Stefanos Kornilios Mitsis Poiitidis
7a9a95583c
Merge pull request #190 from Sonicadvance1/fix_128bit_div
...
Fixes 128bit divisions on AArch64
2020-05-26 09:34:32 +03:00
Stefanos Kornilios Mitsis Poiitidis
ec8eac2836
Merge pull request #197 from Sonicadvance1/remove_xop
...
Removes XOP frontend handling
2020-05-25 17:16:44 +03:00
Stefanos Kornilios Mitsis Poiitidis
a69a821773
Merge pull request #192 from Sonicadvance1/fix_aarch64_llvm9
...
Fixes LLVM on AArch64
2020-05-25 17:06:28 +03:00
Ryan Houdek
45bf39ca0f
Removes XOP frontend handling
...
This keeps the XOP instruction tables around but removes the decoding
assert.
This lets us implement the POP instruction that overlapped the XOP
prefix without having to deal with XOP at all.
I don't believe we have any real need to support XOP, but if it is a
thing later then it'll be easy to add back
2020-05-25 00:53:14 -07:00
Ryan Houdek
7c98f5a3ac
Fixes LLVM on AArch64
...
LLVM-9 doesn have the include, while LLVM-10 does
2020-05-24 00:19:54 -07:00
Ryan Houdek
61c0fde8fe
Fixes ASHR on AArch64
...
Smaller than 8byte wasn't sexting to the full register so it was causing issues
2020-05-24 00:18:51 -07:00
Ryan Houdek
600ad7f597
Fixes 128bit divisions on AArch64
...
These were only doing 64bit divisions.
Function calls are a bit ugly but once the new RA lands we can clean them up
2020-05-23 23:45:52 -07:00
Ryan Houdek
64556e3eac
Ensures passmanager gets the syscall handler during core init
2020-05-14 01:04:04 -07:00
Ryan Houdek
d2cf07a4fd
Passes the running mode through from FEX to FEXCore
2020-05-14 00:56:31 -07:00
Stefanos Kornilios Mitsis Poiitidis
a65bb87fa8
Merge pull request #183 from Sonicadvance1/fix_frontend_imm
...
Fixes a bug in the frontend regarding constants
2020-05-12 19:52:43 +03:00
Ryan Houdek
0b88984005
Fixes a bug in the frontend regarding constants
...
Optimization passes found a bug in the frontend regarding constants.
Specifically in the case of instructions that have a 32bit constant that
are sign extended to 64bit was being incorrectly set to have a size of
32bits (the size of the constant).
So an optimization pass would rematerialize constants of the expected
size, thus ~0ULL would turn in to ~0U and generate invalid code.
Now watch the destination size and set the constant to that size, which
matches what its size will be.
2020-05-11 16:10:50 -07:00
Ryan Houdek
e6e4634651
Ensures LLVM JIT doesn't crash if DebugInfo isn't provided
...
IR doesn't have DebugInfo, all the other backends already handle this
2020-05-09 12:21:54 -07:00
Ryan Houdek
718b3e6b4c
Refactors PassManager to ingest IREmitter class instead of OpDispatcher
...
The Passmanager passes need very little or zero x86 knowledge to do
their work. They don't need the x86 OpDispatcher handling code at all.
Convert this entirely over to using the IREmitter class so it can
actually optimize IR code from the IRLoader frontend
2020-05-09 12:14:59 -07:00
Ryan Houdek
dc87134af3
Switches the OpcodeDispatcher over to using the new IREmitter class
2020-05-08 19:08:14 -07:00
Ryan Houdek
c25f2845d3
Adds the ability for the frontend codeloader to load IR
...
This will be used by the IR unit test loader.
Will give the backend IR so it doesn't even parse any x86 code
2020-05-07 17:40:29 -07:00
Ryan Houdek
66dc5ebc54
Merge pull request #179 from FEX-Emu/skmp/fix-exit-code
...
Propagate guest exit code
2020-05-01 16:58:07 -07:00
Stefanos Kornilios Mitsis Poiitidis
d58961ef1e
Propagate guest exit code
2020-05-02 02:34:50 +03:00
Stefanos Kornilios Mitsis Poiitidis
3c9be4625c
Add silent log
2020-05-02 01:45:22 +03:00
Ryan Houdek
0d89de2799
Merge pull request #175 from Sonicadvance1/xmm_to_mmx
...
Implements three XMM <-> MMX conversion instructions
2020-05-01 00:52:46 -07:00
Scott Mansell
22a557b2db
Remove stray debug code
2020-05-01 19:12:00 +12:00
Ryan Houdek
deb5bd8fe8
Implements three XMM <-> MMX conversion instructions
...
CVTPI2PS does MMX->XMM integer to float conversion with lower 64bit
insert
CVTPS2PI/CVTTPS2PI does an XMM->XMM float to integer conversion
2020-04-29 19:03:48 -07:00
Ryan Houdek
9848f7eac8
Implements the XLAT instruction
...
