May 2005 - Wine-devel - list.winehq.org

Re: Added the callback item processing for LISTVIEW_EndEditLabelT function.
by Dimitrie Paun 19 May '05

19 May '05

> Added the callback item processing for LISTVIEW_EndEditLabelT function. I think this would be a bit more explicit and correct (is_textW() returns false if pszText == NULL) if the test is the other way around: + if (lpItem && lpItem->hdr.pszText == LPSTR_TEXTCALLBACKW) + { + LISTVIEW_InvalidateItem(infoPtr, infoPtr->nEditLabelItem); + } + else + { + ZeroMemory(&dispInfo, sizeof(dispInfo)); + dispInfo.item.mask = LVIF_TEXT; + dispInfo.item.iItem = infoPtr->nEditLabelItem; + dispInfo.item.iSubItem = 0; + dispInfo.item.pszText = pszText; + dispInfo.item.cchTextMax = textlenT(pszText, isW); + return LISTVIEW_SetItemT(infoPtr, &dispInfo.item, isW); + } -- Dimi.

2 1

Re: Implement NtQuerySystemInformation(SystemTimeOfDayInformation)
by Rein Klazes 19 May '05

19 May '05

On Tue, 17 May 2005 23:06:51 +0200, you wrote: > + RtlSystemTimeToLocalTime(&sys_time, &local_time); > + /* liExpTimeZoneBias is not the same as tzi.Bias since it takes > + * disabled DST correction into account (tested on WinXP). */ > + local_diff.QuadPart = sys_time.QuadPart - local_time.QuadPart; You mean the "Automatically adjust clock for daylight saving"? Wine does not follow this silly flag and at a couple of places it depends on it. I doubt whether it should. > + > + sti->liKeBootTime = NTDLL_BootTime; > + sti->liKeSystemTime = sys_time; > + sti->liExpTimeZoneBias = local_diff; > + sti->uCurrentTimeZoneId = RtlQueryTimeZoneInformation(&tzi); This does not look correct. AFAIK RtlQueryTimeZoneInformation returns an NT-status code, 0 is no success, otherwise it is an error code. Have a look at GetTimeZoneInformation how to compute a TimeZoneId. Rein.

2 3

Re: [crypt32] take 3: CryptProtectData/CryptUnprotectData helper functions
by Mike McCormack 18 May '05

18 May '05

Kees Cook wrote: > Corrected for MESSAGE() use and a botch of wine_dbg_sprintf usage. Julliard won't commit dead code, so you have to make sure that each function that you send is used by something else. How about sending a patch implementing CryptProtectData()? Mike

1 0

Re: [crypt32] CryptProtectData/CryptUnprotectData helper functions
by Alexandre Julliard 18 May '05

