February 2016 - Wine-devel - list.winehq.org

Re: [PATCH 3/6] d3d11: Accept SM5 shaders.
by Henri Verbeet 17 Feb '16

17 Feb '16

On 16 February 2016 at 20:48, Matteo Bruni <mbruni(a)codeweavers.com> wrote: > diff --git a/dlls/d3d11/shader.c b/dlls/d3d11/shader.c > index 0e277fc..61dbd4b 100644 > --- a/dlls/d3d11/shader.c > +++ b/dlls/d3d11/shader.c > @@ -389,7 +389,7 @@ static HRESULT d3d_vertex_shader_init(struct d3d_vertex_shader *shader, struct d > desc.byte_code = shader_info.shader_code; > desc.input_signature = &input_signature; > desc.output_signature = &output_signature; > - desc.max_version = 4; > + desc.max_version = 5; I think this should depend on the feature level. I.e., I'd expect creating a SM5 shader to fail on a D3D_FEATURE_LEVEL_10_0 device. (And we'd need feature level support to properly skip the test in patch 6/6, otherwise creating the device would succeed on SM4 hardware, but creating the shader would fail.)

1 0

Re: [PATCH 1/6] wined3d: Add a setting to check relative addressing indices when accessing uniforms.
by Henri Verbeet 17 Feb '16

17 Feb '16

On 16 February 2016 at 20:48, Matteo Bruni <mbruni(a)codeweavers.com> wrote: > @@ -2258,9 +2258,25 @@ static void shader_glsl_get_register_name(const struct wined3d_shader_register * > if (reg->idx[0].rel_addr) > { > if (reg->idx[0].offset) > - sprintf(register_name, "%s_c[%s + %u]", prefix, rel_param0.param_str, reg->idx[0].offset); > + { > + if (wined3d_settings.check_float_constants) > + sprintf(register_name, "(%s + %u >= 0 && %s + %u < %u ? %s_c[%s + %u] : vec4(0.0))", > + rel_param0.param_str, reg->idx[0].offset, > + rel_param0.param_str, reg->idx[0].offset, shader->limits->constant_float, > + prefix, rel_param0.param_str, reg->idx[0].offset); > + else > + sprintf(register_name, "%s_c[%s + %u]", prefix, > + rel_param0.param_str, reg->idx[0].offset); > + } > else > - sprintf(register_name, "%s_c[%s]", prefix, rel_param0.param_str); > + { > + if (wined3d_settings.check_float_constants) > + sprintf(register_name, "(%s >= 0 && %s < %u ? %s_c[%s] : vec4(0.0))", > + rel_param0.param_str, rel_param0.param_str, shader->limits->constant_float, > + prefix, rel_param0.param_str); > + else > + sprintf(register_name, "%s_c[%s]", prefix, rel_param0.param_str); > + } > } > else > { I'd probably just drop the optimization for 0 offsets for check_float_constants. I.e.: if (wined3d_settings.check_float_constants) ... else if (reg->idx[0].offset) ... else ...

1 0

Re: [PATCH] gdi32/tests: Check values returned from GetPath
by Huw Davies 17 Feb '16

