February 2012 - Wine-devel - list.winehq.org

Re: [PATCH 1/2] ddraw/tests: Add some IDirect3DViewport interface tests (try 2)
by Henri Verbeet 02 Feb '12

02 Feb '12

On 2 February 2012 14:57, Stefan Dösinger <stefan(a)codeweavers.com> wrote: > +static ULONG get_refcount(IUnknown *test_iface) > +{ > + IUnknown_AddRef(test_iface); > + return IUnknown_Release(test_iface); > +} ... > + LONG ref; ... > + ref = get_refcount((IUnknown *)d3d); > + ok(ref == 2, "IDirect3D refcount is %d\n", ref); Well yeah, get_refcount() returns ULONG now, but you should fix the type of "ref" and the format strings as well.

2 1

Re: [PATCH 1/2] ddraw/tests: Add some IDirect3DViewport interface tests (try 2)
by Marvin 02 Feb '12

02 Feb '12

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 http://testbot.winehq.org/JobDetails.pl?Key=16784 Your paranoid android. === WNT4WSSP6 (32 bit ddraw2) === ddraw2.c:421: Test failed: Got unexpected count 3. ddraw2.c:430: Test failed: Got unexpected clip rect {14, 33, 646, 237}, expected {14, 33, 646, 486}.

1 0

Re: [PATCH] ddraw/tests: Add some IDirect3DViewport interface tests
by Henri Verbeet 02 Feb '12

02 Feb '12

On 2 February 2012 12:56, Stefan Dösinger <stefan(a)codeweavers.com> wrote: > +static LONG get_refcount(IUnknown *test_iface) > +{ > + IUnknown_AddRef(test_iface); > + return IUnknown_Release(test_iface); > +} Release is supposed to return a ULONG. > + hr = IDirect3DViewport_QueryInterface(viewport, &IID_IDirect3DViewport2, (void **)&viewport2); > + ok(SUCCEEDED(hr) || broken(hr == E_FAIL), "Failed to QI IDirect3DViewport2, hr %#x.\n", hr); I think this would make the test fail on old versions of ddraw that simply don't know about IDirect3DViewport2. > + hr = IDirect3DViewport3_QueryInterface(viewport2, &IID_IUnknown, (void **)&unknown); This looks like it was intended to be called on viewport3. > + /* FIXME: This test suggests that IDirect3D, IDirect3D2 and IDirect3D3 are linked > + * to IDirectDraw. However, they are linked to whatever interface is used to QI the > + * first IDirect3D? interface. If DDraw1 is replaced with DDraw4 here the tests break */ Makes sense. I'm not sure it really should be part of this patch though.

1 0

please test native behaviour with (un)pluggable audio devices
by Joerg-Cyril.Hoehle＠t-systems.com 02 Feb '12

02 Feb '12

Hi, please run the mmdevapi render and capture testsuites while unplugging (or deactivating) audio HW. I'd like to hear how native behaves prior to implementing error handling in Wine's drivers. One way to run the testsuites is as follows: 1. Download winetest-xyz.exe from test.winehq; 2. winetest-xyz.exe /x will extract all exes into some directory; 3. set WINETEST_DEBUG=2 4. set WINETEST_INTERACTIVE=1 (not used yet, but soon) 5. mmdevapi_test.exe render 6. mmdevapi_test.exe capture Esp., I'd like to know whether ReleaseBuffer ever signals an error or if it's always the job of GetBuffer to do so. Then I'd like to know whether primitives such as GetBufferSize fail too when the device is unplugged. Note that the testsuite may crash because it doesn't expect loosing devices mid-way. Thank you, Jörg Höhle

1 0

Re: dinput: Support INFINITE as number of effect iterations (try 2)
by Vitaliy Margolen 01 Feb '12

01 Feb '12

On 01/28/2012 05:56 PM, Marcel Hasler wrote: > According to the DirectX reference, IDirectInputEffect::Start accepts > INFINITE (aka -1) for the number of iterations. However, passing -1 to > Linux via input_event results in no effect being played at all. This > patch sets the number of iterations to the maximum signed value > allowed if INFINITE is passed. > > This fixes bug 29712. > > This patch supersedes patch 83233. It does the same thing but adds a > comment and should be easier to understand. In general looks good. Would be nice to point out that linux/input.h says that any durations should not exceed 0x7fff as a maximum value. Please don't make code "easier to read" by adding needless if / else. Your first patch was fine. Vitaliy.

