seahorser

Of course. I simply write a good alternative solution for people having similar major issues with crappy bios. If I post that workaround in another thread alone, no one is going to find it as a solution to USB connected to motherboard with bugged bios and WinSetupFromUSB can't do it this time.

I was find the fastest method possible that work on all computers both old, new and with bugged BIOS. I copy I386 to the root of hard disk's active partition (usually a second hard disk in the active partition that will be any partition, not only the first one.) Delete \ntldr form that partition (the partition which now has the i386 folder). If \ntldr is hidden, read only etc., attrib -R -S -H \ntldr then delete it or use the Total Commander or any other orthodox file manager On TC the Ctrl-T make a new tab, Ctrl-W to close a tab, Tab goes to other panel, Numeric +/- makes a set selection/deselection, * makes the selected one deselected, and the deselected selected, Insert selects/deselects a file or folder, Alt-left arrow to go back in history, Alt-right arrow to go forward in history, hidden file view Configuration->Options->Display, side by side compare by content two files or two folders eg. a backup copy and the working one. Ctrl-PageDown to go inside the ms packed two character and a bottom line extension like .ex_, .iso, .cab, .zip, .7z etc. by iso.wcx and Total7zip.wcx plugins which without TC to see them I have to call expand.exe on two separete folders for one file and for the other which had to work. Ctrl-D for folder bookmarks, separate editor with F4, internal TC file associations and many more copy \i386\setupldr.bin ..\ntldr copy \i386\ntdetect.com .. So on the root I now have setupldr.bin renamed as ntldr and ntdetect.com Inside \i386\txtsetup.sif replace the line [SetupData] as: [SetupData] BootPath = "\I386\" SetupSourceDevice = "\GLOBAL??\C:" and save it. copy \i386\txtsetup.sif .. So now on the root I have the modded txtsetup.sif Connect that hard disk to the computer which the real setup will take place and boot from it. P.S. The second hard disk should have at that partition boot record as NTLDR (the Process PBR button on bootice tool and press Install/Config button if needed). The MBR on that hard disk should be Windows NT 5.x/6.x at which the bootice came handy because I was used a backup taken place earlier of the MBR and any PBRs to revert that disk to the previous state. (\ is the root of the drive)

Unpack wmp11 with 7zip and run wmfdist11.exe The mtppk12.exe (Media Transfer Protocol Porting Kit) and Microsoft MTP Device Driver 1.0.0.0 for XP cab are bs in my case.

Since I came from the XP where there is no roaming folder inside the appdata or LOCALAPPDATA variable (eg. the set console command (cmd) indicates that appdata points to C:\Documents and Settings\username\Application Data), now it's clear why they had to change that. It must be cache data that had different usage from the user settings. Probably XP wasn't able handle cache/user data that efficient on true roaming profiles which reside on a server. Wonder why they don't see that at the XP era. Anyway the overwhelming majority don't use domain server and roaming profiles.

I don't have roaming profiles but on a new user which belongs to administrators group, the AppData goes to C:\Users\username\AppData\Roaming. I think that it should be C:\Users\username\AppData\Local Which system settings affect that? Which one is the correct?

Dear Wunderbar98 where I download the libpng dos binary? I download some files from http://www.libpng.org/pub/png/libpng.html but I only find some source code for linux etc. and nothing like the dos .com or .exe executables.

winnt32.exe /unattend:[file produced by nlite, which shows partition selection in text mode setup when run from usb] and selecting the directory which will windows will install later etc. options from winnt32 wizard ends-up with the usual folders $WIN_NT$.~LS and $WIN_NT$.~BT. Then I modify the \$WIN_NT$.~BT\winnt.sif to include the [Unattended] and other sections which winnt32 removes from my original winnt.sif. I have Autopartition="0" flag at all times in [data] section. When the computer begin the text mode phase, it continue to install windows in C drive without to show any question or the text partition manager screen with the modify winnt.sif while by the originally made winnt.sif by winnt32.exe it show that screen but that file lacks all other options of my nlited unattend file. Is there a combination of settings which ignores the Autopartition="0" and bypass that screen, or, how to enable disk partition menu to show when text mode setup begins from $WIN_NT$.~LS and $WIN_NT$.~BT temporary hard disk folders?

