DOS ain't dead

> Johnson Lam has posted an updated 23-Sep-2011 DRIVERS.ZIP file, with all
> UIDE drivers improved, on his website at --
>
> http://johnson.tmfc.net/dos/driver.html
>
> UIDE-S now runs 30 BIOS disks/diskettes and 8 CD/DVD drives, and it also
> sets the "UIDE$" default name when no CD/DVD drives are found, making it
> more "identical" to the full UIDE. FreeDOS "autoloader" scripts could
> now use UIDE-S, if they can otherwise deal with XMS memory. I am still
> unable to "fit" /N3 switch logic (87 more bytes) in a 7.5K-size UIDE-S!
>
> UIDE2 now leaves the entire 4608-byte driver (exactly 4.5K!) in upper or
> DOS memory, when its /HL switch is given. /HL thus allows a 280-MB HMA
> cache for V7.10 MS-DOS and 425-MB+ for other DOS systems -- My own V6.22
> MS-DOS can set 600-MB reliably (used to CRASH above 585-MB)! The 25-MB
> larger cache may help V7.10 MS-DOS users, and perhaps others, who want a
> "faster" cache for protected mode. UIDE2 is also back to 34 BIOS units
> and 8 CDs/DVDs, "same as UIDE" like before!
>
> UIDE and UIDEJR have "minor" size reductions, and all other drivers have
> merely been re-dated to 23-Sep-2011 for consistency, as always.
>
> I hope protected-mode users like the new UIDE2. Except for differences
> in doing "protected caching", it is now "identical" to UIDE in features,
> still fits in a 7.5K file, and UIDE2 is one of my best drivers ever!

> Johnson Lam has posted a new 30-Sep-2011 DRIVERS.ZIP file on
> his website at http://johnson.tmfc.net/dos/driver.html .
>
> The UIDE, UIDE2 and UIDE-S drivers can now be assembled from
> the UIDE.ASM source file, using a /dPMDVR switch for UIDE2
> and a /dMINDVR switch for UIDE-S ("protected" or "minimum"
> options). The three caching drivers now have common "core"
> logic, and each runs 10 controllers, 34 BIOS disks/diskettes
> and 8 CD/DVD drives. A lot less "upgrade work" for me with
> only 1 source file, also a bit more "reliability" for users!
>
> XMGR, RDISK, and UIDEJR (quite "different" and still has its
> own source file!) are unchanged, and are merely re-dated for
> consistency, as always.

> Johnson Lam has posted an updated 7-Oct-2011 DRIVERS.ZIP file on
> his website, at http://johnson.tmfc.net/dos/driver.html . The
> site still has a "Last Update" date of 9-30, but the files which
> download ARE dated 10-07-2011 and have been verified O.K.
>
> This update is not more driver improvements but fixes a "BUG" in
> some of the "EDD BIOS" logic being used today!
>
> Daniel Nice noted that on 3 of his 5 systems, a USB memory stick
> declared to the BIOS as a "hard disk" caused the UIDE drivers to
> "duplicate" his last actual hard disk! He found the "EDD BIOS"
> is NOT updating the "DPTE" pointer for a "USB stick" disk -- The
> BIOS declares 30 bytes of EDD data, but in fact returns only 26,
> leaving the last actual hard-disk's "DPTE" pointer still there!!
> This is clearly an ERROR in the BIOS, that needs to be CORRECTED
> by those ["children", usually!] who create BIOS programs!
>
> Daniel suggested a "workaround" of setting the "DPTE" pointer to
> all-ones before UIDE issues Int 13h AH=048h. That worked fine!
> The UIDE drivers can again handle all "real" SATA/IDE hard disks
> without getting confused when USB-stick "disks" are also in use.
>
> Users should upgrade immediately to the 7-Oct-2011 UIDE drivers,
> if (A) their system BIOS lets "USB memory sticks" be declared as
> "hard disks" and (B) their system also has one or more REAL hard
> disks attached!
>
> Many Thanks to Daniel, who in fact did all my "Leg Work!" in his
> analysis and resolution of this Bad BIOS "BUG"!

> Johnson Lam has posted a new 16-Oct-2011 DRIVERS.ZIP file on
> his website at http://johnson.tmfc.net/dos/driver.html .
>
> In it, UIDE has again been reduced down to a 7.5K-byte file,
> same as UIDE2, for "boot" diskettes and other systems having
> limited space. So, UIDE-S is no longer needed, and it has
> been eliminated! UIDE2's performance is also improved, and
> all other drivers have merely been re-dated to 16-Oct-2011.
>
> To get UIDE below 7.5K, I had to delete its /M switch. /M
> saved only 256 bytes of HMA, "not enough" to matter, as UIDE
> uses only 4336 bytes of HMA in all cases. Even poor MS-DOS
> V7.10 (short on HMA due to long-filename and Win95/98 logic)
> still has 9100+ bytes of free HMA, and a "BUFFERS=4" command
> in CONFIG.SYS can reduce what the kernel needs! Most other
> DOS systems have MUCH more free HMA and should be no problem
> if loading UIDE with a /H switch. UIDE can be re-assembled
> with "SBUFSZ equ 256", if anyone absolutely requires its 256
> byte binary-search buffer (runs maybe 1% slower)!
>
> The UIDE2 driver now runs a "hair" faster for protected-mode
> users, due to re-adding the old "ScnD" subroutine which uses
> a "scasw" command (not a full binary-search) in deleting old
> cache-table entries. "ScnD" is used with UIDE2's /H or /HL
> switches, as its HMA caches are small enough for such logic.
> Large caches in upper/DOS memory still use the current UIDE2
> "SrchD" routine, for better speed with bigger search tables.
>
> The "UIDE" caching drivers now consist of only --
>
> 1) The UIDE.ASM source file (assembles both drivers).
> 2) The UIDE.SYS driver for up to 4-Gigabyte caches.
> 3) The UIDE2.SYS driver for "fast" caches in protected-mode.
>
> I hope such a "simplification" is of benefit for UIDE users!