This makes the Tier 2 interpreter a little faster.
I calculated by about 3%,
though I hesitate to claim an exact number.
This starts by doubling the trace size limit (to 512),
making it more likely that loops fit in a trace.
The rest of the approach is to only load
`oparg` and `operand` in cases that use them.
The code generator know when these are used.
For `oparg`, it will conditionally emit
```
oparg = CURRENT_OPARG();
```
at the top of the case block.
(The `oparg` variable may be referenced multiple times
by the instructions code block, so it must be in a variable.)
For `operand`, it will use `CURRENT_OPERAND()` directly
instead of referencing the `operand` variable,
which no longer exists.
(There is only one place where this will be used.)
- Ensure that `assert(type_version != 0);` always comes *before* using `type_version`
Also:
- In cases_generator, rename `-v` to from `--verbose` to `--viable`
- Double max trace size to 256
- Add a dependency on executor_cases.c.h for ceval.o
- Mark `_SPECIALIZE_UNPACK_SEQUENCE` as `TIER_ONE_ONLY`
- Add debug output back showing the optimized trace
- Bunch of cleanups to Tools/cases_generator/
* Replace jumps with deopts in tier 2
* Fewer special cases of uop names
* Add target field to uop IR
* Remove more redundant SET_IP and _CHECK_VALIDITY micro-ops
* Extend whitelist of non-escaping API functions.
- There is no longer a separate Python/executor.c file.
- Conventions in Python/bytecodes.c are slightly different -- don't use `goto error`,
you must use `GOTO_ERROR(error)` (same for others like `unused_local_error`).
- The `TIER_ONE` and `TIER_TWO` symbols are only valid in the generated (.c.h) files.
- In Lib/test/support/__init__.py, `Py_C_RECURSION_LIMIT` is imported from `_testcapi`.
- On Windows, in debug mode, stack allocation grows from 8MiB to 12MiB.
- **Beware!** This changes the env vars to enable uops and their debugging
to `PYTHON_UOPS` and `PYTHON_LLTRACE`.
In Python/bytecodes.c, you now write
```
DEOPT_IF(condition);
```
The code generator expands this to
```
DEOPT_IF(condition, opcode);
```
where `opcode` is the name of the unspecialized instruction.
This works inside macro expansions too.
**CAVEAT:** The entire `DEOPT_IF(condition)` statement must be on a single line.
If it isn't, the substitution will fail; an error will be printed by the code generator
and the C compiler will report some errors.
These are the most popular specializations of `LOAD_ATTR` and `STORE_ATTR`
that weren't already viable uops:
* Split LOAD_ATTR_METHOD_WITH_VALUES
* Split LOAD_ATTR_METHOD_NO_DICT
* Split LOAD_ATTR_SLOT
* Split STORE_ATTR_SLOT
* Split STORE_ATTR_INSTANCE_VALUE
Also:
* Add `-v` flag to code generator which prints a list of non-viable uops
(easter-egg: it can print execution counts -- see source)
* Double _Py_UOP_MAX_TRACE_LENGTH to 128
I had dropped one of the DEOPT_IF() calls! :-(
This makes the internal representation in the code generator simpler: there's a list of ops, and a list of macros, and there's no special-casing needed for ops that aren't macros. (There's now special-casing for ops that are also macros, but that's simpler.)
I must have overlooked this when refactoring the code generator.
The Tier 1 interpreter contained a few silly things like
```
goto resume_frame;
STACK_SHRINK(1);
```
(and other variations, some where the unconditional `goto` was hidden in a macro).
* Rename SAVE_IP to _SET_IP
* Rename EXIT_TRACE to _EXIT_TRACE
* Rename SAVE_CURRENT_IP to _SAVE_CURRENT_IP
* Rename INSERT to _INSERT (This is for Ken Jin's abstract interpreter)
* Rename IS_NONE to _IS_NONE
* Rename JUMP_TO_TOP to _JUMP_TO_TOP
This adds a 16-bit inline cache entry to the conditional branch instructions POP_JUMP_IF_{FALSE,TRUE,NONE,NOT_NONE} and their instrumented variants, which is used to keep track of the branch direction.
Each time we encounter these instructions we shift the cache entry left by one and set the bottom bit to whether we jumped.
Then when it's time to translate such a branch to Tier 2 uops, we use the bit count from the cache entry to decided whether to continue translating the "didn't jump" branch or the "jumped" branch.
