jaclaz

I don' t know it is possible that if the ETDSimpleUI.exe is called, also other files are called (and now they are missing). Or maybe the good guys @Elantech have the equivalent of REMming with a ";" the hardcoded entries and keep them all in the source, but at compile time they are not active anymore. If you now rename ETDSimpleUI.exe to (say) ETDnotfound.exe, what happens? If the behaviour is the same (Mouse properties open) it means that the reference to ETDSimpleUi.exe is just a "dummy" placeholder, at least in this build. Still it remains "queer" as the 11.4.14.1 driver is coming from Lenovo, so also the Asus_UI.exe should be disabled. After the above test, try renaming the ETDnotfound.exe to Lenovo.exe. What happens now? Another thing to try would be adding to the \Program Files\Elantech folder the ETDUI.cpl from Z11542 jaclaz

Hmmm. Can you try renaming back Asus_UI.exe to ETDSimpleUI.exe? In ETDCrtl.exe there are hardcoded (snippet from Bintext): It seems like the good guys @Elantech made a "one size fits all" executable that calls *all* OEM files, possibly on a "first come first served" basis. And in ETD.sys there are: jaclaz

You must be joking. It belongs to self-evident truths that Windows 2000 has been the best MS OS ever. this is the updated link to the MS-chess game rules I invented (JFYI): https://msfn.org/board/topic/155290-windows-8-deeper-impressions/?do=findComment&comment=996327 jaclaz

Well, you may happen to find here and there posts by Steve6375: https://msfn.org/board/profile/37276-steve6375/ Author of Easy2boot and also of RMprepUSB (guess where the RM comes from?). https://www.rmprepusb.com/ He may know, though - at first sight - the specs of actual "Nimbus" models were too low to run NT 4.00, maybe they were computers from RM but not anymore "Nimbus", most probably new models belonging to the newish "RM connect": https://en.wikipedia.org/wiki/RM_Education jaclaz

Sure, you just need to change (I already corrected the reference in the beginning to point to the correct strings) the values in the "Strings table" section at the very end of the .inf, currently: I believe that these: can be *anything* and are just suffixes to the "DriverVersion", i.e. (say): will result in: "ELAN Touchpad 11.4.99.9_X86_ETD0108" Please note how even the "original" 11.4.14.1 is completely arbitrary, as the .sys driver file is actually 11.0.0.31. Good. Yep, I believe they can be removed - if not all - in large part. What I would still do would be checking if re-adding the ETDMcpl.dll is useful (likely it has to do with the "control panel", possibly the added tab to mouse properties?) And of course once you have a "final" version, all the entries for files currently remmed out with ";" can be deleted. jaclaz P.S. To clear the note about ETDMcpl.dll, since the entry for it in the .inf is already there, it should be enough to copy to the \Program Files\Elantech\ folder the ETDMcpl.dll (from 11.4.14.1) and see if its presence changes something.

Ok, so I have a new claim. The attached ETD_XP.inf, "only" 300 KB in size can replace the ETD.inf in the ETD_new folder and will manage to install just fine the driver via Setup.exe in XP. jaclaz ETD_XP.zip

So, there is still a difference between installing the SAME set of files via Setup.exe and via dpinst.exe? The former works AND saves settings whilst the second works BUT NOT saves settings? It is a again "queer" but if it works, it works. Knowing which differences there are in the two different ways to install it would be useful (for a "next" similar issue) but not in any way *needed*. Now, there is only a - totally unneeded - further refining step, the ETD.inf is - now - what I would call an unbearable amount of bloat, almost 2 MB, To make a comparison, the original ETD.inf from (Lenovo) ETD.inf from 11.4.14.1 (actual ETD.sys 11.0.0.31) is 480 KB AND the one from the "Medion" driver is (actual ETD.sys 11.137.2.2) an interesting 9 KB (though this latter version has a coinstaller, that could change the game). So, without taking the "Medion" driver as example, it should be possible to reduce the .inf to the "original" 480 KB or less. Surely we could call this ETD_new an "XP only" driver and remove anything connected to Win8 and Win10, and - additionally - there are a huge amount of "GroupOptions" (or whatever) for programs that either do not exist at all in XP or that exist in versions that those added commands cannot "hook" properly. BTW, there are no "groupoptions" in the original 11.4.14.1 .inf, so it is very likely that they - whatever they are - only work in much later driver versions. Comparing these two .inf's (the modified 11.5.20.3 in use against the "Medion" one, which due to its size is the only "human parsable" one) I have a doubt about the ETDMcpl.dll , as it actually is the *only* file that is registered (in both the .inf's): So, it must be *doing something* , very likely that *something* you didn't notice till now. Side-side note about 7GIF, the idea (mine) is to not use cannons to shoot at flies, 7GIF (in my perverted mind ) is the perfect companion to the nice little tool I use to make these animated .gif's (JFYI): https://msfn.org/board/topic/171205-licecap-useful-screenshot-to-animated-gif-recorder/ the two, summed together, are less than the half the size of the (stupid) ETD.inf . jaclaz

