Macintosh-SW/powermac-rom
1
0
Fork 0
mirror of https://github.com/elliotnunn/powermac-rom.git synced 2024-06-06 02:29:33 +00:00
Code Issues Projects Releases Wiki Activity
933bd48f01
powermac-rom/NanoKernel/NKTimers.s

963 lines
17 KiB
ArmAsm
Raw Normal View History

Auto-generate per-file function lists
2018-06-30 08:00:35 +00:00
; AUTO-GENERATED SYMBOL LIST
; IMPORTS:
; NKConsoleLog
; getchar
; printb
; printd
; printh
; printw
; NKIndex
; DeleteID
; LookupID
; NKPoolAllocator
; PoolAllocClear
; PoolFree
; NKScheduler
; CalculateTimeslice
; FlagSchEvaluationIfTaskRequires
; SchRdyTaskNow
; SchTaskUnrdy
; clear_cr0_lt
; major_0x149d4
; NKSync
; EnqueueMessage
; SetEvent
; SignalSemaphore
; NKThud
; panic
; panic_non_interactive
; EXPORTS:
; DequeueTimer (=> NKMPCalls, NKPrimaryIntHandlers, NKSync, NKTasks)
; EnqueueTimer (=> NKMPCalls, NKSync)
; GetTime (=> NKMPCalls, NKScheduler, NKSync, NKTasks)
; InitTMRQs (=> NKInit)
; SetTimesliceFromCurTime (=> NKScheduler)
; StartTimeslicing (=> NKInit)
; TimebaseTicksPerPeriod (=> NKScheduler, NKSync)
; TimerDispatch (=> NKInterrupts)
Initial commit
2017-11-19 04:11:07 +00:00
Local_Panic set *
b panic
InitTMRQs ; OUTSIDE REFERER
Identify all accesses to the Primary System Area Whenever a field was added to the end of the Primary System Area (PSA) record, the offsets of all other fields would change. When reversing NK version 02.27, I noticed that 32 bytes had been added to the end of the PSA between 02.27 and 02.28. To build a byte-perfect 02.27, all references to PSA fields must therefore use the record definition in NKPublic.s instead of a numeric offset. In this commit, all PSA references are identified by field name (_FFF if the field has not yet been reversed).
2018-04-22 09:13:30 +00:00
addi r9, r1, PSA.TimerQueue
Initial commit
2017-11-19 04:11:07 +00:00
lis r8, 0x544d
ori r8, r8, 0x5251
stw r8, 0x0004(r9)
stw r9, 0x0008(r9)
stw r9, 0x000c(r9)
li r8, 0x00
stb r8, 0x0014(r9)
li r8, 0x01
stb r8, 0x0016(r9)
stb r8, 0x0017(r9)
lis r8, 0x7fff
ori r8, r8, 0xffff
mtspr dec, r8
stw r8, 0x0038(r9)
oris r8, r8, 0xffff
stw r8, 0x003c(r9)
mfspr r8, pvr
rlwinm. r8, r8, 0, 0, 14
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 InitTMRQs_0x7c
Initial commit
2017-11-19 04:11:07 +00:00
mflr r30
li r8, 0x40
; r1 = kdp
; r8 = size
Reverse the NanoKernel pool allocator
2018-03-23 09:22:56 +00:00
bl PoolAllocClear
Initial commit
2017-11-19 04:11:07 +00:00
; r8 = ptr
mr. r31, r8
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 Local_Panic
Identify all accesses to the Primary System Area Whenever a field was added to the end of the Primary System Area (PSA) record, the offsets of all other fields would change. When reversing NK version 02.27, I noticed that 32 bytes had been added to the end of the PSA between 02.27 and 02.28. To build a byte-perfect 02.27, all references to PSA fields must therefore use the record definition in NKPublic.s instead of a numeric offset. In this commit, all PSA references are identified by field name (_FFF if the field has not yet been reversed).
2018-04-22 09:13:30 +00:00
stw r31, PSA.OtherTimerQueuePtr(r1)
Initial commit
2017-11-19 04:11:07 +00:00
li r9, 0x07
stb r9, 0x0014(r31)
li r9, 0x01
stb r9, 0x0016(r31)
mtlr r30
InitTMRQs_0x7c
mfspr r8, pvr
rlwinm. r8, r8, 0, 0, 14
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 InitTMRQs_0xb4
Initial commit
2017-11-19 04:11:07 +00:00
mflr r30
li r8, 0x40
; r1 = kdp
; r8 = size
Reverse the NanoKernel pool allocator
2018-03-23 09:22:56 +00:00
bl PoolAllocClear
Initial commit
2017-11-19 04:11:07 +00:00
; r8 = ptr
mr. r31, r8
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 Local_Panic
Identify all accesses to the Primary System Area Whenever a field was added to the end of the Primary System Area (PSA) record, the offsets of all other fields would change. When reversing NK version 02.27, I noticed that 32 bytes had been added to the end of the PSA between 02.27 and 02.28. To build a byte-perfect 02.27, all references to PSA fields must therefore use the record definition in NKPublic.s instead of a numeric offset. In this commit, all PSA references are identified by field name (_FFF if the field has not yet been reversed).