2 5

Re: [PATCH] winealsa: snd_card_get_name uses strdup (Valgrind).
by Christian Costa 01 Feb '12

01 Feb '12

In that case you don't need the if (!err) condition to free the buffer otherwise you leak the string return by strdup. Christian 2012/2/1 Lauri Kenttä <lauri.kentta(a)gmail.com> > --- > dlls/winealsa.drv/mmdevdrv.c | 10 +++++++--- > 1 files changed, 7 insertions(+), 3 deletions(-) > > diff --git a/dlls/winealsa.drv/mmdevdrv.c b/dlls/winealsa.drv/mmdevdrv.c > index 0bedcc5..e434038 100644 > --- a/dlls/winealsa.drv/mmdevdrv.c > +++ b/dlls/winealsa.drv/mmdevdrv.c > @@ -363,7 +363,7 @@ static HRESULT alsa_enum_devices(EDataFlow flow, WCHAR > **ids, char **keys, > for(err = snd_card_next(&card); card != -1 && err >= 0; > err = snd_card_next(&card)){ > char cardpath[64]; > - const char *cardname; > + char *cardname; > WCHAR *cardnameW; > snd_ctl_t *ctl; > DWORD len; > @@ -376,16 +376,18 @@ static HRESULT alsa_enum_devices(EDataFlow flow, > WCHAR **ids, char **keys, > continue; > } > > - if((err = snd_card_get_name(card, (char **)&cardname)) < 0){ > + if((err = snd_card_get_name(card, &cardname)) < 0){ > WARN("Unable to get card name for ALSA device %s: %d (%s)\n", > cardpath, err, snd_strerror(err)); > /* FIXME: Should be localized */ > - cardname = "Unknown soundcard"; > + cardname = strdup("Unknown soundcard"); > } > > len = MultiByteToWideChar(CP_UNIXCP, 0, cardname, -1, NULL, 0); > cardnameW = HeapAlloc(GetProcessHeap(), 0, len * sizeof(WCHAR)); > if(!cardnameW){ > + if(!err) > + free(cardname); > snd_ctl_close(ctl); > return E_OUTOFMEMORY; > } > @@ -394,6 +396,8 @@ static HRESULT alsa_enum_devices(EDataFlow flow, WCHAR > **ids, char **keys, > alsa_get_card_devices(stream, ids, keys, num, ctl, card, > cardnameW); > > HeapFree(GetProcessHeap(), 0, cardnameW); > + if(!err) > + free(cardname); > > snd_ctl_close(ctl); > } > -- > 1.7.9 > > > >

3 2

snd_card_get_name uses strdup (Valgrind).
by Joerg-Cyril.Hoehle＠t-systems.com 01 Feb '12

01 Feb '12

Hi there's a small memory leak in winealsa, tentatively plugged by patches #83353, #83334 and #82900. It would be nice if somebody who feels like struggling with const warnings in gcc would get a fix accepted in git. The dilemma: a literal "foo" is a const char*, free() does not want a const and casts are the root of all evil in C. Thank you, Jörg Höhle

2 1

Re: [PATCH 2/3] winecoreaudio: Implement the lock-free callback design.
by Andrew Eikum 01 Feb '12

01 Feb '12

On Mon, Jan 30, 2012 at 11:37:17AM +0100, Joerg-Cyril.Hoehle(a)t-systems.com wrote: > Fixes bug #29657 > While I like the idea of this patch and how you implemented it, I wonder if it's appropriate for code freeze. The main problem in that bug is that the CA driver can deadlock, or release memory out of order. When working on this patch, were you able to identify specifically where the contention is? The idea of the InTransition state was to handle the lock release in mid-call. Can we add a check for that state in the callbacks to solve this problem less intrusively? The first patch seems fine. Andrew

4 6

Updated clang-static-analyzer results (just in time for 1.4 ;))
by Austin English 01 Feb '12

01 Feb '12

Enjoy: http://www.mediafire.com/?aiomfk1cr9bc61y Using: wine-1.4-rc1-34-g01947e0 LLVM commit 683d1e1d449e6e790ff4c0b6340a4bbf144ff919 / git-svn-id: http://llvm.org/svn/llvm-project/llvm/trunk(a)149340 91177308-0d34-0410-b5e6-96231b3b80d8 austin(a)aw21 ~ $ sha1sum scan-build-2012-01-30-1.tar.bz2 8b3b2573c3595c805431b2ab40c6958805f37497 scan-build-2012-01-30-1.tar.bz2 -- -Austin

2 2