June 2009 - Wine-devel - list.winehq.org

Test box tagged [dr-asus]
by Nikolay Sivov 07 Jun '09

07 Jun '09

Hi. Does anyone no whose box is it (dr-asus)? It has some (actually always 10) test failures for ListView and I think it's possible to find a cause. Same failures on 2k and XP actually. I'd like to contact owner for testing some patches. Thanks.

5 4

cd-roms that need unhide
by Dan Kegel 07 Jun '09

07 Jun '09

Despite jwhite's kernel fix, http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h… I find I have to use 'norock' on some discs, e.g. Macromedia Studio 8, http://bugs.winehq.org/show_bug.cgi?id=9467 Microsoft Dynamics GP demo, http://bugs.winehq.org/show_bug.cgi?id=8178 ... and the stupid DKNY perfume promo DVD-ROM that came in the newspaper today. Running it with Wine yields... a black screen, as the flash player can't get at its data. Mounting the DKNY disc with -o unhide or -o norock lets it work fine with wine. Here's what one sees without either flag: $ ls /media/cdrom/ -l total 3668 dr-xr-xr-x 2 4294967295 4294967295 40 2007-07-24 15:54 AUDIO_TS -r-xr-xr-x 1 4294967295 4294967295 2144972 2007-07-17 20:18 DKNYDeliciousNight -r-xr-xr-x 1 4294967295 4294967295 1602372 2007-07-17 20:18 DKNYDeliciousNight.exe -r-xr-xr-x 1 4294967295 4294967295 1616 2007-02-14 06:34 Readme.txt dr--r--r-- 3 4294967295 4294967295 560 2007-07-24 15:54 VIDEO_TS (The VIDEO_TS directory is not searchable! ) Here's what one sees with unhide: dank(a)lappy:~$ ls -l /media/cdrom/ total 3788 dr-xr-xr-x 2 4294967295 4294967295 40 2007-07-24 15:54 AUDIO_TS -r-xr-xr-x 1 4294967295 4294967295 67 2007-07-05 11:20 autorun.inf dr-xr-xr-x 2 4294967295 4294967295 1032 2007-07-09 12:11 backgroundLoops dr-xr-xr-x 2 4294967295 4294967295 92 2007-06-25 14:48 css -rwxr-xr-x 1 4294967295 4294967295 1024 2007-01-03 08:47 Desktop DB -rwxr-xr-x 1 4294967295 4294967295 2 2007-01-03 08:47 Desktop DF -r-xr-xr-x 1 4294967295 4294967295 2144972 2007-07-17 20:18 DKNYDeliciousNight -r-xr-xr-x 1 4294967295 4294967295 1602372 2007-07-17 20:18 DKNYDeliciousNight.exe -r-xr-xr-x 1 4294967295 4294967295 97566 2007-06-27 14:30 DKNYDeliciousNight.ico dr-xr-xr-x 2 4294967295 4294967295 244 2007-07-13 09:30 exit dr-xr-xr-x 2 4294967295 4294967295 3400 2007-07-17 17:18 flv dr-xr-xr-x 9 4294967295 4294967295 684 2007-07-18 15:41 fscommand -r-xr-xr-x 1 4294967295 4294967295 1616 2007-02-14 06:34 Readme.txt dr-xr-xr-x 2 4294967295 4294967295 240 2007-07-09 12:11 soundFx dr-xr-xr-x 2 4294967295 4294967295 996 2007-07-18 18:50 swf dr--r--r-- 3 4294967295 4294967295 560 2007-07-24 15:54 VIDEO_TS dr-xr-xr-x 2 4294967295 4294967295 140 2007-07-05 09:23 xml (The VIDEO_TS directory is not searchable, but the app works anyway. ) norock is similar, but VIDEO_TS is searchable. Given that these poorly mastered discs are turning up periodically -- widely distributed -- it seems like it might be time for another kernel change, or at least some userland kludge that does a remount. The heuristic I imagine is: If the top level directory has any subdirectories that are completely nonsearchable or nonreadable, remount with unhide. What say? Would this help users more than it would hurt? - Dan

2 2

Re: [PATCH] This patch fixes context menu crashes for edit controls as apparent in vb apps.
by Reece Dunn 07 Jun '09

