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

Re: [PATCH 1/5] d3drm: Add stubs for IDirect3DRMFrame interface. (v3)
by Stefan Dösinger 15 Jun '16

15 Jun '16

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Am 2016-06-15 um 10:35 schrieb Aaryaman Vasishta: > +static HRESULT WINAPI d3drm_frame1_GetVisuals(IDirect3DRMFrame *iface, IDirect3DRMVisualArray **visuals) > +{ > + struct d3drm_frame *frame = impl_from_IDirect3DRMFrame(iface); > + struct d3drm_visual_array *array; > + > + TRACE("iface %p, visuals %p.\n", iface, visuals); > + > + if (!visuals) > + return D3DRMERR_BADVALUE; > + > + if (!(array = d3drm_visual_array_create(frame->nb_visuals, frame->visuals))) > + return E_OUTOFMEMORY; > + > + *visuals = &array->IDirect3DRMVisualArray_iface; > + > + return D3DRM_OK; > +} This is a reimplementation instead of a thunk. Intentional? The same applies to d3drm_frame1_GetTransform. > +static HRESULT WINAPI d3drm_frame1_GetSceneBackgroundDepth(IDirect3DRMFrame *iface, > + IDirectDrawSurface **surface) The indentation here is different. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBCAAGBQJXYcT1AAoJEN0/YqbEcdMwUHcQAIAfKv2BAfasnwRACZsGBcK9 VAftr2zuIgmlMKtQaplsZuv2ZqcyUnydOgzuXQusFfBldeKeD1ukycLx/sONpuZI ggt7AbXcslw9T0wlprTnfl+1Pna/CSbRQWhVAjeB9dCCWblBslipsj9SLr3hTOwH ppQYvk+BtoXb7A+tvVI2ch/Q8HhwWh/Q9SvVg6tUV7gQLrPG0sj4NEjG50I7VN0c PjLgZHlT7JzVDxlGoefhTPXZdL+1muSIIuxVXho+oNx6qFqiyC0+7xxlJEVR2Eyo VBzdczdYd0UcLRb0GOqyJpRcGTpYE06zahKeDyODIXeorFB0NlYFqSTD2H32RVNU M2a+xKgFLqm4zu+cabfRIvFk9wGkIRykwfm8KUY4LyZCZmlQZ21nPCAIsuolRqV6 DrmQAParbMg/fAb6L9/uB0rb/CBPGIhm3mtv00TewBG8VG16uV4vLbvtoiIJJtD4 Z6iY6x948K+jjVVkhGDRfdEJn+nQRDTKIkgsRKic9S5x/nOEyfreVutjq4vULnGk l5hRAzyzOxjP/JTMHhvYtHYbLtqjgXPi3pwhjrfGMWLKAmeRPk3A2FdsUl9A1I3I INR8LmGAo3P7a1GZ21loc5A2iftn29vcAB0mDzcotczOEsinfx7Sc4oF5fIUDZla miBufY1gPaDENTTiRcz5 =hvk0 -----END PGP SIGNATURE-----

1 0

Re: ddraw/tests: A spelling fix for a d3d ok() message.
by Francois Gouget 15 Jun '16

15 Jun '16

Anything wrong with this patch? On Sun, 29 May 2016, Francois Gouget wrote: > Signed-off-by: Francois Gouget <fgouget(a)free.fr> > --- > dlls/ddraw/tests/d3d.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/dlls/ddraw/tests/d3d.c b/dlls/ddraw/tests/d3d.c > index d877020..0e9a43b 100644 > --- a/dlls/ddraw/tests/d3d.c > +++ b/dlls/ddraw/tests/d3d.c > @@ -1155,7 +1155,7 @@ static void Direct3D1Test(void) > /* Interface consistency check. */ > hr = IDirect3DDevice_GetDirect3D(Direct3DDevice1, &Direct3D_alt); > ok(hr == D3D_OK, "IDirect3DDevice_GetDirect3D failed: %08x\n", hr); > - ok(Direct3D_alt == Direct3D1, "Direct3D1 struct pointer missmatch: %p != %p\n", Direct3D_alt, Direct3D1); > + ok(Direct3D_alt == Direct3D1, "Direct3D1 struct pointer mismatch: %p != %p\n", Direct3D_alt, Direct3D1); > IDirect3D_Release(Direct3D_alt); > > memset(&desc, 0, sizeof(desc)); > -- > 2.8.1 > > > > -- Francois Gouget <fgouget(a)free.fr> http://fgouget.free.fr/ Indifference will certainly be the downfall of mankind, but who cares?

