蚂蚁的天空II － 工作、生活和梦想

For almost one week i trace in and out spoosv to find the source the Bug comes from and today i reduce the area:

it comes from original Layered sample! - The implementation for overlapped send operations is NOT bug free.

i must think out myself at this point:

Why i spend so much time to make it out?

Because i break my incremental development process myself:

Layered->our framework->Feature 1->Feature 2

Do you know what's incremental development?

Regret myself...

• System Services for the Windows Server 2003 Family and Windows XP Operating Systems

When i attach spoolsv.exe with windbg, the bug do NOT appear! Ahh...

Trace in and out spoolsv.exe for two days with WinDbg, but still have no idea about the strange bug - if i trace the LPR print, then the bug fly no where. 555555555...

SpoolerMBThread:

ntoskrnl.exe!KiSwapThread+0xc5
ntoskrnl.exe!KiSuspendThread+0x29
ntoskrnl.exe!KiSwapThread+0xfc
USER32.DLL!NtUserWaitMessage+0xb
USER32.DLL!DialogBox2+0x216
USER32.DLL!InternalDialogBox+0xd1
USER32.DLL!SoftModalMessageBox+0x752
USER32.DLL!MessageBoxWorker+0x247
USER32.DLL!MessageBoxExW+0x77
USER32.DLL!MessageBoxW+0x49
localspl.dll!WinStationMessage+0x165
localspl.dll!MyMessageBox+0xc4
localspl.dll!SpoolerMBThread+0x16
KERNEL32.dll!BaseThreadStart+0x52

PortThread
ntoskrnl.exe!KiSwapThread+0xc5
ntoskrnl.exe!KiSuspendThread+0x29
ntoskrnl.exe!KiSwapThread+0xfc
ntdll.dll!NtDelayExecution+0xb
KERNEL32.dll!SleepEx+0x32
KERNEL32.dll!Sleep+0xb
tcpmon.dll!CTcpJob::StartDocW+0x8b
tcpmon.dll!CLPRJob::StartJob+0x9d
tcpmon.dll!CLPRJob::StartDocW+0x75
tcpmon.dll!CLPRPort::StartDocW+0x5b
tcpmon.dll!CPort::StartDocW+0x65
tcpmon.dll!CPortMgr::StartDocPort+0x3f
tcpmon.dll!StartDocPort+0x1b
E_SL2345.DLL!DllEntryPoint+0x2b4
localspl.dll!PrintingDirectlyToPort+0x156
localspl.dll!LocalStartDocPrinter+0x54
SPOOLSS.DLL!StartDocPrinterW+0x2d
localspl.dll!PrintRawJob+0x35
localspl.dll!PrintDocumentOnPrintProcessor+0x4c
localspl.dll!PrintDocumentThruPrintProcessor+0x19f
localspl.dll!PortThread+0x24a
KERNEL32.dll!BaseThreadStart+0x52

SchedulerThread:

ntoskrnl.exe!KiSwapThread+0xc5
ntoskrnl.exe!KiSuspendThread+0x29
ntoskrnl.exe!KiSwapThread+0xfc
ntdll.dll!ZwWaitForSingleObject+0xb
KERNEL32.dll!WaitForSingleObjectEx+0x71
KERNEL32.dll!WaitForSingleObject+0xf
localspl.dll!SchedulerThread+0x30
KERNEL32.dll!BaseThreadStart+0x52

To go through the NTLM proxy, i use WinInet API do some works(wininet.dll: if not me, who?) Then i got some trouble when using APIs, so i decide to sniffe HTTP with sniffit - but it does NOT work on Japanese OS(ant: why?)

So i have to quick and dirty fix the lsp to capture HTTP myself >_<

OBJ=$(SRC:.cpp=.obj)

!ifndef NODEBUG
O=.\Debug
{$S}.cpp{$O}.obj::
$(CPP)\
-Fo$(O)\$(NULL)\
$(CDEBUG)\
$(CPPFLAGS)\
$(CVARSDLL)\
-DSTRICT\
/Od\
-D_DEBUG\
/c $<
!else
O=.\Release
{$S}.cpp{$O}.obj::
$(CPP)\
-Fo$(O)\$(NULL)\
$(CDEBUG)\
$(CPPFLAGS)\
$(CVARSDLL)\
-DSTRICT\
-DNODEBUG\
/c $<
!endif

all:MAKE_DIR $(MOD).dll

MAKE_DIR:
-if not exist $(O) mkdir $(O)

$(MOD).dll:$(O)\$(OBJ)
$(link)\
$(dlllflags) \
-base:$(BASE)\
-out:$(O)\$(MOD).dll\
$(O)\$(OBJ)\
$(conlibsdll)

clean:
-rd /q /s $(O)

I DID keep the Date in my bad memory

And i miss her so-----------------------------------------------------...much!