jumper

I tested gimp-2.8.14-setup.exe and was unable to reproduce the previously reported problems (most of the reported files do no exist). Please use ImportPatcher40 to check the problem gimp files with "Test by loading" disabled. Then compare that missing api list to ChangeLog.7 (or .8) and Ktree results in core.ini->kexbasen to find potential sources of the problem. Checking "Use API Hook" in an app's KernelEx properties causes Kernelex.dll to look for a plug-in named ApiHook.dll. That plug-in and a debug console to go with it is almost ready. There is no performance penalty if "Use API Hook" is not checked or ApiHook.dll is not found, and will be almost none when api logging (filtered to console or raw to file) is disabled in the debug console (can be enabled/disabled on-the-fly!). Edit by jumper, 28 March 2015 - 07:36 AM.: ApiHook.8.7z Put apihook.dll in KernelEx folder. Rename DebugWindow.exe to KexApiLog.exe (to match the icon) if you like and run from anywhere to begin api monitoring. Right-click in window for options. Try setting filter to Include:"7d00" to see just apis implemented in kexbasen. Edit by jumper, 30 March 2015 - 03:23 PM.: Disable KernelEx extensions on these two files!

Webfonts look fixable for O12 and FF9 in 9x. Can we find or create a small test case to test api support against?

Release.8.7z KernelEx v4.5.2015.8 is nearly finally ready. The XnConvert issue turned out to be very easy to find (using ProcWin) and fix (SHGetImageList needed to not exist or return E_NOTIMPL). There are three different codes for "Not Implemented"--SHGetImageList was returning the wrong one. I have now defined three new constants in common.h to make picking the right one easier. I have also reviewed the return codes for all new and most existing stubs (except Advapi32) and updated many of them. See tentative change log below. The v4.5.2015.9 changes are what I backed out to get to a more stable release. Between 1/24 and 2/28 I was learning the ApiHook interface (very powerful!) and working on improving the performance of the ApiLog/DebugWindow tool to make it usable without rebooting. KernelEx v4.5.2015.9 by jumper [TENTATIVE]2015-01-24Kexbases update: Kernel32:CreateProcessW_fix now chains to CreateProcessA_fixKexbasen new: Ntdll: 70 apis forwarded to Advapi32, 11 to Kernel32Ntdll: 5 new failure stubs: NtCreateProfile, NtOpenFile, NtQueryInformationProcess, NtQueryInformationThread, NtQuerySystemInformationKexbasen forwards: Oleaut32:442/RegisterTypeLibForUser, 443/UnRegisterTypeLibForUser => *RegisterTypeLibMore delay-load improvements---------------------------------------KernelEx v4.5.2015.8 by jumper2015-03-06 Increased ALLOC_CAPACITY in apiconfmgr.cpp and apilib.cpp from 10 to 162015-03-05common\common.h #define ERROR_NI 120L #define HRESULT_NI 0x80004001L #define STATUS_NI 0xC0000002L #define seUNIMPL_FUNC(name,params,err)Kexbasen\winspool\_winspool_stubs.c rsUNIMPL_FUNC(GetPrinterDataExA, ERROR_NI, 7); rsUNIMPL_FUNC(GetPrinterDataExW, ERROR_NI, 7); rsUNIMPL_FUNC(SetPrinterDataExA, ERROR_NI, 6); rsUNIMPL_FUNC(SetPrinterDataExW, ERROR_NI, 6);Kexbases\gdi32\_gdi32_stubs.c rsUNIMPL_FUNC(SetDCBrushColor, CLR_INVALID, 2); rsUNIMPL_FUNC(SetDCPenColor, CLR_INVALID, 2); rsUNIMPL_FUNC(GetDCBrushColor, CLR_INVALID, 1); rsUNIMPL_FUNC(GetDCPenColor, CLR_INVALID, 