2018-04-22 09:13:30 +00:00
stw r31, PSA._364(r1)
Initial commit
2017-11-19 04:11:07 +00:00
li r9, 0x08
stb r9, 0x0014(r31)
li r9, 0x01
stb r9, 0x0016(r31)
mtlr r30
InitTMRQs_0xb4
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
; Activate the NanoDebugger (whatever that is...)
lwz r30, KDP.PA_ConfigInfo(r1)
lhz r31, NKConfigurationInfo.Debug(r30)
cmplwi r31, NKConfigurationInfo.DebugThreshold
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 @nodebug
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
lwz r31, NKConfigurationInfo.DebugFlags(r30)
rlwinm. r8, r31, 0, NKConfigurationInfo.NanodbgrFlagBit, NKConfigurationInfo.NanodbgrFlagBit
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 @nodebug
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
lwz r8, KDP.NanoKernelInfo + NKNanoKernelInfo.ConfigFlags(r1)
_bset r8, r8, NKNanoKernelInfo.NanodbgrFlagBit
stw r8, KDP.NanoKernelInfo + NKNanoKernelInfo.ConfigFlags(r1)
mflr r30
li r8, Timer.Size
Reverse the NanoKernel pool allocator
2018-03-23 09:22:56 +00:00
bl PoolAllocClear ; one of those weird queue structures
Initial commit
2017-11-19 04:11:07 +00:00
mr. r31, r8
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 Local_Panic
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
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
li r9, Timer.kKind6
stb r9, Timer.Kind(r31)
li r9, 1
stb r9, Timer.KeepAfterFiring(r31)
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
bl GetTime
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
stw r8, Timer.Time(r31)
stw r9, Timer.Time+4(r31)
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r31
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
bl EnqueueTimer
Initial commit
2017-11-19 04:11:07 +00:00
_log 'Nanodebugger activated.^n'
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
mtlr r30
@nodebug
Initial commit
2017-11-19 04:11:07 +00:00
blr
TimerTable
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
dc.l TimerFireUnknownKind - NKTop ; Timer.kKind0
dc.l TimerFire1 - NKTop ; Timer.kKind1
dc.l TimerFire2 - NKTop ; Timer.kKind2
dc.l TimerFire3 - NKTop ; Timer.kKind3
dc.l TimerFire4 - NKTop ; Timer.kKind4
dc.l TimerFire5 - NKTop ; Timer.kKind5
dc.l TimerFire6 - NKTop ; Timer.kKind6
dc.l TimerFire7 - NKTop ; Timer.kKind7
dc.l TimerFire8 - NKTop ; Timer.kKind8
Initial commit
2017-11-19 04:11:07 +00:00
TimerDispatch ; OUTSIDE REFERER
mflr r19
mfsprg r18, 0
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 r19, EWA.TimerDispatchLR(r18)
Initial commit
2017-11-19 04:11:07 +00:00
TimerDispatch_0x30 ; OUTSIDE REFERER
mfspr r8, pvr
rlwinm. r8, r8, 0, 0, 14
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 @is_601
Initial commit
2017-11-19 04:11:07 +00:00
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
;not 601
@gettime_loop_non_601
Initial commit
2017-11-19 04:11:07 +00:00
mftbu r8
mftb r9
mftbu r16
cmpw r8, r16
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
bne- @gettime_loop_non_601
b @common
Initial commit
2017-11-19 04:11:07 +00:00
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
@is_601
@gettime_loop_601
Initial commit
2017-11-19 04:11:07 +00:00
mfspr r8, rtcu
mfspr r9, rtcl
mfspr r16, rtcu
cmpw r8, r16
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
bne- @gettime_loop_601
dialect POWER
liu r16, 1000000000 >> 16
oril r16, r16, 1000000000 & 0xffff
mfmq r17
mul r8, r16, r8
mfmq r16
mtmq r17
Initial commit
2017-11-19 04:11:07 +00:00
mfxer r17
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
a r9, r16, r9
aze r8, r8
Initial commit
2017-11-19 04:11:07 +00:00
mtxer r17
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
dialect PowerPC
@common
Initial commit
2017-11-19 04:11:07 +00:00
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
lbz r19, EWA.GlobalTimeIsValid(r18)
addi r30, r18, EWA.Base
cmpwi r19, 1
lwz r16, EWA.GlobalTime - EWA.Base(r30)
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 timer_earlier_than_sometime
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 r17, EWA.GlobalTime + 4 - EWA.Base(r30)
_b_if_time_gt r16, r8, timer_earlier_than_sometime
@skipbranch
Initial commit
2017-11-19 04:11:07 +00:00
li r19, 0x00
stw r30, -0x0254(r18)
stb r19, 0x0017(r30)
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
b TimerFire4_0x10
Initial commit
2017-11-19 04:11:07 +00:00
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
timer_earlier_than_sometime
Identify all accesses to the Primary System Area Whenever a field was added to the end of the Primary System Area (PSA) record, the offsets of all other fields would change. When reversing NK version 02.27, I noticed that 32 bytes had been added to the end of the PSA between 02.27 and 02.28. To build a byte-perfect 02.27, all references to PSA fields must therefore use the record definition in NKPublic.s instead of a numeric offset. In this commit, all PSA references are identified by field name (_FFF if the field has not yet been reversed).
