jaclaz

I still don't understand what you did (nor what you wanted to obtain ), what (the heck) does it mean the "before" and "after" and the half commented commands? An insmod is an installable module, you only need to install it once per session, but it is pretty much binary, either it is installed or is it not. You should NOT use menu.lst entries, but rather commands on command line, as they will give you some feedback. (and report that feedback as well as the commands actualy issued) If the set of commands you are using are these ones: They make no sense whatsoever. 1) you map an existing (compressed) image residing on floppy to current hard disk (which may be *whatever* depending on what has been run before and on the setup) and the result could be anything (and the contary of it) depending on the contents of the image 2) you hook the mapping 3) you set root to the floppy 4) you chainload the OS file on floppy In those the image is simply mapped and is not changed by any of the commands you listed, and BTW you never access it from grub4dos. As it is now it seems to me like the classic recipe for making a vodka filled watermelon[1] in the shorter version if you have not 12-24 h time jaclaz [1] https://www.tablespoon.com/recipes/vodka-infused-watermelon/cbc4b24f-2748-4f52-af7b-b804cbf070d7 Follow steps 1-3, then, at step 4, just drink the vodka.

Without knowing the exact set of commands you used is hard to say anything about the 06 and 0E type of partition, there may be different reasons why that happens, including the pre-existing values. Latest versions of Partnew have been amended as said here: https://msfn.org/board/topic/180597-exploring-setup-p-isdetectbus-and-grub4dos-8½-times-win98se-on-usb-drive/?do=findComment&comment=1173227 to allow to choose (if needed) between 06 and 0E. The F0 (floppy) vs. F8 (hard disk media) is likely to be "hardcoded" in the FAT tool (but of course can be later corrected). The geometry issues may depend: BPB_HiddSec zero depend on the exact procedure sey of commands you used (again it can be later corrected) BPB_SecPerClus is "queer", while it is possible that grubdos uses a sector per cluster table different from DOS, it is unlikely, at least for "normal" or "common" volume sizes BPB_FATSz16 I am not sure to what field this corresponds to? There could be a more generic issue (unless it is explicited in the map command) with the geometry of the device, which may (or may not) be connected to the above. The catedit script is a simple tool to more easily write hex values to given addresses, and of course it can be used to create a FAT32 formatted drive. Essentially what the format command does on DOS (and also on later windows) for FAT 12/16/32[1] is: 1) copy a hardcoded bootsector (containing the bootcode) from itself to the target 2) make some (relatively simple but not trivial) calculations to determine which values to write in some fields of the bootsector and determine the size of each FAT 3) write, just after the reserved sectors a bunch of (fixed value) bytes and write them again at the end of the first FAT #1 above (and complications of it, see note [1] can be done with the internal dd of grub4dos #2 is ? and then catedit can be used to write the calculated values #3 can be done no problem via dd or via catedit So what remains to be written explicited is "only" the calculation part, which may in itself be "easy" for a given set of sizes/geometries or more complex if the idea is that of a "generic" formatting tool. capable of formatting *any* volume. I made some time ago a couple spreadsheets and a small batch (though it won't run on DOS) that may be of inspiration to you for FAT 12/16, start around here: https://msfn.org/board/topic/152097-on-superfloppies-and-their-images/?do=findComment&comment=988732 you might need to read more of the topic to understand the genesis of the spreadsheet, however. Using the same approach/method for FAT32 is entirely possible. jaclaz [1] FAT32 is a tad bit more complicated as the bootsector code is multi-sector (but the sector past the first one are "static" in content if I recall correctly) and there is a backup copy of the bootsector(s)

@FranceBB You see? @KawaiiBunnyRU It is about "photographic proof", not "testimonials", see the original thread: This said, I have seen things you people wouldn't believe. Diskless thin clients running custom OSes. I watched terminals failing to connect to mainframes with blinking cursors. All those moments ... https://en.wikipedia.org/wiki/Tears_in_rain_monologue jaclaz

JFYI, grub4dos has an external module called FAT that allows to crate a FAT filesystem directly from grub4dos, part of grubutils: https://github.com/chenall/grubutils/releases/tag/2015-02-15 Creating the MBR partition table can be done manually or with internal command partnew and writing the Magic Bytes at the end of it can be done manually or using the small Catedit batch: http://reboot.pro/topic/18783-release-cateditg4b-script-to-write-hex-values/ jaclaz