1);Kexbases\kernel32\_kernel32_stubs.c seUNIMPL_FUNC(SetDllDirectoryA, 1, ERROR_NI); seUNIMPL_FUNC(SetDllDirectoryW, 1, ERROR_NI);Kexbases\shell32\_shell32_stubs.c seUNIMPL_FUNC(SHCreateShellItem, 4, ERROR_NI); rsUNIMPL_FUNC(SHOpenFolderAndSelectItems, HRESULT_NI, 4);Kexbases\user32\_user32_stubs.c seUNIMPL_FUNC(SetLayeredWindowAttributes, 4, ERROR_NI);2015-02-28Kexbases stubs: Shell32:727/SHGetImageList now correctly returns E_NOTIMPLKexbasen stubs: Oleaut32:442/RegisterTypeLibForUser, 443/UnRegisterTypeLibForUser now correctly return E_NOTIMPLCore: ApiHook enabled new shared buffers: hConsole, bLogFileNow working: XnConvert needed SHGetImageList to not exist or return E_NOTIMPL---------------------------------------KernelEx v4.5.2015.7 by jumper2015-01-22Kexbases bugfix: Kernel32:CreateProcessA (fix)---------------------------------------KernelEx v4.5.2015.6 by jumper, Xeno86, Tihiy2015-01-20New installer for partial updates: InstallDLLs.pif (and helper file x.bat)Copied from 4.5.3beta (master) into apilibs\kexbases\kernel32\ : unikernel32.c, process.c, thread.c apilibs\kexbases\user32\ : desktop.cpp, desktop.h common\ : listhead.h, kexcoresdk.h, kstructs.h core\ : internals.cpp, internals.hRemoved / merged into thread.c: CreateFiberEx.c, CreateThread_fix.c, OpenThread.cRemoved / merged into process.c: ExitProcess_fix.c, GetProcessId.c, IsWow64Process.c, RemoteDesktop.cKexbases: Kernel32:CreateProcessA (fix) Kernel32:CreateProcessAsUserA (new) Kernel32:CreateProcessW (new, replaces forward to unicows) Kernel32:CreateRemoteThread (new)---------------------------------------KernelEx v4.5.2015.5 by jumper2015-01-17Kexbasen new: Ntdll:<77 crt apis> forwarded to Msvcrt, +one other---------------------------------------KernelEx v4.5.2015.4 by jumper2015-01-14Resources: simplified version.h-based update methodPrep: FORWARD_TO support, better ordinal support, output cosmeticsKexcrt: debug messages upon NULL pointer to StrCpy()Kexbasen new: Gdi32:GdiAlphaBlend=>Msimg32:AlphaBlend Gdi32:GdiGradientFill=>Msimg32:GradientFill Gdi32:GdiTransparentBlt=>Msimg32:TransparentBltKexbases stubs: Kernel32:HeapQueryInformation=z5e ;Msvcr100 Ntdll:vDbgPrintExWithPrefix=t5 ;ReactOS 2014-06-11Now working without Kexstubs: Qupzilla 1.1.5 Slimboat 1.1.50 (with QtWebKit4 4.7.4 from Qupzilla 1.1.5) about 19 ReactOS DLLs---------------------------------------KernelEx v4.5.2015.3 by jumper2015-01-06kexcrt: StrCpy tweakKexbasen: FORWARD_TOKexbases stubs: Kernel32:*ActCtx* ;See Kexstubs topic Kernel32:*SList* ;Msvcr100---------------------------------------KernelEx v4.5.2 by jumper2014-12-27Resources: updated all product and file versions and resource languageskexcrt: fixed level3 warnings, 64-bit returns no longer truncated to 32-bitsKexbasen: DLL delay loading standardized for simplicity and robustness---------------------------------------KernelEx v4.5.1 by jumper2014-12-19Resources: updated product and file versionsKexbasen: added Oleaut32 templateKexbasen stubs: Oleaut32:442/RegisterTypeLibForUser, 443/UnRegisterTypeLibForUserKexbases fixes: Kernel32:VirtualProtect - don't let GCC apps disable write accessKexbases stubs: Shell32:727/SHGetImageList---------------------------------------KernelEx v4.5.0 by jumper2014-12-15rebuild of official 4.5.2 source---------------------------------------