3 4

Re: [PATCH 5/5] d3dx9: Turn an ERR into a WARN.
by Matteo Bruni 15 Jun '16

15 Jun '16

2016-06-15 9:51 GMT+02:00 Paul Gofman <gofmanp(a)gmail.com>: > Hello Matteo, > > I am afraid that if application passes effect data broken an > arbitrary way then probably almost every ERR and FIXME should be changed > to WARN (at least in my code, but probably in the rest of effect > framework either). E. g., we have FIXME on not yet supported relative > addressing, but it can also happen if the application passes broken > effect code. Or do you see any difference for this case with parameter > name? I do not yet, as actually preshader and effect are compiled > together, and it is no way to get such a situation (at least if there is > some feature in fxc compiler which I am not aware of) other than > modifying effect bytecode. > > Regards, > Paul. The general idea is that FIXME refers to something that isn't (completely / correctly) implemented in Wine, so e.g. FIXME is the right choice for relative addressing. ERR is the correct choice for "internal errors" in Wine (i.e. something that shouldn't happen unless there is a bug in Wine - the ERR("Unexpected register set %u.\n", ...) is a good example of that) and WARN should be used when the application can trigger the problematic condition by misusing the API or similar. Specifically WRT this patch I just happened to notice that ERR while looking through the code and it seems to me that WARN fits better. It's not a big deal either way.

1 0

Re: [PATCH] wined3d: Implement icb dcl support in GLSL.
by Józef Kucia 15 Jun '16

15 Jun '16

This implementation has some issues, e.g. a generated GLSL shader will produce a compilation error when the immediate constant buffer contains an unsigned integer which represents NaN. Morever, from testing a similar implementation, it seems that some drivers don't handle shaders with huge arrays very well and such shaders perform poorly. My current impression is that we should implement immediate constant buffers using a uniform variable. Also, a test would be nice. On Mon, Jun 13, 2016 at 6:41 PM, Guillaume Charifi <guillaume.charifi(a)sfr.fr> wrote: > Signed-off-by: Guillaume Charifi <guillaume.charifi(a)sfr.fr> > --- > dlls/wined3d/glsl_shader.c | 15 ++++++++++++++- > 1 file changed, 14 insertions(+), 1 deletion(-) > > diff --git a/dlls/wined3d/glsl_shader.c b/dlls/wined3d/glsl_shader.c > index 55affc1..def224c 100644 > --- a/dlls/wined3d/glsl_shader.c > +++ b/dlls/wined3d/glsl_shader.c > @@ -1886,6 +1886,19 @@ static void shader_generate_glsl_declarations(const struct wined3d_context *cont > if (shader->limits->constant_bool > 0 && reg_maps->boolean_constants) > shader_addline(buffer, "uniform bool %s_b[%u];\n", prefix, shader->limits->constant_bool); > > + if (reg_maps->icb) > + { > + shader_addline(buffer, "const vec4 %s_icb[%u] = vec4[%u](\n", > + prefix, reg_maps->icb->element_count / 4, reg_maps->icb->element_count / 4); > + for (i = 0; i < reg_maps->icb->element_count / 4; i++) { > + shader_glsl_append_imm_vec4(buffer, (float *)&reg_maps->icb->data[4 * i]); > + if (i != reg_maps->icb->element_count / 4 - 1) > + shader_addline(buffer, ","); > + shader_addline(buffer, "\n"); > + } > + shader_addline(buffer, ");\n"); > + } I think it would be better to store the number of vectors instead of the element count in "reg_maps->icb". Not that it matters much, but this block of code would certainly look nicer. > + > for (i = 0; i < WINED3D_MAX_CBS; ++i) > { > if (reg_maps->cb_sizes[i]) > @@ -8561,7 +8574,7 @@ static const SHADER_HANDLER shader_glsl_instruction_handler_table[WINED3DSIH_TAB > /* WINED3DSIH_DCL_GLOBAL_FLAGS */ shader_glsl_nop, > /* WINED3DSIH_DCL_HS_FORK_PHASE_INSTANCE_COUNT */ NULL, > /* WINED3DSIH_DCL_HS_MAX_TESSFACTOR */ NULL, > - /* WINED3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER */ NULL, > + /* WINED3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER */ shader_glsl_nop, > /* WINED3DSIH_DCL_INPUT */ shader_glsl_nop, > /* WINED3DSIH_DCL_INPUT_CONTROL_POINT_COUNT */ NULL, > /* WINED3DSIH_DCL_INPUT_PRIMITIVE */ shader_glsl_nop, > -- > 2.7.4 > > >