P.S. and as a side-side note: http://www.xtreme-lab.net/7gif/en/index.html needs not to be actually installed, you only need to extract contents of the .zip in a folder. You can slow down, speed up reproduction besides pause and still frames, and it works just fine with the ETD_install.gif If you want I can make an animated .gif video of it running ETD_install.gif ... ... no, wait ... jaclaz

I don't know. This is up to you. The key point is that the Setup.exe from 11.4.14.1 in the new folder "runs just fine" (here) using a (slightly) modified ,inf file from 11.5.20.3 (with only a few filenames changed and the reference to ETD.cat removed). While you reported earlier (but maybe there was something else at hand, some leftover from previous attempts, the actual *need* to remove the reference to Etd.cat, whatever) that the only way to run it was renaming it as Setip.exe AND NOT touching in any way not only the .inf but also any of the files in the source folder. Until I found a way via devcon to add the "fake" *ETD0108 device, I had no way to see/test what happens as the Setup.exe or the dpinst.exe refused to run because they found no compatible hardware. After I could do the tests, my claim is [1]: 1) running the setup.exe from the ETD_new folder with my modified .inf installs the driver just fine and uses the files you already tested as working 2) whatever happens here - possibly after a reboot - i.e. whether the driver works at all and if it works it keeps the setting I cannot test nor know 3) running the ETDUn_inst.exe I can not as it "hooks" to the *whatever* real id of the device (and not to my "fake" *ETD108) 4) after having anyway cleaned manually: 5) running the dpinst.exe is EXACTLY the same as running setup.exe in #1 above 6) whatever happens here (same as #2 above) - possibly after a reboot - i.e. whether the driver works at all and if it works it keeps the setting I cannot test nor know 7) running the dpinst.exe /U <full path to the .inf> removes everything was installed and - possibly after a reboot - the system is in the same "clean" state as it was in #4 above, with the exception of the HKLM "Elantech" registry hey that needs to be deleted manually So, if I were you I would try first starting from #4 and only later, having checked that the uninstall in #7 is effective (or cleaning manually what remains) go back to #1. Of course if any of the above do not even start running, you need to find why. If they run, and hopefully install correctly, and even more hopefully they install correctly AND keep the settings, we don't need anything else, it would only be a matter of cleaning/reducing the .inf to make it a decent size, remove the stupid windows 8 and windows 10 stuff, etc. So, as long as you clean the two logs before, you can later extract the last part (starting from the last line after your cleaning of them or first line related to Elan) and state whether you made 1,2,3,4,5,6,7 or 4,5,6,7,0 ,1,2,3. jaclaz [1] actually I also have the additional claim that since it works here, if it doesn't there it is your fault

Well, it plays just fine in Opera 12.something series, Presto. It plays fine in chrome (Iron 49.0.2623.112), though it is very slow to load. It plays fine in Basilisk/Serpent 2019.08.16 It plays fine in Firefox 47.1 It is as widely compatible as an animated .gif can be made. I am not sure to have understood what he meant, it is entirely possible that if you have multiple etd_whatever files in the driverstore the stupid windows attempts to use one before another, but that shouldn't affect the device name . Anyway doing a registry wide search on the Registry with Registrar Lite (Reglite) or use any other registry tool that has search with multiple results. for strings *like* Elan Elantech ETDCtrl.exe ETD.sys etc. should clear easily if this is the case. jaclaz

