iOS does not support Thread Local Storage.
Disabling it for now until a solution is found.
Print a compiler warning informing developers about this issue.
The lpnSize parameter for GetComputerNameEx specifies the total
size of the buffer (in characters).
However, the current code calculated the amount of bytes.
Since only GetComputerNameExA was used and because sizeof(CHAR) == 1
the result was correct but the math was wrong.
Credit goes to @byteboon
Depending on the windows target version (_WIN32_WINNT), the used
SDK and the build configuration the linker will see multiple
libraries exporting the same symbols.
To prevent ugly hacks (e.g. modifying cmake's default system
libraries or fragile library linking order chains) we prefix
these functions with "winpr_" and create corresponding defines
to keep the current api names.
On input, the lpnSize [in, out] parameter for GetComputerNameEx()
specifies the total size of the buffer (in characters).
Several functions in ntlm.c were off by one which caused ntlm to fail
if the netbios hostname's strlen was exactly MAX_COMPUTERNAME_LENGTH.
PathMakePathA:
- This function had an endless loop if no native delimiter was in the string
- Use SHCreateDirectoryExA on Windows
- Replaced old code with a new implementation
TestWLog:
- Windows has no "/tmp" by default
- Use GetKnownPath(KNOWN_PATH_TEMP) for the WLog "outputfilepath"
- Use correct SetLastError values in GetModuleFileName
- Fix wrong return codes in GetModuleFileName
- Build the TestLibraryA/TestLibraryB libraries always shared and
put them in the test output directory
- TestLibraryGetModuleFileName always returned success
- Improve TestLibraryGetModuleFileName to also check last error values
and insufficient buffer sizes
- Change TestLibraryGetProcAddress and TestLibraryLoadLibrary to load
the TestLibrary from the test executable's directory
TestNtCreateFile, TestPipeCreateNamedPipeOverlapped
- These tests are currently only expected to succeed on _WIN32
- Also reflect the reverse meaning of this fact in the return values
TestSynchWaitableTimer, TestSynchWaitableTimerAPC:
- These tests are currently expected to fail on __APPLE__
- Also reflect the reverse meaning of this fact in the return values
This logic makes sure that we don't forget to fix the tests if the
corresponding WinPR implementations are fixed.
TestLibrary:
- TestLibraryA and TestLibraryB must always get built as shared libraries
Let the compiler know that we're comparing a volatile value.
Otherwise the compiler might nuke the comparison operation
and produce code that will spin endlessly.
The SYNCHRONIZATION_BARRIER_FLAGS_SPIN_ONLY flag caused this test
to run extremely long if the system has very few processors.
Although this is expected (thread starvation) this will cause a
intolerably long execution time for automated tests.
Changed the number of threads to be calculated dyamically based
on the number of processors.
Also do proper cleanup to prevent memory leaks.
The current experimental/incomplete WinPR timer queue implementation
has several race conditions.
This commit fixes a segfault caused by not unklinking freed timers
from the timer queue timers list.
pool:
- the winpr implementation fallback was not used on older windows editions
- drop useless and conflicting TP_CALLBACK_ENVIRON_V3
- fix race conditions by using use proper one-time initialization
- on win32 WinPR tried to load several pool/callback_environment functions
from kernel32.dll but since these are defined as inline functions in the
windows headers, no windows edition has ever exported them in any dll.
- removed callback_environment.c and added corresponding static inline
function to pool.h
- fix segfault in TestPoolWork: CloseThreadpoolWork() must not be called
if there is a cleanup group associated with the work object since calling
CloseThreadpoolCleanupGroupMember() already releases the work object
sync:
- The windows headers incorrectly define InitializeCriticalEx support if
_WIN32_WINNT >= 0x0403 instead of >= 0x0600 (Vista)
- created a compatible define to deal with this issue
On some operating systems sched_yield is a stub returning returning -1.
In that case use usleep which should at least trigger a context switch
if any thread is waiting.
Half of the tests expects strings in little endian byte order, half of
the tests expects byte order based on a current architecture, so it
obviously can't work on big endian machines. Wide char strings use
always little endian encoding since commit f722dc5. Use only strings
in little endian to make the tests endian-independent.