I didn't say otherwise! It was revert to UP. BIOS reported one processor on any clean XP 2003 etc. installation I was trying, so both Device Manager and Task Manager shows what they think they find. Also I was try to force install the MP HAL (also try all the other options it has that menu) using F5 key on a new XP setup, but again windows shows one core. The problem solved by reloading the BIOS default settings (although the settings are the same). PS. Some years ago I was see the following peculiar problem: The PC doesn't POST and doesn't boot. Believing that the GPU had fail (from my previous experience with PCs I was thinking the capacitor plague issue), I replace it with another GPU and the PC POST/boot as usual. I then replace the new GPU with the old GPU and the PC, again, doesn't POST or boot. I was try many times add/remove memory DIMMs, changing PSU, keyboad etc. The PC by no god wasn't able to POST/boot with the old GPU. Then with the new GPU I reflash the BIOS from pure DOS. Then I install the old GPU again and the PC POST/boot as should be! Maybe that was the time it revert to UP, don't really know. Now I have the new GPU which is faster laying around because the old is fanless (although I had a small fan to keep it cooler in summer times).

If it is exists and it isn't something else, like BIOS rootkit which control the other core individually from the OS by side channel, it must be hidden. There isn't any setting like this anywhere in the BIOS tabs. I can take pictures of the BIOS if you don't believe me.

Problem solved by reload default BIOS settings (F9 in this m/b) and suddenly 2 processors appear in device manager. The previous BIOS settings and the "new" reloaded defaults are the same as I double check them before and after!!! Another mysterious aspect is that before the reload to defaults action, I had try to change the NoExecute memory protection and the Enhanced Halt State to Enable but then the XP stays forever in a blank screen so I have to press the hardware reset button. XP detect the reset and they boot to safe mode just fine, but then they won't boot to normal mode as long as BIOS NoExecute memory protection and Enhanced Halt State settings are enable. When I set them to disable XP boot as always with one processor available. Probably ACPI BIOS got "corrupted" as described by Mark Shuttleworth at wikipedia's ACPI article. Now with ACPI multiprocessor PC in the computer's device manager and the two performance columns in the task manager, wPrime benchmark needs half the time to finish when wPrime advanced settings is set to 2 threads. It even boot whatever the setting I have at NoExecute memory protection and Enhanced Halt State. After 4 days of waste research the conclusion is that the defaults BIOS values are not internally equal as those I had seen on the screen, although they appear the same.

Msconfig shows one CPU and the numproc value isn't in use (none of the options is in use). In fact I never use Msconfig. Occasionally I use autoruns. Also the new test installations I was saying in my first post didn't have any chance to run msconfig. This motherboard BIOS doesn't have the OS compatibility setting which had earlier motherboards. What you mean by reset the BIOS? Load default values or re-flash it again with the same BIOS version? If is that the case, that should mean that a very sophisticate BIOS level firmware is sitting there...

Device manager shows one processor. The CPU is Core 2 Duo. I don't think my BIOS has the ability to disable cores. BIOS has: Enhanced Halt State, Intel virtualization tech, CPU thermal throttling, Intel speedstep tech, No execute memory protection, Intel C-STATE options. I have try some combinations with enable and disable settings but windows still shows one processor on the Device Manager, whatever the Computer is ACPI multiprocessor PC or ACPI uniprocessor PC.

On the performance tab I remember that I see how many cores a CPU had by separate "CPU Usage History" columns (one graph for each core). Also on the View menu there is an item called "CPU History"->One Graph Per CPU. My old XP and any new installations I make for that matter, even with different Hardware Abstraction Layer than ACPI (by pressing the F5 key early at text mode setup), all show one graph at CPU Usage History where my CPU isn't a single core CPU. How I get Task Manager like this ? PS. I feel that I'm going loose on my memory :(

