCC 10 error with -Werror=narrowingJon Turney2020-10-102-3/+1
| | | | | | | | | | | | | | | Cherry-pick of c5bdf60ac46401a51a7e974333d9622966e22d67. ../../../../src/winsup/cygwin/pinfo.cc: In member function 'DWORD pinfo::status_exit(DWORD)': ../../../../src/winsup/cygwin/ntdll.h:21:68: error: narrowing conversion of '-536870295' from 'NTSTATUS' {aka 'int'} to 'unsigned int' [-Wnarrowing] ../../../../src/winsup/cygwin/pinfo.cc:136:10: note: in expansion of macro 'STATUS_ILLEGAL_DLL_PSEUDO_RELOCATION' ../../../../src/winsup/cygwin/sigproc.cc: In member function 'DWORD child_info::proc_retry(HANDLE)': ../../../../src/winsup/cygwin/ntdll.h:21:68: error: narrowing conversion of '-536870295' from 'NTSTATUS' {aka 'int'} to 'unsigned int' [-Wnarrowing] ../../../../src/winsup/cygwin/sigproc.cc:1120:10: note: in expansion of macro 'STATUS_ILLEGAL_DLL_PSEUDO_RELOCATION' NT error statuses seem to be variously DWORD (unsigned) or NTSTATUS (signed)? So use the one which doesn't cause problems here.
* Cygwin: avoid GCC 10 error with -Werror=parenthesesJon Turney2020-10-101-1/+1
| | | | | | | Cherry-pick of 3bb579a43c09d18217a19035054817591eb8d870. ../../../../src/winsup/cygwin/fhandler_socket_inet.cc: In member function 'ssize_t fhandler_socket_wsock::send_internal(_WSAMSG*, int)': ../../../../src/winsup/cygwin/fhandler_socket_inet.cc:1381:69: error: suggest parentheses around assignment used as truth value [-Werror=parentheses]
* Cygwin: avoid GCC 10 error with -Werror=narrowingJon Turney2020-10-101-2/+2
| | | | | | | | | | | | Cherry-pick of 129c9844a6c40d5dee658151c2ff2c461a5a1365. ../../../../src/winsup/cygwin/fhandler_console.cc: In member function 'const unsigned char* fhandler_console::write_normal(const unsigned char*, const unsigned char*)': ../../../../src/winsup/cygwin/fhandler_console.cc:2782:8: error: narrowing conversion of '-2' from 'int' to 'long unsigned int' [-Wnarrowing] ../../../../src/winsup/cygwin/fhandler_console.cc:2786:8: error: narrowing conversion of '-1' from 'int' to 'long unsigned int' [-Wnarrowing] ../../../../src/winsup/cygwin/fhandler_console.cc:2836:8: error: narrowing conversion of '-2' from 'int' to 'long unsigned int' [-Wnarrowing] ../../../../src/winsup/cygwin/fhandler_console.cc:2840:8: error: narrowing conversion of '-1' from 'int' to 'long unsigned int' [-Wnarrowing] A mbtowc_p function returns an int, so that seems the correct type to use here.
* Cygwin: winlean.h: remove most of the extended memory APIKen Brown2020-10-104-43/+17
| | | | | | | | | | | | | | | | | | | Cherry pick of c1f7c4d1b6d7424d7081fd382ac83a631b7671ff. This was added as a temporary measure in commit e18f7f99 because it wasn't yet in the mingw-w64 headers. With one exception, it is now in the current release of the headers (version 8.0.0), so we don't need it in winlean.h. The exception is that VirtualAlloc2 is declared conditionally in <w32api/memoryapi.h>, but the compilation of Cygwin requires it to always be declared, even though it will only be executed on systems that support it. So retain the declaration in winlean.h. And add "WINAPI" to the declaration, as in memoryapi.h. Add a check that version >= 8 of the mingw-w64 headers is intalled. Also revert commit 3d136011, which was a related temporary workaround.
* Cygwin: strace: ignore GCC exceptionscygwin-3_1_7-releaseKen Brown2020-08-201-0/+8
| | | | | | | Any C++ app that calls 'throw' on 64-bit Cygwin results in an exception of type STATUS_GCC_THROW (0x20474343) generated by the C++ runtime. Don't pollute the strace output by printing information about this and other GCC exceptions.
* Cygwin: add header defining GCC exception codesKen Brown2020-08-202-9/+20
| | | | | Include it in exceptions.cc instead of defining the exception codes there.