Since the current winpr implementation for overlapped operations is
incomplete and buggy, all affected functions will now fail if they are
called with a set FILE_FLAG_OVERLAPPED flag or a non-null pointer to
a OVERLAPPED structure.
winpr/nt:
- use proper one-time initialization on win32
- fix TestNtCreateFile
- fix broken/incomplete _RtlAnsiStringToUnicodeString
- unimplemented functions return appropriate error codes
winpr/pipe:
- improved TestPipeCreateNamedPipe
- rewrite the completely broken TestPipeCreateNamedPipeOverlapped test
rdtk:
- improve test and don't blindly return success
winpr/synch:
- fix race condition in TestSynchTimerQueue
winpr/ssspi:
- fix TestEnumerateSecurityPackages printf output
- fix TestQuerySecurityPackageInfo printf output
winpr/environment:
- fix GetEnvironmentStrings printf output
winpr/comm:
- unimplemented functions return appropriate error codes
winpr/io:
- unimplemented functions return appropriate error codes
winpr/thread:
- implement SwitchToThread() via sched_yield()
This patch fixes NTLM authentication to work properly on a big endian
machines. Freerdp exited with the following error without recent commits:
[09:50:20:914] [13821:13822] [ERROR][com.freerdp.core.transport] - BIO_read returned an error: error:14094438:SSL routines:ssl3_read_bytes:tlsv1 alert internal error
[09:50:20:914] [13821:13822] [ERROR][com.freerdp.core] - freerdp_set_last_error ERRCONNECT_CONNECT_TRANSPORT_FAILED [0x2000D]
[09:50:20:914] [13821:13822] [ERROR][com.freerdp.client.x11] - Freerdp connect error exit status 1
https://github.com/FreeRDP/FreeRDP/issues/2520
Data in ntlm_av_pair_list are accessed directly, which doesn't work on
big endian machines currently. The recieved data are stored as little
endian. Use conversion macros from endian.h to load and store the data
properly.
https://github.com/FreeRDP/FreeRDP/issues/2520
All WCHAR strings are stored as little endian after commit 12dfc5e9,
therefor CharUpperBuffW and CharLowerBuffW have to be changed appropriately
in order to fix NTLM authentication.
https://github.com/FreeRDP/FreeRDP/issues/2520
Unicode conversions doesn't work on big endian machines currently.
The strings are stored as little endian. Use conversion macros from
endian.h to load and store the data properly.
Let's use wide char strings always as little endian. It seems that
Windows API also always expects data to be little endian, so it
makes sense to require wide char strings as little endian also.
The patches fixes transformations between UTF8 and UTF16 only, which are
used by freerdp. UTF32 transformations are not used by freerdp.
https://github.com/FreeRDP/FreeRDP/issues/2520
win32/msvc cc does not recognize the %z format specifier which caused
invalid references and segfaults on win32.
Until FreeRDP gets format specifier macros we'll cast size_t to
unsigned long and use the %lu specifier.
Also simplified winpr_backtrace_symbols() a little bit and fixed it
to allocate the correct amount of bytes for the return buffer.
On Windows we seem to have to load the TestLibrary[AB] test libraries
from in same folder the test executable runs.
Also removed the empty RemoveDllDirectory, SetDefaultDllDirectories,
AddDllDirectory tests and the redundant FreeLibrary test.
TestLibrary now works and succeeds on Win32.
sadasd
- PathCchFindExtensionA had an off-by-one error when verifying the
required null termination
- TestPathCchFindExtension used unicode strings when testing the
*A (ASCII) functions
- The PathAllocCombineW implementation (which is still buggy has
hell) used strlen to calculate the lenght of unicode strings
TestPath now succeeds on WIN32
GetLastError() was not always checked for ERROR_PIPE_CONNECTED which
indicates success if ConnectNamePipe returns FALSE.
TestPipe now also succeeds on Win32
TestSynchTimerQueue:
- fixed race condition
TestSynchWaitableTimerAPC:
- Use WaitForSingleObjectEx since the thread must be in an alterable state
TestSynch is now expected to succeed on WIN32
- Mutex is recursive on Windows; as a consequence we have to use
the pthread PTHREAD_MUTEX_RECURSIVE type
- Adapt MutexCloseHandle accordingly
- ReleaseMutex returned TRUE even if pthread_mutex_unlock failed
- Fixed and improved the TestSynchMutex ctest
SetEventFileDescriptor overrides the internal file descriptor of the
event but didn't close it. Now if the descriptor is closed if it isn't
marked as attached.
comm tests require a serial device for testing. If the test environment
isn't available the tests will return errors therefore the tests are
now disabled per default. They can be (re-)enabled by using the cmake
option BUILD_COMM_TESTS.
If a target is linked against libraries with cmake
(target_link_libraries) and the libraries are not marked as PRIVATE
they are "exported" and in case a other target is linked against this
target it is also linked against *all* (not private) libraries.
Without declaring private libraries PRIVATE a lot of over linking
(linking against unneeded libraries) was done.
- Added missing ConvertFromUnicode checks
- If ConvertToUnicode allocates memory, guarantee the null termination
similar to ConvertFromUnicode's implementation
- Fixed some TestUnicodeConversion.c CTest return values
- Added some CTests for ConvertFromUnicode and ConvertToUnicode
- Misc code and protocol hardening fixes in the surrounding code regions
that have been touched
ConvertToUnicode can't be used if the destination is a static buffer.