Well, you should recreate a new "source" directory identical to the one you can see on top left of the video/gif. The files are the same that you have in the \Program Files\Elantech folder plus: dpinst.exe dpinst.xml a couple Eula files (I left in it only Italian and English but you can re-add the whole lot or add none) the ETD.sys (from 11.4.14.1) the modified ETD.inf (that originates from 11.5.20.3 and where I only removed the .cat reference and changed files to reflect your "final" \Program Files\Elantech folder Here is anyway the detailed dir.txt of it. BUT BEFORE doing any attempt, you need to clean the system which namely means: 1) uninstall from device manager the touchpad device 2) verify that you have no ETD.sys file in \Windows\drivers\ 3) verify that you have no OEMnn.inf and OEMnn.PNF file in \Windows\inf related to Elantech installs 4) verify that you have no etd_inserthereacrazylongstring folder in \WINDOWS\system32\DRVSTORE\ 5) verify that you have not any Elantech related item in Add/Remove programs 6) delete the Registry HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Elantech key and all its subkeys AND "clean" the SetupApi.log and DPINST.log, Then you need to run in couples, i.e. you first run EITHER the Setup/setip.exe from the unmodified 11.5.20.3 folder OR run from the same folder the dpinst.exe,. Following, you run the ETDUn_inst.exe OR the dpinst.exe /U from the same folder. Then you try doing the same from the "ETD_new" folder, after having verified that all 6 points above check (as long as install and uninstall the same .inf and folder it shoudl work nicely). If you setup your desktop similar to what I showed you in the video, you will have most relevant places of the 6 above under control in "real time". Unfortunately you will probably need intermediate reboots to check what happens after a driver install or uninstall, so you will have to re-compose the desktop each time, I used 7-zip because it is more compact and updates faster when copying/deleting files, but you can use a set of explorer windows. or a batch in another command prompt *like* (I am using C: you will need to use D: of course) : DIR C:\Windows\drivers\ETD.sys DIR C:\Windows\inf\oem15.* <- in my case it is OEM15 adapt to your situation DIR C:\WINDOWS\system32\DRVSTORE\etd_* Get: https://docs.microsoft.com/en-us/sysinternals/downloads/regjump regjump HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Elantech jaclaz

OK, it doesn't really matter, devcon is a "strange beast", and I am not at all an expert in its use. Anyway, further update. The Setup (or Setip) .exe seems (a lot) like a wrapper around DPINST.EXE. I.e. all the Setup does with (from a command prompt opened in a directory one level higher than the one where the driver files are): .\x86\Setup.exe can be replicated with the command: .\x86\dpinst.exe /C (the /C is optional, what actually happens is logged anyway in C:\Windows\DPINST.LOG) Here the ETDUn_inst.exe does not work (as I need to force with devcon the device *ETD0108, whilst the ETDUn_inst is probably intended to uninstal the actual "real" device, i.e. the "ACPI\ETD0108\4&1B919E6A&0" you have). Anyway the following command: .\x86\dpinst.exe /U C:\appoggio\Daveh\ETD_new\x86\ETD.inf /D /C Seemingly does all that is needed (the /C is optional as seen above BUT - for some reasons the path to the .inf neeeds to be a FULL path and you'll need to change it according to your situation), namely: 1) it removes the device 2) it removes the etd_inserthereacrazylongstring folder from C:\WINDOWS\system32\DRVSTORE\ 3) it deletes the the OEMnn.inf and OEMnn.PNF from C:\WINDOWS\inf\ 4) it deletes the ETD.SYS from C:\WINDOWS\system32\drivers\ 5) it removes files from C:\Program Files\Elantech\ 6) it removes the Elantech entry from Add/Removeprograms You should open the: C:\Windows\SetupApi.log and the: C:\Windows\DPINST.log with a text editor and remove all the last lines related to Elantech related attempts, so that it will be easier to try the above DPINST commands and compare the log entries with the ones created when Setup.exe and ETDUn_inst.exe are run. jaclaz

No, it wasn't, not Cluster Size, not in BIOS. The above sentence (presumably aimed to explain what you meant) makes no sense whatsoever, and in the following the only thing I can understand is "I apologize for my apparent lack of information": There is no apparent lack of information in your posts, you simply state things that are not accurate, probably it is only a naming issue, you call things with NOT their names. It is perfectly fine of course, but if your intention is to share your experiences (and expect readers to understand what you report) you need to be accurate and use the names of things as they are used by all the rest of the people, particularly if you want to be effective/concise. jaclaz