17 Feb '16

On Wed, Feb 17, 2016 at 02:01:25PM +1100, Alistair Leslie-Hughes wrote: > diff --git a/dlls/gdi32/tests/metafile.c b/dlls/gdi32/tests/metafile.c > index 6804a96..b3f164d 100644 > --- a/dlls/gdi32/tests/metafile.c > +++ b/dlls/gdi32/tests/metafile.c > @@ -3361,12 +3361,46 @@ static void test_emf_polybezier(void) > DeleteEnhMetaFile(hemf); > } > > +static const unsigned char EMF_PATH_BITS[] = > +{ > + 0x01, 0x00, 0x00, 0x00, 0x6c, 0x00, 0x00, 0x00, > + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, > + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, > + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, > + 0xdd, 0xff, 0xff, 0xff, 0xdd, 0xff, 0xff, 0xff, > + 0x20, 0x45, 0x4d, 0x46, 0x00, 0x00, 0x01, 0x00, > + 0xe0, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, > + 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, > + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, > + 0x00, 0x05, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, > + 0xc4, 0x01, 0x00, 0x00, 0x69, 0x01, 0x00, 0x00, > + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, > + 0x00, 0x00, 0x00, 0x00, 0xe3, 0xe3, 0x06, 0x00, > + 0x1c, 0x83, 0x05, 0x00, 0x3b, 0x00, 0x00, 0x00, > + 0x08, 0x00, 0x00, 0x00, 0x1b, 0x00, 0x00, 0x00, > + 0x10, 0x00, 0x00, 0x00, 0x32, 0x00, 0x00, 0x00, > + 0x32, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, > + 0x10, 0x00, 0x00, 0x00, 0x32, 0x00, 0x00, 0x00, > + 0x96, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, > + 0x10, 0x00, 0x00, 0x00, 0x96, 0x00, 0x00, 0x00, > + 0x96, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, > + 0x10, 0x00, 0x00, 0x00, 0x96, 0x00, 0x00, 0x00, > + 0x32, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x00, > + 0x10, 0x00, 0x00, 0x00, 0x32, 0x00, 0x00, 0x00, > + 0x32, 0x00, 0x00, 0x00, 0x3c, 0x00, 0x00, 0x00, > + 0x08, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x00, 0x00, > + 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, > + 0x10, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00 > +}; > + > static void test_emf_GetPath(void) > { > HDC hdcMetafile; > HENHMETAFILE hemf; > BOOL ret; > int size; > + POINT *points; > + BYTE *types; > > SetLastError(0xdeadbeef); > hdcMetafile = CreateEnhMetaFileA(GetDC(0), NULL, NULL, NULL); > @@ -3387,10 +3421,37 @@ static void test_emf_GetPath(void) > > size = GetPath(hdcMetafile, NULL, NULL, 0); > todo_wine ok( size == 5, "GetPath returned %d.\n", size); > + if(size == 5) > + { > + points = HeapAlloc(GetProcessHeap(), 0, size * sizeof(POINT)); > + types = HeapAlloc(GetProcessHeap(), 0, size ); > + > + GetPath(hdcMetafile, points, types, size); > + > + ok(types[0] == PT_MOVETO, "unexpected value %d\n", types[0]); > + ok(points[0].x == 50 && points[0].y == 50, "unexpected point (%d,%d)\n", points[0].x, points[0].y); > + ok(types[1] == PT_LINETO, "unexpected value %d\n", types[1]); > + ok(points[1].x == 50 && points[1].y == 150, "unexpected point (%d,%d)\n", points[1].x, points[1].y); > + ok(types[2] == PT_LINETO, "unexpected value %d\n", types[2]); > + ok(points[2].x == 150 && points[2].y == 150, "unexpected point (%d,%d)\n", points[2].x, points[2].y); > + ok(types[3] == PT_LINETO, "unexpected value %d\n", types[3]); > + ok(points[3].x == 150 && points[3].y == 50, "unexpected point (%d,%d)\n", points[3].x, points[3].y); > + ok(types[4] == PT_LINETO, "unexpected value %d\n", types[4]); > + ok(points[4].x == 50 && points[4].y == 50, "unexpected point (%d,%d)\n", points[4].x, points[4].y); > + > + HeapFree(GetProcessHeap(), 0, points); > + HeapFree(GetProcessHeap(), 0, types); > + } > > hemf = CloseEnhMetaFile(hdcMetafile); > ok(hemf != 0, "CloseEnhMetaFile error %d\n", GetLastError()); > > + if (compare_emf_bits(hemf, EMF_PATH_BITS, sizeof(EMF_PATH_BITS), "test_emf_GetPath", FALSE) != 0) > + { > + dump_emf_bits(hemf, "test_emf_GetPath"); > + dump_emf_records(hemf, "test_emf_GetPath"); > + } > + > DeleteEnhMetaFile(hemf); > } > Please add a Rectangle() call to the path building code. This will show that the metafile doesn't just consist of path elements. Also, as discussed yesterday, please add a similar test for wmfs (this could be a separate patch). Thanks, Huw.

