This PR is the first of two PRs that replaces earlier PRs #2589 and #2590.
Due to a git branching mishap it was decided to re-partition the new
functionality in two sequential PRs that offer self-contained, new
functionality to sim65.
The functionality in this first PR extends the sim65 simulator in the following ways:
(1) It provides tracing functionality, i.e., the possibility of printing one line of simulator state information per instruction executed.
(2) It provides a memory mapped "sim65 control" peripheral that allows control of (a) the tracing functionality, and (b) the cpu mode.
(3) It provides command-line options to sim65 to enable the tracing, and to override the CPU mode as specified in the program file header.
More detailed information and some discussion can be found in the discussions with the (now retracted) PRs #2589 and #2590.
This PR provides the technical infrastructure inside the sim65 simulator program itself. Once this PR is accepted, a follow-up PR will be posted that adds C and assembly-language support for the new tracing and peripheral features so they can be easily accessed from the CC65 compiler and the CA65 assembler; some examples; and the documentation for these features. The lack of the latter, in this pull request, will be addressed then.
It provides access to a handful of 64-bit counters that count different things:
- clock cycles
- instructions
- number of IRQ processed
- number of NMIs processed
- nanoseconds since 1-1-1970.
This in not ready yet to be pushed as a merge request into the upstream CC65
repository. What's lacking:
- documentation
- tests
And to be discussed:
- do we agree on this implementation direction and interface in principe?
- can I include inttypes.h for printing a 64-bit unsigned value?
- will clock_gettime() work on a Windows build?
This PR is mostly a complete rewrite of the emulator for 6502/65c02 opcodes.
It provides an easier to maintain implementation of the instructions, by using few macros rather than having hand-written code for each function.
All undocumented, previously missing opcodes for 6502 are also implemented.
The patch also includes a detailed documentation of those opcodes, for reference to developers.
This PR should fix one of the milestones listed here for the next version of CC65:
https://github.com/cc65/wiki/wiki/Before-the-next-release
MaxCycleCount is accounted by countdown, eliminating the 1-instruction-overhead issue, and removing the need to compare against a growing TotalCycles.
Makes main.c responsible for counting total cycles, instead of 6502.c, so the size of MaxCycleCount etc. is fully determined in one location.
Makes error.c responsible for PrintCycles instead of paravirt.c, so that it can be treated globally instead of
Return value of main() should be SIM65_ERROR because it is unreachable by design.
long is 64-bit on some platforms, making this inconsistent, added range check to catch overflow.
reduced tests requesting 5 billion cycles to 2^32-1 so they can fun on 32-bit long sim65.
load and run address now configured from header
fix error codes not to conflict with test
fix test/misc/endless.c which is supposed to fail if an endless loop does not occur
The sim65 source code has been a construction site for over a decade.
I was looking for a simple cc65 program execution environment for
regression tests. So I decided to re-purpose sim65 for that task by
removing about everything but the 6502 emulation.
There's no memory mapped i/o emulation whatsoever. Rather exit(),
open(), close(), read() and write() calls are supported by mapping
them through a thin paravirtualization layer to the corresponding
host os functions.
Note: The sim65 6502 emulation provides means to switch between
6502 and 65C02 emulation but currently there are no actual 65C02
opcodes implemented.
implementation. Allow for release candidates to be specified and disinguished.
git-svn-id: svn://svn.cc65.org/cc65/trunk@4260 b7a2c559-68d2-44c3-8de9-860c34a00d81