07 Jun '09

2009/6/7 Michael Martin <martinmnet(a)hotmail.com>: You'll need to include the patch as an attachment (renamed as a text file if your mail client messes up the encoding), instead of including it in the body text (the indentation in the patch you included has been removed by your mail client). > + BOOL selectedMenuItem; > ... > + selectedMenuItem = TrackPopupMenu(popup, TPM_LEFTALIGN | TPM_RIGHTBUTTON | > TMPRETURNCMD, x, y, 0, es->hwndSelf, NULL); > + switch (selectedMenuItem) { Despite TrackPopupMenu having a BOOL return type, using an int would make it more readable (selectedMenuItem is not being used as a boolean value, but as a command index). Alternatively, just use: switch (TrackPopupMenu(...)) { TMPRETURNCMD -- This should be TMP_RETURNCMD. Is this an error when adding the patch to the email? Thanks for contributing to Wine, - Reece

1 0

Re: [2/5] WineD3D: Implement loops with the NV extensions
by Allan Tong 07 Jun '09

07 Jun '09

This is using PUSHA and POPA without setting the NV_vertex_program3 option in the shader. - Allan

1 0

Re: [3/5] WineD3D: Unroll loops in ARB if GL_NV_vertex_program2_option is not supported
by James Mckenzie 06 Jun '09

06 Jun '09

Stefan Dösinger <stefan(a)codeweavers.com> wrote about [3/5] WineD3D: Unroll loops in ARB if GL_NV_vertex_program2_option is not supported Stefan: Could this not a be a function: + for(i = 0; i < MAX_CONST_I; i++) + { + if(int_skip & (1 << i)) + { + args->loop_ctrl[i][0] = 0; + args->loop_ctrl[i][1] = 0; + args->loop_ctrl[i][2] = 0; + } + else + { + args->loop_ctrl[i][0] = stateblock->pixelShaderConstantI[i * 4]; + args->loop_ctrl[i][1] = stateblock->pixelShaderConstantI[i * 4 + 1]; + args->loop_ctrl[i][2] = stateblock->pixelShaderConstantI[i * 4 + 2]; + } + } Also, it looks like you did not count correctly: + if(vshader) + { + /* Count and aL start value are unsigned */ + ret[0] = priv->cur_vs_args->loop_ctrl[idx][0]; + ret[1] = priv->cur_vs_args->loop_ctrl[idx][1]; + /* The step/stride is signed */ + ret[2] = ((char) priv->cur_vs_args->loop_ctrl[idx][3]); + } + else + { + ret[0] = priv->cur_ps_args->loop_ctrl[idx][0]; + ret[1] = priv->cur_ps_args->loop_ctrl[idx][1]; + ret[2] = ((char) priv->cur_ps_args->loop_ctrl[idx][3]); + } + return ret; A little explanation might help here. James McKenzie

1 0

Re: [3/5] WineD3D: Unroll loops in ARB if GL_NV_vertex_program2_option is not supported
by Henri Verbeet 06 Jun '09

06 Jun '09

2009/6/7 Stefan Dösinger <stefan(a)codeweavers.com>: > +/* Not reentrant! */ > +static inline const int *get_int_const(const struct wined3d_shader_instruction *ins, IWineD3DBaseShaderImpl *This, DWORD idx) > +{ > + BOOL vshader = shader_is_vshader_version(This->baseShader.reg_maps.shader_version.type); > + WORD flag = (1 << idx); > + const local_constant *constant; > + struct shader_arb_ctx_priv *priv = ins->ctx->backend_data; > + > + if(This->baseShader.reg_maps.local_int_consts & flag) > + { > + static const int err_ret[4] = {0, 0, 0, 0}; > + /* What good is a if(bool) with a hardcoded local constant? I don't know, but handle it */ > + LIST_FOR_EACH_ENTRY(constant, &This->baseShader.constantsI, local_constant, entry) > + { > + if (constant->idx == idx) > + { > + return (int *) constant->value; > + } > + } > + ERR("Local constant not found\n"); > + return err_ret; > + } > + else > + { > + static int ret[4]; > + ret[3] = 0; > + if(vshader) > + { > + /* Count and aL start value are unsigned */ > + ret[0] = priv->cur_vs_args->loop_ctrl[idx][0]; > + ret[1] = priv->cur_vs_args->loop_ctrl[idx][1]; > + /* The step/stride is signed */ > + ret[2] = ((char) priv->cur_vs_args->loop_ctrl[idx][3]); > + } > + else > + { > + ret[0] = priv->cur_ps_args->loop_ctrl[idx][0]; > + ret[1] = priv->cur_ps_args->loop_ctrl[idx][1]; > + ret[2] = ((char) priv->cur_ps_args->loop_ctrl[idx][3]); > + } > + return ret; > + } > +} I said I wouldn't be reviewing arb_program_shader.c changes (and I'm not, in general), but I couldn't help noticing this bit of code. *blink*? > + const int *control_values = get_int_const(ins, This, ins->src[0].reg.idx); > + list_init(&priv->record); > + priv->recording = TRUE; > + control_frame->outer_loop = TRUE; > + control_frame->loop_control[0] = control_values[0]; > + control_frame->loop_control[1] = control_values[1]; > + control_frame->loop_control[1] = control_values[2]; This is either a typo or redundant.