2018-04-22 09:13:30 +00:00
lwz r30, PSA.TimerQueue + LLL.Next(r1)
Initial commit
2017-11-19 04:11:07 +00:00
lwz r16, 0x0038(r30)
lwz r17, 0x003c(r30)
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
_b_if_time_gt r16, r8, TimerDispatch_0x188
Insert calls to known linked-list macros The presence of InitList, InsertAsNext, InsertAsPrev and RemoveFromList calls should make some hitherto unexplored code easier to read.
2017-12-13 04:10:54 +00:00
RemoveFromList r30, scratch1=r19, scratch2=r20
Initial commit
2017-11-19 04:11:07 +00:00
lwz r19, 0x064c(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
lbz r20, Timer.Kind(r30)
Initial commit
2017-11-19 04:11:07 +00:00
rlwimi r19, r20, 2, 23, 29
cmplwi r20, 0x09
llabel r20, TimerTable
li r21, 0x00
add r20, r20, r19
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
bgel Local_Panic
Initial commit
2017-11-19 04:11:07 +00:00
stb r21, 0x0017(r30)
lwz r20, 0x0000(r20)
add r20, r20, r19
mtlr r20
stw r30, -0x0254(r18)
blr
TimerDispatch_0x144
mfsprg r18, 0
lwz r30, -0x0254(r18)
lbz r19, 0x0016(r30)
cmpwi r19, 0x01
lwz r8, 0x0000(r30)
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 TimerDispatch_0x30
Initial commit
2017-11-19 04:11:07 +00:00
bl DeleteID
mr r8, r30
bl PoolFree
lwz r8, 0x001c(r30)
cmpwi r8, 0x00
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 TimerDispatch_0x180
Initial commit
2017-11-19 04:11:07 +00:00
bl PoolFree
li r8, 0x00
stw r8, 0x001c(r30)
TimerDispatch_0x180:
mfsprg r18, 0
b TimerDispatch_0x30
TimerDispatch_0x188
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 r19, EWA.TimerDispatchLR(r18)
Initial commit
2017-11-19 04:11:07 +00:00
mtlr r19
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
b SetTimesliceFromCurTimeAndTripTime
Initial commit
2017-11-19 04:11:07 +00:00
StartTimeslicing ; OUTSIDE REFERER
mfsprg r19, 0
li r8, 1
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
stb r8, EWA.GlobalTimeIsValid(r19)
Initial commit
2017-11-19 04:11:07 +00:00
li r8, 0
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
stw r8, EWA.GlobalTime(r19)
stw r8, EWA.GlobalTime + 4(r19)
Initial commit
2017-11-19 04:11:07 +00:00
mflr r19
_log 'Starting timeslicing^n'
mtlr r19
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
; CLOB r8/r9, r16-r21
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
SetTimeslice
Initial commit
2017-11-19 04:11:07 +00:00
mflr r19
bl GetTime
mtlr r19
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
; ARG TimeBase r8/r9 curTime
; CLOB r16-r21
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
SetTimesliceFromCurTime
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
; This should get the most distant time???