* Cygwin: main exception handler (64-bit): continue GCC exceptionsKen Brown2020-08-182-0/+23
| | | | | | | | | | | | | | | | | This is necessary in order to be consistent with the following comment in the definition of _Unwind_RaiseException() in the GCC source file libgcc/unwind-seh.c: The exception handler installed in crt0 will continue any GCC exception that reaches there (and isn't marked non-continuable). Previously we failed to do this and, as a consequence, the C++ runtime didn't call std::terminate after an unhandled exception. This fixes the problem reported here: https://cygwin.com/pipermail/cygwin/2019-October/242795.html https://sourceware.org/pipermail/cygwin/2020-August/245897.html
* libm/stdlib: Realloc when shrinking by 2* or moreKeith Packard via Newlib2020-08-171-3/+3
| | | | | | This reduces memory usage when reallocating objects much smaller. Signed-off-by: Keith Packard <keithp@keithp.com>
* libm/stdlib: don't read past source in nano_reallocKeith Packard via Newlib2020-08-171-2/+4
| | | | | | | Save the computed block size and use it to avoid reading past the end of the source block. Signed-off-by: Keith Packard <keithp@keithp.com>
* Cygwin: pty: Change the timing of set_locale() call again.Takashi Yano via Cygwin-patches2020-08-172-4/+18
| | | | | | | | - After commit 095972ce5b1d319915501a7e381802914bed790c, charset conversion in mintty is broken if charset is set to other than UTF-8. This seems to be caused because mintty does not set locale yet at fork() call. This patch changes the timing of set_locale() call again to avoid this issue.
* Cygwin: pty: Change the timing of setup_locale() call.Takashi Yano via Cygwin-patches2020-08-131-4/+4
| | | | | | - If native app is exec()'ed in a new pty, setup_locale() loses the chance to be called. For example, with "mintty -e cmd", charset conversion does not work as expected. This patch fixes the issue.
* libc/stdlib: Fix build failure in nano_callocCraig Blackmore2020-08-131-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | commit 588a5e1ddebdf6d74391c7409680ea20e050c0e1 added a non-reentrant call to nano_malloc which causes a build failure if INTERNAL_NEWLIB is defined. Here is a snippet of the error: In file included from .../newlib/newlib/libc/stdlib/nano-mallocr.c:38: .../newlib/newlib/libc/include/malloc.h:42:25: note: expected 'struct _reent *' but argument is of type 'ptrdiff_t' {aka 'int'} 42 | extern void *_malloc_r (struct _reent *, size_t); | ^~~~~~~~~~~~~~~ .../newlib/newlib/libc/stdlib/nano-mallocr.c:67:22: error: too few arguments to function '_malloc_r' 67 | #define nano_malloc _malloc_r | ^~~~~~~~~ .../newlib/newlib/libc/stdlib/nano-mallocr.c:456:11: note: in expansion of macro 'nano_malloc' 456 | mem = nano_malloc(bytes); | ^~~~~~~~~~~ In file included from .../newlib/newlib/libc/stdlib/nano-mallocr.c:38: .../newlib/newlib/libc/include/malloc.h:42:14: note: declared here 42 | extern void *_malloc_r (struct _reent *, size_t); | ^~~~~~~~~ .../newlib/newlib/libc/stdlib/nano-mallocr.c:43: warning: "assert" redefined 43 | #define assert(x) ((void)0) | This patch adds a missing RCALL to the args when calling nano_malloc from nano_calloc, so that if the call is reentrant, reent_ptr is passed as the first argument. The variable `bytes` (also added in 588a5e1d) has been changed from a `ptrdiff_t` to `malloc_size_t` as it does not need to be signed. It is used to store the product of two unsigned malloc_size_t variables and then iff there was no overflow is it passed to malloc and memset which both expect size_t which is unsigned. Signed-off-by: Craig Blackmore <craig.blackmore@embecosm.com>
* libc/stdlib: Use __builtin_mul_overflow for reallocarray and callocKeith Packard via Newlib2020-08-123-15/+22
| | | | | | | This built-in function (available in both gcc and clang) is more efficient and generates shorter code than open-coding the test. Signed-off-by: Keith Packard <keithp@keithp.com>
* libm/machine/riscv: Add custom fma/sqrt functions when supported [v2]Keith Packard via Newlib2020-08-127-5/+237
| | | | | | | Check for HW FMA and SQRT support and use those instructions in place of software implementations. Signed-off-by: Keith Packard <keithp@keithp.com>
* Cygwin: pty: Add a workaround for issue of starting a lot of mintty.Takashi Yano2020-08-111-4/+11
| | | | | | | | | - If a lot of mintty are started in a short time from a mintty, some of them hang with empty screen, crash immediately or hang on exiting mintty. The following report seems to be related to this issue. https://cygwin.com/pipermail/cygwin/2020-August/245751.html The cause is not clear at all, but this patch seems to solve the issue.
