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

Re: webservices: allocate enough memory for WS_XML_TEXT_TYPE_DOUBLE (Valgrind)
by Hans Leidekker 13 Dec '16

13 Dec '16

On Mon, 2016-12-12 at 16:20 -0600, Austin English wrote: > This avoids a stack smashing error when running > dlls/webservices/tests/writer.c under Valgrind. > > diff --git a/dlls/webservices/writer.c b/dlls/webservices/writer.c > index c1f1b0c..07b32d8 100644 > --- a/dlls/webservices/writer.c > +++ b/dlls/webservices/writer.c > @@ -1369,7 +1369,7 @@ static HRESULT text_to_utf8text( const WS_XML_TEXT *text, WS_XML_UTF8_TEXT **ret > case WS_XML_TEXT_TYPE_DOUBLE: > { > const WS_XML_DOUBLE_TEXT *double_text = (const WS_XML_DOUBLE_TEXT *)text; > - unsigned char buf[24]; /* "-1.1111111111111111E-308" */ > + unsigned char buf[48]; /* "-1.1111111111111111E-308" */ > unsigned short fpword; > ULONG len; Please add a comment with motivation for the increased buffer size.

1 0

Re: [PATCH] advapi32: Add tests for StartTrace().
by Marvin 13 Dec '16

13 Dec '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=27322 Your paranoid android. === wxppro (32 bit eventlog) === eventlog.c:1170: Test failed: Expected ERROR_BAD_LENGTH, got 87 eventlog.c:1216: Test failed: Expected success, got 161 eventlog.c:1220: Test failed: Expected ERROR_ALREADY_EXISTS, got 161 === w2003std (32 bit eventlog) === eventlog.c:1170: Test failed: Expected ERROR_BAD_LENGTH, got 87 eventlog.c:1220: Test failed: Expected ERROR_ALREADY_EXISTS, got 32

1 0

Re: [PATCH] mmdevapi: Fix pointer type cast
by Andrew Eikum 12 Dec '16

12 Dec '16

On Sat, Dec 10, 2016 at 08:54:51AM +0300, Nikolay Sivov wrote: > diff --git a/dlls/mmdevapi/devenum.c b/dlls/mmdevapi/devenum.c > index 221fb39..b189e8e 100644 > --- a/dlls/mmdevapi/devenum.c > +++ b/dlls/mmdevapi/devenum.c > @@ -613,7 +613,7 @@ static HRESULT WINAPI MMDevice_Activate(IMMDevice *iface, REFIID riid, DWORD cls > if (SUCCEEDED(hr)) > { > IPersistPropertyBag *ppb; > - hr = IUnknown_QueryInterface((IUnknown*)*ppv, &IID_IPersistPropertyBag, (void*)&ppb); > + hr = IUnknown_QueryInterface((IUnknown*)*ppv, &IID_IPersistPropertyBag, (void **)&ppb); > if (SUCCEEDED(hr)) > { > /* ::Load cannot assume the interface stays alive after the function returns, Seems like this (and other similar cleanup patches) should be deferred until after freeze, right? Andrew

3 2

Re: [2/2 v2] winex11: Use XIAllDevices to select for XI_RawMotion
by Henri Verbeet 12 Dec '16