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 r18, PSA.TimerQueue + LLL.Next(r1)
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
lwz r16, Timer.Time(r18)
lwz r17, Timer.Time+4(r18)
; ARG TimeBase r8/r9 curTime, TimeBase r16/r17 TripTime
; CLOB r18-r21
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
SetTimesliceFromCurTimeAndTripTime
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
mfxer r20
mfsprg r19, 0
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
lis r21, 0x7fff
lbz r18, EWA.GlobalTimeIsValid(r19)
ori r21, r21, 0xffff
cmpwi r18, 1
; r16/r17 = soonest(last timer, global PSA time if available)
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 global_time_invalid
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
lwz r18, EWA.GlobalTime(r19)
lwz r19, EWA.GlobalTime+4(r19)
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
_b_if_time_le r16, r18, last_timer_fires_sooner
Initial commit
2017-11-19 04:11:07 +00:00
mr r17, r19
mr r16, r18
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
last_timer_fires_sooner
global_time_invalid
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
; Subtract the current time (or what we were passed in r8/r9) from that time
Initial commit
2017-11-19 04:11:07 +00:00
subfc r17, r9, r17
subfe. r16, r8, r16
mtxer r20
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
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 @that_time_has_passed ; hi bit of r16 = 1
bne @that_time_is_in_future ;
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
cmplw r16, r21 ; typo? should be r17???
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 @that_time_is_in_future ; will never be taken...
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
; When the times are roughly equal?
Initial commit
2017-11-19 04:11:07 +00:00
mtspr dec, r17
blr
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
@that_time_is_in_future
Initial commit
2017-11-19 04:11:07 +00:00
mtspr dec, r21
blr
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
@that_time_has_passed
Initial commit
2017-11-19 04:11:07 +00:00
mtspr dec, r21
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
mtspr dec, r16 ; this makes nearly no sense!
Initial commit
2017-11-19 04:11:07 +00:00
blr
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
; TimerFire0
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFireUnknownKind
Initial commit
2017-11-19 04:11:07 +00:00
_log 'TimerInformation.kind is zero??^n'
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
; TimerFire1
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire1 ; OUTSIDE REFERER
Initial commit
2017-11-19 04:11:07 +00:00
bl Local_Panic
lwz r18, 0x0018(r30)
stw r16, 0x0080(r18)
stw r17, 0x0084(r18)
lwz r8, 0x0018(r30)
li r16, 0x00
lbz r17, 0x0018(r8)
lwz r19, 0x0088(r8)
cmpwi r17, 0x00
stw r16, 0x011c(r19)
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 TimerFire1_0x64
Initial commit
2017-11-19 04:11:07 +00:00
addi r16, r8, 0x08
Insert calls to known linked-list macros The presence of InitList, InsertAsNext, InsertAsPrev and RemoveFromList calls should make some hitherto unexplored code easier to read.
2017-12-13 04:10:54 +00:00
RemoveFromList r16, scratch1=r17, scratch2=r19
Initial commit
2017-11-19 04:11:07 +00:00
li r17, 0x01
stb r17, 0x0019(r8)
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
bl SchRdyTaskNow
Initial commit
2017-11-19 04:11:07 +00:00
bl CalculateTimeslice
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
bl FlagSchEvaluationIfTaskRequires
Initial commit
2017-11-19 04:11:07 +00:00
b TimerDispatch_0x144
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
TimerFire1_0x64
Initial commit
2017-11-19 04:11:07 +00:00
lwz r16, 0x0064(r8)
rlwinm. r16, r16, 0, 30, 30
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
; TimerFire2
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire2 ; OUTSIDE REFERER
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 TimerDispatch_0x144
Initial commit
2017-11-19 04:11:07 +00:00
bl Local_Panic
lwz r18, 0x0018(r30)
stw r16, 0x0080(r18)
stw r17, 0x0084(r18)
lwz r8, 0x0018(r30)
li r16, -0x7270
lbz r17, 0x0018(r8)
lwz r18, 0x0088(r8)
cmpwi r17, 0x00
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 TimerFire3_0x8
Initial commit
2017-11-19 04:11:07 +00:00
stw r16, 0x011c(r18)
lwz r8, 0x0008(r8)
lwz r8, 0x0000(r8)
; r8 = id
bl LookupID
cmpwi r9, Queue.kIDClass
cmpwi cr1, r9, 0x05
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 TimerFire2_0x8c
beq cr1, TimerFire2_0x7c
Initial commit
2017-11-19 04:11:07 +00:00
cmpwi r9, 0x09
cmpwi cr1, r9, 0x06
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 TimerFire2_0x6c
bne cr1, Local_Panic
Initial commit
2017-11-19 04:11:07 +00:00
lwz r16, 0x0020(r8)
addi r16, r16, -0x01
stw r16, 0x0020(r8)
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
b TimerFire2_0x98
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire2_0x6c
Initial commit
2017-11-19 04:11:07 +00:00
lwz r16, 0x001c(r8)
addi r16, r16, -0x01
stw r16, 0x001c(r8)
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
b TimerFire2_0x98
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire2_0x7c
Initial commit
2017-11-19 04:11:07 +00:00
lwz r16, 0x001c(r8)
addi r16, r16, -0x01
stw r16, 0x001c(r8)
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
b TimerFire2_0x98
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire2_0x8c
Initial commit
2017-11-19 04:11:07 +00:00
lwz r16, 0x002c(r8)
addi r16, r16, -0x01
stw r16, 0x002c(r8)
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
TimerFire2_0x98
Initial commit
2017-11-19 04:11:07 +00:00
lwz r8, 0x0018(r30)
addi r16, r8, 0x08
Insert calls to known linked-list macros The presence of InitList, InsertAsNext, InsertAsPrev and RemoveFromList calls should make some hitherto unexplored code easier to read.
