May 2015 - Wine-Devel - list.winehq.org

Re: [PATCH 1/2] msvcp110: Add _File_size implementation and test.

by YongHao Hu

Hi, Piotr. Thank you very much for all your comments. They are of great help. On 2015年05月27日 18:30, Piotr Caban wrote: > On 05/27/15 04:35, YongHao Hu wrote: >> +/* filesystem */ >> +static unsigned long long (__cdecl*p_tr2_sys__File_size)(char const*); > You should avoid using "long long" type. Please e.g. use ULONGLONG > instead. > >> + mkdir("tr2_test_dir"); >> + file = CreateFileA("tr2_test_dir/f1", GENERIC_WRITE, 0, NULL, >> CREATE_ALWAYS, 0, NULL); >> + SetFilePointerEx(file, large, NULL, FILE_BEGIN); >> + SetEndOfFile(file); > Please test return values of CreateFile, SetFilePointerEx and > SetEndOfFile. > >> + large.LowPart = 4000000000u; >> + large.HighPart = 120; >> + file = CreateFileA("tr2_test_dir/f2", GENERIC_WRITE, 0, NULL, >> CREATE_ALWAYS, 0, NULL); >> + SetFilePointerEx(file, large, NULL, FILE_BEGIN); >> + SetEndOfFile(file); >> + val = p_tr2_sys__File_size("tr2_test_dir/f2"); >> + ok(val == 519396075520 || broken(val == 0), "file_size is >> %llu\n", val); /* not enough storage on the testbots */ >> + CloseHandle(file); > Please remove this test. It's not a good idea to create ~500GB file in > the tests. > >> + val = p_tr2_sys__File_size("tr2_test_dir"); >> + ok(val == 0, "file_size is %llu\n", val); >> + >> + val = p_tr2_sys__File_size("tr2_test_dir/not_exists_file"); >> + ok(val == 0, "file_size is %llu\n", val); > It would be nice to test GetLastError and errno value after this calls. > > What happens if NULL is passed as the argument? > >> +ULONGLONG __cdecl tr2_sys__File_size(const char* path) >> +{ >> + WIN32_FILE_ATTRIBUTE_DATA fad; > Please add a trace message. > >> + if(!GetFileAttributesExA(path, GetFileExInfoStandard, &fad)) >> + return (ULONGLONG)0; > This cast is not needed, you can just return 0 here. > >> + if(fad.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) >> + return (ULONGLONG)0; >> + return ((ULONGLONG)(fad.nFileSizeHigh) << >> (sizeof(fad.nFileSizeLow)*8)) + fad.nFileSizeLow; > I think that it's better to simply write 32 instead of > sizeof(fad.nFileSizeLow)*8 here. > > Thanks, > Piotr

May 28, 2015

Re: libport: Fix build on PowerPC

by André Hentschel

Am 27.05.2015 um 11:18 schrieb Grazvydas Ignotas: > On Tue, May 26, 2015 at 11:16 PM, André Hentschel <nerv(a)dawncrow.de> wrote: >> It was broken by recent commits >> --- >> libs/port/interlocked.c | 15 +++++++++++++++ > > Maybe we could just drop all this powerpc specific code, especially > considering missing cmpxchg64 implementation (better use mutex > fallback there) and that some other platforms (ARM/ARM64) don't have > any of this anyway and just use the fallbacks. A better idea is to implement cmpxchg64 in assembler, i already talked with slackner about that. it'll take i while as i have a lot of work to do in wine alone, and i also have a lot of other things to do...

May 27, 2015

Fix GetRawInputDeviceList() to return ERROR_INSUFFICIENT_BUFFER

by achurch＠achurch.org

(Note: I am not subscribed to the list; please CC me on any replies.) When GetRawInputDeviceList() is passed a buffer which is too small, the function should set the last error code to ERROR_INSUFFICIENT_BUFFER and return failure (see <http://msdn.microsoft.com/en-us/library/windows/desktop/ms645598(v=vs.85).a…>). However, the current Wine implementation does not set the last error code, meaning that callers expecting to see ERROR_INSUFFICIENT_BUFFER will behave incorrectly. I've confirmed that Windows itself (at least XP and 8) matches the documented behavior. See also: https://bugs.winehq.org/show_bug.cgi?id=37667 --Andrew Church http://achurch.org/ diff -urN wine-1.7.43-orig/dlls/user32/input.c wine-1.7.43/dlls/user32/input.c --- wine-1.7.43-orig/dlls/user32/input.c 2015-05-15 22:36:39 +0900 +++ wine-1.7.43/dlls/user32/input.c 2015-05-25 22:47:49 +0900 @@ -507,6 +507,7 @@ if (*device_count < 2) { *device_count = 2; + SetLastError(ERROR_INSUFFICIENT_BUFFER); return ~0U; } diff -urN wine-1.7.43-orig/dlls/user32/tests/Makefile.in wine-1.7.43/dlls/user32/tests/Makefile.in --- wine-1.7.43-orig/dlls/user32/tests/Makefile.in 2015-05-15 22:36:39 +0900 +++ wine-1.7.43/dlls/user32/tests/Makefile.in 2015-05-25 22:48:27 +0900 @@ -17,6 +17,7 @@ menu.c \ monitor.c \ msg.c \ + rawinput.c \ resource.c \ scroll.c \ static.c \ diff -urN wine-1.7.43-orig/dlls/user32/tests/rawinput.c wine-1.7.43/dlls/user32/tests/rawinput.c --- wine-1.7.43-orig/dlls/user32/tests/rawinput.c 1970-01-01 09:00:00 +0900 +++ wine-1.7.43/dlls/user32/tests/rawinput.c 2015-05-25 22:44:36 +0900 @@ -0,0 +1,43 @@ + /* Unit test suite for the rawinput functions + * + * Copyright 2015 Andrew Church + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#define WINVER 0x0501 + +#include "wine/test.h" +#include "windef.h" +#include "winbase.h" +#include "winuser.h" + +static void GRIDLShortArrayTest() +{ + RAWINPUTDEVICELIST devices[1]; + UINT device_count = 1; + int result, error; + + result = GetRawInputDeviceList(devices, &device_count, sizeof(*devices)); + ok(result == ~0U, "wrong result: %d\n", result); + ok(device_count >= 2, "wrong device_count: %d\n", device_count); + error = GetLastError(); + ok(error == ERROR_INSUFFICIENT_BUFFER, "wrong error code: %d\n", error); +} + +START_TEST(rawinput) +{ + GRIDLShortArrayTest(); +}