2 5

Re: [PATCH] wined3d: Properly handle immediate NaN and Inf in GLSL backend.
by Józef Kucia 15 Jun '16

15 Jun '16

In my opinion, we do not want to solve the issue with immediate constant buffers this way. However, this patch might still be useful. Some games (Magicka 2) use shaders which contain NaN values in immediate constants. This patch represents a one way to solve the issue with NaN values in immediate constants. If we decide that we want to apply this change we should be more careful though. We probably want to use this for SM4+ shaders only, or use it only when GL_ARB_shader_bit_encoding is available. On a side note, "uintBitsToFloat" cannot be used as a constant expression (at least in GLSL 1.30), so to make this work with the immediate constant buffers patch you would have to drop "const". On Mon, Jun 13, 2016 at 9:30 PM, Guillaume Charifi <guillaume.charifi(a)sfr.fr> wrote: > Signed-off-by: Guillaume Charifi <guillaume.charifi(a)sfr.fr> > --- > dlls/wined3d/glsl_shader.c | 36 +++++++++++++++++++++++++----------- > 1 file changed, 25 insertions(+), 11 deletions(-) > > diff --git a/dlls/wined3d/glsl_shader.c b/dlls/wined3d/glsl_shader.c > index def224c..fde4c5e 100644 > --- a/dlls/wined3d/glsl_shader.c > +++ b/dlls/wined3d/glsl_shader.c > @@ -34,6 +34,8 @@ > > #include <limits.h> > #include <stdio.h> > +#include <stdint.h> > +#include <inttypes.h> We avoid C99 in Wine. unsigned int and %#xu should be good enough. > #ifdef HAVE_FLOAT_H > # include <float.h> > #endif > @@ -321,14 +323,26 @@ static const char *shader_glsl_get_version(const struct wined3d_gl_info *gl_info > return "#version 120"; > } > > +/* Allow raw NaN and Inf in GLSL code. */ > +void wined3d_glsl_ftoa(float value, char *s) The function should be static. > +{ > + if (!isfinite(value)) > + { > + /* We need a buffer of at least 28 characters. */ > + snprintf(s, 28, "uintBitsToFloat(%"PRIu32"u)", *(uint32_t *)&value); It seems that 29 bytes are needed at least (including the null character). > + } > + else > + wined3d_ftoa(value, s); > +} In wined3d, generally, if one block in if-else statement needs curly braces we use curly braces for all blocks in this statement. else { wined3d_ftoa(value, s); } > + > static void shader_glsl_append_imm_vec4(struct wined3d_string_buffer *buffer, const float *values) > { > - char str[4][17]; > + char str[4][28]; > > - wined3d_ftoa(values[0], str[0]); > - wined3d_ftoa(values[1], str[1]); > - wined3d_ftoa(values[2], str[2]); > - wined3d_ftoa(values[3], str[3]); > + wined3d_glsl_ftoa(values[0], str[0]); > + wined3d_glsl_ftoa(values[1], str[1]); > + wined3d_glsl_ftoa(values[2], str[2]); > + wined3d_glsl_ftoa(values[3], str[3]); > shader_addline(buffer, "vec4(%s, %s, %s, %s)", str[0], str[1], str[2], str[3]); > } > > @@ -2299,7 +2313,7 @@ static void shader_glsl_get_register_name(const struct wined3d_shader_register * > const struct wined3d_gl_info *gl_info = ins->ctx->gl_info; > const char *prefix = shader_glsl_get_prefix(version->type); > struct glsl_src_param rel_param0, rel_param1; > - char imm_str[4][17]; > + char imm_str[4][28]; > > if (reg->idx[0].offset != ~0U && reg->idx[0].rel_addr) > shader_glsl_add_src_param(ins, reg->idx[0].rel_addr, WINED3DSP_WRITEMASK_0, &rel_param0); > @@ -2513,7 +2527,7 @@ static void shader_glsl_get_register_name(const struct wined3d_shader_register * > switch (reg->data_type) > { > case WINED3D_DATA_FLOAT: > - wined3d_ftoa(*(const float *)reg->immconst_data, register_name); > + wined3d_glsl_ftoa(*(const float *)reg->immconst_data, register_name); > break; > case WINED3D_DATA_INT: > sprintf(register_name, "%#x", reg->immconst_data[0]); > @@ -2533,10 +2547,10 @@ static void shader_glsl_get_register_name(const struct wined3d_shader_register * > switch (reg->data_type) > { > case WINED3D_DATA_FLOAT: > - wined3d_ftoa(*(const float *)&reg->immconst_data[0], imm_str[0]); > - wined3d_ftoa(*(const float *)&reg->immconst_data[1], imm_str[1]); > - wined3d_ftoa(*(const float *)&reg->immconst_data[2], imm_str[2]); > - wined3d_ftoa(*(const float *)&reg->immconst_data[3], imm_str[3]); > + wined3d_glsl_ftoa(*(const float *)&reg->immconst_data[0], imm_str[0]); > + wined3d_glsl_ftoa(*(const float *)&reg->immconst_data[1], imm_str[1]); > + wined3d_glsl_ftoa(*(const float *)&reg->immconst_data[2], imm_str[2]); > + wined3d_glsl_ftoa(*(const float *)&reg->immconst_data[3], imm_str[3]); > sprintf(register_name, "vec4(%s, %s, %s, %s)", > imm_str[0], imm_str[1], imm_str[2], imm_str[3]); It looks like shader_glsl_append_imm_vec4() could be called there instead.