* libm/machine/arm: Add optimized fmaf and fma when availableKeith Packard via Newlib2020-08-106-5/+125
| | | | | | | | | | | When HAVE_FAST_FMAF is set, use the vfma.f32 instruction, when HAVE_FAST_FMA is set, use the vfma.f64 instruction. Usually the compiler built-ins will already have inlined these instructions, but provide these symbols for cases where that doesn't work instead of falling back to the (inaccurate) common code versions. Signed-off-by: Keith Packard <keithp@keithp.com>
* libm: Detect fast fmaf supportKeith Packard via Newlib2020-08-101-0/+8
| | | | | | | Anything with fast FMA is assumed to have fast FMAF, along with 32-bit arms that advertise 32-bit FP support and __ARM_FEATURE_FMA Signed-off-by: Keith Packard <keithp@keithp.com>
* libm: ARM without HW double does not have fast FMAKeith Packard via Newlib2020-08-101-1/+1
| | | | | | | | 32-bit ARM processors with HW float (but not HW double) may define __ARM_FEATURE_FMA, but that only means they have fast FMA for 32-bit floats. Signed-off-by: Keith Packard <keithp@keithp.com>
* libm/math: ensure that expf(-huge) sets FE_UNDERFLOW exceptionKeith Packard via Newlib2020-08-101-1/+1
| | | | | | | | It was calling __math_uflow(0) instead of __math_uflowf(0), which resulted in no exception being set on machines with exception support for float but not double. Signed-off-by: Keith Packard <keithp@keithp.com>
* Cygwin: cygserver: build with -Wimplicit-fallthrough=5Ken Brown2020-08-074-3/+5
| | | | Define the pseudo keyword 'fallthrough' in woutsup.h to support this.
* Cygwin: Use documented QueryWorkingSetEx() in dumperJon Turney2020-08-071-29/+7
| | | | | In dumper, use the documented QueryWorkingSetEx(), rather than the undocumented NtQueryVirtualMemory() with MemoryWorkingSetExInformation.
* MSP430: Increase the amount of main memory available in sim ld scriptsJozef Lawrynowicz2020-08-072-2/+2
| | | | | | | | | | | | | | | | The main memory region of the GDB simulator ends at address 0xFFBF, but the simulator linker scripts do not make full use of this available memory. >From 61f3d212741acee583e21ff2c2808775584ecad6 Mon Sep 17 00:00:00 2001 From: Jozef Lawrynowicz <jozef.l@mittosystems.com> Date: Mon, 3 Aug 2020 19:38:23 +0100 Subject: [PATCH 2/2] MSP430: Increase the amount of main memory available in sim ld scripts The main memory region of the GDB simulator ends at address 0xFFBF, but the simulator linker scripts do not make full use of this available memory.
* MSP430: Word align __*_array_start symbols in sim linker scriptsJozef Lawrynowicz2020-08-072-0/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | __{preinit,init,fini}_array_start symbols must be word aligned in linker scripts. If the section preceding the __*_array_start symbol has an odd size, then a NULL byte will be present between the start symbol and the .*_array section itself, when the section gets automatically word-aligned. This results in a branch to an invalid address when the CRT startup code tries to run through the functions listed in the array sections. >From de115144d05ecbaa82c9c737cc261715ca4b7d67 Mon Sep 17 00:00:00 2001 From: Jozef Lawrynowicz <jozef.l@mittosystems.com> Date: Mon, 3 Aug 2020 19:09:46 +0100 Subject: [PATCH 1/2] MSP430: Word align __*_array_start symbols in sim linker scripts __{preinit,init,fini}_array_start symbols must be word aligned in linker scripts. If the section preceding the __*_array_start symbol has an odd size, then a NULL byte will be present between the start symbol and the .*_array section itself, when the section gets automatically word-aligned. This results in a branch to an invalid address when the CRT startup code tries to run through the functions listed in the array sections.
* Cygwin: cygserver: build with -Wimplicit-fallthrough=4 -WerrorCorinna Vinschen2020-08-071-1/+1
| | | | Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
* Cygwin: utils: build with -Wimplicit-fallthrough=4 -WerrorCorinna Vinschen2020-08-072-2/+4
| | | | Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
* Cygwin: utils: convert usage() to proper noreturn function throughoutCorinna Vinschen2020-08-0723-111/+55
| | | | Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
* Cygwin: utils: cygcheck: avoid GCC warning concatenating stringsCorinna Vinschen2020-08-071-1/+5
| | | | Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
generated by cgit v1.2.3 (git 2.25.1) at 2025-07-21 07:43:16 +0000