2017-12-13 04:10:54 +00:00
RemoveFromList r16, scratch1=r17, scratch2=r18
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
bl SchRdyTaskNow
Initial commit
2017-11-19 04:11:07 +00:00
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
; TimerFire3
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire3 ; OUTSIDE REFERER
bl FlagSchEvaluationIfTaskRequires
Initial commit
2017-11-19 04:11:07 +00:00
b TimerDispatch_0x144
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
TimerFire3_0x8 ; OUTSIDE REFERER
Initial commit
2017-11-19 04:11:07 +00:00
b Local_Panic
; major_0x13258
; Dead code -- probably removed from TimerTable
lwz r8, 0x0018(r30)
; r8 = id
bl LookupID
cmpwi r9, Queue.kIDClass
mr r31, r8
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 major_0x13258_0x68
Initial commit
2017-11-19 04:11:07 +00:00
lwz r16, 0x0024(r31)
lwz r8, 0x001c(r30)
cmpwi r16, 0x00
cmpwi cr1, r8, 0x00
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 major_0x13258_0x40
Initial commit
2017-11-19 04:11:07 +00:00
lwz r17, 0x0028(r31)
mr. r8, r17
lwz r17, 0x0008(r17)
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 major_0x13258_0x68
Initial commit
2017-11-19 04:11:07 +00:00
stw r17, 0x0028(r31)
b major_0x13258_0x4c
major_0x13258_0x40
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 cr1, major_0x13258_0x68
Initial commit
2017-11-19 04:11:07 +00:00
li r16, 0x00
stw r16, 0x001c(r30)
major_0x13258_0x4c
lwz r16, 0x0020(r30)
lwz r17, 0x0024(r30)
lwz r18, 0x0028(r30)
stw r16, 0x0010(r8)
stw r17, 0x0014(r8)
stw r18, 0x0018(r8)
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
bl EnqueueMessage ; Message *r8, Queue *r31
Initial commit
2017-11-19 04:11:07 +00:00
major_0x13258_0x68
lwz r8, 0x0034(r30)
; r8 = id
bl LookupID
cmpwi r9, Semaphore.kIDClass
mr r31, r8
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 major_0x13258_0x80
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
bl SignalSemaphore
Initial commit
2017-11-19 04:11:07 +00:00
major_0x13258_0x80
lwz r8, 0x002c(r30)
; r8 = id
bl LookupID
cmpwi r9, EventGroup.kIDClass
mr r31, r8
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
; TimerFire4
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire4 ; OUTSIDE REFERER
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 TimerFire4_0xc
Initial commit
2017-11-19 04:11:07 +00:00
lwz r8, 0x0030(r30)
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
bl SetEvent
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire4_0xc
Initial commit
2017-11-19 04:11:07 +00:00
b TimerDispatch_0x144
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
TimerFire4_0x10 ; OUTSIDE REFERER
Initial commit
2017-11-19 04:11:07 +00:00
mfsprg r28, 0
lwz r29, -0x0008(r28)
mr r8, r29
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
bl SchTaskUnrdy
Initial commit
2017-11-19 04:11:07 +00:00
lbz r17, 0x0019(r29)
cmpwi r17, 0x02
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 TimerFire4_0x64
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r29
lwz r16, 0x0038(r30)
lwz r17, 0x003c(r30)
bl clear_cr0_lt
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 TimerFire4_0x50
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r29
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
bl SchRdyTaskNow
Initial commit
2017-11-19 04:11:07 +00:00
bl CalculateTimeslice
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
b TimerFire5_0x8
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire4_0x50
Initial commit
2017-11-19 04:11:07 +00:00
li r18, 0x02
stb r18, 0x0019(r29)
mr r8, r29
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
bl SchRdyTaskNow
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
b TimerFire5_0x8
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire4_0x64
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r29
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
; TimerFire5
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire5 ; OUTSIDE REFERER
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
bl SchRdyTaskNow
Initial commit
2017-11-19 04:11:07 +00:00
bl major_0x149d4
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
TimerFire5_0x8 ; OUTSIDE REFERER
bl FlagSchEvaluationIfTaskRequires
Initial commit
2017-11-19 04:11:07 +00:00
mfsprg r18, 0
b TimerDispatch_0x30
; major_0x13364
; Dead code -- probably removed from TimerTable
_log 'Heartbeat: Ext '
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
lwz r16, KDP.NanoKernelInfo + NKNanoKernelInfo.ExternalIntCount(r1)
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r16
bl printd
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
_log 'Alerts '
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
lwz r16, KDP.NanoKernelInfo + NKNanoKernelInfo.AlertCount(r1)
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r16
bl printd
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
_log 'Blue cpu-'
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
lwz r17, PSA.PA_BlueTask(r1)
lhz r16, Task.CPUIndex(r17)
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r16
