Macintosh-SW/mac-rom
1
0
Fork 0
mirror of https://github.com/elliotnunn/mac-rom.git synced 2025-01-15 12:30:53 +00:00
Code Issues Projects Releases Wiki Activity
mac-rom/OS/StartMgr/StartErr.a

399 lines
17 KiB
Plaintext
Raw Normal View History

Bring in CubeE sources Resource forks are included only for .rsrc files. These are DeRezzed into their data fork. 'ckid' resources, from the Projector VCS, are not included. The Tools directory, containing mostly junk, is also excluded.
2017-09-17 17:09:04 +08:00
;__________________________________________________________________________________________________
;
; File: StartErr.a
;
; Contains: System Error Handler & Debugger Interface
; This file contains the error handling interface code for Macintosh.
;
; This code can be divided into three pieces:
;
; 1) _Debugger trap handler
; 2) _SysError trap handler
; 3) Exception/interrupt handler
;
; The key to the new interface is the MacJmp low-memory vector. If nil, it
; implies that the current 'debugger' is the DeepShit error handler, otherwise
; it points to the entry point for alternate debugger, be it Macsbug, MPS remote
; or Joe Blow's turboProbe.
;
; The _Debugger trap interface is in the form of
; Debugger (Arg1, Arg2, ... ArgN, SizeOfParams, CmdWord)
;
; SizeOfParams (INTEGER) is the number of bytes occupied by Arg1...ArgN
; CmdWord (INTEGER) is the dispatch word (0...N) for the debugger
;
; The _SysError trap interface is the same, i.e. error value in D0.W
;
; The Exception/interrupt handler & _SysError both save registers in syserror space
; ($7FC80), then they and the _Debugger trap handler attempt to dispatch either
; to an installed debugger (MacJmp <> 0) or (if not _Debugger) to the DeepShit
; handler.
;
; The MacJmp (or MacJmpFlag for 32 bit clean machine) vector has three high-byte flags
;
; | Debugger Running | Debugger Initialized | Debugger Installed | 5 unused bits |
;
; Debugger Running (bit set) => the debugger is currently executing
; Debugger Initialized => Able to handle system errors
; Debugger Installed => MacJmp vector pts to usable debugger
;
; Written by: Ken Krugler
;
; Copyright: <09> 1985-1993 by Apple Computer, Inc., all rights reserved.
;
; Change History (most recent first):
;
; <SM8> 11/9/93 KW added some eieioSTP macros. Only expands for CygnusX1 ROM
; <SM7> 6/3/93 SAM Added some Supports24Bit around the DSHandler.
; <SM6> 12/1/92 RB Under SuperMario ROMs, change the system error for No patch to
; the system error that saiz this is an older system. When someone
; boots a SM ROM with a system less than 7.1 they should get that
; error and not a 'can't find patch' system error.
; <SM5> 11/24/92 HY Change Parity NMI check. Wrong branch sense (changed bne to beq)
; <SM4> 11/3/92 rab Roll in Horror changes. <H5> 5/4/92 JC Change Parity NMI patch
; to check specifically for PGC bit in Hardware External features
; flags on RBV type machines instead of checking for V8 flag.
; <SM3> 8/19/92 CSS Update from Reality:
; <13> 8/18/92 DTY Change the name of BootGlobs to StartGlobals to avoid confusion
; with BootGlobals, which is used by the boot code in Boot[1-3].a.
; <SM2> 5/16/92 kc Roll in Horror. Comments follow:
; <H7> 5/29/91 CCH Backed out revision 6.
; <H6> 5/29/91 WM Changed MOVE to SR into OR to SR
; <H5> 5/10/91 HJR Backed out of last revisions changes since a much cleaner
; implementation was done in PowerMgr.a.
; <H4> 4/29/91 HJR Added NMIExcpPatch to NMIExcp in order allow NMI on portables
; with power-cycling.
; <H3> 12/11/90 JJ Mac LC: Change reference to VISAChipBit to be V8ChipBit.
; <H3> 12/11/90 JJ Mac LC: Changes references to VISAChipBit to V8ChipBit.
; <H2> 10/25/90 CCH Added support for ReAnimator when forRomulator equate is set.
; <12> 1/15/92 KC Fix "If last BRA.S to import is to next instruction, BAD CODE
; will be generated" assembler error.
; <11> 10/1/91 JSM Don<6F>t use eclipseDebug.
