June 2011 - Wine-Devel - list.winehq.org

Re: gcc 4.6 warning report

by David Laight

On Tue, Jun 28, 2011 at 03:18:47PM -0700, Juan Lang wrote: > >> This is a false positive. ?h_addr_list is declared as a char **, and > >> technically it is, but gethostbyname only returns IPv4 addresses, i.e. > >> ones that can fit in a DWORD. > > > > That doesn't sound like a problem that would give that warning. > > Why not? A cast of a pointer to a DWORD with 64-bit pointers would > cause truncation, if it were indeed a pointer that were being cast. > It isn't, even though it's declared that way. But 'char **h_addr_list;' is a pointer to an array of pointers to address buffers (defined as char) - not a pointer to an array of addresses. IIRC the 'something' will be an IPv4 address - which might be held as a 32 bit quantity rather that an array of char. But nowhere should you cast the 32bit int (DWORD) to/from a pointer type. David -- David Laight: david(a)l8s.co.uk

June 29, 2011

Re: wine bug 27600

by wylda＠volny.cz

> > Maybe gcc 4.4.5 is too old or i'm extremly dumb today :-/ > > > > Regards, > > W. > > > I'm not a WINE developer, just scratching my own > itch :) > > I have the same gcc version. Well, this might be a > legitimate issue, > another important thing would be to see the > ./configure output log. Thanks for interest :) That gives me hope, i won't be out of wine game ;) > I've merged the patch with the include lines > modified, so they would be > identical to the include files mostly used in the > other wineserver c files. > This is just a guess, I have no idea why it > doesn't work. There is also > today's commit, that changed include/winternl.h: > "ntdll: Encode the > function pointers in the vectored handler list", > but it's not related. > > The patch is rebased against today's commits - > http://dl.dropbox.com/u/6901628/raw.patch Nice to have all that 9+ changes in one box. So later today i will try to apply them on current fresh git tree and upload config.log if any trouble. If anything else comes to your mind in next 10hours don't hesitate to ask for, so i could report that too. Thanks, W. -- IHNED.cz je nový, přehlednější a rychlejší. Přesvědčte se na: www.ihned.cz

June 29, 2011

Re: wine bug 27600

by Vincas Miliūnas

On 06/29/2011 01:25 AM, wylda(a)volny.cz wrote: >> It fails on the #include "winternl.h" line and >> these are standard >> includes, you can try to replace >> >> [code] >> #include <stdarg.h> >> #include <stdio.h> >> [/code] >> >> code fragment with >> >> [code] >> #include "config.h" >> #include "wine/port.h" >> >> #include <assert.h> >> #include <stdarg.h> >> #include <stdio.h> >> [/code] >> >> maybe that will change something. > Unfortunately that's too generic for me (i can't handle code) :-( I > would need it in more detail way or a patch. But i'm not asking you > from that. I do not want to waste developers time ;) > > >> Others that have successfully built with the patch >> applied are WINE's >> testbot and two persons from >> http://bugs.winehq.org/show_bug.cgi?id=20395 > > Maybe gcc 4.4.5 is too old or i'm extremly dumb today :-/ > > Regards, > W. > > > I'm not a WINE developer, just scratching my own itch :) I have the same gcc version. Well, this might be a legitimate issue, another important thing would be to see the ./configure output log. I've merged the patch with the include lines modified, so they would be identical to the include files mostly used in the other wineserver c files. This is just a guess, I have no idea why it doesn't work. There is also today's commit, that changed include/winternl.h: "ntdll: Encode the function pointers in the vectored handler list", but it's not related. The patch is rebased against today's commits - http://dl.dropbox.com/u/6901628/raw.patch There are some paranormal stuff in wineserver, like in the first revision of this patch (I haven't tried since then), if I removed the user.h include (which is not used), some magical force was preventing from successfully copying lParam value in the peek_message function (it had the correct value from the source, but the target was reset back to some constant) for each WM_INPUT message I posted, while other messages where working just fine.

June 28, 2011

Re: wine bug 27600

by wylda＠volny.cz

> It fails on the #include "winternl.h" line and > these are standard > includes, you can try to replace > > [code] > #include <stdarg.h> > #include <stdio.h> > [/code] > > code fragment with > > [code] > #include "config.h" > #include "wine/port.h" > > #include <assert.h> > #include <stdarg.h> > #include <stdio.h> > [/code] > > maybe that will change something. Unfortunately that's too generic for me (i can't handle code) :-( I would need it in more detail way or a patch. But i'm not asking you from that. I do not want to waste developers time ;) > Others that have successfully built with the patch > applied are WINE's > testbot and two persons from > http://bugs.winehq.org/show_bug.cgi?id=20395 Maybe gcc 4.4.5 is too old or i'm extremly dumb today :-/ Regards, W. -- IHNED.cz je nový, přehlednější a rychlejší. Přesvědčte se na: www.ihned.cz