May 27, 2015

Re: [PATCH] kernel32: Return ERROR_CALL_NOT_IMPLEMENTED in GetUserPreferredUILanguages.

by Alexandre Julliard

Mark Harmstone <hellas(a)burntcomma.com> writes: > I want to include GetUserPreferredUILanguages in a test I'm writing, but can't at present as > it doesn't return the stub error code. Not that it really matters, but I don't see why you'd need to change Wine in order to be able to write a test. -- Alexandre Julliard julliard(a)winehq.org

May 27, 2015

Re: [PATCH 2/2] msvcp110: Add _Equivalent implementation and test.

by Piotr Caban

On 05/27/15 04:36, YongHao Hu wrote: > +static void test_tr2_sys__Equivalent(void) > +{ > + BOOL val; > + int i; > + FILE *file; > + struct { > + char const *file_path1; > + char const *file_path2; > + BOOL equivalent; It will make more sense to define it as int. > + } tests[] = { > + { "f1", "tr2_test_dir", -1 }, > + { "tr2_test_dir", "tr2_test_dir", -1 }, > + { "not_exists_file" , "not_exists_file", -1 }, > + { "tr2_test_dir/./f1", "tr2_test_dir/f2", FALSE }, > + { "tr2_test_dir/f1" , "tr2_test_dir/f1", TRUE }, > + { "tr2_test_dir\\f1", "tr2_test_dir/././f1", TRUE }, > + { "tr2_test_dir/../tr2_test_dir/f1", "tr2_test_dir/f1", TRUE }, Please add following test: { NULL, NULL, ... }, { "not_existing_file", "tr2_test_dir/f1", ... } The tests are also failing when run from VirtualBox share. It's caused by the virtual machine bug (their probably generate a random FileIndexHigh/Low every time file is opened). Could you please create the test files inside temp directory to workaround this bug? Cheers, Piotr

May 27, 2015

Re: [PATCH 1/2] msvcp110: Add _File_size implementation and test.

by Piotr Caban

> + ok(val == 519396075520 || broken(val == 0), "file_size is %llu\n", > val); /* not enough storage on the testbots */ I've missed one more thing. Please don't use %llu format in printf. You can see how to print the value in msvcp90/tests/ios.c file.

May 27, 2015

Re: [PATCH 1/2] msvcp110: Add _File_size implementation and test.

by Piotr Caban

On 05/27/15 04:35, YongHao Hu wrote: > +/* filesystem */ > +static unsigned long long (__cdecl*p_tr2_sys__File_size)(char const*); You should avoid using "long long" type. Please e.g. use ULONGLONG instead. > + mkdir("tr2_test_dir"); > + file = CreateFileA("tr2_test_dir/f1", GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, 0, NULL); > + SetFilePointerEx(file, large, NULL, FILE_BEGIN); > + SetEndOfFile(file); Please test return values of CreateFile, SetFilePointerEx and SetEndOfFile. > + large.LowPart = 4000000000u; > + large.HighPart = 120; > + file = CreateFileA("tr2_test_dir/f2", GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, 0, NULL); > + SetFilePointerEx(file, large, NULL, FILE_BEGIN); > + SetEndOfFile(file); > + val = p_tr2_sys__File_size("tr2_test_dir/f2"); > + ok(val == 519396075520 || broken(val == 0), "file_size is %llu\n", val); /* not enough storage on the testbots */ > + CloseHandle(file); Please remove this test. It's not a good idea to create ~500GB file in the tests. > + val = p_tr2_sys__File_size("tr2_test_dir"); > + ok(val == 0, "file_size is %llu\n", val); > + > + val = p_tr2_sys__File_size("tr2_test_dir/not_exists_file"); > + ok(val == 0, "file_size is %llu\n", val); It would be nice to test GetLastError and errno value after this calls. What happens if NULL is passed as the argument? > +ULONGLONG __cdecl tr2_sys__File_size(const char* path) > +{ > + WIN32_FILE_ATTRIBUTE_DATA fad; Please add a trace message. > + if(!GetFileAttributesExA(path, GetFileExInfoStandard, &fad)) > + return (ULONGLONG)0; This cast is not needed, you can just return 0 here. > + if(fad.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) > + return (ULONGLONG)0; > + return ((ULONGLONG)(fad.nFileSizeHigh) << (sizeof(fad.nFileSizeLow)*8)) + fad.nFileSizeLow; I think that it's better to simply write 32 instead of sizeof(fad.nFileSizeLow)*8 here. Thanks, Piotr