; <10> 9/30/91 JSM Don<6F>t use is32BitClean conditional, all future ROMs will be.
; <9> 9/16/91 JSM Cleanup header.
; <8> 6/12/91 LN removed #include 'HardwareEqu.a'
; <7> 10/22/90 JJ Rex V8: Change VISAChipBit to V8ChipBit.
; <6> 7/2/90 CCH Removed a fatally redundant MOVEC xxx,VBR.
; <5> 6/27/90 CCH Added code to be nice to the ReAnimator nub so that it's
; exception vectors don't get initialized over. Also moved the
; initialization of VBR (Cpu <20> 020) to AFTER we actually have
; useful vectors in the exception vector table.
; <4> 4/14/90 JJ Add runtime check to avoid seeing parity error NMI on Elsie.
; <3> 1/11/90 CCH Added include of <20>HardwarePrivateEqu.a<>.
; <2> 12/26/89 GMR RPU parity is now disabled after the error, just as the IIci was
; handled.
; <2.1> 12/7/89 GMR COMMENT:NEEDED FOR ZONE5: Updated NMI handler for Zone-5 parity
; interrupts.
; <2.0> 8/22/89 SES Removed references to nFiles.
; <1.9> 7/17/89 GMR fixed bug when blasting warm start on MMU machines: call
; _SwapMMUMode before looking at bootglobs and after.
; <1.8> 5/26/89 rwh fixed changed names of BootGlobals fields
; <1.7> 5/19/89 GMR Use equate for parity error code (dsParityErr)
; <1.5> 5/16/89 GMR Added parity code in NMI handler, to set error code to 101 if
; parity caused NMI on RBV based machines.
; <1.6> 5/16/89 rwh (for gmr) cleared warmstart flag after parity error, so ram gets
; re-initialized.
; <1.4> 5/11/89 SWC Changed conditionals from machine- to feature-based for
; universal ROM.
; <1.3> 2/24/89 SES Added setup of VBR for '020 machines and greater.
; <1.2> 2/17/89 CSL Added MacJmpFlag to make MacJmp a clean 32 bit vector.
; <1.1> 11/10/88 CCH Fixed Header.
; <1.0> 11/9/88 CCH Adding to EASE.
; <<3C>1.1> 9/23/88 CCH Got rid of inc.sum.d and empty nFiles
; <1.0> 2/10/88 BBM Adding file for the first time into EASE<53>
; <C914> 10/29/87 rwh Port to Modern Victorian (onMvMac)
; <C882> 9/3/87 RDC Fix code for new MacsBug - make NMI handler set error code(reg
; D0) = 13 on entry to SysError - have Syserror check only for
; D0=13 for NMI exceptions - provide alternate entry point into
; SysError for BadTrap routine
; <C875> 9/2/87 MSH Port to HcMac (Laguna)
; <C447> 11/20/86 DAH Clear NMIFlag in SysErrInit.
; <A375> 11/8/86 CSL fixed launching into Microbug.
; <C206> 10/9/86 bbm Modified to mpw aincludes.
; <C152> 9/18/86 WRL Turned this into a separately compiled PROC.
; <C84> 7/29/86 WRL Removed nub debugger installation.
; <C28> 6/3/86 CSL Added changes for Aladddin (onMacPP).
; <C1> 4/14/86 RDC Added changes for 68020 Reno project (NuMac)
; - Changed variable for NMI debounce
; - Removed init of Nubdebugger for now
; 11/4/85 JTC Save 8 bytes of address/bus error info in SExxx space for new
; debuggers.
; 11/2/85 JTC Fix handling of exception frame. Pass whole business to
; debugger, saving up for a grand RTE finale. Must dummy a
; format-0 frame for 010 and 020.
; 7/23/85 RDC added changes for MidMac
; - code to setup and handle true NMI exception
; - debounce of NMI int
; 7/13/85 JTC Modified SysErr interface for 010/020 stack handling.
; 5/2/85 KWK new MacJmp flags, NMI signals init call
; 5/1/85 KWK Significant re-structure for flexible debugger call
; 4/30/85 KWK new today
;__________________________________________________________________________________________________
BLANKS ON
STRING ASIS
IF CPU >= 20 THEN ; <1.3>
MACHINE MC68020 ; <1.3>
ENDIF ; <1.3>
PRINT OFF ; <C152>
LOAD 'StandardEqu.d'
INCLUDE 'HardwarePrivateEqu.a'
INCLUDE 'UniversalEqu.a' ; <1.5>
INCLUDE 'MMUEqu.a' ; <C152>
INCLUDE 'BootEqu.a' ; <C152>
PRINT ON ; <C152>
;----------------------------------------------------------------------
; Routine SysErrInit
; Arguments A3.L (input) Ram overlay offset value
; A6.L (input) Return address
; Function Set up exception vectors, nub memory, etc.
