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

Re: [PATCH 3/8] d3drm: Implement IDirect3DRMDevice::CreateDeviceFromClipper.
by Stefan Dösinger Aug. 6, 2015

Aug. 6, 2015

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Am 2015-08-06 um 00:49 schrieb Aaryaman Vasishta:> + IDirect3DRMDevice *device1; ... hr = > + IDirect3DRMDevice *device1; > ... > hr = IDirect3DRM_CreateDeviceFromClipper(d3drm1, clipper, &driver, 0, 0, &device1); > - todo_wine ok(hr == D3DRMERR_BADVALUE, "Expected hr == D3DRMERR_BADVALUE, got %x.\n", hr); > - if (SUCCEEDED(hr)) > - IDirect3DRMDevice_Release(device1); > + ok(hr == D3DRMERR_BADVALUE, "Expected hr == D3DRMERR_BADVALUE, got %x.\n", hr); > + ok(device1 == NULL, "Expected device returned == NULL, got %p.\n", device1); Set device1 = (IDirect3DRMDevice *)0xdeadbeef before calling CreateDeviceFromClipper, otherwise the device == NULL check isn't worth much at the end. (You don't know if its NULL because IDirect3DRM::CreateDeviceFromClipper set it to NULL or because it happened to randomly contain NULL at the start of the function. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBAgAGBQJVw1NZAAoJEN0/YqbEcdMwSkcP/1NMXYG2m/OkDgVhrAg3kLB3 RRu4egHLkAPpxx9Ms0iAqfl9NtbhtlQpFBjOdwmdElOQbjRZFVpU5pUhFQO4jgbo ywlrvkXYJvmsYWjMTzNFKM0vZaNO6uRQK/pg/QWwqCW+xlunCyHa8vWLzu4u34sB FqGfjOeS/FoQkH2IYnFjpSV12Org9wJC+GVwqw8h1yfgiW3lKaCgRUcA0pRNgC5o l4OlkLhnAC9EqvdSHDgW+A9UkslVDMZq7iq5XFNCasRKqPixCoEsXXu8LH3JakCI 3WOn0z8bgelcLl8Z6xr4dXO5iKvw2Ff1DxKPnmQrykzY2E4OQkZdjqMznXKNqq3g SVXqHq23KuU+CMkimeyDLd8z7brIQfud5W2Ru9/PIWrftuU3y+TgKjZR9awC7pT/ dXF8cCwPOa33MEOUAL2syunC/nAf7RDhBeCnWtj0G9wKGXWdnqFlzIaKVr4sO0wQ RO457ySvXReGkhxe7Df178VQfst9jppCA0FR7m+MlNe8BP31gNEfUvHi0HxYnD4r lyOXA3ltnhgirKhR070wHOPVbR2ZqWuwe4Qv+QAmwYJup95IU6UsXDBcU5VVHNqa 9U+Mb9+FTfcPkgKX60AHJiWdgMsPvuYIKrNoYPBLexv+x1uYdCTr6iB1YQnHCQee jyRW8O7NLo/lxDIAJsh0 =5ZQO -----END PGP SIGNATURE-----

2 2

Re: [PATCH 2/2] msvcp110: Add tr2_sys__Stat and tr2_sys__Lstat implementation and test.（try2 resend)
by Piotr Caban Aug. 6, 2015

Aug. 6, 2015

On 08/05/15 07:46, YongHao Hu wrote: > +enum file_type tr2_sys__Stat(char const* path, int* err_code) > +enum file_type tr2_sys__Lstat(char const* path, int* err_code) Calling conventions are missing here. Thanks, Piotr

