Commit Graph

26 Commits

Author SHA1 Message Date
Irit Katriel b400219df5
gh-98831: rewrite RAISE_VARARGS in the instruction definition DSL (#101306) 2023-01-25 22:29:56 +00:00
Irit Katriel 19f90d6b97
gh-98831: add variable stack effect support to cases generator (#101309) 2023-01-25 20:41:03 +00:00
Guido van Rossum 395871e511
GH-98831: Elaborate some cases_generator tests (#101299)
* Make macro test more elaborate
* Add test for 'register inst()'
2023-01-25 08:55:46 -08:00
Irit Katriel 1a9d8c750b
gh-98831: rewrite pattern matching opcodes in the instruction definition DSL (#101287) 2023-01-24 22:39:13 +00:00
Irit Katriel 8a2d4f4e8e
gh-98831: register instructions have 0 pushes and pops (#101163) 2023-01-19 21:02:56 +00:00
Guido van Rossum 80e3e3423c
GH-98831: Implement array support in cases generator (#100912)
You can now write things like this:
```
inst(BUILD_STRING, (pieces[oparg] -- str)) { ... }
inst(LIST_APPEND, (list, unused[oparg-1], v -- list, unused[oparg-1])) { ... }
```
Note that array output effects are only partially supported (they must be named `unused` or correspond to an input effect).
2023-01-17 15:59:19 -08:00
Guido van Rossum 5134ef4878
GH-98831: Identify instructions that don't use oparg (#100957)
For these the instr_format field uses IX instead of IB.
Register instructions use IX, IB, IBBX, IBBB, etc.

Also: Include the closing '}' in Block.tokens, for completeness
2023-01-13 17:06:45 -08:00
Guido van Rossum 3f3c78e32f
GH-98831: Refactor instr format code and change to enum (#100895) 2023-01-09 15:53:01 -08:00
Guido van Rossum 9ffbc58f5c
GH-98831: Add some tests for generate_cases.py (#100763)
- This doesn't cover everything (far from it) but it's a start.
- This uses pytest, which isn't ideal, but was quickest to get started.

Co-authored-by: Kumar Aditya <59607654+kumaraditya303@users.noreply.github.com>
2023-01-06 08:04:20 -08:00
Guido van Rossum 14b7f00fdf
GH-98831: Update generate_cases.py: register inst, opcode_metadata.h (#100735)
(These aren't used yet, but may be coming soon,
and it's easier to keep this tool the same between branches.)

Added a sanity check for all this to compile.c.

Co-authored-by: Irit Katriel <iritkatriel@yahoo.com>
2023-01-05 13:01:07 -08:00
Guido van Rossum 08e5594cf3
GH-98831: Modernize a ton of simpler instructions (#100545)
* load_const and load_fast aren't families for now
* Don't decref unmoved names
* Modernize GET_ANEXT
* Modernize GET_AWAITABLE
* Modernize ASYNC_GEN_WRAP
* Modernize YIELD_VALUE
* Modernize POP_EXCEPT (in more than one way)
* Modernize PREP_RERAISE_STAR
* Modernize LOAD_ASSERTION_ERROR
* Modernize LOAD_BUILD_CLASS
* Modernize STORE_NAME
* Modernize LOAD_NAME
* Modernize LOAD_CLASSDEREF
* Modernize LOAD_DEREF
* Modernize STORE_DEREF
* Modernize COPY_FREE_VARS (mark it as done)
* Modernize LIST_TO_TUPLE
* Modernize LIST_EXTEND
* Modernize SET_UPDATE
* Modernize SETUP_ANNOTATIONS
* Modernize DICT_UPDATE
* Modernize DICT_MERGE
* Modernize MAP_ADD
* Modernize IS_OP
* Modernize CONTAINS_OP
* Modernize CHECK_EXC_MATCH
* Modernize IMPORT_NAME
* Modernize IMPORT_STAR
* Modernize IMPORT_FROM
* Modernize JUMP_FORWARD (mark it as done)
* Modernize JUMP_BACKWARD (mark it as done)
2022-12-27 17:11:03 -08:00
Guido van Rossum 9cdd2fa63b
GH-98831: Add DECREF_INPUTS(), expanding to DECREF() each stack input (#100205)
The presence of this macro indicates that a particular instruction
may be considered for conversion to a register-based format
(see https://github.com/faster-cpython/ideas/issues/485).

An invariant (currently unchecked) is that `DEOPT_IF()` may only
occur *before* `DECREF_INPUTS()`, and `ERROR_IF()` may only occur
*after* it. One reason not to check this is that there are a few
places where we insert *two* `DECREF_INPUTS()` calls, in different
branches of the code. The invariant checking would have to be able
to do some flow control analysis to understand this.