Ok, there must be *something* on your install (maybe SP3 changes something? I doubt it) or setting or some other program (like antivirus/anti-whatever) that prevents the running on your system of a Setup.exe with unsigned drivers (or possibly a Group policy, I have NO idea what could it be). Here Setup.exe (from 11.4.14.1) runs just fine, both with the original .inf from 11.5.20.3 and with it modified (removed the .cat file, changed files to the ones you are actually using, etc.). Of course I cannot say if the driver works or not. At first sight the only difference I can find is that devcon (as well as .inf install from device manager "Have disk") does not add anything to the driverstore. Doing a few random tests, it does seems like ,inf installing checks date/version of drivers and installs "whatever it seems better" (but no real prove of this). Anyway, here everything works as it should (as far as I can test it). Find attached a modified 11.5.20.3 .inf and a (quick and half-@§§ed as always) small video (actually animated .gif) showing you what happens here. Let me know when you got the file so that I can remove it from attachments. (EDIT: Ok, removed) jaclaz

No, "basic" electric tape can "resist" those tempoerature, but it will become a mess (the adhesive part). You want so-called Kapton tape (for permanent protection). https://en.wikipedia.org/wiki/Kapton If the issue is thermal paste smearing, put less paste or use a masking with common painter/bodywork paper tape (and remove it once it's done). jaclaz

Good. The other two commands should be: devcon.exe driverfiles "ACPI\ETD0108\4&1B919E6A&0" and devcon.exe drivernodes "ACPI\ETD0108\4&1B919E6A&0" jaclaz

No, there is something that simply does not make sense. Here I am trying to replicate the situation, using devcon too create the *ETD0108 node and here the Setip.exe (setup.exe from 11.4.14.1 renamed) has no issues with an ETD.inf from 11.5.20.3 modified and without the .cat file. Give me some time I'll see if I can come out with something that is hopefully reproducible on the real hardware on your machine. Can you confirm that - besides the .inf - the only *needed* files (all coming from 11.4.14.1) are the ones I listed earlier?: ASUS_UI.exe ETDApix.dll ETDCmds.dll ETDCtrl.exe ETDFavorite.dll ETDUn_inst.exe ETDUninst.dll or, maybe better, can you make a DIR "C:\Program Files\Elantech\*.*">C:\elantech.dir of your system (with the driver "fully" installed and working)? I lost track of which "other" files (if any) you have there (i.e. the .avi's). BTW, also can you get devcon.exe from here (so that we will be using the same version): https://msfn.org/board/topic/173201-gavottes-ramdisk-automation-package/?do=findComment&comment=1091396 and run these commands (and post results): devcon.exe listclass mouse devcon.exe hwids =mouse devcon.exe stack =mouse devcon.exe driverfiles *mouse* devcon.exe driverfiles "HID\Vid_13ee&Pid_0003&Rev_0000" <- here replace with the *whatever* is the ID of the Elantecg trackpad devcon.exe drivernodes *mouse* devcon.exe drivernodes "HID\Vid_13ee&Pid_0003&Rev_0000" <- here replace with the *whatever* is the ID of the Elantecg trackpad devcon findall =mouse devcon classfilter mouse upper jaclaz

Hmmm. It's queer. So the .inf installing must be changing *something else* outside HKLM\SYSTEM\CurrentControlSet\Control\Elantech\ and that something is part of the "magic" (otherwise restoring the saved .regs should have brought the situation to an "as before" state). BUT the .inf installing does not change anything when over-installing on the working/saving settings driver. It makes no sense. jaclaz

And NO (again). Cluster size is set nowhere "in BIOS", it is one of the parameters of the file system, cluster size is a documented parameter and MS has a KB on its defaults (that shouldn't be changed unless there is some reason) for NT based systems: https://support.microsoft.com/en-us/help/140365/default-cluster-size-for-ntfs-fat-and-exfat Win9x/Me format can and will use 32K sector clusters on "large" volumes (>32 GB), similar to NT 3.51 in the above. jaclaz

Good. So you can try deleting the : HKLM\SYSTEM\CurrentControlSet\Control\Elantech\OtherSetting what happens? If you can still "keep" the settings, try deleting the whole: HKLM\SYSTEM\CurrentControlSet\Control\Elantech\ before or later likely either the driver won't work anymore or however you won't be able to change settings. At this point try again "overinstalling" via the .inf. What happens? If by any chance the driver works again and keep settings, the "magic" is not in HKLM\SYSTEM\CurrentControlSet\Control\Elantech\, but if it doesnt', make the second set of files: jaclaz