1 0

Re: [3/4] msvcirt: Implement filebuf::setmode
by Piotr Caban Aug. 6, 2015

Aug. 6, 2015

On 08/05/15 20:32, Iván Matellanes wrote: > diff --git a/dlls/msvcirt/msvcirt.c b/dlls/msvcirt/msvcirt.c > index bf89650..0f82f30 100644 > --- a/dlls/msvcirt/msvcirt.c > +++ b/dlls/msvcirt/msvcirt.c > @@ -1009,8 +1009,16 @@ streambuf* __thiscall filebuf_setbuf(filebuf *this, char *buffer, int length) > DEFINE_THISCALL_WRAPPER(filebuf_setmode, 8) > int __thiscall filebuf_setmode(filebuf *this, int mode) > { > - FIXME("(%p %d) stub\n", this, mode); > - return 0; > + int previous; > + > + TRACE("(%p %d)\n", this, mode); > + if (mode != filebuf_text && mode != filebuf_binary) > + return -1; > + > + streambuf_lock(&this->base); > + previous = _setmode(this->fd, mode); > + streambuf_unlock(&this->base); > + return previous; > } The function should empty the buffers before it changes the mode. Also it will be cleaner if filebuf_text and filebuf_binary definition is changed so _O_TEXT and _O_BINARY is used instead of its value. Thanks, Piotr

1 0

D3DRM Implementation patches upto CreateDeviceFromSurface
by Aaryaman Vasishta Aug. 5, 2015

Aug. 5, 2015

CreateDeviceFromD3D will be pushed as soon as patch 9 is finalised. The main concern right now is the thunking of the Release methods of the device. This concern comes up from the fact that the CreateDeviceFromSurface methods from all 3 versions behave the same way except for one difference in version 3, which I mention below. When you create a device from IDirect3DRM3::CreateDeviceFromSurface and release it, native will also destroy the depth surface attached to it (if it created it internally), otherwise it'll let the application keep the reference to the depth surface it created. Other than the above peculiarity, the behaviour is identical. I've attached all the commits which lead to the point of implementing CreateDeviceFromSurface. Please, review the patches as much as you'd like and if possible, provide feedback and suggestions on any improvements that can be made, possible comments required, any stupid mistakes I made etc. :) Thank you! Aaryaman

2 20

Re: [PATCH 1/2] d3dx9: Correctly handle whitespace character in D3DXCreateText. (try 2 resend)
by Marvin Aug. 5, 2015

Aug. 5, 2015

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at https://testbot.winehq.org/JobDetails.pl?Key=15799 Your paranoid android. === w8 (32 bit mesh) === mesh.c:8258: Test failed: Mesh 26: new_num_vertices == 4, expected 6. mesh.c:8272: Test failed: Mesh 26: indices[3] == 1, expected 3 mesh.c:8272: Test failed: Mesh 26: indices[4] == 3, expected 4 mesh.c:8272: Test failed: Mesh 26: indices[5] == 2, expected 5 mesh.c:8313: Test failed: Mesh 26: vertex_remap_ptr[3] == 4, expected 3 mesh.c:8313: Test failed: Mesh 26: vertex_remap_ptr[4] == -1, expected 4 mesh.c:8313: Test failed: Mesh 26: vertex_remap_ptr[5] == -1, expected 5 mesh.c:8328: Test failed: Mesh 26: Got (3.000000, 0.000000, 0.000000) for vertex 3 position, expected (2.000000, 3.000000, 0.000000). mesh.c:8328: Test failed: Mesh 26: Got (255, 255, 255, 255) for vertex 3 undefined color, expected (0, 0, 0, 0). mesh.c:8328: Test failed: Mesh 26: Got (2.000000, 3.000000, 0.000000) for vertex 4 position, expected (3.000000, 0.000000, 0.000000). mesh.c:8328: Test failed: Mesh 26: Got (255, 255, 255, 255) for vertex 5 undefined color, expected (1, 1, 1, 1).

