Re: [Mingw-w64-public] [Mingw-users] Debugging with GDB on Windows / MinGW is painfully slow
On 2012-7-21 11:38, K. Frank wrote: > As I mentioned above, my gdb version is 7.3.0. > >> >You can try a recent gdb (mostly the gdb build from gdb cvs HEAD) > If anybody knows of a recent mingw or mingw-w64 build of gdb > that addresses this issue, please chime in. > You can try my build of gdb CVS (32bit) see: http://forums.codeblocks.org/index.php/topic,11301.msg77000.html#msg77000 asmwarrior -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [Mingw-users] Debugging with GDB on Windows / MinGW is painfully slow
Hello Eran and asmwarrior!
I have seen something similar with a mingw-w64 build.
(I have taken the liberty of cross-posting this reply to the mingw-w64
list.)
On Fri, Jul 20, 2012 at 9:46 PM, asmwarrior wrote:
> On 2012-7-20 21:10, Eran Ifrah wrote:
>> Hello,
>>
>> I am not sure whether this issue is related to MinGW or to GDB.
>> Since I don't have this issue on Linux / Mac and it only happens on Windows
>> 7 + MinGW, I decided to post it here - maybe someone else on this list face
>> this issue
I am running a 64-bit mingw-w64 build on 64-bit windows 7, and see
slow gdb startup with Qt applications. (I haven't tested whether
startup is also slow with simple non-Qt console programs.)
>> in the past and can shed some light / provide some good tips on the matter.
>>
>> Now to the actual problem:
>> When I debug my application which consists of many shared libraries ("dll"s,
>> around 35 - 40) and I have a breakpoint set in one of the shared libraries
>> the startup time of gdb is *very* slow ( I am talking here about 10 mins ).
In my case I do not have pre-set breakpoints.
When I run gdb on a small Qt test program, (e.g. "gdb test.exe")
the gdb prompt appears essentially immediately, but when I then
type "run" it takes about three minutes for the test application to
actually launch (i.e., for the gui to appear).
Once I get past that slow startup, debugging seems to run normally.
>> ...
>> Some info about my environment:
>>
>> MinGW 4.6.1 (TDM) 32 bit - I tried the official one 4.7 from MinGW with no
>> luck - the problem persists
>> GDB 7.05 (note that I tried all versions of gdbs available on MinGW SF
>> download page, from 6.8.5 -> 7.4 with no luck)
>> Windows 7 64bit
In my case I am running:
g++ (GCC) 4.7.0 20110829 (experimental)
GNU gdb (GDB) 7.3.0.20110829-cvs
from a native 64-bit Ruben "personal build" of g++ 4.7.0, and
Qt 4.8.0-rc1.
(I don't have any reason to think the slow gdb startup is related
to Qt -- I just haven't bothered to test it on non-Qt applications.)
It is worth noting that this problem first showed up for me (or at
least became noticeably worse) recently -- I think the last time
I upgraded my compiler / Qt combination. I don't recall, offhand,
the g++ build / gdb version that I had been using previously that
did not have the slow startup problem (or at least not as badly).
>> ...
>> Any advise?
No, but if anybody gets this issue sorted out, I would love to hear
about it.
>> ...
>> Eran Ifrah
>> Author of the cross platform, open source C++ IDE: http://www.codelite.org
>>
> Hi, Eran
>
> I meet such problem years before, but it was fixed in the gdb trunk months
> ago.
> I'm not sure the mingw gdb official 7.4 have this fixed, but from my test,
> the slow loading problem is already solved.
As I mentioned above, my gdb version is 7.3.0.
> You can try a recent gdb (mostly the gdb build from gdb cvs HEAD)
If anybody knows of a recent mingw or mingw-w64 build of gdb
that addresses this issue, please chime in.
> ...
> asmwarrior
Thanks for any updates on this problem.