StandardName and DaylightName were invalid and timezone
redirection didn't work.
This regression was introduced with PR #3151
When nothing is declared, only export symbols defined
with WINPR_API or FREERDP_API defined.
Override this setting if BUILD_TESTING to allow tests
access to internal functions usually not exposed.
Extend winpr and client/common to support a new option "/buildconfig".
When used build the following build specific information is print:
* cmake options
* cflags
* compiler
* target architecture
* cmake build type
There exist two definitions of WINPR_FILE: in file/ and in nt/.
Both are different definitions used differently. Therefore split them
into WINPR_FILE and WINPR_NT_FILE.
Visual Studio 2010 use a compiler that supports only C89, which
only supports declaring variable at top of a local scope. Moving
scope variable to the top of function should solve this problem.
Don't set the identifier to "winpr" as default value because journald
will use the programs name as default if no identifier is set.
This way a program using WLog doesn't need to set an identifier (except
it want something different then it's name).
* only expose necessary functions and types in header
* don't expose appender internals
* add generic function WLog_ConfigureAppender to have the possibility
to configure appender specific settings
* detect appender availability if WLog_SetLogAppenderType or
WLog_Appender_New return FALSE or NULL respectively the appender isn't
available or the initialization failed. This is very useful for the
use with optional appenders.
* add Free to the appender interface. At the time of the Free the
appender is known and available so it can be called directly (instead
of calling the right function according to the type)
* make all appender internal function static
* all appenders return the generic wLogAppender type now. Typecasts
are internally done where necessary this abstracts the appenders more
cleanly
build-config.h should contain configure/compile time settings that are
relevant for projects that use FreeRDP.
For example the compiled in plugin search paths.
This appender allows to receive the logs over a network connection using UDP packets.
You can see the logs using a listening netcat, for example: nc -ul 127.0.0.1 20000.
* expose EnvironmentBlockToEnvpA
* cleanup includes in process.c
* removed unused "flag" variable in _CreateProcessExA
* make ProcessHandleCloseHandle static
winpr/libwinpr/utils/wlog/wlog.c: In function ‘WLog_PrintMessageVA’:
winpr/libwinpr/utils/wlog/wlog.c:234:7: warning: ‘status’ may be
used uninitialized in this function [-Wmaybe-uninitialized]
BOOL status;
^
gcc 4.9.2
wlog used to return an int but the only meaning
of the return value was:
* negative ... error
* 0 or positive ... success
but the positve returned value was 1 or some id of some
subsystem, nothing meaningful for the caller.
For a more meaningful returnvalue we now use BOOL.
If something goes wrong FALSE is returned.
It was supported in freerdp 1.0.2 but not supported in lastest master.
We should take empty password if it is explicitly specified with /v option.
If a password is not specified, we could first try SAM file. If the user entry does not exist, prompt for password.
The used method for initalizing and registering the handle creators
had certain disadvantages like that it was necessary to have the creators
initialized before the first call to CreateFile. In case of comm it
wouldn't have been possible to open an comm port without calling any
other call related function first.
Now the handle creators are initialized the first time CreateFile is
called. This way it is also possible to ensure a certain order of the
handler invocations.
This commit also splits out the client side named pipe code from
the generic file handling (now called generic.c instead of file.c) and
uses the new handle creator infrastructure to register.
When launching a new process stdin/stdout/stderr can be redirected by
passing the corresponding HANDLEs in the passed lpStartupInfo structure.
This is required for example if a pipe should be used as stdin/stdout.
If set in lpStartupInfo the file descriptor of the passed handle is
now used. The test was updated accordingly.
winsock.h pulls in a lot of defines and dependencies that are not
required and partially unwanted in winpr's core (for parts that are not
related to network). In order to get rid of this dependency and have an
independent defines for extended winpr functions the WINPR_FD_* defines
are used internally (and for exposed functions). Where required, like in
WSAEventSelect, the FD_* is mapped to WINPR_FD_*.
WSAEventSelect did ignore the lNetworkEvents argument.
In case this argument is 0, the non blocking socket must
be set to blocking again to mimic windows behavior.
Allows the WinPR HANDLE functions WaitForSingleObject and
WaitForMultipleObjects to signal in case of write events.
This is used by CreateFileDescriptor and SetEventFileDescriptor,
which got an API change accomodating for this new feature.
If SA_SIGINFO isn't set in the flags sa_handler is used
instead of sa_sigaction.
This fixes also the compiler warning:
FreeRDP/winpr/libwinpr/thread/process.c: In function ‘_CreateProcessExA’:
FreeRDP/winpr/libwinpr/thread/process.c:282:20: warning: assignment from
incompatible pointer type [enabled by default]
The calling thread of CreateProcess can be in any library and
can have arbitrary signal masks and handlers.
We now save the caller's mask and block all signals before forking.