May 27, 2015

Re: libport: Fix build on PowerPC

by Grazvydas Ignotas

On Tue, May 26, 2015 at 11:16 PM, André Hentschel <nerv(a)dawncrow.de> wrote: > It was broken by recent commits > --- > libs/port/interlocked.c | 15 +++++++++++++++ Maybe we could just drop all this powerpc specific code, especially considering missing cmpxchg64 implementation (better use mutex fallback there) and that some other platforms (ARM/ARM64) don't have any of this anyway and just use the fallbacks. Gražvydas

May 27, 2015

Re: [PATCH] wined3d: fix color in range comparison

by Stefan Dösinger

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Am 2015-05-26 um 14:41 schrieb Sergey Isakov: > +#define RED_KEY(x) (x & 0xff0000) > +#define GREEN_KEY(x) (x & 0x00ff00) > +#define BLUE_KEY(x) (x & 0x0000ff) > + return ((RED_KEY(color) >= RED_KEY(color_key->color_space_low_value)) && > + (RED_KEY(color) <= RED_KEY(color_key->color_space_high_value)) && > + (GREEN_KEY(color) >= GREEN_KEY(color_key->color_space_low_value)) && > + (GREEN_KEY(color) <= GREEN_KEY(color_key->color_space_high_value)) && > + (BLUE_KEY(color) >= BLUE_KEY(color_key->color_space_low_value)) && > + (BLUE_KEY(color) <= BLUE_KEY(color_key->color_space_high_value))); > +#undef RED_KEY > +#undef GREEN_KEY > +#undef BLUE_KEY The basic idea is most likely right, but it needs tests. Do you have a game that needs this behavior? The way you're extracting individual colors from the key is wrong. The bitmasks depend on the texture format. The code you're modifying is unused on any modern GPU. On GPUs that support ARB_fragment_program or GLSL we do the color key comparison in the fragment shader nowadays. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBAgAGBQJVZXSWAAoJEN0/YqbEcdMwbEUP/RM4QUsbdSzhSLyIz5j00Epb AW1ZcNOg61yrfymjHdhJkqfdNmEiKQFsRshD+ZQAxVHqMsZLTxXtRFfBB14WtmNI Ng5jwwiAzDqlmpjftlZ4CmkR1w7DMxH6Nt4e9OQBlDKShSIoGngL6RremV+2Wq+f nNC6HfdmUiAIReVvI2aMm5B9C5L624SKLvoAZNI8QW+HJeZeESSGmT9bx4IRVQwl QX+aKojT43CJwq6UpZRu7ZkniotrmO9aDC+6S3S0od0ZJWSP5Q9upCNWNcN+iPG7 9kCPwZMsEAxNUjvh/wQx8xE1+wiR7Ugjl/X6yTN8HFesFHChrqC2UeHUAkUKU++L WC4kwijfbktKSZnIiSvrTAzWvLOeCZMifFQqm8aqmlOqyvbNwyOJjh6yM/nqZi+c 2aEdEIMY5aaoswDIlyGRRN57zeWtqjZVmg7UzVxmZqfiocmUVxgdmdRv8BvFi0RF R4A6YTDP5EYb33SvH3qQE5kIrtDdtL8qtanmBsuHnhhwd1rdgeeq/9N0tqSjH1GQ s8vsXVmd3qU+mQMzyaxN9Cc9RGxGNgd+g1LkkoPqe+aVUw/JDzvAdMDGpy63zjhM QUlfS771f+Zt43wku/20XP1engXXti4m+KfnSOS3nCdKO1xRKOGHAMlUZzd8+jhU 2rWKuh42D3mSMDk1OhLR =IdO/ -----END PGP SIGNATURE-----

May 27, 2015

Re: winedump: Add more image debug types

by Michael Stefaniuc

On 05/26/2015 10:59 PM, André Hentschel wrote: > My main goal is to destroy the ccache from Michael S. ;) Did you just ask for your patch to be Rejected? But I don't mind anymore since I got my new beefy box as compilation is fast enough with that ;-P bye michael > --- > include/winnt.h | 4 ++++ > tools/winedump/pe.c | 8 ++++++++ > 2 files changed, 12 insertions(+) > > diff --git a/include/winnt.h b/include/winnt.h > index b25c25b..4c89750 100644 > --- a/include/winnt.h > +++ b/include/winnt.h

May 27, 2015