Drugwash

Very interesting details! Can't wait to test this technique (be it in alpha/beta stage - I'm used to that). I suspect my sudden GDI crashes may occur due to overloaded registry (16MB+ all in all) and I'm extremely curious if this technique can help in this case. And who else would be the best one to take a stab at it...?

I have fixed the confilict is SVN.Thanks, KEXBASES.DLL compiles fine now. There are a few other minor issues, such as warnings and a compilation error for setupkex.exe in the debug version; apparently there's no definition for printf (mentioned in debug.h) anywhere on my setup: neither in PSDK, nor in VC98 or throughout the KernelEx project. I have attached the complete log of a compilation session (both Release and Debug), as well as a patch for project files that includes Language neutral overall and changes some output paths for easier overview; hopefully someone finds these changes useful and fingers crossed I didn't do something stupid. kexproject_patch.7z

I see; let's hope it's been fixed.Thank you for fixing the source archive name. Too bad. I'm using Total Commander with plug-ins for all archive operations, but none of those can correctly handle NSIS extraction. Oh well... It's just function names, not the code lines. I don't think the debug symbols are being (correctly) loaded. Anyway, it's Miranda's plug-in called Crash Dumper that creates the report and it offers more details, like currently loaded modules (Miranda-wise and system-wise). I do have the MSLU package from MS, that includes unicows.pdb. Hmm? Wouldn't compile? I compile with VC6 under 98SE myself! The error you are getting is most likely caused because you didn't build KEXCRT first. (I should include build instructions). I figured out about kexcrt and built it beforehand, but kexbases.dll still won't compile; I suspect some conflict with PSDK. Here's the full log of compilation attempt: --------------------Configuration: KernelEx Base Shared - Win32 Release-------------------- Performing Custom Build Step on .\kexbases.def Creating library .\Release\kernel32ord.lib and object .\Release\kernel32ord.exp Compiling resources... Compiling... _kernel32_apilist.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' _kernel32_stubs.c CopyFileEx.c CreateFileA_fix.c CreateThread_fix.c DeleteCriticalSection.c GetFileSizeEx.c GlobalMemoryStatusEx.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' HeapLocks.c InitializeCriticalSectionAndSpinCount.c Jobs.c KEXVersion.c LockFileEx.c MapViewOfFile.c MoveFileExA.c MoveFileWithProgressA.c OpenThread.c SetFilePointerEx.c ThreadPool.c TryEnterCriticalSection.c uilang.c unikernel32.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' version.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' VirtualAllocEx.c _user32_apilist.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' D:\free\KernelEx\apilibs\kexbases\User32\_user32_apilist.c(64) : error C2440: 'type cast' : cannot convert from '' to 'unsigned long' Context does not allow for disambiguation of overloaded function _user32_stubs.c CallWindowProcA_fix.c DisableProcessWindowsGhosting.c EnableWindow.c ForegroundWindow.c GetAncestor.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' GetMouseMovePointsEx.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' IsHungAppWindow.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' LockWorkStation.c MapVirtualKey_fix.c SendMessage_fix.c UberUSER.c uniuser32.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' UpdateLayeredWindow.c _gdi32_apilist.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' _gdi32_stubs.c FontResourceExA.c GetGlyphOutlineA_fix.c TextOut.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' unigdi32.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' _advapi32_apilist.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' _advapi32_stubs.c RegDisablePredefinedCache.c RegOpenCurrentUser.c security.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' D:\free\KernelEx\apilibs\kexbases\Advapi32\security.c(457) : warning C4018: '<' : signed/unsigned mismatch D:\free\KernelEx\apilibs\kexbases\Advapi32\security.c(1056) : warning C4244: '=' : conversion from 'unsigned long' to 'unsigned char', possible loss of data D:\free\KernelEx\apilibs\kexbases\Advapi32\security.c(1057) : warning C4244: '=' : conversion from 'unsigned long' to 'unsigned short', possible loss of data D:\free\KernelEx\apilibs\kexbases\Advapi32\security.c(1277) : warning C4018: '>' : signed/unsigned mismatch D:\free\KernelEx\apilibs\kexbases\Advapi32\security.c(1281) : warning C4018: '>' : signed/unsigned mismatch TraceMessage.c uniadvapi32.c _comdlg32_apilist.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' PrintDlgEx.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' _version_apilist.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' universion.c _shell32_apilist.c .\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed .\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' _shell32_stubs.c CommandLineToArgvW.c IsUserAnAdmin.c common.c main.c D:\free\KernelEx\apilibs\kexbases\auxdecl.h(134) : error C2733: second C linkage of overloaded function 'GetAltTabInfoA' not allowed D:\free\KernelEx\apilibs\kexbases\auxdecl.h(134) : see declaration of 'GetAltTabInfoA' Error executing cl.exe.</P> <P>kexbases.dll - 20 error(s), 5 warning(s)

