Commit Graph

110 Commits

Author SHA1 Message Date
gbeauche
6a7c8f7e83 Add PowerPC tester glue for Microlib CPU core 2003-11-28 15:12:37 +00:00
gbeauche
3bea82fa1c fix merge, hunks were missing 2003-11-27 23:59:00 +00:00
gbeauche
8ca440d0b5 Fix SRAW on non PowerPC platforms. 2003-11-27 23:53:41 +00:00
gbeauche
8711c4afd6 Add support for external results file for non PowerPC platforms. 2003-11-27 23:52:19 +00:00
gbeauche
687b9c5a74 I manually synchronize with Kheperix code. 2003-11-27 20:07:33 +00:00
gbeauche
aee3e05e4b The JIT should work now but there is an extra bottleneck causing it to
not match Kheperix speeds while executing Linux/ppc binaries.

Fix x86 DYNGEN_OP_FLAGS settings. Only allocate translation cache in .data
on PPC since x86 can do jumps anywhere.
2003-11-27 11:09:38 +00:00
gbeauche
2bacb2fd01 Workaround CR expectations in MODE_68K execution 2003-11-27 11:06:23 +00:00
gbeauche
d7ac6a0e68 Fix SLW & SRW, an x86 does not work the same way as a ppc 2003-11-27 10:53:37 +00:00
gbeauche
ae2d91912c fix dummy includes 2003-11-27 10:06:27 +00:00
gbeauche
36ce9c07e6 Statically allocate the translation cache on PowerPC. This makes it possible
to generate direct bl instructions for function invokation.
2003-11-27 00:26:35 +00:00
gbeauche
e30001bc00 Fix BCCTR & BCLR. However, conditions are still wrong somehow, disabled
this case. Factored & optimized branch instructions.
2003-11-26 23:58:14 +00:00
gbeauche
2eba241021 self credit cpu emulator ;-) 2003-11-25 10:27:59 +00:00
gbeauche
73d51962f6 Merge in-progress PowerPC "JIT1" engine for AMD64, IA-32, PPC.
The merge probably got wrong as there are some problems probably due to the
experiment begining with CR deferred evaluation. With nbench/ppc, performance
improvement was around 2x. With nbench on x86, performance improvement was
around 4x on average.

