February 2015 - Wine-devel - list.winehq.org

Re: [PATCH 3/5] wined3d: Remove ARB_/EXT_TEXTURE_ENV_ADD.
by Henri Verbeet 26 Feb '15

26 Feb '15

On 24 February 2015 at 13:38, Matteo Bruni <mbruni(a)codeweavers.com> wrote: > Unused. > --- But they should be used by ffp_fragment_get_caps(), since set_tex_op() uses that functionality.

3 3

Re: [3/3] dpnet: Convert Address components to use a list
by Stefan Dösinger 26 Feb '15

26 Feb '15

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi Alistair, Alexandre made me aware that I missed a blaring issue here: Am 2015-02-26 um 04:32 schrieb Alistair Leslie-Hughes: > + temp = heap_realloc(This->components, sizeof(struct > component *) * This->comp_count+1 ); X * This->comp_count+1 won't do what you want it to do. Also he wants the array to be grown by doubling it. If the expected component count is low as you say, allocate the expected count at object creation (e.g. 4), and then double the array in case you need to grow it. Also, in terms of style, you can write sizeof(*This->components) instead of sizeof(struct component *). That way you won't have to change the code if you change the type of This->components. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBAgAGBQJU7tiXAAoJEN0/YqbEcdMw6NwP/iWzcMCi48YkZ5iu52GrQDTo 86BL0M0VjFy72VDJq58pXvrK+7skm7JlI+jj/hdF0dtXg7ga6OCHqMwdyKdpUisB P93rCUVNOWvtqWltjct3UjdFM0PUD+IU9s5r3nX6sYMxwU++Kt31q5Cxq5APwxUO XplBpKeoRKtx44xgX0o84/hv6MijePDbocn+sQ3gBSyG4UdC4lt9F1Ze57r87DoW F0LkapFizu+45sRVWm6d6ANL4VL+pq3vVmjlIFRUPerVlIRRGEXnE3/5RqBegOUw MWFqRzj7LicQsy2tyan01o+OoUqGQL6ynesKj+I2LMf76i5JlSPGxoYMYlsk3vE+ JXn74dqP9iX6+Rhk7mv6q4iOQNOYsH5BraRWWLCT3vzr5DZESk+pjqn3ECKPkkZQ S88g9ismdDLkQ5pSNT/tbxeV2oRQSa4cTPp8aSNa5ttZtvwJhoHZjUGtdxAU7Go2 DcJuSbmTXFQo50Iaaw8Qlb+ZPbUVE0rjcALkKH74Bhq5vSbRkA1eu69A2TQOB9S3 x1sVVT78YhhWkl6qjESrhP0NFApNiCunUIK2U110aagOHGUyFaRix5DNN3lCO0ZT n9bvdXbZGCgg5RhxgIaqvTuGBhVZh882tpWecazipmYePUmO+4GzGwWGfxfyPymO qzUzMMRIGsukENMDUaus =3JhD -----END PGP SIGNATURE-----

1 0

Re: [1/3] dpnet: Add check for mismatched string lengths
by Stefan Dösinger 26 Feb '15