From my first post all this time I experience sudden lockups, freezes and reboots. Several times windows setup tell that "some xxxx line in txtsetup.sif is invalid", grub4dos dies after recognize usb etc. The same UFD with exactly the same files, sometimes boots to windows setup and others freeze etc on this old BIOS. When I remove some stuff using nlite it may go to the famous 7b error, but it never arrive to the screen which shows disk/partitions to choose for the installation to continue. It is not a hard disk controller driver issue either because the disk is IDE. To investigate further after usb --init I chainload to DOS IO.SYS. This time grub4dos provides me with D drive which is the UFD. Nor the DOS setup (winnt.exe), neither a simple copy of the UFD I386 folder to the internal IDE disk came with success. DOS copy reboot after it copies some 200MB of files. With debug 3 in the winsetup.lst I had found that grub4dos throws some warnings on find and geometry commands "MBR cylinders(xxxxx) is not equal to the BIOS one(xxxx). MBR total sectors(xxxxxxxxx) is greater that the BIOS one (xxxxxxxx). Some buggy BIOSes could hang when you access sectors exceeding the BIOS limit. Initial estimation blah-blah-blah, err=1 blah-blah" for the UFD, but it throws err=0 for the internal IDE. That same UFD throws err=0 on another newer PC at which both WinSetupFromUSB and the Rufus work fine to the end of the windows setup. Do you believe the "second thought" approach could work in my case?

I don't know at which UFD format scheme you refer to. My current UFD format is made from fbinst from within WinSetupFromUSB 1.9 using as format options chs zip force and import the grldr to the primary data area and also copied to the root of the UFD. I boot straight to grub4dos without any menus or scripts. Find command give me a list of: (ud) (fd0) (fd0,0) (hd0,0) (hd0,1) (hd0,2) (hd0,3) , map ( and TAB give me fd0 hd0 rd ud. geometry (fd0): drive 0x00(CHS): C/H/S=370/255/63, Sector Count/Size=5944050/512 partition num: 0, active, Filesystem type is fat32, partition type 0x0C Filesystem type is fb, using whole disk geometry (hd0): drive 0x80(LBA): C/H/S=38913/255/63, Sector count/Size 625137345/512 partition num: 0, active, NTFS, 0x07 partition num: 1, NTFS, 0x07 partition num: 2, NTFS, 0x07 partition num: 3, FAT32, 0x0C geometry (hd1): Error 21 selected disk does not exist geometry (hd2): Error 21 selected disk does not exist map (fd0) (hd1): floppies_orig=1, harddrives_orig=1, floppies_curr=1, harddrives_curr=2 map --floppies=0 map --harddrives=2 map --hook (no output) usb --init: Found 1 USB devices. Device Num: 0x82; { Now, form that point on (because of usb --init), geometry (fd0) reports the same numbers as above (drive 0x00(CHS): C/H/S=370/255/63, Sector Count/Size=5944050/512) Error 25 Disk read error. geometry (hd1) now reports Error 21 selected disk does not exist } geometry (hd0): the same as above drive 0x80(LBA) etc. geometry (hd2): drive 0x82(LBA) C/H/S=7539/255/63 sector count / size 121101568/512 partition num: 0, active, fat32, 0x0C find --set-root /WINSETUP/XPpSP3.ISO: (hd2,0) root: (hd2,0) Filesystem type is fat32, partition type 0x0C ls /WINSETUP/XPpSP3.ISO: XPpSP3.ISO map --mem /WINSETUP/XPpSP3.ISO (0xff) map --rehook root (0xff): Filesystem type is iso9660, using whole disk Volume Name is: "XPpSP3". root: (0xff) Filesystem type is iso9660, using whole disk Volume Name is: "XPpSP3". chainloader /I386/SETUPLDR.BIN: Will boot NTLDR from drive=0xff, partition=0xff(hidden sectors=0x0) boot: The system encountered an I/O error accessing multi(0)disk(0)rdisk(1)partition(1) Setup cannot continue. Press any key to exit. That error indicate I have to edit txtsetup.sif inside XPpSP3.ISO and change the rdisk(1) to rdisk(2). Then setup goes well until it is freeze on Mylex EXR2000 SCSI atapter. Again I remove all Mylex EXR2000 and dac2w2k entries from txtsetup.sif as I don't have any SCSI adapter. This time setup freeze at Kernel Security Provider.

