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

Re: [PATCH 3/9] d3d9/tests: Add test for regression in IDirect3DDevice9_Reset().
by Marvin 16 Aug '16

16 Aug '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=25061 Your paranoid android. === w864 (64 bit device) === device.c:4392: Test failed: cw is 0xf60, expected 0x7f. device.c:4401: Test failed: Callback cw is 0xf60, expected 0x7f. === w1064 (64 bit device) === device.c:4392: Test failed: cw is 0xf60, expected 0x7f. device.c:4401: Test failed: Callback cw is 0xf60, expected 0x7f. === w1064 (32 bit d3d9ex) === d3d9ex.c:2411: Test failed: Got unexpected hr 0, i=0. d3d9ex.c:2413: Test failed: Got unexpected hr 0, i=0. d3d9ex.c:2411: Test failed: Got unexpected hr 0, i=1. d3d9ex.c:2413: Test failed: Got unexpected hr 0, i=1. === w1064 (64 bit d3d9ex) === d3d9ex.c:2411: Test failed: Got unexpected hr 0, i=0. d3d9ex.c:2413: Test failed: Got unexpected hr 0, i=0. d3d9ex.c:2411: Test failed: Got unexpected hr 0, i=1. d3d9ex.c:2413: Test failed: Got unexpected hr 0, i=1.

1 0

Re: d3dx9: Add D3DXFrameFind stub.
by Nikolay Sivov 16 Aug '16

16 Aug '16

On 12.08.2016 13:00, Andrey Gusev wrote: > +D3DXFRAME * WINAPI D3DXFrameFind(const D3DXFRAME *frame_root, const char *name) > +{ > + FIXME("frame_root %p, name %s stub!\n", frame_root, debugstr_a(name)); > + > + return NULL; > +} This looks trivial enough to implement. It's probably just going from pFrameFirstChild -> pFrameSibling. The question is does it check grandchildren, but it's easy to test.

2 1

Re: [PATCH 2/2] d3drm: Implement IDirect3DRMViewport*::Clear. (v3 resend)
by Henri Verbeet 16 Aug '16

16 Aug '16

On 15 August 2016 at 21:50, Aaryaman Vasishta <jem456.vasishta(a)gmail.com> wrote: > + memcpy(&driver, &IID_IDirect3DRGBDevice, sizeof(GUID)); Why not "driver = IID_IDirect3DRGBDevice;"? It probably also makes sense to introduce a helper function to create d3drm devices at some point. > + hr = IDirect3DMaterial_SetMaterial(viewport->material, &mat); > + > + return hr; "return IDirect3DMaterial_SetMaterial(viewport->material, &mat);" > + if (FAILED(hr = IDirect3DViewport_Clear(viewport->d3d_viewport, 1, &clear_rect, clear_flags))) > + return hr; Formatting.

1 0

Re: [PATCH] po: Update Simplified Chinese translation.
by Jactry Zeng 16 Aug '16

16 Aug '16

It looks good to me. Thanks! 2016-08-16 14:28 GMT+08:00 Mingcong Bai <jeffbaichina(a)gmail.com>: > Signed-off-by: Mingcong Bai <jeffbai(a)aosc.xyz> > --- > po/zh_CN.po | 342 ++++++++++++++++++++++++++++++ > ------------------------------ > 1 file changed, 171 insertions(+), 171 deletions(-) > > > > -- Regards, Jactry Zeng

1 0

Re: [PATCH 2/9] d3d8/tests: Add test for regression in IDirect3DDevice8_Reset().
by Henri Verbeet 16 Aug '16

16 Aug '16

On 16 August 2016 at 12:51, Henri Verbeet <hverbeet(a)codeweavers.com> wrote: > Signed-off-by: Henri Verbeet <hverbeet(a)codeweavers.com> > Or well, maybe I should hold off on that for a bit. It works here though.

