Program Segment Prefix

(Redirected from Program segment prefix)

The Program Segment Prefix (PSP) is a data structure used in DOS systems to store the state of a program. It resembles the Zero Page in the CP/M operating system. The PSP has the following structure:

Offset Size Contents
00h–01h 2 bytes (code) CP/M-80-like exit (always contains INT 20h)[1][2]
02h–03h word (2 bytes) Segment of the first byte beyond the memory allocated to the program
04h byte Reserved
05h–09h 5 bytes (code) CP/M-80-like far call entry into DOS, and program segment size[1][3]
0Ah–0Dh dword (4 bytes) Terminate address of previous program (old INT 22h)
0Eh–11h dword Break address of previous program (old INT 23h)
12h–15h dword Critical error address of previous program (old INT 24h)
16h–17h word Parent's PSP segment (usually COMMAND.COM - internal)
18h–2Bh 20 bytes Job File Table (JFT) (internal)
2Ch–2Dh word Environment segment
2Eh–31h dword SS:SP on entry to last INT 21h call (internal)
32h–33h word JFT size (internal)
34h–37h dword Pointer to JFT (internal)
38h–3Bh dword Pointer to previous PSP (only used by SHARE in DOS 3.3 and later)
3Ch–3Fh 4 bytes Reserved
40h–41h word DOS version to return (DOS 5 and later, alterable via SETVER in DOS 5 and later)
42h–4Fh 14 bytes Reserved
50h–52h 3 bytes (code) Unix-like far call entry into DOS (always contains INT 21h + RETF)
53h–54h 2 bytes Reserved
55h–5Bh 7 bytes Reserved (can be used to make first FCB into an extended FCB)
5Ch–6Bh 16 bytes Unopened Standard FCB 1
6Ch–7Fh 20 bytes Unopened Standard FCB 2 (overwritten if FCB 1 is opened)
80h 1 byte Number of bytes on command-line
81h–FFh 127 bytes Command-line tail (terminated by a 0Dh)[4][5]

The PSP is most often used to get the command line arguments of a DOS program; for example, the command "FOO.EXE /A /F" executes FOO.EXE with the arguments '/A' and '/F'.

If the PSP entry for the command line length is non-zero and the pointer to the environment segment is neither 0000h nor FFFFh, programs should first try to retrieve the command line from the environment variable %CMDLINE% before extracting it from the PSP. This way, it is possible to pass command lines longer than 126 characters to applications.

The segment address of the PSP is passed in the DS register when the program is executed. It can also be determined later by using Int 21h function 51h or Int 21h function 62h. Either function will return the PSP address in register BX.[6]

Alternatively, in .COM programs loaded at offset 100h, one can address the PSP directly just by using the offsets listed above. Offset 000h points to the beginning of the PSP, 0FFh points to the end, etc.

For example, the following code displays the command line arguments:

org   100h      ; .COM - not using ds

; INT 21h subfunction 9 requires '$' to terminate string
xor   bx,bx
mov   bl,[80h]
cmp   bl,7Eh
 ja   exit      ; preventing overflow

mov   byte [bx+81h],'$'

; print the string
mov   ah,9
mov   dx,81h
int   21h

exit:
mov   ax,4C00h  ; subfunction 4C
int   21h

In DOS 1.x, it was necessary for the CS (Code Segment) register to contain the same segment as the PSP at program termination, thus standard programming practice involved saving the DS register (since the DS register is loaded with the PSP segment) along with a zero word to the stack at program start and terminating the program with a RETF instruction, which would pop the saved segment value off the stack and jump to address 0 of the PSP, which contained an INT 20h instruction.

; save
push  ds
xor   ax,ax
push  ax

; move to the default data group (@data)
mov   ax,@data
mov   ds,ax

; print message in mess1 (21h subfunction 9)
mov   dx,mess1
mov   ah,9
int   21h

retf

If the executable was a .COM file, this procedure was unnecessary and the program could be terminated merely with a direct INT 20h instruction or else calling INT 21h function 0. However, the programmer still had to ensure that the CS register contained the segment address of the PSP at program termination. Thus,

jmp   start

mess1 db 'Hello world!$'

start:
mov   dx,mess1
mov   ah,9
int   21h

int   20h

In DOS 2.x and higher, program termination was accomplished instead with INT 21h function 4Ch which did not require the CS register to contain the segment value of the PSP.