1 0

Re: [PATCH 3/3] advapi32/tests: Add more invalid parameter tests for NtAccessCheck.
by Marvin 17 Feb '16

17 Feb '16

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=20604 Your paranoid android. === w1064 (32 bit security) === security.c:4011: Test failed: MACHINE\Software owner SID (S-1-5-18) != Administrators SID (S-1-5-32-544). security.c:4052: Test failed: Builtin Users ACE has unexpected flags (0x12 != 0xa) security.c:4056: Test failed: Builtin Users ACE has unexpected mask (0x20019 != 0x80000000) security.c:4065: Test failed: Builtin Admins ACE has unexpected flags (0x12 != 0x0)

1 0

Re: [PATCH 1/2] advapi32/tests: Add invalid paramater tests for AccessCheck.
by Sebastian Lackner 17 Feb '16

17 Feb '16

On 15.02.2016 21:52, Qian Hong wrote: > > Signed-off-by: Qian Hong <qhong(a)codeweavers.com> > --- > dlls/advapi32/tests/security.c | 53 ++++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 53 insertions(+) > > > > 0001-advapi32-tests-Add-invalid-paramater-tests-for-AccessC.txt > > > diff --git a/dlls/advapi32/tests/security.c b/dlls/advapi32/tests/security.c > index 742ede1..69c0f77 100644 > --- a/dlls/advapi32/tests/security.c > +++ b/dlls/advapi32/tests/security.c > @@ -1411,6 +1411,59 @@ static void test_AccessCheck(void) > GetLastError()); > trace("AccessCheck with MAXIMUM_ALLOWED got Access 0x%08x\n", Access); > > + /* Null PrivSet with null PrivSetLen pointer */ > + SetLastError(0xdeadbeef); > + Access = AccessStatus = 0x1abe11ed; > + ret = AccessCheck(SecurityDescriptor, Token, KEY_READ, &Mapping, > + NULL, NULL, &Access, &AccessStatus); > + err = GetLastError(); > + ok(!ret && err == ERROR_NOACCESS, "AccessCheck should have " > + "failed with ERROR_NOACCESS, instead of %d\n", err); > + ok(Access == 0x1abe11ed && AccessStatus == 0x1abe11ed, > + "Access and/or AccessStatus were changed!\n"); > + > + /* Null PrivSet with zero PrivSetLen */ > + SetLastError(0xdeadbeef); > + Access = AccessStatus = 0x1abe11ed; > + PrivSetLen = 0; > + ret = AccessCheck(SecurityDescriptor, Token, KEY_READ, &Mapping, > + 0, &PrivSetLen, &Access, &AccessStatus); > + err = GetLastError(); > +todo_wine > + ok(!ret && err == ERROR_INSUFFICIENT_BUFFER, "AccessCheck should have " > + "failed with ERROR_INSUFFICIENT_BUFFER, instead of %d\n", err); > +todo_wine > + ok(PrivSetLen == sizeof(PRIVILEGE_SET), "PrivSetLen returns %d, expects %d\n", PrivSetLen, sizeof(PRIVILEGE_SET)); This will lead to compile warnings on 64-bit. > + ok(Access == 0x1abe11ed && AccessStatus == 0x1abe11ed, > + "Access and/or AccessStatus were changed!\n"); > + > + /* Valid PrivSet with zero PrivSetLen */ > + SetLastError(0xdeadbeef); > + Access = AccessStatus = 0x1abe11ed; > + PrivSetLen = 0; > + ret = AccessCheck(SecurityDescriptor, Token, KEY_READ, &Mapping, > + PrivSet, &PrivSetLen, &Access, &AccessStatus); > + err = GetLastError(); > +todo_wine > + ok(!ret && err == ERROR_INSUFFICIENT_BUFFER, "AccessCheck should have " > + "failed with ERROR_INSUFFICIENT_BUFFER, instead of %d\n", err); > +todo_wine > + ok(Access == 0x1abe11ed && AccessStatus == 0x1abe11ed, > + "Access and/or AccessStatus were changed!\n"); > + > + PrivSetLen = FIELD_OFFSET(PRIVILEGE_SET, Privilege[16]); > + > + /* Null PrivSet with valid PrivSetLen */ > + SetLastError(0xdeadbeef); > + Access = AccessStatus = 0x1abe11ed; > + ret = AccessCheck(SecurityDescriptor, Token, KEY_READ, &Mapping, > + 0, &PrivSetLen, &Access, &AccessStatus); > + err = GetLastError(); > + ok(!ret && err == ERROR_NOACCESS, "AccessCheck should have " > + "failed with ERROR_NOACCESS, instead of %d\n", err); > + ok(Access == 0x1abe11ed && AccessStatus == 0x1abe11ed, > + "Access and/or AccessStatus were changed!\n"); > + > /* Access denied by SD */ > SetLastError(0xdeadbeef); > Access = AccessStatus = 0x1abe11ed; > > >