Didn't try because the main script winsetup.lst needs RDSK variable. I set RDSK=1 in shfthd.bat as I already know that by usb --init one more device (the hd1) appears to grub4dos. The whole setup procedure from start to finish after my modifications to the shifthd.bat runs fine on the newer PC but on old BIOS (AOPEN AK86-L) it freezes when windows setup loads kernel debugger dll. I also try several e820cycles values within winsetup.lst and some grub4dos versions with fbinst without any success. I try all and some of your commands and I get hd2 (0x82) when execute usb --init, so I didn't proceed further. My modifications (mostly comment out to shifthd.bat) as I know that UFD is detected as fdd and later using usb --init became hd1 and the rest are the originals from winsetupfromusb 1.9. Is there any known issue with windows setup and/or WinSetupFromUSB when the setup freeze at kernel debugger dll? shifthd.bat original menu.lst original shifthd.bat original winsetup.lst

My BIOS doesn't have USB-HDD option and I can't able to to boot with USB-FDD option. Only USB-ZIP is working. drive 0x80(LBA): C/H/S 38913/255/63, sector count / size 625137345 / 512 partition num: 0, active, NTFS, 0x07 partition num: 1, NTFS, 0x07 partition num: 2, NTFS, 0x07 partition num: 3, FAT32, 0x0C When I use BOOTICE's UltraISO v2+ using boot code USB-ZIP+ v2, "hide boot part" to hidden with boot partition size 32MB FAT16 Grub4Dos (grldr), "Data partition" FAT32 with size the rest of space I get: drive 0x00(CHS): C/H/S 30020/64/63 sector count / size 121037824/512 partition num: 0, active, FAT16, 0x06 partition num: 3, unknown, 0x0C partition num: 3 is actually a FAT32, I don't know why grub4dos say it is unknown. In fact if I switch partitions 0 with 3 using BOOTICE, or Unhide the hidden partition, the grub4dos then report: num: 0, unknown, 0x0C num: 3, active, FAT16, 0x06 So, when the FAT16 partition is hidden only the partition ID is changing from 0x06 (unhide) to 0x16 (hidden) but the 2nd FAT32 partition is always reported as unknown where it is clearly formatted as FAT32. Then I format it with BOOTICE U.V2+ hide boot part to none, boot partition size the whole space as FAT32 (grldr), set PBR to grub4dos, copy the grldr to root and get: drive 0x00(CHS) 30036/255/63, sector count / size 121103360/512 partition num: 3, active, FAT32, 0x0C Filesystem type is FAT32, using whole disk. usb --init Found 1 USB devices. Device Num: 0x81; geometry (fd0) : Error 25 disk read error geometry (hd1) : drive 0x81(LBA) C/H/S=7539/255/63 sector count / size 121103616/512 partition num: 3, active, FAT32, 0x0C Then reformat with U V2+ hide boot part: hidden, play with Set accessible and unhide a bit then I get for (fd0): drive 0x00(CHS) 17/255/63, sector count / size 65536/512 partition num: 0, FAT32, 0x0C partition num: 3, active, FAT16, 0x16 Filesystem type is FAT16, using whole disk. This time I get the right type for the 2nd partition and I don't know why and don't know what that "using whole disk" means. using usb --init Found 1 USB devices. Device Num: 0x81; geometry (fd0) : Error 25 disk read error geometry (hd1) : drive 0x81(LBA) C/H/S=7539/255/63 sector count / size 121103616/512 partition num: 0, FAT32, 0x0C partition num: 3, active, FAT16, 0x16 On all cases the shifthd.bat hangs when it gets to the plop boot manager. Another issue just before it goes to execute the plop is when it is say "The highlighted menu entry will auto start after 10 seconds" but there isn't any menu entry so probably shifthd.bat has a bug. When I give usb --init and then root (hd1) it hangs. The winsetup.lst is the one from WinSetupFromUSB. I try to insert some usb --init, map (fd0) (hd1), map --hook etc. here and there but it hangs or goes to file not found. My latest attempt with only one partition indicates that WinSetupFromUSB has issues which are not depend to my initial setup which had two partitions.

