Macintosh-SW/powermac-rom
1
0
Fork 0
mirror of https://github.com/elliotnunn/powermac-rom.git synced 2024-11-29 12:49:58 +00:00
Code Issues Projects Releases Wiki Activity
0bd13db3d2
powermac-rom/NanoKernel/NKReplacementInit.s

478 lines
9.2 KiB
ArmAsm
Raw Normal View History

Initial commit
2017-11-19 04:11:07 +00:00
; sprg0 = old KDP/EWA/r1 ptr
; r3 = PA_NanoKernelCode
; r4 = physical base of our global area
; r5 = NoIdeaR23
; r6 = PA_EDP or zero?
; r7 = probably ROMHeader.ROMRelease ('rom vers', e.g. 0x10B5 is 1.0<EFBFBD>5)
InitReplacement
crset cr5_eq
li r0, 0
; Position and initialise the kernel globals, IRP to KDP inclusive.
; (subset of builtin kernel)
; Zero from IRP (r4) to KDP (r4 + 10 pages)
lisori r12, kKDPfromIRP
mr r13, r4
@wipeloop
subic. r12, r12, 4
stwx r0, r13, r12
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bgt @wipeloop
Initial commit
2017-11-19 04:11:07 +00:00
; Copy the old KDP to r4 + 10 pages.
; (r1 becomes our main ptr and r4 is discarded)
mfsprg r11, 0
lisori r1, kKDPfromIRP
add r1, r1, r4
li r12, 4096
@kdp_copyloop
subic. r12, r12, 4
lwzx r10, r11, r12
stwx r10, r1, r12
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bgt @kdp_copyloop
Initial commit
2017-11-19 04:11:07 +00:00
; IRP goes at the base of the area we were given.
; Fill with repeating pattern and point EWA at it.
lisori r12, -kKDPfromIRP
add r12, r12, r1
stw r12, EWA.PA_IRP(r1)
bl InitIRP ; clobbers r10 and r12
; Play with some of the other values we were given
; Leave ROMRelease in r23.
mr r23, r7
; If no EDP (Emulator Data Page) pointer was provided,
; then put the EDP above our new KDP.
cmpwi r6, 0
stw r11, KDP.OldKDP(r1)
stw r9, 0x05a4(r1)
; discarded
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bne @emulatordata_ptr_provided
Initial commit
2017-11-19 04:11:07 +00:00
addi r6, r1, 0x1000
@emulatordata_ptr_provided
; Save a few bits
stw r6, 0x05a8(r1)
stw r3, KDP.PA_NanoKernelCode(r1)
stw r5, PSA.NoIdeaR23(r1)
stw r1, EWA.PA_KDP(r1)
addi r12, r1, -0x340 ; get the base of the main CPU struct
li r10, -1
stw r10, CPU.ID(r12)
lwz r3, KDP.PA_ConfigInfo(r1)
bl LookupInterruptHandler
stw r7, KDP.PA_InterruptHandler(r1)
; Clearly changed our mind about where we might be.
bl @x
@x mflr r12
subi r12, r12, @x - NKTop
stw r12, KDP.PA_NanoKernelCode(r1)
; FDP
llabel r10, FDP
add r12, r10, r12
stw r12, KDP.PA_FDP(r1)
; Do something terrible with the CPU features
Reverse userspace sync primitives Namely queues, semaphores, critical regions, event groups and "notifications". The MP calls implementing these services have been named after their MPLibrary wrapper functions. This convention will be followed in the future (no more NKCreateEvent).
2018-02-17 16:57:01 +00:00
lwz r12, EWA.Flags(r1)
Initial commit
2017-11-19 04:11:07 +00:00
li r10, 0x00
rlwimi r10, r12, 0, 12, 15
rlwimi r10, r12, 0, 28, 30
stw r10, -0x0968(r1)
; Cook up a MSR:
; MSR_EE = 1
; MSR_PR = 1
; MSR_FP = 0
; MSR_ME = 0
; MSR_FE0 = 0
; MSR_SE = 0
; MSR_BE = 0
; MSR_FE1 = 0
; MSR_IP = preserved
; MSR_IR = 1
; MSR_DR = 1
; MSR_RI = 0
; MSR_LE = 0
mfmsr r12
andi. r12, r12, 0x0040
ori r12, r12, 0xd032
stw r12, PSA.UserModeMSR(r1)
; Set SPRG0 (for this CPU at least)
mtsprg 0, r1
; r11 still contains the OLD EWA ptr (which is also KDP/PSA ptr?)