K. Frank
--
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
___
Mingw-w64-public mailing list
Mingw-w64-public@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [patch] ws2tcpip.h: EAI_NODATA should be int
2012/7/20 Ozkan Sezer : > On 7/20/12, Corinna Vinschen wrote: >> On Jul 20 20:15, Corinna Vinschen wrote: >>> On Jul 20 19:45, Kai Tietz wrote: >>> > 2012/7/20 Corinna Vinschen : >>> > > Hi, >>> > > >>> > > The error codes returned by getaddrinfo and getnameinfo are defined >>> > > as int values. The below patch fixes that for EAI_NODATA. Ok? >>> > > >>> > > >>> > > Thanks, >>> > > Corinna >>> > >>> > Hmm, I think we should keep these kind of values long for LLP64 and >>> > ILP targets. >>> >>> Sorry, but that doesn't make sense, IMHO. >>> >>> getaddrinfo returns int: >>> >>> http://msdn.microsoft.com/en-us/library/windows/desktop/ms738520%28v=vs.85%29.aspx >>> >>> getnameinfo returns int: >>> >>> http://msdn.microsoft.com/en-us/library/windows/desktop/ms738532%28v=vs.85%29.aspx >>> >>> gai_strerror takes an int as parameter: >>> >>> http://msdn.microsoft.com/en-us/library/windows/desktop/ms738514%28v=vs.85%29.aspx >> >> Oh, I forgot something else. All other EAI_xxx values are defined >> in terms of WSA error values. The WSA error values are defined as >> int values anyway, see psdk_inc/_wsa_errnos.h. So EAI_NODATA is the >> only one defined as long value right now. >> >> >> Corinna >> > > Corinna's arguments make sense to me. I'm OK with whatever will > be the final decision. > > -- > O.S. So, patch is ok. I just wanted to clarify this before ok-ing patch. Thanks, Kai -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [patch] ws2tcpip.h: EAI_NODATA should be int
On 7/20/12, Corinna Vinschen wrote: > On Jul 20 20:15, Corinna Vinschen wrote: >> On Jul 20 19:45, Kai Tietz wrote: >> > 2012/7/20 Corinna Vinschen : >> > > Hi, >> > > >> > > The error codes returned by getaddrinfo and getnameinfo are defined >> > > as int values. The below patch fixes that for EAI_NODATA. Ok? >> > > >> > > >> > > Thanks, >> > > Corinna >> > >> > Hmm, I think we should keep these kind of values long for LLP64 and >> > ILP targets. >> >> Sorry, but that doesn't make sense, IMHO. >> >> getaddrinfo returns int: >> >> http://msdn.microsoft.com/en-us/library/windows/desktop/ms738520%28v=vs.85%29.aspx >> >> getnameinfo returns int: >> >> http://msdn.microsoft.com/en-us/library/windows/desktop/ms738532%28v=vs.85%29.aspx >> >> gai_strerror takes an int as parameter: >> >> http://msdn.microsoft.com/en-us/library/windows/desktop/ms738514%28v=vs.85%29.aspx > > Oh, I forgot something else. All other EAI_xxx values are defined > in terms of WSA error values. The WSA error values are defined as > int values anyway, see psdk_inc/_wsa_errnos.h. So EAI_NODATA is the > only one defined as long value right now. > > > Corinna > Corinna's arguments make sense to me. I'm OK with whatever will be the final decision. -- O.S. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [patch] ws2tcpip.h: EAI_NODATA should be int
On Jul 20 20:15, Corinna Vinschen wrote: > On Jul 20 19:45, Kai Tietz wrote: > > 2012/7/20 Corinna Vinschen : > > > Hi, > > > > > > The error codes returned by getaddrinfo and getnameinfo are defined > > > as int values. The below patch fixes that for EAI_NODATA. Ok? > > > > > > > > > Thanks, > > > Corinna > > > > Hmm, I think we should keep these kind of values long for LLP64 and > > ILP targets. > > Sorry, but that doesn't make sense, IMHO. > > getaddrinfo returns int: > > http://msdn.microsoft.com/en-us/library/windows/desktop/ms738520%28v=vs.85%29.aspx > > getnameinfo returns int: > > http://msdn.microsoft.com/en-us/library/windows/desktop/ms738532%28v=vs.85%29.aspx > > gai_strerror takes an int as parameter: > > http://msdn.microsoft.com/en-us/library/windows/desktop/ms738514%28v=vs.85%29.aspx Oh, I forgot something else. All other EAI_xxx values are defined in terms of WSA error values. The WSA error values are defined as int values anyway, see psdk_inc/_wsa_errnos.h. So EAI_NODATA is the only one defined as long value right now. Corinna -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [patch] ws2tcpip.h: EAI_NODATA should be int
