14bf4912d1
Fixing CF decode.
2013-10-09 21:15:23 -07:00
7b97d475c2
Hacked in shader disassembler.
...
Checking this in before I break it. Has tons of issues.
Turns out the 360 GPU uses an Adreno (rAdeon)-like r2xx shader format.
VC++'s handling of bitfields is absolutely terrible.
2013-10-09 20:28:35 -07:00
17be429244
Microcode disassembler skeleton.
2013-10-07 19:58:24 -07:00
651954ccae
Starting to drive command buffer actions down to a graphics driver.
2013-10-06 21:09:58 -07:00
371075f154
More GPU decoding.
2013-10-06 19:22:02 -07:00
99bde2d67e
Fixing terrible vsldoi implementation. Still terrible.
2013-10-06 17:45:59 -07:00
87481f5f4c
Fixing vperm.
2013-10-06 16:25:41 -07:00
b77403e754
Fixing lvsl/lvsr.
2013-10-06 16:25:23 -07:00
74692232c9
Add trace thread mask.
...
Makes seeing only thread 1 easier.
2013-10-06 14:35:03 -07:00
1fecb95512
Switching altivec disasm to die on failure.
2013-10-06 12:35:24 -07:00
debf21437e
Untested emulated stvlx/stvrx.
2013-10-06 11:22:35 -07:00
e5ca996a08
Xmm read/write cleanup.
2013-10-06 11:21:50 -07:00
9ebbc3535f
vsldoi.
2013-10-06 11:21:38 -07:00
cdb13775e2
Alignment option in xmm read/write.
2013-10-06 10:25:41 -07:00
26e22e2b16
Untested and nasty emulated vperm.
2013-10-06 10:25:30 -07:00
2b798df631
lvsl/lvsr/lvlx/lvrx
2013-10-06 09:48:42 -07:00
e8429b75e0
Significantly faster byte swapping on xmm read/write.
2013-10-04 22:29:19 -07:00
d3bf7813ea
Improve perf of vcmp cr6 update.
2013-10-04 13:32:11 -07:00
e2842ea4b4
fmaddsx, fmsubx, fmsubsx, fcmpu.
2013-10-04 10:30:57 -07:00
091957e72e
Untested srawx/sradix (makes things run further, at least).
2013-10-04 09:34:36 -07:00
0f5d0d992d
vupkd3d d3dcolor (untested).
2013-10-03 21:06:40 -07:00
611902a97e
Fixing vspltw and vrlimi.
2013-10-03 20:19:11 -07:00
bdb5b0eae2
Fixing alignment of xmm load/store.
2013-10-03 20:18:56 -07:00
6d46b51ed4
A few vcmp*'s, float_16_to_32 bit of vupkd3d128 (untested).
2013-10-02 23:25:05 -07:00
21d273e85b
Tracing FPRs and random fixes.
2013-09-30 22:43:29 -07:00
37caa5cea1
vnmsubfp, fixing vrlimi128.
2013-09-30 21:14:23 -07:00
f5b99d7448
Fixing some altivec instruction decode issues.
2013-09-30 17:54:49 -07:00
26ec76ef35
A lot of progress on altivec instructions.
...
Some bad results, still, and some instructions are not decoding right.
2013-09-29 21:05:48 -07:00
3662118246
vmaddfp/vmaddcfp/vmulfp/vsubfp.
2013-09-29 00:02:25 -07:00
5652a150e1
vand/vnor/vor/vxor.
2013-09-28 23:45:16 -07:00
526b1a85b7
Simple altivec loads/stores.
2013-09-28 23:14:43 -07:00
f398ccba56
Skeleton emitter file.
2013-09-28 19:52:33 -07:00
51184f73ea
Disassembling a few instructions.
2013-09-28 19:37:34 -07:00
f4f66ff7f1
Beginning Altivec disassembly/emitting.
2013-09-28 09:14:04 -07:00
b0481472f2
Finally adding thread wait so we kill the spin loop.
2013-09-25 20:55:22 -07:00
d16ab71252
Fixing %d print.
2013-09-25 20:28:42 -07:00
3e0d9af57e
Placeholders for memory access breakpoints.