Note that many instructions, especially specialized ones,
can't be converted to use this macro straightforwardly.
This is because the generator currently only generates plain
`Py_DECREF(variable)` statements, and cannot generate
things like `_Py_DECREF_SPECIALIZED()` let alone deal with
`_PyList_AppendTakeRef()`.
2022-12-16 20:45:55 -08:00
Mark Shannon 6997e77bdf
GH-100222: Redefine _Py_CODEUNIT as a union to clarify structure of code unit. (GH-100223) 2022-12-14 11:12:53 +00:00
Guido van Rossum 1cfa704f64
GH-98831: Generate things in the input order (#100123)
This makes it easier to see what changed in the generated code
when converting an instruction to super or macro.
2022-12-08 15:54:07 -08:00
Guido van Rossum c85be734d1
GH-98831: Typed stack effects, and more instructions converted (#99764)
Stack effects can now have a type, e.g. `inst(X, (left, right -- jump/uint64_t)) { ... }`.

Instructions converted to the non-legacy format:

* COMPARE_OP
* COMPARE_OP_FLOAT_JUMP
* COMPARE_OP_INT_JUMP
* COMPARE_OP_STR_JUMP
* STORE_ATTR
* DELETE_ATTR
* STORE_GLOBAL
* STORE_ATTR_INSTANCE_VALUE
* STORE_ATTR_WITH_HINT
* STORE_ATTR_SLOT, and complete the store_attr family
* Complete the store_subscr family: STORE_SUBSCR{,DICT,LIST_INT}
  (STORE_SUBSCR was alread half converted,
  but wasn't using cache effects yet.)
* DELETE_SUBSCR
* PRINT_EXPR
* INTERPRETER_EXIT (a bit weird, ends in return)
* RETURN_VALUE
* GET_AITER (had to restructure it some)
  The original had mysterious `SET_TOP(NULL)` before `goto error`.
  I assume those just account for `obj` having been decref'ed,
  so I got rid of them in favor of the cleanup implied by `ERROR_IF()`.
* LIST_APPEND (a bit unhappy with it)
* SET_ADD (also a bit unhappy with it)

Various other improvements/refactorings as well.
2022-12-08 13:31:27 -08:00
Guido van Rossum acf9184e6b
GH-98831: Support cache effects in super- and macro instructions (#99601) 2022-12-02 19:57:30 -08:00
Guido van Rossum 8f18ac04d3
GH-98831: Add `macro` and `op` and their implementation to DSL (#99495)
Newly supported interpreter definition syntax:
- `op(NAME, (input_stack_effects -- output_stack_effects)) { ... }`
- `macro(NAME) = OP1 + OP2;`

Also some other random improvements:
- Convert `WITH_EXCEPT_START` to use stack effects
- Fix lexer to balk at unrecognized characters, e.g. `@`
- Fix moved output names; support object pointers in cache
- Introduce `error()` method to print errors
- Introduce read_uint16(p) as equivalent to `*p`

Co-authored-by: Brandt Bucher <brandtbucher@gmail.com>
2022-11-22 16:04:57 -08:00
Guido van Rossum 4f5e1cb00a
GH-98831: Refactor and fix cases generator (#99526)
Also complete cache effects for BINARY_SUBSCR family.
2022-11-17 17:06:07 -08:00
Guido van Rossum e37744f289
GH-98831: Implement basic cache effects (#99313) 2022-11-15 19:59:19 -08:00
Brandt Bucher c7f5708714
GH-98686: Get rid of "adaptive" and "quick" instructions (GH-99182) 2022-11-09 10:50:09 -08:00
Guido van Rossum f1a654648b
GH-98831: Simple input-output stack effects for bytecodes.c (#99120) 2022-11-08 08:22:56 -08:00
Guido van Rossum 7dcd28eb41
GH-98831: Implement super-instruction generation (#99084)
Co-authored-by: C.A.M. Gerlach <CAM.Gerlach@Gerlach.CAM>
2022-11-06 09:40:47 -08:00
Guido van Rossum d04899abb0
GH-99104: Update headers for bytecodes.c and generate_cases.py (#99112)
Also tweak the labels near the end of bytecodes.c.
2022-11-04 17:40:43 -07:00
Guido van Rossum c885623e9f
GH-98831: Auto-generate PREDICTED() macro calls (#99102)
Co-authored-by: Brandt Bucher <brandtbucher@gmail.com>
2022-11-04 15:30:17 -07:00
Guido van Rossum 916af11a97
GH-98831: Remove redundant extract_cases.py script (GH-99065)
Debt I owe from PR GH-98830.
2022-11-03 09:50:35 -07:00
Guido van Rossum 41bc101dd6
GH-98831: "Generate" the interpreter (#98830)
The switch cases (really TARGET(opcode) macros) have been moved from ceval.c to generated_cases.c.h. That file is generated from instruction definitions in bytecodes.c (which impersonates a C file so the C code it contains can be edited without custom support in e.g. VS Code).

The code generator lives in Tools/cases_generator (it has a README.md explaining how it works). The DSL used to describe the instructions is a work in progress, described in https://github.com/faster-cpython/ideas/blob/main/3.12/interpreter_definition.md.

This is surely a work-in-progress. An easy next step could be auto-generating super-instructions.

**IMPORTANT: Merge Conflicts**

If you get a merge conflict for instruction implementations in ceval.c, your best bet is to port your changes to bytecodes.c. That file looks almost the same as the original cases, except instead of `TARGET(NAME)` it uses `inst(NAME)`, and the trailing `DISPATCH()` call is omitted (the code generator adds it automatically).
2022-11-02 21:31:26 -07:00