2 2

Re: [PATCH 2/3] ntdll: Improve invalid parameter handling in NtAccessCheck. (try 2)
by Marvin 17 Feb '16

17 Feb '16

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=20603 Your paranoid android. === w1064 (32 bit security) === security.c:3892: Test failed: MACHINE\Software owner SID (S-1-5-18) != Administrators SID (S-1-5-32-544). security.c:3933: Test failed: Builtin Users ACE has unexpected flags (0x12 != 0xa) security.c:3937: Test failed: Builtin Users ACE has unexpected mask (0x20019 != 0x80000000) security.c:3946: Test failed: Builtin Admins ACE has unexpected flags (0x12 != 0x0)

1 0

Re: [PATCH 2/2] ntdll: Improve invalid paramater handling in NtAccessCheck.
by Sebastian Lackner 17 Feb '16

17 Feb '16

On 15.02.2016 21:53, Qian Hong wrote: > > Signed-off-by: Qian Hong <qhong(a)codeweavers.com> > --- > dlls/advapi32/tests/security.c | 4 ---- > dlls/ntdll/sec.c | 11 ++++++++++- > 2 files changed, 10 insertions(+), 5 deletions(-) > > > > 0002-ntdll-Improve-invalid-paramater-handling-in-NtAccessCh.txt > > > diff --git a/dlls/advapi32/tests/security.c b/dlls/advapi32/tests/security.c > index 69c0f77..08666dd 100644 > --- a/dlls/advapi32/tests/security.c > +++ b/dlls/advapi32/tests/security.c > @@ -1429,10 +1429,8 @@ static void test_AccessCheck(void) > ret = AccessCheck(SecurityDescriptor, Token, KEY_READ, &Mapping, > 0, &PrivSetLen, &Access, &AccessStatus); > err = GetLastError(); > -todo_wine > ok(!ret && err == ERROR_INSUFFICIENT_BUFFER, "AccessCheck should have " > "failed with ERROR_INSUFFICIENT_BUFFER, instead of %d\n", err); > -todo_wine > ok(PrivSetLen == sizeof(PRIVILEGE_SET), "PrivSetLen returns %d, expects %d\n", PrivSetLen, sizeof(PRIVILEGE_SET)); > ok(Access == 0x1abe11ed && AccessStatus == 0x1abe11ed, > "Access and/or AccessStatus were changed!\n"); > @@ -1444,10 +1442,8 @@ todo_wine > ret = AccessCheck(SecurityDescriptor, Token, KEY_READ, &Mapping, > PrivSet, &PrivSetLen, &Access, &AccessStatus); > err = GetLastError(); > -todo_wine > ok(!ret && err == ERROR_INSUFFICIENT_BUFFER, "AccessCheck should have " > "failed with ERROR_INSUFFICIENT_BUFFER, instead of %d\n", err); > -todo_wine > ok(Access == 0x1abe11ed && AccessStatus == 0x1abe11ed, > "Access and/or AccessStatus were changed!\n"); > > diff --git a/dlls/ntdll/sec.c b/dlls/ntdll/sec.c > index 125c86e..c32ae0c 100644 > --- a/dlls/ntdll/sec.c > +++ b/dlls/ntdll/sec.c > @@ -1586,7 +1586,16 @@ NtAccessCheck( > SecurityDescriptor, ClientToken, DesiredAccess, GenericMapping, > PrivilegeSet, ReturnLength, GrantedAccess, AccessStatus); > > - if (!PrivilegeSet || !ReturnLength) > + if (!ReturnLength) > + return STATUS_ACCESS_VIOLATION; > + > + if (*ReturnLength == 0) > + { > + *ReturnLength = sizeof(PRIVILEGE_SET); > + return STATUS_BUFFER_TOO_SMALL; > + } This looks a bit hacky. The code below assumes that *ReturnLength > FIELD_OFFSET( PRIVILEGE_SET, Privilege ), so it would be interesting to know what happens for sizes 0 ... 8. > + > + if (!PrivilegeSet) > return STATUS_ACCESS_VIOLATION; > > SERVER_START_REQ( access_check ) > > >