26 Feb '15

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I'm sorry, I missed one problem in my earlier reviews of this change: Am 2015-02-26 um 04:31 schrieb Alistair Leslie-Hughes: > case DPNA_DATATYPE_STRING: > + if (((strlenW((WCHAR*)lpvData)+1)*sizeof(WCHAR)) != dwDataSize) > + return DPNERR_INVALIDPARAM; > heap_free(entry->data.string); At the time when you do the check and the error return you have already allocated a new entry and added it to the list. Thus if the check fails you'll leave a zombie component attached to the address object. This also applies to the already existing DWORD size check. You can move the check in front of the loop that searches if such a component name already exists. I also recommend to to write extra tests for these cases: What happens if a new component with an invalid data size is added? Does GetComponentByName return something? What happens if a component already exists and you try to re-add it with an invalid data? Does the old data stay intact? The other two patches look good to me and seem to apply without this one. I'll inform Alexandre on IRC. The subject of patch 3 is broken though, maybe Alexandre can correct it. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBAgAGBQJU7tO3AAoJEN0/YqbEcdMw6z8P/AoWZwr4ysFdMvMBdqlZTVTH bUFYX62n8gj/IADpL1TIq8NN87Q2H2gv+V9p+6IOcvymMkDOVa/jfK+tBtZVLjmy MoLDx/+hJ2Ef8LteV2qiFO4y50btO32ibPrxY6C/jdJ5gj8dHrsjOQQSUiiAHAfQ N4Hl1j9BVblH7elBEa/Anuk+1aArnXncYU3q2wI50HRfgVShDo+3LO8jA4XzQyQM U3Gk0n5YeRapnPH0WAq9FH8o45dUFg3vtkx8M0wpzIHNGsZn91/RlIHOh7m5O9Xx S6mGU9xkhIDoWsRP+I0dIROa/QDVM7aTw8z8jpPVVwc7gS/z7k95BT/skSXw/ryH s4zOZvoS+P4vtJGgQrwnZDpEywJu7p6hdvWcUBEM85ZWvqh2Ui4em9WrW0nsmFeL VC9WcEFG4N8vzAGRK6yF2luyKrygj8KZGukZ2HCJGrjK0pAnln+BSs2V2M4cZMd6 uJus8bKJMcM2oGoB5UbpHMKRlIgSXrO0nwibUYJ02Kp3dpTUG0oHSXer+HFzImvA 2VLinduPKpA0g9XSXtCdUaXZghpQf6K4e7lwDyoikgFZs5x5G6qNYQIx2MtrJwJ2 oQ0YILD8thFi1QQrH7DLqQAGA0kAarDs0wClct5kQ3uiiT3t9a3Rklql/EYPdz3m VvgO1NqMg7yqPq0KMGKl =5ibs -----END PGP SIGNATURE-----

1 0

Re: [PATCH 1/3] services: Monitor process handles for process quit
by Alexandre Julliard 26 Feb '15

26 Feb '15

Andrew Eikum <aeikum(a)codeweavers.com> writes: > @@ -1636,13 +1638,30 @@ DWORD events_loop(void) > > do > { > - err = WaitForMultipleObjects(2, wait_handles, FALSE, timeout); > + num_handles = 2; > + > + /* monitor tracked process handles for process end */ > + EnterCriticalSection(&timeout_queue_cs); > + LIST_FOR_EACH_ENTRY(iter, &timeout_queue, struct timeout_queue_elem, entry) > + { > + if(num_handles == max_handles) > + { > + max_handles++; > + wait_handles = HeapReAlloc(GetProcessHeap(), 0, wait_handles, sizeof(HANDLE) * max_handles); > + } You don't need heap allocation, you can't wait for more than MAXIMUM_WAIT_OBJECTS anyway. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: comctl32: SysDateTimePick32 control add Alt+down hot key likewindows (try 2)
by Nikolay Sivov 26 Feb '15