bl printb
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
_log 'state-'
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
lbz r16, Task.State(r17)
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r16
bl printb
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
_log 'scr-'
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 r16, KDP.PA_ECB(r1)
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
lwz r18, KDP.PostIntMaskInit(r1)
lwz r16, ContextBlock.CR(r16)
Initial commit
2017-11-19 04:11:07 +00:00
and r16, r16, r18
mr r8, r16
bl printw
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
_log 'mcr-'
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
lwz r16, PSA.MCR(r1)
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r16
bl printw
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
_log 'IPL-'
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
lwz r16, KDP.PA_EmulatorIplValue(r1)
lhz r16, 0(r16)
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r16
bl printh
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
_log 'eSR-'
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 r16, KDP.PA_ECB(r1)
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
lwz r16, ContextBlock.r25(r16)
andi. r16, r16, 7
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r16
bl printb
_log '^n'
Reverse power management functions This is part of the work to get the mini running well. The kcPowerDispatch and kcCacheDispatch entry points were explored. The dead code implementing a Timer "Heartbeat" was also used to label some structures better.
2018-03-09 11:46:39 +00:00
Initial commit
2017-11-19 04:11:07 +00:00
mfxer r19
lwz r16, 0x0038(r30)
lwz r17, 0x003c(r30)
lwz r18, 0x0f2c(r1)
slwi r18, r18, 3
addc r17, r17, r18
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
; TimerFire7
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire7 ; OUTSIDE REFERER
Initial commit
2017-11-19 04:11:07 +00:00
addze r16, r16
stw r16, 0x0038(r30)
stw r17, 0x003c(r30)
mtxer r19
mr r8, r30
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
bl EnqueueTimer
Initial commit
2017-11-19 04:11:07 +00:00
b TimerDispatch_0x144
; major_0x134d8
; Dead code -- probably removed from TimerTable
Identify all accesses to the Primary System Area Whenever a field was added to the end of the Primary System Area (PSA) record, the offsets of all other fields would change. When reversing NK version 02.27, I noticed that 32 bytes had been added to the end of the PSA between 02.27 and 02.28. To build a byte-perfect 02.27, all references to PSA fields must therefore use the record definition in NKPublic.s instead of a numeric offset. In this commit, all PSA references are identified by field name (_FFF if the field has not yet been reversed).
2018-04-22 09:13:30 +00:00
lwz r18, PSA.DecClockRateHzCopy(r1)
Initial commit
2017-11-19 04:11:07 +00:00
lwz r19, 0x0f88(r1)
subf. r19, r18, r19
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
ble TimerFire8_0x1c
Initial commit
2017-11-19 04:11:07 +00:00
srwi r19, r19, 11
mfxer r20
major_0x134d8_0x18
mftbu r16
mftb r17, 0x10c
mftbu r18
cmpw r16, r18
li r18, 0x00
bne- major_0x134d8_0x18
mttb r18
addc r17, r17, r19
addze r16, r16
mttbu r16
mttb r17
Identify all accesses to the Primary System Area Whenever a field was added to the end of the Primary System Area (PSA) record, the offsets of all other fields would change. When reversing NK version 02.27, I noticed that 32 bytes had been added to the end of the PSA between 02.27 and 02.28. To build a byte-perfect 02.27, all references to PSA fields must therefore use the record definition in NKPublic.s instead of a numeric offset. In this commit, all PSA references are identified by field name (_FFF if the field has not yet been reversed).
2018-04-22 09:13:30 +00:00
lwz r18, PSA.DecClockRateHzCopy(r1)
Initial commit
2017-11-19 04:11:07 +00:00
srwi r18, r18, 11
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
; TimerFire8
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire8 ; OUTSIDE REFERER
Initial commit
2017-11-19 04:11:07 +00:00
addc r17, r17, r18
addze r16, r16
stw r16, 0x0038(r30)
stw r17, 0x003c(r30)
mtxer r20
mr r8, r30
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
bl EnqueueTimer
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire8_0x1c ; OUTSIDE REFERER
Initial commit
2017-11-19 04:11:07 +00:00
b TimerDispatch_0x144
; major_0x13544
; Dead code -- probably removed from TimerTable
Identify all accesses to the Primary System Area Whenever a field was added to the end of the Primary System Area (PSA) record, the offsets of all other fields would change. When reversing NK version 02.27, I noticed that 32 bytes had been added to the end of the PSA between 02.27 and 02.28. To build a byte-perfect 02.27, all references to PSA fields must therefore use the record definition in NKPublic.s instead of a numeric offset. In this commit, all PSA references are identified by field name (_FFF if the field has not yet been reversed).