See also

edit

References

edit
  1. ^ a b Taylor, Roger; Lemmons, Phil (June 1982). "Upward migration - Part 1: Translators - Using translation programs to move CP/M-86 programs to CP/M and MS-DOS" [Using translation programs to move CP/M programs to CP/M-86 and MS-DOS] (PDF). BYTE. Vol. 7, no. 6. BYTE Publications Inc. pp. 321–322, 324, 326, 328, 330, 332, 334, 336, 338, 340, 342, 344 [342, 344]. ISSN 0360-5280. CODEN BYTEDJ. Archived (PDF) from the original on 2020-01-16. Retrieved 2020-01-15. […] Gaining Access to CP/M-86 […] Gaining access to CP/M-86 requires placing the function code in the CL register, placing the byte parameter in the DL register or placing the word parameter in the DX register, placing the data segment in the DS register (the data segment is usually not changed for a converted program), and executing a software interrupt, INT #224. The result is returned in the AL register if it is a byte value; if the result is a word value, it is returned in both the AX and BX registers. Double-word values are returned with the offset in the BX registers and the segment in the ES register. Conversion of programs from CP/M-80 to CP/M-86, then, requires replacing the call to location 5 with the software interrupt INT #224. Another necessary change involves the warm boot. Under CP/M-80, the warm boot may be accessed by a system call with a function code of 0 for a jump to location 0. CP/M-86, however, does not support the jump to location 0. As a result, you must change this program exit in the translated program if the program is to run correctly. Provided that the call to location 5 is replaced with INT #224, that the warm boot change is made, and that the registers are mapped correctly, there should be little problem in getting the translated program to access the CP/M-86 system functions. […] Gaining Access to MS-DOS […] Although MS-DOS has a "preferred" mechanism through a soft-ware interrupt, INT #33, for accessing the system, an additional mechanism is provided for "preexisting" programs that is compatible with CP/M-80 calling conventions, at least for functions in the range of 0-36. As far as system calls within the allowed function range are concerned, the programmer doesn't have to do anything to translated programs to get them to run under MS-DOS other than to correctly map the registers. MS-DOS also supports the warm boot function of CP/M-80. A jump to location 0 under MS-DOS executes a software interrupt, INT #32, which is functionally a program end and the normal way to exit from a program. […] [1] [2][3][4][5][6][7][8][9][10][11][12][13][14] (13 pages)
  2. ^ Paul, Matthias R. (2002-10-07) [2000]. "Re: Run a COM file". Newsgroupalt.msdos.programmer. Archived from the original on 2017-09-03. Retrieved 2017-09-03. [15] (NB. Has details on DOS COM program calling conventions.)
  3. ^ Necasek, Michal (2011-09-13). "Who needs the address wraparound, anyway?". OS/2 Museum. Archived from the original on 2020-02-19. Retrieved 2020-02-19. […] 86-DOS, and hence PC DOS/MS-DOS, used a clever trick. The byte at offset 5 of the PSP contained a far call opcode (9Ah); the word at offset 6 of the PSP contained the appropriate value to indicate program segment size, and also the offset part of the far call. The word at offset 8, which served as the segment part of the far call, was crafted such that when combined with the offset, it would wrap around (a well understood feature of the 8086 CPU) and point to address 0:C0h, which contains interrupt vector 30h. […] the CALL 5 interface works even in DOS emulation under Windows NT and OS/2, and those systems most certainly cannot run with the A20 line disabled. How does that work then? […] Rather than chopping off address bits, the system mirrors the five bytes at 0:C0h at 1000C0h. The same technique had been in fact used in DOS 5 and above running with DOS=HIGH. In that case, DOS makes sure that linear address 1000C0h contains the appropriate far call. […]
  4. ^ Paul, Matthias R. (1997-07-01) [1994-01-01]. MSDOSTIPs — Tips für den Umgang mit MS-DOS 5.0-7 (in German). Archived from the original on 2017-08-22. Retrieved 2013-10-25. {{cite book}}: |work= ignored (help) (NB. MSDOSTIP.TXT is part of MPDOSTIP.ZIP, maintained up to 2001 and distributed on many sites at the time. The provided link points to a HTML-converted older version of the MSDOSTIP.TXT file.) [16]
  5. ^ Paul, Matthias R. (1997-05-01) [1995-03-01]. "Hinweise zu JPSofts 4DOS 5.5b/c, 5.51, 5.52a und NDOS". MPDOSTIP (in German). Archived from the original on 2016-11-04. Retrieved 2015-05-08. (NB. The provided link points to a HTML-converted version of the 4DOS5TIP.TXT file, which is part of the MPDOSTIP.ZIP collection.) [17]
  6. ^ "INT 21h,62h - Get PSP address (DOS 3.x)". Archived from the original on 2012-02-07.
  7. ^ a b c Schulman, Andrew; Brown, Ralf D.; Maxey, David; Michels, Raymond J.; Kyle, Jim (1994) [November 1993]. Undocumented DOS: A programmer's guide to reserved MS-DOS functions and data structures - expanded to include MS-DOS 6, Novell DOS and Windows 3.1 (2 ed.). Reading, Massachusetts, USA: Addison Wesley. ISBN 0-201-63287-X. (xviii+856+vi pages, 3.5"-floppy) Errata: [18][19]

Further reading

edit
edit