2013-09-25 20:13:43 -07:00
3706b9a3ce
Ignoring MmFreePhysicalMemory (for now).
...
Games seem to be calling this with 0, which is bogus, so the bug is
elsewhere.
2013-09-25 18:38:18 -07:00
536fbbff88
Adding misc Vd functions.
2013-09-25 18:37:30 -07:00
3a6ef8cdf9
Fixing VdQueryVideoMode.
2013-09-25 18:37:13 -07:00
03a68d4c49
PerformanceCounter.
2013-09-25 18:36:34 -07:00
b7fd4f975d
Fixing ringbuffer pointers and no-op instructions.
2013-09-25 00:46:47 -07:00
03494c6b88
Stubbing methods and cleanup.
2013-09-25 00:46:29 -07:00
91f62918f6
Adding vblank interrupt.
2013-09-25 00:46:09 -07:00
998467ce05
Fixing cntlzd.
2013-09-25 00:21:41 -07:00
8424a668bf
Basic implementation of events.
2013-09-24 21:26:45 -07:00
474ecea277
Adding stubs for SetEvent/ResetEvent.
2013-09-23 22:44:34 -07:00
be1ce73a30
Fixing WaitForSingleObject timeout args.
2013-09-23 22:15:43 -07:00
d0d30ab719
Implementing object table. Not complete, but better.
2013-09-22 18:36:06 -07:00
3e0db586c0
RtlNtStatusToDosError, for successes.
2013-09-22 16:45:41 -07:00
d5b8a2091a
Switching tracing order to prevent skipping traces of branch instructions.
2013-09-06 23:21:08 -07:00
1a3fed942d
Skeleton for KeWaitForSingleObject.
...
Trying to figure out where D3D is getting its object from. It never seems
to create a timer/event/etc.
2013-06-30 10:27:51 -07:00
39ef8d8263
Properly using XapiThreadStartup routines, if present.
2013-06-30 10:27:06 -07:00
8d5e877a03
VdGetCurrentDisplayGamma and VdQueryVideoFlags.
2013-06-30 08:21:45 -07:00
c9a39a8b62
Dumping module info on startup and logging unimplemented kernel calls.
2013-06-30 07:53:17 -07:00
a62426948b
Implementing enough ExGetXConfigSetting for D3D.
2013-06-30 07:52:53 -07:00
6f602d120a
Adding thread ID to trace lines and cleaning up their formatting.
2013-06-30 06:09:08 -07:00
052c70efce
Fixing broken opcode 63 stuff.
2013-06-02 12:01:43 -07:00
b3e93a0d1d
srad.
...
Probably right.
2013-06-02 11:33:07 -07:00
97b4fdb041
Fixing misnamed register enum.
2013-06-02 10:32:07 -07:00
eab588f29f
Fixed some additional stack offsets in DbgPrint.
2013-06-02 10:12:27 -07:00
e5bfe7264b
Corrected issues with DbgPrint & _vnsprintf.
2013-06-02 09:44:31 -07:00
fbaaebdb41
Oops. Tab!
2013-06-02 09:43:27 -07:00
100b87846e
Merge branch 'master' of https://github.com/benvanik/xenia
2013-06-02 08:08:17 -07:00
9bc490579e
64-bit shim memory macros.
2013-06-02 08:07:40 -07:00
c9688bb3e9
DbgPrint & _vnsprintf implementations. VERY HACKY.
2013-06-02 08:07:03 -07:00
b97b8226b2
Fixed a crash when adding a FunctionSymbol with a NULL address.
2013-06-02 07:57:38 -07:00
105aee506d
dcbf disasm.
2013-06-02 07:56:17 -07:00
388c622ecd
fill in some more xenos registers
2013-06-02 17:11:54 +10:00
cf57481a0e
Adding basic register names.
2013-06-01 22:51:09 -07:00
e00e665a09
Basic command buffer processing.
2013-06-01 21:49:14 -07:00
ba16a0d847
Trying to setup writeback and add basic secondary command execution.
2013-06-01 03:08:31 -07:00
5b0fa8f17e
Basic ringbuffer worker. Super hacky, not yet writing back to CPU.