NO, it is NOT, never has been and never wiil be. It is in the root directory of the active partition (the volume that normally gets drive letter C:\ ), along with NTLDR and NTDETECT.COM, Just like NTLDR and NTDETECT.COM it may be not visible (since it is a system file) in Explorer unless you set it to show hidden and system files) and additionally it nay be set as read only, but is there alright. I wonder however how you can paste into a file that you cannot find in a hex editor. jaclaz

To be picky (as I am, BTW) no, that (nice) thread was started no less than three years after EOS, not three days, this may be a relevant difference. Yep, that's why waiting a few years might be needed to have more meaningful entries, still the photo taken about the scaremongering since the 14th Jan is nice, particularly because it is the same day a (really serious) insecurity in Windows 10 was made public. jaclaz

I believe you might be confusing the number of copies (or amount of redundancy) which is 2 in a two way mirror, with the number of drives in the drive pool (3 or 4 in your case). This latter, from what I understand, is not connected with the former. Basically your (say) 500 GB of data is first "divided" into 256 MB units called "slabs", which are "doubled" (i.e. "mirrored") and then "dispersed" on the disks that part of the pool. You do not have anymore 500 GB of data, but rather some 512,000/256=2,000 * 2 = 4,000 "slabs" of which "roughly" are written 1,000 on each of the 4 disks. When you remove one disk, the whole stuff is (or should be) re-assembled making a new copy of each slab that was on the removed disk, and again dispersing them on the available disks, so each of the three disks should have about 1,333 slabs. Compare with: https://techcommunity.microsoft.com/t5/storage-at-microsoft/deep-dive-the-storage-pool-in-storage-spaces-direct/ba-p/425959 jaclaz

.Well, it depends a lot on how (EXACTLY) you wrote/modified the BOOT.INI. (You didn't copy and paste from the reference post, did you?) You need (obviously) a hex editor or viewer to check the contents of the file, each and every text editor (let alone the stupid board software) may render the non-printable character 0x1B (i.e. the actual "Esc" in "Escape sequence"): http://ascii-table.com/ansi-escape-sequences.php in a different way, besides encoding and what not, and if the text editor actually allows to insert that special chaaracter. jaclaz

To hopefully clarify the concept (to both you and Caliber), seemingly this TuneUp thingy: 1) takes as source a "default" (and I presume with path/name hardcoded into it) XP Kernel ntoskrnl.exe 2) makes a copy of it (as "TUKernel.exe") 3) patches this copy changing in it *something* related to boot screen, allowing to show a different one 4) adds a boot.ini entry with Kernel=TUKernel.exe It is not that the boot.ini modification is in any way "different" to load the Russian patch, the daniel_k version or the TuneUp version it is a provision there since good ol' NT 4.0 (and probably also 3.x) to allow for using different kernels, originally implemented for testing/debugging purposes. You cannot normally rename ntkrnpae.exe to ntoskrnl.exe because (I presume) it is a system protected file, so the alternative is to either disable the file protection (temporarily or "forever") or try to patch/modify the TuneUp thingy to point to ntkrnpae.exe instead of ntoskrnl.exe as source file of the TuneUp tool, or at least this is what I suggested as a possible workaround to solve that problem (and that problem only). I explained why I believe that the whole thing is futile and how - if I got the final scope - (IMHO) futile as well there may be other ways to obtain the final result (a different, custom bootscreen), but if Caliber wants to use the daniel_k patch, why not? Now, for no apparent reason something as well futile, but that at least it might have some uses (as an example for visually impaired XP users), colours in BOOT.INI thanks to ANSI (and fujianabc): http://reboot.pro/topic/10122-colorful-boot-menu-with-ansi-escape-code-in-bootini/ jaclaz