4 8

Re: [PATCH] wined3d: Correct memory size for Geforece 9600 GT (resend)
by Henri Verbeet 15 Jun '16

15 Jun '16

On 15 June 2016 at 09:48, Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> wrote: > Signed-off-by: Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> > --- > dlls/wined3d/directx.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > This isn't necessarily wrong, but why is this needed/more correct?

2 2

Fwd: [v4 2/4] winejoystick.drv/joystick_osx.c: fixes Slider mapping
by DavidL 15 Jun '16

15 Jun '16

Sorry, keep hitting reply instead of reply-all to make sure the list-serve has a copy. ---------- Forwarded message ---------- From: DavidL <david.dljunk(a)gmail.com> Date: Tue, Jun 14, 2016 at 12:05 AM Subject: Re: [v4 2/4] winejoystick.drv/joystick_osx.c: fixes Slider mapping To: Ken Thomases <ken(a)codeweavers.com> As far as I can tell, both. It both arbitrarily picks which is first for any given run (so yes I can definitely confirm that bug report's description) and the game seems to read input from both inputs regardless of which is first - although which is first does still affect the game. I created virtual joystick in Controllermate which I mapped my physical joystick to but mapping X to Y and Y to X, so that when I pulled back on the yoke it would give both an X reading and a Y reading at the same time if it were reading input from both the virtual and physical joystick. In game, that's what happened, even when the trace logs said it was only recording information from one joystick (and in one case, Red Baron, it was explicitly reading from both in the trace log). Which joystick was found first would affect behavior somewhat (hard to describe, basically how out of whack the motions became!), but regardless I could get it sometimes explicitly, sometimes implicitly to read multiple inputs. Maybe this is an anomaly with virtual joysticks. Not sure, shouldn't be. On Mon, Jun 13, 2016 at 11:41 PM, Ken Thomases <ken(a)codeweavers.com> wrote: > On Jun 14, 2016, at 1:23 AM, DavidL <david.dljunk(a)gmail.com> wrote: > > > On Mon, Jun 13, 2016 at 2:46 PM, Ken Thomases <ken(a)codeweavers.com> wrote: > >> Is there reason to expect that these elements (sliders, dials, wheels) >> will be listed after the "proper" axes (x, y, z, etc.) in the children >> array? Since the proper axes are specific and these elements should only >> be mapped to unused axes, I think they should be enumerated in a separate >> pass after everything else has had a shot. Note that >> collect_joystick_elements() is recursive, so it's not right to do it in a >> separate loop within collect_joystick_elements(). Probably, you'll want to >> call it twice from open_joystick(), with a flag to indicate which mode it >> should operate in. > > > This where the Microsoft source documentation get *really* fuzzy. There > are all sorts of possible mappings in the rare case where there is both a > HID/Dinput Z-axis and a slider/dial/wheel. Microsoft gave 4-5 different > variations of how it could go and in some of these mappings, the slider > would take precedence over the reported HID Z-axis when mapping to the > Winmm Z-axis such that slider -> WinmmZ is correct and the HID Z-axis is > then ignored. So I decided to follow the Linux Winmm version and simply > have whichever element is listed by the joystick first, wins. If it is the > axis, it's the axis, if it is the slider, it's the slider. This actually is > one