2013-06-01 01:13:44 -07:00
3667821be3
ldarx & stdcx disasm.
2013-05-31 22:01:35 -07:00
31199dd41d
VdHSIOCalibrationLock.
2013-05-31 21:40:15 -07:00
4e6e27d9ac
Zero KeCertMonitorData export.
2013-05-31 21:33:19 -07:00
ef0032ecff
Wiring up register read/write from JIT -> GPU.
2013-05-31 21:23:26 -07:00
edf3a9155b
The stack address needs to start at the end of the allocated memory for the stack, not the beginning.
2013-05-31 20:55:52 -07:00
d53458023e
Fixing/implementing static TLS entries.
2013-05-31 10:01:32 -07:00
38efd3837d
Adding support for detecting the save/rest fpr and vmx blocks.
2013-05-31 01:53:30 -07:00
e6195c4086
fmadd, orc, extsh, extsw.
2013-05-31 01:00:00 -07:00
9bb36c7404
Fixing decoding of the rld* instructions.
2013-05-31 00:38:26 -07:00
dc16653164
Fixing decoding of some of the opcode 63 instructions.
2013-05-31 00:09:10 -07:00
14c3b8a382
Adding a case for tail calls.
...
Still missing the forward tail call case, as there is legit code that
does weird branches like that.
2013-05-30 23:41:25 -07:00
e7665c11eb
Adding basic constant checks and updates to memory instructions.
...
Only those instructions likely to have immediate values have been updated.
It'd be nice to go further than this and a) centralize the ea calculations
b) move this all to the analysis pass c) generate mov's in Read/WriteMemory
using the immediates as bases instead of all the register work.
Also fixing a broken stwcx that was causing crashes.
2013-05-30 23:20:05 -07:00
78086684c6
Now tracking constant values for most operations.
...
This will make it easier to enable memory access optimizations and hacks.
Not all operations are updating the constant registers, only those that
are likely to participate in constant value building. Constant values are
cleared on every basic block, though this may be too zealous it prevents
weird fall-through cases.
2013-05-30 22:50:32 -07:00
7990146001
xex2 loader now guesses sizes to allocate.
...
This may not be right in all cases, but seems to work.
2013-05-29 23:26:34 -07:00
f307bc9eb1
Adding --target= and --trace_registers flags.
...
--target=[xex|iso] makes it possible to specify the thing to run from a
flagfile.
2013-05-29 21:45:54 -07:00
37530d1438
Adding some required video export variables.
2013-05-29 21:26:29 -07:00
6950b21424
Reworking the memory system to not commit 3gb and to properly alloc data.
...
Now only 512MB is committed on startup. Loaded XEXs are placed into their
required addresses in the 0x8... range. Kernel structures are allocated
from the normal heap like other data. There should no longer be any magical
pointers.
2013-05-29 21:00:55 -07:00
61f7f6d28e
Starting work on some of the Vd kernel functions.
2013-05-28 20:00:06 -07:00
2cecc02787
Skeleton GPU files.
2013-05-27 21:45:55 -07:00
16baef3591
Starting physical memory allocation routines.
...
These currently don't do anything special w.r.t. memory addresses/etc,
but will in the future.
2013-05-27 20:09:40 -07:00
f78e7945d4
Fixing malloc alighment and duplicate calls to NtAllocateVirtualMemory.
2013-05-27 19:58:20 -07:00
1d0ec64d40
Adding srd/srw. Fixing shamts in other shifts.
2013-05-27 05:35:41 -07:00
778efdb276
fcfid(z), fctid(z) -- both largely untested.
2013-05-27 05:22:38 -07:00
84fb802166
fadd(s), fdiv(s), fmul(s), fsub(s), fsel, frsp, fabs, fmr, fneg, fnmsubs.
2013-05-27 04:09:07 -07:00
fe6df31280
Finishing off the kernel method conversion.
2013-05-26 23:06:09 -07:00
64eb2aa2b3
Converting RTL methods to the new style.
2013-05-26 23:05:05 -07:00
c4e0347932
NULLs!
2013-05-26 22:53:46 -07:00
Ben Vanik