1 0

Re: [1/5] WineD3D: Store the bumpmap ARB constants in ARB structures
by Henri Verbeet 06 Jun '09

06 Jun '09

I think the actual patch is missing.

1 0

[setupapi/tests] pSetupGetField is A or W depending on OS
by Paul Vriens 06 Jun '09

06 Jun '09

Hi, Sent to wine-devel for some feedback. I was checking for more lstrcmpW's (not supported on Win9x) and found one in setupapi/tests/parser.c. After some checking it turns out that it returns an ANSI string on Win9x and a Unicode one on NT4 and higher. There is no export of pSetupGetField[AW] on any of the platforms. The attached patch takes care of that but I have a feeling this patch could be written in a better way. Suggestions, hints, remarks? -- Cheers, Paul. diff --git a/dlls/setupapi/tests/parser.c b/dlls/setupapi/tests/parser.c index a67a5cd..3b4682e 100644 --- a/dlls/setupapi/tests/parser.c +++ b/dlls/setupapi/tests/parser.c @@ -32,14 +32,16 @@ /* function pointers */ static HMODULE hSetupAPI; -static LPCWSTR (WINAPI *pSetupGetField)(PINFCONTEXT,DWORD); +static LPCSTR (WINAPI *pSetupGetFieldA)(PINFCONTEXT,DWORD); +static LPCWSTR (WINAPI *pSetupGetFieldW)(PINFCONTEXT,DWORD); static BOOL (WINAPI *pSetupEnumInfSectionsA)( HINF hinf, UINT index, PSTR buffer, DWORD size, UINT *need ); static void init_function_pointers(void) { hSetupAPI = GetModuleHandleA("setupapi.dll"); - pSetupGetField = (void *)GetProcAddress(hSetupAPI, "pSetupGetField"); + pSetupGetFieldA = (void *)GetProcAddress(hSetupAPI, "pSetupGetField"); + pSetupGetFieldW = (void *)GetProcAddress(hSetupAPI, "pSetupGetField"); pSetupEnumInfSectionsA = (void *)GetProcAddress(hSetupAPI, "SetupEnumInfSectionsA" ); } @@ -497,7 +499,14 @@ static const char *contents = "[Version]\n" "[Strings]\n" "RTMQFE_NAME = \"RTMQFE\"\n"; -static const WCHAR getfield_res[][20] = +static const CHAR getfield_resA[][20] = +{ + "RTMQFE", + "%RTMGFE_NAME%", + "SP1RTM", +}; + +static const WCHAR getfield_resW[][20] = { {'R','T','M','Q','F','E',0}, {'%','R','T','M','G','F','E','_','N','A','M','E','%',0}, @@ -509,10 +518,18 @@ static void test_pSetupGetField(void) UINT err; BOOL ret; HINF hinf; - LPCWSTR field; + LPCSTR fieldA; + LPCWSTR fieldW; INFCONTEXT context; int i; int len; + BOOL unicode = TRUE; + + /* Nice, pSetupGetField is either A or W depending on the Windows version! */ + SetLastError(0xdeadbeef); + lstrcmpW(NULL, NULL); + if (GetLastError() == ERROR_CALL_NOT_IMPLEMENTED) + unicode = FALSE; hinf = test_file_contents( contents, &err ); ok( hinf != NULL, "Expected valid INF file\n" ); @@ -524,23 +541,47 @@ static void test_pSetupGetField(void) for ( i = 0; i < 3; i++ ) { - field = pSetupGetField( &context, i ); - ok( field != NULL, "Failed to get field %i\n", i ); - ok( !lstrcmpW( getfield_res[i], field ), "Wrong string returned\n" ); + if (unicode) + { + fieldW = pSetupGetFieldW( &context, i ); + ok( fieldW != NULL, "Failed to get field %i\n", i ); + ok( !lstrcmpW( getfield_resW[i], fieldW ), "Wrong string returned\n" ); + } + else + { + fieldA = pSetupGetFieldA( &context, i ); + ok( fieldA != NULL, "Failed to get field %i\n", i ); + ok( !lstrcmpA( getfield_resA[i], fieldA ), "Wrong string returned\n" ); + } } - field = pSetupGetField( &context, 3 ); - ok( field != NULL, "Failed to get field 3\n" ); - len = lstrlenW( field ); - ok( len == 511 /* NT4, W2K, XP and W2K3 */ || - len == 4096 /* Vista */ || - len == 256 /* Win9x and WinME */, - "Unexpected length, got %d\n", len ); - - field = pSetupGetField( &context, 4 ); - ok( field == NULL, "Expected NULL, got %p\n", field ); - ok( GetLastError() == ERROR_INVALID_PARAMETER, - "Expected ERROR_INVALID_PARAMETER, got %u\n", GetLastError() ); + if (unicode) + { + fieldW = pSetupGetFieldW( &context, 3 ); + ok( fieldW != NULL, "Failed to get field 3\n" ); + len = lstrlenW( fieldW ); + ok( len == 511 || /* NT4, W2K, XP and W2K3 */ + len == 4096, /* Vista */ + "Unexpected length, got %d\n", len ); + + fieldW = pSetupGetFieldW( &context, 4 ); + ok( fieldW == NULL, "Expected NULL, got %p\n", fieldW ); + ok( GetLastError() == ERROR_INVALID_PARAMETER, + "Expected ERROR_INVALID_PARAMETER, got %u\n", GetLastError() ); + } + else + { + fieldA = pSetupGetFieldA( &context, 3 ); + ok( fieldA != NULL, "Failed to get field 3\n" ); + len = lstrlenA( fieldA ); + ok( len == 511, /* Win9x, WinME */ + "Unexpected length, got %d\n", len ); + + fieldA = pSetupGetFieldA( &context, 4 ); + ok( fieldA == NULL, "Expected NULL, got %p\n", fieldA ); + ok( GetLastError() == ERROR_INVALID_PARAMETER, + "Expected ERROR_INVALID_PARAMETER, got %u\n", GetLastError() ); + } SetupCloseInfFile( hinf ); }

