Date   : Sun, 19 Feb 1984 13:36:43 CST
From   : mknox@ut-ngp.ARPA
Subject: SELDSK bug in CP/M-86

I just encountered a rather interesting bug in the IBM-PC BIOS implementation
of CP/M-86 (found it the hard way, of course).

   Perform a SELDSK BIOS call (BIOS call number 9) to a disk (it doesn't
   matter what one), specifying that it is a 'new disk'.  It will work
   correctly.

   Without doing any disk READ or WRITE functions, now do another SELDSK
   call for the same disk (again specifying that it is a 'new disk').
   The DISK parameter block returned will UNCONDITIONALLY specify that the
   disk is SINGLE DENSITY!!!

-----

Why would anyone do two disk selects in a row to the same drive?  One case
is an application program that selects a disk upon startup, and is then
instructed by the user to 'log in a new diskette in that drive'.  Since
the density and allocation map may have changed a fresh SELDSK is necessary.

Curiously, any read/write between the SELDSK function fixes the problem.
Other BIOS calls do not.