Commit Graph

17 Commits

Author SHA1 Message Date
Jakob Stoklund Olesen
60daf3e76b Separate immediate and value operands in the instruction format.
Instruction formats are now identified by a signature that doesn't
include the ordering of value operands relative to immediate operands.

This means that the BinaryRev instruction format becomes redundant, so
delete it. The isub_imm instruction was the only one using that format.
Rename it to irsub_imm to make it clear what it does now that it is
printed as 'irsub_imm v2, 45'.
2017-03-10 11:20:39 -08:00
Jakob Stoklund Olesen
582a048089 Convert return formats to value lists.
With the Return and ReturnReg formats converted to using value lists for
storing their arguments, thee are no remaining instruction formats with
variable argument lists in boxed storage.

The Return and ReturnReg formats are also going to be merged since
they are identical now.
2017-03-09 15:56:33 -08:00
Jakob Stoklund Olesen
1135a89af9 Convert the Branch and Jump instruction formats to value_list.
The Branch format also stores its fixed argument in the value list. This
requires the value pool to be passed to a few more functions.

Note that this actually makes the Branch and Jump variants of
InstructionData identical. The instruction format hashing does not yet
understand that all value operands are stored in the value list. We'll
fix that in a later patch.

Also convert IndirectCall, noting that Call and IndirectCall remain
separate instruction formats because they have different immediate
fields.
2017-03-09 15:38:12 -08:00
Jakob Stoklund Olesen
364b8e5f0a Use value lists for call arguments.
Add a new kind of instruction format that keeps all of its value
arguments in a value list. These value lists are all allocated out of
the dfg.value_lists memory pool.

Instruction formats with the value_list property set store *all* of
their value arguments in a single value list. There is no distinction
between fixed arguments and variable arguments.

Change the Call instruction format to use the value list representation
for its arguments.

This change is only the beginning. The intent is to eliminate the
boxed_storage instruction formats completely. Value lists use less
memory, and when the transition is complete, InstructionData will have a
trivial Drop implementation.
2017-03-09 14:34:07 -08:00
Jakob Stoklund Olesen
58756e5d34 Add is_call and is_return instruction attributes. 2017-03-08 14:48:50 -08:00
Jakob Stoklund Olesen
25677d1bd8 Add vsplit and vconcat instructions.
Add support for two new type variable functions: half_vector() and
double_vector().

Use these two instructions to break down unsupported SIMD types and
build them up again.
2017-03-07 14:31:57 -08:00
Angus Holder
3d0240d244 Documentation fix for what appears to be a minor copy-paste mistake. 2017-02-22 09:33:17 -08:00
Jakob Stoklund Olesen
20ff2f0025 Add a return_reg instruction to the base instruction set.
Register-style return is used by all RISC architectures, so it is
natural to have a shared instruction representation.
2017-02-21 13:05:17 -08:00
Dominik Inführ
93aa2b456e added Opcode flags methods
generate `is_branch`, `is_terminator`, `can_trap` methods for `enum
Opcode`.
2016-12-02 15:50:28 -08:00
Jakob Stoklund Olesen
19ac05577c Fill in boilerplate for Intel and ARM targets.
The intel, arm32, and arm32 targets were only defined in the meta
language previously. Add Rust implementations too.

This is mostly boilerplate, except for the unit tests in the
registers.rs files.
2016-11-23 10:42:07 -08:00
Jakob Stoklund Olesen
d76280afba Add expansion patterns for large immediates.
Expand foo_imm into iconst + foo.
2016-11-10 15:55:54 -08:00
Jakob Stoklund Olesen
09204ca14a Move ast, xform, and legalize modules.
- cdsl.ast defines classes representing abstract syntax trees.
- cdsl.xform defines classes for instruction transformations.
- base.legalize defines legalization patterns.
2016-11-08 12:33:50 -08:00
Jakob Stoklund Olesen
5fa322f797 Split out instruction definitions.
- cdsl.instructions defines the Instruction class.
- base.instructions defines the base instruction set.
2016-11-08 12:33:50 -08:00
Jakob Stoklund Olesen
2a15130518 Split out instruction formats.
- cdsl.formats provides classes for describing instruction formats.
- base.formats provides concrete instruction format definitions.
2016-11-08 11:48:05 -08:00
Jakob Stoklund Olesen
6eaa8eb382 Move formats, entities, and immediates to the base package.
- base.formats defines instruction formats.
- base.entities defines kinds of entity references.
- base.immediates defines kinds of imediate operands.
2016-11-08 11:06:37 -08:00
Jakob Stoklund Olesen
2fe61e83f6 Split out predicates and settings.
- cdsl.predicates defines classes for describing predicates.
- cdsl.settings defines classes for describing settings.
- base.settings defines shared settings.
2016-11-08 10:37:17 -08:00
Jakob Stoklund Olesen
f8545574b5 Move ValueType into a new cdsl top-level module.
We want to separate the Python classes that make up the DSL used to
define the Cretonne language from the concrete definitions.

- cdsl.types defines the ValueType class hierarchy.
- base.types defines the concrete types.
2016-11-08 10:12:47 -08:00