Commit Graph

713 Commits

Author SHA1 Message Date
Ryan Houdek
ab2b499fa6 [ARMv7] Minor integer fastmem optimization.
This is a one instruction optimization for integer loadstores.
Makes sure to enable nop padding in some cases where a fault can still happen and cause us to overwrite other instructions that aren't meant to be.
2014-12-20 19:41:33 -06:00
Pierre Bourdon
24a2ca4d28 Merge pull request #1703 from FioraAeterna/saveregs
JIT: free up a register by eliminating RCODE_POINTERS
2014-12-20 16:45:08 +01:00
degasus
5f322cc9e8 Jit: move constants out of class
This fixes a linking issues on debug builds.
2014-12-17 23:42:10 +01:00
skidau
5313dd1b8c Merge pull request #1702 from CarlKenner/DebugSymbols
Fix many bugs with the Symbols menu (when run with -d argument).
2014-12-17 22:04:56 +11:00
CarlKenner
6e7eb52211 Rename menu items, stop using PanicAlertT, remove -d when running from visual studio.
I accidentally committed my addition of the -d option that I used for testing, now it's fixed.
2014-12-17 01:12:33 +10:30
CarlKenner
84ae817c18 Remove white space at ends of lines. 2014-12-17 01:12:08 +10:30
CarlKenner
cd5d354acc Load 4 column map files like in American Mensa Academy.
Unfortunately the map files in American Mensa Academy don't correspond to the release version.
But at least now if other games use those map files we will be able to load them.
2014-12-17 01:12:03 +10:30
CarlKenner
dfd915eb53 Merge bad map file loading into the original function. 2014-12-17 01:11:58 +10:30
CarlKenner
e246aaf419 Add "Load bad map file" option for map files on disc that don't quite match.
Currently it is very simple and naive, but filters out most of the bad matches.
2014-12-17 01:11:41 +10:30
CarlKenner
f54d9e33c2 Fix many bugs with the Symbols menu (when run with -d argument).
The Symbols menu is now fully useable.
2014-12-17 01:11:36 +10:30
skidau
ce8b3d83a8 Merge pull request #1693 from magumagu/memcheck-fastmem
JIT: use fastmem loads in MMU mode.
2014-12-16 13:07:53 +11:00
Fiora
76ca42b758 JIT: free up a register by eliminating RCODE_POINTERS
Also use shorter code in the dispatcher if we can get away with it (e.g. on
Windows where the relevant memory is in the low 2GB).
2014-12-14 18:19:37 -08:00
skidau
d02eb3ca59 Merge pull request #1640 from rohit-n/switch-default
Silence some -Wswitch-default warnings.
2014-12-13 19:47:06 +11:00
magumagu
e479606b07 JIT: simplify code using MOVTwo. 2014-12-12 22:29:49 -08:00
magumagu
43d56febc4 JIT: use fastmem loads in MMU mode.
Even in games that require MMU mode, loads outside the area specified by
the BAT are rare, so fastmem is a substantial improvement.

All of the interesting changes are in the backpatch handler, to make it
generate DSI exceptions correctly.
2014-12-12 22:08:46 -08:00
Rohit Nirmal
b030d29067 Silence some -Wswitch-default warnings. 2014-12-11 22:23:05 -06:00
skidau
110d079c1e Enable block-linking with MMU. Blocks that cause an ISI exception are not linked. 2014-12-12 13:25:51 +11:00
Fiora
92aad1140e JIT: fix MOVD_xmm typoes in ConvertDoubleToSingle
Ordering, plus MOVQ instead of MOVD
2014-12-09 04:19:44 -08:00
skidau
1ce3696e8b Merge pull request #1623 from FioraAeterna/fixppcfp
JIT: fix PPC_FP snan/qnan handling
2014-12-09 14:25:23 +11:00
Tillmann Karras
6e928d08a0 Jit64: clean up casts in memory operands 2014-12-07 20:23:27 +01:00
Ryan Houdek
6d6b425061 Merge pull request #1647 from magumagu/mmu-flag-cleanup
Cleanup: remove unnecessary uses of the bMMU flag
2014-12-07 01:08:30 -06:00
Ryan Houdek
b9c6ee93b6 Merge pull request #1661 from FioraAeterna/fixloadstorebind
JIT: fix possible panicalert in loadstore
2014-12-07 00:54:59 -06:00
Fiora
68972da1e5 JIT: fix possible panicalert in loadstore
Didn't bind address register tcorrectly in a very rare case.
2014-12-06 16:09:13 -08:00
Fiora
3d12849967 JIT: fix snan/qnan handling in float loads/stores
Also simplify the conversion code with some suggestions by flacs; might even
be slightly faster now despite handling more cases.
2014-12-06 13:07:24 -08:00
skidau
d485acdb51 Stored a copy of the PTE in the TLB like the real hardware does.
Updated PTE.R bit on Write and Instruction fetch.
Added code to read the PTE from MEM2 if the PTE is stored there.
Refactored the two hash functions to reduce code duplication.
Updated save state version.
2014-12-06 10:28:34 +11:00
skidau
997681b65a Removed the tag check in InvalidateTLBEntry. All four TLB entries are always cleared on each invalidate command.
Initialised the TLB cache to start from a consistent state on reset.
2014-12-05 19:56:45 +11:00
skidau
693f413364 Updated C bit on TLB cache hits.
Added TLB state to the save state file.
2014-12-05 14:29:13 +11:00
Ryan Houdek
d8cb976bba Merge pull request #1639 from Sonicadvance1/aarch64_improvements
Aarch64 improvements
2014-12-03 19:15:55 -06:00
Lioncash
c715a4f8ef Jit: Remove unnecessary include 2014-12-03 17:30:57 -05:00
magumagu
6b34b1b0d0 JIT: remove unnecessary uses of MMU flag. 2014-12-03 14:06:06 -08:00
Lioncash
241c94d337 Merge pull request #1645 from magumagu/jit-unused-flag
JIT: remove unused flag.
2014-12-03 16:13:31 -05:00
Ryan Houdek
71e4e67ae1 [AArch64] Only flush registers that are needed with interpreter fallback.
We try to keep as many registers as possible in callee saved registers, so if we have guest registers in the correct registers and the interpreter
call we are falling back to doesn't need the registers then we can dump just those ones. Which means we don't have to dump 100% of our register state
when falling to the interpreter.
2014-12-02 21:08:38 -06:00
Ryan Houdek
71b77f3173 [AArch64] Makes some integer instructions more clear what they're doing.
ComputeRC was a bit unclear by using 64bit registers for setting the immediate and then calling SXTW on a 6b4it register which is just a bit obscure.
When the source register is an immediate in cntlzwx, just use the built in GCC function instead of our own implementing for counting leading zeros.
2014-12-02 21:08:38 -06:00
Ryan Houdek
8dfb8d8ad5 [AArch64] Implements HLE function injection 2014-12-02 21:08:38 -06:00
Ryan Houdek
51ad798105 [AArch64] Implements block linking.
Before block linking was enabled but it wasn't ever implemented.
Implements link blocks and destroy block functions and moves the downcount check in the WriteExit function so it doesn't get overwritten when linking.
2014-12-02 21:08:38 -06:00
Ryan Houdek
ca04601b14 [AArch64] Fixes the dispatcher
Changes the dispatcher to make sure to we are saving the LR(X30) to the stack. Also makes sure to keep the stack aligned.
AArch64's AAPCS64 mandates the stack to be quad-word aligned.