12 Dec '16

On 18 November 2016 at 12:43, Carlos Garnacho <carlosg(a)gnome.org> wrote: > Change the strategy used to get raw motion from slave devices. Instead > of selecting for XI2 events for every slave device individually, do it > for XIAllDevices, and store the current device's relative X/Y valuators > so they can be quickly looked up in the XI_RawMotion events received. > I think this makes sense, and in my (limited) testing this correctly handles the case of attaching a new input device. I do have a few minor comments though: > +static void update_relative_valuators(XIAnyClassInfo **valuators, > + int n_valuators) This should probably just be "static void update_relative_valuators( XIAnyClassInfo **valuators, int n_valuators )" > + free( thread_data->x_rel_valuator ); > + free( thread_data->y_rel_valuator ); ... > + thread_data->x_rel_valuator = malloc( sizeof(XIValuatorClassInfo) ); > + memcpy ( thread_data->x_rel_valuator, class, sizeof(XIValuatorClassInfo) ); We generally avoid malloc()/free() in Wine code and use HeapAlloc()/HeapFree() instead. The main consideration has to do with the way address space is allocated for Win32 applications, effectively making malloc() space a bit more scarce than HeapAlloc() space. We'd probably prefer the sizeof as "sizeof(*thread_data->x_rel_valuator)", or perhaps "sizeof(*class)". I'm not sure how much we should worry about this in practice, but when multiple slaves are attached to the master, and they're not completely idle, update_relative_valuators() can end up getting called quite a bit. It doesn't seem hard to at least do the allocation only once, and only update the data in update_relative_valuators(). Perhaps it's also fine the way it is though. > - XISetMask( mask_bits, XI_ButtonPress ); This was added intentionally (see also https://bugs.winehq.org/show_bug.cgi?id=27522#c18) Perhaps it's no longer needed, but that would need to be a separate patch. > + pointer_info = pXIQueryDevice( data->display, data->xi2_core_pointer, &count ); > + update_relative_valuators( pointer_info->classes, pointer_info->num_classes ); > + pXIFreeDeviceInfo( pointer_info ); ... > + pXISelectEvents( data->display, DefaultRootWindow( data->display ), &mask, 1 ); I'm not too concerned about it, but can this race? I.e. switching slaves between when update_relative_valuators() is called above and when XISelectEvents() takes effect. Applies to querying the devices list as well. > + if (thread_data->xi2_current_slave == 0) We'd typically write that "if (!thread_data->xi2_current_slave)", although I don't think anyone would be overly concerned about it. > + for (i = 0; i <= max ( x_rel->number, y_rel->number ); i++) > + { > +»······if (!XIMaskIsSet( event->valuators.mask, i )) continue; > +»······val = *values++; > +»······if (i == x_rel->number) > +»······{ > + input.u.mi.dx = dx = val; > +»······ if (x_rel->min < x_rel->max) > + input.u.mi.dx = val * (virtual_rect.right - virtual_rect.left) > + / (x_rel->max - x_rel->min); > +»······} > +»······if (i == y_rel->number) > +»······{ > +»······ input.u.mi.dy = dy = val; > +»······ if (y_rel->min < y_rel->max) You're using tabs for indentation here.

2 1

Re: [PATCH] shlwapi: Avoid 'else if' for IStream_fnQueryInterface.
by Nikolay Sivov 12 Dec '16

12 Dec '16

On Mon, 12 Dec 2016 13:24:38 +0200 Andrey Gusev <andrey.goosev(a)gmail.com> wrote: > Signed-off-by: Andrey Gusev <andrey.goosev(a)gmail.com> > --- > dlls/shlwapi/regstream.c | 4 +--- > 1 file changed, 1 insertion(+), 3 deletions(-) > > diff --git a/dlls/shlwapi/regstream.c b/dlls/shlwapi/regstream.c > index 1b0ca4e..d886c0c 100644 > --- a/dlls/shlwapi/regstream.c > +++ b/dlls/shlwapi/regstream.c > @@ -67,9 +67,7 @@ static HRESULT WINAPI IStream_fnQueryInterface(IStream *iface, REFIID riid, LPVO > > *ppvObj = NULL; > > - if(IsEqualIID(riid, &IID_IUnknown)) /*IUnknown*/ > - *ppvObj = This; > - else if(IsEqualIID(riid, &IID_IStream)) /*IStream*/ > + if(IsEqualIID(riid, &IID_IUnknown) || IsEqualIID(riid, &IID_IStream)) > *ppvObj = This; > > if(*ppvObj) Please return interface pointer, to IStream_iface. Also TRACE() for this method needs simplification, but it could be a separate change.

1 0

Re: [PATCH resend 13/18] wined3d: Use depth/stencil formats for typeless resources only when WINED3DUSAGE_DEPTHSTENCIL is set.
by Henri Verbeet 12 Dec '16