2018-04-22 09:13:30 +00:00
lwz r19, PSA._36c(r1)
Initial commit
2017-11-19 04:11:07 +00:00
mfxer r20
cmpwi cr1, r19, 0x00
srawi r8, r19, 31
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 cr1, TimerFire6_0x4
Initial commit
2017-11-19 04:11:07 +00:00
major_0x13544_0x14
mftbu r16
mftb r17, 0x10c
mftbu r18
cmpw r16, r18
li r18, 0x00
bne- major_0x13544_0x14
mttb r18
addc r19, r17, r19
adde r18, r16, r8
mttbu r18
mttb r19
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 cr1, major_0x13544_0x64
Initial commit
2017-11-19 04:11:07 +00:00
major_0x13544_0x44
mftbu r18
mftb r19, 0x10c
mftbu r8
cmpw r18, r8
bne- major_0x13544_0x44
subfc r19, r17, r19
subfe. r18, r16, r18
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 major_0x13544_0x44
Initial commit
2017-11-19 04:11:07 +00:00
major_0x13544_0x64
Identify all accesses to the Primary System Area Whenever a field was added to the end of the Primary System Area (PSA) record, the offsets of all other fields would change. When reversing NK version 02.27, I noticed that 32 bytes had been added to the end of the PSA between 02.27 and 02.28. To build a byte-perfect 02.27, all references to PSA fields must therefore use the record definition in NKPublic.s instead of a numeric offset. In this commit, all PSA references are identified by field name (_FFF if the field has not yet been reversed).
2018-04-22 09:13:30 +00:00
lwz r18, PSA._368(r1)
Initial commit
2017-11-19 04:11:07 +00:00
addc r17, r17, r18
addze r16, r16
stw r16, 0x0038(r30)
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
; TimerFire6
Initial commit
2017-11-19 04:11:07 +00:00
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
TimerFire6 ; OUTSIDE REFERER
Initial commit
2017-11-19 04:11:07 +00:00
stw r17, 0x003c(r30)
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
TimerFire6_0x4 ; OUTSIDE REFERER
Initial commit
2017-11-19 04:11:07 +00:00
mtxer r20
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 cr1, TimerDispatch_0x144
Initial commit
2017-11-19 04:11:07 +00:00
mr r8, r30
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
bl EnqueueTimer
Initial commit
2017-11-19 04:11:07 +00:00
b TimerDispatch_0x144
; major_0x135d0
; Dead code -- probably removed from TimerTable
mfxer r19
lwz r16, 0x0038(r30)
lwz r17, 0x003c(r30)
lwz r18, 0x0f2c(r1)
srwi r18, r18, 1
addc r17, r17, r18
addze r16, r16
stw r16, 0x0038(r30)
stw r17, 0x003c(r30)
mtxer r19
mr r8, r30
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
bl EnqueueTimer
Initial commit
2017-11-19 04:11:07 +00:00
bl getchar
cmpwi r8, -0x01
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 TimerDispatch_0x144
Initial commit
2017-11-19 04:11:07 +00:00
bl panic_non_interactive
b TimerDispatch_0x144
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
; ARG Timer *r8
; CLOB r16-r20
EnqueueTimer ; OUTSIDE REFERER
; Keep the trip-time of this timer in r16/r17
lwz r16, Timer.Time(r8)
lwz r17, Timer.Time+4(r8)
Initial commit
2017-11-19 04:11:07 +00:00
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
; r20 = timer being considered
; r18/r19 = trip-time of timer being condidered
Initial commit
2017-11-19 04:11:07 +00:00
lwz r20, PSA.TimerQueue + TimerQueueStruct.LLL + LLL.Next(r1)
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
lwz r18, Timer.Time(r20)
lwz r19, Timer.Time+4(r20)
; First try to insert at head of global TMRQ
Initial commit
2017-11-19 04:11:07 +00:00
cmpw r16, r18
cmplw cr1, r17, r19
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 @insert_further_ahead
blt @insert_at_tail
bge cr1, @insert_further_ahead
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
@insert_at_tail
addi r20, r1, PSA.TimerQueue + TimerQueueStruct.LLL
li r18, 1
stb r18, Timer.Byte3(r8)
; Insert at the very back of the queue
lwz r19, LLL.Freeform(r8)
lwz r9, LLL.Freeform(r20)
stw r9, LLL.Freeform(r8) ; my freeform = considered freeform