Yep, but it is strange - at first sight - that the files in the driverstore were added/changed, unless you also replaced the files there, that folder should be - I believe - only modified by the installing (either via. inf or via Setup/setip) From the results it seem like apart from a few files - corresponding to the "needed" files you found out earlier, that were "changed", all the other ones were "added". Anyway, right now you have a working driver that also "keeps" settings, right? Make a copy (export to .reg) of the whole: HKLM\SYSTEM\CurrentControlSet\Control\Elantech\ say to Elantech_saving_OK.reg then make another copy (export to .reg) of only the: HKLM\SYSTEM\CurrentControlSet\Control\Elantech\OtherSetting say to Elantech_OtherSetting_saving_OK.reg Try re-installing the driver via the .inf WITHOUT uninstalling it via ETDUn_inst.exe NOR in ANY OTHER way (i.e. from the current condition of working and saving setting driver just go to device manager and update driver/have disk), then replace the usual files from 11.4.14.1, that from what I can understand are: ASUS_UI.exe ETDApix.dll ETDCmds.dll ETDCtrl.exe ETDFavorite.dll ETDUn_inst.exe ETDUninst.dll and the ETD.sys Two possible results: 1) the driver having been "overinstalled" has still the capability to save settings (after a reboot, etc.) 2) the driver having been "overinstalled" has lost the capability to save settings If #1 it should definitely mean that there is nothing in the .inf that "removes the magic" If #2 it should definitely mean that there is something in the Setup/setip.exe, that "creates the magic" at install time and that is outside "normal" Registry. Still if #2, the counter-test is to: Make a new copy (export to .reg) of the whole: HKLM\SYSTEM\CurrentControlSet\Control\Elantech\ say to Elantech_saving_NOT.reg then make another new copy (export to .reg) of only the: HKLM\SYSTEM\CurrentControlSet\Control\Elantech\OtherSetting say to Elantech_OtherSetting_saving_OK.reg then, as a first step re-import the previous Elantech_OtherSetting_saving_OK.reg and see what happens (a reboot/logout/login might be necessary). If the driver still doesn't save settings, re-import the previous Elantech_saving_OK.reg and see what happens (a reboot/logout/login might be necessary). If the driver still doesn't save settings, the "magic" is not in HKLM\SYSTEM\CurrentControlSet\Control\Elantech\ and must be looked for elsewhere. jaclaz

Hmmm. I find it hard to believe that nothing changed in the \%program files%\Elantech\ folder if there were changes in: D:\WINDOWS\System32\DrvStore\etd_E13DCF8481E89626AF6C254EEF671D041882CD85\ Anyway, no particular changes in the Registry (that make sense). Attached is a "filtered" version of the file you posted. Unless I missed something (which is well possible) the added keys/values are: and the changed ones are: which at least from their name don't seem like relevant. jaclaz 1st_attempt_filtered.txt

I am not sure to understand what you are after. That's it, the system needs HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\atapi\Parameters\EnableBigLba to 1. Then there may be different ways to implement it, this (using a modified SETUPREG.HIV) is the "best" one in that it is simple and enables BIGLBA also in the TXT setup part: The batch by os2fan2: https://msfn.org/board/topic/75713-48-bit-lba-on-win2k-setup/?do=findComment&comment=590337 is one method, but you could also load the hive in regedit and add/modify the key manually or use offline registry editing, etc. jaclaz

Sure , I also just bought a new Aston Martin should my old Opel Vectra (Vauxhall for you ) fail to start tomorrow morning . jaclaz

Sure you can, using the Setip approach (as you already did and successfully) Yes, please ignore #4. To re-cap: 1) at some time you installed from the Z115203-cab in XP with the .inf and (after you replaced the "key" files with those from tp118w7) the driver worked BUT changed settings were "volatile". 2) then, at some other time you installed the Z115203-cab in XP using the "added" Setup.exe from tp118W7 renamed to Setip.exe and (after you replaced the "key" files with those from tp118w7) the driver worked AND changed settings BECAME persistent. So, since the replacing part was the same in both cases you don't really need to do the replacing of files (which is the same in both cases), you only need to 1) install the 115203-cab in XP via the .inf 2) install the 115203-cab in XP via the Setip approach If any difference is found by the tracking/tracing, that could be enough. There is still the possibility that you need to do the actual replacing AND a reboot for the *magic* to happen, but if it is already found in the above simpler test, there is no need to go all the way through the replacing and reboot. jaclaz