lhz r12, KDP.InfoRecord + InfoRecord.NKNanoKernelInfoVer(r11)
cmpwi r12, 0x0101
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bgt @replaces_later_than_0101
Initial commit
2017-11-19 04:11:07 +00:00
;
lwz r12, KDP.PA_ECB_Old(r1)
stw r12, EWA.PA_ContextBlock(r1)
lwz r12, 0x660(r1)
oris r12, r12, 0x20
stw r12, EWA.Flags(r1)
lwz r12, 0x0664(r1)
Reverse userspace sync primitives Namely queues, semaphores, critical regions, event groups and "notifications". The MP calls implementing these services have been named after their MPLibrary wrapper functions. This convention will be followed in the future (no more NKCreateEvent).
2018-02-17 16:57:01 +00:00
stw r12, EWA.Enables(r1) ; boy, better figure out what this is
Initial commit
2017-11-19 04:11:07 +00:00
b @endif
@replaces_later_than_0101
; Obviously cannot replace a v2 NanoKernel like myself
cmpwi r12, 0x0200
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bge CancelReplacement
Initial commit
2017-11-19 04:11:07 +00:00
lwz r12, EWA.PA_ContextBlock(r11)
stw r12, EWA.PA_ContextBlock(r1)
lwz r12, EWA.Flags(r11)
oris r12, r12, 0x20
stw r12, EWA.Flags(r1)
lwz r12, -0x000c(r11)
Reverse userspace sync primitives Namely queues, semaphores, critical regions, event groups and "notifications". The MP calls implementing these services have been named after their MPLibrary wrapper functions. This convention will be followed in the future (no more NKCreateEvent).
2018-02-17 16:57:01 +00:00
stw r12, EWA.Enables(r1)
Initial commit
2017-11-19 04:11:07 +00:00
@endif
lwz r12, 0x0340(r11)
Reverse PPC-68k context switching (incomplete)
2018-03-31 08:17:23 +00:00
lwz r10, KDP.LA_NCB(r11)
Initial commit
2017-11-19 04:11:07 +00:00
cmpw r12, r10
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
beq replace_old_kernel_0x198
Reverse PPC-68k context switching (incomplete)
2018-03-31 08:17:23 +00:00
stw r12, KDP.LA_NCB(r1)
Initial commit
2017-11-19 04:11:07 +00:00
stw r0, 0x06b4(r1)
lwz r10, 0x05b0(r11)
stw r10, 0x06c0(r1)
Reverse PPC-68k context switching (incomplete)
2018-03-31 08:17:23 +00:00
lwz r10, KDP.LA_NCB(r11)
Initial commit
2017-11-19 04:11:07 +00:00
stw r10, 0x06c4(r1)
lwz r10, 0x05b8(r11)
stw r10, 0x06c8(r1)
lwz r10, 0x05bc(r11)
stw r10, 0x06cc(r1)
stw r0, 0x06d0(r1)
stw r0, 0x06d4(r1)
stw r0, 0x06d8(r1)
stw r0, 0x06dc(r1)
stw r0, 0x06e0(r1)
stw r0, 0x06e4(r1)
stw r0, 0x06e8(r1)
stw r0, 0x06ec(r1)
stw r0, 0x06f0(r1)
stw r0, 0x06f4(r1)
stw r0, 0x06f8(r1)
stw r0, 0x06fc(r1)
replace_old_kernel_0x198
; Adjust a few KDP pointers to point into the new KDP
lwz r12, KDP.PA_PageMapStart(r1)
subf r12, r11, r12
add r12, r12, r1
stw r12, KDP.PA_PageMapStart(r1)
lwz r12, KDP.PA_PageMapEnd(r1)
subf r12, r11, r12
add r12, r12, r1
stw r12, KDP.PA_PageMapEnd(r1)
lwz r12, 0x05e8(r1)
subf r12, r11, r12
add r12, r12, r1
stw r12, 0x05e8(r1)
; Wipe KDP's NKInfo and ProcessorInfo
li r12, 0x200
addi r10, r1, KDP.NanoKernelInfo
@wipeloop
subic. r12, r12, 4
stwx r0, r10, r12
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bgt @wipeloop
Initial commit
2017-11-19 04:11:07 +00:00
; r9 = physical base of kernel
li r12, 0
addi r10, r1, KDP.InfoRecord
bl MoveRecord ; (NanoKernelCode, NewKDPInfoRecord, OldKDP, 0)