3 2

Re: [PATCH 1/3] advapi32/tests: Add more invalid parameter tests for AccessCheck.
by Marvin 17 Feb '16

17 Feb '16

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=20602 Your paranoid android. === w1064 (32 bit security) === security.c:3905: Test failed: MACHINE\Software owner SID (S-1-5-18) != Administrators SID (S-1-5-32-544). security.c:3946: Test failed: Builtin Users ACE has unexpected flags (0x12 != 0xa) security.c:3950: Test failed: Builtin Users ACE has unexpected mask (0x20019 != 0x80000000) security.c:3959: Test failed: Builtin Admins ACE has unexpected flags (0x12 != 0x0)

1 0

Re: [PATCH] gdi32: Remove no-op abs(3) call.
by Huw Davies 17 Feb '16

17 Feb '16

On Tue, Feb 16, 2016 at 01:55:34PM -0700, Charles Davis wrote: > Signed-off-by: Charles Davis <cdavis5x(a)gmail.com> > --- > dlls/gdi32/pen.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/dlls/gdi32/pen.c b/dlls/gdi32/pen.c > index 2ae2f2b..6793934 100644 > --- a/dlls/gdi32/pen.c > +++ b/dlls/gdi32/pen.c > @@ -195,7 +195,7 @@ HPEN WINAPI ExtCreatePen( DWORD style, DWORD width, > if (logbrush.lbStyle == BS_DIBPATTERN) logbrush.lbStyle = BS_DIBPATTERNPT; > > penPtr->logpen.elpPenStyle = style; > - penPtr->logpen.elpWidth = abs(width); > + penPtr->logpen.elpWidth = width; > penPtr->logpen.elpBrushStyle = logbrush.lbStyle; > penPtr->logpen.elpColor = logbrush.lbColor; > penPtr->logpen.elpHatch = brush->lbHatch; This breaks a test: ../../../tools/runtest -q -P wine -T ../../.. -M gdi32.dll -p gdi32_test.exe.so pen && touch pen.ok pen.c:440: Test failed: expected 123, got ffffff85 width is actually treated as signed. Huw.

1 0

Re: [PATCH] advapi32, ntdll: Skip test_redirection tests if there is not enough privileges
by Marvin 17 Feb '16

17 Feb '16

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=20601 Your paranoid android. === build (build) === Patch failed to apply

1 0