18 May '05

Kees Cook <kees(a)outflux.net> writes: > +static int > +hexprint(const char *s, unsigned char *p, int n) > +{ > + char report[80]; > + int r=-1; > + snprintf(report,16,"%14s:", s); > + while (--n >= 0) > + { > + if (r++ % 20 == 19) > + { > + wine_dbg_printf("%s\n",report); > + snprintf(report,16,"%14s ", ""); > + } > + sprintf(report+strlen(report)," %02x", *p++); > + } > + wine_dbg_printf("%s\n",report); You should use wine_dbg_sprintf here and return a string. > +static > +void serialize_dword(DWORD value,BYTE ** ptr) > +{ > + /*TRACE("called\n");*/ > + > + *((DWORD*)*ptr)=value; > + *ptr+=sizeof(DWORD); This (and other similar things later on) isn't safe for CPUs that don't allow unaligned accesses, you have to use memcpy instead. > +static > +BOOL fill_protect_data(struct protect_data_t * pInfo, LPCWSTR szDataDescr, > + HCRYPTPROV hProv) > +{ > + DWORD dwStrLen; > + > + TRACE("called\n"); > + > + if (!pInfo) return FALSE; > + > + dwStrLen=lstrlenW(szDataDescr); > + > + memset(pInfo,0,sizeof(*pInfo)); > + > + pInfo->count0=0x0001; > + > + pInfo->info0.cbData=strlen(crypt_magic_str)+1; > + pInfo->info0.pbData=strdup(crypt_magic_str); You don't want to use strdup, you should use HeapAlloc and friends (especially since you use HeapFree later on). > +static void > +crypt_report_func_input(DATA_BLOB* pDataIn, > + DATA_BLOB* pOptionalEntropy, > + CRYPTPROTECT_PROMPTSTRUCT* pPromptStruct, > + DWORD dwFlags) > +{ > + wine_dbg_printf("\tpPromptStruct: 0x%x\n",(unsigned int)pPromptStruct); > + if (pPromptStruct) > + { > + wine_dbg_printf("\t\tcbSize: 0x%x\n",(unsigned int)pPromptStruct->cbSize); > + wine_dbg_printf("\t\tdwPromptFlags: 0x%x\n",(unsigned int)pPromptStruct->dwPromptFlags); > + wine_dbg_printf("\t\thwndApp: 0x%x\n",(unsigned int)pPromptStruct->hwndApp); > + wine_dbg_printf("\t\tszPrompt: 0x%x %s\n", > + (unsigned int)pPromptStruct->szPrompt, > + pPromptStruct->szPrompt ? debugstr_w(pPromptStruct->szPrompt) > + : ""); > + } > + wine_dbg_printf("\tdwFlags: 0x%04x\n",(unsigned int)dwFlags); > + wine_dbg_printf("\tpDataIn->cbData: %u\n",(unsigned int)pDataIn->cbData); > + wine_dbg_printf("\tpDataIn->pbData: 0x%x\n",(unsigned int)pDataIn->pbData); > + hexprint("pbData", pDataIn->pbData, pDataIn->cbData); > + if (pOptionalEntropy) > + { > + wine_dbg_printf("\tpOptionalEntropy->cbData: %u\n",(unsigned int)pOptionalEntropy->cbData); > + wine_dbg_printf("\tpOptionalEntropy->pbData: 0x%x\n",(unsigned int)pOptionalEntropy->pbData); > + hexprint("pbData", pOptionalEntropy->pbData, pOptionalEntropy->cbData); > + wine_dbg_printf("\t\t%s\n",debugstr_an(pOptionalEntropy->pbData,pOptionalEntropy->cbData)); > + } > + > +} > + > +static void > +announce_bad_opaque_data() > +{ > + wine_dbg_printf("CryptUnprotectData received the following pDataIn DATA_BLOB that seems to\n"); > + wine_dbg_printf("have NOT been generated by Wine:\n"); > +} You should use the TRACE/FIXME macros here, not raw wine_dbg_printf. -- Alexandre Julliard julliard(a)winehq.org

4 6

Re: [crypt32] CryptProtectData/CryptUnprotectData helper functions
by Kees Cook 18 May '05

18 May '05

On Wed, May 18, 2005 at 12:29:58PM -0700, Kees Cook wrote: > Take two. Ignore... this patch is broken, and I need to do an update of wine_dbg_printf() -> MESSAGE() -- Kees Cook @outflux.net

1 0

Re: [wine] Re: A better Beep
by Marcelo Duarte 18 May '05

18 May '05

David Lee Lambert escreveu: >On Tue, May 17, 2005 at 10:43:25AM -0300, Marcelo Duarte wrote: > > >>Whats happens if a program calls bepp with invalid values for dwFreq and >>dwDur on a linux console? >> >> > >The bell becomes inaudible until reset by another terminal-control >sequence. However, the special value (duration=-1) does not have the >same semantics; it's simply ignored. > >On my box that has a PC speaker, the bell is silent when any of the >following are true: > >freq < 21 >dur == 0 >dur > 1999 > >I can't hear any tones above 12000 Hz or so, but I'm not sure whether >that's my hearing, the speaker-hardware, or the kernel's decision. > >Should I add range-checks on frequency and duration? > > > >>(the programmer thinks that dwFreq and dwDur are ignored by Win95). >> >> > >That's what the API documentation says. Are they used on WinXP or Win2k? > > > If you have programs to test, send me.

1 0

MSN Webcam
by Joaquín Fernández 18 May '05

18 May '05

hmmmm... can you use a webcam in msn under wine???? If yes, can you explain to me how to do it? I was not able to run msn messenger in wine. Regards Joaquín -- -----BEGIN PGP PUBLIC KEY BLOCK----- Version: GnuPG v1.2.4 (GNU/Linux) mQFCBEFI/gARAwDd2+ojasT3rCyRktSw+Ix3m+yoxSD0NkpMLlunmJxwvn6wKZVl mDw76/Zu9mqDWWeSGdSl+60T7fDLrJZSEB45O9T5jdujj01GFeer7xuiuHBTFw8o CXqD/hzhqYc46ecAoIQQjZ2qZtOWLPRBbegK/nyOIguNAv9QGiKPLBS8o0ksxEUp EfLAExVmu6Zp693uKGf6XrBWNcLriuwRPr1mjy3N/bhMlqc3vcTeUBwxiUuX5h2P NQgB3d2AbJS6oEvhmZL0Bn/8Ij/MSvVrartmCXuw9eSx0aMC/R7Kw9TtUfxFVUGx fQKwoA9BXNElPLcNohbBS/fH87IMMxCJyn+rmTeNCEcUEQ7UgvVCdlzZ8+L4PdlH qGR81nhZVEwPRnSSesLpSHRC1QQVoceBeb7PICr/b2eZiKMX+bQ+Sm9hcXXDrW4g TWFudWVsIEZlcm7DoW5kZXogUXVpbGVzIDxqb2FxdWluQGNhc2FkZWFsYWJhbnph Lm9yZz6IXgQTEQIAHgUCQUj+AAIbAwYLCQgHAwIDFQIDAxYCAQIeAQIXgAAKCRBH 677/q7xL4e15AJwNfSpeaXXMH2EjuKblfeBe51MrUgCcCP5jEnpXrDXLWULIW5yD t6VdHlU= =a8BM -----END PGP PUBLIC KEY BLOCK-----

