Re: [Freedos-user] [Freedos-devel] EMM386 new release 2.20, new HIMEM 3.20

[Japheth <ma...@ja...>]

> Since low-level changes were made to this version of EMM386, please keep an 
> eye out for any problems and report them as soon as you can.  While I don't 

The IF bug fix may have disclosed some previously hidden bugs which are due to 
a careless handling of flags. The following is an excerpt from EMM386.ASM, 
which is an interrupt 13h handler:

NEW13 PROC FAR
         BTS     CS:[Flags],INT13Activ           ; The beginning.
         BTR     CS:[Flags],NeedBuffer           ; Rather disable always.
         PUSH    ECX
         MOV     ECX,8                           ; The Statusbits partly
@@Loop: MOV     CS:[ChanFlags+ECX*2-2],1100B    ; = ModePrgrd+PagePrgrd
         LOOP    @@Loop	                        ; initialise
         POP     ECX
         PUSHF                                   ; Now call the old INT
         CALL    DWORD PTR CS:[OLD13]	        ; If an error occurs,
         PUSHF                                   ; ... there is nothing !
	JC      @@Bye
         BTR     CS:[Flags],NeedBuffer           ; Is it necessary to copy
         JNC     @@Bye				; back the buffer?
HLT13:  HLT                                     ; Hello, 32-Bit World !
@@Bye:  BTR     CS:[Flags],INT13Activ           ; Bye bye (Flags
         POPF				        ; the BTR CY-flag erased !)
         RETF    2		                ; delete flags on stack
NEW13 ENDP

The "RETF 2" at the end not only causes the TF to be lost, but also IF to be 
cleared now.

There is similiar code for Int 40h.

I had to test this with Win98 debug. The FD debug, although more powerful 
regarding 386 opcodes and registers, is either unable to display the correct 
state of IF or even worse, it clears this flag permanently.