June 28, 2011

Re: gcc 4.6 warning report

by Juan Lang

>> This is a false positive. h_addr_list is declared as a char **, and >> technically it is, but gethostbyname only returns IPv4 addresses, i.e. >> ones that can fit in a DWORD. > > That doesn't sound like a problem that would give that warning. Why not? A cast of a pointer to a DWORD with 64-bit pointers would cause truncation, if it were indeed a pointer that were being cast. It isn't, even though it's declared that way. --Juan

June 28, 2011

Re: wine bug 27600

by Vincas Miliūnas

On 06/29/2011 12:28 AM, wylda(a)volny.cz wrote: >> Looks like there is some external interference, >> try compiling in a clean >> bash environment; some googling suggests that >> LIBRARY_PATH variable >> should be unset - >> https://bbs.archlinux.org/viewtopic.php?id=30418 >> First be sure it can compile the original master >> branch without issues. > > Still fails here :-/ > > * succesfully compiled vanilla wine-1.3.23-50-g2497a91 > * make distclean > * applied all nine patches > * unset LIBRARY_PATH > * ./tools/make_requests //3 files updated > * ./configure > * make -j 16 > > Debian 32 squeeze, gcc v4.4.5, no CFLAGS override. > > > It fails on the #include "winternl.h" line and these are standard includes, you can try to replace [code] #include <stdarg.h> #include <stdio.h> [/code] code fragment with [code] #include "config.h" #include "wine/port.h" #include <assert.h> #include <stdarg.h> #include <stdio.h> [/code] maybe that will change something. Others that have successfully built with the patch applied are WINE's testbot and two persons from http://bugs.winehq.org/show_bug.cgi?id=20395

June 28, 2011

Re: gcc 4.6 warning report

by David Laight

On Tue, Jun 28, 2011 at 12:51:02PM -0700, Juan Lang wrote: > ../../../dlls/netapi32/nbt.c:580:30: warning: cast from pointer to > integer of different size [-Wpointer-to-int-cast] > > This is a false positive. h_addr_list is declared as a char **, and > technically it is, but gethostbyname only returns IPv4 addresses, i.e. > ones that can fit in a DWORD. That doesn't sound like a problem that would give that warning. David -- David Laight: david(a)l8s.co.uk

June 28, 2011

Re: wine bug 27600

by wylda＠volny.cz

> Looks like there is some external interference, > try compiling in a clean > bash environment; some googling suggests that > LIBRARY_PATH variable > should be unset - > https://bbs.archlinux.org/viewtopic.php?id=30418 > First be sure it can compile the original master > branch without issues. Still fails here :-/ * succesfully compiled vanilla wine-1.3.23-50-g2497a91 * make distclean * applied all nine patches * unset LIBRARY_PATH * ./tools/make_requests //3 files updated * ./configure * make -j 16 Debian 32 squeeze, gcc v4.4.5, no CFLAGS override. -- IHNED.cz je nový, přehlednější a rychlejší. Přesvědčte se na: www.ihned.cz

June 28, 2011

Re: wine bug 27600

by Vincas Miliūnas

On 06/29/2011 12:10 AM, wylda(a)volny.cz wrote: > So i another try: > * rm -rf * in my git tree. > * git checkout -f 2497a91f6b1326988dd5ea56dc4052cbcc2e521d > * applied all nine patches > * ./tools/make_requests > * configure + make with -O2 and got again: > >> In file included from raw_input.c:28: >> ../include/winternl.h:2359: error: expected >> declaration specifiers >> or ‘...’ before ‘va_list’ >> ../include/winternl.h:2529: error: expected >> declaration specifiers >> or ‘...’ before ‘va_list’ >> ../include/winternl.h:2530: error: expected >> declaration specifiers >> or ‘...’ before ‘va_list’ >> > > > Looks like there is some external interference, try compiling in a clean bash environment; some googling suggests that LIBRARY_PATH variable should be unset - https://bbs.archlinux.org/viewtopic.php?id=30418 First be sure it can compile the original master branch without issues. Compilation with the patch after make distclean works for me.

June 28, 2011

Re: wine bug 27600

by wylda＠volny.cz

So i another try: * rm -rf * in my git tree. * git checkout -f 2497a91f6b1326988dd5ea56dc4052cbcc2e521d * applied all nine patches * ./tools/make_requests * configure + make with -O2 and got again: > > In file included from raw_input.c:28: > ../include/winternl.h:2359: error: expected > declaration specifiers > or ‘...’ before ‘va_list’ > ../include/winternl.h:2529: error: expected > declaration specifiers > or ‘...’ before ‘va_list’ > ../include/winternl.h:2530: error: expected > declaration specifiers > or ‘...’ before ‘va_list’ > -- IHNED.cz je nový, přehlednější a rychlejší. Přesvědčte se na: www.ihned.cz

June 28, 2011