At best all I can get are a bunch of registers and by hitting any key the screen displays several more pages of them. Then it hangs. What I have to do with those DX,BP,AX etc. registers? I try to follow that approach but I'm stuck in mud. I barely understand 30% of what they say. That is not step-by-step walk-though. He jump to conclusions without to present us how he got there. What commands gave to debug (I assume that he got boot from the internal hard disk to DOS then run debug.exe). How I can get the values for LLLL and SSSS?

That BIOS you describe is very "queer". M/B is AOPEN AK86-L Grub4dos (before usb --init) sees 4 FDD partitions (fd0,0) (fd0,1) etc. That thread is huge although I'll try it. 1): I had wrote in my first post I use version 1.9 2): I just try the 1.0 version. At fbinst I reformat the UFD by using chs + zip + force options then I drag-drop grldr 0.4.6a. Test it to see if it boot to grub4dos and is OK. Then I continue with "Add to USB disk" and the two advanced options mention earlier. The results are exactly the same. The script tries to shift the drive and fail (hang) on plop boot manager. Even Ctrl-Alt-Del does not respond. Do you think it is appropriate to try even earlier WinSetupFromUSB versions and/or try earlier grub4dos? If I boot from UFD straight to DOS (no grub4dos) and then start ghost.exe from A: it fails, but the same ghost executable runs fine from internal C: in pure DOS. If I boot straight from the internal C: disk to DOS then DOS doesn't see the UFD (eg. A:) although UFD has FAT32 partitions and its there. Maybe that issue is because UFD has partitions? Also the spfdisk.exe at start it states that UFD has overlapped partitions. Alas, spfdisk does not have the ability to switch to floppy mode and check partitions there.

My old BIOS has only USB-ZIP and USB-FDD and support USB2.0. It only boot from USB when my UFD is formatted with 2 partitions. The first one had to be FAT16 32MB hidden, and the second FAT32. I create successfully and boot those partitions either through BOOTICE using the U+V2 button, with dummy.sys driver and with WinSetupFromUSB v1.9 FBinstTool. I try to install XP from USB to the internal IDE disk which I know is recognized by WinXP SP3 official ISO disk, so I select the 1st option and check the "Don't check for and install grub4dos MBR" and "Do not copy and use DPMS" at the advanced options, as I already know my UDF can start the grldr and I don't need Driver Pack Mass Storage. I also try both grub4dos 0.4.5c and 0.4.6a versions. When grub4dos script try to load the plop boot manager it hangs. The same issue which hangs plop on this old BIOS happen on another new PC, alas that new PC can successfully install XP without any double partition tricks using the Rufus 2.18 way faster as it uses USB2.0 speeds. I modify the script to just load the plop and from plop's menu I select USB1.1 mode 2 which is work through out the whole process on both PC but it is very very very sloooooow. To try to solve this issue I give init --usb command which give me drive 81 for USB on old BIOS but then I can't access files through fdd device with ls. Is it possible shifthd.bat or other script has some bug when USB has two partitions (one is hidden)? I also try to boot that old BIOS PC with UFD straight through MS-DOS (himem.sys in config.sys and smartdrv.exe) and then start winnt.exe from the UFD drive, which DOS gave it letter C to begin with XP setup. The 1st step of the winnt.exe XP setup process goes very fast like USB2.0 speed, but it suddenly reboot after it process a bunch of files. I think that either I don't need plop but I don't know how to modify the grub4dos script to begin with usb --init command so whole setup can start from there, or I need to load some DOS USB driver on old BIOS in case of MS-DOS setup. Although by using DOS setup I end with an ugly installation in drive D. Please advise

When installing from C: which is a DOS FAT32 partition on the drive, the XP setup ends up to the D: drive letter as expected which is NTFS partition. How I get the C: drive letter (SystemDrive) on Windows XP which begin to install into the second partition (or into 3rd partition etc.)? Is it possible for the XP setup to use the C: drive letter when was used initially by the setup at any other stages like T-whatever? I want to have Windows in C: where the C is assigned to second partition (or 3rd etc.).