2 1

Re: list.h: document the more convenient iteration method
by Dimi Paun 18 May '05

18 May '05

On Wed, 2005-05-18 at 08:01 -0400, Dimi Paun wrote: > * struct list *cursor; > - * LIST_FOR_EACH( cursor, &global_gadgets ) > + * LIST_FOR_EACH_ENTRY( cursor, &global_gadgets, struct gadget *, > gadget ) > * { Looking at this, why do we need cursor at all for the LIST_FOR_EACH_ENTRY? Also the type and iteration var should come first, to be simiar for loops in pretty much any language out there. I think it should just be: LIST_FOR_EACH_ENTRY( struct gadget *, gadget, &global_gadgets ) Ah, I see now the problem. We would need to end it with LIST_END instead of just doing {...}. Not such a big deal, we do the same for server requests too :) But I can see why we would like to keep cursor. If so, it should come at the end: LIST_FOR_EACH_ENTRY(struct gadget *,gadget,&global_gadgets,cursor) -- Dimi Paun <dimi(a)lattica.com> Lattica, Inc.

1 0

Re: Small cleanup in list handling
by Robert Shearman 18 May '05

18 May '05

Dimi Paun wrote: >Index: dlls/ntdll/exception.c >=================================================================== >RCS file: /var/cvs/wine/dlls/ntdll/exception.c,v >retrieving revision 1.75 >diff -u -p -r1.75 exception.c >--- dlls/ntdll/exception.c 13 May 2005 13:56:47 -0000 1.75 >+++ dlls/ntdll/exception.c 13 May 2005 19:01:31 -0000 >@@ -480,7 +480,8 @@ ULONG WINAPI RtlRemoveVectoredExceptionH > RtlEnterCriticalSection( &vectored_handlers_section ); > LIST_FOR_EACH( ptr, &vectored_handlers ) > { >- if (ptr == &((VECTORED_HANDLER *)handler)->entry) >+ VECTORED_HANDLER *curr_handler = LIST_ENTRY( ptr, VECTORED_HANDLER, entry ); >+ if (curr_handler == handler) > { > list_remove( ptr ); > ret = TRUE; > > This patch has already been committed, but for future reference code of this form: LIST_FOR_EACH( cursor, list) { TYPE list_entry = LIST_ENTRY( cursor, TYPE, entry ); ... Can be replaced by the following for IMHO, slightly more readable code: LIST_FOR_EACH_ENTRY( list_entry, list, TYPE, entry) { ... Rob

2 1

Re: MSN Webcam
by Maarten Lankhorst 18 May '05

18 May '05

MediaHost (TM) wrote: > Looks interesting! Any idea, when your patches get committed into the > "official" releases? > >> Maarten Lankhorst wrote: >> >>> Status: Nearly complete, need testers. >>> >>> http://wiki.winehq.org/MSN_Messenger_webcam_support >>> >>> You need the native quartz dll, which get installed with internet >>> explorer, as the builtin quartz doesn't work properly when changing >>> media format on initialisation. >> >> Possibly a problem with the code I wrote. Can you provide more >> details on this? >> >>> IFilterGraph Reconnect isn't implemented, and for some reason *looks >>> strangely at Robert* instead of a FIXME() there's a TRACE telling it >>> is a stub... >> >> I didn't write any of the IFilterGraph implementation, but you are >> right that it is crazy. >> >> Rob >> Writing some final bugfixes at the moment, so I think a matter of days before it'll be done in a state that julliard might commit.. There were 2 small reasons why builtin quartz wasn't working 1. if format changed, it would need to reconnect with new size (Quartz, submitted a patch) 2. if format wasn't changed, I used my own optimized code, that wrote 960 bytes to a wrong place, so it crashed. (Totally my fault) btw, nice name and signature...

1 0