; Uses A2
;----------------------------------------------------------------------
SysErrInit PROC EXPORT ; <C152>
IMPORT GenExcps ; <C152>
IMPORT IRQException ; <C152>
IMPORT NMIExcp ; <C152>
SUB.L A0,A0 ; point A0 at start of RAM
IF NOT forRomulator THEN ; <T2>
MOVE.L A0,D0 ; setup the VBR for '020s and greater <1.3>
MOVEC D0,VBR ; <1.3>
ENDIF
ADDQ.W #BusErrVct,A0 ; start stuffing at the bus error vector <1.4>
LEA GenExcps,A1
MOVEQ #10-1,D0 ; setup exception handlers for <1.4>
@0 MOVE.L A1,(A0)+ ; bus error thru line 1111 emulator vectors
ADDQ.W #2,A1 ; <1.4>
DBRA D0,@0
MOVEQ #13-1,D0 ; point unassigned (vector 12) thru spurious <1.4>
@1 MOVE.L A1,(A0)+ ; interrupt vectors to the "undefined" handler
DBRA D0,@1 ; (System Error #11)
LEA IRQException,A2 ; point to the IRQ exception handler
MOVEQ #6-1,D0 ; stuff the IRQ pointer into first 6 autovectors<1.4>
@2 MOVE.L A2,(A0)+
DBRA D0,@2
LEA NMIExcp,A2 ; level 7 (NMI) has a special routine to de-bounce the
MOVE.L A2,(A0)+ ; programmer's switch on some machines (else don't care)
MOVEQ #32-1,D0 ; point TRAP #0 vector thru unassigned (63) <1.4>
@3 MOVE.L A1,(A0)+ ; to the "undefined" handler (System Error #11)
DBRA D0,@3 ; since they aren't user-definable
IF forRomulator THEN ; <T2>
SUB.L A0,A0 ; point A0 at start of RAM
TestInRam a1 ; only allow vector-nuking if running in RAM <bg>
beq.s @dontNukeMe ; skip vector restoration if running from ROM <bg>
movec vbr,a1 ; get OUR VBR
; ... and put them in the standard vector tbl
move.l BusErrVct(a1),BusErrVct(a0) ; then copy in our vectors
move.l AddrVector(a1),AddrVector(a0)
move.l IlglVector(a1),IlglVector(a0)
move.l ZeroVector(a1),ZeroVector(a0)
move.l CheckVector(a1),CheckVector(a0)
move.l TrapVVector(a1),TrapVVector(a0)
move.l PrivlgVector(a1),PrivlgVector(a0)
move.l TraceVector(a1),TraceVector(a0)
move.l DebugVector(a1),DebugVector(a0)
move.l FmtErrVect(a1),FmtErrVect(a0)
move.l BadIntVector(a1),BadIntVector(a0)
move.l BkptVector(a1),BkptVector(a0)
@dontNukeMe ; <bg>
MOVE.L A0,D0 ; setup the VBR for '020s and greater
MOVEC D0,VBR ;
ENDIF ;
; now clear out the SysErr data space. Starts at $7FC80 and continues for 128 bytes. Used to save off
; registers and maintain other nub state information (nub data ptr, etc).
CLR.B NMIFlag ; clear debouncing flag <C447>
LEA SEVars,A0 ; pt to start of nub memory
MOVEQ #(SEVSize/4)-1,D0 ; data size
@4 CLR.L (A0)+ ; clear it
DBRA D0,@4 ; loop
; set up the default debugger (serial interface to MPS debugger)
CLR.L MacJmp ; init debugger flag for no debugger <C1/14Apr86>
CLR.B MacJmpFlag ; init debugger flag for 32 bit clean machine <v1.2>
RTS6
ENDP ; <C152>
;----------------------------------------------------------------------
; Routine ToDeepShit
; Arguments (SP) (input) Return address to exception receiver
; xxxx(SP) CPU and exception type stack frame <13Jul85>
; Function Receives all the hardware exceptions. It derives the
; deepShit ID from the return address on the stack and then
; invokes the deepShit manager.