of the ways Microsoft says to decide proper HID mappings to Winmm (and > it seems like that for the elements within a joystick, the elements are > always reported in the same order such that a user shouldn't experience the > axes mappings changing each time they start up Wine/reconnect a device*). > There is no 1 correct mapping for all cases. I guess ideally we would have > a way of allowing the user to remap the joystick axes in the registry > (maybe that would work even now? I don't know - I've seen some emails on > that already in the wine-devel list serve but I don't know what the status > of that kind of capability is) but even Microsoft states on of their pages > that this is a pretty rare corner case. > > > OK, fine by me. > > > *Note: in my testing this is *not* the same for OS X (I guess macOS now) > reporting back devices - which human interface device is reported first > seems to be random in my testing when I had both a virtual and physical > joystick setup. Historically Winmm games would only take input from the > first reported device - under Mac-Wine, I've noticed the games seem to be > taking input from both my physical and virtual joysticks (sometimes > regardless of what the game was saying it was doing!). This has both good > and bad potential and is not necessarily behavior we should spend effort > trying to get rid of. Not sure how the Linux version behaves in this > situation in terms of joystick order and input from multiple sources. > > > It was taking input from both joysticks in the same run? At the same > time? If so, can you figure out how that's happening? It really > shouldn't, as near as I can see. > > Or was it just arbitrarily picking which was first and therefore active > for any given run? If this, then it's < > https://bugs.winehq.org/show_bug.cgi?id=38997>, more or less. > > -Ken > >

2 4

Re: [v2 02/10] msvcirt: Implement ostream constructors/destructors
by Piotr Caban 15 Jun '16

15 Jun '16

On 06/14/16 19:01, Iván Matellanes wrote: > On 14/06/16 16:20, Piotr Caban wrote: >> On 06/14/16 16:40, Iván Matellanes wrote: >>> /* ??_Dostream@@QAEXXZ */ >>> @@ -2287,23 +2341,48 @@ ostream* __thiscall ostream_assign(ostream >>> *this, const ostream *rhs) >>> DEFINE_THISCALL_WRAPPER(ostream_vbase_dtor, 4) >>> void __thiscall ostream_vbase_dtor(ostream *this) >>> { >>> - FIXME("(%p) stub\n", this); >>> + ios *base = ostream_to_ios(this); >>> + >>> + TRACE("(%p)\n", this); >>> + >>> + ostream_dtor(base); >>> + ios_dtor(base); >>> } >> The base should be computed using ostream_get_ios while calling ios_dtor: >> TRACE("(%p)\n", this); >> ostream_dtor(ostream_to_ios(this)); >> ios_dtor(ostream_get_ios(this)); > Do you mind explaining why is that required? I can't see the reason ostream_vbase_dtor can be used to destroy object that has virtual base in any position pointer by vbtable. ostream_get_ios computes it using object's vbtable. You can easily see it by overwriting ostream->vbtable with garbage, the application will crash then on Windows. On the other hand ostream_dtor is a virtual function that assumes "this" is at fixed offset from "base" that is passed to it. It's done this way because it has no access to vbtable. It's why you need to use ostream_to_ios in this case. There's no real need to call ostream_dtor here because it's an empty function (I don't know if it should be called here or if ostream_vbase_dtor should only destroy the vbase). Thanks, Piotr