Besides it being IMNSHO opinion a perfect example of "looking for troubles" (actually "senselessly looking for troubles"), if we generalize the question to "How can I replace a "system" file which is protected by WFP (Windows File Protection) sometimes called "SFC"?", then the answer is "You shouldn't and you cannot (according by Microsoft)." BUT, there are a couple good ol' tricks possible making use of a "hacked" SFC.DLL or (IMHO simpler/cleaner) an empty SFCFILES.DLL: https://web.archive.org/web/20110719154642/http://www.vorck.com/windows/software.html Will either method work for a kernel file? Who knows? Try and see, but be warned that it is entirely possible that the system will be botched, so have handy a TESTED recovery/restore solution for the OS. I would rather check if - by any chance - in your TuneUp program (whatever it is) the path and/or name of the "source" kernel is hardcoded and hexedit it to point to the 4 GB patched kernel. As a side note, IF the question is a different one, i.e. "How to change XP boot screen?" then there are different answers, possibly around here: http://reboot.pro/topic/291-winbuilder-051-beta-4/page-3#entry2585 http://reboot.pro/topic/339-boot-logo-for-winbuilders-pe-build-post-ideas-here/ or maybe try another tool that works differently (and at least in theory "better"): Start looking for it around here: https://web.archive.org/web/20060715224654/http://www.stardock.com/products/bootskin/ (to have an idea of the story behind it) Then, this is current: https://www.stardock.com/products/bootskin/downloads.asp jaclaz

Can you expand on it? I don't understand what the question is. What is the 1, what is the 3 , what (the heck) do you mean by "mix up the kernel"? jaclaz

NO. Meaning that sector size and cluster size are "independent". The sector size "belongs" to the disk, the cluster size "belongs" to the file system. Here the disk is exposed as having a logical sector size of 2048 bytes, that allows you to have a 1 sector/cluster cluster size, i.e. 2048, but also a multiple of it, i.e. as an example 2 sectors/cluster size i.e. the "normal" 4096 bytes cluster (most current NTFS filesystem of not trivial size will have a cluster size of 8 sectors * 512 bytes=4096 or 1 sector * 4096 bytes=4096). The 2.2 Tb original limit depends on the 32-bit size of the MBR partition table that sets at 2^32-1=4,294,967,295 the max number of sectors accessible (though there is a work around to roughly double that with a "special" partitioning) so the limit is: 512 bytes sector 4,294,967,295*512=2,199,023,255,040 <- this is what is normally exposed by 512 bytes sectored disk or by AF disks (4096/512) but not by "Native 4K" disks 1024 bytes sector 4,294,967,295*1024=4,398,046,510,080 2048 bytes sector 4,294,967,295*2048=8,796,093,020,160 <- this is what this driver does with a 5 TB disk 4096 bytes sector 4,294,967,295*4096=17,592,186,040,320 <- this is what a number of USB external hard disk controller do to allow larger disk sizes. jaclaz

Yep, the issue is known for previous versions of Windows, evidently the (stupid) Windows 10 has not been fixed, see: https://support.microsoft.com/en-sg/help/937251/disk-drive-numbers-may-not-correspond-to-the-sata-channel-numbers-when in practice it depends on how fast the device is in communicating with the plug 'n play subsytem, so that you may (or may be not) able to change order switching (SATA) ports. Drive letters are assigned to drives (that do not mean disk drives, but rather partitions or more exactly volumes). On BIOS, disks are normally partitioned MBR style, so you can have both primary partitions that are also volumes or extended partitions that can contain one or more volumes. On UEFI, disks are normally partitioned GPT style, so you have only primary partitions that are also volumes. Drive letters are assigned in Windows automatically (unless intentionally/manually assigned) to volumes along a set of "rules" that depend (in part) on the order of the disks (the actual diks drives and that get rather complicated where extended partitions and logical volumes are involved), but once they are assigned they are "static" unless something changes, the drive letter is assigned to the volume (even more properly to a disk extent) based on the Disk Signature and the offset (and size) of the volume, the info is stored in the Registry and is independent from the disk order in DIsk Manager or Diskpart. So the issue usually comes not with drive letters assigned to volumes on the internal disks, that are pretty much "static" (physically), but if you have disks that you plan to hot-swap (or hot disconnect/re-connect) there could be issues, though It is difficult to say what can happens if, If you use USB devices and/or network mappings, this may be of use: https://www.uwe-sieber.de/usbdlm_e.html jaclaz

I would also like to see a picture of the moron rectius "person" that upgraded a 1 GB PC to windows 10, but unfortunately there are all the Privacy Laws, GDPR and what not, so that is not possible . jaclaz