2 1

Re: [PATCH] msvcrt: Add _wcreate_locale (try2)
by Piotr Caban 16 Aug '16

16 Aug '16

On 08/15/16 22:02, Daniel Lehman wrote: > + lcl = p_wcreate_locale(LC_ALL, german); You should probably skip this test on systems that doesn't support German locale. Thanks, Piotr

1 0

[v2 PATCH] inetcomm: Add MimeOleObjectFromMoniker stub
by Andrey Gusev 15 Aug '16

15 Aug '16

> FIXME("(0x%08x, %p, %p, %s, %p %p) stub\n", bindf, moniker, binding, debugstr_guid(riid), out, moniker_new); You missed a comma.

1 0

Re: [v4] riched20: allow setting background text color
by Huw Davies 15 Aug '16

15 Aug '16

On Sun, Aug 14, 2016 at 03:07:25PM -0600, Kenneth Haley wrote: > fixes https://bugs.winehq.org/show_bug.cgi?id=36625 Actually I missed that you hadn't added a Signed-off-by: line, git format-patch -s will do this for you. Please resend with one. Thanks, Huw.

1 0

Re: [PATCH] ntdll: Set Rip in for longjmp in RtlRestoreContext
by Sebastian Lackner 15 Aug '16

15 Aug '16

On 11.08.2016 19:27, Daniel Lehman wrote: > Some tests will fail but are unrelated to this patch: > https://urldefense.proofpoint.com/v2/url?u=http-3A__test.winehq.org_data_88… > https://urldefense.proofpoint.com/v2/url?u=http-3A__test.winehq.org_data_88… > https://urldefense.proofpoint.com/v2/url?u=http-3A__test.winehq.org_data_88… > > > 0001-ntdll-Set-Rip-in-for-longjmp-in-RtlRestoreContext.txt > > > From eaab70ee4ddc866526df0341853ccdb599cf7674 Mon Sep 17 00:00:00 2001 > From: Daniel Lehman <dlehman(a)esri.com> > Date: Mon, 1 Aug 2016 14:52:32 -0700 > Subject: [PATCH] ntdll: Set Rip in for longjmp in RtlRestoreContext > > Signed-off-by: Daniel Lehman <dlehman(a)esri.com> > --- > dlls/ntdll/signal_x86_64.c | 1 + > dlls/ntdll/tests/exception.c | 202 +++++++++++++++++++++++++++++++++++++++++++ > 2 files changed, 203 insertions(+) > > diff --git a/dlls/ntdll/signal_x86_64.c b/dlls/ntdll/signal_x86_64.c > index 4c88536..598e08a 100644 > --- a/dlls/ntdll/signal_x86_64.c > +++ b/dlls/ntdll/signal_x86_64.c > @@ -3513,6 +3513,7 @@ void WINAPI RtlRestoreContext( CONTEXT *context, EXCEPTION_RECORD *rec ) > context->R13 = jmp->R13; > context->R14 = jmp->R14; > context->R15 = jmp->R15; > + context->Rip = jmp->Rip; > context->u.s.Xmm6 = jmp->Xmm6; > context->u.s.Xmm7 = jmp->Xmm7; > context->u.s.Xmm8 = jmp->Xmm8; Please note that this also modifies the behavior of RtlUnwindEx (which gets the target_ip passed directly as a function argument). With your change it will be ignored, which might need additional (at least manual) tests to make sure this is correct. > diff --git a/dlls/ntdll/tests/exception.c b/dlls/ntdll/tests/exception.c > index a086436..4846769 100644 > --- a/dlls/ntdll/tests/exception.c > +++ b/dlls/ntdll/tests/exception.c > @@ -53,10 +53,44 @@ static NTSTATUS (WINAPI *pNtSetInformationProcess)(HANDLE, PROCESSINFOCLASS, PV > static BOOL (WINAPI *pIsWow64Process)(HANDLE, PBOOL); > > #if defined(__x86_64__) > +typedef struct _SETJMP_FLOAT128 > +{ > + unsigned __int64 DECLSPEC_ALIGN(16) Part[2]; > +} SETJMP_FLOAT128; > + > +typedef struct _JUMP_BUFFER > +{ > + unsigned __int64 Frame; > + unsigned __int64 Rbx; > + unsigned __int64 Rsp; > + unsigned __int64 Rbp; > + unsigned __int64 Rsi; > + unsigned __int64 Rdi; > + unsigned __int64 R12; > + unsigned __int64 R13; > + unsigned __int64 R14; > + unsigned __int64 R15; > + unsigned __int64 Rip; > + unsigned __int64 Spare; > + SETJMP_FLOAT128 Xmm6; > + SETJMP_FLOAT128 Xmm7; > + SETJMP_FLOAT128 Xmm8; > + SETJMP_FLOAT128 Xmm9; > + SETJMP_FLOAT128 Xmm10; > + SETJMP_FLOAT128 Xmm11; > + SETJMP_FLOAT128 Xmm12; > + SETJMP_FLOAT128 Xmm13; > + SETJMP_FLOAT128 Xmm14; > + SETJMP_FLOAT128 Xmm15; > +} _JUMP_BUFFER; > + > static BOOLEAN (CDECL *pRtlAddFunctionTable)(RUNTIME_FUNCTION*, DWORD, DWORD64); > static BOOLEAN (CDECL *pRtlDeleteFunctionTable)(RUNTIME_FUNCTION*); > static BOOLEAN (CDECL *pRtlInstallFunctionTableCallback)(DWORD64, DWORD64, DWORD, PGET_RUNTIME_FUNCTION_CALLBACK, PVOID, PCWSTR); > static PRUNTIME_FUNCTION (WINAPI *pRtlLookupFunctionEntry)(ULONG64, ULONG64*, UNWIND_HISTORY_TABLE*); > +static VOID (WINAPI *pRtlCaptureContext)(CONTEXT*); > +static VOID (CDECL *pRtlRestoreContext)(CONTEXT*, EXCEPTION_RECORD*); > +static int (CDECL *p_setjmp)(_JUMP_BUFFER*); > #endif > > #ifdef __i386__ > @@ -1631,6 +1665,164 @@ static void test_virtual_unwind(void) > call_virtual_unwind( i, &tests[i] ); > } > > +static int consolidate_dummy_called; > +static PVOID CALLBACK test_consolidate_dummy(EXCEPTION_RECORD *rec) > +{ > + CONTEXT *ctx = (CONTEXT *)rec->ExceptionInformation[1]; > + consolidate_dummy_called = 1; > + ok(ctx->Rip == 0xdeadbeef, "test_consolidate_dummy failed for Rip, expected: 0xdeadbeef, got: %lx\n", ctx->Rip); > + return (PVOID)rec->ExceptionInformation[2]; > +} > + > +static void test_restore_context(void) > +{ > + XMM_SAVE_AREA32 *fltsave; > + EXCEPTION_RECORD rec; > + volatile int done; /* gcc 4.5.2 rearranges 'done' and RtlCaptureContext */ > + volatile int pass; A better way to make sure the compiler does not rearrange the code would be to use Interlocked*() functions. > + _JUMP_BUFFER buf; > + CONTEXT ctx; > + > + if (!pRtlRestoreContext || !pRtlCaptureContext || !p_setjmp) > + { > + skip("RtlCaptureContext/RtlRestoreContext/_setjmp not found\n"); > + return; > + } > + > + /* RtlRestoreContext(NULL, NULL); crashes on Windows */ > + > + /* test simple case of capture and restore context */ > + pass = 0; > + done = 0; > + pRtlCaptureContext(&ctx); > + switch (pass) > + { > + case 0: > + ok(!done, "done on first pass\n"); > + break; > + case 1: > + ok(done, "not done on second pass\n"); > + break; > + default: > + ok(0, "bad pass = %i done = %i\n", pass, done); > + break; > + } > + ++pass; > + if (done) goto test_longjmp; > + done = 1; I don't think the second variable adds much value here. A simplified version like below would test the same things: pass = 