Fixes the dispatcher from infinite looping due to a downcount check jumping to the dispatcher. This was because checking exceptions and the state
pointer wouldn't reset the global conditional flags. So it would leave the timing/exception, jump to the start of the dispatcher and then jump back
again due to the conditional branch.
2014-12-02 21:08:38 -06:00
Ryan Houdek
c3c80e9440 [AArch64] Improves the register cache.
Removes the REG_AWAY nonsense I was doing. I've got to get the JIT more up to speed before thinking of insane register cache things.
Also fixes a bug in immediate setting where if the register being set to an immediate already had a host register tied to it then it wouldn't free the
register it had. Resulting in register exhaustion.
2014-12-02 21:08:38 -06:00
Ryan Houdek
08660c89ad Fix register usage detection in PPCAnalyst.
lmw/stmw weren't properly setting input and output registers since they use multiple registers.
dcbz was just missing a flag in the instruction tables.
2014-12-02 16:12:33 -06:00
Fiora
863fb91cad JIT: fix profiler call
Sometimes this seems to require a 64-bit offset, so use CallFunction instead
2014-12-02 10:17:58 -08:00
Lioncash
af5b1063bf Merge pull request #1619 from FioraAeterna/removepxor
JIT: remove unnecessary pxor in paired stores
2014-12-01 15:50:31 -05:00
Ryan Houdek
4946c6876d Merge pull request #1607 from Sonicadvance1/armv7-optimizations
A whole bunch of ARMv7 optimizations and minor fixes.
2014-11-30 15:48:27 -06:00
Fiora
644a661bf1 JIT: remove unnecessary pxor in paired stores
I don't think this is necessary anymore?
2014-11-30 08:11:45 -08:00
Ryan Houdek
cd13d2d66f [ARM32] Fix a couple bugs in the paired loadstore routines.
This code was obviously wrong, we were sign extending 8 bit unsigned values and loading from the wrong offset as well.
This fixes a bug in Muramasa where some colours were going insane.
2014-11-30 04:10:17 +00:00
Ryan Houdek
b848365f78 [ARM32] Minor optimization in paired loadstores.
When the offset can fit in the instruction encoding make sure to do so.
2014-11-30 04:10:16 +00:00
Ryan Houdek
52c6fb180b [ARM32] Moves loadstore extend flag to backpatch code. 2014-11-30 04:10:16 +00:00
Ryan Houdek
32dc105aa3 [ARM32] Eat a register to store our memory base.
This saves at least two instructions per fastmem operation.
2014-11-30 04:10:16 +00:00
Ryan Houdek
e2f8286415 [ARM32] Adds unsigned compares.
Superscedes PR #1131.
Optimizes cmpi a bit.
2014-11-30 04:10:16 +00:00
Ryan Houdek
581ab9edec [ARM32] cntlzwx optimization if source is immediate 2014-11-30 04:10:16 +00:00
Ryan Houdek
9c82adb14f [ARM32] negx optimization if source is immediate 2014-11-30 04:10:15 +00:00
Ryan Houdek
1dbb39f791 [ARM32] srawix optimization when source is an immediate. 2014-11-30 04:10:15 +00:00