kebera@alzabo.ocunix.on.ca (Krishna E. Bera) (05/17/91)
Is there any effort underway to produce a ROM version of MS-DOS or clone that is: 1. inexpensive 2. easy to upgrade (by changing a board, for example) 3. standardized in interface (i.e., DOS interrupt calls) 4. bootable (obviously), but DIP switch selectable from a disk 5. preferably not patented ? If this has been dismissed in earlier discussion, my apologies and please e-mail me a summary. - -- Krishna E. Bera Programmer/Analyst kebera@alzabo.ocunix.on.ca MIL Systems Engineering Inc. Ottawa, Canada
padgett%tccslr.dnet@mmc.com (Padgett Peterson) (05/22/91)
>From: "Krishna E. Bera" <kebera@alzabo.ocunix.on.ca> >Is there any effort underway to produce a ROM version of MS-DOS or >clone that is: > 1. inexpensive > 2. easy to upgrade (by changing a board, for example) > 3. standardized in interface (i.e., DOS interrupt calls) > 4. bootable (obviously), but DIP switch selectable from a disk > 5. preferably not patented Not that I am aware of though several laptops have attempted this. The major problems would be: 1) Hardware is always more expensive than software to produce 2) Would make it difficult to upgrade 3) Would provide no protection from viruses - too many popular programs and peripherals rely on tailoring the BIOS (e.g. hard disk controllers) MBR (e.g. FDISK), and DOS (most TSRs) in approved methods. Unfortunately many of these methods can also be used by malicious software. 4) Undocumented necessities (such as necessary to use a CD-ROM or NETWARE). 5) "Bug" fixed would be much more expensive. Warmly, Padgett
PHYS169@csc.canterbury.ac.nz (Mark Aitchison, U of Canty; Physics) (05/23/91)
padgett%tccslr.dnet@mmc.com (Padgett Peterson) writes: >>From: "Krishna E. Bera" <kebera@alzabo.ocunix.on.ca> > >>Is there any effort underway to produce a ROM version of MS-DOS... > Not that I am aware of though several laptops have attempted this. yes there is; DRDOS 5.0 has been ROMable for about a year, and it seems MSDOS 5 will available in ROM as well. It is a good idea (even though I doubt it was planned as an anti-virus measure), but there are some popular simple virus methods it won't help against. The great benefit would be, I think (I haven't actually tried it though): (a) Viruses that change not the interrupt vector but substitute a jump at the start of the original code would be stopped, as would viruses that change specific locations inside specific versions of DOS to hook in virus code. (b) Viruses that modify executables (assuming all DOS utilities are in ROM) (c) Viruses that add new .COM files with the same name (assuming all directories in the path, and the current dir, are in ROM also) (d) Boot Sector & MBR viruses (again, with reasonable assumptions about the implementation) What this leaves is a few loopholes like blatently reducing the TOM or changing interrupt vectors, that are relatively easy to spot. > The major problems would be: > 1) Hardware is always more expensive than software to produce The cost of ROM chips is amazingingly cheap - most motherboards have empty ROM slots. Even if it needs an extra card with some fancing hardware for paging in and out large amounts of ROM (a la EMS), the cost of the card is likely to be small compared with modern costs of operating system software. The problem is that most people don't "see" the cost of the DOS because it is either lumped in with the cost of the hardware, or pirated. Software in ROM can be cheaper because the manufacturer doesn't need to charge more to get over half of the users not paying for the product. > 2) Would make it difficult to upgrade So long as it is a good product, like MSDOS 3.3 or DRDOS 5, people tend to stick with one version of operating system for a long time - long compared with the time before they wish they had a newer BIOS, for example. This is true for enough people to make it worthwhile; besides, the cost of the chips is very very small (say $10). > 3) Would provide no protection from viruses - too many popular programs > and peripherals rely on tailoring the BIOS (e.g. hard disk controllers) > MBR (e.g. FDISK), and DOS (most TSRs) in approved methods. Unfortunately > many of these methods can also be used by malicious software. The MBR need only contain the partition table - it need never be executed. A TSR or virus can't make "sneaky" hidden changes to DOS, they must result in obvious effects like changes to vectors, memory allocation, etc. The only worry would be a virus that targets specific TSR's. > 4) Undocumented necessities (such as necessary to use a CD-ROM or NETWARE). > 5) "Bug" fixes would be much more expensive. You would want to change the DOS fairly infrequently, but I dodn't see a big problem there. Perhaps some Digital Research bod could answer how CD-ROMS or NETWARE work in the ROM version of DRDOS, but I guess such things are possible. Of course, another option would be for a 386 CPU to manage memory so that it is as good as ROM - or possibly better (with IO trapping?), except for the time while it is loading the DOS, and so vunerable. In fact, a virus that makes full use of a 386 to protect itself would be very worring - perhaps enough added to teh BIOS to protect that stage would be the answer. Mark Aitchison, Physics, University of CAnterbury, New Zealand.