foo2=foo1 foo3=foo2 nul=foo3

http://archive.org and http://archive.org/details/AllAmeri1936 still not working here (w/ or w/o JS). Twitter is the same. I see Wait... Connecting... Wait... Connecting... (OB1) or Connecting... Connected... Connecting... (FF2) in the status bar for all these sites but never https://:/. OffByOne still gives a Ubuntu server "HTTP/1.1 301 Moved Permanently" message. In OB1 http://details.archive.org/AllAmeri1936 gives what appears to be the home page, but also containing a page not found message. In FF2 it's just an empty "Page not found: We’re sorry, the page you have requested is not available" under a sparse menu bar. http://details.archive.org/movies is interesting. In summary, not working in FF2 for me!

The latest Gimp (2.8.4) has files with bogus internal build timestamps (1970, 2040, etc.). ImportPatcher.41 can now handle them. The "...local folder" option now defaults to ON when ImportPatcher launches and retains any user changes throughout the session. (In IP.40 it was auto-reset to ON for EXEs and OFF for DLLs at each file load.) WM_CTLCOLORSTATIC messages are now processed to (hopefully) prevent Revolutions Pack from "customizing" the background color of the controls: case WM_CTLCOLORSTATIC: SetBkMode ((HDC)wParam, TRANSPARENT); return (int)GetStockObject (LTGRAY_BRUSH);In case that doesn't work, one of the controls was resized so it should look better anyway.UPXing an app with standard parameters converts all dependencies but the first API of each DLL into delay-loads that can currently only be detected by profiling. In the case of FF10, GetProcessIoCounters is provided by kexbases so it's not a problem. In addition to the cursor changing to an hourglass during processing, the Analyze and Patch buttons and large text area are now disabled (grayed) as well. All return to normal to signify when processing is done. This is a very minor release, so major changes such as Ordinal support and better logging options will have to wait. Also: Invalid timestamps and file alignments are detected and reported in the log (if enabled), but not reported in the ini file/text box. Older features "Link to patched copies" and "Unbind broken bindings" have not been tested recently and may not work! ImportPatcher.41.7z

Was working just last week in FF2, but not now with or without javascript. Wayback machine still works. http://web.archive.org/ forwards to http://archive.org/web/ and works. Search of Internet Archive works with results page summaries, but links redirect to https and "The connection was reset" in FF2.

Profile it in Dependency Walker on both platforms and compare the debug output.

What file(s) did you set the compatibility mode on and what worked? If the driver didn't load, then it didn't work.... The network adapter is not working, correct?

Depends doesn't know about KernelEx and is generally incompatible with it. Because Depends is hooking various apis in order to profile, I think you need to set the Kex compatibility modes on Depends as well as on the app being profiled.

The second is easy: stretch the wallpaper in Irfanview, etc., and resave. For the first you might try using autoruns to find and disable the backup process. Or try deleting the *.da0 files and creating folders with the same names.

GetUserDefaultUILanguage is already in Kexbases. I had similar results with SM2.32, but no crashes. Definitely a work in progress.

ImportPatcher.40.7z ImportPatcher.40 Interactive gui with drag and dropCheckboxes: [_] Start dependency search in local folder [_] Test by loading (KernelEx) [_] Process delay imports [_] Walk dependencies [_] Unbind broken bindings [_] Show APIs for missing DLLs [_] Target Win9x (Subsystem version 4.0) [_] Save settings as .ini file [_] Create detailed .log file Buttons:[ Analyze only ] [ Patch ] [ Cancel ] Large editable results box Still TBD:- ordinal support - local walk - sub/folder processing - unUPX

SeaMonkey 2.32 loads with the following two stubs: [shell32.dll] SHEnumerateUnreadMailAccountsW=r0x80004001s4 ;SM2.32 SHSetUnreadMailCountW=r0x80004001s3 ;SM2.32 Using it now to post this reply! [Edit]And now editing the post in FF2 to get the line breaks to appear[/edit]

R8 will have api logging so we can begin tracking down many of these pesky issues. I'm using the apihook code with some improvements to make it easy to use. @Drugwash -- use threading settings from 4.5.2 if you need to. We are ignoring standard libraries and providing our own crt code. The .text warnings are because I .merged sections and used dosstub.exe for size in VC5 on those. Jusr remove the .merge and .stub linker options.