1 0

Re: reg: implement reg query operation
by Stefan Dösinger Aug. 5, 2015

Aug. 5, 2015

Hi, > +#define MAX_NAME_BUFFER 255 > +#define MAX_VALUE_BUFFER 16383 I don't like this. Ideally the code should query the lengths from the Registry APIs and allocate appropriate buffers. If the Reg* functions guarantee a maximum size use defines from the public headers. > +static const WCHAR fmt_value[] = {' ',' ',' ','%','s',' ','%','s',' ','%','s','\n',0}; This is used in print_value() only, you can move it into this function. > +static const WCHAR no_name[] = { 'n','o','_','n','a','m','e',0 }; Is this used for the default value? Is "no_name" the right choice? Does this need localization? > + hkey = get_rootkey(key_name); > + subkey_name = strchrW(key_name, '\\'); > + if (NULL == hkey || NULL == subkey_name) > + { > + reg_printfW(fmt_err, key_name); > + return 1; > + } Are you sure that this is the right behavior? I haven't checked, but I can imagine that I can e.g. have a string value directly in HKEY_CURRENT_USER, in which case subkey_name would be NULL. > + value_buffer = HeapAlloc(GetProcessHeap(), 0, MAX_VALUE_BUFFER); Please check for allocation errors. If you do stay with the fixed size buffer, do you need HeapAlloc at all? Can't you just put this on the stack? On a related note, passing value_buffer to the two helper functions but not its length, and setting value_len in those functions is ugly. This way the helper functions depend on an implementation detail of their caller. Yeah, the #define makes it a bit better, but it's still not pretty. > + run_reg_exe("reg Query HKCU\\" KEY_BASE " /ve", &r); > + ok(r == REG_EXIT_SUCCESS, "got exit code %d, expected 0\n", r); > + > + run_reg_exe("reg Query HKCU\\" KEY_BASE " /v test0", &r); > + ok(r == REG_EXIT_SUCCESS, "got exit code %d, expected 0\n", r); > + > + run_reg_exe("reg Query HKCU\\" KEY_BASE " /s", &r); > + ok(r == REG_EXIT_SUCCESS, "got exit code %d, expected 0\n", r); > + > + run_reg_exe("reg delete HKCU\\" KEY_BASE " /f", &r); > + ok(r == REG_EXIT_SUCCESS, "got exit code %d, expected 0\n", r); Some tests for error cases would be helpful as well, including all the error checks you have in your code. I expect scripts to parse the output of this command. Please add some tests that check the output. There's infrastructure for reading command output in other program tests that you can use. The following are style issues. I'm pointing them out while I'm at it, please fix them as well. > +static void query_sub(HKEY hkey, WCHAR* key_name, WCHAR* value_buffer) (Style) WCHAR *key_name please. "WCHAR*" is more common in C++. > + HKEY hsubkey = NULL; > + HKEY hkey = NULL; (Style) Please drop the h from these names. > + ret = RegOpenKeyExW( hkey, subkey_name+1, 0, KEY_READ, &hsubkey ); (Style) inconsistent space after parenthesis. > + if (ret == ERROR_SUCCESS ) Same > + static const WCHAR* reg_type[] = { > + type_none, > + type_sz, (Style) { placement

2 1

Re: widl: Added support for -ns_prefix option.
by Sebastian Lackner Aug. 5, 2015

Aug. 5, 2015

On 04.08.2015 14:26, Jacek Caban wrote: > --- > tools/widl/header.c | 10 ++++++++-- > tools/widl/parser.l | 3 +++ > tools/widl/typetree.c | 12 ++++++++++-- > tools/widl/widl.c | 6 ++++++ > tools/widl/widl.h | 1 + > tools/widl/widl.man.in | 2 ++ > 6 files changed, 30 insertions(+), 4 deletions(-) > [...] > winrt_mode = 1; > break; > + case RT_NS_PREFIX: > + use_abi_namespace = 1; > case WIN32_OPTION: > do_win32 = 1; > do_win64 = 0; Is there maybe a break missing?

