Instead of a primary opcode lookup table with 64 entries and a few
smaller tables with 4-2048 entries, use a single 64 * 2048 (128K)
entry table to dispatch opcodes.
Helps with performance, since we avoid the function call overhead for
some frequently-used instructions (e.g. branch, integer, floating point).
Saves ~2 seconds from the time to Welcome to Macintosh (same measurement
methodology as #125)
Secondarily also makes opcode registration/decoding a bit more uniform,
and scannable, since it's now all in initialize_ppc_opcode_table.
Replace it wth an explicit opcode parameter that is passed around. That
is both slightly easier to reason about (to trace where it comes from)
and slightly faster, since it can be read from a register.
On my machine takes booting to "Welcome to Macintosh" being output in
a verbose boot of Mac OS X 10.2.8 from 31.8s to 30.6s (average of 5
runs, measured using deterministic mode and looking at when execution
reaches PC 0x90004a88).
ppc_opcode16 and other functions are only needed in the implementation in
ppcexec.cpp, they don't need to be in the header.
fp_return_double and fp_return_uint64 have no uses (as of 2141a72b87)
can can thus be removed altogether.
Similarly ppc_fpu_off has no uses (as of bb3f4e596e)
and can be removed.
Adds support for a --deterministic command-line option that makes
repeated runs the same:
- Keyboard and mouse input is ignored
- The sound server does a periodic pull from the DMA channel (so that
it gets drained), but only does so via a periodic timer (instead of
being driven by a cubeb callback, which could arrive at different
times)
- Disk image writes are disabled (reads of a modified area still
work via an in-memory copy)
- NVRAM writes are disabled
- The current time that ViaCuda initializes the guest OS is always the
same.
This makes execution exactly the same each time, which should
make debugging of more subtle issues easier.
To validate that the deterministic mode is working, I've added a
periodic log of the current "time" (measured in cycle count), PC
and opcode. When comparing two runs with --log-no-uptime, the generated
log files are identical.
Shutdown will enter the debugger or quit depending on the execution mode.
Quit is different from shutdown since it is triggered outside the guest by using the host Quit menu item.
There's no reason for it to be a global, we always set it and use it
in instruction implementations, and we never read it directly.
Perhaps the compiler could optimize this away, but it's better to be
simpler (and also be easier to read).
- Rename DEC to DEC_S and add DEC_U.
- MQ, RTCL_U, RTCU_U, and DEC_U should cause an illegal instruction program exception for non-MPC601 CPUs. The exception handler of classic Mac OS uses this to emulate the instruction.
- For mtspr, the SPRs RTCL_U, RTCU_U, and DEC_U are treated as no-op on MPC601.
- For debugging, use the supervisor instead of the user SPR number as the index for storing the values for RTC, TB, and DEC.
- For debugging, RTC, TB, and DEC should be updated after each access. Previously, mfspr and mtspr would only update the half of RTC and TB that was being accessed instead of both halves.
The first option is a flag that enables MPC601 (POWER) instructions for CPUs that are not MPC601.
This can be useful for the following reasons:
1) To produce results similar to classic Mac OS which emulates MPC601 instructions on CPUs that don't implement MPC601 instructions. This option is used to compare the risu traces produced in Mac OS 9 on a G3 or G4 with DPPC.
2) May increase performance in apps that use POWER instructions on emulated machines with CPUs that are not MPC601. It is not known if any such apps exist but there could be since Apple included MPC601 emulation in classic Mac OS.
Use "logical" since the functions deal with multiple bits instead of a single boolean value and because the 601 manual calls them Logical Instructions.
Use "ppc" for the enums because logical_and is defined elsewhere and because the original DPPC code used these names for those functions.
There are cases where when it's necessary (e.g. given uninitialized NVRAM,
the Beige G3 with the 10.2 install CD inserted will update the boot
device and restart to boot from it).
Restart support was done by wrapping the ppc_exec function in a loop and
checking for a restart power off reason. We also need to disconnect all
event listeners, since they will be recreated when the machine is
re-initialized.
Add MPC601 variants. Variants that decrement and test the ctr are invalid bon't don't appear to trigger an exception. The manual says MPC601 can decrement the counter. Other CPUs do not decrement the counter but will branch based on the value.
Typing Control-C in Terminal app causes an interrupt signal that should enter the DPPC debugger but this only worked once since the signal handler never returned. Even if the signal handler reenabled the signal somehow, it calls enter_debugger recursively which is strange since the earlier calls to enter_debugger would never return.
Now the signal handler just sets a flag (power_on) which can be used to exit any loop (emulator loops, stepping loops, disassembly loops, dumping loops).
Main always calls enter_debugger now which calls the ppc_exec loop. The power_on flag will exit the ppc_exec loop to return to the debugger. Recursion of enter_debugger is eliminated except for calls to loguru's ABORT_F.
An enum power_off_reason is used to indicate why the power_on flag is set to false and to determine what happens next.
It was always changing CR1 (starting at CR bit 4) instead of the CR selected by crfD.
Also, it was clearing all but the FL,FG,FE,FU bits of FPRF of FPSCR.
It's used in the main emulator loop (ppc_exec_inner), and the function
call overhead adds up.
By inlining it, time to boot to the Finder using a 7.1.2 install CD
and a 6100 ROM goes from ~6700ms to ~6400ms (with clang 14 on a
M2 Max)