0; pRtlCaptureContext(&ctx); ok(pass < 2, "unexpected pass = %d\n", pass); if (!pass++) pRtlRestoreContext(&ctx, NULL); > + > + pRtlRestoreContext(&ctx, NULL); > + > +test_longjmp: I do not see any need to use goto here, regular if () checks would also do the job. > + fltsave = (XMM_SAVE_AREA32 *)((char*)&ctx + 0x100); It is better to avoid such hardcoded constants. Why not access the field directly? > + done = 0; > + pRtlCaptureContext(&ctx); What is the purpose of the RtlCaptureContext call here? The result does not seem to be used. After RtlRestoreContext() the ctx struct has been filled with the jump buffer values. > + p_setjmp(&buf); > + if (done) > + { > + ok(buf.Rbx == ctx.Rbx, "RtlRestoreContext:longjump failed for Rbx, expected: %lx, got: %lx\n", buf.Rbx, ctx.Rbx); > + ok(buf.Rsp == ctx.Rsp, "RtlRestoreContext:longjump failed for Rsp, expected: %lx, got: %lx\n", buf.Rsp, ctx.Rsp); > + ok(buf.Rbp == ctx.Rbp, "RtlRestoreContext:longjump failed for Rbp, expected: %lx, got: %lx\n", buf.Rbp, ctx.Rbp); > + ok(buf.Rsi == ctx.Rsi, "RtlRestoreContext:longjump failed for Rsi, expected: %lx, got: %lx\n", buf.Rsi, ctx.Rsi); > + ok(buf.Rdi == ctx.Rdi, "RtlRestoreContext:longjump failed for Rdi, expected: %lx, got: %lx\n", buf.Rdi, ctx.Rdi); > + ok(buf.R12 == ctx.R12, "RtlRestoreContext:longjump failed for R12, expected: %lx, got: %lx\n", buf.R12, ctx.R12); > + ok(buf.R13 == ctx.R13, "RtlRestoreContext:longjump failed for R13, expected: %lx, got: %lx\n", buf.R13, ctx.R13); > + ok(buf.R14 == ctx.R14, "RtlRestoreContext:longjump failed for R14, expected: %lx, got: %lx\n", buf.R14, ctx.R14); > + ok(buf.R15 == ctx.R15, "RtlRestoreContext:longjump failed for R15, expected: %lx, got: %lx\n", buf.R15, ctx.R15); > + > + ok(buf.Xmm6.Part[1] == fltsave->XmmRegisters[6].High && buf.Xmm6.Part[0] == fltsave->XmmRegisters[6].Low, > + "RtlRestoreContext:longjump failed for Xmm6, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm6.Part[1], buf.Xmm6.Part[0], fltsave->XmmRegisters[6].High, fltsave->XmmRegisters[6].Low); > + ok(buf.Xmm7.Part[1] == fltsave->XmmRegisters[7].High && buf.Xmm7.Part[0] == fltsave->XmmRegisters[7].Low, > + "RtlRestoreContext:longjump failed for Xmm7, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm7.Part[1], buf.Xmm7.Part[0], fltsave->XmmRegisters[7].High, fltsave->XmmRegisters[7].Low); > + ok(buf.Xmm8.Part[1] == fltsave->XmmRegisters[8].High && buf.Xmm8.Part[0] == fltsave->XmmRegisters[8].Low, > + "RtlRestoreContext:longjump failed for Xmm8, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm8.Part[1], buf.Xmm8.Part[0], fltsave->XmmRegisters[8].High, fltsave->XmmRegisters[8].Low); > + ok(buf.Xmm9.Part[1] == fltsave->XmmRegisters[9].High && buf.Xmm9.Part[0] == fltsave->XmmRegisters[9].Low, > + "RtlRestoreContext:longjump failed for Xmm9, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm9.Part[1], buf.Xmm9.Part[0], fltsave->XmmRegisters[9].High, fltsave->XmmRegisters[9].Low); > + ok(buf.Xmm10.Part[1] == fltsave->XmmRegisters[10].High && buf.Xmm10.Part[0] == fltsave->XmmRegisters[10].Low, > + "RtlRestoreContext:longjump failed for Xmm10, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm10.Part[1], buf.Xmm10.Part[0], fltsave->XmmRegisters[10].High, fltsave->XmmRegisters[10].Low); > + ok(buf.Xmm11.Part[1] == fltsave->XmmRegisters[11].High && buf.Xmm11.Part[0] == fltsave->XmmRegisters[11].Low, > + "RtlRestoreContext:longjump failed for Xmm11, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm11.Part[1], buf.Xmm11.Part[0], fltsave->XmmRegisters[11].High, fltsave->XmmRegisters[11].Low); > + ok(buf.Xmm12.Part[1] == fltsave->XmmRegisters[12].High && buf.Xmm12.Part[0] == fltsave->XmmRegisters[12].Low, > + "RtlRestoreContext:longjump failed for Xmm12, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm12.Part[1], buf.Xmm12.Part[0], fltsave->XmmRegisters[12].High, fltsave->XmmRegisters[12].Low); > + ok(buf.Xmm13.Part[1] == fltsave->XmmRegisters[13].High && buf.Xmm13.Part[0] == fltsave->XmmRegisters[13].Low, > + "RtlRestoreContext:longjump failed for Xmm13, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm13.Part[1], buf.Xmm13.Part[0], fltsave->XmmRegisters[13].High, fltsave->XmmRegisters[13].Low); > + ok(buf.Xmm14.Part[1] == fltsave->XmmRegisters[14].High && buf.Xmm14.Part[0] == fltsave->XmmRegisters[14].Low, > + "RtlRestoreContext:longjump failed for Xmm14, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm14.Part[1], buf.Xmm14.Part[0], fltsave->XmmRegisters[14].High, fltsave->XmmRegisters[14].Low); > + ok(buf.Xmm15.Part[1] == fltsave->XmmRegisters[15].High && buf.Xmm15.Part[0] == fltsave->XmmRegisters[15].Low, > + "RtlRestoreContext:longjump failed for Xmm15, expected: %lx%08lx, got: %lx%08lx\n", > + buf.Xmm15.Part[1], buf.Xmm15.Part[0], fltsave->XmmRegisters[15].High, fltsave->XmmRegisters[15].Low); > + > + goto test_consolidate; Similar to above, it would be better to avoid goto. All the code below could just be moved to an else branch. > + } > + done = 1; > + > + rec.ExceptionCode = STATUS_LONGJUMP; > + rec.NumberParameters = 1; > + rec.ExceptionInformation[0] = (DWORD64)&buf; > + > + ctx.Rbx = 0xdeadbeef; > + ctx.Rsp = 0xdeadbeef; > + ctx.Rbp = 0xdeadbeef; > + ctx.Rsi = 0xdeadbeef; > + ctx.Rdi = 0xdeadbeef; > + ctx.R12 = 0xdeadbeef; > + ctx.R13 = 0xdeadbeef; > + ctx.R14 = 0xdeadbeef; > + ctx.R15 = 0xdeadbeef; > + > + fltsave->XmmRegisters[6].High = 0xbaadf00d; > + fltsave->XmmRegisters[6].Low = 0xdeadbeef; > + fltsave->XmmRegisters[7].High = 0xbaadf00d; > + fltsave->XmmRegisters[7].Low = 0xdeadbeef; > + fltsave->XmmRegisters[8].High = 0xbaadf00d; > + fltsave->XmmRegisters[8].Low = 0xdeadbeef; > + fltsave->XmmRegisters[9].High = 0xbaadf00d; > + fltsave->XmmRegisters[9].Low = 0xdeadbeef; > + fltsave->XmmRegisters[10].High = 0xbaadf00d; > + fltsave->XmmRegisters[10].Low = 0xdeadbeef; > + fltsave->XmmRegisters[11].High = 0xbaadf00d; > + fltsave->XmmRegisters[11].Low = 0xdeadbeef; > + fltsave->XmmRegisters[12].High = 0xbaadf00d; > + fltsave->XmmRegisters[12].Low = 0xdeadbeef; > + fltsave->XmmRegisters[13].High = 0xbaadf00d; > + fltsave->XmmRegisters[13].Low = 0xdeadbeef; > + fltsave->XmmRegisters[14].High = 0xbaadf00d; > + fltsave->XmmRegisters[14].Low = 0xdeadbeef; > + fltsave->XmmRegisters[15].High = 0xbaadf00d; > + fltsave->XmmRegisters[15].Low = 0xdeadbeef; It would look better if you would use a loop or just a memset() for something like that. > + > + pRtlRestoreContext(&ctx, &rec); > + > +test_consolidate: > + done = 0; > + > + pRtlCaptureContext(&ctx); > + if (done) > + { > + ok(consolidate_dummy_called, "test_consolidate_dummy not called\n"); > + return; > + } > + done = 1; > + > + rec.ExceptionCode = STATUS_UNWIND_CONSOLIDATE; > + rec.NumberParameters = 3; > + rec.ExceptionInformation[0] = (DWORD64)test_consolidate_dummy; > + rec.ExceptionInformation[1] = (DWORD64)&ctx; > + rec.ExceptionInformation[2] = ctx.Rip; > + ctx.Rip = 0xdeadbeef; > + > + pRtlRestoreContext(&ctx, &rec); > +} > + > static RUNTIME_FUNCTION* CALLBACK dynamic_unwind_callback( DWORD64 pc, PVOID context ) > { > static const int code_offset = 1024; > @@ -2155,6 +2347,9 @@ static void test_vectored_continue_handler(void) > START_TEST(exception) > { > HMODULE hntdll = GetModuleHandleA("ntdll.dll"); > +#if defined(__x86_64__) > + HMODULE hmsvcrt = LoadLibraryA("msvcrt.dll"); > +#endif > > code_mem = VirtualAlloc(NULL, 65536, MEM_RESERVE | MEM_COMMIT, PAGE_EXECUTE_READWRITE); > if(!code_mem) { > @@ -2267,6 +2462,12 @@ START_TEST(exception) > "RtlInstallFunctionTableCallback" ); > pRtlLookupFunctionEntry = (void *)GetProcAddress( hntdll, > "RtlLookupFunctionEntry" ); > + pRtlCaptureContext = (void *)GetProcAddress( hntdll, > + "RtlCaptureContext" ); > + pRtlRestoreContext = (void *)GetProcAddress( hntdll, > + "RtlRestoreContext" ); > + p_setjmp = (void *)GetProcAddress( hmsvcrt, > + "_setjmp" ); > > test_debug_registers(); > test_outputdebugstring(1, FALSE); > @@ -2275,6 +2476,7 @@ START_TEST(exception) > test_breakpoint(1); > test_vectored_continue_handler(); > test_virtual_unwind(); > + test_restore_context(); > > if (pRtlAddFunctionTable && pRtlDeleteFunctionTable && pRtlInstallFunctionTableCallback && pRtlLookupFunctionEntry) > test_dynamic_unwind(); > -- 1.9.5 > > >

1 0

RE: TestBot job 25000 results: [PATCH] ntdll: Set Rip in for longjmp in RtlRestoreContext
by Daniel Lehman 15 Aug '16

15 Aug '16

> ntdll: > exception.c:523: Test failed: 42: Unexpected eip 0x2d0000/0x2d0001 > exception: unhandled exception 80000004 at 002D0000 > exception.c:520: Test failed: 42: Wrong exception code > c0000027/80000004 > exception.c:523: Test failed: 42: Unexpected eip 0x77baed97/0x2d0001 ... > > ntdll: > exception.c:1233: Test failed: Got exception offset 0x1b, expected 0x19 > exception.c:1242: Test failed: Got exception offset 0x19, expected 0x17 these are existing failures

1 0