stw r10, KDP.InfoRecord + InfoRecord.InfoRecordPtr(r1)
stw r0, KDP.InfoRecord + InfoRecord.Zero(r1)
lhz r12, KDP.InfoRecord + InfoRecord.NKProcessorStateLen(r1)
addi r10, r1, PSA.ProcessorState
lwz r9, KDP.InfoRecord + InfoRecord.NKProcessorStatePtr(r1)
bl MoveRecord ; (OldProcessorState, NewPSAProcessorState, OldKDP, ProcessorStateLen)
stw r10, KDP.InfoRecord + InfoRecord.NKProcessorStatePtr(r1)
lhz r12, KDP.InfoRecord + InfoRecord.NKHWInfoLen(r1)
lwz r10, EWA.PA_IRP(r1)
addi r10, r10, IRP.HWInfo
lwz r9, KDP.InfoRecord + InfoRecord.NKHWInfoPtr(r1)
bl MoveRecord ; (OldHWInfo, NewIRPHWInfo, OldKDP, HWInfoLen)
stw r10, KDP.InfoRecord + InfoRecord.NKHWInfoPtr(r1)
lhz r12, KDP.InfoRecord + InfoRecord.NKProcessorInfoLen(r1)
addi r10, r1, KDP.ProcessorInfo
lwz r9, KDP.InfoRecord + InfoRecord.NKProcessorInfoPtr(r1)
bl MoveRecord ; (OldProcessorInfo, NewKDPProcessorInfo, OldKDP, ProcessorInfoLen)
stw r10, KDP.InfoRecord + InfoRecord.NKProcessorInfoPtr(r1)
stw r10, KDP.InfoRecord + InfoRecord.NKProcessorInfoPtr2(r1)
lhz r10, KDP.InfoRecord + InfoRecord.NKProcessorInfoVer(r1)
cmplwi r10, 0x0112
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bge @ProcessorInfo_version_already_current
Initial commit
2017-11-19 04:11:07 +00:00
li r12, 160
li r10, 0x0112
sth r12, KDP.InfoRecord + InfoRecord.NKProcessorInfoLen(r1)
sth r12, KDP.InfoRecord + InfoRecord.NKProcessorInfoLen2(r1)
sth r10, KDP.InfoRecord + InfoRecord.NKProcessorInfoVer(r1)
sth r10, KDP.InfoRecord + InfoRecord.NKProcessorInfoVer2(r1)
@ProcessorInfo_version_already_current
lhz r12, KDP.InfoRecord + InfoRecord.NKDiagInfoLen(r1)
addi r10, r1, PSA.DiagInfo
lwz r9, KDP.InfoRecord + InfoRecord.NKDiagInfoPtr(r1)
bl MoveRecord ; (OldDiagInfo, NewPSADiagInfo, OldKDP, DiagInfoLen)
stw r10, KDP.InfoRecord + InfoRecord.NKDiagInfoPtr(r1)
lhz r12, KDP.InfoRecord + InfoRecord.NKSystemInfoLen(r1)
lwz r10, EWA.PA_IRP(r1)
addi r10, r10, IRP.SystemInfo
lwz r9, KDP.InfoRecord + InfoRecord.NKSystemInfoPtr(r1)
bl MoveRecord ; (OldSystemInfo, NewIRPSystemInfo, OldKDP, SystemInfoLen)
stw r10, KDP.InfoRecord + InfoRecord.NKSystemInfoPtr(r1)
lhz r12, KDP.InfoRecord + InfoRecord.NKNanoKernelInfoLen(r1)
addi r10, r1, KDP.NanoKernelInfo
lwz r9, KDP.InfoRecord + InfoRecord.NKNanoKernelInfoPtr(r1)
bl MoveRecord ; (OldNanoKernelInfo, NewKDPNanoKernelInfo, OldKDP, NanoKernelInfoLen)
stw r10, KDP.InfoRecord + InfoRecord.NKNanoKernelInfoPtr(r1)
li r12, 0x160
sth r12, KDP.InfoRecord + InfoRecord.NKNanoKernelInfoLen(r1)
li r12, kNanoKernelVersion
sth r12, KDP.InfoRecord + InfoRecord.NKNanoKernelInfoVer(r1)
lwz r8, KDP.ProcessorInfo + NKProcessorInfo.DecClockRateHz(r1)
stw r8, PSA.DecClockRateHzCopy(r1)
; Play with ConfigFlags
lwz r8, KDP.NanoKernelInfo + NKNanoKernelInfo.ConfigFlags(r1)
_bset r8, r8, 31 ; always set bit 31
if &TYPE('NKShowLog') != 'UNDEFINED'
_bset r8, r8, 28 ; see if someone can test this
endif
cmplwi r23, 0x27f3 ; set bit 27 on ROM 2.7f3 or later
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
blt @oldrom ; means later than PDM and Cordyceps
Initial commit
2017-11-19 04:11:07 +00:00
_bset r8, r8, 27
@oldrom
stw r8, KDP.NanoKernelInfo + NKNanoKernelInfo.ConfigFlags(r1)
; Say hello.