2 1

Re: msxml3: Handle NULL parameter in IXMLDOMElement::get_ownerDocument(). (try 2)
by Nikolay Sivov Aug. 5, 2015

Aug. 5, 2015

On 31.07.2015 12:12, Zhenbo Li wrote: > try 2: > Test get_ownerDocument for more xml nodes. > > --- > dlls/msxml3/node.c | 2 ++ > dlls/msxml3/tests/domdoc.c | 51 > ++++++++++++++++++++++++++++++++++++++++++++++ > 2 files changed, 53 insertions(+) > Hi. This looks fine to me. It would be nice to test attribute node too (I don't think it appears in children list, but it's been a while since I looked into it), but it's not a blocking issue.

1 0

Re:TestBot job 15747 results: quartz: waveparser support wav file generated by ffmpeg
by Changhui LIU Aug. 5, 2015

Aug. 5, 2015

Hi Alexandre: The error code 80040256 is VFW_E_NO_AUDIO_HARDWARE, According to MSDN, the detail info is: Cannot play back the audio stream: no audio hardware is available, or the hardware is not supported. So these test failed on 64 bit windows is not caused by my patch: http://source.winehq.org/patches/data/113403 ------------------ Regards, Changhui Liu ------------------ Original ------------------ From: "Marvin"<testbot(a)winehq.org>; Date: Tue, Aug 4, 2015 07:04 PM To: "liuchanghui"<liuchanghui(a)linuxdeepin.com>; Subject: TestBot job 15747 results: quartz: waveparser support wav file generated by ffmpeg VM Status Number of test failures build completed w2000pro completed 0 wxppro completed 0 w2003std completed 0 wvistau64 completed 2 w2008s64 completed 2 w7u completed 0 w8 completed 0 wvistau64 completed 2 w2008s64 completed 2 === build (build) === No build failures found === w2000pro (32 bit filtergraph) === No test failures found === wxppro (32 bit filtergraph) === No test failures found === w2003std (32 bit filtergraph) === No test failures found === wvistau64 (32 bit filtergraph) === quartz: filtergraph.c:165: Test failed: RenderFile returned: 80040256 filtergraph.c:83: Test failed: Cannot run the graph returned: 0 === w2008s64 (32 bit filtergraph) === quartz: filtergraph.c:165: Test failed: RenderFile returned: 80040256 filtergraph.c:83: Test failed: Cannot run the graph returned: 0 === w7u (32 bit filtergraph) === No test failures found === w8 (32 bit filtergraph) === No test failures found === wvistau64 (64 bit filtergraph) === quartz: filtergraph.c:165: Test failed: RenderFile returned: 80040256 filtergraph.c:83: Test failed: Cannot run the graph returned: 0 === w2008s64 (64 bit filtergraph) === quartz: filtergraph.c:165: Test failed: RenderFile returned: 80040256 filtergraph.c:83: Test failed: Cannot run the graph returned: 0

1 0

Re: [PATCH] d3dx9: Correctly handle whitespace character in D3DXCreateText. (try 2)
by Matteo Bruni Aug. 4, 2015