After fork:
- child resets the handlers and unblocks all signals.
- parent restores the caller's original signal mask.
This patch contains:
* checks for malloc return value + treat callers;
* modified malloc() + ZeroMemory() to calloc();
* misc fixes of micro errors seen during the code audit:
** some invalid checks in gcc.c, also there were some possible
integer overflow. This is interesting because at the end the data are parsed
and freed directly, so it's a vulnerability in some kind of dead code (at least
useless);
** fixed usage of GetComputerNameExA with just one call, when 2 were used
in misc places. According to MSDN GetComputerNameA() is supposed to return
an error when called with NULL;
** there were a bug in the command line parsing of shadow;
** in freerdp_dynamic_channel_collection_add() the size of array was multiplied
by 4 instead of 2 on resize
Unless there is an invalid parameter or internal error, ResetEvent
must return TRUE, even if the object is in nonsignaled state.
Also fixed and improved the CTest TestSynchEvent to test for the
expected/correct SetEvent and ResetEvent return values.
Note: This commit does NOT really fix PathAllocCombine!
- print a warning message that the function is buggy and added a code
comment describing the issues
- fix misuse of the S_FALSE HRESULT in error conditions
- prevent some segfaults
- check result of HeapAlloc
Fortunately PathAllocCombine is unused in FreeRDP
Fix multiple inplementation errors including the incorrect usage of S_FALSE.
Make this function behave like the Windows 8 implementation and
the according MSDN specification.
- return S_OK if the extension was NOT found
- if no extension was found, ppszExt must point to the string's terminating null
- return E_INVALIDARG if pszPath is not null-terminated within the cchPath range
- return E_INVALIDARG if pszPath is NULL
- return E_INVALIDARG if ppszExt is NULL
- return E_INVALIDARG if cchPath is Zero
- return E_NOTIMPL instead of S_OK in PathPathCchFindExtensionW()
Also extended/fixed the TestPathCchFindExtension ctest
Fix incorrect usage of S_FALSE which was used to indicate errors
although it is a HRESULT success code.
Make this function behave like the Windows 8 implementation and
the according MSDN specification.
- return E_INVALIDARG instead of S_FALSE if pszPath is NULL
- return E_INVALIDARG instead of S_FALSE if pszMore is NULL
- return E_INVALIDARG if cchPath is zero
- return E_INVALIDARG if cchPath is greater than PATHCCH_MAX_CCH
- return HRESULT_FROM_WIN32(ERROR_FILENAME_EXCED_RANGE) if the combined
path size exceeds cchPath (this is the same HRESULT which the Windows
implementation returns in this case and which is referred to as
PATHCCH_E_FILENAME_TOO_LONG on msdn)
Also extended/fixed the TestPathCchAppend ctest
And again S_FALSE was incorrectly used to indicate an error.
Fix this and be more compatible to the Windows 8 implementation of
this function:
- return E_INVALIDARG instead of S_FALSE if pszPath is NULL
- return E_INVALIDARG instead of S_FALSE if cchPath < 4
- return E_INVALIDARG if cchPath > PATHCCH_MAX_CCH
- allow stripping of "\\?\c:" and don't require "\\?\c:\"
- verify that the character at the drive letter position is
actually an alphabetic character
- since the passed pszPath must not necessarily be null terminated
and we always have enough space after the memmove we can always
ensure the null termination of the stripped result
Also extended/fixed the TestPathCchStripPrefixctest
The HRESULT S_FALSE does not indicate an error:
- return E_INVALIDARG instead of S_FALSE
- return HRESULT_FROM_WIN32(ERROR_INSUFFICIENT_BUFFER) instead of S_FALSE
Also extended/fixed the corresponding ctest
The HRESULT S_FALSE does not indicate an error:
- return E_INVALIDARG instead of S_FALSE
- return HRESULT_FROM_WIN32(ERROR_INSUFFICIENT_BUFFER) instead of S_FALSE
Also extended/fixed the corresponding ctest
The HRESULT S_FALSE does not indicate an error:
- return E_INVALIDARG instead of S_FALSE
- return HRESULT_FROM_WIN32(ERROR_INSUFFICIENT_BUFFER) instead of S_FALSE
Also extended/fixed the corresponding ctest
Exposing lzcnt in crt.h might causes compiler errors (redefinition) with
recent versions of gcc (>=4.9) when winpr is included in other projects.
As lzcnt isn't part of crt according to MSDN and also shouldn't be
exported by default it was moved to intrin.h.
The related test was also moved to the top level directory of winpr.
The ntlm_construct_challenge_target_info function can potentially pass NULL as
argument to the ntlm_av_pair_add function (for example DnsDomainName.Buffer).
This NULL finally lands in the CopyMemory (which is macro to the memcpy(3)
function) which can't handle NULL.