The counter is initialized to a pattern of alternating ones and zeros to avoid bias.
The .pyc file magic number is updated. There's a new test, some fixes for existing tests, and a few miscellaneous cleanups.
Also remove NOP instructions.
The "stubs" are not optimized in this fashion (their SAVE_IP should always be preserved since it's where to jump next, and they don't contain NOPs by their nature).
Instead of using `GO_TO_INSTRUCTION(CALL_PY_EXACT_ARGS)` we just add the macro elements of the latter to the macro for the former. This requires lengthening the uops array in struct opcode_macro_expansion. (It also required changes to stacking.py that were merged already.)
I was comparing the last preceding poke with the *last* peek,
rather than the *first* peek.
Unfortunately this bug obscured another bug:
When the last preceding poke is UNUSED, the first peek disappears,
leaving the variable unassigned. This is how I fixed it:
- Rename CopyEffect to CopyItem.
- Change CopyItem to contain StackItems instead of StackEffects.
- Update those StackItems when adjusting the manager higher or lower.
- Assert that those StackItems' offsets are equivalent.
- Other clever things.
---------
Co-authored-by: Irit Katriel <1055913+iritkatriel@users.noreply.github.com>
* Add missing includes.
* Remove unused includes.
* Update old include/symbol names to newer names.
* Mention at least one included symbol.
* Sort includes.
* Update Tools/cases_generator/generate_cases.py used to generated
pycore_opcode_metadata.h.
* Update Parser/asdl_c.py used to generate pycore_ast.h.
* Cleanup also includes in _testcapimodule.c and _testinternalcapi.c.
* pycore_intrinsics.h does nothing if included twice
(add #ifndef and #define).
* Update Tools/cases_generator/generate_cases.py to generate the
Py_BUILD_CORE test.
* _bz2, _lzma, _opcode and zlib extensions now define the
Py_BUILD_CORE_MODULE macro to use internal headers
(pycore_code.h, pycore_intrinsics.h and pycore_blocks_output_buffer.h).
This finishes the work begun in gh-107760. When, while projecting a superblock, we encounter a call to a short, simple function, the superblock will now enter the function using `_PUSH_FRAME`, continue through it, and leave it using `_POP_FRAME`, and then continue through the original code. Multiple frame pushes and pops are even possible. It is also possible to stop appending to the superblock in the middle of a called function, when running out of space or encountering an unsupported bytecode.
* Split `CALL_PY_EXACT_ARGS` into uops
This is only the first step for doing `CALL` in Tier 2.
The next step involves tracing into the called code object and back.
After that we'll have to do the remaining `CALL` specialization.
Finally we'll have to deal with `KW_NAMES`.
Note: this moves setting `frame->return_offset` directly in front of
`DISPATCH_INLINED()`, to make it easier to move it into `_PUSH_FRAME`.
- Generalize the syntax for the type of a stack effect to allow a trailing `*`,
so we can declare something as e.g. `PyCodeObject *`.
- When generating assignments for stack effects,
the type of the value on the stack should be the default (i.e., `PyObject *`)
even when the variable copied to/from it has a different type,
so that an appropriate cast is generated
However, not when the variable is an array --
then the type is taken from the variable (as it is always `PyObject **`).
This fixes two tiny defects in analysis.py that I didn't catch on time in #107564:
- `get_var_names` in `check_macro_consistency` should skip `UNUSED` names.
- Fix an occurrence of `is UNUSED` (should be `==`).
Introducing a new file, stacking.py, that takes over several responsibilities related to symbolic evaluation of push/pop operations, with more generality.
This mostly extracts a whole bunch of stuff out of generate_cases.py into separate files, but there are a few other things going on here.
- analysis.py: `Analyzer` etc.
- instructions.py: `Instruction` etc.
- flags.py: `InstructionFlags`, `variable_used`, `variable_used_unspecialized`
- formatting.py: `Formatter` etc.
- Rename parser.py to parsing.py, to avoid conflict with stdlib parser.py
- Blackify most things
- Fix most mypy errors
- Remove output filenames from Generator state, add them to `write_instructions()` etc.
- Fix unit tests
This restores a corner case: when the generator is run with working directory set to Tools/cases_generator, the source filenames listed in the generated provenance header should be relative to the repo root directory.