bl InitScreenConsole
_log 'Hello from the replacement multitasking NanoKernel. Version: '
mr r8, r12
bl printh
_log '^n Old KDP: '
mr r8, r11
bl printw
_log ' new KDP: '
mr r8, r1
bl printw
_log ' new irp: '
lwz r8, EWA.PA_IRP(r1)
mr r8, r8
bl printw
_log 'ROM vers: '
mr r8, r23
bl printh
_log '^n'
; Jump back into the common code path of Init.s
; The Emulator ContextBlock is expected in r6.
lwz r6, KDP.PA_ECB(r1)
b InitHighLevel
; MoveRecord
; r9 = base of kernel???
; Seems to be code to relocate some old structures.
MoveRecord ; OUTSIDE REFERER
; Check whether the old structure is in KDP
;
lwz r22, KDP.PA_ConfigInfo(r1)
lwz r22, NKConfigurationInfo.LA_InfoRecord(r22)
subf r9, r22, r9 ; r9 = offset of old address in irp
cmplwi r9, 0x1000
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bge @kdp
Initial commit
2017-11-19 04:11:07 +00:00
add r21, r9, r11 ; r21 = the old address if it had been in KDP instead?
@0x18
; r9 = offset of old structure in old parent page
; r10 = destination
; r12 = length
;
@loop
subic. r12, r12, 4
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
blt @exit_loop
Initial commit
2017-11-19 04:11:07 +00:00
lwzx r9, r21, r12
stwx r9, r10, r12
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
bgt @loop
Initial commit
2017-11-19 04:11:07 +00:00
@exit_loop
lwz r22, KDP.PA_ConfigInfo(r1)
lwz r22, NKConfigurationInfo.LA_KernelData(r22)
subf r10, r1, r10
lisori r21, -9 * 4096
cmpw r10, r21 ; if dest is nearer than 9 pages below kdp...
Remove unnecessary static branch hints These + and - characters in a branch mnemonic can cause the assembler to produce conditional branch instructions with that hint the branch predictor. The default for forward branches is -, and for backward branches is +. If a mnemonic is issued with the opposite sign, then bit 10 of the instruction (the LSB of the BO field) is set. My long-retired "ppcdisasm" script inserted these hints unconditionally, despite 98% of them not being required. The code is much cleaner now. I read in some old MPW release notes that PPCAsm and PPCLink together exhibit a quirk when linking conditional branches to imported symbols. PPCAsm always assembles these conditional branches as if they were forward branches (that is, a + hint will always set the hint bit, and a - hint will never). I hoped to use this property to divine whether the NanoKernel was linked from one or many assembly files, but I was frustrated by the lack of conditional branches between files.
2018-04-15 07:05:02 +00:00
blt @0x50
Initial commit
2017-11-19 04:11:07 +00:00
add r10, r10, r22
blr
@0x50
lwz r22, KDP.PA_ConfigInfo(r1)
lwz r22, NKConfigurationInfo.LA_InfoRecord(r22)
lwz r21, EWA.PA_IRP(r1)
add r10, r10, r1
subf r10, r21, r10
add r10, r10, r22
blr
@kdp
add r9, r9, r22
lwz r22, KDP.PA_ConfigInfo(r1)
lwz r22, NKConfigurationInfo.LA_KernelData(r22)
subf r9, r22, r9 ; r9 now equals an offset from old_kdp
add r21, r9, r11 ; r21 = address in new_kdp
b @0x18
Reference in New Issue Copy Permalink