You could do it if you wanted to. Untap the bottle (not the vodka one ) and let the genius out.

This reminds me the WLL update project has stalled for a long time, having my fingers stuck in too many other pies. Maybe you could chime in with a few ideas, time allowing. Please note that most of the links expired (those at uploaded.to) but I still have the complete package somewhere around, if you wanna look at the (ugly ) code. I wish PassingBy would get some spare time too, coz he really knows what he's doing.

Heh... no problems so far with regular 9x applications. I've tried however to run the Unicode version of Miranda IM 0.8 alpha #28 self-built (that's my test bed for KernelEx ) and am constantly getting crashes in UNICOWS.DLL: 7F2DAF19 (UNICOWS 7F2D0000): (filename not available) (0): gwcslen 7F2D5645 (UNICOWS 7F2D0000): (filename not available) (0): GodotToCpgCchOnHeap 7F2D866E (UNICOWS 7F2D0000): (filename not available) (0): GodotTransmitMessage 7F2E8A71 (UNICOWS 7F2D0000): (filename not available) (0): GodotDefWindowProcW 0043988E (MIRANDA32 00400000): D:\miranda\trunk\miranda\src\modules\utils\hyperlink.cpp (200): HyperlinkWndProc BFF7363B (KERNEL32 BFF70000): (filename not available) (0): Ordinal121 BFF9443B (KERNEL32 BFF70000): (filename not available) (0): UTUnRegister At some point it was mentioned KEXBASEN.DLL. 7D002CB4 (KEXBASEN 7D000000): (filename not available) (0): (function-name not available) The crash occurs when selecting Options or Accounts in Miranda's main menu. Other dialogs such as Profile Manager, View/Change My Details or Find/Add Contacts display question marks in certain places. I tried to build the debug version of KernelEx with the debug symbols, hoping it would offer more details, but it wouldn't recognize the pdb. Also, KEXBASES.DLL wouldn't compile with VC6 under 98SE. And the debug symbols for UNICOWS.DLL (unicows.pdb) wouldn't be recognized either, although it's in the same folder with unicows.dll (both in System and in Miranda's folder, just to make sure). Operating system: Win98SE KernelEx core: 4.0.11 Application: Miranda IM 0.8 built from SVN with VC6-SP6 If other information is required, please ask.

Install and run CodeStuff Starter, it will tell you what processes are running and what files (dependencies) are loaded for each of them. As long as a process is running, all its dependencies will be locked. To free a file, its parent process must be closed/killed. But system files can be shared between multiple processes, so all of those processes should be stopped before being able to operate on that file. Kernel files cannot be operated on since they're always running and cannot be shut down; that's why any change to such files (installing KernelEx, for example) needs a reboot, so that key files can be operated on while in DOS mode, before being loaded by the Windows subsystem.