2 2

Re: [PATCH] riched20: Remove duplicated condition and duplicated test
by Huw Davies 15 Jun '16

15 Jun '16

On Tue, Jun 14, 2016 at 08:56:41PM +0300, Nikolay Sivov wrote: > diff --git a/dlls/riched20/richole.c b/dlls/riched20/richole.c > index 1013e67..a2f23c9 100644 > --- a/dlls/riched20/richole.c > +++ b/dlls/riched20/richole.c > @@ -1976,7 +1976,7 @@ static HRESULT range_Collapse(LONG bStart, LONG *start, LONG *end) > if (*end == *start) > return S_FALSE; > > - if (bStart == tomEnd || bStart == tomFalse) > + if (bStart == tomEnd) > *start = *end; > else > *end = *start; > > diff --git a/dlls/riched20/tests/richole.c b/dlls/riched20/tests/richole.c > index 6da3ce3..33043fd 100644 > --- a/dlls/riched20/tests/richole.c > +++ b/dlls/riched20/tests/richole.c > @@ -1361,18 +1361,6 @@ static void test_ITextRange_Collapse(void) > first = 4, lim = 8; > hres = ITextDocument_Range(txtDoc, first, lim, &txtRge); > ok(hres == S_OK, "got 0x%08x\n", hres); > - hres = ITextRange_Collapse(txtRge, tomTrue); > - ok(hres == S_OK, "ITextRange_Collapse\n"); > - hres = ITextRange_GetStart(txtRge, &start); > - ok(hres == S_OK, "got 0x%08x\n", hres); > - ok(start == 4, "got wrong start value: %d\n", start); > - hres = ITextRange_GetEnd(txtRge, &end); > - ok(hres == S_OK, "got 0x%08x\n", hres); > - ok(end == 4, "got wrong end value: %d\n", end); > - ITextRange_Release(txtRge); > - > - hres = ITextDocument_Range(txtDoc, first, lim, &txtRge); > - ok(hres == S_OK, "got 0x%08x\n", hres); > hres = ITextRange_Collapse(txtRge, tomStart); > ok(hres == S_OK, "ITextRange_Collapse\n"); > hres = ITextRange_GetStart(txtRge, &start); > @@ -1385,18 +1373,6 @@ static void test_ITextRange_Collapse(void) > > hres = ITextDocument_Range(txtDoc, first, lim, &txtRge); > ok(hres == S_OK, "got 0x%08x\n", hres); > - hres = ITextRange_Collapse(txtRge, tomFalse); > - ok(hres == S_OK, "ITextRange_Collapse\n"); > - hres = ITextRange_GetStart(txtRge, &start); > - ok(hres == S_OK, "got 0x%08x\n", hres); > - ok(start == 8, "got wrong start value: %d\n", start); > - hres = ITextRange_GetEnd(txtRge, &end); > - ok(hres == S_OK, "got 0x%08x\n", hres); > - ok(end == 8, "got wrong end value: %d\n", end); > - ITextRange_Release(txtRge); > - > - hres = ITextDocument_Range(txtDoc, first, lim, &txtRge); > - ok(hres == S_OK, "got 0x%08x\n", hres); > hres = ITextRange_Collapse(txtRge, tomEnd); > ok(hres == S_OK, "ITextRange_Collapse\n"); > hres = ITextRange_GetStart(txtRge, &start); I don't think we should remove the extra tests, especially as tomTrue != tomStart and it's documentented to work with either. Leaving the tomFalse tests in is fine too. Huw.

2 1

Re: [PATCH] kernel32: implement GetLargePageMinimum
by Sebastian Lackner 15 Jun '16