26 Feb '15

On 25.02.2015 10:56, Changhui Liu wrote: > Superseded patch 109085 > http://source.winehq.org/patches/data/109085 > > Change log: > 1, Pass test on all Windows. > > 2, Add DTN_DROPDOWN and DTN_CLOSEUPnotification test. > > 3, > >>>/ + case WM_SYSKEYDOWN: > />>/ + if (wParam == VK_DOWN) > />>/ + { > />>/ + POINT pt; > />>/ + RECT rect; > />>/ + GetClientRect(hwnd, &rect); > />>/ + pt.x = rect.right - GetSystemMetrics(SM_CXVSCROLL)/2; > />>/ + pt.y = rect.top + GetSystemMetrics(SM_CYHSCROLL)/2; > />>/ + > />>/ + return DATETIME_LButtonDown (infoPtr, (SHORT)pt.x, (SHORT)pt.y); > />>/ + } > />>/ + > />> >>>If it really works this way you should just make it drop directly, >>without faking click point coordinates. > > Because most of code of the directly implement are same as DATETIME_LButtonDown, > and too many repeat code looks bad. > So faking click point coordinates can reuse code. > I'm not saying you should duplicate it, but reuse it in a clean way without hit testing - move relevant code somewhere and use it. > + if (WM_NOTIFY == message) > + { > + NMHDR* hdr = (NMHDR*)lParam; > + > + if (hdr->code == DTN_DROPDOWN) > + { > + g_hWndMonthCal = (HWND)SendMessageA(hdr->hwndFrom, DTM_GETMONTHCAL, 0, 0); > + trace("open SysMonthCal32=%p\n", g_hWndMonthCal ); > + } > + else if (hdr->code == DTN_CLOSEUP) > + { > + trace("close SysMonthCal32=%p\n", g_hWndMonthCal); > + } > + } That's not a notification test as you don't check anything. Please add a proper message sequence test, we have tons of examples for those.

2 1

Re: urlmon: Added IInternetBindInfo implementation to BindStatusCallback object.
by Marvin 25 Feb '15

25 Feb '15

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=11740 Your paranoid android. === w2000pro (32 bit url) === url.c:2683: Test failed: Could not get IInternetBindInfo interface: 80004002 url: unhandled exception c0000005 at 0044A249

1 0

Re: (try 6)[1/5] imm32: Move thread data from TLSEntry to an internal list
by Alexandre Julliard 25 Feb '15

25 Feb '15

Aric Stewart <aric(a)codeweavers.com> writes: > @@ -1512,11 +1589,15 @@ BOOL WINAPI ImmGetConversionStatus( > */ > HWND WINAPI ImmGetDefaultIMEWnd(HWND hWnd) > { > - if (IMM_GetThreadData()->hwndDefault == NULL) > - IMM_GetThreadData()->hwndDefault = CreateWindowExW( WS_EX_TOOLWINDOW, > + HWND ret; > + IMMThreadData* thread_data = IMM_GetThreadData(0); > + if (thread_data->hwndDefault == NULL) > + thread_data->hwndDefault = CreateWindowExW( WS_EX_TOOLWINDOW, > szwIME, NULL, WS_POPUP, 0, 0, 1, 1, 0, 0, 0, 0); > - TRACE("Default is %p\n",IMM_GetThreadData()->hwndDefault); > - return IMM_GetThreadData()->hwndDefault; > + ret = thread_data->hwndDefault; > + LeaveCriticalSection(&threaddata_cs); > + TRACE("Default is %p\n",ret); > + return ret; You can't create a window while holding the critical section, this will cause deadlocks. -- Alexandre Julliard julliard(a)winehq.org

2 1

Re: mshtml: Correctly handle S_OK with NULL URL return from TranslateUrl.
by Marvin 25 Feb '15

25 Feb '15

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=11724 Your paranoid android. === wvistau64 (32 bit htmldoc) === The test timed out

2 1

Re: comctl32: SysDateTimePick32 control add Alt+down hot key likewindows (try 2)
by Marvin 25 Feb '15

25 Feb '15

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=11722 Your paranoid android. === wvistau64 (32 bit datetime) === The test timed out

1 0

Re: [PATCH 3/3] scrrun: Implement HashVal for floating point keys (try2)
by Marvin 24 Feb '15

24 Feb '15

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=11712 Your paranoid android. === wvistau64 (64 bit dictionary) === dictionary.c:349: Test failed: got hash 0x000001f4 dictionary.c:374: Test failed: got hash 0x000001f4 === w2008s64 (64 bit dictionary) === dictionary.c:349: Test failed: got hash 0x000001f4 dictionary.c:374: Test failed: got hash 0x000001f4

1 0