On Jul 20 19:45, Kai Tietz wrote: > 2012/7/20 Corinna Vinschen : > > Hi, > > > > The error codes returned by getaddrinfo and getnameinfo are defined > > as int values. The below patch fixes that for EAI_NODATA. Ok? > > > > > > Thanks, > > Corinna > > Hmm, I think we should keep these kind of values long for LLP64 and > ILP targets. Sorry, but that doesn't make sense, IMHO. getaddrinfo returns int: http://msdn.microsoft.com/en-us/library/windows/desktop/ms738520%28v=vs.85%29.aspx getnameinfo returns int: http://msdn.microsoft.com/en-us/library/windows/desktop/ms738532%28v=vs.85%29.aspx gai_strerror takes an int as parameter: http://msdn.microsoft.com/en-us/library/windows/desktop/ms738514%28v=vs.85%29.aspx Corinna -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [patch] winuser.h: HELP_xxx constants are UINT, not LONG
2012/7/20 Corinna Vinschen : > Hi, > > per MSDN, the WinHelp function takes a UINT as command parameter. > THis is refelcted in our winuser.h, but for some reason most of > the related HELP_xxx defines used as commands are defined as long. > The below patch fixes that. Ok? > > > Thanks, > Corinna > > > * winuser.h: Convert all HELP_xxx command macros to int macros. > > > Index: winuser.h > === > --- winuser.h (revision 5238) > +++ winuser.h (working copy) > @@ -4182,20 +4182,20 @@ >__MINGW_TYPEDEF_AW(PHELPWININFO) >__MINGW_TYPEDEF_AW(LPHELPWININFO) > > -#define HELP_CONTEXT 0x0001L > -#define HELP_QUIT 0x0002L > -#define HELP_INDEX 0x0003L > -#define HELP_CONTENTS 0x0003L > -#define HELP_HELPONHELP 0x0004L > -#define HELP_SETINDEX 0x0005L > -#define HELP_SETCONTENTS 0x0005L > -#define HELP_CONTEXTPOPUP 0x0008L > -#define HELP_FORCEFILE 0x0009L > -#define HELP_KEY 0x0101L > -#define HELP_COMMAND 0x0102L > -#define HELP_PARTIALKEY 0x0105L > -#define HELP_MULTIKEY 0x0201L > -#define HELP_SETWINPOS 0x0203L > +#define HELP_CONTEXT 0x0001 > +#define HELP_QUIT 0x0002 > +#define HELP_INDEX 0x0003 > +#define HELP_CONTENTS 0x0003 > +#define HELP_HELPONHELP 0x0004 > +#define HELP_SETINDEX 0x0005 > +#define HELP_SETCONTENTS 0x0005 > +#define HELP_CONTEXTPOPUP 0x0008 > +#define HELP_FORCEFILE 0x0009 > +#define HELP_KEY 0x0101 > +#define HELP_COMMAND 0x0102 > +#define HELP_PARTIALKEY 0x0105 > +#define HELP_MULTIKEY 0x0201 > +#define HELP_SETWINPOS 0x0203 > #define HELP_CONTEXTMENU 0x000a > #define HELP_FINDER 0x000b > #define HELP_WM_HELP 0x000c Hmm, interestingly. Well, long and int simply didn't made much diffference here. So yes, patch looks ok to me. Regards, Kai -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [patch] ws2tcpip.h: EAI_NODATA should be int
2012/7/20 Corinna Vinschen : > Hi, > > The error codes returned by getaddrinfo and getnameinfo are defined > as int values. The below patch fixes that for EAI_NODATA. Ok? > > > Thanks, > Corinna Hmm, I think we should keep these kind of values long for LLP64 and ILP targets. Ozkan, any strong opinion about this? Regards, Kai -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [patch] directx/dxerr[89].h: Define empty, non-debug DXTRACE_MSG statement as int
This is ok. Thanks, Kai -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
Re: [Mingw-w64-public] [patch] kcom.h: YA GUID
Yes, this is ok. Thanks, Kai -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
[Mingw-w64-public] [patch] kcom.h: YA GUID
Another one... Ok?
Thanks,
Corinna
* kcom.h (STATIC_KoCreateObject): Drop 'L' modifier from Data1
member in GUID definition.