Aug. 4, 2015

2015-08-02 22:29 GMT+02:00 Józef Kucia <joseph.kucia(a)gmail.com>: > Fixes bug 38976. > > Try 2: Remove useless hr assignments. > --- > dlls/d3dx9_36/mesh.c | 4 ++ > dlls/d3dx9_36/tests/mesh.c | 168 ++++++++++++++++++++++++--------------------- > 2 files changed, 94 insertions(+), 78 deletions(-) > > diff --git a/dlls/d3dx9_36/mesh.c b/dlls/d3dx9_36/mesh.c > index d0459da..39d279d 100644 > --- a/dlls/d3dx9_36/mesh.c > +++ b/dlls/d3dx9_36/mesh.c > @@ -5847,6 +5847,10 @@ static HRESULT triangulate(struct triangulation_array *triangulations) > int i; > struct point2d_index *idx_ptr; > > + /* Glyphs without outlines do not generate any vertices. */ > + if (!glyph->outlines.count) > + return D3D_OK; > + > for (i = 0; i < glyph->outlines.count; i++) > nb_vertices += glyph->outlines.items[i].count; > > diff --git a/dlls/d3dx9_36/tests/mesh.c b/dlls/d3dx9_36/tests/mesh.c > index 0243a61..16c79ea 100644 > --- a/dlls/d3dx9_36/tests/mesh.c > +++ b/dlls/d3dx9_36/tests/mesh.c > @@ -3617,57 +3617,17 @@ static HRESULT create_outline(struct glyphinfo *glyph, void *raw_outline, int da > return S_OK; > } > > -static BOOL compute_text_mesh(struct mesh *mesh, HDC hdc, const char *text, > - float deviation, float extrusion, float otmEMSquare) > +static BOOL compute_text_mesh(struct mesh *mesh, const char *text, > + float deviation, float extrusion, float otmEMSquare, const struct glyphinfo *glyphs) > { > - HRESULT hr = E_FAIL; > DWORD nb_vertices, nb_faces; > DWORD nb_corners, nb_outline_points; > int textlen = 0; > - float offset_x; > - char *raw_outline = NULL; > - struct glyphinfo *glyphs = NULL; > - GLYPHMETRICS gm; > int i; > struct vertex *vertex_ptr; > face *face_ptr; > > - if (deviation == 0.0f) > - deviation = 1.0f / otmEMSquare; > - > textlen = strlen(text); > - glyphs = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, textlen * sizeof(*glyphs)); > - if (!glyphs) { > - hr = E_OUTOFMEMORY; > - goto error; > - } > - > - offset_x = 0.0f; > - for (i = 0; i < textlen; i++) > - { > - /* get outline points from data returned from GetGlyphOutline */ > - const MAT2 identity = {{0, 1}, {0, 0}, {0, 0}, {0, 1}}; > - int datasize; > - > - glyphs[i].offset_x = offset_x; > - > - datasize = GetGlyphOutlineA(hdc, text[i], GGO_NATIVE, &gm, 0, NULL, &identity); > - if (datasize < 0) { > - hr = E_FAIL; > - goto error; > - } > - HeapFree(GetProcessHeap(), 0, raw_outline); > - raw_outline = HeapAlloc(GetProcessHeap(), 0, datasize); > - if (!raw_outline) { > - hr = E_OUTOFMEMORY; > - goto error; > - } > - datasize = GetGlyphOutlineA(hdc, text[i], GGO_NATIVE, &gm, datasize, raw_outline, &identity); > - > - create_outline(&glyphs[i], raw_outline, datasize, deviation, otmEMSquare); > - > - offset_x += gm.gmCellIncX / (float)otmEMSquare; > - } > > /* corner points need an extra vertex for the different side faces normals */ > nb_corners = 0; > @@ -3691,7 +3651,7 @@ static BOOL compute_text_mesh(struct mesh *mesh, HDC hdc, const char *text, > nb_faces = nb_outline_points * 2; > > if (!new_mesh(mesh, nb_vertices, nb_faces)) > - goto error; > + return FALSE; > > /* convert 2D vertices and faces into 3D mesh */ > vertex_ptr = mesh->vertices; > @@ -3801,24 +3761,11 @@ static BOOL compute_text_mesh(struct mesh *mesh, HDC hdc, const char *text, > vertex_ptr++; > } > > - hr = D3D_OK; > -error: > - if (glyphs) { > - for (i = 0; i < textlen; i++) > - { > - int j; > - for (j = 0; j < glyphs[i].outlines.count; j++) > - HeapFree(GetProcessHeap(), 0, glyphs[i].outlines.items[j].items); > - HeapFree(GetProcessHeap(), 0, glyphs[i].outlines.items); > - } > - HeapFree(GetProcessHeap(), 0, glyphs); > - } > - HeapFree(GetProcessHeap(), 0, raw_outline); > - > - return hr == D3D_OK; > + return TRUE; > } > > -static void compare_text_outline_mesh(const char *name, ID3DXMesh *d3dxmesh, struct mesh *mesh, int textlen, float extrusion) > +static void compare_text_outline_mesh(const char *name, ID3DXMesh *d3dxmesh, struct mesh *mesh, > + size_t textlen, float extrusion, const struct glyphinfo *glyphs) > { > HRESULT hr; > DWORD number_of_vertices, number_of_faces; > @@ -3936,13 +3883,20 @@ static void compare_text_outline_mesh(const char *name, ID3DXMesh *d3dxmesh, str > > first_vtx1 = vtx_idx1; > first_vtx2 = vtx_idx2; > - for (; vtx_idx1 < number_of_vertices; vtx_idx1++) { > - if (vertices[vtx_idx1].normal.z != 0) > - break; > - } > - for (; vtx_idx2 < mesh->number_of_vertices; vtx_idx2++) { > - if (mesh->vertices[vtx_idx2].normal.z != 0) > - break; > + /* Glyphs without outlines do not generate any vertices. */ > + if (glyphs[i].outlines.count > 0) > + { > + for (; vtx_idx1 < number_of_vertices; vtx_idx1++) > + { > + if (vertices[vtx_idx1].normal.z != 0) > + break; > + } > + > + for (; vtx_idx2 < mesh->number_of_vertices; vtx_idx2++) > + { > + if (mesh->vertices[vtx_idx2].normal.z != 0) > + break; > + } > } > nb_outline_vertices1 = vtx_idx1 - first_vtx1; > nb_outline_vertices2 = vtx_idx2 - first_vtx2; > @@ -4124,15 +4078,19 @@ error: > static void test_createtext(IDirect3DDevice9 *device, HDC hdc, const char *text, float deviation, float extrusion) > { > HRESULT hr; > - ID3DXMesh *d3dxmesh; > - struct mesh mesh; > + ID3DXMesh *d3dxmesh = NULL; > + struct mesh mesh = {0}; > char name[256]; > OUTLINETEXTMETRICA otm; > GLYPHMETRICS gm; > + struct glyphinfo *glyphs = NULL; > GLYPHMETRICSFLOAT *glyphmetrics_float = HeapAlloc(GetProcessHeap(), 0, sizeof(GLYPHMETRICSFLOAT) * strlen(text)); > int i; > LOGFONTA lf; > + float offset_x; > + size_t textlen; > HFONT font = NULL, oldfont = NULL; > + char *raw_outline = NULL; > > sprintf(name, "text ('%s', %f, %f)", text, deviation, extrusion); > > @@ -4141,7 +4099,7 @@ static void test_createtext(IDirect3DDevice9 *device, HDC hdc, const char *text, > if (hr != D3D_OK) > { > skip("Couldn't create text with D3DXCreateText\n"); > - return; > + goto error; > } > > /* must select a modified font having lfHeight = otm.otmEMSquare before > @@ -4149,21 +4107,25 @@ static void test_createtext(IDirect3DDevice9 *device, HDC hdc, const char *text, > if (!GetObjectA(GetCurrentObject(hdc, OBJ_FONT), sizeof(lf), &lf) > || !GetOutlineTextMetricsA(hdc, sizeof(otm), &otm)) > { > - d3dxmesh->lpVtbl->Release(d3dxmesh); > skip("Couldn't get text outline\n"); > - return; > + goto error; > } > lf.lfHeight = otm.otmEMSquare; > lf.lfWidth = 0; > if (!