By turning `assert(kwnames == NULL)` into a macro that is not in the "forbidden" list, many instructions that formerly were skipped because they contained such an assert (but no other mention of `kwnames`) are now supported in Tier 2. This covers 10 instructions in total (all specializations of `CALL` that invoke some C code):
- `CALL_NO_KW_TYPE_1`
- `CALL_NO_KW_STR_1`
- `CALL_NO_KW_TUPLE_1`
- `CALL_NO_KW_BUILTIN_O`
- `CALL_NO_KW_BUILTIN_FAST`
- `CALL_NO_KW_LEN`
- `CALL_NO_KW_ISINSTANCE`
- `CALL_NO_KW_METHOD_DESCRIPTOR_O`
- `CALL_NO_KW_METHOD_DESCRIPTOR_NOARGS`
- `CALL_NO_KW_METHOD_DESCRIPTOR_FAST`
This moves EXIT_TRACE, SAVE_IP, JUMP_TO_TOP, and
_POP_JUMP_IF_{FALSE,TRUE} from ceval.c to bytecodes.c.
They are no less special than before, but this way
they are discoverable o the copy-and-patch tooling.
During superblock generation, a JUMP_BACKWARD instruction is translated to either a JUMP_TO_TOP micro-op (when the target of the jump is exactly the beginning of the superblock, closing the loop), or a SAVE_IP + EXIT_TRACE pair, when the jump goes elsewhere.
The new JUMP_TO_TOP instruction includes a CHECK_EVAL_BREAKER() call, so a closed loop can still be interrupted.
- Hand-written uops JUMP_IF_{TRUE,FALSE}.
These peek at the top of the stack.
The jump target (in superblock space) is absolute.
- Hand-written translation for POP_JUMP_IF_{TRUE,FALSE},
assuming the jump is unlikely.
Once we implement jump-likelihood profiling,
we can implement the jump-unlikely case (in another PR).
- Tests (including some test cleanup).
- Improvements to len(ex) and ex[i] to expose the whole trace.
This adds several of unspecialized opcodes to superblocks:
TO_BOOL, BINARY_SUBSCR, STORE_SUBSCR,
UNPACK_SEQUENCE, LOAD_GLOBAL, LOAD_ATTR,
COMPARE_OP, BINARY_OP.
While we may not want that eventually, for now this helps finding bugs.
There is a rudimentary test checking for UNPACK_SEQUENCE.
Once we're ready to undo this, that would be simple:
just replace the call to variable_used_unspecialized
with a call to variable_used (as shown in a comment).
Or add individual opcdes to FORBIDDEN_NAMES_IN_UOPS.
Instead of special-casing specific instructions,
we add a few more special values to the 'size' field of expansions,
so in the future we can automatically handle
additional super-instructions in the generator.
- Tweak uops debugging output
- Fix the bug from gh-106290
- Rename `SET_IP` to `SAVE_IP` (per https://github.com/faster-cpython/ideas/issues/558)
- Add a `SAVE_IP` uop at the start of the trace (ditto)
- Allow `unbound_local_error`; this gives us uops for `LOAD_FAST_CHECK`, `LOAD_CLOSURE`, and `DELETE_FAST`
- Longer traces
- Support `STORE_FAST_LOAD_FAST`, `STORE_FAST_STORE_FAST`
- Add deps on pycore_uops.h to Makefile(.pre.in)
This produces longer traces (superblocks?).
Also improved debug output (uop names are now printed instead of numeric opcodes). This would be simpler if the numeric opcode values were generated by generate_cases.py, but that's another project.
Refactored some code in generate_cases.py so the essential algorithm for cache effects is only run once. (Deciding which effects are used and what the total cache size is, regardless of what's used.)
Added a new, experimental, tracing optimizer and interpreter (a.k.a. "tier 2"). This currently pessimizes, so don't use yet -- this is infrastructure so we can experiment with optimizing passes. To enable it, pass ``-Xuops`` or set ``PYTHONUOPS=1``. To get debug output, set ``PYTHONUOPSDEBUG=N`` where ``N`` is a debug level (0-4, where 0 is no debug output and 4 is excessively verbose).
All of this code is likely to change dramatically before the 3.13 feature freeze. But this is a first step.
Mark Shannon
Irit Katriel
Guido van Rossum
Brandt Bucher
Carl Meyer
Heinz-Alexander Fuetterer
Alex Waygood
Victor Stinner
Dong-hee Na
Ken Jin
Nikita Sobolev
Kevin Diem
Kirill Podoprigora
Benjamin Peterson