Index: kcom.h
===
--- kcom.h (revision 5238)
+++ kcom.h (working copy)
@@ -27,7 +27,7 @@
extern "C" {
#endif
-#define STATIC_KoCreateObject 0x72CF721CL, 0x525A, 0x11D1, 0x9A, 0xA1, 0x00,
0xA0, 0xC9, 0x22, 0x31, 0x96
+#define STATIC_KoCreateObject 0x72CF721C, 0x525A, 0x11D1, 0x9A, 0xA1, 0x00,
0xA0, 0xC9, 0x22, 0x31, 0x96
DEFINE_GUIDSTRUCT("72CF721C-525A-11D1-9AA1-00A0C9223196", KoCreateObject);
#define KOSTRING_CreateObject L"{72CF721C-525A-11D1-9AA1-00A0C9223196}"
--
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
___
Mingw-w64-public mailing list
Mingw-w64-public@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
[Mingw-w64-public] [patch] directx/dxerr[89].h: Define empty, non-debug DXTRACE_MSG statement as int
Hi, directx/dxerr8.h and directx/dxerr9.h define the empty DXTRACE_MSG statement in the non-debug case as #define DXTRACE_MSG(str) (0L) The "L" qualifier is useless here, so just remove it. Ok? Thanks, Corinna * directx/dxerr8.h (DXTRACE_MSG): Drop long qualifier from empty non-debug definition. * directx/dxerr9.h: Ditto. Index: directx/dxerr8.h === --- directx/dxerr8.h(revision 5238) +++ directx/dxerr8.h(working copy) @@ -42,7 +42,7 @@ #define DXTRACE_ERR(str,hr)DXTrace(__FILE__,(DWORD)__LINE__,hr,str,TRUE) #define DXTRACE_ERR_NOMSGBOX(str,hr) DXTrace(__FILE__,(DWORD)__LINE__,hr,str,FALSE) #else -#define DXTRACE_MSG(str) (0L) +#define DXTRACE_MSG(str) (0) #define DXTRACE_ERR(str,hr)(hr) #define DXTRACE_ERR_NOMSGBOX(str,hr) (hr) #endif Index: directx/dxerr9.h === --- directx/dxerr9.h(revision 5238) +++ directx/dxerr9.h(working copy) @@ -42,7 +42,7 @@ #define DXTRACE_ERR(str,hr)DXTrace(__FILE__,(DWORD)__LINE__,hr,str,FALSE) #define DXTRACE_ERR_MSGBOX(str,hr) DXTrace(__FILE__,(DWORD)__LINE__,hr,str,TRUE) #else -#define DXTRACE_MSG(str) (0L) +#define DXTRACE_MSG(str) (0) #define DXTRACE_ERR(str,hr)(hr) #define DXTRACE_ERR_MSGBOX(str,hr) (hr) #endif -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
[Mingw-w64-public] [patch] ws2tcpip.h: EAI_NODATA should be int
Hi,
The error codes returned by getaddrinfo and getnameinfo are defined
as int values. The below patch fixes that for EAI_NODATA. Ok?
Thanks,
Corinna
* ws2tcpip.h (EAI_NODATA): Define as int value.
Index: ws2tcpip.h
===
--- ws2tcpip.h (revision 5238)
+++ ws2tcpip.h (working copy)
@@ -186,7 +186,7 @@
#define EAI_SERVICE WSATYPE_NOT_FOUND
#define EAI_SOCKTYPE WSAESOCKTNOSUPPORT
-#define EAI_NODATA 11004L /* WSANO_DATA */
+#define EAI_NODATA 11004 /* WSANO_DATA */
typedef struct addrinfo {
int ai_flags;
--
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
___
Mingw-w64-public mailing list
Mingw-w64-public@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
[Mingw-w64-public] [patch] winuser.h: HELP_xxx constants are UINT, not LONG
Hi, per MSDN, the WinHelp function takes a UINT as command parameter. THis is refelcted in our winuser.h, but for some reason most of the related HELP_xxx defines used as commands are defined as long. The below patch fixes that. Ok? Thanks, Corinna * winuser.h: Convert all HELP_xxx command macros to int macros. Index: winuser.h === --- winuser.h (revision 5238) +++ winuser.h (working copy) @@ -4182,20 +4182,20 @@ __MINGW_TYPEDEF_AW(PHELPWININFO) __MINGW_TYPEDEF_AW(LPHELPWININFO) -#define HELP_CONTEXT 0x0001L -#define HELP_QUIT 0x0002L -#define HELP_INDEX 0x0003L -#define HELP_CONTENTS 0x0003L -#define HELP_HELPONHELP 0x0004L -#define HELP_SETINDEX 0x0005L -#define HELP_SETCONTENTS 0x0005L -#define HELP_CONTEXTPOPUP 0x0008L -#define HELP_FORCEFILE 0x0009L -#define HELP_KEY 0x0101L -#define HELP_COMMAND 0x0102L -#define HELP_PARTIALKEY 0x0105L -#define HELP_MULTIKEY 0x0201L -#define HELP_SETWINPOS 0x0203L +#define HELP_CONTEXT 0x0001 +#define HELP_QUIT 0x0002 +#define HELP_INDEX 0x0003 +#define HELP_CONTENTS 0x0003 +#define HELP_HELPONHELP 0x0004 +#define HELP_SETINDEX 0x0005 +#define HELP_SETCONTENTS 0x0005 +#define HELP_CONTEXTPOPUP 0x0008 +#define HELP_FORCEFILE 0x0009 +#define HELP_KEY 0x0101 +#define HELP_COMMAND 0x0102 +#define HELP_PARTIALKEY 0x0105 +#define HELP_MULTIKEY 0x0201 +#define HELP_SETWINPOS 0x0203 #define HELP_CONTEXTMENU 0x000a #define HELP_FINDER 0x000b #define HELP_WM_HELP 0x000c -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