12 Dec '16

On 9 December 2016 at 11:30, Józef Kucia <jkucia(a)codeweavers.com> wrote: > const struct wined3d_format *wined3d_get_format(const struct wined3d_gl_info *gl_info, > - enum wined3d_format_id format_id) > + enum wined3d_format_id format_id, unsigned int resource_usage) > { > + const struct wined3d_format *format; > int idx = get_format_idx(format_id); > + unsigned int i; > > if (idx == -1) > { > - FIXME("Can't find format %s (%#x) in the format lookup table\n", > + FIXME("Can't find format %s (%#x) in the format lookup table.\n", > + debug_d3dformat(format_id), format_id); > + return &gl_info->formats[get_format_idx(WINED3DFMT_UNKNOWN)]; > + } > + > + format = &gl_info->formats[idx]; > + > + if (resource_usage & WINED3DUSAGE_DEPTHSTENCIL && wined3d_format_is_typeless(format)) > + { > + for (i = 0; i < ARRAY_SIZE(typeless_depth_stencil_formats); ++i) > + { > + if (typeless_depth_stencil_formats[i].typeless_id == format_id) > + return &gl_info->formats[WINED3D_FORMAT_COUNT + i]; > + } > + > + FIXME("Cannot find depth/stencil typeless format %s (%#x).\n", > debug_d3dformat(format_id), format_id); > - /* Get the caller a valid pointer */ > - idx = get_format_idx(WINED3DFMT_UNKNOWN); > + return &gl_info->formats[get_format_idx(WINED3DFMT_UNKNOWN)]; > } > > - return &gl_info->formats[idx]; > + return format; > } > I do have some concerns about whether returning different wined3d_format structures for different usages/bind flags is better than having different GL internal formats for different usages/bind flags within the same wined3d_format structure, but lets give it a try.

1 0

Re: [v2] preloader: Allow a debugger to find ld.so's rendezvous-struct
by Alexandre Julliard 12 Dec '16

12 Dec '16

Keno Fischer <keno(a)juliacomputing.com> writes: > @@ -123,6 +123,18 @@ static struct wine_preload_info preload_info[] = > { 0, 0 } /* end of list */ > }; > > +/* A fake .dynamic section through which we can inform the debugger of the > + address of the chain-loaded dynamic linker's rendezvous struct */ > +__asm__( > +".section .dynamic,\"aw\"\n" > +".global _DYNAMIC\n" > +"_DYNAMIC: .quad 21\n" // DT_DEBUG > +".quad 0\n" > +".quad 0\n" > +".quad 0\n" > +".previous\n" > +); This is causing a warning here: gcc -m32 -c -o preloader.o preloader.c -I. -I../include -D__WINESRC__ -fno-builtin -Wall -pipe \ -fno-strict-aliasing -Wdeclaration-after-statement -Wempty-body -Wignored-qualifiers \ -Wstrict-prototypes -Wtype-limits -Wunused-but-set-parameter -Wvla -Wwrite-strings -Wpointer-arith \ -Wlogical-op -gdwarf-2 -gstrict-dwarf -fno-omit-frame-pointer -Werror -g -O2 -fno-diagnostics-show-caret {standard input}: Assembler messages: {standard input}:5: Warning: setting incorrect section attributes for .dynamic -- Alexandre Julliard julliard(a)winehq.org

2 6

Re: [PATCH] po: Update Simplified Chinese translation.
by Jactry Zeng 11 Dec '16

11 Dec '16

This looks good to me. Thank you, Mingcong! 2016-12-11 0:12 GMT+08:00 Mingcong Bai <jeffbaichina(a)gmail.com>: > Signed-off-by: Mingcong Bai <jeffbai(a)aosc.xyz> > --- > po/zh_CN.po | 588 ++++++++++++++++++++++++++++++ > +++--------------------------- > 1 file changed, 324 insertions(+), 264 deletions(-) > > > > -- Regards, Jactry Zeng

1 0

Source.winehq.org down?
by Filip Frackiewicz 10 Dec '16

10 Dec '16

Sorry for the bother but is source.winehq.org down for anyone? Filip

3 2

Re: [PATCH] ws2_32: Add support to disable networking with envvar WINENONET=1
by Stefan Dösinger 10 Dec '16

10 Dec '16

Am 2016-12-07 um 06:27 schrieb Bruno Jesus: > @@ -2433,6 +2438,12 @@ static int WS2_send( int fd, struct ws2_async *wsa, int flags ) > union generic_unix_sockaddr unix_addr; > int n, ret; > > + if (network_disabled) > + { > + errno = EHOSTUNREACH; > + return -1; > + } > + Shouldn't it be possible to achieve a similar effect with a LD_PRELOAD library?

2 2