; Modified to call yoghurt manager with stack cleansed of all frame <13Jul85>
; information. Idea is to save Reg/Stack info then RTE to DS manager, <13Jul85>
; using the RTE to clear whatever frame info is there. <13Jul85>
; Nowadays, pass entire frame (sans PC and SR) to debugger, for later RTE <02Nov85> JTC
;----------------------------------------------------------------------
ToDeepShit PROC EXPORT ; <C152>
IMPORT GenExcps ; <C152>
IMPORT SysErr2 ; <C152>
MOVE.W #$2700,SR ; turn off interrupts
TST.B MacJmpFlag ; are the debuggers in control here?
BMI.S @0 ; yes, don't bother saving regs
MOVEM.L A0-A7/D0-D7,SEVars ; save all regs for debug
; Compute small integer exception index 1,2,... into D0, based on address atop the stack. <13Jul85>
@0 LEA GenExcps,A0 ; get address of receiver table
MOVE.L (SP)+,D0 ; get return address
SUB.L A0,D0 ; compute delta
LSR #1,D0 ; divide by 2
; Now clear stack of exception info, saving SR and PC unless re-entered. A special case <13Jul85>
; is class 0 exception on the 68000, which has 4 words of info on top of the usual SR/PC. <13Jul85>
Build updated MainCode image (matches Mac OS ROM 9.6.1+)
2017-09-20 19:58:54 +08:00
TST.B $BFF
BMI.S @3
Bring in CubeE sources Resource forks are included only for .rsrc files. These are DeRezzed into their data fork. 'ckid' resources, from the Projector VCS, are not included. The Tools directory, containing mostly junk, is also excluded.
2017-09-17 17:09:04 +08:00
MOVE (SP)+,SESR ; SR on top for all CPUs <13Jul85>
MOVE.L (SP)+,SEPC ; next is the saved PC <13Jul85>
Build updated MainCode image (matches Mac OS ROM 9.6.1+)
2017-09-20 19:58:54 +08:00
BRA SysErr2 ; bypass nonsaving pop <13Jul85><1.4>
Bring in CubeE sources Resource forks are included only for .rsrc files. These are DeRezzed into their data fork. 'ckid' resources, from the Projector VCS, are not included. The Tools directory, containing mostly junk, is also excluded.
2017-09-17 17:09:04 +08:00
@3 ADDQ.W #6,SP ; kill SR and PC <13Jul85><1.4>
Build updated MainCode image (matches Mac OS ROM 9.6.1+)
2017-09-20 19:58:54 +08:00
BRA SysErr2
Bring in CubeE sources Resource forks are included only for .rsrc files. These are DeRezzed into their data fork. 'ckid' resources, from the Projector VCS, are not included. The Tools directory, containing mostly junk, is also excluded.
2017-09-17 17:09:04 +08:00
NOP ; appease the assembler gods <12> kc
ENDP ; <C152>
; This is the receiver table for all exceptions (except 1010). Look at ToDeepShit
; code to understand BSR.S stuff
GenExcps PROC EXPORT ; <C152>
IMPORT ToDeepShit ; <C152>
EXPORT IRQException ; <C152>
EXPORT NMIExcp ; <C152>
BSR.S ToDeepShit ; (1) bus error
BSR.S ToDeepShit ; (2) address error
BSR.S ToDeepShit ; (3) illegal instruction
BSR.S ToDeepShit ; (4) zero divide
BSR.S ToDeepShit ; (5) check instruction
BSR.S ToDeepShit ; (6) trapv instruction
BSR.S ToDeepShit ; (7) privilege violation
BSR.S ToDeepShit ; (8) trace
BSR.S ToDeepShit ; (9) line 1010 (temporarily)
BSR.S ToDeepShit ; (10) line 1111
BSR.S ToDeepShit ; (11) other exceptions (TRAP calls, spurious int, etc)
BSR.S ToDeepShit ; (12) Place holder for no trap number exception
IRQException
BSR.S ToDeepShit ; (13) interrupt exceptions
NMIExcp BSET #7,NMIFlag ; use high bit of NMI flag as indicator <C1/14Apr86>
BEQ.S IRQException ; process first interrupt <23Jul85><1.4>
RTE ; else just ignore <23Jul85>
ENDP ; <C152>
;----------------------------------------------------------------------
; Routine SystemError
; Arguments D0.W (input) System error number
; Function Handle system errors. Allows consistant interface to
; debuggers.