(font = CreateFontIndirectA(&lf))) > { > - d3dxmesh->lpVtbl->Release(d3dxmesh); > skip("Couldn't create the modified font\n"); > - return; > + goto error; > } > + > + textlen = strlen(text); > + glyphs = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, textlen * sizeof(*glyphs)); > + if (!glyphs) > + goto error; > + > oldfont = SelectObject(hdc, font); > > - for (i = 0; i < strlen(text); i++) > + for (i = 0; i < textlen; i++) > { > const MAT2 identity = {{0, 1}, {0, 0}, {0, 0}, {0, 1}}; > GetGlyphOutlineA(hdc, text[i], GGO_NATIVE, &gm, 0, NULL, &identity); > @@ -4175,8 +4137,42 @@ static void test_createtext(IDirect3DDevice9 *device, HDC hdc, const char *text, > compare_float(glyphmetrics_float[i].gmfCellIncY, gm.gmCellIncY / (float)otm.otmEMSquare); > } > > + if (deviation == 0.0f) > + deviation = 1.0f / otm.otmEMSquare; > + > + offset_x = 0.0f; > + for (i = 0; i < textlen; i++) > + { > + /* get outline points from data returned from GetGlyphOutline */ > + const MAT2 identity = {{0, 1}, {0, 0}, {0, 0}, {0, 1}}; > + int datasize; > + > + glyphs[i].offset_x = offset_x; > + > + datasize = GetGlyphOutlineA(hdc, text[i], GGO_NATIVE, &gm, 0, NULL, &identity); > + if (datasize < 0) > + { > + SelectObject(hdc, oldfont); > + goto error; > + } > + HeapFree(GetProcessHeap(), 0, raw_outline); > + raw_outline = HeapAlloc(GetProcessHeap(), 0, datasize); > + if (!raw_outline) > + { > + SelectObject(hdc, oldfont); > + goto error; > + } > + datasize = GetGlyphOutlineA(hdc, text[i], GGO_NATIVE, &gm, datasize, raw_outline, &identity); > + > + create_outline(&glyphs[i], raw_outline, datasize, deviation, otm.otmEMSquare); > + > + offset_x += gm.gmCellIncX / (float)otm.otmEMSquare; > + } > + > + SelectObject(hdc, oldfont); > + > ZeroMemory(&mesh, sizeof(mesh)); > - if (!compute_text_mesh(&mesh, hdc, text, deviation, extrusion, otm.otmEMSquare)) > + if (!compute_text_mesh(&mesh, text, deviation, extrusion, otm.otmEMSquare, glyphs)) > { > skip("Couldn't create mesh\n"); > d3dxmesh->lpVtbl->Release(d3dxmesh); > @@ -4184,13 +4180,26 @@ static void test_createtext(IDirect3DDevice9 *device, HDC hdc, const char *text, > } > mesh.fvf = D3DFVF_XYZ | D3DFVF_NORMAL; > > - compare_text_outline_mesh(name, d3dxmesh, &mesh, strlen(text), extrusion); > + compare_text_outline_mesh(name, d3dxmesh, &mesh, textlen, extrusion, glyphs); > > +error: > free_mesh(&mesh); > > - d3dxmesh->lpVtbl->Release(d3dxmesh); > - SelectObject(hdc, oldfont); > + if (d3dxmesh) d3dxmesh->lpVtbl->Release(d3dxmesh); > HeapFree(GetProcessHeap(), 0, glyphmetrics_float); > + > + if (glyphs) > + { > + for (i = 0; i < textlen; i++) > + { > + int j; > + for (j = 0; j < glyphs[i].outlines.count; j++) > + HeapFree(GetProcessHeap(), 0, glyphs[i].outlines.items[j].items); > + HeapFree(GetProcessHeap(), 0, glyphs[i].outlines.items); > + } > + HeapFree(GetProcessHeap(), 0, glyphs); > + } > + HeapFree(GetProcessHeap(), 0, raw_outline); > } It looks like there is a minor (preexisting) issue here, the font is not deleted. Maybe fix that in a followup patch, while you're at it? The patch itself is fine to me though.

1 0