lwz r9, LLL.Next(r20)
stw r9, LLL.Next(r8) ; my next = next of considered
stw r20, LLL.Prev(r8) ; my prev = considered
stw r8, LLL.Prev(r9) ; prev of next of considered = me
stw r8, LLL.Next(r20) ; next of considered = me
stw r19, LLL.Freeform(r8) ; my freeform = my original freeform
Work on the scheduler and interrupt handling code
2018-03-23 09:23:06 +00:00
b SetTimeslice
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
@insert_further_ahead
lwz r20, PSA.TimerQueue + TimerQueueStruct.LLL + LLL.Prev(r1)
@searchloop
lwz r18, Timer.Time(r20)
lwz r19, Timer.Time+4(r20)
Initial commit
2017-11-19 04:11:07 +00:00
cmpw r16, r18
cmplw cr1, r17, r19
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 @insert_after_this_one
blt @next
bge cr1, @insert_after_this_one
Reverse Event Group MP calls
2017-11-19 09:59:35 +00:00
@next
lwz r20, LLL.Prev(r20)
b @searchloop
@insert_after_this_one
li r18, 1
stb r18, Timer.Byte3(r8)
lwz r19, LLL.Freeform(r8)
lwz r9, LLL.Freeform(r20)
stw r9, LLL.Freeform(r8) ; my freeform = considered freeform
lwz r9, LLL.Next(r20)
stw r9, LLL.Next(r8) ; my next = next of considered
stw r20, LLL.Prev(r8) ; my prev = considered
stw r8, LLL.Prev(r9) ; prev of next of considered = me
stw r8, LLL.Next(r20) ; next of considered = me
stw r19, LLL.Freeform(r8) ; my freeform = my original freeform
Initial commit
2017-11-19 04:11:07 +00:00
blr
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
; Remove a Timer from the global timer firing queue (TMRQ).
; If the Timer was to be the next to fire, then perform the
; standard decrementer rollover adjustment.
Initial commit
2017-11-19 04:11:07 +00:00
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
; ARG Timer *r8
Initial commit
2017-11-19 04:11:07 +00:00
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
DequeueTimer
lwz r16, Timer.QueueLLL + LLL.FreeForm(r8)
cmpwi r16, 0
lwz r18, PSA.TimerQueue + TimerQueueStruct.LLL + LLL.Next(r1)
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 Local_Panic
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
Insert calls to known linked-list macros The presence of InitList, InsertAsNext, InsertAsPrev and RemoveFromList calls should make some hitherto unexplored code easier to read.
2017-12-13 04:10:54 +00:00
RemoveFromList r8, scratch1=r16, scratch2=r17
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
li r16, 0
Initial commit
2017-11-19 04:11:07 +00:00
cmpw r18, r8
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
stb r16, Timer.Byte3(r8)
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 SetTimeslice
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
Initial commit
2017-11-19 04:11:07 +00:00
blr
; Get the number of timebase ticks in a specified period
; ARG long r8 period (positive for ms, negative for us)
TimebaseTicksPerPeriod
mr. r17, r8
li r19, 250
lwz r9, KDP.ProcessorInfo + NKProcessorInfo.DecClockRateHz(r1)
bgt+ @period_positive
blt+ @period_negative
li r8, 0
li r9, 0
blr ; fail
@period_negative
neg r17, r17
lisori r19, 250000
@period_positive
divw r19, r9, r19
mullw r9, r19, r17
mulhw r8, r19, r17
srwi r9, r9, 2
rlwimi r9, r8, 30, 0, 1
srwi r8, r8, 2
blr
; RET long r8 tbu, long r9 tbl
; CLOB r16, r17
GetTime
mfpvr r8
rlwinm. r8, r8, 0, 0, 14
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 @is_601
Initial commit
2017-11-19 04:11:07 +00:00
@retry_timebase:
mftbu r8
mftb r9
mftbu r16
cmpw r8, r16
bne- @retry_timebase
b @return
@is_601
dialect POWER ; disassembled this in POWER mode!
@retry_rtc
mfrtcu r8
mfrtcl r9
mfrtcu r16
cmp 0, r8, r16
dialect PowerPC
bne- @retry_rtc ; POWER chokes on hints?
dialect POWER
liu r16, 1000000000 >> 16
oril r16, r16, 1000000000 & 0xffff
mfmq r17
mul r8, r16, r8
mfmq r16
mtmq r17
mfxer r17
a r9, r16, r9
aze r8, r8
mtxer r17
dialect POWERPC
@return
blr
Reference in New Issue Copy Permalink