This is a fairly straightforward instruction, it's just <segment>:[rBX + AL]
Needs the previous changes since we weren't handling GS correctly
2020-04-29 18:26:03 -07:00
Ryan Houdek
6a98e93666
Implements the segment reading and writing instructions
2020-04-29 18:25:46 -07:00
Ryan Houdek
a0f412d85a
Update the new register segment reading and writing table flags
...
These are instructions that can executed at any CPL level
2020-04-29 18:25:46 -07:00
Ryan Houdek
53798f194e
Fix the Frontend's handling of the GS prefix
...
We were accidently treating it as FS each time it was encountered
2020-04-29 18:25:46 -07:00
Ryan Houdek
e49a66f740
Implements support for the three loop instructions
...
loop, loope, loopne
2020-04-29 17:20:09 -07:00
Ryan Houdek
8f2f046fc8
Merge pull request #157 from Sonicadvance1/rcp
...
Implements support for rcpps/rcpss
2020-04-29 16:03:38 -07:00
Ryan Houdek
056d017ac8
Merge pull request #160 from Sonicadvance1/horizontal_arith
...
Implements four new vector horizontal ALU ops
2020-04-29 12:56:26 -07:00
Ryan Houdek
f99482ed84
Implements four new vector horizontal ALU ops
...
haddps, hsubps, haddpd, hsubpd
2020-04-29 12:47:27 -07:00
Ryan Houdek
f0229fba38
Implements the two new ops in the LLVM JIT
2020-04-29 12:47:27 -07:00
Ryan Houdek
f6814de357
Implements the two new ops in the x86-64 JIT
2020-04-29 12:47:27 -07:00
Ryan Houdek
1645e65ff3
Implements the two new ops in the AArch64 JIT
2020-04-29 12:47:27 -07:00
Ryan Houdek
1c572e71c7
Implements the two new ops in the interpreter
2020-04-29 12:47:27 -07:00
Ryan Houdek
fee4aeaf88
Merge pull request #167 from Sonicadvance1/fix_exec
...
Fixes loading EXEC applications in to memory.
2020-04-29 12:31:33 -07:00
Ryan Houdek
2005a2ea3d
Merge pull request #164 from Sonicadvance1/fwait
...
Implements fwait as a nop
2020-04-29 12:30:45 -07:00
Ryan Houdek
f1403e5f54
Merge pull request #161 from Sonicadvance1/jrcxz
...
Implements JRCXZ in the OpcodeDispatcher
2020-04-29 12:30:25 -07:00
Ryan Houdek
3d782f0930
Merge pull request #156 from Sonicadvance1/movsldup
...
Implements support for movsldup
2020-04-29 12:27:37 -07:00
Ryan Houdek
169875c34d
Merge pull request #155 from Sonicadvance1/non_temporal
...
Implements two non-temporal XMM move instructions
2020-04-29 12:27:22 -07:00
Ryan Houdek
536eb7ec3e
Merge pull request #154 from Sonicadvance1/more_satalu
...
Implements six more saturating arithmetic instructions
2020-04-29 12:27:10 -07:00
Ryan Houdek
1ee9070e8f
Merge pull request #150 from Sonicadvance1/fix_x86_code_overflow
...
Fixes crash that occurs when the x86-64 JIT code cache gets full
2020-04-29 12:26:11 -07:00
Ryan Houdek
a0537ed630
Merge pull request #149 from Sonicadvance1/fix_syscall_memleak
...
Fixes memory leak on shutdown for syscall handlers
2020-04-29 12:25:55 -07:00
Ryan Houdek
2f56c475e2
Fixes memory leak on shutdown for syscall handlers
...
Switches it over to a unique_ptr to ensure it doesn't happen again.
2020-04-26 18:15:52 -07:00
Ryan Houdek
3528b30684
Fixes loading EXEC applications in to memory.
...
If we are loading EXEC then we actually need to start loading in to the
lower 32bits of memory.
This lets the frontend code loader handle the memory base instead of
hiding it behind some core magic instead.
This lets us get EXEC programs working.
So now we support EXEC, DYN, and static applications
2020-04-26 18:11:48 -07:00
Ryan Houdek
f3f63251ea
Merge pull request #124 from Sonicadvance1/fix_cache_clear_crash
...
Fixes crashes when code cache is cleared
2020-04-26 18:02:54 -07:00
Ryan Houdek
ec87131939
Implements fwait as a nop
...
Since we don't yet support instruction exceptions this is nothing but a
nop.
2020-04-25 23:53:30 -07:00
Ryan Houdek
c726e3ef51
Implements JRCXZ in the OpcodeDispatcher
...
This only operates on 64bit and 32bit RCX register. the 16bit encoding
isn't possible in 64bit mode.
2020-04-25 22:06:46 -07:00
Ryan Houdek
ad467c5b30
Adds new vector float negate and float add horizontal IR ops
2020-04-25 17:55:37 -07:00
Ryan Houdek
4ae33147f5
Implements support for rcpps/rcpss
2020-04-24 23:35:33 -07:00
Ryan Houdek
a0c0384340
Ensures reciprocal hits the precision we expect in the x86-64 JIT
...
Otherwise unit tests can fail
2020-04-24 23:34:47 -07:00
Ryan Houdek
73632556ce
Implements support for movsldup
2020-04-24 23:13:51 -07:00