Incompatible change: instr_info_t has a new field in the middle. But since
insertion of PPC_I(XXX) identifiers is auto-generated, there is no problem.
2003-11-24 23:45:52 +00:00
gbeauche
7968a20100 Handle "JIT1" engine but disable it for now since there are some problems
with SheepShaver integration from Kheperix.
2003-11-24 23:39:35 +00:00
gbeauche
09a774ae51 Add link to B2 Unix/config.{guess,sub} 2003-11-24 22:19:35 +00:00
gbeauche
2a0f750a83 Optimize memory accesses on little endian systems that can do unaligned
accesses to memory. Fix build when vm.hpp is included in a C program.
2003-11-24 21:30:17 +00:00
gbeauche
04349eebce Optimized bswap_32() for AMD64 2003-11-24 21:20:47 +00:00
gbeauche
1dbe1179c6 Merge in "keycodes" support from Basilisk II. e.g. make French keyboard
layout work correctly for me.
2003-11-21 17:01:33 +00:00
gbeauche
5b950fa2ef fix shm screen image allocation. 2003-11-20 16:24:57 +00:00
gbeauche
b7c917e6d1 little endian fixes to name registry 2003-11-20 15:54:10 +00:00
gbeauche
e9f3546539 Remove even more obsolete code. Drop TBL/TBU registers, they are manually
handled through the mftb instruction accessor.
2003-11-11 11:44:34 +00:00
gbeauche
cf0ed72f24 Remove obsolete code related to PPC_NO_FPSCR_UPDATE, PPC_LAZY_PC_UPDATE,
PPC_LAZY_CC_UPDATE, PPC_HAVE_SPLIT_CR defines.
2003-11-11 11:32:27 +00:00
gbeauche
b66d8ef433 Fix "ignoresegv" case to actually skip the faulty instruction. Merge
conditions to skip instruction on SIGSEGVfrom PowerPC native mode. The
instruction skipper takes care to set the output register to 0.
2003-11-10 16:23:58 +00:00
gbeauche
26ec1b8899 Merge Set_pthread_attr() from Basilisk II. 2003-11-10 16:05:52 +00:00
gbeauche
0260210ddf - XLM_IRQ_NEST is always in native byte order format since any write to
this variable go through {Enable,Disable}Interrupt().
- Add Ether thunks but only for WORDS_BIGENDIAN case since we do need more
complicated translation functions.
2003-11-10 15:11:44 +00:00
gbeauche
cbb8efd492 little endian fixes 2003-11-10 14:18:34 +00:00
gbeauche
cd86ff9e94 - Start emulating the FPSCR. Fix mtfsf, mffs.
- Implement mftbr so that MacOS can fully boot with extensions. However,
  using clock() is probably not the right solution. Patching UpTime from
  DriverServicesLib et al. may be a better solution.
2003-11-09 15:39:30 +00:00
gbeauche
59e6227c08 fix mullwo & divw on invalid inputs 2003-11-09 07:19:39 +00:00
gbeauche
aebcb7a6bb New testing framework faster to compile and more flexible. i.e. we now
generate 350K+ instructions. This exhausts errors for mullwo & divw.
2003-11-08 11:57:04 +00:00
gbeauche
4b73163083 Fix PPC_LAZY_CC_UPDATE build. TODO: remove since this is slower. 2003-11-04 22:01:36 +00:00
gbeauche
175dfeea02 fix lfs/stfs breakage introduced with latest FPR type change 2003-11-04 20:56:21 +00:00
gbeauche
8c40d739b6 Add some statistics for interrupt handling, Execute68k/Trap, MacOS & NativeOp 2003-11-04 20:48:29 +00:00
gbeauche
42e1cabc94 Move variables for compile statistics to powerpc_cpu private data 2003-11-04 20:45:46 +00:00
gbeauche
30bd089279 PowerPC floating-point registers are now an union of uint64 & double. This
eases FP load/stores.
2003-11-04 15:03:15 +00:00
gbeauche
8ddf749ed5 fix vm_do_read_memory_8() 2003-11-04 15:00:02 +00:00
gbeauche
a42281aad1 Implement partial block cache invalidation. Rewrite core cached blocks
execution loop with a Duff's device. Gather some predecode time statistics.
This shows that only around 2% of total emulation time is spent for
predecoding the instructions.
2003-11-03 21:28:32 +00:00
gbeauche
f0ea192460 Optimized pointers to non virtual member functions. This reduces space
and overhead since runtime checks are eliminated. Actually, it yields
up to 10% performance improvement with specialized decoders.
2003-11-02 14:48:20 +00:00
gbeauche
d956d3c4ca add specialized instruction decoders (disabled for now) 2003-11-01 17:07:17 +00:00
gbeauche
066af7452a fix ppc-execute.o dependency 2003-11-01 17:03:55 +00:00
gbeauche
89d0f9ca29 Integrate spcflags handling code to kpx_cpu core. We can also remove
oldish EXEC_RETURN handling with a throw/catch mechanism since we
do have a dependency on extra conditions (invalidated cache) that
prevents fast execution loops.
2003-11-01 15:15:31 +00:00
gbeauche
9ce43c6cf3 Fix ASYNC_IRQ build but locks may still happen. Note that with a predecode
cache, checking for pending interrupts may not be the bottle neck nowadays.
2003-10-26 14:16:40 +00:00
gbeauche
60d34a6816 Rewrite interrupts handling code so that the emulator can work with a
predecode cache. This implies to run in interpreted mode only while
processing EmulOps or other native (nested) runs.

Note that the FLIGHT_RECORDER with a predecode cache gets slower than
without caching at all.
2003-10-26 13:59:04 +00:00
gbeauche
d766049d59 - enable multicore cpu emulation with ASYNC_IRQ
- move atomic_* operations to main_unix so that they could use spinlocks or
  other platform-specific locking mechanisms
2003-10-26 09:14:14 +00:00
gbeauche
7312739738 fix MakeExecutable patch for little endian systems 2003-10-26 08:48:48 +00:00
gbeauche
46a1b4ada5 allow DGA & Xshm only on local displays 2003-10-26 07:54:02 +00:00
cebix
5f0a739cc7 - fixed compilation problems under BeOS
- boot drive wasn't set correctly
2003-10-26 00:32:31 +00:00
gbeauche
ccf89d9efb Preserve CR in execute_68k(). This enables MacOS 8.6 to work. ;-) 2003-10-19 21:37:43 +00:00
gbeauche
cb13fe3007 Log both r24 (m68k emulator PC) & stack pointer in SheepShaver mode only 2003-10-19 21:36:21 +00:00
gbeauche
9a05805a27 - Fix ADDME & ADDZE decoders, add RA==R0 testers
- Increase predecode cache size to 32K entries
- Enable PPC_EXECUTE_DUMP_STATE for predecode cache as well
2003-10-18 13:43:25 +00:00
gbeauche
1b9876889e - Record address range of block to invalidate. i.e. icbi records ranges
and isync actually invalidate caches
2003-10-12 06:44:04 +00:00