I use Media Player. QuickTime 4 also works well.

Yes, FF2..20 supports frames/iframes. Check browser.frames.enabled in about:config and any content-blocking plugins you might have. If you right-click inside the frame, you should see "This Frame >" in the context menu. Try some of the submenu options.

In the stack dump, I'm seeing way too many duplicate return addresses in Kernel32 (bff6....) and Kexbases (bfa4....). Looks like a stack leak is happening. The crash is in module FOXITREADER.EXE, so I'll need to get that to have any hope in determining what lead up to the crash. The last reference to Kexbases is bfa4d24b; what version are you using?

Release.7.7z KernelEx v4.5.2015.7 by jumper2015-01-22Kexbases bugfix: Kernel32:CreateProcessA (fix)Good new: all crashes are in KEXBASES.DLL at xxxx:bfa56000 in the CreateProcessA code I merged from 4.5.3beta. I've rewritten the fix including removing the offending line of code.

I'll repost my original #74 below. Original post #75 was Release 6 and changelog. I'll skip reposting this one. Thanks for the detailed reports. I'll do some debugging, then probably just revert to 5. I'll be more careful in using 4.5.3beta code! @MiKl - Are the Thunderbird 12.x and SeaMonkey 8 and 9 hangs a regression or progress compaired to 4.5.2 and recent updates? [answered!] @loblo - You may remember from the Kexstubs discussion last year that we were unable to find a working definition for FindActCtxSectionStringW; its mere presence caused some apps to crash. ReactOS' Ole32.dll needs it, however, and so may some other apps. So I've exported it as FindActCtxSectionStrin_W so ImportPatcher can easily mod a file into working. FindActCtxSectionGuid is not currently known to be needed and might also be dangerous, so I've treated it the same way. @tErmY - Thanks for testing. This is a BIG deal, however--regressions are not good. Can I correctly assume that FlashBlock 1.5.14.2 worked with v.4 and just not v.5? Please define "seems to have broken" and be very specific! If you are also using Kexstubs, please note the order of your "contents=" line in core.ini (mine reads "contents=std,Kstub822,kexbases,kexbasen") @Drugwash - I plan to do a cumulative delta source package at release 10. Hopefully you will be able to to build a full release for us at that time. Until then, don't feel obligated to manually merge and build every version. Maybe TmEE can help us with hosting or we can open a SourceForge account.

loblo's crash is in the XTREE code, while schwups's crash is in sstring, however, all roads are leading back to a failure in the 'new' constructor. 'new' is implemented via HeapAlloc, so I've made the heap growable and added the HEAP_GENERATE_EXCEPTIONS flag to HeapCreate. I've also tweaked my stack dump code so it should work better. And it should appear just once right before the main crash. schwups, please also include the standard error text as it contains the register values needed to fully trace the assembly code. For everyone not experiencing mprexe problems, I've added 78 new Ntdll function forwards, all needed by something I've tested in the last year (including IE8, KM74, and ReactOS). Release_5.7z 4_5_2015_5_source_diff.7z

Release.4 sources posted above. Release.5 will have 77 new Ntdll forwards to Msvcrt (from my stubs.ini!) About 19 dlls from ReactOS should now load. Processing the debug data now. Looks like we may multiple issues. Thanks, everyone! p.s. Cloudme should now load in Slimboat, but be prepared to wait if you have a slow connection or computer--they are both very slow!

Avoiding the problem is of no help in solving the problem! Please try to reproduce the "MPREXE caused an invalid page fault in module KERNELEX.DLL" problem with first release.0 and then release.4. The latter should detect the NULL pointer before the crash and display first a "storage #x is NULL" messagebox and then a deep stack dump messagebox. I need to know the value of #x and the contents of the stack dump (should be on the clipboard).

What is the exact MB and what bios setup options are available?