Few minor issues encountered along the way: • The (fixed?) osdn mirror for the tar source package reports 'page not found'; may be a temporary glitch. Changing the path to point to switch instead of osdn fixes the issue. • May be a good idea to mention 'src' or 'source' in archive's name; initially I thought it's just the usual non-installer alternative for the compiled application. • Speaking of non-installer... I'd pretty much like to have a zip/7z alternative of the application, for analysis/translation/etc purposes; the NSIS installer doesn't easily allow manual extraction of the content and not everybody has a development environment at hand to build everything up from the sources. That being said, the new KernelEx is welcome and hopefully will bring much happiness for us die-hard 9x-ers. Thank you Xeno86 and Tihiy for your efforts!

Reboot to plain DOS (hold F8 while booting) and replace the files. It's good practice to keep backup copies in case anything goes wrong. If there's too many files to replace manually: - create a folder in the root drive (say C:\newfiles) - create another folder for backups (say C:\backup) - copy the replacement files in newfiles - copy the original files in backup - create a batch file that copies new files to the desired location(s) and place it in newfiles - reboot to DOS and run the batch file If any of the original files has Read-only, Hidden and/or System attributes, remove them before rebooting to DOS.

Out of all the info presented along this thread, the power source voltage reports are the most important and yet appropriate measures have not been taken. First of all, get a good source of light and have a close look at all the big capacitors on the mobo; if any of them is gonflated, you got your culprit. Thing is, there can be more of them. And thing is, they can also go silently dead, meaning aspect is normal but capacity is lost and moreover, they may draw an important amount of flyback current that brings the power source to its knees. This notwithstanding, the PSU itself may be damaged - bad capacitors inside it - so it cannot supply the required current anymore, hence low voltage under heavy duty. You may visually check the capacitors inside the PSU if you manage to open it without damaging it's internals or cables. In any of the cases, the (visually) bad capacitors can be replaced by a skilled technician (I've done it a few times, including on my old Jetway SiS630 mobo - 9 caps replaced). Those that do not present any visual sign of damage would be harder to 'sniff'. If there's no sign of bad capacitors on the mobo, you may try to replace the PSU with a good (tested) one and see if that fixes the issue. But remember there can be bad caps on the mobo that do not present visual signs of damage. Good luck!

I remember having had similar problems with a machine when screensaver was off. Try setting it to one minute below power off value. Also have a look in the BIOS, make sure APM/ACPI are enabled (this may impact certain setups so first take note of the current settings before changing anything) and check the Video off setting on the Power management page (depending on BIOS type).

For what it's worth, it's not happening for me, with UberSkin installed. And maybe it's not about UberSkin... try the 256-colors patched Explorer.exe, maybe it fixes this issue too. My setup is Win98SE (Explorer.exe 4.72.3612.1710), UberSkin 8.3.6, MSI GeForce4 Ti4200 VTD8X with MSI driver 45.32. Dual-view IBM G50 and HP56 monitors (both 15").

Ah, I had overlooked VMM32.VXD. Thanks for pointing out those tools, will have a look inside VMM32 soon.

Impressive, I must say. So... we must enhance the old VxD. But there's a small problem: no such VCDFSD.VXD on my system. Maybe you're referring to CDVSD.VXD, which indeed exists. Anyway, thanks for the extensive research. I'll have to read this again later, coz it's almost 5 AM here and my eyelids are too heavy now.

Well, the problem may be with the COPY2GB patch, in this case. Check any available documentation on it and/or contact the author, if possible.

My, my... isn't this a small world...

Let me put it this way: an "object" is either available for everybody or not available at all. That'd be my policy if I owned YouTube. Once they allow preferential settings, they're no better than the owners or whoever uploaded that material. In fact, they've been [insert your words here] all along, for offering such preferential options, to begin with.Anyway, let's drop it since I'm not gonna visit that site anymore for as long as I live. On another note, I wonder if anybody else noticed a small glitch with this version of Flash or it's just my system... At certain times, the bottom bar with progress slider and all on YouTube, in full screen, would slide out of sync horizontally for a split second.