No. The problem lies in the fact that noone has a clear idea of what testing or validation is, and programs are written, more often than not by lazy/sloppy programmers, but also by good, attentive and knowledgeable ones, that do not know anything about the actual processes involved (because there are always one or more middlemen involved, so that the actual final customer, the guy/gal that actually has to use the program or maintain/service the machine never communicates directly with them and viceversa). The program is anyway always late and what is released is invariably - in the best cases - an early beta, and all the bug finding and reporting is demandated to the client, that normally has no idea on how this is done (remember, the client is the one that pays the money to buy something that he has not the capabilities to develop in-house). It is peculiar (and I believe rather unique to the (largely bad) software industry to provide "solutions" that are not working (or not fully working) AND to put the blame on the clients (you know, again, the ones that actully pay for the service) OR asking them to lose hours, days, weeks into troubleshooting the crap they produced. More loosely it is called "professionalism" and most of the software industry has no idea of what it is. Imagine you go to your barber and he after taking your (good) money for a haircut gives you a pair of scissors telling you "If in the next few days you see some tuft sticking out of your head just cut it off" or - which is what the large software companies do - "You also owe me 100 bucks for a week of assistance, if you see some tuft sticking out you just drive here and I will cut it", or - alternatively - "In case you notice some hairs sticking out call me an I will be happy for a mere 200 bucks to come to your house and cut them level with the rest". Then you go to get your car from the mechanic and he tells you "Ok, I changed the tires, I am pretty sure that I did tighten the nuts properly, should you lose a wheel, don't worry, it is covered by warranty, you just bring me back the old wheel and I will fit a new tire on it for free". Then you go to a restaurant and the cook tells you "The food I prepare is fine, should you by any chance get poisoned i will reimburse you part of the bill". jaclaz

There are more than these, including a few Commercial tools that claim to be able to convert back from dynamic to basic, among them there is partition Wizard 4.2 (free version): https://www.sevenforums.com/tutorials/26829-convert-dynamic-disk-basic-disk.html But I wouldn't trust any automated method and would like to see what is actually on the device. The differences between a (MBR style) dynamic disk and a basic disk (single device, i.e. simple not spanned) are trivial and can be fixed in no time: 1) partition ID in the MBR (42 vs 07) 2) excess sectors at the end of the disk (LDM database) so a quick edit with TinyHexer or similar wouldn't be a problem, but here there is the issue of the Disk seen as invalid (before the "missing" volume). I suspect that here the issue could be that of a conflict between the dynamic disk data and the GPT (second copy of) partition table, but the RAID reference is perplexing. So before attempting a cure, I would like to diagnose the illness. jaclaz

You can use Qemu for running Windows 98. Qemu (there are Windows ports) is however a bit complex to configure/set the right command line parameters. You can try (but no guarantee it will work on your Windows 10) the good ol' Qemu Manager that can still be obtained via the Wayback Machine, Qemu Manager offers a graphical (and very convenient) interface to the VM: http://web.archive.org/web/20120607071209/http://www.davereyn.co.uk/download.htm http://web.archive.org/web/20120506095037/http://www.davereyn.co.uk/qem/setupqemuk70.exe or http://web.archive.org/web/20120506093950/http://www.davereyn.co.uk/qem/qman70.zip If you are OK with a command line only tool, you can get recent Qemu ports here: https://qemu.weilnetz.de/ jaclaz

WinsetupfromUSB is basically meant to setup winnt from USB, NOT boot an XP from USB and though it can be used to install an XP on a USB it is a dual stage procedure, see FAQ #7, point #4 here: http://www.winsetupfromusb.com/faq/ You can however use it to boot a PE (either a PE 1.x, i.e. based on XP, like a BartPE, or a later version), compare with FAQ #1 here: http://www.winsetupfromusb.com/faq/ It is not clear what you mean by a "XP PE", a PE 1.x would normally use SETUPLDR.BIN (and not NTLDR) jaclaz