1 0

Одговор: Одговор: Latest Git build 7aeffc442cc1 for Wine 1.1.23
by Milan Kostić 06 Jun '09

06 Jun '09

2009/6/6, Henri Verbeet <hverbeet(a)gmail.com>: > 2009/6/6 Milan Kostić <smoki00790(a)gmail.com>: >> Deafulting to fbo is good, but it should be more good to leave also >> prior working support for backbuffer on older cards who not have >> EXT_framebuffer_object. Reverting to >> "OffscreenRenderingMode"="backbuffer" is not working as expected for >> me on r200 Mesa/DRI drivers, for example in Max Payne 2 which now >> can't be even starting with backbuffer, Rayman 2 and Dungen Siege 2 >> also. In 1.1.22>= they worked. >> > Please do a regression test to find the exact commit that breaks > "OffscreenRenderingMode"="backbuffer", and file a bug report. > Sorry i can't do regression test right now, slow connection (dial-up).

1 0

Re: gdiplus: Add test cases for GdipDrawCurve [corrected]
by Andrew Eikum 06 Jun '09

06 Jun '09

This patch was submitted back on Tuesday and I haven't received a response one way or the other about it. Does anyone see anything immediately wrong with it? I more-or-less copied the functionality of test_GdipDrawBezier right above, testing each of the different input possibilities for correctness. It passes 100% on WinXP SP3 and Win7 RC1, although there are failures in Wine's GdipDrawCurve implementation. Thanks for taking a look, Andrew

3 6