; Stack is cleansed of all trap parameters, so that return with RTE <13Jul85>
; depends on information restored from SEPC and SESR. <13Jul85>
; Nowadays, debugger is called with exception parameters on stack. <02Nov85> JTC
; So must dummy up a format-0 frame when SystemError is called. <02Nov85> JTC
;----------------------------------------------------------------------
SystemError PROC EXPORT ; <C152>
EXPORT SysErr1 ; <C882>
EXPORT SysErr2 ; <C152>
IMPORT DebugProlog ; <C152>
IMPORT DSErrorHandler ; <C152>
MOVEM.L D0-D7/A0-A7,SEVars ; save regs up high in nub space
; entry point from BadTrap routine <C882>
SysErr1 MOVE.W SR,SESR ; save SR <13Jul85>
MOVE.L (SP)+,SEPC ; clean PC for fall-through to SysErr2 <13Jul85>
CLR.W -(SP) ; <02Nov85> JTC
; entry point for standard system exceptions (address error, etc)
SysErr2
TST.B MacJmpFlag ; re-entrancy? (presumed PLus for SystemError) <v1.2>
BMI.S @0 ; yes, skip reg saving
MOVE.L SP,SEA7 ; save true PC
@0 ; deleted obsolete stack-cleansing instructions <13Jul85>
Build updated MainCode image (matches Mac OS ROM 9.6.1+)
2017-09-20 19:58:54 +08:00
BSET.B #7,MacJmpFlag ; prevent re-entrancy, vote Republican <v1.2>
Bring in CubeE sources Resource forks are included only for .rsrc files. These are DeRezzed into their data fork. 'ckid' resources, from the Projector VCS, are not included. The Tools directory, containing mostly junk, is also excluded.
2017-09-17 17:09:04 +08:00
cmpi.w #dsNoPatch,d0 ; is this Error can't find patch ? <SM6> rb
beq.s @PreCubeE ; change the error message then <SM6> rb
cmpi.w #dsBadPatch,d0 ; is this a can't load patch message ? <SM6> rb
bne.s @SaveErr ; if not, don't touch it <SM6> rb
@PreCubeE ; SM ROMs don't have patches, user booted 7.0 <SM6> rb
moveq #dsOldSystem,d0 ; set error (system on disk is too old) <SM6> rb
@SaveErr ; <SM6> rb
MOVE.W D0,DSErrCode ; Save the possibly modified ErrCode <SM7>
Build updated MainCode image (matches Mac OS ROM 9.6.1+)
2017-09-20 19:58:54 +08:00
MOVE.L MacJmp,A1
MOVE.B MacJmpFlag,D3
ADD.B D3,D3
Bring in CubeE sources Resource forks are included only for .rsrc files. These are DeRezzed into their data fork. 'ckid' resources, from the Projector VCS, are not included. The Tools directory, containing mostly junk, is also excluded.
2017-09-17 17:09:04 +08:00
BPL.S Call2DS ; yes, but not initialized (init bit not set)
BNE.S CallDB ; yes, initialized, call 'em
; We got a system error w/un-initialized debuggers, if NMI then assume that user wants to try
; to get the debugger up and running.
Call2DS CMP.W #13,D0 ; Exception?
BNE.S CallDS ; no, call the deepshit handler
MOVEQ #0,D0 ; set special system error (0) for initialization
BRA.S CallDB ; and call the debugger
CallDS LEA DSErrorHandler,A1 ; point at system alert handler
CallDB Move.L A1,D1 ; check if A1 is nil <A375>
BEQ.S CallDS ; if nil, go to DSErrorHandler <A375>
JSR (A1) ; call the debugger/system error handler
TST.W D0 ; any error?
BNE.S CallDS ; yes, call deepshit
SysErrExit
BCLR #7,MacJmpFlag ; NOT in debugger any longer <v1.2>
MOVEM.L SEVars,D0-D7/A0-A7 ; restore regs
; In order to return with RTE, must set up vanilla 000 or 4-word 010/020 frame. <13Jul85>
MOVE.L SEPC,-(SP) ; push user's PC
MOVE.W SESR,-(SP) ; restore status reg
RTE ; and return to caller
ENDP ; <C152>
END
Reference in New Issue Copy Permalink