@cdob I wouldn't worry about that as well. In the worst case (mainboard failure), to recover/access data one can always [1]: 1) make a GPT partition table 2) correct sector/cluster size in the NTFS PBR 3) access the volume from a later OS that undestands GPT or use the Paragon GPT driver on XP (if applicable, would need to be tested) or : 1) use the disk in an external USB case that does 4096 bytes translation 2) correct the MBR and NTFS PBR (more or less what is done in the mentioned thread about switching between 512 and 4096 bytes sector, that approach needs a complicated partitioning scheme to allow for the automatical switch, but manually it is not an issue at all to change a few bytes in the MBR and PBR) I will re-state (if it was needed) that filesystem access is based on clusters and as long as the clusters are the (regular for a NTFS volume of that size) 4096 bytes (i.e. a multiple of *any* physical sector size possible) everything can be "fixed". The issue may come if the cluster size is smaller than that, i.e. 2048 bytes (but i believe it would have needed to be "forced" when formatting), then using the GPT approach is the only way to access the whole volume (through 512 bytes/sector). Anyway, if I were Snear, I would re-create the partition with the "new" 2048 sectors-before paradigm (or 64 or 128, etc.), which is more suitable to a 4096 physical sector and (normally) 512 bytes logical sector disk, and that could make it a tad bit faster in use. jaclaz [1] actually, once said that very few things are as ugly as a hybrid MBR/GPT disk, I see nothing preventing the creation of the GPT structures even now and test the disk connected "normally" (i.e. exposing 512 bytes sector) on a later OS, let's say 7, if the structures are present, changing the involved data (partition table entry and geometry data in the PBR) is just a bunch of bytes and this way one would have a tested "way out" solution in case of disaster.

I.e. like *any* AF disk around, nothing "new" or "peculiar". @daniel_k Good , and interesting links , still none of those shed any light if this is a feature of the specific chip/chipset (triggered or not triggered by this or that driver version) or if it is a feature of this (but not that) driver version (and thus can - at least in theory - be replicated in other drivers). Maybe - just maybe - it is possible to have a "filter driver" (not entirely unlike - say - cfadisk.sys or dummydisk.sys or karyionix's diskmod.sys) capable to replicate the behaviour on other controllers. jaclaz

Good, thanks . In the meantime, I found version 3.1.1540.68 (if needed/for testing) in an archive on driversguide ATI_SB700_IDE_Floppy.zip, I am attaching it. Maybe - just maybe - having a number of "near enough" versions someone with the right knowledge of capability (as an example daniel_k ) may be able to understand what exactly makes these drivers so special, though we don't know if the "variable sector size" is something that is in the driver only or if it is a specific feature of the actual controller chip and the software only "switches modes". A (hopefully "pristine" and unmodified) version 3.1.1540.86 can be found on Acer's site, as a driver for the model Aspire 4535G: https://www.acer.com/ac/en/ID/content/support-product/1141;-; jaclaz ATI_SB700_IDE_Floppy.zip

So, none were included in MS XP SP3? There are - here and there - reports that V3.1.1540.25 is very slow, so, if the V3.1.1540.86 works (and works fast) it should be the "recommended" version. Any source (besides the Dell OEM XP Pro install CD/DVD or *wherever* you found it)? @daniel_k What does NAPA series mean? As a personal favour (only if you wish, of course ): Please, please, please, do not use the .rar format as it is NOT a "universal" format, a lot of other tools (but WinRar) have issues with it. jaclaz

Thank you , re-edited, hopefully this time getting it right, my previous post. jaclaz

The report seemingly mentions using the built-in MS driver (or maybe it is another AMD driver, this is what is not entirely clear), and - still if I get this right ahcix86.sys V3.1.1540.25 works ahcix86.sys V3.1.1540.86 (included in SP3) works or does NOT work? ahcix86.sys V3.1.1540.127 (newer) does NOT work and cannot be even installed ahcix86.sys V3.2.1540.35 does NOT work ahcix86.sys V3.3.1540.29 does NOT work I re-edited my previous post to underline that at the moment that the source could be either MS or AMD. Let's wait for Snear to confirm the above and/or post a more ordered list of what works/doesn't work and the actual source of each file. Once we have a definite list, we can pass the info to Fernando (who is AFAIK the most knowledgeable about SATA/AHCI and RAID drivers ): https://www.win-raid.com/t3024f45-Guide-Integration-of-AMD-s-AHCI-RAID-drivers-into-a-WinXP-W-k-CD.html he may well know much more than us on the various versions. jaclaz

Why would you need a Secure FTP server on an intranet? Wouldn't a "normal" FTP one do? Just in case: https://www.ferrobackup.com/map-ftp-as-disk.html http://www.killprog.com/fdrve.html And also with built-in features in XP: http://cybernetnews.com/cybernotes-map-a-ftp-to-a-drive-in-windows/ jaclaz