While testing Flash 9.0.159.0 on random YouTube videos, I stumbled into another racist stupidity: This video is not available in your country. It's about Krezip-I apologize, and I have no idea who that is or what the video is about. But if that place has become a selective marketing place, then they can shovel that tube right up <there>. I will terminate my account on YouTube effective immediately and never ever visit that site again or advise anyone to; quite the opposite. [EDIT] YouTube account deleted.

Excel Viewer LoadLibraryW("\mso.dll") returned NULL by thread 0xFFFFD8F9. Error: This function is only valid in Win32 mode (120). GetProcAddress(0xBFF70000 [c:\windows\system\KERNEL32.DLL], "QueryActCtxW") called from "c:\program files\microsoft office\office11\XLVIEW.EXE" at address 0x30088869 and returned NULL by thread 0xFFFFD8F9. Error: The specified module could not be found (126).Word Viewer LoadLibraryW(" h(((( H H aVbaX") called from "c:\program files\microsoft office\office11\WORDVIEW.EXE" at address 0x30165056 by thread 0xFFF9A669. LoadLibraryW(" h(((( H H aVbaX") returned NULL by thread 0xFFF9A669. Error: This function is only valid in Win32 mode (120). LoadLibraryW("\mso.dll") called from "c:\program files\microsoft office\office11\WORDVIEW.EXE" at address 0x3060F361 by thread 0xFFF9A669. LoadLibraryW("\mso.dll") returned NULL by thread 0xFFF9A669. Error: This function is only valid in Win32 mode (120). GetProcAddress(0xBFF70000 [c:\windows\system\KERNEL32.DLL], "QueryActCtxW") called from "c:\program files\microsoft office\office11\WORDVIEW.EXE" at address 0x3017451B and returned NULL by thread 0xFFF9A669. Error: The specified module could not be found (126).mso.dll is a delayed load and can be found when path is correctly registered, but LoadLibraryW in kernel32.dll might not work correctly and there's also QueryActCtxW missing. On top of that, whadda heck is that weirdly-named file that WORDVIEW.EXE is trying to load in Unicode mode? I'm afraid that's as far as we can go with Word and Excel Viewer 2003 on 9x systems.

What about SmoothText, wasn't that one designed for Win2000 or am I mistaken? It's WildBill's.

Thanks for the information, dencorso. The screenshot above pertains to the library found in the NT4 packages on MDGx's site; the one installed on my system is actually older (the original one?): 4.87.1959.1877 / 5.0.1877.6 as reported by FileInfo. I remember having some trouble in the past - can't tell what exactly, though - and I had to downgrade this file (and possibly some others I forgot about). I was about to say that MDGx was wrong saying it can be found in CRYPT9X.EXE, but then I checked the inf and it's being renamed at install time, that's why I didn't notice it at the first glance. Given the past experience (blurry as it is), I'm a little reluctant to installing that package again, for fear that I may screw up the system, but eventually I might do it after backing up all the involved files. It may take a while to notice all the possible side-effects though and by then I might forget the culprit (yeah, got a short memory ).

MDGx, please have a look here; information provided by the FileInfo plug-in in Total Commander:

Dunno what kind of machines you guys have, but my Soyo Pentium III reads SYSTEM.DAT: 16,470,048 bytes and USER.DAT: 2,621,472 bytes and have no such problems. Granted, it only has 256MB RAM and no Gigabit Ethernet adapter (just a regular 10/100MB one).

Thank you but I'm unable to buy online so I have to stick to real stores, be them first-hand or second-hand. I'm also short of money so it'll have to wait for a while. Come to think about it, older games are much better; storyline is solid and can take days or weeks to complete. Nowadays' games are over in a couple hours tops, as if they were movies. Oh wait... they actually are! All recent games are half made of movies supposed to bind the chapters, blah-blah; a lame excuse for cutting down the gameplay to the point one'd rather lay back on the couch, have a beer and watch a movie. O, tempora...

Please check again, EvanD; I got the right version here just as advertised, downloaded yesterday.