15 Jun '16

On 14.06.2016 09:09, Marcus Meissner wrote: > From: Austin English <austinenglish(a)gmail.com> > > https://bugs.winehq.org/show_bug.cgi?id=18745 > > Report 2MB minimum page size on x86 32bit and 64bit on Linux. > > This assumes that all x86 processors and the 64/32bit emulation mode > have hugetable support and it is always enabled. > > Signed-off-by: Marcus Meissner <marcus(a)jet.franken.de> > Signed-off-by: Austin English <austinenglish(a)gmail.com> As pointed out on IRC, the order of the sign-offs is wrong here. Your own signoff should always be the last because you have to take responsibility when you send a patch. > --- > dlls/kernel32/cpu.c | 15 +++++++++++++++ > dlls/kernel32/kernel32.spec | 2 +- > dlls/kernel32/tests/process.c | 16 ++++++++++++++++ > 3 files changed, 32 insertions(+), 1 deletion(-) > > diff --git a/dlls/kernel32/cpu.c b/dlls/kernel32/cpu.c > index f48fcf0..f8b2c3d 100644 > --- a/dlls/kernel32/cpu.c > +++ b/dlls/kernel32/cpu.c > @@ -291,3 +291,18 @@ err: > } > return TRUE; > } > + > +/*********************************************************************** > + * GetLargePageMinimum (KERNEL32.@) > + */ > +SIZE_T WINAPI GetLargePageMinimum(void) > +{ > +#ifdef linux I do not see any need for that #ifdef linux. Its more architecture specific than linux specific. Dmitry already mentioned the same concern on the mailing list. > +#if defined(__i386___) || defined(__x86_64__) > + /* Linux x86 hugepages are minimum 2 MB */ > + return 2*1024*1024; > +#endif > +#endif > + FIXME("stub: not implemented on your platform/architecture.\n"); > + return 0; > +} > diff --git a/dlls/kernel32/kernel32.spec b/dlls/kernel32/kernel32.spec > index ec7155b6e..0fc4dcb 100644 > --- a/dlls/kernel32/kernel32.spec > +++ b/dlls/kernel32/kernel32.spec > @@ -696,7 +696,7 @@ > @ stdcall GetHandleInformation(long ptr) > @ stub -i386 GetLSCallbackTarget > @ stub -i386 GetLSCallbackTemplate > -# @ stub GetLargePageMinimum > +@ stdcall GetLargePageMinimum() > @ stdcall GetLargestConsoleWindowSize(long) > @ stdcall GetLastError() > @ stub GetLinguistLangSize > diff --git a/dlls/kernel32/tests/process.c b/dlls/kernel32/tests/process.c > index 61dcdc2..e9acfd3 100644 > --- a/dlls/kernel32/tests/process.c > +++ b/dlls/kernel32/tests/process.c > @@ -88,6 +88,7 @@ static BOOL (WINAPI *pProcess32Next)(HANDLE, PROCESSENTRY32*); > static BOOL (WINAPI *pThread32First)(HANDLE, THREADENTRY32*); > static BOOL (WINAPI *pThread32Next)(HANDLE, THREADENTRY32*); > static BOOL (WINAPI *pGetLogicalProcessorInformationEx)(LOGICAL_PROCESSOR_RELATIONSHIP,SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*,DWORD*); > +static SIZE_T (WINAPI *pGetLargePageMinimum)(void); > > /* ############################### */ > static char base[MAX_PATH]; > @@ -252,6 +253,7 @@ static BOOL init(void) > pThread32First = (void *)GetProcAddress(hkernel32, "Thread32First"); > pThread32Next = (void *)GetProcAddress(hkernel32, "Thread32Next"); > pGetLogicalProcessorInformationEx = (void *)GetProcAddress(hkernel32, "GetLogicalProcessorInformationEx"); > + pGetLargePageMinimum = (void *)GetProcAddress(hkernel32, "GetLargePageMinimum"); > > return TRUE; > } > @@ -3138,6 +3140,19 @@ static void test_GetLogicalProcessorInformationEx(void) > HeapFree(GetProcessHeap(), 0, info); > } > > +static void test_largepages(void) > +{ > + SIZE_T size; > + > + if (!pGetLargePageMinimum) { > + skip("No GetLargePageMinimum support.\n"); > + return; > + } > + size = pGetLargePageMinimum(); > + > + ok((size == 0) || (size == 2*1024*1024) || (size == 4*1024*1024), "GetLargePageMinimum reports %ld size\n", size); > +} > + > START_TEST(process) > { > HANDLE job; > @@ -3210,6 +3225,7 @@ START_TEST(process) > test_GetNumaProcessorNode(); > test_session_info(); > test_GetLogicalProcessorInformationEx(); > + test_largepages(); > > /* things that can be tested: > * lookup: check the way program to be executed is searched >

3 3